0 files changed, 0 insertions, 0 deletions
diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig
index fbef3f471bd0..7e6bc0b3a589 100644
--- a/drivers/gpu/drm/Kconfig
+++ b/drivers/gpu/drm/Kconfig
@@ -26,6 +26,11 @@ menuconfig DRM
 	  details.  You should also select and configure AGP
 	  (/dev/agpgart) support if it is available for your platform.
 
+menu "DRM debugging options"
+depends on DRM
+source "drivers/gpu/drm/Kconfig.debug"
+endmenu
+
 if DRM
 
 config DRM_MIPI_DBI
@@ -37,65 +42,6 @@ config DRM_MIPI_DSI
 	bool
 	depends on DRM
 
-config DRM_DEBUG_MM
-	bool "Insert extra checks and debug info into the DRM range managers"
-	default n
-	depends on DRM
-	depends on STACKTRACE_SUPPORT
-	select STACKDEPOT
-	help
-	  Enable allocation tracking of memory manager and leak detection on
-	  shutdown.
-
-	  Recommended for driver developers only.
-
-	  If in doubt, say "N".
-
-config DRM_USE_DYNAMIC_DEBUG
-	bool "use dynamic debug to implement drm.debug"
-	default n
-	depends on BROKEN
-	depends on DRM
-	depends on DYNAMIC_DEBUG || DYNAMIC_DEBUG_CORE
-	depends on JUMP_LABEL
-	help
-	  Use dynamic-debug to avoid drm_debug_enabled() runtime overheads.
-	  Due to callsite counts in DRM drivers (~4k in amdgpu) and 56
-	  bytes per callsite, the .data costs can be substantial, and
-	  are therefore configurable.
-
-config DRM_KUNIT_TEST_HELPERS
-	tristate
-	depends on DRM && KUNIT
-	select DRM_KMS_HELPER
-	help
-	  KUnit Helpers for KMS drivers.
-
-config DRM_KUNIT_TEST
-	tristate "KUnit tests for DRM" if !KUNIT_ALL_TESTS
-	depends on DRM && KUNIT && MMU
-	select DRM_BUDDY
-	select DRM_DISPLAY_DP_HELPER
-	select DRM_DISPLAY_HDMI_STATE_HELPER
-	select DRM_DISPLAY_HELPER
-	select DRM_EXEC
-	select DRM_EXPORT_FOR_TESTS if m
-	select DRM_GEM_SHMEM_HELPER
-	select DRM_KUNIT_TEST_HELPERS
-	select DRM_LIB_RANDOM
-	select PRIME_NUMBERS
-	default KUNIT_ALL_TESTS
-	help
-	  This builds unit tests for DRM. This option is not useful for
-	  distributions or general kernels, but only for kernel
-	  developers working on DRM and associated drivers.
-
-	  For more information on KUnit and unit tests in general,
-	  please refer to the KUnit documentation in
-	  Documentation/dev-tools/kunit/.
-
-	  If in doubt, say "N".
-
 config DRM_KMS_HELPER
 	tristate
 	depends on DRM
@@ -188,7 +134,7 @@ config DRM_DEBUG_DP_MST_TOPOLOGY_REFS
         bool "Enable refcount backtrace history in the DP MST helpers"
 	depends on STACKTRACE_SUPPORT
         select STACKDEPOT
-        depends on DRM_KMS_HELPER
+        select DRM_KMS_HELPER
         depends on DEBUG_KERNEL
         depends on EXPERT
         help
@@ -242,28 +188,12 @@ source "drivers/gpu/drm/display/Kconfig"
 config DRM_TTM
 	tristate
 	depends on DRM && MMU
+	select SHMEM
 	help
 	  GPU memory management subsystem for devices with multiple
 	  GPU memory types. Will be enabled automatically if a device driver
 	  uses it.
 
-config DRM_TTM_KUNIT_TEST
-        tristate "KUnit tests for TTM" if !KUNIT_ALL_TESTS
-        default n
-        depends on DRM && KUNIT && MMU && (UML || COMPILE_TEST)
-        select DRM_TTM
-        select DRM_BUDDY
-        select DRM_EXPORT_FOR_TESTS if m
-        select DRM_KUNIT_TEST_HELPERS
-        default KUNIT_ALL_TESTS
-        help
-          Enables unit tests for TTM, a GPU memory manager subsystem used
-          to manage memory buffers. This option is mostly useful for kernel
-          developers. It depends on (UML || COMPILE_TEST) since no other driver
-          which uses TTM can be loaded while running the tests.
-
-          If in doubt, say "N".
-
 config DRM_EXEC
 	tristate
 	depends on DRM
@@ -278,6 +208,15 @@ config DRM_GPUVM
 	  GPU-VM representation providing helpers to manage a GPUs virtual
 	  address space
 
+config DRM_GPUSVM
+	tristate
+	depends on DRM && DEVICE_PRIVATE
+	select HMM_MIRROR
+	select MMU_NOTIFIER
+	help
+	  GPU-SVM representation providing helpers to manage a GPUs shared
+	  virtual memory
+
 config DRM_BUDDY
 	tristate
 	depends on DRM
@@ -326,7 +265,7 @@ config DRM_SCHED
 	tristate
 	depends on DRM
 
-source "drivers/gpu/drm/i2c/Kconfig"
+source "drivers/gpu/drm/sysfb/Kconfig"
 
 source "drivers/gpu/drm/arm/Kconfig"
 
@@ -336,6 +275,8 @@ source "drivers/gpu/drm/amd/amdgpu/Kconfig"
 
 source "drivers/gpu/drm/nouveau/Kconfig"
 
+source "drivers/gpu/drm/nova/Kconfig"
+
 source "drivers/gpu/drm/i915/Kconfig"
 
 source "drivers/gpu/drm/xe/Kconfig"
@@ -441,19 +382,25 @@ source "drivers/gpu/drm/mcde/Kconfig"
 
 source "drivers/gpu/drm/tidss/Kconfig"
 
+source "drivers/gpu/drm/adp/Kconfig"
+
 source "drivers/gpu/drm/xlnx/Kconfig"
 
 source "drivers/gpu/drm/gud/Kconfig"
 
+source "drivers/gpu/drm/sitronix/Kconfig"
+
 source "drivers/gpu/drm/solomon/Kconfig"
 
 source "drivers/gpu/drm/sprd/Kconfig"
 
 source "drivers/gpu/drm/imagination/Kconfig"
 
+source "drivers/gpu/drm/tyr/Kconfig"
+
 config DRM_HYPERV
 	tristate "DRM Support for Hyper-V synthetic video device"
-	depends on DRM && PCI && MMU && HYPERV
+	depends on DRM && PCI && HYPERV_VMBUS
 	select DRM_CLIENT_SELECTION
 	select DRM_KMS_HELPER
 	select DRM_GEM_SHMEM_HELPER
@@ -465,9 +412,6 @@ config DRM_HYPERV
 
 	 If M is selected the module will be called hyperv_drm.
 
-config DRM_EXPORT_FOR_TESTS
-	bool
-
 # Separate option as not all DRM drivers use it
 config DRM_PANEL_BACKLIGHT_QUIRKS
 	tristate
@@ -480,20 +424,6 @@ config DRM_PRIVACY_SCREEN
 	bool
 	default n
 
-config DRM_WERROR
-	bool "Compile the drm subsystem with warnings as errors"
-	depends on DRM && EXPERT
-	depends on !WERROR
-	default n
-	help
-	  A kernel build should not cause any compiler warnings, and this
-	  enables the '-Werror' flag to enforce that rule in the drm subsystem.
-
-	  The drm subsystem enables more warnings than the kernel default, so
-	  this config option is disabled by default.
-
-	  If in doubt, say N.
-
 endif
 
 # Separate option because drm_panel_orientation_quirks.c is shared with fbdev
diff --git a/drivers/gpu/drm/Kconfig.debug b/drivers/gpu/drm/Kconfig.debug
new file mode 100644
index 000000000000..05dc43c0b8c5
--- /dev/null
+++ b/drivers/gpu/drm/Kconfig.debug
@@ -0,0 +1,117 @@
+config DRM_USE_DYNAMIC_DEBUG
+	bool "use dynamic debug to implement drm.debug"
+	default n
+	depends on BROKEN
+	depends on DRM
+	depends on DYNAMIC_DEBUG || DYNAMIC_DEBUG_CORE
+	depends on JUMP_LABEL
+	help
+	 Use dynamic-debug to avoid drm_debug_enabled() runtime overheads.
+	 Due to callsite counts in DRM drivers (~4k in amdgpu) and 56
+	 bytes per callsite, the .data costs can be substantial, and
+	 are therefore configurable.
+
+config DRM_WERROR
+	bool "Compile the drm subsystem with warnings as errors"
+	depends on DRM && EXPERT
+	depends on !WERROR
+	default n
+	help
+	  A kernel build should not cause any compiler warnings, and this
+	  enables the '-Werror' flag to enforce that rule in the drm subsystem.
+
+	  The drm subsystem enables more warnings than the kernel default, so
+	  this config option is disabled by default.
+
+	  If in doubt, say N.
+
+config DRM_HEADER_TEST
+	bool "Ensure DRM headers are self-contained and pass kernel-doc"
+	depends on DRM && EXPERT && BROKEN
+	default n
+	help
+	  Ensure the DRM subsystem headers both under drivers/gpu/drm and
+	  include/drm compile, are self-contained, have header guards, and have
+	  no kernel-doc warnings.
+
+	  If in doubt, say N.
+
+config DRM_DEBUG_MM
+	bool "Insert extra checks and debug info into the DRM range managers"
+	default n
+	depends on DRM
+	depends on STACKTRACE_SUPPORT
+	select STACKDEPOT
+	help
+	  Enable allocation tracking of memory manager and leak detection on
+	  shutdown.
+
+	  Recommended for driver developers only.
+
+	  If in doubt, say "N".
+
+config DRM_KUNIT_TEST_HELPERS
+	tristate
+	depends on DRM && KUNIT
+	select DRM_KMS_HELPER
+	help
+	  KUnit Helpers for KMS drivers.
+
+config DRM_KUNIT_TEST
+	tristate "KUnit tests for DRM" if !KUNIT_ALL_TESTS
+	depends on DRM && KUNIT && MMU
+	select DRM_BRIDGE_CONNECTOR
+	select DRM_BUDDY
+	select DRM_DISPLAY_DP_HELPER
+	select DRM_DISPLAY_HDMI_STATE_HELPER
+	select DRM_DISPLAY_HELPER
+	select DRM_EXEC
+	select DRM_EXPORT_FOR_TESTS if m
+	select DRM_GEM_SHMEM_HELPER
+	select DRM_KUNIT_TEST_HELPERS
+	select DRM_LIB_RANDOM
+	select DRM_SYSFB_HELPER
+	select PRIME_NUMBERS
+	default KUNIT_ALL_TESTS
+	help
+	  This builds unit tests for DRM. This option is not useful for
+	  distributions or general kernels, but only for kernel
+	  developers working on DRM and associated drivers.
+
+	  For more information on KUnit and unit tests in general,
+	  please refer to the KUnit documentation in
+	  Documentation/dev-tools/kunit/.
+
+	  If in doubt, say "N".
+
+config DRM_TTM_KUNIT_TEST
+	tristate "KUnit tests for TTM" if !KUNIT_ALL_TESTS
+	default n
+	depends on DRM && KUNIT && MMU && (UML || COMPILE_TEST)
+	select DRM_TTM
+	select DRM_BUDDY
+	select DRM_EXPORT_FOR_TESTS if m
+	select DRM_KUNIT_TEST_HELPERS
+	default KUNIT_ALL_TESTS
+	help
+	  Enables unit tests for TTM, a GPU memory manager subsystem used
+	  to manage memory buffers. This option is mostly useful for kernel
+	  developers. It depends on (UML || COMPILE_TEST) since no other driver
+	  which uses TTM can be loaded while running the tests.
+
+	  If in doubt, say "N".
+
+config DRM_SCHED_KUNIT_TEST
+	tristate "KUnit tests for the DRM scheduler" if !KUNIT_ALL_TESTS
+	select DRM_SCHED
+	depends on DRM && KUNIT
+	default KUNIT_ALL_TESTS
+	help
+	  Choose this option to build unit tests for the DRM scheduler.
+
+	  Recommended for driver developers only.
+
+	  If in doubt, say "N".
+
+config DRM_EXPORT_FOR_TESTS
+	bool
diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile
index 19fb370fbc56..0e1c668b46d2 100644
--- a/drivers/gpu/drm/Makefile
+++ b/drivers/gpu/drm/Makefile
@@ -6,7 +6,7 @@
 CFLAGS-$(CONFIG_DRM_USE_DYNAMIC_DEBUG)	+= -DDYNAMIC_DEBUG_MODULE
 
 # Unconditionally enable W=1 warnings locally
-# --- begin copy-paste W=1 warnings from scripts/Makefile.extrawarn
+# --- begin copy-paste W=1 warnings from scripts/Makefile.warn
 subdir-ccflags-y += -Wextra -Wunused -Wno-unused-parameter
 subdir-ccflags-y += $(call cc-option, -Wrestrict)
 subdir-ccflags-y += -Wmissing-format-attribute
@@ -41,6 +41,7 @@ drm-y := \
 	drm_bridge.o \
 	drm_cache.o \
 	drm_color_mgmt.o \
+	drm_colorop.o \
 	drm_connector.o \
 	drm_crtc.o \
 	drm_displayid.o \
@@ -76,7 +77,8 @@ drm-y := \
 drm-$(CONFIG_DRM_CLIENT) += \
 	drm_client.o \
 	drm_client_event.o \
-	drm_client_modeset.o
+	drm_client_modeset.o \
+	drm_client_sysrq.o
 drm-$(CONFIG_DRM_LIB_RANDOM) += lib/drm_random.o
 drm-$(CONFIG_COMPAT) += drm_ioc32.o
 drm-$(CONFIG_DRM_PANEL) += drm_panel.o
@@ -105,6 +107,11 @@ obj-$(CONFIG_DRM_PANEL_BACKLIGHT_QUIRKS) += drm_panel_backlight_quirks.o
 obj-$(CONFIG_DRM_EXEC) += drm_exec.o
 obj-$(CONFIG_DRM_GPUVM) += drm_gpuvm.o
 
+drm_gpusvm_helper-y := \
+	drm_gpusvm.o\
+	drm_pagemap.o
+obj-$(CONFIG_DRM_GPUSVM) += drm_gpusvm_helper.o
+
 obj-$(CONFIG_DRM_BUDDY) += drm_buddy.o
 
 drm_dma_helper-y := drm_gem_dma_helper.o
@@ -133,9 +140,9 @@ obj-$(CONFIG_DRM_TTM_HELPER) += drm_ttm_helper.o
 drm_kms_helper-y := \
 	drm_atomic_helper.o \
 	drm_atomic_state_helper.o \
+	drm_bridge_helper.o \
 	drm_crtc_helper.o \
 	drm_damage_helper.o \
-	drm_encoder_slave.o \
 	drm_flip_work.o \
 	drm_format_helper.o \
 	drm_gem_atomic_helper.o \
@@ -145,7 +152,8 @@ drm_kms_helper-y := \
 	drm_plane_helper.o \
 	drm_probe_helper.o \
 	drm_self_refresh_helper.o \
-	drm_simple_kms_helper.o
+	drm_simple_kms_helper.o \
+	drm_vblank_helper.o
 drm_kms_helper-$(CONFIG_DRM_PANEL_BRIDGE) += bridge/panel.o
 drm_kms_helper-$(CONFIG_DRM_FBDEV_EMULATION) += drm_fb_helper.o
 obj-$(CONFIG_DRM_KMS_HELPER) += drm_kms_helper.o
@@ -176,6 +184,7 @@ obj-$(CONFIG_DRM_VMWGFX)+= vmwgfx/
 obj-$(CONFIG_DRM_VGEM)	+= vgem/
 obj-$(CONFIG_DRM_VKMS)	+= vkms/
 obj-$(CONFIG_DRM_NOUVEAU) +=nouveau/
+obj-$(CONFIG_DRM_NOVA) += nova/
 obj-$(CONFIG_DRM_EXYNOS) +=exynos/
 obj-$(CONFIG_DRM_ROCKCHIP) +=rockchip/
 obj-$(CONFIG_DRM_GMA500) += gma500/
@@ -198,28 +207,49 @@ obj-$(CONFIG_DRM_INGENIC) += ingenic/
 obj-$(CONFIG_DRM_LOGICVC) += logicvc/
 obj-$(CONFIG_DRM_MEDIATEK) += mediatek/
 obj-$(CONFIG_DRM_MESON)	+= meson/
-obj-y			+= i2c/
 obj-y			+= panel/
 obj-y			+= bridge/
 obj-$(CONFIG_DRM_FSL_DCU) += fsl-dcu/
 obj-$(CONFIG_DRM_ETNAVIV) += etnaviv/
 obj-y			+= hisilicon/
 obj-y			+= mxsfb/
+obj-y			+= sysfb/
 obj-y			+= tiny/
 obj-$(CONFIG_DRM_PL111) += pl111/
 obj-$(CONFIG_DRM_TVE200) += tve200/
+obj-$(CONFIG_DRM_ADP) += adp/
 obj-$(CONFIG_DRM_XEN) += xen/
 obj-$(CONFIG_DRM_VBOXVIDEO) += vboxvideo/
 obj-$(CONFIG_DRM_LIMA)  += lima/
 obj-$(CONFIG_DRM_PANFROST) += panfrost/
 obj-$(CONFIG_DRM_PANTHOR) += panthor/
+obj-$(CONFIG_DRM_TYR) += tyr/
 obj-$(CONFIG_DRM_ASPEED_GFX) += aspeed/
 obj-$(CONFIG_DRM_MCDE) += mcde/
 obj-$(CONFIG_DRM_TIDSS) += tidss/
 obj-y			+= xlnx/
 obj-y			+= gud/
 obj-$(CONFIG_DRM_HYPERV) += hyperv/
+obj-y			+= sitronix/
 obj-y			+= solomon/
 obj-$(CONFIG_DRM_SPRD) += sprd/
 obj-$(CONFIG_DRM_LOONGSON) += loongson/
 obj-$(CONFIG_DRM_POWERVR) += imagination/
+
+# Ensure drm headers are self-contained and pass kernel-doc
+hdrtest-files := \
+	$(shell cd $(src) && find . -maxdepth 1 -name 'drm_*.h') \
+	$(shell cd $(src) && find display lib -name '*.h')
+
+always-$(CONFIG_DRM_HEADER_TEST) += \
+	$(patsubst %.h,%.hdrtest, $(hdrtest-files))
+
+# Include the header twice to detect missing include guard.
+quiet_cmd_hdrtest = HDRTEST $(patsubst %.hdrtest,%.h,$@)
+      cmd_hdrtest = \
+		$(CC) $(c_flags) -fsyntax-only -x c /dev/null -include $< -include $<; \
+		 PYTHONDONTWRITEBYTECODE=1 $(PYTHON3) $(KERNELDOC) -none $(if $(CONFIG_WERROR)$(CONFIG_DRM_WERROR),-Werror) $<; \
+		touch $@
+
+$(obj)/%.hdrtest: $(src)/%.h FORCE
+	$(call if_changed_dep,hdrtest)
diff --git a/drivers/gpu/drm/adp/Kconfig b/drivers/gpu/drm/adp/Kconfig
new file mode 100644
index 000000000000..9fcc27eb200d
--- /dev/null
+++ b/drivers/gpu/drm/adp/Kconfig
@@ -0,0 +1,17 @@
+# SPDX-License-Identifier: GPL-2.0-only OR MIT
+config DRM_ADP
+	tristate "DRM Support for pre-DCP Apple display controllers"
+	depends on DRM && OF && ARM64
+	depends on ARCH_APPLE || COMPILE_TEST
+	select DRM_KMS_HELPER
+	select DRM_BRIDGE_CONNECTOR
+	select DRM_DISPLAY_HELPER
+	select DRM_KMS_DMA_HELPER
+	select DRM_GEM_DMA_HELPER
+	select DRM_PANEL_BRIDGE
+	select VIDEOMODE_HELPERS
+	select DRM_MIPI_DSI
+	help
+	  Chose this option if you have an Apple Arm laptop with a touchbar.
+
+	  If M is selected, this module will be called adpdrm.
diff --git a/drivers/gpu/drm/adp/Makefile b/drivers/gpu/drm/adp/Makefile
new file mode 100644
index 000000000000..8e7b618edd35
--- /dev/null
+++ b/drivers/gpu/drm/adp/Makefile
@@ -0,0 +1,5 @@
+# SPDX-License-Identifier: GPL-2.0-only OR MIT
+
+adpdrm-y := adp_drv.o
+adpdrm-mipi-y := adp-mipi.o
+obj-$(CONFIG_DRM_ADP) += adpdrm.o adpdrm-mipi.o
diff --git a/drivers/gpu/drm/adp/adp-mipi.c b/drivers/gpu/drm/adp/adp-mipi.c
new file mode 100644
index 000000000000..cba7d32150a9
--- /dev/null
+++ b/drivers/gpu/drm/adp/adp-mipi.c
@@ -0,0 +1,277 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include <linux/component.h>
+#include <linux/iopoll.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+
+#include <drm/drm_bridge.h>
+#include <drm/drm_mipi_dsi.h>
+
+#define DSI_GEN_HDR 0x6c
+#define DSI_GEN_PLD_DATA 0x70
+
+#define DSI_CMD_PKT_STATUS 0x74
+
+#define GEN_PLD_R_EMPTY BIT(4)
+#define GEN_PLD_W_FULL BIT(3)
+#define GEN_PLD_W_EMPTY BIT(2)
+#define GEN_CMD_FULL BIT(1)
+#define GEN_CMD_EMPTY BIT(0)
+#define GEN_RD_CMD_BUSY BIT(6)
+#define CMD_PKT_STATUS_TIMEOUT_US 20000
+
+struct adp_mipi_drv_private {
+	struct mipi_dsi_host dsi;
+	struct drm_bridge bridge;
+	struct drm_bridge *next_bridge;
+	void __iomem *mipi;
+};
+
+#define mipi_to_adp(x) container_of(x, struct adp_mipi_drv_private, dsi)
+
+static int adp_dsi_gen_pkt_hdr_write(struct adp_mipi_drv_private *adp, u32 hdr_val)
+{
+	int ret;
+	u32 val, mask;
+
+	ret = readl_poll_timeout(adp->mipi + DSI_CMD_PKT_STATUS,
+				 val, !(val & GEN_CMD_FULL), 1000,
+				 CMD_PKT_STATUS_TIMEOUT_US);
+	if (ret) {
+		dev_err(adp->dsi.dev, "failed to get available command FIFO\n");
+		return ret;
+	}
+
+	writel(hdr_val, adp->mipi + DSI_GEN_HDR);
+
+	mask = GEN_CMD_EMPTY | GEN_PLD_W_EMPTY;
+	ret = readl_poll_timeout(adp->mipi + DSI_CMD_PKT_STATUS,
+				 val, (val & mask) == mask,
+				 1000, CMD_PKT_STATUS_TIMEOUT_US);
+	if (ret) {
+		dev_err(adp->dsi.dev, "failed to write command FIFO\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static int adp_dsi_write(struct adp_mipi_drv_private *adp,
+			 const struct mipi_dsi_packet *packet)
+{
+	const u8 *tx_buf = packet->payload;
+	int len = packet->payload_length, pld_data_bytes = sizeof(u32), ret;
+	__le32 word;
+	u32 val;
+
+	while (len) {
+		if (len < pld_data_bytes) {
+			word = 0;
+			memcpy(&word, tx_buf, len);
+			writel(le32_to_cpu(word), adp->mipi + DSI_GEN_PLD_DATA);
+			len = 0;
+		} else {
+			memcpy(&word, tx_buf, pld_data_bytes);
+			writel(le32_to_cpu(word), adp->mipi + DSI_GEN_PLD_DATA);
+			tx_buf += pld_data_bytes;
+			len -= pld_data_bytes;
+		}
+
+		ret = readl_poll_timeout(adp->mipi + DSI_CMD_PKT_STATUS,
+					 val, !(val & GEN_PLD_W_FULL), 1000,
+					 CMD_PKT_STATUS_TIMEOUT_US);
+		if (ret) {
+			dev_err(adp->dsi.dev,
+				"failed to get available write payload FIFO\n");
+			return ret;
+		}
+	}
+
+	word = 0;
+	memcpy(&word, packet->header, sizeof(packet->header));
+	return adp_dsi_gen_pkt_hdr_write(adp, le32_to_cpu(word));
+}
+
+static int adp_dsi_read(struct adp_mipi_drv_private *adp,
+			const struct mipi_dsi_msg *msg)
+{
+	int i, j, ret, len = msg->rx_len;
+	u8 *buf = msg->rx_buf;
+	u32 val;
+
+	/* Wait end of the read operation */
+	ret = readl_poll_timeout(adp->mipi + DSI_CMD_PKT_STATUS,
+				 val, !(val & GEN_RD_CMD_BUSY),
+				 1000, CMD_PKT_STATUS_TIMEOUT_US);
+	if (ret) {
+		dev_err(adp->dsi.dev, "Timeout during read operation\n");
+		return ret;
+	}
+
+	for (i = 0; i < len; i += 4) {
+		/* Read fifo must not be empty before all bytes are read */
+		ret = readl_poll_timeout(adp->mipi + DSI_CMD_PKT_STATUS,
+					 val, !(val & GEN_PLD_R_EMPTY),
+					 1000, CMD_PKT_STATUS_TIMEOUT_US);
+		if (ret) {
+			dev_err(adp->dsi.dev, "Read payload FIFO is empty\n");
+			return ret;
+		}
+
+		val = readl(adp->mipi + DSI_GEN_PLD_DATA);
+		for (j = 0; j < 4 && j + i < len; j++)
+			buf[i + j] = val >> (8 * j);
+	}
+
+	return ret;
+}
+
+static ssize_t adp_dsi_host_transfer(struct mipi_dsi_host *host,
+				     const struct mipi_dsi_msg *msg)
+{
+	struct adp_mipi_drv_private *adp = mipi_to_adp(host);
+	struct mipi_dsi_packet packet;
+	int ret, nb_bytes;
+
+	ret = mipi_dsi_create_packet(&packet, msg);
+	if (ret) {
+		dev_err(adp->dsi.dev, "failed to create packet: %d\n", ret);
+		return ret;
+	}
+
+	ret = adp_dsi_write(adp, &packet);
+	if (ret)
+		return ret;
+
+	if (msg->rx_buf && msg->rx_len) {
+		ret = adp_dsi_read(adp, msg);
+		if (ret)
+			return ret;
+		nb_bytes = msg->rx_len;
+	} else {
+		nb_bytes = packet.size;
+	}
+
+	return nb_bytes;
+}
+
+static int adp_dsi_bind(struct device *dev, struct device *master, void *data)
+{
+	return 0;
+}
+
+static void adp_dsi_unbind(struct device *dev, struct device *master, void *data)
+{
+}
+
+static const struct component_ops adp_dsi_component_ops = {
+	.bind	= adp_dsi_bind,
+	.unbind	= adp_dsi_unbind,
+};
+
+static int adp_dsi_host_attach(struct mipi_dsi_host *host,
+			       struct mipi_dsi_device *dev)
+{
+	struct adp_mipi_drv_private *adp = mipi_to_adp(host);
+	struct drm_bridge *next;
+	int ret;
+
+	next = devm_drm_of_get_bridge(adp->dsi.dev, adp->dsi.dev->of_node, 1, 0);
+	if (IS_ERR(next))
+		return PTR_ERR(next);
+
+	adp->next_bridge = next;
+
+	drm_bridge_add(&adp->bridge);
+
+	ret = component_add(host->dev, &adp_dsi_component_ops);
+	if (ret) {
+		pr_err("failed to add dsi_host component: %d\n", ret);
+		drm_bridge_remove(&adp->bridge);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int adp_dsi_host_detach(struct mipi_dsi_host *host,
+			       struct mipi_dsi_device *dev)
+{
+	struct adp_mipi_drv_private *adp = mipi_to_adp(host);
+
+	component_del(host->dev, &adp_dsi_component_ops);
+	drm_bridge_remove(&adp->bridge);
+	return 0;
+}
+
+static const struct mipi_dsi_host_ops adp_dsi_host_ops = {
+	.transfer = adp_dsi_host_transfer,
+	.attach = adp_dsi_host_attach,
+	.detach = adp_dsi_host_detach,
+};
+
+static int adp_dsi_bridge_attach(struct drm_bridge *bridge,
+				 struct drm_encoder *encoder,
+				 enum drm_bridge_attach_flags flags)
+{
+	struct adp_mipi_drv_private *adp =
+		container_of(bridge, struct adp_mipi_drv_private, bridge);
+
+	return drm_bridge_attach(encoder, adp->next_bridge, bridge, flags);
+}
+
+static const struct drm_bridge_funcs adp_dsi_bridge_funcs = {
+	.attach	= adp_dsi_bridge_attach,
+};
+
+static int adp_mipi_probe(struct platform_device *pdev)
+{
+	struct adp_mipi_drv_private *adp;
+
+	adp = devm_drm_bridge_alloc(&pdev->dev, struct adp_mipi_drv_private,
+				    bridge, &adp_dsi_bridge_funcs);
+	if (IS_ERR(adp))
+		return PTR_ERR(adp);
+
+	adp->mipi = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(adp->mipi)) {
+		dev_err(&pdev->dev, "failed to map mipi mmio");
+		return PTR_ERR(adp->mipi);
+	}
+
+	adp->dsi.dev = &pdev->dev;
+	adp->dsi.ops = &adp_dsi_host_ops;
+	adp->bridge.of_node = pdev->dev.of_node;
+	adp->bridge.type = DRM_MODE_CONNECTOR_DSI;
+	dev_set_drvdata(&pdev->dev, adp);
+	return mipi_dsi_host_register(&adp->dsi);
+}
+
+static void adp_mipi_remove(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct adp_mipi_drv_private *adp = dev_get_drvdata(dev);
+
+	mipi_dsi_host_unregister(&adp->dsi);
+}
+
+static const struct of_device_id adp_mipi_of_match[] = {
+	{ .compatible = "apple,h7-display-pipe-mipi", },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, adp_mipi_of_match);
+
+static struct platform_driver adp_mipi_platform_driver = {
+	.driver = {
+		.name = "adp-mipi",
+		.of_match_table = adp_mipi_of_match,
+	},
+	.probe = adp_mipi_probe,
+	.remove = adp_mipi_remove,
+};
+
+module_platform_driver(adp_mipi_platform_driver);
+
+MODULE_DESCRIPTION("Apple Display Pipe MIPI driver");
+MODULE_LICENSE("GPL");
diff --git a/drivers/gpu/drm/adp/adp_drv.c b/drivers/gpu/drm/adp/adp_drv.c
new file mode 100644
index 000000000000..4554cf75565e
--- /dev/null
+++ b/drivers/gpu/drm/adp/adp_drv.c
@@ -0,0 +1,614 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include <linux/component.h>
+#include <linux/iopoll.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+
+#include <drm/drm_atomic.h>
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_bridge.h>
+#include <drm/drm_bridge_connector.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_fb_dma_helper.h>
+#include <drm/drm_framebuffer.h>
+#include <drm/drm_gem_atomic_helper.h>
+#include <drm/drm_gem_dma_helper.h>
+#include <drm/drm_gem_framebuffer_helper.h>
+#include <drm/drm_of.h>
+#include <drm/drm_print.h>
+#include <drm/drm_probe_helper.h>
+#include <drm/drm_vblank.h>
+
+#define ADP_INT_STATUS 0x34
+#define ADP_INT_STATUS_INT_MASK 0x7
+#define ADP_INT_STATUS_VBLANK 0x1
+#define ADP_CTRL 0x100
+#define ADP_CTRL_VBLANK_ON 0x12
+#define ADP_CTRL_FIFO_ON 0x601
+#define ADP_SCREEN_SIZE 0x0c
+#define ADP_SCREEN_HSIZE GENMASK(15, 0)
+#define ADP_SCREEN_VSIZE GENMASK(31, 16)
+
+#define ADBE_FIFO 0x10c0
+#define ADBE_FIFO_SYNC 0xc0000000
+
+#define ADBE_BLEND_BYPASS 0x2020
+#define ADBE_BLEND_EN1 0x2028
+#define ADBE_BLEND_EN2 0x2074
+#define ADBE_BLEND_EN3 0x202c
+#define ADBE_BLEND_EN4 0x2034
+#define ADBE_MASK_BUF 0x2200
+
+#define ADBE_SRC_START 0x4040
+#define ADBE_SRC_SIZE 0x4048
+#define ADBE_DST_START 0x4050
+#define ADBE_DST_SIZE 0x4054
+#define ADBE_STRIDE 0x4038
+#define ADBE_FB_BASE 0x4030
+
+#define ADBE_LAYER_EN1 0x4020
+#define ADBE_LAYER_EN2 0x4068
+#define ADBE_LAYER_EN3 0x40b4
+#define ADBE_LAYER_EN4 0x40f4
+#define ADBE_SCALE_CTL 0x40ac
+#define ADBE_SCALE_CTL_BYPASS 0x100000
+
+#define ADBE_LAYER_CTL 0x1038
+#define ADBE_LAYER_CTL_ENABLE 0x10000
+
+#define ADBE_PIX_FMT 0x402c
+#define ADBE_PIX_FMT_XRGB32 0x53e4001
+
+static int adp_open(struct inode *inode, struct file *filp)
+{
+	/*
+	 * The modesetting driver does not check the non-desktop connector
+	 * property and keeps the device open and locked. If the touchbar daemon
+	 * opens the device first, modesetting breaks the whole X session.
+	 * Simply refuse to open the device for X11 server processes as
+	 * workaround.
+	 */
+	if (current->comm[0] == 'X')
+		return -EBUSY;
+
+	return drm_open(inode, filp);
+}
+
+static const struct file_operations adp_fops = {
+	.owner          = THIS_MODULE,
+	.open           = adp_open,
+	.release        = drm_release,
+	.unlocked_ioctl = drm_ioctl,
+	.compat_ioctl   = drm_compat_ioctl,
+	.poll           = drm_poll,
+	.read           = drm_read,
+	.llseek         = noop_llseek,
+	.mmap           = drm_gem_mmap,
+	.fop_flags      = FOP_UNSIGNED_OFFSET,
+	DRM_GEM_DMA_UNMAPPED_AREA_FOPS
+};
+
+static int adp_drm_gem_dumb_create(struct drm_file *file_priv,
+					struct drm_device *drm,
+					struct drm_mode_create_dumb *args)
+{
+	args->height = ALIGN(args->height, 64);
+	args->size = args->pitch * args->height;
+
+	return drm_gem_dma_dumb_create_internal(file_priv, drm, args);
+}
+
+static const struct drm_driver adp_driver = {
+	.driver_features = DRIVER_GEM | DRIVER_MODESET | DRIVER_ATOMIC,
+	.fops = &adp_fops,
+	DRM_GEM_DMA_DRIVER_OPS_VMAP_WITH_DUMB_CREATE(adp_drm_gem_dumb_create),
+	.name = "adp",
+	.desc = "Apple Display Pipe DRM Driver",
+	.major = 0,
+	.minor = 1,
+};
+
+struct adp_drv_private {
+	struct drm_device drm;
+	struct drm_crtc crtc;
+	struct drm_encoder *encoder;
+	struct drm_connector *connector;
+	struct drm_bridge *next_bridge;
+	void __iomem *be;
+	void __iomem *fe;
+	u32 *mask_buf;
+	u64 mask_buf_size;
+	dma_addr_t mask_iova;
+	int be_irq;
+	int fe_irq;
+	struct drm_pending_vblank_event *event;
+};
+
+#define to_adp(x) container_of(x, struct adp_drv_private, drm)
+#define crtc_to_adp(x) container_of(x, struct adp_drv_private, crtc)
+
+static int adp_plane_atomic_check(struct drm_plane *plane,
+				    struct drm_atomic_state *state)
+{
+	struct drm_plane_state *new_plane_state;
+	struct drm_crtc_state *crtc_state;
+
+	new_plane_state = drm_atomic_get_new_plane_state(state, plane);
+
+	if (!new_plane_state->crtc)
+		return 0;
+
+	crtc_state = drm_atomic_get_crtc_state(state, new_plane_state->crtc);
+	if (IS_ERR(crtc_state))
+		return PTR_ERR(crtc_state);
+
+	return drm_atomic_helper_check_plane_state(new_plane_state,
+						   crtc_state,
+						   DRM_PLANE_NO_SCALING,
+						   DRM_PLANE_NO_SCALING,
+						   true, true);
+}
+
+static void adp_plane_atomic_update(struct drm_plane *plane,
+				    struct drm_atomic_state *state)
+{
+	struct adp_drv_private *adp;
+	struct drm_rect src_rect;
+	struct drm_gem_dma_object *obj;
+	struct drm_framebuffer *fb;
+	struct drm_plane_state *new_state = drm_atomic_get_new_plane_state(state, plane);
+	u32 src_pos, src_size, dst_pos, dst_size;
+
+	if (!plane || !new_state)
+		return;
+
+	fb = new_state->fb;
+	if (!fb)
+		return;
+	adp = to_adp(plane->dev);
+
+	drm_rect_fp_to_int(&src_rect, &new_state->src);
+	src_pos = src_rect.x1 << 16 | src_rect.y1;
+	dst_pos = new_state->dst.x1 << 16 | new_state->dst.y1;
+	src_size = drm_rect_width(&src_rect) << 16 | drm_rect_height(&src_rect);
+	dst_size = drm_rect_width(&new_state->dst) << 16 |
+		drm_rect_height(&new_state->dst);
+	writel(src_pos, adp->be + ADBE_SRC_START);
+	writel(src_size, adp->be + ADBE_SRC_SIZE);
+	writel(dst_pos, adp->be + ADBE_DST_START);
+	writel(dst_size, adp->be + ADBE_DST_SIZE);
+	writel(fb->pitches[0], adp->be + ADBE_STRIDE);
+	obj = drm_fb_dma_get_gem_obj(fb, 0);
+	if (obj)
+		writel(obj->dma_addr + fb->offsets[0], adp->be + ADBE_FB_BASE);
+
+	writel(BIT(0), adp->be + ADBE_LAYER_EN1);
+	writel(BIT(0), adp->be + ADBE_LAYER_EN2);
+	writel(BIT(0), adp->be + ADBE_LAYER_EN3);
+	writel(BIT(0), adp->be + ADBE_LAYER_EN4);
+	writel(ADBE_SCALE_CTL_BYPASS, adp->be + ADBE_SCALE_CTL);
+	writel(ADBE_LAYER_CTL_ENABLE | BIT(0), adp->be + ADBE_LAYER_CTL);
+	writel(ADBE_PIX_FMT_XRGB32, adp->be + ADBE_PIX_FMT);
+}
+
+static void adp_plane_atomic_disable(struct drm_plane *plane,
+				     struct drm_atomic_state *state)
+{
+	struct adp_drv_private *adp = to_adp(plane->dev);
+
+	writel(0x0, adp->be + ADBE_LAYER_EN1);
+	writel(0x0, adp->be + ADBE_LAYER_EN2);
+	writel(0x0, adp->be + ADBE_LAYER_EN3);
+	writel(0x0, adp->be + ADBE_LAYER_EN4);
+	writel(ADBE_LAYER_CTL_ENABLE, adp->be + ADBE_LAYER_CTL);
+}
+
+static const struct drm_plane_helper_funcs adp_plane_helper_funcs = {
+	.atomic_check = adp_plane_atomic_check,
+	.atomic_update = adp_plane_atomic_update,
+	.atomic_disable = adp_plane_atomic_disable,
+	DRM_GEM_SHADOW_PLANE_HELPER_FUNCS
+};
+
+static const struct drm_plane_funcs adp_plane_funcs = {
+	.update_plane = drm_atomic_helper_update_plane,
+	.disable_plane = drm_atomic_helper_disable_plane,
+	DRM_GEM_SHADOW_PLANE_FUNCS
+};
+
+static const u32 plane_formats[] = {
+	DRM_FORMAT_XRGB8888,
+};
+
+#define ALL_CRTCS 1
+
+static struct drm_plane *adp_plane_new(struct adp_drv_private *adp)
+{
+	struct drm_device *drm = &adp->drm;
+	struct drm_plane *plane;
+
+	plane = __drmm_universal_plane_alloc(drm, sizeof(struct drm_plane), 0,
+					     ALL_CRTCS, &adp_plane_funcs,
+					     plane_formats, ARRAY_SIZE(plane_formats),
+					     NULL, DRM_PLANE_TYPE_PRIMARY, "plane");
+	if (IS_ERR(plane)) {
+		drm_err(drm, "failed to allocate plane");
+		return plane;
+	}
+
+	drm_plane_helper_add(plane, &adp_plane_helper_funcs);
+	return plane;
+}
+
+static void adp_enable_vblank(struct adp_drv_private *adp)
+{
+	u32 cur_ctrl;
+
+	writel(ADP_INT_STATUS_INT_MASK, adp->fe + ADP_INT_STATUS);
+
+	cur_ctrl = readl(adp->fe + ADP_CTRL);
+	writel(cur_ctrl | ADP_CTRL_VBLANK_ON, adp->fe + ADP_CTRL);
+}
+
+static int adp_crtc_enable_vblank(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct adp_drv_private *adp = to_adp(dev);
+
+	adp_enable_vblank(adp);
+
+	return 0;
+}
+
+static void adp_disable_vblank(struct adp_drv_private *adp)
+{
+	u32 cur_ctrl;
+
+	cur_ctrl = readl(adp->fe + ADP_CTRL);
+	writel(cur_ctrl & ~ADP_CTRL_VBLANK_ON, adp->fe + ADP_CTRL);
+	writel(ADP_INT_STATUS_INT_MASK, adp->fe + ADP_INT_STATUS);
+}
+
+static void adp_crtc_disable_vblank(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct adp_drv_private *adp = to_adp(dev);
+
+	adp_disable_vblank(adp);
+}
+
+static void adp_crtc_atomic_enable(struct drm_crtc *crtc,
+				   struct drm_atomic_state *state)
+{
+	struct adp_drv_private *adp = crtc_to_adp(crtc);
+
+	writel(BIT(0), adp->be + ADBE_BLEND_EN2);
+	writel(BIT(4), adp->be + ADBE_BLEND_EN1);
+	writel(BIT(0), adp->be + ADBE_BLEND_EN3);
+	writel(BIT(0), adp->be + ADBE_BLEND_BYPASS);
+	writel(BIT(0), adp->be + ADBE_BLEND_EN4);
+	drm_crtc_vblank_on(crtc);
+}
+
+static void adp_crtc_atomic_disable(struct drm_crtc *crtc,
+				    struct drm_atomic_state *state)
+{
+	struct adp_drv_private *adp = crtc_to_adp(crtc);
+	struct drm_crtc_state *old_state = drm_atomic_get_old_crtc_state(state, crtc);
+
+	drm_atomic_helper_disable_planes_on_crtc(old_state, false);
+
+	writel(0x0, adp->be + ADBE_BLEND_EN2);
+	writel(0x0, adp->be + ADBE_BLEND_EN1);
+	writel(0x0, adp->be + ADBE_BLEND_EN3);
+	writel(0x0, adp->be + ADBE_BLEND_BYPASS);
+	writel(0x0, adp->be + ADBE_BLEND_EN4);
+	drm_crtc_vblank_off(crtc);
+}
+
+static void adp_crtc_atomic_flush(struct drm_crtc *crtc,
+				  struct drm_atomic_state *state)
+{
+	u32 frame_num = 1;
+	unsigned long flags;
+	struct adp_drv_private *adp = crtc_to_adp(crtc);
+	struct drm_crtc_state *new_state = drm_atomic_get_new_crtc_state(state, crtc);
+	u64 new_size = ALIGN(new_state->mode.hdisplay *
+			     new_state->mode.vdisplay * 4, PAGE_SIZE);
+
+	if (new_size != adp->mask_buf_size) {
+		if (adp->mask_buf)
+			dma_free_coherent(crtc->dev->dev, adp->mask_buf_size,
+					  adp->mask_buf, adp->mask_iova);
+		adp->mask_buf = NULL;
+		if (new_size != 0) {
+			adp->mask_buf = dma_alloc_coherent(crtc->dev->dev, new_size,
+							   &adp->mask_iova, GFP_KERNEL);
+			memset(adp->mask_buf, 0xFF, new_size);
+			writel(adp->mask_iova, adp->be + ADBE_MASK_BUF);
+		}
+		adp->mask_buf_size = new_size;
+	}
+	writel(ADBE_FIFO_SYNC | frame_num, adp->be + ADBE_FIFO);
+	//FIXME: use adbe flush interrupt
+	if (crtc->state->event) {
+		struct drm_pending_vblank_event *event = crtc->state->event;
+
+		crtc->state->event = NULL;
+		spin_lock_irqsave(&crtc->dev->event_lock, flags);
+
+		if (drm_crtc_vblank_get(crtc) != 0)
+			drm_crtc_send_vblank_event(crtc, event);
+		else
+			adp->event = event;
+
+		spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+	}
+}
+
+static const struct drm_crtc_funcs adp_crtc_funcs = {
+	.destroy = drm_crtc_cleanup,
+	.set_config = drm_atomic_helper_set_config,
+	.page_flip = drm_atomic_helper_page_flip,
+	.reset = drm_atomic_helper_crtc_reset,
+	.atomic_duplicate_state = drm_atomic_helper_crtc_duplicate_state,
+	.atomic_destroy_state = drm_atomic_helper_crtc_destroy_state,
+	.enable_vblank = adp_crtc_enable_vblank,
+	.disable_vblank = adp_crtc_disable_vblank,
+};
+
+
+static const struct drm_crtc_helper_funcs adp_crtc_helper_funcs = {
+	.atomic_enable = adp_crtc_atomic_enable,
+	.atomic_disable = adp_crtc_atomic_disable,
+	.atomic_flush = adp_crtc_atomic_flush,
+};
+
+static int adp_setup_crtc(struct adp_drv_private *adp)
+{
+	struct drm_device *drm = &adp->drm;
+	struct drm_plane *primary;
+	int ret;
+
+	primary = adp_plane_new(adp);
+	if (IS_ERR(primary))
+		return PTR_ERR(primary);
+
+	ret = drm_crtc_init_with_planes(drm, &adp->crtc, primary,
+					NULL, &adp_crtc_funcs, NULL);
+	if (ret)
+		return ret;
+
+	drm_crtc_helper_add(&adp->crtc, &adp_crtc_helper_funcs);
+	return 0;
+}
+
+static const struct drm_mode_config_funcs adp_mode_config_funcs = {
+	.fb_create = drm_gem_fb_create_with_dirty,
+	.atomic_check = drm_atomic_helper_check,
+	.atomic_commit = drm_atomic_helper_commit,
+};
+
+static int adp_setup_mode_config(struct adp_drv_private *adp)
+{
+	struct drm_device *drm = &adp->drm;
+	int ret;
+	u32 size;
+
+	ret = drmm_mode_config_init(drm);
+	if (ret)
+		return ret;
+
+	/*
+	 * Query screen size restrict the frame buffer size to the screen size
+	 * aligned to the next multiple of 64. This is not necessary but can be
+	 * used as simple check for non-desktop devices.
+	 * Xorg's modesetting driver does not care about the connector
+	 * "non-desktop" property. The max frame buffer width or height can be
+	 * easily checked and a device can be reject if the max width/height is
+	 * smaller than 120 for example.
+	 * Any touchbar daemon is not limited by this small framebuffer size.
+	 */
+	size = readl(adp->fe + ADP_SCREEN_SIZE);
+
+	drm->mode_config.min_width = 32;
+	drm->mode_config.min_height = 32;
+	drm->mode_config.max_width = ALIGN(FIELD_GET(ADP_SCREEN_HSIZE, size), 64);
+	drm->mode_config.max_height = ALIGN(FIELD_GET(ADP_SCREEN_VSIZE, size), 64);
+	drm->mode_config.preferred_depth = 24;
+	drm->mode_config.prefer_shadow = 0;
+	drm->mode_config.funcs = &adp_mode_config_funcs;
+
+	ret = adp_setup_crtc(adp);
+	if (ret) {
+		drm_err(drm, "failed to create crtc");
+		return ret;
+	}
+
+	adp->encoder = drmm_plain_encoder_alloc(drm, NULL, DRM_MODE_ENCODER_DSI, NULL);
+	if (IS_ERR(adp->encoder)) {
+		drm_err(drm, "failed to init encoder");
+		return PTR_ERR(adp->encoder);
+	}
+	adp->encoder->possible_crtcs = ALL_CRTCS;
+
+	ret = drm_bridge_attach(adp->encoder, adp->next_bridge, NULL,
+				DRM_BRIDGE_ATTACH_NO_CONNECTOR);
+	if (ret) {
+		drm_err(drm, "failed to init bridge chain");
+		return ret;
+	}
+
+	adp->connector = drm_bridge_connector_init(drm, adp->encoder);
+	if (IS_ERR(adp->connector))
+		return PTR_ERR(adp->connector);
+
+	drm_connector_attach_encoder(adp->connector, adp->encoder);
+
+	ret = drm_vblank_init(drm, drm->mode_config.num_crtc);
+	if (ret < 0) {
+		drm_err(drm, "failed to initialize vblank");
+		return ret;
+	}
+
+	drm_mode_config_reset(drm);
+
+	return 0;
+}
+
+static int adp_parse_of(struct platform_device *pdev, struct adp_drv_private *adp)
+{
+	struct device *dev = &pdev->dev;
+
+	adp->be = devm_platform_ioremap_resource_byname(pdev, "be");
+	if (IS_ERR(adp->be)) {
+		dev_err(dev, "failed to map display backend mmio");
+		return PTR_ERR(adp->be);
+	}
+
+	adp->fe = devm_platform_ioremap_resource_byname(pdev, "fe");
+	if (IS_ERR(adp->fe)) {
+		dev_err(dev, "failed to map display pipe mmio");
+		return PTR_ERR(adp->fe);
+	}
+
+	adp->be_irq = platform_get_irq_byname(pdev, "be");
+	if (adp->be_irq < 0)
+		return adp->be_irq;
+
+	adp->fe_irq = platform_get_irq_byname(pdev, "fe");
+	if (adp->fe_irq < 0)
+		return adp->fe_irq;
+
+	return 0;
+}
+
+static irqreturn_t adp_fe_irq(int irq, void *arg)
+{
+	struct adp_drv_private *adp = (struct adp_drv_private *)arg;
+	u32 int_status;
+	u32 int_ctl;
+
+	int_status = readl(adp->fe + ADP_INT_STATUS);
+	if (int_status & ADP_INT_STATUS_VBLANK) {
+		drm_crtc_handle_vblank(&adp->crtc);
+		spin_lock(&adp->crtc.dev->event_lock);
+		if (adp->event) {
+			int_ctl = readl(adp->fe + ADP_CTRL);
+			if ((int_ctl & 0xF00) == 0x600) {
+				drm_crtc_send_vblank_event(&adp->crtc, adp->event);
+				adp->event = NULL;
+				drm_crtc_vblank_put(&adp->crtc);
+			}
+		}
+		spin_unlock(&adp->crtc.dev->event_lock);
+	}
+
+	writel(int_status, adp->fe + ADP_INT_STATUS);
+
+
+	return IRQ_HANDLED;
+}
+
+static int adp_drm_bind(struct device *dev)
+{
+	struct drm_device *drm = dev_get_drvdata(dev);
+	struct adp_drv_private *adp = to_adp(drm);
+	int err;
+
+	writel(ADP_CTRL_FIFO_ON, adp->fe + ADP_CTRL);
+
+	adp->next_bridge = drmm_of_get_bridge(&adp->drm, dev->of_node, 0, 0);
+	if (IS_ERR(adp->next_bridge)) {
+		dev_err(dev, "failed to find next bridge");
+		return PTR_ERR(adp->next_bridge);
+	}
+
+	err = adp_setup_mode_config(adp);
+	if (err < 0)
+		return err;
+
+	err = request_irq(adp->fe_irq, adp_fe_irq, 0, "adp-fe", adp);
+	if (err)
+		return err;
+
+	err = drm_dev_register(&adp->drm, 0);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static void adp_drm_unbind(struct device *dev)
+{
+	struct drm_device *drm = dev_get_drvdata(dev);
+	struct adp_drv_private *adp = to_adp(drm);
+
+	drm_dev_unregister(drm);
+	drm_atomic_helper_shutdown(drm);
+	free_irq(adp->fe_irq, adp);
+}
+
+static const struct component_master_ops adp_master_ops = {
+	.bind	= adp_drm_bind,
+	.unbind = adp_drm_unbind,
+};
+
+static int compare_dev(struct device *dev, void *data)
+{
+	return dev->of_node == data;
+}
+
+static int adp_probe(struct platform_device *pdev)
+{
+	struct device_node *port;
+	struct component_match *match = NULL;
+	struct adp_drv_private *adp;
+	int err;
+
+	adp = devm_drm_dev_alloc(&pdev->dev, &adp_driver, struct adp_drv_private, drm);
+	if (IS_ERR(adp))
+		return PTR_ERR(adp);
+
+	dev_set_drvdata(&pdev->dev, &adp->drm);
+
+	err = adp_parse_of(pdev, adp);
+	if (err < 0)
+		return err;
+
+	port = of_graph_get_remote_node(pdev->dev.of_node, 0, 0);
+	if (!port)
+		return -ENODEV;
+
+	drm_of_component_match_add(&pdev->dev, &match, compare_dev, port);
+	of_node_put(port);
+
+	return component_master_add_with_match(&pdev->dev, &adp_master_ops, match);
+}
+
+static void adp_remove(struct platform_device *pdev)
+{
+	component_master_del(&pdev->dev, &adp_master_ops);
+	dev_set_drvdata(&pdev->dev, NULL);
+}
+
+static const struct of_device_id adp_of_match[] = {
+	{ .compatible = "apple,h7-display-pipe", },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, adp_of_match);
+
+static struct platform_driver adp_platform_driver = {
+	.driver = {
+		.name = "adp",
+		.of_match_table = adp_of_match,
+	},
+	.probe = adp_probe,
+	.remove = adp_remove,
+};
+
+module_platform_driver(adp_platform_driver);
+
+MODULE_DESCRIPTION("Apple Display Pipe DRM driver");
+MODULE_LICENSE("GPL");
diff --git a/drivers/gpu/drm/amd/amdgpu/Kconfig b/drivers/gpu/drm/amd/amdgpu/Kconfig
index 1a11cab741ac..7f515be5185d 100644
--- a/drivers/gpu/drm/amd/amdgpu/Kconfig
+++ b/drivers/gpu/drm/amd/amdgpu/Kconfig
@@ -2,7 +2,7 @@
 
 config DRM_AMDGPU
 	tristate "AMD GPU"
-	depends on DRM && PCI && MMU
+	depends on DRM && PCI
 	depends on !UML
 	select FW_LOADER
 	select DRM_CLIENT
@@ -43,14 +43,16 @@ config DRM_AMDGPU_SI
 	bool "Enable amdgpu support for SI parts"
 	depends on DRM_AMDGPU
 	help
-	  Choose this option if you want to enable experimental support
+	  Choose this option if you want to enable support
 	  for SI (Southern Islands) asics.
 
-	  SI is already supported in radeon. Experimental support for SI
-	  in amdgpu will be disabled by default and is still provided by
-	  radeon. Use module options to override this:
+	  SI (Southern Islands) are first generation GCN GPUs,
+	  supported by both drivers: radeon (old) and amdgpu (new).
+	  By default, SI dedicated GPUs are supported by amdgpu.
 
-	  radeon.si_support=0 amdgpu.si_support=1
+	  Use module options to override this:
+	  To use radeon for SI,
+	  radeon.si_support=1 amdgpu.si_support=0
 
 config DRM_AMDGPU_CIK
 	bool "Enable amdgpu support for CIK parts"
@@ -59,16 +61,21 @@ config DRM_AMDGPU_CIK
 	  Choose this option if you want to enable support for CIK (Sea
 	  Islands) asics.
 
-	  CIK is already supported in radeon. Support for CIK in amdgpu
-	  will be disabled by default and is still provided by radeon.
-	  Use module options to override this:
+	  CIK (Sea Islands) are second generation GCN GPUs,
+	  supported by both drivers: radeon (old) and amdgpu (new).
+	  By default,
+	  CIK dedicated GPUs are supported by amdgpu
+	  CIK APUs are supported by radeon
 
+	  Use module options to override this:
+	  To use amdgpu for CIK,
 	  radeon.cik_support=0 amdgpu.cik_support=1
+	  To use radeon for CIK,
+	  radeon.cik_support=1 amdgpu.cik_support=0
 
 config DRM_AMDGPU_USERPTR
 	bool "Always enable userptr write support"
 	depends on DRM_AMDGPU
-	depends on MMU
 	select HMM_MIRROR
 	select MMU_NOTIFIER
 	help
@@ -77,7 +84,7 @@ config DRM_AMDGPU_USERPTR
 
 config DRM_AMD_ISP
 	bool "Enable AMD Image Signal Processor IP support"
-	depends on DRM_AMDGPU
+	depends on DRM_AMDGPU && ACPI
 	select MFD_CORE
 	select PM_GENERIC_DOMAINS if PM
 	help
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile
index 5b21674b07fb..c88760fb52ea 100644
--- a/drivers/gpu/drm/amd/amdgpu/Makefile
+++ b/drivers/gpu/drm/amd/amdgpu/Makefile
@@ -37,7 +37,8 @@ ccflags-y := -I$(FULL_AMD_PATH)/include/asic_reg \
 	-I$(FULL_AMD_DISPLAY_PATH)/modules/inc \
 	-I$(FULL_AMD_DISPLAY_PATH)/dc \
 	-I$(FULL_AMD_DISPLAY_PATH)/amdgpu_dm \
-	-I$(FULL_AMD_PATH)/amdkfd
+	-I$(FULL_AMD_PATH)/amdkfd \
+	-I$(FULL_AMD_PATH)/ras/ras_mgr
 
 # Locally disable W=1 warnings enabled in drm subsystem Makefile
 subdir-ccflags-y += -Wno-override-init
@@ -65,7 +66,8 @@ amdgpu-y += amdgpu_device.o amdgpu_doorbell_mgr.o amdgpu_kms.o \
 	amdgpu_umc.o smu_v11_0_i2c.o amdgpu_fru_eeprom.o amdgpu_rap.o \
 	amdgpu_fw_attestation.o amdgpu_securedisplay.o \
 	amdgpu_eeprom.o amdgpu_mca.o amdgpu_psp_ta.o amdgpu_lsdma.o \
-	amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o amdgpu_dev_coredump.o
+	amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o amdgpu_dev_coredump.o \
+	amdgpu_cper.o amdgpu_userq_fence.o amdgpu_eviction_fence.o amdgpu_ip.o
 
 amdgpu-$(CONFIG_PROC_FS) += amdgpu_fdinfo.o
 
@@ -76,14 +78,15 @@ amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o \
 	dce_v8_0.o gfx_v7_0.o cik_sdma.o uvd_v4_2.o vce_v2_0.o
 
 amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce_v6_0.o \
-	uvd_v3_1.o
+	uvd_v3_1.o vce_v1_0.o
 
 amdgpu-y += \
 	vi.o mxgpu_vi.o nbio_v6_1.o soc15.o emu_soc.o mxgpu_ai.o nbio_v7_0.o vega10_reg_init.o \
 	vega20_reg_init.o nbio_v7_4.o nbio_v2_3.o nv.o arct_reg_init.o mxgpu_nv.o \
 	nbio_v7_2.o hdp_v4_0.o hdp_v5_0.o aldebaran_reg_init.o aldebaran.o soc21.o soc24.o \
 	sienna_cichlid.o smu_v13_0_10.o nbio_v4_3.o hdp_v6_0.o nbio_v7_7.o hdp_v5_2.o lsdma_v6_0.o \
-	nbio_v7_9.o aqua_vanjaram.o nbio_v7_11.o lsdma_v7_0.o hdp_v7_0.o nbif_v6_3_1.o
+	nbio_v7_9.o aqua_vanjaram.o nbio_v7_11.o lsdma_v7_0.o hdp_v7_0.o nbif_v6_3_1.o \
+	cyan_skillfish_reg_init.o
 
 # add DF block
 amdgpu-y += \
@@ -136,7 +139,6 @@ amdgpu-y += \
 # add DCE block
 amdgpu-y += \
 	dce_v10_0.o \
-	dce_v11_0.o \
 	amdgpu_vkms.o
 
 # add GFX block
@@ -173,7 +175,10 @@ amdgpu-y += \
 amdgpu-y += \
 	amdgpu_mes.o \
 	mes_v11_0.o \
-	mes_v12_0.o
+	mes_v12_0.o \
+
+# add GFX userqueue support
+amdgpu-y += mes_userqueue.o
 
 # add UVD block
 amdgpu-y += \
@@ -252,6 +257,8 @@ amdgpu-y += \
 # add amdkfd interfaces
 amdgpu-y += amdgpu_amdkfd.o
 
+# add gfx usermode queue
+amdgpu-y += amdgpu_userq.o
 
 ifneq ($(CONFIG_HSA_AMD),)
 AMDKFD_PATH := ../amdkfd
@@ -318,4 +325,9 @@ amdgpu-y += \
 	isp_v4_1_1.o
 endif
 
+AMD_GPU_RAS_PATH := ../ras
+AMD_GPU_RAS_FULL_PATH := $(FULL_AMD_PATH)/ras
+include $(AMD_GPU_RAS_FULL_PATH)/Makefile
+amdgpu-y += $(AMD_GPU_RAS_FILES)
+
 obj-$(CONFIG_DRM_AMDGPU)+= amdgpu.o
diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran.c b/drivers/gpu/drm/amd/amdgpu/aldebaran.c
index e13fbd974141..daa7b23bc775 100644
--- a/drivers/gpu/drm/amd/amdgpu/aldebaran.c
+++ b/drivers/gpu/drm/amd/amdgpu/aldebaran.c
@@ -71,18 +71,33 @@ aldebaran_get_reset_handler(struct amdgpu_reset_control *reset_ctl,
 	return NULL;
 }
 
+static inline uint32_t aldebaran_get_ip_block_mask(struct amdgpu_device *adev)
+{
+	uint32_t ip_block_mask = BIT(AMD_IP_BLOCK_TYPE_GFX) |
+				 BIT(AMD_IP_BLOCK_TYPE_SDMA);
+
+	if (adev->aid_mask)
+		ip_block_mask |= BIT(AMD_IP_BLOCK_TYPE_IH);
+
+	return ip_block_mask;
+}
+
 static int aldebaran_mode2_suspend_ip(struct amdgpu_device *adev)
 {
+	uint32_t ip_block_mask = aldebaran_get_ip_block_mask(adev);
+	uint32_t ip_block;
 	int r, i;
 
+	/* Skip suspend of SDMA IP versions >= 4.4.2. They are multi-aid */
+	if (adev->aid_mask)
+		ip_block_mask &= ~BIT(AMD_IP_BLOCK_TYPE_SDMA);
+
 	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
 	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
 
 	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
-		if (!(adev->ip_blocks[i].version->type ==
-			      AMD_IP_BLOCK_TYPE_GFX ||
-		      adev->ip_blocks[i].version->type ==
-			      AMD_IP_BLOCK_TYPE_SDMA))
+		ip_block = BIT(adev->ip_blocks[i].version->type);
+		if (!(ip_block_mask & ip_block))
 			continue;
 
 		r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
@@ -200,8 +215,10 @@ aldebaran_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl,
 static int aldebaran_mode2_restore_ip(struct amdgpu_device *adev)
 {
 	struct amdgpu_firmware_info *ucode_list[AMDGPU_UCODE_ID_MAXIMUM];
+	uint32_t ip_block_mask = aldebaran_get_ip_block_mask(adev);
 	struct amdgpu_firmware_info *ucode;
 	struct amdgpu_ip_block *cmn_block;
+	struct amdgpu_ip_block *ih_block;
 	int ucode_count = 0;
 	int i, r;
 
@@ -243,6 +260,18 @@ static int aldebaran_mode2_restore_ip(struct amdgpu_device *adev)
 	if (r)
 		return r;
 
+	if (ip_block_mask & BIT(AMD_IP_BLOCK_TYPE_IH)) {
+		ih_block = amdgpu_device_ip_get_ip_block(adev,
+							 AMD_IP_BLOCK_TYPE_IH);
+		if (unlikely(!ih_block)) {
+			dev_err(adev->dev, "Failed to get IH handle\n");
+			return -EINVAL;
+		}
+		r = amdgpu_ip_block_resume(ih_block);
+		if (r)
+			return r;
+	}
+
 	/* Reinit GFXHUB */
 	adev->gfxhub.funcs->init(adev);
 	r = adev->gfxhub.funcs->gart_enable(adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 69895fccb474..9f9774f58ce1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -63,6 +63,7 @@
 #include "kgd_pp_interface.h"
 
 #include "amd_shared.h"
+#include "amdgpu_utils.h"
 #include "amdgpu_mode.h"
 #include "amdgpu_ih.h"
 #include "amdgpu_irq.h"
@@ -109,9 +110,12 @@
 #include "amdgpu_mca.h"
 #include "amdgpu_aca.h"
 #include "amdgpu_ras.h"
+#include "amdgpu_cper.h"
 #include "amdgpu_xcp.h"
 #include "amdgpu_seq64.h"
 #include "amdgpu_reg_state.h"
+#include "amdgpu_userq.h"
+#include "amdgpu_eviction_fence.h"
 #if defined(CONFIG_DRM_AMD_ISP)
 #include "amdgpu_isp.h"
 #endif
@@ -227,7 +231,7 @@ extern int amdgpu_force_asic_type;
 extern int amdgpu_smartshift_bias;
 extern int amdgpu_use_xgmi_p2p;
 extern int amdgpu_mtype_local;
-extern bool enforce_isolation;
+extern int amdgpu_enforce_isolation;
 #ifdef CONFIG_HSA_AMD
 extern int sched_policy;
 extern bool debug_evictions;
@@ -265,8 +269,10 @@ extern int amdgpu_umsch_mm_fwlog;
 
 extern int amdgpu_user_partt_mode;
 extern int amdgpu_agp;
+extern int amdgpu_rebar;
 
 extern int amdgpu_wbrf;
+extern int amdgpu_user_queue;
 
 #define AMDGPU_VM_MAX_NUM_CTX			4096
 #define AMDGPU_SG_THRESHOLD			(256*1024*1024)
@@ -352,7 +358,6 @@ enum amdgpu_kiq_irq {
 	AMDGPU_CP_KIQ_IRQ_DRIVER0 = 0,
 	AMDGPU_CP_KIQ_IRQ_LAST
 };
-#define SRIOV_USEC_TIMEOUT  1200000 /* wait 12 * 100ms for SRIOV */
 #define MAX_KIQ_REG_WAIT       5000 /* in usecs, 5ms */
 #define MAX_KIQ_REG_BAILOUT_INTERVAL   5 /* in msecs, 5ms */
 #define MAX_KIQ_REG_TRY 1000
@@ -367,13 +372,15 @@ void amdgpu_device_ip_get_clockgating_state(struct amdgpu_device *adev,
 					    u64 *flags);
 int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev,
 				   enum amd_ip_block_type block_type);
+bool amdgpu_device_ip_is_hw(struct amdgpu_device *adev,
+			    enum amd_ip_block_type block_type);
 bool amdgpu_device_ip_is_valid(struct amdgpu_device *adev,
 			      enum amd_ip_block_type block_type);
 int amdgpu_ip_block_suspend(struct amdgpu_ip_block *ip_block);
 
 int amdgpu_ip_block_resume(struct amdgpu_ip_block *ip_block);
 
-#define AMDGPU_MAX_IP_NUM 16
+#define AMDGPU_MAX_IP_NUM AMD_IP_BLOCK_TYPE_NUM
 
 struct amdgpu_ip_block_status {
 	bool valid;
@@ -415,6 +422,7 @@ bool amdgpu_get_bios(struct amdgpu_device *adev);
 bool amdgpu_read_bios(struct amdgpu_device *adev);
 bool amdgpu_soc15_read_bios_from_rom(struct amdgpu_device *adev,
 				     u8 *bios, u32 length_bytes);
+void amdgpu_bios_release(struct amdgpu_device *adev);
 /*
  * Clocks
  */
@@ -429,7 +437,6 @@ struct amdgpu_clock {
 	uint32_t default_mclk;
 	uint32_t default_sclk;
 	uint32_t default_dispclk;
-	uint32_t current_dispclk;
 	uint32_t dp_extclk;
 	uint32_t max_pixel_clock;
 };
@@ -465,9 +472,6 @@ struct amdgpu_sa_manager {
 	void				*cpu_ptr;
 };
 
-int amdgpu_fence_slab_init(void);
-void amdgpu_fence_slab_fini(void);
-
 /*
  * IRQS.
  */
@@ -487,7 +491,6 @@ struct amdgpu_flip_work {
 	bool				async;
 };
 
-
 /*
  * file private structure
  */
@@ -500,6 +503,11 @@ struct amdgpu_fpriv {
 	struct mutex		bo_list_lock;
 	struct idr		bo_list_handles;
 	struct amdgpu_ctx_mgr	ctx_mgr;
+	struct amdgpu_userq_mgr	userq_mgr;
+
+	/* Eviction fence infra */
+	struct amdgpu_eviction_fence_mgr evf_mgr;
+
 	/** GPU partition selection */
 	uint32_t		xcp_id;
 };
@@ -511,12 +519,62 @@ int amdgpu_file_to_fpriv(struct file *filp, struct amdgpu_fpriv **fpriv);
  */
 #define AMDGPU_MAX_WB 1024	/* Reserve at most 1024 WB slots for amdgpu-owned rings. */
 
+/**
+ * amdgpu_wb - This struct is used for small GPU memory allocation.
+ *
+ * This struct is used to allocate a small amount of GPU memory that can be
+ * used to shadow certain states into the memory. This is especially useful for
+ * providing easy CPU access to some states without requiring register access
+ * (e.g., if some block is power gated, reading register may be problematic).
+ *
+ * Note: the term writeback was initially used because many of the amdgpu
+ * components had some level of writeback memory, and this struct initially
+ * described those components.
+ */
 struct amdgpu_wb {
+
+	/**
+	 * @wb_obj:
+	 *
+	 * Buffer Object used for the writeback memory.
+	 */
 	struct amdgpu_bo	*wb_obj;
-	volatile uint32_t	*wb;
+
+	/**
+	 * @wb:
+	 *
+	 * Pointer to the first writeback slot. In terms of CPU address
+	 * this value can be accessed directly by using the offset as an index.
+	 * For the GPU address, it is necessary to use gpu_addr and the offset.
+	 */
+	uint32_t		*wb;
+
+	/**
+	 * @gpu_addr:
+	 *
+	 * Writeback base address in the GPU.
+	 */
 	uint64_t		gpu_addr;
-	u32			num_wb;	/* Number of wb slots actually reserved for amdgpu. */
+
+	/**
+	 * @num_wb:
+	 *
+	 * Number of writeback slots reserved for amdgpu.
+	 */
+	u32			num_wb;
+
+	/**
+	 * @used:
+	 *
+	 * Track the writeback slot already used.
+	 */
 	unsigned long		used[DIV_ROUND_UP(AMDGPU_MAX_WB, BITS_PER_LONG)];
+
+	/**
+	 * @lock:
+	 *
+	 * Protects read and write of the used field array.
+	 */
 	spinlock_t		lock;
 };
 
@@ -550,6 +608,7 @@ struct amdgpu_allowed_register_entry {
  *                   are reset depends on the ASIC. Notably doesn't reset IPs
  *                   shared with the CPU on APUs or the memory controllers (so
  *                   VRAM is not lost). Not available on all ASICs.
+ * @AMD_RESET_LINK: Triggers SW-UP link reset on other GPUs
  * @AMD_RESET_BACO: BACO (Bus Alive, Chip Off) method powers off and on the card
  *                  but without powering off the PCI bus. Suitable only for
  *                  discrete GPUs.
@@ -567,6 +626,7 @@ enum amd_reset_method {
 	AMD_RESET_METHOD_MODE0,
 	AMD_RESET_METHOD_MODE1,
 	AMD_RESET_METHOD_MODE2,
+	AMD_RESET_METHOD_LINK,
 	AMD_RESET_METHOD_BACO,
 	AMD_RESET_METHOD_PCI,
 	AMD_RESET_METHOD_ON_INIT,
@@ -663,7 +723,7 @@ int amdgpu_cs_wait_fences_ioctl(struct drm_device *dev, void *data,
 /* VRAM scratch page for HDP bug, default vram page */
 struct amdgpu_mem_scratch {
 	struct amdgpu_bo		*robj;
-	volatile uint32_t		*ptr;
+	uint32_t			*ptr;
 	u64				gpu_addr;
 };
 
@@ -694,6 +754,7 @@ typedef void (*amdgpu_block_wreg_t)(struct amdgpu_device*, uint32_t, uint32_t, u
 struct amdgpu_mmio_remap {
 	u32 reg_offset;
 	resource_size_t bus_addr;
+	struct amdgpu_bo *bo;
 };
 
 /* Define the HW IP blocks will be used in driver , add more if necessary */
@@ -761,13 +822,25 @@ struct amdgpu_ip_map_info {
 					uint32_t mask);
 };
 
+enum amdgpu_uid_type {
+	AMDGPU_UID_TYPE_XCD,
+	AMDGPU_UID_TYPE_AID,
+	AMDGPU_UID_TYPE_SOC,
+	AMDGPU_UID_TYPE_MAX
+};
+
+#define AMDGPU_UID_INST_MAX 8 /* max number of instances for each UID type */
+
+struct amdgpu_uid {
+	uint64_t uid[AMDGPU_UID_TYPE_MAX][AMDGPU_UID_INST_MAX];
+	struct amdgpu_device *adev;
+};
+
 struct amd_powerplay {
 	void *pp_handle;
 	const struct amd_pm_funcs *pp_funcs;
 };
 
-struct ip_discovery_top;
-
 /* polaris10 kickers */
 #define ASICID_IS_P20(did, rid)		(((did == 0x67DF) && \
 					 ((rid == 0xE3) || \
@@ -820,6 +893,12 @@ struct amdgpu_mqd_prop {
 	uint32_t hqd_queue_priority;
 	bool allow_tunneling;
 	bool hqd_active;
+	uint64_t shadow_addr;
+	uint64_t gds_bkup_addr;
+	uint64_t csa_addr;
+	uint64_t fence_address;
+	bool tmz_queue;
+	bool kernel_queue;
 };
 
 struct amdgpu_mqd {
@@ -828,6 +907,15 @@ struct amdgpu_mqd {
 			struct amdgpu_mqd_prop *p);
 };
 
+struct amdgpu_pcie_reset_ctx {
+	bool in_link_reset;
+	bool occurs_dpc;
+	bool audio_suspended;
+	struct pci_dev *swus;
+	struct pci_saved_state *swus_pcistate;
+	struct pci_saved_state *swds_pcistate;
+};
+
 /*
  * Custom Init levels could be defined for different situations where a full
  * initialization of all hardware blocks are not expected. Sample cases are
@@ -852,10 +940,12 @@ struct amdgpu_init_level {
 struct amdgpu_reset_domain;
 struct amdgpu_fru_info;
 
-/*
- * Non-zero (true) if the GPU has VRAM. Zero (false) otherwise.
- */
-#define AMDGPU_HAS_VRAM(_adev) ((_adev)->gmc.real_vram_size)
+enum amdgpu_enforce_isolation_mode {
+	AMDGPU_ENFORCE_ISOLATION_DISABLE = 0,
+	AMDGPU_ENFORCE_ISOLATION_ENABLE = 1,
+	AMDGPU_ENFORCE_ISOLATION_ENABLE_LEGACY = 2,
+	AMDGPU_ENFORCE_ISOLATION_NO_CLEANER_SHADER = 3,
+};
 
 struct amdgpu_device {
 	struct device			*dev;
@@ -882,8 +972,7 @@ struct amdgpu_device {
 	struct notifier_block		acpi_nb;
 	struct notifier_block		pm_nb;
 	struct amdgpu_i2c_chan		*i2c_bus[AMDGPU_MAX_I2C_BUS];
-	struct debugfs_blob_wrapper     debugfs_vbios_blob;
-	struct debugfs_blob_wrapper     debugfs_discovery_blob;
+	struct debugfs_blob_wrapper debugfs_vbios_blob;
 	struct mutex			srbm_mutex;
 	/* GRBM index mutex. Protects concurrent access to GRBM index */
 	struct mutex                    grbm_idx_mutex;
@@ -973,6 +1062,9 @@ struct amdgpu_device {
 		u32			log2_max_MBps;
 	} mm_stats;
 
+	/* discovery*/
+	struct amdgpu_discovery_info discovery;
+
 	/* display */
 	bool				enable_virtual_display;
 	struct amdgpu_vkms_output       *amdgpu_vkms_output;
@@ -1060,9 +1152,6 @@ struct amdgpu_device {
 	/* for userq and VM fences */
 	struct amdgpu_seq64		seq64;
 
-	/* KFD */
-	struct amdgpu_kfd_dev		kfd;
-
 	/* UMC */
 	struct amdgpu_umc		umc;
 
@@ -1080,6 +1169,19 @@ struct amdgpu_device {
 	bool                            enable_uni_mes;
 	struct amdgpu_mes               mes;
 	struct amdgpu_mqd               mqds[AMDGPU_HW_IP_NUM];
+	const struct amdgpu_userq_funcs *userq_funcs[AMDGPU_HW_IP_NUM];
+
+	/* xarray used to retrieve the user queue fence driver reference
+	 * in the EOP interrupt handler to signal the particular user
+	 * queue fence.
+	 */
+	struct xarray			userq_xa;
+	/**
+	 * @userq_doorbell_xa: Global user queue map (doorbell index → queue)
+	 * Key: doorbell_index (unique global identifier for the queue)
+	 * Value: struct amdgpu_usermode_queue
+	 */
+	struct xarray userq_doorbell_xa;
 
 	/* df */
 	struct amdgpu_df                df;
@@ -1090,6 +1192,9 @@ struct amdgpu_device {
 	/* ACA */
 	struct amdgpu_aca		aca;
 
+	/* CPER */
+	struct amdgpu_cper		cper;
+
 	struct amdgpu_ip_block          ip_blocks[AMDGPU_MAX_IP_NUM];
 	uint32_t		        harvest_ip_mask;
 	int				num_ip_blocks;
@@ -1119,6 +1224,7 @@ struct amdgpu_device {
 	bool				in_s3;
 	bool				in_s4;
 	bool				in_s0ix;
+	suspend_state_t			last_suspend_state;
 
 	enum pp_mp1_state               mp1_state;
 	struct amdgpu_doorbell_index doorbell_index;
@@ -1149,11 +1255,14 @@ struct amdgpu_device {
 	struct ratelimit_state		throttling_logging_rs;
 	uint32_t                        ras_hw_enabled;
 	uint32_t                        ras_enabled;
+	bool                            ras_default_ecc_enabled;
 
 	bool                            no_hw_access;
 	struct pci_saved_state          *pci_state;
 	pci_channel_state_t		pci_channel_state;
 
+	struct amdgpu_pcie_reset_ctx	pcie_reset_ctx;
+
 	/* Track auto wait count on s_barrier settings */
 	bool				barrier_has_auto_waitcnt;
 
@@ -1164,8 +1273,6 @@ struct amdgpu_device {
 
 	struct list_head                ras_list;
 
-	struct ip_discovery_top         *ip_top;
-
 	struct amdgpu_reset_domain	*reset_domain;
 
 	struct mutex			benchmark_mutex;
@@ -1186,12 +1293,36 @@ struct amdgpu_device {
 	bool                            debug_use_vram_fw_buf;
 	bool                            debug_enable_ras_aca;
 	bool                            debug_exp_resets;
+	bool                            debug_disable_gpu_ring_reset;
+	bool                            debug_vm_userptr;
+	bool                            debug_disable_ce_logs;
+	bool                            debug_enable_ce_cs;
 
-	bool				enforce_isolation[MAX_XCP];
-	/* Added this mutex for cleaner shader isolation between GFX and compute processes */
+	/* Protection for the following isolation structure */
 	struct mutex                    enforce_isolation_mutex;
+	enum amdgpu_enforce_isolation_mode	enforce_isolation[MAX_XCP];
+	struct amdgpu_isolation {
+		void			*owner;
+		struct dma_fence	*spearhead;
+		struct amdgpu_sync	active;
+		struct amdgpu_sync	prev;
+	} isolation[MAX_XCP];
 
 	struct amdgpu_init_level *init_lvl;
+
+	/* This flag is used to determine how VRAM allocations are handled for APUs
+	 * in KFD: VRAM or GTT.
+	 */
+	bool                            apu_prefer_gtt;
+
+	bool                            userq_halt_for_enforce_isolation;
+	struct work_struct              userq_reset_work;
+	struct amdgpu_uid *uid_info;
+
+	/* KFD
+	 * Must be last --ends in a flexible-array member.
+	 */
+	struct amdgpu_kfd_dev		kfd;
 };
 
 static inline uint32_t amdgpu_ip_version(const struct amdgpu_device *adev,
@@ -1225,6 +1356,11 @@ static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_device *bdev)
 	return container_of(bdev, struct amdgpu_device, mman.bdev);
 }
 
+static inline bool amdgpu_is_multi_aid(struct amdgpu_device *adev)
+{
+	return !!adev->aid_mask;
+}
+
 int amdgpu_device_init(struct amdgpu_device *adev,
 		       uint32_t flags);
 void amdgpu_device_fini_hw(struct amdgpu_device *adev);
@@ -1276,7 +1412,8 @@ void amdgpu_device_indirect_wreg64(struct amdgpu_device *adev,
 void amdgpu_device_indirect_wreg64_ext(struct amdgpu_device *adev,
 				   u64 reg_addr, u64 reg_data);
 u32 amdgpu_device_get_rev_id(struct amdgpu_device *adev);
-bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type);
+bool amdgpu_device_asic_has_dc_support(struct pci_dev *pdev,
+				       enum amd_asic_type asic_type);
 bool amdgpu_device_has_dc_support(struct amdgpu_device *adev);
 
 void amdgpu_device_set_sriov_virtual_display(struct amdgpu_device *adev);
@@ -1403,11 +1540,6 @@ int emu_soc_asic_init(struct amdgpu_device *adev);
 #define amdgpu_asic_read_bios_from_rom(adev, b, l) (adev)->asic_funcs->read_bios_from_rom((adev), (b), (l))
 #define amdgpu_asic_read_register(adev, se, sh, offset, v)((adev)->asic_funcs->read_register((adev), (se), (sh), (offset), (v)))
 #define amdgpu_asic_get_config_memsize(adev) (adev)->asic_funcs->get_config_memsize((adev))
-#define amdgpu_asic_flush_hdp(adev, r) \
-	((adev)->asic_funcs->flush_hdp ? (adev)->asic_funcs->flush_hdp((adev), (r)) : (adev)->hdp.funcs->flush_hdp((adev), (r)))
-#define amdgpu_asic_invalidate_hdp(adev, r) \
-	((adev)->asic_funcs->invalidate_hdp ? (adev)->asic_funcs->invalidate_hdp((adev), (r)) : \
-	 ((adev)->hdp.funcs->invalidate_hdp ? (adev)->hdp.funcs->invalidate_hdp((adev), (r)) : (void)0))
 #define amdgpu_asic_need_full_reset(adev) (adev)->asic_funcs->need_full_reset((adev))
 #define amdgpu_asic_init_doorbell_index(adev) (adev)->asic_funcs->init_doorbell_index((adev))
 #define amdgpu_asic_get_pcie_usage(adev, cnt0, cnt1) ((adev)->asic_funcs->get_pcie_usage((adev), (cnt0), (cnt1)))
@@ -1446,16 +1578,17 @@ void amdgpu_device_program_register_sequence(struct amdgpu_device *adev,
 					     const u32 array_size);
 
 int amdgpu_device_mode1_reset(struct amdgpu_device *adev);
-bool amdgpu_device_supports_atpx(struct drm_device *dev);
-bool amdgpu_device_supports_px(struct drm_device *dev);
-bool amdgpu_device_supports_boco(struct drm_device *dev);
-bool amdgpu_device_supports_smart_shift(struct drm_device *dev);
-int amdgpu_device_supports_baco(struct drm_device *dev);
+int amdgpu_device_link_reset(struct amdgpu_device *adev);
+bool amdgpu_device_supports_atpx(struct amdgpu_device *adev);
+bool amdgpu_device_supports_px(struct amdgpu_device *adev);
+bool amdgpu_device_supports_boco(struct amdgpu_device *adev);
+bool amdgpu_device_supports_smart_shift(struct amdgpu_device *adev);
+int amdgpu_device_supports_baco(struct amdgpu_device *adev);
 void amdgpu_device_detect_runtime_pm_mode(struct amdgpu_device *adev);
 bool amdgpu_device_is_peer_accessible(struct amdgpu_device *adev,
 				      struct amdgpu_device *peer_adev);
-int amdgpu_device_baco_enter(struct drm_device *dev);
-int amdgpu_device_baco_exit(struct drm_device *dev);
+int amdgpu_device_baco_enter(struct amdgpu_device *adev);
+int amdgpu_device_baco_exit(struct amdgpu_device *adev);
 
 void amdgpu_device_flush_hdp(struct amdgpu_device *adev,
 		struct amdgpu_ring *ring);
@@ -1470,6 +1603,9 @@ void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev,
 struct dma_fence *amdgpu_device_get_gang(struct amdgpu_device *adev);
 struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,
 					    struct dma_fence *gang);
+struct dma_fence *amdgpu_device_enforce_isolation(struct amdgpu_device *adev,
+						  struct amdgpu_ring *ring,
+						  struct amdgpu_job *job);
 bool amdgpu_device_has_display_hardware(struct amdgpu_device *adev);
 ssize_t amdgpu_get_soft_full_reset_mask(struct amdgpu_ring *ring);
 ssize_t amdgpu_show_reset_mask(char *buf, uint32_t supported_reset);
@@ -1502,8 +1638,8 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 				 struct drm_file *file_priv);
 void amdgpu_driver_release_kms(struct drm_device *dev);
 
-int amdgpu_device_ip_suspend(struct amdgpu_device *adev);
 int amdgpu_device_prepare(struct drm_device *dev);
+void amdgpu_device_complete(struct drm_device *dev);
 int amdgpu_device_suspend(struct drm_device *dev, bool fbcon);
 int amdgpu_device_resume(struct drm_device *dev, bool fbcon);
 u32 amdgpu_get_vblank_counter_kms(struct drm_crtc *crtc);
@@ -1554,7 +1690,8 @@ int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
 						u8 perf_req, bool advertise);
 int amdgpu_acpi_power_shift_control(struct amdgpu_device *adev,
 				    u8 dev_state, bool drv_state);
-int amdgpu_acpi_smart_shift_update(struct drm_device *dev, enum amdgpu_ss ss_state);
+int amdgpu_acpi_smart_shift_update(struct amdgpu_device *adev,
+				   enum amdgpu_ss ss_state);
 int amdgpu_acpi_pcie_notify_device_ready(struct amdgpu_device *adev);
 int amdgpu_acpi_get_tmr_info(struct amdgpu_device *adev, u64 *tmr_offset,
 			     u64 *tmr_size);
@@ -1585,19 +1722,24 @@ static inline void amdgpu_acpi_release(void) { }
 static inline bool amdgpu_acpi_is_power_shift_control_supported(void) { return false; }
 static inline int amdgpu_acpi_power_shift_control(struct amdgpu_device *adev,
 						  u8 dev_state, bool drv_state) { return 0; }
-static inline int amdgpu_acpi_smart_shift_update(struct drm_device *dev,
-						 enum amdgpu_ss ss_state) { return 0; }
+static inline int amdgpu_acpi_smart_shift_update(struct amdgpu_device *adev,
+						 enum amdgpu_ss ss_state)
+{
+	return 0;
+}
 static inline void amdgpu_acpi_get_backlight_caps(struct amdgpu_dm_backlight_caps *caps) { }
 #endif
 
 #if defined(CONFIG_ACPI) && defined(CONFIG_SUSPEND)
 bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev);
 bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev);
-void amdgpu_choose_low_power_state(struct amdgpu_device *adev);
 #else
 static inline bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev) { return false; }
 static inline bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev) { return false; }
-static inline void amdgpu_choose_low_power_state(struct amdgpu_device *adev) { }
+#endif
+
+#if defined(CONFIG_DRM_AMD_ISP)
+int amdgpu_acpi_get_isp4_dev(struct acpi_device **dev);
 #endif
 
 void amdgpu_register_gpu_instance(struct amdgpu_device *adev);
@@ -1643,4 +1785,24 @@ extern const struct attribute_group amdgpu_flash_attr_group;
 
 void amdgpu_set_init_level(struct amdgpu_device *adev,
 			   enum amdgpu_init_lvl_id lvl);
+
+static inline int amdgpu_device_bus_status_check(struct amdgpu_device *adev)
+{
+       u32 status;
+       int r;
+
+       r = pci_read_config_dword(adev->pdev, PCI_COMMAND, &status);
+       if (r || PCI_POSSIBLE_ERROR(status)) {
+		dev_err(adev->dev, "device lost from bus!");
+		return -ENODEV;
+       }
+
+       return 0;
+}
+
+void amdgpu_device_set_uid(struct amdgpu_uid *uid_info,
+			   enum amdgpu_uid_type type, uint8_t inst,
+			   uint64_t uid);
+uint64_t amdgpu_device_get_uid(struct amdgpu_uid *uid_info,
+			       enum amdgpu_uid_type type, uint8_t inst);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
index 9d6345146495..9b3180449150 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
@@ -30,16 +30,6 @@
 
 typedef int bank_handler_t(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type, void *data);
 
-struct aca_banks {
-	int nr_banks;
-	struct list_head list;
-};
-
-struct aca_hwip {
-	int hwid;
-	int mcatype;
-};
-
 static struct aca_hwip aca_hwid_mcatypes[ACA_HWIP_TYPE_COUNT] = {
 	ACA_BANK_HWID(SMU,	0x01,	0x01),
 	ACA_BANK_HWID(PCS_XGMI, 0x50,	0x00),
@@ -86,6 +76,7 @@ static void aca_banks_release(struct aca_banks *banks)
 	list_for_each_entry_safe(node, tmp, &banks->list, node) {
 		list_del(&node->node);
 		kvfree(node);
+		banks->nr_banks--;
 	}
 }
 
@@ -111,7 +102,7 @@ static struct aca_regs_dump {
 	{"STATUS",		ACA_REG_IDX_STATUS},
 	{"ADDR",		ACA_REG_IDX_ADDR},
 	{"MISC",		ACA_REG_IDX_MISC0},
-	{"CONFIG",		ACA_REG_IDX_CONFG},
+	{"CONFIG",		ACA_REG_IDX_CONFIG},
 	{"IPID",		ACA_REG_IDX_IPID},
 	{"SYND",		ACA_REG_IDX_SYND},
 	{"DESTAT",		ACA_REG_IDX_DESTAT},
@@ -125,11 +116,40 @@ static void aca_smu_bank_dump(struct amdgpu_device *adev, int idx, int total, st
 	u64 event_id = qctx ? qctx->evid.event_id : RAS_EVENT_INVALID_ID;
 	int i;
 
+	if (adev->debug_disable_ce_logs &&
+	    bank->smu_err_type == ACA_SMU_TYPE_CE &&
+	    !ACA_BANK_ERR_IS_DEFFERED(bank))
+		return;
+
 	RAS_EVENT_LOG(adev, event_id, HW_ERR "Accelerator Check Architecture events logged\n");
 	/* plus 1 for output format, e.g: ACA[08/08]: xxxx */
 	for (i = 0; i < ARRAY_SIZE(aca_regs); i++)
 		RAS_EVENT_LOG(adev, event_id, HW_ERR "ACA[%02d/%02d].%s=0x%016llx\n",
 			      idx + 1, total, aca_regs[i].name, bank->regs[aca_regs[i].reg_idx]);
+
+	if (ACA_REG__STATUS__SCRUB(bank->regs[ACA_REG_IDX_STATUS]))
+		RAS_EVENT_LOG(adev, event_id, HW_ERR "hardware error logged by the scrubber\n");
+}
+
+static bool aca_bank_hwip_is_matched(struct aca_bank *bank, enum aca_hwip_type type)
+{
+
+	struct aca_hwip *hwip;
+	int hwid, mcatype;
+	u64 ipid;
+
+	if (!bank || type == ACA_HWIP_TYPE_UNKNOW)
+		return false;
+
+	hwip = &aca_hwid_mcatypes[type];
+	if (!hwip->hwid)
+		return false;
+
+	ipid = bank->regs[ACA_REG_IDX_IPID];
+	hwid = ACA_REG__IPID__HARDWAREID(ipid);
+	mcatype = ACA_REG__IPID__MCATYPE(ipid);
+
+	return hwip->hwid == hwid && hwip->mcatype == mcatype;
 }
 
 static int aca_smu_get_valid_aca_banks(struct amdgpu_device *adev, enum aca_smu_type type,
@@ -168,7 +188,16 @@ static int aca_smu_get_valid_aca_banks(struct amdgpu_device *adev, enum aca_smu_
 		if (ret)
 			return ret;
 
-		bank.type = type;
+		bank.smu_err_type = type;
+
+		/*
+		 * Poison being consumed when injecting a UE while running background workloads,
+		 * which are unexpected.
+		 */
+		if (type == ACA_SMU_TYPE_UE &&
+		    ACA_REG__STATUS__POISON(bank.regs[ACA_REG_IDX_STATUS]) &&
+		    !aca_bank_hwip_is_matched(&bank, ACA_HWIP_TYPE_UMC))
+			continue;
 
 		aca_smu_bank_dump(adev, i, count, &bank, qctx);
 
@@ -180,31 +209,14 @@ static int aca_smu_get_valid_aca_banks(struct amdgpu_device *adev, enum aca_smu_
 	return 0;
 }
 
-static bool aca_bank_hwip_is_matched(struct aca_bank *bank, enum aca_hwip_type type)
-{
-
-	struct aca_hwip *hwip;
-	int hwid, mcatype;
-	u64 ipid;
-
-	if (!bank || type == ACA_HWIP_TYPE_UNKNOW)
-		return false;
-
-	hwip = &aca_hwid_mcatypes[type];
-	if (!hwip->hwid)
-		return false;
-
-	ipid = bank->regs[ACA_REG_IDX_IPID];
-	hwid = ACA_REG__IPID__HARDWAREID(ipid);
-	mcatype = ACA_REG__IPID__MCATYPE(ipid);
-
-	return hwip->hwid == hwid && hwip->mcatype == mcatype;
-}
-
 static bool aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, enum aca_smu_type type)
 {
 	const struct aca_bank_ops *bank_ops = handle->bank_ops;
 
+	/* Parse all deferred errors with UMC aca handle */
+	if (ACA_BANK_ERR_IS_DEFFERED(bank))
+		return handle->hwip == ACA_HWIP_TYPE_UMC;
+
 	if (!aca_bank_hwip_is_matched(bank, handle->hwip))
 		return false;
 
@@ -227,6 +239,7 @@ static struct aca_bank_error *new_bank_error(struct aca_error *aerr, struct aca_
 
 	mutex_lock(&aerr->lock);
 	list_add_tail(&bank_error->node, &aerr->list);
+	aerr->nr_errors++;
 	mutex_unlock(&aerr->lock);
 
 	return bank_error;
@@ -394,6 +407,56 @@ static bool aca_bank_should_update(struct amdgpu_device *adev, enum aca_smu_type
 	return ret;
 }
 
+static void aca_banks_generate_cper(struct amdgpu_device *adev,
+				    enum aca_smu_type type,
+				    struct aca_banks *banks,
+				    int count)
+{
+	struct aca_bank_node *node;
+	struct aca_bank *bank;
+	int r;
+
+	if (!adev->cper.enabled)
+		return;
+
+	if (!banks || !count) {
+		dev_warn(adev->dev, "fail to generate cper records\n");
+		return;
+	}
+
+	/* UEs must be encoded into separate CPER entries */
+	if (type == ACA_SMU_TYPE_UE) {
+		struct aca_banks de_banks;
+
+		aca_banks_init(&de_banks);
+		list_for_each_entry(node, &banks->list, node) {
+			bank = &node->bank;
+			if (bank->aca_err_type == ACA_ERROR_TYPE_DEFERRED) {
+				r = aca_banks_add_bank(&de_banks, bank);
+				if (r)
+					dev_warn(adev->dev, "fail to add de banks, ret = %d\n", r);
+			} else {
+				if (amdgpu_cper_generate_ue_record(adev, bank))
+					dev_warn(adev->dev, "fail to generate ue cper records\n");
+			}
+		}
+
+		if (!list_empty(&de_banks.list)) {
+			if (amdgpu_cper_generate_ce_records(adev, &de_banks, de_banks.nr_banks))
+				dev_warn(adev->dev, "fail to generate de cper records\n");
+		}
+
+		aca_banks_release(&de_banks);
+	} else {
+		/*
+		 * SMU_TYPE_CE banks are combined into 1 CPER entries,
+		 * they could be CEs or DEs or both
+		 */
+		if (amdgpu_cper_generate_ce_records(adev, banks, count))
+			dev_warn(adev->dev, "fail to generate ce cper records\n");
+	}
+}
+
 static int aca_banks_update(struct amdgpu_device *adev, enum aca_smu_type type,
 			    bank_handler_t handler, struct ras_query_context *qctx, void *data)
 {
@@ -431,6 +494,8 @@ static int aca_banks_update(struct amdgpu_device *adev, enum aca_smu_type type,
 	if (ret)
 		goto err_release_banks;
 
+	aca_banks_generate_cper(adev, type, &banks, count);
+
 err_release_banks:
 	aca_banks_release(&banks);
 
@@ -516,6 +581,10 @@ static int __aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *h
 	if (ret)
 		return ret;
 
+	/* DEs may contain in CEs or UEs */
+	if (type != ACA_ERROR_TYPE_DEFERRED)
+		aca_log_aca_error(handle, ACA_ERROR_TYPE_DEFERRED, err_data);
+
 	return aca_log_aca_error(handle, type, err_data);
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h
index f3289d289913..38c88897e1ec 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h
@@ -76,12 +76,16 @@ struct ras_query_context;
 #define mmSMNAID_XCD1_MCA_SMU		0x38430400	/* SMN AID XCD1 */
 #define mmSMNXCD_XCD0_MCA_SMU		0x40430400	/* SMN XCD XCD0 */
 
+#define ACA_BANK_ERR_IS_DEFFERED(bank)                                \
+	(ACA_REG__STATUS__POISON((bank)->regs[ACA_REG_IDX_STATUS]) || \
+	 ACA_REG__STATUS__DEFERRED((bank)->regs[ACA_REG_IDX_STATUS]))
+
 enum aca_reg_idx {
 	ACA_REG_IDX_CTL			= 0,
 	ACA_REG_IDX_STATUS		= 1,
 	ACA_REG_IDX_ADDR		= 2,
 	ACA_REG_IDX_MISC0		= 3,
-	ACA_REG_IDX_CONFG		= 4,
+	ACA_REG_IDX_CONFIG		= 4,
 	ACA_REG_IDX_IPID		= 5,
 	ACA_REG_IDX_SYND		= 6,
 	ACA_REG_IDX_DESTAT		= 8,
@@ -108,13 +112,20 @@ enum aca_error_type {
 };
 
 enum aca_smu_type {
+	ACA_SMU_TYPE_INVALID = -1,
 	ACA_SMU_TYPE_UE = 0,
 	ACA_SMU_TYPE_CE,
 	ACA_SMU_TYPE_COUNT,
 };
 
+struct aca_hwip {
+	int hwid;
+	int mcatype;
+};
+
 struct aca_bank {
-	enum aca_smu_type type;
+	enum aca_error_type aca_err_type;
+	enum aca_smu_type smu_err_type;
 	u64 regs[ACA_MAX_REGS_COUNT];
 };
 
@@ -123,6 +134,11 @@ struct aca_bank_node {
 	struct list_head node;
 };
 
+struct aca_banks {
+	int nr_banks;
+	struct list_head list;
+};
+
 struct aca_bank_info {
 	int die_id;
 	int socket_id;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
index deb0785350e8..381ef205b0df 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
@@ -302,17 +302,19 @@ static int acp_hw_init(struct amdgpu_ip_block *ip_block)
 		adev->acp.acp_res[2].end = adev->acp.acp_res[2].start;
 
 		adev->acp.acp_cell[0].name = "acp_audio_dma";
+		adev->acp.acp_cell[0].id = 0;
 		adev->acp.acp_cell[0].num_resources = 3;
 		adev->acp.acp_cell[0].resources = &adev->acp.acp_res[0];
 		adev->acp.acp_cell[0].platform_data = &adev->asic_type;
 		adev->acp.acp_cell[0].pdata_size = sizeof(adev->asic_type);
 
 		adev->acp.acp_cell[1].name = "designware-i2s";
+		adev->acp.acp_cell[1].id = 1;
 		adev->acp.acp_cell[1].num_resources = 1;
 		adev->acp.acp_cell[1].resources = &adev->acp.acp_res[1];
 		adev->acp.acp_cell[1].platform_data = &i2s_pdata[0];
 		adev->acp.acp_cell[1].pdata_size = sizeof(struct i2s_platform_data);
-		r = mfd_add_hotplug_devices(adev->acp.parent, adev->acp.acp_cell, 2);
+		r = mfd_add_devices(adev->acp.parent, 0, adev->acp.acp_cell, 2, NULL, 0, NULL);
 		if (r)
 			goto failure;
 		r = device_for_each_child(adev->acp.parent, &adev->acp.acp_genpd->gpd,
@@ -410,30 +412,34 @@ static int acp_hw_init(struct amdgpu_ip_block *ip_block)
 		adev->acp.acp_res[4].end = adev->acp.acp_res[4].start;
 
 		adev->acp.acp_cell[0].name = "acp_audio_dma";
+		adev->acp.acp_cell[0].id = 0;
 		adev->acp.acp_cell[0].num_resources = 5;
 		adev->acp.acp_cell[0].resources = &adev->acp.acp_res[0];
 		adev->acp.acp_cell[0].platform_data = &adev->asic_type;
 		adev->acp.acp_cell[0].pdata_size = sizeof(adev->asic_type);
 
 		adev->acp.acp_cell[1].name = "designware-i2s";
+		adev->acp.acp_cell[1].id = 1;
 		adev->acp.acp_cell[1].num_resources = 1;
 		adev->acp.acp_cell[1].resources = &adev->acp.acp_res[1];
 		adev->acp.acp_cell[1].platform_data = &i2s_pdata[0];
 		adev->acp.acp_cell[1].pdata_size = sizeof(struct i2s_platform_data);
 
 		adev->acp.acp_cell[2].name = "designware-i2s";
+		adev->acp.acp_cell[2].id = 2;
 		adev->acp.acp_cell[2].num_resources = 1;
 		adev->acp.acp_cell[2].resources = &adev->acp.acp_res[2];
 		adev->acp.acp_cell[2].platform_data = &i2s_pdata[1];
 		adev->acp.acp_cell[2].pdata_size = sizeof(struct i2s_platform_data);
 
 		adev->acp.acp_cell[3].name = "designware-i2s";
+		adev->acp.acp_cell[3].id = 3;
 		adev->acp.acp_cell[3].num_resources = 1;
 		adev->acp.acp_cell[3].resources = &adev->acp.acp_res[3];
 		adev->acp.acp_cell[3].platform_data = &i2s_pdata[2];
 		adev->acp.acp_cell[3].pdata_size = sizeof(struct i2s_platform_data);
 
-		r = mfd_add_hotplug_devices(adev->acp.parent, adev->acp.acp_cell, ACP_DEVS);
+		r = mfd_add_devices(adev->acp.parent, 0, adev->acp.acp_cell, ACP_DEVS, NULL, 0, NULL);
 		if (r)
 			goto failure;
 
@@ -579,7 +585,7 @@ static int acp_resume(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-static bool acp_is_idle(void *handle)
+static bool acp_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
index b8d4e07d2043..d31460a9e958 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
@@ -394,6 +394,10 @@ static int amdgpu_atif_query_backlight_caps(struct amdgpu_atif *atif)
 			characteristics.max_input_signal;
 	atif->backlight_caps.ac_level = characteristics.ac_level;
 	atif->backlight_caps.dc_level = characteristics.dc_level;
+	atif->backlight_caps.data_points = characteristics.number_of_points;
+	memcpy(atif->backlight_caps.luminance_data,
+	       characteristics.data_points,
+	       sizeof(atif->backlight_caps.luminance_data));
 out:
 	kfree(info);
 	return err;
@@ -503,7 +507,6 @@ static int amdgpu_atif_handler(struct amdgpu_device *adev,
 				pm_runtime_get_sync(adev_to_drm(adev)->dev);
 				/* Just fire off a uevent and let userspace tell us what to do */
 				drm_helper_hpd_irq_event(adev_to_drm(adev));
-				pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 				pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 			}
 		}
@@ -807,18 +810,18 @@ int amdgpu_acpi_power_shift_control(struct amdgpu_device *adev,
 /**
  * amdgpu_acpi_smart_shift_update - update dGPU device state to SBIOS
  *
- * @dev: drm_device pointer
+ * @adev: amdgpu device pointer
  * @ss_state: current smart shift event
  *
  * returns 0 on success,
  * otherwise return error number.
  */
-int amdgpu_acpi_smart_shift_update(struct drm_device *dev, enum amdgpu_ss ss_state)
+int amdgpu_acpi_smart_shift_update(struct amdgpu_device *adev,
+				   enum amdgpu_ss ss_state)
 {
-	struct amdgpu_device *adev = drm_to_adev(dev);
 	int r;
 
-	if (!amdgpu_device_supports_smart_shift(dev))
+	if (!amdgpu_device_supports_smart_shift(adev))
 		return 0;
 
 	switch (ss_state) {
@@ -1277,11 +1280,7 @@ void amdgpu_acpi_get_backlight_caps(struct amdgpu_dm_backlight_caps *caps)
 {
 	struct amdgpu_atif *atif = &amdgpu_acpi_priv.atif;
 
-	caps->caps_valid = atif->backlight_caps.caps_valid;
-	caps->min_input_signal = atif->backlight_caps.min_input_signal;
-	caps->max_input_signal = atif->backlight_caps.max_input_signal;
-	caps->ac_level = atif->backlight_caps.ac_level;
-	caps->dc_level = atif->backlight_caps.dc_level;
+	memcpy(caps, &atif->backlight_caps, sizeof(*caps));
 }
 
 /**
@@ -1532,23 +1531,35 @@ bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev)
 	return true;
 #endif /* CONFIG_AMD_PMC */
 }
+#endif /* CONFIG_SUSPEND */
 
-/**
- * amdgpu_choose_low_power_state
- *
- * @adev: amdgpu_device_pointer
- *
- * Choose the target low power state for the GPU
- */
-void amdgpu_choose_low_power_state(struct amdgpu_device *adev)
-{
-	if (adev->in_runpm)
-		return;
+#if IS_ENABLED(CONFIG_DRM_AMD_ISP)
+static const struct acpi_device_id isp_sensor_ids[] = {
+	{ "OMNI5C10" },
+	{ }
+};
 
-	if (amdgpu_acpi_is_s0ix_active(adev))
-		adev->in_s0ix = true;
-	else if (amdgpu_acpi_is_s3_active(adev))
-		adev->in_s3 = true;
+static int isp_match_acpi_device_ids(struct device *dev, const void *data)
+{
+	return acpi_match_device(data, dev) ? 1 : 0;
 }
 
-#endif /* CONFIG_SUSPEND */
+int amdgpu_acpi_get_isp4_dev(struct acpi_device **dev)
+{
+	struct device *pdev __free(put_device) = NULL;
+	struct acpi_device *acpi_pdev;
+
+	pdev = bus_find_device(&platform_bus_type, NULL, isp_sensor_ids,
+			       isp_match_acpi_device_ids);
+	if (!pdev)
+		return -EINVAL;
+
+	acpi_pdev = ACPI_COMPANION(pdev);
+	if (!acpi_pdev)
+		return -ENODEV;
+
+	*dev = acpi_pdev;
+
+	return 0;
+}
+#endif /* CONFIG_DRM_AMD_ISP */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 2c1b38c5cfc6..a2879d2b7c8e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -248,18 +248,42 @@ void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
 		kgd2kfd_interrupt(adev->kfd.dev, ih_ring_entry);
 }
 
-void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool run_pm)
+void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool suspend_proc)
+{
+	if (adev->kfd.dev) {
+		if (adev->in_s0ix)
+			kgd2kfd_stop_sched_all_nodes(adev->kfd.dev);
+		else
+			kgd2kfd_suspend(adev->kfd.dev, suspend_proc);
+	}
+}
+
+int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool resume_proc)
+{
+	int r = 0;
+
+	if (adev->kfd.dev) {
+		if (adev->in_s0ix)
+			r = kgd2kfd_start_sched_all_nodes(adev->kfd.dev);
+		else
+			r = kgd2kfd_resume(adev->kfd.dev, resume_proc);
+	}
+
+	return r;
+}
+
+void amdgpu_amdkfd_suspend_process(struct amdgpu_device *adev)
 {
 	if (adev->kfd.dev)
-		kgd2kfd_suspend(adev->kfd.dev, run_pm);
+		kgd2kfd_suspend_process(adev->kfd.dev);
 }
 
-int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool run_pm)
+int amdgpu_amdkfd_resume_process(struct amdgpu_device *adev)
 {
 	int r = 0;
 
 	if (adev->kfd.dev)
-		r = kgd2kfd_resume(adev->kfd.dev, run_pm);
+		r = kgd2kfd_resume_process(adev->kfd.dev);
 
 	return r;
 }
@@ -368,6 +392,9 @@ void amdgpu_amdkfd_free_gtt_mem(struct amdgpu_device *adev, void **mem_obj)
 {
 	struct amdgpu_bo **bo = (struct amdgpu_bo **) mem_obj;
 
+	if (!bo || !*bo)
+		return;
+
 	(void)amdgpu_bo_reserve(*bo, true);
 	amdgpu_bo_kunmap(*bo);
 	amdgpu_bo_unpin(*bo);
@@ -459,7 +486,7 @@ void amdgpu_amdkfd_get_local_mem_info(struct amdgpu_device *adev,
 		else
 			mem_info->local_mem_size_private =
 					KFD_XCP_MEMORY_SIZE(adev, xcp->id);
-	} else if (adev->flags & AMD_IS_APU) {
+	} else if (adev->apu_prefer_gtt) {
 		mem_info->local_mem_size_public = (ttm_tt_pages_limit() << PAGE_SHIFT);
 		mem_info->local_mem_size_private = 0;
 	} else {
@@ -555,48 +582,6 @@ out_put:
 	return r;
 }
 
-uint8_t amdgpu_amdkfd_get_xgmi_hops_count(struct amdgpu_device *dst,
-					  struct amdgpu_device *src)
-{
-	struct amdgpu_device *peer_adev = src;
-	struct amdgpu_device *adev = dst;
-	int ret = amdgpu_xgmi_get_hops_count(adev, peer_adev);
-
-	if (ret < 0) {
-		DRM_ERROR("amdgpu: failed to get  xgmi hops count between node %d and %d. ret = %d\n",
-			adev->gmc.xgmi.physical_node_id,
-			peer_adev->gmc.xgmi.physical_node_id, ret);
-		ret = 0;
-	}
-	return  (uint8_t)ret;
-}
-
-int amdgpu_amdkfd_get_xgmi_bandwidth_mbytes(struct amdgpu_device *dst,
-					    struct amdgpu_device *src,
-					    bool is_min)
-{
-	struct amdgpu_device *adev = dst, *peer_adev;
-	int num_links;
-
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(9, 4, 2))
-		return 0;
-
-	if (src)
-		peer_adev = src;
-
-	/* num links returns 0 for indirect peers since indirect route is unknown. */
-	num_links = is_min ? 1 : amdgpu_xgmi_get_num_links(adev, peer_adev);
-	if (num_links < 0) {
-		DRM_ERROR("amdgpu: failed to get xgmi num links between node %d and %d. ret = %d\n",
-			adev->gmc.xgmi.physical_node_id,
-			peer_adev->gmc.xgmi.physical_node_id, num_links);
-		num_links = 0;
-	}
-
-	/* Aldebaran xGMI DPM is defeatured so assume x16 x 25Gbps for bandwidth. */
-	return (num_links * 16 * 25000)/BITS_PER_BYTE;
-}
-
 int amdgpu_amdkfd_get_pcie_bandwidth_mbytes(struct amdgpu_device *adev, bool is_min)
 {
 	int num_lanes_shift = (is_min ? ffs(adev->pm.pcie_mlw_mask) :
@@ -681,7 +666,7 @@ int amdgpu_amdkfd_submit_ib(struct amdgpu_device *adev,
 		goto err;
 	}
 
-	ret = amdgpu_job_alloc(adev, NULL, NULL, NULL, 1, &job);
+	ret = amdgpu_job_alloc(adev, NULL, NULL, NULL, 1, &job, 0);
 	if (ret)
 		goto err;
 
@@ -788,12 +773,12 @@ int amdgpu_amdkfd_send_close_event_drain_irq(struct amdgpu_device *adev,
 
 int amdgpu_amdkfd_check_and_lock_kfd(struct amdgpu_device *adev)
 {
-	return kgd2kfd_check_and_lock_kfd();
+	return kgd2kfd_check_and_lock_kfd(adev->kfd.dev);
 }
 
 void amdgpu_amdkfd_unlock_kfd(struct amdgpu_device *adev)
 {
-	kgd2kfd_unlock_kfd();
+	kgd2kfd_unlock_kfd(adev->kfd.dev);
 }
 
 
@@ -818,7 +803,7 @@ u64 amdgpu_amdkfd_xcp_memory_size(struct amdgpu_device *adev, int xcp_id)
 		}
 		do_div(tmp, adev->xcp_mgr->num_xcp_per_mem_partition);
 		return ALIGN_DOWN(tmp, PAGE_SIZE);
-	} else if (adev->flags & AMD_IS_APU) {
+	} else if (adev->apu_prefer_gtt) {
 		return (ttm_tt_pages_limit() << PAGE_SHIFT);
 	} else {
 		return adev->gmc.real_vram_size;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index 8af67f18500a..8bdfcde2029b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -47,6 +47,7 @@ enum TLB_FLUSH_TYPE {
 };
 
 struct amdgpu_device;
+struct kfd_process_device;
 struct amdgpu_reset_context;
 
 enum kfd_mem_attachment_type {
@@ -70,7 +71,7 @@ struct kgd_mem {
 	struct mutex lock;
 	struct amdgpu_bo *bo;
 	struct dma_buf *dmabuf;
-	struct hmm_range *range;
+	struct amdgpu_hmm_range *range;
 	struct list_head attachments;
 	/* protected by amdkfd_process_info.lock */
 	struct list_head validate_list;
@@ -106,11 +107,13 @@ struct amdgpu_kfd_dev {
 	bool init_complete;
 	struct work_struct reset_work;
 
-	/* HMM page migration MEMORY_DEVICE_PRIVATE mapping */
-	struct dev_pagemap pgmap;
-
 	/* Client for KFD BO GEM handle allocations */
 	struct drm_client_dev client;
+
+	/* HMM page migration MEMORY_DEVICE_PRIVATE mapping
+	 * Must be last --ends in a flexible-array member.
+	 */
+	struct dev_pagemap pgmap;
 };
 
 enum kgd_engine_type {
@@ -153,8 +156,10 @@ struct amdkfd_process_info {
 int amdgpu_amdkfd_init(void);
 void amdgpu_amdkfd_fini(void);
 
-void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool run_pm);
-int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool run_pm);
+void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool suspend_proc);
+int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool resume_proc);
+void amdgpu_amdkfd_suspend_process(struct amdgpu_device *adev);
+int amdgpu_amdkfd_resume_process(struct amdgpu_device *adev);
 void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
 			const void *ih_ring_entry);
 void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev);
@@ -192,7 +197,7 @@ int kfd_debugfs_kfd_mem_limits(struct seq_file *m, void *data);
 #if IS_ENABLED(CONFIG_HSA_AMD)
 bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm);
 struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f);
-int amdgpu_amdkfd_remove_fence_on_pt_pd_bos(struct amdgpu_bo *bo);
+void amdgpu_amdkfd_remove_all_eviction_fences(struct amdgpu_bo *bo);
 int amdgpu_amdkfd_evict_userptr(struct mmu_interval_notifier *mni,
 				unsigned long cur_seq, struct kgd_mem *mem);
 int amdgpu_amdkfd_bo_validate_and_fence(struct amdgpu_bo *bo,
@@ -212,9 +217,8 @@ struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f)
 }
 
 static inline
-int amdgpu_amdkfd_remove_fence_on_pt_pd_bos(struct amdgpu_bo *bo)
+void amdgpu_amdkfd_remove_all_eviction_fences(struct amdgpu_bo *bo)
 {
-	return 0;
 }
 
 static inline
@@ -254,11 +258,6 @@ int amdgpu_amdkfd_get_dmabuf_info(struct amdgpu_device *adev, int dma_buf_fd,
 				  uint64_t *bo_size, void *metadata_buffer,
 				  size_t buffer_size, uint32_t *metadata_size,
 				  uint32_t *flags, int8_t *xcp_id);
-uint8_t amdgpu_amdkfd_get_xgmi_hops_count(struct amdgpu_device *dst,
-					  struct amdgpu_device *src);
-int amdgpu_amdkfd_get_xgmi_bandwidth_mbytes(struct amdgpu_device *dst,
-					    struct amdgpu_device *src,
-					    bool is_min);
 int amdgpu_amdkfd_get_pcie_bandwidth_mbytes(struct amdgpu_device *adev, bool is_min);
 int amdgpu_amdkfd_send_close_event_drain_irq(struct amdgpu_device *adev,
 					uint32_t *payload);
@@ -299,14 +298,10 @@ bool amdgpu_amdkfd_compute_active(struct amdgpu_device *adev, uint32_t node_id);
 	(&((struct amdgpu_fpriv *)					\
 		((struct drm_file *)(drm_priv))->driver_priv)->vm)
 
-int amdgpu_amdkfd_gpuvm_set_vm_pasid(struct amdgpu_device *adev,
-				     struct amdgpu_vm *avm, u32 pasid);
 int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev,
 					struct amdgpu_vm *avm,
 					void **process_info,
 					struct dma_fence **ef);
-void amdgpu_amdkfd_gpuvm_release_process_vm(struct amdgpu_device *adev,
-					void *drm_priv);
 uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *drm_priv);
 size_t amdgpu_amdkfd_get_available_memory(struct amdgpu_device *adev,
 					uint8_t xcp_id);
@@ -420,18 +415,22 @@ struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf);
 bool kgd2kfd_device_init(struct kfd_dev *kfd,
 			 const struct kgd2kfd_shared_resources *gpu_resources);
 void kgd2kfd_device_exit(struct kfd_dev *kfd);
-void kgd2kfd_suspend(struct kfd_dev *kfd, bool run_pm);
-int kgd2kfd_resume(struct kfd_dev *kfd, bool run_pm);
+void kgd2kfd_suspend(struct kfd_dev *kfd, bool suspend_proc);
+int kgd2kfd_resume(struct kfd_dev *kfd, bool resume_proc);
+void kgd2kfd_suspend_process(struct kfd_dev *kfd);
+int kgd2kfd_resume_process(struct kfd_dev *kfd);
 int kgd2kfd_pre_reset(struct kfd_dev *kfd,
 		      struct amdgpu_reset_context *reset_context);
 int kgd2kfd_post_reset(struct kfd_dev *kfd);
 void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry);
 void kgd2kfd_set_sram_ecc_flag(struct kfd_dev *kfd);
 void kgd2kfd_smi_event_throttle(struct kfd_dev *kfd, uint64_t throttle_bitmask);
-int kgd2kfd_check_and_lock_kfd(void);
-void kgd2kfd_unlock_kfd(void);
+int kgd2kfd_check_and_lock_kfd(struct kfd_dev *kfd);
+void kgd2kfd_unlock_kfd(struct kfd_dev *kfd);
 int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id);
+int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd);
 int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id);
+int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd);
 bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id);
 bool kgd2kfd_vmfault_fast_path(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry,
 			       bool retry_fault);
@@ -463,11 +462,20 @@ static inline void kgd2kfd_device_exit(struct kfd_dev *kfd)
 {
 }
 
-static inline void kgd2kfd_suspend(struct kfd_dev *kfd, bool run_pm)
+static inline void kgd2kfd_suspend(struct kfd_dev *kfd, bool suspend_proc)
+{
+}
+
+static inline int kgd2kfd_resume(struct kfd_dev *kfd, bool resume_proc)
+{
+	return 0;
+}
+
+static inline void kgd2kfd_suspend_process(struct kfd_dev *kfd)
 {
 }
 
-static inline int kgd2kfd_resume(struct kfd_dev *kfd, bool run_pm)
+static inline int kgd2kfd_resume_process(struct kfd_dev *kfd)
 {
 	return 0;
 }
@@ -498,12 +506,12 @@ void kgd2kfd_smi_event_throttle(struct kfd_dev *kfd, uint64_t throttle_bitmask)
 {
 }
 
-static inline int kgd2kfd_check_and_lock_kfd(void)
+static inline int kgd2kfd_check_and_lock_kfd(struct kfd_dev *kfd)
 {
 	return 0;
 }
 
-static inline void kgd2kfd_unlock_kfd(void)
+static inline void kgd2kfd_unlock_kfd(struct kfd_dev *kfd)
 {
 }
 
@@ -512,11 +520,21 @@ static inline int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id)
 	return 0;
 }
 
+static inline int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd)
+{
+	return 0;
+}
+
 static inline int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id)
 {
 	return 0;
 }
 
+static inline int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd)
+{
+	return 0;
+}
+
 static inline bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id)
 {
 	return false;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c
index 8dfdb18197c4..7e9f7a280c1b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c
@@ -189,8 +189,9 @@ const struct kfd2kgd_calls aldebaran_kfd2kgd = {
 	.set_address_watch = kgd_gfx_aldebaran_set_address_watch,
 	.clear_address_watch = kgd_gfx_v9_clear_address_watch,
 	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
-	.build_grace_period_packet_info = kgd_gfx_v9_build_grace_period_packet_info,
+	.build_dequeue_wait_counts_packet_info = kgd_gfx_v9_build_dequeue_wait_counts_packet_info,
 	.program_trap_handler_settings = kgd_gfx_v9_program_trap_handler_settings,
 	.hqd_get_pq_addr = kgd_gfx_v9_hqd_get_pq_addr,
 	.hqd_reset = kgd_gfx_v9_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v9_hqd_sdma_get_doorbell
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c
index 9abf29b58ac7..1105a09e55dc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c
@@ -320,7 +320,7 @@ static void set_barrier_auto_waitcnt(struct amdgpu_device *adev, bool enable_wai
 	if (!down_read_trylock(&adev->reset_domain->sem))
 		return;
 
-	amdgpu_amdkfd_suspend(adev, false);
+	amdgpu_amdkfd_suspend(adev, true);
 
 	if (suspend_resume_compute_scheduler(adev, true))
 		goto out;
@@ -333,7 +333,7 @@ static void set_barrier_auto_waitcnt(struct amdgpu_device *adev, bool enable_wai
 out:
 	suspend_resume_compute_scheduler(adev, false);
 
-	amdgpu_amdkfd_resume(adev, false);
+	amdgpu_amdkfd_resume(adev, true);
 
 	up_read(&adev->reset_domain->sem);
 }
@@ -415,9 +415,10 @@ const struct kfd2kgd_calls arcturus_kfd2kgd = {
 	.set_address_watch = kgd_gfx_v9_set_address_watch,
 	.clear_address_watch = kgd_gfx_v9_clear_address_watch,
 	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
-	.build_grace_period_packet_info = kgd_gfx_v9_build_grace_period_packet_info,
+	.build_dequeue_wait_counts_packet_info = kgd_gfx_v9_build_dequeue_wait_counts_packet_info,
 	.get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy,
 	.program_trap_handler_settings = kgd_gfx_v9_program_trap_handler_settings,
 	.hqd_get_pq_addr = kgd_gfx_v9_hqd_get_pq_addr,
-	.hqd_reset = kgd_gfx_v9_hqd_reset
+	.hqd_reset = kgd_gfx_v9_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v9_hqd_sdma_get_doorbell
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gc_9_4_3.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gc_9_4_3.c
index e2ae714a700f..89a45a9218f3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gc_9_4_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gc_9_4_3.c
@@ -509,6 +509,17 @@ static uint32_t kgd_gfx_v9_4_3_clear_address_watch(struct amdgpu_device *adev,
 	return 0;
 }
 
+static uint32_t kgd_gfx_v9_4_3_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+						     int engine, int queue)
+{
+	uint32_t reg_offset = get_sdma_rlc_reg_offset(adev, engine, queue);
+	uint32_t status = RREG32(regSDMA_RLC0_CONTEXT_STATUS + reg_offset);
+	uint32_t doorbell_off = RREG32(regSDMA_RLC0_DOORBELL_OFFSET + reg_offset);
+	bool is_active = !!REG_GET_FIELD(status, SDMA_RLC0_CONTEXT_STATUS, SELECTED);
+
+	return is_active ? doorbell_off >> 2 : 0;
+}
+
 const struct kfd2kgd_calls gc_9_4_3_kfd2kgd = {
 	.program_sh_mem_settings = kgd_gfx_v9_program_sh_mem_settings,
 	.set_pasid_vmid_mapping = kgd_gfx_v9_4_3_set_pasid_vmid_mapping,
@@ -530,8 +541,8 @@ const struct kfd2kgd_calls gc_9_4_3_kfd2kgd = {
 	.get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy,
 	.program_trap_handler_settings =
 				kgd_gfx_v9_program_trap_handler_settings,
-	.build_grace_period_packet_info =
-				kgd_gfx_v9_build_grace_period_packet_info,
+	.build_dequeue_wait_counts_packet_info =
+				kgd_gfx_v9_build_dequeue_wait_counts_packet_info,
 	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
 	.enable_debug_trap = kgd_aldebaran_enable_debug_trap,
 	.disable_debug_trap = kgd_gfx_v9_4_3_disable_debug_trap,
@@ -543,5 +554,6 @@ const struct kfd2kgd_calls gc_9_4_3_kfd2kgd = {
 	.set_address_watch = kgd_gfx_v9_4_3_set_address_watch,
 	.clear_address_watch = kgd_gfx_v9_4_3_clear_address_watch,
 	.hqd_get_pq_addr = kgd_gfx_v9_hqd_get_pq_addr,
-	.hqd_reset = kgd_gfx_v9_hqd_reset
+	.hqd_reset = kgd_gfx_v9_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v9_4_3_hqd_sdma_get_doorbell
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
index 62176d607bef..0239114fb6c4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
@@ -352,7 +352,7 @@ static int kgd_hqd_dump(struct amdgpu_device *adev,
 		(*dump)[i++][1] = RREG32_SOC15_IP(GC, addr);		\
 	} while (0)
 
-	*dump = kmalloc(HQD_N_REGS*2*sizeof(uint32_t), GFP_KERNEL);
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
 	if (*dump == NULL)
 		return -ENOMEM;
 
@@ -449,7 +449,7 @@ static int kgd_hqd_sdma_dump(struct amdgpu_device *adev,
 #undef HQD_N_REGS
 #define HQD_N_REGS (19+6+7+10)
 
-	*dump = kmalloc(HQD_N_REGS*2*sizeof(uint32_t), GFP_KERNEL);
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
 	if (*dump == NULL)
 		return -ENOMEM;
 
@@ -1021,25 +1021,25 @@ void kgd_gfx_v10_get_iq_wait_times(struct amdgpu_device *adev,
 	*wait_times = RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_IQ_WAIT_TIME2));
 }
 
-void kgd_gfx_v10_build_grace_period_packet_info(struct amdgpu_device *adev,
+void kgd_gfx_v10_build_dequeue_wait_counts_packet_info(struct amdgpu_device *adev,
 						uint32_t wait_times,
-						uint32_t grace_period,
+						uint32_t sch_wave,
+						uint32_t que_sleep,
 						uint32_t *reg_offset,
 						uint32_t *reg_data)
 {
 	*reg_data = wait_times;
 
-	/*
-	 * The CP cannont handle a 0 grace period input and will result in
-	 * an infinite grace period being set so set to 1 to prevent this.
-	 */
-	if (grace_period == 0)
-		grace_period = 1;
-
-	*reg_data = REG_SET_FIELD(*reg_data,
-			CP_IQ_WAIT_TIME2,
-			SCH_WAVE,
-			grace_period);
+	if (sch_wave)
+		*reg_data = REG_SET_FIELD(*reg_data,
+				CP_IQ_WAIT_TIME2,
+				SCH_WAVE,
+				sch_wave);
+	if (que_sleep)
+		*reg_data = REG_SET_FIELD(*reg_data,
+				CP_IQ_WAIT_TIME2,
+				QUE_SLEEP,
+				que_sleep);
 
 	*reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_IQ_WAIT_TIME2);
 }
@@ -1084,6 +1084,12 @@ uint64_t kgd_gfx_v10_hqd_reset(struct amdgpu_device *adev,
 	return 0;
 }
 
+uint32_t kgd_gfx_v10_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					   int engine, int queue)
+{
+	return 0;
+}
+
 const struct kfd2kgd_calls gfx_v10_kfd2kgd = {
 	.program_sh_mem_settings = kgd_program_sh_mem_settings,
 	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
@@ -1109,8 +1115,9 @@ const struct kfd2kgd_calls gfx_v10_kfd2kgd = {
 	.set_address_watch = kgd_gfx_v10_set_address_watch,
 	.clear_address_watch = kgd_gfx_v10_clear_address_watch,
 	.get_iq_wait_times = kgd_gfx_v10_get_iq_wait_times,
-	.build_grace_period_packet_info = kgd_gfx_v10_build_grace_period_packet_info,
+	.build_dequeue_wait_counts_packet_info = kgd_gfx_v10_build_dequeue_wait_counts_packet_info,
 	.program_trap_handler_settings = program_trap_handler_settings,
 	.hqd_get_pq_addr = kgd_gfx_v10_hqd_get_pq_addr,
-	.hqd_reset = kgd_gfx_v10_hqd_reset
+	.hqd_reset = kgd_gfx_v10_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v10_hqd_sdma_get_doorbell
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h
index 9efd2dd4fdd7..a4c607c88178 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h
@@ -51,9 +51,10 @@ uint32_t kgd_gfx_v10_clear_address_watch(struct amdgpu_device *adev,
 void kgd_gfx_v10_get_iq_wait_times(struct amdgpu_device *adev,
 				uint32_t *wait_times,
 				uint32_t inst);
-void kgd_gfx_v10_build_grace_period_packet_info(struct amdgpu_device *adev,
+void kgd_gfx_v10_build_dequeue_wait_counts_packet_info(struct amdgpu_device *adev,
 					       uint32_t wait_times,
-					       uint32_t grace_period,
+					       uint32_t sch_wave,
+					       uint32_t que_sleep,
 					       uint32_t *reg_offset,
 					       uint32_t *reg_data);
 uint64_t kgd_gfx_v10_hqd_get_pq_addr(struct amdgpu_device *adev,
@@ -65,3 +66,5 @@ uint64_t kgd_gfx_v10_hqd_reset(struct amdgpu_device *adev,
 			      uint32_t queue_id,
 			      uint32_t inst,
 			      unsigned int utimeout);
+uint32_t kgd_gfx_v10_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					   int engine, int queue);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c
index c718bedda0ca..f2278a0937ff 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c
@@ -338,7 +338,7 @@ static int hqd_dump_v10_3(struct amdgpu_device *adev,
 		(*dump)[i++][1] = RREG32_SOC15_IP(GC, addr);		\
 	} while (0)
 
-	*dump = kmalloc(HQD_N_REGS*2*sizeof(uint32_t), GFP_KERNEL);
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
 	if (*dump == NULL)
 		return -ENOMEM;
 
@@ -435,7 +435,7 @@ static int hqd_sdma_dump_v10_3(struct amdgpu_device *adev,
 #undef HQD_N_REGS
 #define HQD_N_REGS (19+6+7+12)
 
-	*dump = kmalloc(HQD_N_REGS*2*sizeof(uint32_t), GFP_KERNEL);
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
 	if (*dump == NULL)
 		return -ENOMEM;
 
@@ -673,7 +673,7 @@ const struct kfd2kgd_calls gfx_v10_3_kfd2kgd = {
 	.set_vm_context_page_table_base = set_vm_context_page_table_base_v10_3,
 	.program_trap_handler_settings = program_trap_handler_settings_v10_3,
 	.get_iq_wait_times = kgd_gfx_v10_get_iq_wait_times,
-	.build_grace_period_packet_info = kgd_gfx_v10_build_grace_period_packet_info,
+	.build_dequeue_wait_counts_packet_info = kgd_gfx_v10_build_dequeue_wait_counts_packet_info,
 	.enable_debug_trap = kgd_gfx_v10_enable_debug_trap,
 	.disable_debug_trap = kgd_gfx_v10_disable_debug_trap,
 	.validate_trap_override_request = kgd_gfx_v10_validate_trap_override_request,
@@ -682,5 +682,6 @@ const struct kfd2kgd_calls gfx_v10_3_kfd2kgd = {
 	.set_address_watch = kgd_gfx_v10_set_address_watch,
 	.clear_address_watch = kgd_gfx_v10_clear_address_watch,
 	.hqd_get_pq_addr = kgd_gfx_v10_hqd_get_pq_addr,
-	.hqd_reset = kgd_gfx_v10_hqd_reset
+	.hqd_reset = kgd_gfx_v10_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v10_hqd_sdma_get_doorbell
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c
index a4ba49cb22db..aaccf0b9947d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c
@@ -323,7 +323,7 @@ static int hqd_dump_v11(struct amdgpu_device *adev,
 		(*dump)[i++][1] = RREG32(addr);		\
 	} while (0)
 
-	*dump = kmalloc(HQD_N_REGS*2*sizeof(uint32_t), GFP_KERNEL);
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
 	if (*dump == NULL)
 		return -ENOMEM;
 
@@ -420,7 +420,7 @@ static int hqd_sdma_dump_v11(struct amdgpu_device *adev,
 #undef HQD_N_REGS
 #define HQD_N_REGS (7+11+1+12+12)
 
-	*dump = kmalloc(HQD_N_REGS*2*sizeof(uint32_t), GFP_KERNEL);
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
 	if (*dump == NULL)
 		return -ENOMEM;
 
@@ -800,6 +800,12 @@ static uint64_t kgd_gfx_v11_hqd_reset(struct amdgpu_device *adev,
 	return 0;
 }
 
+static uint32_t kgd_gfx_v11_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+						  int engine, int queue)
+{
+	return 0;
+}
+
 const struct kfd2kgd_calls gfx_v11_kfd2kgd = {
 	.program_sh_mem_settings = program_sh_mem_settings_v11,
 	.set_pasid_vmid_mapping = set_pasid_vmid_mapping_v11,
@@ -824,5 +830,6 @@ const struct kfd2kgd_calls gfx_v11_kfd2kgd = {
 	.set_address_watch = kgd_gfx_v11_set_address_watch,
 	.clear_address_watch = kgd_gfx_v11_clear_address_watch,
 	.hqd_get_pq_addr = kgd_gfx_v11_hqd_get_pq_addr,
-	.hqd_reset = kgd_gfx_v11_hqd_reset
+	.hqd_reset = kgd_gfx_v11_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v11_hqd_sdma_get_doorbell
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v12.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v12.c
index 0dfe7093bd8a..e0ceab400b2d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v12.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v12.c
@@ -115,7 +115,7 @@ static int hqd_dump_v12(struct amdgpu_device *adev,
 		(*dump)[i++][1] = RREG32(addr);		\
 	} while (0)
 
-	*dump = kmalloc(HQD_N_REGS*2*sizeof(uint32_t), GFP_KERNEL);
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
 	if (*dump == NULL)
 		return -ENOMEM;
 
@@ -146,7 +146,7 @@ static int hqd_sdma_dump_v12(struct amdgpu_device *adev,
 #undef HQD_N_REGS
 #define HQD_N_REGS (last_reg - first_reg + 1)
 
-	*dump = kmalloc(HQD_N_REGS*2*sizeof(uint32_t), GFP_KERNEL);
+	*dump = kmalloc_array(HQD_N_REGS, sizeof(**dump), GFP_KERNEL);
 	if (*dump == NULL)
 		return -ENOMEM;
 
@@ -361,6 +361,12 @@ static uint32_t kgd_gfx_v12_clear_address_watch(struct amdgpu_device *adev,
 	return 0;
 }
 
+static uint32_t kgd_gfx_v12_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+						 int engine, int queue)
+{
+	return 0;
+}
+
 const struct kfd2kgd_calls gfx_v12_kfd2kgd = {
 	.init_interrupts = init_interrupts_v12,
 	.hqd_dump = hqd_dump_v12,
@@ -374,4 +380,5 @@ const struct kfd2kgd_calls gfx_v12_kfd2kgd = {
 	.set_wave_launch_mode = kgd_gfx_v12_set_wave_launch_mode,
 	.set_address_watch = kgd_gfx_v12_set_address_watch,
 	.clear_address_watch = kgd_gfx_v12_clear_address_watch,
+	.hqd_sdma_get_doorbell = kgd_gfx_v12_hqd_sdma_get_doorbell
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
index ca4a6b82817f..df77558e03ef 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
@@ -561,6 +561,13 @@ static uint32_t read_vmid_from_vmfault_reg(struct amdgpu_device *adev)
 	return REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
 }
 
+static uint32_t kgd_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					  int engine, int queue)
+
+{
+	return 0;
+}
+
 const struct kfd2kgd_calls gfx_v7_kfd2kgd = {
 	.program_sh_mem_settings = kgd_program_sh_mem_settings,
 	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
@@ -578,4 +585,5 @@ const struct kfd2kgd_calls gfx_v7_kfd2kgd = {
 	.set_scratch_backing_va = set_scratch_backing_va,
 	.set_vm_context_page_table_base = set_vm_context_page_table_base,
 	.read_vmid_from_vmfault_reg = read_vmid_from_vmfault_reg,
+	.hqd_sdma_get_doorbell = kgd_hqd_sdma_get_doorbell,
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
index 0f3e2944edd7..e68c0fa8d751 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
@@ -582,6 +582,13 @@ static void set_vm_context_page_table_base(struct amdgpu_device *adev,
 			lower_32_bits(page_table_base));
 }
 
+static uint32_t kgd_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					  int engine, int queue)
+
+{
+	return 0;
+}
+
 const struct kfd2kgd_calls gfx_v8_kfd2kgd = {
 	.program_sh_mem_settings = kgd_program_sh_mem_settings,
 	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
@@ -599,4 +606,5 @@ const struct kfd2kgd_calls gfx_v8_kfd2kgd = {
 			get_atc_vmid_pasid_mapping_info,
 	.set_scratch_backing_va = set_scratch_backing_va,
 	.set_vm_context_page_table_base = set_vm_context_page_table_base,
+	.hqd_sdma_get_doorbell = kgd_hqd_sdma_get_doorbell,
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
index 441568163e20..088d09cc7a72 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
@@ -1077,25 +1077,25 @@ void kgd_gfx_v9_get_cu_occupancy(struct amdgpu_device *adev,
 				adev->gfx.cu_info.max_waves_per_simd;
 }
 
-void kgd_gfx_v9_build_grace_period_packet_info(struct amdgpu_device *adev,
+void kgd_gfx_v9_build_dequeue_wait_counts_packet_info(struct amdgpu_device *adev,
 		uint32_t wait_times,
-		uint32_t grace_period,
+		uint32_t sch_wave,
+		uint32_t que_sleep,
 		uint32_t *reg_offset,
 		uint32_t *reg_data)
 {
 	*reg_data = wait_times;
 
-	/*
-	 * The CP cannot handle a 0 grace period input and will result in
-	 * an infinite grace period being set so set to 1 to prevent this.
-	 */
-	if (grace_period == 0)
-		grace_period = 1;
-
-	*reg_data = REG_SET_FIELD(*reg_data,
-			CP_IQ_WAIT_TIME2,
-			SCH_WAVE,
-			grace_period);
+	if (sch_wave)
+		*reg_data = REG_SET_FIELD(*reg_data,
+				CP_IQ_WAIT_TIME2,
+				SCH_WAVE,
+				sch_wave);
+	if (que_sleep)
+		*reg_data = REG_SET_FIELD(*reg_data,
+				CP_IQ_WAIT_TIME2,
+				QUE_SLEEP,
+				que_sleep);
 
 	*reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_IQ_WAIT_TIME2);
 }
@@ -1131,9 +1131,6 @@ uint64_t kgd_gfx_v9_hqd_get_pq_addr(struct amdgpu_device *adev,
 	uint32_t low, high;
 	uint64_t queue_addr = 0;
 
-	if (!amdgpu_gpu_recovery)
-		return 0;
-
 	kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
 	amdgpu_gfx_rlc_enter_safe_mode(adev, inst);
 
@@ -1182,9 +1179,6 @@ uint64_t kgd_gfx_v9_hqd_reset(struct amdgpu_device *adev,
 	uint32_t low, high, pipe_reset_data = 0;
 	uint64_t queue_addr = 0;
 
-	if (!amdgpu_gpu_recovery)
-		return 0;
-
 	kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
 	amdgpu_gfx_rlc_enter_safe_mode(adev, inst);
 
@@ -1229,6 +1223,13 @@ unlock_out:
 	return queue_addr;
 }
 
+uint32_t kgd_gfx_v9_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					  int engine, int queue)
+
+{
+	return 0;
+}
+
 const struct kfd2kgd_calls gfx_v9_kfd2kgd = {
 	.program_sh_mem_settings = kgd_gfx_v9_program_sh_mem_settings,
 	.set_pasid_vmid_mapping = kgd_gfx_v9_set_pasid_vmid_mapping,
@@ -1254,9 +1255,10 @@ const struct kfd2kgd_calls gfx_v9_kfd2kgd = {
 	.set_address_watch = kgd_gfx_v9_set_address_watch,
 	.clear_address_watch = kgd_gfx_v9_clear_address_watch,
 	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
-	.build_grace_period_packet_info = kgd_gfx_v9_build_grace_period_packet_info,
+	.build_dequeue_wait_counts_packet_info = kgd_gfx_v9_build_dequeue_wait_counts_packet_info,
 	.get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy,
 	.program_trap_handler_settings = kgd_gfx_v9_program_trap_handler_settings,
 	.hqd_get_pq_addr = kgd_gfx_v9_hqd_get_pq_addr,
-	.hqd_reset = kgd_gfx_v9_hqd_reset
+	.hqd_reset = kgd_gfx_v9_hqd_reset,
+	.hqd_sdma_get_doorbell = kgd_gfx_v9_hqd_sdma_get_doorbell
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h
index b6a91a552aa4..704452ca62f8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h
@@ -97,9 +97,10 @@ uint32_t kgd_gfx_v9_clear_address_watch(struct amdgpu_device *adev,
 void kgd_gfx_v9_get_iq_wait_times(struct amdgpu_device *adev,
 				uint32_t *wait_times,
 				uint32_t inst);
-void kgd_gfx_v9_build_grace_period_packet_info(struct amdgpu_device *adev,
+void kgd_gfx_v9_build_dequeue_wait_counts_packet_info(struct amdgpu_device *adev,
 					       uint32_t wait_times,
-					       uint32_t grace_period,
+					       uint32_t sch_wave,
+					       uint32_t que_sleep,
 					       uint32_t *reg_offset,
 					       uint32_t *reg_data);
 uint64_t kgd_gfx_v9_hqd_get_pq_addr(struct amdgpu_device *adev,
@@ -111,3 +112,5 @@ uint64_t kgd_gfx_v9_hqd_reset(struct amdgpu_device *adev,
 			      uint32_t queue_id,
 			      uint32_t inst,
 			      unsigned int utimeout);
+uint32_t kgd_gfx_v9_hqd_sdma_get_doorbell(struct amdgpu_device *adev,
+					  int engine, int queue);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index 1e998f972c30..b1c24c8fa686 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -197,7 +197,7 @@ int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev,
 			return -EINVAL;
 
 		vram_size = KFD_XCP_MEMORY_SIZE(adev, xcp_id);
-		if (adev->flags & AMD_IS_APU) {
+		if (adev->apu_prefer_gtt) {
 			system_mem_needed = size;
 			ttm_mem_needed = size;
 		}
@@ -213,19 +213,35 @@ int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev,
 	spin_lock(&kfd_mem_limit.mem_limit_lock);
 
 	if (kfd_mem_limit.system_mem_used + system_mem_needed >
-	    kfd_mem_limit.max_system_mem_limit)
+	    kfd_mem_limit.max_system_mem_limit) {
 		pr_debug("Set no_system_mem_limit=1 if using shared memory\n");
+		if (!no_system_mem_limit) {
+			ret = -ENOMEM;
+			goto release;
+		}
+	}
 
-	if ((kfd_mem_limit.system_mem_used + system_mem_needed >
-	     kfd_mem_limit.max_system_mem_limit && !no_system_mem_limit) ||
-	    (kfd_mem_limit.ttm_mem_used + ttm_mem_needed >
-	     kfd_mem_limit.max_ttm_mem_limit) ||
-	    (adev && xcp_id >= 0 && adev->kfd.vram_used[xcp_id] + vram_needed >
-	     vram_size - reserved_for_pt - reserved_for_ras - atomic64_read(&adev->vram_pin_size))) {
+	if (kfd_mem_limit.ttm_mem_used + ttm_mem_needed >
+		kfd_mem_limit.max_ttm_mem_limit) {
 		ret = -ENOMEM;
 		goto release;
 	}
 
+	/*if is_app_apu is false and apu_prefer_gtt is true, it is an APU with
+	 * carve out < gtt. In that case, VRAM allocation will go to gtt domain, skip
+	 * VRAM check since ttm_mem_limit check already cover this allocation
+	 */
+
+	if (adev && xcp_id >= 0 && (!adev->apu_prefer_gtt || adev->gmc.is_app_apu)) {
+		uint64_t vram_available =
+			vram_size - reserved_for_pt - reserved_for_ras -
+			atomic64_read(&adev->vram_pin_size);
+		if (adev->kfd.vram_used[xcp_id] + vram_needed > vram_available) {
+			ret = -ENOMEM;
+			goto release;
+		}
+	}
+
 	/* Update memory accounting by decreasing available system
 	 * memory, TTM memory and GPU memory as computed above
 	 */
@@ -234,7 +250,7 @@ int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev,
 	if (adev && xcp_id >= 0) {
 		adev->kfd.vram_used[xcp_id] += vram_needed;
 		adev->kfd.vram_used_aligned[xcp_id] +=
-				(adev->flags & AMD_IS_APU) ?
+				adev->apu_prefer_gtt ?
 				vram_needed :
 				ALIGN(vram_needed, VRAM_AVAILABLITY_ALIGN);
 	}
@@ -262,7 +278,7 @@ void amdgpu_amdkfd_unreserve_mem_limit(struct amdgpu_device *adev,
 
 		if (adev) {
 			adev->kfd.vram_used[xcp_id] -= size;
-			if (adev->flags & AMD_IS_APU) {
+			if (adev->apu_prefer_gtt) {
 				adev->kfd.vram_used_aligned[xcp_id] -= size;
 				kfd_mem_limit.system_mem_used -= size;
 				kfd_mem_limit.ttm_mem_used -= size;
@@ -370,40 +386,32 @@ static int amdgpu_amdkfd_remove_eviction_fence(struct amdgpu_bo *bo,
 	return 0;
 }
 
-int amdgpu_amdkfd_remove_fence_on_pt_pd_bos(struct amdgpu_bo *bo)
+/**
+ * amdgpu_amdkfd_remove_all_eviction_fences - Remove all eviction fences
+ * @bo: the BO where to remove the evictions fences from.
+ *
+ * This functions should only be used on release when all references to the BO
+ * are already dropped. We remove the eviction fence from the private copy of
+ * the dma_resv object here since that is what is used during release to
+ * determine of the BO is idle or not.
+ */
+void amdgpu_amdkfd_remove_all_eviction_fences(struct amdgpu_bo *bo)
 {
-	struct amdgpu_bo *root = bo;
-	struct amdgpu_vm_bo_base *vm_bo;
-	struct amdgpu_vm *vm;
-	struct amdkfd_process_info *info;
-	struct amdgpu_amdkfd_fence *ef;
-	int ret;
-
-	/* we can always get vm_bo from root PD bo.*/
-	while (root->parent)
-		root = root->parent;
+	struct dma_resv *resv = &bo->tbo.base._resv;
+	struct dma_fence *fence, *stub;
+	struct dma_resv_iter cursor;
 
-	vm_bo = root->vm_bo;
-	if (!vm_bo)
-		return 0;
+	dma_resv_assert_held(resv);
 
-	vm = vm_bo->vm;
-	if (!vm)
-		return 0;
-
-	info = vm->process_info;
-	if (!info || !info->eviction_fence)
-		return 0;
-
-	ef = container_of(dma_fence_get(&info->eviction_fence->base),
-			struct amdgpu_amdkfd_fence, base);
-
-	BUG_ON(!dma_resv_trylock(bo->tbo.base.resv));
-	ret = amdgpu_amdkfd_remove_eviction_fence(bo, ef);
-	dma_resv_unlock(bo->tbo.base.resv);
+	stub = dma_fence_get_stub();
+	dma_resv_for_each_fence(&cursor, resv, DMA_RESV_USAGE_BOOKKEEP, fence) {
+		if (!to_amdgpu_amdkfd_fence(fence))
+			continue;
 
-	dma_fence_put(&ef->base);
-	return ret;
+		dma_resv_replace_fences(resv, fence->context, stub,
+					DMA_RESV_USAGE_BOOKKEEP);
+	}
+	dma_fence_put(stub);
 }
 
 static int amdgpu_amdkfd_bo_validate(struct amdgpu_bo *bo, uint32_t domain,
@@ -499,10 +507,11 @@ static int vm_update_pds(struct amdgpu_vm *vm, struct amdgpu_sync *sync)
 	if (ret)
 		return ret;
 
-	return amdgpu_sync_fence(sync, vm->last_update);
+	return amdgpu_sync_fence(sync, vm->last_update, GFP_KERNEL);
 }
 
-static uint64_t get_pte_flags(struct amdgpu_device *adev, struct kgd_mem *mem)
+static uint64_t get_pte_flags(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			      struct kgd_mem *mem)
 {
 	uint32_t mapping_flags = AMDGPU_VM_PAGE_READABLE |
 				 AMDGPU_VM_MTYPE_DEFAULT;
@@ -512,7 +521,7 @@ static uint64_t get_pte_flags(struct amdgpu_device *adev, struct kgd_mem *mem)
 	if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_EXECUTABLE)
 		mapping_flags |= AMDGPU_VM_PAGE_EXECUTABLE;
 
-	return amdgpu_gem_va_map_flags(adev, mapping_flags);
+	return mapping_flags;
 }
 
 /**
@@ -603,12 +612,6 @@ kfd_mem_dmamap_dmabuf(struct kfd_mem_attachment *attachment)
 {
 	struct ttm_operation_ctx ctx = {.interruptible = true};
 	struct amdgpu_bo *bo = attachment->bo_va->base.bo;
-	int ret;
-
-	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU);
-	ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
-	if (ret)
-		return ret;
 
 	amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
 	return ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
@@ -890,7 +893,7 @@ static int kfd_mem_attach(struct amdgpu_device *adev, struct kgd_mem *mem,
 	 * if peer device has large BAR. In contrast, access over xGMI is
 	 * allowed for both small and large BAR configurations of peer device
 	 */
-	if ((adev != bo_adev && !(adev->flags & AMD_IS_APU)) &&
+	if ((adev != bo_adev && !adev->apu_prefer_gtt) &&
 	    ((mem->domain == AMDGPU_GEM_DOMAIN_VRAM) ||
 	     (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL) ||
 	     (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP))) {
@@ -975,7 +978,7 @@ static int kfd_mem_attach(struct amdgpu_device *adev, struct kgd_mem *mem,
 			goto unwind;
 		}
 		attachment[i]->va = va;
-		attachment[i]->pte_flags = get_pte_flags(adev, mem);
+		attachment[i]->pte_flags = get_pte_flags(adev, vm, mem);
 		attachment[i]->adev = adev;
 		list_add(&attachment[i]->list, &mem->attachments);
 
@@ -1054,7 +1057,7 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr,
 	struct amdkfd_process_info *process_info = mem->process_info;
 	struct amdgpu_bo *bo = mem->bo;
 	struct ttm_operation_ctx ctx = { true, false };
-	struct hmm_range *range;
+	struct amdgpu_hmm_range *range;
 	int ret = 0;
 
 	mutex_lock(&process_info->lock);
@@ -1086,8 +1089,15 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr,
 		return 0;
 	}
 
-	ret = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages, &range);
+	range = amdgpu_hmm_range_alloc(NULL);
+	if (unlikely(!range)) {
+		ret = -ENOMEM;
+		goto unregister_out;
+	}
+
+	ret = amdgpu_ttm_tt_get_user_pages(bo, range);
 	if (ret) {
+		amdgpu_hmm_range_free(range);
 		if (ret == -EAGAIN)
 			pr_debug("Failed to get user pages, try again\n");
 		else
@@ -1100,6 +1110,9 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr,
 		pr_err("%s: Failed to reserve BO\n", __func__);
 		goto release_out;
 	}
+
+	amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, range);
+
 	amdgpu_bo_placement_from_domain(bo, mem->domain);
 	ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 	if (ret)
@@ -1107,7 +1120,7 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr,
 	amdgpu_bo_unreserve(bo);
 
 release_out:
-	amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, range);
+	amdgpu_hmm_range_free(range);
 unregister_out:
 	if (ret)
 		amdgpu_hmm_unregister(bo);
@@ -1261,9 +1274,13 @@ static int unmap_bo_from_gpuvm(struct kgd_mem *mem,
 
 	(void)amdgpu_vm_bo_unmap(adev, bo_va, entry->va);
 
+	/* VM entity stopped if process killed, don't clear freed pt bo */
+	if (!amdgpu_vm_ready(vm))
+		return 0;
+
 	(void)amdgpu_vm_clear_freed(adev, vm, &bo_va->last_pt_update);
 
-	(void)amdgpu_sync_fence(sync, bo_va->last_pt_update);
+	(void)amdgpu_sync_fence(sync, bo_va->last_pt_update, GFP_KERNEL);
 
 	return 0;
 }
@@ -1287,7 +1304,7 @@ static int update_gpuvm_pte(struct kgd_mem *mem,
 		return ret;
 	}
 
-	return amdgpu_sync_fence(sync, bo_va->last_pt_update);
+	return amdgpu_sync_fence(sync, bo_va->last_pt_update, GFP_KERNEL);
 }
 
 static int map_bo_to_gpuvm(struct kgd_mem *mem,
@@ -1529,27 +1546,6 @@ static void amdgpu_amdkfd_gpuvm_unpin_bo(struct amdgpu_bo *bo)
 	amdgpu_bo_unreserve(bo);
 }
 
-int amdgpu_amdkfd_gpuvm_set_vm_pasid(struct amdgpu_device *adev,
-				     struct amdgpu_vm *avm, u32 pasid)
-
-{
-	int ret;
-
-	/* Free the original amdgpu allocated pasid,
-	 * will be replaced with kfd allocated pasid.
-	 */
-	if (avm->pasid) {
-		amdgpu_pasid_free(avm->pasid);
-		amdgpu_vm_set_pasid(adev, avm, 0);
-	}
-
-	ret = amdgpu_vm_set_pasid(adev, avm, pasid);
-	if (ret)
-		return ret;
-
-	return 0;
-}
-
 int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev,
 					   struct amdgpu_vm *avm,
 					   void **process_info,
@@ -1607,27 +1603,6 @@ void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
 	}
 }
 
-void amdgpu_amdkfd_gpuvm_release_process_vm(struct amdgpu_device *adev,
-					    void *drm_priv)
-{
-	struct amdgpu_vm *avm;
-
-	if (WARN_ON(!adev || !drm_priv))
-		return;
-
-	avm = drm_priv_to_vm(drm_priv);
-
-	pr_debug("Releasing process vm %p\n", avm);
-
-	/* The original pasid of amdgpu vm has already been
-	 * released during making a amdgpu vm to a compute vm
-	 * The current pasid is managed by kfd and will be
-	 * released on kfd process destroy. Set amdgpu pasid
-	 * to 0 to avoid duplicate release.
-	 */
-	amdgpu_vm_release_compute(adev, avm);
-}
-
 uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *drm_priv)
 {
 	struct amdgpu_vm *avm = drm_priv_to_vm(drm_priv);
@@ -1682,13 +1657,17 @@ size_t amdgpu_amdkfd_get_available_memory(struct amdgpu_device *adev,
 	uint64_t vram_available, system_mem_available, ttm_mem_available;
 
 	spin_lock(&kfd_mem_limit.mem_limit_lock);
-	vram_available = KFD_XCP_MEMORY_SIZE(adev, xcp_id)
-		- adev->kfd.vram_used_aligned[xcp_id]
-		- atomic64_read(&adev->vram_pin_size)
-		- reserved_for_pt
-		- reserved_for_ras;
+	if (adev->apu_prefer_gtt && !adev->gmc.is_app_apu)
+		vram_available = KFD_XCP_MEMORY_SIZE(adev, xcp_id)
+			- adev->kfd.vram_used_aligned[xcp_id];
+	else
+		vram_available = KFD_XCP_MEMORY_SIZE(adev, xcp_id)
+			- adev->kfd.vram_used_aligned[xcp_id]
+			- atomic64_read(&adev->vram_pin_size)
+			- reserved_for_pt
+			- reserved_for_ras;
 
-	if (adev->flags & AMD_IS_APU) {
+	if (adev->apu_prefer_gtt) {
 		system_mem_available = no_system_mem_limit ?
 					kfd_mem_limit.max_system_mem_limit :
 					kfd_mem_limit.max_system_mem_limit -
@@ -1736,7 +1715,7 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
 	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) {
 		domain = alloc_domain = AMDGPU_GEM_DOMAIN_VRAM;
 
-		if (adev->flags & AMD_IS_APU) {
+		if (adev->apu_prefer_gtt) {
 			domain = AMDGPU_GEM_DOMAIN_GTT;
 			alloc_domain = AMDGPU_GEM_DOMAIN_GTT;
 			alloc_flags = 0;
@@ -1948,7 +1927,7 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
 	if (amdgpu_ttm_tt_get_usermm(mem->bo->tbo.ttm)) {
 		amdgpu_hmm_unregister(mem->bo);
 		mutex_lock(&process_info->notifier_lock);
-		amdgpu_ttm_tt_discard_user_pages(mem->bo->tbo.ttm, mem->range);
+		amdgpu_hmm_range_free(mem->range);
 		mutex_unlock(&process_info->notifier_lock);
 	}
 
@@ -1986,9 +1965,7 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
 	 */
 	if (size) {
 		if (!is_imported &&
-		   (mem->bo->preferred_domains == AMDGPU_GEM_DOMAIN_VRAM ||
-		   ((adev->flags & AMD_IS_APU) &&
-		    mem->bo->preferred_domains == AMDGPU_GEM_DOMAIN_GTT)))
+		   mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM)
 			*size = bo_size;
 		else
 			*size = 0;
@@ -2361,10 +2338,9 @@ void amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(struct kgd_mem *mem)
 int amdgpu_amdkfd_gpuvm_get_vm_fault_info(struct amdgpu_device *adev,
 					  struct kfd_vm_fault_info *mem)
 {
-	if (atomic_read(&adev->gmc.vm_fault_info_updated) == 1) {
+	if (atomic_read_acquire(&adev->gmc.vm_fault_info_updated) == 1) {
 		*mem = *adev->gmc.vm_fault_info;
-		mb(); /* make sure read happened */
-		atomic_set(&adev->gmc.vm_fault_info_updated, 0);
+		atomic_set_release(&adev->gmc.vm_fault_info_updated, 0);
 	}
 	return 0;
 }
@@ -2414,7 +2390,7 @@ static int import_obj_create(struct amdgpu_device *adev,
 	(*mem)->bo = bo;
 	(*mem)->va = va;
 	(*mem)->domain = (bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) &&
-			 !(adev->flags & AMD_IS_APU) ?
+			 !adev->apu_prefer_gtt ?
 			 AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT;
 
 	(*mem)->mapped_to_gpu_memory = 0;
@@ -2575,7 +2551,7 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
 
 		bo = mem->bo;
 
-		amdgpu_ttm_tt_discard_user_pages(bo->tbo.ttm, mem->range);
+		amdgpu_hmm_range_free(mem->range);
 		mem->range = NULL;
 
 		/* BO reservations and getting user pages (hmm_range_fault)
@@ -2599,10 +2575,14 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
 			}
 		}
 
+		mem->range = amdgpu_hmm_range_alloc(NULL);
+		if (unlikely(!mem->range))
+			return -ENOMEM;
 		/* Get updated user pages */
-		ret = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages,
-						   &mem->range);
+		ret = amdgpu_ttm_tt_get_user_pages(bo, mem->range);
 		if (ret) {
+			amdgpu_hmm_range_free(mem->range);
+			mem->range = NULL;
 			pr_debug("Failed %d to get user pages\n", ret);
 
 			/* Return -EFAULT bad address error as success. It will
@@ -2615,9 +2595,28 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
 			if (ret != -EFAULT)
 				return ret;
 
+			/* If applications unmap memory before destroying the userptr
+			 * from the KFD, trigger a segmentation fault in VM debug mode.
+			 */
+			if (amdgpu_ttm_adev(bo->tbo.bdev)->debug_vm_userptr) {
+				struct kfd_process *p;
+
+				pr_err("Pid %d unmapped memory before destroying userptr at GPU addr 0x%llx\n",
+								pid_nr(process_info->pid), mem->va);
+
+				// Send GPU VM fault to user space
+				p = kfd_lookup_process_by_pid(process_info->pid);
+				if (p) {
+					kfd_signal_vm_fault_event_with_userptr(p, mem->va);
+					kfd_unref_process(p);
+				}
+			}
+
 			ret = 0;
 		}
 
+		amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, mem->range);
+
 		mutex_lock(&process_info->notifier_lock);
 
 		/* Mark the BO as valid unless it was invalidated
@@ -2756,8 +2755,8 @@ static int confirm_valid_user_pages_locked(struct amdkfd_process_info *process_i
 			continue;
 
 		/* Only check mem with hmm range associated */
-		valid = amdgpu_ttm_tt_get_user_pages_done(
-					mem->bo->tbo.ttm, mem->range);
+		valid = amdgpu_hmm_range_valid(mem->range);
+		amdgpu_hmm_range_free(mem->range);
 
 		mem->range = NULL;
 		if (!valid) {
@@ -2969,7 +2968,7 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence __rcu *
 		}
 		dma_resv_for_each_fence(&cursor, bo->tbo.base.resv,
 					DMA_RESV_USAGE_KERNEL, fence) {
-			ret = amdgpu_sync_fence(&sync_obj, fence);
+			ret = amdgpu_sync_fence(&sync_obj, fence, GFP_KERNEL);
 			if (ret) {
 				pr_debug("Memory eviction: Sync BO fence failed. Try again\n");
 				goto validate_map_fail;
@@ -3013,9 +3012,22 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence __rcu *
 		struct amdgpu_device *adev = amdgpu_ttm_adev(
 			peer_vm->root.bo->tbo.bdev);
 
+		struct amdgpu_fpriv *fpriv =
+			container_of(peer_vm, struct amdgpu_fpriv, vm);
+
+		ret = amdgpu_vm_bo_update(adev, fpriv->prt_va, false);
+		if (ret) {
+			dev_dbg(adev->dev,
+				"Memory eviction: handle PRT moved failed, pid %8d. Try again.\n",
+				pid_nr(process_info->pid));
+			goto validate_map_fail;
+		}
+
 		ret = amdgpu_vm_handle_moved(adev, peer_vm, &exec.ticket);
 		if (ret) {
-			pr_debug("Memory eviction: handle moved failed. Try again\n");
+			dev_dbg(adev->dev,
+				"Memory eviction: handle moved failed, pid %8d. Try again.\n",
+				pid_nr(process_info->pid));
 			goto validate_map_fail;
 		}
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
index 093141ad6ed0..763f2b8dcf13 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
@@ -36,13 +36,6 @@
 #include "atombios_encoders.h"
 #include "bif/bif_4_1_d.h"
 
-static void amdgpu_atombios_lookup_i2c_gpio_quirks(struct amdgpu_device *adev,
-					  ATOM_GPIO_I2C_ASSIGMENT *gpio,
-					  u8 index)
-{
-
-}
-
 static struct amdgpu_i2c_bus_rec amdgpu_atombios_get_bus_rec_for_i2c_gpio(ATOM_GPIO_I2C_ASSIGMENT *gpio)
 {
 	struct amdgpu_i2c_bus_rec i2c;
@@ -108,9 +101,6 @@ struct amdgpu_i2c_bus_rec amdgpu_atombios_lookup_i2c_gpio(struct amdgpu_device *
 
 		gpio = &i2c_info->asGPIO_Info[0];
 		for (i = 0; i < num_indices; i++) {
-
-			amdgpu_atombios_lookup_i2c_gpio_quirks(adev, gpio, i);
-
 			if (gpio->sucI2cId.ucAccess == id) {
 				i2c = amdgpu_atombios_get_bus_rec_for_i2c_gpio(gpio);
 				break;
@@ -142,8 +132,6 @@ void amdgpu_atombios_i2c_init(struct amdgpu_device *adev)
 
 		gpio = &i2c_info->asGPIO_Info[0];
 		for (i = 0; i < num_indices; i++) {
-			amdgpu_atombios_lookup_i2c_gpio_quirks(adev, gpio, i);
-
 			i2c = amdgpu_atombios_get_bus_rec_for_i2c_gpio(gpio);
 
 			if (i2c.valid) {
@@ -156,6 +144,38 @@ void amdgpu_atombios_i2c_init(struct amdgpu_device *adev)
 	}
 }
 
+void amdgpu_atombios_oem_i2c_init(struct amdgpu_device *adev, u8 i2c_id)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	ATOM_GPIO_I2C_ASSIGMENT *gpio;
+	struct amdgpu_i2c_bus_rec i2c;
+	int index = GetIndexIntoMasterTable(DATA, GPIO_I2C_Info);
+	struct _ATOM_GPIO_I2C_INFO *i2c_info;
+	uint16_t data_offset, size;
+	int i, num_indices;
+	char stmp[32];
+
+	if (amdgpu_atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) {
+		i2c_info = (struct _ATOM_GPIO_I2C_INFO *)(ctx->bios + data_offset);
+
+		num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+			sizeof(ATOM_GPIO_I2C_ASSIGMENT);
+
+		gpio = &i2c_info->asGPIO_Info[0];
+		for (i = 0; i < num_indices; i++) {
+			i2c = amdgpu_atombios_get_bus_rec_for_i2c_gpio(gpio);
+
+			if (i2c.valid && i2c.i2c_id == i2c_id) {
+				sprintf(stmp, "OEM 0x%x", i2c.i2c_id);
+				adev->i2c_bus[i] = amdgpu_i2c_create(adev_to_drm(adev), &i2c, stmp);
+				break;
+			}
+			gpio = (ATOM_GPIO_I2C_ASSIGMENT *)
+				((u8 *)gpio + sizeof(ATOM_GPIO_I2C_ASSIGMENT));
+		}
+	}
+}
+
 struct amdgpu_gpio_rec
 amdgpu_atombios_lookup_gpio(struct amdgpu_device *adev,
 			    u8 id)
@@ -686,7 +706,6 @@ int amdgpu_atombios_get_clock_info(struct amdgpu_device *adev)
 		}
 		adev->clock.dp_extclk =
 			le16_to_cpu(firmware_info->info_21.usUniphyDPModeExtClkFreq);
-		adev->clock.current_dispclk = adev->clock.default_dispclk;
 
 		adev->clock.max_pixel_clock = le16_to_cpu(firmware_info->info.usMaxPixelClock);
 		if (adev->clock.max_pixel_clock == 0)
@@ -1796,16 +1815,43 @@ static ssize_t amdgpu_atombios_get_vbios_version(struct device *dev,
 	return sysfs_emit(buf, "%s\n", ctx->vbios_pn);
 }
 
+static ssize_t amdgpu_atombios_get_vbios_build(struct device *dev,
+					       struct device_attribute *attr,
+					       char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct atom_context *ctx = adev->mode_info.atom_context;
+
+	return sysfs_emit(buf, "%s\n", ctx->build_num);
+}
+
 static DEVICE_ATTR(vbios_version, 0444, amdgpu_atombios_get_vbios_version,
 		   NULL);
+static DEVICE_ATTR(vbios_build, 0444, amdgpu_atombios_get_vbios_build, NULL);
 
 static struct attribute *amdgpu_vbios_version_attrs[] = {
-	&dev_attr_vbios_version.attr,
-	NULL
+	&dev_attr_vbios_version.attr, &dev_attr_vbios_build.attr, NULL
 };
 
+static umode_t amdgpu_vbios_version_attrs_is_visible(struct kobject *kobj,
+						     struct attribute *attr,
+						     int index)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct atom_context *ctx = adev->mode_info.atom_context;
+
+	if (attr == &dev_attr_vbios_build.attr && !strlen(ctx->build_num))
+		return 0;
+
+	return attr->mode;
+}
+
 const struct attribute_group amdgpu_vbios_version_attr_group = {
-	.attrs = amdgpu_vbios_version_attrs
+	.attrs = amdgpu_vbios_version_attrs,
+	.is_visible = amdgpu_vbios_version_attrs_is_visible,
 };
 
 int amdgpu_atombios_sysfs_init(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h
index 0e16432d9a72..867bc5c5ce67 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h
@@ -136,6 +136,7 @@ amdgpu_atombios_lookup_gpio(struct amdgpu_device *adev,
 struct amdgpu_i2c_bus_rec amdgpu_atombios_lookup_i2c_gpio(struct amdgpu_device *adev,
 							  uint8_t id);
 void amdgpu_atombios_i2c_init(struct amdgpu_device *adev);
+void amdgpu_atombios_oem_i2c_init(struct amdgpu_device *adev, u8 i2c_id);
 
 bool amdgpu_atombios_has_dce_engine_info(struct amdgpu_device *adev);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
index f873dd3cae16..636385c80f64 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
@@ -181,19 +181,22 @@ int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev)
 	u8 frev, crev;
 	int usage_bytes = 0;
 
-	if (amdgpu_atom_parse_data_header(ctx, index, NULL, &frev, &crev, &data_offset)) {
-		if (frev == 2 && crev == 1) {
-			fw_usage_v2_1 =
-				(struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset);
-			amdgpu_atomfirmware_allocate_fb_v2_1(adev,
-					fw_usage_v2_1,
-					&usage_bytes);
-		} else if (frev >= 2 && crev >= 2) {
-			fw_usage_v2_2 =
-				(struct vram_usagebyfirmware_v2_2 *)(ctx->bios + data_offset);
-			amdgpu_atomfirmware_allocate_fb_v2_2(adev,
-					fw_usage_v2_2,
-					&usage_bytes);
+	/* Skip atomfirmware allocation for SRIOV VFs when dynamic crit regn is enabled */
+	if (!(amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled)) {
+		if (amdgpu_atom_parse_data_header(ctx, index, NULL, &frev, &crev, &data_offset)) {
+			if (frev == 2 && crev == 1) {
+				fw_usage_v2_1 =
+					(struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset);
+				amdgpu_atomfirmware_allocate_fb_v2_1(adev,
+						fw_usage_v2_1,
+						&usage_bytes);
+			} else if (frev >= 2 && crev >= 2) {
+				fw_usage_v2_2 =
+					(struct vram_usagebyfirmware_v2_2 *)(ctx->bios + data_offset);
+				amdgpu_atomfirmware_allocate_fb_v2_2(adev,
+						fw_usage_v2_2,
+						&usage_bytes);
+			}
 		}
 	}
 
@@ -281,6 +284,9 @@ static int convert_atom_mem_type_to_vram_type(struct amdgpu_device *adev,
 		case ATOM_DGPU_VRAM_TYPE_GDDR6:
 			vram_type = AMDGPU_VRAM_TYPE_GDDR6;
 			break;
+		case ATOM_DGPU_VRAM_TYPE_HBM3E:
+			vram_type = AMDGPU_VRAM_TYPE_HBM3E;
+			break;
 		default:
 			vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
 			break;
@@ -549,9 +555,10 @@ bool amdgpu_atomfirmware_mem_ecc_supported(struct amdgpu_device *adev)
 	u16 data_offset, size;
 	union umc_info *umc_info;
 	u8 frev, crev;
-	bool ecc_default_enabled = false;
+	bool mem_ecc_enabled = false;
 	u8 umc_config;
 	u32 umc_config1;
+	adev->ras_default_ecc_enabled = false;
 
 	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
 			umc_info);
@@ -563,20 +570,22 @@ bool amdgpu_atomfirmware_mem_ecc_supported(struct amdgpu_device *adev)
 			switch (crev) {
 			case 1:
 				umc_config = le32_to_cpu(umc_info->v31.umc_config);
-				ecc_default_enabled =
+				mem_ecc_enabled =
 					(umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false;
 				break;
 			case 2:
 				umc_config = le32_to_cpu(umc_info->v32.umc_config);
-				ecc_default_enabled =
+				mem_ecc_enabled =
 					(umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false;
 				break;
 			case 3:
 				umc_config = le32_to_cpu(umc_info->v33.umc_config);
 				umc_config1 = le32_to_cpu(umc_info->v33.umc_config1);
-				ecc_default_enabled =
+				mem_ecc_enabled =
 					((umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ||
 					 (umc_config1 & UMC_CONFIG1__ENABLE_ECC_CAPABLE)) ? true : false;
+				adev->ras_default_ecc_enabled =
+					(umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false;
 				break;
 			default:
 				/* unsupported crev */
@@ -585,9 +594,12 @@ bool amdgpu_atomfirmware_mem_ecc_supported(struct amdgpu_device *adev)
 		} else if (frev == 4) {
 			switch (crev) {
 			case 0:
+				umc_config = le32_to_cpu(umc_info->v40.umc_config);
 				umc_config1 = le32_to_cpu(umc_info->v40.umc_config1);
-				ecc_default_enabled =
+				mem_ecc_enabled =
 					(umc_config1 & UMC_CONFIG1__ENABLE_ECC_CAPABLE) ? true : false;
+				adev->ras_default_ecc_enabled =
+					(umc_config & UMC_CONFIG__DEFAULT_MEM_ECC_ENABLE) ? true : false;
 				break;
 			default:
 				/* unsupported crev */
@@ -599,7 +611,7 @@ bool amdgpu_atomfirmware_mem_ecc_supported(struct amdgpu_device *adev)
 		}
 	}
 
-	return ecc_default_enabled;
+	return mem_ecc_enabled;
 }
 
 /*
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
index 423fd2eebe1e..35d04e69aec0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
@@ -84,18 +84,26 @@ static bool check_atom_bios(struct amdgpu_device *adev, size_t size)
 	return false;
 }
 
+void amdgpu_bios_release(struct amdgpu_device *adev)
+{
+	kfree(adev->bios);
+	adev->bios = NULL;
+	adev->bios_size = 0;
+}
+
 /* If you boot an IGP board with a discrete card as the primary,
  * the IGP rom is not accessible via the rom bar as the IGP rom is
  * part of the system bios.  On boot, the system bios puts a
  * copy of the igp rom at the start of vram if a discrete card is
  * present.
- * For SR-IOV, the vbios image is also put in VRAM in the VF.
+ * For SR-IOV, if dynamic critical region is not enabled,
+ * the vbios image is also put at the start of VRAM in the VF.
  */
 static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev)
 {
-	uint8_t __iomem *bios;
+	uint8_t __iomem *bios = NULL;
 	resource_size_t vram_base;
-	resource_size_t size = 256 * 1024; /* ??? */
+	u32 size = 256U * 1024U; /* ??? */
 
 	if (!(adev->flags & AMD_IS_APU))
 		if (amdgpu_device_need_post(adev))
@@ -107,21 +115,36 @@ static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev)
 
 	adev->bios = NULL;
 	vram_base = pci_resource_start(adev->pdev, 0);
-	bios = ioremap_wc(vram_base, size);
-	if (!bios)
-		return false;
 
 	adev->bios = kmalloc(size, GFP_KERNEL);
-	if (!adev->bios) {
-		iounmap(bios);
+	if (!adev->bios)
 		return false;
+
+	/* For SRIOV with dynamic critical region is enabled,
+	 * the vbios image is put at a dynamic offset of VRAM in the VF.
+	 * If dynamic critical region is disabled, follow the existing logic as on baremetal.
+	 */
+	if (amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled) {
+		if (amdgpu_virt_get_dynamic_data_info(adev,
+				AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID, adev->bios, &size)) {
+			amdgpu_bios_release(adev);
+			return false;
+		}
+	} else {
+		bios = ioremap_wc(vram_base, size);
+		if (!bios) {
+			amdgpu_bios_release(adev);
+			return false;
+		}
+
+		memcpy_fromio(adev->bios, bios, size);
+		iounmap(bios);
 	}
+
 	adev->bios_size = size;
-	memcpy_fromio(adev->bios, bios, size);
-	iounmap(bios);
 
 	if (!check_atom_bios(adev, size)) {
-		kfree(adev->bios);
+		amdgpu_bios_release(adev);
 		return false;
 	}
 
@@ -149,7 +172,7 @@ bool amdgpu_read_bios(struct amdgpu_device *adev)
 	pci_unmap_rom(adev->pdev, bios);
 
 	if (!check_atom_bios(adev, size)) {
-		kfree(adev->bios);
+		amdgpu_bios_release(adev);
 		return false;
 	}
 
@@ -189,7 +212,7 @@ static bool amdgpu_read_bios_from_rom(struct amdgpu_device *adev)
 	amdgpu_asic_read_bios_from_rom(adev, adev->bios, len);
 
 	if (!check_atom_bios(adev, len)) {
-		kfree(adev->bios);
+		amdgpu_bios_release(adev);
 		return false;
 	}
 
@@ -225,7 +248,8 @@ static bool amdgpu_read_platform_bios(struct amdgpu_device *adev)
 
 	return true;
 free_bios:
-	kfree(adev->bios);
+	amdgpu_bios_release(adev);
+
 	return false;
 }
 
@@ -327,7 +351,7 @@ static bool amdgpu_atrm_get_bios(struct amdgpu_device *adev)
 	}
 
 	if (!check_atom_bios(adev, size)) {
-		kfree(adev->bios);
+		amdgpu_bios_release(adev);
 		return false;
 	}
 	adev->bios_size = size;
@@ -392,7 +416,7 @@ static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
 					     GFP_KERNEL);
 
 			if (!check_atom_bios(adev, vhdr->ImageLength)) {
-				kfree(adev->bios);
+				amdgpu_bios_release(adev);
 				return false;
 			}
 			adev->bios_size = vhdr->ImageLength;
@@ -439,6 +463,13 @@ success:
 	return true;
 }
 
+static bool amdgpu_prefer_rom_resource(struct amdgpu_device *adev)
+{
+	struct resource *res = &adev->pdev->resource[PCI_ROM_RESOURCE];
+
+	return (res->flags & IORESOURCE_ROM_SHADOW);
+}
+
 static bool amdgpu_get_bios_dgpu(struct amdgpu_device *adev)
 {
 	if (amdgpu_atrm_get_bios(adev)) {
@@ -457,14 +488,27 @@ static bool amdgpu_get_bios_dgpu(struct amdgpu_device *adev)
 		goto success;
 	}
 
-	if (amdgpu_read_platform_bios(adev)) {
-		dev_info(adev->dev, "Fetched VBIOS from platform\n");
-		goto success;
-	}
+	if (amdgpu_prefer_rom_resource(adev)) {
+		if (amdgpu_read_bios(adev)) {
+			dev_info(adev->dev, "Fetched VBIOS from ROM BAR\n");
+			goto success;
+		}
 
-	if (amdgpu_read_bios(adev)) {
-		dev_info(adev->dev, "Fetched VBIOS from ROM BAR\n");
-		goto success;
+		if (amdgpu_read_platform_bios(adev)) {
+			dev_info(adev->dev, "Fetched VBIOS from platform\n");
+			goto success;
+		}
+
+	} else {
+		if (amdgpu_read_platform_bios(adev)) {
+			dev_info(adev->dev, "Fetched VBIOS from platform\n");
+			goto success;
+		}
+
+		if (amdgpu_read_bios(adev)) {
+			dev_info(adev->dev, "Fetched VBIOS from ROM BAR\n");
+			goto success;
+		}
 	}
 
 	if (amdgpu_read_bios_from_rom(adev)) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
index 702f6610d024..66fb37b64388 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
@@ -184,43 +184,36 @@ void amdgpu_bo_list_put(struct amdgpu_bo_list *list)
 int amdgpu_bo_create_list_entry_array(struct drm_amdgpu_bo_list_in *in,
 				      struct drm_amdgpu_bo_list_entry **info_param)
 {
-	const void __user *uptr = u64_to_user_ptr(in->bo_info_ptr);
 	const uint32_t info_size = sizeof(struct drm_amdgpu_bo_list_entry);
+	const void __user *uptr = u64_to_user_ptr(in->bo_info_ptr);
+	const uint32_t bo_info_size = in->bo_info_size;
+	const uint32_t bo_number = in->bo_number;
 	struct drm_amdgpu_bo_list_entry *info;
-	int r;
-
-	info = kvmalloc_array(in->bo_number, info_size, GFP_KERNEL);
-	if (!info)
-		return -ENOMEM;
 
 	/* copy the handle array from userspace to a kernel buffer */
-	r = -EFAULT;
-	if (likely(info_size == in->bo_info_size)) {
-		unsigned long bytes = in->bo_number *
-			in->bo_info_size;
-
-		if (copy_from_user(info, uptr, bytes))
-			goto error_free;
-
+	if (likely(info_size == bo_info_size)) {
+		info = vmemdup_array_user(uptr, bo_number, info_size);
+		if (IS_ERR(info))
+			return PTR_ERR(info);
 	} else {
-		unsigned long bytes = min(in->bo_info_size, info_size);
+		const uint32_t bytes = min(bo_info_size, info_size);
 		unsigned i;
 
-		memset(info, 0, in->bo_number * info_size);
-		for (i = 0; i < in->bo_number; ++i) {
-			if (copy_from_user(&info[i], uptr, bytes))
-				goto error_free;
+		info = kvmalloc_array(bo_number, info_size, GFP_KERNEL);
+		if (!info)
+			return -ENOMEM;
 
-			uptr += in->bo_info_size;
+		memset(info, 0, bo_number * info_size);
+		for (i = 0; i < bo_number; ++i, uptr += bo_info_size) {
+			if (copy_from_user(&info[i], uptr, bytes)) {
+				kvfree(info);
+				return -EFAULT;
+			}
 		}
 	}
 
 	*info_param = info;
 	return 0;
-
-error_free:
-	kvfree(info);
-	return r;
 }
 
 int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
index 555cd6d877c3..2b5e7c46a39d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
@@ -38,8 +38,7 @@ struct amdgpu_bo_list_entry {
 	struct amdgpu_bo		*bo;
 	struct amdgpu_bo_va		*bo_va;
 	uint32_t			priority;
-	struct page			**user_pages;
-	struct hmm_range		*range;
+	struct amdgpu_hmm_range		*range;
 	bool				user_invalidated;
 };
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
index 68bce6a6d09d..004a6a9d6b9f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
@@ -252,83 +252,22 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 
 		if (!adev->pm.fw) {
 			switch (adev->asic_type) {
-			case CHIP_TAHITI:
-				strcpy(fw_name, "radeon/tahiti_smc.bin");
-				break;
-			case CHIP_PITCAIRN:
-				if ((adev->pdev->revision == 0x81) &&
-				    ((adev->pdev->device == 0x6810) ||
-				    (adev->pdev->device == 0x6811))) {
-					info->is_kicker = true;
-					strcpy(fw_name, "radeon/pitcairn_k_smc.bin");
-				} else {
-					strcpy(fw_name, "radeon/pitcairn_smc.bin");
-				}
-				break;
-			case CHIP_VERDE:
-				if (((adev->pdev->device == 0x6820) &&
-					((adev->pdev->revision == 0x81) ||
-					(adev->pdev->revision == 0x83))) ||
-				    ((adev->pdev->device == 0x6821) &&
-					((adev->pdev->revision == 0x83) ||
-					(adev->pdev->revision == 0x87))) ||
-				    ((adev->pdev->revision == 0x87) &&
-					((adev->pdev->device == 0x6823) ||
-					(adev->pdev->device == 0x682b)))) {
-					info->is_kicker = true;
-					strcpy(fw_name, "radeon/verde_k_smc.bin");
-				} else {
-					strcpy(fw_name, "radeon/verde_smc.bin");
-				}
-				break;
-			case CHIP_OLAND:
-				if (((adev->pdev->revision == 0x81) &&
-					((adev->pdev->device == 0x6600) ||
-					(adev->pdev->device == 0x6604) ||
-					(adev->pdev->device == 0x6605) ||
-					(adev->pdev->device == 0x6610))) ||
-				    ((adev->pdev->revision == 0x83) &&
-					(adev->pdev->device == 0x6610))) {
-					info->is_kicker = true;
-					strcpy(fw_name, "radeon/oland_k_smc.bin");
-				} else {
-					strcpy(fw_name, "radeon/oland_smc.bin");
-				}
-				break;
-			case CHIP_HAINAN:
-				if (((adev->pdev->revision == 0x81) &&
-					(adev->pdev->device == 0x6660)) ||
-				    ((adev->pdev->revision == 0x83) &&
-					((adev->pdev->device == 0x6660) ||
-					(adev->pdev->device == 0x6663) ||
-					(adev->pdev->device == 0x6665) ||
-					 (adev->pdev->device == 0x6667)))) {
-					info->is_kicker = true;
-					strcpy(fw_name, "radeon/hainan_k_smc.bin");
-				} else if ((adev->pdev->revision == 0xc3) &&
-					 (adev->pdev->device == 0x6665)) {
-					info->is_kicker = true;
-					strcpy(fw_name, "radeon/banks_k_2_smc.bin");
-				} else {
-					strcpy(fw_name, "radeon/hainan_smc.bin");
-				}
-				break;
 			case CHIP_BONAIRE:
 				if ((adev->pdev->revision == 0x80) ||
 					(adev->pdev->revision == 0x81) ||
 					(adev->pdev->device == 0x665f)) {
 					info->is_kicker = true;
-					strcpy(fw_name, "amdgpu/bonaire_k_smc.bin");
+					strscpy(fw_name, "amdgpu/bonaire_k_smc.bin");
 				} else {
-					strcpy(fw_name, "amdgpu/bonaire_smc.bin");
+					strscpy(fw_name, "amdgpu/bonaire_smc.bin");
 				}
 				break;
 			case CHIP_HAWAII:
 				if (adev->pdev->revision == 0x80) {
 					info->is_kicker = true;
-					strcpy(fw_name, "amdgpu/hawaii_k_smc.bin");
+					strscpy(fw_name, "amdgpu/hawaii_k_smc.bin");
 				} else {
-					strcpy(fw_name, "amdgpu/hawaii_smc.bin");
+					strscpy(fw_name, "amdgpu/hawaii_smc.bin");
 				}
 				break;
 			case CHIP_TOPAZ:
@@ -338,76 +277,76 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 				    ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0xD1)) ||
 				    ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0xD3))) {
 					info->is_kicker = true;
-					strcpy(fw_name, "amdgpu/topaz_k_smc.bin");
+					strscpy(fw_name, "amdgpu/topaz_k_smc.bin");
 				} else
-					strcpy(fw_name, "amdgpu/topaz_smc.bin");
+					strscpy(fw_name, "amdgpu/topaz_smc.bin");
 				break;
 			case CHIP_TONGA:
 				if (((adev->pdev->device == 0x6939) && (adev->pdev->revision == 0xf1)) ||
 				    ((adev->pdev->device == 0x6938) && (adev->pdev->revision == 0xf1))) {
 					info->is_kicker = true;
-					strcpy(fw_name, "amdgpu/tonga_k_smc.bin");
+					strscpy(fw_name, "amdgpu/tonga_k_smc.bin");
 				} else
-					strcpy(fw_name, "amdgpu/tonga_smc.bin");
+					strscpy(fw_name, "amdgpu/tonga_smc.bin");
 				break;
 			case CHIP_FIJI:
-				strcpy(fw_name, "amdgpu/fiji_smc.bin");
+				strscpy(fw_name, "amdgpu/fiji_smc.bin");
 				break;
 			case CHIP_POLARIS11:
 				if (type == CGS_UCODE_ID_SMU) {
 					if (ASICID_IS_P21(adev->pdev->device, adev->pdev->revision)) {
 						info->is_kicker = true;
-						strcpy(fw_name, "amdgpu/polaris11_k_smc.bin");
+						strscpy(fw_name, "amdgpu/polaris11_k_smc.bin");
 					} else if (ASICID_IS_P31(adev->pdev->device, adev->pdev->revision)) {
 						info->is_kicker = true;
-						strcpy(fw_name, "amdgpu/polaris11_k2_smc.bin");
+						strscpy(fw_name, "amdgpu/polaris11_k2_smc.bin");
 					} else {
-						strcpy(fw_name, "amdgpu/polaris11_smc.bin");
+						strscpy(fw_name, "amdgpu/polaris11_smc.bin");
 					}
 				} else if (type == CGS_UCODE_ID_SMU_SK) {
-					strcpy(fw_name, "amdgpu/polaris11_smc_sk.bin");
+					strscpy(fw_name, "amdgpu/polaris11_smc_sk.bin");
 				}
 				break;
 			case CHIP_POLARIS10:
 				if (type == CGS_UCODE_ID_SMU) {
 					if (ASICID_IS_P20(adev->pdev->device, adev->pdev->revision)) {
 						info->is_kicker = true;
-						strcpy(fw_name, "amdgpu/polaris10_k_smc.bin");
+						strscpy(fw_name, "amdgpu/polaris10_k_smc.bin");
 					} else if (ASICID_IS_P30(adev->pdev->device, adev->pdev->revision)) {
 						info->is_kicker = true;
-						strcpy(fw_name, "amdgpu/polaris10_k2_smc.bin");
+						strscpy(fw_name, "amdgpu/polaris10_k2_smc.bin");
 					} else {
-						strcpy(fw_name, "amdgpu/polaris10_smc.bin");
+						strscpy(fw_name, "amdgpu/polaris10_smc.bin");
 					}
 				} else if (type == CGS_UCODE_ID_SMU_SK) {
-					strcpy(fw_name, "amdgpu/polaris10_smc_sk.bin");
+					strscpy(fw_name, "amdgpu/polaris10_smc_sk.bin");
 				}
 				break;
 			case CHIP_POLARIS12:
 				if (ASICID_IS_P23(adev->pdev->device, adev->pdev->revision)) {
 					info->is_kicker = true;
-					strcpy(fw_name, "amdgpu/polaris12_k_smc.bin");
+					strscpy(fw_name, "amdgpu/polaris12_k_smc.bin");
 				} else {
-					strcpy(fw_name, "amdgpu/polaris12_smc.bin");
+					strscpy(fw_name, "amdgpu/polaris12_smc.bin");
 				}
 				break;
 			case CHIP_VEGAM:
-				strcpy(fw_name, "amdgpu/vegam_smc.bin");
+				strscpy(fw_name, "amdgpu/vegam_smc.bin");
 				break;
 			case CHIP_VEGA10:
 				if ((adev->pdev->device == 0x687f) &&
 					((adev->pdev->revision == 0xc0) ||
 					(adev->pdev->revision == 0xc1) ||
 					(adev->pdev->revision == 0xc3)))
-					strcpy(fw_name, "amdgpu/vega10_acg_smc.bin");
+					strscpy(fw_name, "amdgpu/vega10_acg_smc.bin");
 				else
-					strcpy(fw_name, "amdgpu/vega10_smc.bin");
+					strscpy(fw_name, "amdgpu/vega10_smc.bin");
 				break;
 			case CHIP_VEGA12:
-				strcpy(fw_name, "amdgpu/vega12_smc.bin");
+				strscpy(fw_name, "amdgpu/vega12_smc.bin");
 				break;
 			case CHIP_VEGA20:
-				strcpy(fw_name, "amdgpu/vega20_smc.bin");
+				strscpy(fw_name, "amdgpu/vega20_smc.bin");
 				break;
 			default:
 				DRM_ERROR("SMC firmware not supported\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
index 344e0a9ee08a..9f96d568acf2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
@@ -398,30 +398,28 @@ static void amdgpu_connector_add_common_modes(struct drm_encoder *encoder,
 	struct drm_display_mode *mode = NULL;
 	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
 	int i;
-	static const struct mode_size {
+	int n;
+	struct mode_size {
+		char name[DRM_DISPLAY_MODE_LEN];
 		int w;
 		int h;
-	} common_modes[17] = {
-		{ 640,  480},
-		{ 720,  480},
-		{ 800,  600},
-		{ 848,  480},
-		{1024,  768},
-		{1152,  768},
-		{1280,  720},
-		{1280,  800},
-		{1280,  854},
-		{1280,  960},
-		{1280, 1024},
-		{1440,  900},
-		{1400, 1050},
-		{1680, 1050},
-		{1600, 1200},
-		{1920, 1080},
-		{1920, 1200}
+	} common_modes[] = {
+		{  "640x480",  640,  480},
+		{  "800x600",  800,  600},
+		{ "1024x768", 1024,  768},
+		{ "1280x720", 1280,  720},
+		{ "1280x800", 1280,  800},
+		{"1280x1024", 1280, 1024},
+		{ "1440x900", 1440,  900},
+		{"1680x1050", 1680, 1050},
+		{"1600x1200", 1600, 1200},
+		{"1920x1080", 1920, 1080},
+		{"1920x1200", 1920, 1200}
 	};
 
-	for (i = 0; i < 17; i++) {
+	n = ARRAY_SIZE(common_modes);
+
+	for (i = 0; i < n; i++) {
 		if (amdgpu_encoder->devices & (ATOM_DEVICE_TV_SUPPORT)) {
 			if (common_modes[i].w > 1024 ||
 			    common_modes[i].h > 768)
@@ -434,12 +432,11 @@ static void amdgpu_connector_add_common_modes(struct drm_encoder *encoder,
 			     common_modes[i].h == native_mode->vdisplay))
 				continue;
 		}
-		if (common_modes[i].w < 320 || common_modes[i].h < 200)
-			continue;
 
 		mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h, 60, false, false, false);
 		if (!mode)
 			return;
+		strscpy(mode->name, common_modes[i].name, DRM_DISPLAY_MODE_LEN);
 
 		drm_mode_probed_add(connector, mode);
 	}
@@ -674,7 +671,7 @@ static int amdgpu_connector_lvds_get_modes(struct drm_connector *connector)
 }
 
 static enum drm_mode_status amdgpu_connector_lvds_mode_valid(struct drm_connector *connector,
-					     struct drm_display_mode *mode)
+					     const struct drm_display_mode *mode)
 {
 	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
 
@@ -737,10 +734,8 @@ amdgpu_connector_lvds_detect(struct drm_connector *connector, bool force)
 
 	amdgpu_connector_update_scratch_regs(connector, ret);
 
-	if (!drm_kms_helper_is_poll_worker()) {
-		pm_runtime_mark_last_busy(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker())
 		pm_runtime_put_autosuspend(connector->dev->dev);
-	}
 
 	return ret;
 }
@@ -839,7 +834,7 @@ static int amdgpu_connector_vga_get_modes(struct drm_connector *connector)
 }
 
 static enum drm_mode_status amdgpu_connector_vga_mode_valid(struct drm_connector *connector,
-					    struct drm_display_mode *mode)
+					    const struct drm_display_mode *mode)
 {
 	struct drm_device *dev = connector->dev;
 	struct amdgpu_device *adev = drm_to_adev(dev);
@@ -922,10 +917,8 @@ amdgpu_connector_vga_detect(struct drm_connector *connector, bool force)
 	amdgpu_connector_update_scratch_regs(connector, ret);
 
 out:
-	if (!drm_kms_helper_is_poll_worker()) {
-		pm_runtime_mark_last_busy(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker())
 		pm_runtime_put_autosuspend(connector->dev->dev);
-	}
 
 	return ret;
 }
@@ -1149,10 +1142,8 @@ out:
 	amdgpu_connector_update_scratch_regs(connector, ret);
 
 exit:
-	if (!drm_kms_helper_is_poll_worker()) {
-		pm_runtime_mark_last_busy(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker())
 		pm_runtime_put_autosuspend(connector->dev->dev);
-	}
 
 	return ret;
 }
@@ -1195,29 +1186,69 @@ static void amdgpu_connector_dvi_force(struct drm_connector *connector)
 		amdgpu_connector->use_digital = true;
 }
 
+/**
+ * amdgpu_max_hdmi_pixel_clock - Return max supported HDMI (TMDS) pixel clock
+ * @adev: pointer to amdgpu_device
+ *
+ * Return: maximum supported HDMI (TMDS) pixel clock in KHz.
+ */
+static int amdgpu_max_hdmi_pixel_clock(const struct amdgpu_device *adev)
+{
+	if (adev->asic_type >= CHIP_POLARIS10)
+		return 600000;
+	else if (adev->asic_type >= CHIP_TONGA)
+		return 300000;
+	else
+		return 297000;
+}
+
+/**
+ * amdgpu_connector_dvi_mode_valid - Validate a mode on DVI/HDMI connectors
+ * @connector: DRM connector to validate the mode on
+ * @mode: display mode to validate
+ *
+ * Validate the given display mode on DVI and HDMI connectors, including
+ * analog signals on DVI-I.
+ *
+ * Return: drm_mode_status indicating whether the mode is valid.
+ */
 static enum drm_mode_status amdgpu_connector_dvi_mode_valid(struct drm_connector *connector,
-					    struct drm_display_mode *mode)
+					    const struct drm_display_mode *mode)
 {
 	struct drm_device *dev = connector->dev;
 	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	const int max_hdmi_pixel_clock = amdgpu_max_hdmi_pixel_clock(adev);
+	const int max_dvi_single_link_pixel_clock = 165000;
+	int max_digital_pixel_clock_khz;
 
 	/* XXX check mode bandwidth */
 
-	if (amdgpu_connector->use_digital && (mode->clock > 165000)) {
-		if ((amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I) ||
-		    (amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D) ||
-		    (amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_HDMI_TYPE_B)) {
-			return MODE_OK;
-		} else if (connector->display_info.is_hdmi) {
-			/* HDMI 1.3+ supports max clock of 340 Mhz */
-			if (mode->clock > 340000)
-				return MODE_CLOCK_HIGH;
-			else
-				return MODE_OK;
-		} else {
-			return MODE_CLOCK_HIGH;
+	if (amdgpu_connector->use_digital) {
+		switch (amdgpu_connector->connector_object_id) {
+		case CONNECTOR_OBJECT_ID_HDMI_TYPE_A:
+			max_digital_pixel_clock_khz = max_hdmi_pixel_clock;
+			break;
+		case CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I:
+		case CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D:
+			max_digital_pixel_clock_khz = max_dvi_single_link_pixel_clock;
+			break;
+		case CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I:
+		case CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D:
+		case CONNECTOR_OBJECT_ID_HDMI_TYPE_B:
+			max_digital_pixel_clock_khz = max_dvi_single_link_pixel_clock * 2;
+			break;
 		}
+
+		/* When the display EDID claims that it's an HDMI display,
+		 * we use the HDMI encoder mode of the display HW,
+		 * so we should verify against the max HDMI clock here.
+		 */
+		if (connector->display_info.is_hdmi)
+			max_digital_pixel_clock_khz = max_hdmi_pixel_clock;
+
+		if (mode->clock > max_digital_pixel_clock_khz)
+			return MODE_CLOCK_HIGH;
 	}
 
 	/* check against the max pixel clock */
@@ -1449,10 +1480,8 @@ amdgpu_connector_dp_detect(struct drm_connector *connector, bool force)
 
 	amdgpu_connector_update_scratch_regs(connector, ret);
 out:
-	if (!drm_kms_helper_is_poll_worker()) {
-		pm_runtime_mark_last_busy(connector->dev->dev);
+	if (!drm_kms_helper_is_poll_worker())
 		pm_runtime_put_autosuspend(connector->dev->dev);
-	}
 
 	if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort ||
 	    connector->connector_type == DRM_MODE_CONNECTOR_eDP)
@@ -1464,7 +1493,7 @@ out:
 }
 
 static enum drm_mode_status amdgpu_connector_dp_mode_valid(struct drm_connector *connector,
-					   struct drm_display_mode *mode)
+					   const struct drm_display_mode *mode)
 {
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = amdgpu_connector->con_priv;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c
new file mode 100644
index 000000000000..425a3e564360
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c
@@ -0,0 +1,591 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/list.h>
+#include "amdgpu.h"
+
+static const guid_t MCE			= CPER_NOTIFY_MCE;
+static const guid_t CMC			= CPER_NOTIFY_CMC;
+static const guid_t BOOT		= BOOT_TYPE;
+
+static const guid_t CRASHDUMP		= AMD_CRASHDUMP;
+static const guid_t RUNTIME		= AMD_GPU_NONSTANDARD_ERROR;
+
+static void __inc_entry_length(struct cper_hdr *hdr, uint32_t size)
+{
+	hdr->record_length += size;
+}
+
+static void amdgpu_cper_get_timestamp(struct cper_timestamp *timestamp)
+{
+	struct tm tm;
+	time64_t now = ktime_get_real_seconds();
+
+	time64_to_tm(now, 0, &tm);
+	timestamp->seconds = tm.tm_sec;
+	timestamp->minutes = tm.tm_min;
+	timestamp->hours = tm.tm_hour;
+	timestamp->flag = 0;
+	timestamp->day = tm.tm_mday;
+	timestamp->month = 1 + tm.tm_mon;
+	timestamp->year = (1900 + tm.tm_year) % 100;
+	timestamp->century = (1900 + tm.tm_year) / 100;
+}
+
+void amdgpu_cper_entry_fill_hdr(struct amdgpu_device *adev,
+				struct cper_hdr *hdr,
+				enum amdgpu_cper_type type,
+				enum cper_error_severity sev)
+{
+	char record_id[16];
+
+	hdr->signature[0]		= 'C';
+	hdr->signature[1]		= 'P';
+	hdr->signature[2]		= 'E';
+	hdr->signature[3]		= 'R';
+	hdr->revision			= CPER_HDR_REV_1;
+	hdr->signature_end		= 0xFFFFFFFF;
+	hdr->error_severity		= sev;
+
+	hdr->valid_bits.platform_id	= 1;
+	hdr->valid_bits.timestamp	= 1;
+
+	amdgpu_cper_get_timestamp(&hdr->timestamp);
+
+	snprintf(record_id, 9, "%d:%X",
+		 (adev->smuio.funcs && adev->smuio.funcs->get_socket_id) ?
+			 adev->smuio.funcs->get_socket_id(adev) :
+			 0,
+		 atomic_inc_return(&adev->cper.unique_id));
+	memcpy(hdr->record_id, record_id, 8);
+
+	snprintf(hdr->platform_id, 16, "0x%04X:0x%04X",
+		 adev->pdev->vendor, adev->pdev->device);
+	/* pmfw version should be part of creator_id according to CPER spec */
+	snprintf(hdr->creator_id, 16, "%s", CPER_CREATOR_ID_AMDGPU);
+
+	switch (type) {
+	case AMDGPU_CPER_TYPE_BOOT:
+		hdr->notify_type = BOOT;
+		break;
+	case AMDGPU_CPER_TYPE_FATAL:
+	case AMDGPU_CPER_TYPE_BP_THRESHOLD:
+		hdr->notify_type = MCE;
+		break;
+	case AMDGPU_CPER_TYPE_RUNTIME:
+		if (sev == CPER_SEV_NON_FATAL_CORRECTED)
+			hdr->notify_type = CMC;
+		else
+			hdr->notify_type = MCE;
+		break;
+	default:
+		dev_err(adev->dev, "Unknown CPER Type\n");
+		break;
+	}
+
+	__inc_entry_length(hdr, HDR_LEN);
+}
+
+static int amdgpu_cper_entry_fill_section_desc(struct amdgpu_device *adev,
+					       struct cper_sec_desc *section_desc,
+					       bool bp_threshold,
+					       bool poison,
+					       enum cper_error_severity sev,
+					       guid_t sec_type,
+					       uint32_t section_length,
+					       uint32_t section_offset)
+{
+	section_desc->revision_minor		= CPER_SEC_MINOR_REV_1;
+	section_desc->revision_major		= CPER_SEC_MAJOR_REV_22;
+	section_desc->sec_offset		= section_offset;
+	section_desc->sec_length		= section_length;
+	section_desc->valid_bits.fru_text	= 1;
+	section_desc->flag_bits.primary		= 1;
+	section_desc->severity			= sev;
+	section_desc->sec_type			= sec_type;
+
+	snprintf(section_desc->fru_text, 20, "OAM%d",
+		 (adev->smuio.funcs && adev->smuio.funcs->get_socket_id) ?
+			 adev->smuio.funcs->get_socket_id(adev) :
+			 0);
+
+	if (bp_threshold)
+		section_desc->flag_bits.exceed_err_threshold = 1;
+	if (poison)
+		section_desc->flag_bits.latent_err = 1;
+
+	return 0;
+}
+
+int amdgpu_cper_entry_fill_fatal_section(struct amdgpu_device *adev,
+					 struct cper_hdr *hdr,
+					 uint32_t idx,
+					 struct cper_sec_crashdump_reg_data reg_data)
+{
+	struct cper_sec_desc *section_desc;
+	struct cper_sec_crashdump_fatal *section;
+
+	section_desc = (struct cper_sec_desc *)((uint8_t *)hdr + SEC_DESC_OFFSET(idx));
+	section = (struct cper_sec_crashdump_fatal *)((uint8_t *)hdr +
+		   FATAL_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	amdgpu_cper_entry_fill_section_desc(adev, section_desc, false, false,
+					    CPER_SEV_FATAL, CRASHDUMP, FATAL_SEC_LEN,
+					    FATAL_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	section->body.reg_ctx_type = CPER_CTX_TYPE_CRASH;
+	section->body.reg_arr_size = sizeof(reg_data);
+	section->body.data = reg_data;
+
+	__inc_entry_length(hdr, SEC_DESC_LEN + FATAL_SEC_LEN);
+
+	return 0;
+}
+
+int amdgpu_cper_entry_fill_runtime_section(struct amdgpu_device *adev,
+					   struct cper_hdr *hdr,
+					   uint32_t idx,
+					   enum cper_error_severity sev,
+					   uint32_t *reg_dump,
+					   uint32_t reg_count)
+{
+	struct cper_sec_desc *section_desc;
+	struct cper_sec_nonstd_err *section;
+	bool poison;
+
+	poison = sev != CPER_SEV_NON_FATAL_CORRECTED;
+	section_desc = (struct cper_sec_desc *)((uint8_t *)hdr + SEC_DESC_OFFSET(idx));
+	section = (struct cper_sec_nonstd_err *)((uint8_t *)hdr +
+		   NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	amdgpu_cper_entry_fill_section_desc(adev, section_desc, false, poison,
+					    sev, RUNTIME, NONSTD_SEC_LEN,
+					    NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	reg_count = umin(reg_count, CPER_ACA_REG_COUNT);
+
+	section->hdr.valid_bits.err_info_cnt = 1;
+	section->hdr.valid_bits.err_context_cnt = 1;
+
+	section->info.error_type = RUNTIME;
+	section->info.ms_chk_bits.err_type_valid = 1;
+	section->ctx.reg_ctx_type = CPER_CTX_TYPE_CRASH;
+	section->ctx.reg_arr_size = sizeof(section->ctx.reg_dump);
+
+	memcpy(section->ctx.reg_dump, reg_dump, reg_count * sizeof(uint32_t));
+
+	__inc_entry_length(hdr, SEC_DESC_LEN + NONSTD_SEC_LEN);
+
+	return 0;
+}
+
+int amdgpu_cper_entry_fill_bad_page_threshold_section(struct amdgpu_device *adev,
+						      struct cper_hdr *hdr,
+						      uint32_t idx)
+{
+	struct cper_sec_desc *section_desc;
+	struct cper_sec_nonstd_err *section;
+	uint32_t socket_id;
+
+	section_desc = (struct cper_sec_desc *)((uint8_t *)hdr + SEC_DESC_OFFSET(idx));
+	section = (struct cper_sec_nonstd_err *)((uint8_t *)hdr +
+		   NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	amdgpu_cper_entry_fill_section_desc(adev, section_desc, true, false,
+					    CPER_SEV_FATAL, RUNTIME, NONSTD_SEC_LEN,
+					    NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+	section->hdr.valid_bits.err_info_cnt = 1;
+	section->hdr.valid_bits.err_context_cnt = 1;
+
+	section->info.error_type = RUNTIME;
+	section->info.valid_bits.ms_chk = 1;
+	section->info.ms_chk_bits.err_type_valid = 1;
+	section->info.ms_chk_bits.err_type = 1;
+	section->info.ms_chk_bits.pcc = 1;
+	section->ctx.reg_ctx_type = CPER_CTX_TYPE_CRASH;
+	section->ctx.reg_arr_size = sizeof(section->ctx.reg_dump);
+
+	/* Hardcoded Reg dump for bad page threshold CPER */
+	socket_id = (adev->smuio.funcs && adev->smuio.funcs->get_socket_id) ?
+				adev->smuio.funcs->get_socket_id(adev) :
+				0;
+	section->ctx.reg_dump[CPER_ACA_REG_CTL_LO]    = 0x1;
+	section->ctx.reg_dump[CPER_ACA_REG_CTL_HI]    = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_STATUS_LO] = 0x137;
+	section->ctx.reg_dump[CPER_ACA_REG_STATUS_HI] = 0xB0000000;
+	section->ctx.reg_dump[CPER_ACA_REG_ADDR_LO]   = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_ADDR_HI]   = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_MISC0_LO]  = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_MISC0_HI]  = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_CONFIG_LO] = 0x2;
+	section->ctx.reg_dump[CPER_ACA_REG_CONFIG_HI] = 0x1ff;
+	section->ctx.reg_dump[CPER_ACA_REG_IPID_LO]   = (socket_id / 4) & 0x01;
+	section->ctx.reg_dump[CPER_ACA_REG_IPID_HI]   = 0x096 | (((socket_id % 4) & 0x3) << 12);
+	section->ctx.reg_dump[CPER_ACA_REG_SYND_LO]   = 0x0;
+	section->ctx.reg_dump[CPER_ACA_REG_SYND_HI]   = 0x0;
+
+	__inc_entry_length(hdr, SEC_DESC_LEN + NONSTD_SEC_LEN);
+
+	return 0;
+}
+
+struct cper_hdr *amdgpu_cper_alloc_entry(struct amdgpu_device *adev,
+					 enum amdgpu_cper_type type,
+					 uint16_t section_count)
+{
+	struct cper_hdr *hdr;
+	uint32_t size = 0;
+
+	size += HDR_LEN;
+	size += (SEC_DESC_LEN * section_count);
+
+	switch (type) {
+	case AMDGPU_CPER_TYPE_RUNTIME:
+	case AMDGPU_CPER_TYPE_BP_THRESHOLD:
+		size += (NONSTD_SEC_LEN * section_count);
+		break;
+	case AMDGPU_CPER_TYPE_FATAL:
+		size += (FATAL_SEC_LEN * section_count);
+		break;
+	case AMDGPU_CPER_TYPE_BOOT:
+		size += (BOOT_SEC_LEN * section_count);
+		break;
+	default:
+		dev_err(adev->dev, "Unknown CPER Type!\n");
+		return NULL;
+	}
+
+	hdr = kzalloc(size, GFP_KERNEL);
+	if (!hdr)
+		return NULL;
+
+	/* Save this early */
+	hdr->sec_cnt = section_count;
+
+	return hdr;
+}
+
+int amdgpu_cper_generate_ue_record(struct amdgpu_device *adev,
+				   struct aca_bank *bank)
+{
+	struct cper_hdr *fatal = NULL;
+	struct cper_sec_crashdump_reg_data reg_data = { 0 };
+	struct amdgpu_ring *ring = &adev->cper.ring_buf;
+	int ret;
+
+	fatal = amdgpu_cper_alloc_entry(adev, AMDGPU_CPER_TYPE_FATAL, 1);
+	if (!fatal) {
+		dev_err(adev->dev, "fail to alloc cper entry for ue record\n");
+		return -ENOMEM;
+	}
+
+	reg_data.status_lo = lower_32_bits(bank->regs[ACA_REG_IDX_STATUS]);
+	reg_data.status_hi = upper_32_bits(bank->regs[ACA_REG_IDX_STATUS]);
+	reg_data.addr_lo   = lower_32_bits(bank->regs[ACA_REG_IDX_ADDR]);
+	reg_data.addr_hi   = upper_32_bits(bank->regs[ACA_REG_IDX_ADDR]);
+	reg_data.ipid_lo   = lower_32_bits(bank->regs[ACA_REG_IDX_IPID]);
+	reg_data.ipid_hi   = upper_32_bits(bank->regs[ACA_REG_IDX_IPID]);
+	reg_data.synd_lo   = lower_32_bits(bank->regs[ACA_REG_IDX_SYND]);
+	reg_data.synd_hi   = upper_32_bits(bank->regs[ACA_REG_IDX_SYND]);
+
+	amdgpu_cper_entry_fill_hdr(adev, fatal, AMDGPU_CPER_TYPE_FATAL, CPER_SEV_FATAL);
+	ret = amdgpu_cper_entry_fill_fatal_section(adev, fatal, 0, reg_data);
+	if (ret)
+		return ret;
+
+	amdgpu_cper_ring_write(ring, fatal, fatal->record_length);
+	kfree(fatal);
+
+	return 0;
+}
+
+int amdgpu_cper_generate_bp_threshold_record(struct amdgpu_device *adev)
+{
+	struct cper_hdr *bp_threshold = NULL;
+	struct amdgpu_ring *ring = &adev->cper.ring_buf;
+	int ret;
+
+	bp_threshold = amdgpu_cper_alloc_entry(adev, AMDGPU_CPER_TYPE_BP_THRESHOLD, 1);
+	if (!bp_threshold) {
+		dev_err(adev->dev, "fail to alloc cper entry for bad page threshold record\n");
+		return -ENOMEM;
+	}
+
+	amdgpu_cper_entry_fill_hdr(adev, bp_threshold,
+				   AMDGPU_CPER_TYPE_BP_THRESHOLD,
+				   CPER_SEV_FATAL);
+	ret = amdgpu_cper_entry_fill_bad_page_threshold_section(adev, bp_threshold, 0);
+	if (ret)
+		return ret;
+
+	amdgpu_cper_ring_write(ring, bp_threshold, bp_threshold->record_length);
+	kfree(bp_threshold);
+
+	return 0;
+}
+
+static enum cper_error_severity amdgpu_aca_err_type_to_cper_sev(struct amdgpu_device *adev,
+								enum aca_error_type aca_err_type)
+{
+	switch (aca_err_type) {
+	case ACA_ERROR_TYPE_UE:
+		return CPER_SEV_FATAL;
+	case ACA_ERROR_TYPE_CE:
+		return CPER_SEV_NON_FATAL_CORRECTED;
+	case ACA_ERROR_TYPE_DEFERRED:
+		return CPER_SEV_NON_FATAL_UNCORRECTED;
+	default:
+		dev_err(adev->dev, "Unknown ACA error type!\n");
+		return CPER_SEV_FATAL;
+	}
+}
+
+int amdgpu_cper_generate_ce_records(struct amdgpu_device *adev,
+				    struct aca_banks *banks,
+				    uint16_t bank_count)
+{
+	struct cper_hdr *corrected = NULL;
+	enum cper_error_severity sev = CPER_SEV_NON_FATAL_CORRECTED;
+	struct amdgpu_ring *ring = &adev->cper.ring_buf;
+	uint32_t reg_data[CPER_ACA_REG_COUNT] = { 0 };
+	struct aca_bank_node *node;
+	struct aca_bank *bank;
+	uint32_t i = 0;
+	int ret;
+
+	corrected = amdgpu_cper_alloc_entry(adev, AMDGPU_CPER_TYPE_RUNTIME, bank_count);
+	if (!corrected) {
+		dev_err(adev->dev, "fail to allocate cper entry for ce records\n");
+		return -ENOMEM;
+	}
+
+	/* Raise severity if any DE is detected in the ACA bank list */
+	list_for_each_entry(node, &banks->list, node) {
+		bank = &node->bank;
+		if (bank->aca_err_type == ACA_ERROR_TYPE_DEFERRED) {
+			sev = CPER_SEV_NON_FATAL_UNCORRECTED;
+			break;
+		}
+	}
+
+	amdgpu_cper_entry_fill_hdr(adev, corrected, AMDGPU_CPER_TYPE_RUNTIME, sev);
+
+	/* Combine CE and DE in cper record */
+	list_for_each_entry(node, &banks->list, node) {
+		bank = &node->bank;
+		reg_data[CPER_ACA_REG_CTL_LO]    = lower_32_bits(bank->regs[ACA_REG_IDX_CTL]);
+		reg_data[CPER_ACA_REG_CTL_HI]    = upper_32_bits(bank->regs[ACA_REG_IDX_CTL]);
+		reg_data[CPER_ACA_REG_STATUS_LO] = lower_32_bits(bank->regs[ACA_REG_IDX_STATUS]);
+		reg_data[CPER_ACA_REG_STATUS_HI] = upper_32_bits(bank->regs[ACA_REG_IDX_STATUS]);
+		reg_data[CPER_ACA_REG_ADDR_LO]   = lower_32_bits(bank->regs[ACA_REG_IDX_ADDR]);
+		reg_data[CPER_ACA_REG_ADDR_HI]   = upper_32_bits(bank->regs[ACA_REG_IDX_ADDR]);
+		reg_data[CPER_ACA_REG_MISC0_LO]  = lower_32_bits(bank->regs[ACA_REG_IDX_MISC0]);
+		reg_data[CPER_ACA_REG_MISC0_HI]  = upper_32_bits(bank->regs[ACA_REG_IDX_MISC0]);
+		reg_data[CPER_ACA_REG_CONFIG_LO] = lower_32_bits(bank->regs[ACA_REG_IDX_CONFIG]);
+		reg_data[CPER_ACA_REG_CONFIG_HI] = upper_32_bits(bank->regs[ACA_REG_IDX_CONFIG]);
+		reg_data[CPER_ACA_REG_IPID_LO]   = lower_32_bits(bank->regs[ACA_REG_IDX_IPID]);
+		reg_data[CPER_ACA_REG_IPID_HI]   = upper_32_bits(bank->regs[ACA_REG_IDX_IPID]);
+		reg_data[CPER_ACA_REG_SYND_LO]   = lower_32_bits(bank->regs[ACA_REG_IDX_SYND]);
+		reg_data[CPER_ACA_REG_SYND_HI]   = upper_32_bits(bank->regs[ACA_REG_IDX_SYND]);
+
+		ret = amdgpu_cper_entry_fill_runtime_section(adev, corrected, i++,
+				amdgpu_aca_err_type_to_cper_sev(adev, bank->aca_err_type),
+				reg_data, CPER_ACA_REG_COUNT);
+		if (ret)
+			return ret;
+	}
+
+	amdgpu_cper_ring_write(ring, corrected, corrected->record_length);
+	kfree(corrected);
+
+	return 0;
+}
+
+static bool amdgpu_cper_is_hdr(struct amdgpu_ring *ring, u64 pos)
+{
+	struct cper_hdr *chdr;
+
+	chdr = (struct cper_hdr *)&(ring->ring[pos]);
+	return strcmp(chdr->signature, "CPER") ? false : true;
+}
+
+static u32 amdgpu_cper_ring_get_ent_sz(struct amdgpu_ring *ring, u64 pos)
+{
+	struct cper_hdr *chdr;
+	u64 p;
+	u32 chunk, rec_len = 0;
+
+	chdr = (struct cper_hdr *)&(ring->ring[pos]);
+	chunk = ring->ring_size - (pos << 2);
+
+	if (!strcmp(chdr->signature, "CPER")) {
+		rec_len = chdr->record_length;
+		goto calc;
+	}
+
+	/* ring buffer is not full, no cper data after ring->wptr */
+	if (ring->count_dw)
+		goto calc;
+
+	for (p = pos + 1; p <= ring->buf_mask; p++) {
+		chdr = (struct cper_hdr *)&(ring->ring[p]);
+		if (!strcmp(chdr->signature, "CPER")) {
+			rec_len = (p - pos) << 2;
+			goto calc;
+		}
+	}
+
+calc:
+	if (!rec_len)
+		return chunk;
+	else
+		return umin(rec_len, chunk);
+}
+
+void amdgpu_cper_ring_write(struct amdgpu_ring *ring, void *src, int count)
+{
+	u64 pos, wptr_old, rptr;
+	int rec_cnt_dw = count >> 2;
+	u32 chunk, ent_sz;
+	u8 *s = (u8 *)src;
+
+	if (count >= ring->ring_size - 4) {
+		dev_err(ring->adev->dev,
+			"CPER data size(%d) is larger than ring size(%d)\n",
+			count, ring->ring_size - 4);
+
+		return;
+	}
+
+	mutex_lock(&ring->adev->cper.ring_lock);
+
+	wptr_old = ring->wptr;
+	rptr = *ring->rptr_cpu_addr & ring->ptr_mask;
+
+	while (count) {
+		ent_sz = amdgpu_cper_ring_get_ent_sz(ring, ring->wptr);
+		chunk = umin(ent_sz, count);
+
+		memcpy(&ring->ring[ring->wptr], s, chunk);
+
+		ring->wptr += (chunk >> 2);
+		ring->wptr &= ring->ptr_mask;
+		count -= chunk;
+		s += chunk;
+	}
+
+	if (ring->count_dw < rec_cnt_dw)
+		ring->count_dw = 0;
+
+	/* the buffer is overflow, adjust rptr */
+	if (((wptr_old < rptr) && (rptr <= ring->wptr)) ||
+	    ((ring->wptr < wptr_old) && (wptr_old < rptr)) ||
+	    ((rptr <= ring->wptr) && (ring->wptr < wptr_old))) {
+		pos = (ring->wptr + 1) & ring->ptr_mask;
+
+		do {
+			ent_sz = amdgpu_cper_ring_get_ent_sz(ring, pos);
+
+			rptr += (ent_sz >> 2);
+			rptr &= ring->ptr_mask;
+			*ring->rptr_cpu_addr = rptr;
+
+			pos = rptr;
+		} while (!amdgpu_cper_is_hdr(ring, rptr));
+	}
+
+	if (ring->count_dw >= rec_cnt_dw)
+		ring->count_dw -= rec_cnt_dw;
+	mutex_unlock(&ring->adev->cper.ring_lock);
+}
+
+static u64 amdgpu_cper_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	return *(ring->rptr_cpu_addr);
+}
+
+static u64 amdgpu_cper_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	return ring->wptr;
+}
+
+static const struct amdgpu_ring_funcs cper_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_CPER,
+	.align_mask = 0xff,
+	.support_64bit_ptrs = false,
+	.get_rptr = amdgpu_cper_ring_get_rptr,
+	.get_wptr = amdgpu_cper_ring_get_wptr,
+};
+
+static int amdgpu_cper_ring_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &(adev->cper.ring_buf);
+
+	mutex_init(&adev->cper.ring_lock);
+
+	ring->adev = NULL;
+	ring->ring_obj = NULL;
+	ring->use_doorbell = false;
+	ring->no_scheduler = true;
+	ring->funcs = &cper_ring_funcs;
+
+	sprintf(ring->name, "cper");
+	return amdgpu_ring_init(adev, ring, CPER_MAX_RING_SIZE, NULL, 0,
+				AMDGPU_RING_PRIO_DEFAULT, NULL);
+}
+
+int amdgpu_cper_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!amdgpu_aca_is_enabled(adev) && !amdgpu_sriov_ras_cper_en(adev))
+		return 0;
+
+	r = amdgpu_cper_ring_init(adev);
+	if (r) {
+		dev_err(adev->dev, "failed to initialize cper ring, r = %d\n", r);
+		return r;
+	}
+
+	mutex_init(&adev->cper.cper_lock);
+
+	adev->cper.enabled = true;
+	adev->cper.max_count = CPER_MAX_ALLOWED_COUNT;
+
+	return 0;
+}
+
+int amdgpu_cper_fini(struct amdgpu_device *adev)
+{
+	if (!amdgpu_aca_is_enabled(adev) && !amdgpu_sriov_ras_cper_en(adev))
+		return 0;
+
+	adev->cper.enabled = false;
+
+	amdgpu_ring_fini(&(adev->cper.ring_buf));
+	adev->cper.count = 0;
+	adev->cper.wptr = 0;
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h
new file mode 100644
index 000000000000..353421807387
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h
@@ -0,0 +1,105 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_CPER_H__
+#define __AMDGPU_CPER_H__
+
+#include "amd_cper.h"
+#include "amdgpu_aca.h"
+
+#define CPER_MAX_ALLOWED_COUNT		0x1000
+#define CPER_MAX_RING_SIZE		0X100000
+#define HDR_LEN				(sizeof(struct cper_hdr))
+#define SEC_DESC_LEN			(sizeof(struct cper_sec_desc))
+
+#define BOOT_SEC_LEN			(sizeof(struct cper_sec_crashdump_boot))
+#define FATAL_SEC_LEN			(sizeof(struct cper_sec_crashdump_fatal))
+#define NONSTD_SEC_LEN			(sizeof(struct cper_sec_nonstd_err))
+
+#define SEC_DESC_OFFSET(idx)		(HDR_LEN + (SEC_DESC_LEN * idx))
+
+#define BOOT_SEC_OFFSET(count, idx)	(HDR_LEN + (SEC_DESC_LEN * count) + (BOOT_SEC_LEN * idx))
+#define FATAL_SEC_OFFSET(count, idx)	(HDR_LEN + (SEC_DESC_LEN * count) + (FATAL_SEC_LEN * idx))
+#define NONSTD_SEC_OFFSET(count, idx)	(HDR_LEN + (SEC_DESC_LEN * count) + (NONSTD_SEC_LEN * idx))
+
+enum amdgpu_cper_type {
+	AMDGPU_CPER_TYPE_RUNTIME,
+	AMDGPU_CPER_TYPE_FATAL,
+	AMDGPU_CPER_TYPE_BOOT,
+	AMDGPU_CPER_TYPE_BP_THRESHOLD,
+};
+
+struct amdgpu_cper {
+	bool enabled;
+
+	atomic_t unique_id;
+	struct mutex cper_lock;
+
+	/* Lifetime CPERs generated */
+	uint32_t count;
+	uint32_t max_count;
+
+	uint32_t wptr;
+
+	void *ring[CPER_MAX_ALLOWED_COUNT];
+	struct amdgpu_ring ring_buf;
+	struct mutex ring_lock;
+};
+
+void amdgpu_cper_entry_fill_hdr(struct amdgpu_device *adev,
+				struct cper_hdr *hdr,
+				enum amdgpu_cper_type type,
+				enum cper_error_severity sev);
+int amdgpu_cper_entry_fill_fatal_section(struct amdgpu_device *adev,
+					 struct cper_hdr *hdr,
+					 uint32_t idx,
+					 struct cper_sec_crashdump_reg_data reg_data);
+int amdgpu_cper_entry_fill_runtime_section(struct amdgpu_device *adev,
+					   struct cper_hdr *hdr,
+					   uint32_t idx,
+					   enum cper_error_severity sev,
+					   uint32_t *reg_dump,
+					   uint32_t reg_count);
+int amdgpu_cper_entry_fill_bad_page_threshold_section(struct amdgpu_device *adev,
+						      struct cper_hdr *hdr,
+						      uint32_t section_idx);
+
+struct cper_hdr *amdgpu_cper_alloc_entry(struct amdgpu_device *adev,
+					 enum amdgpu_cper_type type,
+					 uint16_t section_count);
+/* UE must be encoded into separated cper entries, 1 UE 1 cper */
+int amdgpu_cper_generate_ue_record(struct amdgpu_device *adev,
+				   struct aca_bank *bank);
+/* CEs and DEs are combined into 1 cper entry */
+int amdgpu_cper_generate_ce_records(struct amdgpu_device *adev,
+				    struct aca_banks *banks,
+				    uint16_t bank_count);
+/* Bad page threshold is encoded into separated cper entry */
+int amdgpu_cper_generate_bp_threshold_record(struct amdgpu_device *adev);
+void amdgpu_cper_ring_write(struct amdgpu_ring *ring,
+			void *src, int count);
+int amdgpu_cper_init(struct amdgpu_device *adev);
+int amdgpu_cper_fini(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 5cc5f59e3018..ecdfe6cb36cc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -40,6 +40,7 @@
 #include "amdgpu_gmc.h"
 #include "amdgpu_gem.h"
 #include "amdgpu_ras.h"
+#include "amdgpu_hmm.h"
 
 static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p,
 				 struct amdgpu_device *adev,
@@ -178,25 +179,17 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
 	unsigned int num_ibs[AMDGPU_CS_GANG_SIZE] = { };
 	struct amdgpu_vm *vm = &fpriv->vm;
-	uint64_t *chunk_array_user;
 	uint64_t *chunk_array;
 	uint32_t uf_offset = 0;
 	size_t size;
 	int ret;
 	int i;
 
-	chunk_array = kvmalloc_array(cs->in.num_chunks, sizeof(uint64_t),
-				     GFP_KERNEL);
-	if (!chunk_array)
-		return -ENOMEM;
-
-	/* get chunks */
-	chunk_array_user = u64_to_user_ptr(cs->in.chunks);
-	if (copy_from_user(chunk_array, chunk_array_user,
-			   sizeof(uint64_t)*cs->in.num_chunks)) {
-		ret = -EFAULT;
-		goto free_chunk;
-	}
+	chunk_array = memdup_array_user(u64_to_user_ptr(cs->in.chunks),
+					cs->in.num_chunks,
+					sizeof(uint64_t));
+	if (IS_ERR(chunk_array))
+		return PTR_ERR(chunk_array);
 
 	p->nchunks = cs->in.num_chunks;
 	p->chunks = kvmalloc_array(p->nchunks, sizeof(struct amdgpu_cs_chunk),
@@ -209,7 +202,6 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
 	for (i = 0; i < p->nchunks; i++) {
 		struct drm_amdgpu_cs_chunk __user *chunk_ptr = NULL;
 		struct drm_amdgpu_cs_chunk user_chunk;
-		uint32_t __user *cdata;
 
 		chunk_ptr = u64_to_user_ptr(chunk_array[i]);
 		if (copy_from_user(&user_chunk, chunk_ptr,
@@ -222,20 +214,16 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
 		p->chunks[i].length_dw = user_chunk.length_dw;
 
 		size = p->chunks[i].length_dw;
-		cdata = u64_to_user_ptr(user_chunk.chunk_data);
 
-		p->chunks[i].kdata = kvmalloc_array(size, sizeof(uint32_t),
-						    GFP_KERNEL);
-		if (p->chunks[i].kdata == NULL) {
-			ret = -ENOMEM;
+		p->chunks[i].kdata = vmemdup_array_user(u64_to_user_ptr(user_chunk.chunk_data),
+							size,
+							sizeof(uint32_t));
+		if (IS_ERR(p->chunks[i].kdata)) {
+			ret = PTR_ERR(p->chunks[i].kdata);
 			i--;
 			goto free_partial_kdata;
 		}
 		size *= sizeof(uint32_t);
-		if (copy_from_user(p->chunks[i].kdata, cdata, size)) {
-			ret = -EFAULT;
-			goto free_partial_kdata;
-		}
 
 		/* Assume the worst on the following checks */
 		ret = -EINVAL;
@@ -286,17 +274,36 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
 		}
 	}
 
-	if (!p->gang_size) {
+	if (!p->gang_size || (amdgpu_sriov_vf(p->adev) && p->gang_size > 1)) {
 		ret = -EINVAL;
 		goto free_all_kdata;
 	}
 
 	for (i = 0; i < p->gang_size; ++i) {
 		ret = amdgpu_job_alloc(p->adev, vm, p->entities[i], vm,
-				       num_ibs[i], &p->jobs[i]);
+				       num_ibs[i], &p->jobs[i],
+				       p->filp->client_id);
 		if (ret)
 			goto free_all_kdata;
-		p->jobs[i]->enforce_isolation = p->adev->enforce_isolation[fpriv->xcp_id];
+		switch (p->adev->enforce_isolation[fpriv->xcp_id]) {
+		case AMDGPU_ENFORCE_ISOLATION_DISABLE:
+		default:
+			p->jobs[i]->enforce_isolation = false;
+			p->jobs[i]->run_cleaner_shader = false;
+			break;
+		case AMDGPU_ENFORCE_ISOLATION_ENABLE:
+			p->jobs[i]->enforce_isolation = true;
+			p->jobs[i]->run_cleaner_shader = true;
+			break;
+		case AMDGPU_ENFORCE_ISOLATION_ENABLE_LEGACY:
+			p->jobs[i]->enforce_isolation = true;
+			p->jobs[i]->run_cleaner_shader = false;
+			break;
+		case AMDGPU_ENFORCE_ISOLATION_NO_CLEANER_SHADER:
+			p->jobs[i]->enforce_isolation = true;
+			p->jobs[i]->run_cleaner_shader = false;
+			break;
+		}
 	}
 	p->gang_leader = p->jobs[p->gang_leader_idx];
 
@@ -349,10 +356,20 @@ static int amdgpu_cs_p2_ib(struct amdgpu_cs_parser *p,
 	ring = amdgpu_job_ring(job);
 	ib = &job->ibs[job->num_ibs++];
 
+	/* submissions to kernel queues are disabled */
+	if (ring->no_user_submission)
+		return -EINVAL;
+
 	/* MM engine doesn't support user fences */
 	if (p->uf_bo && ring->funcs->no_user_fence)
 		return -EINVAL;
 
+	if (!p->adev->debug_enable_ce_cs &&
+	    chunk_ib->flags & AMDGPU_IB_FLAG_CE) {
+		dev_err_ratelimited(p->adev->dev, "CE CS is blocked, use debug=0x400 to override\n");
+		return -EINVAL;
+	}
+
 	if (chunk_ib->ip_type == AMDGPU_HW_IP_GFX &&
 	    chunk_ib->flags & AMDGPU_IB_FLAG_PREEMPT) {
 		if (chunk_ib->flags & AMDGPU_IB_FLAG_CE)
@@ -373,7 +390,7 @@ static int amdgpu_cs_p2_ib(struct amdgpu_cs_parser *p,
 			   chunk_ib->ib_bytes : 0,
 			   AMDGPU_IB_POOL_DELAYED, ib);
 	if (r) {
-		DRM_ERROR("Failed to get ib !\n");
+		drm_err(adev_to_drm(p->adev), "Failed to get ib !\n");
 		return r;
 	}
 
@@ -428,7 +445,7 @@ static int amdgpu_cs_p2_dependencies(struct amdgpu_cs_parser *p,
 			dma_fence_put(old);
 		}
 
-		r = amdgpu_sync_fence(&p->sync, fence);
+		r = amdgpu_sync_fence(&p->sync, fence, GFP_KERNEL);
 		dma_fence_put(fence);
 		if (r)
 			return r;
@@ -445,12 +462,12 @@ static int amdgpu_syncobj_lookup_and_add(struct amdgpu_cs_parser *p,
 
 	r = drm_syncobj_find_fence(p->filp, handle, point, flags, &fence);
 	if (r) {
-		DRM_ERROR("syncobj %u failed to find fence @ %llu (%d)!\n",
+		drm_err(adev_to_drm(p->adev), "syncobj %u failed to find fence @ %llu (%d)!\n",
 			  handle, point, r);
 		return r;
 	}
 
-	r = amdgpu_sync_fence(&p->sync, fence);
+	r = amdgpu_sync_fence(&p->sync, fence, GFP_KERNEL);
 	dma_fence_put(fence);
 	return r;
 }
@@ -691,7 +708,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 	 */
 	const s64 us_upper_bound = 200000;
 
-	if (!adev->mm_stats.log2_max_MBps) {
+	if ((!adev->mm_stats.log2_max_MBps) || !ttm_resource_manager_used(&adev->mman.vram_mgr.manager)) {
 		*max_bytes = 0;
 		*max_vis_bytes = 0;
 		return;
@@ -873,26 +890,18 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
 		bool userpage_invalidated = false;
 		struct amdgpu_bo *bo = e->bo;
-		int i;
-
-		e->user_pages = kvcalloc(bo->tbo.ttm->num_pages,
-					 sizeof(struct page *),
-					 GFP_KERNEL);
-		if (!e->user_pages) {
-			DRM_ERROR("kvmalloc_array failure\n");
-			r = -ENOMEM;
-			goto out_free_user_pages;
-		}
 
-		r = amdgpu_ttm_tt_get_user_pages(bo, e->user_pages, &e->range);
-		if (r) {
-			kvfree(e->user_pages);
-			e->user_pages = NULL;
+		e->range = amdgpu_hmm_range_alloc(NULL);
+		if (unlikely(!e->range))
+			return -ENOMEM;
+
+		r = amdgpu_ttm_tt_get_user_pages(bo, e->range);
+		if (r)
 			goto out_free_user_pages;
-		}
 
 		for (i = 0; i < bo->tbo.ttm->num_pages; i++) {
-			if (bo->tbo.ttm->pages[i] != e->user_pages[i]) {
+			if (bo->tbo.ttm->pages[i] !=
+				hmm_pfn_to_page(e->range->hmm_range.hmm_pfns[i])) {
 				userpage_invalidated = true;
 				break;
 			}
@@ -936,7 +945,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 		}
 
 		if (amdgpu_ttm_tt_is_userptr(e->bo->tbo.ttm) &&
-		    e->user_invalidated && e->user_pages) {
+		    e->user_invalidated) {
 			amdgpu_bo_placement_from_domain(e->bo,
 							AMDGPU_GEM_DOMAIN_CPU);
 			r = ttm_bo_validate(&e->bo->tbo, &e->bo->placement,
@@ -945,11 +954,8 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 				goto out_free_user_pages;
 
 			amdgpu_ttm_tt_set_user_pages(e->bo->tbo.ttm,
-						     e->user_pages);
+						     e->range);
 		}
-
-		kvfree(e->user_pages);
-		e->user_pages = NULL;
 	}
 
 	amdgpu_cs_get_threshold_for_moves(p->adev, &p->bytes_moved_threshold,
@@ -960,7 +966,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	r = amdgpu_vm_validate(p->adev, &fpriv->vm, NULL,
 			       amdgpu_cs_bo_validate, p);
 	if (r) {
-		DRM_ERROR("amdgpu_vm_validate() failed.\n");
+		drm_err(adev_to_drm(p->adev), "amdgpu_vm_validate() failed.\n");
 		goto out_free_user_pages;
 	}
 
@@ -989,13 +995,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 
 out_free_user_pages:
 	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
-		struct amdgpu_bo *bo = e->bo;
-
-		if (!e->user_pages)
-			continue;
-		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, e->range);
-		kvfree(e->user_pages);
-		e->user_pages = NULL;
+		amdgpu_hmm_range_free(e->range);
 		e->range = NULL;
 	}
 	mutex_unlock(&p->bo_list->bo_list_mutex);
@@ -1038,13 +1038,13 @@ static int amdgpu_cs_patch_ibs(struct amdgpu_cs_parser *p,
 		va_start = ib->gpu_addr & AMDGPU_GMC_HOLE_MASK;
 		r = amdgpu_cs_find_mapping(p, va_start, &aobj, &m);
 		if (r) {
-			DRM_ERROR("IB va_start is invalid\n");
+			drm_err(adev_to_drm(p->adev), "IB va_start is invalid\n");
 			return r;
 		}
 
 		if ((va_start + ib->length_dw * 4) >
 		    (m->last + 1) * AMDGPU_GPU_PAGE_SIZE) {
-			DRM_ERROR("IB va_start+ib_bytes is invalid\n");
+			drm_err(adev_to_drm(p->adev), "IB va_start+ib_bytes is invalid\n");
 			return -EINVAL;
 		}
 
@@ -1111,11 +1111,14 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 			struct drm_gpu_scheduler *sched = entity->rq->sched;
 			struct amdgpu_ring *ring = to_amdgpu_ring(sched);
 
-			if (amdgpu_vmid_uses_reserved(adev, vm, ring->vm_hub))
+			if (amdgpu_vmid_uses_reserved(vm, ring->vm_hub))
 				return -EINVAL;
 		}
 	}
 
+	if (!amdgpu_vm_ready(vm))
+		return -EINVAL;
+
 	r = amdgpu_vm_clear_freed(adev, vm, NULL);
 	if (r)
 		return r;
@@ -1124,7 +1127,8 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 	if (r)
 		return r;
 
-	r = amdgpu_sync_fence(&p->sync, fpriv->prt_va->last_pt_update);
+	r = amdgpu_sync_fence(&p->sync, fpriv->prt_va->last_pt_update,
+			      GFP_KERNEL);
 	if (r)
 		return r;
 
@@ -1135,7 +1139,8 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 		if (r)
 			return r;
 
-		r = amdgpu_sync_fence(&p->sync, bo_va->last_pt_update);
+		r = amdgpu_sync_fence(&p->sync, bo_va->last_pt_update,
+				      GFP_KERNEL);
 		if (r)
 			return r;
 	}
@@ -1154,7 +1159,8 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 		if (r)
 			return r;
 
-		r = amdgpu_sync_fence(&p->sync, bo_va->last_pt_update);
+		r = amdgpu_sync_fence(&p->sync, bo_va->last_pt_update,
+				      GFP_KERNEL);
 		if (r)
 			return r;
 	}
@@ -1167,7 +1173,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 	if (r)
 		return r;
 
-	r = amdgpu_sync_fence(&p->sync, vm->last_update);
+	r = amdgpu_sync_fence(&p->sync, vm->last_update, GFP_KERNEL);
 	if (r)
 		return r;
 
@@ -1209,7 +1215,7 @@ static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
 	r = amdgpu_ctx_wait_prev_fence(p->ctx, p->entities[p->gang_leader_idx]);
 	if (r) {
 		if (r != -ERESTARTSYS)
-			DRM_ERROR("amdgpu_ctx_wait_prev_fence failed.\n");
+			drm_err(adev_to_drm(p->adev), "amdgpu_ctx_wait_prev_fence failed.\n");
 		return r;
 	}
 
@@ -1248,7 +1254,8 @@ static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
 			continue;
 		}
 
-		r = amdgpu_sync_fence(&p->gang_leader->explicit_sync, fence);
+		r = amdgpu_sync_fence(&p->gang_leader->explicit_sync, fence,
+				      GFP_KERNEL);
 		dma_fence_put(fence);
 		if (r)
 			return r;
@@ -1319,8 +1326,8 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	 */
 	r = 0;
 	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
-		r |= !amdgpu_ttm_tt_get_user_pages_done(e->bo->tbo.ttm,
-							e->range);
+		r |= !amdgpu_hmm_range_valid(e->range);
+		amdgpu_hmm_range_free(e->range);
 		e->range = NULL;
 	}
 	if (r) {
@@ -1421,7 +1428,7 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 
 	r = amdgpu_cs_parser_init(&parser, adev, filp, data);
 	if (r) {
-		DRM_ERROR_RATELIMITED("Failed to initialize parser %d!\n", r);
+		drm_err_ratelimited(dev, "Failed to initialize parser %d!\n", r);
 		return r;
 	}
 
@@ -1436,9 +1443,9 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 	r = amdgpu_cs_parser_bos(&parser, data);
 	if (r) {
 		if (r == -ENOMEM)
-			DRM_ERROR("Not enough memory for command submission!\n");
+			drm_err(dev, "Not enough memory for command submission!\n");
 		else if (r != -ERESTARTSYS && r != -EAGAIN)
-			DRM_DEBUG("Failed to process the buffer list %d!\n", r);
+			drm_dbg(dev, "Failed to process the buffer list %d!\n", r);
 		goto error_fini;
 	}
 
@@ -1737,30 +1744,21 @@ int amdgpu_cs_wait_fences_ioctl(struct drm_device *dev, void *data,
 {
 	struct amdgpu_device *adev = drm_to_adev(dev);
 	union drm_amdgpu_wait_fences *wait = data;
-	uint32_t fence_count = wait->in.fence_count;
-	struct drm_amdgpu_fence *fences_user;
 	struct drm_amdgpu_fence *fences;
 	int r;
 
 	/* Get the fences from userspace */
-	fences = kmalloc_array(fence_count, sizeof(struct drm_amdgpu_fence),
-			GFP_KERNEL);
-	if (fences == NULL)
-		return -ENOMEM;
-
-	fences_user = u64_to_user_ptr(wait->in.fences);
-	if (copy_from_user(fences, fences_user,
-		sizeof(struct drm_amdgpu_fence) * fence_count)) {
-		r = -EFAULT;
-		goto err_free_fences;
-	}
+	fences = memdup_array_user(u64_to_user_ptr(wait->in.fences),
+				   wait->in.fence_count,
+				   sizeof(struct drm_amdgpu_fence));
+	if (IS_ERR(fences))
+		return PTR_ERR(fences);
 
 	if (wait->in.wait_all)
 		r = amdgpu_cs_wait_all_fences(adev, filp, wait, fences);
 	else
 		r = amdgpu_cs_wait_any_fence(adev, filp, wait, fences);
 
-err_free_fences:
 	kfree(fences);
 
 	return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
index cfdf558b48b6..02138aa55793 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
@@ -109,7 +109,7 @@ int amdgpu_unmap_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	struct drm_exec exec;
 	int r;
 
-	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT, 0);
+	drm_exec_init(&exec, 0, 0);
 	drm_exec_until_all_locked(&exec) {
 		r = amdgpu_vm_lock_pd(vm, &exec, 0);
 		if (likely(!r))
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
index c43d1b6e5d66..afedea02188d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
@@ -236,7 +236,7 @@ static int amdgpu_ctx_init_entity(struct amdgpu_ctx *ctx, u32 hw_ip,
 		r = amdgpu_xcp_select_scheds(adev, hw_ip, hw_prio, fpriv,
 						&num_scheds, &scheds);
 		if (r)
-			goto cleanup_entity;
+			goto error_free_entity;
 	}
 
 	/* disable load balance if the hw engine retains context among dependent jobs */
@@ -919,7 +919,7 @@ long amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr, long timeout)
 	return timeout;
 }
 
-void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr)
+static void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr)
 {
 	struct amdgpu_ctx *ctx;
 	struct idr *idp;
@@ -944,24 +944,13 @@ void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr)
 				drm_sched_entity_fini(entity);
 			}
 		}
+		kref_put(&ctx->refcount, amdgpu_ctx_fini);
 	}
 }
 
 void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr)
 {
-	struct amdgpu_ctx *ctx;
-	struct idr *idp;
-	uint32_t id;
-
 	amdgpu_ctx_mgr_entity_fini(mgr);
-
-	idp = &mgr->ctx_handles;
-
-	idr_for_each_entry(idp, ctx, id) {
-		if (kref_put(&ctx->refcount, amdgpu_ctx_fini) != 1)
-			DRM_ERROR("ctx %p is still alive\n", ctx);
-	}
-
 	idr_destroy(&mgr->ctx_handles);
 	mutex_destroy(&mgr->lock);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h
index 85376baaa92f..090dfe86f75b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h
@@ -92,7 +92,6 @@ int amdgpu_ctx_wait_prev_fence(struct amdgpu_ctx *ctx,
 
 void amdgpu_ctx_mgr_init(struct amdgpu_ctx_mgr *mgr,
 			 struct amdgpu_device *adev);
-void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr);
 long amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr, long timeout);
 void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr);
 void amdgpu_ctx_mgr_usage(struct amdgpu_ctx_mgr *mgr,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
index 49ca8c814455..62d43b8cbe58 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
@@ -129,7 +129,6 @@ static int  amdgpu_debugfs_process_reg_op(bool read, struct file *f,
 	if (use_bank) {
 		if ((sh_bank != 0xFFFFFFFF && sh_bank >= adev->gfx.config.max_sh_per_se) ||
 		    (se_bank != 0xFFFFFFFF && se_bank >= adev->gfx.config.max_shader_engines)) {
-			pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 			pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 			amdgpu_virt_disable_access_debugfs(adev);
 			return -EINVAL;
@@ -179,7 +178,6 @@ end:
 	if (pm_pg_lock)
 		mutex_unlock(&adev->pm.mutex);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	amdgpu_virt_disable_access_debugfs(adev);
@@ -255,7 +253,6 @@ static ssize_t amdgpu_debugfs_regs2_op(struct file *f, char __user *buf, u32 off
 	if (rd->id.use_grbm) {
 		if ((rd->id.grbm.sh != 0xFFFFFFFF && rd->id.grbm.sh >= adev->gfx.config.max_sh_per_se) ||
 		    (rd->id.grbm.se != 0xFFFFFFFF && rd->id.grbm.se >= adev->gfx.config.max_shader_engines)) {
-			pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 			pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 			amdgpu_virt_disable_access_debugfs(adev);
 			mutex_unlock(&rd->lock);
@@ -310,7 +307,6 @@ end:
 
 	mutex_unlock(&rd->lock);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	amdgpu_virt_disable_access_debugfs(adev);
@@ -446,7 +442,6 @@ static ssize_t amdgpu_debugfs_gprwave_read(struct file *f, char __user *buf, siz
 	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, rd->id.xcc_id);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	if (!x) {
@@ -557,7 +552,6 @@ static ssize_t amdgpu_debugfs_regs_pcie_read(struct file *f, char __user *buf,
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -617,7 +611,6 @@ static ssize_t amdgpu_debugfs_regs_pcie_write(struct file *f, const char __user
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -676,7 +669,6 @@ static ssize_t amdgpu_debugfs_regs_didt_read(struct file *f, char __user *buf,
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -736,7 +728,6 @@ static ssize_t amdgpu_debugfs_regs_didt_write(struct file *f, const char __user
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -795,7 +786,6 @@ static ssize_t amdgpu_debugfs_regs_smc_read(struct file *f, char __user *buf,
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -855,7 +845,6 @@ static ssize_t amdgpu_debugfs_regs_smc_write(struct file *f, const char __user *
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	amdgpu_virt_disable_access_debugfs(adev);
 	return r;
@@ -1003,7 +992,6 @@ static ssize_t amdgpu_debugfs_sensor_read(struct file *f, char __user *buf,
 
 	r = amdgpu_dpm_read_sensor(adev, idx, &values[0], &valuesize);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	if (r) {
@@ -1094,7 +1082,6 @@ static ssize_t amdgpu_debugfs_wave_read(struct file *f, char __user *buf,
 	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	if (!x) {
@@ -1192,7 +1179,6 @@ static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf,
 	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	while (size) {
@@ -1266,7 +1252,6 @@ static ssize_t amdgpu_debugfs_gfxoff_residency_read(struct file *f, char __user
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1315,7 +1300,6 @@ static ssize_t amdgpu_debugfs_gfxoff_residency_write(struct file *f, const char
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1365,7 +1349,6 @@ static ssize_t amdgpu_debugfs_gfxoff_count_read(struct file *f, char __user *buf
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1414,7 +1397,6 @@ static ssize_t amdgpu_debugfs_gfxoff_write(struct file *f, const char __user *bu
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1460,7 +1442,6 @@ static ssize_t amdgpu_debugfs_gfxoff_read(struct file *f, char __user *buf,
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1501,7 +1482,6 @@ static ssize_t amdgpu_debugfs_gfxoff_status_read(struct file *f, char __user *bu
 
 	r = result;
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return r;
@@ -1701,7 +1681,6 @@ static int amdgpu_debugfs_test_ib_show(struct seq_file *m, void *unused)
 
 	up_write(&adev->reset_domain->sem);
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return 0;
@@ -1721,7 +1700,6 @@ static int amdgpu_debugfs_evict_vram(void *data, u64 *val)
 
 	*val = amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM);
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return 0;
@@ -1742,7 +1720,6 @@ static int amdgpu_debugfs_evict_gtt(void *data, u64 *val)
 
 	*val = amdgpu_ttm_evict_resources(adev, TTM_PL_TT);
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return 0;
@@ -1762,7 +1739,6 @@ static int amdgpu_debugfs_benchmark(void *data, u64 val)
 
 	r = amdgpu_benchmark(adev, val);
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return r;
@@ -1786,7 +1762,7 @@ static int amdgpu_debugfs_vm_info_show(struct seq_file *m, void *unused)
 
 		ti = amdgpu_vm_get_task_info_vm(vm);
 		if (ti) {
-			seq_printf(m, "pid:%d\tProcess:%s ----------\n", ti->pid, ti->process_name);
+			seq_printf(m, "pid:%d\tProcess:%s ----------\n", ti->task.pid, ti->process_name);
 			amdgpu_vm_put_task_info(ti);
 		}
 
@@ -1902,7 +1878,7 @@ no_preempt:
 			continue;
 		}
 		job = to_amdgpu_job(s_job);
-		if (preempted && (&job->hw_fence) == fence)
+		if (preempted && (&job->hw_fence->base) == fence)
 			/* mark the job as preempted */
 			job->preemption_status |= AMDGPU_IB_PREEMPTED;
 	}
@@ -1990,7 +1966,7 @@ static int amdgpu_debugfs_sclk_set(void *data, u64 val)
 	uint32_t max_freq, min_freq;
 	struct amdgpu_device *adev = (struct amdgpu_device *)data;
 
-	if (amdgpu_sriov_vf(adev) && !amdgpu_sriov_is_pp_one_vf(adev))
+	if (amdgpu_sriov_multi_vf_mode(adev))
 		return -EINVAL;
 
 	ret = pm_runtime_get_sync(adev_to_drm(adev)->dev);
@@ -2014,7 +1990,6 @@ static int amdgpu_debugfs_sclk_set(void *data, u64 val)
 		ret = -EINVAL;
 
 out:
-	pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 
 	return ret;
@@ -2105,6 +2080,7 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev)
 	amdgpu_rap_debugfs_init(adev);
 	amdgpu_securedisplay_debugfs_init(adev);
 	amdgpu_fw_attestation_debugfs_init(adev);
+	amdgpu_psp_debugfs_init(adev);
 
 	debugfs_create_file("amdgpu_evict_vram", 0400, root, adev,
 			    &amdgpu_evict_vram_fops);
@@ -2122,14 +2098,68 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev)
 	debugfs_create_blob("amdgpu_vbios", 0444, root,
 			    &adev->debugfs_vbios_blob);
 
-	adev->debugfs_discovery_blob.data = adev->mman.discovery_bin;
-	adev->debugfs_discovery_blob.size = adev->mman.discovery_tmr_size;
-	debugfs_create_blob("amdgpu_discovery", 0444, root,
-			    &adev->debugfs_discovery_blob);
+	if (adev->discovery.debugfs_blob.size)
+		debugfs_create_blob("amdgpu_discovery", 0444, root,
+				    &adev->discovery.debugfs_blob);
 
 	return 0;
 }
 
+static int amdgpu_pt_info_read(struct seq_file *m, void *unused)
+{
+	struct drm_file *file;
+	struct amdgpu_fpriv *fpriv;
+	struct amdgpu_bo *root_bo;
+	struct amdgpu_device *adev;
+	int r;
+
+	file = m->private;
+	if (!file)
+		return -EINVAL;
+
+	adev = drm_to_adev(file->minor->dev);
+	fpriv = file->driver_priv;
+	if (!fpriv || !fpriv->vm.root.bo)
+		return -ENODEV;
+
+	root_bo = amdgpu_bo_ref(fpriv->vm.root.bo);
+	r = amdgpu_bo_reserve(root_bo, true);
+	if (r) {
+		amdgpu_bo_unref(&root_bo);
+		return -EINVAL;
+	}
+
+	seq_printf(m, "pd_address: 0x%llx\n", amdgpu_gmc_pd_addr(fpriv->vm.root.bo));
+	seq_printf(m, "max_pfn: 0x%llx\n", adev->vm_manager.max_pfn);
+	seq_printf(m, "num_level: 0x%x\n", adev->vm_manager.num_level);
+	seq_printf(m, "block_size: 0x%x\n", adev->vm_manager.block_size);
+	seq_printf(m, "fragment_size: 0x%x\n", adev->vm_manager.fragment_size);
+
+	amdgpu_bo_unreserve(root_bo);
+	amdgpu_bo_unref(&root_bo);
+
+	return 0;
+}
+
+static int amdgpu_pt_info_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, amdgpu_pt_info_read, inode->i_private);
+}
+
+static const struct file_operations amdgpu_pt_info_fops = {
+	.owner = THIS_MODULE,
+	.open = amdgpu_pt_info_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+void amdgpu_debugfs_vm_init(struct drm_file *file)
+{
+	debugfs_create_file("vm_pagetable_info", 0444, file->debugfs_client, file,
+			    &amdgpu_pt_info_fops);
+}
+
 #else
 int amdgpu_debugfs_init(struct amdgpu_device *adev)
 {
@@ -2139,4 +2169,7 @@ int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
 {
 	return 0;
 }
+void amdgpu_debugfs_vm_init(struct drm_file *file)
+{
+}
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h
index 0425432d8659..e7b3c38e5186 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h
@@ -33,4 +33,5 @@ void amdgpu_debugfs_fence_init(struct amdgpu_device *adev);
 void amdgpu_debugfs_firmware_init(struct amdgpu_device *adev);
 void amdgpu_debugfs_gem_init(struct amdgpu_device *adev);
 void amdgpu_debugfs_mes_event_log_init(struct amdgpu_device *adev);
+void amdgpu_debugfs_vm_init(struct drm_file *file);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c
index 824f9da5b6ce..4e2fe6674db8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c
@@ -217,13 +217,12 @@ amdgpu_devcoredump_read(char *buffer, loff_t offset, size_t count,
 	drm_printf(&p, "version: " AMDGPU_COREDUMP_VERSION "\n");
 	drm_printf(&p, "kernel: " UTS_RELEASE "\n");
 	drm_printf(&p, "module: " KBUILD_MODNAME "\n");
-	drm_printf(&p, "time: %lld.%09ld\n", coredump->reset_time.tv_sec,
-		   coredump->reset_time.tv_nsec);
+	drm_printf(&p, "time: %ptSp\n", &coredump->reset_time);
 
-	if (coredump->reset_task_info.pid)
+	if (coredump->reset_task_info.task.pid)
 		drm_printf(&p, "process_name: %s PID: %d\n",
 			   coredump->reset_task_info.process_name,
-			   coredump->reset_task_info.pid);
+			   coredump->reset_task_info.task.pid);
 
 	/* SOC Information */
 	drm_printf(&p, "\nSOC Information\n");
@@ -364,5 +363,9 @@ void amdgpu_coredump(struct amdgpu_device *adev, bool skip_vram_check,
 
 	dev_coredumpm(dev->dev, THIS_MODULE, coredump, 0, GFP_NOWAIT,
 		      amdgpu_devcoredump_read, amdgpu_devcoredump_free);
+
+	drm_info(dev, "AMDGPU device coredump file has been created\n");
+	drm_info(dev, "Check your /sys/class/drm/card%d/device/devcoredump/data\n",
+		 dev->primary->index);
 }
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index d100bb7a137c..58c3ffe707d1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -71,6 +71,7 @@
 
 #include "amdgpu_xgmi.h"
 #include "amdgpu_ras.h"
+#include "amdgpu_ras_mgr.h"
 #include "amdgpu_pmu.h"
 #include "amdgpu_fru_eeprom.h"
 #include "amdgpu_reset.h"
@@ -85,6 +86,7 @@
 
 #if IS_ENABLED(CONFIG_X86)
 #include <asm/intel-family.h>
+#include <asm/cpu_device_id.h>
 #endif
 
 MODULE_FIRMWARE("amdgpu/vega10_gpu_info.bin");
@@ -94,6 +96,7 @@ MODULE_FIRMWARE("amdgpu/picasso_gpu_info.bin");
 MODULE_FIRMWARE("amdgpu/raven2_gpu_info.bin");
 MODULE_FIRMWARE("amdgpu/arcturus_gpu_info.bin");
 MODULE_FIRMWARE("amdgpu/navi12_gpu_info.bin");
+MODULE_FIRMWARE("amdgpu/cyan_skillfish_gpu_info.bin");
 
 #define AMDGPU_RESUME_MS		2000
 #define AMDGPU_MAX_RETRY_LIMIT		2
@@ -102,6 +105,9 @@ MODULE_FIRMWARE("amdgpu/navi12_gpu_info.bin");
 #define AMDGPU_PCIE_INDEX_HI_FALLBACK (0x44 >> 2)
 #define AMDGPU_PCIE_DATA_FALLBACK (0x3C >> 2)
 
+#define AMDGPU_VBIOS_SKIP (1U << 0)
+#define AMDGPU_VBIOS_OPTIONAL (1U << 1)
+
 static const struct drm_driver amdgpu_kms_driver;
 
 const char *amdgpu_asic_name[] = {
@@ -174,6 +180,12 @@ struct amdgpu_init_level amdgpu_init_minimal_xgmi = {
 		BIT(AMD_IP_BLOCK_TYPE_PSP)
 };
 
+static int amdgpu_device_ip_resume_phase1(struct amdgpu_device *adev);
+static int amdgpu_device_ip_resume_phase2(struct amdgpu_device *adev);
+static int amdgpu_device_ip_resume_phase3(struct amdgpu_device *adev);
+
+static void amdgpu_device_load_switch_state(struct amdgpu_device *adev);
+
 static inline bool amdgpu_ip_member_of_hwini(struct amdgpu_device *adev,
 					     enum amd_ip_block_type block)
 {
@@ -224,8 +236,26 @@ static ssize_t amdgpu_device_get_pcie_replay_count(struct device *dev,
 static DEVICE_ATTR(pcie_replay_count, 0444,
 		amdgpu_device_get_pcie_replay_count, NULL);
 
+static int amdgpu_device_attr_sysfs_init(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	if (amdgpu_nbio_is_replay_cnt_supported(adev))
+		ret = sysfs_create_file(&adev->dev->kobj,
+					&dev_attr_pcie_replay_count.attr);
+
+	return ret;
+}
+
+static void amdgpu_device_attr_sysfs_fini(struct amdgpu_device *adev)
+{
+	if (amdgpu_nbio_is_replay_cnt_supported(adev))
+		sysfs_remove_file(&adev->dev->kobj,
+				  &dev_attr_pcie_replay_count.attr);
+}
+
 static ssize_t amdgpu_sysfs_reg_state_get(struct file *f, struct kobject *kobj,
-					  struct bin_attribute *attr, char *buf,
+					  const struct bin_attribute *attr, char *buf,
 					  loff_t ppos, size_t count)
 {
 	struct device *dev = kobj_to_dev(kobj);
@@ -261,8 +291,8 @@ static ssize_t amdgpu_sysfs_reg_state_get(struct file *f, struct kobject *kobj,
 	return bytes_read;
 }
 
-BIN_ATTR(reg_state, 0444, amdgpu_sysfs_reg_state_get, NULL,
-	 AMDGPU_SYS_REG_STATE_END);
+static const BIN_ATTR(reg_state, 0444, amdgpu_sysfs_reg_state_get, NULL,
+		      AMDGPU_SYS_REG_STATE_END);
 
 int amdgpu_reg_state_sysfs_init(struct amdgpu_device *adev)
 {
@@ -389,19 +419,16 @@ static const struct attribute_group amdgpu_board_attrs_group = {
 
 static void amdgpu_device_get_pcie_info(struct amdgpu_device *adev);
 
-
 /**
  * amdgpu_device_supports_px - Is the device a dGPU with ATPX power control
  *
- * @dev: drm_device pointer
+ * @adev: amdgpu device pointer
  *
  * Returns true if the device is a dGPU with ATPX power control,
  * otherwise return false.
  */
-bool amdgpu_device_supports_px(struct drm_device *dev)
+bool amdgpu_device_supports_px(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = drm_to_adev(dev);
-
 	if ((adev->flags & AMD_IS_PX) && !amdgpu_is_atpx_hybrid())
 		return true;
 	return false;
@@ -410,15 +437,13 @@ bool amdgpu_device_supports_px(struct drm_device *dev)
 /**
  * amdgpu_device_supports_boco - Is the device a dGPU with ACPI power resources
  *
- * @dev: drm_device pointer
+ * @adev: amdgpu device pointer
  *
  * Returns true if the device is a dGPU with ACPI power control,
  * otherwise return false.
  */
-bool amdgpu_device_supports_boco(struct drm_device *dev)
+bool amdgpu_device_supports_boco(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = drm_to_adev(dev);
-
 	if (!IS_ENABLED(CONFIG_HOTPLUG_PCI_PCIE))
 		return false;
 
@@ -431,29 +456,24 @@ bool amdgpu_device_supports_boco(struct drm_device *dev)
 /**
  * amdgpu_device_supports_baco - Does the device support BACO
  *
- * @dev: drm_device pointer
+ * @adev: amdgpu device pointer
  *
  * Return:
  * 1 if the device supports BACO;
  * 3 if the device supports MACO (only works if BACO is supported)
  * otherwise return 0.
  */
-int amdgpu_device_supports_baco(struct drm_device *dev)
+int amdgpu_device_supports_baco(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = drm_to_adev(dev);
-
 	return amdgpu_asic_supports_baco(adev);
 }
 
 void amdgpu_device_detect_runtime_pm_mode(struct amdgpu_device *adev)
 {
-	struct drm_device *dev;
 	int bamaco_support;
 
-	dev = adev_to_drm(adev);
-
 	adev->pm.rpm_mode = AMDGPU_RUNPM_NONE;
-	bamaco_support = amdgpu_device_supports_baco(dev);
+	bamaco_support = amdgpu_device_supports_baco(adev);
 
 	switch (amdgpu_runtime_pm) {
 	case 2:
@@ -473,10 +493,12 @@ void amdgpu_device_detect_runtime_pm_mode(struct amdgpu_device *adev)
 		break;
 	case -1:
 	case -2:
-		if (amdgpu_device_supports_px(dev)) { /* enable PX as runtime mode */
+		if (amdgpu_device_supports_px(adev)) {
+			/* enable PX as runtime mode */
 			adev->pm.rpm_mode = AMDGPU_RUNPM_PX;
 			dev_info(adev->dev, "Using ATPX for runtime pm\n");
-		} else if (amdgpu_device_supports_boco(dev)) { /* enable boco as runtime mode */
+		} else if (amdgpu_device_supports_boco(adev)) {
+			/* enable boco as runtime mode */
 			adev->pm.rpm_mode = AMDGPU_RUNPM_BOCO;
 			dev_info(adev->dev, "Using BOCO for runtime pm\n");
 		} else {
@@ -490,12 +512,13 @@ void amdgpu_device_detect_runtime_pm_mode(struct amdgpu_device *adev)
 				break;
 			case CHIP_VEGA10:
 				/* enable BACO as runpm mode if noretry=0 */
-				if (!adev->gmc.noretry)
+				if (!adev->gmc.noretry && !amdgpu_passthrough(adev))
 					adev->pm.rpm_mode = AMDGPU_RUNPM_BACO;
 				break;
 			default:
 				/* enable BACO as runpm mode on CI+ */
-				adev->pm.rpm_mode = AMDGPU_RUNPM_BACO;
+				if (!amdgpu_passthrough(adev))
+					adev->pm.rpm_mode = AMDGPU_RUNPM_BACO;
 				break;
 			}
 
@@ -524,14 +547,14 @@ no_runtime_pm:
  * amdgpu_device_supports_smart_shift - Is the device dGPU with
  * smart shift support
  *
- * @dev: drm_device pointer
+ * @adev: amdgpu device pointer
  *
  * Returns true if the device is a dGPU with Smart Shift support,
  * otherwise returns false.
  */
-bool amdgpu_device_supports_smart_shift(struct drm_device *dev)
+bool amdgpu_device_supports_smart_shift(struct amdgpu_device *adev)
 {
-	return (amdgpu_device_supports_boco(dev) &&
+	return (amdgpu_device_supports_boco(adev) &&
 		amdgpu_acpi_is_power_shift_control_supported());
 }
 
@@ -1265,14 +1288,14 @@ u32 amdgpu_device_get_rev_id(struct amdgpu_device *adev)
  */
 static uint32_t amdgpu_invalid_rreg(struct amdgpu_device *adev, uint32_t reg)
 {
-	DRM_ERROR("Invalid callback to read register 0x%04X\n", reg);
+	dev_err(adev->dev, "Invalid callback to read register 0x%04X\n", reg);
 	BUG();
 	return 0;
 }
 
 static uint32_t amdgpu_invalid_rreg_ext(struct amdgpu_device *adev, uint64_t reg)
 {
-	DRM_ERROR("Invalid callback to read register 0x%llX\n", reg);
+	dev_err(adev->dev, "Invalid callback to read register 0x%llX\n", reg);
 	BUG();
 	return 0;
 }
@@ -1289,15 +1312,17 @@ static uint32_t amdgpu_invalid_rreg_ext(struct amdgpu_device *adev, uint64_t reg
  */
 static void amdgpu_invalid_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v)
 {
-	DRM_ERROR("Invalid callback to write register 0x%04X with 0x%08X\n",
-		  reg, v);
+	dev_err(adev->dev,
+		"Invalid callback to write register 0x%04X with 0x%08X\n", reg,
+		v);
 	BUG();
 }
 
 static void amdgpu_invalid_wreg_ext(struct amdgpu_device *adev, uint64_t reg, uint32_t v)
 {
-	DRM_ERROR("Invalid callback to write register 0x%llX with 0x%08X\n",
-		  reg, v);
+	dev_err(adev->dev,
+		"Invalid callback to write register 0x%llX with 0x%08X\n", reg,
+		v);
 	BUG();
 }
 
@@ -1313,14 +1338,15 @@ static void amdgpu_invalid_wreg_ext(struct amdgpu_device *adev, uint64_t reg, ui
  */
 static uint64_t amdgpu_invalid_rreg64(struct amdgpu_device *adev, uint32_t reg)
 {
-	DRM_ERROR("Invalid callback to read 64 bit register 0x%04X\n", reg);
+	dev_err(adev->dev, "Invalid callback to read 64 bit register 0x%04X\n",
+		reg);
 	BUG();
 	return 0;
 }
 
 static uint64_t amdgpu_invalid_rreg64_ext(struct amdgpu_device *adev, uint64_t reg)
 {
-	DRM_ERROR("Invalid callback to read register 0x%llX\n", reg);
+	dev_err(adev->dev, "Invalid callback to read register 0x%llX\n", reg);
 	BUG();
 	return 0;
 }
@@ -1337,15 +1363,17 @@ static uint64_t amdgpu_invalid_rreg64_ext(struct amdgpu_device *adev, uint64_t r
  */
 static void amdgpu_invalid_wreg64(struct amdgpu_device *adev, uint32_t reg, uint64_t v)
 {
-	DRM_ERROR("Invalid callback to write 64 bit register 0x%04X with 0x%08llX\n",
-		  reg, v);
+	dev_err(adev->dev,
+		"Invalid callback to write 64 bit register 0x%04X with 0x%08llX\n",
+		reg, v);
 	BUG();
 }
 
 static void amdgpu_invalid_wreg64_ext(struct amdgpu_device *adev, uint64_t reg, uint64_t v)
 {
-	DRM_ERROR("Invalid callback to write 64 bit register 0x%llX with 0x%08llX\n",
-		  reg, v);
+	dev_err(adev->dev,
+		"Invalid callback to write 64 bit register 0x%llX with 0x%08llX\n",
+		reg, v);
 	BUG();
 }
 
@@ -1363,8 +1391,9 @@ static void amdgpu_invalid_wreg64_ext(struct amdgpu_device *adev, uint64_t reg,
 static uint32_t amdgpu_block_invalid_rreg(struct amdgpu_device *adev,
 					  uint32_t block, uint32_t reg)
 {
-	DRM_ERROR("Invalid callback to read register 0x%04X in block 0x%04X\n",
-		  reg, block);
+	dev_err(adev->dev,
+		"Invalid callback to read register 0x%04X in block 0x%04X\n",
+		reg, block);
 	BUG();
 	return 0;
 }
@@ -1384,11 +1413,23 @@ static void amdgpu_block_invalid_wreg(struct amdgpu_device *adev,
 				      uint32_t block,
 				      uint32_t reg, uint32_t v)
 {
-	DRM_ERROR("Invalid block callback to write register 0x%04X in block 0x%04X with 0x%08X\n",
-		  reg, block, v);
+	dev_err(adev->dev,
+		"Invalid block callback to write register 0x%04X in block 0x%04X with 0x%08X\n",
+		reg, block, v);
 	BUG();
 }
 
+static uint32_t amdgpu_device_get_vbios_flags(struct amdgpu_device *adev)
+{
+	if (hweight32(adev->aid_mask) && (adev->flags & AMD_IS_APU))
+		return AMDGPU_VBIOS_SKIP;
+
+	if (hweight32(adev->aid_mask) && amdgpu_passthrough(adev))
+		return AMDGPU_VBIOS_OPTIONAL;
+
+	return 0;
+}
+
 /**
  * amdgpu_device_asic_init - Wrapper for atom asic_init
  *
@@ -1398,18 +1439,28 @@ static void amdgpu_block_invalid_wreg(struct amdgpu_device *adev,
  */
 static int amdgpu_device_asic_init(struct amdgpu_device *adev)
 {
+	uint32_t flags;
+	bool optional;
 	int ret;
 
 	amdgpu_asic_pre_asic_init(adev);
+	flags = amdgpu_device_get_vbios_flags(adev);
+	optional = !!(flags & (AMDGPU_VBIOS_OPTIONAL | AMDGPU_VBIOS_SKIP));
 
 	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
 	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
 	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0) ||
 	    amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0)) {
 		amdgpu_psp_wait_for_bootloader(adev);
+		if (optional && !adev->bios)
+			return 0;
+
 		ret = amdgpu_atomfirmware_asic_init(adev, true);
 		return ret;
 	} else {
+		if (optional && !adev->bios)
+			return 0;
+
 		return amdgpu_atom_asic_init(adev->mode_info.atom_context);
 	}
 
@@ -1627,9 +1678,9 @@ int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev)
 	int rbar_size = pci_rebar_bytes_to_size(adev->gmc.real_vram_size);
 	struct pci_bus *root;
 	struct resource *res;
+	int max_size, r;
 	unsigned int i;
 	u16 cmd;
-	int r;
 
 	if (!IS_ENABLED(CONFIG_PHYS_ADDR_T_64BIT))
 		return 0;
@@ -1638,9 +1689,21 @@ int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev)
 	if (amdgpu_sriov_vf(adev))
 		return 0;
 
+	if (!amdgpu_rebar)
+		return 0;
+
+	/* resizing on Dell G5 SE platforms causes problems with runtime pm */
+	if ((amdgpu_runtime_pm != 0) &&
+	    adev->pdev->vendor == PCI_VENDOR_ID_ATI &&
+	    adev->pdev->device == 0x731f &&
+	    adev->pdev->subsystem_vendor == PCI_VENDOR_ID_DELL)
+		return 0;
+
 	/* PCI_EXT_CAP_ID_VNDR extended capability is located at 0x100 */
 	if (!pci_find_ext_capability(adev->pdev, PCI_EXT_CAP_ID_VNDR))
-		DRM_WARN("System can't access extended configuration space, please check!!\n");
+		dev_warn(
+			adev->dev,
+			"System can't access extended configuration space, please check!!\n");
 
 	/* skip if the bios has already enabled large BAR */
 	if (adev->gmc.real_vram_size &&
@@ -1663,28 +1726,27 @@ int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev)
 		return 0;
 
 	/* Limit the BAR size to what is available */
-	rbar_size = min(fls(pci_rebar_get_possible_sizes(adev->pdev, 0)) - 1,
-			rbar_size);
+	max_size = pci_rebar_get_max_size(adev->pdev, 0);
+	if (max_size < 0)
+		return 0;
+	rbar_size = min(max_size, rbar_size);
 
 	/* Disable memory decoding while we change the BAR addresses and size */
 	pci_read_config_word(adev->pdev, PCI_COMMAND, &cmd);
 	pci_write_config_word(adev->pdev, PCI_COMMAND,
 			      cmd & ~PCI_COMMAND_MEMORY);
 
-	/* Free the VRAM and doorbell BAR, we most likely need to move both. */
+	/* Tear down doorbell as resizing will release BARs */
 	amdgpu_doorbell_fini(adev);
-	if (adev->asic_type >= CHIP_BONAIRE)
-		pci_release_resource(adev->pdev, 2);
-
-	pci_release_resource(adev->pdev, 0);
 
-	r = pci_resize_resource(adev->pdev, 0, rbar_size);
+	r = pci_resize_resource(adev->pdev, 0, rbar_size,
+				(adev->asic_type >= CHIP_BONAIRE) ? 1 << 5
+								  : 1 << 2);
 	if (r == -ENOSPC)
-		DRM_INFO("Not enough PCI address space for a large BAR.");
+		dev_info(adev->dev,
+			 "Not enough PCI address space for a large BAR.");
 	else if (r && r != -ENOTSUPP)
-		DRM_ERROR("Problem resizing BAR0 (%d).", r);
-
-	pci_assign_unassigned_bus_resources(adev->pdev->bus);
+		dev_err(adev->dev, "Problem resizing BAR0 (%d).", r);
 
 	/* When the doorbell or fb BAR isn't available we have no chance of
 	 * using the device.
@@ -1698,14 +1760,6 @@ int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev)
 	return 0;
 }
 
-static bool amdgpu_device_read_bios(struct amdgpu_device *adev)
-{
-	if (hweight32(adev->aid_mask) && (adev->flags & AMD_IS_APU))
-		return false;
-
-	return true;
-}
-
 /*
  * GPU helpers function.
  */
@@ -1720,12 +1774,15 @@ static bool amdgpu_device_read_bios(struct amdgpu_device *adev)
  */
 bool amdgpu_device_need_post(struct amdgpu_device *adev)
 {
-	uint32_t reg;
+	uint32_t reg, flags;
 
 	if (amdgpu_sriov_vf(adev))
 		return false;
 
-	if (!amdgpu_device_read_bios(adev))
+	flags = amdgpu_device_get_vbios_flags(adev);
+	if (flags & AMDGPU_VBIOS_SKIP)
+		return false;
+	if ((flags & AMDGPU_VBIOS_OPTIONAL) && !adev->bios)
 		return false;
 
 	if (amdgpu_passthrough(adev)) {
@@ -1789,8 +1846,8 @@ bool amdgpu_device_seamless_boot_supported(struct amdgpu_device *adev)
 	case 0:
 		return false;
 	default:
-		DRM_ERROR("Invalid value for amdgpu.seamless: %d\n",
-			  amdgpu_seamless);
+		dev_err(adev->dev, "Invalid value for amdgpu.seamless: %d\n",
+			amdgpu_seamless);
 		return false;
 	}
 
@@ -1826,6 +1883,42 @@ static bool amdgpu_device_pcie_dynamic_switching_supported(struct amdgpu_device
 	return true;
 }
 
+static bool amdgpu_device_aspm_support_quirk(struct amdgpu_device *adev)
+{
+	/* Enabling ASPM causes randoms hangs on Tahiti and Oland on Zen4.
+	 * It's unclear if this is a platform-specific or GPU-specific issue.
+	 * Disable ASPM on SI for the time being.
+	 */
+	if (adev->family == AMDGPU_FAMILY_SI)
+		return true;
+
+#if IS_ENABLED(CONFIG_X86)
+	struct cpuinfo_x86 *c = &cpu_data(0);
+
+	if (!(amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(12, 0, 0) ||
+		  amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(12, 0, 1)))
+		return false;
+
+	if (c->x86 == 6 &&
+		adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN5) {
+		switch (c->x86_model) {
+		case VFM_MODEL(INTEL_ALDERLAKE):
+		case VFM_MODEL(INTEL_ALDERLAKE_L):
+		case VFM_MODEL(INTEL_RAPTORLAKE):
+		case VFM_MODEL(INTEL_RAPTORLAKE_P):
+		case VFM_MODEL(INTEL_RAPTORLAKE_S):
+			return true;
+		default:
+			return false;
+		}
+	} else {
+		return false;
+	}
+#else
+	return false;
+#endif
+}
+
 /**
  * amdgpu_device_should_use_aspm - check if the device should program ASPM
  *
@@ -1850,7 +1943,7 @@ bool amdgpu_device_should_use_aspm(struct amdgpu_device *adev)
 	}
 	if (adev->flags & AMD_IS_APU)
 		return false;
-	if (!(adev->pm.pp_feature & PP_PCIE_DPM_MASK))
+	if (amdgpu_device_aspm_support_quirk(adev))
 		return false;
 	return pcie_aspm_enabled(adev->pdev);
 }
@@ -1937,7 +2030,7 @@ static void amdgpu_device_check_smu_prv_buffer_size(struct amdgpu_device *adev)
 		return;
 
 	if (!is_os_64) {
-		DRM_WARN("Not 64-bit OS, feature not supported\n");
+		dev_warn(adev->dev, "Not 64-bit OS, feature not supported\n");
 		goto def_value;
 	}
 	si_meminfo(&si);
@@ -1952,7 +2045,7 @@ static void amdgpu_device_check_smu_prv_buffer_size(struct amdgpu_device *adev)
 		if (total_memory < dram_size_seven_GB)
 			goto def_value1;
 	} else {
-		DRM_WARN("Smu memory pool size not supported\n");
+		dev_warn(adev->dev, "Smu memory pool size not supported\n");
 		goto def_value;
 	}
 	adev->pm.smu_prv_buffer_size = amdgpu_smu_memory_pool_size << 28;
@@ -1960,7 +2053,7 @@ static void amdgpu_device_check_smu_prv_buffer_size(struct amdgpu_device *adev)
 	return;
 
 def_value1:
-	DRM_WARN("No enough system memory\n");
+	dev_warn(adev->dev, "No enough system memory\n");
 def_value:
 	adev->pm.smu_prv_buffer_size = 0;
 }
@@ -2068,8 +2161,31 @@ static int amdgpu_device_check_arguments(struct amdgpu_device *adev)
 
 	adev->firmware.load_type = amdgpu_ucode_get_load_type(adev, amdgpu_fw_load_type);
 
-	for (i = 0; i < MAX_XCP; i++)
-		adev->enforce_isolation[i] = !!enforce_isolation;
+	for (i = 0; i < MAX_XCP; i++) {
+		switch (amdgpu_enforce_isolation) {
+		case -1:
+		case 0:
+		default:
+			/* disable */
+			adev->enforce_isolation[i] = AMDGPU_ENFORCE_ISOLATION_DISABLE;
+			break;
+		case 1:
+			/* enable */
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_ENABLE;
+			break;
+		case 2:
+			/* enable legacy mode */
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_ENABLE_LEGACY;
+			break;
+		case 3:
+			/* enable only process isolation without submitting cleaner shader */
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_NO_CLEANER_SHADER;
+			break;
+		}
+	}
 
 	return 0;
 }
@@ -2089,7 +2205,8 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev,
 	struct drm_device *dev = pci_get_drvdata(pdev);
 	int r;
 
-	if (amdgpu_device_supports_px(dev) && state == VGA_SWITCHEROO_OFF)
+	if (amdgpu_device_supports_px(drm_to_adev(dev)) &&
+	    state == VGA_SWITCHEROO_OFF)
 		return;
 
 	if (state == VGA_SWITCHEROO_ON) {
@@ -2101,12 +2218,13 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev,
 		amdgpu_device_load_pci_state(pdev);
 		r = pci_enable_device(pdev);
 		if (r)
-			DRM_WARN("pci_enable_device failed (%d)\n", r);
+			dev_warn(&pdev->dev, "pci_enable_device failed (%d)\n",
+				 r);
 		amdgpu_device_resume(dev, true);
 
 		dev->switch_power_state = DRM_SWITCH_POWER_ON;
 	} else {
-		pr_info("switched off\n");
+		dev_info(&pdev->dev, "switched off\n");
 		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
 		amdgpu_device_prepare(dev);
 		amdgpu_device_suspend(dev, true);
@@ -2173,8 +2291,9 @@ int amdgpu_device_ip_set_clockgating_state(void *dev,
 		r = adev->ip_blocks[i].version->funcs->set_clockgating_state(
 			&adev->ip_blocks[i], state);
 		if (r)
-			DRM_ERROR("set_clockgating_state of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+			dev_err(adev->dev,
+				"set_clockgating_state of IP block <%s> failed %d\n",
+				adev->ip_blocks[i].version->funcs->name, r);
 	}
 	return r;
 }
@@ -2207,8 +2326,9 @@ int amdgpu_device_ip_set_powergating_state(void *dev,
 		r = adev->ip_blocks[i].version->funcs->set_powergating_state(
 			&adev->ip_blocks[i], state);
 		if (r)
-			DRM_ERROR("set_powergating_state of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+			dev_err(adev->dev,
+				"set_powergating_state of IP block <%s> failed %d\n",
+				adev->ip_blocks[i].version->funcs->name, r);
 	}
 	return r;
 }
@@ -2233,7 +2353,8 @@ void amdgpu_device_ip_get_clockgating_state(struct amdgpu_device *adev,
 		if (!adev->ip_blocks[i].status.valid)
 			continue;
 		if (adev->ip_blocks[i].version->funcs->get_clockgating_state)
-			adev->ip_blocks[i].version->funcs->get_clockgating_state((void *)adev, flags);
+			adev->ip_blocks[i].version->funcs->get_clockgating_state(
+				&adev->ip_blocks[i], flags);
 	}
 }
 
@@ -2269,7 +2390,7 @@ int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev,
 }
 
 /**
- * amdgpu_device_ip_is_valid - is the hardware IP enabled
+ * amdgpu_device_ip_is_hw - is the hardware IP enabled
  *
  * @adev: amdgpu_device pointer
  * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
@@ -2277,6 +2398,27 @@ int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev,
  * Check if the hardware IP is enable or not.
  * Returns true if it the IP is enable, false if not.
  */
+bool amdgpu_device_ip_is_hw(struct amdgpu_device *adev,
+			    enum amd_ip_block_type block_type)
+{
+	int i;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (adev->ip_blocks[i].version->type == block_type)
+			return adev->ip_blocks[i].status.hw;
+	}
+	return false;
+}
+
+/**
+ * amdgpu_device_ip_is_valid - is the hardware IP valid
+ *
+ * @adev: amdgpu_device pointer
+ * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ *
+ * Check if the hardware IP is valid or not.
+ * Returns true if it the IP is valid, false if not.
+ */
 bool amdgpu_device_ip_is_valid(struct amdgpu_device *adev,
 			       enum amd_ip_block_type block_type)
 {
@@ -2337,6 +2479,34 @@ int amdgpu_device_ip_block_version_cmp(struct amdgpu_device *adev,
 	return 1;
 }
 
+static const char *ip_block_names[] = {
+	[AMD_IP_BLOCK_TYPE_COMMON] = "common",
+	[AMD_IP_BLOCK_TYPE_GMC] = "gmc",
+	[AMD_IP_BLOCK_TYPE_IH] = "ih",
+	[AMD_IP_BLOCK_TYPE_SMC] = "smu",
+	[AMD_IP_BLOCK_TYPE_PSP] = "psp",
+	[AMD_IP_BLOCK_TYPE_DCE] = "dce",
+	[AMD_IP_BLOCK_TYPE_GFX] = "gfx",
+	[AMD_IP_BLOCK_TYPE_SDMA] = "sdma",
+	[AMD_IP_BLOCK_TYPE_UVD] = "uvd",
+	[AMD_IP_BLOCK_TYPE_VCE] = "vce",
+	[AMD_IP_BLOCK_TYPE_ACP] = "acp",
+	[AMD_IP_BLOCK_TYPE_VCN] = "vcn",
+	[AMD_IP_BLOCK_TYPE_MES] = "mes",
+	[AMD_IP_BLOCK_TYPE_JPEG] = "jpeg",
+	[AMD_IP_BLOCK_TYPE_VPE] = "vpe",
+	[AMD_IP_BLOCK_TYPE_UMSCH_MM] = "umsch_mm",
+	[AMD_IP_BLOCK_TYPE_ISP] = "isp",
+	[AMD_IP_BLOCK_TYPE_RAS] = "ras",
+};
+
+static const char *ip_block_name(struct amdgpu_device *adev, enum amd_ip_block_type type)
+{
+	int idx = (int)type;
+
+	return idx < ARRAY_SIZE(ip_block_names) ? ip_block_names[idx] : "unknown";
+}
+
 /**
  * amdgpu_device_ip_block_add
  *
@@ -2365,8 +2535,13 @@ int amdgpu_device_ip_block_add(struct amdgpu_device *adev,
 		break;
 	}
 
-	dev_info(adev->dev, "detected ip block number %d <%s>\n",
-		 adev->num_ip_blocks, ip_block_version->funcs->name);
+	dev_info(adev->dev, "detected ip block number %d <%s_v%d_%d_%d> (%s)\n",
+		 adev->num_ip_blocks,
+		 ip_block_name(adev, ip_block_version->type),
+		 ip_block_version->major,
+		 ip_block_version->minor,
+		 ip_block_version->rev,
+		 ip_block_version->funcs->name);
 
 	adev->ip_blocks[adev->num_ip_blocks].adev = adev;
 
@@ -2423,9 +2598,11 @@ static void amdgpu_device_enable_virtual_display(struct amdgpu_device *adev)
 			}
 		}
 
-		DRM_INFO("virtual display string:%s, %s:virtual_display:%d, num_crtc:%d\n",
-			 amdgpu_virtual_display, pci_address_name,
-			 adev->enable_virtual_display, adev->mode_info.num_crtc);
+		dev_info(
+			adev->dev,
+			"virtual display string:%s, %s:virtual_display:%d, num_crtc:%d\n",
+			amdgpu_virtual_display, pci_address_name,
+			adev->enable_virtual_display, adev->mode_info.num_crtc);
 
 		kfree(pciaddstr);
 	}
@@ -2436,8 +2613,9 @@ void amdgpu_device_set_sriov_virtual_display(struct amdgpu_device *adev)
 	if (amdgpu_sriov_vf(adev) && !adev->enable_virtual_display) {
 		adev->mode_info.num_crtc = 1;
 		adev->enable_virtual_display = true;
-		DRM_INFO("virtual_display:%d, num_crtc:%d\n",
-			 adev->enable_virtual_display, adev->mode_info.num_crtc);
+		dev_info(adev->dev, "virtual_display:%d, num_crtc:%d\n",
+			 adev->enable_virtual_display,
+			 adev->mode_info.num_crtc);
 	}
 }
 
@@ -2459,9 +2637,6 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
 
 	adev->firmware.gpu_info_fw = NULL;
 
-	if (adev->mman.discovery_bin)
-		return 0;
-
 	switch (adev->asic_type) {
 	default:
 		return 0;
@@ -2483,8 +2658,15 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
 		chip_name = "arcturus";
 		break;
 	case CHIP_NAVI12:
+		if (adev->discovery.bin)
+			return 0;
 		chip_name = "navi12";
 		break;
+	case CHIP_CYAN_SKILLFISH:
+		if (adev->discovery.bin)
+			return 0;
+		chip_name = "cyan_skillfish";
+		break;
 	}
 
 	err = amdgpu_ucode_request(adev, &adev->firmware.gpu_info_fw,
@@ -2564,6 +2746,24 @@ out:
 	return err;
 }
 
+static void amdgpu_uid_init(struct amdgpu_device *adev)
+{
+	/* Initialize the UID for the device */
+	adev->uid_info = kzalloc(sizeof(struct amdgpu_uid), GFP_KERNEL);
+	if (!adev->uid_info) {
+		dev_warn(adev->dev, "Failed to allocate memory for UID\n");
+		return;
+	}
+	adev->uid_info->adev = adev;
+}
+
+static void amdgpu_uid_fini(struct amdgpu_device *adev)
+{
+	/* Free the UID memory */
+	kfree(adev->uid_info);
+	adev->uid_info = NULL;
+}
+
 /**
  * amdgpu_device_ip_early_init - run early init for hardware IPs
  *
@@ -2578,8 +2778,9 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 {
 	struct amdgpu_ip_block *ip_block;
 	struct pci_dev *parent;
+	bool total, skip_bios;
+	uint32_t bios_flags;
 	int i, r;
-	bool total;
 
 	amdgpu_device_enable_virtual_display(adev);
 
@@ -2587,6 +2788,10 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 		r = amdgpu_virt_request_full_gpu(adev, true);
 		if (r)
 			return r;
+
+		r = amdgpu_virt_init_critical_region(adev);
+		if (r)
+			return r;
 	}
 
 	switch (adev->asic_type) {
@@ -2643,6 +2848,13 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 		break;
 	}
 
+	/* Check for IP version 9.4.3 with A0 hardware */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) &&
+	    !amdgpu_device_get_rev_id(adev)) {
+		dev_err(adev->dev, "Unsupported A0 hardware\n");
+		return -ENODEV;	/* device unsupported - no device error */
+	}
+
 	if (amdgpu_has_atpx() &&
 	    (amdgpu_is_atpx_hybrid() ||
 	     amdgpu_has_atpx_dgpu_power_cntl()) &&
@@ -2655,7 +2867,6 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 		adev->has_pr3 = parent ? pci_pr3_present(parent) : false;
 	}
 
-
 	adev->pm.pp_feature = amdgpu_pp_feature_mask;
 	if (amdgpu_sriov_vf(adev) || sched_policy == KFD_SCHED_POLICY_NO_HWS)
 		adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
@@ -2664,21 +2875,29 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 	if (!amdgpu_device_pcie_dynamic_switching_supported(adev))
 		adev->pm.pp_feature &= ~PP_PCIE_DPM_MASK;
 
+	adev->virt.is_xgmi_node_migrate_enabled = false;
+	if (amdgpu_sriov_vf(adev)) {
+		adev->virt.is_xgmi_node_migrate_enabled =
+			amdgpu_ip_version((adev), GC_HWIP, 0) == IP_VERSION(9, 4, 4);
+	}
+
 	total = true;
 	for (i = 0; i < adev->num_ip_blocks; i++) {
 		ip_block = &adev->ip_blocks[i];
 
 		if ((amdgpu_ip_block_mask & (1 << i)) == 0) {
-			DRM_WARN("disabled ip block: %d <%s>\n",
-				  i, adev->ip_blocks[i].version->funcs->name);
+			dev_warn(adev->dev, "disabled ip block: %d <%s>\n", i,
+				 adev->ip_blocks[i].version->funcs->name);
 			adev->ip_blocks[i].status.valid = false;
 		} else if (ip_block->version->funcs->early_init) {
 			r = ip_block->version->funcs->early_init(ip_block);
 			if (r == -ENOENT) {
 				adev->ip_blocks[i].status.valid = false;
 			} else if (r) {
-				DRM_ERROR("early_init of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+				dev_err(adev->dev,
+					"early_init of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
 				total = false;
 			} else {
 				adev->ip_blocks[i].status.valid = true;
@@ -2692,16 +2911,31 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 			if (r)
 				return r;
 
+			bios_flags = amdgpu_device_get_vbios_flags(adev);
+			skip_bios = !!(bios_flags & AMDGPU_VBIOS_SKIP);
 			/* Read BIOS */
-			if (amdgpu_device_read_bios(adev)) {
-				if (!amdgpu_get_bios(adev))
+			if (!skip_bios) {
+				bool optional =
+					!!(bios_flags & AMDGPU_VBIOS_OPTIONAL);
+				if (!amdgpu_get_bios(adev) && !optional)
 					return -EINVAL;
 
-				r = amdgpu_atombios_init(adev);
-				if (r) {
-					dev_err(adev->dev, "amdgpu_atombios_init failed\n");
-					amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL, 0, 0);
-					return r;
+				if (optional && !adev->bios)
+					dev_info(
+						adev->dev,
+						"VBIOS image optional, proceeding without VBIOS image");
+
+				if (adev->bios) {
+					r = amdgpu_atombios_init(adev);
+					if (r) {
+						dev_err(adev->dev,
+							"amdgpu_atombios_init failed\n");
+						amdgpu_vf_error_put(
+							adev,
+							AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL,
+							0, 0);
+						return r;
+					}
 				}
 			}
 
@@ -2714,6 +2948,11 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 	if (!total)
 		return -ENODEV;
 
+	if (adev->gmc.xgmi.supported)
+		amdgpu_xgmi_early_init(adev);
+
+	if (amdgpu_is_multi_aid(adev))
+		amdgpu_uid_init(adev);
 	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX);
 	if (ip_block->status.valid != false)
 		amdgpu_amdkfd_device_probe(adev);
@@ -2741,8 +2980,10 @@ static int amdgpu_device_ip_hw_init_phase1(struct amdgpu_device *adev)
 		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH) {
 			r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]);
 			if (r) {
-				DRM_ERROR("hw_init of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+				dev_err(adev->dev,
+					"hw_init of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
 				return r;
 			}
 			adev->ip_blocks[i].status.hw = true;
@@ -2766,8 +3007,9 @@ static int amdgpu_device_ip_hw_init_phase2(struct amdgpu_device *adev)
 			continue;
 		r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]);
 		if (r) {
-			DRM_ERROR("hw_init of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+			dev_err(adev->dev,
+				"hw_init of IP block <%s> failed %d\n",
+				adev->ip_blocks[i].version->funcs->name, r);
 			return r;
 		}
 		adev->ip_blocks[i].status.hw = true;
@@ -2805,8 +3047,11 @@ static int amdgpu_device_fw_loading(struct amdgpu_device *adev)
 			} else {
 				r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]);
 				if (r) {
-					DRM_ERROR("hw_init of IP block <%s> failed %d\n",
-							  adev->ip_blocks[i].version->funcs->name, r);
+					dev_err(adev->dev,
+						"hw_init of IP block <%s> failed %d\n",
+						adev->ip_blocks[i]
+							.version->funcs->name,
+						r);
 					return r;
 				}
 				adev->ip_blocks[i].status.hw = true;
@@ -2823,6 +3068,12 @@ static int amdgpu_device_fw_loading(struct amdgpu_device *adev)
 
 static int amdgpu_device_init_schedulers(struct amdgpu_device *adev)
 {
+	struct drm_sched_init_args args = {
+		.ops = &amdgpu_sched_ops,
+		.num_rqs = DRM_SCHED_PRIORITY_COUNT,
+		.timeout_wq = adev->reset_domain->wq,
+		.dev = adev->dev,
+	};
 	long timeout;
 	int r, i;
 
@@ -2848,32 +3099,36 @@ static int amdgpu_device_init_schedulers(struct amdgpu_device *adev)
 			break;
 		}
 
-		r = drm_sched_init(&ring->sched, &amdgpu_sched_ops, NULL,
-				   DRM_SCHED_PRIORITY_COUNT,
-				   ring->num_hw_submission, 0,
-				   timeout, adev->reset_domain->wq,
-				   ring->sched_score, ring->name,
-				   adev->dev);
+		args.timeout = timeout;
+		args.credit_limit = ring->num_hw_submission;
+		args.score = ring->sched_score;
+		args.name = ring->name;
+
+		r = drm_sched_init(&ring->sched, &args);
 		if (r) {
-			DRM_ERROR("Failed to create scheduler on ring %s.\n",
-				  ring->name);
+			dev_err(adev->dev,
+				"Failed to create scheduler on ring %s.\n",
+				ring->name);
 			return r;
 		}
 		r = amdgpu_uvd_entity_init(adev, ring);
 		if (r) {
-			DRM_ERROR("Failed to create UVD scheduling entity on ring %s.\n",
-				  ring->name);
+			dev_err(adev->dev,
+				"Failed to create UVD scheduling entity on ring %s.\n",
+				ring->name);
 			return r;
 		}
 		r = amdgpu_vce_entity_init(adev, ring);
 		if (r) {
-			DRM_ERROR("Failed to create VCE scheduling entity on ring %s.\n",
-				  ring->name);
+			dev_err(adev->dev,
+				"Failed to create VCE scheduling entity on ring %s.\n",
+				ring->name);
 			return r;
 		}
 	}
 
-	amdgpu_xcp_update_partition_sched_list(adev);
+	if (adev->xcp_mgr)
+		amdgpu_xcp_update_partition_sched_list(adev);
 
 	return 0;
 }
@@ -2905,8 +3160,10 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
 		if (adev->ip_blocks[i].version->funcs->sw_init) {
 			r = adev->ip_blocks[i].version->funcs->sw_init(&adev->ip_blocks[i]);
 			if (r) {
-				DRM_ERROR("sw_init of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+				dev_err(adev->dev,
+					"sw_init of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
 				goto init_failed;
 			}
 		}
@@ -2920,7 +3177,8 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
 			/* need to do common hw init early so everything is set up for gmc */
 			r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]);
 			if (r) {
-				DRM_ERROR("hw_init %d failed %d\n", i, r);
+				dev_err(adev->dev, "hw_init %d failed %d\n", i,
+					r);
 				goto init_failed;
 			}
 			adev->ip_blocks[i].status.hw = true;
@@ -2932,17 +3190,21 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
 
 			r = amdgpu_device_mem_scratch_init(adev);
 			if (r) {
-				DRM_ERROR("amdgpu_mem_scratch_init failed %d\n", r);
+				dev_err(adev->dev,
+					"amdgpu_mem_scratch_init failed %d\n",
+					r);
 				goto init_failed;
 			}
 			r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]);
 			if (r) {
-				DRM_ERROR("hw_init %d failed %d\n", i, r);
+				dev_err(adev->dev, "hw_init %d failed %d\n", i,
+					r);
 				goto init_failed;
 			}
 			r = amdgpu_device_wb_init(adev);
 			if (r) {
-				DRM_ERROR("amdgpu_device_wb_init failed %d\n", r);
+				dev_err(adev->dev,
+					"amdgpu_device_wb_init failed %d\n", r);
 				goto init_failed;
 			}
 			adev->ip_blocks[i].status.hw = true;
@@ -2954,14 +3216,16 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
 							       AMDGPU_GEM_DOMAIN_GTT,
 							       AMDGPU_CSA_SIZE);
 				if (r) {
-					DRM_ERROR("allocate CSA failed %d\n", r);
+					dev_err(adev->dev,
+						"allocate CSA failed %d\n", r);
 					goto init_failed;
 				}
 			}
 
 			r = amdgpu_seq64_init(adev);
 			if (r) {
-				DRM_ERROR("allocate seq64 failed %d\n", r);
+				dev_err(adev->dev, "allocate seq64 failed %d\n",
+					r);
 				goto init_failed;
 			}
 		}
@@ -3056,6 +3320,9 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
 
 	amdgpu_fru_get_product_info(adev);
 
+	if (!amdgpu_sriov_vf(adev) || amdgpu_sriov_ras_cper_en(adev))
+		r = amdgpu_cper_init(adev);
+
 init_failed:
 
 	return r;
@@ -3099,6 +3366,8 @@ static bool amdgpu_device_check_vram_lost(struct amdgpu_device *adev)
 	 * always assumed to be lost.
 	 */
 	switch (amdgpu_asic_reset_method(adev)) {
+	case AMD_RESET_METHOD_LEGACY:
+	case AMD_RESET_METHOD_LINK:
 	case AMD_RESET_METHOD_BACO:
 	case AMD_RESET_METHOD_MODE1:
 		return true;
@@ -3147,8 +3416,10 @@ int amdgpu_device_set_cg_state(struct amdgpu_device *adev,
 			r = adev->ip_blocks[i].version->funcs->set_clockgating_state(&adev->ip_blocks[i],
 										     state);
 			if (r) {
-				DRM_ERROR("set_clockgating_state(gate) of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+				dev_err(adev->dev,
+					"set_clockgating_state(gate) of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
 				return r;
 			}
 		}
@@ -3174,18 +3445,21 @@ int amdgpu_device_set_pg_state(struct amdgpu_device *adev,
 		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX ||
 		     adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SDMA))
 			continue;
-		/* skip CG for VCE/UVD, it's handled specially */
+		/* skip CG for VCE/UVD/VPE, it's handled specially */
 		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD &&
 		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE &&
 		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCN &&
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VPE &&
 		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_JPEG &&
 		    adev->ip_blocks[i].version->funcs->set_powergating_state) {
 			/* enable powergating to save power */
 			r = adev->ip_blocks[i].version->funcs->set_powergating_state(&adev->ip_blocks[i],
 											state);
 			if (r) {
-				DRM_ERROR("set_powergating_state(gate) of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+				dev_err(adev->dev,
+					"set_powergating_state(gate) of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
 				return r;
 			}
 		}
@@ -3212,7 +3486,7 @@ static int amdgpu_device_enable_mgpu_fan_boost(void)
 	for (i = 0; i < mgpu_info.num_dgpu; i++) {
 		gpu_ins = &(mgpu_info.gpu_ins[i]);
 		adev = gpu_ins->adev;
-		if (!(adev->flags & AMD_IS_APU) &&
+		if (!(adev->flags & AMD_IS_APU || amdgpu_sriov_multi_vf_mode(adev)) &&
 		    !gpu_ins->mgpu_fan_enabled) {
 			ret = amdgpu_dpm_enable_mgpu_fan_boost(adev);
 			if (ret)
@@ -3251,8 +3525,10 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev)
 		if (adev->ip_blocks[i].version->funcs->late_init) {
 			r = adev->ip_blocks[i].version->funcs->late_init(&adev->ip_blocks[i]);
 			if (r) {
-				DRM_ERROR("late_init of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+				dev_err(adev->dev,
+					"late_init of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
 				return r;
 			}
 		}
@@ -3261,7 +3537,7 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev)
 
 	r = amdgpu_ras_late_init(adev);
 	if (r) {
-		DRM_ERROR("amdgpu_ras_late_init failed %d", r);
+		dev_err(adev->dev, "amdgpu_ras_late_init failed %d", r);
 		return r;
 	}
 
@@ -3275,7 +3551,7 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev)
 
 	r = amdgpu_device_enable_mgpu_fan_boost();
 	if (r)
-		DRM_ERROR("enable mgpu fan boost failed (%d).\n", r);
+		dev_err(adev->dev, "enable mgpu fan boost failed (%d).\n", r);
 
 	/* For passthrough configuration on arcturus and aldebaran, enable special handling SBR */
 	if (amdgpu_passthrough(adev) &&
@@ -3308,7 +3584,9 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev)
 				r = amdgpu_xgmi_set_pstate(gpu_instance->adev,
 						AMDGPU_XGMI_PSTATE_MIN);
 				if (r) {
-					DRM_ERROR("pstate setting failed (%d).\n", r);
+					dev_err(adev->dev,
+						"pstate setting failed (%d).\n",
+						r);
 					break;
 				}
 			}
@@ -3322,17 +3600,19 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev)
 
 static void amdgpu_ip_block_hw_fini(struct amdgpu_ip_block *ip_block)
 {
+	struct amdgpu_device *adev = ip_block->adev;
 	int r;
 
 	if (!ip_block->version->funcs->hw_fini) {
-		DRM_ERROR("hw_fini of IP block <%s> not defined\n",
-			  ip_block->version->funcs->name);
+		dev_err(adev->dev, "hw_fini of IP block <%s> not defined\n",
+			ip_block->version->funcs->name);
 	} else {
 		r = ip_block->version->funcs->hw_fini(ip_block);
 		/* XXX handle errors */
 		if (r) {
-			DRM_DEBUG("hw_fini of IP block <%s> failed %d\n",
-				  ip_block->version->funcs->name, r);
+			dev_dbg(adev->dev,
+				"hw_fini of IP block <%s> failed %d\n",
+				ip_block->version->funcs->name, r);
 		}
 	}
 
@@ -3373,15 +3653,17 @@ static int amdgpu_device_ip_fini_early(struct amdgpu_device *adev)
 
 		r = adev->ip_blocks[i].version->funcs->early_fini(&adev->ip_blocks[i]);
 		if (r) {
-			DRM_DEBUG("early_fini of IP block <%s> failed %d\n",
-				  adev->ip_blocks[i].version->funcs->name, r);
+			dev_dbg(adev->dev,
+				"early_fini of IP block <%s> failed %d\n",
+				adev->ip_blocks[i].version->funcs->name, r);
 		}
 	}
 
 	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
 	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
 
-	amdgpu_amdkfd_suspend(adev, false);
+	amdgpu_amdkfd_suspend(adev, true);
+	amdgpu_userq_suspend(adev);
 
 	/* Workaround for ASICs need to disable SMC first */
 	amdgpu_device_smu_fini_early(adev);
@@ -3395,7 +3677,22 @@ static int amdgpu_device_ip_fini_early(struct amdgpu_device *adev)
 
 	if (amdgpu_sriov_vf(adev)) {
 		if (amdgpu_virt_release_full_gpu(adev, false))
-			DRM_ERROR("failed to release exclusive mode on fini\n");
+			dev_err(adev->dev,
+				"failed to release exclusive mode on fini\n");
+	}
+
+	/*
+	 * Driver reload on the APU can fail due to firmware validation because
+	 * the PSP is always running, as it is shared across the whole SoC.
+	 * This same issue does not occur on dGPU because it has a mechanism
+	 * that checks whether the PSP is running. A solution for those issues
+	 * in the APU is to trigger a GPU reset, but this should be done during
+	 * the unload phase to avoid adding boot latency and screen flicker.
+	 */
+	if ((adev->flags & AMD_IS_APU) && !adev->gmc.is_app_apu) {
+		r = amdgpu_asic_reset(adev);
+		if (r)
+			dev_err(adev->dev, "asic reset on %s failed\n", __func__);
 	}
 
 	return 0;
@@ -3416,6 +3713,8 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev)
 {
 	int i, r;
 
+	amdgpu_cper_fini(adev);
+
 	if (amdgpu_sriov_vf(adev) && adev->virt.ras_init_done)
 		amdgpu_virt_release_ras_err_handler_data(adev);
 
@@ -3435,13 +3734,16 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev)
 			amdgpu_device_mem_scratch_fini(adev);
 			amdgpu_ib_pool_fini(adev);
 			amdgpu_seq64_fini(adev);
+			amdgpu_doorbell_fini(adev);
 		}
 		if (adev->ip_blocks[i].version->funcs->sw_fini) {
 			r = adev->ip_blocks[i].version->funcs->sw_fini(&adev->ip_blocks[i]);
 			/* XXX handle errors */
 			if (r) {
-				DRM_DEBUG("sw_fini of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
+				dev_dbg(adev->dev,
+					"sw_fini of IP block <%s> failed %d\n",
+					adev->ip_blocks[i].version->funcs->name,
+					r);
 			}
 		}
 		adev->ip_blocks[i].status.sw = false;
@@ -3457,6 +3759,7 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev)
 	}
 
 	amdgpu_ras_fini(adev);
+	amdgpu_uid_fini(adev);
 
 	return 0;
 }
@@ -3474,7 +3777,7 @@ static void amdgpu_device_delayed_init_work_handler(struct work_struct *work)
 
 	r = amdgpu_ib_ring_tests(adev);
 	if (r)
-		DRM_ERROR("ib ring test failed (%d).\n", r);
+		dev_err(adev->dev, "ib ring test failed (%d).\n", r);
 }
 
 static void amdgpu_device_delay_enable_gfx_off(struct work_struct *work)
@@ -3502,7 +3805,7 @@ static void amdgpu_device_delay_enable_gfx_off(struct work_struct *work)
  */
 static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev)
 {
-	int i, r;
+	int i, r, rec;
 
 	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
 	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
@@ -3523,13 +3826,25 @@ static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev)
 		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_DCE)
 			continue;
 
-		/* XXX handle errors */
 		r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
 		if (r)
-			return r;
+			goto unwind;
 	}
 
 	return 0;
+unwind:
+	rec = amdgpu_device_ip_resume_phase3(adev);
+	if (rec)
+		dev_err(adev->dev,
+			"amdgpu_device_ip_resume_phase3 failed during unwind: %d\n",
+			rec);
+
+	amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_ALLOW);
+
+	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_GATE);
+	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_GATE);
+
+	return r;
 }
 
 /**
@@ -3545,7 +3860,7 @@ static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev)
  */
 static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
 {
-	int i, r;
+	int i, r, rec;
 
 	if (adev->in_s0ix)
 		amdgpu_dpm_gfx_state_change(adev, sGpuChangeState_D3Entry);
@@ -3568,6 +3883,13 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
 			    adev, adev->ip_blocks[i].version->type))
 			continue;
 
+		/* Since we skip suspend for S0i3, we need to cancel the delayed
+		 * idle work here as the suspend callback never gets called.
+		 */
+		if (adev->in_s0ix &&
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX &&
+		    amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 0, 0))
+			cancel_delayed_work_sync(&adev->gfx.idle_work);
 		/* skip suspend of gfx/mes and psp for S0ix
 		 * gfx is in gfxoff state, so on resume it will exit gfxoff just
 		 * like at runtime. PSP is also part of the always on hardware
@@ -3599,24 +3921,52 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
 		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP)
 			continue;
 
-		/* XXX handle errors */
 		r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]);
-		adev->ip_blocks[i].status.hw = false;
+		if (r)
+			goto unwind;
 
 		/* handle putting the SMC in the appropriate state */
 		if (!amdgpu_sriov_vf(adev)) {
 			if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) {
 				r = amdgpu_dpm_set_mp1_state(adev, adev->mp1_state);
 				if (r) {
-					DRM_ERROR("SMC failed to set mp1 state %d, %d\n",
-							adev->mp1_state, r);
-					return r;
+					dev_err(adev->dev,
+						"SMC failed to set mp1 state %d, %d\n",
+						adev->mp1_state, r);
+					goto unwind;
 				}
 			}
 		}
 	}
 
 	return 0;
+unwind:
+	/* suspend phase 2 = resume phase 1 + resume phase 2 */
+	rec = amdgpu_device_ip_resume_phase1(adev);
+	if (rec) {
+		dev_err(adev->dev,
+			"amdgpu_device_ip_resume_phase1 failed during unwind: %d\n",
+			rec);
+		return r;
+	}
+
+	rec = amdgpu_device_fw_loading(adev);
+	if (rec) {
+		dev_err(adev->dev,
+			"amdgpu_device_fw_loading failed during unwind: %d\n",
+			rec);
+		return r;
+	}
+
+	rec = amdgpu_device_ip_resume_phase2(adev);
+	if (rec) {
+		dev_err(adev->dev,
+			"amdgpu_device_ip_resume_phase2 failed during unwind: %d\n",
+			rec);
+		return r;
+	}
+
+	return r;
 }
 
 /**
@@ -3630,7 +3980,7 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
  * in each IP into a state suitable for suspend.
  * Returns 0 on success, negative error code on failure.
  */
-int amdgpu_device_ip_suspend(struct amdgpu_device *adev)
+static int amdgpu_device_ip_suspend(struct amdgpu_device *adev)
 {
 	int r;
 
@@ -3893,12 +4243,14 @@ static void amdgpu_device_detect_sriov_bios(struct amdgpu_device *adev)
 /**
  * amdgpu_device_asic_has_dc_support - determine if DC supports the asic
  *
+ * @pdev : pci device context
  * @asic_type: AMD asic type
  *
  * Check if there is DC (new modesetting infrastructre) support for an asic.
  * returns true if DC has support, false if not.
  */
-bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type)
+bool amdgpu_device_asic_has_dc_support(struct pci_dev *pdev,
+				       enum amd_asic_type asic_type)
 {
 	switch (asic_type) {
 #ifdef CONFIG_DRM_AMDGPU_SI
@@ -3912,25 +4264,13 @@ bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type)
 	case CHIP_PITCAIRN:
 	case CHIP_VERDE:
 	case CHIP_OLAND:
-		/*
-		 * We have systems in the wild with these ASICs that require
-		 * LVDS and VGA support which is not supported with DC.
-		 *
-		 * Fallback to the non-DC driver here by default so as not to
-		 * cause regressions.
-		 */
-#if defined(CONFIG_DRM_AMD_DC_SI)
-		return amdgpu_dc > 0;
-#else
-		return false;
-#endif
-	case CHIP_BONAIRE:
+		return amdgpu_dc != 0 && IS_ENABLED(CONFIG_DRM_AMD_DC_SI);
 	case CHIP_KAVERI:
 	case CHIP_KABINI:
 	case CHIP_MULLINS:
 		/*
 		 * We have systems in the wild with these ASICs that require
-		 * VGA support which is not supported with DC.
+		 * TRAVIS and NUTMEG support which is not supported with DC.
 		 *
 		 * Fallback to the non-DC driver here by default so as not to
 		 * cause regressions.
@@ -3941,7 +4281,9 @@ bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type)
 #else
 	default:
 		if (amdgpu_dc > 0)
-			DRM_INFO_ONCE("Display Core has been requested via kernel parameter but isn't supported by ASIC, ignoring\n");
+			dev_info_once(
+				&pdev->dev,
+				"Display Core has been requested via kernel parameter but isn't supported by ASIC, ignoring\n");
 		return false;
 #endif
 	}
@@ -3960,7 +4302,7 @@ bool amdgpu_device_has_dc_support(struct amdgpu_device *adev)
 	    (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK))
 		return false;
 
-	return amdgpu_device_asic_has_dc_support(adev->asic_type);
+	return amdgpu_device_asic_has_dc_support(adev->pdev, adev->asic_type);
 }
 
 static void amdgpu_device_xgmi_reset_func(struct work_struct *__work)
@@ -3982,13 +4324,13 @@ static void amdgpu_device_xgmi_reset_func(struct work_struct *__work)
 	if (amdgpu_asic_reset_method(adev) == AMD_RESET_METHOD_BACO) {
 
 		task_barrier_enter(&hive->tb);
-		adev->asic_reset_res = amdgpu_device_baco_enter(adev_to_drm(adev));
+		adev->asic_reset_res = amdgpu_device_baco_enter(adev);
 
 		if (adev->asic_reset_res)
 			goto fail;
 
 		task_barrier_exit(&hive->tb);
-		adev->asic_reset_res = amdgpu_device_baco_exit(adev_to_drm(adev));
+		adev->asic_reset_res = amdgpu_device_baco_exit(adev);
 
 		if (adev->asic_reset_res)
 			goto fail;
@@ -4002,7 +4344,8 @@ static void amdgpu_device_xgmi_reset_func(struct work_struct *__work)
 
 fail:
 	if (adev->asic_reset_res)
-		DRM_WARN("ASIC reset failed with error, %d for drm dev, %s",
+		dev_warn(adev->dev,
+			 "ASIC reset failed with error, %d for drm dev, %s",
 			 adev->asic_reset_res, adev_to_drm(adev)->unique);
 	amdgpu_put_xgmi_hive(hive);
 }
@@ -4015,66 +4358,53 @@ static int amdgpu_device_get_job_timeout_settings(struct amdgpu_device *adev)
 	long timeout;
 	int ret = 0;
 
-	/*
-	 * By default timeout for non compute jobs is 10000
-	 * and 60000 for compute jobs.
-	 * In SR-IOV or passthrough mode, timeout for compute
-	 * jobs are 60000 by default.
-	 */
-	adev->gfx_timeout = msecs_to_jiffies(10000);
-	adev->sdma_timeout = adev->video_timeout = adev->gfx_timeout;
-	if (amdgpu_sriov_vf(adev))
-		adev->compute_timeout = amdgpu_sriov_is_pp_one_vf(adev) ?
-					msecs_to_jiffies(60000) : msecs_to_jiffies(10000);
-	else
-		adev->compute_timeout =  msecs_to_jiffies(60000);
+	/* By default timeout for all queues is 2 sec */
+	adev->gfx_timeout = adev->compute_timeout = adev->sdma_timeout =
+		adev->video_timeout = msecs_to_jiffies(2000);
 
-	if (strnlen(input, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) {
-		while ((timeout_setting = strsep(&input, ",")) &&
-				strnlen(timeout_setting, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) {
-			ret = kstrtol(timeout_setting, 0, &timeout);
-			if (ret)
-				return ret;
+	if (!strnlen(input, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH))
+		return 0;
 
-			if (timeout == 0) {
-				index++;
-				continue;
-			} else if (timeout < 0) {
-				timeout = MAX_SCHEDULE_TIMEOUT;
-				dev_warn(adev->dev, "lockup timeout disabled");
-				add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK);
-			} else {
-				timeout = msecs_to_jiffies(timeout);
-			}
+	while ((timeout_setting = strsep(&input, ",")) &&
+	       strnlen(timeout_setting, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) {
+		ret = kstrtol(timeout_setting, 0, &timeout);
+		if (ret)
+			return ret;
 
-			switch (index++) {
-			case 0:
-				adev->gfx_timeout = timeout;
-				break;
-			case 1:
-				adev->compute_timeout = timeout;
-				break;
-			case 2:
-				adev->sdma_timeout = timeout;
-				break;
-			case 3:
-				adev->video_timeout = timeout;
-				break;
-			default:
-				break;
-			}
+		if (timeout == 0) {
+			index++;
+			continue;
+		} else if (timeout < 0) {
+			timeout = MAX_SCHEDULE_TIMEOUT;
+			dev_warn(adev->dev, "lockup timeout disabled");
+			add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK);
+		} else {
+			timeout = msecs_to_jiffies(timeout);
 		}
-		/*
-		 * There is only one value specified and
-		 * it should apply to all non-compute jobs.
-		 */
-		if (index == 1) {
-			adev->sdma_timeout = adev->video_timeout = adev->gfx_timeout;
-			if (amdgpu_sriov_vf(adev) || amdgpu_passthrough(adev))
-				adev->compute_timeout = adev->gfx_timeout;
+
+		switch (index++) {
+		case 0:
+			adev->gfx_timeout = timeout;
+			break;
+		case 1:
+			adev->compute_timeout = timeout;
+			break;
+		case 2:
+			adev->sdma_timeout = timeout;
+			break;
+		case 3:
+			adev->video_timeout = timeout;
+			break;
+		default:
+			break;
 		}
 	}
 
+	/* When only one value specified apply it to all queues. */
+	if (index == 1)
+		adev->gfx_timeout = adev->compute_timeout = adev->sdma_timeout =
+			adev->video_timeout = timeout;
+
 	return ret;
 }
 
@@ -4115,11 +4445,6 @@ static bool amdgpu_device_check_iommu_remap(struct amdgpu_device *adev)
 }
 #endif
 
-static const struct attribute *amdgpu_dev_attributes[] = {
-	&dev_attr_pcie_replay_count.attr,
-	NULL
-};
-
 static void amdgpu_device_set_mcbp(struct amdgpu_device *adev)
 {
 	if (amdgpu_mcbp == 1)
@@ -4131,7 +4456,56 @@ static void amdgpu_device_set_mcbp(struct amdgpu_device *adev)
 		adev->gfx.mcbp = true;
 
 	if (adev->gfx.mcbp)
-		DRM_INFO("MCBP is enabled\n");
+		dev_info(adev->dev, "MCBP is enabled\n");
+}
+
+static int amdgpu_device_sys_interface_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_atombios_sysfs_init(adev);
+	if (r)
+		drm_err(&adev->ddev,
+			"registering atombios sysfs failed (%d).\n", r);
+
+	r = amdgpu_pm_sysfs_init(adev);
+	if (r)
+		dev_err(adev->dev, "registering pm sysfs failed (%d).\n", r);
+
+	r = amdgpu_ucode_sysfs_init(adev);
+	if (r) {
+		adev->ucode_sysfs_en = false;
+		dev_err(adev->dev, "Creating firmware sysfs failed (%d).\n", r);
+	} else
+		adev->ucode_sysfs_en = true;
+
+	r = amdgpu_device_attr_sysfs_init(adev);
+	if (r)
+		dev_err(adev->dev, "Could not create amdgpu device attr\n");
+
+	r = devm_device_add_group(adev->dev, &amdgpu_board_attrs_group);
+	if (r)
+		dev_err(adev->dev,
+			"Could not create amdgpu board attributes\n");
+
+	amdgpu_fru_sysfs_init(adev);
+	amdgpu_reg_state_sysfs_init(adev);
+	amdgpu_xcp_sysfs_init(adev);
+
+	return r;
+}
+
+static void amdgpu_device_sys_interface_fini(struct amdgpu_device *adev)
+{
+	if (adev->pm.sysfs_initialized)
+		amdgpu_pm_sysfs_fini(adev);
+	if (adev->ucode_sysfs_en)
+		amdgpu_ucode_sysfs_fini(adev);
+	amdgpu_device_attr_sysfs_fini(adev);
+	amdgpu_fru_sysfs_fini(adev);
+
+	amdgpu_reg_state_sysfs_fini(adev);
+	amdgpu_xcp_sysfs_fini(adev);
 }
 
 /**
@@ -4147,7 +4521,6 @@ static void amdgpu_device_set_mcbp(struct amdgpu_device *adev)
 int amdgpu_device_init(struct amdgpu_device *adev,
 		       uint32_t flags)
 {
-	struct drm_device *ddev = adev_to_drm(adev);
 	struct pci_dev *pdev = adev->pdev;
 	int r, i;
 	bool px = false;
@@ -4199,9 +4572,11 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	adev->audio_endpt_rreg = &amdgpu_block_invalid_rreg;
 	adev->audio_endpt_wreg = &amdgpu_block_invalid_wreg;
 
-	DRM_INFO("initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X 0x%02X).\n",
-		 amdgpu_asic_name[adev->asic_type], pdev->vendor, pdev->device,
-		 pdev->subsystem_vendor, pdev->subsystem_device, pdev->revision);
+	dev_info(
+		adev->dev,
+		"initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X 0x%02X).\n",
+		amdgpu_asic_name[adev->asic_type], pdev->vendor, pdev->device,
+		pdev->subsystem_vendor, pdev->subsystem_device, pdev->revision);
 
 	/* mutex initialization are all done here so we
 	 * can recall function without having locking issues
@@ -4216,7 +4591,6 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	mutex_init(&adev->grbm_idx_mutex);
 	mutex_init(&adev->mn_lock);
 	mutex_init(&adev->virt.vf_errors.lock);
-	mutex_init(&adev->virt.rlcg_reg_lock);
 	hash_init(adev->mn_hash);
 	mutex_init(&adev->psp.mutex);
 	mutex_init(&adev->notifier_lock);
@@ -4225,7 +4599,14 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	mutex_init(&adev->gfx.reset_sem_mutex);
 	/* Initialize the mutex for cleaner shader isolation between GFX and compute processes */
 	mutex_init(&adev->enforce_isolation_mutex);
-	mutex_init(&adev->gfx.kfd_sch_mutex);
+	for (i = 0; i < MAX_XCP; ++i) {
+		adev->isolation[i].spearhead = dma_fence_get_stub();
+		amdgpu_sync_create(&adev->isolation[i].active);
+		amdgpu_sync_create(&adev->isolation[i].prev);
+	}
+	mutex_init(&adev->gfx.userq_sch_mutex);
+	mutex_init(&adev->gfx.workload_profile_mutex);
+	mutex_init(&adev->vcn.workload_profile_mutex);
 
 	amdgpu_device_init_apu_flags(adev);
 
@@ -4242,14 +4623,19 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	spin_lock_init(&adev->se_cac_idx_lock);
 	spin_lock_init(&adev->audio_endpt_idx_lock);
 	spin_lock_init(&adev->mm_stats.lock);
+	spin_lock_init(&adev->virt.rlcg_reg_lock);
 	spin_lock_init(&adev->wb.lock);
 
+	xa_init_flags(&adev->userq_xa, XA_FLAGS_LOCK_IRQ);
+
 	INIT_LIST_HEAD(&adev->reset_list);
 
 	INIT_LIST_HEAD(&adev->ras_list);
 
 	INIT_LIST_HEAD(&adev->pm.od_kobj_list);
 
+	xa_init(&adev->userq_doorbell_xa);
+
 	INIT_DELAYED_WORK(&adev->delayed_init_work,
 			  amdgpu_device_delayed_init_work_handler);
 	INIT_DELAYED_WORK(&adev->gfx.gfx_off_delay_work,
@@ -4271,6 +4657,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	}
 
 	INIT_WORK(&adev->xgmi_reset_work, amdgpu_device_xgmi_reset_func);
+	INIT_WORK(&adev->userq_reset_work, amdgpu_userq_reset_work);
 
 	adev->gfx.gfx_off_req_count = 1;
 	adev->gfx.gfx_off_residency = 0;
@@ -4286,10 +4673,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	 * for throttling interrupt) = 60 seconds.
 	 */
 	ratelimit_state_init(&adev->throttling_logging_rs, (60 - 1) * HZ, 1);
-	ratelimit_state_init(&adev->virt.ras_telemetry_rs, 5 * HZ, 1);
 
 	ratelimit_set_flags(&adev->throttling_logging_rs, RATELIMIT_MSG_ON_RELEASE);
-	ratelimit_set_flags(&adev->virt.ras_telemetry_rs, RATELIMIT_MSG_ON_RELEASE);
 
 	/* Registers mapping */
 	/* TODO: block userspace mapping of io register */
@@ -4308,8 +4693,10 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	if (!adev->rmmio)
 		return -ENOMEM;
 
-	DRM_INFO("register mmio base: 0x%08X\n", (uint32_t)adev->rmmio_base);
-	DRM_INFO("register mmio size: %u\n", (unsigned int)adev->rmmio_size);
+	dev_info(adev->dev, "register mmio base: 0x%08X\n",
+		 (uint32_t)adev->rmmio_base);
+	dev_info(adev->dev, "register mmio size: %u\n",
+		 (unsigned int)adev->rmmio_size);
 
 	/*
 	 * Reset domain needs to be present early, before XGMI hive discovered
@@ -4321,7 +4708,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 		return -ENOMEM;
 
 	/* detect hw virtualization here */
-	amdgpu_detect_virtualization(adev);
+	amdgpu_virt_init(adev);
 
 	amdgpu_device_get_pcie_info(adev);
 
@@ -4344,10 +4731,17 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	if (r)
 		return r;
 
-	/* Get rid of things like offb */
-	r = aperture_remove_conflicting_pci_devices(adev->pdev, amdgpu_kms_driver.name);
-	if (r)
-		return r;
+	/*
+	 * No need to remove conflicting FBs for non-display class devices.
+	 * This prevents the sysfb from being freed accidently.
+	 */
+	if ((pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA ||
+	    (pdev->class >> 8) == PCI_CLASS_DISPLAY_OTHER) {
+		/* Get rid of things like offb */
+		r = aperture_remove_conflicting_pci_devices(adev->pdev, amdgpu_kms_driver.name);
+		if (r)
+			return r;
+	}
 
 	/* Enable TMZ based on IP_VERSION */
 	amdgpu_gmc_tmz_set(adev);
@@ -4439,7 +4833,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 			r = -EINVAL;
 			goto failed;
 		}
-		DRM_INFO("GPU posting now...\n");
+		dev_info(adev->dev, "GPU posting now...\n");
 		r = amdgpu_device_asic_init(adev);
 		if (r) {
 			dev_err(adev->dev, "gpu post error!\n");
@@ -4465,8 +4859,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 				goto failed;
 			}
 			/* init i2c buses */
-			if (!amdgpu_device_has_dc_support(adev))
-				amdgpu_atombios_i2c_init(adev);
+			amdgpu_i2c_init(adev);
 		}
 	}
 
@@ -4538,39 +4931,14 @@ fence_driver_init:
 		flush_delayed_work(&adev->delayed_init_work);
 	}
 
+	if (adev->init_lvl->level == AMDGPU_INIT_LEVEL_MINIMAL_XGMI)
+		amdgpu_xgmi_reset_on_init(adev);
 	/*
 	 * Place those sysfs registering after `late_init`. As some of those
 	 * operations performed in `late_init` might affect the sysfs
 	 * interfaces creating.
 	 */
-	r = amdgpu_atombios_sysfs_init(adev);
-	if (r)
-		drm_err(&adev->ddev,
-			"registering atombios sysfs failed (%d).\n", r);
-
-	r = amdgpu_pm_sysfs_init(adev);
-	if (r)
-		DRM_ERROR("registering pm sysfs failed (%d).\n", r);
-
-	r = amdgpu_ucode_sysfs_init(adev);
-	if (r) {
-		adev->ucode_sysfs_en = false;
-		DRM_ERROR("Creating firmware sysfs failed (%d).\n", r);
-	} else
-		adev->ucode_sysfs_en = true;
-
-	r = sysfs_create_files(&adev->dev->kobj, amdgpu_dev_attributes);
-	if (r)
-		dev_err(adev->dev, "Could not create amdgpu device attr\n");
-
-	r = devm_device_add_group(adev->dev, &amdgpu_board_attrs_group);
-	if (r)
-		dev_err(adev->dev,
-			"Could not create amdgpu board attributes\n");
-
-	amdgpu_fru_sysfs_init(adev);
-	amdgpu_reg_state_sysfs_init(adev);
-	amdgpu_xcp_cfg_sysfs_init(adev);
+	r = amdgpu_device_sys_interface_init(adev);
 
 	if (IS_ENABLED(CONFIG_PERF_EVENTS))
 		r = amdgpu_pmu_init(adev);
@@ -4588,7 +4956,7 @@ fence_driver_init:
 	if ((adev->pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA)
 		vga_client_register(adev->pdev, amdgpu_device_vga_set_decode);
 
-	px = amdgpu_device_supports_px(ddev);
+	px = amdgpu_device_supports_px(adev);
 
 	if (px || (!dev_is_removable(&adev->pdev->dev) &&
 				apple_gmux_detect(NULL, NULL)))
@@ -4598,9 +4966,6 @@ fence_driver_init:
 	if (px)
 		vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain);
 
-	if (adev->init_lvl->level == AMDGPU_INIT_LEVEL_MINIMAL_XGMI)
-		amdgpu_xgmi_reset_on_init(adev);
-
 	amdgpu_device_check_iommu_direct_map(adev);
 
 	adev->pm_nb.notifier_call = amdgpu_device_pm_notifier;
@@ -4692,15 +5057,7 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev)
 	}
 	amdgpu_fence_driver_hw_fini(adev);
 
-	if (adev->pm.sysfs_initialized)
-		amdgpu_pm_sysfs_fini(adev);
-	if (adev->ucode_sysfs_en)
-		amdgpu_ucode_sysfs_fini(adev);
-	sysfs_remove_files(&adev->dev->kobj, amdgpu_dev_attributes);
-	amdgpu_fru_sysfs_fini(adev);
-
-	amdgpu_reg_state_sysfs_fini(adev);
-	amdgpu_xcp_cfg_sysfs_fini(adev);
+	amdgpu_device_sys_interface_fini(adev);
 
 	/* disable ras feature must before hw fini */
 	amdgpu_ras_pre_fini(adev);
@@ -4723,7 +5080,7 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev)
 
 void amdgpu_device_fini_sw(struct amdgpu_device *adev)
 {
-	int idx;
+	int i, idx;
 	bool px;
 
 	amdgpu_device_ip_fini(adev);
@@ -4731,23 +5088,30 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev)
 	amdgpu_ucode_release(&adev->firmware.gpu_info_fw);
 	adev->accel_working = false;
 	dma_fence_put(rcu_dereference_protected(adev->gang_submit, true));
+	for (i = 0; i < MAX_XCP; ++i) {
+		dma_fence_put(adev->isolation[i].spearhead);
+		amdgpu_sync_free(&adev->isolation[i].active);
+		amdgpu_sync_free(&adev->isolation[i].prev);
+	}
 
 	amdgpu_reset_fini(adev);
 
 	/* free i2c buses */
-	if (!amdgpu_device_has_dc_support(adev))
-		amdgpu_i2c_fini(adev);
-
-	if (amdgpu_emu_mode != 1)
-		amdgpu_atombios_fini(adev);
+	amdgpu_i2c_fini(adev);
 
-	kfree(adev->bios);
-	adev->bios = NULL;
+	if (adev->bios) {
+		if (amdgpu_emu_mode != 1)
+			amdgpu_atombios_fini(adev);
+		amdgpu_bios_release(adev);
+	}
 
 	kfree(adev->fru_info);
 	adev->fru_info = NULL;
 
-	px = amdgpu_device_supports_px(adev_to_drm(adev));
+	kfree(adev->xcp_mgr);
+	adev->xcp_mgr = NULL;
+
+	px = amdgpu_device_supports_px(adev);
 
 	if (px || (!dev_is_removable(&adev->pdev->dev) &&
 				apple_gmux_detect(NULL, NULL)))
@@ -4763,20 +5127,20 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev)
 
 		iounmap(adev->rmmio);
 		adev->rmmio = NULL;
-		amdgpu_doorbell_fini(adev);
 		drm_dev_exit(idx);
 	}
 
 	if (IS_ENABLED(CONFIG_PERF_EVENTS))
 		amdgpu_pmu_fini(adev);
-	if (adev->mman.discovery_bin)
+	if (adev->discovery.bin)
 		amdgpu_discovery_fini(adev);
 
 	amdgpu_reset_put_reset_domain(adev->reset_domain);
 	adev->reset_domain = NULL;
 
 	kfree(adev->pci_state);
-
+	kfree(adev->pcie_reset_ctx.swds_pcistate);
+	kfree(adev->pcie_reset_ctx.swus_pcistate);
 }
 
 /**
@@ -4796,9 +5160,21 @@ static int amdgpu_device_evict_resources(struct amdgpu_device *adev)
 	if (!adev->in_s4 && (adev->flags & AMD_IS_APU))
 		return 0;
 
+	/* No need to evict when going to S5 through S4 callbacks */
+	if (system_state == SYSTEM_POWER_OFF)
+		return 0;
+
 	ret = amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM);
-	if (ret)
-		DRM_WARN("evicting device resources failed\n");
+	if (ret) {
+		dev_warn(adev->dev, "evicting device resources failed\n");
+		return ret;
+	}
+
+	if (adev->in_s4) {
+		ret = ttm_device_prepare_hibernation(&adev->mman.bdev);
+		if (ret)
+			dev_err(adev->dev, "prepare hibernation failed, %d\n", ret);
+	}
 	return ret;
 }
 
@@ -4812,28 +5188,20 @@ static int amdgpu_device_evict_resources(struct amdgpu_device *adev)
  * @data: data
  *
  * This function is called when the system is about to suspend or hibernate.
- * It is used to evict resources from the device before the system goes to
- * sleep while there is still access to swap.
+ * It is used to set the appropriate flags so that eviction can be optimized
+ * in the pm prepare callback.
  */
 static int amdgpu_device_pm_notifier(struct notifier_block *nb, unsigned long mode,
 				     void *data)
 {
 	struct amdgpu_device *adev = container_of(nb, struct amdgpu_device, pm_nb);
-	int r;
 
 	switch (mode) {
 	case PM_HIBERNATION_PREPARE:
 		adev->in_s4 = true;
-		fallthrough;
-	case PM_SUSPEND_PREPARE:
-		r = amdgpu_device_evict_resources(adev);
-		/*
-		 * This is considered non-fatal at this time because
-		 * amdgpu_device_prepare() will also fatally evict resources.
-		 * See https://gitlab.freedesktop.org/drm/amd/-/issues/3781
-		 */
-		if (r)
-			drm_warn(adev_to_drm(adev), "Failed to evict resources, freeze active processes if problems occur: %d\n", r);
+		break;
+	case PM_POST_HIBERNATION:
+		adev->in_s4 = false;
 		break;
 	}
 
@@ -4854,15 +5222,13 @@ int amdgpu_device_prepare(struct drm_device *dev)
 	struct amdgpu_device *adev = drm_to_adev(dev);
 	int i, r;
 
-	amdgpu_choose_low_power_state(adev);
-
 	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
 		return 0;
 
 	/* Evict the majority of BOs before starting suspend sequence */
 	r = amdgpu_device_evict_resources(adev);
 	if (r)
-		goto unprepare;
+		return r;
 
 	flush_delayed_work(&adev->gfx.gfx_off_delay_work);
 
@@ -4873,15 +5239,32 @@ int amdgpu_device_prepare(struct drm_device *dev)
 			continue;
 		r = adev->ip_blocks[i].version->funcs->prepare_suspend(&adev->ip_blocks[i]);
 		if (r)
-			goto unprepare;
+			return r;
 	}
 
 	return 0;
+}
 
-unprepare:
-	adev->in_s0ix = adev->in_s3 = adev->in_s4 = false;
+/**
+ * amdgpu_device_complete - complete power state transition
+ *
+ * @dev: drm dev pointer
+ *
+ * Undo the changes from amdgpu_device_prepare. This will be
+ * called on all resume transitions, including those that failed.
+ */
+void amdgpu_device_complete(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int i;
 
-	return r;
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (!adev->ip_blocks[i].version->funcs->complete)
+			continue;
+		adev->ip_blocks[i].version->funcs->complete(&adev->ip_blocks[i]);
+	}
 }
 
 /**
@@ -4897,7 +5280,7 @@ unprepare:
 int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients)
 {
 	struct amdgpu_device *adev = drm_to_adev(dev);
-	int r = 0;
+	int r, rec;
 
 	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
 		return 0;
@@ -4905,44 +5288,125 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients)
 	adev->in_suspend = true;
 
 	if (amdgpu_sriov_vf(adev)) {
+		if (!adev->in_runpm)
+			amdgpu_amdkfd_suspend_process(adev);
 		amdgpu_virt_fini_data_exchange(adev);
 		r = amdgpu_virt_request_full_gpu(adev, false);
 		if (r)
 			return r;
 	}
 
-	if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DEV_D3))
-		DRM_WARN("smart shift update failed\n");
+	r = amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DEV_D3);
+	if (r)
+		goto unwind_sriov;
 
 	if (notify_clients)
-		drm_client_dev_suspend(adev_to_drm(adev), false);
+		drm_client_dev_suspend(adev_to_drm(adev));
 
 	cancel_delayed_work_sync(&adev->delayed_init_work);
 
 	amdgpu_ras_suspend(adev);
 
-	amdgpu_device_ip_suspend_phase1(adev);
+	r = amdgpu_device_ip_suspend_phase1(adev);
+	if (r)
+		goto unwind_smartshift;
 
-	if (!adev->in_s0ix)
-		amdgpu_amdkfd_suspend(adev, adev->in_runpm);
+	amdgpu_amdkfd_suspend(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);
+	r = amdgpu_userq_suspend(adev);
+	if (r)
+		goto unwind_ip_phase1;
 
 	r = amdgpu_device_evict_resources(adev);
 	if (r)
-		return r;
+		goto unwind_userq;
 
 	amdgpu_ttm_set_buffer_funcs_status(adev, false);
 
 	amdgpu_fence_driver_hw_fini(adev);
 
-	amdgpu_device_ip_suspend_phase2(adev);
+	r = amdgpu_device_ip_suspend_phase2(adev);
+	if (r)
+		goto unwind_evict;
 
 	if (amdgpu_sriov_vf(adev))
 		amdgpu_virt_release_full_gpu(adev, false);
 
-	r = amdgpu_dpm_notify_rlc_state(adev, false);
+	return 0;
+
+unwind_evict:
+	if (adev->mman.buffer_funcs_ring->sched.ready)
+		amdgpu_ttm_set_buffer_funcs_status(adev, true);
+	amdgpu_fence_driver_hw_init(adev);
+
+unwind_userq:
+	rec = amdgpu_userq_resume(adev);
+	if (rec) {
+		dev_warn(adev->dev, "failed to re-initialize user queues: %d\n", rec);
+		return r;
+	}
+	rec = amdgpu_amdkfd_resume(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);
+	if (rec) {
+		dev_warn(adev->dev, "failed to re-initialize kfd: %d\n", rec);
+		return r;
+	}
+
+unwind_ip_phase1:
+	/* suspend phase 1 = resume phase 3 */
+	rec = amdgpu_device_ip_resume_phase3(adev);
+	if (rec) {
+		dev_warn(adev->dev, "failed to re-initialize IPs phase1: %d\n", rec);
+		return r;
+	}
+
+unwind_smartshift:
+	rec = amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DEV_D0);
+	if (rec) {
+		dev_warn(adev->dev, "failed to re-update smart shift: %d\n", rec);
+		return r;
+	}
+
+	if (notify_clients)
+		drm_client_dev_resume(adev_to_drm(adev));
+
+	amdgpu_ras_resume(adev);
+
+unwind_sriov:
+	if (amdgpu_sriov_vf(adev)) {
+		rec = amdgpu_virt_request_full_gpu(adev, true);
+		if (rec) {
+			dev_warn(adev->dev, "failed to reinitialize sriov: %d\n", rec);
+			return r;
+		}
+	}
+
+	adev->in_suspend = adev->in_s0ix = adev->in_s3 = false;
+
+	return r;
+}
+
+static inline int amdgpu_virt_resume(struct amdgpu_device *adev)
+{
+	int r;
+	unsigned int prev_physical_node_id = adev->gmc.xgmi.physical_node_id;
+
+	/* During VM resume, QEMU programming of VF MSIX table (register GFXMSIX_VECT0_ADDR_LO)
+	 * may not work. The access could be blocked by nBIF protection as VF isn't in
+	 * exclusive access mode. Exclusive access is enabled now, disable/enable MSIX
+	 * so that QEMU reprograms MSIX table.
+	 */
+	amdgpu_restore_msix(adev);
+
+	r = adev->gfxhub.funcs->get_xgmi_info(adev);
 	if (r)
 		return r;
 
+	dev_info(adev->dev, "xgmi node, old id %d, new id %d\n",
+		prev_physical_node_id, adev->gmc.xgmi.physical_node_id);
+
+	adev->vm_manager.vram_base_offset = adev->gfxhub.funcs->get_mc_fb_offset(adev);
+	adev->vm_manager.vram_base_offset +=
+		adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
+
 	return 0;
 }
 
@@ -4967,6 +5431,12 @@ int amdgpu_device_resume(struct drm_device *dev, bool notify_clients)
 			return r;
 	}
 
+	if (amdgpu_virt_xgmi_migrate_enabled(adev)) {
+		r = amdgpu_virt_resume(adev);
+		if (r)
+			goto exit;
+	}
+
 	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
 		return 0;
 
@@ -4987,11 +5457,13 @@ int amdgpu_device_resume(struct drm_device *dev, bool notify_clients)
 		goto exit;
 	}
 
-	if (!adev->in_s0ix) {
-		r = amdgpu_amdkfd_resume(adev, adev->in_runpm);
-		if (r)
-			goto exit;
-	}
+	r = amdgpu_amdkfd_resume(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);
+	if (r)
+		goto exit;
+
+	r = amdgpu_userq_resume(adev);
+	if (r)
+		goto exit;
 
 	r = amdgpu_device_ip_late_init(adev);
 	if (r)
@@ -5003,6 +5475,9 @@ exit:
 	if (amdgpu_sriov_vf(adev)) {
 		amdgpu_virt_init_data_exchange(adev);
 		amdgpu_virt_release_full_gpu(adev, true);
+
+		if (!r && !adev->in_runpm)
+			r = amdgpu_amdkfd_resume_process(adev);
 	}
 
 	if (r)
@@ -5012,7 +5487,7 @@ exit:
 	flush_delayed_work(&adev->delayed_init_work);
 
 	if (notify_clients)
-		drm_client_dev_resume(adev_to_drm(adev), false);
+		drm_client_dev_resume(adev_to_drm(adev));
 
 	amdgpu_ras_resume(adev);
 
@@ -5037,13 +5512,12 @@ exit:
 		dev->dev->power.disable_depth--;
 #endif
 	}
-	adev->in_suspend = false;
 
-	if (adev->enable_mes)
-		amdgpu_mes_self_test(adev);
+	amdgpu_vram_mgr_clear_reset_blocks(adev);
+	adev->in_suspend = false;
 
-	if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DEV_D0))
-		DRM_WARN("smart shift update failed\n");
+	if (amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DEV_D0))
+		dev_warn(adev->dev, "smart shift update failed\n");
 
 	return 0;
 }
@@ -5278,6 +5752,7 @@ static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
 	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) ||
 	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
 	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0) ||
 	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3))
 		amdgpu_ras_resume(adev);
 
@@ -5368,7 +5843,8 @@ int amdgpu_device_mode1_reset(struct amdgpu_device *adev)
 	u32 i;
 	int ret = 0;
 
-	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
+	if (adev->bios)
+		amdgpu_atombios_scratch_regs_engine_hung(adev, true);
 
 	dev_info(adev->dev, "GPU mode1 reset\n");
 
@@ -5410,7 +5886,8 @@ int amdgpu_device_mode1_reset(struct amdgpu_device *adev)
 		goto mode1_reset_failed;
 	}
 
-	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+	if (adev->bios)
+		amdgpu_atombios_scratch_regs_engine_hung(adev, false);
 
 	return 0;
 
@@ -5419,6 +5896,29 @@ mode1_reset_failed:
 	return ret;
 }
 
+int amdgpu_device_link_reset(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	dev_info(adev->dev, "GPU link reset\n");
+
+	if (!amdgpu_reset_in_dpc(adev))
+		ret = amdgpu_dpm_link_reset(adev);
+
+	if (ret)
+		goto link_reset_failed;
+
+	ret = amdgpu_psp_wait_for_bootloader(adev);
+	if (ret)
+		goto link_reset_failed;
+
+	return 0;
+
+link_reset_failed:
+	dev_err(adev->dev, "GPU link reset failed\n");
+	return ret;
+}
+
 int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
 				 struct amdgpu_reset_context *reset_context)
 {
@@ -5443,11 +5943,6 @@ int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
 		if (!amdgpu_ring_sched_ready(ring))
 			continue;
 
-		/* Clear job fence from fence drv to avoid force_completion
-		 * leave NULL and vm flush fence in fence drv
-		 */
-		amdgpu_fence_driver_clear_job_fences(ring);
-
 		/* after all hw jobs are reset, hw fence is meaningless, so force_completion */
 		amdgpu_fence_driver_force_completion(ring);
 	}
@@ -5531,6 +6026,7 @@ int amdgpu_device_reinit_after_reset(struct amdgpu_reset_context *reset_context)
 		amdgpu_set_init_level(tmp_adev, init_level);
 		if (full_reset) {
 			/* post card */
+			amdgpu_reset_set_dpc_status(tmp_adev, false);
 			amdgpu_ras_clear_err_state(tmp_adev);
 			r = amdgpu_device_asic_init(tmp_adev);
 			if (r) {
@@ -5548,7 +6044,9 @@ int amdgpu_device_reinit_after_reset(struct amdgpu_reset_context *reset_context)
 					amdgpu_coredump(tmp_adev, false, vram_lost, reset_context->job);
 
 				if (vram_lost) {
-					DRM_INFO("VRAM is lost due to GPU reset!\n");
+					dev_info(
+						tmp_adev->dev,
+						"VRAM is lost due to GPU reset!\n");
 					amdgpu_inc_vram_lost(tmp_adev);
 				}
 
@@ -5589,7 +6087,11 @@ int amdgpu_device_reinit_after_reset(struct amdgpu_reset_context *reset_context)
 				if (r)
 					goto out;
 
-				drm_client_dev_resume(adev_to_drm(tmp_adev), false);
+				r = amdgpu_userq_post_reset(tmp_adev, vram_lost);
+				if (r)
+					goto out;
+
+				drm_client_dev_resume(adev_to_drm(tmp_adev));
 
 				/*
 				 * The GPU enters bad state once faulty pages
@@ -5723,6 +6225,7 @@ static void amdgpu_device_set_mp1_state(struct amdgpu_device *adev)
 
 	switch (amdgpu_asic_reset_method(adev)) {
 	case AMD_RESET_METHOD_MODE1:
+	case AMD_RESET_METHOD_LINK:
 		adev->mp1_state = PP_MP1_STATE_SHUTDOWN;
 		break;
 	case AMD_RESET_METHOD_MODE2:
@@ -5810,6 +6313,7 @@ static inline void amdgpu_device_stop_pending_resets(struct amdgpu_device *adev)
 	if (!amdgpu_sriov_vf(adev))
 		cancel_work(&adev->reset_work);
 #endif
+	cancel_work(&adev->userq_reset_work);
 
 	if (adev->kfd.dev)
 		cancel_work(&adev->kfd.reset_work);
@@ -5826,117 +6330,76 @@ static int amdgpu_device_health_check(struct list_head *device_list_handle)
 {
 	struct amdgpu_device *tmp_adev;
 	int ret = 0;
-	u32 status;
 
 	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
-		pci_read_config_dword(tmp_adev->pdev, PCI_COMMAND, &status);
-		if (PCI_POSSIBLE_ERROR(status)) {
-			dev_err(tmp_adev->dev, "device lost from bus!");
-			ret = -ENODEV;
-		}
+		ret |= amdgpu_device_bus_status_check(tmp_adev);
 	}
 
 	return ret;
 }
 
-/**
- * amdgpu_device_gpu_recover - reset the asic and recover scheduler
- *
- * @adev: amdgpu_device pointer
- * @job: which job trigger hang
- * @reset_context: amdgpu reset context pointer
- *
- * Attempt to reset the GPU if it has hung (all asics).
- * Attempt to do soft-reset or full-reset and reinitialize Asic
- * Returns 0 for success or an error on failure.
- */
-
-int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
-			      struct amdgpu_job *job,
-			      struct amdgpu_reset_context *reset_context)
+static void amdgpu_device_recovery_prepare(struct amdgpu_device *adev,
+					  struct list_head *device_list,
+					  struct amdgpu_hive_info *hive)
 {
-	struct list_head device_list, *device_list_handle =  NULL;
-	bool job_signaled = false;
-	struct amdgpu_hive_info *hive = NULL;
 	struct amdgpu_device *tmp_adev = NULL;
-	int i, r = 0;
-	bool need_emergency_restart = false;
-	bool audio_suspended = false;
-	int retry_limit = AMDGPU_MAX_RETRY_LIMIT;
-
-	/*
-	 * If it reaches here because of hang/timeout and a RAS error is
-	 * detected at the same time, let RAS recovery take care of it.
-	 */
-	if (amdgpu_ras_is_err_state(adev, AMDGPU_RAS_BLOCK__ANY) &&
-	    !amdgpu_sriov_vf(adev) &&
-	    reset_context->src != AMDGPU_RESET_SRC_RAS) {
-		dev_dbg(adev->dev,
-			"Gpu recovery from source: %d yielding to RAS error recovery handling",
-			reset_context->src);
-		return 0;
-	}
-	/*
-	 * Special case: RAS triggered and full reset isn't supported
-	 */
-	need_emergency_restart = amdgpu_ras_need_emergency_restart(adev);
-
-	/*
-	 * Flush RAM to disk so that after reboot
-	 * the user can read log and see why the system rebooted.
-	 */
-	if (need_emergency_restart && amdgpu_ras_get_context(adev) &&
-		amdgpu_ras_get_context(adev)->reboot) {
-		DRM_WARN("Emergency reboot.");
-
-		ksys_sync_helper();
-		emergency_restart();
-	}
-
-	dev_info(adev->dev, "GPU %s begin!\n",
-		need_emergency_restart ? "jobs stop":"reset");
-
-	if (!amdgpu_sriov_vf(adev))
-		hive = amdgpu_get_xgmi_hive(adev);
-	if (hive)
-		mutex_lock(&hive->hive_lock);
 
-	reset_context->job = job;
-	reset_context->hive = hive;
 	/*
 	 * Build list of devices to reset.
 	 * In case we are in XGMI hive mode, resort the device list
 	 * to put adev in the 1st position.
 	 */
-	INIT_LIST_HEAD(&device_list);
 	if (!amdgpu_sriov_vf(adev) && (adev->gmc.xgmi.num_physical_nodes > 1) && hive) {
 		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
-			list_add_tail(&tmp_adev->reset_list, &device_list);
+			list_add_tail(&tmp_adev->reset_list, device_list);
 			if (adev->shutdown)
 				tmp_adev->shutdown = true;
+			if (amdgpu_reset_in_dpc(adev))
+				tmp_adev->pcie_reset_ctx.in_link_reset = true;
 		}
-		if (!list_is_first(&adev->reset_list, &device_list))
-			list_rotate_to_front(&adev->reset_list, &device_list);
-		device_list_handle = &device_list;
+		if (!list_is_first(&adev->reset_list, device_list))
+			list_rotate_to_front(&adev->reset_list, device_list);
 	} else {
-		list_add_tail(&adev->reset_list, &device_list);
-		device_list_handle = &device_list;
+		list_add_tail(&adev->reset_list, device_list);
 	}
+}
 
-	if (!amdgpu_sriov_vf(adev)) {
-		r = amdgpu_device_health_check(device_list_handle);
-		if (r)
-			goto end_reset;
-	}
+static void amdgpu_device_recovery_get_reset_lock(struct amdgpu_device *adev,
+						  struct list_head *device_list)
+{
+	struct amdgpu_device *tmp_adev = NULL;
 
-	/* We need to lock reset domain only once both for XGMI and single device */
-	tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device,
-				    reset_list);
+	if (list_empty(device_list))
+		return;
+	tmp_adev =
+		list_first_entry(device_list, struct amdgpu_device, reset_list);
 	amdgpu_device_lock_reset_domain(tmp_adev->reset_domain);
+}
 
-	/* block all schedulers and reset given job's ring */
-	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+static void amdgpu_device_recovery_put_reset_lock(struct amdgpu_device *adev,
+						  struct list_head *device_list)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+
+	if (list_empty(device_list))
+		return;
+	tmp_adev =
+		list_first_entry(device_list, struct amdgpu_device, reset_list);
+	amdgpu_device_unlock_reset_domain(tmp_adev->reset_domain);
+}
 
+static void amdgpu_device_halt_activities(struct amdgpu_device *adev,
+					  struct amdgpu_job *job,
+					  struct amdgpu_reset_context *reset_context,
+					  struct list_head *device_list,
+					  struct amdgpu_hive_info *hive,
+					  bool need_emergency_restart)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+	int i;
+
+	/* block all schedulers and reset given job's ring */
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
 		amdgpu_device_set_mp1_state(tmp_adev);
 
 		/*
@@ -5950,7 +6413,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 		 * some audio codec errors.
 		 */
 		if (!amdgpu_device_suspend_display_audio(tmp_adev))
-			audio_suspended = true;
+			tmp_adev->pcie_reset_ctx.audio_suspended = true;
 
 		amdgpu_ras_set_error_query_ready(tmp_adev, false);
 
@@ -5964,13 +6427,15 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 		 */
 		amdgpu_unregister_gpu_instance(tmp_adev);
 
-		drm_client_dev_suspend(adev_to_drm(tmp_adev), false);
+		drm_client_dev_suspend(adev_to_drm(tmp_adev));
 
 		/* disable ras on ALL IPs */
-		if (!need_emergency_restart &&
-		      amdgpu_device_ip_need_full_reset(tmp_adev))
+		if (!need_emergency_restart && !amdgpu_reset_in_dpc(adev) &&
+		    amdgpu_device_ip_need_full_reset(tmp_adev))
 			amdgpu_ras_suspend(tmp_adev);
 
+		amdgpu_userq_pre_reset(tmp_adev);
+
 		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 			struct amdgpu_ring *ring = tmp_adev->rings[i];
 
@@ -5984,24 +6449,18 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 		}
 		atomic_inc(&tmp_adev->gpu_reset_counter);
 	}
+}
 
-	if (need_emergency_restart)
-		goto skip_sched_resume;
-
-	/*
-	 * Must check guilty signal here since after this point all old
-	 * HW fences are force signaled.
-	 *
-	 * job->base holds a reference to parent fence
-	 */
-	if (job && dma_fence_is_signaled(&job->hw_fence)) {
-		job_signaled = true;
-		dev_info(adev->dev, "Guilty job already signaled, skipping HW reset");
-		goto skip_hw_reset;
-	}
+static int amdgpu_device_asic_reset(struct amdgpu_device *adev,
+			      struct list_head *device_list,
+			      struct amdgpu_reset_context *reset_context)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+	int retry_limit = AMDGPU_MAX_RETRY_LIMIT;
+	int r = 0;
 
 retry:	/* Rest of adevs pre asic reset from XGMI hive. */
-	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
 		r = amdgpu_device_pre_asic_reset(tmp_adev, reset_context);
 		/*TODO Should we stop ?*/
 		if (r) {
@@ -6014,6 +6473,11 @@ retry:	/* Rest of adevs pre asic reset from XGMI hive. */
 	/* Actual ASIC resets if needed.*/
 	/* Host driver will handle XGMI hive reset for SRIOV */
 	if (amdgpu_sriov_vf(adev)) {
+
+		/* Bail out of reset early */
+		if (amdgpu_ras_is_rma(adev))
+			return -ENODEV;
+
 		if (amdgpu_ras_get_fed_status(adev) || amdgpu_virt_rcvd_ras_interrupt(adev)) {
 			dev_dbg(adev->dev, "Detected RAS error, wait for FLR completion\n");
 			amdgpu_ras_set_fed(adev, true);
@@ -6028,12 +6492,12 @@ retry:	/* Rest of adevs pre asic reset from XGMI hive. */
 		if (r)
 			adev->asic_reset_res = r;
 	} else {
-		r = amdgpu_do_asic_reset(device_list_handle, reset_context);
+		r = amdgpu_do_asic_reset(device_list, reset_context);
 		if (r && r == -EAGAIN)
 			goto retry;
 	}
 
-	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
 		/*
 		 * Drop any pending non scheduler resets queued before reset is done.
 		 * Any reset scheduled after this point would be valid. Scheduler resets
@@ -6043,10 +6507,18 @@ retry:	/* Rest of adevs pre asic reset from XGMI hive. */
 		amdgpu_device_stop_pending_resets(tmp_adev);
 	}
 
-skip_hw_reset:
+	return r;
+}
+
+static int amdgpu_device_sched_resume(struct list_head *device_list,
+			      struct amdgpu_reset_context *reset_context,
+			      bool   job_signaled)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+	int i, r = 0;
 
 	/* Post ASIC reset for all devs .*/
-	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
 
 		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 			struct amdgpu_ring *ring = tmp_adev->rings[i];
@@ -6060,30 +6532,45 @@ skip_hw_reset:
 		if (!drm_drv_uses_atomic_modeset(adev_to_drm(tmp_adev)) && !job_signaled)
 			drm_helper_resume_force_mode(adev_to_drm(tmp_adev));
 
-		if (tmp_adev->asic_reset_res)
-			r = tmp_adev->asic_reset_res;
-
-		tmp_adev->asic_reset_res = 0;
-
-		if (r) {
+		if (tmp_adev->asic_reset_res) {
 			/* bad news, how to tell it to userspace ?
 			 * for ras error, we should report GPU bad status instead of
 			 * reset failure
 			 */
 			if (reset_context->src != AMDGPU_RESET_SRC_RAS ||
 			    !amdgpu_ras_eeprom_check_err_threshold(tmp_adev))
-				dev_info(tmp_adev->dev, "GPU reset(%d) failed\n",
-					atomic_read(&tmp_adev->gpu_reset_counter));
-			amdgpu_vf_error_put(tmp_adev, AMDGIM_ERROR_VF_GPU_RESET_FAIL, 0, r);
+				dev_info(
+					tmp_adev->dev,
+					"GPU reset(%d) failed with error %d \n",
+					atomic_read(
+						&tmp_adev->gpu_reset_counter),
+					tmp_adev->asic_reset_res);
+			amdgpu_vf_error_put(tmp_adev,
+					    AMDGIM_ERROR_VF_GPU_RESET_FAIL, 0,
+					    tmp_adev->asic_reset_res);
+			if (!r)
+				r = tmp_adev->asic_reset_res;
+			tmp_adev->asic_reset_res = 0;
 		} else {
-			dev_info(tmp_adev->dev, "GPU reset(%d) succeeded!\n", atomic_read(&tmp_adev->gpu_reset_counter));
-			if (amdgpu_acpi_smart_shift_update(adev_to_drm(tmp_adev), AMDGPU_SS_DEV_D0))
-				DRM_WARN("smart shift update failed\n");
+			dev_info(tmp_adev->dev, "GPU reset(%d) succeeded!\n",
+				 atomic_read(&tmp_adev->gpu_reset_counter));
+			if (amdgpu_acpi_smart_shift_update(tmp_adev,
+							   AMDGPU_SS_DEV_D0))
+				dev_warn(tmp_adev->dev,
+					 "smart shift update failed\n");
 		}
 	}
 
-skip_sched_resume:
-	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+	return r;
+}
+
+static void amdgpu_device_gpu_resume(struct amdgpu_device *adev,
+			      struct list_head *device_list,
+			      bool   need_emergency_restart)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
 		/* unlock kfd: SRIOV would do it separately */
 		if (!need_emergency_restart && !amdgpu_sriov_vf(tmp_adev))
 			amdgpu_amdkfd_post_reset(tmp_adev);
@@ -6094,18 +6581,124 @@ skip_sched_resume:
 		if (!adev->kfd.init_complete)
 			amdgpu_amdkfd_device_init(adev);
 
-		if (audio_suspended)
+		if (tmp_adev->pcie_reset_ctx.audio_suspended)
 			amdgpu_device_resume_display_audio(tmp_adev);
 
 		amdgpu_device_unset_mp1_state(tmp_adev);
 
 		amdgpu_ras_set_error_query_ready(tmp_adev, true);
+
 	}
+}
 
-	tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device,
-					    reset_list);
-	amdgpu_device_unlock_reset_domain(tmp_adev->reset_domain);
 
+/**
+ * amdgpu_device_gpu_recover - reset the asic and recover scheduler
+ *
+ * @adev: amdgpu_device pointer
+ * @job: which job trigger hang
+ * @reset_context: amdgpu reset context pointer
+ *
+ * Attempt to reset the GPU if it has hung (all asics).
+ * Attempt to do soft-reset or full-reset and reinitialize Asic
+ * Returns 0 for success or an error on failure.
+ */
+
+int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
+			      struct amdgpu_job *job,
+			      struct amdgpu_reset_context *reset_context)
+{
+	struct list_head device_list;
+	bool job_signaled = false;
+	struct amdgpu_hive_info *hive = NULL;
+	int r = 0;
+	bool need_emergency_restart = false;
+
+	/*
+	 * If it reaches here because of hang/timeout and a RAS error is
+	 * detected at the same time, let RAS recovery take care of it.
+	 */
+	if (amdgpu_ras_is_err_state(adev, AMDGPU_RAS_BLOCK__ANY) &&
+	    !amdgpu_sriov_vf(adev) &&
+	    reset_context->src != AMDGPU_RESET_SRC_RAS) {
+		dev_dbg(adev->dev,
+			"Gpu recovery from source: %d yielding to RAS error recovery handling",
+			reset_context->src);
+		return 0;
+	}
+
+	/*
+	 * Special case: RAS triggered and full reset isn't supported
+	 */
+	need_emergency_restart = amdgpu_ras_need_emergency_restart(adev);
+
+	/*
+	 * Flush RAM to disk so that after reboot
+	 * the user can read log and see why the system rebooted.
+	 */
+	if (need_emergency_restart && amdgpu_ras_get_context(adev) &&
+		amdgpu_ras_get_context(adev)->reboot) {
+		dev_warn(adev->dev, "Emergency reboot.");
+
+		ksys_sync_helper();
+		emergency_restart();
+	}
+
+	dev_info(adev->dev, "GPU %s begin!. Source:  %d\n",
+		 need_emergency_restart ? "jobs stop" : "reset",
+		 reset_context->src);
+
+	if (!amdgpu_sriov_vf(adev))
+		hive = amdgpu_get_xgmi_hive(adev);
+	if (hive)
+		mutex_lock(&hive->hive_lock);
+
+	reset_context->job = job;
+	reset_context->hive = hive;
+	INIT_LIST_HEAD(&device_list);
+
+	amdgpu_device_recovery_prepare(adev, &device_list, hive);
+
+	if (!amdgpu_sriov_vf(adev)) {
+		r = amdgpu_device_health_check(&device_list);
+		if (r)
+			goto end_reset;
+	}
+
+	/* Cannot be called after locking reset domain */
+	amdgpu_ras_pre_reset(adev, &device_list);
+
+	/* We need to lock reset domain only once both for XGMI and single device */
+	amdgpu_device_recovery_get_reset_lock(adev, &device_list);
+
+	amdgpu_device_halt_activities(adev, job, reset_context, &device_list,
+				      hive, need_emergency_restart);
+	if (need_emergency_restart)
+		goto skip_sched_resume;
+	/*
+	 * Must check guilty signal here since after this point all old
+	 * HW fences are force signaled.
+	 *
+	 * job->base holds a reference to parent fence
+	 */
+	if (job && dma_fence_is_signaled(&job->hw_fence->base)) {
+		job_signaled = true;
+		dev_info(adev->dev, "Guilty job already signaled, skipping HW reset");
+		goto skip_hw_reset;
+	}
+
+	r = amdgpu_device_asic_reset(adev, &device_list, reset_context);
+	if (r)
+		goto reset_unlock;
+skip_hw_reset:
+	r = amdgpu_device_sched_resume(&device_list, reset_context, job_signaled);
+	if (r)
+		goto reset_unlock;
+skip_sched_resume:
+	amdgpu_device_gpu_resume(adev, &device_list, need_emergency_restart);
+reset_unlock:
+	amdgpu_device_recovery_put_reset_lock(adev, &device_list);
+	amdgpu_ras_post_reset(adev, &device_list);
 end_reset:
 	if (hive) {
 		mutex_unlock(&hive->hive_lock);
@@ -6116,6 +6709,19 @@ end_reset:
 		dev_info(adev->dev, "GPU reset end with ret = %d\n", r);
 
 	atomic_set(&adev->reset_domain->reset_res, r);
+
+	if (!r) {
+		struct amdgpu_task_info *ti = NULL;
+
+		if (job)
+			ti = amdgpu_vm_get_task_info_pasid(adev, job->pasid);
+
+		drm_dev_wedged_event(adev_to_drm(adev), DRM_WEDGE_RECOVERY_NONE,
+				     ti ? &ti->task : NULL);
+
+		amdgpu_vm_put_task_info(ti);
+	}
+
 	return r;
 }
 
@@ -6433,12 +7039,11 @@ bool amdgpu_device_is_peer_accessible(struct amdgpu_device *adev,
 #endif
 }
 
-int amdgpu_device_baco_enter(struct drm_device *dev)
+int amdgpu_device_baco_enter(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
 
-	if (!amdgpu_device_supports_baco(dev))
+	if (!amdgpu_device_supports_baco(adev))
 		return -ENOTSUPP;
 
 	if (ras && adev->ras_enabled &&
@@ -6448,13 +7053,12 @@ int amdgpu_device_baco_enter(struct drm_device *dev)
 	return amdgpu_dpm_baco_enter(adev);
 }
 
-int amdgpu_device_baco_exit(struct drm_device *dev)
+int amdgpu_device_baco_exit(struct amdgpu_device *adev)
 {
-	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
 	int ret = 0;
 
-	if (!amdgpu_device_supports_baco(dev))
+	if (!amdgpu_device_supports_baco(adev))
 		return -ENOTSUPP;
 
 	ret = amdgpu_dpm_baco_exit(adev);
@@ -6485,45 +7089,52 @@ pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_sta
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
 	struct amdgpu_device *adev = drm_to_adev(dev);
-	int i;
+	struct amdgpu_hive_info *hive __free(xgmi_put_hive) =
+		amdgpu_get_xgmi_hive(adev);
+	struct amdgpu_reset_context reset_context;
+	struct list_head device_list;
 
-	DRM_INFO("PCI error: detected callback, state(%d)!!\n", state);
-
-	if (adev->gmc.xgmi.num_physical_nodes > 1) {
-		DRM_WARN("No support for XGMI hive yet...");
-		return PCI_ERS_RESULT_DISCONNECT;
-	}
+	dev_info(adev->dev, "PCI error: detected callback!!\n");
 
 	adev->pci_channel_state = state;
 
 	switch (state) {
 	case pci_channel_io_normal:
+		dev_info(adev->dev, "pci_channel_io_normal: state(%d)!!\n", state);
 		return PCI_ERS_RESULT_CAN_RECOVER;
-	/* Fatal error, prepare for slot reset */
 	case pci_channel_io_frozen:
-		/*
-		 * Locking adev->reset_domain->sem will prevent any external access
-		 * to GPU during PCI error recovery
-		 */
-		amdgpu_device_lock_reset_domain(adev->reset_domain);
-		amdgpu_device_set_mp1_state(adev);
-
-		/*
-		 * Block any work scheduling as we do for regular GPU reset
-		 * for the duration of the recovery
-		 */
-		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
-			struct amdgpu_ring *ring = adev->rings[i];
-
-			if (!amdgpu_ring_sched_ready(ring))
-				continue;
+		/* Fatal error, prepare for slot reset */
+		dev_info(adev->dev, "pci_channel_io_frozen: state(%d)!!\n", state);
+		if (hive) {
+			/* Hive devices should be able to support FW based
+			 * link reset on other devices, if not return.
+			 */
+			if (!amdgpu_dpm_is_link_reset_supported(adev)) {
+				dev_warn(adev->dev,
+					 "No support for XGMI hive yet...\n");
+				return PCI_ERS_RESULT_DISCONNECT;
+			}
+			/* Set dpc status only if device is part of hive
+			 * Non-hive devices should be able to recover after
+			 * link reset.
+			 */
+			amdgpu_reset_set_dpc_status(adev, true);
 
-			drm_sched_stop(&ring->sched, NULL);
+			mutex_lock(&hive->hive_lock);
 		}
-		atomic_inc(&adev->gpu_reset_counter);
+		memset(&reset_context, 0, sizeof(reset_context));
+		INIT_LIST_HEAD(&device_list);
+
+		amdgpu_device_recovery_prepare(adev, &device_list, hive);
+		amdgpu_device_recovery_get_reset_lock(adev, &device_list);
+		amdgpu_device_halt_activities(adev, NULL, &reset_context, &device_list,
+					      hive, false);
+		if (hive)
+			mutex_unlock(&hive->hive_lock);
 		return PCI_ERS_RESULT_NEED_RESET;
 	case pci_channel_io_perm_failure:
 		/* Permanent error, prepare for device removal */
+		dev_info(adev->dev, "pci_channel_io_perm_failure: state(%d)!!\n", state);
 		return PCI_ERS_RESULT_DISCONNECT;
 	}
 
@@ -6536,8 +7147,10 @@ pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_sta
  */
 pci_ers_result_t amdgpu_pci_mmio_enabled(struct pci_dev *pdev)
 {
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
 
-	DRM_INFO("PCI error: mmio enabled callback!!\n");
+	dev_info(adev->dev, "PCI error: mmio enabled callback!!\n");
 
 	/* TODO - dump whatever for debugging purposes */
 
@@ -6561,27 +7174,38 @@ pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
 	struct amdgpu_device *adev = drm_to_adev(dev);
-	int r, i;
 	struct amdgpu_reset_context reset_context;
-	u32 memsize;
+	struct amdgpu_device *tmp_adev;
+	struct amdgpu_hive_info *hive;
 	struct list_head device_list;
+	struct pci_dev *link_dev;
+	int r = 0, i, timeout;
+	u32 memsize;
+	u16 status;
 
-	/* PCI error slot reset should be skipped During RAS recovery */
-	if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4)) &&
-	    amdgpu_ras_in_recovery(adev))
-		return PCI_ERS_RESULT_RECOVERED;
-
-	DRM_INFO("PCI error: slot reset callback!!\n");
+	dev_info(adev->dev, "PCI error: slot reset callback!!\n");
 
 	memset(&reset_context, 0, sizeof(reset_context));
 
-	INIT_LIST_HEAD(&device_list);
-	list_add_tail(&adev->reset_list, &device_list);
+	if (adev->pcie_reset_ctx.swus)
+		link_dev = adev->pcie_reset_ctx.swus;
+	else
+		link_dev = adev->pdev;
+	/* wait for asic to come out of reset, timeout = 10s */
+	timeout = 10000;
+	do {
+		usleep_range(10000, 10500);
+		r = pci_read_config_word(link_dev, PCI_VENDOR_ID, &status);
+		timeout -= 10;
+	} while (timeout > 0 && (status != PCI_VENDOR_ID_ATI) &&
+		 (status != PCI_VENDOR_ID_AMD));
 
-	/* wait for asic to come out of reset */
-	msleep(500);
+	if ((status != PCI_VENDOR_ID_ATI) && (status != PCI_VENDOR_ID_AMD)) {
+		r = -ETIME;
+		goto out;
+	}
 
+	amdgpu_device_load_switch_state(adev);
 	/* Restore PCI confspace */
 	amdgpu_device_load_pci_state(pdev);
 
@@ -6601,26 +7225,40 @@ pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
 	reset_context.method = AMD_RESET_METHOD_NONE;
 	reset_context.reset_req_dev = adev;
 	set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
-	set_bit(AMDGPU_SKIP_HW_RESET, &reset_context.flags);
-
-	adev->no_hw_access = true;
-	r = amdgpu_device_pre_asic_reset(adev, &reset_context);
-	adev->no_hw_access = false;
-	if (r)
-		goto out;
+	set_bit(AMDGPU_SKIP_COREDUMP, &reset_context.flags);
+	INIT_LIST_HEAD(&device_list);
 
-	r = amdgpu_do_asic_reset(&device_list, &reset_context);
+	hive = amdgpu_get_xgmi_hive(adev);
+	if (hive) {
+		mutex_lock(&hive->hive_lock);
+		reset_context.hive = hive;
+		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+			tmp_adev->pcie_reset_ctx.in_link_reset = true;
+			list_add_tail(&tmp_adev->reset_list, &device_list);
+		}
+	} else {
+		set_bit(AMDGPU_SKIP_HW_RESET, &reset_context.flags);
+		list_add_tail(&adev->reset_list, &device_list);
+	}
 
+	r = amdgpu_device_asic_reset(adev, &device_list, &reset_context);
 out:
 	if (!r) {
 		if (amdgpu_device_cache_pci_state(adev->pdev))
 			pci_restore_state(adev->pdev);
-
-		DRM_INFO("PCIe error recovery succeeded\n");
+		dev_info(adev->dev, "PCIe error recovery succeeded\n");
 	} else {
-		DRM_ERROR("PCIe error recovery failed, err:%d", r);
-		amdgpu_device_unset_mp1_state(adev);
-		amdgpu_device_unlock_reset_domain(adev->reset_domain);
+		dev_err(adev->dev, "PCIe error recovery failed, err:%d\n", r);
+		if (hive) {
+			list_for_each_entry(tmp_adev, &device_list, reset_list)
+				amdgpu_device_unset_mp1_state(tmp_adev);
+		}
+		amdgpu_device_recovery_put_reset_lock(adev, &device_list);
+	}
+
+	if (hive) {
+		mutex_unlock(&hive->hive_lock);
+		amdgpu_put_xgmi_hive(hive);
 	}
 
 	return r ? PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_RECOVERED;
@@ -6637,26 +7275,95 @@ void amdgpu_pci_resume(struct pci_dev *pdev)
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
 	struct amdgpu_device *adev = drm_to_adev(dev);
-	int i;
-
+	struct list_head device_list;
+	struct amdgpu_hive_info *hive = NULL;
+	struct amdgpu_device *tmp_adev = NULL;
 
-	DRM_INFO("PCI error: resume callback!!\n");
+	dev_info(adev->dev, "PCI error: resume callback!!\n");
 
 	/* Only continue execution for the case of pci_channel_io_frozen */
 	if (adev->pci_channel_state != pci_channel_io_frozen)
 		return;
 
-	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
-		struct amdgpu_ring *ring = adev->rings[i];
+	INIT_LIST_HEAD(&device_list);
 
-		if (!amdgpu_ring_sched_ready(ring))
-			continue;
+	hive = amdgpu_get_xgmi_hive(adev);
+	if (hive) {
+		mutex_lock(&hive->hive_lock);
+		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
+			tmp_adev->pcie_reset_ctx.in_link_reset = false;
+			list_add_tail(&tmp_adev->reset_list, &device_list);
+		}
+	} else
+		list_add_tail(&adev->reset_list, &device_list);
+
+	amdgpu_device_sched_resume(&device_list, NULL, NULL);
+	amdgpu_device_gpu_resume(adev, &device_list, false);
+	amdgpu_device_recovery_put_reset_lock(adev, &device_list);
 
-		drm_sched_start(&ring->sched, 0);
+	if (hive) {
+		mutex_unlock(&hive->hive_lock);
+		amdgpu_put_xgmi_hive(hive);
 	}
+}
+
+static void amdgpu_device_cache_switch_state(struct amdgpu_device *adev)
+{
+	struct pci_dev *swus, *swds;
+	int r;
 
-	amdgpu_device_unset_mp1_state(adev);
-	amdgpu_device_unlock_reset_domain(adev->reset_domain);
+	swds = pci_upstream_bridge(adev->pdev);
+	if (!swds || swds->vendor != PCI_VENDOR_ID_ATI ||
+	    pci_pcie_type(swds) != PCI_EXP_TYPE_DOWNSTREAM)
+		return;
+	swus = pci_upstream_bridge(swds);
+	if (!swus ||
+	    (swus->vendor != PCI_VENDOR_ID_ATI &&
+	     swus->vendor != PCI_VENDOR_ID_AMD) ||
+	    pci_pcie_type(swus) != PCI_EXP_TYPE_UPSTREAM)
+		return;
+
+	/* If already saved, return */
+	if (adev->pcie_reset_ctx.swus)
+		return;
+	/* Upstream bridge is ATI, assume it's SWUS/DS architecture */
+	r = pci_save_state(swds);
+	if (r)
+		return;
+	adev->pcie_reset_ctx.swds_pcistate = pci_store_saved_state(swds);
+
+	r = pci_save_state(swus);
+	if (r)
+		return;
+	adev->pcie_reset_ctx.swus_pcistate = pci_store_saved_state(swus);
+
+	adev->pcie_reset_ctx.swus = swus;
+}
+
+static void amdgpu_device_load_switch_state(struct amdgpu_device *adev)
+{
+	struct pci_dev *pdev;
+	int r;
+
+	if (!adev->pcie_reset_ctx.swds_pcistate ||
+	    !adev->pcie_reset_ctx.swus_pcistate)
+		return;
+
+	pdev = adev->pcie_reset_ctx.swus;
+	r = pci_load_saved_state(pdev, adev->pcie_reset_ctx.swus_pcistate);
+	if (!r) {
+		pci_restore_state(pdev);
+	} else {
+		dev_warn(adev->dev, "Failed to load SWUS state, err:%d\n", r);
+		return;
+	}
+
+	pdev = pci_upstream_bridge(adev->pdev);
+	r = pci_load_saved_state(pdev, adev->pcie_reset_ctx.swds_pcistate);
+	if (!r)
+		pci_restore_state(pdev);
+	else
+		dev_warn(adev->dev, "Failed to load SWDS state, err:%d\n", r);
 }
 
 bool amdgpu_device_cache_pci_state(struct pci_dev *pdev)
@@ -6675,14 +7382,16 @@ bool amdgpu_device_cache_pci_state(struct pci_dev *pdev)
 		adev->pci_state = pci_store_saved_state(pdev);
 
 		if (!adev->pci_state) {
-			DRM_ERROR("Failed to store PCI saved state");
+			dev_err(adev->dev, "Failed to store PCI saved state");
 			return false;
 		}
 	} else {
-		DRM_WARN("Failed to save PCI state, err:%d\n", r);
+		dev_warn(adev->dev, "Failed to save PCI state, err:%d\n", r);
 		return false;
 	}
 
+	amdgpu_device_cache_switch_state(adev);
+
 	return true;
 }
 
@@ -6700,7 +7409,7 @@ bool amdgpu_device_load_pci_state(struct pci_dev *pdev)
 	if (!r) {
 		pci_restore_state(pdev);
 	} else {
-		DRM_WARN("Failed to load PCI state, err:%d\n", r);
+		dev_warn(adev->dev, "Failed to load PCI state, err:%d\n", r);
 		return false;
 	}
 
@@ -6717,10 +7426,17 @@ void amdgpu_device_flush_hdp(struct amdgpu_device *adev,
 	if (adev->gmc.xgmi.connected_to_cpu)
 		return;
 
-	if (ring && ring->funcs->emit_hdp_flush)
+	if (ring && ring->funcs->emit_hdp_flush) {
 		amdgpu_ring_emit_hdp_flush(ring);
-	else
-		amdgpu_asic_flush_hdp(adev, ring);
+		return;
+	}
+
+	if (!ring && amdgpu_sriov_runtime(adev)) {
+		if (!amdgpu_kiq_hdp_flush(adev))
+			return;
+	}
+
+	amdgpu_hdp_flush(adev, ring);
 }
 
 void amdgpu_device_invalidate_hdp(struct amdgpu_device *adev,
@@ -6733,7 +7449,7 @@ void amdgpu_device_invalidate_hdp(struct amdgpu_device *adev,
 	if (adev->gmc.xgmi.connected_to_cpu)
 		return;
 
-	amdgpu_asic_invalidate_hdp(adev, ring);
+	amdgpu_hdp_invalidate(adev, ring);
 }
 
 int amdgpu_in_reset(struct amdgpu_device *adev)
@@ -6844,22 +7560,117 @@ struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,
 {
 	struct dma_fence *old = NULL;
 
+	dma_fence_get(gang);
 	do {
 		dma_fence_put(old);
 		old = amdgpu_device_get_gang(adev);
 		if (old == gang)
 			break;
 
-		if (!dma_fence_is_signaled(old))
+		if (!dma_fence_is_signaled(old)) {
+			dma_fence_put(gang);
 			return old;
+		}
 
 	} while (cmpxchg((struct dma_fence __force **)&adev->gang_submit,
 			 old, gang) != old);
 
+	/*
+	 * Drop it once for the exchanged reference in adev and once for the
+	 * thread local reference acquired in amdgpu_device_get_gang().
+	 */
+	dma_fence_put(old);
 	dma_fence_put(old);
 	return NULL;
 }
 
+/**
+ * amdgpu_device_enforce_isolation - enforce HW isolation
+ * @adev: the amdgpu device pointer
+ * @ring: the HW ring the job is supposed to run on
+ * @job: the job which is about to be pushed to the HW ring
+ *
+ * Makes sure that only one client at a time can use the GFX block.
+ * Returns: The dependency to wait on before the job can be pushed to the HW.
+ * The function is called multiple times until NULL is returned.
+ */
+struct dma_fence *amdgpu_device_enforce_isolation(struct amdgpu_device *adev,
+						  struct amdgpu_ring *ring,
+						  struct amdgpu_job *job)
+{
+	struct amdgpu_isolation *isolation = &adev->isolation[ring->xcp_id];
+	struct drm_sched_fence *f = job->base.s_fence;
+	struct dma_fence *dep;
+	void *owner;
+	int r;
+
+	/*
+	 * For now enforce isolation only for the GFX block since we only need
+	 * the cleaner shader on those rings.
+	 */
+	if (ring->funcs->type != AMDGPU_RING_TYPE_GFX &&
+	    ring->funcs->type != AMDGPU_RING_TYPE_COMPUTE)
+		return NULL;
+
+	/*
+	 * All submissions where enforce isolation is false are handled as if
+	 * they come from a single client. Use ~0l as the owner to distinct it
+	 * from kernel submissions where the owner is NULL.
+	 */
+	owner = job->enforce_isolation ? f->owner : (void *)~0l;
+
+	mutex_lock(&adev->enforce_isolation_mutex);
+
+	/*
+	 * The "spearhead" submission is the first one which changes the
+	 * ownership to its client. We always need to wait for it to be
+	 * pushed to the HW before proceeding with anything.
+	 */
+	if (&f->scheduled != isolation->spearhead &&
+	    !dma_fence_is_signaled(isolation->spearhead)) {
+		dep = isolation->spearhead;
+		goto out_grab_ref;
+	}
+
+	if (isolation->owner != owner) {
+
+		/*
+		 * Wait for any gang to be assembled before switching to a
+		 * different owner or otherwise we could deadlock the
+		 * submissions.
+		 */
+		if (!job->gang_submit) {
+			dep = amdgpu_device_get_gang(adev);
+			if (!dma_fence_is_signaled(dep))
+				goto out_return_dep;
+			dma_fence_put(dep);
+		}
+
+		dma_fence_put(isolation->spearhead);
+		isolation->spearhead = dma_fence_get(&f->scheduled);
+		amdgpu_sync_move(&isolation->active, &isolation->prev);
+		trace_amdgpu_isolation(isolation->owner, owner);
+		isolation->owner = owner;
+	}
+
+	/*
+	 * Specifying the ring here helps to pipeline submissions even when
+	 * isolation is enabled. If that is not desired for testing NULL can be
+	 * used instead of the ring to enforce a CPU round trip while switching
+	 * between clients.
+	 */
+	dep = amdgpu_sync_peek_fence(&isolation->prev, ring);
+	r = amdgpu_sync_fence(&isolation->active, &f->finished, GFP_NOWAIT);
+	if (r)
+		dev_warn(adev->dev, "OOM tracking isolation\n");
+
+out_grab_ref:
+	dma_fence_get(dep);
+out_return_dep:
+	mutex_unlock(&adev->enforce_isolation_mutex);
+	return dep;
+}
+
 bool amdgpu_device_has_display_hardware(struct amdgpu_device *adev)
 {
 	switch (adev->asic_type) {
@@ -6919,9 +7730,11 @@ uint32_t amdgpu_device_wait_on_rreg(struct amdgpu_device *adev,
 		tmp_ = RREG32(reg_addr);
 		loop--;
 		if (!loop) {
-			DRM_WARN("Register(%d) [%s] failed to reach value 0x%08x != 0x%08xn",
-				  inst, reg_name, (uint32_t)expected_value,
-				  (uint32_t)(tmp_ & (mask)));
+			dev_warn(
+				adev->dev,
+				"Register(%d) [%s] failed to reach value 0x%08x != 0x%08xn",
+				inst, reg_name, (uint32_t)expected_value,
+				(uint32_t)(tmp_ & (mask)));
 			ret = -ETIMEDOUT;
 			break;
 		}
@@ -6972,3 +7785,53 @@ ssize_t amdgpu_show_reset_mask(char *buf, uint32_t supported_reset)
 	size += sysfs_emit_at(buf, size, "\n");
 	return size;
 }
+
+void amdgpu_device_set_uid(struct amdgpu_uid *uid_info,
+			   enum amdgpu_uid_type type, uint8_t inst,
+			   uint64_t uid)
+{
+	if (!uid_info)
+		return;
+
+	if (type >= AMDGPU_UID_TYPE_MAX) {
+		dev_err_once(uid_info->adev->dev, "Invalid UID type %d\n",
+			     type);
+		return;
+	}
+
+	if (inst >= AMDGPU_UID_INST_MAX) {
+		dev_err_once(uid_info->adev->dev, "Invalid UID instance %d\n",
+			     inst);
+		return;
+	}
+
+	if (uid_info->uid[type][inst] != 0) {
+		dev_warn_once(
+			uid_info->adev->dev,
+			"Overwriting existing UID %llu for type %d instance %d\n",
+			uid_info->uid[type][inst], type, inst);
+	}
+
+	uid_info->uid[type][inst] = uid;
+}
+
+u64 amdgpu_device_get_uid(struct amdgpu_uid *uid_info,
+			  enum amdgpu_uid_type type, uint8_t inst)
+{
+	if (!uid_info)
+		return 0;
+
+	if (type >= AMDGPU_UID_TYPE_MAX) {
+		dev_err_once(uid_info->adev->dev, "Invalid UID type %d\n",
+			     type);
+		return 0;
+	}
+
+	if (inst >= AMDGPU_UID_INST_MAX) {
+		dev_err_once(uid_info->adev->dev, "Invalid UID instance %d\n",
+			     inst);
+		return 0;
+	}
+
+	return uid_info->uid[type][inst];
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
index 949d74eff294..fa2a22dfa048 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
@@ -107,14 +107,22 @@
 #include "vcn_v5_0_1.h"
 #include "jpeg_v5_0_0.h"
 #include "jpeg_v5_0_1.h"
+#include "amdgpu_ras_mgr.h"
 
 #include "amdgpu_vpe.h"
 #if defined(CONFIG_DRM_AMD_ISP)
 #include "amdgpu_isp.h"
 #endif
 
-#define FIRMWARE_IP_DISCOVERY "amdgpu/ip_discovery.bin"
-MODULE_FIRMWARE(FIRMWARE_IP_DISCOVERY);
+MODULE_FIRMWARE("amdgpu/ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/vega10_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/vega12_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/vega20_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/raven_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/raven2_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/picasso_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/arcturus_ip_discovery.bin");
+MODULE_FIRMWARE("amdgpu/aldebaran_ip_discovery.bin");
 
 #define mmIP_DISCOVERY_VERSION  0x16A00
 #define mmRCC_CONFIG_MEMSIZE	0xde3
@@ -247,9 +255,9 @@ static int amdgpu_discovery_read_binary_from_sysmem(struct amdgpu_device *adev,
 	pos = tmr_offset + tmr_size - DISCOVERY_TMR_OFFSET;
 
 	/* This region is read-only and reserved from system use */
-	discv_regn = memremap(pos, adev->mman.discovery_tmr_size, MEMREMAP_WC);
+	discv_regn = memremap(pos, adev->discovery.size, MEMREMAP_WC);
 	if (discv_regn) {
-		memcpy(binary, discv_regn, adev->mman.discovery_tmr_size);
+		memcpy(binary, discv_regn, adev->discovery.size);
 		memunmap(discv_regn);
 		return 0;
 	}
@@ -263,12 +271,13 @@ static int amdgpu_discovery_read_binary_from_sysmem(struct amdgpu_device *adev,
 static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev,
 						 uint8_t *binary)
 {
+	bool sz_valid = true;
 	uint64_t vram_size;
-	u32 msg;
 	int i, ret = 0;
+	u32 msg;
 
 	if (!amdgpu_sriov_vf(adev)) {
-		/* It can take up to a second for IFWI init to complete on some dGPUs,
+		/* It can take up to two second for IFWI init to complete on some dGPUs,
 		 * but generally it should be in the 60-100ms range.  Normally this starts
 		 * as soon as the device gets power so by the time the OS loads this has long
 		 * completed.  However, when a card is hotplugged via e.g., USB4, we need to
@@ -276,7 +285,7 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev,
 		 * continue.
 		 */
 
-		for (i = 0; i < 1000; i++) {
+		for (i = 0; i < 2000; i++) {
 			msg = RREG32(mmMP0_SMN_C2PMSG_33);
 			if (msg & 0x80000000)
 				break;
@@ -284,38 +293,62 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev,
 		}
 	}
 
-	vram_size = (uint64_t)RREG32(mmRCC_CONFIG_MEMSIZE) << 20;
+	vram_size = RREG32(mmRCC_CONFIG_MEMSIZE);
+	if (!vram_size || vram_size == U32_MAX)
+		sz_valid = false;
+	else
+		vram_size <<= 20;
+
+	/*
+	 * If in VRAM, discovery TMR is marked for reservation. If it is in system mem,
+	 * then it is not required to be reserved.
+	 */
+	if (sz_valid) {
+		if (amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled) {
+			/* For SRIOV VFs with dynamic critical region enabled,
+			 * we will get the IPD binary via below call.
+			 * If dynamic critical is disabled, fall through to normal seq.
+			 */
+			if (amdgpu_virt_get_dynamic_data_info(adev,
+						AMD_SRIOV_MSG_IPD_TABLE_ID, binary,
+						&adev->discovery.size)) {
+				dev_err(adev->dev,
+						"failed to read discovery info from dynamic critical region.");
+				ret = -EINVAL;
+				goto exit;
+			}
+		} else {
+			uint64_t pos = vram_size - DISCOVERY_TMR_OFFSET;
 
-	if (vram_size) {
-		uint64_t pos = vram_size - DISCOVERY_TMR_OFFSET;
-		amdgpu_device_vram_access(adev, pos, (uint32_t *)binary,
-					  adev->mman.discovery_tmr_size, false);
+			amdgpu_device_vram_access(adev, pos, (uint32_t *)binary,
+					adev->discovery.size, false);
+			adev->discovery.reserve_tmr = true;
+		}
 	} else {
 		ret = amdgpu_discovery_read_binary_from_sysmem(adev, binary);
 	}
 
+	if (ret)
+		dev_err(adev->dev,
+			"failed to read discovery info from memory, vram size read: %llx",
+			vram_size);
+exit:
 	return ret;
 }
 
-static int amdgpu_discovery_read_binary_from_file(struct amdgpu_device *adev, uint8_t *binary)
+static int amdgpu_discovery_read_binary_from_file(struct amdgpu_device *adev,
+							uint8_t *binary,
+							const char *fw_name)
 {
 	const struct firmware *fw;
-	const char *fw_name;
 	int r;
 
-	switch (amdgpu_discovery) {
-	case 2:
-		fw_name = FIRMWARE_IP_DISCOVERY;
-		break;
-	default:
-		dev_warn(adev->dev, "amdgpu_discovery is not set properly\n");
-		return -EINVAL;
-	}
-
-	r = request_firmware(&fw, fw_name, adev->dev);
+	r = firmware_request_nowarn(&fw, fw_name, adev->dev);
 	if (r) {
-		dev_err(adev->dev, "can't load firmware \"%s\"\n",
-			fw_name);
+		if (amdgpu_discovery == 2)
+			dev_err(adev->dev, "can't load firmware \"%s\"\n", fw_name);
+		else
+			drm_info(&adev->ddev, "Optional firmware \"%s\" was not found\n", fw_name);
 		return r;
 	}
 
@@ -378,6 +411,7 @@ static void amdgpu_discovery_harvest_config_quirk(struct amdgpu_device *adev)
 static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev,
 					   struct binary_header *bhdr)
 {
+	uint8_t *discovery_bin = adev->discovery.bin;
 	struct table_info *info;
 	uint16_t checksum;
 	uint16_t offset;
@@ -387,14 +421,14 @@ static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev,
 	checksum = le16_to_cpu(info->checksum);
 
 	struct nps_info_header *nhdr =
-		(struct nps_info_header *)(adev->mman.discovery_bin + offset);
+		(struct nps_info_header *)(discovery_bin + offset);
 
 	if (le32_to_cpu(nhdr->table_id) != NPS_INFO_TABLE_ID) {
 		dev_dbg(adev->dev, "invalid ip discovery nps info table id\n");
 		return -EINVAL;
 	}
 
-	if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
+	if (!amdgpu_discovery_verify_checksum(discovery_bin + offset,
 					      le32_to_cpu(nhdr->size_bytes),
 					      checksum)) {
 		dev_dbg(adev->dev, "invalid nps info data table checksum\n");
@@ -404,55 +438,88 @@ static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev,
 	return 0;
 }
 
+static const char *amdgpu_discovery_get_fw_name(struct amdgpu_device *adev)
+{
+	if (amdgpu_discovery == 2) {
+		/* Assume there is valid discovery TMR in VRAM even if binary is sideloaded */
+		adev->discovery.reserve_tmr = true;
+		return "amdgpu/ip_discovery.bin";
+	}
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		return "amdgpu/vega10_ip_discovery.bin";
+	case CHIP_VEGA12:
+		return "amdgpu/vega12_ip_discovery.bin";
+	case CHIP_RAVEN:
+		if (adev->apu_flags & AMD_APU_IS_RAVEN2)
+			return "amdgpu/raven2_ip_discovery.bin";
+		else if (adev->apu_flags & AMD_APU_IS_PICASSO)
+			return "amdgpu/picasso_ip_discovery.bin";
+		else
+			return "amdgpu/raven_ip_discovery.bin";
+	case CHIP_VEGA20:
+		return "amdgpu/vega20_ip_discovery.bin";
+	case CHIP_ARCTURUS:
+		return "amdgpu/arcturus_ip_discovery.bin";
+	case CHIP_ALDEBARAN:
+		return "amdgpu/aldebaran_ip_discovery.bin";
+	default:
+		return NULL;
+	}
+}
+
 static int amdgpu_discovery_init(struct amdgpu_device *adev)
 {
 	struct table_info *info;
 	struct binary_header *bhdr;
+	uint8_t *discovery_bin;
+	const char *fw_name;
 	uint16_t offset;
 	uint16_t size;
 	uint16_t checksum;
 	int r;
 
-	adev->mman.discovery_tmr_size = DISCOVERY_TMR_SIZE;
-	adev->mman.discovery_bin = kzalloc(adev->mman.discovery_tmr_size, GFP_KERNEL);
-	if (!adev->mman.discovery_bin)
+	adev->discovery.bin = kzalloc(DISCOVERY_TMR_SIZE, GFP_KERNEL);
+	if (!adev->discovery.bin)
 		return -ENOMEM;
+	adev->discovery.size = DISCOVERY_TMR_SIZE;
+	adev->discovery.debugfs_blob.data = adev->discovery.bin;
+	adev->discovery.debugfs_blob.size = adev->discovery.size;
 
+	discovery_bin = adev->discovery.bin;
 	/* Read from file if it is the preferred option */
-	if (amdgpu_discovery == 2) {
-		dev_info(adev->dev, "use ip discovery information from file");
-		r = amdgpu_discovery_read_binary_from_file(adev, adev->mman.discovery_bin);
-
-		if (r) {
-			dev_err(adev->dev, "failed to read ip discovery binary from file\n");
-			r = -EINVAL;
+	fw_name = amdgpu_discovery_get_fw_name(adev);
+	if (fw_name != NULL) {
+		drm_dbg(&adev->ddev, "use ip discovery information from file");
+		r = amdgpu_discovery_read_binary_from_file(adev, discovery_bin,
+							   fw_name);
+		if (r)
 			goto out;
-		}
-
 	} else {
-		r = amdgpu_discovery_read_binary_from_mem(
-			adev, adev->mman.discovery_bin);
+		drm_dbg(&adev->ddev, "use ip discovery information from memory");
+		r = amdgpu_discovery_read_binary_from_mem(adev, discovery_bin);
 		if (r)
 			goto out;
 	}
 
 	/* check the ip discovery binary signature */
-	if (!amdgpu_discovery_verify_binary_signature(adev->mman.discovery_bin)) {
+	if (!amdgpu_discovery_verify_binary_signature(discovery_bin)) {
 		dev_err(adev->dev,
 			"get invalid ip discovery binary signature\n");
 		r = -EINVAL;
 		goto out;
 	}
 
-	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	bhdr = (struct binary_header *)discovery_bin;
 
 	offset = offsetof(struct binary_header, binary_checksum) +
 		sizeof(bhdr->binary_checksum);
 	size = le16_to_cpu(bhdr->binary_size) - offset;
 	checksum = le16_to_cpu(bhdr->binary_checksum);
 
-	if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
-					      size, checksum)) {
+	if (!amdgpu_discovery_verify_checksum(discovery_bin + offset, size,
+					      checksum)) {
 		dev_err(adev->dev, "invalid ip discovery binary checksum\n");
 		r = -EINVAL;
 		goto out;
@@ -464,15 +531,16 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
 
 	if (offset) {
 		struct ip_discovery_header *ihdr =
-			(struct ip_discovery_header *)(adev->mman.discovery_bin + offset);
+			(struct ip_discovery_header *)(discovery_bin + offset);
 		if (le32_to_cpu(ihdr->signature) != DISCOVERY_TABLE_SIGNATURE) {
 			dev_err(adev->dev, "invalid ip discovery data table signature\n");
 			r = -EINVAL;
 			goto out;
 		}
 
-		if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
-						      le16_to_cpu(ihdr->size), checksum)) {
+		if (!amdgpu_discovery_verify_checksum(discovery_bin + offset,
+						      le16_to_cpu(ihdr->size),
+						      checksum)) {
 			dev_err(adev->dev, "invalid ip discovery data table checksum\n");
 			r = -EINVAL;
 			goto out;
@@ -485,7 +553,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
 
 	if (offset) {
 		struct gpu_info_header *ghdr =
-			(struct gpu_info_header *)(adev->mman.discovery_bin + offset);
+			(struct gpu_info_header *)(discovery_bin + offset);
 
 		if (le32_to_cpu(ghdr->table_id) != GC_TABLE_ID) {
 			dev_err(adev->dev, "invalid ip discovery gc table id\n");
@@ -493,8 +561,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
 			goto out;
 		}
 
-		if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
-						      le32_to_cpu(ghdr->size), checksum)) {
+		if (!amdgpu_discovery_verify_checksum(discovery_bin + offset,
+						      le32_to_cpu(ghdr->size),
+						      checksum)) {
 			dev_err(adev->dev, "invalid gc data table checksum\n");
 			r = -EINVAL;
 			goto out;
@@ -507,7 +576,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
 
 	if (offset) {
 		struct harvest_info_header *hhdr =
-			(struct harvest_info_header *)(adev->mman.discovery_bin + offset);
+			(struct harvest_info_header *)(discovery_bin + offset);
 
 		if (le32_to_cpu(hhdr->signature) != HARVEST_TABLE_SIGNATURE) {
 			dev_err(adev->dev, "invalid ip discovery harvest table signature\n");
@@ -515,8 +584,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
 			goto out;
 		}
 
-		if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
-						      sizeof(struct harvest_table), checksum)) {
+		if (!amdgpu_discovery_verify_checksum(
+			    discovery_bin + offset,
+			    sizeof(struct harvest_table), checksum)) {
 			dev_err(adev->dev, "invalid harvest data table checksum\n");
 			r = -EINVAL;
 			goto out;
@@ -529,7 +599,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
 
 	if (offset) {
 		struct vcn_info_header *vhdr =
-			(struct vcn_info_header *)(adev->mman.discovery_bin + offset);
+			(struct vcn_info_header *)(discovery_bin + offset);
 
 		if (le32_to_cpu(vhdr->table_id) != VCN_INFO_TABLE_ID) {
 			dev_err(adev->dev, "invalid ip discovery vcn table id\n");
@@ -537,8 +607,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
 			goto out;
 		}
 
-		if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
-						      le32_to_cpu(vhdr->size_bytes), checksum)) {
+		if (!amdgpu_discovery_verify_checksum(
+			    discovery_bin + offset,
+			    le32_to_cpu(vhdr->size_bytes), checksum)) {
 			dev_err(adev->dev, "invalid vcn data table checksum\n");
 			r = -EINVAL;
 			goto out;
@@ -551,7 +622,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
 
 	if (0 && offset) {
 		struct mall_info_header *mhdr =
-			(struct mall_info_header *)(adev->mman.discovery_bin + offset);
+			(struct mall_info_header *)(discovery_bin + offset);
 
 		if (le32_to_cpu(mhdr->table_id) != MALL_INFO_TABLE_ID) {
 			dev_err(adev->dev, "invalid ip discovery mall table id\n");
@@ -559,8 +630,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
 			goto out;
 		}
 
-		if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset,
-						      le32_to_cpu(mhdr->size_bytes), checksum)) {
+		if (!amdgpu_discovery_verify_checksum(
+			    discovery_bin + offset,
+			    le32_to_cpu(mhdr->size_bytes), checksum)) {
 			dev_err(adev->dev, "invalid mall data table checksum\n");
 			r = -EINVAL;
 			goto out;
@@ -570,8 +642,8 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
 	return 0;
 
 out:
-	kfree(adev->mman.discovery_bin);
-	adev->mman.discovery_bin = NULL;
+	kfree(adev->discovery.bin);
+	adev->discovery.bin = NULL;
 	if ((amdgpu_discovery != 2) &&
 	    (RREG32(mmIP_DISCOVERY_VERSION) == 4))
 		amdgpu_ras_query_boot_status(adev, 4);
@@ -583,20 +655,23 @@ static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev);
 void amdgpu_discovery_fini(struct amdgpu_device *adev)
 {
 	amdgpu_discovery_sysfs_fini(adev);
-	kfree(adev->mman.discovery_bin);
-	adev->mman.discovery_bin = NULL;
+	kfree(adev->discovery.bin);
+	adev->discovery.bin = NULL;
 }
 
-static int amdgpu_discovery_validate_ip(const struct ip_v4 *ip)
+static int amdgpu_discovery_validate_ip(struct amdgpu_device *adev,
+					uint8_t instance, uint16_t hw_id)
 {
-	if (ip->instance_number >= HWIP_MAX_INSTANCE) {
-		DRM_ERROR("Unexpected instance_number (%d) from ip discovery blob\n",
-			  ip->instance_number);
+	if (instance >= HWIP_MAX_INSTANCE) {
+		dev_err(adev->dev,
+			"Unexpected instance_number (%d) from ip discovery blob\n",
+			instance);
 		return -EINVAL;
 	}
-	if (le16_to_cpu(ip->hw_id) >= HW_ID_MAX) {
-		DRM_ERROR("Unexpected hw_id (%d) from ip discovery blob\n",
-			  le16_to_cpu(ip->hw_id));
+	if (hw_id >= HW_ID_MAX) {
+		dev_err(adev->dev,
+			"Unexpected hw_id (%d) from ip discovery blob\n",
+			hw_id);
 		return -EINVAL;
 	}
 
@@ -606,36 +681,41 @@ static int amdgpu_discovery_validate_ip(const struct ip_v4 *ip)
 static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev,
 						uint32_t *vcn_harvest_count)
 {
+	uint8_t *discovery_bin = adev->discovery.bin;
 	struct binary_header *bhdr;
 	struct ip_discovery_header *ihdr;
 	struct die_header *dhdr;
-	struct ip_v4 *ip;
+	struct ip *ip;
 	uint16_t die_offset, ip_offset, num_dies, num_ips;
+	uint16_t hw_id;
+	uint8_t inst;
 	int i, j;
 
-	bhdr = (struct binary_header *)adev->mman.discovery_bin;
-	ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin +
-			le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
+	bhdr = (struct binary_header *)discovery_bin;
+	ihdr = (struct ip_discovery_header
+			*)(discovery_bin +
+			   le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
 	num_dies = le16_to_cpu(ihdr->num_dies);
 
 	/* scan harvest bit of all IP data structures */
 	for (i = 0; i < num_dies; i++) {
 		die_offset = le16_to_cpu(ihdr->die_info[i].die_offset);
-		dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset);
+		dhdr = (struct die_header *)(discovery_bin + die_offset);
 		num_ips = le16_to_cpu(dhdr->num_ips);
 		ip_offset = die_offset + sizeof(*dhdr);
 
 		for (j = 0; j < num_ips; j++) {
-			ip = (struct ip_v4 *)(adev->mman.discovery_bin + ip_offset);
-
-			if (amdgpu_discovery_validate_ip(ip))
+			ip = (struct ip *)(discovery_bin + ip_offset);
+			inst = ip->number_instance;
+			hw_id = le16_to_cpu(ip->hw_id);
+			if (amdgpu_discovery_validate_ip(adev, inst, hw_id))
 				goto next_ip;
 
-			if (le16_to_cpu(ip->variant) == 1) {
-				switch (le16_to_cpu(ip->hw_id)) {
+			if (ip->harvest == 1) {
+				switch (hw_id) {
 				case VCN_HWID:
 					(*vcn_harvest_count)++;
-					if (ip->instance_number == 0) {
+					if (inst == 0) {
 						adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN0;
 						adev->vcn.inst_mask &=
 							~AMDGPU_VCN_HARVEST_VCN0;
@@ -657,10 +737,8 @@ static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev,
 				}
 			}
 next_ip:
-			if (ihdr->base_addr_64_bit)
-				ip_offset += struct_size(ip, base_address_64, ip->num_base_address);
-			else
-				ip_offset += struct_size(ip, base_address, ip->num_base_address);
+			ip_offset += struct_size(ip, base_address,
+						 ip->num_base_address);
 		}
 	}
 }
@@ -669,13 +747,14 @@ static void amdgpu_discovery_read_from_harvest_table(struct amdgpu_device *adev,
 						     uint32_t *vcn_harvest_count,
 						     uint32_t *umc_harvest_count)
 {
+	uint8_t *discovery_bin = adev->discovery.bin;
 	struct binary_header *bhdr;
 	struct harvest_table *harvest_info;
 	u16 offset;
 	int i;
 	uint32_t umc_harvest_config = 0;
 
-	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	bhdr = (struct binary_header *)discovery_bin;
 	offset = le16_to_cpu(bhdr->table_list[HARVEST_INFO].offset);
 
 	if (!offset) {
@@ -683,7 +762,7 @@ static void amdgpu_discovery_read_from_harvest_table(struct amdgpu_device *adev,
 		return;
 	}
 
-	harvest_info = (struct harvest_table *)(adev->mman.discovery_bin + offset);
+	harvest_info = (struct harvest_table *)(discovery_bin + offset);
 
 	for (i = 0; i < 32; i++) {
 		if (le16_to_cpu(harvest_info->list[i].hw_id) == 0)
@@ -979,8 +1058,8 @@ static void ip_disc_release(struct kobject *kobj)
 						       kobj);
 	struct amdgpu_device *adev = ip_top->adev;
 
-	adev->ip_top = NULL;
 	kfree(ip_top);
+	adev->discovery.ip_top = NULL;
 }
 
 static uint8_t amdgpu_discovery_get_harvest_info(struct amdgpu_device *adev,
@@ -991,7 +1070,9 @@ static uint8_t amdgpu_discovery_get_harvest_info(struct amdgpu_device *adev,
 	/* Until a uniform way is figured, get mask based on hwid */
 	switch (hw_id) {
 	case VCN_HWID:
-		harvest = ((1 << inst) & adev->vcn.inst_mask) == 0;
+		/* VCN vs UVD+VCE */
+		if (!amdgpu_ip_version(adev, VCE_HWIP, 0))
+			harvest = ((1 << inst) & adev->vcn.inst_mask) == 0;
 		break;
 	case DMU_HWID:
 		if (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK)
@@ -1018,7 +1099,10 @@ static int amdgpu_discovery_sysfs_ips(struct amdgpu_device *adev,
 				      const size_t _ip_offset, const int num_ips,
 				      bool reg_base_64)
 {
+	uint8_t *discovery_bin = adev->discovery.bin;
 	int ii, jj, kk, res;
+	uint16_t hw_id;
+	uint8_t inst;
 
 	DRM_DEBUG("num_ips:%d", num_ips);
 
@@ -1033,9 +1117,11 @@ static int amdgpu_discovery_sysfs_ips(struct amdgpu_device *adev,
 			struct ip_v4 *ip;
 			struct ip_hw_instance *ip_hw_instance;
 
-			ip = (struct ip_v4 *)(adev->mman.discovery_bin + ip_offset);
-			if (amdgpu_discovery_validate_ip(ip) ||
-			    le16_to_cpu(ip->hw_id) != ii)
+			ip = (struct ip_v4 *)(discovery_bin + ip_offset);
+			inst = ip->instance_number;
+			hw_id = le16_to_cpu(ip->hw_id);
+			if (amdgpu_discovery_validate_ip(adev, inst, hw_id) ||
+			    hw_id != ii)
 				goto next_ip;
 
 			DRM_DEBUG("match:%d @ ip_offset:%zu", ii, ip_offset);
@@ -1118,17 +1204,20 @@ next_ip:
 
 static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev)
 {
+	struct ip_discovery_top *ip_top = adev->discovery.ip_top;
+	uint8_t *discovery_bin = adev->discovery.bin;
 	struct binary_header *bhdr;
 	struct ip_discovery_header *ihdr;
 	struct die_header *dhdr;
-	struct kset *die_kset = &adev->ip_top->die_kset;
+	struct kset *die_kset = &ip_top->die_kset;
 	u16 num_dies, die_offset, num_ips;
 	size_t ip_offset;
 	int ii, res;
 
-	bhdr = (struct binary_header *)adev->mman.discovery_bin;
-	ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin +
-					      le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
+	bhdr = (struct binary_header *)discovery_bin;
+	ihdr = (struct ip_discovery_header
+			*)(discovery_bin +
+			   le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
 	num_dies = le16_to_cpu(ihdr->num_dies);
 
 	DRM_DEBUG("number of dies: %d\n", num_dies);
@@ -1137,7 +1226,7 @@ static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev)
 		struct ip_die_entry *ip_die_entry;
 
 		die_offset = le16_to_cpu(ihdr->die_info[ii].die_offset);
-		dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset);
+		dhdr = (struct die_header *)(discovery_bin + die_offset);
 		num_ips = le16_to_cpu(dhdr->num_ips);
 		ip_offset = die_offset + sizeof(*dhdr);
 
@@ -1171,30 +1260,32 @@ static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev)
 
 static int amdgpu_discovery_sysfs_init(struct amdgpu_device *adev)
 {
+	uint8_t *discovery_bin = adev->discovery.bin;
+	struct ip_discovery_top *ip_top;
 	struct kset *die_kset;
 	int res, ii;
 
-	if (!adev->mman.discovery_bin)
+	if (!discovery_bin)
 		return -EINVAL;
 
-	adev->ip_top = kzalloc(sizeof(*adev->ip_top), GFP_KERNEL);
-	if (!adev->ip_top)
+	ip_top = kzalloc(sizeof(*ip_top), GFP_KERNEL);
+	if (!ip_top)
 		return -ENOMEM;
 
-	adev->ip_top->adev = adev;
-
-	res = kobject_init_and_add(&adev->ip_top->kobj, &ip_discovery_ktype,
+	ip_top->adev = adev;
+	adev->discovery.ip_top = ip_top;
+	res = kobject_init_and_add(&ip_top->kobj, &ip_discovery_ktype,
 				   &adev->dev->kobj, "ip_discovery");
 	if (res) {
 		DRM_ERROR("Couldn't init and add ip_discovery/");
 		goto Err;
 	}
 
-	die_kset = &adev->ip_top->die_kset;
+	die_kset = &ip_top->die_kset;
 	kobject_set_name(&die_kset->kobj, "%s", "die");
-	die_kset->kobj.parent = &adev->ip_top->kobj;
+	die_kset->kobj.parent = &ip_top->kobj;
 	die_kset->kobj.ktype = &die_kobj_ktype;
-	res = kset_register(&adev->ip_top->die_kset);
+	res = kset_register(&ip_top->die_kset);
 	if (res) {
 		DRM_ERROR("Couldn't register die_kset");
 		goto Err;
@@ -1208,7 +1299,7 @@ static int amdgpu_discovery_sysfs_init(struct amdgpu_device *adev)
 
 	return res;
 Err:
-	kobject_put(&adev->ip_top->kobj);
+	kobject_put(&ip_top->kobj);
 	return res;
 }
 
@@ -1253,10 +1344,11 @@ static void amdgpu_discovery_sysfs_die_free(struct ip_die_entry *ip_die_entry)
 
 static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev)
 {
+	struct ip_discovery_top *ip_top = adev->discovery.ip_top;
 	struct list_head *el, *tmp;
 	struct kset *die_kset;
 
-	die_kset = &adev->ip_top->die_kset;
+	die_kset = &ip_top->die_kset;
 	spin_lock(&die_kset->list_lock);
 	list_for_each_prev_safe(el, tmp, &die_kset->list) {
 		list_del_init(el);
@@ -1265,8 +1357,8 @@ static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev)
 		spin_lock(&die_kset->list_lock);
 	}
 	spin_unlock(&die_kset->list_lock);
-	kobject_put(&adev->ip_top->die_kset.kobj);
-	kobject_put(&adev->ip_top->kobj);
+	kobject_put(&ip_top->die_kset.kobj);
+	kobject_put(&ip_top->kobj);
 }
 
 /* ================================================== */
@@ -1277,35 +1369,39 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev)
 	struct binary_header *bhdr;
 	struct ip_discovery_header *ihdr;
 	struct die_header *dhdr;
+	uint8_t *discovery_bin;
 	struct ip_v4 *ip;
 	uint16_t die_offset;
 	uint16_t ip_offset;
 	uint16_t num_dies;
+	uint32_t wafl_ver;
 	uint16_t num_ips;
+	uint16_t hw_id;
+	uint8_t inst;
 	int hw_ip;
 	int i, j, k;
 	int r;
 
 	r = amdgpu_discovery_init(adev);
-	if (r) {
-		DRM_ERROR("amdgpu_discovery_init failed\n");
+	if (r)
 		return r;
-	}
-
+	discovery_bin = adev->discovery.bin;
+	wafl_ver = 0;
 	adev->gfx.xcc_mask = 0;
 	adev->sdma.sdma_mask = 0;
 	adev->vcn.inst_mask = 0;
 	adev->jpeg.inst_mask = 0;
-	bhdr = (struct binary_header *)adev->mman.discovery_bin;
-	ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin +
-			le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
+	bhdr = (struct binary_header *)discovery_bin;
+	ihdr = (struct ip_discovery_header
+			*)(discovery_bin +
+			   le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
 	num_dies = le16_to_cpu(ihdr->num_dies);
 
 	DRM_DEBUG("number of dies: %d\n", num_dies);
 
 	for (i = 0; i < num_dies; i++) {
 		die_offset = le16_to_cpu(ihdr->die_info[i].die_offset);
-		dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset);
+		dhdr = (struct die_header *)(discovery_bin + die_offset);
 		num_ips = le16_to_cpu(dhdr->num_ips);
 		ip_offset = die_offset + sizeof(*dhdr);
 
@@ -1319,9 +1415,11 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev)
 				le16_to_cpu(dhdr->die_id), num_ips);
 
 		for (j = 0; j < num_ips; j++) {
-			ip = (struct ip_v4 *)(adev->mman.discovery_bin + ip_offset);
+			ip = (struct ip_v4 *)(discovery_bin + ip_offset);
 
-			if (amdgpu_discovery_validate_ip(ip))
+			inst = ip->instance_number;
+			hw_id = le16_to_cpu(ip->hw_id);
+			if (amdgpu_discovery_validate_ip(adev, inst, hw_id))
 				goto next_ip;
 
 			num_base_address = ip->num_base_address;
@@ -1390,6 +1488,10 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev)
 				adev->gfx.xcc_mask |=
 					(1U << ip->instance_number);
 
+			if (!wafl_ver && le16_to_cpu(ip->hw_id) == WAFLC_HWID)
+				wafl_ver = IP_VERSION_FULL(ip->major, ip->minor,
+							   ip->revision, 0, 0);
+
 			for (k = 0; k < num_base_address; k++) {
 				/*
 				 * convert the endianness of base addresses in place,
@@ -1455,22 +1557,32 @@ next_ip:
 		}
 	}
 
+	if (wafl_ver && !adev->ip_versions[XGMI_HWIP][0])
+		adev->ip_versions[XGMI_HWIP][0] = wafl_ver;
+
 	return 0;
 }
 
 static void amdgpu_discovery_harvest_ip(struct amdgpu_device *adev)
 {
+	uint8_t *discovery_bin = adev->discovery.bin;
+	struct ip_discovery_header *ihdr;
+	struct binary_header *bhdr;
 	int vcn_harvest_count = 0;
 	int umc_harvest_count = 0;
+	uint16_t offset, ihdr_ver;
 
+	bhdr = (struct binary_header *)discovery_bin;
+	offset = le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset);
+	ihdr = (struct ip_discovery_header *)(discovery_bin + offset);
+	ihdr_ver = le16_to_cpu(ihdr->version);
 	/*
 	 * Harvest table does not fit Navi1x and legacy GPUs,
 	 * so read harvest bit per IP data structure to set
 	 * harvest configuration.
 	 */
 	if (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(10, 2, 0) &&
-	    amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 3) &&
-	    amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 4)) {
+	    ihdr_ver <= 2) {
 		if ((adev->pdev->device == 0x731E &&
 			(adev->pdev->revision == 0xC6 ||
 			 adev->pdev->revision == 0xC7)) ||
@@ -1509,22 +1621,23 @@ union gc_info {
 
 static int amdgpu_discovery_get_gfx_info(struct amdgpu_device *adev)
 {
+	uint8_t *discovery_bin = adev->discovery.bin;
 	struct binary_header *bhdr;
 	union gc_info *gc_info;
 	u16 offset;
 
-	if (!adev->mman.discovery_bin) {
+	if (!discovery_bin) {
 		DRM_ERROR("ip discovery uninitialized\n");
 		return -EINVAL;
 	}
 
-	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	bhdr = (struct binary_header *)discovery_bin;
 	offset = le16_to_cpu(bhdr->table_list[GC].offset);
 
 	if (!offset)
 		return 0;
 
-	gc_info = (union gc_info *)(adev->mman.discovery_bin + offset);
+	gc_info = (union gc_info *)(discovery_bin + offset);
 
 	switch (le16_to_cpu(gc_info->v1.header.version_major)) {
 	case 1:
@@ -1617,24 +1730,25 @@ union mall_info {
 
 static int amdgpu_discovery_get_mall_info(struct amdgpu_device *adev)
 {
+	uint8_t *discovery_bin = adev->discovery.bin;
 	struct binary_header *bhdr;
 	union mall_info *mall_info;
 	u32 u, mall_size_per_umc, m_s_present, half_use;
 	u64 mall_size;
 	u16 offset;
 
-	if (!adev->mman.discovery_bin) {
+	if (!discovery_bin) {
 		DRM_ERROR("ip discovery uninitialized\n");
 		return -EINVAL;
 	}
 
-	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	bhdr = (struct binary_header *)discovery_bin;
 	offset = le16_to_cpu(bhdr->table_list[MALL_INFO].offset);
 
 	if (!offset)
 		return 0;
 
-	mall_info = (union mall_info *)(adev->mman.discovery_bin + offset);
+	mall_info = (union mall_info *)(discovery_bin + offset);
 
 	switch (le16_to_cpu(mall_info->v1.header.version_major)) {
 	case 1:
@@ -1673,12 +1787,13 @@ union vcn_info {
 
 static int amdgpu_discovery_get_vcn_info(struct amdgpu_device *adev)
 {
+	uint8_t *discovery_bin = adev->discovery.bin;
 	struct binary_header *bhdr;
 	union vcn_info *vcn_info;
 	u16 offset;
 	int v;
 
-	if (!adev->mman.discovery_bin) {
+	if (!discovery_bin) {
 		DRM_ERROR("ip discovery uninitialized\n");
 		return -EINVAL;
 	}
@@ -1693,13 +1808,13 @@ static int amdgpu_discovery_get_vcn_info(struct amdgpu_device *adev)
 		return -EINVAL;
 	}
 
-	bhdr = (struct binary_header *)adev->mman.discovery_bin;
+	bhdr = (struct binary_header *)discovery_bin;
 	offset = le16_to_cpu(bhdr->table_list[VCN_INFO].offset);
 
 	if (!offset)
 		return 0;
 
-	vcn_info = (union vcn_info *)(adev->mman.discovery_bin + offset);
+	vcn_info = (union vcn_info *)(discovery_bin + offset);
 
 	switch (le16_to_cpu(vcn_info->v1.header.version_major)) {
 	case 1:
@@ -1759,6 +1874,7 @@ int amdgpu_discovery_get_nps_info(struct amdgpu_device *adev,
 				  struct amdgpu_gmc_memrange **ranges,
 				  int *range_cnt, bool refresh)
 {
+	uint8_t *discovery_bin = adev->discovery.bin;
 	struct amdgpu_gmc_memrange *mem_ranges;
 	struct binary_header *bhdr;
 	union nps_info *nps_info;
@@ -1775,13 +1891,13 @@ int amdgpu_discovery_get_nps_info(struct amdgpu_device *adev,
 			return r;
 		nps_info = &nps_data;
 	} else {
-		if (!adev->mman.discovery_bin) {
+		if (!discovery_bin) {
 			dev_err(adev->dev,
 				"fetch mem range failed, ip discovery uninitialized\n");
 			return -EINVAL;
 		}
 
-		bhdr = (struct binary_header *)adev->mman.discovery_bin;
+		bhdr = (struct binary_header *)discovery_bin;
 		offset = le16_to_cpu(bhdr->table_list[NPS_INFO].offset);
 
 		if (!offset)
@@ -1791,8 +1907,7 @@ int amdgpu_discovery_get_nps_info(struct amdgpu_device *adev,
 		if (amdgpu_discovery_verify_npsinfo(adev, bhdr))
 			return -ENOENT;
 
-		nps_info =
-			(union nps_info *)(adev->mman.discovery_bin + offset);
+		nps_info = (union nps_info *)(discovery_bin + offset);
 	}
 
 	switch (le16_to_cpu(nps_info->v1.header.version_major)) {
@@ -1864,6 +1979,7 @@ static int amdgpu_discovery_set_common_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		amdgpu_device_ip_block_add(adev, &soc21_common_ip_block);
 		break;
 	case IP_VERSION(12, 0, 0):
@@ -1919,6 +2035,7 @@ static int amdgpu_discovery_set_gmc_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		amdgpu_device_ip_block_add(adev, &gmc_v11_0_ip_block);
 		break;
 	case IP_VERSION(12, 0, 0):
@@ -1998,6 +2115,7 @@ static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(11, 0, 12):
 	case IP_VERSION(11, 0, 13):
 	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 2):
 		amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block);
 		break;
 	case IP_VERSION(11, 0, 8):
@@ -2029,6 +2147,7 @@ static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev)
 		break;
 	case IP_VERSION(14, 0, 2):
 	case IP_VERSION(14, 0, 3):
+	case IP_VERSION(14, 0, 5):
 		amdgpu_device_ip_block_add(adev, &psp_v14_0_ip_block);
 		break;
 	default:
@@ -2056,13 +2175,17 @@ static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(11, 0, 5):
 	case IP_VERSION(11, 0, 9):
 	case IP_VERSION(11, 0, 7):
-	case IP_VERSION(11, 0, 8):
 	case IP_VERSION(11, 0, 11):
 	case IP_VERSION(11, 0, 12):
 	case IP_VERSION(11, 0, 13):
 	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 2):
 		amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block);
 		break;
+	case IP_VERSION(11, 0, 8):
+		if (adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2)
+			amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block);
+		break;
 	case IP_VERSION(12, 0, 0):
 	case IP_VERSION(12, 0, 1):
 		amdgpu_device_ip_block_add(adev, &smu_v12_0_ip_block);
@@ -2079,6 +2202,7 @@ static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(13, 0, 10):
 	case IP_VERSION(13, 0, 11):
 	case IP_VERSION(13, 0, 14):
+	case IP_VERSION(13, 0, 12):
 		amdgpu_device_ip_block_add(adev, &smu_v13_0_ip_block);
 		break;
 	case IP_VERSION(14, 0, 0):
@@ -2086,6 +2210,7 @@ static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(14, 0, 2):
 	case IP_VERSION(14, 0, 3):
 	case IP_VERSION(14, 0, 4):
+	case IP_VERSION(14, 0, 5):
 		amdgpu_device_ip_block_add(adev, &smu_v14_0_ip_block);
 		break;
 	default:
@@ -2137,6 +2262,7 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev)
 		case IP_VERSION(3, 2, 1):
 		case IP_VERSION(3, 5, 0):
 		case IP_VERSION(3, 5, 1):
+		case IP_VERSION(3, 6, 0):
 		case IP_VERSION(4, 1, 0):
 			/* TODO: Fix IP version. DC code expects version 4.0.1 */
 			if (adev->ip_versions[DCE_HWIP][0] == IP_VERSION(4, 1, 0))
@@ -2215,6 +2341,7 @@ static int amdgpu_discovery_set_gc_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		amdgpu_device_ip_block_add(adev, &gfx_v11_0_ip_block);
 		break;
 	case IP_VERSION(12, 0, 0):
@@ -2270,6 +2397,7 @@ static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(6, 1, 0):
 	case IP_VERSION(6, 1, 1):
 	case IP_VERSION(6, 1, 2):
+	case IP_VERSION(6, 1, 3):
 		amdgpu_device_ip_block_add(adev, &sdma_v6_0_ip_block);
 		break;
 	case IP_VERSION(7, 0, 0):
@@ -2282,6 +2410,21 @@ static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev)
 			amdgpu_ip_version(adev, SDMA0_HWIP, 0));
 		return -EINVAL;
 	}
+
+	return 0;
+}
+
+static int amdgpu_discovery_set_ras_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 12):
+	case IP_VERSION(13, 0, 14):
+		amdgpu_device_ip_block_add(adev, &ras_v1_0_ip_block);
+		break;
+	default:
+		break;
+	}
 	return 0;
 }
 
@@ -2393,6 +2536,7 @@ static int amdgpu_discovery_set_mes_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		amdgpu_device_ip_block_add(adev, &mes_v11_0_ip_block);
 		adev->enable_mes = true;
 		adev->enable_mes_kiq = true;
@@ -2480,9 +2624,16 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 
 	switch (adev->asic_type) {
 	case CHIP_VEGA10:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
 		vega10_reg_base_init(adev);
 		adev->sdma.num_instances = 2;
+		adev->sdma.sdma_mask = 3;
 		adev->gmc.num_umc = 4;
+		adev->gfx.xcc_mask = 1;
 		adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 0, 0);
 		adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 0, 0);
 		adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 0, 0);
@@ -2502,9 +2653,16 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 		adev->ip_versions[DCI_HWIP][0] = IP_VERSION(12, 0, 0);
 		break;
 	case CHIP_VEGA12:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
 		vega10_reg_base_init(adev);
 		adev->sdma.num_instances = 2;
+		adev->sdma.sdma_mask = 3;
 		adev->gmc.num_umc = 4;
+		adev->gfx.xcc_mask = 1;
 		adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 3, 0);
 		adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 3, 0);
 		adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 0, 1);
@@ -2524,10 +2682,17 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 		adev->ip_versions[DCI_HWIP][0] = IP_VERSION(12, 0, 1);
 		break;
 	case CHIP_RAVEN:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
 		vega10_reg_base_init(adev);
 		adev->sdma.num_instances = 1;
+		adev->sdma.sdma_mask = 1;
 		adev->vcn.num_vcn_inst = 1;
 		adev->gmc.num_umc = 2;
+		adev->gfx.xcc_mask = 1;
 		if (adev->apu_flags & AMD_APU_IS_RAVEN2) {
 			adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 2, 0);
 			adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 2, 0);
@@ -2565,9 +2730,16 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 		}
 		break;
 	case CHIP_VEGA20:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
 		vega20_reg_base_init(adev);
 		adev->sdma.num_instances = 2;
+		adev->sdma.sdma_mask = 3;
 		adev->gmc.num_umc = 8;
+		adev->gfx.xcc_mask = 1;
 		adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 0);
 		adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 0);
 		adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 2, 0);
@@ -2588,10 +2760,17 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 		adev->ip_versions[DCI_HWIP][0] = IP_VERSION(12, 1, 0);
 		break;
 	case CHIP_ARCTURUS:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
 		arct_reg_base_init(adev);
 		adev->sdma.num_instances = 8;
+		adev->sdma.sdma_mask = 0xff;
 		adev->vcn.num_vcn_inst = 2;
 		adev->gmc.num_umc = 8;
+		adev->gfx.xcc_mask = 1;
 		adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 1);
 		adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 1);
 		adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 2, 1);
@@ -2616,10 +2795,17 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 		adev->ip_versions[UVD_HWIP][1] = IP_VERSION(2, 5, 0);
 		break;
 	case CHIP_ALDEBARAN:
+		/* This is not fatal.  We only need the discovery
+		 * binary for sysfs.  We don't need it for a
+		 * functional system.
+		 */
+		amdgpu_discovery_init(adev);
 		aldebaran_reg_base_init(adev);
 		adev->sdma.num_instances = 5;
+		adev->sdma.sdma_mask = 0x1f;
 		adev->vcn.num_vcn_inst = 2;
 		adev->gmc.num_umc = 4;
+		adev->gfx.xcc_mask = 1;
 		adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 2);
 		adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 2);
 		adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 4, 0);
@@ -2641,10 +2827,44 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 		adev->ip_versions[UVD_HWIP][1] = IP_VERSION(2, 6, 0);
 		adev->ip_versions[XGMI_HWIP][0] = IP_VERSION(6, 1, 0);
 		break;
+	case CHIP_CYAN_SKILLFISH:
+		if (adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2) {
+			r = amdgpu_discovery_reg_base_init(adev);
+			if (r)
+				return -EINVAL;
+
+			amdgpu_discovery_harvest_ip(adev);
+			amdgpu_discovery_get_gfx_info(adev);
+			amdgpu_discovery_get_mall_info(adev);
+			amdgpu_discovery_get_vcn_info(adev);
+		} else {
+			cyan_skillfish_reg_base_init(adev);
+			adev->sdma.num_instances = 2;
+			adev->sdma.sdma_mask = 3;
+			adev->gfx.xcc_mask = 1;
+			adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(2, 0, 3);
+			adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(2, 0, 3);
+			adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(5, 0, 1);
+			adev->ip_versions[HDP_HWIP][0] = IP_VERSION(5, 0, 1);
+			adev->ip_versions[SDMA0_HWIP][0] = IP_VERSION(5, 0, 1);
+			adev->ip_versions[SDMA1_HWIP][1] = IP_VERSION(5, 0, 1);
+			adev->ip_versions[DF_HWIP][0] = IP_VERSION(3, 5, 0);
+			adev->ip_versions[NBIO_HWIP][0] = IP_VERSION(2, 1, 1);
+			adev->ip_versions[UMC_HWIP][0] = IP_VERSION(8, 1, 1);
+			adev->ip_versions[MP0_HWIP][0] = IP_VERSION(11, 0, 8);
+			adev->ip_versions[MP1_HWIP][0] = IP_VERSION(11, 0, 8);
+			adev->ip_versions[THM_HWIP][0] = IP_VERSION(11, 0, 1);
+			adev->ip_versions[SMUIO_HWIP][0] = IP_VERSION(11, 0, 8);
+			adev->ip_versions[GC_HWIP][0] = IP_VERSION(10, 1, 3);
+			adev->ip_versions[UVD_HWIP][0] = IP_VERSION(2, 0, 3);
+		}
+		break;
 	default:
 		r = amdgpu_discovery_reg_base_init(adev);
-		if (r)
-			return -EINVAL;
+		if (r) {
+			drm_err(&adev->ddev, "discovery failed: %d\n", r);
+			return r;
+		}
 
 		amdgpu_discovery_harvest_ip(adev);
 		amdgpu_discovery_get_gfx_info(adev);
@@ -2708,6 +2928,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		adev->family = AMDGPU_FAMILY_GC_11_5_0;
 		break;
 	case IP_VERSION(12, 0, 0):
@@ -2733,19 +2954,13 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		adev->flags |= AMD_IS_APU;
 		break;
 	default:
 		break;
 	}
 
-	if (amdgpu_ip_version(adev, XGMI_HWIP, 0) == IP_VERSION(4, 8, 0))
-		adev->gmc.xgmi.supported = true;
-
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4))
-		adev->ip_versions[XGMI_HWIP][0] = IP_VERSION(6, 4, 0);
-
 	/* set NBIO version */
 	switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) {
 	case IP_VERSION(6, 1, 0):
@@ -2766,11 +2981,13 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 		adev->nbio.hdp_flush_reg = &nbio_v7_4_hdp_flush_reg;
 		break;
 	case IP_VERSION(7, 9, 0):
+	case IP_VERSION(7, 9, 1):
 		adev->nbio.funcs = &nbio_v7_9_funcs;
 		adev->nbio.hdp_flush_reg = &nbio_v7_9_hdp_flush_reg;
 		break;
 	case IP_VERSION(7, 11, 0):
 	case IP_VERSION(7, 11, 1):
+	case IP_VERSION(7, 11, 2):
 	case IP_VERSION(7, 11, 3):
 		adev->nbio.funcs = &nbio_v7_11_funcs;
 		adev->nbio.hdp_flush_reg = &nbio_v7_11_hdp_flush_reg;
@@ -2898,6 +3115,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 	case IP_VERSION(11, 0, 10):
 	case IP_VERSION(11, 0, 11):
 	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 2):
 	case IP_VERSION(13, 0, 1):
 	case IP_VERSION(13, 0, 9):
 	case IP_VERSION(13, 0, 10):
@@ -2907,6 +3125,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 		adev->smuio.funcs = &smuio_v13_0_funcs;
 		break;
 	case IP_VERSION(13, 0, 3):
+	case IP_VERSION(13, 0, 11):
 		adev->smuio.funcs = &smuio_v13_0_3_funcs;
 		if (adev->smuio.funcs->get_pkg_type(adev) == AMDGPU_PKG_TYPE_APU) {
 			adev->flags |= AMD_IS_APU;
@@ -2986,6 +3205,10 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
 	if (r)
 		return r;
 
+	r = amdgpu_discovery_set_ras_ip_blocks(adev);
+	if (r)
+		return r;
+
 	if ((adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT &&
 	     !amdgpu_sriov_vf(adev)) ||
 	    (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO && amdgpu_dpm == 1)) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h
index b44d56465c5b..4ce04486cc31 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h
@@ -24,9 +24,21 @@
 #ifndef __AMDGPU_DISCOVERY__
 #define __AMDGPU_DISCOVERY__
 
+#include <linux/debugfs.h>
+
 #define DISCOVERY_TMR_SIZE      (10 << 10)
 #define DISCOVERY_TMR_OFFSET    (64 << 10)
 
+struct ip_discovery_top;
+
+struct amdgpu_discovery_info {
+	struct debugfs_blob_wrapper debugfs_blob;
+	struct ip_discovery_top *ip_top;
+	uint32_t size;
+	uint8_t *bin;
+	bool reserve_tmr;
+};
+
 void amdgpu_discovery_fini(struct amdgpu_device *adev);
 int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
index 35c778426a7c..b5d34797d606 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
@@ -332,8 +332,6 @@ int amdgpu_display_crtc_set_config(struct drm_mode_set *set,
 		if (crtc->enabled)
 			active = true;
 
-	pm_runtime_mark_last_busy(dev->dev);
-
 	adev = drm_to_adev(dev);
 	/* if we have active crtcs and we don't have a power ref,
 	 * take the current one
@@ -1196,13 +1194,14 @@ static int amdgpu_display_get_fb_info(const struct amdgpu_framebuffer *amdgpu_fb
 static int amdgpu_display_gem_fb_verify_and_init(struct drm_device *dev,
 						 struct amdgpu_framebuffer *rfb,
 						 struct drm_file *file_priv,
+						 const struct drm_format_info *info,
 						 const struct drm_mode_fb_cmd2 *mode_cmd,
 						 struct drm_gem_object *obj)
 {
 	int ret;
 
 	rfb->base.obj[0] = obj;
-	drm_helper_mode_fill_fb_struct(dev, &rfb->base, mode_cmd);
+	drm_helper_mode_fill_fb_struct(dev, &rfb->base, info, mode_cmd);
 	/* Verify that the modifier is supported. */
 	if (!drm_any_plane_has_format(dev, mode_cmd->pixel_format,
 				      mode_cmd->modifier[0])) {
@@ -1297,6 +1296,7 @@ static int amdgpu_display_framebuffer_init(struct drm_device *dev,
 struct drm_framebuffer *
 amdgpu_display_user_framebuffer_create(struct drm_device *dev,
 				       struct drm_file *file_priv,
+				       const struct drm_format_info *info,
 				       const struct drm_mode_fb_cmd2 *mode_cmd)
 {
 	struct amdgpu_framebuffer *amdgpu_fb;
@@ -1317,7 +1317,7 @@ amdgpu_display_user_framebuffer_create(struct drm_device *dev,
 	/* Handle is imported dma-buf, so cannot be migrated to VRAM for scanout */
 	bo = gem_to_amdgpu_bo(obj);
 	domains = amdgpu_display_supported_domains(drm_to_adev(dev), bo->flags);
-	if (obj->import_attach && !(domains & AMDGPU_GEM_DOMAIN_GTT)) {
+	if (drm_gem_is_imported(obj) && !(domains & AMDGPU_GEM_DOMAIN_GTT)) {
 		drm_dbg_kms(dev, "Cannot create framebuffer from imported dma_buf\n");
 		drm_gem_object_put(obj);
 		return ERR_PTR(-EINVAL);
@@ -1330,7 +1330,7 @@ amdgpu_display_user_framebuffer_create(struct drm_device *dev,
 	}
 
 	ret = amdgpu_display_gem_fb_verify_and_init(dev, amdgpu_fb, file_priv,
-						    mode_cmd, obj);
+						    info, mode_cmd, obj);
 	if (ret) {
 		kfree(amdgpu_fb);
 		drm_gem_object_put(obj);
@@ -1363,6 +1363,64 @@ static const struct drm_prop_enum_list amdgpu_dither_enum_list[] = {
 	{ AMDGPU_FMT_DITHER_ENABLE, "on" },
 };
 
+/**
+ * DOC: property for adaptive backlight modulation
+ *
+ * The 'adaptive backlight modulation' property is used for the compositor to
+ * directly control the adaptive backlight modulation power savings feature
+ * that is part of DCN hardware.
+ *
+ * The property will be attached specifically to eDP panels that support it.
+ *
+ * The property is by default set to 'sysfs' to allow the sysfs file 'panel_power_savings'
+ * to be able to control it.
+ * If set to 'off' the compositor will ensure it stays off.
+ * The other values 'min', 'bias min', 'bias max', and 'max' will control the
+ * intensity of the power savings.
+ *
+ * Modifying this value can have implications on color accuracy, so tread
+ * carefully.
+ */
+static int amdgpu_display_setup_abm_prop(struct amdgpu_device *adev)
+{
+	const struct drm_prop_enum_list props[] = {
+		{ ABM_SYSFS_CONTROL, "sysfs" },
+		{ ABM_LEVEL_OFF, "off" },
+		{ ABM_LEVEL_MIN, "min" },
+		{ ABM_LEVEL_BIAS_MIN, "bias min" },
+		{ ABM_LEVEL_BIAS_MAX, "bias max" },
+		{ ABM_LEVEL_MAX, "max" },
+	};
+	struct drm_property *prop;
+	int i;
+
+	if (!adev->dc_enabled)
+		return 0;
+
+	prop = drm_property_create(adev_to_drm(adev), DRM_MODE_PROP_ENUM,
+				"adaptive backlight modulation",
+				6);
+	if (!prop)
+		return -ENOMEM;
+
+	for (i = 0; i < ARRAY_SIZE(props); i++) {
+		int ret;
+
+		ret = drm_property_add_enum(prop, props[i].type,
+						props[i].name);
+
+		if (ret) {
+			drm_property_destroy(adev_to_drm(adev), prop);
+
+			return ret;
+		}
+	}
+
+	adev->mode_info.abm_level_property = prop;
+
+	return 0;
+}
+
 int amdgpu_display_modeset_create_props(struct amdgpu_device *adev)
 {
 	int sz;
@@ -1409,7 +1467,7 @@ int amdgpu_display_modeset_create_props(struct amdgpu_device *adev)
 					 "dither",
 					 amdgpu_dither_enum_list, sz);
 
-	return 0;
+	return amdgpu_display_setup_abm_prop(adev);
 }
 
 void amdgpu_display_update_priority(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h
index dfa0d642ac16..49a29bf47a37 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h
@@ -44,6 +44,7 @@ uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev,
 struct drm_framebuffer *
 amdgpu_display_user_framebuffer_create(struct drm_device *dev,
 				       struct drm_file *file_priv,
+				       const struct drm_format_info *info,
 				       const struct drm_mode_fb_cmd2 *mode_cmd);
 const struct drm_format_info *
 amdgpu_lookup_format_info(u32 format, uint64_t modifier);
@@ -54,4 +55,11 @@ int amdgpu_display_resume_helper(struct amdgpu_device *adev);
 int amdgpu_display_get_scanout_buffer(struct drm_plane *plane,
 				      struct drm_scanout_buffer *sb);
 
+#define ABM_SYSFS_CONTROL	-1
+#define ABM_LEVEL_OFF		0
+#define ABM_LEVEL_MIN		1
+#define ABM_LEVEL_BIAS_MIN	2
+#define ABM_LEVEL_BIAS_MAX	3
+#define ABM_LEVEL_MAX		4
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
index 9f627caedc3f..e22cfa7c6d32 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
@@ -43,6 +43,29 @@
 #include <linux/dma-fence-array.h>
 #include <linux/pci-p2pdma.h>
 
+static const struct dma_buf_attach_ops amdgpu_dma_buf_attach_ops;
+
+/**
+ * dma_buf_attach_adev - Helper to get adev of an attachment
+ *
+ * @attach: attachment
+ *
+ * Returns:
+ * A struct amdgpu_device * if the attaching device is an amdgpu device or
+ * partition, NULL otherwise.
+ */
+static struct amdgpu_device *dma_buf_attach_adev(struct dma_buf_attachment *attach)
+{
+	if (attach->importer_ops == &amdgpu_dma_buf_attach_ops) {
+		struct drm_gem_object *obj = attach->importer_priv;
+		struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+		return amdgpu_ttm_adev(bo->tbo.bdev);
+	}
+
+	return NULL;
+}
+
 /**
  * amdgpu_dma_buf_attach - &dma_buf_ops.attach implementation
  *
@@ -54,15 +77,48 @@
 static int amdgpu_dma_buf_attach(struct dma_buf *dmabuf,
 				 struct dma_buf_attachment *attach)
 {
+	struct amdgpu_device *attach_adev = dma_buf_attach_adev(attach);
 	struct drm_gem_object *obj = dmabuf->priv;
 	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	int r;
 
-	if (pci_p2pdma_distance(adev->pdev, attach->dev, false) < 0)
+	/*
+	 * Disable peer-to-peer access for DCC-enabled VRAM surfaces on GFX12+.
+	 * Such buffers cannot be safely accessed over P2P due to device-local
+	 * compression metadata. Fallback to system-memory path instead.
+	 * Device supports GFX12 (GC 12.x or newer)
+	 * BO was created with the AMDGPU_GEM_CREATE_GFX12_DCC flag
+	 *
+	 */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(12, 0, 0) &&
+	    bo->flags & AMDGPU_GEM_CREATE_GFX12_DCC)
 		attach->peer2peer = false;
 
+	/*
+	 * Disable peer-to-peer access for DCC-enabled VRAM surfaces on GFX12+.
+	 * Such buffers cannot be safely accessed over P2P due to device-local
+	 * compression metadata. Fallback to system-memory path instead.
+	 * Device supports GFX12 (GC 12.x or newer)
+	 * BO was created with the AMDGPU_GEM_CREATE_GFX12_DCC flag
+	 *
+	 */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(12, 0, 0) &&
+	    bo->flags & AMDGPU_GEM_CREATE_GFX12_DCC)
+		attach->peer2peer = false;
+
+	if (!amdgpu_dmabuf_is_xgmi_accessible(attach_adev, bo) &&
+	    pci_p2pdma_distance(adev->pdev, attach->dev, false) < 0)
+		attach->peer2peer = false;
+
+	r = dma_resv_lock(bo->tbo.base.resv, NULL);
+	if (r)
+		return r;
+
 	amdgpu_vm_bo_update_shared(bo);
 
+	dma_resv_unlock(bo->tbo.base.resv);
+
 	return 0;
 }
 
@@ -75,11 +131,35 @@ static int amdgpu_dma_buf_attach(struct dma_buf *dmabuf,
  */
 static int amdgpu_dma_buf_pin(struct dma_buf_attachment *attach)
 {
-	struct drm_gem_object *obj = attach->dmabuf->priv;
-	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	struct dma_buf *dmabuf = attach->dmabuf;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(dmabuf->priv);
+	u32 domains = bo->allowed_domains;
+
+	dma_resv_assert_held(dmabuf->resv);
+
+	/* Try pinning into VRAM to allow P2P with RDMA NICs without ODP
+	 * support if all attachments can do P2P. If any attachment can't do
+	 * P2P just pin into GTT instead.
+	 *
+	 * To avoid with conflicting pinnings between GPUs and RDMA when move
+	 * notifiers are disabled, only allow pinning in VRAM when move
+	 * notiers are enabled.
+	 */
+	if (!IS_ENABLED(CONFIG_DMABUF_MOVE_NOTIFY)) {
+		domains &= ~AMDGPU_GEM_DOMAIN_VRAM;
+	} else {
+		list_for_each_entry(attach, &dmabuf->attachments, node)
+			if (!attach->peer2peer)
+				domains &= ~AMDGPU_GEM_DOMAIN_VRAM;
+	}
+
+	if (domains & AMDGPU_GEM_DOMAIN_VRAM)
+		bo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
 
-	/* pin buffer into GTT */
-	return amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT);
+	if (WARN_ON(!domains))
+		return -EINVAL;
+
+	return amdgpu_bo_pin(bo, domains);
 }
 
 /**
@@ -134,9 +214,6 @@ static struct sg_table *amdgpu_dma_buf_map(struct dma_buf_attachment *attach,
 		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 		if (r)
 			return ERR_PTR(r);
-
-	} else if (bo->tbo.resource->mem_type != TTM_PL_TT) {
-		return ERR_PTR(-EBUSY);
 	}
 
 	switch (bo->tbo.resource->mem_type) {
@@ -153,6 +230,11 @@ static struct sg_table *amdgpu_dma_buf_map(struct dma_buf_attachment *attach,
 		break;
 
 	case TTM_PL_VRAM:
+		/* XGMI-accessible memory should never be DMA-mapped */
+		if (WARN_ON(amdgpu_dmabuf_is_xgmi_accessible(
+				dma_buf_attach_adev(attach), bo)))
+			return ERR_PTR(-EINVAL);
+
 		r = amdgpu_vram_mgr_alloc_sgt(adev, bo->tbo.resource, 0,
 					      bo->tbo.base.size, attach->dev,
 					      dir, &sgt);
@@ -184,7 +266,7 @@ static void amdgpu_dma_buf_unmap(struct dma_buf_attachment *attach,
 				 struct sg_table *sgt,
 				 enum dma_data_direction dir)
 {
-	if (sgt->sgl->page_link) {
+	if (sg_page(sgt->sgl)) {
 		dma_unmap_sgtable(attach->dev, sgt, dir, 0);
 		sg_free_table(sgt);
 		kfree(sgt);
@@ -234,6 +316,36 @@ static int amdgpu_dma_buf_begin_cpu_access(struct dma_buf *dma_buf,
 	return ret;
 }
 
+static int amdgpu_dma_buf_vmap(struct dma_buf *dma_buf, struct iosys_map *map)
+{
+	struct drm_gem_object *obj = dma_buf->priv;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	int ret;
+
+	/*
+	 * Pin to keep buffer in place while it's vmap'ed. The actual
+	 * domain is not that important as long as it's mapable. Using
+	 * GTT and VRAM should be compatible with most use cases.
+	 */
+	ret = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT | AMDGPU_GEM_DOMAIN_VRAM);
+	if (ret)
+		return ret;
+	ret = drm_gem_dmabuf_vmap(dma_buf, map);
+	if (ret)
+		amdgpu_bo_unpin(bo);
+
+	return ret;
+}
+
+static void amdgpu_dma_buf_vunmap(struct dma_buf *dma_buf, struct iosys_map *map)
+{
+	struct drm_gem_object *obj = dma_buf->priv;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+	drm_gem_dmabuf_vunmap(dma_buf, map);
+	amdgpu_bo_unpin(bo);
+}
+
 const struct dma_buf_ops amdgpu_dmabuf_ops = {
 	.attach = amdgpu_dma_buf_attach,
 	.pin = amdgpu_dma_buf_pin,
@@ -243,8 +355,8 @@ const struct dma_buf_ops amdgpu_dmabuf_ops = {
 	.release = drm_gem_dmabuf_release,
 	.begin_cpu_access = amdgpu_dma_buf_begin_cpu_access,
 	.mmap = drm_gem_dmabuf_mmap,
-	.vmap = drm_gem_dmabuf_vmap,
-	.vunmap = drm_gem_dmabuf_vunmap,
+	.vmap = amdgpu_dma_buf_vmap,
+	.vunmap = amdgpu_dma_buf_vunmap,
 };
 
 /**
@@ -262,11 +374,23 @@ struct dma_buf *amdgpu_gem_prime_export(struct drm_gem_object *gobj,
 {
 	struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj);
 	struct dma_buf *buf;
+	struct ttm_operation_ctx ctx = {
+		.interruptible = true,
+		.no_wait_gpu = true,
+		/* We opt to avoid OOM on system pages allocations */
+		.gfp_retry_mayfail = true,
+		.allow_res_evict = false,
+	};
+	int ret;
 
 	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm) ||
 	    bo->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID)
 		return ERR_PTR(-EPERM);
 
+	ret = ttm_bo_setup_export(&bo->tbo, &ctx);
+	if (ret)
+		return ERR_PTR(ret);
+
 	buf = drm_gem_prime_export(gobj, flags);
 	if (!IS_ERR(buf))
 		buf->ops = &amdgpu_dmabuf_ops;
@@ -459,7 +583,10 @@ bool amdgpu_dmabuf_is_xgmi_accessible(struct amdgpu_device *adev,
 	struct drm_gem_object *obj = &bo->tbo.base;
 	struct drm_gem_object *gobj;
 
-	if (obj->import_attach) {
+	if (!adev)
+		return false;
+
+	if (drm_gem_is_imported(obj)) {
 		struct dma_buf *dma_buf = obj->import_attach->dmabuf;
 
 		if (dma_buf->ops != &amdgpu_dmabuf_ops)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c
index 3f3662e8b871..3040437d99c2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c
@@ -41,7 +41,8 @@ u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index)
 	if (index < adev->doorbell.num_kernel_doorbells)
 		return readl(adev->doorbell.cpu_addr + index);
 
-	DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index);
+	dev_err(adev->dev, "reading beyond doorbell aperture: 0x%08x!\n",
+		index);
 	return 0;
 }
 
@@ -63,7 +64,8 @@ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v)
 	if (index < adev->doorbell.num_kernel_doorbells)
 		writel(v, adev->doorbell.cpu_addr + index);
 	else
-		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
+		dev_err(adev->dev,
+			"writing beyond doorbell aperture: 0x%08x!\n", index);
 }
 
 /**
@@ -83,7 +85,8 @@ u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index)
 	if (index < adev->doorbell.num_kernel_doorbells)
 		return atomic64_read((atomic64_t *)(adev->doorbell.cpu_addr + index));
 
-	DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index);
+	dev_err(adev->dev, "reading beyond doorbell aperture: 0x%08x!\n",
+		index);
 	return 0;
 }
 
@@ -105,7 +108,8 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v)
 	if (index < adev->doorbell.num_kernel_doorbells)
 		atomic64_set((atomic64_t *)(adev->doorbell.cpu_addr + index), v);
 	else
-		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
+		dev_err(adev->dev,
+			"writing beyond doorbell aperture: 0x%08x!\n", index);
 }
 
 /**
@@ -166,7 +170,8 @@ int amdgpu_doorbell_create_kernel_doorbells(struct amdgpu_device *adev)
 				    NULL,
 				    (void **)&adev->doorbell.cpu_addr);
 	if (r) {
-		DRM_ERROR("Failed to allocate kernel doorbells, err=%d\n", r);
+		dev_err(adev->dev,
+			"Failed to allocate kernel doorbells, err=%d\n", r);
 		return r;
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 95a05b03f799..2dfbddcef9ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -51,6 +51,8 @@
 #include "amdgpu_reset.h"
 #include "amdgpu_sched.h"
 #include "amdgpu_xgmi.h"
+#include "amdgpu_userq.h"
+#include "amdgpu_userq_fence.h"
 #include "../amdxcp/amdgpu_xcp_drv.h"
 
 /*
@@ -121,9 +123,12 @@
  * - 3.59.0 - Cleared VRAM
  * - 3.60.0 - Add AMDGPU_TILING_GFX12_DCC_WRITE_COMPRESS_DISABLE (Vulkan requirement)
  * - 3.61.0 - Contains fix for RV/PCO compute queues
+ * - 3.62.0 - Add AMDGPU_IDS_FLAGS_MODE_PF, AMDGPU_IDS_FLAGS_MODE_VF & AMDGPU_IDS_FLAGS_MODE_PT
+ * - 3.63.0 - GFX12 display DCC supports 256B max compressed block size
+ * - 3.64.0 - Userq IP support query
  */
 #define KMS_DRIVER_MAJOR	3
-#define KMS_DRIVER_MINOR	61
+#define KMS_DRIVER_MINOR	64
 #define KMS_DRIVER_PATCHLEVEL	0
 
 /*
@@ -136,6 +141,11 @@ enum AMDGPU_DEBUG_MASK {
 	AMDGPU_DEBUG_USE_VRAM_FW_BUF = BIT(3),
 	AMDGPU_DEBUG_ENABLE_RAS_ACA = BIT(4),
 	AMDGPU_DEBUG_ENABLE_EXP_RESETS = BIT(5),
+	AMDGPU_DEBUG_DISABLE_GPU_RING_RESET = BIT(6),
+	AMDGPU_DEBUG_SMU_POOL = BIT(7),
+	AMDGPU_DEBUG_VM_USERPTR = BIT(8),
+	AMDGPU_DEBUG_DISABLE_RAS_CE_LOG = BIT(9),
+	AMDGPU_DEBUG_ENABLE_CE_CS = BIT(10)
 };
 
 unsigned int amdgpu_vram_limit = UINT_MAX;
@@ -172,7 +182,8 @@ uint amdgpu_pg_mask = 0xffffffff;
 uint amdgpu_sdma_phase_quantum = 32;
 char *amdgpu_disable_cu;
 char *amdgpu_virtual_display;
-bool enforce_isolation;
+int amdgpu_enforce_isolation = -1;
+int amdgpu_modeset = -1;
 
 /* Specifies the default granularity for SVM, used in buffer
  * migration and restoration of backing memory when handling
@@ -233,6 +244,8 @@ int amdgpu_agp = -1; /* auto */
 int amdgpu_wbrf = -1;
 int amdgpu_damage_clips = -1; /* auto */
 int amdgpu_umsch_mm_fwlog;
+int amdgpu_rebar = -1; /* auto */
+int amdgpu_user_queue = -1;
 
 DECLARE_DYNDBG_CLASSMAP(drm_debug_classes, DD_CLASS_TYPE_DISJOINT_BITS, 0,
 			"DRM_UT_CORE",
@@ -283,6 +296,7 @@ module_param_named(gartsize, amdgpu_gart_size, uint, 0600);
  * DOC: gttsize (int)
  * Restrict the size of GTT domain (for userspace use) in MiB for testing.
  * The default is -1 (Use value specified by TTM).
+ * This parameter is deprecated and will be removed in the future.
  */
 MODULE_PARM_DESC(gttsize, "Size of the GTT userspace domain in megabytes (-1 = auto)");
 module_param_named(gttsize, amdgpu_gtt_size, int, 0600);
@@ -298,7 +312,7 @@ module_param_named(moverate, amdgpu_moverate, int, 0600);
  * DOC: audio (int)
  * Set HDMI/DPAudio. Only affects non-DC display handling. The default is -1 (Enabled), set 0 to disabled it.
  */
-MODULE_PARM_DESC(audio, "Audio enable (-1 = auto, 0 = disable, 1 = enable)");
+MODULE_PARM_DESC(audio, "HDMI/DP Audio enable for non DC displays (-1 = auto, 0 = disable, 1 = enable)");
 module_param_named(audio, amdgpu_audio, int, 0444);
 
 /**
@@ -340,22 +354,16 @@ module_param_named(svm_default_granularity, amdgpu_svm_default_granularity, uint
  * DOC: lockup_timeout (string)
  * Set GPU scheduler timeout value in ms.
  *
- * The format can be [Non-Compute] or [GFX,Compute,SDMA,Video]. That is there can be one or
- * multiple values specified. 0 and negative values are invalidated. They will be adjusted
- * to the default timeout.
+ * The format can be [single value] for setting all timeouts at once or
+ * [GFX,Compute,SDMA,Video] to set individual timeouts.
+ * Negative values mean infinity.
  *
- * - With one value specified, the setting will apply to all non-compute jobs.
- * - With multiple values specified, the first one will be for GFX.
- *   The second one is for Compute. The third and fourth ones are
- *   for SDMA and Video.
- *
- * By default(with no lockup_timeout settings), the timeout for all non-compute(GFX, SDMA and Video)
- * jobs is 10000. The timeout for compute is 60000.
+ * By default(with no lockup_timeout settings), the timeout for all queues is 2000.
  */
-MODULE_PARM_DESC(lockup_timeout, "GPU lockup timeout in ms (default: for bare metal 10000 for non-compute jobs and 60000 for compute jobs; "
-		"for passthrough or sriov, 10000 for all jobs. 0: keep default value. negative: infinity timeout), format: for bare metal [Non-Compute] or [GFX,Compute,SDMA,Video]; "
-		"for passthrough or sriov [all jobs] or [GFX,Compute,SDMA,Video].");
-module_param_string(lockup_timeout, amdgpu_lockup_timeout, sizeof(amdgpu_lockup_timeout), 0444);
+MODULE_PARM_DESC(lockup_timeout,
+		 "GPU lockup timeout in ms (default: 2000. 0: keep default value. negative: infinity timeout), format: [single value for all] or [GFX,Compute,SDMA,Video].");
+module_param_string(lockup_timeout, amdgpu_lockup_timeout,
+		    sizeof(amdgpu_lockup_timeout), 0444);
 
 /**
  * DOC: dpm (int)
@@ -610,39 +618,39 @@ module_param_named(timeout_period, amdgpu_watchdog_timer.period, uint, 0644);
 
 /**
  * DOC: si_support (int)
- * Set SI support driver. This parameter works after set config CONFIG_DRM_AMDGPU_SI. For SI asic, when radeon driver is enabled,
- * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available,
- * otherwise using amdgpu driver.
- */
+ * 1 = enabled, 0 = disabled, -1 = default
+ *
+ * SI (Southern Islands) are first generation GCN GPUs, supported by both
+ * drivers: radeon (old) and amdgpu (new). This parameter controls whether
+ * amdgpu should support SI.
+ * By default, SI dedicated GPUs are supported by amdgpu.
+ * Only relevant when CONFIG_DRM_AMDGPU_SI is enabled to build SI support in amdgpu.
+ * See also radeon.si_support which should be disabled when amdgpu.si_support is
+ * enabled, and vice versa.
+ */
+int amdgpu_si_support = -1;
 #ifdef CONFIG_DRM_AMDGPU_SI
-
-#if IS_ENABLED(CONFIG_DRM_RADEON) || IS_ENABLED(CONFIG_DRM_RADEON_MODULE)
-int amdgpu_si_support;
-MODULE_PARM_DESC(si_support, "SI support (1 = enabled, 0 = disabled (default))");
-#else
-int amdgpu_si_support = 1;
-MODULE_PARM_DESC(si_support, "SI support (1 = enabled (default), 0 = disabled)");
-#endif
-
+MODULE_PARM_DESC(si_support, "SI support (1 = enabled, 0 = disabled, -1 = default)");
 module_param_named(si_support, amdgpu_si_support, int, 0444);
 #endif
 
 /**
  * DOC: cik_support (int)
- * Set CIK support driver. This parameter works after set config CONFIG_DRM_AMDGPU_CIK. For CIK asic, when radeon driver is enabled,
- * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available,
- * otherwise using amdgpu driver.
- */
+ * 1 = enabled, 0 = disabled, -1 = default
+ *
+ * CIK (Sea Islands) are second generation GCN GPUs, supported by both
+ * drivers: radeon (old) and amdgpu (new). This parameter controls whether
+ * amdgpu should support CIK.
+ * By default:
+ * - CIK dedicated GPUs are supported by amdgpu.
+ * - CIK APUs are supported by radeon (except when radeon is not built).
+ * Only relevant when CONFIG_DRM_AMDGPU_CIK is enabled to build CIK support in amdgpu.
+ * See also radeon.cik_support which should be disabled when amdgpu.cik_support is
+ * enabled, and vice versa.
+ */
+int amdgpu_cik_support = -1;
 #ifdef CONFIG_DRM_AMDGPU_CIK
-
-#if IS_ENABLED(CONFIG_DRM_RADEON) || IS_ENABLED(CONFIG_DRM_RADEON_MODULE)
-int amdgpu_cik_support;
-MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled, 0 = disabled (default))");
-#else
-int amdgpu_cik_support = 1;
-MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled (default), 0 = disabled)");
-#endif
-
+MODULE_PARM_DESC(cik_support, "CIK support  (1 = enabled, 0 = disabled, -1 = default)");
 module_param_named(cik_support, amdgpu_cik_support, int, 0444);
 #endif
 
@@ -873,7 +881,7 @@ module_param_named(dcfeaturemask, amdgpu_dc_feature_mask, uint, 0444);
 
 /**
  * DOC: dcdebugmask (uint)
- * Override display features enabled. See enum DC_DEBUG_MASK in drivers/gpu/drm/amd/include/amd_shared.h.
+ * Display debug options. See enum DC_DEBUG_MASK in drivers/gpu/drm/amd/include/amd_shared.h.
  */
 MODULE_PARM_DESC(dcdebugmask, "all debug options disabled (default))");
 module_param_named(dcdebugmask, amdgpu_dc_debug_mask, uint, 0444);
@@ -947,7 +955,7 @@ module_param_named(tmz, amdgpu_tmz, int, 0444);
  */
 MODULE_PARM_DESC(
 	freesync_video,
-	"Enable freesync modesetting optimization feature (0 = off (default), 1 = on)");
+	"Adds additional modes via VRR for refresh changes without a full modeset (0 = off (default), 1 = on)");
 module_param_named(freesync_video, amdgpu_freesync_vid_mode, uint, 0444);
 
 /**
@@ -963,7 +971,7 @@ module_param_named_unsafe(reset_method, amdgpu_reset_method, int, 0644);
  * result in the GPU entering bad status when the number of total
  * faulty pages by ECC exceeds the threshold value.
  */
-MODULE_PARM_DESC(bad_page_threshold, "Bad page threshold(-1 = ignore threshold (default value), 0 = disable bad page retirement, -2 = driver sets threshold)");
+MODULE_PARM_DESC(bad_page_threshold, "Bad page threshold(-1 = ignore threshold (default value), 0 = disable bad page retirement, -2 = threshold determined by a formula, 0 < threshold < max records, user-defined threshold)");
 module_param_named(bad_page_threshold, amdgpu_bad_page_threshold, int, 0444);
 
 MODULE_PARM_DESC(num_kcq, "number of kernel compute queue user want to setup (8 if set to greater than 8 or less than 0, only affect gfx 8+)");
@@ -1027,11 +1035,20 @@ module_param_named(user_partt_mode, amdgpu_user_partt_mode, uint, 0444);
 
 
 /**
- * DOC: enforce_isolation (bool)
- * enforce process isolation between graphics and compute via using the same reserved vmid.
+ * DOC: enforce_isolation (int)
+ * enforce process isolation between graphics and compute.
+ * (-1 = auto, 0 = disable, 1 = enable, 2 = enable legacy mode, 3 = enable without cleaner shader)
  */
-module_param(enforce_isolation, bool, 0444);
-MODULE_PARM_DESC(enforce_isolation, "enforce process isolation between graphics and compute . enforce_isolation = on");
+module_param_named(enforce_isolation, amdgpu_enforce_isolation, int, 0444);
+MODULE_PARM_DESC(enforce_isolation,
+"enforce process isolation between graphics and compute. (-1 = auto, 0 = disable, 1 = enable, 2 = enable legacy mode, 3 = enable without cleaner shader)");
+
+/**
+ * DOC: modeset (int)
+ * Override nomodeset (1 = override, -1 = auto). The default is -1 (auto).
+ */
+MODULE_PARM_DESC(modeset, "Override nomodeset (1 = enable, -1 = auto)");
+module_param_named(modeset, amdgpu_modeset, int, 0444);
 
 /**
  * DOC: seamless (int)
@@ -1049,6 +1066,11 @@ module_param_named(seamless, amdgpu_seamless, int, 0444);
  *   limits the VRAM size reported to ROCm applications to the visible
  *   size, usually 256MB.
  * - 0x4: Disable GPU soft recovery, always do a full reset
+ * - 0x8: Use VRAM for firmware loading
+ * - 0x10: Enable ACA based RAS logging
+ * - 0x20: Enable experimental resets
+ * - 0x40: Disable ring resets
+ * - 0x80: Use VRAM for SMU pool
  */
 MODULE_PARM_DESC(debug_mask, "debug options for amdgpu, disabled by default");
 module_param_named_unsafe(debug_mask, amdgpu_debug_mask, uint, 0444);
@@ -1078,6 +1100,28 @@ MODULE_PARM_DESC(wbrf,
 	"Enable Wifi RFI interference mitigation (0 = disabled, 1 = enabled, -1 = auto(default)");
 module_param_named(wbrf, amdgpu_wbrf, int, 0444);
 
+/**
+ * DOC: rebar (int)
+ * Allow BAR resizing.  Disable this to prevent the driver from attempting
+ * to resize the BAR if the GPU supports it and there is available MMIO space.
+ * Note that this just prevents the driver from resizing the BAR.  The BIOS
+ * may have already resized the BAR at boot time.
+ */
+MODULE_PARM_DESC(rebar, "Resizable BAR (-1 = auto (default), 0 = disable, 1 = enable)");
+module_param_named(rebar, amdgpu_rebar, int, 0444);
+
+/**
+ * DOC: user_queue (int)
+ * Enable user queues on systems that support user queues. Possible values:
+ *
+ * - -1 = auto (ASIC specific default)
+ * -  0 = user queues disabled
+ * -  1 = user queues enabled and kernel queues enabled (if supported)
+ * -  2 = user queues enabled and kernel queues disabled
+ */
+MODULE_PARM_DESC(user_queue, "Enable user queues (-1 = auto (default), 0 = disable, 1 = enable, 2 = enable UQs and disable KQs)");
+module_param_named(user_queue, amdgpu_user_queue, int, 0444);
+
 /* These devices are not supported by amdgpu.
  * They are supported by the mach64, r128, radeon drivers
  */
@@ -1791,7 +1835,6 @@ static const u16 amdgpu_unsupported_pciidlist[] = {
 };
 
 static const struct pci_device_id pciidlist[] = {
-#ifdef CONFIG_DRM_AMDGPU_SI
 	{0x1002, 0x6780, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
 	{0x1002, 0x6784, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
 	{0x1002, 0x6788, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
@@ -1864,8 +1907,6 @@ static const struct pci_device_id pciidlist[] = {
 	{0x1002, 0x6665, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
 	{0x1002, 0x6667, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
 	{0x1002, 0x666F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
-#endif
-#ifdef CONFIG_DRM_AMDGPU_CIK
 	/* Kaveri */
 	{0x1002, 0x1304, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
 	{0x1002, 0x1305, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
@@ -1948,7 +1989,6 @@ static const struct pci_device_id pciidlist[] = {
 	{0x1002, 0x985D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
 	{0x1002, 0x985E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
 	{0x1002, 0x985F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
-#endif
 	/* topaz */
 	{0x1002, 0x6900, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
 	{0x1002, 0x6901, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
@@ -2127,6 +2167,11 @@ static const struct pci_device_id pciidlist[] = {
 	{0x1002, 0x7410, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ALDEBARAN},
 
 	/* CYAN_SKILLFISH */
+	{0x1002, 0x13DB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+	{0x1002, 0x13F9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+	{0x1002, 0x13FA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+	{0x1002, 0x13FB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
+	{0x1002, 0x13FC, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
 	{0x1002, 0x13FE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
 	{0x1002, 0x143F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CYAN_SKILLFISH|AMD_IS_APU},
 
@@ -2183,7 +2228,6 @@ static void amdgpu_get_secondary_funcs(struct amdgpu_device *adev)
 						adev->pdev->bus->number, i);
 		if (p) {
 			pm_runtime_get_sync(&p->dev);
-			pm_runtime_mark_last_busy(&p->dev);
 			pm_runtime_put_autosuspend(&p->dev);
 			pci_dev_put(p);
 		}
@@ -2221,6 +2265,29 @@ static void amdgpu_init_debug_options(struct amdgpu_device *adev)
 		pr_info("debug: enable experimental reset features\n");
 		adev->debug_exp_resets = true;
 	}
+
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_DISABLE_GPU_RING_RESET) {
+		pr_info("debug: ring reset disabled\n");
+		adev->debug_disable_gpu_ring_reset = true;
+	}
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_SMU_POOL) {
+		pr_info("debug: use vram for smu pool\n");
+		adev->pm.smu_debug_mask |= SMU_DEBUG_POOL_USE_VRAM;
+	}
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_VM_USERPTR) {
+		pr_info("debug: VM mode debug for userptr is enabled\n");
+		adev->debug_vm_userptr = true;
+	}
+
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_DISABLE_RAS_CE_LOG) {
+		pr_info("debug: disable kernel logs of correctable errors\n");
+		adev->debug_disable_ce_logs = true;
+	}
+
+	if (amdgpu_debug_mask & AMDGPU_DEBUG_ENABLE_CE_CS) {
+		pr_info("debug: allowing command submission to CE engine\n");
+		adev->debug_enable_ce_cs = true;
+	}
 }
 
 static unsigned long amdgpu_fix_asic_type(struct pci_dev *pdev, unsigned long flags)
@@ -2239,6 +2306,72 @@ static unsigned long amdgpu_fix_asic_type(struct pci_dev *pdev, unsigned long fl
 	return flags;
 }
 
+static bool amdgpu_support_enabled(struct device *dev,
+				   const enum amd_asic_type family)
+{
+	const char *gen;
+	const char *param;
+	int module_param = -1;
+	bool radeon_support_built = IS_ENABLED(CONFIG_DRM_RADEON);
+	bool amdgpu_support_built = false;
+	bool support_by_default = false;
+
+	switch (family) {
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	case CHIP_OLAND:
+	case CHIP_HAINAN:
+		gen = "SI";
+		param = "si_support";
+		module_param = amdgpu_si_support;
+		amdgpu_support_built = IS_ENABLED(CONFIG_DRM_AMDGPU_SI);
+		support_by_default = true;
+		break;
+
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+		support_by_default = true;
+		fallthrough;
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		gen = "CIK";
+		param = "cik_support";
+		module_param = amdgpu_cik_support;
+		amdgpu_support_built = IS_ENABLED(CONFIG_DRM_AMDGPU_CIK);
+		break;
+
+	default:
+		/* All other chips are supported by amdgpu only */
+		return true;
+	}
+
+	if (!amdgpu_support_built) {
+		dev_info(dev, "amdgpu built without %s support\n", gen);
+		return false;
+	}
+
+	if ((module_param == -1 && (support_by_default || !radeon_support_built)) ||
+	    module_param == 1) {
+		if (radeon_support_built)
+			dev_info(dev, "%s support provided by amdgpu.\n"
+				 "Use radeon.%s=1 amdgpu.%s=0 to override.\n",
+				 gen, param, param);
+
+		return true;
+	}
+
+	if (radeon_support_built)
+		dev_info(dev, "%s support provided by radeon.\n"
+			 "Use radeon.%s=0 amdgpu.%s=1 to override.\n",
+			 gen, param, param);
+	else if (module_param == 0)
+		dev_info(dev, "%s support disabled by module param\n", gen);
+
+	return false;
+}
+
 static int amdgpu_pci_probe(struct pci_dev *pdev,
 			    const struct pci_device_id *ent)
 {
@@ -2248,6 +2381,12 @@ static int amdgpu_pci_probe(struct pci_dev *pdev,
 	int ret, retry = 0, i;
 	bool supports_atomic = false;
 
+	if ((pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA ||
+	    (pdev->class >> 8) == PCI_CLASS_DISPLAY_OTHER) {
+		if (drm_firmware_drivers_only() && amdgpu_modeset == -1)
+			return -EINVAL;
+	}
+
 	/* skip devices which are owned by radeon */
 	for (i = 0; i < ARRAY_SIZE(amdgpu_unsupported_pciidlist); i++) {
 		if (amdgpu_unsupported_pciidlist[i] == pdev->device)
@@ -2258,7 +2397,7 @@ static int amdgpu_pci_probe(struct pci_dev *pdev,
 		amdgpu_aspm = 0;
 
 	if (amdgpu_virtual_display ||
-	    amdgpu_device_asic_has_dc_support(flags & AMD_ASIC_MASK))
+	    amdgpu_device_asic_has_dc_support(pdev, flags & AMD_ASIC_MASK))
 		supports_atomic = true;
 
 	if ((flags & AMD_EXP_HW_SUPPORT) && !amdgpu_exp_hw_support) {
@@ -2280,40 +2419,8 @@ static int amdgpu_pci_probe(struct pci_dev *pdev,
 		return -ENOTSUPP;
 	}
 
-#ifdef CONFIG_DRM_AMDGPU_SI
-	if (!amdgpu_si_support) {
-		switch (flags & AMD_ASIC_MASK) {
-		case CHIP_TAHITI:
-		case CHIP_PITCAIRN:
-		case CHIP_VERDE:
-		case CHIP_OLAND:
-		case CHIP_HAINAN:
-			dev_info(&pdev->dev,
-				 "SI support provided by radeon.\n");
-			dev_info(&pdev->dev,
-				 "Use radeon.si_support=0 amdgpu.si_support=1 to override.\n"
-				);
-			return -ENODEV;
-		}
-	}
-#endif
-#ifdef CONFIG_DRM_AMDGPU_CIK
-	if (!amdgpu_cik_support) {
-		switch (flags & AMD_ASIC_MASK) {
-		case CHIP_KAVERI:
-		case CHIP_BONAIRE:
-		case CHIP_HAWAII:
-		case CHIP_KABINI:
-		case CHIP_MULLINS:
-			dev_info(&pdev->dev,
-				 "CIK support provided by radeon.\n");
-			dev_info(&pdev->dev,
-				 "Use radeon.cik_support=0 amdgpu.cik_support=1 to override.\n"
-				);
-			return -ENODEV;
-		}
-	}
-#endif
+	if (!amdgpu_support_enabled(&pdev->dev, flags & AMD_ASIC_MASK))
+		return -ENODEV;
 
 	adev = devm_drm_dev_alloc(&pdev->dev, &amdgpu_kms_driver, typeof(*adev), ddev);
 	if (IS_ERR(adev))
@@ -2380,10 +2487,10 @@ retry_init:
 
 	if (adev->pm.rpm_mode != AMDGPU_RUNPM_NONE) {
 		/* only need to skip on ATPX */
-		if (amdgpu_device_supports_px(ddev))
+		if (amdgpu_device_supports_px(adev))
 			dev_pm_set_driver_flags(ddev->dev, DPM_FLAG_NO_DIRECT_COMPLETE);
 		/* we want direct complete for BOCO */
-		if (amdgpu_device_supports_boco(ddev))
+		if (amdgpu_device_supports_boco(adev))
 			dev_pm_set_driver_flags(ddev->dev, DPM_FLAG_SMART_PREPARE |
 						DPM_FLAG_SMART_SUSPEND |
 						DPM_FLAG_MAY_SKIP_RESUME);
@@ -2392,7 +2499,6 @@ retry_init:
 
 		pm_runtime_allow(ddev->dev);
 
-		pm_runtime_mark_last_busy(ddev->dev);
 		pm_runtime_put_autosuspend(ddev->dev);
 
 		pci_wake_from_d3(pdev, TRUE);
@@ -2416,9 +2522,9 @@ retry_init:
 		 * into D0 state. Then there will be a PMFW-aware D-state
 		 * transition(D0->D3) on runpm suspend.
 		 */
-		if (amdgpu_device_supports_baco(ddev) &&
+		if (amdgpu_device_supports_baco(adev) &&
 		    !(adev->flags & AMD_IS_APU) &&
-		    (adev->asic_type >= CHIP_NAVI10))
+		    adev->asic_type >= CHIP_NAVI10)
 			amdgpu_get_secondary_funcs(adev);
 	}
 
@@ -2435,6 +2541,7 @@ amdgpu_pci_remove(struct pci_dev *pdev)
 	struct drm_device *dev = pci_get_drvdata(pdev);
 	struct amdgpu_device *adev = drm_to_adev(dev);
 
+	amdgpu_ras_eeprom_check_and_recover(adev);
 	amdgpu_xcp_dev_unplug(adev);
 	amdgpu_gmc_prepare_nps_mode_change(adev);
 	drm_dev_unplug(dev);
@@ -2464,6 +2571,10 @@ amdgpu_pci_shutdown(struct pci_dev *pdev)
 	if (amdgpu_ras_intr_triggered())
 		return;
 
+	/* device maybe not resumed here, return immediately in this case */
+	if (adev->in_s4 && adev->in_suspend)
+		return;
+
 	/* if we are running in a VM, make sure the device
 	 * torn down properly on reboot/shutdown.
 	 * unfortunately we can't detect certain
@@ -2471,7 +2582,8 @@ amdgpu_pci_shutdown(struct pci_dev *pdev)
 	 */
 	if (!amdgpu_passthrough(adev))
 		adev->mp1_state = PP_MP1_STATE_UNLOAD;
-	amdgpu_device_ip_suspend(adev);
+	amdgpu_device_prepare(dev);
+	amdgpu_device_suspend(dev, true);
 	adev->mp1_state = PP_MP1_STATE_NONE;
 }
 
@@ -2480,11 +2592,14 @@ static int amdgpu_pmops_prepare(struct device *dev)
 	struct drm_device *drm_dev = dev_get_drvdata(dev);
 	struct amdgpu_device *adev = drm_to_adev(drm_dev);
 
+	/* device maybe not resumed here, return immediately in this case */
+	if (adev->in_s4 && adev->in_suspend)
+		return 0;
+
 	/* Return a positive number here so
 	 * DPM_FLAG_SMART_SUSPEND works properly
 	 */
-	if (amdgpu_device_supports_boco(drm_dev) &&
-	    pm_runtime_suspended(dev))
+	if (amdgpu_device_supports_boco(adev) && pm_runtime_suspended(dev))
 		return 1;
 
 	/* if we will not support s3 or s2i for the device
@@ -2499,7 +2614,7 @@ static int amdgpu_pmops_prepare(struct device *dev)
 
 static void amdgpu_pmops_complete(struct device *dev)
 {
-	/* nothing to do */
+	amdgpu_device_complete(dev_get_drvdata(dev));
 }
 
 static int amdgpu_pmops_suspend(struct device *dev)
@@ -2511,8 +2626,24 @@ static int amdgpu_pmops_suspend(struct device *dev)
 		adev->in_s0ix = true;
 	else if (amdgpu_acpi_is_s3_active(adev))
 		adev->in_s3 = true;
-	if (!adev->in_s0ix && !adev->in_s3)
+	if (!adev->in_s0ix && !adev->in_s3) {
+#if IS_ENABLED(CONFIG_SUSPEND)
+		/* don't allow going deep first time followed by s2idle the next time */
+		if (adev->last_suspend_state != PM_SUSPEND_ON &&
+		    adev->last_suspend_state != pm_suspend_target_state) {
+			drm_err_once(drm_dev, "Unsupported suspend state %d\n",
+				     pm_suspend_target_state);
+			return -EINVAL;
+		}
+#endif
 		return 0;
+	}
+
+#if IS_ENABLED(CONFIG_SUSPEND)
+	/* cache the state last used for suspend */
+	adev->last_suspend_state = pm_suspend_target_state;
+#endif
+
 	return amdgpu_device_suspend(drm_dev, true);
 }
 
@@ -2520,9 +2651,14 @@ static int amdgpu_pmops_suspend_noirq(struct device *dev)
 {
 	struct drm_device *drm_dev = dev_get_drvdata(dev);
 	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+	int r;
 
-	if (amdgpu_acpi_should_gpu_reset(adev))
-		return amdgpu_asic_reset(adev);
+	if (amdgpu_acpi_should_gpu_reset(adev)) {
+		amdgpu_device_lock_reset_domain(adev->reset_domain);
+		r = amdgpu_asic_reset(adev);
+		amdgpu_device_unlock_reset_domain(adev->reset_domain);
+		return r;
+	}
 
 	return 0;
 }
@@ -2555,7 +2691,6 @@ static int amdgpu_pmops_freeze(struct device *dev)
 	int r;
 
 	r = amdgpu_device_suspend(drm_dev, true);
-	adev->in_s4 = false;
 	if (r)
 		return r;
 
@@ -2568,12 +2703,21 @@ static int amdgpu_pmops_thaw(struct device *dev)
 {
 	struct drm_device *drm_dev = dev_get_drvdata(dev);
 
+	/* do not resume device if it's normal hibernation */
+	if (!pm_hibernate_is_recovering() && !pm_hibernation_mode_is_suspend())
+		return 0;
+
 	return amdgpu_device_resume(drm_dev, true);
 }
 
 static int amdgpu_pmops_poweroff(struct device *dev)
 {
 	struct drm_device *drm_dev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+
+	/* device maybe not resumed here, return immediately in this case */
+	if (adev->in_s4 && adev->in_suspend)
+		return 0;
 
 	return amdgpu_device_suspend(drm_dev, true);
 }
@@ -2652,6 +2796,15 @@ static int amdgpu_runtime_idle_check_display(struct device *dev)
 	return 0;
 }
 
+static int amdgpu_runtime_idle_check_userq(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(drm_dev);
+
+	return xa_empty(&adev->userq_doorbell_xa) ? 0 : -EBUSY;
+}
+
 static int amdgpu_pmops_runtime_suspend(struct device *dev)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
@@ -2667,6 +2820,9 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev)
 	ret = amdgpu_runtime_idle_check_display(dev);
 	if (ret)
 		return ret;
+	ret = amdgpu_runtime_idle_check_userq(dev);
+	if (ret)
+		return ret;
 
 	/* wait for all rings to drain before suspending */
 	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
@@ -2720,7 +2876,7 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev)
 		/* nothing to do */
 	} else if ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) ||
 			(adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO)) {
-		amdgpu_device_baco_enter(drm_dev);
+		amdgpu_device_baco_enter(adev);
 	}
 
 	dev_dbg(&pdev->dev, "asic/device is runtime suspended\n");
@@ -2761,7 +2917,7 @@ static int amdgpu_pmops_runtime_resume(struct device *dev)
 		pci_set_master(pdev);
 	} else if ((adev->pm.rpm_mode == AMDGPU_RUNPM_BACO) ||
 			(adev->pm.rpm_mode == AMDGPU_RUNPM_BAMACO)) {
-		amdgpu_device_baco_exit(drm_dev);
+		amdgpu_device_baco_exit(adev);
 	}
 	ret = amdgpu_device_resume(drm_dev, false);
 	if (ret) {
@@ -2788,12 +2944,32 @@ static int amdgpu_pmops_runtime_idle(struct device *dev)
 	}
 
 	ret = amdgpu_runtime_idle_check_display(dev);
+	if (ret)
+		goto done;
 
-	pm_runtime_mark_last_busy(dev);
+	ret = amdgpu_runtime_idle_check_userq(dev);
+done:
 	pm_runtime_autosuspend(dev);
 	return ret;
 }
 
+static int amdgpu_drm_release(struct inode *inode, struct file *filp)
+{
+	struct drm_file *file_priv = filp->private_data;
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
+	struct drm_device *dev = file_priv->minor->dev;
+	int idx;
+
+	if (fpriv && drm_dev_enter(dev, &idx)) {
+		fpriv->evf_mgr.fd_closing = true;
+		amdgpu_eviction_fence_destroy(&fpriv->evf_mgr);
+		amdgpu_userq_mgr_fini(&fpriv->userq_mgr);
+		drm_dev_exit(idx);
+	}
+
+	return drm_release(inode, filp);
+}
+
 long amdgpu_drm_ioctl(struct file *filp,
 		      unsigned int cmd, unsigned long arg)
 {
@@ -2808,22 +2984,21 @@ long amdgpu_drm_ioctl(struct file *filp,
 
 	ret = drm_ioctl(filp, cmd, arg);
 
-	pm_runtime_mark_last_busy(dev->dev);
 out:
 	pm_runtime_put_autosuspend(dev->dev);
 	return ret;
 }
 
 static const struct dev_pm_ops amdgpu_pm_ops = {
-	.prepare = amdgpu_pmops_prepare,
-	.complete = amdgpu_pmops_complete,
-	.suspend = amdgpu_pmops_suspend,
-	.suspend_noirq = amdgpu_pmops_suspend_noirq,
-	.resume = amdgpu_pmops_resume,
-	.freeze = amdgpu_pmops_freeze,
-	.thaw = amdgpu_pmops_thaw,
-	.poweroff = amdgpu_pmops_poweroff,
-	.restore = amdgpu_pmops_restore,
+	.prepare = pm_sleep_ptr(amdgpu_pmops_prepare),
+	.complete = pm_sleep_ptr(amdgpu_pmops_complete),
+	.suspend = pm_sleep_ptr(amdgpu_pmops_suspend),
+	.suspend_noirq = pm_sleep_ptr(amdgpu_pmops_suspend_noirq),
+	.resume = pm_sleep_ptr(amdgpu_pmops_resume),
+	.freeze = pm_sleep_ptr(amdgpu_pmops_freeze),
+	.thaw = pm_sleep_ptr(amdgpu_pmops_thaw),
+	.poweroff = pm_sleep_ptr(amdgpu_pmops_poweroff),
+	.restore = pm_sleep_ptr(amdgpu_pmops_restore),
 	.runtime_suspend = amdgpu_pmops_runtime_suspend,
 	.runtime_resume = amdgpu_pmops_runtime_resume,
 	.runtime_idle = amdgpu_pmops_runtime_idle,
@@ -2845,7 +3020,7 @@ static const struct file_operations amdgpu_driver_kms_fops = {
 	.owner = THIS_MODULE,
 	.open = drm_open,
 	.flush = amdgpu_flush,
-	.release = drm_release,
+	.release = amdgpu_drm_release,
 	.unlocked_ioctl = amdgpu_drm_ioctl,
 	.mmap = drm_gem_mmap,
 	.poll = drm_poll,
@@ -2892,6 +3067,10 @@ const struct drm_ioctl_desc amdgpu_ioctls_kms[] = {
 	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_VA, amdgpu_gem_va_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_OP, amdgpu_gem_op_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_USERPTR, amdgpu_gem_userptr_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_USERQ, amdgpu_userq_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_USERQ_SIGNAL, amdgpu_userq_signal_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_USERQ_WAIT, amdgpu_userq_wait_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_LIST_HANDLES, amdgpu_gem_list_handles_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
 };
 
 static const struct drm_driver amdgpu_kms_driver = {
@@ -2965,7 +3144,7 @@ static struct pci_driver amdgpu_kms_pci_driver = {
 	.probe = amdgpu_pci_probe,
 	.remove = amdgpu_pci_remove,
 	.shutdown = amdgpu_pci_shutdown,
-	.driver.pm = &amdgpu_pm_ops,
+	.driver.pm = pm_ptr(&amdgpu_pm_ops),
 	.err_handler = &amdgpu_pci_err_handler,
 	.dev_groups = amdgpu_sysfs_groups,
 };
@@ -2974,14 +3153,11 @@ static int __init amdgpu_init(void)
 {
 	int r;
 
-	if (drm_firmware_drivers_only())
-		return -EINVAL;
-
 	r = amdgpu_sync_init();
 	if (r)
 		goto error_sync;
 
-	r = amdgpu_fence_slab_init();
+	r = amdgpu_userq_fence_slab_init();
 	if (r)
 		goto error_fence;
 
@@ -3015,7 +3191,7 @@ static void __exit amdgpu_exit(void)
 	amdgpu_unregister_atpx_handler();
 	amdgpu_acpi_release();
 	amdgpu_sync_fini();
-	amdgpu_fence_slab_fini();
+	amdgpu_userq_fence_slab_fini();
 	mmu_notifier_synchronize();
 	amdgpu_xcp_drv_release();
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c
new file mode 100644
index 000000000000..23d7d0b0d625
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c
@@ -0,0 +1,241 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/sched.h>
+#include <drm/drm_exec.h>
+#include "amdgpu.h"
+
+#define work_to_evf_mgr(w, name) container_of(w, struct amdgpu_eviction_fence_mgr, name)
+#define evf_mgr_to_fpriv(e) container_of(e, struct amdgpu_fpriv, evf_mgr)
+
+static const char *
+amdgpu_eviction_fence_get_driver_name(struct dma_fence *fence)
+{
+	return "amdgpu_eviction_fence";
+}
+
+static const char *
+amdgpu_eviction_fence_get_timeline_name(struct dma_fence *f)
+{
+	struct amdgpu_eviction_fence *ef;
+
+	ef = container_of(f, struct amdgpu_eviction_fence, base);
+	return ef->timeline_name;
+}
+
+int
+amdgpu_eviction_fence_replace_fence(struct amdgpu_eviction_fence_mgr *evf_mgr,
+				    struct drm_exec *exec)
+{
+	struct amdgpu_eviction_fence *old_ef, *new_ef;
+	struct drm_gem_object *obj;
+	unsigned long index;
+	int ret;
+
+	if (evf_mgr->ev_fence &&
+	    !dma_fence_is_signaled(&evf_mgr->ev_fence->base))
+		return 0;
+	/*
+	 * Steps to replace eviction fence:
+	 * * lock all objects in exec (caller)
+	 * * create a new eviction fence
+	 * * update new eviction fence in evf_mgr
+	 * * attach the new eviction fence to BOs
+	 * * release the old fence
+	 * * unlock the objects (caller)
+	 */
+	new_ef = amdgpu_eviction_fence_create(evf_mgr);
+	if (!new_ef) {
+		DRM_ERROR("Failed to create new eviction fence\n");
+		return -ENOMEM;
+	}
+
+	/* Update the eviction fence now */
+	spin_lock(&evf_mgr->ev_fence_lock);
+	old_ef = evf_mgr->ev_fence;
+	evf_mgr->ev_fence = new_ef;
+	spin_unlock(&evf_mgr->ev_fence_lock);
+
+	/* Attach the new fence */
+	drm_exec_for_each_locked_object(exec, index, obj) {
+		struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+		if (!bo)
+			continue;
+		ret = amdgpu_eviction_fence_attach(evf_mgr, bo);
+		if (ret) {
+			DRM_ERROR("Failed to attch new eviction fence\n");
+			goto free_err;
+		}
+	}
+
+	/* Free old fence */
+	if (old_ef)
+		dma_fence_put(&old_ef->base);
+	return 0;
+
+free_err:
+	kfree(new_ef);
+	return ret;
+}
+
+static void
+amdgpu_eviction_fence_suspend_worker(struct work_struct *work)
+{
+	struct amdgpu_eviction_fence_mgr *evf_mgr = work_to_evf_mgr(work, suspend_work.work);
+	struct amdgpu_fpriv *fpriv = evf_mgr_to_fpriv(evf_mgr);
+	struct amdgpu_userq_mgr *uq_mgr = &fpriv->userq_mgr;
+	struct amdgpu_eviction_fence *ev_fence;
+
+	mutex_lock(&uq_mgr->userq_mutex);
+	spin_lock(&evf_mgr->ev_fence_lock);
+	ev_fence = evf_mgr->ev_fence;
+	if (ev_fence)
+		dma_fence_get(&ev_fence->base);
+	else
+		goto unlock;
+	spin_unlock(&evf_mgr->ev_fence_lock);
+
+	amdgpu_userq_evict(uq_mgr, ev_fence);
+
+	mutex_unlock(&uq_mgr->userq_mutex);
+	dma_fence_put(&ev_fence->base);
+	return;
+
+unlock:
+	spin_unlock(&evf_mgr->ev_fence_lock);
+	mutex_unlock(&uq_mgr->userq_mutex);
+}
+
+static bool amdgpu_eviction_fence_enable_signaling(struct dma_fence *f)
+{
+	struct amdgpu_eviction_fence_mgr *evf_mgr;
+	struct amdgpu_eviction_fence *ev_fence;
+
+	if (!f)
+		return true;
+
+	ev_fence = to_ev_fence(f);
+	evf_mgr = ev_fence->evf_mgr;
+
+	schedule_delayed_work(&evf_mgr->suspend_work, 0);
+	return true;
+}
+
+static const struct dma_fence_ops amdgpu_eviction_fence_ops = {
+	.get_driver_name = amdgpu_eviction_fence_get_driver_name,
+	.get_timeline_name = amdgpu_eviction_fence_get_timeline_name,
+	.enable_signaling = amdgpu_eviction_fence_enable_signaling,
+};
+
+void amdgpu_eviction_fence_signal(struct amdgpu_eviction_fence_mgr *evf_mgr,
+				  struct amdgpu_eviction_fence *ev_fence)
+{
+	spin_lock(&evf_mgr->ev_fence_lock);
+	dma_fence_signal(&ev_fence->base);
+	spin_unlock(&evf_mgr->ev_fence_lock);
+}
+
+struct amdgpu_eviction_fence *
+amdgpu_eviction_fence_create(struct amdgpu_eviction_fence_mgr *evf_mgr)
+{
+	struct amdgpu_eviction_fence *ev_fence;
+
+	ev_fence = kzalloc(sizeof(*ev_fence), GFP_KERNEL);
+	if (!ev_fence)
+		return NULL;
+
+	ev_fence->evf_mgr = evf_mgr;
+	get_task_comm(ev_fence->timeline_name, current);
+	spin_lock_init(&ev_fence->lock);
+	dma_fence_init64(&ev_fence->base, &amdgpu_eviction_fence_ops,
+			 &ev_fence->lock, evf_mgr->ev_fence_ctx,
+			 atomic_inc_return(&evf_mgr->ev_fence_seq));
+	return ev_fence;
+}
+
+void amdgpu_eviction_fence_destroy(struct amdgpu_eviction_fence_mgr *evf_mgr)
+{
+	struct amdgpu_eviction_fence *ev_fence;
+
+	/* Wait for any pending work to execute */
+	flush_delayed_work(&evf_mgr->suspend_work);
+
+	spin_lock(&evf_mgr->ev_fence_lock);
+	ev_fence = evf_mgr->ev_fence;
+	spin_unlock(&evf_mgr->ev_fence_lock);
+
+	if (!ev_fence)
+		return;
+
+	dma_fence_wait(&ev_fence->base, false);
+
+	/* Last unref of ev_fence */
+	dma_fence_put(&ev_fence->base);
+}
+
+int amdgpu_eviction_fence_attach(struct amdgpu_eviction_fence_mgr *evf_mgr,
+				 struct amdgpu_bo *bo)
+{
+	struct amdgpu_eviction_fence *ev_fence;
+	struct dma_resv *resv = bo->tbo.base.resv;
+	int ret;
+
+	if (!resv)
+		return 0;
+
+	ret = dma_resv_reserve_fences(resv, 1);
+	if (ret) {
+		DRM_DEBUG_DRIVER("Failed to resv fence space\n");
+		return ret;
+	}
+
+	spin_lock(&evf_mgr->ev_fence_lock);
+	ev_fence = evf_mgr->ev_fence;
+	if (ev_fence)
+		dma_resv_add_fence(resv, &ev_fence->base, DMA_RESV_USAGE_BOOKKEEP);
+	spin_unlock(&evf_mgr->ev_fence_lock);
+
+	return 0;
+}
+
+void amdgpu_eviction_fence_detach(struct amdgpu_eviction_fence_mgr *evf_mgr,
+				  struct amdgpu_bo *bo)
+{
+	struct dma_fence *stub = dma_fence_get_stub();
+
+	dma_resv_replace_fences(bo->tbo.base.resv, evf_mgr->ev_fence_ctx,
+				stub, DMA_RESV_USAGE_BOOKKEEP);
+	dma_fence_put(stub);
+}
+
+int amdgpu_eviction_fence_init(struct amdgpu_eviction_fence_mgr *evf_mgr)
+{
+	/* This needs to be done one time per open */
+	atomic_set(&evf_mgr->ev_fence_seq, 0);
+	evf_mgr->ev_fence_ctx = dma_fence_context_alloc(1);
+	spin_lock_init(&evf_mgr->ev_fence_lock);
+
+	INIT_DELAYED_WORK(&evf_mgr->suspend_work, amdgpu_eviction_fence_suspend_worker);
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.h
new file mode 100644
index 000000000000..fcd867b7147d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.h
@@ -0,0 +1,69 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef AMDGPU_EV_FENCE_H_
+#define AMDGPU_EV_FENCE_H_
+
+struct amdgpu_eviction_fence {
+	struct dma_fence base;
+	spinlock_t	 lock;
+	char		 timeline_name[TASK_COMM_LEN];
+	struct amdgpu_eviction_fence_mgr *evf_mgr;
+};
+
+struct amdgpu_eviction_fence_mgr {
+	u64			ev_fence_ctx;
+	atomic_t		ev_fence_seq;
+	spinlock_t		ev_fence_lock;
+	struct amdgpu_eviction_fence *ev_fence;
+	struct delayed_work	suspend_work;
+	uint8_t fd_closing;
+};
+
+/* Eviction fence helper functions */
+struct amdgpu_eviction_fence *
+amdgpu_eviction_fence_create(struct amdgpu_eviction_fence_mgr *evf_mgr);
+
+void
+amdgpu_eviction_fence_destroy(struct amdgpu_eviction_fence_mgr *evf_mgr);
+
+int
+amdgpu_eviction_fence_attach(struct amdgpu_eviction_fence_mgr *evf_mgr,
+			     struct amdgpu_bo *bo);
+
+void
+amdgpu_eviction_fence_detach(struct amdgpu_eviction_fence_mgr *evf_mgr,
+			     struct amdgpu_bo *bo);
+
+int
+amdgpu_eviction_fence_init(struct amdgpu_eviction_fence_mgr *evf_mgr);
+
+void
+amdgpu_eviction_fence_signal(struct amdgpu_eviction_fence_mgr *evf_mgr,
+			     struct amdgpu_eviction_fence *ev_fence);
+
+int
+amdgpu_eviction_fence_replace_fence(struct amdgpu_eviction_fence_mgr *evf_mgr,
+				    struct drm_exec *exec);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c
index 91d638098889..b349bb3676d5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c
@@ -70,6 +70,7 @@ void amdgpu_show_fdinfo(struct drm_printer *p, struct drm_file *file)
 		[AMDGPU_PL_GWS] = "gws",
 		[AMDGPU_PL_OA] = "oa",
 		[AMDGPU_PL_DOORBELL] = "doorbell",
+		[AMDGPU_PL_MMIO_REMAP] = "mmioremap",
 	};
 	unsigned int hw_ip, i;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
index 2f24a6aa13bf..c7843e336310 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
@@ -42,50 +42,14 @@
 #include "amdgpu_reset.h"
 
 /*
- * Fences mark an event in the GPUs pipeline and are used
- * for GPU/CPU synchronization.  When the fence is written,
- * it is expected that all buffers associated with that fence
- * are no longer in use by the associated ring on the GPU and
- * that the relevant GPU caches have been flushed.
- */
-
-struct amdgpu_fence {
-	struct dma_fence base;
-
-	/* RB, DMA, etc. */
-	struct amdgpu_ring		*ring;
-	ktime_t				start_timestamp;
-};
-
-static struct kmem_cache *amdgpu_fence_slab;
-
-int amdgpu_fence_slab_init(void)
-{
-	amdgpu_fence_slab = KMEM_CACHE(amdgpu_fence, SLAB_HWCACHE_ALIGN);
-	if (!amdgpu_fence_slab)
-		return -ENOMEM;
-	return 0;
-}
-
-void amdgpu_fence_slab_fini(void)
-{
-	rcu_barrier();
-	kmem_cache_destroy(amdgpu_fence_slab);
-}
-/*
  * Cast helper
  */
 static const struct dma_fence_ops amdgpu_fence_ops;
-static const struct dma_fence_ops amdgpu_job_fence_ops;
 static inline struct amdgpu_fence *to_amdgpu_fence(struct dma_fence *f)
 {
 	struct amdgpu_fence *__f = container_of(f, struct amdgpu_fence, base);
 
-	if (__f->base.ops == &amdgpu_fence_ops ||
-	    __f->base.ops == &amdgpu_job_fence_ops)
-		return __f;
-
-	return NULL;
+	return __f;
 }
 
 /**
@@ -129,57 +93,32 @@ static u32 amdgpu_fence_read(struct amdgpu_ring *ring)
  * amdgpu_fence_emit - emit a fence on the requested ring
  *
  * @ring: ring the fence is associated with
- * @f: resulting fence object
- * @job: job the fence is embedded in
+ * @af: amdgpu fence input
  * @flags: flags to pass into the subordinate .emit_fence() call
  *
  * Emits a fence command on the requested ring (all asics).
  * Returns 0 on success, -ENOMEM on failure.
  */
-int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, struct amdgpu_job *job,
+int amdgpu_fence_emit(struct amdgpu_ring *ring, struct amdgpu_fence *af,
 		      unsigned int flags)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct dma_fence *fence;
-	struct amdgpu_fence *am_fence;
 	struct dma_fence __rcu **ptr;
 	uint32_t seq;
 	int r;
 
-	if (job == NULL) {
-		/* create a sperate hw fence */
-		am_fence = kmem_cache_alloc(amdgpu_fence_slab, GFP_ATOMIC);
-		if (am_fence == NULL)
-			return -ENOMEM;
-		fence = &am_fence->base;
-		am_fence->ring = ring;
-	} else {
-		/* take use of job-embedded fence */
-		fence = &job->hw_fence;
-	}
+	fence = &af->base;
+	af->ring = ring;
 
 	seq = ++ring->fence_drv.sync_seq;
-	if (job && job->job_run_counter) {
-		/* reinit seq for resubmitted jobs */
-		fence->seqno = seq;
-		/* TO be inline with external fence creation and other drivers */
-		dma_fence_get(fence);
-	} else {
-		if (job) {
-			dma_fence_init(fence, &amdgpu_job_fence_ops,
-				       &ring->fence_drv.lock,
-				       adev->fence_context + ring->idx, seq);
-			/* Against remove in amdgpu_job_{free, free_cb} */
-			dma_fence_get(fence);
-		} else {
-			dma_fence_init(fence, &amdgpu_fence_ops,
-				       &ring->fence_drv.lock,
-				       adev->fence_context + ring->idx, seq);
-		}
-	}
+	dma_fence_init(fence, &amdgpu_fence_ops,
+		       &ring->fence_drv.lock,
+		       adev->fence_context + ring->idx, seq);
 
 	amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr,
 			       seq, flags | AMDGPU_FENCE_FLAG_INT);
+	amdgpu_fence_save_wptr(af);
 	pm_runtime_get_noresume(adev_to_drm(adev)->dev);
 	ptr = &ring->fence_drv.fences[seq & ring->fence_drv.num_fences_mask];
 	if (unlikely(rcu_dereference_protected(*ptr, 1))) {
@@ -204,8 +143,6 @@ int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, struct amd
 	 */
 	rcu_assign_pointer(*ptr, dma_fence_get(fence));
 
-	*f = fence;
-
 	return 0;
 }
 
@@ -280,7 +217,7 @@ bool amdgpu_fence_process(struct amdgpu_ring *ring)
 
 	} while (atomic_cmpxchg(&drv->last_seq, last_seq, seq) != last_seq);
 
-	if (del_timer(&ring->fence_drv.fallback_timer) &&
+	if (timer_delete(&ring->fence_drv.fallback_timer) &&
 	    seq != ring->fence_drv.sync_seq)
 		amdgpu_fence_schedule_fallback(ring);
 
@@ -292,6 +229,7 @@ bool amdgpu_fence_process(struct amdgpu_ring *ring)
 
 	do {
 		struct dma_fence *fence, **ptr;
+		struct amdgpu_fence *am_fence;
 
 		++last_seq;
 		last_seq &= drv->num_fences_mask;
@@ -304,9 +242,14 @@ bool amdgpu_fence_process(struct amdgpu_ring *ring)
 		if (!fence)
 			continue;
 
+		/* Save the wptr in the fence driver so we know what the last processed
+		 * wptr was.  This is required for re-emitting the ring state for
+		 * queues that are reset but are not guilty and thus have no guilty fence.
+		 */
+		am_fence = container_of(fence, struct amdgpu_fence, base);
+		drv->signalled_wptr = am_fence->wptr;
 		dma_fence_signal(fence);
 		dma_fence_put(fence);
-		pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
 		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	} while (last_seq != seq);
 
@@ -322,11 +265,13 @@ bool amdgpu_fence_process(struct amdgpu_ring *ring)
  */
 static void amdgpu_fence_fallback(struct timer_list *t)
 {
-	struct amdgpu_ring *ring = from_timer(ring, t,
-					      fence_drv.fallback_timer);
+	struct amdgpu_ring *ring = timer_container_of(ring, t,
+						      fence_drv.fallback_timer);
 
 	if (amdgpu_fence_process(ring))
-		DRM_WARN("Fence fallback timer expired on ring %s\n", ring->name);
+		dev_warn(ring->adev->dev,
+			 "Fence fallback timer expired on ring %s\n",
+			 ring->name);
 }
 
 /**
@@ -618,7 +563,7 @@ void amdgpu_fence_driver_hw_fini(struct amdgpu_device *adev)
 			amdgpu_irq_put(adev, ring->fence_drv.irq_src,
 				       ring->fence_drv.irq_type);
 
-		del_timer_sync(&ring->fence_drv.fallback_timer);
+		timer_delete_sync(&ring->fence_drv.fallback_timer);
 	}
 }
 
@@ -698,36 +643,6 @@ void amdgpu_fence_driver_hw_init(struct amdgpu_device *adev)
 }
 
 /**
- * amdgpu_fence_driver_clear_job_fences - clear job embedded fences of ring
- *
- * @ring: fence of the ring to be cleared
- *
- */
-void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring)
-{
-	int i;
-	struct dma_fence *old, **ptr;
-
-	for (i = 0; i <= ring->fence_drv.num_fences_mask; i++) {
-		ptr = &ring->fence_drv.fences[i];
-		old = rcu_dereference_protected(*ptr, 1);
-		if (old && old->ops == &amdgpu_job_fence_ops) {
-			struct amdgpu_job *job;
-
-			/* For non-scheduler bad job, i.e. failed ib test, we need to signal
-			 * it right here or we won't be able to track them in fence_drv
-			 * and they will remain unsignaled during sa_bo free.
-			 */
-			job = container_of(old, struct amdgpu_job, hw_fence);
-			if (!job->base.s_fence && !dma_fence_is_signaled(old))
-				dma_fence_signal(old);
-			RCU_INIT_POINTER(*ptr, NULL);
-			dma_fence_put(old);
-		}
-	}
-}
-
-/**
  * amdgpu_fence_driver_set_error - set error code on fences
  * @ring: the ring which contains the fences
  * @error: the error code to set
@@ -764,6 +679,120 @@ void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring)
 	amdgpu_fence_process(ring);
 }
 
+
+/*
+ * Kernel queue reset handling
+ *
+ * The driver can reset individual queues for most engines, but those queues
+ * may contain work from multiple contexts.  Resetting the queue will reset
+ * lose all of that state.  In order to minimize the collateral damage, the
+ * driver will save the ring contents which are not associated with the guilty
+ * context prior to resetting the queue.  After resetting the queue the queue
+ * contents from the other contexts is re-emitted to the rings so that it can
+ * be processed by the engine.  To handle this, we save the queue's write
+ * pointer (wptr) in the fences associated with each context.  If we get a
+ * queue timeout, we can then use the wptrs from the fences to determine
+ * which data needs to be saved out of the queue's ring buffer.
+ */
+
+/**
+ * amdgpu_fence_driver_guilty_force_completion - force signal of specified sequence
+ *
+ * @af: fence of the ring to signal
+ *
+ */
+void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_fence *af)
+{
+	struct dma_fence *unprocessed;
+	struct dma_fence __rcu **ptr;
+	struct amdgpu_fence *fence;
+	struct amdgpu_ring *ring = af->ring;
+	unsigned long flags;
+	u32 seq, last_seq;
+
+	last_seq = amdgpu_fence_read(ring) & ring->fence_drv.num_fences_mask;
+	seq = ring->fence_drv.sync_seq & ring->fence_drv.num_fences_mask;
+
+	/* mark all fences from the guilty context with an error */
+	spin_lock_irqsave(&ring->fence_drv.lock, flags);
+	do {
+		last_seq++;
+		last_seq &= ring->fence_drv.num_fences_mask;
+
+		ptr = &ring->fence_drv.fences[last_seq];
+		rcu_read_lock();
+		unprocessed = rcu_dereference(*ptr);
+
+		if (unprocessed && !dma_fence_is_signaled_locked(unprocessed)) {
+			fence = container_of(unprocessed, struct amdgpu_fence, base);
+
+			if (fence == af)
+				dma_fence_set_error(&fence->base, -ETIME);
+			else if (fence->context == af->context)
+				dma_fence_set_error(&fence->base, -ECANCELED);
+		}
+		rcu_read_unlock();
+	} while (last_seq != seq);
+	spin_unlock_irqrestore(&ring->fence_drv.lock, flags);
+	/* signal the guilty fence */
+	amdgpu_fence_write(ring, (u32)af->base.seqno);
+	amdgpu_fence_process(ring);
+}
+
+void amdgpu_fence_save_wptr(struct amdgpu_fence *af)
+{
+	af->wptr = af->ring->wptr;
+}
+
+static void amdgpu_ring_backup_unprocessed_command(struct amdgpu_ring *ring,
+						   u64 start_wptr, u32 end_wptr)
+{
+	unsigned int first_idx = start_wptr & ring->buf_mask;
+	unsigned int last_idx = end_wptr & ring->buf_mask;
+	unsigned int i;
+
+	/* Backup the contents of the ring buffer. */
+	for (i = first_idx; i != last_idx; ++i, i &= ring->buf_mask)
+		ring->ring_backup[ring->ring_backup_entries_to_copy++] = ring->ring[i];
+}
+
+void amdgpu_ring_backup_unprocessed_commands(struct amdgpu_ring *ring,
+					     struct amdgpu_fence *guilty_fence)
+{
+	struct dma_fence *unprocessed;
+	struct dma_fence __rcu **ptr;
+	struct amdgpu_fence *fence;
+	u64 wptr;
+	u32 seq, last_seq;
+
+	last_seq = amdgpu_fence_read(ring) & ring->fence_drv.num_fences_mask;
+	seq = ring->fence_drv.sync_seq & ring->fence_drv.num_fences_mask;
+	wptr = ring->fence_drv.signalled_wptr;
+	ring->ring_backup_entries_to_copy = 0;
+
+	do {
+		last_seq++;
+		last_seq &= ring->fence_drv.num_fences_mask;
+
+		ptr = &ring->fence_drv.fences[last_seq];
+		rcu_read_lock();
+		unprocessed = rcu_dereference(*ptr);
+
+		if (unprocessed && !dma_fence_is_signaled(unprocessed)) {
+			fence = container_of(unprocessed, struct amdgpu_fence, base);
+
+			/* save everything if the ring is not guilty, otherwise
+			 * just save the content from other contexts.
+			 */
+			if (!guilty_fence || (fence->context != guilty_fence->context))
+				amdgpu_ring_backup_unprocessed_command(ring, wptr,
+								       fence->wptr);
+			wptr = fence->wptr;
+		}
+		rcu_read_unlock();
+	} while (last_seq != seq);
+}
+
 /*
  * Common fence implementation
  */
@@ -778,13 +807,6 @@ static const char *amdgpu_fence_get_timeline_name(struct dma_fence *f)
 	return (const char *)to_amdgpu_fence(f)->ring->name;
 }
 
-static const char *amdgpu_job_fence_get_timeline_name(struct dma_fence *f)
-{
-	struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence);
-
-	return (const char *)to_amdgpu_ring(job->base.sched)->name;
-}
-
 /**
  * amdgpu_fence_enable_signaling - enable signalling on fence
  * @f: fence
@@ -802,23 +824,6 @@ static bool amdgpu_fence_enable_signaling(struct dma_fence *f)
 }
 
 /**
- * amdgpu_job_fence_enable_signaling - enable signalling on job fence
- * @f: fence
- *
- * This is the simliar function with amdgpu_fence_enable_signaling above, it
- * only handles the job embedded fence.
- */
-static bool amdgpu_job_fence_enable_signaling(struct dma_fence *f)
-{
-	struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence);
-
-	if (!timer_pending(&to_amdgpu_ring(job->base.sched)->fence_drv.fallback_timer))
-		amdgpu_fence_schedule_fallback(to_amdgpu_ring(job->base.sched));
-
-	return true;
-}
-
-/**
  * amdgpu_fence_free - free up the fence memory
  *
  * @rcu: RCU callback head
@@ -830,22 +835,7 @@ static void amdgpu_fence_free(struct rcu_head *rcu)
 	struct dma_fence *f = container_of(rcu, struct dma_fence, rcu);
 
 	/* free fence_slab if it's separated fence*/
-	kmem_cache_free(amdgpu_fence_slab, to_amdgpu_fence(f));
-}
-
-/**
- * amdgpu_job_fence_free - free up the job with embedded fence
- *
- * @rcu: RCU callback head
- *
- * Free up the job with embedded fence after the RCU grace period.
- */
-static void amdgpu_job_fence_free(struct rcu_head *rcu)
-{
-	struct dma_fence *f = container_of(rcu, struct dma_fence, rcu);
-
-	/* free job if fence has a parent job */
-	kfree(container_of(f, struct amdgpu_job, hw_fence));
+	kfree(to_amdgpu_fence(f));
 }
 
 /**
@@ -861,19 +851,6 @@ static void amdgpu_fence_release(struct dma_fence *f)
 	call_rcu(&f->rcu, amdgpu_fence_free);
 }
 
-/**
- * amdgpu_job_fence_release - callback that job embedded fence can be freed
- *
- * @f: fence
- *
- * This is the simliar function with amdgpu_fence_release above, it
- * only handles the job embedded fence.
- */
-static void amdgpu_job_fence_release(struct dma_fence *f)
-{
-	call_rcu(&f->rcu, amdgpu_job_fence_free);
-}
-
 static const struct dma_fence_ops amdgpu_fence_ops = {
 	.get_driver_name = amdgpu_fence_get_driver_name,
 	.get_timeline_name = amdgpu_fence_get_timeline_name,
@@ -881,13 +858,6 @@ static const struct dma_fence_ops amdgpu_fence_ops = {
 	.release = amdgpu_fence_release,
 };
 
-static const struct dma_fence_ops amdgpu_job_fence_ops = {
-	.get_driver_name = amdgpu_fence_get_driver_name,
-	.get_timeline_name = amdgpu_job_fence_get_timeline_name,
-	.enable_signaling = amdgpu_job_fence_enable_signaling,
-	.release = amdgpu_job_fence_release,
-};
-
 /*
  * Fence debugfs
  */
@@ -957,7 +927,6 @@ static int gpu_recover_get(void *data, u64 *val)
 
 	*val = atomic_read(&adev->reset_domain->reset_res);
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c
index 09c9194d5bd5..b0082aa7f3c6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c
@@ -31,6 +31,7 @@
 
 #define FRU_EEPROM_MADDR_6      0x60000
 #define FRU_EEPROM_MADDR_8      0x80000
+#define FRU_EEPROM_MADDR_INV    0xFFFFF
 
 static bool is_fru_eeprom_supported(struct amdgpu_device *adev, u32 *fru_addr)
 {
@@ -63,10 +64,10 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev, u32 *fru_addr)
 		switch (adev->asic_type) {
 		case CHIP_VEGA20:
 			/* D161 and D163 are the VG20 server SKUs */
-			if (strnstr(atom_ctx->vbios_pn, "D161",
-				    sizeof(atom_ctx->vbios_pn)) ||
-			    strnstr(atom_ctx->vbios_pn, "D163",
-				    sizeof(atom_ctx->vbios_pn))) {
+			if (atom_ctx && (strnstr(atom_ctx->vbios_pn, "D161",
+						 sizeof(atom_ctx->vbios_pn)) ||
+					 strnstr(atom_ctx->vbios_pn, "D163",
+						 sizeof(atom_ctx->vbios_pn)))) {
 				if (fru_addr)
 					*fru_addr = FRU_EEPROM_MADDR_6;
 				return true;
@@ -78,8 +79,8 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev, u32 *fru_addr)
 			return false;
 		}
 	case IP_VERSION(11, 0, 7):
-		if (strnstr(atom_ctx->vbios_pn, "D603",
-			    sizeof(atom_ctx->vbios_pn))) {
+		if (atom_ctx && strnstr(atom_ctx->vbios_pn, "D603",
+					sizeof(atom_ctx->vbios_pn))) {
 			if (strnstr(atom_ctx->vbios_pn, "D603GLXE",
 				    sizeof(atom_ctx->vbios_pn))) {
 				return false;
@@ -94,8 +95,8 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev, u32 *fru_addr)
 		}
 	case IP_VERSION(13, 0, 2):
 		/* All Aldebaran SKUs have an FRU */
-		if (!strnstr(atom_ctx->vbios_pn, "D673",
-			     sizeof(atom_ctx->vbios_pn)))
+		if (atom_ctx && !strnstr(atom_ctx->vbios_pn, "D673",
+					 sizeof(atom_ctx->vbios_pn)))
 			if (fru_addr)
 				*fru_addr = FRU_EEPROM_MADDR_6;
 		return true;
@@ -104,6 +105,10 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev, u32 *fru_addr)
 			if (fru_addr)
 				*fru_addr = FRU_EEPROM_MADDR_8;
 			return true;
+	case IP_VERSION(13, 0, 12):
+			if (fru_addr)
+				*fru_addr = FRU_EEPROM_MADDR_INV;
+			return true;
 	default:
 		return false;
 	}
@@ -120,6 +125,10 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev)
 	if (!is_fru_eeprom_supported(adev, &fru_addr))
 		return 0;
 
+	/* FRU data avaialble, but no direct EEPROM access */
+	if (fru_addr == FRU_EEPROM_MADDR_INV)
+		return 0;
+
 	if (!adev->fru_info) {
 		adev->fru_info = kzalloc(sizeof(*adev->fru_info), GFP_KERNEL);
 		if (!adev->fru_info)
@@ -135,7 +144,8 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev)
 
 	/* If algo exists, it means that the i2c_adapter's initialized */
 	if (!adev->pm.fru_eeprom_i2c_bus || !adev->pm.fru_eeprom_i2c_bus->algo) {
-		DRM_WARN("Cannot access FRU, EEPROM accessor not initialized");
+		dev_warn(adev->dev,
+			 "Cannot access FRU, EEPROM accessor not initialized");
 		return -ENODEV;
 	}
 
@@ -143,19 +153,22 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev)
 	len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, fru_addr, buf,
 				 sizeof(buf));
 	if (len != 8) {
-		DRM_ERROR("Couldn't read the IPMI Common Header: %d", len);
+		dev_err(adev->dev, "Couldn't read the IPMI Common Header: %d",
+			len);
 		return len < 0 ? len : -EIO;
 	}
 
 	if (buf[0] != 1) {
-		DRM_ERROR("Bad IPMI Common Header version: 0x%02x", buf[0]);
+		dev_err(adev->dev, "Bad IPMI Common Header version: 0x%02x",
+			buf[0]);
 		return -EIO;
 	}
 
 	for (csum = 0; len > 0; len--)
 		csum += buf[len - 1];
 	if (csum) {
-		DRM_ERROR("Bad IPMI Common Header checksum: 0x%02x", csum);
+		dev_err(adev->dev, "Bad IPMI Common Header checksum: 0x%02x",
+			csum);
 		return -EIO;
 	}
 
@@ -170,12 +183,14 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev)
 	/* Read the header of the PIA. */
 	len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addr, buf, 3);
 	if (len != 3) {
-		DRM_ERROR("Couldn't read the Product Info Area header: %d", len);
+		dev_err(adev->dev,
+			"Couldn't read the Product Info Area header: %d", len);
 		return len < 0 ? len : -EIO;
 	}
 
 	if (buf[0] != 1) {
-		DRM_ERROR("Bad IPMI Product Info Area version: 0x%02x", buf[0]);
+		dev_err(adev->dev, "Bad IPMI Product Info Area version: 0x%02x",
+			buf[0]);
 		return -EIO;
 	}
 
@@ -188,14 +203,16 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev)
 	len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addr, pia, size);
 	if (len != size) {
 		kfree(pia);
-		DRM_ERROR("Couldn't read the Product Info Area: %d", len);
+		dev_err(adev->dev, "Couldn't read the Product Info Area: %d",
+			len);
 		return len < 0 ? len : -EIO;
 	}
 
 	for (csum = 0; size > 0; size--)
 		csum += pia[size - 1];
 	if (csum) {
-		DRM_ERROR("Bad Product Info Area checksum: 0x%02x", csum);
+		dev_err(adev->dev, "Bad Product Info Area checksum: 0x%02x",
+			csum);
 		kfree(pia);
 		return -EIO;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
index b2033f8352f5..d2237ce9da70 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
@@ -302,7 +302,6 @@ void amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
 			int pages)
 {
 	unsigned t;
-	unsigned p;
 	int i, j;
 	u64 page_base;
 	/* Starting from VEGA10, system bit must be 0 to mean invalid. */
@@ -316,8 +315,7 @@ void amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
 		return;
 
 	t = offset / AMDGPU_GPU_PAGE_SIZE;
-	p = t / AMDGPU_GPU_PAGES_IN_CPU_PAGE;
-	for (i = 0; i < pages; i++, p++) {
+	for (i = 0; i < pages; i++) {
 		page_base = adev->dummy_page_addr;
 		if (!adev->gart.ptr)
 			continue;
@@ -370,6 +368,42 @@ void amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
 }
 
 /**
+ * amdgpu_gart_map_vram_range - map VRAM pages into the GART page table
+ *
+ * @adev: amdgpu_device pointer
+ * @pa: physical address of the first page to be mapped
+ * @start_page: first page to map in the GART aperture
+ * @num_pages: number of pages to be mapped
+ * @flags: page table entry flags
+ * @dst: CPU address of the GART table
+ *
+ * Binds a BO that is allocated in VRAM to the GART page table
+ * (all ASICs).
+ *
+ * Useful when a kernel BO is located in VRAM but
+ * needs to be accessed from the GART address space.
+ */
+void amdgpu_gart_map_vram_range(struct amdgpu_device *adev, uint64_t pa,
+				uint64_t start_page, uint64_t num_pages,
+				uint64_t flags, void *dst)
+{
+	u32 i, idx;
+
+	/* The SYSTEM flag indicates the pages aren't in VRAM. */
+	WARN_ON_ONCE(flags & AMDGPU_PTE_SYSTEM);
+
+	if (!drm_dev_enter(adev_to_drm(adev), &idx))
+		return;
+
+	for (i = 0; i < num_pages; ++i) {
+		amdgpu_gmc_set_pte_pde(adev, adev->gart.ptr,
+			start_page + i, pa + AMDGPU_GPU_PAGE_SIZE * i, flags);
+	}
+
+	drm_dev_exit(idx);
+}
+
+/**
  * amdgpu_gart_bind - bind pages into the gart page table
  *
  * @adev: amdgpu_device pointer
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h
index 7cc980bf4725..d3118275ddae 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h
@@ -64,5 +64,8 @@ void amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
 		     void *dst);
 void amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset,
 		      int pages, dma_addr_t *dma_addr, uint64_t flags);
+void amdgpu_gart_map_vram_range(struct amdgpu_device *adev, uint64_t pa,
+				uint64_t start_page, uint64_t num_pages,
+				uint64_t flags, void *dst);
 void amdgpu_gart_invalidate_tlb(struct amdgpu_device *adev);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 69429df09477..3e38c5db2987 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -36,6 +36,7 @@
 #include <drm/drm_exec.h>
 #include <drm/drm_gem_ttm_helper.h>
 #include <drm/ttm/ttm_tt.h>
+#include <drm/drm_syncobj.h>
 
 #include "amdgpu.h"
 #include "amdgpu_display.h"
@@ -44,6 +45,114 @@
 #include "amdgpu_xgmi.h"
 #include "amdgpu_vm.h"
 
+static int
+amdgpu_gem_add_input_fence(struct drm_file *filp,
+			   uint64_t syncobj_handles_array,
+			   uint32_t num_syncobj_handles)
+{
+	struct dma_fence *fence;
+	uint32_t *syncobj_handles;
+	int ret, i;
+
+	if (!num_syncobj_handles)
+		return 0;
+
+	syncobj_handles = memdup_user(u64_to_user_ptr(syncobj_handles_array),
+				      size_mul(sizeof(uint32_t), num_syncobj_handles));
+	if (IS_ERR(syncobj_handles))
+		return PTR_ERR(syncobj_handles);
+
+	for (i = 0; i < num_syncobj_handles; i++) {
+
+		if (!syncobj_handles[i]) {
+			ret = -EINVAL;
+			goto free_memdup;
+		}
+
+		ret = drm_syncobj_find_fence(filp, syncobj_handles[i], 0, 0, &fence);
+		if (ret)
+			goto free_memdup;
+
+		dma_fence_wait(fence, false);
+
+		/* TODO: optimize async handling */
+		dma_fence_put(fence);
+	}
+
+free_memdup:
+	kfree(syncobj_handles);
+	return ret;
+}
+
+static int
+amdgpu_gem_update_timeline_node(struct drm_file *filp,
+				uint32_t syncobj_handle,
+				uint64_t point,
+				struct drm_syncobj **syncobj,
+				struct dma_fence_chain **chain)
+{
+	if (!syncobj_handle)
+		return 0;
+
+	/* Find the sync object */
+	*syncobj = drm_syncobj_find(filp, syncobj_handle);
+	if (!*syncobj)
+		return -ENOENT;
+
+	if (!point)
+		return 0;
+
+	/* Allocate the chain node */
+	*chain = dma_fence_chain_alloc();
+	if (!*chain) {
+		drm_syncobj_put(*syncobj);
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+static void
+amdgpu_gem_update_bo_mapping(struct drm_file *filp,
+			     struct amdgpu_bo_va *bo_va,
+			     uint32_t operation,
+			     uint64_t point,
+			     struct dma_fence *fence,
+			     struct drm_syncobj *syncobj,
+			     struct dma_fence_chain *chain)
+{
+	struct amdgpu_bo *bo = bo_va ? bo_va->base.bo : NULL;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct dma_fence *last_update;
+
+	if (!syncobj)
+		return;
+
+	/* Find the last update fence */
+	switch (operation) {
+	case AMDGPU_VA_OP_MAP:
+	case AMDGPU_VA_OP_REPLACE:
+		if (bo && (bo->tbo.base.resv == vm->root.bo->tbo.base.resv))
+			last_update = vm->last_update;
+		else
+			last_update = bo_va->last_pt_update;
+		break;
+	case AMDGPU_VA_OP_UNMAP:
+	case AMDGPU_VA_OP_CLEAR:
+		last_update = fence;
+		break;
+	default:
+		return;
+	}
+
+	/* Add fence to timeline */
+	if (!point)
+		drm_syncobj_replace_fence(syncobj, last_update);
+	else
+		drm_syncobj_add_point(syncobj, chain, last_update, point);
+}
+
 static vm_fault_t amdgpu_gem_fault(struct vm_fault *vmf)
 {
 	struct ttm_buffer_object *bo = vmf->vma->vm_private_data;
@@ -89,7 +198,7 @@ static void amdgpu_gem_object_free(struct drm_gem_object *gobj)
 	struct amdgpu_bo *aobj = gem_to_amdgpu_bo(gobj);
 
 	amdgpu_hmm_unregister(aobj);
-	ttm_bo_put(&aobj->tbo);
+	ttm_bo_fini(&aobj->tbo);
 }
 
 int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
@@ -184,6 +293,15 @@ static int amdgpu_gem_object_open(struct drm_gem_object *obj,
 		bo_va = amdgpu_vm_bo_add(adev, vm, abo);
 	else
 		++bo_va->ref_count;
+
+	/* attach gfx eviction fence */
+	r = amdgpu_eviction_fence_attach(&fpriv->evf_mgr, abo);
+	if (r) {
+		DRM_DEBUG_DRIVER("Failed to attach eviction fence to BO\n");
+		amdgpu_bo_unreserve(abo);
+		return r;
+	}
+
 	amdgpu_bo_unreserve(abo);
 
 	/* Validate and add eviction fence to DMABuf imports with dynamic
@@ -199,7 +317,7 @@ static int amdgpu_gem_object_open(struct drm_gem_object *obj,
 	 */
 	if (!vm->is_compute_context || !vm->process_info)
 		return 0;
-	if (!obj->import_attach ||
+	if (!drm_gem_is_imported(obj) ||
 	    !dma_buf_is_dynamic(obj->import_attach->dmabuf))
 		return 0;
 	mutex_lock_nested(&vm->process_info->lock, 1);
@@ -211,7 +329,7 @@ static int amdgpu_gem_object_open(struct drm_gem_object *obj,
 
 			dev_warn(adev->dev, "validate_and_fence failed: %d\n", r);
 			if (ti) {
-				dev_warn(adev->dev, "pid %d\n", ti->pid);
+				dev_warn(adev->dev, "pid %d\n", ti->task.pid);
 				amdgpu_vm_put_task_info(ti);
 			}
 		}
@@ -247,6 +365,9 @@ static void amdgpu_gem_object_close(struct drm_gem_object *obj,
 			goto out_unlock;
 	}
 
+	if (!amdgpu_vm_is_bo_always_valid(vm, bo))
+		amdgpu_eviction_fence_detach(&fpriv->evf_mgr, bo);
+
 	bo_va = amdgpu_vm_bo_find(vm, bo);
 	if (!bo_va || --bo_va->ref_count)
 		goto out_unlock;
@@ -321,20 +442,8 @@ int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
 	uint32_t handle, initial_domain;
 	int r;
 
-	/* reject DOORBELLs until userspace code to use it is available */
-	if (args->in.domains & AMDGPU_GEM_DOMAIN_DOORBELL)
-		return -EINVAL;
-
 	/* reject invalid gem flags */
-	if (flags & ~(AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-		      AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
-		      AMDGPU_GEM_CREATE_CPU_GTT_USWC |
-		      AMDGPU_GEM_CREATE_VRAM_CLEARED |
-		      AMDGPU_GEM_CREATE_VM_ALWAYS_VALID |
-		      AMDGPU_GEM_CREATE_EXPLICIT_SYNC |
-		      AMDGPU_GEM_CREATE_ENCRYPTED |
-		      AMDGPU_GEM_CREATE_GFX12_DCC |
-		      AMDGPU_GEM_CREATE_DISCARDABLE))
+	if (flags & ~AMDGPU_GEM_CREATE_SETTABLE_MASK)
 		return -EINVAL;
 
 	/* reject invalid gem domains */
@@ -349,6 +458,9 @@ int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
 	/* always clear VRAM */
 	flags |= AMDGPU_GEM_CREATE_VRAM_CLEARED;
 
+	if (args->in.domains & AMDGPU_GEM_DOMAIN_MMIO_REMAP)
+		return -EINVAL;
+
 	/* create a gem object to contain this object in */
 	if (args->in.domains & (AMDGPU_GEM_DOMAIN_GDS |
 	    AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA)) {
@@ -419,7 +531,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 	struct drm_amdgpu_gem_userptr *args = data;
 	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	struct drm_gem_object *gobj;
-	struct hmm_range *range;
+	struct amdgpu_hmm_range *range;
 	struct amdgpu_bo *bo;
 	uint32_t handle;
 	int r;
@@ -460,15 +572,20 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 		goto release_object;
 
 	if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE) {
-		r = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages,
-						 &range);
-		if (r)
+		range = amdgpu_hmm_range_alloc(NULL);
+		if (unlikely(!range))
+			return -ENOMEM;
+		r = amdgpu_ttm_tt_get_user_pages(bo, range);
+		if (r) {
+			amdgpu_hmm_range_free(range);
 			goto release_object;
-
+		}
 		r = amdgpu_bo_reserve(bo, true);
 		if (r)
 			goto user_pages_done;
 
+		amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, range);
+
 		amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
 		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 		amdgpu_bo_unreserve(bo);
@@ -484,8 +601,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 
 user_pages_done:
 	if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE)
-		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, range);
-
+		amdgpu_hmm_range_free(range);
 release_object:
 	drm_gem_object_put(gobj);
 
@@ -638,18 +754,23 @@ out:
  *
  * Update the bo_va directly after setting its address. Errors are not
  * vital here, so they are not reported back to userspace.
+ *
+ * Returns resulting fence if freed BO(s) got cleared from the PT.
+ * otherwise stub fence in case of error.
  */
-static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
-				    struct amdgpu_vm *vm,
-				    struct amdgpu_bo_va *bo_va,
-				    uint32_t operation)
+static struct dma_fence *
+amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
+			struct amdgpu_vm *vm,
+			struct amdgpu_bo_va *bo_va,
+			uint32_t operation)
 {
+	struct dma_fence *fence = dma_fence_get_stub();
 	int r;
 
 	if (!amdgpu_vm_ready(vm))
-		return;
+		return fence;
 
-	r = amdgpu_vm_clear_freed(adev, vm, NULL);
+	r = amdgpu_vm_clear_freed(adev, vm, &fence);
 	if (r)
 		goto error;
 
@@ -665,36 +786,8 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
 error:
 	if (r && r != -ERESTARTSYS)
 		DRM_ERROR("Couldn't update BO_VA (%d)\n", r);
-}
 
-/**
- * amdgpu_gem_va_map_flags - map GEM UAPI flags into hardware flags
- *
- * @adev: amdgpu_device pointer
- * @flags: GEM UAPI flags
- *
- * Returns the GEM UAPI flags mapped into hardware for the ASIC.
- */
-uint64_t amdgpu_gem_va_map_flags(struct amdgpu_device *adev, uint32_t flags)
-{
-	uint64_t pte_flag = 0;
-
-	if (flags & AMDGPU_VM_PAGE_EXECUTABLE)
-		pte_flag |= AMDGPU_PTE_EXECUTABLE;
-	if (flags & AMDGPU_VM_PAGE_READABLE)
-		pte_flag |= AMDGPU_PTE_READABLE;
-	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
-		pte_flag |= AMDGPU_PTE_WRITEABLE;
-	if (flags & AMDGPU_VM_PAGE_PRT)
-		pte_flag |= AMDGPU_PTE_PRT_FLAG(adev);
-	if (flags & AMDGPU_VM_PAGE_NOALLOC)
-		pte_flag |= AMDGPU_PTE_NOALLOC;
-
-	if (adev->gmc.gmc_funcs->map_mtype)
-		pte_flag |= amdgpu_gmc_map_mtype(adev,
-						 flags & AMDGPU_VM_MTYPE_MASK);
-
-	return pte_flag;
+	return fence;
 }
 
 int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
@@ -713,8 +806,10 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	struct amdgpu_bo *abo;
 	struct amdgpu_bo_va *bo_va;
+	struct drm_syncobj *timeline_syncobj = NULL;
+	struct dma_fence_chain *timeline_chain = NULL;
+	struct dma_fence *fence;
 	struct drm_exec exec;
-	uint64_t va_flags;
 	uint64_t vm_size;
 	int r = 0;
 
@@ -774,6 +869,12 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 		abo = NULL;
 	}
 
+	r = amdgpu_gem_add_input_fence(filp,
+				       args->input_fence_syncobj_handles,
+				       args->num_syncobj_handles);
+	if (r)
+		goto error_put_gobj;
+
 	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT |
 		      DRM_EXEC_IGNORE_DUPLICATES, 0);
 	drm_exec_until_all_locked(&exec) {
@@ -802,12 +903,19 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 		bo_va = NULL;
 	}
 
+	r = amdgpu_gem_update_timeline_node(filp,
+					    args->vm_timeline_syncobj_out,
+					    args->vm_timeline_point,
+					    &timeline_syncobj,
+					    &timeline_chain);
+	if (r)
+		goto error;
+
 	switch (args->operation) {
 	case AMDGPU_VA_OP_MAP:
-		va_flags = amdgpu_gem_va_map_flags(adev, args->flags);
 		r = amdgpu_vm_bo_map(adev, bo_va, args->va_address,
 				     args->offset_in_bo, args->map_size,
-				     va_flags);
+				     args->flags);
 		break;
 	case AMDGPU_VA_OP_UNMAP:
 		r = amdgpu_vm_bo_unmap(adev, bo_va, args->va_address);
@@ -819,20 +927,31 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 						args->map_size);
 		break;
 	case AMDGPU_VA_OP_REPLACE:
-		va_flags = amdgpu_gem_va_map_flags(adev, args->flags);
 		r = amdgpu_vm_bo_replace_map(adev, bo_va, args->va_address,
 					     args->offset_in_bo, args->map_size,
-					     va_flags);
+					     args->flags);
 		break;
 	default:
 		break;
 	}
-	if (!r && !(args->flags & AMDGPU_VM_DELAY_UPDATE) && !adev->debug_vm)
-		amdgpu_gem_va_update_vm(adev, &fpriv->vm, bo_va,
-					args->operation);
+	if (!r && !(args->flags & AMDGPU_VM_DELAY_UPDATE) && !adev->debug_vm) {
+		fence = amdgpu_gem_va_update_vm(adev, &fpriv->vm, bo_va,
+						args->operation);
+
+		if (timeline_syncobj)
+			amdgpu_gem_update_bo_mapping(filp, bo_va,
+					     args->operation,
+					     args->vm_timeline_point,
+					     fence, timeline_syncobj,
+					     timeline_chain);
+		else
+			dma_fence_put(fence);
+
+	}
 
 error:
 	drm_exec_fini(&exec);
+error_put_gobj:
 	drm_gem_object_put(gobj);
 	return r;
 }
@@ -844,17 +963,34 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 	struct drm_gem_object *gobj;
 	struct amdgpu_vm_bo_base *base;
 	struct amdgpu_bo *robj;
+	struct drm_exec exec;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	int r;
 
+	if (args->padding)
+		return -EINVAL;
+
 	gobj = drm_gem_object_lookup(filp, args->handle);
 	if (!gobj)
 		return -ENOENT;
 
 	robj = gem_to_amdgpu_bo(gobj);
 
-	r = amdgpu_bo_reserve(robj, false);
-	if (unlikely(r))
-		goto out;
+	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT |
+			  DRM_EXEC_IGNORE_DUPLICATES, 0);
+	drm_exec_until_all_locked(&exec) {
+		r = drm_exec_lock_obj(&exec, gobj);
+		drm_exec_retry_on_contention(&exec);
+		if (r)
+			goto out_exec;
+
+		if (args->op == AMDGPU_GEM_OP_GET_MAPPING_INFO) {
+			r = amdgpu_vm_lock_pd(&fpriv->vm, &exec, 0);
+			drm_exec_retry_on_contention(&exec);
+			if (r)
+				goto out_exec;
+		}
+	}
 
 	switch (args->op) {
 	case AMDGPU_GEM_OP_GET_GEM_CREATE_INFO: {
@@ -865,29 +1001,26 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 		info.alignment = robj->tbo.page_alignment << PAGE_SHIFT;
 		info.domains = robj->preferred_domains;
 		info.domain_flags = robj->flags;
-		amdgpu_bo_unreserve(robj);
+		drm_exec_fini(&exec);
 		if (copy_to_user(out, &info, sizeof(info)))
 			r = -EFAULT;
 		break;
 	}
 	case AMDGPU_GEM_OP_SET_PLACEMENT:
-		if (robj->tbo.base.import_attach &&
+		if (drm_gem_is_imported(&robj->tbo.base) &&
 		    args->value & AMDGPU_GEM_DOMAIN_VRAM) {
 			r = -EINVAL;
-			amdgpu_bo_unreserve(robj);
-			break;
+			goto out_exec;
 		}
 		if (amdgpu_ttm_tt_get_usermm(robj->tbo.ttm)) {
 			r = -EPERM;
-			amdgpu_bo_unreserve(robj);
-			break;
+			goto out_exec;
 		}
 		for (base = robj->vm_bo; base; base = base->next)
 			if (amdgpu_xgmi_same_hive(amdgpu_ttm_adev(robj->tbo.bdev),
 				amdgpu_ttm_adev(base->vm->root.bo->tbo.bdev))) {
 				r = -EINVAL;
-				amdgpu_bo_unreserve(robj);
-				goto out;
+				goto out_exec;
 			}
 
 
@@ -900,17 +1033,146 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 
 		if (robj->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID)
 			amdgpu_vm_bo_invalidate(robj, true);
+		drm_exec_fini(&exec);
+		break;
+	case AMDGPU_GEM_OP_GET_MAPPING_INFO: {
+		struct amdgpu_bo_va *bo_va = amdgpu_vm_bo_find(&fpriv->vm, robj);
+		struct drm_amdgpu_gem_vm_entry *vm_entries;
+		struct amdgpu_bo_va_mapping *mapping;
+		int num_mappings = 0;
+		/*
+		 * num_entries is set as an input to the size of the user-allocated array of
+		 * drm_amdgpu_gem_vm_entry stored at args->value.
+		 * num_entries is sent back as output as the number of mappings the bo has.
+		 * If that number is larger than the size of the array, the ioctl must
+		 * be retried.
+		 */
+		vm_entries = kvcalloc(args->num_entries, sizeof(*vm_entries), GFP_KERNEL);
+		if (!vm_entries)
+			return -ENOMEM;
+
+		amdgpu_vm_bo_va_for_each_valid_mapping(bo_va, mapping) {
+			if (num_mappings < args->num_entries) {
+				vm_entries[num_mappings].addr = mapping->start * AMDGPU_GPU_PAGE_SIZE;
+				vm_entries[num_mappings].size = (mapping->last - mapping->start + 1) * AMDGPU_GPU_PAGE_SIZE;
+				vm_entries[num_mappings].offset = mapping->offset;
+				vm_entries[num_mappings].flags = mapping->flags;
+			}
+			num_mappings += 1;
+		}
+
+		amdgpu_vm_bo_va_for_each_invalid_mapping(bo_va, mapping) {
+			if (num_mappings < args->num_entries) {
+				vm_entries[num_mappings].addr = mapping->start * AMDGPU_GPU_PAGE_SIZE;
+				vm_entries[num_mappings].size = (mapping->last - mapping->start + 1) * AMDGPU_GPU_PAGE_SIZE;
+				vm_entries[num_mappings].offset = mapping->offset;
+				vm_entries[num_mappings].flags = mapping->flags;
+			}
+			num_mappings += 1;
+		}
+
+		drm_exec_fini(&exec);
 
-		amdgpu_bo_unreserve(robj);
+		if (num_mappings > 0 && num_mappings <= args->num_entries)
+			if (copy_to_user(u64_to_user_ptr(args->value), vm_entries, num_mappings * sizeof(*vm_entries)))
+				r = -EFAULT;
+
+		args->num_entries = num_mappings;
+
+		kvfree(vm_entries);
 		break;
+	}
 	default:
-		amdgpu_bo_unreserve(robj);
+		drm_exec_fini(&exec);
 		r = -EINVAL;
 	}
 
-out:
 	drm_gem_object_put(gobj);
 	return r;
+out_exec:
+	drm_exec_fini(&exec);
+	drm_gem_object_put(gobj);
+	return r;
+}
+
+/**
+ * amdgpu_gem_list_handles_ioctl - get information about a process' buffer objects
+ *
+ * @dev: drm device pointer
+ * @data: drm_amdgpu_gem_list_handles
+ * @filp: drm file pointer
+ *
+ * num_entries is set as an input to the size of the entries array.
+ * num_entries is sent back as output as the number of bos in the process.
+ * If that number is larger than the size of the array, the ioctl must
+ * be retried.
+ *
+ * Returns:
+ * 0 for success, -errno for errors.
+ */
+int amdgpu_gem_list_handles_ioctl(struct drm_device *dev, void *data,
+				  struct drm_file *filp)
+{
+	struct drm_amdgpu_gem_list_handles *args = data;
+	struct drm_amdgpu_gem_list_handles_entry *bo_entries;
+	struct drm_gem_object *gobj;
+	int id, ret = 0;
+	int bo_index = 0;
+	int num_bos = 0;
+
+	spin_lock(&filp->table_lock);
+	idr_for_each_entry(&filp->object_idr, gobj, id)
+		num_bos += 1;
+	spin_unlock(&filp->table_lock);
+
+	if (args->num_entries < num_bos) {
+		args->num_entries = num_bos;
+		return 0;
+	}
+
+	if (num_bos == 0) {
+		args->num_entries = 0;
+		return 0;
+	}
+
+	bo_entries = kvcalloc(num_bos, sizeof(*bo_entries), GFP_KERNEL);
+	if (!bo_entries)
+		return -ENOMEM;
+
+	spin_lock(&filp->table_lock);
+	idr_for_each_entry(&filp->object_idr, gobj, id) {
+		struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj);
+		struct drm_amdgpu_gem_list_handles_entry *bo_entry;
+
+		if (bo_index >= num_bos) {
+			ret = -EAGAIN;
+			break;
+		}
+
+		bo_entry = &bo_entries[bo_index];
+
+		bo_entry->size = amdgpu_bo_size(bo);
+		bo_entry->alloc_flags = bo->flags & AMDGPU_GEM_CREATE_SETTABLE_MASK;
+		bo_entry->preferred_domains = bo->preferred_domains;
+		bo_entry->gem_handle = id;
+		bo_entry->alignment = bo->tbo.page_alignment;
+
+		if (bo->tbo.base.import_attach)
+			bo_entry->flags |= AMDGPU_GEM_LIST_HANDLES_FLAG_IS_IMPORT;
+
+		bo_index += 1;
+	}
+	spin_unlock(&filp->table_lock);
+
+	args->num_entries = bo_index;
+
+	if (!ret)
+		if (copy_to_user(u64_to_user_ptr(args->entries), bo_entries, num_bos * sizeof(*bo_entries)))
+			ret = -EFAULT;
+
+	kvfree(bo_entries);
+
+	return ret;
 }
 
 static int amdgpu_gem_align_pitch(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
index 3a8f57900a3a..b558336bc4c6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
@@ -63,13 +63,28 @@ int amdgpu_gem_mmap_ioctl(struct drm_device *dev, void *data,
 			  struct drm_file *filp);
 int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
 			      struct drm_file *filp);
-uint64_t amdgpu_gem_va_map_flags(struct amdgpu_device *adev, uint32_t flags);
 int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 			  struct drm_file *filp);
 int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 			struct drm_file *filp);
+int amdgpu_gem_list_handles_ioctl(struct drm_device *dev, void *data,
+				  struct drm_file *filp);
 
 int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
 				struct drm_file *filp);
 
+#define AMDGPU_GEM_CREATE_SETTABLE_MASK	(AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED | \
+	AMDGPU_GEM_CREATE_NO_CPU_ACCESS | \
+	AMDGPU_GEM_CREATE_CPU_GTT_USWC | \
+	AMDGPU_GEM_CREATE_VRAM_CLEARED | \
+	AMDGPU_GEM_CREATE_VM_ALWAYS_VALID | \
+	AMDGPU_GEM_CREATE_EXPLICIT_SYNC | \
+	AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE | \
+	AMDGPU_GEM_CREATE_ENCRYPTED | \
+	AMDGPU_GEM_CREATE_GFX12_DCC | \
+	AMDGPU_GEM_CREATE_DISCARDABLE | \
+	AMDGPU_GEM_CREATE_COHERENT | \
+	AMDGPU_GEM_CREATE_UNCACHED | \
+	AMDGPU_GEM_CREATE_EXT_COHERENT)
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 784b03abb3a4..8b118c53f351 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -33,6 +33,8 @@
 #include "amdgpu_reset.h"
 #include "amdgpu_xcp.h"
 #include "amdgpu_xgmi.h"
+#include "amdgpu_mes.h"
+#include "nvd.h"
 
 /* delay 0.1 second to enable gfx off feature */
 #define GFX_OFF_DELAY_ENABLE         msecs_to_jiffies(100)
@@ -74,14 +76,15 @@ bool amdgpu_gfx_is_mec_queue_enabled(struct amdgpu_device *adev,
 			adev->gfx.mec_bitmap[xcc_id].queue_bitmap);
 }
 
-int amdgpu_gfx_me_queue_to_bit(struct amdgpu_device *adev,
-			       int me, int pipe, int queue)
+static int amdgpu_gfx_me_queue_to_bit(struct amdgpu_device *adev,
+				      int me, int pipe, int queue)
 {
+	int num_queue_per_pipe = 1; /* we only enable 1 KGQ per pipe */
 	int bit = 0;
 
 	bit += me * adev->gfx.me.num_pipe_per_me
-		* adev->gfx.me.num_queue_per_pipe;
-	bit += pipe * adev->gfx.me.num_queue_per_pipe;
+		* num_queue_per_pipe;
+	bit += pipe * num_queue_per_pipe;
 	bit += queue;
 
 	return bit;
@@ -147,7 +150,7 @@ static bool amdgpu_gfx_is_graphics_multipipe_capable(struct amdgpu_device *adev)
 static bool amdgpu_gfx_is_compute_multipipe_capable(struct amdgpu_device *adev)
 {
 	if (amdgpu_compute_multipipe != -1) {
-		DRM_INFO("amdgpu: forcing compute pipe policy %d\n",
+		dev_info(adev->dev, "amdgpu: forcing compute pipe policy %d\n",
 			 amdgpu_compute_multipipe);
 		return amdgpu_compute_multipipe == 1;
 	}
@@ -238,8 +241,8 @@ void amdgpu_gfx_graphics_queue_acquire(struct amdgpu_device *adev)
 {
 	int i, queue, pipe;
 	bool multipipe_policy = amdgpu_gfx_is_graphics_multipipe_capable(adev);
-	int max_queues_per_me = adev->gfx.me.num_pipe_per_me *
-					adev->gfx.me.num_queue_per_pipe;
+	int num_queue_per_pipe = 1; /* we only enable 1 KGQ per pipe */
+	int max_queues_per_me = adev->gfx.me.num_pipe_per_me * num_queue_per_pipe;
 
 	if (multipipe_policy) {
 		/* policy: amdgpu owns the first queue per pipe at this stage
@@ -247,9 +250,9 @@ void amdgpu_gfx_graphics_queue_acquire(struct amdgpu_device *adev)
 		for (i = 0; i < max_queues_per_me; i++) {
 			pipe = i % adev->gfx.me.num_pipe_per_me;
 			queue = (i / adev->gfx.me.num_pipe_per_me) %
-				adev->gfx.me.num_queue_per_pipe;
+				num_queue_per_pipe;
 
-			set_bit(pipe * adev->gfx.me.num_queue_per_pipe + queue,
+			set_bit(pipe * num_queue_per_pipe + queue,
 				adev->gfx.me.queue_bitmap);
 		}
 	} else {
@@ -258,8 +261,9 @@ void amdgpu_gfx_graphics_queue_acquire(struct amdgpu_device *adev)
 	}
 
 	/* update the number of active graphics rings */
-	adev->gfx.num_gfx_rings =
-		bitmap_weight(adev->gfx.me.queue_bitmap, AMDGPU_MAX_GFX_QUEUES);
+	if (adev->gfx.num_gfx_rings)
+		adev->gfx.num_gfx_rings =
+			bitmap_weight(adev->gfx.me.queue_bitmap, AMDGPU_MAX_GFX_QUEUES);
 }
 
 static int amdgpu_gfx_kiq_acquire(struct amdgpu_device *adev,
@@ -671,7 +675,7 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id)
 		 * generation exposes more than 64 queues. If so, the
 		 * definition of queue_mask needs updating */
 		if (WARN_ON(i > (sizeof(queue_mask)*8))) {
-			DRM_ERROR("Invalid KCQ enabled: %d\n", i);
+			dev_err(adev->dev, "Invalid KCQ enabled: %d\n", i);
 			break;
 		}
 
@@ -680,15 +684,15 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id)
 
 	amdgpu_device_flush_hdp(adev, NULL);
 
-	DRM_INFO("kiq ring mec %d pipe %d q %d\n", kiq_ring->me, kiq_ring->pipe,
-		 kiq_ring->queue);
+	dev_info(adev->dev, "kiq ring mec %d pipe %d q %d\n", kiq_ring->me,
+		 kiq_ring->pipe, kiq_ring->queue);
 
 	spin_lock(&kiq->ring_lock);
 	r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size *
 					adev->gfx.num_compute_rings +
 					kiq->pmf->set_resources_size);
 	if (r) {
-		DRM_ERROR("Failed to lock KIQ (%d).\n", r);
+		dev_err(adev->dev, "Failed to lock KIQ (%d).\n", r);
 		spin_unlock(&kiq->ring_lock);
 		return r;
 	}
@@ -709,7 +713,7 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id)
 	r = amdgpu_ring_test_helper(kiq_ring);
 	spin_unlock(&kiq->ring_lock);
 	if (r)
-		DRM_ERROR("KCQ enable failed\n");
+		dev_err(adev->dev, "KCQ enable failed\n");
 
 	return r;
 }
@@ -731,7 +735,7 @@ int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id)
 			r = amdgpu_mes_map_legacy_queue(adev,
 							&adev->gfx.gfx_ring[j]);
 			if (r) {
-				DRM_ERROR("failed to map gfx queue\n");
+				dev_err(adev->dev, "failed to map gfx queue\n");
 				return r;
 			}
 		}
@@ -745,7 +749,7 @@ int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id)
 		r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size *
 						adev->gfx.num_gfx_rings);
 		if (r) {
-			DRM_ERROR("Failed to lock KIQ (%d).\n", r);
+			dev_err(adev->dev, "Failed to lock KIQ (%d).\n", r);
 			spin_unlock(&kiq->ring_lock);
 			return r;
 		}
@@ -766,23 +770,13 @@ int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id)
 	r = amdgpu_ring_test_helper(kiq_ring);
 	spin_unlock(&kiq->ring_lock);
 	if (r)
-		DRM_ERROR("KGQ enable failed\n");
+		dev_err(adev->dev, "KGQ enable failed\n");
 
 	return r;
 }
 
-/* amdgpu_gfx_off_ctrl - Handle gfx off feature enable/disable
- *
- * @adev: amdgpu_device pointer
- * @bool enable true: enable gfx off feature, false: disable gfx off feature
- *
- * 1. gfx off feature will be enabled by gfx ip after gfx cg gp enabled.
- * 2. other client can send request to disable gfx off feature, the request should be honored.
- * 3. other client can cancel their request of disable gfx off feature
- * 4. other client should not send request to enable gfx off feature before disable gfx off feature.
- */
-
-void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable)
+static void amdgpu_gfx_do_off_ctrl(struct amdgpu_device *adev, bool enable,
+				   bool no_delay)
 {
 	unsigned long delay = GFX_OFF_DELAY_ENABLE;
 
@@ -804,7 +798,7 @@ void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable)
 		if (adev->gfx.gfx_off_req_count == 0 &&
 		    !adev->gfx.gfx_off_state) {
 			/* If going to s2idle, no need to wait */
-			if (adev->in_s0ix) {
+			if (no_delay) {
 				if (!amdgpu_dpm_set_powergating_by_smu(adev,
 						AMD_IP_BLOCK_TYPE_GFX, true, 0))
 					adev->gfx.gfx_off_state = true;
@@ -836,6 +830,43 @@ unlock:
 	mutex_unlock(&adev->gfx.gfx_off_mutex);
 }
 
+/* amdgpu_gfx_off_ctrl - Handle gfx off feature enable/disable
+ *
+ * @adev: amdgpu_device pointer
+ * @bool enable true: enable gfx off feature, false: disable gfx off feature
+ *
+ * 1. gfx off feature will be enabled by gfx ip after gfx cg pg enabled.
+ * 2. other client can send request to disable gfx off feature, the request should be honored.
+ * 3. other client can cancel their request of disable gfx off feature
+ * 4. other client should not send request to enable gfx off feature before disable gfx off feature.
+ *
+ * gfx off allow will be delayed by GFX_OFF_DELAY_ENABLE ms.
+ */
+void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable)
+{
+	/* If going to s2idle, no need to wait */
+	bool no_delay = adev->in_s0ix ? true : false;
+
+	amdgpu_gfx_do_off_ctrl(adev, enable, no_delay);
+}
+
+/* amdgpu_gfx_off_ctrl_immediate - Handle gfx off feature enable/disable
+ *
+ * @adev: amdgpu_device pointer
+ * @bool enable true: enable gfx off feature, false: disable gfx off feature
+ *
+ * 1. gfx off feature will be enabled by gfx ip after gfx cg pg enabled.
+ * 2. other client can send request to disable gfx off feature, the request should be honored.
+ * 3. other client can cancel their request of disable gfx off feature
+ * 4. other client should not send request to enable gfx off feature before disable gfx off feature.
+ *
+ * gfx off allow will be issued immediately.
+ */
+void amdgpu_gfx_off_ctrl_immediate(struct amdgpu_device *adev, bool enable)
+{
+	amdgpu_gfx_do_off_ctrl(adev, enable, true);
+}
+
 int amdgpu_set_gfx_off_residency(struct amdgpu_device *adev, bool value)
 {
 	int r = 0;
@@ -1000,7 +1031,7 @@ int amdgpu_gfx_cp_ecc_error_irq(struct amdgpu_device *adev,
 
 	ih_data.head = *ras_if;
 
-	DRM_ERROR("CP ECC ERROR IRQ\n");
+	dev_err(adev->dev, "CP ECC ERROR IRQ\n");
 	amdgpu_ras_interrupt_dispatch(adev, &ih_data);
 	return 0;
 }
@@ -1072,6 +1103,9 @@ uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg, uint32_t xcc_
 
 	might_sleep();
 	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+		if (amdgpu_in_reset(adev))
+			goto failed_kiq_read;
+
 		msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
 		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
 	}
@@ -1141,6 +1175,8 @@ void amdgpu_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v, uint3
 
 	might_sleep();
 	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+		if (amdgpu_in_reset(adev))
+			goto failed_kiq_write;
 
 		msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
 		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
@@ -1159,6 +1195,75 @@ failed_kiq_write:
 	dev_err(adev->dev, "failed to write reg:%x\n", reg);
 }
 
+int amdgpu_kiq_hdp_flush(struct amdgpu_device *adev)
+{
+	signed long r, cnt = 0;
+	unsigned long flags;
+	uint32_t seq;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
+	struct amdgpu_ring *ring = &kiq->ring;
+
+	if (amdgpu_device_skip_hw_access(adev))
+		return 0;
+
+	if (adev->enable_mes_kiq && adev->mes.ring[0].sched.ready)
+		return amdgpu_mes_hdp_flush(adev);
+
+	if (!ring->funcs->emit_hdp_flush) {
+		return -EOPNOTSUPP;
+	}
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+	r = amdgpu_ring_alloc(ring, 32);
+	if (r)
+		goto failed_unlock;
+
+	amdgpu_ring_emit_hdp_flush(ring);
+	r = amdgpu_fence_emit_polling(ring, &seq, MAX_KIQ_REG_WAIT);
+	if (r)
+		goto failed_undo;
+
+	amdgpu_ring_commit(ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+	/* don't wait anymore for gpu reset case because this way may
+	 * block gpu_recover() routine forever, e.g. this virt_kiq_rreg
+	 * is triggered in TTM and ttm_bo_lock_delayed_workqueue() will
+	 * never return if we keep waiting in virt_kiq_rreg, which cause
+	 * gpu_recover() hang there.
+	 *
+	 * also don't wait anymore for IRQ context
+	 * */
+	if (r < 1 && (amdgpu_in_reset(adev) || in_interrupt()))
+		goto failed_kiq_hdp_flush;
+
+	might_sleep();
+	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+		if (amdgpu_in_reset(adev))
+			goto failed_kiq_hdp_flush;
+
+		msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+	}
+
+	if (cnt > MAX_KIQ_REG_TRY) {
+		dev_err(adev->dev, "failed to flush HDP via KIQ timeout\n");
+		return -ETIMEDOUT;
+	}
+
+	return 0;
+
+failed_undo:
+	amdgpu_ring_undo(ring);
+failed_unlock:
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+failed_kiq_hdp_flush:
+	dev_err(adev->dev, "failed to flush HDP via KIQ\n");
+	return r < 0 ? r : -EIO;
+}
+
 int amdgpu_gfx_get_num_kcq(struct amdgpu_device *adev)
 {
 	if (amdgpu_num_kcq == -1) {
@@ -1324,6 +1429,10 @@ static ssize_t amdgpu_gfx_get_current_compute_partition(struct device *dev,
 	struct amdgpu_device *adev = drm_to_adev(ddev);
 	int mode;
 
+	/* Only minimal precaution taken to reject requests while in reset.*/
+	if (amdgpu_in_reset(adev))
+		return -EPERM;
+
 	mode = amdgpu_xcp_query_partition_mode(adev->xcp_mgr,
 					       AMDGPU_XCP_FL_NONE);
 
@@ -1367,8 +1476,14 @@ static ssize_t amdgpu_gfx_set_compute_partition(struct device *dev,
 		return -EINVAL;
 	}
 
+	/* Don't allow a switch while under reset */
+	if (!down_read_trylock(&adev->reset_domain->sem))
+		return -EPERM;
+
 	ret = amdgpu_xcp_switch_partition_mode(adev->xcp_mgr, mode);
 
+	up_read(&adev->reset_domain->sem);
+
 	if (ret)
 		return ret;
 
@@ -1411,9 +1526,11 @@ static int amdgpu_gfx_run_cleaner_shader_job(struct amdgpu_ring *ring)
 	struct amdgpu_device *adev = ring->adev;
 	struct drm_gpu_scheduler *sched = &ring->sched;
 	struct drm_sched_entity entity;
+	static atomic_t counter;
 	struct dma_fence *f;
 	struct amdgpu_job *job;
 	struct amdgpu_ib *ib;
+	void *owner;
 	int i, r;
 
 	/* Initialize the scheduler entity */
@@ -1424,13 +1541,22 @@ static int amdgpu_gfx_run_cleaner_shader_job(struct amdgpu_ring *ring)
 		goto err;
 	}
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, &entity, NULL,
-				     64, 0,
-				     &job);
+	/*
+	 * Use some unique dummy value as the owner to make sure we execute
+	 * the cleaner shader on each submission. The value just need to change
+	 * for each submission and is otherwise meaningless.
+	 */
+	owner = (void *)(unsigned long)atomic_inc_return(&counter);
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, &entity, owner,
+				     64, 0, &job,
+				     AMDGPU_KERNEL_JOB_ID_CLEANER_SHADER);
 	if (r)
 		goto err;
 
 	job->enforce_isolation = true;
+	/* always run the cleaner shader */
+	job->run_cleaner_shader = true;
 
 	ib = &job->ibs[0];
 	for (i = 0; i <= ring->funcs->align_mask; ++i)
@@ -1517,6 +1643,9 @@ static ssize_t amdgpu_gfx_set_run_cleaner_shader(struct device *dev,
 	if (adev->in_suspend && !adev->in_runpm)
 		return -EPERM;
 
+	if (adev->gfx.disable_kq)
+		return -EPERM;
+
 	ret = kstrtol(buf, 0, &value);
 
 	if (ret)
@@ -1541,7 +1670,6 @@ static ssize_t amdgpu_gfx_set_run_cleaner_shader(struct device *dev,
 
 	ret = amdgpu_gfx_run_cleaner_shader(adev, value);
 
-	pm_runtime_mark_last_busy(ddev->dev);
 	pm_runtime_put_autosuspend(ddev->dev);
 
 	if (ret)
@@ -1559,7 +1687,8 @@ static ssize_t amdgpu_gfx_set_run_cleaner_shader(struct device *dev,
  * Provides the sysfs read interface to get the current settings of the 'enforce_isolation'
  * feature for each GPU partition. Reading from the 'enforce_isolation'
  * sysfs file returns the isolation settings for all partitions, where '0'
- * indicates disabled and '1' indicates enabled.
+ * indicates disabled, '1' indicates enabled, and '2' indicates enabled in legacy mode,
+ * and '3' indicates enabled without cleaner shader.
  *
  * Return: The number of bytes read from the sysfs file.
  */
@@ -1594,9 +1723,12 @@ static ssize_t amdgpu_gfx_get_enforce_isolation(struct device *dev,
  * @count: The size of the input data
  *
  * This function allows control over the 'enforce_isolation' feature, which
- * serializes access to the graphics engine. Writing '1' or '0' to the
- * 'enforce_isolation' sysfs file enables or disables process isolation for
- * each partition. The input should specify the setting for all partitions.
+ * serializes access to the graphics engine. Writing '0' to disable, '1' to
+ * enable isolation with cleaner shader, '2' to enable legacy isolation without
+ * cleaner shader, or '3' to enable process isolation without submitting the
+ * cleaner shader to the 'enforce_isolation' sysfs file sets the isolation mode
+ * for each partition. The input should specify the setting for all
+ * partitions.
  *
  * Return: The number of bytes written to the sysfs file.
  */
@@ -1633,27 +1765,38 @@ static ssize_t amdgpu_gfx_set_enforce_isolation(struct device *dev,
 		return -EINVAL;
 
 	for (i = 0; i < num_partitions; i++) {
-		if (partition_values[i] != 0 && partition_values[i] != 1)
+		if (partition_values[i] != 0 &&
+		    partition_values[i] != 1 &&
+		    partition_values[i] != 2 &&
+		    partition_values[i] != 3)
 			return -EINVAL;
 	}
 
 	mutex_lock(&adev->enforce_isolation_mutex);
-
 	for (i = 0; i < num_partitions; i++) {
-		if (adev->enforce_isolation[i] && !partition_values[i]) {
-			/* Going from enabled to disabled */
-			amdgpu_vmid_free_reserved(adev, AMDGPU_GFXHUB(i));
-			amdgpu_mes_set_enforce_isolation(adev, i, false);
-		} else if (!adev->enforce_isolation[i] && partition_values[i]) {
-			/* Going from disabled to enabled */
-			amdgpu_vmid_alloc_reserved(adev, AMDGPU_GFXHUB(i));
-			amdgpu_mes_set_enforce_isolation(adev, i, true);
+		switch (partition_values[i]) {
+		case 0:
+		default:
+			adev->enforce_isolation[i] = AMDGPU_ENFORCE_ISOLATION_DISABLE;
+			break;
+		case 1:
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_ENABLE;
+			break;
+		case 2:
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_ENABLE_LEGACY;
+			break;
+		case 3:
+			adev->enforce_isolation[i] =
+				AMDGPU_ENFORCE_ISOLATION_NO_CLEANER_SHADER;
+			break;
 		}
-		adev->enforce_isolation[i] = partition_values[i];
 	}
-
 	mutex_unlock(&adev->enforce_isolation_mutex);
 
+	amdgpu_mes_update_enforce_isolation(adev);
+
 	return count;
 }
 
@@ -1898,39 +2041,41 @@ void amdgpu_gfx_cleaner_shader_init(struct amdgpu_device *adev,
 static void amdgpu_gfx_kfd_sch_ctrl(struct amdgpu_device *adev, u32 idx,
 				    bool enable)
 {
-	mutex_lock(&adev->gfx.kfd_sch_mutex);
+	mutex_lock(&adev->gfx.userq_sch_mutex);
 
 	if (enable) {
 		/* If the count is already 0, it means there's an imbalance bug somewhere.
 		 * Note that the bug may be in a different caller than the one which triggers the
 		 * WARN_ON_ONCE.
 		 */
-		if (WARN_ON_ONCE(adev->gfx.kfd_sch_req_count[idx] == 0)) {
+		if (WARN_ON_ONCE(adev->gfx.userq_sch_req_count[idx] == 0)) {
 			dev_err(adev->dev, "Attempted to enable KFD scheduler when reference count is already zero\n");
 			goto unlock;
 		}
 
-		adev->gfx.kfd_sch_req_count[idx]--;
+		adev->gfx.userq_sch_req_count[idx]--;
 
-		if (adev->gfx.kfd_sch_req_count[idx] == 0 &&
-		    adev->gfx.kfd_sch_inactive[idx]) {
+		if (adev->gfx.userq_sch_req_count[idx] == 0 &&
+		    adev->gfx.userq_sch_inactive[idx]) {
 			schedule_delayed_work(&adev->gfx.enforce_isolation[idx].work,
 					      msecs_to_jiffies(adev->gfx.enforce_isolation_time[idx]));
 		}
 	} else {
-		if (adev->gfx.kfd_sch_req_count[idx] == 0) {
+		if (adev->gfx.userq_sch_req_count[idx] == 0) {
 			cancel_delayed_work_sync(&adev->gfx.enforce_isolation[idx].work);
-			if (!adev->gfx.kfd_sch_inactive[idx]) {
-				amdgpu_amdkfd_stop_sched(adev, idx);
-				adev->gfx.kfd_sch_inactive[idx] = true;
+			if (!adev->gfx.userq_sch_inactive[idx]) {
+				amdgpu_userq_stop_sched_for_enforce_isolation(adev, idx);
+				if (adev->kfd.init_complete)
+					amdgpu_amdkfd_stop_sched(adev, idx);
+				adev->gfx.userq_sch_inactive[idx] = true;
 			}
 		}
 
-		adev->gfx.kfd_sch_req_count[idx]++;
+		adev->gfx.userq_sch_req_count[idx]++;
 	}
 
 unlock:
-	mutex_unlock(&adev->gfx.kfd_sch_mutex);
+	mutex_unlock(&adev->gfx.userq_sch_mutex);
 }
 
 /**
@@ -1975,12 +2120,13 @@ void amdgpu_gfx_enforce_isolation_handler(struct work_struct *work)
 				      msecs_to_jiffies(1));
 	} else {
 		/* Tell KFD to resume the runqueue */
-		if (adev->kfd.init_complete) {
-			WARN_ON_ONCE(!adev->gfx.kfd_sch_inactive[idx]);
-			WARN_ON_ONCE(adev->gfx.kfd_sch_req_count[idx]);
-				amdgpu_amdkfd_start_sched(adev, idx);
-				adev->gfx.kfd_sch_inactive[idx] = false;
-		}
+		WARN_ON_ONCE(!adev->gfx.userq_sch_inactive[idx]);
+		WARN_ON_ONCE(adev->gfx.userq_sch_req_count[idx]);
+
+		amdgpu_userq_start_sched_for_enforce_isolation(adev, idx);
+		if (adev->kfd.init_complete)
+			amdgpu_amdkfd_start_sched(adev, idx);
+		adev->gfx.userq_sch_inactive[idx] = false;
 	}
 	mutex_unlock(&adev->enforce_isolation_mutex);
 }
@@ -2004,7 +2150,7 @@ amdgpu_gfx_enforce_isolation_wait_for_kfd(struct amdgpu_device *adev,
 	bool wait = false;
 
 	mutex_lock(&adev->enforce_isolation_mutex);
-	if (adev->enforce_isolation[idx]) {
+	if (adev->enforce_isolation[idx] == AMDGPU_ENFORCE_ISOLATION_ENABLE) {
 		/* set the initial values if nothing is set */
 		if (!adev->gfx.enforce_isolation_jiffies[idx]) {
 			adev->gfx.enforce_isolation_jiffies[idx] = jiffies;
@@ -2071,7 +2217,7 @@ void amdgpu_gfx_enforce_isolation_ring_begin_use(struct amdgpu_ring *ring)
 	amdgpu_gfx_enforce_isolation_wait_for_kfd(adev, idx);
 
 	mutex_lock(&adev->enforce_isolation_mutex);
-	if (adev->enforce_isolation[idx]) {
+	if (adev->enforce_isolation[idx] == AMDGPU_ENFORCE_ISOLATION_ENABLE) {
 		if (adev->kfd.init_complete)
 			sched_work = true;
 	}
@@ -2108,7 +2254,7 @@ void amdgpu_gfx_enforce_isolation_ring_end_use(struct amdgpu_ring *ring)
 		return;
 
 	mutex_lock(&adev->enforce_isolation_mutex);
-	if (adev->enforce_isolation[idx]) {
+	if (adev->enforce_isolation[idx] == AMDGPU_ENFORCE_ISOLATION_ENABLE) {
 		if (adev->kfd.init_complete)
 			sched_work = true;
 	}
@@ -2118,6 +2264,156 @@ void amdgpu_gfx_enforce_isolation_ring_end_use(struct amdgpu_ring *ring)
 		amdgpu_gfx_kfd_sch_ctrl(adev, idx, true);
 }
 
+void amdgpu_gfx_profile_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, gfx.idle_work.work);
+	enum PP_SMC_POWER_PROFILE profile;
+	u32 i, fences = 0;
+	int r;
+
+	if (adev->gfx.num_gfx_rings)
+		profile = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
+	else
+		profile = PP_SMC_POWER_PROFILE_COMPUTE;
+
+	for (i = 0; i < AMDGPU_MAX_GFX_RINGS; ++i)
+		fences += amdgpu_fence_count_emitted(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < (AMDGPU_MAX_COMPUTE_RINGS * AMDGPU_MAX_GC_INSTANCES); ++i)
+		fences += amdgpu_fence_count_emitted(&adev->gfx.compute_ring[i]);
+	if (!fences && !atomic_read(&adev->gfx.total_submission_cnt)) {
+		mutex_lock(&adev->gfx.workload_profile_mutex);
+		if (adev->gfx.workload_profile_active) {
+			r = amdgpu_dpm_switch_power_profile(adev, profile, false);
+			if (r)
+				dev_warn(adev->dev, "(%d) failed to disable %s power profile mode\n", r,
+					 profile == PP_SMC_POWER_PROFILE_FULLSCREEN3D ?
+					 "fullscreen 3D" : "compute");
+			adev->gfx.workload_profile_active = false;
+		}
+		mutex_unlock(&adev->gfx.workload_profile_mutex);
+	} else {
+		schedule_delayed_work(&adev->gfx.idle_work, GFX_PROFILE_IDLE_TIMEOUT);
+	}
+}
+
+void amdgpu_gfx_profile_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	enum PP_SMC_POWER_PROFILE profile;
+	int r;
+
+	if (amdgpu_dpm_is_overdrive_enabled(adev))
+		return;
+
+	if (adev->gfx.num_gfx_rings)
+		profile = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
+	else
+		profile = PP_SMC_POWER_PROFILE_COMPUTE;
+
+	atomic_inc(&adev->gfx.total_submission_cnt);
+
+	cancel_delayed_work_sync(&adev->gfx.idle_work);
+
+	/* We can safely return early here because we've cancelled the
+	 * the delayed work so there is no one else to set it to false
+	 * and we don't care if someone else sets it to true.
+	 */
+	if (adev->gfx.workload_profile_active)
+		return;
+
+	mutex_lock(&adev->gfx.workload_profile_mutex);
+	if (!adev->gfx.workload_profile_active) {
+		r = amdgpu_dpm_switch_power_profile(adev, profile, true);
+		if (r)
+			dev_warn(adev->dev, "(%d) failed to disable %s power profile mode\n", r,
+				 profile == PP_SMC_POWER_PROFILE_FULLSCREEN3D ?
+				 "fullscreen 3D" : "compute");
+		adev->gfx.workload_profile_active = true;
+	}
+	mutex_unlock(&adev->gfx.workload_profile_mutex);
+}
+
+void amdgpu_gfx_profile_ring_end_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (amdgpu_dpm_is_overdrive_enabled(adev))
+		return;
+
+	atomic_dec(&ring->adev->gfx.total_submission_cnt);
+
+	schedule_delayed_work(&ring->adev->gfx.idle_work, GFX_PROFILE_IDLE_TIMEOUT);
+}
+
+/**
+ * amdgpu_gfx_csb_preamble_start - Set CSB preamble start
+ *
+ * @buffer: This is an output variable that gets the PACKET3 preamble setup.
+ *
+ * Return:
+ * return the latest index.
+ */
+u32 amdgpu_gfx_csb_preamble_start(u32 *buffer)
+{
+	u32 count = 0;
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	buffer[count++] = cpu_to_le32(0x80000000);
+	buffer[count++] = cpu_to_le32(0x80000000);
+
+	return count;
+}
+
+/**
+ * amdgpu_gfx_csb_data_parser - Parser CS data
+ *
+ * @adev: amdgpu_device pointer used to get the CS data and other gfx info.
+ * @buffer: This is an output variable that gets the PACKET3 preamble end.
+ * @count: Index to start set the preemble end.
+ *
+ * Return:
+ * return the latest index.
+ */
+u32 amdgpu_gfx_csb_data_parser(struct amdgpu_device *adev, u32 *buffer, u32 count)
+{
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+	u32 i;
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
+				buffer[count++] = cpu_to_le32(ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
+
+				for (i = 0; i < ext->reg_count; i++)
+					buffer[count++] = cpu_to_le32(ext->extent[i]);
+			}
+		}
+	}
+
+	return count;
+}
+
+/**
+ * amdgpu_gfx_csb_preamble_end - Set CSB preamble end
+ *
+ * @buffer: This is an output variable that gets the PACKET3 preamble end.
+ * @count: Index to start set the preemble end.
+ */
+void amdgpu_gfx_csb_preamble_end(u32 *buffer, u32 count)
+{
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
+	buffer[count++] = cpu_to_le32(0);
+}
+
 /*
  * debugfs for to enable/disable gfx job submission to specific core.
  */
@@ -2258,3 +2554,4 @@ void amdgpu_debugfs_compute_sched_mask_init(struct amdgpu_device *adev)
 			    &amdgpu_debugfs_compute_sched_mask_fops);
 #endif
 }
+
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
index 8b5bd63b5773..efd61a1ccc66 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
@@ -57,6 +57,9 @@ enum amdgpu_gfx_pipe_priority {
 #define AMDGPU_GFX_QUEUE_PRIORITY_MINIMUM  0
 #define AMDGPU_GFX_QUEUE_PRIORITY_MAXIMUM  15
 
+/* 1 second timeout */
+#define GFX_PROFILE_IDLE_TIMEOUT	msecs_to_jiffies(1000)
+
 enum amdgpu_gfx_partition {
 	AMDGPU_SPX_PARTITION_MODE = 0,
 	AMDGPU_DPX_PARTITION_MODE = 1,
@@ -167,10 +170,46 @@ struct amdgpu_kiq {
 #define AMDGPU_GFX_MAX_SE 4
 #define AMDGPU_GFX_MAX_SH_PER_SE 2
 
+/**
+ * amdgpu_rb_config - Configure a single Render Backend (RB)
+ *
+ * Bad RBs are fused off and there is a harvest register the driver reads to
+ * determine which RB(s) are fused off so that the driver can configure the
+ * hardware state so that nothing gets sent to them. There are also user
+ * harvest registers that the driver can program to disable additional RBs,
+ * etc., for testing purposes.
+ */
 struct amdgpu_rb_config {
+	/**
+	 * @rb_backend_disable:
+	 *
+	 * The value captured from register RB_BACKEND_DISABLE indicates if the
+	 * RB backend is disabled or not.
+	 */
 	uint32_t rb_backend_disable;
+
+	/**
+	 * @user_rb_backend_disable:
+	 *
+	 * The value captured from register USER_RB_BACKEND_DISABLE indicates
+	 * if the User RB backend is disabled or not.
+	 */
 	uint32_t user_rb_backend_disable;
+
+	/**
+	 * @raster_config:
+	 *
+	 * To set up all of the states, it is necessary to have two registers
+	 * to keep all of the states. This field holds the first register.
+	 */
 	uint32_t raster_config;
+
+	/**
+	 * @raster_config_1:
+	 *
+	 * To set up all of the states, it is necessary to have two registers
+	 * to keep all of the states. This field holds the second register.
+	 */
 	uint32_t raster_config_1;
 };
 
@@ -218,6 +257,13 @@ struct amdgpu_gfx_config {
 	uint32_t macrotile_mode_array[16];
 
 	struct gb_addr_config gb_addr_config_fields;
+
+	/**
+	 * @rb_config:
+	 *
+	 * Matrix that keeps all the Render Backend (color and depth buffer
+	 * handling) configuration on the 3D engine.
+	 */
 	struct amdgpu_rb_config rb_config[AMDGPU_GFX_MAX_SE][AMDGPU_GFX_MAX_SH_PER_SE];
 
 	/* gfx configure feature */
@@ -302,7 +348,8 @@ struct amdgpu_gfx_funcs {
 	void (*init_spm_golden)(struct amdgpu_device *adev);
 	void (*update_perfmon_mgcg)(struct amdgpu_device *adev, bool enable);
 	int (*get_gfx_shadow_info)(struct amdgpu_device *adev,
-				   struct amdgpu_gfx_shadow_info *shadow_info);
+				   struct amdgpu_gfx_shadow_info *shadow_info,
+				   bool skip_check);
 	enum amdgpu_gfx_partition
 			(*query_partition_mode)(struct amdgpu_device *adev);
 	int (*switch_partition_mode)(struct amdgpu_device *adev,
@@ -471,11 +518,19 @@ struct amdgpu_gfx {
 	bool				enable_cleaner_shader;
 	struct amdgpu_isolation_work	enforce_isolation[MAX_XCP];
 	/* Mutex for synchronizing KFD scheduler operations */
-	struct mutex                    kfd_sch_mutex;
-	u64				kfd_sch_req_count[MAX_XCP];
-	bool				kfd_sch_inactive[MAX_XCP];
+	struct mutex                    userq_sch_mutex;
+	u64				userq_sch_req_count[MAX_XCP];
+	bool				userq_sch_inactive[MAX_XCP];
 	unsigned long			enforce_isolation_jiffies[MAX_XCP];
 	unsigned long			enforce_isolation_time[MAX_XCP];
+
+	atomic_t			total_submission_cnt;
+	struct delayed_work		idle_work;
+	bool				workload_profile_active;
+	struct mutex                    workload_profile_mutex;
+
+	bool				disable_kq;
+	bool				disable_uq;
 };
 
 struct amdgpu_gfx_ras_reg_entry {
@@ -495,7 +550,7 @@ struct amdgpu_gfx_ras_mem_id_entry {
 #define amdgpu_gfx_select_se_sh(adev, se, sh, instance, xcc_id) ((adev)->gfx.funcs->select_se_sh((adev), (se), (sh), (instance), (xcc_id)))
 #define amdgpu_gfx_select_me_pipe_q(adev, me, pipe, q, vmid, xcc_id) ((adev)->gfx.funcs->select_me_pipe_q((adev), (me), (pipe), (q), (vmid), (xcc_id)))
 #define amdgpu_gfx_init_spm_golden(adev) (adev)->gfx.funcs->init_spm_golden((adev))
-#define amdgpu_gfx_get_gfx_shadow_info(adev, si) ((adev)->gfx.funcs->get_gfx_shadow_info((adev), (si)))
+#define amdgpu_gfx_get_gfx_shadow_info(adev, si) ((adev)->gfx.funcs->get_gfx_shadow_info((adev), (si), false))
 
 /**
  * amdgpu_gfx_create_bitmask - create a bitmask
@@ -542,11 +597,10 @@ bool amdgpu_gfx_is_high_priority_compute_queue(struct amdgpu_device *adev,
 					       struct amdgpu_ring *ring);
 bool amdgpu_gfx_is_high_priority_graphics_queue(struct amdgpu_device *adev,
 						struct amdgpu_ring *ring);
-int amdgpu_gfx_me_queue_to_bit(struct amdgpu_device *adev, int me,
-			       int pipe, int queue);
 bool amdgpu_gfx_is_me_queue_enabled(struct amdgpu_device *adev, int me,
 				    int pipe, int queue);
 void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable);
+void amdgpu_gfx_off_ctrl_immediate(struct amdgpu_device *adev, bool enable);
 int amdgpu_get_gfx_off_status(struct amdgpu_device *adev, uint32_t *value);
 int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block);
 void amdgpu_gfx_ras_fini(struct amdgpu_device *adev);
@@ -561,6 +615,7 @@ int amdgpu_gfx_cp_ecc_error_irq(struct amdgpu_device *adev,
 				  struct amdgpu_iv_entry *entry);
 uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg, uint32_t xcc_id);
 void amdgpu_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v, uint32_t xcc_id);
+int amdgpu_kiq_hdp_flush(struct amdgpu_device *adev);
 int amdgpu_gfx_get_num_kcq(struct amdgpu_device *adev);
 void amdgpu_gfx_cp_init_microcode(struct amdgpu_device *adev, uint32_t ucode_id);
 
@@ -584,6 +639,14 @@ void amdgpu_gfx_cleaner_shader_init(struct amdgpu_device *adev,
 void amdgpu_gfx_enforce_isolation_handler(struct work_struct *work);
 void amdgpu_gfx_enforce_isolation_ring_begin_use(struct amdgpu_ring *ring);
 void amdgpu_gfx_enforce_isolation_ring_end_use(struct amdgpu_ring *ring);
+
+void amdgpu_gfx_profile_idle_work_handler(struct work_struct *work);
+void amdgpu_gfx_profile_ring_begin_use(struct amdgpu_ring *ring);
+void amdgpu_gfx_profile_ring_end_use(struct amdgpu_ring *ring);
+u32 amdgpu_gfx_csb_preamble_start(u32 *buffer);
+u32 amdgpu_gfx_csb_data_parser(struct amdgpu_device *adev, u32 *buffer, u32 count);
+void amdgpu_gfx_csb_preamble_end(u32 *buffer, u32 count);
+
 void amdgpu_debugfs_gfx_sched_mask_init(struct amdgpu_device *adev);
 void amdgpu_debugfs_compute_sched_mask_init(struct amdgpu_device *adev);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
index 1c19a65e6553..869bceb0fe2c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
@@ -38,6 +38,13 @@
 #include <drm/drm_drv.h>
 #include <drm/ttm/ttm_tt.h>
 
+static const u64 four_gb = 0x100000000ULL;
+
+bool amdgpu_gmc_is_pdb0_enabled(struct amdgpu_device *adev)
+{
+	return adev->gmc.xgmi.connected_to_cpu || amdgpu_virt_xgmi_migrate_enabled(adev);
+}
+
 /**
  * amdgpu_gmc_pdb0_alloc - allocate vram for pdb0
  *
@@ -251,10 +258,20 @@ void amdgpu_gmc_sysvm_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc
 	u64 hive_vram_end = mc->xgmi.node_segment_size * mc->xgmi.num_physical_nodes - 1;
 	mc->vram_start = mc->xgmi.node_segment_size * mc->xgmi.physical_node_id;
 	mc->vram_end = mc->vram_start + mc->xgmi.node_segment_size - 1;
-	mc->gart_start = hive_vram_end + 1;
+	/* node_segment_size may not 4GB aligned on SRIOV, align up is needed. */
+	mc->gart_start = ALIGN(hive_vram_end + 1, four_gb);
 	mc->gart_end = mc->gart_start + mc->gart_size - 1;
-	mc->fb_start = hive_vram_start;
-	mc->fb_end = hive_vram_end;
+	if (amdgpu_virt_xgmi_migrate_enabled(adev)) {
+		/* set mc->vram_start to 0 to switch the returned GPU address of
+		 * amdgpu_bo_create_reserved() from FB aperture to GART aperture.
+		 */
+		mc->vram_start = 0;
+		mc->vram_end = mc->vram_start + mc->mc_vram_size - 1;
+		mc->visible_vram_size = min(mc->visible_vram_size, mc->real_vram_size);
+	} else {
+		mc->fb_start = hive_vram_start;
+		mc->fb_end = hive_vram_end;
+	}
 	dev_info(adev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n",
 			mc->mc_vram_size >> 20, mc->vram_start,
 			mc->vram_end, mc->real_vram_size >> 20);
@@ -269,14 +286,13 @@ void amdgpu_gmc_sysvm_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc
  * @mc: memory controller structure holding memory information
  * @gart_placement: GART placement policy with respect to VRAM
  *
- * Function will place try to place GART before or after VRAM.
+ * Function will try to place GART before or after VRAM.
  * If GART size is bigger than space left then we ajust GART size.
  * Thus function will never fails.
  */
 void amdgpu_gmc_gart_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc,
 			      enum amdgpu_gart_placement gart_placement)
 {
-	const uint64_t four_gb = 0x100000000ULL;
 	u64 size_af, size_bf;
 	/*To avoid the hole, limit the max mc address to AMDGPU_GMC_HOLE_START*/
 	u64 max_mc_address = min(adev->gmc.mc_mask, AMDGPU_GMC_HOLE_START - 1);
@@ -573,6 +589,7 @@ int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev)
 	unsigned vm_inv_engs[AMDGPU_MAX_VMHUBS] = {0};
 	unsigned i;
 	unsigned vmhub, inv_eng;
+	struct amdgpu_ring *shared_ring;
 
 	/* init the vm inv eng for all vmhubs */
 	for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS) {
@@ -580,6 +597,9 @@ int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev)
 		/* reserve engine 5 for firmware */
 		if (adev->enable_mes)
 			vm_inv_engs[i] &= ~(1 << 5);
+		/* reserve engine 6 for uni mes */
+		if (adev->enable_uni_mes)
+			vm_inv_engs[i] &= ~(1 << 6);
 		/* reserve mmhub engine 3 for firmware */
 		if (adev->enable_umsch_mm)
 			vm_inv_engs[i] &= ~(1 << 3);
@@ -591,7 +611,12 @@ int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev)
 
 		if (ring == &adev->mes.ring[0] ||
 		    ring == &adev->mes.ring[1] ||
-		    ring == &adev->umsch_mm.ring)
+		    ring == &adev->umsch_mm.ring ||
+		    ring == &adev->cper.ring_buf)
+			continue;
+
+		/* Skip if the ring is a shared ring */
+		if (amdgpu_sdma_is_shared_inv_eng(adev, ring))
 			continue;
 
 		inv_eng = ffs(vm_inv_engs[vmhub]);
@@ -606,6 +631,21 @@ int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev)
 
 		dev_info(adev->dev, "ring %s uses VM inv eng %u on hub %u\n",
 			 ring->name, ring->vm_inv_eng, ring->vm_hub);
+		/* SDMA has a special packet which allows it to use the same
+		 * invalidation engine for all the rings in one instance.
+		 * Therefore, we do not allocate a separate VM invalidation engine
+		 * for SDMA page rings. Instead, they share the VM invalidation
+		 * engine with the SDMA gfx ring. This change ensures efficient
+		 * resource management and avoids the issue of insufficient VM
+		 * invalidation engines.
+		 */
+		shared_ring = amdgpu_sdma_get_shared_ring(adev, ring);
+		if (shared_ring) {
+			shared_ring->vm_inv_eng = ring->vm_inv_eng;
+			dev_info(adev->dev, "ring %s shares VM invalidation engine %u with ring %s on hub %u\n",
+					ring->name, ring->vm_inv_eng, shared_ring->name, ring->vm_hub);
+			continue;
+		}
 	}
 
 	return 0;
@@ -653,7 +693,7 @@ void amdgpu_gmc_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
 	r = amdgpu_job_alloc_with_ib(ring->adev, &adev->mman.high_pr,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     16 * 4, AMDGPU_IB_POOL_IMMEDIATE,
-				     &job);
+				     &job, AMDGPU_KERNEL_JOB_ID_FLUSH_GPU_TLB);
 	if (r)
 		goto error_alloc;
 
@@ -678,12 +718,10 @@ int amdgpu_gmc_flush_gpu_tlb_pasid(struct amdgpu_device *adev, uint16_t pasid,
 				   uint32_t flush_type, bool all_hub,
 				   uint32_t inst)
 {
-	u32 usec_timeout = amdgpu_sriov_vf(adev) ? SRIOV_USEC_TIMEOUT :
-		adev->usec_timeout;
 	struct amdgpu_ring *ring = &adev->gfx.kiq[inst].ring;
 	struct amdgpu_kiq *kiq = &adev->gfx.kiq[inst];
 	unsigned int ndw;
-	int r;
+	int r, cnt = 0;
 	uint32_t seq;
 
 	/*
@@ -740,10 +778,21 @@ int amdgpu_gmc_flush_gpu_tlb_pasid(struct amdgpu_device *adev, uint16_t pasid,
 
 		amdgpu_ring_commit(ring);
 		spin_unlock(&adev->gfx.kiq[inst].ring_lock);
-		if (amdgpu_fence_wait_polling(ring, seq, usec_timeout) < 1) {
+
+		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+		might_sleep();
+		while (r < 1 && cnt++ < MAX_KIQ_REG_TRY &&
+		       !amdgpu_reset_pending(adev->reset_domain)) {
+			msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+			r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+		}
+
+		if (cnt > MAX_KIQ_REG_TRY) {
 			dev_err(adev->dev, "timeout waiting for kiq fence\n");
 			r = -ETIME;
-		}
+		} else
+			r = 0;
 	}
 
 error_unlock_reset:
@@ -851,6 +900,7 @@ void amdgpu_gmc_tmz_set(struct amdgpu_device *adev)
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		/* Don't enable it by default yet.
 		 */
 		if (amdgpu_tmz < 1) {
@@ -888,6 +938,7 @@ void amdgpu_gmc_noretry_set(struct amdgpu_device *adev)
 				gc_ver == IP_VERSION(9, 4, 2) ||
 				gc_ver == IP_VERSION(9, 4, 3) ||
 				gc_ver == IP_VERSION(9, 4, 4) ||
+				gc_ver == IP_VERSION(9, 5, 0) ||
 				gc_ver >= IP_VERSION(10, 3, 0));
 
 	if (!amdgpu_sriov_xnack_support(adev))
@@ -1009,9 +1060,7 @@ void amdgpu_gmc_init_pdb0(struct amdgpu_device *adev)
 	 */
 	u64 vram_size = adev->gmc.xgmi.node_segment_size * adev->gmc.xgmi.num_physical_nodes;
 	u64 pde0_page_size = (1ULL<<adev->gmc.vmid0_page_table_block_size)<<21;
-	u64 vram_addr = adev->vm_manager.vram_base_offset -
-		adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
-	u64 vram_end = vram_addr + vram_size;
+	u64 vram_addr, vram_end;
 	u64 gart_ptb_gpu_pa = amdgpu_gmc_vram_pa(adev, adev->gart.bo);
 	int idx;
 
@@ -1024,6 +1073,11 @@ void amdgpu_gmc_init_pdb0(struct amdgpu_device *adev)
 	flags |= AMDGPU_PTE_FRAG((adev->gmc.vmid0_page_table_block_size + 9*1));
 	flags |= AMDGPU_PDE_PTE_FLAG(adev);
 
+	vram_addr = adev->vm_manager.vram_base_offset;
+	if (!amdgpu_virt_xgmi_migrate_enabled(adev))
+		vram_addr -= adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
+	vram_end = vram_addr + vram_size;
+
 	/* The first n PDE0 entries are used as PTE,
 	 * pointing to vram
 	 */
@@ -1198,6 +1252,10 @@ static ssize_t current_memory_partition_show(
 	struct amdgpu_device *adev = drm_to_adev(ddev);
 	enum amdgpu_memory_partition mode;
 
+	/* Only minimal precaution taken to reject requests while in reset */
+	if (amdgpu_in_reset(adev))
+		return -EPERM;
+
 	mode = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
 	if ((mode >= ARRAY_SIZE(nps_desc)) ||
 	    (BIT(mode) & AMDGPU_ALL_NPS_MASK) != BIT(mode))
@@ -1393,3 +1451,232 @@ bool amdgpu_gmc_need_reset_on_init(struct amdgpu_device *adev)
 
 	return false;
 }
+
+enum amdgpu_memory_partition
+amdgpu_gmc_get_vf_memory_partition(struct amdgpu_device *adev)
+{
+	switch (adev->gmc.num_mem_partitions) {
+	case 0:
+		return UNKNOWN_MEMORY_PARTITION_MODE;
+	case 1:
+		return AMDGPU_NPS1_PARTITION_MODE;
+	case 2:
+		return AMDGPU_NPS2_PARTITION_MODE;
+	case 4:
+		return AMDGPU_NPS4_PARTITION_MODE;
+	case 8:
+		return AMDGPU_NPS8_PARTITION_MODE;
+	default:
+		return AMDGPU_NPS1_PARTITION_MODE;
+	}
+}
+
+enum amdgpu_memory_partition
+amdgpu_gmc_get_memory_partition(struct amdgpu_device *adev, u32 *supp_modes)
+{
+	enum amdgpu_memory_partition mode = UNKNOWN_MEMORY_PARTITION_MODE;
+
+	if (adev->nbio.funcs &&
+	    adev->nbio.funcs->get_memory_partition_mode)
+		mode = adev->nbio.funcs->get_memory_partition_mode(adev,
+								   supp_modes);
+	else
+		dev_warn(adev->dev, "memory partition mode query is not supported\n");
+
+	return mode;
+}
+
+enum amdgpu_memory_partition
+amdgpu_gmc_query_memory_partition(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev))
+		return amdgpu_gmc_get_vf_memory_partition(adev);
+	else
+		return amdgpu_gmc_get_memory_partition(adev, NULL);
+}
+
+static bool amdgpu_gmc_validate_partition_info(struct amdgpu_device *adev)
+{
+	enum amdgpu_memory_partition mode;
+	u32 supp_modes;
+	bool valid;
+
+	mode = amdgpu_gmc_get_memory_partition(adev, &supp_modes);
+
+	/* Mode detected by hardware not present in supported modes */
+	if ((mode != UNKNOWN_MEMORY_PARTITION_MODE) &&
+	    !(BIT(mode - 1) & supp_modes))
+		return false;
+
+	switch (mode) {
+	case UNKNOWN_MEMORY_PARTITION_MODE:
+	case AMDGPU_NPS1_PARTITION_MODE:
+		valid = (adev->gmc.num_mem_partitions == 1);
+		break;
+	case AMDGPU_NPS2_PARTITION_MODE:
+		valid = (adev->gmc.num_mem_partitions == 2);
+		break;
+	case AMDGPU_NPS4_PARTITION_MODE:
+		valid = (adev->gmc.num_mem_partitions == 3 ||
+			 adev->gmc.num_mem_partitions == 4);
+		break;
+	case AMDGPU_NPS8_PARTITION_MODE:
+		valid = (adev->gmc.num_mem_partitions == 8);
+		break;
+	default:
+		valid = false;
+	}
+
+	return valid;
+}
+
+static bool amdgpu_gmc_is_node_present(int *node_ids, int num_ids, int nid)
+{
+	int i;
+
+	/* Check if node with id 'nid' is present in 'node_ids' array */
+	for (i = 0; i < num_ids; ++i)
+		if (node_ids[i] == nid)
+			return true;
+
+	return false;
+}
+
+static void
+amdgpu_gmc_init_acpi_mem_ranges(struct amdgpu_device *adev,
+				struct amdgpu_mem_partition_info *mem_ranges)
+{
+	struct amdgpu_numa_info numa_info;
+	int node_ids[AMDGPU_MAX_MEM_RANGES];
+	int num_ranges = 0, ret;
+	int num_xcc, xcc_id;
+	uint32_t xcc_mask;
+
+	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
+	xcc_mask = (1U << num_xcc) - 1;
+
+	for_each_inst(xcc_id, xcc_mask)	{
+		ret = amdgpu_acpi_get_mem_info(adev, xcc_id, &numa_info);
+		if (ret)
+			continue;
+
+		if (numa_info.nid == NUMA_NO_NODE) {
+			mem_ranges[0].size = numa_info.size;
+			mem_ranges[0].numa.node = numa_info.nid;
+			num_ranges = 1;
+			break;
+		}
+
+		if (amdgpu_gmc_is_node_present(node_ids, num_ranges,
+					     numa_info.nid))
+			continue;
+
+		node_ids[num_ranges] = numa_info.nid;
+		mem_ranges[num_ranges].numa.node = numa_info.nid;
+		mem_ranges[num_ranges].size = numa_info.size;
+		++num_ranges;
+	}
+
+	adev->gmc.num_mem_partitions = num_ranges;
+}
+
+void amdgpu_gmc_init_sw_mem_ranges(struct amdgpu_device *adev,
+				   struct amdgpu_mem_partition_info *mem_ranges)
+{
+	enum amdgpu_memory_partition mode;
+	u32 start_addr = 0, size;
+	int i, r, l;
+
+	mode = amdgpu_gmc_query_memory_partition(adev);
+
+	switch (mode) {
+	case UNKNOWN_MEMORY_PARTITION_MODE:
+		adev->gmc.num_mem_partitions = 0;
+		break;
+	case AMDGPU_NPS1_PARTITION_MODE:
+		adev->gmc.num_mem_partitions = 1;
+		break;
+	case AMDGPU_NPS2_PARTITION_MODE:
+		adev->gmc.num_mem_partitions = 2;
+		break;
+	case AMDGPU_NPS4_PARTITION_MODE:
+		if (adev->flags & AMD_IS_APU)
+			adev->gmc.num_mem_partitions = 3;
+		else
+			adev->gmc.num_mem_partitions = 4;
+		break;
+	case AMDGPU_NPS8_PARTITION_MODE:
+		adev->gmc.num_mem_partitions = 8;
+		break;
+	default:
+		adev->gmc.num_mem_partitions = 1;
+		break;
+	}
+
+	/* Use NPS range info, if populated */
+	r = amdgpu_gmc_get_nps_memranges(adev, mem_ranges,
+					 &adev->gmc.num_mem_partitions);
+	if (!r) {
+		l = 0;
+		for (i = 1; i < adev->gmc.num_mem_partitions; ++i) {
+			if (mem_ranges[i].range.lpfn >
+			    mem_ranges[i - 1].range.lpfn)
+				l = i;
+		}
+
+	} else {
+		if (!adev->gmc.num_mem_partitions) {
+			dev_warn(adev->dev,
+				 "Not able to detect NPS mode, fall back to NPS1\n");
+			adev->gmc.num_mem_partitions = 1;
+		}
+		/* Fallback to sw based calculation */
+		size = (adev->gmc.real_vram_size + SZ_16M) >> AMDGPU_GPU_PAGE_SHIFT;
+		size /= adev->gmc.num_mem_partitions;
+
+		for (i = 0; i < adev->gmc.num_mem_partitions; ++i) {
+			mem_ranges[i].range.fpfn = start_addr;
+			mem_ranges[i].size =
+				((u64)size << AMDGPU_GPU_PAGE_SHIFT);
+			mem_ranges[i].range.lpfn = start_addr + size - 1;
+			start_addr += size;
+		}
+
+		l = adev->gmc.num_mem_partitions - 1;
+	}
+
+	/* Adjust the last one */
+	mem_ranges[l].range.lpfn =
+		(adev->gmc.real_vram_size >> AMDGPU_GPU_PAGE_SHIFT) - 1;
+	mem_ranges[l].size =
+		adev->gmc.real_vram_size -
+		((u64)mem_ranges[l].range.fpfn << AMDGPU_GPU_PAGE_SHIFT);
+}
+
+int amdgpu_gmc_init_mem_ranges(struct amdgpu_device *adev)
+{
+	bool valid;
+
+	adev->gmc.mem_partitions = kcalloc(AMDGPU_MAX_MEM_RANGES,
+					   sizeof(struct amdgpu_mem_partition_info),
+					   GFP_KERNEL);
+	if (!adev->gmc.mem_partitions)
+		return -ENOMEM;
+
+	if (adev->gmc.is_app_apu)
+		amdgpu_gmc_init_acpi_mem_ranges(adev, adev->gmc.mem_partitions);
+	else
+		amdgpu_gmc_init_sw_mem_ranges(adev, adev->gmc.mem_partitions);
+
+	if (amdgpu_sriov_vf(adev))
+		valid = true;
+	else
+		valid = amdgpu_gmc_validate_partition_info(adev);
+	if (!valid) {
+		/* TODO: handle invalid case */
+		dev_warn(adev->dev,
+			 "Mem ranges not matching with hardware config\n");
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
index 459a30fe239f..727342689d4b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
@@ -29,6 +29,7 @@
 #include <linux/types.h>
 
 #include "amdgpu_irq.h"
+#include "amdgpu_xgmi.h"
 #include "amdgpu_ras.h"
 
 /* VA hole for 48bit addresses on Vega10 */
@@ -61,6 +62,9 @@
  */
 #define AMDGPU_GMC_FAULT_TIMEOUT	5000ULL
 
+/* XNACK flags */
+#define AMDGPU_GMC_XNACK_FLAG_CHAIN BIT(0)
+
 struct firmware;
 
 enum amdgpu_memory_partition {
@@ -80,6 +84,13 @@ enum amdgpu_memory_partition {
 
 #define AMDGPU_GMC_INIT_RESET_NPS  BIT(0)
 
+#define AMDGPU_MAX_MEM_RANGES 8
+
+#define AMDGPU_GMC9_FAULT_SOURCE_DATA_RETRY 0x80
+#define AMDGPU_GMC9_FAULT_SOURCE_DATA_READ  0x40
+#define AMDGPU_GMC9_FAULT_SOURCE_DATA_WRITE 0x20
+#define AMDGPU_GMC9_FAULT_SOURCE_DATA_EXE   0x10
+
 /*
  * GMC page fault information
  */
@@ -148,15 +159,15 @@ struct amdgpu_gmc_funcs {
 				   unsigned pasid);
 	/* enable/disable PRT support */
 	void (*set_prt)(struct amdgpu_device *adev, bool enable);
-	/* map mtype to hardware flags */
-	uint64_t (*map_mtype)(struct amdgpu_device *adev, uint32_t flags);
 	/* get the pde for a given mc addr */
 	void (*get_vm_pde)(struct amdgpu_device *adev, int level,
 			   u64 *dst, u64 *flags);
-	/* get the pte flags to use for a BO VA mapping */
+	/* get the pte flags to use for PTEs */
 	void (*get_vm_pte)(struct amdgpu_device *adev,
-			   struct amdgpu_bo_va_mapping *mapping,
-			   uint64_t *flags);
+			   struct amdgpu_vm *vm,
+			   struct amdgpu_bo *bo,
+			   uint32_t vm_flags,
+			   uint64_t *pte_flags);
 	/* override per-page pte flags */
 	void (*override_vm_pte_flags)(struct amdgpu_device *dev,
 				      struct amdgpu_vm *vm,
@@ -174,28 +185,6 @@ struct amdgpu_gmc_funcs {
 	bool (*need_reset_on_init)(struct amdgpu_device *adev);
 };
 
-struct amdgpu_xgmi_ras {
-	struct amdgpu_ras_block_object ras_block;
-};
-
-struct amdgpu_xgmi {
-	/* from psp */
-	u64 node_id;
-	u64 hive_id;
-	/* fixed per family */
-	u64 node_segment_size;
-	/* physical node (0-3) */
-	unsigned physical_node_id;
-	/* number of nodes (0-4) */
-	unsigned num_physical_nodes;
-	/* gpu list in the same hive */
-	struct list_head head;
-	bool supported;
-	struct ras_common_if *ras_if;
-	bool connected_to_cpu;
-	struct amdgpu_xgmi_ras *ras;
-};
-
 struct amdgpu_mem_partition_info {
 	union {
 		struct {
@@ -322,6 +311,7 @@ struct amdgpu_gmc {
 	struct amdgpu_xgmi xgmi;
 	struct amdgpu_irq_src	ecc_irq;
 	int noretry;
+	uint32_t xnack_flags;
 
 	uint32_t	vmid0_page_table_block_size;
 	uint32_t	vmid0_page_table_depth;
@@ -371,9 +361,10 @@ struct amdgpu_gmc {
 
 #define amdgpu_gmc_emit_flush_gpu_tlb(r, vmid, addr) (r)->adev->gmc.gmc_funcs->emit_flush_gpu_tlb((r), (vmid), (addr))
 #define amdgpu_gmc_emit_pasid_mapping(r, vmid, pasid) (r)->adev->gmc.gmc_funcs->emit_pasid_mapping((r), (vmid), (pasid))
-#define amdgpu_gmc_map_mtype(adev, flags) (adev)->gmc.gmc_funcs->map_mtype((adev),(flags))
 #define amdgpu_gmc_get_vm_pde(adev, level, dst, flags) (adev)->gmc.gmc_funcs->get_vm_pde((adev), (level), (dst), (flags))
-#define amdgpu_gmc_get_vm_pte(adev, mapping, flags) (adev)->gmc.gmc_funcs->get_vm_pte((adev), (mapping), (flags))
+#define amdgpu_gmc_get_vm_pte(adev, vm, bo, vm_flags, pte_flags) \
+	((adev)->gmc.gmc_funcs->get_vm_pte((adev), (vm), (bo), (vm_flags), \
+					   (pte_flags)))
 #define amdgpu_gmc_override_vm_pte_flags(adev, vm, addr, pte_flags)	\
 	(adev)->gmc.gmc_funcs->override_vm_pte_flags			\
 		((adev), (vm), (addr), (pte_flags))
@@ -411,6 +402,7 @@ static inline uint64_t amdgpu_gmc_sign_extend(uint64_t addr)
 	return addr;
 }
 
+bool amdgpu_gmc_is_pdb0_enabled(struct amdgpu_device *adev);
 int amdgpu_gmc_pdb0_alloc(struct amdgpu_device *adev);
 void amdgpu_gmc_get_pde_for_bo(struct amdgpu_bo *bo, int level,
 			       uint64_t *addr, uint64_t *flags);
@@ -472,5 +464,13 @@ int amdgpu_gmc_request_memory_partition(struct amdgpu_device *adev,
 					int nps_mode);
 void amdgpu_gmc_prepare_nps_mode_change(struct amdgpu_device *adev);
 bool amdgpu_gmc_need_reset_on_init(struct amdgpu_device *adev);
-
+enum amdgpu_memory_partition
+amdgpu_gmc_get_vf_memory_partition(struct amdgpu_device *adev);
+enum amdgpu_memory_partition
+amdgpu_gmc_get_memory_partition(struct amdgpu_device *adev, u32 *supp_modes);
+enum amdgpu_memory_partition
+amdgpu_gmc_query_memory_partition(struct amdgpu_device *adev);
+int amdgpu_gmc_init_mem_ranges(struct amdgpu_device *adev);
+void amdgpu_gmc_init_sw_mem_ranges(struct amdgpu_device *adev,
+				   struct amdgpu_mem_partition_info *mem_ranges);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
index 0760e70402ec..895c1e4c6747 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
@@ -284,6 +284,7 @@ int amdgpu_gtt_mgr_init(struct amdgpu_device *adev, uint64_t gtt_size)
 	ttm_resource_manager_init(man, &adev->mman.bdev, gtt_size);
 
 	start = AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GTT_NUM_TRANSFER_WINDOWS;
+	start += amdgpu_vce_required_gart_pages(adev);
 	size = (adev->gmc.gart_size >> PAGE_SHIFT) - start;
 	drm_mm_init(&mgr->mm, start, size);
 	spin_lock_init(&mgr->lock);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
index b6cf801939aa..5a60d69a3e1f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
@@ -22,6 +22,7 @@
  */
 #include "amdgpu.h"
 #include "amdgpu_ras.h"
+#include <uapi/linux/kfd_ioctl.h>
 
 int amdgpu_hdp_ras_sw_init(struct amdgpu_device *adev)
 {
@@ -46,3 +47,38 @@ int amdgpu_hdp_ras_sw_init(struct amdgpu_device *adev)
 	/* hdp ras follows amdgpu_ras_block_late_init_default for late init */
 	return 0;
 }
+
+void amdgpu_hdp_generic_flush(struct amdgpu_device *adev,
+			      struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32((adev->rmmio_remap.reg_offset +
+			KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >>
+			       2,
+		       0);
+		if (adev->nbio.funcs->get_memsize)
+			adev->nbio.funcs->get_memsize(adev);
+	} else {
+		amdgpu_ring_emit_wreg(ring,
+				      (adev->rmmio_remap.reg_offset +
+				       KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >>
+					      2,
+				      0);
+	}
+}
+
+void amdgpu_hdp_invalidate(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (adev->asic_funcs && adev->asic_funcs->invalidate_hdp)
+		adev->asic_funcs->invalidate_hdp(adev, ring);
+	else if (adev->hdp.funcs && adev->hdp.funcs->invalidate_hdp)
+		adev->hdp.funcs->invalidate_hdp(adev, ring);
+}
+
+void amdgpu_hdp_flush(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (adev->asic_funcs && adev->asic_funcs->flush_hdp)
+		adev->asic_funcs->flush_hdp(adev, ring);
+	else if (adev->hdp.funcs && adev->hdp.funcs->flush_hdp)
+		adev->hdp.funcs->flush_hdp(adev, ring);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h
index 7b8a6152dc8d..d9f488fa76b9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h
@@ -44,4 +44,10 @@ struct amdgpu_hdp {
 };
 
 int amdgpu_hdp_ras_sw_init(struct amdgpu_device *adev);
+void amdgpu_hdp_generic_flush(struct amdgpu_device *adev,
+			      struct amdgpu_ring *ring);
+void amdgpu_hdp_invalidate(struct amdgpu_device *adev,
+			   struct amdgpu_ring *ring);
+void amdgpu_hdp_flush(struct amdgpu_device *adev,
+		      struct amdgpu_ring *ring);
 #endif /* __AMDGPU_HDP_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c
index e36fede7f74c..90d26d820bac 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c
@@ -167,19 +167,14 @@ void amdgpu_hmm_unregister(struct amdgpu_bo *bo)
 
 int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier,
 			       uint64_t start, uint64_t npages, bool readonly,
-			       void *owner, struct page **pages,
-			       struct hmm_range **phmm_range)
+			       void *owner,
+			       struct amdgpu_hmm_range *range)
 {
-	struct hmm_range *hmm_range;
 	unsigned long end;
 	unsigned long timeout;
-	unsigned long i;
 	unsigned long *pfns;
 	int r = 0;
-
-	hmm_range = kzalloc(sizeof(*hmm_range), GFP_KERNEL);
-	if (unlikely(!hmm_range))
-		return -ENOMEM;
+	struct hmm_range *hmm_range = &range->hmm_range;
 
 	pfns = kvmalloc_array(npages, sizeof(*pfns), GFP_KERNEL);
 	if (unlikely(!pfns)) {
@@ -222,36 +217,77 @@ retry:
 	hmm_range->start = start;
 	hmm_range->hmm_pfns = pfns;
 
-	/*
-	 * Due to default_flags, all pages are HMM_PFN_VALID or
-	 * hmm_range_fault() fails. FIXME: The pages cannot be touched outside
-	 * the notifier_lock, and mmu_interval_read_retry() must be done first.
-	 */
-	for (i = 0; pages && i < npages; i++)
-		pages[i] = hmm_pfn_to_page(pfns[i]);
-
-	*phmm_range = hmm_range;
-
 	return 0;
 
 out_free_pfns:
 	kvfree(pfns);
+	hmm_range->hmm_pfns = NULL;
 out_free_range:
-	kfree(hmm_range);
-
 	if (r == -EBUSY)
 		r = -EAGAIN;
 	return r;
 }
 
-bool amdgpu_hmm_range_get_pages_done(struct hmm_range *hmm_range)
+/**
+ * amdgpu_hmm_range_valid - check if an HMM range is still valid
+ * @range: pointer to the &struct amdgpu_hmm_range to validate
+ *
+ * Determines whether the given HMM range @range is still valid by
+ * checking for invalidations via the MMU notifier sequence. This is
+ * typically used to verify that the range has not been invalidated
+ * by concurrent address space updates before it is accessed.
+ *
+ * Return:
+ * * true if @range is valid and can be used safely
+ * * false if @range is NULL or has been invalidated
+ */
+bool amdgpu_hmm_range_valid(struct amdgpu_hmm_range *range)
 {
-	bool r;
+	if (!range)
+		return false;
 
-	r = mmu_interval_read_retry(hmm_range->notifier,
-				    hmm_range->notifier_seq);
-	kvfree(hmm_range->hmm_pfns);
-	kfree(hmm_range);
+	return !mmu_interval_read_retry(range->hmm_range.notifier,
+					range->hmm_range.notifier_seq);
+}
 
-	return r;
+/**
+ * amdgpu_hmm_range_alloc - allocate and initialize an AMDGPU HMM range
+ * @bo: optional buffer object to associate with this HMM range
+ *
+ * Allocates memory for amdgpu_hmm_range and associates it with the @bo passed.
+ * The reference count of the @bo is incremented.
+ *
+ * Return:
+ * Pointer to a newly allocated struct amdgpu_hmm_range on success,
+ * or NULL if memory allocation fails.
+ */
+struct amdgpu_hmm_range *amdgpu_hmm_range_alloc(struct amdgpu_bo *bo)
+{
+	struct amdgpu_hmm_range *range;
+
+	range = kzalloc(sizeof(*range), GFP_KERNEL);
+	if (!range)
+		return NULL;
+
+	range->bo = amdgpu_bo_ref(bo);
+	return range;
+}
+
+/**
+ * amdgpu_hmm_range_free - release an AMDGPU HMM range
+ * @range: pointer to the range object to free
+ *
+ * Releases all resources held by @range, including the associated
+ * hmm_pfns and the dropping reference of associated bo if any.
+ *
+ * Return: void
+ */
+void amdgpu_hmm_range_free(struct amdgpu_hmm_range *range)
+{
+	if (!range)
+		return;
+
+	kvfree(range->hmm_range.hmm_pfns);
+	amdgpu_bo_unref(&range->bo);
+	kfree(range);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h
index e2edcd010ccc..140bc9cd57b4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h
@@ -31,13 +31,20 @@
 #include <linux/interval_tree.h>
 #include <linux/mmu_notifier.h>
 
+struct amdgpu_hmm_range {
+	struct hmm_range hmm_range;
+	struct amdgpu_bo *bo;
+};
+
 int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier,
 			       uint64_t start, uint64_t npages, bool readonly,
-			       void *owner, struct page **pages,
-			       struct hmm_range **phmm_range);
-bool amdgpu_hmm_range_get_pages_done(struct hmm_range *hmm_range);
+			       void *owner,
+			       struct amdgpu_hmm_range *range);
 
 #if defined(CONFIG_HMM_MIRROR)
+bool amdgpu_hmm_range_valid(struct amdgpu_hmm_range *range);
+struct amdgpu_hmm_range *amdgpu_hmm_range_alloc(struct amdgpu_bo *bo);
+void amdgpu_hmm_range_free(struct amdgpu_hmm_range *range);
 int amdgpu_hmm_register(struct amdgpu_bo *bo, unsigned long addr);
 void amdgpu_hmm_unregister(struct amdgpu_bo *bo);
 #else
@@ -47,7 +54,20 @@ static inline int amdgpu_hmm_register(struct amdgpu_bo *bo, unsigned long addr)
 		      "add CONFIG_ZONE_DEVICE=y in config file to fix this\n");
 	return -ENODEV;
 }
+
 static inline void amdgpu_hmm_unregister(struct amdgpu_bo *bo) {}
+
+static inline bool amdgpu_hmm_range_valid(struct amdgpu_hmm_range *range)
+{
+	return false;
+}
+
+static inline struct amdgpu_hmm_range *amdgpu_hmm_range_alloc(struct amdgpu_bo *bo)
+{
+	return NULL;
+}
+
+static inline void amdgpu_hmm_range_free(struct amdgpu_hmm_range *range) {}
 #endif
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c
index f0765ccde668..9cb72f0c5277 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c
@@ -24,7 +24,6 @@
  *          Alex Deucher
  */
 
-#include <linux/export.h>
 #include <linux/pci.h>
 
 #include <drm/drm_edid.h>
@@ -185,7 +184,7 @@ struct amdgpu_i2c_chan *amdgpu_i2c_create(struct drm_device *dev,
 		snprintf(i2c->adapter.name, sizeof(i2c->adapter.name),
 			 "AMDGPU i2c hw bus %s", name);
 		i2c->adapter.algo = &amdgpu_atombios_i2c_algo;
-		ret = i2c_add_adapter(&i2c->adapter);
+		ret = devm_i2c_add_adapter(dev->dev, &i2c->adapter);
 		if (ret)
 			goto out_free;
 	} else {
@@ -216,13 +215,23 @@ out_free:
 
 }
 
-void amdgpu_i2c_destroy(struct amdgpu_i2c_chan *i2c)
+void amdgpu_i2c_init(struct amdgpu_device *adev)
 {
-	if (!i2c)
-		return;
-	WARN_ON(i2c->has_aux);
-	i2c_del_adapter(&i2c->adapter);
-	kfree(i2c);
+	if (!adev->is_atom_fw) {
+		if (!amdgpu_device_has_dc_support(adev)) {
+			amdgpu_atombios_i2c_init(adev);
+		} else {
+			switch (adev->asic_type) {
+			case CHIP_POLARIS10:
+			case CHIP_POLARIS11:
+			case CHIP_POLARIS12:
+				amdgpu_atombios_oem_i2c_init(adev, 0x97);
+				break;
+			default:
+				break;
+			}
+		}
+	}
 }
 
 /* remove all the buses */
@@ -230,12 +239,9 @@ void amdgpu_i2c_fini(struct amdgpu_device *adev)
 {
 	int i;
 
-	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++) {
-		if (adev->i2c_bus[i]) {
-			amdgpu_i2c_destroy(adev->i2c_bus[i]);
+	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++)
+		if (adev->i2c_bus[i])
 			adev->i2c_bus[i] = NULL;
-		}
-	}
 }
 
 /* looks up bus based on id */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h
index 21e3d1dad0a1..1d3d3806e0dd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h
@@ -28,6 +28,7 @@ struct amdgpu_i2c_chan *amdgpu_i2c_create(struct drm_device *dev,
 					  const struct amdgpu_i2c_bus_rec *rec,
 					  const char *name);
 void amdgpu_i2c_destroy(struct amdgpu_i2c_chan *i2c);
+void amdgpu_i2c_init(struct amdgpu_device *adev);
 void amdgpu_i2c_fini(struct amdgpu_device *adev);
 struct amdgpu_i2c_chan *
 amdgpu_i2c_lookup(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index 2ea98ec60220..586a58facca1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -128,6 +128,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_ib *ib = &ibs[0];
 	struct dma_fence *tmp = NULL;
+	struct amdgpu_fence *af;
 	bool need_ctx_switch;
 	struct amdgpu_vm *vm;
 	uint64_t fence_ctx;
@@ -138,7 +139,6 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
 	int vmid = AMDGPU_JOB_GET_VMID(job);
 	bool need_pipe_sync = false;
 	unsigned int cond_exec;
-
 	unsigned int i;
 	int r = 0;
 
@@ -149,11 +149,19 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
 	if (job) {
 		vm = job->vm;
 		fence_ctx = job->base.s_fence ?
-			job->base.s_fence->scheduled.context : 0;
+			job->base.s_fence->finished.context : 0;
 		shadow_va = job->shadow_va;
 		csa_va = job->csa_va;
 		gds_va = job->gds_va;
 		init_shadow = job->init_shadow;
+		af = job->hw_fence;
+		/* Save the context of the job for reset handling.
+		 * The driver needs this so it can skip the ring
+		 * contents for guilty contexts.
+		 */
+		af->context = fence_ctx;
+		/* the vm fence is also part of the job's context */
+		job->hw_vm_fence->context = fence_ctx;
 	} else {
 		vm = NULL;
 		fence_ctx = 0;
@@ -161,22 +169,28 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
 		csa_va = 0;
 		gds_va = 0;
 		init_shadow = false;
+		af = kzalloc(sizeof(*af), GFP_ATOMIC);
+		if (!af)
+			return -ENOMEM;
 	}
 
-	if (!ring->sched.ready && !ring->is_mes_queue) {
+	if (!ring->sched.ready) {
 		dev_err(adev->dev, "couldn't schedule ib on ring <%s>\n", ring->name);
-		return -EINVAL;
+		r = -EINVAL;
+		goto free_fence;
 	}
 
-	if (vm && !job->vmid && !ring->is_mes_queue) {
+	if (vm && !job->vmid) {
 		dev_err(adev->dev, "VM IB without ID\n");
-		return -EINVAL;
+		r = -EINVAL;
+		goto free_fence;
 	}
 
 	if ((ib->flags & AMDGPU_IB_FLAGS_SECURE) &&
 	    (!ring->funcs->secure_submission_supported)) {
 		dev_err(adev->dev, "secure submissions not supported on ring <%s>\n", ring->name);
-		return -EINVAL;
+		r = -EINVAL;
+		goto free_fence;
 	}
 
 	alloc_size = ring->funcs->emit_frame_size + num_ibs *
@@ -185,7 +199,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
 	r = amdgpu_ring_alloc(ring, alloc_size);
 	if (r) {
 		dev_err(adev->dev, "scheduling IB failed (%d).\n", r);
-		return r;
+		goto free_fence;
 	}
 
 	need_ctx_switch = ring->current_ctx != fence_ctx;
@@ -282,7 +296,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
 		amdgpu_ring_init_cond_exec(ring, ring->cond_exe_gpu_addr);
 	}
 
-	r = amdgpu_fence_emit(ring, f, job, fence_flags);
+	r = amdgpu_fence_emit(ring, af, fence_flags);
 	if (r) {
 		dev_err(adev->dev, "failed to emit fence (%d)\n", r);
 		if (job && job->vmid)
@@ -290,6 +304,10 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
 		amdgpu_ring_undo(ring);
 		return r;
 	}
+	*f = &af->base;
+	/* get a ref for the job */
+	if (job)
+		dma_fence_get(*f);
 
 	if (ring->funcs->insert_end)
 		ring->funcs->insert_end(ring);
@@ -304,9 +322,23 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs,
 	    ring->hw_prio == AMDGPU_GFX_PIPE_PRIO_HIGH)
 		ring->funcs->emit_wave_limit(ring, false);
 
+	/* Save the wptr associated with this fence.
+	 * This must be last for resets to work properly
+	 * as we need to save the wptr associated with this
+	 * fence so we know what rings contents to backup
+	 * after we reset the queue.
+	 */
+	amdgpu_fence_save_wptr(af);
+
 	amdgpu_ring_ib_end(ring);
 	amdgpu_ring_commit(ring);
+
 	return 0;
+
+free_fence:
+	if (!job)
+		kfree(af);
+	return r;
 }
 
 /**
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
index 8e712a11aba5..9cab36322c16 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
@@ -201,58 +201,34 @@ static int amdgpu_vmid_grab_idle(struct amdgpu_ring *ring,
 	struct amdgpu_device *adev = ring->adev;
 	unsigned vmhub = ring->vm_hub;
 	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
-	struct dma_fence **fences;
-	unsigned i;
 
+	/* If anybody is waiting for a VMID let everybody wait for fairness */
 	if (!dma_fence_is_signaled(ring->vmid_wait)) {
 		*fence = dma_fence_get(ring->vmid_wait);
 		return 0;
 	}
 
-	fences = kmalloc_array(id_mgr->num_ids, sizeof(void *), GFP_KERNEL);
-	if (!fences)
-		return -ENOMEM;
-
 	/* Check if we have an idle VMID */
-	i = 0;
-	list_for_each_entry((*idle), &id_mgr->ids_lru, list) {
+	list_for_each_entry_reverse((*idle), &id_mgr->ids_lru, list) {
 		/* Don't use per engine and per process VMID at the same time */
 		struct amdgpu_ring *r = adev->vm_manager.concurrent_flush ?
 			NULL : ring;
 
-		fences[i] = amdgpu_sync_peek_fence(&(*idle)->active, r);
-		if (!fences[i])
-			break;
-		++i;
+		*fence = amdgpu_sync_peek_fence(&(*idle)->active, r);
+		if (!(*fence))
+			return 0;
 	}
 
-	/* If we can't find a idle VMID to use, wait till one becomes available */
-	if (&(*idle)->list == &id_mgr->ids_lru) {
-		u64 fence_context = adev->vm_manager.fence_context + ring->idx;
-		unsigned seqno = ++adev->vm_manager.seqno[ring->idx];
-		struct dma_fence_array *array;
-		unsigned j;
-
-		*idle = NULL;
-		for (j = 0; j < i; ++j)
-			dma_fence_get(fences[j]);
-
-		array = dma_fence_array_create(i, fences, fence_context,
-					       seqno, true);
-		if (!array) {
-			for (j = 0; j < i; ++j)
-				dma_fence_put(fences[j]);
-			kfree(fences);
-			return -ENOMEM;
-		}
-
-		*fence = dma_fence_get(&array->base);
-		dma_fence_put(ring->vmid_wait);
-		ring->vmid_wait = &array->base;
-		return 0;
-	}
-	kfree(fences);
+	/*
+	 * If we can't find a idle VMID to use, wait on a fence from the least
+	 * recently used in the hope that it will be available soon.
+	 */
+	*idle = NULL;
+	dma_fence_put(ring->vmid_wait);
+	ring->vmid_wait = dma_fence_get(*fence);
 
+	/* This is the reference we return */
+	dma_fence_get(*fence);
 	return 0;
 }
 
@@ -275,58 +251,45 @@ static int amdgpu_vmid_grab_reserved(struct amdgpu_vm *vm,
 {
 	struct amdgpu_device *adev = ring->adev;
 	unsigned vmhub = ring->vm_hub;
-	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
 	uint64_t fence_context = adev->fence_context + ring->idx;
 	bool needs_flush = vm->use_cpu_for_update;
 	uint64_t updates = amdgpu_vm_tlb_seq(vm);
 	int r;
 
-	*id = id_mgr->reserved;
+	*id = vm->reserved_vmid[vmhub];
 	if ((*id)->owner != vm->immediate.fence_context ||
 	    !amdgpu_vmid_compatible(*id, job) ||
 	    (*id)->flushed_updates < updates ||
 	    !(*id)->last_flush ||
 	    ((*id)->last_flush->context != fence_context &&
-	     !dma_fence_is_signaled((*id)->last_flush))) {
+	     !dma_fence_is_signaled((*id)->last_flush)))
+		needs_flush = true;
+
+	if ((*id)->owner != vm->immediate.fence_context ||
+	    (!adev->vm_manager.concurrent_flush && needs_flush)) {
 		struct dma_fence *tmp;
 
-		/* Wait for the gang to be assembled before using a
-		 * reserved VMID or otherwise the gang could deadlock.
+		/* Don't use per engine and per process VMID at the
+		 * same time
 		 */
-		tmp = amdgpu_device_get_gang(adev);
-		if (!dma_fence_is_signaled(tmp) && tmp != job->gang_submit) {
+		if (adev->vm_manager.concurrent_flush)
+			ring = NULL;
+
+		/* to prevent one context starved by another context */
+		(*id)->pd_gpu_addr = 0;
+		tmp = amdgpu_sync_peek_fence(&(*id)->active, ring);
+		if (tmp) {
 			*id = NULL;
-			*fence = tmp;
+			*fence = dma_fence_get(tmp);
 			return 0;
 		}
-		dma_fence_put(tmp);
-
-		/* Make sure the id is owned by the gang before proceeding */
-		if (!job->gang_submit ||
-		    (*id)->owner != vm->immediate.fence_context) {
-
-			/* Don't use per engine and per process VMID at the
-			 * same time
-			 */
-			if (adev->vm_manager.concurrent_flush)
-				ring = NULL;
-
-			/* to prevent one context starved by another context */
-			(*id)->pd_gpu_addr = 0;
-			tmp = amdgpu_sync_peek_fence(&(*id)->active, ring);
-			if (tmp) {
-				*id = NULL;
-				*fence = dma_fence_get(tmp);
-				return 0;
-			}
-		}
-		needs_flush = true;
 	}
 
 	/* Good we can use this VMID. Remember this submission as
 	* user of the VMID.
 	*/
-	r = amdgpu_sync_fence(&(*id)->active, &job->base.s_fence->finished);
+	r = amdgpu_sync_fence(&(*id)->active, &job->base.s_fence->finished,
+			      GFP_ATOMIC);
 	if (r)
 		return r;
 
@@ -385,7 +348,8 @@ static int amdgpu_vmid_grab_used(struct amdgpu_vm *vm,
 		 * user of the VMID.
 		 */
 		r = amdgpu_sync_fence(&(*id)->active,
-				      &job->base.s_fence->finished);
+				      &job->base.s_fence->finished,
+				      GFP_ATOMIC);
 		if (r)
 			return r;
 
@@ -422,7 +386,7 @@ int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 	if (r || !idle)
 		goto error;
 
-	if (amdgpu_vmid_uses_reserved(adev, vm, vmhub)) {
+	if (amdgpu_vmid_uses_reserved(vm, vmhub)) {
 		r = amdgpu_vmid_grab_reserved(vm, ring, job, &id, fence);
 		if (r || !id)
 			goto error;
@@ -437,7 +401,8 @@ int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 
 			/* Remember this submission as user of the VMID */
 			r = amdgpu_sync_fence(&id->active,
-					      &job->base.s_fence->finished);
+					      &job->base.s_fence->finished,
+					      GFP_ATOMIC);
 			if (r)
 				goto error;
 
@@ -474,55 +439,71 @@ error:
 
 /*
  * amdgpu_vmid_uses_reserved - check if a VM will use a reserved VMID
- * @adev: amdgpu_device pointer
  * @vm: the VM to check
  * @vmhub: the VMHUB which will be used
  *
  * Returns: True if the VM will use a reserved VMID.
  */
-bool amdgpu_vmid_uses_reserved(struct amdgpu_device *adev,
-			       struct amdgpu_vm *vm, unsigned int vmhub)
+bool amdgpu_vmid_uses_reserved(struct amdgpu_vm *vm, unsigned int vmhub)
 {
-	return vm->reserved_vmid[vmhub] ||
-		(adev->enforce_isolation[(vm->root.bo->xcp_id != AMDGPU_XCP_NO_PARTITION) ?
-					 vm->root.bo->xcp_id : 0] &&
-		 AMDGPU_IS_GFXHUB(vmhub));
+	return vm->reserved_vmid[vmhub];
 }
 
-int amdgpu_vmid_alloc_reserved(struct amdgpu_device *adev,
+/*
+ * amdgpu_vmid_alloc_reserved - reserve a specific VMID for this vm
+ * @adev: amdgpu device structure
+ * @vm: the VM to reserve an ID for
+ * @vmhub: the VMHUB which should be used
+ *
+ * Mostly used to have a reserved VMID for debugging and SPM.
+ *
+ * Returns: 0 for success, -ENOENT if an ID is already reserved.
+ */
+int amdgpu_vmid_alloc_reserved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			       unsigned vmhub)
 {
 	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct amdgpu_vmid *id;
+	int r = 0;
 
 	mutex_lock(&id_mgr->lock);
-
-	++id_mgr->reserved_use_count;
-	if (!id_mgr->reserved) {
-		struct amdgpu_vmid *id;
-
-		id = list_first_entry(&id_mgr->ids_lru, struct amdgpu_vmid,
-				      list);
-		/* Remove from normal round robin handling */
-		list_del_init(&id->list);
-		id_mgr->reserved = id;
+	if (vm->reserved_vmid[vmhub])
+		goto unlock;
+	if (id_mgr->reserved_vmid) {
+		r = -ENOENT;
+		goto unlock;
 	}
-
+	/* Remove from normal round robin handling */
+	id = list_first_entry(&id_mgr->ids_lru, struct amdgpu_vmid, list);
+	list_del_init(&id->list);
+	vm->reserved_vmid[vmhub] = id;
+	id_mgr->reserved_vmid = true;
 	mutex_unlock(&id_mgr->lock);
+
 	return 0;
+unlock:
+	mutex_unlock(&id_mgr->lock);
+	return r;
 }
 
-void amdgpu_vmid_free_reserved(struct amdgpu_device *adev,
+/*
+ * amdgpu_vmid_free_reserved - free up a reserved VMID again
+ * @adev: amdgpu device structure
+ * @vm: the VM with the reserved ID
+ * @vmhub: the VMHUB which should be used
+ */
+void amdgpu_vmid_free_reserved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			       unsigned vmhub)
 {
 	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
 
 	mutex_lock(&id_mgr->lock);
-	if (!--id_mgr->reserved_use_count) {
-		/* give the reserved ID back to normal round robin */
-		list_add(&id_mgr->reserved->list, &id_mgr->ids_lru);
-		id_mgr->reserved = NULL;
+	if (vm->reserved_vmid[vmhub]) {
+		list_add(&vm->reserved_vmid[vmhub]->list,
+			&id_mgr->ids_lru);
+		vm->reserved_vmid[vmhub] = NULL;
+		id_mgr->reserved_vmid = false;
 	}
-
 	mutex_unlock(&id_mgr->lock);
 }
 
@@ -589,10 +570,17 @@ void amdgpu_vmid_mgr_init(struct amdgpu_device *adev)
 
 		mutex_init(&id_mgr->lock);
 		INIT_LIST_HEAD(&id_mgr->ids_lru);
-		id_mgr->reserved_use_count = 0;
 
-		/* manage only VMIDs not used by KFD */
-		id_mgr->num_ids = adev->vm_manager.first_kfd_vmid;
+		/* for GC <10, SDMA uses MMHUB so use first_kfd_vmid for both GC and MM */
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(10, 0, 0))
+			/* manage only VMIDs not used by KFD */
+			id_mgr->num_ids = adev->vm_manager.first_kfd_vmid;
+		else if (AMDGPU_IS_MMHUB0(i) ||
+			 AMDGPU_IS_MMHUB1(i))
+			id_mgr->num_ids = 16;
+		else
+			/* manage only VMIDs not used by KFD */
+			id_mgr->num_ids = adev->vm_manager.first_kfd_vmid;
 
 		/* skip over VMID 0, since it is the system VM */
 		for (j = 1; j < id_mgr->num_ids; ++j) {
@@ -601,11 +589,6 @@ void amdgpu_vmid_mgr_init(struct amdgpu_device *adev)
 			list_add_tail(&id_mgr->ids[j].list, &id_mgr->ids_lru);
 		}
 	}
-	/* alloc a default reserved vmid to enforce isolation */
-	for (i = 0; i < (adev->xcp_mgr ? adev->xcp_mgr->num_xcps : 1); i++) {
-		if (adev->enforce_isolation[i])
-			amdgpu_vmid_alloc_reserved(adev, AMDGPU_GFXHUB(i));
-	}
 }
 
 /**
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h
index 4012fb2dd08a..b3649cd3af56 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.h
@@ -67,8 +67,7 @@ struct amdgpu_vmid_mgr {
 	unsigned		num_ids;
 	struct list_head	ids_lru;
 	struct amdgpu_vmid	ids[AMDGPU_NUM_VMID];
-	struct amdgpu_vmid	*reserved;
-	unsigned int		reserved_use_count;
+	bool			reserved_vmid;
 };
 
 int amdgpu_pasid_alloc(unsigned int bits);
@@ -78,12 +77,11 @@ void amdgpu_pasid_free_delayed(struct dma_resv *resv,
 
 bool amdgpu_vmid_had_gpu_reset(struct amdgpu_device *adev,
 			       struct amdgpu_vmid *id);
-bool amdgpu_vmid_uses_reserved(struct amdgpu_device *adev,
-			       struct amdgpu_vm *vm, unsigned int vmhub);
-int amdgpu_vmid_alloc_reserved(struct amdgpu_device *adev,
-				unsigned vmhub);
-void amdgpu_vmid_free_reserved(struct amdgpu_device *adev,
-				unsigned vmhub);
+bool amdgpu_vmid_uses_reserved(struct amdgpu_vm *vm, unsigned int vmhub);
+int amdgpu_vmid_alloc_reserved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			       unsigned vmhub);
+void amdgpu_vmid_free_reserved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			       unsigned vmhub);
 int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 		     struct amdgpu_job *job, struct dma_fence **fence);
 void amdgpu_vmid_reset(struct amdgpu_device *adev, unsigned vmhub,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
index 901f8b12c672..a6419246e9c2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
@@ -25,6 +25,7 @@
 
 #include "amdgpu.h"
 #include "amdgpu_ih.h"
+#include "amdgpu_reset.h"
 
 /**
  * amdgpu_ih_ring_init - initialize the IH state
@@ -217,7 +218,7 @@ int amdgpu_ih_process(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih)
 
 restart_ih:
 	count  = AMDGPU_IH_MAX_NUM_IVS;
-	DRM_DEBUG("%s: rptr %d, wptr %d\n", __func__, ih->rptr, wptr);
+	dev_dbg(adev->dev, "%s: rptr %d, wptr %d\n", __func__, ih->rptr, wptr);
 
 	/* Order reading of wptr vs. reading of IH ring data */
 	rmb();
@@ -227,13 +228,23 @@ restart_ih:
 		ih->rptr &= ih->ptr_mask;
 	}
 
-	amdgpu_ih_set_rptr(adev, ih);
+	if (!ih->overflow)
+		amdgpu_ih_set_rptr(adev, ih);
+
 	wake_up_all(&ih->wait_process);
 
 	/* make sure wptr hasn't changed while processing */
 	wptr = amdgpu_ih_get_wptr(adev, ih);
 	if (wptr != ih->rptr)
-		goto restart_ih;
+		if (!ih->overflow)
+			goto restart_ih;
+
+	if (ih->overflow)
+		if (amdgpu_sriov_runtime(adev))
+			WARN_ONCE(!amdgpu_reset_domain_schedule(adev->reset_domain,
+				   &adev->virt.flr_work),
+				  "Failed to queue work! at %s",
+				  __func__);
 
 	return IRQ_HANDLED;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
index 7d4395a5d8ac..f58b6be7fccc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
@@ -56,14 +56,14 @@ struct amdgpu_ih_ring {
 	bool			use_bus_addr;
 
 	struct amdgpu_bo	*ring_obj;
-	volatile uint32_t	*ring;
+	uint32_t		*ring;
 	uint64_t		gpu_addr;
 
 	uint64_t		wptr_addr;
-	volatile uint32_t	*wptr_cpu;
+	uint32_t		*wptr_cpu;
 
 	uint64_t		rptr_addr;
-	volatile uint32_t	*rptr_cpu;
+	uint32_t		*rptr_cpu;
 
 	bool                    enabled;
 	unsigned		rptr;
@@ -72,12 +72,16 @@ struct amdgpu_ih_ring {
 	/* For waiting on IH processing at checkpoint. */
 	wait_queue_head_t wait_process;
 	uint64_t		processed_timestamp;
+	bool overflow;
 };
 
 /* return true if time stamp t2 is after t1 with 48bit wrap around */
 #define amdgpu_ih_ts_after(t1, t2) \
 		(((int64_t)((t2) << 16) - (int64_t)((t1) << 16)) > 0LL)
 
+#define amdgpu_ih_ts_after_or_equal(t1, t2) \
+		(((int64_t)((t2) << 16) - (int64_t)((t1) << 16)) >= 0LL)
+
 /* provided by the ih block */
 struct amdgpu_ih_funcs {
 	/* ring read/write ptr handling, called from interrupt context */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.c
new file mode 100644
index 000000000000..99e1cf4fc955
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.c
@@ -0,0 +1,96 @@
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_ip.h"
+
+static int8_t amdgpu_logical_to_dev_inst(struct amdgpu_device *adev,
+					 enum amd_hw_ip_block_type block,
+					 int8_t inst)
+{
+	int8_t dev_inst;
+
+	switch (block) {
+	case GC_HWIP:
+	case SDMA0_HWIP:
+	/* Both JPEG and VCN as JPEG is only alias of VCN */
+	case VCN_HWIP:
+		dev_inst = adev->ip_map.dev_inst[block][inst];
+		break;
+	default:
+		/* For rest of the IPs, no look up required.
+		 * Assume 'logical instance == physical instance' for all configs. */
+		dev_inst = inst;
+		break;
+	}
+
+	return dev_inst;
+}
+
+static uint32_t amdgpu_logical_to_dev_mask(struct amdgpu_device *adev,
+					   enum amd_hw_ip_block_type block,
+					   uint32_t mask)
+{
+	uint32_t dev_mask = 0;
+	int8_t log_inst, dev_inst;
+
+	while (mask) {
+		log_inst = ffs(mask) - 1;
+		dev_inst = amdgpu_logical_to_dev_inst(adev, block, log_inst);
+		dev_mask |= (1 << dev_inst);
+		mask &= ~(1 << log_inst);
+	}
+
+	return dev_mask;
+}
+
+static void amdgpu_populate_ip_map(struct amdgpu_device *adev,
+				   enum amd_hw_ip_block_type ip_block,
+				   uint32_t inst_mask)
+{
+	int l = 0, i;
+
+	while (inst_mask) {
+		i = ffs(inst_mask) - 1;
+		adev->ip_map.dev_inst[ip_block][l++] = i;
+		inst_mask &= ~(1 << i);
+	}
+	for (; l < HWIP_MAX_INSTANCE; l++)
+		adev->ip_map.dev_inst[ip_block][l] = -1;
+}
+
+void amdgpu_ip_map_init(struct amdgpu_device *adev)
+{
+	u32 ip_map[][2] = {
+		{ GC_HWIP, adev->gfx.xcc_mask },
+		{ SDMA0_HWIP, adev->sdma.sdma_mask },
+		{ VCN_HWIP, adev->vcn.inst_mask },
+	};
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(ip_map); ++i)
+		amdgpu_populate_ip_map(adev, ip_map[i][0], ip_map[i][1]);
+
+	adev->ip_map.logical_to_dev_inst = amdgpu_logical_to_dev_inst;
+	adev->ip_map.logical_to_dev_mask = amdgpu_logical_to_dev_mask;
+}
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/ofa/ga102.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.h
index bea255529993..2490fd322aec 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/ofa/ga102.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.h
@@ -1,5 +1,5 @@
 /*
- * Copyright 2023 Red Hat Inc.
+ * Copyright 2025 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -18,27 +18,12 @@
  * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
  * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
  * OTHER DEALINGS IN THE SOFTWARE.
+ *
  */
-#include "priv.h"
-
-#include <subdev/gsp.h>
-
-#include <nvif/class.h>
 
-static const struct nvkm_engine_func
-ga102_ofa = {
-	.sclass = {
-		{ -1, -1, NVC7FA_VIDEO_OFA },
-		{}
-	}
-};
+#ifndef __AMDGPU_IP_H__
+#define __AMDGPU_IP_H__
 
-int
-ga102_ofa_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst,
-	      struct nvkm_engine **pengine)
-{
-	if (nvkm_gsp_rm(device->gsp))
-		return r535_ofa_new(&ga102_ofa, device, type, inst, pengine);
+void amdgpu_ip_map_init(struct amdgpu_device *adev);
 
-	return -ENODEV;
-}
+#endif /* __AMDGPU_IP_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
index 19ce4da285e8..8112ffc85995 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
@@ -142,8 +142,9 @@ void amdgpu_irq_disable_all(struct amdgpu_device *adev)
 				r = src->funcs->set(adev, src, k,
 						    AMDGPU_IRQ_STATE_DISABLE);
 				if (r)
-					DRM_ERROR("error disabling interrupt (%d)\n",
-						  r);
+					dev_err(adev->dev,
+						"error disabling interrupt (%d)\n",
+						r);
 			}
 		}
 	}
@@ -242,7 +243,7 @@ static bool amdgpu_msi_ok(struct amdgpu_device *adev)
 	return true;
 }
 
-static void amdgpu_restore_msix(struct amdgpu_device *adev)
+void amdgpu_restore_msix(struct amdgpu_device *adev)
 {
 	u16 ctrl;
 
@@ -315,7 +316,7 @@ int amdgpu_irq_init(struct amdgpu_device *adev)
 	adev->irq.irq = irq;
 	adev_to_drm(adev)->max_vblank_count = 0x00ffffff;
 
-	DRM_DEBUG("amdgpu: irq initialized.\n");
+	dev_dbg(adev->dev, "amdgpu: irq initialized.\n");
 	return 0;
 
 free_vectors:
@@ -461,10 +462,10 @@ void amdgpu_irq_dispatch(struct amdgpu_device *adev,
 	src_id = entry.src_id;
 
 	if (client_id >= AMDGPU_IRQ_CLIENTID_MAX) {
-		DRM_DEBUG("Invalid client_id in IV: %d\n", client_id);
+		dev_dbg(adev->dev, "Invalid client_id in IV: %d\n", client_id);
 
 	} else	if (src_id >= AMDGPU_MAX_IRQ_SRC_ID) {
-		DRM_DEBUG("Invalid src_id in IV: %d\n", src_id);
+		dev_dbg(adev->dev, "Invalid src_id in IV: %d\n", src_id);
 
 	} else if (((client_id == AMDGPU_IRQ_CLIENTID_LEGACY) ||
 		    (client_id == SOC15_IH_CLIENTID_ISP)) &&
@@ -472,18 +473,21 @@ void amdgpu_irq_dispatch(struct amdgpu_device *adev,
 		generic_handle_domain_irq(adev->irq.domain, src_id);
 
 	} else if (!adev->irq.client[client_id].sources) {
-		DRM_DEBUG("Unregistered interrupt client_id: %d src_id: %d\n",
-			  client_id, src_id);
+		dev_dbg(adev->dev,
+			"Unregistered interrupt client_id: %d src_id: %d\n",
+			client_id, src_id);
 
 	} else if ((src = adev->irq.client[client_id].sources[src_id])) {
 		r = src->funcs->process(adev, src, &entry);
 		if (r < 0)
-			DRM_ERROR("error processing interrupt (%d)\n", r);
+			dev_err(adev->dev, "error processing interrupt (%d)\n",
+				r);
 		else if (r)
 			handled = true;
 
 	} else {
-		DRM_DEBUG("Unregistered interrupt src_id: %d of client_id:%d\n",
+		dev_dbg(adev->dev,
+			"Unregistered interrupt src_id: %d of client_id:%d\n",
 			src_id, client_id);
 	}
 
@@ -619,6 +623,10 @@ int amdgpu_irq_get(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
 int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
 		   unsigned int type)
 {
+	/* When the threshold is reached,the interrupt source may not be enabled.return -EINVAL */
+	if (amdgpu_ras_is_rma(adev) && !amdgpu_irq_enabled(adev, src, type))
+		return -EINVAL;
+
 	if (!adev->irq.installed)
 		return -ENOENT;
 
@@ -725,10 +733,10 @@ static const struct irq_domain_ops amdgpu_hw_irqdomain_ops = {
  */
 int amdgpu_irq_add_domain(struct amdgpu_device *adev)
 {
-	adev->irq.domain = irq_domain_add_linear(NULL, AMDGPU_MAX_IRQ_SRC_ID,
-						 &amdgpu_hw_irqdomain_ops, adev);
+	adev->irq.domain = irq_domain_create_linear(NULL, AMDGPU_MAX_IRQ_SRC_ID,
+						    &amdgpu_hw_irqdomain_ops, adev);
 	if (!adev->irq.domain) {
-		DRM_ERROR("GPU irq add domain failed\n");
+		dev_err(adev->dev, "GPU irq add domain failed\n");
 		return -ENODEV;
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
index 04c0b4fa17a4..9f0417456abd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
@@ -146,5 +146,6 @@ void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev);
 int amdgpu_irq_add_domain(struct amdgpu_device *adev);
 void amdgpu_irq_remove_domain(struct amdgpu_device *adev);
 unsigned amdgpu_irq_create_mapping(struct amdgpu_device *adev, unsigned src_id);
+void amdgpu_restore_msix(struct amdgpu_device *adev);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c
index 732744488b03..37270c4dab8d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c
@@ -33,6 +33,8 @@
 #include "isp_v4_1_0.h"
 #include "isp_v4_1_1.h"
 
+#define ISP_MC_ADDR_ALIGN (1024 * 32)
+
 /**
  * isp_hw_init - start and test isp block
  *
@@ -124,7 +126,7 @@ static int isp_early_init(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-static bool isp_is_idle(void *handle)
+static bool isp_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
@@ -141,6 +143,181 @@ static int isp_set_powergating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
+static int is_valid_isp_device(struct device *isp_parent, struct device *amdgpu_dev)
+{
+	if (isp_parent != amdgpu_dev)
+		return -EINVAL;
+
+	return 0;
+}
+
+/**
+ * isp_user_buffer_alloc - create user buffer object (BO) for isp
+ *
+ * @dev: isp device handle
+ * @dmabuf: DMABUF handle for isp buffer allocated in system memory
+ * @buf_obj: GPU buffer object handle to initialize
+ * @buf_addr: GPU addr of the pinned BO to initialize
+ *
+ * Imports isp DMABUF to allocate and pin a user BO for isp internal use. It does
+ * GART alloc to generate GPU addr for BO to make it accessible through the
+ * GART aperture for ISP HW.
+ *
+ * This function is exported to allow the V4L2 isp device external to drm device
+ * to create and access the isp user BO.
+ *
+ * Returns:
+ * 0 on success, negative error code otherwise.
+ */
+int isp_user_buffer_alloc(struct device *dev, void *dmabuf,
+			  void **buf_obj, u64 *buf_addr)
+{
+	struct platform_device *ispdev = to_platform_device(dev);
+	const struct isp_platform_data *isp_pdata;
+	struct amdgpu_device *adev;
+	struct mfd_cell *mfd_cell;
+	struct amdgpu_bo *bo;
+	u64 gpu_addr;
+	int ret;
+
+	if (WARN_ON(!ispdev))
+		return -ENODEV;
+
+	if (WARN_ON(!buf_obj))
+		return -EINVAL;
+
+	if (WARN_ON(!buf_addr))
+		return -EINVAL;
+
+	mfd_cell = &ispdev->mfd_cell[0];
+	if (!mfd_cell)
+		return -ENODEV;
+
+	isp_pdata = mfd_cell->platform_data;
+	adev = isp_pdata->adev;
+
+	ret = is_valid_isp_device(ispdev->dev.parent, adev->dev);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_bo_create_isp_user(adev, dmabuf,
+					AMDGPU_GEM_DOMAIN_GTT, &bo, &gpu_addr);
+	if (ret) {
+		drm_err(&adev->ddev, "failed to alloc gart user buffer (%d)", ret);
+		return ret;
+	}
+
+	*buf_obj = (void *)bo;
+	*buf_addr = gpu_addr;
+
+	return 0;
+}
+EXPORT_SYMBOL(isp_user_buffer_alloc);
+
+/**
+ * isp_user_buffer_free - free isp user buffer object (BO)
+ *
+ * @buf_obj: amdgpu isp user BO to free
+ *
+ * unpin and unref BO for isp internal use.
+ *
+ * This function is exported to allow the V4L2 isp device
+ * external to drm device to free the isp user BO.
+ */
+void isp_user_buffer_free(void *buf_obj)
+{
+	amdgpu_bo_free_isp_user(buf_obj);
+}
+EXPORT_SYMBOL(isp_user_buffer_free);
+
+/**
+ * isp_kernel_buffer_alloc - create kernel buffer object (BO) for isp
+ *
+ * @dev: isp device handle
+ * @size: size for the new BO
+ * @buf_obj: GPU BO handle to initialize
+ * @gpu_addr: GPU addr of the pinned BO
+ * @cpu_addr: CPU address mapping of BO
+ *
+ * Allocates and pins a kernel BO for internal isp firmware use.
+ *
+ * This function is exported to allow the V4L2 isp device
+ * external to drm device to create and access the kernel BO.
+ *
+ * Returns:
+ * 0 on success, negative error code otherwise.
+ */
+int isp_kernel_buffer_alloc(struct device *dev, u64 size,
+			    void **buf_obj, u64 *gpu_addr, void **cpu_addr)
+{
+	struct platform_device *ispdev = to_platform_device(dev);
+	struct amdgpu_bo **bo = (struct amdgpu_bo **)buf_obj;
+	const struct isp_platform_data *isp_pdata;
+	struct amdgpu_device *adev;
+	struct mfd_cell *mfd_cell;
+	int ret;
+
+	if (WARN_ON(!ispdev))
+		return -ENODEV;
+
+	if (WARN_ON(!buf_obj))
+		return -EINVAL;
+
+	if (WARN_ON(!gpu_addr))
+		return -EINVAL;
+
+	if (WARN_ON(!cpu_addr))
+		return -EINVAL;
+
+	mfd_cell = &ispdev->mfd_cell[0];
+	if (!mfd_cell)
+		return -ENODEV;
+
+	isp_pdata = mfd_cell->platform_data;
+	adev = isp_pdata->adev;
+
+	ret = is_valid_isp_device(ispdev->dev.parent, adev->dev);
+	if (ret)
+		return ret;
+
+	/* Ensure *bo is NULL so a new BO will be created */
+	*bo = NULL;
+	ret = amdgpu_bo_create_kernel(adev,
+				      size,
+				      ISP_MC_ADDR_ALIGN,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      bo,
+				      gpu_addr,
+				      cpu_addr);
+	if (!cpu_addr || ret) {
+		drm_err(&adev->ddev, "failed to alloc gart kernel buffer (%d)", ret);
+		return ret;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(isp_kernel_buffer_alloc);
+
+/**
+ * isp_kernel_buffer_free - free isp kernel buffer object (BO)
+ *
+ * @buf_obj: amdgpu isp user BO to free
+ * @gpu_addr: GPU addr of isp kernel BO
+ * @cpu_addr: CPU addr of isp kernel BO
+ *
+ * unmaps and unpin a isp kernel BO.
+ *
+ * This function is exported to allow the V4L2 isp device
+ * external to drm device to free the kernel BO.
+ */
+void isp_kernel_buffer_free(void **buf_obj, u64 *gpu_addr, void **cpu_addr)
+{
+	struct amdgpu_bo **bo = (struct amdgpu_bo **)buf_obj;
+
+	amdgpu_bo_free_kernel(bo, gpu_addr, cpu_addr);
+}
+EXPORT_SYMBOL(isp_kernel_buffer_free);
+
 static const struct amd_ip_funcs isp_ip_funcs = {
 	.name = "isp_ip",
 	.early_init = isp_early_init,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h
index b03664c66dd6..d6f4ffa4c97c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h
@@ -28,16 +28,13 @@
 #ifndef __AMDGPU_ISP_H__
 #define __AMDGPU_ISP_H__
 
+#include <drm/amd/isp.h>
+#include <linux/pm_domain.h>
+
 #define ISP_REGS_OFFSET_END 0x629A4
 
 struct amdgpu_isp;
 
-struct isp_platform_data {
-	void *adev;
-	u32 asic_type;
-	resource_size_t base_rmmio_size;
-};
-
 struct isp_funcs {
 	int (*hw_init)(struct amdgpu_isp *isp);
 	int (*hw_fini)(struct amdgpu_isp *isp);
@@ -50,9 +47,11 @@ struct amdgpu_isp {
 	struct mfd_cell *isp_cell;
 	struct resource *isp_res;
 	struct resource *isp_i2c_res;
+	struct resource *isp_gpio_res;
 	struct isp_platform_data *isp_pdata;
 	unsigned int harvest_config;
 	const struct firmware	*fw;
+	struct generic_pm_domain ispgpd;
 };
 
 extern const struct amdgpu_ip_block_version isp_v4_1_0_ip_block;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
index 100f04475943..0a0dcbf0798d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@@ -89,10 +89,10 @@ static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job)
 {
 	struct amdgpu_ring *ring = to_amdgpu_ring(s_job->sched);
 	struct amdgpu_job *job = to_amdgpu_job(s_job);
-	struct amdgpu_task_info *ti;
+	struct drm_wedge_task_info *info = NULL;
+	struct amdgpu_task_info *ti = NULL;
 	struct amdgpu_device *adev = ring->adev;
-	int idx;
-	int r;
+	int idx, r;
 
 	if (!drm_dev_enter(adev_to_drm(adev), &idx)) {
 		dev_info(adev->dev, "%s - device unplugged skipping recovery on scheduler:%s",
@@ -112,6 +112,7 @@ static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job)
 		amdgpu_job_core_dump(adev, job);
 
 	if (amdgpu_gpu_recovery &&
+	    amdgpu_ring_is_reset_type_supported(ring, AMDGPU_RESET_TYPE_SOFT_RESET) &&
 	    amdgpu_ring_soft_recovery(ring, job->vmid, s_job->s_fence->parent)) {
 		dev_err(adev->dev, "ring %s timeout, but soft recovered\n",
 			s_job->sched->name);
@@ -124,36 +125,30 @@ static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job)
 
 	ti = amdgpu_vm_get_task_info_pasid(ring->adev, job->pasid);
 	if (ti) {
-		dev_err(adev->dev,
-			"Process information: process %s pid %d thread %s pid %d\n",
-			ti->process_name, ti->tgid, ti->task_name, ti->pid);
-		amdgpu_vm_put_task_info(ti);
+		amdgpu_vm_print_task_info(adev, ti);
+		info = &ti->task;
 	}
 
-	dma_fence_set_error(&s_job->s_fence->finished, -ETIME);
-
 	/* attempt a per ring reset */
 	if (amdgpu_gpu_recovery &&
+	    amdgpu_ring_is_reset_type_supported(ring, AMDGPU_RESET_TYPE_PER_QUEUE) &&
 	    ring->funcs->reset) {
-		dev_err(adev->dev, "Starting %s ring reset\n", s_job->sched->name);
-		/* stop the scheduler, but don't mess with the
-		 * bad job yet because if ring reset fails
-		 * we'll fall back to full GPU reset.
-		 */
-		drm_sched_wqueue_stop(&ring->sched);
-		r = amdgpu_ring_reset(ring, job->vmid);
+		dev_err(adev->dev, "Starting %s ring reset\n",
+			s_job->sched->name);
+		r = amdgpu_ring_reset(ring, job->vmid, job->hw_fence);
 		if (!r) {
-			if (amdgpu_ring_sched_ready(ring))
-				drm_sched_stop(&ring->sched, s_job);
 			atomic_inc(&ring->adev->gpu_reset_counter);
-			amdgpu_fence_driver_force_completion(ring);
-			if (amdgpu_ring_sched_ready(ring))
-				drm_sched_start(&ring->sched, 0);
+			dev_err(adev->dev, "Ring %s reset succeeded\n",
+				ring->sched.name);
+			drm_dev_wedged_event(adev_to_drm(adev),
+					     DRM_WEDGE_RECOVERY_NONE, info);
 			goto exit;
 		}
-		dev_err(adev->dev, "Ring %s reset failure\n", ring->sched.name);
+		dev_err(adev->dev, "Ring %s reset failed\n", ring->sched.name);
 	}
 
+	dma_fence_set_error(&s_job->s_fence->finished, -ETIME);
+
 	if (amdgpu_device_should_recover_gpu(ring->adev)) {
 		struct amdgpu_reset_context reset_context;
 		memset(&reset_context, 0, sizeof(reset_context));
@@ -179,14 +174,19 @@ static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job)
 	}
 
 exit:
+	amdgpu_vm_put_task_info(ti);
 	drm_dev_exit(idx);
-	return DRM_GPU_SCHED_STAT_NOMINAL;
+	return DRM_GPU_SCHED_STAT_RESET;
 }
 
 int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		     struct drm_sched_entity *entity, void *owner,
-		     unsigned int num_ibs, struct amdgpu_job **job)
+		     unsigned int num_ibs, struct amdgpu_job **job,
+		     u64 drm_client_id)
 {
+	struct amdgpu_fence *af;
+	int r;
+
 	if (num_ibs == 0)
 		return -EINVAL;
 
@@ -194,6 +194,20 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	if (!*job)
 		return -ENOMEM;
 
+	af = kzalloc(sizeof(struct amdgpu_fence), GFP_KERNEL);
+	if (!af) {
+		r = -ENOMEM;
+		goto err_job;
+	}
+	(*job)->hw_fence = af;
+
+	af = kzalloc(sizeof(struct amdgpu_fence), GFP_KERNEL);
+	if (!af) {
+		r = -ENOMEM;
+		goto err_fence;
+	}
+	(*job)->hw_vm_fence = af;
+
 	(*job)->vm = vm;
 
 	amdgpu_sync_create(&(*job)->explicit_sync);
@@ -203,17 +217,27 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	if (!entity)
 		return 0;
 
-	return drm_sched_job_init(&(*job)->base, entity, 1, owner);
+	return drm_sched_job_init(&(*job)->base, entity, 1, owner,
+				  drm_client_id);
+
+err_fence:
+	kfree((*job)->hw_fence);
+err_job:
+	kfree(*job);
+	*job = NULL;
+
+	return r;
 }
 
 int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev,
 			     struct drm_sched_entity *entity, void *owner,
 			     size_t size, enum amdgpu_ib_pool_type pool_type,
-			     struct amdgpu_job **job)
+			     struct amdgpu_job **job, u64 k_job_id)
 {
 	int r;
 
-	r = amdgpu_job_alloc(adev, NULL, entity, owner, 1, job);
+	r = amdgpu_job_alloc(adev, NULL, entity, owner, 1, job,
+			     k_job_id);
 	if (r)
 		return r;
 
@@ -222,7 +246,10 @@ int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev,
 	if (r) {
 		if (entity)
 			drm_sched_job_cleanup(&(*job)->base);
+		kfree((*job)->hw_vm_fence);
+		kfree((*job)->hw_fence);
 		kfree(*job);
+		*job = NULL;
 	}
 
 	return r;
@@ -250,11 +277,11 @@ void amdgpu_job_free_resources(struct amdgpu_job *job)
 	struct dma_fence *f;
 	unsigned i;
 
-	/* Check if any fences where initialized */
+	/* Check if any fences were initialized */
 	if (job->base.s_fence && job->base.s_fence->finished.ops)
 		f = &job->base.s_fence->finished;
-	else if (job->hw_fence.ops)
-		f = &job->hw_fence;
+	else if (job->hw_fence && job->hw_fence->base.ops)
+		f = &job->hw_fence->base;
 	else
 		f = NULL;
 
@@ -270,11 +297,16 @@ static void amdgpu_job_free_cb(struct drm_sched_job *s_job)
 
 	amdgpu_sync_free(&job->explicit_sync);
 
-	/* only put the hw fence if has embedded fence */
-	if (!job->hw_fence.ops)
-		kfree(job);
+	if (job->hw_fence->base.ops)
+		dma_fence_put(&job->hw_fence->base);
+	else
+		kfree(job->hw_fence);
+	if (job->hw_vm_fence->base.ops)
+		dma_fence_put(&job->hw_vm_fence->base);
 	else
-		dma_fence_put(&job->hw_fence);
+		kfree(job->hw_vm_fence);
+
+	kfree(job);
 }
 
 void amdgpu_job_set_gang_leader(struct amdgpu_job *job,
@@ -303,10 +335,16 @@ void amdgpu_job_free(struct amdgpu_job *job)
 	if (job->gang_submit != &job->base.s_fence->scheduled)
 		dma_fence_put(job->gang_submit);
 
-	if (!job->hw_fence.ops)
-		kfree(job);
+	if (job->hw_fence->base.ops)
+		dma_fence_put(&job->hw_fence->base);
+	else
+		kfree(job->hw_fence);
+	if (job->hw_vm_fence->base.ops)
+		dma_fence_put(&job->hw_vm_fence->base);
 	else
-		dma_fence_put(&job->hw_fence);
+		kfree(job->hw_vm_fence);
+
+	kfree(job);
 }
 
 struct dma_fence *amdgpu_job_submit(struct amdgpu_job *job)
@@ -342,32 +380,33 @@ amdgpu_job_prepare_job(struct drm_sched_job *sched_job,
 {
 	struct amdgpu_ring *ring = to_amdgpu_ring(s_entity->rq->sched);
 	struct amdgpu_job *job = to_amdgpu_job(sched_job);
-	struct dma_fence *fence = NULL;
+	struct dma_fence *fence;
 	int r;
 
 	r = drm_sched_entity_error(s_entity);
 	if (r)
 		goto error;
 
-	if (job->gang_submit)
+	if (job->gang_submit) {
 		fence = amdgpu_device_switch_gang(ring->adev, job->gang_submit);
+		if (fence)
+			return fence;
+	}
 
-	if (!fence && job->vm && !job->vmid) {
+	fence = amdgpu_device_enforce_isolation(ring->adev, ring, job);
+	if (fence)
+		return fence;
+
+	if (job->vm && !job->vmid) {
 		r = amdgpu_vmid_grab(job->vm, ring, job, &fence);
 		if (r) {
 			dev_err(ring->adev->dev, "Error getting VM ID (%d)\n", r);
 			goto error;
 		}
-		/*
-		 * The VM structure might be released after the VMID is
-		 * assigned, we had multiple problems with people trying to use
-		 * the VM pointer so better set it to NULL.
-		 */
-		if (!fence)
-			job->vm = NULL;
+		return fence;
 	}
 
-	return fence;
+	return NULL;
 
 error:
 	dma_fence_set_error(&job->base.s_fence->finished, r);
@@ -411,8 +450,24 @@ static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
 	return fence;
 }
 
-#define to_drm_sched_job(sched_job)		\
-		container_of((sched_job), struct drm_sched_job, queue_node)
+/*
+ * This is a duplicate function from DRM scheduler sched_internal.h.
+ * Plan is to remove it when amdgpu_job_stop_all_jobs_on_sched is removed, due
+ * latter being incorrect and racy.
+ *
+ * See https://lore.kernel.org/amd-gfx/44edde63-7181-44fb-a4f7-94e50514f539@amd.com/
+ */
+static struct drm_sched_job *
+drm_sched_entity_queue_pop(struct drm_sched_entity *entity)
+{
+	struct spsc_node *node;
+
+	node = spsc_queue_pop(&entity->job_queue);
+	if (!node)
+		return NULL;
+
+	return container_of(node, struct drm_sched_job, queue_node);
+}
 
 void amdgpu_job_stop_all_jobs_on_sched(struct drm_gpu_scheduler *sched)
 {
@@ -425,7 +480,7 @@ void amdgpu_job_stop_all_jobs_on_sched(struct drm_gpu_scheduler *sched)
 		struct drm_sched_rq *rq = sched->sched_rq[i];
 		spin_lock(&rq->lock);
 		list_for_each_entry(s_entity, &rq->entities, list) {
-			while ((s_job = to_drm_sched_job(spsc_queue_pop(&s_entity->job_queue)))) {
+			while ((s_job = drm_sched_entity_queue_pop(s_entity))) {
 				struct drm_sched_fence *s_fence = s_job->s_fence;
 
 				dma_fence_signal(&s_fence->scheduled);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
index ce6b9ba967ff..7abf069d17d4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
@@ -44,11 +44,28 @@
 struct amdgpu_fence;
 enum amdgpu_ib_pool_type;
 
+/* Internal kernel job ids. (decreasing values, starting from U64_MAX). */
+#define AMDGPU_KERNEL_JOB_ID_VM_UPDATE              (18446744073709551615ULL)
+#define AMDGPU_KERNEL_JOB_ID_VM_UPDATE_PDES         (18446744073709551614ULL)
+#define AMDGPU_KERNEL_JOB_ID_VM_UPDATE_RANGE        (18446744073709551613ULL)
+#define AMDGPU_KERNEL_JOB_ID_VM_PT_CLEAR            (18446744073709551612ULL)
+#define AMDGPU_KERNEL_JOB_ID_TTM_MAP_BUFFER         (18446744073709551611ULL)
+#define AMDGPU_KERNEL_JOB_ID_TTM_ACCESS_MEMORY_SDMA (18446744073709551610ULL)
+#define AMDGPU_KERNEL_JOB_ID_TTM_COPY_BUFFER        (18446744073709551609ULL)
+#define AMDGPU_KERNEL_JOB_ID_CLEAR_ON_RELEASE       (18446744073709551608ULL)
+#define AMDGPU_KERNEL_JOB_ID_MOVE_BLIT              (18446744073709551607ULL)
+#define AMDGPU_KERNEL_JOB_ID_TTM_CLEAR_BUFFER       (18446744073709551606ULL)
+#define AMDGPU_KERNEL_JOB_ID_CLEANER_SHADER         (18446744073709551605ULL)
+#define AMDGPU_KERNEL_JOB_ID_FLUSH_GPU_TLB          (18446744073709551604ULL)
+#define AMDGPU_KERNEL_JOB_ID_KFD_GART_MAP           (18446744073709551603ULL)
+#define AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST          (18446744073709551602ULL)
+
 struct amdgpu_job {
 	struct drm_sched_job    base;
 	struct amdgpu_vm	*vm;
 	struct amdgpu_sync	explicit_sync;
-	struct dma_fence	hw_fence;
+	struct amdgpu_fence	*hw_fence;
+	struct amdgpu_fence	*hw_vm_fence;
 	struct dma_fence	*gang_submit;
 	uint32_t		preamble_status;
 	uint32_t                preemption_status;
@@ -78,6 +95,7 @@ struct amdgpu_job {
 
 	/* enforce isolation */
 	bool			enforce_isolation;
+	bool			run_cleaner_shader;
 
 	uint32_t		num_ibs;
 	struct amdgpu_ib	ibs[];
@@ -90,11 +108,13 @@ static inline struct amdgpu_ring *amdgpu_job_ring(struct amdgpu_job *job)
 
 int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		     struct drm_sched_entity *entity, void *owner,
-		     unsigned int num_ibs, struct amdgpu_job **job);
+		     unsigned int num_ibs, struct amdgpu_job **job,
+		     u64 drm_client_id);
 int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev,
 			     struct drm_sched_entity *entity, void *owner,
 			     size_t size, enum amdgpu_ib_pool_type pool_type,
-			     struct amdgpu_job **job);
+			     struct amdgpu_job **job,
+			     u64 k_job_id);
 void amdgpu_job_set_resources(struct amdgpu_job *job, struct amdgpu_bo *gds,
 			      struct amdgpu_bo *gws, struct amdgpu_bo *oa);
 void amdgpu_job_free_resources(struct amdgpu_job *job);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
index b6d2eb049f54..63ee6ba6a931 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
@@ -33,6 +33,7 @@
 #define JPEG_IDLE_TIMEOUT	msecs_to_jiffies(1000)
 
 static void amdgpu_jpeg_idle_work_handler(struct work_struct *work);
+static void amdgpu_jpeg_reg_dump_fini(struct amdgpu_device *adev);
 
 int amdgpu_jpeg_sw_init(struct amdgpu_device *adev)
 {
@@ -85,6 +86,9 @@ int amdgpu_jpeg_sw_fini(struct amdgpu_device *adev)
 			amdgpu_ring_fini(&adev->jpeg.inst[i].ring_dec[j]);
 	}
 
+	if (adev->jpeg.reg_list)
+		amdgpu_jpeg_reg_dump_fini(adev);
+
 	mutex_destroy(&adev->jpeg.jpeg_pg_lock);
 
 	return 0;
@@ -117,10 +121,12 @@ static void amdgpu_jpeg_idle_work_handler(struct work_struct *work)
 			fences += amdgpu_fence_count_emitted(&adev->jpeg.inst[i].ring_dec[j]);
 	}
 
-	if (!fences && !atomic_read(&adev->jpeg.total_submission_cnt))
+	if (!fences && !atomic_read(&adev->jpeg.total_submission_cnt)) {
+		mutex_lock(&adev->jpeg.jpeg_pg_lock);
 		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_JPEG,
 						       AMD_PG_STATE_GATE);
-	else
+		mutex_unlock(&adev->jpeg.jpeg_pg_lock);
+	} else
 		schedule_delayed_work(&adev->jpeg.idle_work, JPEG_IDLE_TIMEOUT);
 }
 
@@ -190,7 +196,8 @@ static int amdgpu_jpeg_dec_set_reg(struct amdgpu_ring *ring, uint32_t handle,
 	int i, r;
 
 	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
-				     AMDGPU_IB_POOL_DIRECT, &job);
+				     AMDGPU_IB_POOL_DIRECT, &job,
+				     AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
@@ -364,7 +371,7 @@ static int amdgpu_debugfs_jpeg_sched_mask_set(void *data, u64 val)
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
 		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
 			ring = &adev->jpeg.inst[i].ring_dec[j];
-			if (val & (1 << ((i * adev->jpeg.num_jpeg_rings) + j)))
+			if (val & (BIT_ULL((i * adev->jpeg.num_jpeg_rings) + j)))
 				ring->sched.ready = true;
 			else
 				ring->sched.ready = false;
@@ -452,3 +459,149 @@ void amdgpu_jpeg_sysfs_reset_mask_fini(struct amdgpu_device *adev)
 			device_remove_file(adev->dev, &dev_attr_jpeg_reset_mask);
 	}
 }
+
+int amdgpu_jpeg_reg_dump_init(struct amdgpu_device *adev,
+			       const struct amdgpu_hwip_reg_entry *reg, u32 count)
+{
+	adev->jpeg.ip_dump = kcalloc(adev->jpeg.num_jpeg_inst * count,
+				     sizeof(uint32_t), GFP_KERNEL);
+	if (!adev->jpeg.ip_dump) {
+		dev_err(adev->dev,
+			"Failed to allocate memory for JPEG IP Dump\n");
+		return -ENOMEM;
+	}
+	adev->jpeg.reg_list = reg;
+	adev->jpeg.reg_count = count;
+
+	return 0;
+}
+
+static void amdgpu_jpeg_reg_dump_fini(struct amdgpu_device *adev)
+{
+	kfree(adev->jpeg.ip_dump);
+	adev->jpeg.reg_list = NULL;
+	adev->jpeg.reg_count = 0;
+}
+
+void amdgpu_jpeg_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 inst_off, inst_id, is_powered;
+	int i, j;
+
+	if (!adev->jpeg.ip_dump)
+		return;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+		if (adev->jpeg.harvest_config & (1 << i))
+			continue;
+
+		inst_id = GET_INST(JPEG, i);
+		inst_off = i * adev->jpeg.reg_count;
+		/* check power status from UVD_JPEG_POWER_STATUS */
+		adev->jpeg.ip_dump[inst_off] =
+			RREG32(SOC15_REG_ENTRY_OFFSET_INST(adev->jpeg.reg_list[0],
+							   inst_id));
+		is_powered = ((adev->jpeg.ip_dump[inst_off] & 0x1) != 1);
+
+		if (is_powered)
+			for (j = 1; j < adev->jpeg.reg_count; j++)
+				adev->jpeg.ip_dump[inst_off + j] =
+					RREG32(SOC15_REG_ENTRY_OFFSET_INST(adev->jpeg.reg_list[j],
+									   inst_id));
+	}
+}
+
+void amdgpu_jpeg_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	u32 inst_off, is_powered;
+	int i, j;
+
+	if (!adev->jpeg.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->jpeg.num_jpeg_inst);
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+		if (adev->jpeg.harvest_config & (1 << i)) {
+			drm_printf(p, "\nHarvested Instance:JPEG%d Skipping dump\n", i);
+			continue;
+		}
+
+		inst_off = i * adev->jpeg.reg_count;
+		is_powered = ((adev->jpeg.ip_dump[inst_off] & 0x1) != 1);
+
+		if (is_powered) {
+			drm_printf(p, "Active Instance:JPEG%d\n", i);
+			for (j = 0; j < adev->jpeg.reg_count; j++)
+				drm_printf(p, "%-50s \t 0x%08x\n", adev->jpeg.reg_list[j].reg_name,
+					   adev->jpeg.ip_dump[inst_off + j]);
+		} else
+			drm_printf(p, "\nInactive Instance:JPEG%d\n", i);
+	}
+}
+
+static inline bool amdgpu_jpeg_reg_valid(u32 reg)
+{
+	if (reg < JPEG_REG_RANGE_START || reg > JPEG_REG_RANGE_END ||
+	    (reg >= JPEG_ATOMIC_RANGE_START && reg <= JPEG_ATOMIC_RANGE_END))
+		return false;
+	else
+		return true;
+}
+
+/**
+ * amdgpu_jpeg_dec_parse_cs - command submission parser
+ *
+ * @parser: Command submission parser context
+ * @job: the job to parse
+ * @ib: the IB to parse
+ *
+ * Parse the command stream, return -EINVAL for invalid packet,
+ * 0 otherwise
+ */
+
+int amdgpu_jpeg_dec_parse_cs(struct amdgpu_cs_parser *parser,
+			      struct amdgpu_job *job,
+			      struct amdgpu_ib *ib)
+{
+	u32 i, reg, res, cond, type;
+	struct amdgpu_device *adev = parser->adev;
+
+	for (i = 0; i < ib->length_dw ; i += 2) {
+		reg  = CP_PACKETJ_GET_REG(ib->ptr[i]);
+		res  = CP_PACKETJ_GET_RES(ib->ptr[i]);
+		cond = CP_PACKETJ_GET_COND(ib->ptr[i]);
+		type = CP_PACKETJ_GET_TYPE(ib->ptr[i]);
+
+		if (res) /* only support 0 at the moment */
+			return -EINVAL;
+
+		switch (type) {
+		case PACKETJ_TYPE0:
+			if (cond != PACKETJ_CONDITION_CHECK0 ||
+			    !amdgpu_jpeg_reg_valid(reg)) {
+				dev_err(adev->dev, "Invalid packet [0x%08x]!\n", ib->ptr[i]);
+				return -EINVAL;
+			}
+			break;
+		case PACKETJ_TYPE3:
+			if (cond != PACKETJ_CONDITION_CHECK3 ||
+			    !amdgpu_jpeg_reg_valid(reg)) {
+				dev_err(adev->dev, "Invalid packet [0x%08x]!\n", ib->ptr[i]);
+				return -EINVAL;
+			}
+			break;
+		case PACKETJ_TYPE6:
+			if (ib->ptr[i] == CP_PACKETJ_NOP)
+				continue;
+			dev_err(adev->dev, "Invalid packet [0x%08x]!\n", ib->ptr[i]);
+			return -EINVAL;
+		default:
+			dev_err(adev->dev, "Unknown packet type %d !\n", type);
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
index d9cb343a8708..346ae0ab09d3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
@@ -25,11 +25,18 @@
 #define __AMDGPU_JPEG_H__
 
 #include "amdgpu_ras.h"
+#include "amdgpu_cs.h"
 
 #define AMDGPU_MAX_JPEG_INSTANCES	4
 #define AMDGPU_MAX_JPEG_RINGS           10
 #define AMDGPU_MAX_JPEG_RINGS_4_0_3     8
 
+#define JPEG_REG_RANGE_START            0x4000
+#define JPEG_REG_RANGE_END              0x41c2
+#define JPEG_ATOMIC_RANGE_START         0x4120
+#define JPEG_ATOMIC_RANGE_END           0x412A
+
+
 #define AMDGPU_JPEG_HARVEST_JPEG0 (1 << 0)
 #define AMDGPU_JPEG_HARVEST_JPEG1 (1 << 1)
 
@@ -92,6 +99,14 @@
 		*adev->jpeg.inst[inst_idx].dpg_sram_curr_addr++ = value;	\
 	} while (0)
 
+struct amdgpu_hwip_reg_entry;
+
+enum amdgpu_jpeg_caps {
+	AMDGPU_JPEG_RRMT_ENABLED,
+};
+
+#define AMDGPU_JPEG_CAPS(caps) BIT(AMDGPU_JPEG_##caps)
+
 struct amdgpu_jpeg_reg{
 	unsigned jpeg_pitch[AMDGPU_MAX_JPEG_RINGS];
 };
@@ -130,6 +145,10 @@ struct amdgpu_jpeg {
 	uint8_t num_inst_per_aid;
 	bool	indirect_sram;
 	uint32_t supported_reset;
+	uint32_t caps;
+	u32 *ip_dump;
+	u32 reg_count;
+	const struct amdgpu_hwip_reg_entry *reg_list;
 };
 
 int amdgpu_jpeg_sw_init(struct amdgpu_device *adev);
@@ -154,5 +173,12 @@ int amdgpu_jpeg_psp_update_sram(struct amdgpu_device *adev, int inst_idx,
 void amdgpu_debugfs_jpeg_sched_mask_init(struct amdgpu_device *adev);
 int amdgpu_jpeg_sysfs_reset_mask_init(struct amdgpu_device *adev);
 void amdgpu_jpeg_sysfs_reset_mask_fini(struct amdgpu_device *adev);
+int amdgpu_jpeg_reg_dump_init(struct amdgpu_device *adev,
+			       const struct amdgpu_hwip_reg_entry *reg, u32 count);
+void amdgpu_jpeg_dump_ip_state(struct amdgpu_ip_block *ip_block);
+void amdgpu_jpeg_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p);
+int amdgpu_jpeg_dec_parse_cs(struct amdgpu_cs_parser *parser,
+			     struct amdgpu_job *job,
+			     struct amdgpu_ib *ib);
 
 #endif /*__AMDGPU_JPEG_H__*/
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index 98528ee94c15..6ee77f431d56 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -45,6 +45,7 @@
 #include "amdgpu_ras.h"
 #include "amdgpu_reset.h"
 #include "amd_pcie.h"
+#include "amdgpu_userq.h"
 
 void amdgpu_unregister_gpu_instance(struct amdgpu_device *adev)
 {
@@ -90,7 +91,7 @@ void amdgpu_driver_unload_kms(struct drm_device *dev)
 	if (adev->rmmio == NULL)
 		return;
 
-	if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DRV_UNLOAD))
+	if (amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DRV_UNLOAD))
 		DRM_WARN("smart shift update failed\n");
 
 	amdgpu_acpi_fini(adev);
@@ -160,7 +161,7 @@ int amdgpu_driver_load_kms(struct amdgpu_device *adev, unsigned long flags)
 	if (acpi_status)
 		dev_dbg(dev->dev, "Error during ACPI methods call\n");
 
-	if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DRV_LOAD))
+	if (amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DRV_LOAD))
 		DRM_WARN("smart shift update failed\n");
 
 out:
@@ -370,6 +371,26 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 	return 0;
 }
 
+static int amdgpu_userq_metadata_info_gfx(struct amdgpu_device *adev,
+					  struct drm_amdgpu_info *info,
+					  struct drm_amdgpu_info_uq_metadata_gfx *meta)
+{
+	int ret = -EOPNOTSUPP;
+
+	if (adev->gfx.funcs->get_gfx_shadow_info) {
+		struct amdgpu_gfx_shadow_info shadow = {};
+
+		adev->gfx.funcs->get_gfx_shadow_info(adev, &shadow, true);
+		meta->shadow_size = shadow.shadow_size;
+		meta->shadow_alignment = shadow.shadow_alignment;
+		meta->csa_size = shadow.csa_size;
+		meta->csa_alignment = shadow.csa_alignment;
+		ret = 0;
+	}
+
+	return ret;
+}
+
 static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 			     struct drm_amdgpu_info *info,
 			     struct drm_amdgpu_info_hw_ip *result)
@@ -378,6 +399,7 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 	uint32_t ib_size_alignment = 0;
 	enum amd_ip_block_type type;
 	unsigned int num_rings = 0;
+	uint32_t num_slots = 0;
 	unsigned int i, j;
 
 	if (info->query_hw_ip.ip_instance >= AMDGPU_HW_IP_INSTANCE_MAX_COUNT)
@@ -387,24 +409,45 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 	case AMDGPU_HW_IP_GFX:
 		type = AMD_IP_BLOCK_TYPE_GFX;
 		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
-			if (adev->gfx.gfx_ring[i].sched.ready)
+			if (adev->gfx.gfx_ring[i].sched.ready &&
+			    !adev->gfx.gfx_ring[i].no_user_submission)
 				++num_rings;
+
+		if (!adev->gfx.disable_uq) {
+			for (i = 0; i < AMDGPU_MES_MAX_GFX_PIPES; i++)
+				num_slots += hweight32(adev->mes.gfx_hqd_mask[i]);
+		}
+
 		ib_start_alignment = 32;
 		ib_size_alignment = 32;
 		break;
 	case AMDGPU_HW_IP_COMPUTE:
 		type = AMD_IP_BLOCK_TYPE_GFX;
 		for (i = 0; i < adev->gfx.num_compute_rings; i++)
-			if (adev->gfx.compute_ring[i].sched.ready)
+			if (adev->gfx.compute_ring[i].sched.ready &&
+			    !adev->gfx.compute_ring[i].no_user_submission)
 				++num_rings;
+
+		if (!adev->sdma.disable_uq) {
+			for (i = 0; i < AMDGPU_MES_MAX_COMPUTE_PIPES; i++)
+				num_slots += hweight32(adev->mes.compute_hqd_mask[i]);
+		}
+
 		ib_start_alignment = 32;
 		ib_size_alignment = 32;
 		break;
 	case AMDGPU_HW_IP_DMA:
 		type = AMD_IP_BLOCK_TYPE_SDMA;
 		for (i = 0; i < adev->sdma.num_instances; i++)
-			if (adev->sdma.instance[i].ring.sched.ready)
+			if (adev->sdma.instance[i].ring.sched.ready &&
+			    !adev->sdma.instance[i].ring.no_user_submission)
 				++num_rings;
+
+		if (!adev->gfx.disable_uq) {
+			for (i = 0; i < AMDGPU_MES_MAX_SDMA_PIPES; i++)
+				num_slots += hweight32(adev->mes.sdma_hqd_mask[i]);
+		}
+
 		ib_start_alignment = 256;
 		ib_size_alignment = 4;
 		break;
@@ -414,7 +457,8 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 			if (adev->uvd.harvest_config & (1 << i))
 				continue;
 
-			if (adev->uvd.inst[i].ring.sched.ready)
+			if (adev->uvd.inst[i].ring.sched.ready &&
+			    !adev->uvd.inst[i].ring.no_user_submission)
 				++num_rings;
 		}
 		ib_start_alignment = 256;
@@ -423,7 +467,8 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 	case AMDGPU_HW_IP_VCE:
 		type = AMD_IP_BLOCK_TYPE_VCE;
 		for (i = 0; i < adev->vce.num_rings; i++)
-			if (adev->vce.ring[i].sched.ready)
+			if (adev->vce.ring[i].sched.ready &&
+			    !adev->vce.ring[i].no_user_submission)
 				++num_rings;
 		ib_start_alignment = 256;
 		ib_size_alignment = 4;
@@ -435,7 +480,8 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 				continue;
 
 			for (j = 0; j < adev->uvd.num_enc_rings; j++)
-				if (adev->uvd.inst[i].ring_enc[j].sched.ready)
+				if (adev->uvd.inst[i].ring_enc[j].sched.ready &&
+				    !adev->uvd.inst[i].ring_enc[j].no_user_submission)
 					++num_rings;
 		}
 		ib_start_alignment = 256;
@@ -447,7 +493,8 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 			if (adev->vcn.harvest_config & (1 << i))
 				continue;
 
-			if (adev->vcn.inst[i].ring_dec.sched.ready)
+			if (adev->vcn.inst[i].ring_dec.sched.ready &&
+			    !adev->vcn.inst[i].ring_dec.no_user_submission)
 				++num_rings;
 		}
 		ib_start_alignment = 256;
@@ -459,8 +506,9 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 			if (adev->vcn.harvest_config & (1 << i))
 				continue;
 
-			for (j = 0; j < adev->vcn.num_enc_rings; j++)
-				if (adev->vcn.inst[i].ring_enc[j].sched.ready)
+			for (j = 0; j < adev->vcn.inst[i].num_enc_rings; j++)
+				if (adev->vcn.inst[i].ring_enc[j].sched.ready &&
+				    !adev->vcn.inst[i].ring_enc[j].no_user_submission)
 					++num_rings;
 		}
 		ib_start_alignment = 256;
@@ -475,7 +523,8 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 				continue;
 
 			for (j = 0; j < adev->jpeg.num_jpeg_rings; j++)
-				if (adev->jpeg.inst[i].ring_dec[j].sched.ready)
+				if (adev->jpeg.inst[i].ring_dec[j].sched.ready &&
+				    !adev->jpeg.inst[i].ring_dec[j].no_user_submission)
 					++num_rings;
 		}
 		ib_start_alignment = 256;
@@ -483,7 +532,8 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 		break;
 	case AMDGPU_HW_IP_VPE:
 		type = AMD_IP_BLOCK_TYPE_VPE;
-		if (adev->vpe.ring.sched.ready)
+		if (adev->vpe.ring.sched.ready &&
+		    !adev->vpe.ring.no_user_submission)
 			++num_rings;
 		ib_start_alignment = 256;
 		ib_size_alignment = 4;
@@ -539,6 +589,7 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev,
 	}
 	result->capabilities_flags = 0;
 	result->available_rings = (1 << num_rings) - 1;
+	result->userq_num_slots = num_slots;
 	result->ib_start_alignment = ib_start_alignment;
 	result->ib_size_alignment = ib_size_alignment;
 	return 0;
@@ -707,7 +758,8 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		ui64 = atomic64_read(&adev->num_vram_cpu_page_faults);
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_VRAM_USAGE:
-		ui64 = ttm_resource_manager_usage(&adev->mman.vram_mgr.manager);
+		ui64 = ttm_resource_manager_used(&adev->mman.vram_mgr.manager) ?
+			ttm_resource_manager_usage(&adev->mman.vram_mgr.manager) : 0;
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_VIS_VRAM_USAGE:
 		ui64 = amdgpu_vram_mgr_vis_usage(&adev->mman.vram_mgr);
@@ -753,8 +805,8 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		mem.vram.usable_heap_size = adev->gmc.real_vram_size -
 			atomic64_read(&adev->vram_pin_size) -
 			AMDGPU_VM_RESERVED_VRAM;
-		mem.vram.heap_usage =
-			ttm_resource_manager_usage(vram_man);
+		mem.vram.heap_usage = ttm_resource_manager_used(&adev->mman.vram_mgr.manager) ?
+				ttm_resource_manager_usage(vram_man) : 0;
 		mem.vram.max_allocation = mem.vram.usable_heap_size * 3 / 4;
 
 		mem.cpu_accessible_vram.total_heap_size =
@@ -888,6 +940,19 @@ out:
 		if (adev->gfx.config.ta_cntl2_truncate_coord_mode)
 			dev_info->ids_flags |= AMDGPU_IDS_FLAGS_CONFORMANT_TRUNC_COORD;
 
+		/* Gang submit is not supported under SRIOV currently */
+		if (!amdgpu_sriov_vf(adev))
+			dev_info->ids_flags |= AMDGPU_IDS_FLAGS_GANG_SUBMIT;
+
+		if (amdgpu_passthrough(adev))
+			dev_info->ids_flags |= (AMDGPU_IDS_FLAGS_MODE_PT <<
+						AMDGPU_IDS_FLAGS_MODE_SHIFT) &
+						AMDGPU_IDS_FLAGS_MODE_MASK;
+		else if (amdgpu_sriov_vf(adev))
+			dev_info->ids_flags |= (AMDGPU_IDS_FLAGS_MODE_VF <<
+						AMDGPU_IDS_FLAGS_MODE_SHIFT) &
+						AMDGPU_IDS_FLAGS_MODE_MASK;
+
 		vm_size = adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
 		vm_size -= AMDGPU_VA_RESERVED_TOP;
 
@@ -969,6 +1034,8 @@ out:
 			}
 		}
 
+		dev_info->userq_ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+
 		ret = copy_to_user(out, dev_info,
 				   min((size_t)size, sizeof(*dev_info))) ? -EFAULT : 0;
 		kfree(dev_info);
@@ -1284,6 +1351,22 @@ out:
 		return copy_to_user(out, &gpuvm_fault,
 				    min((size_t)size, sizeof(gpuvm_fault))) ? -EFAULT : 0;
 	}
+	case AMDGPU_INFO_UQ_FW_AREAS: {
+		struct drm_amdgpu_info_uq_metadata meta_info = {};
+
+		switch (info->query_hw_ip.type) {
+		case AMDGPU_HW_IP_GFX:
+			ret = amdgpu_userq_metadata_info_gfx(adev, info, &meta_info.gfx);
+			if (ret)
+				return ret;
+
+			ret = copy_to_user(out, &meta_info,
+						min((size_t)size, sizeof(meta_info))) ? -EFAULT : 0;
+			return 0;
+		default:
+			return -EINVAL;
+		}
+	}
 	default:
 		DRM_DEBUG_KMS("Invalid request %d\n", info->query);
 		return -EINVAL;
@@ -1337,13 +1420,11 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 	if (r)
 		goto error_pasid;
 
-	r = amdgpu_vm_init(adev, &fpriv->vm, fpriv->xcp_id);
-	if (r)
-		goto error_pasid;
+	amdgpu_debugfs_vm_init(file_priv);
 
-	r = amdgpu_vm_set_pasid(adev, &fpriv->vm, pasid);
+	r = amdgpu_vm_init(adev, &fpriv->vm, fpriv->xcp_id, pasid);
 	if (r)
-		goto error_vm;
+		goto error_pasid;
 
 	fpriv->prt_va = amdgpu_vm_bo_add(adev, &fpriv->vm, NULL);
 	if (!fpriv->prt_va) {
@@ -1367,6 +1448,14 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 	mutex_init(&fpriv->bo_list_lock);
 	idr_init_base(&fpriv->bo_list_handles, 1);
 
+	r = amdgpu_userq_mgr_init(&fpriv->userq_mgr, file_priv, adev);
+	if (r)
+		DRM_WARN("Can't setup usermode queues, use legacy workload submission only\n");
+
+	r = amdgpu_eviction_fence_init(&fpriv->evf_mgr);
+	if (r)
+		goto error_vm;
+
 	amdgpu_ctx_mgr_init(&fpriv->ctx_mgr, adev);
 
 	file_priv->driver_priv = fpriv;
@@ -1376,15 +1465,12 @@ error_vm:
 	amdgpu_vm_fini(adev, &fpriv->vm);
 
 error_pasid:
-	if (pasid) {
+	if (pasid)
 		amdgpu_pasid_free(pasid);
-		amdgpu_vm_set_pasid(adev, &fpriv->vm, 0);
-	}
 
 	kfree(fpriv);
 
 out_suspend:
-	pm_runtime_mark_last_busy(dev->dev);
 pm_put:
 	pm_runtime_put_autosuspend(dev->dev);
 
@@ -1452,7 +1538,6 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 	kfree(fpriv);
 	file_priv->driver_priv = NULL;
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
index 32b27a1658e7..9c182ce501af 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
@@ -39,42 +39,6 @@ int amdgpu_mes_doorbell_process_slice(struct amdgpu_device *adev)
 		       PAGE_SIZE);
 }
 
-static int amdgpu_mes_kernel_doorbell_get(struct amdgpu_device *adev,
-					 int ip_type, uint64_t *doorbell_index)
-{
-	unsigned int offset, found;
-	struct amdgpu_mes *mes = &adev->mes;
-
-	if (ip_type == AMDGPU_RING_TYPE_SDMA)
-		offset = adev->doorbell_index.sdma_engine[0];
-	else
-		offset = 0;
-
-	found = find_next_zero_bit(mes->doorbell_bitmap, mes->num_mes_dbs, offset);
-	if (found >= mes->num_mes_dbs) {
-		DRM_WARN("No doorbell available\n");
-		return -ENOSPC;
-	}
-
-	set_bit(found, mes->doorbell_bitmap);
-
-	/* Get the absolute doorbell index on BAR */
-	*doorbell_index = mes->db_start_dw_offset + found * 2;
-	return 0;
-}
-
-static void amdgpu_mes_kernel_doorbell_free(struct amdgpu_device *adev,
-					   uint32_t doorbell_index)
-{
-	unsigned int old, rel_index;
-	struct amdgpu_mes *mes = &adev->mes;
-
-	/* Find the relative index of the doorbell in this object */
-	rel_index = (doorbell_index - mes->db_start_dw_offset) / 2;
-	old = test_and_clear_bit(rel_index, mes->doorbell_bitmap);
-	WARN_ON(!old);
-}
-
 static int amdgpu_mes_doorbell_init(struct amdgpu_device *adev)
 {
 	int i;
@@ -83,7 +47,7 @@ static int amdgpu_mes_doorbell_init(struct amdgpu_device *adev)
 	/* Bitmap for dynamic allocation of kernel doorbells */
 	mes->doorbell_bitmap = bitmap_zalloc(PAGE_SIZE / sizeof(u32), GFP_KERNEL);
 	if (!mes->doorbell_bitmap) {
-		DRM_ERROR("Failed to allocate MES doorbell bitmap\n");
+		dev_err(adev->dev, "Failed to allocate MES doorbell bitmap\n");
 		return -ENOMEM;
 	}
 
@@ -126,7 +90,7 @@ static void amdgpu_mes_doorbell_free(struct amdgpu_device *adev)
 
 int amdgpu_mes_init(struct amdgpu_device *adev)
 {
-	int i, r;
+	int i, r, num_pipes;
 
 	adev->mes.adev = adev;
 
@@ -141,28 +105,55 @@ int amdgpu_mes_init(struct amdgpu_device *adev)
 		spin_lock_init(&adev->mes.ring_lock[i]);
 
 	adev->mes.total_max_queue = AMDGPU_FENCE_MES_QUEUE_ID_MASK;
-	adev->mes.vmid_mask_mmhub = 0xffffff00;
-	adev->mes.vmid_mask_gfxhub = 0xffffff00;
+	adev->mes.vmid_mask_mmhub = 0xFF00;
+	adev->mes.vmid_mask_gfxhub = adev->gfx.disable_kq ? 0xFFFE : 0xFF00;
+
+	num_pipes = adev->gfx.me.num_pipe_per_me * adev->gfx.me.num_me;
+	if (num_pipes > AMDGPU_MES_MAX_GFX_PIPES)
+		dev_warn(adev->dev, "more gfx pipes than supported by MES! (%d vs %d)\n",
+			 num_pipes, AMDGPU_MES_MAX_GFX_PIPES);
+
+	for (i = 0; i < AMDGPU_MES_MAX_GFX_PIPES; i++) {
+		if (i >= num_pipes)
+			break;
+		if (amdgpu_ip_version(adev, GC_HWIP, 0) >=
+		    IP_VERSION(12, 0, 0))
+			/*
+			 * GFX V12 has only one GFX pipe, but 8 queues in it.
+			 * GFX pipe 0 queue 0 is being used by Kernel queue.
+			 * Set GFX pipe 0 queue 1-7 for MES scheduling
+			 * mask = 1111 1110b
+			 */
+			adev->mes.gfx_hqd_mask[i] = adev->gfx.disable_kq ? 0xFF : 0xFE;
+		else
+			/*
+			 * GFX pipe 0 queue 0 is being used by Kernel queue.
+			 * Set GFX pipe 0 queue 1 for MES scheduling
+			 * mask = 10b
+			 */
+			adev->mes.gfx_hqd_mask[i] = adev->gfx.disable_kq ? 0x3 : 0x2;
+	}
+
+	num_pipes = adev->gfx.mec.num_pipe_per_mec * adev->gfx.mec.num_mec;
+	if (num_pipes > AMDGPU_MES_MAX_COMPUTE_PIPES)
+		dev_warn(adev->dev, "more compute pipes than supported by MES! (%d vs %d)\n",
+			 num_pipes, AMDGPU_MES_MAX_COMPUTE_PIPES);
 
 	for (i = 0; i < AMDGPU_MES_MAX_COMPUTE_PIPES; i++) {
-		/* use only 1st MEC pipes */
-		if (i >= adev->gfx.mec.num_pipe_per_mec)
-			continue;
-		adev->mes.compute_hqd_mask[i] = 0xc;
+		if (i >= num_pipes)
+			break;
+		adev->mes.compute_hqd_mask[i] = adev->gfx.disable_kq ? 0xF : 0xC;
 	}
 
-	for (i = 0; i < AMDGPU_MES_MAX_GFX_PIPES; i++)
-		adev->mes.gfx_hqd_mask[i] = i ? 0 : 0xfffffffe;
+	num_pipes = adev->sdma.num_instances;
+	if (num_pipes > AMDGPU_MES_MAX_SDMA_PIPES)
+		dev_warn(adev->dev, "more SDMA pipes than supported by MES! (%d vs %d)\n",
+			 num_pipes, AMDGPU_MES_MAX_SDMA_PIPES);
 
 	for (i = 0; i < AMDGPU_MES_MAX_SDMA_PIPES; i++) {
-		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) <
-		    IP_VERSION(6, 0, 0))
-			adev->mes.sdma_hqd_mask[i] = i ? 0 : 0x3fc;
-		/* zero sdma_hqd_mask for non-existent engine */
-		else if (adev->sdma.num_instances == 1)
-			adev->mes.sdma_hqd_mask[i] = i ? 0 : 0xfc;
-		else
-			adev->mes.sdma_hqd_mask[i] = 0xfc;
+		if (i >= num_pipes)
+			break;
+		adev->mes.sdma_hqd_mask[i] = 0xfc;
 	}
 
 	for (i = 0; i < AMDGPU_MAX_MES_PIPES; i++) {
@@ -200,6 +191,20 @@ int amdgpu_mes_init(struct amdgpu_device *adev)
 	if (r)
 		goto error_doorbell;
 
+	if (adev->mes.hung_queue_db_array_size) {
+		r = amdgpu_bo_create_kernel(adev,
+					    adev->mes.hung_queue_db_array_size * sizeof(u32),
+					    PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_GTT,
+					    &adev->mes.hung_queue_db_array_gpu_obj,
+					    &adev->mes.hung_queue_db_array_gpu_addr,
+					    &adev->mes.hung_queue_db_array_cpu_addr);
+		if (r) {
+			dev_warn(adev->dev, "failed to create MES hung db array buffer (%d)", r);
+			goto error_doorbell;
+		}
+	}
+
 	return 0;
 
 error_doorbell:
@@ -225,6 +230,10 @@ void amdgpu_mes_fini(struct amdgpu_device *adev)
 {
 	int i;
 
+	amdgpu_bo_free_kernel(&adev->mes.hung_queue_db_array_gpu_obj,
+			      &adev->mes.hung_queue_db_array_gpu_addr,
+			      &adev->mes.hung_queue_db_array_cpu_addr);
+
 	amdgpu_bo_free_kernel(&adev->mes.event_log_gpu_obj,
 			      &adev->mes.event_log_gpu_addr,
 			      &adev->mes.event_log_cpu_addr);
@@ -246,244 +255,6 @@ void amdgpu_mes_fini(struct amdgpu_device *adev)
 	mutex_destroy(&adev->mes.mutex_hidden);
 }
 
-static void amdgpu_mes_queue_free_mqd(struct amdgpu_mes_queue *q)
-{
-	amdgpu_bo_free_kernel(&q->mqd_obj,
-			      &q->mqd_gpu_addr,
-			      &q->mqd_cpu_ptr);
-}
-
-int amdgpu_mes_create_process(struct amdgpu_device *adev, int pasid,
-			      struct amdgpu_vm *vm)
-{
-	struct amdgpu_mes_process *process;
-	int r;
-
-	/* allocate the mes process buffer */
-	process = kzalloc(sizeof(struct amdgpu_mes_process), GFP_KERNEL);
-	if (!process) {
-		DRM_ERROR("no more memory to create mes process\n");
-		return -ENOMEM;
-	}
-
-	/* allocate the process context bo and map it */
-	r = amdgpu_bo_create_kernel(adev, AMDGPU_MES_PROC_CTX_SIZE, PAGE_SIZE,
-				    AMDGPU_GEM_DOMAIN_GTT,
-				    &process->proc_ctx_bo,
-				    &process->proc_ctx_gpu_addr,
-				    &process->proc_ctx_cpu_ptr);
-	if (r) {
-		DRM_ERROR("failed to allocate process context bo\n");
-		goto clean_up_memory;
-	}
-	memset(process->proc_ctx_cpu_ptr, 0, AMDGPU_MES_PROC_CTX_SIZE);
-
-	/*
-	 * Avoid taking any other locks under MES lock to avoid circular
-	 * lock dependencies.
-	 */
-	amdgpu_mes_lock(&adev->mes);
-
-	/* add the mes process to idr list */
-	r = idr_alloc(&adev->mes.pasid_idr, process, pasid, pasid + 1,
-		      GFP_KERNEL);
-	if (r < 0) {
-		DRM_ERROR("failed to lock pasid=%d\n", pasid);
-		goto clean_up_ctx;
-	}
-
-	INIT_LIST_HEAD(&process->gang_list);
-	process->vm = vm;
-	process->pasid = pasid;
-	process->process_quantum = adev->mes.default_process_quantum;
-	process->pd_gpu_addr = amdgpu_bo_gpu_offset(vm->root.bo);
-
-	amdgpu_mes_unlock(&adev->mes);
-	return 0;
-
-clean_up_ctx:
-	amdgpu_mes_unlock(&adev->mes);
-	amdgpu_bo_free_kernel(&process->proc_ctx_bo,
-			      &process->proc_ctx_gpu_addr,
-			      &process->proc_ctx_cpu_ptr);
-clean_up_memory:
-	kfree(process);
-	return r;
-}
-
-void amdgpu_mes_destroy_process(struct amdgpu_device *adev, int pasid)
-{
-	struct amdgpu_mes_process *process;
-	struct amdgpu_mes_gang *gang, *tmp1;
-	struct amdgpu_mes_queue *queue, *tmp2;
-	struct mes_remove_queue_input queue_input;
-	unsigned long flags;
-	int r;
-
-	/*
-	 * Avoid taking any other locks under MES lock to avoid circular
-	 * lock dependencies.
-	 */
-	amdgpu_mes_lock(&adev->mes);
-
-	process = idr_find(&adev->mes.pasid_idr, pasid);
-	if (!process) {
-		DRM_WARN("pasid %d doesn't exist\n", pasid);
-		amdgpu_mes_unlock(&adev->mes);
-		return;
-	}
-
-	/* Remove all queues from hardware */
-	list_for_each_entry_safe(gang, tmp1, &process->gang_list, list) {
-		list_for_each_entry_safe(queue, tmp2, &gang->queue_list, list) {
-			spin_lock_irqsave(&adev->mes.queue_id_lock, flags);
-			idr_remove(&adev->mes.queue_id_idr, queue->queue_id);
-			spin_unlock_irqrestore(&adev->mes.queue_id_lock, flags);
-
-			queue_input.doorbell_offset = queue->doorbell_off;
-			queue_input.gang_context_addr = gang->gang_ctx_gpu_addr;
-
-			r = adev->mes.funcs->remove_hw_queue(&adev->mes,
-							     &queue_input);
-			if (r)
-				DRM_WARN("failed to remove hardware queue\n");
-		}
-
-		idr_remove(&adev->mes.gang_id_idr, gang->gang_id);
-	}
-
-	idr_remove(&adev->mes.pasid_idr, pasid);
-	amdgpu_mes_unlock(&adev->mes);
-
-	/* free all memory allocated by the process */
-	list_for_each_entry_safe(gang, tmp1, &process->gang_list, list) {
-		/* free all queues in the gang */
-		list_for_each_entry_safe(queue, tmp2, &gang->queue_list, list) {
-			amdgpu_mes_queue_free_mqd(queue);
-			list_del(&queue->list);
-			kfree(queue);
-		}
-		amdgpu_bo_free_kernel(&gang->gang_ctx_bo,
-				      &gang->gang_ctx_gpu_addr,
-				      &gang->gang_ctx_cpu_ptr);
-		list_del(&gang->list);
-		kfree(gang);
-
-	}
-	amdgpu_bo_free_kernel(&process->proc_ctx_bo,
-			      &process->proc_ctx_gpu_addr,
-			      &process->proc_ctx_cpu_ptr);
-	kfree(process);
-}
-
-int amdgpu_mes_add_gang(struct amdgpu_device *adev, int pasid,
-			struct amdgpu_mes_gang_properties *gprops,
-			int *gang_id)
-{
-	struct amdgpu_mes_process *process;
-	struct amdgpu_mes_gang *gang;
-	int r;
-
-	/* allocate the mes gang buffer */
-	gang = kzalloc(sizeof(struct amdgpu_mes_gang), GFP_KERNEL);
-	if (!gang) {
-		return -ENOMEM;
-	}
-
-	/* allocate the gang context bo and map it to cpu space */
-	r = amdgpu_bo_create_kernel(adev, AMDGPU_MES_GANG_CTX_SIZE, PAGE_SIZE,
-				    AMDGPU_GEM_DOMAIN_GTT,
-				    &gang->gang_ctx_bo,
-				    &gang->gang_ctx_gpu_addr,
-				    &gang->gang_ctx_cpu_ptr);
-	if (r) {
-		DRM_ERROR("failed to allocate process context bo\n");
-		goto clean_up_mem;
-	}
-	memset(gang->gang_ctx_cpu_ptr, 0, AMDGPU_MES_GANG_CTX_SIZE);
-
-	/*
-	 * Avoid taking any other locks under MES lock to avoid circular
-	 * lock dependencies.
-	 */
-	amdgpu_mes_lock(&adev->mes);
-
-	process = idr_find(&adev->mes.pasid_idr, pasid);
-	if (!process) {
-		DRM_ERROR("pasid %d doesn't exist\n", pasid);
-		r = -EINVAL;
-		goto clean_up_ctx;
-	}
-
-	/* add the mes gang to idr list */
-	r = idr_alloc(&adev->mes.gang_id_idr, gang, 1, 0,
-		      GFP_KERNEL);
-	if (r < 0) {
-		DRM_ERROR("failed to allocate idr for gang\n");
-		goto clean_up_ctx;
-	}
-
-	gang->gang_id = r;
-	*gang_id = r;
-
-	INIT_LIST_HEAD(&gang->queue_list);
-	gang->process = process;
-	gang->priority = gprops->priority;
-	gang->gang_quantum = gprops->gang_quantum ?
-		gprops->gang_quantum : adev->mes.default_gang_quantum;
-	gang->global_priority_level = gprops->global_priority_level;
-	gang->inprocess_gang_priority = gprops->inprocess_gang_priority;
-	list_add_tail(&gang->list, &process->gang_list);
-
-	amdgpu_mes_unlock(&adev->mes);
-	return 0;
-
-clean_up_ctx:
-	amdgpu_mes_unlock(&adev->mes);
-	amdgpu_bo_free_kernel(&gang->gang_ctx_bo,
-			      &gang->gang_ctx_gpu_addr,
-			      &gang->gang_ctx_cpu_ptr);
-clean_up_mem:
-	kfree(gang);
-	return r;
-}
-
-int amdgpu_mes_remove_gang(struct amdgpu_device *adev, int gang_id)
-{
-	struct amdgpu_mes_gang *gang;
-
-	/*
-	 * Avoid taking any other locks under MES lock to avoid circular
-	 * lock dependencies.
-	 */
-	amdgpu_mes_lock(&adev->mes);
-
-	gang = idr_find(&adev->mes.gang_id_idr, gang_id);
-	if (!gang) {
-		DRM_ERROR("gang id %d doesn't exist\n", gang_id);
-		amdgpu_mes_unlock(&adev->mes);
-		return -EINVAL;
-	}
-
-	if (!list_empty(&gang->queue_list)) {
-		DRM_ERROR("queue list is not empty\n");
-		amdgpu_mes_unlock(&adev->mes);
-		return -EBUSY;
-	}
-
-	idr_remove(&adev->mes.gang_id_idr, gang->gang_id);
-	list_del(&gang->list);
-	amdgpu_mes_unlock(&adev->mes);
-
-	amdgpu_bo_free_kernel(&gang->gang_ctx_bo,
-			      &gang->gang_ctx_gpu_addr,
-			      &gang->gang_ctx_cpu_ptr);
-
-	kfree(gang);
-
-	return 0;
-}
-
 int amdgpu_mes_suspend(struct amdgpu_device *adev)
 {
 	struct mes_suspend_gang_input input;
@@ -503,7 +274,7 @@ int amdgpu_mes_suspend(struct amdgpu_device *adev)
 	r = adev->mes.funcs->suspend_gang(&adev->mes, &input);
 	amdgpu_mes_unlock(&adev->mes);
 	if (r)
-		DRM_ERROR("failed to suspend all gangs");
+		dev_err(adev->dev, "failed to suspend all gangs");
 
 	return r;
 }
@@ -527,306 +298,8 @@ int amdgpu_mes_resume(struct amdgpu_device *adev)
 	r = adev->mes.funcs->resume_gang(&adev->mes, &input);
 	amdgpu_mes_unlock(&adev->mes);
 	if (r)
-		DRM_ERROR("failed to resume all gangs");
-
-	return r;
-}
-
-static int amdgpu_mes_queue_alloc_mqd(struct amdgpu_device *adev,
-				     struct amdgpu_mes_queue *q,
-				     struct amdgpu_mes_queue_properties *p)
-{
-	struct amdgpu_mqd *mqd_mgr = &adev->mqds[p->queue_type];
-	u32 mqd_size = mqd_mgr->mqd_size;
-	int r;
-
-	r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
-				    AMDGPU_GEM_DOMAIN_GTT,
-				    &q->mqd_obj,
-				    &q->mqd_gpu_addr, &q->mqd_cpu_ptr);
-	if (r) {
-		dev_warn(adev->dev, "failed to create queue mqd bo (%d)", r);
-		return r;
-	}
-	memset(q->mqd_cpu_ptr, 0, mqd_size);
-
-	r = amdgpu_bo_reserve(q->mqd_obj, false);
-	if (unlikely(r != 0))
-		goto clean_up;
-
-	return 0;
-
-clean_up:
-	amdgpu_bo_free_kernel(&q->mqd_obj,
-			      &q->mqd_gpu_addr,
-			      &q->mqd_cpu_ptr);
-	return r;
-}
-
-static void amdgpu_mes_queue_init_mqd(struct amdgpu_device *adev,
-				     struct amdgpu_mes_queue *q,
-				     struct amdgpu_mes_queue_properties *p)
-{
-	struct amdgpu_mqd *mqd_mgr = &adev->mqds[p->queue_type];
-	struct amdgpu_mqd_prop mqd_prop = {0};
-
-	mqd_prop.mqd_gpu_addr = q->mqd_gpu_addr;
-	mqd_prop.hqd_base_gpu_addr = p->hqd_base_gpu_addr;
-	mqd_prop.rptr_gpu_addr = p->rptr_gpu_addr;
-	mqd_prop.wptr_gpu_addr = p->wptr_gpu_addr;
-	mqd_prop.queue_size = p->queue_size;
-	mqd_prop.use_doorbell = true;
-	mqd_prop.doorbell_index = p->doorbell_off;
-	mqd_prop.eop_gpu_addr = p->eop_gpu_addr;
-	mqd_prop.hqd_pipe_priority = p->hqd_pipe_priority;
-	mqd_prop.hqd_queue_priority = p->hqd_queue_priority;
-	mqd_prop.hqd_active = false;
-
-	if (p->queue_type == AMDGPU_RING_TYPE_GFX ||
-	    p->queue_type == AMDGPU_RING_TYPE_COMPUTE) {
-		mutex_lock(&adev->srbm_mutex);
-		amdgpu_gfx_select_me_pipe_q(adev, p->ring->me, p->ring->pipe, 0, 0, 0);
-	}
-
-	mqd_mgr->init_mqd(adev, q->mqd_cpu_ptr, &mqd_prop);
-
-	if (p->queue_type == AMDGPU_RING_TYPE_GFX ||
-	    p->queue_type == AMDGPU_RING_TYPE_COMPUTE) {
-		amdgpu_gfx_select_me_pipe_q(adev, 0, 0, 0, 0, 0);
-		mutex_unlock(&adev->srbm_mutex);
-	}
-
-	amdgpu_bo_unreserve(q->mqd_obj);
-}
-
-int amdgpu_mes_add_hw_queue(struct amdgpu_device *adev, int gang_id,
-			    struct amdgpu_mes_queue_properties *qprops,
-			    int *queue_id)
-{
-	struct amdgpu_mes_queue *queue;
-	struct amdgpu_mes_gang *gang;
-	struct mes_add_queue_input queue_input;
-	unsigned long flags;
-	int r;
-
-	memset(&queue_input, 0, sizeof(struct mes_add_queue_input));
-
-	/* allocate the mes queue buffer */
-	queue = kzalloc(sizeof(struct amdgpu_mes_queue), GFP_KERNEL);
-	if (!queue) {
-		DRM_ERROR("Failed to allocate memory for queue\n");
-		return -ENOMEM;
-	}
-
-	/* Allocate the queue mqd */
-	r = amdgpu_mes_queue_alloc_mqd(adev, queue, qprops);
-	if (r)
-		goto clean_up_memory;
-
-	/*
-	 * Avoid taking any other locks under MES lock to avoid circular
-	 * lock dependencies.
-	 */
-	amdgpu_mes_lock(&adev->mes);
-
-	gang = idr_find(&adev->mes.gang_id_idr, gang_id);
-	if (!gang) {
-		DRM_ERROR("gang id %d doesn't exist\n", gang_id);
-		r = -EINVAL;
-		goto clean_up_mqd;
-	}
-
-	/* add the mes gang to idr list */
-	spin_lock_irqsave(&adev->mes.queue_id_lock, flags);
-	r = idr_alloc(&adev->mes.queue_id_idr, queue, 1, 0,
-		      GFP_ATOMIC);
-	if (r < 0) {
-		spin_unlock_irqrestore(&adev->mes.queue_id_lock, flags);
-		goto clean_up_mqd;
-	}
-	spin_unlock_irqrestore(&adev->mes.queue_id_lock, flags);
-	*queue_id = queue->queue_id = r;
-
-	/* allocate a doorbell index for the queue */
-	r = amdgpu_mes_kernel_doorbell_get(adev,
-					  qprops->queue_type,
-					  &qprops->doorbell_off);
-	if (r)
-		goto clean_up_queue_id;
-
-	/* initialize the queue mqd */
-	amdgpu_mes_queue_init_mqd(adev, queue, qprops);
-
-	/* add hw queue to mes */
-	queue_input.process_id = gang->process->pasid;
-
-	queue_input.page_table_base_addr =
-		adev->vm_manager.vram_base_offset + gang->process->pd_gpu_addr -
-		adev->gmc.vram_start;
-
-	queue_input.process_va_start = 0;
-	queue_input.process_va_end =
-		(adev->vm_manager.max_pfn - 1) << AMDGPU_GPU_PAGE_SHIFT;
-	queue_input.process_quantum = gang->process->process_quantum;
-	queue_input.process_context_addr = gang->process->proc_ctx_gpu_addr;
-	queue_input.gang_quantum = gang->gang_quantum;
-	queue_input.gang_context_addr = gang->gang_ctx_gpu_addr;
-	queue_input.inprocess_gang_priority = gang->inprocess_gang_priority;
-	queue_input.gang_global_priority_level = gang->global_priority_level;
-	queue_input.doorbell_offset = qprops->doorbell_off;
-	queue_input.mqd_addr = queue->mqd_gpu_addr;
-	queue_input.wptr_addr = qprops->wptr_gpu_addr;
-	queue_input.wptr_mc_addr = qprops->wptr_mc_addr;
-	queue_input.queue_type = qprops->queue_type;
-	queue_input.paging = qprops->paging;
-	queue_input.is_kfd_process = 0;
-
-	r = adev->mes.funcs->add_hw_queue(&adev->mes, &queue_input);
-	if (r) {
-		DRM_ERROR("failed to add hardware queue to MES, doorbell=0x%llx\n",
-			  qprops->doorbell_off);
-		goto clean_up_doorbell;
-	}
-
-	DRM_DEBUG("MES hw queue was added, pasid=%d, gang id=%d, "
-		  "queue type=%d, doorbell=0x%llx\n",
-		  gang->process->pasid, gang_id, qprops->queue_type,
-		  qprops->doorbell_off);
-
-	queue->ring = qprops->ring;
-	queue->doorbell_off = qprops->doorbell_off;
-	queue->wptr_gpu_addr = qprops->wptr_gpu_addr;
-	queue->queue_type = qprops->queue_type;
-	queue->paging = qprops->paging;
-	queue->gang = gang;
-	queue->ring->mqd_ptr = queue->mqd_cpu_ptr;
-	list_add_tail(&queue->list, &gang->queue_list);
-
-	amdgpu_mes_unlock(&adev->mes);
-	return 0;
-
-clean_up_doorbell:
-	amdgpu_mes_kernel_doorbell_free(adev, qprops->doorbell_off);
-clean_up_queue_id:
-	spin_lock_irqsave(&adev->mes.queue_id_lock, flags);
-	idr_remove(&adev->mes.queue_id_idr, queue->queue_id);
-	spin_unlock_irqrestore(&adev->mes.queue_id_lock, flags);
-clean_up_mqd:
-	amdgpu_mes_unlock(&adev->mes);
-	amdgpu_mes_queue_free_mqd(queue);
-clean_up_memory:
-	kfree(queue);
-	return r;
-}
-
-int amdgpu_mes_remove_hw_queue(struct amdgpu_device *adev, int queue_id)
-{
-	unsigned long flags;
-	struct amdgpu_mes_queue *queue;
-	struct amdgpu_mes_gang *gang;
-	struct mes_remove_queue_input queue_input;
-	int r;
-
-	/*
-	 * Avoid taking any other locks under MES lock to avoid circular
-	 * lock dependencies.
-	 */
-	amdgpu_mes_lock(&adev->mes);
-
-	/* remove the mes gang from idr list */
-	spin_lock_irqsave(&adev->mes.queue_id_lock, flags);
-
-	queue = idr_find(&adev->mes.queue_id_idr, queue_id);
-	if (!queue) {
-		spin_unlock_irqrestore(&adev->mes.queue_id_lock, flags);
-		amdgpu_mes_unlock(&adev->mes);
-		DRM_ERROR("queue id %d doesn't exist\n", queue_id);
-		return -EINVAL;
-	}
-
-	idr_remove(&adev->mes.queue_id_idr, queue_id);
-	spin_unlock_irqrestore(&adev->mes.queue_id_lock, flags);
-
-	DRM_DEBUG("try to remove queue, doorbell off = 0x%llx\n",
-		  queue->doorbell_off);
-
-	gang = queue->gang;
-	queue_input.doorbell_offset = queue->doorbell_off;
-	queue_input.gang_context_addr = gang->gang_ctx_gpu_addr;
-
-	r = adev->mes.funcs->remove_hw_queue(&adev->mes, &queue_input);
-	if (r)
-		DRM_ERROR("failed to remove hardware queue, queue id = %d\n",
-			  queue_id);
-
-	list_del(&queue->list);
-	amdgpu_mes_kernel_doorbell_free(adev, queue->doorbell_off);
-	amdgpu_mes_unlock(&adev->mes);
-
-	amdgpu_mes_queue_free_mqd(queue);
-	kfree(queue);
-	return 0;
-}
-
-int amdgpu_mes_reset_hw_queue(struct amdgpu_device *adev, int queue_id)
-{
-	unsigned long flags;
-	struct amdgpu_mes_queue *queue;
-	struct amdgpu_mes_gang *gang;
-	struct mes_reset_queue_input queue_input;
-	int r;
-
-	/*
-	 * Avoid taking any other locks under MES lock to avoid circular
-	 * lock dependencies.
-	 */
-	amdgpu_mes_lock(&adev->mes);
-
-	/* remove the mes gang from idr list */
-	spin_lock_irqsave(&adev->mes.queue_id_lock, flags);
-
-	queue = idr_find(&adev->mes.queue_id_idr, queue_id);
-	if (!queue) {
-		spin_unlock_irqrestore(&adev->mes.queue_id_lock, flags);
-		amdgpu_mes_unlock(&adev->mes);
-		DRM_ERROR("queue id %d doesn't exist\n", queue_id);
-		return -EINVAL;
-	}
-	spin_unlock_irqrestore(&adev->mes.queue_id_lock, flags);
-
-	DRM_DEBUG("try to reset queue, doorbell off = 0x%llx\n",
-		  queue->doorbell_off);
-
-	gang = queue->gang;
-	queue_input.doorbell_offset = queue->doorbell_off;
-	queue_input.gang_context_addr = gang->gang_ctx_gpu_addr;
-
-	r = adev->mes.funcs->reset_hw_queue(&adev->mes, &queue_input);
-	if (r)
-		DRM_ERROR("failed to reset hardware queue, queue id = %d\n",
-			  queue_id);
+		dev_err(adev->dev, "failed to resume all gangs");
 
-	amdgpu_mes_unlock(&adev->mes);
-
-	return 0;
-}
-
-int amdgpu_mes_reset_hw_queue_mmio(struct amdgpu_device *adev, int queue_type,
-				   int me_id, int pipe_id, int queue_id, int vmid)
-{
-	struct mes_reset_queue_input queue_input;
-	int r;
-
-	queue_input.queue_type = queue_type;
-	queue_input.use_mmio = true;
-	queue_input.me_id = me_id;
-	queue_input.pipe_id = pipe_id;
-	queue_input.queue_id = queue_id;
-	queue_input.vmid = vmid;
-	r = adev->mes.funcs->reset_hw_queue(&adev->mes, &queue_input);
-	if (r)
-		DRM_ERROR("failed to reset hardware queue by mmio, queue id = %d\n",
-			  queue_id);
 	return r;
 }
 
@@ -845,9 +318,11 @@ int amdgpu_mes_map_legacy_queue(struct amdgpu_device *adev,
 	queue_input.mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
 	queue_input.wptr_addr = ring->wptr_gpu_addr;
 
+	amdgpu_mes_lock(&adev->mes);
 	r = adev->mes.funcs->map_legacy_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
 	if (r)
-		DRM_ERROR("failed to map legacy queue\n");
+		dev_err(adev->dev, "failed to map legacy queue\n");
 
 	return r;
 }
@@ -868,9 +343,11 @@ int amdgpu_mes_unmap_legacy_queue(struct amdgpu_device *adev,
 	queue_input.trail_fence_addr = gpu_addr;
 	queue_input.trail_fence_data = seq;
 
+	amdgpu_mes_lock(&adev->mes);
 	r = adev->mes.funcs->unmap_legacy_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
 	if (r)
-		DRM_ERROR("failed to unmap legacy queue\n");
+		dev_err(adev->dev, "failed to unmap legacy queue\n");
 
 	return r;
 }
@@ -880,7 +357,7 @@ int amdgpu_mes_reset_legacy_queue(struct amdgpu_device *adev,
 				  unsigned int vmid,
 				  bool use_mmio)
 {
-	struct mes_reset_legacy_queue_input queue_input;
+	struct mes_reset_queue_input queue_input;
 	int r;
 
 	memset(&queue_input, 0, sizeof(queue_input));
@@ -894,10 +371,67 @@ int amdgpu_mes_reset_legacy_queue(struct amdgpu_device *adev,
 	queue_input.wptr_addr = ring->wptr_gpu_addr;
 	queue_input.vmid = vmid;
 	queue_input.use_mmio = use_mmio;
+	queue_input.is_kq = true;
+	if (ring->funcs->type == AMDGPU_RING_TYPE_GFX)
+		queue_input.legacy_gfx = true;
 
-	r = adev->mes.funcs->reset_legacy_queue(&adev->mes, &queue_input);
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->reset_hw_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
 	if (r)
-		DRM_ERROR("failed to reset legacy queue\n");
+		dev_err(adev->dev, "failed to reset legacy queue\n");
+
+	return r;
+}
+
+int amdgpu_mes_get_hung_queue_db_array_size(struct amdgpu_device *adev)
+{
+	return adev->mes.hung_queue_db_array_size;
+}
+
+int amdgpu_mes_detect_and_reset_hung_queues(struct amdgpu_device *adev,
+					    int queue_type,
+					    bool detect_only,
+					    unsigned int *hung_db_num,
+					    u32 *hung_db_array)
+
+{
+	struct mes_detect_and_reset_queue_input input;
+	u32 *db_array = adev->mes.hung_queue_db_array_cpu_addr;
+	int r, i;
+
+	if (!hung_db_num || !hung_db_array)
+		return -EINVAL;
+
+	if ((queue_type != AMDGPU_RING_TYPE_GFX) &&
+	    (queue_type != AMDGPU_RING_TYPE_COMPUTE) &&
+	    (queue_type != AMDGPU_RING_TYPE_SDMA))
+		return -EINVAL;
+
+	/* Clear the doorbell array before detection */
+	memset(adev->mes.hung_queue_db_array_cpu_addr, AMDGPU_MES_INVALID_DB_OFFSET,
+		adev->mes.hung_queue_db_array_size * sizeof(u32));
+	input.queue_type = queue_type;
+	input.detect_only = detect_only;
+
+	r = adev->mes.funcs->detect_and_reset_hung_queues(&adev->mes,
+							  &input);
+	if (r) {
+		dev_err(adev->dev, "failed to detect and reset\n");
+	} else {
+		*hung_db_num = 0;
+		for (i = 0; i < adev->mes.hung_queue_hqd_info_offset; i++) {
+			if (db_array[i] != AMDGPU_MES_INVALID_DB_OFFSET) {
+				hung_db_array[i] = db_array[i];
+				*hung_db_num += 1;
+			}
+		}
+
+		/*
+		 * TODO: return HQD info for MES scheduled user compute queue reset cases
+		 * stored in hung_db_array hqd info offset to full array size
+		 */
+	}
 
 	return r;
 }
@@ -911,7 +445,7 @@ uint32_t amdgpu_mes_rreg(struct amdgpu_device *adev, uint32_t reg)
 	uint32_t *read_val_ptr;
 
 	if (amdgpu_device_wb_get(adev, &addr_offset)) {
-		DRM_ERROR("critical bug! too many mes readers\n");
+		dev_err(adev->dev, "critical bug! too many mes readers\n");
 		goto error;
 	}
 	read_val_gpu_addr = adev->wb.gpu_addr + (addr_offset * 4);
@@ -921,13 +455,15 @@ uint32_t amdgpu_mes_rreg(struct amdgpu_device *adev, uint32_t reg)
 	op_input.read_reg.buffer_addr = read_val_gpu_addr;
 
 	if (!adev->mes.funcs->misc_op) {
-		DRM_ERROR("mes rreg is not supported!\n");
+		dev_err(adev->dev, "mes rreg is not supported!\n");
 		goto error;
 	}
 
+	amdgpu_mes_lock(&adev->mes);
 	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
+	amdgpu_mes_unlock(&adev->mes);
 	if (r)
-		DRM_ERROR("failed to read reg (0x%x)\n", reg);
+		dev_err(adev->dev, "failed to read reg (0x%x)\n", reg);
 	else
 		val = *(read_val_ptr);
 
@@ -948,14 +484,16 @@ int amdgpu_mes_wreg(struct amdgpu_device *adev,
 	op_input.write_reg.reg_value = val;
 
 	if (!adev->mes.funcs->misc_op) {
-		DRM_ERROR("mes wreg is not supported!\n");
+		dev_err(adev->dev, "mes wreg is not supported!\n");
 		r = -EINVAL;
 		goto error;
 	}
 
+	amdgpu_mes_lock(&adev->mes);
 	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
+	amdgpu_mes_unlock(&adev->mes);
 	if (r)
-		DRM_ERROR("failed to write reg (0x%x)\n", reg);
+		dev_err(adev->dev, "failed to write reg (0x%x)\n", reg);
 
 error:
 	return r;
@@ -975,42 +513,31 @@ int amdgpu_mes_reg_write_reg_wait(struct amdgpu_device *adev,
 	op_input.wrm_reg.mask = mask;
 
 	if (!adev->mes.funcs->misc_op) {
-		DRM_ERROR("mes reg_write_reg_wait is not supported!\n");
+		dev_err(adev->dev, "mes reg_write_reg_wait is not supported!\n");
 		r = -EINVAL;
 		goto error;
 	}
 
+	amdgpu_mes_lock(&adev->mes);
 	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
+	amdgpu_mes_unlock(&adev->mes);
 	if (r)
-		DRM_ERROR("failed to reg_write_reg_wait\n");
+		dev_err(adev->dev, "failed to reg_write_reg_wait\n");
 
 error:
 	return r;
 }
 
-int amdgpu_mes_reg_wait(struct amdgpu_device *adev, uint32_t reg,
-			uint32_t val, uint32_t mask)
+int amdgpu_mes_hdp_flush(struct amdgpu_device *adev)
 {
-	struct mes_misc_op_input op_input;
-	int r;
-
-	op_input.op = MES_MISC_OP_WRM_REG_WAIT;
-	op_input.wrm_reg.reg0 = reg;
-	op_input.wrm_reg.ref = val;
-	op_input.wrm_reg.mask = mask;
-
-	if (!adev->mes.funcs->misc_op) {
-		DRM_ERROR("mes reg wait is not supported!\n");
-		r = -EINVAL;
-		goto error;
-	}
+	uint32_t hdp_flush_req_offset, hdp_flush_done_offset, ref_and_mask;
 
-	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
-	if (r)
-		DRM_ERROR("failed to reg_write_reg_wait\n");
+	hdp_flush_req_offset = adev->nbio.funcs->get_hdp_flush_req_offset(adev);
+	hdp_flush_done_offset = adev->nbio.funcs->get_hdp_flush_done_offset(adev);
+	ref_and_mask = adev->nbio.hdp_flush_reg->ref_and_mask_cp0;
 
-error:
-	return r;
+	return amdgpu_mes_reg_write_reg_wait(adev, hdp_flush_req_offset, hdp_flush_done_offset,
+					     ref_and_mask, ref_and_mask);
 }
 
 int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev,
@@ -1024,7 +551,8 @@ int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev,
 	int r;
 
 	if (!adev->mes.funcs->misc_op) {
-		DRM_ERROR("mes set shader debugger is not supported!\n");
+		dev_err(adev->dev,
+			"mes set shader debugger is not supported!\n");
 		return -EINVAL;
 	}
 
@@ -1048,7 +576,7 @@ int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev,
 
 	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
 	if (r)
-		DRM_ERROR("failed to set_shader_debugger\n");
+		dev_err(adev->dev, "failed to set_shader_debugger\n");
 
 	amdgpu_mes_unlock(&adev->mes);
 
@@ -1062,7 +590,8 @@ int amdgpu_mes_flush_shader_debugger(struct amdgpu_device *adev,
 	int r;
 
 	if (!adev->mes.funcs->misc_op) {
-		DRM_ERROR("mes flush shader debugger is not supported!\n");
+		dev_err(adev->dev,
+			"mes flush shader debugger is not supported!\n");
 		return -EINVAL;
 	}
 
@@ -1074,515 +603,19 @@ int amdgpu_mes_flush_shader_debugger(struct amdgpu_device *adev,
 
 	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
 	if (r)
-		DRM_ERROR("failed to set_shader_debugger\n");
+		dev_err(adev->dev, "failed to set_shader_debugger\n");
 
 	amdgpu_mes_unlock(&adev->mes);
 
 	return r;
 }
 
-static void
-amdgpu_mes_ring_to_queue_props(struct amdgpu_device *adev,
-			       struct amdgpu_ring *ring,
-			       struct amdgpu_mes_queue_properties *props)
-{
-	props->queue_type = ring->funcs->type;
-	props->hqd_base_gpu_addr = ring->gpu_addr;
-	props->rptr_gpu_addr = ring->rptr_gpu_addr;
-	props->wptr_gpu_addr = ring->wptr_gpu_addr;
-	props->wptr_mc_addr =
-		ring->mes_ctx->meta_data_mc_addr + ring->wptr_offs;
-	props->queue_size = ring->ring_size;
-	props->eop_gpu_addr = ring->eop_gpu_addr;
-	props->hqd_pipe_priority = AMDGPU_GFX_PIPE_PRIO_NORMAL;
-	props->hqd_queue_priority = AMDGPU_GFX_QUEUE_PRIORITY_MINIMUM;
-	props->paging = false;
-	props->ring = ring;
-}
-
-#define DEFINE_AMDGPU_MES_CTX_GET_OFFS_ENG(_eng)			\
-do {									\
-       if (id_offs < AMDGPU_MES_CTX_MAX_OFFS)				\
-		return offsetof(struct amdgpu_mes_ctx_meta_data,	\
-				_eng[ring->idx].slots[id_offs]);        \
-       else if (id_offs == AMDGPU_MES_CTX_RING_OFFS)			\
-		return offsetof(struct amdgpu_mes_ctx_meta_data,        \
-				_eng[ring->idx].ring);                  \
-       else if (id_offs == AMDGPU_MES_CTX_IB_OFFS)			\
-		return offsetof(struct amdgpu_mes_ctx_meta_data,        \
-				_eng[ring->idx].ib);                    \
-       else if (id_offs == AMDGPU_MES_CTX_PADDING_OFFS)			\
-		return offsetof(struct amdgpu_mes_ctx_meta_data,        \
-				_eng[ring->idx].padding);               \
-} while(0)
-
-int amdgpu_mes_ctx_get_offs(struct amdgpu_ring *ring, unsigned int id_offs)
-{
-	switch (ring->funcs->type) {
-	case AMDGPU_RING_TYPE_GFX:
-		DEFINE_AMDGPU_MES_CTX_GET_OFFS_ENG(gfx);
-		break;
-	case AMDGPU_RING_TYPE_COMPUTE:
-		DEFINE_AMDGPU_MES_CTX_GET_OFFS_ENG(compute);
-		break;
-	case AMDGPU_RING_TYPE_SDMA:
-		DEFINE_AMDGPU_MES_CTX_GET_OFFS_ENG(sdma);
-		break;
-	default:
-		break;
-	}
-
-	WARN_ON(1);
-	return -EINVAL;
-}
-
-int amdgpu_mes_add_ring(struct amdgpu_device *adev, int gang_id,
-			int queue_type, int idx,
-			struct amdgpu_mes_ctx_data *ctx_data,
-			struct amdgpu_ring **out)
-{
-	struct amdgpu_ring *ring;
-	struct amdgpu_mes_gang *gang;
-	struct amdgpu_mes_queue_properties qprops = {0};
-	int r, queue_id, pasid;
-
-	/*
-	 * Avoid taking any other locks under MES lock to avoid circular
-	 * lock dependencies.
-	 */
-	amdgpu_mes_lock(&adev->mes);
-	gang = idr_find(&adev->mes.gang_id_idr, gang_id);
-	if (!gang) {
-		DRM_ERROR("gang id %d doesn't exist\n", gang_id);
-		amdgpu_mes_unlock(&adev->mes);
-		return -EINVAL;
-	}
-	pasid = gang->process->pasid;
-
-	ring = kzalloc(sizeof(struct amdgpu_ring), GFP_KERNEL);
-	if (!ring) {
-		amdgpu_mes_unlock(&adev->mes);
-		return -ENOMEM;
-	}
-
-	ring->ring_obj = NULL;
-	ring->use_doorbell = true;
-	ring->is_mes_queue = true;
-	ring->mes_ctx = ctx_data;
-	ring->idx = idx;
-	ring->no_scheduler = true;
-
-	if (queue_type == AMDGPU_RING_TYPE_COMPUTE) {
-		int offset = offsetof(struct amdgpu_mes_ctx_meta_data,
-				      compute[ring->idx].mec_hpd);
-		ring->eop_gpu_addr =
-			amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
-	}
-
-	switch (queue_type) {
-	case AMDGPU_RING_TYPE_GFX:
-		ring->funcs = adev->gfx.gfx_ring[0].funcs;
-		ring->me = adev->gfx.gfx_ring[0].me;
-		ring->pipe = adev->gfx.gfx_ring[0].pipe;
-		break;
-	case AMDGPU_RING_TYPE_COMPUTE:
-		ring->funcs = adev->gfx.compute_ring[0].funcs;
-		ring->me = adev->gfx.compute_ring[0].me;
-		ring->pipe = adev->gfx.compute_ring[0].pipe;
-		break;
-	case AMDGPU_RING_TYPE_SDMA:
-		ring->funcs = adev->sdma.instance[0].ring.funcs;
-		break;
-	default:
-		BUG();
-	}
-
-	r = amdgpu_ring_init(adev, ring, 1024, NULL, 0,
-			     AMDGPU_RING_PRIO_DEFAULT, NULL);
-	if (r) {
-		amdgpu_mes_unlock(&adev->mes);
-		goto clean_up_memory;
-	}
-
-	amdgpu_mes_ring_to_queue_props(adev, ring, &qprops);
-
-	dma_fence_wait(gang->process->vm->last_update, false);
-	dma_fence_wait(ctx_data->meta_data_va->last_pt_update, false);
-	amdgpu_mes_unlock(&adev->mes);
-
-	r = amdgpu_mes_add_hw_queue(adev, gang_id, &qprops, &queue_id);
-	if (r)
-		goto clean_up_ring;
-
-	ring->hw_queue_id = queue_id;
-	ring->doorbell_index = qprops.doorbell_off;
-
-	if (queue_type == AMDGPU_RING_TYPE_GFX)
-		sprintf(ring->name, "gfx_%d.%d.%d", pasid, gang_id, queue_id);
-	else if (queue_type == AMDGPU_RING_TYPE_COMPUTE)
-		sprintf(ring->name, "compute_%d.%d.%d", pasid, gang_id,
-			queue_id);
-	else if (queue_type == AMDGPU_RING_TYPE_SDMA)
-		sprintf(ring->name, "sdma_%d.%d.%d", pasid, gang_id,
-			queue_id);
-	else
-		BUG();
-
-	*out = ring;
-	return 0;
-
-clean_up_ring:
-	amdgpu_ring_fini(ring);
-clean_up_memory:
-	kfree(ring);
-	return r;
-}
-
-void amdgpu_mes_remove_ring(struct amdgpu_device *adev,
-			    struct amdgpu_ring *ring)
-{
-	if (!ring)
-		return;
-
-	amdgpu_mes_remove_hw_queue(adev, ring->hw_queue_id);
-	del_timer_sync(&ring->fence_drv.fallback_timer);
-	amdgpu_ring_fini(ring);
-	kfree(ring);
-}
-
 uint32_t amdgpu_mes_get_aggregated_doorbell_index(struct amdgpu_device *adev,
 						   enum amdgpu_mes_priority_level prio)
 {
 	return adev->mes.aggregated_doorbells[prio];
 }
 
-int amdgpu_mes_ctx_alloc_meta_data(struct amdgpu_device *adev,
-				   struct amdgpu_mes_ctx_data *ctx_data)
-{
-	int r;
-
-	r = amdgpu_bo_create_kernel(adev,
-			    sizeof(struct amdgpu_mes_ctx_meta_data),
-			    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
-			    &ctx_data->meta_data_obj,
-			    &ctx_data->meta_data_mc_addr,
-			    &ctx_data->meta_data_ptr);
-	if (r) {
-		dev_warn(adev->dev, "(%d) create CTX bo failed\n", r);
-		return r;
-	}
-
-	if (!ctx_data->meta_data_obj)
-		return -ENOMEM;
-
-	memset(ctx_data->meta_data_ptr, 0,
-	       sizeof(struct amdgpu_mes_ctx_meta_data));
-
-	return 0;
-}
-
-void amdgpu_mes_ctx_free_meta_data(struct amdgpu_mes_ctx_data *ctx_data)
-{
-	if (ctx_data->meta_data_obj)
-		amdgpu_bo_free_kernel(&ctx_data->meta_data_obj,
-				      &ctx_data->meta_data_mc_addr,
-				      &ctx_data->meta_data_ptr);
-}
-
-int amdgpu_mes_ctx_map_meta_data(struct amdgpu_device *adev,
-				 struct amdgpu_vm *vm,
-				 struct amdgpu_mes_ctx_data *ctx_data)
-{
-	struct amdgpu_bo_va *bo_va;
-	struct amdgpu_sync sync;
-	struct drm_exec exec;
-	int r;
-
-	amdgpu_sync_create(&sync);
-
-	drm_exec_init(&exec, 0, 0);
-	drm_exec_until_all_locked(&exec) {
-		r = drm_exec_lock_obj(&exec,
-				      &ctx_data->meta_data_obj->tbo.base);
-		drm_exec_retry_on_contention(&exec);
-		if (unlikely(r))
-			goto error_fini_exec;
-
-		r = amdgpu_vm_lock_pd(vm, &exec, 0);
-		drm_exec_retry_on_contention(&exec);
-		if (unlikely(r))
-			goto error_fini_exec;
-	}
-
-	bo_va = amdgpu_vm_bo_add(adev, vm, ctx_data->meta_data_obj);
-	if (!bo_va) {
-		DRM_ERROR("failed to create bo_va for meta data BO\n");
-		r = -ENOMEM;
-		goto error_fini_exec;
-	}
-
-	r = amdgpu_vm_bo_map(adev, bo_va, ctx_data->meta_data_gpu_addr, 0,
-			     sizeof(struct amdgpu_mes_ctx_meta_data),
-			     AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE |
-			     AMDGPU_PTE_EXECUTABLE);
-
-	if (r) {
-		DRM_ERROR("failed to do bo_map on meta data, err=%d\n", r);
-		goto error_del_bo_va;
-	}
-
-	r = amdgpu_vm_bo_update(adev, bo_va, false);
-	if (r) {
-		DRM_ERROR("failed to do vm_bo_update on meta data\n");
-		goto error_del_bo_va;
-	}
-	amdgpu_sync_fence(&sync, bo_va->last_pt_update);
-
-	r = amdgpu_vm_update_pdes(adev, vm, false);
-	if (r) {
-		DRM_ERROR("failed to update pdes on meta data\n");
-		goto error_del_bo_va;
-	}
-	amdgpu_sync_fence(&sync, vm->last_update);
-
-	amdgpu_sync_wait(&sync, false);
-	drm_exec_fini(&exec);
-
-	amdgpu_sync_free(&sync);
-	ctx_data->meta_data_va = bo_va;
-	return 0;
-
-error_del_bo_va:
-	amdgpu_vm_bo_del(adev, bo_va);
-
-error_fini_exec:
-	drm_exec_fini(&exec);
-	amdgpu_sync_free(&sync);
-	return r;
-}
-
-int amdgpu_mes_ctx_unmap_meta_data(struct amdgpu_device *adev,
-				   struct amdgpu_mes_ctx_data *ctx_data)
-{
-	struct amdgpu_bo_va *bo_va = ctx_data->meta_data_va;
-	struct amdgpu_bo *bo = ctx_data->meta_data_obj;
-	struct amdgpu_vm *vm = bo_va->base.vm;
-	struct dma_fence *fence;
-	struct drm_exec exec;
-	long r;
-
-	drm_exec_init(&exec, 0, 0);
-	drm_exec_until_all_locked(&exec) {
-		r = drm_exec_lock_obj(&exec,
-				      &ctx_data->meta_data_obj->tbo.base);
-		drm_exec_retry_on_contention(&exec);
-		if (unlikely(r))
-			goto out_unlock;
-
-		r = amdgpu_vm_lock_pd(vm, &exec, 0);
-		drm_exec_retry_on_contention(&exec);
-		if (unlikely(r))
-			goto out_unlock;
-	}
-
-	amdgpu_vm_bo_del(adev, bo_va);
-	if (!amdgpu_vm_ready(vm))
-		goto out_unlock;
-
-	r = dma_resv_get_singleton(bo->tbo.base.resv, DMA_RESV_USAGE_BOOKKEEP,
-				   &fence);
-	if (r)
-		goto out_unlock;
-	if (fence) {
-		amdgpu_bo_fence(bo, fence, true);
-		fence = NULL;
-	}
-
-	r = amdgpu_vm_clear_freed(adev, vm, &fence);
-	if (r || !fence)
-		goto out_unlock;
-
-	dma_fence_wait(fence, false);
-	amdgpu_bo_fence(bo, fence, true);
-	dma_fence_put(fence);
-
-out_unlock:
-	if (unlikely(r < 0))
-		dev_err(adev->dev, "failed to clear page tables (%ld)\n", r);
-	drm_exec_fini(&exec);
-
-	return r;
-}
-
-static int amdgpu_mes_test_create_gang_and_queues(struct amdgpu_device *adev,
-					  int pasid, int *gang_id,
-					  int queue_type, int num_queue,
-					  struct amdgpu_ring **added_rings,
-					  struct amdgpu_mes_ctx_data *ctx_data)
-{
-	struct amdgpu_ring *ring;
-	struct amdgpu_mes_gang_properties gprops = {0};
-	int r, j;
-
-	/* create a gang for the process */
-	gprops.priority = AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
-	gprops.gang_quantum = adev->mes.default_gang_quantum;
-	gprops.inprocess_gang_priority = AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
-	gprops.priority_level = AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
-	gprops.global_priority_level = AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
-
-	r = amdgpu_mes_add_gang(adev, pasid, &gprops, gang_id);
-	if (r) {
-		DRM_ERROR("failed to add gang\n");
-		return r;
-	}
-
-	/* create queues for the gang */
-	for (j = 0; j < num_queue; j++) {
-		r = amdgpu_mes_add_ring(adev, *gang_id, queue_type, j,
-					ctx_data, &ring);
-		if (r) {
-			DRM_ERROR("failed to add ring\n");
-			break;
-		}
-
-		DRM_INFO("ring %s was added\n", ring->name);
-		added_rings[j] = ring;
-	}
-
-	return 0;
-}
-
-static int amdgpu_mes_test_queues(struct amdgpu_ring **added_rings)
-{
-	struct amdgpu_ring *ring;
-	int i, r;
-
-	for (i = 0; i < AMDGPU_MES_CTX_MAX_RINGS; i++) {
-		ring = added_rings[i];
-		if (!ring)
-			continue;
-
-		r = amdgpu_ring_test_helper(ring);
-		if (r)
-			return r;
-
-		r = amdgpu_ring_test_ib(ring, 1000 * 10);
-		if (r) {
-			DRM_DEV_ERROR(ring->adev->dev,
-				      "ring %s ib test failed (%d)\n",
-				      ring->name, r);
-			return r;
-		} else
-			DRM_INFO("ring %s ib test pass\n", ring->name);
-	}
-
-	return 0;
-}
-
-int amdgpu_mes_self_test(struct amdgpu_device *adev)
-{
-	struct amdgpu_vm *vm = NULL;
-	struct amdgpu_mes_ctx_data ctx_data = {0};
-	struct amdgpu_ring *added_rings[AMDGPU_MES_CTX_MAX_RINGS] = { NULL };
-	int gang_ids[3] = {0};
-	int queue_types[][2] = { { AMDGPU_RING_TYPE_GFX, 1 },
-				 { AMDGPU_RING_TYPE_COMPUTE, 1 },
-				 { AMDGPU_RING_TYPE_SDMA, 1} };
-	int i, r, pasid, k = 0;
-
-	pasid = amdgpu_pasid_alloc(16);
-	if (pasid < 0) {
-		dev_warn(adev->dev, "No more PASIDs available!");
-		pasid = 0;
-	}
-
-	vm = kzalloc(sizeof(*vm), GFP_KERNEL);
-	if (!vm) {
-		r = -ENOMEM;
-		goto error_pasid;
-	}
-
-	r = amdgpu_vm_init(adev, vm, -1);
-	if (r) {
-		DRM_ERROR("failed to initialize vm\n");
-		goto error_pasid;
-	}
-
-	r = amdgpu_mes_ctx_alloc_meta_data(adev, &ctx_data);
-	if (r) {
-		DRM_ERROR("failed to alloc ctx meta data\n");
-		goto error_fini;
-	}
-
-	ctx_data.meta_data_gpu_addr = AMDGPU_VA_RESERVED_BOTTOM;
-	r = amdgpu_mes_ctx_map_meta_data(adev, vm, &ctx_data);
-	if (r) {
-		DRM_ERROR("failed to map ctx meta data\n");
-		goto error_vm;
-	}
-
-	r = amdgpu_mes_create_process(adev, pasid, vm);
-	if (r) {
-		DRM_ERROR("failed to create MES process\n");
-		goto error_vm;
-	}
-
-	for (i = 0; i < ARRAY_SIZE(queue_types); i++) {
-		/* On GFX v10.3, fw hasn't supported to map sdma queue. */
-		if (amdgpu_ip_version(adev, GC_HWIP, 0) >=
-			    IP_VERSION(10, 3, 0) &&
-		    amdgpu_ip_version(adev, GC_HWIP, 0) <
-			    IP_VERSION(11, 0, 0) &&
-		    queue_types[i][0] == AMDGPU_RING_TYPE_SDMA)
-			continue;
-
-		r = amdgpu_mes_test_create_gang_and_queues(adev, pasid,
-							   &gang_ids[i],
-							   queue_types[i][0],
-							   queue_types[i][1],
-							   &added_rings[k],
-							   &ctx_data);
-		if (r)
-			goto error_queues;
-
-		k += queue_types[i][1];
-	}
-
-	/* start ring test and ib test for MES queues */
-	amdgpu_mes_test_queues(added_rings);
-
-error_queues:
-	/* remove all queues */
-	for (i = 0; i < ARRAY_SIZE(added_rings); i++) {
-		if (!added_rings[i])
-			continue;
-		amdgpu_mes_remove_ring(adev, added_rings[i]);
-	}
-
-	for (i = 0; i < ARRAY_SIZE(gang_ids); i++) {
-		if (!gang_ids[i])
-			continue;
-		amdgpu_mes_remove_gang(adev, gang_ids[i]);
-	}
-
-	amdgpu_mes_destroy_process(adev, pasid);
-
-error_vm:
-	amdgpu_mes_ctx_unmap_meta_data(adev, &ctx_data);
-
-error_fini:
-	amdgpu_vm_fini(adev, vm);
-
-error_pasid:
-	if (pasid)
-		amdgpu_pasid_free(pasid);
-
-	amdgpu_mes_ctx_free_meta_data(&ctx_data);
-	kfree(vm);
-	return 0;
-}
-
 int amdgpu_mes_init_microcode(struct amdgpu_device *adev, int pipe)
 {
 	const struct mes_firmware_header_v1_0 *mes_hdr;
@@ -1670,18 +703,16 @@ out:
 bool amdgpu_mes_suspend_resume_all_supported(struct amdgpu_device *adev)
 {
 	uint32_t mes_rev = adev->mes.sched_version & AMDGPU_MES_VERSION_MASK;
-	bool is_supported = false;
 
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0) &&
-	    amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(12, 0, 0) &&
-	    mes_rev >= 0x63)
-		is_supported = true;
-
-	return is_supported;
+	return ((amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0) &&
+		 amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(12, 0, 0) &&
+		 mes_rev >= 0x63) ||
+		amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(12, 0, 0));
 }
 
 /* Fix me -- node_id is used to identify the correct MES instances in the future */
-int amdgpu_mes_set_enforce_isolation(struct amdgpu_device *adev, uint32_t node_id, bool enable)
+static int amdgpu_mes_set_enforce_isolation(struct amdgpu_device *adev,
+					    uint32_t node_id, bool enable)
 {
 	struct mes_misc_op_input op_input = {0};
 	int r;
@@ -1695,7 +726,9 @@ int amdgpu_mes_set_enforce_isolation(struct amdgpu_device *adev, uint32_t node_i
 		goto error;
 	}
 
+	amdgpu_mes_lock(&adev->mes);
 	r = adev->mes.funcs->misc_op(&adev->mes, &op_input);
+	amdgpu_mes_unlock(&adev->mes);
 	if (r)
 		dev_err(adev->dev, "failed to change_config.\n");
 
@@ -1703,6 +736,23 @@ error:
 	return r;
 }
 
+int amdgpu_mes_update_enforce_isolation(struct amdgpu_device *adev)
+{
+	int i, r = 0;
+
+	if (adev->enable_mes && adev->gfx.enable_cleaner_shader) {
+		mutex_lock(&adev->enforce_isolation_mutex);
+		for (i = 0; i < (adev->xcp_mgr ? adev->xcp_mgr->num_xcps : 1); i++) {
+			if (adev->enforce_isolation[i] == AMDGPU_ENFORCE_ISOLATION_ENABLE)
+				r |= amdgpu_mes_set_enforce_isolation(adev, i, true);
+			else
+				r |= amdgpu_mes_set_enforce_isolation(adev, i, false);
+		}
+		mutex_unlock(&adev->enforce_isolation_mutex);
+	}
+	return r;
+}
+
 #if defined(CONFIG_DEBUG_FS)
 
 static int amdgpu_debugfs_mes_event_log_show(struct seq_file *m, void *unused)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
index 2df2444ee892..e989225b354b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
@@ -41,6 +41,7 @@
 #define AMDGPU_MES_API_VERSION_MASK	0x00fff000
 #define AMDGPU_MES_FEAT_VERSION_MASK	0xff000000
 #define AMDGPU_MES_MSCRATCH_SIZE	0x40000
+#define AMDGPU_MES_INVALID_DB_OFFSET	0xffffffff
 
 enum amdgpu_mes_priority_level {
 	AMDGPU_MES_PRIORITY_LEVEL_LOW       = 0,
@@ -56,7 +57,7 @@ enum amdgpu_mes_priority_level {
 
 struct amdgpu_mes_funcs;
 
-enum admgpu_mes_pipe {
+enum amdgpu_mes_pipe {
 	AMDGPU_MES_SCHED_PIPE = 0,
 	AMDGPU_MES_KIQ_PIPE,
 	AMDGPU_MAX_MES_PIPES = 2,
@@ -111,8 +112,8 @@ struct amdgpu_mes {
 
 	uint32_t                        vmid_mask_gfxhub;
 	uint32_t                        vmid_mask_mmhub;
-	uint32_t                        compute_hqd_mask[AMDGPU_MES_MAX_COMPUTE_PIPES];
 	uint32_t                        gfx_hqd_mask[AMDGPU_MES_MAX_GFX_PIPES];
+	uint32_t                        compute_hqd_mask[AMDGPU_MES_MAX_COMPUTE_PIPES];
 	uint32_t                        sdma_hqd_mask[AMDGPU_MES_MAX_SDMA_PIPES];
 	uint32_t                        aggregated_doorbells[AMDGPU_MES_PRIORITY_NUM_LEVELS];
 	uint32_t                        sch_ctx_offs[AMDGPU_MAX_MES_PIPES];
@@ -143,23 +144,15 @@ struct amdgpu_mes {
 	const struct amdgpu_mes_funcs   *funcs;
 
 	/* mes resource_1 bo*/
-	struct amdgpu_bo    *resource_1;
-	uint64_t            resource_1_gpu_addr;
-	void                *resource_1_addr;
-
-};
-
-struct amdgpu_mes_process {
-	int			pasid;
-	struct			amdgpu_vm *vm;
-	uint64_t		pd_gpu_addr;
-	struct amdgpu_bo 	*proc_ctx_bo;
-	uint64_t 		proc_ctx_gpu_addr;
-	void 			*proc_ctx_cpu_ptr;
-	uint64_t 		process_quantum;
-	struct 			list_head gang_list;
-	uint32_t 		doorbell_index;
-	struct mutex		doorbell_lock;
+	struct amdgpu_bo    *resource_1[AMDGPU_MAX_MES_PIPES];
+	uint64_t            resource_1_gpu_addr[AMDGPU_MAX_MES_PIPES];
+	void                *resource_1_addr[AMDGPU_MAX_MES_PIPES];
+
+	int				hung_queue_db_array_size;
+	int				hung_queue_hqd_info_offset;
+	struct amdgpu_bo		*hung_queue_db_array_gpu_obj;
+	uint64_t			hung_queue_db_array_gpu_addr;
+	void				*hung_queue_db_array_cpu_addr;
 };
 
 struct amdgpu_mes_gang {
@@ -246,18 +239,7 @@ struct mes_add_queue_input {
 struct mes_remove_queue_input {
 	uint32_t	doorbell_offset;
 	uint64_t	gang_context_addr;
-};
-
-struct mes_reset_queue_input {
-	uint32_t	doorbell_offset;
-	uint64_t	gang_context_addr;
-	bool		use_mmio;
-	uint32_t	queue_type;
-	uint32_t	me_id;
-	uint32_t	pipe_id;
-	uint32_t	queue_id;
-	uint32_t	xcc_id;
-	uint32_t	vmid;
+	bool		remove_queue_after_reset;
 };
 
 struct mes_map_legacy_queue_input {
@@ -291,7 +273,7 @@ struct mes_resume_gang_input {
 	uint64_t	gang_context_addr;
 };
 
-struct mes_reset_legacy_queue_input {
+struct mes_reset_queue_input {
 	uint32_t                           queue_type;
 	uint32_t                           doorbell_offset;
 	bool                               use_mmio;
@@ -301,6 +283,20 @@ struct mes_reset_legacy_queue_input {
 	uint64_t                           mqd_addr;
 	uint64_t                           wptr_addr;
 	uint32_t                           vmid;
+	bool                               legacy_gfx;
+	bool                               is_kq;
+};
+
+struct mes_detect_and_reset_queue_input {
+	uint32_t                           queue_type;
+	bool                               detect_only;
+};
+
+struct mes_inv_tlbs_pasid_input {
+	uint32_t        xcc_id;
+	uint16_t        pasid;
+	uint8_t         hub_id;
+	uint8_t         flush_type;
 };
 
 enum mes_misc_opcode {
@@ -388,42 +384,27 @@ struct amdgpu_mes_funcs {
 	int (*misc_op)(struct amdgpu_mes *mes,
 		       struct mes_misc_op_input *input);
 
-	int (*reset_legacy_queue)(struct amdgpu_mes *mes,
-				  struct mes_reset_legacy_queue_input *input);
-
 	int (*reset_hw_queue)(struct amdgpu_mes *mes,
 			      struct mes_reset_queue_input *input);
+
+	int (*detect_and_reset_hung_queues)(struct amdgpu_mes *mes,
+			      struct mes_detect_and_reset_queue_input *input);
+
+
+	int (*invalidate_tlbs_pasid)(struct amdgpu_mes *mes,
+			      struct mes_inv_tlbs_pasid_input *input);
 };
 
 #define amdgpu_mes_kiq_hw_init(adev) (adev)->mes.kiq_hw_init((adev))
 #define amdgpu_mes_kiq_hw_fini(adev) (adev)->mes.kiq_hw_fini((adev))
 
-int amdgpu_mes_ctx_get_offs(struct amdgpu_ring *ring, unsigned int id_offs);
-
 int amdgpu_mes_init_microcode(struct amdgpu_device *adev, int pipe);
 int amdgpu_mes_init(struct amdgpu_device *adev);
 void amdgpu_mes_fini(struct amdgpu_device *adev);
 
-int amdgpu_mes_create_process(struct amdgpu_device *adev, int pasid,
-			      struct amdgpu_vm *vm);
-void amdgpu_mes_destroy_process(struct amdgpu_device *adev, int pasid);
-
-int amdgpu_mes_add_gang(struct amdgpu_device *adev, int pasid,
-			struct amdgpu_mes_gang_properties *gprops,
-			int *gang_id);
-int amdgpu_mes_remove_gang(struct amdgpu_device *adev, int gang_id);
-
 int amdgpu_mes_suspend(struct amdgpu_device *adev);
 int amdgpu_mes_resume(struct amdgpu_device *adev);
 
-int amdgpu_mes_add_hw_queue(struct amdgpu_device *adev, int gang_id,
-			    struct amdgpu_mes_queue_properties *qprops,
-			    int *queue_id);
-int amdgpu_mes_remove_hw_queue(struct amdgpu_device *adev, int queue_id);
-int amdgpu_mes_reset_hw_queue(struct amdgpu_device *adev, int queue_id);
-int amdgpu_mes_reset_hw_queue_mmio(struct amdgpu_device *adev, int queue_type,
-				   int me_id, int pipe_id, int queue_id, int vmid);
-
 int amdgpu_mes_map_legacy_queue(struct amdgpu_device *adev,
 				struct amdgpu_ring *ring);
 int amdgpu_mes_unmap_legacy_queue(struct amdgpu_device *adev,
@@ -435,14 +416,20 @@ int amdgpu_mes_reset_legacy_queue(struct amdgpu_device *adev,
 				  unsigned int vmid,
 				  bool use_mmio);
 
+int amdgpu_mes_get_hung_queue_db_array_size(struct amdgpu_device *adev);
+int amdgpu_mes_detect_and_reset_hung_queues(struct amdgpu_device *adev,
+					    int queue_type,
+					    bool detect_only,
+					    unsigned int *hung_db_num,
+					    u32 *hung_db_array);
+
 uint32_t amdgpu_mes_rreg(struct amdgpu_device *adev, uint32_t reg);
 int amdgpu_mes_wreg(struct amdgpu_device *adev,
 		    uint32_t reg, uint32_t val);
-int amdgpu_mes_reg_wait(struct amdgpu_device *adev, uint32_t reg,
-			uint32_t val, uint32_t mask);
 int amdgpu_mes_reg_write_reg_wait(struct amdgpu_device *adev,
 				  uint32_t reg0, uint32_t reg1,
 				  uint32_t ref, uint32_t mask);
+int amdgpu_mes_hdp_flush(struct amdgpu_device *adev);
 int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev,
 				uint64_t process_context_addr,
 				uint32_t spi_gdbg_per_vmid_cntl,
@@ -451,27 +438,10 @@ int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev,
 				bool trap_en);
 int amdgpu_mes_flush_shader_debugger(struct amdgpu_device *adev,
 				uint64_t process_context_addr);
-int amdgpu_mes_add_ring(struct amdgpu_device *adev, int gang_id,
-			int queue_type, int idx,
-			struct amdgpu_mes_ctx_data *ctx_data,
-			struct amdgpu_ring **out);
-void amdgpu_mes_remove_ring(struct amdgpu_device *adev,
-			    struct amdgpu_ring *ring);
 
 uint32_t amdgpu_mes_get_aggregated_doorbell_index(struct amdgpu_device *adev,
 						   enum amdgpu_mes_priority_level prio);
 
-int amdgpu_mes_ctx_alloc_meta_data(struct amdgpu_device *adev,
-				   struct amdgpu_mes_ctx_data *ctx_data);
-void amdgpu_mes_ctx_free_meta_data(struct amdgpu_mes_ctx_data *ctx_data);
-int amdgpu_mes_ctx_map_meta_data(struct amdgpu_device *adev,
-				 struct amdgpu_vm *vm,
-				 struct amdgpu_mes_ctx_data *ctx_data);
-int amdgpu_mes_ctx_unmap_meta_data(struct amdgpu_device *adev,
-				   struct amdgpu_mes_ctx_data *ctx_data);
-
-int amdgpu_mes_self_test(struct amdgpu_device *adev);
-
 int amdgpu_mes_doorbell_process_slice(struct amdgpu_device *adev);
 
 /*
@@ -534,6 +504,6 @@ static inline void amdgpu_mes_unlock(struct amdgpu_mes *mes)
 
 bool amdgpu_mes_suspend_resume_all_supported(struct amdgpu_device *adev);
 
-int amdgpu_mes_set_enforce_isolation(struct amdgpu_device *adev, uint32_t node_id, bool enable);
+int amdgpu_mes_update_enforce_isolation(struct amdgpu_device *adev);
 
 #endif /* __AMDGPU_MES_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
index 5e3faefc5510..dc8d2f52c7d6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
@@ -326,6 +326,8 @@ struct amdgpu_mode_info {
 	struct drm_property *audio_property;
 	/* FMT dithering */
 	struct drm_property *dither_property;
+	/* Adaptive Backlight Modulation (power feature) */
+	struct drm_property *abm_level_property;
 	/* hardcoded DFP edid from BIOS */
 	const struct drm_edid *bios_hardcoded_edid;
 
@@ -496,8 +498,6 @@ struct amdgpu_crtc {
 	struct drm_connector *connector;
 	/* for dpm */
 	u32 line_time;
-	u32 wm_low;
-	u32 wm_high;
 	u32 lb_vblank_lead_lines;
 	struct drm_display_mode hw_mode;
 	/* for virtual dce */
@@ -609,6 +609,7 @@ struct amdgpu_i2c_adapter {
 	struct i2c_adapter base;
 
 	struct ddc_service *ddc_service;
+	bool oem;
 };
 
 #define TO_DM_AUX(x) container_of((x), struct amdgpu_dm_dp_aux, aux)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c
index d085687a47ea..a974265837f0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c
@@ -53,6 +53,16 @@ u64 amdgpu_nbio_get_pcie_replay_count(struct amdgpu_device *adev)
 	return 0;
 }
 
+bool amdgpu_nbio_is_replay_cnt_supported(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev) || !adev->asic_funcs ||
+	    !adev->asic_funcs->get_pcie_replay_count ||
+	    (!adev->nbio.funcs || !adev->nbio.funcs->get_pcie_replay_count))
+		return false;
+
+	return true;
+}
+
 int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
 {
 	int r;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h
index 79c2f807b9fe..b528de6a01f6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h
@@ -119,4 +119,6 @@ int amdgpu_nbio_ras_sw_init(struct amdgpu_device *adev);
 int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block);
 u64 amdgpu_nbio_get_pcie_replay_count(struct amdgpu_device *adev);
 
+bool amdgpu_nbio_is_replay_cnt_supported(struct amdgpu_device *adev);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 96f4b8904e9a..e08f58de4b17 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -32,6 +32,7 @@
 #include <linux/list.h>
 #include <linux/slab.h>
 #include <linux/dma-buf.h>
+#include <linux/export.h>
 
 #include <drm/drm_drv.h>
 #include <drm/amdgpu_drm.h>
@@ -62,7 +63,7 @@ static void amdgpu_bo_destroy(struct ttm_buffer_object *tbo)
 
 	amdgpu_bo_kunmap(bo);
 
-	if (bo->tbo.base.import_attach)
+	if (drm_gem_is_imported(&bo->tbo.base))
 		drm_prime_gem_destroy(&bo->tbo.base, bo->tbo.sg);
 	drm_gem_object_release(&bo->tbo.base);
 	amdgpu_bo_unref(&bo->parent);
@@ -152,6 +153,14 @@ void amdgpu_bo_placement_from_domain(struct amdgpu_bo *abo, u32 domain)
 		c++;
 	}
 
+	if (domain & AMDGPU_GEM_DOMAIN_MMIO_REMAP) {
+		places[c].fpfn = 0;
+		places[c].lpfn = 0;
+		places[c].mem_type = AMDGPU_PL_MMIO_REMAP;
+		places[c].flags = 0;
+		c++;
+	}
+
 	if (domain & AMDGPU_GEM_DOMAIN_GTT) {
 		places[c].fpfn = 0;
 		places[c].lpfn = 0;
@@ -163,8 +172,8 @@ void amdgpu_bo_placement_from_domain(struct amdgpu_bo *abo, u32 domain)
 		 * When GTT is just an alternative to VRAM make sure that we
 		 * only use it as fallback and still try to fill up VRAM first.
 		 */
-		if (domain & abo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM &&
-		    !(adev->flags & AMD_IS_APU))
+		if (abo->tbo.resource && !(adev->flags & AMD_IS_APU) &&
+		    domain & abo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM)
 			places[c].flags |= TTM_PL_FLAG_FALLBACK;
 		c++;
 	}
@@ -351,7 +360,6 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
 
 	return 0;
 }
-EXPORT_SYMBOL(amdgpu_bo_create_kernel);
 
 /**
  * amdgpu_bo_create_isp_user - create user BO for isp
@@ -420,7 +428,6 @@ error_unreserve:
 
 	return r;
 }
-EXPORT_SYMBOL(amdgpu_bo_create_isp_user);
 
 /**
  * amdgpu_bo_create_kernel_at - create BO for kernel use at specific location
@@ -524,7 +531,6 @@ void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
 	if (cpu_addr)
 		*cpu_addr = NULL;
 }
-EXPORT_SYMBOL(amdgpu_bo_free_kernel);
 
 /**
  * amdgpu_bo_free_isp_user - free BO for isp use
@@ -547,7 +553,6 @@ void amdgpu_bo_free_isp_user(struct amdgpu_bo *bo)
 	}
 	amdgpu_bo_unref(&bo);
 }
-EXPORT_SYMBOL(amdgpu_bo_free_isp_user);
 
 /* Validate bo size is bit bigger than the request domain */
 static bool amdgpu_bo_validate_size(struct amdgpu_device *adev,
@@ -939,7 +944,7 @@ int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain)
 		domain = bo->preferred_domains & domain;
 
 	/* A shared bo cannot be migrated to VRAM */
-	if (bo->tbo.base.import_attach) {
+	if (drm_gem_is_imported(&bo->tbo.base)) {
 		if (domain & AMDGPU_GEM_DOMAIN_GTT)
 			domain = AMDGPU_GEM_DOMAIN_GTT;
 		else
@@ -967,7 +972,7 @@ int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain)
 	 */
 	domain = amdgpu_bo_get_preferred_domain(adev, domain);
 
-	if (bo->tbo.base.import_attach)
+	if (drm_gem_is_imported(&bo->tbo.base))
 		dma_buf_pin(bo->tbo.base.import_attach);
 
 	/* force to pin into visible video ram */
@@ -1018,7 +1023,7 @@ void amdgpu_bo_unpin(struct amdgpu_bo *bo)
 	if (bo->tbo.pin_count)
 		return;
 
-	if (bo->tbo.base.import_attach)
+	if (drm_gem_is_imported(&bo->tbo.base))
 		dma_buf_unpin(bo->tbo.base.import_attach);
 
 	if (bo->tbo.resource->mem_type == TTM_PL_VRAM) {
@@ -1044,7 +1049,8 @@ static const char * const amdgpu_vram_names[] = {
 	"GDDR6",
 	"DDR5",
 	"LPDDR4",
-	"LPDDR5"
+	"LPDDR5",
+	"HBM3E"
 };
 
 /**
@@ -1262,7 +1268,7 @@ void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
 
 	amdgpu_bo_kunmap(abo);
 
-	if (abo->tbo.base.dma_buf && !abo->tbo.base.import_attach &&
+	if (abo->tbo.base.dma_buf && !drm_gem_is_imported(&abo->tbo.base) &&
 	    old_mem && old_mem->mem_type != TTM_PL_SYSTEM)
 		dma_buf_move_notify(abo->tbo.base.dma_buf);
 
@@ -1295,28 +1301,37 @@ void amdgpu_bo_release_notify(struct ttm_buffer_object *bo)
 	if (abo->kfd_bo)
 		amdgpu_amdkfd_release_notify(abo);
 
-	/* We only remove the fence if the resv has individualized. */
-	WARN_ON_ONCE(bo->type == ttm_bo_type_kernel
-			&& bo->base.resv != &bo->base._resv);
-	if (bo->base.resv == &bo->base._resv)
-		amdgpu_amdkfd_remove_fence_on_pt_pd_bos(abo);
+	/*
+	 * We lock the private dma_resv object here and since the BO is about to
+	 * be released nobody else should have a pointer to it.
+	 * So when this locking here fails something is wrong with the reference
+	 * counting.
+	 */
+	if (WARN_ON_ONCE(!dma_resv_trylock(&bo->base._resv)))
+		return;
+
+	amdgpu_amdkfd_remove_all_eviction_fences(abo);
 
 	if (!bo->resource || bo->resource->mem_type != TTM_PL_VRAM ||
 	    !(abo->flags & AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE) ||
 	    adev->in_suspend || drm_dev_is_unplugged(adev_to_drm(adev)))
-		return;
+		goto out;
 
-	if (WARN_ON_ONCE(!dma_resv_trylock(bo->base.resv)))
-		return;
+	r = dma_resv_reserve_fences(&bo->base._resv, 1);
+	if (r)
+		goto out;
 
-	r = amdgpu_fill_buffer(abo, 0, bo->base.resv, &fence, true);
-	if (!WARN_ON(r)) {
-		amdgpu_vram_mgr_set_cleared(bo->resource);
-		amdgpu_bo_fence(abo, fence, false);
-		dma_fence_put(fence);
-	}
+	r = amdgpu_fill_buffer(abo, 0, &bo->base._resv, &fence, true,
+			       AMDGPU_KERNEL_JOB_ID_CLEAR_ON_RELEASE);
+	if (WARN_ON(r))
+		goto out;
+
+	amdgpu_vram_mgr_set_cleared(bo->resource);
+	dma_resv_add_fence(&bo->base._resv, fence, DMA_RESV_USAGE_KERNEL);
+	dma_fence_put(fence);
 
-	dma_resv_unlock(bo->base.resv);
+out:
+	dma_resv_unlock(&bo->base._resv);
 }
 
 /**
@@ -1464,6 +1479,26 @@ u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo)
 }
 
 /**
+ * amdgpu_bo_fb_aper_addr - return FB aperture GPU offset of the VRAM bo
+ * @bo:	amdgpu VRAM buffer object for which we query the offset
+ *
+ * Returns:
+ * current FB aperture GPU offset of the object.
+ */
+u64 amdgpu_bo_fb_aper_addr(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	uint64_t offset, fb_base;
+
+	WARN_ON_ONCE(bo->tbo.resource->mem_type != TTM_PL_VRAM);
+
+	fb_base = adev->gmc.fb_start;
+	fb_base += adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
+	offset = (bo->tbo.resource->start << PAGE_SHIFT) + fb_base;
+	return amdgpu_gmc_sign_extend(offset);
+}
+
+/**
  * amdgpu_bo_gpu_offset_no_check - return GPU offset of bo
  * @bo:	amdgpu object for which we query the offset
  *
@@ -1519,6 +1554,8 @@ uint32_t amdgpu_bo_mem_stats_placement(struct amdgpu_bo *bo)
 		return AMDGPU_PL_OA;
 	case AMDGPU_GEM_DOMAIN_DOORBELL:
 		return AMDGPU_PL_DOORBELL;
+	case AMDGPU_GEM_DOMAIN_MMIO_REMAP:
+		return AMDGPU_PL_MMIO_REMAP;
 	default:
 		return TTM_PL_SYSTEM;
 	}
@@ -1602,6 +1639,9 @@ u64 amdgpu_bo_print_info(int id, struct amdgpu_bo *bo, struct seq_file *m)
 			case AMDGPU_PL_DOORBELL:
 				placement = "DOORBELL";
 				break;
+			case AMDGPU_PL_MMIO_REMAP:
+				placement = "MMIO REMAP";
+				break;
 			case TTM_PL_SYSTEM:
 			default:
 				placement = "CPU";
@@ -1636,7 +1676,11 @@ u64 amdgpu_bo_print_info(int id, struct amdgpu_bo *bo, struct seq_file *m)
 	amdgpu_bo_print_flag(m, bo, VRAM_CONTIGUOUS);
 	amdgpu_bo_print_flag(m, bo, VM_ALWAYS_VALID);
 	amdgpu_bo_print_flag(m, bo, EXPLICIT_SYNC);
-
+	/* Add the gem obj resv fence dump*/
+	if (dma_resv_trylock(bo->tbo.base.resv)) {
+		dma_resv_describe(bo->tbo.base.resv, m);
+		dma_resv_unlock(bo->tbo.base.resv);
+	}
 	seq_puts(m, "\n");
 
 	return size;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
index 375448627f7b..52c2d1731aab 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
@@ -69,7 +69,7 @@ struct amdgpu_bo_va_mapping {
 	uint64_t			last;
 	uint64_t			__subtree_last;
 	uint64_t			offset;
-	uint64_t			flags;
+	uint32_t			flags;
 };
 
 /* User space allocated BO in a VM */
@@ -96,6 +96,7 @@ struct amdgpu_bo_va {
 	 * if non-zero, cannot unmap from GPU because user queues may still access it
 	 */
 	unsigned int			queue_refcount;
+	atomic_t			userq_va_mapped;
 };
 
 struct amdgpu_bo {
@@ -167,6 +168,8 @@ static inline unsigned amdgpu_mem_type_to_domain(u32 mem_type)
 		return AMDGPU_GEM_DOMAIN_OA;
 	case AMDGPU_PL_DOORBELL:
 		return AMDGPU_GEM_DOMAIN_DOORBELL;
+	case AMDGPU_PL_MMIO_REMAP:
+		return AMDGPU_GEM_DOMAIN_MMIO_REMAP;
 	default:
 		break;
 	}
@@ -304,6 +307,7 @@ int amdgpu_bo_sync_wait_resv(struct amdgpu_device *adev, struct dma_resv *resv,
 			     bool intr);
 int amdgpu_bo_sync_wait(struct amdgpu_bo *bo, void *owner, bool intr);
 u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo);
+u64 amdgpu_bo_fb_aper_addr(struct amdgpu_bo *bo);
 u64 amdgpu_bo_gpu_offset_no_check(struct amdgpu_bo *bo);
 uint32_t amdgpu_bo_mem_stats_placement(struct amdgpu_bo *bo);
 uint32_t amdgpu_bo_get_preferred_domain(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
index e5fc80ed06ea..0b10497d487c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
@@ -44,7 +44,7 @@
 #include "amdgpu_securedisplay.h"
 #include "amdgpu_atomfirmware.h"
 
-#define AMD_VBIOS_FILE_MAX_SIZE_B      (1024*1024*3)
+#define AMD_VBIOS_FILE_MAX_SIZE_B      (1024*1024*16)
 
 static int psp_load_smu_fw(struct psp_context *psp);
 static int psp_rap_terminate(struct psp_context *psp);
@@ -153,6 +153,9 @@ static int psp_init_sriov_microcode(struct psp_context *psp)
 		adev->virt.autoload_ucode_id = AMDGPU_UCODE_ID_CP_MES1_DATA;
 		ret = psp_init_cap_microcode(psp, ucode_prefix);
 		break;
+	case IP_VERSION(13, 0, 12):
+		ret = psp_init_ta_microcode(psp, ucode_prefix);
+		break;
 	default:
 		return -EINVAL;
 	}
@@ -193,6 +196,7 @@ static int psp_early_init(struct amdgpu_ip_block *ip_block)
 	case IP_VERSION(11, 0, 9):
 	case IP_VERSION(11, 0, 11):
 	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 2):
 	case IP_VERSION(11, 0, 12):
 	case IP_VERSION(11, 0, 13):
 		psp_v11_0_set_psp_funcs(psp);
@@ -208,11 +212,15 @@ static int psp_early_init(struct amdgpu_ip_block *ip_block)
 		psp->boot_time_tmr = false;
 		fallthrough;
 	case IP_VERSION(13, 0, 6):
-	case IP_VERSION(13, 0, 12):
 	case IP_VERSION(13, 0, 14):
 		psp_v13_0_set_psp_funcs(psp);
 		psp->autoload_supported = false;
 		break;
+	case IP_VERSION(13, 0, 12):
+		psp_v13_0_set_psp_funcs(psp);
+		psp->autoload_supported = false;
+		adev->psp.sup_ifwi_up = !amdgpu_sriov_vf(adev);
+		break;
 	case IP_VERSION(13, 0, 1):
 	case IP_VERSION(13, 0, 3):
 	case IP_VERSION(13, 0, 5):
@@ -244,8 +252,13 @@ static int psp_early_init(struct amdgpu_ip_block *ip_block)
 		break;
 	case IP_VERSION(14, 0, 2):
 	case IP_VERSION(14, 0, 3):
+		adev->psp.sup_ifwi_up = !amdgpu_sriov_vf(adev);
 		psp_v14_0_set_psp_funcs(psp);
 		break;
+	case IP_VERSION(14, 0, 5):
+		psp_v14_0_set_psp_funcs(psp);
+		psp->boot_time_tmr = false;
+		break;
 	default:
 		return -EINVAL;
 	}
@@ -435,7 +448,7 @@ static int psp_sw_init(struct amdgpu_ip_block *ip_block)
 	psp->cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
 	if (!psp->cmd) {
 		dev_err(adev->dev, "Failed to allocate memory to command buffer!\n");
-		ret = -ENOMEM;
+		return -ENOMEM;
 	}
 
 	adev->psp.xgmi_context.supports_extended_data =
@@ -533,7 +546,6 @@ static int psp_sw_fini(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_device *adev = ip_block->adev;
 	struct psp_context *psp = &adev->psp;
-	struct psp_gfx_cmd_resp *cmd = psp->cmd;
 
 	psp_memory_training_fini(psp);
 
@@ -543,8 +555,8 @@ static int psp_sw_fini(struct amdgpu_ip_block *ip_block)
 	amdgpu_ucode_release(&psp->cap_fw);
 	amdgpu_ucode_release(&psp->toc_fw);
 
-	kfree(cmd);
-	cmd = NULL;
+	kfree(psp->cmd);
+	psp->cmd = NULL;
 
 	psp_free_shared_bufs(psp);
 
@@ -563,9 +575,11 @@ static int psp_sw_fini(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
-		 uint32_t reg_val, uint32_t mask, bool check_changed)
+int psp_wait_for(struct psp_context *psp, uint32_t reg_index, uint32_t reg_val,
+		 uint32_t mask, uint32_t flags)
 {
+	bool check_changed = flags & PSP_WAITREG_CHANGED;
+	bool verbose = !(flags & PSP_WAITREG_NOVERBOSE);
 	uint32_t val;
 	int i;
 	struct amdgpu_device *adev = psp->adev;
@@ -585,6 +599,11 @@ int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
 		udelay(1);
 	}
 
+	if (verbose)
+		dev_err(adev->dev,
+			"psp reg (0x%x) wait timed out, mask: %x, read: %x exp: %x",
+			reg_index, mask, val, reg_val);
+
 	return -ETIME;
 }
 
@@ -643,6 +662,14 @@ static const char *psp_gfx_cmd_name(enum psp_gfx_cmd_id cmd_id)
 		return "BOOT_CFG";
 	case GFX_CMD_ID_CONFIG_SQ_PERFMON:
 		return "CONFIG_SQ_PERFMON";
+	case GFX_CMD_ID_FB_FW_RESERV_ADDR:
+		return "FB_FW_RESERV_ADDR";
+	case GFX_CMD_ID_FB_FW_RESERV_EXT_ADDR:
+		return "FB_FW_RESERV_EXT_ADDR";
+	case GFX_CMD_ID_SRIOV_SPATIAL_PART:
+		return "SPATIAL_PARTITION";
+	case GFX_CMD_ID_FB_NPS_MODE:
+		return "NPS_MODE_CHANGE";
 	default:
 		return "UNKNOWN CMD";
 	}
@@ -854,12 +881,12 @@ static int psp_tmr_init(struct psp_context *psp)
 		pptr = amdgpu_sriov_vf(psp->adev) ? &tmr_buf : NULL;
 		ret = amdgpu_bo_create_kernel(psp->adev, tmr_size,
 					      PSP_TMR_ALIGNMENT,
-					      AMDGPU_HAS_VRAM(psp->adev) ?
-					      AMDGPU_GEM_DOMAIN_VRAM :
-					      AMDGPU_GEM_DOMAIN_GTT,
+					      AMDGPU_GEM_DOMAIN_GTT | AMDGPU_GEM_DOMAIN_VRAM,
 					      &psp->tmr_bo, &psp->tmr_mc_addr,
 					      pptr);
 	}
+	if (amdgpu_virt_xgmi_migrate_enabled(psp->adev) && psp->tmr_bo)
+		psp->tmr_mc_addr = amdgpu_bo_fb_aper_addr(psp->tmr_bo);
 
 	return ret;
 }
@@ -973,6 +1000,106 @@ int psp_get_fw_attestation_records_addr(struct psp_context *psp,
 	return ret;
 }
 
+static int psp_get_fw_reservation_info(struct psp_context *psp,
+						   uint32_t cmd_id,
+						   uint64_t *addr,
+						   uint32_t *size)
+{
+	int ret;
+	uint32_t status;
+	struct psp_gfx_cmd_resp *cmd;
+
+	cmd = acquire_psp_cmd_buf(psp);
+
+	cmd->cmd_id = cmd_id;
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd,
+				 psp->fence_buf_mc_addr);
+	if (ret) {
+		release_psp_cmd_buf(psp);
+		return ret;
+	}
+
+	status = cmd->resp.status;
+	if (status == PSP_ERR_UNKNOWN_COMMAND) {
+		release_psp_cmd_buf(psp);
+		*addr = 0;
+		*size = 0;
+		return 0;
+	}
+
+	*addr = (uint64_t)cmd->resp.uresp.fw_reserve_info.reserve_base_address_hi << 32 |
+		cmd->resp.uresp.fw_reserve_info.reserve_base_address_lo;
+	*size = cmd->resp.uresp.fw_reserve_info.reserve_size;
+
+	release_psp_cmd_buf(psp);
+
+	return 0;
+}
+
+int psp_update_fw_reservation(struct psp_context *psp)
+{
+	int ret;
+	uint64_t reserv_addr, reserv_addr_ext;
+	uint32_t reserv_size, reserv_size_ext, mp0_ip_ver;
+	struct amdgpu_device *adev = psp->adev;
+
+	mp0_ip_ver = amdgpu_ip_version(adev, MP0_HWIP, 0);
+
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	switch (mp0_ip_ver) {
+	case IP_VERSION(14, 0, 2):
+		if (adev->psp.sos.fw_version < 0x3b0e0d)
+			return 0;
+		break;
+
+	case IP_VERSION(14, 0, 3):
+		if (adev->psp.sos.fw_version < 0x3a0e14)
+			return 0;
+		break;
+
+	default:
+		return 0;
+	}
+
+	ret = psp_get_fw_reservation_info(psp, GFX_CMD_ID_FB_FW_RESERV_ADDR, &reserv_addr, &reserv_size);
+	if (ret)
+		return ret;
+	ret = psp_get_fw_reservation_info(psp, GFX_CMD_ID_FB_FW_RESERV_EXT_ADDR, &reserv_addr_ext, &reserv_size_ext);
+	if (ret)
+		return ret;
+
+	if (reserv_addr != adev->gmc.real_vram_size - reserv_size) {
+		dev_warn(adev->dev, "reserve fw region is not valid!\n");
+		return 0;
+	}
+
+	amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL, NULL);
+
+	reserv_size = roundup(reserv_size, SZ_1M);
+
+	ret = amdgpu_bo_create_kernel_at(adev, reserv_addr, reserv_size, &adev->mman.fw_reserved_memory, NULL);
+	if (ret) {
+		dev_err(adev->dev, "reserve fw region failed(%d)!\n", ret);
+		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL, NULL);
+		return ret;
+	}
+
+	reserv_size_ext = roundup(reserv_size_ext, SZ_1M);
+
+	ret = amdgpu_bo_create_kernel_at(adev, reserv_addr_ext, reserv_size_ext,
+					 &adev->mman.fw_reserved_memory_extend, NULL);
+	if (ret) {
+		dev_err(adev->dev, "reserve extend fw region failed(%d)!\n", ret);
+		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory_extend, NULL, NULL);
+		return ret;
+	}
+
+	return 0;
+}
+
 static int psp_boot_config_get(struct amdgpu_device *adev, uint32_t *boot_cfg)
 {
 	struct psp_context *psp = &adev->psp;
@@ -1259,6 +1386,11 @@ int psp_ta_load(struct psp_context *psp, struct ta_context *context)
 	psp_copy_fw(psp, context->bin_desc.start_addr,
 		    context->bin_desc.size_bytes);
 
+	if (amdgpu_virt_xgmi_migrate_enabled(psp->adev) &&
+		context->mem_context.shared_bo)
+		context->mem_context.shared_mc_addr =
+			amdgpu_bo_fb_aper_addr(context->mem_context.shared_bo);
+
 	psp_prep_ta_load_cmd_buf(cmd, psp->fw_pri_mc_addr, context);
 
 	ret = psp_cmd_submit_buf(psp, NULL, cmd,
@@ -1407,6 +1539,7 @@ static void psp_xgmi_reflect_topology_info(struct psp_context *psp,
 	uint64_t src_node_id = psp->adev->gmc.xgmi.node_id;
 	uint64_t dst_node_id = node_info.node_id;
 	uint8_t dst_num_hops = node_info.num_hops;
+	uint8_t dst_is_sharing_enabled = node_info.is_sharing_enabled;
 	uint8_t dst_num_links = node_info.num_links;
 
 	hive = amdgpu_get_xgmi_hive(psp->adev);
@@ -1426,13 +1559,20 @@ static void psp_xgmi_reflect_topology_info(struct psp_context *psp,
 				continue;
 
 			mirror_top_info->nodes[j].num_hops = dst_num_hops;
-			/*
-			 * prevent 0 num_links value re-reflection since reflection
+			mirror_top_info->nodes[j].is_sharing_enabled = dst_is_sharing_enabled;
+			/* prevent 0 num_links value re-reflection since reflection
 			 * criteria is based on num_hops (direct or indirect).
-			 *
 			 */
-			if (dst_num_links)
+			if (dst_num_links) {
 				mirror_top_info->nodes[j].num_links = dst_num_links;
+				/* swap src and dst due to frame of reference */
+				for (int k = 0; k < dst_num_links; k++) {
+					mirror_top_info->nodes[j].port_num[k].src_xgmi_port_num =
+						node_info.port_num[k].dst_xgmi_port_num;
+					mirror_top_info->nodes[j].port_num[k].dst_xgmi_port_num =
+						node_info.port_num[k].src_xgmi_port_num;
+				}
+			}
 
 			break;
 		}
@@ -1507,9 +1647,10 @@ int psp_xgmi_get_topology_info(struct psp_context *psp,
 			amdgpu_ip_version(psp->adev, MP0_HWIP, 0) ==
 				IP_VERSION(13, 0, 6) ||
 			amdgpu_ip_version(psp->adev, MP0_HWIP, 0) ==
-				IP_VERSION(13, 0, 14);
-		bool ta_port_num_support = amdgpu_sriov_vf(psp->adev) ? 0 :
-				psp->xgmi_context.xgmi_ta_caps & EXTEND_PEER_LINK_INFO_CMD_FLAG;
+				IP_VERSION(13, 0, 14) ||
+			amdgpu_sriov_vf(psp->adev);
+		bool ta_port_num_support = psp->xgmi_context.xgmi_ta_caps & EXTEND_PEER_LINK_INFO_CMD_FLAG ||
+			amdgpu_sriov_xgmi_ta_ext_peer_link_en(psp->adev);
 
 		/* popluate the shared output buffer rather than the cmd input buffer
 		 * with node_ids as the input for GET_PEER_LINKS command execution.
@@ -1786,34 +1927,47 @@ int psp_ras_initialize(struct psp_context *psp)
 		if (ret)
 			dev_warn(adev->dev, "PSP get boot config failed\n");
 
-		if (!amdgpu_ras_is_supported(psp->adev, AMDGPU_RAS_BLOCK__UMC)) {
-			if (!boot_cfg) {
-				dev_info(adev->dev, "GECC is disabled\n");
-			} else {
-				/* disable GECC in next boot cycle if ras is
-				 * disabled by module parameter amdgpu_ras_enable
-				 * and/or amdgpu_ras_mask, or boot_config_get call
-				 * is failed
-				 */
-				ret = psp_boot_config_set(adev, 0);
-				if (ret)
-					dev_warn(adev->dev, "PSP set boot config failed\n");
-				else
-					dev_warn(adev->dev, "GECC will be disabled in next boot cycle if set amdgpu_ras_enable and/or amdgpu_ras_mask to 0x0\n");
-			}
+		if (boot_cfg == 1 && !adev->ras_default_ecc_enabled &&
+		    amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC)) {
+			dev_warn(adev->dev, "GECC is currently enabled, which may affect performance\n");
+			dev_warn(adev->dev,
+				"To disable GECC, please reboot the system and load the amdgpu driver with the parameter amdgpu_ras_enable=0\n");
 		} else {
-			if (boot_cfg == 1) {
-				dev_info(adev->dev, "GECC is enabled\n");
+			if ((adev->ras_default_ecc_enabled || amdgpu_ras_enable == 1) &&
+				amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC)) {
+				if (boot_cfg == 1) {
+					dev_info(adev->dev, "GECC is enabled\n");
+				} else {
+					/* enable GECC in next boot cycle if it is disabled
+					 * in boot config, or force enable GECC if failed to
+					 * get boot configuration
+					 */
+					ret = psp_boot_config_set(adev, BOOT_CONFIG_GECC);
+					if (ret)
+						dev_warn(adev->dev, "PSP set boot config failed\n");
+					else
+						dev_warn(adev->dev, "GECC will be enabled in next boot cycle\n");
+				}
 			} else {
-				/* enable GECC in next boot cycle if it is disabled
-				 * in boot config, or force enable GECC if failed to
-				 * get boot configuration
-				 */
-				ret = psp_boot_config_set(adev, BOOT_CONFIG_GECC);
-				if (ret)
-					dev_warn(adev->dev, "PSP set boot config failed\n");
-				else
-					dev_warn(adev->dev, "GECC will be enabled in next boot cycle\n");
+				if (!boot_cfg) {
+					if (!adev->ras_default_ecc_enabled &&
+					    amdgpu_ras_enable != 1 &&
+					    amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC))
+						dev_warn(adev->dev, "GECC is disabled, set amdgpu_ras_enable=1 to enable GECC in next boot cycle if needed\n");
+					else
+						dev_info(adev->dev, "GECC is disabled\n");
+				} else {
+					/* disable GECC in next boot cycle if ras is
+					 * disabled by module parameter amdgpu_ras_enable
+					 * and/or amdgpu_ras_mask, or boot_config_get call
+					 * is failed
+					 */
+					ret = psp_boot_config_set(adev, 0);
+					if (ret)
+						dev_warn(adev->dev, "PSP set boot config failed\n");
+					else
+						dev_warn(adev->dev, "GECC will be disabled in next boot cycle if set amdgpu_ras_enable and/or amdgpu_ras_mask to 0x0\n");
+				}
 			}
 		}
 	}
@@ -1840,6 +1994,7 @@ int psp_ras_initialize(struct psp_context *psp)
 	if (adev->gmc.gmc_funcs->query_mem_partition_mode)
 		ras_cmd->ras_in_message.init_flags.nps_mode =
 			adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+	ras_cmd->ras_in_message.init_flags.active_umc_mask = adev->umc.active_mask;
 
 	ret = psp_ta_load(psp, &psp->ras_context.context);
 
@@ -2189,7 +2344,8 @@ static int psp_securedisplay_initialize(struct psp_context *psp)
 
 	if (!psp->securedisplay_context.context.bin_desc.size_bytes ||
 	    !psp->securedisplay_context.context.bin_desc.start_addr) {
-		dev_info(psp->adev->dev, "SECUREDISPLAY: securedisplay ta ucode is not available\n");
+		dev_info(psp->adev->dev,
+			 "SECUREDISPLAY: optional securedisplay ta ucode is not available\n");
 		return 0;
 	}
 
@@ -2205,11 +2361,14 @@ static int psp_securedisplay_initialize(struct psp_context *psp)
 	}
 
 	ret = psp_ta_load(psp, &psp->securedisplay_context.context);
-	if (!ret) {
+	if (!ret && !psp->securedisplay_context.context.resp_status) {
 		psp->securedisplay_context.context.initialized = true;
 		mutex_init(&psp->securedisplay_context.mutex);
-	} else
+	} else {
+		/* don't try again */
+		psp->securedisplay_context.context.bin_desc.size_bytes = 0;
 		return ret;
+	}
 
 	mutex_lock(&psp->securedisplay_context.mutex);
 
@@ -2311,11 +2470,27 @@ bool amdgpu_psp_tos_reload_needed(struct amdgpu_device *adev)
 	return false;
 }
 
+static void psp_update_gpu_addresses(struct amdgpu_device *adev)
+{
+	struct psp_context *psp = &adev->psp;
+
+	if (psp->cmd_buf_bo && psp->cmd_buf_mem) {
+		psp->fw_pri_mc_addr = amdgpu_bo_fb_aper_addr(psp->fw_pri_bo);
+		psp->fence_buf_mc_addr = amdgpu_bo_fb_aper_addr(psp->fence_buf_bo);
+		psp->cmd_buf_mc_addr = amdgpu_bo_fb_aper_addr(psp->cmd_buf_bo);
+	}
+	if (adev->firmware.rbuf && psp->km_ring.ring_mem)
+		psp->km_ring.ring_mem_mc_addr = amdgpu_bo_fb_aper_addr(adev->firmware.rbuf);
+}
+
 static int psp_hw_start(struct psp_context *psp)
 {
 	struct amdgpu_device *adev = psp->adev;
 	int ret;
 
+	if (amdgpu_virt_xgmi_migrate_enabled(adev))
+		psp_update_gpu_addresses(adev);
+
 	if (!amdgpu_sriov_vf(adev)) {
 		if ((is_psp_fw_valid(psp->kdb)) &&
 		    (psp->funcs->bootloader_load_kdb != NULL)) {
@@ -2414,6 +2589,14 @@ static int psp_hw_start(struct psp_context *psp)
 		return ret;
 	}
 
+	if (!amdgpu_in_reset(adev) && !adev->in_suspend) {
+		ret = psp_update_fw_reservation(psp);
+		if (ret) {
+			dev_err(adev->dev, "update fw reservation failed!\n");
+			return ret;
+		}
+	}
+
 	if (amdgpu_sriov_vf(adev) && amdgpu_in_reset(adev))
 		goto skip_pin_bo;
 
@@ -3496,8 +3679,12 @@ int psp_init_sos_microcode(struct psp_context *psp, const char *chip_name)
 	uint8_t *ucode_array_start_addr;
 	int err = 0;
 
-	err = amdgpu_ucode_request(adev, &adev->psp.sos_fw, AMDGPU_UCODE_REQUIRED,
-				   "amdgpu/%s_sos.bin", chip_name);
+	if (amdgpu_is_kicker_fw(adev))
+		err = amdgpu_ucode_request(adev, &adev->psp.sos_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_sos_kicker.bin", chip_name);
+	else
+		err = amdgpu_ucode_request(adev, &adev->psp.sos_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_sos.bin", chip_name);
 	if (err)
 		goto out;
 
@@ -3773,8 +3960,12 @@ int psp_init_ta_microcode(struct psp_context *psp, const char *chip_name)
 	struct amdgpu_device *adev = psp->adev;
 	int err;
 
-	err = amdgpu_ucode_request(adev, &adev->psp.ta_fw, AMDGPU_UCODE_REQUIRED,
-				   "amdgpu/%s_ta.bin", chip_name);
+	if (amdgpu_is_kicker_fw(adev))
+		err = amdgpu_ucode_request(adev, &adev->psp.ta_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_ta_kicker.bin", chip_name);
+	else
+		err = amdgpu_ucode_request(adev, &adev->psp.ta_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_ta.bin", chip_name);
 	if (err)
 		return err;
 
@@ -4000,7 +4191,7 @@ int is_psp_fw_valid(struct psp_bin_desc bin)
 }
 
 static ssize_t amdgpu_psp_vbflash_write(struct file *filp, struct kobject *kobj,
-					struct bin_attribute *bin_attr,
+					const struct bin_attribute *bin_attr,
 					char *buffer, loff_t pos, size_t count)
 {
 	struct device *dev = kobj_to_dev(kobj);
@@ -4036,7 +4227,7 @@ static ssize_t amdgpu_psp_vbflash_write(struct file *filp, struct kobject *kobj,
 }
 
 static ssize_t amdgpu_psp_vbflash_read(struct file *filp, struct kobject *kobj,
-				       struct bin_attribute *bin_attr, char *buffer,
+				       const struct bin_attribute *bin_attr, char *buffer,
 				       loff_t pos, size_t count)
 {
 	struct device *dev = kobj_to_dev(kobj);
@@ -4088,7 +4279,7 @@ rel_buf:
  * Writing to this file will stage an IFWI for update. Reading from this file
  * will trigger the update process.
  */
-static struct bin_attribute psp_vbflash_bin_attr = {
+static const struct bin_attribute psp_vbflash_bin_attr = {
 	.attr = {.name = "psp_vbflash", .mode = 0660},
 	.size = 0,
 	.write = amdgpu_psp_vbflash_write,
@@ -4119,7 +4310,7 @@ static ssize_t amdgpu_psp_vbflash_status(struct device *dev,
 }
 static DEVICE_ATTR(psp_vbflash_status, 0440, amdgpu_psp_vbflash_status, NULL);
 
-static struct bin_attribute *bin_flash_attrs[] = {
+static const struct bin_attribute *const bin_flash_attrs[] = {
 	&psp_vbflash_bin_attr,
 	NULL
 };
@@ -4160,6 +4351,110 @@ const struct attribute_group amdgpu_flash_attr_group = {
 	.is_visible = amdgpu_flash_attr_is_visible,
 };
 
+#if defined(CONFIG_DEBUG_FS)
+static int psp_read_spirom_debugfs_open(struct inode *inode, struct file *filp)
+{
+	struct amdgpu_device *adev = filp->f_inode->i_private;
+	struct spirom_bo *bo_triplet;
+	int ret;
+
+	/* serialize the open() file calling */
+	if (!mutex_trylock(&adev->psp.mutex))
+		return -EBUSY;
+
+	/*
+	 * make sure only one userpace process is alive for dumping so that
+	 * only one memory buffer of AMD_VBIOS_FILE_MAX_SIZE * 2 is consumed.
+	 * let's say the case where one process try opening the file while
+	 * another one has proceeded to read or release. In this way, eliminate
+	 * the use of mutex for read() or release() callback as well.
+	 */
+	if (adev->psp.spirom_dump_trip) {
+		mutex_unlock(&adev->psp.mutex);
+		return -EBUSY;
+	}
+
+	bo_triplet = kzalloc(sizeof(struct spirom_bo), GFP_KERNEL);
+	if (!bo_triplet) {
+		mutex_unlock(&adev->psp.mutex);
+		return -ENOMEM;
+	}
+
+	ret = amdgpu_bo_create_kernel(adev, AMD_VBIOS_FILE_MAX_SIZE_B * 2,
+				      AMDGPU_GPU_PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &bo_triplet->bo,
+				      &bo_triplet->mc_addr,
+				      &bo_triplet->cpu_addr);
+	if (ret)
+		goto rel_trip;
+
+	ret = psp_dump_spirom(&adev->psp, bo_triplet->mc_addr);
+	if (ret)
+		goto rel_bo;
+
+	adev->psp.spirom_dump_trip = bo_triplet;
+	mutex_unlock(&adev->psp.mutex);
+	return 0;
+rel_bo:
+	amdgpu_bo_free_kernel(&bo_triplet->bo, &bo_triplet->mc_addr,
+			      &bo_triplet->cpu_addr);
+rel_trip:
+	kfree(bo_triplet);
+	mutex_unlock(&adev->psp.mutex);
+	dev_err(adev->dev, "Trying IFWI dump fails, err = %d\n", ret);
+	return ret;
+}
+
+static ssize_t psp_read_spirom_debugfs_read(struct file *filp, char __user *buf, size_t size,
+					    loff_t *pos)
+{
+	struct amdgpu_device *adev = filp->f_inode->i_private;
+	struct spirom_bo *bo_triplet = adev->psp.spirom_dump_trip;
+
+	if (!bo_triplet)
+		return -EINVAL;
+
+	return simple_read_from_buffer(buf,
+				       size,
+				       pos, bo_triplet->cpu_addr,
+				       AMD_VBIOS_FILE_MAX_SIZE_B * 2);
+}
+
+static int psp_read_spirom_debugfs_release(struct inode *inode, struct file *filp)
+{
+	struct amdgpu_device *adev = filp->f_inode->i_private;
+	struct spirom_bo *bo_triplet = adev->psp.spirom_dump_trip;
+
+	if (bo_triplet) {
+		amdgpu_bo_free_kernel(&bo_triplet->bo, &bo_triplet->mc_addr,
+				      &bo_triplet->cpu_addr);
+		kfree(bo_triplet);
+	}
+
+	adev->psp.spirom_dump_trip = NULL;
+	return 0;
+}
+
+static const struct file_operations psp_dump_spirom_debugfs_ops = {
+	.owner = THIS_MODULE,
+	.open = psp_read_spirom_debugfs_open,
+	.read = psp_read_spirom_debugfs_read,
+	.release = psp_read_spirom_debugfs_release,
+	.llseek = default_llseek,
+};
+#endif
+
+void amdgpu_psp_debugfs_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+
+	debugfs_create_file_size("psp_spirom_dump", 0444, minor->debugfs_root,
+				 adev, &psp_dump_spirom_debugfs_ops, AMD_VBIOS_FILE_MAX_SIZE_B * 2);
+#endif
+}
+
 const struct amd_ip_funcs psp_ip_funcs = {
 	.name = "psp",
 	.early_init = psp_early_init,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
index 8d5acc415d38..237b624aa51c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
@@ -39,6 +39,29 @@
 #define PSP_TMR_ALIGNMENT	0x100000
 #define PSP_FW_NAME_LEN		0x24
 
+/* VBIOS gfl defines */
+#define MBOX_READY_MASK 0x80000000
+#define MBOX_STATUS_MASK 0x0000FFFF
+#define MBOX_COMMAND_MASK 0x00FF0000
+#define MBOX_READY_FLAG 0x80000000
+#define C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_LO 0x2
+#define C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_HI 0x3
+#define C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE 0x4
+#define C2PMSG_CMD_SPI_GET_ROM_IMAGE_ADDR_LO 0xf
+#define C2PMSG_CMD_SPI_GET_ROM_IMAGE_ADDR_HI 0x10
+#define C2PMSG_CMD_SPI_GET_FLASH_IMAGE 0x11
+
+/* Command register bit 31 set to indicate readiness */
+#define MBOX_TOS_READY_FLAG (GFX_FLAG_RESPONSE)
+#define MBOX_TOS_READY_MASK (GFX_CMD_RESPONSE_MASK | GFX_CMD_STATUS_MASK)
+
+/* Values to check for a successful GFX_CMD response wait. Check against
+ * both status bits and response state - helps to detect a command failure
+ * or other unexpected cases like a device drop reading all 0xFFs
+ */
+#define MBOX_TOS_RESP_FLAG (GFX_FLAG_RESPONSE)
+#define MBOX_TOS_RESP_MASK (GFX_CMD_RESPONSE_MASK | GFX_CMD_STATUS_MASK)
+
 extern const struct attribute_group amdgpu_flash_attr_group;
 
 enum psp_shared_mem_size {
@@ -107,9 +130,13 @@ enum psp_reg_prog_id {
 	PSP_REG_IH_RB_CNTL        = 0,  /* register IH_RB_CNTL */
 	PSP_REG_IH_RB_CNTL_RING1  = 1,  /* register IH_RB_CNTL_RING1 */
 	PSP_REG_IH_RB_CNTL_RING2  = 2,  /* register IH_RB_CNTL_RING2 */
+	PSP_REG_MMHUB_L1_TLB_CNTL = 25,
 	PSP_REG_LAST
 };
 
+#define PSP_WAITREG_CHANGED BIT(0) /* check if the value has changed */
+#define PSP_WAITREG_NOVERBOSE BIT(1) /* No error verbose */
+
 struct psp_funcs {
 	int (*init_microcode)(struct psp_context *psp);
 	int (*wait_for_bootloader)(struct psp_context *psp);
@@ -137,11 +164,14 @@ struct psp_funcs {
 	int (*load_usbc_pd_fw)(struct psp_context *psp, uint64_t fw_pri_mc_addr);
 	int (*read_usbc_pd_fw)(struct psp_context *psp, uint32_t *fw_ver);
 	int (*update_spirom)(struct psp_context *psp, uint64_t fw_pri_mc_addr);
+	int (*dump_spirom)(struct psp_context *psp, uint64_t fw_pri_mc_addr);
 	int (*vbflash_stat)(struct psp_context *psp);
 	int (*fatal_error_recovery_quirk)(struct psp_context *psp);
 	bool (*get_ras_capability)(struct psp_context *psp);
 	bool (*is_aux_sos_load_required)(struct psp_context *psp);
 	bool (*is_reload_needed)(struct psp_context *psp);
+	int (*reg_program_no_ring)(struct psp_context *psp, uint32_t val,
+				   enum psp_reg_prog_id id);
 };
 
 struct ta_funcs {
@@ -319,6 +349,14 @@ struct psp_runtime_scpm_entry {
 	enum psp_runtime_scpm_authentication scpm_status;
 };
 
+#if defined(CONFIG_DEBUG_FS)
+struct spirom_bo {
+	struct amdgpu_bo *bo;
+	uint64_t mc_addr;
+	void *cpu_addr;
+};
+#endif
+
 struct psp_context {
 	struct amdgpu_device		*adev;
 	struct psp_ring			km_ring;
@@ -406,6 +444,9 @@ struct psp_context {
 	char				*vbflash_tmp_buf;
 	size_t				vbflash_image_size;
 	bool				vbflash_done;
+#if defined(CONFIG_DEBUG_FS)
+	struct spirom_bo *spirom_dump_trip;
+#endif
 };
 
 struct amdgpu_psp_funcs {
@@ -464,6 +505,10 @@ struct amdgpu_psp_funcs {
 	((psp)->funcs->update_spirom ? \
 	(psp)->funcs->update_spirom((psp), fw_pri_mc_addr) : -EINVAL)
 
+#define psp_dump_spirom(psp, fw_pri_mc_addr) \
+	((psp)->funcs->dump_spirom ? \
+	(psp)->funcs->dump_spirom((psp), fw_pri_mc_addr) : -EINVAL)
+
 #define psp_vbflash_status(psp) \
 	((psp)->funcs->vbflash_stat ? \
 	(psp)->funcs->vbflash_stat((psp)) : -EINVAL)
@@ -475,6 +520,10 @@ struct amdgpu_psp_funcs {
 #define psp_is_aux_sos_load_required(psp) \
 	((psp)->funcs->is_aux_sos_load_required ? (psp)->funcs->is_aux_sos_load_required((psp)) : 0)
 
+#define psp_reg_program_no_ring(psp, val, id) \
+	((psp)->funcs->reg_program_no_ring ? \
+	(psp)->funcs->reg_program_no_ring((psp), val, id) : -EINVAL)
+
 extern const struct amd_ip_funcs psp_ip_funcs;
 
 extern const struct amdgpu_ip_block_version psp_v3_1_ip_block;
@@ -486,8 +535,8 @@ extern const struct amdgpu_ip_block_version psp_v13_0_ip_block;
 extern const struct amdgpu_ip_block_version psp_v13_0_4_ip_block;
 extern const struct amdgpu_ip_block_version psp_v14_0_ip_block;
 
-extern int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
-			uint32_t field_val, uint32_t mask, bool check_changed);
+int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
+		 uint32_t field_val, uint32_t mask, uint32_t flags);
 extern int psp_wait_for_spirom_update(struct psp_context *psp, uint32_t reg_index,
 			uint32_t field_val, uint32_t mask, uint32_t msec_timeout);
 
@@ -553,7 +602,7 @@ int psp_init_cap_microcode(struct psp_context *psp,
 			  const char *chip_name);
 int psp_get_fw_attestation_records_addr(struct psp_context *psp,
 					uint64_t *output_ptr);
-
+int psp_update_fw_reservation(struct psp_context *psp);
 int psp_load_fw_list(struct psp_context *psp,
 		     struct amdgpu_firmware_info **ucode_list, int ucode_count);
 void psp_copy_fw(struct psp_context *psp, uint8_t *start_addr, uint32_t bin_size);
@@ -569,5 +618,9 @@ bool amdgpu_psp_get_ras_capability(struct psp_context *psp);
 int psp_config_sq_perfmon(struct psp_context *psp, uint32_t xcp_id,
 	bool core_override_enable, bool reg_override_enable, bool perfmon_override_enable);
 bool amdgpu_psp_tos_reload_needed(struct amdgpu_device *adev);
+int amdgpu_psp_reg_program_no_ring(struct psp_context *psp, uint32_t val,
+				   enum psp_reg_prog_id id);
+void amdgpu_psp_debugfs_init(struct amdgpu_device *adev);
+
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
index 38face981c3e..6e8aad91bcd3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
@@ -171,13 +171,9 @@ static ssize_t ta_if_load_debugfs_write(struct file *fp, const char *buf, size_t
 
 	copy_pos += sizeof(uint32_t);
 
-	ta_bin = kzalloc(ta_bin_len, GFP_KERNEL);
-	if (!ta_bin)
-		return -ENOMEM;
-	if (copy_from_user((void *)ta_bin, &buf[copy_pos], ta_bin_len)) {
-		ret = -EFAULT;
-		goto err_free_bin;
-	}
+	ta_bin = memdup_user(&buf[copy_pos], ta_bin_len);
+	if (IS_ERR(ta_bin))
+		return PTR_ERR(ta_bin);
 
 	/* Set TA context and functions */
 	set_ta_context_funcs(psp, ta_type, &context);
@@ -327,13 +323,9 @@ static ssize_t ta_if_invoke_debugfs_write(struct file *fp, const char *buf, size
 		return -EFAULT;
 	copy_pos += sizeof(uint32_t);
 
-	shared_buf = kzalloc(shared_buf_len, GFP_KERNEL);
-	if (!shared_buf)
-		return -ENOMEM;
-	if (copy_from_user((void *)shared_buf, &buf[copy_pos], shared_buf_len)) {
-		ret = -EFAULT;
-		goto err_free_shared_buf;
-	}
+	shared_buf = memdup_user(&buf[copy_pos], shared_buf_len);
+	if (IS_ERR(shared_buf))
+		return PTR_ERR(shared_buf);
 
 	set_ta_context_funcs(psp, ta_type, &context);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c
index 123bcf5c2bb1..bacf888735db 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c
@@ -101,7 +101,6 @@ static ssize_t amdgpu_rap_debugfs_write(struct file *f, const char __user *buf,
 	}
 
 	amdgpu_gfx_off_ctrl(adev, true);
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return size;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index f0924aa3f4e4..2a6cf7963dde 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -41,6 +41,7 @@
 #include "atom.h"
 #include "amdgpu_reset.h"
 #include "amdgpu_psp.h"
+#include "amdgpu_ras_mgr.h"
 
 #ifdef CONFIG_X86_MCE_AMD
 #include <asm/mce.h>
@@ -77,6 +78,7 @@ const char *ras_block_string[] = {
 	"jpeg",
 	"ih",
 	"mpio",
+	"mmsch",
 };
 
 const char *ras_mca_block_string[] = {
@@ -121,12 +123,15 @@ const char *get_ras_block_str(struct ras_common_if *ras_block)
 /* typical ECC bad page rate is 1 bad page per 100MB VRAM */
 #define RAS_BAD_PAGE_COVER              (100 * 1024 * 1024ULL)
 
-#define MAX_UMC_POISON_POLLING_TIME_ASYNC  300  //ms
+#define MAX_UMC_POISON_POLLING_TIME_ASYNC  10
 
 #define AMDGPU_RAS_RETIRE_PAGE_INTERVAL 100  //ms
 
 #define MAX_FLUSH_RETIRE_DWORK_TIMES  100
 
+#define BYPASS_ALLOCATED_ADDRESS        0x0
+#define BYPASS_INITIALIZATION_ADDRESS   0x1
+
 enum amdgpu_ras_retire_page_reservation {
 	AMDGPU_RAS_RETIRE_PAGE_RESERVED,
 	AMDGPU_RAS_RETIRE_PAGE_PENDING,
@@ -135,12 +140,18 @@ enum amdgpu_ras_retire_page_reservation {
 
 atomic_t amdgpu_ras_in_intr = ATOMIC_INIT(0);
 
-static bool amdgpu_ras_check_bad_page_unlock(struct amdgpu_ras *con,
+static int amdgpu_ras_check_bad_page_unlock(struct amdgpu_ras *con,
 				uint64_t addr);
-static bool amdgpu_ras_check_bad_page(struct amdgpu_device *adev,
+static int amdgpu_ras_check_bad_page(struct amdgpu_device *adev,
 				uint64_t addr);
+
+static void amdgpu_ras_critical_region_init(struct amdgpu_device *adev);
+static void amdgpu_ras_critical_region_fini(struct amdgpu_device *adev);
+
 #ifdef CONFIG_X86_MCE_AMD
 static void amdgpu_register_bad_pages_mca_notifier(struct amdgpu_device *adev);
+static void
+amdgpu_unregister_bad_pages_mca_notifier(struct amdgpu_device *adev);
 struct mce_notifier_adev_list {
 	struct amdgpu_device *devs[MAX_GPU_INSTANCE];
 	int num_gpu;
@@ -168,18 +179,16 @@ static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t addre
 	struct eeprom_table_record err_rec;
 	int ret;
 
-	if ((address >= adev->gmc.mc_vram_size) ||
-	    (address >= RAS_UMC_INJECT_ADDR_LIMIT)) {
+	ret = amdgpu_ras_check_bad_page(adev, address);
+	if (ret == -EINVAL) {
 		dev_warn(adev->dev,
-		         "RAS WARN: input address 0x%llx is invalid.\n",
-		         address);
+			"RAS WARN: input address 0x%llx is invalid.\n",
+			address);
 		return -EINVAL;
-	}
-
-	if (amdgpu_ras_check_bad_page(adev, address)) {
+	} else if (ret == 1) {
 		dev_warn(adev->dev,
-			 "RAS WARN: 0x%llx has already been marked as bad page!\n",
-			 address);
+			"RAS WARN: 0x%llx has already been marked as bad page!\n",
+			address);
 		return 0;
 	}
 
@@ -206,6 +215,56 @@ static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t addre
 	return 0;
 }
 
+static int amdgpu_check_address_validity(struct amdgpu_device *adev,
+			uint64_t address, uint64_t flags)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct amdgpu_vram_block_info blk_info;
+	uint64_t page_pfns[32] = {0};
+	int i, ret, count;
+	bool hit = false;
+
+	if (amdgpu_ip_version(adev, UMC_HWIP, 0) < IP_VERSION(12, 0, 0))
+		return 0;
+
+	if (amdgpu_sriov_vf(adev)) {
+		if (amdgpu_virt_check_vf_critical_region(adev, address, &hit))
+			return -EPERM;
+		return hit ? -EACCES : 0;
+	}
+
+	if ((address >= adev->gmc.mc_vram_size) ||
+	    (address >= RAS_UMC_INJECT_ADDR_LIMIT))
+		return -EFAULT;
+
+	count = amdgpu_umc_lookup_bad_pages_in_a_row(adev,
+				address, page_pfns, ARRAY_SIZE(page_pfns));
+	if (count <= 0)
+		return -EPERM;
+
+	for (i = 0; i < count; i++) {
+		memset(&blk_info, 0, sizeof(blk_info));
+		ret = amdgpu_vram_mgr_query_address_block_info(&adev->mman.vram_mgr,
+					page_pfns[i] << AMDGPU_GPU_PAGE_SHIFT, &blk_info);
+		if (!ret) {
+			/* The input address that needs to be checked is allocated by
+			 * current calling process, so it is necessary to exclude
+			 * the calling process.
+			 */
+			if ((flags == BYPASS_ALLOCATED_ADDRESS) &&
+			    ((blk_info.task.pid != task_pid_nr(current)) ||
+				strncmp(blk_info.task.comm, current->comm, TASK_COMM_LEN)))
+				return -EACCES;
+			else if ((flags == BYPASS_INITIALIZATION_ADDRESS) &&
+				(blk_info.task.pid == con->init_task_pid) &&
+				!strncmp(blk_info.task.comm, con->init_task_comm, TASK_COMM_LEN))
+				return -EACCES;
+		}
+	}
+
+	return 0;
+}
+
 static ssize_t amdgpu_ras_debugfs_read(struct file *f, char __user *buf,
 					size_t size, loff_t *pos)
 {
@@ -296,6 +355,8 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f,
 		op = 2;
 	else if (strstr(str, "retire_page") != NULL)
 		op = 3;
+	else if (strstr(str, "check_address") != NULL)
+		op = 4;
 	else if (str[0] && str[1] && str[2] && str[3])
 		/* ascii string, but commands are not matched. */
 		return -EINVAL;
@@ -310,6 +371,15 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f,
 			data->inject.address = address;
 
 			return 0;
+		} else if (op == 4) {
+			if (sscanf(str, "%*s 0x%llx 0x%llx", &address, &value) != 2 &&
+			    sscanf(str, "%*s %llu %llu", &address, &value) != 2)
+				return -EINVAL;
+
+			data->op = op;
+			data->inject.address = address;
+			data->inject.value = value;
+			return 0;
 		}
 
 		if (amdgpu_ras_find_block_id_by_name(block_name, &block_id))
@@ -499,6 +569,9 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f,
 			return size;
 		else
 			return ret;
+	} else if (data.op == 4) {
+		ret = amdgpu_check_address_validity(adev, data.inject.address, data.inject.value);
+		return ret ? ret : size;
 	}
 
 	if (!amdgpu_ras_is_supported(adev, data.head.block))
@@ -512,22 +585,16 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f,
 		ret = amdgpu_ras_feature_enable(adev, &data.head, 1);
 		break;
 	case 2:
-		if ((data.inject.address >= adev->gmc.mc_vram_size &&
-		    adev->gmc.mc_vram_size) ||
-		    (data.inject.address >= RAS_UMC_INJECT_ADDR_LIMIT)) {
-			dev_warn(adev->dev, "RAS WARN: input address "
-					"0x%llx is invalid.",
+		/* umc ce/ue error injection for a bad page is not allowed */
+		if (data.head.block == AMDGPU_RAS_BLOCK__UMC)
+			ret = amdgpu_ras_check_bad_page(adev, data.inject.address);
+		if (ret == -EINVAL) {
+			dev_warn(adev->dev, "RAS WARN: input address 0x%llx is invalid.",
 					data.inject.address);
-			ret = -EINVAL;
 			break;
-		}
-
-		/* umc ce/ue error injection for a bad page is not allowed */
-		if ((data.head.block == AMDGPU_RAS_BLOCK__UMC) &&
-		    amdgpu_ras_check_bad_page(adev, data.inject.address)) {
-			dev_warn(adev->dev, "RAS WARN: inject: 0x%llx has "
-				 "already been marked as bad!\n",
-				 data.inject.address);
+		} else if (ret == 1) {
+			dev_warn(adev->dev, "RAS WARN: inject: 0x%llx has already been marked as bad!\n",
+					data.inject.address);
 			break;
 		}
 
@@ -547,6 +614,8 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f,
 	return size;
 }
 
+static int amdgpu_uniras_clear_badpages_info(struct amdgpu_device *adev);
+
 /**
  * DOC: AMDGPU RAS debugfs EEPROM table reset interface
  *
@@ -571,6 +640,11 @@ static ssize_t amdgpu_ras_debugfs_eeprom_write(struct file *f,
 		(struct amdgpu_device *)file_inode(f)->i_private;
 	int ret;
 
+	if (amdgpu_uniras_enabled(adev)) {
+		ret = amdgpu_uniras_clear_badpages_info(adev);
+		return ret ? ret : size;
+	}
+
 	ret = amdgpu_ras_eeprom_reset_table(
 		&(amdgpu_ras_get_context(adev)->eeprom_control));
 
@@ -1106,6 +1180,9 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev,
 					      err_info->de_count, blk_name);
 			}
 		} else {
+			if (adev->debug_disable_ce_logs)
+				return;
+
 			for_each_ras_error(err_node, err_data) {
 				err_info = &err_node->err_info;
 				mcm_info = &err_info->mcm_info;
@@ -1475,9 +1552,51 @@ out_fini_err_data:
 	return ret;
 }
 
+static int amdgpu_uniras_clear_badpages_info(struct amdgpu_device *adev)
+{
+	struct ras_cmd_dev_handle req = {0};
+	int ret;
+
+	ret = amdgpu_ras_mgr_handle_ras_cmd(adev, RAS_CMD__CLEAR_BAD_PAGE_INFO,
+				&req, sizeof(req), NULL, 0);
+	if (ret) {
+		dev_err(adev->dev, "Failed to clear bad pages info, ret: %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int amdgpu_uniras_query_block_ecc(struct amdgpu_device *adev,
+			struct ras_query_if *info)
+{
+	struct ras_cmd_block_ecc_info_req req = {0};
+	struct ras_cmd_block_ecc_info_rsp rsp = {0};
+	int ret;
+
+	if (!info)
+		return -EINVAL;
+
+	req.block_id = info->head.block;
+	req.subblock_id = info->head.sub_block_index;
+
+	ret = amdgpu_ras_mgr_handle_ras_cmd(adev, RAS_CMD__GET_BLOCK_ECC_STATUS,
+				&req, sizeof(req), &rsp, sizeof(rsp));
+	if (!ret) {
+		info->ce_count = rsp.ce_count;
+		info->ue_count = rsp.ue_count;
+		info->de_count = rsp.de_count;
+	}
+
+	return ret;
+}
+
 int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_if *info)
 {
-	return amdgpu_ras_query_error_status_with_event(adev, info, RAS_EVENT_TYPE_INVALID);
+	if (amdgpu_uniras_enabled(adev))
+		return amdgpu_uniras_query_block_ecc(adev, info);
+	else
+		return amdgpu_ras_query_error_status_with_event(adev, info, RAS_EVENT_TYPE_INVALID);
 }
 
 int amdgpu_ras_reset_error_count(struct amdgpu_device *adev,
@@ -1497,6 +1616,9 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev,
 	    !amdgpu_ras_get_aca_debug_mode(adev))
 		return -EOPNOTSUPP;
 
+	if (amdgpu_sriov_vf(adev))
+		return -EOPNOTSUPP;
+
 	/* skip ras error reset in gpu reset */
 	if ((amdgpu_in_reset(adev) || amdgpu_ras_in_recovery(adev)) &&
 	    ((smu_funcs && smu_funcs->set_debug_mode) ||
@@ -1526,6 +1648,27 @@ int amdgpu_ras_reset_error_status(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int amdgpu_uniras_error_inject(struct amdgpu_device *adev,
+		struct ras_inject_if *info)
+{
+	struct ras_cmd_inject_error_req inject_req;
+	struct ras_cmd_inject_error_rsp rsp;
+
+	if (!info)
+		return -EINVAL;
+
+	memset(&inject_req, 0, sizeof(inject_req));
+	inject_req.block_id = info->head.block;
+	inject_req.subblock_id = info->head.sub_block_index;
+	inject_req.address = info->address;
+	inject_req.error_type = info->head.type;
+	inject_req.instance_mask = info->instance_mask;
+	inject_req.method = info->value;
+
+	return amdgpu_ras_mgr_handle_ras_cmd(adev, RAS_CMD__INJECT_ERROR,
+			&inject_req, sizeof(inject_req), &rsp, sizeof(rsp));
+}
+
 /* wrapper of psp_ras_trigger_error */
 int amdgpu_ras_error_inject(struct amdgpu_device *adev,
 		struct ras_inject_if *info)
@@ -1543,6 +1686,9 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev,
 							info->head.block,
 							info->head.sub_block_index);
 
+	if (amdgpu_uniras_enabled(adev))
+		return amdgpu_uniras_error_inject(adev, info);
+
 	/* inject on guest isn't allowed, return success directly */
 	if (amdgpu_sriov_vf(adev))
 		return 0;
@@ -1687,7 +1833,9 @@ int amdgpu_ras_query_error_count(struct amdgpu_device *adev,
 /* sysfs begin */
 
 static int amdgpu_ras_badpages_read(struct amdgpu_device *adev,
-		struct ras_badpage **bps, unsigned int *count);
+		struct ras_badpage *bps, uint32_t count, uint32_t start);
+static int amdgpu_uniras_badpages_read(struct amdgpu_device *adev,
+		struct ras_badpage *bps, uint32_t count, uint32_t start);
 
 static char *amdgpu_ras_badpage_flags_str(unsigned int flags)
 {
@@ -1733,7 +1881,7 @@ static char *amdgpu_ras_badpage_flags_str(unsigned int flags)
  */
 
 static ssize_t amdgpu_ras_sysfs_badpages_read(struct file *f,
-		struct kobject *kobj, struct bin_attribute *attr,
+		struct kobject *kobj, const struct bin_attribute *attr,
 		char *buf, loff_t ppos, size_t count)
 {
 	struct amdgpu_ras *con =
@@ -1745,19 +1893,50 @@ static ssize_t amdgpu_ras_sysfs_badpages_read(struct file *f,
 	unsigned int end = div64_ul(ppos + count - 1, element_size);
 	ssize_t s = 0;
 	struct ras_badpage *bps = NULL;
-	unsigned int bps_count = 0;
+	int bps_count = 0, i, status;
+	uint64_t address;
 
 	memset(buf, 0, count);
 
-	if (amdgpu_ras_badpages_read(adev, &bps, &bps_count))
+	bps_count = end - start;
+	bps = kmalloc_array(bps_count, sizeof(*bps), GFP_KERNEL);
+	if (!bps)
+		return 0;
+
+	memset(bps, 0, sizeof(*bps) * bps_count);
+
+	if (amdgpu_uniras_enabled(adev))
+		bps_count = amdgpu_uniras_badpages_read(adev, bps, bps_count, start);
+	else
+		bps_count = amdgpu_ras_badpages_read(adev, bps, bps_count, start);
+
+	if (bps_count <= 0) {
+		kfree(bps);
 		return 0;
+	}
+
+	for (i = 0; i < bps_count; i++) {
+		address = ((uint64_t)bps[i].bp) << AMDGPU_GPU_PAGE_SHIFT;
+		if (amdgpu_ras_check_critical_address(adev, address))
+			continue;
+
+		bps[i].size = AMDGPU_GPU_PAGE_SIZE;
+
+		status = amdgpu_vram_mgr_query_page_status(&adev->mman.vram_mgr,
+					address);
+		if (status == -EBUSY)
+			bps[i].flags = AMDGPU_RAS_RETIRE_PAGE_PENDING;
+		else if (status == -ENOENT)
+			bps[i].flags = AMDGPU_RAS_RETIRE_PAGE_FAULT;
+		else
+			bps[i].flags = AMDGPU_RAS_RETIRE_PAGE_RESERVED;
 
-	for (; start < end && start < bps_count; start++)
 		s += scnprintf(&buf[s], element_size + 1,
 				"0x%08x : 0x%08x : %1s\n",
-				bps[start].bp,
-				bps[start].size,
-				amdgpu_ras_badpage_flags_str(bps[start].flags));
+				bps[i].bp,
+				bps[i].size,
+				amdgpu_ras_badpage_flags_str(bps[i].flags));
+	}
 
 	kfree(bps);
 
@@ -1773,12 +1952,42 @@ static ssize_t amdgpu_ras_sysfs_features_read(struct device *dev,
 	return sysfs_emit(buf, "feature mask: 0x%x\n", con->features);
 }
 
+static bool amdgpu_ras_get_version_info(struct amdgpu_device *adev, u32 *major,
+			u32 *minor, u32 *rev)
+{
+	int i;
+
+	if (!adev || !major || !minor || !rev || !amdgpu_uniras_enabled(adev))
+		return false;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_RAS) {
+			*major = adev->ip_blocks[i].version->major;
+			*minor = adev->ip_blocks[i].version->minor;
+			*rev = adev->ip_blocks[i].version->rev;
+			return true;
+		}
+	}
+
+	return false;
+}
+
 static ssize_t amdgpu_ras_sysfs_version_show(struct device *dev,
 		struct device_attribute *attr, char *buf)
 {
 	struct amdgpu_ras *con =
 		container_of(attr, struct amdgpu_ras, version_attr);
-	return sysfs_emit(buf, "table version: 0x%x\n", con->eeprom_control.tbl_hdr.version);
+	u32 major, minor, rev;
+	ssize_t size = 0;
+
+	size += sysfs_emit_at(buf, size, "table version: 0x%x\n",
+			con->eeprom_control.tbl_hdr.version);
+
+	if (amdgpu_ras_get_version_info(con->adev, &major, &minor, &rev))
+		size += sysfs_emit_at(buf, size, "ras version: %u.%u.%u\n",
+			major, minor, rev);
+
+	return size;
 }
 
 static ssize_t amdgpu_ras_sysfs_schema_show(struct device *dev,
@@ -1864,6 +2073,9 @@ int amdgpu_ras_sysfs_create(struct amdgpu_device *adev,
 	if (!obj || obj->attr_inuse)
 		return -EINVAL;
 
+	if (amdgpu_sriov_vf(adev) && !amdgpu_virt_ras_telemetry_block_en(adev, head->block))
+		return 0;
+
 	get_obj(obj);
 
 	snprintf(obj->fs_data.sysfs_name, sizeof(obj->fs_data.sysfs_name),
@@ -2065,8 +2277,8 @@ void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev)
 /* debugfs end */
 
 /* ras fs */
-static BIN_ATTR(gpu_vram_bad_pages, S_IRUGO,
-		amdgpu_ras_sysfs_badpages_read, NULL, 0);
+static const BIN_ATTR(gpu_vram_bad_pages, S_IRUGO,
+		      amdgpu_ras_sysfs_badpages_read, NULL, 0);
 static DEVICE_ATTR(features, S_IRUGO,
 		amdgpu_ras_sysfs_features_read, NULL);
 static DEVICE_ATTR(version, 0444,
@@ -2088,7 +2300,7 @@ static int amdgpu_ras_fs_init(struct amdgpu_device *adev)
 		&con->event_state_attr.attr,
 		NULL
 	};
-	struct bin_attribute *bin_attrs[] = {
+	const struct bin_attribute *bin_attrs[] = {
 		NULL,
 		NULL,
 	};
@@ -2114,11 +2326,10 @@ static int amdgpu_ras_fs_init(struct amdgpu_device *adev)
 
 	if (amdgpu_bad_page_threshold != 0) {
 		/* add bad_page_features entry */
-		bin_attr_gpu_vram_bad_pages.private = NULL;
 		con->badpages_attr = bin_attr_gpu_vram_bad_pages;
+		sysfs_bin_attr_init(&con->badpages_attr);
 		bin_attrs[0] = &con->badpages_attr;
 		group.bin_attrs = bin_attrs;
-		sysfs_bin_attr_init(bin_attrs[0]);
 	}
 
 	r = sysfs_create_group(&adev->dev->kobj, &group);
@@ -2158,7 +2369,7 @@ void amdgpu_ras_interrupt_fatal_error_handler(struct amdgpu_device *adev)
 	/* Fatal error events are handled on host side */
 	if (amdgpu_sriov_vf(adev))
 		return;
-	/**
+	/*
 	 * If the current interrupt is caused by a non-fatal RAS error, skip
 	 * check for fatal error. For fatal errors, FED status of all devices
 	 * in XGMI hive gets set when the first device gets fatal error
@@ -2169,6 +2380,11 @@ void amdgpu_ras_interrupt_fatal_error_handler(struct amdgpu_device *adev)
 	    amdgpu_ras_is_err_state(adev, AMDGPU_RAS_BLOCK__ANY))
 		return;
 
+	if (amdgpu_uniras_enabled(adev)) {
+		amdgpu_ras_mgr_handle_fatal_interrupt(adev, NULL);
+		return;
+	}
+
 	if (adev->nbio.ras &&
 	    adev->nbio.ras->handle_ras_controller_intr_no_bifring)
 		adev->nbio.ras->handle_ras_controller_intr_no_bifring(adev);
@@ -2339,6 +2555,16 @@ int amdgpu_ras_interrupt_dispatch(struct amdgpu_device *adev,
 	struct ras_manager *obj;
 	struct ras_ih_data *data;
 
+	if (amdgpu_uniras_enabled(adev)) {
+		struct ras_ih_info ih_info;
+
+		memset(&ih_info, 0, sizeof(ih_info));
+		ih_info.block = info->head.block;
+		memcpy(&ih_info.iv_entry, info->entry, sizeof(struct amdgpu_iv_entry));
+
+		return amdgpu_ras_mgr_handle_controller_interrupt(adev, &ih_info);
+	}
+
 	obj = amdgpu_ras_find_obj(adev, &info->head);
 	if (!obj)
 		return -EINVAL;
@@ -2533,54 +2759,83 @@ static void amdgpu_ras_query_err_status(struct amdgpu_device *adev)
 	}
 }
 
-/* recovery begin */
-
-/* return 0 on success.
- * caller need free bps.
- */
 static int amdgpu_ras_badpages_read(struct amdgpu_device *adev,
-		struct ras_badpage **bps, unsigned int *count)
+		struct ras_badpage *bps, uint32_t count, uint32_t start)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	struct ras_err_handler_data *data;
-	int i = 0;
-	int ret = 0, status;
+	int r = 0;
+	uint32_t i;
 
 	if (!con || !con->eh_data || !bps || !count)
 		return -EINVAL;
 
 	mutex_lock(&con->recovery_lock);
 	data = con->eh_data;
-	if (!data || data->count == 0) {
-		*bps = NULL;
-		ret = -EINVAL;
-		goto out;
-	}
+	if (start < data->count) {
+		for (i = start; i < data->count; i++) {
+			if (!data->bps[i].ts)
+				continue;
 
-	*bps = kmalloc(sizeof(struct ras_badpage) * data->count, GFP_KERNEL);
-	if (!*bps) {
-		ret = -ENOMEM;
-		goto out;
+			bps[r].bp = data->bps[i].retired_page;
+			r++;
+			if (r >= count)
+				break;
+		}
 	}
+	mutex_unlock(&con->recovery_lock);
 
-	for (; i < data->count; i++) {
-		(*bps)[i] = (struct ras_badpage){
-			.bp = data->bps[i].retired_page,
-			.size = AMDGPU_GPU_PAGE_SIZE,
-			.flags = AMDGPU_RAS_RETIRE_PAGE_RESERVED,
-		};
-		status = amdgpu_vram_mgr_query_page_status(&adev->mman.vram_mgr,
-				data->bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT);
-		if (status == -EBUSY)
-			(*bps)[i].flags = AMDGPU_RAS_RETIRE_PAGE_PENDING;
-		else if (status == -ENOENT)
-			(*bps)[i].flags = AMDGPU_RAS_RETIRE_PAGE_FAULT;
+	return r;
+}
+
+static int amdgpu_uniras_badpages_read(struct amdgpu_device *adev,
+		struct ras_badpage *bps, uint32_t count, uint32_t start)
+{
+	struct ras_cmd_bad_pages_info_req cmd_input;
+	struct ras_cmd_bad_pages_info_rsp *output;
+	uint32_t group, start_group, end_group;
+	uint32_t pos, pos_in_group;
+	int r = 0, i;
+
+	if (!bps || !count)
+		return -EINVAL;
+
+	output = kmalloc(sizeof(*output), GFP_KERNEL);
+	if (!output)
+		return -ENOMEM;
+
+	memset(&cmd_input, 0, sizeof(cmd_input));
+
+	start_group = start / RAS_CMD_MAX_BAD_PAGES_PER_GROUP;
+	end_group = (start + count + RAS_CMD_MAX_BAD_PAGES_PER_GROUP - 1) /
+				RAS_CMD_MAX_BAD_PAGES_PER_GROUP;
+
+	pos = start;
+	for (group = start_group; group < end_group; group++) {
+		memset(output, 0, sizeof(*output));
+		cmd_input.group_index = group;
+		if (amdgpu_ras_mgr_handle_ras_cmd(adev, RAS_CMD__GET_BAD_PAGES,
+			&cmd_input, sizeof(cmd_input), output, sizeof(*output)))
+			goto out;
+
+		if (pos >= output->bp_total_cnt)
+			goto out;
+
+		pos_in_group = pos - group * RAS_CMD_MAX_BAD_PAGES_PER_GROUP;
+		for (i = pos_in_group; i < output->bp_in_group; i++, pos++) {
+			if (!output->records[i].ts)
+				continue;
+
+			bps[r].bp = output->records[i].retired_page;
+			r++;
+			if (r >= count)
+				goto out;
+		}
 	}
 
-	*count = data->count;
 out:
-	mutex_unlock(&con->recovery_lock);
-	return ret;
+	kfree(output);
+	return r;
 }
 
 static void amdgpu_ras_set_fed_all(struct amdgpu_device *adev,
@@ -2629,6 +2884,7 @@ static void amdgpu_ras_do_recovery(struct work_struct *work)
 	struct amdgpu_device *adev = ras->adev;
 	struct list_head device_list, *device_list_handle =  NULL;
 	struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev);
+	unsigned int error_query_mode;
 	enum ras_event_type type;
 
 	if (hive) {
@@ -2657,11 +2913,22 @@ static void amdgpu_ras_do_recovery(struct work_struct *work)
 			device_list_handle = &device_list;
 		}
 
+		if (amdgpu_ras_get_error_query_mode(adev, &error_query_mode)) {
+			if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY) {
+				/* wait 500ms to ensure pmfw polling mca bank info done */
+				msleep(500);
+			}
+		}
+
 		type = amdgpu_ras_get_fatal_error_event(adev);
 		list_for_each_entry(remote_adev,
 				device_list_handle, gmc.xgmi.head) {
-			amdgpu_ras_query_err_status(remote_adev);
-			amdgpu_ras_log_on_err_counter(remote_adev, type);
+			if (amdgpu_uniras_enabled(remote_adev)) {
+				amdgpu_ras_mgr_update_ras_ecc(remote_adev);
+			} else {
+				amdgpu_ras_query_err_status(remote_adev);
+				amdgpu_ras_log_on_err_counter(remote_adev, type);
+			}
 		}
 
 	}
@@ -2713,7 +2980,7 @@ static int amdgpu_ras_realloc_eh_data_space(struct amdgpu_device *adev,
 	unsigned int old_space = data->count + data->space_left;
 	unsigned int new_space = old_space + pages;
 	unsigned int align_space = ALIGN(new_space, 512);
-	void *bps = kmalloc(align_space * sizeof(*data->bps), GFP_KERNEL);
+	void *bps = kmalloc_array(align_space, sizeof(*data->bps), GFP_KERNEL);
 
 	if (!bps) {
 		return -ENOMEM;
@@ -2749,8 +3016,13 @@ static int amdgpu_ras_mca2pa_by_idx(struct amdgpu_device *adev,
 	addr_in.ma.err_addr = bps->address;
 	addr_in.ma.socket_id = socket;
 	addr_in.ma.ch_inst = bps->mem_channel;
-	/* tell RAS TA the node instance is not used */
-	addr_in.ma.node_inst = TA_RAS_INV_NODE;
+	if (!amdgpu_ras_smu_eeprom_supported(adev)) {
+		/* tell RAS TA the node instance is not used */
+		addr_in.ma.node_inst = TA_RAS_INV_NODE;
+	} else {
+		addr_in.ma.umc_inst = bps->mcumc_id;
+		addr_in.ma.node_inst = bps->cu;
+	}
 
 	if (adev->umc.ras && adev->umc.ras->convert_ras_err_addr)
 		ret = adev->umc.ras->convert_ras_err_addr(adev, err_data,
@@ -2796,20 +3068,157 @@ static int amdgpu_ras_mca2pa(struct amdgpu_device *adev,
 		return  -EINVAL;
 }
 
+static int __amdgpu_ras_restore_bad_pages(struct amdgpu_device *adev,
+					struct eeprom_table_record *bps, int count)
+{
+	int j;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_err_handler_data *data = con->eh_data;
+
+	for (j = 0; j < count; j++) {
+		if (amdgpu_ras_check_bad_page_unlock(con,
+			bps[j].retired_page << AMDGPU_GPU_PAGE_SHIFT)) {
+			data->count++;
+			data->space_left--;
+			continue;
+		}
+
+		if (!data->space_left &&
+		    amdgpu_ras_realloc_eh_data_space(adev, data, 256)) {
+			return -ENOMEM;
+		}
+
+		amdgpu_ras_reserve_page(adev, bps[j].retired_page);
+
+		memcpy(&data->bps[data->count], &(bps[j]),
+				sizeof(struct eeprom_table_record));
+		data->count++;
+		data->space_left--;
+		con->bad_page_num++;
+	}
+
+	return 0;
+}
+
+static int __amdgpu_ras_convert_rec_array_from_rom(struct amdgpu_device *adev,
+				struct eeprom_table_record *bps, struct ras_err_data *err_data,
+				enum amdgpu_memory_partition nps)
+{
+	int i = 0;
+	enum amdgpu_memory_partition save_nps;
+
+	save_nps = (bps[0].retired_page >> UMC_NPS_SHIFT) & UMC_NPS_MASK;
+
+	/*old asics just have pa in eeprom*/
+	if (IP_VERSION_MAJ(amdgpu_ip_version(adev, UMC_HWIP, 0)) < 12) {
+		memcpy(err_data->err_addr, bps,
+			sizeof(struct eeprom_table_record) * adev->umc.retire_unit);
+		goto out;
+	}
+
+	for (i = 0; i < adev->umc.retire_unit; i++)
+		bps[i].retired_page &= ~(UMC_NPS_MASK << UMC_NPS_SHIFT);
+
+	if (save_nps) {
+		if (save_nps == nps) {
+			if (amdgpu_umc_pages_in_a_row(adev, err_data,
+					bps[0].retired_page << AMDGPU_GPU_PAGE_SHIFT))
+				return -EINVAL;
+			for (i = 0; i < adev->umc.retire_unit; i++) {
+				err_data->err_addr[i].address = bps[0].address;
+				err_data->err_addr[i].mem_channel = bps[0].mem_channel;
+				err_data->err_addr[i].bank = bps[0].bank;
+				err_data->err_addr[i].err_type = bps[0].err_type;
+				err_data->err_addr[i].mcumc_id = bps[0].mcumc_id;
+			}
+		} else {
+			if (amdgpu_ras_mca2pa_by_idx(adev, &bps[0], err_data))
+				return -EINVAL;
+		}
+	} else {
+		if (bps[0].address == 0) {
+			/* for specific old eeprom data, mca address is not stored,
+			 * calc it from pa
+			 */
+			if (amdgpu_umc_pa2mca(adev, bps[0].retired_page << AMDGPU_GPU_PAGE_SHIFT,
+				&(bps[0].address), AMDGPU_NPS1_PARTITION_MODE))
+				return -EINVAL;
+		}
+
+		if (amdgpu_ras_mca2pa(adev, &bps[0], err_data)) {
+			if (nps == AMDGPU_NPS1_PARTITION_MODE)
+				memcpy(err_data->err_addr, bps,
+					sizeof(struct eeprom_table_record) * adev->umc.retire_unit);
+			else
+				return -EOPNOTSUPP;
+		}
+	}
+
+out:
+	return __amdgpu_ras_restore_bad_pages(adev, err_data->err_addr, adev->umc.retire_unit);
+}
+
+static int __amdgpu_ras_convert_rec_from_rom(struct amdgpu_device *adev,
+				struct eeprom_table_record *bps, struct ras_err_data *err_data,
+				enum amdgpu_memory_partition nps)
+{
+	int i = 0;
+	enum amdgpu_memory_partition save_nps;
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev)) {
+		save_nps = (bps->retired_page >> UMC_NPS_SHIFT) & UMC_NPS_MASK;
+		bps->retired_page &= ~(UMC_NPS_MASK << UMC_NPS_SHIFT);
+	} else {
+		/* if pmfw manages eeprom, save_nps is not stored on eeprom,
+		 * we should always convert mca address into physical address,
+		 * make save_nps different from nps
+		 */
+		save_nps = nps + 1;
+	}
+
+	if (save_nps == nps) {
+		if (amdgpu_umc_pages_in_a_row(adev, err_data,
+				bps->retired_page << AMDGPU_GPU_PAGE_SHIFT))
+			return -EINVAL;
+		for (i = 0; i < adev->umc.retire_unit; i++) {
+			err_data->err_addr[i].address = bps->address;
+			err_data->err_addr[i].mem_channel = bps->mem_channel;
+			err_data->err_addr[i].bank = bps->bank;
+			err_data->err_addr[i].err_type = bps->err_type;
+			err_data->err_addr[i].mcumc_id = bps->mcumc_id;
+		}
+	} else {
+		if (bps->address) {
+			if (amdgpu_ras_mca2pa_by_idx(adev, bps, err_data))
+				return -EINVAL;
+		} else {
+			/* for specific old eeprom data, mca address is not stored,
+			 * calc it from pa
+			 */
+			if (amdgpu_umc_pa2mca(adev, bps->retired_page << AMDGPU_GPU_PAGE_SHIFT,
+				&(bps->address), AMDGPU_NPS1_PARTITION_MODE))
+				return -EINVAL;
+
+			if (amdgpu_ras_mca2pa(adev, bps, err_data))
+				return -EOPNOTSUPP;
+		}
+	}
+
+	return __amdgpu_ras_restore_bad_pages(adev, err_data->err_addr,
+									adev->umc.retire_unit);
+}
+
 /* it deal with vram only. */
 int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev,
 		struct eeprom_table_record *bps, int pages, bool from_rom)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
-	struct ras_err_handler_data *data;
 	struct ras_err_data err_data;
-	struct eeprom_table_record *err_rec;
 	struct amdgpu_ras_eeprom_control *control =
 			&adev->psp.ras_context.ras->eeprom_control;
 	enum amdgpu_memory_partition nps = AMDGPU_NPS1_PARTITION_MODE;
 	int ret = 0;
-	uint32_t i, j, loop_cnt = 1;
-	bool find_pages_per_pa = false;
+	uint32_t i = 0;
 
 	if (!con || !con->eh_data || !bps || pages <= 0)
 		return 0;
@@ -2820,114 +3229,51 @@ int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev,
 				sizeof(struct eeprom_table_record), GFP_KERNEL);
 		if (!err_data.err_addr) {
 			dev_warn(adev->dev, "Failed to alloc UMC error address record in mca2pa conversion!\n");
-			ret = -ENOMEM;
-			goto out;
+			return -ENOMEM;
 		}
 
-		err_rec = err_data.err_addr;
-		loop_cnt = adev->umc.retire_unit;
 		if (adev->gmc.gmc_funcs->query_mem_partition_mode)
 			nps = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
 	}
 
 	mutex_lock(&con->recovery_lock);
-	data = con->eh_data;
-	if (!data) {
-		/* Returning 0 as the absence of eh_data is acceptable */
-		goto free;
-	}
-
-	for (i = 0; i < pages; i++) {
-		if (from_rom &&
-		    control->rec_type == AMDGPU_RAS_EEPROM_REC_MCA) {
-			if (!find_pages_per_pa) {
-				if (amdgpu_ras_mca2pa_by_idx(adev, &bps[i], &err_data)) {
-					if (!i && nps == AMDGPU_NPS1_PARTITION_MODE) {
-						/* may use old RAS TA, use PA to find pages in
-						 * one row
-						 */
-						if (amdgpu_umc_pages_in_a_row(adev, &err_data,
-									      bps[i].retired_page <<
-									      AMDGPU_GPU_PAGE_SHIFT)) {
-							ret = -EINVAL;
-							goto free;
-						} else {
-							find_pages_per_pa = true;
-						}
+
+	if (from_rom) {
+		/* there is no pa recs in V3, so skip pa recs processing */
+		if ((control->tbl_hdr.version < RAS_TABLE_VER_V3) &&
+		    !amdgpu_ras_smu_eeprom_supported(adev)) {
+			for (i = 0; i < pages; i++) {
+				if (control->ras_num_recs - i >= adev->umc.retire_unit) {
+					if ((bps[i].address == bps[i + 1].address) &&
+						(bps[i].mem_channel == bps[i + 1].mem_channel)) {
+						/* deal with retire_unit records a time */
+						ret = __amdgpu_ras_convert_rec_array_from_rom(adev,
+										&bps[i], &err_data, nps);
+						if (ret)
+							con->bad_page_num -= adev->umc.retire_unit;
+						i += (adev->umc.retire_unit - 1);
 					} else {
-						/* unsupported cases */
-						ret = -EOPNOTSUPP;
-						goto free;
-					}
-				}
-			} else {
-				if (amdgpu_umc_pages_in_a_row(adev, &err_data,
-						bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT)) {
-					ret = -EINVAL;
-					goto free;
-				}
-			}
-		} else {
-			if (from_rom && !find_pages_per_pa) {
-				if (bps[i].retired_page & UMC_CHANNEL_IDX_V2) {
-					/* bad page in any NPS mode in eeprom */
-					if (amdgpu_ras_mca2pa_by_idx(adev, &bps[i], &err_data)) {
-						ret = -EINVAL;
-						goto free;
+						break;
 					}
 				} else {
-					/* legacy bad page in eeprom, generated only in
-					 * NPS1 mode
-					 */
-					if (amdgpu_ras_mca2pa(adev, &bps[i], &err_data)) {
-						/* old RAS TA or ASICs which don't support to
-						 * convert addrss via mca address
-						 */
-						if (!i && nps == AMDGPU_NPS1_PARTITION_MODE) {
-							find_pages_per_pa = true;
-							err_rec = &bps[i];
-							loop_cnt = 1;
-						} else {
-							/* non-nps1 mode, old RAS TA
-							 * can't support it
-							 */
-							ret = -EOPNOTSUPP;
-							goto free;
-						}
-					}
+					break;
 				}
-
-				if (!find_pages_per_pa)
-					i += (adev->umc.retire_unit - 1);
-			} else {
-				err_rec = &bps[i];
 			}
 		}
-
-		for (j = 0; j < loop_cnt; j++) {
-			if (amdgpu_ras_check_bad_page_unlock(con,
-				err_rec[j].retired_page << AMDGPU_GPU_PAGE_SHIFT))
-				continue;
-
-			if (!data->space_left &&
-			    amdgpu_ras_realloc_eh_data_space(adev, data, 256)) {
-				ret = -ENOMEM;
-				goto free;
-			}
-
-			amdgpu_ras_reserve_page(adev, err_rec[j].retired_page);
-
-			memcpy(&data->bps[data->count], &(err_rec[j]),
-					sizeof(struct eeprom_table_record));
-			data->count++;
-			data->space_left--;
+		for (; i < pages; i++) {
+			ret = __amdgpu_ras_convert_rec_from_rom(adev,
+				&bps[i], &err_data, nps);
+			if (ret)
+				con->bad_page_num -= adev->umc.retire_unit;
 		}
+
+		con->eh_data->count_saved = con->eh_data->count;
+	} else {
+		ret = __amdgpu_ras_restore_bad_pages(adev, bps, pages);
 	}
 
-free:
 	if (from_rom)
 		kfree(err_data.err_addr);
-out:
 	mutex_unlock(&con->recovery_lock);
 
 	return ret;
@@ -2944,7 +3290,7 @@ int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev,
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	struct ras_err_handler_data *data;
 	struct amdgpu_ras_eeprom_control *control;
-	int save_count, unit_num, bad_page_num, i;
+	int save_count, unit_num, i;
 
 	if (!con || !con->eh_data) {
 		if (new_cnt)
@@ -2953,31 +3299,45 @@ int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev,
 		return 0;
 	}
 
+	if (!con->eeprom_control.is_eeprom_valid) {
+		dev_warn(adev->dev,
+			"Failed to save EEPROM table data because of EEPROM data corruption!");
+		if (new_cnt)
+			*new_cnt = 0;
+
+		return 0;
+	}
+
 	mutex_lock(&con->recovery_lock);
 	control = &con->eeprom_control;
 	data = con->eh_data;
-	bad_page_num = control->ras_num_bad_pages;
-	save_count = data->count - bad_page_num;
+	if (amdgpu_ras_smu_eeprom_supported(adev))
+		unit_num = control->ras_num_recs -
+			control->ras_num_recs_old;
+	else
+		unit_num = data->count / adev->umc.retire_unit -
+			control->ras_num_recs;
+
+	save_count = con->bad_page_num - control->ras_num_bad_pages;
 	mutex_unlock(&con->recovery_lock);
 
-	unit_num = save_count / adev->umc.retire_unit;
 	if (new_cnt)
 		*new_cnt = unit_num;
 
 	/* only new entries are saved */
-	if (save_count > 0) {
-		if (control->rec_type == AMDGPU_RAS_EEPROM_REC_PA) {
+	if (unit_num && save_count) {
+		/*old asics only save pa to eeprom like before*/
+		if (IP_VERSION_MAJ(amdgpu_ip_version(adev, UMC_HWIP, 0)) < 12) {
 			if (amdgpu_ras_eeprom_append(control,
-						     &data->bps[control->ras_num_recs],
-						     save_count)) {
+					&data->bps[data->count_saved], unit_num)) {
 				dev_err(adev->dev, "Failed to save EEPROM table data!");
 				return -EIO;
 			}
 		} else {
 			for (i = 0; i < unit_num; i++) {
 				if (amdgpu_ras_eeprom_append(control,
-						&data->bps[bad_page_num + i * adev->umc.retire_unit],
-						1)) {
+						&data->bps[data->count_saved +
+						i * adev->umc.retire_unit], 1)) {
 					dev_err(adev->dev, "Failed to save EEPROM table data!");
 					return -EIO;
 				}
@@ -2985,6 +3345,7 @@ int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev,
 		}
 
 		dev_info(adev->dev, "Saved %d pages to EEPROM table.\n", save_count);
+		data->count_saved = data->count;
 	}
 
 	return 0;
@@ -2999,7 +3360,7 @@ static int amdgpu_ras_load_bad_pages(struct amdgpu_device *adev)
 	struct amdgpu_ras_eeprom_control *control =
 		&adev->psp.ras_context.ras->eeprom_control;
 	struct eeprom_table_record *bps;
-	int ret;
+	int ret, i = 0;
 
 	/* no bad page record, skip eeprom access */
 	if (control->ras_num_recs == 0 || amdgpu_bad_page_threshold == 0)
@@ -3013,26 +3374,44 @@ static int amdgpu_ras_load_bad_pages(struct amdgpu_device *adev)
 	if (ret) {
 		dev_err(adev->dev, "Failed to load EEPROM table records!");
 	} else {
-		if (control->ras_num_recs > 1 &&
-		    adev->umc.ras && adev->umc.ras->convert_ras_err_addr) {
-			if ((bps[0].address == bps[1].address) &&
-			    (bps[0].mem_channel == bps[1].mem_channel))
-				control->rec_type = AMDGPU_RAS_EEPROM_REC_PA;
-			else
-				control->rec_type = AMDGPU_RAS_EEPROM_REC_MCA;
+		if (adev->umc.ras && adev->umc.ras->convert_ras_err_addr) {
+			/*In V3, there is no pa recs, and some cases(when address==0) may be parsed
+			as pa recs, so add verion check to avoid it.
+			*/
+			if ((control->tbl_hdr.version < RAS_TABLE_VER_V3) &&
+			    !amdgpu_ras_smu_eeprom_supported(adev)) {
+				for (i = 0; i < control->ras_num_recs; i++) {
+					if ((control->ras_num_recs - i) >= adev->umc.retire_unit) {
+						if ((bps[i].address == bps[i + 1].address) &&
+							(bps[i].mem_channel == bps[i + 1].mem_channel)) {
+							control->ras_num_pa_recs += adev->umc.retire_unit;
+							i += (adev->umc.retire_unit - 1);
+						} else {
+							control->ras_num_mca_recs +=
+										(control->ras_num_recs - i);
+							break;
+						}
+					} else {
+						control->ras_num_mca_recs += (control->ras_num_recs - i);
+						break;
+					}
+				}
+			} else {
+				control->ras_num_mca_recs = control->ras_num_recs;
+			}
 		}
 
+		ret = amdgpu_ras_add_bad_pages(adev, bps, control->ras_num_recs, true);
+		if (ret)
+			goto out;
+
 		ret = amdgpu_ras_eeprom_check(control);
 		if (ret)
 			goto out;
 
 		/* HW not usable */
-		if (amdgpu_ras_is_rma(adev)) {
+		if (amdgpu_ras_is_rma(adev))
 			ret = -EHWPOISON;
-			goto out;
-		}
-
-		ret = amdgpu_ras_add_bad_pages(adev, bps, control->ras_num_recs, true);
 	}
 
 out:
@@ -3040,18 +3419,24 @@ out:
 	return ret;
 }
 
-static bool amdgpu_ras_check_bad_page_unlock(struct amdgpu_ras *con,
+static int amdgpu_ras_check_bad_page_unlock(struct amdgpu_ras *con,
 				uint64_t addr)
 {
 	struct ras_err_handler_data *data = con->eh_data;
+	struct amdgpu_device *adev = con->adev;
 	int i;
 
+	if ((addr >= adev->gmc.mc_vram_size &&
+	    adev->gmc.mc_vram_size) ||
+	    (addr >= RAS_UMC_INJECT_ADDR_LIMIT))
+		return -EINVAL;
+
 	addr >>= AMDGPU_GPU_PAGE_SHIFT;
 	for (i = 0; i < data->count; i++)
 		if (addr == data->bps[i].retired_page)
-			return true;
+			return 1;
 
-	return false;
+	return 0;
 }
 
 /*
@@ -3059,11 +3444,11 @@ static bool amdgpu_ras_check_bad_page_unlock(struct amdgpu_ras *con,
  *
  * Note: this check is only for umc block
  */
-static bool amdgpu_ras_check_bad_page(struct amdgpu_device *adev,
+static int amdgpu_ras_check_bad_page(struct amdgpu_device *adev,
 				uint64_t addr)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
-	bool ret = false;
+	int ret = 0;
 
 	if (!con || !con->eh_data)
 		return ret;
@@ -3080,31 +3465,29 @@ static void amdgpu_ras_validate_threshold(struct amdgpu_device *adev,
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 
 	/*
-	 * Justification of value bad_page_cnt_threshold in ras structure
-	 *
-	 * Generally, 0 <= amdgpu_bad_page_threshold <= max record length
-	 * in eeprom or amdgpu_bad_page_threshold == -2, introduce two
-	 * scenarios accordingly.
-	 *
-	 * Bad page retirement enablement:
-	 *    - If amdgpu_bad_page_threshold = -2,
-	 *      bad_page_cnt_threshold = typical value by formula.
-	 *
-	 *    - When the value from user is 0 < amdgpu_bad_page_threshold <
-	 *      max record length in eeprom, use it directly.
-	 *
-	 * Bad page retirement disablement:
-	 *    - If amdgpu_bad_page_threshold = 0, bad page retirement
-	 *      functionality is disabled, and bad_page_cnt_threshold will
-	 *      take no effect.
+	 * amdgpu_bad_page_threshold is used to config
+	 * the threshold for the number of bad pages.
+	 * -1:  Threshold is set to default value
+	 *      Driver will issue a warning message when threshold is reached
+	 *      and continue runtime services.
+	 * 0:   Disable bad page retirement
+	 *      Driver will not retire bad pages
+	 *      which is intended for debugging purpose.
+	 * -2:  Threshold is determined by a formula
+	 *      that assumes 1 bad page per 100M of local memory.
+	 *      Driver will continue runtime services when threhold is reached.
+	 * 0 < threshold < max number of bad page records in EEPROM,
+	 *      A user-defined threshold is set
+	 *      Driver will halt runtime services when this custom threshold is reached.
 	 */
-
-	if (amdgpu_bad_page_threshold < 0) {
+	if (amdgpu_bad_page_threshold == -2) {
 		u64 val = adev->gmc.mc_vram_size;
 
 		do_div(val, RAS_BAD_PAGE_COVER);
 		con->bad_page_cnt_threshold = min(lower_32_bits(val),
 						  max_count);
+	} else if (amdgpu_bad_page_threshold == -1) {
+		con->bad_page_cnt_threshold = ((con->reserved_pages_in_bytes) >> 21) << 4;
 	} else {
 		con->bad_page_cnt_threshold = min_t(int, max_count,
 						    amdgpu_bad_page_threshold);
@@ -3149,7 +3532,7 @@ static void amdgpu_ras_ecc_log_init(struct ras_ecc_log_info *ecc_log)
 
 	INIT_RADIX_TREE(&ecc_log->de_page_tree, GFP_KERNEL);
 	ecc_log->de_queried_count = 0;
-	ecc_log->prev_de_queried_count = 0;
+	ecc_log->consumption_q_count = 0;
 }
 
 static void amdgpu_ras_ecc_log_fini(struct ras_ecc_log_info *ecc_log)
@@ -3169,7 +3552,7 @@ static void amdgpu_ras_ecc_log_fini(struct ras_ecc_log_info *ecc_log)
 
 	mutex_destroy(&ecc_log->lock);
 	ecc_log->de_queried_count = 0;
-	ecc_log->prev_de_queried_count = 0;
+	ecc_log->consumption_q_count = 0;
 }
 
 static bool amdgpu_ras_schedule_retirement_dwork(struct amdgpu_ras *con,
@@ -3195,7 +3578,6 @@ static void amdgpu_ras_do_page_retirement(struct work_struct *work)
 					      page_retirement_dwork.work);
 	struct amdgpu_device *adev = con->adev;
 	struct ras_err_data err_data;
-	unsigned long err_cnt;
 
 	/* If gpu reset is ongoing, delay retiring the bad pages */
 	if (amdgpu_in_reset(adev) || amdgpu_ras_in_recovery(adev)) {
@@ -3207,13 +3589,9 @@ static void amdgpu_ras_do_page_retirement(struct work_struct *work)
 	amdgpu_ras_error_data_init(&err_data);
 
 	amdgpu_umc_handle_bad_pages(adev, &err_data);
-	err_cnt = err_data.err_addr_cnt;
 
 	amdgpu_ras_error_data_fini(&err_data);
 
-	if (err_cnt && amdgpu_ras_is_rma(adev))
-		amdgpu_ras_reset_gpu(adev);
-
 	amdgpu_ras_schedule_retirement_dwork(con,
 			AMDGPU_RAS_RETIRE_PAGE_INTERVAL);
 }
@@ -3224,58 +3602,39 @@ static int amdgpu_ras_poison_creation_handler(struct amdgpu_device *adev,
 	int ret = 0;
 	struct ras_ecc_log_info *ecc_log;
 	struct ras_query_if info;
-	uint32_t timeout = 0;
+	u32 timeout = MAX_UMC_POISON_POLLING_TIME_ASYNC;
 	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
-	uint64_t de_queried_count;
-	uint32_t new_detect_count, total_detect_count;
-	uint32_t need_query_count = poison_creation_count;
-	bool query_data_timeout = false;
+	u64 de_queried_count;
+	u64 consumption_q_count;
 	enum ras_event_type type = RAS_EVENT_TYPE_POISON_CREATION;
 
 	memset(&info, 0, sizeof(info));
 	info.head.block = AMDGPU_RAS_BLOCK__UMC;
 
 	ecc_log = &ras->umc_ecc_log;
-	total_detect_count = 0;
+	ecc_log->de_queried_count = 0;
+	ecc_log->consumption_q_count = 0;
+
 	do {
 		ret = amdgpu_ras_query_error_status_with_event(adev, &info, type);
 		if (ret)
 			return ret;
 
 		de_queried_count = ecc_log->de_queried_count;
-		if (de_queried_count > ecc_log->prev_de_queried_count) {
-			new_detect_count = de_queried_count - ecc_log->prev_de_queried_count;
-			ecc_log->prev_de_queried_count = de_queried_count;
-			timeout = 0;
-		} else {
-			new_detect_count = 0;
-		}
-
-		if (new_detect_count) {
-			total_detect_count += new_detect_count;
-		} else {
-			if (!timeout && need_query_count)
-				timeout = MAX_UMC_POISON_POLLING_TIME_ASYNC;
+		consumption_q_count = ecc_log->consumption_q_count;
 
-			if (timeout) {
-				if (!--timeout) {
-					query_data_timeout = true;
-					break;
-				}
-				msleep(1);
-			}
-		}
-	} while (total_detect_count < need_query_count);
+		if (de_queried_count && consumption_q_count)
+			break;
 
-	if (query_data_timeout) {
-		dev_warn(adev->dev, "Can't find deferred error! count: %u\n",
-			(need_query_count - total_detect_count));
-		return -ENOENT;
-	}
+		msleep(100);
+	} while (--timeout);
 
-	if (total_detect_count)
+	if (de_queried_count)
 		schedule_delayed_work(&ras->page_retirement_dwork, 0);
 
+	if (amdgpu_ras_is_rma(adev) && atomic_cmpxchg(&ras->rma_in_recovery, 0, 1) == 0)
+		amdgpu_ras_reset_gpu(adev);
+
 	return 0;
 }
 
@@ -3311,6 +3670,12 @@ static int amdgpu_ras_poison_consumption_handler(struct amdgpu_device *adev,
 		reset_flags |= msg.reset;
 	}
 
+	/*
+	 * Try to ensure poison creation handler is completed first
+	 * to set rma if bad page exceed threshold.
+	 */
+	flush_delayed_work(&con->page_retirement_dwork);
+
 	/* for RMA, amdgpu_ras_poison_creation_handler will trigger gpu reset */
 	if (reset_flags && !amdgpu_ras_is_rma(adev)) {
 		if (reset_flags & AMDGPU_RAS_GPU_RESET_MODE1_RESET)
@@ -3320,8 +3685,6 @@ static int amdgpu_ras_poison_consumption_handler(struct amdgpu_device *adev,
 		else
 			reset = reset_flags;
 
-		flush_delayed_work(&con->page_retirement_dwork);
-
 		con->gpu_reset_flags |= reset;
 		amdgpu_ras_reset_gpu(adev);
 
@@ -3351,6 +3714,7 @@ static int amdgpu_ras_page_retirement_thread(void *param)
 		if (kthread_should_stop())
 			break;
 
+		mutex_lock(&con->poison_lock);
 		gpu_reset = 0;
 
 		do {
@@ -3363,7 +3727,8 @@ static int amdgpu_ras_page_retirement_thread(void *param)
 				atomic_sub(poison_creation_count, &con->poison_creation_count);
 				atomic_sub(poison_creation_count, &con->page_retirement_req_cnt);
 			}
-		} while (atomic_read(&con->poison_creation_count));
+		} while (atomic_read(&con->poison_creation_count) &&
+			!atomic_read(&con->poison_consumption_count));
 
 		if (ret != -EIO) {
 			msg_count = kfifo_len(&con->poison_fifo);
@@ -3380,6 +3745,7 @@ static int amdgpu_ras_page_retirement_thread(void *param)
 			/* gpu mode-1 reset is ongoing or just completed ras mode-1 reset */
 			/* Clear poison creation request */
 			atomic_set(&con->poison_creation_count, 0);
+			atomic_set(&con->poison_consumption_count, 0);
 
 			/* Clear poison fifo */
 			amdgpu_ras_clear_poison_fifo(adev);
@@ -3404,9 +3770,12 @@ static int amdgpu_ras_page_retirement_thread(void *param)
 				atomic_sub(msg_count, &con->page_retirement_req_cnt);
 			}
 
+			atomic_set(&con->poison_consumption_count, 0);
+
 			/* Wake up work to save bad pages to eeprom */
 			schedule_delayed_work(&con->page_retirement_dwork, 0);
 		}
+		mutex_unlock(&con->poison_lock);
 	}
 
 	return 0;
@@ -3421,23 +3790,28 @@ int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev)
 	if (!con || amdgpu_sriov_vf(adev))
 		return 0;
 
+	if (amdgpu_uniras_enabled(adev))
+		return 0;
+
 	control = &con->eeprom_control;
+	con->ras_smu_drv = amdgpu_dpm_get_ras_smu_driver(adev);
+
 	ret = amdgpu_ras_eeprom_init(control);
-	if (ret)
-		return ret;
+	control->is_eeprom_valid = !ret;
 
 	if (!adev->umc.ras || !adev->umc.ras->convert_ras_err_addr)
-		control->rec_type = AMDGPU_RAS_EEPROM_REC_PA;
+		control->ras_num_pa_recs = control->ras_num_recs;
 
-	/* default status is MCA storage */
-	if (control->ras_num_recs <= 1 &&
-	    adev->umc.ras && adev->umc.ras->convert_ras_err_addr)
-		control->rec_type = AMDGPU_RAS_EEPROM_REC_MCA;
+	if (adev->umc.ras &&
+	    adev->umc.ras->get_retire_flip_bits)
+		adev->umc.ras->get_retire_flip_bits(adev);
 
-	if (control->ras_num_recs) {
+	if (control->ras_num_recs && control->is_eeprom_valid) {
 		ret = amdgpu_ras_load_bad_pages(adev);
-		if (ret)
-			return ret;
+		if (ret) {
+			control->is_eeprom_valid = false;
+			return 0;
+		}
 
 		amdgpu_dpm_send_hbm_bad_pages_num(
 			adev, control->ras_num_bad_pages);
@@ -3447,9 +3821,16 @@ int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev)
 				adev, control->bad_channel_bitmap);
 			con->update_channel_flag = false;
 		}
+
+		/* The format action is only applied to new ASICs */
+		if (IP_VERSION_MAJ(amdgpu_ip_version(adev, UMC_HWIP, 0)) >= 12 &&
+		    control->tbl_hdr.version < RAS_TABLE_VER_V3)
+			if (!amdgpu_ras_eeprom_reset_table(control))
+				if (amdgpu_ras_save_bad_pages(adev, NULL))
+					dev_warn(adev->dev, "Failed to format RAS EEPROM data in V3 version!\n");
 	}
 
-	return ret;
+	return 0;
 }
 
 int amdgpu_ras_recovery_init(struct amdgpu_device *adev, bool init_bp_info)
@@ -3480,8 +3861,10 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev, bool init_bp_info)
 	}
 
 	mutex_init(&con->recovery_lock);
+	mutex_init(&con->poison_lock);
 	INIT_WORK(&con->recovery_work, amdgpu_ras_do_recovery);
 	atomic_set(&con->in_recovery, 0);
+	atomic_set(&con->rma_in_recovery, 0);
 	con->eeprom_control.bad_channel_bitmap = 0;
 
 	max_eeprom_records_count = amdgpu_ras_eeprom_max_record_count(&con->eeprom_control);
@@ -3499,6 +3882,7 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev, bool init_bp_info)
 	init_waitqueue_head(&con->page_retirement_wq);
 	atomic_set(&con->page_retirement_req_cnt, 0);
 	atomic_set(&con->poison_creation_count, 0);
+	atomic_set(&con->poison_consumption_count, 0);
 	con->page_retirement_thread =
 		kthread_run(amdgpu_ras_page_retirement_thread, adev, "umc_page_retirement");
 	if (IS_ERR(con->page_retirement_thread)) {
@@ -3571,6 +3955,10 @@ static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev)
 	kfree(data);
 	mutex_unlock(&con->recovery_lock);
 
+	amdgpu_ras_critical_region_init(adev);
+#ifdef CONFIG_X86_MCE_AMD
+	amdgpu_unregister_bad_pages_mca_notifier(adev);
+#endif
 	return 0;
 }
 /* recovery end */
@@ -3658,7 +4046,8 @@ static void amdgpu_ras_query_ras_capablity_from_vbios(struct amdgpu_device *adev
 		 */
 		if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(2, 6, 0) ||
 		    amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 0) ||
-		    amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3))
+		    amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3) ||
+		    amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(5, 0, 1))
 			adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN |
 						 1 << AMDGPU_RAS_BLOCK__JPEG);
 		else
@@ -3759,8 +4148,13 @@ init_ras_enabled_flag:
 	adev->ras_enabled = amdgpu_ras_enable == 0 ? 0 :
 		adev->ras_hw_enabled & amdgpu_ras_mask;
 
-	/* aca is disabled by default */
-	adev->aca.is_enabled = false;
+	/* aca is disabled by default except for psp v13_0_6/v13_0_12/v13_0_14 */
+	if (!amdgpu_sriov_vf(adev)) {
+		adev->aca.is_enabled =
+			(amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) ||
+			amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 12) ||
+			amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 14));
+	}
 
 	/* bad page feature is not applicable to specific app platform */
 	if (adev->gmc.is_app_apu &&
@@ -3788,7 +4182,6 @@ static void amdgpu_ras_counte_dw(struct work_struct *work)
 		atomic_set(&con->ras_ue_count, ue_count);
 	}
 
-	pm_runtime_mark_last_busy(dev->dev);
 Out:
 	pm_runtime_put_autosuspend(dev->dev);
 }
@@ -3848,8 +4241,10 @@ static void amdgpu_ras_init_reserved_vram_size(struct amdgpu_device *adev)
 	case IP_VERSION(13, 0, 2):
 	case IP_VERSION(13, 0, 6):
 	case IP_VERSION(13, 0, 12):
+		con->reserved_pages_in_bytes = AMDGPU_RAS_RESERVED_VRAM_SIZE_DEFAULT;
+		break;
 	case IP_VERSION(13, 0, 14):
-		con->reserved_pages_in_bytes = AMDGPU_RAS_RESERVED_VRAM_SIZE;
+		con->reserved_pages_in_bytes = (AMDGPU_RAS_RESERVED_VRAM_SIZE_DEFAULT << 1);
 		break;
 	default:
 		break;
@@ -3989,6 +4384,12 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
 			goto release_con;
 	}
 
+	con->init_task_pid = task_pid_nr(current);
+	get_task_comm(con->init_task_comm, current);
+
+	mutex_init(&con->critical_region_lock);
+	INIT_LIST_HEAD(&con->critical_region_head);
+
 	dev_info(adev->dev, "RAS INFO: ras initialized successfully, "
 		 "hardware ability[%x] ras_mask[%x]\n",
 		 adev->ras_hw_enabled, adev->ras_enabled);
@@ -4268,6 +4669,9 @@ int amdgpu_ras_fini(struct amdgpu_device *adev)
 	if (!adev->ras_enabled || !con)
 		return 0;
 
+	amdgpu_ras_critical_region_fini(adev);
+	mutex_destroy(&con->critical_region_lock);
+
 	list_for_each_entry_safe(ras_node, tmp, &adev->ras_list, node) {
 		if (ras_node->ras_obj) {
 			obj = ras_node->ras_obj;
@@ -4335,8 +4739,10 @@ void amdgpu_ras_clear_err_state(struct amdgpu_device *adev)
 	struct amdgpu_ras *ras;
 
 	ras = amdgpu_ras_get_context(adev);
-	if (ras)
+	if (ras) {
 		ras->ras_err_state = 0;
+		ras->gpu_reset_flags = 0;
+	}
 }
 
 void amdgpu_ras_set_err_poison(struct amdgpu_device *adev,
@@ -4384,6 +4790,9 @@ int amdgpu_ras_mark_ras_event_caller(struct amdgpu_device *adev, enum ras_event_
 	struct ras_event_state *event_state;
 	int ret = 0;
 
+	if (amdgpu_uniras_enabled(adev))
+		return 0;
+
 	if (type >= RAS_EVENT_TYPE_COUNT) {
 		ret = -EINVAL;
 		goto out;
@@ -4434,17 +4843,18 @@ u64 amdgpu_ras_acquire_event_id(struct amdgpu_device *adev, enum ras_event_type
 	return id;
 }
 
-void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev)
+int amdgpu_ras_global_ras_isr(struct amdgpu_device *adev)
 {
 	if (atomic_cmpxchg(&amdgpu_ras_in_intr, 0, 1) == 0) {
 		struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
 		enum ras_event_type type = RAS_EVENT_TYPE_FATAL;
-		u64 event_id;
+		u64 event_id = RAS_EVENT_INVALID_ID;
 
-		if (amdgpu_ras_mark_ras_event(adev, type))
-			return;
+		if (amdgpu_uniras_enabled(adev))
+			return 0;
 
-		event_id = amdgpu_ras_acquire_event_id(adev, type);
+		if (!amdgpu_ras_mark_ras_event(adev, type))
+			event_id = amdgpu_ras_acquire_event_id(adev, type);
 
 		RAS_EVENT_LOG(adev, event_id, "uncorrectable hardware error"
 			      "(ERREVENT_ATHUB_INTERRUPT) detected!\n");
@@ -4453,6 +4863,8 @@ void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev)
 		ras->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE1_RESET;
 		amdgpu_ras_reset_gpu(adev);
 	}
+
+	return -EBUSY;
 }
 
 bool amdgpu_ras_need_emergency_restart(struct amdgpu_device *adev)
@@ -4580,6 +4992,28 @@ static void amdgpu_register_bad_pages_mca_notifier(struct amdgpu_device *adev)
 		notifier_registered = true;
 	}
 }
+static void amdgpu_unregister_bad_pages_mca_notifier(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	if (!notifier_registered && !mce_adev_list.num_gpu)
+		return;
+	for (i = 0, j = 0; i < mce_adev_list.num_gpu; i++) {
+		if (mce_adev_list.devs[i] == adev)
+			mce_adev_list.devs[i] = NULL;
+		if (!mce_adev_list.devs[i])
+			++j;
+	}
+
+	if (j == mce_adev_list.num_gpu) {
+		mce_adev_list.num_gpu = 0;
+		/* Unregister x86 notifier with MCE subsystem. */
+		if (notifier_registered) {
+			mce_unregister_decode_chain(&amdgpu_bad_page_nb);
+			notifier_registered = false;
+		}
+	}
+}
 #endif
 
 struct amdgpu_ras *amdgpu_ras_get_context(struct amdgpu_device *adev)
@@ -5127,9 +5561,9 @@ static void amdgpu_ras_boot_time_error_reporting(struct amdgpu_device *adev,
 			 "socket: %d, aid: %d, fw_status: 0x%x, data abort exception\n",
 			 socket_id, aid_id, fw_status);
 
-	if (AMDGPU_RAS_GPU_ERR_UNKNOWN(boot_error))
+	if (AMDGPU_RAS_GPU_ERR_GENERIC(boot_error))
 		dev_info(adev->dev,
-			 "socket: %d, aid: %d, fw_status: 0x%x, unknown boot time errors\n",
+			 "socket: %d, aid: %d, fw_status: 0x%x, Boot Controller Generic Error\n",
 			 socket_id, aid_id, fw_status);
 }
 
@@ -5171,6 +5605,9 @@ int amdgpu_ras_reserve_page(struct amdgpu_device *adev, uint64_t pfn)
 	uint64_t start = pfn << AMDGPU_GPU_PAGE_SHIFT;
 	int ret = 0;
 
+	if (amdgpu_ras_check_critical_address(adev, start))
+		return 0;
+
 	mutex_lock(&con->page_rsv_lock);
 	ret = amdgpu_vram_mgr_query_page_status(mgr, start);
 	if (ret == -ENOENT)
@@ -5202,8 +5639,110 @@ bool amdgpu_ras_is_rma(struct amdgpu_device *adev)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 
+	if (amdgpu_uniras_enabled(adev))
+		return amdgpu_ras_mgr_is_rma(adev);
+
 	if (!con)
 		return false;
 
 	return con->is_rma;
 }
+
+int amdgpu_ras_add_critical_region(struct amdgpu_device *adev,
+			struct amdgpu_bo *bo)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct amdgpu_vram_mgr_resource *vres;
+	struct ras_critical_region *region;
+	struct drm_buddy_block *block;
+	int ret = 0;
+
+	if (!bo || !bo->tbo.resource)
+		return -EINVAL;
+
+	vres = to_amdgpu_vram_mgr_resource(bo->tbo.resource);
+
+	mutex_lock(&con->critical_region_lock);
+
+	/* Check if the bo had been recorded */
+	list_for_each_entry(region, &con->critical_region_head, node)
+		if (region->bo == bo)
+			goto out;
+
+	/* Record new critical amdgpu bo */
+	list_for_each_entry(block, &vres->blocks, link) {
+		region = kzalloc(sizeof(*region), GFP_KERNEL);
+		if (!region) {
+			ret = -ENOMEM;
+			goto out;
+		}
+		region->bo = bo;
+		region->start = amdgpu_vram_mgr_block_start(block);
+		region->size = amdgpu_vram_mgr_block_size(block);
+		list_add_tail(&region->node, &con->critical_region_head);
+	}
+
+out:
+	mutex_unlock(&con->critical_region_lock);
+
+	return ret;
+}
+
+static void amdgpu_ras_critical_region_init(struct amdgpu_device *adev)
+{
+	amdgpu_ras_add_critical_region(adev, adev->mman.fw_reserved_memory);
+}
+
+static void amdgpu_ras_critical_region_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_critical_region *region, *tmp;
+
+	mutex_lock(&con->critical_region_lock);
+	list_for_each_entry_safe(region, tmp, &con->critical_region_head, node) {
+		list_del(&region->node);
+		kfree(region);
+	}
+	mutex_unlock(&con->critical_region_lock);
+}
+
+bool amdgpu_ras_check_critical_address(struct amdgpu_device *adev, uint64_t addr)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct ras_critical_region *region;
+	bool ret = false;
+
+	mutex_lock(&con->critical_region_lock);
+	list_for_each_entry(region, &con->critical_region_head, node) {
+		if ((region->start <= addr) &&
+		    (addr < (region->start + region->size))) {
+			ret = true;
+			break;
+		}
+	}
+	mutex_unlock(&con->critical_region_lock);
+
+	return ret;
+}
+
+void amdgpu_ras_pre_reset(struct amdgpu_device *adev,
+					  struct list_head *device_list)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
+		if (amdgpu_uniras_enabled(tmp_adev))
+			amdgpu_ras_mgr_pre_reset(tmp_adev);
+	}
+}
+
+void amdgpu_ras_post_reset(struct amdgpu_device *adev,
+					  struct list_head *device_list)
+{
+	struct amdgpu_device *tmp_adev = NULL;
+
+	list_for_each_entry(tmp_adev, device_list, reset_list) {
+		if (amdgpu_uniras_enabled(tmp_adev))
+			amdgpu_ras_mgr_post_reset(tmp_adev);
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
index 82db986c36a0..ff44190d7d98 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
@@ -47,7 +47,7 @@ struct amdgpu_iv_entry;
 #define AMDGPU_RAS_GPU_ERR_AID_ID(x)			AMDGPU_GET_REG_FIELD(x, 12, 11)
 #define AMDGPU_RAS_GPU_ERR_HBM_ID(x)			AMDGPU_GET_REG_FIELD(x, 14, 13)
 #define AMDGPU_RAS_GPU_ERR_DATA_ABORT(x)		AMDGPU_GET_REG_FIELD(x, 29, 29)
-#define AMDGPU_RAS_GPU_ERR_UNKNOWN(x)			AMDGPU_GET_REG_FIELD(x, 30, 30)
+#define AMDGPU_RAS_GPU_ERR_GENERIC(x)			AMDGPU_GET_REG_FIELD(x, 30, 30)
 
 #define AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT	100
 #define AMDGPU_RAS_BOOT_STEADY_STATUS		0xBA
@@ -65,7 +65,7 @@ struct amdgpu_iv_entry;
 
 /* Reserve 8 physical dram row for possible retirement.
  * In worst cases, it will lose 8 * 2MB memory in vram domain */
-#define AMDGPU_RAS_RESERVED_VRAM_SIZE	(16ULL << 20)
+#define AMDGPU_RAS_RESERVED_VRAM_SIZE_DEFAULT	(16ULL << 20)
 /* The high three bits indicates socketid */
 #define AMDGPU_RAS_GET_FEATURES(val)  ((val) & ~AMDGPU_RAS_FEATURES_SOCKETID_MASK)
 
@@ -98,6 +98,7 @@ enum amdgpu_ras_block {
 	AMDGPU_RAS_BLOCK__JPEG,
 	AMDGPU_RAS_BLOCK__IH,
 	AMDGPU_RAS_BLOCK__MPIO,
+	AMDGPU_RAS_BLOCK__MMSCH,
 
 	AMDGPU_RAS_BLOCK__LAST,
 	AMDGPU_RAS_BLOCK__ANY = -1
@@ -491,11 +492,45 @@ struct ras_ecc_err {
 struct ras_ecc_log_info {
 	struct mutex lock;
 	struct radix_tree_root de_page_tree;
-	uint64_t	de_queried_count;
-	uint64_t	prev_de_queried_count;
+	uint64_t de_queried_count;
+	uint64_t consumption_q_count;
+};
+
+struct ras_critical_region {
+	struct list_head node;
+	struct amdgpu_bo *bo;
+	uint64_t start;
+	uint64_t size;
+};
+
+struct ras_eeprom_table_version {
+	uint32_t minor    : 16;
+	uint32_t major    : 16;
+};
+
+struct ras_eeprom_smu_funcs {
+	int (*get_ras_table_version)(struct amdgpu_device *adev,
+							uint32_t *table_version);
+	int (*get_badpage_count)(struct amdgpu_device *adev, uint32_t *count, uint32_t timeout);
+	int (*get_badpage_mca_addr)(struct amdgpu_device *adev, uint16_t index, uint64_t *mca_addr);
+	int (*set_timestamp)(struct amdgpu_device *adev, uint64_t timestamp);
+	int (*get_timestamp)(struct amdgpu_device *adev,
+							uint16_t index, uint64_t *timestamp);
+	int (*get_badpage_ipid)(struct amdgpu_device *adev, uint16_t index, uint64_t *ipid);
+	int (*erase_ras_table)(struct amdgpu_device *adev, uint32_t *result);
+};
+
+enum ras_smu_feature_flags {
+	RAS_SMU_FEATURE_BIT__RAS_EEPROM = BIT_ULL(0),
+};
+
+struct ras_smu_drv {
+	const struct ras_eeprom_smu_funcs *smu_eeprom_funcs;
+	void (*ras_smu_feature_flags)(struct amdgpu_device *adev, uint64_t *flags);
 };
 
 struct amdgpu_ras {
+	void *ras_mgr;
 	/* ras infrastructure */
 	/* for ras itself. */
 	uint32_t features;
@@ -514,6 +549,7 @@ struct amdgpu_ras {
 	/* gpu recovery */
 	struct work_struct recovery_work;
 	atomic_t in_recovery;
+	atomic_t rma_in_recovery;
 	struct amdgpu_device *adev;
 	/* error handler data */
 	struct ras_err_handler_data *eh_data;
@@ -556,6 +592,7 @@ struct amdgpu_ras {
 	struct mutex page_retirement_lock;
 	atomic_t page_retirement_req_cnt;
 	atomic_t poison_creation_count;
+	atomic_t poison_consumption_count;
 	struct mutex page_rsv_lock;
 	DECLARE_KFIFO(poison_fifo, struct ras_poison_msg, 128);
 	struct ras_ecc_log_info  umc_ecc_log;
@@ -569,6 +606,21 @@ struct amdgpu_ras {
 	struct ras_event_manager *event_mgr;
 
 	uint64_t reserved_pages_in_bytes;
+
+	pid_t init_task_pid;
+	char init_task_comm[TASK_COMM_LEN];
+
+	int bad_page_num;
+
+	struct list_head critical_region_head;
+	struct mutex critical_region_lock;
+
+	/* Protect poison injection */
+	struct mutex poison_lock;
+
+	/* Disable/Enable uniras switch */
+	bool uniras_enabled;
+	const struct ras_smu_drv *ras_smu_drv;
 };
 
 struct ras_fs_data {
@@ -607,6 +659,7 @@ struct ras_err_handler_data {
 	struct eeprom_table_record *bps;
 	/* the count of entries */
 	int count;
+	int count_saved;
 	/* the space can place new entries */
 	int space_left;
 };
@@ -795,6 +848,12 @@ amdgpu_ras_block_to_ta(enum amdgpu_ras_block block) {
 		return TA_RAS_BLOCK__VCN;
 	case AMDGPU_RAS_BLOCK__JPEG:
 		return TA_RAS_BLOCK__JPEG;
+	case AMDGPU_RAS_BLOCK__IH:
+		return TA_RAS_BLOCK__IH;
+	case AMDGPU_RAS_BLOCK__MPIO:
+		return TA_RAS_BLOCK__MPIO;
+	case AMDGPU_RAS_BLOCK__MMSCH:
+		return TA_RAS_BLOCK__MMSCH;
 	default:
 		WARN_ONCE(1, "RAS ERROR: unexpected block id %d\n", block);
 		return TA_RAS_BLOCK__UMC;
@@ -881,7 +940,7 @@ static inline void amdgpu_ras_intr_cleared(void)
 	atomic_set(&amdgpu_ras_in_intr, 0);
 }
 
-void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev);
+int amdgpu_ras_global_ras_isr(struct amdgpu_device *adev);
 
 void amdgpu_ras_set_error_query_ready(struct amdgpu_device *adev, bool ready);
 
@@ -966,6 +1025,9 @@ int amdgpu_ras_mark_ras_event_caller(struct amdgpu_device *adev, enum ras_event_
 
 int amdgpu_ras_reserve_page(struct amdgpu_device *adev, uint64_t pfn);
 
+int amdgpu_ras_add_critical_region(struct amdgpu_device *adev, struct amdgpu_bo *bo);
+bool amdgpu_ras_check_critical_address(struct amdgpu_device *adev, uint64_t addr);
+
 int amdgpu_ras_put_poison_req(struct amdgpu_device *adev,
 		enum amdgpu_ras_block block, uint16_t pasid,
 		pasid_notify pasid_fn, void *data, uint32_t reset);
@@ -977,4 +1039,9 @@ void amdgpu_ras_event_log_print(struct amdgpu_device *adev, u64 event_id,
 				const char *fmt, ...);
 
 bool amdgpu_ras_is_rma(struct amdgpu_device *adev);
+
+void amdgpu_ras_pre_reset(struct amdgpu_device *adev,
+					  struct list_head *device_list);
+void amdgpu_ras_post_reset(struct amdgpu_device *adev,
+					  struct list_head *device_list);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
index 52c16bfeccaa..64dd7a81bff5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
@@ -32,6 +32,7 @@
 #include <linux/uaccess.h>
 
 #include "amdgpu_reset.h"
+#include "amdgpu_ras_mgr.h"
 
 /* These are memory addresses as would be seen by one or more EEPROM
  * chips strung on the I2C bus, usually by manipulating pins 1-3 of a
@@ -123,6 +124,8 @@
 					RAS_TABLE_V2_1_INFO_SIZE) \
 					/ RAS_TABLE_RECORD_SIZE)
 
+#define RAS_SMU_MESSAGE_TIMEOUT_MS 1000 /* 1s */
+
 /* Given a zero-based index of an EEPROM RAS record, yields the EEPROM
  * offset off of RAS_TABLE_START.  That is, this is something you can
  * add to control->i2c_address, and then tell I2C layer to read
@@ -161,6 +164,7 @@ static bool __is_ras_eeprom_supported(struct amdgpu_device *adev)
 	case IP_VERSION(13, 0, 10):
 		return true;
 	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 12):
 	case IP_VERSION(13, 0, 14):
 		return (adev->gmc.is_app_apu) ? false : true;
 	default:
@@ -177,7 +181,7 @@ static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev,
 	if (!control)
 		return false;
 
-	if (amdgpu_atomfirmware_ras_rom_addr(adev, &i2c_addr)) {
+	if (adev->bios && amdgpu_atomfirmware_ras_rom_addr(adev, &i2c_addr)) {
 		/* The address given by VBIOS is an 8-bit, wire-format
 		 * address, i.e. the most significant byte.
 		 *
@@ -223,6 +227,7 @@ static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev,
 		return true;
 	case IP_VERSION(13, 0, 6):
 	case IP_VERSION(13, 0, 10):
+	case IP_VERSION(13, 0, 12):
 	case IP_VERSION(13, 0, 14):
 		control->i2c_address = EEPROM_I2C_MADDR_4;
 		return true;
@@ -275,10 +280,11 @@ static int __write_table_header(struct amdgpu_ras_eeprom_control *control)
 	up_read(&adev->reset_domain->sem);
 
 	if (res < 0) {
-		DRM_ERROR("Failed to write EEPROM table header:%d", res);
+		dev_err(adev->dev, "Failed to write EEPROM table header:%d",
+			res);
 	} else if (res < RAS_TABLE_HEADER_SIZE) {
-		DRM_ERROR("Short write:%d out of %d\n",
-			  res, RAS_TABLE_HEADER_SIZE);
+		dev_err(adev->dev, "Short write:%d out of %d\n", res,
+			RAS_TABLE_HEADER_SIZE);
 		res = -EIO;
 	} else {
 		res = 0;
@@ -321,7 +327,8 @@ static int __write_table_ras_info(struct amdgpu_ras_eeprom_control *control)
 
 	buf = kzalloc(RAS_TABLE_V2_1_INFO_SIZE, GFP_KERNEL);
 	if (!buf) {
-		DRM_ERROR("Failed to alloc buf to write table ras info\n");
+		dev_err(adev->dev,
+			"Failed to alloc buf to write table ras info\n");
 		return -ENOMEM;
 	}
 
@@ -336,10 +343,11 @@ static int __write_table_ras_info(struct amdgpu_ras_eeprom_control *control)
 	up_read(&adev->reset_domain->sem);
 
 	if (res < 0) {
-		DRM_ERROR("Failed to write EEPROM table ras info:%d", res);
+		dev_err(adev->dev, "Failed to write EEPROM table ras info:%d",
+			res);
 	} else if (res < RAS_TABLE_V2_1_INFO_SIZE) {
-		DRM_ERROR("Short write:%d out of %d\n",
-			  res, RAS_TABLE_V2_1_INFO_SIZE);
+		dev_err(adev->dev, "Short write:%d out of %d\n", res,
+			RAS_TABLE_V2_1_INFO_SIZE);
 		res = -EIO;
 	} else {
 		res = 0;
@@ -413,9 +421,12 @@ static void amdgpu_ras_set_eeprom_table_version(struct amdgpu_ras_eeprom_control
 
 	switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) {
 	case IP_VERSION(8, 10, 0):
-	case IP_VERSION(12, 0, 0):
 		hdr->version = RAS_TABLE_VER_V2_1;
 		return;
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 5, 0):
+		hdr->version = RAS_TABLE_VER_V3;
+		return;
 	default:
 		hdr->version = RAS_TABLE_VER_V1;
 		return;
@@ -435,42 +446,61 @@ int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control)
 	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
 	struct amdgpu_ras_eeprom_table_ras_info *rai = &control->tbl_rai;
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	u32 erase_res = 0;
 	u8 csum;
 	int res;
 
 	mutex_lock(&control->ras_tbl_mutex);
 
-	hdr->header = RAS_TABLE_HDR_VAL;
-	amdgpu_ras_set_eeprom_table_version(control);
+	if (!amdgpu_ras_smu_eeprom_supported(adev)) {
+		hdr->header = RAS_TABLE_HDR_VAL;
+		amdgpu_ras_set_eeprom_table_version(control);
+
+		if (hdr->version >= RAS_TABLE_VER_V2_1) {
+			hdr->first_rec_offset = RAS_RECORD_START_V2_1;
+			hdr->tbl_size = RAS_TABLE_HEADER_SIZE +
+					RAS_TABLE_V2_1_INFO_SIZE;
+			rai->rma_status = GPU_HEALTH_USABLE;
+
+			control->ras_record_offset = RAS_RECORD_START_V2_1;
+			control->ras_max_record_count = RAS_MAX_RECORD_COUNT_V2_1;
+			/**
+			 * GPU health represented as a percentage.
+			 * 0 means worst health, 100 means fully health.
+			 */
+			rai->health_percent = 100;
+			/* ecc_page_threshold = 0 means disable bad page retirement */
+			rai->ecc_page_threshold = con->bad_page_cnt_threshold;
+		} else {
+			hdr->first_rec_offset = RAS_RECORD_START;
+			hdr->tbl_size = RAS_TABLE_HEADER_SIZE;
 
-	if (hdr->version == RAS_TABLE_VER_V2_1) {
-		hdr->first_rec_offset = RAS_RECORD_START_V2_1;
-		hdr->tbl_size = RAS_TABLE_HEADER_SIZE +
-				RAS_TABLE_V2_1_INFO_SIZE;
-		rai->rma_status = GPU_HEALTH_USABLE;
-		/**
-		 * GPU health represented as a percentage.
-		 * 0 means worst health, 100 means fully health.
-		 */
-		rai->health_percent = 100;
-		/* ecc_page_threshold = 0 means disable bad page retirement */
-		rai->ecc_page_threshold = con->bad_page_cnt_threshold;
+			control->ras_record_offset = RAS_RECORD_START;
+			control->ras_max_record_count = RAS_MAX_RECORD_COUNT;
+		}
+
+		csum = __calc_hdr_byte_sum(control);
+		if (hdr->version >= RAS_TABLE_VER_V2_1)
+			csum += __calc_ras_info_byte_sum(control);
+		csum = -csum;
+		hdr->checksum = csum;
+		res = __write_table_header(control);
+		if (!res && hdr->version > RAS_TABLE_VER_V1)
+			res = __write_table_ras_info(control);
 	} else {
-		hdr->first_rec_offset = RAS_RECORD_START;
-		hdr->tbl_size = RAS_TABLE_HEADER_SIZE;
+		res = amdgpu_ras_smu_erase_ras_table(adev, &erase_res);
+		if (res || erase_res) {
+			dev_warn(adev->dev, "RAS EEPROM reset failed, res:%d result:%d",
+										res, erase_res);
+			if (!res)
+				res = -EIO;
+		}
 	}
 
-	csum = __calc_hdr_byte_sum(control);
-	if (hdr->version == RAS_TABLE_VER_V2_1)
-		csum += __calc_ras_info_byte_sum(control);
-	csum = -csum;
-	hdr->checksum = csum;
-	res = __write_table_header(control);
-	if (!res && hdr->version > RAS_TABLE_VER_V1)
-		res = __write_table_ras_info(control);
-
 	control->ras_num_recs = 0;
 	control->ras_num_bad_pages = 0;
+	control->ras_num_mca_recs = 0;
+	control->ras_num_pa_recs = 0;
 	control->ras_fri = 0;
 
 	amdgpu_dpm_send_hbm_bad_pages_num(adev, control->ras_num_bad_pages);
@@ -546,6 +576,9 @@ bool amdgpu_ras_eeprom_check_err_threshold(struct amdgpu_device *adev)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 
+	if (amdgpu_uniras_enabled(adev))
+		return amdgpu_ras_mgr_check_eeprom_safety_watermark(adev);
+
 	if (!__is_ras_eeprom_supported(adev) ||
 	    !amdgpu_bad_page_threshold)
 		return false;
@@ -558,16 +591,17 @@ bool amdgpu_ras_eeprom_check_err_threshold(struct amdgpu_device *adev)
 			return false;
 
 	if (con->eeprom_control.tbl_hdr.header == RAS_TABLE_HDR_BAD) {
-		if (amdgpu_bad_page_threshold == -1) {
+		if (con->eeprom_control.ras_num_bad_pages > con->bad_page_cnt_threshold)
 			dev_warn(adev->dev, "RAS records:%d exceed threshold:%d",
-				con->eeprom_control.ras_num_bad_pages, con->bad_page_cnt_threshold);
+				 con->eeprom_control.ras_num_bad_pages, con->bad_page_cnt_threshold);
+		if ((amdgpu_bad_page_threshold == -1) ||
+		    (amdgpu_bad_page_threshold == -2)) {
 			dev_warn(adev->dev,
-				"But GPU can be operated due to bad_page_threshold = -1.\n");
+				 "Please consult AMD Service Action Guide (SAG) for appropriate service procedures.\n");
 			return false;
 		} else {
-			dev_warn(adev->dev, "This GPU is in BAD status.");
-			dev_warn(adev->dev, "Please retire it or set a larger "
-				 "threshold value when reloading driver.\n");
+			dev_warn(adev->dev,
+				 "Please consider adjusting the customized threshold.\n");
 			return true;
 		}
 	}
@@ -601,13 +635,13 @@ static int __amdgpu_ras_eeprom_write(struct amdgpu_ras_eeprom_control *control,
 				  buf, buf_size);
 	up_read(&adev->reset_domain->sem);
 	if (res < 0) {
-		DRM_ERROR("Writing %d EEPROM table records error:%d",
-			  num, res);
+		dev_err(adev->dev, "Writing %d EEPROM table records error:%d",
+			num, res);
 	} else if (res < buf_size) {
 		/* Short write, return error.
 		 */
-		DRM_ERROR("Wrote %d records out of %d",
-			  res / RAS_TABLE_RECORD_SIZE, num);
+		dev_err(adev->dev, "Wrote %d records out of %d",
+			res / RAS_TABLE_RECORD_SIZE, num);
 		res = -EIO;
 	} else {
 		res = 0;
@@ -726,11 +760,13 @@ amdgpu_ras_eeprom_append_table(struct amdgpu_ras_eeprom_control *control,
 				     - control->ras_fri)
 		% control->ras_max_record_count;
 
-	if (control->rec_type == AMDGPU_RAS_EEPROM_REC_PA)
-		control->ras_num_bad_pages = control->ras_num_recs;
+	/*old asics only save pa to eeprom like before*/
+	if (IP_VERSION_MAJ(amdgpu_ip_version(adev, UMC_HWIP, 0)) < 12)
+		control->ras_num_pa_recs += num;
 	else
-		control->ras_num_bad_pages =
-			control->ras_num_recs * adev->umc.retire_unit;
+		control->ras_num_mca_recs += num;
+
+	control->ras_num_bad_pages = con->bad_page_num;
 Out:
 	kfree(buf);
 	return res;
@@ -748,24 +784,30 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control)
 	/* Modify the header if it exceeds.
 	 */
 	if (amdgpu_bad_page_threshold != 0 &&
-	    control->ras_num_bad_pages >= ras->bad_page_cnt_threshold) {
+	    control->ras_num_bad_pages > ras->bad_page_cnt_threshold) {
 		dev_warn(adev->dev,
 			"Saved bad pages %d reaches threshold value %d\n",
 			control->ras_num_bad_pages, ras->bad_page_cnt_threshold);
-		control->tbl_hdr.header = RAS_TABLE_HDR_BAD;
-		if (control->tbl_hdr.version == RAS_TABLE_VER_V2_1) {
-			control->tbl_rai.rma_status = GPU_RETIRED__ECC_REACH_THRESHOLD;
-			control->tbl_rai.health_percent = 0;
-		}
 
-		if (amdgpu_bad_page_threshold != -1)
+		if (adev->cper.enabled && !amdgpu_uniras_enabled(adev) &&
+		    amdgpu_cper_generate_bp_threshold_record(adev))
+			dev_warn(adev->dev, "fail to generate bad page threshold cper records\n");
+
+		if ((amdgpu_bad_page_threshold != -1) &&
+		    (amdgpu_bad_page_threshold != -2)) {
+			control->tbl_hdr.header = RAS_TABLE_HDR_BAD;
+			if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1) {
+				control->tbl_rai.rma_status = GPU_RETIRED__ECC_REACH_THRESHOLD;
+				control->tbl_rai.health_percent = 0;
+			}
 			ras->is_rma = true;
+		}
 
 		/* ignore the -ENOTSUPP return value */
 		amdgpu_dpm_send_rma_reason(adev);
 	}
 
-	if (control->tbl_hdr.version == RAS_TABLE_VER_V2_1)
+	if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1)
 		control->tbl_hdr.tbl_size = RAS_TABLE_HEADER_SIZE +
 					    RAS_TABLE_V2_1_INFO_SIZE +
 					    control->ras_num_recs * RAS_TABLE_RECORD_SIZE;
@@ -777,8 +819,9 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control)
 	buf_size = control->ras_num_recs * RAS_TABLE_RECORD_SIZE;
 	buf = kcalloc(control->ras_num_recs, RAS_TABLE_RECORD_SIZE, GFP_KERNEL);
 	if (!buf) {
-		DRM_ERROR("allocating memory for table of size %d bytes failed\n",
-			  control->tbl_hdr.tbl_size);
+		dev_err(adev->dev,
+			"allocating memory for table of size %d bytes failed\n",
+			control->tbl_hdr.tbl_size);
 		res = -ENOMEM;
 		goto Out;
 	}
@@ -790,12 +833,11 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control)
 				 buf, buf_size);
 	up_read(&adev->reset_domain->sem);
 	if (res < 0) {
-		DRM_ERROR("EEPROM failed reading records:%d\n",
-			  res);
+		dev_err(adev->dev, "EEPROM failed reading records:%d\n", res);
 		goto Out;
 	} else if (res < buf_size) {
-		DRM_ERROR("EEPROM read %d out of %d bytes\n",
-			  res, buf_size);
+		dev_err(adev->dev, "EEPROM read %d out of %d bytes\n", res,
+			buf_size);
 		res = -EIO;
 		goto Out;
 	}
@@ -805,8 +847,8 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control)
 	 * now calculate gpu health percent
 	 */
 	if (amdgpu_bad_page_threshold != 0 &&
-	    control->tbl_hdr.version == RAS_TABLE_VER_V2_1 &&
-	    control->ras_num_bad_pages < ras->bad_page_cnt_threshold)
+	    control->tbl_hdr.version >= RAS_TABLE_VER_V2_1 &&
+	    control->ras_num_bad_pages <= ras->bad_page_cnt_threshold)
 		control->tbl_rai.health_percent = ((ras->bad_page_cnt_threshold -
 						   control->ras_num_bad_pages) * 100) /
 						   ras->bad_page_cnt_threshold;
@@ -818,7 +860,7 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control)
 		csum += *pp;
 
 	csum += __calc_hdr_byte_sum(control);
-	if (control->tbl_hdr.version == RAS_TABLE_VER_V2_1)
+	if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1)
 		csum += __calc_ras_info_byte_sum(control);
 	/* avoid sign extension when assigning to "checksum" */
 	csum = -csum;
@@ -831,6 +873,71 @@ Out:
 	return res;
 }
 
+int amdgpu_ras_eeprom_update_record_num(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	int ret, retry = 20;
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev))
+		return 0;
+
+	control->ras_num_recs_old = control->ras_num_recs;
+
+	do {
+		/* 1000ms timeout is long enough, smu_get_badpage_count won't
+		 * return -EBUSY before timeout.
+		 */
+		ret = amdgpu_ras_smu_get_badpage_count(adev,
+			&(control->ras_num_recs), RAS_SMU_MESSAGE_TIMEOUT_MS);
+		if (!ret &&
+		    (control->ras_num_recs_old == control->ras_num_recs)) {
+			/* record number update in PMFW needs some time,
+			 * smu_get_badpage_count may return immediately without
+			 * count update, sleep for a while and retry again.
+			 */
+			msleep(50);
+			retry--;
+		} else {
+			break;
+		}
+	} while (retry);
+
+	/* no update of record number is not a real failure,
+	 * don't print warning here
+	 */
+	if (!ret && (control->ras_num_recs_old == control->ras_num_recs))
+		ret = -EINVAL;
+
+	return ret;
+}
+
+static int amdgpu_ras_smu_eeprom_append(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev) || !con)
+		return 0;
+
+	control->ras_num_bad_pages = con->bad_page_num;
+
+	if (amdgpu_bad_page_threshold != 0 &&
+	    control->ras_num_bad_pages > con->bad_page_cnt_threshold) {
+		dev_warn(adev->dev,
+			"Saved bad pages %d reaches threshold value %d\n",
+			control->ras_num_bad_pages, con->bad_page_cnt_threshold);
+
+		if (adev->cper.enabled && amdgpu_cper_generate_bp_threshold_record(adev))
+			dev_warn(adev->dev, "fail to generate bad page threshold cper records\n");
+
+		if ((amdgpu_bad_page_threshold != -1) &&
+		    (amdgpu_bad_page_threshold != -2))
+			con->is_rma = true;
+	}
+
+	return 0;
+}
+
 /**
  * amdgpu_ras_eeprom_append -- append records to the EEPROM RAS table
  * @control: pointer to control structure
@@ -850,22 +957,30 @@ int amdgpu_ras_eeprom_append(struct amdgpu_ras_eeprom_control *control,
 {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
 	int res, i;
+	uint64_t nps = AMDGPU_NPS1_PARTITION_MODE;
 
 	if (!__is_ras_eeprom_supported(adev))
 		return 0;
 
+	if (amdgpu_ras_smu_eeprom_supported(adev))
+		return amdgpu_ras_smu_eeprom_append(control);
+
 	if (num == 0) {
-		DRM_ERROR("will not append 0 records\n");
+		dev_err(adev->dev, "will not append 0 records\n");
 		return -EINVAL;
 	} else if (num > control->ras_max_record_count) {
-		DRM_ERROR("cannot append %d records than the size of table %d\n",
-			  num, control->ras_max_record_count);
+		dev_err(adev->dev,
+			"cannot append %d records than the size of table %d\n",
+			num, control->ras_max_record_count);
 		return -EINVAL;
 	}
 
+	if (adev->gmc.gmc_funcs->query_mem_partition_mode)
+		nps = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+
 	/* set the new channel index flag */
 	for (i = 0; i < num; i++)
-		record[i].retired_page |= UMC_CHANNEL_IDX_V2;
+		record[i].retired_page |= (nps << UMC_NPS_SHIFT);
 
 	mutex_lock(&control->ras_tbl_mutex);
 
@@ -879,7 +994,7 @@ int amdgpu_ras_eeprom_append(struct amdgpu_ras_eeprom_control *control,
 
 	/* clear channel index flag, the flag is only saved on eeprom */
 	for (i = 0; i < num; i++)
-		record[i].retired_page &= ~UMC_CHANNEL_IDX_V2;
+		record[i].retired_page &= ~(nps << UMC_NPS_SHIFT);
 
 	return res;
 }
@@ -910,13 +1025,13 @@ static int __amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
 				 buf, buf_size);
 	up_read(&adev->reset_domain->sem);
 	if (res < 0) {
-		DRM_ERROR("Reading %d EEPROM table records error:%d",
-			  num, res);
+		dev_err(adev->dev, "Reading %d EEPROM table records error:%d",
+			num, res);
 	} else if (res < buf_size) {
 		/* Short read, return error.
 		 */
-		DRM_ERROR("Read %d records out of %d",
-			  res / RAS_TABLE_RECORD_SIZE, num);
+		dev_err(adev->dev, "Read %d records out of %d",
+			res / RAS_TABLE_RECORD_SIZE, num);
 		res = -EIO;
 	} else {
 		res = 0;
@@ -925,6 +1040,50 @@ static int __amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
 	return res;
 }
 
+int amdgpu_ras_eeprom_read_idx(struct amdgpu_ras_eeprom_control *control,
+			struct eeprom_table_record *record, u32 rec_idx,
+			const u32 num)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	uint64_t ts, end_idx;
+	int i, ret;
+	u64 mca, ipid;
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev))
+		return 0;
+
+	if (!adev->umc.ras || !adev->umc.ras->mca_ipid_parse)
+		return -EOPNOTSUPP;
+
+	end_idx = rec_idx + num;
+	for (i = rec_idx; i < end_idx; i++) {
+		ret = amdgpu_ras_smu_get_badpage_mca_addr(adev, i, &mca);
+		if (ret)
+			return ret;
+
+		ret = amdgpu_ras_smu_get_badpage_ipid(adev, i, &ipid);
+		if (ret)
+			return ret;
+
+		ret = amdgpu_ras_smu_get_timestamp(adev, i, &ts);
+		if (ret)
+			return ret;
+
+		record[i - rec_idx].address = mca;
+		/* retired_page (pa) is unused now */
+		record[i - rec_idx].retired_page = 0x1ULL;
+		record[i - rec_idx].ts = ts;
+		record[i - rec_idx].err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE;
+
+		adev->umc.ras->mca_ipid_parse(adev, ipid,
+			(uint32_t *)&(record[i - rec_idx].cu),
+			(uint32_t *)&(record[i - rec_idx].mem_channel),
+			(uint32_t *)&(record[i - rec_idx].mcumc_id), NULL);
+	}
+
+	return 0;
+}
+
 /**
  * amdgpu_ras_eeprom_read -- read EEPROM
  * @control: pointer to control structure
@@ -946,15 +1105,18 @@ int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
 	u8 *buf, *pp;
 	u32 g0, g1;
 
+	if (amdgpu_ras_smu_eeprom_supported(adev))
+		return amdgpu_ras_eeprom_read_idx(control, record, 0, num);
+
 	if (!__is_ras_eeprom_supported(adev))
 		return 0;
 
 	if (num == 0) {
-		DRM_ERROR("will not read 0 records\n");
+		dev_err(adev->dev, "will not read 0 records\n");
 		return -EINVAL;
 	} else if (num > control->ras_num_recs) {
-		DRM_ERROR("too many records to read:%d available:%d\n",
-			  num, control->ras_num_recs);
+		dev_err(adev->dev, "too many records to read:%d available:%d\n",
+			num, control->ras_num_recs);
 		return -EINVAL;
 	}
 
@@ -1031,7 +1193,7 @@ uint32_t amdgpu_ras_eeprom_max_record_count(struct amdgpu_ras_eeprom_control *co
 	/* get available eeprom table version first before eeprom table init */
 	amdgpu_ras_set_eeprom_table_version(control);
 
-	if (control->tbl_hdr.version == RAS_TABLE_VER_V2_1)
+	if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1)
 		return RAS_MAX_RECORD_COUNT_V2_1;
 	else
 		return RAS_MAX_RECORD_COUNT;
@@ -1117,6 +1279,10 @@ static ssize_t amdgpu_ras_debugfs_table_read(struct file *f, char __user *buf,
 	int res = -EFAULT;
 	size_t data_len;
 
+	/* pmfw manages eeprom data by itself */
+	if (amdgpu_ras_smu_eeprom_supported(adev))
+		return 0;
+
 	mutex_lock(&control->ras_tbl_mutex);
 
 	/* We want *pos - data_len > 0, which means there's
@@ -1276,7 +1442,7 @@ static int __verify_ras_table_checksum(struct amdgpu_ras_eeprom_control *control
 	int buf_size, res;
 	u8  csum, *buf, *pp;
 
-	if (control->tbl_hdr.version == RAS_TABLE_VER_V2_1)
+	if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1)
 		buf_size = RAS_TABLE_HEADER_SIZE +
 			   RAS_TABLE_V2_1_INFO_SIZE +
 			   control->ras_num_recs * RAS_TABLE_RECORD_SIZE;
@@ -1286,7 +1452,8 @@ static int __verify_ras_table_checksum(struct amdgpu_ras_eeprom_control *control
 
 	buf = kzalloc(buf_size, GFP_KERNEL);
 	if (!buf) {
-		DRM_ERROR("Out of memory checking RAS table checksum.\n");
+		dev_err(adev->dev,
+			"Out of memory checking RAS table checksum.\n");
 		return -ENOMEM;
 	}
 
@@ -1295,7 +1462,7 @@ static int __verify_ras_table_checksum(struct amdgpu_ras_eeprom_control *control
 				 control->ras_header_offset,
 				 buf, buf_size);
 	if (res < buf_size) {
-		DRM_ERROR("Partial read for checksum, res:%d\n", res);
+		dev_err(adev->dev, "Partial read for checksum, res:%d\n", res);
 		/* On partial reads, return -EIO.
 		 */
 		if (res >= 0)
@@ -1320,7 +1487,8 @@ static int __read_table_ras_info(struct amdgpu_ras_eeprom_control *control)
 
 	buf = kzalloc(RAS_TABLE_V2_1_INFO_SIZE, GFP_KERNEL);
 	if (!buf) {
-		DRM_ERROR("Failed to alloc buf to read EEPROM table ras info\n");
+		dev_err(adev->dev,
+			"Failed to alloc buf to read EEPROM table ras info\n");
 		return -ENOMEM;
 	}
 
@@ -1332,7 +1500,8 @@ static int __read_table_ras_info(struct amdgpu_ras_eeprom_control *control)
 				 control->i2c_address + control->ras_info_offset,
 				 buf, RAS_TABLE_V2_1_INFO_SIZE);
 	if (res < RAS_TABLE_V2_1_INFO_SIZE) {
-		DRM_ERROR("Failed to read EEPROM table ras info, res:%d", res);
+		dev_err(adev->dev,
+			"Failed to read EEPROM table ras info, res:%d", res);
 		res = res >= 0 ? -EIO : res;
 		goto Out;
 	}
@@ -1344,6 +1513,42 @@ Out:
 	return res == RAS_TABLE_V2_1_INFO_SIZE ? 0 : res;
 }
 
+static int amdgpu_ras_smu_eeprom_init(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	uint64_t local_time;
+	int res;
+
+	ras->is_rma = false;
+
+	if (!__is_ras_eeprom_supported(adev))
+		return 0;
+	mutex_init(&control->ras_tbl_mutex);
+
+	res = amdgpu_ras_smu_get_table_version(adev, &(hdr->version));
+	if (res)
+		return res;
+
+	res = amdgpu_ras_smu_get_badpage_count(adev,
+								&(control->ras_num_recs), 100);
+	if (res)
+		return res;
+
+	local_time = (uint64_t)ktime_get_real_seconds();
+	res = amdgpu_ras_smu_set_timestamp(adev, local_time);
+	if (res)
+		return res;
+
+	control->ras_max_record_count = 4000;
+
+	control->ras_num_mca_recs = 0;
+	control->ras_num_pa_recs = 0;
+
+	return 0;
+}
+
 int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control)
 {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
@@ -1352,6 +1557,9 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control)
 	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
 	int res;
 
+	if (amdgpu_ras_smu_eeprom_supported(adev))
+		return amdgpu_ras_smu_eeprom_init(control);
+
 	ras->is_rma = false;
 
 	if (!__is_ras_eeprom_supported(adev))
@@ -1373,23 +1581,89 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control)
 				 control->i2c_address + control->ras_header_offset,
 				 buf, RAS_TABLE_HEADER_SIZE);
 	if (res < RAS_TABLE_HEADER_SIZE) {
-		DRM_ERROR("Failed to read EEPROM table header, res:%d", res);
+		dev_err(adev->dev, "Failed to read EEPROM table header, res:%d",
+			res);
 		return res >= 0 ? -EIO : res;
 	}
 
 	__decode_table_header_from_buf(hdr, buf);
 
-	if (hdr->version == RAS_TABLE_VER_V2_1) {
+	if (hdr->header != RAS_TABLE_HDR_VAL &&
+	    hdr->header != RAS_TABLE_HDR_BAD) {
+		dev_info(adev->dev, "Creating a new EEPROM table");
+		return amdgpu_ras_eeprom_reset_table(control);
+	}
+
+	switch (hdr->version) {
+	case RAS_TABLE_VER_V2_1:
+	case RAS_TABLE_VER_V3:
 		control->ras_num_recs = RAS_NUM_RECS_V2_1(hdr);
 		control->ras_record_offset = RAS_RECORD_START_V2_1;
 		control->ras_max_record_count = RAS_MAX_RECORD_COUNT_V2_1;
-	} else {
+		break;
+	case RAS_TABLE_VER_V1:
 		control->ras_num_recs = RAS_NUM_RECS(hdr);
 		control->ras_record_offset = RAS_RECORD_START;
 		control->ras_max_record_count = RAS_MAX_RECORD_COUNT;
+		break;
+	default:
+		dev_err(adev->dev,
+			"RAS header invalid, unsupported version: %u",
+			hdr->version);
+		return -EINVAL;
 	}
+
+	if (control->ras_num_recs > control->ras_max_record_count) {
+		dev_err(adev->dev,
+			"RAS header invalid, records in header: %u max allowed :%u",
+			control->ras_num_recs, control->ras_max_record_count);
+		return -EINVAL;
+	}
+
 	control->ras_fri = RAS_OFFSET_TO_INDEX(control, hdr->first_rec_offset);
+	control->ras_num_mca_recs = 0;
+	control->ras_num_pa_recs = 0;
+	return 0;
+}
+
+static int amdgpu_ras_smu_eeprom_check(struct amdgpu_ras_eeprom_control *control)
+{
+	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	if (!__is_ras_eeprom_supported(adev))
+		return 0;
 
+	control->ras_num_bad_pages = ras->bad_page_num;
+
+	if ((ras->bad_page_cnt_threshold < control->ras_num_bad_pages) &&
+	    amdgpu_bad_page_threshold != 0) {
+		dev_warn(adev->dev,
+			"RAS records:%d exceed threshold:%d\n",
+			control->ras_num_bad_pages, ras->bad_page_cnt_threshold);
+		if ((amdgpu_bad_page_threshold == -1) ||
+			(amdgpu_bad_page_threshold == -2)) {
+			dev_warn(adev->dev,
+				 "Please consult AMD Service Action Guide (SAG) for appropriate service procedures\n");
+		} else {
+			ras->is_rma = true;
+			dev_warn(adev->dev,
+				 "User defined threshold is set, runtime service will be halt when threshold is reached\n");
+		}
+
+		return 0;
+	}
+
+	dev_dbg(adev->dev,
+		"Found existing EEPROM table with %d records",
+		control->ras_num_bad_pages);
+
+	/* Warn if we are at 90% of the threshold or above
+	 */
+	if (10 * control->ras_num_bad_pages >= 9 * ras->bad_page_cnt_threshold)
+		dev_warn(adev->dev, "RAS records:%u exceeds 90%% of threshold:%d",
+				control->ras_num_bad_pages,
+				ras->bad_page_cnt_threshold);
 	return 0;
 }
 
@@ -1398,7 +1672,10 @@ int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control)
 	struct amdgpu_device *adev = to_amdgpu_device(control);
 	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
 	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
-	int res;
+	int res = 0;
+
+	if (amdgpu_ras_smu_eeprom_supported(adev))
+		return amdgpu_ras_smu_eeprom_check(control);
 
 	if (!__is_ras_eeprom_supported(adev))
 		return 0;
@@ -1410,17 +1687,14 @@ int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control)
 	if (!__get_eeprom_i2c_addr(adev, control))
 		return -EINVAL;
 
-	if (control->rec_type == AMDGPU_RAS_EEPROM_REC_PA)
-		control->ras_num_bad_pages = control->ras_num_recs;
-	else
-		control->ras_num_bad_pages =
-			control->ras_num_recs * adev->umc.retire_unit;
+	control->ras_num_bad_pages = ras->bad_page_num;
 
 	if (hdr->header == RAS_TABLE_HDR_VAL) {
-		DRM_DEBUG_DRIVER("Found existing EEPROM table with %d records",
-				 control->ras_num_bad_pages);
+		dev_dbg(adev->dev,
+			"Found existing EEPROM table with %d records",
+			control->ras_num_bad_pages);
 
-		if (hdr->version == RAS_TABLE_VER_V2_1) {
+		if (hdr->version >= RAS_TABLE_VER_V2_1) {
 			res = __read_table_ras_info(control);
 			if (res)
 				return res;
@@ -1428,8 +1702,9 @@ int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control)
 
 		res = __verify_ras_table_checksum(control);
 		if (res)
-			DRM_ERROR("RAS table incorrect checksum or error:%d\n",
-				  res);
+			dev_err(adev->dev,
+				"RAS table incorrect checksum or error:%d\n",
+				res);
 
 		/* Warn if we are at 90% of the threshold or above
 		 */
@@ -1439,7 +1714,7 @@ int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control)
 					ras->bad_page_cnt_threshold);
 	} else if (hdr->header == RAS_TABLE_HDR_BAD &&
 		   amdgpu_bad_page_threshold != 0) {
-		if (hdr->version == RAS_TABLE_VER_V2_1) {
+		if (hdr->version >= RAS_TABLE_VER_V2_1) {
 			res = __read_table_ras_info(control);
 			if (res)
 				return res;
@@ -1447,11 +1722,12 @@ int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control)
 
 		res = __verify_ras_table_checksum(control);
 		if (res) {
-			dev_err(adev->dev, "RAS Table incorrect checksum or error:%d\n",
-				  res);
+			dev_err(adev->dev,
+				"RAS Table incorrect checksum or error:%d\n",
+				res);
 			return -EINVAL;
 		}
-		if (ras->bad_page_cnt_threshold > control->ras_num_bad_pages) {
+		if (ras->bad_page_cnt_threshold >= control->ras_num_bad_pages) {
 			/* This means that, the threshold was increased since
 			 * the last time the system was booted, and now,
 			 * ras->bad_page_cnt_threshold - control->num_recs > 0,
@@ -1466,24 +1742,189 @@ int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control)
 			res = amdgpu_ras_eeprom_correct_header_tag(control,
 								   RAS_TABLE_HDR_VAL);
 		} else {
-			dev_err(adev->dev, "RAS records:%d exceed threshold:%d",
+			dev_warn(adev->dev,
+				"RAS records:%d exceed threshold:%d\n",
 				control->ras_num_bad_pages, ras->bad_page_cnt_threshold);
-			if (amdgpu_bad_page_threshold == -1) {
-				dev_warn(adev->dev, "GPU will be initialized due to bad_page_threshold = -1.");
+			if ((amdgpu_bad_page_threshold == -1) ||
+			    (amdgpu_bad_page_threshold == -2)) {
 				res = 0;
+				dev_warn(adev->dev,
+					 "Please consult AMD Service Action Guide (SAG) for appropriate service procedures\n");
 			} else {
 				ras->is_rma = true;
-				dev_err(adev->dev,
-					"RAS records:%d exceed threshold:%d, "
-					"GPU will not be initialized. Replace this GPU or increase the threshold",
-					control->ras_num_bad_pages, ras->bad_page_cnt_threshold);
+				dev_warn(adev->dev,
+					 "User defined threshold is set, runtime service will be halt when threshold is reached\n");
 			}
 		}
-	} else {
-		DRM_INFO("Creating a new EEPROM table");
-
-		res = amdgpu_ras_eeprom_reset_table(control);
 	}
 
 	return res < 0 ? res : 0;
 }
+
+void amdgpu_ras_eeprom_check_and_recover(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+	struct amdgpu_ras_eeprom_control *control;
+	int res;
+
+	if (!__is_ras_eeprom_supported(adev) || !ras ||
+	    amdgpu_ras_smu_eeprom_supported(adev))
+		return;
+	control = &ras->eeprom_control;
+	if (!control->is_eeprom_valid)
+		return;
+	res = __verify_ras_table_checksum(control);
+	if (res) {
+		dev_warn(adev->dev,
+			"RAS table incorrect checksum or error:%d, try to recover\n",
+			res);
+		if (!amdgpu_ras_eeprom_reset_table(control))
+			if (!amdgpu_ras_save_bad_pages(adev, NULL))
+				if (!__verify_ras_table_checksum(control)) {
+					dev_info(adev->dev, "RAS table recovery succeed\n");
+					return;
+				}
+		dev_err(adev->dev, "RAS table recovery failed\n");
+		control->is_eeprom_valid = false;
+	}
+	return;
+}
+
+static const struct ras_smu_drv *amdgpu_ras_get_smu_ras_drv(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
+
+	if (!ras)
+		return NULL;
+
+	return ras->ras_smu_drv;
+}
+
+static uint64_t amdgpu_ras_smu_get_feature_flags(struct amdgpu_device *adev)
+{
+	const struct ras_smu_drv *ras_smu_drv = amdgpu_ras_get_smu_ras_drv(adev);
+	uint64_t flags = 0ULL;
+
+	if (!ras_smu_drv)
+		goto out;
+
+	if (ras_smu_drv->ras_smu_feature_flags)
+		ras_smu_drv->ras_smu_feature_flags(adev, &flags);
+
+out:
+	return flags;
+}
+
+bool amdgpu_ras_smu_eeprom_supported(struct amdgpu_device *adev)
+{
+	const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev);
+	uint64_t flags = 0ULL;
+
+	if (!__is_ras_eeprom_supported(adev) || !smu_ras_drv)
+		return false;
+
+	if (!smu_ras_drv->smu_eeprom_funcs)
+		return false;
+
+	flags = amdgpu_ras_smu_get_feature_flags(adev);
+
+	return !!(flags & RAS_SMU_FEATURE_BIT__RAS_EEPROM);
+}
+
+int amdgpu_ras_smu_get_table_version(struct amdgpu_device *adev,
+				     uint32_t *table_version)
+{
+	const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev);
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev))
+		return -EOPNOTSUPP;
+
+	if (smu_ras_drv->smu_eeprom_funcs->get_ras_table_version)
+		return smu_ras_drv->smu_eeprom_funcs->get_ras_table_version(adev,
+										 table_version);
+	return -EOPNOTSUPP;
+}
+
+int amdgpu_ras_smu_get_badpage_count(struct amdgpu_device *adev,
+				     uint32_t *count, uint32_t timeout)
+{
+	const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev);
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev))
+		return -EOPNOTSUPP;
+
+	if (smu_ras_drv->smu_eeprom_funcs->get_badpage_count)
+		return smu_ras_drv->smu_eeprom_funcs->get_badpage_count(adev,
+									     count, timeout);
+	return -EOPNOTSUPP;
+}
+
+int amdgpu_ras_smu_get_badpage_mca_addr(struct amdgpu_device *adev,
+					uint16_t index, uint64_t *mca_addr)
+{
+	const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev);
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev))
+		return -EOPNOTSUPP;
+
+	if (smu_ras_drv->smu_eeprom_funcs->get_badpage_mca_addr)
+		return smu_ras_drv->smu_eeprom_funcs->get_badpage_mca_addr(adev,
+										index, mca_addr);
+	return -EOPNOTSUPP;
+}
+
+int amdgpu_ras_smu_set_timestamp(struct amdgpu_device *adev,
+				 uint64_t timestamp)
+{
+	const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev);
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev))
+		return -EOPNOTSUPP;
+
+	if (smu_ras_drv->smu_eeprom_funcs->set_timestamp)
+		return smu_ras_drv->smu_eeprom_funcs->set_timestamp(adev,
+									 timestamp);
+	return -EOPNOTSUPP;
+}
+
+int amdgpu_ras_smu_get_timestamp(struct amdgpu_device *adev,
+				 uint16_t index, uint64_t *timestamp)
+{
+	const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev);
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev))
+		return -EOPNOTSUPP;
+
+	if (smu_ras_drv->smu_eeprom_funcs->get_timestamp)
+		return smu_ras_drv->smu_eeprom_funcs->get_timestamp(adev,
+									 index, timestamp);
+	return -EOPNOTSUPP;
+}
+
+int amdgpu_ras_smu_get_badpage_ipid(struct amdgpu_device *adev,
+				    uint16_t index, uint64_t *ipid)
+{
+	const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev);
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev))
+		return -EOPNOTSUPP;
+
+	if (smu_ras_drv->smu_eeprom_funcs->get_badpage_ipid)
+		return smu_ras_drv->smu_eeprom_funcs->get_badpage_ipid(adev,
+									    index, ipid);
+	return -EOPNOTSUPP;
+}
+
+int amdgpu_ras_smu_erase_ras_table(struct amdgpu_device *adev,
+				   uint32_t *result)
+{
+	const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev);
+
+	if (!amdgpu_ras_smu_eeprom_supported(adev))
+		return -EOPNOTSUPP;
+
+	if (smu_ras_drv->smu_eeprom_funcs->erase_ras_table)
+		return smu_ras_drv->smu_eeprom_funcs->erase_ras_table(adev,
+									   result);
+	return -EOPNOTSUPP;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h
index 81d55cb7b397..2e5d63957e71 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h
@@ -28,6 +28,7 @@
 
 #define RAS_TABLE_VER_V1           0x00010000
 #define RAS_TABLE_VER_V2_1         0x00021000
+#define RAS_TABLE_VER_V3           0x00030000
 
 struct amdgpu_device;
 
@@ -43,19 +44,6 @@ enum amdgpu_ras_eeprom_err_type {
 	AMDGPU_RAS_EEPROM_ERR_COUNT,
 };
 
-/*
- * one UMC MCA address could map to multiply physical address (PA),
- * such as 1:16, we use eeprom_table_record.address to store MCA
- * address and use eeprom_table_record.retired_page to save PA.
- *
- * AMDGPU_RAS_EEPROM_REC_PA: one record store one PA
- * AMDGPU_RAS_EEPROM_REC_MCA: one record store one MCA address
- */
-enum amdgpu_ras_eeprom_rec_type {
-	AMDGPU_RAS_EEPROM_REC_PA,
-	AMDGPU_RAS_EEPROM_REC_MCA,
-};
-
 struct amdgpu_ras_eeprom_table_header {
 	uint32_t header;
 	uint32_t version;
@@ -94,12 +82,19 @@ struct amdgpu_ras_eeprom_control {
 	/* Number of records in the table.
 	 */
 	u32 ras_num_recs;
+	u32 ras_num_recs_old;
 
 	/* the bad page number is ras_num_recs or
 	 * ras_num_recs * umc.retire_unit
 	 */
 	u32 ras_num_bad_pages;
 
+	/* Number of records store mca address */
+	u32 ras_num_mca_recs;
+
+	/* Number of records store physical address */
+	u32 ras_num_pa_recs;
+
 	/* First record index to read, 0-based.
 	 * Range is [0, num_recs-1]. This is
 	 * an absolute index, starting right after
@@ -120,7 +115,8 @@ struct amdgpu_ras_eeprom_control {
 	/* Record channel info which occurred bad pages
 	 */
 	u32 bad_channel_bitmap;
-	enum amdgpu_ras_eeprom_rec_type rec_type;
+
+	bool is_eeprom_valid;
 };
 
 /*
@@ -166,6 +162,37 @@ void amdgpu_ras_debugfs_set_ret_size(struct amdgpu_ras_eeprom_control *control);
 
 int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control);
 
+void amdgpu_ras_eeprom_check_and_recover(struct amdgpu_device *adev);
+
+bool amdgpu_ras_smu_eeprom_supported(struct amdgpu_device *adev);
+
+int amdgpu_ras_smu_get_table_version(struct amdgpu_device *adev,
+							uint32_t *table_version);
+
+int amdgpu_ras_smu_get_badpage_count(struct amdgpu_device *adev,
+								uint32_t *count, uint32_t timeout);
+
+int amdgpu_ras_smu_get_badpage_mca_addr(struct amdgpu_device *adev,
+								uint16_t index, uint64_t *mca_addr);
+
+int amdgpu_ras_smu_set_timestamp(struct amdgpu_device *adev,
+										uint64_t timestamp);
+
+int amdgpu_ras_smu_get_timestamp(struct amdgpu_device *adev,
+							uint16_t index, uint64_t *timestamp);
+
+int amdgpu_ras_smu_get_badpage_ipid(struct amdgpu_device *adev,
+								uint16_t index, uint64_t *ipid);
+
+int amdgpu_ras_smu_erase_ras_table(struct amdgpu_device *adev,
+									uint32_t *result);
+
+int amdgpu_ras_eeprom_read_idx(struct amdgpu_ras_eeprom_control *control,
+			struct eeprom_table_record *record, u32 rec_idx,
+			const u32 num);
+
+int amdgpu_ras_eeprom_update_record_num(struct amdgpu_ras_eeprom_control *control);
+
 extern const struct file_operations amdgpu_ras_debugfs_eeprom_size_ops;
 extern const struct file_operations amdgpu_ras_debugfs_eeprom_table_ops;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h
index 50fcd86e1033..be2e56ce1355 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_res_cursor.h
@@ -91,6 +91,7 @@ static inline void amdgpu_res_first(struct ttm_resource *res,
 		break;
 	case TTM_PL_TT:
 	case AMDGPU_PL_DOORBELL:
+	case AMDGPU_PL_MMIO_REMAP:
 		node = to_ttm_range_mgr_node(res)->mm_nodes;
 		while (start >= node->size << PAGE_SHIFT)
 			start -= node++->size << PAGE_SHIFT;
@@ -153,6 +154,7 @@ static inline void amdgpu_res_next(struct amdgpu_res_cursor *cur, uint64_t size)
 		break;
 	case TTM_PL_TT:
 	case AMDGPU_PL_DOORBELL:
+	case AMDGPU_PL_MMIO_REMAP:
 		node = cur->node;
 
 		cur->node = ++node;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c
index dabfbdf6f1ce..28c4ad62f50e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c
@@ -340,6 +340,9 @@ void amdgpu_reset_get_desc(struct amdgpu_reset_context *rst_ctxt, char *buf,
 	case AMDGPU_RESET_SRC_USER:
 		strscpy(buf, "user trigger", len);
 		break;
+	case AMDGPU_RESET_SRC_USERQ:
+		strscpy(buf, "user queue trigger", len);
+		break;
 	default:
 		strscpy(buf, "unknown", len);
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
index 4d9b9701139b..07b4d37f1db6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
@@ -43,6 +43,7 @@ enum AMDGPU_RESET_SRCS {
 	AMDGPU_RESET_SRC_MES,
 	AMDGPU_RESET_SRC_HWS,
 	AMDGPU_RESET_SRC_USER,
+	AMDGPU_RESET_SRC_USERQ,
 };
 
 struct amdgpu_reset_context {
@@ -160,4 +161,16 @@ int amdgpu_reset_do_xgmi_reset_on_init(
 
 bool amdgpu_reset_in_recovery(struct amdgpu_device *adev);
 
+static inline void amdgpu_reset_set_dpc_status(struct amdgpu_device *adev,
+					       bool status)
+{
+	adev->pcie_reset_ctx.occurs_dpc = status;
+	adev->no_hw_access = status;
+}
+
+static inline bool amdgpu_reset_in_dpc(struct amdgpu_device *adev)
+{
+	return adev->pcie_reset_ctx.occurs_dpc;
+}
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
index a6e28fe3f8d6..c596b6df2e2d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
@@ -33,6 +33,7 @@
 
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
+#include "amdgpu_ras_mgr.h"
 #include "atom.h"
 
 /*
@@ -99,6 +100,29 @@ int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned int ndw)
 	return 0;
 }
 
+/**
+ * amdgpu_ring_alloc_reemit - allocate space on the ring buffer for reemit
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @ndw: number of dwords to allocate in the ring buffer
+ *
+ * Allocate @ndw dwords in the ring buffer (all asics).
+ * doesn't check the max_dw limit as we may be reemitting
+ * several submissions.
+ */
+static void amdgpu_ring_alloc_reemit(struct amdgpu_ring *ring, unsigned int ndw)
+{
+	/* Align requested size with padding so unlock_commit can
+	 * pad safely */
+	ndw = (ndw + ring->funcs->align_mask) & ~ring->funcs->align_mask;
+
+	ring->count_dw = ndw;
+	ring->wptr_old = ring->wptr;
+
+	if (ring->funcs->begin_use)
+		ring->funcs->begin_use(ring);
+}
+
 /** amdgpu_ring_insert_nop - insert NOP packets
  *
  * @ring: amdgpu_ring structure holding ring information
@@ -136,8 +160,16 @@ void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
  */
 void amdgpu_ring_generic_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
 {
-	while (ib->length_dw & ring->funcs->align_mask)
-		ib->ptr[ib->length_dw++] = ring->funcs->nop;
+	u32 align_mask = ring->funcs->align_mask;
+	u32 count = ib->length_dw & align_mask;
+
+	if (count) {
+		count = align_mask + 1 - count;
+
+		memset32(&ib->ptr[ib->length_dw], ring->funcs->nop, count);
+
+		ib->length_dw += count;
+	}
 }
 
 /**
@@ -187,14 +219,10 @@ void amdgpu_ring_undo(struct amdgpu_ring *ring)
 }
 
 #define amdgpu_ring_get_gpu_addr(ring, offset)				\
-	(ring->is_mes_queue ?						\
-	 (ring->mes_ctx->meta_data_gpu_addr + offset) :			\
-	 (ring->adev->wb.gpu_addr + offset * 4))
+	 (ring->adev->wb.gpu_addr + offset * 4)
 
 #define amdgpu_ring_get_cpu_addr(ring, offset)				\
-	(ring->is_mes_queue ?						\
-	 (void *)((uint8_t *)(ring->mes_ctx->meta_data_ptr) + offset) : \
-	 (&ring->adev->wb.wb[offset]))
+	 (&ring->adev->wb.wb[offset])
 
 /**
  * amdgpu_ring_init - init driver ring struct.
@@ -243,57 +271,42 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 		ring->sched_score = sched_score;
 		ring->vmid_wait = dma_fence_get_stub();
 
-		if (!ring->is_mes_queue) {
-			ring->idx = adev->num_rings++;
-			adev->rings[ring->idx] = ring;
-		}
+		ring->idx = adev->num_rings++;
+		adev->rings[ring->idx] = ring;
 
 		r = amdgpu_fence_driver_init_ring(ring);
 		if (r)
 			return r;
 	}
 
-	if (ring->is_mes_queue) {
-		ring->rptr_offs = amdgpu_mes_ctx_get_offs(ring,
-				AMDGPU_MES_CTX_RPTR_OFFS);
-		ring->wptr_offs = amdgpu_mes_ctx_get_offs(ring,
-				AMDGPU_MES_CTX_WPTR_OFFS);
-		ring->fence_offs = amdgpu_mes_ctx_get_offs(ring,
-				AMDGPU_MES_CTX_FENCE_OFFS);
-		ring->trail_fence_offs = amdgpu_mes_ctx_get_offs(ring,
-				AMDGPU_MES_CTX_TRAIL_FENCE_OFFS);
-		ring->cond_exe_offs = amdgpu_mes_ctx_get_offs(ring,
-				AMDGPU_MES_CTX_COND_EXE_OFFS);
-	} else {
-		r = amdgpu_device_wb_get(adev, &ring->rptr_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
-			return r;
-		}
+	r = amdgpu_device_wb_get(adev, &ring->rptr_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
+		return r;
+	}
 
-		r = amdgpu_device_wb_get(adev, &ring->wptr_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
-			return r;
-		}
+	r = amdgpu_device_wb_get(adev, &ring->wptr_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
+		return r;
+	}
 
-		r = amdgpu_device_wb_get(adev, &ring->fence_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring fence_offs wb alloc failed\n", r);
-			return r;
-		}
+	r = amdgpu_device_wb_get(adev, &ring->fence_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring fence_offs wb alloc failed\n", r);
+		return r;
+	}
 
-		r = amdgpu_device_wb_get(adev, &ring->trail_fence_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring trail_fence_offs wb alloc failed\n", r);
-			return r;
-		}
+	r = amdgpu_device_wb_get(adev, &ring->trail_fence_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring trail_fence_offs wb alloc failed\n", r);
+		return r;
+	}
 
-		r = amdgpu_device_wb_get(adev, &ring->cond_exe_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring cond_exec_polling wb alloc failed\n", r);
-			return r;
-		}
+	r = amdgpu_device_wb_get(adev, &ring->cond_exe_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring cond_exec_polling wb alloc failed\n", r);
+		return r;
 	}
 
 	ring->fence_gpu_addr =
@@ -324,45 +337,51 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 	/* always set cond_exec_polling to CONTINUE */
 	*ring->cond_exe_cpu_addr = 1;
 
-	r = amdgpu_fence_driver_start_ring(ring, irq_src, irq_type);
-	if (r) {
-		dev_err(adev->dev, "failed initializing fences (%d).\n", r);
-		return r;
-	}
+	if (ring->funcs->type != AMDGPU_RING_TYPE_CPER) {
+		r = amdgpu_fence_driver_start_ring(ring, irq_src, irq_type);
+		if (r) {
+			dev_err(adev->dev, "failed initializing fences (%d).\n", r);
+			return r;
+		}
 
-	max_ibs_dw = ring->funcs->emit_frame_size +
-		     amdgpu_ring_max_ibs(ring->funcs->type) * ring->funcs->emit_ib_size;
-	max_ibs_dw = (max_ibs_dw + ring->funcs->align_mask) & ~ring->funcs->align_mask;
+		max_ibs_dw = ring->funcs->emit_frame_size +
+			     amdgpu_ring_max_ibs(ring->funcs->type) * ring->funcs->emit_ib_size;
+		max_ibs_dw = (max_ibs_dw + ring->funcs->align_mask) & ~ring->funcs->align_mask;
 
-	if (WARN_ON(max_ibs_dw > max_dw))
-		max_dw = max_ibs_dw;
+		if (WARN_ON(max_ibs_dw > max_dw))
+			max_dw = max_ibs_dw;
 
-	ring->ring_size = roundup_pow_of_two(max_dw * 4 * sched_hw_submission);
+		ring->ring_size = roundup_pow_of_two(max_dw * 4 * sched_hw_submission);
+	} else {
+		ring->ring_size = roundup_pow_of_two(max_dw * 4);
+		ring->count_dw = (ring->ring_size - 4) >> 2;
+		/* ring buffer is empty now */
+		ring->wptr = *ring->rptr_cpu_addr = 0;
+	}
 
 	ring->buf_mask = (ring->ring_size / 4) - 1;
 	ring->ptr_mask = ring->funcs->support_64bit_ptrs ?
 		0xffffffffffffffff : ring->buf_mask;
+	/*  Initialize cached_rptr to 0 */
+	ring->cached_rptr = 0;
 
-	/* Allocate ring buffer */
-	if (ring->is_mes_queue) {
-		int offset = 0;
-
-		BUG_ON(ring->ring_size > PAGE_SIZE*4);
-
-		offset = amdgpu_mes_ctx_get_offs(ring,
-					 AMDGPU_MES_CTX_RING_OFFS);
-		ring->gpu_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
-		ring->ring = amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset);
-		amdgpu_ring_clear_ring(ring);
+	if (!ring->ring_backup) {
+		ring->ring_backup = kvzalloc(ring->ring_size, GFP_KERNEL);
+		if (!ring->ring_backup)
+			return -ENOMEM;
+	}
 
-	} else if (ring->ring_obj == NULL) {
-		r = amdgpu_bo_create_kernel(adev, ring->ring_size + ring->funcs->extra_dw, PAGE_SIZE,
+	/* Allocate ring buffer */
+	if (ring->ring_obj == NULL) {
+		r = amdgpu_bo_create_kernel(adev, ring->ring_size + ring->funcs->extra_bytes,
+					    PAGE_SIZE,
 					    AMDGPU_GEM_DOMAIN_GTT,
 					    &ring->ring_obj,
 					    &ring->gpu_addr,
 					    (void **)&ring->ring);
 		if (r) {
 			dev_err(adev->dev, "(%d) ring create failed\n", r);
+			kvfree(ring->ring_backup);
 			return r;
 		}
 		amdgpu_ring_clear_ring(ring);
@@ -392,32 +411,26 @@ void amdgpu_ring_fini(struct amdgpu_ring *ring)
 {
 
 	/* Not to finish a ring which is not initialized */
-	if (!(ring->adev) ||
-	    (!ring->is_mes_queue && !(ring->adev->rings[ring->idx])))
+	if (!(ring->adev) || !(ring->adev->rings[ring->idx]))
 		return;
 
 	ring->sched.ready = false;
 
-	if (!ring->is_mes_queue) {
-		amdgpu_device_wb_free(ring->adev, ring->rptr_offs);
-		amdgpu_device_wb_free(ring->adev, ring->wptr_offs);
+	amdgpu_device_wb_free(ring->adev, ring->rptr_offs);
+	amdgpu_device_wb_free(ring->adev, ring->wptr_offs);
 
-		amdgpu_device_wb_free(ring->adev, ring->cond_exe_offs);
-		amdgpu_device_wb_free(ring->adev, ring->fence_offs);
+	amdgpu_device_wb_free(ring->adev, ring->cond_exe_offs);
+	amdgpu_device_wb_free(ring->adev, ring->fence_offs);
 
-		amdgpu_bo_free_kernel(&ring->ring_obj,
-				      &ring->gpu_addr,
-				      (void **)&ring->ring);
-	} else {
-		kfree(ring->fence_drv.fences);
-	}
+	amdgpu_bo_free_kernel(&ring->ring_obj,
+			      &ring->gpu_addr,
+			      (void **)&ring->ring);
+	kvfree(ring->ring_backup);
+	ring->ring_backup = NULL;
 
 	dma_fence_put(ring->vmid_wait);
 	ring->vmid_wait = NULL;
 	ring->me = 0;
-
-	if (!ring->is_mes_queue)
-		ring->adev->rings[ring->idx] = NULL;
 }
 
 /**
@@ -454,9 +467,7 @@ bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
 {
 	unsigned long flags;
 	ktime_t deadline;
-
-	if (unlikely(ring->adev->debug_disable_soft_recovery))
-		return false;
+	bool ret;
 
 	deadline = ktime_add_us(ktime_get(), 10000);
 
@@ -468,12 +479,16 @@ bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
 		dma_fence_set_error(fence, -ENODATA);
 	spin_unlock_irqrestore(fence->lock, flags);
 
-	atomic_inc(&ring->adev->gpu_reset_counter);
 	while (!dma_fence_is_signaled(fence) &&
 	       ktime_to_ns(ktime_sub(deadline, ktime_get())) > 0)
 		ring->funcs->soft_recovery(ring, vmid);
 
-	return dma_fence_is_signaled(fence);
+	ret = dma_fence_is_signaled(fence);
+	/* increment the counter only if soft reset worked */
+	if (ret)
+		atomic_inc(&ring->adev->gpu_reset_counter);
+
+	return ret;
 }
 
 /*
@@ -481,6 +496,66 @@ bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
  */
 #if defined(CONFIG_DEBUG_FS)
 
+static ssize_t amdgpu_ras_cper_debugfs_read(struct file *f, char __user *buf,
+					    size_t size, loff_t *offset)
+{
+	const uint8_t ring_header_size = 12;
+	struct amdgpu_ring *ring = file_inode(f)->i_private;
+	struct ras_cmd_cper_snapshot_req *snapshot_req __free(kfree) =
+		kzalloc(sizeof(struct ras_cmd_cper_snapshot_req), GFP_KERNEL);
+	struct ras_cmd_cper_snapshot_rsp *snapshot_rsp __free(kfree) =
+		kzalloc(sizeof(struct ras_cmd_cper_snapshot_rsp), GFP_KERNEL);
+	struct ras_cmd_cper_record_req *record_req __free(kfree) =
+		kzalloc(sizeof(struct ras_cmd_cper_record_req), GFP_KERNEL);
+	struct ras_cmd_cper_record_rsp *record_rsp __free(kfree) =
+		kzalloc(sizeof(struct ras_cmd_cper_record_rsp), GFP_KERNEL);
+	uint8_t *ring_header __free(kfree) =
+		kzalloc(ring_header_size, GFP_KERNEL);
+	uint32_t total_cper_num;
+	uint64_t start_cper_id;
+	int r;
+
+	if (!snapshot_req || !snapshot_rsp || !record_req || !record_rsp ||
+	    !ring_header)
+		return -ENOMEM;
+
+	if (!(*offset)) {
+		/* Need at least 12 bytes for the header on the first read */
+		if (size < ring_header_size)
+			return -EINVAL;
+
+		if (copy_to_user(buf, ring_header, ring_header_size))
+			return -EFAULT;
+		buf += ring_header_size;
+		size -= ring_header_size;
+	}
+
+	r = amdgpu_ras_mgr_handle_ras_cmd(ring->adev,
+					  RAS_CMD__GET_CPER_SNAPSHOT,
+					  snapshot_req, sizeof(struct ras_cmd_cper_snapshot_req),
+					  snapshot_rsp, sizeof(struct ras_cmd_cper_snapshot_rsp));
+	if (r || !snapshot_rsp->total_cper_num)
+		return r;
+
+	start_cper_id = snapshot_rsp->start_cper_id;
+	total_cper_num = snapshot_rsp->total_cper_num;
+
+	record_req->buf_ptr = (uint64_t)(uintptr_t)buf;
+	record_req->buf_size = size;
+	record_req->cper_start_id = start_cper_id + *offset;
+	record_req->cper_num = total_cper_num;
+	r = amdgpu_ras_mgr_handle_ras_cmd(ring->adev, RAS_CMD__GET_CPER_RECORD,
+					  record_req, sizeof(struct ras_cmd_cper_record_req),
+					  record_rsp, sizeof(struct ras_cmd_cper_record_rsp));
+	if (r)
+		return r;
+
+	r = *offset ? record_rsp->real_data_size : record_rsp->real_data_size + ring_header_size;
+	(*offset) += record_rsp->real_cper_num;
+
+	return r;
+}
+
 /* Layout of file is 12 bytes consisting of
  * - rptr
  * - wptr
@@ -493,22 +568,31 @@ static ssize_t amdgpu_debugfs_ring_read(struct file *f, char __user *buf,
 {
 	struct amdgpu_ring *ring = file_inode(f)->i_private;
 	uint32_t value, result, early[3];
+	uint64_t p;
 	loff_t i;
 	int r;
 
+	if (ring->funcs->type == AMDGPU_RING_TYPE_CPER && amdgpu_uniras_enabled(ring->adev))
+		return amdgpu_ras_cper_debugfs_read(f, buf, size, pos);
+
 	if (*pos & 3 || size & 3)
 		return -EINVAL;
 
 	result = 0;
 
 	if (*pos < 12) {
+		if (ring->funcs->type == AMDGPU_RING_TYPE_CPER)
+			mutex_lock(&ring->adev->cper.ring_lock);
+
 		early[0] = amdgpu_ring_get_rptr(ring) & ring->buf_mask;
 		early[1] = amdgpu_ring_get_wptr(ring) & ring->buf_mask;
 		early[2] = ring->wptr & ring->buf_mask;
 		for (i = *pos / 4; i < 3 && size; i++) {
 			r = put_user(early[i], (uint32_t *)buf);
-			if (r)
-				return r;
+			if (r) {
+				result = r;
+				goto out;
+			}
 			buf += 4;
 			result += 4;
 			size -= 4;
@@ -516,86 +600,94 @@ static ssize_t amdgpu_debugfs_ring_read(struct file *f, char __user *buf,
 		}
 	}
 
-	while (size) {
-		if (*pos >= (ring->ring_size + 12))
-			return result;
+	if (ring->funcs->type != AMDGPU_RING_TYPE_CPER) {
+		while (size) {
+			if (*pos >= (ring->ring_size + 12))
+				return result;
 
-		value = ring->ring[(*pos - 12)/4];
-		r = put_user(value, (uint32_t *)buf);
-		if (r)
-			return r;
-		buf += 4;
-		result += 4;
-		size -= 4;
-		*pos += 4;
+			value = ring->ring[(*pos - 12)/4];
+			r = put_user(value, (uint32_t *)buf);
+			if (r)
+				return r;
+			buf += 4;
+			result += 4;
+			size -= 4;
+			*pos += 4;
+		}
+	} else {
+		p = early[0];
+		if (early[0] <= early[1])
+			size = (early[1] - early[0]);
+		else
+			size = ring->ring_size - (early[0] - early[1]);
+
+		while (size) {
+			if (p == early[1])
+				goto out;
+
+			value = ring->ring[p];
+			r = put_user(value, (uint32_t *)buf);
+			if (r) {
+				result = r;
+				goto out;
+			}
+
+			buf += 4;
+			result += 4;
+			size--;
+			p++;
+			p &= ring->ptr_mask;
+		}
 	}
 
+out:
+	if (ring->funcs->type == AMDGPU_RING_TYPE_CPER)
+		mutex_unlock(&ring->adev->cper.ring_lock);
+
 	return result;
 }
 
-static const struct file_operations amdgpu_debugfs_ring_fops = {
-	.owner = THIS_MODULE,
-	.read = amdgpu_debugfs_ring_read,
-	.llseek = default_llseek
-};
-
-static ssize_t amdgpu_debugfs_mqd_read(struct file *f, char __user *buf,
-				       size_t size, loff_t *pos)
+static ssize_t amdgpu_debugfs_virt_ring_read(struct file *f, char __user *buf,
+	size_t size, loff_t *pos)
 {
 	struct amdgpu_ring *ring = file_inode(f)->i_private;
-	volatile u32 *mqd;
-	u32 *kbuf;
-	int r, i;
-	uint32_t value, result;
 
 	if (*pos & 3 || size & 3)
 		return -EINVAL;
 
-	kbuf = kmalloc(ring->mqd_size, GFP_KERNEL);
-	if (!kbuf)
-		return -ENOMEM;
-
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0))
-		goto err_free;
+	if (ring->funcs->type == AMDGPU_RING_TYPE_CPER)
+		amdgpu_virt_req_ras_cper_dump(ring->adev, false);
 
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&mqd);
-	if (r)
-		goto err_unreserve;
+	return amdgpu_debugfs_ring_read(f, buf, size, pos);
+}
 
-	/*
-	 * Copy to local buffer to avoid put_user(), which might fault
-	 * and acquire mmap_sem, under reservation_ww_class_mutex.
-	 */
-	for (i = 0; i < ring->mqd_size/sizeof(u32); i++)
-		kbuf[i] = mqd[i];
+static const struct file_operations amdgpu_debugfs_ring_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_ring_read,
+	.llseek = default_llseek
+};
 
-	amdgpu_bo_kunmap(ring->mqd_obj);
-	amdgpu_bo_unreserve(ring->mqd_obj);
+static const struct file_operations amdgpu_debugfs_virt_ring_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_virt_ring_read,
+	.llseek = default_llseek
+};
 
-	result = 0;
-	while (size) {
-		if (*pos >= ring->mqd_size)
-			break;
+static ssize_t amdgpu_debugfs_mqd_read(struct file *f, char __user *buf,
+				       size_t size, loff_t *pos)
+{
+	struct amdgpu_ring *ring = file_inode(f)->i_private;
+	ssize_t bytes = min_t(ssize_t, ring->mqd_size - *pos, size);
+	void *from = ((u8 *)ring->mqd_ptr) + *pos;
 
-		value = kbuf[*pos/4];
-		r = put_user(value, (uint32_t *)buf);
-		if (r)
-			goto err_free;
-		buf += 4;
-		result += 4;
-		size -= 4;
-		*pos += 4;
-	}
+	if (*pos > ring->mqd_size)
+		return 0;
 
-	kfree(kbuf);
-	return result;
+	if (copy_to_user(buf, from, bytes))
+		return -EFAULT;
 
-err_unreserve:
-	amdgpu_bo_unreserve(ring->mqd_obj);
-err_free:
-	kfree(kbuf);
-	return r;
+	*pos += bytes;
+	return bytes;
 }
 
 static const struct file_operations amdgpu_debugfs_mqd_fops = {
@@ -626,9 +718,14 @@ void amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
 	char name[32];
 
 	sprintf(name, "amdgpu_ring_%s", ring->name);
-	debugfs_create_file_size(name, S_IFREG | 0444, root, ring,
-				 &amdgpu_debugfs_ring_fops,
-				 ring->ring_size + 12);
+	if (amdgpu_sriov_vf(adev))
+		debugfs_create_file_size(name, S_IFREG | 0444, root, ring,
+					 &amdgpu_debugfs_virt_ring_fops,
+					 ring->ring_size + 12);
+	else
+		debugfs_create_file_size(name, S_IFREG | 0444, root, ring,
+					 &amdgpu_debugfs_ring_fops,
+					 ring->ring_size + 12);
 
 	if (ring->mqd_obj) {
 		sprintf(name, "amdgpu_mqd_%s", ring->name);
@@ -690,6 +787,7 @@ static void amdgpu_ring_to_mqd_prop(struct amdgpu_ring *ring,
 	prop->eop_gpu_addr = ring->eop_gpu_addr;
 	prop->use_doorbell = ring->use_doorbell;
 	prop->doorbell_index = ring->doorbell_index;
+	prop->kernel_queue = true;
 
 	/* map_queues packet doesn't need activate the queue,
 	 * so only kiq need set this field.
@@ -761,3 +859,69 @@ bool amdgpu_ring_sched_ready(struct amdgpu_ring *ring)
 
 	return true;
 }
+
+void amdgpu_ring_reset_helper_begin(struct amdgpu_ring *ring,
+				    struct amdgpu_fence *guilty_fence)
+{
+	/* Stop the scheduler to prevent anybody else from touching the ring buffer. */
+	drm_sched_wqueue_stop(&ring->sched);
+	/* back up the non-guilty commands */
+	amdgpu_ring_backup_unprocessed_commands(ring, guilty_fence);
+}
+
+int amdgpu_ring_reset_helper_end(struct amdgpu_ring *ring,
+				 struct amdgpu_fence *guilty_fence)
+{
+	unsigned int i;
+	int r;
+
+	/* verify that the ring is functional */
+	r = amdgpu_ring_test_ring(ring);
+	if (r)
+		return r;
+
+	/* signal the guilty fence and set an error on all fences from the context */
+	if (guilty_fence)
+		amdgpu_fence_driver_guilty_force_completion(guilty_fence);
+	/* Re-emit the non-guilty commands */
+	if (ring->ring_backup_entries_to_copy) {
+		amdgpu_ring_alloc_reemit(ring, ring->ring_backup_entries_to_copy);
+		for (i = 0; i < ring->ring_backup_entries_to_copy; i++)
+			amdgpu_ring_write(ring, ring->ring_backup[i]);
+		amdgpu_ring_commit(ring);
+	}
+	/* Start the scheduler again */
+	drm_sched_wqueue_start(&ring->sched);
+	return 0;
+}
+
+bool amdgpu_ring_is_reset_type_supported(struct amdgpu_ring *ring,
+					 u32 reset_type)
+{
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_GFX:
+		if (ring->adev->gfx.gfx_supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_COMPUTE:
+		if (ring->adev->gfx.compute_supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_SDMA:
+		if (ring->adev->sdma.supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_VCN_DEC:
+	case AMDGPU_RING_TYPE_VCN_ENC:
+		if (ring->adev->vcn.supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_VCN_JPEG:
+		if (ring->adev->jpeg.supported_reset & reset_type)
+			return true;
+		break;
+	default:
+		break;
+	}
+	return false;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
index dee5a1b4e572..7a27c6c4bb44 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
@@ -37,7 +37,7 @@ struct amdgpu_job;
 struct amdgpu_vm;
 
 /* max number of rings */
-#define AMDGPU_MAX_RINGS		124
+#define AMDGPU_MAX_RINGS		149
 #define AMDGPU_MAX_HWIP_RINGS		64
 #define AMDGPU_MAX_GFX_RINGS		2
 #define AMDGPU_MAX_SW_GFX_RINGS         2
@@ -82,6 +82,8 @@ enum amdgpu_ring_type {
 	AMDGPU_RING_TYPE_KIQ,
 	AMDGPU_RING_TYPE_MES,
 	AMDGPU_RING_TYPE_UMSCH_MM,
+	AMDGPU_RING_TYPE_CPER,
+	AMDGPU_RING_TYPE_MAX,
 };
 
 enum amdgpu_ib_pool_type {
@@ -113,10 +115,11 @@ struct amdgpu_sched {
  */
 struct amdgpu_fence_driver {
 	uint64_t			gpu_addr;
-	volatile uint32_t		*cpu_addr;
+	uint32_t			*cpu_addr;
 	/* sync_seq is protected by ring emission lock */
 	uint32_t			sync_seq;
 	atomic_t			last_seq;
+	u64				signalled_wptr;
 	bool				initialized;
 	struct amdgpu_irq_src		*irq_src;
 	unsigned			irq_type;
@@ -126,11 +129,33 @@ struct amdgpu_fence_driver {
 	struct dma_fence		**fences;
 };
 
+/*
+ * Fences mark an event in the GPUs pipeline and are used
+ * for GPU/CPU synchronization.  When the fence is written,
+ * it is expected that all buffers associated with that fence
+ * are no longer in use by the associated ring on the GPU and
+ * that the relevant GPU caches have been flushed.
+ */
+
+struct amdgpu_fence {
+	struct dma_fence base;
+
+	/* RB, DMA, etc. */
+	struct amdgpu_ring		*ring;
+	ktime_t				start_timestamp;
+
+	/* wptr for the fence for resets */
+	u64				wptr;
+	/* fence context for resets */
+	u64				context;
+};
+
 extern const struct drm_sched_backend_ops amdgpu_sched_ops;
 
-void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring);
 void amdgpu_fence_driver_set_error(struct amdgpu_ring *ring, int error);
 void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring);
+void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_fence *af);
+void amdgpu_fence_save_wptr(struct amdgpu_fence *af);
 
 int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring);
 int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring,
@@ -140,8 +165,8 @@ void amdgpu_fence_driver_hw_init(struct amdgpu_device *adev);
 void amdgpu_fence_driver_hw_fini(struct amdgpu_device *adev);
 int amdgpu_fence_driver_sw_init(struct amdgpu_device *adev);
 void amdgpu_fence_driver_sw_fini(struct amdgpu_device *adev);
-int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **fence, struct amdgpu_job *job,
-		      unsigned flags);
+int amdgpu_fence_emit(struct amdgpu_ring *ring, struct amdgpu_fence *af,
+		      unsigned int flags);
 int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s,
 			      uint32_t timeout);
 bool amdgpu_fence_process(struct amdgpu_ring *ring);
@@ -163,13 +188,40 @@ void amdgpu_fence_update_start_timestamp(struct amdgpu_ring *ring, uint32_t seq,
 
 /* provided by hw blocks that expose a ring buffer for commands */
 struct amdgpu_ring_funcs {
+	/**
+	 * @type:
+	 *
+	 * GFX, Compute, SDMA, UVD, VCE, VCN, VPE, KIQ, MES, UMSCH, and CPER
+	 * use ring buffers. The type field just identifies which component the
+	 * ring buffer is associated with.
+	 */
 	enum amdgpu_ring_type	type;
 	uint32_t		align_mask;
+
+	/**
+	 * @nop:
+	 *
+	 * Every block in the amdgpu has no-op instructions (e.g., GFX 10
+	 * uses PACKET3(PACKET3_NOP, 0x3FFF), VCN 5 uses VCN_ENC_CMD_NO_OP,
+	 * etc). This field receives the specific no-op for the component
+	 * that initializes the ring.
+	 */
 	u32			nop;
 	bool			support_64bit_ptrs;
 	bool			no_user_fence;
 	bool			secure_submission_supported;
-	unsigned		extra_dw;
+
+	/**
+	 * @extra_bytes:
+	 *
+	 * Optional extra space in bytes that is added to the ring size
+	 * when allocating the BO that holds the contents of the ring.
+	 * This space isn't used for command submission to the ring,
+	 * but is just there to satisfy some hardware requirements or
+	 * implement workarounds. It's up to the implementation of each
+	 * specific ring to initialize this space.
+	 */
+	unsigned		extra_bytes;
 
 	/* ring read/write ptr handling */
 	u64 (*get_rptr)(struct amdgpu_ring *ring);
@@ -235,10 +287,14 @@ struct amdgpu_ring_funcs {
 	void (*patch_cntl)(struct amdgpu_ring *ring, unsigned offset);
 	void (*patch_ce)(struct amdgpu_ring *ring, unsigned offset);
 	void (*patch_de)(struct amdgpu_ring *ring, unsigned offset);
-	int (*reset)(struct amdgpu_ring *ring, unsigned int vmid);
+	int (*reset)(struct amdgpu_ring *ring, unsigned int vmid,
+		     struct amdgpu_fence *timedout_fence);
 	void (*emit_cleaner_shader)(struct amdgpu_ring *ring);
 };
 
+/**
+ * amdgpu_ring - Holds ring information
+ */
 struct amdgpu_ring {
 	struct amdgpu_device		*adev;
 	const struct amdgpu_ring_funcs	*funcs;
@@ -247,16 +303,67 @@ struct amdgpu_ring {
 
 	struct amdgpu_bo	*ring_obj;
 	uint32_t		*ring;
+	/* backups for resets */
+	uint32_t		*ring_backup;
+	unsigned int		ring_backup_entries_to_copy;
 	unsigned		rptr_offs;
 	u64			rptr_gpu_addr;
-	volatile u32		*rptr_cpu_addr;
+	u32			*rptr_cpu_addr;
+
+	/**
+	 * @wptr:
+	 *
+	 * This is part of the Ring buffer implementation and represents the
+	 * write pointer. The wptr determines where the host has written.
+	 */
 	u64			wptr;
+
+	/**
+	 * @wptr_old:
+	 *
+	 * Before update wptr with the new value, usually the old value is
+	 * stored in the wptr_old.
+	 */
 	u64			wptr_old;
 	unsigned		ring_size;
+
+	/**
+	 * @max_dw:
+	 *
+	 * Maximum number of DWords for ring allocation. This information is
+	 * provided at the ring initialization time, and each IP block can
+	 * specify a specific value. Check places that invoke
+	 * amdgpu_ring_init() to see the maximum size per block.
+	 */
 	unsigned		max_dw;
+
+	/**
+	 * @count_dw:
+	 *
+	 * This value starts with the maximum amount of DWords supported by the
+	 * ring. This value is updated based on the ring manipulation.
+	 */
 	int			count_dw;
 	uint64_t		gpu_addr;
+
+	/**
+	 * @ptr_mask:
+	 *
+	 * Some IPs provide support for 64-bit pointers and others for 32-bit
+	 * only; this behavior is component-specific and defined by the field
+	 * support_64bit_ptr. If the IP block supports 64-bits, the mask
+	 * 0xffffffffffffffff is set; otherwise, this value assumes buf_mask.
+	 * Notice that this field is used to keep wptr under a valid range.
+	 */
 	uint64_t		ptr_mask;
+
+	/**
+	 * @buf_mask:
+	 *
+	 * Buffer mask is a value used to keep wptr count under its
+	 * thresholding. Buffer mask initialized during the ring buffer
+	 * initialization time, and it is defined as (ring_size / 4) -1.
+	 */
 	uint32_t		buf_mask;
 	u32			idx;
 	u32			xcc_id;
@@ -274,19 +381,26 @@ struct amdgpu_ring {
 	bool			use_pollmem;
 	unsigned		wptr_offs;
 	u64			wptr_gpu_addr;
-	volatile u32		*wptr_cpu_addr;
+
+	/**
+	 * @wptr_cpu_addr:
+	 *
+	 * This is the CPU address pointer in the writeback slot. This is used
+	 * to commit changes to the GPU.
+	 */
+	u32			*wptr_cpu_addr;
 	unsigned		fence_offs;
 	u64			fence_gpu_addr;
-	volatile u32		*fence_cpu_addr;
+	u32			*fence_cpu_addr;
 	uint64_t		current_ctx;
 	char			name[16];
 	u32                     trail_seq;
 	unsigned		trail_fence_offs;
 	u64			trail_fence_gpu_addr;
-	volatile u32		*trail_fence_cpu_addr;
+	u32			*trail_fence_cpu_addr;
 	unsigned		cond_exe_offs;
 	u64			cond_exe_gpu_addr;
-	volatile u32		*cond_exe_cpu_addr;
+	u32			*cond_exe_cpu_addr;
 	unsigned int		set_q_mode_offs;
 	u32			*set_q_mode_ptr;
 	u64			set_q_mode_token;
@@ -295,18 +409,15 @@ struct amdgpu_ring {
 	struct dma_fence	*vmid_wait;
 	bool			has_compute_vm_bug;
 	bool			no_scheduler;
+	bool			no_user_submission;
 	int			hw_prio;
 	unsigned 		num_hw_submission;
 	atomic_t		*sched_score;
 
-	/* used for mes */
-	bool			is_mes_queue;
-	uint32_t		hw_queue_id;
-	struct amdgpu_mes_ctx_data *mes_ctx;
-
 	bool            is_sw_ring;
 	unsigned int    entry_index;
-
+	/* store the cached rptr to restore after reset */
+	uint64_t cached_rptr;
 };
 
 #define amdgpu_ring_parse_cs(r, p, job, ib) ((r)->funcs->parse_cs((p), (job), (ib)))
@@ -336,7 +447,7 @@ struct amdgpu_ring {
 #define amdgpu_ring_patch_cntl(r, o) ((r)->funcs->patch_cntl((r), (o)))
 #define amdgpu_ring_patch_ce(r, o) ((r)->funcs->patch_ce((r), (o)))
 #define amdgpu_ring_patch_de(r, o) ((r)->funcs->patch_de((r), (o)))
-#define amdgpu_ring_reset(r, v) (r)->funcs->reset((r), (v))
+#define amdgpu_ring_reset(r, v, f) (r)->funcs->reset((r), (v), (f))
 
 unsigned int amdgpu_ring_max_ibs(enum amdgpu_ring_type type);
 int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw);
@@ -369,10 +480,7 @@ static inline void amdgpu_ring_set_preempt_cond_exec(struct amdgpu_ring *ring,
 
 static inline void amdgpu_ring_clear_ring(struct amdgpu_ring *ring)
 {
-	int i = 0;
-	while (i <= ring->buf_mask)
-		ring->ring[i++] = ring->funcs->nop;
-
+	memset32(ring->ring, ring->funcs->nop, ring->buf_mask + 1);
 }
 
 static inline void amdgpu_ring_write(struct amdgpu_ring *ring, uint32_t v)
@@ -431,15 +539,6 @@ static inline void amdgpu_ring_patch_cond_exec(struct amdgpu_ring *ring,
 	ring->ring[offset] = cur - offset;
 }
 
-#define amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset)			\
-	(ring->is_mes_queue && ring->mes_ctx ?				\
-	 (ring->mes_ctx->meta_data_gpu_addr + offset) : 0)
-
-#define amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset)			\
-	(ring->is_mes_queue && ring->mes_ctx ?				\
-	 (void *)((uint8_t *)(ring->mes_ctx->meta_data_ptr) + offset) : \
-	 NULL)
-
 int amdgpu_ring_test_helper(struct amdgpu_ring *ring);
 
 void amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
@@ -470,4 +569,12 @@ int amdgpu_ib_pool_init(struct amdgpu_device *adev);
 void amdgpu_ib_pool_fini(struct amdgpu_device *adev);
 int amdgpu_ib_ring_tests(struct amdgpu_device *adev);
 bool amdgpu_ring_sched_ready(struct amdgpu_ring *ring);
+void amdgpu_ring_backup_unprocessed_commands(struct amdgpu_ring *ring,
+					     struct amdgpu_fence *guilty_fence);
+void amdgpu_ring_reset_helper_begin(struct amdgpu_ring *ring,
+				    struct amdgpu_fence *guilty_fence);
+int amdgpu_ring_reset_helper_end(struct amdgpu_ring *ring,
+				 struct amdgpu_fence *guilty_fence);
+bool amdgpu_ring_is_reset_type_supported(struct amdgpu_ring *ring,
+					 u32 reset_type);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c
index 1c66da1c3fb4..7e7d6c3865bc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring_mux.c
@@ -124,7 +124,7 @@ static void amdgpu_mux_resubmit_chunks(struct amdgpu_ring_mux *mux)
 		}
 	}
 
-	del_timer(&mux->resubmit_timer);
+	timer_delete(&mux->resubmit_timer);
 	mux->s_resubmit = false;
 }
 
@@ -135,7 +135,8 @@ static void amdgpu_ring_mux_schedule_resubmit(struct amdgpu_ring_mux *mux)
 
 static void amdgpu_mux_resubmit_fallback(struct timer_list *t)
 {
-	struct amdgpu_ring_mux *mux = from_timer(mux, t, resubmit_timer);
+	struct amdgpu_ring_mux *mux = timer_container_of(mux, t,
+							 resubmit_timer);
 
 	if (!spin_trylock(&mux->lock)) {
 		amdgpu_ring_mux_schedule_resubmit(mux);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c
index db5791e1a7ce..5aa830a02d80 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c
@@ -89,7 +89,7 @@ void amdgpu_gfx_rlc_exit_safe_mode(struct amdgpu_device *adev, int xcc_id)
 int amdgpu_gfx_rlc_init_sr(struct amdgpu_device *adev, u32 dws)
 {
 	const u32 *src_ptr;
-	volatile u32 *dst_ptr;
+	u32 *dst_ptr;
 	u32 i;
 	int r;
 
@@ -189,7 +189,7 @@ int amdgpu_gfx_rlc_init_cpt(struct amdgpu_device *adev)
 void amdgpu_gfx_rlc_setup_cp_table(struct amdgpu_device *adev)
 {
 	const __le32 *fw_data;
-	volatile u32 *dst_ptr;
+	u32 *dst_ptr;
 	int me, i, max_me;
 	u32 bo_offset = 0;
 	u32 table_offset, table_size;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h
index fce22d3f816b..2ce310b31942 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h
@@ -237,7 +237,21 @@ struct amdgpu_rlc_funcs {
 	void (*unset_safe_mode)(struct amdgpu_device *adev, int xcc_id);
 	int  (*init)(struct amdgpu_device *adev);
 	u32  (*get_csb_size)(struct amdgpu_device *adev);
-	void (*get_csb_buffer)(struct amdgpu_device *adev, volatile u32 *buffer);
+
+	/**
+	 * @get_csb_buffer: Get the clear state to be put into the hardware.
+	 *
+	 * The parameter adev is used to get the CS data and other gfx info,
+	 * and buffer is the RLC CS pointer
+	 *
+	 * Sometimes, the user space puts a request to clear the state in the
+	 * command buffer; this function provides the clear state that gets put
+	 * into the hardware. Note that the driver programs Clear State
+	 * Indirect Buffer (CSB) explicitly when it sets up the kernel rings,
+	 * and it also provides a pointer to it which is used by the firmware
+	 * to load the clear state in some cases.
+	 */
+	void (*get_csb_buffer)(struct amdgpu_device *adev, u32 *buffer);
 	int  (*get_cp_table_num)(struct amdgpu_device *adev);
 	int  (*resume)(struct amdgpu_device *adev);
 	void (*stop)(struct amdgpu_device *adev);
@@ -261,19 +275,19 @@ struct amdgpu_rlc {
 	/* for power gating */
 	struct amdgpu_bo        *save_restore_obj;
 	uint64_t                save_restore_gpu_addr;
-	volatile uint32_t       *sr_ptr;
+	uint32_t		*sr_ptr;
 	const u32               *reg_list;
 	u32                     reg_list_size;
 	/* for clear state */
 	struct amdgpu_bo        *clear_state_obj;
 	uint64_t                clear_state_gpu_addr;
-	volatile uint32_t       *cs_ptr;
+	uint32_t		*cs_ptr;
 	const struct cs_section_def   *cs_data;
 	u32                     clear_state_size;
 	/* for cp tables */
 	struct amdgpu_bo        *cp_table_obj;
 	uint64_t                cp_table_gpu_addr;
-	volatile uint32_t       *cp_table_ptr;
+	uint32_t		*cp_table_ptr;
 	u32                     cp_table_size;
 
 	/* safe mode for updating CG/PG state */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
index 174badca27e7..8b8a04138711 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
@@ -25,6 +25,9 @@
 #include "amdgpu.h"
 #include "amdgpu_sdma.h"
 #include "amdgpu_ras.h"
+#include "amdgpu_reset.h"
+#include "gc/gc_10_1_0_offset.h"
+#include "gc/gc_10_3_0_sh_mask.h"
 
 #define AMDGPU_CSA_SDMA_SIZE 64
 /* SDMA CSA reside in the 3rd page of CSA */
@@ -75,22 +78,14 @@ uint64_t amdgpu_sdma_get_csa_mc_addr(struct amdgpu_ring *ring,
 	if (amdgpu_sriov_vf(adev) || vmid == 0 || !adev->gfx.mcbp)
 		return 0;
 
-	if (ring->is_mes_queue) {
-		uint32_t offset = 0;
+	r = amdgpu_sdma_get_index_from_ring(ring, &index);
 
-		offset = offsetof(struct amdgpu_mes_ctx_meta_data,
-				  sdma[ring->idx].sdma_meta_data);
-		csa_mc_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
-	} else {
-		r = amdgpu_sdma_get_index_from_ring(ring, &index);
-
-		if (r || index > 31)
-			csa_mc_addr = 0;
-		else
-			csa_mc_addr = amdgpu_csa_vaddr(adev) +
-				AMDGPU_CSA_SDMA_OFFSET +
-				index * AMDGPU_CSA_SDMA_SIZE;
-	}
+	if (r || index > 31)
+		csa_mc_addr = 0;
+	else
+		csa_mc_addr = amdgpu_csa_vaddr(adev) +
+			AMDGPU_CSA_SDMA_OFFSET +
+			index * AMDGPU_CSA_SDMA_SIZE;
 
 	return csa_mc_addr;
 }
@@ -355,23 +350,44 @@ int amdgpu_sdma_ras_sw_init(struct amdgpu_device *adev)
 static int amdgpu_debugfs_sdma_sched_mask_set(void *data, u64 val)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)data;
-	u32 i;
+	u64 i, num_ring;
 	u64 mask = 0;
-	struct amdgpu_ring *ring;
+	struct amdgpu_ring *ring, *page = NULL;
 
 	if (!adev)
 		return -ENODEV;
 
-	mask = BIT_ULL(adev->sdma.num_instances) - 1;
+	/* Determine the number of rings per SDMA instance
+	 * (1 for sdma gfx ring, 2 if page queue exists)
+	 */
+	if (adev->sdma.has_page_queue)
+		num_ring = 2;
+	else
+		num_ring = 1;
+
+	/* Calculate the maximum possible mask value
+	 * based on the number of SDMA instances and rings
+	*/
+	mask = BIT_ULL(adev->sdma.num_instances * num_ring) - 1;
+
 	if ((val & mask) == 0)
 		return -EINVAL;
 
 	for (i = 0; i < adev->sdma.num_instances; ++i) {
 		ring = &adev->sdma.instance[i].ring;
-		if (val & BIT_ULL(i))
+		if (adev->sdma.has_page_queue)
+			page = &adev->sdma.instance[i].page;
+		if (val & BIT_ULL(i * num_ring))
 			ring->sched.ready = true;
 		else
 			ring->sched.ready = false;
+
+		if (page) {
+			if (val & BIT_ULL(i * num_ring + 1))
+				page->sched.ready = true;
+			else
+				page->sched.ready = false;
+		}
 	}
 	/* publish sched.ready flag update effective immediately across smp */
 	smp_rmb();
@@ -381,16 +397,37 @@ static int amdgpu_debugfs_sdma_sched_mask_set(void *data, u64 val)
 static int amdgpu_debugfs_sdma_sched_mask_get(void *data, u64 *val)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)data;
-	u32 i;
+	u64 i, num_ring;
 	u64 mask = 0;
-	struct amdgpu_ring *ring;
+	struct amdgpu_ring *ring, *page = NULL;
 
 	if (!adev)
 		return -ENODEV;
+
+	/* Determine the number of rings per SDMA instance
+	 * (1 for sdma gfx ring, 2 if page queue exists)
+	 */
+	if (adev->sdma.has_page_queue)
+		num_ring = 2;
+	else
+		num_ring = 1;
+
 	for (i = 0; i < adev->sdma.num_instances; ++i) {
 		ring = &adev->sdma.instance[i].ring;
+		if (adev->sdma.has_page_queue)
+			page = &adev->sdma.instance[i].page;
+
 		if (ring->sched.ready)
-			mask |= 1 << i;
+			mask |= BIT_ULL(i * num_ring);
+		else
+			mask &= ~BIT_ULL(i * num_ring);
+
+		if (page) {
+			if (page->sched.ready)
+				mask |= BIT_ULL(i * num_ring + 1);
+			else
+				mask &= ~BIT_ULL(i * num_ring + 1);
+		}
 	}
 
 	*val = mask;
@@ -460,3 +497,115 @@ void amdgpu_sdma_sysfs_reset_mask_fini(struct amdgpu_device *adev)
 			device_remove_file(adev->dev, &dev_attr_sdma_reset_mask);
 	}
 }
+
+struct amdgpu_ring *amdgpu_sdma_get_shared_ring(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (adev->sdma.has_page_queue &&
+	    (ring->me < adev->sdma.num_instances) &&
+	    (ring == &adev->sdma.instance[ring->me].ring))
+		return &adev->sdma.instance[ring->me].page;
+	else
+		return NULL;
+}
+
+/**
+* amdgpu_sdma_is_shared_inv_eng - Check if a ring is an SDMA ring that shares a VM invalidation engine
+* @adev: Pointer to the AMDGPU device structure
+* @ring: Pointer to the ring structure to check
+*
+* This function checks if the given ring is an SDMA ring that shares a VM invalidation engine.
+* It returns true if the ring is such an SDMA ring, false otherwise.
+*/
+bool amdgpu_sdma_is_shared_inv_eng(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	int i = ring->me;
+
+	if (!adev->sdma.has_page_queue || i >= adev->sdma.num_instances)
+		return false;
+
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+		return (ring == &adev->sdma.instance[i].page);
+	else
+		return false;
+}
+
+static int amdgpu_sdma_soft_reset(struct amdgpu_device *adev, u32 instance_id)
+{
+	struct amdgpu_sdma_instance *sdma_instance = &adev->sdma.instance[instance_id];
+
+	if (sdma_instance->funcs->soft_reset_kernel_queue)
+		return sdma_instance->funcs->soft_reset_kernel_queue(adev, instance_id);
+
+	return -EOPNOTSUPP;
+}
+
+/**
+ * amdgpu_sdma_reset_engine - Reset a specific SDMA engine
+ * @adev: Pointer to the AMDGPU device
+ * @instance_id: Logical ID of the SDMA engine instance to reset
+ * @caller_handles_kernel_queues: Skip kernel queue processing. Caller
+ * will handle it.
+ *
+ * Returns: 0 on success, or a negative error code on failure.
+ */
+int amdgpu_sdma_reset_engine(struct amdgpu_device *adev, uint32_t instance_id,
+			     bool caller_handles_kernel_queues)
+{
+	int ret = 0;
+	struct amdgpu_sdma_instance *sdma_instance = &adev->sdma.instance[instance_id];
+	struct amdgpu_ring *gfx_ring = &sdma_instance->ring;
+	struct amdgpu_ring *page_ring = &sdma_instance->page;
+
+	mutex_lock(&sdma_instance->engine_reset_mutex);
+
+	if (!caller_handles_kernel_queues) {
+		/* Stop the scheduler's work queue for the GFX and page rings if they are running.
+		 * This ensures that no new tasks are submitted to the queues while
+		 * the reset is in progress.
+		 */
+		drm_sched_wqueue_stop(&gfx_ring->sched);
+
+		if (adev->sdma.has_page_queue)
+			drm_sched_wqueue_stop(&page_ring->sched);
+	}
+
+	if (sdma_instance->funcs->stop_kernel_queue) {
+		sdma_instance->funcs->stop_kernel_queue(gfx_ring);
+		if (adev->sdma.has_page_queue)
+			sdma_instance->funcs->stop_kernel_queue(page_ring);
+	}
+
+	/* Perform the SDMA reset for the specified instance */
+	ret = amdgpu_sdma_soft_reset(adev, instance_id);
+	if (ret) {
+		dev_err(adev->dev, "Failed to reset SDMA logical instance %u\n", instance_id);
+		goto exit;
+	}
+
+	if (sdma_instance->funcs->start_kernel_queue) {
+		sdma_instance->funcs->start_kernel_queue(gfx_ring);
+		if (adev->sdma.has_page_queue)
+			sdma_instance->funcs->start_kernel_queue(page_ring);
+	}
+
+exit:
+	if (!caller_handles_kernel_queues) {
+		/* Restart the scheduler's work queue for the GFX and page rings
+		 * if they were stopped by this function. This allows new tasks
+		 * to be submitted to the queues after the reset is complete.
+		 */
+		if (!ret) {
+			amdgpu_fence_driver_force_completion(gfx_ring);
+			drm_sched_wqueue_start(&gfx_ring->sched);
+			if (adev->sdma.has_page_queue) {
+				amdgpu_fence_driver_force_completion(page_ring);
+				drm_sched_wqueue_start(&page_ring->sched);
+			}
+		}
+	}
+	mutex_unlock(&sdma_instance->engine_reset_mutex);
+
+	return ret;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h
index 5f60736051d1..34311f32be4c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h
@@ -50,6 +50,12 @@ enum amdgpu_sdma_irq {
 
 #define NUM_SDMA(x) hweight32(x)
 
+struct amdgpu_sdma_funcs {
+	int (*stop_kernel_queue)(struct amdgpu_ring *ring);
+	int (*start_kernel_queue)(struct amdgpu_ring *ring);
+	int (*soft_reset_kernel_queue)(struct amdgpu_device *adev, u32 instance_id);
+};
+
 struct amdgpu_sdma_instance {
 	/* SDMA firmware */
 	const struct firmware	*fw;
@@ -64,6 +70,11 @@ struct amdgpu_sdma_instance {
 	struct amdgpu_bo	*sdma_fw_obj;
 	uint64_t		sdma_fw_gpu_addr;
 	uint32_t		*sdma_fw_ptr;
+	struct mutex		engine_reset_mutex;
+	/* track guilty state of GFX and PAGE queues */
+	bool			gfx_guilty;
+	bool			page_guilty;
+	const struct amdgpu_sdma_funcs   *funcs;
 };
 
 enum amdgpu_sdma_ras_memory_id {
@@ -102,6 +113,7 @@ struct amdgpu_sdma {
 	struct amdgpu_sdma_instance instance[AMDGPU_MAX_SDMA_INSTANCES];
 	struct amdgpu_irq_src	trap_irq;
 	struct amdgpu_irq_src	illegal_inst_irq;
+	struct amdgpu_irq_src	fence_irq;
 	struct amdgpu_irq_src	ecc_irq;
 	struct amdgpu_irq_src	vm_hole_irq;
 	struct amdgpu_irq_src	doorbell_invalid_irq;
@@ -118,6 +130,9 @@ struct amdgpu_sdma {
 	struct amdgpu_sdma_ras	*ras;
 	uint32_t		*ip_dump;
 	uint32_t 		supported_reset;
+	struct list_head	reset_callback_list;
+	bool			no_user_submission;
+	bool			disable_uq;
 };
 
 /*
@@ -157,6 +172,9 @@ struct amdgpu_buffer_funcs {
 				 uint32_t byte_count);
 };
 
+int amdgpu_sdma_reset_engine(struct amdgpu_device *adev, uint32_t instance_id,
+			     bool caller_handles_kernel_queues);
+
 #define amdgpu_emit_copy_buffer(adev, ib, s, d, b, t) (adev)->mman.buffer_funcs->emit_copy_buffer((ib),  (s), (d), (b), (t))
 #define amdgpu_emit_fill_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_fill_buffer((ib), (s), (d), (b))
 
@@ -180,4 +198,7 @@ int amdgpu_sdma_ras_sw_init(struct amdgpu_device *adev);
 void amdgpu_debugfs_sdma_sched_mask_init(struct amdgpu_device *adev);
 int amdgpu_sdma_sysfs_reset_mask_init(struct amdgpu_device *adev);
 void amdgpu_sdma_sysfs_reset_mask_fini(struct amdgpu_device *adev);
+bool amdgpu_sdma_is_shared_inv_eng(struct amdgpu_device *adev, struct amdgpu_ring *ring);
+struct amdgpu_ring *amdgpu_sdma_get_shared_ring(struct amdgpu_device *adev,
+	struct amdgpu_ring *ring);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c
index 41ebe690eeff..3739be1b71e0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c
@@ -159,7 +159,6 @@ static ssize_t amdgpu_securedisplay_debugfs_write(struct file *f, const char __u
 		dev_err(adev->dev, "Invalid input: %s\n", str);
 	}
 
-	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
 
 	return size;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c
index e22cb2b5cd92..a0b479d5fff1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c
@@ -45,7 +45,11 @@
  */
 static inline u64 amdgpu_seq64_get_va_base(struct amdgpu_device *adev)
 {
-	return AMDGPU_VA_RESERVED_SEQ64_START(adev);
+	u64 addr = AMDGPU_VA_RESERVED_SEQ64_START(adev);
+
+	addr = amdgpu_gmc_sign_extend(addr);
+
+	return addr;
 }
 
 /**
@@ -88,9 +92,11 @@ int amdgpu_seq64_map(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		goto error;
 	}
 
-	seq64_addr = amdgpu_seq64_get_va_base(adev);
-	r = amdgpu_vm_bo_map(adev, *bo_va, seq64_addr, 0, AMDGPU_VA_RESERVED_SEQ64_SIZE,
-			     AMDGPU_PTE_READABLE);
+	seq64_addr = amdgpu_seq64_get_va_base(adev) & AMDGPU_GMC_HOLE_MASK;
+
+	r = amdgpu_vm_bo_map(adev, *bo_va, seq64_addr, 0,
+			     AMDGPU_VA_RESERVED_SEQ64_SIZE,
+			     AMDGPU_VM_PAGE_READABLE | AMDGPU_VM_MTYPE_UC);
 	if (r) {
 		DRM_ERROR("failed to do bo_map on userq sem, err=%d\n", r);
 		amdgpu_vm_bo_del(adev, *bo_va);
@@ -133,7 +139,7 @@ void amdgpu_seq64_unmap(struct amdgpu_device *adev, struct amdgpu_fpriv *fpriv)
 
 	vm = &fpriv->vm;
 
-	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT, 0);
+	drm_exec_init(&exec, 0, 0);
 	drm_exec_until_all_locked(&exec) {
 		r = amdgpu_vm_lock_pd(vm, &exec, 0);
 		if (likely(!r))
@@ -156,6 +162,7 @@ error:
  *
  * @adev: amdgpu_device pointer
  * @va: VA to access the seq in process address space
+ * @gpu_addr: GPU address to access the seq
  * @cpu_addr: CPU address to access the seq
  *
  * Alloc a 64 bit memory from seq64 pool.
@@ -163,7 +170,8 @@ error:
  * Returns:
  * 0 on success or a negative error code on failure
  */
-int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *va, u64 **cpu_addr)
+int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *va,
+		       u64 *gpu_addr, u64 **cpu_addr)
 {
 	unsigned long bit_pos;
 
@@ -172,7 +180,12 @@ int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *va, u64 **cpu_addr)
 		return -ENOSPC;
 
 	__set_bit(bit_pos, adev->seq64.used);
+
 	*va = bit_pos * sizeof(u64) + amdgpu_seq64_get_va_base(adev);
+
+	if (gpu_addr)
+		*gpu_addr = bit_pos * sizeof(u64) + adev->seq64.gpu_addr;
+
 	*cpu_addr = bit_pos + adev->seq64.cpu_base_addr;
 
 	return 0;
@@ -233,7 +246,7 @@ int amdgpu_seq64_init(struct amdgpu_device *adev)
 	 */
 	r = amdgpu_bo_create_kernel(adev, AMDGPU_VA_RESERVED_SEQ64_SIZE,
 				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
-				    &adev->seq64.sbo, NULL,
+				    &adev->seq64.sbo, &adev->seq64.gpu_addr,
 				    (void **)&adev->seq64.cpu_base_addr);
 	if (r) {
 		dev_warn(adev->dev, "(%d) create seq64 failed\n", r);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h
index 4203b2ab318d..26a249aaaee1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h
@@ -32,13 +32,14 @@
 struct amdgpu_seq64 {
 	struct amdgpu_bo *sbo;
 	u32 num_sem;
+	u64 gpu_addr;
 	u64 *cpu_base_addr;
 	DECLARE_BITMAP(used, AMDGPU_MAX_SEQ64_SLOTS);
 };
 
 void amdgpu_seq64_fini(struct amdgpu_device *adev);
 int amdgpu_seq64_init(struct amdgpu_device *adev);
-int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *gpu_addr, u64 **cpu_addr);
+int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *va, u64 *gpu_addr, u64 **cpu_addr);
 void amdgpu_seq64_free(struct amdgpu_device *adev, u64 gpu_addr);
 int amdgpu_seq64_map(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		     struct amdgpu_bo_va **bo_va);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
index c586ab4c911b..d6ae9974c952 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
@@ -135,11 +135,16 @@ static bool amdgpu_sync_add_later(struct amdgpu_sync *sync, struct dma_fence *f)
 	struct amdgpu_sync_entry *e;
 
 	hash_for_each_possible(sync->fences, e, node, f->context) {
-		if (unlikely(e->fence->context != f->context))
-			continue;
+		if (dma_fence_is_signaled(e->fence)) {
+			dma_fence_put(e->fence);
+			e->fence = dma_fence_get(f);
+			return true;
+		}
 
-		amdgpu_sync_keep_later(&e->fence, f);
-		return true;
+		if (likely(e->fence->context == f->context)) {
+			amdgpu_sync_keep_later(&e->fence, f);
+			return true;
+		}
 	}
 	return false;
 }
@@ -149,10 +154,12 @@ static bool amdgpu_sync_add_later(struct amdgpu_sync *sync, struct dma_fence *f)
  *
  * @sync: sync object to add fence to
  * @f: fence to sync to
+ * @flags: memory allocation flags to use when allocating sync entry
  *
  * Add the fence to the sync object.
  */
-int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f)
+int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f,
+		      gfp_t flags)
 {
 	struct amdgpu_sync_entry *e;
 
@@ -162,7 +169,7 @@ int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f)
 	if (amdgpu_sync_add_later(sync, f))
 		return 0;
 
-	e = kmem_cache_alloc(amdgpu_sync_slab, GFP_KERNEL);
+	e = kmem_cache_alloc(amdgpu_sync_slab, flags);
 	if (!e)
 		return -ENOMEM;
 
@@ -242,14 +249,13 @@ int amdgpu_sync_resv(struct amdgpu_device *adev, struct amdgpu_sync *sync,
 
 	if (resv == NULL)
 		return -EINVAL;
-
-	/* TODO: Use DMA_RESV_USAGE_READ here */
-	dma_resv_for_each_fence(&cursor, resv, DMA_RESV_USAGE_BOOKKEEP, f) {
+	/* Implicitly sync only to KERNEL, WRITE and READ */
+	dma_resv_for_each_fence(&cursor, resv, DMA_RESV_USAGE_READ, f) {
 		dma_fence_chain_for_each(f, f) {
 			struct dma_fence *tmp = dma_fence_chain_contained(f);
 
 			if (amdgpu_sync_test_fence(adev, mode, owner, tmp)) {
-				r = amdgpu_sync_fence(sync, f);
+				r = amdgpu_sync_fence(sync, f, GFP_KERNEL);
 				dma_fence_put(f);
 				if (r)
 					return r;
@@ -281,7 +287,7 @@ int amdgpu_sync_kfd(struct amdgpu_sync *sync, struct dma_resv *resv)
 		if (fence_owner != AMDGPU_FENCE_OWNER_KFD)
 			continue;
 
-		r = amdgpu_sync_fence(sync, f);
+		r = amdgpu_sync_fence(sync, f, GFP_KERNEL);
 		if (r)
 			break;
 	}
@@ -388,7 +394,7 @@ int amdgpu_sync_clone(struct amdgpu_sync *source, struct amdgpu_sync *clone)
 	hash_for_each_safe(source->fences, i, tmp, e, node) {
 		f = e->fence;
 		if (!dma_fence_is_signaled(f)) {
-			r = amdgpu_sync_fence(clone, f);
+			r = amdgpu_sync_fence(clone, f, GFP_KERNEL);
 			if (r)
 				return r;
 		} else {
@@ -400,6 +406,25 @@ int amdgpu_sync_clone(struct amdgpu_sync *source, struct amdgpu_sync *clone)
 }
 
 /**
+ * amdgpu_sync_move - move all fences from src to dst
+ *
+ * @src: source of the fences, empty after function
+ * @dst: destination for the fences
+ *
+ * Moves all fences from source to destination. All fences in destination are
+ * freed and source is empty after the function call.
+ */
+void amdgpu_sync_move(struct amdgpu_sync *src, struct amdgpu_sync *dst)
+{
+	unsigned int i;
+
+	amdgpu_sync_free(dst);
+
+	for (i = 0; i < HASH_SIZE(src->fences); ++i)
+		hlist_move_list(&src->fences[i], &dst->fences[i]);
+}
+
+/**
  * amdgpu_sync_push_to_job - push fences into job
  * @sync: sync object to get the fences from
  * @job: job to push the fences into
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h
index e3272dce798d..51eb4382c91e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h
@@ -47,7 +47,8 @@ struct amdgpu_sync {
 };
 
 void amdgpu_sync_create(struct amdgpu_sync *sync);
-int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f);
+int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f,
+		      gfp_t flags);
 int amdgpu_sync_resv(struct amdgpu_device *adev, struct amdgpu_sync *sync,
 		     struct dma_resv *resv, enum amdgpu_sync_mode mode,
 		     void *owner);
@@ -56,6 +57,7 @@ struct dma_fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync,
 				     struct amdgpu_ring *ring);
 struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync);
 int amdgpu_sync_clone(struct amdgpu_sync *source, struct amdgpu_sync *clone);
+void amdgpu_sync_move(struct amdgpu_sync *src, struct amdgpu_sync *dst);
 int amdgpu_sync_push_to_job(struct amdgpu_sync *sync, struct amdgpu_job *job);
 int amdgpu_sync_wait(struct amdgpu_sync *sync, bool intr);
 void amdgpu_sync_free(struct amdgpu_sync *sync);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
index 383fce40d4dd..d13e64a69e25 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -167,25 +167,23 @@ TRACE_EVENT(amdgpu_cs_ioctl,
 	    TP_PROTO(struct amdgpu_job *job),
 	    TP_ARGS(job),
 	    TP_STRUCT__entry(
-			     __field(uint64_t, sched_job_id)
 			     __string(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
-			     __field(unsigned int, context)
-			     __field(unsigned int, seqno)
+			     __field(u64, context)
+			     __field(u64, seqno)
 			     __field(struct dma_fence *, fence)
 			     __string(ring, to_amdgpu_ring(job->base.sched)->name)
 			     __field(u32, num_ibs)
 			     ),
 
 	    TP_fast_assign(
-			   __entry->sched_job_id = job->base.id;
 			   __assign_str(timeline);
 			   __entry->context = job->base.s_fence->finished.context;
 			   __entry->seqno = job->base.s_fence->finished.seqno;
 			   __assign_str(ring);
 			   __entry->num_ibs = job->num_ibs;
 			   ),
-	    TP_printk("sched_job=%llu, timeline=%s, context=%u, seqno=%u, ring_name=%s, num_ibs=%u",
-		      __entry->sched_job_id, __get_str(timeline), __entry->context,
+	    TP_printk("timeline=%s, fence=%llu:%llu, ring_name=%s, num_ibs=%u",
+		      __get_str(timeline), __entry->context,
 		      __entry->seqno, __get_str(ring), __entry->num_ibs)
 );
 
@@ -193,24 +191,22 @@ TRACE_EVENT(amdgpu_sched_run_job,
 	    TP_PROTO(struct amdgpu_job *job),
 	    TP_ARGS(job),
 	    TP_STRUCT__entry(
-			     __field(uint64_t, sched_job_id)
 			     __string(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
-			     __field(unsigned int, context)
-			     __field(unsigned int, seqno)
+			     __field(u64, context)
+			     __field(u64, seqno)
 			     __string(ring, to_amdgpu_ring(job->base.sched)->name)
 			     __field(u32, num_ibs)
 			     ),
 
 	    TP_fast_assign(
-			   __entry->sched_job_id = job->base.id;
 			   __assign_str(timeline);
 			   __entry->context = job->base.s_fence->finished.context;
 			   __entry->seqno = job->base.s_fence->finished.seqno;
 			   __assign_str(ring);
 			   __entry->num_ibs = job->num_ibs;
 			   ),
-	    TP_printk("sched_job=%llu, timeline=%s, context=%u, seqno=%u, ring_name=%s, num_ibs=%u",
-		      __entry->sched_job_id, __get_str(timeline), __entry->context,
+	    TP_printk("timeline=%s, fence=%llu:%llu, ring_name=%s, num_ibs=%u",
+		      __get_str(timeline), __entry->context,
 		      __entry->seqno, __get_str(ring), __entry->num_ibs)
 );
 
@@ -457,6 +453,38 @@ DEFINE_EVENT(amdgpu_pasid, amdgpu_pasid_freed,
 	    TP_ARGS(pasid)
 );
 
+TRACE_EVENT(amdgpu_isolation,
+	    TP_PROTO(void *prev, void *next),
+	    TP_ARGS(prev, next),
+	    TP_STRUCT__entry(
+			     __field(void *, prev)
+			     __field(void *, next)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->prev = prev;
+			   __entry->next = next;
+			   ),
+	    TP_printk("prev=%p, next=%p",
+		      __entry->prev,
+		      __entry->next)
+);
+
+TRACE_EVENT(amdgpu_cleaner_shader,
+	    TP_PROTO(struct amdgpu_ring *ring, struct dma_fence *fence),
+	    TP_ARGS(ring, fence),
+	    TP_STRUCT__entry(
+			     __string(ring, ring->name)
+			     __field(u64, seqno)
+			     ),
+
+	    TP_fast_assign(
+			   __assign_str(ring);
+			   __entry->seqno = fence->seqno;
+			   ),
+	    TP_printk("ring=%s, seqno=%Lu", __get_str(ring), __entry->seqno)
+);
+
 TRACE_EVENT(amdgpu_bo_list_set,
 	    TP_PROTO(struct amdgpu_bo_list *list, struct amdgpu_bo *bo),
 	    TP_ARGS(list, bo),
@@ -519,23 +547,19 @@ TRACE_EVENT(amdgpu_ib_pipe_sync,
 	    TP_ARGS(sched_job, fence),
 	    TP_STRUCT__entry(
 			     __string(ring, sched_job->base.sched->name)
-			     __field(uint64_t, id)
 			     __field(struct dma_fence *, fence)
-			     __field(uint64_t, ctx)
-			     __field(unsigned, seqno)
+			     __field(u64, ctx)
+			     __field(u64, seqno)
 			     ),
 
 	    TP_fast_assign(
 			   __assign_str(ring);
-			   __entry->id = sched_job->base.id;
 			   __entry->fence = fence;
 			   __entry->ctx = fence->context;
 			   __entry->seqno = fence->seqno;
 			   ),
-	    TP_printk("job ring=%s, id=%llu, need pipe sync to fence=%p, context=%llu, seq=%u",
-		      __get_str(ring), __entry->id,
-		      __entry->fence, __entry->ctx,
-		      __entry->seqno)
+	    TP_printk("job ring=%s need pipe sync to fence=%llu:%llu",
+		      __get_str(ring), __entry->ctx, __entry->seqno)
 );
 
 TRACE_EVENT(amdgpu_reset_reg_dumps,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index 01ae2f88dec8..2b931e855abd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -123,6 +123,7 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 	case AMDGPU_PL_GWS:
 	case AMDGPU_PL_OA:
 	case AMDGPU_PL_DOORBELL:
+	case AMDGPU_PL_MMIO_REMAP:
 		placement->num_placement = 0;
 		return;
 
@@ -187,7 +188,6 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
 	struct amdgpu_job *job;
 	void *cpu_addr;
 	uint64_t flags;
-	unsigned int i;
 	int r;
 
 	BUG_ON(adev->mman.buffer_funcs->copy_max_bytes <
@@ -226,7 +226,8 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
 	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.high_pr,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     num_dw * 4 + num_bytes,
-				     AMDGPU_IB_POOL_DELAYED, &job);
+				     AMDGPU_IB_POOL_DELAYED, &job,
+				     AMDGPU_KERNEL_JOB_ID_TTM_MAP_BUFFER);
 	if (r)
 		return r;
 
@@ -253,16 +254,9 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
 		dma_addr = &bo->ttm->dma_address[mm_cur->start >> PAGE_SHIFT];
 		amdgpu_gart_map(adev, 0, num_pages, dma_addr, flags, cpu_addr);
 	} else {
-		dma_addr_t dma_address;
-
-		dma_address = mm_cur->start;
-		dma_address += adev->vm_manager.vram_base_offset;
+		u64 pa = mm_cur->start + adev->vm_manager.vram_base_offset;
 
-		for (i = 0; i < num_pages; ++i) {
-			amdgpu_gart_map(adev, i << PAGE_SHIFT, 1, &dma_address,
-					flags, cpu_addr);
-			dma_address += PAGE_SIZE;
-		}
+		amdgpu_gart_map_vram_range(adev, pa, 0, num_pages, flags, cpu_addr);
 	}
 
 	dma_fence_put(amdgpu_job_submit(job));
@@ -284,12 +278,13 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
  * move and different for a BO to BO copy.
  *
  */
-int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
-			       const struct amdgpu_copy_mem *src,
-			       const struct amdgpu_copy_mem *dst,
-			       uint64_t size, bool tmz,
-			       struct dma_resv *resv,
-			       struct dma_fence **f)
+__attribute__((nonnull))
+static int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
+				      const struct amdgpu_copy_mem *src,
+				      const struct amdgpu_copy_mem *dst,
+				      uint64_t size, bool tmz,
+				      struct dma_resv *resv,
+				      struct dma_fence **f)
 {
 	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
 	struct amdgpu_res_cursor src_mm, dst_mm;
@@ -299,7 +294,8 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
 	struct amdgpu_bo *abo_src, *abo_dst;
 
 	if (!adev->mman.buffer_funcs_enabled) {
-		DRM_ERROR("Trying to move memory with ring turned off.\n");
+		dev_err(adev->dev,
+			"Trying to move memory with ring turned off.\n");
 		return -EINVAL;
 	}
 
@@ -362,9 +358,7 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
 	}
 error:
 	mutex_unlock(&adev->mman.gtt_window_lock);
-	if (f)
-		*f = dma_fence_get(fence);
-	dma_fence_put(fence);
+	*f = fence;
 	return r;
 }
 
@@ -405,7 +399,7 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo,
 		struct dma_fence *wipe_fence = NULL;
 
 		r = amdgpu_fill_buffer(abo, 0, NULL, &wipe_fence,
-				       false);
+				       false, AMDGPU_KERNEL_JOB_ID_MOVE_BLIT);
 		if (r) {
 			goto error;
 		} else if (wipe_fence) {
@@ -446,7 +440,8 @@ bool amdgpu_res_cpu_visible(struct amdgpu_device *adev,
 		return false;
 
 	if (res->mem_type == TTM_PL_SYSTEM || res->mem_type == TTM_PL_TT ||
-	    res->mem_type == AMDGPU_PL_PREEMPT || res->mem_type == AMDGPU_PL_DOORBELL)
+	    res->mem_type == AMDGPU_PL_PREEMPT || res->mem_type == AMDGPU_PL_DOORBELL ||
+	    res->mem_type == AMDGPU_PL_MMIO_REMAP)
 		return true;
 
 	if (res->mem_type != TTM_PL_VRAM)
@@ -537,10 +532,12 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict,
 	    old_mem->mem_type == AMDGPU_PL_GWS ||
 	    old_mem->mem_type == AMDGPU_PL_OA ||
 	    old_mem->mem_type == AMDGPU_PL_DOORBELL ||
+	    old_mem->mem_type == AMDGPU_PL_MMIO_REMAP ||
 	    new_mem->mem_type == AMDGPU_PL_GDS ||
 	    new_mem->mem_type == AMDGPU_PL_GWS ||
 	    new_mem->mem_type == AMDGPU_PL_OA ||
-	    new_mem->mem_type == AMDGPU_PL_DOORBELL) {
+	    new_mem->mem_type == AMDGPU_PL_DOORBELL ||
+	    new_mem->mem_type == AMDGPU_PL_MMIO_REMAP) {
 		/* Nothing to save here */
 		amdgpu_bo_move_notify(bo, evict, new_mem);
 		ttm_bo_move_null(bo, new_mem);
@@ -628,6 +625,12 @@ static int amdgpu_ttm_io_mem_reserve(struct ttm_device *bdev,
 		mem->bus.is_iomem = true;
 		mem->bus.caching = ttm_uncached;
 		break;
+	case AMDGPU_PL_MMIO_REMAP:
+		mem->bus.offset = mem->start << PAGE_SHIFT;
+		mem->bus.offset += adev->rmmio_remap.bus_addr;
+		mem->bus.is_iomem = true;
+		mem->bus.caching = ttm_uncached;
+		break;
 	default:
 		return -EINVAL;
 	}
@@ -645,6 +648,8 @@ static unsigned long amdgpu_ttm_io_mem_pfn(struct ttm_buffer_object *bo,
 
 	if (bo->resource->mem_type == AMDGPU_PL_DOORBELL)
 		return ((uint64_t)(adev->doorbell.base + cursor.start)) >> PAGE_SHIFT;
+	else if (bo->resource->mem_type == AMDGPU_PL_MMIO_REMAP)
+		return ((uint64_t)(adev->rmmio_remap.bus_addr + cursor.start)) >> PAGE_SHIFT;
 
 	return (adev->gmc.aper_base + cursor.start) >> PAGE_SHIFT;
 }
@@ -692,10 +697,11 @@ struct amdgpu_ttm_tt {
  * memory and start HMM tracking CPU page table update
  *
  * Calling function must call amdgpu_ttm_tt_userptr_range_done() once and only
- * once afterwards to stop HMM tracking
+ * once afterwards to stop HMM tracking. Its the caller responsibility to ensure
+ * that range is a valid memory and it is freed too.
  */
-int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages,
-				 struct hmm_range **range)
+int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo,
+				 struct amdgpu_hmm_range *range)
 {
 	struct ttm_tt *ttm = bo->tbo.ttm;
 	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
@@ -705,9 +711,6 @@ int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages,
 	bool readonly;
 	int r = 0;
 
-	/* Make sure get_user_pages_done() can cleanup gracefully */
-	*range = NULL;
-
 	mm = bo->notifier.mm;
 	if (unlikely(!mm)) {
 		DRM_DEBUG_DRIVER("BO is not registered?\n");
@@ -731,7 +734,7 @@ int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages,
 
 	readonly = amdgpu_ttm_tt_is_readonly(ttm);
 	r = amdgpu_hmm_range_get_pages(&bo->notifier, start, ttm->num_pages,
-				       readonly, NULL, pages, range);
+				       readonly, NULL, range);
 out_unlock:
 	mmap_read_unlock(mm);
 	if (r)
@@ -742,38 +745,6 @@ out_unlock:
 	return r;
 }
 
-/* amdgpu_ttm_tt_discard_user_pages - Discard range and pfn array allocations
- */
-void amdgpu_ttm_tt_discard_user_pages(struct ttm_tt *ttm,
-				      struct hmm_range *range)
-{
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-
-	if (gtt && gtt->userptr && range)
-		amdgpu_hmm_range_get_pages_done(range);
-}
-
-/*
- * amdgpu_ttm_tt_get_user_pages_done - stop HMM track the CPU page table change
- * Check if the pages backing this ttm range have been invalidated
- *
- * Returns: true if pages are still valid
- */
-bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm,
-				       struct hmm_range *range)
-{
-	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
-
-	if (!gtt || !gtt->userptr || !range)
-		return false;
-
-	DRM_DEBUG_DRIVER("user_pages_done 0x%llx pages 0x%x\n",
-		gtt->userptr, ttm->num_pages);
-
-	WARN_ONCE(!range->hmm_pfns, "No user pages to check\n");
-
-	return !amdgpu_hmm_range_get_pages_done(range);
-}
 #endif
 
 /*
@@ -783,12 +754,12 @@ bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm,
  * that backs user memory and will ultimately be mapped into the device
  * address space.
  */
-void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages)
+void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct amdgpu_hmm_range *range)
 {
 	unsigned long i;
 
 	for (i = 0; i < ttm->num_pages; ++i)
-		ttm->pages[i] = pages ? pages[i] : NULL;
+		ttm->pages[i] = range ? hmm_pfn_to_page(range->hmm_range.hmm_pfns[i]) : NULL;
 }
 
 /*
@@ -934,7 +905,7 @@ static int amdgpu_ttm_backend_bind(struct ttm_device *bdev,
 	if (gtt->userptr) {
 		r = amdgpu_ttm_tt_pin_userptr(bdev, ttm);
 		if (r) {
-			DRM_ERROR("failed to pin userptr\n");
+			dev_err(adev->dev, "failed to pin userptr\n");
 			return r;
 		}
 	} else if (ttm->page_flags & TTM_TT_FLAG_EXTERNAL) {
@@ -1060,7 +1031,7 @@ static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev,
 	/* if the pages have userptr pinning then clear that first */
 	if (gtt->userptr) {
 		amdgpu_ttm_tt_unpin_userptr(bdev, ttm);
-	} else if (ttm->sg && gtt->gobj->import_attach) {
+	} else if (ttm->sg && drm_gem_is_imported(gtt->gobj)) {
 		struct dma_buf_attachment *attach;
 
 		attach = gtt->gobj->import_attach;
@@ -1354,10 +1325,11 @@ uint64_t amdgpu_ttm_tt_pde_flags(struct ttm_tt *ttm, struct ttm_resource *mem)
 
 	if (mem && (mem->mem_type == TTM_PL_TT ||
 		    mem->mem_type == AMDGPU_PL_DOORBELL ||
-		    mem->mem_type == AMDGPU_PL_PREEMPT)) {
+		    mem->mem_type == AMDGPU_PL_PREEMPT ||
+		    mem->mem_type == AMDGPU_PL_MMIO_REMAP)) {
 		flags |= AMDGPU_PTE_SYSTEM;
 
-		if (ttm->caching == ttm_cached)
+		if (ttm && ttm->caching == ttm_cached)
 			flags |= AMDGPU_PTE_SNOOPED;
 	}
 
@@ -1509,10 +1481,12 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
 	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.high_pr,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     num_dw * 4, AMDGPU_IB_POOL_DELAYED,
-				     &job);
+				     &job,
+				     AMDGPU_KERNEL_JOB_ID_TTM_ACCESS_MEMORY_SDMA);
 	if (r)
 		goto out;
 
+	mutex_lock(&adev->mman.gtt_window_lock);
 	amdgpu_res_first(abo->tbo.resource, offset, len, &src_mm);
 	src_addr = amdgpu_ttm_domain_start(adev, bo->resource->mem_type) +
 		src_mm.start;
@@ -1527,6 +1501,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
 	WARN_ON(job->ibs[0].length_dw > num_dw);
 
 	fence = amdgpu_job_submit(job);
+	mutex_unlock(&adev->mman.gtt_window_lock);
 
 	if (!dma_fence_wait_timeout(fence, false, adev->sdma_timeout))
 		r = -ETIMEDOUT;
@@ -1781,25 +1756,21 @@ static int amdgpu_ttm_reserve_tmr(struct amdgpu_device *adev)
 						 &ctx->c2p_bo,
 						 NULL);
 		if (ret) {
-			DRM_ERROR("alloc c2p_bo failed(%d)!\n", ret);
+			dev_err(adev->dev, "alloc c2p_bo failed(%d)!\n", ret);
 			amdgpu_ttm_training_reserve_vram_fini(adev);
 			return ret;
 		}
 		ctx->init = PSP_MEM_TRAIN_RESERVE_SUCCESS;
 	}
 
-	if (!adev->gmc.is_app_apu) {
-		ret = amdgpu_bo_create_kernel_at(
-			adev, adev->gmc.real_vram_size - reserve_size,
-			reserve_size, &adev->mman.fw_reserved_memory, NULL);
-		if (ret) {
-			DRM_ERROR("alloc tmr failed(%d)!\n", ret);
-			amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory,
-					      NULL, NULL);
-			return ret;
-		}
-	} else {
-		DRM_DEBUG_DRIVER("backdoor fw loading path for PSP TMR, no reservation needed\n");
+	ret = amdgpu_bo_create_kernel_at(
+		adev, adev->gmc.real_vram_size - reserve_size, reserve_size,
+		&adev->mman.fw_reserved_memory, NULL);
+	if (ret) {
+		dev_err(adev->dev, "alloc tmr failed(%d)!\n", ret);
+		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL,
+				      NULL);
+		return ret;
 	}
 
 	return 0;
@@ -1821,7 +1792,7 @@ static int amdgpu_ttm_pools_init(struct amdgpu_device *adev)
 	for (i = 0; i < adev->gmc.num_mem_partitions; i++) {
 		ttm_pool_init(&adev->mman.ttm_pools[i], adev->dev,
 			      adev->gmc.mem_partitions[i].numa.node,
-			      false, false);
+			      TTM_ALLOCATION_POOL_BENEFICIAL_ORDER(get_order(SZ_2M)));
 	}
 	return 0;
 }
@@ -1840,6 +1811,59 @@ static void amdgpu_ttm_pools_fini(struct amdgpu_device *adev)
 	adev->mman.ttm_pools = NULL;
 }
 
+/**
+ * amdgpu_ttm_mmio_remap_bo_init - Allocate the singleton 4K MMIO_REMAP BO
+ * @adev: amdgpu device
+ *
+ * Allocates a one-page (4K) GEM BO in AMDGPU_GEM_DOMAIN_MMIO_REMAP when the
+ * hardware exposes a remap base (adev->rmmio_remap.bus_addr) and the host
+ * PAGE_SIZE is <= AMDGPU_GPU_PAGE_SIZE (4K). The BO is created as a regular
+ * GEM object (amdgpu_bo_create).
+ *
+ * Return:
+ *  * 0 on success or intentional skip (feature not present/unsupported)
+ *  * negative errno on allocation failure
+ */
+static int amdgpu_ttm_mmio_remap_bo_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_bo_param bp;
+	int r;
+
+	/* Skip if HW doesn't expose remap, or if PAGE_SIZE > AMDGPU_GPU_PAGE_SIZE (4K). */
+	if (!adev->rmmio_remap.bus_addr || PAGE_SIZE > AMDGPU_GPU_PAGE_SIZE)
+		return 0;
+
+	memset(&bp, 0, sizeof(bp));
+
+	/* Create exactly one GEM BO in the MMIO_REMAP domain. */
+	bp.type        = ttm_bo_type_device;          /* userspace-mappable GEM */
+	bp.size        = AMDGPU_GPU_PAGE_SIZE;        /* 4K */
+	bp.byte_align  = AMDGPU_GPU_PAGE_SIZE;
+	bp.domain      = AMDGPU_GEM_DOMAIN_MMIO_REMAP;
+	bp.flags       = 0;
+	bp.resv        = NULL;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
+
+	r = amdgpu_bo_create(adev, &bp, &adev->rmmio_remap.bo);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * amdgpu_ttm_mmio_remap_bo_fini - Free the singleton MMIO_REMAP BO
+ * @adev: amdgpu device
+ *
+ * Frees the kernel-owned MMIO_REMAP BO if it was allocated by
+ * amdgpu_ttm_mmio_remap_bo_init().
+ */
+static void amdgpu_ttm_mmio_remap_bo_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_unref(&adev->rmmio_remap.bo);
+	adev->rmmio_remap.bo = NULL;
+}
+
 /*
  * amdgpu_ttm_init - Init the memory management (ttm) as well as various
  * gtt/vram related fields.
@@ -1861,25 +1885,31 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
 	r = ttm_device_init(&adev->mman.bdev, &amdgpu_bo_driver, adev->dev,
 			       adev_to_drm(adev)->anon_inode->i_mapping,
 			       adev_to_drm(adev)->vma_offset_manager,
-			       adev->need_swiotlb,
-			       dma_addressing_limited(adev->dev));
+			       (adev->need_swiotlb ?
+				TTM_ALLOCATION_POOL_USE_DMA_ALLOC : 0) |
+			       (dma_addressing_limited(adev->dev) ?
+				TTM_ALLOCATION_POOL_USE_DMA32 : 0) |
+			       TTM_ALLOCATION_POOL_BENEFICIAL_ORDER(get_order(SZ_2M)));
 	if (r) {
-		DRM_ERROR("failed initializing buffer object driver(%d).\n", r);
+		dev_err(adev->dev,
+			"failed initializing buffer object driver(%d).\n", r);
 		return r;
 	}
 
 	r = amdgpu_ttm_pools_init(adev);
 	if (r) {
-		DRM_ERROR("failed to init ttm pools(%d).\n", r);
+		dev_err(adev->dev, "failed to init ttm pools(%d).\n", r);
 		return r;
 	}
 	adev->mman.initialized = true;
 
-	/* Initialize VRAM pool with all of VRAM divided into pages */
-	r = amdgpu_vram_mgr_init(adev);
-	if (r) {
-		DRM_ERROR("Failed initializing VRAM heap.\n");
-		return r;
+	if (!adev->gmc.is_app_apu) {
+		/* Initialize VRAM pool with all of VRAM divided into pages */
+		r = amdgpu_vram_mgr_init(adev);
+		if (r) {
+			dev_err(adev->dev, "Failed initializing VRAM heap.\n");
+			return r;
+		}
 	}
 
 	/* Change the size here instead of the init above so only lpfn is affected */
@@ -1908,19 +1938,19 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
 		return r;
 
 	/*
-	 *The reserved vram for driver must be pinned to the specified
-	 *place on the VRAM, so reserve it early.
+	 * The reserved VRAM for the driver must be pinned to a specific
+	 * location in VRAM, so reserve it early.
 	 */
 	r = amdgpu_ttm_drv_reserve_vram_init(adev);
 	if (r)
 		return r;
 
 	/*
-	 * only NAVI10 and onwards ASIC support for IP discovery.
-	 * If IP discovery enabled, a block of memory should be
-	 * reserved for IP discovey.
+	 * only NAVI10 and later ASICs support IP discovery.
+	 * If IP discovery is enabled, a block of memory should be
+	 * reserved for it.
 	 */
-	if (adev->mman.discovery_bin) {
+	if (adev->discovery.reserve_tmr) {
 		r = amdgpu_ttm_reserve_tmr(adev);
 		if (r)
 			return r;
@@ -1958,63 +1988,89 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
 		DRM_DEBUG_DRIVER("Skipped stolen memory reservation\n");
 	}
 
-	DRM_INFO("amdgpu: %uM of VRAM memory ready\n",
+	dev_info(adev->dev, "amdgpu: %uM of VRAM memory ready\n",
 		 (unsigned int)(adev->gmc.real_vram_size / (1024 * 1024)));
 
 	/* Compute GTT size, either based on TTM limit
 	 * or whatever the user passed on module init.
 	 */
-	if (amdgpu_gtt_size == -1)
-		gtt_size = ttm_tt_pages_limit() << PAGE_SHIFT;
-	else
-		gtt_size = (uint64_t)amdgpu_gtt_size << 20;
+	gtt_size = ttm_tt_pages_limit() << PAGE_SHIFT;
+	if (amdgpu_gtt_size != -1) {
+		uint64_t configured_size = (uint64_t)amdgpu_gtt_size << 20;
+
+		drm_warn(&adev->ddev,
+			"Configuring gttsize via module parameter is deprecated, please use ttm.pages_limit\n");
+		if (gtt_size != configured_size)
+			drm_warn(&adev->ddev,
+				"GTT size has been set as %llu but TTM size has been set as %llu, this is unusual\n",
+				configured_size, gtt_size);
+
+		gtt_size = configured_size;
+	}
 
 	/* Initialize GTT memory pool */
 	r = amdgpu_gtt_mgr_init(adev, gtt_size);
 	if (r) {
-		DRM_ERROR("Failed initializing GTT heap.\n");
+		dev_err(adev->dev, "Failed initializing GTT heap.\n");
 		return r;
 	}
-	DRM_INFO("amdgpu: %uM of GTT memory ready.\n",
+	dev_info(adev->dev, "amdgpu: %uM of GTT memory ready.\n",
 		 (unsigned int)(gtt_size / (1024 * 1024)));
 
+	if (adev->flags & AMD_IS_APU) {
+		if (adev->gmc.real_vram_size < gtt_size)
+			adev->apu_prefer_gtt = true;
+	}
+
 	/* Initialize doorbell pool on PCI BAR */
 	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_DOORBELL, adev->doorbell.size / PAGE_SIZE);
 	if (r) {
-		DRM_ERROR("Failed initializing doorbell heap.\n");
+		dev_err(adev->dev, "Failed initializing doorbell heap.\n");
 		return r;
 	}
 
 	/* Create a boorbell page for kernel usages */
 	r = amdgpu_doorbell_create_kernel_doorbells(adev);
 	if (r) {
-		DRM_ERROR("Failed to initialize kernel doorbells.\n");
+		dev_err(adev->dev, "Failed to initialize kernel doorbells.\n");
+		return r;
+	}
+
+	/* Initialize MMIO-remap pool (single page 4K) */
+	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_MMIO_REMAP, 1);
+	if (r) {
+		dev_err(adev->dev, "Failed initializing MMIO-remap heap.\n");
 		return r;
 	}
 
+	/* Allocate the singleton MMIO_REMAP BO (4K) if supported */
+	r = amdgpu_ttm_mmio_remap_bo_init(adev);
+	if (r)
+		return r;
+
 	/* Initialize preemptible memory pool */
 	r = amdgpu_preempt_mgr_init(adev);
 	if (r) {
-		DRM_ERROR("Failed initializing PREEMPT heap.\n");
+		dev_err(adev->dev, "Failed initializing PREEMPT heap.\n");
 		return r;
 	}
 
 	/* Initialize various on-chip memory pools */
 	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_GDS, adev->gds.gds_size);
 	if (r) {
-		DRM_ERROR("Failed initializing GDS heap.\n");
+		dev_err(adev->dev, "Failed initializing GDS heap.\n");
 		return r;
 	}
 
 	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_GWS, adev->gds.gws_size);
 	if (r) {
-		DRM_ERROR("Failed initializing gws heap.\n");
+		dev_err(adev->dev, "Failed initializing gws heap.\n");
 		return r;
 	}
 
 	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_OA, adev->gds.oa_size);
 	if (r) {
-		DRM_ERROR("Failed initializing oa heap.\n");
+		dev_err(adev->dev, "Failed initializing oa heap.\n");
 		return r;
 	}
 	if (amdgpu_bo_create_kernel(adev, PAGE_SIZE, PAGE_SIZE,
@@ -2046,12 +2102,16 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev)
 		/* return the FW reserved memory back to VRAM */
 		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL,
 				      NULL);
+		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory_extend, NULL,
+				      NULL);
 		if (adev->mman.stolen_reserved_size)
 			amdgpu_bo_free_kernel(&adev->mman.stolen_reserved_memory,
 					      NULL, NULL);
 	}
 	amdgpu_bo_free_kernel(&adev->mman.sdma_access_bo, NULL,
 					&adev->mman.sdma_access_ptr);
+
+	amdgpu_ttm_mmio_remap_bo_fini(adev);
 	amdgpu_ttm_fw_reserve_vram_fini(adev);
 	amdgpu_ttm_drv_reserve_vram_fini(adev);
 
@@ -2064,16 +2124,20 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev)
 		drm_dev_exit(idx);
 	}
 
-	amdgpu_vram_mgr_fini(adev);
+	if (!adev->gmc.is_app_apu)
+		amdgpu_vram_mgr_fini(adev);
 	amdgpu_gtt_mgr_fini(adev);
 	amdgpu_preempt_mgr_fini(adev);
+	amdgpu_doorbell_fini(adev);
+
 	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_GDS);
 	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_GWS);
 	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_OA);
 	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_DOORBELL);
+	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_MMIO_REMAP);
 	ttm_device_fini(&adev->mman.bdev);
 	adev->mman.initialized = false;
-	DRM_INFO("amdgpu: ttm finalized\n");
+	dev_info(adev->dev, "amdgpu: ttm finalized\n");
 }
 
 /**
@@ -2105,8 +2169,9 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
 					  DRM_SCHED_PRIORITY_KERNEL, &sched,
 					  1, NULL);
 		if (r) {
-			DRM_ERROR("Failed setting up TTM BO move entity (%d)\n",
-				  r);
+			dev_err(adev->dev,
+				"Failed setting up TTM BO move entity (%d)\n",
+				r);
 			return;
 		}
 
@@ -2114,15 +2179,18 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
 					  DRM_SCHED_PRIORITY_NORMAL, &sched,
 					  1, NULL);
 		if (r) {
-			DRM_ERROR("Failed setting up TTM BO move entity (%d)\n",
-				  r);
+			dev_err(adev->dev,
+				"Failed setting up TTM BO move entity (%d)\n",
+				r);
 			goto error_free_entity;
 		}
 	} else {
 		drm_sched_entity_destroy(&adev->mman.high_pr);
 		drm_sched_entity_destroy(&adev->mman.low_pr);
-		dma_fence_put(man->move);
-		man->move = NULL;
+		/* Drop all the old fences since re-creating the scheduler entities
+		 * will allocate new contexts.
+		 */
+		ttm_resource_manager_cleanup(man);
 	}
 
 	/* this just adjusts TTM size idea, which sets lpfn to the correct value */
@@ -2145,7 +2213,7 @@ static int amdgpu_ttm_prepare_job(struct amdgpu_device *adev,
 				  struct dma_resv *resv,
 				  bool vm_needs_flush,
 				  struct amdgpu_job **job,
-				  bool delayed)
+				  bool delayed, u64 k_job_id)
 {
 	enum amdgpu_ib_pool_type pool = direct_submit ?
 		AMDGPU_IB_POOL_DIRECT :
@@ -2155,7 +2223,7 @@ static int amdgpu_ttm_prepare_job(struct amdgpu_device *adev,
 						    &adev->mman.high_pr;
 	r = amdgpu_job_alloc_with_ib(adev, entity,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
-				     num_dw * 4, pool, job);
+				     num_dw * 4, pool, job, k_job_id);
 	if (r)
 		return r;
 
@@ -2186,7 +2254,8 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 	int r;
 
 	if (!direct_submit && !ring->sched.ready) {
-		DRM_ERROR("Trying to move memory with ring turned off.\n");
+		dev_err(adev->dev,
+			"Trying to move memory with ring turned off.\n");
 		return -EINVAL;
 	}
 
@@ -2194,7 +2263,8 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 	num_loops = DIV_ROUND_UP(byte_count, max_bytes);
 	num_dw = ALIGN(num_loops * adev->mman.buffer_funcs->copy_num_dw, 8);
 	r = amdgpu_ttm_prepare_job(adev, direct_submit, num_dw,
-				   resv, vm_needs_flush, &job, false);
+				   resv, vm_needs_flush, &job, false,
+				   AMDGPU_KERNEL_JOB_ID_TTM_COPY_BUFFER);
 	if (r)
 		return r;
 
@@ -2221,7 +2291,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 
 error_free:
 	amdgpu_job_free(job);
-	DRM_ERROR("Error scheduling IBs (%d)\n", r);
+	dev_err(adev->dev, "Error scheduling IBs (%d)\n", r);
 	return r;
 }
 
@@ -2229,7 +2299,8 @@ static int amdgpu_ttm_fill_mem(struct amdgpu_ring *ring, uint32_t src_data,
 			       uint64_t dst_addr, uint32_t byte_count,
 			       struct dma_resv *resv,
 			       struct dma_fence **fence,
-			       bool vm_needs_flush, bool delayed)
+			       bool vm_needs_flush, bool delayed,
+			       u64 k_job_id)
 {
 	struct amdgpu_device *adev = ring->adev;
 	unsigned int num_loops, num_dw;
@@ -2242,7 +2313,7 @@ static int amdgpu_ttm_fill_mem(struct amdgpu_ring *ring, uint32_t src_data,
 	num_loops = DIV_ROUND_UP_ULL(byte_count, max_bytes);
 	num_dw = ALIGN(num_loops * adev->mman.buffer_funcs->fill_num_dw, 8);
 	r = amdgpu_ttm_prepare_job(adev, false, num_dw, resv, vm_needs_flush,
-				   &job, delayed);
+				   &job, delayed, k_job_id);
 	if (r)
 		return r;
 
@@ -2281,7 +2352,7 @@ int amdgpu_ttm_clear_buffer(struct amdgpu_bo *bo,
 	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
 	struct amdgpu_res_cursor cursor;
 	u64 addr;
-	int r;
+	int r = 0;
 
 	if (!adev->mman.buffer_funcs_enabled)
 		return -EINVAL;
@@ -2312,7 +2383,8 @@ int amdgpu_ttm_clear_buffer(struct amdgpu_bo *bo,
 			goto err;
 
 		r = amdgpu_ttm_fill_mem(ring, 0, addr, size, resv,
-					&next, true, true);
+					&next, true, true,
+					AMDGPU_KERNEL_JOB_ID_TTM_CLEAR_BUFFER);
 		if (r)
 			goto err;
 
@@ -2331,7 +2403,8 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
 			uint32_t src_data,
 			struct dma_resv *resv,
 			struct dma_fence **f,
-			bool delayed)
+			bool delayed,
+			u64 k_job_id)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
 	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
@@ -2340,7 +2413,8 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
 	int r;
 
 	if (!adev->mman.buffer_funcs_enabled) {
-		DRM_ERROR("Trying to clear memory with ring turned off.\n");
+		dev_err(adev->dev,
+			"Trying to clear memory with ring turned off.\n");
 		return -EINVAL;
 	}
 
@@ -2360,7 +2434,7 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
 			goto error;
 
 		r = amdgpu_ttm_fill_mem(ring, src_data, to, cur_size, resv,
-					&next, true, delayed);
+					&next, true, delayed, k_job_id);
 		if (r)
 			goto error;
 
@@ -2400,7 +2474,7 @@ int amdgpu_ttm_evict_resources(struct amdgpu_device *adev, int mem_type)
 		man = ttm_manager_type(&adev->mman.bdev, mem_type);
 		break;
 	default:
-		DRM_ERROR("Trying to evict invalid memory type\n");
+		dev_err(adev->dev, "Trying to evict invalid memory type\n");
 		return -EINVAL;
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
index 208b7d1d8a27..577ee04ce0bf 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
@@ -28,13 +28,15 @@
 #include <drm/gpu_scheduler.h>
 #include <drm/ttm/ttm_placement.h>
 #include "amdgpu_vram_mgr.h"
+#include "amdgpu_hmm.h"
 
 #define AMDGPU_PL_GDS		(TTM_PL_PRIV + 0)
 #define AMDGPU_PL_GWS		(TTM_PL_PRIV + 1)
 #define AMDGPU_PL_OA		(TTM_PL_PRIV + 2)
 #define AMDGPU_PL_PREEMPT	(TTM_PL_PRIV + 3)
 #define AMDGPU_PL_DOORBELL	(TTM_PL_PRIV + 4)
-#define __AMDGPU_PL_NUM	(TTM_PL_PRIV + 5)
+#define AMDGPU_PL_MMIO_REMAP	(TTM_PL_PRIV + 5)
+#define __AMDGPU_PL_NUM	(TTM_PL_PRIV + 6)
 
 #define AMDGPU_GTT_MAX_TRANSFER_SIZE	512
 #define AMDGPU_GTT_NUM_TRANSFER_WINDOWS	2
@@ -81,11 +83,9 @@ struct amdgpu_mman {
 	uint64_t		stolen_reserved_offset;
 	uint64_t		stolen_reserved_size;
 
-	/* discovery */
-	uint8_t				*discovery_bin;
-	uint32_t			discovery_tmr_size;
 	/* fw reserved memory */
 	struct amdgpu_bo		*fw_reserved_memory;
+	struct amdgpu_bo		*fw_reserved_memory_extend;
 
 	/* firmware VRAM reservation */
 	u64		fw_vram_usage_start_offset;
@@ -154,6 +154,7 @@ int amdgpu_vram_mgr_reserve_range(struct amdgpu_vram_mgr *mgr,
 				  uint64_t start, uint64_t size);
 int amdgpu_vram_mgr_query_page_status(struct amdgpu_vram_mgr *mgr,
 				      uint64_t start);
+void amdgpu_vram_mgr_clear_reset_blocks(struct amdgpu_device *adev);
 
 bool amdgpu_res_cpu_visible(struct amdgpu_device *adev,
 			    struct ttm_resource *res);
@@ -167,12 +168,6 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 		       struct dma_resv *resv,
 		       struct dma_fence **fence, bool direct_submit,
 		       bool vm_needs_flush, uint32_t copy_flags);
-int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
-			       const struct amdgpu_copy_mem *src,
-			       const struct amdgpu_copy_mem *dst,
-			       uint64_t size, bool tmz,
-			       struct dma_resv *resv,
-			       struct dma_fence **f);
 int amdgpu_ttm_clear_buffer(struct amdgpu_bo *bo,
 			    struct dma_resv *resv,
 			    struct dma_fence **fence);
@@ -180,38 +175,25 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
 			uint32_t src_data,
 			struct dma_resv *resv,
 			struct dma_fence **fence,
-			bool delayed);
+			bool delayed,
+			u64 k_job_id);
 
 int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo);
 void amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo);
 uint64_t amdgpu_ttm_domain_start(struct amdgpu_device *adev, uint32_t type);
 
 #if IS_ENABLED(CONFIG_DRM_AMDGPU_USERPTR)
-int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages,
-				 struct hmm_range **range);
-void amdgpu_ttm_tt_discard_user_pages(struct ttm_tt *ttm,
-				      struct hmm_range *range);
-bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm,
-				       struct hmm_range *range);
+int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo,
+				 struct amdgpu_hmm_range *range);
 #else
 static inline int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo,
-					       struct page **pages,
-					       struct hmm_range **range)
+					       struct amdgpu_hmm_range *range)
 {
 	return -EPERM;
 }
-static inline void amdgpu_ttm_tt_discard_user_pages(struct ttm_tt *ttm,
-						    struct hmm_range *range)
-{
-}
-static inline bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm,
-						     struct hmm_range *range)
-{
-	return false;
-}
 #endif
 
-void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages);
+void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct amdgpu_hmm_range *range);
 int amdgpu_ttm_tt_get_userptr(const struct ttm_buffer_object *tbo,
 			      uint64_t *user_addr);
 int amdgpu_ttm_tt_set_userptr(struct ttm_buffer_object *bo,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
index cf700824b960..e96f24e9ad57 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
@@ -30,6 +30,11 @@
 
 #define AMDGPU_UCODE_NAME_MAX		(128)
 
+static const struct kicker_device kicker_device_list[] = {
+	{0x744B, 0x00},
+	{0x7551, 0xC8}
+};
+
 static void amdgpu_ucode_print_common_hdr(const struct common_firmware_header *hdr)
 {
 	DRM_DEBUG("size_bytes: %u\n", le32_to_cpu(hdr->size_bytes));
@@ -765,8 +770,10 @@ FW_VERSION_ATTR(sdma_fw_version, 0444, sdma.instance[0].fw_version);
 FW_VERSION_ATTR(sdma2_fw_version, 0444, sdma.instance[1].fw_version);
 FW_VERSION_ATTR(vcn_fw_version, 0444, vcn.fw_version);
 FW_VERSION_ATTR(dmcu_fw_version, 0444, dm.dmcu_fw_version);
+FW_VERSION_ATTR(dmcub_fw_version, 0444, dm.dmcub_fw_version);
 FW_VERSION_ATTR(mes_fw_version, 0444, mes.sched_version & AMDGPU_MES_VERSION_MASK);
 FW_VERSION_ATTR(mes_kiq_fw_version, 0444, mes.kiq_version & AMDGPU_MES_VERSION_MASK);
+FW_VERSION_ATTR(pldm_fw_version, 0444, firmware.pldm_version);
 
 static struct attribute *fw_attrs[] = {
 	&dev_attr_vce_fw_version.attr, &dev_attr_uvd_fw_version.attr,
@@ -779,8 +786,9 @@ static struct attribute *fw_attrs[] = {
 	&dev_attr_ta_ras_fw_version.attr, &dev_attr_ta_xgmi_fw_version.attr,
 	&dev_attr_smc_fw_version.attr, &dev_attr_sdma_fw_version.attr,
 	&dev_attr_sdma2_fw_version.attr, &dev_attr_vcn_fw_version.attr,
-	&dev_attr_dmcu_fw_version.attr, &dev_attr_imu_fw_version.attr,
-	&dev_attr_mes_fw_version.attr, &dev_attr_mes_kiq_fw_version.attr,
+	&dev_attr_dmcu_fw_version.attr, &dev_attr_dmcub_fw_version.attr,
+	&dev_attr_imu_fw_version.attr, &dev_attr_mes_fw_version.attr,
+	&dev_attr_mes_kiq_fw_version.attr, &dev_attr_pldm_fw_version.attr,
 	NULL
 };
 
@@ -1152,6 +1160,9 @@ int amdgpu_ucode_init_bo(struct amdgpu_device *adev)
 		adev->firmware.max_ucodes = AMDGPU_UCODE_ID_MAXIMUM;
 	}
 
+	if (amdgpu_virt_xgmi_migrate_enabled(adev) && adev->firmware.fw_buf)
+		adev->firmware.fw_buf_mc = amdgpu_bo_fb_aper_addr(adev->firmware.fw_buf);
+
 	for (i = 0; i < adev->firmware.max_ucodes; i++) {
 		ucode = &adev->firmware.ucode[i];
 		if (ucode->fw) {
@@ -1216,6 +1227,7 @@ static const char *amdgpu_ucode_legacy_naming(struct amdgpu_device *adev, int bl
 		case IP_VERSION(11, 0, 13):
 			return "beige_goby";
 		case IP_VERSION(11, 5, 0):
+		case IP_VERSION(11, 5, 2):
 			return "vangogh";
 		case IP_VERSION(12, 0, 1):
 			return "green_sardine";
@@ -1383,6 +1395,19 @@ static const char *amdgpu_ucode_legacy_naming(struct amdgpu_device *adev, int bl
 	return NULL;
 }
 
+bool amdgpu_is_kicker_fw(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(kicker_device_list); i++) {
+		if (adev->pdev->device == kicker_device_list[i].device &&
+		    adev->pdev->revision == kicker_device_list[i].revision)
+			return true;
+	}
+
+	return false;
+}
+
 void amdgpu_ucode_ip_version_decode(struct amdgpu_device *adev, int block_type, char *ucode_prefix, int len)
 {
 	int maj, min, rev;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
index 4eedd92f000b..6349aad6da35 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
@@ -25,6 +25,8 @@
 
 #include "amdgpu_socbb.h"
 
+#define RS64_FW_UC_START_ADDR_LO 0x3000
+
 struct common_firmware_header {
 	uint32_t size_bytes; /* size of the entire header+image(s) in bytes */
 	uint32_t header_size_bytes; /* size of just the header in bytes */
@@ -600,6 +602,12 @@ struct amdgpu_firmware {
 
 	void *fw_buf_ptr;
 	uint64_t fw_buf_mc;
+	uint32_t pldm_version;
+};
+
+struct kicker_device{
+	unsigned short device;
+	u8 revision;
 };
 
 void amdgpu_ucode_print_mc_hdr(const struct common_firmware_header *hdr);
@@ -629,5 +637,6 @@ amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type);
 const char *amdgpu_ucode_name(enum AMDGPU_UCODE_ID ucode_id);
 
 void amdgpu_ucode_ip_version_decode(struct amdgpu_device *adev, int block_type, char *ucode_prefix, int len);
+bool amdgpu_is_kicker_fw(struct amdgpu_device *adev);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
index eafe20d8fe0b..3f0b0e9af4f3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
@@ -24,6 +24,7 @@
 #include <linux/sort.h>
 #include "amdgpu.h"
 #include "umc_v6_7.h"
+#include "amdgpu_ras_mgr.h"
 #define MAX_UMC_POISON_POLLING_TIME_SYNC   20  //ms
 
 #define MAX_UMC_HASH_STRING_SIZE  256
@@ -96,67 +97,96 @@ void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev,
 {
 	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct amdgpu_ras_eeprom_control *control = &con->eeprom_control;
 	unsigned int error_query_mode;
 	int ret = 0;
 	unsigned long err_count;
 
 	amdgpu_ras_get_error_query_mode(adev, &error_query_mode);
 
+	err_data->err_addr =
+		kcalloc(adev->umc.max_ras_err_cnt_per_query,
+			sizeof(struct eeprom_table_record), GFP_KERNEL);
+
+	/* still call query_ras_error_address to clear error status
+	 * even NOMEM error is encountered
+	 */
+	if (!err_data->err_addr)
+		dev_warn(adev->dev,
+			"Failed to alloc memory for umc error address record!\n");
+	else
+		err_data->err_addr_len = adev->umc.max_ras_err_cnt_per_query;
+
 	mutex_lock(&con->page_retirement_lock);
-	ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(con->umc_ecc));
-	if (ret == -EOPNOTSUPP &&
-	    error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) {
-		if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
-		    adev->umc.ras->ras_block.hw_ops->query_ras_error_count)
-		    adev->umc.ras->ras_block.hw_ops->query_ras_error_count(adev, ras_error_status);
-
-		if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
-		    adev->umc.ras->ras_block.hw_ops->query_ras_error_address &&
-		    adev->umc.max_ras_err_cnt_per_query) {
-			err_data->err_addr =
-				kcalloc(adev->umc.max_ras_err_cnt_per_query,
-					sizeof(struct eeprom_table_record), GFP_KERNEL);
-
-			/* still call query_ras_error_address to clear error status
-			 * even NOMEM error is encountered
-			 */
-			if(!err_data->err_addr)
-				dev_warn(adev->dev, "Failed to alloc memory for "
-						"umc error address record!\n");
-			else
-				err_data->err_addr_len = adev->umc.max_ras_err_cnt_per_query;
-
-			/* umc query_ras_error_address is also responsible for clearing
-			 * error status
-			 */
-			adev->umc.ras->ras_block.hw_ops->query_ras_error_address(adev, ras_error_status);
+	if (!amdgpu_ras_smu_eeprom_supported(adev)) {
+		ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(con->umc_ecc));
+		if (ret == -EOPNOTSUPP &&
+		    error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) {
+			if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
+			    adev->umc.ras->ras_block.hw_ops->query_ras_error_count)
+				adev->umc.ras->ras_block.hw_ops->query_ras_error_count(adev,
+								ras_error_status);
+
+			if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
+			    adev->umc.ras->ras_block.hw_ops->query_ras_error_address &&
+			    adev->umc.max_ras_err_cnt_per_query) {
+				err_data->err_addr =
+					kcalloc(adev->umc.max_ras_err_cnt_per_query,
+						sizeof(struct eeprom_table_record), GFP_KERNEL);
+
+				/* still call query_ras_error_address to clear error status
+				 * even NOMEM error is encountered
+				 */
+				if (!err_data->err_addr)
+					dev_warn(adev->dev,
+						"Failed to alloc memory for umc error address record!\n");
+				else
+					err_data->err_addr_len =
+						adev->umc.max_ras_err_cnt_per_query;
+
+				/* umc query_ras_error_address is also responsible for clearing
+				 * error status
+				 */
+				adev->umc.ras->ras_block.hw_ops->query_ras_error_address(adev,
+								ras_error_status);
+			}
+		} else if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY ||
+		    (!ret && error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY)) {
+			if (adev->umc.ras &&
+			    adev->umc.ras->ecc_info_query_ras_error_count)
+				adev->umc.ras->ecc_info_query_ras_error_count(adev,
+								ras_error_status);
+
+			if (adev->umc.ras &&
+			    adev->umc.ras->ecc_info_query_ras_error_address &&
+			    adev->umc.max_ras_err_cnt_per_query) {
+				err_data->err_addr =
+					kcalloc(adev->umc.max_ras_err_cnt_per_query,
+						sizeof(struct eeprom_table_record), GFP_KERNEL);
+
+				/* still call query_ras_error_address to clear error status
+				 * even NOMEM error is encountered
+				 */
+				if (!err_data->err_addr)
+					dev_warn(adev->dev,
+						"Failed to alloc memory for umc error address record!\n");
+				else
+					err_data->err_addr_len =
+						adev->umc.max_ras_err_cnt_per_query;
+
+				/* umc query_ras_error_address is also responsible for clearing
+				 * error status
+				 */
+				adev->umc.ras->ecc_info_query_ras_error_address(adev,
+								ras_error_status);
+			}
 		}
-	} else if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY ||
-	    (!ret && error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY)) {
-		if (adev->umc.ras &&
-		    adev->umc.ras->ecc_info_query_ras_error_count)
-		    adev->umc.ras->ecc_info_query_ras_error_count(adev, ras_error_status);
-
-		if (adev->umc.ras &&
-		    adev->umc.ras->ecc_info_query_ras_error_address &&
-		    adev->umc.max_ras_err_cnt_per_query) {
-			err_data->err_addr =
-				kcalloc(adev->umc.max_ras_err_cnt_per_query,
-					sizeof(struct eeprom_table_record), GFP_KERNEL);
-
-			/* still call query_ras_error_address to clear error status
-			 * even NOMEM error is encountered
-			 */
-			if(!err_data->err_addr)
-				dev_warn(adev->dev, "Failed to alloc memory for "
-						"umc error address record!\n");
-			else
-				err_data->err_addr_len = adev->umc.max_ras_err_cnt_per_query;
-
-			/* umc query_ras_error_address is also responsible for clearing
-			 * error status
-			 */
-			adev->umc.ras->ecc_info_query_ras_error_address(adev, ras_error_status);
+	} else {
+		if (!amdgpu_ras_eeprom_update_record_num(control)) {
+			err_data->err_addr_cnt = err_data->de_count =
+				control->ras_num_recs -	control->ras_num_recs_old;
+			amdgpu_ras_eeprom_read_idx(control, err_data->err_addr,
+				control->ras_num_recs_old, err_data->de_count);
 		}
 	}
 
@@ -166,7 +196,7 @@ void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev,
 		if ((amdgpu_bad_page_threshold != 0) &&
 			err_data->err_addr_cnt) {
 			amdgpu_ras_add_bad_pages(adev, err_data->err_addr,
-						err_data->err_addr_cnt, false);
+				err_data->err_addr_cnt, amdgpu_ras_smu_eeprom_supported(adev));
 			amdgpu_ras_save_bad_pages(adev, &err_count);
 
 			amdgpu_dpm_send_hbm_bad_pages_num(adev,
@@ -244,6 +274,15 @@ int amdgpu_umc_pasid_poison_handler(struct amdgpu_device *adev,
 			}
 
 			amdgpu_ras_error_data_fini(&err_data);
+		} else if (amdgpu_uniras_enabled(adev)) {
+			struct ras_ih_info ih_info = {0};
+
+			ih_info.block = block;
+			ih_info.pasid = pasid;
+			ih_info.reset = reset;
+			ih_info.pasid_fn = pasid_fn;
+			ih_info.data = data;
+			amdgpu_ras_mgr_handle_consumer_interrupt(adev, &ih_info);
 		} else {
 			struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 			int ret;
@@ -252,6 +291,7 @@ int amdgpu_umc_pasid_poison_handler(struct amdgpu_device *adev,
 				block, pasid, pasid_fn, data, reset);
 			if (!ret) {
 				atomic_inc(&con->page_retirement_req_cnt);
+				atomic_inc(&con->poison_consumption_count);
 				wake_up(&con->page_retirement_wq);
 			}
 		}
@@ -387,6 +427,45 @@ int amdgpu_umc_fill_error_record(struct ras_err_data *err_data,
 	return 0;
 }
 
+static int amdgpu_umc_loop_all_aid(struct amdgpu_device *adev, umc_func func,
+				   void *data)
+{
+	uint32_t umc_node_inst;
+	uint32_t node_inst;
+	uint32_t umc_inst;
+	uint32_t ch_inst;
+	int ret;
+
+	/*
+	 * This loop is done based on the following -
+	 * umc.active mask = mask of active umc instances across all nodes
+	 * umc.umc_inst_num = maximum number of umc instancess per node
+	 * umc.node_inst_num = maximum number of node instances
+	 * Channel instances are not assumed to be harvested.
+	 */
+	dev_dbg(adev->dev, "active umcs :%lx umc_inst per node: %d",
+		adev->umc.active_mask, adev->umc.umc_inst_num);
+	for_each_set_bit(umc_node_inst, &(adev->umc.active_mask),
+			 adev->umc.node_inst_num * adev->umc.umc_inst_num) {
+		node_inst = umc_node_inst / adev->umc.umc_inst_num;
+		umc_inst = umc_node_inst % adev->umc.umc_inst_num;
+		LOOP_UMC_CH_INST(ch_inst) {
+			dev_dbg(adev->dev,
+				"node_inst :%d umc_inst: %d ch_inst: %d",
+				node_inst, umc_inst, ch_inst);
+			ret = func(adev, node_inst, umc_inst, ch_inst, data);
+			if (ret) {
+				dev_err(adev->dev,
+					"Node %d umc %d ch %d func returns %d\n",
+					node_inst, umc_inst, ch_inst, ret);
+				return ret;
+			}
+		}
+	}
+
+	return 0;
+}
+
 int amdgpu_umc_loop_channels(struct amdgpu_device *adev,
 			umc_func func, void *data)
 {
@@ -395,6 +474,9 @@ int amdgpu_umc_loop_channels(struct amdgpu_device *adev,
 	uint32_t ch_inst         = 0;
 	int ret = 0;
 
+	if (adev->aid_mask)
+		return amdgpu_umc_loop_all_aid(adev, func, data);
+
 	if (adev->umc.node_inst_num) {
 		LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) {
 			ret = func(adev, node_inst, umc_inst, ch_inst, data);
@@ -487,6 +569,7 @@ int amdgpu_umc_lookup_bad_pages_in_a_row(struct amdgpu_device *adev,
 		pfns[i] = err_data.err_addr[i].retired_page;
 	}
 	ret = i;
+	adev->umc.err_addr_cnt = err_data.err_addr_cnt;
 
 out:
 	kfree(err_data.err_addr);
@@ -519,3 +602,26 @@ int amdgpu_umc_mca_to_addr(struct amdgpu_device *adev,
 
 	return 0;
 }
+
+int amdgpu_umc_pa2mca(struct amdgpu_device *adev,
+		uint64_t pa, uint64_t *mca, enum amdgpu_memory_partition nps)
+{
+	struct ta_ras_query_address_input addr_in;
+	struct ta_ras_query_address_output addr_out;
+	int ret;
+
+	/* nps: the pa belongs to */
+	addr_in.pa.pa = pa | ((uint64_t)nps << 58);
+	addr_in.addr_type = TA_RAS_PA_TO_MCA;
+	ret = psp_ras_query_address(&adev->psp, &addr_in, &addr_out);
+	if (ret) {
+		dev_warn(adev->dev, "Failed to query RAS MCA address for 0x%llx",
+			pa);
+
+		return ret;
+	}
+
+	*mca = addr_out.ma.err_addr;
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
index a4a7e61817aa..28dff750c47e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
@@ -71,6 +71,25 @@
  */
 #define UMC_CHANNEL_IDX_V2	BIT_ULL(47)
 
+/*
+ * save nps value to eeprom_table_record.retired_page[47:40],
+ * the channel index flag above will be retired.
+ */
+#define UMC_NPS_SHIFT 40
+#define UMC_NPS_MASK 0xffULL
+
+/* three column bits and one row bit in MCA address flip
+ * in bad page retirement
+ */
+#define RETIRE_FLIP_BITS_NUM 4
+
+struct amdgpu_umc_flip_bits {
+	uint32_t flip_bits_in_pa[RETIRE_FLIP_BITS_NUM];
+	uint32_t flip_row_bit;
+	uint32_t r13_in_pa;
+	uint32_t bit_num;
+};
+
 typedef int (*umc_func)(struct amdgpu_device *adev, uint32_t node_inst,
 			uint32_t umc_inst, uint32_t ch_inst, void *data);
 
@@ -93,6 +112,9 @@ struct amdgpu_umc_ras {
 			bool dump_addr);
 	uint32_t (*get_die_id_from_pa)(struct amdgpu_device *adev,
 			uint64_t mca_addr, uint64_t retired_page);
+	void (*get_retire_flip_bits)(struct amdgpu_device *adev);
+	void (*mca_ipid_parse)(struct amdgpu_device *adev, uint64_t ipid,
+			uint32_t *did, uint32_t *ch, uint32_t *umc_inst, uint32_t *sid);
 };
 
 struct amdgpu_umc_funcs {
@@ -123,6 +145,10 @@ struct amdgpu_umc {
 
 	/* active mask for umc node instance */
 	unsigned long active_mask;
+
+	struct amdgpu_umc_flip_bits flip_bits;
+
+	unsigned long err_addr_cnt;
 };
 
 int amdgpu_umc_ras_sw_init(struct amdgpu_device *adev);
@@ -165,4 +191,6 @@ int amdgpu_umc_mca_to_addr(struct amdgpu_device *adev,
 			uint64_t err_addr, uint32_t ch, uint32_t umc,
 			uint32_t node, uint32_t socket,
 			struct ta_ras_query_address_output *addr_out, bool dump_addr);
+int amdgpu_umc_pa2mca(struct amdgpu_device *adev,
+		uint64_t pa, uint64_t *mca, enum amdgpu_memory_partition nps);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c
index dde15c6a96e1..cd707d70a0bf 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c
@@ -32,462 +32,7 @@
 #include "amdgpu_umsch_mm.h"
 #include "umsch_mm_v4_0.h"
 
-struct umsch_mm_test_ctx_data {
-	uint8_t process_csa[PAGE_SIZE];
-	uint8_t vpe_ctx_csa[PAGE_SIZE];
-	uint8_t vcn_ctx_csa[PAGE_SIZE];
-};
-
-struct umsch_mm_test_mqd_data {
-	uint8_t vpe_mqd[PAGE_SIZE];
-	uint8_t vcn_mqd[PAGE_SIZE];
-};
-
-struct umsch_mm_test_ring_data {
-	uint8_t vpe_ring[PAGE_SIZE];
-	uint8_t vpe_ib[PAGE_SIZE];
-	uint8_t vcn_ring[PAGE_SIZE];
-	uint8_t vcn_ib[PAGE_SIZE];
-};
-
-struct umsch_mm_test_queue_info {
-	uint64_t mqd_addr;
-	uint64_t csa_addr;
-	uint32_t doorbell_offset_0;
-	uint32_t doorbell_offset_1;
-	enum UMSCH_SWIP_ENGINE_TYPE engine;
-};
-
-struct umsch_mm_test {
-	struct amdgpu_bo	*ctx_data_obj;
-	uint64_t		ctx_data_gpu_addr;
-	uint32_t		*ctx_data_cpu_addr;
-
-	struct amdgpu_bo	*mqd_data_obj;
-	uint64_t		mqd_data_gpu_addr;
-	uint32_t		*mqd_data_cpu_addr;
-
-	struct amdgpu_bo	*ring_data_obj;
-	uint64_t		ring_data_gpu_addr;
-	uint32_t		*ring_data_cpu_addr;
-
-
-	struct amdgpu_vm	*vm;
-	struct amdgpu_bo_va	*bo_va;
-	uint32_t		pasid;
-	uint32_t		vm_cntx_cntl;
-	uint32_t		num_queues;
-};
-
-static int map_ring_data(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			  struct amdgpu_bo *bo, struct amdgpu_bo_va **bo_va,
-			  uint64_t addr, uint32_t size)
-{
-	struct amdgpu_sync sync;
-	struct drm_exec exec;
-	int r;
-
-	amdgpu_sync_create(&sync);
-
-	drm_exec_init(&exec, 0, 0);
-	drm_exec_until_all_locked(&exec) {
-		r = drm_exec_lock_obj(&exec, &bo->tbo.base);
-		drm_exec_retry_on_contention(&exec);
-		if (unlikely(r))
-			goto error_fini_exec;
-
-		r = amdgpu_vm_lock_pd(vm, &exec, 0);
-		drm_exec_retry_on_contention(&exec);
-		if (unlikely(r))
-			goto error_fini_exec;
-	}
-
-	*bo_va = amdgpu_vm_bo_add(adev, vm, bo);
-	if (!*bo_va) {
-		r = -ENOMEM;
-		goto error_fini_exec;
-	}
-
-	r = amdgpu_vm_bo_map(adev, *bo_va, addr, 0, size,
-			     AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE |
-			     AMDGPU_PTE_EXECUTABLE);
-
-	if (r)
-		goto error_del_bo_va;
-
-
-	r = amdgpu_vm_bo_update(adev, *bo_va, false);
-	if (r)
-		goto error_del_bo_va;
-
-	amdgpu_sync_fence(&sync, (*bo_va)->last_pt_update);
-
-	r = amdgpu_vm_update_pdes(adev, vm, false);
-	if (r)
-		goto error_del_bo_va;
-
-	amdgpu_sync_fence(&sync, vm->last_update);
-
-	amdgpu_sync_wait(&sync, false);
-	drm_exec_fini(&exec);
-
-	amdgpu_sync_free(&sync);
-
-	return 0;
-
-error_del_bo_va:
-	amdgpu_vm_bo_del(adev, *bo_va);
-	amdgpu_sync_free(&sync);
-
-error_fini_exec:
-	drm_exec_fini(&exec);
-	amdgpu_sync_free(&sync);
-	return r;
-}
-
-static int unmap_ring_data(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			    struct amdgpu_bo *bo, struct amdgpu_bo_va *bo_va,
-			    uint64_t addr)
-{
-	struct drm_exec exec;
-	long r;
-
-	drm_exec_init(&exec, 0, 0);
-	drm_exec_until_all_locked(&exec) {
-		r = drm_exec_lock_obj(&exec, &bo->tbo.base);
-		drm_exec_retry_on_contention(&exec);
-		if (unlikely(r))
-			goto out_unlock;
-
-		r = amdgpu_vm_lock_pd(vm, &exec, 0);
-		drm_exec_retry_on_contention(&exec);
-		if (unlikely(r))
-			goto out_unlock;
-	}
-
-
-	r = amdgpu_vm_bo_unmap(adev, bo_va, addr);
-	if (r)
-		goto out_unlock;
-
-	amdgpu_vm_bo_del(adev, bo_va);
-
-out_unlock:
-	drm_exec_fini(&exec);
-
-	return r;
-}
-
-static void setup_vpe_queue(struct amdgpu_device *adev,
-			    struct umsch_mm_test *test,
-			    struct umsch_mm_test_queue_info *qinfo)
-{
-	struct MQD_INFO *mqd = (struct MQD_INFO *)test->mqd_data_cpu_addr;
-	uint64_t ring_gpu_addr = test->ring_data_gpu_addr;
-
-	mqd->rb_base_lo = (ring_gpu_addr >> 8);
-	mqd->rb_base_hi = (ring_gpu_addr >> 40);
-	mqd->rb_size = PAGE_SIZE / 4;
-	mqd->wptr_val = 0;
-	mqd->rptr_val = 0;
-	mqd->unmapped = 1;
-
-	if (adev->vpe.collaborate_mode)
-		memcpy(++mqd, test->mqd_data_cpu_addr, sizeof(struct MQD_INFO));
-
-	qinfo->mqd_addr = test->mqd_data_gpu_addr;
-	qinfo->csa_addr = test->ctx_data_gpu_addr +
-		offsetof(struct umsch_mm_test_ctx_data, vpe_ctx_csa);
-	qinfo->doorbell_offset_0 = 0;
-	qinfo->doorbell_offset_1 = 0;
-}
-
-static void setup_vcn_queue(struct amdgpu_device *adev,
-			    struct umsch_mm_test *test,
-			    struct umsch_mm_test_queue_info *qinfo)
-{
-}
-
-static int add_test_queue(struct amdgpu_device *adev,
-			  struct umsch_mm_test *test,
-			  struct umsch_mm_test_queue_info *qinfo)
-{
-	struct umsch_mm_add_queue_input queue_input = {};
-	int r;
-
-	queue_input.process_id = test->pasid;
-	queue_input.page_table_base_addr = amdgpu_gmc_pd_addr(test->vm->root.bo);
-
-	queue_input.process_va_start = 0;
-	queue_input.process_va_end = (adev->vm_manager.max_pfn - 1) << AMDGPU_GPU_PAGE_SHIFT;
-
-	queue_input.process_quantum = 100000; /* 10ms */
-	queue_input.process_csa_addr = test->ctx_data_gpu_addr +
-				       offsetof(struct umsch_mm_test_ctx_data, process_csa);
-
-	queue_input.context_quantum = 10000; /* 1ms */
-	queue_input.context_csa_addr = qinfo->csa_addr;
-
-	queue_input.inprocess_context_priority = CONTEXT_PRIORITY_LEVEL_NORMAL;
-	queue_input.context_global_priority_level = CONTEXT_PRIORITY_LEVEL_NORMAL;
-	queue_input.doorbell_offset_0 = qinfo->doorbell_offset_0;
-	queue_input.doorbell_offset_1 = qinfo->doorbell_offset_1;
-
-	queue_input.engine_type = qinfo->engine;
-	queue_input.mqd_addr = qinfo->mqd_addr;
-	queue_input.vm_context_cntl = test->vm_cntx_cntl;
-
-	amdgpu_umsch_mm_lock(&adev->umsch_mm);
-	r = adev->umsch_mm.funcs->add_queue(&adev->umsch_mm, &queue_input);
-	amdgpu_umsch_mm_unlock(&adev->umsch_mm);
-	if (r)
-		return r;
-
-	return 0;
-}
-
-static int remove_test_queue(struct amdgpu_device *adev,
-			     struct umsch_mm_test *test,
-			     struct umsch_mm_test_queue_info *qinfo)
-{
-	struct umsch_mm_remove_queue_input queue_input = {};
-	int r;
-
-	queue_input.doorbell_offset_0 = qinfo->doorbell_offset_0;
-	queue_input.doorbell_offset_1 = qinfo->doorbell_offset_1;
-	queue_input.context_csa_addr = qinfo->csa_addr;
-
-	amdgpu_umsch_mm_lock(&adev->umsch_mm);
-	r = adev->umsch_mm.funcs->remove_queue(&adev->umsch_mm, &queue_input);
-	amdgpu_umsch_mm_unlock(&adev->umsch_mm);
-	if (r)
-		return r;
-
-	return 0;
-}
-
-static int submit_vpe_queue(struct amdgpu_device *adev, struct umsch_mm_test *test)
-{
-	struct MQD_INFO *mqd = (struct MQD_INFO *)test->mqd_data_cpu_addr;
-	uint32_t *ring = test->ring_data_cpu_addr +
-		offsetof(struct umsch_mm_test_ring_data, vpe_ring) / 4;
-	uint32_t *ib = test->ring_data_cpu_addr +
-		offsetof(struct umsch_mm_test_ring_data, vpe_ib) / 4;
-	uint64_t ib_gpu_addr = test->ring_data_gpu_addr +
-		offsetof(struct umsch_mm_test_ring_data, vpe_ib);
-	uint32_t *fence = ib + 2048 / 4;
-	uint64_t fence_gpu_addr = ib_gpu_addr + 2048;
-	const uint32_t test_pattern = 0xdeadbeef;
-	int i;
-
-	ib[0] = VPE_CMD_HEADER(VPE_CMD_OPCODE_FENCE, 0);
-	ib[1] = lower_32_bits(fence_gpu_addr);
-	ib[2] = upper_32_bits(fence_gpu_addr);
-	ib[3] = test_pattern;
-
-	ring[0] = VPE_CMD_HEADER(VPE_CMD_OPCODE_INDIRECT, 0);
-	ring[1] = (ib_gpu_addr & 0xffffffe0);
-	ring[2] = upper_32_bits(ib_gpu_addr);
-	ring[3] = 4;
-	ring[4] = 0;
-	ring[5] = 0;
-
-	mqd->wptr_val = (6 << 2);
-	if (adev->vpe.collaborate_mode)
-		(++mqd)->wptr_val = (6 << 2);
-
-	WDOORBELL32(adev->umsch_mm.agdb_index[CONTEXT_PRIORITY_LEVEL_NORMAL], mqd->wptr_val);
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		if (*fence == test_pattern)
-			return 0;
-		udelay(1);
-	}
-
-	dev_err(adev->dev, "vpe queue submission timeout\n");
-
-	return -ETIMEDOUT;
-}
-
-static int submit_vcn_queue(struct amdgpu_device *adev, struct umsch_mm_test *test)
-{
-	return 0;
-}
-
-static int setup_umsch_mm_test(struct amdgpu_device *adev,
-			  struct umsch_mm_test *test)
-{
-	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB0(0)];
-	int r;
-
-	test->vm_cntx_cntl = hub->vm_cntx_cntl;
-
-	test->vm = kzalloc(sizeof(*test->vm), GFP_KERNEL);
-	if (!test->vm) {
-		r = -ENOMEM;
-		return r;
-	}
-
-	r = amdgpu_vm_init(adev, test->vm, -1);
-	if (r)
-		goto error_free_vm;
-
-	r = amdgpu_pasid_alloc(16);
-	if (r < 0)
-		goto error_fini_vm;
-	test->pasid = r;
-
-	r = amdgpu_bo_create_kernel(adev, sizeof(struct umsch_mm_test_ctx_data),
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
-				    &test->ctx_data_obj,
-				    &test->ctx_data_gpu_addr,
-				    (void **)&test->ctx_data_cpu_addr);
-	if (r)
-		goto error_free_pasid;
-
-	memset(test->ctx_data_cpu_addr, 0, sizeof(struct umsch_mm_test_ctx_data));
-
-	r = amdgpu_bo_create_kernel(adev, PAGE_SIZE,
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
-				    &test->mqd_data_obj,
-				    &test->mqd_data_gpu_addr,
-				    (void **)&test->mqd_data_cpu_addr);
-	if (r)
-		goto error_free_ctx_data_obj;
-
-	memset(test->mqd_data_cpu_addr, 0, PAGE_SIZE);
-
-	r = amdgpu_bo_create_kernel(adev, sizeof(struct umsch_mm_test_ring_data),
-				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
-				    &test->ring_data_obj,
-				    NULL,
-				    (void **)&test->ring_data_cpu_addr);
-	if (r)
-		goto error_free_mqd_data_obj;
-
-	memset(test->ring_data_cpu_addr, 0, sizeof(struct umsch_mm_test_ring_data));
-
-	test->ring_data_gpu_addr = AMDGPU_VA_RESERVED_BOTTOM;
-	r = map_ring_data(adev, test->vm, test->ring_data_obj, &test->bo_va,
-			  test->ring_data_gpu_addr, sizeof(struct umsch_mm_test_ring_data));
-	if (r)
-		goto error_free_ring_data_obj;
-
-	return 0;
-
-error_free_ring_data_obj:
-	amdgpu_bo_free_kernel(&test->ring_data_obj, NULL,
-			      (void **)&test->ring_data_cpu_addr);
-error_free_mqd_data_obj:
-	amdgpu_bo_free_kernel(&test->mqd_data_obj, &test->mqd_data_gpu_addr,
-			      (void **)&test->mqd_data_cpu_addr);
-error_free_ctx_data_obj:
-	amdgpu_bo_free_kernel(&test->ctx_data_obj, &test->ctx_data_gpu_addr,
-			      (void **)&test->ctx_data_cpu_addr);
-error_free_pasid:
-	amdgpu_pasid_free(test->pasid);
-error_fini_vm:
-	amdgpu_vm_fini(adev, test->vm);
-error_free_vm:
-	kfree(test->vm);
-
-	return r;
-}
-
-static void cleanup_umsch_mm_test(struct amdgpu_device *adev,
-				  struct umsch_mm_test *test)
-{
-	unmap_ring_data(adev, test->vm, test->ring_data_obj,
-			test->bo_va, test->ring_data_gpu_addr);
-	amdgpu_bo_free_kernel(&test->mqd_data_obj, &test->mqd_data_gpu_addr,
-			      (void **)&test->mqd_data_cpu_addr);
-	amdgpu_bo_free_kernel(&test->ring_data_obj, NULL,
-			      (void **)&test->ring_data_cpu_addr);
-	amdgpu_bo_free_kernel(&test->ctx_data_obj, &test->ctx_data_gpu_addr,
-			       (void **)&test->ctx_data_cpu_addr);
-	amdgpu_pasid_free(test->pasid);
-	amdgpu_vm_fini(adev, test->vm);
-	kfree(test->vm);
-}
-
-static int setup_test_queues(struct amdgpu_device *adev,
-			     struct umsch_mm_test *test,
-			     struct umsch_mm_test_queue_info *qinfo)
-{
-	int i, r;
-
-	for (i = 0; i < test->num_queues; i++) {
-		if (qinfo[i].engine == UMSCH_SWIP_ENGINE_TYPE_VPE)
-			setup_vpe_queue(adev, test, &qinfo[i]);
-		else
-			setup_vcn_queue(adev, test, &qinfo[i]);
-
-		r = add_test_queue(adev, test, &qinfo[i]);
-		if (r)
-			return r;
-	}
-
-	return 0;
-}
-
-static int submit_test_queues(struct amdgpu_device *adev,
-			      struct umsch_mm_test *test,
-			      struct umsch_mm_test_queue_info *qinfo)
-{
-	int i, r;
-
-	for (i = 0; i < test->num_queues; i++) {
-		if (qinfo[i].engine == UMSCH_SWIP_ENGINE_TYPE_VPE)
-			r = submit_vpe_queue(adev, test);
-		else
-			r = submit_vcn_queue(adev, test);
-		if (r)
-			return r;
-	}
-
-	return 0;
-}
-
-static void cleanup_test_queues(struct amdgpu_device *adev,
-			      struct umsch_mm_test *test,
-			      struct umsch_mm_test_queue_info *qinfo)
-{
-	int i;
-
-	for (i = 0; i < test->num_queues; i++)
-		remove_test_queue(adev, test, &qinfo[i]);
-}
-
-static int umsch_mm_test(struct amdgpu_device *adev)
-{
-	struct umsch_mm_test_queue_info qinfo[] = {
-		{ .engine = UMSCH_SWIP_ENGINE_TYPE_VPE },
-	};
-	struct umsch_mm_test test = { .num_queues = ARRAY_SIZE(qinfo) };
-	int r;
-
-	r = setup_umsch_mm_test(adev, &test);
-	if (r)
-		return r;
-
-	r = setup_test_queues(adev, &test, qinfo);
-	if (r)
-		goto cleanup;
-
-	r = submit_test_queues(adev, &test, qinfo);
-	if (r)
-		goto cleanup;
-
-	cleanup_test_queues(adev, &test, qinfo);
-	cleanup_umsch_mm_test(adev, &test);
-
-	return 0;
-
-cleanup:
-	cleanup_test_queues(adev, &test, qinfo);
-	cleanup_umsch_mm_test(adev, &test);
-	return r;
-}
+MODULE_FIRMWARE("amdgpu/umsch_mm_4_0_0.bin");
 
 int amdgpu_umsch_mm_submit_pkt(struct amdgpu_umsch_mm *umsch, void *pkt, int ndws)
 {
@@ -581,14 +126,14 @@ int amdgpu_umsch_mm_init_microcode(struct amdgpu_umsch_mm *umsch)
 	switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) {
 	case IP_VERSION(4, 0, 5):
 	case IP_VERSION(4, 0, 6):
-		fw_name = "amdgpu/umsch_mm_4_0_0.bin";
+		fw_name = "4_0_0";
 		break;
 	default:
-		break;
+		return -EINVAL;
 	}
 
 	r = amdgpu_ucode_request(adev, &adev->umsch_mm.fw, AMDGPU_UCODE_REQUIRED,
-				 "%s", fw_name);
+				 "amdgpu/umsch_mm_%s.bin", fw_name);
 	if (r) {
 		release_firmware(adev->umsch_mm.fw);
 		adev->umsch_mm.fw = NULL;
@@ -792,7 +337,7 @@ static int umsch_mm_late_init(struct amdgpu_ip_block *ip_block)
 	if (amdgpu_in_reset(adev) || adev->in_s0ix || adev->in_suspend)
 		return 0;
 
-	return umsch_mm_test(adev);
+	return 0;
 }
 
 static int umsch_mm_sw_init(struct amdgpu_ip_block *ip_block)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c
new file mode 100644
index 000000000000..9a969175900e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c
@@ -0,0 +1,1482 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <drm/drm_auth.h>
+#include <drm/drm_exec.h>
+#include <linux/pm_runtime.h>
+#include <drm/drm_drv.h>
+
+#include "amdgpu.h"
+#include "amdgpu_reset.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_userq.h"
+#include "amdgpu_hmm.h"
+#include "amdgpu_userq_fence.h"
+
+u32 amdgpu_userq_get_supported_ip_mask(struct amdgpu_device *adev)
+{
+	int i;
+	u32 userq_ip_mask = 0;
+
+	for (i = 0; i < AMDGPU_HW_IP_NUM; i++) {
+		if (adev->userq_funcs[i])
+			userq_ip_mask |= (1 << i);
+	}
+
+	return userq_ip_mask;
+}
+
+static bool amdgpu_userq_is_reset_type_supported(struct amdgpu_device *adev,
+				enum amdgpu_ring_type ring_type, int reset_type)
+{
+
+	if (ring_type < 0 || ring_type >= AMDGPU_RING_TYPE_MAX)
+		return false;
+
+	switch (ring_type) {
+	case AMDGPU_RING_TYPE_GFX:
+		if (adev->gfx.gfx_supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_COMPUTE:
+		if (adev->gfx.compute_supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_SDMA:
+		if (adev->sdma.supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_VCN_DEC:
+	case AMDGPU_RING_TYPE_VCN_ENC:
+		if (adev->vcn.supported_reset & reset_type)
+			return true;
+		break;
+	case AMDGPU_RING_TYPE_VCN_JPEG:
+		if (adev->jpeg.supported_reset & reset_type)
+			return true;
+		break;
+	default:
+		break;
+	}
+	return false;
+}
+
+static void amdgpu_userq_gpu_reset(struct amdgpu_device *adev)
+{
+	if (amdgpu_device_should_recover_gpu(adev)) {
+		amdgpu_reset_domain_schedule(adev->reset_domain,
+					     &adev->userq_reset_work);
+		/* Wait for the reset job to complete */
+		flush_work(&adev->userq_reset_work);
+	}
+}
+
+static int
+amdgpu_userq_detect_and_reset_queues(struct amdgpu_userq_mgr *uq_mgr)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const int queue_types[] = {
+		AMDGPU_RING_TYPE_COMPUTE,
+		AMDGPU_RING_TYPE_GFX,
+		AMDGPU_RING_TYPE_SDMA
+	};
+	const int num_queue_types = ARRAY_SIZE(queue_types);
+	bool gpu_reset = false;
+	int r = 0;
+	int i;
+
+	/* Warning if current process mutex is not held */
+	WARN_ON(!mutex_is_locked(&uq_mgr->userq_mutex));
+
+	if (unlikely(adev->debug_disable_gpu_ring_reset)) {
+		dev_err(adev->dev, "userq reset disabled by debug mask\n");
+		return 0;
+	}
+
+	/*
+	 * If GPU recovery feature is disabled system-wide,
+	 * skip all reset detection logic
+	 */
+	if (!amdgpu_gpu_recovery)
+		return 0;
+
+	/*
+	 * Iterate through all queue types to detect and reset problematic queues
+	 * Process each queue type in the defined order
+	 */
+	for (i = 0; i < num_queue_types; i++) {
+		int ring_type = queue_types[i];
+		const struct amdgpu_userq_funcs *funcs = adev->userq_funcs[ring_type];
+
+		if (!amdgpu_userq_is_reset_type_supported(adev, ring_type, AMDGPU_RESET_TYPE_PER_QUEUE))
+				continue;
+
+		if (atomic_read(&uq_mgr->userq_count[ring_type]) > 0 &&
+		    funcs && funcs->detect_and_reset) {
+			r = funcs->detect_and_reset(adev, ring_type);
+			if (r) {
+				gpu_reset = true;
+				break;
+			}
+		}
+	}
+
+	if (gpu_reset)
+		amdgpu_userq_gpu_reset(adev);
+
+	return r;
+}
+
+static int amdgpu_userq_buffer_va_list_add(struct amdgpu_usermode_queue *queue,
+					   struct amdgpu_bo_va_mapping *va_map, u64 addr)
+{
+	struct amdgpu_userq_va_cursor *va_cursor;
+	struct userq_va_list;
+
+	va_cursor = kzalloc(sizeof(*va_cursor), GFP_KERNEL);
+	if (!va_cursor)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&va_cursor->list);
+	va_cursor->gpu_addr = addr;
+	atomic_set(&va_map->bo_va->userq_va_mapped, 1);
+	list_add(&va_cursor->list, &queue->userq_va_list);
+
+	return 0;
+}
+
+int amdgpu_userq_input_va_validate(struct amdgpu_usermode_queue *queue,
+				   u64 addr, u64 expected_size)
+{
+	struct amdgpu_bo_va_mapping *va_map;
+	struct amdgpu_vm *vm = queue->vm;
+	u64 user_addr;
+	u64 size;
+	int r = 0;
+
+	user_addr = (addr & AMDGPU_GMC_HOLE_MASK) >> AMDGPU_GPU_PAGE_SHIFT;
+	size = expected_size >> AMDGPU_GPU_PAGE_SHIFT;
+
+	r = amdgpu_bo_reserve(vm->root.bo, false);
+	if (r)
+		return r;
+
+	va_map = amdgpu_vm_bo_lookup_mapping(vm, user_addr);
+	if (!va_map) {
+		r = -EINVAL;
+		goto out_err;
+	}
+	/* Only validate the userq whether resident in the VM mapping range */
+	if (user_addr >= va_map->start  &&
+	    va_map->last - user_addr + 1 >= size) {
+		amdgpu_userq_buffer_va_list_add(queue, va_map, user_addr);
+		amdgpu_bo_unreserve(vm->root.bo);
+		return 0;
+	}
+
+	r = -EINVAL;
+out_err:
+	amdgpu_bo_unreserve(vm->root.bo);
+	return r;
+}
+
+static bool amdgpu_userq_buffer_va_mapped(struct amdgpu_vm *vm, u64 addr)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	bool r;
+
+	if (amdgpu_bo_reserve(vm->root.bo, false))
+		return false;
+
+	mapping = amdgpu_vm_bo_lookup_mapping(vm, addr);
+	if (!IS_ERR_OR_NULL(mapping) && atomic_read(&mapping->bo_va->userq_va_mapped))
+		r = true;
+	else
+		r = false;
+	amdgpu_bo_unreserve(vm->root.bo);
+
+	return r;
+}
+
+static bool amdgpu_userq_buffer_vas_mapped(struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_userq_va_cursor *va_cursor, *tmp;
+	int r = 0;
+
+	list_for_each_entry_safe(va_cursor, tmp, &queue->userq_va_list, list) {
+		r += amdgpu_userq_buffer_va_mapped(queue->vm, va_cursor->gpu_addr);
+		dev_dbg(queue->userq_mgr->adev->dev,
+			"validate the userq mapping:%p va:%llx r:%d\n",
+			queue, va_cursor->gpu_addr, r);
+	}
+
+	if (r != 0)
+		return true;
+
+	return false;
+}
+
+static void amdgpu_userq_buffer_va_list_del(struct amdgpu_bo_va_mapping *mapping,
+					    struct amdgpu_userq_va_cursor *va_cursor)
+{
+	atomic_set(&mapping->bo_va->userq_va_mapped, 0);
+	list_del(&va_cursor->list);
+	kfree(va_cursor);
+}
+
+static int amdgpu_userq_buffer_vas_list_cleanup(struct amdgpu_device *adev,
+						struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_userq_va_cursor *va_cursor, *tmp;
+	struct amdgpu_bo_va_mapping *mapping;
+	int r;
+
+	r = amdgpu_bo_reserve(queue->vm->root.bo, false);
+	if (r)
+		return r;
+
+	list_for_each_entry_safe(va_cursor, tmp, &queue->userq_va_list, list) {
+		mapping = amdgpu_vm_bo_lookup_mapping(queue->vm, va_cursor->gpu_addr);
+		if (!mapping) {
+			r = -EINVAL;
+			goto err;
+		}
+		dev_dbg(adev->dev, "delete the userq:%p va:%llx\n",
+			queue, va_cursor->gpu_addr);
+		amdgpu_userq_buffer_va_list_del(mapping, va_cursor);
+	}
+err:
+	amdgpu_bo_unreserve(queue->vm->root.bo);
+	return r;
+}
+
+static int
+amdgpu_userq_preempt_helper(struct amdgpu_userq_mgr *uq_mgr,
+			  struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *userq_funcs =
+		adev->userq_funcs[queue->queue_type];
+	bool found_hung_queue = false;
+	int r = 0;
+
+	if (queue->state == AMDGPU_USERQ_STATE_MAPPED) {
+		r = userq_funcs->preempt(uq_mgr, queue);
+		if (r) {
+			queue->state = AMDGPU_USERQ_STATE_HUNG;
+			found_hung_queue = true;
+		} else {
+			queue->state = AMDGPU_USERQ_STATE_PREEMPTED;
+		}
+	}
+
+	if (found_hung_queue)
+		amdgpu_userq_detect_and_reset_queues(uq_mgr);
+
+	return r;
+}
+
+static int
+amdgpu_userq_restore_helper(struct amdgpu_userq_mgr *uq_mgr,
+			struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *userq_funcs =
+		adev->userq_funcs[queue->queue_type];
+	int r = 0;
+
+	if (queue->state == AMDGPU_USERQ_STATE_PREEMPTED) {
+		r = userq_funcs->restore(uq_mgr, queue);
+		if (r) {
+			queue->state = AMDGPU_USERQ_STATE_HUNG;
+		} else {
+			queue->state = AMDGPU_USERQ_STATE_MAPPED;
+		}
+	}
+
+	return r;
+}
+
+static int
+amdgpu_userq_unmap_helper(struct amdgpu_userq_mgr *uq_mgr,
+			  struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *userq_funcs =
+		adev->userq_funcs[queue->queue_type];
+	bool found_hung_queue = false;
+	int r = 0;
+
+	if ((queue->state == AMDGPU_USERQ_STATE_MAPPED) ||
+		(queue->state == AMDGPU_USERQ_STATE_PREEMPTED)) {
+		r = userq_funcs->unmap(uq_mgr, queue);
+		if (r) {
+			queue->state = AMDGPU_USERQ_STATE_HUNG;
+			found_hung_queue = true;
+		} else {
+			queue->state = AMDGPU_USERQ_STATE_UNMAPPED;
+		}
+	}
+
+	if (found_hung_queue)
+		amdgpu_userq_detect_and_reset_queues(uq_mgr);
+
+	return r;
+}
+
+static int
+amdgpu_userq_map_helper(struct amdgpu_userq_mgr *uq_mgr,
+			struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *userq_funcs =
+		adev->userq_funcs[queue->queue_type];
+	int r = 0;
+
+	if (queue->state == AMDGPU_USERQ_STATE_UNMAPPED) {
+		r = userq_funcs->map(uq_mgr, queue);
+		if (r) {
+			queue->state = AMDGPU_USERQ_STATE_HUNG;
+			amdgpu_userq_detect_and_reset_queues(uq_mgr);
+		} else {
+			queue->state = AMDGPU_USERQ_STATE_MAPPED;
+		}
+	}
+
+	return r;
+}
+
+static int
+amdgpu_userq_wait_for_last_fence(struct amdgpu_userq_mgr *uq_mgr,
+				 struct amdgpu_usermode_queue *queue)
+{
+	struct dma_fence *f = queue->last_fence;
+	int ret = 0;
+
+	if (f && !dma_fence_is_signaled(f)) {
+		ret = dma_fence_wait_timeout(f, true, MAX_SCHEDULE_TIMEOUT);
+		if (ret <= 0) {
+			drm_file_err(uq_mgr->file, "Timed out waiting for fence=%llu:%llu\n",
+				     f->context, f->seqno);
+			queue->state = AMDGPU_USERQ_STATE_HUNG;
+			return -ETIME;
+		}
+	}
+
+	return ret;
+}
+
+static void
+amdgpu_userq_cleanup(struct amdgpu_userq_mgr *uq_mgr,
+		     struct amdgpu_usermode_queue *queue,
+		     int queue_id)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *uq_funcs = adev->userq_funcs[queue->queue_type];
+
+	/* Wait for mode-1 reset to complete */
+	down_read(&adev->reset_domain->sem);
+
+	/* Drop the userq reference. */
+	amdgpu_userq_buffer_vas_list_cleanup(adev, queue);
+	uq_funcs->mqd_destroy(uq_mgr, queue);
+	amdgpu_userq_fence_driver_free(queue);
+	/* Use interrupt-safe locking since IRQ handlers may access these XArrays */
+	xa_erase_irq(&uq_mgr->userq_mgr_xa, (unsigned long)queue_id);
+	xa_erase_irq(&adev->userq_doorbell_xa, queue->doorbell_index);
+	queue->userq_mgr = NULL;
+	list_del(&queue->userq_va_list);
+	kfree(queue);
+
+	up_read(&adev->reset_domain->sem);
+}
+
+static struct amdgpu_usermode_queue *
+amdgpu_userq_find(struct amdgpu_userq_mgr *uq_mgr, int qid)
+{
+	return xa_load(&uq_mgr->userq_mgr_xa, qid);
+}
+
+void
+amdgpu_userq_ensure_ev_fence(struct amdgpu_userq_mgr *uq_mgr,
+			     struct amdgpu_eviction_fence_mgr *evf_mgr)
+{
+	struct amdgpu_eviction_fence *ev_fence;
+
+retry:
+	/* Flush any pending resume work to create ev_fence */
+	flush_delayed_work(&uq_mgr->resume_work);
+
+	mutex_lock(&uq_mgr->userq_mutex);
+	spin_lock(&evf_mgr->ev_fence_lock);
+	ev_fence = evf_mgr->ev_fence;
+	spin_unlock(&evf_mgr->ev_fence_lock);
+	if (!ev_fence || dma_fence_is_signaled(&ev_fence->base)) {
+		mutex_unlock(&uq_mgr->userq_mutex);
+		/*
+		 * Looks like there was no pending resume work,
+		 * add one now to create a valid eviction fence
+		 */
+		schedule_delayed_work(&uq_mgr->resume_work, 0);
+		goto retry;
+	}
+}
+
+int amdgpu_userq_create_object(struct amdgpu_userq_mgr *uq_mgr,
+			       struct amdgpu_userq_obj *userq_obj,
+			       int size)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct amdgpu_bo_param bp;
+	int r;
+
+	memset(&bp, 0, sizeof(bp));
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_GTT;
+	bp.flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
+		   AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	bp.type = ttm_bo_type_kernel;
+	bp.size = size;
+	bp.resv = NULL;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
+
+	r = amdgpu_bo_create(adev, &bp, &userq_obj->obj);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to allocate BO for userqueue (%d)", r);
+		return r;
+	}
+
+	r = amdgpu_bo_reserve(userq_obj->obj, true);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to reserve BO to map (%d)", r);
+		goto free_obj;
+	}
+
+	r = amdgpu_ttm_alloc_gart(&(userq_obj->obj)->tbo);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to alloc GART for userqueue object (%d)", r);
+		goto unresv;
+	}
+
+	r = amdgpu_bo_kmap(userq_obj->obj, &userq_obj->cpu_ptr);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to map BO for userqueue (%d)", r);
+		goto unresv;
+	}
+
+	userq_obj->gpu_addr = amdgpu_bo_gpu_offset(userq_obj->obj);
+	amdgpu_bo_unreserve(userq_obj->obj);
+	memset(userq_obj->cpu_ptr, 0, size);
+	return 0;
+
+unresv:
+	amdgpu_bo_unreserve(userq_obj->obj);
+
+free_obj:
+	amdgpu_bo_unref(&userq_obj->obj);
+	return r;
+}
+
+void amdgpu_userq_destroy_object(struct amdgpu_userq_mgr *uq_mgr,
+				 struct amdgpu_userq_obj *userq_obj)
+{
+	amdgpu_bo_kunmap(userq_obj->obj);
+	amdgpu_bo_unref(&userq_obj->obj);
+}
+
+uint64_t
+amdgpu_userq_get_doorbell_index(struct amdgpu_userq_mgr *uq_mgr,
+				struct amdgpu_db_info *db_info,
+				struct drm_file *filp)
+{
+	uint64_t index;
+	struct drm_gem_object *gobj;
+	struct amdgpu_userq_obj *db_obj = db_info->db_obj;
+	int r, db_size;
+
+	gobj = drm_gem_object_lookup(filp, db_info->doorbell_handle);
+	if (gobj == NULL) {
+		drm_file_err(uq_mgr->file, "Can't find GEM object for doorbell\n");
+		return -EINVAL;
+	}
+
+	db_obj->obj = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
+	drm_gem_object_put(gobj);
+
+	r = amdgpu_bo_reserve(db_obj->obj, true);
+	if (r) {
+		drm_file_err(uq_mgr->file, "[Usermode queues] Failed to pin doorbell object\n");
+		goto unref_bo;
+	}
+
+	/* Pin the BO before generating the index, unpin in queue destroy */
+	r = amdgpu_bo_pin(db_obj->obj, AMDGPU_GEM_DOMAIN_DOORBELL);
+	if (r) {
+		drm_file_err(uq_mgr->file, "[Usermode queues] Failed to pin doorbell object\n");
+		goto unresv_bo;
+	}
+
+	switch (db_info->queue_type) {
+	case AMDGPU_HW_IP_GFX:
+	case AMDGPU_HW_IP_COMPUTE:
+	case AMDGPU_HW_IP_DMA:
+		db_size = sizeof(u64);
+		break;
+	default:
+		drm_file_err(uq_mgr->file, "[Usermode queues] IP %d not support\n",
+			     db_info->queue_type);
+		r = -EINVAL;
+		goto unpin_bo;
+	}
+
+	index = amdgpu_doorbell_index_on_bar(uq_mgr->adev, db_obj->obj,
+					     db_info->doorbell_offset, db_size);
+	drm_dbg_driver(adev_to_drm(uq_mgr->adev),
+		       "[Usermode queues] doorbell index=%lld\n", index);
+	amdgpu_bo_unreserve(db_obj->obj);
+	return index;
+
+unpin_bo:
+	amdgpu_bo_unpin(db_obj->obj);
+unresv_bo:
+	amdgpu_bo_unreserve(db_obj->obj);
+unref_bo:
+	amdgpu_bo_unref(&db_obj->obj);
+	return r;
+}
+
+static int
+amdgpu_userq_destroy(struct drm_file *filp, int queue_id)
+{
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_userq_mgr *uq_mgr = &fpriv->userq_mgr;
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct amdgpu_usermode_queue *queue;
+	int r = 0;
+
+	cancel_delayed_work_sync(&uq_mgr->resume_work);
+	mutex_lock(&uq_mgr->userq_mutex);
+
+	queue = amdgpu_userq_find(uq_mgr, queue_id);
+	if (!queue) {
+		drm_dbg_driver(adev_to_drm(uq_mgr->adev), "Invalid queue id to destroy\n");
+		mutex_unlock(&uq_mgr->userq_mutex);
+		return -EINVAL;
+	}
+	amdgpu_userq_wait_for_last_fence(uq_mgr, queue);
+	r = amdgpu_bo_reserve(queue->db_obj.obj, true);
+	if (!r) {
+		amdgpu_bo_unpin(queue->db_obj.obj);
+		amdgpu_bo_unreserve(queue->db_obj.obj);
+	}
+	amdgpu_bo_unref(&queue->db_obj.obj);
+	atomic_dec(&uq_mgr->userq_count[queue->queue_type]);
+#if defined(CONFIG_DEBUG_FS)
+	debugfs_remove_recursive(queue->debugfs_queue);
+#endif
+	amdgpu_userq_detect_and_reset_queues(uq_mgr);
+	r = amdgpu_userq_unmap_helper(uq_mgr, queue);
+	/*TODO: It requires a reset for userq hw unmap error*/
+	if (unlikely(r != AMDGPU_USERQ_STATE_UNMAPPED)) {
+		drm_warn(adev_to_drm(uq_mgr->adev), "trying to destroy a HW mapping userq\n");
+		queue->state = AMDGPU_USERQ_STATE_HUNG;
+	}
+	amdgpu_userq_cleanup(uq_mgr, queue, queue_id);
+	mutex_unlock(&uq_mgr->userq_mutex);
+
+	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+
+	return r;
+}
+
+static int amdgpu_userq_priority_permit(struct drm_file *filp,
+					int priority)
+{
+	if (priority < AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_HIGH)
+		return 0;
+
+	if (capable(CAP_SYS_NICE))
+		return 0;
+
+	if (drm_is_current_master(filp))
+		return 0;
+
+	return -EACCES;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_mqd_info_read(struct seq_file *m, void *unused)
+{
+	struct amdgpu_usermode_queue *queue = m->private;
+	struct amdgpu_bo *bo;
+	int r;
+
+	if (!queue || !queue->mqd.obj)
+		return -EINVAL;
+
+	bo = amdgpu_bo_ref(queue->mqd.obj);
+	r = amdgpu_bo_reserve(bo, true);
+	if (r) {
+		amdgpu_bo_unref(&bo);
+		return -EINVAL;
+	}
+
+	seq_printf(m, "queue_type: %d\n", queue->queue_type);
+	seq_printf(m, "mqd_gpu_address: 0x%llx\n", amdgpu_bo_gpu_offset(queue->mqd.obj));
+
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+
+	return 0;
+}
+
+static int amdgpu_mqd_info_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, amdgpu_mqd_info_read, inode->i_private);
+}
+
+static const struct file_operations amdgpu_mqd_info_fops = {
+	.owner = THIS_MODULE,
+	.open = amdgpu_mqd_info_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+#endif
+
+static int
+amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args)
+{
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_userq_mgr *uq_mgr = &fpriv->userq_mgr;
+	struct amdgpu_device *adev = uq_mgr->adev;
+	const struct amdgpu_userq_funcs *uq_funcs;
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_db_info db_info;
+	char *queue_name;
+	bool skip_map_queue;
+	u32 qid;
+	uint64_t index;
+	int r = 0;
+	int priority =
+		(args->in.flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_MASK) >>
+		AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_SHIFT;
+
+	r = amdgpu_userq_priority_permit(filp, priority);
+	if (r)
+		return r;
+
+	r = pm_runtime_get_sync(adev_to_drm(adev)->dev);
+	if (r < 0) {
+		drm_file_err(uq_mgr->file, "pm_runtime_get_sync() failed for userqueue create\n");
+		pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
+		return r;
+	}
+
+	/*
+	 * There could be a situation that we are creating a new queue while
+	 * the other queues under this UQ_mgr are suspended. So if there is any
+	 * resume work pending, wait for it to get done.
+	 *
+	 * This will also make sure we have a valid eviction fence ready to be used.
+	 */
+	amdgpu_userq_ensure_ev_fence(&fpriv->userq_mgr, &fpriv->evf_mgr);
+
+	uq_funcs = adev->userq_funcs[args->in.ip_type];
+	if (!uq_funcs) {
+		drm_file_err(uq_mgr->file, "Usermode queue is not supported for this IP (%u)\n",
+			     args->in.ip_type);
+		r = -EINVAL;
+		goto unlock;
+	}
+
+	queue = kzalloc(sizeof(struct amdgpu_usermode_queue), GFP_KERNEL);
+	if (!queue) {
+		drm_file_err(uq_mgr->file, "Failed to allocate memory for queue\n");
+		r = -ENOMEM;
+		goto unlock;
+	}
+
+	INIT_LIST_HEAD(&queue->userq_va_list);
+	queue->doorbell_handle = args->in.doorbell_handle;
+	queue->queue_type = args->in.ip_type;
+	queue->vm = &fpriv->vm;
+	queue->priority = priority;
+
+	db_info.queue_type = queue->queue_type;
+	db_info.doorbell_handle = queue->doorbell_handle;
+	db_info.db_obj = &queue->db_obj;
+	db_info.doorbell_offset = args->in.doorbell_offset;
+
+	/* Validate the userq virtual address.*/
+	if (amdgpu_userq_input_va_validate(queue, args->in.queue_va, args->in.queue_size) ||
+	    amdgpu_userq_input_va_validate(queue, args->in.rptr_va, AMDGPU_GPU_PAGE_SIZE) ||
+	    amdgpu_userq_input_va_validate(queue, args->in.wptr_va, AMDGPU_GPU_PAGE_SIZE)) {
+		r = -EINVAL;
+		kfree(queue);
+		goto unlock;
+	}
+
+	/* Convert relative doorbell offset into absolute doorbell index */
+	index = amdgpu_userq_get_doorbell_index(uq_mgr, &db_info, filp);
+	if (index == (uint64_t)-EINVAL) {
+		drm_file_err(uq_mgr->file, "Failed to get doorbell for queue\n");
+		kfree(queue);
+		r = -EINVAL;
+		goto unlock;
+	}
+
+	queue->doorbell_index = index;
+	xa_init_flags(&queue->fence_drv_xa, XA_FLAGS_ALLOC);
+	r = amdgpu_userq_fence_driver_alloc(adev, queue);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to alloc fence driver\n");
+		goto unlock;
+	}
+
+	r = uq_funcs->mqd_create(uq_mgr, &args->in, queue);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to create Queue\n");
+		amdgpu_userq_fence_driver_free(queue);
+		kfree(queue);
+		goto unlock;
+	}
+
+	/* Wait for mode-1 reset to complete */
+	down_read(&adev->reset_domain->sem);
+	r = xa_err(xa_store_irq(&adev->userq_doorbell_xa, index, queue, GFP_KERNEL));
+	if (r) {
+		kfree(queue);
+		up_read(&adev->reset_domain->sem);
+		goto unlock;
+	}
+
+	r = xa_alloc(&uq_mgr->userq_mgr_xa, &qid, queue, XA_LIMIT(1, AMDGPU_MAX_USERQ_COUNT), GFP_KERNEL);
+	if (r) {
+		drm_file_err(uq_mgr->file, "Failed to allocate a queue id\n");
+		amdgpu_userq_fence_driver_free(queue);
+		uq_funcs->mqd_destroy(uq_mgr, queue);
+		kfree(queue);
+		r = -ENOMEM;
+		up_read(&adev->reset_domain->sem);
+		goto unlock;
+	}
+	up_read(&adev->reset_domain->sem);
+	queue->userq_mgr = uq_mgr;
+
+	/* don't map the queue if scheduling is halted */
+	if (adev->userq_halt_for_enforce_isolation &&
+	    ((queue->queue_type == AMDGPU_HW_IP_GFX) ||
+	     (queue->queue_type == AMDGPU_HW_IP_COMPUTE)))
+		skip_map_queue = true;
+	else
+		skip_map_queue = false;
+	if (!skip_map_queue) {
+		r = amdgpu_userq_map_helper(uq_mgr, queue);
+		if (r) {
+			drm_file_err(uq_mgr->file, "Failed to map Queue\n");
+			xa_erase(&uq_mgr->userq_mgr_xa, qid);
+			amdgpu_userq_fence_driver_free(queue);
+			uq_funcs->mqd_destroy(uq_mgr, queue);
+			kfree(queue);
+			goto unlock;
+		}
+	}
+
+	queue_name = kasprintf(GFP_KERNEL, "queue-%d", qid);
+	if (!queue_name) {
+		r = -ENOMEM;
+		goto unlock;
+	}
+
+#if defined(CONFIG_DEBUG_FS)
+	/* Queue dentry per client to hold MQD information   */
+	queue->debugfs_queue = debugfs_create_dir(queue_name, filp->debugfs_client);
+	debugfs_create_file("mqd_info", 0444, queue->debugfs_queue, queue, &amdgpu_mqd_info_fops);
+#endif
+	kfree(queue_name);
+
+	args->out.queue_id = qid;
+	atomic_inc(&uq_mgr->userq_count[queue->queue_type]);
+
+unlock:
+	mutex_unlock(&uq_mgr->userq_mutex);
+
+	return r;
+}
+
+static int amdgpu_userq_input_args_validate(struct drm_device *dev,
+					union drm_amdgpu_userq *args,
+					struct drm_file *filp)
+{
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	switch (args->in.op) {
+	case AMDGPU_USERQ_OP_CREATE:
+		if (args->in.flags & ~(AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_MASK |
+				       AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE))
+			return -EINVAL;
+		/* Usermode queues are only supported for GFX IP as of now */
+		if (args->in.ip_type != AMDGPU_HW_IP_GFX &&
+		    args->in.ip_type != AMDGPU_HW_IP_DMA &&
+		    args->in.ip_type != AMDGPU_HW_IP_COMPUTE) {
+			drm_file_err(filp, "Usermode queue doesn't support IP type %u\n",
+				     args->in.ip_type);
+			return -EINVAL;
+		}
+
+		if ((args->in.flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE) &&
+		    (args->in.ip_type != AMDGPU_HW_IP_GFX) &&
+		    (args->in.ip_type != AMDGPU_HW_IP_COMPUTE) &&
+		    !amdgpu_is_tmz(adev)) {
+			drm_file_err(filp, "Secure only supported on GFX/Compute queues\n");
+			return -EINVAL;
+		}
+
+		if (args->in.queue_va == AMDGPU_BO_INVALID_OFFSET ||
+		    args->in.queue_va == 0 ||
+		    args->in.queue_size == 0) {
+			drm_file_err(filp, "invalidate userq queue va or size\n");
+			return -EINVAL;
+		}
+		if (!args->in.wptr_va || !args->in.rptr_va) {
+			drm_file_err(filp, "invalidate userq queue rptr or wptr\n");
+			return -EINVAL;
+		}
+		break;
+	case AMDGPU_USERQ_OP_FREE:
+		if (args->in.ip_type ||
+		    args->in.doorbell_handle ||
+		    args->in.doorbell_offset ||
+		    args->in.flags ||
+		    args->in.queue_va ||
+		    args->in.queue_size ||
+		    args->in.rptr_va ||
+		    args->in.wptr_va ||
+		    args->in.mqd ||
+		    args->in.mqd_size)
+			return -EINVAL;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int amdgpu_userq_ioctl(struct drm_device *dev, void *data,
+		       struct drm_file *filp)
+{
+	union drm_amdgpu_userq *args = data;
+	int r;
+
+	if (amdgpu_userq_input_args_validate(dev, args, filp) < 0)
+		return -EINVAL;
+
+	switch (args->in.op) {
+	case AMDGPU_USERQ_OP_CREATE:
+		r = amdgpu_userq_create(filp, args);
+		if (r)
+			drm_file_err(filp, "Failed to create usermode queue\n");
+		break;
+
+	case AMDGPU_USERQ_OP_FREE:
+		r = amdgpu_userq_destroy(filp, args->in.queue_id);
+		if (r)
+			drm_file_err(filp, "Failed to destroy usermode queue\n");
+		break;
+
+	default:
+		drm_dbg_driver(dev, "Invalid user queue op specified: %d\n", args->in.op);
+		return -EINVAL;
+	}
+
+	return r;
+}
+
+static int
+amdgpu_userq_restore_all(struct amdgpu_userq_mgr *uq_mgr)
+{
+	struct amdgpu_usermode_queue *queue;
+	unsigned long queue_id;
+	int ret = 0, r;
+
+	/* Resume all the queues for this process */
+	xa_for_each(&uq_mgr->userq_mgr_xa, queue_id, queue) {
+
+		if (!amdgpu_userq_buffer_vas_mapped(queue)) {
+			drm_file_err(uq_mgr->file,
+				     "trying restore queue without va mapping\n");
+			queue->state = AMDGPU_USERQ_STATE_INVALID_VA;
+			continue;
+		}
+
+		r = amdgpu_userq_restore_helper(uq_mgr, queue);
+		if (r)
+			ret = r;
+	}
+
+	if (ret)
+		drm_file_err(uq_mgr->file, "Failed to map all the queues\n");
+	return ret;
+}
+
+static int amdgpu_userq_validate_vm(void *param, struct amdgpu_bo *bo)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+
+	amdgpu_bo_placement_from_domain(bo, bo->allowed_domains);
+	return ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+}
+
+/* Handle all BOs on the invalidated list, validate them and update the PTs */
+static int
+amdgpu_userq_bo_validate(struct amdgpu_device *adev, struct drm_exec *exec,
+			 struct amdgpu_vm *vm)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo *bo;
+	int ret;
+
+	spin_lock(&vm->status_lock);
+	while (!list_empty(&vm->invalidated)) {
+		bo_va = list_first_entry(&vm->invalidated,
+					 struct amdgpu_bo_va,
+					 base.vm_status);
+		spin_unlock(&vm->status_lock);
+
+		bo = bo_va->base.bo;
+		ret = drm_exec_prepare_obj(exec, &bo->tbo.base, 2);
+		if (unlikely(ret))
+			return ret;
+
+		amdgpu_bo_placement_from_domain(bo, bo->allowed_domains);
+		ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+		if (ret)
+			return ret;
+
+		/* This moves the bo_va to the done list */
+		ret = amdgpu_vm_bo_update(adev, bo_va, false);
+		if (ret)
+			return ret;
+
+		spin_lock(&vm->status_lock);
+	}
+	spin_unlock(&vm->status_lock);
+
+	return 0;
+}
+
+/* Make sure the whole VM is ready to be used */
+static int
+amdgpu_userq_vm_validate(struct amdgpu_userq_mgr *uq_mgr)
+{
+	struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr);
+	bool invalidated = false, new_addition = false;
+	struct ttm_operation_ctx ctx = { true, false };
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct amdgpu_hmm_range *range;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	unsigned long key, tmp_key;
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo *bo;
+	struct drm_exec exec;
+	struct xarray xa;
+	int ret;
+
+	xa_init(&xa);
+
+retry_lock:
+	drm_exec_init(&exec, DRM_EXEC_IGNORE_DUPLICATES, 0);
+	drm_exec_until_all_locked(&exec) {
+		ret = amdgpu_vm_lock_pd(vm, &exec, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(ret))
+			goto unlock_all;
+
+		ret = amdgpu_vm_lock_done_list(vm, &exec, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(ret))
+			goto unlock_all;
+
+		/* This validates PDs, PTs and per VM BOs */
+		ret = amdgpu_vm_validate(adev, vm, NULL,
+					 amdgpu_userq_validate_vm,
+					 NULL);
+		if (unlikely(ret))
+			goto unlock_all;
+
+		/* This locks and validates the remaining evicted BOs */
+		ret = amdgpu_userq_bo_validate(adev, &exec, vm);
+		drm_exec_retry_on_contention(&exec);
+		if (unlikely(ret))
+			goto unlock_all;
+	}
+
+	if (invalidated) {
+		xa_for_each(&xa, tmp_key, range) {
+			bo = range->bo;
+			amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU);
+			ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+			if (ret)
+				goto unlock_all;
+
+			amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, range);
+
+			amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
+			ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+			if (ret)
+				goto unlock_all;
+		}
+		invalidated = false;
+	}
+
+	ret = amdgpu_vm_handle_moved(adev, vm, NULL);
+	if (ret)
+		goto unlock_all;
+
+	key = 0;
+	/* Validate User Ptr BOs */
+	list_for_each_entry(bo_va, &vm->done, base.vm_status) {
+		bo = bo_va->base.bo;
+		if (!bo)
+			continue;
+
+		if (!amdgpu_ttm_tt_is_userptr(bo->tbo.ttm))
+			continue;
+
+		range = xa_load(&xa, key);
+		if (range && range->bo != bo) {
+			xa_erase(&xa, key);
+			amdgpu_hmm_range_free(range);
+			range = NULL;
+		}
+
+		if (!range) {
+			range = amdgpu_hmm_range_alloc(bo);
+			if (!range) {
+				ret = -ENOMEM;
+				goto unlock_all;
+			}
+
+			xa_store(&xa, key, range, GFP_KERNEL);
+			new_addition = true;
+		}
+		key++;
+	}
+
+	if (new_addition) {
+		drm_exec_fini(&exec);
+		xa_for_each(&xa, tmp_key, range) {
+			if (!range)
+				continue;
+			bo = range->bo;
+			ret = amdgpu_ttm_tt_get_user_pages(bo, range);
+			if (ret)
+				goto unlock_all;
+		}
+
+		invalidated = true;
+		new_addition = false;
+		goto retry_lock;
+	}
+
+	ret = amdgpu_vm_update_pdes(adev, vm, false);
+	if (ret)
+		goto unlock_all;
+
+	/*
+	 * We need to wait for all VM updates to finish before restarting the
+	 * queues. Using the done list like that is now ok since everything is
+	 * locked in place.
+	 */
+	list_for_each_entry(bo_va, &vm->done, base.vm_status)
+		dma_fence_wait(bo_va->last_pt_update, false);
+	dma_fence_wait(vm->last_update, false);
+
+	ret = amdgpu_eviction_fence_replace_fence(&fpriv->evf_mgr, &exec);
+	if (ret)
+		drm_file_err(uq_mgr->file, "Failed to replace eviction fence\n");
+
+unlock_all:
+	drm_exec_fini(&exec);
+	xa_for_each(&xa, tmp_key, range) {
+		if (!range)
+			continue;
+		bo = range->bo;
+		amdgpu_hmm_range_free(range);
+	}
+	xa_destroy(&xa);
+	return ret;
+}
+
+static void amdgpu_userq_restore_worker(struct work_struct *work)
+{
+	struct amdgpu_userq_mgr *uq_mgr = work_to_uq_mgr(work, resume_work.work);
+	struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr);
+	int ret;
+
+	flush_delayed_work(&fpriv->evf_mgr.suspend_work);
+
+	mutex_lock(&uq_mgr->userq_mutex);
+
+	ret = amdgpu_userq_vm_validate(uq_mgr);
+	if (ret) {
+		drm_file_err(uq_mgr->file, "Failed to validate BOs to restore\n");
+		goto unlock;
+	}
+
+	ret = amdgpu_userq_restore_all(uq_mgr);
+	if (ret) {
+		drm_file_err(uq_mgr->file, "Failed to restore all queues\n");
+		goto unlock;
+	}
+
+unlock:
+	mutex_unlock(&uq_mgr->userq_mutex);
+}
+
+static int
+amdgpu_userq_evict_all(struct amdgpu_userq_mgr *uq_mgr)
+{
+	struct amdgpu_usermode_queue *queue;
+	unsigned long queue_id;
+	int ret = 0, r;
+
+	amdgpu_userq_detect_and_reset_queues(uq_mgr);
+	/* Try to unmap all the queues in this process ctx */
+	xa_for_each(&uq_mgr->userq_mgr_xa, queue_id, queue) {
+		r = amdgpu_userq_preempt_helper(uq_mgr, queue);
+		if (r)
+			ret = r;
+	}
+
+	if (ret)
+		drm_file_err(uq_mgr->file, "Couldn't unmap all the queues\n");
+	return ret;
+}
+
+void amdgpu_userq_reset_work(struct work_struct *work)
+{
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  userq_reset_work);
+	struct amdgpu_reset_context reset_context;
+
+	memset(&reset_context, 0, sizeof(reset_context));
+
+	reset_context.method = AMD_RESET_METHOD_NONE;
+	reset_context.reset_req_dev = adev;
+	reset_context.src = AMDGPU_RESET_SRC_USERQ;
+	set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+	/*set_bit(AMDGPU_SKIP_COREDUMP, &reset_context.flags);*/
+
+	amdgpu_device_gpu_recover(adev, NULL, &reset_context);
+}
+
+static int
+amdgpu_userq_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr)
+{
+	struct amdgpu_usermode_queue *queue;
+	unsigned long queue_id;
+	int ret;
+
+	xa_for_each(&uq_mgr->userq_mgr_xa, queue_id, queue) {
+		struct dma_fence *f = queue->last_fence;
+
+		if (!f || dma_fence_is_signaled(f))
+			continue;
+		ret = dma_fence_wait_timeout(f, true, msecs_to_jiffies(100));
+		if (ret <= 0) {
+			drm_file_err(uq_mgr->file, "Timed out waiting for fence=%llu:%llu\n",
+				     f->context, f->seqno);
+			return -ETIMEDOUT;
+		}
+	}
+
+	return 0;
+}
+
+void
+amdgpu_userq_evict(struct amdgpu_userq_mgr *uq_mgr,
+		   struct amdgpu_eviction_fence *ev_fence)
+{
+	struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr);
+	struct amdgpu_eviction_fence_mgr *evf_mgr = &fpriv->evf_mgr;
+	struct amdgpu_device *adev = uq_mgr->adev;
+	int ret;
+
+	/* Wait for any pending userqueue fence work to finish */
+	ret = amdgpu_userq_wait_for_signal(uq_mgr);
+	if (ret)
+		dev_err(adev->dev, "Not evicting userqueue, timeout waiting for work\n");
+
+	ret = amdgpu_userq_evict_all(uq_mgr);
+	if (ret)
+		dev_err(adev->dev, "Failed to evict userqueue\n");
+
+	/* Signal current eviction fence */
+	amdgpu_eviction_fence_signal(evf_mgr, ev_fence);
+
+	if (evf_mgr->fd_closing) {
+		cancel_delayed_work_sync(&uq_mgr->resume_work);
+		return;
+	}
+
+	/* Schedule a resume work */
+	schedule_delayed_work(&uq_mgr->resume_work, 0);
+}
+
+int amdgpu_userq_mgr_init(struct amdgpu_userq_mgr *userq_mgr, struct drm_file *file_priv,
+			  struct amdgpu_device *adev)
+{
+	mutex_init(&userq_mgr->userq_mutex);
+	xa_init_flags(&userq_mgr->userq_mgr_xa, XA_FLAGS_ALLOC);
+	userq_mgr->adev = adev;
+	userq_mgr->file = file_priv;
+
+	INIT_DELAYED_WORK(&userq_mgr->resume_work, amdgpu_userq_restore_worker);
+	return 0;
+}
+
+void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr)
+{
+	struct amdgpu_usermode_queue *queue;
+	unsigned long queue_id;
+
+	cancel_delayed_work_sync(&userq_mgr->resume_work);
+
+	mutex_lock(&userq_mgr->userq_mutex);
+	amdgpu_userq_detect_and_reset_queues(userq_mgr);
+	xa_for_each(&userq_mgr->userq_mgr_xa, queue_id, queue) {
+		amdgpu_userq_wait_for_last_fence(userq_mgr, queue);
+		amdgpu_userq_unmap_helper(userq_mgr, queue);
+		amdgpu_userq_cleanup(userq_mgr, queue, queue_id);
+	}
+
+	xa_destroy(&userq_mgr->userq_mgr_xa);
+	mutex_unlock(&userq_mgr->userq_mutex);
+	mutex_destroy(&userq_mgr->userq_mutex);
+}
+
+int amdgpu_userq_suspend(struct amdgpu_device *adev)
+{
+	u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm;
+	unsigned long queue_id;
+	int r;
+
+	if (!ip_mask)
+		return 0;
+
+	xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) {
+		uqm = queue->userq_mgr;
+		cancel_delayed_work_sync(&uqm->resume_work);
+		guard(mutex)(&uqm->userq_mutex);
+		amdgpu_userq_detect_and_reset_queues(uqm);
+		if (adev->in_s0ix)
+			r = amdgpu_userq_preempt_helper(uqm, queue);
+		else
+			r = amdgpu_userq_unmap_helper(uqm, queue);
+		if (r)
+			return r;
+	}
+	return 0;
+}
+
+int amdgpu_userq_resume(struct amdgpu_device *adev)
+{
+	u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm;
+	unsigned long queue_id;
+	int r;
+
+	if (!ip_mask)
+		return 0;
+
+	xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) {
+		uqm = queue->userq_mgr;
+		guard(mutex)(&uqm->userq_mutex);
+		if (adev->in_s0ix)
+			r = amdgpu_userq_restore_helper(uqm, queue);
+		else
+			r = amdgpu_userq_map_helper(uqm, queue);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+int amdgpu_userq_stop_sched_for_enforce_isolation(struct amdgpu_device *adev,
+						  u32 idx)
+{
+	u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm;
+	unsigned long queue_id;
+	int ret = 0, r;
+
+	/* only need to stop gfx/compute */
+	if (!(ip_mask & ((1 << AMDGPU_HW_IP_GFX) | (1 << AMDGPU_HW_IP_COMPUTE))))
+		return 0;
+
+	if (adev->userq_halt_for_enforce_isolation)
+		dev_warn(adev->dev, "userq scheduling already stopped!\n");
+	adev->userq_halt_for_enforce_isolation = true;
+	xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) {
+		uqm = queue->userq_mgr;
+		cancel_delayed_work_sync(&uqm->resume_work);
+		mutex_lock(&uqm->userq_mutex);
+		if (((queue->queue_type == AMDGPU_HW_IP_GFX) ||
+		     (queue->queue_type == AMDGPU_HW_IP_COMPUTE)) &&
+		    (queue->xcp_id == idx)) {
+			amdgpu_userq_detect_and_reset_queues(uqm);
+			r = amdgpu_userq_preempt_helper(uqm, queue);
+			if (r)
+				ret = r;
+		}
+		mutex_unlock(&uqm->userq_mutex);
+	}
+
+	return ret;
+}
+
+int amdgpu_userq_start_sched_for_enforce_isolation(struct amdgpu_device *adev,
+						   u32 idx)
+{
+	u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm;
+	unsigned long queue_id;
+	int ret = 0, r;
+
+	/* only need to stop gfx/compute */
+	if (!(ip_mask & ((1 << AMDGPU_HW_IP_GFX) | (1 << AMDGPU_HW_IP_COMPUTE))))
+		return 0;
+
+	if (!adev->userq_halt_for_enforce_isolation)
+		dev_warn(adev->dev, "userq scheduling already started!\n");
+	adev->userq_halt_for_enforce_isolation = false;
+	xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) {
+		uqm = queue->userq_mgr;
+		mutex_lock(&uqm->userq_mutex);
+			if (((queue->queue_type == AMDGPU_HW_IP_GFX) ||
+			     (queue->queue_type == AMDGPU_HW_IP_COMPUTE)) &&
+			    (queue->xcp_id == idx)) {
+				r = amdgpu_userq_restore_helper(uqm, queue);
+				if (r)
+					ret = r;
+			}
+		mutex_unlock(&uqm->userq_mutex);
+	}
+
+	return ret;
+}
+
+int amdgpu_userq_gem_va_unmap_validate(struct amdgpu_device *adev,
+				       struct amdgpu_bo_va_mapping *mapping,
+				       uint64_t saddr)
+{
+	u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev);
+	struct amdgpu_bo_va *bo_va = mapping->bo_va;
+	struct dma_resv *resv = bo_va->base.bo->tbo.base.resv;
+	int ret = 0;
+
+	if (!ip_mask)
+		return 0;
+
+	dev_warn_once(adev->dev, "now unmapping a vital queue va:%llx\n", saddr);
+	/**
+	 * The userq VA mapping reservation should include the eviction fence,
+	 * if the eviction fence can't signal successfully during unmapping,
+	 * then driver will warn to flag this improper unmap of the userq VA.
+	 * Note: The eviction fence may be attached to different BOs, and this
+	 * unmap is only for one kind of userq VAs, so at this point suppose
+	 * the eviction fence is always unsignaled.
+	 */
+	if (!dma_resv_test_signaled(resv, DMA_RESV_USAGE_BOOKKEEP)) {
+		ret = dma_resv_wait_timeout(resv, DMA_RESV_USAGE_BOOKKEEP, true,
+					    MAX_SCHEDULE_TIMEOUT);
+		if (ret <= 0)
+			return -EBUSY;
+	}
+
+	return 0;
+}
+
+void amdgpu_userq_pre_reset(struct amdgpu_device *adev)
+{
+	const struct amdgpu_userq_funcs *userq_funcs;
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_mgr *uqm;
+	unsigned long queue_id;
+
+	xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) {
+		uqm = queue->userq_mgr;
+		cancel_delayed_work_sync(&uqm->resume_work);
+		if (queue->state == AMDGPU_USERQ_STATE_MAPPED) {
+			amdgpu_userq_wait_for_last_fence(uqm, queue);
+			userq_funcs = adev->userq_funcs[queue->queue_type];
+			userq_funcs->unmap(uqm, queue);
+			/* just mark all queues as hung at this point.
+			 * if unmap succeeds, we could map again
+			 * in amdgpu_userq_post_reset() if vram is not lost
+			 */
+			queue->state = AMDGPU_USERQ_STATE_HUNG;
+			amdgpu_userq_fence_driver_force_completion(queue);
+		}
+	}
+}
+
+int amdgpu_userq_post_reset(struct amdgpu_device *adev, bool vram_lost)
+{
+	/* if any queue state is AMDGPU_USERQ_STATE_UNMAPPED
+	 * at this point, we should be able to map it again
+	 * and continue if vram is not lost.
+	 */
+	struct amdgpu_userq_mgr *uqm;
+	struct amdgpu_usermode_queue *queue;
+	const struct amdgpu_userq_funcs *userq_funcs;
+	unsigned long queue_id;
+	int r = 0;
+
+	xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) {
+		uqm = queue->userq_mgr;
+		if (queue->state == AMDGPU_USERQ_STATE_HUNG && !vram_lost) {
+			userq_funcs = adev->userq_funcs[queue->queue_type];
+			/* Re-map queue */
+			r = userq_funcs->map(uqm, queue);
+			if (r) {
+				dev_err(adev->dev, "Failed to remap queue %ld\n", queue_id);
+				continue;
+			}
+			queue->state = AMDGPU_USERQ_STATE_MAPPED;
+		}
+	}
+
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h
new file mode 100644
index 000000000000..c37444427a14
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h
@@ -0,0 +1,161 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef AMDGPU_USERQ_H_
+#define AMDGPU_USERQ_H_
+#include "amdgpu_eviction_fence.h"
+
+#define AMDGPU_MAX_USERQ_COUNT 512
+
+#define to_ev_fence(f) container_of(f, struct amdgpu_eviction_fence, base)
+#define uq_mgr_to_fpriv(u) container_of(u, struct amdgpu_fpriv, userq_mgr)
+#define work_to_uq_mgr(w, name) container_of(w, struct amdgpu_userq_mgr, name)
+
+enum amdgpu_userq_state {
+	AMDGPU_USERQ_STATE_UNMAPPED = 0,
+	AMDGPU_USERQ_STATE_MAPPED,
+	AMDGPU_USERQ_STATE_PREEMPTED,
+	AMDGPU_USERQ_STATE_HUNG,
+	AMDGPU_USERQ_STATE_INVALID_VA,
+};
+
+struct amdgpu_mqd_prop;
+
+struct amdgpu_userq_obj {
+	void		 *cpu_ptr;
+	uint64_t	 gpu_addr;
+	struct amdgpu_bo *obj;
+};
+
+struct amdgpu_userq_va_cursor {
+	u64			gpu_addr;
+	struct list_head	list;
+};
+
+struct amdgpu_usermode_queue {
+	int			queue_type;
+	enum amdgpu_userq_state state;
+	uint64_t		doorbell_handle;
+	uint64_t		doorbell_index;
+	uint64_t		flags;
+	struct amdgpu_mqd_prop	*userq_prop;
+	struct amdgpu_userq_mgr *userq_mgr;
+	struct amdgpu_vm	*vm;
+	struct amdgpu_userq_obj mqd;
+	struct amdgpu_userq_obj	db_obj;
+	struct amdgpu_userq_obj fw_obj;
+	struct amdgpu_userq_obj wptr_obj;
+	struct xarray		fence_drv_xa;
+	struct amdgpu_userq_fence_driver *fence_drv;
+	struct dma_fence	*last_fence;
+	u32			xcp_id;
+	int			priority;
+	struct dentry		*debugfs_queue;
+
+	struct list_head	userq_va_list;
+};
+
+struct amdgpu_userq_funcs {
+	int (*mqd_create)(struct amdgpu_userq_mgr *uq_mgr,
+			  struct drm_amdgpu_userq_in *args,
+			  struct amdgpu_usermode_queue *queue);
+	void (*mqd_destroy)(struct amdgpu_userq_mgr *uq_mgr,
+			    struct amdgpu_usermode_queue *uq);
+	int (*unmap)(struct amdgpu_userq_mgr *uq_mgr,
+		     struct amdgpu_usermode_queue *queue);
+	int (*map)(struct amdgpu_userq_mgr *uq_mgr,
+		   struct amdgpu_usermode_queue *queue);
+	int (*preempt)(struct amdgpu_userq_mgr *uq_mgr,
+		   struct amdgpu_usermode_queue *queue);
+	int (*restore)(struct amdgpu_userq_mgr *uq_mgr,
+		   struct amdgpu_usermode_queue *queue);
+	int (*detect_and_reset)(struct amdgpu_device *adev,
+		  int queue_type);
+};
+
+/* Usermode queues for gfx */
+struct amdgpu_userq_mgr {
+	/**
+	 * @userq_mgr_xa: Per-process user queue map (queue ID → queue)
+	 * Key: queue_id (unique ID within the process's userq manager)
+	 * Value: struct amdgpu_usermode_queue
+	 */
+	struct xarray			userq_mgr_xa;
+	struct mutex			userq_mutex;
+	struct amdgpu_device		*adev;
+	struct delayed_work		resume_work;
+	struct drm_file			*file;
+	atomic_t                        userq_count[AMDGPU_RING_TYPE_MAX];
+};
+
+struct amdgpu_db_info {
+	uint64_t doorbell_handle;
+	uint32_t queue_type;
+	uint32_t doorbell_offset;
+	struct amdgpu_userq_obj	*db_obj;
+};
+
+int amdgpu_userq_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
+
+int amdgpu_userq_mgr_init(struct amdgpu_userq_mgr *userq_mgr, struct drm_file *file_priv,
+			  struct amdgpu_device *adev);
+
+void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr);
+
+int amdgpu_userq_create_object(struct amdgpu_userq_mgr *uq_mgr,
+			       struct amdgpu_userq_obj *userq_obj,
+			       int size);
+
+void amdgpu_userq_destroy_object(struct amdgpu_userq_mgr *uq_mgr,
+				 struct amdgpu_userq_obj *userq_obj);
+
+void amdgpu_userq_evict(struct amdgpu_userq_mgr *uq_mgr,
+			struct amdgpu_eviction_fence *ev_fence);
+
+void amdgpu_userq_ensure_ev_fence(struct amdgpu_userq_mgr *userq_mgr,
+				  struct amdgpu_eviction_fence_mgr *evf_mgr);
+
+uint64_t amdgpu_userq_get_doorbell_index(struct amdgpu_userq_mgr *uq_mgr,
+					 struct amdgpu_db_info *db_info,
+					     struct drm_file *filp);
+
+u32 amdgpu_userq_get_supported_ip_mask(struct amdgpu_device *adev);
+
+int amdgpu_userq_suspend(struct amdgpu_device *adev);
+int amdgpu_userq_resume(struct amdgpu_device *adev);
+
+int amdgpu_userq_stop_sched_for_enforce_isolation(struct amdgpu_device *adev,
+						  u32 idx);
+int amdgpu_userq_start_sched_for_enforce_isolation(struct amdgpu_device *adev,
+						   u32 idx);
+void amdgpu_userq_reset_work(struct work_struct *work);
+void amdgpu_userq_pre_reset(struct amdgpu_device *adev);
+int amdgpu_userq_post_reset(struct amdgpu_device *adev, bool vram_lost);
+
+int amdgpu_userq_input_va_validate(struct amdgpu_usermode_queue *queue,
+				   u64 addr, u64 expected_size);
+int amdgpu_userq_gem_va_unmap_validate(struct amdgpu_device *adev,
+				       struct amdgpu_bo_va_mapping *mapping,
+				       uint64_t saddr);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c
new file mode 100644
index 000000000000..eba9fb359047
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c
@@ -0,0 +1,1011 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/kref.h>
+#include <linux/slab.h>
+#include <linux/dma-fence-unwrap.h>
+
+#include <drm/drm_exec.h>
+#include <drm/drm_syncobj.h>
+
+#include "amdgpu.h"
+#include "amdgpu_userq_fence.h"
+
+static const struct dma_fence_ops amdgpu_userq_fence_ops;
+static struct kmem_cache *amdgpu_userq_fence_slab;
+
+int amdgpu_userq_fence_slab_init(void)
+{
+	amdgpu_userq_fence_slab = kmem_cache_create("amdgpu_userq_fence",
+						    sizeof(struct amdgpu_userq_fence),
+						    0,
+						    SLAB_HWCACHE_ALIGN,
+						    NULL);
+	if (!amdgpu_userq_fence_slab)
+		return -ENOMEM;
+
+	return 0;
+}
+
+void amdgpu_userq_fence_slab_fini(void)
+{
+	rcu_barrier();
+	kmem_cache_destroy(amdgpu_userq_fence_slab);
+}
+
+static inline struct amdgpu_userq_fence *to_amdgpu_userq_fence(struct dma_fence *f)
+{
+	if (!f || f->ops != &amdgpu_userq_fence_ops)
+		return NULL;
+
+	return container_of(f, struct amdgpu_userq_fence, base);
+}
+
+static u64 amdgpu_userq_fence_read(struct amdgpu_userq_fence_driver *fence_drv)
+{
+	return le64_to_cpu(*fence_drv->cpu_addr);
+}
+
+static void
+amdgpu_userq_fence_write(struct amdgpu_userq_fence_driver *fence_drv,
+			 u64 seq)
+{
+	if (fence_drv->cpu_addr)
+		*fence_drv->cpu_addr = cpu_to_le64(seq);
+}
+
+int amdgpu_userq_fence_driver_alloc(struct amdgpu_device *adev,
+				    struct amdgpu_usermode_queue *userq)
+{
+	struct amdgpu_userq_fence_driver *fence_drv;
+	unsigned long flags;
+	int r;
+
+	fence_drv = kzalloc(sizeof(*fence_drv), GFP_KERNEL);
+	if (!fence_drv)
+		return -ENOMEM;
+
+	/* Acquire seq64 memory */
+	r = amdgpu_seq64_alloc(adev, &fence_drv->va, &fence_drv->gpu_addr,
+			       &fence_drv->cpu_addr);
+	if (r)
+		goto free_fence_drv;
+
+	memset(fence_drv->cpu_addr, 0, sizeof(u64));
+
+	kref_init(&fence_drv->refcount);
+	INIT_LIST_HEAD(&fence_drv->fences);
+	spin_lock_init(&fence_drv->fence_list_lock);
+
+	fence_drv->adev = adev;
+	fence_drv->context = dma_fence_context_alloc(1);
+	get_task_comm(fence_drv->timeline_name, current);
+
+	xa_lock_irqsave(&adev->userq_xa, flags);
+	r = xa_err(__xa_store(&adev->userq_xa, userq->doorbell_index,
+			      fence_drv, GFP_KERNEL));
+	xa_unlock_irqrestore(&adev->userq_xa, flags);
+	if (r)
+		goto free_seq64;
+
+	userq->fence_drv = fence_drv;
+
+	return 0;
+
+free_seq64:
+	amdgpu_seq64_free(adev, fence_drv->va);
+free_fence_drv:
+	kfree(fence_drv);
+
+	return r;
+}
+
+static void amdgpu_userq_walk_and_drop_fence_drv(struct xarray *xa)
+{
+	struct amdgpu_userq_fence_driver *fence_drv;
+	unsigned long index;
+
+	if (xa_empty(xa))
+		return;
+
+	xa_lock(xa);
+	xa_for_each(xa, index, fence_drv) {
+		__xa_erase(xa, index);
+		amdgpu_userq_fence_driver_put(fence_drv);
+	}
+
+	xa_unlock(xa);
+}
+
+void
+amdgpu_userq_fence_driver_free(struct amdgpu_usermode_queue *userq)
+{
+	amdgpu_userq_walk_and_drop_fence_drv(&userq->fence_drv_xa);
+	xa_destroy(&userq->fence_drv_xa);
+	/* Drop the fence_drv reference held by user queue */
+	amdgpu_userq_fence_driver_put(userq->fence_drv);
+}
+
+void amdgpu_userq_fence_driver_process(struct amdgpu_userq_fence_driver *fence_drv)
+{
+	struct amdgpu_userq_fence *userq_fence, *tmp;
+	struct dma_fence *fence;
+	unsigned long flags;
+	u64 rptr;
+	int i;
+
+	if (!fence_drv)
+		return;
+
+	spin_lock_irqsave(&fence_drv->fence_list_lock, flags);
+	rptr = amdgpu_userq_fence_read(fence_drv);
+
+	list_for_each_entry_safe(userq_fence, tmp, &fence_drv->fences, link) {
+		fence = &userq_fence->base;
+
+		if (rptr < fence->seqno)
+			break;
+
+		dma_fence_signal(fence);
+
+		for (i = 0; i < userq_fence->fence_drv_array_count; i++)
+			amdgpu_userq_fence_driver_put(userq_fence->fence_drv_array[i]);
+
+		list_del(&userq_fence->link);
+		dma_fence_put(fence);
+	}
+	spin_unlock_irqrestore(&fence_drv->fence_list_lock, flags);
+}
+
+void amdgpu_userq_fence_driver_destroy(struct kref *ref)
+{
+	struct amdgpu_userq_fence_driver *fence_drv = container_of(ref,
+					 struct amdgpu_userq_fence_driver,
+					 refcount);
+	struct amdgpu_userq_fence_driver *xa_fence_drv;
+	struct amdgpu_device *adev = fence_drv->adev;
+	struct amdgpu_userq_fence *fence, *tmp;
+	struct xarray *xa = &adev->userq_xa;
+	unsigned long index, flags;
+	struct dma_fence *f;
+
+	spin_lock_irqsave(&fence_drv->fence_list_lock, flags);
+	list_for_each_entry_safe(fence, tmp, &fence_drv->fences, link) {
+		f = &fence->base;
+
+		if (!dma_fence_is_signaled(f)) {
+			dma_fence_set_error(f, -ECANCELED);
+			dma_fence_signal(f);
+		}
+
+		list_del(&fence->link);
+		dma_fence_put(f);
+	}
+	spin_unlock_irqrestore(&fence_drv->fence_list_lock, flags);
+
+	xa_lock_irqsave(xa, flags);
+	xa_for_each(xa, index, xa_fence_drv)
+		if (xa_fence_drv == fence_drv)
+			__xa_erase(xa, index);
+	xa_unlock_irqrestore(xa, flags);
+
+	/* Free seq64 memory */
+	amdgpu_seq64_free(adev, fence_drv->va);
+	kfree(fence_drv);
+}
+
+void amdgpu_userq_fence_driver_get(struct amdgpu_userq_fence_driver *fence_drv)
+{
+	kref_get(&fence_drv->refcount);
+}
+
+void amdgpu_userq_fence_driver_put(struct amdgpu_userq_fence_driver *fence_drv)
+{
+	kref_put(&fence_drv->refcount, amdgpu_userq_fence_driver_destroy);
+}
+
+static int amdgpu_userq_fence_alloc(struct amdgpu_userq_fence **userq_fence)
+{
+	*userq_fence = kmem_cache_alloc(amdgpu_userq_fence_slab, GFP_ATOMIC);
+	return *userq_fence ? 0 : -ENOMEM;
+}
+
+static int amdgpu_userq_fence_create(struct amdgpu_usermode_queue *userq,
+				     struct amdgpu_userq_fence *userq_fence,
+				     u64 seq, struct dma_fence **f)
+{
+	struct amdgpu_userq_fence_driver *fence_drv;
+	struct dma_fence *fence;
+	unsigned long flags;
+
+	fence_drv = userq->fence_drv;
+	if (!fence_drv)
+		return -EINVAL;
+
+	spin_lock_init(&userq_fence->lock);
+	INIT_LIST_HEAD(&userq_fence->link);
+	fence = &userq_fence->base;
+	userq_fence->fence_drv = fence_drv;
+
+	dma_fence_init64(fence, &amdgpu_userq_fence_ops, &userq_fence->lock,
+			 fence_drv->context, seq);
+
+	amdgpu_userq_fence_driver_get(fence_drv);
+	dma_fence_get(fence);
+
+	if (!xa_empty(&userq->fence_drv_xa)) {
+		struct amdgpu_userq_fence_driver *stored_fence_drv;
+		unsigned long index, count = 0;
+		int i = 0;
+
+		xa_lock(&userq->fence_drv_xa);
+		xa_for_each(&userq->fence_drv_xa, index, stored_fence_drv)
+			count++;
+
+		userq_fence->fence_drv_array =
+			kvmalloc_array(count,
+				       sizeof(struct amdgpu_userq_fence_driver *),
+				       GFP_ATOMIC);
+
+		if (userq_fence->fence_drv_array) {
+			xa_for_each(&userq->fence_drv_xa, index, stored_fence_drv) {
+				userq_fence->fence_drv_array[i] = stored_fence_drv;
+				__xa_erase(&userq->fence_drv_xa, index);
+				i++;
+			}
+		}
+
+		userq_fence->fence_drv_array_count = i;
+		xa_unlock(&userq->fence_drv_xa);
+	} else {
+		userq_fence->fence_drv_array = NULL;
+		userq_fence->fence_drv_array_count = 0;
+	}
+
+	/* Check if hardware has already processed the job */
+	spin_lock_irqsave(&fence_drv->fence_list_lock, flags);
+	if (!dma_fence_is_signaled(fence))
+		list_add_tail(&userq_fence->link, &fence_drv->fences);
+	else
+		dma_fence_put(fence);
+
+	spin_unlock_irqrestore(&fence_drv->fence_list_lock, flags);
+
+	*f = fence;
+
+	return 0;
+}
+
+static const char *amdgpu_userq_fence_get_driver_name(struct dma_fence *f)
+{
+	return "amdgpu_userq_fence";
+}
+
+static const char *amdgpu_userq_fence_get_timeline_name(struct dma_fence *f)
+{
+	struct amdgpu_userq_fence *fence = to_amdgpu_userq_fence(f);
+
+	return fence->fence_drv->timeline_name;
+}
+
+static bool amdgpu_userq_fence_signaled(struct dma_fence *f)
+{
+	struct amdgpu_userq_fence *fence = to_amdgpu_userq_fence(f);
+	struct amdgpu_userq_fence_driver *fence_drv = fence->fence_drv;
+	u64 rptr, wptr;
+
+	rptr = amdgpu_userq_fence_read(fence_drv);
+	wptr = fence->base.seqno;
+
+	if (rptr >= wptr)
+		return true;
+
+	return false;
+}
+
+static void amdgpu_userq_fence_free(struct rcu_head *rcu)
+{
+	struct dma_fence *fence = container_of(rcu, struct dma_fence, rcu);
+	struct amdgpu_userq_fence *userq_fence = to_amdgpu_userq_fence(fence);
+	struct amdgpu_userq_fence_driver *fence_drv = userq_fence->fence_drv;
+
+	/* Release the fence driver reference */
+	amdgpu_userq_fence_driver_put(fence_drv);
+
+	kvfree(userq_fence->fence_drv_array);
+	kmem_cache_free(amdgpu_userq_fence_slab, userq_fence);
+}
+
+static void amdgpu_userq_fence_release(struct dma_fence *f)
+{
+	call_rcu(&f->rcu, amdgpu_userq_fence_free);
+}
+
+static const struct dma_fence_ops amdgpu_userq_fence_ops = {
+	.get_driver_name = amdgpu_userq_fence_get_driver_name,
+	.get_timeline_name = amdgpu_userq_fence_get_timeline_name,
+	.signaled = amdgpu_userq_fence_signaled,
+	.release = amdgpu_userq_fence_release,
+};
+
+/**
+ * amdgpu_userq_fence_read_wptr - Read the userq wptr value
+ *
+ * @queue: user mode queue structure pointer
+ * @wptr: write pointer value
+ *
+ * Read the wptr value from userq's MQD. The userq signal IOCTL
+ * creates a dma_fence for the shared buffers that expects the
+ * RPTR value written to seq64 memory >= WPTR.
+ *
+ * Returns wptr value on success, error on failure.
+ */
+static int amdgpu_userq_fence_read_wptr(struct amdgpu_usermode_queue *queue,
+					u64 *wptr)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_bo *bo;
+	u64 addr, *ptr;
+	int r;
+
+	r = amdgpu_bo_reserve(queue->vm->root.bo, false);
+	if (r)
+		return r;
+
+	addr = queue->userq_prop->wptr_gpu_addr;
+	addr &= AMDGPU_GMC_HOLE_MASK;
+
+	mapping = amdgpu_vm_bo_lookup_mapping(queue->vm, addr >> PAGE_SHIFT);
+	if (!mapping) {
+		amdgpu_bo_unreserve(queue->vm->root.bo);
+		DRM_ERROR("Failed to lookup amdgpu_bo_va_mapping\n");
+		return -EINVAL;
+	}
+
+	bo = amdgpu_bo_ref(mapping->bo_va->base.bo);
+	amdgpu_bo_unreserve(queue->vm->root.bo);
+	r = amdgpu_bo_reserve(bo, true);
+	if (r) {
+		amdgpu_bo_unref(&bo);
+		DRM_ERROR("Failed to reserve userqueue wptr bo");
+		return r;
+	}
+
+	r = amdgpu_bo_kmap(bo, (void **)&ptr);
+	if (r) {
+		DRM_ERROR("Failed mapping the userqueue wptr bo");
+		goto map_error;
+	}
+
+	*wptr = le64_to_cpu(*ptr);
+
+	amdgpu_bo_kunmap(bo);
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+
+	return 0;
+
+map_error:
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+
+	return r;
+}
+
+static void amdgpu_userq_fence_cleanup(struct dma_fence *fence)
+{
+	dma_fence_put(fence);
+}
+
+static void
+amdgpu_userq_fence_driver_set_error(struct amdgpu_userq_fence *fence,
+				    int error)
+{
+	struct amdgpu_userq_fence_driver *fence_drv = fence->fence_drv;
+	unsigned long flags;
+	struct dma_fence *f;
+
+	spin_lock_irqsave(&fence_drv->fence_list_lock, flags);
+
+	f = rcu_dereference_protected(&fence->base,
+				      lockdep_is_held(&fence_drv->fence_list_lock));
+	if (f && !dma_fence_is_signaled_locked(f))
+		dma_fence_set_error(f, error);
+	spin_unlock_irqrestore(&fence_drv->fence_list_lock, flags);
+}
+
+void
+amdgpu_userq_fence_driver_force_completion(struct amdgpu_usermode_queue *userq)
+{
+	struct dma_fence *f = userq->last_fence;
+
+	if (f) {
+		struct amdgpu_userq_fence *fence = to_amdgpu_userq_fence(f);
+		struct amdgpu_userq_fence_driver *fence_drv = fence->fence_drv;
+		u64 wptr = fence->base.seqno;
+
+		amdgpu_userq_fence_driver_set_error(fence, -ECANCELED);
+		amdgpu_userq_fence_write(fence_drv, wptr);
+		amdgpu_userq_fence_driver_process(fence_drv);
+
+	}
+}
+
+int amdgpu_userq_signal_ioctl(struct drm_device *dev, void *data,
+			      struct drm_file *filp)
+{
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_userq_mgr *userq_mgr = &fpriv->userq_mgr;
+	struct drm_amdgpu_userq_signal *args = data;
+	struct drm_gem_object **gobj_write = NULL;
+	struct drm_gem_object **gobj_read = NULL;
+	struct amdgpu_usermode_queue *queue;
+	struct amdgpu_userq_fence *userq_fence;
+	struct drm_syncobj **syncobj = NULL;
+	u32 *bo_handles_write, num_write_bo_handles;
+	u32 *syncobj_handles, num_syncobj_handles;
+	u32 *bo_handles_read, num_read_bo_handles;
+	int r, i, entry, rentry, wentry;
+	struct dma_fence *fence;
+	struct drm_exec exec;
+	u64 wptr;
+
+	num_syncobj_handles = args->num_syncobj_handles;
+	syncobj_handles = memdup_user(u64_to_user_ptr(args->syncobj_handles),
+				      size_mul(sizeof(u32), num_syncobj_handles));
+	if (IS_ERR(syncobj_handles))
+		return PTR_ERR(syncobj_handles);
+
+	/* Array of pointers to the looked up syncobjs */
+	syncobj = kmalloc_array(num_syncobj_handles, sizeof(*syncobj), GFP_KERNEL);
+	if (!syncobj) {
+		r = -ENOMEM;
+		goto free_syncobj_handles;
+	}
+
+	for (entry = 0; entry < num_syncobj_handles; entry++) {
+		syncobj[entry] = drm_syncobj_find(filp, syncobj_handles[entry]);
+		if (!syncobj[entry]) {
+			r = -ENOENT;
+			goto free_syncobj;
+		}
+	}
+
+	num_read_bo_handles = args->num_bo_read_handles;
+	bo_handles_read = memdup_user(u64_to_user_ptr(args->bo_read_handles),
+				      sizeof(u32) * num_read_bo_handles);
+	if (IS_ERR(bo_handles_read)) {
+		r = PTR_ERR(bo_handles_read);
+		goto free_syncobj;
+	}
+
+	/* Array of pointers to the GEM read objects */
+	gobj_read = kmalloc_array(num_read_bo_handles, sizeof(*gobj_read), GFP_KERNEL);
+	if (!gobj_read) {
+		r = -ENOMEM;
+		goto free_bo_handles_read;
+	}
+
+	for (rentry = 0; rentry < num_read_bo_handles; rentry++) {
+		gobj_read[rentry] = drm_gem_object_lookup(filp, bo_handles_read[rentry]);
+		if (!gobj_read[rentry]) {
+			r = -ENOENT;
+			goto put_gobj_read;
+		}
+	}
+
+	num_write_bo_handles = args->num_bo_write_handles;
+	bo_handles_write = memdup_user(u64_to_user_ptr(args->bo_write_handles),
+				       sizeof(u32) * num_write_bo_handles);
+	if (IS_ERR(bo_handles_write)) {
+		r = PTR_ERR(bo_handles_write);
+		goto put_gobj_read;
+	}
+
+	/* Array of pointers to the GEM write objects */
+	gobj_write = kmalloc_array(num_write_bo_handles, sizeof(*gobj_write), GFP_KERNEL);
+	if (!gobj_write) {
+		r = -ENOMEM;
+		goto free_bo_handles_write;
+	}
+
+	for (wentry = 0; wentry < num_write_bo_handles; wentry++) {
+		gobj_write[wentry] = drm_gem_object_lookup(filp, bo_handles_write[wentry]);
+		if (!gobj_write[wentry]) {
+			r = -ENOENT;
+			goto put_gobj_write;
+		}
+	}
+
+	/* Retrieve the user queue */
+	queue = xa_load(&userq_mgr->userq_mgr_xa, args->queue_id);
+	if (!queue) {
+		r = -ENOENT;
+		goto put_gobj_write;
+	}
+
+	r = amdgpu_userq_fence_read_wptr(queue, &wptr);
+	if (r)
+		goto put_gobj_write;
+
+	r = amdgpu_userq_fence_alloc(&userq_fence);
+	if (r)
+		goto put_gobj_write;
+
+	/* We are here means UQ is active, make sure the eviction fence is valid */
+	amdgpu_userq_ensure_ev_fence(&fpriv->userq_mgr, &fpriv->evf_mgr);
+
+	/* Create a new fence */
+	r = amdgpu_userq_fence_create(queue, userq_fence, wptr, &fence);
+	if (r) {
+		mutex_unlock(&userq_mgr->userq_mutex);
+		kmem_cache_free(amdgpu_userq_fence_slab, userq_fence);
+		goto put_gobj_write;
+	}
+
+	dma_fence_put(queue->last_fence);
+	queue->last_fence = dma_fence_get(fence);
+	mutex_unlock(&userq_mgr->userq_mutex);
+
+	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT,
+		      (num_read_bo_handles + num_write_bo_handles));
+
+	/* Lock all BOs with retry handling */
+	drm_exec_until_all_locked(&exec) {
+		r = drm_exec_prepare_array(&exec, gobj_read, num_read_bo_handles, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (r) {
+			amdgpu_userq_fence_cleanup(fence);
+			goto exec_fini;
+		}
+
+		r = drm_exec_prepare_array(&exec, gobj_write, num_write_bo_handles, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (r) {
+			amdgpu_userq_fence_cleanup(fence);
+			goto exec_fini;
+		}
+	}
+
+	for (i = 0; i < num_read_bo_handles; i++) {
+		if (!gobj_read || !gobj_read[i]->resv)
+			continue;
+
+		dma_resv_add_fence(gobj_read[i]->resv, fence,
+				   DMA_RESV_USAGE_READ);
+	}
+
+	for (i = 0; i < num_write_bo_handles; i++) {
+		if (!gobj_write || !gobj_write[i]->resv)
+			continue;
+
+		dma_resv_add_fence(gobj_write[i]->resv, fence,
+				   DMA_RESV_USAGE_WRITE);
+	}
+
+	/* Add the created fence to syncobj/BO's */
+	for (i = 0; i < num_syncobj_handles; i++)
+		drm_syncobj_replace_fence(syncobj[i], fence);
+
+	/* drop the reference acquired in fence creation function */
+	dma_fence_put(fence);
+
+exec_fini:
+	drm_exec_fini(&exec);
+put_gobj_write:
+	while (wentry-- > 0)
+		drm_gem_object_put(gobj_write[wentry]);
+	kfree(gobj_write);
+free_bo_handles_write:
+	kfree(bo_handles_write);
+put_gobj_read:
+	while (rentry-- > 0)
+		drm_gem_object_put(gobj_read[rentry]);
+	kfree(gobj_read);
+free_bo_handles_read:
+	kfree(bo_handles_read);
+free_syncobj:
+	while (entry-- > 0)
+		if (syncobj[entry])
+			drm_syncobj_put(syncobj[entry]);
+	kfree(syncobj);
+free_syncobj_handles:
+	kfree(syncobj_handles);
+
+	return r;
+}
+
+int amdgpu_userq_wait_ioctl(struct drm_device *dev, void *data,
+			    struct drm_file *filp)
+{
+	u32 *syncobj_handles, *timeline_points, *timeline_handles, *bo_handles_read, *bo_handles_write;
+	u32 num_syncobj, num_read_bo_handles, num_write_bo_handles;
+	struct drm_amdgpu_userq_fence_info *fence_info = NULL;
+	struct drm_amdgpu_userq_wait *wait_info = data;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_userq_mgr *userq_mgr = &fpriv->userq_mgr;
+	struct amdgpu_usermode_queue *waitq;
+	struct drm_gem_object **gobj_write;
+	struct drm_gem_object **gobj_read;
+	struct dma_fence **fences = NULL;
+	u16 num_points, num_fences = 0;
+	int r, i, rentry, wentry, cnt;
+	struct drm_exec exec;
+
+	num_read_bo_handles = wait_info->num_bo_read_handles;
+	bo_handles_read = memdup_user(u64_to_user_ptr(wait_info->bo_read_handles),
+				      size_mul(sizeof(u32), num_read_bo_handles));
+	if (IS_ERR(bo_handles_read))
+		return PTR_ERR(bo_handles_read);
+
+	num_write_bo_handles = wait_info->num_bo_write_handles;
+	bo_handles_write = memdup_user(u64_to_user_ptr(wait_info->bo_write_handles),
+				       size_mul(sizeof(u32), num_write_bo_handles));
+	if (IS_ERR(bo_handles_write)) {
+		r = PTR_ERR(bo_handles_write);
+		goto free_bo_handles_read;
+	}
+
+	num_syncobj = wait_info->num_syncobj_handles;
+	syncobj_handles = memdup_user(u64_to_user_ptr(wait_info->syncobj_handles),
+				      size_mul(sizeof(u32), num_syncobj));
+	if (IS_ERR(syncobj_handles)) {
+		r = PTR_ERR(syncobj_handles);
+		goto free_bo_handles_write;
+	}
+
+	num_points = wait_info->num_syncobj_timeline_handles;
+	timeline_handles = memdup_user(u64_to_user_ptr(wait_info->syncobj_timeline_handles),
+				       sizeof(u32) * num_points);
+	if (IS_ERR(timeline_handles)) {
+		r = PTR_ERR(timeline_handles);
+		goto free_syncobj_handles;
+	}
+
+	timeline_points = memdup_user(u64_to_user_ptr(wait_info->syncobj_timeline_points),
+				      sizeof(u32) * num_points);
+	if (IS_ERR(timeline_points)) {
+		r = PTR_ERR(timeline_points);
+		goto free_timeline_handles;
+	}
+
+	gobj_read = kmalloc_array(num_read_bo_handles, sizeof(*gobj_read), GFP_KERNEL);
+	if (!gobj_read) {
+		r = -ENOMEM;
+		goto free_timeline_points;
+	}
+
+	for (rentry = 0; rentry < num_read_bo_handles; rentry++) {
+		gobj_read[rentry] = drm_gem_object_lookup(filp, bo_handles_read[rentry]);
+		if (!gobj_read[rentry]) {
+			r = -ENOENT;
+			goto put_gobj_read;
+		}
+	}
+
+	gobj_write = kmalloc_array(num_write_bo_handles, sizeof(*gobj_write), GFP_KERNEL);
+	if (!gobj_write) {
+		r = -ENOMEM;
+		goto put_gobj_read;
+	}
+
+	for (wentry = 0; wentry < num_write_bo_handles; wentry++) {
+		gobj_write[wentry] = drm_gem_object_lookup(filp, bo_handles_write[wentry]);
+		if (!gobj_write[wentry]) {
+			r = -ENOENT;
+			goto put_gobj_write;
+		}
+	}
+
+	drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT,
+		      (num_read_bo_handles + num_write_bo_handles));
+
+	/* Lock all BOs with retry handling */
+	drm_exec_until_all_locked(&exec) {
+		r = drm_exec_prepare_array(&exec, gobj_read, num_read_bo_handles, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (r) {
+			drm_exec_fini(&exec);
+			goto put_gobj_write;
+		}
+
+		r = drm_exec_prepare_array(&exec, gobj_write, num_write_bo_handles, 1);
+		drm_exec_retry_on_contention(&exec);
+		if (r) {
+			drm_exec_fini(&exec);
+			goto put_gobj_write;
+		}
+	}
+
+	if (!wait_info->num_fences) {
+		if (num_points) {
+			struct dma_fence_unwrap iter;
+			struct dma_fence *fence;
+			struct dma_fence *f;
+
+			for (i = 0; i < num_points; i++) {
+				r = drm_syncobj_find_fence(filp, timeline_handles[i],
+							   timeline_points[i],
+							   DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT,
+							   &fence);
+				if (r)
+					goto exec_fini;
+
+				dma_fence_unwrap_for_each(f, &iter, fence)
+					num_fences++;
+
+				dma_fence_put(fence);
+			}
+		}
+
+		/* Count syncobj's fence */
+		for (i = 0; i < num_syncobj; i++) {
+			struct dma_fence *fence;
+
+			r = drm_syncobj_find_fence(filp, syncobj_handles[i],
+						   0,
+						   DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT,
+						   &fence);
+			if (r)
+				goto exec_fini;
+
+			num_fences++;
+			dma_fence_put(fence);
+		}
+
+		/* Count GEM objects fence */
+		for (i = 0; i < num_read_bo_handles; i++) {
+			struct dma_resv_iter resv_cursor;
+			struct dma_fence *fence;
+
+			dma_resv_for_each_fence(&resv_cursor, gobj_read[i]->resv,
+						DMA_RESV_USAGE_READ, fence)
+				num_fences++;
+		}
+
+		for (i = 0; i < num_write_bo_handles; i++) {
+			struct dma_resv_iter resv_cursor;
+			struct dma_fence *fence;
+
+			dma_resv_for_each_fence(&resv_cursor, gobj_write[i]->resv,
+						DMA_RESV_USAGE_WRITE, fence)
+				num_fences++;
+		}
+
+		/*
+		 * Passing num_fences = 0 means that userspace doesn't want to
+		 * retrieve userq_fence_info. If num_fences = 0 we skip filling
+		 * userq_fence_info and return the actual number of fences on
+		 * args->num_fences.
+		 */
+		wait_info->num_fences = num_fences;
+	} else {
+		/* Array of fence info */
+		fence_info = kmalloc_array(wait_info->num_fences, sizeof(*fence_info), GFP_KERNEL);
+		if (!fence_info) {
+			r = -ENOMEM;
+			goto exec_fini;
+		}
+
+		/* Array of fences */
+		fences = kmalloc_array(wait_info->num_fences, sizeof(*fences), GFP_KERNEL);
+		if (!fences) {
+			r = -ENOMEM;
+			goto free_fence_info;
+		}
+
+		/* Retrieve GEM read objects fence */
+		for (i = 0; i < num_read_bo_handles; i++) {
+			struct dma_resv_iter resv_cursor;
+			struct dma_fence *fence;
+
+			dma_resv_for_each_fence(&resv_cursor, gobj_read[i]->resv,
+						DMA_RESV_USAGE_READ, fence) {
+				if (WARN_ON_ONCE(num_fences >= wait_info->num_fences)) {
+					r = -EINVAL;
+					goto free_fences;
+				}
+
+				fences[num_fences++] = fence;
+				dma_fence_get(fence);
+			}
+		}
+
+		/* Retrieve GEM write objects fence */
+		for (i = 0; i < num_write_bo_handles; i++) {
+			struct dma_resv_iter resv_cursor;
+			struct dma_fence *fence;
+
+			dma_resv_for_each_fence(&resv_cursor, gobj_write[i]->resv,
+						DMA_RESV_USAGE_WRITE, fence) {
+				if (WARN_ON_ONCE(num_fences >= wait_info->num_fences)) {
+					r = -EINVAL;
+					goto free_fences;
+				}
+
+				fences[num_fences++] = fence;
+				dma_fence_get(fence);
+			}
+		}
+
+		if (num_points) {
+			struct dma_fence_unwrap iter;
+			struct dma_fence *fence;
+			struct dma_fence *f;
+
+			for (i = 0; i < num_points; i++) {
+				r = drm_syncobj_find_fence(filp, timeline_handles[i],
+							   timeline_points[i],
+							   DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT,
+							   &fence);
+				if (r)
+					goto free_fences;
+
+				dma_fence_unwrap_for_each(f, &iter, fence) {
+					if (WARN_ON_ONCE(num_fences >= wait_info->num_fences)) {
+						r = -EINVAL;
+						goto free_fences;
+					}
+
+					dma_fence_get(f);
+					fences[num_fences++] = f;
+				}
+
+				dma_fence_put(fence);
+			}
+		}
+
+		/* Retrieve syncobj's fence */
+		for (i = 0; i < num_syncobj; i++) {
+			struct dma_fence *fence;
+
+			r = drm_syncobj_find_fence(filp, syncobj_handles[i],
+						   0,
+						   DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT,
+						   &fence);
+			if (r)
+				goto free_fences;
+
+			if (WARN_ON_ONCE(num_fences >= wait_info->num_fences)) {
+				r = -EINVAL;
+				goto free_fences;
+			}
+
+			fences[num_fences++] = fence;
+		}
+
+		/*
+		 * Keep only the latest fences to reduce the number of values
+		 * given back to userspace.
+		 */
+		num_fences = dma_fence_dedup_array(fences, num_fences);
+
+		waitq = xa_load(&userq_mgr->userq_mgr_xa, wait_info->waitq_id);
+		if (!waitq) {
+			r = -EINVAL;
+			goto free_fences;
+		}
+
+		for (i = 0, cnt = 0; i < num_fences; i++) {
+			struct amdgpu_userq_fence_driver *fence_drv;
+			struct amdgpu_userq_fence *userq_fence;
+			u32 index;
+
+			userq_fence = to_amdgpu_userq_fence(fences[i]);
+			if (!userq_fence) {
+				/*
+				 * Just waiting on other driver fences should
+				 * be good for now
+				 */
+				r = dma_fence_wait(fences[i], true);
+				if (r) {
+					dma_fence_put(fences[i]);
+					goto free_fences;
+				}
+
+				dma_fence_put(fences[i]);
+				continue;
+			}
+
+			fence_drv = userq_fence->fence_drv;
+			/*
+			 * We need to make sure the user queue release their reference
+			 * to the fence drivers at some point before queue destruction.
+			 * Otherwise, we would gather those references until we don't
+			 * have any more space left and crash.
+			 */
+			r = xa_alloc(&waitq->fence_drv_xa, &index, fence_drv,
+				     xa_limit_32b, GFP_KERNEL);
+			if (r)
+				goto free_fences;
+
+			amdgpu_userq_fence_driver_get(fence_drv);
+
+			/* Store drm syncobj's gpu va address and value */
+			fence_info[cnt].va = fence_drv->va;
+			fence_info[cnt].value = fences[i]->seqno;
+
+			dma_fence_put(fences[i]);
+			/* Increment the actual userq fence count */
+			cnt++;
+		}
+
+		wait_info->num_fences = cnt;
+		/* Copy userq fence info to user space */
+		if (copy_to_user(u64_to_user_ptr(wait_info->out_fences),
+				 fence_info, wait_info->num_fences * sizeof(*fence_info))) {
+			r = -EFAULT;
+			goto free_fences;
+		}
+
+		kfree(fences);
+		kfree(fence_info);
+	}
+
+	drm_exec_fini(&exec);
+	for (i = 0; i < num_read_bo_handles; i++)
+		drm_gem_object_put(gobj_read[i]);
+	kfree(gobj_read);
+
+	for (i = 0; i < num_write_bo_handles; i++)
+		drm_gem_object_put(gobj_write[i]);
+	kfree(gobj_write);
+
+	kfree(timeline_points);
+	kfree(timeline_handles);
+	kfree(syncobj_handles);
+	kfree(bo_handles_write);
+	kfree(bo_handles_read);
+
+	return 0;
+
+free_fences:
+	while (num_fences-- > 0)
+		dma_fence_put(fences[num_fences]);
+	kfree(fences);
+free_fence_info:
+	kfree(fence_info);
+exec_fini:
+	drm_exec_fini(&exec);
+put_gobj_write:
+	while (wentry-- > 0)
+		drm_gem_object_put(gobj_write[wentry]);
+	kfree(gobj_write);
+put_gobj_read:
+	while (rentry-- > 0)
+		drm_gem_object_put(gobj_read[rentry]);
+	kfree(gobj_read);
+free_timeline_points:
+	kfree(timeline_points);
+free_timeline_handles:
+	kfree(timeline_handles);
+free_syncobj_handles:
+	kfree(syncobj_handles);
+free_bo_handles_write:
+	kfree(bo_handles_write);
+free_bo_handles_read:
+	kfree(bo_handles_read);
+
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.h
new file mode 100644
index 000000000000..d76add2afc77
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.h
@@ -0,0 +1,77 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_USERQ_FENCE_H__
+#define __AMDGPU_USERQ_FENCE_H__
+
+#include <linux/types.h>
+
+#include "amdgpu_userq.h"
+
+struct amdgpu_userq_fence {
+	struct dma_fence base;
+	/*
+	 * This lock is necessary to synchronize the
+	 * userqueue dma fence operations.
+	 */
+	spinlock_t lock;
+	struct list_head link;
+	unsigned long fence_drv_array_count;
+	struct amdgpu_userq_fence_driver *fence_drv;
+	struct amdgpu_userq_fence_driver **fence_drv_array;
+};
+
+struct amdgpu_userq_fence_driver {
+	struct kref refcount;
+	u64 va;
+	u64 gpu_addr;
+	u64 *cpu_addr;
+	u64 context;
+	/*
+	 * This lock is necesaary to synchronize the access
+	 * to the fences list by the fence driver.
+	 */
+	spinlock_t fence_list_lock;
+	struct list_head fences;
+	struct amdgpu_device *adev;
+	char timeline_name[TASK_COMM_LEN];
+};
+
+int amdgpu_userq_fence_slab_init(void);
+void amdgpu_userq_fence_slab_fini(void);
+
+void amdgpu_userq_fence_driver_get(struct amdgpu_userq_fence_driver *fence_drv);
+void amdgpu_userq_fence_driver_put(struct amdgpu_userq_fence_driver *fence_drv);
+int amdgpu_userq_fence_driver_alloc(struct amdgpu_device *adev,
+				    struct amdgpu_usermode_queue *userq);
+void amdgpu_userq_fence_driver_free(struct amdgpu_usermode_queue *userq);
+void amdgpu_userq_fence_driver_process(struct amdgpu_userq_fence_driver *fence_drv);
+void amdgpu_userq_fence_driver_force_completion(struct amdgpu_usermode_queue *userq);
+void amdgpu_userq_fence_driver_destroy(struct kref *ref);
+int amdgpu_userq_signal_ioctl(struct drm_device *dev, void *data,
+			      struct drm_file *filp);
+int amdgpu_userq_wait_ioctl(struct drm_device *dev, void *data,
+			    struct drm_file *filp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_utils.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_utils.h
new file mode 100644
index 000000000000..1e40ca3b1584
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_utils.h
@@ -0,0 +1,91 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef AMDGPU_UTILS_H_
+#define AMDGPU_UTILS_H_
+
+/* ---------- Generic 2‑bit capability attribute encoding ----------
+ * 00 INVALID, 01 RO, 10 WO, 11 RW
+ */
+enum amdgpu_cap_attr {
+	AMDGPU_CAP_ATTR_INVALID = 0,
+	AMDGPU_CAP_ATTR_RO      = 1 << 0,
+	AMDGPU_CAP_ATTR_WO      = 1 << 1,
+	AMDGPU_CAP_ATTR_RW      = (AMDGPU_CAP_ATTR_RO | AMDGPU_CAP_ATTR_WO),
+};
+
+#define AMDGPU_CAP_ATTR_BITS 2
+#define AMDGPU_CAP_ATTR_MAX  ((1U << AMDGPU_CAP_ATTR_BITS) - 1)
+
+/* Internal helper to build helpers for a given enum NAME */
+#define DECLARE_ATTR_CAP_CLASS_HELPERS(NAME)							\
+enum { NAME##_BITMAP_BITS = NAME##_COUNT * AMDGPU_CAP_ATTR_BITS };				\
+struct NAME##_caps {										\
+	DECLARE_BITMAP(bmap, NAME##_BITMAP_BITS);						\
+};												\
+static inline unsigned int NAME##_ATTR_START(enum NAME##_cap_id cap)				\
+{ return (unsigned int)cap * AMDGPU_CAP_ATTR_BITS; }						\
+static inline void NAME##_attr_init(struct NAME##_caps *c)					\
+{ if (c) bitmap_zero(c->bmap, NAME##_BITMAP_BITS); }						\
+static inline int NAME##_attr_set(struct NAME##_caps *c,					\
+				  enum NAME##_cap_id cap, enum amdgpu_cap_attr attr)		\
+{												\
+	if (!c)											\
+		return -EINVAL;									\
+	if (cap >= NAME##_COUNT)								\
+		return -EINVAL;									\
+	if ((unsigned int)attr > AMDGPU_CAP_ATTR_MAX)						\
+		return -EINVAL;									\
+	bitmap_write(c->bmap, (unsigned long)attr,						\
+			NAME##_ATTR_START(cap), AMDGPU_CAP_ATTR_BITS);				\
+	return 0;										\
+}												\
+static inline int NAME##_attr_get(const struct NAME##_caps *c,					\
+				  enum NAME##_cap_id cap, enum amdgpu_cap_attr *out)		\
+{												\
+	unsigned long v;									\
+	if (!c || !out)										\
+		return -EINVAL;									\
+	if (cap >= NAME##_COUNT)								\
+		return -EINVAL;									\
+	v = bitmap_read(c->bmap, NAME##_ATTR_START(cap), AMDGPU_CAP_ATTR_BITS);			\
+	*out = (enum amdgpu_cap_attr)v;								\
+	return 0;										\
+}												\
+static inline bool NAME##_cap_is_ro(const struct NAME##_caps *c, enum NAME##_cap_id id)		\
+{ enum amdgpu_cap_attr a; return !NAME##_attr_get(c, id, &a) && a == AMDGPU_CAP_ATTR_RO; }	\
+static inline bool NAME##_cap_is_wo(const struct NAME##_caps *c, enum NAME##_cap_id id)		\
+{ enum amdgpu_cap_attr a; return !NAME##_attr_get(c, id, &a) && a == AMDGPU_CAP_ATTR_WO; }	\
+static inline bool NAME##_cap_is_rw(const struct NAME##_caps *c, enum NAME##_cap_id id)		\
+{ enum amdgpu_cap_attr a; return !NAME##_attr_get(c, id, &a) && a == AMDGPU_CAP_ATTR_RW; }
+
+/* Element expander for enum creation */
+#define _CAP_ENUM_ELEM(x) x,
+
+/* Public macro: declare enum + helpers from an X‑macro list */
+#define DECLARE_ATTR_CAP_CLASS(NAME, LIST_MACRO)						\
+	enum NAME##_cap_id { LIST_MACRO(_CAP_ENUM_ELEM) NAME##_COUNT };				\
+	DECLARE_ATTR_CAP_CLASS_HELPERS(NAME)
+
+#endif /* AMDGPU_UTILS_H_ */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
index 74758b5ffc6c..5c38f0d30c87 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
@@ -1136,7 +1136,8 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo,
 	r = amdgpu_job_alloc_with_ib(ring->adev, &adev->uvd.entity,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     64, direct ? AMDGPU_IB_POOL_DIRECT :
-				     AMDGPU_IB_POOL_DELAYED, &job);
+				     AMDGPU_IB_POOL_DELAYED, &job,
+				     AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
index b9060bcd4806..a7d8f1ce6ac2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
@@ -41,6 +41,9 @@
 #define VCE_IDLE_TIMEOUT	msecs_to_jiffies(1000)
 
 /* Firmware Names */
+#ifdef CONFIG_DRM_AMDGPU_SI
+#define FIRMWARE_VCE_V1_0	"amdgpu/vce_1_0_0.bin"
+#endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
 #define FIRMWARE_BONAIRE	"amdgpu/bonaire_vce.bin"
 #define FIRMWARE_KABINI	"amdgpu/kabini_vce.bin"
@@ -61,6 +64,9 @@
 #define FIRMWARE_VEGA12		"amdgpu/vega12_vce.bin"
 #define FIRMWARE_VEGA20		"amdgpu/vega20_vce.bin"
 
+#ifdef CONFIG_DRM_AMDGPU_SI
+MODULE_FIRMWARE(FIRMWARE_VCE_V1_0);
+#endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
 MODULE_FIRMWARE(FIRMWARE_BONAIRE);
 MODULE_FIRMWARE(FIRMWARE_KABINI);
@@ -88,82 +94,93 @@ static int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 				      bool direct, struct dma_fence **fence);
 
 /**
- * amdgpu_vce_sw_init - allocate memory, load vce firmware
+ * amdgpu_vce_firmware_name() - determine the firmware file name for VCE
  *
  * @adev: amdgpu_device pointer
- * @size: size for the new BO
  *
- * First step to get VCE online, allocate memory and load the firmware
+ * Each chip that has VCE IP may need a different firmware.
+ * This function returns the name of the VCE firmware file
+ * appropriate for the current chip.
  */
-int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
+static const char *amdgpu_vce_firmware_name(struct amdgpu_device *adev)
 {
-	const char *fw_name;
-	const struct common_firmware_header *hdr;
-	unsigned int ucode_version, version_major, version_minor, binary_id;
-	int i, r;
-
 	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_PITCAIRN:
+	case CHIP_TAHITI:
+	case CHIP_VERDE:
+		return FIRMWARE_VCE_V1_0;
+#endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
 	case CHIP_BONAIRE:
-		fw_name = FIRMWARE_BONAIRE;
-		break;
+		return FIRMWARE_BONAIRE;
 	case CHIP_KAVERI:
-		fw_name = FIRMWARE_KAVERI;
-		break;
+		return FIRMWARE_KAVERI;
 	case CHIP_KABINI:
-		fw_name = FIRMWARE_KABINI;
-		break;
+		return FIRMWARE_KABINI;
 	case CHIP_HAWAII:
-		fw_name = FIRMWARE_HAWAII;
-		break;
+		return FIRMWARE_HAWAII;
 	case CHIP_MULLINS:
-		fw_name = FIRMWARE_MULLINS;
-		break;
+		return FIRMWARE_MULLINS;
 #endif
 	case CHIP_TONGA:
-		fw_name = FIRMWARE_TONGA;
-		break;
+		return  FIRMWARE_TONGA;
 	case CHIP_CARRIZO:
-		fw_name = FIRMWARE_CARRIZO;
-		break;
+		return  FIRMWARE_CARRIZO;
 	case CHIP_FIJI:
-		fw_name = FIRMWARE_FIJI;
-		break;
+		return  FIRMWARE_FIJI;
 	case CHIP_STONEY:
-		fw_name = FIRMWARE_STONEY;
-		break;
+		return  FIRMWARE_STONEY;
 	case CHIP_POLARIS10:
-		fw_name = FIRMWARE_POLARIS10;
-		break;
+		return  FIRMWARE_POLARIS10;
 	case CHIP_POLARIS11:
-		fw_name = FIRMWARE_POLARIS11;
-		break;
+		return  FIRMWARE_POLARIS11;
 	case CHIP_POLARIS12:
-		fw_name = FIRMWARE_POLARIS12;
-		break;
+		return  FIRMWARE_POLARIS12;
 	case CHIP_VEGAM:
-		fw_name = FIRMWARE_VEGAM;
-		break;
+		return  FIRMWARE_VEGAM;
 	case CHIP_VEGA10:
-		fw_name = FIRMWARE_VEGA10;
-		break;
+		return  FIRMWARE_VEGA10;
 	case CHIP_VEGA12:
-		fw_name = FIRMWARE_VEGA12;
-		break;
+		return  FIRMWARE_VEGA12;
 	case CHIP_VEGA20:
-		fw_name = FIRMWARE_VEGA20;
-		break;
+		return  FIRMWARE_VEGA20;
 
 	default:
-		return -EINVAL;
+		return NULL;
 	}
+}
+
+/**
+ * amdgpu_vce_early_init() - try to load VCE firmware
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tries to load the VCE firmware.
+ *
+ * When not found, returns ENOENT so that the driver can
+ * still load and initialize the rest of the IP blocks.
+ * The GPU can function just fine without VCE, they will just
+ * not support video encoding.
+ */
+int amdgpu_vce_early_init(struct amdgpu_device *adev)
+{
+	const char *fw_name = amdgpu_vce_firmware_name(adev);
+	const struct common_firmware_header *hdr;
+	unsigned int ucode_version, version_major, version_minor, binary_id;
+	int r;
+
+	if (!fw_name)
+		return -ENOENT;
 
 	r = amdgpu_ucode_request(adev, &adev->vce.fw, AMDGPU_UCODE_REQUIRED, "%s", fw_name);
 	if (r) {
-		dev_err(adev->dev, "amdgpu_vce: Can't validate firmware \"%s\"\n",
-			fw_name);
+		dev_err(adev->dev,
+			"amdgpu_vce: Firmware \"%s\" not found or failed to validate (%d)\n",
+			fw_name, r);
+
 		amdgpu_ucode_release(&adev->vce.fw);
-		return r;
+		return -ENOENT;
 	}
 
 	hdr = (const struct common_firmware_header *)adev->vce.fw->data;
@@ -172,11 +189,35 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
 	version_major = (ucode_version >> 20) & 0xfff;
 	version_minor = (ucode_version >> 8) & 0xfff;
 	binary_id = ucode_version & 0xff;
-	DRM_INFO("Found VCE firmware Version: %d.%d Binary ID: %d\n",
+	dev_info(adev->dev, "Found VCE firmware Version: %d.%d Binary ID: %d\n",
 		version_major, version_minor, binary_id);
 	adev->vce.fw_version = ((version_major << 24) | (version_minor << 16) |
 				(binary_id << 8));
 
+	return 0;
+}
+
+/**
+ * amdgpu_vce_sw_init() - allocate memory for VCE BO
+ *
+ * @adev: amdgpu_device pointer
+ * @size: size for the new BO
+ *
+ * First step to get VCE online: allocate memory for VCE BO.
+ * The VCE firmware binary is copied into the VCE BO later,
+ * in amdgpu_vce_resume. The VCE executes its code from the
+ * VCE BO and also uses the space in this BO for its stack and data.
+ *
+ * Ideally this BO should be placed in VRAM for optimal performance,
+ * although technically it also runs from system RAM (albeit slowly).
+ */
+int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
+{
+	int i, r;
+
+	if (!adev->vce.fw)
+		return -ENOENT;
+
 	r = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE,
 				    AMDGPU_GEM_DOMAIN_VRAM |
 				    AMDGPU_GEM_DOMAIN_GTT,
@@ -285,40 +326,23 @@ int amdgpu_vce_suspend(struct amdgpu_device *adev)
  */
 int amdgpu_vce_resume(struct amdgpu_device *adev)
 {
-	void *cpu_addr;
 	const struct common_firmware_header *hdr;
 	unsigned int offset;
-	int r, idx;
+	int idx;
 
 	if (adev->vce.vcpu_bo == NULL)
 		return -EINVAL;
 
-	r = amdgpu_bo_reserve(adev->vce.vcpu_bo, false);
-	if (r) {
-		dev_err(adev->dev, "(%d) failed to reserve VCE bo\n", r);
-		return r;
-	}
-
-	r = amdgpu_bo_kmap(adev->vce.vcpu_bo, &cpu_addr);
-	if (r) {
-		amdgpu_bo_unreserve(adev->vce.vcpu_bo);
-		dev_err(adev->dev, "(%d) VCE map failed\n", r);
-		return r;
-	}
-
 	hdr = (const struct common_firmware_header *)adev->vce.fw->data;
 	offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
 
 	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
-		memcpy_toio(cpu_addr, adev->vce.fw->data + offset,
+		memset_io(adev->vce.cpu_addr, 0, amdgpu_bo_size(adev->vce.vcpu_bo));
+		memcpy_toio(adev->vce.cpu_addr, adev->vce.fw->data + offset,
 			    adev->vce.fw->size - offset);
 		drm_dev_exit(idx);
 	}
 
-	amdgpu_bo_kunmap(adev->vce.vcpu_bo);
-
-	amdgpu_bo_unreserve(adev->vce.vcpu_bo);
-
 	return 0;
 }
 
@@ -427,6 +451,24 @@ void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
 }
 
 /**
+ * amdgpu_vce_required_gart_pages() - gets number of GART pages required by VCE
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns how many GART pages we need before GTT for the VCE IP block.
+ * For VCE1, see vce_v1_0_ensure_vcpu_bo_32bit_addr for details.
+ * For VCE2+, this is not needed so return zero.
+ */
+u32 amdgpu_vce_required_gart_pages(struct amdgpu_device *adev)
+{
+	/* VCE IP block not added yet, so can't use amdgpu_ip_version */
+	if (adev->family == AMDGPU_FAMILY_SI)
+		return 512;
+
+	return 0;
+}
+
+/**
  * amdgpu_vce_get_create_msg - generate a VCE create msg
  *
  * @ring: ring we should submit the msg to
@@ -449,7 +491,7 @@ static int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
 	r = amdgpu_job_alloc_with_ib(ring->adev, &ring->adev->vce.entity,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
-				     &job);
+				     &job, AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
@@ -540,7 +582,8 @@ static int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     ib_size_dw * 4,
 				     direct ? AMDGPU_IB_POOL_DIRECT :
-				     AMDGPU_IB_POOL_DELAYED, &job);
+				     AMDGPU_IB_POOL_DELAYED, &job,
+				     AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
index 6e53f872d084..1c3464ce5037 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
@@ -51,14 +51,17 @@ struct amdgpu_vce {
 	struct drm_sched_entity	entity;
 	uint32_t                srbm_soft_reset;
 	unsigned		num_rings;
+	uint32_t		keyselect;
 };
 
+int amdgpu_vce_early_init(struct amdgpu_device *adev);
 int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size);
 int amdgpu_vce_sw_fini(struct amdgpu_device *adev);
 int amdgpu_vce_entity_init(struct amdgpu_device *adev, struct amdgpu_ring *ring);
 int amdgpu_vce_suspend(struct amdgpu_device *adev);
 int amdgpu_vce_resume(struct amdgpu_device *adev);
 void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp);
+u32 amdgpu_vce_required_gart_pages(struct amdgpu_device *adev);
 int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, struct amdgpu_job *job,
 			     struct amdgpu_ib *ib);
 int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
index 83faf6e6788a..5e0786ea911b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
@@ -92,48 +92,56 @@ MODULE_FIRMWARE(FIRMWARE_VCN5_0_0);
 MODULE_FIRMWARE(FIRMWARE_VCN5_0_1);
 
 static void amdgpu_vcn_idle_work_handler(struct work_struct *work);
+static void amdgpu_vcn_reg_dump_fini(struct amdgpu_device *adev);
 
-int amdgpu_vcn_early_init(struct amdgpu_device *adev)
+int amdgpu_vcn_early_init(struct amdgpu_device *adev, int i)
 {
 	char ucode_prefix[25];
-	int r, i;
+	int r;
 
+	adev->vcn.inst[i].adev = adev;
+	adev->vcn.inst[i].inst = i;
 	amdgpu_ucode_ip_version_decode(adev, UVD_HWIP, ucode_prefix, sizeof(ucode_prefix));
-	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
-		if (i == 1 && amdgpu_ip_version(adev, UVD_HWIP, 0) ==  IP_VERSION(4, 0, 6))
-			r = amdgpu_ucode_request(adev, &adev->vcn.inst[i].fw,
-						 AMDGPU_UCODE_REQUIRED,
-						 "amdgpu/%s_%d.bin", ucode_prefix, i);
-		else
-			r = amdgpu_ucode_request(adev, &adev->vcn.inst[i].fw,
+
+	if (i != 0 && adev->vcn.per_inst_fw) {
+		r = amdgpu_ucode_request(adev, &adev->vcn.inst[i].fw,
+					 AMDGPU_UCODE_REQUIRED,
+					 "amdgpu/%s_%d.bin", ucode_prefix, i);
+		if (r)
+			amdgpu_ucode_release(&adev->vcn.inst[i].fw);
+	} else {
+		if (!adev->vcn.inst[0].fw) {
+			r = amdgpu_ucode_request(adev, &adev->vcn.inst[0].fw,
 						 AMDGPU_UCODE_REQUIRED,
 						 "amdgpu/%s.bin", ucode_prefix);
-		if (r) {
-			amdgpu_ucode_release(&adev->vcn.inst[i].fw);
-			return r;
+			if (r)
+				amdgpu_ucode_release(&adev->vcn.inst[0].fw);
+		} else {
+			r = 0;
 		}
+		adev->vcn.inst[i].fw = adev->vcn.inst[0].fw;
 	}
+
 	return r;
 }
 
-int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
+int amdgpu_vcn_sw_init(struct amdgpu_device *adev, int i)
 {
 	unsigned long bo_size;
 	const struct common_firmware_header *hdr;
 	unsigned char fw_check;
 	unsigned int fw_shared_size, log_offset;
-	int i, r;
-
-	INIT_DELAYED_WORK(&adev->vcn.idle_work, amdgpu_vcn_idle_work_handler);
-	mutex_init(&adev->vcn.vcn_pg_lock);
-	mutex_init(&adev->vcn.vcn1_jpeg1_workaround);
-	atomic_set(&adev->vcn.total_submission_cnt, 0);
-	for (i = 0; i < adev->vcn.num_vcn_inst; i++)
-		atomic_set(&adev->vcn.inst[i].dpg_enc_submission_cnt, 0);
+	int r;
 
+	mutex_init(&adev->vcn.inst[i].vcn1_jpeg1_workaround);
+	mutex_init(&adev->vcn.inst[i].vcn_pg_lock);
+	mutex_init(&adev->vcn.inst[i].engine_reset_mutex);
+	atomic_set(&adev->vcn.inst[i].total_submission_cnt, 0);
+	INIT_DELAYED_WORK(&adev->vcn.inst[i].idle_work, amdgpu_vcn_idle_work_handler);
+	atomic_set(&adev->vcn.inst[i].dpg_enc_submission_cnt, 0);
 	if ((adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) &&
 	    (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG))
-		adev->vcn.indirect_sram = true;
+		adev->vcn.inst[i].indirect_sram = true;
 
 	/*
 	 * Some Steam Deck's BIOS versions are incompatible with the
@@ -146,18 +154,19 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
 		const char *bios_ver = dmi_get_system_info(DMI_BIOS_VERSION);
 
 		if (bios_ver && (!strncmp("F7A0113", bios_ver, 7) ||
-		     !strncmp("F7A0114", bios_ver, 7))) {
-			adev->vcn.indirect_sram = false;
+				 !strncmp("F7A0114", bios_ver, 7))) {
+			adev->vcn.inst[i].indirect_sram = false;
 			dev_info(adev->dev,
-				"Steam Deck quirk: indirect SRAM disabled on BIOS %s\n", bios_ver);
+				 "Steam Deck quirk: indirect SRAM disabled on BIOS %s\n", bios_ver);
 		}
 	}
 
 	/* from vcn4 and above, only unified queue is used */
-	adev->vcn.using_unified_queue =
+	adev->vcn.inst[i].using_unified_queue =
 		amdgpu_ip_version(adev, UVD_HWIP, 0) >= IP_VERSION(4, 0, 0);
 
-	hdr = (const struct common_firmware_header *)adev->vcn.inst[0].fw->data;
+	hdr = (const struct common_firmware_header *)adev->vcn.inst[i].fw->data;
+	adev->vcn.inst[i].fw_version = le32_to_cpu(hdr->ucode_version);
 	adev->vcn.fw_version = le32_to_cpu(hdr->ucode_version);
 
 	/* Bit 20-23, it is encode major and non-zero for new naming convention.
@@ -175,16 +184,17 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
 		enc_major = fw_check;
 		dec_ver = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xf;
 		vep = (le32_to_cpu(hdr->ucode_version) >> 28) & 0xf;
-		DRM_INFO("Found VCN firmware Version ENC: %u.%u DEC: %u VEP: %u Revision: %u\n",
-			enc_major, enc_minor, dec_ver, vep, fw_rev);
+		dev_info(adev->dev,
+			 "[VCN instance %d] Found VCN firmware Version ENC: %u.%u DEC: %u VEP: %u Revision: %u\n",
+			 i, enc_major, enc_minor, dec_ver, vep, fw_rev);
 	} else {
 		unsigned int version_major, version_minor, family_id;
 
 		family_id = le32_to_cpu(hdr->ucode_version) & 0xff;
 		version_major = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xff;
 		version_minor = (le32_to_cpu(hdr->ucode_version) >> 8) & 0xff;
-		DRM_INFO("Found VCN firmware Version: %u.%u Family ID: %u\n",
-			version_major, version_minor, family_id);
+		dev_info(adev->dev, "[VCN instance %d] Found VCN firmware Version: %u.%u Family ID: %u\n",
+			 i, version_major, version_minor, family_id);
 	}
 
 	bo_size = AMDGPU_VCN_STACK_SIZE + AMDGPU_VCN_CONTEXT_SIZE;
@@ -207,82 +217,81 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
 	if (amdgpu_vcnfw_log)
 		bo_size += AMDGPU_VCNFW_LOG_SIZE;
 
-	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
-		if (adev->vcn.harvest_config & (1 << i))
-			continue;
+	r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM |
+				    AMDGPU_GEM_DOMAIN_GTT,
+				    &adev->vcn.inst[i].vcpu_bo,
+				    &adev->vcn.inst[i].gpu_addr,
+				    &adev->vcn.inst[i].cpu_addr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate vcn bo\n", r);
+		return r;
+	}
+
+	adev->vcn.inst[i].fw_shared.cpu_addr = adev->vcn.inst[i].cpu_addr +
+		bo_size - fw_shared_size;
+	adev->vcn.inst[i].fw_shared.gpu_addr = adev->vcn.inst[i].gpu_addr +
+		bo_size - fw_shared_size;
+
+	adev->vcn.inst[i].fw_shared.mem_size = fw_shared_size;
 
-		r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE,
+	if (amdgpu_vcnfw_log) {
+		adev->vcn.inst[i].fw_shared.cpu_addr -= AMDGPU_VCNFW_LOG_SIZE;
+		adev->vcn.inst[i].fw_shared.gpu_addr -= AMDGPU_VCNFW_LOG_SIZE;
+		adev->vcn.inst[i].fw_shared.log_offset = log_offset;
+	}
+
+	if (adev->vcn.inst[i].indirect_sram) {
+		r = amdgpu_bo_create_kernel(adev, 64 * 2 * 4, PAGE_SIZE,
 					    AMDGPU_GEM_DOMAIN_VRAM |
 					    AMDGPU_GEM_DOMAIN_GTT,
-					    &adev->vcn.inst[i].vcpu_bo,
-					    &adev->vcn.inst[i].gpu_addr,
-					    &adev->vcn.inst[i].cpu_addr);
+					    &adev->vcn.inst[i].dpg_sram_bo,
+					    &adev->vcn.inst[i].dpg_sram_gpu_addr,
+					    &adev->vcn.inst[i].dpg_sram_cpu_addr);
 		if (r) {
-			dev_err(adev->dev, "(%d) failed to allocate vcn bo\n", r);
+			dev_err(adev->dev, "VCN %d (%d) failed to allocate DPG bo\n", i, r);
 			return r;
 		}
-
-		adev->vcn.inst[i].fw_shared.cpu_addr = adev->vcn.inst[i].cpu_addr +
-				bo_size - fw_shared_size;
-		adev->vcn.inst[i].fw_shared.gpu_addr = adev->vcn.inst[i].gpu_addr +
-				bo_size - fw_shared_size;
-
-		adev->vcn.inst[i].fw_shared.mem_size = fw_shared_size;
-
-		if (amdgpu_vcnfw_log) {
-			adev->vcn.inst[i].fw_shared.cpu_addr -= AMDGPU_VCNFW_LOG_SIZE;
-			adev->vcn.inst[i].fw_shared.gpu_addr -= AMDGPU_VCNFW_LOG_SIZE;
-			adev->vcn.inst[i].fw_shared.log_offset = log_offset;
-		}
-
-		if (adev->vcn.indirect_sram) {
-			r = amdgpu_bo_create_kernel(adev, 64 * 2 * 4, PAGE_SIZE,
-					AMDGPU_GEM_DOMAIN_VRAM |
-					AMDGPU_GEM_DOMAIN_GTT,
-					&adev->vcn.inst[i].dpg_sram_bo,
-					&adev->vcn.inst[i].dpg_sram_gpu_addr,
-					&adev->vcn.inst[i].dpg_sram_cpu_addr);
-			if (r) {
-				dev_err(adev->dev, "VCN %d (%d) failed to allocate DPG bo\n", i, r);
-				return r;
-			}
-		}
 	}
 
 	return 0;
 }
 
-int amdgpu_vcn_sw_fini(struct amdgpu_device *adev)
+void amdgpu_vcn_sw_fini(struct amdgpu_device *adev, int i)
 {
-	int i, j;
+	int j;
 
-	for (j = 0; j < adev->vcn.num_vcn_inst; ++j) {
-		if (adev->vcn.harvest_config & (1 << j))
-			continue;
+	if (adev->vcn.harvest_config & (1 << i))
+		return;
 
-		amdgpu_bo_free_kernel(
-			&adev->vcn.inst[j].dpg_sram_bo,
-			&adev->vcn.inst[j].dpg_sram_gpu_addr,
-			(void **)&adev->vcn.inst[j].dpg_sram_cpu_addr);
+	amdgpu_bo_free_kernel(
+		&adev->vcn.inst[i].dpg_sram_bo,
+		&adev->vcn.inst[i].dpg_sram_gpu_addr,
+		(void **)&adev->vcn.inst[i].dpg_sram_cpu_addr);
 
-		kvfree(adev->vcn.inst[j].saved_bo);
+	kvfree(adev->vcn.inst[i].saved_bo);
 
-		amdgpu_bo_free_kernel(&adev->vcn.inst[j].vcpu_bo,
-					  &adev->vcn.inst[j].gpu_addr,
-					  (void **)&adev->vcn.inst[j].cpu_addr);
+	amdgpu_bo_free_kernel(&adev->vcn.inst[i].vcpu_bo,
+			      &adev->vcn.inst[i].gpu_addr,
+			      (void **)&adev->vcn.inst[i].cpu_addr);
 
-		amdgpu_ring_fini(&adev->vcn.inst[j].ring_dec);
+	amdgpu_ring_fini(&adev->vcn.inst[i].ring_dec);
 
-		for (i = 0; i < adev->vcn.num_enc_rings; ++i)
-			amdgpu_ring_fini(&adev->vcn.inst[j].ring_enc[i]);
+	for (j = 0; j < adev->vcn.inst[i].num_enc_rings; ++j)
+		amdgpu_ring_fini(&adev->vcn.inst[i].ring_enc[j]);
 
-		amdgpu_ucode_release(&adev->vcn.inst[j].fw);
+	if (adev->vcn.per_inst_fw) {
+		amdgpu_ucode_release(&adev->vcn.inst[i].fw);
+	} else {
+		amdgpu_ucode_release(&adev->vcn.inst[0].fw);
+		adev->vcn.inst[i].fw = NULL;
 	}
 
-	mutex_destroy(&adev->vcn.vcn1_jpeg1_workaround);
-	mutex_destroy(&adev->vcn.vcn_pg_lock);
+	if (adev->vcn.reg_list)
+		amdgpu_vcn_reg_dump_fini(adev);
 
-	return 0;
+	mutex_destroy(&adev->vcn.inst[i].vcn_pg_lock);
+	mutex_destroy(&adev->vcn.inst[i].vcn1_jpeg1_workaround);
 }
 
 bool amdgpu_vcn_is_disabled_vcn(struct amdgpu_device *adev, enum vcn_ring_type type, uint32_t vcn_instance)
@@ -300,179 +309,230 @@ bool amdgpu_vcn_is_disabled_vcn(struct amdgpu_device *adev, enum vcn_ring_type t
 	return ret;
 }
 
-int amdgpu_vcn_save_vcpu_bo(struct amdgpu_device *adev)
+static int amdgpu_vcn_save_vcpu_bo_inst(struct amdgpu_device *adev, int i)
 {
 	unsigned int size;
 	void *ptr;
-	int i, idx;
+	int idx;
 
-	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
-		if (adev->vcn.harvest_config & (1 << i))
-			continue;
-		if (adev->vcn.inst[i].vcpu_bo == NULL)
-			return 0;
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+	if (adev->vcn.inst[i].vcpu_bo == NULL)
+		return 0;
 
-		size = amdgpu_bo_size(adev->vcn.inst[i].vcpu_bo);
-		ptr = adev->vcn.inst[i].cpu_addr;
+	size = amdgpu_bo_size(adev->vcn.inst[i].vcpu_bo);
+	ptr = adev->vcn.inst[i].cpu_addr;
 
-		adev->vcn.inst[i].saved_bo = kvmalloc(size, GFP_KERNEL);
-		if (!adev->vcn.inst[i].saved_bo)
-			return -ENOMEM;
+	adev->vcn.inst[i].saved_bo = kvmalloc(size, GFP_KERNEL);
+	if (!adev->vcn.inst[i].saved_bo)
+		return -ENOMEM;
 
-		if (drm_dev_enter(adev_to_drm(adev), &idx)) {
-			memcpy_fromio(adev->vcn.inst[i].saved_bo, ptr, size);
-			drm_dev_exit(idx);
-		}
+	if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+		memcpy_fromio(adev->vcn.inst[i].saved_bo, ptr, size);
+		drm_dev_exit(idx);
 	}
 
 	return 0;
 }
 
-int amdgpu_vcn_suspend(struct amdgpu_device *adev)
+int amdgpu_vcn_save_vcpu_bo(struct amdgpu_device *adev)
+{
+	int ret, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		ret = amdgpu_vcn_save_vcpu_bo_inst(adev, i);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+int amdgpu_vcn_suspend(struct amdgpu_device *adev, int i)
 {
 	bool in_ras_intr = amdgpu_ras_intr_triggered();
 
-	cancel_delayed_work_sync(&adev->vcn.idle_work);
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
 
-	/* err_event_athub will corrupt VCPU buffer, so we need to
+	/* err_event_athub and dpc recovery will corrupt VCPU buffer, so we need to
 	 * restore fw data and clear buffer in amdgpu_vcn_resume() */
-	if (in_ras_intr)
+	if (in_ras_intr || adev->pcie_reset_ctx.in_link_reset)
 		return 0;
 
-	return amdgpu_vcn_save_vcpu_bo(adev);
+	return amdgpu_vcn_save_vcpu_bo_inst(adev, i);
 }
 
-int amdgpu_vcn_resume(struct amdgpu_device *adev)
+int amdgpu_vcn_resume(struct amdgpu_device *adev, int i)
 {
 	unsigned int size;
 	void *ptr;
-	int i, idx;
+	int idx;
 
-	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
-		if (adev->vcn.harvest_config & (1 << i))
-			continue;
-		if (adev->vcn.inst[i].vcpu_bo == NULL)
-			return -EINVAL;
+	if (adev->vcn.harvest_config & (1 << i))
+		return 0;
+	if (adev->vcn.inst[i].vcpu_bo == NULL)
+		return -EINVAL;
+
+	size = amdgpu_bo_size(adev->vcn.inst[i].vcpu_bo);
+	ptr = adev->vcn.inst[i].cpu_addr;
 
-		size = amdgpu_bo_size(adev->vcn.inst[i].vcpu_bo);
-		ptr = adev->vcn.inst[i].cpu_addr;
+	if (adev->vcn.inst[i].saved_bo != NULL) {
+		if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+			memcpy_toio(ptr, adev->vcn.inst[i].saved_bo, size);
+			drm_dev_exit(idx);
+		}
+		kvfree(adev->vcn.inst[i].saved_bo);
+		adev->vcn.inst[i].saved_bo = NULL;
+	} else {
+		const struct common_firmware_header *hdr;
+		unsigned int offset;
 
-		if (adev->vcn.inst[i].saved_bo != NULL) {
+		hdr = (const struct common_firmware_header *)adev->vcn.inst[i].fw->data;
+		if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+			offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
 			if (drm_dev_enter(adev_to_drm(adev), &idx)) {
-				memcpy_toio(ptr, adev->vcn.inst[i].saved_bo, size);
+				memcpy_toio(adev->vcn.inst[i].cpu_addr,
+					    adev->vcn.inst[i].fw->data + offset,
+					    le32_to_cpu(hdr->ucode_size_bytes));
 				drm_dev_exit(idx);
 			}
-			kvfree(adev->vcn.inst[i].saved_bo);
-			adev->vcn.inst[i].saved_bo = NULL;
-		} else {
-			const struct common_firmware_header *hdr;
-			unsigned int offset;
-
-			hdr = (const struct common_firmware_header *)adev->vcn.inst[i].fw->data;
-			if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
-				offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
-				if (drm_dev_enter(adev_to_drm(adev), &idx)) {
-					memcpy_toio(adev->vcn.inst[i].cpu_addr,
-						    adev->vcn.inst[i].fw->data + offset,
-						    le32_to_cpu(hdr->ucode_size_bytes));
-					drm_dev_exit(idx);
-				}
-				size -= le32_to_cpu(hdr->ucode_size_bytes);
-				ptr += le32_to_cpu(hdr->ucode_size_bytes);
-			}
-			memset_io(ptr, 0, size);
+			size -= le32_to_cpu(hdr->ucode_size_bytes);
+			ptr += le32_to_cpu(hdr->ucode_size_bytes);
 		}
+		memset_io(ptr, 0, size);
 	}
+
 	return 0;
 }
 
+void amdgpu_vcn_get_profile(struct amdgpu_device *adev)
+{
+	int r;
+
+	mutex_lock(&adev->vcn.workload_profile_mutex);
+
+	if (adev->vcn.workload_profile_active) {
+		mutex_unlock(&adev->vcn.workload_profile_mutex);
+		return;
+	}
+	r = amdgpu_dpm_switch_power_profile(adev, PP_SMC_POWER_PROFILE_VIDEO,
+					    true);
+	if (r)
+		dev_warn(adev->dev,
+			 "(%d) failed to enable video power profile mode\n", r);
+	else
+		adev->vcn.workload_profile_active = true;
+	mutex_unlock(&adev->vcn.workload_profile_mutex);
+}
+
+void amdgpu_vcn_put_profile(struct amdgpu_device *adev)
+{
+	bool pg = true;
+	int r, i;
+
+	mutex_lock(&adev->vcn.workload_profile_mutex);
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.inst[i].cur_state != AMD_PG_STATE_GATE) {
+			pg = false;
+			break;
+		}
+	}
+
+	if (pg) {
+		r = amdgpu_dpm_switch_power_profile(
+			adev, PP_SMC_POWER_PROFILE_VIDEO, false);
+		if (r)
+			dev_warn(
+				adev->dev,
+				"(%d) failed to disable video power profile mode\n",
+				r);
+		else
+			adev->vcn.workload_profile_active = false;
+	}
+
+	mutex_unlock(&adev->vcn.workload_profile_mutex);
+}
+
 static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
 {
-	struct amdgpu_device *adev =
-		container_of(work, struct amdgpu_device, vcn.idle_work.work);
+	struct amdgpu_vcn_inst *vcn_inst =
+		container_of(work, struct amdgpu_vcn_inst, idle_work.work);
+	struct amdgpu_device *adev = vcn_inst->adev;
 	unsigned int fences = 0, fence[AMDGPU_MAX_VCN_INSTANCES] = {0};
-	unsigned int i, j;
-	int r = 0;
+	unsigned int i = vcn_inst->inst, j;
 
-	for (j = 0; j < adev->vcn.num_vcn_inst; ++j) {
-		if (adev->vcn.harvest_config & (1 << j))
-			continue;
-
-		for (i = 0; i < adev->vcn.num_enc_rings; ++i)
-			fence[j] += amdgpu_fence_count_emitted(&adev->vcn.inst[j].ring_enc[i]);
+	if (adev->vcn.harvest_config & (1 << i))
+		return;
 
-		/* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */
-		if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG &&
-		    !adev->vcn.using_unified_queue) {
-			struct dpg_pause_state new_state;
+	for (j = 0; j < adev->vcn.inst[i].num_enc_rings; ++j)
+		fence[i] += amdgpu_fence_count_emitted(&vcn_inst->ring_enc[j]);
 
-			if (fence[j] ||
-				unlikely(atomic_read(&adev->vcn.inst[j].dpg_enc_submission_cnt)))
-				new_state.fw_based = VCN_DPG_STATE__PAUSE;
-			else
-				new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+	/* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG &&
+	    !adev->vcn.inst[i].using_unified_queue) {
+		struct dpg_pause_state new_state;
 
-			adev->vcn.pause_dpg_mode(adev, j, &new_state);
-		}
+		if (fence[i] ||
+		    unlikely(atomic_read(&vcn_inst->dpg_enc_submission_cnt)))
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
 
-		fence[j] += amdgpu_fence_count_emitted(&adev->vcn.inst[j].ring_dec);
-		fences += fence[j];
+		adev->vcn.inst[i].pause_dpg_mode(vcn_inst, &new_state);
 	}
 
-	if (!fences && !atomic_read(&adev->vcn.total_submission_cnt)) {
-		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
-		       AMD_PG_STATE_GATE);
-		r = amdgpu_dpm_switch_power_profile(adev, PP_SMC_POWER_PROFILE_VIDEO,
-				false);
-		if (r)
-			dev_warn(adev->dev, "(%d) failed to disable video power profile mode\n", r);
+	fence[i] += amdgpu_fence_count_emitted(&vcn_inst->ring_dec);
+	fences += fence[i];
+
+	if (!fences && !atomic_read(&vcn_inst->total_submission_cnt)) {
+		mutex_lock(&vcn_inst->vcn_pg_lock);
+		vcn_inst->set_pg_state(vcn_inst, AMD_PG_STATE_GATE);
+		mutex_unlock(&vcn_inst->vcn_pg_lock);
+		amdgpu_vcn_put_profile(adev);
+
 	} else {
-		schedule_delayed_work(&adev->vcn.idle_work, VCN_IDLE_TIMEOUT);
+		schedule_delayed_work(&vcn_inst->idle_work, VCN_IDLE_TIMEOUT);
 	}
 }
 
 void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	int r = 0;
+	struct amdgpu_vcn_inst *vcn_inst = &adev->vcn.inst[ring->me];
 
-	atomic_inc(&adev->vcn.total_submission_cnt);
+	atomic_inc(&vcn_inst->total_submission_cnt);
 
-	if (!cancel_delayed_work_sync(&adev->vcn.idle_work)) {
-		r = amdgpu_dpm_switch_power_profile(adev, PP_SMC_POWER_PROFILE_VIDEO,
-				true);
-		if (r)
-			dev_warn(adev->dev, "(%d) failed to switch to video power profile mode\n", r);
-	}
+	cancel_delayed_work_sync(&vcn_inst->idle_work);
 
-	mutex_lock(&adev->vcn.vcn_pg_lock);
-	amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
-	       AMD_PG_STATE_UNGATE);
+	mutex_lock(&vcn_inst->vcn_pg_lock);
+	vcn_inst->set_pg_state(vcn_inst, AMD_PG_STATE_UNGATE);
 
 	/* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */
 	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG &&
-	    !adev->vcn.using_unified_queue) {
+	    !vcn_inst->using_unified_queue) {
 		struct dpg_pause_state new_state;
 
 		if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC) {
-			atomic_inc(&adev->vcn.inst[ring->me].dpg_enc_submission_cnt);
+			atomic_inc(&vcn_inst->dpg_enc_submission_cnt);
 			new_state.fw_based = VCN_DPG_STATE__PAUSE;
 		} else {
 			unsigned int fences = 0;
 			unsigned int i;
 
-			for (i = 0; i < adev->vcn.num_enc_rings; ++i)
-				fences += amdgpu_fence_count_emitted(&adev->vcn.inst[ring->me].ring_enc[i]);
+			for (i = 0; i < vcn_inst->num_enc_rings; ++i)
+				fences += amdgpu_fence_count_emitted(&vcn_inst->ring_enc[i]);
 
-			if (fences || atomic_read(&adev->vcn.inst[ring->me].dpg_enc_submission_cnt))
+			if (fences || atomic_read(&vcn_inst->dpg_enc_submission_cnt))
 				new_state.fw_based = VCN_DPG_STATE__PAUSE;
 			else
 				new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
 		}
 
-		adev->vcn.pause_dpg_mode(adev, ring->me, &new_state);
+		vcn_inst->pause_dpg_mode(vcn_inst, &new_state);
 	}
-	mutex_unlock(&adev->vcn.vcn_pg_lock);
+	mutex_unlock(&vcn_inst->vcn_pg_lock);
+	amdgpu_vcn_get_profile(adev);
 }
 
 void amdgpu_vcn_ring_end_use(struct amdgpu_ring *ring)
@@ -482,12 +542,13 @@ void amdgpu_vcn_ring_end_use(struct amdgpu_ring *ring)
 	/* Only set DPG pause for VCN3 or below, VCN4 and above will be handled by FW */
 	if (ring->adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG &&
 	    ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC &&
-	    !adev->vcn.using_unified_queue)
+	    !adev->vcn.inst[ring->me].using_unified_queue)
 		atomic_dec(&ring->adev->vcn.inst[ring->me].dpg_enc_submission_cnt);
 
-	atomic_dec(&ring->adev->vcn.total_submission_cnt);
+	atomic_dec(&ring->adev->vcn.inst[ring->me].total_submission_cnt);
 
-	schedule_delayed_work(&ring->adev->vcn.idle_work, VCN_IDLE_TIMEOUT);
+	schedule_delayed_work(&ring->adev->vcn.inst[ring->me].idle_work,
+			      VCN_IDLE_TIMEOUT);
 }
 
 int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring)
@@ -505,7 +566,7 @@ int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring)
 	r = amdgpu_ring_alloc(ring, 3);
 	if (r)
 		return r;
-	amdgpu_ring_write(ring, PACKET0(adev->vcn.internal.scratch9, 0));
+	amdgpu_ring_write(ring, PACKET0(adev->vcn.inst[ring->me].internal.scratch9, 0));
 	amdgpu_ring_write(ring, 0xDEADBEEF);
 	amdgpu_ring_commit(ring);
 	for (i = 0; i < adev->usec_timeout; i++) {
@@ -565,19 +626,19 @@ static int amdgpu_vcn_dec_send_msg(struct amdgpu_ring *ring,
 
 	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
 				     64, AMDGPU_IB_POOL_DIRECT,
-				     &job);
+				     &job, AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		goto err;
 
 	ib = &job->ibs[0];
-	ib->ptr[0] = PACKET0(adev->vcn.internal.data0, 0);
+	ib->ptr[0] = PACKET0(adev->vcn.inst[ring->me].internal.data0, 0);
 	ib->ptr[1] = addr;
-	ib->ptr[2] = PACKET0(adev->vcn.internal.data1, 0);
+	ib->ptr[2] = PACKET0(adev->vcn.inst[ring->me].internal.data1, 0);
 	ib->ptr[3] = addr >> 32;
-	ib->ptr[4] = PACKET0(adev->vcn.internal.cmd, 0);
+	ib->ptr[4] = PACKET0(adev->vcn.inst[ring->me].internal.cmd, 0);
 	ib->ptr[5] = 0;
 	for (i = 6; i < 16; i += 2) {
-		ib->ptr[i] = PACKET0(adev->vcn.internal.nop, 0);
+		ib->ptr[i] = PACKET0(adev->vcn.inst[ring->me].internal.nop, 0);
 		ib->ptr[i+1] = 0;
 	}
 	ib->length_dw = 16;
@@ -740,12 +801,12 @@ static int amdgpu_vcn_dec_sw_send_msg(struct amdgpu_ring *ring,
 	uint32_t ib_pack_in_dw;
 	int i, r;
 
-	if (adev->vcn.using_unified_queue)
+	if (adev->vcn.inst[ring->me].using_unified_queue)
 		ib_size_dw += 8;
 
 	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
 				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
-				     &job);
+				     &job, AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		goto err;
 
@@ -753,7 +814,7 @@ static int amdgpu_vcn_dec_sw_send_msg(struct amdgpu_ring *ring,
 	ib->length_dw = 0;
 
 	/* single queue headers */
-	if (adev->vcn.using_unified_queue) {
+	if (adev->vcn.inst[ring->me].using_unified_queue) {
 		ib_pack_in_dw = sizeof(struct amdgpu_vcn_decode_buffer) / sizeof(uint32_t)
 						+ 4 + 2; /* engine info + decoding ib in dw */
 		ib_checksum = amdgpu_vcn_unified_ring_ib_header(ib, ib_pack_in_dw, false);
@@ -772,7 +833,7 @@ static int amdgpu_vcn_dec_sw_send_msg(struct amdgpu_ring *ring,
 	for (i = ib->length_dw; i < ib_size_dw; ++i)
 		ib->ptr[i] = 0x0;
 
-	if (adev->vcn.using_unified_queue)
+	if (adev->vcn.inst[ring->me].using_unified_queue)
 		amdgpu_vcn_unified_ring_ib_checksum(&ib_checksum, ib_pack_in_dw);
 
 	r = amdgpu_job_submit_direct(job, ring, &f);
@@ -870,12 +931,12 @@ static int amdgpu_vcn_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
 	uint64_t addr;
 	int i, r;
 
-	if (adev->vcn.using_unified_queue)
+	if (adev->vcn.inst[ring->me].using_unified_queue)
 		ib_size_dw += 8;
 
 	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
 				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
-				     &job);
+				     &job, AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
@@ -884,7 +945,7 @@ static int amdgpu_vcn_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
 
 	ib->length_dw = 0;
 
-	if (adev->vcn.using_unified_queue)
+	if (adev->vcn.inst[ring->me].using_unified_queue)
 		ib_checksum = amdgpu_vcn_unified_ring_ib_header(ib, 0x11, true);
 
 	ib->ptr[ib->length_dw++] = 0x00000018;
@@ -906,7 +967,7 @@ static int amdgpu_vcn_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
 	for (i = ib->length_dw; i < ib_size_dw; ++i)
 		ib->ptr[i] = 0x0;
 
-	if (adev->vcn.using_unified_queue)
+	if (adev->vcn.inst[ring->me].using_unified_queue)
 		amdgpu_vcn_unified_ring_ib_checksum(&ib_checksum, 0x11);
 
 	r = amdgpu_job_submit_direct(job, ring, &f);
@@ -937,12 +998,12 @@ static int amdgpu_vcn_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
 	uint64_t addr;
 	int i, r;
 
-	if (adev->vcn.using_unified_queue)
+	if (adev->vcn.inst[ring->me].using_unified_queue)
 		ib_size_dw += 8;
 
 	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
 				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
-				     &job);
+				     &job, AMDGPU_KERNEL_JOB_ID_VCN_RING_TEST);
 	if (r)
 		return r;
 
@@ -951,7 +1012,7 @@ static int amdgpu_vcn_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
 
 	ib->length_dw = 0;
 
-	if (adev->vcn.using_unified_queue)
+	if (adev->vcn.inst[ring->me].using_unified_queue)
 		ib_checksum = amdgpu_vcn_unified_ring_ib_header(ib, 0x11, true);
 
 	ib->ptr[ib->length_dw++] = 0x00000018;
@@ -973,7 +1034,7 @@ static int amdgpu_vcn_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
 	for (i = ib->length_dw; i < ib_size_dw; ++i)
 		ib->ptr[i] = 0x0;
 
-	if (adev->vcn.using_unified_queue)
+	if (adev->vcn.inst[ring->me].using_unified_queue)
 		amdgpu_vcn_unified_ring_ib_checksum(&ib_checksum, 0x11);
 
 	r = amdgpu_job_submit_direct(job, ring, &f);
@@ -1058,36 +1119,32 @@ enum amdgpu_ring_priority_level amdgpu_vcn_get_enc_ring_prio(int ring)
 	}
 }
 
-void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev)
+void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev, int i)
 {
-	int i;
 	unsigned int idx;
 
 	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
 		const struct common_firmware_header *hdr;
 
-		for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
-			if (adev->vcn.harvest_config & (1 << i))
-				continue;
-
-			hdr = (const struct common_firmware_header *)adev->vcn.inst[i].fw->data;
-			/* currently only support 2 FW instances */
-			if (i >= 2) {
-				dev_info(adev->dev, "More then 2 VCN FW instances!\n");
-				break;
-			}
-			idx = AMDGPU_UCODE_ID_VCN + i;
-			adev->firmware.ucode[idx].ucode_id = idx;
-			adev->firmware.ucode[idx].fw = adev->vcn.inst[i].fw;
-			adev->firmware.fw_size +=
-				ALIGN(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE);
-
-			if (amdgpu_ip_version(adev, UVD_HWIP, 0) ==
-			    IP_VERSION(4, 0, 3) ||
-			    amdgpu_ip_version(adev, UVD_HWIP, 0) ==
-			    IP_VERSION(5, 0, 1))
-				break;
+		if (adev->vcn.harvest_config & (1 << i))
+			return;
+
+		if ((amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(4, 0, 3) ||
+		     amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(5, 0, 1))
+		    && (i > 0))
+			return;
+
+		hdr = (const struct common_firmware_header *)adev->vcn.inst[i].fw->data;
+		/* currently only support 2 FW instances */
+		if (i >= 2) {
+			dev_info(adev->dev, "More then 2 VCN FW instances!\n");
+			return;
 		}
+		idx = AMDGPU_UCODE_ID_VCN + i;
+		adev->firmware.ucode[idx].ucode_id = idx;
+		adev->firmware.ucode[idx].fw = adev->vcn.inst[i].fw;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE);
 	}
 }
 
@@ -1100,7 +1157,7 @@ static ssize_t amdgpu_debugfs_vcn_fwlog_read(struct file *f, char __user *buf,
 {
 	struct amdgpu_vcn_inst *vcn;
 	void *log_buf;
-	volatile struct amdgpu_vcn_fwlog *plog;
+	struct amdgpu_vcn_fwlog *plog;
 	unsigned int read_pos, write_pos, available, i, read_bytes = 0;
 	unsigned int read_num[2] = {0};
 
@@ -1113,7 +1170,7 @@ static ssize_t amdgpu_debugfs_vcn_fwlog_read(struct file *f, char __user *buf,
 
 	log_buf = vcn->fw_shared.cpu_addr + vcn->fw_shared.mem_size;
 
-	plog = (volatile struct amdgpu_vcn_fwlog *)log_buf;
+	plog = (struct amdgpu_vcn_fwlog *)log_buf;
 	read_pos = plog->rptr;
 	write_pos = plog->wptr;
 
@@ -1180,11 +1237,11 @@ void amdgpu_debugfs_vcn_fwlog_init(struct amdgpu_device *adev, uint8_t i,
 void amdgpu_vcn_fwlog_init(struct amdgpu_vcn_inst *vcn)
 {
 #if defined(CONFIG_DEBUG_FS)
-	volatile uint32_t *flag = vcn->fw_shared.cpu_addr;
+	uint32_t *flag = vcn->fw_shared.cpu_addr;
 	void *fw_log_cpu_addr = vcn->fw_shared.cpu_addr + vcn->fw_shared.mem_size;
 	uint64_t fw_log_gpu_addr = vcn->fw_shared.gpu_addr + vcn->fw_shared.mem_size;
-	volatile struct amdgpu_vcn_fwlog *log_buf = fw_log_cpu_addr;
-	volatile struct amdgpu_fw_shared_fw_logging *fw_log = vcn->fw_shared.cpu_addr
+	struct amdgpu_vcn_fwlog *log_buf = fw_log_cpu_addr;
+	struct amdgpu_fw_shared_fw_logging *fw_log = vcn->fw_shared.cpu_addr
 							 + vcn->fw_shared.log_offset;
 	*flag |= cpu_to_le32(AMDGPU_VCN_FW_LOGGING_FLAG);
 	fw_log->is_enabled = 1;
@@ -1390,10 +1447,191 @@ void amdgpu_debugfs_vcn_sched_mask_init(struct amdgpu_device *adev)
 	struct dentry *root = minor->debugfs_root;
 	char name[32];
 
-	if (adev->vcn.num_vcn_inst <= 1 || !adev->vcn.using_unified_queue)
+	if (adev->vcn.num_vcn_inst <= 1 || !adev->vcn.inst[0].using_unified_queue)
 		return;
 	sprintf(name, "amdgpu_vcn_sched_mask");
 	debugfs_create_file(name, 0600, root, adev,
 			    &amdgpu_debugfs_vcn_sched_mask_fops);
 #endif
 }
+
+/**
+ * vcn_set_powergating_state - set VCN block powergating state
+ *
+ * @ip_block: amdgpu_ip_block pointer
+ * @state: power gating state
+ *
+ * Set VCN block powergating state
+ */
+int vcn_set_powergating_state(struct amdgpu_ip_block *ip_block,
+			      enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int ret = 0, i;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+		struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[i];
+
+		ret |= vinst->set_pg_state(vinst, state);
+	}
+
+	return ret;
+}
+
+/**
+ * amdgpu_vcn_reset_engine - Reset a specific VCN engine
+ * @adev: Pointer to the AMDGPU device
+ * @instance_id: VCN engine instance to reset
+ *
+ * Returns: 0 on success, or a negative error code on failure.
+ */
+static int amdgpu_vcn_reset_engine(struct amdgpu_device *adev,
+				   uint32_t instance_id)
+{
+	struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[instance_id];
+	int r, i;
+
+	mutex_lock(&vinst->engine_reset_mutex);
+	/* Stop the scheduler's work queue for the dec and enc rings if they are running.
+	 * This ensures that no new tasks are submitted to the queues while
+	 * the reset is in progress.
+	 */
+	drm_sched_wqueue_stop(&vinst->ring_dec.sched);
+	for (i = 0; i < vinst->num_enc_rings; i++)
+		drm_sched_wqueue_stop(&vinst->ring_enc[i].sched);
+
+	/* Perform the VCN reset for the specified instance */
+	r = vinst->reset(vinst);
+	if (r)
+		goto unlock;
+	r = amdgpu_ring_test_ring(&vinst->ring_dec);
+	if (r)
+		goto unlock;
+	for (i = 0; i < vinst->num_enc_rings; i++) {
+		r = amdgpu_ring_test_ring(&vinst->ring_enc[i]);
+		if (r)
+			goto unlock;
+	}
+	amdgpu_fence_driver_force_completion(&vinst->ring_dec);
+	for (i = 0; i < vinst->num_enc_rings; i++)
+		amdgpu_fence_driver_force_completion(&vinst->ring_enc[i]);
+
+	/* Restart the scheduler's work queue for the dec and enc rings
+	 * if they were stopped by this function. This allows new tasks
+	 * to be submitted to the queues after the reset is complete.
+	 */
+	drm_sched_wqueue_start(&vinst->ring_dec.sched);
+	for (i = 0; i < vinst->num_enc_rings; i++)
+		drm_sched_wqueue_start(&vinst->ring_enc[i].sched);
+
+unlock:
+	mutex_unlock(&vinst->engine_reset_mutex);
+
+	return r;
+}
+
+/**
+ * amdgpu_vcn_ring_reset - Reset a VCN ring
+ * @ring: ring to reset
+ * @vmid: vmid of guilty job
+ * @timedout_fence: fence of timed out job
+ *
+ * This helper is for VCN blocks without unified queues because
+ * resetting the engine resets all queues in that case.  With
+ * unified queues we have one queue per engine.
+ * Returns: 0 on success, or a negative error code on failure.
+ */
+int amdgpu_vcn_ring_reset(struct amdgpu_ring *ring,
+			  unsigned int vmid,
+			  struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (adev->vcn.inst[ring->me].using_unified_queue)
+		return -EINVAL;
+
+	return amdgpu_vcn_reset_engine(adev, ring->me);
+}
+
+int amdgpu_vcn_reg_dump_init(struct amdgpu_device *adev,
+			     const struct amdgpu_hwip_reg_entry *reg, u32 count)
+{
+	adev->vcn.ip_dump = kcalloc(adev->vcn.num_vcn_inst * count,
+				     sizeof(uint32_t), GFP_KERNEL);
+	if (!adev->vcn.ip_dump)
+		return -ENOMEM;
+	adev->vcn.reg_list = reg;
+	adev->vcn.reg_count = count;
+
+	return 0;
+}
+
+static void amdgpu_vcn_reg_dump_fini(struct amdgpu_device *adev)
+{
+	kfree(adev->vcn.ip_dump);
+	adev->vcn.ip_dump = NULL;
+	adev->vcn.reg_list = NULL;
+	adev->vcn.reg_count = 0;
+}
+
+void amdgpu_vcn_dump_ip_state(struct amdgpu_ip_block *ip_block)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	bool is_powered;
+	u32 inst_off;
+
+	if (!adev->vcn.ip_dump)
+		return;
+
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		inst_off = i * adev->vcn.reg_count;
+		/* mmUVD_POWER_STATUS is always readable and is the first in reg_list */
+		adev->vcn.ip_dump[inst_off] =
+			RREG32(SOC15_REG_ENTRY_OFFSET_INST(adev->vcn.reg_list[0], i));
+		is_powered = (adev->vcn.ip_dump[inst_off] &
+			      UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF) !=
+			      UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+
+		if (is_powered)
+			for (j = 1; j < adev->vcn.reg_count; j++)
+				adev->vcn.ip_dump[inst_off + j] =
+				RREG32(SOC15_REG_ENTRY_OFFSET_INST(adev->vcn.reg_list[j], i));
+	}
+}
+
+void amdgpu_vcn_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
+{
+	struct amdgpu_device *adev = ip_block->adev;
+	int i, j;
+	bool is_powered;
+	u32 inst_off;
+
+	if (!adev->vcn.ip_dump)
+		return;
+
+	drm_printf(p, "num_instances:%d\n", adev->vcn.num_vcn_inst);
+	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (adev->vcn.harvest_config & (1 << i)) {
+			drm_printf(p, "\nHarvested Instance:VCN%d Skipping dump\n", i);
+			continue;
+		}
+
+		inst_off = i * adev->vcn.reg_count;
+		is_powered = (adev->vcn.ip_dump[inst_off] &
+			      UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF) !=
+			      UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+
+		if (is_powered) {
+			drm_printf(p, "\nActive Instance:VCN%d\n", i);
+			for (j = 0; j < adev->vcn.reg_count; j++)
+				drm_printf(p, "%-50s \t 0x%08x\n", adev->vcn.reg_list[j].reg_name,
+					   adev->vcn.ip_dump[inst_off + j]);
+		} else {
+			drm_printf(p, "\nInactive Instance:VCN%d\n", i);
+		}
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
index adaf4388ad28..82624b44e661 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
@@ -66,7 +66,6 @@
 #define VCN_ENC_CMD_REG_WAIT		0x0000000c
 
 #define VCN_AON_SOC_ADDRESS_2_0 	0x1f800
-#define VCN1_AON_SOC_ADDRESS_3_0 	0x48000
 #define VCN_VID_IP_ADDRESS_2_0		0x0
 #define VCN_AON_IP_ADDRESS_2_0		0x30000
 
@@ -101,7 +100,8 @@
 
 #define SOC15_DPG_MODE_OFFSET(ip, inst_idx, reg) 						\
 	({											\
-		uint32_t internal_reg_offset, addr;						\
+		/* To avoid a -Wunused-but-set-variable warning. */				\
+		uint32_t internal_reg_offset __maybe_unused, addr;				\
 		bool video_range, video1_range, aon_range, aon1_range;				\
 												\
 		addr = (adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg);		\
@@ -162,7 +162,8 @@
 
 #define SOC24_DPG_MODE_OFFSET(ip, inst_idx, reg)						\
 	({											\
-		uint32_t internal_reg_offset, addr;						\
+		/* To avoid a -Wunused-but-set-variable warning. */				\
+		uint32_t internal_reg_offset __maybe_unused, addr;				\
 		bool video_range, video1_range, aon_range, aon1_range;				\
 												\
 		addr = (adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg);		\
@@ -238,6 +239,14 @@
 
 #define AMDGPU_DRM_KEY_INJECT_WORKAROUND_VCNFW_ASD_HANDSHAKING 2
 
+struct amdgpu_hwip_reg_entry;
+
+enum amdgpu_vcn_caps {
+	AMDGPU_VCN_RRMT_ENABLED,
+};
+
+#define AMDGPU_VCN_CAPS(caps) BIT(AMDGPU_VCN_##caps)
+
 enum fw_queue_mode {
 	FW_QUEUE_RING_RESET = 1,
 	FW_QUEUE_DPG_HOLD_OFF = 2,
@@ -289,6 +298,8 @@ struct amdgpu_vcn_fw_shared {
 };
 
 struct amdgpu_vcn_inst {
+	struct amdgpu_device	*adev;
+	int			inst;
 	struct amdgpu_bo	*vcpu_bo;
 	void			*cpu_addr;
 	uint64_t		gpu_addr;
@@ -310,6 +321,22 @@ struct amdgpu_vcn_inst {
 	const struct firmware	*fw; /* VCN firmware */
 	uint8_t			vcn_config;
 	uint32_t		vcn_codec_disable_mask;
+	atomic_t		total_submission_cnt;
+	struct mutex		vcn_pg_lock;
+	enum amd_powergating_state cur_state;
+	struct delayed_work	idle_work;
+	unsigned		fw_version;
+	unsigned		num_enc_rings;
+	bool			indirect_sram;
+	struct amdgpu_vcn_reg	 internal;
+	struct mutex		vcn1_jpeg1_workaround;
+	int (*pause_dpg_mode)(struct amdgpu_vcn_inst *vinst,
+			      struct dpg_pause_state *new_state);
+	int (*set_pg_state)(struct amdgpu_vcn_inst *vinst,
+			    enum amd_powergating_state state);
+	int (*reset)(struct amdgpu_vcn_inst *vinst);
+	bool using_unified_queue;
+	struct mutex		engine_reset_mutex;
 };
 
 struct amdgpu_vcn_ras {
@@ -317,34 +344,30 @@ struct amdgpu_vcn_ras {
 };
 
 struct amdgpu_vcn {
-	unsigned		fw_version;
-	struct delayed_work	idle_work;
-	unsigned		num_enc_rings;
-	enum amd_powergating_state cur_state;
-	bool			indirect_sram;
-
 	uint8_t	num_vcn_inst;
 	struct amdgpu_vcn_inst	 inst[AMDGPU_MAX_VCN_INSTANCES];
-	struct amdgpu_vcn_reg	 internal;
-	struct mutex		 vcn_pg_lock;
-	struct mutex		vcn1_jpeg1_workaround;
-	atomic_t		 total_submission_cnt;
 
 	unsigned	harvest_config;
-	int (*pause_dpg_mode)(struct amdgpu_device *adev,
-		int inst_idx, struct dpg_pause_state *new_state);
 
 	struct ras_common_if    *ras_if;
 	struct amdgpu_vcn_ras   *ras;
 
 	uint16_t inst_mask;
 	uint8_t	num_inst_per_aid;
-	bool using_unified_queue;
 
 	/* IP reg dump */
 	uint32_t		*ip_dump;
 
 	uint32_t		supported_reset;
+	uint32_t		caps;
+
+	bool			per_inst_fw;
+	unsigned		fw_version;
+
+	bool			workload_profile_active;
+	struct mutex            workload_profile_mutex;
+	u32 reg_count;
+	const struct amdgpu_hwip_reg_entry *reg_list;
 };
 
 struct amdgpu_fw_shared_rb_ptrs_struct {
@@ -480,7 +503,7 @@ struct amdgpu_vcn5_fw_shared {
 	struct amdgpu_fw_shared_rb_setup rb_setup;
 	struct amdgpu_fw_shared_smu_interface_info smu_dpm_interface;
 	struct amdgpu_fw_shared_drm_key_wa drm_key_wa;
-	uint8_t pad3[9];
+	uint8_t pad3[404];
 };
 
 #define VCN_BLOCK_ENCODE_DISABLE_MASK 0x80
@@ -493,11 +516,11 @@ enum vcn_ring_type {
 	VCN_UNIFIED_RING,
 };
 
-int amdgpu_vcn_early_init(struct amdgpu_device *adev);
-int amdgpu_vcn_sw_init(struct amdgpu_device *adev);
-int amdgpu_vcn_sw_fini(struct amdgpu_device *adev);
-int amdgpu_vcn_suspend(struct amdgpu_device *adev);
-int amdgpu_vcn_resume(struct amdgpu_device *adev);
+int amdgpu_vcn_early_init(struct amdgpu_device *adev, int i);
+int amdgpu_vcn_sw_init(struct amdgpu_device *adev, int i);
+void amdgpu_vcn_sw_fini(struct amdgpu_device *adev, int i);
+int amdgpu_vcn_suspend(struct amdgpu_device *adev, int i);
+int amdgpu_vcn_resume(struct amdgpu_device *adev, int i);
 void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring);
 void amdgpu_vcn_ring_end_use(struct amdgpu_ring *ring);
 
@@ -515,7 +538,7 @@ int amdgpu_vcn_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout);
 
 enum amdgpu_ring_priority_level amdgpu_vcn_get_enc_ring_prio(int ring);
 
-void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev);
+void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev, int i);
 
 void amdgpu_vcn_fwlog_init(struct amdgpu_vcn_inst *vcn);
 void amdgpu_debugfs_vcn_fwlog_init(struct amdgpu_device *adev,
@@ -535,4 +558,16 @@ int amdgpu_vcn_sysfs_reset_mask_init(struct amdgpu_device *adev);
 void amdgpu_vcn_sysfs_reset_mask_fini(struct amdgpu_device *adev);
 void amdgpu_debugfs_vcn_sched_mask_init(struct amdgpu_device *adev);
 
+int vcn_set_powergating_state(struct amdgpu_ip_block *ip_block,
+			      enum amd_powergating_state state);
+int amdgpu_vcn_ring_reset(struct amdgpu_ring *ring,
+			  unsigned int vmid,
+			  struct amdgpu_fence *guilty_fence);
+int amdgpu_vcn_reg_dump_init(struct amdgpu_device *adev,
+			     const struct amdgpu_hwip_reg_entry *reg, u32 count);
+void amdgpu_vcn_dump_ip_state(struct amdgpu_ip_block *ip_block);
+void amdgpu_vcn_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p);
+void amdgpu_vcn_get_profile(struct amdgpu_device *adev);
+void amdgpu_vcn_put_profile(struct amdgpu_device *adev);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
index 0af469ec6fcc..47a6ce4fdc74 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
@@ -44,6 +44,18 @@
 		vf2pf_info->ucode_info[ucode].version = ver; \
 	} while (0)
 
+#define mmRCC_CONFIG_MEMSIZE    0xde3
+
+const char *amdgpu_virt_dynamic_crit_table_name[] = {
+	"IP DISCOVERY",
+	"VBIOS IMG",
+	"RAS TELEMETRY",
+	"DATA EXCHANGE",
+	"BAD PAGE INFO",
+	"INIT HEADER",
+	"LAST",
+};
+
 bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev)
 {
 	/* By now all MMIO pages except mailbox are blocked */
@@ -150,9 +162,10 @@ void amdgpu_virt_request_init_data(struct amdgpu_device *adev)
 		virt->ops->req_init_data(adev);
 
 	if (adev->virt.req_init_data_ver > 0)
-		DRM_INFO("host supports REQ_INIT_DATA handshake\n");
+		dev_info(adev->dev, "host supports REQ_INIT_DATA handshake of critical_region_version %d\n",
+				 adev->virt.req_init_data_ver);
 	else
-		DRM_WARN("host doesn't support REQ_INIT_DATA handshake\n");
+		dev_warn(adev->dev, "host doesn't support REQ_INIT_DATA handshake\n");
 }
 
 /**
@@ -205,12 +218,12 @@ int amdgpu_virt_alloc_mm_table(struct amdgpu_device *adev)
 				    &adev->virt.mm_table.gpu_addr,
 				    (void *)&adev->virt.mm_table.cpu_addr);
 	if (r) {
-		DRM_ERROR("failed to alloc mm table and error = %d.\n", r);
+		dev_err(adev->dev, "failed to alloc mm table and error = %d.\n", r);
 		return r;
 	}
 
 	memset((void *)adev->virt.mm_table.cpu_addr, 0, PAGE_SIZE);
-	DRM_INFO("MM table gpu addr = 0x%llx, cpu addr = %p.\n",
+	dev_info(adev->dev, "MM table gpu addr = 0x%llx, cpu addr = %p.\n",
 		 adev->virt.mm_table.gpu_addr,
 		 adev->virt.mm_table.cpu_addr);
 	return 0;
@@ -390,7 +403,9 @@ static void amdgpu_virt_ras_reserve_bps(struct amdgpu_device *adev)
 			if (amdgpu_bo_create_kernel_at(adev, bp << AMDGPU_GPU_PAGE_SHIFT,
 							AMDGPU_GPU_PAGE_SIZE,
 							&bo, NULL))
-				DRM_DEBUG("RAS WARN: reserve vram for retired page %llx fail\n", bp);
+				dev_dbg(adev->dev,
+						"RAS WARN: reserve vram for retired page %llx fail\n",
+						bp);
 			data->bps_bo[i] = bo;
 		}
 		data->last_reserved = i + 1;
@@ -598,8 +613,8 @@ static int amdgpu_virt_write_vf2pf_data(struct amdgpu_device *adev)
 	vf2pf_info->driver_cert = 0;
 	vf2pf_info->os_info.all = 0;
 
-	vf2pf_info->fb_usage =
-		ttm_resource_manager_usage(&adev->mman.vram_mgr.manager) >> 20;
+	vf2pf_info->fb_usage = ttm_resource_manager_used(&adev->mman.vram_mgr.manager) ?
+		 ttm_resource_manager_usage(&adev->mman.vram_mgr.manager) >> 20 : 0;
 	vf2pf_info->fb_vis_usage =
 		amdgpu_vram_mgr_vis_usage(&adev->mman.vram_mgr) >> 20;
 	vf2pf_info->fb_size = adev->gmc.real_vram_size >> 20;
@@ -614,10 +629,11 @@ static int amdgpu_virt_write_vf2pf_data(struct amdgpu_device *adev)
 	vf2pf_info->decode_usage = 0;
 
 	vf2pf_info->dummy_page_addr = (uint64_t)adev->dummy_page_addr;
-	vf2pf_info->mes_info_addr = (uint64_t)adev->mes.resource_1_gpu_addr;
-
-	if (adev->mes.resource_1) {
-		vf2pf_info->mes_info_size = adev->mes.resource_1->tbo.base.size;
+	if (amdgpu_sriov_is_mes_info_enable(adev)) {
+		vf2pf_info->mes_info_addr =
+			(uint64_t)(adev->mes.resource_1_gpu_addr[0] + AMDGPU_GPU_PAGE_SIZE);
+		vf2pf_info->mes_info_size =
+			adev->mes.resource_1[0]->tbo.base.size - AMDGPU_GPU_PAGE_SIZE;
 	}
 	vf2pf_info->checksum =
 		amd_sriov_msg_checksum(
@@ -657,10 +673,34 @@ out:
 	schedule_delayed_work(&(adev->virt.vf2pf_work), adev->virt.vf2pf_update_interval_ms);
 }
 
+static int amdgpu_virt_read_exchange_data_from_mem(struct amdgpu_device *adev, uint32_t *pfvf_data)
+{
+	uint32_t dataexchange_offset =
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].offset;
+	uint32_t dataexchange_size =
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].size_kb << 10;
+	uint64_t pos = 0;
+
+	dev_info(adev->dev,
+			"Got data exchange info from dynamic crit_region_table at offset 0x%x with size of 0x%x bytes.\n",
+			dataexchange_offset, dataexchange_size);
+
+	if (!IS_ALIGNED(dataexchange_offset, 4) || !IS_ALIGNED(dataexchange_size, 4)) {
+		dev_err(adev->dev, "Data exchange data not aligned to 4 bytes\n");
+		return -EINVAL;
+	}
+
+	pos = (uint64_t)dataexchange_offset;
+	amdgpu_device_vram_access(adev, pos, pfvf_data,
+			dataexchange_size, false);
+
+	return 0;
+}
+
 void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev)
 {
 	if (adev->virt.vf2pf_update_interval_ms != 0) {
-		DRM_INFO("clean up the vf2pf work item\n");
+		dev_info(adev->dev, "clean up the vf2pf work item\n");
 		cancel_delayed_work_sync(&adev->virt.vf2pf_work);
 		adev->virt.vf2pf_update_interval_ms = 0;
 	}
@@ -668,13 +708,15 @@ void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev)
 
 void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev)
 {
+	uint32_t *pfvf_data = NULL;
+
 	adev->virt.fw_reserve.p_pf2vf = NULL;
 	adev->virt.fw_reserve.p_vf2pf = NULL;
 	adev->virt.vf2pf_update_interval_ms = 0;
 	adev->virt.vf2pf_update_retry_cnt = 0;
 
 	if (adev->mman.fw_vram_usage_va && adev->mman.drv_vram_usage_va) {
-		DRM_WARN("Currently fw_vram and drv_vram should not have values at the same time!");
+		dev_warn(adev->dev, "Currently fw_vram and drv_vram should not have values at the same time!");
 	} else if (adev->mman.fw_vram_usage_va || adev->mman.drv_vram_usage_va) {
 		/* go through this logic in ip_init and reset to init workqueue*/
 		amdgpu_virt_exchange_data(adev);
@@ -683,11 +725,34 @@ void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev)
 		schedule_delayed_work(&(adev->virt.vf2pf_work), msecs_to_jiffies(adev->virt.vf2pf_update_interval_ms));
 	} else if (adev->bios != NULL) {
 		/* got through this logic in early init stage to get necessary flags, e.g. rlcg_acc related*/
-		adev->virt.fw_reserve.p_pf2vf =
-			(struct amd_sriov_msg_pf2vf_info_header *)
-			(adev->bios + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10));
+		if (adev->virt.req_init_data_ver == GPU_CRIT_REGION_V2) {
+			pfvf_data =
+				kzalloc(adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].size_kb << 10,
+					GFP_KERNEL);
+			if (!pfvf_data) {
+				dev_err(adev->dev, "Failed to allocate memory for pfvf_data\n");
+				return;
+			}
 
-		amdgpu_virt_read_pf2vf_data(adev);
+			if (amdgpu_virt_read_exchange_data_from_mem(adev, pfvf_data))
+				goto free_pfvf_data;
+
+			adev->virt.fw_reserve.p_pf2vf =
+				(struct amd_sriov_msg_pf2vf_info_header *)pfvf_data;
+
+			amdgpu_virt_read_pf2vf_data(adev);
+
+free_pfvf_data:
+			kfree(pfvf_data);
+			pfvf_data = NULL;
+			adev->virt.fw_reserve.p_pf2vf = NULL;
+		} else {
+			adev->virt.fw_reserve.p_pf2vf =
+				(struct amd_sriov_msg_pf2vf_info_header *)
+				(adev->bios + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB_V1 << 10));
+
+			amdgpu_virt_read_pf2vf_data(adev);
+		}
 	}
 }
 
@@ -700,23 +765,38 @@ void amdgpu_virt_exchange_data(struct amdgpu_device *adev)
 
 	if (adev->mman.fw_vram_usage_va || adev->mman.drv_vram_usage_va) {
 		if (adev->mman.fw_vram_usage_va) {
-			adev->virt.fw_reserve.p_pf2vf =
-				(struct amd_sriov_msg_pf2vf_info_header *)
-				(adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10));
-			adev->virt.fw_reserve.p_vf2pf =
-				(struct amd_sriov_msg_vf2pf_info_header *)
-				(adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB << 10));
-			adev->virt.fw_reserve.ras_telemetry =
-				(adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB << 10));
+			if (adev->virt.req_init_data_ver == GPU_CRIT_REGION_V2) {
+				adev->virt.fw_reserve.p_pf2vf =
+					(struct amd_sriov_msg_pf2vf_info_header *)
+					(adev->mman.fw_vram_usage_va +
+					adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].offset);
+				adev->virt.fw_reserve.p_vf2pf =
+					(struct amd_sriov_msg_vf2pf_info_header *)
+					(adev->mman.fw_vram_usage_va +
+					adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].offset +
+					(AMD_SRIOV_MSG_SIZE_KB << 10));
+				adev->virt.fw_reserve.ras_telemetry =
+					(adev->mman.fw_vram_usage_va +
+					adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID].offset);
+			} else {
+				adev->virt.fw_reserve.p_pf2vf =
+					(struct amd_sriov_msg_pf2vf_info_header *)
+					(adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB_V1 << 10));
+				adev->virt.fw_reserve.p_vf2pf =
+					(struct amd_sriov_msg_vf2pf_info_header *)
+					(adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB_V1 << 10));
+				adev->virt.fw_reserve.ras_telemetry =
+					(adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB_V1 << 10));
+			}
 		} else if (adev->mman.drv_vram_usage_va) {
 			adev->virt.fw_reserve.p_pf2vf =
 				(struct amd_sriov_msg_pf2vf_info_header *)
-				(adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10));
+				(adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB_V1 << 10));
 			adev->virt.fw_reserve.p_vf2pf =
 				(struct amd_sriov_msg_vf2pf_info_header *)
-				(adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB << 10));
+				(adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB_V1 << 10));
 			adev->virt.fw_reserve.ras_telemetry =
-				(adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB << 10));
+				(adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB_V1 << 10));
 		}
 
 		amdgpu_virt_read_pf2vf_data(adev);
@@ -739,7 +819,7 @@ void amdgpu_virt_exchange_data(struct amdgpu_device *adev)
 	}
 }
 
-void amdgpu_detect_virtualization(struct amdgpu_device *adev)
+static u32 amdgpu_virt_init_detect_asic(struct amdgpu_device *adev)
 {
 	uint32_t reg;
 
@@ -775,8 +855,17 @@ void amdgpu_detect_virtualization(struct amdgpu_device *adev)
 			adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
 	}
 
+	return reg;
+}
+
+static bool amdgpu_virt_init_req_data(struct amdgpu_device *adev, u32 reg)
+{
+	bool is_sriov = false;
+
 	/* we have the ability to check now */
 	if (amdgpu_sriov_vf(adev)) {
+		is_sriov = true;
+
 		switch (adev->asic_type) {
 		case CHIP_TONGA:
 		case CHIP_FIJI:
@@ -805,10 +894,252 @@ void amdgpu_detect_virtualization(struct amdgpu_device *adev)
 			amdgpu_virt_request_init_data(adev);
 			break;
 		default: /* other chip doesn't support SRIOV */
-			DRM_ERROR("Unknown asic type: %d!\n", adev->asic_type);
+			is_sriov = false;
+			dev_err(adev->dev, "Unknown asic type: %d!\n", adev->asic_type);
 			break;
 		}
 	}
+
+	return is_sriov;
+}
+
+static void amdgpu_virt_init_ras(struct amdgpu_device *adev)
+{
+	ratelimit_state_init(&adev->virt.ras.ras_error_cnt_rs, 5 * HZ, 1);
+	ratelimit_state_init(&adev->virt.ras.ras_cper_dump_rs, 5 * HZ, 1);
+	ratelimit_state_init(&adev->virt.ras.ras_chk_criti_rs, 5 * HZ, 1);
+
+	ratelimit_set_flags(&adev->virt.ras.ras_error_cnt_rs,
+			    RATELIMIT_MSG_ON_RELEASE);
+	ratelimit_set_flags(&adev->virt.ras.ras_cper_dump_rs,
+			    RATELIMIT_MSG_ON_RELEASE);
+	ratelimit_set_flags(&adev->virt.ras.ras_chk_criti_rs,
+			    RATELIMIT_MSG_ON_RELEASE);
+
+	mutex_init(&adev->virt.ras.ras_telemetry_mutex);
+	mutex_init(&adev->virt.access_req_mutex);
+
+	adev->virt.ras.cper_rptr = 0;
+}
+
+static uint8_t amdgpu_virt_crit_region_calc_checksum(uint8_t *buf_start, uint8_t *buf_end)
+{
+	uint32_t sum = 0;
+
+	if (buf_start >= buf_end)
+		return 0;
+
+	for (; buf_start < buf_end; buf_start++)
+		sum += buf_start[0];
+
+	return 0xffffffff - sum;
+}
+
+int amdgpu_virt_init_critical_region(struct amdgpu_device *adev)
+{
+	struct amd_sriov_msg_init_data_header *init_data_hdr = NULL;
+	u64 init_hdr_offset = adev->virt.init_data_header.offset;
+	u64 init_hdr_size = (u64)adev->virt.init_data_header.size_kb << 10; /* KB → bytes */
+	u64 vram_size;
+	u64 end;
+	int r = 0;
+	uint8_t checksum = 0;
+
+	/* Skip below init if critical region version != v2 */
+	if (adev->virt.req_init_data_ver != GPU_CRIT_REGION_V2)
+		return 0;
+
+	if (init_hdr_offset < 0) {
+		dev_err(adev->dev, "Invalid init header offset\n");
+		return -EINVAL;
+	}
+
+	vram_size = RREG32(mmRCC_CONFIG_MEMSIZE);
+	if (!vram_size || vram_size == U32_MAX)
+		return -EINVAL;
+	vram_size <<= 20;
+
+	if (check_add_overflow(init_hdr_offset, init_hdr_size, &end) || end > vram_size) {
+		dev_err(adev->dev, "init_data_header exceeds VRAM size, exiting\n");
+		return -EINVAL;
+	}
+
+	/* Allocate for init_data_hdr */
+	init_data_hdr = kzalloc(sizeof(struct amd_sriov_msg_init_data_header), GFP_KERNEL);
+	if (!init_data_hdr)
+		return -ENOMEM;
+
+	amdgpu_device_vram_access(adev, (uint64_t)init_hdr_offset, (uint32_t *)init_data_hdr,
+					sizeof(struct amd_sriov_msg_init_data_header), false);
+
+	/* Table validation */
+	if (strncmp(init_data_hdr->signature,
+				AMDGPU_SRIOV_CRIT_DATA_SIGNATURE,
+				AMDGPU_SRIOV_CRIT_DATA_SIG_LEN) != 0) {
+		dev_err(adev->dev, "Invalid init data signature: %.4s\n",
+			init_data_hdr->signature);
+		r = -EINVAL;
+		goto out;
+	}
+
+	checksum = amdgpu_virt_crit_region_calc_checksum(
+			(uint8_t *)&init_data_hdr->initdata_offset,
+			(uint8_t *)init_data_hdr +
+			sizeof(struct amd_sriov_msg_init_data_header));
+	if (checksum != init_data_hdr->checksum) {
+		dev_err(adev->dev, "Found unmatching checksum from calculation 0x%x and init_data 0x%x\n",
+				checksum, init_data_hdr->checksum);
+		r = -EINVAL;
+		goto out;
+	}
+
+	memset(&adev->virt.crit_regn, 0, sizeof(adev->virt.crit_regn));
+	memset(adev->virt.crit_regn_tbl, 0, sizeof(adev->virt.crit_regn_tbl));
+
+	adev->virt.crit_regn.offset = init_data_hdr->initdata_offset;
+	adev->virt.crit_regn.size_kb = init_data_hdr->initdata_size_in_kb;
+
+	/* Validation and initialization for each table entry */
+	if (IS_SRIOV_CRIT_REGN_ENTRY_VALID(init_data_hdr, AMD_SRIOV_MSG_IPD_TABLE_ID)) {
+		if (!init_data_hdr->ip_discovery_size_in_kb ||
+				init_data_hdr->ip_discovery_size_in_kb > DISCOVERY_TMR_SIZE) {
+			dev_err(adev->dev, "Invalid %s size: 0x%x\n",
+				amdgpu_virt_dynamic_crit_table_name[AMD_SRIOV_MSG_IPD_TABLE_ID],
+				init_data_hdr->ip_discovery_size_in_kb);
+			r = -EINVAL;
+			goto out;
+		}
+
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_IPD_TABLE_ID].offset =
+			init_data_hdr->ip_discovery_offset;
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_IPD_TABLE_ID].size_kb =
+			init_data_hdr->ip_discovery_size_in_kb;
+	}
+
+	if (IS_SRIOV_CRIT_REGN_ENTRY_VALID(init_data_hdr, AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID)) {
+		if (!init_data_hdr->vbios_img_size_in_kb) {
+			dev_err(adev->dev, "Invalid %s size: 0x%x\n",
+				amdgpu_virt_dynamic_crit_table_name[AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID],
+				init_data_hdr->vbios_img_size_in_kb);
+			r = -EINVAL;
+			goto out;
+		}
+
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID].offset =
+			init_data_hdr->vbios_img_offset;
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID].size_kb =
+			init_data_hdr->vbios_img_size_in_kb;
+	}
+
+	if (IS_SRIOV_CRIT_REGN_ENTRY_VALID(init_data_hdr, AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID)) {
+		if (!init_data_hdr->ras_tele_info_size_in_kb) {
+			dev_err(adev->dev, "Invalid %s size: 0x%x\n",
+				amdgpu_virt_dynamic_crit_table_name[AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID],
+				init_data_hdr->ras_tele_info_size_in_kb);
+			r = -EINVAL;
+			goto out;
+		}
+
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID].offset =
+			init_data_hdr->ras_tele_info_offset;
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID].size_kb =
+			init_data_hdr->ras_tele_info_size_in_kb;
+	}
+
+	if (IS_SRIOV_CRIT_REGN_ENTRY_VALID(init_data_hdr, AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID)) {
+		if (!init_data_hdr->dataexchange_size_in_kb) {
+			dev_err(adev->dev, "Invalid %s size: 0x%x\n",
+				amdgpu_virt_dynamic_crit_table_name[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID],
+				init_data_hdr->dataexchange_size_in_kb);
+			r = -EINVAL;
+			goto out;
+		}
+
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].offset =
+			init_data_hdr->dataexchange_offset;
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].size_kb =
+			init_data_hdr->dataexchange_size_in_kb;
+	}
+
+	if (IS_SRIOV_CRIT_REGN_ENTRY_VALID(init_data_hdr, AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID)) {
+		if (!init_data_hdr->bad_page_size_in_kb) {
+			dev_err(adev->dev, "Invalid %s size: 0x%x\n",
+				amdgpu_virt_dynamic_crit_table_name[AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID],
+				init_data_hdr->bad_page_size_in_kb);
+			r = -EINVAL;
+			goto out;
+		}
+
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID].offset =
+			init_data_hdr->bad_page_info_offset;
+		adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID].size_kb =
+			init_data_hdr->bad_page_size_in_kb;
+	}
+
+	/* Validation for critical region info */
+	if (adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_IPD_TABLE_ID].size_kb > DISCOVERY_TMR_SIZE) {
+		dev_err(adev->dev, "Invalid IP discovery size: 0x%x\n",
+				adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_IPD_TABLE_ID].size_kb);
+		r = -EINVAL;
+		goto out;
+	}
+
+	/* reserved memory starts from crit region base offset with the size of 5MB */
+	adev->mman.fw_vram_usage_start_offset = adev->virt.crit_regn.offset;
+	adev->mman.fw_vram_usage_size = adev->virt.crit_regn.size_kb << 10;
+	dev_info(adev->dev,
+		"critical region v%d requested to reserve memory start at %08llx with %llu KB.\n",
+			init_data_hdr->version,
+			adev->mman.fw_vram_usage_start_offset,
+			adev->mman.fw_vram_usage_size >> 10);
+
+	adev->virt.is_dynamic_crit_regn_enabled = true;
+
+out:
+	kfree(init_data_hdr);
+	init_data_hdr = NULL;
+
+	return r;
+}
+
+int amdgpu_virt_get_dynamic_data_info(struct amdgpu_device *adev,
+	int data_id, uint8_t *binary, u32 *size)
+{
+	uint32_t data_offset = 0;
+	uint32_t data_size = 0;
+	enum amd_sriov_msg_table_id_enum data_table_id = data_id;
+
+	if (data_table_id >= AMD_SRIOV_MSG_MAX_TABLE_ID)
+		return -EINVAL;
+
+	data_offset = adev->virt.crit_regn_tbl[data_table_id].offset;
+	data_size = adev->virt.crit_regn_tbl[data_table_id].size_kb << 10;
+
+	/* Validate on input params */
+	if (!binary || !size || *size < (uint64_t)data_size)
+		return -EINVAL;
+
+	/* Proceed to copy the dynamic content */
+	amdgpu_device_vram_access(adev,
+			(uint64_t)data_offset, (uint32_t *)binary, data_size, false);
+	*size = (uint64_t)data_size;
+
+	dev_dbg(adev->dev,
+		"Got %s info from dynamic crit_region_table at offset 0x%x with size of 0x%x bytes.\n",
+		amdgpu_virt_dynamic_crit_table_name[data_id], data_offset, data_size);
+
+	return 0;
+}
+
+void amdgpu_virt_init(struct amdgpu_device *adev)
+{
+	bool is_sriov = false;
+	uint32_t reg = amdgpu_virt_init_detect_asic(adev);
+
+	is_sriov = amdgpu_virt_init_req_data(adev, reg);
+
+	if (is_sriov)
+		amdgpu_virt_init_ras(adev);
 }
 
 static bool amdgpu_virt_access_debugfs_is_mmio(struct amdgpu_device *adev)
@@ -1017,6 +1348,7 @@ u32 amdgpu_virt_rlcg_reg_rw(struct amdgpu_device *adev, u32 offset, u32 v, u32 f
 	void *scratch_reg2;
 	void *scratch_reg3;
 	void *spare_int;
+	unsigned long flags;
 
 	if (!adev->gfx.rlc.rlcg_reg_access_supported) {
 		dev_err(adev->dev,
@@ -1038,7 +1370,7 @@ u32 amdgpu_virt_rlcg_reg_rw(struct amdgpu_device *adev, u32 offset, u32 v, u32 f
 	scratch_reg2 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg2;
 	scratch_reg3 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg3;
 
-	mutex_lock(&adev->virt.rlcg_reg_lock);
+	spin_lock_irqsave(&adev->virt.rlcg_reg_lock, flags);
 
 	if (reg_access_ctrl->spare_int)
 		spare_int = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->spare_int;
@@ -1097,7 +1429,7 @@ u32 amdgpu_virt_rlcg_reg_rw(struct amdgpu_device *adev, u32 offset, u32 v, u32 f
 
 	ret = readl(scratch_reg0);
 
-	mutex_unlock(&adev->virt.rlcg_reg_lock);
+	spin_unlock_irqrestore(&adev->virt.rlcg_reg_lock, flags);
 
 	return ret;
 }
@@ -1246,7 +1578,8 @@ amdgpu_ras_block_to_sriov(struct amdgpu_device *adev, enum amdgpu_ras_block bloc
 	case AMDGPU_RAS_BLOCK__MPIO:
 		return RAS_TELEMETRY_GPU_BLOCK_MPIO;
 	default:
-		dev_err(adev->dev, "Unsupported SRIOV RAS telemetry block 0x%x\n", block);
+		dev_warn(adev->dev, "Unsupported SRIOV RAS telemetry block 0x%x\n",
+			      block);
 		return RAS_TELEMETRY_GPU_BLOCK_COUNT;
 	}
 }
@@ -1260,7 +1593,7 @@ static int amdgpu_virt_cache_host_error_counts(struct amdgpu_device *adev,
 	checksum = host_telemetry->header.checksum;
 	used_size = host_telemetry->header.used_size;
 
-	if (used_size > (AMD_SRIOV_RAS_TELEMETRY_SIZE_KB << 10))
+	if (used_size > (AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1 << 10))
 		return 0;
 
 	tmp = kmemdup(&host_telemetry->body.error_count, used_size, GFP_KERNEL);
@@ -1282,14 +1615,19 @@ static int amdgpu_virt_req_ras_err_count_internal(struct amdgpu_device *adev, bo
 {
 	struct amdgpu_virt *virt = &adev->virt;
 
+	if (!virt->ops || !virt->ops->req_ras_err_count)
+		return -EOPNOTSUPP;
+
 	/* Host allows 15 ras telemetry requests per 60 seconds. Afterwhich, the Host
 	 * will ignore incoming guest messages. Ratelimit the guest messages to
 	 * prevent guest self DOS.
 	 */
-	if (__ratelimit(&adev->virt.ras_telemetry_rs) || force_update) {
+	if (__ratelimit(&virt->ras.ras_error_cnt_rs) || force_update) {
+		mutex_lock(&virt->ras.ras_telemetry_mutex);
 		if (!virt->ops->req_ras_err_count(adev))
 			amdgpu_virt_cache_host_error_counts(adev,
-				adev->virt.fw_reserve.ras_telemetry);
+				virt->fw_reserve.ras_telemetry);
+		mutex_unlock(&virt->ras.ras_telemetry_mutex);
 	}
 
 	return 0;
@@ -1320,6 +1658,103 @@ int amdgpu_virt_req_ras_err_count(struct amdgpu_device *adev, enum amdgpu_ras_bl
 	return 0;
 }
 
+static int
+amdgpu_virt_write_cpers_to_ring(struct amdgpu_device *adev,
+				struct amdsriov_ras_telemetry *host_telemetry,
+				u32 *more)
+{
+	struct amd_sriov_ras_cper_dump *cper_dump = NULL;
+	struct cper_hdr *entry = NULL;
+	struct amdgpu_ring *ring = &adev->cper.ring_buf;
+	uint32_t checksum, used_size, i;
+	int ret = 0;
+
+	checksum = host_telemetry->header.checksum;
+	used_size = host_telemetry->header.used_size;
+
+	if (used_size > (AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1 << 10))
+		return -EINVAL;
+
+	cper_dump = kmemdup(&host_telemetry->body.cper_dump, used_size, GFP_KERNEL);
+	if (!cper_dump)
+		return -ENOMEM;
+
+	if (checksum != amd_sriov_msg_checksum(cper_dump, used_size, 0, 0)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	*more = cper_dump->more;
+
+	if (cper_dump->wptr < adev->virt.ras.cper_rptr) {
+		dev_warn(
+			adev->dev,
+			"guest specified rptr that was too high! guest rptr: 0x%llx, host rptr: 0x%llx\n",
+			adev->virt.ras.cper_rptr, cper_dump->wptr);
+
+		adev->virt.ras.cper_rptr = cper_dump->wptr;
+		goto out;
+	}
+
+	entry = (struct cper_hdr *)&cper_dump->buf[0];
+
+	for (i = 0; i < cper_dump->count; i++) {
+		amdgpu_cper_ring_write(ring, entry, entry->record_length);
+		entry = (struct cper_hdr *)((char *)entry +
+					    entry->record_length);
+	}
+
+	if (cper_dump->overflow_count)
+		dev_warn(adev->dev,
+			 "host reported CPER overflow of 0x%llx entries!\n",
+			 cper_dump->overflow_count);
+
+	adev->virt.ras.cper_rptr = cper_dump->wptr;
+out:
+	kfree(cper_dump);
+
+	return ret;
+}
+
+static int amdgpu_virt_req_ras_cper_dump_internal(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	int ret = 0;
+	uint32_t more = 0;
+
+	if (!virt->ops || !virt->ops->req_ras_cper_dump)
+		return -EOPNOTSUPP;
+
+	do {
+		if (!virt->ops->req_ras_cper_dump(adev, virt->ras.cper_rptr))
+			ret = amdgpu_virt_write_cpers_to_ring(
+				adev, virt->fw_reserve.ras_telemetry, &more);
+		else
+			ret = 0;
+	} while (more && !ret);
+
+	return ret;
+}
+
+int amdgpu_virt_req_ras_cper_dump(struct amdgpu_device *adev, bool force_update)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	int ret = 0;
+
+	if (!amdgpu_sriov_ras_cper_en(adev))
+		return -EOPNOTSUPP;
+
+	if ((__ratelimit(&virt->ras.ras_cper_dump_rs) || force_update) &&
+	    down_read_trylock(&adev->reset_domain->sem)) {
+		mutex_lock(&virt->ras.ras_telemetry_mutex);
+		ret = amdgpu_virt_req_ras_cper_dump_internal(adev);
+		mutex_unlock(&virt->ras.ras_telemetry_mutex);
+		up_read(&adev->reset_domain->sem);
+	}
+
+	return ret;
+}
+
 int amdgpu_virt_ras_telemetry_post_reset(struct amdgpu_device *adev)
 {
 	unsigned long ue_count, ce_count;
@@ -1331,3 +1766,82 @@ int amdgpu_virt_ras_telemetry_post_reset(struct amdgpu_device *adev)
 
 	return 0;
 }
+
+bool amdgpu_virt_ras_telemetry_block_en(struct amdgpu_device *adev,
+					enum amdgpu_ras_block block)
+{
+	enum amd_sriov_ras_telemetry_gpu_block sriov_block;
+
+	sriov_block = amdgpu_ras_block_to_sriov(adev, block);
+
+	if (sriov_block >= RAS_TELEMETRY_GPU_BLOCK_COUNT ||
+	    !amdgpu_sriov_ras_telemetry_block_en(adev, sriov_block))
+		return false;
+
+	return true;
+}
+
+/*
+ * amdgpu_virt_request_bad_pages() - request bad pages
+ * @adev: amdgpu device.
+ * Send command to GPU hypervisor to write new bad pages into the shared PF2VF region
+ */
+void amdgpu_virt_request_bad_pages(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+
+	if (virt->ops && virt->ops->req_bad_pages)
+		virt->ops->req_bad_pages(adev);
+}
+
+static int amdgpu_virt_cache_chk_criti_hit(struct amdgpu_device *adev,
+					   struct amdsriov_ras_telemetry *host_telemetry,
+					   bool *hit)
+{
+	struct amd_sriov_ras_chk_criti *tmp = NULL;
+	uint32_t checksum, used_size;
+
+	checksum = host_telemetry->header.checksum;
+	used_size = host_telemetry->header.used_size;
+
+	if (used_size > (AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1 << 10))
+		return 0;
+
+	tmp = kmemdup(&host_telemetry->body.chk_criti, used_size, GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+	if (checksum != amd_sriov_msg_checksum(tmp, used_size, 0, 0))
+		goto out;
+
+	if (hit)
+		*hit = tmp->hit ? true : false;
+
+out:
+	kfree(tmp);
+
+	return 0;
+}
+
+int amdgpu_virt_check_vf_critical_region(struct amdgpu_device *adev, u64 addr, bool *hit)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	int r = -EPERM;
+
+	if (!virt->ops || !virt->ops->req_ras_chk_criti)
+		return -EOPNOTSUPP;
+
+	/* Host allows 15 ras telemetry requests per 60 seconds. Afterwhich, the Host
+	 * will ignore incoming guest messages. Ratelimit the guest messages to
+	 * prevent guest self DOS.
+	 */
+	if (__ratelimit(&virt->ras.ras_chk_criti_rs)) {
+		mutex_lock(&virt->ras.ras_telemetry_mutex);
+		if (!virt->ops->req_ras_chk_criti(adev, addr))
+			r = amdgpu_virt_cache_chk_criti_hit(
+				adev, virt->fw_reserve.ras_telemetry, hit);
+		mutex_unlock(&virt->ras.ras_telemetry_mutex);
+	}
+
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
index 5381b8d596e6..01d5bca2dee1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
@@ -54,6 +54,12 @@
 
 #define AMDGPU_VF2PF_UPDATE_MAX_RETRY_LIMIT 2
 
+/* Signature used to validate the SR-IOV dynamic critical region init data header ("INDA") */
+#define AMDGPU_SRIOV_CRIT_DATA_SIGNATURE "INDA"
+#define AMDGPU_SRIOV_CRIT_DATA_SIG_LEN   4
+
+#define IS_SRIOV_CRIT_REGN_ENTRY_VALID(hdr, id) ((hdr)->valid_tables & (1 << (id)))
+
 enum amdgpu_sriov_vf_mode {
 	SRIOV_VF_MODE_BARE_METAL = 0,
 	SRIOV_VF_MODE_ONE_VF,
@@ -96,6 +102,9 @@ struct amdgpu_virt_ops {
 					enum amdgpu_ras_block block);
 	bool (*rcvd_ras_intr)(struct amdgpu_device *adev);
 	int (*req_ras_err_count)(struct amdgpu_device *adev);
+	int (*req_ras_cper_dump)(struct amdgpu_device *adev, u64 vf_rptr);
+	int (*req_bad_pages)(struct amdgpu_device *adev);
+	int (*req_ras_chk_criti)(struct amdgpu_device *adev, u64 addr);
 };
 
 /*
@@ -140,15 +149,21 @@ enum AMDGIM_FEATURE_FLAG {
 	AMDGIM_FEATURE_MES_INFO_ENABLE = (1 << 8),
 	AMDGIM_FEATURE_RAS_CAPS = (1 << 9),
 	AMDGIM_FEATURE_RAS_TELEMETRY = (1 << 10),
+	AMDGIM_FEATURE_RAS_CPER = (1 << 11),
+	AMDGIM_FEATURE_XGMI_TA_EXT_PEER_LINK = (1 << 12),
 };
 
 enum AMDGIM_REG_ACCESS_FLAG {
 	/* Use PSP to program IH_RB_CNTL */
-	AMDGIM_FEATURE_IH_REG_PSP_EN     = (1 << 0),
+	AMDGIM_FEATURE_IH_REG_PSP_EN      = (1 << 0),
 	/* Use RLC to program MMHUB regs */
-	AMDGIM_FEATURE_MMHUB_REG_RLC_EN  = (1 << 1),
+	AMDGIM_FEATURE_MMHUB_REG_RLC_EN   = (1 << 1),
 	/* Use RLC to program GC regs */
-	AMDGIM_FEATURE_GC_REG_RLC_EN     = (1 << 2),
+	AMDGIM_FEATURE_GC_REG_RLC_EN      = (1 << 2),
+	/* Use PSP to program L1_TLB_CNTL */
+	AMDGIM_FEATURE_L1_TLB_CNTL_PSP_EN = (1 << 3),
+	/* Use RLCG to program SQ_CONFIG1 */
+	AMDGIM_FEATURE_REG_ACCESS_SQ_CONFIG = (1 << 4),
 };
 
 struct amdgim_pf2vf_info_v1 {
@@ -242,6 +257,23 @@ struct amdgpu_virt_ras_err_handler_data {
 	int last_reserved;
 };
 
+struct amdgpu_virt_ras {
+	struct ratelimit_state ras_error_cnt_rs;
+	struct ratelimit_state ras_cper_dump_rs;
+	struct ratelimit_state ras_chk_criti_rs;
+	struct mutex ras_telemetry_mutex;
+	uint64_t cper_rptr;
+};
+
+#define AMDGPU_VIRT_CAPS_LIST(X) X(AMDGPU_VIRT_CAP_POWER_LIMIT)
+
+DECLARE_ATTR_CAP_CLASS(amdgpu_virt, AMDGPU_VIRT_CAPS_LIST);
+
+struct amdgpu_virt_region {
+	uint32_t offset;
+	uint32_t size_kb;
+};
+
 /* GPU virtualization */
 struct amdgpu_virt {
 	uint32_t			caps;
@@ -251,11 +283,16 @@ struct amdgpu_virt {
 	uint32_t			reg_val_offs;
 	struct amdgpu_irq_src		ack_irq;
 	struct amdgpu_irq_src		rcv_irq;
+
 	struct work_struct		flr_work;
+	struct work_struct		req_bad_pages_work;
+	struct work_struct		handle_bad_pages_work;
+
 	struct amdgpu_mm_table		mm_table;
 	const struct amdgpu_virt_ops	*ops;
 	struct amdgpu_vf_error_buffer	vf_errors;
 	struct amdgpu_virt_fw_reserve	fw_reserve;
+	struct amdgpu_virt_caps virt_caps;
 	uint32_t gim_feature;
 	uint32_t reg_access_mode;
 	int req_init_data_ver;
@@ -264,6 +301,12 @@ struct amdgpu_virt {
 	bool ras_init_done;
 	uint32_t reg_access;
 
+	/* dynamic(v2) critical regions */
+	struct amdgpu_virt_region init_data_header;
+	struct amdgpu_virt_region crit_regn;
+	struct amdgpu_virt_region crit_regn_tbl[AMD_SRIOV_MSG_MAX_TABLE_ID];
+	bool is_dynamic_crit_regn_enabled;
+
 	/* vf2pf message */
 	struct delayed_work vf2pf_work;
 	uint32_t vf2pf_update_interval_ms;
@@ -279,13 +322,18 @@ struct amdgpu_virt {
 	/* the ucode id to signal the autoload */
 	uint32_t autoload_ucode_id;
 
-	struct mutex rlcg_reg_lock;
+	/* Spinlock to protect access to the RLCG register interface */
+	spinlock_t rlcg_reg_lock;
+
+	struct mutex access_req_mutex;
 
 	union amd_sriov_ras_caps ras_en_caps;
 	union amd_sriov_ras_caps ras_telemetry_en_caps;
-
-	struct ratelimit_state ras_telemetry_rs;
+	struct amdgpu_virt_ras ras;
 	struct amd_sriov_ras_telemetry_error_count count_cache;
+
+	/* hibernate and resume with different VF feature for xgmi enabled system */
+	bool is_xgmi_node_migrate_enabled;
 };
 
 struct amdgpu_video_codec_info;
@@ -321,9 +369,17 @@ struct amdgpu_video_codec_info;
 (amdgpu_sriov_vf((adev)) && \
 	((adev)->virt.reg_access & (AMDGIM_FEATURE_GC_REG_RLC_EN)))
 
+#define amdgpu_sriov_reg_indirect_l1_tlb_cntl(adev) \
+(amdgpu_sriov_vf((adev)) && \
+	((adev)->virt.reg_access & (AMDGIM_FEATURE_L1_TLB_CNTL_PSP_EN)))
+
 #define amdgpu_sriov_rlcg_error_report_enabled(adev) \
         (amdgpu_sriov_reg_indirect_mmhub(adev) || amdgpu_sriov_reg_indirect_gc(adev))
 
+#define amdgpu_sriov_reg_access_sq_config(adev) \
+(amdgpu_sriov_vf((adev)) && \
+	((adev)->virt.reg_access & (AMDGIM_FEATURE_REG_ACCESS_SQ_CONFIG)))
+
 #define amdgpu_passthrough(adev) \
 ((adev)->virt.caps & AMDGPU_PASSTHROUGH_MODE)
 
@@ -339,6 +395,12 @@ struct amdgpu_video_codec_info;
 #define amdgpu_sriov_ras_telemetry_block_en(adev, sriov_blk) \
 (amdgpu_sriov_ras_telemetry_en((adev)) && (adev)->virt.ras_telemetry_en_caps.all & BIT(sriov_blk))
 
+#define amdgpu_sriov_ras_cper_en(adev) \
+((adev)->virt.gim_feature & AMDGIM_FEATURE_RAS_CPER)
+
+#define amdgpu_sriov_xgmi_ta_ext_peer_link_en(adev) \
+((adev)->virt.gim_feature & AMDGIM_FEATURE_XGMI_TA_EXT_PEER_LINK)
+
 static inline bool is_virtual_machine(void)
 {
 #if defined(CONFIG_X86)
@@ -352,6 +414,8 @@ static inline bool is_virtual_machine(void)
 
 #define amdgpu_sriov_is_pp_one_vf(adev) \
 	((adev)->virt.gim_feature & AMDGIM_FEATURE_PP_ONE_VF)
+#define amdgpu_sriov_multi_vf_mode(adev) \
+	(amdgpu_sriov_vf(adev) && !amdgpu_sriov_is_pp_one_vf(adev))
 #define amdgpu_sriov_is_debug(adev) \
 	((!amdgpu_in_reset(adev)) && adev->virt.tdr_debug)
 #define amdgpu_sriov_is_normal(adev) \
@@ -362,6 +426,10 @@ static inline bool is_virtual_machine(void)
 	((adev)->virt.gim_feature & AMDGIM_FEATURE_VCN_RB_DECOUPLE)
 #define amdgpu_sriov_is_mes_info_enable(adev) \
 	((adev)->virt.gim_feature & AMDGIM_FEATURE_MES_INFO_ENABLE)
+
+#define amdgpu_virt_xgmi_migrate_enabled(adev) \
+	((adev)->virt.is_xgmi_node_migrate_enabled && (adev)->gmc.xgmi.node_segment_size != 0)
+
 bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev);
 void amdgpu_virt_init_setting(struct amdgpu_device *adev);
 int amdgpu_virt_request_full_gpu(struct amdgpu_device *adev, bool init);
@@ -377,7 +445,11 @@ void amdgpu_virt_release_ras_err_handler_data(struct amdgpu_device *adev);
 void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev);
 void amdgpu_virt_exchange_data(struct amdgpu_device *adev);
 void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev);
-void amdgpu_detect_virtualization(struct amdgpu_device *adev);
+void amdgpu_virt_init(struct amdgpu_device *adev);
+
+int amdgpu_virt_init_critical_region(struct amdgpu_device *adev);
+int amdgpu_virt_get_dynamic_data_info(struct amdgpu_device *adev,
+	int data_id, uint8_t *binary, u32 *size);
 
 bool amdgpu_virt_can_access_debugfs(struct amdgpu_device *adev);
 int amdgpu_virt_enable_access_debugfs(struct amdgpu_device *adev);
@@ -405,5 +477,10 @@ u32 amdgpu_virt_rlcg_reg_rw(struct amdgpu_device *adev, u32 offset, u32 v, u32 f
 bool amdgpu_virt_get_ras_capability(struct amdgpu_device *adev);
 int amdgpu_virt_req_ras_err_count(struct amdgpu_device *adev, enum amdgpu_ras_block block,
 				  struct ras_err_data *err_data);
+int amdgpu_virt_req_ras_cper_dump(struct amdgpu_device *adev, bool force_update);
 int amdgpu_virt_ras_telemetry_post_reset(struct amdgpu_device *adev);
+bool amdgpu_virt_ras_telemetry_block_en(struct amdgpu_device *adev,
+					enum amdgpu_ras_block block);
+void amdgpu_virt_request_bad_pages(struct amdgpu_device *adev);
+int amdgpu_virt_check_vf_critical_region(struct amdgpu_device *adev, u64 addr, bool *hit);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c
index 03308261f894..79bad9cbe2ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c
@@ -14,7 +14,6 @@
 #include "dce_v8_0.h"
 #endif
 #include "dce_v10_0.h"
-#include "dce_v11_0.h"
 #include "ivsrcid/ivsrcid_vislands30.h"
 #include "amdgpu_vkms.h"
 #include "amdgpu_display.h"
@@ -188,8 +187,8 @@ static int amdgpu_vkms_crtc_init(struct drm_device *dev, struct drm_crtc *crtc,
 	amdgpu_crtc->connector = NULL;
 	amdgpu_crtc->vsync_timer_enabled = AMDGPU_IRQ_STATE_DISABLE;
 
-	hrtimer_init(&amdgpu_crtc->vblank_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
-	amdgpu_crtc->vblank_timer.function = &amdgpu_vkms_vblank_simulate;
+	hrtimer_setup(&amdgpu_crtc->vblank_timer, &amdgpu_vkms_vblank_simulate, CLOCK_MONOTONIC,
+		      HRTIMER_MODE_REL);
 
 	return ret;
 }
@@ -581,13 +580,6 @@ static int amdgpu_vkms_hw_init(struct amdgpu_ip_block *ip_block)
 	case CHIP_TONGA:
 		dce_v10_0_disable_dce(adev);
 		break;
-	case CHIP_CARRIZO:
-	case CHIP_STONEY:
-	case CHIP_POLARIS10:
-	case CHIP_POLARIS11:
-	case CHIP_VEGAM:
-		dce_v11_0_disable_dce(adev);
-		break;
 	case CHIP_TOPAZ:
 #ifdef CONFIG_DRM_AMDGPU_SI
 	case CHIP_HAINAN:
@@ -627,7 +619,7 @@ static int amdgpu_vkms_resume(struct amdgpu_ip_block *ip_block)
 	return drm_mode_config_helper_resume(adev_to_drm(ip_block->adev));
 }
 
-static bool amdgpu_vkms_is_idle(void *handle)
+static bool amdgpu_vkms_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 5c07777d3239..a67285118c37 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -128,43 +128,14 @@ struct amdgpu_vm_tlb_seq_struct {
 };
 
 /**
- * amdgpu_vm_set_pasid - manage pasid and vm ptr mapping
- *
- * @adev: amdgpu_device pointer
- * @vm: amdgpu_vm pointer
- * @pasid: the pasid the VM is using on this GPU
- *
- * Set the pasid this VM is using on this GPU, can also be used to remove the
- * pasid by passing in zero.
+ * amdgpu_vm_assert_locked - check if VM is correctly locked
+ * @vm: the VM which schould be tested
  *
+ * Asserts that the VM root PD is locked.
  */
-int amdgpu_vm_set_pasid(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			u32 pasid)
+static void amdgpu_vm_assert_locked(struct amdgpu_vm *vm)
 {
-	int r;
-
-	if (vm->pasid == pasid)
-		return 0;
-
-	if (vm->pasid) {
-		r = xa_err(xa_erase_irq(&adev->vm_manager.pasids, vm->pasid));
-		if (r < 0)
-			return r;
-
-		vm->pasid = 0;
-	}
-
-	if (pasid) {
-		r = xa_err(xa_store_irq(&adev->vm_manager.pasids, pasid, vm,
-					GFP_KERNEL));
-		if (r < 0)
-			return r;
-
-		vm->pasid = pasid;
-	}
-
-
-	return 0;
+	dma_resv_assert_held(vm->root.bo->tbo.base.resv);
 }
 
 /**
@@ -181,6 +152,7 @@ static void amdgpu_vm_bo_evicted(struct amdgpu_vm_bo_base *vm_bo)
 	struct amdgpu_bo *bo = vm_bo->bo;
 
 	vm_bo->moved = true;
+	amdgpu_vm_assert_locked(vm);
 	spin_lock(&vm_bo->vm->status_lock);
 	if (bo->tbo.type == ttm_bo_type_kernel)
 		list_move(&vm_bo->vm_status, &vm->evicted);
@@ -198,6 +170,7 @@ static void amdgpu_vm_bo_evicted(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_moved(struct amdgpu_vm_bo_base *vm_bo)
 {
+	amdgpu_vm_assert_locked(vm_bo->vm);
 	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->moved);
 	spin_unlock(&vm_bo->vm->status_lock);
@@ -213,6 +186,7 @@ static void amdgpu_vm_bo_moved(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_idle(struct amdgpu_vm_bo_base *vm_bo)
 {
+	amdgpu_vm_assert_locked(vm_bo->vm);
 	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->idle);
 	spin_unlock(&vm_bo->vm->status_lock);
@@ -260,6 +234,7 @@ static void amdgpu_vm_bo_evicted_user(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_relocated(struct amdgpu_vm_bo_base *vm_bo)
 {
+	amdgpu_vm_assert_locked(vm_bo->vm);
 	if (vm_bo->bo->parent) {
 		spin_lock(&vm_bo->vm->status_lock);
 		list_move(&vm_bo->vm_status, &vm_bo->vm->relocated);
@@ -279,6 +254,7 @@ static void amdgpu_vm_bo_relocated(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_done(struct amdgpu_vm_bo_base *vm_bo)
 {
+	amdgpu_vm_assert_locked(vm_bo->vm);
 	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->done);
 	spin_unlock(&vm_bo->vm->status_lock);
@@ -295,10 +271,13 @@ static void amdgpu_vm_bo_reset_state_machine(struct amdgpu_vm *vm)
 {
 	struct amdgpu_vm_bo_base *vm_bo, *tmp;
 
+	amdgpu_vm_assert_locked(vm);
+
 	spin_lock(&vm->status_lock);
 	list_splice_init(&vm->done, &vm->invalidated);
 	list_for_each_entry(vm_bo, &vm->invalidated, vm_status)
 		vm_bo->moved = true;
+
 	list_for_each_entry_safe(vm_bo, tmp, &vm->idle, vm_status) {
 		struct amdgpu_bo *bo = vm_bo->bo;
 
@@ -327,6 +306,7 @@ static void amdgpu_vm_update_shared(struct amdgpu_vm_bo_base *base)
 	uint32_t bo_memtype = amdgpu_bo_mem_stats_placement(bo);
 	bool shared;
 
+	dma_resv_assert_held(bo->tbo.base.resv);
 	spin_lock(&vm->status_lock);
 	shared = drm_gem_object_is_shared_for_memory_stats(&bo->tbo.base);
 	if (base->shared != shared) {
@@ -485,6 +465,46 @@ int amdgpu_vm_lock_pd(struct amdgpu_vm *vm, struct drm_exec *exec,
 }
 
 /**
+ * amdgpu_vm_lock_done_list - lock all BOs on the done list
+ * @vm: vm providing the BOs
+ * @exec: drm execution context
+ * @num_fences: number of extra fences to reserve
+ *
+ * Lock the BOs on the done list in the DRM execution context.
+ */
+int amdgpu_vm_lock_done_list(struct amdgpu_vm *vm, struct drm_exec *exec,
+			     unsigned int num_fences)
+{
+	struct list_head *prev = &vm->done;
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo *bo;
+	int ret;
+
+	/* We can only trust prev->next while holding the lock */
+	spin_lock(&vm->status_lock);
+	while (!list_is_head(prev->next, &vm->done)) {
+		bo_va = list_entry(prev->next, typeof(*bo_va), base.vm_status);
+
+		bo = bo_va->base.bo;
+		if (bo) {
+			amdgpu_bo_ref(bo);
+			spin_unlock(&vm->status_lock);
+
+			ret = drm_exec_prepare_obj(exec, &bo->tbo.base, 1);
+			amdgpu_bo_unref(&bo);
+			if (unlikely(ret))
+				return ret;
+
+			spin_lock(&vm->status_lock);
+		}
+		prev = prev->next;
+	}
+	spin_unlock(&vm->status_lock);
+
+	return 0;
+}
+
+/**
  * amdgpu_vm_move_to_lru_tail - move all BOs to the end of LRU
  *
  * @adev: amdgpu device pointer
@@ -616,18 +636,7 @@ int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		spin_unlock(&vm->status_lock);
 
 		bo = bo_base->bo;
-
-		if (dma_resv_locking_ctx(bo->tbo.base.resv) != ticket) {
-			struct amdgpu_task_info *ti = amdgpu_vm_get_task_info_vm(vm);
-
-			pr_warn_ratelimited("Evicted user BO is not reserved\n");
-			if (ti) {
-				pr_warn_ratelimited("pid %d\n", ti->pid);
-				amdgpu_vm_put_task_info(ti);
-			}
-
-			return -EINVAL;
-		}
+		dma_resv_assert_held(bo->tbo.base.resv);
 
 		r = validate(param, bo);
 		if (r)
@@ -654,22 +663,31 @@ int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
  * Check if all VM PDs/PTs are ready for updates
  *
  * Returns:
- * True if VM is not evicting.
+ * True if VM is not evicting and all VM entities are not stopped
  */
 bool amdgpu_vm_ready(struct amdgpu_vm *vm)
 {
-	bool empty;
 	bool ret;
 
+	amdgpu_vm_assert_locked(vm);
+
 	amdgpu_vm_eviction_lock(vm);
 	ret = !vm->evicting;
 	amdgpu_vm_eviction_unlock(vm);
 
 	spin_lock(&vm->status_lock);
-	empty = list_empty(&vm->evicted);
+	ret &= list_empty(&vm->evicted);
 	spin_unlock(&vm->status_lock);
 
-	return ret && empty;
+	spin_lock(&vm->immediate.lock);
+	ret &= !vm->immediate.stopped;
+	spin_unlock(&vm->immediate.lock);
+
+	spin_lock(&vm->delayed.lock);
+	ret &= !vm->delayed.stopped;
+	spin_unlock(&vm->delayed.lock);
+
+	return ret;
 }
 
 /**
@@ -754,6 +772,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 		    bool need_pipe_sync)
 {
 	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_isolation *isolation = &adev->isolation[ring->xcp_id];
 	unsigned vmhub = ring->vm_hub;
 	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
 	struct amdgpu_vmid *id = &id_mgr->ids[job->vmid];
@@ -761,8 +780,9 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 	bool gds_switch_needed = ring->funcs->emit_gds_switch &&
 		job->gds_switch_needed;
 	bool vm_flush_needed = job->vm_needs_flush;
-	struct dma_fence *fence = NULL;
+	bool cleaner_shader_needed = false;
 	bool pasid_mapping_needed = false;
+	struct dma_fence *fence = NULL;
 	unsigned int patch;
 	int r;
 
@@ -785,8 +805,13 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 	pasid_mapping_needed &= adev->gmc.gmc_funcs->emit_pasid_mapping &&
 		ring->funcs->emit_wreg;
 
+	cleaner_shader_needed = job->run_cleaner_shader &&
+		adev->gfx.enable_cleaner_shader &&
+		ring->funcs->emit_cleaner_shader && job->base.s_fence &&
+		&job->base.s_fence->scheduled == isolation->spearhead;
+
 	if (!vm_flush_needed && !gds_switch_needed && !need_pipe_sync &&
-	    !(job->enforce_isolation && !job->vmid))
+	    !cleaner_shader_needed)
 		return 0;
 
 	amdgpu_ring_ib_begin(ring);
@@ -797,9 +822,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 	if (need_pipe_sync)
 		amdgpu_ring_emit_pipeline_sync(ring);
 
-	if (adev->gfx.enable_cleaner_shader &&
-	    ring->funcs->emit_cleaner_shader &&
-	    job->enforce_isolation)
+	if (cleaner_shader_needed)
 		ring->funcs->emit_cleaner_shader(ring);
 
 	if (vm_flush_needed) {
@@ -813,7 +836,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 	if (spm_update_needed && adev->gfx.rlc.funcs->update_spm_vmid)
 		adev->gfx.rlc.funcs->update_spm_vmid(adev, ring, job->vmid);
 
-	if (!ring->is_mes_queue && ring->funcs->emit_gds_switch &&
+	if (ring->funcs->emit_gds_switch &&
 	    gds_switch_needed) {
 		amdgpu_ring_emit_gds_switch(ring, job->vmid, job->gds_base,
 					    job->gds_size, job->gws_base,
@@ -821,10 +844,13 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 					    job->oa_size);
 	}
 
-	if (vm_flush_needed || pasid_mapping_needed) {
-		r = amdgpu_fence_emit(ring, &fence, NULL, 0);
+	if (vm_flush_needed || pasid_mapping_needed || cleaner_shader_needed) {
+		r = amdgpu_fence_emit(ring, job->hw_vm_fence, 0);
 		if (r)
 			return r;
+		fence = &job->hw_vm_fence->base;
+		/* get a ref for the job */
+		dma_fence_get(fence);
 	}
 
 	if (vm_flush_needed) {
@@ -843,6 +869,18 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 		id->pasid_mapping = dma_fence_get(fence);
 		mutex_unlock(&id_mgr->lock);
 	}
+
+	/*
+	 * Make sure that all other submissions wait for the cleaner shader to
+	 * finish before we push them to the HW.
+	 */
+	if (cleaner_shader_needed) {
+		trace_amdgpu_cleaner_shader(ring, fence);
+		mutex_lock(&adev->enforce_isolation_mutex);
+		dma_fence_put(isolation->spearhead);
+		isolation->spearhead = dma_fence_get(fence);
+		mutex_unlock(&adev->enforce_isolation_mutex);
+	}
 	dma_fence_put(fence);
 
 	amdgpu_ring_patch_cond_exec(ring, patch);
@@ -934,6 +972,8 @@ int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
 	LIST_HEAD(relocated);
 	int r, idx;
 
+	amdgpu_vm_assert_locked(vm);
+
 	spin_lock(&vm->status_lock);
 	list_splice_init(&vm->relocated, &relocated);
 	spin_unlock(&vm->status_lock);
@@ -949,7 +989,8 @@ int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
 	params.vm = vm;
 	params.immediate = immediate;
 
-	r = vm->update_funcs->prepare(&params, NULL);
+	r = vm->update_funcs->prepare(&params, NULL,
+				      AMDGPU_KERNEL_JOB_ID_VM_UPDATE_PDES);
 	if (r)
 		goto error;
 
@@ -1028,7 +1069,7 @@ amdgpu_vm_tlb_flush(struct amdgpu_vm_update_params *params,
 	}
 
 	/* Prepare a TLB flush fence to be attached to PTs */
-	if (!params->unlocked && vm->is_compute_context) {
+	if (!params->unlocked) {
 		amdgpu_vm_tlb_fence_create(params->adev, vm, fence);
 
 		/* Makes sure no PD/PT is freed before the flush */
@@ -1118,7 +1159,8 @@ int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		dma_fence_put(tmp);
 	}
 
-	r = vm->update_funcs->prepare(&params, sync);
+	r = vm->update_funcs->prepare(&params, sync,
+				      AMDGPU_KERNEL_JOB_ID_VM_UPDATE_RANGE);
 	if (r)
 		goto error_free;
 
@@ -1254,7 +1296,7 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
 	} else {
 		struct drm_gem_object *obj = &bo->tbo.base;
 
-		if (obj->import_attach && bo_va->is_xgmi) {
+		if (drm_gem_is_imported(obj) && bo_va->is_xgmi) {
 			struct dma_buf *dma_buf = obj->import_attach->dmabuf;
 			struct drm_gem_object *gobj = dma_buf->priv;
 			struct amdgpu_bo *abo = gem_to_amdgpu_bo(gobj);
@@ -1311,13 +1353,14 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
 		/* normally,bo_va->flags only contians READABLE and WIRTEABLE bit go here
 		 * but in case of something, we filter the flags in first place
 		 */
-		if (!(mapping->flags & AMDGPU_PTE_READABLE))
+		if (!(mapping->flags & AMDGPU_VM_PAGE_READABLE))
 			update_flags &= ~AMDGPU_PTE_READABLE;
-		if (!(mapping->flags & AMDGPU_PTE_WRITEABLE))
+		if (!(mapping->flags & AMDGPU_VM_PAGE_WRITEABLE))
 			update_flags &= ~AMDGPU_PTE_WRITEABLE;
 
 		/* Apply ASIC specific mapping flags */
-		amdgpu_gmc_get_vm_pte(adev, mapping, &update_flags);
+		amdgpu_gmc_get_vm_pte(adev, vm, bo, mapping->flags,
+				      &update_flags);
 
 		trace_amdgpu_vm_bo_update(mapping);
 
@@ -1458,7 +1501,7 @@ static void amdgpu_vm_free_mapping(struct amdgpu_device *adev,
 				   struct amdgpu_bo_va_mapping *mapping,
 				   struct dma_fence *fence)
 {
-	if (mapping->flags & AMDGPU_PTE_PRT_FLAG(adev))
+	if (mapping->flags & AMDGPU_VM_PAGE_PRT)
 		amdgpu_vm_add_prt_cb(adev, fence);
 	kfree(mapping);
 }
@@ -1614,7 +1657,7 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 		 * validation
 		 */
 		if (vm->is_compute_context &&
-		    bo_va->base.bo->tbo.base.import_attach &&
+		    drm_gem_is_imported(&bo_va->base.bo->tbo.base) &&
 		    (!bo_va->base.bo->tbo.resource ||
 		     bo_va->base.bo->tbo.resource->mem_type == TTM_PL_SYSTEM))
 			amdgpu_vm_bo_evicted_user(&bo_va->base);
@@ -1737,7 +1780,7 @@ static void amdgpu_vm_bo_insert_map(struct amdgpu_device *adev,
 	list_add(&mapping->list, &bo_va->invalids);
 	amdgpu_vm_it_insert(mapping, &vm->va);
 
-	if (mapping->flags & AMDGPU_PTE_PRT_FLAG(adev))
+	if (mapping->flags & AMDGPU_VM_PAGE_PRT)
 		amdgpu_vm_prt_get(adev);
 
 	if (amdgpu_vm_is_bo_always_valid(vm, bo) && !bo_va->base.moved)
@@ -1797,7 +1840,7 @@ static int amdgpu_vm_verify_parameters(struct amdgpu_device *adev,
 int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 		     struct amdgpu_bo_va *bo_va,
 		     uint64_t saddr, uint64_t offset,
-		     uint64_t size, uint64_t flags)
+		     uint64_t size, uint32_t flags)
 {
 	struct amdgpu_bo_va_mapping *mapping, *tmp;
 	struct amdgpu_bo *bo = bo_va->base.bo;
@@ -1856,7 +1899,7 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
 			     struct amdgpu_bo_va *bo_va,
 			     uint64_t saddr, uint64_t offset,
-			     uint64_t size, uint64_t flags)
+			     uint64_t size, uint32_t flags)
 {
 	struct amdgpu_bo_va_mapping *mapping;
 	struct amdgpu_bo *bo = bo_va->base.bo;
@@ -1912,6 +1955,7 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 	struct amdgpu_bo_va_mapping *mapping;
 	struct amdgpu_vm *vm = bo_va->base.vm;
 	bool valid = true;
+	int r;
 
 	saddr /= AMDGPU_GPU_PAGE_SIZE;
 
@@ -1932,6 +1976,17 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 			return -ENOENT;
 	}
 
+	/* It's unlikely to happen that the mapping userq hasn't been idled
+	 * during user requests GEM unmap IOCTL except for forcing the unmap
+	 * from user space.
+	 */
+	if (unlikely(atomic_read(&bo_va->userq_va_mapped) > 0)) {
+		r = amdgpu_userq_gem_va_unmap_validate(adev, mapping, saddr);
+		if (unlikely(r == -EBUSY))
+			dev_warn_once(adev->dev,
+				      "Attempt to unmap an active userq buffer\n");
+	}
+
 	list_del(&mapping->list);
 	amdgpu_vm_it_remove(mapping, &vm->va);
 	mapping->bo_va = NULL;
@@ -2038,7 +2093,7 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
 		struct amdgpu_bo *bo = before->bo_va->base.bo;
 
 		amdgpu_vm_it_insert(before, &vm->va);
-		if (before->flags & AMDGPU_PTE_PRT_FLAG(adev))
+		if (before->flags & AMDGPU_VM_PAGE_PRT)
 			amdgpu_vm_prt_get(adev);
 
 		if (amdgpu_vm_is_bo_always_valid(vm, bo) &&
@@ -2053,7 +2108,7 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
 		struct amdgpu_bo *bo = after->bo_va->base.bo;
 
 		amdgpu_vm_it_insert(after, &vm->va);
-		if (after->flags & AMDGPU_PTE_PRT_FLAG(adev))
+		if (after->flags & AMDGPU_VM_PAGE_PRT)
 			amdgpu_vm_prt_get(adev);
 
 		if (amdgpu_vm_is_bo_always_valid(vm, bo) &&
@@ -2378,10 +2433,11 @@ void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t min_vm_size,
 	else
 		adev->vm_manager.fragment_size = amdgpu_vm_fragment_size;
 
-	DRM_INFO("vm size is %u GB, %u levels, block size is %u-bit, fragment size is %u-bit\n",
-		 vm_size, adev->vm_manager.num_level + 1,
-		 adev->vm_manager.block_size,
-		 adev->vm_manager.fragment_size);
+	dev_info(
+		adev->dev,
+		"vm size is %u GB, %u levels, block size is %u-bit, fragment size is %u-bit\n",
+		vm_size, adev->vm_manager.num_level + 1,
+		adev->vm_manager.block_size, adev->vm_manager.fragment_size);
 }
 
 /**
@@ -2392,13 +2448,11 @@ void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t min_vm_size,
  */
 long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout)
 {
-	timeout = dma_resv_wait_timeout(vm->root.bo->tbo.base.resv,
-					DMA_RESV_USAGE_BOOKKEEP,
-					true, timeout);
+	timeout = drm_sched_entity_flush(&vm->immediate, timeout);
 	if (timeout <= 0)
 		return timeout;
 
-	return dma_fence_wait_timeout(vm->last_unlocked, true, timeout);
+	return drm_sched_entity_flush(&vm->delayed, timeout);
 }
 
 static void amdgpu_vm_destroy_task_info(struct kref *kref)
@@ -2430,7 +2484,8 @@ amdgpu_vm_get_vm_from_pasid(struct amdgpu_device *adev, u32 pasid)
  */
 void amdgpu_vm_put_task_info(struct amdgpu_task_info *task_info)
 {
-	kref_put(&task_info->refcount, amdgpu_vm_destroy_task_info);
+	if (task_info)
+		kref_put(&task_info->refcount, amdgpu_vm_destroy_task_info);
 }
 
 /**
@@ -2490,11 +2545,11 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
 	if (!vm->task_info)
 		return;
 
-	if (vm->task_info->pid == current->pid)
+	if (vm->task_info->task.pid == current->pid)
 		return;
 
-	vm->task_info->pid = current->pid;
-	get_task_comm(vm->task_info->task_name, current);
+	vm->task_info->task.pid = current->pid;
+	get_task_comm(vm->task_info->task.comm, current);
 
 	if (current->group_leader->mm != current->mm)
 		return;
@@ -2509,6 +2564,7 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
  * @adev: amdgpu_device pointer
  * @vm: requested vm
  * @xcp_id: GPU partition selection id
+ * @pasid: the pasid the VM is using on this GPU
  *
  * Init @vm fields.
  *
@@ -2516,7 +2572,7 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
  * 0 for success, error for failure.
  */
 int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-		   int32_t xcp_id)
+		   int32_t xcp_id, uint32_t pasid)
 {
 	struct amdgpu_bo *root_bo;
 	struct amdgpu_bo_vm *root;
@@ -2534,8 +2590,6 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	spin_lock_init(&vm->status_lock);
 	INIT_LIST_HEAD(&vm->freed);
 	INIT_LIST_HEAD(&vm->done);
-	INIT_LIST_HEAD(&vm->pt_freed);
-	INIT_WORK(&vm->pt_free_work, amdgpu_vm_pt_free_work);
 	INIT_KFIFO(vm->faults);
 
 	r = amdgpu_vm_init_entities(adev, vm);
@@ -2549,8 +2603,8 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
 				    AMDGPU_VM_USE_CPU_FOR_GFX);
 
-	DRM_DEBUG_DRIVER("VM update mode is %s\n",
-			 vm->use_cpu_for_update ? "CPU" : "SDMA");
+	dev_dbg(adev->dev, "VM update mode is %s\n",
+		vm->use_cpu_for_update ? "CPU" : "SDMA");
 	WARN_ONCE((vm->use_cpu_for_update &&
 		   !amdgpu_gmc_vram_full_visible(&adev->gmc)),
 		  "CPU update of VM recommended only for large BAR system\n");
@@ -2592,7 +2646,16 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 
 	r = amdgpu_vm_create_task_info(vm);
 	if (r)
-		DRM_DEBUG("Failed to create task info for VM\n");
+		dev_dbg(adev->dev, "Failed to create task info for VM\n");
+
+	/* Store new PASID in XArray (if non-zero) */
+	if (pasid != 0) {
+		r = xa_err(xa_store_irq(&adev->vm_manager.pasids, pasid, vm, GFP_KERNEL));
+		if (r < 0)
+			goto error_free_root;
+
+		vm->pasid = pasid;
+	}
 
 	amdgpu_bo_unreserve(vm->root.bo);
 	amdgpu_bo_unref(&root_bo);
@@ -2600,6 +2663,11 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	return 0;
 
 error_free_root:
+	/* If PASID was partially set, erase it from XArray before failing */
+	if (vm->pasid != 0) {
+		xa_erase_irq(&adev->vm_manager.pasids, vm->pasid);
+		vm->pasid = 0;
+	}
 	amdgpu_vm_pt_free_root(adev, vm);
 	amdgpu_bo_unreserve(vm->root.bo);
 	amdgpu_bo_unref(&root_bo);
@@ -2643,8 +2711,8 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	/* Update VM state */
 	vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
 				    AMDGPU_VM_USE_CPU_FOR_COMPUTE);
-	DRM_DEBUG_DRIVER("VM update mode is %s\n",
-			 vm->use_cpu_for_update ? "CPU" : "SDMA");
+	dev_dbg(adev->dev, "VM update mode is %s\n",
+		vm->use_cpu_for_update ? "CPU" : "SDMA");
 	WARN_ONCE((vm->use_cpu_for_update &&
 		   !amdgpu_gmc_vram_full_visible(&adev->gmc)),
 		  "CPU update of VM recommended only for large BAR system\n");
@@ -2674,20 +2742,6 @@ unreserve_bo:
 	return r;
 }
 
-/**
- * amdgpu_vm_release_compute - release a compute vm
- * @adev: amdgpu_device pointer
- * @vm: a vm turned into compute vm by calling amdgpu_vm_make_compute
- *
- * This is a correspondant of amdgpu_vm_make_compute. It decouples compute
- * pasid from vm. Compute should stop use of vm after this call.
- */
-void amdgpu_vm_release_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm)
-{
-	amdgpu_vm_set_pasid(adev, vm, 0);
-	vm->is_compute_context = false;
-}
-
 static int amdgpu_vm_stats_is_zero(struct amdgpu_vm *vm)
 {
 	for (int i = 0; i < __AMDGPU_PL_NUM; ++i) {
@@ -2717,11 +2771,13 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 
 	amdgpu_amdkfd_gpuvm_destroy_cb(adev, vm);
 
-	flush_work(&vm->pt_free_work);
-
 	root = amdgpu_bo_ref(vm->root.bo);
 	amdgpu_bo_reserve(root, true);
-	amdgpu_vm_set_pasid(adev, vm, 0);
+	/* Remove PASID mapping before destroying VM */
+	if (vm->pasid != 0) {
+		xa_erase_irq(&adev->vm_manager.pasids, vm->pasid);
+		vm->pasid = 0;
+	}
 	dma_fence_wait(vm->last_unlocked, false);
 	dma_fence_put(vm->last_unlocked);
 	dma_fence_wait(vm->last_tlb_flush, false);
@@ -2731,7 +2787,7 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	dma_fence_put(vm->last_tlb_flush);
 
 	list_for_each_entry_safe(mapping, tmp, &vm->freed, list) {
-		if (mapping->flags & AMDGPU_PTE_PRT_FLAG(adev) && prt_fini_needed) {
+		if (mapping->flags & AMDGPU_VM_PAGE_PRT && prt_fini_needed) {
 			amdgpu_vm_prt_fini(adev, vm);
 			prt_fini_needed = false;
 		}
@@ -2762,10 +2818,7 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	dma_fence_put(vm->last_update);
 
 	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++) {
-		if (vm->reserved_vmid[i]) {
-			amdgpu_vmid_free_reserved(adev, i);
-			vm->reserved_vmid[i] = false;
-		}
+		amdgpu_vmid_free_reserved(adev, vm, i);
 	}
 
 	ttm_lru_bulk_move_fini(&adev->mman.bdev, &vm->lru_bulk_move);
@@ -2775,7 +2828,7 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 
 		dev_warn(adev->dev,
 			 "VM memory stats for proc %s(%d) task %s(%d) is non-zero when fini\n",
-			 ti->process_name, ti->pid, ti->task_name, ti->tgid);
+			 ti->process_name, ti->task.pid, ti->task.comm, ti->tgid);
 	}
 
 	amdgpu_vm_put_task_info(vm->task_info);
@@ -2790,8 +2843,6 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
  */
 void amdgpu_vm_manager_init(struct amdgpu_device *adev)
 {
-	unsigned i;
-
 	/* Concurrent flushes are only possible starting with Vega10 and
 	 * are broken on Navi10 and Navi14.
 	 */
@@ -2800,11 +2851,6 @@ void amdgpu_vm_manager_init(struct amdgpu_device *adev)
 					      adev->asic_type == CHIP_NAVI14);
 	amdgpu_vmid_mgr_init(adev);
 
-	adev->vm_manager.fence_context =
-		dma_fence_context_alloc(AMDGPU_MAX_RINGS);
-	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
-		adev->vm_manager.seqno[i] = 0;
-
 	spin_lock_init(&adev->vm_manager.prt_lock);
 	atomic_set(&adev->vm_manager.num_prt_users, 0);
 
@@ -2861,6 +2907,7 @@ int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 	union drm_amdgpu_vm *args = data;
 	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
 
 	/* No valid flags defined yet */
 	if (args->in.flags)
@@ -2869,17 +2916,9 @@ int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 	switch (args->in.op) {
 	case AMDGPU_VM_OP_RESERVE_VMID:
 		/* We only have requirement to reserve vmid from gfxhub */
-		if (!fpriv->vm.reserved_vmid[AMDGPU_GFXHUB(0)]) {
-			amdgpu_vmid_alloc_reserved(adev, AMDGPU_GFXHUB(0));
-			fpriv->vm.reserved_vmid[AMDGPU_GFXHUB(0)] = true;
-		}
-
-		break;
+		return amdgpu_vmid_alloc_reserved(adev, vm, AMDGPU_GFXHUB(0));
 	case AMDGPU_VM_OP_UNRESERVE_VMID:
-		if (fpriv->vm.reserved_vmid[AMDGPU_GFXHUB(0)]) {
-			amdgpu_vmid_free_reserved(adev, AMDGPU_GFXHUB(0));
-			fpriv->vm.reserved_vmid[AMDGPU_GFXHUB(0)] = false;
-		}
+		amdgpu_vmid_free_reserved(adev, vm, AMDGPU_GFXHUB(0));
 		break;
 	default:
 		return -EINVAL;
@@ -2983,7 +3022,7 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
 error_unlock:
 	amdgpu_bo_unreserve(root);
 	if (r < 0)
-		DRM_ERROR("Can't handle page fault (%d)\n", r);
+		dev_err(adev->dev, "Can't handle page fault (%d)\n", r);
 
 error_unref:
 	amdgpu_bo_unref(&root);
@@ -3017,6 +3056,8 @@ void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m)
 	unsigned int total_done_objs = 0;
 	unsigned int id = 0;
 
+	amdgpu_vm_assert_locked(vm);
+
 	spin_lock(&vm->status_lock);
 	seq_puts(m, "\tIdle BOs:\n");
 	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status) {
@@ -3157,3 +3198,12 @@ bool amdgpu_vm_is_bo_always_valid(struct amdgpu_vm *vm, struct amdgpu_bo *bo)
 {
 	return bo && bo->tbo.base.resv == vm->root.bo->tbo.base.resv;
 }
+
+void amdgpu_vm_print_task_info(struct amdgpu_device *adev,
+			       struct amdgpu_task_info *task_info)
+{
+	dev_err(adev->dev,
+		" Process %s pid %d thread %s pid %d\n",
+		task_info->process_name, task_info->tgid,
+		task_info->task.comm, task_info->task.pid);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index a3e128e373bc..15d757c016cb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -236,9 +236,8 @@ struct amdgpu_vm_pte_funcs {
 };
 
 struct amdgpu_task_info {
+	struct drm_wedge_task_info task;
 	char		process_name[TASK_COMM_LEN];
-	char		task_name[TASK_COMM_LEN];
-	pid_t		pid;
 	pid_t		tgid;
 	struct kref	refcount;
 };
@@ -309,7 +308,7 @@ struct amdgpu_vm_update_params {
 struct amdgpu_vm_update_funcs {
 	int (*map_table)(struct amdgpu_bo_vm *bo);
 	int (*prepare)(struct amdgpu_vm_update_params *p,
-		       struct amdgpu_sync *sync);
+		       struct amdgpu_sync *sync, u64 k_job_id);
 	int (*update)(struct amdgpu_vm_update_params *p,
 		      struct amdgpu_bo_vm *bo, uint64_t pe, uint64_t addr,
 		      unsigned count, uint32_t incr, uint64_t flags);
@@ -350,12 +349,16 @@ struct amdgpu_vm {
 	/* Memory statistics for this vm, protected by status_lock */
 	struct amdgpu_mem_stats stats[__AMDGPU_PL_NUM];
 
+	/*
+	 * The following lists contain amdgpu_vm_bo_base objects for either
+	 * PDs, PTs or per VM BOs. The state transits are:
+	 *
+	 * evicted -> relocated (PDs, PTs) or moved (per VM BOs) -> idle
+	 */
+
 	/* Per-VM and PT BOs who needs a validation */
 	struct list_head	evicted;
 
-	/* BOs for user mode queues that need a validation */
-	struct list_head	evicted_user;
-
 	/* PT BOs which relocated and their parent need an update */
 	struct list_head	relocated;
 
@@ -365,18 +368,28 @@ struct amdgpu_vm {
 	/* All BOs of this VM not currently in the state machine */
 	struct list_head	idle;
 
+	/*
+	 * The following lists contain amdgpu_vm_bo_base objects for BOs which
+	 * have their own dma_resv object and not depend on the root PD. Their
+	 * state transits are:
+	 *
+	 * evicted_user or invalidated -> done
+	 */
+
+	/* BOs for user mode queues that need a validation */
+	struct list_head	evicted_user;
+
 	/* regular invalidated BOs, but not yet updated in the PT */
 	struct list_head	invalidated;
 
-	/* BO mappings freed, but not yet updated in the PT */
-	struct list_head	freed;
-
 	/* BOs which are invalidated, has been updated in the PTs */
 	struct list_head        done;
 
-	/* PT BOs scheduled to free and fill with zero if vm_resv is not hold */
-	struct list_head	pt_freed;
-	struct work_struct	pt_free_work;
+	/*
+	 * This list contains amdgpu_bo_va_mapping objects which have been freed
+	 * but not updated in the PTs
+	 */
+	struct list_head	freed;
 
 	/* contains the page directory */
 	struct amdgpu_vm_bo_base     root;
@@ -399,7 +412,7 @@ struct amdgpu_vm {
 	struct dma_fence	*last_unlocked;
 
 	unsigned int		pasid;
-	bool			reserved_vmid[AMDGPU_MAX_VMHUBS];
+	struct amdgpu_vmid	*reserved_vmid[AMDGPU_MAX_VMHUBS];
 
 	/* Flag to indicate if VM tables are updated by CPU or GPU (SDMA) */
 	bool					use_cpu_for_update;
@@ -440,10 +453,6 @@ struct amdgpu_vm_manager {
 	unsigned int				first_kfd_vmid;
 	bool					concurrent_flush;
 
-	/* Handling of VM fences */
-	u64					fence_context;
-	unsigned				seqno[AMDGPU_MAX_RINGS];
-
 	uint64_t				max_pfn;
 	uint32_t				num_level;
 	uint32_t				block_size;
@@ -487,16 +496,14 @@ extern const struct amdgpu_vm_update_funcs amdgpu_vm_sdma_funcs;
 void amdgpu_vm_manager_init(struct amdgpu_device *adev);
 void amdgpu_vm_manager_fini(struct amdgpu_device *adev);
 
-int amdgpu_vm_set_pasid(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			u32 pasid);
-
 long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout);
-int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, int32_t xcp_id);
+int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, int32_t xcp_id, uint32_t pasid);
 int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm);
-void amdgpu_vm_release_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 int amdgpu_vm_lock_pd(struct amdgpu_vm *vm, struct drm_exec *exec,
 		      unsigned int num_fences);
+int amdgpu_vm_lock_done_list(struct amdgpu_vm *vm, struct drm_exec *exec,
+			     unsigned int num_fences);
 bool amdgpu_vm_ready(struct amdgpu_vm *vm);
 uint64_t amdgpu_vm_generation(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
@@ -544,11 +551,11 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
 int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 		     struct amdgpu_bo_va *bo_va,
 		     uint64_t addr, uint64_t offset,
-		     uint64_t size, uint64_t flags);
+		     uint64_t size, uint32_t flags);
 int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
 			     struct amdgpu_bo_va *bo_va,
 			     uint64_t addr, uint64_t offset,
-			     uint64_t size, uint64_t flags);
+			     uint64_t size, uint32_t flags);
 int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 		       struct amdgpu_bo_va *bo_va,
 		       uint64_t addr);
@@ -673,4 +680,12 @@ void amdgpu_vm_tlb_fence_create(struct amdgpu_device *adev,
 				 struct amdgpu_vm *vm,
 				 struct dma_fence **fence);
 
+void amdgpu_vm_print_task_info(struct amdgpu_device *adev,
+			       struct amdgpu_task_info *task_info);
+
+#define amdgpu_vm_bo_va_for_each_valid_mapping(bo_va, mapping) \
+		list_for_each_entry(mapping, &(bo_va)->valids, list)
+#define amdgpu_vm_bo_va_for_each_invalid_mapping(bo_va, mapping) \
+		list_for_each_entry(mapping, &(bo_va)->invalids, list)
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c
index 0c1ef5850a5e..22e2e5b47341 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c
@@ -40,12 +40,14 @@ static int amdgpu_vm_cpu_map_table(struct amdgpu_bo_vm *table)
  *
  * @p: see amdgpu_vm_update_params definition
  * @sync: sync obj with fences to wait on
+ * @k_job_id: the id for tracing/debug purposes
  *
  * Returns:
  * Negativ errno, 0 for success.
  */
 static int amdgpu_vm_cpu_prepare(struct amdgpu_vm_update_params *p,
-				 struct amdgpu_sync *sync)
+				 struct amdgpu_sync *sync,
+				 u64 k_job_id)
 {
 	if (!sync)
 		return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
index b0bf21682115..f794fb1cc06e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
@@ -26,6 +26,7 @@
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
 #include "amdgpu_vm.h"
+#include "amdgpu_job.h"
 
 /*
  * amdgpu_vm_pt_cursor - state for for_each_amdgpu_vm_pt
@@ -395,7 +396,8 @@ int amdgpu_vm_pt_clear(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	params.vm = vm;
 	params.immediate = immediate;
 
-	r = vm->update_funcs->prepare(&params, NULL);
+	r = vm->update_funcs->prepare(&params, NULL,
+				      AMDGPU_KERNEL_JOB_ID_VM_PT_CLEAR);
 	if (r)
 		goto exit;
 
@@ -547,27 +549,6 @@ static void amdgpu_vm_pt_free(struct amdgpu_vm_bo_base *entry)
 	amdgpu_bo_unref(&entry->bo);
 }
 
-void amdgpu_vm_pt_free_work(struct work_struct *work)
-{
-	struct amdgpu_vm_bo_base *entry, *next;
-	struct amdgpu_vm *vm;
-	LIST_HEAD(pt_freed);
-
-	vm = container_of(work, struct amdgpu_vm, pt_free_work);
-
-	spin_lock(&vm->status_lock);
-	list_splice_init(&vm->pt_freed, &pt_freed);
-	spin_unlock(&vm->status_lock);
-
-	/* flush_work in amdgpu_vm_fini ensure vm->root.bo is valid. */
-	amdgpu_bo_reserve(vm->root.bo, true);
-
-	list_for_each_entry_safe(entry, next, &pt_freed, vm_status)
-		amdgpu_vm_pt_free(entry);
-
-	amdgpu_bo_unreserve(vm->root.bo);
-}
-
 /**
  * amdgpu_vm_pt_free_list - free PD/PT levels
  *
@@ -580,19 +561,15 @@ void amdgpu_vm_pt_free_list(struct amdgpu_device *adev,
 			    struct amdgpu_vm_update_params *params)
 {
 	struct amdgpu_vm_bo_base *entry, *next;
-	struct amdgpu_vm *vm = params->vm;
 	bool unlocked = params->unlocked;
 
 	if (list_empty(&params->tlb_flush_waitlist))
 		return;
 
-	if (unlocked) {
-		spin_lock(&vm->status_lock);
-		list_splice_init(&params->tlb_flush_waitlist, &vm->pt_freed);
-		spin_unlock(&vm->status_lock);
-		schedule_work(&vm->pt_free_work);
-		return;
-	}
+	/*
+	 * unlocked unmap clear page table leaves, warning to free the page entry.
+	 */
+	WARN_ON(unlocked);
 
 	list_for_each_entry_safe(entry, next, &params->tlb_flush_waitlist, vm_status)
 		amdgpu_vm_pt_free(entry);
@@ -900,7 +877,15 @@ int amdgpu_vm_ptes_update(struct amdgpu_vm_update_params *params,
 		incr = (uint64_t)AMDGPU_GPU_PAGE_SIZE << shift;
 		mask = amdgpu_vm_pt_entries_mask(adev, cursor.level);
 		pe_start = ((cursor.pfn >> shift) & mask) * 8;
-		entry_end = ((uint64_t)mask + 1) << shift;
+
+		if (cursor.level < AMDGPU_VM_PTB && params->unlocked)
+			/*
+			 * MMU notifier callback unlocked unmap huge page, leave is PDE entry,
+			 * only clear one entry. Next entry search again for PDE or PTE leave.
+			 */
+			entry_end = 1ULL << shift;
+		else
+			entry_end = ((uint64_t)mask + 1) << shift;
 		entry_end += cursor.pfn & ~(entry_end - 1);
 		entry_end = min(entry_end, end);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
index 46d9fb433ab2..36805dcfa159 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
@@ -40,7 +40,7 @@ static int amdgpu_vm_sdma_map_table(struct amdgpu_bo_vm *table)
 
 /* Allocate a new job for @count PTE updates */
 static int amdgpu_vm_sdma_alloc_job(struct amdgpu_vm_update_params *p,
-				    unsigned int count)
+				    unsigned int count, u64 k_job_id)
 {
 	enum amdgpu_ib_pool_type pool = p->immediate ? AMDGPU_IB_POOL_IMMEDIATE
 		: AMDGPU_IB_POOL_DELAYED;
@@ -56,7 +56,7 @@ static int amdgpu_vm_sdma_alloc_job(struct amdgpu_vm_update_params *p,
 	ndw = min(ndw, AMDGPU_VM_SDMA_MAX_NUM_DW);
 
 	r = amdgpu_job_alloc_with_ib(p->adev, entity, AMDGPU_FENCE_OWNER_VM,
-				     ndw * 4, pool, &p->job);
+				     ndw * 4, pool, &p->job, k_job_id);
 	if (r)
 		return r;
 
@@ -69,16 +69,17 @@ static int amdgpu_vm_sdma_alloc_job(struct amdgpu_vm_update_params *p,
  *
  * @p: see amdgpu_vm_update_params definition
  * @sync: amdgpu_sync object with fences to wait for
+ * @k_job_id: identifier of the job, for tracing purpose
  *
  * Returns:
  * Negativ errno, 0 for success.
  */
 static int amdgpu_vm_sdma_prepare(struct amdgpu_vm_update_params *p,
-				  struct amdgpu_sync *sync)
+				  struct amdgpu_sync *sync, u64 k_job_id)
 {
 	int r;
 
-	r = amdgpu_vm_sdma_alloc_job(p, 0);
+	r = amdgpu_vm_sdma_alloc_job(p, 0, k_job_id);
 	if (r)
 		return r;
 
@@ -249,7 +250,8 @@ static int amdgpu_vm_sdma_update(struct amdgpu_vm_update_params *p,
 			if (r)
 				return r;
 
-			r = amdgpu_vm_sdma_alloc_job(p, count);
+			r = amdgpu_vm_sdma_alloc_job(p, count,
+						     AMDGPU_KERNEL_JOB_ID_VM_UPDATE);
 			if (r)
 				return r;
 		}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_tlb_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_tlb_fence.c
index 51cddfa3f1e8..5d26797356a3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_tlb_fence.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_tlb_fence.c
@@ -71,7 +71,6 @@ static void amdgpu_tlb_fence_work(struct work_struct *work)
 }
 
 static const struct dma_fence_ops amdgpu_tlb_fence_ops = {
-	.use_64bit_seqno = true,
 	.get_driver_name = amdgpu_tlb_fence_get_driver_name,
 	.get_timeline_name = amdgpu_tlb_fence_get_timeline_name
 };
@@ -101,8 +100,8 @@ void amdgpu_vm_tlb_fence_create(struct amdgpu_device *adev, struct amdgpu_vm *vm
 	INIT_WORK(&f->work, amdgpu_tlb_fence_work);
 	spin_lock_init(&f->lock);
 
-	dma_fence_init(&f->base, &amdgpu_tlb_fence_ops, &f->lock,
-		       vm->tlb_fence_context, atomic64_read(&vm->tlb_seq));
+	dma_fence_init64(&f->base, &amdgpu_tlb_fence_ops, &f->lock,
+			 vm->tlb_fence_context, atomic64_read(&vm->tlb_seq));
 
 	/* TODO: We probably need a separate wq here */
 	dma_fence_get(&f->base);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c
index 121ee17b522b..aa78c2ee9e21 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c
@@ -322,6 +322,26 @@ static int vpe_early_init(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
+static bool vpe_need_dpm0_at_power_down(struct amdgpu_device *adev)
+{
+	switch (amdgpu_ip_version(adev, VPE_HWIP, 0)) {
+	case IP_VERSION(6, 1, 1):
+		return adev->pm.fw_version < 0x0a640500;
+	default:
+		return false;
+	}
+}
+
+static int vpe_get_dpm_level(struct amdgpu_device *adev)
+{
+	struct amdgpu_vpe *vpe = &adev->vpe;
+
+	if (!adev->pm.dpm_enabled)
+		return 0;
+
+	return RREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_request_lv));
+}
+
 static void vpe_idle_work_handler(struct work_struct *work)
 {
 	struct amdgpu_device *adev =
@@ -329,11 +349,17 @@ static void vpe_idle_work_handler(struct work_struct *work)
 	unsigned int fences = 0;
 
 	fences += amdgpu_fence_count_emitted(&adev->vpe.ring);
+	if (fences)
+		goto reschedule;
+
+	if (vpe_need_dpm0_at_power_down(adev) && vpe_get_dpm_level(adev) != 0)
+		goto reschedule;
+
+	amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE, AMD_PG_STATE_GATE);
+	return;
 
-	if (fences == 0)
-		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE, AMD_PG_STATE_GATE);
-	else
-		schedule_delayed_work(&adev->vpe.idle_work, VPE_IDLE_TIMEOUT);
+reschedule:
+	schedule_delayed_work(&adev->vpe.idle_work, VPE_IDLE_TIMEOUT);
 }
 
 static int vpe_common_init(struct amdgpu_vpe *vpe)
@@ -379,9 +405,10 @@ static int vpe_sw_init(struct amdgpu_ip_block *ip_block)
 	if (ret)
 		goto out;
 
-	/* TODO: Add queue reset mask when FW fully supports it */
 	adev->vpe.supported_reset =
 		 amdgpu_get_soft_full_reset_mask(&adev->vpe.ring);
+	if (!amdgpu_sriov_vf(adev))
+		adev->vpe.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
 	ret = amdgpu_vpe_sysfs_reset_mask_init(adev);
 	if (ret)
 		goto out;
@@ -435,6 +462,8 @@ static int vpe_hw_fini(struct amdgpu_ip_block *ip_block)
 	struct amdgpu_device *adev = ip_block->adev;
 	struct amdgpu_vpe *vpe = &adev->vpe;
 
+	cancel_delayed_work_sync(&adev->vpe.idle_work);
+
 	vpe_ring_stop(vpe);
 
 	/* Power off VPE */
@@ -445,10 +474,6 @@ static int vpe_hw_fini(struct amdgpu_ip_block *ip_block)
 
 static int vpe_suspend(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = ip_block->adev;
-
-	cancel_delayed_work_sync(&adev->vpe.idle_work);
-
 	return vpe_hw_fini(ip_block);
 }
 
@@ -874,6 +899,27 @@ static void vpe_ring_end_use(struct amdgpu_ring *ring)
 	schedule_delayed_work(&adev->vpe.idle_work, VPE_IDLE_TIMEOUT);
 }
 
+static int vpe_ring_reset(struct amdgpu_ring *ring,
+			  unsigned int vmid,
+			  struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	r = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE,
+						   AMD_PG_STATE_GATE);
+	if (r)
+		return r;
+	r = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE,
+						   AMD_PG_STATE_UNGATE);
+	if (r)
+		return r;
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
 static ssize_t amdgpu_get_vpe_reset_mask(struct device *dev,
 						struct device_attribute *attr,
 						char *buf)
@@ -942,6 +988,7 @@ static const struct amdgpu_ring_funcs vpe_ring_funcs = {
 	.preempt_ib = vpe_ring_preempt_ib,
 	.begin_use = vpe_ring_begin_use,
 	.end_use = vpe_ring_end_use,
+	.reset = vpe_ring_reset,
 };
 
 static void vpe_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
index ff5e52025266..9d934c07fa6b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
@@ -24,11 +24,11 @@
 
 #include <linux/dma-mapping.h>
 #include <drm/ttm/ttm_range_manager.h>
+#include <drm/drm_drv.h>
 
 #include "amdgpu.h"
 #include "amdgpu_vm.h"
 #include "amdgpu_res_cursor.h"
-#include "amdgpu_atomfirmware.h"
 #include "atom.h"
 
 #define AMDGPU_MAX_SG_SEGMENT_SIZE	(2UL << 30)
@@ -234,6 +234,9 @@ static umode_t amdgpu_vram_attrs_is_visible(struct kobject *kobj,
 	    !adev->gmc.vram_vendor)
 		return 0;
 
+	if (!ttm_resource_manager_used(&adev->mman.vram_mgr.manager))
+		return 0;
+
 	return attr->mode;
 }
 
@@ -396,43 +399,33 @@ out:
 	return ret;
 }
 
-static void amdgpu_dummy_vram_mgr_debug(struct ttm_resource_manager *man,
-				  struct drm_printer *printer)
+int amdgpu_vram_mgr_query_address_block_info(struct amdgpu_vram_mgr *mgr,
+			uint64_t address, struct amdgpu_vram_block_info *info)
 {
-	DRM_DEBUG_DRIVER("Dummy vram mgr debug\n");
-}
-
-static bool amdgpu_dummy_vram_mgr_compatible(struct ttm_resource_manager *man,
-				       struct ttm_resource *res,
-				       const struct ttm_place *place,
-				       size_t size)
-{
-	DRM_DEBUG_DRIVER("Dummy vram mgr compatible\n");
-	return false;
-}
+	struct amdgpu_vram_mgr_resource *vres;
+	struct drm_buddy_block *block;
+	u64 start, size;
+	int ret = -ENOENT;
 
-static bool amdgpu_dummy_vram_mgr_intersects(struct ttm_resource_manager *man,
-				       struct ttm_resource *res,
-				       const struct ttm_place *place,
-				       size_t size)
-{
-	DRM_DEBUG_DRIVER("Dummy vram mgr intersects\n");
-	return true;
-}
+	mutex_lock(&mgr->lock);
+	list_for_each_entry(vres, &mgr->allocated_vres_list, vres_node) {
+		list_for_each_entry(block, &vres->blocks, link) {
+			start = amdgpu_vram_mgr_block_start(block);
+			size = amdgpu_vram_mgr_block_size(block);
+			if ((start <= address) && (address < (start + size))) {
+				info->start = start;
+				info->size = size;
+				memcpy(&info->task, &vres->task, sizeof(vres->task));
+				ret = 0;
+				goto out;
+			}
+		}
+	}
 
-static void amdgpu_dummy_vram_mgr_del(struct ttm_resource_manager *man,
-				struct ttm_resource *res)
-{
-	DRM_DEBUG_DRIVER("Dummy vram mgr deleted\n");
-}
+out:
+	mutex_unlock(&mgr->lock);
 
-static int amdgpu_dummy_vram_mgr_new(struct ttm_resource_manager *man,
-			       struct ttm_buffer_object *tbo,
-			       const struct ttm_place *place,
-			       struct ttm_resource **res)
-{
-	DRM_DEBUG_DRIVER("Dummy vram mgr new\n");
-	return -ENOSPC;
+	return ret;
 }
 
 /**
@@ -463,7 +456,7 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man,
 	int r;
 
 	lpfn = (u64)place->lpfn << PAGE_SHIFT;
-	if (!lpfn)
+	if (!lpfn || lpfn > man->size)
 		lpfn = man->size;
 
 	fpfn = (u64)place->fpfn << PAGE_SHIFT;
@@ -568,6 +561,10 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man,
 			remaining_size -= size;
 	}
 
+	vres->task.pid = task_pid_nr(current);
+	get_task_comm(vres->task.comm, current);
+	list_add_tail(&vres->vres_node, &mgr->allocated_vres_list);
+
 	if (bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS && adjust_dcc_size) {
 		struct drm_buddy_block *dcc_block;
 		unsigned long dcc_start;
@@ -645,12 +642,15 @@ static void amdgpu_vram_mgr_del(struct ttm_resource_manager *man,
 	uint64_t vis_usage = 0;
 
 	mutex_lock(&mgr->lock);
+
+	list_del(&vres->vres_node);
+	memset(&vres->task, 0, sizeof(vres->task));
+
 	list_for_each_entry(block, &vres->blocks, link)
 		vis_usage += amdgpu_vram_mgr_vis_size(adev, block);
 
-	amdgpu_vram_mgr_do_reserve(man);
-
 	drm_buddy_free_list(mm, &vres->blocks, vres->flags);
+	amdgpu_vram_mgr_do_reserve(man);
 	mutex_unlock(&mgr->lock);
 
 	atomic64_sub(vis_usage, &mgr->vis_usage);
@@ -783,6 +783,23 @@ uint64_t amdgpu_vram_mgr_vis_usage(struct amdgpu_vram_mgr *mgr)
 }
 
 /**
+ * amdgpu_vram_mgr_clear_reset_blocks - reset clear blocks
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Reset the cleared drm buddy blocks.
+ */
+void amdgpu_vram_mgr_clear_reset_blocks(struct amdgpu_device *adev)
+{
+	struct amdgpu_vram_mgr *mgr = &adev->mman.vram_mgr;
+	struct drm_buddy *mm = &mgr->mm;
+
+	mutex_lock(&mgr->lock);
+	drm_buddy_reset_clear(mm, false);
+	mutex_unlock(&mgr->lock);
+}
+
+/**
  * amdgpu_vram_mgr_intersects - test each drm buddy block for intersection
  *
  * @man: TTM memory type manager
@@ -879,14 +896,6 @@ static void amdgpu_vram_mgr_debug(struct ttm_resource_manager *man,
 	mutex_unlock(&mgr->lock);
 }
 
-static const struct ttm_resource_manager_func amdgpu_dummy_vram_mgr_func = {
-	.alloc	= amdgpu_dummy_vram_mgr_new,
-	.free	= amdgpu_dummy_vram_mgr_del,
-	.intersects = amdgpu_dummy_vram_mgr_intersects,
-	.compatible = amdgpu_dummy_vram_mgr_compatible,
-	.debug	= amdgpu_dummy_vram_mgr_debug
-};
-
 static const struct ttm_resource_manager_func amdgpu_vram_mgr_func = {
 	.alloc	= amdgpu_vram_mgr_new,
 	.free	= amdgpu_vram_mgr_del,
@@ -908,24 +917,22 @@ int amdgpu_vram_mgr_init(struct amdgpu_device *adev)
 	struct ttm_resource_manager *man = &mgr->manager;
 	int err;
 
+	man->cg = drmm_cgroup_register_region(adev_to_drm(adev), "vram", adev->gmc.real_vram_size);
+	if (IS_ERR(man->cg))
+		return PTR_ERR(man->cg);
 	ttm_resource_manager_init(man, &adev->mman.bdev,
 				  adev->gmc.real_vram_size);
 
 	mutex_init(&mgr->lock);
 	INIT_LIST_HEAD(&mgr->reservations_pending);
 	INIT_LIST_HEAD(&mgr->reserved_pages);
+	INIT_LIST_HEAD(&mgr->allocated_vres_list);
 	mgr->default_page_size = PAGE_SIZE;
 
-	if (!adev->gmc.is_app_apu) {
-		man->func = &amdgpu_vram_mgr_func;
-
-		err = drm_buddy_init(&mgr->mm, man->size, PAGE_SIZE);
-		if (err)
-			return err;
-	} else {
-		man->func = &amdgpu_dummy_vram_mgr_func;
-		DRM_INFO("Setup dummy vram mgr\n");
-	}
+	man->func = &amdgpu_vram_mgr_func;
+	err = drm_buddy_init(&mgr->mm, man->size, PAGE_SIZE);
+	if (err)
+		return err;
 
 	ttm_set_driver_manager(&adev->mman.bdev, TTM_PL_VRAM, &mgr->manager);
 	ttm_resource_manager_set_used(man, true);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h
index b256cbc2bc27..5f5fd9a911c2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.h
@@ -35,12 +35,26 @@ struct amdgpu_vram_mgr {
 	struct list_head reserved_pages;
 	atomic64_t vis_usage;
 	u64 default_page_size;
+	struct list_head allocated_vres_list;
+};
+
+struct amdgpu_vres_task {
+	pid_t pid;
+	char  comm[TASK_COMM_LEN];
+};
+
+struct amdgpu_vram_block_info {
+	u64 start;
+	u64 size;
+	struct amdgpu_vres_task task;
 };
 
 struct amdgpu_vram_mgr_resource {
 	struct ttm_resource base;
 	struct list_head blocks;
 	unsigned long flags;
+	struct list_head vres_node;
+	struct amdgpu_vres_task task;
 };
 
 static inline u64 amdgpu_vram_mgr_block_start(struct drm_buddy_block *block)
@@ -66,7 +80,13 @@ to_amdgpu_vram_mgr_resource(struct ttm_resource *res)
 
 static inline void amdgpu_vram_mgr_set_cleared(struct ttm_resource *res)
 {
-	to_amdgpu_vram_mgr_resource(res)->flags |= DRM_BUDDY_CLEARED;
+	struct amdgpu_vram_mgr_resource *ares = to_amdgpu_vram_mgr_resource(res);
+
+	WARN_ON(ares->flags & DRM_BUDDY_CLEARED);
+	ares->flags |= DRM_BUDDY_CLEARED;
 }
 
+int amdgpu_vram_mgr_query_address_block_info(struct amdgpu_vram_mgr *mgr,
+		uint64_t address, struct amdgpu_vram_block_info *info);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
index 23b6f7a4aa4a..1083db8cea2e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
@@ -27,6 +27,9 @@
 #include <drm/drm_drv.h>
 #include "../amdxcp/amdgpu_xcp_drv.h"
 
+static void amdgpu_xcp_sysfs_entries_init(struct amdgpu_xcp_mgr *xcp_mgr);
+static void amdgpu_xcp_sysfs_entries_update(struct amdgpu_xcp_mgr *xcp_mgr);
+
 static int __amdgpu_xcp_run(struct amdgpu_xcp_mgr *xcp_mgr,
 			    struct amdgpu_xcp_ip *xcp_ip, int xcp_state)
 {
@@ -117,6 +120,25 @@ static void __amdgpu_xcp_add_block(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id,
 	xcp->valid = true;
 }
 
+static void __amdgpu_xcp_set_unique_id(struct amdgpu_xcp_mgr *xcp_mgr,
+				       int xcp_id)
+{
+	struct amdgpu_xcp *xcp = &xcp_mgr->xcp[xcp_id];
+	struct amdgpu_device *adev = xcp_mgr->adev;
+	uint32_t inst_mask;
+	uint64_t uid;
+	int i;
+
+	if (!amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_GFX, &inst_mask) &&
+	    inst_mask) {
+		i = GET_INST(GC, (ffs(inst_mask) - 1));
+		uid = amdgpu_device_get_uid(xcp_mgr->adev->uid_info,
+					    AMDGPU_UID_TYPE_XCD, i);
+		if (uid)
+			xcp->unique_id = uid;
+	}
+}
+
 int amdgpu_xcp_init(struct amdgpu_xcp_mgr *xcp_mgr, int num_xcps, int mode)
 {
 	struct amdgpu_device *adev = xcp_mgr->adev;
@@ -155,6 +177,7 @@ int amdgpu_xcp_init(struct amdgpu_xcp_mgr *xcp_mgr, int num_xcps, int mode)
 			else
 				xcp_mgr->xcp[i].mem_id = mem_id;
 		}
+		__amdgpu_xcp_set_unique_id(xcp_mgr, i);
 	}
 
 	xcp_mgr->num_xcps = num_xcps;
@@ -189,7 +212,7 @@ static int __amdgpu_xcp_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr,
 
 		goto out;
 	}
-
+	amdgpu_xcp_sysfs_entries_update(xcp_mgr);
 out:
 	mutex_unlock(&xcp_mgr->xcp_lock);
 
@@ -215,15 +238,27 @@ int amdgpu_xcp_restore_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr)
 	return __amdgpu_xcp_switch_partition_mode(xcp_mgr, xcp_mgr->mode);
 }
 
-int amdgpu_xcp_query_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags)
+static bool __amdgpu_xcp_is_cached_mode_valid(struct amdgpu_xcp_mgr *xcp_mgr)
 {
-	int mode;
+	if (!xcp_mgr->funcs || !xcp_mgr->funcs->query_partition_mode)
+		return true;
 
 	if (!amdgpu_sriov_vf(xcp_mgr->adev) &&
 	    xcp_mgr->mode == AMDGPU_XCP_MODE_NONE)
-		return xcp_mgr->mode;
+		return true;
 
-	if (!xcp_mgr->funcs || !xcp_mgr->funcs->query_partition_mode)
+	if (xcp_mgr->mode != AMDGPU_XCP_MODE_NONE &&
+	    xcp_mgr->mode != AMDGPU_XCP_MODE_TRANS)
+		return true;
+
+	return false;
+}
+
+int amdgpu_xcp_query_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags)
+{
+	int mode;
+
+	if (__amdgpu_xcp_is_cached_mode_valid(xcp_mgr))
 		return xcp_mgr->mode;
 
 	if (!(flags & AMDGPU_XCP_FL_LOCKED))
@@ -263,9 +298,10 @@ static int amdgpu_xcp_dev_alloc(struct amdgpu_device *adev)
 		if (ret == -ENOSPC) {
 			dev_warn(adev->dev,
 			"Skip xcp node #%d when out of drm node resource.", i);
-			return 0;
+			ret = 0;
+			goto out;
 		} else if (ret) {
-			return ret;
+			goto out;
 		}
 
 		/* Redirect all IOCTLs to the primary device */
@@ -278,9 +314,14 @@ static int amdgpu_xcp_dev_alloc(struct amdgpu_device *adev)
 		p_ddev->vma_offset_manager = ddev->vma_offset_manager;
 		p_ddev->driver = &amdgpu_partition_driver;
 		adev->xcp_mgr->xcp[i].ddev = p_ddev;
+
+		dev_set_drvdata(p_ddev->dev, &adev->xcp_mgr->xcp[i]);
 	}
+	ret = 0;
+out:
+	amdgpu_xcp_sysfs_entries_init(adev->xcp_mgr);
 
-	return 0;
+	return ret;
 }
 
 int amdgpu_xcp_mgr_init(struct amdgpu_device *adev, int init_mode,
@@ -288,6 +329,7 @@ int amdgpu_xcp_mgr_init(struct amdgpu_device *adev, int init_mode,
 			struct amdgpu_xcp_mgr_funcs *xcp_funcs)
 {
 	struct amdgpu_xcp_mgr *xcp_mgr;
+	int i;
 
 	if (!xcp_funcs || !xcp_funcs->get_ip_details)
 		return -EINVAL;
@@ -306,6 +348,8 @@ int amdgpu_xcp_mgr_init(struct amdgpu_device *adev, int init_mode,
 		amdgpu_xcp_init(xcp_mgr, init_num_xcps, init_mode);
 
 	adev->xcp_mgr = xcp_mgr;
+	for (i = 0; i < MAX_XCP; ++i)
+		xcp_mgr->xcp[i].xcp_mgr = xcp_mgr;
 
 	return amdgpu_xcp_dev_alloc(adev);
 }
@@ -382,6 +426,7 @@ void amdgpu_xcp_dev_unplug(struct amdgpu_device *adev)
 		p_ddev->primary->dev = adev->xcp_mgr->xcp[i].pdev;
 		p_ddev->driver =  adev->xcp_mgr->xcp[i].driver;
 		p_ddev->vma_offset_manager = adev->xcp_mgr->xcp[i].vma_offset_manager;
+		amdgpu_xcp_drm_dev_free(p_ddev);
 	}
 }
 
@@ -433,6 +478,223 @@ void amdgpu_xcp_release_sched(struct amdgpu_device *adev,
 	}
 }
 
+int amdgpu_xcp_select_scheds(struct amdgpu_device *adev,
+			     u32 hw_ip, u32 hw_prio,
+			     struct amdgpu_fpriv *fpriv,
+			     unsigned int *num_scheds,
+			     struct drm_gpu_scheduler ***scheds)
+{
+	u32 sel_xcp_id;
+	int i;
+	struct amdgpu_xcp_mgr *xcp_mgr = adev->xcp_mgr;
+
+	if (fpriv->xcp_id == AMDGPU_XCP_NO_PARTITION) {
+		u32 least_ref_cnt = ~0;
+
+		fpriv->xcp_id = 0;
+		for (i = 0; i < xcp_mgr->num_xcps; i++) {
+			u32 total_ref_cnt;
+
+			total_ref_cnt = atomic_read(&xcp_mgr->xcp[i].ref_cnt);
+			if (total_ref_cnt < least_ref_cnt) {
+				fpriv->xcp_id = i;
+				least_ref_cnt = total_ref_cnt;
+			}
+		}
+	}
+	sel_xcp_id = fpriv->xcp_id;
+
+	if (xcp_mgr->xcp[sel_xcp_id].gpu_sched[hw_ip][hw_prio].num_scheds) {
+		*num_scheds =
+			xcp_mgr->xcp[fpriv->xcp_id].gpu_sched[hw_ip][hw_prio].num_scheds;
+		*scheds =
+			xcp_mgr->xcp[fpriv->xcp_id].gpu_sched[hw_ip][hw_prio].sched;
+		atomic_inc(&adev->xcp_mgr->xcp[sel_xcp_id].ref_cnt);
+		dev_dbg(adev->dev, "Selected partition #%d", sel_xcp_id);
+	} else {
+		dev_err(adev->dev, "Failed to schedule partition #%d.", sel_xcp_id);
+		return -ENOENT;
+	}
+
+	return 0;
+}
+
+static void amdgpu_set_xcp_id(struct amdgpu_device *adev,
+			      uint32_t inst_idx,
+			      struct amdgpu_ring *ring)
+{
+	int xcp_id;
+	enum AMDGPU_XCP_IP_BLOCK ip_blk;
+	uint32_t inst_mask;
+
+	ring->xcp_id = AMDGPU_XCP_NO_PARTITION;
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)
+		adev->gfx.enforce_isolation[0].xcp_id = ring->xcp_id;
+	if ((adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE) ||
+	    (ring->funcs->type == AMDGPU_RING_TYPE_CPER))
+		return;
+
+	inst_mask = 1 << inst_idx;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_HW_IP_GFX:
+	case AMDGPU_RING_TYPE_COMPUTE:
+	case AMDGPU_RING_TYPE_KIQ:
+		ip_blk = AMDGPU_XCP_GFX;
+		break;
+	case AMDGPU_RING_TYPE_SDMA:
+		ip_blk = AMDGPU_XCP_SDMA;
+		break;
+	case AMDGPU_RING_TYPE_VCN_ENC:
+	case AMDGPU_RING_TYPE_VCN_JPEG:
+		ip_blk = AMDGPU_XCP_VCN;
+		break;
+	default:
+		dev_err(adev->dev, "Not support ring type %d!", ring->funcs->type);
+		return;
+	}
+
+	for (xcp_id = 0; xcp_id < adev->xcp_mgr->num_xcps; xcp_id++) {
+		if (adev->xcp_mgr->xcp[xcp_id].ip[ip_blk].inst_mask & inst_mask) {
+			ring->xcp_id = xcp_id;
+			dev_dbg(adev->dev, "ring:%s xcp_id :%u", ring->name,
+				ring->xcp_id);
+			if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)
+				adev->gfx.enforce_isolation[xcp_id].xcp_id = xcp_id;
+			break;
+		}
+	}
+}
+
+static void amdgpu_xcp_gpu_sched_update(struct amdgpu_device *adev,
+					struct amdgpu_ring *ring,
+					unsigned int sel_xcp_id)
+{
+	unsigned int *num_gpu_sched;
+
+	num_gpu_sched = &adev->xcp_mgr->xcp[sel_xcp_id]
+			.gpu_sched[ring->funcs->type][ring->hw_prio].num_scheds;
+	adev->xcp_mgr->xcp[sel_xcp_id].gpu_sched[ring->funcs->type][ring->hw_prio]
+			.sched[(*num_gpu_sched)++] = &ring->sched;
+	dev_dbg(adev->dev, "%s :[%d] gpu_sched[%d][%d] = %d",
+		ring->name, sel_xcp_id, ring->funcs->type,
+		ring->hw_prio, *num_gpu_sched);
+}
+
+static int amdgpu_xcp_sched_list_update(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	int i;
+
+	for (i = 0; i < MAX_XCP; i++) {
+		atomic_set(&adev->xcp_mgr->xcp[i].ref_cnt, 0);
+		memset(adev->xcp_mgr->xcp[i].gpu_sched, 0, sizeof(adev->xcp_mgr->xcp->gpu_sched));
+	}
+
+	if (adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE)
+		return 0;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		ring = adev->rings[i];
+		if (!ring || !ring->sched.ready || ring->no_scheduler)
+			continue;
+
+		amdgpu_xcp_gpu_sched_update(adev, ring, ring->xcp_id);
+
+		/* VCN may be shared by two partitions under CPX MODE in certain
+		 * configs.
+		 */
+		if ((ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC ||
+		     ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG) &&
+		    (adev->xcp_mgr->num_xcps > adev->vcn.num_vcn_inst))
+			amdgpu_xcp_gpu_sched_update(adev, ring, ring->xcp_id + 1);
+	}
+
+	return 0;
+}
+
+int amdgpu_xcp_update_partition_sched_list(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->num_rings; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE ||
+			ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+			amdgpu_set_xcp_id(adev, ring->xcc_id, ring);
+		else
+			amdgpu_set_xcp_id(adev, ring->me, ring);
+	}
+
+	return amdgpu_xcp_sched_list_update(adev);
+}
+
+void amdgpu_xcp_update_supported_modes(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	struct amdgpu_device *adev = xcp_mgr->adev;
+
+	xcp_mgr->supp_xcp_modes = 0;
+
+	switch (NUM_XCC(adev->gfx.xcc_mask)) {
+	case 8:
+		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
+					  BIT(AMDGPU_DPX_PARTITION_MODE) |
+					  BIT(AMDGPU_QPX_PARTITION_MODE) |
+					  BIT(AMDGPU_CPX_PARTITION_MODE);
+		break;
+	case 6:
+		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
+					  BIT(AMDGPU_TPX_PARTITION_MODE) |
+					  BIT(AMDGPU_CPX_PARTITION_MODE);
+		break;
+	case 4:
+		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
+					  BIT(AMDGPU_DPX_PARTITION_MODE) |
+					  BIT(AMDGPU_CPX_PARTITION_MODE);
+		break;
+	case 2:
+		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
+					  BIT(AMDGPU_CPX_PARTITION_MODE);
+		break;
+	case 1:
+		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
+					  BIT(AMDGPU_CPX_PARTITION_MODE);
+		break;
+
+	default:
+		break;
+	}
+}
+
+int amdgpu_xcp_pre_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags)
+{
+	/* TODO:
+	 * Stop user queues and threads, and make sure GPU is empty of work.
+	 */
+
+	if (flags & AMDGPU_XCP_OPS_KFD)
+		amdgpu_amdkfd_device_fini_sw(xcp_mgr->adev);
+
+	return 0;
+}
+
+int amdgpu_xcp_post_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags)
+{
+	int ret = 0;
+
+	if (flags & AMDGPU_XCP_OPS_KFD) {
+		amdgpu_amdkfd_device_probe(xcp_mgr->adev);
+		amdgpu_amdkfd_device_init(xcp_mgr->adev);
+		/* If KFD init failed, return failure */
+		if (!xcp_mgr->adev->kfd.init_complete)
+			ret = -EIO;
+	}
+
+	return ret;
+}
+
+/*====================== xcp sysfs - configuration ======================*/
 #define XCP_CFG_SYSFS_RES_ATTR_SHOW(_name)                         \
 	static ssize_t amdgpu_xcp_res_sysfs_##_name##_show(        \
 		struct amdgpu_xcp_res_details *xcp_res, char *buf) \
@@ -635,7 +897,7 @@ static const struct attribute *xcp_attrs[] = {
 	NULL,
 };
 
-void amdgpu_xcp_cfg_sysfs_init(struct amdgpu_device *adev)
+static void amdgpu_xcp_cfg_sysfs_init(struct amdgpu_device *adev)
 {
 	struct amdgpu_xcp_res_details *xcp_res;
 	struct amdgpu_xcp_cfg *xcp_cfg;
@@ -703,16 +965,16 @@ err1:
 	kobject_put(&xcp_cfg->kobj);
 }
 
-void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev)
+static void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev)
 {
 	struct amdgpu_xcp_res_details *xcp_res;
 	struct amdgpu_xcp_cfg *xcp_cfg;
 	int i;
 
-	if (!adev->xcp_mgr)
+	if (!adev->xcp_mgr || !adev->xcp_mgr->xcp_cfg)
 		return;
 
-	xcp_cfg =  adev->xcp_mgr->xcp_cfg;
+	xcp_cfg = adev->xcp_mgr->xcp_cfg;
 	for (i = 0; i < xcp_cfg->num_res; i++) {
 		xcp_res = &xcp_cfg->xcp_res[i];
 		kobject_put(&xcp_res->kobj);
@@ -722,3 +984,124 @@ void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev)
 	sysfs_remove_files(&xcp_cfg->kobj, xcp_attrs);
 	kobject_put(&xcp_cfg->kobj);
 }
+
+/*====================== xcp sysfs - data entries ======================*/
+
+#define to_xcp(x) container_of(x, struct amdgpu_xcp, kobj)
+
+static ssize_t xcp_metrics_show(struct kobject *kobj,
+				struct kobj_attribute *attr, char *buf)
+{
+	struct amdgpu_xcp *xcp = to_xcp(kobj);
+	struct amdgpu_xcp_mgr *xcp_mgr;
+	ssize_t size;
+
+	xcp_mgr = xcp->xcp_mgr;
+	size = amdgpu_dpm_get_xcp_metrics(xcp_mgr->adev, xcp->id, NULL);
+	if (size <= 0)
+		return size;
+
+	if (size > PAGE_SIZE)
+		return -ENOSPC;
+
+	return amdgpu_dpm_get_xcp_metrics(xcp_mgr->adev, xcp->id, buf);
+}
+
+static umode_t amdgpu_xcp_attrs_is_visible(struct kobject *kobj,
+					   struct attribute *attr, int n)
+{
+	struct amdgpu_xcp *xcp = to_xcp(kobj);
+
+	if (!xcp || !xcp->valid)
+		return 0;
+
+	return attr->mode;
+}
+
+static struct kobj_attribute xcp_sysfs_metrics = __ATTR_RO(xcp_metrics);
+
+static struct attribute *amdgpu_xcp_attrs[] = {
+	&xcp_sysfs_metrics.attr,
+	NULL,
+};
+
+static const struct attribute_group amdgpu_xcp_attrs_group = {
+	.attrs = amdgpu_xcp_attrs,
+	.is_visible = amdgpu_xcp_attrs_is_visible
+};
+
+static const struct kobj_type xcp_sysfs_ktype = {
+	.sysfs_ops = &kobj_sysfs_ops,
+};
+
+static void amdgpu_xcp_sysfs_entries_fini(struct amdgpu_xcp_mgr *xcp_mgr, int n)
+{
+	struct amdgpu_xcp *xcp;
+
+	for (n--; n >= 0; n--) {
+		xcp = &xcp_mgr->xcp[n];
+		if (!xcp->ddev || !xcp->valid)
+			continue;
+		sysfs_remove_group(&xcp->kobj, &amdgpu_xcp_attrs_group);
+		kobject_put(&xcp->kobj);
+	}
+}
+
+static void amdgpu_xcp_sysfs_entries_init(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	struct amdgpu_xcp *xcp;
+	int i, r;
+
+	for (i = 0; i < MAX_XCP; i++) {
+		/* Redirect all IOCTLs to the primary device */
+		xcp = &xcp_mgr->xcp[i];
+		if (!xcp->ddev)
+			break;
+		r = kobject_init_and_add(&xcp->kobj, &xcp_sysfs_ktype,
+					 &xcp->ddev->dev->kobj, "xcp");
+		if (r)
+			goto out;
+
+		r = sysfs_create_group(&xcp->kobj, &amdgpu_xcp_attrs_group);
+		if (r)
+			goto out;
+	}
+
+	return;
+out:
+	kobject_put(&xcp->kobj);
+}
+
+static void amdgpu_xcp_sysfs_entries_update(struct amdgpu_xcp_mgr *xcp_mgr)
+{
+	struct amdgpu_xcp *xcp;
+	int i;
+
+	for (i = 0; i < MAX_XCP; i++) {
+		/* Redirect all IOCTLs to the primary device */
+		xcp = &xcp_mgr->xcp[i];
+		if (!xcp->ddev)
+			continue;
+		sysfs_update_group(&xcp->kobj, &amdgpu_xcp_attrs_group);
+	}
+
+	return;
+}
+
+void amdgpu_xcp_sysfs_init(struct amdgpu_device *adev)
+{
+	if (!adev->xcp_mgr)
+		return;
+
+	amdgpu_xcp_cfg_sysfs_init(adev);
+
+	return;
+}
+
+void amdgpu_xcp_sysfs_fini(struct amdgpu_device *adev)
+{
+	if (!adev->xcp_mgr)
+		return;
+	amdgpu_xcp_sysfs_entries_fini(adev->xcp_mgr, MAX_XCP);
+	amdgpu_xcp_cfg_sysfs_fini(adev);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h
index b63f53242c57..1928d9e224fc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h
@@ -39,6 +39,8 @@
 
 #define AMDGPU_XCP_NO_PARTITION (~0)
 
+#define AMDGPU_XCP_OPS_KFD	(1 << 0)
+
 struct amdgpu_fpriv;
 
 enum AMDGPU_XCP_IP_BLOCK {
@@ -108,6 +110,9 @@ struct amdgpu_xcp {
 	struct drm_driver *driver;
 	struct drm_vma_offset_manager *vma_offset_manager;
 	struct amdgpu_sched	gpu_sched[AMDGPU_HW_IP_NUM][AMDGPU_RING_PRIO_MAX];
+	struct amdgpu_xcp_mgr *xcp_mgr;
+	struct kobject kobj;
+	uint64_t unique_id;
 };
 
 struct amdgpu_xcp_mgr {
@@ -142,10 +147,6 @@ struct amdgpu_xcp_mgr_funcs {
 	int (*suspend)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
 	int (*prepare_resume)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
 	int (*resume)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
-	int (*select_scheds)(struct amdgpu_device *adev,
-				  u32 hw_ip, u32 hw_prio, struct amdgpu_fpriv *fpriv,
-				  unsigned int *num_scheds, struct drm_gpu_scheduler ***scheds);
-	int (*update_partition_sched_list)(struct amdgpu_device *adev);
 };
 
 int amdgpu_xcp_prepare_suspend(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id);
@@ -174,18 +175,17 @@ int amdgpu_xcp_open_device(struct amdgpu_device *adev,
 			   struct drm_file *file_priv);
 void amdgpu_xcp_release_sched(struct amdgpu_device *adev,
 			      struct amdgpu_ctx_entity *entity);
-
-void amdgpu_xcp_cfg_sysfs_init(struct amdgpu_device *adev);
-void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev);
-
-#define amdgpu_xcp_select_scheds(adev, e, c, d, x, y) \
-	((adev)->xcp_mgr && (adev)->xcp_mgr->funcs && \
-	(adev)->xcp_mgr->funcs->select_scheds ? \
-	(adev)->xcp_mgr->funcs->select_scheds((adev), (e), (c), (d), (x), (y)) : -ENOENT)
-#define amdgpu_xcp_update_partition_sched_list(adev) \
-	((adev)->xcp_mgr && (adev)->xcp_mgr->funcs && \
-	(adev)->xcp_mgr->funcs->update_partition_sched_list ? \
-	(adev)->xcp_mgr->funcs->update_partition_sched_list(adev) : 0)
+int amdgpu_xcp_select_scheds(struct amdgpu_device *adev,
+			     u32 hw_ip, u32 hw_prio,
+			     struct amdgpu_fpriv *fpriv,
+			     unsigned int *num_scheds,
+			     struct drm_gpu_scheduler ***scheds);
+void amdgpu_xcp_update_supported_modes(struct amdgpu_xcp_mgr *xcp_mgr);
+int amdgpu_xcp_update_partition_sched_list(struct amdgpu_device *adev);
+int amdgpu_xcp_pre_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags);
+int amdgpu_xcp_post_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags);
+void amdgpu_xcp_sysfs_init(struct amdgpu_device *adev);
+void amdgpu_xcp_sysfs_fini(struct amdgpu_device *adev);
 
 static inline int amdgpu_xcp_get_num_xcp(struct amdgpu_xcp_mgr *xcp_mgr)
 {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
index 74b4349e345a..aad530c46a9f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
@@ -294,17 +294,53 @@ static const struct amdgpu_pcs_ras_field xgmi3x16_pcs_ras_fields[] = {
 	 SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, RxCMDPktErr)},
 };
 
+int amdgpu_xgmi_get_ext_link(struct amdgpu_device *adev, int link_num)
+{
+	int link_map_6_4_x[8] = { 0, 3, 1, 2, 7, 6, 4, 5 };
+
+	if (adev->gmc.xgmi.num_physical_nodes <= 1)
+		return -EINVAL;
+
+	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
+		if (link_num < ARRAY_SIZE(link_map_6_4_x))
+			return link_map_6_4_x[link_num];
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return -EINVAL;
+}
+
 static u32 xgmi_v6_4_get_link_status(struct amdgpu_device *adev, int global_link_num)
 {
-	const u32 smnpcs_xgmi3x16_pcs_state_hist1 = 0x11a00070;
-	const int xgmi_inst = 2;
-	u32 link_inst;
+	const u32 smn_xgmi_6_4_pcs_state_hist1[2] = { 0x11a00070, 0x11b00070 };
+	const u32 smn_xgmi_6_4_1_pcs_state_hist1[2] = { 0x12100070,
+							0x11b00070 };
+	u32 i, n;
 	u64 addr;
 
-	link_inst = global_link_num % xgmi_inst;
+	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+	case IP_VERSION(6, 4, 0):
+		n = ARRAY_SIZE(smn_xgmi_6_4_pcs_state_hist1);
+		addr = smn_xgmi_6_4_pcs_state_hist1[global_link_num % n];
+		break;
+	case IP_VERSION(6, 4, 1):
+		n = ARRAY_SIZE(smn_xgmi_6_4_1_pcs_state_hist1);
+		addr = smn_xgmi_6_4_1_pcs_state_hist1[global_link_num % n];
+		break;
+	default:
+		return U32_MAX;
+	}
+
+	i = global_link_num / n;
+
+	if (!(adev->aid_mask & BIT(i)))
+		return U32_MAX;
 
-	addr = (smnpcs_xgmi3x16_pcs_state_hist1 | (link_inst << 20)) +
-		adev->asic_funcs->encode_ext_smn_addressing(global_link_num / xgmi_inst);
+	addr += adev->asic_funcs->encode_ext_smn_addressing(i);
 
 	return RREG32_PCIE_EXT(addr);
 }
@@ -313,8 +349,12 @@ int amdgpu_get_xgmi_link_status(struct amdgpu_device *adev, int global_link_num)
 {
 	u32 xgmi_state_reg_val;
 
+	if (adev->gmc.xgmi.num_physical_nodes <= 1)
+		return -EINVAL;
+
 	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
 	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
 		xgmi_state_reg_val = xgmi_v6_4_get_link_status(adev, global_link_num);
 		break;
 	default:
@@ -818,28 +858,71 @@ int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_dev
  * num_hops[2:0] = number of hops
  */
 int amdgpu_xgmi_get_hops_count(struct amdgpu_device *adev,
-		struct amdgpu_device *peer_adev)
+			       struct amdgpu_device *peer_adev)
 {
 	struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
 	uint8_t num_hops_mask = 0x7;
 	int i;
 
+	if (!adev->gmc.xgmi.supported)
+		return 0;
+
 	for (i = 0 ; i < top->num_nodes; ++i)
 		if (top->nodes[i].node_id == peer_adev->gmc.xgmi.node_id)
 			return top->nodes[i].num_hops & num_hops_mask;
-	return	-EINVAL;
+
+	dev_err(adev->dev, "Failed to get xgmi hops count for peer %d.\n",
+		peer_adev->gmc.xgmi.physical_node_id);
+
+	return 0;
 }
 
-int amdgpu_xgmi_get_num_links(struct amdgpu_device *adev,
-		struct amdgpu_device *peer_adev)
+int amdgpu_xgmi_get_bandwidth(struct amdgpu_device *adev, struct amdgpu_device *peer_adev,
+			      enum amdgpu_xgmi_bw_mode bw_mode, enum amdgpu_xgmi_bw_unit bw_unit,
+			      uint32_t *min_bw, uint32_t *max_bw)
 {
-	struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
-	int i;
+	bool peer_mode = bw_mode == AMDGPU_XGMI_BW_MODE_PER_PEER;
+	int unit_scale = bw_unit == AMDGPU_XGMI_BW_UNIT_MBYTES ? 1000 : 1;
+	int num_lanes = adev->gmc.xgmi.max_width;
+	int speed = adev->gmc.xgmi.max_speed;
+	int num_links = !peer_mode ? 1 : -1;
 
-	for (i = 0 ; i < top->num_nodes; ++i)
-		if (top->nodes[i].node_id == peer_adev->gmc.xgmi.node_id)
-			return top->nodes[i].num_links;
-	return	-EINVAL;
+	if (!(min_bw && max_bw))
+		return -EINVAL;
+
+	*min_bw = 0;
+	*max_bw = 0;
+
+	if (!adev->gmc.xgmi.supported)
+		return -ENODATA;
+
+	if (peer_mode && !peer_adev)
+		return -EINVAL;
+
+	if (peer_mode) {
+		struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
+		int i;
+
+		for (i = 0 ; i < top->num_nodes; ++i) {
+			if (top->nodes[i].node_id != peer_adev->gmc.xgmi.node_id)
+				continue;
+
+			num_links =  top->nodes[i].num_links;
+			break;
+		}
+	}
+
+	if (num_links == -1) {
+		dev_err(adev->dev, "Failed to get number of xgmi links for peer %d.\n",
+			peer_adev->gmc.xgmi.physical_node_id);
+	} else if (num_links) {
+		int per_link_bw = (speed * num_lanes * unit_scale)/BITS_PER_BYTE;
+
+		*min_bw = per_link_bw;
+		*max_bw = num_links * per_link_bw;
+	}
+
+	return 0;
 }
 
 bool amdgpu_xgmi_get_is_sharing_enabled(struct amdgpu_device *adev,
@@ -885,28 +968,6 @@ static int amdgpu_xgmi_initialize_hive_get_data_partition(struct amdgpu_hive_inf
 	return 0;
 }
 
-static void amdgpu_xgmi_fill_topology_info(struct amdgpu_device *adev,
-	struct amdgpu_device *peer_adev)
-{
-	struct psp_xgmi_topology_info *top_info = &adev->psp.xgmi_context.top_info;
-	struct psp_xgmi_topology_info *peer_info = &peer_adev->psp.xgmi_context.top_info;
-
-	for (int i = 0; i < peer_info->num_nodes; i++) {
-		if (peer_info->nodes[i].node_id == adev->gmc.xgmi.node_id) {
-			for (int j = 0; j < top_info->num_nodes; j++) {
-				if (top_info->nodes[j].node_id == peer_adev->gmc.xgmi.node_id) {
-					peer_info->nodes[i].num_hops = top_info->nodes[j].num_hops;
-					peer_info->nodes[i].is_sharing_enabled =
-							top_info->nodes[j].is_sharing_enabled;
-					peer_info->nodes[i].num_links =
-							top_info->nodes[j].num_links;
-					return;
-				}
-			}
-		}
-	}
-}
-
 int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
 {
 	struct psp_xgmi_topology_info *top_info;
@@ -992,11 +1053,6 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
 				/* To do: continue with some node failed or disable the whole hive*/
 				goto exit_unlock;
 			}
-
-			/* fill the topology info for peers instead of getting from PSP */
-			list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
-				amdgpu_xgmi_fill_topology_info(adev, tmp_adev);
-			}
 		} else {
 			/* get latest topology info for each device from psp */
 			list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
@@ -1123,11 +1179,13 @@ static int xgmi_v6_4_0_aca_bank_parser(struct aca_handle *handle, struct aca_ban
 		if (ext_error_code != 0 && ext_error_code != 9)
 			count = 0ULL;
 
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
 		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE, count);
 		break;
 	case ACA_SMU_TYPE_CE:
 		count = ext_error_code == 6 ? count : 0ULL;
-		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_CE, count);
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type, count);
 		break;
 	default:
 		return -EINVAL;
@@ -1162,6 +1220,7 @@ static int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev, struct ras_comm
 
 	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
 	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
 		r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL,
 					&xgmi_v6_4_0_aca_info, NULL);
 		if (r)
@@ -1221,6 +1280,7 @@ static void amdgpu_xgmi_legacy_reset_ras_error_count(struct amdgpu_device *adev)
 
 	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
 	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
 		for (i = 0; i < ARRAY_SIZE(xgmi3x16_pcs_err_status_reg_v6_4); i++)
 			pcs_clear_status(adev,
 					xgmi3x16_pcs_err_status_reg_v6_4[i]);
@@ -1255,6 +1315,7 @@ static void amdgpu_xgmi_reset_ras_error_count(struct amdgpu_device *adev)
 {
 	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
 	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
 		xgmi_v6_4_0_reset_ras_error_count(adev);
 		break;
 	default:
@@ -1280,7 +1341,9 @@ static int amdgpu_xgmi_query_pcs_error_status(struct amdgpu_device *adev,
 		if (amdgpu_ip_version(adev, XGMI_HWIP, 0) ==
 		    IP_VERSION(6, 1, 0) ||
 		    amdgpu_ip_version(adev, XGMI_HWIP, 0) ==
-		    IP_VERSION(6, 4, 0)) {
+		    IP_VERSION(6, 4, 0) ||
+		    amdgpu_ip_version(adev, XGMI_HWIP, 0) ==
+		    IP_VERSION(6, 4, 1)) {
 			pcs_ras_fields = &xgmi3x16_pcs_ras_fields[0];
 			field_array_size = ARRAY_SIZE(xgmi3x16_pcs_ras_fields);
 		} else {
@@ -1388,6 +1451,7 @@ static void amdgpu_xgmi_legacy_query_ras_error_count(struct amdgpu_device *adev,
 
 	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
 	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
 		/* check xgmi3x16 pcs error */
 		for (i = 0; i < ARRAY_SIZE(xgmi3x16_pcs_err_status_reg_v6_4); i++) {
 			data = RREG32_PCIE(xgmi3x16_pcs_err_status_reg_v6_4[i]);
@@ -1484,6 +1548,7 @@ static void amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev,
 {
 	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
 	case IP_VERSION(6, 4, 0):
+	case IP_VERSION(6, 4, 1):
 		xgmi_v6_4_0_query_ras_error_count(adev, ras_error_status);
 		break;
 	default:
@@ -1671,3 +1736,43 @@ int amdgpu_xgmi_request_nps_change(struct amdgpu_device *adev,
 
 	return r;
 }
+
+bool amdgpu_xgmi_same_hive(struct amdgpu_device *adev,
+			   struct amdgpu_device *bo_adev)
+{
+	return (amdgpu_use_xgmi_p2p && adev != bo_adev &&
+		adev->gmc.xgmi.hive_id &&
+		adev->gmc.xgmi.hive_id == bo_adev->gmc.xgmi.hive_id);
+}
+
+void amdgpu_xgmi_early_init(struct amdgpu_device *adev)
+{
+	if (!adev->gmc.xgmi.supported)
+		return;
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 4, 1):
+	case IP_VERSION(9, 4, 2):
+		/* 25 GT/s */
+		adev->gmc.xgmi.max_speed = 25;
+		adev->gmc.xgmi.max_width = 16;
+		break;
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+	case IP_VERSION(9, 5, 0):
+		/* 32 GT/s */
+		adev->gmc.xgmi.max_speed = 32;
+		adev->gmc.xgmi.max_width = 16;
+		break;
+	default:
+		break;
+	}
+}
+
+void amgpu_xgmi_set_max_speed_width(struct amdgpu_device *adev,
+				    uint16_t max_speed, uint8_t max_width)
+{
+	adev->gmc.xgmi.max_speed = max_speed;
+	adev->gmc.xgmi.max_width = max_width;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
index d1282b4c6348..5f36aff17e79 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
@@ -23,7 +23,6 @@
 #define __AMDGPU_XGMI_H__
 
 #include <drm/task_barrier.h>
-#include "amdgpu_psp.h"
 #include "amdgpu_ras.h"
 
 struct amdgpu_hive_info {
@@ -55,29 +54,63 @@ struct amdgpu_pcs_ras_field {
 	uint32_t pcs_err_shift;
 };
 
-extern struct amdgpu_xgmi_ras  xgmi_ras;
+/**
+ * Bandwidth range reporting comes in two modes.
+ *
+ * PER_LINK - range for any xgmi link
+ * PER_PEER - range of max of single xgmi link to max of multiple links based on source peer
+ */
+enum amdgpu_xgmi_bw_mode {
+	AMDGPU_XGMI_BW_MODE_PER_LINK = 0,
+	AMDGPU_XGMI_BW_MODE_PER_PEER
+};
+
+enum amdgpu_xgmi_bw_unit {
+	AMDGPU_XGMI_BW_UNIT_GBYTES = 0,
+	AMDGPU_XGMI_BW_UNIT_MBYTES
+};
+
+struct amdgpu_xgmi_ras {
+	struct amdgpu_ras_block_object ras_block;
+};
+extern struct amdgpu_xgmi_ras xgmi_ras;
+
+struct amdgpu_xgmi {
+	/* from psp */
+	u64 node_id;
+	u64 hive_id;
+	/* fixed per family */
+	u64 node_segment_size;
+	/* physical node (0-3) */
+	unsigned physical_node_id;
+	/* number of nodes (0-4) */
+	unsigned num_physical_nodes;
+	/* gpu list in the same hive */
+	struct list_head head;
+	bool supported;
+	struct ras_common_if *ras_if;
+	bool connected_to_cpu;
+	struct amdgpu_xgmi_ras *ras;
+	uint16_t max_speed;
+	uint8_t max_width;
+};
+
 struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev);
 void amdgpu_put_xgmi_hive(struct amdgpu_hive_info *hive);
 int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev);
 int amdgpu_xgmi_add_device(struct amdgpu_device *adev);
 int amdgpu_xgmi_remove_device(struct amdgpu_device *adev);
 int amdgpu_xgmi_set_pstate(struct amdgpu_device *adev, int pstate);
-int amdgpu_xgmi_get_hops_count(struct amdgpu_device *adev,
-		struct amdgpu_device *peer_adev);
-int amdgpu_xgmi_get_num_links(struct amdgpu_device *adev,
-		struct amdgpu_device *peer_adev);
+int amdgpu_xgmi_get_hops_count(struct amdgpu_device *adev, struct amdgpu_device *peer_adev);
+int amdgpu_xgmi_get_bandwidth(struct amdgpu_device *adev, struct amdgpu_device *peer_adev,
+			      enum amdgpu_xgmi_bw_mode bw_mode, enum amdgpu_xgmi_bw_unit bw_unit,
+			      uint32_t *min_bw, uint32_t *max_bw);
 bool amdgpu_xgmi_get_is_sharing_enabled(struct amdgpu_device *adev,
 					struct amdgpu_device *peer_adev);
 uint64_t amdgpu_xgmi_get_relative_phy_addr(struct amdgpu_device *adev,
 					   uint64_t addr);
-static inline bool amdgpu_xgmi_same_hive(struct amdgpu_device *adev,
-		struct amdgpu_device *bo_adev)
-{
-	return (amdgpu_use_xgmi_p2p &&
-		adev != bo_adev &&
-		adev->gmc.xgmi.hive_id &&
-		adev->gmc.xgmi.hive_id == bo_adev->gmc.xgmi.hive_id);
-}
+bool amdgpu_xgmi_same_hive(struct amdgpu_device *adev,
+			   struct amdgpu_device *bo_adev);
 int amdgpu_xgmi_ras_sw_init(struct amdgpu_device *adev);
 int amdgpu_xgmi_reset_on_init(struct amdgpu_device *adev);
 
@@ -86,5 +119,15 @@ int amdgpu_xgmi_request_nps_change(struct amdgpu_device *adev,
 				   int req_nps_mode);
 int amdgpu_get_xgmi_link_status(struct amdgpu_device *adev,
 				int global_link_num);
+int amdgpu_xgmi_get_ext_link(struct amdgpu_device *adev, int link_num);
+
+void amdgpu_xgmi_early_init(struct amdgpu_device *adev);
+uint32_t amdgpu_xgmi_get_max_bandwidth(struct amdgpu_device *adev);
+
+void amgpu_xgmi_set_max_speed_width(struct amdgpu_device *adev,
+				    uint16_t max_speed, uint8_t max_width);
+
+/* Cleanup macro for use with __free(xgmi_put_hive) */
+DEFINE_FREE(xgmi_put_hive, struct amdgpu_hive_info *, if (_T) amdgpu_put_xgmi_hive(_T))
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h
index b4f9c2f4e92c..3cdb1e0eca37 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h
@@ -23,26 +23,84 @@
 #ifndef AMDGV_SRIOV_MSG__H_
 #define AMDGV_SRIOV_MSG__H_
 
-/* unit in kilobytes */
-#define AMD_SRIOV_MSG_VBIOS_OFFSET	     0
-#define AMD_SRIOV_MSG_VBIOS_SIZE_KB	     64
-#define AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB AMD_SRIOV_MSG_VBIOS_SIZE_KB
-#define AMD_SRIOV_MSG_DATAEXCHANGE_SIZE_KB   4
-#define AMD_SRIOV_MSG_TMR_OFFSET_KB	     2048
-#define AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB	     2
-#define AMD_SRIOV_RAS_TELEMETRY_SIZE_KB	     64
+#define AMD_SRIOV_MSG_SIZE_KB                           1
+
 /*
- * layout
+ * layout v1
  * 0           64KB        65KB        66KB           68KB                   132KB
  * |   VBIOS   |   PF2VF   |   VF2PF   |   Bad Page   | RAS Telemetry Region | ...
  * |   64KB    |   1KB     |   1KB     |   2KB        | 64KB                 | ...
  */
 
-#define AMD_SRIOV_MSG_SIZE_KB                   1
-#define AMD_SRIOV_MSG_PF2VF_OFFSET_KB           AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB
-#define AMD_SRIOV_MSG_VF2PF_OFFSET_KB           (AMD_SRIOV_MSG_PF2VF_OFFSET_KB + AMD_SRIOV_MSG_SIZE_KB)
-#define AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB        (AMD_SRIOV_MSG_VF2PF_OFFSET_KB + AMD_SRIOV_MSG_SIZE_KB)
-#define AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB   (AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB + AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB)
+/*
+ * layout v2 (offsets are dynamically allocated and the offsets below are examples)
+ * 0           1KB         64KB        65KB        66KB           68KB                   132KB
+ * |  INITD_H  |   VBIOS   |   PF2VF   |   VF2PF   |   Bad Page   | RAS Telemetry Region | ...
+ * |   1KB     |   64KB    |   1KB     |   1KB     |   2KB        | 64KB                 | ...
+ *
+ * Note: PF2VF + VF2PF + Bad Page = DataExchange region (allocated contiguously)
+ */
+
+/* v1 layout sizes */
+#define AMD_SRIOV_MSG_VBIOS_SIZE_KB_V1			64
+#define AMD_SRIOV_MSG_PF2VF_SIZE_KB_V1			1
+#define AMD_SRIOV_MSG_VF2PF_SIZE_KB_V1			1
+#define AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB_V1		2
+#define AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1		64
+#define AMD_SRIOV_MSG_DATAEXCHANGE_SIZE_KB_V1		\
+	(AMD_SRIOV_MSG_PF2VF_SIZE_KB_V1 + AMD_SRIOV_MSG_VF2PF_SIZE_KB_V1 + \
+	 AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB_V1)
+
+/* v1 offsets */
+#define AMD_SRIOV_MSG_VBIOS_OFFSET_V1			0
+#define AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB_V1		AMD_SRIOV_MSG_VBIOS_SIZE_KB_V1
+#define AMD_SRIOV_MSG_TMR_OFFSET_KB			2048
+#define AMD_SRIOV_MSG_PF2VF_OFFSET_KB_V1		AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB_V1
+#define AMD_SRIOV_MSG_VF2PF_OFFSET_KB_V1		\
+	(AMD_SRIOV_MSG_PF2VF_OFFSET_KB_V1 + AMD_SRIOV_MSG_SIZE_KB)
+#define AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB_V1		\
+	(AMD_SRIOV_MSG_VF2PF_OFFSET_KB_V1 + AMD_SRIOV_MSG_SIZE_KB)
+#define AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB_V1	\
+	(AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB_V1 + AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB_V1)
+#define AMD_SRIOV_MSG_INIT_DATA_TOT_SIZE_KB_V1		\
+	(AMD_SRIOV_MSG_VBIOS_SIZE_KB_V1 + AMD_SRIOV_MSG_DATAEXCHANGE_SIZE_KB_V1 + \
+	 AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1)
+
+enum amd_sriov_crit_region_version {
+	GPU_CRIT_REGION_V1 = 1,
+	GPU_CRIT_REGION_V2 = 2,
+};
+
+/* v2 layout offset enum (in order of allocation) */
+enum amd_sriov_msg_table_id_enum {
+	AMD_SRIOV_MSG_IPD_TABLE_ID = 0,
+	AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID,
+	AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID,
+	AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID,
+	AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID,
+	AMD_SRIOV_MSG_INITD_H_TABLE_ID,
+	AMD_SRIOV_MSG_MAX_TABLE_ID,
+};
+
+struct amd_sriov_msg_init_data_header {
+	char     signature[4];  /* "INDA"  */
+	uint32_t version;
+	uint32_t checksum;
+	uint32_t initdata_offset; /* 0 */
+	uint32_t initdata_size_in_kb; /* 5MB */
+	uint32_t valid_tables;
+	uint32_t vbios_img_offset;
+	uint32_t vbios_img_size_in_kb;
+	uint32_t dataexchange_offset;
+	uint32_t dataexchange_size_in_kb;
+	uint32_t ras_tele_info_offset;
+	uint32_t ras_tele_info_size_in_kb;
+	uint32_t ip_discovery_offset;
+	uint32_t ip_discovery_size_in_kb;
+	uint32_t bad_page_info_offset;
+	uint32_t bad_page_size_in_kb;
+	uint32_t reserved[8];
+};
 
 /*
  * PF2VF history log:
@@ -97,21 +155,25 @@ union amd_sriov_msg_feature_flags {
 		uint32_t pp_one_vf_mode		: 1;
 		uint32_t reg_indirect_acc	: 1;
 		uint32_t av1_support		: 1;
-		uint32_t vcn_rb_decouple 	: 1;
+		uint32_t vcn_rb_decouple	: 1;
 		uint32_t mes_info_dump_enable	: 1;
 		uint32_t ras_caps		: 1;
 		uint32_t ras_telemetry		: 1;
-		uint32_t reserved		: 21;
+		uint32_t ras_cper		: 1;
+		uint32_t xgmi_ta_ext_peer_link	: 1;
+		uint32_t reserved		: 19;
 	} flags;
 	uint32_t all;
 };
 
 union amd_sriov_reg_access_flags {
 	struct {
-		uint32_t vf_reg_access_ih	: 1;
-		uint32_t vf_reg_access_mmhub	: 1;
-		uint32_t vf_reg_access_gc	: 1;
-		uint32_t reserved		: 29;
+		uint32_t vf_reg_access_ih		: 1;
+		uint32_t vf_reg_access_mmhub		: 1;
+		uint32_t vf_reg_access_gc		: 1;
+		uint32_t vf_reg_access_l1_tlb_cntl	: 1;
+		uint32_t vf_reg_access_sq_config	: 1;
+		uint32_t reserved			: 27;
 	} flags;
 	uint32_t all;
 };
@@ -137,8 +199,9 @@ union amd_sriov_ras_caps {
 		uint64_t block_jpeg			: 1;
 		uint64_t block_ih			: 1;
 		uint64_t block_mpio			: 1;
+		uint64_t block_mmsch			: 1;
 		uint64_t poison_propogation_mode	: 1;
-		uint64_t reserved			: 44;
+		uint64_t reserved			: 43;
 	} bits;
 	uint64_t all;
 };
@@ -328,21 +391,29 @@ enum amd_sriov_mailbox_request_message {
 	MB_REQ_MSG_READY_TO_RESET = 201,
 	MB_REQ_MSG_RAS_POISON = 202,
 	MB_REQ_RAS_ERROR_COUNT = 203,
+	MB_REQ_RAS_CPER_DUMP = 204,
+	MB_REQ_RAS_BAD_PAGES = 205,
 };
 
 /* mailbox message send from host to guest  */
 enum amd_sriov_mailbox_response_message {
-	MB_RES_MSG_CLR_MSG_BUF = 0,
-	MB_RES_MSG_READY_TO_ACCESS_GPU = 1,
-	MB_RES_MSG_FLR_NOTIFICATION,
-	MB_RES_MSG_FLR_NOTIFICATION_COMPLETION,
-	MB_RES_MSG_SUCCESS,
-	MB_RES_MSG_FAIL,
-	MB_RES_MSG_QUERY_ALIVE,
-	MB_RES_MSG_GPU_INIT_DATA_READY,
-	MB_RES_MSG_RAS_ERROR_COUNT_READY = 11,
-
-	MB_RES_MSG_TEXT_MESSAGE = 255
+	MB_RES_MSG_CLR_MSG_BUF			= 0,
+	MB_RES_MSG_READY_TO_ACCESS_GPU		= 1,
+	MB_RES_MSG_FLR_NOTIFICATION		= 2,
+	MB_RES_MSG_FLR_NOTIFICATION_COMPLETION  = 3,
+	MB_RES_MSG_SUCCESS			= 4,
+	MB_RES_MSG_FAIL				= 5,
+	MB_RES_MSG_QUERY_ALIVE			= 6,
+	MB_RES_MSG_GPU_INIT_DATA_READY		= 7,
+	MB_RES_MSG_RAS_POISON_READY		= 8,
+	MB_RES_MSG_PF_SOFT_FLR_NOTIFICATION	= 9,
+	MB_RES_MSG_GPU_RMA			= 10,
+	MB_RES_MSG_RAS_ERROR_COUNT_READY	= 11,
+	MB_REQ_RAS_CPER_DUMP_READY		= 14,
+	MB_RES_MSG_RAS_BAD_PAGES_READY		= 15,
+	MB_RES_MSG_RAS_BAD_PAGES_NOTIFICATION	= 16,
+	MB_RES_MSG_UNRECOV_ERR_NOTIFICATION	= 17,
+	MB_RES_MSG_TEXT_MESSAGE			= 255
 };
 
 enum amd_sriov_ras_telemetry_gpu_block {
@@ -386,11 +457,25 @@ struct amd_sriov_ras_telemetry_error_count {
 	} block[RAS_TELEMETRY_GPU_BLOCK_COUNT];
 };
 
+struct amd_sriov_ras_cper_dump {
+	uint32_t more;
+	uint64_t overflow_count;
+	uint64_t count;
+	uint64_t wptr;
+	uint32_t buf[];
+};
+
+struct amd_sriov_ras_chk_criti {
+	uint32_t hit;
+};
+
 struct amdsriov_ras_telemetry {
 	struct amd_sriov_ras_telemetry_header header;
 
 	union {
 		struct amd_sriov_ras_telemetry_error_count error_count;
+		struct amd_sriov_ras_cper_dump cper_dump;
+		struct amd_sriov_ras_chk_criti chk_criti;
 	} body;
 };
 
diff --git a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c
index e157d6d857b6..f9e2edf5260b 100644
--- a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c
+++ b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c
@@ -29,12 +29,11 @@
 #include "gfx_v9_4_3.h"
 #include "gfxhub_v1_2.h"
 #include "sdma_v4_4_2.h"
+#include "amdgpu_ip.h"
 
 #define XCP_INST_MASK(num_inst, xcp_id)                                        \
 	(num_inst ? GENMASK(num_inst - 1, 0) << (xcp_id * num_inst) : 0)
 
-#define AMDGPU_XCP_OPS_KFD	(1 << 0)
-
 void aqua_vanjaram_doorbell_index_init(struct amdgpu_device *adev)
 {
 	int i;
@@ -62,233 +61,6 @@ void aqua_vanjaram_doorbell_index_init(struct amdgpu_device *adev)
 	adev->doorbell_index.max_assignment = AMDGPU_DOORBELL_LAYOUT1_MAX_ASSIGNMENT << 1;
 }
 
-static bool aqua_vanjaram_xcp_vcn_shared(struct amdgpu_device *adev)
-{
-	return (adev->xcp_mgr->num_xcps > adev->vcn.num_vcn_inst);
-}
-
-static void aqua_vanjaram_set_xcp_id(struct amdgpu_device *adev,
-			     uint32_t inst_idx, struct amdgpu_ring *ring)
-{
-	int xcp_id;
-	enum AMDGPU_XCP_IP_BLOCK ip_blk;
-	uint32_t inst_mask;
-
-	ring->xcp_id = AMDGPU_XCP_NO_PARTITION;
-	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)
-		adev->gfx.enforce_isolation[0].xcp_id = ring->xcp_id;
-	if (adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE)
-		return;
-
-	inst_mask = 1 << inst_idx;
-
-	switch (ring->funcs->type) {
-	case AMDGPU_HW_IP_GFX:
-	case AMDGPU_RING_TYPE_COMPUTE:
-	case AMDGPU_RING_TYPE_KIQ:
-		ip_blk = AMDGPU_XCP_GFX;
-		break;
-	case AMDGPU_RING_TYPE_SDMA:
-		ip_blk = AMDGPU_XCP_SDMA;
-		break;
-	case AMDGPU_RING_TYPE_VCN_ENC:
-	case AMDGPU_RING_TYPE_VCN_JPEG:
-		ip_blk = AMDGPU_XCP_VCN;
-		break;
-	default:
-		DRM_ERROR("Not support ring type %d!", ring->funcs->type);
-		return;
-	}
-
-	for (xcp_id = 0; xcp_id < adev->xcp_mgr->num_xcps; xcp_id++) {
-		if (adev->xcp_mgr->xcp[xcp_id].ip[ip_blk].inst_mask & inst_mask) {
-			ring->xcp_id = xcp_id;
-			dev_dbg(adev->dev, "ring:%s xcp_id :%u", ring->name,
-				ring->xcp_id);
-			if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)
-				adev->gfx.enforce_isolation[xcp_id].xcp_id = xcp_id;
-			break;
-		}
-	}
-}
-
-static void aqua_vanjaram_xcp_gpu_sched_update(
-		struct amdgpu_device *adev,
-		struct amdgpu_ring *ring,
-		unsigned int sel_xcp_id)
-{
-	unsigned int *num_gpu_sched;
-
-	num_gpu_sched = &adev->xcp_mgr->xcp[sel_xcp_id]
-			.gpu_sched[ring->funcs->type][ring->hw_prio].num_scheds;
-	adev->xcp_mgr->xcp[sel_xcp_id].gpu_sched[ring->funcs->type][ring->hw_prio]
-			.sched[(*num_gpu_sched)++] = &ring->sched;
-	DRM_DEBUG("%s :[%d] gpu_sched[%d][%d] = %d", ring->name,
-			sel_xcp_id, ring->funcs->type,
-			ring->hw_prio, *num_gpu_sched);
-}
-
-static int aqua_vanjaram_xcp_sched_list_update(
-		struct amdgpu_device *adev)
-{
-	struct amdgpu_ring *ring;
-	int i;
-
-	for (i = 0; i < MAX_XCP; i++) {
-		atomic_set(&adev->xcp_mgr->xcp[i].ref_cnt, 0);
-		memset(adev->xcp_mgr->xcp[i].gpu_sched, 0, sizeof(adev->xcp_mgr->xcp->gpu_sched));
-	}
-
-	if (adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE)
-		return 0;
-
-	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
-		ring = adev->rings[i];
-		if (!ring || !ring->sched.ready || ring->no_scheduler)
-			continue;
-
-		aqua_vanjaram_xcp_gpu_sched_update(adev, ring, ring->xcp_id);
-
-		/* VCN may be shared by two partitions under CPX MODE in certain
-		 * configs.
-		 */
-		if ((ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC ||
-		     ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG) &&
-		    aqua_vanjaram_xcp_vcn_shared(adev))
-			aqua_vanjaram_xcp_gpu_sched_update(adev, ring, ring->xcp_id + 1);
-	}
-
-	return 0;
-}
-
-static int aqua_vanjaram_update_partition_sched_list(struct amdgpu_device *adev)
-{
-	int i;
-
-	for (i = 0; i < adev->num_rings; i++) {
-		struct amdgpu_ring *ring = adev->rings[i];
-
-		if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE ||
-			ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
-			aqua_vanjaram_set_xcp_id(adev, ring->xcc_id, ring);
-		else
-			aqua_vanjaram_set_xcp_id(adev, ring->me, ring);
-	}
-
-	return aqua_vanjaram_xcp_sched_list_update(adev);
-}
-
-static int aqua_vanjaram_select_scheds(
-		struct amdgpu_device *adev,
-		u32 hw_ip,
-		u32 hw_prio,
-		struct amdgpu_fpriv *fpriv,
-		unsigned int *num_scheds,
-		struct drm_gpu_scheduler ***scheds)
-{
-	u32 sel_xcp_id;
-	int i;
-
-	if (fpriv->xcp_id == AMDGPU_XCP_NO_PARTITION) {
-		u32 least_ref_cnt = ~0;
-
-		fpriv->xcp_id = 0;
-		for (i = 0; i < adev->xcp_mgr->num_xcps; i++) {
-			u32 total_ref_cnt;
-
-			total_ref_cnt = atomic_read(&adev->xcp_mgr->xcp[i].ref_cnt);
-			if (total_ref_cnt < least_ref_cnt) {
-				fpriv->xcp_id = i;
-				least_ref_cnt = total_ref_cnt;
-			}
-		}
-	}
-	sel_xcp_id = fpriv->xcp_id;
-
-	if (adev->xcp_mgr->xcp[sel_xcp_id].gpu_sched[hw_ip][hw_prio].num_scheds) {
-		*num_scheds = adev->xcp_mgr->xcp[fpriv->xcp_id].gpu_sched[hw_ip][hw_prio].num_scheds;
-		*scheds = adev->xcp_mgr->xcp[fpriv->xcp_id].gpu_sched[hw_ip][hw_prio].sched;
-		atomic_inc(&adev->xcp_mgr->xcp[sel_xcp_id].ref_cnt);
-		DRM_DEBUG("Selected partition #%d", sel_xcp_id);
-	} else {
-		DRM_ERROR("Failed to schedule partition #%d.", sel_xcp_id);
-		return -ENOENT;
-	}
-
-	return 0;
-}
-
-static int8_t aqua_vanjaram_logical_to_dev_inst(struct amdgpu_device *adev,
-					 enum amd_hw_ip_block_type block,
-					 int8_t inst)
-{
-	int8_t dev_inst;
-
-	switch (block) {
-	case GC_HWIP:
-	case SDMA0_HWIP:
-	/* Both JPEG and VCN as JPEG is only alias of VCN */
-	case VCN_HWIP:
-		dev_inst = adev->ip_map.dev_inst[block][inst];
-		break;
-	default:
-		/* For rest of the IPs, no look up required.
-		 * Assume 'logical instance == physical instance' for all configs. */
-		dev_inst = inst;
-		break;
-	}
-
-	return dev_inst;
-}
-
-static uint32_t aqua_vanjaram_logical_to_dev_mask(struct amdgpu_device *adev,
-					 enum amd_hw_ip_block_type block,
-					 uint32_t mask)
-{
-	uint32_t dev_mask = 0;
-	int8_t log_inst, dev_inst;
-
-	while (mask) {
-		log_inst = ffs(mask) - 1;
-		dev_inst = aqua_vanjaram_logical_to_dev_inst(adev, block, log_inst);
-		dev_mask |= (1 << dev_inst);
-		mask &= ~(1 << log_inst);
-	}
-
-	return dev_mask;
-}
-
-static void aqua_vanjaram_populate_ip_map(struct amdgpu_device *adev,
-					  enum amd_hw_ip_block_type ip_block,
-					  uint32_t inst_mask)
-{
-	int l = 0, i;
-
-	while (inst_mask) {
-		i = ffs(inst_mask) - 1;
-		adev->ip_map.dev_inst[ip_block][l++] = i;
-		inst_mask &= ~(1 << i);
-	}
-	for (; l < HWIP_MAX_INSTANCE; l++)
-		adev->ip_map.dev_inst[ip_block][l] = -1;
-}
-
-void aqua_vanjaram_ip_map_init(struct amdgpu_device *adev)
-{
-	u32 ip_map[][2] = {
-		{ GC_HWIP, adev->gfx.xcc_mask },
-		{ SDMA0_HWIP, adev->sdma.sdma_mask },
-		{ VCN_HWIP, adev->vcn.inst_mask },
-	};
-	int i;
-
-	for (i = 0; i < ARRAY_SIZE(ip_map); ++i)
-		aqua_vanjaram_populate_ip_map(adev, ip_map[i][0], ip_map[i][1]);
-
-	adev->ip_map.logical_to_dev_inst = aqua_vanjaram_logical_to_dev_inst;
-	adev->ip_map.logical_to_dev_mask = aqua_vanjaram_logical_to_dev_mask;
-}
-
 /* Fixed pattern for smn addressing on different AIDs:
  *   bit[34]: indicate cross AID access
  *   bit[33:32]: indicate target AID id
@@ -352,11 +124,14 @@ static int aqua_vanjaram_query_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr)
 
 	if (adev->nbio.funcs->get_compute_partition_mode) {
 		mode = adev->nbio.funcs->get_compute_partition_mode(adev);
-		if (mode != derv_mode)
+		if (mode != derv_mode) {
 			dev_warn(
 				adev->dev,
 				"Mismatch in compute partition mode - reported : %d derived : %d",
 				mode, derv_mode);
+			if (derv_mode == AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE)
+				amdgpu_device_bus_status_check(adev);
+		}
 	}
 
 	return mode;
@@ -447,52 +222,74 @@ static int __aqua_vanjaram_get_xcp_ip_info(struct amdgpu_xcp_mgr *xcp_mgr, int x
 	return 0;
 }
 
-static int aqua_vanjaram_get_xcp_res_info(struct amdgpu_xcp_mgr *xcp_mgr,
-					  int mode,
-					  struct amdgpu_xcp_cfg *xcp_cfg)
+static int __aqua_vanjaram_get_px_mode_info(struct amdgpu_xcp_mgr *xcp_mgr,
+					    int px_mode, int *num_xcp,
+					    uint16_t *nps_modes)
 {
 	struct amdgpu_device *adev = xcp_mgr->adev;
-	int max_res[AMDGPU_XCP_RES_MAX] = {};
-	bool res_lt_xcp;
-	int num_xcp, i;
-	u16 nps_modes;
+	uint32_t gc_ver = amdgpu_ip_version(adev, GC_HWIP, 0);
 
-	if (!(xcp_mgr->supp_xcp_modes & BIT(mode)))
+	if (!num_xcp || !nps_modes || !(xcp_mgr->supp_xcp_modes & BIT(px_mode)))
 		return -EINVAL;
 
-	max_res[AMDGPU_XCP_RES_XCC] = NUM_XCC(adev->gfx.xcc_mask);
-	max_res[AMDGPU_XCP_RES_DMA] = adev->sdma.num_instances;
-	max_res[AMDGPU_XCP_RES_DEC] = adev->vcn.num_vcn_inst;
-	max_res[AMDGPU_XCP_RES_JPEG] = adev->jpeg.num_jpeg_inst;
-
-	switch (mode) {
+	switch (px_mode) {
 	case AMDGPU_SPX_PARTITION_MODE:
-		num_xcp = 1;
-		nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE);
+		*num_xcp = 1;
+		*nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE);
 		break;
 	case AMDGPU_DPX_PARTITION_MODE:
-		num_xcp = 2;
-		nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE);
+		*num_xcp = 2;
+		*nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
+			     BIT(AMDGPU_NPS2_PARTITION_MODE);
 		break;
 	case AMDGPU_TPX_PARTITION_MODE:
-		num_xcp = 3;
-		nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
-			    BIT(AMDGPU_NPS4_PARTITION_MODE);
+		*num_xcp = 3;
+		*nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
+			     BIT(AMDGPU_NPS4_PARTITION_MODE);
 		break;
 	case AMDGPU_QPX_PARTITION_MODE:
-		num_xcp = 4;
-		nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
-			    BIT(AMDGPU_NPS4_PARTITION_MODE);
+		*num_xcp = 4;
+		*nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
+			     BIT(AMDGPU_NPS4_PARTITION_MODE);
+		if (gc_ver == IP_VERSION(9, 5, 0))
+			*nps_modes |= BIT(AMDGPU_NPS2_PARTITION_MODE);
 		break;
 	case AMDGPU_CPX_PARTITION_MODE:
-		num_xcp = NUM_XCC(adev->gfx.xcc_mask);
-		nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
-			    BIT(AMDGPU_NPS4_PARTITION_MODE);
+		*num_xcp = NUM_XCC(adev->gfx.xcc_mask);
+		*nps_modes = BIT(AMDGPU_NPS1_PARTITION_MODE) |
+			     BIT(AMDGPU_NPS4_PARTITION_MODE);
+		if (gc_ver == IP_VERSION(9, 5, 0))
+			*nps_modes |= BIT(AMDGPU_NPS2_PARTITION_MODE);
 		break;
 	default:
 		return -EINVAL;
 	}
 
+	return 0;
+}
+
+static int aqua_vanjaram_get_xcp_res_info(struct amdgpu_xcp_mgr *xcp_mgr,
+					  int mode,
+					  struct amdgpu_xcp_cfg *xcp_cfg)
+{
+	struct amdgpu_device *adev = xcp_mgr->adev;
+	int max_res[AMDGPU_XCP_RES_MAX] = {};
+	bool res_lt_xcp;
+	int num_xcp, i, r;
+	u16 nps_modes;
+
+	if (!(xcp_mgr->supp_xcp_modes & BIT(mode)))
+		return -EINVAL;
+
+	max_res[AMDGPU_XCP_RES_XCC] = NUM_XCC(adev->gfx.xcc_mask);
+	max_res[AMDGPU_XCP_RES_DMA] = adev->sdma.num_instances;
+	max_res[AMDGPU_XCP_RES_DEC] = adev->vcn.num_vcn_inst;
+	max_res[AMDGPU_XCP_RES_JPEG] = adev->jpeg.num_jpeg_inst;
+
+	r = __aqua_vanjaram_get_px_mode_info(xcp_mgr, mode, &num_xcp, &nps_modes);
+	if (r)
+		return r;
+
 	xcp_cfg->compatible_nps_modes =
 		(adev->gmc.supported_nps_modes & nps_modes);
 	xcp_cfg->num_res = ARRAY_SIZE(max_res);
@@ -541,27 +338,31 @@ static bool __aqua_vanjaram_is_valid_mode(struct amdgpu_xcp_mgr *xcp_mgr,
 					  enum amdgpu_gfx_partition mode)
 {
 	struct amdgpu_device *adev = xcp_mgr->adev;
-	int num_xcc, num_xccs_per_xcp;
+	int num_xcc, num_xccs_per_xcp, r;
+	int num_xcp, nps_mode;
+	u16 supp_nps_modes;
+	bool comp_mode;
+
+	nps_mode = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
+	r = __aqua_vanjaram_get_px_mode_info(xcp_mgr, mode, &num_xcp,
+					       &supp_nps_modes);
+	if (r)
+		return false;
 
+	comp_mode = !!(BIT(nps_mode) & supp_nps_modes);
 	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
 	switch (mode) {
 	case AMDGPU_SPX_PARTITION_MODE:
-		return adev->gmc.num_mem_partitions == 1 && num_xcc > 0;
+		return comp_mode && num_xcc > 0;
 	case AMDGPU_DPX_PARTITION_MODE:
-		return adev->gmc.num_mem_partitions <= 2 && (num_xcc % 4) == 0;
+		return comp_mode && (num_xcc % 4) == 0;
 	case AMDGPU_TPX_PARTITION_MODE:
-		return (adev->gmc.num_mem_partitions == 1 ||
-			adev->gmc.num_mem_partitions == 3) &&
-		       ((num_xcc % 3) == 0);
+		return comp_mode && ((num_xcc % 3) == 0);
 	case AMDGPU_QPX_PARTITION_MODE:
 		num_xccs_per_xcp = num_xcc / 4;
-		return (adev->gmc.num_mem_partitions == 1 ||
-			adev->gmc.num_mem_partitions == 4) &&
-		       (num_xccs_per_xcp >= 2);
+		return comp_mode && (num_xccs_per_xcp >= 2);
 	case AMDGPU_CPX_PARTITION_MODE:
-		return ((num_xcc > 1) &&
-		       (adev->gmc.num_mem_partitions == 1 || adev->gmc.num_mem_partitions == 4) &&
-		       (num_xcc % adev->gmc.num_mem_partitions) == 0);
+		return comp_mode && (num_xcc > 1);
 	default:
 		return false;
 	}
@@ -569,72 +370,6 @@ static bool __aqua_vanjaram_is_valid_mode(struct amdgpu_xcp_mgr *xcp_mgr,
 	return false;
 }
 
-static int __aqua_vanjaram_pre_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags)
-{
-	/* TODO:
-	 * Stop user queues and threads, and make sure GPU is empty of work.
-	 */
-
-	if (flags & AMDGPU_XCP_OPS_KFD)
-		amdgpu_amdkfd_device_fini_sw(xcp_mgr->adev);
-
-	return 0;
-}
-
-static int __aqua_vanjaram_post_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags)
-{
-	int ret = 0;
-
-	if (flags & AMDGPU_XCP_OPS_KFD) {
-		amdgpu_amdkfd_device_probe(xcp_mgr->adev);
-		amdgpu_amdkfd_device_init(xcp_mgr->adev);
-		/* If KFD init failed, return failure */
-		if (!xcp_mgr->adev->kfd.init_complete)
-			ret = -EIO;
-	}
-
-	return ret;
-}
-
-static void
-__aqua_vanjaram_update_supported_modes(struct amdgpu_xcp_mgr *xcp_mgr)
-{
-	struct amdgpu_device *adev = xcp_mgr->adev;
-
-	xcp_mgr->supp_xcp_modes = 0;
-
-	switch (NUM_XCC(adev->gfx.xcc_mask)) {
-	case 8:
-		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
-					  BIT(AMDGPU_DPX_PARTITION_MODE) |
-					  BIT(AMDGPU_QPX_PARTITION_MODE) |
-					  BIT(AMDGPU_CPX_PARTITION_MODE);
-		break;
-	case 6:
-		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
-					  BIT(AMDGPU_TPX_PARTITION_MODE) |
-					  BIT(AMDGPU_CPX_PARTITION_MODE);
-		break;
-	case 4:
-		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
-					  BIT(AMDGPU_DPX_PARTITION_MODE) |
-					  BIT(AMDGPU_CPX_PARTITION_MODE);
-		break;
-	/* this seems only existing in emulation phase */
-	case 2:
-		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
-					  BIT(AMDGPU_CPX_PARTITION_MODE);
-		break;
-	case 1:
-		xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) |
-					  BIT(AMDGPU_CPX_PARTITION_MODE);
-		break;
-
-	default:
-		break;
-	}
-}
-
 static void __aqua_vanjaram_update_available_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr)
 {
 	int mode;
@@ -672,7 +407,8 @@ static int aqua_vanjaram_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr,
 		return -EINVAL;
 	}
 
-	if (adev->kfd.init_complete && !amdgpu_in_reset(adev))
+	if (adev->kfd.init_complete && !amdgpu_in_reset(adev) &&
+		!adev->in_suspend)
 		flags |= AMDGPU_XCP_OPS_KFD;
 
 	if (flags & AMDGPU_XCP_OPS_KFD) {
@@ -681,7 +417,7 @@ static int aqua_vanjaram_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr,
 			goto out;
 	}
 
-	ret = __aqua_vanjaram_pre_partition_switch(xcp_mgr, flags);
+	ret = amdgpu_xcp_pre_partition_switch(xcp_mgr, flags);
 	if (ret)
 		goto unlock;
 
@@ -694,7 +430,7 @@ static int aqua_vanjaram_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr,
 	*num_xcps = num_xcc / num_xcc_per_xcp;
 	amdgpu_xcp_init(xcp_mgr, *num_xcps, mode);
 
-	ret = __aqua_vanjaram_post_partition_switch(xcp_mgr, flags);
+	ret = amdgpu_xcp_post_partition_switch(xcp_mgr, flags);
 	if (!ret)
 		__aqua_vanjaram_update_available_partition_mode(xcp_mgr);
 unlock:
@@ -777,9 +513,6 @@ struct amdgpu_xcp_mgr_funcs aqua_vanjaram_xcp_funcs = {
 	.get_ip_details = &aqua_vanjaram_get_xcp_ip_details,
 	.get_xcp_res_info = &aqua_vanjaram_get_xcp_res_info,
 	.get_xcp_mem_id = &aqua_vanjaram_get_xcp_mem_id,
-	.select_scheds = &aqua_vanjaram_select_scheds,
-	.update_partition_sched_list =
-		&aqua_vanjaram_update_partition_sched_list
 };
 
 static int aqua_vanjaram_xcp_mgr_init(struct amdgpu_device *adev)
@@ -794,7 +527,7 @@ static int aqua_vanjaram_xcp_mgr_init(struct amdgpu_device *adev)
 	if (ret)
 		return ret;
 
-	__aqua_vanjaram_update_supported_modes(adev->xcp_mgr);
+	amdgpu_xcp_update_supported_modes(adev->xcp_mgr);
 	/* TODO: Default memory node affinity init */
 
 	return ret;
@@ -834,7 +567,7 @@ int aqua_vanjaram_init_soc_config(struct amdgpu_device *adev)
 	if (ret)
 		return ret;
 
-	aqua_vanjaram_ip_map_init(adev);
+	amdgpu_ip_map_init(adev);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/atom.c b/drivers/gpu/drm/amd/amdgpu/atom.c
index 81d195d366ce..7a063e44d429 100644
--- a/drivers/gpu/drm/amd/amdgpu/atom.c
+++ b/drivers/gpu/drm/amd/amdgpu/atom.c
@@ -1246,6 +1246,10 @@ static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index,
 	ectx.last_jump_jiffies = 0;
 	if (ws) {
 		ectx.ws = kcalloc(4, ws, GFP_KERNEL);
+		if (!ectx.ws) {
+			ret = -ENOMEM;
+			goto free;
+		}
 		ectx.ws_size = ws;
 	} else {
 		ectx.ws = NULL;
@@ -1444,6 +1448,7 @@ static void atom_get_vbios_pn(struct atom_context *ctx)
 		if (vbios_str == NULL)
 			vbios_str += sizeof(BIOS_ATOM_PREFIX) - 1;
 	}
+	OPTIMIZER_HIDE_VAR(vbios_str);
 	if (vbios_str != NULL && *vbios_str == 0)
 		vbios_str++;
 
@@ -1493,6 +1498,28 @@ static void atom_get_vbios_version(struct atom_context *ctx)
 	}
 }
 
+static void atom_get_vbios_build(struct atom_context *ctx)
+{
+	unsigned char *atom_rom_hdr;
+	unsigned char *str;
+	uint16_t base, len;
+
+	base = CU16(ATOM_ROM_TABLE_PTR);
+	atom_rom_hdr = CSTR(base);
+
+	str = CSTR(CU16(base + ATOM_ROM_CFG_PTR));
+	/* Skip config string */
+	while (str < atom_rom_hdr && *str++)
+		;
+	/* Skip change list string */
+	while (str < atom_rom_hdr && *str++)
+		;
+
+	len = min(atom_rom_hdr - str, STRLEN_NORMAL);
+	if (len)
+		strscpy(ctx->build_num, str, len);
+}
+
 struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios)
 {
 	int base;
@@ -1553,6 +1580,7 @@ struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios)
 	atom_get_vbios_pn(ctx);
 	atom_get_vbios_date(ctx);
 	atom_get_vbios_version(ctx);
+	atom_get_vbios_build(ctx);
 
 	return ctx;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/atom.h b/drivers/gpu/drm/amd/amdgpu/atom.h
index b807f6639a4c..825ff28731f5 100644
--- a/drivers/gpu/drm/amd/amdgpu/atom.h
+++ b/drivers/gpu/drm/amd/amdgpu/atom.h
@@ -37,6 +37,7 @@ struct drm_device;
 #define ATOM_ROM_MAGIC		"ATOM"
 #define ATOM_ROM_MAGIC_PTR	4
 
+#define ATOM_ROM_CFG_PTR 0xC
 #define ATOM_ROM_MSG_PTR	0x10
 #define ATOM_ROM_CMD_PTR	0x1E
 #define ATOM_ROM_DATA_PTR	0x20
@@ -151,6 +152,7 @@ struct atom_context {
 	uint32_t version;
 	uint8_t vbios_ver_str[STRLEN_NORMAL];
 	uint8_t date[STRLEN_NORMAL];
+	uint8_t build_num[STRLEN_NORMAL];
 };
 
 extern int amdgpu_atom_debug;
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_dp.c b/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
index 622634c08c7b..492813ab1b54 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
@@ -430,7 +430,7 @@ void amdgpu_atombios_dp_set_link_config(struct drm_connector *connector,
 }
 
 int amdgpu_atombios_dp_mode_valid_helper(struct drm_connector *connector,
-				  struct drm_display_mode *mode)
+				  const struct drm_display_mode *mode)
 {
 	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 	struct amdgpu_connector_atom_dig *dig_connector;
@@ -458,8 +458,8 @@ bool amdgpu_atombios_dp_needs_link_train(struct amdgpu_connector *amdgpu_connect
 	u8 link_status[DP_LINK_STATUS_SIZE];
 	struct amdgpu_connector_atom_dig *dig = amdgpu_connector->con_priv;
 
-	if (drm_dp_dpcd_read_link_status(&amdgpu_connector->ddc_bus->aux, link_status)
-	    <= 0)
+	if (drm_dp_dpcd_read_link_status(&amdgpu_connector->ddc_bus->aux,
+					 link_status) < 0)
 		return false;
 	if (drm_dp_channel_eq_ok(link_status, dig->dp_lane_count))
 		return false;
@@ -616,7 +616,7 @@ amdgpu_atombios_dp_link_train_cr(struct amdgpu_atombios_dp_link_train_info *dp_i
 		drm_dp_link_train_clock_recovery_delay(dp_info->aux, dp_info->dpcd);
 
 		if (drm_dp_dpcd_read_link_status(dp_info->aux,
-						 dp_info->link_status) <= 0) {
+						 dp_info->link_status) < 0) {
 			DRM_ERROR("displayport link status failed\n");
 			break;
 		}
@@ -681,7 +681,7 @@ amdgpu_atombios_dp_link_train_ce(struct amdgpu_atombios_dp_link_train_info *dp_i
 		drm_dp_link_train_channel_eq_delay(dp_info->aux, dp_info->dpcd);
 
 		if (drm_dp_dpcd_read_link_status(dp_info->aux,
-						 dp_info->link_status) <= 0) {
+						 dp_info->link_status) < 0) {
 			DRM_ERROR("displayport link status failed\n");
 			break;
 		}
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_dp.h b/drivers/gpu/drm/amd/amdgpu/atombios_dp.h
index f59d85eaddf0..3e24acf8133f 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_dp.h
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_dp.h
@@ -32,7 +32,7 @@ int amdgpu_atombios_dp_get_panel_mode(struct drm_encoder *encoder,
 void amdgpu_atombios_dp_set_link_config(struct drm_connector *connector,
 				 const struct drm_display_mode *mode);
 int amdgpu_atombios_dp_mode_valid_helper(struct drm_connector *connector,
-				  struct drm_display_mode *mode);
+				  const struct drm_display_mode *mode);
 bool amdgpu_atombios_dp_needs_link_train(struct amdgpu_connector *amdgpu_connector);
 void amdgpu_atombios_dp_set_rx_power_state(struct drm_connector *connector,
 				    u8 power_state);
diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c
index 08d6787893b3..9cd63b4177bf 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik.c
@@ -2148,7 +2148,7 @@ static int cik_common_resume(struct amdgpu_ip_block *ip_block)
 	return cik_common_hw_init(ip_block);
 }
 
-static bool cik_common_is_idle(void *handle)
+static bool cik_common_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_ih.c b/drivers/gpu/drm/amd/amdgpu/cik_ih.c
index 444563486769..876a3256dba4 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_ih.c
@@ -156,6 +156,9 @@ static int cik_ih_irq_init(struct amdgpu_device *adev)
 	/* enable irqs */
 	cik_ih_enable_interrupts(adev);
 
+	if (adev->irq.ih_soft.ring_size)
+		adev->irq.ih_soft.enabled = true;
+
 	return 0;
 }
 
@@ -192,6 +195,9 @@ static u32 cik_ih_get_wptr(struct amdgpu_device *adev,
 
 	wptr = le32_to_cpu(*ih->wptr_cpu);
 
+	if (ih == &adev->irq.ih_soft)
+		goto out;
+
 	if (wptr & IH_RB_WPTR__RB_OVERFLOW_MASK) {
 		wptr &= ~IH_RB_WPTR__RB_OVERFLOW_MASK;
 		/* When a ring buffer overflow happen start parsing interrupt
@@ -211,6 +217,8 @@ static u32 cik_ih_get_wptr(struct amdgpu_device *adev,
 		tmp &= ~IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK;
 		WREG32(mmIH_RB_CNTL, tmp);
 	}
+
+out:
 	return (wptr & ih->ptr_mask);
 }
 
@@ -306,6 +314,10 @@ static int cik_ih_sw_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true);
+	if (r)
+		return r;
+
 	r = amdgpu_irq_init(adev);
 
 	return r;
@@ -345,9 +357,9 @@ static int cik_ih_resume(struct amdgpu_ip_block *ip_block)
 	return cik_ih_hw_init(ip_block);
 }
 
-static bool cik_ih_is_idle(void *handle)
+static bool cik_ih_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
index d9bd8f3f17e2..9e8715b4739d 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
@@ -56,6 +56,8 @@ static void cik_sdma_set_buffer_funcs(struct amdgpu_device *adev);
 static void cik_sdma_set_vm_pte_funcs(struct amdgpu_device *adev);
 static int cik_sdma_soft_reset(struct amdgpu_ip_block *ip_block);
 
+u32 amdgpu_cik_gpu_check_soft_reset(struct amdgpu_device *adev);
+
 MODULE_FIRMWARE("amdgpu/bonaire_sdma.bin");
 MODULE_FIRMWARE("amdgpu/bonaire_sdma1.bin");
 MODULE_FIRMWARE("amdgpu/hawaii_sdma.bin");
@@ -67,9 +69,6 @@ MODULE_FIRMWARE("amdgpu/kabini_sdma1.bin");
 MODULE_FIRMWARE("amdgpu/mullins_sdma.bin");
 MODULE_FIRMWARE("amdgpu/mullins_sdma1.bin");
 
-u32 amdgpu_cik_gpu_check_soft_reset(struct amdgpu_device *adev);
-
-
 static void cik_sdma_free_microcode(struct amdgpu_device *adev)
 {
 	int i;
@@ -993,14 +992,9 @@ static int cik_sdma_sw_fini(struct amdgpu_ip_block *ip_block)
 
 static int cik_sdma_hw_init(struct amdgpu_ip_block *ip_block)
 {
-	int r;
 	struct amdgpu_device *adev = ip_block->adev;
 
-	r = cik_sdma_start(adev);
-	if (r)
-		return r;
-
-	return r;
+	return cik_sdma_start(adev);
 }
 
 static int cik_sdma_hw_fini(struct amdgpu_ip_block *ip_block)
@@ -1025,9 +1019,9 @@ static int cik_sdma_resume(struct amdgpu_ip_block *ip_block)
 	return cik_sdma_hw_init(ip_block);
 }
 
-static bool cik_sdma_is_idle(void *handle)
+static bool cik_sdma_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS2);
 
 	if (tmp & (SRBM_STATUS2__SDMA_BUSY_MASK |
@@ -1040,14 +1034,10 @@ static bool cik_sdma_is_idle(void *handle)
 static int cik_sdma_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned i;
-	u32 tmp;
 	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(mmSRBM_STATUS2) & (SRBM_STATUS2__SDMA_BUSY_MASK |
-				SRBM_STATUS2__SDMA1_BUSY_MASK);
-
-		if (!tmp)
+		if (cik_sdma_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/cikd.h b/drivers/gpu/drm/amd/amdgpu/cikd.h
index 06088d52d81c..8aca4f2734f2 100644
--- a/drivers/gpu/drm/amd/amdgpu/cikd.h
+++ b/drivers/gpu/drm/amd/amdgpu/cikd.h
@@ -51,8 +51,14 @@
 #define HPD4_REGISTER_OFFSET                 (0x1813 - 0x1807)
 #define HPD5_REGISTER_OFFSET                 (0x1816 - 0x1807)
 
-#define BONAIRE_GB_ADDR_CONFIG_GOLDEN        0x12010001
-#define HAWAII_GB_ADDR_CONFIG_GOLDEN         0x12011003
+/* audio endpt instance offsets */
+#define AUD0_REGISTER_OFFSET                 (0x1780 - 0x1780)
+#define AUD1_REGISTER_OFFSET                 (0x1786 - 0x1780)
+#define AUD2_REGISTER_OFFSET                 (0x178c - 0x1780)
+#define AUD3_REGISTER_OFFSET                 (0x1792 - 0x1780)
+#define AUD4_REGISTER_OFFSET                 (0x1798 - 0x1780)
+#define AUD5_REGISTER_OFFSET                 (0x179d - 0x1780)
+#define AUD6_REGISTER_OFFSET                 (0x17a4 - 0x1780)
 
 #define		PIPEID(x)					((x) << 0)
 #define		MEID(x)						((x) << 2)
diff --git a/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c b/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c
new file mode 100644
index 000000000000..ed1e25661706
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c
@@ -0,0 +1,56 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "nv.h"
+
+#include "soc15_common.h"
+#include "soc15_hw_ip.h"
+#include "cyan_skillfish_ip_offset.h"
+
+int cyan_skillfish_reg_base_init(struct amdgpu_device *adev)
+{
+	/* HW has more IP blocks,  only initialized the blocke needed by driver */
+	uint32_t i;
+
+	adev->gfx.xcc_mask = 1;
+	for (i = 0 ; i < MAX_INSTANCE ; ++i) {
+		adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i]));
+		adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i]));
+		adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i]));
+		adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
+		adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i]));
+		adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i]));
+		adev->reg_offset[VCN_HWIP][i] = (uint32_t *)(&(UVD0_BASE.instance[i]));
+		adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i]));
+		adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DMU_BASE.instance[i]));
+		adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i]));
+		adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i]));
+		adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i]));
+		adev->reg_offset[CLK_HWIP][i] = (uint32_t *)(&(CLK_BASE.instance[i]));
+	}
+	return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/cz_ih.c b/drivers/gpu/drm/amd/amdgpu/cz_ih.c
index 82586b76aeda..bc7a2e06ab5f 100644
--- a/drivers/gpu/drm/amd/amdgpu/cz_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/cz_ih.c
@@ -157,6 +157,9 @@ static int cz_ih_irq_init(struct amdgpu_device *adev)
 	/* enable interrupts */
 	cz_ih_enable_interrupts(adev);
 
+	if (adev->irq.ih_soft.ring_size)
+		adev->irq.ih_soft.enabled = true;
+
 	return 0;
 }
 
@@ -194,6 +197,9 @@ static u32 cz_ih_get_wptr(struct amdgpu_device *adev,
 
 	wptr = le32_to_cpu(*ih->wptr_cpu);
 
+	if (ih == &adev->irq.ih_soft)
+		goto out;
+
 	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
 		goto out;
 
@@ -297,6 +303,10 @@ static int cz_ih_sw_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true);
+	if (r)
+		return r;
+
 	r = amdgpu_irq_init(adev);
 
 	return r;
@@ -341,9 +351,9 @@ static int cz_ih_resume(struct amdgpu_ip_block *ip_block)
 	return cz_ih_hw_init(ip_block);
 }
 
-static bool cz_ih_is_idle(void *handle)
+static bool cz_ih_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
index c5e3d2251b18..72ca6538b2e4 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
@@ -1141,8 +1141,7 @@ static void dce_v10_0_program_watermarks(struct amdgpu_device *adev,
 
 	/* save values for DPM */
 	amdgpu_crtc->line_time = line_time;
-	amdgpu_crtc->wm_high = latency_watermark_a;
-	amdgpu_crtc->wm_low = latency_watermark_b;
+
 	/* Save number of lines the linebuffer leads before the scanout */
 	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
 }
@@ -1462,17 +1461,12 @@ static int dce_v10_0_audio_init(struct amdgpu_device *adev)
 
 static void dce_v10_0_audio_fini(struct amdgpu_device *adev)
 {
-	int i;
-
 	if (!amdgpu_audio)
 		return;
 
 	if (!adev->mode_info.audio.enabled)
 		return;
 
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
-		dce_v10_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-
 	adev->mode_info.audio.enabled = false;
 }
 
@@ -2970,7 +2964,7 @@ static int dce_v10_0_resume(struct amdgpu_ip_block *ip_block)
 	return amdgpu_display_resume_helper(adev);
 }
 
-static bool dce_v10_0_is_idle(void *handle)
+static bool dce_v10_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
@@ -3075,7 +3069,7 @@ static int dce_v10_0_set_hpd_irq_state(struct amdgpu_device *adev,
 	u32 tmp;
 
 	if (hpd >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", hpd);
+		DRM_DEBUG("invalid hpd %d\n", hpd);
 		return 0;
 	}
 
@@ -3227,7 +3221,7 @@ static void dce_v10_0_hpd_int_ack(struct amdgpu_device *adev,
 	u32 tmp;
 
 	if (hpd >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", hpd);
+		DRM_DEBUG("invalid hpd %d\n", hpd);
 		return;
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
deleted file mode 100644
index ea42a4472bf6..000000000000
--- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
+++ /dev/null
@@ -1,3824 +0,0 @@
-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
- * OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#include <drm/drm_edid.h>
-#include <drm/drm_fourcc.h>
-#include <drm/drm_modeset_helper.h>
-#include <drm/drm_modeset_helper_vtables.h>
-#include <drm/drm_vblank.h>
-
-#include "amdgpu.h"
-#include "amdgpu_pm.h"
-#include "amdgpu_i2c.h"
-#include "vid.h"
-#include "atom.h"
-#include "amdgpu_atombios.h"
-#include "atombios_crtc.h"
-#include "atombios_encoders.h"
-#include "amdgpu_pll.h"
-#include "amdgpu_connectors.h"
-#include "amdgpu_display.h"
-#include "dce_v11_0.h"
-
-#include "dce/dce_11_0_d.h"
-#include "dce/dce_11_0_sh_mask.h"
-#include "dce/dce_11_0_enum.h"
-#include "oss/oss_3_0_d.h"
-#include "oss/oss_3_0_sh_mask.h"
-#include "gmc/gmc_8_1_d.h"
-#include "gmc/gmc_8_1_sh_mask.h"
-
-#include "ivsrcid/ivsrcid_vislands30.h"
-
-static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev);
-static void dce_v11_0_set_irq_funcs(struct amdgpu_device *adev);
-static void dce_v11_0_hpd_int_ack(struct amdgpu_device *adev, int hpd);
-
-static const u32 crtc_offsets[] =
-{
-	CRTC0_REGISTER_OFFSET,
-	CRTC1_REGISTER_OFFSET,
-	CRTC2_REGISTER_OFFSET,
-	CRTC3_REGISTER_OFFSET,
-	CRTC4_REGISTER_OFFSET,
-	CRTC5_REGISTER_OFFSET,
-	CRTC6_REGISTER_OFFSET
-};
-
-static const u32 hpd_offsets[] =
-{
-	HPD0_REGISTER_OFFSET,
-	HPD1_REGISTER_OFFSET,
-	HPD2_REGISTER_OFFSET,
-	HPD3_REGISTER_OFFSET,
-	HPD4_REGISTER_OFFSET,
-	HPD5_REGISTER_OFFSET
-};
-
-static const uint32_t dig_offsets[] = {
-	DIG0_REGISTER_OFFSET,
-	DIG1_REGISTER_OFFSET,
-	DIG2_REGISTER_OFFSET,
-	DIG3_REGISTER_OFFSET,
-	DIG4_REGISTER_OFFSET,
-	DIG5_REGISTER_OFFSET,
-	DIG6_REGISTER_OFFSET,
-	DIG7_REGISTER_OFFSET,
-	DIG8_REGISTER_OFFSET
-};
-
-static const struct {
-	uint32_t        reg;
-	uint32_t        vblank;
-	uint32_t        vline;
-	uint32_t        hpd;
-
-} interrupt_status_offsets[] = { {
-	.reg = mmDISP_INTERRUPT_STATUS,
-	.vblank = DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK
-}, {
-	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE,
-	.vblank = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK
-}, {
-	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE2,
-	.vblank = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK
-}, {
-	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE3,
-	.vblank = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK
-}, {
-	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE4,
-	.vblank = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK
-}, {
-	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE5,
-	.vblank = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK,
-	.vline = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK,
-	.hpd = DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK
-} };
-
-static const u32 cz_golden_settings_a11[] =
-{
-	mmCRTC_DOUBLE_BUFFER_CONTROL, 0x00010101, 0x00010000,
-	mmFBC_MISC, 0x1f311fff, 0x14300000,
-};
-
-static const u32 cz_mgcg_cgcg_init[] =
-{
-	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
-	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
-};
-
-static const u32 stoney_golden_settings_a11[] =
-{
-	mmCRTC_DOUBLE_BUFFER_CONTROL, 0x00010101, 0x00010000,
-	mmFBC_MISC, 0x1f311fff, 0x14302000,
-};
-
-static const u32 polaris11_golden_settings_a11[] =
-{
-	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
-	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
-	mmFBC_DEBUG1, 0xffffffff, 0x00000008,
-	mmFBC_MISC, 0x9f313fff, 0x14302008,
-	mmHDMI_CONTROL, 0x313f031f, 0x00000011,
-};
-
-static const u32 polaris10_golden_settings_a11[] =
-{
-	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
-	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
-	mmFBC_MISC, 0x9f313fff, 0x14302008,
-	mmHDMI_CONTROL, 0x313f031f, 0x00000011,
-};
-
-static void dce_v11_0_init_golden_registers(struct amdgpu_device *adev)
-{
-	switch (adev->asic_type) {
-	case CHIP_CARRIZO:
-		amdgpu_device_program_register_sequence(adev,
-							cz_mgcg_cgcg_init,
-							ARRAY_SIZE(cz_mgcg_cgcg_init));
-		amdgpu_device_program_register_sequence(adev,
-							cz_golden_settings_a11,
-							ARRAY_SIZE(cz_golden_settings_a11));
-		break;
-	case CHIP_STONEY:
-		amdgpu_device_program_register_sequence(adev,
-							stoney_golden_settings_a11,
-							ARRAY_SIZE(stoney_golden_settings_a11));
-		break;
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS12:
-		amdgpu_device_program_register_sequence(adev,
-							polaris11_golden_settings_a11,
-							ARRAY_SIZE(polaris11_golden_settings_a11));
-		break;
-	case CHIP_POLARIS10:
-	case CHIP_VEGAM:
-		amdgpu_device_program_register_sequence(adev,
-							polaris10_golden_settings_a11,
-							ARRAY_SIZE(polaris10_golden_settings_a11));
-		break;
-	default:
-		break;
-	}
-}
-
-static u32 dce_v11_0_audio_endpt_rreg(struct amdgpu_device *adev,
-				     u32 block_offset, u32 reg)
-{
-	unsigned long flags;
-	u32 r;
-
-	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
-	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
-	r = RREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset);
-	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
-
-	return r;
-}
-
-static void dce_v11_0_audio_endpt_wreg(struct amdgpu_device *adev,
-				      u32 block_offset, u32 reg, u32 v)
-{
-	unsigned long flags;
-
-	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
-	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
-	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset, v);
-	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
-}
-
-static u32 dce_v11_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
-{
-	if (crtc < 0 || crtc >= adev->mode_info.num_crtc)
-		return 0;
-	else
-		return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]);
-}
-
-static void dce_v11_0_pageflip_interrupt_init(struct amdgpu_device *adev)
-{
-	unsigned i;
-
-	/* Enable pflip interrupts */
-	for (i = 0; i < adev->mode_info.num_crtc; i++)
-		amdgpu_irq_get(adev, &adev->pageflip_irq, i);
-}
-
-static void dce_v11_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
-{
-	unsigned i;
-
-	/* Disable pflip interrupts */
-	for (i = 0; i < adev->mode_info.num_crtc; i++)
-		amdgpu_irq_put(adev, &adev->pageflip_irq, i);
-}
-
-/**
- * dce_v11_0_page_flip - pageflip callback.
- *
- * @adev: amdgpu_device pointer
- * @crtc_id: crtc to cleanup pageflip on
- * @crtc_base: new address of the crtc (GPU MC address)
- * @async: asynchronous flip
- *
- * Triggers the actual pageflip by updating the primary
- * surface base address.
- */
-static void dce_v11_0_page_flip(struct amdgpu_device *adev,
-				int crtc_id, u64 crtc_base, bool async)
-{
-	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
-	struct drm_framebuffer *fb = amdgpu_crtc->base.primary->fb;
-	u32 tmp;
-
-	/* flip immediate for async, default is vsync */
-	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
-			    GRPH_SURFACE_UPDATE_IMMEDIATE_EN, async ? 1 : 0);
-	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-	/* update pitch */
-	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset,
-	       fb->pitches[0] / fb->format->cpp[0]);
-	/* update the scanout addresses */
-	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
-	       upper_32_bits(crtc_base));
-	/* writing to the low address triggers the update */
-	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
-	       lower_32_bits(crtc_base));
-	/* post the write */
-	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
-}
-
-static int dce_v11_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
-					u32 *vbl, u32 *position)
-{
-	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
-		return -EINVAL;
-
-	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
-	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
-
-	return 0;
-}
-
-/**
- * dce_v11_0_hpd_sense - hpd sense callback.
- *
- * @adev: amdgpu_device pointer
- * @hpd: hpd (hotplug detect) pin
- *
- * Checks if a digital monitor is connected (evergreen+).
- * Returns true if connected, false if not connected.
- */
-static bool dce_v11_0_hpd_sense(struct amdgpu_device *adev,
-			       enum amdgpu_hpd_id hpd)
-{
-	bool connected = false;
-
-	if (hpd >= adev->mode_info.num_hpd)
-		return connected;
-
-	if (RREG32(mmDC_HPD_INT_STATUS + hpd_offsets[hpd]) &
-	    DC_HPD_INT_STATUS__DC_HPD_SENSE_MASK)
-		connected = true;
-
-	return connected;
-}
-
-/**
- * dce_v11_0_hpd_set_polarity - hpd set polarity callback.
- *
- * @adev: amdgpu_device pointer
- * @hpd: hpd (hotplug detect) pin
- *
- * Set the polarity of the hpd pin (evergreen+).
- */
-static void dce_v11_0_hpd_set_polarity(struct amdgpu_device *adev,
-				      enum amdgpu_hpd_id hpd)
-{
-	u32 tmp;
-	bool connected = dce_v11_0_hpd_sense(adev, hpd);
-
-	if (hpd >= adev->mode_info.num_hpd)
-		return;
-
-	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
-	if (connected)
-		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 0);
-	else
-		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 1);
-	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
-}
-
-/**
- * dce_v11_0_hpd_init - hpd setup callback.
- *
- * @adev: amdgpu_device pointer
- *
- * Setup the hpd pins used by the card (evergreen+).
- * Enable the pin, set the polarity, and enable the hpd interrupts.
- */
-static void dce_v11_0_hpd_init(struct amdgpu_device *adev)
-{
-	struct drm_device *dev = adev_to_drm(adev);
-	struct drm_connector *connector;
-	struct drm_connector_list_iter iter;
-	u32 tmp;
-
-	drm_connector_list_iter_begin(dev, &iter);
-	drm_for_each_connector_iter(connector, &iter) {
-		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-
-		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
-			continue;
-
-		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
-		    connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
-			/* don't try to enable hpd on eDP or LVDS avoid breaking the
-			 * aux dp channel on imac and help (but not completely fix)
-			 * https://bugzilla.redhat.com/show_bug.cgi?id=726143
-			 * also avoid interrupt storms during dpms.
-			 */
-			tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
-			tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 0);
-			WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
-			continue;
-		}
-
-		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 1);
-		WREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
-
-		tmp = RREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd]);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
-				    DC_HPD_CONNECT_INT_DELAY,
-				    AMDGPU_HPD_CONNECT_INT_DELAY_IN_MS);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
-				    DC_HPD_DISCONNECT_INT_DELAY,
-				    AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS);
-		WREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
-
-		dce_v11_0_hpd_int_ack(adev, amdgpu_connector->hpd.hpd);
-		dce_v11_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
-		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
-	}
-	drm_connector_list_iter_end(&iter);
-}
-
-/**
- * dce_v11_0_hpd_fini - hpd tear down callback.
- *
- * @adev: amdgpu_device pointer
- *
- * Tear down the hpd pins used by the card (evergreen+).
- * Disable the hpd interrupts.
- */
-static void dce_v11_0_hpd_fini(struct amdgpu_device *adev)
-{
-	struct drm_device *dev = adev_to_drm(adev);
-	struct drm_connector *connector;
-	struct drm_connector_list_iter iter;
-	u32 tmp;
-
-	drm_connector_list_iter_begin(dev, &iter);
-	drm_for_each_connector_iter(connector, &iter) {
-		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-
-		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
-			continue;
-
-		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 0);
-		WREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
-
-		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
-	}
-	drm_connector_list_iter_end(&iter);
-}
-
-static u32 dce_v11_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
-{
-	return mmDC_GPIO_HPD_A;
-}
-
-static bool dce_v11_0_is_display_hung(struct amdgpu_device *adev)
-{
-	u32 crtc_hung = 0;
-	u32 crtc_status[6];
-	u32 i, j, tmp;
-
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
-		if (REG_GET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN)) {
-			crtc_status[i] = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
-			crtc_hung |= (1 << i);
-		}
-	}
-
-	for (j = 0; j < 10; j++) {
-		for (i = 0; i < adev->mode_info.num_crtc; i++) {
-			if (crtc_hung & (1 << i)) {
-				tmp = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
-				if (tmp != crtc_status[i])
-					crtc_hung &= ~(1 << i);
-			}
-		}
-		if (crtc_hung == 0)
-			return false;
-		udelay(100);
-	}
-
-	return true;
-}
-
-static void dce_v11_0_set_vga_render_state(struct amdgpu_device *adev,
-					   bool render)
-{
-	u32 tmp;
-
-	/* Lockout access through VGA aperture*/
-	tmp = RREG32(mmVGA_HDP_CONTROL);
-	if (render)
-		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 0);
-	else
-		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
-	WREG32(mmVGA_HDP_CONTROL, tmp);
-
-	/* disable VGA render */
-	tmp = RREG32(mmVGA_RENDER_CONTROL);
-	if (render)
-		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 1);
-	else
-		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
-	WREG32(mmVGA_RENDER_CONTROL, tmp);
-}
-
-static int dce_v11_0_get_num_crtc (struct amdgpu_device *adev)
-{
-	int num_crtc = 0;
-
-	switch (adev->asic_type) {
-	case CHIP_CARRIZO:
-		num_crtc = 3;
-		break;
-	case CHIP_STONEY:
-		num_crtc = 2;
-		break;
-	case CHIP_POLARIS10:
-	case CHIP_VEGAM:
-		num_crtc = 6;
-		break;
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS12:
-		num_crtc = 5;
-		break;
-	default:
-		num_crtc = 0;
-	}
-	return num_crtc;
-}
-
-void dce_v11_0_disable_dce(struct amdgpu_device *adev)
-{
-	/*Disable VGA render and enabled crtc, if has DCE engine*/
-	if (amdgpu_atombios_has_dce_engine_info(adev)) {
-		u32 tmp;
-		int crtc_enabled, i;
-
-		dce_v11_0_set_vga_render_state(adev, false);
-
-		/*Disable crtc*/
-		for (i = 0; i < dce_v11_0_get_num_crtc(adev); i++) {
-			crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
-									 CRTC_CONTROL, CRTC_MASTER_EN);
-			if (crtc_enabled) {
-				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
-				tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
-				tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
-				WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
-				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
-			}
-		}
-	}
-}
-
-static void dce_v11_0_program_fmt(struct drm_encoder *encoder)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
-	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
-	int bpc = 0;
-	u32 tmp = 0;
-	enum amdgpu_connector_dither dither = AMDGPU_FMT_DITHER_DISABLE;
-
-	if (connector) {
-		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-		bpc = amdgpu_connector_get_monitor_bpc(connector);
-		dither = amdgpu_connector->dither;
-	}
-
-	/* LVDS/eDP FMT is set up by atom */
-	if (amdgpu_encoder->devices & ATOM_DEVICE_LCD_SUPPORT)
-		return;
-
-	/* not needed for analog */
-	if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1) ||
-	    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2))
-		return;
-
-	if (bpc == 0)
-		return;
-
-	switch (bpc) {
-	case 6:
-		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
-			/* XXX sort out optimal dither settings */
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 0);
-		} else {
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 0);
-		}
-		break;
-	case 8:
-		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
-			/* XXX sort out optimal dither settings */
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 1);
-		} else {
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 1);
-		}
-		break;
-	case 10:
-		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
-			/* XXX sort out optimal dither settings */
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 2);
-		} else {
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
-			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 2);
-		}
-		break;
-	default:
-		/* not needed */
-		break;
-	}
-
-	WREG32(mmFMT_BIT_DEPTH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-}
-
-
-/* display watermark setup */
-/**
- * dce_v11_0_line_buffer_adjust - Set up the line buffer
- *
- * @adev: amdgpu_device pointer
- * @amdgpu_crtc: the selected display controller
- * @mode: the current display mode on the selected display
- * controller
- *
- * Setup up the line buffer allocation for
- * the selected display controller (CIK).
- * Returns the line buffer size in pixels.
- */
-static u32 dce_v11_0_line_buffer_adjust(struct amdgpu_device *adev,
-				       struct amdgpu_crtc *amdgpu_crtc,
-				       struct drm_display_mode *mode)
-{
-	u32 tmp, buffer_alloc, i, mem_cfg;
-	u32 pipe_offset = amdgpu_crtc->crtc_id;
-	/*
-	 * Line Buffer Setup
-	 * There are 6 line buffers, one for each display controllers.
-	 * There are 3 partitions per LB. Select the number of partitions
-	 * to enable based on the display width.  For display widths larger
-	 * than 4096, you need use to use 2 display controllers and combine
-	 * them using the stereo blender.
-	 */
-	if (amdgpu_crtc->base.enabled && mode) {
-		if (mode->crtc_hdisplay < 1920) {
-			mem_cfg = 1;
-			buffer_alloc = 2;
-		} else if (mode->crtc_hdisplay < 2560) {
-			mem_cfg = 2;
-			buffer_alloc = 2;
-		} else if (mode->crtc_hdisplay < 4096) {
-			mem_cfg = 0;
-			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
-		} else {
-			DRM_DEBUG_KMS("Mode too big for LB!\n");
-			mem_cfg = 0;
-			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
-		}
-	} else {
-		mem_cfg = 1;
-		buffer_alloc = 0;
-	}
-
-	tmp = RREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, LB_MEMORY_CTRL, LB_MEMORY_CONFIG, mem_cfg);
-	WREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
-	tmp = REG_SET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATED, buffer_alloc);
-	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset, tmp);
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
-		if (REG_GET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATION_COMPLETED))
-			break;
-		udelay(1);
-	}
-
-	if (amdgpu_crtc->base.enabled && mode) {
-		switch (mem_cfg) {
-		case 0:
-		default:
-			return 4096 * 2;
-		case 1:
-			return 1920 * 2;
-		case 2:
-			return 2560 * 2;
-		}
-	}
-
-	/* controller not enabled, so no lb used */
-	return 0;
-}
-
-/**
- * cik_get_number_of_dram_channels - get the number of dram channels
- *
- * @adev: amdgpu_device pointer
- *
- * Look up the number of video ram channels (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the number of dram channels
- */
-static u32 cik_get_number_of_dram_channels(struct amdgpu_device *adev)
-{
-	u32 tmp = RREG32(mmMC_SHARED_CHMAP);
-
-	switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
-	case 0:
-	default:
-		return 1;
-	case 1:
-		return 2;
-	case 2:
-		return 4;
-	case 3:
-		return 8;
-	case 4:
-		return 3;
-	case 5:
-		return 6;
-	case 6:
-		return 10;
-	case 7:
-		return 12;
-	case 8:
-		return 16;
-	}
-}
-
-struct dce10_wm_params {
-	u32 dram_channels; /* number of dram channels */
-	u32 yclk;          /* bandwidth per dram data pin in kHz */
-	u32 sclk;          /* engine clock in kHz */
-	u32 disp_clk;      /* display clock in kHz */
-	u32 src_width;     /* viewport width */
-	u32 active_time;   /* active display time in ns */
-	u32 blank_time;    /* blank time in ns */
-	bool interlaced;    /* mode is interlaced */
-	fixed20_12 vsc;    /* vertical scale ratio */
-	u32 num_heads;     /* number of active crtcs */
-	u32 bytes_per_pixel; /* bytes per pixel display + overlay */
-	u32 lb_size;       /* line buffer allocated to pipe */
-	u32 vtaps;         /* vertical scaler taps */
-};
-
-/**
- * dce_v11_0_dram_bandwidth - get the dram bandwidth
- *
- * @wm: watermark calculation data
- *
- * Calculate the raw dram bandwidth (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the dram bandwidth in MBytes/s
- */
-static u32 dce_v11_0_dram_bandwidth(struct dce10_wm_params *wm)
-{
-	/* Calculate raw DRAM Bandwidth */
-	fixed20_12 dram_efficiency; /* 0.7 */
-	fixed20_12 yclk, dram_channels, bandwidth;
-	fixed20_12 a;
-
-	a.full = dfixed_const(1000);
-	yclk.full = dfixed_const(wm->yclk);
-	yclk.full = dfixed_div(yclk, a);
-	dram_channels.full = dfixed_const(wm->dram_channels * 4);
-	a.full = dfixed_const(10);
-	dram_efficiency.full = dfixed_const(7);
-	dram_efficiency.full = dfixed_div(dram_efficiency, a);
-	bandwidth.full = dfixed_mul(dram_channels, yclk);
-	bandwidth.full = dfixed_mul(bandwidth, dram_efficiency);
-
-	return dfixed_trunc(bandwidth);
-}
-
-/**
- * dce_v11_0_dram_bandwidth_for_display - get the dram bandwidth for display
- *
- * @wm: watermark calculation data
- *
- * Calculate the dram bandwidth used for display (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the dram bandwidth for display in MBytes/s
- */
-static u32 dce_v11_0_dram_bandwidth_for_display(struct dce10_wm_params *wm)
-{
-	/* Calculate DRAM Bandwidth and the part allocated to display. */
-	fixed20_12 disp_dram_allocation; /* 0.3 to 0.7 */
-	fixed20_12 yclk, dram_channels, bandwidth;
-	fixed20_12 a;
-
-	a.full = dfixed_const(1000);
-	yclk.full = dfixed_const(wm->yclk);
-	yclk.full = dfixed_div(yclk, a);
-	dram_channels.full = dfixed_const(wm->dram_channels * 4);
-	a.full = dfixed_const(10);
-	disp_dram_allocation.full = dfixed_const(3); /* XXX worse case value 0.3 */
-	disp_dram_allocation.full = dfixed_div(disp_dram_allocation, a);
-	bandwidth.full = dfixed_mul(dram_channels, yclk);
-	bandwidth.full = dfixed_mul(bandwidth, disp_dram_allocation);
-
-	return dfixed_trunc(bandwidth);
-}
-
-/**
- * dce_v11_0_data_return_bandwidth - get the data return bandwidth
- *
- * @wm: watermark calculation data
- *
- * Calculate the data return bandwidth used for display (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the data return bandwidth in MBytes/s
- */
-static u32 dce_v11_0_data_return_bandwidth(struct dce10_wm_params *wm)
-{
-	/* Calculate the display Data return Bandwidth */
-	fixed20_12 return_efficiency; /* 0.8 */
-	fixed20_12 sclk, bandwidth;
-	fixed20_12 a;
-
-	a.full = dfixed_const(1000);
-	sclk.full = dfixed_const(wm->sclk);
-	sclk.full = dfixed_div(sclk, a);
-	a.full = dfixed_const(10);
-	return_efficiency.full = dfixed_const(8);
-	return_efficiency.full = dfixed_div(return_efficiency, a);
-	a.full = dfixed_const(32);
-	bandwidth.full = dfixed_mul(a, sclk);
-	bandwidth.full = dfixed_mul(bandwidth, return_efficiency);
-
-	return dfixed_trunc(bandwidth);
-}
-
-/**
- * dce_v11_0_dmif_request_bandwidth - get the dmif bandwidth
- *
- * @wm: watermark calculation data
- *
- * Calculate the dmif bandwidth used for display (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the dmif bandwidth in MBytes/s
- */
-static u32 dce_v11_0_dmif_request_bandwidth(struct dce10_wm_params *wm)
-{
-	/* Calculate the DMIF Request Bandwidth */
-	fixed20_12 disp_clk_request_efficiency; /* 0.8 */
-	fixed20_12 disp_clk, bandwidth;
-	fixed20_12 a, b;
-
-	a.full = dfixed_const(1000);
-	disp_clk.full = dfixed_const(wm->disp_clk);
-	disp_clk.full = dfixed_div(disp_clk, a);
-	a.full = dfixed_const(32);
-	b.full = dfixed_mul(a, disp_clk);
-
-	a.full = dfixed_const(10);
-	disp_clk_request_efficiency.full = dfixed_const(8);
-	disp_clk_request_efficiency.full = dfixed_div(disp_clk_request_efficiency, a);
-
-	bandwidth.full = dfixed_mul(b, disp_clk_request_efficiency);
-
-	return dfixed_trunc(bandwidth);
-}
-
-/**
- * dce_v11_0_available_bandwidth - get the min available bandwidth
- *
- * @wm: watermark calculation data
- *
- * Calculate the min available bandwidth used for display (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the min available bandwidth in MBytes/s
- */
-static u32 dce_v11_0_available_bandwidth(struct dce10_wm_params *wm)
-{
-	/* Calculate the Available bandwidth. Display can use this temporarily but not in average. */
-	u32 dram_bandwidth = dce_v11_0_dram_bandwidth(wm);
-	u32 data_return_bandwidth = dce_v11_0_data_return_bandwidth(wm);
-	u32 dmif_req_bandwidth = dce_v11_0_dmif_request_bandwidth(wm);
-
-	return min(dram_bandwidth, min(data_return_bandwidth, dmif_req_bandwidth));
-}
-
-/**
- * dce_v11_0_average_bandwidth - get the average available bandwidth
- *
- * @wm: watermark calculation data
- *
- * Calculate the average available bandwidth used for display (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the average available bandwidth in MBytes/s
- */
-static u32 dce_v11_0_average_bandwidth(struct dce10_wm_params *wm)
-{
-	/* Calculate the display mode Average Bandwidth
-	 * DisplayMode should contain the source and destination dimensions,
-	 * timing, etc.
-	 */
-	fixed20_12 bpp;
-	fixed20_12 line_time;
-	fixed20_12 src_width;
-	fixed20_12 bandwidth;
-	fixed20_12 a;
-
-	a.full = dfixed_const(1000);
-	line_time.full = dfixed_const(wm->active_time + wm->blank_time);
-	line_time.full = dfixed_div(line_time, a);
-	bpp.full = dfixed_const(wm->bytes_per_pixel);
-	src_width.full = dfixed_const(wm->src_width);
-	bandwidth.full = dfixed_mul(src_width, bpp);
-	bandwidth.full = dfixed_mul(bandwidth, wm->vsc);
-	bandwidth.full = dfixed_div(bandwidth, line_time);
-
-	return dfixed_trunc(bandwidth);
-}
-
-/**
- * dce_v11_0_latency_watermark - get the latency watermark
- *
- * @wm: watermark calculation data
- *
- * Calculate the latency watermark (CIK).
- * Used for display watermark bandwidth calculations
- * Returns the latency watermark in ns
- */
-static u32 dce_v11_0_latency_watermark(struct dce10_wm_params *wm)
-{
-	/* First calculate the latency in ns */
-	u32 mc_latency = 2000; /* 2000 ns. */
-	u32 available_bandwidth = dce_v11_0_available_bandwidth(wm);
-	u32 worst_chunk_return_time = (512 * 8 * 1000) / available_bandwidth;
-	u32 cursor_line_pair_return_time = (128 * 4 * 1000) / available_bandwidth;
-	u32 dc_latency = 40000000 / wm->disp_clk; /* dc pipe latency */
-	u32 other_heads_data_return_time = ((wm->num_heads + 1) * worst_chunk_return_time) +
-		(wm->num_heads * cursor_line_pair_return_time);
-	u32 latency = mc_latency + other_heads_data_return_time + dc_latency;
-	u32 max_src_lines_per_dst_line, lb_fill_bw, line_fill_time;
-	u32 tmp, dmif_size = 12288;
-	fixed20_12 a, b, c;
-
-	if (wm->num_heads == 0)
-		return 0;
-
-	a.full = dfixed_const(2);
-	b.full = dfixed_const(1);
-	if ((wm->vsc.full > a.full) ||
-	    ((wm->vsc.full > b.full) && (wm->vtaps >= 3)) ||
-	    (wm->vtaps >= 5) ||
-	    ((wm->vsc.full >= a.full) && wm->interlaced))
-		max_src_lines_per_dst_line = 4;
-	else
-		max_src_lines_per_dst_line = 2;
-
-	a.full = dfixed_const(available_bandwidth);
-	b.full = dfixed_const(wm->num_heads);
-	a.full = dfixed_div(a, b);
-	tmp = div_u64((u64) dmif_size * (u64) wm->disp_clk, mc_latency + 512);
-	tmp = min(dfixed_trunc(a), tmp);
-
-	lb_fill_bw = min(tmp, wm->disp_clk * wm->bytes_per_pixel / 1000);
-
-	a.full = dfixed_const(max_src_lines_per_dst_line * wm->src_width * wm->bytes_per_pixel);
-	b.full = dfixed_const(1000);
-	c.full = dfixed_const(lb_fill_bw);
-	b.full = dfixed_div(c, b);
-	a.full = dfixed_div(a, b);
-	line_fill_time = dfixed_trunc(a);
-
-	if (line_fill_time < wm->active_time)
-		return latency;
-	else
-		return latency + (line_fill_time - wm->active_time);
-
-}
-
-/**
- * dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display - check
- * average and available dram bandwidth
- *
- * @wm: watermark calculation data
- *
- * Check if the display average bandwidth fits in the display
- * dram bandwidth (CIK).
- * Used for display watermark bandwidth calculations
- * Returns true if the display fits, false if not.
- */
-static bool dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(struct dce10_wm_params *wm)
-{
-	if (dce_v11_0_average_bandwidth(wm) <=
-	    (dce_v11_0_dram_bandwidth_for_display(wm) / wm->num_heads))
-		return true;
-	else
-		return false;
-}
-
-/**
- * dce_v11_0_average_bandwidth_vs_available_bandwidth - check
- * average and available bandwidth
- *
- * @wm: watermark calculation data
- *
- * Check if the display average bandwidth fits in the display
- * available bandwidth (CIK).
- * Used for display watermark bandwidth calculations
- * Returns true if the display fits, false if not.
- */
-static bool dce_v11_0_average_bandwidth_vs_available_bandwidth(struct dce10_wm_params *wm)
-{
-	if (dce_v11_0_average_bandwidth(wm) <=
-	    (dce_v11_0_available_bandwidth(wm) / wm->num_heads))
-		return true;
-	else
-		return false;
-}
-
-/**
- * dce_v11_0_check_latency_hiding - check latency hiding
- *
- * @wm: watermark calculation data
- *
- * Check latency hiding (CIK).
- * Used for display watermark bandwidth calculations
- * Returns true if the display fits, false if not.
- */
-static bool dce_v11_0_check_latency_hiding(struct dce10_wm_params *wm)
-{
-	u32 lb_partitions = wm->lb_size / wm->src_width;
-	u32 line_time = wm->active_time + wm->blank_time;
-	u32 latency_tolerant_lines;
-	u32 latency_hiding;
-	fixed20_12 a;
-
-	a.full = dfixed_const(1);
-	if (wm->vsc.full > a.full)
-		latency_tolerant_lines = 1;
-	else {
-		if (lb_partitions <= (wm->vtaps + 1))
-			latency_tolerant_lines = 1;
-		else
-			latency_tolerant_lines = 2;
-	}
-
-	latency_hiding = (latency_tolerant_lines * line_time + wm->blank_time);
-
-	if (dce_v11_0_latency_watermark(wm) <= latency_hiding)
-		return true;
-	else
-		return false;
-}
-
-/**
- * dce_v11_0_program_watermarks - program display watermarks
- *
- * @adev: amdgpu_device pointer
- * @amdgpu_crtc: the selected display controller
- * @lb_size: line buffer size
- * @num_heads: number of display controllers in use
- *
- * Calculate and program the display watermarks for the
- * selected display controller (CIK).
- */
-static void dce_v11_0_program_watermarks(struct amdgpu_device *adev,
-					struct amdgpu_crtc *amdgpu_crtc,
-					u32 lb_size, u32 num_heads)
-{
-	struct drm_display_mode *mode = &amdgpu_crtc->base.mode;
-	struct dce10_wm_params wm_low, wm_high;
-	u32 active_time;
-	u32 line_time = 0;
-	u32 latency_watermark_a = 0, latency_watermark_b = 0;
-	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
-
-	if (amdgpu_crtc->base.enabled && num_heads && mode) {
-		active_time = (u32) div_u64((u64)mode->crtc_hdisplay * 1000000,
-					    (u32)mode->clock);
-		line_time = (u32) div_u64((u64)mode->crtc_htotal * 1000000,
-					  (u32)mode->clock);
-		line_time = min_t(u32, line_time, 65535);
-
-		/* watermark for high clocks */
-		if (adev->pm.dpm_enabled) {
-			wm_high.yclk =
-				amdgpu_dpm_get_mclk(adev, false) * 10;
-			wm_high.sclk =
-				amdgpu_dpm_get_sclk(adev, false) * 10;
-		} else {
-			wm_high.yclk = adev->pm.current_mclk * 10;
-			wm_high.sclk = adev->pm.current_sclk * 10;
-		}
-
-		wm_high.disp_clk = mode->clock;
-		wm_high.src_width = mode->crtc_hdisplay;
-		wm_high.active_time = active_time;
-		wm_high.blank_time = line_time - wm_high.active_time;
-		wm_high.interlaced = false;
-		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
-			wm_high.interlaced = true;
-		wm_high.vsc = amdgpu_crtc->vsc;
-		wm_high.vtaps = 1;
-		if (amdgpu_crtc->rmx_type != RMX_OFF)
-			wm_high.vtaps = 2;
-		wm_high.bytes_per_pixel = 4; /* XXX: get this from fb config */
-		wm_high.lb_size = lb_size;
-		wm_high.dram_channels = cik_get_number_of_dram_channels(adev);
-		wm_high.num_heads = num_heads;
-
-		/* set for high clocks */
-		latency_watermark_a = min_t(u32, dce_v11_0_latency_watermark(&wm_high), 65535);
-
-		/* possibly force display priority to high */
-		/* should really do this at mode validation time... */
-		if (!dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_high) ||
-		    !dce_v11_0_average_bandwidth_vs_available_bandwidth(&wm_high) ||
-		    !dce_v11_0_check_latency_hiding(&wm_high) ||
-		    (adev->mode_info.disp_priority == 2)) {
-			DRM_DEBUG_KMS("force priority to high\n");
-		}
-
-		/* watermark for low clocks */
-		if (adev->pm.dpm_enabled) {
-			wm_low.yclk =
-				amdgpu_dpm_get_mclk(adev, true) * 10;
-			wm_low.sclk =
-				amdgpu_dpm_get_sclk(adev, true) * 10;
-		} else {
-			wm_low.yclk = adev->pm.current_mclk * 10;
-			wm_low.sclk = adev->pm.current_sclk * 10;
-		}
-
-		wm_low.disp_clk = mode->clock;
-		wm_low.src_width = mode->crtc_hdisplay;
-		wm_low.active_time = active_time;
-		wm_low.blank_time = line_time - wm_low.active_time;
-		wm_low.interlaced = false;
-		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
-			wm_low.interlaced = true;
-		wm_low.vsc = amdgpu_crtc->vsc;
-		wm_low.vtaps = 1;
-		if (amdgpu_crtc->rmx_type != RMX_OFF)
-			wm_low.vtaps = 2;
-		wm_low.bytes_per_pixel = 4; /* XXX: get this from fb config */
-		wm_low.lb_size = lb_size;
-		wm_low.dram_channels = cik_get_number_of_dram_channels(adev);
-		wm_low.num_heads = num_heads;
-
-		/* set for low clocks */
-		latency_watermark_b = min_t(u32, dce_v11_0_latency_watermark(&wm_low), 65535);
-
-		/* possibly force display priority to high */
-		/* should really do this at mode validation time... */
-		if (!dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_low) ||
-		    !dce_v11_0_average_bandwidth_vs_available_bandwidth(&wm_low) ||
-		    !dce_v11_0_check_latency_hiding(&wm_low) ||
-		    (adev->mode_info.disp_priority == 2)) {
-			DRM_DEBUG_KMS("force priority to high\n");
-		}
-		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
-	}
-
-	/* select wm A */
-	wm_mask = RREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 1);
-	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_a);
-	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
-	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-	/* select wm B */
-	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 2);
-	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_b);
-	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
-	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-	/* restore original selection */
-	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, wm_mask);
-
-	/* save values for DPM */
-	amdgpu_crtc->line_time = line_time;
-	amdgpu_crtc->wm_high = latency_watermark_a;
-	amdgpu_crtc->wm_low = latency_watermark_b;
-	/* Save number of lines the linebuffer leads before the scanout */
-	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
-}
-
-/**
- * dce_v11_0_bandwidth_update - program display watermarks
- *
- * @adev: amdgpu_device pointer
- *
- * Calculate and program the display watermarks and line
- * buffer allocation (CIK).
- */
-static void dce_v11_0_bandwidth_update(struct amdgpu_device *adev)
-{
-	struct drm_display_mode *mode = NULL;
-	u32 num_heads = 0, lb_size;
-	int i;
-
-	amdgpu_display_update_priority(adev);
-
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		if (adev->mode_info.crtcs[i]->base.enabled)
-			num_heads++;
-	}
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		mode = &adev->mode_info.crtcs[i]->base.mode;
-		lb_size = dce_v11_0_line_buffer_adjust(adev, adev->mode_info.crtcs[i], mode);
-		dce_v11_0_program_watermarks(adev, adev->mode_info.crtcs[i],
-					    lb_size, num_heads);
-	}
-}
-
-static void dce_v11_0_audio_get_connected_pins(struct amdgpu_device *adev)
-{
-	int i;
-	u32 offset, tmp;
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
-		offset = adev->mode_info.audio.pin[i].offset;
-		tmp = RREG32_AUDIO_ENDPT(offset,
-					 ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT);
-		if (((tmp &
-		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY_MASK) >>
-		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY__SHIFT) == 1)
-			adev->mode_info.audio.pin[i].connected = false;
-		else
-			adev->mode_info.audio.pin[i].connected = true;
-	}
-}
-
-static struct amdgpu_audio_pin *dce_v11_0_audio_get_pin(struct amdgpu_device *adev)
-{
-	int i;
-
-	dce_v11_0_audio_get_connected_pins(adev);
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
-		if (adev->mode_info.audio.pin[i].connected)
-			return &adev->mode_info.audio.pin[i];
-	}
-	DRM_ERROR("No connected audio pins found!\n");
-	return NULL;
-}
-
-static void dce_v11_0_afmt_audio_select_pin(struct drm_encoder *encoder)
-{
-	struct amdgpu_device *adev = drm_to_adev(encoder->dev);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	u32 tmp;
-
-	if (!dig || !dig->afmt || !dig->afmt->pin)
-		return;
-
-	tmp = RREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, dig->afmt->pin->id);
-	WREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset, tmp);
-}
-
-static void dce_v11_0_audio_write_latency_fields(struct drm_encoder *encoder,
-						struct drm_display_mode *mode)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	struct drm_connector *connector;
-	struct drm_connector_list_iter iter;
-	struct amdgpu_connector *amdgpu_connector = NULL;
-	u32 tmp;
-	int interlace = 0;
-
-	if (!dig || !dig->afmt || !dig->afmt->pin)
-		return;
-
-	drm_connector_list_iter_begin(dev, &iter);
-	drm_for_each_connector_iter(connector, &iter) {
-		if (connector->encoder == encoder) {
-			amdgpu_connector = to_amdgpu_connector(connector);
-			break;
-		}
-	}
-	drm_connector_list_iter_end(&iter);
-
-	if (!amdgpu_connector) {
-		DRM_ERROR("Couldn't find encoder's connector\n");
-		return;
-	}
-
-	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
-		interlace = 1;
-	if (connector->latency_present[interlace]) {
-		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
-				    VIDEO_LIPSYNC, connector->video_latency[interlace]);
-		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
-				    AUDIO_LIPSYNC, connector->audio_latency[interlace]);
-	} else {
-		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
-				    VIDEO_LIPSYNC, 0);
-		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
-				    AUDIO_LIPSYNC, 0);
-	}
-	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
-			   ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC, tmp);
-}
-
-static void dce_v11_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	struct drm_connector *connector;
-	struct drm_connector_list_iter iter;
-	struct amdgpu_connector *amdgpu_connector = NULL;
-	u32 tmp;
-	u8 *sadb = NULL;
-	int sad_count;
-
-	if (!dig || !dig->afmt || !dig->afmt->pin)
-		return;
-
-	drm_connector_list_iter_begin(dev, &iter);
-	drm_for_each_connector_iter(connector, &iter) {
-		if (connector->encoder == encoder) {
-			amdgpu_connector = to_amdgpu_connector(connector);
-			break;
-		}
-	}
-	drm_connector_list_iter_end(&iter);
-
-	if (!amdgpu_connector) {
-		DRM_ERROR("Couldn't find encoder's connector\n");
-		return;
-	}
-
-	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector->edid, &sadb);
-	if (sad_count < 0) {
-		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
-		sad_count = 0;
-	}
-
-	/* program the speaker allocation */
-	tmp = RREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
-				 ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER);
-	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
-			    DP_CONNECTION, 0);
-	/* set HDMI mode */
-	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
-			    HDMI_CONNECTION, 1);
-	if (sad_count)
-		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
-				    SPEAKER_ALLOCATION, sadb[0]);
-	else
-		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
-				    SPEAKER_ALLOCATION, 5); /* stereo */
-	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
-			   ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER, tmp);
-
-	kfree(sadb);
-}
-
-static void dce_v11_0_audio_write_sad_regs(struct drm_encoder *encoder)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	struct drm_connector *connector;
-	struct drm_connector_list_iter iter;
-	struct amdgpu_connector *amdgpu_connector = NULL;
-	struct cea_sad *sads;
-	int i, sad_count;
-
-	static const u16 eld_reg_to_type[][2] = {
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0, HDMI_AUDIO_CODING_TYPE_PCM },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1, HDMI_AUDIO_CODING_TYPE_AC3 },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2, HDMI_AUDIO_CODING_TYPE_MPEG1 },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3, HDMI_AUDIO_CODING_TYPE_MP3 },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4, HDMI_AUDIO_CODING_TYPE_MPEG2 },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5, HDMI_AUDIO_CODING_TYPE_AAC_LC },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6, HDMI_AUDIO_CODING_TYPE_DTS },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7, HDMI_AUDIO_CODING_TYPE_ATRAC },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9, HDMI_AUDIO_CODING_TYPE_EAC3 },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10, HDMI_AUDIO_CODING_TYPE_DTS_HD },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11, HDMI_AUDIO_CODING_TYPE_MLP },
-		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
-	};
-
-	if (!dig || !dig->afmt || !dig->afmt->pin)
-		return;
-
-	drm_connector_list_iter_begin(dev, &iter);
-	drm_for_each_connector_iter(connector, &iter) {
-		if (connector->encoder == encoder) {
-			amdgpu_connector = to_amdgpu_connector(connector);
-			break;
-		}
-	}
-	drm_connector_list_iter_end(&iter);
-
-	if (!amdgpu_connector) {
-		DRM_ERROR("Couldn't find encoder's connector\n");
-		return;
-	}
-
-	sad_count = drm_edid_to_sad(amdgpu_connector->edid, &sads);
-	if (sad_count < 0)
-		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
-	if (sad_count <= 0)
-		return;
-	BUG_ON(!sads);
-
-	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
-		u32 tmp = 0;
-		u8 stereo_freqs = 0;
-		int max_channels = -1;
-		int j;
-
-		for (j = 0; j < sad_count; j++) {
-			struct cea_sad *sad = &sads[j];
-
-			if (sad->format == eld_reg_to_type[i][1]) {
-				if (sad->channels > max_channels) {
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							    MAX_CHANNELS, sad->channels);
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							    DESCRIPTOR_BYTE_2, sad->byte2);
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							    SUPPORTED_FREQUENCIES, sad->freq);
-					max_channels = sad->channels;
-				}
-
-				if (sad->format == HDMI_AUDIO_CODING_TYPE_PCM)
-					stereo_freqs |= sad->freq;
-				else
-					break;
-			}
-		}
-
-		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-				    SUPPORTED_FREQUENCIES_STEREO, stereo_freqs);
-		WREG32_AUDIO_ENDPT(dig->afmt->pin->offset, eld_reg_to_type[i][0], tmp);
-	}
-
-	kfree(sads);
-}
-
-static void dce_v11_0_audio_enable(struct amdgpu_device *adev,
-				  struct amdgpu_audio_pin *pin,
-				  bool enable)
-{
-	if (!pin)
-		return;
-
-	WREG32_AUDIO_ENDPT(pin->offset, ixAZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
-			   enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
-}
-
-static const u32 pin_offsets[] =
-{
-	AUD0_REGISTER_OFFSET,
-	AUD1_REGISTER_OFFSET,
-	AUD2_REGISTER_OFFSET,
-	AUD3_REGISTER_OFFSET,
-	AUD4_REGISTER_OFFSET,
-	AUD5_REGISTER_OFFSET,
-	AUD6_REGISTER_OFFSET,
-	AUD7_REGISTER_OFFSET,
-};
-
-static int dce_v11_0_audio_init(struct amdgpu_device *adev)
-{
-	int i;
-
-	if (!amdgpu_audio)
-		return 0;
-
-	adev->mode_info.audio.enabled = true;
-
-	switch (adev->asic_type) {
-	case CHIP_CARRIZO:
-	case CHIP_STONEY:
-		adev->mode_info.audio.num_pins = 7;
-		break;
-	case CHIP_POLARIS10:
-	case CHIP_VEGAM:
-		adev->mode_info.audio.num_pins = 8;
-		break;
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS12:
-		adev->mode_info.audio.num_pins = 6;
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
-		adev->mode_info.audio.pin[i].channels = -1;
-		adev->mode_info.audio.pin[i].rate = -1;
-		adev->mode_info.audio.pin[i].bits_per_sample = -1;
-		adev->mode_info.audio.pin[i].status_bits = 0;
-		adev->mode_info.audio.pin[i].category_code = 0;
-		adev->mode_info.audio.pin[i].connected = false;
-		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
-		adev->mode_info.audio.pin[i].id = i;
-		/* disable audio.  it will be set up later */
-		/* XXX remove once we switch to ip funcs */
-		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-	}
-
-	return 0;
-}
-
-static void dce_v11_0_audio_fini(struct amdgpu_device *adev)
-{
-	int i;
-
-	if (!amdgpu_audio)
-		return;
-
-	if (!adev->mode_info.audio.enabled)
-		return;
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
-		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-
-	adev->mode_info.audio.enabled = false;
-}
-
-/*
- * update the N and CTS parameters for a given pixel clock rate
- */
-static void dce_v11_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	u32 tmp;
-
-	tmp = RREG32(mmHDMI_ACR_32_0 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_0, HDMI_ACR_CTS_32, acr.cts_32khz);
-	WREG32(mmHDMI_ACR_32_0 + dig->afmt->offset, tmp);
-	tmp = RREG32(mmHDMI_ACR_32_1 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_1, HDMI_ACR_N_32, acr.n_32khz);
-	WREG32(mmHDMI_ACR_32_1 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_ACR_44_0 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_0, HDMI_ACR_CTS_44, acr.cts_44_1khz);
-	WREG32(mmHDMI_ACR_44_0 + dig->afmt->offset, tmp);
-	tmp = RREG32(mmHDMI_ACR_44_1 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_1, HDMI_ACR_N_44, acr.n_44_1khz);
-	WREG32(mmHDMI_ACR_44_1 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_ACR_48_0 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_0, HDMI_ACR_CTS_48, acr.cts_48khz);
-	WREG32(mmHDMI_ACR_48_0 + dig->afmt->offset, tmp);
-	tmp = RREG32(mmHDMI_ACR_48_1 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_1, HDMI_ACR_N_48, acr.n_48khz);
-	WREG32(mmHDMI_ACR_48_1 + dig->afmt->offset, tmp);
-
-}
-
-/*
- * build a HDMI Video Info Frame
- */
-static void dce_v11_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
-					       void *buffer, size_t size)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	uint8_t *frame = buffer + 3;
-	uint8_t *header = buffer;
-
-	WREG32(mmAFMT_AVI_INFO0 + dig->afmt->offset,
-		frame[0x0] | (frame[0x1] << 8) | (frame[0x2] << 16) | (frame[0x3] << 24));
-	WREG32(mmAFMT_AVI_INFO1 + dig->afmt->offset,
-		frame[0x4] | (frame[0x5] << 8) | (frame[0x6] << 16) | (frame[0x7] << 24));
-	WREG32(mmAFMT_AVI_INFO2 + dig->afmt->offset,
-		frame[0x8] | (frame[0x9] << 8) | (frame[0xA] << 16) | (frame[0xB] << 24));
-	WREG32(mmAFMT_AVI_INFO3 + dig->afmt->offset,
-		frame[0xC] | (frame[0xD] << 8) | (header[1] << 24));
-}
-
-static void dce_v11_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
-	u32 dto_phase = 24 * 1000;
-	u32 dto_modulo = clock;
-	u32 tmp;
-
-	if (!dig || !dig->afmt)
-		return;
-
-	/* XXX two dtos; generally use dto0 for hdmi */
-	/* Express [24MHz / target pixel clock] as an exact rational
-	 * number (coefficient of two integer numbers.  DCCG_AUDIO_DTOx_PHASE
-	 * is the numerator, DCCG_AUDIO_DTOx_MODULE is the denominator
-	 */
-	tmp = RREG32(mmDCCG_AUDIO_DTO_SOURCE);
-	tmp = REG_SET_FIELD(tmp, DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL,
-			    amdgpu_crtc->crtc_id);
-	WREG32(mmDCCG_AUDIO_DTO_SOURCE, tmp);
-	WREG32(mmDCCG_AUDIO_DTO0_PHASE, dto_phase);
-	WREG32(mmDCCG_AUDIO_DTO0_MODULE, dto_modulo);
-}
-
-/*
- * update the info frames with the data from the current display mode
- */
-static void dce_v11_0_afmt_setmode(struct drm_encoder *encoder,
-				  struct drm_display_mode *mode)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
-	u8 buffer[HDMI_INFOFRAME_HEADER_SIZE + HDMI_AVI_INFOFRAME_SIZE];
-	struct hdmi_avi_infoframe frame;
-	ssize_t err;
-	u32 tmp;
-	int bpc = 8;
-
-	if (!dig || !dig->afmt)
-		return;
-
-	/* Silent, r600_hdmi_enable will raise WARN for us */
-	if (!dig->afmt->enabled)
-		return;
-
-	/* hdmi deep color mode general control packets setup, if bpc > 8 */
-	if (encoder->crtc) {
-		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
-		bpc = amdgpu_crtc->bpc;
-	}
-
-	/* disable audio prior to setting up hw */
-	dig->afmt->pin = dce_v11_0_audio_get_pin(adev);
-	dce_v11_0_audio_enable(adev, dig->afmt->pin, false);
-
-	dce_v11_0_audio_set_dto(encoder, mode->clock);
-
-	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1);
-	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp); /* send null packets when required */
-
-	WREG32(mmAFMT_AUDIO_CRC_CONTROL + dig->afmt->offset, 0x1000);
-
-	tmp = RREG32(mmHDMI_CONTROL + dig->afmt->offset);
-	switch (bpc) {
-	case 0:
-	case 6:
-	case 8:
-	case 16:
-	default:
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 0);
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
-		DRM_DEBUG("%s: Disabling hdmi deep color for %d bpc.\n",
-			  connector->name, bpc);
-		break;
-	case 10:
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 1);
-		DRM_DEBUG("%s: Enabling hdmi deep color 30 for 10 bpc.\n",
-			  connector->name);
-		break;
-	case 12:
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
-		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 2);
-		DRM_DEBUG("%s: Enabling hdmi deep color 36 for 12 bpc.\n",
-			  connector->name);
-		break;
-	}
-	WREG32(mmHDMI_CONTROL + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1); /* send null packets when required */
-	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, 1); /* send general control packets */
-	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, 1); /* send general control packets every frame */
-	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
-	/* enable audio info frames (frames won't be set until audio is enabled) */
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
-	/* required for audio info values to be updated */
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_CONT, 1);
-	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset);
-	/* required for audio info values to be updated */
-	tmp = REG_SET_FIELD(tmp, AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
-	WREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
-	/* anything other than 0 */
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, 2);
-	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
-
-	WREG32(mmHDMI_GC + dig->afmt->offset, 0); /* unset HDMI_GC_AVMUTE */
-
-	tmp = RREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset);
-	/* set the default audio delay */
-	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, 1);
-	/* should be suffient for all audio modes and small enough for all hblanks */
-	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_PACKETS_PER_LINE, 3);
-	WREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
-	/* allow 60958 channel status fields to be updated */
-	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
-	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset);
-	if (bpc > 8)
-		/* clear SW CTS value */
-		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 0);
-	else
-		/* select SW CTS value */
-		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 1);
-	/* allow hw to sent ACR packets when required */
-	tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, 1);
-	WREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset, tmp);
-
-	dce_v11_0_afmt_update_ACR(encoder, mode->clock);
-
-	tmp = RREG32(mmAFMT_60958_0 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, 1);
-	WREG32(mmAFMT_60958_0 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmAFMT_60958_1 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, 2);
-	WREG32(mmAFMT_60958_1 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmAFMT_60958_2 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, 3);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, 4);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, 5);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, 6);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, 7);
-	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, 8);
-	WREG32(mmAFMT_60958_2 + dig->afmt->offset, tmp);
-
-	dce_v11_0_audio_write_speaker_allocation(encoder);
-
-	WREG32(mmAFMT_AUDIO_PACKET_CONTROL2 + dig->afmt->offset,
-	       (0xff << AFMT_AUDIO_PACKET_CONTROL2__AFMT_AUDIO_CHANNEL_ENABLE__SHIFT));
-
-	dce_v11_0_afmt_audio_select_pin(encoder);
-	dce_v11_0_audio_write_sad_regs(encoder);
-	dce_v11_0_audio_write_latency_fields(encoder, mode);
-
-	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, connector, mode);
-	if (err < 0) {
-		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
-		return;
-	}
-
-	err = hdmi_avi_infoframe_pack(&frame, buffer, sizeof(buffer));
-	if (err < 0) {
-		DRM_ERROR("failed to pack AVI infoframe: %zd\n", err);
-		return;
-	}
-
-	dce_v11_0_afmt_update_avi_infoframe(encoder, buffer, sizeof(buffer));
-
-	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
-	/* enable AVI info frames */
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_SEND, 1);
-	/* required for audio info values to be updated */
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_CONT, 1);
-	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
-	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AVI_INFO_LINE, 2);
-	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
-
-	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
-	/* send audio packets */
-	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, 1);
-	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
-
-	WREG32(mmAFMT_RAMP_CONTROL0 + dig->afmt->offset, 0x00FFFFFF);
-	WREG32(mmAFMT_RAMP_CONTROL1 + dig->afmt->offset, 0x007FFFFF);
-	WREG32(mmAFMT_RAMP_CONTROL2 + dig->afmt->offset, 0x00000001);
-	WREG32(mmAFMT_RAMP_CONTROL3 + dig->afmt->offset, 0x00000001);
-
-	/* enable audio after to setting up hw */
-	dce_v11_0_audio_enable(adev, dig->afmt->pin, true);
-}
-
-static void dce_v11_0_afmt_enable(struct drm_encoder *encoder, bool enable)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-
-	if (!dig || !dig->afmt)
-		return;
-
-	/* Silent, r600_hdmi_enable will raise WARN for us */
-	if (enable && dig->afmt->enabled)
-		return;
-	if (!enable && !dig->afmt->enabled)
-		return;
-
-	if (!enable && dig->afmt->pin) {
-		dce_v11_0_audio_enable(adev, dig->afmt->pin, false);
-		dig->afmt->pin = NULL;
-	}
-
-	dig->afmt->enabled = enable;
-
-	DRM_DEBUG("%sabling AFMT interface @ 0x%04X for encoder 0x%x\n",
-		  enable ? "En" : "Dis", dig->afmt->offset, amdgpu_encoder->encoder_id);
-}
-
-static int dce_v11_0_afmt_init(struct amdgpu_device *adev)
-{
-	int i;
-
-	for (i = 0; i < adev->mode_info.num_dig; i++)
-		adev->mode_info.afmt[i] = NULL;
-
-	/* DCE11 has audio blocks tied to DIG encoders */
-	for (i = 0; i < adev->mode_info.num_dig; i++) {
-		adev->mode_info.afmt[i] = kzalloc(sizeof(struct amdgpu_afmt), GFP_KERNEL);
-		if (adev->mode_info.afmt[i]) {
-			adev->mode_info.afmt[i]->offset = dig_offsets[i];
-			adev->mode_info.afmt[i]->id = i;
-		} else {
-			int j;
-			for (j = 0; j < i; j++) {
-				kfree(adev->mode_info.afmt[j]);
-				adev->mode_info.afmt[j] = NULL;
-			}
-			return -ENOMEM;
-		}
-	}
-	return 0;
-}
-
-static void dce_v11_0_afmt_fini(struct amdgpu_device *adev)
-{
-	int i;
-
-	for (i = 0; i < adev->mode_info.num_dig; i++) {
-		kfree(adev->mode_info.afmt[i]);
-		adev->mode_info.afmt[i] = NULL;
-	}
-}
-
-static const u32 vga_control_regs[6] =
-{
-	mmD1VGA_CONTROL,
-	mmD2VGA_CONTROL,
-	mmD3VGA_CONTROL,
-	mmD4VGA_CONTROL,
-	mmD5VGA_CONTROL,
-	mmD6VGA_CONTROL,
-};
-
-static void dce_v11_0_vga_enable(struct drm_crtc *crtc, bool enable)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	u32 vga_control;
-
-	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
-	if (enable)
-		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control | 1);
-	else
-		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control);
-}
-
-static void dce_v11_0_grph_enable(struct drm_crtc *crtc, bool enable)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-
-	if (enable)
-		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 1);
-	else
-		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 0);
-}
-
-static int dce_v11_0_crtc_do_set_base(struct drm_crtc *crtc,
-				     struct drm_framebuffer *fb,
-				     int x, int y, int atomic)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct drm_framebuffer *target_fb;
-	struct drm_gem_object *obj;
-	struct amdgpu_bo *abo;
-	uint64_t fb_location, tiling_flags;
-	uint32_t fb_format, fb_pitch_pixels;
-	u32 fb_swap = REG_SET_FIELD(0, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP, ENDIAN_NONE);
-	u32 pipe_config;
-	u32 tmp, viewport_w, viewport_h;
-	int r;
-	bool bypass_lut = false;
-
-	/* no fb bound */
-	if (!atomic && !crtc->primary->fb) {
-		DRM_DEBUG_KMS("No FB bound\n");
-		return 0;
-	}
-
-	if (atomic)
-		target_fb = fb;
-	else
-		target_fb = crtc->primary->fb;
-
-	/* If atomic, assume fb object is pinned & idle & fenced and
-	 * just update base pointers
-	 */
-	obj = target_fb->obj[0];
-	abo = gem_to_amdgpu_bo(obj);
-	r = amdgpu_bo_reserve(abo, false);
-	if (unlikely(r != 0))
-		return r;
-
-	if (!atomic) {
-		abo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
-		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM);
-		if (unlikely(r != 0)) {
-			amdgpu_bo_unreserve(abo);
-			return -EINVAL;
-		}
-	}
-	fb_location = amdgpu_bo_gpu_offset(abo);
-
-	amdgpu_bo_get_tiling_flags(abo, &tiling_flags);
-	amdgpu_bo_unreserve(abo);
-
-	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
-
-	switch (target_fb->format->format) {
-	case DRM_FORMAT_C8:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 0);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
-		break;
-	case DRM_FORMAT_XRGB4444:
-	case DRM_FORMAT_ARGB4444:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 2);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN16);
-#endif
-		break;
-	case DRM_FORMAT_XRGB1555:
-	case DRM_FORMAT_ARGB1555:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN16);
-#endif
-		break;
-	case DRM_FORMAT_BGRX5551:
-	case DRM_FORMAT_BGRA5551:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 5);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN16);
-#endif
-		break;
-	case DRM_FORMAT_RGB565:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN16);
-#endif
-		break;
-	case DRM_FORMAT_XRGB8888:
-	case DRM_FORMAT_ARGB8888:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN32);
-#endif
-		break;
-	case DRM_FORMAT_XRGB2101010:
-	case DRM_FORMAT_ARGB2101010:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN32);
-#endif
-		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
-		bypass_lut = true;
-		break;
-	case DRM_FORMAT_BGRX1010102:
-	case DRM_FORMAT_BGRA1010102:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 4);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN32);
-#endif
-		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
-		bypass_lut = true;
-		break;
-	case DRM_FORMAT_XBGR8888:
-	case DRM_FORMAT_ABGR8888:
-		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_RED_CROSSBAR, 2);
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_BLUE_CROSSBAR, 2);
-#ifdef __BIG_ENDIAN
-		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
-					ENDIAN_8IN32);
-#endif
-		break;
-	default:
-		DRM_ERROR("Unsupported screen format %p4cc\n",
-			  &target_fb->format->format);
-		return -EINVAL;
-	}
-
-	if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_2D_TILED_THIN1) {
-		unsigned bankw, bankh, mtaspect, tile_split, num_banks;
-
-		bankw = AMDGPU_TILING_GET(tiling_flags, BANK_WIDTH);
-		bankh = AMDGPU_TILING_GET(tiling_flags, BANK_HEIGHT);
-		mtaspect = AMDGPU_TILING_GET(tiling_flags, MACRO_TILE_ASPECT);
-		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
-		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
-
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_NUM_BANKS, num_banks);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
-					  ARRAY_2D_TILED_THIN1);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_TILE_SPLIT,
-					  tile_split);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_WIDTH, bankw);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_HEIGHT, bankh);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MACRO_TILE_ASPECT,
-					  mtaspect);
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MICRO_TILE_MODE,
-					  ADDR_SURF_MICRO_TILING_DISPLAY);
-	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
-		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
-					  ARRAY_1D_TILED_THIN1);
-	}
-
-	fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_PIPE_CONFIG,
-				  pipe_config);
-
-	dce_v11_0_vga_enable(crtc, false);
-
-	/* Make sure surface address is updated at vertical blank rather than
-	 * horizontal blank
-	 */
-	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
-			    GRPH_SURFACE_UPDATE_H_RETRACE_EN, 0);
-	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
-	       upper_32_bits(fb_location));
-	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
-	       upper_32_bits(fb_location));
-	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
-	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
-	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
-	       (u32) fb_location & GRPH_SECONDARY_SURFACE_ADDRESS__GRPH_SECONDARY_SURFACE_ADDRESS_MASK);
-	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
-	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
-
-	/*
-	 * The LUT only has 256 slots for indexing by a 8 bpc fb. Bypass the LUT
-	 * for > 8 bpc scanout to avoid truncation of fb indices to 8 msb's, to
-	 * retain the full precision throughout the pipeline.
-	 */
-	tmp = RREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset);
-	if (bypass_lut)
-		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 1);
-	else
-		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 0);
-	WREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset, tmp);
-
-	if (bypass_lut)
-		DRM_DEBUG_KMS("Bypassing hardware LUT due to 10 bit fb scanout.\n");
-
-	WREG32(mmGRPH_SURFACE_OFFSET_X + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmGRPH_SURFACE_OFFSET_Y + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmGRPH_X_START + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmGRPH_Y_START + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmGRPH_X_END + amdgpu_crtc->crtc_offset, target_fb->width);
-	WREG32(mmGRPH_Y_END + amdgpu_crtc->crtc_offset, target_fb->height);
-
-	fb_pitch_pixels = target_fb->pitches[0] / target_fb->format->cpp[0];
-	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset, fb_pitch_pixels);
-
-	dce_v11_0_grph_enable(crtc, true);
-
-	WREG32(mmLB_DESKTOP_HEIGHT + amdgpu_crtc->crtc_offset,
-	       target_fb->height);
-
-	x &= ~3;
-	y &= ~1;
-	WREG32(mmVIEWPORT_START + amdgpu_crtc->crtc_offset,
-	       (x << 16) | y);
-	viewport_w = crtc->mode.hdisplay;
-	viewport_h = (crtc->mode.vdisplay + 1) & ~1;
-	WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset,
-	       (viewport_w << 16) | viewport_h);
-
-	/* set pageflip to happen anywhere in vblank interval */
-	WREG32(mmCRTC_MASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0);
-
-	if (!atomic && fb && fb != crtc->primary->fb) {
-		abo = gem_to_amdgpu_bo(fb->obj[0]);
-		r = amdgpu_bo_reserve(abo, true);
-		if (unlikely(r != 0))
-			return r;
-		amdgpu_bo_unpin(abo);
-		amdgpu_bo_unreserve(abo);
-	}
-
-	/* Bytes per pixel may have changed */
-	dce_v11_0_bandwidth_update(adev);
-
-	return 0;
-}
-
-static void dce_v11_0_set_interleave(struct drm_crtc *crtc,
-				     struct drm_display_mode *mode)
-{
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	u32 tmp;
-
-	tmp = RREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset);
-	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
-		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 1);
-	else
-		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 0);
-	WREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset, tmp);
-}
-
-static void dce_v11_0_crtc_load_lut(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	u16 *r, *g, *b;
-	int i;
-	u32 tmp;
-
-	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
-
-	tmp = RREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, INPUT_CSC_CONTROL, INPUT_CSC_GRPH_MODE, 0);
-	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_BYPASS, 1);
-	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, INPUT_GAMMA_CONTROL, GRPH_INPUT_GAMMA_MODE, 0);
-	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
-
-	WREG32(mmDC_LUT_BLACK_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmDC_LUT_BLACK_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmDC_LUT_BLACK_OFFSET_RED + amdgpu_crtc->crtc_offset, 0);
-
-	WREG32(mmDC_LUT_WHITE_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0xffff);
-	WREG32(mmDC_LUT_WHITE_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0xffff);
-	WREG32(mmDC_LUT_WHITE_OFFSET_RED + amdgpu_crtc->crtc_offset, 0xffff);
-
-	WREG32(mmDC_LUT_RW_MODE + amdgpu_crtc->crtc_offset, 0);
-	WREG32(mmDC_LUT_WRITE_EN_MASK + amdgpu_crtc->crtc_offset, 0x00000007);
-
-	WREG32(mmDC_LUT_RW_INDEX + amdgpu_crtc->crtc_offset, 0);
-	r = crtc->gamma_store;
-	g = r + crtc->gamma_size;
-	b = g + crtc->gamma_size;
-	for (i = 0; i < 256; i++) {
-		WREG32(mmDC_LUT_30_COLOR + amdgpu_crtc->crtc_offset,
-		       ((*r++ & 0xffc0) << 14) |
-		       ((*g++ & 0xffc0) << 4) |
-		       (*b++ >> 6));
-	}
-
-	tmp = RREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, GRPH_DEGAMMA_MODE, 0);
-	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, CURSOR_DEGAMMA_MODE, 0);
-	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, CURSOR2_DEGAMMA_MODE, 0);
-	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, GAMUT_REMAP_CONTROL, GRPH_GAMUT_REMAP_MODE, 0);
-	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, REGAMMA_CONTROL, GRPH_REGAMMA_MODE, 0);
-	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	tmp = RREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, OUTPUT_CSC_CONTROL, OUTPUT_CSC_GRPH_MODE, 0);
-	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-
-	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
-	WREG32(mmDENORM_CONTROL + amdgpu_crtc->crtc_offset, 0);
-	/* XXX this only needs to be programmed once per crtc at startup,
-	 * not sure where the best place for it is
-	 */
-	tmp = RREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, ALPHA_CONTROL, CURSOR_ALPHA_BLND_ENA, 1);
-	WREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-}
-
-static int dce_v11_0_pick_dig_encoder(struct drm_encoder *encoder)
-{
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-
-	switch (amdgpu_encoder->encoder_id) {
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
-		if (dig->linkb)
-			return 1;
-		else
-			return 0;
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
-		if (dig->linkb)
-			return 3;
-		else
-			return 2;
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
-		if (dig->linkb)
-			return 5;
-		else
-			return 4;
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
-		return 6;
-	default:
-		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
-		return 0;
-	}
-}
-
-/**
- * dce_v11_0_pick_pll - Allocate a PPLL for use by the crtc.
- *
- * @crtc: drm crtc
- *
- * Returns the PPLL (Pixel PLL) to be used by the crtc.  For DP monitors
- * a single PPLL can be used for all DP crtcs/encoders.  For non-DP
- * monitors a dedicated PPLL must be used.  If a particular board has
- * an external DP PLL, return ATOM_PPLL_INVALID to skip PLL programming
- * as there is no need to program the PLL itself.  If we are not able to
- * allocate a PLL, return ATOM_PPLL_INVALID to skip PLL programming to
- * avoid messing up an existing monitor.
- *
- * Asic specific PLL information
- *
- * DCE 10.x
- * Tonga
- * - PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP)
- * CI
- * - PPLL0, PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP) and DAC
- *
- */
-static u32 dce_v11_0_pick_pll(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	u32 pll_in_use;
-	int pll;
-
-	if ((adev->asic_type == CHIP_POLARIS10) ||
-	    (adev->asic_type == CHIP_POLARIS11) ||
-	    (adev->asic_type == CHIP_POLARIS12) ||
-	    (adev->asic_type == CHIP_VEGAM)) {
-		struct amdgpu_encoder *amdgpu_encoder =
-			to_amdgpu_encoder(amdgpu_crtc->encoder);
-		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-
-		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
-			return ATOM_DP_DTO;
-
-		switch (amdgpu_encoder->encoder_id) {
-		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
-			if (dig->linkb)
-				return ATOM_COMBOPHY_PLL1;
-			else
-				return ATOM_COMBOPHY_PLL0;
-		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
-			if (dig->linkb)
-				return ATOM_COMBOPHY_PLL3;
-			else
-				return ATOM_COMBOPHY_PLL2;
-		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
-			if (dig->linkb)
-				return ATOM_COMBOPHY_PLL5;
-			else
-				return ATOM_COMBOPHY_PLL4;
-		default:
-			DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
-			return ATOM_PPLL_INVALID;
-		}
-	}
-
-	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder))) {
-		if (adev->clock.dp_extclk)
-			/* skip PPLL programming if using ext clock */
-			return ATOM_PPLL_INVALID;
-		else {
-			/* use the same PPLL for all DP monitors */
-			pll = amdgpu_pll_get_shared_dp_ppll(crtc);
-			if (pll != ATOM_PPLL_INVALID)
-				return pll;
-		}
-	} else {
-		/* use the same PPLL for all monitors with the same clock */
-		pll = amdgpu_pll_get_shared_nondp_ppll(crtc);
-		if (pll != ATOM_PPLL_INVALID)
-			return pll;
-	}
-
-	/* XXX need to determine what plls are available on each DCE11 part */
-	pll_in_use = amdgpu_pll_get_use_mask(crtc);
-	if (adev->flags & AMD_IS_APU) {
-		if (!(pll_in_use & (1 << ATOM_PPLL1)))
-			return ATOM_PPLL1;
-		if (!(pll_in_use & (1 << ATOM_PPLL0)))
-			return ATOM_PPLL0;
-		DRM_ERROR("unable to allocate a PPLL\n");
-		return ATOM_PPLL_INVALID;
-	} else {
-		if (!(pll_in_use & (1 << ATOM_PPLL2)))
-			return ATOM_PPLL2;
-		if (!(pll_in_use & (1 << ATOM_PPLL1)))
-			return ATOM_PPLL1;
-		if (!(pll_in_use & (1 << ATOM_PPLL0)))
-			return ATOM_PPLL0;
-		DRM_ERROR("unable to allocate a PPLL\n");
-		return ATOM_PPLL_INVALID;
-	}
-	return ATOM_PPLL_INVALID;
-}
-
-static void dce_v11_0_lock_cursor(struct drm_crtc *crtc, bool lock)
-{
-	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	uint32_t cur_lock;
-
-	cur_lock = RREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset);
-	if (lock)
-		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 1);
-	else
-		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 0);
-	WREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset, cur_lock);
-}
-
-static void dce_v11_0_hide_cursor(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
-	u32 tmp;
-
-	tmp = RREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 0);
-	WREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-}
-
-static void dce_v11_0_show_cursor(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
-	u32 tmp;
-
-	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
-	       upper_32_bits(amdgpu_crtc->cursor_addr));
-	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
-	       lower_32_bits(amdgpu_crtc->cursor_addr));
-
-	tmp = RREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
-	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 1);
-	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_MODE, 2);
-	WREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
-}
-
-static int dce_v11_0_cursor_move_locked(struct drm_crtc *crtc,
-					int x, int y)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct amdgpu_device *adev = drm_to_adev(crtc->dev);
-	int xorigin = 0, yorigin = 0;
-
-	amdgpu_crtc->cursor_x = x;
-	amdgpu_crtc->cursor_y = y;
-
-	/* avivo cursor are offset into the total surface */
-	x += crtc->x;
-	y += crtc->y;
-	DRM_DEBUG("x %d y %d c->x %d c->y %d\n", x, y, crtc->x, crtc->y);
-
-	if (x < 0) {
-		xorigin = min(-x, amdgpu_crtc->max_cursor_width - 1);
-		x = 0;
-	}
-	if (y < 0) {
-		yorigin = min(-y, amdgpu_crtc->max_cursor_height - 1);
-		y = 0;
-	}
-
-	WREG32(mmCUR_POSITION + amdgpu_crtc->crtc_offset, (x << 16) | y);
-	WREG32(mmCUR_HOT_SPOT + amdgpu_crtc->crtc_offset, (xorigin << 16) | yorigin);
-	WREG32(mmCUR_SIZE + amdgpu_crtc->crtc_offset,
-	       ((amdgpu_crtc->cursor_width - 1) << 16) | (amdgpu_crtc->cursor_height - 1));
-
-	return 0;
-}
-
-static int dce_v11_0_crtc_cursor_move(struct drm_crtc *crtc,
-				      int x, int y)
-{
-	int ret;
-
-	dce_v11_0_lock_cursor(crtc, true);
-	ret = dce_v11_0_cursor_move_locked(crtc, x, y);
-	dce_v11_0_lock_cursor(crtc, false);
-
-	return ret;
-}
-
-static int dce_v11_0_crtc_cursor_set2(struct drm_crtc *crtc,
-				      struct drm_file *file_priv,
-				      uint32_t handle,
-				      uint32_t width,
-				      uint32_t height,
-				      int32_t hot_x,
-				      int32_t hot_y)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_gem_object *obj;
-	struct amdgpu_bo *aobj;
-	int ret;
-
-	if (!handle) {
-		/* turn off cursor */
-		dce_v11_0_hide_cursor(crtc);
-		obj = NULL;
-		goto unpin;
-	}
-
-	if ((width > amdgpu_crtc->max_cursor_width) ||
-	    (height > amdgpu_crtc->max_cursor_height)) {
-		DRM_ERROR("bad cursor width or height %d x %d\n", width, height);
-		return -EINVAL;
-	}
-
-	obj = drm_gem_object_lookup(file_priv, handle);
-	if (!obj) {
-		DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, amdgpu_crtc->crtc_id);
-		return -ENOENT;
-	}
-
-	aobj = gem_to_amdgpu_bo(obj);
-	ret = amdgpu_bo_reserve(aobj, false);
-	if (ret != 0) {
-		drm_gem_object_put(obj);
-		return ret;
-	}
-
-	aobj->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
-	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM);
-	amdgpu_bo_unreserve(aobj);
-	if (ret) {
-		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
-		drm_gem_object_put(obj);
-		return ret;
-	}
-	amdgpu_crtc->cursor_addr = amdgpu_bo_gpu_offset(aobj);
-
-	dce_v11_0_lock_cursor(crtc, true);
-
-	if (width != amdgpu_crtc->cursor_width ||
-	    height != amdgpu_crtc->cursor_height ||
-	    hot_x != amdgpu_crtc->cursor_hot_x ||
-	    hot_y != amdgpu_crtc->cursor_hot_y) {
-		int x, y;
-
-		x = amdgpu_crtc->cursor_x + amdgpu_crtc->cursor_hot_x - hot_x;
-		y = amdgpu_crtc->cursor_y + amdgpu_crtc->cursor_hot_y - hot_y;
-
-		dce_v11_0_cursor_move_locked(crtc, x, y);
-
-		amdgpu_crtc->cursor_width = width;
-		amdgpu_crtc->cursor_height = height;
-		amdgpu_crtc->cursor_hot_x = hot_x;
-		amdgpu_crtc->cursor_hot_y = hot_y;
-	}
-
-	dce_v11_0_show_cursor(crtc);
-	dce_v11_0_lock_cursor(crtc, false);
-
-unpin:
-	if (amdgpu_crtc->cursor_bo) {
-		struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
-		ret = amdgpu_bo_reserve(aobj, true);
-		if (likely(ret == 0)) {
-			amdgpu_bo_unpin(aobj);
-			amdgpu_bo_unreserve(aobj);
-		}
-		drm_gem_object_put(amdgpu_crtc->cursor_bo);
-	}
-
-	amdgpu_crtc->cursor_bo = obj;
-	return 0;
-}
-
-static void dce_v11_0_cursor_reset(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-
-	if (amdgpu_crtc->cursor_bo) {
-		dce_v11_0_lock_cursor(crtc, true);
-
-		dce_v11_0_cursor_move_locked(crtc, amdgpu_crtc->cursor_x,
-					     amdgpu_crtc->cursor_y);
-
-		dce_v11_0_show_cursor(crtc);
-
-		dce_v11_0_lock_cursor(crtc, false);
-	}
-}
-
-static int dce_v11_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
-				    u16 *blue, uint32_t size,
-				    struct drm_modeset_acquire_ctx *ctx)
-{
-	dce_v11_0_crtc_load_lut(crtc);
-
-	return 0;
-}
-
-static void dce_v11_0_crtc_destroy(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-
-	drm_crtc_cleanup(crtc);
-	kfree(amdgpu_crtc);
-}
-
-static const struct drm_crtc_funcs dce_v11_0_crtc_funcs = {
-	.cursor_set2 = dce_v11_0_crtc_cursor_set2,
-	.cursor_move = dce_v11_0_crtc_cursor_move,
-	.gamma_set = dce_v11_0_crtc_gamma_set,
-	.set_config = amdgpu_display_crtc_set_config,
-	.destroy = dce_v11_0_crtc_destroy,
-	.page_flip_target = amdgpu_display_crtc_page_flip_target,
-	.get_vblank_counter = amdgpu_get_vblank_counter_kms,
-	.enable_vblank = amdgpu_enable_vblank_kms,
-	.disable_vblank = amdgpu_disable_vblank_kms,
-	.get_vblank_timestamp = drm_crtc_vblank_helper_get_vblank_timestamp,
-};
-
-static void dce_v11_0_crtc_dpms(struct drm_crtc *crtc, int mode)
-{
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	unsigned type;
-
-	switch (mode) {
-	case DRM_MODE_DPMS_ON:
-		amdgpu_crtc->enabled = true;
-		amdgpu_atombios_crtc_enable(crtc, ATOM_ENABLE);
-		dce_v11_0_vga_enable(crtc, true);
-		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
-		dce_v11_0_vga_enable(crtc, false);
-		/* Make sure VBLANK and PFLIP interrupts are still enabled */
-		type = amdgpu_display_crtc_idx_to_irq_type(adev,
-						amdgpu_crtc->crtc_id);
-		amdgpu_irq_update(adev, &adev->crtc_irq, type);
-		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
-		drm_crtc_vblank_on(crtc);
-		dce_v11_0_crtc_load_lut(crtc);
-		break;
-	case DRM_MODE_DPMS_STANDBY:
-	case DRM_MODE_DPMS_SUSPEND:
-	case DRM_MODE_DPMS_OFF:
-		drm_crtc_vblank_off(crtc);
-		if (amdgpu_crtc->enabled) {
-			dce_v11_0_vga_enable(crtc, true);
-			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
-			dce_v11_0_vga_enable(crtc, false);
-		}
-		amdgpu_atombios_crtc_enable(crtc, ATOM_DISABLE);
-		amdgpu_crtc->enabled = false;
-		break;
-	}
-	/* adjust pm to dpms */
-	amdgpu_dpm_compute_clocks(adev);
-}
-
-static void dce_v11_0_crtc_prepare(struct drm_crtc *crtc)
-{
-	/* disable crtc pair power gating before programming */
-	amdgpu_atombios_crtc_powergate(crtc, ATOM_DISABLE);
-	amdgpu_atombios_crtc_lock(crtc, ATOM_ENABLE);
-	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
-}
-
-static void dce_v11_0_crtc_commit(struct drm_crtc *crtc)
-{
-	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
-	amdgpu_atombios_crtc_lock(crtc, ATOM_DISABLE);
-}
-
-static void dce_v11_0_crtc_disable(struct drm_crtc *crtc)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-	struct amdgpu_atom_ss ss;
-	int i;
-
-	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
-	if (crtc->primary->fb) {
-		int r;
-		struct amdgpu_bo *abo;
-
-		abo = gem_to_amdgpu_bo(crtc->primary->fb->obj[0]);
-		r = amdgpu_bo_reserve(abo, true);
-		if (unlikely(r))
-			DRM_ERROR("failed to reserve abo before unpin\n");
-		else {
-			amdgpu_bo_unpin(abo);
-			amdgpu_bo_unreserve(abo);
-		}
-	}
-	/* disable the GRPH */
-	dce_v11_0_grph_enable(crtc, false);
-
-	amdgpu_atombios_crtc_powergate(crtc, ATOM_ENABLE);
-
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		if (adev->mode_info.crtcs[i] &&
-		    adev->mode_info.crtcs[i]->enabled &&
-		    i != amdgpu_crtc->crtc_id &&
-		    amdgpu_crtc->pll_id == adev->mode_info.crtcs[i]->pll_id) {
-			/* one other crtc is using this pll don't turn
-			 * off the pll
-			 */
-			goto done;
-		}
-	}
-
-	switch (amdgpu_crtc->pll_id) {
-	case ATOM_PPLL0:
-	case ATOM_PPLL1:
-	case ATOM_PPLL2:
-		/* disable the ppll */
-		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
-						 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
-		break;
-	case ATOM_COMBOPHY_PLL0:
-	case ATOM_COMBOPHY_PLL1:
-	case ATOM_COMBOPHY_PLL2:
-	case ATOM_COMBOPHY_PLL3:
-	case ATOM_COMBOPHY_PLL4:
-	case ATOM_COMBOPHY_PLL5:
-		/* disable the ppll */
-		amdgpu_atombios_crtc_program_pll(crtc, ATOM_CRTC_INVALID, amdgpu_crtc->pll_id,
-						 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
-		break;
-	default:
-		break;
-	}
-done:
-	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
-	amdgpu_crtc->adjusted_clock = 0;
-	amdgpu_crtc->encoder = NULL;
-	amdgpu_crtc->connector = NULL;
-}
-
-static int dce_v11_0_crtc_mode_set(struct drm_crtc *crtc,
-				  struct drm_display_mode *mode,
-				  struct drm_display_mode *adjusted_mode,
-				  int x, int y, struct drm_framebuffer *old_fb)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-
-	if (!amdgpu_crtc->adjusted_clock)
-		return -EINVAL;
-
-	if ((adev->asic_type == CHIP_POLARIS10) ||
-	    (adev->asic_type == CHIP_POLARIS11) ||
-	    (adev->asic_type == CHIP_POLARIS12) ||
-	    (adev->asic_type == CHIP_VEGAM)) {
-		struct amdgpu_encoder *amdgpu_encoder =
-			to_amdgpu_encoder(amdgpu_crtc->encoder);
-		int encoder_mode =
-			amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder);
-
-		/* SetPixelClock calculates the plls and ss values now */
-		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id,
-						 amdgpu_crtc->pll_id,
-						 encoder_mode, amdgpu_encoder->encoder_id,
-						 adjusted_mode->clock, 0, 0, 0, 0,
-						 amdgpu_crtc->bpc, amdgpu_crtc->ss_enabled, &amdgpu_crtc->ss);
-	} else {
-		amdgpu_atombios_crtc_set_pll(crtc, adjusted_mode);
-	}
-	amdgpu_atombios_crtc_set_dtd_timing(crtc, adjusted_mode);
-	dce_v11_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
-	amdgpu_atombios_crtc_overscan_setup(crtc, mode, adjusted_mode);
-	amdgpu_atombios_crtc_scaler_setup(crtc);
-	dce_v11_0_cursor_reset(crtc);
-	/* update the hw version fpr dpm */
-	amdgpu_crtc->hw_mode = *adjusted_mode;
-
-	return 0;
-}
-
-static bool dce_v11_0_crtc_mode_fixup(struct drm_crtc *crtc,
-				     const struct drm_display_mode *mode,
-				     struct drm_display_mode *adjusted_mode)
-{
-	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-	struct drm_encoder *encoder;
-
-	/* assign the encoder to the amdgpu crtc to avoid repeated lookups later */
-	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
-		if (encoder->crtc == crtc) {
-			amdgpu_crtc->encoder = encoder;
-			amdgpu_crtc->connector = amdgpu_get_connector_for_encoder(encoder);
-			break;
-		}
-	}
-	if ((amdgpu_crtc->encoder == NULL) || (amdgpu_crtc->connector == NULL)) {
-		amdgpu_crtc->encoder = NULL;
-		amdgpu_crtc->connector = NULL;
-		return false;
-	}
-	if (!amdgpu_display_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
-		return false;
-	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
-		return false;
-	/* pick pll */
-	amdgpu_crtc->pll_id = dce_v11_0_pick_pll(crtc);
-	/* if we can't get a PPLL for a non-DP encoder, fail */
-	if ((amdgpu_crtc->pll_id == ATOM_PPLL_INVALID) &&
-	    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
-		return false;
-
-	return true;
-}
-
-static int dce_v11_0_crtc_set_base(struct drm_crtc *crtc, int x, int y,
-				  struct drm_framebuffer *old_fb)
-{
-	return dce_v11_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
-}
-
-static int dce_v11_0_crtc_set_base_atomic(struct drm_crtc *crtc,
-					 struct drm_framebuffer *fb,
-					 int x, int y, enum mode_set_atomic state)
-{
-	return dce_v11_0_crtc_do_set_base(crtc, fb, x, y, 1);
-}
-
-static const struct drm_crtc_helper_funcs dce_v11_0_crtc_helper_funcs = {
-	.dpms = dce_v11_0_crtc_dpms,
-	.mode_fixup = dce_v11_0_crtc_mode_fixup,
-	.mode_set = dce_v11_0_crtc_mode_set,
-	.mode_set_base = dce_v11_0_crtc_set_base,
-	.mode_set_base_atomic = dce_v11_0_crtc_set_base_atomic,
-	.prepare = dce_v11_0_crtc_prepare,
-	.commit = dce_v11_0_crtc_commit,
-	.disable = dce_v11_0_crtc_disable,
-	.get_scanout_position = amdgpu_crtc_get_scanout_position,
-};
-
-static void dce_v11_0_panic_flush(struct drm_plane *plane)
-{
-	struct drm_framebuffer *fb;
-	struct amdgpu_crtc *amdgpu_crtc;
-	struct amdgpu_device *adev;
-	uint32_t fb_format;
-
-	if (!plane->fb)
-		return;
-
-	fb = plane->fb;
-	amdgpu_crtc = to_amdgpu_crtc(plane->crtc);
-	adev = drm_to_adev(fb->dev);
-
-	/* Disable DC tiling */
-	fb_format = RREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset);
-	fb_format &= ~GRPH_CONTROL__GRPH_ARRAY_MODE_MASK;
-	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
-
-}
-
-static const struct drm_plane_helper_funcs dce_v11_0_drm_primary_plane_helper_funcs = {
-	.get_scanout_buffer = amdgpu_display_get_scanout_buffer,
-	.panic_flush = dce_v11_0_panic_flush,
-};
-
-static int dce_v11_0_crtc_init(struct amdgpu_device *adev, int index)
-{
-	struct amdgpu_crtc *amdgpu_crtc;
-
-	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
-			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
-	if (amdgpu_crtc == NULL)
-		return -ENOMEM;
-
-	drm_crtc_init(adev_to_drm(adev), &amdgpu_crtc->base, &dce_v11_0_crtc_funcs);
-
-	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
-	amdgpu_crtc->crtc_id = index;
-	adev->mode_info.crtcs[index] = amdgpu_crtc;
-
-	amdgpu_crtc->max_cursor_width = 128;
-	amdgpu_crtc->max_cursor_height = 128;
-	adev_to_drm(adev)->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
-	adev_to_drm(adev)->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
-
-	switch (amdgpu_crtc->crtc_id) {
-	case 0:
-	default:
-		amdgpu_crtc->crtc_offset = CRTC0_REGISTER_OFFSET;
-		break;
-	case 1:
-		amdgpu_crtc->crtc_offset = CRTC1_REGISTER_OFFSET;
-		break;
-	case 2:
-		amdgpu_crtc->crtc_offset = CRTC2_REGISTER_OFFSET;
-		break;
-	case 3:
-		amdgpu_crtc->crtc_offset = CRTC3_REGISTER_OFFSET;
-		break;
-	case 4:
-		amdgpu_crtc->crtc_offset = CRTC4_REGISTER_OFFSET;
-		break;
-	case 5:
-		amdgpu_crtc->crtc_offset = CRTC5_REGISTER_OFFSET;
-		break;
-	}
-
-	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
-	amdgpu_crtc->adjusted_clock = 0;
-	amdgpu_crtc->encoder = NULL;
-	amdgpu_crtc->connector = NULL;
-	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v11_0_crtc_helper_funcs);
-	drm_plane_helper_add(amdgpu_crtc->base.primary, &dce_v11_0_drm_primary_plane_helper_funcs);
-
-	return 0;
-}
-
-static int dce_v11_0_early_init(struct amdgpu_ip_block *ip_block)
-{
-	struct amdgpu_device *adev = ip_block->adev;
-
-	adev->audio_endpt_rreg = &dce_v11_0_audio_endpt_rreg;
-	adev->audio_endpt_wreg = &dce_v11_0_audio_endpt_wreg;
-
-	dce_v11_0_set_display_funcs(adev);
-
-	adev->mode_info.num_crtc = dce_v11_0_get_num_crtc(adev);
-
-	switch (adev->asic_type) {
-	case CHIP_CARRIZO:
-		adev->mode_info.num_hpd = 6;
-		adev->mode_info.num_dig = 9;
-		break;
-	case CHIP_STONEY:
-		adev->mode_info.num_hpd = 6;
-		adev->mode_info.num_dig = 9;
-		break;
-	case CHIP_POLARIS10:
-	case CHIP_VEGAM:
-		adev->mode_info.num_hpd = 6;
-		adev->mode_info.num_dig = 6;
-		break;
-	case CHIP_POLARIS11:
-	case CHIP_POLARIS12:
-		adev->mode_info.num_hpd = 5;
-		adev->mode_info.num_dig = 5;
-		break;
-	default:
-		/* FIXME: not supported yet */
-		return -EINVAL;
-	}
-
-	dce_v11_0_set_irq_funcs(adev);
-
-	return 0;
-}
-
-static int dce_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
-{
-	int r, i;
-	struct amdgpu_device *adev = ip_block->adev;
-
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
-		if (r)
-			return r;
-	}
-
-	for (i = VISLANDS30_IV_SRCID_D1_GRPH_PFLIP; i < 20; i += 2) {
-		r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, i, &adev->pageflip_irq);
-		if (r)
-			return r;
-	}
-
-	/* HPD hotplug */
-	r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_HOTPLUG_DETECT_A, &adev->hpd_irq);
-	if (r)
-		return r;
-
-	adev_to_drm(adev)->mode_config.funcs = &amdgpu_mode_funcs;
-
-	adev_to_drm(adev)->mode_config.async_page_flip = true;
-
-	adev_to_drm(adev)->mode_config.max_width = 16384;
-	adev_to_drm(adev)->mode_config.max_height = 16384;
-
-	adev_to_drm(adev)->mode_config.preferred_depth = 24;
-	adev_to_drm(adev)->mode_config.prefer_shadow = 1;
-
-	adev_to_drm(adev)->mode_config.fb_modifiers_not_supported = true;
-
-	r = amdgpu_display_modeset_create_props(adev);
-	if (r)
-		return r;
-
-	adev_to_drm(adev)->mode_config.max_width = 16384;
-	adev_to_drm(adev)->mode_config.max_height = 16384;
-
-
-	/* allocate crtcs */
-	for (i = 0; i < adev->mode_info.num_crtc; i++) {
-		r = dce_v11_0_crtc_init(adev, i);
-		if (r)
-			return r;
-	}
-
-	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
-		amdgpu_display_print_display_setup(adev_to_drm(adev));
-	else
-		return -EINVAL;
-
-	/* setup afmt */
-	r = dce_v11_0_afmt_init(adev);
-	if (r)
-		return r;
-
-	r = dce_v11_0_audio_init(adev);
-	if (r)
-		return r;
-
-	/* Disable vblank IRQs aggressively for power-saving */
-	/* XXX: can this be enabled for DC? */
-	adev_to_drm(adev)->vblank_disable_immediate = true;
-
-	r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc);
-	if (r)
-		return r;
-
-	INIT_DELAYED_WORK(&adev->hotplug_work,
-		  amdgpu_display_hotplug_work_func);
-
-	drm_kms_helper_poll_init(adev_to_drm(adev));
-
-	adev->mode_info.mode_config_initialized = true;
-	return 0;
-}
-
-static int dce_v11_0_sw_fini(struct amdgpu_ip_block *ip_block)
-{
-	struct amdgpu_device *adev = ip_block->adev;
-
-	drm_edid_free(adev->mode_info.bios_hardcoded_edid);
-
-	drm_kms_helper_poll_fini(adev_to_drm(adev));
-
-	dce_v11_0_audio_fini(adev);
-
-	dce_v11_0_afmt_fini(adev);
-
-	drm_mode_config_cleanup(adev_to_drm(adev));
-	adev->mode_info.mode_config_initialized = false;
-
-	return 0;
-}
-
-static int dce_v11_0_hw_init(struct amdgpu_ip_block *ip_block)
-{
-	int i;
-	struct amdgpu_device *adev = ip_block->adev;
-
-	dce_v11_0_init_golden_registers(adev);
-
-	/* disable vga render */
-	dce_v11_0_set_vga_render_state(adev, false);
-	/* init dig PHYs, disp eng pll */
-	amdgpu_atombios_crtc_powergate_init(adev);
-	amdgpu_atombios_encoder_init_dig(adev);
-	if ((adev->asic_type == CHIP_POLARIS10) ||
-	    (adev->asic_type == CHIP_POLARIS11) ||
-	    (adev->asic_type == CHIP_POLARIS12) ||
-	    (adev->asic_type == CHIP_VEGAM)) {
-		amdgpu_atombios_crtc_set_dce_clock(adev, adev->clock.default_dispclk,
-						   DCE_CLOCK_TYPE_DISPCLK, ATOM_GCK_DFS);
-		amdgpu_atombios_crtc_set_dce_clock(adev, 0,
-						   DCE_CLOCK_TYPE_DPREFCLK, ATOM_GCK_DFS);
-	} else {
-		amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
-	}
-
-	/* initialize hpd */
-	dce_v11_0_hpd_init(adev);
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
-		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-	}
-
-	dce_v11_0_pageflip_interrupt_init(adev);
-
-	return 0;
-}
-
-static int dce_v11_0_hw_fini(struct amdgpu_ip_block *ip_block)
-{
-	int i;
-	struct amdgpu_device *adev = ip_block->adev;
-
-	dce_v11_0_hpd_fini(adev);
-
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
-		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-	}
-
-	dce_v11_0_pageflip_interrupt_fini(adev);
-
-	flush_delayed_work(&adev->hotplug_work);
-
-	return 0;
-}
-
-static int dce_v11_0_suspend(struct amdgpu_ip_block *ip_block)
-{
-	struct amdgpu_device *adev = ip_block->adev;
-	int r;
-
-	r = amdgpu_display_suspend_helper(adev);
-	if (r)
-		return r;
-
-	adev->mode_info.bl_level =
-		amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
-
-	return dce_v11_0_hw_fini(ip_block);
-}
-
-static int dce_v11_0_resume(struct amdgpu_ip_block *ip_block)
-{
-	struct amdgpu_device *adev = ip_block->adev;
-	int ret;
-
-	amdgpu_atombios_encoder_set_backlight_level_to_reg(adev,
-							   adev->mode_info.bl_level);
-
-	ret = dce_v11_0_hw_init(ip_block);
-
-	/* turn on the BL */
-	if (adev->mode_info.bl_encoder) {
-		u8 bl_level = amdgpu_display_backlight_get_level(adev,
-								  adev->mode_info.bl_encoder);
-		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
-						    bl_level);
-	}
-	if (ret)
-		return ret;
-
-	return amdgpu_display_resume_helper(adev);
-}
-
-static bool dce_v11_0_is_idle(void *handle)
-{
-	return true;
-}
-
-static int dce_v11_0_soft_reset(struct amdgpu_ip_block *ip_block)
-{
-	u32 srbm_soft_reset = 0, tmp;
-	struct amdgpu_device *adev = ip_block->adev;
-
-	if (dce_v11_0_is_display_hung(adev))
-		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
-
-	if (srbm_soft_reset) {
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-		tmp |= srbm_soft_reset;
-		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
-		WREG32(mmSRBM_SOFT_RESET, tmp);
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-
-		udelay(50);
-
-		tmp &= ~srbm_soft_reset;
-		WREG32(mmSRBM_SOFT_RESET, tmp);
-		tmp = RREG32(mmSRBM_SOFT_RESET);
-
-		/* Wait a little for things to settle down */
-		udelay(50);
-	}
-	return 0;
-}
-
-static void dce_v11_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
-						     int crtc,
-						     enum amdgpu_interrupt_state state)
-{
-	u32 lb_interrupt_mask;
-
-	if (crtc >= adev->mode_info.num_crtc) {
-		DRM_DEBUG("invalid crtc %d\n", crtc);
-		return;
-	}
-
-	switch (state) {
-	case AMDGPU_IRQ_STATE_DISABLE:
-		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
-		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
-						  VBLANK_INTERRUPT_MASK, 0);
-		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
-		break;
-	case AMDGPU_IRQ_STATE_ENABLE:
-		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
-		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
-						  VBLANK_INTERRUPT_MASK, 1);
-		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
-		break;
-	default:
-		break;
-	}
-}
-
-static void dce_v11_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
-						    int crtc,
-						    enum amdgpu_interrupt_state state)
-{
-	u32 lb_interrupt_mask;
-
-	if (crtc >= adev->mode_info.num_crtc) {
-		DRM_DEBUG("invalid crtc %d\n", crtc);
-		return;
-	}
-
-	switch (state) {
-	case AMDGPU_IRQ_STATE_DISABLE:
-		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
-		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
-						  VLINE_INTERRUPT_MASK, 0);
-		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
-		break;
-	case AMDGPU_IRQ_STATE_ENABLE:
-		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
-		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
-						  VLINE_INTERRUPT_MASK, 1);
-		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
-		break;
-	default:
-		break;
-	}
-}
-
-static int dce_v11_0_set_hpd_irq_state(struct amdgpu_device *adev,
-					struct amdgpu_irq_src *source,
-					unsigned hpd,
-					enum amdgpu_interrupt_state state)
-{
-	u32 tmp;
-
-	if (hpd >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", hpd);
-		return 0;
-	}
-
-	switch (state) {
-	case AMDGPU_IRQ_STATE_DISABLE:
-		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 0);
-		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
-		break;
-	case AMDGPU_IRQ_STATE_ENABLE:
-		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
-		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 1);
-		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
-		break;
-	default:
-		break;
-	}
-
-	return 0;
-}
-
-static int dce_v11_0_set_crtc_irq_state(struct amdgpu_device *adev,
-					struct amdgpu_irq_src *source,
-					unsigned type,
-					enum amdgpu_interrupt_state state)
-{
-	switch (type) {
-	case AMDGPU_CRTC_IRQ_VBLANK1:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 0, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VBLANK2:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 1, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VBLANK3:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 2, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VBLANK4:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 3, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VBLANK5:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 4, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VBLANK6:
-		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 5, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VLINE1:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 0, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VLINE2:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 1, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VLINE3:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 2, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VLINE4:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 3, state);
-		break;
-	case AMDGPU_CRTC_IRQ_VLINE5:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 4, state);
-		break;
-	 case AMDGPU_CRTC_IRQ_VLINE6:
-		dce_v11_0_set_crtc_vline_interrupt_state(adev, 5, state);
-		break;
-	default:
-		break;
-	}
-	return 0;
-}
-
-static int dce_v11_0_set_pageflip_irq_state(struct amdgpu_device *adev,
-					    struct amdgpu_irq_src *src,
-					    unsigned type,
-					    enum amdgpu_interrupt_state state)
-{
-	u32 reg;
-
-	if (type >= adev->mode_info.num_crtc) {
-		DRM_ERROR("invalid pageflip crtc %d\n", type);
-		return -EINVAL;
-	}
-
-	reg = RREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type]);
-	if (state == AMDGPU_IRQ_STATE_DISABLE)
-		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
-		       reg & ~GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
-	else
-		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
-		       reg | GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
-
-	return 0;
-}
-
-static int dce_v11_0_pageflip_irq(struct amdgpu_device *adev,
-				  struct amdgpu_irq_src *source,
-				  struct amdgpu_iv_entry *entry)
-{
-	unsigned long flags;
-	unsigned crtc_id;
-	struct amdgpu_crtc *amdgpu_crtc;
-	struct amdgpu_flip_work *works;
-
-	crtc_id = (entry->src_id - 8) >> 1;
-	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
-
-	if (crtc_id >= adev->mode_info.num_crtc) {
-		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
-		return -EINVAL;
-	}
-
-	if (RREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id]) &
-	    GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK)
-		WREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id],
-		       GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK);
-
-	/* IRQ could occur when in initial stage */
-	if(amdgpu_crtc == NULL)
-		return 0;
-
-	spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
-	works = amdgpu_crtc->pflip_works;
-	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED){
-		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
-						 "AMDGPU_FLIP_SUBMITTED(%d)\n",
-						 amdgpu_crtc->pflip_status,
-						 AMDGPU_FLIP_SUBMITTED);
-		spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
-		return 0;
-	}
-
-	/* page flip completed. clean up */
-	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
-	amdgpu_crtc->pflip_works = NULL;
-
-	/* wakeup usersapce */
-	if(works->event)
-		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
-
-	spin_unlock_irqrestore(&adev_to_drm(adev)->event_lock, flags);
-
-	drm_crtc_vblank_put(&amdgpu_crtc->base);
-	schedule_work(&works->unpin_work);
-
-	return 0;
-}
-
-static void dce_v11_0_hpd_int_ack(struct amdgpu_device *adev,
-				  int hpd)
-{
-	u32 tmp;
-
-	if (hpd >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", hpd);
-		return;
-	}
-
-	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
-	tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_ACK, 1);
-	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
-}
-
-static void dce_v11_0_crtc_vblank_int_ack(struct amdgpu_device *adev,
-					  int crtc)
-{
-	u32 tmp;
-
-	if (crtc < 0 || crtc >= adev->mode_info.num_crtc) {
-		DRM_DEBUG("invalid crtc %d\n", crtc);
-		return;
-	}
-
-	tmp = RREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc]);
-	tmp = REG_SET_FIELD(tmp, LB_VBLANK_STATUS, VBLANK_ACK, 1);
-	WREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc], tmp);
-}
-
-static void dce_v11_0_crtc_vline_int_ack(struct amdgpu_device *adev,
-					 int crtc)
-{
-	u32 tmp;
-
-	if (crtc < 0 || crtc >= adev->mode_info.num_crtc) {
-		DRM_DEBUG("invalid crtc %d\n", crtc);
-		return;
-	}
-
-	tmp = RREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc]);
-	tmp = REG_SET_FIELD(tmp, LB_VLINE_STATUS, VLINE_ACK, 1);
-	WREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc], tmp);
-}
-
-static int dce_v11_0_crtc_irq(struct amdgpu_device *adev,
-				struct amdgpu_irq_src *source,
-				struct amdgpu_iv_entry *entry)
-{
-	unsigned crtc = entry->src_id - 1;
-	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
-	unsigned int irq_type = amdgpu_display_crtc_idx_to_irq_type(adev,
-								    crtc);
-
-	switch (entry->src_data[0]) {
-	case 0: /* vblank */
-		if (disp_int & interrupt_status_offsets[crtc].vblank)
-			dce_v11_0_crtc_vblank_int_ack(adev, crtc);
-		else
-			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
-
-		if (amdgpu_irq_enabled(adev, source, irq_type)) {
-			drm_handle_vblank(adev_to_drm(adev), crtc);
-		}
-		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
-
-		break;
-	case 1: /* vline */
-		if (disp_int & interrupt_status_offsets[crtc].vline)
-			dce_v11_0_crtc_vline_int_ack(adev, crtc);
-		else
-			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
-
-		DRM_DEBUG("IH: D%d vline\n", crtc + 1);
-
-		break;
-	default:
-		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
-		break;
-	}
-
-	return 0;
-}
-
-static int dce_v11_0_hpd_irq(struct amdgpu_device *adev,
-			     struct amdgpu_irq_src *source,
-			     struct amdgpu_iv_entry *entry)
-{
-	uint32_t disp_int, mask;
-	unsigned hpd;
-
-	if (entry->src_data[0] >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
-		return 0;
-	}
-
-	hpd = entry->src_data[0];
-	disp_int = RREG32(interrupt_status_offsets[hpd].reg);
-	mask = interrupt_status_offsets[hpd].hpd;
-
-	if (disp_int & mask) {
-		dce_v11_0_hpd_int_ack(adev, hpd);
-		schedule_delayed_work(&adev->hotplug_work, 0);
-		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
-	}
-
-	return 0;
-}
-
-static int dce_v11_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
-					  enum amd_clockgating_state state)
-{
-	return 0;
-}
-
-static int dce_v11_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
-					  enum amd_powergating_state state)
-{
-	return 0;
-}
-
-static const struct amd_ip_funcs dce_v11_0_ip_funcs = {
-	.name = "dce_v11_0",
-	.early_init = dce_v11_0_early_init,
-	.sw_init = dce_v11_0_sw_init,
-	.sw_fini = dce_v11_0_sw_fini,
-	.hw_init = dce_v11_0_hw_init,
-	.hw_fini = dce_v11_0_hw_fini,
-	.suspend = dce_v11_0_suspend,
-	.resume = dce_v11_0_resume,
-	.is_idle = dce_v11_0_is_idle,
-	.soft_reset = dce_v11_0_soft_reset,
-	.set_clockgating_state = dce_v11_0_set_clockgating_state,
-	.set_powergating_state = dce_v11_0_set_powergating_state,
-};
-
-static void
-dce_v11_0_encoder_mode_set(struct drm_encoder *encoder,
-			  struct drm_display_mode *mode,
-			  struct drm_display_mode *adjusted_mode)
-{
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-
-	amdgpu_encoder->pixel_clock = adjusted_mode->clock;
-
-	/* need to call this here rather than in prepare() since we need some crtc info */
-	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
-
-	/* set scaler clears this on some chips */
-	dce_v11_0_set_interleave(encoder->crtc, mode);
-
-	if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) {
-		dce_v11_0_afmt_enable(encoder, true);
-		dce_v11_0_afmt_setmode(encoder, adjusted_mode);
-	}
-}
-
-static void dce_v11_0_encoder_prepare(struct drm_encoder *encoder)
-{
-	struct amdgpu_device *adev = drm_to_adev(encoder->dev);
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
-
-	if ((amdgpu_encoder->active_device &
-	     (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
-	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
-	     ENCODER_OBJECT_ID_NONE)) {
-		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
-		if (dig) {
-			dig->dig_encoder = dce_v11_0_pick_dig_encoder(encoder);
-			if (amdgpu_encoder->active_device & ATOM_DEVICE_DFP_SUPPORT)
-				dig->afmt = adev->mode_info.afmt[dig->dig_encoder];
-		}
-	}
-
-	amdgpu_atombios_scratch_regs_lock(adev, true);
-
-	if (connector) {
-		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
-
-		/* select the clock/data port if it uses a router */
-		if (amdgpu_connector->router.cd_valid)
-			amdgpu_i2c_router_select_cd_port(amdgpu_connector);
-
-		/* turn eDP panel on for mode set */
-		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP)
-			amdgpu_atombios_encoder_set_edp_panel_power(connector,
-							     ATOM_TRANSMITTER_ACTION_POWER_ON);
-	}
-
-	/* this is needed for the pll/ss setup to work correctly in some cases */
-	amdgpu_atombios_encoder_set_crtc_source(encoder);
-	/* set up the FMT blocks */
-	dce_v11_0_program_fmt(encoder);
-}
-
-static void dce_v11_0_encoder_commit(struct drm_encoder *encoder)
-{
-	struct drm_device *dev = encoder->dev;
-	struct amdgpu_device *adev = drm_to_adev(dev);
-
-	/* need to call this here as we need the crtc set up */
-	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
-	amdgpu_atombios_scratch_regs_lock(adev, false);
-}
-
-static void dce_v11_0_encoder_disable(struct drm_encoder *encoder)
-{
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct amdgpu_encoder_atom_dig *dig;
-
-	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
-
-	if (amdgpu_atombios_encoder_is_digital(encoder)) {
-		if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI)
-			dce_v11_0_afmt_enable(encoder, false);
-		dig = amdgpu_encoder->enc_priv;
-		dig->dig_encoder = -1;
-	}
-	amdgpu_encoder->active_device = 0;
-}
-
-/* these are handled by the primary encoders */
-static void dce_v11_0_ext_prepare(struct drm_encoder *encoder)
-{
-
-}
-
-static void dce_v11_0_ext_commit(struct drm_encoder *encoder)
-{
-
-}
-
-static void
-dce_v11_0_ext_mode_set(struct drm_encoder *encoder,
-		      struct drm_display_mode *mode,
-		      struct drm_display_mode *adjusted_mode)
-{
-
-}
-
-static void dce_v11_0_ext_disable(struct drm_encoder *encoder)
-{
-
-}
-
-static void
-dce_v11_0_ext_dpms(struct drm_encoder *encoder, int mode)
-{
-
-}
-
-static const struct drm_encoder_helper_funcs dce_v11_0_ext_helper_funcs = {
-	.dpms = dce_v11_0_ext_dpms,
-	.prepare = dce_v11_0_ext_prepare,
-	.mode_set = dce_v11_0_ext_mode_set,
-	.commit = dce_v11_0_ext_commit,
-	.disable = dce_v11_0_ext_disable,
-	/* no detect for TMDS/LVDS yet */
-};
-
-static const struct drm_encoder_helper_funcs dce_v11_0_dig_helper_funcs = {
-	.dpms = amdgpu_atombios_encoder_dpms,
-	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
-	.prepare = dce_v11_0_encoder_prepare,
-	.mode_set = dce_v11_0_encoder_mode_set,
-	.commit = dce_v11_0_encoder_commit,
-	.disable = dce_v11_0_encoder_disable,
-	.detect = amdgpu_atombios_encoder_dig_detect,
-};
-
-static const struct drm_encoder_helper_funcs dce_v11_0_dac_helper_funcs = {
-	.dpms = amdgpu_atombios_encoder_dpms,
-	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
-	.prepare = dce_v11_0_encoder_prepare,
-	.mode_set = dce_v11_0_encoder_mode_set,
-	.commit = dce_v11_0_encoder_commit,
-	.detect = amdgpu_atombios_encoder_dac_detect,
-};
-
-static void dce_v11_0_encoder_destroy(struct drm_encoder *encoder)
-{
-	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
-		amdgpu_atombios_encoder_fini_backlight(amdgpu_encoder);
-	kfree(amdgpu_encoder->enc_priv);
-	drm_encoder_cleanup(encoder);
-	kfree(amdgpu_encoder);
-}
-
-static const struct drm_encoder_funcs dce_v11_0_encoder_funcs = {
-	.destroy = dce_v11_0_encoder_destroy,
-};
-
-static void dce_v11_0_encoder_add(struct amdgpu_device *adev,
-				 uint32_t encoder_enum,
-				 uint32_t supported_device,
-				 u16 caps)
-{
-	struct drm_device *dev = adev_to_drm(adev);
-	struct drm_encoder *encoder;
-	struct amdgpu_encoder *amdgpu_encoder;
-
-	/* see if we already added it */
-	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
-		amdgpu_encoder = to_amdgpu_encoder(encoder);
-		if (amdgpu_encoder->encoder_enum == encoder_enum) {
-			amdgpu_encoder->devices |= supported_device;
-			return;
-		}
-
-	}
-
-	/* add a new one */
-	amdgpu_encoder = kzalloc(sizeof(struct amdgpu_encoder), GFP_KERNEL);
-	if (!amdgpu_encoder)
-		return;
-
-	encoder = &amdgpu_encoder->base;
-	switch (adev->mode_info.num_crtc) {
-	case 1:
-		encoder->possible_crtcs = 0x1;
-		break;
-	case 2:
-	default:
-		encoder->possible_crtcs = 0x3;
-		break;
-	case 3:
-		encoder->possible_crtcs = 0x7;
-		break;
-	case 4:
-		encoder->possible_crtcs = 0xf;
-		break;
-	case 5:
-		encoder->possible_crtcs = 0x1f;
-		break;
-	case 6:
-		encoder->possible_crtcs = 0x3f;
-		break;
-	}
-
-	amdgpu_encoder->enc_priv = NULL;
-
-	amdgpu_encoder->encoder_enum = encoder_enum;
-	amdgpu_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
-	amdgpu_encoder->devices = supported_device;
-	amdgpu_encoder->rmx_type = RMX_OFF;
-	amdgpu_encoder->underscan_type = UNDERSCAN_OFF;
-	amdgpu_encoder->is_ext_encoder = false;
-	amdgpu_encoder->caps = caps;
-
-	switch (amdgpu_encoder->encoder_id) {
-	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
-	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
-		drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-				 DRM_MODE_ENCODER_DAC, NULL);
-		drm_encoder_helper_add(encoder, &dce_v11_0_dac_helper_funcs);
-		break;
-	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
-	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
-		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
-			amdgpu_encoder->rmx_type = RMX_FULL;
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_LVDS, NULL);
-			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder);
-		} else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) {
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_DAC, NULL);
-			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
-		} else {
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_TMDS, NULL);
-			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
-		}
-		drm_encoder_helper_add(encoder, &dce_v11_0_dig_helper_funcs);
-		break;
-	case ENCODER_OBJECT_ID_SI170B:
-	case ENCODER_OBJECT_ID_CH7303:
-	case ENCODER_OBJECT_ID_EXTERNAL_SDVOA:
-	case ENCODER_OBJECT_ID_EXTERNAL_SDVOB:
-	case ENCODER_OBJECT_ID_TITFP513:
-	case ENCODER_OBJECT_ID_VT1623:
-	case ENCODER_OBJECT_ID_HDMI_SI1930:
-	case ENCODER_OBJECT_ID_TRAVIS:
-	case ENCODER_OBJECT_ID_NUTMEG:
-		/* these are handled by the primary encoders */
-		amdgpu_encoder->is_ext_encoder = true;
-		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_LVDS, NULL);
-		else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT))
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_DAC, NULL);
-		else
-			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
-					 DRM_MODE_ENCODER_TMDS, NULL);
-		drm_encoder_helper_add(encoder, &dce_v11_0_ext_helper_funcs);
-		break;
-	}
-}
-
-static const struct amdgpu_display_funcs dce_v11_0_display_funcs = {
-	.bandwidth_update = &dce_v11_0_bandwidth_update,
-	.vblank_get_counter = &dce_v11_0_vblank_get_counter,
-	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
-	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
-	.hpd_sense = &dce_v11_0_hpd_sense,
-	.hpd_set_polarity = &dce_v11_0_hpd_set_polarity,
-	.hpd_get_gpio_reg = &dce_v11_0_hpd_get_gpio_reg,
-	.page_flip = &dce_v11_0_page_flip,
-	.page_flip_get_scanoutpos = &dce_v11_0_crtc_get_scanoutpos,
-	.add_encoder = &dce_v11_0_encoder_add,
-	.add_connector = &amdgpu_connector_add,
-};
-
-static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev)
-{
-	adev->mode_info.funcs = &dce_v11_0_display_funcs;
-}
-
-static const struct amdgpu_irq_src_funcs dce_v11_0_crtc_irq_funcs = {
-	.set = dce_v11_0_set_crtc_irq_state,
-	.process = dce_v11_0_crtc_irq,
-};
-
-static const struct amdgpu_irq_src_funcs dce_v11_0_pageflip_irq_funcs = {
-	.set = dce_v11_0_set_pageflip_irq_state,
-	.process = dce_v11_0_pageflip_irq,
-};
-
-static const struct amdgpu_irq_src_funcs dce_v11_0_hpd_irq_funcs = {
-	.set = dce_v11_0_set_hpd_irq_state,
-	.process = dce_v11_0_hpd_irq,
-};
-
-static void dce_v11_0_set_irq_funcs(struct amdgpu_device *adev)
-{
-	if (adev->mode_info.num_crtc > 0)
-		adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VLINE1 + adev->mode_info.num_crtc;
-	else
-		adev->crtc_irq.num_types = 0;
-	adev->crtc_irq.funcs = &dce_v11_0_crtc_irq_funcs;
-
-	adev->pageflip_irq.num_types = adev->mode_info.num_crtc;
-	adev->pageflip_irq.funcs = &dce_v11_0_pageflip_irq_funcs;
-
-	adev->hpd_irq.num_types = adev->mode_info.num_hpd;
-	adev->hpd_irq.funcs = &dce_v11_0_hpd_irq_funcs;
-}
-
-const struct amdgpu_ip_block_version dce_v11_0_ip_block =
-{
-	.type = AMD_IP_BLOCK_TYPE_DCE,
-	.major = 11,
-	.minor = 0,
-	.rev = 0,
-	.funcs = &dce_v11_0_ip_funcs,
-};
-
-const struct amdgpu_ip_block_version dce_v11_2_ip_block =
-{
-	.type = AMD_IP_BLOCK_TYPE_DCE,
-	.major = 11,
-	.minor = 2,
-	.rev = 0,
-	.funcs = &dce_v11_0_ip_funcs,
-};
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
index 915804a6a1d7..acc887a58518 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
@@ -40,18 +40,25 @@
 #include "amdgpu_connectors.h"
 #include "amdgpu_display.h"
 
+#include "dce_v6_0.h"
+#include "sid.h"
+
 #include "bif/bif_3_0_d.h"
 #include "bif/bif_3_0_sh_mask.h"
+
 #include "oss/oss_1_0_d.h"
 #include "oss/oss_1_0_sh_mask.h"
+
 #include "gca/gfx_6_0_d.h"
 #include "gca/gfx_6_0_sh_mask.h"
+#include "gca/gfx_7_2_enum.h"
+
 #include "gmc/gmc_6_0_d.h"
 #include "gmc/gmc_6_0_sh_mask.h"
+
 #include "dce/dce_6_0_d.h"
 #include "dce/dce_6_0_sh_mask.h"
-#include "gca/gfx_7_2_enum.h"
-#include "dce_v6_0.h"
+
 #include "si_enums.h"
 
 static void dce_v6_0_set_display_funcs(struct amdgpu_device *adev);
@@ -59,31 +66,31 @@ static void dce_v6_0_set_irq_funcs(struct amdgpu_device *adev);
 
 static const u32 crtc_offsets[6] =
 {
-	SI_CRTC0_REGISTER_OFFSET,
-	SI_CRTC1_REGISTER_OFFSET,
-	SI_CRTC2_REGISTER_OFFSET,
-	SI_CRTC3_REGISTER_OFFSET,
-	SI_CRTC4_REGISTER_OFFSET,
-	SI_CRTC5_REGISTER_OFFSET
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET
 };
 
 static const u32 hpd_offsets[] =
 {
-	mmDC_HPD1_INT_STATUS - mmDC_HPD1_INT_STATUS,
-	mmDC_HPD2_INT_STATUS - mmDC_HPD1_INT_STATUS,
-	mmDC_HPD3_INT_STATUS - mmDC_HPD1_INT_STATUS,
-	mmDC_HPD4_INT_STATUS - mmDC_HPD1_INT_STATUS,
-	mmDC_HPD5_INT_STATUS - mmDC_HPD1_INT_STATUS,
-	mmDC_HPD6_INT_STATUS - mmDC_HPD1_INT_STATUS,
+	HPD0_REGISTER_OFFSET,
+	HPD1_REGISTER_OFFSET,
+	HPD2_REGISTER_OFFSET,
+	HPD3_REGISTER_OFFSET,
+	HPD4_REGISTER_OFFSET,
+	HPD5_REGISTER_OFFSET
 };
 
 static const uint32_t dig_offsets[] = {
-	SI_CRTC0_REGISTER_OFFSET,
-	SI_CRTC1_REGISTER_OFFSET,
-	SI_CRTC2_REGISTER_OFFSET,
-	SI_CRTC3_REGISTER_OFFSET,
-	SI_CRTC4_REGISTER_OFFSET,
-	SI_CRTC5_REGISTER_OFFSET,
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET,
 	(0x13830 - 0x7030) >> 2,
 };
 
@@ -206,9 +213,9 @@ static void dce_v6_0_page_flip(struct amdgpu_device *adev,
 	/* update the scanout addresses */
 	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
 	       upper_32_bits(crtc_base));
+	/* writing to the low address triggers the update */
 	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
 	       (u32)crtc_base);
-
 	/* post the write */
 	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
 }
@@ -218,11 +225,11 @@ static int dce_v6_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
 {
 	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
 		return -EINVAL;
+
 	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
 	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
 
 	return 0;
-
 }
 
 /**
@@ -242,7 +249,8 @@ static bool dce_v6_0_hpd_sense(struct amdgpu_device *adev,
 	if (hpd >= adev->mode_info.num_hpd)
 		return connected;
 
-	if (RREG32(mmDC_HPD1_INT_STATUS + hpd_offsets[hpd]) & DC_HPD1_INT_STATUS__DC_HPD1_SENSE_MASK)
+	if (RREG32(mmDC_HPD1_INT_STATUS + hpd_offsets[hpd]) &
+	    DC_HPD1_INT_STATUS__DC_HPD1_SENSE_MASK)
 		connected = true;
 
 	return connected;
@@ -279,7 +287,7 @@ static void dce_v6_0_hpd_int_ack(struct amdgpu_device *adev,
 	u32 tmp;
 
 	if (hpd >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", hpd);
+		DRM_DEBUG("invalid hpd %d\n", hpd);
 		return;
 	}
 
@@ -370,13 +378,41 @@ static u32 dce_v6_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
 	return mmDC_GPIO_HPD_A;
 }
 
+static bool dce_v6_0_is_display_hung(struct amdgpu_device *adev)
+{
+	u32 crtc_hung = 0;
+	u32 crtc_status[6];
+	u32 i, j, tmp;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (RREG32(mmCRTC_CONTROL + crtc_offsets[i]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK) {
+			crtc_status[i] = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+			crtc_hung |= (1 << i);
+		}
+	}
+
+	for (j = 0; j < 10; j++) {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (crtc_hung & (1 << i)) {
+				tmp = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+				if (tmp != crtc_status[i])
+					crtc_hung &= ~(1 << i);
+			}
+		}
+		if (crtc_hung == 0)
+			return false;
+		udelay(100);
+	}
+
+	return true;
+}
+
 static void dce_v6_0_set_vga_render_state(struct amdgpu_device *adev,
 					  bool render)
 {
 	if (!render)
 		WREG32(mmVGA_RENDER_CONTROL,
-			RREG32(mmVGA_RENDER_CONTROL) & VGA_VSTATUS_CNTL);
-
+		       RREG32(mmVGA_RENDER_CONTROL) & ~VGA_RENDER_CONTROL__VGA_VSTATUS_CNTL_MASK);
 }
 
 static int dce_v6_0_get_num_crtc(struct amdgpu_device *adev)
@@ -419,7 +455,6 @@ void dce_v6_0_disable_dce(struct amdgpu_device *adev)
 
 static void dce_v6_0_program_fmt(struct drm_encoder *encoder)
 {
-
 	struct drm_device *dev = encoder->dev;
 	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
@@ -895,8 +930,8 @@ static void dce_v6_0_program_watermarks(struct amdgpu_device *adev,
 		wm_high.dram_channels = dram_channels;
 		wm_high.num_heads = num_heads;
 
-		if (adev->pm.dpm_enabled) {
 		/* watermark for low clocks */
+		if (adev->pm.dpm_enabled) {
 			wm_low.yclk =
 				amdgpu_dpm_get_mclk(adev, true) * 10;
 			wm_low.sclk =
@@ -976,16 +1011,16 @@ static void dce_v6_0_program_watermarks(struct amdgpu_device *adev,
 	/* select wm A */
 	arb_control3 = RREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset);
 	tmp = arb_control3;
-	tmp &= ~LATENCY_WATERMARK_MASK(3);
-	tmp |= LATENCY_WATERMARK_MASK(1);
+	tmp &= ~(3 << DPG_PIPE_ARBITRATION_CONTROL3__URGENCY_WATERMARK_MASK__SHIFT);
+	tmp |= (1 << DPG_PIPE_ARBITRATION_CONTROL3__URGENCY_WATERMARK_MASK__SHIFT);
 	WREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset, tmp);
 	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset,
 	       ((latency_watermark_a << DPG_PIPE_URGENCY_CONTROL__URGENCY_LOW_WATERMARK__SHIFT)  |
 		(line_time << DPG_PIPE_URGENCY_CONTROL__URGENCY_HIGH_WATERMARK__SHIFT)));
 	/* select wm B */
 	tmp = RREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset);
-	tmp &= ~LATENCY_WATERMARK_MASK(3);
-	tmp |= LATENCY_WATERMARK_MASK(2);
+	tmp &= ~(3 << DPG_PIPE_ARBITRATION_CONTROL3__URGENCY_WATERMARK_MASK__SHIFT);
+	tmp |= (2 << DPG_PIPE_ARBITRATION_CONTROL3__URGENCY_WATERMARK_MASK__SHIFT);
 	WREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset, tmp);
 	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset,
 	       ((latency_watermark_b << DPG_PIPE_URGENCY_CONTROL__URGENCY_LOW_WATERMARK__SHIFT) |
@@ -999,13 +1034,26 @@ static void dce_v6_0_program_watermarks(struct amdgpu_device *adev,
 
 	/* save values for DPM */
 	amdgpu_crtc->line_time = line_time;
-	amdgpu_crtc->wm_high = latency_watermark_a;
 
 	/* Save number of lines the linebuffer leads before the scanout */
 	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
 }
 
 /* watermark setup */
+/**
+ * dce_v6_0_line_buffer_adjust - Set up the line buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @mode: the current display mode on the selected display
+ * controller
+ * @other_mode: the display mode of another display controller
+ *              that may be sharing the line buffer
+ *
+ * Setup up the line buffer allocation for
+ * the selected display controller (CIK).
+ * Returns the line buffer size in pixels.
+ */
 static u32 dce_v6_0_line_buffer_adjust(struct amdgpu_device *adev,
 				   struct amdgpu_crtc *amdgpu_crtc,
 				   struct drm_display_mode *mode,
@@ -1040,7 +1088,7 @@ static u32 dce_v6_0_line_buffer_adjust(struct amdgpu_device *adev,
 	}
 
 	WREG32(mmDC_LB_MEMORY_SPLIT + amdgpu_crtc->crtc_offset,
-	       DC_LB_MEMORY_CONFIG(tmp));
+	       (tmp << DC_LB_MEMORY_SPLIT__DC_LB_MEMORY_CONFIG__SHIFT));
 
 	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset,
 	       (buffer_alloc << PIPE0_DMIF_BUFFER_CONTROL__DMIF_BUFFERS_ALLOCATED__SHIFT));
@@ -1257,6 +1305,7 @@ static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 offset;
 	struct drm_connector *connector;
 	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
@@ -1278,6 +1327,11 @@ static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
 	};
 
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	offset = dig->afmt->pin->offset;
+
 	drm_connector_list_iter_begin(dev, &iter);
 	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
@@ -1299,7 +1353,7 @@ static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 		return;
 
 	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
-		u32 tmp = 0;
+		u32 value = 0;
 		u8 stereo_freqs = 0;
 		int max_channels = -1;
 		int j;
@@ -1309,12 +1363,12 @@ static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 
 			if (sad->format == eld_reg_to_type[i][1]) {
 				if (sad->channels > max_channels) {
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							MAX_CHANNELS, sad->channels);
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							DESCRIPTOR_BYTE_2, sad->byte2);
-					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-							SUPPORTED_FREQUENCIES, sad->freq);
+					value = (sad->channels <<
+						AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__MAX_CHANNELS__SHIFT) |
+					       (sad->byte2 <<
+						AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__DESCRIPTOR_BYTE_2__SHIFT) |
+					       (sad->freq <<
+						AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES__SHIFT);
 					max_channels = sad->channels;
 				}
 
@@ -1325,13 +1379,13 @@ static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 			}
 		}
 
-		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
-				SUPPORTED_FREQUENCIES_STEREO, stereo_freqs);
-		WREG32_AUDIO_ENDPT(dig->afmt->pin->offset, eld_reg_to_type[i][0], tmp);
+		value |= (stereo_freqs <<
+			AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES_STEREO__SHIFT);
+
+		WREG32_AUDIO_ENDPT(offset, eld_reg_to_type[i][0], value);
 	}
 
 	kfree(sads);
-
 }
 
 static void dce_v6_0_audio_enable(struct amdgpu_device *adev,
@@ -1347,13 +1401,13 @@ static void dce_v6_0_audio_enable(struct amdgpu_device *adev,
 
 static const u32 pin_offsets[7] =
 {
-	(0x1780 - 0x1780),
-	(0x1786 - 0x1780),
-	(0x178c - 0x1780),
-	(0x1792 - 0x1780),
-	(0x1798 - 0x1780),
-	(0x179d - 0x1780),
-	(0x17a4 - 0x1780),
+	AUD0_REGISTER_OFFSET,
+	AUD1_REGISTER_OFFSET,
+	AUD2_REGISTER_OFFSET,
+	AUD3_REGISTER_OFFSET,
+	AUD4_REGISTER_OFFSET,
+	AUD5_REGISTER_OFFSET,
+	AUD6_REGISTER_OFFSET,
 };
 
 static int dce_v6_0_audio_init(struct amdgpu_device *adev)
@@ -1386,6 +1440,8 @@ static int dce_v6_0_audio_init(struct amdgpu_device *adev)
 		adev->mode_info.audio.pin[i].connected = false;
 		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
 		adev->mode_info.audio.pin[i].id = i;
+		/* disable audio.  it will be set up later */
+		/* XXX remove once we switch to ip funcs */
 		dce_v6_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
 	}
 
@@ -1394,17 +1450,12 @@ static int dce_v6_0_audio_init(struct amdgpu_device *adev)
 
 static void dce_v6_0_audio_fini(struct amdgpu_device *adev)
 {
-	int i;
-
 	if (!amdgpu_audio)
 		return;
 
 	if (!adev->mode_info.audio.enabled)
 		return;
 
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
-		dce_v6_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-
 	adev->mode_info.audio.enabled = false;
 }
 
@@ -1835,7 +1886,7 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc,
 	struct amdgpu_bo *abo;
 	uint64_t fb_location, tiling_flags;
 	uint32_t fb_format, fb_pitch_pixels, pipe_config;
-	u32 fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_NONE);
+	u32 fb_swap = (GRPH_ENDIAN_NONE << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 	u32 viewport_w, viewport_h;
 	int r;
 	bool bypass_lut = false;
@@ -1875,76 +1926,76 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc,
 
 	switch (target_fb->format->format) {
 	case DRM_FORMAT_C8:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_8BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_INDEXED));
+		fb_format = ((GRPH_DEPTH_8BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_INDEXED << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 		break;
 	case DRM_FORMAT_XRGB4444:
 	case DRM_FORMAT_ARGB4444:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB4444));
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB4444 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	case DRM_FORMAT_XRGB1555:
 	case DRM_FORMAT_ARGB1555:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB1555));
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB1555 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	case DRM_FORMAT_BGRX5551:
 	case DRM_FORMAT_BGRA5551:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_BGRA5551));
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_BGRA5551 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	case DRM_FORMAT_RGB565:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB565));
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB565 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	case DRM_FORMAT_XRGB8888:
 	case DRM_FORMAT_ARGB8888:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_32BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB8888));
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB8888 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN32);
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	case DRM_FORMAT_XRGB2101010:
 	case DRM_FORMAT_ARGB2101010:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_32BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB2101010));
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB2101010 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN32);
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
 		bypass_lut = true;
 		break;
 	case DRM_FORMAT_BGRX1010102:
 	case DRM_FORMAT_BGRA1010102:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_32BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_BGRA1010102));
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_BGRA1010102 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN32);
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
 		bypass_lut = true;
 		break;
 	case DRM_FORMAT_XBGR8888:
 	case DRM_FORMAT_ABGR8888:
-		fb_format = (GRPH_DEPTH(GRPH_DEPTH_32BPP) |
-			     GRPH_FORMAT(GRPH_FORMAT_ARGB8888));
-		fb_swap = (GRPH_RED_CROSSBAR(GRPH_RED_SEL_B) |
-			   GRPH_BLUE_CROSSBAR(GRPH_BLUE_SEL_R));
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB8888 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+		fb_swap = ((GRPH_RED_SEL_B << GRPH_SWAP_CNTL__GRPH_RED_CROSSBAR__SHIFT) |
+			   (GRPH_BLUE_SEL_R << GRPH_SWAP_CNTL__GRPH_BLUE_CROSSBAR__SHIFT));
 #ifdef __BIG_ENDIAN
-		fb_swap |= GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN32);
+		fb_swap |= (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
 #endif
 		break;
 	default:
@@ -1962,18 +2013,18 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc,
 		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
 		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
 
-		fb_format |= GRPH_NUM_BANKS(num_banks);
-		fb_format |= GRPH_ARRAY_MODE(GRPH_ARRAY_2D_TILED_THIN1);
-		fb_format |= GRPH_TILE_SPLIT(tile_split);
-		fb_format |= GRPH_BANK_WIDTH(bankw);
-		fb_format |= GRPH_BANK_HEIGHT(bankh);
-		fb_format |= GRPH_MACRO_TILE_ASPECT(mtaspect);
+		fb_format |= (num_banks << GRPH_CONTROL__GRPH_NUM_BANKS__SHIFT);
+		fb_format |= (GRPH_ARRAY_2D_TILED_THIN1 << GRPH_CONTROL__GRPH_ARRAY_MODE__SHIFT);
+		fb_format |= (tile_split << GRPH_CONTROL__GRPH_TILE_SPLIT__SHIFT);
+		fb_format |= (bankw << GRPH_CONTROL__GRPH_BANK_WIDTH__SHIFT);
+		fb_format |= (bankh << GRPH_CONTROL__GRPH_BANK_HEIGHT__SHIFT);
+		fb_format |= (mtaspect << GRPH_CONTROL__GRPH_MACRO_TILE_ASPECT__SHIFT);
 	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
-		fb_format |= GRPH_ARRAY_MODE(GRPH_ARRAY_1D_TILED_THIN1);
+		fb_format |= (GRPH_ARRAY_1D_TILED_THIN1 << GRPH_CONTROL__GRPH_ARRAY_MODE__SHIFT);
 	}
 
 	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
-	fb_format |= GRPH_PIPE_CONFIG(pipe_config);
+	fb_format |= (pipe_config << GRPH_CONTROL__GRPH_PIPE_CONFIG__SHIFT);
 
 	dce_v6_0_vga_enable(crtc, false);
 
@@ -1989,7 +2040,7 @@ static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc,
 	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
 	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
 	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
-	       (u32) fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
+	       (u32) fb_location & GRPH_SECONDARY_SURFACE_ADDRESS__GRPH_SECONDARY_SURFACE_ADDRESS_MASK);
 	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
 	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
 
@@ -2057,14 +2108,13 @@ static void dce_v6_0_set_interleave(struct drm_crtc *crtc,
 
 	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
 		WREG32(mmDATA_FORMAT + amdgpu_crtc->crtc_offset,
-		       INTERLEAVE_EN);
+			DATA_FORMAT__INTERLEAVE_EN_MASK);
 	else
 		WREG32(mmDATA_FORMAT + amdgpu_crtc->crtc_offset, 0);
 }
 
 static void dce_v6_0_crtc_load_lut(struct drm_crtc *crtc)
 {
-
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
 	struct amdgpu_device *adev = drm_to_adev(dev);
@@ -2074,15 +2124,15 @@ static void dce_v6_0_crtc_load_lut(struct drm_crtc *crtc)
 	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
 
 	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << INPUT_CSC_CONTROL__INPUT_CSC_GRPH_MODE__SHIFT) |
-		(0 << INPUT_CSC_CONTROL__INPUT_CSC_OVL_MODE__SHIFT)));
+	       ((INPUT_CSC_BYPASS << INPUT_CSC_CONTROL__INPUT_CSC_GRPH_MODE__SHIFT) |
+		(INPUT_CSC_BYPASS << INPUT_CSC_CONTROL__INPUT_CSC_OVL_MODE__SHIFT)));
 	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset,
 	       PRESCALE_GRPH_CONTROL__GRPH_PRESCALE_BYPASS_MASK);
 	WREG32(mmPRESCALE_OVL_CONTROL + amdgpu_crtc->crtc_offset,
 	       PRESCALE_OVL_CONTROL__OVL_PRESCALE_BYPASS_MASK);
 	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << INPUT_GAMMA_CONTROL__GRPH_INPUT_GAMMA_MODE__SHIFT) |
-		(0 << INPUT_GAMMA_CONTROL__OVL_INPUT_GAMMA_MODE__SHIFT)));
+	       ((INPUT_GAMMA_USE_LUT << INPUT_GAMMA_CONTROL__GRPH_INPUT_GAMMA_MODE__SHIFT) |
+		(INPUT_GAMMA_USE_LUT << INPUT_GAMMA_CONTROL__OVL_INPUT_GAMMA_MODE__SHIFT)));
 
 	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
 
@@ -2109,19 +2159,19 @@ static void dce_v6_0_crtc_load_lut(struct drm_crtc *crtc)
 	}
 
 	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << DEGAMMA_CONTROL__GRPH_DEGAMMA_MODE__SHIFT) |
-		(0 << DEGAMMA_CONTROL__OVL_DEGAMMA_MODE__SHIFT) |
-		ICON_DEGAMMA_MODE(0) |
-		(0 << DEGAMMA_CONTROL__CURSOR_DEGAMMA_MODE__SHIFT)));
+	       ((DEGAMMA_BYPASS << DEGAMMA_CONTROL__GRPH_DEGAMMA_MODE__SHIFT) |
+		(DEGAMMA_BYPASS << DEGAMMA_CONTROL__OVL_DEGAMMA_MODE__SHIFT) |
+		(DEGAMMA_BYPASS << DEGAMMA_CONTROL__ICON_DEGAMMA_MODE__SHIFT) |
+		(DEGAMMA_BYPASS << DEGAMMA_CONTROL__CURSOR_DEGAMMA_MODE__SHIFT)));
 	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << GAMUT_REMAP_CONTROL__GRPH_GAMUT_REMAP_MODE__SHIFT) |
-		(0 << GAMUT_REMAP_CONTROL__OVL_GAMUT_REMAP_MODE__SHIFT)));
+	       ((GAMUT_REMAP_BYPASS << GAMUT_REMAP_CONTROL__GRPH_GAMUT_REMAP_MODE__SHIFT) |
+		(GAMUT_REMAP_BYPASS << GAMUT_REMAP_CONTROL__OVL_GAMUT_REMAP_MODE__SHIFT)));
 	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << REGAMMA_CONTROL__GRPH_REGAMMA_MODE__SHIFT) |
-		(0 << REGAMMA_CONTROL__OVL_REGAMMA_MODE__SHIFT)));
+	       ((REGAMMA_BYPASS << REGAMMA_CONTROL__GRPH_REGAMMA_MODE__SHIFT) |
+		(REGAMMA_BYPASS << REGAMMA_CONTROL__OVL_REGAMMA_MODE__SHIFT)));
 	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset,
-	       ((0 << OUTPUT_CSC_CONTROL__OUTPUT_CSC_GRPH_MODE__SHIFT) |
-		(0 << OUTPUT_CSC_CONTROL__OUTPUT_CSC_OVL_MODE__SHIFT)));
+	       ((OUTPUT_CSC_BYPASS << OUTPUT_CSC_CONTROL__OUTPUT_CSC_GRPH_MODE__SHIFT) |
+		(OUTPUT_CSC_BYPASS << OUTPUT_CSC_CONTROL__OUTPUT_CSC_OVL_MODE__SHIFT)));
 	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
 	WREG32(0x1a50 + amdgpu_crtc->crtc_offset, 0);
 
@@ -2216,8 +2266,6 @@ static void dce_v6_0_hide_cursor(struct drm_crtc *crtc)
 	WREG32(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
 	       (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
 	       (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
-
-
 }
 
 static void dce_v6_0_show_cursor(struct drm_crtc *crtc)
@@ -2234,7 +2282,6 @@ static void dce_v6_0_show_cursor(struct drm_crtc *crtc)
 	       CUR_CONTROL__CURSOR_EN_MASK |
 	       (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
 	       (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
-
 }
 
 static int dce_v6_0_cursor_move_locked(struct drm_crtc *crtc,
@@ -2545,7 +2592,6 @@ static bool dce_v6_0_crtc_mode_fixup(struct drm_crtc *crtc,
 				     const struct drm_display_mode *mode,
 				     struct drm_display_mode *adjusted_mode)
 {
-
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
 	struct drm_encoder *encoder;
@@ -2618,7 +2664,7 @@ static void dce_v6_0_panic_flush(struct drm_plane *plane)
 
 	/* Disable DC tiling */
 	fb_format = RREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset);
-	fb_format &= ~GRPH_ARRAY_MODE(0x7);
+	fb_format &= ~GRPH_CONTROL__GRPH_ARRAY_MODE_MASK;
 	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
 
 }
@@ -2694,7 +2740,6 @@ static int dce_v6_0_early_init(struct amdgpu_ip_block *ip_block)
 static int dce_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	int r, i;
-	bool ret;
 	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
@@ -2738,8 +2783,7 @@ static int dce_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
 			return r;
 	}
 
-	ret = amdgpu_atombios_get_connector_info_from_object_table(adev);
-	if (ret)
+	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
 		amdgpu_display_print_display_setup(adev_to_drm(adev));
 	else
 		return -EINVAL;
@@ -2865,14 +2909,35 @@ static int dce_v6_0_resume(struct amdgpu_ip_block *ip_block)
 	return amdgpu_display_resume_helper(adev);
 }
 
-static bool dce_v6_0_is_idle(void *handle)
+static bool dce_v6_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
 
 static int dce_v6_0_soft_reset(struct amdgpu_ip_block *ip_block)
 {
-	DRM_INFO("xxxx: dce_v6_0_soft_reset --- no impl!!\n");
+	u32 srbm_soft_reset = 0, tmp;
+	struct amdgpu_device *adev = ip_block->adev;
+
+	if (dce_v6_0_is_display_hung(adev))
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
 	return 0;
 }
 
@@ -2889,22 +2954,22 @@ static void dce_v6_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
 
 	switch (crtc) {
 	case 0:
-		reg_block = SI_CRTC0_REGISTER_OFFSET;
+		reg_block = CRTC0_REGISTER_OFFSET;
 		break;
 	case 1:
-		reg_block = SI_CRTC1_REGISTER_OFFSET;
+		reg_block = CRTC1_REGISTER_OFFSET;
 		break;
 	case 2:
-		reg_block = SI_CRTC2_REGISTER_OFFSET;
+		reg_block = CRTC2_REGISTER_OFFSET;
 		break;
 	case 3:
-		reg_block = SI_CRTC3_REGISTER_OFFSET;
+		reg_block = CRTC3_REGISTER_OFFSET;
 		break;
 	case 4:
-		reg_block = SI_CRTC4_REGISTER_OFFSET;
+		reg_block = CRTC4_REGISTER_OFFSET;
 		break;
 	case 5:
-		reg_block = SI_CRTC5_REGISTER_OFFSET;
+		reg_block = CRTC5_REGISTER_OFFSET;
 		break;
 	default:
 		DRM_DEBUG("invalid crtc %d\n", crtc);
@@ -2914,12 +2979,12 @@ static void dce_v6_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
 	switch (state) {
 	case AMDGPU_IRQ_STATE_DISABLE:
 		interrupt_mask = RREG32(mmINT_MASK + reg_block);
-		interrupt_mask &= ~VBLANK_INT_MASK;
+		interrupt_mask &= ~INT_MASK__VBLANK_INT_MASK;
 		WREG32(mmINT_MASK + reg_block, interrupt_mask);
 		break;
 	case AMDGPU_IRQ_STATE_ENABLE:
 		interrupt_mask = RREG32(mmINT_MASK + reg_block);
-		interrupt_mask |= VBLANK_INT_MASK;
+		interrupt_mask |= INT_MASK__VBLANK_INT_MASK;
 		WREG32(mmINT_MASK + reg_block, interrupt_mask);
 		break;
 	default:
@@ -2934,28 +2999,28 @@ static void dce_v6_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
 
 }
 
-static int dce_v6_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
+static int dce_v6_0_set_hpd_irq_state(struct amdgpu_device *adev,
 					    struct amdgpu_irq_src *src,
-					    unsigned type,
+					    unsigned hpd,
 					    enum amdgpu_interrupt_state state)
 {
 	u32 dc_hpd_int_cntl;
 
-	if (type >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", type);
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hpd %d\n", hpd);
 		return 0;
 	}
 
 	switch (state) {
 	case AMDGPU_IRQ_STATE_DISABLE:
-		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type]);
-		dc_hpd_int_cntl &= ~DC_HPDx_INT_EN;
-		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type], dc_hpd_int_cntl);
+		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+		dc_hpd_int_cntl &= ~DC_HPD1_INT_CONTROL__DC_HPD1_INT_EN_MASK;
+		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], dc_hpd_int_cntl);
 		break;
 	case AMDGPU_IRQ_STATE_ENABLE:
-		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type]);
-		dc_hpd_int_cntl |= DC_HPDx_INT_EN;
-		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type], dc_hpd_int_cntl);
+		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+		dc_hpd_int_cntl |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_EN_MASK;
+		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], dc_hpd_int_cntl);
 		break;
 	default:
 		break;
@@ -2964,7 +3029,7 @@ static int dce_v6_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int dce_v6_0_set_crtc_interrupt_state(struct amdgpu_device *adev,
+static int dce_v6_0_set_crtc_irq_state(struct amdgpu_device *adev,
 					     struct amdgpu_irq_src *src,
 					     unsigned type,
 					     enum amdgpu_interrupt_state state)
@@ -3024,7 +3089,7 @@ static int dce_v6_0_crtc_irq(struct amdgpu_device *adev,
 	switch (entry->src_data[0]) {
 	case 0: /* vblank */
 		if (disp_int & interrupt_status_offsets[crtc].vblank)
-			WREG32(mmVBLANK_STATUS + crtc_offsets[crtc], VBLANK_ACK);
+			WREG32(mmVBLANK_STATUS + crtc_offsets[crtc], VBLANK_STATUS__VBLANK_ACK_MASK);
 		else
 			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
 
@@ -3035,7 +3100,7 @@ static int dce_v6_0_crtc_irq(struct amdgpu_device *adev,
 		break;
 	case 1: /* vline */
 		if (disp_int & interrupt_status_offsets[crtc].vline)
-			WREG32(mmVLINE_STATUS + crtc_offsets[crtc], VLINE_ACK);
+			WREG32(mmVLINE_STATUS + crtc_offsets[crtc], VLINE_STATUS__VLINE_ACK_MASK);
 		else
 			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
 
@@ -3049,7 +3114,7 @@ static int dce_v6_0_crtc_irq(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int dce_v6_0_set_pageflip_interrupt_state(struct amdgpu_device *adev,
+static int dce_v6_0_set_pageflip_irq_state(struct amdgpu_device *adev,
 						 struct amdgpu_irq_src *src,
 						 unsigned type,
 						 enum amdgpu_interrupt_state state)
@@ -3100,7 +3165,7 @@ static int dce_v6_0_pageflip_irq(struct amdgpu_device *adev,
 
 	spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags);
 	works = amdgpu_crtc->pflip_works;
-	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED){
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED) {
 		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
 						"AMDGPU_FLIP_SUBMITTED(%d)\n",
 						amdgpu_crtc->pflip_status,
@@ -3148,7 +3213,6 @@ static int dce_v6_0_hpd_irq(struct amdgpu_device *adev,
 	}
 
 	return 0;
-
 }
 
 static int dce_v6_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
@@ -3178,12 +3242,10 @@ static const struct amd_ip_funcs dce_v6_0_ip_funcs = {
 	.set_powergating_state = dce_v6_0_set_powergating_state,
 };
 
-static void
-dce_v6_0_encoder_mode_set(struct drm_encoder *encoder,
+static void dce_v6_0_encoder_mode_set(struct drm_encoder *encoder,
 			  struct drm_display_mode *mode,
 			  struct drm_display_mode *adjusted_mode)
 {
-
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
 
@@ -3203,7 +3265,6 @@ dce_v6_0_encoder_mode_set(struct drm_encoder *encoder,
 
 static void dce_v6_0_encoder_prepare(struct drm_encoder *encoder)
 {
-
 	struct amdgpu_device *adev = drm_to_adev(encoder->dev);
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
@@ -3243,7 +3304,6 @@ static void dce_v6_0_encoder_prepare(struct drm_encoder *encoder)
 
 static void dce_v6_0_encoder_commit(struct drm_encoder *encoder)
 {
-
 	struct drm_device *dev = encoder->dev;
 	struct amdgpu_device *adev = drm_to_adev(dev);
 
@@ -3254,7 +3314,6 @@ static void dce_v6_0_encoder_commit(struct drm_encoder *encoder)
 
 static void dce_v6_0_encoder_disable(struct drm_encoder *encoder)
 {
-
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig;
 	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
@@ -3281,8 +3340,7 @@ static void dce_v6_0_ext_commit(struct drm_encoder *encoder)
 
 }
 
-static void
-dce_v6_0_ext_mode_set(struct drm_encoder *encoder,
+static void dce_v6_0_ext_mode_set(struct drm_encoder *encoder,
 		      struct drm_display_mode *mode,
 		      struct drm_display_mode *adjusted_mode)
 {
@@ -3294,8 +3352,7 @@ static void dce_v6_0_ext_disable(struct drm_encoder *encoder)
 
 }
 
-static void
-dce_v6_0_ext_dpms(struct drm_encoder *encoder, int mode)
+static void dce_v6_0_ext_dpms(struct drm_encoder *encoder, int mode)
 {
 
 }
@@ -3366,7 +3423,6 @@ static void dce_v6_0_encoder_add(struct amdgpu_device *adev,
 			amdgpu_encoder->devices |= supported_device;
 			return;
 		}
-
 	}
 
 	/* add a new one */
@@ -3473,17 +3529,17 @@ static void dce_v6_0_set_display_funcs(struct amdgpu_device *adev)
 }
 
 static const struct amdgpu_irq_src_funcs dce_v6_0_crtc_irq_funcs = {
-	.set = dce_v6_0_set_crtc_interrupt_state,
+	.set = dce_v6_0_set_crtc_irq_state,
 	.process = dce_v6_0_crtc_irq,
 };
 
 static const struct amdgpu_irq_src_funcs dce_v6_0_pageflip_irq_funcs = {
-	.set = dce_v6_0_set_pageflip_interrupt_state,
+	.set = dce_v6_0_set_pageflip_irq_state,
 	.process = dce_v6_0_pageflip_irq,
 };
 
 static const struct amdgpu_irq_src_funcs dce_v6_0_hpd_irq_funcs = {
-	.set = dce_v6_0_set_hpd_interrupt_state,
+	.set = dce_v6_0_set_hpd_irq_state,
 	.process = dce_v6_0_hpd_irq,
 };
 
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
index f2edc0fece5b..2ccd6aad8dd6 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
@@ -271,7 +271,7 @@ static void dce_v8_0_hpd_int_ack(struct amdgpu_device *adev,
 	u32 tmp;
 
 	if (hpd >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", hpd);
+		DRM_DEBUG("invalid hpd %d\n", hpd);
 		return;
 	}
 
@@ -1096,8 +1096,7 @@ static void dce_v8_0_program_watermarks(struct amdgpu_device *adev,
 
 	/* save values for DPM */
 	amdgpu_crtc->line_time = line_time;
-	amdgpu_crtc->wm_high = latency_watermark_a;
-	amdgpu_crtc->wm_low = latency_watermark_b;
+
 	/* Save number of lines the linebuffer leads before the scanout */
 	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
 }
@@ -1395,13 +1394,13 @@ static void dce_v8_0_audio_enable(struct amdgpu_device *adev,
 }
 
 static const u32 pin_offsets[7] = {
-	(0x1780 - 0x1780),
-	(0x1786 - 0x1780),
-	(0x178c - 0x1780),
-	(0x1792 - 0x1780),
-	(0x1798 - 0x1780),
-	(0x179d - 0x1780),
-	(0x17a4 - 0x1780),
+	AUD0_REGISTER_OFFSET,
+	AUD1_REGISTER_OFFSET,
+	AUD2_REGISTER_OFFSET,
+	AUD3_REGISTER_OFFSET,
+	AUD4_REGISTER_OFFSET,
+	AUD5_REGISTER_OFFSET,
+	AUD6_REGISTER_OFFSET,
 };
 
 static int dce_v8_0_audio_init(struct amdgpu_device *adev)
@@ -1443,17 +1442,12 @@ static int dce_v8_0_audio_init(struct amdgpu_device *adev)
 
 static void dce_v8_0_audio_fini(struct amdgpu_device *adev)
 {
-	int i;
-
 	if (!amdgpu_audio)
 		return;
 
 	if (!adev->mode_info.audio.enabled)
 		return;
 
-	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
-		dce_v8_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
-
 	adev->mode_info.audio.enabled = false;
 }
 
@@ -2887,7 +2881,7 @@ static int dce_v8_0_resume(struct amdgpu_ip_block *ip_block)
 	return amdgpu_display_resume_helper(adev);
 }
 
-static bool dce_v8_0_is_idle(void *handle)
+static bool dce_v8_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	return true;
 }
@@ -3021,7 +3015,7 @@ static void dce_v8_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
 	}
 }
 
-static int dce_v8_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
+static int dce_v8_0_set_hpd_irq_state(struct amdgpu_device *adev,
 					    struct amdgpu_irq_src *src,
 					    unsigned type,
 					    enum amdgpu_interrupt_state state)
@@ -3029,7 +3023,7 @@ static int dce_v8_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
 	u32 dc_hpd_int_cntl;
 
 	if (type >= adev->mode_info.num_hpd) {
-		DRM_DEBUG("invalid hdp %d\n", type);
+		DRM_DEBUG("invalid hpd %d\n", type);
 		return 0;
 	}
 
@@ -3051,7 +3045,7 @@ static int dce_v8_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int dce_v8_0_set_crtc_interrupt_state(struct amdgpu_device *adev,
+static int dce_v8_0_set_crtc_irq_state(struct amdgpu_device *adev,
 					     struct amdgpu_irq_src *src,
 					     unsigned type,
 					     enum amdgpu_interrupt_state state)
@@ -3136,7 +3130,7 @@ static int dce_v8_0_crtc_irq(struct amdgpu_device *adev,
 	return 0;
 }
 
-static int dce_v8_0_set_pageflip_interrupt_state(struct amdgpu_device *adev,
+static int dce_v8_0_set_pageflip_irq_state(struct amdgpu_device *adev,
 						 struct amdgpu_irq_src *src,
 						 unsigned type,
 						 enum amdgpu_interrupt_state state)
@@ -3547,17 +3541,17 @@ static void dce_v8_0_set_display_funcs(struct amdgpu_device *adev)
 }
 
 static const struct amdgpu_irq_src_funcs dce_v8_0_crtc_irq_funcs = {
-	.set = dce_v8_0_set_crtc_interrupt_state,
+	.set = dce_v8_0_set_crtc_irq_state,
 	.process = dce_v8_0_crtc_irq,
 };
 
 static const struct amdgpu_irq_src_funcs dce_v8_0_pageflip_irq_funcs = {
-	.set = dce_v8_0_set_pageflip_interrupt_state,
+	.set = dce_v8_0_set_pageflip_irq_state,
 	.process = dce_v8_0_pageflip_irq,
 };
 
 static const struct amdgpu_irq_src_funcs dce_v8_0_hpd_irq_funcs = {
-	.set = dce_v8_0_set_hpd_interrupt_state,
+	.set = dce_v8_0_set_hpd_irq_state,
 	.process = dce_v8_0_hpd_irq,
 };
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
index 5ba263fe5512..d75b9940f248 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
@@ -40,7 +40,6 @@
 #include "ivsrcid/gfx/irqsrcs_gfx_10_1.h"
 
 #include "soc15.h"
-#include "soc15d.h"
 #include "soc15_common.h"
 #include "clearstate_gfx10.h"
 #include "v10_structs.h"
@@ -369,11 +368,6 @@ static const struct amdgpu_hwip_reg_entry gc_reg_list_10_1[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_DEBUG_INST_ADDR),
 	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_LX6_CORE_PDEBUG_INST),
 	/* cp header registers */
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME2_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
 	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MES_HEADER_DUMP),
 	/* SE status registers */
 	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE0),
@@ -422,7 +416,16 @@ static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_10[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_SUSPEND_CNTL_STACK_OFFSET),
 	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_SUSPEND_CNTL_STACK_DW_CNT),
 	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_SUSPEND_WG_STATE_OFFSET),
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_DEQUEUE_STATUS)
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_DEQUEUE_STATUS),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
 };
 
 static const struct amdgpu_hwip_reg_entry gc_gfx_queue_reg_list_10[] = {
@@ -449,7 +452,32 @@ static const struct amdgpu_hwip_reg_entry gc_gfx_queue_reg_list_10[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_MQD_BASE_ADDR),
 	SOC15_REG_ENTRY_STR(GC, 0, mmCP_GFX_MQD_BASE_ADDR_HI),
 	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_WPTR_POLL_ADDR_LO),
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_WPTR_POLL_ADDR_HI)
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_RB_WPTR_POLL_ADDR_HI),
+	/* gfx header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
 };
 
 static const struct soc15_reg_golden golden_settings_gc_10_1[] = {
@@ -3790,12 +3818,65 @@ static void gfx10_kiq_invalidate_tlbs(struct amdgpu_ring *kiq_ring,
 	gfx_v10_0_ring_invalidate_tlbs(kiq_ring, pasid, flush_type, all_hub, 1);
 }
 
+static void gfx_v10_0_kiq_reset_hw_queue(struct amdgpu_ring *kiq_ring, uint32_t queue_type,
+					uint32_t me_id, uint32_t pipe_id, uint32_t queue_id,
+					uint32_t xcc_id, uint32_t vmid)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	unsigned i;
+	uint32_t tmp;
+
+	/* enter save mode */
+	amdgpu_gfx_rlc_enter_safe_mode(adev, xcc_id);
+	mutex_lock(&adev->srbm_mutex);
+	nv_grbm_select(adev, me_id, pipe_id, queue_id, 0);
+
+	if (queue_type == AMDGPU_RING_TYPE_COMPUTE) {
+		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 0x2);
+		WREG32_SOC15(GC, 0, mmSPI_COMPUTE_QUEUE_RESET, 0x1);
+		/* wait till dequeue take effects */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout)
+			dev_err(adev->dev, "fail to wait on hqd deactive\n");
+	} else if (queue_type == AMDGPU_RING_TYPE_GFX) {
+		WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX,
+			     (uint32_t)(0x1 << GRBM_GFX_INDEX__SE_BROADCAST_WRITES__SHIFT));
+		tmp = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid);
+		if (pipe_id == 0)
+			tmp = REG_SET_FIELD(tmp, CP_VMID_RESET, PIPE0_QUEUES, 1 << queue_id);
+		else
+			tmp = REG_SET_FIELD(tmp, CP_VMID_RESET, PIPE1_QUEUES, 1 << queue_id);
+		WREG32_SOC15(GC, 0, mmCP_VMID_RESET, tmp);
+
+		/* wait till dequeue take effects */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32_SOC15(GC, 0, mmCP_GFX_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		if (i >= adev->usec_timeout)
+			dev_err(adev->dev, "failed to wait on gfx hqd deactivate\n");
+	} else {
+		dev_err(adev->dev, "reset queue_type(%d) not supported\n", queue_type);
+	}
+
+	nv_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	/* exit safe mode */
+	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
+}
+
 static const struct kiq_pm4_funcs gfx_v10_0_kiq_pm4_funcs = {
 	.kiq_set_resources = gfx10_kiq_set_resources,
 	.kiq_map_queues = gfx10_kiq_map_queues,
 	.kiq_unmap_queues = gfx10_kiq_unmap_queues,
 	.kiq_query_status = gfx10_kiq_query_status,
 	.kiq_invalidate_tlbs = gfx10_kiq_invalidate_tlbs,
+	.kiq_reset_hw_queue = gfx_v10_0_kiq_reset_hw_queue,
 	.set_resources_size = 8,
 	.map_queues_size = 7,
 	.unmap_queues_size = 6,
@@ -3994,7 +4075,7 @@ static int gfx_v10_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	struct dma_fence *f = NULL;
 	unsigned int index;
 	uint64_t gpu_addr;
-	volatile uint32_t *cpu_ptr;
+	uint32_t *cpu_ptr;
 	long r;
 
 	memset(&ib, 0, sizeof(ib));
@@ -4241,12 +4322,9 @@ static u32 gfx_v10_0_get_csb_size(struct amdgpu_device *adev)
 	return count;
 }
 
-static void gfx_v10_0_get_csb_buffer(struct amdgpu_device *adev,
-				    volatile u32 *buffer)
+static void gfx_v10_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
-	u32 count = 0, i;
-	const struct cs_section_def *sect = NULL;
-	const struct cs_extent_def *ext = NULL;
+	u32 count = 0;
 	int ctx_reg_offset;
 
 	if (adev->gfx.rlc.cs_data == NULL)
@@ -4254,39 +4332,15 @@ static void gfx_v10_0_get_csb_buffer(struct amdgpu_device *adev,
 	if (buffer == NULL)
 		return;
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
-	buffer[count++] = cpu_to_le32(0x80000000);
-	buffer[count++] = cpu_to_le32(0x80000000);
-
-	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
-		for (ext = sect->section; ext->extent != NULL; ++ext) {
-			if (sect->id == SECT_CONTEXT) {
-				buffer[count++] =
-					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
-				buffer[count++] = cpu_to_le32(ext->reg_index -
-						PACKET3_SET_CONTEXT_REG_START);
-				for (i = 0; i < ext->reg_count; i++)
-					buffer[count++] = cpu_to_le32(ext->extent[i]);
-			} else {
-				return;
-			}
-		}
-	}
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
 
-	ctx_reg_offset =
-		SOC15_REG_OFFSET(GC, 0, mmPA_SC_TILE_STEERING_OVERRIDE) - PACKET3_SET_CONTEXT_REG_START;
+	ctx_reg_offset = SOC15_REG_OFFSET(GC, 0, mmPA_SC_TILE_STEERING_OVERRIDE) - PACKET3_SET_CONTEXT_REG_START;
 	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 1));
 	buffer[count++] = cpu_to_le32(ctx_reg_offset);
 	buffer[count++] = cpu_to_le32(adev->gfx.config.pa_sc_tile_steering_override);
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
-	buffer[count++] = cpu_to_le32(0);
+	amdgpu_gfx_csb_preamble_end(buffer, count);
 }
 
 static void gfx_v10_0_rlc_fini(struct amdgpu_device *adev)
@@ -4700,6 +4754,9 @@ static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
 	int i, j, k, r, ring_id = 0;
 	int xcc_id = 0;
 	struct amdgpu_device *adev = ip_block->adev;
+	int num_queue_per_pipe = 1; /* we only enable 1 KGQ per pipe */
+
+	INIT_DELAYED_WORK(&adev->gfx.idle_work, amdgpu_gfx_profile_idle_work_handler);
 
 	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
 	case IP_VERSION(10, 1, 10):
@@ -4709,7 +4766,7 @@ static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
 	case IP_VERSION(10, 1, 4):
 		adev->gfx.me.num_me = 1;
 		adev->gfx.me.num_pipe_per_me = 1;
-		adev->gfx.me.num_queue_per_pipe = 1;
+		adev->gfx.me.num_queue_per_pipe = 8;
 		adev->gfx.mec.num_mec = 2;
 		adev->gfx.mec.num_pipe_per_mec = 4;
 		adev->gfx.mec.num_queue_per_pipe = 8;
@@ -4724,7 +4781,7 @@ static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
 	case IP_VERSION(10, 3, 7):
 		adev->gfx.me.num_me = 1;
 		adev->gfx.me.num_pipe_per_me = 2;
-		adev->gfx.me.num_queue_per_pipe = 1;
+		adev->gfx.me.num_queue_per_pipe = 2;
 		adev->gfx.mec.num_mec = 2;
 		adev->gfx.mec.num_pipe_per_mec = 4;
 		adev->gfx.mec.num_queue_per_pipe = 4;
@@ -4739,8 +4796,26 @@ static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
 		break;
 	}
 	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(10, 1, 10):
+	case IP_VERSION(10, 1, 1):
+	case IP_VERSION(10, 1, 2):
+		adev->gfx.cleaner_shader_ptr = gfx_10_1_10_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_10_1_10_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version >= 101 &&
+		    adev->gfx.pfp_fw_version  >= 158 &&
+		    adev->gfx.mec_fw_version >= 151) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
 	case IP_VERSION(10, 3, 0):
+	case IP_VERSION(10, 3, 1):
 	case IP_VERSION(10, 3, 2):
+	case IP_VERSION(10, 3, 3):
 	case IP_VERSION(10, 3, 4):
 	case IP_VERSION(10, 3, 5):
 		adev->gfx.cleaner_shader_ptr = gfx_10_3_0_cleaner_shader_hex;
@@ -4756,6 +4831,34 @@ static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
 			}
 		}
 		break;
+	case IP_VERSION(10, 3, 6):
+		adev->gfx.cleaner_shader_ptr = gfx_10_3_0_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_10_3_0_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version >= 14 &&
+		    adev->gfx.pfp_fw_version >= 17 &&
+		    adev->gfx.mec_fw_version >= 24) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(10, 3, 7):
+		adev->gfx.cleaner_shader_ptr = gfx_10_3_0_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_10_3_0_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version >= 4 &&
+		    adev->gfx.pfp_fw_version >= 9 &&
+		    adev->gfx.mec_fw_version >= 12) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
 	default:
 		adev->gfx.enable_cleaner_shader = false;
 		break;
@@ -4816,7 +4919,7 @@ static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
 
 	/* set up the gfx ring */
 	for (i = 0; i < adev->gfx.me.num_me; i++) {
-		for (j = 0; j < adev->gfx.me.num_queue_per_pipe; j++) {
+		for (j = 0; j < num_queue_per_pipe; j++) {
 			for (k = 0; k < adev->gfx.me.num_pipe_per_me; k++) {
 				if (!amdgpu_gfx_is_me_queue_enabled(adev, i, k, j))
 					continue;
@@ -4848,11 +4951,16 @@ static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
 			}
 		}
 	}
-	/* TODO: Add queue reset mask when FW fully supports it */
+
 	adev->gfx.gfx_supported_reset =
 		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
 	adev->gfx.compute_supported_reset =
 		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+	if (!amdgpu_sriov_vf(adev) &&
+	    !adev->debug_disable_gpu_ring_reset) {
+		adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	}
 
 	r = amdgpu_gfx_kiq_init(adev, GFX10_MEC_HPD_SIZE, 0);
 	if (r) {
@@ -6044,7 +6152,7 @@ static int gfx_v10_0_cp_gfx_load_pfp_microcode(struct amdgpu_device *adev)
 	}
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	tmp = RREG32_SOC15(GC, 0, mmCP_PFP_IC_BASE_CNTL);
 	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, VMID, 0);
@@ -6122,7 +6230,7 @@ static int gfx_v10_0_cp_gfx_load_ce_microcode(struct amdgpu_device *adev)
 	}
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	tmp = RREG32_SOC15(GC, 0, mmCP_CE_IC_BASE_CNTL);
 	tmp = REG_SET_FIELD(tmp, CP_CE_IC_BASE_CNTL, VMID, 0);
@@ -6199,7 +6307,7 @@ static int gfx_v10_0_cp_gfx_load_me_microcode(struct amdgpu_device *adev)
 	}
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	tmp = RREG32_SOC15(GC, 0, mmCP_ME_IC_BASE_CNTL);
 	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, VMID, 0);
@@ -6574,7 +6682,7 @@ static int gfx_v10_0_cp_compute_load_microcode(struct amdgpu_device *adev)
 	}
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	tmp = RREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_CNTL);
 	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0);
@@ -6781,22 +6889,9 @@ static int gfx_v10_0_kgq_init_queue(struct amdgpu_ring *ring, bool reset)
 static int gfx_v10_0_cp_async_gfx_ring_resume(struct amdgpu_device *adev)
 {
 	int r, i;
-	struct amdgpu_ring *ring;
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
-		ring = &adev->gfx.gfx_ring[i];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			return r;
-
-		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v10_0_kgq_init_queue(ring, false);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
+		r = gfx_v10_0_kgq_init_queue(&adev->gfx.gfx_ring[i], false);
 		if (r)
 			return r;
 	}
@@ -7103,55 +7198,24 @@ static int gfx_v10_0_kcq_init_queue(struct amdgpu_ring *ring, bool restore)
 
 static int gfx_v10_0_kiq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring;
-	int r;
-
-	ring = &adev->gfx.kiq[0].ring;
-
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0))
-		return r;
-
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (unlikely(r != 0)) {
-		amdgpu_bo_unreserve(ring->mqd_obj);
-		return r;
-	}
-
-	gfx_v10_0_kiq_init_queue(ring);
-	amdgpu_bo_kunmap(ring->mqd_obj);
-	ring->mqd_ptr = NULL;
-	amdgpu_bo_unreserve(ring->mqd_obj);
+	gfx_v10_0_kiq_init_queue(&adev->gfx.kiq[0].ring);
 	return 0;
 }
 
 static int gfx_v10_0_kcq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = NULL;
-	int r = 0, i;
+	int i, r;
 
 	gfx_v10_0_cp_compute_enable(adev, true);
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		ring = &adev->gfx.compute_ring[i];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			goto done;
-		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v10_0_kcq_init_queue(ring, false);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
+		r = gfx_v10_0_kcq_init_queue(&adev->gfx.compute_ring[i],
+					     false);
 		if (r)
-			goto done;
+			return r;
 	}
 
-	r = amdgpu_gfx_enable_kcq(adev, 0);
-done:
-	return r;
+	return amdgpu_gfx_enable_kcq(adev, 0);
 }
 
 static int gfx_v10_0_cp_resume(struct amdgpu_device *adev)
@@ -7467,6 +7531,8 @@ static int gfx_v10_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_device *adev = ip_block->adev;
 
+	cancel_delayed_work_sync(&adev->gfx.idle_work);
+
 	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.bad_op_irq, 0);
@@ -7511,9 +7577,9 @@ static int gfx_v10_0_resume(struct amdgpu_ip_block *ip_block)
 	return gfx_v10_0_hw_init(ip_block);
 }
 
-static bool gfx_v10_0_is_idle(void *handle)
+static bool gfx_v10_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (REG_GET_FIELD(RREG32_SOC15(GC, 0, mmGRBM_STATUS),
 				GRBM_STATUS, GUI_ACTIVE))
@@ -7602,19 +7668,17 @@ static int gfx_v10_0_soft_reset(struct amdgpu_ip_block *ip_block)
 		/* Disable MEC parsing/prefetching */
 		gfx_v10_0_cp_compute_enable(adev, false);
 
-		if (grbm_soft_reset) {
-			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
-			tmp |= grbm_soft_reset;
-			dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
-			WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
-			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+		tmp |= grbm_soft_reset;
+		dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
 
-			udelay(50);
+		udelay(50);
 
-			tmp &= ~grbm_soft_reset;
-			WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
-			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
-		}
+		tmp &= ~grbm_soft_reset;
+		WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
 
 		/* Wait a little for things to settle down */
 		udelay(50);
@@ -8431,9 +8495,9 @@ static int gfx_v10_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
-static void gfx_v10_0_get_clockgating_state(void *handle, u64 *flags)
+static void gfx_v10_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	/* AMD_CG_SUPPORT_GFX_FGCG */
@@ -8984,21 +9048,6 @@ static void gfx_v10_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
 							   ref, mask);
 }
 
-static void gfx_v10_0_ring_soft_recovery(struct amdgpu_ring *ring,
-					 unsigned int vmid)
-{
-	struct amdgpu_device *adev = ring->adev;
-	uint32_t value = 0;
-
-	value = REG_SET_FIELD(value, SQ_CMD, CMD, 0x03);
-	value = REG_SET_FIELD(value, SQ_CMD, MODE, 0x01);
-	value = REG_SET_FIELD(value, SQ_CMD, CHECK_VMID, 1);
-	value = REG_SET_FIELD(value, SQ_CMD, VM_ID, vmid);
-	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
-	WREG32_SOC15(GC, 0, mmSQ_CMD, value);
-	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
-}
-
 static void
 gfx_v10_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
 				      uint32_t me, uint32_t pipe,
@@ -9460,7 +9509,9 @@ static void gfx_v10_ring_insert_nop(struct amdgpu_ring *ring, uint32_t num_nop)
 	amdgpu_ring_insert_nop(ring, num_nop - 1);
 }
 
-static int gfx_v10_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid)
+static int gfx_v10_0_reset_kgq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
@@ -9470,15 +9521,14 @@ static int gfx_v10_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid)
 	u64 addr;
 	int r;
 
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
-
 	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
 		return -EINVAL;
 
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
 	spin_lock_irqsave(&kiq->ring_lock, flags);
 
-	if (amdgpu_ring_alloc(kiq_ring, 5 + 7 + 7 + kiq->pmf->map_queues_size)) {
+	if (amdgpu_ring_alloc(kiq_ring, 5 + 7 + 7)) {
 		spin_unlock_irqrestore(&kiq->ring_lock, flags);
 		return -ENOMEM;
 	}
@@ -9498,37 +9548,37 @@ static int gfx_v10_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid)
 			       0, 1, 0x20);
 	gfx_v10_0_ring_emit_reg_wait(kiq_ring,
 				     SOC15_REG_OFFSET(GC, 0, mmCP_VMID_RESET), 0, 0xffffffff);
-	kiq->pmf->kiq_map_queues(kiq_ring, ring);
 	amdgpu_ring_commit(kiq_ring);
-
-	spin_unlock_irqrestore(&kiq->ring_lock, flags);
-
 	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
 	if (r)
 		return r;
 
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0)) {
-		DRM_ERROR("fail to resv mqd_obj\n");
+	r = gfx_v10_0_kgq_init_queue(ring, true);
+	if (r) {
+		DRM_ERROR("fail to init kgq\n");
 		return r;
 	}
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (!r) {
-		r = gfx_v10_0_kgq_init_queue(ring, true);
-		amdgpu_bo_kunmap(ring->mqd_obj);
-		ring->mqd_ptr = NULL;
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+
+	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size)) {
+		spin_unlock_irqrestore(&kiq->ring_lock, flags);
+		return -ENOMEM;
 	}
-	amdgpu_bo_unreserve(ring->mqd_obj);
-	if (r) {
-		DRM_ERROR("fail to unresv mqd_obj\n");
+	kiq->pmf->kiq_map_queues(kiq_ring, ring);
+	amdgpu_ring_commit(kiq_ring);
+	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+	if (r)
 		return r;
-	}
 
-	return amdgpu_ring_test_ring(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
 }
 
 static int gfx_v10_0_reset_kcq(struct amdgpu_ring *ring,
-			       unsigned int vmid)
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
@@ -9536,12 +9586,11 @@ static int gfx_v10_0_reset_kcq(struct amdgpu_ring *ring,
 	unsigned long flags;
 	int i, r;
 
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
-
 	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
 		return -EINVAL;
 
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
 	spin_lock_irqsave(&kiq->ring_lock, flags);
 
 	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
@@ -9552,9 +9601,8 @@ static int gfx_v10_0_reset_kcq(struct amdgpu_ring *ring,
 	kiq->pmf->kiq_unmap_queues(kiq_ring, ring, RESET_QUEUES,
 				   0, 0);
 	amdgpu_ring_commit(kiq_ring);
-	spin_unlock_irqrestore(&kiq->ring_lock, flags);
-
 	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
 	if (r)
 		return r;
 
@@ -9577,20 +9625,9 @@ static int gfx_v10_0_reset_kcq(struct amdgpu_ring *ring,
 		return r;
 	}
 
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0)) {
-		dev_err(adev->dev, "fail to resv mqd_obj\n");
-		return r;
-	}
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (!r) {
-		r = gfx_v10_0_kcq_init_queue(ring, true);
-		amdgpu_bo_kunmap(ring->mqd_obj);
-		ring->mqd_ptr = NULL;
-	}
-	amdgpu_bo_unreserve(ring->mqd_obj);
+	r = gfx_v10_0_kcq_init_queue(ring, true);
 	if (r) {
-		dev_err(adev->dev, "fail to unresv mqd_obj\n");
+		dev_err(adev->dev, "fail to init kcq\n");
 		return r;
 	}
 
@@ -9601,13 +9638,12 @@ static int gfx_v10_0_reset_kcq(struct amdgpu_ring *ring,
 	}
 	kiq->pmf->kiq_map_queues(kiq_ring, ring);
 	amdgpu_ring_commit(kiq_ring);
-	spin_unlock_irqrestore(&kiq->ring_lock, flags);
-
 	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
 	if (r)
 		return r;
 
-	return amdgpu_ring_test_ring(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
 }
 
 static void gfx_v10_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
@@ -9639,9 +9675,14 @@ static void gfx_v10_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printe
 			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
 				drm_printf(p, "\nmec %d, pipe %d, queue %d\n", i, j, k);
 				for (reg = 0; reg < reg_count; reg++) {
-					drm_printf(p, "%-50s \t 0x%08x\n",
-						   gc_cp_reg_list_10[reg].reg_name,
-						   adev->gfx.ip_dump_compute_queues[index + reg]);
+					if (i && gc_cp_reg_list_10[reg].reg_offset == mmCP_MEC_ME1_HEADER_DUMP)
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   "mmCP_MEC_ME2_HEADER_DUMP",
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
+					else
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   gc_cp_reg_list_10[reg].reg_name,
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
 				}
 				index += reg_count;
 			}
@@ -9702,9 +9743,13 @@ static void gfx_v10_ip_dump(struct amdgpu_ip_block *ip_block)
 				nv_grbm_select(adev, adev->gfx.me.num_me + i, j, k, 0);
 
 				for (reg = 0; reg < reg_count; reg++) {
-					adev->gfx.ip_dump_compute_queues[index + reg] =
-						RREG32(SOC15_REG_ENTRY_OFFSET(
-							gc_cp_reg_list_10[reg]));
+					if (i && gc_cp_reg_list_10[reg].reg_offset == mmCP_MEC_ME1_HEADER_DUMP)
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME2_HEADER_DUMP));
+					else
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_ENTRY_OFFSET(
+								       gc_cp_reg_list_10[reg]));
 				}
 				index += reg_count;
 			}
@@ -9748,6 +9793,20 @@ static void gfx_v10_0_ring_emit_cleaner_shader(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, 0);  /* RESERVED field, programmed to zero */
 }
 
+static void gfx_v10_0_ring_begin_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_begin_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_begin_use(ring);
+}
+
+static void gfx_v10_0_ring_end_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_end_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_end_use(ring);
+}
+
 static const struct amd_ip_funcs gfx_v10_0_ip_funcs = {
 	.name = "gfx_v10_0",
 	.early_init = gfx_v10_0_early_init,
@@ -9819,12 +9878,11 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_gfx = {
 	.emit_wreg = gfx_v10_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v10_0_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait,
-	.soft_recovery = gfx_v10_0_ring_soft_recovery,
 	.emit_mem_sync = gfx_v10_0_emit_mem_sync,
 	.reset = gfx_v10_0_reset_kgq,
 	.emit_cleaner_shader = gfx_v10_0_ring_emit_cleaner_shader,
-	.begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use,
-	.end_use = amdgpu_gfx_enforce_isolation_ring_end_use,
+	.begin_use = gfx_v10_0_ring_begin_use,
+	.end_use = gfx_v10_0_ring_end_use,
 };
 
 static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_compute = {
@@ -9860,12 +9918,11 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_compute = {
 	.emit_wreg = gfx_v10_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v10_0_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait,
-	.soft_recovery = gfx_v10_0_ring_soft_recovery,
 	.emit_mem_sync = gfx_v10_0_emit_mem_sync,
 	.reset = gfx_v10_0_reset_kcq,
 	.emit_cleaner_shader = gfx_v10_0_ring_emit_cleaner_shader,
-	.begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use,
-	.end_use = amdgpu_gfx_enforce_isolation_ring_end_use,
+	.begin_use = gfx_v10_0_ring_begin_use,
+	.end_use = gfx_v10_0_ring_end_use,
 };
 
 static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_kiq = {
@@ -9895,6 +9952,7 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_kiq = {
 	.emit_wreg = gfx_v10_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v10_0_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait,
+	.emit_hdp_flush = gfx_v10_0_ring_emit_hdp_flush,
 };
 
 static void gfx_v10_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h
index 663c2572d440..f67569ccf9f6 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h
@@ -21,6 +21,41 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
+/* Define the cleaner shader gfx_10_1_10 */
+static const u32 gfx_10_1_10_cleaner_shader_hex[] = {
+	0xb0804004, 0xbf8a0000,
+	0xbf068100, 0xbf840023,
+	0xbe8203b8, 0xbefc0380,
+	0x7e008480, 0x7e028480,
+	0x7e048480, 0x7e068480,
+	0x7e088480, 0x7e0a8480,
+	0x7e0c8480, 0x7e0e8480,
+	0xbefc0302, 0x80828802,
+	0xbf84fff5, 0xbe8203ff,
+	0x80000000, 0x87020102,
+	0xbf840012, 0xbefe03c1,
+	0xbeff03c1, 0xd7650001,
+	0x0001007f, 0xd7660001,
+	0x0002027e, 0x16020288,
+	0xbe8203bf, 0xbefc03c1,
+	0xd9382000, 0x00020201,
+	0xd9386040, 0x00040401,
+	0xd70f6a01, 0x000202ff,
+	0x00000400, 0x80828102,
+	0xbf84fff7, 0xbefc03ff,
+	0x00000068, 0xbe803000,
+	0xbe813000, 0xbe823000,
+	0xbe833000, 0x80fc847c,
+	0xbf84fffa, 0xbeea0480,
+	0xbeec0480, 0xbeee0480,
+	0xbef00480, 0xbef20480,
+	0xbef40480, 0xbef60480,
+	0xbef80480, 0xbefa0480,
+	0xbf810000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+	0xbf9f0000, 0xbf9f0000,
+};
+
 /* Define the cleaner shader gfx_10_3_0 */
 static const u32 gfx_10_3_0_cleaner_shader_hex[] = {
 	0xb0804004, 0xbf8a0000,
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm b/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm
new file mode 100644
index 000000000000..54f7ed9e2801
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm
@@ -0,0 +1,125 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+// This shader is to clean LDS, SGPRs and VGPRs. It is  first 64 Dwords or 256 bytes of 256 Dwords cleaner shader.
+
+// GFX10.1 : Clear SGPRs, VGPRs and LDS
+//   Launch 32 waves per CU (16 per SIMD) as a workgroup (threadgroup) to fill every wave slot
+//   Waves are "wave32" and have 64 VGPRs each, which uses all 1024 VGPRs per SIMD
+//   Waves are launched in "CU" mode, and the workgroup shares 64KB of LDS (half of the WGP's LDS)
+//      It takes 2 workgroups to use all of LDS: one on each CU of the WGP
+//   Each wave clears SGPRs 0 - 107
+//   Each wave clears VGPRs 0 - 63
+//   The first wave of the workgroup clears its 64KB of LDS
+//   The shader starts with "S_BARRIER" to ensure SPI has launched all waves of the workgroup
+//       before any wave in the workgroup could end.  Without this, it is possible not all SGPRs get cleared.
+
+
+shader main
+  asic(GFX10.1)
+  type(CS)
+  wave_size(32)
+// Note: original source code from SQ team
+//
+// Create 32 waves in a threadgroup (CS waves)
+// Each allocates 64 VGPRs
+// The workgroup allocates all of LDS (64kbytes)
+//
+// Takes about 2500 clocks to run.
+//   (theorhetical fastest = 1024clks vgpr + 640lds = 1660 clks)
+//
+  S_BARRIER
+  s_cmp_eq_u32 s0, 1                                // Bit0 is set, sgpr0 is set then clear VGPRS and LDS as FW set COMPUTE_USER_DATA_0
+  s_cbranch_scc0  label_0023                        // Clean VGPRs and LDS if sgpr0 of wave is set, scc = (s0 == 1)
+
+  s_mov_b32     s2, 0x00000038  // Loop 64/8=8 times  (loop unrolled for performance)
+  s_mov_b32     m0, 0
+  //
+  // CLEAR VGPRs
+  //
+label_0005:
+  v_movreld_b32     v0, 0
+  v_movreld_b32     v1, 0
+  v_movreld_b32     v2, 0
+  v_movreld_b32     v3, 0
+  v_movreld_b32     v4, 0
+  v_movreld_b32     v5, 0
+  v_movreld_b32     v6, 0
+  v_movreld_b32     v7, 0
+  s_mov_b32         m0, s2
+  s_sub_u32     s2, s2, 8
+  s_cbranch_scc0  label_0005
+  //
+  s_mov_b32     s2, 0x80000000                       // Bit31 is first_wave
+  s_and_b32     s2, s2, s1                           // sgpr0 has tg_size (first_wave) term as in ucode only COMPUTE_PGM_RSRC2.tg_size_en is set
+  s_cbranch_scc0  label_0023                         // Clean LDS if its first wave of ThreadGroup/WorkGroup
+  // CLEAR LDS
+  //
+  s_mov_b32 exec_lo, 0xffffffff
+  s_mov_b32 exec_hi, 0xffffffff
+  v_mbcnt_lo_u32_b32  v1, exec_hi, 0          // Set V1 to thread-ID (0..63)
+  v_mbcnt_hi_u32_b32  v1, exec_lo, v1        // Set V1 to thread-ID (0..63)
+  v_mul_u32_u24  v1, 0x00000008, v1          // * 8, so each thread is a double-dword address (8byte)
+  s_mov_b32     s2, 0x00000003f                    // 64 loop iterations
+  s_mov_b32     m0, 0xffffffff
+  // Clear all of LDS space
+  // Each FirstWave of WorkGroup clears 64kbyte block
+
+label_001F:
+  ds_write2_b64  v1, v[2:3], v[2:3] offset1:32
+  ds_write2_b64  v1, v[4:5], v[4:5] offset0:64 offset1:96
+  v_add_co_u32     v1, vcc, 0x00000400, v1
+  s_sub_u32     s2, s2, 1
+  s_cbranch_scc0  label_001F
+
+  //
+  // CLEAR SGPRs
+  //
+label_0023:
+  s_mov_b32     m0, 0x00000068  // Loop 108/4=27 times  (loop unrolled for performance)
+label_sgpr_loop:
+  s_movreld_b32     s0, s0
+  s_movreld_b32     s1, s0
+  s_movreld_b32     s2, s0
+  s_movreld_b32     s3, s0
+  s_sub_u32         m0, m0, 4
+  s_cbranch_scc0  label_sgpr_loop
+
+  //clear vcc
+  s_mov_b64 vcc, 0          //clear vcc
+  //s_setreg_imm32_b32 hw_reg_shader_flat_scratch_lo, 0   //clear  flat scratch lo SGPR
+  //s_setreg_imm32_b32 hw_reg_shader_flat_scratch_hi, 0    //clear  flat scratch hi SGPR
+  s_mov_b64 ttmp0, 0        //Clear ttmp0 and ttmp1
+  s_mov_b64 ttmp2, 0        //Clear ttmp2 and ttmp3
+  s_mov_b64 ttmp4, 0        //Clear ttmp4 and ttmp5
+  s_mov_b64 ttmp6, 0        //Clear ttmp6 and ttmp7
+  s_mov_b64 ttmp8, 0        //Clear ttmp8 and ttmp9
+  s_mov_b64 ttmp10, 0       //Clear ttmp10 and ttmp11
+  s_mov_b64 ttmp12, 0       //Clear ttmp12 and ttmp13
+  s_mov_b64 ttmp14, 0       //Clear ttmp14 and ttmp15
+
+ s_endpgm
+
+end
+
+
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
index 56c06b72a70a..8a2ee2de390f 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
@@ -29,7 +29,6 @@
 #include "amdgpu_gfx.h"
 #include "amdgpu_psp.h"
 #include "amdgpu_smu.h"
-#include "amdgpu_atomfirmware.h"
 #include "imu_v11_0.h"
 #include "soc21.h"
 #include "nvd.h"
@@ -42,7 +41,6 @@
 #include "ivsrcid/gfx/irqsrcs_gfx_11_0_0.h"
 
 #include "soc15.h"
-#include "soc15d.h"
 #include "clearstate_gfx11.h"
 #include "v11_structs.h"
 #include "gfx_v11_0.h"
@@ -50,6 +48,8 @@
 #include "gfx_v11_0_3.h"
 #include "nbio_v4_3.h"
 #include "mes_v11_0.h"
+#include "mes_userqueue.h"
+#include "amdgpu_userq_fence.h"
 
 #define GFX11_NUM_GFX_RINGS		1
 #define GFX11_MEC_HPD_SIZE	2048
@@ -64,10 +64,28 @@
 #define regPC_CONFIG_CNTL_1		0x194d
 #define regPC_CONFIG_CNTL_1_BASE_IDX	1
 
+#define regCP_GFX_MQD_CONTROL_DEFAULT                                             0x00000100
+#define regCP_GFX_HQD_VMID_DEFAULT                                                0x00000000
+#define regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT                                      0x00000000
+#define regCP_GFX_HQD_QUANTUM_DEFAULT                                             0x00000a01
+#define regCP_GFX_HQD_CNTL_DEFAULT                                                0x00a00000
+#define regCP_RB_DOORBELL_CONTROL_DEFAULT                                         0x00000000
+#define regCP_GFX_HQD_RPTR_DEFAULT                                                0x00000000
+
+#define regCP_HQD_EOP_CONTROL_DEFAULT                                             0x00000006
+#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT                                     0x00000000
+#define regCP_MQD_CONTROL_DEFAULT                                                 0x00000100
+#define regCP_HQD_PQ_CONTROL_DEFAULT                                              0x00308509
+#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT                                     0x00000000
+#define regCP_HQD_PQ_RPTR_DEFAULT                                                 0x00000000
+#define regCP_HQD_PERSISTENT_STATE_DEFAULT                                        0x0be05501
+#define regCP_HQD_IB_CONTROL_DEFAULT                                              0x00300000
+
 MODULE_FIRMWARE("amdgpu/gc_11_0_0_pfp.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_0_me.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_0_mec.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_0_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_rlc_kicker.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_0_rlc_1.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_0_toc.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_1_pfp.bin");
@@ -98,6 +116,10 @@ MODULE_FIRMWARE("amdgpu/gc_11_5_2_pfp.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_5_2_me.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_5_2_mec.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_5_2_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_pfp.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_me.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_mec.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_rlc.bin");
 
 static const struct amdgpu_hwip_reg_entry gc_reg_list_11_0[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS),
@@ -158,9 +180,13 @@ static const struct amdgpu_hwip_reg_entry gc_reg_list_11_0[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_INSTR_PNTR),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_CPC_STATUS),
 	/* cp header registers */
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
 	/* SE status registers */
 	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE0),
@@ -211,7 +237,16 @@ static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_11[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_CNTL_STACK_OFFSET),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_CNTL_STACK_DW_CNT),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_WG_STATE_OFFSET),
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_STATUS)
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_STATUS),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
 };
 
 static const struct amdgpu_hwip_reg_entry gc_gfx_queue_reg_list_11[] = {
@@ -240,7 +275,24 @@ static const struct amdgpu_hwip_reg_entry gc_gfx_queue_reg_list_11[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_LO),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_HI),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_CMD_BUFSZ),
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BUFSZ)
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BUFSZ),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
 };
 
 static const struct soc15_reg_golden golden_settings_gc_11_0[] = {
@@ -551,7 +603,7 @@ static int gfx_v11_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	struct dma_fence *f = NULL;
 	unsigned index;
 	uint64_t gpu_addr;
-	volatile uint32_t *cpu_ptr;
+	uint32_t *cpu_ptr;
 	long r;
 
 	/* MES KIQ fw hasn't indirect buffer support for now */
@@ -561,33 +613,18 @@ static int gfx_v11_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 
 	memset(&ib, 0, sizeof(ib));
 
-	if (ring->is_mes_queue) {
-		uint32_t padding, offset;
-
-		offset = amdgpu_mes_ctx_get_offs(ring, AMDGPU_MES_CTX_IB_OFFS);
-		padding = amdgpu_mes_ctx_get_offs(ring,
-						  AMDGPU_MES_CTX_PADDING_OFFS);
-
-		ib.gpu_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
-		ib.ptr = amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset);
-
-		gpu_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, padding);
-		cpu_ptr = amdgpu_mes_ctx_get_offs_cpu_addr(ring, padding);
-		*cpu_ptr = cpu_to_le32(0xCAFEDEAD);
-	} else {
-		r = amdgpu_device_wb_get(adev, &index);
-		if (r)
-			return r;
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
+		return r;
 
-		gpu_addr = adev->wb.gpu_addr + (index * 4);
-		adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
-		cpu_ptr = &adev->wb.wb[index];
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
+	cpu_ptr = &adev->wb.wb[index];
 
-		r = amdgpu_ib_get(adev, NULL, 20, AMDGPU_IB_POOL_DIRECT, &ib);
-		if (r) {
-			DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
-			goto err1;
-		}
+	r = amdgpu_ib_get(adev, NULL, 20, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err1;
 	}
 
 	ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3);
@@ -614,12 +651,10 @@ static int gfx_v11_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	else
 		r = -EINVAL;
 err2:
-	if (!ring->is_mes_queue)
-		amdgpu_ib_free(&ib, NULL);
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
 err1:
-	if (!ring->is_mes_queue)
-		amdgpu_device_wb_free(adev, index);
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
@@ -725,6 +760,10 @@ static int gfx_v11_0_init_microcode(struct amdgpu_device *adev)
 			err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
 						   AMDGPU_UCODE_REQUIRED,
 						   "amdgpu/gc_11_0_0_rlc_1.bin");
+		else if (amdgpu_is_kicker_fw(adev))
+			err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_rlc_kicker.bin", ucode_prefix);
 		else
 			err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
 						   AMDGPU_UCODE_REQUIRED,
@@ -811,12 +850,9 @@ static u32 gfx_v11_0_get_csb_size(struct amdgpu_device *adev)
 	return count;
 }
 
-static void gfx_v11_0_get_csb_buffer(struct amdgpu_device *adev,
-				    volatile u32 *buffer)
+static void gfx_v11_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
-	u32 count = 0, i;
-	const struct cs_section_def *sect = NULL;
-	const struct cs_extent_def *ext = NULL;
+	u32 count = 0;
 	int ctx_reg_offset;
 
 	if (adev->gfx.rlc.cs_data == NULL)
@@ -824,39 +860,15 @@ static void gfx_v11_0_get_csb_buffer(struct amdgpu_device *adev,
 	if (buffer == NULL)
 		return;
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
-	buffer[count++] = cpu_to_le32(0x80000000);
-	buffer[count++] = cpu_to_le32(0x80000000);
-
-	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
-		for (ext = sect->section; ext->extent != NULL; ++ext) {
-			if (sect->id == SECT_CONTEXT) {
-				buffer[count++] =
-					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
-				buffer[count++] = cpu_to_le32(ext->reg_index -
-						PACKET3_SET_CONTEXT_REG_START);
-				for (i = 0; i < ext->reg_count; i++)
-					buffer[count++] = cpu_to_le32(ext->extent[i]);
-			} else {
-				return;
-			}
-		}
-	}
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
 
-	ctx_reg_offset =
-		SOC15_REG_OFFSET(GC, 0, regPA_SC_TILE_STEERING_OVERRIDE) - PACKET3_SET_CONTEXT_REG_START;
+	ctx_reg_offset = SOC15_REG_OFFSET(GC, 0, regPA_SC_TILE_STEERING_OVERRIDE) - PACKET3_SET_CONTEXT_REG_START;
 	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 1));
 	buffer[count++] = cpu_to_le32(ctx_reg_offset);
 	buffer[count++] = cpu_to_le32(adev->gfx.config.pa_sc_tile_steering_override);
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
-	buffer[count++] = cpu_to_le32(0);
+	amdgpu_gfx_csb_preamble_end(buffer, count);
 }
 
 static void gfx_v11_0_rlc_fini(struct amdgpu_device *adev)
@@ -1037,14 +1049,21 @@ static void gfx_v11_0_select_me_pipe_q(struct amdgpu_device *adev,
 #define MQD_FWWORKAREA_SIZE       484
 #define MQD_FWWORKAREA_ALIGNMENT  256
 
-static int gfx_v11_0_get_gfx_shadow_info(struct amdgpu_device *adev,
+static void gfx_v11_0_get_gfx_shadow_info_nocheck(struct amdgpu_device *adev,
 					 struct amdgpu_gfx_shadow_info *shadow_info)
 {
-	if (adev->gfx.cp_gfx_shadow) {
-		shadow_info->shadow_size = MQD_SHADOW_BASE_SIZE;
-		shadow_info->shadow_alignment = MQD_SHADOW_BASE_ALIGNMENT;
-		shadow_info->csa_size = MQD_FWWORKAREA_SIZE;
-		shadow_info->csa_alignment = MQD_FWWORKAREA_ALIGNMENT;
+	shadow_info->shadow_size = MQD_SHADOW_BASE_SIZE;
+	shadow_info->shadow_alignment = MQD_SHADOW_BASE_ALIGNMENT;
+	shadow_info->csa_size = MQD_FWWORKAREA_SIZE;
+	shadow_info->csa_alignment = MQD_FWWORKAREA_ALIGNMENT;
+}
+
+static int gfx_v11_0_get_gfx_shadow_info(struct amdgpu_device *adev,
+					 struct amdgpu_gfx_shadow_info *shadow_info,
+					 bool skip_check)
+{
+	if (adev->gfx.cp_gfx_shadow || skip_check) {
+		gfx_v11_0_get_gfx_shadow_info_nocheck(adev, shadow_info);
 		return 0;
 	} else {
 		memset(shadow_info, 0, sizeof(struct amdgpu_gfx_shadow_info));
@@ -1087,6 +1106,7 @@ static int gfx_v11_0_gpu_early_init(struct amdgpu_device *adev)
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		adev->gfx.config.max_hw_contexts = 8;
 		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
 		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
@@ -1116,6 +1136,10 @@ static int gfx_v11_0_gfx_ring_init(struct amdgpu_device *adev, int ring_id,
 
 	ring->ring_obj = NULL;
 	ring->use_doorbell = true;
+	if (adev->gfx.disable_kq) {
+		ring->no_scheduler = true;
+		ring->no_user_submission = true;
+	}
 
 	if (!ring_id)
 		ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
@@ -1548,29 +1572,26 @@ static void gfx_v11_0_alloc_ip_dump(struct amdgpu_device *adev)
 
 static int gfx_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
-	int i, j, k, r, ring_id = 0;
+	int i, j, k, r, ring_id;
 	int xcc_id = 0;
 	struct amdgpu_device *adev = ip_block->adev;
+	int num_queue_per_pipe = 1; /* we only enable 1 KGQ per pipe */
+
+	INIT_DELAYED_WORK(&adev->gfx.idle_work, amdgpu_gfx_profile_idle_work_handler);
 
 	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
 	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 1):
 	case IP_VERSION(11, 0, 2):
 	case IP_VERSION(11, 0, 3):
-		adev->gfx.me.num_me = 1;
-		adev->gfx.me.num_pipe_per_me = 1;
-		adev->gfx.me.num_queue_per_pipe = 1;
-		adev->gfx.mec.num_mec = 2;
-		adev->gfx.mec.num_pipe_per_mec = 4;
-		adev->gfx.mec.num_queue_per_pipe = 4;
-		break;
-	case IP_VERSION(11, 0, 1):
 	case IP_VERSION(11, 0, 4):
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		adev->gfx.me.num_me = 1;
 		adev->gfx.me.num_pipe_per_me = 1;
-		adev->gfx.me.num_queue_per_pipe = 1;
+		adev->gfx.me.num_queue_per_pipe = 2;
 		adev->gfx.mec.num_mec = 1;
 		adev->gfx.mec.num_pipe_per_mec = 4;
 		adev->gfx.mec.num_queue_per_pipe = 4;
@@ -1589,6 +1610,35 @@ static int gfx_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 	case IP_VERSION(11, 0, 0):
 	case IP_VERSION(11, 0, 2):
 	case IP_VERSION(11, 0, 3):
+		if (!adev->gfx.disable_uq &&
+		    adev->gfx.me_fw_version  >= 2420 &&
+		    adev->gfx.pfp_fw_version >= 2580 &&
+		    adev->gfx.mec_fw_version >= 2650 &&
+		    adev->mes.fw_version[0] >= 120) {
+			adev->userq_funcs[AMDGPU_HW_IP_GFX] = &userq_mes_funcs;
+			adev->userq_funcs[AMDGPU_HW_IP_COMPUTE] = &userq_mes_funcs;
+		}
+		break;
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 4):
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
+		/* add firmware version checks here */
+		if (0 && !adev->gfx.disable_uq) {
+			adev->userq_funcs[AMDGPU_HW_IP_GFX] = &userq_mes_funcs;
+			adev->userq_funcs[AMDGPU_HW_IP_COMPUTE] = &userq_mes_funcs;
+		}
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
 		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
 		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
 		if (adev->gfx.me_fw_version  >= 2280 &&
@@ -1603,6 +1653,63 @@ static int gfx_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 			}
 		}
 		break;
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 4):
+		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+		if (adev->gfx.pfp_fw_version >= 102 &&
+		    adev->gfx.mec_fw_version >= 66 &&
+		    adev->mes.fw_version[0] >= 128) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(11, 5, 0):
+	case IP_VERSION(11, 5, 1):
+		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+		if (adev->gfx.mec_fw_version >= 26 &&
+		    adev->mes.fw_version[0] >= 114) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(11, 5, 2):
+		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version  >= 12 &&
+		    adev->gfx.pfp_fw_version >= 15 &&
+		    adev->gfx.mec_fw_version >= 15) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
+	case IP_VERSION(11, 5, 3):
+		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version  >= 7 &&
+		    adev->gfx.pfp_fw_version >= 8 &&
+		    adev->gfx.mec_fw_version >= 8) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
 	default:
 		adev->gfx.enable_cleaner_shader = false;
 		break;
@@ -1664,37 +1771,42 @@ static int gfx_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 		return r;
 	}
 
-	/* set up the gfx ring */
-	for (i = 0; i < adev->gfx.me.num_me; i++) {
-		for (j = 0; j < adev->gfx.me.num_queue_per_pipe; j++) {
-			for (k = 0; k < adev->gfx.me.num_pipe_per_me; k++) {
-				if (!amdgpu_gfx_is_me_queue_enabled(adev, i, k, j))
-					continue;
-
-				r = gfx_v11_0_gfx_ring_init(adev, ring_id,
-							    i, k, j);
-				if (r)
-					return r;
-				ring_id++;
+	if (adev->gfx.num_gfx_rings) {
+		ring_id = 0;
+		/* set up the gfx ring */
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < num_queue_per_pipe; j++) {
+				for (k = 0; k < adev->gfx.me.num_pipe_per_me; k++) {
+					if (!amdgpu_gfx_is_me_queue_enabled(adev, i, k, j))
+						continue;
+
+					r = gfx_v11_0_gfx_ring_init(adev, ring_id,
+								    i, k, j);
+					if (r)
+						return r;
+					ring_id++;
+				}
 			}
 		}
 	}
 
-	ring_id = 0;
-	/* set up the compute queues - allocate horizontally across pipes */
-	for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
-		for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
-			for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
-				if (!amdgpu_gfx_is_mec_queue_enabled(adev, 0, i,
-								     k, j))
-					continue;
+	if (adev->gfx.num_compute_rings) {
+		ring_id = 0;
+		/* set up the compute queues - allocate horizontally across pipes */
+		for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
+			for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
+				for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+					if (!amdgpu_gfx_is_mec_queue_enabled(adev, 0, i,
+									     k, j))
+						continue;
 
-				r = gfx_v11_0_compute_ring_init(adev, ring_id,
-								i, k, j);
-				if (r)
-					return r;
+					r = gfx_v11_0_compute_ring_init(adev, ring_id,
+									i, k, j);
+					if (r)
+						return r;
 
-				ring_id++;
+					ring_id++;
+				}
 			}
 		}
 	}
@@ -1708,12 +1820,19 @@ static int gfx_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 	case IP_VERSION(11, 0, 2):
 	case IP_VERSION(11, 0, 3):
 		if ((adev->gfx.me_fw_version >= 2280) &&
-			    (adev->gfx.mec_fw_version >= 2410)) {
-				adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
-				adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		    (adev->gfx.mec_fw_version >= 2410) &&
+		    !amdgpu_sriov_vf(adev) &&
+		    !adev->debug_disable_gpu_ring_reset) {
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+			adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
 		}
 		break;
 	default:
+		if (!amdgpu_sriov_vf(adev) &&
+		    !adev->debug_disable_gpu_ring_reset) {
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+			adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		}
 		break;
 	}
 
@@ -2321,7 +2440,7 @@ static int gfx_v11_0_rlc_load_microcode(struct amdgpu_device *adev)
 			if (version_minor == 3)
 				gfx_v11_0_load_rlcp_rlcv_microcode(adev);
 		}
-		
+
 		return 0;
 	}
 
@@ -2391,7 +2510,7 @@ static int gfx_v11_0_config_me_cache(struct amdgpu_device *adev, uint64_t addr)
 	}
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_BASE_CNTL);
 	tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, VMID, 0);
@@ -2435,7 +2554,7 @@ static int gfx_v11_0_config_pfp_cache(struct amdgpu_device *adev, uint64_t addr)
 	}
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_CNTL);
 	tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, VMID, 0);
@@ -2480,7 +2599,7 @@ static int gfx_v11_0_config_mec_cache(struct amdgpu_device *adev, uint64_t addr)
 	}
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_CNTL);
 	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0);
@@ -2926,7 +3045,8 @@ static int gfx_v11_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev)
 			    IP_VERSION(11, 0, 4) ||
 		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 5, 0) ||
 		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 5, 1) ||
-		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 5, 2))
+		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 5, 2) ||
+		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 5, 3))
 			bootload_status = RREG32_SOC15(GC, 0,
 					regRLC_RLCS_BOOTLOAD_STATUS_gc_11_0_1);
 		else
@@ -3115,7 +3235,7 @@ static int gfx_v11_0_cp_gfx_load_pfp_microcode_rs64(struct amdgpu_device *adev)
 	amdgpu_bo_unreserve(adev->gfx.pfp.pfp_fw_data_obj);
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_LO,
 		lower_32_bits(adev->gfx.pfp.pfp_fw_gpu_addr));
@@ -3333,7 +3453,7 @@ static int gfx_v11_0_cp_gfx_load_me_microcode_rs64(struct amdgpu_device *adev)
 	amdgpu_bo_unreserve(adev->gfx.me.me_fw_data_obj);
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_LO,
 		lower_32_bits(adev->gfx.me.me_fw_gpu_addr));
@@ -3768,7 +3888,7 @@ static int gfx_v11_0_cp_compute_load_microcode(struct amdgpu_device *adev)
 	}
 
 	memcpy(fw, fw_data, fw_size);
-	
+
 	amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_obj);
 	amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
 
@@ -3958,7 +4078,7 @@ static void gfx_v11_0_gfx_mqd_set_priority(struct amdgpu_device *adev,
 	if (prop->hqd_pipe_priority == AMDGPU_GFX_PIPE_PRIO_HIGH)
 		priority = 1;
 
-	tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_QUEUE_PRIORITY);
+	tmp = regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUEUE_PRIORITY, PRIORITY_LEVEL, priority);
 	mqd->cp_gfx_hqd_queue_priority = tmp;
 }
@@ -3980,14 +4100,14 @@ static int gfx_v11_0_gfx_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr);
 
 	/* set up mqd control */
-	tmp = RREG32_SOC15(GC, 0, regCP_GFX_MQD_CONTROL);
+	tmp = regCP_GFX_MQD_CONTROL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, VMID, 0);
 	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, PRIV_STATE, 1);
 	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, CACHE_POLICY, 0);
 	mqd->cp_gfx_mqd_control = tmp;
 
 	/* set up gfx_hqd_vimd with 0x0 to indicate the ring buffer's vmid */
-	tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_VMID);
+	tmp = regCP_GFX_HQD_VMID_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_VMID, VMID, 0);
 	mqd->cp_gfx_hqd_vmid = 0;
 
@@ -3995,7 +4115,7 @@ static int gfx_v11_0_gfx_mqd_init(struct amdgpu_device *adev, void *m,
 	gfx_v11_0_gfx_mqd_set_priority(adev, mqd, prop);
 
 	/* set up time quantum */
-	tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_QUANTUM);
+	tmp = regCP_GFX_HQD_QUANTUM_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUANTUM, QUANTUM_EN, 1);
 	mqd->cp_gfx_hqd_quantum = tmp;
 
@@ -4017,16 +4137,20 @@ static int gfx_v11_0_gfx_mqd_init(struct amdgpu_device *adev, void *m,
 
 	/* set up the gfx_hqd_control, similar as CP_RB0_CNTL */
 	rb_bufsz = order_base_2(prop->queue_size / 4) - 1;
-	tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_CNTL);
+	tmp = regCP_GFX_HQD_CNTL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BUFSZ, rb_bufsz);
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BLKSZ, rb_bufsz - 2);
 #ifdef __BIG_ENDIAN
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, BUF_SWAP, 1);
 #endif
+	if (prop->tmz_queue)
+		tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, TMZ_MATCH, 1);
+	if (!prop->kernel_queue)
+		tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_NON_PRIV, 1);
 	mqd->cp_gfx_hqd_cntl = tmp;
 
 	/* set up cp_doorbell_control */
-	tmp = RREG32_SOC15(GC, 0, regCP_RB_DOORBELL_CONTROL);
+	tmp = regCP_RB_DOORBELL_CONTROL_DEFAULT;
 	if (prop->use_doorbell) {
 		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
 				    DOORBELL_OFFSET, prop->doorbell_index);
@@ -4038,11 +4162,21 @@ static int gfx_v11_0_gfx_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_rb_doorbell_control = tmp;
 
 	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
-	mqd->cp_gfx_hqd_rptr = RREG32_SOC15(GC, 0, regCP_GFX_HQD_RPTR);
+	mqd->cp_gfx_hqd_rptr = regCP_GFX_HQD_RPTR_DEFAULT;
 
 	/* active the queue */
 	mqd->cp_gfx_hqd_active = 1;
 
+	/* set gfx UQ items */
+	mqd->shadow_base_lo = lower_32_bits(prop->shadow_addr);
+	mqd->shadow_base_hi = upper_32_bits(prop->shadow_addr);
+	mqd->gds_bkup_base_lo = lower_32_bits(prop->gds_bkup_addr);
+	mqd->gds_bkup_base_hi = upper_32_bits(prop->gds_bkup_addr);
+	mqd->fw_work_area_base_lo = lower_32_bits(prop->csa_addr);
+	mqd->fw_work_area_base_hi = upper_32_bits(prop->csa_addr);
+	mqd->fence_address_lo = lower_32_bits(prop->fence_address);
+	mqd->fence_address_hi = upper_32_bits(prop->fence_address);
+
 	return 0;
 }
 
@@ -4077,22 +4211,9 @@ static int gfx_v11_0_kgq_init_queue(struct amdgpu_ring *ring, bool reset)
 static int gfx_v11_0_cp_async_gfx_ring_resume(struct amdgpu_device *adev)
 {
 	int r, i;
-	struct amdgpu_ring *ring;
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
-		ring = &adev->gfx.gfx_ring[i];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			return r;
-
-		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v11_0_kgq_init_queue(ring, false);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
+		r = gfx_v11_0_kgq_init_queue(&adev->gfx.gfx_ring[i], false);
 		if (r)
 			return r;
 	}
@@ -4124,14 +4245,14 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
 
 	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
-	tmp = RREG32_SOC15(GC, 0, regCP_HQD_EOP_CONTROL);
+	tmp = regCP_HQD_EOP_CONTROL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
 			(order_base_2(GFX11_MEC_HPD_SIZE / 4) - 1));
 
 	mqd->cp_hqd_eop_control = tmp;
 
 	/* enable doorbell? */
-	tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL);
+	tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT;
 
 	if (prop->use_doorbell) {
 		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
@@ -4160,7 +4281,7 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr);
 
 	/* set MQD vmid to 0 */
-	tmp = RREG32_SOC15(GC, 0, regCP_MQD_CONTROL);
+	tmp = regCP_MQD_CONTROL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
 	mqd->cp_mqd_control = tmp;
 
@@ -4170,7 +4291,7 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
 
 	/* set up the HQD, this is similar to CP_RB0_CNTL */
-	tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_CONTROL);
+	tmp = regCP_HQD_PQ_CONTROL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
 			    (order_base_2(prop->queue_size / 4) - 1));
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
@@ -4178,8 +4299,12 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 1);
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TUNNEL_DISPATCH,
 			    prop->allow_tunneling);
-	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
-	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	if (prop->kernel_queue) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	}
+	if (prop->tmz_queue)
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TMZ, 1);
 	mqd->cp_hqd_pq_control = tmp;
 
 	/* set the wb address whether it's enabled or not */
@@ -4196,7 +4321,7 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	tmp = 0;
 	/* enable the doorbell if requested */
 	if (prop->use_doorbell) {
-		tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL);
+		tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT;
 		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
 				DOORBELL_OFFSET, prop->doorbell_index);
 
@@ -4211,17 +4336,17 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_hqd_pq_doorbell_control = tmp;
 
 	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
-	mqd->cp_hqd_pq_rptr = RREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR);
+	mqd->cp_hqd_pq_rptr = regCP_HQD_PQ_RPTR_DEFAULT;
 
 	/* set the vmid for the queue */
 	mqd->cp_hqd_vmid = 0;
 
-	tmp = RREG32_SOC15(GC, 0, regCP_HQD_PERSISTENT_STATE);
+	tmp = regCP_HQD_PERSISTENT_STATE_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x55);
 	mqd->cp_hqd_persistent_state = tmp;
 
 	/* set MIN_IB_AVAIL_SIZE */
-	tmp = RREG32_SOC15(GC, 0, regCP_HQD_IB_CONTROL);
+	tmp = regCP_HQD_IB_CONTROL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3);
 	mqd->cp_hqd_ib_control = tmp;
 
@@ -4231,6 +4356,10 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 
 	mqd->cp_hqd_active = prop->hqd_active;
 
+	/* set UQ fenceaddress */
+	mqd->fence_address_lo = lower_32_bits(prop->fence_address);
+	mqd->fence_address_hi = upper_32_bits(prop->fence_address);
+
 	return 0;
 }
 
@@ -4414,57 +4543,24 @@ static int gfx_v11_0_kcq_init_queue(struct amdgpu_ring *ring, bool reset)
 
 static int gfx_v11_0_kiq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring;
-	int r;
-
-	ring = &adev->gfx.kiq[0].ring;
-
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0))
-		return r;
-
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (unlikely(r != 0)) {
-		amdgpu_bo_unreserve(ring->mqd_obj);
-		return r;
-	}
-
-	gfx_v11_0_kiq_init_queue(ring);
-	amdgpu_bo_kunmap(ring->mqd_obj);
-	ring->mqd_ptr = NULL;
-	amdgpu_bo_unreserve(ring->mqd_obj);
-	ring->sched.ready = true;
+	gfx_v11_0_kiq_init_queue(&adev->gfx.kiq[0].ring);
 	return 0;
 }
 
 static int gfx_v11_0_kcq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = NULL;
-	int r = 0, i;
+	int i, r;
 
 	if (!amdgpu_async_gfx_ring)
 		gfx_v11_0_cp_compute_enable(adev, true);
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		ring = &adev->gfx.compute_ring[i];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			goto done;
-		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v11_0_kcq_init_queue(ring, false);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
+		r = gfx_v11_0_kcq_init_queue(&adev->gfx.compute_ring[i], false);
 		if (r)
-			goto done;
+			return r;
 	}
 
-	r = amdgpu_gfx_enable_kcq(adev, 0);
-done:
-	return r;
+	return amdgpu_gfx_enable_kcq(adev, 0);
 }
 
 static int gfx_v11_0_cp_resume(struct amdgpu_device *adev)
@@ -4517,11 +4613,23 @@ static int gfx_v11_0_cp_resume(struct amdgpu_device *adev)
 			return r;
 	}
 
-	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
-		ring = &adev->gfx.gfx_ring[i];
-		r = amdgpu_ring_test_helper(ring);
-		if (r)
-			return r;
+	if (adev->gfx.disable_kq) {
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+			ring = &adev->gfx.gfx_ring[i];
+			/* we don't want to set ring->ready */
+			r = amdgpu_ring_test_ring(ring);
+			if (r)
+				return r;
+		}
+		if (amdgpu_async_gfx_ring)
+			amdgpu_gfx_disable_kgq(adev, 0);
+	} else {
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+			ring = &adev->gfx.gfx_ring[i];
+			r = amdgpu_ring_test_helper(ring);
+			if (r)
+				return r;
+		}
 	}
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
@@ -4549,10 +4657,9 @@ static int gfx_v11_0_gfxhub_enable(struct amdgpu_device *adev)
 	if (r)
 		return r;
 
-	adev->hdp.funcs->flush_hdp(adev, NULL);
+	amdgpu_device_flush_hdp(adev, NULL);
 
-	value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ?
-		false : true;
+	value = amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS;
 
 	adev->gfxhub.funcs->set_fault_enable_default(adev, value);
 	/* TODO investigate why this and the hdp flush above is needed,
@@ -4730,16 +4837,63 @@ static int gfx_v11_0_hw_init(struct amdgpu_ip_block *ip_block)
 	return r;
 }
 
+static int gfx_v11_0_set_userq_eop_interrupts(struct amdgpu_device *adev,
+					      bool enable)
+{
+	unsigned int irq_type;
+	int m, p, r;
+
+	if (adev->userq_funcs[AMDGPU_HW_IP_GFX]) {
+		for (m = 0; m < adev->gfx.me.num_me; m++) {
+			for (p = 0; p < adev->gfx.me.num_pipe_per_me; p++) {
+				irq_type = AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP + p;
+				if (enable)
+					r = amdgpu_irq_get(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				else
+					r = amdgpu_irq_put(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	if (adev->userq_funcs[AMDGPU_HW_IP_COMPUTE]) {
+		for (m = 0; m < adev->gfx.mec.num_mec; ++m) {
+			for (p = 0; p < adev->gfx.mec.num_pipe_per_mec; p++) {
+				irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+					+ (m * adev->gfx.mec.num_pipe_per_mec)
+					+ p;
+				if (enable)
+					r = amdgpu_irq_get(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				else
+					r = amdgpu_irq_put(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
 static int gfx_v11_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_device *adev = ip_block->adev;
 
+	cancel_delayed_work_sync(&adev->gfx.idle_work);
+
 	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.bad_op_irq, 0);
+	gfx_v11_0_set_userq_eop_interrupts(adev, false);
 
 	if (!adev->no_hw_access) {
-		if (amdgpu_async_gfx_ring) {
+		if (amdgpu_async_gfx_ring &&
+		    !adev->gfx.disable_kq) {
 			if (amdgpu_gfx_disable_kgq(adev, 0))
 				DRM_ERROR("KGQ disable failed\n");
 		}
@@ -4778,9 +4932,9 @@ static int gfx_v11_0_resume(struct amdgpu_ip_block *ip_block)
 	return gfx_v11_0_hw_init(ip_block);
 }
 
-static bool gfx_v11_0_is_idle(void *handle)
+static bool gfx_v11_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (REG_GET_FIELD(RREG32_SOC15(GC, 0, regGRBM_STATUS),
 				GRBM_STATUS, GUI_ACTIVE))
@@ -5065,11 +5219,36 @@ static int gfx_v11_0_early_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_device *adev = ip_block->adev;
 
+	switch (amdgpu_user_queue) {
+	case -1:
+	case 0:
+	default:
+		adev->gfx.disable_kq = false;
+		adev->gfx.disable_uq = true;
+		break;
+	case 1:
+		adev->gfx.disable_kq = false;
+		adev->gfx.disable_uq = false;
+		break;
+	case 2:
+		adev->gfx.disable_kq = true;
+		adev->gfx.disable_uq = false;
+		break;
+	}
+
 	adev->gfx.funcs = &gfx_v11_0_gfx_funcs;
 
-	adev->gfx.num_gfx_rings = GFX11_NUM_GFX_RINGS;
-	adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
-					  AMDGPU_MAX_COMPUTE_RINGS);
+	if (adev->gfx.disable_kq) {
+		/* We need one GFX ring temporarily to set up
+		 * the clear state.
+		 */
+		adev->gfx.num_gfx_rings = 1;
+		adev->gfx.num_compute_rings = 0;
+	} else {
+		adev->gfx.num_gfx_rings = GFX11_NUM_GFX_RINGS;
+		adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+						  AMDGPU_MAX_COMPUTE_RINGS);
+	}
 
 	gfx_v11_0_set_kiq_pm4_funcs(adev);
 	gfx_v11_0_set_ring_funcs(adev);
@@ -5100,6 +5279,11 @@ static int gfx_v11_0_late_init(struct amdgpu_ip_block *ip_block)
 	r = amdgpu_irq_get(adev, &adev->gfx.bad_op_irq, 0);
 	if (r)
 		return r;
+
+	r = gfx_v11_0_set_userq_eop_interrupts(adev, true);
+	if (r)
+		return r;
+
 	return 0;
 }
 
@@ -5448,6 +5632,7 @@ static void gfx_v11_cntl_power_gating(struct amdgpu_device *adev, bool enable)
 		case IP_VERSION(11, 5, 0):
 		case IP_VERSION(11, 5, 1):
 		case IP_VERSION(11, 5, 2):
+		case IP_VERSION(11, 5, 3):
 			WREG32_SOC15(GC, 0, regRLC_PG_DELAY_3, RLC_PG_DELAY_3_DEFAULT_GC_11_0_1);
 			break;
 		default:
@@ -5485,6 +5670,7 @@ static int gfx_v11_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		if (!enable)
 			amdgpu_gfx_off_ctrl(adev, false);
 
@@ -5518,6 +5704,7 @@ static int gfx_v11_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 	        gfx_v11_0_update_gfx_clock_gating(adev,
 	                        state ==  AMD_CG_STATE_GATE);
 	        break;
@@ -5528,9 +5715,9 @@ static int gfx_v11_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
-static void gfx_v11_0_get_clockgating_state(void *handle, u64 *flags)
+static void gfx_v11_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	/* AMD_CG_SUPPORT_GFX_MGCG */
@@ -5677,8 +5864,6 @@ static void gfx_v11_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	u32 header, control = 0;
 
-	BUG_ON(ib->flags & AMDGPU_IB_FLAG_CE);
-
 	header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
 
 	control |= ib->length_dw | (vmid << 24);
@@ -5689,15 +5874,11 @@ static void gfx_v11_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 		if (flags & AMDGPU_IB_PREEMPTED)
 			control |= INDIRECT_BUFFER_PRE_RESUME(1);
 
-		if (vmid)
+		if (vmid && !ring->adev->gfx.rs64_enable)
 			gfx_v11_0_ring_emit_de_meta(ring,
-				    (!amdgpu_sriov_vf(ring->adev) && flags & AMDGPU_IB_PREEMPTED) ? true : false);
+				!amdgpu_sriov_vf(ring->adev) && (flags & AMDGPU_IB_PREEMPTED));
 	}
 
-	if (ring->is_mes_queue)
-		/* inherit vmid from mqd */
-		control |= 0x400000;
-
 	amdgpu_ring_write(ring, header);
 	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
 	amdgpu_ring_write(ring,
@@ -5717,10 +5898,6 @@ static void gfx_v11_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
 	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
 
-	if (ring->is_mes_queue)
-		/* inherit vmid from mqd */
-		control |= 0x40000000;
-
 	/* Currently, there is a high possibility to get wave ID mismatch
 	 * between ME and GDS, leading to a hw deadlock, because ME generates
 	 * different wave IDs than the GDS expects. This situation happens
@@ -5778,8 +5955,7 @@ static void gfx_v11_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
 	amdgpu_ring_write(ring, upper_32_bits(addr));
 	amdgpu_ring_write(ring, lower_32_bits(seq));
 	amdgpu_ring_write(ring, upper_32_bits(seq));
-	amdgpu_ring_write(ring, ring->is_mes_queue ?
-			 (ring->hw_queue_id | AMDGPU_FENCE_MES_QUEUE_FLAG) : 0);
+	amdgpu_ring_write(ring, 0);
 }
 
 static void gfx_v11_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
@@ -5807,10 +5983,7 @@ static void gfx_v11_0_ring_invalidate_tlbs(struct amdgpu_ring *ring,
 static void gfx_v11_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 					 unsigned vmid, uint64_t pd_addr)
 {
-	if (ring->is_mes_queue)
-		gfx_v11_0_ring_invalidate_tlbs(ring, 0, 0, false, 0);
-	else
-		amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	/* compute doesn't have PFP */
 	if (ring->funcs->type == AMDGPU_RING_TYPE_GFX) {
@@ -6039,28 +6212,13 @@ static void gfx_v11_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume)
 	void *de_payload_cpu_addr;
 	int cnt;
 
-	if (ring->is_mes_queue) {
-		offset = offsetof(struct amdgpu_mes_ctx_meta_data,
-				  gfx[0].gfx_meta_data) +
-			offsetof(struct v10_gfx_meta_data, de_payload);
-		de_payload_gpu_addr =
-			amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
-		de_payload_cpu_addr =
-			amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset);
-
-		offset = offsetof(struct amdgpu_mes_ctx_meta_data,
-				  gfx[0].gds_backup) +
-			offsetof(struct v10_gfx_meta_data, de_payload);
-		gds_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
-	} else {
-		offset = offsetof(struct v10_gfx_meta_data, de_payload);
-		de_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
-		de_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
+	offset = offsetof(struct v10_gfx_meta_data, de_payload);
+	de_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
+	de_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
 
-		gds_addr = ALIGN(amdgpu_csa_vaddr(ring->adev) +
-				 AMDGPU_CSA_SIZE - adev->gds.gds_size,
-				 PAGE_SIZE);
-	}
+	gds_addr = ALIGN(amdgpu_csa_vaddr(ring->adev) +
+			 AMDGPU_CSA_SIZE - adev->gds.gds_size,
+			 PAGE_SIZE);
 
 	de_payload.gds_backup_addrlo = lower_32_bits(gds_addr);
 	de_payload.gds_backup_addrhi = upper_32_bits(gds_addr);
@@ -6147,21 +6305,6 @@ static void gfx_v11_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
 			       ref, mask, 0x20);
 }
 
-static void gfx_v11_0_ring_soft_recovery(struct amdgpu_ring *ring,
-					 unsigned vmid)
-{
-	struct amdgpu_device *adev = ring->adev;
-	uint32_t value = 0;
-
-	value = REG_SET_FIELD(value, SQ_CMD, CMD, 0x03);
-	value = REG_SET_FIELD(value, SQ_CMD, MODE, 0x01);
-	value = REG_SET_FIELD(value, SQ_CMD, CHECK_VMID, 1);
-	value = REG_SET_FIELD(value, SQ_CMD, VM_ID, vmid);
-	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
-	WREG32_SOC15(GC, 0, regSQ_CMD, value);
-	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
-}
-
 static void
 gfx_v11_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
 				      uint32_t me, uint32_t pipe,
@@ -6299,25 +6442,23 @@ static int gfx_v11_0_eop_irq(struct amdgpu_device *adev,
 			     struct amdgpu_irq_src *source,
 			     struct amdgpu_iv_entry *entry)
 {
-	int i;
+	u32 doorbell_offset = entry->src_data[0];
 	u8 me_id, pipe_id, queue_id;
 	struct amdgpu_ring *ring;
-	uint32_t mes_queue_id = entry->src_data[0];
+	int i;
 
 	DRM_DEBUG("IH: CP EOP\n");
 
-	if (adev->enable_mes && (mes_queue_id & AMDGPU_FENCE_MES_QUEUE_FLAG)) {
-		struct amdgpu_mes_queue *queue;
-
-		mes_queue_id &= AMDGPU_FENCE_MES_QUEUE_ID_MASK;
+	if (adev->enable_mes && doorbell_offset) {
+		struct amdgpu_userq_fence_driver *fence_drv = NULL;
+		struct xarray *xa = &adev->userq_xa;
+		unsigned long flags;
 
-		spin_lock(&adev->mes.queue_id_lock);
-		queue = idr_find(&adev->mes.queue_id_idr, mes_queue_id);
-		if (queue) {
-			DRM_DEBUG("process mes queue id = %d\n", mes_queue_id);
-			amdgpu_fence_process(queue->ring);
-		}
-		spin_unlock(&adev->mes.queue_id_lock);
+		xa_lock_irqsave(xa, flags);
+		fence_drv = xa_load(xa, doorbell_offset);
+		if (fence_drv)
+			amdgpu_userq_fence_driver_process(fence_drv);
+		xa_unlock_irqrestore(xa, flags);
 	} else {
 		me_id = (entry->ring_id & 0x0c) >> 2;
 		pipe_id = (entry->ring_id & 0x03) >> 0;
@@ -6484,27 +6625,29 @@ static void gfx_v11_0_handle_priv_fault(struct amdgpu_device *adev,
 	pipe_id = (entry->ring_id & 0x03) >> 0;
 	queue_id = (entry->ring_id & 0x70) >> 4;
 
-	switch (me_id) {
-	case 0:
-		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
-			ring = &adev->gfx.gfx_ring[i];
-			if (ring->me == me_id && ring->pipe == pipe_id &&
-			    ring->queue == queue_id)
-				drm_sched_fault(&ring->sched);
-		}
-		break;
-	case 1:
-	case 2:
-		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-			ring = &adev->gfx.compute_ring[i];
-			if (ring->me == me_id && ring->pipe == pipe_id &&
-			    ring->queue == queue_id)
-				drm_sched_fault(&ring->sched);
+	if (!adev->gfx.disable_kq) {
+		switch (me_id) {
+		case 0:
+			for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+				ring = &adev->gfx.gfx_ring[i];
+				if (ring->me == me_id && ring->pipe == pipe_id &&
+				    ring->queue == queue_id)
+					drm_sched_fault(&ring->sched);
+			}
+			break;
+		case 1:
+		case 2:
+			for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+				ring = &adev->gfx.compute_ring[i];
+				if (ring->me == me_id && ring->pipe == pipe_id &&
+				    ring->queue == queue_id)
+					drm_sched_fault(&ring->sched);
+			}
+			break;
+		default:
+			BUG();
+			break;
 		}
-		break;
-	default:
-		BUG();
-		break;
 	}
 }
 
@@ -6612,32 +6755,90 @@ static void gfx_v11_0_emit_mem_sync(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, gcr_cntl); /* GCR_CNTL */
 }
 
-static int gfx_v11_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid)
+static bool gfx_v11_pipe_reset_support(struct amdgpu_device *adev)
+{
+	/* Disable the pipe reset until the CPFW fully support it.*/
+	dev_warn_once(adev->dev, "The CPFW hasn't support pipe reset yet.\n");
+	return false;
+}
+
+
+static int gfx_v11_reset_gfx_pipe(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
+	uint32_t reset_pipe = 0, clean_pipe = 0;
 	int r;
 
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
+	if (!gfx_v11_pipe_reset_support(adev))
+		return -EOPNOTSUPP;
 
-	r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, false);
-	if (r)
-		return r;
+	gfx_v11_0_set_safe_mode(adev, 0);
+	mutex_lock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
 
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0)) {
-		dev_err(adev->dev, "fail to resv mqd_obj\n");
-		return r;
+	switch (ring->pipe) {
+	case 0:
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   PFP_PIPE0_RESET, 1);
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   ME_PIPE0_RESET, 1);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   PFP_PIPE0_RESET, 0);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   ME_PIPE0_RESET, 0);
+		break;
+	case 1:
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   PFP_PIPE1_RESET, 1);
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   ME_PIPE1_RESET, 1);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   PFP_PIPE1_RESET, 0);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   ME_PIPE1_RESET, 0);
+		break;
+	default:
+		break;
 	}
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (!r) {
-		r = gfx_v11_0_kgq_init_queue(ring, true);
-		amdgpu_bo_kunmap(ring->mqd_obj);
-		ring->mqd_ptr = NULL;
+
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, reset_pipe);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, clean_pipe);
+
+	r = (RREG32(SOC15_REG_OFFSET(GC, 0, regCP_GFX_RS64_INSTR_PNTR1)) << 2) -
+						RS64_FW_UC_START_ADDR_LO;
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	gfx_v11_0_unset_safe_mode(adev, 0);
+
+	dev_info(adev->dev, "The ring %s pipe reset to the ME firmware start PC: %s\n", ring->name,
+			r == 0 ? "successfully" : "failed");
+	/* FIXME: Sometimes driver can't cache the ME firmware start PC correctly,
+	 * so the pipe reset status relies on the later gfx ring test result.
+	 */
+	return 0;
+}
+
+static int gfx_v11_0_reset_kgq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
+	r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, false);
+	if (r) {
+
+		dev_warn(adev->dev, "reset via MES failed and try pipe reset %d\n", r);
+		r = gfx_v11_reset_gfx_pipe(ring);
+		if (r)
+			return r;
 	}
-	amdgpu_bo_unreserve(ring->mqd_obj);
+
+	r = gfx_v11_0_kgq_init_queue(ring, true);
 	if (r) {
-		dev_err(adev->dev, "fail to unresv mqd_obj\n");
+		dev_err(adev->dev, "failed to init kgq\n");
 		return r;
 	}
 
@@ -6647,37 +6848,159 @@ static int gfx_v11_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid)
 		return r;
 	}
 
-	return amdgpu_ring_test_ring(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
 }
 
-static int gfx_v11_0_reset_kcq(struct amdgpu_ring *ring, unsigned int vmid)
+static int gfx_v11_0_reset_compute_pipe(struct amdgpu_ring *ring)
+{
+
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reset_pipe = 0, clean_pipe = 0;
+	int r;
+
+	if (!gfx_v11_pipe_reset_support(adev))
+		return -EOPNOTSUPP;
+
+	gfx_v11_0_set_safe_mode(adev, 0);
+	mutex_lock(&adev->srbm_mutex);
+	soc21_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+
+	reset_pipe = RREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL);
+	clean_pipe = reset_pipe;
+
+	if (adev->gfx.rs64_enable) {
+
+		switch (ring->pipe) {
+		case 0:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE0_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE0_RESET, 0);
+			break;
+		case 1:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE1_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE1_RESET, 0);
+			break;
+		case 2:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE2_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE2_RESET, 0);
+			break;
+		case 3:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE3_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE3_RESET, 0);
+			break;
+		default:
+			break;
+		}
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, reset_pipe);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, clean_pipe);
+		r = (RREG32_SOC15(GC, 0, regCP_MEC_RS64_INSTR_PNTR) << 2) -
+					RS64_FW_UC_START_ADDR_LO;
+	} else {
+		if (ring->me == 1) {
+			switch (ring->pipe) {
+			case 0:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE0_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE0_RESET, 0);
+				break;
+			case 1:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE1_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE1_RESET, 0);
+				break;
+			case 2:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE2_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE2_RESET, 0);
+				break;
+			case 3:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE3_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE3_RESET, 0);
+				break;
+			default:
+				break;
+			}
+			/* mec1 fw pc: CP_MEC1_INSTR_PNTR */
+		} else {
+			switch (ring->pipe) {
+			case 0:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE0_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE0_RESET, 0);
+				break;
+			case 1:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE1_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE1_RESET, 0);
+				break;
+			case 2:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE2_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE2_RESET, 0);
+				break;
+			case 3:
+				reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE3_RESET, 1);
+				clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME2_PIPE3_RESET, 0);
+				break;
+			default:
+				break;
+			}
+			/* mec2 fw pc: CP:CP_MEC2_INSTR_PNTR */
+		}
+		WREG32_SOC15(GC, 0, regCP_MEC_CNTL, reset_pipe);
+		WREG32_SOC15(GC, 0, regCP_MEC_CNTL, clean_pipe);
+		r = RREG32(SOC15_REG_OFFSET(GC, 0, regCP_MEC1_INSTR_PNTR));
+	}
+
+	soc21_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	gfx_v11_0_unset_safe_mode(adev, 0);
+
+	dev_info(adev->dev, "The ring %s pipe resets to MEC FW start PC: %s\n", ring->name,
+			r == 0 ? "successfully" : "failed");
+	/*FIXME:Sometimes driver can't cache the MEC firmware start PC correctly, so the pipe
+	 * reset status relies on the compute ring test result.
+	 */
+	return 0;
+}
+
+static int gfx_v11_0_reset_kcq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
 {
 	struct amdgpu_device *adev = ring->adev;
 	int r = 0;
 
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
 
 	r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, true);
 	if (r) {
-		dev_err(adev->dev, "reset via MMIO failed %d\n", r);
-		return r;
+		dev_warn(adev->dev, "fail(%d) to reset kcq and try pipe reset\n", r);
+		r = gfx_v11_0_reset_compute_pipe(ring);
+		if (r)
+			return r;
 	}
 
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0)) {
-		dev_err(adev->dev, "fail to resv mqd_obj\n");
-		return r;
-	}
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (!r) {
-		r = gfx_v11_0_kcq_init_queue(ring, true);
-		amdgpu_bo_kunmap(ring->mqd_obj);
-		ring->mqd_ptr = NULL;
-	}
-	amdgpu_bo_unreserve(ring->mqd_obj);
+	r = gfx_v11_0_kcq_init_queue(ring, true);
 	if (r) {
-		dev_err(adev->dev, "fail to unresv mqd_obj\n");
+		dev_err(adev->dev, "fail to init kcq\n");
 		return r;
 	}
 	r = amdgpu_mes_map_legacy_queue(adev, ring);
@@ -6686,7 +7009,7 @@ static int gfx_v11_0_reset_kcq(struct amdgpu_ring *ring, unsigned int vmid)
 		return r;
 	}
 
-	return amdgpu_ring_test_ring(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
 }
 
 static void gfx_v11_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
@@ -6718,9 +7041,14 @@ static void gfx_v11_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printe
 			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
 				drm_printf(p, "\nmec %d, pipe %d, queue %d\n", i, j, k);
 				for (reg = 0; reg < reg_count; reg++) {
-					drm_printf(p, "%-50s \t 0x%08x\n",
-						   gc_cp_reg_list_11[reg].reg_name,
-						   adev->gfx.ip_dump_compute_queues[index + reg]);
+					if (i && gc_cp_reg_list_11[reg].reg_offset == regCP_MEC_ME1_HEADER_DUMP)
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   "regCP_MEC_ME2_HEADER_DUMP",
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
+					else
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   gc_cp_reg_list_11[reg].reg_name,
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
 				}
 				index += reg_count;
 			}
@@ -6780,9 +7108,16 @@ static void gfx_v11_ip_dump(struct amdgpu_ip_block *ip_block)
 				/* ME0 is for GFX so start from 1 for CP */
 				soc21_grbm_select(adev, adev->gfx.me.num_me + i, j, k, 0);
 				for (reg = 0; reg < reg_count; reg++) {
-					adev->gfx.ip_dump_compute_queues[index + reg] =
-						RREG32(SOC15_REG_ENTRY_OFFSET(
-							gc_cp_reg_list_11[reg]));
+					if (i &&
+					    gc_cp_reg_list_11[reg].reg_offset ==
+						    regCP_MEC_ME1_HEADER_DUMP)
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_OFFSET(GC, 0,
+							       regCP_MEC_ME2_HEADER_DUMP));
+					else
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_ENTRY_OFFSET(
+								       gc_cp_reg_list_11[reg]));
 				}
 				index += reg_count;
 			}
@@ -6826,6 +7161,20 @@ static void gfx_v11_0_ring_emit_cleaner_shader(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, 0);  /* RESERVED field, programmed to zero */
 }
 
+static void gfx_v11_0_ring_begin_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_begin_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_begin_use(ring);
+}
+
+static void gfx_v11_0_ring_end_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_end_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_end_use(ring);
+}
+
 static const struct amd_ip_funcs gfx_v11_0_ip_funcs = {
 	.name = "gfx_v11_0",
 	.early_init = gfx_v11_0_early_init,
@@ -6896,12 +7245,11 @@ static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_gfx = {
 	.emit_wreg = gfx_v11_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v11_0_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v11_0_ring_emit_reg_write_reg_wait,
-	.soft_recovery = gfx_v11_0_ring_soft_recovery,
 	.emit_mem_sync = gfx_v11_0_emit_mem_sync,
 	.reset = gfx_v11_0_reset_kgq,
 	.emit_cleaner_shader = gfx_v11_0_ring_emit_cleaner_shader,
-	.begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use,
-	.end_use = amdgpu_gfx_enforce_isolation_ring_end_use,
+	.begin_use = gfx_v11_0_ring_begin_use,
+	.end_use = gfx_v11_0_ring_end_use,
 };
 
 static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_compute = {
@@ -6938,12 +7286,11 @@ static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_compute = {
 	.emit_wreg = gfx_v11_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v11_0_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v11_0_ring_emit_reg_write_reg_wait,
-	.soft_recovery = gfx_v11_0_ring_soft_recovery,
 	.emit_mem_sync = gfx_v11_0_emit_mem_sync,
 	.reset = gfx_v11_0_reset_kcq,
 	.emit_cleaner_shader = gfx_v11_0_ring_emit_cleaner_shader,
-	.begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use,
-	.end_use = amdgpu_gfx_enforce_isolation_ring_end_use,
+	.begin_use = gfx_v11_0_ring_begin_use,
+	.end_use = gfx_v11_0_ring_end_use,
 };
 
 static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_kiq = {
@@ -6973,6 +7320,7 @@ static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_kiq = {
 	.emit_wreg = gfx_v11_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v11_0_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v11_0_ring_emit_reg_write_reg_wait,
+	.emit_hdp_flush = gfx_v11_0_ring_emit_hdp_flush,
 };
 
 static void gfx_v11_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
index 2523221a2519..d01d2712cf57 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
@@ -29,7 +29,6 @@
 #include "amdgpu_gfx.h"
 #include "amdgpu_psp.h"
 #include "amdgpu_smu.h"
-#include "amdgpu_atomfirmware.h"
 #include "imu_v12_0.h"
 #include "soc24.h"
 #include "nvd.h"
@@ -37,21 +36,40 @@
 #include "gc/gc_12_0_0_offset.h"
 #include "gc/gc_12_0_0_sh_mask.h"
 #include "soc24_enum.h"
-#include "ivsrcid/gfx/irqsrcs_gfx_11_0_0.h"
+#include "ivsrcid/gfx/irqsrcs_gfx_12_0_0.h"
 
 #include "soc15.h"
-#include "soc15d.h"
 #include "clearstate_gfx12.h"
 #include "v12_structs.h"
 #include "gfx_v12_0.h"
 #include "nbif_v6_3_1.h"
 #include "mes_v12_0.h"
+#include "mes_userqueue.h"
+#include "amdgpu_userq_fence.h"
 
 #define GFX12_NUM_GFX_RINGS	1
 #define GFX12_MEC_HPD_SIZE	2048
 
 #define RLCG_UCODE_LOADING_START_ADDRESS	0x00002000L
 
+#define regCP_GFX_MQD_CONTROL_DEFAULT                                             0x00000100
+#define regCP_GFX_HQD_VMID_DEFAULT                                                0x00000000
+#define regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT                                      0x00000000
+#define regCP_GFX_HQD_QUANTUM_DEFAULT                                             0x00000a01
+#define regCP_GFX_HQD_CNTL_DEFAULT                                                0x00f00000
+#define regCP_RB_DOORBELL_CONTROL_DEFAULT                                         0x00000000
+#define regCP_GFX_HQD_RPTR_DEFAULT                                                0x00000000
+
+#define regCP_HQD_EOP_CONTROL_DEFAULT                                             0x00000006
+#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT                                     0x00000000
+#define regCP_MQD_CONTROL_DEFAULT                                                 0x00000100
+#define regCP_HQD_PQ_CONTROL_DEFAULT                                              0x00308509
+#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT                                     0x00000000
+#define regCP_HQD_PQ_RPTR_DEFAULT                                                 0x00000000
+#define regCP_HQD_PERSISTENT_STATE_DEFAULT                                        0x0be05501
+#define regCP_HQD_IB_CONTROL_DEFAULT                                              0x00300000
+
+
 MODULE_FIRMWARE("amdgpu/gc_12_0_0_pfp.bin");
 MODULE_FIRMWARE("amdgpu/gc_12_0_0_me.bin");
 MODULE_FIRMWARE("amdgpu/gc_12_0_0_mec.bin");
@@ -61,6 +79,7 @@ MODULE_FIRMWARE("amdgpu/gc_12_0_1_pfp.bin");
 MODULE_FIRMWARE("amdgpu/gc_12_0_1_me.bin");
 MODULE_FIRMWARE("amdgpu/gc_12_0_1_mec.bin");
 MODULE_FIRMWARE("amdgpu/gc_12_0_1_rlc.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_rlc_kicker.bin");
 MODULE_FIRMWARE("amdgpu/gc_12_0_1_toc.bin");
 
 static const struct amdgpu_hwip_reg_entry gc_reg_list_12_0[] = {
@@ -117,11 +136,14 @@ static const struct amdgpu_hwip_reg_entry gc_reg_list_12_0[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_RS64_INSTR_PNTR0),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_GFX_RS64_INSTR_PNTR1),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_RS64_INSTR_PNTR),
-
 	/* cp header registers */
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_MES_HEADER_DUMP),
 	/* SE status registers */
 	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE0),
@@ -170,7 +192,16 @@ static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_12[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_CNTL_STACK_OFFSET),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_CNTL_STACK_DW_CNT),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_WG_STATE_OFFSET),
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_STATUS)
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_STATUS),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
 };
 
 static const struct amdgpu_hwip_reg_entry gc_gfx_queue_reg_list_12[] = {
@@ -199,7 +230,24 @@ static const struct amdgpu_hwip_reg_entry gc_gfx_queue_reg_list_12[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_LO),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BASE_HI),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_CMD_BUFSZ),
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BUFSZ)
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_IB1_BUFSZ),
+	/* cp header registers */
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_ME_HEADER_DUMP),
 };
 
 static const struct soc15_reg_golden golden_settings_gc_12_0_rev0[] = {
@@ -449,7 +497,7 @@ static int gfx_v12_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	struct dma_fence *f = NULL;
 	unsigned index;
 	uint64_t gpu_addr;
-	volatile uint32_t *cpu_ptr;
+	uint32_t *cpu_ptr;
 	long r;
 
 	/* MES KIQ fw hasn't indirect buffer support for now */
@@ -459,33 +507,18 @@ static int gfx_v12_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 
 	memset(&ib, 0, sizeof(ib));
 
-	if (ring->is_mes_queue) {
-		uint32_t padding, offset;
-
-		offset = amdgpu_mes_ctx_get_offs(ring, AMDGPU_MES_CTX_IB_OFFS);
-		padding = amdgpu_mes_ctx_get_offs(ring,
-						  AMDGPU_MES_CTX_PADDING_OFFS);
-
-		ib.gpu_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
-		ib.ptr = amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset);
-
-		gpu_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, padding);
-		cpu_ptr = amdgpu_mes_ctx_get_offs_cpu_addr(ring, padding);
-		*cpu_ptr = cpu_to_le32(0xCAFEDEAD);
-	} else {
-		r = amdgpu_device_wb_get(adev, &index);
-		if (r)
-			return r;
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r)
+		return r;
 
-		gpu_addr = adev->wb.gpu_addr + (index * 4);
-		adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
-		cpu_ptr = &adev->wb.wb[index];
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
+	cpu_ptr = &adev->wb.wb[index];
 
-		r = amdgpu_ib_get(adev, NULL, 16, AMDGPU_IB_POOL_DIRECT, &ib);
-		if (r) {
-			dev_err(adev->dev, "amdgpu: failed to get ib (%ld).\n", r);
-			goto err1;
-		}
+	r = amdgpu_ib_get(adev, NULL, 16, AMDGPU_IB_POOL_DIRECT, &ib);
+	if (r) {
+		dev_err(adev->dev, "amdgpu: failed to get ib (%ld).\n", r);
+		goto err1;
 	}
 
 	ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3);
@@ -512,12 +545,10 @@ static int gfx_v12_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	else
 		r = -EINVAL;
 err2:
-	if (!ring->is_mes_queue)
-		amdgpu_ib_free(&ib, NULL);
+	amdgpu_ib_free(&ib, NULL);
 	dma_fence_put(f);
 err1:
-	if (!ring->is_mes_queue)
-		amdgpu_device_wb_free(adev, index);
+	amdgpu_device_wb_free(adev, index);
 	return r;
 }
 
@@ -556,7 +587,7 @@ out:
 
 static int gfx_v12_0_init_microcode(struct amdgpu_device *adev)
 {
-	char ucode_prefix[15];
+	char ucode_prefix[30];
 	int err;
 	const struct rlc_firmware_header_v2_0 *rlc_hdr;
 	uint16_t version_major;
@@ -583,9 +614,14 @@ static int gfx_v12_0_init_microcode(struct amdgpu_device *adev)
 	amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_RS64_ME_P0_STACK);
 
 	if (!amdgpu_sriov_vf(adev)) {
-		err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
-					   AMDGPU_UCODE_REQUIRED,
-					   "amdgpu/%s_rlc.bin", ucode_prefix);
+		if (amdgpu_is_kicker_fw(adev))
+			err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_rlc_kicker.bin", ucode_prefix);
+		else
+			err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
+						   AMDGPU_UCODE_REQUIRED,
+						   "amdgpu/%s_rlc.bin", ucode_prefix);
 		if (err)
 			goto out;
 		rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
@@ -649,8 +685,7 @@ static u32 gfx_v12_0_get_csb_size(struct amdgpu_device *adev)
 	return count;
 }
 
-static void gfx_v12_0_get_csb_buffer(struct amdgpu_device *adev,
-				     volatile u32 *buffer)
+static void gfx_v12_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
 	u32 count = 0, clustercount = 0, i;
 	const struct cs_section_def *sect = NULL;
@@ -865,6 +900,34 @@ static void gfx_v12_0_select_me_pipe_q(struct amdgpu_device *adev,
 	soc24_grbm_select(adev, me, pipe, q, vm);
 }
 
+/* all sizes are in bytes */
+#define MQD_SHADOW_BASE_SIZE      73728
+#define MQD_SHADOW_BASE_ALIGNMENT 256
+#define MQD_FWWORKAREA_SIZE       484
+#define MQD_FWWORKAREA_ALIGNMENT  256
+
+static void gfx_v12_0_get_gfx_shadow_info_nocheck(struct amdgpu_device *adev,
+						  struct amdgpu_gfx_shadow_info *shadow_info)
+{
+	shadow_info->shadow_size = MQD_SHADOW_BASE_SIZE;
+	shadow_info->shadow_alignment = MQD_SHADOW_BASE_ALIGNMENT;
+	shadow_info->csa_size = MQD_FWWORKAREA_SIZE;
+	shadow_info->csa_alignment = MQD_FWWORKAREA_ALIGNMENT;
+}
+
+static int gfx_v12_0_get_gfx_shadow_info(struct amdgpu_device *adev,
+					 struct amdgpu_gfx_shadow_info *shadow_info,
+					 bool skip_check)
+{
+	if (adev->gfx.cp_gfx_shadow || skip_check) {
+		gfx_v12_0_get_gfx_shadow_info_nocheck(adev, shadow_info);
+		return 0;
+	}
+
+	memset(shadow_info, 0, sizeof(struct amdgpu_gfx_shadow_info));
+	return -EINVAL;
+}
+
 static const struct amdgpu_gfx_funcs gfx_v12_0_gfx_funcs = {
 	.get_gpu_clock_counter = &gfx_v12_0_get_gpu_clock_counter,
 	.select_se_sh = &gfx_v12_0_select_se_sh,
@@ -873,6 +936,7 @@ static const struct amdgpu_gfx_funcs gfx_v12_0_gfx_funcs = {
 	.read_wave_vgprs = &gfx_v12_0_read_wave_vgprs,
 	.select_me_pipe_q = &gfx_v12_0_select_me_pipe_q,
 	.update_perfmon_mgcg = &gfx_v12_0_update_perf_clk,
+	.get_gfx_shadow_info = &gfx_v12_0_get_gfx_shadow_info,
 };
 
 static int gfx_v12_0_gpu_early_init(struct amdgpu_device *adev)
@@ -1330,14 +1394,17 @@ static int gfx_v12_0_sw_init(struct amdgpu_ip_block *ip_block)
 	unsigned num_compute_rings;
 	int xcc_id = 0;
 	struct amdgpu_device *adev = ip_block->adev;
+	int num_queue_per_pipe = 1; /* we only enable 1 KGQ per pipe */
+
+	INIT_DELAYED_WORK(&adev->gfx.idle_work, amdgpu_gfx_profile_idle_work_handler);
 
 	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
 	case IP_VERSION(12, 0, 0):
 	case IP_VERSION(12, 0, 1):
 		adev->gfx.me.num_me = 1;
 		adev->gfx.me.num_pipe_per_me = 1;
-		adev->gfx.me.num_queue_per_pipe = 1;
-		adev->gfx.mec.num_mec = 2;
+		adev->gfx.me.num_queue_per_pipe = 8;
+		adev->gfx.mec.num_mec = 1;
 		adev->gfx.mec.num_pipe_per_mec = 2;
 		adev->gfx.mec.num_queue_per_pipe = 4;
 		break;
@@ -1354,6 +1421,22 @@ static int gfx_v12_0_sw_init(struct amdgpu_ip_block *ip_block)
 	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
 	case IP_VERSION(12, 0, 0):
 	case IP_VERSION(12, 0, 1):
+		if (!adev->gfx.disable_uq &&
+		    adev->gfx.me_fw_version  >= 2780 &&
+		    adev->gfx.pfp_fw_version >= 2840 &&
+		    adev->gfx.mec_fw_version >= 3050 &&
+		    adev->mes.fw_version[0] >= 123) {
+			adev->userq_funcs[AMDGPU_HW_IP_GFX] = &userq_mes_funcs;
+			adev->userq_funcs[AMDGPU_HW_IP_COMPUTE] = &userq_mes_funcs;
+		}
+		break;
+	default:
+		break;
+	}
+
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 0, 1):
 		if (adev->gfx.me_fw_version  >= 2480 &&
 		    adev->gfx.pfp_fw_version >= 2530 &&
 		    adev->gfx.mec_fw_version >= 2680 &&
@@ -1365,36 +1448,38 @@ static int gfx_v12_0_sw_init(struct amdgpu_ip_block *ip_block)
 		break;
 	}
 
-	/* recalculate compute rings to use based on hardware configuration */
-	num_compute_rings = (adev->gfx.mec.num_pipe_per_mec *
-			     adev->gfx.mec.num_queue_per_pipe) / 2;
-	adev->gfx.num_compute_rings = min(adev->gfx.num_compute_rings,
-					  num_compute_rings);
+	if (adev->gfx.num_compute_rings) {
+		/* recalculate compute rings to use based on hardware configuration */
+		num_compute_rings = (adev->gfx.mec.num_pipe_per_mec *
+				     adev->gfx.mec.num_queue_per_pipe) / 2;
+		adev->gfx.num_compute_rings = min(adev->gfx.num_compute_rings,
+						  num_compute_rings);
+	}
 
 	/* EOP Event */
 	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
-			      GFX_11_0_0__SRCID__CP_EOP_INTERRUPT,
+			      GFX_12_0_0__SRCID__CP_EOP_INTERRUPT,
 			      &adev->gfx.eop_irq);
 	if (r)
 		return r;
 
 	/* Bad opcode Event */
 	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
-			      GFX_11_0_0__SRCID__CP_BAD_OPCODE_ERROR,
+			      GFX_12_0_0__SRCID__CP_BAD_OPCODE_ERROR,
 			      &adev->gfx.bad_op_irq);
 	if (r)
 		return r;
 
 	/* Privileged reg */
 	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
-			      GFX_11_0_0__SRCID__CP_PRIV_REG_FAULT,
+			      GFX_12_0_0__SRCID__CP_PRIV_REG_FAULT,
 			      &adev->gfx.priv_reg_irq);
 	if (r)
 		return r;
 
 	/* Privileged inst */
 	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
-			      GFX_11_0_0__SRCID__CP_PRIV_INSTR_FAULT,
+			      GFX_12_0_0__SRCID__CP_PRIV_INSTR_FAULT,
 			      &adev->gfx.priv_inst_irq);
 	if (r)
 		return r;
@@ -1415,37 +1500,41 @@ static int gfx_v12_0_sw_init(struct amdgpu_ip_block *ip_block)
 		return r;
 	}
 
-	/* set up the gfx ring */
-	for (i = 0; i < adev->gfx.me.num_me; i++) {
-		for (j = 0; j < adev->gfx.me.num_queue_per_pipe; j++) {
-			for (k = 0; k < adev->gfx.me.num_pipe_per_me; k++) {
-				if (!amdgpu_gfx_is_me_queue_enabled(adev, i, k, j))
-					continue;
-
-				r = gfx_v12_0_gfx_ring_init(adev, ring_id,
-							    i, k, j);
-				if (r)
-					return r;
-				ring_id++;
+	if (adev->gfx.num_gfx_rings) {
+		/* set up the gfx ring */
+		for (i = 0; i < adev->gfx.me.num_me; i++) {
+			for (j = 0; j < num_queue_per_pipe; j++) {
+				for (k = 0; k < adev->gfx.me.num_pipe_per_me; k++) {
+					if (!amdgpu_gfx_is_me_queue_enabled(adev, i, k, j))
+						continue;
+
+					r = gfx_v12_0_gfx_ring_init(adev, ring_id,
+								    i, k, j);
+					if (r)
+						return r;
+					ring_id++;
+				}
 			}
 		}
 	}
 
-	ring_id = 0;
-	/* set up the compute queues - allocate horizontally across pipes */
-	for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
-		for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
-			for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
-				if (!amdgpu_gfx_is_mec_queue_enabled(adev,
-								0, i, k, j))
-					continue;
+	if (adev->gfx.num_compute_rings) {
+		ring_id = 0;
+		/* set up the compute queues - allocate horizontally across pipes */
+		for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
+			for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
+				for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+					if (!amdgpu_gfx_is_mec_queue_enabled(adev,
+									     0, i, k, j))
+						continue;
 
-				r = gfx_v12_0_compute_ring_init(adev, ring_id,
-								i, k, j);
-				if (r)
-					return r;
+					r = gfx_v12_0_compute_ring_init(adev, ring_id,
+									i, k, j);
+					if (r)
+						return r;
 
-				ring_id++;
+					ring_id++;
+				}
 			}
 		}
 	}
@@ -1458,10 +1547,15 @@ static int gfx_v12_0_sw_init(struct amdgpu_ip_block *ip_block)
 	case IP_VERSION(12, 0, 0):
 	case IP_VERSION(12, 0, 1):
 		if ((adev->gfx.me_fw_version >= 2660) &&
-			    (adev->gfx.mec_fw_version >= 2920)) {
-				adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
-				adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		    (adev->gfx.mec_fw_version >= 2920) &&
+		    !amdgpu_sriov_vf(adev) &&
+		    !adev->debug_disable_gpu_ring_reset) {
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+			adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
 		}
+		break;
+	default:
+		break;
 	}
 
 	if (!adev->enable_mes_kiq) {
@@ -2306,7 +2400,7 @@ static int gfx_v12_0_cp_gfx_load_pfp_microcode_rs64(struct amdgpu_device *adev)
 	amdgpu_bo_unreserve(adev->gfx.pfp.pfp_fw_data_obj);
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_LO,
 		lower_32_bits(adev->gfx.pfp.pfp_fw_gpu_addr));
@@ -2437,7 +2531,7 @@ static int gfx_v12_0_cp_gfx_load_me_microcode_rs64(struct amdgpu_device *adev)
 				      (void **)&adev->gfx.me.me_fw_data_ptr);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to create me data bo\n", r);
-		gfx_v12_0_pfp_fini(adev);
+		gfx_v12_0_me_fini(adev);
 		return r;
 	}
 
@@ -2450,7 +2544,7 @@ static int gfx_v12_0_cp_gfx_load_me_microcode_rs64(struct amdgpu_device *adev)
 	amdgpu_bo_unreserve(adev->gfx.me.me_fw_data_obj);
 
 	if (amdgpu_emu_mode == 1)
-		adev->hdp.funcs->flush_hdp(adev, NULL);
+		amdgpu_device_flush_hdp(adev, NULL);
 
 	WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_LO,
 		lower_32_bits(adev->gfx.me.me_fw_gpu_addr));
@@ -2619,7 +2713,6 @@ static int gfx_v12_0_cp_gfx_resume(struct amdgpu_device *adev)
 	u32 tmp;
 	u32 rb_bufsz;
 	u64 rb_addr, rptr_addr, wptr_gpu_addr;
-	u32 i;
 
 	/* Set the write pointer delay */
 	WREG32_SOC15(GC, 0, regCP_RB_WPTR_DELAY, 0);
@@ -2674,12 +2767,6 @@ static int gfx_v12_0_cp_gfx_resume(struct amdgpu_device *adev)
 
 	/* start the ring */
 	gfx_v12_0_cp_gfx_start(adev);
-
-	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
-		ring = &adev->gfx.gfx_ring[i];
-		ring->sched.ready = true;
-	}
-
 	return 0;
 }
 
@@ -2891,25 +2978,25 @@ static int gfx_v12_0_gfx_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr);
 
 	/* set up mqd control */
-	tmp = RREG32_SOC15(GC, 0, regCP_GFX_MQD_CONTROL);
+	tmp = regCP_GFX_MQD_CONTROL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, VMID, 0);
 	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, PRIV_STATE, 1);
 	tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, CACHE_POLICY, 0);
 	mqd->cp_gfx_mqd_control = tmp;
 
 	/* set up gfx_hqd_vimd with 0x0 to indicate the ring buffer's vmid */
-	tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_VMID);
+	tmp = regCP_GFX_HQD_VMID_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_VMID, VMID, 0);
 	mqd->cp_gfx_hqd_vmid = 0;
 
 	/* set up default queue priority level
 	 * 0x0 = low priority, 0x1 = high priority */
-	tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_QUEUE_PRIORITY);
+	tmp = regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUEUE_PRIORITY, PRIORITY_LEVEL, 0);
 	mqd->cp_gfx_hqd_queue_priority = tmp;
 
 	/* set up time quantum */
-	tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_QUANTUM);
+	tmp = regCP_GFX_HQD_QUANTUM_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUANTUM, QUANTUM_EN, 1);
 	mqd->cp_gfx_hqd_quantum = tmp;
 
@@ -2931,16 +3018,20 @@ static int gfx_v12_0_gfx_mqd_init(struct amdgpu_device *adev, void *m,
 
 	/* set up the gfx_hqd_control, similar as CP_RB0_CNTL */
 	rb_bufsz = order_base_2(prop->queue_size / 4) - 1;
-	tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_CNTL);
+	tmp = regCP_GFX_HQD_CNTL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BUFSZ, rb_bufsz);
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BLKSZ, rb_bufsz - 2);
 #ifdef __BIG_ENDIAN
 	tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, BUF_SWAP, 1);
 #endif
+	if (prop->tmz_queue)
+		tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, TMZ_MATCH, 1);
+	if (!prop->kernel_queue)
+		tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_NON_PRIV, 1);
 	mqd->cp_gfx_hqd_cntl = tmp;
 
 	/* set up cp_doorbell_control */
-	tmp = RREG32_SOC15(GC, 0, regCP_RB_DOORBELL_CONTROL);
+	tmp = regCP_RB_DOORBELL_CONTROL_DEFAULT;
 	if (prop->use_doorbell) {
 		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
 				    DOORBELL_OFFSET, prop->doorbell_index);
@@ -2952,11 +3043,19 @@ static int gfx_v12_0_gfx_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_rb_doorbell_control = tmp;
 
 	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
-	mqd->cp_gfx_hqd_rptr = RREG32_SOC15(GC, 0, regCP_GFX_HQD_RPTR);
+	mqd->cp_gfx_hqd_rptr = regCP_GFX_HQD_RPTR_DEFAULT;
 
 	/* active the queue */
 	mqd->cp_gfx_hqd_active = 1;
 
+	/* set gfx UQ items */
+	mqd->shadow_base_lo = lower_32_bits(prop->shadow_addr);
+	mqd->shadow_base_hi = upper_32_bits(prop->shadow_addr);
+	mqd->fw_work_area_base_lo = lower_32_bits(prop->csa_addr);
+	mqd->fw_work_area_base_hi = upper_32_bits(prop->csa_addr);
+	mqd->fence_address_lo = lower_32_bits(prop->fence_address);
+	mqd->fence_address_hi = upper_32_bits(prop->fence_address);
+
 	return 0;
 }
 
@@ -2990,41 +3089,19 @@ static int gfx_v12_0_kgq_init_queue(struct amdgpu_ring *ring, bool reset)
 
 static int gfx_v12_0_cp_async_gfx_ring_resume(struct amdgpu_device *adev)
 {
-	int r, i;
-	struct amdgpu_ring *ring;
+	int i, r;
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
-		ring = &adev->gfx.gfx_ring[i];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			goto done;
-
-		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v12_0_kgq_init_queue(ring, false);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
+		r = gfx_v12_0_kgq_init_queue(&adev->gfx.gfx_ring[i], false);
 		if (r)
-			goto done;
+			return r;
 	}
 
 	r = amdgpu_gfx_enable_kgq(adev, 0);
 	if (r)
-		goto done;
-
-	r = gfx_v12_0_cp_gfx_start(adev);
-	if (r)
-		goto done;
+		return r;
 
-	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
-		ring = &adev->gfx.gfx_ring[i];
-		ring->sched.ready = true;
-	}
-done:
-	return r;
+	return gfx_v12_0_cp_gfx_start(adev);
 }
 
 static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
@@ -3047,14 +3124,14 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
 
 	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
-	tmp = RREG32_SOC15(GC, 0, regCP_HQD_EOP_CONTROL);
+	tmp = regCP_HQD_EOP_CONTROL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
 			(order_base_2(GFX12_MEC_HPD_SIZE / 4) - 1));
 
 	mqd->cp_hqd_eop_control = tmp;
 
 	/* enable doorbell? */
-	tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL);
+	tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT;
 
 	if (prop->use_doorbell) {
 		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
@@ -3083,7 +3160,7 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr);
 
 	/* set MQD vmid to 0 */
-	tmp = RREG32_SOC15(GC, 0, regCP_MQD_CONTROL);
+	tmp = regCP_MQD_CONTROL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
 	mqd->cp_mqd_control = tmp;
 
@@ -3093,15 +3170,19 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
 
 	/* set up the HQD, this is similar to CP_RB0_CNTL */
-	tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_CONTROL);
+	tmp = regCP_HQD_PQ_CONTROL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
 			    (order_base_2(prop->queue_size / 4) - 1));
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
 			    (order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1));
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 1);
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TUNNEL_DISPATCH, 0);
-	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
-	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	if (prop->kernel_queue) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	}
+	if (prop->tmz_queue)
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TMZ, 1);
 	mqd->cp_hqd_pq_control = tmp;
 
 	/* set the wb address whether it's enabled or not */
@@ -3118,7 +3199,7 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	tmp = 0;
 	/* enable the doorbell if requested */
 	if (prop->use_doorbell) {
-		tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL);
+		tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT;
 		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
 				DOORBELL_OFFSET, prop->doorbell_index);
 
@@ -3133,17 +3214,17 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 	mqd->cp_hqd_pq_doorbell_control = tmp;
 
 	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
-	mqd->cp_hqd_pq_rptr = RREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR);
+	mqd->cp_hqd_pq_rptr = regCP_HQD_PQ_RPTR_DEFAULT;
 
 	/* set the vmid for the queue */
 	mqd->cp_hqd_vmid = 0;
 
-	tmp = RREG32_SOC15(GC, 0, regCP_HQD_PERSISTENT_STATE);
+	tmp = regCP_HQD_PERSISTENT_STATE_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x55);
 	mqd->cp_hqd_persistent_state = tmp;
 
 	/* set MIN_IB_AVAIL_SIZE */
-	tmp = RREG32_SOC15(GC, 0, regCP_HQD_IB_CONTROL);
+	tmp = regCP_HQD_IB_CONTROL_DEFAULT;
 	tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3);
 	mqd->cp_hqd_ib_control = tmp;
 
@@ -3153,6 +3234,10 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
 
 	mqd->cp_hqd_active = prop->hqd_active;
 
+	/* set UQ fenceaddress */
+	mqd->fence_address_lo = lower_32_bits(prop->fence_address);
+	mqd->fence_address_hi = upper_32_bits(prop->fence_address);
+
 	return 0;
 }
 
@@ -3337,57 +3422,25 @@ static int gfx_v12_0_kcq_init_queue(struct amdgpu_ring *ring, bool reset)
 
 static int gfx_v12_0_kiq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring;
-	int r;
-
-	ring = &adev->gfx.kiq[0].ring;
-
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0))
-		return r;
-
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (unlikely(r != 0)) {
-		amdgpu_bo_unreserve(ring->mqd_obj);
-		return r;
-	}
-
-	gfx_v12_0_kiq_init_queue(ring);
-	amdgpu_bo_kunmap(ring->mqd_obj);
-	ring->mqd_ptr = NULL;
-	amdgpu_bo_unreserve(ring->mqd_obj);
-	ring->sched.ready = true;
+	gfx_v12_0_kiq_init_queue(&adev->gfx.kiq[0].ring);
+	adev->gfx.kiq[0].ring.sched.ready = true;
 	return 0;
 }
 
 static int gfx_v12_0_kcq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = NULL;
-	int r = 0, i;
+	int i, r;
 
 	if (!amdgpu_async_gfx_ring)
 		gfx_v12_0_cp_compute_enable(adev, true);
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		ring = &adev->gfx.compute_ring[i];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			goto done;
-		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v12_0_kcq_init_queue(ring, false);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
+		r = gfx_v12_0_kcq_init_queue(&adev->gfx.compute_ring[i], false);
 		if (r)
-			goto done;
+			return r;
 	}
 
-	r = amdgpu_gfx_enable_kcq(adev, 0);
-done:
-	return r;
+	return amdgpu_gfx_enable_kcq(adev, 0);
 }
 
 static int gfx_v12_0_cp_resume(struct amdgpu_device *adev)
@@ -3469,10 +3522,9 @@ static int gfx_v12_0_gfxhub_enable(struct amdgpu_device *adev)
 	if (r)
 		return r;
 
-	adev->hdp.funcs->flush_hdp(adev, NULL);
+	amdgpu_device_flush_hdp(adev, NULL);
 
-	value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ?
-		false : true;
+	value = amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS;
 
 	adev->gfxhub.funcs->set_fault_enable_default(adev, value);
 	/* TODO investigate why this and the hdp flush above is needed,
@@ -3643,14 +3695,60 @@ static int gfx_v12_0_hw_init(struct amdgpu_ip_block *ip_block)
 	return r;
 }
 
+static int gfx_v12_0_set_userq_eop_interrupts(struct amdgpu_device *adev,
+					      bool enable)
+{
+	unsigned int irq_type;
+	int m, p, r;
+
+	if (adev->userq_funcs[AMDGPU_HW_IP_GFX]) {
+		for (m = 0; m < adev->gfx.me.num_me; m++) {
+			for (p = 0; p < adev->gfx.me.num_pipe_per_me; p++) {
+				irq_type = AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP + p;
+				if (enable)
+					r = amdgpu_irq_get(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				else
+					r = amdgpu_irq_put(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	if (adev->userq_funcs[AMDGPU_HW_IP_COMPUTE]) {
+		for (m = 0; m < adev->gfx.mec.num_mec; ++m) {
+			for (p = 0; p < adev->gfx.mec.num_pipe_per_mec; p++) {
+				irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+					+ (m * adev->gfx.mec.num_pipe_per_mec)
+					+ p;
+				if (enable)
+					r = amdgpu_irq_get(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				else
+					r = amdgpu_irq_put(adev, &adev->gfx.eop_irq,
+							   irq_type);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
 static int gfx_v12_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_device *adev = ip_block->adev;
 	uint32_t tmp;
 
+	cancel_delayed_work_sync(&adev->gfx.idle_work);
+
 	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.bad_op_irq, 0);
+	gfx_v12_0_set_userq_eop_interrupts(adev, false);
 
 	if (!adev->no_hw_access) {
 		if (amdgpu_async_gfx_ring) {
@@ -3693,9 +3791,9 @@ static int gfx_v12_0_resume(struct amdgpu_ip_block *ip_block)
 	return gfx_v12_0_hw_init(ip_block);
 }
 
-static bool gfx_v12_0_is_idle(void *handle)
+static bool gfx_v12_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (REG_GET_FIELD(RREG32_SOC15(GC, 0, regGRBM_STATUS),
 				GRBM_STATUS, GUI_ACTIVE))
@@ -3739,11 +3837,33 @@ static int gfx_v12_0_early_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_device *adev = ip_block->adev;
 
+	switch (amdgpu_user_queue) {
+	case -1:
+	case 0:
+	default:
+		adev->gfx.disable_kq = false;
+		adev->gfx.disable_uq = true;
+		break;
+	case 1:
+		adev->gfx.disable_kq = false;
+		adev->gfx.disable_uq = false;
+		break;
+	case 2:
+		adev->gfx.disable_kq = true;
+		adev->gfx.disable_uq = false;
+		break;
+	}
+
 	adev->gfx.funcs = &gfx_v12_0_gfx_funcs;
 
-	adev->gfx.num_gfx_rings = GFX12_NUM_GFX_RINGS;
-	adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
-					  AMDGPU_MAX_COMPUTE_RINGS);
+	if (adev->gfx.disable_kq) {
+		adev->gfx.num_gfx_rings = 0;
+		adev->gfx.num_compute_rings = 0;
+	} else {
+		adev->gfx.num_gfx_rings = GFX12_NUM_GFX_RINGS;
+		adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+						  AMDGPU_MAX_COMPUTE_RINGS);
+	}
 
 	gfx_v12_0_set_kiq_pm4_funcs(adev);
 	gfx_v12_0_set_ring_funcs(adev);
@@ -3774,6 +3894,10 @@ static int gfx_v12_0_late_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	r = gfx_v12_0_set_userq_eop_interrupts(adev, true);
+	if (r)
+		return r;
+
 	return 0;
 }
 
@@ -4147,9 +4271,9 @@ static int gfx_v12_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
-static void gfx_v12_0_get_clockgating_state(void *handle, u64 *flags)
+static void gfx_v12_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	/* AMD_CG_SUPPORT_GFX_MGCG */
@@ -4213,45 +4337,17 @@ static u64 gfx_v12_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
 static void gfx_v12_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	uint32_t *wptr_saved;
-	uint32_t *is_queue_unmap;
-	uint64_t aggregated_db_index;
-	uint32_t mqd_size = adev->mqds[AMDGPU_HW_IP_GFX].mqd_size;
-	uint64_t wptr_tmp;
-
-	if (ring->is_mes_queue) {
-		wptr_saved = (uint32_t *)(ring->mqd_ptr + mqd_size);
-		is_queue_unmap = (uint32_t *)(ring->mqd_ptr + mqd_size +
-					      sizeof(uint32_t));
-		aggregated_db_index =
-			amdgpu_mes_get_aggregated_doorbell_index(adev,
-								 ring->hw_prio);
-
-		wptr_tmp = ring->wptr & ring->buf_mask;
-		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, wptr_tmp);
-		*wptr_saved = wptr_tmp;
-		/* assume doorbell always being used by mes mapped queue */
-		if (*is_queue_unmap) {
-			WDOORBELL64(aggregated_db_index, wptr_tmp);
-			WDOORBELL64(ring->doorbell_index, wptr_tmp);
-		} else {
-			WDOORBELL64(ring->doorbell_index, wptr_tmp);
 
-			if (*is_queue_unmap)
-				WDOORBELL64(aggregated_db_index, wptr_tmp);
-		}
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
 	} else {
-		if (ring->use_doorbell) {
-			/* XXX check if swapping is necessary on BE */
-			atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
-				     ring->wptr);
-			WDOORBELL64(ring->doorbell_index, ring->wptr);
-		} else {
-			WREG32_SOC15(GC, 0, regCP_RB0_WPTR,
-				     lower_32_bits(ring->wptr));
-			WREG32_SOC15(GC, 0, regCP_RB0_WPTR_HI,
-				     upper_32_bits(ring->wptr));
-		}
+		WREG32_SOC15(GC, 0, regCP_RB0_WPTR,
+			     lower_32_bits(ring->wptr));
+		WREG32_SOC15(GC, 0, regCP_RB0_WPTR_HI,
+			     upper_32_bits(ring->wptr));
 	}
 }
 
@@ -4276,42 +4372,14 @@ static u64 gfx_v12_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
 static void gfx_v12_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
-	uint32_t *wptr_saved;
-	uint32_t *is_queue_unmap;
-	uint64_t aggregated_db_index;
-	uint32_t mqd_size = adev->mqds[AMDGPU_HW_IP_COMPUTE].mqd_size;
-	uint64_t wptr_tmp;
-
-	if (ring->is_mes_queue) {
-		wptr_saved = (uint32_t *)(ring->mqd_ptr + mqd_size);
-		is_queue_unmap = (uint32_t *)(ring->mqd_ptr + mqd_size +
-					      sizeof(uint32_t));
-		aggregated_db_index =
-			amdgpu_mes_get_aggregated_doorbell_index(adev,
-								 ring->hw_prio);
-
-		wptr_tmp = ring->wptr & ring->buf_mask;
-		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, wptr_tmp);
-		*wptr_saved = wptr_tmp;
-		/* assume doorbell always used by mes mapped queue */
-		if (*is_queue_unmap) {
-			WDOORBELL64(aggregated_db_index, wptr_tmp);
-			WDOORBELL64(ring->doorbell_index, wptr_tmp);
-		} else {
-			WDOORBELL64(ring->doorbell_index, wptr_tmp);
 
-			if (*is_queue_unmap)
-				WDOORBELL64(aggregated_db_index, wptr_tmp);
-		}
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell) {
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
+			     ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
 	} else {
-		/* XXX check if swapping is necessary on BE */
-		if (ring->use_doorbell) {
-			atomic64_set((atomic64_t *)ring->wptr_cpu_addr,
-				     ring->wptr);
-			WDOORBELL64(ring->doorbell_index, ring->wptr);
-		} else {
-			BUG(); /* only DOORBELL method supported on gfx12 now */
-		}
+		BUG(); /* only DOORBELL method supported on gfx12 now */
 	}
 }
 
@@ -4352,16 +4420,10 @@ static void gfx_v12_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	u32 header, control = 0;
 
-	BUG_ON(ib->flags & AMDGPU_IB_FLAG_CE);
-
 	header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
 
 	control |= ib->length_dw | (vmid << 24);
 
-	if (ring->is_mes_queue)
-		/* inherit vmid from mqd */
-		control |= 0x400000;
-
 	amdgpu_ring_write(ring, header);
 	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
 	amdgpu_ring_write(ring,
@@ -4381,10 +4443,6 @@ static void gfx_v12_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
 	unsigned vmid = AMDGPU_JOB_GET_VMID(job);
 	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
 
-	if (ring->is_mes_queue)
-		/* inherit vmid from mqd */
-		control |= 0x40000000;
-
 	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
 	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
 	amdgpu_ring_write(ring,
@@ -4424,8 +4482,7 @@ static void gfx_v12_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
 	amdgpu_ring_write(ring, upper_32_bits(addr));
 	amdgpu_ring_write(ring, lower_32_bits(seq));
 	amdgpu_ring_write(ring, upper_32_bits(seq));
-	amdgpu_ring_write(ring, ring->is_mes_queue ?
-			 (ring->hw_queue_id | AMDGPU_FENCE_MES_QUEUE_FLAG) : 0);
+	amdgpu_ring_write(ring, 0);
 }
 
 static void gfx_v12_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
@@ -4453,10 +4510,7 @@ static void gfx_v12_0_ring_invalidate_tlbs(struct amdgpu_ring *ring,
 static void gfx_v12_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 					 unsigned vmid, uint64_t pd_addr)
 {
-	if (ring->is_mes_queue)
-		gfx_v12_0_ring_invalidate_tlbs(ring, 0, 0, false, 0);
-	else
-		amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
 
 	/* compute doesn't have PFP */
 	if (ring->funcs->type == AMDGPU_RING_TYPE_GFX) {
@@ -4647,21 +4701,6 @@ static void gfx_v12_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
 			       ref, mask, 0x20);
 }
 
-static void gfx_v12_0_ring_soft_recovery(struct amdgpu_ring *ring,
-					 unsigned vmid)
-{
-	struct amdgpu_device *adev = ring->adev;
-	uint32_t value = 0;
-
-	value = REG_SET_FIELD(value, SQ_CMD, CMD, 0x03);
-	value = REG_SET_FIELD(value, SQ_CMD, MODE, 0x01);
-	value = REG_SET_FIELD(value, SQ_CMD, CHECK_VMID, 1);
-	value = REG_SET_FIELD(value, SQ_CMD, VM_ID, vmid);
-	amdgpu_gfx_rlc_enter_safe_mode(adev, 0);
-	WREG32_SOC15(GC, 0, regSQ_CMD, value);
-	amdgpu_gfx_rlc_exit_safe_mode(adev, 0);
-}
-
 static void
 gfx_v12_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
 				      uint32_t me, uint32_t pipe,
@@ -4790,25 +4829,23 @@ static int gfx_v12_0_eop_irq(struct amdgpu_device *adev,
 			     struct amdgpu_irq_src *source,
 			     struct amdgpu_iv_entry *entry)
 {
-	int i;
+	u32 doorbell_offset = entry->src_data[0];
 	u8 me_id, pipe_id, queue_id;
 	struct amdgpu_ring *ring;
-	uint32_t mes_queue_id = entry->src_data[0];
+	int i;
 
 	DRM_DEBUG("IH: CP EOP\n");
 
-	if (adev->enable_mes && (mes_queue_id & AMDGPU_FENCE_MES_QUEUE_FLAG)) {
-		struct amdgpu_mes_queue *queue;
+	if (adev->enable_mes && doorbell_offset) {
+		struct amdgpu_userq_fence_driver *fence_drv = NULL;
+		struct xarray *xa = &adev->userq_xa;
+		unsigned long flags;
 
-		mes_queue_id &= AMDGPU_FENCE_MES_QUEUE_ID_MASK;
-
-		spin_lock(&adev->mes.queue_id_lock);
-		queue = idr_find(&adev->mes.queue_id_idr, mes_queue_id);
-		if (queue) {
-			DRM_DEBUG("process mes queue id = %d\n", mes_queue_id);
-			amdgpu_fence_process(queue->ring);
-		}
-		spin_unlock(&adev->mes.queue_id_lock);
+		xa_lock_irqsave(xa, flags);
+		fence_drv = xa_load(xa, doorbell_offset);
+		if (fence_drv)
+			amdgpu_userq_fence_driver_process(fence_drv);
+		xa_unlock_irqrestore(xa, flags);
 	} else {
 		me_id = (entry->ring_id & 0x0c) >> 2;
 		pipe_id = (entry->ring_id & 0x03) >> 0;
@@ -4975,27 +5012,29 @@ static void gfx_v12_0_handle_priv_fault(struct amdgpu_device *adev,
 	pipe_id = (entry->ring_id & 0x03) >> 0;
 	queue_id = (entry->ring_id & 0x70) >> 4;
 
-	switch (me_id) {
-	case 0:
-		for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
-			ring = &adev->gfx.gfx_ring[i];
-			if (ring->me == me_id && ring->pipe == pipe_id &&
-			    ring->queue == queue_id)
-				drm_sched_fault(&ring->sched);
-		}
-		break;
-	case 1:
-	case 2:
-		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-			ring = &adev->gfx.compute_ring[i];
-			if (ring->me == me_id && ring->pipe == pipe_id &&
-			    ring->queue == queue_id)
-				drm_sched_fault(&ring->sched);
+	if (!adev->gfx.disable_kq) {
+		switch (me_id) {
+		case 0:
+			for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+				ring = &adev->gfx.gfx_ring[i];
+				if (ring->me == me_id && ring->pipe == pipe_id &&
+				    ring->queue == queue_id)
+					drm_sched_fault(&ring->sched);
+			}
+			break;
+		case 1:
+		case 2:
+			for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+				ring = &adev->gfx.compute_ring[i];
+				if (ring->me == me_id && ring->pipe == pipe_id &&
+				    ring->queue == queue_id)
+					drm_sched_fault(&ring->sched);
+			}
+			break;
+		default:
+			BUG();
+			break;
 		}
-		break;
-	default:
-		BUG();
-		break;
 	}
 }
 
@@ -5201,34 +5240,89 @@ static void gfx_v12_ip_dump(struct amdgpu_ip_block *ip_block)
 	amdgpu_gfx_off_ctrl(adev, true);
 }
 
-static int gfx_v12_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid)
+static bool gfx_v12_pipe_reset_support(struct amdgpu_device *adev)
+{
+	/* Disable the pipe reset until the CPFW fully support it.*/
+	dev_warn_once(adev->dev, "The CPFW hasn't support pipe reset yet.\n");
+	return false;
+}
+
+static int gfx_v12_reset_gfx_pipe(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
+	uint32_t reset_pipe = 0, clean_pipe = 0;
 	int r;
 
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
+	if (!gfx_v12_pipe_reset_support(adev))
+		return -EOPNOTSUPP;
+
+	gfx_v12_0_set_safe_mode(adev, 0);
+	mutex_lock(&adev->srbm_mutex);
+	soc24_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+
+	switch (ring->pipe) {
+	case 0:
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   PFP_PIPE0_RESET, 1);
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   ME_PIPE0_RESET, 1);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   PFP_PIPE0_RESET, 0);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   ME_PIPE0_RESET, 0);
+		break;
+	case 1:
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   PFP_PIPE1_RESET, 1);
+		reset_pipe = REG_SET_FIELD(reset_pipe, CP_ME_CNTL,
+					   ME_PIPE1_RESET, 1);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   PFP_PIPE1_RESET, 0);
+		clean_pipe = REG_SET_FIELD(clean_pipe, CP_ME_CNTL,
+					   ME_PIPE1_RESET, 0);
+		break;
+	default:
+		break;
+	}
+
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, reset_pipe);
+	WREG32_SOC15(GC, 0, regCP_ME_CNTL, clean_pipe);
+
+	r = (RREG32(SOC15_REG_OFFSET(GC, 0, regCP_GFX_RS64_INSTR_PNTR1)) << 2) -
+					RS64_FW_UC_START_ADDR_LO;
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	gfx_v12_0_unset_safe_mode(adev, 0);
+
+	dev_info(adev->dev, "The ring %s pipe reset: %s\n", ring->name,
+			r == 0 ? "successfully" : "failed");
+	/* Sometimes the ME start pc counter can't cache correctly, so the
+	 * PC check only as a reference and pipe reset result rely on the
+	 * later ring test.
+	 */
+	return 0;
+}
+
+static int gfx_v12_0_reset_kgq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
 
 	r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, false);
 	if (r) {
-		dev_err(adev->dev, "reset via MES failed %d\n", r);
-		return r;
+		dev_warn(adev->dev, "reset via MES failed and try pipe reset %d\n", r);
+		r = gfx_v12_reset_gfx_pipe(ring);
+		if (r)
+			return r;
 	}
 
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0)) {
-		dev_err(adev->dev, "fail to resv mqd_obj\n");
-		return r;
-	}
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (!r) {
-		r = gfx_v12_0_kgq_init_queue(ring, true);
-		amdgpu_bo_kunmap(ring->mqd_obj);
-		ring->mqd_ptr = NULL;
-	}
-	amdgpu_bo_unreserve(ring->mqd_obj);
+	r = gfx_v12_0_kgq_init_queue(ring, true);
 	if (r) {
-		DRM_ERROR("fail to unresv mqd_obj\n");
+		dev_err(adev->dev, "failed to init kgq\n");
 		return r;
 	}
 
@@ -5238,37 +5332,112 @@ static int gfx_v12_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid)
 		return r;
 	}
 
-	return amdgpu_ring_test_ring(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static int gfx_v12_0_reset_compute_pipe(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reset_pipe = 0, clean_pipe = 0;
+	int r = 0;
+
+	if (!gfx_v12_pipe_reset_support(adev))
+		return -EOPNOTSUPP;
+
+	gfx_v12_0_set_safe_mode(adev, 0);
+	mutex_lock(&adev->srbm_mutex);
+	soc24_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+
+	reset_pipe = RREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL);
+	clean_pipe = reset_pipe;
+
+	if (adev->gfx.rs64_enable) {
+		switch (ring->pipe) {
+		case 0:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE0_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE0_RESET, 0);
+			break;
+		case 1:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE1_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE1_RESET, 0);
+			break;
+		case 2:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE2_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE2_RESET, 0);
+			break;
+		case 3:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE3_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_RS64_CNTL,
+						   MEC_PIPE3_RESET, 0);
+			break;
+		default:
+			break;
+		}
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, reset_pipe);
+		WREG32_SOC15(GC, 0, regCP_MEC_RS64_CNTL, clean_pipe);
+		r = (RREG32_SOC15(GC, 0, regCP_MEC_RS64_INSTR_PNTR) << 2) -
+				RS64_FW_UC_START_ADDR_LO;
+	} else {
+		switch (ring->pipe) {
+		case 0:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE0_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE0_RESET, 0);
+			break;
+		case 1:
+			reset_pipe = REG_SET_FIELD(reset_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE1_RESET, 1);
+			clean_pipe = REG_SET_FIELD(clean_pipe, CP_MEC_CNTL,
+							   MEC_ME1_PIPE1_RESET, 0);
+			break;
+		default:
+		break;
+		}
+		WREG32_SOC15(GC, 0, regCP_MEC_CNTL, reset_pipe);
+		WREG32_SOC15(GC, 0, regCP_MEC_CNTL, clean_pipe);
+		/* Doesn't find the F32 MEC instruction pointer register, and suppose
+		 * the driver won't run into the F32 mode.
+		 */
+	}
+
+	soc24_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+	gfx_v12_0_unset_safe_mode(adev, 0);
+
+	dev_info(adev->dev, "The ring %s pipe resets: %s\n", ring->name,
+			r == 0 ? "successfully" : "failed");
+	/* Need the ring test to verify the pipe reset result.*/
+	return 0;
 }
 
-static int gfx_v12_0_reset_kcq(struct amdgpu_ring *ring, unsigned int vmid)
+static int gfx_v12_0_reset_kcq(struct amdgpu_ring *ring,
+			       unsigned int vmid,
+			       struct amdgpu_fence *timedout_fence)
 {
 	struct amdgpu_device *adev = ring->adev;
 	int r;
 
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
 
 	r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, true);
 	if (r) {
-		dev_err(adev->dev, "reset via MMIO failed %d\n", r);
-		return r;
+		dev_warn(adev->dev, "fail(%d) to reset kcq  and try pipe reset\n", r);
+		r = gfx_v12_0_reset_compute_pipe(ring);
+		if (r)
+			return r;
 	}
 
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0)) {
-		DRM_ERROR("fail to resv mqd_obj\n");
-		return r;
-	}
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (!r) {
-		r = gfx_v12_0_kcq_init_queue(ring, true);
-		amdgpu_bo_kunmap(ring->mqd_obj);
-		ring->mqd_ptr = NULL;
-	}
-	amdgpu_bo_unreserve(ring->mqd_obj);
+	r = gfx_v12_0_kcq_init_queue(ring, true);
 	if (r) {
-		DRM_ERROR("fail to unresv mqd_obj\n");
+		dev_err(adev->dev, "failed to init kcq\n");
 		return r;
 	}
 	r = amdgpu_mes_map_legacy_queue(adev, ring);
@@ -5277,7 +5446,21 @@ static int gfx_v12_0_reset_kcq(struct amdgpu_ring *ring, unsigned int vmid)
 		return r;
 	}
 
-	return amdgpu_ring_test_ring(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
+static void gfx_v12_0_ring_begin_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_begin_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_begin_use(ring);
+}
+
+static void gfx_v12_0_ring_end_use(struct amdgpu_ring *ring)
+{
+	amdgpu_gfx_profile_ring_end_use(ring);
+
+	amdgpu_gfx_enforce_isolation_ring_end_use(ring);
 }
 
 static const struct amd_ip_funcs gfx_v12_0_ip_funcs = {
@@ -5341,12 +5524,11 @@ static const struct amdgpu_ring_funcs gfx_v12_0_ring_funcs_gfx = {
 	.emit_wreg = gfx_v12_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v12_0_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v12_0_ring_emit_reg_write_reg_wait,
-	.soft_recovery = gfx_v12_0_ring_soft_recovery,
 	.emit_mem_sync = gfx_v12_0_emit_mem_sync,
 	.reset = gfx_v12_0_reset_kgq,
 	.emit_cleaner_shader = gfx_v12_0_ring_emit_cleaner_shader,
-	.begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use,
-	.end_use = amdgpu_gfx_enforce_isolation_ring_end_use,
+	.begin_use = gfx_v12_0_ring_begin_use,
+	.end_use = gfx_v12_0_ring_end_use,
 };
 
 static const struct amdgpu_ring_funcs gfx_v12_0_ring_funcs_compute = {
@@ -5380,12 +5562,11 @@ static const struct amdgpu_ring_funcs gfx_v12_0_ring_funcs_compute = {
 	.emit_wreg = gfx_v12_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v12_0_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v12_0_ring_emit_reg_write_reg_wait,
-	.soft_recovery = gfx_v12_0_ring_soft_recovery,
 	.emit_mem_sync = gfx_v12_0_emit_mem_sync,
 	.reset = gfx_v12_0_reset_kcq,
 	.emit_cleaner_shader = gfx_v12_0_ring_emit_cleaner_shader,
-	.begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use,
-	.end_use = amdgpu_gfx_enforce_isolation_ring_end_use,
+	.begin_use = gfx_v12_0_ring_begin_use,
+	.end_use = gfx_v12_0_ring_end_use,
 };
 
 static const struct amdgpu_ring_funcs gfx_v12_0_ring_funcs_kiq = {
@@ -5415,6 +5596,7 @@ static const struct amdgpu_ring_funcs gfx_v12_0_ring_funcs_kiq = {
 	.emit_wreg = gfx_v12_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v12_0_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v12_0_ring_emit_reg_write_reg_wait,
+	.emit_hdp_flush = gfx_v12_0_ring_emit_hdp_flush,
 };
 
 static void gfx_v12_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
index f26e2cdec07a..80565392313f 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
@@ -28,19 +28,33 @@
 #include "amdgpu_gfx.h"
 #include "amdgpu_ucode.h"
 #include "clearstate_si.h"
+#include "si.h"
+#include "sid.h"
+
 #include "bif/bif_3_0_d.h"
 #include "bif/bif_3_0_sh_mask.h"
+
 #include "oss/oss_1_0_d.h"
 #include "oss/oss_1_0_sh_mask.h"
+
 #include "gca/gfx_6_0_d.h"
 #include "gca/gfx_6_0_sh_mask.h"
+#include "gca/gfx_7_2_enum.h"
+
 #include "gmc/gmc_6_0_d.h"
 #include "gmc/gmc_6_0_sh_mask.h"
+
 #include "dce/dce_6_0_d.h"
 #include "dce/dce_6_0_sh_mask.h"
-#include "gca/gfx_7_2_enum.h"
+
 #include "si_enums.h"
-#include "si.h"
+
+#define TAHITI_GB_ADDR_CONFIG_GOLDEN        0x12011003
+#define VERDE_GB_ADDR_CONFIG_GOLDEN         0x12010002
+#define HAINAN_GB_ADDR_CONFIG_GOLDEN        0x02010001
+
+#define GFX6_NUM_GFX_RINGS     1
+#define GFX6_NUM_COMPUTE_RINGS 2
 
 static void gfx_v6_0_set_ring_funcs(struct amdgpu_device *adev);
 static void gfx_v6_0_set_irq_funcs(struct amdgpu_device *adev);
@@ -72,7 +86,7 @@ MODULE_FIRMWARE("amdgpu/hainan_ce.bin");
 MODULE_FIRMWARE("amdgpu/hainan_rlc.bin");
 
 static u32 gfx_v6_0_get_csb_size(struct amdgpu_device *adev);
-static void gfx_v6_0_get_csb_buffer(struct amdgpu_device *adev, volatile u32 *buffer);
+static void gfx_v6_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer);
 //static void gfx_v6_0_init_cp_pg_table(struct amdgpu_device *adev);
 static void gfx_v6_0_init_pg(struct amdgpu_device *adev);
 
@@ -1721,10 +1735,14 @@ static void gfx_v6_0_constants_init(struct amdgpu_device *adev)
 	gfx_v6_0_get_cu_info(adev);
 	gfx_v6_0_config_init(adev);
 
-	WREG32(mmCP_QUEUE_THRESHOLDS, ((0x16 << CP_QUEUE_THRESHOLDS__ROQ_IB1_START__SHIFT) |
-				       (0x2b << CP_QUEUE_THRESHOLDS__ROQ_IB2_START__SHIFT)));
-	WREG32(mmCP_MEQ_THRESHOLDS, (0x30 << CP_MEQ_THRESHOLDS__MEQ1_START__SHIFT) |
-				    (0x60 << CP_MEQ_THRESHOLDS__MEQ2_START__SHIFT));
+	WREG32(mmCP_QUEUE_THRESHOLDS,
+		((0x16 << CP_QUEUE_THRESHOLDS__ROQ_IB1_START__SHIFT) |
+		(0x2b << CP_QUEUE_THRESHOLDS__ROQ_IB2_START__SHIFT)));
+
+	/* set HW defaults for 3D engine */
+	WREG32(mmCP_MEQ_THRESHOLDS,
+		(0x30 << CP_MEQ_THRESHOLDS__MEQ1_START__SHIFT) |
+		(0x60 << CP_MEQ_THRESHOLDS__MEQ2_START__SHIFT));
 
 	sx_debug_1 = RREG32(mmSX_DEBUG_1);
 	WREG32(mmSX_DEBUG_1, sx_debug_1);
@@ -2336,7 +2354,7 @@ static void gfx_v6_0_ring_emit_wreg(struct amdgpu_ring *ring,
 static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
 {
 	const u32 *src_ptr;
-	volatile u32 *dst_ptr;
+	u32 *dst_ptr;
 	u32 dws;
 	u64 reg_list_mc_addr;
 	const struct cs_section_def *cs_data;
@@ -2837,47 +2855,23 @@ static u32 gfx_v6_0_get_csb_size(struct amdgpu_device *adev)
 	return count;
 }
 
-static void gfx_v6_0_get_csb_buffer(struct amdgpu_device *adev,
-				    volatile u32 *buffer)
+static void gfx_v6_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
-	u32 count = 0, i;
-	const struct cs_section_def *sect = NULL;
-	const struct cs_extent_def *ext = NULL;
+	u32 count = 0;
 
 	if (adev->gfx.rlc.cs_data == NULL)
 		return;
 	if (buffer == NULL)
 		return;
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
-	buffer[count++] = cpu_to_le32(0x80000000);
-	buffer[count++] = cpu_to_le32(0x80000000);
-
-	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
-		for (ext = sect->section; ext->extent != NULL; ++ext) {
-			if (sect->id == SECT_CONTEXT) {
-				buffer[count++] =
-					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
-				buffer[count++] = cpu_to_le32(ext->reg_index - 0xa000);
-				for (i = 0; i < ext->reg_count; i++)
-					buffer[count++] = cpu_to_le32(ext->extent[i]);
-			} else {
-				return;
-			}
-		}
-	}
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
 
 	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 1));
 	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
 	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config);
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
-	buffer[count++] = cpu_to_le32(0);
+	amdgpu_gfx_csb_preamble_end(buffer, count);
 }
 
 static void gfx_v6_0_init_pg(struct amdgpu_device *adev)
@@ -3108,6 +3102,11 @@ static int gfx_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
 			return r;
 	}
 
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+
 	return r;
 }
 
@@ -3167,9 +3166,9 @@ static int gfx_v6_0_resume(struct amdgpu_ip_block *ip_block)
 	return gfx_v6_0_hw_init(ip_block);
 }
 
-static bool gfx_v6_0_is_idle(void *handle)
+static bool gfx_v6_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK)
 		return false;
@@ -3183,7 +3182,7 @@ static int gfx_v6_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (gfx_v6_0_is_idle(adev))
+		if (gfx_v6_0_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index 84745b2453ab..2b7aba22ecc1 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -55,6 +55,9 @@
 #define GFX7_NUM_GFX_RINGS     1
 #define GFX7_MEC_HPD_SIZE      2048
 
+#define BONAIRE_GB_ADDR_CONFIG_GOLDEN        0x12010001
+#define HAWAII_GB_ADDR_CONFIG_GOLDEN         0x12011003
+
 static void gfx_v7_0_set_ring_funcs(struct amdgpu_device *adev);
 static void gfx_v7_0_set_irq_funcs(struct amdgpu_device *adev);
 static void gfx_v7_0_set_gds_init(struct amdgpu_device *adev);
@@ -880,7 +883,7 @@ static const u32 kalindi_rlc_save_restore_register_list[] = {
 };
 
 static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev);
-static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, volatile u32 *buffer);
+static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer);
 static void gfx_v7_0_init_pg(struct amdgpu_device *adev);
 static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev);
 
@@ -3879,70 +3882,24 @@ static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev)
 	return count;
 }
 
-static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev,
-				    volatile u32 *buffer)
+static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
-	u32 count = 0, i;
-	const struct cs_section_def *sect = NULL;
-	const struct cs_extent_def *ext = NULL;
+	u32 count = 0;
 
 	if (adev->gfx.rlc.cs_data == NULL)
 		return;
 	if (buffer == NULL)
 		return;
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
-	buffer[count++] = cpu_to_le32(0x80000000);
-	buffer[count++] = cpu_to_le32(0x80000000);
-
-	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
-		for (ext = sect->section; ext->extent != NULL; ++ext) {
-			if (sect->id == SECT_CONTEXT) {
-				buffer[count++] =
-					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
-				buffer[count++] = cpu_to_le32(ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
-				for (i = 0; i < ext->reg_count; i++)
-					buffer[count++] = cpu_to_le32(ext->extent[i]);
-			} else {
-				return;
-			}
-		}
-	}
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
 
 	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 2));
 	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
-	switch (adev->asic_type) {
-	case CHIP_BONAIRE:
-		buffer[count++] = cpu_to_le32(0x16000012);
-		buffer[count++] = cpu_to_le32(0x00000000);
-		break;
-	case CHIP_KAVERI:
-		buffer[count++] = cpu_to_le32(0x00000000); /* XXX */
-		buffer[count++] = cpu_to_le32(0x00000000);
-		break;
-	case CHIP_KABINI:
-	case CHIP_MULLINS:
-		buffer[count++] = cpu_to_le32(0x00000000); /* XXX */
-		buffer[count++] = cpu_to_le32(0x00000000);
-		break;
-	case CHIP_HAWAII:
-		buffer[count++] = cpu_to_le32(0x3a00161a);
-		buffer[count++] = cpu_to_le32(0x0000002e);
-		break;
-	default:
-		buffer[count++] = cpu_to_le32(0x00000000);
-		buffer[count++] = cpu_to_le32(0x00000000);
-		break;
-	}
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
+	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config);
+	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config_1);
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
-	buffer[count++] = cpu_to_le32(0);
+	amdgpu_gfx_csb_preamble_end(buffer, count);
 }
 
 static void gfx_v7_0_init_pg(struct amdgpu_device *adev)
@@ -4442,6 +4399,11 @@ static int gfx_v7_0_sw_init(struct amdgpu_ip_block *ip_block)
 
 	gfx_v7_0_gpu_early_init(adev);
 
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+
 	return r;
 }
 
@@ -4515,9 +4477,9 @@ static int gfx_v7_0_resume(struct amdgpu_ip_block *ip_block)
 	return gfx_v7_0_hw_init(ip_block);
 }
 
-static bool gfx_v7_0_is_idle(void *handle)
+static bool gfx_v7_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK)
 		return false;
@@ -4926,76 +4888,6 @@ static void gfx_v7_0_emit_mem_sync_compute(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, 0x0000000A);	/* poll interval */
 }
 
-static void gfx_v7_0_wait_reg_mem(struct amdgpu_ring *ring, int eng_sel,
-				  int mem_space, int opt, uint32_t addr0,
-				  uint32_t addr1, uint32_t ref, uint32_t mask,
-				  uint32_t inv)
-{
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
-	amdgpu_ring_write(ring,
-			  /* memory (1) or register (0) */
-			  (WAIT_REG_MEM_MEM_SPACE(mem_space) |
-			   WAIT_REG_MEM_OPERATION(opt) | /* wait */
-			   WAIT_REG_MEM_FUNCTION(3) |  /* equal */
-			   WAIT_REG_MEM_ENGINE(eng_sel)));
-
-	if (mem_space)
-		BUG_ON(addr0 & 0x3); /* Dword align */
-	amdgpu_ring_write(ring, addr0);
-	amdgpu_ring_write(ring, addr1);
-	amdgpu_ring_write(ring, ref);
-	amdgpu_ring_write(ring, mask);
-	amdgpu_ring_write(ring, inv); /* poll interval */
-}
-
-static void gfx_v7_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
-					uint32_t val, uint32_t mask)
-{
-	gfx_v7_0_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20);
-}
-
-static int gfx_v7_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid)
-{
-	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
-	struct amdgpu_ring *kiq_ring = &kiq->ring;
-	unsigned long flags;
-	u32 tmp;
-	int r;
-
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
-
-	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
-		return -EINVAL;
-
-	spin_lock_irqsave(&kiq->ring_lock, flags);
-
-	if (amdgpu_ring_alloc(kiq_ring, 5)) {
-		spin_unlock_irqrestore(&kiq->ring_lock, flags);
-		return -ENOMEM;
-	}
-
-	tmp = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid);
-	gfx_v7_0_ring_emit_wreg(kiq_ring, mmCP_VMID_RESET, tmp);
-	amdgpu_ring_commit(kiq_ring);
-
-	spin_unlock_irqrestore(&kiq->ring_lock, flags);
-
-	r = amdgpu_ring_test_ring(kiq_ring);
-	if (r)
-		return r;
-
-	if (amdgpu_ring_alloc(ring, 7 + 12 + 5))
-		return -ENOMEM;
-	gfx_v7_0_ring_emit_fence_gfx(ring, ring->fence_drv.gpu_addr,
-				     ring->fence_drv.sync_seq, AMDGPU_FENCE_FLAG_EXEC);
-	gfx_v7_0_ring_emit_reg_wait(ring, mmCP_VMID_RESET, 0, 0xffff);
-	gfx_v7_0_ring_emit_wreg(ring, mmCP_VMID_RESET, 0);
-
-	return amdgpu_ring_test_ring(ring);
-}
-
 static const struct amd_ip_funcs gfx_v7_0_ip_funcs = {
 	.name = "gfx_v7_0",
 	.early_init = gfx_v7_0_early_init,
@@ -5045,7 +4937,6 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = {
 	.emit_wreg = gfx_v7_0_ring_emit_wreg,
 	.soft_recovery = gfx_v7_0_ring_soft_recovery,
 	.emit_mem_sync = gfx_v7_0_emit_mem_sync,
-	.reset = gfx_v7_0_reset_kgq,
 };
 
 static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 6a025438f9d0..1c87375e1dd5 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -1220,51 +1220,24 @@ out:
 	return err;
 }
 
-static void gfx_v8_0_get_csb_buffer(struct amdgpu_device *adev,
-				    volatile u32 *buffer)
+static void gfx_v8_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
-	u32 count = 0, i;
-	const struct cs_section_def *sect = NULL;
-	const struct cs_extent_def *ext = NULL;
+	u32 count = 0;
 
 	if (adev->gfx.rlc.cs_data == NULL)
 		return;
 	if (buffer == NULL)
 		return;
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
-	buffer[count++] = cpu_to_le32(0x80000000);
-	buffer[count++] = cpu_to_le32(0x80000000);
-
-	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
-		for (ext = sect->section; ext->extent != NULL; ++ext) {
-			if (sect->id == SECT_CONTEXT) {
-				buffer[count++] =
-					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
-				buffer[count++] = cpu_to_le32(ext->reg_index -
-						PACKET3_SET_CONTEXT_REG_START);
-				for (i = 0; i < ext->reg_count; i++)
-					buffer[count++] = cpu_to_le32(ext->extent[i]);
-			} else {
-				return;
-			}
-		}
-	}
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
 
 	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 2));
-	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG -
-			PACKET3_SET_CONTEXT_REG_START);
+	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
 	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config);
 	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config_1);
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
-	buffer[count++] = cpu_to_le32(0);
+	amdgpu_gfx_csb_preamble_end(buffer, count);
 }
 
 static int gfx_v8_0_cp_jump_table_num(struct amdgpu_device *adev)
@@ -2050,6 +2023,11 @@ static int gfx_v8_0_sw_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+
 	return 0;
 }
 
@@ -4666,6 +4644,7 @@ static int gfx_v8_0_kcq_init_queue(struct amdgpu_ring *ring)
 			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct vi_mqd_allocation));
 		/* reset ring buffer */
 		ring->wptr = 0;
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, 0);
 		amdgpu_ring_clear_ring(ring);
 	}
 	return 0;
@@ -4683,60 +4662,25 @@ static void gfx_v8_0_set_mec_doorbell_range(struct amdgpu_device *adev)
 
 static int gfx_v8_0_kiq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring;
-	int r;
-
-	ring = &adev->gfx.kiq[0].ring;
-
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0))
-		return r;
-
-	r = amdgpu_bo_kmap(ring->mqd_obj, &ring->mqd_ptr);
-	if (unlikely(r != 0)) {
-		amdgpu_bo_unreserve(ring->mqd_obj);
-		return r;
-	}
-
-	gfx_v8_0_kiq_init_queue(ring);
-	amdgpu_bo_kunmap(ring->mqd_obj);
-	ring->mqd_ptr = NULL;
-	amdgpu_bo_unreserve(ring->mqd_obj);
+	gfx_v8_0_kiq_init_queue(&adev->gfx.kiq[0].ring);
 	return 0;
 }
 
 static int gfx_v8_0_kcq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = NULL;
-	int r = 0, i;
+	int i, r;
 
 	gfx_v8_0_cp_compute_enable(adev, true);
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		ring = &adev->gfx.compute_ring[i];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			goto done;
-		r = amdgpu_bo_kmap(ring->mqd_obj, &ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v8_0_kcq_init_queue(ring);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
+		r = gfx_v8_0_kcq_init_queue(&adev->gfx.compute_ring[i]);
 		if (r)
-			goto done;
+			return r;
 	}
 
 	gfx_v8_0_set_mec_doorbell_range(adev);
 
-	r = gfx_v8_0_kiq_kcq_enable(adev);
-	if (r)
-		goto done;
-
-done:
-	return r;
+	return gfx_v8_0_kiq_kcq_enable(adev);
 }
 
 static int gfx_v8_0_cp_test_all_rings(struct amdgpu_device *adev)
@@ -4851,9 +4795,9 @@ static int gfx_v8_0_kcq_disable(struct amdgpu_device *adev)
 	return r;
 }
 
-static bool gfx_v8_0_is_idle(void *handle)
+static bool gfx_v8_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (REG_GET_FIELD(RREG32(mmGRBM_STATUS), GRBM_STATUS, GUI_ACTIVE)
 		|| RREG32(mmGRBM_STATUS2) != 0x8)
@@ -4892,7 +4836,7 @@ static int gfx_v8_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (gfx_v8_0_is_idle(adev))
+		if (gfx_v8_0_is_idle(ip_block))
 			return 0;
 
 		udelay(1);
@@ -5452,9 +5396,9 @@ static int gfx_v8_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
-static void gfx_v8_0_get_clockgating_state(void *handle, u64 *flags)
+static void gfx_v8_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
@@ -6400,34 +6344,6 @@ static void gfx_v8_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
 	amdgpu_ring_write(ring, val);
 }
 
-static void gfx_v8_0_wait_reg_mem(struct amdgpu_ring *ring, int eng_sel,
-				  int mem_space, int opt, uint32_t addr0,
-				  uint32_t addr1, uint32_t ref, uint32_t mask,
-				  uint32_t inv)
-{
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
-	amdgpu_ring_write(ring,
-			  /* memory (1) or register (0) */
-			  (WAIT_REG_MEM_MEM_SPACE(mem_space) |
-			   WAIT_REG_MEM_OPERATION(opt) | /* wait */
-			   WAIT_REG_MEM_FUNCTION(3) |  /* equal */
-			   WAIT_REG_MEM_ENGINE(eng_sel)));
-
-	if (mem_space)
-		BUG_ON(addr0 & 0x3); /* Dword align */
-	amdgpu_ring_write(ring, addr0);
-	amdgpu_ring_write(ring, addr1);
-	amdgpu_ring_write(ring, ref);
-	amdgpu_ring_write(ring, mask);
-	amdgpu_ring_write(ring, inv); /* poll interval */
-}
-
-static void gfx_v8_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
-					uint32_t val, uint32_t mask)
-{
-	gfx_v8_0_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20);
-}
-
 static void gfx_v8_0_ring_soft_recovery(struct amdgpu_ring *ring, unsigned vmid)
 {
 	struct amdgpu_device *adev = ring->adev;
@@ -6904,48 +6820,6 @@ static void gfx_v8_0_emit_wave_limit(struct amdgpu_ring *ring, bool enable)
 
 }
 
-static int gfx_v8_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid)
-{
-	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
-	struct amdgpu_ring *kiq_ring = &kiq->ring;
-	unsigned long flags;
-	u32 tmp;
-	int r;
-
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
-
-	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
-		return -EINVAL;
-
-	spin_lock_irqsave(&kiq->ring_lock, flags);
-
-	if (amdgpu_ring_alloc(kiq_ring, 5)) {
-		spin_unlock_irqrestore(&kiq->ring_lock, flags);
-		return -ENOMEM;
-	}
-
-	tmp = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid);
-	gfx_v8_0_ring_emit_wreg(kiq_ring, mmCP_VMID_RESET, tmp);
-	amdgpu_ring_commit(kiq_ring);
-
-	spin_unlock_irqrestore(&kiq->ring_lock, flags);
-
-	r = amdgpu_ring_test_ring(kiq_ring);
-	if (r)
-		return r;
-
-	if (amdgpu_ring_alloc(ring, 7 + 12 + 5))
-		return -ENOMEM;
-	gfx_v8_0_ring_emit_fence_gfx(ring, ring->fence_drv.gpu_addr,
-				     ring->fence_drv.sync_seq, AMDGPU_FENCE_FLAG_EXEC);
-	gfx_v8_0_ring_emit_reg_wait(ring, mmCP_VMID_RESET, 0, 0xffff);
-	gfx_v8_0_ring_emit_wreg(ring, mmCP_VMID_RESET, 0);
-
-	return amdgpu_ring_test_ring(ring);
-}
-
 static const struct amd_ip_funcs gfx_v8_0_ip_funcs = {
 	.name = "gfx_v8_0",
 	.early_init = gfx_v8_0_early_init,
@@ -7011,7 +6885,6 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
 	.emit_wreg = gfx_v8_0_ring_emit_wreg,
 	.soft_recovery = gfx_v8_0_ring_soft_recovery,
 	.emit_mem_sync = gfx_v8_0_emit_mem_sync,
-	.reset = gfx_v8_0_reset_kgq,
 };
 
 static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
@@ -7071,6 +6944,7 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_kiq = {
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.emit_rreg = gfx_v8_0_ring_emit_rreg,
 	.emit_wreg = gfx_v8_0_ring_emit_wreg,
+	.emit_hdp_flush = gfx_v8_0_ring_emit_hdp_flush,
 };
 
 static void gfx_v8_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index 0dce4421418c..0148d7ff34d9 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -225,17 +225,36 @@ static const struct amdgpu_hwip_reg_entry gc_reg_list_9[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_SMU_SAFE_MODE),
 	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_INT_STAT),
 	SOC15_REG_ENTRY_STR(GC, 0, mmRLC_GPM_GENERAL_6),
-	/* cp header registers */
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME2_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
 	/* SE status registers */
 	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE0),
 	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE1),
 	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE2),
-	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE3)
+	SOC15_REG_ENTRY_STR(GC, 0, mmGRBM_STATUS_SE3),
+	/* packet headers */
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_CE_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_PFP_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_ME_HEADER_DUMP)
 };
 
 static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_9[] = {
@@ -277,6 +296,14 @@ static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_9[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_WPTR_LO),
 	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_PQ_WPTR_HI),
 	SOC15_REG_ENTRY_STR(GC, 0, mmCP_HQD_GFX_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, mmCP_MEC_ME1_HEADER_DUMP)
 };
 
 enum ta_ras_gfx_subblock {
@@ -1269,6 +1296,7 @@ static void gfx_v9_0_check_fw_write_wait(struct amdgpu_device *adev)
 	adev->gfx.mec_fw_write_wait = false;
 
 	if ((amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 1)) &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 2)) &&
 	    ((adev->gfx.mec_fw_version < 0x000001a5) ||
 	     (adev->gfx.mec_feature_version < 46) ||
 	     (adev->gfx.pfp_fw_version < 0x000000b7) ||
@@ -1620,45 +1648,18 @@ static u32 gfx_v9_0_get_csb_size(struct amdgpu_device *adev)
 	return count;
 }
 
-static void gfx_v9_0_get_csb_buffer(struct amdgpu_device *adev,
-				    volatile u32 *buffer)
+static void gfx_v9_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer)
 {
-	u32 count = 0, i;
-	const struct cs_section_def *sect = NULL;
-	const struct cs_extent_def *ext = NULL;
+	u32 count = 0;
 
 	if (adev->gfx.rlc.cs_data == NULL)
 		return;
 	if (buffer == NULL)
 		return;
 
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
-	buffer[count++] = cpu_to_le32(0x80000000);
-	buffer[count++] = cpu_to_le32(0x80000000);
-
-	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
-		for (ext = sect->section; ext->extent != NULL; ++ext) {
-			if (sect->id == SECT_CONTEXT) {
-				buffer[count++] =
-					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
-				buffer[count++] = cpu_to_le32(ext->reg_index -
-						PACKET3_SET_CONTEXT_REG_START);
-				for (i = 0; i < ext->reg_count; i++)
-					buffer[count++] = cpu_to_le32(ext->extent[i]);
-			} else {
-				return;
-			}
-		}
-	}
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
-	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
-
-	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
-	buffer[count++] = cpu_to_le32(0);
+	count = amdgpu_gfx_csb_preamble_start(buffer);
+	count = amdgpu_gfx_csb_data_parser(adev, buffer, count);
+	amdgpu_gfx_csb_preamble_end(buffer, count);
 }
 
 static void gfx_v9_0_init_always_on_cu_mask(struct amdgpu_device *adev)
@@ -2233,6 +2234,25 @@ static int gfx_v9_0_sw_init(struct amdgpu_ip_block *ip_block)
 	}
 
 	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 0, 1):
+	case IP_VERSION(9, 2, 1):
+	case IP_VERSION(9, 4, 0):
+	case IP_VERSION(9, 2, 2):
+	case IP_VERSION(9, 1, 0):
+	case IP_VERSION(9, 3, 0):
+		adev->gfx.cleaner_shader_ptr = gfx_9_4_2_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_9_4_2_cleaner_shader_hex);
+		if (adev->gfx.me_fw_version  >= 167 &&
+		    adev->gfx.pfp_fw_version >= 196 &&
+		    adev->gfx.mec_fw_version >= 474) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
 	case IP_VERSION(9, 4, 2):
 		adev->gfx.cleaner_shader_ptr = gfx_9_4_2_cleaner_shader_hex;
 		adev->gfx.cleaner_shader_size = sizeof(gfx_9_4_2_cleaner_shader_hex);
@@ -2389,6 +2409,8 @@ static int gfx_v9_0_sw_init(struct amdgpu_ip_block *ip_block)
 		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
 	adev->gfx.compute_supported_reset =
 		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+	if (!amdgpu_sriov_vf(adev) && !adev->debug_disable_gpu_ring_reset)
+		adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
 
 	r = amdgpu_gfx_kiq_init(adev, GFX9_MEC_HPD_SIZE, 0);
 	if (r) {
@@ -2627,6 +2649,9 @@ static void gfx_v9_0_init_sq_config(struct amdgpu_device *adev)
 				!READ_ONCE(adev->barrier_has_auto_waitcnt));
 		WREG32_SOC15(GC, 0, mmSQ_CONFIG, tmp);
 		break;
+	case IP_VERSION(9, 4, 2):
+		gfx_v9_4_2_init_sq(adev);
+		break;
 	default:
 		break;
 	}
@@ -2637,7 +2662,10 @@ static void gfx_v9_0_constants_init(struct amdgpu_device *adev)
 	u32 tmp;
 	int i;
 
-	WREG32_FIELD15_RLC(GC, 0, GRBM_CNTL, READ_TIMEOUT, 0xff);
+	if (!amdgpu_sriov_vf(adev) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 2)) {
+		WREG32_FIELD15_RLC(GC, 0, GRBM_CNTL, READ_TIMEOUT, 0xff);
+	}
 
 	gfx_v9_0_tiling_mode_table_init(adev);
 
@@ -3887,55 +3915,23 @@ static int gfx_v9_0_kcq_init_queue(struct amdgpu_ring *ring, bool restore)
 
 static int gfx_v9_0_kiq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring;
-	int r;
-
-	ring = &adev->gfx.kiq[0].ring;
-
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0))
-		return r;
-
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (unlikely(r != 0)) {
-		amdgpu_bo_unreserve(ring->mqd_obj);
-		return r;
-	}
-
-	gfx_v9_0_kiq_init_queue(ring);
-	amdgpu_bo_kunmap(ring->mqd_obj);
-	ring->mqd_ptr = NULL;
-	amdgpu_bo_unreserve(ring->mqd_obj);
+	gfx_v9_0_kiq_init_queue(&adev->gfx.kiq[0].ring);
 	return 0;
 }
 
 static int gfx_v9_0_kcq_resume(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring = NULL;
-	int r = 0, i;
+	int i, r;
 
 	gfx_v9_0_cp_compute_enable(adev, true);
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		ring = &adev->gfx.compute_ring[i];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			goto done;
-		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v9_0_kcq_init_queue(ring, false);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
+		r = gfx_v9_0_kcq_init_queue(&adev->gfx.compute_ring[i], false);
 		if (r)
-			goto done;
+			return r;
 	}
 
-	r = amdgpu_gfx_enable_kcq(adev, 0);
-done:
-	return r;
+	return amdgpu_gfx_enable_kcq(adev, 0);
 }
 
 static int gfx_v9_0_cp_resume(struct amdgpu_device *adev)
@@ -4042,7 +4038,8 @@ static int gfx_v9_0_hw_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2))
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) &&
+	    !amdgpu_sriov_vf(adev))
 		gfx_v9_4_2_set_power_brake_sequence(adev);
 
 	return r;
@@ -4110,9 +4107,9 @@ static int gfx_v9_0_resume(struct amdgpu_ip_block *ip_block)
 	return gfx_v9_0_hw_init(ip_block);
 }
 
-static bool gfx_v9_0_is_idle(void *handle)
+static bool gfx_v9_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (REG_GET_FIELD(RREG32_SOC15(GC, 0, mmGRBM_STATUS),
 				GRBM_STATUS, GUI_ACTIVE))
@@ -4127,7 +4124,7 @@ static int gfx_v9_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (gfx_v9_0_is_idle(adev))
+		if (gfx_v9_0_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
@@ -4177,19 +4174,17 @@ static int gfx_v9_0_soft_reset(struct amdgpu_ip_block *ip_block)
 		/* Disable MEC parsing/prefetching */
 		gfx_v9_0_cp_compute_enable(adev, false);
 
-		if (grbm_soft_reset) {
-			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
-			tmp |= grbm_soft_reset;
-			dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
-			WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
-			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+		tmp |= grbm_soft_reset;
+		dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
 
-			udelay(50);
+		udelay(50);
 
-			tmp &= ~grbm_soft_reset;
-			WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
-			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
-		}
+		tmp &= ~grbm_soft_reset;
+		WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
 
 		/* Wait a little for things to settle down */
 		udelay(50);
@@ -5241,7 +5236,7 @@ static int gfx_v9_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 	case IP_VERSION(9, 1, 0):
 	case IP_VERSION(9, 3, 0):
 		if (!enable)
-			amdgpu_gfx_off_ctrl(adev, false);
+			amdgpu_gfx_off_ctrl_immediate(adev, false);
 
 		if (adev->pg_flags & AMD_PG_SUPPORT_RLC_SMU_HS) {
 			gfx_v9_0_enable_sck_slow_down_on_power_up(adev, true);
@@ -5263,10 +5258,10 @@ static int gfx_v9_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 		gfx_v9_0_update_gfx_mg_power_gating(adev, enable);
 
 		if (enable)
-			amdgpu_gfx_off_ctrl(adev, true);
+			amdgpu_gfx_off_ctrl_immediate(adev, true);
 		break;
 	case IP_VERSION(9, 2, 1):
-		amdgpu_gfx_off_ctrl(adev, enable);
+		amdgpu_gfx_off_ctrl_immediate(adev, enable);
 		break;
 	default:
 		break;
@@ -5301,9 +5296,9 @@ static int gfx_v9_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
-static void gfx_v9_0_get_clockgating_state(void *handle, u64 *flags)
+static void gfx_v9_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
@@ -5468,16 +5463,8 @@ static void gfx_v9_0_ring_patch_ce_meta(struct amdgpu_ring *ring,
 
 	payload_size = sizeof(struct v9_ce_ib_state);
 
-	if (ring->is_mes_queue) {
-		payload_offset = offsetof(struct amdgpu_mes_ctx_meta_data,
-					  gfx[0].gfx_meta_data) +
-			offsetof(struct v9_gfx_meta_data, ce_payload);
-		ce_payload_cpu_addr =
-			amdgpu_mes_ctx_get_offs_cpu_addr(ring, payload_offset);
-	} else {
-		payload_offset = offsetof(struct v9_gfx_meta_data, ce_payload);
-		ce_payload_cpu_addr = adev->virt.csa_cpu_addr + payload_offset;
-	}
+	payload_offset = offsetof(struct v9_gfx_meta_data, ce_payload);
+	ce_payload_cpu_addr = adev->virt.csa_cpu_addr + payload_offset;
 
 	if (offset + (payload_size >> 2) <= ring->buf_mask + 1) {
 		memcpy((void *)&ring->ring[offset], ce_payload_cpu_addr, payload_size);
@@ -5500,16 +5487,8 @@ static void gfx_v9_0_ring_patch_de_meta(struct amdgpu_ring *ring,
 
 	payload_size = sizeof(struct v9_de_ib_state);
 
-	if (ring->is_mes_queue) {
-		payload_offset = offsetof(struct amdgpu_mes_ctx_meta_data,
-					  gfx[0].gfx_meta_data) +
-			offsetof(struct v9_gfx_meta_data, de_payload);
-		de_payload_cpu_addr =
-			amdgpu_mes_ctx_get_offs_cpu_addr(ring, payload_offset);
-	} else {
-		payload_offset = offsetof(struct v9_gfx_meta_data, de_payload);
-		de_payload_cpu_addr = adev->virt.csa_cpu_addr + payload_offset;
-	}
+	payload_offset = offsetof(struct v9_gfx_meta_data, de_payload);
+	de_payload_cpu_addr = adev->virt.csa_cpu_addr + payload_offset;
 
 	((struct v9_de_ib_state *)de_payload_cpu_addr)->ib_completion_status =
 		IB_COMPLETION_STATUS_PREEMPTED;
@@ -5699,19 +5678,9 @@ static void gfx_v9_0_ring_emit_ce_meta(struct amdgpu_ring *ring, bool resume)
 
 	cnt = (sizeof(ce_payload) >> 2) + 4 - 2;
 
-	if (ring->is_mes_queue) {
-		offset = offsetof(struct amdgpu_mes_ctx_meta_data,
-				  gfx[0].gfx_meta_data) +
-			offsetof(struct v9_gfx_meta_data, ce_payload);
-		ce_payload_gpu_addr =
-			amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
-		ce_payload_cpu_addr =
-			amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset);
-	} else {
-		offset = offsetof(struct v9_gfx_meta_data, ce_payload);
-		ce_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
-		ce_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
-	}
+	offset = offsetof(struct v9_gfx_meta_data, ce_payload);
+	ce_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
+	ce_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
 
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt));
 	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(2) |
@@ -5797,28 +5766,13 @@ static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume, bo
 	void *de_payload_cpu_addr;
 	int cnt;
 
-	if (ring->is_mes_queue) {
-		offset = offsetof(struct amdgpu_mes_ctx_meta_data,
-				  gfx[0].gfx_meta_data) +
-			offsetof(struct v9_gfx_meta_data, de_payload);
-		de_payload_gpu_addr =
-			amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
-		de_payload_cpu_addr =
-			amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset);
-
-		offset = offsetof(struct amdgpu_mes_ctx_meta_data,
-				  gfx[0].gds_backup) +
-			offsetof(struct v9_gfx_meta_data, de_payload);
-		gds_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
-	} else {
-		offset = offsetof(struct v9_gfx_meta_data, de_payload);
-		de_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
-		de_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
+	offset = offsetof(struct v9_gfx_meta_data, de_payload);
+	de_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
+	de_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
 
-		gds_addr = ALIGN(amdgpu_csa_vaddr(ring->adev) +
-				 AMDGPU_CSA_SIZE - adev->gds.gds_size,
-				 PAGE_SIZE);
-	}
+	gds_addr = ALIGN(amdgpu_csa_vaddr(ring->adev) +
+			 AMDGPU_CSA_SIZE - adev->gds.gds_size,
+			 PAGE_SIZE);
 
 	if (usegds) {
 		de_payload.gds_backup_addrlo = lower_32_bits(gds_addr);
@@ -7219,53 +7173,9 @@ static void gfx_v9_ring_insert_nop(struct amdgpu_ring *ring, uint32_t num_nop)
 	amdgpu_ring_insert_nop(ring, num_nop - 1);
 }
 
-static int gfx_v9_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid)
-{
-	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
-	struct amdgpu_ring *kiq_ring = &kiq->ring;
-	unsigned long flags;
-	u32 tmp;
-	int r;
-
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
-
-	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
-		return -EINVAL;
-
-	spin_lock_irqsave(&kiq->ring_lock, flags);
-
-	if (amdgpu_ring_alloc(kiq_ring, 5)) {
-		spin_unlock_irqrestore(&kiq->ring_lock, flags);
-		return -ENOMEM;
-	}
-
-	tmp = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid);
-	gfx_v9_0_ring_emit_wreg(kiq_ring,
-				 SOC15_REG_OFFSET(GC, 0, mmCP_VMID_RESET), tmp);
-	amdgpu_ring_commit(kiq_ring);
-
-	spin_unlock_irqrestore(&kiq->ring_lock, flags);
-
-	r = amdgpu_ring_test_ring(kiq_ring);
-	if (r)
-		return r;
-
-	if (amdgpu_ring_alloc(ring, 7 + 7 + 5))
-		return -ENOMEM;
-	gfx_v9_0_ring_emit_fence(ring, ring->fence_drv.gpu_addr,
-				 ring->fence_drv.sync_seq, AMDGPU_FENCE_FLAG_EXEC);
-	gfx_v9_0_ring_emit_reg_wait(ring,
-				    SOC15_REG_OFFSET(GC, 0, mmCP_VMID_RESET), 0, 0xffff);
-	gfx_v9_0_ring_emit_wreg(ring,
-				SOC15_REG_OFFSET(GC, 0, mmCP_VMID_RESET), 0);
-
-	return amdgpu_ring_test_ring(ring);
-}
-
 static int gfx_v9_0_reset_kcq(struct amdgpu_ring *ring,
-			      unsigned int vmid)
+			      unsigned int vmid,
+			      struct amdgpu_fence *timedout_fence)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_kiq *kiq = &adev->gfx.kiq[0];
@@ -7273,12 +7183,11 @@ static int gfx_v9_0_reset_kcq(struct amdgpu_ring *ring,
 	unsigned long flags;
 	int i, r;
 
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
-
 	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
 		return -EINVAL;
 
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
 	spin_lock_irqsave(&kiq->ring_lock, flags);
 
 	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
@@ -7315,20 +7224,9 @@ static int gfx_v9_0_reset_kcq(struct amdgpu_ring *ring,
 		return r;
 	}
 
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0)){
-		dev_err(adev->dev, "fail to resv mqd_obj\n");
-		return r;
-	}
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (!r) {
-		r = gfx_v9_0_kcq_init_queue(ring, true);
-		amdgpu_bo_kunmap(ring->mqd_obj);
-		ring->mqd_ptr = NULL;
-	}
-	amdgpu_bo_unreserve(ring->mqd_obj);
+	r = gfx_v9_0_kcq_init_queue(ring, true);
 	if (r) {
-		dev_err(adev->dev, "fail to unresv mqd_obj\n");
+		dev_err(adev->dev, "fail to init kcq\n");
 		return r;
 	}
 	spin_lock_irqsave(&kiq->ring_lock, flags);
@@ -7339,13 +7237,13 @@ static int gfx_v9_0_reset_kcq(struct amdgpu_ring *ring,
 	}
 	kiq->pmf->kiq_map_queues(kiq_ring, ring);
 	amdgpu_ring_commit(kiq_ring);
-	spin_unlock_irqrestore(&kiq->ring_lock, flags);
 	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
 	if (r) {
 		DRM_ERROR("fail to remap queue\n");
 		return r;
 	}
-	return amdgpu_ring_test_ring(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
 }
 
 static void gfx_v9_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
@@ -7377,9 +7275,14 @@ static void gfx_v9_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer
 			for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
 				drm_printf(p, "\nmec %d, pipe %d, queue %d\n", i, j, k);
 				for (reg = 0; reg < reg_count; reg++) {
-					drm_printf(p, "%-50s \t 0x%08x\n",
-						   gc_cp_reg_list_9[reg].reg_name,
-						   adev->gfx.ip_dump_compute_queues[index + reg]);
+					if (i && gc_cp_reg_list_9[reg].reg_offset == mmCP_MEC_ME1_HEADER_DUMP)
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   "mmCP_MEC_ME2_HEADER_DUMP",
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
+					else
+						drm_printf(p, "%-50s \t 0x%08x\n",
+							   gc_cp_reg_list_9[reg].reg_name,
+							   adev->gfx.ip_dump_compute_queues[index + reg]);
 				}
 				index += reg_count;
 			}
@@ -7416,9 +7319,13 @@ static void gfx_v9_ip_dump(struct amdgpu_ip_block *ip_block)
 				soc15_grbm_select(adev, 1 + i, j, k, 0, 0);
 
 				for (reg = 0; reg < reg_count; reg++) {
-					adev->gfx.ip_dump_compute_queues[index + reg] =
-						RREG32(SOC15_REG_ENTRY_OFFSET(
-							gc_cp_reg_list_9[reg]));
+					if (i && gc_cp_reg_list_9[reg].reg_offset == mmCP_MEC_ME1_HEADER_DUMP)
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME2_HEADER_DUMP));
+					else
+						adev->gfx.ip_dump_compute_queues[index + reg] =
+							RREG32(SOC15_REG_ENTRY_OFFSET(
+								       gc_cp_reg_list_9[reg]));
 				}
 				index += reg_count;
 			}
@@ -7432,8 +7339,14 @@ static void gfx_v9_ip_dump(struct amdgpu_ip_block *ip_block)
 
 static void gfx_v9_0_ring_emit_cleaner_shader(struct amdgpu_ring *ring)
 {
+	struct amdgpu_device *adev = ring->adev;
+
 	/* Emit the cleaner shader */
-	amdgpu_ring_write(ring, PACKET3(PACKET3_RUN_CLEANER_SHADER, 0));
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2))
+		amdgpu_ring_write(ring, PACKET3(PACKET3_RUN_CLEANER_SHADER, 0));
+	else
+		amdgpu_ring_write(ring, PACKET3(PACKET3_RUN_CLEANER_SHADER_9_0, 0));
+
 	amdgpu_ring_write(ring, 0);  /* RESERVED field, programmed to zero */
 }
 
@@ -7540,7 +7453,6 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = {
 	.emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait,
 	.soft_recovery = gfx_v9_0_ring_soft_recovery,
 	.emit_mem_sync = gfx_v9_0_emit_mem_sync,
-	.reset = gfx_v9_0_reset_kgq,
 	.emit_cleaner_shader = gfx_v9_0_ring_emit_cleaner_shader,
 	.begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use,
 	.end_use = amdgpu_gfx_enforce_isolation_ring_end_use,
@@ -7674,6 +7586,7 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_kiq = {
 	.emit_wreg = gfx_v9_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v9_0_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait,
+	.emit_hdp_flush = gfx_v9_0_ring_emit_hdp_flush,
 };
 
 static void gfx_v9_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c
index f53b379d8971..6028afd81690 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c
@@ -27,7 +27,6 @@
 #include "amdgpu_gfx.h"
 #include "soc15.h"
 #include "soc15d.h"
-#include "amdgpu_atomfirmware.h"
 #include "amdgpu_pm.h"
 
 #include "gc/gc_9_4_1_offset.h"
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c
index d81449f9d822..8058ea91ecaf 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c
@@ -748,6 +748,18 @@ void gfx_v9_4_2_init_golden_registers(struct amdgpu_device *adev,
 	}
 }
 
+void gfx_v9_4_2_init_sq(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	if (adev->gfx.mec_fw_version >= 98) {
+		adev->gmc.xnack_flags |= AMDGPU_GMC_XNACK_FLAG_CHAIN;
+		data = RREG32_SOC15(GC, 0, regSQ_CONFIG1);
+		data = REG_SET_FIELD(data, SQ_CONFIG1, DISABLE_XNACK_CHECK_IN_RETRY_DISABLE, 1);
+		WREG32_SOC15(GC, 0, regSQ_CONFIG1, data);
+	}
+}
+
 void gfx_v9_4_2_debug_trap_config_init(struct amdgpu_device *adev,
 				uint32_t first_vmid,
 				uint32_t last_vmid)
@@ -1547,7 +1559,7 @@ static void gfx_v9_4_2_log_utc_edc_count(struct amdgpu_device *adev,
 {
 	uint32_t bank, way, mem;
 	static const char * const vml2_way_str[] = { "BIGK", "4K" };
-	static const char * const utcl2_rounter_str[] = { "VMC", "APT" };
+	static const char * const utcl2_router_str[] = { "VMC", "APT" };
 
 	mem = instance % blk->num_mem_blocks;
 	way = (instance / blk->num_mem_blocks) % blk->num_ways;
@@ -1568,7 +1580,7 @@ static void gfx_v9_4_2_log_utc_edc_count(struct amdgpu_device *adev,
 		dev_info(
 			adev->dev,
 			"GFX SubBlock UTCL2_ROUTER_IFIF%d_GROUP0_%s, SED %d, DED %d\n",
-			bank, utcl2_rounter_str[mem], sec_cnt, ded_cnt);
+			bank, utcl2_router_str[mem], sec_cnt, ded_cnt);
 		break;
 	case ATC_L2_CACHE_2M:
 		dev_info(
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h
index 7584624b641c..a603724c1dfc 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h
@@ -28,6 +28,7 @@ void gfx_v9_4_2_debug_trap_config_init(struct amdgpu_device *adev,
 				uint32_t first_vmid, uint32_t last_vmid);
 void gfx_v9_4_2_init_golden_registers(struct amdgpu_device *adev,
 				      uint32_t die_id);
+void gfx_v9_4_2_init_sq(struct amdgpu_device *adev);
 void gfx_v9_4_2_set_power_brake_sequence(struct amdgpu_device *adev);
 int gfx_v9_4_2_do_edc_gpr_workarounds(struct amdgpu_device *adev);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
index 2ba185875baa..cbb74ffc4792 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
@@ -105,9 +105,6 @@ static const struct amdgpu_hwip_reg_entry gc_reg_list_9_4_3[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, regRLC_SMU_SAFE_MODE),
 	SOC15_REG_ENTRY_STR(GC, 0, regRLC_INT_STAT),
 	SOC15_REG_ENTRY_STR(GC, 0, regRLC_GPM_GENERAL_6),
-	/* cp header registers */
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
-	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME2_HEADER_DUMP),
 	/* SE status registers */
 	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE0),
 	SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE1),
@@ -154,6 +151,14 @@ static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_9_4_3[] = {
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_LO),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_HI),
 	SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_GFX_STATUS),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
+	SOC15_REG_ENTRY_STR(GC, 0, regCP_MEC_ME1_HEADER_DUMP),
 };
 
 struct amdgpu_gfx_ras gfx_v9_4_3_ras;
@@ -349,18 +354,7 @@ static void gfx_v9_4_3_init_golden_registers(struct amdgpu_device *adev)
 
 		WREG32_SOC15(GC, dev_inst, regGB_ADDR_CONFIG,
 			     GOLDEN_GB_ADDR_CONFIG);
-		if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0)) {
-			WREG32_FIELD15_PREREG(GC, dev_inst, TCP_UTCL1_CNTL2, SPARE, 0x1);
-		} else {
-			/* Golden settings applied by driver for ASIC with rev_id 0 */
-			if (adev->rev_id == 0) {
-				WREG32_FIELD15_PREREG(GC, dev_inst, TCP_UTCL1_CNTL1,
-						      REDUCE_FIFO_DEPTH_BY_2, 2);
-			} else {
-				WREG32_FIELD15_PREREG(GC, dev_inst, TCP_UTCL1_CNTL2,
-						      SPARE, 0x1);
-			}
-		}
+		WREG32_FIELD15_PREREG(GC, dev_inst, TCP_UTCL1_CNTL2, SPARE, 0x1);
 	}
 }
 
@@ -563,17 +557,6 @@ out:
 	return err;
 }
 
-static bool gfx_v9_4_3_should_disable_gfxoff(struct pci_dev *pdev)
-{
-	return true;
-}
-
-static void gfx_v9_4_3_check_if_need_gfxoff(struct amdgpu_device *adev)
-{
-	if (gfx_v9_4_3_should_disable_gfxoff(adev->pdev))
-		adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
-}
-
 static int gfx_v9_4_3_init_cp_compute_microcode(struct amdgpu_device *adev,
 					  const char *chip_name)
 {
@@ -600,8 +583,6 @@ static int gfx_v9_4_3_init_cp_compute_microcode(struct amdgpu_device *adev,
 	adev->gfx.mec2_fw_version = adev->gfx.mec_fw_version;
 	adev->gfx.mec2_feature_version = adev->gfx.mec_feature_version;
 
-	gfx_v9_4_3_check_if_need_gfxoff(adev);
-
 out:
 	if (err)
 		amdgpu_ucode_release(&adev->gfx.mec_fw);
@@ -891,12 +872,13 @@ static int gfx_v9_4_3_aca_bank_parser(struct aca_handle *handle,
 
 	switch (type) {
 	case ACA_SMU_TYPE_UE:
-		ret = aca_error_cache_log_bank_error(handle, &info,
-						     ACA_ERROR_TYPE_UE, 1ULL);
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type, 1ULL);
 		break;
 	case ACA_SMU_TYPE_CE:
-		ret = aca_error_cache_log_bank_error(handle, &info,
-						     ACA_ERROR_TYPE_CE, ACA_REG__MISC0__ERRCNT(misc0));
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type,
+						     ACA_REG__MISC0__ERRCNT(misc0));
 		break;
 	default:
 		return -EINVAL;
@@ -937,28 +919,15 @@ static const struct aca_info gfx_v9_4_3_aca_info = {
 
 static int gfx_v9_4_3_gpu_early_init(struct amdgpu_device *adev)
 {
-	u32 gb_addr_config;
-
 	adev->gfx.funcs = &gfx_v9_4_3_gfx_funcs;
 	adev->gfx.ras = &gfx_v9_4_3_ras;
 
-	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
-	case IP_VERSION(9, 4, 3):
-	case IP_VERSION(9, 4, 4):
-	case IP_VERSION(9, 5, 0):
-		adev->gfx.config.max_hw_contexts = 8;
-		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
-		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
-		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
-		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
-		gb_addr_config = RREG32_SOC15(GC, GET_INST(GC, 0), regGB_ADDR_CONFIG);
-		break;
-	default:
-		BUG();
-		break;
-	}
-
-	adev->gfx.config.gb_addr_config = gb_addr_config;
+	adev->gfx.config.max_hw_contexts = 8;
+	adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+	adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+	adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+	adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+	adev->gfx.config.gb_addr_config = GOLDEN_GB_ADDR_CONFIG;
 
 	adev->gfx.config.gb_addr_config_fields.num_pipes = 1 <<
 			REG_GET_FIELD(
@@ -1179,7 +1148,17 @@ static int gfx_v9_4_3_sw_init(struct amdgpu_ip_block *ip_block)
 	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
 	case IP_VERSION(9, 4, 3):
 	case IP_VERSION(9, 4, 4):
-		if (adev->gfx.mec_fw_version >= 155) {
+		if ((adev->gfx.mec_fw_version >= 155) &&
+		    !amdgpu_sriov_vf(adev) &&
+		    !adev->debug_disable_gpu_ring_reset) {
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_PIPE;
+		}
+		break;
+	case IP_VERSION(9, 5, 0):
+		if ((adev->gfx.mec_fw_version >= 21) &&
+		    !amdgpu_sriov_vf(adev) &&
+		    !adev->debug_disable_gpu_ring_reset) {
 			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
 			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_PIPE;
 		}
@@ -1298,6 +1277,22 @@ static void gfx_v9_4_3_xcc_init_gds_vmid(struct amdgpu_device *adev, int xcc_id)
 	}
 }
 
+/* For ASICs that needs xnack chain and MEC version supports, set SG_CONFIG1
+ * DISABLE_XNACK_CHECK_IN_RETRY_DISABLE bit and inform KFD to set xnack_chain
+ * bit in SET_RESOURCES
+ */
+static void gfx_v9_4_3_xcc_init_sq(struct amdgpu_device *adev, int xcc_id)
+{
+	uint32_t data;
+
+	if (!(adev->gmc.xnack_flags & AMDGPU_GMC_XNACK_FLAG_CHAIN))
+		return;
+
+	data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regSQ_CONFIG1);
+	data = REG_SET_FIELD(data, SQ_CONFIG1, DISABLE_XNACK_CHECK_IN_RETRY_DISABLE, 1);
+	WREG32_SOC15(GC, xcc_id, regSQ_CONFIG1, data);
+}
+
 static void gfx_v9_4_3_xcc_constants_init(struct amdgpu_device *adev,
 					  int xcc_id)
 {
@@ -1342,6 +1337,7 @@ static void gfx_v9_4_3_xcc_constants_init(struct amdgpu_device *adev,
 
 	gfx_v9_4_3_xcc_init_compute_vmid(adev, xcc_id);
 	gfx_v9_4_3_xcc_init_gds_vmid(adev, xcc_id);
+	gfx_v9_4_3_xcc_init_sq(adev, xcc_id);
 }
 
 static void gfx_v9_4_3_constants_init(struct amdgpu_device *adev)
@@ -1354,6 +1350,22 @@ static void gfx_v9_4_3_constants_init(struct amdgpu_device *adev)
 	adev->gfx.config.db_debug2 =
 		RREG32_SOC15(GC, GET_INST(GC, 0), regDB_DEBUG2);
 
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	/* ToDo: GC 9.4.4 */
+	case IP_VERSION(9, 4, 3):
+		if (adev->gfx.mec_fw_version >= 184 &&
+		    (amdgpu_sriov_reg_access_sq_config(adev) ||
+		     !amdgpu_sriov_vf(adev)))
+			adev->gmc.xnack_flags |= AMDGPU_GMC_XNACK_FLAG_CHAIN;
+		break;
+	case IP_VERSION(9, 5, 0):
+		if (adev->gfx.mec_fw_version >= 23)
+			adev->gmc.xnack_flags |= AMDGPU_GMC_XNACK_FLAG_CHAIN;
+		break;
+	default:
+		break;
+	}
+
 	for (i = 0; i < num_xcc; i++)
 		gfx_v9_4_3_xcc_constants_init(adev, i);
 }
@@ -1369,10 +1381,8 @@ static void gfx_v9_4_3_xcc_init_pg(struct amdgpu_device *adev, int xcc_id)
 {
 	/*
 	 * Rlc save restore list is workable since v2_1.
-	 * And it's needed by gfxoff feature.
 	 */
-	if (adev->gfx.rlc.is_rlc_v2_1)
-		gfx_v9_4_3_xcc_enable_save_restore_machine(adev, xcc_id);
+	gfx_v9_4_3_xcc_enable_save_restore_machine(adev, xcc_id);
 }
 
 static void gfx_v9_4_3_xcc_disable_gpa_mode(struct amdgpu_device *adev, int xcc_id)
@@ -1857,7 +1867,7 @@ static int gfx_v9_4_3_xcc_mqd_init(struct amdgpu_ring *ring, int xcc_id)
 				    DOORBELL_SOURCE, 0);
 		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
 				    DOORBELL_HIT, 0);
-		if (amdgpu_sriov_vf(adev))
+		if (amdgpu_sriov_multi_vf_mode(adev))
 			tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
 					    DOORBELL_MODE, 1);
 	} else {
@@ -2144,7 +2154,8 @@ static int gfx_v9_4_3_xcc_kiq_init_queue(struct amdgpu_ring *ring, int xcc_id)
 	return 0;
 }
 
-static int gfx_v9_4_3_xcc_kcq_init_queue(struct amdgpu_ring *ring, int xcc_id, bool restore)
+static void gfx_v9_4_3_xcc_kcq_init_queue(struct amdgpu_ring *ring, int xcc_id,
+					  bool restore)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct v9_mqd *mqd = ring->mqd_ptr;
@@ -2178,8 +2189,6 @@ static int gfx_v9_4_3_xcc_kcq_init_queue(struct amdgpu_ring *ring, int xcc_id, b
 		atomic64_set((atomic64_t *)&adev->wb.wb[ring->wptr_offs], 0);
 		amdgpu_ring_clear_ring(ring);
 	}
-
-	return 0;
 }
 
 static int gfx_v9_4_3_xcc_kcq_fini_register(struct amdgpu_device *adev, int xcc_id)
@@ -2205,55 +2214,25 @@ static int gfx_v9_4_3_xcc_kcq_fini_register(struct amdgpu_device *adev, int xcc_
 
 static int gfx_v9_4_3_xcc_kiq_resume(struct amdgpu_device *adev, int xcc_id)
 {
-	struct amdgpu_ring *ring;
-	int r;
-
-	ring = &adev->gfx.kiq[xcc_id].ring;
-
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0))
-		return r;
-
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (unlikely(r != 0)) {
-		amdgpu_bo_unreserve(ring->mqd_obj);
-		return r;
-	}
-
-	gfx_v9_4_3_xcc_kiq_init_queue(ring, xcc_id);
-	amdgpu_bo_kunmap(ring->mqd_obj);
-	ring->mqd_ptr = NULL;
-	amdgpu_bo_unreserve(ring->mqd_obj);
+	gfx_v9_4_3_xcc_kiq_init_queue(&adev->gfx.kiq[xcc_id].ring, xcc_id);
 	return 0;
 }
 
 static int gfx_v9_4_3_xcc_kcq_resume(struct amdgpu_device *adev, int xcc_id)
 {
-	struct amdgpu_ring *ring = NULL;
-	int r = 0, i;
+	struct amdgpu_ring *ring;
+	int i;
 
 	gfx_v9_4_3_xcc_cp_compute_enable(adev, true, xcc_id);
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
-		ring = &adev->gfx.compute_ring[i + xcc_id * adev->gfx.num_compute_rings];
-
-		r = amdgpu_bo_reserve(ring->mqd_obj, false);
-		if (unlikely(r != 0))
-			goto done;
-		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-		if (!r) {
-			r = gfx_v9_4_3_xcc_kcq_init_queue(ring, xcc_id, false);
-			amdgpu_bo_kunmap(ring->mqd_obj);
-			ring->mqd_ptr = NULL;
-		}
-		amdgpu_bo_unreserve(ring->mqd_obj);
-		if (r)
-			goto done;
+		ring = &adev->gfx.compute_ring[i + xcc_id *
+			adev->gfx.num_compute_rings];
+
+		gfx_v9_4_3_xcc_kcq_init_queue(ring, xcc_id, false);
 	}
 
-	r = amdgpu_gfx_enable_kcq(adev, xcc_id);
-done:
-	return r;
+	return amdgpu_gfx_enable_kcq(adev, xcc_id);
 }
 
 static int gfx_v9_4_3_xcc_cp_resume(struct amdgpu_device *adev, int xcc_id)
@@ -2312,7 +2291,9 @@ static int gfx_v9_4_3_cp_resume(struct amdgpu_device *adev)
 		r = amdgpu_xcp_init(adev->xcp_mgr, num_xcp, mode);
 
 	} else {
-		if (amdgpu_xcp_query_partition_mode(adev->xcp_mgr,
+		if (adev->in_suspend)
+			amdgpu_xcp_restore_partition_mode(adev->xcp_mgr);
+		else if (amdgpu_xcp_query_partition_mode(adev->xcp_mgr,
 						    AMDGPU_XCP_FL_NONE) ==
 		    AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE)
 			r = amdgpu_xcp_switch_partition_mode(
@@ -2415,9 +2396,9 @@ static int gfx_v9_4_3_resume(struct amdgpu_ip_block *ip_block)
 	return gfx_v9_4_3_hw_init(ip_block);
 }
 
-static bool gfx_v9_4_3_is_idle(void *handle)
+static bool gfx_v9_4_3_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int i, num_xcc;
 
 	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
@@ -2435,7 +2416,7 @@ static int gfx_v9_4_3_wait_for_idle(struct amdgpu_ip_block *ip_block)
 	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (gfx_v9_4_3_is_idle(adev))
+		if (gfx_v9_4_3_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
@@ -2481,19 +2462,17 @@ static int gfx_v9_4_3_soft_reset(struct amdgpu_ip_block *ip_block)
 		/* Disable MEC parsing/prefetching */
 		gfx_v9_4_3_xcc_cp_compute_enable(adev, false, 0);
 
-		if (grbm_soft_reset) {
-			tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET);
-			tmp |= grbm_soft_reset;
-			dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
-			WREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET, tmp);
-			tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET);
+		tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET);
+		tmp |= grbm_soft_reset;
+		dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET);
 
-			udelay(50);
+		udelay(50);
 
-			tmp &= ~grbm_soft_reset;
-			WREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET, tmp);
-			tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET);
-		}
+		tmp &= ~grbm_soft_reset;
+		WREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET, tmp);
+		tmp = RREG32_SOC15(GC, GET_INST(GC, 0), regGRBM_SOFT_RESET);
 
 		/* Wait a little for things to settle down */
 		udelay(50);
@@ -2795,22 +2774,16 @@ static int gfx_v9_4_3_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 		return 0;
 
 	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
-	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
-	case IP_VERSION(9, 4, 3):
-	case IP_VERSION(9, 4, 4):
-		for (i = 0; i < num_xcc; i++)
-			gfx_v9_4_3_xcc_update_gfx_clock_gating(
-				adev, state == AMD_CG_STATE_GATE, i);
-		break;
-	default:
-		break;
-	}
+	for (i = 0; i < num_xcc; i++)
+		gfx_v9_4_3_xcc_update_gfx_clock_gating(
+			adev, state == AMD_CG_STATE_GATE, i);
+
 	return 0;
 }
 
-static void gfx_v9_4_3_get_clockgating_state(void *handle, u64 *flags)
+static void gfx_v9_4_3_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
@@ -3519,9 +3492,7 @@ static int gfx_v9_4_3_unmap_done(struct amdgpu_device *adev, uint32_t me,
 
 static bool gfx_v9_4_3_pipe_reset_support(struct amdgpu_device *adev)
 {
-	/*TODO: Need check gfx9.4.4 mec fw whether supports pipe reset as well.*/
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) &&
-			adev->gfx.mec_fw_version >= 0x0000009b)
+	if (!!(adev->gfx.compute_supported_reset & AMDGPU_RESET_TYPE_PER_PIPE))
 		return true;
 	else
 		dev_warn_once(adev->dev, "Please use the latest MEC version to see whether support pipe reset\n");
@@ -3584,20 +3555,21 @@ static int gfx_v9_4_3_reset_hw_pipe(struct amdgpu_ring *ring)
 }
 
 static int gfx_v9_4_3_reset_kcq(struct amdgpu_ring *ring,
-				unsigned int vmid)
+				unsigned int vmid,
+				struct amdgpu_fence *timedout_fence)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_kiq *kiq = &adev->gfx.kiq[ring->xcc_id];
 	struct amdgpu_ring *kiq_ring = &kiq->ring;
+	int reset_mode = AMDGPU_RESET_TYPE_PER_QUEUE;
 	unsigned long flags;
 	int r;
 
-	if (amdgpu_sriov_vf(adev))
-		return -EINVAL;
-
 	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
 		return -EINVAL;
 
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+
 	spin_lock_irqsave(&kiq->ring_lock, flags);
 
 	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
@@ -3623,30 +3595,19 @@ static int gfx_v9_4_3_reset_kcq(struct amdgpu_ring *ring,
 		dev_err(adev->dev, "fail to wait on hqd deactive and will try pipe reset\n");
 
 pipe_reset:
-	if(r) {
+	if (r) {
+		if (!(adev->gfx.compute_supported_reset & AMDGPU_RESET_TYPE_PER_PIPE))
+			return -EOPNOTSUPP;
 		r = gfx_v9_4_3_reset_hw_pipe(ring);
+		reset_mode = AMDGPU_RESET_TYPE_PER_PIPE;
 		dev_info(adev->dev, "ring: %s pipe reset :%s\n", ring->name,
 				r ? "failed" : "successfully");
 		if (r)
 			return r;
 	}
 
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0)){
-		dev_err(adev->dev, "fail to resv mqd_obj\n");
-		return r;
-	}
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
-	if (!r) {
-		r = gfx_v9_4_3_xcc_kcq_init_queue(ring, ring->xcc_id, true);
-		amdgpu_bo_kunmap(ring->mqd_obj);
-		ring->mqd_ptr = NULL;
-	}
-	amdgpu_bo_unreserve(ring->mqd_obj);
-	if (r) {
-		dev_err(adev->dev, "fail to unresv mqd_obj\n");
-		return r;
-	}
+	gfx_v9_4_3_xcc_kcq_init_queue(ring, ring->xcc_id, true);
+
 	spin_lock_irqsave(&kiq->ring_lock, flags);
 	r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size);
 	if (r) {
@@ -3655,14 +3616,24 @@ pipe_reset:
 	}
 	kiq->pmf->kiq_map_queues(kiq_ring, ring);
 	amdgpu_ring_commit(kiq_ring);
-	spin_unlock_irqrestore(&kiq->ring_lock, flags);
-
 	r = amdgpu_ring_test_ring(kiq_ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
 	if (r) {
+		if (reset_mode == AMDGPU_RESET_TYPE_PER_QUEUE)
+			goto pipe_reset;
+
 		dev_err(adev->dev, "fail to remap queue\n");
 		return r;
 	}
-	return amdgpu_ring_test_ring(ring);
+
+	if (reset_mode == AMDGPU_RESET_TYPE_PER_QUEUE) {
+		r = amdgpu_ring_test_ring(ring);
+		if (r)
+			goto pipe_reset;
+	}
+
+
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
 }
 
 enum amdgpu_gfx_cp_ras_mem_id {
@@ -4641,12 +4612,21 @@ static void gfx_v9_4_3_ip_print(struct amdgpu_ip_block *ip_block, struct drm_pri
 						   "\nxcc:%d mec:%d, pipe:%d, queue:%d\n",
 						    xcc_id, i, j, k);
 					for (reg = 0; reg < reg_count; reg++) {
-						drm_printf(p,
-							   "%-50s \t 0x%08x\n",
-							   gc_cp_reg_list_9_4_3[reg].reg_name,
-							   adev->gfx.ip_dump_compute_queues
-								[xcc_offset + inst_offset +
-								reg]);
+						if (i && gc_cp_reg_list_9_4_3[reg].reg_offset ==
+						    regCP_MEC_ME1_HEADER_DUMP)
+							drm_printf(p,
+								   "%-50s \t 0x%08x\n",
+								   "regCP_MEC_ME2_HEADER_DUMP",
+								   adev->gfx.ip_dump_compute_queues
+								   [xcc_offset + inst_offset +
+								    reg]);
+						else
+							drm_printf(p,
+								   "%-50s \t 0x%08x\n",
+								   gc_cp_reg_list_9_4_3[reg].reg_name,
+								   adev->gfx.ip_dump_compute_queues
+								   [xcc_offset + inst_offset +
+								    reg]);
 					}
 					inst_offset += reg_count;
 				}
@@ -4695,12 +4675,20 @@ static void gfx_v9_4_3_ip_dump(struct amdgpu_ip_block *ip_block)
 							  GET_INST(GC, xcc_id));
 
 					for (reg = 0; reg < reg_count; reg++) {
-						adev->gfx.ip_dump_compute_queues
-							[xcc_offset +
-							 inst_offset + reg] =
-							RREG32(SOC15_REG_ENTRY_OFFSET_INST(
-								gc_cp_reg_list_9_4_3[reg],
-								GET_INST(GC, xcc_id)));
+						if (i && gc_cp_reg_list_9_4_3[reg].reg_offset ==
+						    regCP_MEC_ME1_HEADER_DUMP)
+							adev->gfx.ip_dump_compute_queues
+								[xcc_offset +
+								 inst_offset + reg] =
+								RREG32(SOC15_REG_OFFSET(GC, GET_INST(GC, xcc_id),
+											regCP_MEC_ME2_HEADER_DUMP));
+						else
+							adev->gfx.ip_dump_compute_queues
+								[xcc_offset +
+								 inst_offset + reg] =
+								RREG32(SOC15_REG_ENTRY_OFFSET_INST(
+									       gc_cp_reg_list_9_4_3[reg],
+									       GET_INST(GC, xcc_id)));
 					}
 					inst_offset += reg_count;
 				}
@@ -4808,6 +4796,7 @@ static const struct amdgpu_ring_funcs gfx_v9_4_3_ring_funcs_kiq = {
 	.emit_wreg = gfx_v9_4_3_ring_emit_wreg,
 	.emit_reg_wait = gfx_v9_4_3_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = gfx_v9_4_3_ring_emit_reg_write_reg_wait,
+	.emit_hdp_flush = gfx_v9_4_3_ring_emit_hdp_flush,
 };
 
 static void gfx_v9_4_3_set_ring_funcs(struct amdgpu_device *adev)
@@ -4867,34 +4856,13 @@ static void gfx_v9_4_3_set_rlc_funcs(struct amdgpu_device *adev)
 
 static void gfx_v9_4_3_set_gds_init(struct amdgpu_device *adev)
 {
-	/* init asci gds info */
-	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
-	case IP_VERSION(9, 4, 3):
-	case IP_VERSION(9, 4, 4):
-	case IP_VERSION(9, 5, 0):
-		/* 9.4.3 removed all the GDS internal memory,
-		 * only support GWS opcode in kernel, like barrier
-		 * semaphore.etc */
-		adev->gds.gds_size = 0;
-		break;
-	default:
-		adev->gds.gds_size = 0x10000;
-		break;
-	}
-
-	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
-	case IP_VERSION(9, 4, 3):
-	case IP_VERSION(9, 4, 4):
-	case IP_VERSION(9, 5, 0):
-		/* deprecated for 9.4.3, no usage at all */
-		adev->gds.gds_compute_max_wave_id = 0;
-		break;
-	default:
-		/* this really depends on the chip */
-		adev->gds.gds_compute_max_wave_id = 0x7ff;
-		break;
-	}
+	/* 9.4.3 variants removed all the GDS internal memory,
+	 * only support GWS opcode in kernel, like barrier
+	 * semaphore.etc */
 
+	/* init asic gds info */
+	adev->gds.gds_size = 0;
+	adev->gds.gds_compute_max_wave_id = 0;
 	adev->gds.gws_size = 64;
 	adev->gds.oa_size = 16;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
index 0e3ddea7b8e0..a7bfc9f41d0e 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
@@ -92,12 +92,12 @@ static void gfxhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
 {
 	uint64_t value;
 
-	/* Program the AGP BAR */
-	WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BASE, 0);
-	WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
-	WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
-
 	if (!amdgpu_sriov_vf(adev) || adev->asic_type <= CHIP_VEGA10) {
+		/* Program the AGP BAR */
+		WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BASE, 0);
+		WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BOT, adev->gmc.agp_start >> 24);
+		WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_TOP, adev->gmc.agp_end >> 24);
+
 		/* Program the system aperture low logical page number. */
 		WREG32_SOC15_RLC(GC, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
 			min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c
index 5470cef7e9bd..6c03bf9f1ae8 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c
@@ -74,6 +74,8 @@ static void gfxhub_v1_2_setup_vm_pt_regs(struct amdgpu_device *adev,
 static void gfxhub_v1_2_xcc_init_gart_aperture_regs(struct amdgpu_device *adev,
 						    uint32_t xcc_mask)
 {
+	uint64_t gart_start = amdgpu_virt_xgmi_migrate_enabled(adev) ?
+			adev->gmc.vram_start : adev->gmc.fb_start;
 	uint64_t pt_base;
 	int i;
 
@@ -91,10 +93,10 @@ static void gfxhub_v1_2_xcc_init_gart_aperture_regs(struct amdgpu_device *adev,
 		if (adev->gmc.pdb0_bo) {
 			WREG32_SOC15(GC, GET_INST(GC, i),
 				     regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
-				     (u32)(adev->gmc.fb_start >> 12));
+				     (u32)(gart_start >> 12));
 			WREG32_SOC15(GC, GET_INST(GC, i),
 				     regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
-				     (u32)(adev->gmc.fb_start >> 44));
+				     (u32)(gart_start >> 44));
 
 			WREG32_SOC15(GC, GET_INST(GC, i),
 				     regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
@@ -180,7 +182,7 @@ gfxhub_v1_2_xcc_init_system_aperture_regs(struct amdgpu_device *adev,
 		/* In the case squeezing vram into GART aperture, we don't use
 		 * FB aperture and AGP aperture. Disable them.
 		 */
-		if (adev->gmc.pdb0_bo) {
+		if (adev->gmc.pdb0_bo && adev->gmc.xgmi.connected_to_cpu) {
 			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_FB_LOCATION_TOP, 0);
 			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_FB_LOCATION_BASE, 0x00FFFFFF);
 			WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_AGP_TOP, 0);
@@ -313,6 +315,16 @@ gfxhub_v1_2_xcc_disable_identity_aperture(struct amdgpu_device *adev,
 	}
 }
 
+static inline bool
+gfxhub_v1_2_per_process_xnack_support(struct amdgpu_device *adev)
+{
+	/*
+	 * TODO: Check if this function is really needed, so far only 9.4.3
+	 * variants use GFXHUB 1.2
+	 */
+	return !!adev->aid_mask;
+}
+
 static void gfxhub_v1_2_xcc_setup_vmid_config(struct amdgpu_device *adev,
 					      uint32_t xcc_mask)
 {
@@ -355,7 +367,7 @@ static void gfxhub_v1_2_xcc_setup_vmid_config(struct amdgpu_device *adev,
 					    PAGE_TABLE_BLOCK_SIZE,
 					    block_size);
 			/* Send no-retry XNACK on fault to suppress VM fault storm.
-			 * On 9.4.2 and 9.4.3, XNACK can be enabled in
+			 * On 9.4.3 variants, XNACK can be enabled in
 			 * the SQ per-process.
 			 * Retry faults need to be enabled for that to work.
 			 */
@@ -363,14 +375,8 @@ static void gfxhub_v1_2_xcc_setup_vmid_config(struct amdgpu_device *adev,
 				tmp, VM_CONTEXT1_CNTL,
 				RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
 				!adev->gmc.noretry ||
-					amdgpu_ip_version(adev, GC_HWIP, 0) ==
-						IP_VERSION(9, 4, 2) ||
-					amdgpu_ip_version(adev, GC_HWIP, 0) ==
-						IP_VERSION(9, 4, 3) ||
-					amdgpu_ip_version(adev, GC_HWIP, 0) ==
-						IP_VERSION(9, 4, 4) ||
-					amdgpu_ip_version(adev, GC_HWIP, 0) ==
-						IP_VERSION(9, 5, 0));
+					gfxhub_v1_2_per_process_xnack_support(
+						adev));
 			WREG32_SOC15_OFFSET(GC, GET_INST(GC, j), regVM_CONTEXT1_CNTL,
 					    i * hub->ctx_distance, tmp);
 			WREG32_SOC15_OFFSET(GC, GET_INST(GC, j),
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c
index 17509f32f61a..deb95fab02df 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c
@@ -505,42 +505,6 @@ static void gfxhub_v2_1_init(struct amdgpu_device *adev)
 	hub->vmhub_funcs = &gfxhub_v2_1_vmhub_funcs;
 }
 
-static int gfxhub_v2_1_get_xgmi_info(struct amdgpu_device *adev)
-{
-	u32 xgmi_lfb_cntl = RREG32_SOC15(GC, 0, mmGCMC_VM_XGMI_LFB_CNTL);
-	u32 max_region =
-		REG_GET_FIELD(xgmi_lfb_cntl, GCMC_VM_XGMI_LFB_CNTL, PF_MAX_REGION);
-	u32 max_num_physical_nodes   = 0;
-	u32 max_physical_node_id     = 0;
-
-	switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
-	case IP_VERSION(4, 8, 0):
-		max_num_physical_nodes   = 4;
-		max_physical_node_id     = 3;
-		break;
-	default:
-		return -EINVAL;
-	}
-
-	/* PF_MAX_REGION=0 means xgmi is disabled */
-	if (max_region) {
-		adev->gmc.xgmi.num_physical_nodes = max_region + 1;
-		if (adev->gmc.xgmi.num_physical_nodes > max_num_physical_nodes)
-			return -EINVAL;
-
-		adev->gmc.xgmi.physical_node_id =
-			REG_GET_FIELD(xgmi_lfb_cntl, GCMC_VM_XGMI_LFB_CNTL, PF_LFB_REGION);
-		if (adev->gmc.xgmi.physical_node_id > max_physical_node_id)
-			return -EINVAL;
-
-		adev->gmc.xgmi.node_segment_size = REG_GET_FIELD(
-			RREG32_SOC15(GC, 0, mmGCMC_VM_XGMI_LFB_SIZE),
-			GCMC_VM_XGMI_LFB_SIZE, PF_LFB_SIZE) << 24;
-	}
-
-	return 0;
-}
-
 static void gfxhub_v2_1_utcl2_harvest(struct amdgpu_device *adev)
 {
 	int i;
@@ -696,7 +660,6 @@ const struct amdgpu_gfxhub_funcs gfxhub_v2_1_funcs = {
 	.gart_disable = gfxhub_v2_1_gart_disable,
 	.set_fault_enable_default = gfxhub_v2_1_set_fault_enable_default,
 	.init = gfxhub_v2_1_init,
-	.get_xgmi_info = gfxhub_v2_1_get_xgmi_info,
 	.utcl2_harvest = gfxhub_v2_1_utcl2_harvest,
 	.mode2_save_regs = gfxhub_v2_1_save_regs,
 	.mode2_restore_regs = gfxhub_v2_1_restore_regs,
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
index 9bedca9a79c6..ce6e04242c52 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
@@ -103,8 +103,10 @@ static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev,
 	uint32_t vmhub_index = entry->client_id == SOC15_IH_CLIENTID_VMC ?
 			       AMDGPU_MMHUB0(0) : AMDGPU_GFXHUB(0);
 	struct amdgpu_vmhub *hub = &adev->vmhub[vmhub_index];
-	bool retry_fault = !!(entry->src_data[1] & 0x80);
-	bool write_fault = !!(entry->src_data[1] & 0x20);
+	bool retry_fault = !!(entry->src_data[1] &
+			      AMDGPU_GMC9_FAULT_SOURCE_DATA_RETRY);
+	bool write_fault = !!(entry->src_data[1] &
+			      AMDGPU_GMC9_FAULT_SOURCE_DATA_WRITE);
 	struct amdgpu_task_info *task_info;
 	uint32_t status = 0;
 	u64 addr;
@@ -164,10 +166,7 @@ static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev,
 		entry->src_id, entry->ring_id, entry->vmid, entry->pasid);
 	task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
 	if (task_info) {
-		dev_err(adev->dev,
-			" in process %s pid %d thread %s pid %d\n",
-			task_info->process_name, task_info->tgid,
-			task_info->task_name, task_info->pid);
+		amdgpu_vm_print_task_info(adev, task_info);
 		amdgpu_vm_put_task_info(task_info);
 	}
 
@@ -268,7 +267,7 @@ static void gmc_v10_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
 	ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng;
 
 	/* flush hdp cache */
-	adev->hdp.funcs->flush_hdp(adev, NULL);
+	amdgpu_device_flush_hdp(adev, NULL);
 
 	/* This is necessary for SRIOV as well as for GFXOFF to function
 	 * properly under bare metal
@@ -428,10 +427,6 @@ static void gmc_v10_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned int
 	struct amdgpu_device *adev = ring->adev;
 	uint32_t reg;
 
-	/* MES fw manages IH_VMID_x_LUT updating */
-	if (ring->is_mes_queue)
-		return;
-
 	if (ring->vm_hub == AMDGPU_GFXHUB(0))
 		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid;
 	else
@@ -473,24 +468,6 @@ static void gmc_v10_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned int
  * 0 valid
  */
 
-static uint64_t gmc_v10_0_map_mtype(struct amdgpu_device *adev, uint32_t flags)
-{
-	switch (flags) {
-	case AMDGPU_VM_MTYPE_DEFAULT:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_NC);
-	case AMDGPU_VM_MTYPE_NC:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_NC);
-	case AMDGPU_VM_MTYPE_WC:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_WC);
-	case AMDGPU_VM_MTYPE_CC:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_CC);
-	case AMDGPU_VM_MTYPE_UC:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_UC);
-	default:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_NC);
-	}
-}
-
 static void gmc_v10_0_get_vm_pde(struct amdgpu_device *adev, int level,
 				 uint64_t *addr, uint64_t *flags)
 {
@@ -515,21 +492,39 @@ static void gmc_v10_0_get_vm_pde(struct amdgpu_device *adev, int level,
 }
 
 static void gmc_v10_0_get_vm_pte(struct amdgpu_device *adev,
-				 struct amdgpu_bo_va_mapping *mapping,
+				 struct amdgpu_vm *vm,
+				 struct amdgpu_bo *bo,
+				 uint32_t vm_flags,
 				 uint64_t *flags)
 {
-	struct amdgpu_bo *bo = mapping->bo_va->base.bo;
-
-	*flags &= ~AMDGPU_PTE_EXECUTABLE;
-	*flags |= mapping->flags & AMDGPU_PTE_EXECUTABLE;
+	if (vm_flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		*flags |= AMDGPU_PTE_EXECUTABLE;
+	else
+		*flags &= ~AMDGPU_PTE_EXECUTABLE;
 
-	*flags &= ~AMDGPU_PTE_MTYPE_NV10_MASK;
-	*flags |= (mapping->flags & AMDGPU_PTE_MTYPE_NV10_MASK);
+	switch (vm_flags & AMDGPU_VM_MTYPE_MASK) {
+	case AMDGPU_VM_MTYPE_DEFAULT:
+	case AMDGPU_VM_MTYPE_NC:
+	default:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_WC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_WC);
+		break;
+	case AMDGPU_VM_MTYPE_CC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_CC);
+		break;
+	case AMDGPU_VM_MTYPE_UC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_UC);
+		break;
+	}
 
-	*flags &= ~AMDGPU_PTE_NOALLOC;
-	*flags |= (mapping->flags & AMDGPU_PTE_NOALLOC);
+	if (vm_flags & AMDGPU_VM_PAGE_NOALLOC)
+		*flags |= AMDGPU_PTE_NOALLOC;
+	else
+		*flags &= ~AMDGPU_PTE_NOALLOC;
 
-	if (mapping->flags & AMDGPU_PTE_PRT) {
+	if (vm_flags & AMDGPU_VM_PAGE_PRT) {
 		*flags |= AMDGPU_PTE_PRT;
 		*flags |= AMDGPU_PTE_SNOOPED;
 		*flags |= AMDGPU_PTE_LOG;
@@ -570,7 +565,6 @@ static const struct amdgpu_gmc_funcs gmc_v10_0_gmc_funcs = {
 	.flush_gpu_tlb_pasid = gmc_v10_0_flush_gpu_tlb_pasid,
 	.emit_flush_gpu_tlb = gmc_v10_0_emit_flush_gpu_tlb,
 	.emit_pasid_mapping = gmc_v10_0_emit_pasid_mapping,
-	.map_mtype = gmc_v10_0_map_mtype,
 	.get_vm_pde = gmc_v10_0_get_vm_pde,
 	.get_vm_pte = gmc_v10_0_get_vm_pte,
 	.get_vbios_fb_size = gmc_v10_0_get_vbios_fb_size,
@@ -969,10 +963,9 @@ static int gmc_v10_0_gart_enable(struct amdgpu_device *adev)
 	adev->hdp.funcs->init_registers(adev);
 
 	/* Flush HDP after it is initialized */
-	adev->hdp.funcs->flush_hdp(adev, NULL);
+	amdgpu_device_flush_hdp(adev, NULL);
 
-	value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ?
-		false : true;
+	value = amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS;
 
 	if (!adev->in_s0ix)
 		adev->gfxhub.funcs->set_fault_enable_default(adev, value);
@@ -1076,7 +1069,7 @@ static int gmc_v10_0_resume(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-static bool gmc_v10_0_is_idle(void *handle)
+static bool gmc_v10_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* MC is always ready in GMC v10.*/
 	return true;
@@ -1115,9 +1108,9 @@ static int gmc_v10_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 		return athub_v2_0_set_clockgating(adev, state);
 }
 
-static void gmc_v10_0_get_clockgating_state(void *handle, u64 *flags)
+static void gmc_v10_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 3) ||
 	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 4))
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
index 72751ab4c766..ba59ee8e398a 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
@@ -103,12 +103,41 @@ static int gmc_v11_0_process_interrupt(struct amdgpu_device *adev,
 	uint32_t vmhub_index = entry->client_id == SOC21_IH_CLIENTID_VMC ?
 			       AMDGPU_MMHUB0(0) : AMDGPU_GFXHUB(0);
 	struct amdgpu_vmhub *hub = &adev->vmhub[vmhub_index];
+	bool retry_fault = !!(entry->src_data[1] &
+			      AMDGPU_GMC9_FAULT_SOURCE_DATA_RETRY);
+	bool write_fault = !!(entry->src_data[1] &
+			      AMDGPU_GMC9_FAULT_SOURCE_DATA_WRITE);
 	uint32_t status = 0;
 	u64 addr;
 
 	addr = (u64)entry->src_data[0] << 12;
 	addr |= ((u64)entry->src_data[1] & 0xf) << 44;
 
+	if (retry_fault) {
+		/* Returning 1 here also prevents sending the IV to the KFD */
+
+		/* Process it only if it's the first fault for this address */
+		if (entry->ih != &adev->irq.ih_soft &&
+		    amdgpu_gmc_filter_faults(adev, entry->ih, addr, entry->pasid,
+					     entry->timestamp))
+			return 1;
+
+		/* Delegate it to a different ring if the hardware hasn't
+		 * already done it.
+		 */
+		if (entry->ih == &adev->irq.ih) {
+			amdgpu_irq_delegate(adev, entry, 8);
+			return 1;
+		}
+
+		/* Try to handle the recoverable page faults by filling page
+		 * tables
+		 */
+		if (amdgpu_vm_handle_fault(adev, entry->pasid, 0, 0, addr,
+					   entry->timestamp, write_fault))
+			return 1;
+	}
+
 	if (!amdgpu_sriov_vf(adev)) {
 		/*
 		 * Issue a dummy read to wait for the status register to
@@ -134,10 +163,7 @@ static int gmc_v11_0_process_interrupt(struct amdgpu_device *adev,
 			entry->src_id, entry->ring_id, entry->vmid, entry->pasid);
 		task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
 		if (task_info) {
-			dev_err(adev->dev,
-				" in process %s pid %d thread %s pid %d)\n",
-				task_info->process_name, task_info->tgid,
-				task_info->task_name, task_info->pid);
+			amdgpu_vm_print_task_info(adev, task_info);
 			amdgpu_vm_put_task_info(task_info);
 		}
 
@@ -229,7 +255,7 @@ static void gmc_v11_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
 	ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng;
 
 	/* flush hdp cache */
-	adev->hdp.funcs->flush_hdp(adev, NULL);
+	amdgpu_device_flush_hdp(adev, NULL);
 
 	/* This is necessary for SRIOV as well as for GFXOFF to function
 	 * properly under bare metal
@@ -393,10 +419,6 @@ static void gmc_v11_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned int
 	struct amdgpu_device *adev = ring->adev;
 	uint32_t reg;
 
-	/* MES fw manages IH_VMID_x_LUT updating */
-	if (ring->is_mes_queue)
-		return;
-
 	if (ring->vm_hub == AMDGPU_GFXHUB(0))
 		reg = SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT) + vmid;
 	else
@@ -437,24 +459,6 @@ static void gmc_v11_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned int
  * 0 valid
  */
 
-static uint64_t gmc_v11_0_map_mtype(struct amdgpu_device *adev, uint32_t flags)
-{
-	switch (flags) {
-	case AMDGPU_VM_MTYPE_DEFAULT:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_NC);
-	case AMDGPU_VM_MTYPE_NC:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_NC);
-	case AMDGPU_VM_MTYPE_WC:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_WC);
-	case AMDGPU_VM_MTYPE_CC:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_CC);
-	case AMDGPU_VM_MTYPE_UC:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_UC);
-	default:
-		return AMDGPU_PTE_MTYPE_NV10(0ULL, MTYPE_NC);
-	}
-}
-
 static void gmc_v11_0_get_vm_pde(struct amdgpu_device *adev, int level,
 				 uint64_t *addr, uint64_t *flags)
 {
@@ -479,21 +483,39 @@ static void gmc_v11_0_get_vm_pde(struct amdgpu_device *adev, int level,
 }
 
 static void gmc_v11_0_get_vm_pte(struct amdgpu_device *adev,
-				 struct amdgpu_bo_va_mapping *mapping,
+				 struct amdgpu_vm *vm,
+				 struct amdgpu_bo *bo,
+				 uint32_t vm_flags,
 				 uint64_t *flags)
 {
-	struct amdgpu_bo *bo = mapping->bo_va->base.bo;
-
-	*flags &= ~AMDGPU_PTE_EXECUTABLE;
-	*flags |= mapping->flags & AMDGPU_PTE_EXECUTABLE;
+	if (vm_flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		*flags |= AMDGPU_PTE_EXECUTABLE;
+	else
+		*flags &= ~AMDGPU_PTE_EXECUTABLE;
 
-	*flags &= ~AMDGPU_PTE_MTYPE_NV10_MASK;
-	*flags |= (mapping->flags & AMDGPU_PTE_MTYPE_NV10_MASK);
+	switch (vm_flags & AMDGPU_VM_MTYPE_MASK) {
+	case AMDGPU_VM_MTYPE_DEFAULT:
+	case AMDGPU_VM_MTYPE_NC:
+	default:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_WC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_WC);
+		break;
+	case AMDGPU_VM_MTYPE_CC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_CC);
+		break;
+	case AMDGPU_VM_MTYPE_UC:
+		*flags = AMDGPU_PTE_MTYPE_NV10(*flags, MTYPE_UC);
+		break;
+	}
 
-	*flags &= ~AMDGPU_PTE_NOALLOC;
-	*flags |= (mapping->flags & AMDGPU_PTE_NOALLOC);
+	if (vm_flags & AMDGPU_VM_PAGE_NOALLOC)
+		*flags |= AMDGPU_PTE_NOALLOC;
+	else
+		*flags &= ~AMDGPU_PTE_NOALLOC;
 
-	if (mapping->flags & AMDGPU_PTE_PRT) {
+	if (vm_flags & AMDGPU_VM_PAGE_PRT) {
 		*flags |= AMDGPU_PTE_PRT;
 		*flags |= AMDGPU_PTE_SNOOPED;
 		*flags |= AMDGPU_PTE_LOG;
@@ -534,7 +556,6 @@ static const struct amdgpu_gmc_funcs gmc_v11_0_gmc_funcs = {
 	.flush_gpu_tlb_pasid = gmc_v11_0_flush_gpu_tlb_pasid,
 	.emit_flush_gpu_tlb = gmc_v11_0_emit_flush_gpu_tlb,
 	.emit_pasid_mapping = gmc_v11_0_emit_pasid_mapping,
-	.map_mtype = gmc_v11_0_map_mtype,
 	.get_vm_pde = gmc_v11_0_get_vm_pde,
 	.get_vm_pte = gmc_v11_0_get_vm_pte,
 	.get_vbios_fb_size = gmc_v11_0_get_vbios_fb_size,
@@ -579,6 +600,7 @@ static void gmc_v11_0_set_mmhub_funcs(struct amdgpu_device *adev)
 		break;
 	case IP_VERSION(3, 3, 0):
 	case IP_VERSION(3, 3, 1):
+	case IP_VERSION(3, 3, 2):
 		adev->mmhub.funcs = &mmhub_v3_3_funcs;
 		break;
 	default:
@@ -596,6 +618,7 @@ static void gmc_v11_0_set_gfxhub_funcs(struct amdgpu_device *adev)
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		adev->gfxhub.funcs = &gfxhub_v11_5_0_funcs;
 		break;
 	default:
@@ -750,6 +773,18 @@ static int gmc_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 	adev->gmc.vram_type = vram_type;
 	adev->gmc.vram_vendor = vram_vendor;
 
+	/* The mall_size is already calculated as mall_size_per_umc * num_umc.
+	 * However, for gfx1151, which features a 2-to-1 UMC mapping,
+	 * the result must be multiplied by 2 to determine the actual mall size.
+	 */
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 5, 1):
+		adev->gmc.mall_size *= 2;
+		break;
+	default:
+		break;
+	}
+
 	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
 	case IP_VERSION(11, 0, 0):
 	case IP_VERSION(11, 0, 1):
@@ -759,6 +794,7 @@ static int gmc_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 	case IP_VERSION(11, 5, 2):
+	case IP_VERSION(11, 5, 3):
 		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
 		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
 		/*
@@ -829,7 +865,7 @@ static int gmc_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 	 * amdgpu graphics/compute will use VMIDs 1-7
 	 * amdkfd will use VMIDs 8-15
 	 */
-	adev->vm_manager.first_kfd_vmid = 8;
+	adev->vm_manager.first_kfd_vmid = adev->gfx.disable_kq ? 1 : 8;
 
 	amdgpu_vm_manager_init(adev);
 
@@ -896,10 +932,9 @@ static int gmc_v11_0_gart_enable(struct amdgpu_device *adev)
 		return r;
 
 	/* Flush HDP after it is initialized */
-	adev->hdp.funcs->flush_hdp(adev, NULL);
+	amdgpu_device_flush_hdp(adev, NULL);
 
-	value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ?
-		false : true;
+	value = amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS;
 
 	adev->mmhub.funcs->set_fault_enable_default(adev, value);
 	gmc_v11_0_flush_gpu_tlb(adev, 0, AMDGPU_MMHUB0(0), 0);
@@ -984,7 +1019,7 @@ static int gmc_v11_0_resume(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-static bool gmc_v11_0_is_idle(void *handle)
+static bool gmc_v11_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* MC is always ready in GMC v11.*/
 	return true;
@@ -1009,9 +1044,9 @@ static int gmc_v11_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	return athub_v3_0_set_clockgating(adev, state);
 }
 
-static void gmc_v11_0_get_clockgating_state(void *handle, u64 *flags)
+static void gmc_v11_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->mmhub.funcs->get_clockgating(adev, flags);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c
index b749f1c3f6a9..7a9d6894e321 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c
@@ -91,6 +91,10 @@ static int gmc_v12_0_process_interrupt(struct amdgpu_device *adev,
 				       struct amdgpu_iv_entry *entry)
 {
 	struct amdgpu_vmhub *hub;
+	bool retry_fault = !!(entry->src_data[1] &
+			      AMDGPU_GMC9_FAULT_SOURCE_DATA_RETRY);
+	bool write_fault = !!(entry->src_data[1] &
+			      AMDGPU_GMC9_FAULT_SOURCE_DATA_WRITE);
 	uint32_t status = 0;
 	u64 addr;
 
@@ -102,6 +106,31 @@ static int gmc_v12_0_process_interrupt(struct amdgpu_device *adev,
 	else
 		hub = &adev->vmhub[AMDGPU_GFXHUB(0)];
 
+	if (retry_fault) {
+		/* Returning 1 here also prevents sending the IV to the KFD */
+
+		/* Process it only if it's the first fault for this address */
+		if (entry->ih != &adev->irq.ih_soft &&
+		    amdgpu_gmc_filter_faults(adev, entry->ih, addr, entry->pasid,
+					     entry->timestamp))
+			return 1;
+
+		/* Delegate it to a different ring if the hardware hasn't
+		 * already done it.
+		 */
+		if (entry->ih == &adev->irq.ih) {
+			amdgpu_irq_delegate(adev, entry, 8);
+			return 1;
+		}
+
+		/* Try to handle the recoverable page faults by filling page
+		 * tables
+		 */
+		if (amdgpu_vm_handle_fault(adev, entry->pasid, 0, 0, addr,
+					   entry->timestamp, write_fault))
+			return 1;
+	}
+
 	if (!amdgpu_sriov_vf(adev)) {
 		/*
 		 * Issue a dummy read to wait for the status register to
@@ -127,10 +156,7 @@ static int gmc_v12_0_process_interrupt(struct amdgpu_device *adev,
 			entry->src_id, entry->ring_id, entry->vmid, entry->pasid);
 		task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
 		if (task_info) {
-			dev_err(adev->dev,
-				" in process %s pid %d thread %s pid %d)\n",
-				task_info->process_name, task_info->tgid,
-				task_info->task_name, task_info->pid);
+			amdgpu_vm_print_task_info(adev, task_info);
 			amdgpu_vm_put_task_info(task_info);
 		}
 
@@ -297,7 +323,7 @@ static void gmc_v12_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
 		return;
 
 	/* flush hdp cache */
-	adev->hdp.funcs->flush_hdp(adev, NULL);
+	amdgpu_device_flush_hdp(adev, NULL);
 
 	/* This is necessary for SRIOV as well as for GFXOFF to function
 	 * properly under bare metal
@@ -315,9 +341,7 @@ static void gmc_v12_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
 		return;
 	}
 
-	mutex_lock(&adev->mman.gtt_window_lock);
 	gmc_v12_0_flush_vm_hub(adev, vmid, vmhub, 0);
-	mutex_unlock(&adev->mman.gtt_window_lock);
 	return;
 }
 
@@ -339,6 +363,22 @@ static void gmc_v12_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
 	uint16_t queried;
 	int vmid, i;
 
+	if (adev->enable_uni_mes && adev->mes.ring[AMDGPU_MES_SCHED_PIPE].sched.ready &&
+	    (adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x84) {
+		struct mes_inv_tlbs_pasid_input input = {0};
+		input.pasid = pasid;
+		input.flush_type = flush_type;
+		input.hub_id = AMDGPU_GFXHUB(0);
+		/* MES will invalidate all gc_hub for the device from master */
+		adev->mes.funcs->invalidate_tlbs_pasid(&adev->mes, &input);
+		if (all_hub) {
+			/* Only need to invalidate mm_hub now, gfx12 only support one mmhub */
+			input.hub_id = AMDGPU_MMHUB0(0);
+			adev->mes.funcs->invalidate_tlbs_pasid(&adev->mes, &input);
+		}
+		return;
+	}
+
 	for (vmid = 1; vmid < 16; vmid++) {
 		bool valid;
 
@@ -413,10 +453,6 @@ static void gmc_v12_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid
 	struct amdgpu_device *adev = ring->adev;
 	uint32_t reg;
 
-	/* MES fw manages IH_VMID_x_LUT updating */
-	if (ring->is_mes_queue)
-		return;
-
 	if (ring->vm_hub == AMDGPU_GFXHUB(0))
 		reg = SOC15_REG_OFFSET(OSSSYS, 0, regIH_VMID_0_LUT) + vmid;
 	else
@@ -460,20 +496,6 @@ static void gmc_v12_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid
  * 0 valid
  */
 
-static uint64_t gmc_v12_0_map_mtype(struct amdgpu_device *adev, uint32_t flags)
-{
-	switch (flags) {
-	case AMDGPU_VM_MTYPE_DEFAULT:
-		return AMDGPU_PTE_MTYPE_GFX12(0ULL, MTYPE_NC);
-	case AMDGPU_VM_MTYPE_NC:
-		return AMDGPU_PTE_MTYPE_GFX12(0ULL, MTYPE_NC);
-	case AMDGPU_VM_MTYPE_UC:
-		return AMDGPU_PTE_MTYPE_GFX12(0ULL, MTYPE_UC);
-	default:
-		return AMDGPU_PTE_MTYPE_GFX12(0ULL, MTYPE_NC);
-	}
-}
-
 static void gmc_v12_0_get_vm_pde(struct amdgpu_device *adev, int level,
 				 uint64_t *addr, uint64_t *flags)
 {
@@ -497,21 +519,35 @@ static void gmc_v12_0_get_vm_pde(struct amdgpu_device *adev, int level,
 }
 
 static void gmc_v12_0_get_vm_pte(struct amdgpu_device *adev,
-				 struct amdgpu_bo_va_mapping *mapping,
+				 struct amdgpu_vm *vm,
+				 struct amdgpu_bo *bo,
+				 uint32_t vm_flags,
 				 uint64_t *flags)
 {
-	struct amdgpu_bo *bo = mapping->bo_va->base.bo;
-	struct amdgpu_device *bo_adev;
-	bool coherent, is_system;
-
+	if (vm_flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		*flags |= AMDGPU_PTE_EXECUTABLE;
+	else
+		*flags &= ~AMDGPU_PTE_EXECUTABLE;
 
-	*flags &= ~AMDGPU_PTE_EXECUTABLE;
-	*flags |= mapping->flags & AMDGPU_PTE_EXECUTABLE;
+	switch (vm_flags & AMDGPU_VM_MTYPE_MASK) {
+	case AMDGPU_VM_MTYPE_DEFAULT:
+		*flags = AMDGPU_PTE_MTYPE_GFX12(*flags, MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_NC:
+	default:
+		*flags = AMDGPU_PTE_MTYPE_GFX12(*flags, MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_UC:
+		*flags = AMDGPU_PTE_MTYPE_GFX12(*flags, MTYPE_UC);
+		break;
+	}
 
-	*flags &= ~AMDGPU_PTE_MTYPE_GFX12_MASK;
-	*flags |= (mapping->flags & AMDGPU_PTE_MTYPE_GFX12_MASK);
+	if (vm_flags & AMDGPU_VM_PAGE_NOALLOC)
+		*flags |= AMDGPU_PTE_NOALLOC;
+	else
+		*flags &= ~AMDGPU_PTE_NOALLOC;
 
-	if (mapping->flags & AMDGPU_PTE_PRT_GFX12) {
+	if (vm_flags & AMDGPU_VM_PAGE_PRT) {
 		*flags |= AMDGPU_PTE_PRT_GFX12;
 		*flags |= AMDGPU_PTE_SNOOPED;
 		*flags |= AMDGPU_PTE_SYSTEM;
@@ -519,25 +555,11 @@ static void gmc_v12_0_get_vm_pte(struct amdgpu_device *adev,
 		*flags &= ~AMDGPU_PTE_VALID;
 	}
 
-	if (!bo)
-		return;
-
-	if (bo->flags & (AMDGPU_GEM_CREATE_COHERENT |
-			       AMDGPU_GEM_CREATE_UNCACHED))
-		*flags = AMDGPU_PTE_MTYPE_GFX12(*flags, MTYPE_UC);
-
-	bo_adev = amdgpu_ttm_adev(bo->tbo.bdev);
-	coherent = bo->flags & AMDGPU_GEM_CREATE_COHERENT;
-	is_system = (bo->tbo.resource->mem_type == TTM_PL_TT) ||
-		(bo->tbo.resource->mem_type == AMDGPU_PL_PREEMPT);
-
 	if (bo && bo->flags & AMDGPU_GEM_CREATE_GFX12_DCC)
 		*flags |= AMDGPU_PTE_DCC;
 
-	/* WA for HW bug */
-	if (is_system || ((bo_adev != adev) && coherent))
-		*flags = AMDGPU_PTE_MTYPE_GFX12(*flags, MTYPE_NC);
-
+	if (bo && bo->flags & AMDGPU_GEM_CREATE_UNCACHED)
+		*flags = AMDGPU_PTE_MTYPE_GFX12(*flags, MTYPE_UC);
 }
 
 static unsigned gmc_v12_0_get_vbios_fb_size(struct amdgpu_device *adev)
@@ -567,7 +589,6 @@ static const struct amdgpu_gmc_funcs gmc_v12_0_gmc_funcs = {
 	.flush_gpu_tlb_pasid = gmc_v12_0_flush_gpu_tlb_pasid,
 	.emit_flush_gpu_tlb = gmc_v12_0_emit_flush_gpu_tlb,
 	.emit_pasid_mapping = gmc_v12_0_emit_pasid_mapping,
-	.map_mtype = gmc_v12_0_map_mtype,
 	.get_vm_pde = gmc_v12_0_get_vm_pde,
 	.get_vm_pte = gmc_v12_0_get_vm_pte,
 	.get_vbios_fb_size = gmc_v12_0_get_vbios_fb_size,
@@ -837,7 +858,7 @@ static int gmc_v12_0_sw_init(struct amdgpu_ip_block *ip_block)
 	 * amdgpu graphics/compute will use VMIDs 1-7
 	 * amdkfd will use VMIDs 8-15
 	 */
-	adev->vm_manager.first_kfd_vmid = 8;
+	adev->vm_manager.first_kfd_vmid = adev->gfx.disable_kq ? 1 : 8;
 
 	amdgpu_vm_manager_init(adev);
 
@@ -898,10 +919,9 @@ static int gmc_v12_0_gart_enable(struct amdgpu_device *adev)
 		return r;
 
 	/* Flush HDP after it is initialized */
-	adev->hdp.funcs->flush_hdp(adev, NULL);
+	amdgpu_device_flush_hdp(adev, NULL);
 
-	value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ?
-		false : true;
+	value = amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS;
 
 	adev->mmhub.funcs->set_fault_enable_default(adev, value);
 	gmc_v12_0_flush_gpu_tlb(adev, 0, AMDGPU_MMHUB0(0), 0);
@@ -984,7 +1004,7 @@ static int gmc_v12_0_resume(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-static bool gmc_v12_0_is_idle(void *handle)
+static bool gmc_v12_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* MC is always ready in GMC v11.*/
 	return true;
@@ -1009,9 +1029,9 @@ static int gmc_v12_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	return athub_v4_1_0_set_clockgating(adev, state);
 }
 
-static void gmc_v12_0_get_clockgating_state(void *handle, u64 *flags)
+static void gmc_v12_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->mmhub.funcs->get_clockgating(adev, flags);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
index 2245dda92021..a8ec95f42926 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
@@ -213,7 +213,7 @@ static void gmc_v6_0_vram_gtt_location(struct amdgpu_device *adev,
 
 	amdgpu_gmc_set_agp_default(adev, mc);
 	amdgpu_gmc_vram_location(adev, mc, base);
-	amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_BEST_FIT);
+	amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_LOW);
 }
 
 static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
@@ -249,7 +249,7 @@ static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
 
 		/* disable VGA render */
 		tmp = RREG32(mmVGA_RENDER_CONTROL);
-		tmp &= ~VGA_VSTATUS_CNTL;
+		tmp &= VGA_RENDER_CONTROL__VGA_VSTATUS_CNTL_MASK;
 		WREG32(mmVGA_RENDER_CONTROL, tmp);
 	}
 	/* Update configuration */
@@ -382,7 +382,9 @@ static void gmc_v6_0_get_vm_pde(struct amdgpu_device *adev, int level,
 }
 
 static void gmc_v6_0_get_vm_pte(struct amdgpu_device *adev,
-				struct amdgpu_bo_va_mapping *mapping,
+				struct amdgpu_vm *vm,
+				struct amdgpu_bo *bo,
+				uint32_t vm_flags,
 				uint64_t *flags)
 {
 	*flags &= ~AMDGPU_PTE_EXECUTABLE;
@@ -608,36 +610,33 @@ static void gmc_v6_0_gart_disable(struct amdgpu_device *adev)
 }
 
 static void gmc_v6_0_vm_decode_fault(struct amdgpu_device *adev,
-				     u32 status, u32 addr, u32 mc_client)
+				     u32 status, u32 addr)
 {
 	u32 mc_id;
 	u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
 	u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 					PROTECTIONS);
-	char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff,
-		(mc_client >> 8) & 0xff, mc_client & 0xff, 0 };
 
 	mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 			      MEMORY_CLIENT_ID);
 
-	dev_err(adev->dev, "VM fault (0x%02x, vmid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
+	dev_err(adev->dev, "VM fault (0x%02x, vmid %d) at page %u, %s from %d\n",
 	       protections, vmid, addr,
 	       REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 			     MEMORY_CLIENT_RW) ?
-	       "write" : "read", block, mc_client, mc_id);
+	       "write" : "read", mc_id);
 }
 
-/*
 static const u32 mc_cg_registers[] = {
-	MC_HUB_MISC_HUB_CG,
-	MC_HUB_MISC_SIP_CG,
-	MC_HUB_MISC_VM_CG,
-	MC_XPB_CLK_GAT,
-	ATC_MISC_CG,
-	MC_CITF_MISC_WR_CG,
-	MC_CITF_MISC_RD_CG,
-	MC_CITF_MISC_VM_CG,
-	VM_L2_CG,
+	mmMC_HUB_MISC_HUB_CG,
+	mmMC_HUB_MISC_SIP_CG,
+	mmMC_HUB_MISC_VM_CG,
+	mmMC_XPB_CLK_GAT,
+	mmATC_MISC_CG,
+	mmMC_CITF_MISC_WR_CG,
+	mmMC_CITF_MISC_RD_CG,
+	mmMC_CITF_MISC_VM_CG,
+	mmVM_L2_CG,
 };
 
 static const u32 mc_cg_ls_en[] = {
@@ -672,7 +671,7 @@ static void gmc_v6_0_enable_mc_ls(struct amdgpu_device *adev,
 
 	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
 		orig = data = RREG32(mc_cg_registers[i]);
-		if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_MC_LS))
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS))
 			data |= mc_cg_ls_en[i];
 		else
 			data &= ~mc_cg_ls_en[i];
@@ -689,7 +688,7 @@ static void gmc_v6_0_enable_mc_mgcg(struct amdgpu_device *adev,
 
 	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
 		orig = data = RREG32(mc_cg_registers[i]);
-		if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_MC_MGCG))
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
 			data |= mc_cg_en[i];
 		else
 			data &= ~mc_cg_en[i];
@@ -705,7 +704,7 @@ static void gmc_v6_0_enable_bif_mgls(struct amdgpu_device *adev,
 
 	orig = data = RREG32_PCIE(ixPCIE_CNTL2);
 
-	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_BIF_LS)) {
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS)) {
 		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_LS_EN, 1);
 		data = REG_SET_FIELD(data, PCIE_CNTL2, MST_MEM_LS_EN, 1);
 		data = REG_SET_FIELD(data, PCIE_CNTL2, REPLAY_MEM_LS_EN, 1);
@@ -728,7 +727,7 @@ static void gmc_v6_0_enable_hdp_mgcg(struct amdgpu_device *adev,
 
 	orig = data = RREG32(mmHDP_HOST_PATH_CNTL);
 
-	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_HDP_MGCG))
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_MGCG))
 		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 0);
 	else
 		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 1);
@@ -744,7 +743,7 @@ static void gmc_v6_0_enable_hdp_ls(struct amdgpu_device *adev,
 
 	orig = data = RREG32(mmHDP_MEM_POWER_LS);
 
-	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_HDP_LS))
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
 		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 1);
 	else
 		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 0);
@@ -752,7 +751,6 @@ static void gmc_v6_0_enable_hdp_ls(struct amdgpu_device *adev,
 	if (orig != data)
 		WREG32(mmHDP_MEM_POWER_LS, data);
 }
-*/
 
 static int gmc_v6_0_convert_vram_type(int mc_seq_vram_type)
 {
@@ -957,9 +955,9 @@ static int gmc_v6_0_resume(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-static bool gmc_v6_0_is_idle(void *handle)
+static bool gmc_v6_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
@@ -976,7 +974,7 @@ static int gmc_v6_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (gmc_v6_0_is_idle(adev))
+		if (gmc_v6_0_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
@@ -1072,6 +1070,12 @@ static int gmc_v6_0_process_interrupt(struct amdgpu_device *adev,
 {
 	u32 addr, status;
 
+	/* Delegate to the soft IRQ handler ring */
+	if (adev->irq.ih_soft.enabled && entry->ih != &adev->irq.ih_soft) {
+		amdgpu_irq_delegate(adev, entry, 4);
+		return 1;
+	}
+
 	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
 	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
 	WREG32_P(mmVM_CONTEXT1_CNTL2, 1, ~1);
@@ -1079,6 +1083,10 @@ static int gmc_v6_0_process_interrupt(struct amdgpu_device *adev,
 	if (!addr && !status)
 		return 0;
 
+	amdgpu_vm_update_fault_cache(adev, entry->pasid,
+				     ((u64)addr) << AMDGPU_GPU_PAGE_SHIFT,
+				     status, AMDGPU_GFXHUB(0));
+
 	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_FIRST)
 		gmc_v6_0_set_fault_enable_default(adev, false);
 
@@ -1089,7 +1097,7 @@ static int gmc_v6_0_process_interrupt(struct amdgpu_device *adev,
 			addr);
 		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
 			status);
-		gmc_v6_0_vm_decode_fault(adev, status, addr, 0);
+		gmc_v6_0_vm_decode_fault(adev, status, addr);
 	}
 
 	return 0;
@@ -1098,6 +1106,20 @@ static int gmc_v6_0_process_interrupt(struct amdgpu_device *adev,
 static int gmc_v6_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
+	struct amdgpu_device *adev = ip_block->adev;
+	bool gate = false;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		gmc_v6_0_enable_mc_mgcg(adev, gate);
+		gmc_v6_0_enable_mc_ls(adev, gate);
+	}
+	gmc_v6_0_enable_bif_mgls(adev, gate);
+	gmc_v6_0_enable_hdp_mgcg(adev, gate);
+	gmc_v6_0_enable_hdp_ls(adev, gate);
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
index 9aac4b1101e3..fbd0bf147f50 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
@@ -504,7 +504,9 @@ static void gmc_v7_0_get_vm_pde(struct amdgpu_device *adev, int level,
 }
 
 static void gmc_v7_0_get_vm_pte(struct amdgpu_device *adev,
-				struct amdgpu_bo_va_mapping *mapping,
+				struct amdgpu_vm *vm,
+				struct amdgpu_bo *bo,
+				uint32_t vm_flags,
 				uint64_t *flags)
 {
 	*flags &= ~AMDGPU_PTE_EXECUTABLE;
@@ -1066,7 +1068,7 @@ static int gmc_v7_0_sw_init(struct amdgpu_ip_block *ip_block)
 					GFP_KERNEL);
 	if (!adev->gmc.vm_fault_info)
 		return -ENOMEM;
-	atomic_set(&adev->gmc.vm_fault_info_updated, 0);
+	atomic_set_release(&adev->gmc.vm_fault_info_updated, 0);
 
 	return 0;
 }
@@ -1142,9 +1144,9 @@ static int gmc_v7_0_resume(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-static bool gmc_v7_0_is_idle(void *handle)
+static bool gmc_v7_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
@@ -1157,17 +1159,10 @@ static bool gmc_v7_0_is_idle(void *handle)
 static int gmc_v7_0_wait_for_idle(struct amdgpu_ip_block *ip_block)
 {
 	unsigned int i;
-	u32 tmp;
 	struct amdgpu_device *adev = ip_block->adev;
 
 	for (i = 0; i < adev->usec_timeout; i++) {
-		/* read MC_STATUS */
-		tmp = RREG32(mmSRBM_STATUS) & (SRBM_STATUS__MCB_BUSY_MASK |
-					       SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
-					       SRBM_STATUS__MCC_BUSY_MASK |
-					       SRBM_STATUS__MCD_BUSY_MASK |
-					       SRBM_STATUS__VMC_BUSY_MASK);
-		if (!tmp)
+		if (gmc_v7_0_is_idle(ip_block))
 			return 0;
 		udelay(1);
 	}
@@ -1266,6 +1261,12 @@ static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev,
 {
 	u32 addr, status, mc_client, vmid;
 
+	/* Delegate to the soft IRQ handler ring */
+	if (adev->irq.ih_soft.enabled && entry->ih != &adev->irq.ih_soft) {
+		amdgpu_irq_delegate(adev, entry, 4);
+		return 1;
+	}
+
 	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
 	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
 	mc_client = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT);
@@ -1295,7 +1296,7 @@ static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev,
 	vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 			     VMID);
 	if (amdgpu_amdkfd_is_kfd_vmid(adev, vmid)
-		&& !atomic_read(&adev->gmc.vm_fault_info_updated)) {
+		&& !atomic_read_acquire(&adev->gmc.vm_fault_info_updated)) {
 		struct kfd_vm_fault_info *info = adev->gmc.vm_fault_info;
 		u32 protections = REG_GET_FIELD(status,
 					VM_CONTEXT1_PROTECTION_FAULT_STATUS,
@@ -1311,8 +1312,7 @@ static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev,
 		info->prot_read = protections & 0x8 ? true : false;
 		info->prot_write = protections & 0x10 ? true : false;
 		info->prot_exec = protections & 0x20 ? true : false;
-		mb();
-		atomic_set(&adev->gmc.vm_fault_info_updated, 1);
+		atomic_set_release(&adev->gmc.vm_fault_info_updated, 1);
 	}
 
 	return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
index d06585207c33..6551b60f2584 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
@@ -716,11 +716,15 @@ static void gmc_v8_0_get_vm_pde(struct amdgpu_device *adev, int level,
 }
 
 static void gmc_v8_0_get_vm_pte(struct amdgpu_device *adev,
-				struct amdgpu_bo_va_mapping *mapping,
+				struct amdgpu_vm *vm,
+				struct amdgpu_bo *bo,
+				uint32_t vm_flags,
 				uint64_t *flags)
 {
-	*flags &= ~AMDGPU_PTE_EXECUTABLE;
-	*flags |= mapping->flags & AMDGPU_PTE_EXECUTABLE;
+	if (vm_flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		*flags |= AMDGPU_PTE_EXECUTABLE;
+	else
+		*flags &= ~AMDGPU_PTE_EXECUTABLE;
 	*flags &= ~AMDGPU_PTE_PRT;
 }
 
@@ -1179,7 +1183,7 @@ static int gmc_v8_0_sw_init(struct amdgpu_ip_block *ip_block)
 					GFP_KERNEL);
 	if (!adev->gmc.vm_fault_info)
 		return -ENOMEM;
-	atomic_set(&adev->gmc.vm_fault_info_updated, 0);
+	atomic_set_release(&adev->gmc.vm_fault_info_updated, 0);
 
 	return 0;
 }
@@ -1263,9 +1267,9 @@ static int gmc_v8_0_resume(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-static bool gmc_v8_0_is_idle(void *handle)
+static bool gmc_v8_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
@@ -1435,6 +1439,12 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
 		return 0;
 	}
 
+	/* Delegate to the soft IRQ handler ring */
+	if (adev->irq.ih_soft.enabled && entry->ih != &adev->irq.ih_soft) {
+		amdgpu_irq_delegate(adev, entry, 4);
+		return 1;
+	}
+
 	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
 	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
 	mc_client = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT);
@@ -1458,9 +1468,7 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
 
 		task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
 		if (task_info) {
-			dev_err(adev->dev, " for process %s pid %d thread %s pid %d\n",
-				task_info->process_name, task_info->tgid,
-				task_info->task_name, task_info->pid);
+			amdgpu_vm_print_task_info(adev, task_info);
 			amdgpu_vm_put_task_info(task_info);
 		}
 
@@ -1476,7 +1484,7 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
 	vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
 			     VMID);
 	if (amdgpu_amdkfd_is_kfd_vmid(adev, vmid)
-		&& !atomic_read(&adev->gmc.vm_fault_info_updated)) {
+		&& !atomic_read_acquire(&adev->gmc.vm_fault_info_updated)) {
 		struct kfd_vm_fault_info *info = adev->gmc.vm_fault_info;
 		u32 protections = REG_GET_FIELD(status,
 					VM_CONTEXT1_PROTECTION_FAULT_STATUS,
@@ -1492,8 +1500,7 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
 		info->prot_read = protections & 0x8 ? true : false;
 		info->prot_write = protections & 0x10 ? true : false;
 		info->prot_exec = protections & 0x20 ? true : false;
-		mb();
-		atomic_set(&adev->gmc.vm_fault_info_updated, 1);
+		atomic_set_release(&adev->gmc.vm_fault_info_updated, 1);
 	}
 
 	return 0;
@@ -1686,9 +1693,9 @@ static int gmc_v8_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
-static void gmc_v8_0_get_clockgating_state(void *handle, u64 *flags)
+static void gmc_v8_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int data;
 
 	if (amdgpu_sriov_vf(adev))
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 291549765c38..8ad7519f7b58 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -78,8 +78,6 @@
 #define mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION_DCN2                                                          0x05ea
 #define mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION_DCN2_BASE_IDX                                                 2
 
-#define MAX_MEM_RANGES 8
-
 static const char * const gfxhub_client_ids[] = {
 	"CB",
 	"DB",
@@ -546,8 +544,10 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
 				      struct amdgpu_irq_src *source,
 				      struct amdgpu_iv_entry *entry)
 {
-	bool retry_fault = !!(entry->src_data[1] & 0x80);
-	bool write_fault = !!(entry->src_data[1] & 0x20);
+	bool retry_fault = !!(entry->src_data[1] &
+			      AMDGPU_GMC9_FAULT_SOURCE_DATA_RETRY);
+	bool write_fault = !!(entry->src_data[1] &
+			      AMDGPU_GMC9_FAULT_SOURCE_DATA_WRITE);
 	uint32_t status = 0, cid = 0, rw = 0, fed = 0;
 	struct amdgpu_task_info *task_info;
 	struct amdgpu_vmhub *hub;
@@ -636,10 +636,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
 
 	task_info = amdgpu_vm_get_task_info_pasid(adev, entry->pasid);
 	if (task_info) {
-		dev_err(adev->dev,
-			" for process %s pid %d thread %s pid %d)\n",
-			task_info->process_name, task_info->tgid,
-			task_info->task_name, task_info->pid);
+		amdgpu_vm_print_task_info(adev, task_info);
 		amdgpu_vm_put_task_info(task_info);
 	}
 
@@ -647,9 +644,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
 		addr, entry->client_id,
 		soc15_ih_clientid_name[entry->client_id]);
 
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+	if (amdgpu_is_multi_aid(adev))
 		dev_err(adev->dev, "  cookie node_id %d fault from die %s%d%s\n",
 			node_id, node_id % 4 == 3 ? "RSV" : "AID", node_id / 4,
 			node_id % 4 == 1 ? ".XCD0" : node_id % 4 == 2 ? ".XCD1" : "");
@@ -798,9 +793,7 @@ static bool gmc_v9_0_use_invalidate_semaphore(struct amdgpu_device *adev,
 				       uint32_t vmhub)
 {
 	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+	    amdgpu_is_multi_aid(adev))
 		return false;
 
 	return ((vmhub == AMDGPU_MMHUB0(0) ||
@@ -1082,27 +1075,6 @@ static void gmc_v9_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned int v
  * 0 valid
  */
 
-static uint64_t gmc_v9_0_map_mtype(struct amdgpu_device *adev, uint32_t flags)
-
-{
-	switch (flags) {
-	case AMDGPU_VM_MTYPE_DEFAULT:
-		return AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_NC);
-	case AMDGPU_VM_MTYPE_NC:
-		return AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_NC);
-	case AMDGPU_VM_MTYPE_WC:
-		return AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_WC);
-	case AMDGPU_VM_MTYPE_RW:
-		return AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_RW);
-	case AMDGPU_VM_MTYPE_CC:
-		return AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_CC);
-	case AMDGPU_VM_MTYPE_UC:
-		return AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_UC);
-	default:
-		return AMDGPU_PTE_MTYPE_VG10(0ULL, MTYPE_NC);
-	}
-}
-
 static void gmc_v9_0_get_vm_pde(struct amdgpu_device *adev, int level,
 				uint64_t *addr, uint64_t *flags)
 {
@@ -1130,8 +1102,9 @@ static void gmc_v9_0_get_vm_pde(struct amdgpu_device *adev, int level,
 }
 
 static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev,
+					 struct amdgpu_vm *vm,
 					 struct amdgpu_bo *bo,
-					 struct amdgpu_bo_va_mapping *mapping,
+					 uint32_t vm_flags,
 					 uint64_t *flags)
 {
 	struct amdgpu_device *bo_adev = amdgpu_ttm_adev(bo->tbo.bdev);
@@ -1141,7 +1114,6 @@ static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev,
 				     AMDGPU_GEM_CREATE_EXT_COHERENT);
 	bool ext_coherent = bo->flags & AMDGPU_GEM_CREATE_EXT_COHERENT;
 	bool uncached = bo->flags & AMDGPU_GEM_CREATE_UNCACHED;
-	struct amdgpu_vm *vm = mapping->bo_va->base.vm;
 	unsigned int mtype_local, mtype;
 	uint32_t gc_ip_version = amdgpu_ip_version(adev, GC_HWIP, 0);
 	bool snoop = false;
@@ -1171,7 +1143,7 @@ static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev,
 					mtype = MTYPE_UC;
 				else
 					mtype = MTYPE_NC;
-				if (mapping->bo_va->is_xgmi)
+				if (amdgpu_xgmi_same_hive(adev, bo_adev))
 					snoop = true;
 			}
 		} else {
@@ -1212,10 +1184,7 @@ static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev,
 		if (uncached) {
 			mtype = MTYPE_UC;
 		} else if (ext_coherent) {
-			if (gc_ip_version == IP_VERSION(9, 5, 0) || adev->rev_id)
-				mtype = is_local ? MTYPE_CC : MTYPE_UC;
-			else
-				mtype = MTYPE_UC;
+			mtype = is_local ? MTYPE_CC : MTYPE_UC;
 		} else if (adev->flags & AMD_IS_APU) {
 			mtype = is_local ? mtype_local : MTYPE_NC;
 		} else {
@@ -1249,24 +1218,43 @@ static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev,
 }
 
 static void gmc_v9_0_get_vm_pte(struct amdgpu_device *adev,
-				struct amdgpu_bo_va_mapping *mapping,
+				struct amdgpu_vm *vm,
+				struct amdgpu_bo *bo,
+				uint32_t vm_flags,
 				uint64_t *flags)
 {
-	struct amdgpu_bo *bo = mapping->bo_va->base.bo;
-
-	*flags &= ~AMDGPU_PTE_EXECUTABLE;
-	*flags |= mapping->flags & AMDGPU_PTE_EXECUTABLE;
+	if (vm_flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		*flags |= AMDGPU_PTE_EXECUTABLE;
+	else
+		*flags &= ~AMDGPU_PTE_EXECUTABLE;
 
-	*flags &= ~AMDGPU_PTE_MTYPE_VG10_MASK;
-	*flags |= mapping->flags & AMDGPU_PTE_MTYPE_VG10_MASK;
+	switch (vm_flags & AMDGPU_VM_MTYPE_MASK) {
+	case AMDGPU_VM_MTYPE_DEFAULT:
+	case AMDGPU_VM_MTYPE_NC:
+	default:
+		*flags = AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_WC:
+		*flags |= AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_WC);
+		break;
+	case AMDGPU_VM_MTYPE_RW:
+		*flags |= AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_RW);
+		break;
+	case AMDGPU_VM_MTYPE_CC:
+		*flags |= AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_CC);
+		break;
+	case AMDGPU_VM_MTYPE_UC:
+		*flags |= AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_UC);
+		break;
+	}
 
-	if (mapping->flags & AMDGPU_PTE_PRT) {
+	if (vm_flags & AMDGPU_VM_PAGE_PRT) {
 		*flags |= AMDGPU_PTE_PRT;
 		*flags &= ~AMDGPU_PTE_VALID;
 	}
 
 	if ((*flags & AMDGPU_PTE_VALID) && bo)
-		gmc_v9_0_get_coherence_flags(adev, bo, mapping, flags);
+		gmc_v9_0_get_coherence_flags(adev, vm, bo, vm_flags, flags);
 }
 
 static void gmc_v9_0_override_vm_pte_flags(struct amdgpu_device *adev,
@@ -1278,9 +1266,8 @@ static void gmc_v9_0_override_vm_pte_flags(struct amdgpu_device *adev,
 	/* Only GFX 9.4.3 APUs associate GPUs with NUMA nodes. Local system
 	 * memory can use more efficient MTYPEs.
 	 */
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 3) &&
-	    amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 4) &&
-	    amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 5, 0))
+	if (!(adev->flags & AMD_IS_APU) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 3))
 		return;
 
 	/* Only direct-mapped memory allows us to determine the NUMA node from
@@ -1336,7 +1323,7 @@ static void gmc_v9_0_override_vm_pte_flags(struct amdgpu_device *adev,
 				mtype_local = MTYPE_CC;
 
 			*flags = AMDGPU_PTE_MTYPE_VG10(*flags, mtype_local);
-		} else if (adev->rev_id) {
+		} else {
 			/* MTYPE_UC case */
 			*flags = AMDGPU_PTE_MTYPE_VG10(*flags, MTYPE_CC);
 		}
@@ -1388,46 +1375,6 @@ static unsigned int gmc_v9_0_get_vbios_fb_size(struct amdgpu_device *adev)
 	return size;
 }
 
-static enum amdgpu_memory_partition
-gmc_v9_0_get_memory_partition(struct amdgpu_device *adev, u32 *supp_modes)
-{
-	enum amdgpu_memory_partition mode = UNKNOWN_MEMORY_PARTITION_MODE;
-
-	if (adev->nbio.funcs->get_memory_partition_mode)
-		mode = adev->nbio.funcs->get_memory_partition_mode(adev,
-								   supp_modes);
-
-	return mode;
-}
-
-static enum amdgpu_memory_partition
-gmc_v9_0_query_vf_memory_partition(struct amdgpu_device *adev)
-{
-	switch (adev->gmc.num_mem_partitions) {
-	case 0:
-		return UNKNOWN_MEMORY_PARTITION_MODE;
-	case 1:
-		return AMDGPU_NPS1_PARTITION_MODE;
-	case 2:
-		return AMDGPU_NPS2_PARTITION_MODE;
-	case 4:
-		return AMDGPU_NPS4_PARTITION_MODE;
-	default:
-		return AMDGPU_NPS1_PARTITION_MODE;
-	}
-
-	return AMDGPU_NPS1_PARTITION_MODE;
-}
-
-static enum amdgpu_memory_partition
-gmc_v9_0_query_memory_partition(struct amdgpu_device *adev)
-{
-	if (amdgpu_sriov_vf(adev))
-		return gmc_v9_0_query_vf_memory_partition(adev);
-
-	return gmc_v9_0_get_memory_partition(adev, NULL);
-}
-
 static bool gmc_v9_0_need_reset_on_init(struct amdgpu_device *adev)
 {
 	if (adev->nbio.funcs && adev->nbio.funcs->is_nps_switch_requested &&
@@ -1444,12 +1391,11 @@ static const struct amdgpu_gmc_funcs gmc_v9_0_gmc_funcs = {
 	.flush_gpu_tlb_pasid = gmc_v9_0_flush_gpu_tlb_pasid,
 	.emit_flush_gpu_tlb = gmc_v9_0_emit_flush_gpu_tlb,
 	.emit_pasid_mapping = gmc_v9_0_emit_pasid_mapping,
-	.map_mtype = gmc_v9_0_map_mtype,
 	.get_vm_pde = gmc_v9_0_get_vm_pde,
 	.get_vm_pte = gmc_v9_0_get_vm_pte,
 	.override_vm_pte_flags = gmc_v9_0_override_vm_pte_flags,
 	.get_vbios_fb_size = gmc_v9_0_get_vbios_fb_size,
-	.query_mem_partition_mode = &gmc_v9_0_query_memory_partition,
+	.query_mem_partition_mode = &amdgpu_gmc_query_memory_partition,
 	.request_mem_partition_mode = &amdgpu_gmc_request_memory_partition,
 	.need_reset_on_init = &gmc_v9_0_need_reset_on_init,
 };
@@ -1498,14 +1444,13 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev)
 			adev->umc.channel_idx_tbl = &umc_v6_7_channel_idx_tbl_second[0][0];
 		break;
 	case IP_VERSION(12, 0, 0):
+	case IP_VERSION(12, 5, 0):
 		adev->umc.max_ras_err_cnt_per_query =
 			UMC_V12_0_TOTAL_CHANNEL_NUM(adev) * UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL;
 		adev->umc.channel_inst_num = UMC_V12_0_CHANNEL_INSTANCE_NUM;
 		adev->umc.umc_inst_num = UMC_V12_0_UMC_INSTANCE_NUM;
 		adev->umc.node_inst_num /= UMC_V12_0_UMC_INSTANCE_NUM;
 		adev->umc.channel_offs = UMC_V12_0_PER_CHANNEL_OFFSET;
-		adev->umc.active_mask = adev->aid_mask;
-		adev->umc.retire_unit = UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL;
 		if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu)
 			adev->umc.ras = &umc_v12_0_ras;
 		break;
@@ -1524,6 +1469,7 @@ static void gmc_v9_0_set_mmhub_funcs(struct amdgpu_device *adev)
 		adev->mmhub.funcs = &mmhub_v1_7_funcs;
 		break;
 	case IP_VERSION(1, 8, 0):
+	case IP_VERSION(1, 8, 1):
 		adev->mmhub.funcs = &mmhub_v1_8_funcs;
 		break;
 	default:
@@ -1556,9 +1502,7 @@ static void gmc_v9_0_set_mmhub_ras_funcs(struct amdgpu_device *adev)
 
 static void gmc_v9_0_set_gfxhub_funcs(struct amdgpu_device *adev)
 {
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+	if (amdgpu_is_multi_aid(adev))
 		adev->gfxhub.funcs = &gfxhub_v1_2_funcs;
 	else
 		adev->gfxhub.funcs = &gfxhub_v1_0_funcs;
@@ -1595,23 +1539,38 @@ static void gmc_v9_0_set_xgmi_ras_funcs(struct amdgpu_device *adev)
 
 static void gmc_v9_0_init_nps_details(struct amdgpu_device *adev)
 {
+	enum amdgpu_memory_partition mode;
+	uint32_t supp_modes;
+	int i;
+
 	adev->gmc.supported_nps_modes = 0;
 
 	if (amdgpu_sriov_vf(adev) || (adev->flags & AMD_IS_APU))
 		return;
 
-	/*TODO: Check PSP version also which supports NPS switch. Otherwise keep
+	mode = amdgpu_gmc_get_memory_partition(adev, &supp_modes);
+
+	/* Mode detected by hardware and supported modes available */
+	if ((mode != UNKNOWN_MEMORY_PARTITION_MODE) && supp_modes) {
+		while ((i = ffs(supp_modes))) {
+			if (AMDGPU_ALL_NPS_MASK & BIT(i))
+				adev->gmc.supported_nps_modes |= BIT(i);
+			supp_modes &= supp_modes - 1;
+		}
+	} else {
+		/*TODO: Check PSP version also which supports NPS switch. Otherwise keep
 	 * supported modes as 0.
 	 */
-	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
-	case IP_VERSION(9, 4, 3):
-	case IP_VERSION(9, 4, 4):
-		adev->gmc.supported_nps_modes =
-			BIT(AMDGPU_NPS1_PARTITION_MODE) |
-			BIT(AMDGPU_NPS4_PARTITION_MODE);
-		break;
-	default:
-		break;
+		switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+		case IP_VERSION(9, 4, 3):
+		case IP_VERSION(9, 4, 4):
+			adev->gmc.supported_nps_modes =
+				BIT(AMDGPU_NPS1_PARTITION_MODE) |
+				BIT(AMDGPU_NPS4_PARTITION_MODE);
+			break;
+		default:
+			break;
+		}
 	}
 }
 
@@ -1625,9 +1584,7 @@ static int gmc_v9_0_early_init(struct amdgpu_ip_block *ip_block)
 	 */
 	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 0) ||
 	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+	    amdgpu_is_multi_aid(adev))
 		adev->gmc.xgmi.supported = true;
 
 	if (amdgpu_ip_version(adev, XGMI_HWIP, 0) == IP_VERSION(6, 1, 0)) {
@@ -1636,8 +1593,7 @@ static int gmc_v9_0_early_init(struct amdgpu_ip_block *ip_block)
 			adev->smuio.funcs->is_host_gpu_xgmi_supported(adev);
 	}
 
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4)) {
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) {
 		enum amdgpu_pkg_type pkg_type =
 			adev->smuio.funcs->get_pkg_type(adev);
 		/* On GFXIP 9.4.3. APU, there is no physical VRAM domain present
@@ -1715,7 +1671,7 @@ static void gmc_v9_0_vram_gtt_location(struct amdgpu_device *adev,
 
 	/* add the xgmi offset of the physical node */
 	base += adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size;
-	if (adev->gmc.xgmi.connected_to_cpu) {
+	if (amdgpu_gmc_is_pdb0_enabled(adev)) {
 		amdgpu_gmc_sysvm_location(adev, mc);
 	} else {
 		amdgpu_gmc_vram_location(adev, mc, base);
@@ -1830,7 +1786,7 @@ static int gmc_v9_0_gart_init(struct amdgpu_device *adev)
 		return 0;
 	}
 
-	if (adev->gmc.xgmi.connected_to_cpu) {
+	if (amdgpu_gmc_is_pdb0_enabled(adev)) {
 		adev->gmc.vmid0_page_table_depth = 1;
 		adev->gmc.vmid0_page_table_block_size = 12;
 	} else {
@@ -1856,7 +1812,7 @@ static int gmc_v9_0_gart_init(struct amdgpu_device *adev)
 		if (r)
 			return r;
 
-		if (adev->gmc.xgmi.connected_to_cpu)
+		if (amdgpu_gmc_is_pdb0_enabled(adev))
 			r = amdgpu_gmc_pdb0_alloc(adev);
 	}
 
@@ -1878,192 +1834,25 @@ static void gmc_v9_0_save_registers(struct amdgpu_device *adev)
 		adev->gmc.sdpif_register = RREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0);
 }
 
-static bool gmc_v9_0_validate_partition_info(struct amdgpu_device *adev)
-{
-	enum amdgpu_memory_partition mode;
-	u32 supp_modes;
-	bool valid;
-
-	mode = gmc_v9_0_get_memory_partition(adev, &supp_modes);
-
-	/* Mode detected by hardware not present in supported modes */
-	if ((mode != UNKNOWN_MEMORY_PARTITION_MODE) &&
-	    !(BIT(mode - 1) & supp_modes))
-		return false;
-
-	switch (mode) {
-	case UNKNOWN_MEMORY_PARTITION_MODE:
-	case AMDGPU_NPS1_PARTITION_MODE:
-		valid = (adev->gmc.num_mem_partitions == 1);
-		break;
-	case AMDGPU_NPS2_PARTITION_MODE:
-		valid = (adev->gmc.num_mem_partitions == 2);
-		break;
-	case AMDGPU_NPS4_PARTITION_MODE:
-		valid = (adev->gmc.num_mem_partitions == 3 ||
-			 adev->gmc.num_mem_partitions == 4);
-		break;
-	default:
-		valid = false;
-	}
-
-	return valid;
-}
-
-static bool gmc_v9_0_is_node_present(int *node_ids, int num_ids, int nid)
-{
-	int i;
-
-	/* Check if node with id 'nid' is present in 'node_ids' array */
-	for (i = 0; i < num_ids; ++i)
-		if (node_ids[i] == nid)
-			return true;
-
-	return false;
-}
-
-static void
-gmc_v9_0_init_acpi_mem_ranges(struct amdgpu_device *adev,
-			      struct amdgpu_mem_partition_info *mem_ranges)
-{
-	struct amdgpu_numa_info numa_info;
-	int node_ids[MAX_MEM_RANGES];
-	int num_ranges = 0, ret;
-	int num_xcc, xcc_id;
-	uint32_t xcc_mask;
-
-	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
-	xcc_mask = (1U << num_xcc) - 1;
-
-	for_each_inst(xcc_id, xcc_mask)	{
-		ret = amdgpu_acpi_get_mem_info(adev, xcc_id, &numa_info);
-		if (ret)
-			continue;
-
-		if (numa_info.nid == NUMA_NO_NODE) {
-			mem_ranges[0].size = numa_info.size;
-			mem_ranges[0].numa.node = numa_info.nid;
-			num_ranges = 1;
-			break;
-		}
-
-		if (gmc_v9_0_is_node_present(node_ids, num_ranges,
-					     numa_info.nid))
-			continue;
-
-		node_ids[num_ranges] = numa_info.nid;
-		mem_ranges[num_ranges].numa.node = numa_info.nid;
-		mem_ranges[num_ranges].size = numa_info.size;
-		++num_ranges;
-	}
-
-	adev->gmc.num_mem_partitions = num_ranges;
-}
-
-static void
-gmc_v9_0_init_sw_mem_ranges(struct amdgpu_device *adev,
-			    struct amdgpu_mem_partition_info *mem_ranges)
+static void gmc_v9_4_3_init_vram_info(struct amdgpu_device *adev)
 {
-	enum amdgpu_memory_partition mode;
-	u32 start_addr = 0, size;
-	int i, r, l;
-
-	mode = gmc_v9_0_query_memory_partition(adev);
+	static const u32 regBIF_BIOS_SCRATCH_4 = 0x50;
+	u32 vram_info;
 
-	switch (mode) {
-	case UNKNOWN_MEMORY_PARTITION_MODE:
-		adev->gmc.num_mem_partitions = 0;
-		break;
-	case AMDGPU_NPS1_PARTITION_MODE:
-		adev->gmc.num_mem_partitions = 1;
-		break;
-	case AMDGPU_NPS2_PARTITION_MODE:
-		adev->gmc.num_mem_partitions = 2;
-		break;
-	case AMDGPU_NPS4_PARTITION_MODE:
-		if (adev->flags & AMD_IS_APU)
-			adev->gmc.num_mem_partitions = 3;
-		else
-			adev->gmc.num_mem_partitions = 4;
-		break;
-	default:
-		adev->gmc.num_mem_partitions = 1;
-		break;
-	}
-
-	/* Use NPS range info, if populated */
-	r = amdgpu_gmc_get_nps_memranges(adev, mem_ranges,
-					 &adev->gmc.num_mem_partitions);
-	if (!r) {
-		l = 0;
-		for (i = 1; i < adev->gmc.num_mem_partitions; ++i) {
-			if (mem_ranges[i].range.lpfn >
-			    mem_ranges[i - 1].range.lpfn)
-				l = i;
-		}
-
-	} else {
-		if (!adev->gmc.num_mem_partitions) {
-			dev_err(adev->dev,
-				"Not able to detect NPS mode, fall back to NPS1");
-			adev->gmc.num_mem_partitions = 1;
-		}
-		/* Fallback to sw based calculation */
-		size = (adev->gmc.real_vram_size + SZ_16M) >> AMDGPU_GPU_PAGE_SHIFT;
-		size /= adev->gmc.num_mem_partitions;
-
-		for (i = 0; i < adev->gmc.num_mem_partitions; ++i) {
-			mem_ranges[i].range.fpfn = start_addr;
-			mem_ranges[i].size =
-				((u64)size << AMDGPU_GPU_PAGE_SHIFT);
-			mem_ranges[i].range.lpfn = start_addr + size - 1;
-			start_addr += size;
-		}
-
-		l = adev->gmc.num_mem_partitions - 1;
-	}
-
-	/* Adjust the last one */
-	mem_ranges[l].range.lpfn =
-		(adev->gmc.real_vram_size >> AMDGPU_GPU_PAGE_SHIFT) - 1;
-	mem_ranges[l].size =
-		adev->gmc.real_vram_size -
-		((u64)mem_ranges[l].range.fpfn << AMDGPU_GPU_PAGE_SHIFT);
-}
-
-static int gmc_v9_0_init_mem_ranges(struct amdgpu_device *adev)
-{
-	bool valid;
+	adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM;
+	adev->gmc.vram_width = 128 * 64;
 
-	adev->gmc.mem_partitions = kcalloc(MAX_MEM_RANGES,
-					   sizeof(struct amdgpu_mem_partition_info),
-					   GFP_KERNEL);
-	if (!adev->gmc.mem_partitions)
-		return -ENOMEM;
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM3E;
 
-	/* TODO : Get the range from PSP/Discovery for dGPU */
-	if (adev->gmc.is_app_apu)
-		gmc_v9_0_init_acpi_mem_ranges(adev, adev->gmc.mem_partitions);
-	else
-		gmc_v9_0_init_sw_mem_ranges(adev, adev->gmc.mem_partitions);
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) &&
+		adev->rev_id == 0x3)
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM3E;
 
-	if (amdgpu_sriov_vf(adev))
-		valid = true;
-	else
-		valid = gmc_v9_0_validate_partition_info(adev);
-	if (!valid) {
-		/* TODO: handle invalid case */
-		dev_WARN(adev->dev,
-			 "Mem ranges not matching with hardware config");
+	if (!(adev->flags & AMD_IS_APU) && !amdgpu_sriov_vf(adev)) {
+		vram_info = RREG32(regBIF_BIOS_SCRATCH_4);
+		adev->gmc.vram_vendor = vram_info & 0xF;
 	}
-
-	return 0;
-}
-
-static void gmc_v9_4_3_init_vram_info(struct amdgpu_device *adev)
-{
-	adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM;
-	adev->gmc.vram_width = 128 * 64;
 }
 
 static int gmc_v9_0_sw_init(struct amdgpu_ip_block *ip_block)
@@ -2078,9 +1867,7 @@ static int gmc_v9_0_sw_init(struct amdgpu_ip_block *ip_block)
 
 	spin_lock_init(&adev->gmc.invalidate_lock);
 
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0)) {
+	if (amdgpu_is_multi_aid(adev)) {
 		gmc_v9_4_3_init_vram_info(adev);
 	} else if (!adev->bios) {
 		if (adev->flags & AMD_IS_APU) {
@@ -2230,10 +2017,8 @@ static int gmc_v9_0_sw_init(struct amdgpu_ip_block *ip_block)
 
 	amdgpu_gmc_get_vbios_allocations(adev);
 
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0)) {
-		r = gmc_v9_0_init_mem_ranges(adev);
+	if (amdgpu_is_multi_aid(adev)) {
+		r = amdgpu_gmc_init_mem_ranges(adev);
 		if (r)
 			return r;
 	}
@@ -2261,9 +2046,7 @@ static int gmc_v9_0_sw_init(struct amdgpu_ip_block *ip_block)
 	adev->vm_manager.first_kfd_vmid =
 		(amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) ||
 		 amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) ||
-		 amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-		 amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
-		 amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0)) ?
+		 amdgpu_is_multi_aid(adev)) ?
 			3 :
 			8;
 
@@ -2275,9 +2058,7 @@ static int gmc_v9_0_sw_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+	if (amdgpu_is_multi_aid(adev))
 		amdgpu_gmc_sysfs_init(adev);
 
 	return 0;
@@ -2287,9 +2068,7 @@ static int gmc_v9_0_sw_fini(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_device *adev = ip_block->adev;
 
-	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0))
+	if (amdgpu_is_multi_aid(adev))
 		amdgpu_gmc_sysfs_fini(adev);
 
 	amdgpu_gmc_ras_fini(adev);
@@ -2363,7 +2142,7 @@ static int gmc_v9_0_gart_enable(struct amdgpu_device *adev)
 {
 	int r;
 
-	if (adev->gmc.xgmi.connected_to_cpu)
+	if (amdgpu_gmc_is_pdb0_enabled(adev))
 		amdgpu_gmc_init_pdb0(adev);
 
 	if (adev->gart.bo == NULL) {
@@ -2410,13 +2189,6 @@ static int gmc_v9_0_hw_init(struct amdgpu_ip_block *ip_block)
 	adev->gmc.flush_tlb_needs_extra_type_2 =
 		amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 0) &&
 		adev->gmc.xgmi.num_physical_nodes;
-	/*
-	 * TODO: This workaround is badly documented and had a buggy
-	 * implementation. We should probably verify what we do here.
-	 */
-	adev->gmc.flush_tlb_needs_extra_type_0 =
-		amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) &&
-		adev->rev_id == 0;
 
 	/* The sequence of these two function calls matters.*/
 	gmc_v9_0_init_golden_registers(adev);
@@ -2434,7 +2206,7 @@ static int gmc_v9_0_hw_init(struct amdgpu_ip_block *ip_block)
 	adev->hdp.funcs->init_registers(adev);
 
 	/* After HDP is initialized, flush HDP.*/
-	adev->hdp.funcs->flush_hdp(adev, NULL);
+	amdgpu_device_flush_hdp(adev, NULL);
 
 	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
 		value = false;
@@ -2528,7 +2300,7 @@ static int gmc_v9_0_resume(struct amdgpu_ip_block *ip_block)
 	 * information again.
 	 */
 	if (adev->gmc.reset_flags & AMDGPU_GMC_INIT_RESET_NPS) {
-		gmc_v9_0_init_sw_mem_ranges(adev, adev->gmc.mem_partitions);
+		amdgpu_gmc_init_sw_mem_ranges(adev, adev->gmc.mem_partitions);
 		adev->gmc.reset_flags &= ~AMDGPU_GMC_INIT_RESET_NPS;
 	}
 
@@ -2541,7 +2313,7 @@ static int gmc_v9_0_resume(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-static bool gmc_v9_0_is_idle(void *handle)
+static bool gmc_v9_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* MC is always ready in GMC v9.*/
 	return true;
@@ -2571,9 +2343,9 @@ static int gmc_v9_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
-static void gmc_v9_0_get_clockgating_state(void *handle, u64 *flags)
+static void gmc_v9_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	adev->mmhub.funcs->get_clockgating(adev, flags);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
index 194026e9be33..e6c0d86d3486 100644
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
@@ -21,7 +21,6 @@
  *
  */
 #include "amdgpu.h"
-#include "amdgpu_atombios.h"
 #include "hdp_v4_0.h"
 #include "amdgpu_ras.h"
 
@@ -37,17 +36,6 @@
 #define HDP_MEM_POWER_CTRL__RC_MEM_POWER_LS_EN_MASK     0x00020000L
 #define mmHDP_MEM_POWER_CTRL_BASE_IDX   0
 
-static void hdp_v4_0_flush_hdp(struct amdgpu_device *adev,
-				struct amdgpu_ring *ring)
-{
-	if (!ring || !ring->funcs->emit_wreg) {
-		WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-		RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
-	} else {
-		amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-	}
-}
-
 static void hdp_v4_0_invalidate_hdp(struct amdgpu_device *adev,
 				    struct amdgpu_ring *ring)
 {
@@ -181,7 +169,7 @@ struct amdgpu_hdp_ras hdp_v4_0_ras = {
 };
 
 const struct amdgpu_hdp_funcs hdp_v4_0_funcs = {
-	.flush_hdp = hdp_v4_0_flush_hdp,
+	.flush_hdp = amdgpu_hdp_generic_flush,
 	.invalidate_hdp = hdp_v4_0_invalidate_hdp,
 	.update_clock_gating = hdp_v4_0_update_clock_gating,
 	.get_clock_gating_state = hdp_v4_0_get_clockgating_state,
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c
index d3962d469088..8bc001dc9f63 100644
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c
@@ -21,24 +21,12 @@
  *
  */
 #include "amdgpu.h"
-#include "amdgpu_atombios.h"
 #include "hdp_v5_0.h"
 
 #include "hdp/hdp_5_0_0_offset.h"
 #include "hdp/hdp_5_0_0_sh_mask.h"
 #include <uapi/linux/kfd_ioctl.h>
 
-static void hdp_v5_0_flush_hdp(struct amdgpu_device *adev,
-				struct amdgpu_ring *ring)
-{
-	if (!ring || !ring->funcs->emit_wreg) {
-		WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-		RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
-	} else {
-		amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-	}
-}
-
 static void hdp_v5_0_invalidate_hdp(struct amdgpu_device *adev,
 				    struct amdgpu_ring *ring)
 {
@@ -218,7 +206,7 @@ static void hdp_v5_0_init_registers(struct amdgpu_device *adev)
 }
 
 const struct amdgpu_hdp_funcs hdp_v5_0_funcs = {
-	.flush_hdp = hdp_v5_0_flush_hdp,
+	.flush_hdp = amdgpu_hdp_generic_flush,
 	.invalidate_hdp = hdp_v5_0_invalidate_hdp,
 	.update_clock_gating = hdp_v5_0_update_clock_gating,
 	.get_clock_gating_state = hdp_v5_0_get_clockgating_state,
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c b/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c
index f52552c5fa27..40940b4ab400 100644
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c
@@ -21,7 +21,6 @@
  *
  */
 #include "amdgpu.h"
-#include "amdgpu_atombios.h"
 #include "hdp_v5_2.h"
 
 #include "hdp/hdp_5_2_1_offset.h"
@@ -34,7 +33,17 @@ static void hdp_v5_2_flush_hdp(struct amdgpu_device *adev,
 	if (!ring || !ring->funcs->emit_wreg) {
 		WREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2,
 			0);
-		RREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
+		if (amdgpu_sriov_vf(adev)) {
+			/* this is fine because SR_IOV doesn't remap the register */
+			RREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
+		} else {
+			/* We just need to read back a register to post the write.
+			 * Reading back the remapped register causes problems on
+			 * some platforms so just read back the memory size register.
+			 */
+			if (adev->nbio.funcs->get_memsize)
+				adev->nbio.funcs->get_memsize(adev);
+		}
 	} else {
 		amdgpu_ring_emit_wreg(ring,
 			(adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2,
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c
index 6948fe9956ce..ec20daf4272c 100644
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c
@@ -21,7 +21,6 @@
  *
  */
 #include "amdgpu.h"
-#include "amdgpu_atombios.h"
 #include "hdp_v6_0.h"
 
 #include "hdp/hdp_6_0_0_offset.h"
@@ -31,17 +30,6 @@
 #define regHDP_CLK_CNTL_V6_1	0xd5
 #define regHDP_CLK_CNTL_V6_1_BASE_IDX 0
 
-static void hdp_v6_0_flush_hdp(struct amdgpu_device *adev,
-				struct amdgpu_ring *ring)
-{
-	if (!ring || !ring->funcs->emit_wreg) {
-		WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-		RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
-	} else {
-		amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-	}
-}
-
 static void hdp_v6_0_update_clock_gating(struct amdgpu_device *adev,
 					 bool enable)
 {
@@ -150,7 +138,7 @@ static void hdp_v6_0_get_clockgating_state(struct amdgpu_device *adev,
 }
 
 const struct amdgpu_hdp_funcs hdp_v6_0_funcs = {
-	.flush_hdp = hdp_v6_0_flush_hdp,
+	.flush_hdp = amdgpu_hdp_generic_flush,
 	.update_clock_gating = hdp_v6_0_update_clock_gating,
 	.get_clock_gating_state = hdp_v6_0_get_clockgating_state,
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c
index 63820329f67e..ed1debc03507 100644
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c
@@ -21,24 +21,12 @@
  *
  */
 #include "amdgpu.h"
-#include "amdgpu_atombios.h"
 #include "hdp_v7_0.h"
 
 #include "hdp/hdp_7_0_0_offset.h"
 #include "hdp/hdp_7_0_0_sh_mask.h"
 #include <uapi/linux/kfd_ioctl.h>
 
-static void hdp_v7_0_flush_hdp(struct amdgpu_device *adev,
-				struct amdgpu_ring *ring)
-{
-	if (!ring || !ring->funcs->emit_wreg) {
-		WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-		RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
-	} else {
-		amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-	}
-}
-
 static void hdp_v7_0_update_clock_gating(struct amdgpu_device *adev,
 					 bool enable)
 {
@@ -138,7 +126,7 @@ static void hdp_v7_0_get_clockgating_state(struct amdgpu_device *adev,
 }
 
 const struct amdgpu_hdp_funcs hdp_v7_0_funcs = {
-	.flush_hdp = hdp_v7_0_flush_hdp,
+	.flush_hdp = amdgpu_hdp_generic_flush,
 	.update_clock_gating = hdp_v7_0_update_clock_gating,
 	.get_clock_gating_state = hdp_v7_0_get_clockgating_state,
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
index 8ac3d3282268..01cadf898c00 100644
--- a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
@@ -157,6 +157,9 @@ static int iceland_ih_irq_init(struct amdgpu_device *adev)
 	/* enable interrupts */
 	iceland_ih_enable_interrupts(adev);
 
+	if (adev->irq.ih_soft.ring_size)
+		adev->irq.ih_soft.enabled = true;
+
 	return 0;
 }
 
@@ -194,6 +197,9 @@ static u32 iceland_ih_get_wptr(struct amdgpu_device *adev,
 
 	wptr = le32_to_cpu(*ih->wptr_cpu);
 
+	if (ih == &adev->irq.ih_soft)
+		goto out;
+
 	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
 		goto out;
 
@@ -296,6 +302,10 @@ static int iceland_ih_sw_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true);
+	if (r)
+		return r;
+
 	r = amdgpu_irq_init(adev);
 
 	return r;
@@ -335,9 +345,9 @@ static int iceland_ih_resume(struct amdgpu_ip_block *ip_block)
 	return iceland_ih_hw_init(ip_block);
 }
 
-static bool iceland_ih_is_idle(void *handle)
+static bool iceland_ih_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	u32 tmp = RREG32(mmSRBM_STATUS);
 
 	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c b/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c
index f8a485164437..333e9c30c091 100644
--- a/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v6_0.c
@@ -349,6 +349,7 @@ static int ih_v6_0_irq_init(struct amdgpu_device *adev)
 			if (ret)
 				return ret;
 		}
+		ih[i]->overflow = false;
 	}
 
 	/* update doorbell range for ih ring 0 */
@@ -446,7 +447,10 @@ static u32 ih_v6_0_get_wptr(struct amdgpu_device *adev,
 	wptr = RREG32_NO_KIQ(ih_regs->ih_rb_wptr);
 	if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
 		goto out;
-	wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+	if (!amdgpu_sriov_vf(adev))
+		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+	else
+		ih->overflow = true;
 
 	/* When a ring buffer overflow happen start parsing interrupt
 	 * from the last not overwritten vector (wptr + 32). Hopefully
@@ -583,8 +587,7 @@ static int ih_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
 	/* use gpu virtual address for ih ring
 	 * until ih_checken is programmed to allow
 	 * use bus address for ih ring by psp bl */
-	use_bus_addr =
-		(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) ? false : true;
+	use_bus_addr = adev->firmware.load_type != AMDGPU_FW_LOAD_PSP;
 	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, IH_RING_SIZE, use_bus_addr);
 	if (r)
 		return r;
@@ -652,7 +655,7 @@ static int ih_v6_0_resume(struct amdgpu_ip_block *ip_block)
 	return ih_v6_0_hw_init(ip_block);
 }
 
-static bool ih_v6_0_is_idle(void *handle)
+static bool ih_v6_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* todo */
 	return true;
@@ -768,9 +771,9 @@ static int ih_v6_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
-static void ih_v6_0_get_clockgating_state(void *handle, u64 *flags)
+static void ih_v6_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (!RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL))
 		*flags |= AMD_CG_SUPPORT_IH_CG;
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c
index dd0042efceec..95b3f4e55ec3 100644
--- a/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c
@@ -562,8 +562,7 @@ static int ih_v6_1_sw_init(struct amdgpu_ip_block *ip_block)
 	/* use gpu virtual address for ih ring
 	 * until ih_checken is programmed to allow
 	 * use bus address for ih ring by psp bl */
-	use_bus_addr =
-		(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) ? false : true;
+	use_bus_addr = adev->firmware.load_type != AMDGPU_FW_LOAD_PSP;
 	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 256 * 1024, use_bus_addr);
 	if (r)
 		return r;
@@ -631,7 +630,7 @@ static int ih_v6_1_resume(struct amdgpu_ip_block *ip_block)
 	return ih_v6_1_hw_init(ip_block);
 }
 
-static bool ih_v6_1_is_idle(void *handle)
+static bool ih_v6_1_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* todo */
 	return true;
@@ -749,9 +748,9 @@ static int ih_v6_1_set_powergating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
-static void ih_v6_1_get_clockgating_state(void *handle, u64 *flags)
+static void ih_v6_1_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (!RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL))
 		*flags |= AMD_CG_SUPPORT_IH_CG;
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c
index 8f9b15c171f3..b32ea4129c61 100644
--- a/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c
@@ -552,8 +552,7 @@ static int ih_v7_0_sw_init(struct amdgpu_ip_block *ip_block)
 	/* use gpu virtual address for ih ring
 	 * until ih_checken is programmed to allow
 	 * use bus address for ih ring by psp bl */
-	use_bus_addr =
-		(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) ? false : true;
+	use_bus_addr = adev->firmware.load_type != AMDGPU_FW_LOAD_PSP;
 	r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 256 * 1024, use_bus_addr);
 	if (r)
 		return r;
@@ -621,7 +620,7 @@ static int ih_v7_0_resume(struct amdgpu_ip_block *ip_block)
 	return ih_v7_0_hw_init(ip_block);
 }
 
-static bool ih_v7_0_is_idle(void *handle)
+static bool ih_v7_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
 	/* todo */
 	return true;
@@ -739,9 +738,9 @@ static int ih_v7_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 	return 0;
 }
 
-static void ih_v7_0_get_clockgating_state(void *handle, u64 *flags)
+static void ih_v7_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	if (!RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL))
 		*flags |= AMD_CG_SUPPORT_IH_CG;
diff --git a/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c b/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c
index aeca5c08ea2f..cc626036ed9c 100644
--- a/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c
@@ -32,6 +32,7 @@
 #include "gc/gc_11_0_0_sh_mask.h"
 
 MODULE_FIRMWARE("amdgpu/gc_11_0_0_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_0_0_imu_kicker.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_1_imu.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_2_imu.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_0_3_imu.bin");
@@ -39,6 +40,7 @@ MODULE_FIRMWARE("amdgpu/gc_11_0_4_imu.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_5_0_imu.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_5_1_imu.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_5_2_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_imu.bin");
 
 static int imu_v11_0_init_microcode(struct amdgpu_device *adev)
 {
@@ -50,8 +52,12 @@ static int imu_v11_0_init_microcode(struct amdgpu_device *adev)
 	DRM_DEBUG("\n");
 
 	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
-	err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
-				   "amdgpu/%s_imu.bin", ucode_prefix);
+	if (amdgpu_is_kicker_fw(adev))
+		err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_imu_kicker.bin", ucode_prefix);
+	else
+		err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_imu.bin", ucode_prefix);
 	if (err)
 		goto out;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/imu_v12_0.c b/drivers/gpu/drm/amd/amdgpu/imu_v12_0.c
index df898dbb746e..58cd87db8061 100644
--- a/drivers/gpu/drm/amd/amdgpu/imu_v12_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/imu_v12_0.c
@@ -34,12 +34,13 @@
 
 MODULE_FIRMWARE("amdgpu/gc_12_0_0_imu.bin");
 MODULE_FIRMWARE("amdgpu/gc_12_0_1_imu.bin");
+MODULE_FIRMWARE("amdgpu/gc_12_0_1_imu_kicker.bin");
 
 #define TRANSFER_RAM_MASK	0x001c0000
 
 static int imu_v12_0_init_microcode(struct amdgpu_device *adev)
 {
-	char ucode_prefix[15];
+	char ucode_prefix[30];
 	int err;
 	const struct imu_firmware_header_v1_0 *imu_hdr;
 	struct amdgpu_firmware_info *info = NULL;
@@ -47,8 +48,12 @@ static int imu_v12_0_init_microcode(struct amdgpu_device *adev)
 	DRM_DEBUG("\n");
 
 	amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix));
-	err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
-				   "amdgpu/%s_imu.bin", ucode_prefix);
+	if (amdgpu_is_kicker_fw(adev))
+		err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_imu_kicker.bin", ucode_prefix);
+	else
+		err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED,
+					   "amdgpu/%s_imu.bin", ucode_prefix);
 	if (err)
 		goto out;
 
@@ -362,7 +367,7 @@ static void program_imu_rlc_ram(struct amdgpu_device *adev,
 static void imu_v12_0_program_rlc_ram(struct amdgpu_device *adev)
 {
 	u32 reg_data, size = 0;
-	const u32 *data;
+	const u32 *data = NULL;
 	int r = -EINVAL;
 
 	WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_INDEX, 0x2);
diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.c b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.c
index 964c29ef25dc..0027a639c7e6 100644
--- a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.c
@@ -50,26 +50,29 @@ static int isp_v4_1_0_hw_init(struct amdgpu_isp *isp)
 
 	isp_base = adev->rmmio_base;
 
-	isp->isp_cell = kcalloc(2, sizeof(struct mfd_cell), GFP_KERNEL);
+	isp->isp_cell = kcalloc(3, sizeof(struct mfd_cell), GFP_KERNEL);
 	if (!isp->isp_cell) {
 		r = -ENOMEM;
-		DRM_ERROR("%s: isp mfd cell alloc failed\n", __func__);
+		drm_err(&adev->ddev,
+			"%s: isp mfd cell alloc failed\n", __func__);
 		goto failure;
 	}
 
-	num_res = MAX_ISP410_MEM_RES + MAX_ISP410_SENSOR_RES + MAX_ISP410_INT_SRC;
+	num_res = MAX_ISP410_MEM_RES + MAX_ISP410_INT_SRC;
 	isp->isp_res = kcalloc(num_res, sizeof(struct resource),
 			       GFP_KERNEL);
 	if (!isp->isp_res) {
 		r = -ENOMEM;
-		DRM_ERROR("%s: isp mfd res alloc failed\n", __func__);
+		drm_err(&adev->ddev,
+			"%s: isp mfd res alloc failed\n", __func__);
 		goto failure;
 	}
 
 	isp->isp_pdata = kzalloc(sizeof(*isp->isp_pdata), GFP_KERNEL);
 	if (!isp->isp_pdata) {
 		r = -ENOMEM;
-		DRM_ERROR("%s: isp platform data alloc failed\n", __func__);
+		drm_err(&adev->ddev,
+			"%s: isp platform data alloc failed\n", __func__);
 		goto failure;
 	}
 
@@ -88,14 +91,7 @@ static int isp_v4_1_0_hw_init(struct amdgpu_isp *isp)
 	isp->isp_res[1].start = isp_base + ISP410_PHY0_OFFSET;
 	isp->isp_res[1].end = isp_base + ISP410_PHY0_OFFSET + ISP410_PHY0_SIZE;
 
-	isp->isp_res[2].name = "isp_gpio_sensor0_reg";
-	isp->isp_res[2].flags = IORESOURCE_MEM;
-	isp->isp_res[2].start = isp_base + ISP410_GPIO_SENSOR0_OFFSET;
-	isp->isp_res[2].end = isp_base + ISP410_GPIO_SENSOR0_OFFSET +
-			      ISP410_GPIO_SENSOR0_SIZE;
-
-	for (idx = MAX_ISP410_MEM_RES + MAX_ISP410_SENSOR_RES, int_idx = 0;
-	     idx < num_res; idx++, int_idx++) {
+	for (idx = MAX_ISP410_MEM_RES, int_idx = 0; idx < num_res; idx++, int_idx++) {
 		isp->isp_res[idx].name = "isp_4_1_0_irq";
 		isp->isp_res[idx].flags = IORESOURCE_IRQ;
 		isp->isp_res[idx].start =
@@ -110,11 +106,12 @@ static int isp_v4_1_0_hw_init(struct amdgpu_isp *isp)
 	isp->isp_cell[0].platform_data = isp->isp_pdata;
 	isp->isp_cell[0].pdata_size = sizeof(struct isp_platform_data);
 
-	isp->isp_i2c_res = kcalloc(1, sizeof(struct resource),
-				   GFP_KERNEL);
+	/* initialize isp i2c platform data */
+	isp->isp_i2c_res = kcalloc(1, sizeof(struct resource), GFP_KERNEL);
 	if (!isp->isp_i2c_res) {
 		r = -ENOMEM;
-		DRM_ERROR("%s: isp mfd res alloc failed\n", __func__);
+		drm_err(&adev->ddev,
+			"%s: isp mfd res alloc failed\n", __func__);
 		goto failure;
 	}
 
@@ -129,9 +126,31 @@ static int isp_v4_1_0_hw_init(struct amdgpu_isp *isp)
 	isp->isp_cell[1].platform_data = isp->isp_pdata;
 	isp->isp_cell[1].pdata_size = sizeof(struct isp_platform_data);
 
-	r = mfd_add_hotplug_devices(isp->parent, isp->isp_cell, 2);
+	/* initialize isp gpiochip platform data */
+	isp->isp_gpio_res = kcalloc(1, sizeof(struct resource), GFP_KERNEL);
+	if (!isp->isp_gpio_res) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev,
+			"%s: isp gpio res alloc failed\n", __func__);
+		goto failure;
+	}
+
+	isp->isp_gpio_res[0].name = "isp_gpio_reg";
+	isp->isp_gpio_res[0].flags = IORESOURCE_MEM;
+	isp->isp_gpio_res[0].start = isp_base + ISP410_GPIO_SENSOR_OFFSET;
+	isp->isp_gpio_res[0].end = isp_base + ISP410_GPIO_SENSOR_OFFSET +
+				   ISP410_GPIO_SENSOR_SIZE;
+
+	isp->isp_cell[2].name = "amdisp-pinctrl";
+	isp->isp_cell[2].num_resources = 1;
+	isp->isp_cell[2].resources = &isp->isp_gpio_res[0];
+	isp->isp_cell[2].platform_data = isp->isp_pdata;
+	isp->isp_cell[2].pdata_size = sizeof(struct isp_platform_data);
+
+	r = mfd_add_hotplug_devices(isp->parent, isp->isp_cell, 3);
 	if (r) {
-		DRM_ERROR("%s: add mfd hotplug device failed\n", __func__);
+		drm_err(&adev->ddev,
+			"%s: add mfd hotplug device failed\n", __func__);
 		goto failure;
 	}
 
@@ -143,6 +162,7 @@ failure:
 	kfree(isp->isp_res);
 	kfree(isp->isp_cell);
 	kfree(isp->isp_i2c_res);
+	kfree(isp->isp_gpio_res);
 
 	return r;
 }
@@ -155,6 +175,7 @@ static int isp_v4_1_0_hw_fini(struct amdgpu_isp *isp)
 	kfree(isp->isp_cell);
 	kfree(isp->isp_pdata);
 	kfree(isp->isp_i2c_res);
+	kfree(isp->isp_gpio_res);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.h b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.h
index 7db24c0f1080..4d239198edd0 100644
--- a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_0.h
@@ -42,8 +42,8 @@
 #define ISP410_I2C0_OFFSET 0x66400
 #define ISP410_I2C0_SIZE 0x100
 
-#define ISP410_GPIO_SENSOR0_OFFSET 0x6613C
-#define ISP410_GPIO_SENSOR0_SIZE 0x4
+#define ISP410_GPIO_SENSOR_OFFSET 0x6613C
+#define ISP410_GPIO_SENSOR_SIZE 0x54
 
 void isp_v4_1_0_set_isp_funcs(struct amdgpu_isp *isp);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c
index b56f27295468..4258d3e0b706 100644
--- a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c
+++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c
@@ -25,9 +25,18 @@
  *
  */
 
+#include <linux/gpio/machine.h>
 #include "amdgpu.h"
 #include "isp_v4_1_1.h"
 
+MODULE_FIRMWARE("amdgpu/isp_4_1_1.bin");
+
+#define ISP_PERFORMANCE_STATE_LOW 0
+#define ISP_PERFORMANCE_STATE_HIGH 1
+
+#define ISP_HIGH_PERFORMANC_XCLK 788
+#define ISP_HIGH_PERFORMANC_ICLK 788
+
 static const unsigned int isp_4_1_1_int_srcid[MAX_ISP411_INT_SRC] = {
 	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT9,
 	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT10,
@@ -39,41 +48,205 @@ static const unsigned int isp_4_1_1_int_srcid[MAX_ISP411_INT_SRC] = {
 	ISP_4_1__SRCID__ISP_RINGBUFFER_WPT16
 };
 
+static struct gpiod_lookup_table isp_gpio_table = {
+	.dev_id = "amd_isp_capture",
+	.table = {
+		GPIO_LOOKUP("AMDI0030:00", 85, "enable_isp", GPIO_ACTIVE_HIGH),
+		{ }
+	},
+};
+
+static struct gpiod_lookup_table isp_sensor_gpio_table = {
+	.dev_id = "i2c-ov05c10",
+	.table = {
+		GPIO_LOOKUP("amdisp-pinctrl", 0, "enable", GPIO_ACTIVE_HIGH),
+		{ }
+	},
+};
+
+static int isp_poweroff(struct generic_pm_domain *genpd)
+{
+	struct amdgpu_isp *isp = container_of(genpd, struct amdgpu_isp, ispgpd);
+	struct amdgpu_device *adev = isp->adev;
+
+	return amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ISP, true, 0);
+}
+
+static int isp_poweron(struct generic_pm_domain *genpd)
+{
+	struct amdgpu_isp *isp = container_of(genpd, struct amdgpu_isp, ispgpd);
+	struct amdgpu_device *adev = isp->adev;
+
+	return amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ISP, false, 0);
+}
+
+static int isp_set_performance_state(struct generic_pm_domain *genpd,
+				     unsigned int state)
+{
+	struct amdgpu_isp *isp = container_of(genpd, struct amdgpu_isp, ispgpd);
+	struct amdgpu_device *adev = isp->adev;
+	u32 iclk, xclk;
+	int ret;
+
+	switch (state) {
+	case ISP_PERFORMANCE_STATE_HIGH:
+		xclk = ISP_HIGH_PERFORMANC_XCLK;
+		iclk = ISP_HIGH_PERFORMANC_ICLK;
+		break;
+	case ISP_PERFORMANCE_STATE_LOW:
+		/* isp runs at default lowest clock-rate on power-on, do nothing */
+		return 0;
+	default:
+		return -EINVAL;
+	}
+
+	ret = amdgpu_dpm_set_soft_freq_range(adev, PP_ISPXCLK, xclk, 0);
+	if (ret) {
+		drm_err(&adev->ddev, "failed to set xclk %u to %u: %d\n",
+			xclk, state, ret);
+		return ret;
+	}
+
+	ret = amdgpu_dpm_set_soft_freq_range(adev, PP_ISPICLK, iclk, 0);
+	if (ret) {
+		drm_err(&adev->ddev, "failed to set iclk %u to %u: %d\n",
+			iclk, state, ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int isp_genpd_add_device(struct device *dev, void *data)
+{
+	struct generic_pm_domain *gpd = data;
+	struct platform_device *pdev = container_of(dev, struct platform_device, dev);
+	struct amdgpu_isp *isp = container_of(gpd, struct amdgpu_isp, ispgpd);
+	struct amdgpu_device *adev = isp->adev;
+	int ret;
+
+	if (!pdev)
+		return -EINVAL;
+
+	if (!dev->type->name) {
+		drm_dbg(&adev->ddev, "Invalid device type to add\n");
+		goto exit;
+	}
+
+	if (strcmp(dev->type->name, "mfd_device")) {
+		drm_dbg(&adev->ddev, "Invalid isp mfd device %s to add\n", pdev->mfd_cell->name);
+		goto exit;
+	}
+
+	ret = pm_genpd_add_device(gpd, dev);
+	if (ret) {
+		drm_err(&adev->ddev, "Failed to add dev %s to genpd %d\n",
+			pdev->mfd_cell->name, ret);
+		return -ENODEV;
+	}
+
+exit:
+	/* Continue to add */
+	return 0;
+}
+
+static int isp_genpd_remove_device(struct device *dev, void *data)
+{
+	struct generic_pm_domain *gpd = data;
+	struct platform_device *pdev = container_of(dev, struct platform_device, dev);
+	struct amdgpu_isp *isp = container_of(gpd, struct amdgpu_isp, ispgpd);
+	struct amdgpu_device *adev = isp->adev;
+	int ret;
+
+	if (!pdev)
+		return -EINVAL;
+
+	if (!dev->type->name) {
+		drm_dbg(&adev->ddev, "Invalid device type to remove\n");
+		goto exit;
+	}
+
+	if (strcmp(dev->type->name, "mfd_device")) {
+		drm_dbg(&adev->ddev, "Invalid isp mfd device %s to remove\n",
+			pdev->mfd_cell->name);
+		goto exit;
+	}
+
+	ret = pm_genpd_remove_device(dev);
+	if (ret) {
+		drm_err(&adev->ddev, "Failed to remove dev from genpd %d\n", ret);
+		return -ENODEV;
+	}
+
+exit:
+	/* Continue to remove */
+	return 0;
+}
+
 static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp)
 {
+	const struct software_node *amd_camera_node, *isp4_node;
 	struct amdgpu_device *adev = isp->adev;
+	struct acpi_device *acpi_dev;
 	int idx, int_idx, num_res, r;
 	u64 isp_base;
 
 	if (adev->rmmio_size == 0 || adev->rmmio_size < 0x5289)
 		return -EINVAL;
 
+	r = amdgpu_acpi_get_isp4_dev(&acpi_dev);
+	if (r) {
+		drm_dbg(&adev->ddev, "Invalid isp platform detected (%d)", r);
+		/* allow GPU init to progress */
+		return 0;
+	}
+
+	/* add GPIO resources required for OMNI5C10 sensor */
+	if (!strcmp("OMNI5C10", acpi_device_hid(acpi_dev))) {
+		gpiod_add_lookup_table(&isp_gpio_table);
+		gpiod_add_lookup_table(&isp_sensor_gpio_table);
+	}
+
 	isp_base = adev->rmmio_base;
 
-	isp->isp_cell = kcalloc(2, sizeof(struct mfd_cell), GFP_KERNEL);
+	isp->ispgpd.name = "ISP_v_4_1_1";
+	isp->ispgpd.power_off = isp_poweroff;
+	isp->ispgpd.power_on = isp_poweron;
+	isp->ispgpd.set_performance_state = isp_set_performance_state;
+
+	r = pm_genpd_init(&isp->ispgpd, NULL, true);
+	if (r) {
+		drm_err(&adev->ddev, "failed to initialize genpd (%d)\n", r);
+		return -EINVAL;
+	}
+
+	isp->isp_cell = kcalloc(3, sizeof(struct mfd_cell), GFP_KERNEL);
 	if (!isp->isp_cell) {
 		r = -ENOMEM;
-		DRM_ERROR("%s: isp mfd cell alloc failed\n", __func__);
+		drm_err(&adev->ddev, "isp mfd cell alloc failed (%d)\n", r);
 		goto failure;
 	}
 
-	num_res = MAX_ISP411_MEM_RES + MAX_ISP411_SENSOR_RES + MAX_ISP411_INT_SRC;
+	num_res = MAX_ISP411_MEM_RES + MAX_ISP411_INT_SRC;
 
 	isp->isp_res = kcalloc(num_res, sizeof(struct resource),
 			       GFP_KERNEL);
 	if (!isp->isp_res) {
 		r = -ENOMEM;
-		DRM_ERROR("%s: isp mfd res alloc failed\n", __func__);
+		drm_err(&adev->ddev, "isp mfd resource alloc failed (%d)\n", r);
 		goto failure;
 	}
 
 	isp->isp_pdata = kzalloc(sizeof(*isp->isp_pdata), GFP_KERNEL);
 	if (!isp->isp_pdata) {
 		r = -ENOMEM;
-		DRM_ERROR("%s: isp platform data alloc failed\n", __func__);
+		drm_err(&adev->ddev, "isp platform data alloc failed (%d)\n", r);
 		goto failure;
 	}
 
+	amd_camera_node = (const struct software_node *)acpi_dev->driver_data;
+	isp4_node = software_node_find_by_name(amd_camera_node, "isp4");
+
 	/* initialize isp platform data */
 	isp->isp_pdata->adev = (void *)adev;
 	isp->isp_pdata->asic_type = adev->asic_type;
@@ -89,14 +262,7 @@ static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp)
 	isp->isp_res[1].start = isp_base + ISP411_PHY0_OFFSET;
 	isp->isp_res[1].end = isp_base + ISP411_PHY0_OFFSET + ISP411_PHY0_SIZE;
 
-	isp->isp_res[2].name = "isp_4_1_1_sensor0_reg";
-	isp->isp_res[2].flags = IORESOURCE_MEM;
-	isp->isp_res[2].start = isp_base + ISP411_GPIO_SENSOR0_OFFSET;
-	isp->isp_res[2].end = isp_base + ISP411_GPIO_SENSOR0_OFFSET +
-			      ISP411_GPIO_SENSOR0_SIZE;
-
-	for (idx = MAX_ISP411_MEM_RES + MAX_ISP411_SENSOR_RES, int_idx = 0;
-	     idx < num_res; idx++, int_idx++) {
+	for (idx = MAX_ISP411_MEM_RES, int_idx = 0; idx < num_res; idx++, int_idx++) {
 		isp->isp_res[idx].name = "isp_4_1_1_irq";
 		isp->isp_res[idx].flags = IORESOURCE_IRQ;
 		isp->isp_res[idx].start =
@@ -109,12 +275,14 @@ static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp)
 	isp->isp_cell[0].num_resources = num_res;
 	isp->isp_cell[0].resources = &isp->isp_res[0];
 	isp->isp_cell[0].platform_data = isp->isp_pdata;
+	isp->isp_cell[0].swnode = isp4_node;
 	isp->isp_cell[0].pdata_size = sizeof(struct isp_platform_data);
 
+	/* initialize isp i2c platform data */
 	isp->isp_i2c_res = kcalloc(1, sizeof(struct resource), GFP_KERNEL);
 	if (!isp->isp_i2c_res) {
 		r = -ENOMEM;
-		DRM_ERROR("%s: isp mfd res alloc failed\n", __func__);
+		drm_err(&adev->ddev, "isp mfd res alloc failed (%d)\n", r);
 		goto failure;
 	}
 
@@ -129,9 +297,43 @@ static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp)
 	isp->isp_cell[1].platform_data = isp->isp_pdata;
 	isp->isp_cell[1].pdata_size = sizeof(struct isp_platform_data);
 
+	/* initialize isp gpiochip platform data */
+	isp->isp_gpio_res = kcalloc(1, sizeof(struct resource), GFP_KERNEL);
+	if (!isp->isp_gpio_res) {
+		r = -ENOMEM;
+		drm_err(&adev->ddev, "isp gpio resource alloc failed (%d)\n", r);
+		goto failure;
+	}
+
+	isp->isp_gpio_res[0].name = "isp_gpio_reg";
+	isp->isp_gpio_res[0].flags = IORESOURCE_MEM;
+	isp->isp_gpio_res[0].start = isp_base + ISP411_GPIO_SENSOR_OFFSET;
+	isp->isp_gpio_res[0].end = isp_base + ISP411_GPIO_SENSOR_OFFSET +
+				   ISP411_GPIO_SENSOR_SIZE;
+
+	isp->isp_cell[2].name = "amdisp-pinctrl";
+	isp->isp_cell[2].num_resources = 1;
+	isp->isp_cell[2].resources = &isp->isp_gpio_res[0];
+	isp->isp_cell[2].platform_data = isp->isp_pdata;
+	isp->isp_cell[2].pdata_size = sizeof(struct isp_platform_data);
+
+	/* add only amd_isp_capture and amd_isp_i2c_designware to genpd */
 	r = mfd_add_hotplug_devices(isp->parent, isp->isp_cell, 2);
 	if (r) {
-		DRM_ERROR("%s: add mfd hotplug device failed\n", __func__);
+		drm_err(&adev->ddev, "add mfd hotplug device failed (%d)\n", r);
+		goto failure;
+	}
+
+	r = device_for_each_child(isp->parent, &isp->ispgpd,
+				  isp_genpd_add_device);
+	if (r) {
+		drm_err(&adev->ddev, "failed to add devices to genpd (%d)\n", r);
+		goto failure;
+	}
+
+	r = mfd_add_hotplug_devices(isp->parent, &isp->isp_cell[2], 1);
+	if (r) {
+		drm_err(&adev->ddev, "add pinctl hotplug device failed (%d)\n", r);
 		goto failure;
 	}
 
@@ -143,18 +345,23 @@ failure:
 	kfree(isp->isp_res);
 	kfree(isp->isp_cell);
 	kfree(isp->isp_i2c_res);
+	kfree(isp->isp_gpio_res);
 
 	return r;
 }
 
 static int isp_v4_1_1_hw_fini(struct amdgpu_isp *isp)
 {
+	device_for_each_child(isp->parent, NULL,
+			      isp_genpd_remove_device);
+
 	mfd_remove_devices(isp->parent);
 
 	kfree(isp->isp_res);
 	kfree(isp->isp_cell);
 	kfree(isp->isp_pdata);
 	kfree(isp->isp_i2c_res);
+	kfree(isp->isp_gpio_res);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.h b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.h
index 40887ddeb08c..fe45d70d87f1 100644
--- a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.h
+++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.h
@@ -33,7 +33,6 @@
 #include "ivsrcid/isp/irqsrcs_isp_4_1.h"
 
 #define MAX_ISP411_MEM_RES 2
-#define MAX_ISP411_SENSOR_RES 1
 #define MAX_ISP411_INT_SRC 8
 
 #define ISP411_PHY0_OFFSET 0x66700
@@ -42,8 +41,8 @@
 #define ISP411_I2C0_OFFSET 0x66400
 #define ISP411_I2C0_SIZE 0x100
 
-#define ISP411_GPIO_SENSOR0_OFFSET 0x6613C
-#define ISP411_GPIO_SENSOR0_SIZE 0x4
+#define ISP411_GPIO_SENSOR_OFFSET 0x6613C
+#define ISP411_GPIO_SENSOR_SIZE 0x54
 
 void isp_v4_1_1_set_isp_funcs(struct amdgpu_isp *isp);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c
index 03b8b7cd5229..b5bb7f4d607c 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c
@@ -557,7 +557,7 @@ static const struct amdgpu_ring_funcs jpeg_v1_0_decode_ring_vm_funcs = {
 	.nop = PACKET0(0x81ff, 0),
 	.support_64bit_ptrs = false,
 	.no_user_fence = true,
-	.extra_dw = 64,
+	.extra_bytes = 256,
 	.get_rptr = jpeg_v1_0_decode_ring_get_rptr,
 	.get_wptr = jpeg_v1_0_decode_ring_get_wptr,
 	.set_wptr = jpeg_v1_0_decode_ring_set_wptr,
@@ -604,15 +604,15 @@ static void jpeg_v1_0_set_irq_funcs(struct amdgpu_device *adev)
 static void jpeg_v1_0_ring_begin_use(struct amdgpu_ring *ring)
 {
 	struct	amdgpu_device *adev = ring->adev;
-	bool	set_clocks = !cancel_delayed_work_sync(&adev->vcn.idle_work);
+	bool	set_clocks = !cancel_delayed_work_sync(&adev->vcn.inst[0].idle_work);
 	int		cnt = 0;
 
-	mutex_lock(&adev->vcn.vcn1_jpeg1_workaround);
+	mutex_lock(&adev->vcn.inst[0].vcn1_jpeg1_workaround);
 
 	if (amdgpu_fence_wait_empty(&adev->vcn.inst->ring_dec))
 		DRM_ERROR("JPEG dec: vcn dec ring may not be empty\n");
 
-	for (cnt = 0; cnt < adev->vcn.num_enc_rings; cnt++) {
+	for (cnt = 0; cnt < adev->vcn.inst[0].num_enc_rings; cnt++) {
 		if (amdgpu_fence_wait_empty(&adev->vcn.inst->ring_enc[cnt]))
 			DRM_ERROR("JPEG dec: vcn enc ring[%d] may not be empty\n", cnt);
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c
index 7c9251c03815..27c76bd424cf 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c
@@ -23,7 +23,6 @@
 
 #include "amdgpu.h"
 #include "amdgpu_jpeg.h"
-#include "amdgpu_cs.h"
 #include "amdgpu_pm.h"
 #include "soc15.h"
 #include "soc15d.h"
@@ -33,6 +32,22 @@
 #include "vcn/vcn_2_0_0_sh_mask.h"
 #include "ivsrcid/vcn/irqsrcs_vcn_2_0.h"
 
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_2_0[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_UV_PITCH),
+};
+
 static void jpeg_v2_0_set_dec_ring_funcs(struct amdgpu_device *adev);
 static void jpeg_v2_0_set_irq_funcs(struct amdgpu_device *adev);
 static int jpeg_v2_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
@@ -98,7 +113,17 @@ static int jpeg_v2_0_sw_init(struct amdgpu_ip_block *ip_block)
 	adev->jpeg.internal.jpeg_pitch[0] = mmUVD_JPEG_PITCH_INTERNAL_OFFSET;
 	adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_PITCH);
 
-	return 0;
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_2_0, ARRAY_SIZE(jpeg_reg_list_2_0));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
 }
 
 /**
@@ -117,6 +142,8 @@ static int jpeg_v2_0_sw_fini(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+
 	r = amdgpu_jpeg_sw_fini(adev);
 
 	return r;
@@ -655,9 +682,9 @@ void jpeg_v2_0_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count)
 	}
 }
 
-static bool jpeg_v2_0_is_idle(void *handle)
+static bool jpeg_v2_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 
 	return ((RREG32_SOC15(JPEG, 0, mmUVD_JRBC_STATUS) &
 		UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
@@ -682,7 +709,7 @@ static int jpeg_v2_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	bool enable = (state == AMD_CG_STATE_GATE);
 
 	if (enable) {
-		if (!jpeg_v2_0_is_idle(adev))
+		if (!jpeg_v2_0_is_idle(ip_block))
 			return -EBUSY;
 		jpeg_v2_0_enable_clock_gating(adev);
 	} else {
@@ -739,6 +766,22 @@ static int jpeg_v2_0_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int jpeg_v2_0_ring_reset(struct amdgpu_ring *ring,
+				unsigned int vmid,
+				struct amdgpu_fence *timedout_fence)
+{
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = jpeg_v2_0_stop(ring->adev);
+	if (r)
+		return r;
+	r = jpeg_v2_0_start(ring->adev);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
 static const struct amd_ip_funcs jpeg_v2_0_ip_funcs = {
 	.name = "jpeg_v2_0",
 	.early_init = jpeg_v2_0_early_init,
@@ -752,6 +795,8 @@ static const struct amd_ip_funcs jpeg_v2_0_ip_funcs = {
 	.wait_for_idle = jpeg_v2_0_wait_for_idle,
 	.set_clockgating_state = jpeg_v2_0_set_clockgating_state,
 	.set_powergating_state = jpeg_v2_0_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
 };
 
 static const struct amdgpu_ring_funcs jpeg_v2_0_dec_ring_vm_funcs = {
@@ -760,7 +805,7 @@ static const struct amdgpu_ring_funcs jpeg_v2_0_dec_ring_vm_funcs = {
 	.get_rptr = jpeg_v2_0_dec_ring_get_rptr,
 	.get_wptr = jpeg_v2_0_dec_ring_get_wptr,
 	.set_wptr = jpeg_v2_0_dec_ring_set_wptr,
-	.parse_cs = jpeg_v2_dec_ring_parse_cs,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
 	.emit_frame_size =
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
 		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
@@ -782,6 +827,7 @@ static const struct amdgpu_ring_funcs jpeg_v2_0_dec_ring_vm_funcs = {
 	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
 	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v2_0_ring_reset,
 };
 
 static void jpeg_v2_0_set_dec_ring_funcs(struct amdgpu_device *adev)
@@ -807,58 +853,3 @@ const struct amdgpu_ip_block_version jpeg_v2_0_ip_block = {
 		.rev = 0,
 		.funcs = &jpeg_v2_0_ip_funcs,
 };
-
-/**
- * jpeg_v2_dec_ring_parse_cs - command submission parser
- *
- * @parser: Command submission parser context
- * @job: the job to parse
- * @ib: the IB to parse
- *
- * Parse the command stream, return -EINVAL for invalid packet,
- * 0 otherwise
- */
-int jpeg_v2_dec_ring_parse_cs(struct amdgpu_cs_parser *parser,
-			      struct amdgpu_job *job,
-			      struct amdgpu_ib *ib)
-{
-	u32 i, reg, res, cond, type;
-	struct amdgpu_device *adev = parser->adev;
-
-	for (i = 0; i < ib->length_dw ; i += 2) {
-		reg  = CP_PACKETJ_GET_REG(ib->ptr[i]);
-		res  = CP_PACKETJ_GET_RES(ib->ptr[i]);
-		cond = CP_PACKETJ_GET_COND(ib->ptr[i]);
-		type = CP_PACKETJ_GET_TYPE(ib->ptr[i]);
-
-		if (res) /* only support 0 at the moment */
-			return -EINVAL;
-
-		switch (type) {
-		case PACKETJ_TYPE0:
-			if (cond != PACKETJ_CONDITION_CHECK0 || reg < JPEG_REG_RANGE_START ||
-			    reg > JPEG_REG_RANGE_END) {
-				dev_err(adev->dev, "Invalid packet [0x%08x]!\n", ib->ptr[i]);
-				return -EINVAL;
-			}
-			break;
-		case PACKETJ_TYPE3:
-			if (cond != PACKETJ_CONDITION_CHECK3 || reg < JPEG_REG_RANGE_START ||
-			    reg > JPEG_REG_RANGE_END) {
-				dev_err(adev->dev, "Invalid packet [0x%08x]!\n", ib->ptr[i]);
-				return -EINVAL;
-			}
-			break;
-		case PACKETJ_TYPE6:
-			if (ib->ptr[i] == CP_PACKETJ_NOP)
-				continue;
-			dev_err(adev->dev, "Invalid packet [0x%08x]!\n", ib->ptr[i]);
-			return -EINVAL;
-		default:
-			dev_err(adev->dev, "Unknown packet type %d !\n", type);
-			return -EINVAL;
-		}
-	}
-
-	return 0;
-}
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.h
index 63fadda7a673..654e43e83e2c 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.h
@@ -45,9 +45,6 @@
 
 #define JRBC_DEC_EXTERNAL_REG_WRITE_ADDR				0x18000
 
-#define JPEG_REG_RANGE_START						0x4000
-#define JPEG_REG_RANGE_END						0x41c2
-
 void jpeg_v2_0_dec_ring_insert_start(struct amdgpu_ring *ring);
 void jpeg_v2_0_dec_ring_insert_end(struct amdgpu_ring *ring);
 void jpeg_v2_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
@@ -60,9 +57,6 @@ void jpeg_v2_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
 				unsigned vmid, uint64_t pd_addr);
 void jpeg_v2_0_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val);
 void jpeg_v2_0_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count);
-int jpeg_v2_dec_ring_parse_cs(struct amdgpu_cs_parser *parser,
-			      struct amdgpu_job *job,
-			      struct amdgpu_ib *ib);
 
 extern const struct amdgpu_ip_block_version jpeg_v2_0_ip_block;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
index 11f6af2646e7..20983f126b49 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
@@ -36,6 +36,22 @@
 
 #define JPEG25_MAX_HW_INSTANCES_ARCTURUS			2
 
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_2_5[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_UV_PITCH),
+};
+
 static void jpeg_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev);
 static void jpeg_v2_5_set_irq_funcs(struct amdgpu_device *adev);
 static int jpeg_v2_5_set_powergating_state(struct amdgpu_ip_block *ip_block,
@@ -147,7 +163,17 @@ static int jpeg_v2_5_sw_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
-	return 0;
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_2_5, ARRAY_SIZE(jpeg_reg_list_2_5));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
 }
 
 /**
@@ -166,6 +192,8 @@ static int jpeg_v2_5_sw_fini(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+
 	r = amdgpu_jpeg_sw_fini(adev);
 
 	return r;
@@ -310,6 +338,44 @@ static void jpeg_v2_5_enable_clock_gating(struct amdgpu_device *adev, int inst)
 	WREG32_SOC15(JPEG, inst, mmJPEG_CGC_GATE, data);
 }
 
+static void jpeg_v2_5_start_inst(struct amdgpu_device *adev, int i)
+{
+	struct amdgpu_ring *ring = adev->jpeg.inst[i].ring_dec;
+	/* disable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JPEG_POWER_STATUS), 0,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	/* JPEG disable CGC */
+	jpeg_v2_5_disable_clock_gating(adev, i);
+
+	/* MJPEG global tiling registers */
+	WREG32_SOC15(JPEG, i, mmJPEG_DEC_GFX8_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(JPEG, i, mmJPEG_DEC_GFX10_ADDR_CONFIG,
+		adev->gfx.config.gb_addr_config);
+
+	/* enable JMI channel */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JMI_CNTL), 0,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	/* enable System Interrupt for JRBC */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmJPEG_SYS_INT_EN),
+		JPEG_SYS_INT_EN__DJRBC_MASK,
+		~JPEG_SYS_INT_EN__DJRBC_MASK);
+
+	WREG32_SOC15(JPEG, i, mmUVD_LMI_JRBC_RB_VMID, 0);
+	WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+	WREG32_SOC15(JPEG, i, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+		lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, i, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+		upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_RPTR, 0);
+	WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_WPTR, 0);
+	WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_CNTL, 0x00000002L);
+	WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_WPTR);
+}
+
 /**
  * jpeg_v2_5_start - start JPEG block
  *
@@ -319,52 +385,33 @@ static void jpeg_v2_5_enable_clock_gating(struct amdgpu_device *adev, int inst)
  */
 static int jpeg_v2_5_start(struct amdgpu_device *adev)
 {
-	struct amdgpu_ring *ring;
 	int i;
 
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
 		if (adev->jpeg.harvest_config & (1 << i))
 			continue;
+		jpeg_v2_5_start_inst(adev, i);
 
-		ring = adev->jpeg.inst[i].ring_dec;
-		/* disable anti hang mechanism */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JPEG_POWER_STATUS), 0,
-			~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
-
-		/* JPEG disable CGC */
-		jpeg_v2_5_disable_clock_gating(adev, i);
-
-		/* MJPEG global tiling registers */
-		WREG32_SOC15(JPEG, i, mmJPEG_DEC_GFX8_ADDR_CONFIG,
-			adev->gfx.config.gb_addr_config);
-		WREG32_SOC15(JPEG, i, mmJPEG_DEC_GFX10_ADDR_CONFIG,
-			adev->gfx.config.gb_addr_config);
-
-		/* enable JMI channel */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JMI_CNTL), 0,
-			~UVD_JMI_CNTL__SOFT_RESET_MASK);
-
-		/* enable System Interrupt for JRBC */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmJPEG_SYS_INT_EN),
-			JPEG_SYS_INT_EN__DJRBC_MASK,
-			~JPEG_SYS_INT_EN__DJRBC_MASK);
-
-		WREG32_SOC15(JPEG, i, mmUVD_LMI_JRBC_RB_VMID, 0);
-		WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
-		WREG32_SOC15(JPEG, i, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
-			lower_32_bits(ring->gpu_addr));
-		WREG32_SOC15(JPEG, i, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
-			upper_32_bits(ring->gpu_addr));
-		WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_RPTR, 0);
-		WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_WPTR, 0);
-		WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_CNTL, 0x00000002L);
-		WREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_SIZE, ring->ring_size / 4);
-		ring->wptr = RREG32_SOC15(JPEG, i, mmUVD_JRBC_RB_WPTR);
 	}
 
 	return 0;
 }
 
+static void jpeg_v2_5_stop_inst(struct amdgpu_device *adev, int i)
+{
+	/* reset JMI */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JMI_CNTL),
+		UVD_JMI_CNTL__SOFT_RESET_MASK,
+		~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	jpeg_v2_5_enable_clock_gating(adev, i);
+
+	/* enable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JPEG_POWER_STATUS),
+		UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
+		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+}
+
 /**
  * jpeg_v2_5_stop - stop JPEG block
  *
@@ -379,18 +426,7 @@ static int jpeg_v2_5_stop(struct amdgpu_device *adev)
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
 		if (adev->jpeg.harvest_config & (1 << i))
 			continue;
-
-		/* reset JMI */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JMI_CNTL),
-			UVD_JMI_CNTL__SOFT_RESET_MASK,
-			~UVD_JMI_CNTL__SOFT_RESET_MASK);
-
-		jpeg_v2_5_enable_clock_gating(adev, i);
-
-		/* enable anti hang mechanism */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, i, mmUVD_JPEG_POWER_STATUS),
-			UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
-			~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+		jpeg_v2_5_stop_inst(adev, i);
 	}
 
 	return 0;
@@ -482,9 +518,9 @@ static void jpeg_v2_6_dec_ring_insert_end(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, (1 << (ring->me * 2 + 14)));
 }
 
-static bool jpeg_v2_5_is_idle(void *handle)
+static bool jpeg_v2_5_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int i, ret = 1;
 
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
@@ -530,7 +566,7 @@ static int jpeg_v2_5_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 			continue;
 
 		if (enable) {
-			if (!jpeg_v2_5_is_idle(adev))
+			if (!jpeg_v2_5_is_idle(ip_block))
 				return -EBUSY;
 			jpeg_v2_5_enable_clock_gating(adev, i);
 		} else {
@@ -610,6 +646,16 @@ static int jpeg_v2_5_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int jpeg_v2_5_ring_reset(struct amdgpu_ring *ring,
+				unsigned int vmid,
+				struct amdgpu_fence *timedout_fence)
+{
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	jpeg_v2_5_stop_inst(ring->adev, ring->me);
+	jpeg_v2_5_start_inst(ring->adev, ring->me);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
 static const struct amd_ip_funcs jpeg_v2_5_ip_funcs = {
 	.name = "jpeg_v2_5",
 	.early_init = jpeg_v2_5_early_init,
@@ -623,6 +669,8 @@ static const struct amd_ip_funcs jpeg_v2_5_ip_funcs = {
 	.wait_for_idle = jpeg_v2_5_wait_for_idle,
 	.set_clockgating_state = jpeg_v2_5_set_clockgating_state,
 	.set_powergating_state = jpeg_v2_5_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
 };
 
 static const struct amd_ip_funcs jpeg_v2_6_ip_funcs = {
@@ -638,6 +686,8 @@ static const struct amd_ip_funcs jpeg_v2_6_ip_funcs = {
 	.wait_for_idle = jpeg_v2_5_wait_for_idle,
 	.set_clockgating_state = jpeg_v2_5_set_clockgating_state,
 	.set_powergating_state = jpeg_v2_5_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
 };
 
 static const struct amdgpu_ring_funcs jpeg_v2_5_dec_ring_vm_funcs = {
@@ -646,7 +696,7 @@ static const struct amdgpu_ring_funcs jpeg_v2_5_dec_ring_vm_funcs = {
 	.get_rptr = jpeg_v2_5_dec_ring_get_rptr,
 	.get_wptr = jpeg_v2_5_dec_ring_get_wptr,
 	.set_wptr = jpeg_v2_5_dec_ring_set_wptr,
-	.parse_cs = jpeg_v2_dec_ring_parse_cs,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
 	.emit_frame_size =
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
 		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
@@ -668,6 +718,7 @@ static const struct amdgpu_ring_funcs jpeg_v2_5_dec_ring_vm_funcs = {
 	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
 	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v2_5_ring_reset,
 };
 
 static const struct amdgpu_ring_funcs jpeg_v2_6_dec_ring_vm_funcs = {
@@ -676,7 +727,7 @@ static const struct amdgpu_ring_funcs jpeg_v2_6_dec_ring_vm_funcs = {
 	.get_rptr = jpeg_v2_5_dec_ring_get_rptr,
 	.get_wptr = jpeg_v2_5_dec_ring_get_wptr,
 	.set_wptr = jpeg_v2_5_dec_ring_set_wptr,
-	.parse_cs = jpeg_v2_dec_ring_parse_cs,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
 	.emit_frame_size =
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
 		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
@@ -698,6 +749,7 @@ static const struct amdgpu_ring_funcs jpeg_v2_6_dec_ring_vm_funcs = {
 	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
 	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v2_5_ring_reset,
 };
 
 static void jpeg_v2_5_set_dec_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c
index 4eca65ea9053..d1a011c40ba2 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c
@@ -34,6 +34,22 @@
 
 #define mmUVD_JPEG_PITCH_INTERNAL_OFFSET	0x401f
 
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_3_0[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_UV_PITCH),
+};
+
 static void jpeg_v3_0_set_dec_ring_funcs(struct amdgpu_device *adev);
 static void jpeg_v3_0_set_irq_funcs(struct amdgpu_device *adev);
 static int jpeg_v3_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
@@ -112,7 +128,17 @@ static int jpeg_v3_0_sw_init(struct amdgpu_ip_block *ip_block)
 	adev->jpeg.internal.jpeg_pitch[0] = mmUVD_JPEG_PITCH_INTERNAL_OFFSET;
 	adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_PITCH);
 
-	return 0;
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_3_0, ARRAY_SIZE(jpeg_reg_list_3_0));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
 }
 
 /**
@@ -131,6 +157,8 @@ static int jpeg_v3_0_sw_fini(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+
 	r = amdgpu_jpeg_sw_fini(adev);
 
 	return r;
@@ -445,9 +473,9 @@ static void jpeg_v3_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
 	}
 }
 
-static bool jpeg_v3_0_is_idle(void *handle)
+static bool jpeg_v3_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret = 1;
 
 	ret &= (((RREG32_SOC15(JPEG, 0, mmUVD_JRBC_STATUS) &
@@ -473,7 +501,7 @@ static int jpeg_v3_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	bool enable = state == AMD_CG_STATE_GATE;
 
 	if (enable) {
-		if (!jpeg_v3_0_is_idle(adev))
+		if (!jpeg_v3_0_is_idle(ip_block))
 			return -EBUSY;
 		jpeg_v3_0_enable_clock_gating(adev);
 	} else {
@@ -530,6 +558,22 @@ static int jpeg_v3_0_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int jpeg_v3_0_ring_reset(struct amdgpu_ring *ring,
+				unsigned int vmid,
+				struct amdgpu_fence *timedout_fence)
+{
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = jpeg_v3_0_stop(ring->adev);
+	if (r)
+		return r;
+	r = jpeg_v3_0_start(ring->adev);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
 static const struct amd_ip_funcs jpeg_v3_0_ip_funcs = {
 	.name = "jpeg_v3_0",
 	.early_init = jpeg_v3_0_early_init,
@@ -543,6 +587,8 @@ static const struct amd_ip_funcs jpeg_v3_0_ip_funcs = {
 	.wait_for_idle = jpeg_v3_0_wait_for_idle,
 	.set_clockgating_state = jpeg_v3_0_set_clockgating_state,
 	.set_powergating_state = jpeg_v3_0_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
 };
 
 static const struct amdgpu_ring_funcs jpeg_v3_0_dec_ring_vm_funcs = {
@@ -551,7 +597,7 @@ static const struct amdgpu_ring_funcs jpeg_v3_0_dec_ring_vm_funcs = {
 	.get_rptr = jpeg_v3_0_dec_ring_get_rptr,
 	.get_wptr = jpeg_v3_0_dec_ring_get_wptr,
 	.set_wptr = jpeg_v3_0_dec_ring_set_wptr,
-	.parse_cs = jpeg_v2_dec_ring_parse_cs,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
 	.emit_frame_size =
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
 		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
@@ -573,6 +619,7 @@ static const struct amdgpu_ring_funcs jpeg_v3_0_dec_ring_vm_funcs = {
 	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
 	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v3_0_ring_reset,
 };
 
 static void jpeg_v3_0_set_dec_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
index 0aef1f64afd0..33db2c1ae6cc 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
@@ -36,13 +36,28 @@
 
 #define regUVD_JPEG_PITCH_INTERNAL_OFFSET                  0x401f
 
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_4_0[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH),
+};
+
 static int jpeg_v4_0_start_sriov(struct amdgpu_device *adev);
 static void jpeg_v4_0_set_dec_ring_funcs(struct amdgpu_device *adev);
 static void jpeg_v4_0_set_irq_funcs(struct amdgpu_device *adev);
 static int jpeg_v4_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
 				enum amd_powergating_state state);
 static void jpeg_v4_0_set_ras_funcs(struct amdgpu_device *adev);
-
 static void jpeg_v4_0_dec_ring_set_wptr(struct amdgpu_ring *ring);
 
 /**
@@ -123,14 +138,18 @@ static int jpeg_v4_0_sw_init(struct amdgpu_ip_block *ip_block)
 	r = amdgpu_jpeg_ras_sw_init(adev);
 	if (r)
 		return r;
-	/* TODO: Add queue reset mask when FW fully supports it */
-	adev->jpeg.supported_reset =
-		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
-	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_4_0, ARRAY_SIZE(jpeg_reg_list_4_0));
 	if (r)
 		return r;
 
-	return 0;
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
 }
 
 /**
@@ -614,9 +633,9 @@ static void jpeg_v4_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
 	}
 }
 
-static bool jpeg_v4_0_is_idle(void *handle)
+static bool jpeg_v4_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret = 1;
 
 	ret &= (((RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS) &
@@ -642,7 +661,7 @@ static int jpeg_v4_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 	bool enable = state == AMD_CG_STATE_GATE;
 
 	if (enable) {
-		if (!jpeg_v4_0_is_idle(adev))
+		if (!jpeg_v4_0_is_idle(ip_block))
 			return -EBUSY;
 		jpeg_v4_0_enable_clock_gating(adev);
 	} else {
@@ -704,6 +723,22 @@ static int jpeg_v4_0_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int jpeg_v4_0_ring_reset(struct amdgpu_ring *ring,
+				unsigned int vmid,
+				struct amdgpu_fence *timedout_fence)
+{
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = jpeg_v4_0_stop(ring->adev);
+	if (r)
+		return r;
+	r = jpeg_v4_0_start(ring->adev);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
 static const struct amd_ip_funcs jpeg_v4_0_ip_funcs = {
 	.name = "jpeg_v4_0",
 	.early_init = jpeg_v4_0_early_init,
@@ -717,6 +752,8 @@ static const struct amd_ip_funcs jpeg_v4_0_ip_funcs = {
 	.wait_for_idle = jpeg_v4_0_wait_for_idle,
 	.set_clockgating_state = jpeg_v4_0_set_clockgating_state,
 	.set_powergating_state = jpeg_v4_0_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
 };
 
 static const struct amdgpu_ring_funcs jpeg_v4_0_dec_ring_vm_funcs = {
@@ -725,7 +762,7 @@ static const struct amdgpu_ring_funcs jpeg_v4_0_dec_ring_vm_funcs = {
 	.get_rptr = jpeg_v4_0_dec_ring_get_rptr,
 	.get_wptr = jpeg_v4_0_dec_ring_get_wptr,
 	.set_wptr = jpeg_v4_0_dec_ring_set_wptr,
-	.parse_cs = jpeg_v2_dec_ring_parse_cs,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
 	.emit_frame_size =
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
 		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
@@ -747,6 +784,7 @@ static const struct amdgpu_ring_funcs jpeg_v4_0_dec_ring_vm_funcs = {
 	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
 	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v4_0_ring_reset,
 };
 
 static void jpeg_v4_0_set_dec_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
index 88f9771c1686..aae7328973d1 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
@@ -59,10 +59,53 @@ static int amdgpu_ih_srcid_jpeg[] = {
 	VCN_4_0__SRCID__JPEG7_DECODE
 };
 
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_4_0_3[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_SYS_INT_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_STATUS),
+};
+
 static inline bool jpeg_v4_0_3_normalizn_reqd(struct amdgpu_device *adev)
 {
-	return amdgpu_sriov_vf(adev) ||
-	       (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4));
+	return (adev->jpeg.caps & AMDGPU_JPEG_CAPS(RRMT_ENABLED)) == 0;
+}
+
+static inline int jpeg_v4_0_3_core_reg_offset(u32 pipe)
+{
+	if (pipe)
+		return ((0x40 * pipe) - 0xc80);
+	else
+		return 0;
 }
 
 /**
@@ -106,6 +149,18 @@ static int jpeg_v4_0_3_sw_init(struct amdgpu_ip_block *ip_block)
 			return r;
 	}
 
+	/* JPEG DJPEG POISON EVENT */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_4_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
+	if (r)
+		return r;
+
+	/* JPEG EJPEG POISON EVENT */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_4_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
+	if (r)
+		return r;
+
 	r = amdgpu_jpeg_sw_init(adev);
 	if (r)
 		return r;
@@ -144,10 +199,8 @@ static int jpeg_v4_0_3_sw_init(struct amdgpu_ip_block *ip_block)
 			adev->jpeg.internal.jpeg_pitch[j] =
 				regUVD_JRBC0_UVD_JRBC_SCRATCH0_INTERNAL_OFFSET;
 			adev->jpeg.inst[i].external.jpeg_pitch[j] =
-				SOC15_REG_OFFSET1(
-					JPEG, jpeg_inst,
-					regUVD_JRBC0_UVD_JRBC_SCRATCH0,
-					(j ? (0x40 * j - 0xc80) : 0));
+				SOC15_REG_OFFSET1(JPEG, jpeg_inst, regUVD_JRBC0_UVD_JRBC_SCRATCH0,
+						  jpeg_v4_0_3_core_reg_offset(j));
 		}
 	}
 
@@ -159,13 +212,16 @@ static int jpeg_v4_0_3_sw_init(struct amdgpu_ip_block *ip_block)
 		}
 	}
 
-	/* TODO: Add queue reset mask when FW fully supports it */
-	adev->jpeg.supported_reset =
-		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
-	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_4_0_3, ARRAY_SIZE(jpeg_reg_list_4_0_3));
 	if (r)
 		return r;
 
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
 	return 0;
 }
 
@@ -186,6 +242,7 @@ static int jpeg_v4_0_3_sw_fini(struct amdgpu_ip_block *ip_block)
 		return r;
 
 	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+
 	r = amdgpu_jpeg_sw_fini(adev);
 
 	return r;
@@ -331,6 +388,11 @@ static int jpeg_v4_0_3_hw_init(struct amdgpu_ip_block *ip_block)
 			}
 		}
 	} else {
+		/* This flag is not set for VF, assumed to be disabled always */
+		if (RREG32_SOC15(VCN, GET_INST(VCN, 0), regVCN_RRMT_CNTL) &
+		    0x100)
+			adev->jpeg.caps |= AMDGPU_JPEG_CAPS(RRMT_ENABLED);
+
 		for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
 			jpeg_inst = GET_INST(JPEG, i);
 
@@ -382,6 +444,9 @@ static int jpeg_v4_0_3_hw_fini(struct amdgpu_ip_block *ip_block)
 			ret = jpeg_v4_0_3_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
 	}
 
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG) && !amdgpu_sriov_vf(adev))
+		amdgpu_irq_put(adev, &adev->jpeg.inst->ras_poison_irq, 0);
+
 	return ret;
 }
 
@@ -475,6 +540,75 @@ static void jpeg_v4_0_3_enable_clock_gating(struct amdgpu_device *adev, int inst
 	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CGC_GATE, data);
 }
 
+static void jpeg_v4_0_3_start_inst(struct amdgpu_device *adev, int inst)
+{
+	int jpeg_inst = GET_INST(JPEG, inst);
+
+	WREG32_SOC15(JPEG, jpeg_inst, regUVD_PGFSM_CONFIG,
+		     1 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT);
+	SOC15_WAIT_ON_RREG(JPEG, jpeg_inst, regUVD_PGFSM_STATUS,
+			   UVD_PGFSM_STATUS__UVDJ_PWR_ON <<
+			   UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT,
+			   UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK);
+
+	/* disable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS),
+		 0, ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+	/* JPEG disable CGC */
+	jpeg_v4_0_3_disable_clock_gating(adev, inst);
+
+	/* MJPEG global tiling registers */
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_DEC_GFX8_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_DEC_GFX10_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* enable JMI channel */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL), 0,
+		 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+}
+
+static void jpeg_v4_0_3_start_jrbc(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int jpeg_inst = GET_INST(JPEG, ring->me);
+	int reg_offset = jpeg_v4_0_3_core_reg_offset(ring->pipe);
+
+	/* enable System Interrupt for JRBC */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regJPEG_SYS_INT_EN),
+		 JPEG_SYS_INT_EN__DJRBC0_MASK << ring->pipe,
+		 ~(JPEG_SYS_INT_EN__DJRBC0_MASK << ring->pipe));
+
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_LMI_JRBC_RB_VMID,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC0_UVD_JRBC_RB_CNTL,
+			    reg_offset,
+			    (0x00000001L | 0x00000002L));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+			    reg_offset, lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+			    reg_offset, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC0_UVD_JRBC_RB_RPTR,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC0_UVD_JRBC_RB_WPTR,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC0_UVD_JRBC_RB_CNTL,
+			    reg_offset, 0x00000002L);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC0_UVD_JRBC_RB_SIZE,
+			    reg_offset, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15_OFFSET(JPEG, jpeg_inst, regUVD_JRBC0_UVD_JRBC_RB_WPTR,
+					 reg_offset);
+}
+
 /**
  * jpeg_v4_0_3_start - start JPEG block
  *
@@ -485,84 +619,36 @@ static void jpeg_v4_0_3_enable_clock_gating(struct amdgpu_device *adev, int inst
 static int jpeg_v4_0_3_start(struct amdgpu_device *adev)
 {
 	struct amdgpu_ring *ring;
-	int i, j, jpeg_inst;
+	int i, j;
 
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
-		jpeg_inst = GET_INST(JPEG, i);
-
-		WREG32_SOC15(JPEG, jpeg_inst, regUVD_PGFSM_CONFIG,
-			     1 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT);
-		SOC15_WAIT_ON_RREG(
-			JPEG, jpeg_inst, regUVD_PGFSM_STATUS,
-			UVD_PGFSM_STATUS__UVDJ_PWR_ON
-				<< UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT,
-			UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK);
-
-		/* disable anti hang mechanism */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst,
-					  regUVD_JPEG_POWER_STATUS),
-			 0, ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
-
-		/* JPEG disable CGC */
-		jpeg_v4_0_3_disable_clock_gating(adev, i);
-
-		/* MJPEG global tiling registers */
-		WREG32_SOC15(JPEG, jpeg_inst, regJPEG_DEC_GFX8_ADDR_CONFIG,
-			     adev->gfx.config.gb_addr_config);
-		WREG32_SOC15(JPEG, jpeg_inst, regJPEG_DEC_GFX10_ADDR_CONFIG,
-			     adev->gfx.config.gb_addr_config);
-
-		/* enable JMI channel */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL), 0,
-			 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
-
+		jpeg_v4_0_3_start_inst(adev, i);
 		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
-			unsigned int reg_offset = (j?(0x40 * j - 0xc80):0);
-
 			ring = &adev->jpeg.inst[i].ring_dec[j];
-
-			/* enable System Interrupt for JRBC */
-			WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst,
-						  regJPEG_SYS_INT_EN),
-				 JPEG_SYS_INT_EN__DJRBC0_MASK << j,
-				 ~(JPEG_SYS_INT_EN__DJRBC0_MASK << j));
-
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JMI0_UVD_LMI_JRBC_RB_VMID,
-					    reg_offset, 0);
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JRBC0_UVD_JRBC_RB_CNTL,
-					    reg_offset,
-					    (0x00000001L | 0x00000002L));
-			WREG32_SOC15_OFFSET(
-				JPEG, jpeg_inst,
-				regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_LOW,
-				reg_offset, lower_32_bits(ring->gpu_addr));
-			WREG32_SOC15_OFFSET(
-				JPEG, jpeg_inst,
-				regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
-				reg_offset, upper_32_bits(ring->gpu_addr));
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JRBC0_UVD_JRBC_RB_RPTR,
-					    reg_offset, 0);
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JRBC0_UVD_JRBC_RB_WPTR,
-					    reg_offset, 0);
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JRBC0_UVD_JRBC_RB_CNTL,
-					    reg_offset, 0x00000002L);
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JRBC0_UVD_JRBC_RB_SIZE,
-					    reg_offset, ring->ring_size / 4);
-			ring->wptr = RREG32_SOC15_OFFSET(
-				JPEG, jpeg_inst, regUVD_JRBC0_UVD_JRBC_RB_WPTR,
-				reg_offset);
+			jpeg_v4_0_3_start_jrbc(ring);
 		}
 	}
 
 	return 0;
 }
 
+static void jpeg_v4_0_3_stop_inst(struct amdgpu_device *adev, int inst)
+{
+	int jpeg_inst = GET_INST(JPEG, inst);
+	/* reset JMI */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL),
+		 UVD_JMI_CNTL__SOFT_RESET_MASK,
+		 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+	jpeg_v4_0_3_enable_clock_gating(adev, inst);
+
+	/* enable anti hang mechanism */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS),
+		 UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
+		 ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+}
+
 /**
  * jpeg_v4_0_3_stop - stop JPEG block
  *
@@ -572,31 +658,10 @@ static int jpeg_v4_0_3_start(struct amdgpu_device *adev)
  */
 static int jpeg_v4_0_3_stop(struct amdgpu_device *adev)
 {
-	int i, jpeg_inst;
+	int i;
 
-	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
-		jpeg_inst = GET_INST(JPEG, i);
-		/* reset JMI */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL),
-			 UVD_JMI_CNTL__SOFT_RESET_MASK,
-			 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
-
-		jpeg_v4_0_3_enable_clock_gating(adev, i);
-
-		/* enable anti hang mechanism */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst,
-					  regUVD_JPEG_POWER_STATUS),
-			 UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
-			 ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
-
-		WREG32_SOC15(JPEG, jpeg_inst, regUVD_PGFSM_CONFIG,
-			     2 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT);
-		SOC15_WAIT_ON_RREG(
-			JPEG, jpeg_inst, regUVD_PGFSM_STATUS,
-			UVD_PGFSM_STATUS__UVDJ_PWR_OFF
-				<< UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT,
-			UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK);
-	}
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i)
+		jpeg_v4_0_3_stop_inst(adev, i);
 
 	return 0;
 }
@@ -612,9 +677,8 @@ static uint64_t jpeg_v4_0_3_dec_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	struct amdgpu_device *adev = ring->adev;
 
-	return RREG32_SOC15_OFFSET(
-		JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_RPTR,
-		ring->pipe ? (0x40 * ring->pipe - 0xc80) : 0);
+	return RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_RPTR,
+				   jpeg_v4_0_3_core_reg_offset(ring->pipe));
 }
 
 /**
@@ -630,14 +694,12 @@ static uint64_t jpeg_v4_0_3_dec_ring_get_wptr(struct amdgpu_ring *ring)
 
 	if (ring->use_doorbell)
 		return adev->wb.wb[ring->wptr_offs];
-	else
-		return RREG32_SOC15_OFFSET(
-			JPEG, GET_INST(JPEG, ring->me),
-			regUVD_JRBC0_UVD_JRBC_RB_WPTR,
-			ring->pipe ? (0x40 * ring->pipe - 0xc80) : 0);
+
+	return RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_WPTR,
+				   jpeg_v4_0_3_core_reg_offset(ring->pipe));
 }
 
-static void jpeg_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+void jpeg_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring)
 {
 	/* JPEG engine access for HDP flush doesn't work when RRMT is enabled.
 	 * This is a workaround to avoid any HDP flush through JPEG ring.
@@ -659,10 +721,8 @@ static void jpeg_v4_0_3_dec_ring_set_wptr(struct amdgpu_ring *ring)
 		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
 		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
 	} else {
-		WREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me),
-				    regUVD_JRBC0_UVD_JRBC_RB_WPTR,
-				    (ring->pipe ? (0x40 * ring->pipe - 0xc80) :
-						  0),
+		WREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_WPTR,
+				    jpeg_v4_0_3_core_reg_offset(ring->pipe),
 				    lower_32_bits(ring->wptr));
 	}
 }
@@ -907,21 +967,17 @@ void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count)
 	}
 }
 
-static bool jpeg_v4_0_3_is_idle(void *handle)
+static bool jpeg_v4_0_3_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	bool ret = false;
 	int i, j;
 
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
 		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
-			unsigned int reg_offset = (j?(0x40 * j - 0xc80):0);
-
-			ret &= ((RREG32_SOC15_OFFSET(
-					 JPEG, GET_INST(JPEG, i),
-					 regUVD_JRBC0_UVD_JRBC_STATUS,
-					 reg_offset) &
-				 UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
+			ret &= ((RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, i),
+				regUVD_JRBC0_UVD_JRBC_STATUS, jpeg_v4_0_3_core_reg_offset(j)) &
+				UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
 				UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
 		}
 	}
@@ -937,13 +993,10 @@ static int jpeg_v4_0_3_wait_for_idle(struct amdgpu_ip_block *ip_block)
 
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
 		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
-			unsigned int reg_offset = (j?(0x40 * j - 0xc80):0);
-
-			ret &= SOC15_WAIT_ON_RREG_OFFSET(
-				JPEG, GET_INST(JPEG, i),
-				regUVD_JRBC0_UVD_JRBC_STATUS, reg_offset,
+			ret &= (SOC15_WAIT_ON_RREG_OFFSET(JPEG, GET_INST(JPEG, i),
+				regUVD_JRBC0_UVD_JRBC_STATUS, jpeg_v4_0_3_core_reg_offset(j),
 				UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK,
-				UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+				UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK));
 		}
 	}
 	return ret;
@@ -958,7 +1011,7 @@ static int jpeg_v4_0_3_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
 		if (enable) {
-			if (!jpeg_v4_0_3_is_idle(adev))
+			if (!jpeg_v4_0_3_is_idle(ip_block))
 				return -EBUSY;
 			jpeg_v4_0_3_enable_clock_gating(adev, i);
 		} else {
@@ -1001,6 +1054,14 @@ static int jpeg_v4_0_3_set_interrupt_state(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int jpeg_v4_0_3_set_ras_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
 static int jpeg_v4_0_3_process_interrupt(struct amdgpu_device *adev,
 				      struct amdgpu_irq_src *source,
 				      struct amdgpu_iv_entry *entry)
@@ -1055,6 +1116,44 @@ static int jpeg_v4_0_3_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static void jpeg_v4_0_3_core_stall_reset(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int jpeg_inst = GET_INST(JPEG, ring->me);
+	int reg_offset = jpeg_v4_0_3_core_reg_offset(ring->pipe);
+
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_JMI_CLIENT_STALL,
+			    reg_offset, 0x1F);
+	SOC15_WAIT_ON_RREG_OFFSET(JPEG, jpeg_inst,
+				  regUVD_JMI0_UVD_JMI_CLIENT_CLEAN_STATUS,
+				  reg_offset, 0x1F, 0x1F);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_JPEG_LMI_DROP,
+			    reg_offset, 0x1F);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 1 << ring->pipe);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_JMI_CLIENT_STALL,
+			    reg_offset, 0x00);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_JPEG_LMI_DROP,
+			    reg_offset, 0x00);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 0x00);
+}
+
+static int jpeg_v4_0_3_ring_reset(struct amdgpu_ring *ring,
+				  unsigned int vmid,
+				  struct amdgpu_fence *timedout_fence)
+{
+	if (amdgpu_sriov_vf(ring->adev))
+		return -EOPNOTSUPP;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	jpeg_v4_0_3_core_stall_reset(ring);
+	jpeg_v4_0_3_start_jrbc(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
 static const struct amd_ip_funcs jpeg_v4_0_3_ip_funcs = {
 	.name = "jpeg_v4_0_3",
 	.early_init = jpeg_v4_0_3_early_init,
@@ -1068,6 +1167,8 @@ static const struct amd_ip_funcs jpeg_v4_0_3_ip_funcs = {
 	.wait_for_idle = jpeg_v4_0_3_wait_for_idle,
 	.set_clockgating_state = jpeg_v4_0_3_set_clockgating_state,
 	.set_powergating_state = jpeg_v4_0_3_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
 };
 
 static const struct amdgpu_ring_funcs jpeg_v4_0_3_dec_ring_vm_funcs = {
@@ -1076,7 +1177,7 @@ static const struct amdgpu_ring_funcs jpeg_v4_0_3_dec_ring_vm_funcs = {
 	.get_rptr = jpeg_v4_0_3_dec_ring_get_rptr,
 	.get_wptr = jpeg_v4_0_3_dec_ring_get_wptr,
 	.set_wptr = jpeg_v4_0_3_dec_ring_set_wptr,
-	.parse_cs = jpeg_v2_dec_ring_parse_cs,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
 	.emit_frame_size =
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
 		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
@@ -1099,6 +1200,7 @@ static const struct amdgpu_ring_funcs jpeg_v4_0_3_dec_ring_vm_funcs = {
 	.emit_wreg = jpeg_v4_0_3_dec_ring_emit_wreg,
 	.emit_reg_wait = jpeg_v4_0_3_dec_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v4_0_3_ring_reset,
 };
 
 static void jpeg_v4_0_3_set_dec_ring_funcs(struct amdgpu_device *adev)
@@ -1122,6 +1224,11 @@ static const struct amdgpu_irq_src_funcs jpeg_v4_0_3_irq_funcs = {
 	.process = jpeg_v4_0_3_process_interrupt,
 };
 
+static const struct amdgpu_irq_src_funcs jpeg_v4_0_3_ras_irq_funcs = {
+	.set = jpeg_v4_0_3_set_ras_interrupt_state,
+	.process = amdgpu_jpeg_process_poison_irq,
+};
+
 static void jpeg_v4_0_3_set_irq_funcs(struct amdgpu_device *adev)
 {
 	int i;
@@ -1130,6 +1237,9 @@ static void jpeg_v4_0_3_set_irq_funcs(struct amdgpu_device *adev)
 		adev->jpeg.inst->irq.num_types += adev->jpeg.num_jpeg_rings;
 	}
 	adev->jpeg.inst->irq.funcs = &jpeg_v4_0_3_irq_funcs;
+
+	adev->jpeg.inst->ras_poison_irq.num_types = 1;
+	adev->jpeg.inst->ras_poison_irq.funcs = &jpeg_v4_0_3_ras_irq_funcs;
 }
 
 const struct amdgpu_ip_block_version jpeg_v4_0_3_ip_block = {
@@ -1226,9 +1336,47 @@ static void jpeg_v4_0_3_reset_ras_error_count(struct amdgpu_device *adev)
 		jpeg_v4_0_3_inst_reset_ras_error_count(adev, i);
 }
 
+static uint32_t jpeg_v4_0_3_query_poison_by_instance(struct amdgpu_device *adev,
+		uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_JPEG_V4_0_3_JPEG0:
+		reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG0_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG0_STATUS, POISONED_PF);
+		break;
+	case AMDGPU_JPEG_V4_0_3_JPEG1:
+		reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG1_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG1_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in JPEG%d sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool jpeg_v4_0_3_query_ras_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst = 0, sub = 0, poison_stat = 0;
+
+	for (inst = 0; inst < adev->jpeg.num_jpeg_inst; inst++)
+		for (sub = 0; sub < AMDGPU_JPEG_V4_0_3_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+			jpeg_v4_0_3_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
 static const struct amdgpu_ras_block_hw_ops jpeg_v4_0_3_ras_hw_ops = {
 	.query_ras_error_count = jpeg_v4_0_3_query_ras_error_count,
 	.reset_ras_error_count = jpeg_v4_0_3_reset_ras_error_count,
+	.query_poison_status = jpeg_v4_0_3_query_ras_poison_status,
 };
 
 static int jpeg_v4_0_3_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
@@ -1245,11 +1393,13 @@ static int jpeg_v4_0_3_aca_bank_parser(struct aca_handle *handle, struct aca_ban
 	misc0 = bank->regs[ACA_REG_IDX_MISC0];
 	switch (type) {
 	case ACA_SMU_TYPE_UE:
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
 		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE,
 						     1ULL);
 		break;
 	case ACA_SMU_TYPE_CE:
-		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_CE,
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type,
 						     ACA_REG__MISC0__ERRCNT(misc0));
 		break;
 	default:
@@ -1303,6 +1453,13 @@ static int jpeg_v4_0_3_ras_late_init(struct amdgpu_device *adev, struct ras_comm
 	if (r)
 		return r;
 
+	if (amdgpu_ras_is_supported(adev, ras_block->block) &&
+		adev->jpeg.inst->ras_poison_irq.funcs) {
+		r = amdgpu_irq_get(adev, &adev->jpeg.inst->ras_poison_irq, 0);
+		if (r)
+			goto late_fini;
+	}
+
 	r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__JPEG,
 				&jpeg_v4_0_3_aca_info, NULL);
 	if (r)
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h
index 747a3e5f6856..2e110d04af84 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h
@@ -46,6 +46,13 @@
 
 #define JRBC_DEC_EXTERNAL_REG_WRITE_ADDR				0x18000
 
+enum amdgpu_jpeg_v4_0_3_sub_block {
+	AMDGPU_JPEG_V4_0_3_JPEG0 = 0,
+	AMDGPU_JPEG_V4_0_3_JPEG1,
+
+	AMDGPU_JPEG_V4_0_3_MAX_SUB_BLOCK,
+};
+
 extern const struct amdgpu_ip_block_version jpeg_v4_0_3_ip_block;
 
 void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring,
@@ -56,6 +63,7 @@ void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
 				unsigned int flags);
 void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
 					unsigned int vmid, uint64_t pd_addr);
+void jpeg_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring);
 void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count);
 void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring);
 void jpeg_v4_0_3_dec_ring_insert_end(struct amdgpu_ring *ring);
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
index 6b3656984957..54fd9c800c40 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
@@ -46,11 +46,26 @@
 #define regJPEG_CGC_GATE_INTERNAL_OFFSET		0x4160
 #define regUVD_NO_OP_INTERNAL_OFFSET			0x0029
 
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_4_0_5[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH),
+};
+
 static void jpeg_v4_0_5_set_dec_ring_funcs(struct amdgpu_device *adev);
 static void jpeg_v4_0_5_set_irq_funcs(struct amdgpu_device *adev);
 static int jpeg_v4_0_5_set_powergating_state(struct amdgpu_ip_block *ip_block,
 				enum amd_powergating_state state);
-
 static void jpeg_v4_0_5_dec_ring_set_wptr(struct amdgpu_ring *ring);
 
 static int amdgpu_ih_clientid_jpeg[] = {
@@ -58,6 +73,8 @@ static int amdgpu_ih_clientid_jpeg[] = {
 	SOC15_IH_CLIENTID_VCN1
 };
 
+
+
 /**
  * jpeg_v4_0_5_early_init - set function pointers
  *
@@ -153,9 +170,14 @@ static int jpeg_v4_0_5_sw_init(struct amdgpu_ip_block *ip_block)
 		adev->jpeg.inst[i].external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, i, regUVD_JPEG_PITCH);
 	}
 
-	/* TODO: Add queue reset mask when FW fully supports it */
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_4_0_5, ARRAY_SIZE(jpeg_reg_list_4_0_5));
+	if (r)
+		return r;
+
 	adev->jpeg.supported_reset =
 		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
 	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
 	if (r)
 		return r;
@@ -627,9 +649,9 @@ static void jpeg_v4_0_5_dec_ring_set_wptr(struct amdgpu_ring *ring)
 	}
 }
 
-static bool jpeg_v4_0_5_is_idle(void *handle)
+static bool jpeg_v4_0_5_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int i, ret = 1;
 
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
@@ -664,7 +686,7 @@ static int jpeg_v4_0_5_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	struct amdgpu_device *adev = ip_block->adev;
-	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+	bool enable = state == AMD_CG_STATE_GATE;
 	int i;
 
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
@@ -672,7 +694,7 @@ static int jpeg_v4_0_5_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 			continue;
 
 		if (enable) {
-			if (!jpeg_v4_0_5_is_idle(adev))
+			if (!jpeg_v4_0_5_is_idle(ip_block))
 				return -EBUSY;
 
 			jpeg_v4_0_5_enable_clock_gating(adev, i);
@@ -746,6 +768,22 @@ static int jpeg_v4_0_5_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int jpeg_v4_0_5_ring_reset(struct amdgpu_ring *ring,
+				  unsigned int vmid,
+				  struct amdgpu_fence *timedout_fence)
+{
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = jpeg_v4_0_5_stop(ring->adev);
+	if (r)
+		return r;
+	r = jpeg_v4_0_5_start(ring->adev);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
 static const struct amd_ip_funcs jpeg_v4_0_5_ip_funcs = {
 	.name = "jpeg_v4_0_5",
 	.early_init = jpeg_v4_0_5_early_init,
@@ -759,6 +797,8 @@ static const struct amd_ip_funcs jpeg_v4_0_5_ip_funcs = {
 	.wait_for_idle = jpeg_v4_0_5_wait_for_idle,
 	.set_clockgating_state = jpeg_v4_0_5_set_clockgating_state,
 	.set_powergating_state = jpeg_v4_0_5_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
 };
 
 static const struct amdgpu_ring_funcs jpeg_v4_0_5_dec_ring_vm_funcs = {
@@ -767,7 +807,7 @@ static const struct amdgpu_ring_funcs jpeg_v4_0_5_dec_ring_vm_funcs = {
 	.get_rptr = jpeg_v4_0_5_dec_ring_get_rptr,
 	.get_wptr = jpeg_v4_0_5_dec_ring_get_wptr,
 	.set_wptr = jpeg_v4_0_5_dec_ring_set_wptr,
-	.parse_cs = jpeg_v2_dec_ring_parse_cs,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
 	.emit_frame_size =
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
 		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
@@ -789,6 +829,7 @@ static const struct amdgpu_ring_funcs jpeg_v4_0_5_dec_ring_vm_funcs = {
 	.emit_wreg = jpeg_v2_0_dec_ring_emit_wreg,
 	.emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v4_0_5_ring_reset,
 };
 
 static void jpeg_v4_0_5_set_dec_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c
index d5cf0f2799d4..46bf15dce2bd 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c
@@ -34,6 +34,22 @@
 #include "ivsrcid/vcn/irqsrcs_vcn_5_0.h"
 #include "jpeg_v5_0_0.h"
 
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_5_0[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_CNTL),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_RB_SIZE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH),
+};
+
 static void jpeg_v5_0_0_set_dec_ring_funcs(struct amdgpu_device *adev);
 static void jpeg_v5_0_0_set_irq_funcs(struct amdgpu_device *adev);
 static int jpeg_v5_0_0_set_powergating_state(struct amdgpu_ip_block *ip_block,
@@ -100,13 +116,17 @@ static int jpeg_v5_0_0_sw_init(struct amdgpu_ip_block *ip_block)
 	adev->jpeg.internal.jpeg_pitch[0] = regUVD_JPEG_PITCH_INTERNAL_OFFSET;
 	adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_PITCH);
 
-	/* TODO: Add queue reset mask when FW fully supports it */
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_5_0, ARRAY_SIZE(jpeg_reg_list_5_0));
+	if (r)
+		return r;
+
 	adev->jpeg.supported_reset =
 		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
 	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
-	if (r)
-		return r;
-	return 0;
+
+	return r;
 }
 
 /**
@@ -539,9 +559,9 @@ static void jpeg_v5_0_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
 	}
 }
 
-static bool jpeg_v5_0_0_is_idle(void *handle)
+static bool jpeg_v5_0_0_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	int ret = 1;
 
 	ret &= (((RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS) &
@@ -564,10 +584,10 @@ static int jpeg_v5_0_0_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					  enum amd_clockgating_state state)
 {
 	struct amdgpu_device *adev = ip_block->adev;
-	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+	bool enable = state == AMD_CG_STATE_GATE;
 
 	if (enable) {
-		if (!jpeg_v5_0_0_is_idle(adev))
+		if (!jpeg_v5_0_0_is_idle(ip_block))
 			return -EBUSY;
 		jpeg_v5_0_0_enable_clock_gating(adev);
 	} else {
@@ -624,6 +644,22 @@ static int jpeg_v5_0_0_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int jpeg_v5_0_0_ring_reset(struct amdgpu_ring *ring,
+				  unsigned int vmid,
+				  struct amdgpu_fence *timedout_fence)
+{
+	int r;
+
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	r = jpeg_v5_0_0_stop(ring->adev);
+	if (r)
+		return r;
+	r = jpeg_v5_0_0_start(ring->adev);
+	if (r)
+		return r;
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
 static const struct amd_ip_funcs jpeg_v5_0_0_ip_funcs = {
 	.name = "jpeg_v5_0_0",
 	.early_init = jpeg_v5_0_0_early_init,
@@ -637,6 +673,8 @@ static const struct amd_ip_funcs jpeg_v5_0_0_ip_funcs = {
 	.wait_for_idle = jpeg_v5_0_0_wait_for_idle,
 	.set_clockgating_state = jpeg_v5_0_0_set_clockgating_state,
 	.set_powergating_state = jpeg_v5_0_0_set_powergating_state,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
 };
 
 static const struct amdgpu_ring_funcs jpeg_v5_0_0_dec_ring_vm_funcs = {
@@ -645,7 +683,7 @@ static const struct amdgpu_ring_funcs jpeg_v5_0_0_dec_ring_vm_funcs = {
 	.get_rptr = jpeg_v5_0_0_dec_ring_get_rptr,
 	.get_wptr = jpeg_v5_0_0_dec_ring_get_wptr,
 	.set_wptr = jpeg_v5_0_0_dec_ring_set_wptr,
-	.parse_cs = jpeg_v2_dec_ring_parse_cs,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
 	.emit_frame_size =
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
 		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
@@ -667,6 +705,7 @@ static const struct amdgpu_ring_funcs jpeg_v5_0_0_dec_ring_vm_funcs = {
 	.emit_wreg = jpeg_v4_0_3_dec_ring_emit_wreg,
 	.emit_reg_wait = jpeg_v4_0_3_dec_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v5_0_0_ring_reset,
 };
 
 static void jpeg_v5_0_0_set_dec_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c
index 40d4c32a8c2a..ab0bf880d3d8 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c
@@ -28,15 +28,18 @@
 #include "soc15d.h"
 #include "jpeg_v4_0_3.h"
 #include "jpeg_v5_0_1.h"
+#include "mmsch_v5_0.h"
 
 #include "vcn/vcn_5_0_0_offset.h"
 #include "vcn/vcn_5_0_0_sh_mask.h"
 #include "ivsrcid/vcn/irqsrcs_vcn_5_0.h"
 
+static int jpeg_v5_0_1_start_sriov(struct amdgpu_device *adev);
 static void jpeg_v5_0_1_set_dec_ring_funcs(struct amdgpu_device *adev);
 static void jpeg_v5_0_1_set_irq_funcs(struct amdgpu_device *adev);
 static int jpeg_v5_0_1_set_powergating_state(struct amdgpu_ip_block *ip_block,
 					     enum amd_powergating_state state);
+static void jpeg_v5_0_1_set_ras_funcs(struct amdgpu_device *adev);
 static void jpeg_v5_0_1_dec_ring_set_wptr(struct amdgpu_ring *ring);
 
 static int amdgpu_ih_srcid_jpeg[] = {
@@ -52,6 +55,47 @@ static int amdgpu_ih_srcid_jpeg[] = {
 	VCN_5_0__SRCID__JPEG9_DECODE,
 };
 
+static const struct amdgpu_hwip_reg_entry jpeg_reg_list_5_0_1[] = {
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC8_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC8_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC8_UVD_JRBC_STATUS),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC9_UVD_JRBC_RB_RPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC9_UVD_JRBC_RB_WPTR),
+	SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC9_UVD_JRBC_STATUS),
+};
+
 static int jpeg_v5_0_1_core_reg_offset(u32 pipe)
 {
 	if (pipe <= AMDGPU_MAX_JPEG_RINGS_4_0_3)
@@ -77,6 +121,7 @@ static int jpeg_v5_0_1_early_init(struct amdgpu_ip_block *ip_block)
 	adev->jpeg.num_jpeg_rings = AMDGPU_MAX_JPEG_RINGS;
 	jpeg_v5_0_1_set_dec_ring_funcs(adev);
 	jpeg_v5_0_1_set_irq_funcs(adev);
+	jpeg_v5_0_1_set_ras_funcs(adev);
 
 	return 0;
 }
@@ -101,6 +146,17 @@ static int jpeg_v5_0_1_sw_init(struct amdgpu_ip_block *ip_block)
 		if (r)
 			return r;
 	}
+	/* JPEG DJPEG POISON EVENT */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_5_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
+	if (r)
+		return r;
+
+	/* JPEG EJPEG POISON EVENT */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+			VCN_5_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq);
+	if (r)
+		return r;
 
 	r = amdgpu_jpeg_sw_init(adev);
 	if (r)
@@ -115,21 +171,16 @@ static int jpeg_v5_0_1_sw_init(struct amdgpu_ip_block *ip_block)
 
 		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
 			ring = &adev->jpeg.inst[i].ring_dec[j];
-			ring->use_doorbell = false;
+			ring->use_doorbell = true;
 			ring->vm_hub = AMDGPU_MMHUB0(adev->jpeg.inst[i].aid_id);
 			if (!amdgpu_sriov_vf(adev)) {
 				ring->doorbell_index =
 					(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
 					1 + j + 11 * jpeg_inst;
 			} else {
-				if (j < 4)
-					ring->doorbell_index =
-						(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
-						4 + j + 32 * jpeg_inst;
-				else
-					ring->doorbell_index =
-						(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
-						8 + j + 32 * jpeg_inst;
+				ring->doorbell_index =
+					(adev->doorbell_index.vcn.vcn_ring0_1 << 1) +
+					2 + j + 32 * jpeg_inst;
 			}
 			sprintf(ring->name, "jpeg_dec_%d.%d", adev->jpeg.inst[i].aid_id, j);
 			r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0,
@@ -145,7 +196,25 @@ static int jpeg_v5_0_1_sw_init(struct amdgpu_ip_block *ip_block)
 		}
 	}
 
-	return 0;
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG)) {
+		r = amdgpu_jpeg_ras_sw_init(adev);
+		if (r) {
+			dev_err(adev->dev, "Failed to initialize jpeg ras block!\n");
+			return r;
+		}
+	}
+
+	r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_5_0_1, ARRAY_SIZE(jpeg_reg_list_5_0_1));
+	if (r)
+		return r;
+
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
+	if (!amdgpu_sriov_vf(adev))
+		adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+
+	return r;
 }
 
 /**
@@ -164,6 +233,8 @@ static int jpeg_v5_0_1_sw_fini(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
+
 	r = amdgpu_jpeg_sw_fini(adev);
 
 	return r;
@@ -182,7 +253,10 @@ static int jpeg_v5_0_1_hw_init(struct amdgpu_ip_block *ip_block)
 	int i, j, r, jpeg_inst;
 
 	if (amdgpu_sriov_vf(adev)) {
-		/* jpeg_v5_0_1_start_sriov(adev); */
+		r = jpeg_v5_0_1_start_sriov(adev);
+		if (r)
+			return r;
+
 		for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
 			for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
 				ring = &adev->jpeg.inst[i].ring_dec[j];
@@ -194,6 +268,9 @@ static int jpeg_v5_0_1_hw_init(struct amdgpu_ip_block *ip_block)
 		}
 		return 0;
 	}
+	if (RREG32_SOC15(VCN, GET_INST(VCN, 0), regVCN_RRMT_CNTL) & 0x100)
+		adev->jpeg.caps |= AMDGPU_JPEG_CAPS(RRMT_ENABLED);
+
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
 		jpeg_inst = GET_INST(JPEG, i);
 		ring = adev->jpeg.inst[i].ring_dec;
@@ -206,7 +283,7 @@ static int jpeg_v5_0_1_hw_init(struct amdgpu_ip_block *ip_block)
 			ring = &adev->jpeg.inst[i].ring_dec[j];
 			if (ring->use_doorbell)
 				WREG32_SOC15_OFFSET(VCN, GET_INST(VCN, i), regVCN_JPEG_DB_CTRL,
-						    (ring->pipe ? (ring->pipe - 0x15) : 0),
+						    ring->pipe,
 						    ring->doorbell_index <<
 						    VCN_JPEG_DB_CTRL__OFFSET__SHIFT |
 						    VCN_JPEG_DB_CTRL__EN_MASK);
@@ -233,8 +310,13 @@ static int jpeg_v5_0_1_hw_fini(struct amdgpu_ip_block *ip_block)
 
 	cancel_delayed_work_sync(&adev->jpeg.idle_work);
 
-	if (adev->jpeg.cur_state != AMD_PG_STATE_GATE)
-		ret = jpeg_v5_0_1_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
+	if (!amdgpu_sriov_vf(adev)) {
+		if (adev->jpeg.cur_state != AMD_PG_STATE_GATE)
+			ret = jpeg_v5_0_1_set_powergating_state(ip_block, AMD_PG_STATE_GATE);
+	}
+
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG) && !amdgpu_sriov_vf(adev))
+		amdgpu_irq_put(adev, &adev->jpeg.inst->ras_poison_irq, 0);
 
 	return ret;
 }
@@ -281,11 +363,10 @@ static int jpeg_v5_0_1_resume(struct amdgpu_ip_block *ip_block)
 	return r;
 }
 
-static int jpeg_v5_0_1_disable_antihang(struct amdgpu_device *adev, int inst_idx)
+static void jpeg_v5_0_1_init_inst(struct amdgpu_device *adev, int i)
 {
-	int jpeg_inst;
+	int jpeg_inst = GET_INST(JPEG, i);
 
-	jpeg_inst = GET_INST(JPEG, inst_idx);
 	/* disable anti hang mechanism */
 	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS), 0,
 		 ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
@@ -294,19 +375,187 @@ static int jpeg_v5_0_1_disable_antihang(struct amdgpu_device *adev, int inst_idx
 	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS), 0,
 		 ~UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK);
 
-	return 0;
+	/* MJPEG global tiling registers */
+	WREG32_SOC15(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG,
+		     adev->gfx.config.gb_addr_config);
+
+	/* enable JMI channel */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL), 0,
+		 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
 }
 
-static int jpeg_v5_0_1_enable_antihang(struct amdgpu_device *adev, int inst_idx)
+static void jpeg_v5_0_1_deinit_inst(struct amdgpu_device *adev, int i)
 {
-	int jpeg_inst;
+	int jpeg_inst = GET_INST(JPEG, i);
+	/* reset JMI */
+	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL),
+		 UVD_JMI_CNTL__SOFT_RESET_MASK,
+		 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
 
-	jpeg_inst = GET_INST(JPEG, inst_idx);
 	/* enable anti hang mechanism */
 	WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS),
 		 UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
-		~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+		 ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+}
+
+static void jpeg_v5_0_1_init_jrbc(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 reg, data, mask;
+	int jpeg_inst = GET_INST(JPEG, ring->me);
+	int reg_offset = ring->pipe ? jpeg_v5_0_1_core_reg_offset(ring->pipe) : 0;
+
+	/* enable System Interrupt for JRBC */
+	reg = SOC15_REG_OFFSET(JPEG, jpeg_inst, regJPEG_SYS_INT_EN);
+	if (ring->pipe < AMDGPU_MAX_JPEG_RINGS_4_0_3) {
+		data = JPEG_SYS_INT_EN__DJRBC0_MASK << ring->pipe;
+		mask = ~(JPEG_SYS_INT_EN__DJRBC0_MASK << ring->pipe);
+		WREG32_P(reg, data, mask);
+	} else {
+		data = JPEG_SYS_INT_EN__DJRBC0_MASK << (ring->pipe+12);
+		mask = ~(JPEG_SYS_INT_EN__DJRBC0_MASK << (ring->pipe+12));
+		WREG32_P(reg, data, mask);
+	}
+
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_LMI_JRBC_RB_VMID,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC_RB_CNTL,
+			    reg_offset,
+			    (0x00000001L | 0x00000002L));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+			    reg_offset, lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+			    reg_offset, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC_RB_RPTR,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC_RB_WPTR,
+			    reg_offset, 0);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC_RB_CNTL,
+			    reg_offset, 0x00000002L);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JRBC_RB_SIZE,
+			    reg_offset, ring->ring_size / 4);
+	ring->wptr = RREG32_SOC15_OFFSET(JPEG, jpeg_inst, regUVD_JRBC_RB_WPTR,
+					 reg_offset);
+}
+
+static int jpeg_v5_0_1_start_sriov(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	uint64_t ctx_addr;
+	uint32_t param, resp, expected;
+	uint32_t tmp, timeout;
+
+	struct amdgpu_mm_table *table = &adev->virt.mm_table;
+	uint32_t *table_loc;
+	uint32_t table_size;
+	uint32_t size, size_dw, item_offset;
+	uint32_t init_status;
+	int i, j, jpeg_inst;
+
+	struct mmsch_v5_0_cmd_direct_write
+		direct_wt = { {0} };
+	struct mmsch_v5_0_cmd_end end = { {0} };
+	struct mmsch_v5_0_init_header header;
 
+	direct_wt.cmd_header.command_type =
+		MMSCH_COMMAND__DIRECT_REG_WRITE;
+	end.cmd_header.command_type =
+		MMSCH_COMMAND__END;
+
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+		jpeg_inst = GET_INST(JPEG, i);
+
+		memset(&header, 0, sizeof(struct mmsch_v5_0_init_header));
+		header.version = MMSCH_VERSION;
+		header.total_size = sizeof(struct mmsch_v5_0_init_header) >> 2;
+
+		table_loc = (uint32_t *)table->cpu_addr;
+		table_loc += header.total_size;
+
+		item_offset = header.total_size;
+
+		for (j = 0; j < adev->jpeg.num_jpeg_rings; j++) {
+			ring = &adev->jpeg.inst[i].ring_dec[j];
+			table_size = 0;
+
+			tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW);
+			MMSCH_V5_0_INSERT_DIRECT_WT(tmp, lower_32_bits(ring->gpu_addr));
+			tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH);
+			MMSCH_V5_0_INSERT_DIRECT_WT(tmp, upper_32_bits(ring->gpu_addr));
+			tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JRBC_RB_SIZE);
+			MMSCH_V5_0_INSERT_DIRECT_WT(tmp, ring->ring_size / 4);
+
+			if (j < 5) {
+				header.mjpegdec0[j].table_offset = item_offset;
+				header.mjpegdec0[j].init_status = 0;
+				header.mjpegdec0[j].table_size = table_size;
+			} else {
+				header.mjpegdec1[j - 5].table_offset = item_offset;
+				header.mjpegdec1[j - 5].init_status = 0;
+				header.mjpegdec1[j - 5].table_size = table_size;
+			}
+			header.total_size += table_size;
+			item_offset += table_size;
+		}
+
+		MMSCH_V5_0_INSERT_END();
+
+		/* send init table to MMSCH */
+		size = sizeof(struct mmsch_v5_0_init_header);
+		table_loc = (uint32_t *)table->cpu_addr;
+		memcpy((void *)table_loc, &header, size);
+
+		ctx_addr = table->gpu_addr;
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr));
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr));
+
+		tmp = RREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_VMID);
+		tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+		tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_VMID, tmp);
+
+		size = header.total_size;
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_SIZE, size);
+
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_RESP, 0);
+
+		param = 0x00000001;
+		WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_HOST, param);
+		tmp = 0;
+		timeout = 1000;
+		resp = 0;
+		expected = MMSCH_VF_MAILBOX_RESP__OK;
+		init_status =
+			((struct mmsch_v5_0_init_header *)(table_loc))->mjpegdec0[i].init_status;
+		while (resp != expected) {
+			resp = RREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_RESP);
+
+			if (resp != 0)
+				break;
+			udelay(10);
+			tmp = tmp + 10;
+			if (tmp >= timeout) {
+				DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\
+					" waiting for regMMSCH_VF_MAILBOX_RESP "\
+					"(expected=0x%08x, readback=0x%08x)\n",
+					tmp, expected, resp);
+				return -EBUSY;
+			}
+		}
+		if (resp != expected && resp != MMSCH_VF_MAILBOX_RESP__INCOMPLETE &&
+				init_status != MMSCH_VF_ENGINE_STATUS__PASS)
+			DRM_ERROR("MMSCH init status is incorrect! readback=0x%08x, header init status for jpeg: %x\n",
+					resp, init_status);
+
+	}
 	return 0;
 }
 
@@ -320,69 +569,13 @@ static int jpeg_v5_0_1_enable_antihang(struct amdgpu_device *adev, int inst_idx)
 static int jpeg_v5_0_1_start(struct amdgpu_device *adev)
 {
 	struct amdgpu_ring *ring;
-	int i, j, jpeg_inst, r;
+	int i, j;
 
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
-		jpeg_inst = GET_INST(JPEG, i);
-
-		/* disable antihang */
-		r = jpeg_v5_0_1_disable_antihang(adev, i);
-		if (r)
-			return r;
-
-		/* MJPEG global tiling registers */
-		WREG32_SOC15(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG,
-			     adev->gfx.config.gb_addr_config);
-
-		/* enable JMI channel */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL), 0,
-			 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
-
+		jpeg_v5_0_1_init_inst(adev, i);
 		for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) {
-			int reg_offset = (j ? jpeg_v5_0_1_core_reg_offset(j) : 0);
-			u32 reg, data, mask;
-
 			ring = &adev->jpeg.inst[i].ring_dec[j];
-
-			/* enable System Interrupt for JRBC */
-			reg = SOC15_REG_OFFSET(JPEG, jpeg_inst, regJPEG_SYS_INT_EN);
-			if (j < AMDGPU_MAX_JPEG_RINGS_4_0_3) {
-				data = JPEG_SYS_INT_EN__DJRBC0_MASK << j;
-				mask = ~(JPEG_SYS_INT_EN__DJRBC0_MASK << j);
-				WREG32_P(reg, data, mask);
-			} else {
-				data = JPEG_SYS_INT_EN__DJRBC0_MASK << (j+12);
-				mask = ~(JPEG_SYS_INT_EN__DJRBC0_MASK << (j+12));
-				WREG32_P(reg, data, mask);
-			}
-
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_LMI_JRBC_RB_VMID,
-					    reg_offset, 0);
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JRBC_RB_CNTL,
-					    reg_offset,
-					    (0x00000001L | 0x00000002L));
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
-					    reg_offset, lower_32_bits(ring->gpu_addr));
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
-					    reg_offset, upper_32_bits(ring->gpu_addr));
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JRBC_RB_RPTR,
-					    reg_offset, 0);
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JRBC_RB_WPTR,
-					    reg_offset, 0);
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JRBC_RB_CNTL,
-					    reg_offset, 0x00000002L);
-			WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
-					    regUVD_JRBC_RB_SIZE,
-					    reg_offset, ring->ring_size / 4);
-			ring->wptr = RREG32_SOC15_OFFSET(JPEG, jpeg_inst, regUVD_JRBC_RB_WPTR,
-							 reg_offset);
+			jpeg_v5_0_1_init_jrbc(ring);
 		}
 	}
 
@@ -398,20 +591,10 @@ static int jpeg_v5_0_1_start(struct amdgpu_device *adev)
  */
 static int jpeg_v5_0_1_stop(struct amdgpu_device *adev)
 {
-	int i, jpeg_inst, r;
-
-	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
-		jpeg_inst = GET_INST(JPEG, i);
-		/* reset JMI */
-		WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL),
-			 UVD_JMI_CNTL__SOFT_RESET_MASK,
-			 ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+	int i;
 
-		/* enable antihang */
-		r = jpeg_v5_0_1_enable_antihang(adev, i);
-		if (r)
-			return r;
-	}
+	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i)
+		jpeg_v5_0_1_deinit_inst(adev, i);
 
 	return 0;
 }
@@ -471,9 +654,9 @@ static void jpeg_v5_0_1_dec_ring_set_wptr(struct amdgpu_ring *ring)
 	}
 }
 
-static bool jpeg_v5_0_1_is_idle(void *handle)
+static bool jpeg_v5_0_1_is_idle(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_device *adev = ip_block->adev;
 	bool ret = false;
 	int i, j;
 
@@ -514,7 +697,7 @@ static int jpeg_v5_0_1_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 					     enum amd_clockgating_state state)
 {
 	struct amdgpu_device *adev = ip_block->adev;
-	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+	bool enable = state == AMD_CG_STATE_GATE;
 
 	int i;
 
@@ -522,7 +705,7 @@ static int jpeg_v5_0_1_set_clockgating_state(struct amdgpu_ip_block *ip_block,
 		return 0;
 
 	for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
-		if (!jpeg_v5_0_1_is_idle(adev))
+		if (!jpeg_v5_0_1_is_idle(ip_block))
 			return -EBUSY;
 	}
 
@@ -535,6 +718,11 @@ static int jpeg_v5_0_1_set_powergating_state(struct amdgpu_ip_block *ip_block,
 	struct amdgpu_device *adev = ip_block->adev;
 	int ret;
 
+	if (amdgpu_sriov_vf(adev)) {
+		adev->jpeg.cur_state = AMD_PG_STATE_UNGATE;
+		return 0;
+	}
+
 	if (state == adev->jpeg.cur_state)
 		return 0;
 
@@ -557,6 +745,16 @@ static int jpeg_v5_0_1_set_interrupt_state(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int jpeg_v5_0_1_set_ras_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+
+
 static int jpeg_v5_0_1_process_interrupt(struct amdgpu_device *adev,
 					 struct amdgpu_irq_src *source,
 					 struct amdgpu_iv_entry *entry)
@@ -617,6 +815,41 @@ static int jpeg_v5_0_1_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static void jpeg_v5_0_1_core_stall_reset(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int jpeg_inst = GET_INST(JPEG, ring->me);
+	int reg_offset = ring->pipe ? jpeg_v5_0_1_core_reg_offset(ring->pipe) : 0;
+
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_JMI_CLIENT_STALL,
+			    reg_offset, 0x1F);
+	SOC15_WAIT_ON_RREG_OFFSET(JPEG, jpeg_inst,
+				  regUVD_JMI0_UVD_JMI_CLIENT_CLEAN_STATUS,
+				  reg_offset, 0x1F, 0x1F);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_JPEG_LMI_DROP,
+			    reg_offset, 0x1F);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 1 << ring->pipe);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_UVD_JMI_CLIENT_STALL,
+			    reg_offset, 0x00);
+	WREG32_SOC15_OFFSET(JPEG, jpeg_inst,
+			    regUVD_JMI0_JPEG_LMI_DROP,
+			    reg_offset, 0x00);
+	WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 0x00);
+}
+
+static int jpeg_v5_0_1_ring_reset(struct amdgpu_ring *ring,
+				  unsigned int vmid,
+				  struct amdgpu_fence *timedout_fence)
+{
+	amdgpu_ring_reset_helper_begin(ring, timedout_fence);
+	jpeg_v5_0_1_core_stall_reset(ring);
+	jpeg_v5_0_1_init_jrbc(ring);
+	return amdgpu_ring_reset_helper_end(ring, timedout_fence);
+}
+
 static const struct amd_ip_funcs jpeg_v5_0_1_ip_funcs = {
 	.name = "jpeg_v5_0_1",
 	.early_init = jpeg_v5_0_1_early_init,
@@ -635,8 +868,8 @@ static const struct amd_ip_funcs jpeg_v5_0_1_ip_funcs = {
 	.post_soft_reset = NULL,
 	.set_clockgating_state = jpeg_v5_0_1_set_clockgating_state,
 	.set_powergating_state = jpeg_v5_0_1_set_powergating_state,
-	.dump_ip_state = NULL,
-	.print_ip_state = NULL,
+	.dump_ip_state = amdgpu_jpeg_dump_ip_state,
+	.print_ip_state = amdgpu_jpeg_print_ip_state,
 };
 
 static const struct amdgpu_ring_funcs jpeg_v5_0_1_dec_ring_vm_funcs = {
@@ -645,6 +878,7 @@ static const struct amdgpu_ring_funcs jpeg_v5_0_1_dec_ring_vm_funcs = {
 	.get_rptr = jpeg_v5_0_1_dec_ring_get_rptr,
 	.get_wptr = jpeg_v5_0_1_dec_ring_get_wptr,
 	.set_wptr = jpeg_v5_0_1_dec_ring_set_wptr,
+	.parse_cs = amdgpu_jpeg_dec_parse_cs,
 	.emit_frame_size =
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
 		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
@@ -655,6 +889,7 @@ static const struct amdgpu_ring_funcs jpeg_v5_0_1_dec_ring_vm_funcs = {
 	.emit_ib = jpeg_v4_0_3_dec_ring_emit_ib,
 	.emit_fence = jpeg_v4_0_3_dec_ring_emit_fence,
 	.emit_vm_flush = jpeg_v4_0_3_dec_ring_emit_vm_flush,
+	.emit_hdp_flush = jpeg_v4_0_3_ring_emit_hdp_flush,
 	.test_ring = amdgpu_jpeg_dec_ring_test_ring,
 	.test_ib = amdgpu_jpeg_dec_ring_test_ib,
 	.insert_nop = jpeg_v4_0_3_dec_ring_nop,
@@ -666,6 +901,7 @@ static const struct amdgpu_ring_funcs jpeg_v5_0_1_dec_ring_vm_funcs = {
 	.emit_wreg = jpeg_v4_0_3_dec_ring_emit_wreg,
 	.emit_reg_wait = jpeg_v4_0_3_dec_ring_emit_reg_wait,
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+	.reset = jpeg_v5_0_1_ring_reset,
 };
 
 static void jpeg_v5_0_1_set_dec_ring_funcs(struct amdgpu_device *adev)
@@ -689,6 +925,11 @@ static const struct amdgpu_irq_src_funcs jpeg_v5_0_1_irq_funcs = {
 	.process = jpeg_v5_0_1_process_interrupt,
 };
 
+static const struct amdgpu_irq_src_funcs jpeg_v5_0_1_ras_irq_funcs = {
+	.set = jpeg_v5_0_1_set_ras_interrupt_state,
+	.process = amdgpu_jpeg_process_poison_irq,
+};
+
 static void jpeg_v5_0_1_set_irq_funcs(struct amdgpu_device *adev)
 {
 	int i;
@@ -697,6 +938,10 @@ static void jpeg_v5_0_1_set_irq_funcs(struct amdgpu_device *adev)
 		adev->jpeg.inst->irq.num_types += adev->jpeg.num_jpeg_rings;
 
 	adev->jpeg.inst->irq.funcs = &jpeg_v5_0_1_irq_funcs;
+
+	adev->jpeg.inst->ras_poison_irq.num_types = 1;
+	adev->jpeg.inst->ras_poison_irq.funcs = &jpeg_v5_0_1_ras_irq_funcs;
+
 }
 
 const struct amdgpu_ip_block_version jpeg_v5_0_1_ip_block = {
@@ -706,3 +951,151 @@ const struct amdgpu_ip_block_version jpeg_v5_0_1_ip_block = {
 	.rev = 1,
 	.funcs = &jpeg_v5_0_1_ip_funcs,
 };
+
+static uint32_t jpeg_v5_0_1_query_poison_by_instance(struct amdgpu_device *adev,
+		uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_JPEG_V5_0_1_JPEG0:
+		reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG0_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG0_STATUS, POISONED_PF);
+		break;
+	case AMDGPU_JPEG_V5_0_1_JPEG1:
+		reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG1_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG1_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in JPEG%d sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool jpeg_v5_0_1_query_ras_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst = 0, sub = 0, poison_stat = 0;
+
+	for (inst = 0; inst < adev->jpeg.num_jpeg_inst; inst++)
+		for (sub = 0; sub < AMDGPU_JPEG_V5_0_1_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+			jpeg_v5_0_1_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
+static const struct amdgpu_ras_block_hw_ops jpeg_v5_0_1_ras_hw_ops = {
+	.query_poison_status = jpeg_v5_0_1_query_ras_poison_status,
+};
+
+static int jpeg_v5_0_1_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
+				      enum aca_smu_type type, void *data)
+{
+	struct aca_bank_info info;
+	u64 misc0;
+	int ret;
+
+	ret = aca_bank_info_decode(bank, &info);
+	if (ret)
+		return ret;
+
+	misc0 = bank->regs[ACA_REG_IDX_MISC0];
+	switch (type) {
+	case ACA_SMU_TYPE_UE:
+		bank->aca_err_type = ACA_ERROR_TYPE_UE;
+		ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE,
+						     1ULL);
+		break;
+	case ACA_SMU_TYPE_CE:
+		bank->aca_err_type = ACA_ERROR_TYPE_CE;
+		ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type,
+						     ACA_REG__MISC0__ERRCNT(misc0));
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return ret;
+}
+
+/* reference to smu driver if header file */
+static int jpeg_v5_0_1_err_codes[] = {
+	16, 17, 18, 19, 20, 21, 22, 23, /* JPEG[0-9][S|D] */
+	24, 25, 26, 27, 28, 29, 30, 31,
+	48, 49, 50, 51,
+};
+
+static bool jpeg_v5_0_1_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+					 enum aca_smu_type type, void *data)
+{
+	u32 instlo;
+
+	instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+	instlo &= GENMASK(31, 1);
+
+	if (instlo != mmSMNAID_AID0_MCA_SMU)
+		return false;
+
+	if (aca_bank_check_error_codes(handle->adev, bank,
+				       jpeg_v5_0_1_err_codes,
+				       ARRAY_SIZE(jpeg_v5_0_1_err_codes)))
+		return false;
+
+	return true;
+}
+
+static const struct aca_bank_ops jpeg_v5_0_1_aca_bank_ops = {
+	.aca_bank_parser = jpeg_v5_0_1_aca_bank_parser,
+	.aca_bank_is_valid = jpeg_v5_0_1_aca_bank_is_valid,
+};
+
+static const struct aca_info jpeg_v5_0_1_aca_info = {
+	.hwip = ACA_HWIP_TYPE_SMU,
+	.mask = ACA_ERROR_UE_MASK,
+	.bank_ops = &jpeg_v5_0_1_aca_bank_ops,
+};
+
+static int jpeg_v5_0_1_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+	int r;
+
+	r = amdgpu_ras_block_late_init(adev, ras_block);
+	if (r)
+		return r;
+
+	r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__JPEG,
+				&jpeg_v5_0_1_aca_info, NULL);
+	if (r)
+		goto late_fini;
+
+	if (amdgpu_ras_is_supported(adev, ras_block->block) &&
+		adev->jpeg.inst->ras_poison_irq.funcs) {
+		r = amdgpu_irq_get(adev, &adev->jpeg.inst->ras_poison_irq, 0);
+		if (r)
+			goto late_fini;
+	}
+
+	return 0;
+
+late_fini:
+	amdgpu_ras_block_late_fini(adev, ras_block);
+
+	return r;
+}
+
+static struct amdgpu_jpeg_ras jpeg_v5_0_1_ras = {
+	.ras_block = {
+		.hw_ops = &jpeg_v5_0_1_ras_hw_ops,
+		.ras_late_init = jpeg_v5_0_1_ras_late_init,
+	},
+};
+
+static void jpeg_v5_0_1_set_ras_funcs(struct amdgpu_device *adev)
+{
+	adev->jpeg.ras = &jpeg_v5_0_1_ras;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h
index 8ce146c00bb6..a7e58d5fb246 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h
@@ -26,4 +26,86 @@
 
 extern const struct amdgpu_ip_block_version jpeg_v5_0_1_ip_block;
 
-#endif /* __JPEG_V5_0_0_H__ */
+#define regUVD_JRBC0_UVD_JRBC_SCRATCH0_INTERNAL_OFFSET			0x4094
+#define regUVD_JRBC_EXTERNAL_MCM_ADDR_INTERNAL_OFFSET			0x1bffe
+
+#define regUVD_JRBC0_UVD_JRBC_RB_WPTR                                                         0x0640
+#define regUVD_JRBC0_UVD_JRBC_RB_WPTR_BASE_IDX                                                1
+#define regUVD_JRBC0_UVD_JRBC_STATUS                                                          0x0649
+#define regUVD_JRBC0_UVD_JRBC_STATUS_BASE_IDX                                                 1
+#define regUVD_JRBC0_UVD_JRBC_RB_RPTR                                                         0x064a
+#define regUVD_JRBC0_UVD_JRBC_RB_RPTR_BASE_IDX                                                1
+#define regUVD_JRBC1_UVD_JRBC_RB_WPTR                                                         0x0000
+#define regUVD_JRBC1_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC1_UVD_JRBC_STATUS                                                          0x0009
+#define regUVD_JRBC1_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC1_UVD_JRBC_RB_RPTR                                                         0x000a
+#define regUVD_JRBC1_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC2_UVD_JRBC_RB_WPTR                                                         0x0040
+#define regUVD_JRBC2_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC2_UVD_JRBC_STATUS                                                          0x0049
+#define regUVD_JRBC2_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC2_UVD_JRBC_RB_RPTR                                                         0x004a
+#define regUVD_JRBC2_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC3_UVD_JRBC_RB_WPTR                                                         0x0080
+#define regUVD_JRBC3_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC3_UVD_JRBC_STATUS                                                          0x0089
+#define regUVD_JRBC3_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC3_UVD_JRBC_RB_RPTR                                                         0x008a
+#define regUVD_JRBC3_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC4_UVD_JRBC_RB_WPTR                                                         0x00c0
+#define regUVD_JRBC4_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC4_UVD_JRBC_STATUS                                                          0x00c9
+#define regUVD_JRBC4_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC4_UVD_JRBC_RB_RPTR                                                         0x00ca
+#define regUVD_JRBC4_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC5_UVD_JRBC_RB_WPTR                                                         0x0100
+#define regUVD_JRBC5_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC5_UVD_JRBC_STATUS                                                          0x0109
+#define regUVD_JRBC5_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC5_UVD_JRBC_RB_RPTR                                                         0x010a
+#define regUVD_JRBC5_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC6_UVD_JRBC_RB_WPTR                                                         0x0140
+#define regUVD_JRBC6_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC6_UVD_JRBC_STATUS                                                          0x0149
+#define regUVD_JRBC6_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC6_UVD_JRBC_RB_RPTR                                                         0x014a
+#define regUVD_JRBC6_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC7_UVD_JRBC_RB_WPTR                                                         0x0180
+#define regUVD_JRBC7_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC7_UVD_JRBC_STATUS                                                          0x0189
+#define regUVD_JRBC7_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC7_UVD_JRBC_RB_RPTR                                                         0x018a
+#define regUVD_JRBC7_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC8_UVD_JRBC_RB_WPTR                                                         0x01c0
+#define regUVD_JRBC8_UVD_JRBC_RB_WPTR_BASE_IDX                                                0
+#define regUVD_JRBC8_UVD_JRBC_STATUS                                                          0x01c9
+#define regUVD_JRBC8_UVD_JRBC_STATUS_BASE_IDX                                                 0
+#define regUVD_JRBC8_UVD_JRBC_RB_RPTR                                                         0x01ca
+#define regUVD_JRBC8_UVD_JRBC_RB_RPTR_BASE_IDX                                                0
+#define regUVD_JRBC9_UVD_JRBC_RB_WPTR                                                         0x0440
+#define regUVD_JRBC9_UVD_JRBC_RB_WPTR_BASE_IDX                                                1
+#define regUVD_JRBC9_UVD_JRBC_STATUS                                                          0x0449
+#define regUVD_JRBC9_UVD_JRBC_STATUS_BASE_IDX                                                 1
+#define regUVD_JRBC9_UVD_JRBC_RB_RPTR                                                         0x044a
+#define regUVD_JRBC9_UVD_JRBC_RB_RPTR_BASE_IDX                                                1
+#define regUVD_JMI0_JPEG_LMI_DROP                                                             0x0663
+#define regUVD_JMI0_JPEG_LMI_DROP_BASE_IDX                                                    1
+#define regUVD_JMI0_UVD_JMI_CLIENT_STALL                                                      0x067a
+#define regUVD_JMI0_UVD_JMI_CLIENT_STALL_BASE_IDX                                             1
+#define regUVD_JMI0_UVD_JMI_CLIENT_CLEAN_STATUS                                               0x067b
+#define regUVD_JMI0_UVD_JMI_CLIENT_CLEAN_STATUS_BASE_IDX                                      1
+#define regJPEG_CORE_RST_CTRL                                                                 0x072e
+#define regJPEG_CORE_RST_CTRL_BASE_IDX                                                        1
+
+#define regVCN_RRMT_CNTL                          0x0940
+#define regVCN_RRMT_CNTL_BASE_IDX                 1
+
+enum amdgpu_jpeg_v5_0_1_sub_block {
+	AMDGPU_JPEG_V5_0_1_JPEG0 = 0,
+	AMDGPU_JPEG_V5_0_1_JPEG1,
+
+	AMDGPU_JPEG_V5_0_1_MAX_SUB_BLOCK,
+};
+
+#endif /* __JPEG_V5_0_1_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c
new file mode 100644
index 000000000000..64cae89357b6
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c
@@ -0,0 +1,501 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drm_drv.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "mes_userqueue.h"
+#include "amdgpu_userq_fence.h"
+
+#define AMDGPU_USERQ_PROC_CTX_SZ PAGE_SIZE
+#define AMDGPU_USERQ_GANG_CTX_SZ PAGE_SIZE
+
+static int
+mes_userq_map_gtt_bo_to_gart(struct amdgpu_bo *bo)
+{
+	int ret;
+
+	ret = amdgpu_bo_reserve(bo, true);
+	if (ret) {
+		DRM_ERROR("Failed to reserve bo. ret %d\n", ret);
+		goto err_reserve_bo_failed;
+	}
+
+	ret = amdgpu_ttm_alloc_gart(&bo->tbo);
+	if (ret) {
+		DRM_ERROR("Failed to bind bo to GART. ret %d\n", ret);
+		goto err_map_bo_gart_failed;
+	}
+
+	amdgpu_bo_unreserve(bo);
+	bo = amdgpu_bo_ref(bo);
+
+	return 0;
+
+err_map_bo_gart_failed:
+	amdgpu_bo_unreserve(bo);
+err_reserve_bo_failed:
+	return ret;
+}
+
+static int
+mes_userq_create_wptr_mapping(struct amdgpu_userq_mgr *uq_mgr,
+			      struct amdgpu_usermode_queue *queue,
+			      uint64_t wptr)
+{
+	struct amdgpu_bo_va_mapping *wptr_mapping;
+	struct amdgpu_vm *wptr_vm;
+	struct amdgpu_userq_obj *wptr_obj = &queue->wptr_obj;
+	int ret;
+
+	wptr_vm = queue->vm;
+	ret = amdgpu_bo_reserve(wptr_vm->root.bo, false);
+	if (ret)
+		return ret;
+
+	wptr &= AMDGPU_GMC_HOLE_MASK;
+	wptr_mapping = amdgpu_vm_bo_lookup_mapping(wptr_vm, wptr >> PAGE_SHIFT);
+	amdgpu_bo_unreserve(wptr_vm->root.bo);
+	if (!wptr_mapping) {
+		DRM_ERROR("Failed to lookup wptr bo\n");
+		return -EINVAL;
+	}
+
+	wptr_obj->obj = wptr_mapping->bo_va->base.bo;
+	if (wptr_obj->obj->tbo.base.size > PAGE_SIZE) {
+		DRM_ERROR("Requested GART mapping for wptr bo larger than one page\n");
+		return -EINVAL;
+	}
+
+	ret = mes_userq_map_gtt_bo_to_gart(wptr_obj->obj);
+	if (ret) {
+		DRM_ERROR("Failed to map wptr bo to GART\n");
+		return ret;
+	}
+
+	queue->wptr_obj.gpu_addr = amdgpu_bo_gpu_offset_no_check(wptr_obj->obj);
+	return 0;
+}
+
+static int convert_to_mes_priority(int priority)
+{
+	switch (priority) {
+	case AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_NORMAL_LOW:
+	default:
+		return AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
+	case AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_LOW:
+		return AMDGPU_MES_PRIORITY_LEVEL_LOW;
+	case AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_NORMAL_HIGH:
+		return AMDGPU_MES_PRIORITY_LEVEL_MEDIUM;
+	case AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_HIGH:
+		return AMDGPU_MES_PRIORITY_LEVEL_HIGH;
+	}
+}
+
+static int mes_userq_map(struct amdgpu_userq_mgr *uq_mgr,
+			 struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct amdgpu_userq_obj *ctx = &queue->fw_obj;
+	struct amdgpu_mqd_prop *userq_props = queue->userq_prop;
+	struct mes_add_queue_input queue_input;
+	int r;
+
+	memset(&queue_input, 0x0, sizeof(struct mes_add_queue_input));
+
+	queue_input.process_va_start = 0;
+	queue_input.process_va_end = adev->vm_manager.max_pfn - 1;
+
+	/* set process quantum to 10 ms and gang quantum to 1 ms as default */
+	queue_input.process_quantum = 100000;
+	queue_input.gang_quantum = 10000;
+	queue_input.paging = false;
+
+	queue_input.process_context_addr = ctx->gpu_addr;
+	queue_input.gang_context_addr = ctx->gpu_addr + AMDGPU_USERQ_PROC_CTX_SZ;
+	queue_input.inprocess_gang_priority = AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
+	queue_input.gang_global_priority_level = convert_to_mes_priority(queue->priority);
+
+	queue_input.process_id = queue->vm->pasid;
+	queue_input.queue_type = queue->queue_type;
+	queue_input.mqd_addr = queue->mqd.gpu_addr;
+	queue_input.wptr_addr = userq_props->wptr_gpu_addr;
+	queue_input.queue_size = userq_props->queue_size >> 2;
+	queue_input.doorbell_offset = userq_props->doorbell_index;
+	queue_input.page_table_base_addr = amdgpu_gmc_pd_addr(queue->vm->root.bo);
+	queue_input.wptr_mc_addr = queue->wptr_obj.gpu_addr;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->add_hw_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r) {
+		DRM_ERROR("Failed to map queue in HW, err (%d)\n", r);
+		return r;
+	}
+
+	DRM_DEBUG_DRIVER("Queue (doorbell:%d) mapped successfully\n", userq_props->doorbell_index);
+	return 0;
+}
+
+static int mes_userq_unmap(struct amdgpu_userq_mgr *uq_mgr,
+			   struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct mes_remove_queue_input queue_input;
+	struct amdgpu_userq_obj *ctx = &queue->fw_obj;
+	int r;
+
+	memset(&queue_input, 0x0, sizeof(struct mes_remove_queue_input));
+	queue_input.doorbell_offset = queue->doorbell_index;
+	queue_input.gang_context_addr = ctx->gpu_addr + AMDGPU_USERQ_PROC_CTX_SZ;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->remove_hw_queue(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		DRM_ERROR("Failed to unmap queue in HW, err (%d)\n", r);
+	return r;
+}
+
+static int mes_userq_create_ctx_space(struct amdgpu_userq_mgr *uq_mgr,
+				      struct amdgpu_usermode_queue *queue,
+				      struct drm_amdgpu_userq_in *mqd_user)
+{
+	struct amdgpu_userq_obj *ctx = &queue->fw_obj;
+	int r, size;
+
+	/*
+	 * The FW expects at least one page space allocated for
+	 * process ctx and gang ctx each. Create an object
+	 * for the same.
+	 */
+	size = AMDGPU_USERQ_PROC_CTX_SZ + AMDGPU_USERQ_GANG_CTX_SZ;
+	r = amdgpu_userq_create_object(uq_mgr, ctx, size);
+	if (r) {
+		DRM_ERROR("Failed to allocate ctx space bo for userqueue, err:%d\n", r);
+		return r;
+	}
+
+	return 0;
+}
+
+static int mes_userq_detect_and_reset(struct amdgpu_device *adev,
+				      int queue_type)
+{
+	int db_array_size = amdgpu_mes_get_hung_queue_db_array_size(adev);
+	struct mes_detect_and_reset_queue_input input;
+	struct amdgpu_usermode_queue *queue;
+	unsigned int hung_db_num = 0;
+	unsigned long queue_id;
+	u32 db_array[8];
+	bool found_hung_queue = false;
+	int r, i;
+
+	if (db_array_size > 8) {
+		dev_err(adev->dev, "DB array size (%d vs 8) too small\n",
+			db_array_size);
+		return -EINVAL;
+	}
+
+	memset(&input, 0x0, sizeof(struct mes_detect_and_reset_queue_input));
+
+	input.queue_type = queue_type;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = amdgpu_mes_detect_and_reset_hung_queues(adev, queue_type, false,
+						    &hung_db_num, db_array);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r) {
+		dev_err(adev->dev, "Failed to detect and reset queues, err (%d)\n", r);
+	} else if (hung_db_num) {
+		xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) {
+			if (queue->queue_type == queue_type) {
+				for (i = 0; i < hung_db_num; i++) {
+					if (queue->doorbell_index == db_array[i]) {
+						queue->state = AMDGPU_USERQ_STATE_HUNG;
+						found_hung_queue = true;
+						atomic_inc(&adev->gpu_reset_counter);
+						amdgpu_userq_fence_driver_force_completion(queue);
+						drm_dev_wedged_event(adev_to_drm(adev), DRM_WEDGE_RECOVERY_NONE, NULL);
+					}
+				}
+			}
+		}
+	}
+
+	if (found_hung_queue) {
+		/* Resume scheduling after hang recovery */
+		r = amdgpu_mes_resume(adev);
+	}
+
+	return r;
+}
+
+static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr,
+				struct drm_amdgpu_userq_in *args_in,
+				struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct amdgpu_mqd *mqd_hw_default = &adev->mqds[queue->queue_type];
+	struct drm_amdgpu_userq_in *mqd_user = args_in;
+	struct amdgpu_mqd_prop *userq_props;
+	int r;
+
+	/* Structure to initialize MQD for userqueue using generic MQD init function */
+	userq_props = kzalloc(sizeof(struct amdgpu_mqd_prop), GFP_KERNEL);
+	if (!userq_props) {
+		DRM_ERROR("Failed to allocate memory for userq_props\n");
+		return -ENOMEM;
+	}
+
+	r = amdgpu_userq_create_object(uq_mgr, &queue->mqd, mqd_hw_default->mqd_size);
+	if (r) {
+		DRM_ERROR("Failed to create MQD object for userqueue\n");
+		goto free_props;
+	}
+
+	/* Initialize the MQD BO with user given values */
+	userq_props->wptr_gpu_addr = mqd_user->wptr_va;
+	userq_props->rptr_gpu_addr = mqd_user->rptr_va;
+	userq_props->queue_size = mqd_user->queue_size;
+	userq_props->hqd_base_gpu_addr = mqd_user->queue_va;
+	userq_props->mqd_gpu_addr = queue->mqd.gpu_addr;
+	userq_props->use_doorbell = true;
+	userq_props->doorbell_index = queue->doorbell_index;
+	userq_props->fence_address = queue->fence_drv->gpu_addr;
+
+	if (queue->queue_type == AMDGPU_HW_IP_COMPUTE) {
+		struct drm_amdgpu_userq_mqd_compute_gfx11 *compute_mqd;
+
+		if (mqd_user->mqd_size != sizeof(*compute_mqd)) {
+			DRM_ERROR("Invalid compute IP MQD size\n");
+			r = -EINVAL;
+			goto free_mqd;
+		}
+
+		compute_mqd = memdup_user(u64_to_user_ptr(mqd_user->mqd), mqd_user->mqd_size);
+		if (IS_ERR(compute_mqd)) {
+			DRM_ERROR("Failed to read user MQD\n");
+			r = -ENOMEM;
+			goto free_mqd;
+		}
+
+		r = amdgpu_userq_input_va_validate(queue, compute_mqd->eop_va,
+						   2048);
+		if (r)
+			goto free_mqd;
+
+		userq_props->eop_gpu_addr = compute_mqd->eop_va;
+		userq_props->hqd_pipe_priority = AMDGPU_GFX_PIPE_PRIO_NORMAL;
+		userq_props->hqd_queue_priority = AMDGPU_GFX_QUEUE_PRIORITY_MINIMUM;
+		userq_props->hqd_active = false;
+		userq_props->tmz_queue =
+			mqd_user->flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE;
+		kfree(compute_mqd);
+	} else if (queue->queue_type == AMDGPU_HW_IP_GFX) {
+		struct drm_amdgpu_userq_mqd_gfx11 *mqd_gfx_v11;
+		struct amdgpu_gfx_shadow_info shadow_info;
+
+		if (adev->gfx.funcs->get_gfx_shadow_info) {
+			adev->gfx.funcs->get_gfx_shadow_info(adev, &shadow_info, true);
+		} else {
+			r = -EINVAL;
+			goto free_mqd;
+		}
+
+		if (mqd_user->mqd_size != sizeof(*mqd_gfx_v11) || !mqd_user->mqd) {
+			DRM_ERROR("Invalid GFX MQD\n");
+			r = -EINVAL;
+			goto free_mqd;
+		}
+
+		mqd_gfx_v11 = memdup_user(u64_to_user_ptr(mqd_user->mqd), mqd_user->mqd_size);
+		if (IS_ERR(mqd_gfx_v11)) {
+			DRM_ERROR("Failed to read user MQD\n");
+			r = -ENOMEM;
+			goto free_mqd;
+		}
+
+		userq_props->shadow_addr = mqd_gfx_v11->shadow_va;
+		userq_props->csa_addr = mqd_gfx_v11->csa_va;
+		userq_props->tmz_queue =
+			mqd_user->flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE;
+
+		r = amdgpu_userq_input_va_validate(queue, mqd_gfx_v11->shadow_va,
+						   shadow_info.shadow_size);
+		if (r)
+			goto free_mqd;
+		r = amdgpu_userq_input_va_validate(queue, mqd_gfx_v11->csa_va,
+						   shadow_info.csa_size);
+		if (r)
+			goto free_mqd;
+
+		kfree(mqd_gfx_v11);
+	} else if (queue->queue_type == AMDGPU_HW_IP_DMA) {
+		struct drm_amdgpu_userq_mqd_sdma_gfx11 *mqd_sdma_v11;
+
+		if (mqd_user->mqd_size != sizeof(*mqd_sdma_v11) || !mqd_user->mqd) {
+			DRM_ERROR("Invalid SDMA MQD\n");
+			r = -EINVAL;
+			goto free_mqd;
+		}
+
+		mqd_sdma_v11 = memdup_user(u64_to_user_ptr(mqd_user->mqd), mqd_user->mqd_size);
+		if (IS_ERR(mqd_sdma_v11)) {
+			DRM_ERROR("Failed to read sdma user MQD\n");
+			r = -ENOMEM;
+			goto free_mqd;
+		}
+		r = amdgpu_userq_input_va_validate(queue, mqd_sdma_v11->csa_va,
+						   32);
+		if (r)
+			goto free_mqd;
+
+		userq_props->csa_addr = mqd_sdma_v11->csa_va;
+		kfree(mqd_sdma_v11);
+	}
+
+	queue->userq_prop = userq_props;
+
+	r = mqd_hw_default->init_mqd(adev, (void *)queue->mqd.cpu_ptr, userq_props);
+	if (r) {
+		DRM_ERROR("Failed to initialize MQD for userqueue\n");
+		goto free_mqd;
+	}
+
+	/* Create BO for FW operations */
+	r = mes_userq_create_ctx_space(uq_mgr, queue, mqd_user);
+	if (r) {
+		DRM_ERROR("Failed to allocate BO for userqueue (%d)", r);
+		goto free_mqd;
+	}
+
+	/* FW expects WPTR BOs to be mapped into GART */
+	r = mes_userq_create_wptr_mapping(uq_mgr, queue, userq_props->wptr_gpu_addr);
+	if (r) {
+		DRM_ERROR("Failed to create WPTR mapping\n");
+		goto free_ctx;
+	}
+
+	return 0;
+
+free_ctx:
+	amdgpu_userq_destroy_object(uq_mgr, &queue->fw_obj);
+
+free_mqd:
+	amdgpu_userq_destroy_object(uq_mgr, &queue->mqd);
+
+free_props:
+	kfree(userq_props);
+
+	return r;
+}
+
+static void
+mes_userq_mqd_destroy(struct amdgpu_userq_mgr *uq_mgr,
+		      struct amdgpu_usermode_queue *queue)
+{
+	amdgpu_userq_destroy_object(uq_mgr, &queue->fw_obj);
+	kfree(queue->userq_prop);
+	amdgpu_userq_destroy_object(uq_mgr, &queue->mqd);
+}
+
+static int mes_userq_preempt(struct amdgpu_userq_mgr *uq_mgr,
+				struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct mes_suspend_gang_input queue_input;
+	struct amdgpu_userq_obj *ctx = &queue->fw_obj;
+	signed long timeout = 2100000; /* 2100 ms */
+	u64 fence_gpu_addr;
+	u32 fence_offset;
+	u64 *fence_ptr;
+	int i, r;
+
+	if (queue->state != AMDGPU_USERQ_STATE_MAPPED)
+		return 0;
+	r = amdgpu_device_wb_get(adev, &fence_offset);
+	if (r)
+		return r;
+
+	fence_gpu_addr = adev->wb.gpu_addr + (fence_offset * 4);
+	fence_ptr = (u64 *)&adev->wb.wb[fence_offset];
+	*fence_ptr = 0;
+
+	memset(&queue_input, 0x0, sizeof(struct mes_suspend_gang_input));
+	queue_input.gang_context_addr = ctx->gpu_addr + AMDGPU_USERQ_PROC_CTX_SZ;
+	queue_input.suspend_fence_addr = fence_gpu_addr;
+	queue_input.suspend_fence_value = 1;
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->suspend_gang(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r) {
+		DRM_ERROR("Failed to suspend gang: %d\n", r);
+		goto out;
+	}
+
+	for (i = 0; i < timeout; i++) {
+		if (*fence_ptr == 1)
+			goto out;
+		udelay(1);
+	}
+	r = -ETIMEDOUT;
+
+out:
+	amdgpu_device_wb_free(adev, fence_offset);
+	return r;
+}
+
+static int mes_userq_restore(struct amdgpu_userq_mgr *uq_mgr,
+				struct amdgpu_usermode_queue *queue)
+{
+	struct amdgpu_device *adev = uq_mgr->adev;
+	struct mes_resume_gang_input queue_input;
+	struct amdgpu_userq_obj *ctx = &queue->fw_obj;
+	int r;
+
+	if (queue->state == AMDGPU_USERQ_STATE_HUNG)
+		return -EINVAL;
+	if (queue->state != AMDGPU_USERQ_STATE_PREEMPTED)
+		return 0;
+
+	memset(&queue_input, 0x0, sizeof(struct mes_resume_gang_input));
+	queue_input.gang_context_addr = ctx->gpu_addr + AMDGPU_USERQ_PROC_CTX_SZ;
+
+	amdgpu_mes_lock(&adev->mes);
+	r = adev->mes.funcs->resume_gang(&adev->mes, &queue_input);
+	amdgpu_mes_unlock(&adev->mes);
+	if (r)
+		dev_err(adev->dev, "Failed to resume queue, err (%d)\n", r);
+	return r;
+}
+
+const struct amdgpu_userq_funcs userq_mes_funcs = {
+	.mqd_create = mes_userq_mqd_create,
+	.mqd_destroy = mes_userq_mqd_destroy,
+	.unmap = mes_userq_unmap,
+	.map = mes_userq_map,
+	.detect_and_reset = mes_userq_detect_and_reset,
+	.preempt = mes_userq_preempt,
+	.restore = mes_userq_restore,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_userqueue.h b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.h
new file mode 100644
index 000000000000..090ae8897770
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.h
@@ -0,0 +1,30 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef MES_USERQ_H
+#define MES_USERQ_H
+#include "amdgpu_userq.h"
+
+extern const struct amdgpu_userq_funcs userq_mes_funcs;
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
index 65f389eb65e5..3a52754b5cad 100644
--- a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
@@ -54,6 +54,8 @@ MODULE_FIRMWARE("amdgpu/gc_11_5_1_mes_2.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_5_1_mes1.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_5_2_mes_2.bin");
 MODULE_FIRMWARE("amdgpu/gc_11_5_2_mes1.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_mes_2.bin");
+MODULE_FIRMWARE("amdgpu/gc_11_5_3_mes1.bin");
 
 static int mes_v11_0_hw_init(struct amdgpu_ip_block *ip_block);
 static int mes_v11_0_hw_fini(struct amdgpu_ip_block *ip_block);
@@ -62,6 +64,10 @@ static int mes_v11_0_kiq_hw_fini(struct amdgpu_device *adev);
 
 #define MES_EOP_SIZE   2048
 #define GFX_MES_DRAM_SIZE	0x80000
+#define MES11_HW_RESOURCE_1_SIZE (128 * AMDGPU_GPU_PAGE_SIZE)
+
+#define MES11_HUNG_DB_OFFSET_ARRAY_SIZE 8 /* [0:3] = db offset, [4:7] = hqd info */
+#define MES11_HUNG_HQD_INFO_OFFSET	4
 
 static void mes_v11_0_ring_set_wptr(struct amdgpu_ring *ring)
 {
@@ -284,6 +290,23 @@ static int convert_to_mes_queue_type(int queue_type)
 	return -1;
 }
 
+static int convert_to_mes_priority_level(int priority_level)
+{
+	switch (priority_level) {
+	case AMDGPU_MES_PRIORITY_LEVEL_LOW:
+		return AMD_PRIORITY_LEVEL_LOW;
+	case AMDGPU_MES_PRIORITY_LEVEL_NORMAL:
+	default:
+		return AMD_PRIORITY_LEVEL_NORMAL;
+	case AMDGPU_MES_PRIORITY_LEVEL_MEDIUM:
+		return AMD_PRIORITY_LEVEL_MEDIUM;
+	case AMDGPU_MES_PRIORITY_LEVEL_HIGH:
+		return AMD_PRIORITY_LEVEL_HIGH;
+	case AMDGPU_MES_PRIORITY_LEVEL_REALTIME:
+		return AMD_PRIORITY_LEVEL_REALTIME;
+	}
+}
+
 static int mes_v11_0_add_hw_queue(struct amdgpu_mes *mes,
 				  struct mes_add_queue_input *input)
 {
@@ -307,9 +330,9 @@ static int mes_v11_0_add_hw_queue(struct amdgpu_mes *mes,
 	mes_add_queue_pkt.gang_quantum = input->gang_quantum;
 	mes_add_queue_pkt.gang_context_addr = input->gang_context_addr;
 	mes_add_queue_pkt.inprocess_gang_priority =
-		input->inprocess_gang_priority;
+		convert_to_mes_priority_level(input->inprocess_gang_priority);
 	mes_add_queue_pkt.gang_global_priority_level =
-		input->gang_global_priority_level;
+		convert_to_mes_priority_level(input->gang_global_priority_level);
 	mes_add_queue_pkt.doorbell_offset = input->doorbell_offset;
 	mes_add_queue_pkt.mqd_addr = input->mqd_addr;
 
@@ -346,6 +369,7 @@ static int mes_v11_0_remove_hw_queue(struct amdgpu_mes *mes,
 				     struct mes_remove_queue_input *input)
 {
 	union MESAPI__REMOVE_QUEUE mes_remove_queue_pkt;
+	uint32_t mes_rev = mes->sched_version & AMDGPU_MES_VERSION_MASK;
 
 	memset(&mes_remove_queue_pkt, 0, sizeof(mes_remove_queue_pkt));
 
@@ -356,6 +380,9 @@ static int mes_v11_0_remove_hw_queue(struct amdgpu_mes *mes,
 	mes_remove_queue_pkt.doorbell_offset = input->doorbell_offset;
 	mes_remove_queue_pkt.gang_context_addr = input->gang_context_addr;
 
+	if (mes_rev >= 0x60)
+		mes_remove_queue_pkt.remove_queue_after_reset = input->remove_queue_after_reset;
+
 	return mes_v11_0_submit_pkt_and_poll_completion(mes,
 			&mes_remove_queue_pkt, sizeof(mes_remove_queue_pkt),
 			offsetof(union MESAPI__REMOVE_QUEUE, api_status));
@@ -455,31 +482,6 @@ static int mes_v11_0_reset_queue_mmio(struct amdgpu_mes *mes, uint32_t queue_typ
 	return r;
 }
 
-static int mes_v11_0_reset_hw_queue(struct amdgpu_mes *mes,
-				    struct mes_reset_queue_input *input)
-{
-	if (input->use_mmio)
-		return mes_v11_0_reset_queue_mmio(mes, input->queue_type,
-						  input->me_id, input->pipe_id,
-						  input->queue_id, input->vmid);
-
-	union MESAPI__RESET mes_reset_queue_pkt;
-
-	memset(&mes_reset_queue_pkt, 0, sizeof(mes_reset_queue_pkt));
-
-	mes_reset_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
-	mes_reset_queue_pkt.header.opcode = MES_SCH_API_RESET;
-	mes_reset_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
-
-	mes_reset_queue_pkt.doorbell_offset = input->doorbell_offset;
-	mes_reset_queue_pkt.gang_context_addr = input->gang_context_addr;
-	/*mes_reset_queue_pkt.reset_queue_only = 1;*/
-
-	return mes_v11_0_submit_pkt_and_poll_completion(mes,
-			&mes_reset_queue_pkt, sizeof(mes_reset_queue_pkt),
-			offsetof(union MESAPI__REMOVE_QUEUE, api_status));
-}
-
 static int mes_v11_0_map_legacy_queue(struct amdgpu_mes *mes,
 				      struct mes_map_legacy_queue_input *input)
 {
@@ -646,8 +648,9 @@ static int mes_v11_0_misc_op(struct amdgpu_mes *mes,
 		break;
 	case MES_MISC_OP_CHANGE_CONFIG:
 		if ((mes->adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) < 0x63) {
-			dev_err(mes->adev->dev, "MES FW versoin must be larger than 0x63 to support limit single process feature.\n");
-			return -EINVAL;
+			dev_warn_once(mes->adev->dev,
+				      "MES FW version must be larger than 0x63 to support limit single process feature.\n");
+			return 0;
 		}
 		misc_pkt.opcode = MESAPI_MISC__CHANGE_CONFIG;
 		misc_pkt.change_config.opcode =
@@ -691,7 +694,8 @@ static int mes_v11_0_set_hw_resources(struct amdgpu_mes *mes)
 			mes->compute_hqd_mask[i];
 
 	for (i = 0; i < MAX_GFX_PIPES; i++)
-		mes_set_hw_res_pkt.gfx_hqd_mask[i] = mes->gfx_hqd_mask[i];
+		mes_set_hw_res_pkt.gfx_hqd_mask[i] =
+			mes->gfx_hqd_mask[i];
 
 	for (i = 0; i < MAX_SDMA_PIPES; i++)
 		mes_set_hw_res_pkt.sdma_hqd_mask[i] = mes->sdma_hqd_mask[i];
@@ -714,13 +718,19 @@ static int mes_v11_0_set_hw_resources(struct amdgpu_mes *mes)
 	mes_set_hw_res_pkt.enable_reg_active_poll = 1;
 	mes_set_hw_res_pkt.enable_level_process_quantum_check = 1;
 	mes_set_hw_res_pkt.oversubscription_timer = 50;
+	if ((mes->adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x7f)
+		mes_set_hw_res_pkt.enable_lr_compute_wa = 1;
+	else
+		dev_info_once(mes->adev->dev,
+			      "MES FW version must be >= 0x7f to enable LR compute workaround.\n");
+
 	if (amdgpu_mes_log_enable) {
 		mes_set_hw_res_pkt.enable_mes_event_int_logging = 1;
 		mes_set_hw_res_pkt.event_intr_history_gpu_mc_ptr =
 					mes->event_log_gpu_addr;
 	}
 
-	if (enforce_isolation)
+	if (adev->enforce_isolation[0] == AMDGPU_ENFORCE_ISOLATION_ENABLE)
 		mes_set_hw_res_pkt.limit_single_process = 1;
 
 	return mes_v11_0_submit_pkt_and_poll_completion(mes,
@@ -730,9 +740,6 @@ static int mes_v11_0_set_hw_resources(struct amdgpu_mes *mes)
 
 static int mes_v11_0_set_hw_resources_1(struct amdgpu_mes *mes)
 {
-	int size = 128 * PAGE_SIZE;
-	int ret = 0;
-	struct amdgpu_device *adev = mes->adev;
 	union MESAPI_SET_HW_RESOURCES_1 mes_set_hw_res_pkt;
 	memset(&mes_set_hw_res_pkt, 0, sizeof(mes_set_hw_res_pkt));
 
@@ -741,25 +748,20 @@ static int mes_v11_0_set_hw_resources_1(struct amdgpu_mes *mes)
 	mes_set_hw_res_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
 	mes_set_hw_res_pkt.enable_mes_info_ctx = 1;
 
-	ret = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE,
-				AMDGPU_GEM_DOMAIN_VRAM,
-				&mes->resource_1,
-				&mes->resource_1_gpu_addr,
-				&mes->resource_1_addr);
-	if (ret) {
-		dev_err(adev->dev, "(%d) failed to create mes resource_1 bo\n", ret);
-		return ret;
+	mes_set_hw_res_pkt.cleaner_shader_fence_mc_addr = mes->resource_1_gpu_addr[0];
+	if (amdgpu_sriov_is_mes_info_enable(mes->adev)) {
+		mes_set_hw_res_pkt.mes_info_ctx_mc_addr =
+			mes->resource_1_gpu_addr[0] + AMDGPU_GPU_PAGE_SIZE;
+		mes_set_hw_res_pkt.mes_info_ctx_size = MES11_HW_RESOURCE_1_SIZE;
 	}
 
-	mes_set_hw_res_pkt.mes_info_ctx_mc_addr = mes->resource_1_gpu_addr;
-	mes_set_hw_res_pkt.mes_info_ctx_size = mes->resource_1->tbo.base.size;
 	return mes_v11_0_submit_pkt_and_poll_completion(mes,
 			&mes_set_hw_res_pkt, sizeof(mes_set_hw_res_pkt),
 			offsetof(union MESAPI_SET_HW_RESOURCES_1, api_status));
 }
 
-static int mes_v11_0_reset_legacy_queue(struct amdgpu_mes *mes,
-					struct mes_reset_legacy_queue_input *input)
+static int mes_v11_0_reset_hw_queue(struct amdgpu_mes *mes,
+				    struct mes_reset_queue_input *input)
 {
 	union MESAPI__RESET mes_reset_queue_pkt;
 
@@ -777,7 +779,7 @@ static int mes_v11_0_reset_legacy_queue(struct amdgpu_mes *mes,
 	mes_reset_queue_pkt.queue_type =
 		convert_to_mes_queue_type(input->queue_type);
 
-	if (mes_reset_queue_pkt.queue_type == MES_QUEUE_TYPE_GFX) {
+	if (input->legacy_gfx) {
 		mes_reset_queue_pkt.reset_legacy_gfx = 1;
 		mes_reset_queue_pkt.pipe_id_lp = input->pipe_id;
 		mes_reset_queue_pkt.queue_id_lp = input->queue_id;
@@ -795,6 +797,32 @@ static int mes_v11_0_reset_legacy_queue(struct amdgpu_mes *mes,
 			offsetof(union MESAPI__RESET, api_status));
 }
 
+static int mes_v11_0_detect_and_reset_hung_queues(struct amdgpu_mes *mes,
+						  struct mes_detect_and_reset_queue_input *input)
+{
+	union MESAPI__RESET mes_reset_queue_pkt;
+
+	memset(&mes_reset_queue_pkt, 0, sizeof(mes_reset_queue_pkt));
+
+	mes_reset_queue_pkt.header.type = MES_API_TYPE_SCHEDULER;
+	mes_reset_queue_pkt.header.opcode = MES_SCH_API_RESET;
+	mes_reset_queue_pkt.header.dwsize = API_FRAME_SIZE_IN_DWORDS;
+
+	mes_reset_queue_pkt.queue_type =
+		convert_to_mes_queue_type(input->queue_type);
+	mes_reset_queue_pkt.doorbell_offset_addr =
+		mes->hung_queue_db_array_gpu_addr;
+
+	if (input->detect_only)
+		mes_reset_queue_pkt.hang_detect_only = 1;
+	else
+		mes_reset_queue_pkt.hang_detect_then_reset = 1;
+
+	return mes_v11_0_submit_pkt_and_poll_completion(mes,
+			&mes_reset_queue_pkt, sizeof(mes_reset_queue_pkt),
+			offsetof(union MESAPI__RESET, api_status));
+}
+
 static const struct amdgpu_mes_funcs mes_v11_0_funcs = {
 	.add_hw_queue = mes_v11_0_add_hw_queue,
 	.remove_hw_queue = mes_v11_0_remove_hw_queue,
@@ -803,12 +831,12 @@ static const struct amdgpu_mes_funcs mes_v11_0_funcs = {
 	.suspend_gang = mes_v11_0_suspend_gang,
 	.resume_gang = mes_v11_0_resume_gang,
 	.misc_op = mes_v11_0_misc_op,
-	.reset_legacy_queue = mes_v11_0_reset_legacy_queue,
 	.reset_hw_queue = mes_v11_0_reset_hw_queue,
+	.detect_and_reset_hung_queues = mes_v11_0_detect_and_reset_hung_queues,
 };
 
 static int mes_v11_0_allocate_ucode_buffer(struct amdgpu_device *adev,
-					   enum admgpu_mes_pipe pipe)
+					   enum amdgpu_mes_pipe pipe)
 {
 	int r;
 	const struct mes_firmware_header_v1_0 *mes_hdr;
@@ -843,7 +871,7 @@ static int mes_v11_0_allocate_ucode_buffer(struct amdgpu_device *adev,
 }
 
 static int mes_v11_0_allocate_ucode_data_buffer(struct amdgpu_device *adev,
-						enum admgpu_mes_pipe pipe)
+						enum amdgpu_mes_pipe pipe)
 {
 	int r;
 	const struct mes_firmware_header_v1_0 *mes_hdr;
@@ -884,7 +912,7 @@ static int mes_v11_0_allocate_ucode_data_buffer(struct amdgpu_device *adev,
 }
 
 static void mes_v11_0_free_ucode_buffers(struct amdgpu_device *adev,
-					 enum admgpu_mes_pipe pipe)
+					 enum amdgpu_mes_pipe pipe)
 {
 	amdgpu_bo_free_kernel(&adev->mes.data_fw_obj[pipe],
 			      &adev->mes.data_fw_gpu_addr[pipe],
@@ -899,6 +927,10 @@ static void mes_v11_0_get_fw_version(struct amdgpu_device *adev)
 {
 	int pipe;
 
+	/* return early if we have already fetched these */
+	if (adev->mes.sched_version && adev->mes.kiq_version)
+		return;
+
 	/* get MES scheduler/KIQ versions */
 	mutex_lock(&adev->srbm_mutex);
 
@@ -982,7 +1014,7 @@ static void mes_v11_0_enable(struct amdgpu_device *adev, bool enable)
 
 /* This function is for backdoor MES firmware */
 static int mes_v11_0_load_microcode(struct amdgpu_device *adev,
-				    enum admgpu_mes_pipe pipe, bool prime_icache)
+				    enum amdgpu_mes_pipe pipe, bool prime_icache)
 {
 	int r;
 	uint32_t data;
@@ -1054,7 +1086,7 @@ static int mes_v11_0_load_microcode(struct amdgpu_device *adev,
 }
 
 static int mes_v11_0_allocate_eop_buf(struct amdgpu_device *adev,
-				      enum admgpu_mes_pipe pipe)
+				      enum amdgpu_mes_pipe pipe)
 {
 	int r;
 	u32 *eop;
@@ -1265,7 +1297,7 @@ static int mes_v11_0_kiq_enable_queue(struct amdgpu_device *adev)
 }
 
 static int mes_v11_0_queue_init(struct amdgpu_device *adev,
-				enum admgpu_mes_pipe pipe)
+				enum amdgpu_mes_pipe pipe)
 {
 	struct amdgpu_ring *ring;
 	int r;
@@ -1348,7 +1380,7 @@ static int mes_v11_0_kiq_ring_init(struct amdgpu_device *adev)
 }
 
 static int mes_v11_0_mqd_sw_init(struct amdgpu_device *adev,
-				 enum admgpu_mes_pipe pipe)
+				 enum amdgpu_mes_pipe pipe)
 {
 	int r, mqd_size = sizeof(struct v11_compute_mqd);
 	struct amdgpu_ring *ring;
@@ -1389,7 +1421,7 @@ static int mes_v11_0_mqd_sw_init(struct amdgpu_device *adev,
 static int mes_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 {
 	struct amdgpu_device *adev = ip_block->adev;
-	int pipe, r;
+	int pipe, r, bo_size;
 
 	adev->mes.funcs = &mes_v11_0_funcs;
 	adev->mes.kiq_hw_init = &mes_v11_0_kiq_hw_init;
@@ -1424,6 +1456,23 @@ static int mes_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	bo_size = AMDGPU_GPU_PAGE_SIZE;
+	if (amdgpu_sriov_is_mes_info_enable(adev))
+		bo_size += MES11_HW_RESOURCE_1_SIZE;
+
+	/* Only needed for AMDGPU_MES_SCHED_PIPE on MES 11*/
+	r = amdgpu_bo_create_kernel(adev,
+				    bo_size,
+				    PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM,
+				    &adev->mes.resource_1[0],
+				    &adev->mes.resource_1_gpu_addr[0],
+				    &adev->mes.resource_1_addr[0]);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create mes resource_1 bo\n", r);
+		return r;
+	}
+
 	return 0;
 }
 
@@ -1432,6 +1481,9 @@ static int mes_v11_0_sw_fini(struct amdgpu_ip_block *ip_block)
 	struct amdgpu_device *adev = ip_block->adev;
 	int pipe;
 
+	amdgpu_bo_free_kernel(&adev->mes.resource_1[0], &adev->mes.resource_1_gpu_addr[0],
+			      &adev->mes.resource_1_addr[0]);
+
 	for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
 		kfree(adev->mes.mqd_backup[pipe]);
 
@@ -1619,7 +1671,7 @@ static int mes_v11_0_hw_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		goto failure;
 
-	if (amdgpu_sriov_is_mes_info_enable(adev)) {
+	if ((adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x50) {
 		r = mes_v11_0_set_hw_resources_1(&adev->mes);
 		if (r) {
 			DRM_ERROR("failed mes_v11_0_set_hw_resources_1, r=%d\n", r);
@@ -1633,6 +1685,10 @@ static int mes_v11_0_hw_init(struct amdgpu_ip_block *ip_block)
 		goto failure;
 	}
 
+	r = amdgpu_mes_update_enforce_isolation(adev);
+	if (r)
+		goto failure;
+
 out:
 	/*
 	 * Disable KIQ ring usage from the driver once MES is enabled.
@@ -1651,34 +1707,17 @@ failure:
 
 static int mes_v11_0_hw_fini(struct amdgpu_ip_block *ip_block)
 {
-	struct amdgpu_device *adev = ip_block->adev;
-	if (amdgpu_sriov_is_mes_info_enable(adev)) {
-		amdgpu_bo_free_kernel(&adev->mes.resource_1, &adev->mes.resource_1_gpu_addr,
-					&adev->mes.resource_1_addr);
-	}
 	return 0;
 }
 
 static int mes_v11_0_suspend(struct amdgpu_ip_block *ip_block)
 {
-	int r;
-
-	r = amdgpu_mes_suspend(ip_block->adev);
-	if (r)
-		return r;
-
 	return mes_v11_0_hw_fini(ip_block);
 }
 
 static int mes_v11_0_resume(struct amdgpu_ip_block *ip_block)
 {
-	int r;
-
-	r = mes_v11_0_hw_init(ip_block);
-	if (r)
-		return r;
-
-	return amdgpu_mes_resume(ip_block->adev);
+	return mes_v11_0_hw_init(ip_block);
 }
 
 static int mes_v11_0_early_init(struct amdgpu_ip_block *ip_block)
@@ -1686,6 +1725,9 @@ static int mes_v11_0_early_init(struct amdgpu_ip_block *ip_block)
 	struct amdgpu_device *adev = ip_block->adev;
 	int pipe, r;
 
+	adev->mes.hung_queue_db_array_size = MES11_HUNG_DB_OFFSET_ARRAY_SIZE;
+	adev->mes.hung_queue_hqd_info_offset = MES11_HUNG_HQD_INFO_OFFSET;
+
 	for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
 		if (!adev->enable_mes_kiq && pipe == AMDGPU_MES_KIQ_PIPE)
 			continue;
@@ -1697,22 +1739,10 @@ static int mes_v11_0_early_init(struct amdgpu_ip_block *ip_block)
 	return 0;
 }
 
-static int mes_v11_0_late_init(struct amdgpu_ip_block *ip_block)
-{
-	struct amdgpu_device *adev = ip_block->adev;
-
-	/* it's only intended for use in mes_self_test case, not for s0ix and reset */
-	if (!amdgpu_in_reset(adev) && !adev->in_s0ix && !adev->in_suspend &&
-	    (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(11, 0, 3)))
-		amdgpu_mes_self_test(adev);
-
-	return 0;
-}
-
 static const struct amd_ip_funcs mes_v11_0_ip_funcs = {
 	.name = "mes_v11_0",
 	.early_init = mes_v11_0_early_init,
-	.late_init = mes_v11_0_late_init,
+	.late_init = NULL,
 	.sw_init = mes_v11_0_sw_init,
 	.sw_fini = mes_v11_0_sw_fini,
 	.hw_init = mes_v11_0_hw_init,
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c