0 files changed, 0 insertions, 0 deletions
diff --git a/drivers/Kconfig b/drivers/Kconfig
index 7c556c5ac4fd..c0f1fb893ec0 100644
--- a/drivers/Kconfig
+++ b/drivers/Kconfig
@@ -10,6 +10,12 @@ source "drivers/cxl/Kconfig"
 source "drivers/pcmcia/Kconfig"
 source "drivers/rapidio/Kconfig"
 
+config PC104
+	bool "PC/104 support" if EXPERT
+	help
+	  Expose PC/104 form factor device drivers and options available for
+	  selection and configuration. Enable this option if your target
+	  machine has a PC/104 bus.
 
 source "drivers/base/Kconfig"
 
@@ -77,6 +83,8 @@ source "drivers/pps/Kconfig"
 
 source "drivers/ptp/Kconfig"
 
+source "drivers/dpll/Kconfig"
+
 source "drivers/pinctrl/Kconfig"
 
 source "drivers/gpio/Kconfig"
@@ -153,6 +161,8 @@ source "drivers/greybus/Kconfig"
 
 source "drivers/comedi/Kconfig"
 
+source "drivers/gpib/Kconfig"
+
 source "drivers/staging/Kconfig"
 
 source "drivers/platform/Kconfig"
@@ -209,8 +219,6 @@ source "drivers/thunderbolt/Kconfig"
 
 source "drivers/android/Kconfig"
 
-source "drivers/gpu/trace/Kconfig"
-
 source "drivers/nvdimm/Kconfig"
 
 source "drivers/dax/Kconfig"
@@ -245,6 +253,6 @@ source "drivers/hte/Kconfig"
 
 source "drivers/cdx/Kconfig"
 
-source "drivers/dpll/Kconfig"
+source "drivers/resctrl/Kconfig"
 
 endmenu
diff --git a/drivers/Makefile b/drivers/Makefile
index b5749cf67044..ccc05f1eae3e 100644
--- a/drivers/Makefile
+++ b/drivers/Makefile
@@ -150,6 +150,7 @@ obj-$(CONFIG_VHOST_IOTLB)	+= vhost/
 obj-$(CONFIG_VHOST)		+= vhost/
 obj-$(CONFIG_GREYBUS)		+= greybus/
 obj-$(CONFIG_COMEDI)		+= comedi/
+obj-$(CONFIG_GPIB)		+= gpib/
 obj-$(CONFIG_STAGING)		+= staging/
 obj-y				+= platform/
 
@@ -160,8 +161,8 @@ obj-$(CONFIG_RPMSG)		+= rpmsg/
 obj-$(CONFIG_SOUNDWIRE)		+= soundwire/
 
 # Virtualization drivers
-obj-$(CONFIG_VIRT_DRIVERS)	+= virt/
-obj-$(subst m,y,$(CONFIG_HYPERV))	+= hv/
+obj-y				+= virt/
+obj-$(CONFIG_HYPERV)		+= hv/
 
 obj-$(CONFIG_PM_DEVFREQ)	+= devfreq/
 obj-$(CONFIG_EXTCON)		+= extcon/
@@ -194,5 +195,7 @@ obj-$(CONFIG_HTE)		+= hte/
 obj-$(CONFIG_DRM_ACCEL)		+= accel/
 obj-$(CONFIG_CDX_BUS)		+= cdx/
 obj-$(CONFIG_DPLL)		+= dpll/
+obj-y				+= resctrl/
 
+obj-$(CONFIG_DIBS)		+= dibs/
 obj-$(CONFIG_S390)		+= s390/
diff --git a/drivers/accel/Kconfig b/drivers/accel/Kconfig
index 5b9490367a39..bdf48ccafcf2 100644
--- a/drivers/accel/Kconfig
+++ b/drivers/accel/Kconfig
@@ -25,8 +25,10 @@ menuconfig DRM_ACCEL
 	  and debugfs).
 
 source "drivers/accel/amdxdna/Kconfig"
+source "drivers/accel/ethosu/Kconfig"
 source "drivers/accel/habanalabs/Kconfig"
 source "drivers/accel/ivpu/Kconfig"
 source "drivers/accel/qaic/Kconfig"
+source "drivers/accel/rocket/Kconfig"
 
 endif
diff --git a/drivers/accel/Makefile b/drivers/accel/Makefile
index a301fb6089d4..1d3a7251b950 100644
--- a/drivers/accel/Makefile
+++ b/drivers/accel/Makefile
@@ -1,6 +1,8 @@
 # SPDX-License-Identifier: GPL-2.0-only
 
 obj-$(CONFIG_DRM_ACCEL_AMDXDNA)		+= amdxdna/
+obj-$(CONFIG_DRM_ACCEL_ARM_ETHOSU)	+= ethosu/
 obj-$(CONFIG_DRM_ACCEL_HABANALABS)	+= habanalabs/
 obj-$(CONFIG_DRM_ACCEL_IVPU)		+= ivpu/
 obj-$(CONFIG_DRM_ACCEL_QAIC)		+= qaic/
+obj-$(CONFIG_DRM_ACCEL_ROCKET)		+= rocket/
+\ No newline at end of file
diff --git a/drivers/accel/amdxdna/Makefile b/drivers/accel/amdxdna/Makefile
index 0e9adf6890a0..6344aaf523fa 100644
--- a/drivers/accel/amdxdna/Makefile
+++ b/drivers/accel/amdxdna/Makefile
@@ -14,7 +14,9 @@ amdxdna-y := \
 	amdxdna_mailbox.o \
 	amdxdna_mailbox_helper.o \
 	amdxdna_pci_drv.o \
+	amdxdna_pm.o \
 	amdxdna_sysfs.o \
+	amdxdna_ubuf.o \
 	npu1_regs.o \
 	npu2_regs.o \
 	npu4_regs.o \
diff --git a/drivers/accel/amdxdna/TODO b/drivers/accel/amdxdna/TODO
index ad8ac6e315b6..0e4bbebeaedf 100644
--- a/drivers/accel/amdxdna/TODO
+++ b/drivers/accel/amdxdna/TODO
@@ -1,2 +1 @@
 - Add debugfs support
-- Add debug BO support
diff --git a/drivers/accel/amdxdna/aie2_ctx.c b/drivers/accel/amdxdna/aie2_ctx.c
index e04549f64d69..42d876a427c5 100644
--- a/drivers/accel/amdxdna/aie2_ctx.c
+++ b/drivers/accel/amdxdna/aie2_ctx.c
@@ -21,6 +21,7 @@
 #include "amdxdna_gem.h"
 #include "amdxdna_mailbox.h"
 #include "amdxdna_pci_drv.h"
+#include "amdxdna_pm.h"
 
 static bool force_cmdlist;
 module_param(force_cmdlist, bool, 0600);
@@ -46,6 +47,17 @@ static void aie2_job_put(struct amdxdna_sched_job *job)
 	kref_put(&job->refcnt, aie2_job_release);
 }
 
+static void aie2_hwctx_status_shift_stop(struct amdxdna_hwctx *hwctx)
+{
+	 hwctx->old_status = hwctx->status;
+	 hwctx->status = HWCTX_STAT_STOP;
+}
+
+static void aie2_hwctx_status_restore(struct amdxdna_hwctx *hwctx)
+{
+	hwctx->status = hwctx->old_status;
+}
+
 /* The bad_job is used in aie2_sched_job_timedout, otherwise, set it to NULL */
 static void aie2_hwctx_stop(struct amdxdna_dev *xdna, struct amdxdna_hwctx *hwctx,
 			    struct drm_sched_job *bad_job)
@@ -77,7 +89,7 @@ static int aie2_hwctx_restart(struct amdxdna_dev *xdna, struct amdxdna_hwctx *hw
 		goto out;
 	}
 
-	ret = aie2_config_cu(hwctx);
+	ret = aie2_config_cu(hwctx, NULL);
 	if (ret) {
 		XDNA_ERR(xdna, "Config cu failed, ret %d", ret);
 		goto out;
@@ -89,25 +101,6 @@ out:
 	return ret;
 }
 
-void aie2_restart_ctx(struct amdxdna_client *client)
-{
-	struct amdxdna_dev *xdna = client->xdna;
-	struct amdxdna_hwctx *hwctx;
-	unsigned long hwctx_id;
-
-	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
-	mutex_lock(&client->hwctx_lock);
-	amdxdna_for_each_hwctx(client, hwctx_id, hwctx) {
-		if (hwctx->status != HWCTX_STAT_STOP)
-			continue;
-
-		hwctx->status = hwctx->old_status;
-		XDNA_DBG(xdna, "Resetting %s", hwctx->name);
-		aie2_hwctx_restart(xdna, hwctx);
-	}
-	mutex_unlock(&client->hwctx_lock);
-}
-
 static struct dma_fence *aie2_cmd_get_out_fence(struct amdxdna_hwctx *hwctx, u64 seq)
 {
 	struct dma_fence *fence, *out_fence = NULL;
@@ -141,34 +134,46 @@ static void aie2_hwctx_wait_for_idle(struct amdxdna_hwctx *hwctx)
 	dma_fence_put(fence);
 }
 
-void aie2_hwctx_suspend(struct amdxdna_hwctx *hwctx)
+static int aie2_hwctx_suspend_cb(struct amdxdna_hwctx *hwctx, void *arg)
 {
 	struct amdxdna_dev *xdna = hwctx->client->xdna;
 
+	aie2_hwctx_wait_for_idle(hwctx);
+	aie2_hwctx_stop(xdna, hwctx, NULL);
+	aie2_hwctx_status_shift_stop(hwctx);
+
+	return 0;
+}
+
+void aie2_hwctx_suspend(struct amdxdna_client *client)
+{
+	struct amdxdna_dev *xdna = client->xdna;
+
 	/*
 	 * Command timeout is unlikely. But if it happens, it doesn't
 	 * break the system. aie2_hwctx_stop() will destroy mailbox
 	 * and abort all commands.
 	 */
 	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
-	aie2_hwctx_wait_for_idle(hwctx);
-	aie2_hwctx_stop(xdna, hwctx, NULL);
-	hwctx->old_status = hwctx->status;
-	hwctx->status = HWCTX_STAT_STOP;
+	amdxdna_hwctx_walk(client, NULL, aie2_hwctx_suspend_cb);
 }
 
-void aie2_hwctx_resume(struct amdxdna_hwctx *hwctx)
+static int aie2_hwctx_resume_cb(struct amdxdna_hwctx *hwctx, void *arg)
 {
 	struct amdxdna_dev *xdna = hwctx->client->xdna;
 
+	aie2_hwctx_status_restore(hwctx);
+	return aie2_hwctx_restart(xdna, hwctx);
+}
+
+int aie2_hwctx_resume(struct amdxdna_client *client)
+{
 	/*
 	 * The resume path cannot guarantee that mailbox channel can be
 	 * regenerated. If this happen, when submit message to this
 	 * mailbox channel, error will return.
 	 */
-	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
-	hwctx->status = hwctx->old_status;
-	aie2_hwctx_restart(xdna, hwctx);
+	return amdxdna_hwctx_walk(client, NULL, aie2_hwctx_resume_cb);
 }
 
 static void
@@ -177,12 +182,13 @@ aie2_sched_notify(struct amdxdna_sched_job *job)
 	struct dma_fence *fence = job->fence;
 
 	trace_xdna_job(&job->base, job->hwctx->name, "signaled fence", job->seq);
+
+	amdxdna_pm_suspend_put(job->hwctx->client->xdna);
 	job->hwctx->priv->completed++;
 	dma_fence_signal(fence);
 
 	up(&job->hwctx->priv->job_sem);
 	job->job_done = true;
-	dma_fence_put(fence);
 	mmput_async(job->mm);
 	aie2_job_put(job);
 }
@@ -192,15 +198,18 @@ aie2_sched_resp_handler(void *handle, void __iomem *data, size_t size)
 {
 	struct amdxdna_sched_job *job = handle;
 	struct amdxdna_gem_obj *cmd_abo;
-	u32 ret = 0;
+	int ret = 0;
 	u32 status;
 
 	cmd_abo = job->cmd_bo;
 
-	if (unlikely(!data))
+	if (unlikely(job->job_timeout)) {
+		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_TIMEOUT);
+		ret = -EINVAL;
 		goto out;
+	}
 
-	if (unlikely(size != sizeof(u32))) {
+	if (unlikely(!data) || unlikely(size != sizeof(u32))) {
 		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_ABORT);
 		ret = -EINVAL;
 		goto out;
@@ -219,11 +228,10 @@ out:
 }
 
 static int
-aie2_sched_nocmd_resp_handler(void *handle, void __iomem *data, size_t size)
+aie2_sched_drvcmd_resp_handler(void *handle, void __iomem *data, size_t size)
 {
 	struct amdxdna_sched_job *job = handle;
-	u32 ret = 0;
-	u32 status;
+	int ret = 0;
 
 	if (unlikely(!data))
 		goto out;
@@ -233,8 +241,7 @@ aie2_sched_nocmd_resp_handler(void *handle, void __iomem *data, size_t size)
 		goto out;
 	}
 
-	status = readl(data);
-	XDNA_DBG(job->hwctx->client->xdna, "Resp status 0x%x", status);
+	job->drv_cmd->result = readl(data);
 
 out:
 	aie2_sched_notify(job);
@@ -250,9 +257,16 @@ aie2_sched_cmdlist_resp_handler(void *handle, void __iomem *data, size_t size)
 	u32 fail_cmd_status;
 	u32 fail_cmd_idx;
 	u32 cmd_status;
-	u32 ret = 0;
+	int ret = 0;
 
 	cmd_abo = job->cmd_bo;
+
+	if (unlikely(job->job_timeout)) {
+		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_TIMEOUT);
+		ret = -EINVAL;
+		goto out;
+	}
+
 	if (unlikely(!data) || unlikely(size != sizeof(u32) * 3)) {
 		amdxdna_cmd_set_state(cmd_abo, ERT_CMD_STATE_ABORT);
 		ret = -EINVAL;
@@ -307,8 +321,18 @@ aie2_sched_job_run(struct drm_sched_job *sched_job)
 	kref_get(&job->refcnt);
 	fence = dma_fence_get(job->fence);
 
-	if (unlikely(!cmd_abo)) {
-		ret = aie2_sync_bo(hwctx, job, aie2_sched_nocmd_resp_handler);
+	if (job->drv_cmd) {
+		switch (job->drv_cmd->opcode) {
+		case SYNC_DEBUG_BO:
+			ret = aie2_sync_bo(hwctx, job, aie2_sched_drvcmd_resp_handler);
+			break;
+		case ATTACH_DEBUG_BO:
+			ret = aie2_config_debug_bo(hwctx, job, aie2_sched_drvcmd_resp_handler);
+			break;
+		default:
+			ret = -EINVAL;
+			break;
+		}
 		goto out;
 	}
 
@@ -355,13 +379,14 @@ aie2_sched_job_timedout(struct drm_sched_job *sched_job)
 
 	xdna = hwctx->client->xdna;
 	trace_xdna_job(sched_job, hwctx->name, "job timedout", job->seq);
+	job->job_timeout = true;
 	mutex_lock(&xdna->dev_lock);
 	aie2_hwctx_stop(xdna, hwctx, sched_job);
 
 	aie2_hwctx_restart(xdna, hwctx);
 	mutex_unlock(&xdna->dev_lock);
 
-	return DRM_GPU_SCHED_STAT_NOMINAL;
+	return DRM_GPU_SCHED_STAT_RESET;
 }
 
 static const struct drm_sched_backend_ops sched_ops = {
@@ -524,13 +549,12 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
 		.num_rqs = DRM_SCHED_PRIORITY_COUNT,
 		.credit_limit = HWCTX_MAX_CMDS,
 		.timeout = msecs_to_jiffies(HWCTX_MAX_TIMEOUT),
-		.name = hwctx->name,
+		.name = "amdxdna_js",
 		.dev = xdna->ddev.dev,
 	};
 	struct drm_gpu_scheduler *sched;
 	struct amdxdna_hwctx_priv *priv;
 	struct amdxdna_gem_obj *heap;
-	struct amdxdna_dev_hdl *ndev;
 	int i, ret;
 
 	priv = kzalloc(sizeof(*hwctx->priv), GFP_KERNEL);
@@ -566,7 +590,7 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
 			.size = MAX_CHAIN_CMDBUF_SIZE,
 		};
 
-		abo = amdxdna_drm_alloc_dev_bo(&xdna->ddev, &args, client->filp, true);
+		abo = amdxdna_drm_alloc_dev_bo(&xdna->ddev, &args, client->filp);
 		if (IS_ERR(abo)) {
 			ret = PTR_ERR(abo);
 			goto free_cmd_bufs;
@@ -603,10 +627,14 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
 		goto free_entity;
 	}
 
+	ret = amdxdna_pm_resume_get(xdna);
+	if (ret)
+		goto free_col_list;
+
 	ret = aie2_alloc_resource(hwctx);
 	if (ret) {
 		XDNA_ERR(xdna, "Alloc hw resource failed, ret %d", ret);
-		goto free_col_list;
+		goto suspend_put;
 	}
 
 	ret = aie2_map_host_buf(xdna->dev_handle, hwctx->fw_ctx_id,
@@ -621,10 +649,9 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
 		XDNA_ERR(xdna, "Create syncobj failed, ret %d", ret);
 		goto release_resource;
 	}
+	amdxdna_pm_suspend_put(xdna);
 
 	hwctx->status = HWCTX_STAT_INIT;
-	ndev = xdna->dev_handle;
-	ndev->hwctx_num++;
 	init_waitqueue_head(&priv->job_free_wq);
 
 	XDNA_DBG(xdna, "hwctx %s init completed", hwctx->name);
@@ -633,6 +660,8 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
 
 release_resource:
 	aie2_release_resource(hwctx);
+suspend_put:
+	amdxdna_pm_suspend_put(xdna);
 free_col_list:
 	kfree(hwctx->col_list);
 free_entity:
@@ -655,26 +684,25 @@ free_priv:
 
 void aie2_hwctx_fini(struct amdxdna_hwctx *hwctx)
 {
-	struct amdxdna_dev_hdl *ndev;
 	struct amdxdna_dev *xdna;
 	int idx;
 
 	xdna = hwctx->client->xdna;
-	ndev = xdna->dev_handle;
-	ndev->hwctx_num--;
 
 	XDNA_DBG(xdna, "%s sequence number %lld", hwctx->name, hwctx->priv->seq);
-	drm_sched_entity_destroy(&hwctx->priv->entity);
-
 	aie2_hwctx_wait_for_idle(hwctx);
 
 	/* Request fw to destroy hwctx and cancel the rest pending requests */
 	aie2_release_resource(hwctx);
 
+	mutex_unlock(&xdna->dev_lock);
+	drm_sched_entity_destroy(&hwctx->priv->entity);
+
 	/* Wait for all submitted jobs to be completed or canceled */
 	wait_event(hwctx->priv->job_free_wq,
 		   atomic64_read(&hwctx->job_submit_cnt) ==
 		   atomic64_read(&hwctx->job_free_cnt));
+	mutex_lock(&xdna->dev_lock);
 
 	drm_sched_fini(&hwctx->priv->sched);
 	aie2_ctx_syncobj_destroy(hwctx);
@@ -690,6 +718,14 @@ void aie2_hwctx_fini(struct amdxdna_hwctx *hwctx)
 	kfree(hwctx->cus);
 }
 
+static int aie2_config_cu_resp_handler(void *handle, void __iomem *data, size_t size)
+{
+	struct amdxdna_hwctx *hwctx = handle;
+
+	amdxdna_pm_suspend_put(hwctx->client->xdna);
+	return 0;
+}
+
 static int aie2_hwctx_cu_config(struct amdxdna_hwctx *hwctx, void *buf, u32 size)
 {
 	struct amdxdna_hwctx_param_config_cu *config = buf;
@@ -721,10 +757,14 @@ static int aie2_hwctx_cu_config(struct amdxdna_hwctx *hwctx, void *buf, u32 size
 	if (!hwctx->cus)
 		return -ENOMEM;
 
-	ret = aie2_config_cu(hwctx);
+	ret = amdxdna_pm_resume_get(xdna);
+	if (ret)
+		goto free_cus;
+
+	ret = aie2_config_cu(hwctx, aie2_config_cu_resp_handler);
 	if (ret) {
 		XDNA_ERR(xdna, "Config CU to firmware failed, ret %d", ret);
-		goto free_cus;
+		goto pm_suspend_put;
 	}
 
 	wmb(); /* To avoid locking in command submit when check status */
@@ -732,12 +772,82 @@ static int aie2_hwctx_cu_config(struct amdxdna_hwctx *hwctx, void *buf, u32 size
 
 	return 0;
 
+pm_suspend_put:
+	amdxdna_pm_suspend_put(xdna);
 free_cus:
 	kfree(hwctx->cus);
 	hwctx->cus = NULL;
 	return ret;
 }
 
+static void aie2_cmd_wait(struct amdxdna_hwctx *hwctx, u64 seq)
+{
+	struct dma_fence *out_fence = aie2_cmd_get_out_fence(hwctx, seq);
+
+	if (!out_fence) {
+		XDNA_ERR(hwctx->client->xdna, "Failed to get fence");
+		return;
+	}
+
+	dma_fence_wait_timeout(out_fence, false, MAX_SCHEDULE_TIMEOUT);
+	dma_fence_put(out_fence);
+}
+
+static int aie2_hwctx_cfg_debug_bo(struct amdxdna_hwctx *hwctx, u32 bo_hdl,
+				   bool attach)
+{
+	struct amdxdna_client *client = hwctx->client;
+	struct amdxdna_dev *xdna = client->xdna;
+	struct amdxdna_drv_cmd cmd = { 0 };
+	struct amdxdna_gem_obj *abo;
+	u64 seq;
+	int ret;
+
+	abo = amdxdna_gem_get_obj(client, bo_hdl, AMDXDNA_BO_DEV);
+	if (!abo) {
+		XDNA_ERR(xdna, "Get bo %d failed", bo_hdl);
+		return -EINVAL;
+	}
+
+	if (attach) {
+		if (abo->assigned_hwctx != AMDXDNA_INVALID_CTX_HANDLE) {
+			ret = -EBUSY;
+			goto put_obj;
+		}
+		cmd.opcode = ATTACH_DEBUG_BO;
+	} else {
+		if (abo->assigned_hwctx != hwctx->id) {
+			ret = -EINVAL;
+			goto put_obj;
+		}
+		cmd.opcode = DETACH_DEBUG_BO;
+	}
+
+	ret = amdxdna_cmd_submit(client, &cmd, AMDXDNA_INVALID_BO_HANDLE,
+				 &bo_hdl, 1, hwctx->id, &seq);
+	if (ret) {
+		XDNA_ERR(xdna, "Submit command failed");
+		goto put_obj;
+	}
+
+	aie2_cmd_wait(hwctx, seq);
+	if (cmd.result) {
+		XDNA_ERR(xdna, "Response failure 0x%x", cmd.result);
+		goto put_obj;
+	}
+
+	if (attach)
+		abo->assigned_hwctx = hwctx->id;
+	else
+		abo->assigned_hwctx = AMDXDNA_INVALID_CTX_HANDLE;
+
+	XDNA_DBG(xdna, "Config debug BO %d to %s", bo_hdl, hwctx->name);
+
+put_obj:
+	amdxdna_gem_put_obj(abo);
+	return ret;
+}
+
 int aie2_hwctx_config(struct amdxdna_hwctx *hwctx, u32 type, u64 value, void *buf, u32 size)
 {
 	struct amdxdna_dev *xdna = hwctx->client->xdna;
@@ -747,14 +857,40 @@ int aie2_hwctx_config(struct amdxdna_hwctx *hwctx, u32 type, u64 value, void *bu
 	case DRM_AMDXDNA_HWCTX_CONFIG_CU:
 		return aie2_hwctx_cu_config(hwctx, buf, size);
 	case DRM_AMDXDNA_HWCTX_ASSIGN_DBG_BUF:
+		return aie2_hwctx_cfg_debug_bo(hwctx, (u32)value, true);
 	case DRM_AMDXDNA_HWCTX_REMOVE_DBG_BUF:
-		return -EOPNOTSUPP;
+		return aie2_hwctx_cfg_debug_bo(hwctx, (u32)value, false);
 	default:
 		XDNA_DBG(xdna, "Not supported type %d", type);
 		return -EOPNOTSUPP;
 	}
 }
 
+int aie2_hwctx_sync_debug_bo(struct amdxdna_hwctx *hwctx, u32 debug_bo_hdl)
+{
+	struct amdxdna_client *client = hwctx->client;
+	struct amdxdna_dev *xdna = client->xdna;
+	struct amdxdna_drv_cmd cmd = { 0 };
+	u64 seq;
+	int ret;
+
+	cmd.opcode = SYNC_DEBUG_BO;
+	ret = amdxdna_cmd_submit(client, &cmd, AMDXDNA_INVALID_BO_HANDLE,
+				 &debug_bo_hdl, 1, hwctx->id, &seq);
+	if (ret) {
+		XDNA_ERR(xdna, "Submit command failed");
+		return ret;
+	}
+
+	aie2_cmd_wait(hwctx, seq);
+	if (cmd.result) {
+		XDNA_ERR(xdna, "Response failure 0x%x", cmd.result);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
 static int aie2_populate_range(struct amdxdna_gem_obj *abo)
 {
 	struct amdxdna_dev *xdna = to_xdna_dev(to_gobj(abo)->dev);
@@ -848,17 +984,22 @@ int aie2_cmd_submit(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job,
 		goto up_sem;
 	}
 
-	ret = drm_sched_job_init(&job->base, &hwctx->priv->entity, 1, hwctx);
+	ret = drm_sched_job_init(&job->base, &hwctx->priv->entity, 1, hwctx,
+				 hwctx->client->filp->client_id);
 	if (ret) {
 		XDNA_ERR(xdna, "DRM job init failed, ret %d", ret);
 		goto free_chain;
 	}
 
+	ret = amdxdna_pm_resume_get(xdna);
+	if (ret)
+		goto cleanup_job;
+
 retry:
 	ret = drm_gem_lock_reservations(job->bos, job->bo_cnt, &acquire_ctx);
 	if (ret) {
 		XDNA_WARN(xdna, "Failed to lock BOs, ret %d", ret);
-		goto cleanup_job;
+		goto suspend_put;
 	}
 
 	for (i = 0; i < job->bo_cnt; i++) {
@@ -866,7 +1007,7 @@ retry:
 		if (ret) {
 			XDNA_WARN(xdna, "Failed to reserve fences %d", ret);
 			drm_gem_unlock_reservations(job->bos, job->bo_cnt, &acquire_ctx);
-			goto cleanup_job;
+			goto suspend_put;
 		}
 	}
 
@@ -881,12 +1022,12 @@ retry:
 					msecs_to_jiffies(HMM_RANGE_DEFAULT_TIMEOUT);
 			} else if (time_after(jiffies, timeout)) {
 				ret = -ETIME;
-				goto cleanup_job;
+				goto suspend_put;
 			}
 
 			ret = aie2_populate_range(abo);
 			if (ret)
-				goto cleanup_job;
+				goto suspend_put;
 			goto retry;
 		}
 	}
@@ -912,6 +1053,8 @@ retry:
 
 	return 0;
 
+suspend_put:
+	amdxdna_pm_suspend_put(xdna);
 cleanup_job:
 	drm_sched_job_cleanup(&job->base);
 free_chain:
diff --git a/drivers/accel/amdxdna/aie2_error.c b/drivers/accel/amdxdna/aie2_error.c
index 5ee905632a39..d452008ec4f4 100644
--- a/drivers/accel/amdxdna/aie2_error.c
+++ b/drivers/accel/amdxdna/aie2_error.c
@@ -13,6 +13,7 @@
 
 #include "aie2_msg_priv.h"
 #include "aie2_pci.h"
+#include "amdxdna_error.h"
 #include "amdxdna_mailbox.h"
 #include "amdxdna_pci_drv.h"
 
@@ -46,6 +47,7 @@ enum aie_module_type {
 	AIE_MEM_MOD = 0,
 	AIE_CORE_MOD,
 	AIE_PL_MOD,
+	AIE_UNKNOWN_MOD,
 };
 
 enum aie_error_category {
@@ -143,6 +145,31 @@ static const struct aie_event_category aie_ml_shim_tile_event_cat[] = {
 	EVENT_CATEGORY(74U, AIE_ERROR_LOCK),
 };
 
+static const enum amdxdna_error_num aie_cat_err_num_map[] = {
+	[AIE_ERROR_SATURATION] = AMDXDNA_ERROR_NUM_AIE_SATURATION,
+	[AIE_ERROR_FP] = AMDXDNA_ERROR_NUM_AIE_FP,
+	[AIE_ERROR_STREAM] = AMDXDNA_ERROR_NUM_AIE_STREAM,
+	[AIE_ERROR_ACCESS] = AMDXDNA_ERROR_NUM_AIE_ACCESS,
+	[AIE_ERROR_BUS] = AMDXDNA_ERROR_NUM_AIE_BUS,
+	[AIE_ERROR_INSTRUCTION] = AMDXDNA_ERROR_NUM_AIE_INSTRUCTION,
+	[AIE_ERROR_ECC] = AMDXDNA_ERROR_NUM_AIE_ECC,
+	[AIE_ERROR_LOCK] = AMDXDNA_ERROR_NUM_AIE_LOCK,
+	[AIE_ERROR_DMA] = AMDXDNA_ERROR_NUM_AIE_DMA,
+	[AIE_ERROR_MEM_PARITY] = AMDXDNA_ERROR_NUM_AIE_MEM_PARITY,
+	[AIE_ERROR_UNKNOWN] = AMDXDNA_ERROR_NUM_UNKNOWN,
+};
+
+static_assert(ARRAY_SIZE(aie_cat_err_num_map) == AIE_ERROR_UNKNOWN + 1);
+
+static const enum amdxdna_error_module aie_err_mod_map[] = {
+	[AIE_MEM_MOD] = AMDXDNA_ERROR_MODULE_AIE_MEMORY,
+	[AIE_CORE_MOD] = AMDXDNA_ERROR_MODULE_AIE_CORE,
+	[AIE_PL_MOD] = AMDXDNA_ERROR_MODULE_AIE_PL,
+	[AIE_UNKNOWN_MOD] = AMDXDNA_ERROR_MODULE_UNKNOWN,
+};
+
+static_assert(ARRAY_SIZE(aie_err_mod_map) == AIE_UNKNOWN_MOD + 1);
+
 static enum aie_error_category
 aie_get_error_category(u8 row, u8 event_id, enum aie_module_type mod_type)
 {
@@ -176,12 +203,40 @@ aie_get_error_category(u8 row, u8 event_id, enum aie_module_type mod_type)
 		if (event_id != lut[i].event_id)
 			continue;
 
+		if (lut[i].category > AIE_ERROR_UNKNOWN)
+			return AIE_ERROR_UNKNOWN;
+
 		return lut[i].category;
 	}
 
 	return AIE_ERROR_UNKNOWN;
 }
 
+static void aie2_update_last_async_error(struct amdxdna_dev_hdl *ndev, void *err_info, u32 num_err)
+{
+	struct aie_error *errs = err_info;
+	enum amdxdna_error_module err_mod;
+	enum aie_error_category aie_err;
+	enum amdxdna_error_num err_num;
+	struct aie_error *last_err;
+
+	last_err = &errs[num_err - 1];
+	if (last_err->mod_type >= AIE_UNKNOWN_MOD) {
+		err_num = aie_cat_err_num_map[AIE_ERROR_UNKNOWN];
+		err_mod = aie_err_mod_map[AIE_UNKNOWN_MOD];
+	} else {
+		aie_err = aie_get_error_category(last_err->row,
+						 last_err->event_id,
+						 last_err->mod_type);
+		err_num = aie_cat_err_num_map[aie_err];
+		err_mod = aie_err_mod_map[last_err->mod_type];
+	}
+
+	ndev->last_async_err.err_code = AMDXDNA_ERROR_ENCODE(err_num, err_mod);
+	ndev->last_async_err.ts_us = ktime_to_us(ktime_get_real());
+	ndev->last_async_err.ex_err_code = AMDXDNA_EXTRA_ERR_ENCODE(last_err->row, last_err->col);
+}
+
 static u32 aie2_error_backtrack(struct amdxdna_dev_hdl *ndev, void *err_info, u32 num_err)
 {
 	struct aie_error *errs = err_info;
@@ -264,29 +319,14 @@ static void aie2_error_worker(struct work_struct *err_work)
 	}
 
 	mutex_lock(&xdna->dev_lock);
+	aie2_update_last_async_error(e->ndev, info->payload, info->err_cnt);
+
 	/* Re-sent this event to firmware */
 	if (aie2_error_event_send(e))
 		XDNA_WARN(xdna, "Unable to register async event");
 	mutex_unlock(&xdna->dev_lock);
 }
 
-int aie2_error_async_events_send(struct amdxdna_dev_hdl *ndev)
-{
-	struct amdxdna_dev *xdna = ndev->xdna;
-	struct async_event *e;
-	int i, ret;
-
-	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
-	for (i = 0; i < ndev->async_events->event_cnt; i++) {
-		e = &ndev->async_events->event[i];
-		ret = aie2_error_event_send(e);
-		if (ret)
-			return ret;
-	}
-
-	return 0;
-}
-
 void aie2_error_async_events_free(struct amdxdna_dev_hdl *ndev)
 {
 	struct amdxdna_dev *xdna = ndev->xdna;
@@ -341,6 +381,10 @@ int aie2_error_async_events_alloc(struct amdxdna_dev_hdl *ndev)
 		e->size = ASYNC_BUF_SIZE;
 		e->resp.status = MAX_AIE2_STATUS_CODE;
 		INIT_WORK(&e->work, aie2_error_worker);
+
+		ret = aie2_error_event_send(e);
+		if (ret)
+			goto free_wq;
 	}
 
 	ndev->async_events = events;
@@ -349,6 +393,8 @@ int aie2_error_async_events_alloc(struct amdxdna_dev_hdl *ndev)
 		 events->event_cnt, events->size);
 	return 0;
 
+free_wq:
+	destroy_workqueue(events->wq);
 free_buf:
 	dma_free_noncoherent(xdna->ddev.dev, events->size, events->buf,
 			     events->addr, DMA_FROM_DEVICE);
@@ -356,3 +402,18 @@ free_events:
 	kfree(events);
 	return ret;
 }
+
+int aie2_get_array_async_error(struct amdxdna_dev_hdl *ndev, struct amdxdna_drm_get_array *args)
+{
+	struct amdxdna_dev *xdna = ndev->xdna;
+
+	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
+
+	args->num_element = 1;
+	args->element_size = sizeof(ndev->last_async_err);
+	if (copy_to_user(u64_to_user_ptr(args->buffer),
+			 &ndev->last_async_err, args->element_size))
+		return -EFAULT;
+
+	return 0;
+}
diff --git a/drivers/accel/amdxdna/aie2_message.c b/drivers/accel/amdxdna/aie2_message.c
index 82412eec9a4b..d493bb1c3360 100644
--- a/drivers/accel/amdxdna/aie2_message.c
+++ b/drivers/accel/amdxdna/aie2_message.c
@@ -27,6 +27,8 @@
 #define DECLARE_AIE2_MSG(name, op) \
 	DECLARE_XDNA_MSG_COMMON(name, op, MAX_AIE2_STATUS_CODE)
 
+#define EXEC_MSG_OPS(xdna)	((xdna)->dev_handle->exec_msg_ops)
+
 static int aie2_send_mgmt_msg_wait(struct amdxdna_dev_hdl *ndev,
 				   struct xdna_mailbox_msg *msg)
 {
@@ -37,7 +39,7 @@ static int aie2_send_mgmt_msg_wait(struct amdxdna_dev_hdl *ndev,
 	if (!ndev->mgmt_chann)
 		return -ENODEV;
 
-	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
+	drm_WARN_ON(&xdna->ddev, xdna->rpm_on && !mutex_is_locked(&xdna->dev_lock));
 	ret = xdna_send_msg_wait(xdna, ndev->mgmt_chann, msg);
 	if (ret == -ETIME) {
 		xdna_mailbox_stop_channel(ndev->mgmt_chann);
@@ -45,7 +47,7 @@ static int aie2_send_mgmt_msg_wait(struct amdxdna_dev_hdl *ndev,
 		ndev->mgmt_chann = NULL;
 	}
 
-	if (!ret && *hdl->data != AIE2_STATUS_SUCCESS) {
+	if (!ret && *hdl->status != AIE2_STATUS_SUCCESS) {
 		XDNA_ERR(xdna, "command opcode 0x%x failed, status 0x%x",
 			 msg->opcode, *hdl->data);
 		ret = -EINVAL;
@@ -208,6 +210,14 @@ int aie2_create_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwct
 	hwctx->fw_ctx_id = resp.context_id;
 	WARN_ONCE(hwctx->fw_ctx_id == -1, "Unexpected context id");
 
+	if (ndev->force_preempt_enabled) {
+		ret = aie2_runtime_cfg(ndev, AIE2_RT_CFG_FORCE_PREEMPT, &hwctx->fw_ctx_id);
+		if (ret) {
+			XDNA_ERR(xdna, "failed to enable force preempt %d", ret);
+			return ret;
+		}
+	}
+
 	cq_pair = &resp.cq_pair[0];
 	x2i.mb_head_ptr_reg = AIE2_MBOX_OFF(ndev, cq_pair->x2i_q.head_addr);
 	x2i.mb_tail_ptr_reg = AIE2_MBOX_OFF(ndev, cq_pair->x2i_q.tail_addr);
@@ -233,6 +243,7 @@ int aie2_create_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwct
 		ret = -EINVAL;
 		goto out_destroy_context;
 	}
+	ndev->hwctx_num++;
 
 	XDNA_DBG(xdna, "%s mailbox channel irq: %d, msix_id: %d",
 		 hwctx->name, ret, resp.msix_id);
@@ -267,6 +278,7 @@ int aie2_destroy_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwc
 		 hwctx->fw_ctx_id);
 	hwctx->priv->mbox_chann = NULL;
 	hwctx->fw_ctx_id = -1;
+	ndev->hwctx_num--;
 
 	return ret;
 }
@@ -290,18 +302,25 @@ int aie2_map_host_buf(struct amdxdna_dev_hdl *ndev, u32 context_id, u64 addr, u6
 	return 0;
 }
 
+static int amdxdna_hwctx_col_map(struct amdxdna_hwctx *hwctx, void *arg)
+{
+	u32 *bitmap = arg;
+
+	*bitmap |= GENMASK(hwctx->start_col + hwctx->num_col - 1, hwctx->start_col);
+
+	return 0;
+}
+
 int aie2_query_status(struct amdxdna_dev_hdl *ndev, char __user *buf,
 		      u32 size, u32 *cols_filled)
 {
 	DECLARE_AIE2_MSG(aie_column_info, MSG_OP_QUERY_COL_STATUS);
 	struct amdxdna_dev *xdna = ndev->xdna;
 	struct amdxdna_client *client;
-	struct amdxdna_hwctx *hwctx;
-	unsigned long hwctx_id;
 	dma_addr_t dma_addr;
 	u32 aie_bitmap = 0;
 	u8 *buff_addr;
-	int ret, idx;
+	int ret;
 
 	buff_addr = dma_alloc_noncoherent(xdna->ddev.dev, size, &dma_addr,
 					  DMA_FROM_DEVICE, GFP_KERNEL);
@@ -309,12 +328,8 @@ int aie2_query_status(struct amdxdna_dev_hdl *ndev, char __user *buf,
 		return -ENOMEM;
 
 	/* Go through each hardware context and mark the AIE columns that are active */
-	list_for_each_entry(client, &xdna->client_list, node) {
-		idx = srcu_read_lock(&client->hwctx_srcu);
-		amdxdna_for_each_hwctx(client, hwctx_id, hwctx)
-			aie_bitmap |= amdxdna_hwctx_col_map(hwctx);
-		srcu_read_unlock(&client->hwctx_srcu, idx);
-	}
+	list_for_each_entry(client, &xdna->client_list, node)
+		amdxdna_hwctx_walk(client, &aie_bitmap, amdxdna_hwctx_col_map);
 
 	*cols_filled = 0;
 	req.dump_buff_addr = dma_addr;
@@ -329,11 +344,6 @@ int aie2_query_status(struct amdxdna_dev_hdl *ndev, char __user *buf,
 		goto fail;
 	}
 
-	if (resp.status != AIE2_STATUS_SUCCESS) {
-		XDNA_ERR(xdna, "Query NPU status failed, status 0x%x", resp.status);
-		ret = -EINVAL;
-		goto fail;
-	}
 	XDNA_DBG(xdna, "Query NPU status completed");
 
 	if (size < resp.size) {
@@ -355,6 +365,55 @@ fail:
 	return ret;
 }
 
+int aie2_query_telemetry(struct amdxdna_dev_hdl *ndev,
+			 char __user *buf, u32 size,
+			 struct amdxdna_drm_query_telemetry_header *header)
+{
+	DECLARE_AIE2_MSG(get_telemetry, MSG_OP_GET_TELEMETRY);
+	struct amdxdna_dev *xdna = ndev->xdna;
+	dma_addr_t dma_addr;
+	u8 *addr;
+	int ret;
+
+	if (header->type >= MAX_TELEMETRY_TYPE)
+		return -EINVAL;
+
+	addr = dma_alloc_noncoherent(xdna->ddev.dev, size, &dma_addr,
+				     DMA_FROM_DEVICE, GFP_KERNEL);
+	if (!addr)
+		return -ENOMEM;
+
+	req.buf_addr = dma_addr;
+	req.buf_size = size;
+	req.type = header->type;
+
+	drm_clflush_virt_range(addr, size); /* device can access */
+	ret = aie2_send_mgmt_msg_wait(ndev, &msg);
+	if (ret) {
+		XDNA_ERR(xdna, "Query telemetry failed, status %d", ret);
+		goto free_buf;
+	}
+
+	if (size < resp.size) {
+		ret = -EINVAL;
+		XDNA_ERR(xdna, "Bad buffer size. Available: %u. Needs: %u", size, resp.size);
+		goto free_buf;
+	}
+
+	if (copy_to_user(buf, addr, resp.size)) {
+		ret = -EFAULT;
+		XDNA_ERR(xdna, "Failed to copy telemetry to user space");
+		goto free_buf;
+	}
+
+	header->major = resp.major;
+	header->minor = resp.minor;
+
+free_buf:
+	dma_free_noncoherent(xdna->ddev.dev, size, addr, dma_addr, DMA_FROM_DEVICE);
+	return ret;
+}
+
 int aie2_register_asyn_event_msg(struct amdxdna_dev_hdl *ndev, dma_addr_t addr, u32 size,
 				 void *handle, int (*cb)(void*, void __iomem *, size_t))
 {
@@ -374,15 +433,17 @@ int aie2_register_asyn_event_msg(struct amdxdna_dev_hdl *ndev, dma_addr_t addr,
 	return xdna_mailbox_send_msg(ndev->mgmt_chann, &msg, TX_TIMEOUT);
 }
 
-int aie2_config_cu(struct amdxdna_hwctx *hwctx)
+int aie2_config_cu(struct amdxdna_hwctx *hwctx,
+		   int (*notify_cb)(void *, void __iomem *, size_t))
 {
 	struct mailbox_channel *chann = hwctx->priv->mbox_chann;
 	struct amdxdna_dev *xdna = hwctx->client->xdna;
 	u32 shift = xdna->dev_info->dev_mem_buf_shift;
-	DECLARE_AIE2_MSG(config_cu, MSG_OP_CONFIG_CU);
+	struct config_cu_req req = { 0 };
+	struct xdna_mailbox_msg msg;
 	struct drm_gem_object *gobj;
 	struct amdxdna_gem_obj *abo;
-	int ret, i;
+	int i;
 
 	if (!chann)
 		return -ENODEV;
@@ -420,191 +481,386 @@ int aie2_config_cu(struct amdxdna_hwctx *hwctx)
 	}
 	req.num_cus = hwctx->cus->num_cus;
 
-	ret = xdna_send_msg_wait(xdna, chann, &msg);
-	if (ret == -ETIME)
-		aie2_destroy_context(xdna->dev_handle, hwctx);
+	msg.send_data = (u8 *)&req;
+	msg.send_size = sizeof(req);
+	msg.handle = hwctx;
+	msg.opcode = MSG_OP_CONFIG_CU;
+	msg.notify_cb = notify_cb;
+	return xdna_mailbox_send_msg(chann, &msg, TX_TIMEOUT);
+}
 
-	if (resp.status == AIE2_STATUS_SUCCESS) {
-		XDNA_DBG(xdna, "Configure %d CUs, ret %d", req.num_cus, ret);
-		return 0;
-	}
+static int aie2_init_exec_cu_req(struct amdxdna_gem_obj *cmd_bo, void *req,
+				 size_t *size, u32 *msg_op)
+{
+	struct execute_buffer_req *cu_req = req;
+	u32 cmd_len;
+	void *cmd;
 
-	XDNA_ERR(xdna, "Command opcode 0x%x failed, status 0x%x ret %d",
-		 msg.opcode, resp.status, ret);
-	return ret;
+	cmd = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
+	if (cmd_len > sizeof(cu_req->payload))
+		return -EINVAL;
+
+	cu_req->cu_idx = amdxdna_cmd_get_cu_idx(cmd_bo);
+	if (cu_req->cu_idx == INVALID_CU_IDX)
+		return -EINVAL;
+
+	memcpy(cu_req->payload, cmd, cmd_len);
+
+	*size = sizeof(*cu_req);
+	*msg_op = MSG_OP_EXECUTE_BUFFER_CF;
+	return 0;
 }
 
-int aie2_execbuf(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job,
-		 int (*notify_cb)(void *, void __iomem *, size_t))
+static int aie2_init_exec_dpu_req(struct amdxdna_gem_obj *cmd_bo, void *req,
+				  size_t *size, u32 *msg_op)
 {
-	struct mailbox_channel *chann = hwctx->priv->mbox_chann;
-	struct amdxdna_dev *xdna = hwctx->client->xdna;
-	struct amdxdna_gem_obj *cmd_abo = job->cmd_bo;
-	union {
-		struct execute_buffer_req ebuf;
-		struct exec_dpu_req dpu;
-	} req;
-	struct xdna_mailbox_msg msg;
-	u32 payload_len;
-	void *payload;
-	int cu_idx;
-	int ret;
-	u32 op;
+	struct exec_dpu_req *dpu_req = req;
+	struct amdxdna_cmd_start_npu *sn;
+	u32 cmd_len;
 
-	if (!chann)
-		return -ENODEV;
+	sn = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
+	if (cmd_len - sizeof(*sn) > sizeof(dpu_req->payload))
+		return -EINVAL;
 
-	payload = amdxdna_cmd_get_payload(cmd_abo, &payload_len);
-	if (!payload) {
-		XDNA_ERR(xdna, "Invalid command, cannot get payload");
+	dpu_req->cu_idx = amdxdna_cmd_get_cu_idx(cmd_bo);
+	if (dpu_req->cu_idx == INVALID_CU_IDX)
 		return -EINVAL;
-	}
 
-	cu_idx = amdxdna_cmd_get_cu_idx(cmd_abo);
-	if (cu_idx < 0) {
-		XDNA_DBG(xdna, "Invalid cu idx");
+	dpu_req->inst_buf_addr = sn->buffer;
+	dpu_req->inst_size = sn->buffer_size;
+	dpu_req->inst_prop_cnt = sn->prop_count;
+	memcpy(dpu_req->payload, sn->prop_args, cmd_len - sizeof(*sn));
+
+	*size = sizeof(*dpu_req);
+	*msg_op = MSG_OP_EXEC_DPU;
+	return 0;
+}
+
+static void aie2_init_exec_chain_req(void *req, u64 slot_addr, size_t size, u32 cmd_cnt)
+{
+	struct cmd_chain_req *chain_req = req;
+
+	chain_req->buf_addr = slot_addr;
+	chain_req->buf_size = size;
+	chain_req->count = cmd_cnt;
+}
+
+static void aie2_init_npu_chain_req(void *req, u64 slot_addr, size_t size, u32 cmd_cnt)
+{
+	struct cmd_chain_npu_req *npu_chain_req = req;
+
+	npu_chain_req->flags = 0;
+	npu_chain_req->reserved = 0;
+	npu_chain_req->buf_addr = slot_addr;
+	npu_chain_req->buf_size = size;
+	npu_chain_req->count = cmd_cnt;
+}
+
+static int
+aie2_cmdlist_fill_cf(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size)
+{
+	struct cmd_chain_slot_execbuf_cf *cf_slot = slot;
+	u32 cmd_len;
+	void *cmd;
+
+	cmd = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
+	if (*size < sizeof(*cf_slot) + cmd_len)
 		return -EINVAL;
-	}
 
-	op = amdxdna_cmd_get_op(cmd_abo);
-	switch (op) {
+	cf_slot->cu_idx = amdxdna_cmd_get_cu_idx(cmd_bo);
+	if (cf_slot->cu_idx == INVALID_CU_IDX)
+		return -EINVAL;
+
+	cf_slot->arg_cnt = cmd_len / sizeof(u32);
+	memcpy(cf_slot->args, cmd, cmd_len);
+	/* Accurate slot size to hint firmware to do necessary copy */
+	*size = sizeof(*cf_slot) + cmd_len;
+	return 0;
+}
+
+static int
+aie2_cmdlist_fill_dpu(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size)
+{
+	struct cmd_chain_slot_dpu *dpu_slot = slot;
+	struct amdxdna_cmd_start_npu *sn;
+	u32 cmd_len;
+	u32 arg_sz;
+
+	sn = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
+	arg_sz = cmd_len - sizeof(*sn);
+	if (cmd_len < sizeof(*sn) || arg_sz > MAX_DPU_ARGS_SIZE)
+		return -EINVAL;
+
+	if (*size < sizeof(*dpu_slot) + arg_sz)
+		return -EINVAL;
+
+	dpu_slot->cu_idx = amdxdna_cmd_get_cu_idx(cmd_bo);
+	if (dpu_slot->cu_idx == INVALID_CU_IDX)
+		return -EINVAL;
+
+	dpu_slot->inst_buf_addr = sn->buffer;
+	dpu_slot->inst_size = sn->buffer_size;
+	dpu_slot->inst_prop_cnt = sn->prop_count;
+	dpu_slot->arg_cnt = arg_sz / sizeof(u32);
+	memcpy(dpu_slot->args, sn->prop_args, arg_sz);
+
+	/* Accurate slot size to hint firmware to do necessary copy */
+	*size = sizeof(*dpu_slot) + arg_sz;
+	return 0;
+}
+
+static int aie2_cmdlist_unsupp(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size)
+{
+	return -EOPNOTSUPP;
+}
+
+static u32 aie2_get_chain_msg_op(u32 cmd_op)
+{
+	switch (cmd_op) {
 	case ERT_START_CU:
-		if (unlikely(payload_len > sizeof(req.ebuf.payload)))
-			XDNA_DBG(xdna, "Invalid ebuf payload len: %d", payload_len);
-		req.ebuf.cu_idx = cu_idx;
-		memcpy(req.ebuf.payload, payload, sizeof(req.ebuf.payload));
-		msg.send_size = sizeof(req.ebuf);
-		msg.opcode = MSG_OP_EXECUTE_BUFFER_CF;
-		break;
-	case ERT_START_NPU: {
-		struct amdxdna_cmd_start_npu *sn = payload;
-
-		if (unlikely(payload_len - sizeof(*sn) > sizeof(req.dpu.payload)))
-			XDNA_DBG(xdna, "Invalid dpu payload len: %d", payload_len);
-		req.dpu.inst_buf_addr = sn->buffer;
-		req.dpu.inst_size = sn->buffer_size;
-		req.dpu.inst_prop_cnt = sn->prop_count;
-		req.dpu.cu_idx = cu_idx;
-		memcpy(req.dpu.payload, sn->prop_args, sizeof(req.dpu.payload));
-		msg.send_size = sizeof(req.dpu);
-		msg.opcode = MSG_OP_EXEC_DPU;
+		return MSG_OP_CHAIN_EXEC_BUFFER_CF;
+	case ERT_START_NPU:
+		return MSG_OP_CHAIN_EXEC_DPU;
+	default:
 		break;
 	}
-	default:
-		XDNA_DBG(xdna, "Invalid ERT cmd op code: %d", op);
+
+	return MSG_OP_MAX_OPCODE;
+}
+
+static struct aie2_exec_msg_ops legacy_exec_message_ops = {
+	.init_cu_req = aie2_init_exec_cu_req,
+	.init_dpu_req = aie2_init_exec_dpu_req,
+	.init_chain_req = aie2_init_exec_chain_req,
+	.fill_cf_slot = aie2_cmdlist_fill_cf,
+	.fill_dpu_slot = aie2_cmdlist_fill_dpu,
+	.fill_preempt_slot = aie2_cmdlist_unsupp,
+	.fill_elf_slot = aie2_cmdlist_unsupp,
+	.get_chain_msg_op = aie2_get_chain_msg_op,
+};
+
+static int
+aie2_cmdlist_fill_npu_cf(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size)
+{
+	struct cmd_chain_slot_npu *npu_slot = slot;
+	u32 cmd_len;
+	void *cmd;
+
+	cmd = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
+	if (*size < sizeof(*npu_slot) + cmd_len)
 		return -EINVAL;
-	}
-	msg.handle = job;
-	msg.notify_cb = notify_cb;
-	msg.send_data = (u8 *)&req;
-	print_hex_dump_debug("cmd: ", DUMP_PREFIX_OFFSET, 16, 4, &req,
-			     0x40, false);
 
-	ret = xdna_mailbox_send_msg(chann, &msg, TX_TIMEOUT);
-	if (ret) {
-		XDNA_ERR(xdna, "Send message failed");
-		return ret;
-	}
+	npu_slot->cu_idx = amdxdna_cmd_get_cu_idx(cmd_bo);
+	if (npu_slot->cu_idx == INVALID_CU_IDX)
+		return -EINVAL;
+
+	memset(npu_slot, 0, sizeof(*npu_slot));
+	npu_slot->type = EXEC_NPU_TYPE_NON_ELF;
+	npu_slot->arg_cnt = cmd_len / sizeof(u32);
+	memcpy(npu_slot->args, cmd, cmd_len);
 
+	*size = sizeof(*npu_slot) + cmd_len;
 	return 0;
 }
 
 static int
-aie2_cmdlist_fill_one_slot_cf(void *cmd_buf, u32 offset,
-			      struct amdxdna_gem_obj *abo, u32 *size)
+aie2_cmdlist_fill_npu_dpu(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size)
 {
-	struct cmd_chain_slot_execbuf_cf *buf = cmd_buf + offset;
-	int cu_idx = amdxdna_cmd_get_cu_idx(abo);
-	u32 payload_len;
-	void *payload;
+	struct cmd_chain_slot_npu *npu_slot = slot;
+	struct amdxdna_cmd_start_npu *sn;
+	u32 cmd_len;
+	u32 arg_sz;
+
+	sn = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
+	arg_sz = cmd_len - sizeof(*sn);
+	if (cmd_len < sizeof(*sn) || arg_sz > MAX_NPU_ARGS_SIZE)
+		return -EINVAL;
 
-	if (cu_idx < 0)
+	if (*size < sizeof(*npu_slot) + arg_sz)
 		return -EINVAL;
 
-	payload = amdxdna_cmd_get_payload(abo, &payload_len);
-	if (!payload)
+	npu_slot->cu_idx = amdxdna_cmd_get_cu_idx(cmd_bo);
+	if (npu_slot->cu_idx == INVALID_CU_IDX)
 		return -EINVAL;
 
-	if (!slot_has_space(*buf, offset, payload_len))
-		return -ENOSPC;
+	memset(npu_slot, 0, sizeof(*npu_slot));
+	npu_slot->type = EXEC_NPU_TYPE_PARTIAL_ELF;
+	npu_slot->inst_buf_addr = sn->buffer;
+	npu_slot->inst_size = sn->buffer_size;
+	npu_slot->inst_prop_cnt = sn->prop_count;
+	npu_slot->arg_cnt = arg_sz / sizeof(u32);
+	memcpy(npu_slot->args, sn->prop_args, arg_sz);
 
-	buf->cu_idx = cu_idx;
-	buf->arg_cnt = payload_len / sizeof(u32);
-	memcpy(buf->args, payload, payload_len);
-	/* Accurate buf size to hint firmware to do necessary copy */
-	*size = sizeof(*buf) + payload_len;
+	*size = sizeof(*npu_slot) + arg_sz;
 	return 0;
 }
 
 static int
-aie2_cmdlist_fill_one_slot_dpu(void *cmd_buf, u32 offset,
-			       struct amdxdna_gem_obj *abo, u32 *size)
+aie2_cmdlist_fill_npu_preempt(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size)
 {
-	struct cmd_chain_slot_dpu *buf = cmd_buf + offset;
-	int cu_idx = amdxdna_cmd_get_cu_idx(abo);
-	struct amdxdna_cmd_start_npu *sn;
-	u32 payload_len;
-	void *payload;
+	struct cmd_chain_slot_npu *npu_slot = slot;
+	struct amdxdna_cmd_preempt_data *pd;
+	u32 cmd_len;
 	u32 arg_sz;
 
-	if (cu_idx < 0)
+	pd = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
+	arg_sz = cmd_len - sizeof(*pd);
+	if (cmd_len < sizeof(*pd) || arg_sz > MAX_NPU_ARGS_SIZE)
 		return -EINVAL;
 
-	payload = amdxdna_cmd_get_payload(abo, &payload_len);
-	if (!payload)
+	if (*size < sizeof(*npu_slot) + arg_sz)
 		return -EINVAL;
-	sn = payload;
-	arg_sz = payload_len - sizeof(*sn);
-	if (payload_len < sizeof(*sn) || arg_sz > MAX_DPU_ARGS_SIZE)
+
+	npu_slot->cu_idx = amdxdna_cmd_get_cu_idx(cmd_bo);
+	if (npu_slot->cu_idx == INVALID_CU_IDX)
 		return -EINVAL;
 
-	if (!slot_has_space(*buf, offset, arg_sz))
-		return -ENOSPC;
+	memset(npu_slot, 0, sizeof(*npu_slot));
+	npu_slot->type = EXEC_NPU_TYPE_PREEMPT;
+	npu_slot->inst_buf_addr = pd->inst_buf;
+	npu_slot->save_buf_addr = pd->save_buf;
+	npu_slot->restore_buf_addr = pd->restore_buf;
+	npu_slot->inst_size = pd->inst_size;
+	npu_slot->save_size = pd->save_size;
+	npu_slot->restore_size = pd->restore_size;
+	npu_slot->inst_prop_cnt = pd->inst_prop_cnt;
+	npu_slot->arg_cnt = arg_sz / sizeof(u32);
+	memcpy(npu_slot->args, pd->prop_args, arg_sz);
+
+	*size = sizeof(*npu_slot) + arg_sz;
+	return 0;
+}
+
+static int
+aie2_cmdlist_fill_npu_elf(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size)
+{
+	struct cmd_chain_slot_npu *npu_slot = slot;
+	struct amdxdna_cmd_preempt_data *pd;
+	u32 cmd_len;
+	u32 arg_sz;
+
+	pd = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
+	arg_sz = cmd_len - sizeof(*pd);
+	if (cmd_len < sizeof(*pd) || arg_sz > MAX_NPU_ARGS_SIZE)
+		return -EINVAL;
 
-	buf->inst_buf_addr = sn->buffer;
-	buf->inst_size = sn->buffer_size;
-	buf->inst_prop_cnt = sn->prop_count;
-	buf->cu_idx = cu_idx;
-	buf->arg_cnt = arg_sz / sizeof(u32);
-	memcpy(buf->args, sn->prop_args, arg_sz);
+	if (*size < sizeof(*npu_slot) + arg_sz)
+		return -EINVAL;
 
-	/* Accurate buf size to hint firmware to do necessary copy */
-	*size = sizeof(*buf) + arg_sz;
+	memset(npu_slot, 0, sizeof(*npu_slot));
+	npu_slot->type = EXEC_NPU_TYPE_ELF;
+	npu_slot->inst_buf_addr = pd->inst_buf;
+	npu_slot->save_buf_addr = pd->save_buf;
+	npu_slot->restore_buf_addr = pd->restore_buf;
+	npu_slot->inst_size = pd->inst_size;
+	npu_slot->save_size = pd->save_size;
+	npu_slot->restore_size = pd->restore_size;
+	npu_slot->inst_prop_cnt = pd->inst_prop_cnt;
+	npu_slot->arg_cnt = 1;
+	npu_slot->args[0] = AIE2_EXEC_BUFFER_KERNEL_OP_TXN;
+
+	*size = struct_size(npu_slot, args, npu_slot->arg_cnt);
 	return 0;
 }
 
-static int
-aie2_cmdlist_fill_one_slot(u32 op, struct amdxdna_gem_obj *cmdbuf_abo, u32 offset,
-			   struct amdxdna_gem_obj *abo, u32 *size)
+static u32 aie2_get_npu_chain_msg_op(u32 cmd_op)
+{
+	return MSG_OP_CHAIN_EXEC_NPU;
+}
+
+static struct aie2_exec_msg_ops npu_exec_message_ops = {
+	.init_cu_req = aie2_init_exec_cu_req,
+	.init_dpu_req = aie2_init_exec_dpu_req,
+	.init_chain_req = aie2_init_npu_chain_req,
+	.fill_cf_slot = aie2_cmdlist_fill_npu_cf,
+	.fill_dpu_slot = aie2_cmdlist_fill_npu_dpu,
+	.fill_preempt_slot = aie2_cmdlist_fill_npu_preempt,
+	.fill_elf_slot = aie2_cmdlist_fill_npu_elf,
+	.get_chain_msg_op = aie2_get_npu_chain_msg_op,
+};
+
+static int aie2_init_exec_req(void *req, struct amdxdna_gem_obj *cmd_abo,
+			      size_t *size, u32 *msg_op)
 {
-	u32 this_op = amdxdna_cmd_get_op(abo);
-	void *cmd_buf = cmdbuf_abo->mem.kva;
+	struct amdxdna_dev *xdna = cmd_abo->client->xdna;
 	int ret;
+	u32 op;
 
-	if (this_op != op) {
-		ret = -EINVAL;
-		goto done;
-	}
 
+	op = amdxdna_cmd_get_op(cmd_abo);
 	switch (op) {
 	case ERT_START_CU:
-		ret = aie2_cmdlist_fill_one_slot_cf(cmd_buf, offset, abo, size);
+		ret = EXEC_MSG_OPS(xdna)->init_cu_req(cmd_abo, req, size, msg_op);
+		if (ret) {
+			XDNA_DBG(xdna, "Init CU req failed ret %d", ret);
+			return ret;
+		}
 		break;
 	case ERT_START_NPU:
-		ret = aie2_cmdlist_fill_one_slot_dpu(cmd_buf, offset, abo, size);
+		ret = EXEC_MSG_OPS(xdna)->init_dpu_req(cmd_abo, req, size, msg_op);
+		if (ret) {
+			XDNA_DBG(xdna, "Init DPU req failed ret %d", ret);
+			return ret;
+		}
+
 		break;
 	default:
+		XDNA_ERR(xdna, "Unsupported op %d", op);
 		ret = -EOPNOTSUPP;
+		break;
 	}
 
-done:
-	if (ret) {
-		XDNA_ERR(abo->client->xdna, "Can't fill slot for cmd op %d ret %d",
-			 op, ret);
+	return ret;
+}
+
+static int
+aie2_cmdlist_fill_slot(void *slot, struct amdxdna_gem_obj *cmd_abo,
+		       size_t *size, u32 *cmd_op)
+{
+	struct amdxdna_dev *xdna = cmd_abo->client->xdna;
+	int ret;
+	u32 op;
+
+	op = amdxdna_cmd_get_op(cmd_abo);
+	if (*cmd_op == ERT_INVALID_CMD)
+		*cmd_op = op;
+	else if (op != *cmd_op)
+		return -EINVAL;
+
+	switch (op) {
+	case ERT_START_CU:
+		ret = EXEC_MSG_OPS(xdna)->fill_cf_slot(cmd_abo, slot, size);
+		break;
+	case ERT_START_NPU:
+		ret = EXEC_MSG_OPS(xdna)->fill_dpu_slot(cmd_abo, slot, size);
+		break;
+	case ERT_START_NPU_PREEMPT:
+		if (!AIE2_FEATURE_ON(xdna->dev_handle, AIE2_PREEMPT))
+			return -EOPNOTSUPP;
+		ret = EXEC_MSG_OPS(xdna)->fill_preempt_slot(cmd_abo, slot, size);
+		break;
+	case ERT_START_NPU_PREEMPT_ELF:
+		if (!AIE2_FEATURE_ON(xdna->dev_handle, AIE2_PREEMPT))
+			return -EOPNOTSUPP;
+		ret = EXEC_MSG_OPS(xdna)->fill_elf_slot(cmd_abo, slot, size);
+		break;
+	default:
+		XDNA_INFO(xdna, "Unsupported op %d", op);
+		ret = -EOPNOTSUPP;
+		break;
 	}
+
 	return ret;
 }
 
+void aie2_msg_init(struct amdxdna_dev_hdl *ndev)
+{
+	if (AIE2_FEATURE_ON(ndev, AIE2_NPU_COMMAND))
+		ndev->exec_msg_ops = &npu_exec_message_ops;
+	else
+		ndev->exec_msg_ops = &legacy_exec_message_ops;
+}
+
 static inline struct amdxdna_gem_obj *
 aie2_cmdlist_get_cmd_buf(struct amdxdna_sched_job *job)
 {
@@ -613,29 +869,36 @@ aie2_cmdlist_get_cmd_buf(struct amdxdna_sched_job *job)
 	return job->hwctx->priv->cmd_buf[idx];
 }
 
-static void
-aie2_cmdlist_prepare_request(struct cmd_chain_req *req,
-			     struct amdxdna_gem_obj *cmdbuf_abo, u32 size, u32 cnt)
+int aie2_execbuf(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job,
+		 int (*notify_cb)(void *, void __iomem *, size_t))
 {
-	req->buf_addr = cmdbuf_abo->mem.dev_addr;
-	req->buf_size = size;
-	req->count = cnt;
-	drm_clflush_virt_range(cmdbuf_abo->mem.kva, size);
-	XDNA_DBG(cmdbuf_abo->client->xdna, "Command buf addr 0x%llx size 0x%x count %d",
-		 req->buf_addr, size, cnt);
-}
+	struct mailbox_channel *chann = hwctx->priv->mbox_chann;
+	struct amdxdna_dev *xdna = hwctx->client->xdna;
+	struct amdxdna_gem_obj *cmd_abo = job->cmd_bo;
+	struct xdna_mailbox_msg msg;
+	union exec_req req;
+	int ret;
 
-static inline u32
-aie2_cmd_op_to_msg_op(u32 op)
-{
-	switch (op) {
-	case ERT_START_CU:
-		return MSG_OP_CHAIN_EXEC_BUFFER_CF;
-	case ERT_START_NPU:
-		return MSG_OP_CHAIN_EXEC_DPU;
-	default:
-		return MSG_OP_MAX_OPCODE;
+	if (!chann)
+		return -ENODEV;
+
+	ret = aie2_init_exec_req(&req, cmd_abo, &msg.send_size, &msg.opcode);
+	if (ret)
+		return ret;
+
+	msg.handle = job;
+	msg.notify_cb = notify_cb;
+	msg.send_data = (u8 *)&req;
+	print_hex_dump_debug("cmd: ", DUMP_PREFIX_OFFSET, 16, 4, &req,
+			     0x40, false);
+
+	ret = xdna_mailbox_send_msg(chann, &msg, TX_TIMEOUT);
+	if (ret) {
+		XDNA_ERR(xdna, "Send message failed");
+		return ret;
 	}
+
+	return 0;
 }
 
 int aie2_cmdlist_multi_execbuf(struct amdxdna_hwctx *hwctx,
@@ -646,12 +909,13 @@ int aie2_cmdlist_multi_execbuf(struct amdxdna_hwctx *hwctx,
 	struct mailbox_channel *chann = hwctx->priv->mbox_chann;
 	struct amdxdna_client *client = hwctx->client;
 	struct amdxdna_gem_obj *cmd_abo = job->cmd_bo;
+	struct amdxdna_dev *xdna = client->xdna;
 	struct amdxdna_cmd_chain *payload;
 	struct xdna_mailbox_msg msg;
-	struct cmd_chain_req req;
+	union exec_chain_req req;
 	u32 payload_len;
 	u32 offset = 0;
-	u32 size;
+	size_t size;
 	int ret;
 	u32 op;
 	u32 i;
@@ -662,41 +926,42 @@ int aie2_cmdlist_multi_execbuf(struct amdxdna_hwctx *hwctx,
 	    payload_len < struct_size(payload, data, payload->command_count))
 		return -EINVAL;
 
+	op = ERT_INVALID_CMD;
 	for (i = 0; i < payload->command_count; i++) {
 		u32 boh = (u32)(payload->data[i]);
 		struct amdxdna_gem_obj *abo;
 
 		abo = amdxdna_gem_get_obj(client, boh, AMDXDNA_BO_CMD);
 		if (!abo) {
-			XDNA_ERR(client->xdna, "Failed to find cmd BO %d", boh);
+			XDNA_ERR(xdna, "Failed to find cmd BO %d", boh);
 			return -ENOENT;
 		}
 
-		/* All sub-cmd should have same op, use the first one. */
-		if (i == 0)
-			op = amdxdna_cmd_get_op(abo);
-
-		ret = aie2_cmdlist_fill_one_slot(op, cmdbuf_abo, offset, abo, &size);
+		size = cmdbuf_abo->mem.size - offset;
+		ret = aie2_cmdlist_fill_slot(cmdbuf_abo->mem.kva + offset,
+					     abo, &size, &op);
 		amdxdna_gem_put_obj(abo);
 		if (ret)
-			return -EINVAL;
+			return ret;
 
 		offset += size;
 	}
+	msg.opcode = EXEC_MSG_OPS(xdna)->get_chain_msg_op(op);
+	if (msg.opcode == MSG_OP_MAX_OPCODE)
+		return -EOPNOTSUPP;
 
 	/* The offset is the accumulated total size of the cmd buffer */
-	aie2_cmdlist_prepare_request(&req, cmdbuf_abo, offset, payload->command_count);
+	EXEC_MSG_OPS(xdna)->init_chain_req(&req, cmdbuf_abo->mem.dev_addr,
+					   offset, payload->command_count);
+	drm_clflush_virt_range(cmdbuf_abo->mem.kva, offset);
 
-	msg.opcode = aie2_cmd_op_to_msg_op(op);
-	if (msg.opcode == MSG_OP_MAX_OPCODE)
-		return -EOPNOTSUPP;
 	msg.handle = job;
 	msg.notify_cb = notify_cb;
 	msg.send_data = (u8 *)&req;
 	msg.send_size = sizeof(req);
 	ret = xdna_mailbox_send_msg(chann, &msg, TX_TIMEOUT);
 	if (ret) {
-		XDNA_ERR(hwctx->client->xdna, "Send message failed");
+		XDNA_ERR(xdna, "Send message failed");
 		return ret;
 	}
 
@@ -709,23 +974,27 @@ int aie2_cmdlist_single_execbuf(struct amdxdna_hwctx *hwctx,
 {
 	struct amdxdna_gem_obj *cmdbuf_abo = aie2_cmdlist_get_cmd_buf(job);
 	struct mailbox_channel *chann = hwctx->priv->mbox_chann;
+	struct amdxdna_dev *xdna = hwctx->client->xdna;
 	struct amdxdna_gem_obj *cmd_abo = job->cmd_bo;
 	struct xdna_mailbox_msg msg;
-	struct cmd_chain_req req;
-	u32 size;
+	union exec_chain_req req;
+	u32 op = ERT_INVALID_CMD;
+	size_t size;
 	int ret;
-	u32 op;
 
-	op = amdxdna_cmd_get_op(cmd_abo);
-	ret = aie2_cmdlist_fill_one_slot(op, cmdbuf_abo, 0, cmd_abo, &size);
+	size = cmdbuf_abo->mem.size;
+	ret = aie2_cmdlist_fill_slot(cmdbuf_abo->mem.kva, cmd_abo, &size, &op);
 	if (ret)
 		return ret;
 
-	aie2_cmdlist_prepare_request(&req, cmdbuf_abo, size, 1);
-
-	msg.opcode = aie2_cmd_op_to_msg_op(op);
+	msg.opcode = EXEC_MSG_OPS(xdna)->get_chain_msg_op(op);
 	if (msg.opcode == MSG_OP_MAX_OPCODE)
 		return -EOPNOTSUPP;
+
+	EXEC_MSG_OPS(xdna)->init_chain_req(&req, cmdbuf_abo->mem.dev_addr,
+					   size, 1);
+	drm_clflush_virt_range(cmdbuf_abo->mem.kva, size);
+
 	msg.handle = job;
 	msg.notify_cb = notify_cb;
 	msg.send_data = (u8 *)&req;
@@ -750,7 +1019,7 @@ int aie2_sync_bo(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job,
 	int ret = 0;
 
 	req.src_addr = 0;
-	req.dst_addr = abo->mem.dev_addr - hwctx->client->dev_heap->mem.dev_addr;
+	req.dst_addr = amdxdna_dev_bo_offset(abo);
 	req.size = abo->mem.size;
 
 	/* Device to Host */
@@ -774,3 +1043,32 @@ int aie2_sync_bo(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job,
 
 	return 0;
 }
+
+int aie2_config_debug_bo(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job,
+			 int (*notify_cb)(void *, void __iomem *, size_t))
+{
+	struct mailbox_channel *chann = hwctx->priv->mbox_chann;
+	struct amdxdna_gem_obj *abo = to_xdna_obj(job->bos[0]);
+	struct amdxdna_dev *xdna = hwctx->client->xdna;
+	struct config_debug_bo_req req;
+	struct xdna_mailbox_msg msg;
+
+	if (job->drv_cmd->opcode == ATTACH_DEBUG_BO)
+		req.config = DEBUG_BO_REGISTER;
+	else
+		req.config = DEBUG_BO_UNREGISTER;
+
+	req.offset = amdxdna_dev_bo_offset(abo);
+	req.size = abo->mem.size;
+
+	XDNA_DBG(xdna, "offset 0x%llx size 0x%llx config %d",
+		 req.offset, req.size, req.config);
+
+	msg.handle = job;
+	msg.notify_cb = notify_cb;
+	msg.send_data = (u8 *)&req;
+	msg.send_size = sizeof(req);
+	msg.opcode = MSG_OP_CONFIG_DEBUG_BO;
+
+	return xdna_mailbox_send_msg(chann, &msg, TX_TIMEOUT);
+}
diff --git a/drivers/accel/amdxdna/aie2_msg_priv.h b/drivers/accel/amdxdna/aie2_msg_priv.h
index 6df9065b13f6..1c957a6298d3 100644
--- a/drivers/accel/amdxdna/aie2_msg_priv.h
+++ b/drivers/accel/amdxdna/aie2_msg_priv.h
@@ -9,7 +9,8 @@
 enum aie2_msg_opcode {
 	MSG_OP_CREATE_CONTEXT              = 0x2,
 	MSG_OP_DESTROY_CONTEXT             = 0x3,
-	MSG_OP_SYNC_BO			   = 0x7,
+	MSG_OP_GET_TELEMETRY               = 0x4,
+	MSG_OP_SYNC_BO                     = 0x7,
 	MSG_OP_EXECUTE_BUFFER_CF           = 0xC,
 	MSG_OP_QUERY_COL_STATUS            = 0xD,
 	MSG_OP_QUERY_AIE_TILE_INFO         = 0xE,
@@ -18,6 +19,8 @@ enum aie2_msg_opcode {
 	MSG_OP_CONFIG_CU                   = 0x11,
 	MSG_OP_CHAIN_EXEC_BUFFER_CF        = 0x12,
 	MSG_OP_CHAIN_EXEC_DPU              = 0x13,
+	MSG_OP_CONFIG_DEBUG_BO             = 0x14,
+	MSG_OP_CHAIN_EXEC_NPU              = 0x18,
 	MSG_OP_MAX_XRT_OPCODE,
 	MSG_OP_SUSPEND                     = 0x101,
 	MSG_OP_RESUME                      = 0x102,
@@ -135,6 +138,28 @@ struct destroy_ctx_resp {
 	enum aie2_msg_status	status;
 } __packed;
 
+enum telemetry_type {
+	TELEMETRY_TYPE_DISABLED,
+	TELEMETRY_TYPE_HEALTH,
+	TELEMETRY_TYPE_ERROR_INFO,
+	TELEMETRY_TYPE_PROFILING,
+	TELEMETRY_TYPE_DEBUG,
+	MAX_TELEMETRY_TYPE
+};
+
+struct get_telemetry_req {
+	enum telemetry_type	type;
+	__u64	buf_addr;
+	__u32	buf_size;
+} __packed;
+
+struct get_telemetry_resp {
+	__u32	major;
+	__u32	minor;
+	__u32	size;
+	enum aie2_msg_status	status;
+} __packed;
+
 struct execute_buffer_req {
 	__u32	cu_idx;
 	__u32	payload[19];
@@ -148,6 +173,18 @@ struct exec_dpu_req {
 	__u32	payload[35];
 } __packed;
 
+enum exec_npu_type {
+	EXEC_NPU_TYPE_NON_ELF		= 0x1,
+	EXEC_NPU_TYPE_PARTIAL_ELF	= 0x2,
+	EXEC_NPU_TYPE_PREEMPT		= 0x3,
+	EXEC_NPU_TYPE_ELF		= 0x4,
+};
+
+union exec_req {
+	struct execute_buffer_req ebuf;
+	struct exec_dpu_req dpu_req;
+};
+
 struct execute_buffer_resp {
 	enum aie2_msg_status	status;
 } __packed;
@@ -319,9 +356,6 @@ struct async_event_msg_resp {
 } __packed;
 
 #define MAX_CHAIN_CMDBUF_SIZE SZ_4K
-#define slot_has_space(slot, offset, payload_size)		\
-	(MAX_CHAIN_CMDBUF_SIZE >= (offset) + (payload_size) +	\
-	 sizeof(typeof(slot)))
 
 struct cmd_chain_slot_execbuf_cf {
 	__u32 cu_idx;
@@ -339,12 +373,41 @@ struct cmd_chain_slot_dpu {
 	__u32 args[] __counted_by(arg_cnt);
 };
 
+#define MAX_NPU_ARGS_SIZE (26 * sizeof(__u32))
+#define AIE2_EXEC_BUFFER_KERNEL_OP_TXN	3
+struct cmd_chain_slot_npu {
+	enum exec_npu_type type;
+	u64 inst_buf_addr;
+	u64 save_buf_addr;
+	u64 restore_buf_addr;
+	u32 inst_size;
+	u32 save_size;
+	u32 restore_size;
+	u32 inst_prop_cnt;
+	u32 cu_idx;
+	u32 arg_cnt;
+	u32 args[] __counted_by(arg_cnt);
+} __packed;
+
 struct cmd_chain_req {
 	__u64 buf_addr;
 	__u32 buf_size;
 	__u32 count;
 } __packed;
 
+struct cmd_chain_npu_req {
+	u32 flags;
+	u32 reserved;
+	u64 buf_addr;
+	u32 buf_size;
+	u32 count;
+} __packed;
+
+union exec_chain_req {
+	struct cmd_chain_npu_req npu_req;
+	struct cmd_chain_req req;
+};
+
 struct cmd_chain_resp {
 	enum aie2_msg_status	status;
 	__u32			fail_cmd_idx;
@@ -365,4 +428,21 @@ struct sync_bo_req {
 struct sync_bo_resp {
 	enum aie2_msg_status	status;
 } __packed;
+
+#define DEBUG_BO_UNREGISTER 0
+#define DEBUG_BO_REGISTER   1
+struct config_debug_bo_req {
+	__u64	offset;
+	__u64	size;
+	/*
+	 * config operations.
+	 *   DEBUG_BO_REGISTER: Register debug buffer
+	 *   DEBUG_BO_UNREGISTER: Unregister debug buffer
+	 */
+	__u32	config;
+} __packed;
+
+struct config_debug_bo_resp {
+	enum aie2_msg_status	status;
+} __packed;
 #endif /* _AIE2_MSG_PRIV_H_ */
diff --git a/drivers/accel/amdxdna/aie2_pci.c b/drivers/accel/amdxdna/aie2_pci.c
index c6cf7068d23c..ceef1c502e9e 100644
--- a/drivers/accel/amdxdna/aie2_pci.c
+++ b/drivers/accel/amdxdna/aie2_pci.c
@@ -10,6 +10,7 @@
 #include <drm/drm_managed.h>
 #include <drm/drm_print.h>
 #include <drm/gpu_scheduler.h>
+#include <linux/cleanup.h>
 #include <linux/errno.h>
 #include <linux/firmware.h>
 #include <linux/iommu.h>
@@ -24,6 +25,7 @@
 #include "amdxdna_gem.h"
 #include "amdxdna_mailbox.h"
 #include "amdxdna_pci_drv.h"
+#include "amdxdna_pm.h"
 
 static int aie2_max_col = XRS_MAX_COL;
 module_param(aie2_max_col, uint, 0600);
@@ -53,6 +55,7 @@ struct mgmt_mbox_chann_info {
 
 static int aie2_check_protocol(struct amdxdna_dev_hdl *ndev, u32 fw_major, u32 fw_minor)
 {
+	const struct aie2_fw_feature_tbl *feature;
 	struct amdxdna_dev *xdna = ndev->xdna;
 
 	/*
@@ -76,6 +79,17 @@ static int aie2_check_protocol(struct amdxdna_dev_hdl *ndev, u32 fw_major, u32 f
 		XDNA_ERR(xdna, "Firmware minor version smaller than supported");
 		return -EINVAL;
 	}
+
+	for (feature = ndev->priv->fw_feature_tbl; feature && feature->min_minor;
+	     feature++) {
+		if (fw_minor < feature->min_minor)
+			continue;
+		if (feature->max_minor > 0 && fw_minor > feature->max_minor)
+			continue;
+
+		set_bit(feature->feature, &ndev->feature_mask);
+	}
+
 	return 0;
 }
 
@@ -169,6 +183,10 @@ int aie2_runtime_cfg(struct amdxdna_dev_hdl *ndev,
 		if (cfg->category != category)
 			continue;
 
+		if (cfg->feature_mask &&
+		    bitmap_subset(&cfg->feature_mask, &ndev->feature_mask, AIE2_FEATURE_MAX))
+			continue;
+
 		value = val ? *val : cfg->value;
 		ret = aie2_set_runtime_cfg(ndev, cfg->type, value);
 		if (ret) {
@@ -222,15 +240,6 @@ static int aie2_mgmt_fw_init(struct amdxdna_dev_hdl *ndev)
 		return ret;
 	}
 
-	if (!ndev->async_events)
-		return 0;
-
-	ret = aie2_error_async_events_send(ndev);
-	if (ret) {
-		XDNA_ERR(ndev->xdna, "Send async events failed");
-		return ret;
-	}
-
 	return 0;
 }
 
@@ -256,6 +265,8 @@ static int aie2_mgmt_fw_query(struct amdxdna_dev_hdl *ndev)
 		return ret;
 	}
 
+	ndev->total_col = min(aie2_max_col, ndev->metadata.cols);
+
 	return 0;
 }
 
@@ -337,6 +348,7 @@ static void aie2_hw_stop(struct amdxdna_dev *xdna)
 	ndev->mbox = NULL;
 	aie2_psp_stop(ndev->psp_hdl);
 	aie2_smu_fini(ndev);
+	aie2_error_async_events_free(ndev);
 	pci_disable_device(pdev);
 
 	ndev->dev_status = AIE2_DEV_INIT;
@@ -423,6 +435,18 @@ static int aie2_hw_start(struct amdxdna_dev *xdna)
 		goto destroy_mgmt_chann;
 	}
 
+	ret = aie2_mgmt_fw_query(ndev);
+	if (ret) {
+		XDNA_ERR(xdna, "failed to query fw, ret %d", ret);
+		goto destroy_mgmt_chann;
+	}
+
+	ret = aie2_error_async_events_alloc(ndev);
+	if (ret) {
+		XDNA_ERR(xdna, "Allocate async events failed, ret %d", ret);
+		goto destroy_mgmt_chann;
+	}
+
 	ndev->dev_status = AIE2_DEV_START;
 
 	return 0;
@@ -440,6 +464,39 @@ disable_dev:
 	return ret;
 }
 
+static int aie2_hw_suspend(struct amdxdna_dev *xdna)
+{
+	struct amdxdna_client *client;
+
+	guard(mutex)(&xdna->dev_lock);
+	list_for_each_entry(client, &xdna->client_list, node)
+		aie2_hwctx_suspend(client);
+
+	aie2_hw_stop(xdna);
+
+	return 0;
+}
+
+static int aie2_hw_resume(struct amdxdna_dev *xdna)
+{
+	struct amdxdna_client *client;
+	int ret;
+
+	ret = aie2_hw_start(xdna);
+	if (ret) {
+		XDNA_ERR(xdna, "Start hardware failed, %d", ret);
+		return ret;
+	}
+
+	list_for_each_entry(client, &xdna->client_list, node) {
+		ret = aie2_hwctx_resume(client);
+		if (ret)
+			break;
+	}
+
+	return ret;
+}
+
 static int aie2_init(struct amdxdna_dev *xdna)
 {
 	struct pci_dev *pdev = to_pci_dev(xdna->ddev.dev);
@@ -520,22 +577,15 @@ static int aie2_init(struct amdxdna_dev *xdna)
 	if (!ndev->psp_hdl) {
 		XDNA_ERR(xdna, "failed to create psp");
 		ret = -ENOMEM;
-		goto free_irq;
+		goto release_fw;
 	}
 	xdna->dev_handle = ndev;
 
 	ret = aie2_hw_start(xdna);
 	if (ret) {
 		XDNA_ERR(xdna, "start npu failed, ret %d", ret);
-		goto free_irq;
-	}
-
-	ret = aie2_mgmt_fw_query(ndev);
-	if (ret) {
-		XDNA_ERR(xdna, "Query firmware failed, ret %d", ret);
-		goto stop_hw;
+		goto release_fw;
 	}
-	ndev->total_col = min(aie2_max_col, ndev->metadata.cols);
 
 	xrs_cfg.clk_list.num_levels = ndev->max_dpm_level + 1;
 	for (i = 0; i < xrs_cfg.clk_list.num_levels; i++)
@@ -552,34 +602,13 @@ static int aie2_init(struct amdxdna_dev *xdna)
 		goto stop_hw;
 	}
 
-	ret = aie2_error_async_events_alloc(ndev);
-	if (ret) {
-		XDNA_ERR(xdna, "Allocate async events failed, ret %d", ret);
-		goto stop_hw;
-	}
-
-	ret = aie2_error_async_events_send(ndev);
-	if (ret) {
-		XDNA_ERR(xdna, "Send async events failed, ret %d", ret);
-		goto async_event_free;
-	}
-
-	/* Issue a command to make sure firmware handled async events */
-	ret = aie2_query_firmware_version(ndev, &ndev->xdna->fw_ver);
-	if (ret) {
-		XDNA_ERR(xdna, "Re-query firmware version failed");
-		goto async_event_free;
-	}
-
 	release_firmware(fw);
+	aie2_msg_init(ndev);
+	amdxdna_pm_init(xdna);
 	return 0;
 
-async_event_free:
-	aie2_error_async_events_free(ndev);
 stop_hw:
 	aie2_hw_stop(xdna);
-free_irq:
-	pci_free_irq_vectors(pdev);
 release_fw:
 	release_firmware(fw);
 
@@ -588,12 +617,8 @@ release_fw:
 
 static void aie2_fini(struct amdxdna_dev *xdna)
 {
-	struct pci_dev *pdev = to_pci_dev(xdna->ddev.dev);
-	struct amdxdna_dev_hdl *ndev = xdna->dev_handle;
-
+	amdxdna_pm_fini(xdna);
 	aie2_hw_stop(xdna);
-	aie2_error_async_events_free(ndev);
-	pci_free_irq_vectors(pdev);
 }
 
 static int aie2_get_aie_status(struct amdxdna_client *client,
@@ -752,66 +777,182 @@ static int aie2_get_clock_metadata(struct amdxdna_client *client,
 	return ret;
 }
 
+static int aie2_hwctx_status_cb(struct amdxdna_hwctx *hwctx, void *arg)
+{
+	struct amdxdna_drm_hwctx_entry *tmp __free(kfree) = NULL;
+	struct amdxdna_drm_get_array *array_args = arg;
+	struct amdxdna_drm_hwctx_entry __user *buf;
+	u32 size;
+
+	if (!array_args->num_element)
+		return -EINVAL;
+
+	tmp = kzalloc(sizeof(*tmp), GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+	tmp->pid = hwctx->client->pid;
+	tmp->context_id = hwctx->id;
+	tmp->start_col = hwctx->start_col;
+	tmp->num_col = hwctx->num_col;
+	tmp->command_submissions = hwctx->priv->seq;
+	tmp->command_completions = hwctx->priv->completed;
+	tmp->pasid = hwctx->client->pasid;
+	tmp->priority = hwctx->qos.priority;
+	tmp->gops = hwctx->qos.gops;
+	tmp->fps = hwctx->qos.fps;
+	tmp->dma_bandwidth = hwctx->qos.dma_bandwidth;
+	tmp->latency = hwctx->qos.latency;
+	tmp->frame_exec_time = hwctx->qos.frame_exec_time;
+	tmp->state = AMDXDNA_HWCTX_STATE_ACTIVE;
+
+	buf = u64_to_user_ptr(array_args->buffer);
+	size = min(sizeof(*tmp), array_args->element_size);
+
+	if (copy_to_user(buf, tmp, size))
+		return -EFAULT;
+
+	array_args->buffer += size;
+	array_args->num_element--;
+
+	return 0;
+}
+
 static int aie2_get_hwctx_status(struct amdxdna_client *client,
 				 struct amdxdna_drm_get_info *args)
 {
-	struct amdxdna_drm_query_hwctx __user *buf;
+	struct amdxdna_drm_get_array array_args;
 	struct amdxdna_dev *xdna = client->xdna;
-	struct amdxdna_drm_query_hwctx *tmp;
 	struct amdxdna_client *tmp_client;
-	struct amdxdna_hwctx *hwctx;
-	unsigned long hwctx_id;
-	bool overflow = false;
-	u32 req_bytes = 0;
-	u32 hw_i = 0;
-	int ret = 0;
-	int idx;
+	int ret;
 
 	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
 
-	tmp = kzalloc(sizeof(*tmp), GFP_KERNEL);
-	if (!tmp)
+	array_args.element_size = sizeof(struct amdxdna_drm_query_hwctx);
+	array_args.buffer = args->buffer;
+	array_args.num_element = args->buffer_size / array_args.element_size;
+	list_for_each_entry(tmp_client, &xdna->client_list, node) {
+		ret = amdxdna_hwctx_walk(tmp_client, &array_args,
+					 aie2_hwctx_status_cb);
+		if (ret)
+			break;
+	}
+
+	args->buffer_size -= (u32)(array_args.buffer - args->buffer);
+	return 0;
+}
+
+static int aie2_query_resource_info(struct amdxdna_client *client,
+				    struct amdxdna_drm_get_info *args)
+{
+	struct amdxdna_drm_get_resource_info res_info;
+	const struct amdxdna_dev_priv *priv;
+	struct amdxdna_dev_hdl *ndev;
+	struct amdxdna_dev *xdna;
+
+	xdna = client->xdna;
+	ndev = xdna->dev_handle;
+	priv = ndev->priv;
+
+	res_info.npu_clk_max = priv->dpm_clk_tbl[ndev->max_dpm_level].hclk;
+	res_info.npu_tops_max = ndev->max_tops;
+	res_info.npu_task_max = priv->hwctx_limit;
+	res_info.npu_tops_curr = ndev->curr_tops;
+	res_info.npu_task_curr = ndev->hwctx_num;
+
+	if (copy_to_user(u64_to_user_ptr(args->buffer), &res_info, sizeof(res_info)))
+		return -EFAULT;
+
+	return 0;
+}
+
+static int aie2_fill_hwctx_map(struct amdxdna_hwctx *hwctx, void *arg)
+{
+	struct amdxdna_dev *xdna = hwctx->client->xdna;
+	u32 *map = arg;
+
+	if (hwctx->fw_ctx_id >= xdna->dev_handle->priv->hwctx_limit) {
+		XDNA_ERR(xdna, "Invalid fw ctx id %d/%d ", hwctx->fw_ctx_id,
+			 xdna->dev_handle->priv->hwctx_limit);
+		return -EINVAL;
+	}
+
+	map[hwctx->fw_ctx_id] = hwctx->id;
+	return 0;
+}
+
+static int aie2_get_telemetry(struct amdxdna_client *client,
+			      struct amdxdna_drm_get_info *args)
+{
+	struct amdxdna_drm_query_telemetry_header *header __free(kfree) = NULL;
+	u32 telemetry_data_sz, header_sz, elem_num;
+	struct amdxdna_dev *xdna = client->xdna;
+	struct amdxdna_client *tmp_client;
+	int ret;
+
+	elem_num = xdna->dev_handle->priv->hwctx_limit;
+	header_sz = struct_size(header, map, elem_num);
+	if (args->buffer_size <= header_sz) {
+		XDNA_ERR(xdna, "Invalid buffer size");
+		return -EINVAL;
+	}
+
+	telemetry_data_sz = args->buffer_size - header_sz;
+	if (telemetry_data_sz > SZ_4M) {
+		XDNA_ERR(xdna, "Buffer size is too big, %d", telemetry_data_sz);
+		return -EINVAL;
+	}
+
+	header = kzalloc(header_sz, GFP_KERNEL);
+	if (!header)
 		return -ENOMEM;
 
-	buf = u64_to_user_ptr(args->buffer);
+	if (copy_from_user(header, u64_to_user_ptr(args->buffer), sizeof(*header))) {
+		XDNA_ERR(xdna, "Failed to copy telemetry header from user");
+		return -EFAULT;
+	}
+
+	header->map_num_elements = elem_num;
 	list_for_each_entry(tmp_client, &xdna->client_list, node) {
-		idx = srcu_read_lock(&tmp_client->hwctx_srcu);
-		amdxdna_for_each_hwctx(tmp_client, hwctx_id, hwctx) {
-			req_bytes += sizeof(*tmp);
-			if (args->buffer_size < req_bytes) {
-				/* Continue iterating to get the required size */
-				overflow = true;
-				continue;
-			}
-
-			memset(tmp, 0, sizeof(*tmp));
-			tmp->pid = tmp_client->pid;
-			tmp->context_id = hwctx->id;
-			tmp->start_col = hwctx->start_col;
-			tmp->num_col = hwctx->num_col;
-			tmp->command_submissions = hwctx->priv->seq;
-			tmp->command_completions = hwctx->priv->completed;
-
-			if (copy_to_user(&buf[hw_i], tmp, sizeof(*tmp))) {
-				ret = -EFAULT;
-				srcu_read_unlock(&tmp_client->hwctx_srcu, idx);
-				goto out;
-			}
-			hw_i++;
-		}
-		srcu_read_unlock(&tmp_client->hwctx_srcu, idx);
+		ret = amdxdna_hwctx_walk(tmp_client, &header->map,
+					 aie2_fill_hwctx_map);
+		if (ret)
+			return ret;
 	}
 
-	if (overflow) {
-		XDNA_ERR(xdna, "Invalid buffer size. Given: %u Need: %u.",
-			 args->buffer_size, req_bytes);
-		ret = -EINVAL;
+	ret = aie2_query_telemetry(xdna->dev_handle,
+				   u64_to_user_ptr(args->buffer + header_sz),
+				   telemetry_data_sz, header);
+	if (ret) {
+		XDNA_ERR(xdna, "Query telemetry failed ret %d", ret);
+		return ret;
 	}
 
-out:
-	kfree(tmp);
-	args->buffer_size = req_bytes;
-	return ret;
+	if (copy_to_user(u64_to_user_ptr(args->buffer), header, header_sz)) {
+		XDNA_ERR(xdna, "Copy header failed");
+		return -EFAULT;
+	}
+
+	return 0;
+}
+
+static int aie2_get_preempt_state(struct amdxdna_client *client,
+				  struct amdxdna_drm_get_info *args)
+{
+	struct amdxdna_drm_attribute_state state = {};
+	struct amdxdna_dev *xdna = client->xdna;
+	struct amdxdna_dev_hdl *ndev;
+
+	ndev = xdna->dev_handle;
+	if (args->param == DRM_AMDXDNA_GET_FORCE_PREEMPT_STATE)
+		state.state = ndev->force_preempt_enabled;
+	else if (args->param == DRM_AMDXDNA_GET_FRAME_BOUNDARY_PREEMPT_STATE)
+		state.state = ndev->frame_boundary_preempt;
+
+	if (copy_to_user(u64_to_user_ptr(args->buffer), &state, sizeof(state)))
+		return -EFAULT;
+
+	return 0;
 }
 
 static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_info *args)
@@ -822,6 +963,10 @@ static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_i
 	if (!drm_dev_enter(&xdna->ddev, &idx))
 		return -ENODEV;
 
+	ret = amdxdna_pm_resume_get(xdna);
+	if (ret)
+		goto dev_exit;
+
 	switch (args->param) {
 	case DRM_AMDXDNA_QUERY_AIE_STATUS:
 		ret = aie2_get_aie_status(client, args);
@@ -844,12 +989,93 @@ static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_i
 	case DRM_AMDXDNA_GET_POWER_MODE:
 		ret = aie2_get_power_mode(client, args);
 		break;
+	case DRM_AMDXDNA_QUERY_TELEMETRY:
+		ret = aie2_get_telemetry(client, args);
+		break;
+	case DRM_AMDXDNA_QUERY_RESOURCE_INFO:
+		ret = aie2_query_resource_info(client, args);
+		break;
+	case DRM_AMDXDNA_GET_FORCE_PREEMPT_STATE:
+	case DRM_AMDXDNA_GET_FRAME_BOUNDARY_PREEMPT_STATE:
+		ret = aie2_get_preempt_state(client, args);
+		break;
 	default:
 		XDNA_ERR(xdna, "Not supported request parameter %u", args->param);
 		ret = -EOPNOTSUPP;
 	}
+
+	amdxdna_pm_suspend_put(xdna);
 	XDNA_DBG(xdna, "Got param %d", args->param);
 
+dev_exit:
+	drm_dev_exit(idx);
+	return ret;
+}
+
+static int aie2_query_ctx_status_array(struct amdxdna_client *client,
+				       struct amdxdna_drm_get_array *args)
+{
+	struct amdxdna_drm_get_array array_args;
+	struct amdxdna_dev *xdna = client->xdna;
+	struct amdxdna_client *tmp_client;
+	int ret;
+
+	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
+
+	if (args->element_size > SZ_4K || args->num_element > SZ_1K) {
+		XDNA_DBG(xdna, "Invalid element size %d or number of element %d",
+			 args->element_size, args->num_element);
+		return -EINVAL;
+	}
+
+	array_args.element_size = min(args->element_size,
+				      sizeof(struct amdxdna_drm_hwctx_entry));
+	array_args.buffer = args->buffer;
+	array_args.num_element = args->num_element * args->element_size /
+				array_args.element_size;
+	list_for_each_entry(tmp_client, &xdna->client_list, node) {
+		ret = amdxdna_hwctx_walk(tmp_client, &array_args,
+					 aie2_hwctx_status_cb);
+		if (ret)
+			break;
+	}
+
+	args->element_size = array_args.element_size;
+	args->num_element = (u32)((array_args.buffer - args->buffer) /
+				  args->element_size);
+
+	return 0;
+}
+
+static int aie2_get_array(struct amdxdna_client *client,
+			  struct amdxdna_drm_get_array *args)
+{
+	struct amdxdna_dev *xdna = client->xdna;
+	int ret, idx;
+
+	if (!drm_dev_enter(&xdna->ddev, &idx))
+		return -ENODEV;
+
+	ret = amdxdna_pm_resume_get(xdna);
+	if (ret)
+		goto dev_exit;
+
+	switch (args->param) {
+	case DRM_AMDXDNA_HW_CONTEXT_ALL:
+		ret = aie2_query_ctx_status_array(client, args);
+		break;
+	case DRM_AMDXDNA_HW_LAST_ASYNC_ERR:
+		ret = aie2_get_array_async_error(xdna->dev_handle, args);
+		break;
+	default:
+		XDNA_ERR(xdna, "Not supported request parameter %u", args->param);
+		ret = -EOPNOTSUPP;
+	}
+
+	amdxdna_pm_suspend_put(xdna);
+	XDNA_DBG(xdna, "Got param %d", args->param);
+
+dev_exit:
 	drm_dev_exit(idx);
 	return ret;
 }
@@ -879,6 +1105,38 @@ static int aie2_set_power_mode(struct amdxdna_client *client,
 	return aie2_pm_set_mode(xdna->dev_handle, power_mode);
 }
 
+static int aie2_set_preempt_state(struct amdxdna_client *client,
+				  struct amdxdna_drm_set_state *args)
+{
+	struct amdxdna_dev_hdl *ndev = client->xdna->dev_handle;
+	struct amdxdna_drm_attribute_state state;
+	u32 val;
+	int ret;
+
+	if (copy_from_user(&state, u64_to_user_ptr(args->buffer), sizeof(state)))
+		return -EFAULT;
+
+	if (state.state > 1)
+		return -EINVAL;
+
+	if (XDNA_MBZ_DBG(client->xdna, state.pad, sizeof(state.pad)))
+		return -EINVAL;
+
+	if (args->param == DRM_AMDXDNA_SET_FORCE_PREEMPT) {
+		ndev->force_preempt_enabled = state.state;
+	} else if (args->param == DRM_AMDXDNA_SET_FRAME_BOUNDARY_PREEMPT) {
+		val = state.state;
+		ret = aie2_runtime_cfg(ndev, AIE2_RT_CFG_FRAME_BOUNDARY_PREEMPT,
+				       &val);
+		if (ret)
+			return ret;
+
+		ndev->frame_boundary_preempt = state.state;
+	}
+
+	return 0;
+}
+
 static int aie2_set_state(struct amdxdna_client *client,
 			  struct amdxdna_drm_set_state *args)
 {
@@ -888,32 +1146,42 @@ static int aie2_set_state(struct amdxdna_client *client,
 	if (!drm_dev_enter(&xdna->ddev, &idx))
 		return -ENODEV;
 
+	ret = amdxdna_pm_resume_get(xdna);
+	if (ret)
+		goto dev_exit;
+
 	switch (args->param) {
 	case DRM_AMDXDNA_SET_POWER_MODE:
 		ret = aie2_set_power_mode(client, args);
 		break;
+	case DRM_AMDXDNA_SET_FORCE_PREEMPT:
+	case DRM_AMDXDNA_SET_FRAME_BOUNDARY_PREEMPT:
+		ret = aie2_set_preempt_state(client, args);
+		break;
 	default:
 		XDNA_ERR(xdna, "Not supported request parameter %u", args->param);
 		ret = -EOPNOTSUPP;
 		break;
 	}
 
+	amdxdna_pm_suspend_put(xdna);
+dev_exit:
 	drm_dev_exit(idx);
 	return ret;
 }
 
 const struct amdxdna_dev_ops aie2_ops = {
-	.init           = aie2_init,
-	.fini           = aie2_fini,
-	.resume         = aie2_hw_start,
-	.suspend        = aie2_hw_stop,
-	.get_aie_info   = aie2_get_info,
-	.set_aie_state	= aie2_set_state,
-	.hwctx_init     = aie2_hwctx_init,
-	.hwctx_fini     = aie2_hwctx_fini,
-	.hwctx_config   = aie2_hwctx_config,
-	.cmd_submit     = aie2_cmd_submit,
+	.init = aie2_init,
+	.fini = aie2_fini,
+	.resume = aie2_hw_resume,
+	.suspend = aie2_hw_suspend,
+	.get_aie_info = aie2_get_info,
+	.set_aie_state = aie2_set_state,
+	.hwctx_init = aie2_hwctx_init,
+	.hwctx_fini = aie2_hwctx_fini,
+	.hwctx_config = aie2_hwctx_config,
+	.hwctx_sync_debug_bo = aie2_hwctx_sync_debug_bo,
+	.cmd_submit = aie2_cmd_submit,
 	.hmm_invalidate = aie2_hmm_invalidate,
-	.hwctx_suspend  = aie2_hwctx_suspend,
-	.hwctx_resume   = aie2_hwctx_resume,
+	.get_array = aie2_get_array,
 };
diff --git a/drivers/accel/amdxdna/aie2_pci.h b/drivers/accel/amdxdna/aie2_pci.h
index 385914840eaa..a5f9c42155d1 100644
--- a/drivers/accel/amdxdna/aie2_pci.h
+++ b/drivers/accel/amdxdna/aie2_pci.h
@@ -110,12 +110,15 @@ struct aie_metadata {
 enum rt_config_category {
 	AIE2_RT_CFG_INIT,
 	AIE2_RT_CFG_CLK_GATING,
+	AIE2_RT_CFG_FORCE_PREEMPT,
+	AIE2_RT_CFG_FRAME_BOUNDARY_PREEMPT,
 };
 
 struct rt_config {
 	u32	type;
 	u32	value;
 	u32	category;
+	unsigned long feature_mask;
 };
 
 struct dpm_clk_freq {
@@ -156,6 +159,19 @@ enum aie2_dev_status {
 	AIE2_DEV_START,
 };
 
+struct aie2_exec_msg_ops {
+	int (*init_cu_req)(struct amdxdna_gem_obj *cmd_bo, void *req,
+			   size_t *size, u32 *msg_op);
+	int (*init_dpu_req)(struct amdxdna_gem_obj *cmd_bo, void *req,
+			    size_t *size, u32 *msg_op);
+	void (*init_chain_req)(void *req, u64 slot_addr, size_t size, u32 cmd_cnt);
+	int (*fill_cf_slot)(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size);
+	int (*fill_dpu_slot)(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size);
+	int (*fill_preempt_slot)(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size);
+	int (*fill_elf_slot)(struct amdxdna_gem_obj *cmd_bo, void *slot, size_t *size);
+	u32 (*get_chain_msg_op)(u32 cmd_op);
+};
+
 struct amdxdna_dev_hdl {
 	struct amdxdna_dev		*xdna;
 	const struct amdxdna_dev_priv	*priv;
@@ -173,6 +189,8 @@ struct amdxdna_dev_hdl {
 	u32				total_col;
 	struct aie_version		version;
 	struct aie_metadata		metadata;
+	unsigned long			feature_mask;
+	struct aie2_exec_msg_ops	*exec_msg_ops;
 
 	/* power management and clock*/
 	enum amdxdna_power_mode_type	pw_mode;
@@ -182,6 +200,10 @@ struct amdxdna_dev_hdl {
 	u32				clk_gating;
 	u32				npuclk_freq;
 	u32				hclk_freq;
+	u32				max_tops;
+	u32				curr_tops;
+	u32				force_preempt_enabled;
+	u32				frame_boundary_preempt;
 
 	/* Mailbox and the management channel */
 	struct mailbox			*mbox;
@@ -190,6 +212,8 @@ struct amdxdna_dev_hdl {
 
 	enum aie2_dev_status		dev_status;
 	u32				hwctx_num;
+
+	struct amdxdna_async_error	last_async_err;
 };
 
 #define DEFINE_BAR_OFFSET(reg_name, bar, reg_addr) \
@@ -204,12 +228,27 @@ struct aie2_hw_ops {
 	int (*set_dpm)(struct amdxdna_dev_hdl *ndev, u32 dpm_level);
 };
 
+enum aie2_fw_feature {
+	AIE2_NPU_COMMAND,
+	AIE2_PREEMPT,
+	AIE2_FEATURE_MAX
+};
+
+struct aie2_fw_feature_tbl {
+	enum aie2_fw_feature feature;
+	u32 max_minor;
+	u32 min_minor;
+};
+
+#define AIE2_FEATURE_ON(ndev, feature)	test_bit(feature, &(ndev)->feature_mask)
+
 struct amdxdna_dev_priv {
 	const char			*fw_path;
 	u64				protocol_major;
 	u64				protocol_minor;
 	const struct rt_config		*rt_config;
 	const struct dpm_clk_freq	*dpm_clk_tbl;
+	const struct aie2_fw_feature_tbl *fw_feature_tbl;
 
 #define COL_ALIGN_NONE   0
 #define COL_ALIGN_NATURE 1
@@ -217,6 +256,7 @@ struct amdxdna_dev_priv {
 	u32				mbox_dev_addr;
 	/* If mbox_size is 0, use BAR size. See MBOX_SIZE macro */
 	u32				mbox_size;
+	u32				hwctx_limit;
 	u32				sram_dev_addr;
 	struct aie2_bar_off_pair	sram_offs[SRAM_MAX_INDEX];
 	struct aie2_bar_off_pair	psp_regs_off[PSP_MAX_REGS];
@@ -234,6 +274,7 @@ extern const struct dpm_clk_freq npu1_dpm_clk_table[];
 extern const struct dpm_clk_freq npu4_dpm_clk_table[];
 extern const struct rt_config npu1_default_rt_cfg[];
 extern const struct rt_config npu4_default_rt_cfg[];
+extern const struct aie2_fw_feature_tbl npu4_fw_feature_table[];
 
 /* aie2_smu.c */
 int aie2_smu_init(struct amdxdna_dev_hdl *ndev);
@@ -253,10 +294,12 @@ void aie2_psp_stop(struct psp_device *psp);
 /* aie2_error.c */
 int aie2_error_async_events_alloc(struct amdxdna_dev_hdl *ndev);
 void aie2_error_async_events_free(struct amdxdna_dev_hdl *ndev);
-int aie2_error_async_events_send(struct amdxdna_dev_hdl *ndev);
 int aie2_error_async_msg_thread(void *data);
+int aie2_get_array_async_error(struct amdxdna_dev_hdl *ndev,
+			       struct amdxdna_drm_get_array *args);
 
 /* aie2_message.c */
+void aie2_msg_init(struct amdxdna_dev_hdl *ndev);
 int aie2_suspend_fw(struct amdxdna_dev_hdl *ndev);
 int aie2_resume_fw(struct amdxdna_dev_hdl *ndev);
 int aie2_set_runtime_cfg(struct amdxdna_dev_hdl *ndev, u32 type, u64 value);
@@ -270,9 +313,13 @@ int aie2_create_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwct
 int aie2_destroy_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwctx);
 int aie2_map_host_buf(struct amdxdna_dev_hdl *ndev, u32 context_id, u64 addr, u64 size);
 int aie2_query_status(struct amdxdna_dev_hdl *ndev, char __user *buf, u32 size, u32 *cols_filled);
+int aie2_query_telemetry(struct amdxdna_dev_hdl *ndev,
+			 char __user *buf, u32 size,
+			 struct amdxdna_drm_query_telemetry_header *header);
 int aie2_register_asyn_event_msg(struct amdxdna_dev_hdl *ndev, dma_addr_t addr, u32 size,
 				 void *handle, int (*cb)(void*, void __iomem *, size_t));
-int aie2_config_cu(struct amdxdna_hwctx *hwctx);
+int aie2_config_cu(struct amdxdna_hwctx *hwctx,
+		   int (*notify_cb)(void *, void __iomem *, size_t));
 int aie2_execbuf(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job,
 		 int (*notify_cb)(void *, void __iomem *, size_t));
 int aie2_cmdlist_single_execbuf(struct amdxdna_hwctx *hwctx,
@@ -283,15 +330,17 @@ int aie2_cmdlist_multi_execbuf(struct amdxdna_hwctx *hwctx,
 			       int (*notify_cb)(void *, void __iomem *, size_t));
 int aie2_sync_bo(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job,
 		 int (*notify_cb)(void *, void __iomem *, size_t));
+int aie2_config_debug_bo(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job,
+			 int (*notify_cb)(void *, void __iomem *, size_t));
 
 /* aie2_hwctx.c */
 int aie2_hwctx_init(struct amdxdna_hwctx *hwctx);
 void aie2_hwctx_fini(struct amdxdna_hwctx *hwctx);
 int aie2_hwctx_config(struct amdxdna_hwctx *hwctx, u32 type, u64 value, void *buf, u32 size);
-void aie2_hwctx_suspend(struct amdxdna_hwctx *hwctx);
-void aie2_hwctx_resume(struct amdxdna_hwctx *hwctx);
+int aie2_hwctx_sync_debug_bo(struct amdxdna_hwctx *hwctx, u32 debug_bo_hdl);
+void aie2_hwctx_suspend(struct amdxdna_client *client);
+int aie2_hwctx_resume(struct amdxdna_client *client);
 int aie2_cmd_submit(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job, u64 *seq);
 void aie2_hmm_invalidate(struct amdxdna_gem_obj *abo, unsigned long cur_seq);
-void aie2_restart_ctx(struct amdxdna_client *client);
 
 #endif /* _AIE2_PCI_H_ */
diff --git a/drivers/accel/amdxdna/aie2_psp.c b/drivers/accel/amdxdna/aie2_psp.c
index dc3a072ce3b6..f28a060a8810 100644
--- a/drivers/accel/amdxdna/aie2_psp.c
+++ b/drivers/accel/amdxdna/aie2_psp.c
@@ -126,8 +126,8 @@ struct psp_device *aie2m_psp_create(struct drm_device *ddev, struct psp_config *
 	psp->ddev = ddev;
 	memcpy(psp->psp_regs, conf->psp_regs, sizeof(psp->psp_regs));
 
-	psp->fw_buf_sz = ALIGN(conf->fw_size, PSP_FW_ALIGN) + PSP_FW_ALIGN;
-	psp->fw_buffer = drmm_kmalloc(ddev, psp->fw_buf_sz, GFP_KERNEL);
+	psp->fw_buf_sz = ALIGN(conf->fw_size, PSP_FW_ALIGN);
+	psp->fw_buffer = drmm_kmalloc(ddev, psp->fw_buf_sz + PSP_FW_ALIGN, GFP_KERNEL);
 	if (!psp->fw_buffer) {
 		drm_err(ddev, "no memory for fw buffer");
 		return NULL;
diff --git a/drivers/accel/amdxdna/aie2_smu.c b/drivers/accel/amdxdna/aie2_smu.c
index d303701b0ded..bd94ee96c2bc 100644
--- a/drivers/accel/amdxdna/aie2_smu.c
+++ b/drivers/accel/amdxdna/aie2_smu.c
@@ -11,6 +11,7 @@
 
 #include "aie2_pci.h"
 #include "amdxdna_pci_drv.h"
+#include "amdxdna_pm.h"
 
 #define SMU_RESULT_OK		1
 
@@ -22,6 +23,13 @@
 #define AIE2_SMU_SET_SOFT_DPMLEVEL	0x7
 #define AIE2_SMU_SET_HARD_DPMLEVEL	0x8
 
+#define NPU4_DPM_TOPS(ndev, dpm_level) \
+({ \
+	typeof(ndev) _ndev = ndev; \
+	(4096 * (_ndev)->total_col * \
+	 (_ndev)->priv->dpm_clk_tbl[dpm_level].hclk / 1000000); \
+})
+
 static int aie2_smu_exec(struct amdxdna_dev_hdl *ndev, u32 reg_cmd,
 			 u32 reg_arg, u32 *out)
 {
@@ -59,12 +67,16 @@ int npu1_set_dpm(struct amdxdna_dev_hdl *ndev, u32 dpm_level)
 	u32 freq;
 	int ret;
 
+	ret = amdxdna_pm_resume_get(ndev->xdna);
+	if (ret)
+		return ret;
+
 	ret = aie2_smu_exec(ndev, AIE2_SMU_SET_MPNPUCLK_FREQ,
 			    ndev->priv->dpm_clk_tbl[dpm_level].npuclk, &freq);
 	if (ret) {
 		XDNA_ERR(ndev->xdna, "Set npu clock to %d failed, ret %d\n",
 			 ndev->priv->dpm_clk_tbl[dpm_level].npuclk, ret);
-		return ret;
+		goto suspend_put;
 	}
 	ndev->npuclk_freq = freq;
 
@@ -73,49 +85,78 @@ int npu1_set_dpm(struct amdxdna_dev_hdl *ndev, u32 dpm_level)
 	if (ret) {
 		XDNA_ERR(ndev->xdna, "Set h clock to %d failed, ret %d\n",
 			 ndev->priv->dpm_clk_tbl[dpm_level].hclk, ret);
-		return ret;
+		goto suspend_put;
 	}
+
+	amdxdna_pm_suspend_put(ndev->xdna);
 	ndev->hclk_freq = freq;
 	ndev->dpm_level = dpm_level;
+	ndev->max_tops = 2 * ndev->total_col;
+	ndev->curr_tops = ndev->max_tops * freq / 1028;
 
 	XDNA_DBG(ndev->xdna, "MP-NPU clock %d, H clock %d\n",
 		 ndev->npuclk_freq, ndev->hclk_freq);
 
 	return 0;
+
+suspend_put:
+	amdxdna_pm_suspend_put(ndev->xdna);
+	return ret;
 }
 
 int npu4_set_dpm(struct amdxdna_dev_hdl *ndev, u32 dpm_level)
 {
 	int ret;
 
+	ret = amdxdna_pm_resume_get(ndev->xdna);
+	if (ret)
+		return ret;
+
 	ret = aie2_smu_exec(ndev, AIE2_SMU_SET_HARD_DPMLEVEL, dpm_level, NULL);
 	if (ret) {
 		XDNA_ERR(ndev->xdna, "Set hard dpm level %d failed, ret %d ",
 			 dpm_level, ret);
-		return ret;
+		goto suspend_put;
 	}
 
 	ret = aie2_smu_exec(ndev, AIE2_SMU_SET_SOFT_DPMLEVEL, dpm_level, NULL);
 	if (ret) {
 		XDNA_ERR(ndev->xdna, "Set soft dpm level %d failed, ret %d",
 			 dpm_level, ret);
-		return ret;
+		goto suspend_put;
 	}
 
+	amdxdna_pm_suspend_put(ndev->xdna);
 	ndev->npuclk_freq = ndev->priv->dpm_clk_tbl[dpm_level].npuclk;
 	ndev->hclk_freq = ndev->priv->dpm_clk_tbl[dpm_level].hclk;
 	ndev->dpm_level = dpm_level;
+	ndev->max_tops = NPU4_DPM_TOPS(ndev, ndev->max_dpm_level);
+	ndev->curr_tops = NPU4_DPM_TOPS(ndev, dpm_level);
 
 	XDNA_DBG(ndev->xdna, "MP-NPU clock %d, H clock %d\n",
 		 ndev->npuclk_freq, ndev->hclk_freq);
 
 	return 0;
+
+suspend_put:
+	amdxdna_pm_suspend_put(ndev->xdna);
+	return ret;
 }
 
 int aie2_smu_init(struct amdxdna_dev_hdl *ndev)
 {
 	int ret;
 
+	/*
+	 * Failing to set power off indicates an unrecoverable hardware or
+	 * firmware error.
+	 */
+	ret = aie2_smu_exec(ndev, AIE2_SMU_POWER_OFF, 0, NULL);
+	if (ret) {
+		XDNA_ERR(ndev->xdna, "Access power failed, ret %d", ret);
+		return ret;
+	}
+
 	ret = aie2_smu_exec(ndev, AIE2_SMU_POWER_ON, 0, NULL);
 	if (ret) {
 		XDNA_ERR(ndev->xdna, "Power on failed, ret %d", ret);
diff --git a/drivers/accel/amdxdna/amdxdna_ctx.c b/drivers/accel/amdxdna/amdxdna_ctx.c
index be073224bd69..d17aef89a0ad 100644
--- a/drivers/accel/amdxdna/amdxdna_ctx.c
+++ b/drivers/accel/amdxdna/amdxdna_ctx.c
@@ -60,32 +60,6 @@ static struct dma_fence *amdxdna_fence_create(struct amdxdna_hwctx *hwctx)
 	return &fence->base;
 }
 
-void amdxdna_hwctx_suspend(struct amdxdna_client *client)
-{
-	struct amdxdna_dev *xdna = client->xdna;
-	struct amdxdna_hwctx *hwctx;
-	unsigned long hwctx_id;
-
-	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
-	mutex_lock(&client->hwctx_lock);
-	amdxdna_for_each_hwctx(client, hwctx_id, hwctx)
-		xdna->dev_info->ops->hwctx_suspend(hwctx);
-	mutex_unlock(&client->hwctx_lock);
-}
-
-void amdxdna_hwctx_resume(struct amdxdna_client *client)
-{
-	struct amdxdna_dev *xdna = client->xdna;
-	struct amdxdna_hwctx *hwctx;
-	unsigned long hwctx_id;
-
-	drm_WARN_ON(&xdna->ddev, !mutex_is_locked(&xdna->dev_lock));
-	mutex_lock(&client->hwctx_lock);
-	amdxdna_for_each_hwctx(client, hwctx_id, hwctx)
-		xdna->dev_info->ops->hwctx_resume(hwctx);
-	mutex_unlock(&client->hwctx_lock);
-}
-
 static void amdxdna_hwctx_destroy_rcu(struct amdxdna_hwctx *hwctx,
 				      struct srcu_struct *ss)
 {
@@ -94,14 +68,30 @@ static void amdxdna_hwctx_destroy_rcu(struct amdxdna_hwctx *hwctx,
 	synchronize_srcu(ss);
 
 	/* At this point, user is not able to submit new commands */
-	mutex_lock(&xdna->dev_lock);
 	xdna->dev_info->ops->hwctx_fini(hwctx);
-	mutex_unlock(&xdna->dev_lock);
 
 	kfree(hwctx->name);
 	kfree(hwctx);
 }
 
+int amdxdna_hwctx_walk(struct amdxdna_client *client, void *arg,
+		       int (*walk)(struct amdxdna_hwctx *hwctx, void *arg))
+{
+	struct amdxdna_hwctx *hwctx;
+	unsigned long hwctx_id;
+	int ret = 0, idx;
+
+	idx = srcu_read_lock(&client->hwctx_srcu);
+	amdxdna_for_each_hwctx(client, hwctx_id, hwctx) {
+		ret = walk(hwctx, arg);
+		if (ret)
+			break;
+	}
+	srcu_read_unlock(&client->hwctx_srcu, idx);
+
+	return ret;
+}
+
 void *amdxdna_cmd_get_payload(struct amdxdna_gem_obj *abo, u32 *size)
 {
 	struct amdxdna_cmd *cmd = abo->mem.kva;
@@ -123,14 +113,14 @@ void *amdxdna_cmd_get_payload(struct amdxdna_gem_obj *abo, u32 *size)
 	return &cmd->data[num_masks];
 }
 
-int amdxdna_cmd_get_cu_idx(struct amdxdna_gem_obj *abo)
+u32 amdxdna_cmd_get_cu_idx(struct amdxdna_gem_obj *abo)
 {
 	struct amdxdna_cmd *cmd = abo->mem.kva;
 	u32 num_masks, i;
 	u32 *cu_mask;
 
 	if (amdxdna_cmd_get_op(abo) == ERT_CMD_CHAIN)
-		return -1;
+		return INVALID_CU_IDX;
 
 	num_masks = 1 + FIELD_GET(AMDXDNA_CMD_EXTRA_CU_MASK, cmd->header);
 	cu_mask = cmd->data;
@@ -139,7 +129,7 @@ int amdxdna_cmd_get_cu_idx(struct amdxdna_gem_obj *abo)
 			return ffs(cu_mask[i]) - 1;
 	}
 
-	return -1;
+	return INVALID_CU_IDX;
 }
 
 /*
@@ -152,16 +142,12 @@ void amdxdna_hwctx_remove_all(struct amdxdna_client *client)
 	struct amdxdna_hwctx *hwctx;
 	unsigned long hwctx_id;
 
-	mutex_lock(&client->hwctx_lock);
 	amdxdna_for_each_hwctx(client, hwctx_id, hwctx) {
 		XDNA_DBG(client->xdna, "PID %d close HW context %d",
 			 client->pid, hwctx->id);
 		xa_erase(&client->hwctx_xa, hwctx->id);
-		mutex_unlock(&client->hwctx_lock);
 		amdxdna_hwctx_destroy_rcu(hwctx, &client->hwctx_srcu);
-		mutex_lock(&client->hwctx_lock);
 	}
-	mutex_unlock(&client->hwctx_lock);
 }
 
 int amdxdna_drm_create_hwctx_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
@@ -175,19 +161,14 @@ int amdxdna_drm_create_hwctx_ioctl(struct drm_device *dev, void *data, struct dr
 	if (args->ext || args->ext_flags)
 		return -EINVAL;
 
-	if (!drm_dev_enter(dev, &idx))
-		return -ENODEV;
-
 	hwctx = kzalloc(sizeof(*hwctx), GFP_KERNEL);
-	if (!hwctx) {
-		ret = -ENOMEM;
-		goto exit;
-	}
+	if (!hwctx)
+		return -ENOMEM;
 
 	if (copy_from_user(&hwctx->qos, u64_to_user_ptr(args->qos_p), sizeof(hwctx->qos))) {
 		XDNA_ERR(xdna, "Access QoS info failed");
-		ret = -EFAULT;
-		goto free_hwctx;
+		kfree(hwctx);
+		return -EFAULT;
 	}
 
 	hwctx->client = client;
@@ -195,30 +176,36 @@ int amdxdna_drm_create_hwctx_ioctl(struct drm_device *dev, void *data, struct dr
 	hwctx->num_tiles = args->num_tiles;
 	hwctx->mem_size = args->mem_size;
 	hwctx->max_opc = args->max_opc;
-	ret = xa_alloc_cyclic(&client->hwctx_xa, &hwctx->id, hwctx,
-			      XA_LIMIT(AMDXDNA_INVALID_CTX_HANDLE + 1, MAX_HWCTX_ID),
-			      &client->next_hwctxid, GFP_KERNEL);
-	if (ret < 0) {
-		XDNA_ERR(xdna, "Allocate hwctx ID failed, ret %d", ret);
+
+	guard(mutex)(&xdna->dev_lock);
+
+	if (!drm_dev_enter(dev, &idx)) {
+		ret = -ENODEV;
 		goto free_hwctx;
 	}
 
-	hwctx->name = kasprintf(GFP_KERNEL, "hwctx.%d.%d", client->pid, hwctx->id);
+	ret = xdna->dev_info->ops->hwctx_init(hwctx);
+	if (ret) {
+		XDNA_ERR(xdna, "Init hwctx failed, ret %d", ret);
+		goto dev_exit;
+	}
+
+	hwctx->name = kasprintf(GFP_KERNEL, "hwctx.%d.%d", client->pid, hwctx->fw_ctx_id);
 	if (!hwctx->name) {
 		ret = -ENOMEM;
-		goto rm_id;
+		goto fini_hwctx;
 	}
 
-	mutex_lock(&xdna->dev_lock);
-	ret = xdna->dev_info->ops->hwctx_init(hwctx);
-	if (ret) {
-		mutex_unlock(&xdna->dev_lock);
-		XDNA_ERR(xdna, "Init hwctx failed, ret %d", ret);
+	ret = xa_alloc_cyclic(&client->hwctx_xa, &hwctx->id, hwctx,
+			      XA_LIMIT(AMDXDNA_INVALID_CTX_HANDLE + 1, MAX_HWCTX_ID),
+			      &client->next_hwctxid, GFP_KERNEL);
+	if (ret < 0) {
+		XDNA_ERR(xdna, "Allocate hwctx ID failed, ret %d", ret);
 		goto free_name;
 	}
+
 	args->handle = hwctx->id;
 	args->syncobj_handle = hwctx->syncobj_hdl;
-	mutex_unlock(&xdna->dev_lock);
 
 	atomic64_set(&hwctx->job_submit_cnt, 0);
 	atomic64_set(&hwctx->job_free_cnt, 0);
@@ -228,12 +215,12 @@ int amdxdna_drm_create_hwctx_ioctl(struct drm_device *dev, void *data, struct dr
 
 free_name:
 	kfree(hwctx->name);
-rm_id:
-	xa_erase(&client->hwctx_xa, hwctx->id);
+fini_hwctx:
+	xdna->dev_info->ops->hwctx_fini(hwctx);
+dev_exit:
+	drm_dev_exit(idx);
 free_hwctx:
 	kfree(hwctx);
-exit:
-	drm_dev_exit(idx);
 	return ret;
 }
 
@@ -251,6 +238,7 @@ int amdxdna_drm_destroy_hwctx_ioctl(struct drm_device *dev, void *data, struct d
 	if (!drm_dev_enter(dev, &idx))
 		return -ENODEV;
 
+	mutex_lock(&xdna->dev_lock);
 	hwctx = xa_erase(&client->hwctx_xa, args->handle);
 	if (!hwctx) {
 		ret = -EINVAL;
@@ -267,6 +255,7 @@ int amdxdna_drm_destroy_hwctx_ioctl(struct drm_device *dev, void *data, struct d
 
 	XDNA_DBG(xdna, "PID %d destroyed HW context %d", client->pid, args->handle);
 out:
+	mutex_unlock(&xdna->dev_lock);
 	drm_dev_exit(idx);
 	return ret;
 }
@@ -339,6 +328,38 @@ unlock_srcu:
 	return ret;
 }
 
+int amdxdna_hwctx_sync_debug_bo(struct amdxdna_client *client, u32 debug_bo_hdl)
+{
+	struct amdxdna_dev *xdna = client->xdna;
+	struct amdxdna_hwctx *hwctx;
+	struct amdxdna_gem_obj *abo;
+	struct drm_gem_object *gobj;
+	int ret, idx;
+
+	if (!xdna->dev_info->ops->hwctx_sync_debug_bo)
+		return -EOPNOTSUPP;
+
+	gobj = drm_gem_object_lookup(client->filp, debug_bo_hdl);
+	if (!gobj)
+		return -EINVAL;
+
+	abo = to_xdna_obj(gobj);
+	guard(mutex)(&xdna->dev_lock);
+	idx = srcu_read_lock(&client->hwctx_srcu);
+	hwctx = xa_load(&client->hwctx_xa, abo->assigned_hwctx);
+	if (!hwctx) {
+		ret = -EINVAL;
+		goto unlock_srcu;
+	}
+
+	ret = xdna->dev_info->ops->hwctx_sync_debug_bo(hwctx, debug_bo_hdl);
+
+unlock_srcu:
+	srcu_read_unlock(&client->hwctx_srcu, idx);
+	drm_gem_object_put(gobj);
+	return ret;
+}
+
 static void
 amdxdna_arg_bos_put(struct amdxdna_sched_job *job)
 {
@@ -401,9 +422,11 @@ void amdxdna_sched_job_cleanup(struct amdxdna_sched_job *job)
 	trace_amdxdna_debug_point(job->hwctx->name, job->seq, "job release");
 	amdxdna_arg_bos_put(job);
 	amdxdna_gem_put_obj(job->cmd_bo);
+	dma_fence_put(job->fence);
 }
 
 int amdxdna_cmd_submit(struct amdxdna_client *client,
+		       struct amdxdna_drv_cmd *drv_cmd,
 		       u32 cmd_bo_hdl, u32 *arg_bo_hdls, u32 arg_bo_cnt,
 		       u32 hwctx_hdl, u64 *seq)
 {
@@ -417,6 +440,8 @@ int amdxdna_cmd_submit(struct amdxdna_client *client,
 	if (!job)
 		return -ENOMEM;
 
+	job->drv_cmd = drv_cmd;
+
 	if (cmd_bo_hdl != AMDXDNA_INVALID_BO_HANDLE) {
 		job->cmd_bo = amdxdna_gem_get_obj(client, cmd_bo_hdl, AMDXDNA_BO_CMD);
 		if (!job->cmd_bo) {
@@ -424,8 +449,6 @@ int amdxdna_cmd_submit(struct amdxdna_client *client,
 			ret = -EINVAL;
 			goto free_job;
 		}
-	} else {
-		job->cmd_bo = NULL;
 	}
 
 	ret = amdxdna_arg_bos_lookup(client, job, arg_bo_hdls, arg_bo_cnt);
@@ -443,11 +466,6 @@ int amdxdna_cmd_submit(struct amdxdna_client *client,
 		goto unlock_srcu;
 	}
 
-	if (hwctx->status != HWCTX_STAT_READY) {
-		XDNA_ERR(xdna, "HW Context is not ready");
-		ret = -EINVAL;
-		goto unlock_srcu;
-	}
 
 	job->hwctx = hwctx;
 	job->mm = current->mm;
@@ -524,7 +542,7 @@ static int amdxdna_drm_submit_execbuf(struct amdxdna_client *client,
 		}
 	}
 
-	ret = amdxdna_cmd_submit(client, cmd_bo_hdl, arg_bo_hdls,
+	ret = amdxdna_cmd_submit(client, NULL, cmd_bo_hdl, arg_bo_hdls,
 				 args->arg_count, args->hwctx, &args->seq);
 	if (ret)
 		XDNA_DBG(xdna, "Submit cmds failed, ret %d", ret);
diff --git a/drivers/accel/amdxdna/amdxdna_ctx.h b/drivers/accel/amdxdna/amdxdna_ctx.h
index f0a4a8586d85..b6151244d64f 100644
--- a/drivers/accel/amdxdna/amdxdna_ctx.h
+++ b/drivers/accel/amdxdna/amdxdna_ctx.h
@@ -13,9 +13,12 @@
 struct amdxdna_hwctx_priv;
 
 enum ert_cmd_opcode {
-	ERT_START_CU      = 0,
-	ERT_CMD_CHAIN     = 19,
-	ERT_START_NPU     = 20,
+	ERT_START_CU = 0,
+	ERT_CMD_CHAIN = 19,
+	ERT_START_NPU = 20,
+	ERT_START_NPU_PREEMPT = 21,
+	ERT_START_NPU_PREEMPT_ELF = 22,
+	ERT_INVALID_CMD	= ~0U,
 };
 
 enum ert_cmd_state {
@@ -54,6 +57,21 @@ struct amdxdna_cmd_chain {
 	u64 data[] __counted_by(command_count);
 };
 
+/*
+ * Interpretation of the beginning of data payload for ERT_START_NPU_PREEMPT in
+ * amdxdna_cmd. The rest of the payload in amdxdna_cmd is regular kernel args.
+ */
+struct amdxdna_cmd_preempt_data {
+	u64 inst_buf;	    /* instruction buffer address */
+	u64 save_buf;	    /* save buffer address */
+	u64 restore_buf;    /* restore buffer address */
+	u32 inst_size;	    /* size of instruction buffer in bytes */
+	u32 save_size;	    /* size of save buffer in bytes */
+	u32 restore_size;   /* size of restore buffer in bytes */
+	u32 inst_prop_cnt;  /* properties count */
+	u32 prop_args[];    /* properties and regular kernel arguments */
+};
+
 /* Exec buffer command header format */
 #define AMDXDNA_CMD_STATE		GENMASK(3, 0)
 #define AMDXDNA_CMD_EXTRA_CU_MASK	GENMASK(11, 10)
@@ -64,6 +82,8 @@ struct amdxdna_cmd {
 	u32 data[];
 };
 
+#define INVALID_CU_IDX		(~0U)
+
 struct amdxdna_hwctx {
 	struct amdxdna_client		*client;
 	struct amdxdna_hwctx_priv	*priv;
@@ -95,6 +115,17 @@ struct amdxdna_hwctx {
 #define drm_job_to_xdna_job(j) \
 	container_of(j, struct amdxdna_sched_job, base)
 
+enum amdxdna_job_opcode {
+	SYNC_DEBUG_BO,
+	ATTACH_DEBUG_BO,
+	DETACH_DEBUG_BO,
+};
+
+struct amdxdna_drv_cmd {
+	enum amdxdna_job_opcode	opcode;
+	u32			result;
+};
+
 struct amdxdna_sched_job {
 	struct drm_sched_job	base;
 	struct kref		refcnt;
@@ -105,7 +136,9 @@ struct amdxdna_sched_job {
 	/* user can wait on this fence */
 	struct dma_fence	*out_fence;
 	bool			job_done;
+	bool			job_timeout;
 	u64			seq;
+	struct amdxdna_drv_cmd	*drv_cmd;
 	struct amdxdna_gem_obj	*cmd_bo;
 	size_t			bo_cnt;
 	struct drm_gem_object	*bos[] __counted_by(bo_cnt);
@@ -137,21 +170,17 @@ amdxdna_cmd_get_state(struct amdxdna_gem_obj *abo)
 }
 
 void *amdxdna_cmd_get_payload(struct amdxdna_gem_obj *abo, u32 *size);
-int amdxdna_cmd_get_cu_idx(struct amdxdna_gem_obj *abo);
-
-static inline u32 amdxdna_hwctx_col_map(struct amdxdna_hwctx *hwctx)
-{
-	return GENMASK(hwctx->start_col + hwctx->num_col - 1,
-		       hwctx->start_col);
-}
+u32 amdxdna_cmd_get_cu_idx(struct amdxdna_gem_obj *abo);
 
 void amdxdna_sched_job_cleanup(struct amdxdna_sched_job *job);
 void amdxdna_hwctx_remove_all(struct amdxdna_client *client);
-void amdxdna_hwctx_suspend(struct amdxdna_client *client);
-void amdxdna_hwctx_resume(struct amdxdna_client *client);
+int amdxdna_hwctx_walk(struct amdxdna_client *client, void *arg,
+		       int (*walk)(struct amdxdna_hwctx *hwctx, void *arg));
+int amdxdna_hwctx_sync_debug_bo(struct amdxdna_client *client, u32 debug_bo_hdl);
 
 int amdxdna_cmd_submit(struct amdxdna_client *client,
-		       u32 cmd_bo_hdls, u32 *arg_bo_hdls, u32 arg_bo_cnt,
+		       struct amdxdna_drv_cmd *drv_cmd, u32 cmd_bo_hdls,
+		       u32 *arg_bo_hdls, u32 arg_bo_cnt,
 		       u32 hwctx_hdl, u64 *seq);
 
 int amdxdna_cmd_wait(struct amdxdna_client *client, u32 hwctx_hdl,
diff --git a/drivers/accel/amdxdna/amdxdna_error.h b/drivers/accel/amdxdna/amdxdna_error.h
new file mode 100644
index 000000000000..c51de86ec12b
--- /dev/null
+++ b/drivers/accel/amdxdna/amdxdna_error.h
@@ -0,0 +1,59 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2025, Advanced Micro Devices, Inc.
+ */
+
+#ifndef _AMDXDNA_ERROR_H_
+#define _AMDXDNA_ERROR_H_
+
+#include <linux/bitfield.h>
+#include <linux/bits.h>
+
+#define AMDXDNA_ERR_DRV_AIE		4
+#define AMDXDNA_ERR_SEV_CRITICAL	3
+#define AMDXDNA_ERR_CLASS_AIE		2
+
+#define AMDXDNA_ERR_NUM_MASK		GENMASK_U64(15, 0)
+#define AMDXDNA_ERR_DRV_MASK		GENMASK_U64(23, 16)
+#define AMDXDNA_ERR_SEV_MASK		GENMASK_U64(31, 24)
+#define AMDXDNA_ERR_MOD_MASK		GENMASK_U64(39, 32)
+#define AMDXDNA_ERR_CLASS_MASK		GENMASK_U64(47, 40)
+
+enum amdxdna_error_num {
+	AMDXDNA_ERROR_NUM_AIE_SATURATION = 3,
+	AMDXDNA_ERROR_NUM_AIE_FP,
+	AMDXDNA_ERROR_NUM_AIE_STREAM,
+	AMDXDNA_ERROR_NUM_AIE_ACCESS,
+	AMDXDNA_ERROR_NUM_AIE_BUS,
+	AMDXDNA_ERROR_NUM_AIE_INSTRUCTION,
+	AMDXDNA_ERROR_NUM_AIE_ECC,
+	AMDXDNA_ERROR_NUM_AIE_LOCK,
+	AMDXDNA_ERROR_NUM_AIE_DMA,
+	AMDXDNA_ERROR_NUM_AIE_MEM_PARITY,
+	AMDXDNA_ERROR_NUM_UNKNOWN = 15,
+};
+
+enum amdxdna_error_module {
+	AMDXDNA_ERROR_MODULE_AIE_CORE = 3,
+	AMDXDNA_ERROR_MODULE_AIE_MEMORY,
+	AMDXDNA_ERROR_MODULE_AIE_SHIM,
+	AMDXDNA_ERROR_MODULE_AIE_NOC,
+	AMDXDNA_ERROR_MODULE_AIE_PL,
+	AMDXDNA_ERROR_MODULE_UNKNOWN = 8,
+};
+
+#define AMDXDNA_ERROR_ENCODE(err_num, err_mod)				\
+	(FIELD_PREP(AMDXDNA_ERR_NUM_MASK, err_num) |			\
+	 FIELD_PREP_CONST(AMDXDNA_ERR_DRV_MASK, AMDXDNA_ERR_DRV_AIE) |	\
+	 FIELD_PREP_CONST(AMDXDNA_ERR_SEV_MASK, AMDXDNA_ERR_SEV_CRITICAL) | \
+	 FIELD_PREP(AMDXDNA_ERR_MOD_MASK, err_mod) |			\
+	 FIELD_PREP_CONST(AMDXDNA_ERR_CLASS_MASK, AMDXDNA_ERR_CLASS_AIE))
+
+#define AMDXDNA_EXTRA_ERR_COL_MASK	GENMASK_U64(7, 0)
+#define AMDXDNA_EXTRA_ERR_ROW_MASK	GENMASK_U64(15, 8)
+
+#define AMDXDNA_EXTRA_ERR_ENCODE(row, col)				\
+	(FIELD_PREP(AMDXDNA_EXTRA_ERR_COL_MASK, col) |			\
+	 FIELD_PREP(AMDXDNA_EXTRA_ERR_ROW_MASK, row))
+
+#endif /* _AMDXDNA_ERROR_H_ */
diff --git a/drivers/accel/amdxdna/amdxdna_gem.c b/drivers/accel/amdxdna/amdxdna_gem.c
index 26831ec69f89..dfa916eeb2d9 100644
--- a/drivers/accel/amdxdna/amdxdna_gem.c
+++ b/drivers/accel/amdxdna/amdxdna_gem.c
@@ -8,6 +8,7 @@
 #include <drm/drm_device.h>
 #include <drm/drm_gem.h>
 #include <drm/drm_gem_shmem_helper.h>
+#include <drm/drm_print.h>
 #include <drm/gpu_scheduler.h>
 #include <linux/dma-buf.h>
 #include <linux/dma-direct.h>
@@ -18,46 +19,86 @@
 #include "amdxdna_ctx.h"
 #include "amdxdna_gem.h"
 #include "amdxdna_pci_drv.h"
+#include "amdxdna_ubuf.h"
 
 #define XDNA_MAX_CMD_BO_SIZE	SZ_32K
 
 MODULE_IMPORT_NS("DMA_BUF");
 
 static int
-amdxdna_gem_insert_node_locked(struct amdxdna_gem_obj *abo, bool use_vmap)
+amdxdna_gem_heap_alloc(struct amdxdna_gem_obj *abo)
 {
 	struct amdxdna_client *client = abo->client;
 	struct amdxdna_dev *xdna = client->xdna;
 	struct amdxdna_mem *mem = &abo->mem;
+	struct amdxdna_gem_obj *heap;
 	u64 offset;
 	u32 align;
 	int ret;
 
+	mutex_lock(&client->mm_lock);
+
+	heap = client->dev_heap;
+	if (!heap) {
+		ret = -EINVAL;
+		goto unlock_out;
+	}
+
+	if (heap->mem.userptr == AMDXDNA_INVALID_ADDR) {
+		XDNA_ERR(xdna, "Invalid dev heap userptr");
+		ret = -EINVAL;
+		goto unlock_out;
+	}
+
+	if (mem->size == 0 || mem->size > heap->mem.size) {
+		XDNA_ERR(xdna, "Invalid dev bo size 0x%lx, limit 0x%lx",
+			 mem->size, heap->mem.size);
+		ret = -EINVAL;
+		goto unlock_out;
+	}
+
 	align = 1 << max(PAGE_SHIFT, xdna->dev_info->dev_mem_buf_shift);
-	ret = drm_mm_insert_node_generic(&abo->dev_heap->mm, &abo->mm_node,
+	ret = drm_mm_insert_node_generic(&heap->mm, &abo->mm_node,
 					 mem->size, align,
 					 0, DRM_MM_INSERT_BEST);
 	if (ret) {
 		XDNA_ERR(xdna, "Failed to alloc dev bo memory, ret %d", ret);
-		return ret;
+		goto unlock_out;
 	}
 
 	mem->dev_addr = abo->mm_node.start;
-	offset = mem->dev_addr - abo->dev_heap->mem.dev_addr;
-	mem->userptr = abo->dev_heap->mem.userptr + offset;
-	mem->pages = &abo->dev_heap->base.pages[offset >> PAGE_SHIFT];
-	mem->nr_pages = mem->size >> PAGE_SHIFT;
-
-	if (use_vmap) {
-		mem->kva = vmap(mem->pages, mem->nr_pages, VM_MAP, PAGE_KERNEL);
-		if (!mem->kva) {
-			XDNA_ERR(xdna, "Failed to vmap");
-			drm_mm_remove_node(&abo->mm_node);
-			return -EFAULT;
-		}
-	}
+	offset = mem->dev_addr - heap->mem.dev_addr;
+	mem->userptr = heap->mem.userptr + offset;
+	mem->kva = heap->mem.kva + offset;
 
-	return 0;
+	drm_gem_object_get(to_gobj(heap));
+
+unlock_out:
+	mutex_unlock(&client->mm_lock);
+
+	return ret;
+}
+
+static void
+amdxdna_gem_destroy_obj(struct amdxdna_gem_obj *abo)
+{
+	mutex_destroy(&abo->lock);
+	kfree(abo);
+}
+
+static void
+amdxdna_gem_heap_free(struct amdxdna_gem_obj *abo)
+{
+	struct amdxdna_gem_obj *heap;
+
+	mutex_lock(&abo->client->mm_lock);
+
+	drm_mm_remove_node(&abo->mm_node);
+
+	heap = abo->client->dev_heap;
+	drm_gem_object_put(to_gobj(heap));
+
+	mutex_unlock(&abo->client->mm_lock);
 }
 
 static bool amdxdna_hmm_invalidate(struct mmu_interval_notifier *mni,
@@ -213,6 +254,20 @@ free_map:
 	return ret;
 }
 
+static void amdxdna_gem_dev_obj_free(struct drm_gem_object *gobj)
+{
+	struct amdxdna_dev *xdna = to_xdna_dev(gobj->dev);
+	struct amdxdna_gem_obj *abo = to_xdna_obj(gobj);
+
+	XDNA_DBG(xdna, "BO type %d xdna_addr 0x%llx", abo->type, abo->mem.dev_addr);
+	if (abo->pinned)
+		amdxdna_gem_unpin(abo);
+
+	amdxdna_gem_heap_free(abo);
+	drm_gem_object_release(gobj);
+	amdxdna_gem_destroy_obj(abo);
+}
+
 static int amdxdna_insert_pages(struct amdxdna_gem_obj *abo,
 				struct vm_area_struct *vma)
 {
@@ -243,7 +298,7 @@ static int amdxdna_insert_pages(struct amdxdna_gem_obj *abo,
 
 	vma->vm_private_data = NULL;
 	vma->vm_ops = NULL;
-	ret = dma_buf_mmap(to_gobj(abo)->dma_buf, vma, 0);
+	ret = dma_buf_mmap(abo->dma_buf, vma, 0);
 	if (ret) {
 		XDNA_ERR(xdna, "Failed to mmap dma buf %d", ret);
 		return ret;
@@ -338,10 +393,45 @@ static const struct dma_buf_ops amdxdna_dmabuf_ops = {
 	.vunmap = drm_gem_dmabuf_vunmap,
 };
 
+static int amdxdna_gem_obj_vmap(struct amdxdna_gem_obj *abo, void **vaddr)
+{
+	struct iosys_map map = IOSYS_MAP_INIT_VADDR(NULL);
+	int ret;
+
+	if (is_import_bo(abo))
+		ret = dma_buf_vmap_unlocked(abo->dma_buf, &map);
+	else
+		ret = drm_gem_vmap(to_gobj(abo), &map);
+
+	*vaddr = map.vaddr;
+	return ret;
+}
+
+static void amdxdna_gem_obj_vunmap(struct amdxdna_gem_obj *abo)
+{
+	struct iosys_map map;
+
+	if (!abo->mem.kva)
+		return;
+
+	iosys_map_set_vaddr(&map, abo->mem.kva);
+
+	if (is_import_bo(abo))
+		dma_buf_vunmap_unlocked(abo->dma_buf, &map);
+	else
+		drm_gem_vunmap(to_gobj(abo), &map);
+}
+
 static struct dma_buf *amdxdna_gem_prime_export(struct drm_gem_object *gobj, int flags)
 {
+	struct amdxdna_gem_obj *abo = to_xdna_obj(gobj);
 	DEFINE_DMA_BUF_EXPORT_INFO(exp_info);
 
+	if (abo->dma_buf) {
+		get_dma_buf(abo->dma_buf);
+		return abo->dma_buf;
+	}
+
 	exp_info.ops = &amdxdna_dmabuf_ops;
 	exp_info.size = gobj->size;
 	exp_info.flags = flags;
@@ -364,7 +454,6 @@ static void amdxdna_gem_obj_free(struct drm_gem_object *gobj)
 {
 	struct amdxdna_dev *xdna = to_xdna_dev(gobj->dev);
 	struct amdxdna_gem_obj *abo = to_xdna_obj(gobj);
-	struct iosys_map map = IOSYS_MAP_INIT_VADDR(abo->mem.kva);
 
 	XDNA_DBG(xdna, "BO type %d xdna_addr 0x%llx", abo->type, abo->mem.dev_addr);
 
@@ -374,23 +463,10 @@ static void amdxdna_gem_obj_free(struct drm_gem_object *gobj)
 	if (abo->pinned)
 		amdxdna_gem_unpin(abo);
 
-	if (abo->type == AMDXDNA_BO_DEV) {
-		mutex_lock(&abo->client->mm_lock);
-		drm_mm_remove_node(&abo->mm_node);
-		mutex_unlock(&abo->client->mm_lock);
-
-		vunmap(abo->mem.kva);
-		drm_gem_object_put(to_gobj(abo->dev_heap));
-		drm_gem_object_release(gobj);
-		mutex_destroy(&abo->lock);
-		kfree(abo);
-		return;
-	}
-
 	if (abo->type == AMDXDNA_BO_DEV_HEAP)
 		drm_mm_takedown(&abo->mm);
 
-	drm_gem_vunmap(gobj, &map);
+	amdxdna_gem_obj_vunmap(abo);
 	mutex_destroy(&abo->lock);
 
 	if (is_import_bo(abo)) {
@@ -402,7 +478,7 @@ static void amdxdna_gem_obj_free(struct drm_gem_object *gobj)
 }
 
 static const struct drm_gem_object_funcs amdxdna_gem_dev_obj_funcs = {
-	.free = amdxdna_gem_obj_free,
+	.free = amdxdna_gem_dev_obj_free,
 };
 
 static const struct drm_gem_object_funcs amdxdna_gem_shmem_funcs = {
@@ -454,6 +530,68 @@ amdxdna_gem_create_object_cb(struct drm_device *dev, size_t size)
 	return to_gobj(abo);
 }
 
+static struct amdxdna_gem_obj *
+amdxdna_gem_create_shmem_object(struct drm_device *dev, size_t size)
+{
+	struct drm_gem_shmem_object *shmem = drm_gem_shmem_create(dev, size);
+
+	if (IS_ERR(shmem))
+		return ERR_CAST(shmem);
+
+	shmem->map_wc = false;
+	return to_xdna_obj(&shmem->base);
+}
+
+static struct amdxdna_gem_obj *
+amdxdna_gem_create_ubuf_object(struct drm_device *dev, struct amdxdna_drm_create_bo *args)
+{
+	struct amdxdna_dev *xdna = to_xdna_dev(dev);
+	enum amdxdna_ubuf_flag flags = 0;
+	struct amdxdna_drm_va_tbl va_tbl;
+	struct drm_gem_object *gobj;
+	struct dma_buf *dma_buf;
+
+	if (copy_from_user(&va_tbl, u64_to_user_ptr(args->vaddr), sizeof(va_tbl))) {
+		XDNA_DBG(xdna, "Access va table failed");
+		return ERR_PTR(-EINVAL);
+	}
+
+	if (va_tbl.num_entries) {
+		if (args->type == AMDXDNA_BO_CMD)
+			flags |= AMDXDNA_UBUF_FLAG_MAP_DMA;
+
+		dma_buf = amdxdna_get_ubuf(dev, flags, va_tbl.num_entries,
+					   u64_to_user_ptr(args->vaddr + sizeof(va_tbl)));
+	} else {
+		dma_buf = dma_buf_get(va_tbl.dmabuf_fd);
+	}
+
+	if (IS_ERR(dma_buf))
+		return ERR_CAST(dma_buf);
+
+	gobj = amdxdna_gem_prime_import(dev, dma_buf);
+	if (IS_ERR(gobj)) {
+		dma_buf_put(dma_buf);
+		return ERR_CAST(gobj);
+	}
+
+	dma_buf_put(dma_buf);
+
+	return to_xdna_obj(gobj);
+}
+
+static struct amdxdna_gem_obj *
+amdxdna_gem_create_object(struct drm_device *dev,
+			  struct amdxdna_drm_create_bo *args)
+{
+	size_t aligned_sz = PAGE_ALIGN(args->size);
+
+	if (args->vaddr)
+		return amdxdna_gem_create_ubuf_object(dev, args);
+
+	return amdxdna_gem_create_shmem_object(dev, aligned_sz);
+}
+
 struct drm_gem_object *
 amdxdna_gem_prime_import(struct drm_device *dev, struct dma_buf *dma_buf)
 {
@@ -505,16 +643,12 @@ amdxdna_drm_alloc_shmem(struct drm_device *dev,
 			struct drm_file *filp)
 {
 	struct amdxdna_client *client = filp->driver_priv;
-	struct drm_gem_shmem_object *shmem;
 	struct amdxdna_gem_obj *abo;
 
-	shmem = drm_gem_shmem_create(dev, args->size);
-	if (IS_ERR(shmem))
-		return ERR_CAST(shmem);
-
-	shmem->map_wc = false;
+	abo = amdxdna_gem_create_object(dev, args);
+	if (IS_ERR(abo))
+		return ERR_CAST(abo);
 
-	abo = to_xdna_obj(&shmem->base);
 	abo->client = client;
 	abo->type = AMDXDNA_BO_SHMEM;
 
@@ -528,7 +662,6 @@ amdxdna_drm_create_dev_heap(struct drm_device *dev,
 {
 	struct amdxdna_client *client = filp->driver_priv;
 	struct amdxdna_dev *xdna = to_xdna_dev(dev);
-	struct drm_gem_shmem_object *shmem;
 	struct amdxdna_gem_obj *abo;
 	int ret;
 
@@ -545,26 +678,31 @@ amdxdna_drm_create_dev_heap(struct drm_device *dev,
 		goto mm_unlock;
 	}
 
-	shmem = drm_gem_shmem_create(dev, args->size);
-	if (IS_ERR(shmem)) {
-		ret = PTR_ERR(shmem);
+	abo = amdxdna_gem_create_object(dev, args);
+	if (IS_ERR(abo)) {
+		ret = PTR_ERR(abo);
 		goto mm_unlock;
 	}
 
-	shmem->map_wc = false;
-	abo = to_xdna_obj(&shmem->base);
-
 	abo->type = AMDXDNA_BO_DEV_HEAP;
 	abo->client = client;
 	abo->mem.dev_addr = client->xdna->dev_info->dev_mem_base;
 	drm_mm_init(&abo->mm, abo->mem.dev_addr, abo->mem.size);
 
+	ret = amdxdna_gem_obj_vmap(abo, &abo->mem.kva);
+	if (ret) {
+		XDNA_ERR(xdna, "Vmap heap bo failed, ret %d", ret);
+		goto release_obj;
+	}
+
 	client->dev_heap = abo;
 	drm_gem_object_get(to_gobj(abo));
 	mutex_unlock(&client->mm_lock);
 
 	return abo;
 
+release_obj:
+	drm_gem_object_put(to_gobj(abo));
 mm_unlock:
 	mutex_unlock(&client->mm_lock);
 	return ERR_PTR(ret);
@@ -573,58 +711,32 @@ mm_unlock:
 struct amdxdna_gem_obj *
 amdxdna_drm_alloc_dev_bo(struct drm_device *dev,
 			 struct amdxdna_drm_create_bo *args,
-			 struct drm_file *filp, bool use_vmap)
+			 struct drm_file *filp)
 {
 	struct amdxdna_client *client = filp->driver_priv;
 	struct amdxdna_dev *xdna = to_xdna_dev(dev);
 	size_t aligned_sz = PAGE_ALIGN(args->size);
-	struct amdxdna_gem_obj *abo, *heap;
+	struct amdxdna_gem_obj *abo;
 	int ret;
 
-	mutex_lock(&client->mm_lock);
-	heap = client->dev_heap;
-	if (!heap) {
-		ret = -EINVAL;
-		goto mm_unlock;
-	}
-
-	if (heap->mem.userptr == AMDXDNA_INVALID_ADDR) {
-		XDNA_ERR(xdna, "Invalid dev heap userptr");
-		ret = -EINVAL;
-		goto mm_unlock;
-	}
-
-	if (args->size > heap->mem.size) {
-		XDNA_ERR(xdna, "Invalid dev bo size 0x%llx, limit 0x%lx",
-			 args->size, heap->mem.size);
-		ret = -EINVAL;
-		goto mm_unlock;
-	}
-
 	abo = amdxdna_gem_create_obj(&xdna->ddev, aligned_sz);
-	if (IS_ERR(abo)) {
-		ret = PTR_ERR(abo);
-		goto mm_unlock;
-	}
+	if (IS_ERR(abo))
+		return abo;
+
 	to_gobj(abo)->funcs = &amdxdna_gem_dev_obj_funcs;
 	abo->type = AMDXDNA_BO_DEV;
 	abo->client = client;
-	abo->dev_heap = heap;
-	ret = amdxdna_gem_insert_node_locked(abo, use_vmap);
+
+	ret = amdxdna_gem_heap_alloc(abo);
 	if (ret) {
 		XDNA_ERR(xdna, "Failed to alloc dev bo memory, ret %d", ret);
-		goto mm_unlock;
+		amdxdna_gem_destroy_obj(abo);
+		return ERR_PTR(ret);
 	}
 
-	drm_gem_object_get(to_gobj(heap));
 	drm_gem_private_object_init(&xdna->ddev, to_gobj(abo), aligned_sz);
 
-	mutex_unlock(&client->mm_lock);
 	return abo;
-
-mm_unlock:
-	mutex_unlock(&client->mm_lock);
-	return ERR_PTR(ret);
 }
 
 static struct amdxdna_gem_obj *
@@ -633,9 +745,7 @@ amdxdna_drm_create_cmd_bo(struct drm_device *dev,
 			  struct drm_file *filp)
 {
 	struct amdxdna_dev *xdna = to_xdna_dev(dev);
-	struct drm_gem_shmem_object *shmem;
 	struct amdxdna_gem_obj *abo;
-	struct iosys_map map;
 	int ret;
 
 	if (args->size > XDNA_MAX_CMD_BO_SIZE) {
@@ -648,27 +758,23 @@ amdxdna_drm_create_cmd_bo(struct drm_device *dev,
 		return ERR_PTR(-EINVAL);
 	}
 
-	shmem = drm_gem_shmem_create(dev, args->size);
-	if (IS_ERR(shmem))
-		return ERR_CAST(shmem);
-
-	shmem->map_wc = false;
-	abo = to_xdna_obj(&shmem->base);
+	abo = amdxdna_gem_create_object(dev, args);
+	if (IS_ERR(abo))
+		return ERR_CAST(abo);
 
 	abo->type = AMDXDNA_BO_CMD;
 	abo->client = filp->driver_priv;
 
-	ret = drm_gem_vmap(to_gobj(abo), &map);
+	ret = amdxdna_gem_obj_vmap(abo, &abo->mem.kva);
 	if (ret) {
 		XDNA_ERR(xdna, "Vmap cmd bo failed, ret %d", ret);
 		goto release_obj;
 	}
-	abo->mem.kva = map.vaddr;
 
 	return abo;
 
 release_obj:
-	drm_gem_shmem_free(shmem);
+	drm_gem_object_put(to_gobj(abo));
 	return ERR_PTR(ret);
 }
 
@@ -679,7 +785,7 @@ int amdxdna_drm_create_bo_ioctl(struct drm_device *dev, void *data, struct drm_f
 	struct amdxdna_gem_obj *abo;
 	int ret;
 
-	if (args->flags || args->vaddr || !args->size)
+	if (args->flags)
 		return -EINVAL;
 
 	XDNA_DBG(xdna, "BO arg type %d vaddr 0x%llx size 0x%llx flags 0x%llx",
@@ -692,7 +798,7 @@ int amdxdna_drm_create_bo_ioctl(struct drm_device *dev, void *data, struct drm_f
 		abo = amdxdna_drm_create_dev_heap(dev, args, filp);
 		break;
 	case AMDXDNA_BO_DEV:
-		abo = amdxdna_drm_alloc_dev_bo(dev, args, filp, false);
+		abo = amdxdna_drm_alloc_dev_bo(dev, args, filp);
 		break;
 	case AMDXDNA_BO_CMD:
 		abo = amdxdna_drm_create_cmd_bo(dev, args, filp);
@@ -724,20 +830,13 @@ int amdxdna_gem_pin_nolock(struct amdxdna_gem_obj *abo)
 	struct amdxdna_dev *xdna = to_xdna_dev(to_gobj(abo)->dev);
 	int ret;
 
+	if (abo->type == AMDXDNA_BO_DEV)
+		abo = abo->client->dev_heap;
+
 	if (is_import_bo(abo))
 		return 0;
 
-	switch (abo->type) {
-	case AMDXDNA_BO_SHMEM:
-	case AMDXDNA_BO_DEV_HEAP:
-		ret = drm_gem_shmem_pin(&abo->base);
-		break;
-	case AMDXDNA_BO_DEV:
-		ret = drm_gem_shmem_pin(&abo->dev_heap->base);
-		break;
-	default:
-		ret = -EOPNOTSUPP;
-	}
+	ret = drm_gem_shmem_pin(&abo->base);
 
 	XDNA_DBG(xdna, "BO type %d ret %d", abo->type, ret);
 	return ret;
@@ -747,9 +846,6 @@ int amdxdna_gem_pin(struct amdxdna_gem_obj *abo)
 {
 	int ret;
 
-	if (abo->type == AMDXDNA_BO_DEV)
-		abo = abo->dev_heap;
-
 	mutex_lock(&abo->lock);
 	ret = amdxdna_gem_pin_nolock(abo);
 	mutex_unlock(&abo->lock);
@@ -759,12 +855,12 @@ int amdxdna_gem_pin(struct amdxdna_gem_obj *abo)
 
 void amdxdna_gem_unpin(struct amdxdna_gem_obj *abo)
 {
+	if (abo->type == AMDXDNA_BO_DEV)
+		abo = abo->client->dev_heap;
+
 	if (is_import_bo(abo))
 		return;
 
-	if (abo->type == AMDXDNA_BO_DEV)
-		abo = abo->dev_heap;
-
 	mutex_lock(&abo->lock);
 	drm_gem_shmem_unpin(&abo->base);
 	mutex_unlock(&abo->lock);
@@ -855,16 +951,21 @@ int amdxdna_drm_sync_bo_ioctl(struct drm_device *dev,
 
 	if (is_import_bo(abo))
 		drm_clflush_sg(abo->base.sgt);
-	else if (abo->type == AMDXDNA_BO_DEV)
-		drm_clflush_pages(abo->mem.pages, abo->mem.nr_pages);
-	else
+	else if (abo->mem.kva)
+		drm_clflush_virt_range(abo->mem.kva + args->offset, args->size);
+	else if (abo->base.pages)
 		drm_clflush_pages(abo->base.pages, gobj->size >> PAGE_SHIFT);
+	else
+		drm_WARN(&xdna->ddev, 1, "Can not get flush memory");
 
 	amdxdna_gem_unpin(abo);
 
 	XDNA_DBG(xdna, "Sync bo %d offset 0x%llx, size 0x%llx\n",
 		 args->handle, args->offset, args->size);
 
+	if (args->direction == SYNC_DIRECT_FROM_DEVICE)
+		ret = amdxdna_hwctx_sync_debug_bo(abo->client, args->handle);
+
 put_obj:
 	drm_gem_object_put(gobj);
 	return ret;
diff --git a/drivers/accel/amdxdna/amdxdna_gem.h b/drivers/accel/amdxdna/amdxdna_gem.h
index aee97e971d6d..f79fc7f3c93b 100644
--- a/drivers/accel/amdxdna/amdxdna_gem.h
+++ b/drivers/accel/amdxdna/amdxdna_gem.h
@@ -7,6 +7,7 @@
 #define _AMDXDNA_GEM_H_
 
 #include <linux/hmm.h>
+#include "amdxdna_pci_drv.h"
 
 struct amdxdna_umap {
 	struct vm_area_struct		*vma;
@@ -41,7 +42,6 @@ struct amdxdna_gem_obj {
 
 	/* Below members is uninitialized when needed */
 	struct drm_mm			mm; /* For AMDXDNA_BO_DEV_HEAP */
-	struct amdxdna_gem_obj		*dev_heap; /* For AMDXDNA_BO_DEV */
 	struct drm_mm_node		mm_node; /* For AMDXDNA_BO_DEV */
 	u32				assigned_hwctx;
 	struct dma_buf			*dma_buf;
@@ -63,6 +63,11 @@ static inline void amdxdna_gem_put_obj(struct amdxdna_gem_obj *abo)
 	drm_gem_object_put(to_gobj(abo));
 }
 
+static inline u64 amdxdna_dev_bo_offset(struct amdxdna_gem_obj *abo)
+{
+	return abo->mem.dev_addr - abo->client->dev_heap->mem.dev_addr;
+}
+
 void amdxdna_umap_put(struct amdxdna_umap *mapp);
 
 struct drm_gem_object *
@@ -72,7 +77,7 @@ amdxdna_gem_prime_import(struct drm_device *dev, struct dma_buf *dma_buf);
 struct amdxdna_gem_obj *
 amdxdna_drm_alloc_dev_bo(struct drm_device *dev,
 			 struct amdxdna_drm_create_bo *args,
-			 struct drm_file *filp, bool use_vmap);
+			 struct drm_file *filp);
 
 int amdxdna_gem_pin_nolock(struct amdxdna_gem_obj *abo);
 int amdxdna_gem_pin(struct amdxdna_gem_obj *abo);
diff --git a/drivers/accel/amdxdna/amdxdna_mailbox.c b/drivers/accel/amdxdna/amdxdna_mailbox.c
index da1ac89bb78f..858df97cd3fb 100644
--- a/drivers/accel/amdxdna/amdxdna_mailbox.c
+++ b/drivers/accel/amdxdna/amdxdna_mailbox.c
@@ -194,7 +194,8 @@ static void mailbox_release_msg(struct mailbox_channel *mb_chann,
 {
 	MB_DBG(mb_chann, "msg_id 0x%x msg opcode 0x%x",
 	       mb_msg->pkg.header.id, mb_msg->pkg.header.opcode);
-	mb_msg->notify_cb(mb_msg->handle, NULL, 0);
+	if (mb_msg->notify_cb)
+		mb_msg->notify_cb(mb_msg->handle, NULL, 0);
 	kfree(mb_msg);
 }
 
@@ -248,7 +249,7 @@ mailbox_get_resp(struct mailbox_channel *mb_chann, struct xdna_msg_header *heade
 {
 	struct mailbox_msg *mb_msg;
 	int msg_id;
-	int ret;
+	int ret = 0;
 
 	msg_id = header->id;
 	if (!mailbox_validate_msgid(msg_id)) {
@@ -265,9 +266,11 @@ mailbox_get_resp(struct mailbox_channel *mb_chann, struct xdna_msg_header *heade
 
 	MB_DBG(mb_chann, "opcode 0x%x size %d id 0x%x",
 	       header->opcode, header->total_size, header->id);
-	ret = mb_msg->notify_cb(mb_msg->handle, data, header->total_size);
-	if (unlikely(ret))
-		MB_ERR(mb_chann, "Message callback ret %d", ret);
+	if (mb_msg->notify_cb) {
+		ret = mb_msg->notify_cb(mb_msg->handle, data, header->total_size);
+		if (unlikely(ret))
+			MB_ERR(mb_chann, "Message callback ret %d", ret);
+	}
 
 	kfree(mb_msg);
 	return ret;
@@ -513,6 +516,7 @@ xdna_mailbox_create_channel(struct mailbox *mb,
 	}
 
 	mb_chann->bad_state = false;
+	mailbox_reg_write(mb_chann, mb_chann->iohub_int_addr, 0);
 
 	MB_DBG(mb_chann, "Mailbox channel created (irq: %d)", mb_chann->msix_irq);
 	return mb_chann;
diff --git a/drivers/accel/amdxdna/amdxdna_mailbox_helper.h b/drivers/accel/amdxdna/amdxdna_mailbox_helper.h
index 710ff8873d61..556c712cad0a 100644
--- a/drivers/accel/amdxdna/amdxdna_mailbox_helper.h
+++ b/drivers/accel/amdxdna/amdxdna_mailbox_helper.h
@@ -16,16 +16,18 @@ struct xdna_notify {
 	u32			*data;
 	size_t			size;
 	int			error;
+	u32			*status;
 };
 
-#define DECLARE_XDNA_MSG_COMMON(name, op, status)			\
+#define DECLARE_XDNA_MSG_COMMON(name, op, s)				\
 	struct name##_req	req = { 0 };				\
-	struct name##_resp	resp = { status	};			\
+	struct name##_resp	resp = { .status = s };			\
 	struct xdna_notify	hdl = {					\
 		.error = 0,						\
 		.data = (u32 *)&resp,					\
 		.size = sizeof(resp),					\
 		.comp = COMPLETION_INITIALIZER_ONSTACK(hdl.comp),	\
+		.status = (u32 *)&resp.status,				\
 	};								\
 	struct xdna_mailbox_msg msg = {					\
 		.send_data = (u8 *)&req,				\
diff --git a/drivers/accel/amdxdna/amdxdna_pci_drv.c b/drivers/accel/amdxdna/amdxdna_pci_drv.c
index f2bf1d374cc7..1973ab67721b 100644
--- a/drivers/accel/amdxdna/amdxdna_pci_drv.c
+++ b/drivers/accel/amdxdna/amdxdna_pci_drv.c
@@ -13,13 +13,11 @@
 #include <drm/gpu_scheduler.h>
 #include <linux/iommu.h>
 #include <linux/pci.h>
-#include <linux/pm_runtime.h>
 
 #include "amdxdna_ctx.h"
 #include "amdxdna_gem.h"
 #include "amdxdna_pci_drv.h"
-
-#define AMDXDNA_AUTOSUSPEND_DELAY	5000 /* milliseconds */
+#include "amdxdna_pm.h"
 
 MODULE_FIRMWARE("amdnpu/1502_00/npu.sbin");
 MODULE_FIRMWARE("amdnpu/17f0_10/npu.sbin");
@@ -27,6 +25,18 @@ MODULE_FIRMWARE("amdnpu/17f0_11/npu.sbin");
 MODULE_FIRMWARE("amdnpu/17f0_20/npu.sbin");
 
 /*
+ * 0.0: Initial version
+ * 0.1: Support getting all hardware contexts by DRM_IOCTL_AMDXDNA_GET_ARRAY
+ * 0.2: Support getting last error hardware error
+ * 0.3: Support firmware debug buffer
+ * 0.4: Support getting resource information
+ * 0.5: Support getting telemetry data
+ * 0.6: Support preemption
+ */
+#define AMDXDNA_DRIVER_MAJOR		0
+#define AMDXDNA_DRIVER_MINOR		6
+
+/*
  * Bind the driver base on (vendor_id, device_id) pair and later use the
  * (device_id, rev_id) pair as a key to select the devices. The devices with
  * same device_id have very similar interface to host driver.
@@ -54,17 +64,9 @@ static int amdxdna_drm_open(struct drm_device *ddev, struct drm_file *filp)
 	struct amdxdna_client *client;
 	int ret;
 
-	ret = pm_runtime_resume_and_get(ddev->dev);
-	if (ret) {
-		XDNA_ERR(xdna, "Failed to get rpm, ret %d", ret);
-		return ret;
-	}
-
 	client = kzalloc(sizeof(*client), GFP_KERNEL);
-	if (!client) {
-		ret = -ENOMEM;
-		goto put_rpm;
-	}
+	if (!client)
+		return -ENOMEM;
 
 	client->pid = pid_nr(rcu_access_pointer(filp->pid));
 	client->xdna = xdna;
@@ -81,7 +83,6 @@ static int amdxdna_drm_open(struct drm_device *ddev, struct drm_file *filp)
 		ret = -ENODEV;
 		goto unbind_sva;
 	}
-	mutex_init(&client->hwctx_lock);
 	init_srcu_struct(&client->hwctx_srcu);
 	xa_init_flags(&client->hwctx_xa, XA_FLAGS_ALLOC);
 	mutex_init(&client->mm_lock);
@@ -100,9 +101,6 @@ unbind_sva:
 	iommu_sva_unbind_device(client->sva);
 failed:
 	kfree(client);
-put_rpm:
-	pm_runtime_mark_last_busy(ddev->dev);
-	pm_runtime_put_autosuspend(ddev->dev);
 
 	return ret;
 }
@@ -116,7 +114,6 @@ static void amdxdna_drm_close(struct drm_device *ddev, struct drm_file *filp)
 
 	xa_destroy(&client->hwctx_xa);
 	cleanup_srcu_struct(&client->hwctx_srcu);
-	mutex_destroy(&client->hwctx_lock);
 	mutex_destroy(&client->mm_lock);
 	if (client->dev_heap)
 		drm_gem_object_put(to_gobj(client->dev_heap));
@@ -125,8 +122,6 @@ static void amdxdna_drm_close(struct drm_device *ddev, struct drm_file *filp)
 
 	XDNA_DBG(xdna, "pid %d closed", client->pid);
 	kfree(client);
-	pm_runtime_mark_last_busy(ddev->dev);
-	pm_runtime_put_autosuspend(ddev->dev);
 }
 
 static int amdxdna_flush(struct file *f, fl_owner_t id)
@@ -142,8 +137,8 @@ static int amdxdna_flush(struct file *f, fl_owner_t id)
 
 	mutex_lock(&xdna->dev_lock);
 	list_del_init(&client->node);
-	mutex_unlock(&xdna->dev_lock);
 	amdxdna_hwctx_remove_all(client);
+	mutex_unlock(&xdna->dev_lock);
 
 	drm_dev_exit(idx);
 	return 0;
@@ -166,6 +161,23 @@ static int amdxdna_drm_get_info_ioctl(struct drm_device *dev, void *data, struct
 	return ret;
 }
 
+static int amdxdna_drm_get_array_ioctl(struct drm_device *dev, void *data,
+				       struct drm_file *filp)
+{
+	struct amdxdna_client *client = filp->driver_priv;
+	struct amdxdna_dev *xdna = to_xdna_dev(dev);
+	struct amdxdna_drm_get_array *args = data;
+
+	if (!xdna->dev_info->ops->get_array)
+		return -EOPNOTSUPP;
+
+	if (args->pad || !args->num_element || !args->element_size)
+		return -EINVAL;
+
+	guard(mutex)(&xdna->dev_lock);
+	return xdna->dev_info->ops->get_array(client, args);
+}
+
 static int amdxdna_drm_set_state_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 {
 	struct amdxdna_client *client = filp->driver_priv;
@@ -197,6 +209,7 @@ static const struct drm_ioctl_desc amdxdna_drm_ioctls[] = {
 	DRM_IOCTL_DEF_DRV(AMDXDNA_EXEC_CMD, amdxdna_drm_submit_cmd_ioctl, 0),
 	/* AIE hardware */
 	DRM_IOCTL_DEF_DRV(AMDXDNA_GET_INFO, amdxdna_drm_get_info_ioctl, 0),
+	DRM_IOCTL_DEF_DRV(AMDXDNA_GET_ARRAY, amdxdna_drm_get_array_ioctl, 0),
 	DRM_IOCTL_DEF_DRV(AMDXDNA_SET_STATE, amdxdna_drm_set_state_ioctl, DRM_ROOT_ONLY),
 };
 
@@ -220,6 +233,8 @@ const struct drm_driver amdxdna_drm_drv = {
 	.fops = &amdxdna_fops,
 	.name = "amdxdna_accel_driver",
 	.desc = "AMD XDNA DRM implementation",
+	.major = AMDXDNA_DRIVER_MAJOR,
+	.minor = AMDXDNA_DRIVER_MINOR,
 	.open = amdxdna_drm_open,
 	.postclose = amdxdna_drm_close,
 	.ioctls = amdxdna_drm_ioctls,
@@ -285,19 +300,12 @@ static int amdxdna_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 		goto failed_dev_fini;
 	}
 
-	pm_runtime_set_autosuspend_delay(dev, AMDXDNA_AUTOSUSPEND_DELAY);
-	pm_runtime_use_autosuspend(dev);
-	pm_runtime_allow(dev);
-
 	ret = drm_dev_register(&xdna->ddev, 0);
 	if (ret) {
 		XDNA_ERR(xdna, "DRM register failed, ret %d", ret);
-		pm_runtime_forbid(dev);
 		goto failed_sysfs_fini;
 	}
 
-	pm_runtime_mark_last_busy(dev);
-	pm_runtime_put_autosuspend(dev);
 	return 0;
 
 failed_sysfs_fini:
@@ -314,14 +322,10 @@ destroy_notifier_wq:
 static void amdxdna_remove(struct pci_dev *pdev)
 {
 	struct amdxdna_dev *xdna = pci_get_drvdata(pdev);
-	struct device *dev = &pdev->dev;
 	struct amdxdna_client *client;
 
 	destroy_workqueue(xdna->notifier_wq);
 
-	pm_runtime_get_noresume(dev);
-	pm_runtime_forbid(dev);
-
 	drm_dev_unplug(&xdna->ddev);
 	amdxdna_sysfs_fini(xdna);
 
@@ -330,11 +334,8 @@ static void amdxdna_remove(struct pci_dev *pdev)
 					  struct amdxdna_client, node);
 	while (client) {
 		list_del_init(&client->node);
-		mutex_unlock(&xdna->dev_lock);
-
 		amdxdna_hwctx_remove_all(client);
 
-		mutex_lock(&xdna->dev_lock);
 		client = list_first_entry_or_null(&xdna->client_list,
 						  struct amdxdna_client, node);
 	}
@@ -343,89 +344,9 @@ static void amdxdna_remove(struct pci_dev *pdev)
 	mutex_unlock(&xdna->dev_lock);
 }
 
-static int amdxdna_dev_suspend_nolock(struct amdxdna_dev *xdna)
-{
-	if (xdna->dev_info->ops->suspend)
-		xdna->dev_info->ops->suspend(xdna);
-
-	return 0;
-}
-
-static int amdxdna_dev_resume_nolock(struct amdxdna_dev *xdna)
-{
-	if (xdna->dev_info->ops->resume)
-		return xdna->dev_info->ops->resume(xdna);
-
-	return 0;
-}
-
-static int amdxdna_pmops_suspend(struct device *dev)
-{
-	struct amdxdna_dev *xdna = pci_get_drvdata(to_pci_dev(dev));
-	struct amdxdna_client *client;
-
-	mutex_lock(&xdna->dev_lock);
-	list_for_each_entry(client, &xdna->client_list, node)
-		amdxdna_hwctx_suspend(client);
-
-	amdxdna_dev_suspend_nolock(xdna);
-	mutex_unlock(&xdna->dev_lock);
-
-	return 0;
-}
-
-static int amdxdna_pmops_resume(struct device *dev)
-{
-	struct amdxdna_dev *xdna = pci_get_drvdata(to_pci_dev(dev));
-	struct amdxdna_client *client;
-	int ret;
-
-	XDNA_INFO(xdna, "firmware resuming...");
-	mutex_lock(&xdna->dev_lock);
-	ret = amdxdna_dev_resume_nolock(xdna);
-	if (ret) {
-		XDNA_ERR(xdna, "resume NPU firmware failed");
-		mutex_unlock(&xdna->dev_lock);
-		return ret;
-	}
-
-	XDNA_INFO(xdna, "hardware context resuming...");
-	list_for_each_entry(client, &xdna->client_list, node)
-		amdxdna_hwctx_resume(client);
-	mutex_unlock(&xdna->dev_lock);
-
-	return 0;
-}
-
-static int amdxdna_rpmops_suspend(struct device *dev)
-{
-	struct amdxdna_dev *xdna = pci_get_drvdata(to_pci_dev(dev));
-	int ret;
-
-	mutex_lock(&xdna->dev_lock);
-	ret = amdxdna_dev_suspend_nolock(xdna);
-	mutex_unlock(&xdna->dev_lock);
-
-	XDNA_DBG(xdna, "Runtime suspend done ret: %d", ret);
-	return ret;
-}
-
-static int amdxdna_rpmops_resume(struct device *dev)
-{
-	struct amdxdna_dev *xdna = pci_get_drvdata(to_pci_dev(dev));
-	int ret;
-
-	mutex_lock(&xdna->dev_lock);
-	ret = amdxdna_dev_resume_nolock(xdna);
-	mutex_unlock(&xdna->dev_lock);
-
-	XDNA_DBG(xdna, "Runtime resume done ret: %d", ret);
-	return ret;
-}
-
 static const struct dev_pm_ops amdxdna_pm_ops = {
-	SYSTEM_SLEEP_PM_OPS(amdxdna_pmops_suspend, amdxdna_pmops_resume)
-	RUNTIME_PM_OPS(amdxdna_rpmops_suspend, amdxdna_rpmops_resume, NULL)
+	SYSTEM_SLEEP_PM_OPS(amdxdna_pm_suspend, amdxdna_pm_resume)
+	RUNTIME_PM_OPS(amdxdna_pm_suspend, amdxdna_pm_resume, NULL)
 };
 
 static struct pci_driver amdxdna_pci_driver = {
diff --git a/drivers/accel/amdxdna/amdxdna_pci_drv.h b/drivers/accel/amdxdna/amdxdna_pci_drv.h
index ab79600911aa..c99477f5e454 100644
--- a/drivers/accel/amdxdna/amdxdna_pci_drv.h
+++ b/drivers/accel/amdxdna/amdxdna_pci_drv.h
@@ -6,6 +6,7 @@
 #ifndef _AMDXDNA_PCI_DRV_H_
 #define _AMDXDNA_PCI_DRV_H_
 
+#include <drm/drm_print.h>
 #include <linux/workqueue.h>
 #include <linux/xarray.h>
 
@@ -50,16 +51,16 @@ struct amdxdna_dev_ops {
 	int (*init)(struct amdxdna_dev *xdna);
 	void (*fini)(struct amdxdna_dev *xdna);
 	int (*resume)(struct amdxdna_dev *xdna);
-	void (*suspend)(struct amdxdna_dev *xdna);
+	int (*suspend)(struct amdxdna_dev *xdna);
 	int (*hwctx_init)(struct amdxdna_hwctx *hwctx);
 	void (*hwctx_fini)(struct amdxdna_hwctx *hwctx);
 	int (*hwctx_config)(struct amdxdna_hwctx *hwctx, u32 type, u64 value, void *buf, u32 size);
+	int (*hwctx_sync_debug_bo)(struct amdxdna_hwctx *hwctx, u32 debug_bo_hdl);
 	void (*hmm_invalidate)(struct amdxdna_gem_obj *abo, unsigned long cur_seq);
-	void (*hwctx_suspend)(struct amdxdna_hwctx *hwctx);
-	void (*hwctx_resume)(struct amdxdna_hwctx *hwctx);
 	int (*cmd_submit)(struct amdxdna_hwctx *hwctx, struct amdxdna_sched_job *job, u64 *seq);
 	int (*get_aie_info)(struct amdxdna_client *client, struct amdxdna_drm_get_info *args);
 	int (*set_aie_state)(struct amdxdna_client *client, struct amdxdna_drm_set_state *args);
+	int (*get_array)(struct amdxdna_client *client, struct amdxdna_drm_get_array *args);
 };
 
 /*
@@ -100,6 +101,7 @@ struct amdxdna_dev {
 	struct amdxdna_fw_ver		fw_ver;
 	struct rw_semaphore		notifier_lock; /* for mmu notifier*/
 	struct workqueue_struct		*notifier_wq;
+	bool				rpm_on;
 };
 
 /*
@@ -118,8 +120,6 @@ struct amdxdna_device_id {
 struct amdxdna_client {
 	struct list_head		node;
 	pid_t				pid;
-	struct mutex			hwctx_lock; /* protect hwctx */
-	/* do NOT wait this srcu when hwctx_lock is held */
 	struct srcu_struct		hwctx_srcu;
 	struct xarray			hwctx_xa;
 	u32				next_hwctxid;
diff --git a/drivers/accel/amdxdna/amdxdna_pm.c b/drivers/accel/amdxdna/amdxdna_pm.c
new file mode 100644
index 000000000000..fa38e65d617c
--- /dev/null
+++ b/drivers/accel/amdxdna/amdxdna_pm.c
@@ -0,0 +1,94 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2025, Advanced Micro Devices, Inc.
+ */
+
+#include <drm/amdxdna_accel.h>
+#include <drm/drm_drv.h>
+#include <linux/pm_runtime.h>
+
+#include "amdxdna_pm.h"
+
+#define AMDXDNA_AUTOSUSPEND_DELAY	5000 /* milliseconds */
+
+int amdxdna_pm_suspend(struct device *dev)
+{
+	struct amdxdna_dev *xdna = to_xdna_dev(dev_get_drvdata(dev));
+	int ret = -EOPNOTSUPP;
+	bool rpm;
+
+	if (xdna->dev_info->ops->suspend) {
+		rpm = xdna->rpm_on;
+		xdna->rpm_on = false;
+		ret = xdna->dev_info->ops->suspend(xdna);
+		xdna->rpm_on = rpm;
+	}
+
+	XDNA_DBG(xdna, "Suspend done ret %d", ret);
+	return ret;
+}
+
+int amdxdna_pm_resume(struct device *dev)
+{
+	struct amdxdna_dev *xdna = to_xdna_dev(dev_get_drvdata(dev));
+	int ret = -EOPNOTSUPP;
+	bool rpm;
+
+	if (xdna->dev_info->ops->resume) {
+		rpm = xdna->rpm_on;
+		xdna->rpm_on = false;
+		ret = xdna->dev_info->ops->resume(xdna);
+		xdna->rpm_on = rpm;
+	}
+
+	XDNA_DBG(xdna, "Resume done ret %d", ret);
+	return ret;
+}
+
+int amdxdna_pm_resume_get(struct amdxdna_dev *xdna)
+{
+	struct device *dev = xdna->ddev.dev;
+	int ret;
+
+	if (!xdna->rpm_on)
+		return 0;
+
+	ret = pm_runtime_resume_and_get(dev);
+	if (ret) {
+		XDNA_ERR(xdna, "Resume failed: %d", ret);
+		pm_runtime_set_suspended(dev);
+	}
+
+	return ret;
+}
+
+void amdxdna_pm_suspend_put(struct amdxdna_dev *xdna)
+{
+	struct device *dev = xdna->ddev.dev;
+
+	if (!xdna->rpm_on)
+		return;
+
+	pm_runtime_put_autosuspend(dev);
+}
+
+void amdxdna_pm_init(struct amdxdna_dev *xdna)
+{
+	struct device *dev = xdna->ddev.dev;
+
+	pm_runtime_set_active(dev);
+	pm_runtime_set_autosuspend_delay(dev, AMDXDNA_AUTOSUSPEND_DELAY);
+	pm_runtime_use_autosuspend(dev);
+	pm_runtime_allow(dev);
+	pm_runtime_put_autosuspend(dev);
+	xdna->rpm_on = true;
+}
+
+void amdxdna_pm_fini(struct amdxdna_dev *xdna)
+{
+	struct device *dev = xdna->ddev.dev;
+
+	xdna->rpm_on = false;
+	pm_runtime_get_noresume(dev);
+	pm_runtime_forbid(dev);
+}
diff --git a/drivers/accel/amdxdna/amdxdna_pm.h b/drivers/accel/amdxdna/amdxdna_pm.h
new file mode 100644
index 000000000000..77b2d6e45570
--- /dev/null
+++ b/drivers/accel/amdxdna/amdxdna_pm.h
@@ -0,0 +1,18 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2025, Advanced Micro Devices, Inc.
+ */
+
+#ifndef _AMDXDNA_PM_H_
+#define _AMDXDNA_PM_H_
+
+#include "amdxdna_pci_drv.h"
+
+int amdxdna_pm_suspend(struct device *dev);
+int amdxdna_pm_resume(struct device  *dev);
+int amdxdna_pm_resume_get(struct amdxdna_dev *xdna);
+void amdxdna_pm_suspend_put(struct amdxdna_dev *xdna);
+void amdxdna_pm_init(struct amdxdna_dev *xdna);
+void amdxdna_pm_fini(struct amdxdna_dev *xdna);
+
+#endif /* _AMDXDNA_PM_H_ */
diff --git a/drivers/accel/amdxdna/amdxdna_ubuf.c b/drivers/accel/amdxdna/amdxdna_ubuf.c
new file mode 100644
index 000000000000..077b2261cf2a
--- /dev/null
+++ b/drivers/accel/amdxdna/amdxdna_ubuf.c
@@ -0,0 +1,232 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2025, Advanced Micro Devices, Inc.
+ */
+
+#include <drm/amdxdna_accel.h>
+#include <drm/drm_device.h>
+#include <drm/drm_print.h>
+#include <linux/dma-buf.h>
+#include <linux/pagemap.h>
+#include <linux/vmalloc.h>
+
+#include "amdxdna_pci_drv.h"
+#include "amdxdna_ubuf.h"
+
+struct amdxdna_ubuf_priv {
+	struct page **pages;
+	u64 nr_pages;
+	enum amdxdna_ubuf_flag flags;
+	struct mm_struct *mm;
+};
+
+static struct sg_table *amdxdna_ubuf_map(struct dma_buf_attachment *attach,
+					 enum dma_data_direction direction)
+{
+	struct amdxdna_ubuf_priv *ubuf = attach->dmabuf->priv;
+	struct sg_table *sg;
+	int ret;
+
+	sg = kzalloc(sizeof(*sg), GFP_KERNEL);
+	if (!sg)
+		return ERR_PTR(-ENOMEM);
+
+	ret = sg_alloc_table_from_pages(sg, ubuf->pages, ubuf->nr_pages, 0,
+					ubuf->nr_pages << PAGE_SHIFT, GFP_KERNEL);
+	if (ret)
+		return ERR_PTR(ret);
+
+	if (ubuf->flags & AMDXDNA_UBUF_FLAG_MAP_DMA) {
+		ret = dma_map_sgtable(attach->dev, sg, direction, 0);
+		if (ret)
+			return ERR_PTR(ret);
+	}
+
+	return sg;
+}
+
+static void amdxdna_ubuf_unmap(struct dma_buf_attachment *attach,
+			       struct sg_table *sg,
+			       enum dma_data_direction direction)
+{
+	struct amdxdna_ubuf_priv *ubuf = attach->dmabuf->priv;
+
+	if (ubuf->flags & AMDXDNA_UBUF_FLAG_MAP_DMA)
+		dma_unmap_sgtable(attach->dev, sg, direction, 0);
+
+	sg_free_table(sg);
+	kfree(sg);
+}
+
+static void amdxdna_ubuf_release(struct dma_buf *dbuf)
+{
+	struct amdxdna_ubuf_priv *ubuf = dbuf->priv;
+
+	unpin_user_pages(ubuf->pages, ubuf->nr_pages);
+	kvfree(ubuf->pages);
+	atomic64_sub(ubuf->nr_pages, &ubuf->mm->pinned_vm);
+	mmdrop(ubuf->mm);
+	kfree(ubuf);
+}
+
+static vm_fault_t amdxdna_ubuf_vm_fault(struct vm_fault *vmf)
+{
+	struct vm_area_struct *vma = vmf->vma;
+	struct amdxdna_ubuf_priv *ubuf;
+	unsigned long pfn;
+	pgoff_t pgoff;
+
+	ubuf = vma->vm_private_data;
+	pgoff = (vmf->address - vma->vm_start) >> PAGE_SHIFT;
+
+	pfn = page_to_pfn(ubuf->pages[pgoff]);
+	return vmf_insert_pfn(vma, vmf->address, pfn);
+}
+
+static const struct vm_operations_struct amdxdna_ubuf_vm_ops = {
+	.fault = amdxdna_ubuf_vm_fault,
+};
+
+static int amdxdna_ubuf_mmap(struct dma_buf *dbuf, struct vm_area_struct *vma)
+{
+	struct amdxdna_ubuf_priv *ubuf = dbuf->priv;
+
+	vma->vm_ops = &amdxdna_ubuf_vm_ops;
+	vma->vm_private_data = ubuf;
+	vm_flags_set(vma, VM_PFNMAP | VM_DONTEXPAND | VM_DONTDUMP);
+
+	return 0;
+}
+
+static int amdxdna_ubuf_vmap(struct dma_buf *dbuf, struct iosys_map *map)
+{
+	struct amdxdna_ubuf_priv *ubuf = dbuf->priv;
+	void *kva;
+
+	kva = vmap(ubuf->pages, ubuf->nr_pages, VM_MAP, PAGE_KERNEL);
+	if (!kva)
+		return -EINVAL;
+
+	iosys_map_set_vaddr(map, kva);
+	return 0;
+}
+
+static void amdxdna_ubuf_vunmap(struct dma_buf *dbuf, struct iosys_map *map)
+{
+	vunmap(map->vaddr);
+}
+
+static const struct dma_buf_ops amdxdna_ubuf_dmabuf_ops = {
+	.map_dma_buf = amdxdna_ubuf_map,
+	.unmap_dma_buf = amdxdna_ubuf_unmap,
+	.release = amdxdna_ubuf_release,
+	.mmap = amdxdna_ubuf_mmap,
+	.vmap = amdxdna_ubuf_vmap,
+	.vunmap = amdxdna_ubuf_vunmap,
+};
+
+struct dma_buf *amdxdna_get_ubuf(struct drm_device *dev,
+				 enum amdxdna_ubuf_flag flags,
+				 u32 num_entries, void __user *va_entries)
+{
+	struct amdxdna_dev *xdna = to_xdna_dev(dev);
+	unsigned long lock_limit, new_pinned;
+	struct amdxdna_drm_va_entry *va_ent;
+	struct amdxdna_ubuf_priv *ubuf;
+	u32 npages, start = 0;
+	struct dma_buf *dbuf;
+	int i, ret;
+	DEFINE_DMA_BUF_EXPORT_INFO(exp_info);
+
+	if (!can_do_mlock())
+		return ERR_PTR(-EPERM);
+
+	ubuf = kzalloc(sizeof(*ubuf), GFP_KERNEL);
+	if (!ubuf)
+		return ERR_PTR(-ENOMEM);
+
+	ubuf->flags = flags;
+	ubuf->mm = current->mm;
+	mmgrab(ubuf->mm);
+
+	va_ent = kvcalloc(num_entries, sizeof(*va_ent), GFP_KERNEL);
+	if (!va_ent) {
+		ret = -ENOMEM;
+		goto free_ubuf;
+	}
+
+	if (copy_from_user(va_ent, va_entries, sizeof(*va_ent) * num_entries)) {
+		XDNA_DBG(xdna, "Access va entries failed");
+		ret = -EINVAL;
+		goto free_ent;
+	}
+
+	for (i = 0, exp_info.size = 0; i < num_entries; i++) {
+		if (!IS_ALIGNED(va_ent[i].vaddr, PAGE_SIZE) ||
+		    !IS_ALIGNED(va_ent[i].len, PAGE_SIZE)) {
+			XDNA_ERR(xdna, "Invalid address or len %llx, %llx",
+				 va_ent[i].vaddr, va_ent[i].len);
+			ret = -EINVAL;
+			goto free_ent;
+		}
+
+		exp_info.size += va_ent[i].len;
+	}
+
+	ubuf->nr_pages = exp_info.size >> PAGE_SHIFT;
+	lock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
+	new_pinned = atomic64_add_return(ubuf->nr_pages, &ubuf->mm->pinned_vm);
+	if (new_pinned > lock_limit && !capable(CAP_IPC_LOCK)) {
+		XDNA_DBG(xdna, "New pin %ld, limit %ld, cap %d",
+			 new_pinned, lock_limit, capable(CAP_IPC_LOCK));
+		ret = -ENOMEM;
+		goto sub_pin_cnt;
+	}
+
+	ubuf->pages = kvmalloc_array(ubuf->nr_pages, sizeof(*ubuf->pages), GFP_KERNEL);
+	if (!ubuf->pages) {
+		ret = -ENOMEM;
+		goto sub_pin_cnt;
+	}
+
+	for (i = 0; i < num_entries; i++) {
+		npages = va_ent[i].len >> PAGE_SHIFT;
+
+		ret = pin_user_pages_fast(va_ent[i].vaddr, npages,
+					  FOLL_WRITE | FOLL_LONGTERM,
+					  &ubuf->pages[start]);
+		if (ret < 0 || ret != npages) {
+			ret = -ENOMEM;
+			XDNA_ERR(xdna, "Failed to pin pages ret %d", ret);
+			goto destroy_pages;
+		}
+
+		start += ret;
+	}
+
+	exp_info.ops = &amdxdna_ubuf_dmabuf_ops;
+	exp_info.priv = ubuf;
+	exp_info.flags = O_RDWR | O_CLOEXEC;
+
+	dbuf = dma_buf_export(&exp_info);
+	if (IS_ERR(dbuf)) {
+		ret = PTR_ERR(dbuf);
+		goto destroy_pages;
+	}
+	kvfree(va_ent);
+
+	return dbuf;
+
+destroy_pages:
+	if (start)
+		unpin_user_pages(ubuf->pages, start);
+	kvfree(ubuf->pages);
+sub_pin_cnt:
+	atomic64_sub(ubuf->nr_pages, &ubuf->mm->pinned_vm);
+free_ent:
+	kvfree(va_ent);
+free_ubuf:
+	mmdrop(ubuf->mm);
+	kfree(ubuf);
+	return ERR_PTR(ret);
+}
diff --git a/drivers/accel/amdxdna/amdxdna_ubuf.h b/drivers/accel/amdxdna/amdxdna_ubuf.h
new file mode 100644
index 000000000000..e5cb3bdb3ec9
--- /dev/null
+++ b/drivers/accel/amdxdna/amdxdna_ubuf.h
@@ -0,0 +1,19 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2025, Advanced Micro Devices, Inc.
+ */
+#ifndef _AMDXDNA_UBUF_H_
+#define _AMDXDNA_UBUF_H_
+
+#include <drm/drm_device.h>
+#include <linux/dma-buf.h>
+
+enum amdxdna_ubuf_flag {
+	AMDXDNA_UBUF_FLAG_MAP_DMA = 1,
+};
+
+struct dma_buf *amdxdna_get_ubuf(struct drm_device *dev,
+				 enum amdxdna_ubuf_flag flags,
+				 u32 num_entries, void __user *va_entries);
+
+#endif /* _AMDXDNA_UBUF_H_ */
diff --git a/drivers/accel/amdxdna/npu1_regs.c b/drivers/accel/amdxdna/npu1_regs.c
index e4f6dac7d00f..ec407f3b48fc 100644
--- a/drivers/accel/amdxdna/npu1_regs.c
+++ b/drivers/accel/amdxdna/npu1_regs.c
@@ -46,6 +46,7 @@
 
 const struct rt_config npu1_default_rt_cfg[] = {
 	{ 2, 1, AIE2_RT_CFG_INIT }, /* PDI APP LOAD MODE */
+	{ 4, 1, AIE2_RT_CFG_INIT }, /* Debug BO */
 	{ 1, 1, AIE2_RT_CFG_CLK_GATING }, /* Clock gating on */
 	{ 0 },
 };
@@ -62,16 +63,23 @@ const struct dpm_clk_freq npu1_dpm_clk_table[] = {
 	{ 0 }
 };
 
+static const struct aie2_fw_feature_tbl npu1_fw_feature_table[] = {
+	{ .feature = AIE2_NPU_COMMAND, .min_minor = 8 },
+	{ 0 }
+};
+
 static const struct amdxdna_dev_priv npu1_dev_priv = {
 	.fw_path        = "amdnpu/1502_00/npu.sbin",
 	.protocol_major = 0x5,
 	.protocol_minor = 0x7,
 	.rt_config	= npu1_default_rt_cfg,
 	.dpm_clk_tbl	= npu1_dpm_clk_table,
+	.fw_feature_tbl = npu1_fw_feature_table,
 	.col_align	= COL_ALIGN_NONE,
 	.mbox_dev_addr  = NPU1_MBOX_BAR_BASE,
 	.mbox_size      = 0, /* Use BAR size */
 	.sram_dev_addr  = NPU1_SRAM_BAR_BASE,
+	.hwctx_limit    = 6,
 	.sram_offs      = {
 		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU1_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
 		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU1_SRAM, MPNPU_SRAM_I2X_MAILBOX_15),
diff --git a/drivers/accel/amdxdna/npu2_regs.c b/drivers/accel/amdxdna/npu2_regs.c
index a081cac75ee0..86f87d0d1354 100644
--- a/drivers/accel/amdxdna/npu2_regs.c
+++ b/drivers/accel/amdxdna/npu2_regs.c
@@ -67,10 +67,12 @@ static const struct amdxdna_dev_priv npu2_dev_priv = {
 	.protocol_minor = 0x6,
 	.rt_config	= npu4_default_rt_cfg,
 	.dpm_clk_tbl	= npu4_dpm_clk_table,
+	.fw_feature_tbl = npu4_fw_feature_table,
 	.col_align	= COL_ALIGN_NATURE,
 	.mbox_dev_addr  = NPU2_MBOX_BAR_BASE,
 	.mbox_size      = 0, /* Use BAR size */
 	.sram_dev_addr  = NPU2_SRAM_BAR_BASE,
+	.hwctx_limit    = 16,
 	.sram_offs      = {
 		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU2_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
 		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU2_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
diff --git a/drivers/accel/amdxdna/npu4_regs.c b/drivers/accel/amdxdna/npu4_regs.c
index 9f2e33182ec6..986a5f28ba24 100644
--- a/drivers/accel/amdxdna/npu4_regs.c
+++ b/drivers/accel/amdxdna/npu4_regs.c
@@ -63,10 +63,14 @@
 
 const struct rt_config npu4_default_rt_cfg[] = {
 	{ 5, 1, AIE2_RT_CFG_INIT }, /* PDI APP LOAD MODE */
+	{ 10, 1, AIE2_RT_CFG_INIT }, /* DEBUG BUF */
+	{ 14, 0, AIE2_RT_CFG_INIT, BIT_U64(AIE2_PREEMPT) }, /* Frame boundary preemption */
 	{ 1, 1, AIE2_RT_CFG_CLK_GATING }, /* Clock gating on */
 	{ 2, 1, AIE2_RT_CFG_CLK_GATING }, /* Clock gating on */
 	{ 3, 1, AIE2_RT_CFG_CLK_GATING }, /* Clock gating on */
 	{ 4, 1, AIE2_RT_CFG_CLK_GATING }, /* Clock gating on */
+	{ 13, 0, AIE2_RT_CFG_FORCE_PREEMPT },
+	{ 14, 0, AIE2_RT_CFG_FRAME_BOUNDARY_PREEMPT },
 	{ 0 },
 };
 
@@ -82,16 +86,24 @@ const struct dpm_clk_freq npu4_dpm_clk_table[] = {
 	{ 0 }
 };
 
+const struct aie2_fw_feature_tbl npu4_fw_feature_table[] = {
+	{ .feature = AIE2_NPU_COMMAND, .min_minor = 15 },
+	{ .feature = AIE2_PREEMPT, .min_minor = 12 },
+	{ 0 }
+};
+
 static const struct amdxdna_dev_priv npu4_dev_priv = {
 	.fw_path        = "amdnpu/17f0_10/npu.sbin",
 	.protocol_major = 0x6,
 	.protocol_minor = 12,
 	.rt_config	= npu4_default_rt_cfg,
 	.dpm_clk_tbl	= npu4_dpm_clk_table,
+	.fw_feature_tbl = npu4_fw_feature_table,
 	.col_align	= COL_ALIGN_NATURE,
 	.mbox_dev_addr  = NPU4_MBOX_BAR_BASE,
 	.mbox_size      = 0, /* Use BAR size */
 	.sram_dev_addr  = NPU4_SRAM_BAR_BASE,
+	.hwctx_limit    = 16,
 	.sram_offs      = {
 		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU4_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
 		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU4_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
diff --git a/drivers/accel/amdxdna/npu5_regs.c b/drivers/accel/amdxdna/npu5_regs.c
index 5f1cf83461c4..75ad97f0b937 100644
--- a/drivers/accel/amdxdna/npu5_regs.c
+++ b/drivers/accel/amdxdna/npu5_regs.c
@@ -67,10 +67,12 @@ static const struct amdxdna_dev_priv npu5_dev_priv = {
 	.protocol_minor = 12,
 	.rt_config	= npu4_default_rt_cfg,
 	.dpm_clk_tbl	= npu4_dpm_clk_table,
+	.fw_feature_tbl = npu4_fw_feature_table,
 	.col_align	= COL_ALIGN_NATURE,
 	.mbox_dev_addr  = NPU5_MBOX_BAR_BASE,
 	.mbox_size      = 0, /* Use BAR size */
 	.sram_dev_addr  = NPU5_SRAM_BAR_BASE,
+	.hwctx_limit    = 16,
 	.sram_offs      = {
 		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU5_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
 		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU5_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
diff --git a/drivers/accel/amdxdna/npu6_regs.c b/drivers/accel/amdxdna/npu6_regs.c
index 94a7005685a7..758dc013fe13 100644
--- a/drivers/accel/amdxdna/npu6_regs.c
+++ b/drivers/accel/amdxdna/npu6_regs.c
@@ -67,10 +67,12 @@ static const struct amdxdna_dev_priv npu6_dev_priv = {
 	.protocol_minor = 12,
 	.rt_config	= npu4_default_rt_cfg,
 	.dpm_clk_tbl	= npu4_dpm_clk_table,
+	.fw_feature_tbl = npu4_fw_feature_table,
 	.col_align	= COL_ALIGN_NATURE,
 	.mbox_dev_addr  = NPU6_MBOX_BAR_BASE,
 	.mbox_size      = 0, /* Use BAR size */
 	.sram_dev_addr  = NPU6_SRAM_BAR_BASE,
+	.hwctx_limit    = 16,
 	.sram_offs      = {
 		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU6_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
 		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU6_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
diff --git a/drivers/accel/drm_accel.c b/drivers/accel/drm_accel.c
index aa826033b0ce..ca3357acd127 100644
--- a/drivers/accel/drm_accel.c
+++ b/drivers/accel/drm_accel.c
@@ -20,8 +20,6 @@
 
 DEFINE_XARRAY_ALLOC(accel_minors_xa);
 
-static struct dentry *accel_debugfs_root;
-
 static const struct device_type accel_sysfs_device_minor = {
 	.name = "accel_minor"
 };
@@ -74,17 +72,6 @@ static const struct drm_info_list accel_debugfs_list[] = {
 #define ACCEL_DEBUGFS_ENTRIES ARRAY_SIZE(accel_debugfs_list)
 
 /**
- * accel_debugfs_init() - Initialize debugfs for device
- * @dev: Pointer to the device instance.
- *
- * This function creates a root directory for the device in debugfs.
- */
-void accel_debugfs_init(struct drm_device *dev)
-{
-	drm_debugfs_dev_init(dev, accel_debugfs_root);
-}
-
-/**
  * accel_debugfs_register() - Register debugfs for device
  * @dev: Pointer to the device instance.
  *
@@ -194,7 +181,6 @@ static const struct file_operations accel_stub_fops = {
 void accel_core_exit(void)
 {
 	unregister_chrdev(ACCEL_MAJOR, "accel");
-	debugfs_remove(accel_debugfs_root);
 	accel_sysfs_destroy();
 	WARN_ON(!xa_empty(&accel_minors_xa));
 }
@@ -209,8 +195,6 @@ int __init accel_core_init(void)
 		goto error;
 	}
 
-	accel_debugfs_root = debugfs_create_dir("accel", NULL);
-
 	ret = register_chrdev(ACCEL_MAJOR, "accel", &accel_stub_fops);
 	if (ret < 0)
 		DRM_ERROR("Cannot register ACCEL major: %d\n", ret);
diff --git a/drivers/accel/ethosu/Kconfig b/drivers/accel/ethosu/Kconfig
new file mode 100644
index 000000000000..d25f9b3eb317
--- /dev/null
+++ b/drivers/accel/ethosu/Kconfig
@@ -0,0 +1,11 @@
+# SPDX-License-Identifier: GPL-2.0-only
+
+config DRM_ACCEL_ARM_ETHOSU
+	tristate "Arm Ethos-U65/U85 NPU"
+	depends on HAS_IOMEM
+	depends on DRM_ACCEL
+	select DRM_GEM_DMA_HELPER
+	select DRM_SCHED
+	select GENERIC_ALLOCATOR
+	help
+	  Enables driver for Arm Ethos-U65/U85 NPUs
diff --git a/drivers/accel/ethosu/Makefile b/drivers/accel/ethosu/Makefile
new file mode 100644
index 000000000000..17db5a600416
--- /dev/null
+++ b/drivers/accel/ethosu/Makefile
@@ -0,0 +1,4 @@
+# SPDX-License-Identifier: GPL-2.0-only
+
+obj-$(CONFIG_DRM_ACCEL_ARM_ETHOSU) := ethosu.o
+ethosu-y += ethosu_drv.o ethosu_gem.o ethosu_job.o
diff --git a/drivers/accel/ethosu/ethosu_device.h b/drivers/accel/ethosu/ethosu_device.h
new file mode 100644
index 000000000000..b189fa783d6a
--- /dev/null
+++ b/drivers/accel/ethosu/ethosu_device.h
@@ -0,0 +1,197 @@
+/* SPDX-License-Identifier: GPL-2.0-only or MIT */
+/* Copyright 2025 Arm, Ltd. */
+
+#ifndef __ETHOSU_DEVICE_H__
+#define __ETHOSU_DEVICE_H__
+
+#include <linux/bitfield.h>
+#include <linux/bits.h>
+#include <linux/types.h>
+
+#include <drm/drm_device.h>
+#include <drm/gpu_scheduler.h>
+
+#include <drm/ethosu_accel.h>
+
+struct clk;
+struct gen_pool;
+
+#define NPU_REG_ID		0x0000
+#define NPU_REG_STATUS		0x0004
+#define NPU_REG_CMD		0x0008
+#define NPU_REG_RESET		0x000c
+#define NPU_REG_QBASE		0x0010
+#define NPU_REG_QBASE_HI	0x0014
+#define NPU_REG_QREAD		0x0018
+#define NPU_REG_QCONFIG		0x001c
+#define NPU_REG_QSIZE		0x0020
+#define NPU_REG_PROT		0x0024
+#define NPU_REG_CONFIG		0x0028
+#define NPU_REG_REGIONCFG	0x003c
+#define NPU_REG_AXILIMIT0	0x0040		// U65
+#define NPU_REG_AXILIMIT1	0x0044		// U65
+#define NPU_REG_AXILIMIT2	0x0048		// U65
+#define NPU_REG_AXILIMIT3	0x004c		// U65
+#define NPU_REG_MEM_ATTR0	0x0040		// U85
+#define NPU_REG_MEM_ATTR1	0x0044		// U85
+#define NPU_REG_MEM_ATTR2	0x0048		// U85
+#define NPU_REG_MEM_ATTR3	0x004c		// U85
+#define NPU_REG_AXI_SRAM	0x0050		// U85
+#define NPU_REG_AXI_EXT		0x0054		// U85
+
+#define NPU_REG_BASEP(x)	(0x0080 + (x) * 8)
+#define NPU_REG_BASEP_HI(x)	(0x0084 + (x) * 8)
+#define NPU_BASEP_REGION_MAX	8
+
+#define ID_ARCH_MAJOR_MASK	GENMASK(31, 28)
+#define ID_ARCH_MINOR_MASK	GENMASK(27, 20)
+#define ID_ARCH_PATCH_MASK	GENMASK(19, 16)
+#define ID_VER_MAJOR_MASK	GENMASK(11, 8)
+#define ID_VER_MINOR_MASK	GENMASK(7, 4)
+
+#define CONFIG_MACS_PER_CC_MASK	GENMASK(3, 0)
+#define CONFIG_CMD_STREAM_VER_MASK	GENMASK(7, 4)
+
+#define STATUS_STATE_RUNNING	BIT(0)
+#define STATUS_IRQ_RAISED	BIT(1)
+#define STATUS_BUS_STATUS	BIT(2)
+#define STATUS_RESET_STATUS	BIT(3)
+#define STATUS_CMD_PARSE_ERR	BIT(4)
+#define STATUS_CMD_END_REACHED	BIT(5)
+
+#define CMD_CLEAR_IRQ		BIT(1)
+#define CMD_TRANSITION_TO_RUN	BIT(0)
+
+#define RESET_PENDING_CSL	BIT(1)
+#define RESET_PENDING_CPL	BIT(0)
+
+#define PROT_ACTIVE_CSL		BIT(1)
+
+enum ethosu_cmds {
+	NPU_OP_CONV = 0x2,
+	NPU_OP_DEPTHWISE = 0x3,
+	NPU_OP_POOL = 0x5,
+	NPU_OP_ELEMENTWISE = 0x6,
+	NPU_OP_RESIZE = 0x7,	// U85 only
+	NPU_OP_DMA_START = 0x10,
+	NPU_SET_IFM_PAD_TOP = 0x100,
+	NPU_SET_IFM_PAD_LEFT = 0x101,
+	NPU_SET_IFM_PAD_RIGHT = 0x102,
+	NPU_SET_IFM_PAD_BOTTOM = 0x103,
+	NPU_SET_IFM_DEPTH_M1 = 0x104,
+	NPU_SET_IFM_PRECISION = 0x105,
+	NPU_SET_IFM_BROADCAST = 0x108,
+	NPU_SET_IFM_WIDTH0_M1 = 0x10a,
+	NPU_SET_IFM_HEIGHT0_M1 = 0x10b,
+	NPU_SET_IFM_HEIGHT1_M1 = 0x10c,
+	NPU_SET_IFM_REGION = 0x10f,
+	NPU_SET_OFM_WIDTH_M1 = 0x111,
+	NPU_SET_OFM_HEIGHT_M1 = 0x112,
+	NPU_SET_OFM_DEPTH_M1 = 0x113,
+	NPU_SET_OFM_PRECISION = 0x114,
+	NPU_SET_OFM_WIDTH0_M1 = 0x11a,
+	NPU_SET_OFM_HEIGHT0_M1 = 0x11b,
+	NPU_SET_OFM_HEIGHT1_M1 = 0x11c,
+	NPU_SET_OFM_REGION = 0x11f,
+	NPU_SET_KERNEL_WIDTH_M1 = 0x120,
+	NPU_SET_KERNEL_HEIGHT_M1 = 0x121,
+	NPU_SET_KERNEL_STRIDE = 0x122,
+	NPU_SET_WEIGHT_REGION = 0x128,
+	NPU_SET_SCALE_REGION = 0x129,
+	NPU_SET_DMA0_SRC_REGION = 0x130,
+	NPU_SET_DMA0_DST_REGION = 0x131,
+	NPU_SET_DMA0_SIZE0 = 0x132,
+	NPU_SET_DMA0_SIZE1 = 0x133,
+	NPU_SET_IFM2_BROADCAST = 0x180,
+	NPU_SET_IFM2_PRECISION = 0x185,
+	NPU_SET_IFM2_WIDTH0_M1 = 0x18a,
+	NPU_SET_IFM2_HEIGHT0_M1 = 0x18b,
+	NPU_SET_IFM2_HEIGHT1_M1 = 0x18c,
+	NPU_SET_IFM2_REGION = 0x18f,
+	NPU_SET_IFM_BASE0 = 0x4000,
+	NPU_SET_IFM_BASE1 = 0x4001,
+	NPU_SET_IFM_BASE2 = 0x4002,
+	NPU_SET_IFM_BASE3 = 0x4003,
+	NPU_SET_IFM_STRIDE_X = 0x4004,
+	NPU_SET_IFM_STRIDE_Y = 0x4005,
+	NPU_SET_IFM_STRIDE_C = 0x4006,
+	NPU_SET_OFM_BASE0 = 0x4010,
+	NPU_SET_OFM_BASE1 = 0x4011,
+	NPU_SET_OFM_BASE2 = 0x4012,
+	NPU_SET_OFM_BASE3 = 0x4013,
+	NPU_SET_OFM_STRIDE_X = 0x4014,
+	NPU_SET_OFM_STRIDE_Y = 0x4015,
+	NPU_SET_OFM_STRIDE_C = 0x4016,
+	NPU_SET_WEIGHT_BASE = 0x4020,
+	NPU_SET_WEIGHT_LENGTH = 0x4021,
+	NPU_SET_SCALE_BASE = 0x4022,
+	NPU_SET_SCALE_LENGTH = 0x4023,
+	NPU_SET_DMA0_SRC = 0x4030,
+	NPU_SET_DMA0_DST = 0x4031,
+	NPU_SET_DMA0_LEN = 0x4032,
+	NPU_SET_DMA0_SRC_STRIDE0 = 0x4033,
+	NPU_SET_DMA0_SRC_STRIDE1 = 0x4034,
+	NPU_SET_DMA0_DST_STRIDE0 = 0x4035,
+	NPU_SET_DMA0_DST_STRIDE1 = 0x4036,
+	NPU_SET_IFM2_BASE0 = 0x4080,
+	NPU_SET_IFM2_BASE1 = 0x4081,
+	NPU_SET_IFM2_BASE2 = 0x4082,
+	NPU_SET_IFM2_BASE3 = 0x4083,
+	NPU_SET_IFM2_STRIDE_X = 0x4084,
+	NPU_SET_IFM2_STRIDE_Y = 0x4085,
+	NPU_SET_IFM2_STRIDE_C = 0x4086,
+	NPU_SET_WEIGHT1_BASE = 0x4090,
+	NPU_SET_WEIGHT1_LENGTH = 0x4091,
+	NPU_SET_SCALE1_BASE = 0x4092,
+	NPU_SET_WEIGHT2_BASE = 0x4092,
+	NPU_SET_SCALE1_LENGTH = 0x4093,
+	NPU_SET_WEIGHT2_LENGTH = 0x4093,
+	NPU_SET_WEIGHT3_BASE = 0x4094,
+	NPU_SET_WEIGHT3_LENGTH = 0x4095,
+};
+
+#define ETHOSU_SRAM_REGION	2	/* Matching Vela compiler */
+
+/**
+ * struct ethosu_device - Ethosu device
+ */
+struct ethosu_device {
+	/** @base: Base drm_device. */
+	struct drm_device base;
+
+	/** @iomem: CPU mapping of the registers. */
+	void __iomem *regs;
+
+	void __iomem *sram;
+	struct gen_pool *srampool;
+	dma_addr_t sramphys;
+
+	struct clk_bulk_data *clks;
+	int num_clks;
+	int irq;
+
+	struct drm_ethosu_npu_info npu_info;
+
+	struct ethosu_job *in_flight_job;
+	/* For in_flight_job and ethosu_job_hw_submit() */
+	struct mutex job_lock;
+
+	/* For dma_fence */
+	spinlock_t fence_lock;
+
+	struct drm_gpu_scheduler sched;
+	/* For ethosu_job_do_push() */
+	struct mutex sched_lock;
+	u64 fence_context;
+	u64 emit_seqno;
+};
+
+#define to_ethosu_device(drm_dev) \
+	((struct ethosu_device *)container_of(drm_dev, struct ethosu_device, base))
+
+static inline bool ethosu_is_u65(const struct ethosu_device *ethosudev)
+{
+	return FIELD_GET(ID_ARCH_MAJOR_MASK, ethosudev->npu_info.id) == 1;
+}
+
+#endif
diff --git a/drivers/accel/ethosu/ethosu_drv.c b/drivers/accel/ethosu/ethosu_drv.c
new file mode 100644
index 000000000000..e05a69bf5574
--- /dev/null
+++ b/drivers/accel/ethosu/ethosu_drv.c
@@ -0,0 +1,403 @@
+// SPDX-License-Identifier: GPL-2.0-only or MIT
+// Copyright (C) 2025 Arm, Ltd.
+
+#include <linux/bitfield.h>
+#include <linux/clk.h>
+#include <linux/genalloc.h>
+#include <linux/io.h>
+#include <linux/iopoll.h>
+#include <linux/module.h>
+#include <linux/mod_devicetable.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+
+#include <drm/drm_drv.h>
+#include <drm/drm_ioctl.h>
+#include <drm/drm_utils.h>
+#include <drm/drm_gem.h>
+#include <drm/drm_accel.h>
+#include <drm/ethosu_accel.h>
+
+#include "ethosu_drv.h"
+#include "ethosu_device.h"
+#include "ethosu_gem.h"
+#include "ethosu_job.h"
+
+static int ethosu_ioctl_dev_query(struct drm_device *ddev, void *data,
+				  struct drm_file *file)
+{
+	struct ethosu_device *ethosudev = to_ethosu_device(ddev);
+	struct drm_ethosu_dev_query *args = data;
+
+	if (!args->pointer) {
+		switch (args->type) {
+		case DRM_ETHOSU_DEV_QUERY_NPU_INFO:
+			args->size = sizeof(ethosudev->npu_info);
+			return 0;
+		default:
+			return -EINVAL;
+		}
+	}
+
+	switch (args->type) {
+	case DRM_ETHOSU_DEV_QUERY_NPU_INFO:
+		if (args->size < offsetofend(struct drm_ethosu_npu_info, sram_size))
+			return -EINVAL;
+		return copy_struct_to_user(u64_to_user_ptr(args->pointer),
+					   args->size,
+					   &ethosudev->npu_info,
+					   sizeof(ethosudev->npu_info), NULL);
+	default:
+		return -EINVAL;
+	}
+}
+
+#define ETHOSU_BO_FLAGS		DRM_ETHOSU_BO_NO_MMAP
+
+static int ethosu_ioctl_bo_create(struct drm_device *ddev, void *data,
+				  struct drm_file *file)
+{
+	struct drm_ethosu_bo_create *args = data;
+	int cookie, ret;
+
+	if (!drm_dev_enter(ddev, &cookie))
+		return -ENODEV;
+
+	if (!args->size || (args->flags & ~ETHOSU_BO_FLAGS)) {
+		ret = -EINVAL;
+		goto out_dev_exit;
+	}
+
+	ret = ethosu_gem_create_with_handle(file, ddev, &args->size,
+					    args->flags, &args->handle);
+
+out_dev_exit:
+	drm_dev_exit(cookie);
+	return ret;
+}
+
+static int ethosu_ioctl_bo_wait(struct drm_device *ddev, void *data,
+				struct drm_file *file)
+{
+	struct drm_ethosu_bo_wait *args = data;
+	int cookie, ret;
+	unsigned long timeout = drm_timeout_abs_to_jiffies(args->timeout_ns);
+
+	if (args->pad)
+		return -EINVAL;
+
+	if (!drm_dev_enter(ddev, &cookie))
+		return -ENODEV;
+
+	ret = drm_gem_dma_resv_wait(file, args->handle, true, timeout);
+
+	drm_dev_exit(cookie);
+	return ret;
+}
+
+static int ethosu_ioctl_bo_mmap_offset(struct drm_device *ddev, void *data,
+				       struct drm_file *file)
+{
+	struct drm_ethosu_bo_mmap_offset *args = data;
+	struct drm_gem_object *obj;
+
+	if (args->pad)
+		return -EINVAL;
+
+	obj = drm_gem_object_lookup(file, args->handle);
+	if (!obj)
+		return -ENOENT;
+
+	args->offset = drm_vma_node_offset_addr(&obj->vma_node);
+	drm_gem_object_put(obj);
+	return 0;
+}
+
+static int ethosu_ioctl_cmdstream_bo_create(struct drm_device *ddev, void *data,
+					    struct drm_file *file)
+{
+	struct drm_ethosu_cmdstream_bo_create *args = data;
+	int cookie, ret;
+
+	if (!drm_dev_enter(ddev, &cookie))
+		return -ENODEV;
+
+	if (!args->size || !args->data || args->pad || args->flags) {
+		ret = -EINVAL;
+		goto out_dev_exit;
+	}
+
+	args->flags |= DRM_ETHOSU_BO_NO_MMAP;
+
+	ret = ethosu_gem_cmdstream_create(file, ddev, args->size, args->data,
+					  args->flags, &args->handle);
+
+out_dev_exit:
+	drm_dev_exit(cookie);
+	return ret;
+}
+
+static int ethosu_open(struct drm_device *ddev, struct drm_file *file)
+{
+	int ret = 0;
+
+	if (!try_module_get(THIS_MODULE))
+		return -EINVAL;
+
+	struct ethosu_file_priv __free(kfree) *priv = kzalloc(sizeof(*priv), GFP_KERNEL);
+	if (!priv) {
+		ret = -ENOMEM;
+		goto err_put_mod;
+	}
+	priv->edev = to_ethosu_device(ddev);
+
+	ret = ethosu_job_open(priv);
+	if (ret)
+		goto err_put_mod;
+
+	file->driver_priv = no_free_ptr(priv);
+	return 0;
+
+err_put_mod:
+	module_put(THIS_MODULE);
+	return ret;
+}
+
+static void ethosu_postclose(struct drm_device *ddev, struct drm_file *file)
+{
+	ethosu_job_close(file->driver_priv);
+	kfree(file->driver_priv);
+	module_put(THIS_MODULE);
+}
+
+static const struct drm_ioctl_desc ethosu_drm_driver_ioctls[] = {
+#define ETHOSU_IOCTL(n, func, flags) \
+	DRM_IOCTL_DEF_DRV(ETHOSU_##n, ethosu_ioctl_##func, flags)
+
+	ETHOSU_IOCTL(DEV_QUERY, dev_query, 0),
+	ETHOSU_IOCTL(BO_CREATE, bo_create, 0),
+	ETHOSU_IOCTL(BO_WAIT, bo_wait, 0),
+	ETHOSU_IOCTL(BO_MMAP_OFFSET, bo_mmap_offset, 0),
+	ETHOSU_IOCTL(CMDSTREAM_BO_CREATE, cmdstream_bo_create, 0),
+	ETHOSU_IOCTL(SUBMIT, submit, 0),
+};
+
+DEFINE_DRM_ACCEL_FOPS(ethosu_drm_driver_fops);
+
+/*
+ * Ethosu driver version:
+ * - 1.0 - initial interface
+ */
+static const struct drm_driver ethosu_drm_driver = {
+	.driver_features = DRIVER_COMPUTE_ACCEL | DRIVER_GEM,
+	.open = ethosu_open,
+	.postclose = ethosu_postclose,
+	.ioctls = ethosu_drm_driver_ioctls,
+	.num_ioctls = ARRAY_SIZE(ethosu_drm_driver_ioctls),
+	.fops = &ethosu_drm_driver_fops,
+	.name = "ethosu",
+	.desc = "Arm Ethos-U Accel driver",
+	.major = 1,
+	.minor = 0,
+
+	.gem_create_object = ethosu_gem_create_object,
+};
+
+#define U65_DRAM_AXI_LIMIT_CFG	0x1f3f0002
+#define U65_SRAM_AXI_LIMIT_CFG	0x1f3f00b0
+#define U85_AXI_EXT_CFG		0x00021f3f
+#define U85_AXI_SRAM_CFG	0x00021f3f
+#define U85_MEM_ATTR0_CFG	0x00000000
+#define U85_MEM_ATTR2_CFG	0x000000b7
+
+static int ethosu_reset(struct ethosu_device *ethosudev)
+{
+	int ret;
+	u32 reg;
+
+	writel_relaxed(RESET_PENDING_CSL, ethosudev->regs + NPU_REG_RESET);
+	ret = readl_poll_timeout(ethosudev->regs + NPU_REG_STATUS, reg,
+				 !FIELD_GET(STATUS_RESET_STATUS, reg),
+				 USEC_PER_MSEC, USEC_PER_SEC);
+	if (ret)
+		return ret;
+
+	if (!FIELD_GET(PROT_ACTIVE_CSL, readl_relaxed(ethosudev->regs + NPU_REG_PROT))) {
+		dev_warn(ethosudev->base.dev, "Could not reset to non-secure mode (PROT = %x)\n",
+			 readl_relaxed(ethosudev->regs + NPU_REG_PROT));
+	}
+
+	/*
+	 * Assign region 2 (SRAM) to AXI M0 (AXILIMIT0),
+	 * everything else to AXI M1 (AXILIMIT2)
+	 */
+	writel_relaxed(0x0000aa8a, ethosudev->regs + NPU_REG_REGIONCFG);
+	if (ethosu_is_u65(ethosudev)) {
+		writel_relaxed(U65_SRAM_AXI_LIMIT_CFG, ethosudev->regs + NPU_REG_AXILIMIT0);
+		writel_relaxed(U65_DRAM_AXI_LIMIT_CFG, ethosudev->regs + NPU_REG_AXILIMIT2);
+	} else {
+		writel_relaxed(U85_AXI_SRAM_CFG, ethosudev->regs + NPU_REG_AXI_SRAM);
+		writel_relaxed(U85_AXI_EXT_CFG, ethosudev->regs + NPU_REG_AXI_EXT);
+		writel_relaxed(U85_MEM_ATTR0_CFG, ethosudev->regs + NPU_REG_MEM_ATTR0);	// SRAM
+		writel_relaxed(U85_MEM_ATTR2_CFG, ethosudev->regs + NPU_REG_MEM_ATTR2);	// DRAM
+	}
+
+	if (ethosudev->sram)
+		memset_io(ethosudev->sram, 0, ethosudev->npu_info.sram_size);
+
+	return 0;
+}
+
+static int ethosu_device_resume(struct device *dev)
+{
+	struct ethosu_device *ethosudev = dev_get_drvdata(dev);
+	int ret;
+
+	ret = clk_bulk_prepare_enable(ethosudev->num_clks, ethosudev->clks);
+	if (ret)
+		return ret;
+
+	ret = ethosu_reset(ethosudev);
+	if (!ret)
+		return 0;
+
+	clk_bulk_disable_unprepare(ethosudev->num_clks, ethosudev->clks);
+	return ret;
+}
+
+static int ethosu_device_suspend(struct device *dev)
+{
+	struct ethosu_device *ethosudev = dev_get_drvdata(dev);
+
+	clk_bulk_disable_unprepare(ethosudev->num_clks, ethosudev->clks);
+	return 0;
+}
+
+static int ethosu_sram_init(struct ethosu_device *ethosudev)
+{
+	ethosudev->npu_info.sram_size = 0;
+
+	ethosudev->srampool = of_gen_pool_get(ethosudev->base.dev->of_node, "sram", 0);
+	if (!ethosudev->srampool)
+		return 0;
+
+	ethosudev->npu_info.sram_size = gen_pool_size(ethosudev->srampool);
+
+	ethosudev->sram = (void __iomem *)gen_pool_dma_alloc(ethosudev->srampool,
+							     ethosudev->npu_info.sram_size,
+							     &ethosudev->sramphys);
+	if (!ethosudev->sram) {
+		dev_err(ethosudev->base.dev, "failed to allocate from SRAM pool\n");
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+static int ethosu_init(struct ethosu_device *ethosudev)
+{
+	int ret;
+	u32 id, config;
+
+	ret = ethosu_device_resume(ethosudev->base.dev);
+	if (ret)
+		return ret;
+
+	pm_runtime_set_autosuspend_delay(ethosudev->base.dev, 50);
+	pm_runtime_use_autosuspend(ethosudev->base.dev);
+	ret = devm_pm_runtime_set_active_enabled(ethosudev->base.dev);
+	if (ret)
+		return ret;
+	pm_runtime_get_noresume(ethosudev->base.dev);
+
+	ethosudev->npu_info.id = id = readl_relaxed(ethosudev->regs + NPU_REG_ID);
+	ethosudev->npu_info.config = config = readl_relaxed(ethosudev->regs + NPU_REG_CONFIG);
+
+	ethosu_sram_init(ethosudev);
+
+	dev_info(ethosudev->base.dev,
+		 "Ethos-U NPU, arch v%ld.%ld.%ld, rev r%ldp%ld, cmd stream ver%ld, %d MACs, %dKB SRAM\n",
+		 FIELD_GET(ID_ARCH_MAJOR_MASK, id),
+		 FIELD_GET(ID_ARCH_MINOR_MASK, id),
+		 FIELD_GET(ID_ARCH_PATCH_MASK, id),
+		 FIELD_GET(ID_VER_MAJOR_MASK, id),
+		 FIELD_GET(ID_VER_MINOR_MASK, id),
+		 FIELD_GET(CONFIG_CMD_STREAM_VER_MASK, config),
+		 1 << FIELD_GET(CONFIG_MACS_PER_CC_MASK, config),
+		 ethosudev->npu_info.sram_size / 1024);
+
+	return 0;
+}
+
+static int ethosu_probe(struct platform_device *pdev)
+{
+	int ret;
+	struct ethosu_device *ethosudev;
+
+	ethosudev = devm_drm_dev_alloc(&pdev->dev, &ethosu_drm_driver,
+				       struct ethosu_device, base);
+	if (IS_ERR(ethosudev))
+		return -ENOMEM;
+	platform_set_drvdata(pdev, ethosudev);
+
+	dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(40));
+
+	ethosudev->regs = devm_platform_ioremap_resource(pdev, 0);
+
+	ethosudev->num_clks = devm_clk_bulk_get_all(&pdev->dev, &ethosudev->clks);
+	if (ethosudev->num_clks < 0)
+		return ethosudev->num_clks;
+
+	ret = ethosu_job_init(ethosudev);
+	if (ret)
+		return ret;
+
+	ret = ethosu_init(ethosudev);
+	if (ret)
+		return ret;
+
+	ret = drm_dev_register(&ethosudev->base, 0);
+	if (ret)
+		pm_runtime_dont_use_autosuspend(ethosudev->base.dev);
+
+	pm_runtime_put_autosuspend(ethosudev->base.dev);
+	return ret;
+}
+
+static void ethosu_remove(struct platform_device *pdev)
+{
+	struct ethosu_device *ethosudev = dev_get_drvdata(&pdev->dev);
+
+	drm_dev_unregister(&ethosudev->base);
+	ethosu_job_fini(ethosudev);
+	if (ethosudev->sram)
+		gen_pool_free(ethosudev->srampool, (unsigned long)ethosudev->sram,
+			      ethosudev->npu_info.sram_size);
+}
+
+static const struct of_device_id dt_match[] = {
+	{ .compatible = "arm,ethos-u65" },
+	{ .compatible = "arm,ethos-u85" },
+	{}
+};
+MODULE_DEVICE_TABLE(of, dt_match);
+
+static DEFINE_RUNTIME_DEV_PM_OPS(ethosu_pm_ops,
+				 ethosu_device_suspend,
+				 ethosu_device_resume,
+				 NULL);
+
+static struct platform_driver ethosu_driver = {
+	.probe = ethosu_probe,
+	.remove = ethosu_remove,
+	.driver = {
+		.name = "ethosu",
+		.pm = pm_ptr(&ethosu_pm_ops),
+		.of_match_table = dt_match,
+	},
+};
+module_platform_driver(ethosu_driver);
+
+MODULE_AUTHOR("Rob Herring <robh@kernel.org>");
+MODULE_DESCRIPTION("Arm Ethos-U Accel Driver");
+MODULE_LICENSE("Dual MIT/GPL");
diff --git a/drivers/accel/ethosu/ethosu_drv.h b/drivers/accel/ethosu/ethosu_drv.h
new file mode 100644
index 000000000000..9e21dfe94184
--- /dev/null
+++ b/drivers/accel/ethosu/ethosu_drv.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0-only OR MIT */
+/* Copyright 2025 Arm, Ltd. */
+#ifndef __ETHOSU_DRV_H__
+#define __ETHOSU_DRV_H__
+
+#include <drm/gpu_scheduler.h>
+
+struct ethosu_device;
+
+struct ethosu_file_priv {
+	struct ethosu_device *edev;
+	struct drm_sched_entity sched_entity;
+};
+
+#endif
diff --git a/drivers/accel/ethosu/ethosu_gem.c b/drivers/accel/ethosu/ethosu_gem.c
new file mode 100644
index 000000000000..473b5f5d7514
--- /dev/null
+++ b/drivers/accel/ethosu/ethosu_gem.c
@@ -0,0 +1,704 @@
+// SPDX-License-Identifier: GPL-2.0-only or MIT
+/* Copyright 2025 Arm, Ltd. */
+
+#include <linux/err.h>
+#include <linux/slab.h>
+
+#include <drm/ethosu_accel.h>
+
+#include "ethosu_device.h"
+#include "ethosu_gem.h"
+
+static void ethosu_gem_free_object(struct drm_gem_object *obj)
+{
+	struct ethosu_gem_object *bo = to_ethosu_bo(obj);
+
+	kfree(bo->info);
+	drm_gem_free_mmap_offset(&bo->base.base);
+	drm_gem_dma_free(&bo->base);
+}
+
+static int ethosu_gem_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma)
+{
+	struct ethosu_gem_object *bo = to_ethosu_bo(obj);
+
+	/* Don't allow mmap on objects that have the NO_MMAP flag set. */
+	if (bo->flags & DRM_ETHOSU_BO_NO_MMAP)
+		return -EINVAL;
+
+	return drm_gem_dma_object_mmap(obj, vma);
+}
+
+static const struct drm_gem_object_funcs ethosu_gem_funcs = {
+	.free = ethosu_gem_free_object,
+	.print_info = drm_gem_dma_object_print_info,
+	.get_sg_table = drm_gem_dma_object_get_sg_table,
+	.vmap = drm_gem_dma_object_vmap,
+	.mmap = ethosu_gem_mmap,
+	.vm_ops = &drm_gem_dma_vm_ops,
+};
+
+/**
+ * ethosu_gem_create_object - Implementation of driver->gem_create_object.
+ * @ddev: DRM device
+ * @size: Size in bytes of the memory the object will reference
+ *
+ * This lets the GEM helpers allocate object structs for us, and keep
+ * our BO stats correct.
+ */
+struct drm_gem_object *ethosu_gem_create_object(struct drm_device *ddev, size_t size)
+{
+	struct ethosu_gem_object *obj;
+
+	obj = kzalloc(sizeof(*obj), GFP_KERNEL);
+	if (!obj)
+		return ERR_PTR(-ENOMEM);
+
+	obj->base.base.funcs = &ethosu_gem_funcs;
+	return &obj->base.base;
+}
+
+/**
+ * ethosu_gem_create_with_handle() - Create a GEM object and attach it to a handle.
+ * @file: DRM file.
+ * @ddev: DRM device.
+ * @size: Size of the GEM object to allocate.
+ * @flags: Combination of drm_ethosu_bo_flags flags.
+ * @handle: Pointer holding the handle pointing to the new GEM object.
+ *
+ * Return: Zero on success
+ */
+int ethosu_gem_create_with_handle(struct drm_file *file,
+				  struct drm_device *ddev,
+				  u64 *size, u32 flags, u32 *handle)
+{
+	struct drm_gem_dma_object *mem;
+	struct ethosu_gem_object *bo;
+	int ret;
+
+	mem = drm_gem_dma_create(ddev, *size);
+	if (IS_ERR(mem))
+		return PTR_ERR(mem);
+
+	bo = to_ethosu_bo(&mem->base);
+	bo->flags = flags;
+
+	/*
+	 * Allocate an id of idr table where the obj is registered
+	 * and handle has the id what user can see.
+	 */
+	ret = drm_gem_handle_create(file, &mem->base, handle);
+	if (!ret)
+		*size = bo->base.base.size;
+
+	/* drop reference from allocate - handle holds it now. */
+	drm_gem_object_put(&mem->base);
+
+	return ret;
+}
+
+struct dma {
+	s8 region;
+	u64 len;
+	u64 offset;
+	s64 stride[2];
+};
+
+struct dma_state {
+	u16 size0;
+	u16 size1;
+	s8 mode;
+	struct dma src;
+	struct dma dst;
+};
+
+struct buffer {
+	u64 base;
+	u32 length;
+	s8 region;
+};
+
+struct feat_matrix {
+	u64 base[4];
+	s64 stride_x;
+	s64 stride_y;
+	s64 stride_c;
+	s8 region;
+	u8 broadcast;
+	u16 stride_kernel;
+	u16 precision;
+	u16 depth;
+	u16 width;
+	u16 width0;
+	u16 height[3];
+	u8 pad_top;
+	u8 pad_left;
+	u8 pad_bottom;
+	u8 pad_right;
+};
+
+struct cmd_state {
+	struct dma_state dma;
+	struct buffer scale[2];
+	struct buffer weight[4];
+	struct feat_matrix ofm;
+	struct feat_matrix ifm;
+	struct feat_matrix ifm2;
+};
+
+static void cmd_state_init(struct cmd_state *st)
+{
+	/* Initialize to all 1s to detect missing setup */
+	memset(st, 0xff, sizeof(*st));
+}
+
+static u64 cmd_to_addr(u32 *cmd)
+{
+	return ((u64)((cmd[0] & 0xff0000) << 16)) | cmd[1];
+}
+
+static u64 dma_length(struct ethosu_validated_cmdstream_info *info,
+		      struct dma_state *dma_st, struct dma *dma)
+{
+	s8 mode = dma_st->mode;
+	u64 len = dma->len;
+
+	if (mode >= 1) {
+		len += dma->stride[0];
+		len *= dma_st->size0;
+	}
+	if (mode == 2) {
+		len += dma->stride[1];
+		len *= dma_st->size1;
+	}
+	if (dma->region >= 0)
+		info->region_size[dma->region] = max(info->region_size[dma->region],
+						     len + dma->offset);
+
+	return len;
+}
+
+static u64 feat_matrix_length(struct ethosu_validated_cmdstream_info *info,
+			      struct feat_matrix *fm,
+			      u32 x, u32 y, u32 c)
+{
+	u32 element_size, storage = fm->precision >> 14;
+	int tile = 0;
+	u64 addr;
+
+	if (fm->region < 0)
+		return U64_MAX;
+
+	switch (storage) {
+	case 0:
+		if (x >= fm->width0 + 1) {
+			x -= fm->width0 + 1;
+			tile += 1;
+		}
+		if (y >= fm->height[tile] + 1) {
+			y -= fm->height[tile] + 1;
+			tile += 2;
+		}
+		break;
+	case 1:
+		if (y >= fm->height[1] + 1) {
+			y -= fm->height[1] + 1;
+			tile = 2;
+		} else if (y >= fm->height[0] + 1) {
+			y -= fm->height[0] + 1;
+			tile = 1;
+		}
+		break;
+	}
+	if (fm->base[tile] == U64_MAX)
+		return U64_MAX;
+
+	addr = fm->base[tile] + y * fm->stride_y;
+
+	switch ((fm->precision >> 6) & 0x3) { // format
+	case 0: //nhwc:
+		addr += x * fm->stride_x + c;
+		break;
+	case 1: //nhcwb16:
+		element_size = BIT((fm->precision >> 1) & 0x3);
+
+		addr += (c / 16) * fm->stride_c + (16 * x + (c & 0xf)) * element_size;
+		break;
+	}
+
+	info->region_size[fm->region] = max(info->region_size[fm->region], addr + 1);
+
+	return addr;
+}
+
+static int calc_sizes(struct drm_device *ddev,
+		      struct ethosu_validated_cmdstream_info *info,
+		      u16 op, struct cmd_state *st,
+		      bool ifm, bool ifm2, bool weight, bool scale)
+{
+	u64 len;
+
+	if (ifm) {
+		if (st->ifm.stride_kernel == U16_MAX)
+			return -EINVAL;
+		u32 stride_y = ((st->ifm.stride_kernel >> 8) & 0x2) +
+			((st->ifm.stride_kernel >> 1) & 0x1) + 1;
+		u32 stride_x = ((st->ifm.stride_kernel >> 5) & 0x2) +
+			(st->ifm.stride_kernel & 0x1) + 1;
+		u32 ifm_height = st->ofm.height[2] * stride_y +
+			st->ifm.height[2] - (st->ifm.pad_top + st->ifm.pad_bottom);
+		u32 ifm_width  = st->ofm.width * stride_x +
+			st->ifm.width - (st->ifm.pad_left + st->ifm.pad_right);
+
+		len = feat_matrix_length(info, &st->ifm, ifm_width,
+					 ifm_height, st->ifm.depth);
+		dev_dbg(ddev->dev, "op %d: IFM:%d:0x%llx-0x%llx\n",
+			op, st->ifm.region, st->ifm.base[0], len);
+		if (len == U64_MAX)
+			return -EINVAL;
+	}
+
+	if (ifm2) {
+		len = feat_matrix_length(info, &st->ifm2, st->ifm.depth,
+					 0, st->ofm.depth);
+		dev_dbg(ddev->dev, "op %d: IFM2:%d:0x%llx-0x%llx\n",
+			op, st->ifm2.region, st->ifm2.base[0], len);
+		if (len == U64_MAX)
+			return -EINVAL;
+	}
+
+	if (weight) {
+		dev_dbg(ddev->dev, "op %d: W:%d:0x%llx-0x%llx\n",
+			op, st->weight[0].region, st->weight[0].base,
+			st->weight[0].base + st->weight[0].length - 1);
+		if (st->weight[0].region < 0 || st->weight[0].base == U64_MAX ||
+		    st->weight[0].length == U32_MAX)
+			return -EINVAL;
+		info->region_size[st->weight[0].region] =
+			max(info->region_size[st->weight[0].region],
+			    st->weight[0].base + st->weight[0].length);
+	}
+
+	if (scale) {
+		dev_dbg(ddev->dev, "op %d: S:%d:0x%llx-0x%llx\n",
+			op, st->scale[0].region, st->scale[0].base,
+			st->scale[0].base + st->scale[0].length - 1);
+		if (st->scale[0].region < 0 || st->scale[0].base == U64_MAX ||
+		    st->scale[0].length == U32_MAX)
+			return -EINVAL;
+		info->region_size[st->scale[0].region] =
+			max(info->region_size[st->scale[0].region],
+			    st->scale[0].base + st->scale[0].length);
+	}
+
+	len = feat_matrix_length(info, &st->ofm, st->ofm.width,
+				 st->ofm.height[2], st->ofm.depth);
+	dev_dbg(ddev->dev, "op %d: OFM:%d:0x%llx-0x%llx\n",
+		op, st->ofm.region, st->ofm.base[0], len);
+	if (len == U64_MAX)
+		return -EINVAL;
+	info->output_region[st->ofm.region] = true;
+
+	return 0;
+}
+
+static int calc_sizes_elemwise(struct drm_device *ddev,
+			       struct ethosu_validated_cmdstream_info *info,
+			       u16 op, struct cmd_state *st,
+			       bool ifm, bool ifm2)
+{
+	u32 height, width, depth;
+	u64 len;
+
+	if (ifm) {
+		height = st->ifm.broadcast & 0x1 ? 0 : st->ofm.height[2];
+		width = st->ifm.broadcast & 0x2 ? 0 : st->ofm.width;
+		depth = st->ifm.broadcast & 0x4 ? 0 : st->ofm.depth;
+
+		len = feat_matrix_length(info, &st->ifm, width,
+					 height, depth);
+		dev_dbg(ddev->dev, "op %d: IFM:%d:0x%llx-0x%llx\n",
+			op, st->ifm.region, st->ifm.base[0], len);
+		if (len == U64_MAX)
+			return -EINVAL;
+	}
+
+	if (ifm2) {
+		height = st->ifm2.broadcast & 0x1 ? 0 : st->ofm.height[2];
+		width = st->ifm2.broadcast & 0x2 ? 0 : st->ofm.width;
+		depth = st->ifm2.broadcast & 0x4 ? 0 : st->ofm.depth;
+
+		len = feat_matrix_length(info, &st->ifm2, width,
+					 height, depth);
+		dev_dbg(ddev->dev, "op %d: IFM2:%d:0x%llx-0x%llx\n",
+			op, st->ifm2.region, st->ifm2.base[0], len);
+		if (len == U64_MAX)
+			return -EINVAL;
+	}
+
+	len = feat_matrix_length(info, &st->ofm, st->ofm.width,
+				 st->ofm.height[2], st->ofm.depth);
+	dev_dbg(ddev->dev, "op %d: OFM:%d:0x%llx-0x%llx\n",
+		op, st->ofm.region, st->ofm.base[0], len);
+	if (len == U64_MAX)
+		return -EINVAL;
+	info->output_region[st->ofm.region] = true;
+
+	return 0;
+}
+
+static int ethosu_gem_cmdstream_copy_and_validate(struct drm_device *ddev,
+						  u32 __user *ucmds,
+						  struct ethosu_gem_object *bo,
+						  u32 size)
+{
+	struct ethosu_validated_cmdstream_info __free(kfree) *info = kzalloc(sizeof(*info), GFP_KERNEL);
+	struct ethosu_device *edev = to_ethosu_device(ddev);
+	u32 *bocmds = bo->base.vaddr;
+	struct cmd_state st;
+	int i, ret;
+
+	if (!info)
+		return -ENOMEM;
+	info->cmd_size = size;
+
+	cmd_state_init(&st);
+
+	for (i = 0; i < size / 4; i++) {
+		bool use_ifm, use_ifm2, use_scale;
+		u64 dstlen, srclen;
+		u16 cmd, param;
+		u32 cmds[2];
+		u64 addr;
+
+		if (get_user(cmds[0], ucmds++))
+			return -EFAULT;
+
+		bocmds[i] = cmds[0];
+
+		cmd = cmds[0];
+		param = cmds[0] >> 16;
+
+		if (cmd & 0x4000) {
+			if (get_user(cmds[1], ucmds++))
+				return -EFAULT;
+
+			i++;
+			bocmds[i] = cmds[1];
+			addr = cmd_to_addr(cmds);
+		}
+
+		switch (cmd) {
+		case NPU_OP_DMA_START:
+			srclen = dma_length(info, &st.dma, &st.dma.src);
+			dstlen = dma_length(info, &st.dma, &st.dma.dst);
+
+			if (st.dma.dst.region >= 0)
+				info->output_region[st.dma.dst.region] = true;
+			dev_dbg(ddev->dev, "cmd: DMA SRC:%d:0x%llx+0x%llx DST:%d:0x%llx+0x%llx\n",
+				st.dma.src.region, st.dma.src.offset, srclen,
+				st.dma.dst.region, st.dma.dst.offset, dstlen);
+			break;
+		case NPU_OP_CONV:
+		case NPU_OP_DEPTHWISE:
+			use_ifm2 = param & 0x1;  // weights_ifm2
+			use_scale = !(st.ofm.precision & 0x100);
+			ret = calc_sizes(ddev, info, cmd, &st, true, use_ifm2,
+					 !use_ifm2, use_scale);
+			if (ret)
+				return ret;
+			break;
+		case NPU_OP_POOL:
+			use_ifm = param != 0x4;  // pooling mode
+			use_scale = !(st.ofm.precision & 0x100);
+			ret = calc_sizes(ddev, info, cmd, &st, use_ifm, false,
+					 false, use_scale);
+			if (ret)
+				return ret;
+			break;
+		case NPU_OP_ELEMENTWISE:
+			use_ifm2 = !((st.ifm2.broadcast == 8) || (param == 5) ||
+				(param == 6) || (param == 7) || (param == 0x24));
+			use_ifm = st.ifm.broadcast != 8;
+			ret = calc_sizes_elemwise(ddev, info, cmd, &st, use_ifm, use_ifm2);
+			if (ret)
+				return ret;
+			break;
+		case NPU_OP_RESIZE: // U85 only
+			WARN_ON(1); // TODO
+			break;
+		case NPU_SET_KERNEL_WIDTH_M1:
+			st.ifm.width = param;
+			break;
+		case NPU_SET_KERNEL_HEIGHT_M1:
+			st.ifm.height[2] = param;
+			break;
+		case NPU_SET_KERNEL_STRIDE:
+			st.ifm.stride_kernel = param;
+			break;
+		case NPU_SET_IFM_PAD_TOP:
+			st.ifm.pad_top = param & 0x7f;
+			break;
+		case NPU_SET_IFM_PAD_LEFT:
+			st.ifm.pad_left = param & 0x7f;
+			break;
+		case NPU_SET_IFM_PAD_RIGHT:
+			st.ifm.pad_right = param & 0xff;
+			break;
+		case NPU_SET_IFM_PAD_BOTTOM:
+			st.ifm.pad_bottom = param & 0xff;
+			break;
+		case NPU_SET_IFM_DEPTH_M1:
+			st.ifm.depth = param;
+			break;
+		case NPU_SET_IFM_PRECISION:
+			st.ifm.precision = param;
+			break;
+		case NPU_SET_IFM_BROADCAST:
+			st.ifm.broadcast = param;
+			break;
+		case NPU_SET_IFM_REGION:
+			st.ifm.region = param & 0x7f;
+			break;
+		case NPU_SET_IFM_WIDTH0_M1:
+			st.ifm.width0 = param;
+			break;
+		case NPU_SET_IFM_HEIGHT0_M1:
+			st.ifm.height[0] = param;
+			break;
+		case NPU_SET_IFM_HEIGHT1_M1:
+			st.ifm.height[1] = param;
+			break;
+		case NPU_SET_IFM_BASE0:
+		case NPU_SET_IFM_BASE1:
+		case NPU_SET_IFM_BASE2:
+		case NPU_SET_IFM_BASE3:
+			st.ifm.base[cmd & 0x3] = addr;
+			break;
+		case NPU_SET_IFM_STRIDE_X:
+			st.ifm.stride_x = addr;
+			break;
+		case NPU_SET_IFM_STRIDE_Y:
+			st.ifm.stride_y = addr;
+			break;
+		case NPU_SET_IFM_STRIDE_C:
+			st.ifm.stride_c = addr;
+			break;
+
+		case NPU_SET_OFM_WIDTH_M1:
+			st.ofm.width = param;
+			break;
+		case NPU_SET_OFM_HEIGHT_M1:
+			st.ofm.height[2] = param;
+			break;
+		case NPU_SET_OFM_DEPTH_M1:
+			st.ofm.depth = param;
+			break;
+		case NPU_SET_OFM_PRECISION:
+			st.ofm.precision = param;
+			break;
+		case NPU_SET_OFM_REGION:
+			st.ofm.region = param & 0x7;
+			break;
+		case NPU_SET_OFM_WIDTH0_M1:
+			st.ofm.width0 = param;
+			break;
+		case NPU_SET_OFM_HEIGHT0_M1:
+			st.ofm.height[0] = param;
+			break;
+		case NPU_SET_OFM_HEIGHT1_M1:
+			st.ofm.height[1] = param;
+			break;
+		case NPU_SET_OFM_BASE0:
+		case NPU_SET_OFM_BASE1:
+		case NPU_SET_OFM_BASE2:
+		case NPU_SET_OFM_BASE3:
+			st.ofm.base[cmd & 0x3] = addr;
+			break;
+		case NPU_SET_OFM_STRIDE_X:
+			st.ofm.stride_x = addr;
+			break;
+		case NPU_SET_OFM_STRIDE_Y:
+			st.ofm.stride_y = addr;
+			break;
+		case NPU_SET_OFM_STRIDE_C:
+			st.ofm.stride_c = addr;
+			break;
+
+		case NPU_SET_IFM2_BROADCAST:
+			st.ifm2.broadcast = param;
+			break;
+		case NPU_SET_IFM2_PRECISION:
+			st.ifm2.precision = param;
+			break;
+		case NPU_SET_IFM2_REGION:
+			st.ifm2.region = param & 0x7;
+			break;
+		case NPU_SET_IFM2_WIDTH0_M1:
+			st.ifm2.width0 = param;
+			break;
+		case NPU_SET_IFM2_HEIGHT0_M1:
+			st.ifm2.height[0] = param;
+			break;
+		case NPU_SET_IFM2_HEIGHT1_M1:
+			st.ifm2.height[1] = param;
+			break;
+		case NPU_SET_IFM2_BASE0:
+		case NPU_SET_IFM2_BASE1:
+		case NPU_SET_IFM2_BASE2:
+		case NPU_SET_IFM2_BASE3:
+			st.ifm2.base[cmd & 0x3] = addr;
+			break;
+		case NPU_SET_IFM2_STRIDE_X:
+			st.ifm2.stride_x = addr;
+			break;
+		case NPU_SET_IFM2_STRIDE_Y:
+			st.ifm2.stride_y = addr;
+			break;
+		case NPU_SET_IFM2_STRIDE_C:
+			st.ifm2.stride_c = addr;
+			break;
+
+		case NPU_SET_WEIGHT_REGION:
+			st.weight[0].region = param & 0x7;
+			break;
+		case NPU_SET_SCALE_REGION:
+			st.scale[0].region = param & 0x7;
+			break;
+		case NPU_SET_WEIGHT_BASE:
+			st.weight[0].base = addr;
+			break;
+		case NPU_SET_WEIGHT_LENGTH:
+			st.weight[0].length = cmds[1];
+			break;
+		case NPU_SET_SCALE_BASE:
+			st.scale[0].base = addr;
+			break;
+		case NPU_SET_SCALE_LENGTH:
+			st.scale[0].length = cmds[1];
+			break;
+		case NPU_SET_WEIGHT1_BASE:
+			st.weight[1].base = addr;
+			break;
+		case NPU_SET_WEIGHT1_LENGTH:
+			st.weight[1].length = cmds[1];
+			break;
+		case NPU_SET_SCALE1_BASE: // NPU_SET_WEIGHT2_BASE (U85)
+			if (ethosu_is_u65(edev))
+				st.scale[1].base = addr;
+			else
+				st.weight[2].base = addr;
+			break;
+		case NPU_SET_SCALE1_LENGTH: // NPU_SET_WEIGHT2_LENGTH (U85)
+			if (ethosu_is_u65(edev))
+				st.scale[1].length = cmds[1];
+			else
+				st.weight[1].length = cmds[1];
+			break;
+		case NPU_SET_WEIGHT3_BASE:
+			st.weight[3].base = addr;
+			break;
+		case NPU_SET_WEIGHT3_LENGTH:
+			st.weight[3].length = cmds[1];
+			break;
+
+		case NPU_SET_DMA0_SRC_REGION:
+			if (param & 0x100)
+				st.dma.src.region = -1;
+			else
+				st.dma.src.region = param & 0x7;
+			st.dma.mode = (param >> 9) & 0x3;
+			break;
+		case NPU_SET_DMA0_DST_REGION:
+			if (param & 0x100)
+				st.dma.dst.region = -1;
+			else
+				st.dma.dst.region = param & 0x7;
+			break;
+		case NPU_SET_DMA0_SIZE0:
+			st.dma.size0 = param;
+			break;
+		case NPU_SET_DMA0_SIZE1:
+			st.dma.size1 = param;
+			break;
+		case NPU_SET_DMA0_SRC_STRIDE0:
+			st.dma.src.stride[0] = ((s64)addr << 24) >> 24;
+			break;
+		case NPU_SET_DMA0_SRC_STRIDE1:
+			st.dma.src.stride[1] = ((s64)addr << 24) >> 24;
+			break;
+		case NPU_SET_DMA0_DST_STRIDE0:
+			st.dma.dst.stride[0] = ((s64)addr << 24) >> 24;
+			break;
+		case NPU_SET_DMA0_DST_STRIDE1:
+			st.dma.dst.stride[1] = ((s64)addr << 24) >> 24;
+			break;
+		case NPU_SET_DMA0_SRC:
+			st.dma.src.offset = addr;
+			break;
+		case NPU_SET_DMA0_DST:
+			st.dma.dst.offset = addr;
+			break;
+		case NPU_SET_DMA0_LEN:
+			st.dma.src.len = st.dma.dst.len = addr;
+			break;
+		default:
+			break;
+		}
+	}
+
+	for (i = 0; i < NPU_BASEP_REGION_MAX; i++) {
+		if (!info->region_size[i])
+			continue;
+		dev_dbg(ddev->dev, "region %d max size: 0x%llx\n",
+			i, info->region_size[i]);
+	}
+
+	bo->info = no_free_ptr(info);
+	return 0;
+}
+
+/**
+ * ethosu_gem_cmdstream_create() - Create a GEM object and attach it to a handle.
+ * @file: DRM file.
+ * @ddev: DRM device.
+ * @exclusive_vm: Exclusive VM. Not NULL if the GEM object can't be shared.
+ * @size: Size of the GEM object to allocate.
+ * @flags: Combination of drm_ethosu_bo_flags flags.
+ * @handle: Pointer holding the handle pointing to the new GEM object.
+ *
+ * Return: Zero on success
+ */
+int ethosu_gem_cmdstream_create(struct drm_file *file,
+				struct drm_device *ddev,
+				u32 size, u64 data, u32 flags, u32 *handle)
+{
+	int ret;
+	struct drm_gem_dma_object *mem;
+	struct ethosu_gem_object *bo;
+
+	mem = drm_gem_dma_create(ddev, size);
+	if (IS_ERR(mem))
+		return PTR_ERR(mem);
+
+	bo = to_ethosu_bo(&mem->base);
+	bo->flags = flags;
+
+	ret = ethosu_gem_cmdstream_copy_and_validate(ddev,
+						     (void __user *)(uintptr_t)data,
+						     bo, size);
+	if (ret)
+		goto fail;
+
+	/*
+	 * Allocate an id of idr table where the obj is registered
+	 * and handle has the id what user can see.
+	 */
+	ret = drm_gem_handle_create(file, &mem->base, handle);
+
+fail:
+	/* drop reference from allocate - handle holds it now. */
+	drm_gem_object_put(&mem->base);
+
+	return ret;
+}
diff --git a/drivers/accel/ethosu/ethosu_gem.h b/drivers/accel/ethosu/ethosu_gem.h
new file mode 100644
index 000000000000..3922895a60fb
--- /dev/null
+++ b/drivers/accel/ethosu/ethosu_gem.h
@@ -0,0 +1,46 @@
+/* SPDX-License-Identifier: GPL-2.0 or MIT */
+/* Copyright 2025 Arm, Ltd. */
+
+#ifndef __ETHOSU_GEM_H__
+#define __ETHOSU_GEM_H__
+
+#include "ethosu_device.h"
+#include <drm/drm_gem_dma_helper.h>
+
+struct ethosu_validated_cmdstream_info {
+	u32 cmd_size;
+	u64 region_size[NPU_BASEP_REGION_MAX];
+	bool output_region[NPU_BASEP_REGION_MAX];
+};
+
+/**
+ * struct ethosu_gem_object - Driver specific GEM object.
+ */
+struct ethosu_gem_object {
+	/** @base: Inherit from drm_gem_shmem_object. */
+	struct drm_gem_dma_object base;
+
+	struct ethosu_validated_cmdstream_info *info;
+
+	/** @flags: Combination of drm_ethosu_bo_flags flags. */
+	u32 flags;
+};
+
+static inline
+struct ethosu_gem_object *to_ethosu_bo(struct drm_gem_object *obj)
+{
+	return container_of(to_drm_gem_dma_obj(obj), struct ethosu_gem_object, base);
+}
+
+struct drm_gem_object *ethosu_gem_create_object(struct drm_device *ddev,
+						size_t size);
+
+int ethosu_gem_create_with_handle(struct drm_file *file,
+				  struct drm_device *ddev,
+				  u64 *size, u32 flags, uint32_t *handle);
+
+int ethosu_gem_cmdstream_create(struct drm_file *file,
+				struct drm_device *ddev,
+				u32 size, u64 data, u32 flags, u32 *handle);
+
+#endif /* __ETHOSU_GEM_H__ */
diff --git a/drivers/accel/ethosu/ethosu_job.c b/drivers/accel/ethosu/ethosu_job.c
new file mode 100644
index 000000000000..26e7a2f64d71
--- /dev/null
+++ b/drivers/accel/ethosu/ethosu_job.c
@@ -0,0 +1,497 @@
+// SPDX-License-Identifier: GPL-2.0-only OR MIT
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+/* Copyright 2025 Arm, Ltd. */
+
+#include <linux/bitfield.h>
+#include <linux/genalloc.h>
+#include <linux/interrupt.h>
+#include <linux/iopoll.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+
+#include <drm/drm_file.h>
+#include <drm/drm_gem.h>
+#include <drm/drm_gem_dma_helper.h>
+#include <drm/drm_print.h>
+#include <drm/ethosu_accel.h>
+
+#include "ethosu_device.h"
+#include "ethosu_drv.h"
+#include "ethosu_gem.h"
+#include "ethosu_job.h"
+
+#define JOB_TIMEOUT_MS 500
+
+static struct ethosu_job *to_ethosu_job(struct drm_sched_job *sched_job)
+{
+	return container_of(sched_job, struct ethosu_job, base);
+}
+
+static const char *ethosu_fence_get_driver_name(struct dma_fence *fence)
+{
+	return "ethosu";
+}
+
+static const char *ethosu_fence_get_timeline_name(struct dma_fence *fence)
+{
+	return "ethosu-npu";
+}
+
+static const struct dma_fence_ops ethosu_fence_ops = {
+	.get_driver_name = ethosu_fence_get_driver_name,
+	.get_timeline_name = ethosu_fence_get_timeline_name,
+};
+
+static void ethosu_job_hw_submit(struct ethosu_device *dev, struct ethosu_job *job)
+{
+	struct drm_gem_dma_object *cmd_bo = to_drm_gem_dma_obj(job->cmd_bo);
+	struct ethosu_validated_cmdstream_info *cmd_info = to_ethosu_bo(job->cmd_bo)->info;
+
+	for (int i = 0; i < job->region_cnt; i++) {
+		struct drm_gem_dma_object *bo;
+		int region = job->region_bo_num[i];
+
+		bo = to_drm_gem_dma_obj(job->region_bo[i]);
+		writel_relaxed(lower_32_bits(bo->dma_addr), dev->regs + NPU_REG_BASEP(region));
+		writel_relaxed(upper_32_bits(bo->dma_addr), dev->regs + NPU_REG_BASEP_HI(region));
+		dev_dbg(dev->base.dev, "Region %d base addr = %pad\n", region, &bo->dma_addr);
+	}
+
+	if (job->sram_size) {
+		writel_relaxed(lower_32_bits(dev->sramphys),
+			       dev->regs + NPU_REG_BASEP(ETHOSU_SRAM_REGION));
+		writel_relaxed(upper_32_bits(dev->sramphys),
+			       dev->regs + NPU_REG_BASEP_HI(ETHOSU_SRAM_REGION));
+		dev_dbg(dev->base.dev, "Region %d base addr = %pad (SRAM)\n",
+			ETHOSU_SRAM_REGION, &dev->sramphys);
+	}
+
+	writel_relaxed(lower_32_bits(cmd_bo->dma_addr), dev->regs + NPU_REG_QBASE);
+	writel_relaxed(upper_32_bits(cmd_bo->dma_addr), dev->regs + NPU_REG_QBASE_HI);
+	writel_relaxed(cmd_info->cmd_size, dev->regs + NPU_REG_QSIZE);
+
+	writel(CMD_TRANSITION_TO_RUN, dev->regs + NPU_REG_CMD);
+
+	dev_dbg(dev->base.dev,
+		"Submitted cmd at %pad to core\n", &cmd_bo->dma_addr);
+}
+
+static int ethosu_acquire_object_fences(struct ethosu_job *job)
+{
+	int i, ret;
+	struct drm_gem_object **bos = job->region_bo;
+	struct ethosu_validated_cmdstream_info *info = to_ethosu_bo(job->cmd_bo)->info;
+
+	for (i = 0; i < job->region_cnt; i++) {
+		bool is_write;
+
+		if (!bos[i])
+			break;
+
+		ret = dma_resv_reserve_fences(bos[i]->resv, 1);
+		if (ret)
+			return ret;
+
+		is_write = info->output_region[job->region_bo_num[i]];
+		ret = drm_sched_job_add_implicit_dependencies(&job->base, bos[i],
+							      is_write);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static void ethosu_attach_object_fences(struct ethosu_job *job)
+{
+	int i;
+	struct dma_fence *fence = job->inference_done_fence;
+	struct drm_gem_object **bos = job->region_bo;
+	struct ethosu_validated_cmdstream_info *info = to_ethosu_bo(job->cmd_bo)->info;
+
+	for (i = 0; i < job->region_cnt; i++)
+		if (info->output_region[job->region_bo_num[i]])
+			dma_resv_add_fence(bos[i]->resv, fence, DMA_RESV_USAGE_WRITE);
+}
+
+static int ethosu_job_push(struct ethosu_job *job)
+{
+	struct ww_acquire_ctx acquire_ctx;
+	int ret;
+
+	ret = drm_gem_lock_reservations(job->region_bo, job->region_cnt, &acquire_ctx);
+	if (ret)
+		return ret;
+
+	ret = ethosu_acquire_object_fences(job);
+	if (ret)
+		goto out;
+
+	ret = pm_runtime_resume_and_get(job->dev->base.dev);
+	if (!ret) {
+		guard(mutex)(&job->dev->sched_lock);
+
+		drm_sched_job_arm(&job->base);
+		job->inference_done_fence = dma_fence_get(&job->base.s_fence->finished);
+		kref_get(&job->refcount); /* put by scheduler job completion */
+		drm_sched_entity_push_job(&job->base);
+		ethosu_attach_object_fences(job);
+	}
+
+out:
+	drm_gem_unlock_reservations(job->region_bo, job->region_cnt, &acquire_ctx);
+	return ret;
+}
+
+static void ethosu_job_cleanup(struct kref *ref)
+{
+	struct ethosu_job *job = container_of(ref, struct ethosu_job,
+						refcount);
+	unsigned int i;
+
+	pm_runtime_put_autosuspend(job->dev->base.dev);
+
+	dma_fence_put(job->done_fence);
+	dma_fence_put(job->inference_done_fence);
+
+	for (i = 0; i < job->region_cnt; i++)
+		drm_gem_object_put(job->region_bo[i]);
+
+	drm_gem_object_put(job->cmd_bo);
+
+	kfree(job);
+}
+
+static void ethosu_job_put(struct ethosu_job *job)
+{
+	kref_put(&job->refcount, ethosu_job_cleanup);
+}
+
+static void ethosu_job_free(struct drm_sched_job *sched_job)
+{
+	struct ethosu_job *job = to_ethosu_job(sched_job);
+
+	drm_sched_job_cleanup(sched_job);
+	ethosu_job_put(job);
+}
+
+static struct dma_fence *ethosu_job_run(struct drm_sched_job *sched_job)
+{
+	struct ethosu_job *job = to_ethosu_job(sched_job);
+	struct ethosu_device *dev = job->dev;
+	struct dma_fence *fence = job->done_fence;
+
+	if (unlikely(job->base.s_fence->finished.error))
+		return NULL;
+
+	dma_fence_init(fence, &ethosu_fence_ops, &dev->fence_lock,
+		       dev->fence_context, ++dev->emit_seqno);
+	dma_fence_get(fence);
+
+	scoped_guard(mutex, &dev->job_lock) {
+		dev->in_flight_job = job;
+		ethosu_job_hw_submit(dev, job);
+	}
+
+	return fence;
+}
+
+static void ethosu_job_handle_irq(struct ethosu_device *dev)
+{
+	u32 status = readl_relaxed(dev->regs + NPU_REG_STATUS);
+
+	if (status & (STATUS_BUS_STATUS | STATUS_CMD_PARSE_ERR)) {
+		dev_err(dev->base.dev, "Error IRQ - %x\n", status);
+		drm_sched_fault(&dev->sched);
+		return;
+	}
+
+	scoped_guard(mutex, &dev->job_lock) {
+		if (dev->in_flight_job) {
+			dma_fence_signal(dev->in_flight_job->done_fence);
+			dev->in_flight_job = NULL;
+		}
+	}
+}
+
+static irqreturn_t ethosu_job_irq_handler_thread(int irq, void *data)
+{
+	struct ethosu_device *dev = data;
+
+	ethosu_job_handle_irq(dev);
+
+	return IRQ_HANDLED;
+}
+
+static irqreturn_t ethosu_job_irq_handler(int irq, void *data)
+{
+	struct ethosu_device *dev = data;
+	u32 status = readl_relaxed(dev->regs + NPU_REG_STATUS);
+
+	if (!(status & STATUS_IRQ_RAISED))
+		return IRQ_NONE;
+
+	writel_relaxed(CMD_CLEAR_IRQ, dev->regs + NPU_REG_CMD);
+	return IRQ_WAKE_THREAD;
+}
+
+static enum drm_gpu_sched_stat ethosu_job_timedout(struct drm_sched_job *bad)
+{
+	struct ethosu_job *job = to_ethosu_job(bad);
+	struct ethosu_device *dev = job->dev;
+	bool running;
+	u32 *bocmds = to_drm_gem_dma_obj(job->cmd_bo)->vaddr;
+	u32 cmdaddr;
+
+	cmdaddr = readl_relaxed(dev->regs + NPU_REG_QREAD);
+	running = FIELD_GET(STATUS_STATE_RUNNING, readl_relaxed(dev->regs + NPU_REG_STATUS));
+
+	if (running) {
+		int ret;
+		u32 reg;
+
+		ret = readl_relaxed_poll_timeout(dev->regs + NPU_REG_QREAD,
+						 reg,
+						 reg != cmdaddr,
+						 USEC_PER_MSEC, 100 * USEC_PER_MSEC);
+
+		/* If still running and progress is being made, just return */
+		if (!ret)
+			return DRM_GPU_SCHED_STAT_NO_HANG;
+	}
+
+	dev_err(dev->base.dev, "NPU sched timed out: NPU %s, cmdstream offset 0x%x: 0x%x\n",
+		running ? "running" : "stopped",
+		cmdaddr, bocmds[cmdaddr / 4]);
+
+	drm_sched_stop(&dev->sched, bad);
+
+	scoped_guard(mutex, &dev->job_lock)
+		dev->in_flight_job = NULL;
+
+	/* Proceed with reset now. */
+	pm_runtime_force_suspend(dev->base.dev);
+	pm_runtime_force_resume(dev->base.dev);
+
+	/* Restart the scheduler */
+	drm_sched_start(&dev->sched, 0);
+
+	return DRM_GPU_SCHED_STAT_RESET;
+}
+
+static const struct drm_sched_backend_ops ethosu_sched_ops = {
+	.run_job = ethosu_job_run,
+	.timedout_job = ethosu_job_timedout,
+	.free_job = ethosu_job_free
+};
+
+int ethosu_job_init(struct ethosu_device *edev)
+{
+	struct device *dev = edev->base.dev;
+	struct drm_sched_init_args args = {
+		.ops = &ethosu_sched_ops,
+		.num_rqs = DRM_SCHED_PRIORITY_COUNT,
+		.credit_limit = 1,
+		.timeout = msecs_to_jiffies(JOB_TIMEOUT_MS),
+		.name = dev_name(dev),
+		.dev = dev,
+	};
+	int ret;
+
+	spin_lock_init(&edev->fence_lock);
+	ret = devm_mutex_init(dev, &edev->job_lock);
+	if (ret)
+		return ret;
+	ret = devm_mutex_init(dev, &edev->sched_lock);
+	if (ret)
+		return ret;
+
+	edev->irq = platform_get_irq(to_platform_device(dev), 0);
+	if (edev->irq < 0)
+		return edev->irq;
+
+	ret = devm_request_threaded_irq(dev, edev->irq,
+					ethosu_job_irq_handler,
+					ethosu_job_irq_handler_thread,
+					IRQF_SHARED, KBUILD_MODNAME,
+					edev);
+	if (ret) {
+		dev_err(dev, "failed to request irq\n");
+		return ret;
+	}
+
+	edev->fence_context = dma_fence_context_alloc(1);
+
+	ret = drm_sched_init(&edev->sched, &args);
+	if (ret) {
+		dev_err(dev, "Failed to create scheduler: %d\n", ret);
+		goto err_sched;
+	}
+
+	return 0;
+
+err_sched:
+	drm_sched_fini(&edev->sched);
+	return ret;
+}
+
+void ethosu_job_fini(struct ethosu_device *dev)
+{
+	drm_sched_fini(&dev->sched);
+}
+
+int ethosu_job_open(struct ethosu_file_priv *ethosu_priv)
+{
+	struct ethosu_device *dev = ethosu_priv->edev;
+	struct drm_gpu_scheduler *sched = &dev->sched;
+	int ret;
+
+	ret = drm_sched_entity_init(&ethosu_priv->sched_entity,
+				    DRM_SCHED_PRIORITY_NORMAL,
+				    &sched, 1, NULL);
+	return WARN_ON(ret);
+}
+
+void ethosu_job_close(struct ethosu_file_priv *ethosu_priv)
+{
+	struct drm_sched_entity *entity = &ethosu_priv->sched_entity;
+
+	drm_sched_entity_destroy(entity);
+}
+
+static int ethosu_ioctl_submit_job(struct drm_device *dev, struct drm_file *file,
+				   struct drm_ethosu_job *job)
+{
+	struct ethosu_device *edev = to_ethosu_device(dev);
+	struct ethosu_file_priv *file_priv = file->driver_priv;
+	struct ethosu_job *ejob = NULL;
+	struct ethosu_validated_cmdstream_info *cmd_info;
+	int ret = 0;
+
+	/* BO region 2 is reserved if SRAM is used */
+	if (job->region_bo_handles[ETHOSU_SRAM_REGION] && job->sram_size)
+		return -EINVAL;
+
+	if (edev->npu_info.sram_size < job->sram_size)
+		return -EINVAL;
+
+	ejob = kzalloc(sizeof(*ejob), GFP_KERNEL);
+	if (!ejob)
+		return -ENOMEM;
+
+	kref_init(&ejob->refcount);
+
+	ejob->dev = edev;
+	ejob->sram_size = job->sram_size;
+
+	ejob->done_fence = kzalloc(sizeof(*ejob->done_fence), GFP_KERNEL);
+	if (!ejob->done_fence) {
+		ret = -ENOMEM;
+		goto out_cleanup_job;
+	}
+
+	ret = drm_sched_job_init(&ejob->base,
+				 &file_priv->sched_entity,
+				 1, NULL, file->client_id);
+	if (ret)
+		goto out_put_job;
+
+	ejob->cmd_bo = drm_gem_object_lookup(file, job->cmd_bo);
+	if (!ejob->cmd_bo) {
+		ret = -ENOENT;
+		goto out_cleanup_job;
+	}
+	cmd_info = to_ethosu_bo(ejob->cmd_bo)->info;
+	if (!cmd_info) {
+		ret = -EINVAL;
+		goto out_cleanup_job;
+	}
+
+	for (int i = 0; i < NPU_BASEP_REGION_MAX; i++) {
+		struct drm_gem_object *gem;
+
+		/* Can only omit a BO handle if the region is not used or used for SRAM */
+		if (!job->region_bo_handles[i] &&
+		    (!cmd_info->region_size[i] || (i == ETHOSU_SRAM_REGION && job->sram_size)))
+			continue;
+
+		if (job->region_bo_handles[i] && !cmd_info->region_size[i]) {
+			dev_err(dev->dev,
+				"Cmdstream BO handle %d set for unused region %d\n",
+				job->region_bo_handles[i], i);
+			ret = -EINVAL;
+			goto out_cleanup_job;
+		}
+
+		gem = drm_gem_object_lookup(file, job->region_bo_handles[i]);
+		if (!gem) {
+			dev_err(dev->dev,
+				"Invalid BO handle %d for region %d\n",
+				job->region_bo_handles[i], i);
+			ret = -ENOENT;
+			goto out_cleanup_job;
+		}
+
+		ejob->region_bo[ejob->region_cnt] = gem;
+		ejob->region_bo_num[ejob->region_cnt] = i;
+		ejob->region_cnt++;
+
+		if (to_ethosu_bo(gem)->info) {
+			dev_err(dev->dev,
+				"Cmdstream BO handle %d used for region %d\n",
+				job->region_bo_handles[i], i);
+			ret = -EINVAL;
+			goto out_cleanup_job;
+		}
+
+		/* Verify the command stream doesn't have accesses outside the BO */
+		if (cmd_info->region_size[i] > gem->size) {
+			dev_err(dev->dev,
+				"cmd stream region %d size greater than BO size (%llu > %zu)\n",
+				i, cmd_info->region_size[i], gem->size);
+			ret = -EOVERFLOW;
+			goto out_cleanup_job;
+		}
+	}
+	ret = ethosu_job_push(ejob);
+
+out_cleanup_job:
+	if (ret)
+		drm_sched_job_cleanup(&ejob->base);
+out_put_job:
+	ethosu_job_put(ejob);
+
+	return ret;
+}
+
+int ethosu_ioctl_submit(struct drm_device *dev, void *data, struct drm_file *file)
+{
+	struct drm_ethosu_submit *args = data;
+	int ret = 0;
+	unsigned int i = 0;
+
+	if (args->pad) {
+		drm_dbg(dev, "Reserved field in drm_ethosu_submit struct should be 0.\n");
+		return -EINVAL;
+	}
+
+	struct drm_ethosu_job __free(kvfree) *jobs =
+		kvmalloc_array(args->job_count, sizeof(*jobs), GFP_KERNEL);
+	if (!jobs)
+		return -ENOMEM;
+
+	if (copy_from_user(jobs,
+			   (void __user *)(uintptr_t)args->jobs,
+			   args->job_count * sizeof(*jobs))) {
+		drm_dbg(dev, "Failed to copy incoming job array\n");
+		return -EFAULT;
+	}
+
+	for (i = 0; i < args->job_count; i++) {
+		ret = ethosu_ioctl_submit_job(dev, file, &jobs[i]);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
diff --git a/drivers/accel/ethosu/ethosu_job.h b/drivers/accel/ethosu/ethosu_job.h
new file mode 100644
index 000000000000..ff1cf448d094
--- /dev/null
+++ b/drivers/accel/ethosu/ethosu_job.h
@@ -0,0 +1,40 @@
+/* SPDX-License-Identifier: GPL-2.0-only OR MIT */
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+/* Copyright 2025 Arm, Ltd. */
+
+#ifndef __ETHOSU_JOB_H__
+#define __ETHOSU_JOB_H__
+
+#include <linux/kref.h>
+#include <drm/gpu_scheduler.h>
+
+struct ethosu_device;
+struct ethosu_file_priv;
+
+struct ethosu_job {
+	struct drm_sched_job base;
+	struct ethosu_device *dev;
+
+	struct drm_gem_object *cmd_bo;
+	struct drm_gem_object *region_bo[NPU_BASEP_REGION_MAX];
+	u8 region_bo_num[NPU_BASEP_REGION_MAX];
+	u8 region_cnt;
+	u32 sram_size;
+
+	/* Fence to be signaled by drm-sched once its done with the job */
+	struct dma_fence *inference_done_fence;
+
+	/* Fence to be signaled by IRQ handler when the job is complete. */
+	struct dma_fence *done_fence;
+
+	struct kref refcount;
+};
+
+int ethosu_ioctl_submit(struct drm_device *dev, void *data, struct drm_file *file);
+
+int ethosu_job_init(struct ethosu_device *dev);
+void ethosu_job_fini(struct ethosu_device *dev);
+int ethosu_job_open(struct ethosu_file_priv *ethosu_priv);
+void ethosu_job_close(struct ethosu_file_priv *ethosu_priv);
+
+#endif
diff --git a/drivers/accel/habanalabs/Kconfig b/drivers/accel/habanalabs/Kconfig
index 1919fbb169c7..6d1506acbd72 100644
--- a/drivers/accel/habanalabs/Kconfig
+++ b/drivers/accel/habanalabs/Kconfig
@@ -27,3 +27,26 @@ config DRM_ACCEL_HABANALABS
 
 	  To compile this driver as a module, choose M here: the
 	  module will be called habanalabs.
+
+if DRM_ACCEL_HABANALABS
+
+config HL_HLDIO
+	bool "Habanalabs NVMe Direct I/O (HLDIO)"
+	depends on PCI_P2PDMA
+	depends on BLOCK
+	help
+	  Enable NVMe peer-to-peer direct I/O support for Habanalabs AI
+	  accelerators.
+
+	  This allows direct data transfers between NVMe storage devices
+	  and Habanalabs accelerators without involving system memory,
+	  using PCI peer-to-peer DMA capabilities.
+
+	  Requirements:
+	  - CONFIG_PCI_P2PDMA=y
+	  - NVMe device and Habanalabs accelerator under same PCI root complex
+	  - IOMMU disabled or in passthrough mode
+	  - Hardware supporting PCI P2P DMA
+
+	  If unsure, say N
+endif # DRM_ACCEL_HABANALABS
diff --git a/drivers/accel/habanalabs/common/Makefile b/drivers/accel/habanalabs/common/Makefile
index e6abffea9f87..b6d00de09db5 100644
--- a/drivers/accel/habanalabs/common/Makefile
+++ b/drivers/accel/habanalabs/common/Makefile
@@ -13,3 +13,8 @@ HL_COMMON_FILES := common/habanalabs_drv.o common/device.o common/context.o \
 		common/command_submission.o common/firmware_if.o \
 		common/security.o common/state_dump.o \
 		common/memory_mgr.o common/decoder.o
+
+# Conditionally add HLDIO support
+ifdef CONFIG_HL_HLDIO
+HL_COMMON_FILES += common/hldio.o
+endif
+\ No newline at end of file
diff --git a/drivers/accel/habanalabs/common/debugfs.c b/drivers/accel/habanalabs/common/debugfs.c
index 4b391807e5f2..5f0820b19ccb 100644
--- a/drivers/accel/habanalabs/common/debugfs.c
+++ b/drivers/accel/habanalabs/common/debugfs.c
@@ -6,6 +6,7 @@
  */
 
 #include "habanalabs.h"
+#include "hldio.h"
 #include "../include/hw_ip/mmu/mmu_general.h"
 
 #include <linux/pci.h>
@@ -602,6 +603,198 @@ static int engines_show(struct seq_file *s, void *data)
 	return 0;
 }
 
+#ifdef CONFIG_HL_HLDIO
+/* DIO debugfs functions following the standard pattern */
+static int dio_ssd2hl_show(struct seq_file *s, void *data)
+{
+	struct hl_debugfs_entry *entry = s->private;
+	struct hl_dbg_device_entry *dev_entry = entry->dev_entry;
+	struct hl_device *hdev = dev_entry->hdev;
+
+	if (!hdev->asic_prop.supports_nvme) {
+		seq_puts(s, "NVMe Direct I/O not supported\\n");
+		return 0;
+	}
+
+	seq_puts(s, "Usage: echo \"fd=N va=0xADDR off=N len=N\" > dio_ssd2hl\n");
+	seq_printf(s, "Last transfer: %zu bytes\\n", dev_entry->dio_stats.last_len_read);
+	seq_puts(s, "Note: All parameters must be page-aligned (4KB)\\n");
+
+	return 0;
+}
+
+static ssize_t dio_ssd2hl_write(struct file *file, const char __user *buf,
+				size_t count, loff_t *f_pos)
+{
+	struct seq_file *s = file->private_data;
+	struct hl_debugfs_entry *entry = s->private;
+	struct hl_dbg_device_entry *dev_entry = entry->dev_entry;
+	struct hl_device *hdev = dev_entry->hdev;
+	struct hl_ctx *ctx = hdev->kernel_ctx;
+	char kbuf[128];
+	u64 device_va = 0, off_bytes = 0, len_bytes = 0;
+	u32 fd = 0;
+	size_t len_read = 0;
+	int rc, parsed;
+
+	if (!hdev->asic_prop.supports_nvme)
+		return -EOPNOTSUPP;
+
+	if (count >= sizeof(kbuf))
+		return -EINVAL;
+
+	if (copy_from_user(kbuf, buf, count))
+		return -EFAULT;
+
+	kbuf[count] = 0;
+
+	/* Parse: fd=N va=0xADDR off=N len=N */
+	parsed = sscanf(kbuf, "fd=%u va=0x%llx off=%llu len=%llu",
+			&fd, &device_va, &off_bytes, &len_bytes);
+	if (parsed != 4) {
+		dev_err(hdev->dev, "Invalid format. Expected: fd=N va=0xADDR off=N len=N\\n");
+		return -EINVAL;
+	}
+
+	/* Validate file descriptor */
+	if (fd == 0) {
+		dev_err(hdev->dev, "Invalid file descriptor: %u\\n", fd);
+		return -EINVAL;
+	}
+
+	/* Validate alignment requirements */
+	if (!IS_ALIGNED(device_va, PAGE_SIZE) ||
+	    !IS_ALIGNED(off_bytes, PAGE_SIZE) ||
+	    !IS_ALIGNED(len_bytes, PAGE_SIZE)) {
+		dev_err(hdev->dev,
+			"All parameters must be page-aligned (4KB)\\n");
+		return -EINVAL;
+	}
+
+	/* Validate transfer size */
+	if (len_bytes == 0 || len_bytes > SZ_1G) {
+		dev_err(hdev->dev, "Invalid length: %llu (max 1GB)\\n",
+			len_bytes);
+		return -EINVAL;
+	}
+
+	dev_dbg(hdev->dev, "DIO SSD2HL: fd=%u va=0x%llx off=%llu len=%llu\\n",
+		fd, device_va, off_bytes, len_bytes);
+
+	rc = hl_dio_ssd2hl(hdev, ctx, fd, device_va, off_bytes, len_bytes, &len_read);
+	if (rc < 0) {
+		dev_entry->dio_stats.failed_ops++;
+		dev_err(hdev->dev, "SSD2HL operation failed: %d\\n", rc);
+		return rc;
+	}
+
+	/* Update statistics */
+	dev_entry->dio_stats.total_ops++;
+	dev_entry->dio_stats.successful_ops++;
+	dev_entry->dio_stats.bytes_transferred += len_read;
+	dev_entry->dio_stats.last_len_read = len_read;
+
+	dev_dbg(hdev->dev, "DIO SSD2HL completed: %zu bytes transferred\\n", len_read);
+
+	return count;
+}
+
+static int dio_hl2ssd_show(struct seq_file *s, void *data)
+{
+	seq_puts(s, "HL2SSD (device-to-SSD) transfers not implemented\\n");
+	return 0;
+}
+
+static ssize_t dio_hl2ssd_write(struct file *file, const char __user *buf,
+			       size_t count, loff_t *f_pos)
+{
+	struct seq_file *s = file->private_data;
+	struct hl_debugfs_entry *entry = s->private;
+	struct hl_dbg_device_entry *dev_entry = entry->dev_entry;
+	struct hl_device *hdev = dev_entry->hdev;
+
+	if (!hdev->asic_prop.supports_nvme)
+		return -EOPNOTSUPP;
+
+	dev_dbg(hdev->dev, "HL2SSD operation not implemented\\n");
+	return -EOPNOTSUPP;
+}
+
+static int dio_stats_show(struct seq_file *s, void *data)
+{
+	struct hl_debugfs_entry *entry = s->private;
+	struct hl_dbg_device_entry *dev_entry = entry->dev_entry;
+	struct hl_device *hdev = dev_entry->hdev;
+	struct hl_dio_stats *stats = &dev_entry->dio_stats;
+	u64 avg_bytes_per_op = 0, success_rate = 0;
+
+	if (!hdev->asic_prop.supports_nvme) {
+		seq_puts(s, "NVMe Direct I/O not supported\\n");
+		return 0;
+	}
+
+	if (stats->successful_ops > 0)
+		avg_bytes_per_op = stats->bytes_transferred / stats->successful_ops;
+
+	if (stats->total_ops > 0)
+		success_rate = (stats->successful_ops * 100) / stats->total_ops;
+
+	seq_puts(s, "=== Habanalabs Direct I/O Statistics ===\\n");
+	seq_printf(s, "Total operations:     %llu\\n", stats->total_ops);
+	seq_printf(s, "Successful ops:       %llu\\n", stats->successful_ops);
+	seq_printf(s, "Failed ops:           %llu\\n", stats->failed_ops);
+	seq_printf(s, "Success rate:         %llu%%\\n", success_rate);
+	seq_printf(s, "Total bytes:          %llu\\n", stats->bytes_transferred);
+	seq_printf(s, "Avg bytes per op:     %llu\\n", avg_bytes_per_op);
+	seq_printf(s, "Last transfer:        %zu bytes\\n", stats->last_len_read);
+
+	return 0;
+}
+
+static int dio_reset_show(struct seq_file *s, void *data)
+{
+	seq_puts(s, "Write '1' to reset DIO statistics\\n");
+	return 0;
+}
+
+static ssize_t dio_reset_write(struct file *file, const char __user *buf,
+			       size_t count, loff_t *f_pos)
+{
+	struct seq_file *s = file->private_data;
+	struct hl_debugfs_entry *entry = s->private;
+	struct hl_dbg_device_entry *dev_entry = entry->dev_entry;
+	struct hl_device *hdev = dev_entry->hdev;
+	char kbuf[8];
+	unsigned long val;
+	int rc;
+
+	if (!hdev->asic_prop.supports_nvme)
+		return -EOPNOTSUPP;
+
+	if (count >= sizeof(kbuf))
+		return -EINVAL;
+
+	if (copy_from_user(kbuf, buf, count))
+		return -EFAULT;
+
+	kbuf[count] = 0;
+
+	rc = kstrtoul(kbuf, 0, &val);
+	if (rc)
+		return rc;
+
+	if (val == 1) {
+		memset(&dev_entry->dio_stats, 0, sizeof(dev_entry->dio_stats));
+		dev_dbg(hdev->dev, "DIO statistics reset\\n");
+	} else {
+		dev_err(hdev->dev, "Write '1' to reset statistics\\n");
+		return -EINVAL;
+	}
+
+	return count;
+}
+#endif
+
 static ssize_t hl_memory_scrub(struct file *f, const char __user *buf,
 					size_t count, loff_t *ppos)
 {
@@ -788,6 +981,113 @@ static void hl_access_host_mem(struct hl_device *hdev, u64 addr, u64 *val,
 	}
 }
 
+static void dump_cfg_access_entry(struct hl_device *hdev,
+				  struct hl_debugfs_cfg_access_entry *entry)
+{
+	char *access_type = "";
+	struct tm tm;
+
+	switch (entry->debugfs_type) {
+	case DEBUGFS_READ32:
+		access_type = "READ32 from";
+		break;
+	case DEBUGFS_WRITE32:
+		access_type = "WRITE32 to";
+		break;
+	case DEBUGFS_READ64:
+		access_type = "READ64 from";
+		break;
+	case DEBUGFS_WRITE64:
+		access_type = "WRITE64 to";
+		break;
+	default:
+		dev_err(hdev->dev, "Invalid DEBUGFS access type (%u)\n", entry->debugfs_type);
+		return;
+	}
+
+	time64_to_tm(entry->seconds_since_epoch, 0, &tm);
+	dev_info(hdev->dev,
+		"%ld-%02d-%02d %02d:%02d:%02d (UTC): %s %#llx\n", tm.tm_year + 1900, tm.tm_mon + 1,
+		tm.tm_mday, tm.tm_hour, tm.tm_min, tm.tm_sec, access_type, entry->addr);
+}
+
+void hl_debugfs_cfg_access_history_dump(struct hl_device *hdev)
+{
+	struct hl_debugfs_cfg_access *dbgfs = &hdev->debugfs_cfg_accesses;
+	u32 i, head, count = 0;
+	time64_t entry_time, now;
+	unsigned long flags;
+
+	now = ktime_get_real_seconds();
+
+	spin_lock_irqsave(&dbgfs->lock, flags);
+	head = dbgfs->head;
+	if (head == 0)
+		i = HL_DBGFS_CFG_ACCESS_HIST_LEN - 1;
+	else
+		i = head - 1;
+
+	/* Walk back until timeout or invalid entry */
+	while (dbgfs->cfg_access_list[i].valid) {
+		entry_time = dbgfs->cfg_access_list[i].seconds_since_epoch;
+		/* Stop when entry is older than timeout */
+		if (now - entry_time > HL_DBGFS_CFG_ACCESS_HIST_TIMEOUT_SEC)
+			break;
+
+		/* print single entry under lock */
+		{
+			struct hl_debugfs_cfg_access_entry entry = dbgfs->cfg_access_list[i];
+			/*
+			 * We copy the entry out under lock and then print after
+			 * releasing the lock to minimize time under lock.
+			 */
+			spin_unlock_irqrestore(&dbgfs->lock, flags);
+			dump_cfg_access_entry(hdev, &entry);
+			spin_lock_irqsave(&dbgfs->lock, flags);
+		}
+
+		/* mark consumed */
+		dbgfs->cfg_access_list[i].valid = false;
+
+		if (i == 0)
+			i = HL_DBGFS_CFG_ACCESS_HIST_LEN - 1;
+		else
+			i--;
+		count++;
+		if (count >= HL_DBGFS_CFG_ACCESS_HIST_LEN)
+			break;
+	}
+	spin_unlock_irqrestore(&dbgfs->lock, flags);
+}
+
+static void check_if_cfg_access_and_log(struct hl_device *hdev, u64 addr, size_t access_size,
+					enum debugfs_access_type access_type)
+{
+	struct hl_debugfs_cfg_access *dbgfs_cfg_accesses = &hdev->debugfs_cfg_accesses;
+	struct pci_mem_region *mem_reg = &hdev->pci_mem_region[PCI_REGION_CFG];
+	struct hl_debugfs_cfg_access_entry *new_entry;
+	unsigned long flags;
+
+	/* Check if address is in config memory */
+	if (addr >= mem_reg->region_base &&
+		mem_reg->region_size >= access_size &&
+		addr <= mem_reg->region_base + mem_reg->region_size - access_size) {
+
+		spin_lock_irqsave(&dbgfs_cfg_accesses->lock, flags);
+
+		new_entry = &dbgfs_cfg_accesses->cfg_access_list[dbgfs_cfg_accesses->head];
+		new_entry->seconds_since_epoch = ktime_get_real_seconds();
+		new_entry->addr = addr;
+		new_entry->debugfs_type = access_type;
+		new_entry->valid = true;
+		dbgfs_cfg_accesses->head = (dbgfs_cfg_accesses->head + 1)
+						% HL_DBGFS_CFG_ACCESS_HIST_LEN;
+
+		spin_unlock_irqrestore(&dbgfs_cfg_accesses->lock, flags);
+
+	}
+}
+
 static int hl_access_mem(struct hl_device *hdev, u64 addr, u64 *val,
 				enum debugfs_access_type acc_type)
 {
@@ -805,6 +1105,7 @@ static int hl_access_mem(struct hl_device *hdev, u64 addr, u64 *val,
 			return rc;
 	}
 
+	check_if_cfg_access_and_log(hdev, addr, acc_size, acc_type);
 	rc = hl_access_dev_mem_by_region(hdev, addr, val, acc_type, &found);
 	if (rc) {
 		dev_err(hdev->dev,
@@ -1525,6 +1826,13 @@ static const struct hl_info_list hl_debugfs_list[] = {
 	{"mmu", mmu_show, mmu_asid_va_write},
 	{"mmu_error", mmu_ack_error, mmu_ack_error_value_write},
 	{"engines", engines_show, NULL},
+#ifdef CONFIG_HL_HLDIO
+	/* DIO entries - only created if NVMe is supported */
+	{"dio_ssd2hl", dio_ssd2hl_show, dio_ssd2hl_write},
+	{"dio_stats", dio_stats_show, NULL},
+	{"dio_reset", dio_reset_show, dio_reset_write},
+	{"dio_hl2ssd", dio_hl2ssd_show, dio_hl2ssd_write},
+#endif
 };
 
 static int hl_debugfs_open(struct inode *inode, struct file *file)
@@ -1723,6 +2031,11 @@ static void add_files_to_device(struct hl_device *hdev, struct hl_dbg_device_ent
 				&hdev->asic_prop.server_type);
 
 	for (i = 0, entry = dev_entry->entry_arr ; i < count ; i++, entry++) {
+		/* Skip DIO entries if NVMe is not supported */
+		if (strncmp(hl_debugfs_list[i].name, "dio_", 4) == 0 &&
+		    !hdev->asic_prop.supports_nvme)
+			continue;
+
 		debugfs_create_file(hl_debugfs_list[i].name,
 					0644,
 					root,
@@ -1762,6 +2075,14 @@ int hl_debugfs_device_init(struct hl_device *hdev)
 	spin_lock_init(&dev_entry->userptr_spinlock);
 	mutex_init(&dev_entry->ctx_mem_hash_mutex);
 
+	spin_lock_init(&hdev->debugfs_cfg_accesses.lock);
+	hdev->debugfs_cfg_accesses.head = 0; /* already zero by alloc but explicit init is fine */
+
+#ifdef CONFIG_HL_HLDIO
+	/* Initialize DIO statistics */
+	memset(&dev_entry->dio_stats, 0, sizeof(dev_entry->dio_stats));
+#endif
+
 	return 0;
 }
 
@@ -1780,6 +2101,7 @@ void hl_debugfs_device_fini(struct hl_device *hdev)
 		vfree(entry->state_dump[i]);
 
 	kfree(entry->entry_arr);
+
 }
 
 void hl_debugfs_add_device(struct hl_device *hdev)
@@ -1792,6 +2114,7 @@ void hl_debugfs_add_device(struct hl_device *hdev)
 
 	if (!hdev->asic_prop.fw_security_enabled)
 		add_secured_nodes(dev_entry, dev_entry->root);
+
 }
 
 void hl_debugfs_add_file(struct hl_fpriv *hpriv)
@@ -1924,3 +2247,4 @@ void hl_debugfs_set_state_dump(struct hl_device *hdev, char *data,
 
 	up_write(&dev_entry->state_dump_sem);
 }
+
diff --git a/drivers/accel/habanalabs/common/device.c b/drivers/accel/habanalabs/common/device.c
index 68eebed3b050..999c92d7036e 100644
--- a/drivers/accel/habanalabs/common/device.c
+++ b/drivers/accel/habanalabs/common/device.c
@@ -1066,28 +1066,11 @@ static bool is_pci_link_healthy(struct hl_device *hdev)
 	return (device_id == hdev->pdev->device);
 }
 
-static void stringify_time_of_last_heartbeat(struct hl_device *hdev, char *time_str, size_t size,
-						bool is_pq_hb)
-{
-	time64_t seconds = is_pq_hb ? hdev->heartbeat_debug_info.last_pq_heartbeat_ts
-					: hdev->heartbeat_debug_info.last_eq_heartbeat_ts;
-	struct tm tm;
-
-	if (!seconds)
-		return;
-
-	time64_to_tm(seconds, 0, &tm);
-
-	snprintf(time_str, size, "%ld-%02d-%02d %02d:%02d:%02d (UTC)",
-		tm.tm_year + 1900, tm.tm_mon, tm.tm_mday, tm.tm_hour, tm.tm_min, tm.tm_sec);
-}
-
 static bool hl_device_eq_heartbeat_received(struct hl_device *hdev)
 {
 	struct eq_heartbeat_debug_info *heartbeat_debug_info = &hdev->heartbeat_debug_info;
 	u32 cpu_q_id = heartbeat_debug_info->cpu_queue_id, pq_pi_mask = (HL_QUEUE_LENGTH << 1) - 1;
 	struct asic_fixed_properties *prop = &hdev->asic_prop;
-	char pq_time_str[64] = "N/A", eq_time_str[64] = "N/A";
 
 	if (!prop->cpucp_info.eq_health_check_supported)
 		return true;
@@ -1095,17 +1078,15 @@ static bool hl_device_eq_heartbeat_received(struct hl_device *hdev)
 	if (!hdev->eq_heartbeat_received) {
 		dev_err(hdev->dev, "EQ heartbeat event was not received!\n");
 
-		stringify_time_of_last_heartbeat(hdev, pq_time_str, sizeof(pq_time_str), true);
-		stringify_time_of_last_heartbeat(hdev, eq_time_str, sizeof(eq_time_str), false);
 		dev_err(hdev->dev,
-			"EQ: {CI %u, HB counter %u, last HB time: %s}, PQ: {PI: %u, CI: %u (%u), last HB time: %s}\n",
+			"EQ: {CI %u, HB counter %u, last HB time: %ptTs}, PQ: {PI: %u, CI: %u (%u), last HB time: %ptTs}\n",
 			hdev->event_queue.ci,
 			heartbeat_debug_info->heartbeat_event_counter,
-			eq_time_str,
+			&hdev->heartbeat_debug_info.last_eq_heartbeat_ts,
 			hdev->kernel_queues[cpu_q_id].pi,
 			atomic_read(&hdev->kernel_queues[cpu_q_id].ci),
 			atomic_read(&hdev->kernel_queues[cpu_q_id].ci) & pq_pi_mask,
-			pq_time_str);
+			&hdev->heartbeat_debug_info.last_pq_heartbeat_ts);
 
 		hl_eq_dump(hdev, &hdev->event_queue);
 
@@ -1649,6 +1630,11 @@ int hl_device_reset(struct hl_device *hdev, u32 flags)
 	from_watchdog_thread = !!(flags & HL_DRV_RESET_FROM_WD_THR);
 	reset_upon_device_release = hdev->reset_upon_device_release && from_dev_release;
 
+	if (hdev->cpld_shutdown) {
+		dev_err(hdev->dev, "Cannot reset device, cpld is shutdown! Device is NOT usable\n");
+		return -EIO;
+	}
+
 	if (!hard_reset && (hl_device_status(hdev) == HL_DEVICE_STATUS_MALFUNCTION)) {
 		dev_dbg(hdev->dev, "soft-reset isn't supported on a malfunctioning device\n");
 		return 0;
@@ -2595,6 +2581,14 @@ void hl_device_fini(struct hl_device *hdev)
 	if (rc)
 		dev_err(hdev->dev, "hw_fini failed in device fini while removing device %d\n", rc);
 
+	/* Reset the H/W (if it accessible). It will be in idle state after this returns */
+	if (!hdev->cpld_shutdown) {
+		rc = hdev->asic_funcs->hw_fini(hdev, true, false);
+		if (rc)
+			dev_err(hdev->dev,
+				"hw_fini failed in device fini while removing device %d\n", rc);
+	}
+
 	hdev->fw_loader.fw_comp_loaded = FW_TYPE_NONE;
 
 	/* Release kernel context */
@@ -2962,3 +2956,13 @@ void hl_handle_clk_change_event(struct hl_device *hdev, u16 event_type, u64 *eve
 
 	mutex_unlock(&clk_throttle->lock);
 }
+
+void hl_eq_cpld_shutdown_event_handle(struct hl_device *hdev, u16 event_id, u64 *event_mask)
+{
+	hl_handle_critical_hw_err(hdev, event_id, event_mask);
+	*event_mask |= HL_NOTIFIER_EVENT_DEVICE_UNAVAILABLE;
+
+	/* Avoid any new accesses to the H/W */
+	hdev->disabled = true;
+	hdev->cpld_shutdown = true;
+}
diff --git a/drivers/accel/habanalabs/common/habanalabs.h b/drivers/accel/habanalabs/common/habanalabs.h
index 6f27ce4fa01b..d94c2ba22a6a 100644
--- a/drivers/accel/habanalabs/common/habanalabs.h
+++ b/drivers/accel/habanalabs/common/habanalabs.h
@@ -90,7 +90,9 @@ struct hl_fpriv;
 #define HL_COMMON_USER_CQ_INTERRUPT_ID	0xFFF
 #define HL_COMMON_DEC_INTERRUPT_ID	0xFFE
 
-#define HL_STATE_DUMP_HIST_LEN		5
+#define HL_STATE_DUMP_HIST_LEN			5
+#define HL_DBGFS_CFG_ACCESS_HIST_LEN		20
+#define HL_DBGFS_CFG_ACCESS_HIST_TIMEOUT_SEC	2 /* 2s */
 
 /* Default value for device reset trigger , an invalid value */
 #define HL_RESET_TRIGGER_DEFAULT	0xFF
@@ -702,6 +704,7 @@ struct hl_hints_range {
  * @supports_advanced_cpucp_rc: true if new cpucp opcodes are supported.
  * @supports_engine_modes: true if changing engines/engine_cores modes is supported.
  * @support_dynamic_resereved_fw_size: true if we support dynamic reserved size for fw.
+ * @supports_nvme: indicates whether the asic supports NVMe P2P DMA.
  */
 struct asic_fixed_properties {
 	struct hw_queue_properties	*hw_queues_props;
@@ -822,6 +825,7 @@ struct asic_fixed_properties {
 	u8				supports_advanced_cpucp_rc;
 	u8				supports_engine_modes;
 	u8				support_dynamic_resereved_fw_size;
+	u8				supports_nvme;
 };
 
 /**
@@ -2274,6 +2278,9 @@ struct hl_vm {
 	u8			init_done;
 };
 
+#ifdef CONFIG_HL_HLDIO
+#include "hldio.h"
+#endif
 
 /*
  * DEBUG, PROFILING STRUCTURE
@@ -2344,7 +2351,6 @@ struct hl_fpriv {
 	struct mutex			ctx_lock;
 };
 
-
 /*
  * DebugFS
  */
@@ -2372,6 +2378,7 @@ struct hl_debugfs_entry {
 	struct hl_dbg_device_entry	*dev_entry;
 };
 
+
 /**
  * struct hl_dbg_device_entry - ASIC specific debugfs manager.
  * @root: root dentry.
@@ -2403,6 +2410,7 @@ struct hl_debugfs_entry {
  * @i2c_addr: generic u8 debugfs file for address value to use in i2c_data_read.
  * @i2c_reg: generic u8 debugfs file for register value to use in i2c_data_read.
  * @i2c_len: generic u8 debugfs file for length value to use in i2c_data_read.
+ * @dio_stats: Direct I/O statistics
  */
 struct hl_dbg_device_entry {
 	struct dentry			*root;
@@ -2434,6 +2442,35 @@ struct hl_dbg_device_entry {
 	u8				i2c_addr;
 	u8				i2c_reg;
 	u8				i2c_len;
+#ifdef CONFIG_HL_HLDIO
+	struct hl_dio_stats	dio_stats;
+#endif
+};
+
+/**
+ * struct hl_debugfs_cfg_access_entry - single debugfs config access object, member of
+ * hl_debugfs_cfg_access.
+ * @seconds_since_epoch: seconds since January 1, 1970, used for time comparisons.
+ * @debugfs_type: the debugfs operation requested, can be READ32, WRITE32, READ64 or WRITE64.
+ * @addr: the requested address to access.
+ * @valid: if set, this entry has valid data for dumping at interrupt time.
+ */
+struct hl_debugfs_cfg_access_entry {
+	ktime_t				seconds_since_epoch;
+	enum debugfs_access_type	debugfs_type;
+	u64				addr;
+	bool				valid;
+};
+
+/**
+ * struct hl_debugfs_cfg_access - saves debugfs config region access requests history.
+ * @cfg_access_list: list of objects describing config region access requests.
+ * @head: next valid index to add new entry to in cfg_access_list.
+ */
+struct hl_debugfs_cfg_access {
+	struct hl_debugfs_cfg_access_entry	cfg_access_list[HL_DBGFS_CFG_ACCESS_HIST_LEN];
+	u32					head;
+	spinlock_t			lock; /* protects head and entries */
 };
 
 /**
@@ -3281,6 +3318,7 @@ struct eq_heartbeat_debug_info {
  * @hl_chip_info: ASIC's sensors information.
  * @device_status_description: device status description.
  * @hl_debugfs: device's debugfs manager.
+ * @debugfs_cfg_accesses: list of last debugfs config region accesses.
  * @cb_pool: list of pre allocated CBs.
  * @cb_pool_lock: protects the CB pool.
  * @internal_cb_pool_virt_addr: internal command buffer pool virtual address.
@@ -3305,6 +3343,7 @@ struct eq_heartbeat_debug_info {
  * @captured_err_info: holds information about errors.
  * @reset_info: holds current device reset information.
  * @heartbeat_debug_info: counters used to debug heartbeat failures.
+ * @hldio: describes habanalabs direct storage interaction interface.
  * @irq_affinity_mask: mask of available CPU cores for user and decoder interrupt handling.
  * @stream_master_qid_arr: pointer to array with QIDs of master streams.
  * @fw_inner_major_ver: the major of current loaded preboot inner version.
@@ -3357,6 +3396,7 @@ struct eq_heartbeat_debug_info {
  *                    addresses.
  * @is_in_dram_scrub: true if dram scrub operation is on going.
  * @disabled: is device disabled.
+ * @cpld_shutdown: is cpld shutdown.
  * @late_init_done: is late init stage was done during initialization.
  * @hwmon_initialized: is H/W monitor sensors was initialized.
  * @reset_on_lockup: true if a reset should be done in case of stuck CS, false
@@ -3461,6 +3501,7 @@ struct hl_device {
 	struct hwmon_chip_info		*hl_chip_info;
 
 	struct hl_dbg_device_entry	hl_debugfs;
+	struct hl_debugfs_cfg_access	debugfs_cfg_accesses;
 
 	struct list_head		cb_pool;
 	spinlock_t			cb_pool_lock;
@@ -3496,7 +3537,9 @@ struct hl_device {
 	struct hl_reset_info		reset_info;
 
 	struct eq_heartbeat_debug_info	heartbeat_debug_info;
-
+#ifdef CONFIG_HL_HLDIO
+	struct hl_dio			hldio;
+#endif
 	cpumask_t			irq_affinity_mask;
 
 	u32				*stream_master_qid_arr;
@@ -3532,6 +3575,7 @@ struct hl_device {
 	u16				cpu_pci_msb_addr;
 	u8				is_in_dram_scrub;
 	u8				disabled;
+	u8				cpld_shutdown;
 	u8				late_init_done;
 	u8				hwmon_initialized;
 	u8				reset_on_lockup;
@@ -4089,6 +4133,7 @@ void hl_init_cpu_for_irq(struct hl_device *hdev);
 void hl_set_irq_affinity(struct hl_device *hdev, int irq);
 void hl_eq_heartbeat_event_handle(struct hl_device *hdev);
 void hl_handle_clk_change_event(struct hl_device *hdev, u16 event_type, u64 *event_mask);
+void hl_eq_cpld_shutdown_event_handle(struct hl_device *hdev, u16 event_id, u64 *event_mask);
 
 #ifdef CONFIG_DEBUG_FS
 
@@ -4110,6 +4155,7 @@ void hl_debugfs_add_ctx_mem_hash(struct hl_device *hdev, struct hl_ctx *ctx);
 void hl_debugfs_remove_ctx_mem_hash(struct hl_device *hdev, struct hl_ctx *ctx);
 void hl_debugfs_set_state_dump(struct hl_device *hdev, char *data,
 					unsigned long length);
+void hl_debugfs_cfg_access_history_dump(struct hl_device *hdev);
 
 #else
 
@@ -4185,6 +4231,10 @@ static inline void hl_debugfs_set_state_dump(struct hl_device *hdev,
 {
 }
 
+static inline void hl_debugfs_cfg_access_history_dump(struct hl_device *hdev)
+{
+}
+
 #endif
 
 /* Security */
diff --git a/drivers/accel/habanalabs/common/habanalabs_ioctl.c b/drivers/accel/habanalabs/common/habanalabs_ioctl.c
index dc80ca921d90..fdfdabc85e54 100644
--- a/drivers/accel/habanalabs/common/habanalabs_ioctl.c
+++ b/drivers/accel/habanalabs/common/habanalabs_ioctl.c
@@ -961,6 +961,12 @@ static int send_fw_generic_request(struct hl_device *hdev, struct hl_info_args *
 	case HL_PASSTHROUGH_VERSIONS:
 		need_input_buff = false;
 		break;
+	case  HL_GET_ERR_COUNTERS_CMD:
+		need_input_buff = true;
+		break;
+	case HL_GET_P_STATE:
+		need_input_buff = false;
+		break;
 	default:
 		return -EINVAL;
 	}
diff --git a/drivers/accel/habanalabs/common/hldio.c b/drivers/accel/habanalabs/common/hldio.c
new file mode 100644
index 000000000000..083ae5610875
--- /dev/null
+++ b/drivers/accel/habanalabs/common/hldio.c
@@ -0,0 +1,437 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/*
+ * Copyright 2024 HabanaLabs, Ltd.
+ * All Rights Reserved.
+ */
+
+#include "habanalabs.h"
+#include "hldio.h"
+#include <generated/uapi/linux/version.h>
+#include <linux/pci-p2pdma.h>
+#include <linux/blkdev.h>
+#include <linux/vmalloc.h>
+
+/*
+ * NVMe Direct I/O implementation for habanalabs driver
+ *
+ * ASSUMPTIONS
+ * ===========
+ * 1. No IOMMU (well, technically it can work with IOMMU, but it is *almost useless).
+ * 2. Only READ operations (can extend in the future).
+ * 3. No sparse files (can overcome this in the future).
+ * 4. Kernel version >= 6.9
+ * 5. Requiring page alignment is OK (I don't see a solution to this one right,
+ *    now, how do we read partial pages?)
+ * 6. Kernel compiled with CONFIG_PCI_P2PDMA. This requires a CUSTOM kernel.
+ *    Theoretically I have a slight idea on how this could be solvable, but it
+ *    is probably inacceptable for the upstream. Also may not work in the end.
+ * 7. Either make sure our cards and disks are under the same PCI bridge, or
+ *    compile a custom kernel to hack around this.
+ */
+
+#define IO_STABILIZE_TIMEOUT 10000000 /* 10 seconds in microseconds */
+
+/*
+ * This struct contains all the useful data I could milk out of the file handle
+ * provided by the user.
+ * @TODO: right now it is retrieved on each IO, but can be done once with some
+ * dedicated IOCTL, call it for example HL_REGISTER_HANDLE.
+ */
+struct hl_dio_fd {
+	/* Back pointer in case we need it in async completion */
+	struct hl_ctx *ctx;
+	/* Associated fd struct */
+	struct file *filp;
+};
+
+/*
+ * This is a single IO descriptor
+ */
+struct hl_direct_io {
+	struct hl_dio_fd f;
+	struct kiocb kio;
+	struct bio_vec *bv;
+	struct iov_iter iter;
+	u64 device_va;
+	u64 off_bytes;
+	u64 len_bytes;
+	u32 type;
+};
+
+bool hl_device_supports_nvme(struct hl_device *hdev)
+{
+	return hdev->asic_prop.supports_nvme;
+}
+
+static int hl_dio_fd_register(struct hl_ctx *ctx, int fd, struct hl_dio_fd *f)
+{
+	struct hl_device *hdev = ctx->hdev;
+	struct block_device *bd;
+	struct super_block *sb;
+	struct inode *inode;
+	struct gendisk *gd;
+	struct device *disk_dev;
+	int rc;
+
+	f->filp = fget(fd);
+	if (!f->filp) {
+		rc = -ENOENT;
+		goto out;
+	}
+
+	if (!(f->filp->f_flags & O_DIRECT)) {
+		dev_err(hdev->dev, "file is not in the direct mode\n");
+		rc = -EINVAL;
+		goto fput;
+	}
+
+	if (!f->filp->f_op->read_iter) {
+		dev_err(hdev->dev, "read iter is not supported, need to fall back to legacy\n");
+		rc = -EINVAL;
+		goto fput;
+	}
+
+	inode = file_inode(f->filp);
+	sb = inode->i_sb;
+	bd = sb->s_bdev;
+	gd = bd->bd_disk;
+
+	if (inode->i_blocks << sb->s_blocksize_bits < i_size_read(inode)) {
+		dev_err(hdev->dev, "sparse files are not currently supported\n");
+		rc = -EINVAL;
+		goto fput;
+	}
+
+	if (!bd || !gd) {
+		dev_err(hdev->dev, "invalid block device\n");
+		rc = -ENODEV;
+		goto fput;
+	}
+	/* Get the underlying device from the block device */
+	disk_dev = disk_to_dev(gd);
+	if (!dma_pci_p2pdma_supported(disk_dev)) {
+		dev_err(hdev->dev, "device does not support PCI P2P DMA\n");
+		rc = -EOPNOTSUPP;
+		goto fput;
+	}
+
+	/*
+	 * @TODO: Maybe we need additional checks here
+	 */
+
+	f->ctx = ctx;
+	rc = 0;
+
+	goto out;
+fput:
+	fput(f->filp);
+out:
+	return rc;
+}
+
+static void hl_dio_fd_unregister(struct hl_dio_fd *f)
+{
+	fput(f->filp);
+}
+
+static long hl_dio_count_io(struct hl_device *hdev)
+{
+	s64 sum = 0;
+	int i;
+
+	for_each_possible_cpu(i)
+		sum += per_cpu(*hdev->hldio.inflight_ios, i);
+
+	return sum;
+}
+
+static bool hl_dio_get_iopath(struct hl_ctx *ctx)
+{
+	struct hl_device *hdev = ctx->hdev;
+
+	if (hdev->hldio.io_enabled) {
+		this_cpu_inc(*hdev->hldio.inflight_ios);
+
+		/* Avoid race conditions */
+		if (!hdev->hldio.io_enabled) {
+			this_cpu_dec(*hdev->hldio.inflight_ios);
+			return false;
+		}
+
+		hl_ctx_get(ctx);
+
+		return true;
+	}
+
+	return false;
+}
+
+static void hl_dio_put_iopath(struct hl_ctx *ctx)
+{
+	struct hl_device *hdev = ctx->hdev;
+
+	hl_ctx_put(ctx);
+	this_cpu_dec(*hdev->hldio.inflight_ios);
+}
+
+static void hl_dio_set_io_enabled(struct hl_device *hdev, bool enabled)
+{
+	hdev->hldio.io_enabled = enabled;
+}
+
+static bool hl_dio_validate_io(struct hl_device *hdev, struct hl_direct_io *io)
+{
+	if ((u64)io->device_va & ~PAGE_MASK) {
+		dev_dbg(hdev->dev, "device address must be 4K aligned\n");
+		return false;
+	}
+
+	if (io->len_bytes & ~PAGE_MASK) {
+		dev_dbg(hdev->dev, "IO length must be 4K aligned\n");
+		return false;
+	}
+
+	if (io->off_bytes & ~PAGE_MASK) {
+		dev_dbg(hdev->dev, "IO offset must be 4K aligned\n");
+		return false;
+	}
+
+	return true;
+}
+
+static struct page *hl_dio_va2page(struct hl_device *hdev, struct hl_ctx *ctx, u64 device_va)
+{
+	struct hl_dio *hldio = &hdev->hldio;
+	u64 device_pa;
+	int rc, i;
+
+	rc = hl_mmu_va_to_pa(ctx, device_va, &device_pa);
+	if (rc) {
+		dev_err(hdev->dev, "device virtual address translation error: %#llx (%d)",
+				device_va, rc);
+		return NULL;
+	}
+
+	for (i = 0 ; i < hldio->np2prs ; ++i) {
+		if (device_pa >= hldio->p2prs[i].device_pa &&
+		    device_pa < hldio->p2prs[i].device_pa + hldio->p2prs[i].size)
+			return hldio->p2prs[i].p2ppages[(device_pa - hldio->p2prs[i].device_pa) >>
+				PAGE_SHIFT];
+	}
+
+	return NULL;
+}
+
+static ssize_t hl_direct_io(struct hl_device *hdev, struct hl_direct_io *io)
+{
+	u64 npages, device_va;
+	ssize_t rc;
+	int i;
+
+	if (!hl_dio_validate_io(hdev, io))
+		return -EINVAL;
+
+	if (!hl_dio_get_iopath(io->f.ctx)) {
+		dev_info(hdev->dev, "can't schedule a new IO, IO is disabled\n");
+		return -ESHUTDOWN;
+	}
+
+	init_sync_kiocb(&io->kio, io->f.filp);
+	io->kio.ki_pos = io->off_bytes;
+
+	npages = (io->len_bytes >> PAGE_SHIFT);
+
+	/* @TODO: this can be implemented smarter, vmalloc in iopath is not
+	 * ideal. Maybe some variation of genpool. Number of pages may differ
+	 * greatly, so maybe even use pools of different sizes and chose the
+	 * closest one.
+	 */
+	io->bv = vzalloc(npages * sizeof(struct bio_vec));
+	if (!io->bv)
+		return -ENOMEM;
+
+	for (i = 0, device_va = io->device_va; i < npages ; ++i, device_va += PAGE_SIZE) {
+		io->bv[i].bv_page = hl_dio_va2page(hdev, io->f.ctx, device_va);
+		if (!io->bv[i].bv_page) {
+			dev_err(hdev->dev, "error getting page struct for device va %#llx",
+					device_va);
+			rc = -EFAULT;
+			goto cleanup;
+		}
+		io->bv[i].bv_offset = 0;
+		io->bv[i].bv_len = PAGE_SIZE;
+	}
+
+	iov_iter_bvec(&io->iter, io->type, io->bv, 1, io->len_bytes);
+	if (io->f.filp->f_op && io->f.filp->f_op->read_iter)
+		rc = io->f.filp->f_op->read_iter(&io->kio, &io->iter);
+	else
+		rc = -EINVAL;
+
+cleanup:
+	vfree(io->bv);
+	hl_dio_put_iopath(io->f.ctx);
+
+	dev_dbg(hdev->dev, "IO ended with %ld\n", rc);
+
+	return rc;
+}
+
+/*
+ * @TODO: This function can be used as a callback for io completion under
+ * kio->ki_complete in order to implement async IO.
+ * Note that on more recent kernels there is no ret2.
+ */
+__maybe_unused static void hl_direct_io_complete(struct kiocb *kio, long ret, long ret2)
+{
+	struct hl_direct_io *io = container_of(kio, struct hl_direct_io, kio);
+
+	dev_dbg(io->f.ctx->hdev->dev, "IO completed with %ld\n", ret);
+
+	/* Do something to copy result to user / notify completion */
+
+	hl_dio_put_iopath(io->f.ctx);
+
+	hl_dio_fd_unregister(&io->f);
+}
+
+/*
+ * DMA disk to ASIC, wait for results. Must be invoked from the user context
+ */
+int hl_dio_ssd2hl(struct hl_device *hdev, struct hl_ctx *ctx, int fd,
+		  u64 device_va, off_t off_bytes, size_t len_bytes,
+		  size_t *len_read)
+{
+	struct hl_direct_io *io;
+	ssize_t rc;
+
+	dev_dbg(hdev->dev, "SSD2HL fd=%d va=%#llx len=%#lx\n", fd, device_va, len_bytes);
+
+	io = kzalloc(sizeof(*io), GFP_KERNEL);
+	if (!io) {
+		rc = -ENOMEM;
+		goto out;
+	}
+
+	*io = (struct hl_direct_io){
+		.device_va = device_va,
+		.len_bytes = len_bytes,
+		.off_bytes = off_bytes,
+		.type = READ,
+	};
+
+	rc = hl_dio_fd_register(ctx, fd, &io->f);
+	if (rc)
+		goto kfree_io;
+
+	rc = hl_direct_io(hdev, io);
+	if (rc >= 0) {
+		*len_read = rc;
+		rc = 0;
+	}
+
+	/* This shall be called only in the case of a sync IO */
+	hl_dio_fd_unregister(&io->f);
+kfree_io:
+	kfree(io);
+out:
+	return rc;
+}
+
+static void hl_p2p_region_fini(struct hl_device *hdev, struct hl_p2p_region *p2pr)
+{
+	if (p2pr->p2ppages) {
+		vfree(p2pr->p2ppages);
+		p2pr->p2ppages = NULL;
+	}
+
+	if (p2pr->p2pmem) {
+		dev_dbg(hdev->dev, "freeing P2P mem from %p, size=%#llx\n",
+				p2pr->p2pmem, p2pr->size);
+		pci_free_p2pmem(hdev->pdev, p2pr->p2pmem, p2pr->size);
+		p2pr->p2pmem = NULL;
+	}
+}
+
+void hl_p2p_region_fini_all(struct hl_device *hdev)
+{
+	int i;
+
+	for (i = 0 ; i < hdev->hldio.np2prs ; ++i)
+		hl_p2p_region_fini(hdev, &hdev->hldio.p2prs[i]);
+
+	kvfree(hdev->hldio.p2prs);
+	hdev->hldio.p2prs = NULL;
+	hdev->hldio.np2prs = 0;
+}
+
+int hl_p2p_region_init(struct hl_device *hdev, struct hl_p2p_region *p2pr)
+{
+	void *addr;
+	int rc, i;
+
+	/* Start by publishing our p2p memory */
+	rc = pci_p2pdma_add_resource(hdev->pdev, p2pr->bar, p2pr->size, p2pr->bar_offset);
+	if (rc) {
+		dev_err(hdev->dev, "error adding p2p resource: %d\n", rc);
+		goto err;
+	}
+
+	/* Alloc all p2p mem */
+	p2pr->p2pmem = pci_alloc_p2pmem(hdev->pdev, p2pr->size);
+	if (!p2pr->p2pmem) {
+		dev_err(hdev->dev, "error allocating p2p memory\n");
+		rc = -ENOMEM;
+		goto err;
+	}
+
+	p2pr->p2ppages = vmalloc((p2pr->size >> PAGE_SHIFT) * sizeof(struct page *));
+	if (!p2pr->p2ppages) {
+		rc = -ENOMEM;
+		goto err;
+	}
+
+	for (i = 0, addr = p2pr->p2pmem ; i < (p2pr->size >> PAGE_SHIFT) ; ++i, addr += PAGE_SIZE) {
+		p2pr->p2ppages[i] = virt_to_page(addr);
+		if (!p2pr->p2ppages[i]) {
+			rc = -EFAULT;
+			goto err;
+		}
+	}
+
+	return 0;
+err:
+	hl_p2p_region_fini(hdev, p2pr);
+	return rc;
+}
+
+int hl_dio_start(struct hl_device *hdev)
+{
+	dev_dbg(hdev->dev, "initializing HLDIO\n");
+
+	/* Initialize the IO counter and enable IO */
+	hdev->hldio.inflight_ios = alloc_percpu(s64);
+	if (!hdev->hldio.inflight_ios)
+		return -ENOMEM;
+
+	hl_dio_set_io_enabled(hdev, true);
+
+	return 0;
+}
+
+void hl_dio_stop(struct hl_device *hdev)
+{
+	dev_dbg(hdev->dev, "deinitializing HLDIO\n");
+
+	if (hdev->hldio.io_enabled) {
+		/* Wait for all the IO to finish */
+		hl_dio_set_io_enabled(hdev, false);
+		hl_poll_timeout_condition(hdev, !hl_dio_count_io(hdev), 1000, IO_STABILIZE_TIMEOUT);
+	}
+
+	if (hdev->hldio.inflight_ios) {
+		free_percpu(hdev->hldio.inflight_ios);
+		hdev->hldio.inflight_ios = NULL;
+	}
+}
diff --git a/drivers/accel/habanalabs/common/hldio.h b/drivers/accel/habanalabs/common/hldio.h
new file mode 100644
index 000000000000..2874388f2851
--- /dev/null
+++ b/drivers/accel/habanalabs/common/hldio.h
@@ -0,0 +1,146 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * hldio.h - NVMe Direct I/O (HLDIO) infrastructure for Habana Labs Driver
+ *
+ * This feature requires specific hardware setup and must not be built
+ * under COMPILE_TEST.
+ */
+
+#ifndef __HL_HLDIO_H__
+#define __HL_HLDIO_H__
+
+#include <linux/types.h>
+#include <linux/fs.h>
+#include <linux/seq_file.h>
+#include <linux/ktime.h>     /* ktime functions */
+#include <linux/delay.h>     /* usleep_range */
+#include <linux/kernel.h>    /* might_sleep_if */
+#include <linux/errno.h>     /* error codes */
+
+/* Forward declarations */
+struct hl_device;
+struct file;
+
+/* Enable only if Kconfig selected */
+#ifdef CONFIG_HL_HLDIO
+/**
+ * struct hl_p2p_region - describes a single P2P memory region
+ * @p2ppages: array of page structs for the P2P memory
+ * @p2pmem: virtual address of the P2P memory region
+ * @device_pa: physical address on the device
+ * @bar_offset: offset within the BAR
+ * @size: size of the region in bytes
+ * @bar: BAR number containing this region
+ */
+struct hl_p2p_region {
+	struct page **p2ppages;
+	void *p2pmem;
+	u64 device_pa;
+	u64 bar_offset;
+	u64 size;
+	int bar;
+};
+
+/**
+ * struct hl_dio_stats - Direct I/O statistics
+ * @total_ops: total number of operations attempted
+ * @successful_ops: number of successful operations
+ * @failed_ops: number of failed operations
+ * @bytes_transferred: total bytes successfully transferred
+ * @last_len_read: length of the last read operation
+ */
+struct hl_dio_stats {
+	u64 total_ops;
+	u64 successful_ops;
+	u64 failed_ops;
+	u64 bytes_transferred;
+	size_t last_len_read;
+};
+
+/**
+ * struct hl_dio - describes habanalabs direct storage interaction interface
+ * @p2prs: array of p2p regions
+ * @inflight_ios: percpu counter for inflight ios
+ * @np2prs: number of elements in p2prs
+ * @io_enabled: 1 if io is enabled 0 otherwise
+ */
+struct hl_dio {
+	struct hl_p2p_region *p2prs;
+	s64 __percpu *inflight_ios;
+	u8 np2prs;
+	u8 io_enabled;
+};
+
+int hl_dio_ssd2hl(struct hl_device *hdev, struct hl_ctx *ctx, int fd,
+		  u64 device_va, off_t off_bytes, size_t len_bytes,
+		  size_t *len_read);
+void hl_p2p_region_fini_all(struct hl_device *hdev);
+int hl_p2p_region_init(struct hl_device *hdev, struct hl_p2p_region *p2pr);
+int hl_dio_start(struct hl_device *hdev);
+void hl_dio_stop(struct hl_device *hdev);
+
+/* Init/teardown */
+int hl_hldio_init(struct hl_device *hdev);
+void hl_hldio_fini(struct hl_device *hdev);
+
+/* File operations */
+long hl_hldio_ioctl(struct file *filep, unsigned int cmd, unsigned long arg);
+
+/* DebugFS hooks */
+#ifdef CONFIG_DEBUG_FS
+void hl_hldio_debugfs_init(struct hl_device *hdev);
+void hl_hldio_debugfs_fini(struct hl_device *hdev);
+#else
+static inline void hl_hldio_debugfs_init(struct hl_device *hdev) { }
+static inline void hl_hldio_debugfs_fini(struct hl_device *hdev) { }
+#endif
+
+#else /* !CONFIG_HL_HLDIO */
+
+struct hl_p2p_region;
+/* Stubs when HLDIO is disabled */
+static inline int hl_dio_ssd2hl(struct hl_device *hdev, struct hl_ctx *ctx, int fd,
+		  u64 device_va, off_t off_bytes, size_t len_bytes,
+		  size_t *len_read)
+{ return -EOPNOTSUPP; }
+static inline void hl_p2p_region_fini_all(struct hl_device *hdev) {}
+static inline int hl_p2p_region_init(struct hl_device *hdev, struct hl_p2p_region *p2pr)
+{ return -EOPNOTSUPP; }
+static inline int hl_dio_start(struct hl_device *hdev) { return -EOPNOTSUPP; }
+static inline void hl_dio_stop(struct hl_device *hdev) {}
+
+static inline int hl_hldio_init(struct hl_device *hdev) { return 0; }
+static inline void hl_hldio_fini(struct hl_device *hdev) { }
+static inline long hl_hldio_ioctl(struct file *f, unsigned int c,
+				  unsigned long a)
+{ return -ENOTTY; }
+static inline void hl_hldio_debugfs_init(struct hl_device *hdev) { }
+static inline void hl_hldio_debugfs_fini(struct hl_device *hdev) { }
+
+#endif /* CONFIG_HL_HLDIO */
+
+/* Simplified polling macro for HLDIO (no simulator support) */
+#define hl_poll_timeout_condition(hdev, cond, sleep_us, timeout_us) \
+({ \
+	ktime_t __timeout = ktime_add_us(ktime_get(), timeout_us); \
+	might_sleep_if(sleep_us); \
+	(void)(hdev); /* keep signature consistent, hdev unused */ \
+	for (;;) { \
+		mb(); /* ensure ordering of memory operations */ \
+		if (cond) \
+			break; \
+		if (timeout_us && ktime_compare(ktime_get(), __timeout) > 0) \
+			break; \
+		if (sleep_us) \
+			usleep_range((sleep_us >> 2) + 1, sleep_us); \
+	} \
+	(cond) ? 0 : -ETIMEDOUT; \
+})
+
+#ifdef CONFIG_HL_HLDIO
+bool hl_device_supports_nvme(struct hl_device *hdev);
+#else
+static inline bool hl_device_supports_nvme(struct hl_device *hdev) { return false; }
+#endif
+
+#endif /* __HL_HLDIO_H__ */
diff --git a/drivers/accel/habanalabs/common/memory.c b/drivers/accel/habanalabs/common/memory.c
index 601fdbe70179..633db4bff46f 100644
--- a/drivers/accel/habanalabs/common/memory.c
+++ b/drivers/accel/habanalabs/common/memory.c
@@ -1829,9 +1829,6 @@ static void hl_release_dmabuf(struct dma_buf *dmabuf)
 	struct hl_dmabuf_priv *hl_dmabuf = dmabuf->priv;
 	struct hl_ctx *ctx;
 
-	if (!hl_dmabuf)
-		return;
-
 	ctx = hl_dmabuf->ctx;
 
 	if (hl_dmabuf->memhash_hnode)
@@ -1840,7 +1837,12 @@ static void hl_release_dmabuf(struct dma_buf *dmabuf)
 	atomic_dec(&ctx->hdev->dmabuf_export_cnt);
 	hl_ctx_put(ctx);
 
-	/* Paired with get_file() in export_dmabuf() */
+	/*
+	 * Paired with get_file() in export_dmabuf().
+	 * 'ctx' can be still used here to get the file pointer, even after hl_ctx_put() was called,
+	 * because releasing the compute device file involves another reference decrement, and it
+	 * would be possible only after calling fput().
+	 */
 	fput(ctx->hpriv->file_priv->filp);
 
 	kfree(hl_dmabuf);
@@ -1859,7 +1861,12 @@ static int export_dmabuf(struct hl_ctx *ctx,
 {
 	DEFINE_DMA_BUF_EXPORT_INFO(exp_info);
 	struct hl_device *hdev = ctx->hdev;
-	int rc, fd;
+	CLASS(get_unused_fd, fd)(flags);
+
+	if (fd < 0) {
+		dev_err(hdev->dev, "failed to get a file descriptor for a dma-buf, %d\n", fd);
+		return fd;
+	}
 
 	exp_info.ops = &habanalabs_dmabuf_ops;
 	exp_info.size = total_size;
@@ -1872,13 +1879,6 @@ static int export_dmabuf(struct hl_ctx *ctx,
 		return PTR_ERR(hl_dmabuf->dmabuf);
 	}
 
-	fd = dma_buf_fd(hl_dmabuf->dmabuf, flags);
-	if (fd < 0) {
-		dev_err(hdev->dev, "failed to get a file descriptor for a dma-buf, %d\n", fd);
-		rc = fd;
-		goto err_dma_buf_put;
-	}
-
 	hl_dmabuf->ctx = ctx;
 	hl_ctx_get(hl_dmabuf->ctx);
 	atomic_inc(&ctx->hdev->dmabuf_export_cnt);
@@ -1890,13 +1890,9 @@ static int export_dmabuf(struct hl_ctx *ctx,
 	get_file(ctx->hpriv->file_priv->filp);
 
 	*dmabuf_fd = fd;
+	fd_install(take_fd(fd), hl_dmabuf->dmabuf->file);
 
 	return 0;
-
-err_dma_buf_put:
-	hl_dmabuf->dmabuf->priv = NULL;
-	dma_buf_put(hl_dmabuf->dmabuf);
-	return rc;
 }
 
 static int validate_export_params_common(struct hl_device *hdev, u64 addr, u64 size, u64 offset)
@@ -2341,7 +2337,7 @@ static int get_user_memory(struct hl_device *hdev, u64 addr, u64 size,
 		if (rc < 0)
 			goto destroy_pages;
 		npages = rc;
-		rc = -EFAULT;
+		rc = -ENOMEM;
 		goto put_pages;
 	}
 	userptr->npages = npages;
diff --git a/drivers/accel/habanalabs/common/memory_mgr.c b/drivers/accel/habanalabs/common/memory_mgr.c
index 99cd83139d46..4401beb99e42 100644
--- a/drivers/accel/habanalabs/common/memory_mgr.c
+++ b/drivers/accel/habanalabs/common/memory_mgr.c
@@ -259,13 +259,8 @@ int hl_mem_mgr_mmap(struct hl_mem_mgr *mmg, struct vm_area_struct *vma,
 		goto put_mem;
 	}
 
-#ifdef _HAS_TYPE_ARG_IN_ACCESS_OK
-	if (!access_ok(VERIFY_WRITE, (void __user *)(uintptr_t)vma->vm_start,
-		       user_mem_size)) {
-#else
 	if (!access_ok((void __user *)(uintptr_t)vma->vm_start,
 		       user_mem_size)) {
-#endif
 		dev_err(mmg->dev, "%s: User pointer is invalid - 0x%lx\n",
 			buf->behavior->topic, vma->vm_start);
 
diff --git a/drivers/accel/habanalabs/common/sysfs.c b/drivers/accel/habanalabs/common/sysfs.c
index 9d58efa2ff38..8f55ba3b4e73 100644
--- a/drivers/accel/habanalabs/common/sysfs.c
+++ b/drivers/accel/habanalabs/common/sysfs.c
@@ -96,14 +96,21 @@ static ssize_t vrm_ver_show(struct device *dev, struct device_attribute *attr, c
 	infineon_second_stage_third_instance =
 					(infineon_second_stage_version >> 16) & mask;
 
-	if (cpucp_info->infineon_second_stage_version)
+	if (cpucp_info->infineon_version && cpucp_info->infineon_second_stage_version)
 		return sprintf(buf, "%#04x %#04x:%#04x:%#04x\n",
 				le32_to_cpu(cpucp_info->infineon_version),
 				infineon_second_stage_first_instance,
 				infineon_second_stage_second_instance,
 				infineon_second_stage_third_instance);
-	else
+	else if (cpucp_info->infineon_second_stage_version)
+		return sprintf(buf, "%#04x:%#04x:%#04x\n",
+				infineon_second_stage_first_instance,
+				infineon_second_stage_second_instance,
+				infineon_second_stage_third_instance);
+	else if (cpucp_info->infineon_version)
 		return sprintf(buf, "%#04x\n", le32_to_cpu(cpucp_info->infineon_version));
+
+	return 0;
 }
 
 static DEVICE_ATTR_RO(vrm_ver);
@@ -446,7 +453,7 @@ static DEVICE_ATTR_RO(parent_device);
 static const struct bin_attribute bin_attr_eeprom = {
 	.attr = {.name = "eeprom", .mode = (0444)},
 	.size = PAGE_SIZE,
-	.read_new = eeprom_read_handler
+	.read = eeprom_read_handler
 };
 
 static struct attribute *hl_dev_attrs[] = {
@@ -479,7 +486,7 @@ static const struct bin_attribute *const hl_dev_bin_attrs[] = {
 
 static struct attribute_group hl_dev_attr_group = {
 	.attrs = hl_dev_attrs,
-	.bin_attrs_new = hl_dev_bin_attrs,
+	.bin_attrs = hl_dev_bin_attrs,
 };
 
 static struct attribute_group hl_dev_clks_attr_group;
diff --git a/drivers/accel/habanalabs/gaudi/gaudi.c b/drivers/accel/habanalabs/gaudi/gaudi.c
index fa893a9b826e..34771d75da9d 100644
--- a/drivers/accel/habanalabs/gaudi/gaudi.c
+++ b/drivers/accel/habanalabs/gaudi/gaudi.c
@@ -4168,10 +4168,29 @@ static int gaudi_mmap(struct hl_device *hdev, struct vm_area_struct *vma,
 	vm_flags_set(vma, VM_IO | VM_PFNMAP | VM_DONTEXPAND | VM_DONTDUMP |
 			VM_DONTCOPY | VM_NORESERVE);
 
+#ifdef _HAS_DMA_MMAP_COHERENT
+	/*
+	 * If dma_alloc_coherent() returns a vmalloc address, set VM_MIXEDMAP
+	 * so vm_insert_page() can handle it safely. Without this, the kernel
+	 * may BUG_ON due to VM_PFNMAP.
+	 */
+	if (is_vmalloc_addr(cpu_addr))
+		vm_flags_set(vma, VM_MIXEDMAP);
+
 	rc = dma_mmap_coherent(hdev->dev, vma, cpu_addr,
 				(dma_addr - HOST_PHYS_BASE), size);
 	if (rc)
 		dev_err(hdev->dev, "dma_mmap_coherent error %d", rc);
+#else
+
+	rc = remap_pfn_range(vma, vma->vm_start,
+				virt_to_phys(cpu_addr) >> PAGE_SHIFT,
+				size, vma->vm_page_prot);
+	if (rc)
+		dev_err(hdev->dev, "remap_pfn_range error %d", rc);
+
+ #endif
+
 
 	return rc;
 }
diff --git a/drivers/accel/habanalabs/gaudi2/gaudi2.c b/drivers/accel/habanalabs/gaudi2/gaudi2.c
index a38b88baadf2..b8c0689dba64 100644
--- a/drivers/accel/habanalabs/gaudi2/gaudi2.c
+++ b/drivers/accel/habanalabs/gaudi2/gaudi2.c
@@ -728,6 +728,354 @@ static const int gaudi2_dma_core_async_event_id[] = {
 	[DMA_CORE_ID_KDMA] = GAUDI2_EVENT_KDMA0_CORE,
 };
 
+const char *gaudi2_engine_id_str[] = {
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_EDMA_0),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_EDMA_1),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_MME),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_TPC_0),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_TPC_1),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_TPC_2),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_TPC_3),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_TPC_4),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_TPC_5),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_DEC_0),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_DEC_1),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_EDMA_0),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_EDMA_1),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_MME),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_TPC_0),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_TPC_1),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_TPC_2),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_TPC_3),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_TPC_4),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_TPC_5),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_DEC_0),
+	__stringify(GAUDI2_DCORE1_ENGINE_ID_DEC_1),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_EDMA_0),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_EDMA_1),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_MME),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_TPC_0),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_TPC_1),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_TPC_2),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_TPC_3),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_TPC_4),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_TPC_5),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_DEC_0),
+	__stringify(GAUDI2_DCORE2_ENGINE_ID_DEC_1),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_EDMA_0),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_EDMA_1),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_MME),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_TPC_0),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_TPC_1),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_TPC_2),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_TPC_3),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_TPC_4),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_TPC_5),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_DEC_0),
+	__stringify(GAUDI2_DCORE3_ENGINE_ID_DEC_1),
+	__stringify(GAUDI2_DCORE0_ENGINE_ID_TPC_6),
+	__stringify(GAUDI2_ENGINE_ID_PDMA_0),
+	__stringify(GAUDI2_ENGINE_ID_PDMA_1),
+	__stringify(GAUDI2_ENGINE_ID_ROT_0),
+	__stringify(GAUDI2_ENGINE_ID_ROT_1),
+	__stringify(GAUDI2_PCIE_ENGINE_ID_DEC_0),
+	__stringify(GAUDI2_PCIE_ENGINE_ID_DEC_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC0_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC0_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC1_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC1_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC2_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC2_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC3_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC3_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC4_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC4_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC5_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC5_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC6_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC6_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC7_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC7_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC8_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC8_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC9_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC9_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC10_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC10_1),
+	__stringify(GAUDI2_ENGINE_ID_NIC11_0),
+	__stringify(GAUDI2_ENGINE_ID_NIC11_1),
+	__stringify(GAUDI2_ENGINE_ID_PCIE),
+	__stringify(GAUDI2_ENGINE_ID_PSOC),
+	__stringify(GAUDI2_ENGINE_ID_ARC_FARM),
+	__stringify(GAUDI2_ENGINE_ID_KDMA),
+	__stringify(GAUDI2_ENGINE_ID_SIZE),
+};
+
+const char *gaudi2_queue_id_str[] = {
+	__stringify(GAUDI2_QUEUE_ID_PDMA_0_0),
+	__stringify(GAUDI2_QUEUE_ID_PDMA_0_1),
+	__stringify(GAUDI2_QUEUE_ID_PDMA_0_2),
+	__stringify(GAUDI2_QUEUE_ID_PDMA_0_3),
+	__stringify(GAUDI2_QUEUE_ID_PDMA_1_0),
+	__stringify(GAUDI2_QUEUE_ID_PDMA_1_1),
+	__stringify(GAUDI2_QUEUE_ID_PDMA_1_2),
+	__stringify(GAUDI2_QUEUE_ID_PDMA_1_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_EDMA_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_EDMA_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_EDMA_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_EDMA_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_EDMA_1_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_EDMA_1_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_EDMA_1_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_EDMA_1_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_MME_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_MME_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_MME_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_MME_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_1_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_1_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_1_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_1_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_2_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_2_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_2_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_2_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_3_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_3_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_3_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_3_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_4_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_4_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_4_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_4_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_5_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_5_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_5_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_5_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_6_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_6_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_6_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE0_TPC_6_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_EDMA_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_EDMA_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_EDMA_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_EDMA_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_EDMA_1_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_EDMA_1_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_EDMA_1_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_EDMA_1_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_MME_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_MME_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_MME_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_MME_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_1_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_1_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_1_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_1_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_2_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_2_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_2_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_2_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_3_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_3_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_3_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_3_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_4_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_4_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_4_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_4_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_5_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_5_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_5_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE1_TPC_5_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_EDMA_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_EDMA_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_EDMA_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_EDMA_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_EDMA_1_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_EDMA_1_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_EDMA_1_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_EDMA_1_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_MME_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_MME_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_MME_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_MME_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_1_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_1_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_1_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_1_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_2_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_2_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_2_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_2_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_3_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_3_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_3_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_3_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_4_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_4_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_4_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_4_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_5_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_5_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_5_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE2_TPC_5_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_EDMA_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_EDMA_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_EDMA_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_EDMA_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_EDMA_1_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_EDMA_1_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_EDMA_1_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_EDMA_1_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_MME_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_MME_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_MME_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_MME_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_0_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_0_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_0_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_0_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_1_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_1_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_1_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_1_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_2_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_2_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_2_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_2_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_3_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_3_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_3_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_3_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_4_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_4_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_4_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_4_3),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_5_0),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_5_1),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_5_2),
+	__stringify(GAUDI2_QUEUE_ID_DCORE3_TPC_5_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_0_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_0_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_0_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_0_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_1_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_1_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_1_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_1_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_2_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_2_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_2_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_2_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_3_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_3_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_3_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_3_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_4_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_4_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_4_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_4_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_5_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_5_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_5_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_5_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_6_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_6_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_6_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_6_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_7_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_7_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_7_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_7_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_8_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_8_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_8_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_8_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_9_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_9_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_9_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_9_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_10_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_10_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_10_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_10_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_11_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_11_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_11_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_11_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_12_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_12_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_12_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_12_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_13_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_13_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_13_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_13_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_14_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_14_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_14_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_14_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_15_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_15_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_15_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_15_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_16_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_16_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_16_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_16_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_17_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_17_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_17_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_17_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_18_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_18_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_18_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_18_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_19_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_19_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_19_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_19_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_20_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_20_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_20_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_20_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_21_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_21_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_21_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_21_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_22_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_22_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_22_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_22_3),
+	__stringify(GAUDI2_QUEUE_ID_NIC_23_0),
+	__stringify(GAUDI2_QUEUE_ID_NIC_23_1),
+	__stringify(GAUDI2_QUEUE_ID_NIC_23_2),
+	__stringify(GAUDI2_QUEUE_ID_NIC_23_3),
+	__stringify(GAUDI2_QUEUE_ID_ROT_0_0),
+	__stringify(GAUDI2_QUEUE_ID_ROT_0_1),
+	__stringify(GAUDI2_QUEUE_ID_ROT_0_2),
+	__stringify(GAUDI2_QUEUE_ID_ROT_0_3),
+	__stringify(GAUDI2_QUEUE_ID_ROT_1_0),
+	__stringify(GAUDI2_QUEUE_ID_ROT_1_1),
+	__stringify(GAUDI2_QUEUE_ID_ROT_1_2),
+	__stringify(GAUDI2_QUEUE_ID_ROT_1_3),
+	__stringify(GAUDI2_QUEUE_ID_CPU_PQ),
+	__stringify(GAUDI2_QUEUE_ID_SIZE),
+};
+
 static const char * const gaudi2_qm_sei_error_cause[GAUDI2_NUM_OF_QM_SEI_ERR_CAUSE] = {
 	"qman sei intr",
 	"arc sei intr"
@@ -3150,7 +3498,6 @@ static int gaudi2_early_init(struct hl_device *hdev)
 	rc = hl_fw_read_preboot_status(hdev);
 	if (rc) {
 		if (hdev->reset_on_preboot_fail)
-			/* we are already on failure flow, so don't check if hw_fini fails. */
 			hdev->asic_funcs->hw_fini(hdev, true, false);
 		goto pci_fini;
 	}
@@ -3162,6 +3509,13 @@ static int gaudi2_early_init(struct hl_device *hdev)
 			dev_err(hdev->dev, "failed to reset HW in dirty state (%d)\n", rc);
 			goto pci_fini;
 		}
+
+		rc = hl_fw_read_preboot_status(hdev);
+		if (rc) {
+			if (hdev->reset_on_preboot_fail)
+				hdev->asic_funcs->hw_fini(hdev, true, false);
+			goto pci_fini;
+		}
 	}
 
 	return 0;
@@ -4836,7 +5190,7 @@ static void gaudi2_halt_engines(struct hl_device *hdev, bool hard_reset, bool fw
 	else
 		wait_timeout_ms = GAUDI2_RESET_WAIT_MSEC;
 
-	if (fw_reset)
+	if (fw_reset || hdev->cpld_shutdown)
 		goto skip_engines;
 
 	gaudi2_stop_dma_qmans(hdev);
@@ -6484,6 +6838,13 @@ static int gaudi2_mmap(struct hl_device *hdev, struct vm_area_struct *vma,
 			VM_DONTCOPY | VM_NORESERVE);
 
 #ifdef _HAS_DMA_MMAP_COHERENT
+	/*
+	 * If dma_alloc_coherent() returns a vmalloc address, set VM_MIXEDMAP
+	 * so vm_insert_page() can handle it safely. Without this, the kernel
+	 * may BUG_ON due to VM_PFNMAP.
+	 */
+	if (is_vmalloc_addr(cpu_addr))
+		vm_flags_set(vma, VM_MIXEDMAP);
 
 	rc = dma_mmap_coherent(hdev->dev, vma, cpu_addr, dma_addr, size);
 	if (rc)
@@ -6774,7 +7135,8 @@ static int gaudi2_validate_cb_address(struct hl_device *hdev, struct hl_cs_parse
 	struct gaudi2_device *gaudi2 = hdev->asic_specific;
 
 	if (!gaudi2_is_queue_enabled(hdev, parser->hw_queue_id)) {
-		dev_err(hdev->dev, "h/w queue %d is disabled\n", parser->hw_queue_id);
+		dev_err(hdev->dev, "h/w queue %s is disabled\n",
+		GAUDI2_QUEUE_ID_TO_STR(parser->hw_queue_id));
 		return -EINVAL;
 	}
 
@@ -7026,7 +7388,8 @@ static int gaudi2_test_queue_send_msg_short(struct hl_device *hdev, u32 hw_queue
 	rc = hl_hw_queue_send_cb_no_cmpl(hdev, hw_queue_id, pkt_size, msg_info->dma_addr);
 	if (rc)
 		dev_err(hdev->dev,
-			"Failed to send msg_short packet to H/W queue %d\n", hw_queue_id);
+			"Failed to send msg_short packet to H/W queue %s\n",
+			GAUDI2_QUEUE_ID_TO_STR(hw_queue_id));
 
 	return rc;
 }
@@ -7052,8 +7415,8 @@ static int gaudi2_test_queue_wait_completion(struct hl_device *hdev, u32 hw_queu
 			timeout_usec);
 
 	if (rc == -ETIMEDOUT) {
-		dev_err(hdev->dev, "H/W queue %d test failed (SOB_OBJ_0 == 0x%x)\n",
-			hw_queue_id, tmp);
+		dev_err(hdev->dev, "H/W queue %s test failed (SOB_OBJ_0 == 0x%x)\n",
+			GAUDI2_QUEUE_ID_TO_STR(hw_queue_id), tmp);
 		rc = -EIO;
 	}
 
@@ -9603,8 +9966,8 @@ static int hl_arc_event_handle(struct hl_device *hdev, u16 event_type,
 		q = (struct hl_engine_arc_dccm_queue_full_irq *) &payload;
 
 		gaudi2_print_event(hdev, event_type, true,
-				"ARC DCCM Full event: EngId: %u, Intr_type: %u, Qidx: %u",
-				engine_id, intr_type, q->queue_index);
+				"ARC DCCM Full event: Eng: %s, Intr_type: %u, Qidx: %u",
+				GAUDI2_ENG_ID_TO_STR(engine_id), intr_type, q->queue_index);
 		return 1;
 	default:
 		gaudi2_print_event(hdev, event_type, true, "Unknown ARC event type");
@@ -10172,7 +10535,7 @@ static void gaudi2_handle_eqe(struct hl_device *hdev, struct hl_eq_entry *eq_ent
 		dev_err(hdev->dev, "CPLD shutdown event, reset reason: 0x%llx\n",
 						le64_to_cpu(eq_entry->data[0]));
 		error_count = GAUDI2_NA_EVENT_CAUSE;
-		event_mask |= HL_NOTIFIER_EVENT_GENERAL_HW_ERR;
+		hl_eq_cpld_shutdown_event_handle(hdev, event_type, &event_mask);
 		break;
 
 	case GAUDI2_EVENT_CPU_PKT_SANITY_FAILED:
@@ -10260,6 +10623,7 @@ reset_device:
 	if (event_mask & HL_NOTIFIER_EVENT_GENERAL_HW_ERR)
 		hl_handle_critical_hw_err(hdev, event_type, &event_mask);
 
+	hl_debugfs_cfg_access_history_dump(hdev);
 	event_mask |= HL_NOTIFIER_EVENT_DEVICE_RESET;
 	hl_device_cond_reset(hdev, reset_flags, event_mask);
 }
@@ -10296,8 +10660,8 @@ static int gaudi2_memset_memory_chunk_using_edma_qm(struct hl_device *hdev,
 
 	rc = hl_hw_queue_send_cb_no_cmpl(hdev, hw_queue_id, pkt_size, phys_addr);
 	if (rc)
-		dev_err(hdev->dev, "Failed to send lin_dma packet to H/W queue %d\n",
-				hw_queue_id);
+		dev_err(hdev->dev, "Failed to send lin_dma packet to H/W queue %s\n",
+				GAUDI2_QUEUE_ID_TO_STR(hw_queue_id));
 
 	return rc;
 }
@@ -10437,7 +10801,7 @@ end:
 				(u64 *)(lin_dma_pkts_arr), DEBUGFS_WRITE64);
 	WREG32(sob_addr, 0);
 
-	kfree(lin_dma_pkts_arr);
+	kvfree(lin_dma_pkts_arr);
 
 	return rc;
 }
diff --git a/drivers/accel/habanalabs/gaudi2/gaudi2P.h b/drivers/accel/habanalabs/gaudi2/gaudi2P.h
index 05117272cac7..bdf5c1bd2d63 100644
--- a/drivers/accel/habanalabs/gaudi2/gaudi2P.h
+++ b/drivers/accel/habanalabs/gaudi2/gaudi2P.h
@@ -240,6 +240,15 @@
 #define GAUDI2_NUM_TESTED_QS		(GAUDI2_QUEUE_ID_CPU_PQ - GAUDI2_QUEUE_ID_PDMA_0_0)
 
 
+extern const char *gaudi2_engine_id_str[];
+extern const char *gaudi2_queue_id_str[];
+
+#define GAUDI2_ENG_ID_TO_STR(initiator) ((initiator) >= GAUDI2_ENGINE_ID_SIZE ? "not found" :	\
+						gaudi2_engine_id_str[initiator])
+
+#define GAUDI2_QUEUE_ID_TO_STR(initiator) ((initiator) >= GAUDI2_QUEUE_ID_SIZE ? "not found" :	\
+						gaudi2_queue_id_str[initiator])
+
 enum gaudi2_reserved_sob_id {
 	GAUDI2_RESERVED_SOB_CS_COMPLETION_FIRST,
 	GAUDI2_RESERVED_SOB_CS_COMPLETION_LAST =
diff --git a/drivers/accel/habanalabs/gaudi2/gaudi2_coresight.c b/drivers/accel/habanalabs/gaudi2/gaudi2_coresight.c
index 2423620ff358..bc3c57bda5cd 100644
--- a/drivers/accel/habanalabs/gaudi2/gaudi2_coresight.c
+++ b/drivers/accel/habanalabs/gaudi2/gaudi2_coresight.c
@@ -2426,7 +2426,7 @@ static int gaudi2_config_bmon(struct hl_device *hdev, struct hl_debug_params *pa
 		WREG32(base_reg + mmBMON_ADDRH_E3_OFFSET, 0);
 		WREG32(base_reg + mmBMON_REDUCTION_OFFSET, 0);
 		WREG32(base_reg + mmBMON_STM_TRC_OFFSET, 0x7 | (0xA << 8));
-		WREG32(base_reg + mmBMON_CR_OFFSET, 0x77 | 0xf << 24);
+		WREG32(base_reg + mmBMON_CR_OFFSET, 0x41);
 	}
 
 	return 0;
diff --git a/drivers/accel/ivpu/Makefile b/drivers/accel/ivpu/Makefile
index 1029e0bab061..dbf76b8a5b4c 100644
--- a/drivers/accel/ivpu/Makefile
+++ b/drivers/accel/ivpu/Makefile
@@ -6,6 +6,7 @@ intel_vpu-y := \
 	ivpu_fw.o \
 	ivpu_fw_log.o \
 	ivpu_gem.o \
+	ivpu_gem_userptr.o \
 	ivpu_hw.o \
 	ivpu_hw_btrs.o \
 	ivpu_hw_ip.o \
diff --git a/drivers/accel/ivpu/ivpu_debugfs.c b/drivers/accel/ivpu/ivpu_debugfs.c
index cd24ccd20ba6..3bd85ee6c26b 100644
--- a/drivers/accel/ivpu/ivpu_debugfs.c
+++ b/drivers/accel/ivpu/ivpu_debugfs.c
@@ -398,35 +398,25 @@ static int dct_active_set(void *data, u64 active_percent)
 
 DEFINE_DEBUGFS_ATTRIBUTE(ivpu_dct_fops, dct_active_get, dct_active_set, "%llu\n");
 
+static void print_priority_band(struct seq_file *s, struct ivpu_hw_info *hw,
+				int band, const char *name)
+{
+	seq_printf(s, "%-9s: grace_period %9u process_grace_period %9u process_quantum %9u\n",
+		   name,
+		   hw->hws.grace_period[band],
+		   hw->hws.process_grace_period[band],
+		   hw->hws.process_quantum[band]);
+}
+
 static int priority_bands_show(struct seq_file *s, void *v)
 {
 	struct ivpu_device *vdev = s->private;
 	struct ivpu_hw_info *hw = vdev->hw;
 
-	for (int band = VPU_JOB_SCHEDULING_PRIORITY_BAND_IDLE;
-	     band < VPU_JOB_SCHEDULING_PRIORITY_BAND_COUNT; band++) {
-		switch (band) {
-		case VPU_JOB_SCHEDULING_PRIORITY_BAND_IDLE:
-			seq_puts(s, "Idle:     ");
-			break;
-
-		case VPU_JOB_SCHEDULING_PRIORITY_BAND_NORMAL:
-			seq_puts(s, "Normal:   ");
-			break;
-
-		case VPU_JOB_SCHEDULING_PRIORITY_BAND_FOCUS:
-			seq_puts(s, "Focus:    ");
-			break;
-
-		case VPU_JOB_SCHEDULING_PRIORITY_BAND_REALTIME:
-			seq_puts(s, "Realtime: ");
-			break;
-		}
-
-		seq_printf(s, "grace_period %9u process_grace_period %9u process_quantum %9u\n",
-			   hw->hws.grace_period[band], hw->hws.process_grace_period[band],
-			   hw->hws.process_quantum[band]);
-	}
+	print_priority_band(s, hw, VPU_JOB_SCHEDULING_PRIORITY_BAND_IDLE, "Idle");
+	print_priority_band(s, hw, VPU_JOB_SCHEDULING_PRIORITY_BAND_NORMAL, "Normal");
+	print_priority_band(s, hw, VPU_JOB_SCHEDULING_PRIORITY_BAND_FOCUS, "Focus");
+	print_priority_band(s, hw, VPU_JOB_SCHEDULING_PRIORITY_BAND_REALTIME, "Realtime");
 
 	return 0;
 }
diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c
index 0e7748c5e117..3d6fccdefdd6 100644
--- a/drivers/accel/ivpu/ivpu_drv.c
+++ b/drivers/accel/ivpu/ivpu_drv.c
@@ -57,7 +57,7 @@ MODULE_PARM_DESC(pll_max_ratio, "Maximum PLL ratio used to set NPU frequency");
 
 int ivpu_sched_mode = IVPU_SCHED_MODE_AUTO;
 module_param_named(sched_mode, ivpu_sched_mode, int, 0444);
-MODULE_PARM_DESC(sched_mode, "Scheduler mode: -1 - Use default scheduler, 0 - Use OS scheduler, 1 - Use HW scheduler");
+MODULE_PARM_DESC(sched_mode, "Scheduler mode: -1 - Use default scheduler, 0 - Use OS scheduler (supported on 27XX - 50XX), 1 - Use HW scheduler");
 
 bool ivpu_disable_mmu_cont_pages;
 module_param_named(disable_mmu_cont_pages, ivpu_disable_mmu_cont_pages, bool, 0444);
@@ -134,6 +134,8 @@ bool ivpu_is_capable(struct ivpu_device *vdev, u32 capability)
 		return true;
 	case DRM_IVPU_CAP_DMA_MEMORY_RANGE:
 		return true;
+	case DRM_IVPU_CAP_BO_CREATE_FROM_USERPTR:
+		return true;
 	case DRM_IVPU_CAP_MANAGE_CMDQ:
 		return vdev->fw->sched_mode == VPU_SCHEDULING_MODE_HW;
 	default:
@@ -200,6 +202,9 @@ static int ivpu_get_param_ioctl(struct drm_device *dev, void *data, struct drm_f
 	case DRM_IVPU_PARAM_CAPABILITIES:
 		args->value = ivpu_is_capable(vdev, args->index);
 		break;
+	case DRM_IVPU_PARAM_PREEMPT_BUFFER_SIZE:
+		args->value = ivpu_fw_preempt_buf_size(vdev);
+		break;
 	default:
 		ret = -EINVAL;
 		break;
@@ -310,6 +315,7 @@ static const struct drm_ioctl_desc ivpu_drm_ioctls[] = {
 	DRM_IOCTL_DEF_DRV(IVPU_CMDQ_CREATE, ivpu_cmdq_create_ioctl, 0),
 	DRM_IOCTL_DEF_DRV(IVPU_CMDQ_DESTROY, ivpu_cmdq_destroy_ioctl, 0),
 	DRM_IOCTL_DEF_DRV(IVPU_CMDQ_SUBMIT, ivpu_cmdq_submit_ioctl, 0),
+	DRM_IOCTL_DEF_DRV(IVPU_BO_CREATE_FROM_USERPTR, ivpu_bo_create_from_userptr_ioctl, 0),
 };
 
 static int ivpu_wait_for_ready(struct ivpu_device *vdev)
@@ -377,8 +383,7 @@ int ivpu_boot(struct ivpu_device *vdev)
 	drm_WARN_ON(&vdev->drm, atomic_read(&vdev->job_timeout_counter));
 	drm_WARN_ON(&vdev->drm, !xa_empty(&vdev->submitted_jobs_xa));
 
-	/* Update boot params located at first 4KB of FW memory */
-	ivpu_fw_boot_params_setup(vdev, ivpu_bo_vaddr(vdev->fw->mem));
+	ivpu_fw_boot_params_setup(vdev, ivpu_bo_vaddr(vdev->fw->mem_bp));
 
 	ret = ivpu_hw_boot_fw(vdev);
 	if (ret) {
@@ -450,6 +455,9 @@ int ivpu_shutdown(struct ivpu_device *vdev)
 static const struct file_operations ivpu_fops = {
 	.owner		= THIS_MODULE,
 	DRM_ACCEL_FOPS,
+#ifdef CONFIG_PROC_FS
+	.show_fdinfo = drm_show_fdinfo,
+#endif
 };
 
 static const struct drm_driver driver = {
@@ -464,6 +472,9 @@ static const struct drm_driver driver = {
 	.ioctls = ivpu_drm_ioctls,
 	.num_ioctls = ARRAY_SIZE(ivpu_drm_ioctls),
 	.fops = &ivpu_fops,
+#ifdef CONFIG_PROC_FS
+	.show_fdinfo = drm_show_memory_stats,
+#endif
 
 	.name = DRIVER_NAME,
 	.desc = DRIVER_DESC,
@@ -677,7 +688,7 @@ static void ivpu_bo_unbind_all_user_contexts(struct ivpu_device *vdev)
 static void ivpu_dev_fini(struct ivpu_device *vdev)
 {
 	ivpu_jobs_abort_all(vdev);
-	ivpu_pm_cancel_recovery(vdev);
+	ivpu_pm_disable_recovery(vdev);
 	ivpu_pm_disable(vdev);
 	ivpu_prepare_for_reset(vdev);
 	ivpu_shutdown(vdev);
@@ -704,6 +715,8 @@ static struct pci_device_id ivpu_pci_ids[] = {
 	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_ARL) },
 	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_LNL) },
 	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_PTL_P) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_WCL) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_NVL) },
 	{ }
 };
 MODULE_DEVICE_TABLE(pci, ivpu_pci_ids);
diff --git a/drivers/accel/ivpu/ivpu_drv.h b/drivers/accel/ivpu/ivpu_drv.h
index 5497e7030e91..5b34b6f50e69 100644
--- a/drivers/accel/ivpu/ivpu_drv.h
+++ b/drivers/accel/ivpu/ivpu_drv.h
@@ -1,6 +1,6 @@
 /* SPDX-License-Identifier: GPL-2.0-only */
 /*
- * Copyright (C) 2020-2024 Intel Corporation
+ * Copyright (C) 2020-2025 Intel Corporation
  */
 
 #ifndef __IVPU_DRV_H__
@@ -26,6 +26,8 @@
 #define PCI_DEVICE_ID_ARL	0xad1d
 #define PCI_DEVICE_ID_LNL	0x643e
 #define PCI_DEVICE_ID_PTL_P	0xb03e
+#define PCI_DEVICE_ID_WCL	0xfd3e
+#define PCI_DEVICE_ID_NVL	0xd71d
 
 #define IVPU_HW_IP_37XX 37
 #define IVPU_HW_IP_40XX 40
@@ -77,6 +79,7 @@
 #define IVPU_DBG_KREF	 BIT(11)
 #define IVPU_DBG_RPM	 BIT(12)
 #define IVPU_DBG_MMU_MAP BIT(13)
+#define IVPU_DBG_IOCTL   BIT(14)
 
 #define ivpu_err(vdev, fmt, ...) \
 	drm_err(&(vdev)->drm, "%s(): " fmt, __func__, ##__VA_ARGS__)
@@ -165,6 +168,7 @@ struct ivpu_device {
 		int boot;
 		int jsm;
 		int tdr;
+		int inference;
 		int autosuspend;
 		int d0i3_entry_msg;
 		int state_dump_msg;
@@ -207,10 +211,11 @@ extern bool ivpu_force_snoop;
 #define IVPU_TEST_MODE_D0I3_MSG_ENABLE    BIT(5)
 #define IVPU_TEST_MODE_MIP_DISABLE        BIT(6)
 #define IVPU_TEST_MODE_DISABLE_TIMEOUTS   BIT(8)
-#define IVPU_TEST_MODE_TURBO		  BIT(9)
-#define IVPU_TEST_MODE_CLK_RELINQ_DISABLE BIT(10)
-#define IVPU_TEST_MODE_CLK_RELINQ_ENABLE  BIT(11)
-#define IVPU_TEST_MODE_D0I2_DISABLE       BIT(12)
+#define IVPU_TEST_MODE_TURBO_ENABLE       BIT(9)
+#define IVPU_TEST_MODE_TURBO_DISABLE      BIT(10)
+#define IVPU_TEST_MODE_CLK_RELINQ_DISABLE BIT(11)
+#define IVPU_TEST_MODE_CLK_RELINQ_ENABLE  BIT(12)
+#define IVPU_TEST_MODE_D0I2_DISABLE       BIT(13)
 extern int ivpu_test_mode;
 
 struct ivpu_file_priv *ivpu_file_priv_get(struct ivpu_file_priv *file_priv);
@@ -240,7 +245,10 @@ static inline int ivpu_hw_ip_gen(struct ivpu_device *vdev)
 	case PCI_DEVICE_ID_LNL:
 		return IVPU_HW_IP_40XX;
 	case PCI_DEVICE_ID_PTL_P:
+	case PCI_DEVICE_ID_WCL:
 		return IVPU_HW_IP_50XX;
+	case PCI_DEVICE_ID_NVL:
+		return IVPU_HW_IP_60XX;
 	default:
 		dump_stack();
 		ivpu_err(vdev, "Unknown NPU IP generation\n");
@@ -256,6 +264,8 @@ static inline int ivpu_hw_btrs_gen(struct ivpu_device *vdev)
 		return IVPU_HW_BTRS_MTL;
 	case PCI_DEVICE_ID_LNL:
 	case PCI_DEVICE_ID_PTL_P:
+	case PCI_DEVICE_ID_WCL:
+	case PCI_DEVICE_ID_NVL:
 		return IVPU_HW_BTRS_LNL;
 	default:
 		dump_stack();
diff --git a/drivers/accel/ivpu/ivpu_fw.c b/drivers/accel/ivpu/ivpu_fw.c
index 9db741695401..48386d2cddbb 100644
--- a/drivers/accel/ivpu/ivpu_fw.c
+++ b/drivers/accel/ivpu/ivpu_fw.c
@@ -17,15 +17,10 @@
 #include "ivpu_ipc.h"
 #include "ivpu_pm.h"
 
-#define FW_GLOBAL_MEM_START	(2ull * SZ_1G)
-#define FW_GLOBAL_MEM_END	(3ull * SZ_1G)
-#define FW_SHARED_MEM_SIZE	SZ_256M /* Must be aligned to FW_SHARED_MEM_ALIGNMENT */
-#define FW_SHARED_MEM_ALIGNMENT	SZ_128K /* VPU MTRR limitation */
-#define FW_RUNTIME_MAX_SIZE	SZ_512M
 #define FW_SHAVE_NN_MAX_SIZE	SZ_2M
-#define FW_RUNTIME_MIN_ADDR	(FW_GLOBAL_MEM_START)
-#define FW_RUNTIME_MAX_ADDR	(FW_GLOBAL_MEM_END - FW_SHARED_MEM_SIZE)
 #define FW_FILE_IMAGE_OFFSET	(VPU_FW_HEADER_SIZE + FW_VERSION_HEADER_SIZE)
+#define FW_PREEMPT_BUF_MIN_SIZE SZ_4K
+#define FW_PREEMPT_BUF_MAX_SIZE SZ_32M
 
 #define WATCHDOG_MSS_REDIRECT	32
 #define WATCHDOG_NCE_REDIRECT	33
@@ -61,12 +56,14 @@ static struct {
 	{ IVPU_HW_IP_40XX, "intel/vpu/vpu_40xx_v0.0.bin" },
 	{ IVPU_HW_IP_50XX, "intel/vpu/vpu_50xx_v1.bin" },
 	{ IVPU_HW_IP_50XX, "intel/vpu/vpu_50xx_v0.0.bin" },
+	{ IVPU_HW_IP_60XX, "intel/vpu/vpu_60xx_v1.bin" },
 };
 
 /* Production fw_names from the table above */
 MODULE_FIRMWARE("intel/vpu/vpu_37xx_v1.bin");
 MODULE_FIRMWARE("intel/vpu/vpu_40xx_v1.bin");
 MODULE_FIRMWARE("intel/vpu/vpu_50xx_v1.bin");
+MODULE_FIRMWARE("intel/vpu/vpu_60xx_v1.bin");
 
 static int ivpu_fw_request(struct ivpu_device *vdev)
 {
@@ -131,9 +128,14 @@ ivpu_fw_check_api_ver_lt(struct ivpu_device *vdev, const struct vpu_firmware_hea
 	return false;
 }
 
-static bool is_within_range(u64 addr, size_t size, u64 range_start, size_t range_size)
+bool ivpu_is_within_range(u64 addr, size_t size, struct ivpu_addr_range *range)
 {
-	if (addr < range_start || addr + size > range_start + range_size)
+	u64 addr_end;
+
+	if (!range || check_add_overflow(addr, size, &addr_end))
+		return false;
+
+	if (addr < range->start || addr_end > range->end)
 		return false;
 
 	return true;
@@ -142,6 +144,12 @@ static bool is_within_range(u64 addr, size_t size, u64 range_start, size_t range
 static u32
 ivpu_fw_sched_mode_select(struct ivpu_device *vdev, const struct vpu_firmware_header *fw_hdr)
 {
+	if (ivpu_hw_ip_gen(vdev) >= IVPU_HW_IP_60XX &&
+	    ivpu_sched_mode == VPU_SCHEDULING_MODE_OS) {
+		ivpu_warn(vdev, "OS sched mode is not supported, using HW mode\n");
+		return VPU_SCHEDULING_MODE_HW;
+	}
+
 	if (ivpu_sched_mode != IVPU_SCHED_MODE_AUTO)
 		return ivpu_sched_mode;
 
@@ -151,11 +159,56 @@ ivpu_fw_sched_mode_select(struct ivpu_device *vdev, const struct vpu_firmware_he
 	return VPU_SCHEDULING_MODE_HW;
 }
 
+static void
+ivpu_preemption_config_parse(struct ivpu_device *vdev, const struct vpu_firmware_header *fw_hdr)
+{
+	struct ivpu_fw_info *fw = vdev->fw;
+	u32 primary_preempt_buf_size, secondary_preempt_buf_size;
+
+	if (fw_hdr->preemption_buffer_1_max_size)
+		primary_preempt_buf_size = fw_hdr->preemption_buffer_1_max_size;
+	else
+		primary_preempt_buf_size = fw_hdr->preemption_buffer_1_size;
+
+	if (fw_hdr->preemption_buffer_2_max_size)
+		secondary_preempt_buf_size = fw_hdr->preemption_buffer_2_max_size;
+	else
+		secondary_preempt_buf_size = fw_hdr->preemption_buffer_2_size;
+
+	ivpu_dbg(vdev, FW_BOOT, "Preemption buffer size, primary: %u, secondary: %u\n",
+		 primary_preempt_buf_size, secondary_preempt_buf_size);
+
+	if (primary_preempt_buf_size < FW_PREEMPT_BUF_MIN_SIZE ||
+	    secondary_preempt_buf_size < FW_PREEMPT_BUF_MIN_SIZE) {
+		ivpu_warn(vdev, "Preemption buffers size too small\n");
+		return;
+	}
+
+	if (primary_preempt_buf_size > FW_PREEMPT_BUF_MAX_SIZE ||
+	    secondary_preempt_buf_size > FW_PREEMPT_BUF_MAX_SIZE) {
+		ivpu_warn(vdev, "Preemption buffers size too big\n");
+		return;
+	}
+
+	if (fw->sched_mode != VPU_SCHEDULING_MODE_HW)
+		return;
+
+	if (ivpu_test_mode & IVPU_TEST_MODE_MIP_DISABLE)
+		return;
+
+	vdev->fw->primary_preempt_buf_size = ALIGN(primary_preempt_buf_size, PAGE_SIZE);
+	vdev->fw->secondary_preempt_buf_size = ALIGN(secondary_preempt_buf_size, PAGE_SIZE);
+}
+
 static int ivpu_fw_parse(struct ivpu_device *vdev)
 {
 	struct ivpu_fw_info *fw = vdev->fw;
 	const struct vpu_firmware_header *fw_hdr = (const void *)fw->file->data;
-	u64 runtime_addr, image_load_addr, runtime_size, image_size;
+	struct ivpu_addr_range fw_image_range;
+	u64 boot_params_addr, boot_params_size;
+	u64 fw_version_addr, fw_version_size;
+	u64 runtime_addr, runtime_size;
+	u64 image_load_addr, image_size;
 
 	if (fw->file->size <= FW_FILE_IMAGE_OFFSET) {
 		ivpu_err(vdev, "Firmware file is too small: %zu\n", fw->file->size);
@@ -167,18 +220,37 @@ static int ivpu_fw_parse(struct ivpu_device *vdev)
 		return -EINVAL;
 	}
 
-	runtime_addr = fw_hdr->boot_params_load_address;
-	runtime_size = fw_hdr->runtime_size;
-	image_load_addr = fw_hdr->image_load_address;
-	image_size = fw_hdr->image_size;
+	boot_params_addr = fw_hdr->boot_params_load_address;
+	boot_params_size = SZ_4K;
 
-	if (runtime_addr < FW_RUNTIME_MIN_ADDR || runtime_addr > FW_RUNTIME_MAX_ADDR) {
-		ivpu_err(vdev, "Invalid firmware runtime address: 0x%llx\n", runtime_addr);
+	if (!ivpu_is_within_range(boot_params_addr, boot_params_size, &vdev->hw->ranges.runtime)) {
+		ivpu_err(vdev, "Invalid boot params address: 0x%llx\n", boot_params_addr);
 		return -EINVAL;
 	}
 
-	if (runtime_size < fw->file->size || runtime_size > FW_RUNTIME_MAX_SIZE) {
-		ivpu_err(vdev, "Invalid firmware runtime size: %llu\n", runtime_size);
+	fw_version_addr = fw_hdr->firmware_version_load_address;
+	fw_version_size = ALIGN(fw_hdr->firmware_version_size, SZ_4K);
+
+	if (fw_version_size != SZ_4K) {
+		ivpu_err(vdev, "Invalid firmware version size: %u\n",
+			 fw_hdr->firmware_version_size);
+		return -EINVAL;
+	}
+
+	if (!ivpu_is_within_range(fw_version_addr, fw_version_size, &vdev->hw->ranges.runtime)) {
+		ivpu_err(vdev, "Invalid firmware version address: 0x%llx\n", fw_version_addr);
+		return -EINVAL;
+	}
+
+	runtime_addr = fw_hdr->image_load_address;
+	runtime_size = fw_hdr->runtime_size - boot_params_size - fw_version_size;
+
+	image_load_addr = fw_hdr->image_load_address;
+	image_size = fw_hdr->image_size;
+
+	if (!ivpu_is_within_range(runtime_addr, runtime_size, &vdev->hw->ranges.runtime)) {
+		ivpu_err(vdev, "Invalid firmware runtime address: 0x%llx and size %llu\n",
+			 runtime_addr, runtime_size);
 		return -EINVAL;
 	}
 
@@ -187,23 +259,25 @@ static int ivpu_fw_parse(struct ivpu_device *vdev)
 		return -EINVAL;
 	}
 
-	if (image_load_addr < runtime_addr ||
-	    image_load_addr + image_size > runtime_addr + runtime_size) {
-		ivpu_err(vdev, "Invalid firmware load address size: 0x%llx and size %llu\n",
+	if (!ivpu_is_within_range(image_load_addr, image_size, &vdev->hw->ranges.runtime)) {
+		ivpu_err(vdev, "Invalid firmware load address: 0x%llx and size %llu\n",
 			 image_load_addr, image_size);
 		return -EINVAL;
 	}
 
-	if (fw_hdr->shave_nn_fw_size > FW_SHAVE_NN_MAX_SIZE) {
-		ivpu_err(vdev, "SHAVE NN firmware is too big: %u\n", fw_hdr->shave_nn_fw_size);
+	if (ivpu_hw_range_init(vdev, &fw_image_range, image_load_addr, image_size))
 		return -EINVAL;
-	}
 
-	if (fw_hdr->entry_point < image_load_addr ||
-	    fw_hdr->entry_point >= image_load_addr + image_size) {
+	if (!ivpu_is_within_range(fw_hdr->entry_point, SZ_4K, &fw_image_range)) {
 		ivpu_err(vdev, "Invalid entry point: 0x%llx\n", fw_hdr->entry_point);
 		return -EINVAL;
 	}
+
+	if (fw_hdr->shave_nn_fw_size > FW_SHAVE_NN_MAX_SIZE) {
+		ivpu_err(vdev, "SHAVE NN firmware is too big: %u\n", fw_hdr->shave_nn_fw_size);
+		return -EINVAL;
+	}
+
 	ivpu_dbg(vdev, FW_BOOT, "Header version: 0x%x, format 0x%x\n",
 		 fw_hdr->header_version, fw_hdr->image_format);
 
@@ -217,6 +291,10 @@ static int ivpu_fw_parse(struct ivpu_device *vdev)
 	if (IVPU_FW_CHECK_API_COMPAT(vdev, fw_hdr, JSM, 3))
 		return -EINVAL;
 
+	fw->boot_params_addr = boot_params_addr;
+	fw->boot_params_size = boot_params_size;
+	fw->fw_version_addr = fw_version_addr;
+	fw->fw_version_size = fw_version_size;
 	fw->runtime_addr = runtime_addr;
 	fw->runtime_size = runtime_size;
 	fw->image_load_offset = image_load_addr - runtime_addr;
@@ -235,22 +313,13 @@ static int ivpu_fw_parse(struct ivpu_device *vdev)
 	fw->sched_mode = ivpu_fw_sched_mode_select(vdev, fw_hdr);
 	ivpu_info(vdev, "Scheduler mode: %s\n", fw->sched_mode ? "HW" : "OS");
 
-	if (fw_hdr->preemption_buffer_1_max_size)
-		fw->primary_preempt_buf_size = fw_hdr->preemption_buffer_1_max_size;
-	else
-		fw->primary_preempt_buf_size = fw_hdr->preemption_buffer_1_size;
+	ivpu_preemption_config_parse(vdev, fw_hdr);
+	ivpu_dbg(vdev, FW_BOOT, "Mid-inference preemption %s supported\n",
+		 ivpu_fw_preempt_buf_size(vdev) ? "is" : "is not");
 
-	if (fw_hdr->preemption_buffer_2_max_size)
-		fw->secondary_preempt_buf_size = fw_hdr->preemption_buffer_2_max_size;
-	else
-		fw->secondary_preempt_buf_size = fw_hdr->preemption_buffer_2_size;
-	ivpu_dbg(vdev, FW_BOOT, "Preemption buffer sizes: primary %u, secondary %u\n",
-		 fw->primary_preempt_buf_size, fw->secondary_preempt_buf_size);
-
-	if (fw_hdr->ro_section_start_address && !is_within_range(fw_hdr->ro_section_start_address,
-								 fw_hdr->ro_section_size,
-								 fw_hdr->image_load_address,
-								 fw_hdr->image_size)) {
+	if (fw_hdr->ro_section_start_address &&
+	    !ivpu_is_within_range(fw_hdr->ro_section_start_address, fw_hdr->ro_section_size,
+				  &fw_image_range)) {
 		ivpu_err(vdev, "Invalid read-only section: start address 0x%llx, size %u\n",
 			 fw_hdr->ro_section_start_address, fw_hdr->ro_section_size);
 		return -EINVAL;
@@ -259,12 +328,18 @@ static int ivpu_fw_parse(struct ivpu_device *vdev)
 	fw->read_only_addr = fw_hdr->ro_section_start_address;
 	fw->read_only_size = fw_hdr->ro_section_size;
 
-	ivpu_dbg(vdev, FW_BOOT, "Size: file %lu image %u runtime %u shavenn %u\n",
-		 fw->file->size, fw->image_size, fw->runtime_size, fw->shave_nn_size);
-	ivpu_dbg(vdev, FW_BOOT, "Address: runtime 0x%llx, load 0x%llx, entry point 0x%llx\n",
-		 fw->runtime_addr, image_load_addr, fw->entry_point);
+	ivpu_dbg(vdev, FW_BOOT, "Boot params: address 0x%llx, size %llu\n",
+		 fw->boot_params_addr, fw->boot_params_size);
+	ivpu_dbg(vdev, FW_BOOT, "FW version:  address 0x%llx, size %llu\n",
+		 fw->fw_version_addr, fw->fw_version_size);
+	ivpu_dbg(vdev, FW_BOOT, "Runtime:     address 0x%llx, size %u\n",
+		 fw->runtime_addr, fw->runtime_size);
+	ivpu_dbg(vdev, FW_BOOT, "Image load offset: 0x%llx, size %u\n",
+		 fw->image_load_offset, fw->image_size);
 	ivpu_dbg(vdev, FW_BOOT, "Read-only section: address 0x%llx, size %u\n",
 		 fw->read_only_addr, fw->read_only_size);
+	ivpu_dbg(vdev, FW_BOOT, "FW entry point: 0x%llx\n", fw->entry_point);
+	ivpu_dbg(vdev, FW_BOOT, "SHAVE NN size: %u\n", fw->shave_nn_size);
 
 	return 0;
 }
@@ -291,39 +366,33 @@ ivpu_fw_init_wa(struct ivpu_device *vdev)
 	IVPU_PRINT_WA(disable_d0i3_msg);
 }
 
-static int ivpu_fw_update_global_range(struct ivpu_device *vdev)
-{
-	struct ivpu_fw_info *fw = vdev->fw;
-	u64 start = ALIGN(fw->runtime_addr + fw->runtime_size, FW_SHARED_MEM_ALIGNMENT);
-	u64 size = FW_SHARED_MEM_SIZE;
-
-	if (start + size > FW_GLOBAL_MEM_END) {
-		ivpu_err(vdev, "No space for shared region, start %lld, size %lld\n", start, size);
-		return -EINVAL;
-	}
-
-	ivpu_hw_range_init(&vdev->hw->ranges.global, start, size);
-	return 0;
-}
-
 static int ivpu_fw_mem_init(struct ivpu_device *vdev)
 {
 	struct ivpu_fw_info *fw = vdev->fw;
-	struct ivpu_addr_range fw_range;
 	int log_verb_size;
 	int ret;
 
-	ret = ivpu_fw_update_global_range(vdev);
-	if (ret)
-		return ret;
+	fw->mem_bp = ivpu_bo_create_runtime(vdev, fw->boot_params_addr, fw->boot_params_size,
+					    DRM_IVPU_BO_WC | DRM_IVPU_BO_MAPPABLE);
+	if (!fw->mem_bp) {
+		ivpu_err(vdev, "Failed to create firmware boot params memory buffer\n");
+		return -ENOMEM;
+	}
 
-	fw_range.start = fw->runtime_addr;
-	fw_range.end = fw->runtime_addr + fw->runtime_size;
-	fw->mem = ivpu_bo_create(vdev, &vdev->gctx, &fw_range, fw->runtime_size,
-				 DRM_IVPU_BO_WC | DRM_IVPU_BO_MAPPABLE);
+	fw->mem_fw_ver = ivpu_bo_create_runtime(vdev, fw->fw_version_addr, fw->fw_version_size,
+						DRM_IVPU_BO_WC | DRM_IVPU_BO_MAPPABLE);
+	if (!fw->mem_fw_ver) {
+		ivpu_err(vdev, "Failed to create firmware version memory buffer\n");
+		ret = -ENOMEM;
+		goto err_free_bp;
+	}
+
+	fw->mem = ivpu_bo_create_runtime(vdev, fw->runtime_addr, fw->runtime_size,
+					 DRM_IVPU_BO_WC | DRM_IVPU_BO_MAPPABLE);
 	if (!fw->mem) {
 		ivpu_err(vdev, "Failed to create firmware runtime memory buffer\n");
-		return -ENOMEM;
+		ret = -ENOMEM;
+		goto err_free_fw_ver;
 	}
 
 	ret = ivpu_mmu_context_set_pages_ro(vdev, &vdev->gctx, fw->read_only_addr,
@@ -372,6 +441,10 @@ err_free_log_crit:
 	ivpu_bo_free(fw->mem_log_crit);
 err_free_fw_mem:
 	ivpu_bo_free(fw->mem);
+err_free_fw_ver:
+	ivpu_bo_free(fw->mem_fw_ver);
+err_free_bp:
+	ivpu_bo_free(fw->mem_bp);
 	return ret;
 }
 
@@ -387,10 +460,14 @@ static void ivpu_fw_mem_fini(struct ivpu_device *vdev)
 	ivpu_bo_free(fw->mem_log_verb);
 	ivpu_bo_free(fw->mem_log_crit);
 	ivpu_bo_free(fw->mem);
+	ivpu_bo_free(fw->mem_fw_ver);
+	ivpu_bo_free(fw->mem_bp);
 
 	fw->mem_log_verb = NULL;
 	fw->mem_log_crit = NULL;
 	fw->mem = NULL;
+	fw->mem_fw_ver = NULL;
+	fw->mem_bp = NULL;
 }
 
 int ivpu_fw_init(struct ivpu_device *vdev)
@@ -483,11 +560,6 @@ static void ivpu_fw_boot_params_print(struct ivpu_device *vdev, struct vpu_boot_
 	ivpu_dbg(vdev, FW_BOOT, "boot_params.cache_defaults[VPU_BOOT_L2_CACHE_CFG_NN].cfg = 0x%x\n",
 		 boot_params->cache_defaults[VPU_BOOT_L2_CACHE_CFG_NN].cfg);
 
-	ivpu_dbg(vdev, FW_BOOT, "boot_params.global_memory_allocator_base = 0x%llx\n",
-		 boot_params->global_memory_allocator_base);
-	ivpu_dbg(vdev, FW_BOOT, "boot_params.global_memory_allocator_size = 0x%x\n",
-		 boot_params->global_memory_allocator_size);
-
 	ivpu_dbg(vdev, FW_BOOT, "boot_params.shave_nn_fw_base = 0x%llx\n",
 		 boot_params->shave_nn_fw_base);
 
@@ -495,10 +567,6 @@ static void ivpu_fw_boot_params_print(struct ivpu_device *vdev, struct vpu_boot_
 		 boot_params->watchdog_irq_mss);
 	ivpu_dbg(vdev, FW_BOOT, "boot_params.watchdog_irq_nce = 0x%x\n",
 		 boot_params->watchdog_irq_nce);
-	ivpu_dbg(vdev, FW_BOOT, "boot_params.host_to_vpu_irq = 0x%x\n",
-		 boot_params->host_to_vpu_irq);
-	ivpu_dbg(vdev, FW_BOOT, "boot_params.job_done_irq = 0x%x\n",
-		 boot_params->job_done_irq);
 
 	ivpu_dbg(vdev, FW_BOOT, "boot_params.host_version_id = 0x%x\n",
 		 boot_params->host_version_id);
@@ -546,6 +614,8 @@ static void ivpu_fw_boot_params_print(struct ivpu_device *vdev, struct vpu_boot_
 		 boot_params->system_time_us);
 	ivpu_dbg(vdev, FW_BOOT, "boot_params.power_profile = 0x%x\n",
 		 boot_params->power_profile);
+	ivpu_dbg(vdev, FW_BOOT, "boot_params.vpu_uses_ecc_mca_signal = 0x%x\n",
+		 boot_params->vpu_uses_ecc_mca_signal);
 }
 
 void ivpu_fw_boot_params_setup(struct ivpu_device *vdev, struct vpu_boot_params *boot_params)
@@ -572,6 +642,7 @@ void ivpu_fw_boot_params_setup(struct ivpu_device *vdev, struct vpu_boot_params
 		return;
 	}
 
+	memset(boot_params, 0, sizeof(*boot_params));
 	vdev->pm->is_warmboot = false;
 
 	boot_params->magic = VPU_BOOT_PARAMS_MAGIC;
@@ -647,6 +718,8 @@ void ivpu_fw_boot_params_setup(struct ivpu_device *vdev, struct vpu_boot_params
 	boot_params->d0i3_entry_vpu_ts = 0;
 	if (IVPU_WA(disable_d0i2))
 		boot_params->power_profile |= BIT(1);
+	boot_params->vpu_uses_ecc_mca_signal =
+		ivpu_hw_uses_ecc_mca_signal(vdev) ? VPU_BOOT_MCA_ECC_BOTH : 0;
 
 	boot_params->system_time_us = ktime_to_us(ktime_get_real());
 	wmb(); /* Flush WC buffers after writing bootparams */
diff --git a/drivers/accel/ivpu/ivpu_fw.h b/drivers/accel/ivpu/ivpu_fw.h
index 9a3935be1c05..00945892b55e 100644
--- a/drivers/accel/ivpu/ivpu_fw.h
+++ b/drivers/accel/ivpu/ivpu_fw.h
@@ -1,6 +1,6 @@
 /* SPDX-License-Identifier: GPL-2.0-only */
 /*
- * Copyright (C) 2020-2024 Intel Corporation
+ * Copyright (C) 2020-2025 Intel Corporation
  */
 
 #ifndef __IVPU_FW_H__
@@ -19,10 +19,16 @@ struct ivpu_fw_info {
 	const struct firmware *file;
 	const char *name;
 	char version[FW_VERSION_STR_SIZE];
+	struct ivpu_bo *mem_bp;
+	struct ivpu_bo *mem_fw_ver;
 	struct ivpu_bo *mem;
 	struct ivpu_bo *mem_shave_nn;
 	struct ivpu_bo *mem_log_crit;
 	struct ivpu_bo *mem_log_verb;
+	u64 boot_params_addr;
+	u64 boot_params_size;
+	u64 fw_version_addr;
+	u64 fw_version_size;
 	u64 runtime_addr;
 	u32 runtime_size;
 	u64 image_load_offset;
@@ -42,14 +48,20 @@ struct ivpu_fw_info {
 	u64 last_heartbeat;
 };
 
+bool ivpu_is_within_range(u64 addr, size_t size, struct ivpu_addr_range *range);
 int ivpu_fw_init(struct ivpu_device *vdev);
 void ivpu_fw_fini(struct ivpu_device *vdev);
 void ivpu_fw_load(struct ivpu_device *vdev);
-void ivpu_fw_boot_params_setup(struct ivpu_device *vdev, struct vpu_boot_params *bp);
+void ivpu_fw_boot_params_setup(struct ivpu_device *vdev, struct vpu_boot_params *boot_params);
 
 static inline bool ivpu_fw_is_cold_boot(struct ivpu_device *vdev)
 {
 	return vdev->fw->entry_point == vdev->fw->cold_boot_entry_point;
 }
 
+static inline u32 ivpu_fw_preempt_buf_size(struct ivpu_device *vdev)
+{
+	return vdev->fw->primary_preempt_buf_size + vdev->fw->secondary_preempt_buf_size;
+}
+
 #endif /* __IVPU_FW_H__ */
diff --git a/drivers/accel/ivpu/ivpu_gem.c b/drivers/accel/ivpu/ivpu_gem.c
index 59cfcf3eaded..ece68f570b7e 100644
--- a/drivers/accel/ivpu/ivpu_gem.c
+++ b/drivers/accel/ivpu/ivpu_gem.c
@@ -15,6 +15,7 @@
 #include <drm/drm_utils.h>
 
 #include "ivpu_drv.h"
+#include "ivpu_fw.h"
 #include "ivpu_gem.h"
 #include "ivpu_hw.h"
 #include "ivpu_mmu.h"
@@ -27,8 +28,8 @@ static const struct drm_gem_object_funcs ivpu_gem_funcs;
 static inline void ivpu_dbg_bo(struct ivpu_device *vdev, struct ivpu_bo *bo, const char *action)
 {
 	ivpu_dbg(vdev, BO,
-		 "%6s: bo %8p vpu_addr %9llx size %8zu ctx %d has_pages %d dma_mapped %d mmu_mapped %d wc %d imported %d\n",
-		 action, bo, bo->vpu_addr, ivpu_bo_size(bo), bo->ctx_id,
+		 "%6s: bo %8p size %9zu ctx %d vpu_addr %9llx pages %d sgt %d mmu_mapped %d wc %d imported %d\n",
+		 action, bo, ivpu_bo_size(bo), bo->ctx_id, bo->vpu_addr,
 		 (bool)bo->base.pages, (bool)bo->base.sgt, bo->mmu_mapped, bo->base.map_wc,
 		 (bool)drm_gem_is_imported(&bo->base.base));
 }
@@ -43,22 +44,47 @@ static inline void ivpu_bo_unlock(struct ivpu_bo *bo)
 	dma_resv_unlock(bo->base.base.resv);
 }
 
+static struct sg_table *ivpu_bo_map_attachment(struct ivpu_device *vdev, struct ivpu_bo *bo)
+{
+	struct sg_table *sgt;
+
+	drm_WARN_ON(&vdev->drm, !bo->base.base.import_attach);
+
+	ivpu_bo_lock(bo);
+
+	sgt = bo->base.sgt;
+	if (!sgt) {
+		sgt = dma_buf_map_attachment(bo->base.base.import_attach, DMA_BIDIRECTIONAL);
+		if (IS_ERR(sgt))
+			ivpu_err(vdev, "Failed to map BO in IOMMU: %ld\n", PTR_ERR(sgt));
+		else
+			bo->base.sgt = sgt;
+	}
+
+	ivpu_bo_unlock(bo);
+
+	return sgt;
+}
+
 /*
- * ivpu_bo_pin() - pin the backing physical pages and map them to VPU.
+ * ivpu_bo_bind() - pin the backing physical pages and map them to VPU.
  *
  * This function pins physical memory pages, then maps the physical pages
  * to IOMMU address space and finally updates the VPU MMU page tables
  * to allow the VPU to translate VPU address to IOMMU address.
  */
-int __must_check ivpu_bo_pin(struct ivpu_bo *bo)
+int __must_check ivpu_bo_bind(struct ivpu_bo *bo)
 {
 	struct ivpu_device *vdev = ivpu_bo_to_vdev(bo);
 	struct sg_table *sgt;
 	int ret = 0;
 
-	ivpu_dbg_bo(vdev, bo, "pin");
+	ivpu_dbg_bo(vdev, bo, "bind");
 
-	sgt = drm_gem_shmem_get_pages_sgt(&bo->base);
+	if (bo->base.base.import_attach)
+		sgt = ivpu_bo_map_attachment(vdev, bo);
+	else
+		sgt = drm_gem_shmem_get_pages_sgt(&bo->base);
 	if (IS_ERR(sgt)) {
 		ret = PTR_ERR(sgt);
 		ivpu_err(vdev, "Failed to map BO in IOMMU: %d\n", ret);
@@ -70,7 +96,7 @@ int __must_check ivpu_bo_pin(struct ivpu_bo *bo)
 	if (!bo->mmu_mapped) {
 		drm_WARN_ON(&vdev->drm, !bo->ctx);
 		ret = ivpu_mmu_context_map_sgt(vdev, bo->ctx, bo->vpu_addr, sgt,
-					       ivpu_bo_is_snooped(bo));
+					       ivpu_bo_is_snooped(bo), ivpu_bo_is_read_only(bo));
 		if (ret) {
 			ivpu_err(vdev, "Failed to map BO in MMU: %d\n", ret);
 			goto unlock;
@@ -99,9 +125,9 @@ ivpu_bo_alloc_vpu_addr(struct ivpu_bo *bo, struct ivpu_mmu_context *ctx,
 	ret = ivpu_mmu_context_insert_node(ctx, range, ivpu_bo_size(bo), &bo->mm_node);
 	if (!ret) {
 		bo->ctx = ctx;
+		bo->ctx_id = ctx->id;
 		bo->vpu_addr = bo->mm_node.start;
-	} else {
-		ivpu_err(vdev, "Failed to add BO to context %u: %d\n", ctx->id, ret);
+		ivpu_dbg_bo(vdev, bo, "vaddr");
 	}
 
 	ivpu_bo_unlock(bo);
@@ -115,7 +141,7 @@ static void ivpu_bo_unbind_locked(struct ivpu_bo *bo)
 {
 	struct ivpu_device *vdev = ivpu_bo_to_vdev(bo);
 
-	lockdep_assert(dma_resv_held(bo->base.base.resv) || !kref_read(&bo->base.base.refcount));
+	dma_resv_assert_held(bo->base.base.resv);
 
 	if (bo->mmu_mapped) {
 		drm_WARN_ON(&vdev->drm, !bo->ctx);
@@ -130,13 +156,15 @@ static void ivpu_bo_unbind_locked(struct ivpu_bo *bo)
 		bo->ctx = NULL;
 	}
 
-	if (drm_gem_is_imported(&bo->base.base))
-		return;
-
 	if (bo->base.sgt) {
-		dma_unmap_sgtable(vdev->drm.dev, bo->base.sgt, DMA_BIDIRECTIONAL, 0);
-		sg_free_table(bo->base.sgt);
-		kfree(bo->base.sgt);
+		if (bo->base.base.import_attach) {
+			dma_buf_unmap_attachment(bo->base.base.import_attach,
+						 bo->base.sgt, DMA_BIDIRECTIONAL);
+		} else {
+			dma_unmap_sgtable(vdev->drm.dev, bo->base.sgt, DMA_BIDIRECTIONAL, 0);
+			sg_free_table(bo->base.sgt);
+			kfree(bo->base.sgt);
+		}
 		bo->base.sgt = NULL;
 	}
 }
@@ -182,10 +210,11 @@ struct drm_gem_object *ivpu_gem_create_object(struct drm_device *dev, size_t siz
 struct drm_gem_object *ivpu_gem_prime_import(struct drm_device *dev,
 					     struct dma_buf *dma_buf)
 {
+	struct ivpu_device *vdev = to_ivpu_device(dev);
 	struct device *attach_dev = dev->dev;
 	struct dma_buf_attachment *attach;
-	struct sg_table *sgt;
 	struct drm_gem_object *obj;
+	struct ivpu_bo *bo;
 	int ret;
 
 	attach = dma_buf_attach(dma_buf, attach_dev);
@@ -194,25 +223,25 @@ struct drm_gem_object *ivpu_gem_prime_import(struct drm_device *dev,
 
 	get_dma_buf(dma_buf);
 
-	sgt = dma_buf_map_attachment_unlocked(attach, DMA_BIDIRECTIONAL);
-	if (IS_ERR(sgt)) {
-		ret = PTR_ERR(sgt);
-		goto fail_detach;
-	}
-
-	obj = drm_gem_shmem_prime_import_sg_table(dev, attach, sgt);
+	obj = drm_gem_shmem_prime_import_sg_table(dev, attach, NULL);
 	if (IS_ERR(obj)) {
 		ret = PTR_ERR(obj);
-		goto fail_unmap;
+		goto fail_detach;
 	}
 
 	obj->import_attach = attach;
 	obj->resv = dma_buf->resv;
 
+	bo = to_ivpu_bo(obj);
+
+	mutex_lock(&vdev->bo_list_lock);
+	list_add_tail(&bo->bo_list_node, &vdev->bo_list);
+	mutex_unlock(&vdev->bo_list_lock);
+
+	ivpu_dbg(vdev, BO, "import: bo %8p size %9zu\n", bo, ivpu_bo_size(bo));
+
 	return obj;
 
-fail_unmap:
-	dma_buf_unmap_attachment_unlocked(attach, sgt, DMA_BIDIRECTIONAL);
 fail_detach:
 	dma_buf_detach(dma_buf, attach);
 	dma_buf_put(dma_buf);
@@ -220,7 +249,7 @@ fail_detach:
 	return ERR_PTR(ret);
 }
 
-static struct ivpu_bo *ivpu_bo_alloc(struct ivpu_device *vdev, u64 size, u32 flags, u32 ctx_id)
+static struct ivpu_bo *ivpu_bo_alloc(struct ivpu_device *vdev, u64 size, u32 flags)
 {
 	struct drm_gem_shmem_object *shmem;
 	struct ivpu_bo *bo;
@@ -238,7 +267,6 @@ static struct ivpu_bo *ivpu_bo_alloc(struct ivpu_device *vdev, u64 size, u32 fla
 		return ERR_CAST(shmem);
 
 	bo = to_ivpu_bo(&shmem->base);
-	bo->ctx_id = ctx_id;
 	bo->base.map_wc = flags & DRM_IVPU_BO_WC;
 	bo->flags = flags;
 
@@ -246,7 +274,7 @@ static struct ivpu_bo *ivpu_bo_alloc(struct ivpu_device *vdev, u64 size, u32 fla
 	list_add_tail(&bo->bo_list_node, &vdev->bo_list);
 	mutex_unlock(&vdev->bo_list_lock);
 
-	ivpu_dbg_bo(vdev, bo, "alloc");
+	ivpu_dbg(vdev, BO, " alloc: bo %8p size %9llu\n", bo, size);
 
 	return bo;
 }
@@ -259,8 +287,8 @@ static int ivpu_gem_bo_open(struct drm_gem_object *obj, struct drm_file *file)
 	struct ivpu_addr_range *range;
 
 	if (bo->ctx) {
-		ivpu_warn(vdev, "Can't add BO to ctx %u: already in ctx %u\n",
-			  file_priv->ctx.id, bo->ctx->id);
+		ivpu_dbg(vdev, IOCTL, "Can't add BO %pe to ctx %u: already in ctx %u\n",
+			 bo, file_priv->ctx.id, bo->ctx->id);
 		return -EALREADY;
 	}
 
@@ -281,23 +309,41 @@ static void ivpu_gem_bo_free(struct drm_gem_object *obj)
 
 	ivpu_dbg_bo(vdev, bo, "free");
 
+	drm_WARN_ON(&vdev->drm, list_empty(&bo->bo_list_node));
+
 	mutex_lock(&vdev->bo_list_lock);
 	list_del(&bo->bo_list_node);
-	mutex_unlock(&vdev->bo_list_lock);
 
 	drm_WARN_ON(&vdev->drm, !drm_gem_is_imported(&bo->base.base) &&
 		    !dma_resv_test_signaled(obj->resv, DMA_RESV_USAGE_READ));
 	drm_WARN_ON(&vdev->drm, ivpu_bo_size(bo) == 0);
 	drm_WARN_ON(&vdev->drm, bo->base.vaddr);
 
+	ivpu_bo_lock(bo);
 	ivpu_bo_unbind_locked(bo);
+	ivpu_bo_unlock(bo);
+
+	mutex_unlock(&vdev->bo_list_lock);
+
 	drm_WARN_ON(&vdev->drm, bo->mmu_mapped);
 	drm_WARN_ON(&vdev->drm, bo->ctx);
 
 	drm_WARN_ON(obj->dev, refcount_read(&bo->base.pages_use_count) > 1);
+	drm_WARN_ON(obj->dev, bo->base.base.vma_node.vm_files.rb_node);
 	drm_gem_shmem_free(&bo->base);
 }
 
+static enum drm_gem_object_status ivpu_gem_status(struct drm_gem_object *obj)
+{
+	struct ivpu_bo *bo = to_ivpu_bo(obj);
+	enum drm_gem_object_status status = 0;
+
+	if (ivpu_bo_is_resident(bo))
+		status |= DRM_GEM_OBJECT_RESIDENT;
+
+	return status;
+}
+
 static const struct drm_gem_object_funcs ivpu_gem_funcs = {
 	.free = ivpu_gem_bo_free,
 	.open = ivpu_gem_bo_open,
@@ -308,6 +354,7 @@ static const struct drm_gem_object_funcs ivpu_gem_funcs = {
 	.vmap = drm_gem_shmem_object_vmap,
 	.vunmap = drm_gem_shmem_object_vunmap,
 	.mmap = drm_gem_shmem_object_mmap,
+	.status = ivpu_gem_status,
 	.vm_ops = &drm_gem_shmem_vm_ops,
 };
 
@@ -320,25 +367,33 @@ int ivpu_bo_create_ioctl(struct drm_device *dev, void *data, struct drm_file *fi
 	struct ivpu_bo *bo;
 	int ret;
 
-	if (args->flags & ~DRM_IVPU_BO_FLAGS)
+	if (args->flags & ~DRM_IVPU_BO_FLAGS) {
+		ivpu_dbg(vdev, IOCTL, "Invalid BO flags 0x%x\n", args->flags);
 		return -EINVAL;
+	}
 
-	if (size == 0)
+	if (size == 0) {
+		ivpu_dbg(vdev, IOCTL, "Invalid BO size %llu\n", args->size);
 		return -EINVAL;
+	}
 
-	bo = ivpu_bo_alloc(vdev, size, args->flags, file_priv->ctx.id);
+	bo = ivpu_bo_alloc(vdev, size, args->flags);
 	if (IS_ERR(bo)) {
-		ivpu_err(vdev, "Failed to allocate BO: %pe (ctx %u size %llu flags 0x%x)",
+		ivpu_dbg(vdev, IOCTL, "Failed to allocate BO: %pe ctx %u size %llu flags 0x%x\n",
 			 bo, file_priv->ctx.id, args->size, args->flags);
 		return PTR_ERR(bo);
 	}
 
+	drm_WARN_ON(&vdev->drm, bo->base.base.handle_count != 0);
+
 	ret = drm_gem_handle_create(file, &bo->base.base, &args->handle);
-	if (ret)
-		ivpu_err(vdev, "Failed to create handle for BO: %pe (ctx %u size %llu flags 0x%x)",
+	if (ret) {
+		ivpu_dbg(vdev, IOCTL, "Failed to create handle for BO: %pe ctx %u size %llu flags 0x%x\n",
 			 bo, file_priv->ctx.id, args->size, args->flags);
-	else
+	} else {
 		args->vpu_addr = bo->vpu_addr;
+		drm_WARN_ON(&vdev->drm, bo->base.base.handle_count != 1);
+	}
 
 	drm_gem_object_put(&bo->base.base);
 
@@ -360,18 +415,21 @@ ivpu_bo_create(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
 	drm_WARN_ON(&vdev->drm, !PAGE_ALIGNED(range->end));
 	drm_WARN_ON(&vdev->drm, !PAGE_ALIGNED(size));
 
-	bo = ivpu_bo_alloc(vdev, size, flags, IVPU_GLOBAL_CONTEXT_MMU_SSID);
+	bo = ivpu_bo_alloc(vdev, size, flags);
 	if (IS_ERR(bo)) {
-		ivpu_err(vdev, "Failed to allocate BO: %pe (vpu_addr 0x%llx size %llu flags 0x%x)",
+		ivpu_err(vdev, "Failed to allocate BO: %pe vpu_addr 0x%llx size %llu flags 0x%x\n",
 			 bo, range->start, size, flags);
 		return NULL;
 	}
 
 	ret = ivpu_bo_alloc_vpu_addr(bo, ctx, range);
-	if (ret)
+	if (ret) {
+		ivpu_err(vdev, "Failed to allocate NPU address for BO: %pe ctx %u size %llu: %d\n",
+			 bo, ctx->id, size, ret);
 		goto err_put;
+	}
 
-	ret = ivpu_bo_pin(bo);
+	ret = ivpu_bo_bind(bo);
 	if (ret)
 		goto err_put;
 
@@ -391,6 +449,21 @@ err_put:
 	return NULL;
 }
 
+struct ivpu_bo *ivpu_bo_create_runtime(struct ivpu_device *vdev, u64 addr, u64 size, u32 flags)
+{
+	struct ivpu_addr_range range;
+
+	if (!ivpu_is_within_range(addr, size, &vdev->hw->ranges.runtime)) {
+		ivpu_err(vdev, "Invalid runtime BO address 0x%llx size %llu\n", addr, size);
+		return NULL;
+	}
+
+	if (ivpu_hw_range_init(vdev, &range, addr, size))
+		return NULL;
+
+	return ivpu_bo_create(vdev, &vdev->gctx, &range, size, flags);
+}
+
 struct ivpu_bo *ivpu_bo_create_global(struct ivpu_device *vdev, u64 size, u32 flags)
 {
 	return ivpu_bo_create(vdev, &vdev->gctx, &vdev->hw->ranges.global, size, flags);
diff --git a/drivers/accel/ivpu/ivpu_gem.h b/drivers/accel/ivpu/ivpu_gem.h
index aa8ff14f7aae..0c3350f22b55 100644
--- a/drivers/accel/ivpu/ivpu_gem.h
+++ b/drivers/accel/ivpu/ivpu_gem.h
@@ -1,6 +1,6 @@
 /* SPDX-License-Identifier: GPL-2.0-only */
 /*
- * Copyright (C) 2020-2023 Intel Corporation
+ * Copyright (C) 2020-2025 Intel Corporation
  */
 #ifndef __IVPU_GEM_H__
 #define __IVPU_GEM_H__
@@ -24,19 +24,22 @@ struct ivpu_bo {
 	bool mmu_mapped;
 };
 
-int ivpu_bo_pin(struct ivpu_bo *bo);
+int ivpu_bo_bind(struct ivpu_bo *bo);
 void ivpu_bo_unbind_all_bos_from_context(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx);
 
 struct drm_gem_object *ivpu_gem_create_object(struct drm_device *dev, size_t size);
 struct drm_gem_object *ivpu_gem_prime_import(struct drm_device *dev, struct dma_buf *dma_buf);
 struct ivpu_bo *ivpu_bo_create(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
 			       struct ivpu_addr_range *range, u64 size, u32 flags);
+struct ivpu_bo *ivpu_bo_create_runtime(struct ivpu_device *vdev, u64 addr, u64 size, u32 flags);
 struct ivpu_bo *ivpu_bo_create_global(struct ivpu_device *vdev, u64 size, u32 flags);
 void ivpu_bo_free(struct ivpu_bo *bo);
 
 int ivpu_bo_create_ioctl(struct drm_device *dev, void *data, struct drm_file *file);
 int ivpu_bo_info_ioctl(struct drm_device *dev, void *data, struct drm_file *file);
 int ivpu_bo_wait_ioctl(struct drm_device *dev, void *data, struct drm_file *file);
+int ivpu_bo_create_from_userptr_ioctl(struct drm_device *dev, void *data,
+				      struct drm_file *file);
 
 void ivpu_bo_list(struct drm_device *dev, struct drm_printer *p);
 void ivpu_bo_list_print(struct drm_device *dev);
@@ -74,6 +77,16 @@ static inline bool ivpu_bo_is_snooped(struct ivpu_bo *bo)
 	return ivpu_bo_cache_mode(bo) == DRM_IVPU_BO_CACHED;
 }
 
+static inline bool ivpu_bo_is_read_only(struct ivpu_bo *bo)
+{
+	return bo->flags & DRM_IVPU_BO_READ_ONLY;
+}
+
+static inline bool ivpu_bo_is_resident(struct ivpu_bo *bo)
+{
+	return !!bo->base.pages;
+}
+
 static inline void *ivpu_to_cpu_addr(struct ivpu_bo *bo, u32 vpu_addr)
 {
 	if (vpu_addr < bo->vpu_addr)
@@ -96,4 +109,9 @@ static inline u32 cpu_to_vpu_addr(struct ivpu_bo *bo, void *cpu_addr)
 	return bo->vpu_addr + (cpu_addr - ivpu_bo_vaddr(bo));
 }
 
+static inline bool ivpu_bo_is_mappable(struct ivpu_bo *bo)
+{
+	return bo->flags & DRM_IVPU_BO_MAPPABLE;
+}
+
 #endif /* __IVPU_GEM_H__ */
diff --git a/drivers/accel/ivpu/ivpu_gem_userptr.c b/drivers/accel/ivpu/ivpu_gem_userptr.c
new file mode 100644
index 000000000000..25ba606164c0
--- /dev/null
+++ b/drivers/accel/ivpu/ivpu_gem_userptr.c
@@ -0,0 +1,213 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2020-2025 Intel Corporation
+ */
+
+#include <linux/dma-buf.h>
+#include <linux/err.h>
+#include <linux/highmem.h>
+#include <linux/mm.h>
+#include <linux/mman.h>
+#include <linux/scatterlist.h>
+#include <linux/slab.h>
+#include <linux/capability.h>
+
+#include <drm/drm_device.h>
+#include <drm/drm_file.h>
+#include <drm/drm_gem.h>
+
+#include "ivpu_drv.h"
+#include "ivpu_gem.h"
+
+static struct sg_table *
+ivpu_gem_userptr_dmabuf_map(struct dma_buf_attachment *attachment,
+			    enum dma_data_direction direction)
+{
+	struct sg_table *sgt = attachment->dmabuf->priv;
+	int ret;
+
+	ret = dma_map_sgtable(attachment->dev, sgt, direction, DMA_ATTR_SKIP_CPU_SYNC);
+	if (ret)
+		return ERR_PTR(ret);
+
+	return sgt;
+}
+
+static void ivpu_gem_userptr_dmabuf_unmap(struct dma_buf_attachment *attachment,
+					  struct sg_table *sgt,
+					  enum dma_data_direction direction)
+{
+	dma_unmap_sgtable(attachment->dev, sgt, direction, DMA_ATTR_SKIP_CPU_SYNC);
+}
+
+static void ivpu_gem_userptr_dmabuf_release(struct dma_buf *dma_buf)
+{
+	struct sg_table *sgt = dma_buf->priv;
+	struct sg_page_iter page_iter;
+	struct page *page;
+
+	for_each_sgtable_page(sgt, &page_iter, 0) {
+		page = sg_page_iter_page(&page_iter);
+		unpin_user_page(page);
+	}
+
+	sg_free_table(sgt);
+	kfree(sgt);
+}
+
+static const struct dma_buf_ops ivpu_gem_userptr_dmabuf_ops = {
+	.map_dma_buf = ivpu_gem_userptr_dmabuf_map,
+	.unmap_dma_buf = ivpu_gem_userptr_dmabuf_unmap,
+	.release = ivpu_gem_userptr_dmabuf_release,
+};
+
+static struct dma_buf *
+ivpu_create_userptr_dmabuf(struct ivpu_device *vdev, void __user *user_ptr,
+			   size_t size, uint32_t flags)
+{
+	struct dma_buf_export_info exp_info = {};
+	struct dma_buf *dma_buf;
+	struct sg_table *sgt;
+	struct page **pages;
+	unsigned long nr_pages = size >> PAGE_SHIFT;
+	unsigned int gup_flags = FOLL_LONGTERM;
+	int ret, i, pinned;
+
+	/* Add FOLL_WRITE only if the BO is not read-only */
+	if (!(flags & DRM_IVPU_BO_READ_ONLY))
+		gup_flags |= FOLL_WRITE;
+
+	pages = kvmalloc_array(nr_pages, sizeof(*pages), GFP_KERNEL);
+	if (!pages)
+		return ERR_PTR(-ENOMEM);
+
+	pinned = pin_user_pages_fast((unsigned long)user_ptr, nr_pages, gup_flags, pages);
+	if (pinned < 0) {
+		ret = pinned;
+		ivpu_dbg(vdev, IOCTL, "Failed to pin user pages: %d\n", ret);
+		goto free_pages_array;
+	}
+
+	if (pinned != nr_pages) {
+		ivpu_dbg(vdev, IOCTL, "Pinned %d pages, expected %lu\n", pinned, nr_pages);
+		ret = -EFAULT;
+		goto unpin_pages;
+	}
+
+	sgt = kmalloc(sizeof(*sgt), GFP_KERNEL);
+	if (!sgt) {
+		ret = -ENOMEM;
+		goto unpin_pages;
+	}
+
+	ret = sg_alloc_table_from_pages(sgt, pages, nr_pages, 0, size, GFP_KERNEL);
+	if (ret) {
+		ivpu_dbg(vdev, IOCTL, "Failed to create sg table: %d\n", ret);
+		goto free_sgt;
+	}
+
+	exp_info.exp_name = "ivpu_userptr_dmabuf";
+	exp_info.owner = THIS_MODULE;
+	exp_info.ops = &ivpu_gem_userptr_dmabuf_ops;
+	exp_info.size = size;
+	exp_info.flags = O_RDWR | O_CLOEXEC;
+	exp_info.priv = sgt;
+
+	dma_buf = dma_buf_export(&exp_info);
+	if (IS_ERR(dma_buf)) {
+		ret = PTR_ERR(dma_buf);
+		ivpu_dbg(vdev, IOCTL, "Failed to export userptr dma-buf: %d\n", ret);
+		goto free_sg_table;
+	}
+
+	kvfree(pages);
+	return dma_buf;
+
+free_sg_table:
+	sg_free_table(sgt);
+free_sgt:
+	kfree(sgt);
+unpin_pages:
+	for (i = 0; i < pinned; i++)
+		unpin_user_page(pages[i]);
+free_pages_array:
+	kvfree(pages);
+	return ERR_PTR(ret);
+}
+
+static struct ivpu_bo *
+ivpu_bo_create_from_userptr(struct ivpu_device *vdev, void __user *user_ptr,
+			    size_t size, uint32_t flags)
+{
+	struct dma_buf *dma_buf;
+	struct drm_gem_object *obj;
+	struct ivpu_bo *bo;
+
+	dma_buf = ivpu_create_userptr_dmabuf(vdev, user_ptr, size, flags);
+	if (IS_ERR(dma_buf))
+		return ERR_CAST(dma_buf);
+
+	obj = ivpu_gem_prime_import(&vdev->drm, dma_buf);
+	if (IS_ERR(obj)) {
+		dma_buf_put(dma_buf);
+		return ERR_CAST(obj);
+	}
+
+	dma_buf_put(dma_buf);
+
+	bo = to_ivpu_bo(obj);
+	bo->flags = flags;
+
+	return bo;
+}
+
+int ivpu_bo_create_from_userptr_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
+{
+	struct drm_ivpu_bo_create_from_userptr *args = data;
+	struct ivpu_file_priv *file_priv = file->driver_priv;
+	struct ivpu_device *vdev = to_ivpu_device(dev);
+	void __user *user_ptr = u64_to_user_ptr(args->user_ptr);
+	struct ivpu_bo *bo;
+	int ret;
+
+	if (args->flags & ~(DRM_IVPU_BO_HIGH_MEM | DRM_IVPU_BO_DMA_MEM | DRM_IVPU_BO_READ_ONLY)) {
+		ivpu_dbg(vdev, IOCTL, "Invalid BO flags: 0x%x\n", args->flags);
+		return -EINVAL;
+	}
+
+	if (!args->user_ptr || !args->size) {
+		ivpu_dbg(vdev, IOCTL, "Userptr or size are zero: ptr %llx size %llu\n",
+			 args->user_ptr, args->size);
+		return -EINVAL;
+	}
+
+	if (!PAGE_ALIGNED(args->user_ptr) || !PAGE_ALIGNED(args->size)) {
+		ivpu_dbg(vdev, IOCTL, "Userptr or size not page aligned: ptr %llx size %llu\n",
+			 args->user_ptr, args->size);
+		return -EINVAL;
+	}
+
+	if (!access_ok(user_ptr, args->size)) {
+		ivpu_dbg(vdev, IOCTL, "Userptr is not accessible: ptr %llx size %llu\n",
+			 args->user_ptr, args->size);
+		return -EFAULT;
+	}
+
+	bo = ivpu_bo_create_from_userptr(vdev, user_ptr, args->size, args->flags);
+	if (IS_ERR(bo))
+		return PTR_ERR(bo);
+
+	ret = drm_gem_handle_create(file, &bo->base.base, &args->handle);
+	if (ret) {
+		ivpu_dbg(vdev, IOCTL, "Failed to create handle for BO: %pe ctx %u size %llu flags 0x%x\n",
+			 bo, file_priv->ctx.id, args->size, args->flags);
+	} else {
+		ivpu_dbg(vdev, BO, "Created userptr BO: handle=%u vpu_addr=0x%llx size=%llu flags=0x%x\n",
+			 args->handle, bo->vpu_addr, args->size, bo->flags);
+		args->vpu_addr = bo->vpu_addr;
+	}
+
+	drm_gem_object_put(&bo->base.base);
+
+	return ret;
+}
diff --git a/drivers/accel/ivpu/ivpu_hw.c b/drivers/accel/ivpu/ivpu_hw.c
index 633160470c93..d69cd0d93569 100644
--- a/drivers/accel/ivpu/ivpu_hw.c
+++ b/drivers/accel/ivpu/ivpu_hw.c
@@ -8,6 +8,8 @@
 #include "ivpu_hw_btrs.h"
 #include "ivpu_hw_ip.h"
 
+#include <asm/msr-index.h>
+#include <asm/msr.h>
 #include <linux/dmi.h>
 #include <linux/fault-inject.h>
 #include <linux/pm_runtime.h>
@@ -20,6 +22,10 @@ module_param_named_unsafe(fail_hw, ivpu_fail_hw, charp, 0444);
 MODULE_PARM_DESC(fail_hw, "<interval>,<probability>,<space>,<times>");
 #endif
 
+#define FW_SHARED_MEM_ALIGNMENT	SZ_512K /* VPU MTRR limitation */
+
+#define ECC_MCA_SIGNAL_ENABLE_MASK	0xff
+
 static char *platform_to_str(u32 platform)
 {
 	switch (platform) {
@@ -94,12 +100,14 @@ static void timeouts_init(struct ivpu_device *vdev)
 		vdev->timeout.boot = -1;
 		vdev->timeout.jsm = -1;
 		vdev->timeout.tdr = -1;
+		vdev->timeout.inference = -1;
 		vdev->timeout.autosuspend = -1;
 		vdev->timeout.d0i3_entry_msg = -1;
 	} else if (ivpu_is_fpga(vdev)) {
 		vdev->timeout.boot = 50;
 		vdev->timeout.jsm = 15000;
 		vdev->timeout.tdr = 30000;
+		vdev->timeout.inference = 900000;
 		vdev->timeout.autosuspend = -1;
 		vdev->timeout.d0i3_entry_msg = 500;
 		vdev->timeout.state_dump_msg = 10000;
@@ -107,6 +115,7 @@ static void timeouts_init(struct ivpu_device *vdev)
 		vdev->timeout.boot = 50;
 		vdev->timeout.jsm = 500;
 		vdev->timeout.tdr = 10000;
+		vdev->timeout.inference = 300000;
 		vdev->timeout.autosuspend = 100;
 		vdev->timeout.d0i3_entry_msg = 100;
 		vdev->timeout.state_dump_msg = 10;
@@ -114,6 +123,7 @@ static void timeouts_init(struct ivpu_device *vdev)
 		vdev->timeout.boot = 1000;
 		vdev->timeout.jsm = 500;
 		vdev->timeout.tdr = 2000;
+		vdev->timeout.inference = 60000;
 		if (ivpu_hw_ip_gen(vdev) == IVPU_HW_IP_37XX)
 			vdev->timeout.autosuspend = 10;
 		else
@@ -143,19 +153,39 @@ static void priority_bands_init(struct ivpu_device *vdev)
 	vdev->hw->hws.process_quantum[VPU_JOB_SCHEDULING_PRIORITY_BAND_REALTIME] = 200000;
 }
 
+int ivpu_hw_range_init(struct ivpu_device *vdev, struct ivpu_addr_range *range, u64 start, u64 size)
+{
+	u64 end;
+
+	if (!range || check_add_overflow(start, size, &end)) {
+		ivpu_err(vdev, "Invalid range: start 0x%llx size %llu\n", start, size);
+		return -EINVAL;
+	}
+
+	range->start = start;
+	range->end = end;
+
+	return 0;
+}
+
 static void memory_ranges_init(struct ivpu_device *vdev)
 {
 	if (ivpu_hw_ip_gen(vdev) == IVPU_HW_IP_37XX) {
-		ivpu_hw_range_init(&vdev->hw->ranges.global, 0x80000000, SZ_512M);
-		ivpu_hw_range_init(&vdev->hw->ranges.user,   0x88000000, 511 * SZ_1M);
-		ivpu_hw_range_init(&vdev->hw->ranges.shave, 0x180000000, SZ_2G);
-		ivpu_hw_range_init(&vdev->hw->ranges.dma,   0x200000000, SZ_128G);
+		ivpu_hw_range_init(vdev, &vdev->hw->ranges.runtime, 0x84800000, SZ_64M);
+		ivpu_hw_range_init(vdev, &vdev->hw->ranges.global,  0x90000000, SZ_256M);
+		ivpu_hw_range_init(vdev, &vdev->hw->ranges.user,    0xa0000000, 511 * SZ_1M);
+		ivpu_hw_range_init(vdev, &vdev->hw->ranges.shave,  0x180000000, SZ_2G);
+		ivpu_hw_range_init(vdev, &vdev->hw->ranges.dma,    0x200000000, SZ_128G);
 	} else {
-		ivpu_hw_range_init(&vdev->hw->ranges.global, 0x80000000, SZ_512M);
-		ivpu_hw_range_init(&vdev->hw->ranges.shave,  0x80000000, SZ_2G);
-		ivpu_hw_range_init(&vdev->hw->ranges.user,  0x100000000, SZ_256G);
+		ivpu_hw_range_init(vdev, &vdev->hw->ranges.runtime, 0x80000000, SZ_64M);
+		ivpu_hw_range_init(vdev, &vdev->hw->ranges.global,  0x90000000, SZ_256M);
+		ivpu_hw_range_init(vdev, &vdev->hw->ranges.shave,   0x80000000, SZ_2G);
+		ivpu_hw_range_init(vdev, &vdev->hw->ranges.user,   0x100000000, SZ_256G);
 		vdev->hw->ranges.dma = vdev->hw->ranges.user;
 	}
+
+	drm_WARN_ON(&vdev->drm, !IS_ALIGNED(vdev->hw->ranges.global.start,
+					    FW_SHARED_MEM_ALIGNMENT));
 }
 
 static int wp_enable(struct ivpu_device *vdev)
@@ -369,3 +399,22 @@ irqreturn_t ivpu_hw_irq_handler(int irq, void *ptr)
 	pm_runtime_mark_last_busy(vdev->drm.dev);
 	return IRQ_HANDLED;
 }
+
+bool ivpu_hw_uses_ecc_mca_signal(struct ivpu_device *vdev)
+{
+	unsigned long long msr_integrity_caps;
+	int ret;
+
+	if (ivpu_hw_ip_gen(vdev) < IVPU_HW_IP_50XX)
+		return false;
+
+	ret = rdmsrq_safe(MSR_INTEGRITY_CAPS, &msr_integrity_caps);
+	if (ret) {
+		ivpu_warn(vdev, "Error reading MSR_INTEGRITY_CAPS: %d", ret);
+		return false;
+	}
+
+	ivpu_dbg(vdev, MISC, "MSR_INTEGRITY_CAPS: 0x%llx\n", msr_integrity_caps);
+
+	return msr_integrity_caps & ECC_MCA_SIGNAL_ENABLE_MASK;
+}
diff --git a/drivers/accel/ivpu/ivpu_hw.h b/drivers/accel/ivpu/ivpu_hw.h
index d79668fe1609..b6d0f0d0dccc 100644
--- a/drivers/accel/ivpu/ivpu_hw.h
+++ b/drivers/accel/ivpu/ivpu_hw.h
@@ -21,6 +21,7 @@ struct ivpu_hw_info {
 		bool (*ip_irq_handler)(struct ivpu_device *vdev, int irq);
 	} irq;
 	struct {
+		struct ivpu_addr_range runtime;
 		struct ivpu_addr_range global;
 		struct ivpu_addr_range user;
 		struct ivpu_addr_range shave;
@@ -51,6 +52,8 @@ struct ivpu_hw_info {
 };
 
 int ivpu_hw_init(struct ivpu_device *vdev);
+int ivpu_hw_range_init(struct ivpu_device *vdev, struct ivpu_addr_range *range, u64 start,
+		       u64 size);
 int ivpu_hw_power_up(struct ivpu_device *vdev);
 int ivpu_hw_power_down(struct ivpu_device *vdev);
 int ivpu_hw_reset(struct ivpu_device *vdev);
@@ -60,6 +63,7 @@ void ivpu_irq_handlers_init(struct ivpu_device *vdev);
 void ivpu_hw_irq_enable(struct ivpu_device *vdev);
 void ivpu_hw_irq_disable(struct ivpu_device *vdev);
 irqreturn_t ivpu_hw_irq_handler(int irq, void *ptr);
+bool ivpu_hw_uses_ecc_mca_signal(struct ivpu_device *vdev);
 
 static inline u32 ivpu_hw_btrs_irq_handler(struct ivpu_device *vdev, int irq)
 {
@@ -71,12 +75,6 @@ static inline u32 ivpu_hw_ip_irq_handler(struct ivpu_device *vdev, int irq)
 	return vdev->hw->irq.ip_irq_handler(vdev, irq);
 }
 
-static inline void ivpu_hw_range_init(struct ivpu_addr_range *range, u64 start, u64 size)
-{
-	range->start = start;
-	range->end = start + size;
-}
-
 static inline u64 ivpu_hw_range_size(const struct ivpu_addr_range *range)
 {
 	return range->end - range->start;
diff --git a/drivers/accel/ivpu/ivpu_hw_btrs.c b/drivers/accel/ivpu/ivpu_hw_btrs.c
index b236c7234daa..06e65c592618 100644
--- a/drivers/accel/ivpu/ivpu_hw_btrs.c
+++ b/drivers/accel/ivpu/ivpu_hw_btrs.c
@@ -33,7 +33,6 @@
 
 #define PLL_CDYN_DEFAULT               0x80
 #define PLL_EPP_DEFAULT                0x80
-#define PLL_CONFIG_DEFAULT             0x0
 #define PLL_REF_CLK_FREQ               50000000ull
 #define PLL_RATIO_TO_FREQ(x)           ((x) * PLL_REF_CLK_FREQ)
 
@@ -303,7 +302,7 @@ static void prepare_wp_request(struct ivpu_device *vdev, struct wp_request *wp,
 		wp->epp = 0;
 	} else {
 		wp->target = hw->pll.pn_ratio;
-		wp->cfg = enable ? PLL_CONFIG_DEFAULT : 0;
+		wp->cfg = 0;
 		wp->cdyn = enable ? PLL_CDYN_DEFAULT : 0;
 		wp->epp = enable ? PLL_EPP_DEFAULT : 0;
 	}
@@ -322,6 +321,14 @@ static int wait_for_pll_lock(struct ivpu_device *vdev, bool enable)
 	return REGB_POLL_FLD(VPU_HW_BTRS_MTL_PLL_STATUS, LOCK, exp_val, PLL_TIMEOUT_US);
 }
 
+static int wait_for_cdyn_deassert(struct ivpu_device *vdev)
+{
+	if (ivpu_hw_btrs_gen(vdev) == IVPU_HW_BTRS_MTL)
+		return 0;
+
+	return REGB_POLL_FLD(VPU_HW_BTRS_LNL_CDYN, CDYN, 0, PLL_TIMEOUT_US);
+}
+
 int ivpu_hw_btrs_wp_drive(struct ivpu_device *vdev, bool enable)
 {
 	struct wp_request wp;
@@ -355,6 +362,14 @@ int ivpu_hw_btrs_wp_drive(struct ivpu_device *vdev, bool enable)
 		return ret;
 	}
 
+	if (!enable) {
+		ret = wait_for_cdyn_deassert(vdev);
+		if (ret) {
+			ivpu_err(vdev, "Timed out waiting for CDYN deassert\n");
+			return ret;
+		}
+	}
+
 	return 0;
 }
 
@@ -674,7 +689,7 @@ bool ivpu_hw_btrs_irq_handler_lnl(struct ivpu_device *vdev, int irq)
 
 	if (REG_TEST_FLD(VPU_HW_BTRS_LNL_INTERRUPT_STAT, SURV_ERR, status)) {
 		ivpu_dbg(vdev, IRQ, "Survivability IRQ\n");
-		queue_work(system_wq, &vdev->irq_dct_work);
+		queue_work(system_percpu_wq, &vdev->irq_dct_work);
 	}
 
 	if (REG_TEST_FLD(VPU_HW_BTRS_LNL_INTERRUPT_STAT, FREQ_CHANGE, status)) {
@@ -753,7 +768,7 @@ int ivpu_hw_btrs_dct_get_request(struct ivpu_device *vdev, bool *enable)
 	}
 }
 
-void ivpu_hw_btrs_dct_set_status(struct ivpu_device *vdev, bool enable, u32 active_percent)
+void ivpu_hw_btrs_dct_set_status(struct ivpu_device *vdev, bool enable, u8 active_percent)
 {
 	u32 val = 0;
 	u32 cmd = enable ? DCT_ENABLE : DCT_DISABLE;
diff --git a/drivers/accel/ivpu/ivpu_hw_btrs.h b/drivers/accel/ivpu/ivpu_hw_btrs.h
index d2d82651976d..c4c10e22f30f 100644
--- a/drivers/accel/ivpu/ivpu_hw_btrs.h
+++ b/drivers/accel/ivpu/ivpu_hw_btrs.h
@@ -36,7 +36,7 @@ u32 ivpu_hw_btrs_dpu_freq_get(struct ivpu_device *vdev);
 bool ivpu_hw_btrs_irq_handler_mtl(struct ivpu_device *vdev, int irq);
 bool ivpu_hw_btrs_irq_handler_lnl(struct ivpu_device *vdev, int irq);
 int ivpu_hw_btrs_dct_get_request(struct ivpu_device *vdev, bool *enable);
-void ivpu_hw_btrs_dct_set_status(struct ivpu_device *vdev, bool enable, u32 dct_percent);
+void ivpu_hw_btrs_dct_set_status(struct ivpu_device *vdev, bool enable, u8 active_percent);
 u32 ivpu_hw_btrs_telemetry_offset_get(struct ivpu_device *vdev);
 u32 ivpu_hw_btrs_telemetry_size_get(struct ivpu_device *vdev);
 u32 ivpu_hw_btrs_telemetry_enable_get(struct ivpu_device *vdev);
diff --git a/drivers/accel/ivpu/ivpu_hw_btrs_lnl_reg.h b/drivers/accel/ivpu/ivpu_hw_btrs_lnl_reg.h
index fff2ef2cada6..a81a9ba540fa 100644
--- a/drivers/accel/ivpu/ivpu_hw_btrs_lnl_reg.h
+++ b/drivers/accel/ivpu/ivpu_hw_btrs_lnl_reg.h
@@ -74,6 +74,9 @@
 #define VPU_HW_BTRS_LNL_PLL_FREQ				0x00000148u
 #define VPU_HW_BTRS_LNL_PLL_FREQ_RATIO_MASK			GENMASK(15, 0)
 
+#define VPU_HW_BTRS_LNL_CDYN					0x0000014cu
+#define VPU_HW_BTRS_LNL_CDYN_CDYN_MASK				GENMASK(15, 0)
+
 #define VPU_HW_BTRS_LNL_TILE_FUSE				0x00000150u
 #define VPU_HW_BTRS_LNL_TILE_FUSE_VALID_MASK			BIT_MASK(0)
 #define VPU_HW_BTRS_LNL_TILE_FUSE_CONFIG_MASK			GENMASK(6, 1)
diff --git a/drivers/accel/ivpu/ivpu_hw_ip.c b/drivers/accel/ivpu/ivpu_hw_ip.c
index 823f6a57dc54..06aa1e7dc50b 100644
--- a/drivers/accel/ivpu/ivpu_hw_ip.c
+++ b/drivers/accel/ivpu/ivpu_hw_ip.c
@@ -683,6 +683,7 @@ static void pwr_island_delay_set(struct ivpu_device *vdev)
 		return;
 
 	switch (ivpu_device_id(vdev)) {
+	case PCI_DEVICE_ID_WCL:
 	case PCI_DEVICE_ID_PTL_P:
 		post = high ? 18 : 0;
 		post1 = 0;
@@ -690,6 +691,13 @@ static void pwr_island_delay_set(struct ivpu_device *vdev)
 		status = high ? 46 : 3;
 		break;
 
+	case PCI_DEVICE_ID_NVL:
+		post = high ? 198 : 17;
+		post1 = 0;
+		post2 = high ? 198 : 17;
+		status = 0;
+		break;
+
 	default:
 		dump_stack();
 		ivpu_err(vdev, "Unknown device ID\n");
@@ -888,6 +896,9 @@ static int soc_cpu_drive_40xx(struct ivpu_device *vdev, bool enable)
 
 static int soc_cpu_enable(struct ivpu_device *vdev)
 {
+	if (ivpu_hw_ip_gen(vdev) >= IVPU_HW_IP_60XX)
+		return 0;
+
 	return soc_cpu_drive_40xx(vdev, true);
 }
 
diff --git a/drivers/accel/ivpu/ivpu_ipc.c b/drivers/accel/ivpu/ivpu_ipc.c
index 39f83225c181..1f13bf95b2b3 100644
--- a/drivers/accel/ivpu/ivpu_ipc.c
+++ b/drivers/accel/ivpu/ivpu_ipc.c
@@ -141,7 +141,6 @@ ivpu_ipc_rx_msg_add(struct ivpu_device *vdev, struct ivpu_ipc_consumer *cons,
 	struct ivpu_ipc_rx_msg *rx_msg;
 
 	lockdep_assert_held(&ipc->cons_lock);
-	lockdep_assert_irqs_disabled();
 
 	rx_msg = kzalloc(sizeof(*rx_msg), GFP_ATOMIC);
 	if (!rx_msg) {
@@ -460,7 +459,7 @@ void ivpu_ipc_irq_handler(struct ivpu_device *vdev)
 		}
 	}
 
-	queue_work(system_wq, &vdev->irq_ipc_work);
+	queue_work(system_percpu_wq, &vdev->irq_ipc_work);
 }
 
 void ivpu_ipc_irq_work_fn(struct work_struct *work)
diff --git a/drivers/accel/ivpu/ivpu_job.c b/drivers/accel/ivpu/ivpu_job.c
index fae8351aa330..4f8564e2878a 100644
--- a/drivers/accel/ivpu/ivpu_job.c
+++ b/drivers/accel/ivpu/ivpu_job.c
@@ -1,6 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0-only
 /*
- * Copyright (C) 2020-2024 Intel Corporation
+ * Copyright (C) 2020-2025 Intel Corporation
  */
 
 #include <drm/drm_file.h>
@@ -34,22 +34,20 @@ static void ivpu_cmdq_ring_db(struct ivpu_device *vdev, struct ivpu_cmdq *cmdq)
 static int ivpu_preemption_buffers_create(struct ivpu_device *vdev,
 					  struct ivpu_file_priv *file_priv, struct ivpu_cmdq *cmdq)
 {
-	u64 primary_size = ALIGN(vdev->fw->primary_preempt_buf_size, PAGE_SIZE);
-	u64 secondary_size = ALIGN(vdev->fw->secondary_preempt_buf_size, PAGE_SIZE);
-
-	if (vdev->fw->sched_mode != VPU_SCHEDULING_MODE_HW ||
-	    ivpu_test_mode & IVPU_TEST_MODE_MIP_DISABLE)
+	if (ivpu_fw_preempt_buf_size(vdev) == 0)
 		return 0;
 
 	cmdq->primary_preempt_buf = ivpu_bo_create(vdev, &file_priv->ctx, &vdev->hw->ranges.user,
-						   primary_size, DRM_IVPU_BO_WC);
+						   vdev->fw->primary_preempt_buf_size,
+						   DRM_IVPU_BO_WC);
 	if (!cmdq->primary_preempt_buf) {
 		ivpu_err(vdev, "Failed to create primary preemption buffer\n");
 		return -ENOMEM;
 	}
 
 	cmdq->secondary_preempt_buf = ivpu_bo_create(vdev, &file_priv->ctx, &vdev->hw->ranges.dma,
-						     secondary_size, DRM_IVPU_BO_WC);
+						     vdev->fw->secondary_preempt_buf_size,
+						     DRM_IVPU_BO_WC);
 	if (!cmdq->secondary_preempt_buf) {
 		ivpu_err(vdev, "Failed to create secondary preemption buffer\n");
 		goto err_free_primary;
@@ -66,20 +64,39 @@ err_free_primary:
 static void ivpu_preemption_buffers_free(struct ivpu_device *vdev,
 					 struct ivpu_file_priv *file_priv, struct ivpu_cmdq *cmdq)
 {
-	if (vdev->fw->sched_mode != VPU_SCHEDULING_MODE_HW)
-		return;
-
 	if (cmdq->primary_preempt_buf)
 		ivpu_bo_free(cmdq->primary_preempt_buf);
 	if (cmdq->secondary_preempt_buf)
 		ivpu_bo_free(cmdq->secondary_preempt_buf);
 }
 
+static int ivpu_preemption_job_init(struct ivpu_device *vdev, struct ivpu_file_priv *file_priv,
+				    struct ivpu_cmdq *cmdq, struct ivpu_job *job)
+{
+	int ret;
+
+	/* Use preemption buffer provided by the user space */
+	if (job->primary_preempt_buf)
+		return 0;
+
+	if (!cmdq->primary_preempt_buf) {
+		/* Allocate per command queue preemption buffers */
+		ret = ivpu_preemption_buffers_create(vdev, file_priv, cmdq);
+		if (ret)
+			return ret;
+	}
+
+	/* Use preemption buffers allocated by the kernel */
+	job->primary_preempt_buf = cmdq->primary_preempt_buf;
+	job->secondary_preempt_buf = cmdq->secondary_preempt_buf;
+
+	return 0;
+}
+
 static struct ivpu_cmdq *ivpu_cmdq_alloc(struct ivpu_file_priv *file_priv)
 {
 	struct ivpu_device *vdev = file_priv->vdev;
 	struct ivpu_cmdq *cmdq;
-	int ret;
 
 	cmdq = kzalloc(sizeof(*cmdq), GFP_KERNEL);
 	if (!cmdq)
@@ -89,10 +106,6 @@ static struct ivpu_cmdq *ivpu_cmdq_alloc(struct ivpu_file_priv *file_priv)
 	if (!cmdq->mem)
 		goto err_free_cmdq;
 
-	ret = ivpu_preemption_buffers_create(vdev, file_priv, cmdq);
-	if (ret)
-		ivpu_warn(vdev, "Failed to allocate preemption buffers, preemption limited\n");
-
 	return cmdq;
 
 err_free_cmdq:
@@ -100,6 +113,43 @@ err_free_cmdq:
 	return NULL;
 }
 
+/**
+ * ivpu_cmdq_get_entry_count - Calculate the number of entries in the command queue.
+ * @cmdq: Pointer to the command queue structure.
+ *
+ * Returns the number of entries that can fit in the command queue memory.
+ */
+static inline u32 ivpu_cmdq_get_entry_count(struct ivpu_cmdq *cmdq)
+{
+	size_t size = ivpu_bo_size(cmdq->mem) - sizeof(struct vpu_job_queue_header);
+
+	return size / sizeof(struct vpu_job_queue_entry);
+}
+
+/**
+ * ivpu_cmdq_get_flags - Get command queue flags based on input flags and test mode.
+ * @vdev: Pointer to the ivpu device structure.
+ * @flags: Input flags to determine the command queue flags.
+ *
+ * Returns the calculated command queue flags, considering both the input flags
+ * and the current test mode settings.
+ */
+static u32 ivpu_cmdq_get_flags(struct ivpu_device *vdev, u32 flags)
+{
+	u32 cmdq_flags = 0;
+
+	if ((flags & DRM_IVPU_CMDQ_FLAG_TURBO) && (ivpu_hw_ip_gen(vdev) >= IVPU_HW_IP_40XX))
+		cmdq_flags |= VPU_JOB_QUEUE_FLAGS_TURBO_MODE;
+
+	/* Test mode can override the TURBO flag coming from the application */
+	if (ivpu_test_mode & IVPU_TEST_MODE_TURBO_ENABLE)
+		cmdq_flags |= VPU_JOB_QUEUE_FLAGS_TURBO_MODE;
+	if (ivpu_test_mode & IVPU_TEST_MODE_TURBO_DISABLE)
+		cmdq_flags &= ~VPU_JOB_QUEUE_FLAGS_TURBO_MODE;
+
+	return cmdq_flags;
+}
+
 static void ivpu_cmdq_free(struct ivpu_file_priv *file_priv, struct ivpu_cmdq *cmdq)
 {
 	ivpu_preemption_buffers_free(file_priv->vdev, file_priv, cmdq);
@@ -107,8 +157,7 @@ static void ivpu_cmdq_free(struct ivpu_file_priv *file_priv, struct ivpu_cmdq *c
 	kfree(cmdq);
 }
 
-static struct ivpu_cmdq *ivpu_cmdq_create(struct ivpu_file_priv *file_priv, u8 priority,
-					  bool is_legacy)
+static struct ivpu_cmdq *ivpu_cmdq_create(struct ivpu_file_priv *file_priv, u8 priority, u32 flags)
 {
 	struct ivpu_device *vdev = file_priv->vdev;
 	struct ivpu_cmdq *cmdq = NULL;
@@ -121,10 +170,6 @@ static struct ivpu_cmdq *ivpu_cmdq_create(struct ivpu_file_priv *file_priv, u8 p
 		ivpu_err(vdev, "Failed to allocate command queue\n");
 		return NULL;
 	}
-
-	cmdq->priority = priority;
-	cmdq->is_legacy = is_legacy;
-
 	ret = xa_alloc_cyclic(&file_priv->cmdq_xa, &cmdq->id, cmdq, file_priv->cmdq_limit,
 			      &file_priv->cmdq_id_next, GFP_KERNEL);
 	if (ret < 0) {
@@ -132,7 +177,15 @@ static struct ivpu_cmdq *ivpu_cmdq_create(struct ivpu_file_priv *file_priv, u8 p
 		goto err_free_cmdq;
 	}
 
-	ivpu_dbg(vdev, JOB, "Command queue %d created, ctx %d\n", cmdq->id, file_priv->ctx.id);
+	cmdq->entry_count = ivpu_cmdq_get_entry_count(cmdq);
+	cmdq->priority = priority;
+
+	cmdq->jobq = (struct vpu_job_queue *)ivpu_bo_vaddr(cmdq->mem);
+	cmdq->jobq->header.engine_idx = VPU_ENGINE_COMPUTE;
+	cmdq->jobq->header.flags = ivpu_cmdq_get_flags(vdev, flags);
+
+	ivpu_dbg(vdev, JOB, "Command queue %d created, ctx %d, flags 0x%08x\n",
+		 cmdq->id, file_priv->ctx.id, cmdq->jobq->header.flags);
 	return cmdq;
 
 err_free_cmdq:
@@ -179,36 +232,25 @@ static int ivpu_register_db(struct ivpu_file_priv *file_priv, struct ivpu_cmdq *
 		ret = ivpu_jsm_register_db(vdev, file_priv->ctx.id, cmdq->db_id,
 					   cmdq->mem->vpu_addr, ivpu_bo_size(cmdq->mem));
 
-	if (!ret)
+	if (!ret) {
 		ivpu_dbg(vdev, JOB, "DB %d registered to cmdq %d ctx %d priority %d\n",
 			 cmdq->db_id, cmdq->id, file_priv->ctx.id, cmdq->priority);
-	else
+	} else {
 		xa_erase(&vdev->db_xa, cmdq->db_id);
+		cmdq->db_id = 0;
+	}
 
 	return ret;
 }
 
-static void ivpu_cmdq_jobq_init(struct ivpu_device *vdev, struct vpu_job_queue *jobq)
+static void ivpu_cmdq_jobq_reset(struct ivpu_device *vdev, struct vpu_job_queue *jobq)
 {
-	jobq->header.engine_idx = VPU_ENGINE_COMPUTE;
 	jobq->header.head = 0;
 	jobq->header.tail = 0;
 
-	if (ivpu_test_mode & IVPU_TEST_MODE_TURBO) {
-		ivpu_dbg(vdev, JOB, "Turbo mode enabled");
-		jobq->header.flags = VPU_JOB_QUEUE_FLAGS_TURBO_MODE;
-	}
-
 	wmb(); /* Flush WC buffer for jobq->header */
 }
 
-static inline u32 ivpu_cmdq_get_entry_count(struct ivpu_cmdq *cmdq)
-{
-	size_t size = ivpu_bo_size(cmdq->mem) - sizeof(struct vpu_job_queue_header);
-
-	return size / sizeof(struct vpu_job_queue_entry);
-}
-
 static int ivpu_cmdq_register(struct ivpu_file_priv *file_priv, struct ivpu_cmdq *cmdq)
 {
 	struct ivpu_device *vdev = file_priv->vdev;
@@ -219,10 +261,7 @@ static int ivpu_cmdq_register(struct ivpu_file_priv *file_priv, struct ivpu_cmdq
 	if (cmdq->db_id)
 		return 0;
 
-	cmdq->entry_count = ivpu_cmdq_get_entry_count(cmdq);
-	cmdq->jobq = (struct vpu_job_queue *)ivpu_bo_vaddr(cmdq->mem);
-
-	ivpu_cmdq_jobq_init(vdev, cmdq->jobq);
+	ivpu_cmdq_jobq_reset(vdev, cmdq->jobq);
 
 	if (vdev->fw->sched_mode == VPU_SCHEDULING_MODE_HW) {
 		ret = ivpu_hws_cmdq_init(file_priv, cmdq, VPU_ENGINE_COMPUTE, cmdq->priority);
@@ -291,9 +330,10 @@ static struct ivpu_cmdq *ivpu_cmdq_acquire_legacy(struct ivpu_file_priv *file_pr
 			break;
 
 	if (!cmdq) {
-		cmdq = ivpu_cmdq_create(file_priv, priority, true);
+		cmdq = ivpu_cmdq_create(file_priv, priority, 0);
 		if (!cmdq)
 			return NULL;
+		cmdq->is_legacy = true;
 	}
 
 	return cmdq;
@@ -308,7 +348,7 @@ static struct ivpu_cmdq *ivpu_cmdq_acquire(struct ivpu_file_priv *file_priv, u32
 
 	cmdq = xa_load(&file_priv->cmdq_xa, cmdq_id);
 	if (!cmdq) {
-		ivpu_warn_ratelimited(vdev, "Failed to find command queue with ID: %u\n", cmdq_id);
+		ivpu_dbg(vdev, IOCTL, "Failed to find command queue with ID: %u\n", cmdq_id);
 		return NULL;
 	}
 
@@ -402,17 +442,14 @@ static int ivpu_cmdq_push_job(struct ivpu_cmdq *cmdq, struct ivpu_job *job)
 	if (unlikely(ivpu_test_mode & IVPU_TEST_MODE_NULL_SUBMISSION))
 		entry->flags = VPU_JOB_FLAGS_NULL_SUBMISSION_MASK;
 
-	if (vdev->fw->sched_mode == VPU_SCHEDULING_MODE_HW) {
-		if (cmdq->primary_preempt_buf) {
-			entry->primary_preempt_buf_addr = cmdq->primary_preempt_buf->vpu_addr;
-			entry->primary_preempt_buf_size = ivpu_bo_size(cmdq->primary_preempt_buf);
-		}
+	if (job->primary_preempt_buf) {
+		entry->primary_preempt_buf_addr = job->primary_preempt_buf->vpu_addr;
+		entry->primary_preempt_buf_size = ivpu_bo_size(job->primary_preempt_buf);
+	}
 
-		if (cmdq->secondary_preempt_buf) {
-			entry->secondary_preempt_buf_addr = cmdq->secondary_preempt_buf->vpu_addr;
-			entry->secondary_preempt_buf_size =
-				ivpu_bo_size(cmdq->secondary_preempt_buf);
-		}
+	if (job->secondary_preempt_buf) {
+		entry->secondary_preempt_buf_addr = job->secondary_preempt_buf->vpu_addr;
+		entry->secondary_preempt_buf_size = ivpu_bo_size(job->secondary_preempt_buf);
 	}
 
 	wmb(); /* Ensure that tail is updated after filling entry */
@@ -497,7 +534,7 @@ ivpu_job_create(struct ivpu_file_priv *file_priv, u32 engine_idx, u32 bo_count)
 	job->bo_count = bo_count;
 	job->done_fence = ivpu_fence_create(vdev);
 	if (!job->done_fence) {
-		ivpu_warn_ratelimited(vdev, "Failed to create a fence\n");
+		ivpu_err(vdev, "Failed to create a fence\n");
 		goto err_free_job;
 	}
 
@@ -527,21 +564,26 @@ static struct ivpu_job *ivpu_job_remove_from_submitted_jobs(struct ivpu_device *
 	return job;
 }
 
-static int ivpu_job_signal_and_destroy(struct ivpu_device *vdev, u32 job_id, u32 job_status)
+bool ivpu_job_handle_engine_error(struct ivpu_device *vdev, u32 job_id, u32 job_status)
 {
-	struct ivpu_job *job;
-
 	lockdep_assert_held(&vdev->submitted_jobs_lock);
 
-	job = xa_load(&vdev->submitted_jobs_xa, job_id);
-	if (!job)
-		return -ENOENT;
+	switch (job_status) {
+	case VPU_JSM_STATUS_PROCESSING_ERR:
+	case VPU_JSM_STATUS_ENGINE_RESET_REQUIRED_MIN ... VPU_JSM_STATUS_ENGINE_RESET_REQUIRED_MAX:
+	{
+		struct ivpu_job *job = xa_load(&vdev->submitted_jobs_xa, job_id);
 
-	if (job_status == VPU_JSM_STATUS_MVNCI_CONTEXT_VIOLATION_HW) {
+		if (!job)
+			return false;
+
+		/* Trigger an engine reset */
 		guard(mutex)(&job->file_priv->lock);
 
+		job->job_status = job_status;
+
 		if (job->file_priv->has_mmu_faults)
-			return 0;
+			return false;
 
 		/*
 		 * Mark context as faulty and defer destruction of the job to jobs abort thread
@@ -549,23 +591,43 @@ static int ivpu_job_signal_and_destroy(struct ivpu_device *vdev, u32 job_id, u32
 		 * status and ensure both are handled in the same way
 		 */
 		job->file_priv->has_mmu_faults = true;
-		queue_work(system_wq, &vdev->context_abort_work);
-		return 0;
+		queue_work(system_percpu_wq, &vdev->context_abort_work);
+		return true;
+	}
+	default:
+		/* Complete job with error status, engine reset not required */
+		break;
 	}
 
-	job = ivpu_job_remove_from_submitted_jobs(vdev, job_id);
+	return false;
+}
+
+static int ivpu_job_signal_and_destroy(struct ivpu_device *vdev, u32 job_id, u32 job_status)
+{
+	struct ivpu_job *job;
+
+	lockdep_assert_held(&vdev->submitted_jobs_lock);
+
+	job = xa_load(&vdev->submitted_jobs_xa, job_id);
 	if (!job)
 		return -ENOENT;
 
-	if (job->file_priv->has_mmu_faults)
-		job_status = DRM_IVPU_JOB_STATUS_ABORTED;
+	ivpu_job_remove_from_submitted_jobs(vdev, job_id);
 
-	job->bos[CMD_BUF_IDX]->job_status = job_status;
+	if (job->job_status == VPU_JSM_STATUS_SUCCESS) {
+		if (job->file_priv->has_mmu_faults)
+			job->job_status = DRM_IVPU_JOB_STATUS_ABORTED;
+		else
+			job->job_status = job_status;
+	}
+
+	job->bos[CMD_BUF_IDX]->job_status = job->job_status;
 	dma_fence_signal(job->done_fence);
 
 	trace_job("done", job);
 	ivpu_dbg(vdev, JOB, "Job complete:  id %3u ctx %2d cmdq_id %u engine %d status 0x%x\n",
-		 job->job_id, job->file_priv->ctx.id, job->cmdq_id, job->engine_idx, job_status);
+		 job->job_id, job->file_priv->ctx.id, job->cmdq_id, job->engine_idx,
+		 job->job_status);
 
 	ivpu_job_destroy(job);
 	ivpu_stop_job_timeout_detection(vdev);
@@ -625,7 +687,6 @@ static int ivpu_job_submit(struct ivpu_job *job, u8 priority, u32 cmdq_id)
 	else
 		cmdq = ivpu_cmdq_acquire(file_priv, cmdq_id);
 	if (!cmdq) {
-		ivpu_warn_ratelimited(vdev, "Failed to get job queue, ctx %d\n", file_priv->ctx.id);
 		ret = -EINVAL;
 		goto err_unlock;
 	}
@@ -636,6 +697,13 @@ static int ivpu_job_submit(struct ivpu_job *job, u8 priority, u32 cmdq_id)
 		goto err_unlock;
 	}
 
+	ret = ivpu_preemption_job_init(vdev, file_priv, cmdq, job);
+	if (ret) {
+		ivpu_err(vdev, "Failed to initialize preemption buffers for job %d: %d\n",
+			 job->job_id, ret);
+		goto err_unlock;
+	}
+
 	job->cmdq_id = cmdq->id;
 
 	is_first_job = xa_empty(&vdev->submitted_jobs_xa);
@@ -689,7 +757,7 @@ err_unlock:
 
 static int
 ivpu_job_prepare_bos_for_submit(struct drm_file *file, struct ivpu_job *job, u32 *buf_handles,
-				u32 buf_count, u32 commands_offset)
+				u32 buf_count, u32 commands_offset, u32 preempt_buffer_index)
 {
 	struct ivpu_file_priv *file_priv = job->file_priv;
 	struct ivpu_device *vdev = file_priv->vdev;
@@ -702,40 +770,58 @@ ivpu_job_prepare_bos_for_submit(struct drm_file *file, struct ivpu_job *job, u32
 	for (i = 0; i < buf_count; i++) {
 		struct drm_gem_object *obj = drm_gem_object_lookup(file, buf_handles[i]);
 
-		if (!obj)
+		if (!obj) {
+			ivpu_dbg(vdev, IOCTL, "Failed to lookup GEM object with handle %u\n",
+				 buf_handles[i]);
 			return -ENOENT;
+		}
 
 		job->bos[i] = to_ivpu_bo(obj);
 
-		ret = ivpu_bo_pin(job->bos[i]);
+		ret = ivpu_bo_bind(job->bos[i]);
 		if (ret)
 			return ret;
 	}
 
 	bo = job->bos[CMD_BUF_IDX];
 	if (!dma_resv_test_signaled(bo->base.base.resv, DMA_RESV_USAGE_READ)) {
-		ivpu_warn(vdev, "Buffer is already in use\n");
+		ivpu_dbg(vdev, IOCTL, "Buffer is already in use by another job\n");
 		return -EBUSY;
 	}
 
 	if (commands_offset >= ivpu_bo_size(bo)) {
-		ivpu_warn(vdev, "Invalid command buffer offset %u\n", commands_offset);
+		ivpu_dbg(vdev, IOCTL, "Invalid commands offset %u for buffer size %zu\n",
+			 commands_offset, ivpu_bo_size(bo));
 		return -EINVAL;
 	}
 
 	job->cmd_buf_vpu_addr = bo->vpu_addr + commands_offset;
 
+	if (preempt_buffer_index) {
+		struct ivpu_bo *preempt_bo = job->bos[preempt_buffer_index];
+
+		if (ivpu_bo_size(preempt_bo) < ivpu_fw_preempt_buf_size(vdev)) {
+			ivpu_dbg(vdev, IOCTL, "Preemption buffer is too small\n");
+			return -EINVAL;
+		}
+		if (ivpu_bo_is_mappable(preempt_bo)) {
+			ivpu_dbg(vdev, IOCTL, "Preemption buffer cannot be mappable\n");
+			return -EINVAL;
+		}
+		job->primary_preempt_buf = preempt_bo;
+	}
+
 	ret = drm_gem_lock_reservations((struct drm_gem_object **)job->bos, buf_count,
 					&acquire_ctx);
 	if (ret) {
-		ivpu_warn(vdev, "Failed to lock reservations: %d\n", ret);
+		ivpu_warn_ratelimited(vdev, "Failed to lock reservations: %d\n", ret);
 		return ret;
 	}
 
 	for (i = 0; i < buf_count; i++) {
 		ret = dma_resv_reserve_fences(job->bos[i]->base.base.resv, 1);
 		if (ret) {
-			ivpu_warn(vdev, "Failed to reserve fences: %d\n", ret);
+			ivpu_warn_ratelimited(vdev, "Failed to reserve fences: %d\n", ret);
 			goto unlock_reservations;
 		}
 	}
@@ -755,7 +841,7 @@ unlock_reservations:
 
 static int ivpu_submit(struct drm_file *file, struct ivpu_file_priv *file_priv, u32 cmdq_id,
 		       u32 buffer_count, u32 engine, void __user *buffers_ptr, u32 cmds_offset,
-		       u8 priority)
+		       u32 preempt_buffer_index, u8 priority)
 {
 	struct ivpu_device *vdev = file_priv->vdev;
 	struct ivpu_job *job;
@@ -782,16 +868,14 @@ static int ivpu_submit(struct drm_file *file, struct ivpu_file_priv *file_priv,
 
 	job = ivpu_job_create(file_priv, engine, buffer_count);
 	if (!job) {
-		ivpu_err(vdev, "Failed to create job\n");
 		ret = -ENOMEM;
 		goto err_exit_dev;
 	}
 
-	ret = ivpu_job_prepare_bos_for_submit(file, job, buf_handles, buffer_count, cmds_offset);
-	if (ret) {
-		ivpu_err(vdev, "Failed to prepare job: %d\n", ret);
+	ret = ivpu_job_prepare_bos_for_submit(file, job, buf_handles, buffer_count, cmds_offset,
+					      preempt_buffer_index);
+	if (ret)
 		goto err_destroy_job;
-	}
 
 	down_read(&vdev->pm->reset_lock);
 	ret = ivpu_job_submit(job, priority, cmdq_id);
@@ -817,58 +901,91 @@ err_free_handles:
 int ivpu_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
 {
 	struct ivpu_file_priv *file_priv = file->driver_priv;
+	struct ivpu_device *vdev = file_priv->vdev;
 	struct drm_ivpu_submit *args = data;
 	u8 priority;
 
-	if (args->engine != DRM_IVPU_ENGINE_COMPUTE)
+	if (args->engine != DRM_IVPU_ENGINE_COMPUTE) {
+		ivpu_dbg(vdev, IOCTL, "Invalid engine %d\n", args->engine);
 		return -EINVAL;
+	}
 
-	if (args->priority > DRM_IVPU_JOB_PRIORITY_REALTIME)
+	if (args->priority > DRM_IVPU_JOB_PRIORITY_REALTIME) {
+		ivpu_dbg(vdev, IOCTL, "Invalid priority %d\n", args->priority);
 		return -EINVAL;
+	}
 
-	if (args->buffer_count == 0 || args->buffer_count > JOB_MAX_BUFFER_COUNT)
+	if (args->buffer_count == 0 || args->buffer_count > JOB_MAX_BUFFER_COUNT) {
+		ivpu_dbg(vdev, IOCTL, "Invalid buffer count %u\n", args->buffer_count);
 		return -EINVAL;
+	}
 
-	if (!IS_ALIGNED(args->commands_offset, 8))
+	if (!IS_ALIGNED(args->commands_offset, 8)) {
+		ivpu_dbg(vdev, IOCTL, "Invalid commands offset %u\n", args->commands_offset);
 		return -EINVAL;
+	}
 
-	if (!file_priv->ctx.id)
+	if (!file_priv->ctx.id) {
+		ivpu_dbg(vdev, IOCTL, "Context not initialized\n");
 		return -EINVAL;
+	}
 
-	if (file_priv->has_mmu_faults)
+	if (file_priv->has_mmu_faults) {
+		ivpu_dbg(vdev, IOCTL, "Context %u has MMU faults\n", file_priv->ctx.id);
 		return -EBADFD;
+	}
 
 	priority = ivpu_job_to_jsm_priority(args->priority);
 
 	return ivpu_submit(file, file_priv, 0, args->buffer_count, args->engine,
-			   (void __user *)args->buffers_ptr, args->commands_offset, priority);
+			   (void __user *)args->buffers_ptr, args->commands_offset, 0, priority);
 }
 
 int ivpu_cmdq_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
 {
 	struct ivpu_file_priv *file_priv = file->driver_priv;
+	struct ivpu_device *vdev = file_priv->vdev;
 	struct drm_ivpu_cmdq_submit *args = data;
 
-	if (!ivpu_is_capable(file_priv->vdev, DRM_IVPU_CAP_MANAGE_CMDQ))
+	if (!ivpu_is_capable(file_priv->vdev, DRM_IVPU_CAP_MANAGE_CMDQ)) {
+		ivpu_dbg(vdev, IOCTL, "Command queue management not supported\n");
 		return -ENODEV;
+	}
 
-	if (args->cmdq_id < IVPU_CMDQ_MIN_ID || args->cmdq_id > IVPU_CMDQ_MAX_ID)
+	if (args->cmdq_id < IVPU_CMDQ_MIN_ID || args->cmdq_id > IVPU_CMDQ_MAX_ID) {
+		ivpu_dbg(vdev, IOCTL, "Invalid command queue ID %u\n", args->cmdq_id);
 		return -EINVAL;
+	}
 
-	if (args->buffer_count == 0 || args->buffer_count > JOB_MAX_BUFFER_COUNT)
+	if (args->buffer_count == 0 || args->buffer_count > JOB_MAX_BUFFER_COUNT) {
+		ivpu_dbg(vdev, IOCTL, "Invalid buffer count %u\n", args->buffer_count);
 		return -EINVAL;
+	}
 
-	if (!IS_ALIGNED(args->commands_offset, 8))
+	if (args->preempt_buffer_index >= args->buffer_count) {
+		ivpu_dbg(vdev, IOCTL, "Invalid preemption buffer index %u\n",
+			 args->preempt_buffer_index);
 		return -EINVAL;
+	}
 
-	if (!file_priv->ctx.id)
+	if (!IS_ALIGNED(args->commands_offset, 8)) {
+		ivpu_dbg(vdev, IOCTL, "Invalid commands offset %u\n", args->commands_offset);
 		return -EINVAL;
+	}
 
-	if (file_priv->has_mmu_faults)
+	if (!file_priv->ctx.id) {
+		ivpu_dbg(vdev, IOCTL, "Context not initialized\n");
+		return -EINVAL;
+	}
+
+	if (file_priv->has_mmu_faults) {
+		ivpu_dbg(vdev, IOCTL, "Context %u has MMU faults\n", file_priv->ctx.id);
 		return -EBADFD;
+	}
 
 	return ivpu_submit(file, file_priv, args->cmdq_id, args->buffer_count, VPU_ENGINE_COMPUTE,
-			   (void __user *)args->buffers_ptr, args->commands_offset, 0);
+			   (void __user *)args->buffers_ptr, args->commands_offset,
+			   args->preempt_buffer_index, 0);
 }
 
 int ivpu_cmdq_create_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
@@ -879,11 +996,15 @@ int ivpu_cmdq_create_ioctl(struct drm_device *dev, void *data, struct drm_file *
 	struct ivpu_cmdq *cmdq;
 	int ret;
 
-	if (!ivpu_is_capable(vdev, DRM_IVPU_CAP_MANAGE_CMDQ))
+	if (!ivpu_is_capable(vdev, DRM_IVPU_CAP_MANAGE_CMDQ)) {
+		ivpu_dbg(vdev, IOCTL, "Command queue management not supported\n");
 		return -ENODEV;
+	}
 
-	if (args->priority > DRM_IVPU_JOB_PRIORITY_REALTIME)
+	if (args->priority > DRM_IVPU_JOB_PRIORITY_REALTIME) {
+		ivpu_dbg(vdev, IOCTL, "Invalid priority %d\n", args->priority);
 		return -EINVAL;
+	}
 
 	ret = ivpu_rpm_get(vdev);
 	if (ret < 0)
@@ -891,7 +1012,7 @@ int ivpu_cmdq_create_ioctl(struct drm_device *dev, void *data, struct drm_file *
 
 	mutex_lock(&file_priv->lock);
 
-	cmdq = ivpu_cmdq_create(file_priv, ivpu_job_to_jsm_priority(args->priority), false);
+	cmdq = ivpu_cmdq_create(file_priv, ivpu_job_to_jsm_priority(args->priority), args->flags);
 	if (cmdq)
 		args->cmdq_id = cmdq->id;
 
@@ -911,8 +1032,10 @@ int ivpu_cmdq_destroy_ioctl(struct drm_device *dev, void *data, struct drm_file
 	u32 cmdq_id = 0;
 	int ret;
 
-	if (!ivpu_is_capable(vdev, DRM_IVPU_CAP_MANAGE_CMDQ))
+	if (!ivpu_is_capable(vdev, DRM_IVPU_CAP_MANAGE_CMDQ)) {
+		ivpu_dbg(vdev, IOCTL, "Command queue management not supported\n");
 		return -ENODEV;
+	}
 
 	ret = ivpu_rpm_get(vdev);
 	if (ret < 0)
@@ -959,7 +1082,9 @@ ivpu_job_done_callback(struct ivpu_device *vdev, struct ivpu_ipc_hdr *ipc_hdr,
 	payload = (struct vpu_ipc_msg_payload_job_done *)&jsm_msg->payload;
 
 	mutex_lock(&vdev->submitted_jobs_lock);
-	ivpu_job_signal_and_destroy(vdev, payload->job_id, payload->job_status);
+	if (!ivpu_job_handle_engine_error(vdev, payload->job_id, payload->job_status))
+		/* No engine error, complete the job normally */
+		ivpu_job_signal_and_destroy(vdev, payload->job_id, payload->job_status);
 	mutex_unlock(&vdev->submitted_jobs_lock);
 }
 
@@ -987,7 +1112,7 @@ void ivpu_context_abort_work_fn(struct work_struct *work)
 
 	if (vdev->fw->sched_mode == VPU_SCHEDULING_MODE_HW)
 		if (ivpu_jsm_reset_engine(vdev, 0))
-			return;
+			goto runtime_put;
 
 	mutex_lock(&vdev->context_list_lock);
 	xa_for_each(&vdev->context_xa, ctx_id, file_priv) {
@@ -1011,7 +1136,7 @@ void ivpu_context_abort_work_fn(struct work_struct *work)
 		goto runtime_put;
 
 	if (ivpu_jsm_hws_resume_engine(vdev, 0))
-		return;
+		goto runtime_put;
 	/*
 	 * In hardware scheduling mode NPU already has stopped processing jobs
 	 * and won't send us any further notifications, thus we have to free job related resources
@@ -1024,6 +1149,5 @@ void ivpu_context_abort_work_fn(struct work_struct *work)
 	mutex_unlock(&vdev->submitted_jobs_lock);
 
 runtime_put:
-	pm_runtime_mark_last_busy(vdev->drm.dev);
 	pm_runtime_put_autosuspend(vdev->drm.dev);
 }
diff --git a/drivers/accel/ivpu/ivpu_job.h b/drivers/accel/ivpu/ivpu_job.h
index 2e301c2eea7b..3ab61e6a5616 100644
--- a/drivers/accel/ivpu/ivpu_job.h
+++ b/drivers/accel/ivpu/ivpu_job.h
@@ -1,6 +1,6 @@
 /* SPDX-License-Identifier: GPL-2.0-only */
 /*
- * Copyright (C) 2020-2024 Intel Corporation
+ * Copyright (C) 2020-2025 Intel Corporation
  */
 
 #ifndef __IVPU_JOB_H__
@@ -15,12 +15,17 @@ struct ivpu_device;
 struct ivpu_file_priv;
 
 /**
- * struct ivpu_cmdq - Object representing device queue used to send jobs.
- * @jobq:	   Pointer to job queue memory shared with the device
- * @mem:           Memory allocated for the job queue, shared with device
- * @entry_count    Number of job entries in the queue
- * @db_id:	   Doorbell assigned to this job queue
- * @db_registered: True if doorbell is registered in device
+ * struct ivpu_cmdq - Represents a command queue for submitting jobs to the VPU.
+ * Tracks queue memory, preemption buffers, and metadata for job management.
+ * @jobq:                Pointer to job queue memory shared with the device
+ * @primary_preempt_buf: Primary preemption buffer for this queue (optional)
+ * @secondary_preempt_buf: Secondary preemption buffer for this queue (optional)
+ * @mem:                 Memory allocated for the job queue, shared with device
+ * @entry_count:         Number of job entries in the queue
+ * @id:                  Unique command queue ID
+ * @db_id:               Doorbell ID assigned to this job queue
+ * @priority:            Priority level of the command queue
+ * @is_legacy:           True if this is a legacy command queue
  */
 struct ivpu_cmdq {
 	struct vpu_job_queue *jobq;
@@ -35,16 +40,22 @@ struct ivpu_cmdq {
 };
 
 /**
- * struct ivpu_job - KMD object that represents batchbuffer / DMA buffer.
- * Each batch / DMA buffer is a job to be submitted and executed by the VPU FW.
- * This is a unit of execution, and be tracked by the job_id for
- * any status reporting from VPU FW through IPC JOB RET/DONE message.
- * @file_priv:		  The client that submitted this job
- * @job_id:		  Job ID for KMD tracking and job status reporting from VPU FW
- * @status:		  Status of the Job from IPC JOB RET/DONE message
- * @batch_buffer:	  CPU vaddr points to the batch buffer memory allocated for the job
- * @submit_status_offset: Offset within batch buffer where job completion handler
-			  will update the job status
+ * struct ivpu_job - Representing a batch or DMA buffer submitted to the VPU.
+ * Each job is a unit of execution, tracked by job_id for status reporting from VPU FW.
+ * The structure holds all resources and metadata needed for job submission, execution,
+ * and completion handling.
+ * @vdev:                Pointer to the VPU device
+ * @file_priv:           The client context that submitted this job
+ * @done_fence:          Fence signaled when job completes
+ * @cmd_buf_vpu_addr:    VPU address of the command buffer for this job
+ * @cmdq_id:             Command queue ID used for submission
+ * @job_id:              Unique job ID for tracking and status reporting
+ * @engine_idx:          Engine index for job execution
+ * @job_status:          Status reported by firmware for this job
+ * @primary_preempt_buf: Primary preemption buffer for job
+ * @secondary_preempt_buf: Secondary preemption buffer for job (optional)
+ * @bo_count:            Number of buffer objects associated with this job
+ * @bos:                 Array of buffer objects used by the job (batch buffer is at index 0)
  */
 struct ivpu_job {
 	struct ivpu_device *vdev;
@@ -54,6 +65,9 @@ struct ivpu_job {
 	u32 cmdq_id;
 	u32 job_id;
 	u32 engine_idx;
+	u32 job_status;
+	struct ivpu_bo *primary_preempt_buf;
+	struct ivpu_bo *secondary_preempt_buf;
 	size_t bo_count;
 	struct ivpu_bo *bos[] __counted_by(bo_count);
 };
@@ -71,6 +85,7 @@ void ivpu_cmdq_abort_all_jobs(struct ivpu_device *vdev, u32 ctx_id, u32 cmdq_id)
 
 void ivpu_job_done_consumer_init(struct ivpu_device *vdev);
 void ivpu_job_done_consumer_fini(struct ivpu_device *vdev);
+bool ivpu_job_handle_engine_error(struct ivpu_device *vdev, u32 job_id, u32 job_status);
 void ivpu_context_abort_work_fn(struct work_struct *work);
 
 void ivpu_jobs_abort_all(struct ivpu_device *vdev);
diff --git a/drivers/accel/ivpu/ivpu_mmu.c b/drivers/accel/ivpu/ivpu_mmu.c
index 5ea010568faa..e1baf6b64935 100644
--- a/drivers/accel/ivpu/ivpu_mmu.c
+++ b/drivers/accel/ivpu/ivpu_mmu.c
@@ -970,7 +970,7 @@ void ivpu_mmu_irq_evtq_handler(struct ivpu_device *vdev)
 		}
 	}
 
-	queue_work(system_wq, &vdev->context_abort_work);
+	queue_work(system_percpu_wq, &vdev->context_abort_work);
 }
 
 void ivpu_mmu_evtq_dump(struct ivpu_device *vdev)
diff --git a/drivers/accel/ivpu/ivpu_mmu_context.c b/drivers/accel/ivpu/ivpu_mmu_context.c
index f0267efa55aa..87ad593ef47d 100644
--- a/drivers/accel/ivpu/ivpu_mmu_context.c
+++ b/drivers/accel/ivpu/ivpu_mmu_context.c
@@ -430,7 +430,7 @@ static void ivpu_mmu_context_unmap_pages(struct ivpu_mmu_context *ctx, u64 vpu_a
 
 int
 ivpu_mmu_context_map_sgt(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
-			 u64 vpu_addr, struct sg_table *sgt,  bool llc_coherent)
+			 u64 vpu_addr, struct sg_table *sgt, bool llc_coherent, bool read_only)
 {
 	size_t start_vpu_addr = vpu_addr;
 	struct scatterlist *sg;
@@ -450,6 +450,8 @@ ivpu_mmu_context_map_sgt(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
 	prot = IVPU_MMU_ENTRY_MAPPED;
 	if (llc_coherent)
 		prot |= IVPU_MMU_ENTRY_FLAG_LLC_COHERENT;
+	if (read_only)
+		prot |= IVPU_MMU_ENTRY_FLAG_RO;
 
 	mutex_lock(&ctx->lock);
 
@@ -527,7 +529,8 @@ ivpu_mmu_context_unmap_sgt(struct ivpu_device *vdev, struct ivpu_mmu_context *ct
 
 	ret = ivpu_mmu_invalidate_tlb(vdev, ctx->id);
 	if (ret)
-		ivpu_warn(vdev, "Failed to invalidate TLB for ctx %u: %d\n", ctx->id, ret);
+		ivpu_warn_ratelimited(vdev, "Failed to invalidate TLB for ctx %u: %d\n",
+				      ctx->id, ret);
 }
 
 int
@@ -568,7 +571,7 @@ void ivpu_mmu_context_init(struct ivpu_device *vdev, struct ivpu_mmu_context *ct
 	mutex_init(&ctx->lock);
 
 	if (!context_id) {
-		start = vdev->hw->ranges.global.start;
+		start = vdev->hw->ranges.runtime.start;
 		end = vdev->hw->ranges.shave.end;
 	} else {
 		start = min_t(u64, vdev->hw->ranges.user.start, vdev->hw->ranges.shave.start);
diff --git a/drivers/accel/ivpu/ivpu_mmu_context.h b/drivers/accel/ivpu/ivpu_mmu_context.h
index f255310968cf..663a11a9db11 100644
--- a/drivers/accel/ivpu/ivpu_mmu_context.h
+++ b/drivers/accel/ivpu/ivpu_mmu_context.h
@@ -42,7 +42,7 @@ int ivpu_mmu_context_insert_node(struct ivpu_mmu_context *ctx, const struct ivpu
 void ivpu_mmu_context_remove_node(struct ivpu_mmu_context *ctx, struct drm_mm_node *node);
 
 int ivpu_mmu_context_map_sgt(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
-			     u64 vpu_addr, struct sg_table *sgt, bool llc_coherent);
+			     u64 vpu_addr, struct sg_table *sgt, bool llc_coherent, bool read_only);
 void ivpu_mmu_context_unmap_sgt(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
 				u64 vpu_addr, struct sg_table *sgt);
 int ivpu_mmu_context_set_pages_ro(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
diff --git a/drivers/accel/ivpu/ivpu_ms.c b/drivers/accel/ivpu/ivpu_ms.c
index 2a043baf10ca..1d9c1cb17924 100644
--- a/drivers/accel/ivpu/ivpu_ms.c
+++ b/drivers/accel/ivpu/ivpu_ms.c
@@ -8,6 +8,7 @@
 
 #include "ivpu_drv.h"
 #include "ivpu_gem.h"
+#include "ivpu_hw.h"
 #include "ivpu_jsm_msg.h"
 #include "ivpu_ms.h"
 #include "ivpu_pm.h"
@@ -37,8 +38,8 @@ int ivpu_ms_start_ioctl(struct drm_device *dev, void *data, struct drm_file *fil
 	struct drm_ivpu_metric_streamer_start *args = data;
 	struct ivpu_device *vdev = file_priv->vdev;
 	struct ivpu_ms_instance *ms;
-	u64 single_buff_size;
 	u32 sample_size;
+	u64 buf_size;
 	int ret;
 
 	if (!args->metric_group_mask || !args->read_period_samples ||
@@ -52,7 +53,8 @@ int ivpu_ms_start_ioctl(struct drm_device *dev, void *data, struct drm_file *fil
 	mutex_lock(&file_priv->ms_lock);
 
 	if (get_instance_by_mask(file_priv, args->metric_group_mask)) {
-		ivpu_err(vdev, "Instance already exists (mask %#llx)\n", args->metric_group_mask);
+		ivpu_dbg(vdev, IOCTL, "Instance already exists (mask %#llx)\n",
+			 args->metric_group_mask);
 		ret = -EALREADY;
 		goto unlock;
 	}
@@ -69,12 +71,18 @@ int ivpu_ms_start_ioctl(struct drm_device *dev, void *data, struct drm_file *fil
 	if (ret)
 		goto err_free_ms;
 
-	single_buff_size = sample_size *
-		((u64)args->read_period_samples * MS_READ_PERIOD_MULTIPLIER);
-	ms->bo = ivpu_bo_create_global(vdev, PAGE_ALIGN(single_buff_size * MS_NUM_BUFFERS),
-				       DRM_IVPU_BO_CACHED | DRM_IVPU_BO_MAPPABLE);
+	buf_size = PAGE_ALIGN((u64)args->read_period_samples * sample_size *
+			      MS_READ_PERIOD_MULTIPLIER * MS_NUM_BUFFERS);
+	if (buf_size > ivpu_hw_range_size(&vdev->hw->ranges.global)) {
+		ivpu_dbg(vdev, IOCTL, "Requested MS buffer size %llu exceeds range size %llu\n",
+			 buf_size, ivpu_hw_range_size(&vdev->hw->ranges.global));
+		ret = -EINVAL;
+		goto err_free_ms;
+	}
+
+	ms->bo = ivpu_bo_create_global(vdev, buf_size, DRM_IVPU_BO_CACHED | DRM_IVPU_BO_MAPPABLE);
 	if (!ms->bo) {
-		ivpu_err(vdev, "Failed to allocate MS buffer (size %llu)\n", single_buff_size);
+		ivpu_dbg(vdev, IOCTL, "Failed to allocate MS buffer (size %llu)\n", buf_size);
 		ret = -ENOMEM;
 		goto err_free_ms;
 	}
@@ -175,7 +183,8 @@ int ivpu_ms_get_data_ioctl(struct drm_device *dev, void *data, struct drm_file *
 
 	ms = get_instance_by_mask(file_priv, args->metric_group_mask);
 	if (!ms) {
-		ivpu_err(vdev, "Instance doesn't exist for mask: %#llx\n", args->metric_group_mask);
+		ivpu_dbg(vdev, IOCTL, "Instance doesn't exist for mask: %#llx\n",
+			 args->metric_group_mask);
 		ret = -EINVAL;
 		goto unlock;
 	}
diff --git a/drivers/accel/ivpu/ivpu_pm.c b/drivers/accel/ivpu/ivpu_pm.c
index ea30db181cd7..480c075d87f6 100644
--- a/drivers/accel/ivpu/ivpu_pm.c
+++ b/drivers/accel/ivpu/ivpu_pm.c
@@ -33,8 +33,11 @@ static unsigned long ivpu_tdr_timeout_ms;
 module_param_named(tdr_timeout_ms, ivpu_tdr_timeout_ms, ulong, 0644);
 MODULE_PARM_DESC(tdr_timeout_ms, "Timeout for device hang detection, in milliseconds, 0 - default");
 
+static unsigned long ivpu_inference_timeout_ms;
+module_param_named(inference_timeout_ms, ivpu_inference_timeout_ms, ulong, 0644);
+MODULE_PARM_DESC(inference_timeout_ms, "Inference maximum duration, in milliseconds, 0 - default");
+
 #define PM_RESCHEDULE_LIMIT     5
-#define PM_TDR_HEARTBEAT_LIMIT  30
 
 static void ivpu_pm_prepare_cold_boot(struct ivpu_device *vdev)
 {
@@ -51,7 +54,7 @@ static void ivpu_pm_prepare_cold_boot(struct ivpu_device *vdev)
 static void ivpu_pm_prepare_warm_boot(struct ivpu_device *vdev)
 {
 	struct ivpu_fw_info *fw = vdev->fw;
-	struct vpu_boot_params *bp = ivpu_bo_vaddr(fw->mem);
+	struct vpu_boot_params *bp = ivpu_bo_vaddr(fw->mem_bp);
 
 	if (!bp->save_restore_ret_address) {
 		ivpu_pm_prepare_cold_boot(vdev);
@@ -183,7 +186,7 @@ void ivpu_pm_trigger_recovery(struct ivpu_device *vdev, const char *reason)
 	if (atomic_cmpxchg(&vdev->pm->reset_pending, 0, 1) == 0) {
 		ivpu_hw_diagnose_failure(vdev);
 		ivpu_hw_irq_disable(vdev); /* Disable IRQ early to protect from IRQ storm */
-		queue_work(system_unbound_wq, &vdev->pm->recovery_work);
+		queue_work(system_dfl_wq, &vdev->pm->recovery_work);
 	}
 }
 
@@ -191,6 +194,10 @@ static void ivpu_job_timeout_work(struct work_struct *work)
 {
 	struct ivpu_pm_info *pm = container_of(work, struct ivpu_pm_info, job_timeout_work.work);
 	struct ivpu_device *vdev = pm->vdev;
+	unsigned long timeout_ms = ivpu_tdr_timeout_ms ? ivpu_tdr_timeout_ms : vdev->timeout.tdr;
+	unsigned long inference_timeout_ms = ivpu_inference_timeout_ms ? ivpu_inference_timeout_ms :
+					     vdev->timeout.inference;
+	u64 inference_max_retries;
 	u64 heartbeat;
 
 	if (ivpu_jsm_get_heartbeat(vdev, 0, &heartbeat) || heartbeat <= vdev->fw->last_heartbeat) {
@@ -198,8 +205,10 @@ static void ivpu_job_timeout_work(struct work_struct *work)
 		goto recovery;
 	}
 
-	if (atomic_fetch_inc(&vdev->job_timeout_counter) > PM_TDR_HEARTBEAT_LIMIT) {
-		ivpu_err(vdev, "Job timeout detected, heartbeat limit exceeded\n");
+	inference_max_retries = DIV_ROUND_UP(inference_timeout_ms, timeout_ms);
+	if (atomic_fetch_inc(&vdev->job_timeout_counter) >= inference_max_retries) {
+		ivpu_err(vdev, "Job timeout detected, heartbeat limit (%lld) exceeded\n",
+			 inference_max_retries);
 		goto recovery;
 	}
 
@@ -217,7 +226,8 @@ void ivpu_start_job_timeout_detection(struct ivpu_device *vdev)
 	unsigned long timeout_ms = ivpu_tdr_timeout_ms ? ivpu_tdr_timeout_ms : vdev->timeout.tdr;
 
 	/* No-op if already queued */
-	queue_delayed_work(system_wq, &vdev->pm->job_timeout_work, msecs_to_jiffies(timeout_ms));
+	queue_delayed_work(system_percpu_wq, &vdev->pm->job_timeout_work,
+			   msecs_to_jiffies(timeout_ms));
 }
 
 void ivpu_stop_job_timeout_detection(struct ivpu_device *vdev)
@@ -350,7 +360,6 @@ int ivpu_rpm_get(struct ivpu_device *vdev)
 
 void ivpu_rpm_put(struct ivpu_device *vdev)
 {
-	pm_runtime_mark_last_busy(vdev->drm.dev);
 	pm_runtime_put_autosuspend(vdev->drm.dev);
 }
 
@@ -408,10 +417,10 @@ void ivpu_pm_init(struct ivpu_device *vdev)
 	ivpu_dbg(vdev, PM, "Autosuspend delay = %d\n", delay);
 }
 
-void ivpu_pm_cancel_recovery(struct ivpu_device *vdev)
+void ivpu_pm_disable_recovery(struct ivpu_device *vdev)
 {
 	drm_WARN_ON(&vdev->drm, delayed_work_pending(&vdev->pm->job_timeout_work));
-	cancel_work_sync(&vdev->pm->recovery_work);
+	disable_work_sync(&vdev->pm->recovery_work);
 }
 
 void ivpu_pm_enable(struct ivpu_device *vdev)
@@ -419,7 +428,6 @@ void ivpu_pm_enable(struct ivpu_device *vdev)
 	struct device *dev = vdev->drm.dev;
 
 	pm_runtime_allow(dev);
-	pm_runtime_mark_last_busy(dev);
 	pm_runtime_put_autosuspend(dev);
 }
 
@@ -493,6 +501,11 @@ void ivpu_pm_irq_dct_work_fn(struct work_struct *work)
 	else
 		ret = ivpu_pm_dct_disable(vdev);
 
-	if (!ret)
-		ivpu_hw_btrs_dct_set_status(vdev, enable, vdev->pm->dct_active_percent);
+	if (!ret) {
+		/* Convert percent to U1.7 format */
+		u8 val = DIV_ROUND_CLOSEST(vdev->pm->dct_active_percent * 128, 100);
+
+		ivpu_hw_btrs_dct_set_status(vdev, enable, val);
+	}
+
 }
diff --git a/drivers/accel/ivpu/ivpu_pm.h b/drivers/accel/ivpu/ivpu_pm.h
index 89b264cc0e3e..a2aa7a27f32e 100644
--- a/drivers/accel/ivpu/ivpu_pm.h
+++ b/drivers/accel/ivpu/ivpu_pm.h
@@ -25,7 +25,7 @@ struct ivpu_pm_info {
 void ivpu_pm_init(struct ivpu_device *vdev);
 void ivpu_pm_enable(struct ivpu_device *vdev);
 void ivpu_pm_disable(struct ivpu_device *vdev);
-void ivpu_pm_cancel_recovery(struct ivpu_device *vdev);
+void ivpu_pm_disable_recovery(struct ivpu_device *vdev);
 
 int ivpu_pm_suspend_cb(struct device *dev);
 int ivpu_pm_resume_cb(struct device *dev);
diff --git a/drivers/accel/ivpu/ivpu_sysfs.c b/drivers/accel/ivpu/ivpu_sysfs.c
index 268ab7744a8b..d250a10caca9 100644
--- a/drivers/accel/ivpu/ivpu_sysfs.c
+++ b/drivers/accel/ivpu/ivpu_sysfs.c
@@ -63,7 +63,8 @@ npu_memory_utilization_show(struct device *dev, struct device_attribute *attr, c
 
 	mutex_lock(&vdev->bo_list_lock);
 	list_for_each_entry(bo, &vdev->bo_list, bo_list_node)
-		total_npu_memory += bo->base.base.size;
+		if (ivpu_bo_is_resident(bo))
+			total_npu_memory += ivpu_bo_size(bo);
 	mutex_unlock(&vdev->bo_list_lock);
 
 	return sysfs_emit(buf, "%lld\n", total_npu_memory);
diff --git a/drivers/accel/ivpu/vpu_jsm_api.h b/drivers/accel/ivpu/vpu_jsm_api.h
index 4b6b2b3d2583..bca6a44dc041 100644
--- a/drivers/accel/ivpu/vpu_jsm_api.h
+++ b/drivers/accel/ivpu/vpu_jsm_api.h
@@ -1,15 +1,16 @@
 /* SPDX-License-Identifier: MIT */
 /*
- * Copyright (c) 2020-2024, Intel Corporation.
+ * Copyright (c) 2020-2025, Intel Corporation.
+ */
+
+/**
+ * @addtogroup Jsm
+ * @{
  */
 
 /**
  * @file
  * @brief JSM shared definitions
- *
- * @ingroup Jsm
- * @brief JSM shared definitions
- * @{
  */
 #ifndef VPU_JSM_API_H
 #define VPU_JSM_API_H
@@ -22,7 +23,7 @@
 /*
  * Minor version changes when API backward compatibility is preserved.
  */
-#define VPU_JSM_API_VER_MINOR 29
+#define VPU_JSM_API_VER_MINOR 33
 
 /*
  * API header changed (field names, documentation, formatting) but API itself has not been changed
@@ -71,9 +72,15 @@
 #define VPU_JSM_STATUS_MVNCI_OUT_OF_RESOURCES		 0xAU
 #define VPU_JSM_STATUS_MVNCI_NOT_IMPLEMENTED		 0xBU
 #define VPU_JSM_STATUS_MVNCI_INTERNAL_ERROR		 0xCU
-/* Job status returned when the job was preempted mid-inference */
+/* @deprecated (use VPU_JSM_STATUS_PREEMPTED_MID_COMMAND instead) */
 #define VPU_JSM_STATUS_PREEMPTED_MID_INFERENCE		 0xDU
+/* Job status returned when the job was preempted mid-command */
+#define VPU_JSM_STATUS_PREEMPTED_MID_COMMAND		 0xDU
+/* Range of status codes that require engine reset */
+#define VPU_JSM_STATUS_ENGINE_RESET_REQUIRED_MIN	 0xEU
 #define VPU_JSM_STATUS_MVNCI_CONTEXT_VIOLATION_HW	 0xEU
+#define VPU_JSM_STATUS_MVNCI_PREEMPTION_TIMED_OUT	 0xFU
+#define VPU_JSM_STATUS_ENGINE_RESET_REQUIRED_MAX	 0x1FU
 
 /*
  * Host <-> VPU IPC channels.
@@ -134,11 +141,21 @@ enum {
 	 *  2. Native fence queues are only supported on VPU 40xx onwards.
 	 */
 	VPU_JOB_QUEUE_FLAGS_USE_NATIVE_FENCE_MASK = (1 << 1U),
-
 	/*
 	 * Enable turbo mode for testing NPU performance; not recommended for regular usage.
 	 */
-	VPU_JOB_QUEUE_FLAGS_TURBO_MODE = (1 << 2U)
+	VPU_JOB_QUEUE_FLAGS_TURBO_MODE = (1 << 2U),
+	/*
+	 * Queue error detection mode flag
+	 * For 'interactive' queues (this bit not set), the FW will identify queues that have not
+	 * completed a job inside the TDR timeout as in error as part of engine reset sequence.
+	 * For 'non-interactive' queues (this bit set), the FW will identify queues that have not
+	 * progressed the heartbeat inside the non-interactive no-progress timeout as in error as
+	 * part of engine reset sequence. Additionally, there is an upper limit applied to these
+	 * queues: even if they progress the heartbeat, if they run longer than non-interactive
+	 * timeout, then the FW will also identify them as in error.
+	 */
+	VPU_JOB_QUEUE_FLAGS_NON_INTERACTIVE = (1 << 3U)
 };
 
 /*
@@ -209,7 +226,7 @@ enum {
  */
 #define VPU_INLINE_CMD_TYPE_FENCE_SIGNAL 0x2
 
-/*
+/**
  * Job scheduling priority bands for both hardware scheduling and OS scheduling.
  */
 enum vpu_job_scheduling_priority_band {
@@ -220,16 +237,16 @@ enum vpu_job_scheduling_priority_band {
 	VPU_JOB_SCHEDULING_PRIORITY_BAND_COUNT = 4,
 };
 
-/*
+/**
  * Job format.
  * Jobs defines the actual workloads to be executed by a given engine.
  */
 struct vpu_job_queue_entry {
-	/**< Address of VPU commands batch buffer */
+	/** Address of VPU commands batch buffer */
 	u64 batch_buf_addr;
-	/**< Job ID */
+	/** Job ID */
 	u32 job_id;
-	/**< Flags bit field, see VPU_JOB_FLAGS_* above */
+	/** Flags bit field, see VPU_JOB_FLAGS_* above */
 	u32 flags;
 	/**
 	 * Doorbell ring timestamp taken by KMD from SoC's global system clock, in
@@ -237,20 +254,20 @@ struct vpu_job_queue_entry {
 	 * to match other profiling timestamps.
 	 */
 	u64 doorbell_timestamp;
-	/**< Extra id for job tracking, used only in the firmware perf traces */
+	/** Extra id for job tracking, used only in the firmware perf traces */
 	u64 host_tracking_id;
-	/**< Address of the primary preemption buffer to use for this job */
+	/** Address of the primary preemption buffer to use for this job */
 	u64 primary_preempt_buf_addr;
-	/**< Size of the primary preemption buffer to use for this job */
+	/** Size of the primary preemption buffer to use for this job */
 	u32 primary_preempt_buf_size;
-	/**< Size of secondary preemption buffer to use for this job */
+	/** Size of secondary preemption buffer to use for this job */
 	u32 secondary_preempt_buf_size;
-	/**< Address of secondary preemption buffer to use for this job */
+	/** Address of secondary preemption buffer to use for this job */
 	u64 secondary_preempt_buf_addr;
 	u64 reserved_0;
 };
 
-/*
+/**
  * Inline command format.
  * Inline commands are the commands executed at scheduler level (typically,
  * synchronization directives). Inline command and job objects must be of
@@ -258,34 +275,36 @@ struct vpu_job_queue_entry {
  */
 struct vpu_inline_cmd {
 	u64 reserved_0;
-	/* Inline command type, see VPU_INLINE_CMD_TYPE_* defines. */
+	/** Inline command type, see VPU_INLINE_CMD_TYPE_* defines. */
 	u32 type;
-	/* Flags bit field, see VPU_JOB_FLAGS_* above. */
+	/** Flags bit field, see VPU_JOB_FLAGS_* above. */
 	u32 flags;
-	/* Inline command payload. Depends on inline command type. */
-	union {
-		/* Fence (wait and signal) commands' payload. */
-		struct {
-			/* Fence object handle. */
+	/** Inline command payload. Depends on inline command type. */
+	union payload {
+		/** Fence (wait and signal) commands' payload. */
+		struct fence {
+			/** Fence object handle. */
 			u64 fence_handle;
-			/* User VA of the current fence value. */
+			/** User VA of the current fence value. */
 			u64 current_value_va;
-			/* User VA of the monitored fence value (read-only). */
+			/** User VA of the monitored fence value (read-only). */
 			u64 monitored_value_va;
-			/* Value to wait for or write in fence location. */
+			/** Value to wait for or write in fence location. */
 			u64 value;
-			/* User VA of the log buffer in which to add log entry on completion. */
+			/** User VA of the log buffer in which to add log entry on completion. */
 			u64 log_buffer_va;
-			/* NPU private data. */
+			/** NPU private data. */
 			u64 npu_private_data;
 		} fence;
-		/* Other commands do not have a payload. */
-		/* Payload definition for future inline commands can be inserted here. */
+		/**
+		 * Other commands do not have a payload:
+		 * Payload definition for future inline commands can be inserted here.
+		 */
 		u64 reserved_1[6];
 	} payload;
 };
 
-/*
+/**
  * Job queue slots can be populated either with job objects or inline command objects.
  */
 union vpu_jobq_slot {
@@ -293,7 +312,7 @@ union vpu_jobq_slot {
 	struct vpu_inline_cmd inline_cmd;
 };
 
-/*
+/**
  * Job queue control registers.
  */
 struct vpu_job_queue_header {
@@ -301,18 +320,18 @@ struct vpu_job_queue_header {
 	u32 head;
 	u32 tail;
 	u32 flags;
-	/* Set to 1 to indicate priority_band field is valid */
+	/** Set to 1 to indicate priority_band field is valid */
 	u32 priority_band_valid;
-	/*
+	/**
 	 * Priority for the work of this job queue, valid only if the HWS is NOT used
-	 * and the `priority_band_valid` is set to 1. It is applied only during
-	 * the VPU_JSM_MSG_REGISTER_DB message processing.
-	 * The device firmware might use the `priority_band` to optimize the power
+	 * and the @ref priority_band_valid is set to 1. It is applied only during
+	 * the @ref VPU_JSM_MSG_REGISTER_DB message processing.
+	 * The device firmware might use the priority_band to optimize the power
 	 * management logic, but it will not affect the order of jobs.
 	 * Available priority bands: @see enum vpu_job_scheduling_priority_band
 	 */
 	u32 priority_band;
-	/* Inside realtime band assigns a further priority, limited to 0..31 range */
+	/** Inside realtime band assigns a further priority, limited to 0..31 range */
 	u32 realtime_priority_level;
 	u32 reserved_0[9];
 };
@@ -337,16 +356,16 @@ enum vpu_trace_entity_type {
 	VPU_TRACE_ENTITY_TYPE_HW_COMPONENT = 2,
 };
 
-/*
+/**
  * HWS specific log buffer header details.
  * Total size is 32 bytes.
  */
 struct vpu_hws_log_buffer_header {
-	/* Written by VPU after adding a log entry. Initialised by host to 0. */
+	/** Written by VPU after adding a log entry. Initialised by host to 0. */
 	u32 first_free_entry_index;
-	/* Incremented by VPU every time the VPU writes the 0th entry; initialised by host to 0. */
+	/** Incremented by VPU every time the VPU writes the 0th entry; initialised by host to 0. */
 	u32 wraparound_count;
-	/*
+	/**
 	 * This is the number of buffers that can be stored in the log buffer provided by the host.
 	 * It is written by host before passing buffer to VPU. VPU should consider it read-only.
 	 */
@@ -354,14 +373,14 @@ struct vpu_hws_log_buffer_header {
 	u64 reserved[2];
 };
 
-/*
+/**
  * HWS specific log buffer entry details.
  * Total size is 32 bytes.
  */
 struct vpu_hws_log_buffer_entry {
-	/* VPU timestamp must be an invariant timer tick (not impacted by DVFS) */
+	/** VPU timestamp must be an invariant timer tick (not impacted by DVFS) */
 	u64 vpu_timestamp;
-	/*
+	/**
 	 * Operation type:
 	 *     0 - context state change
 	 *     1 - queue new work
@@ -371,7 +390,7 @@ struct vpu_hws_log_buffer_entry {
 	 */
 	u32 operation_type;
 	u32 reserved;
-	/* Operation data depends on operation type */
+	/** Operation data depends on operation type */
 	u64 operation_data[2];
 };
 
@@ -381,51 +400,54 @@ enum vpu_hws_native_fence_log_type {
 	VPU_HWS_NATIVE_FENCE_LOG_TYPE_SIGNALS = 2
 };
 
-/* HWS native fence log buffer header. */
+/** HWS native fence log buffer header. */
 struct vpu_hws_native_fence_log_header {
 	union {
 		struct {
-			/* Index of the first free entry in buffer. */
+			/** Index of the first free entry in buffer. */
 			u32 first_free_entry_idx;
-			/* Incremented each time NPU wraps around the buffer to write next entry. */
+			/**
+			 * Incremented whenever the NPU wraps around the buffer and writes
+			 * to the first entry again.
+			 */
 			u32 wraparound_count;
 		};
-		/* Field allowing atomic update of both fields above. */
+		/** Field allowing atomic update of both fields above. */
 		u64 atomic_wraparound_and_entry_idx;
 	};
-	/* Log buffer type, see enum vpu_hws_native_fence_log_type. */
+	/** Log buffer type, see enum vpu_hws_native_fence_log_type. */
 	u64 type;
-	/* Allocated number of entries in the log buffer. */
+	/** Allocated number of entries in the log buffer. */
 	u64 entry_nb;
 	u64 reserved[2];
 };
 
-/* Native fence log operation types. */
+/** Native fence log operation types. */
 enum vpu_hws_native_fence_log_op {
 	VPU_HWS_NATIVE_FENCE_LOG_OP_SIGNAL_EXECUTED = 0,
 	VPU_HWS_NATIVE_FENCE_LOG_OP_WAIT_UNBLOCKED = 1
 };
 
-/* HWS native fence log entry. */
+/** HWS native fence log entry. */
 struct vpu_hws_native_fence_log_entry {
-	/* Newly signaled/unblocked fence value. */
+	/** Newly signaled/unblocked fence value. */
 	u64 fence_value;
-	/* Native fence object handle to which this operation belongs. */
+	/** Native fence object handle to which this operation belongs. */
 	u64 fence_handle;
-	/* Operation type, see enum vpu_hws_native_fence_log_op. */
+	/** Operation type, see enum vpu_hws_native_fence_log_op. */
 	u64 op_type;
 	u64 reserved_0;
-	/*
+	/**
 	 * VPU_HWS_NATIVE_FENCE_LOG_OP_WAIT_UNBLOCKED only: Timestamp at which fence
 	 * wait was started (in NPU SysTime).
 	 */
 	u64 fence_wait_start_ts;
 	u64 reserved_1;
-	/* Timestamp at which fence operation was completed (in NPU SysTime). */
+	/** Timestamp at which fence operation was completed (in NPU SysTime). */
 	u64 fence_end_ts;
 };
 
-/* Native fence log buffer. */
+/** Native fence log buffer. */
 struct vpu_hws_native_fence_log_buffer {
 	struct vpu_hws_native_fence_log_header header;
 	struct vpu_hws_native_fence_log_entry entry[];
@@ -435,10 +457,17 @@ struct vpu_hws_native_fence_log_buffer {
  * Host <-> VPU IPC messages types.
  */
 enum vpu_ipc_msg_type {
+	/** Unsupported command */
 	VPU_JSM_MSG_UNKNOWN = 0xFFFFFFFF,
 
-	/* IPC Host -> Device, Async commands */
+	/** IPC Host -> Device, base id for async commands */
 	VPU_JSM_MSG_ASYNC_CMD = 0x1100,
+	/**
+	 * Reset engine. The NPU cancels all the jobs currently executing on the target
+	 * engine making the engine become idle and then does a HW reset, before returning
+	 * to the host.
+	 * @see struct vpu_ipc_msg_payload_engine_reset
+	 */
 	VPU_JSM_MSG_ENGINE_RESET = VPU_JSM_MSG_ASYNC_CMD,
 	/**
 	 * Preempt engine. The NPU stops (preempts) all the jobs currently
@@ -448,10 +477,24 @@ enum vpu_ipc_msg_type {
 	 * the target engine, but it stops processing them (until the queue doorbell
 	 * is rung again); the host is responsible to reset the job queue, either
 	 * after preemption or when resubmitting jobs to the queue.
+	 * @see vpu_ipc_msg_payload_engine_preempt
 	 */
 	VPU_JSM_MSG_ENGINE_PREEMPT = 0x1101,
+	/**
+	 * OS scheduling doorbell register command
+	 * @see vpu_ipc_msg_payload_register_db
+	 */
 	VPU_JSM_MSG_REGISTER_DB = 0x1102,
+	/**
+	 * OS scheduling doorbell unregister command
+	 * @see vpu_ipc_msg_payload_unregister_db
+	 */
 	VPU_JSM_MSG_UNREGISTER_DB = 0x1103,
+	/**
+	 * Query engine heartbeat. Heartbeat is expected to increase monotonically
+	 * and increase while work is being progressed by NPU.
+	 * @see vpu_ipc_msg_payload_query_engine_hb
+	 */
 	VPU_JSM_MSG_QUERY_ENGINE_HB = 0x1104,
 	VPU_JSM_MSG_GET_POWER_LEVEL_COUNT = 0x1105,
 	VPU_JSM_MSG_GET_POWER_LEVEL = 0x1106,
@@ -477,6 +520,7 @@ enum vpu_ipc_msg_type {
 	 * aborted and removed from internal scheduling queues. All doorbells assigned
 	 * to the host_ssid are unregistered and any internal FW resources belonging to
 	 * the host_ssid are released.
+	 * @see vpu_ipc_msg_payload_ssid_release
 	 */
 	VPU_JSM_MSG_SSID_RELEASE = 0x110e,
 	/**
@@ -504,43 +548,78 @@ enum vpu_ipc_msg_type {
 	 * @see vpu_jsm_metric_streamer_start
 	 */
 	VPU_JSM_MSG_METRIC_STREAMER_INFO = 0x1112,
-	/** Control command: Priority band setup */
+	/**
+	 * Control command: Priority band setup
+	 * @see vpu_ipc_msg_payload_hws_priority_band_setup
+	 */
 	VPU_JSM_MSG_SET_PRIORITY_BAND_SETUP = 0x1113,
-	/** Control command: Create command queue */
+	/**
+	 * Control command: Create command queue
+	 * @see vpu_ipc_msg_payload_hws_create_cmdq
+	 */
 	VPU_JSM_MSG_CREATE_CMD_QUEUE = 0x1114,
-	/** Control command: Destroy command queue */
+	/**
+	 * Control command: Destroy command queue
+	 * @see vpu_ipc_msg_payload_hws_destroy_cmdq
+	 */
 	VPU_JSM_MSG_DESTROY_CMD_QUEUE = 0x1115,
-	/** Control command: Set context scheduling properties */
+	/**
+	 * Control command: Set context scheduling properties
+	 * @see vpu_ipc_msg_payload_hws_set_context_sched_properties
+	 */
 	VPU_JSM_MSG_SET_CONTEXT_SCHED_PROPERTIES = 0x1116,
-	/*
+	/**
 	 * Register a doorbell to notify VPU of new work. The doorbell may later be
 	 * deallocated or reassigned to another context.
+	 * @see vpu_jsm_hws_register_db
 	 */
 	VPU_JSM_MSG_HWS_REGISTER_DB = 0x1117,
-	/** Control command: Log buffer setting */
+	/**
+	 * Control command: Log buffer setting
+	 * @see vpu_ipc_msg_payload_hws_set_scheduling_log
+	 */
 	VPU_JSM_MSG_HWS_SET_SCHEDULING_LOG = 0x1118,
-	/* Control command: Suspend command queue. */
+	/**
+	 * Control command: Suspend command queue.
+	 * @see vpu_ipc_msg_payload_hws_suspend_cmdq
+	 */
 	VPU_JSM_MSG_HWS_SUSPEND_CMDQ = 0x1119,
-	/* Control command: Resume command queue */
+	/**
+	 * Control command: Resume command queue
+	 * @see vpu_ipc_msg_payload_hws_resume_cmdq
+	 */
 	VPU_JSM_MSG_HWS_RESUME_CMDQ = 0x111a,
-	/* Control command: Resume engine after reset */
+	/**
+	 * Control command: Resume engine after reset
+	 * @see vpu_ipc_msg_payload_hws_resume_engine
+	 */
 	VPU_JSM_MSG_HWS_ENGINE_RESUME = 0x111b,
-	/* Control command: Enable survivability/DCT mode */
+	/**
+	 * Control command: Enable survivability/DCT mode
+	 * @see vpu_ipc_msg_payload_pwr_dct_control
+	 */
 	VPU_JSM_MSG_DCT_ENABLE = 0x111c,
-	/* Control command: Disable survivability/DCT mode */
+	/**
+	 * Control command: Disable survivability/DCT mode
+	 * This command has no payload
+	 */
 	VPU_JSM_MSG_DCT_DISABLE = 0x111d,
 	/**
 	 * Dump VPU state. To be used for debug purposes only.
-	 * NOTE: Please introduce new ASYNC commands before this one. *
+	 * This command has no payload.
+	 * NOTE: Please introduce new ASYNC commands before this one.
 	 */
 	VPU_JSM_MSG_STATE_DUMP = 0x11FF,
 
-	/* IPC Host -> Device, General commands */
+	/** IPC Host -> Device, base id for general commands */
 	VPU_JSM_MSG_GENERAL_CMD = 0x1200,
+	/** Unsupported command */
 	VPU_JSM_MSG_BLOB_DEINIT_DEPRECATED = VPU_JSM_MSG_GENERAL_CMD,
 	/**
 	 * Control dyndbg behavior by executing a dyndbg command; equivalent to
-	 * Linux command: `echo '<dyndbg_cmd>' > <debugfs>/dynamic_debug/control`.
+	 * Linux command:
+	 * @verbatim echo '<dyndbg_cmd>' > <debugfs>/dynamic_debug/control @endverbatim
+	 * @see vpu_ipc_msg_payload_dyndbg_control
 	 */
 	VPU_JSM_MSG_DYNDBG_CONTROL = 0x1201,
 	/**
@@ -548,17 +627,35 @@ enum vpu_ipc_msg_type {
 	 */
 	VPU_JSM_MSG_PWR_D0I3_ENTER = 0x1202,
 
-	/* IPC Device -> Host, Job completion */
+	/**
+	 * IPC Device -> Host, Job completion
+	 * @see struct vpu_ipc_msg_payload_job_done
+	 */
 	VPU_JSM_MSG_JOB_DONE = 0x2100,
-	/* IPC Device -> Host, Fence signalled */
+	/**
+	 * IPC Device -> Host, Fence signalled
+	 * @see vpu_ipc_msg_payload_native_fence_signalled
+	 */
 	VPU_JSM_MSG_NATIVE_FENCE_SIGNALLED = 0x2101,
 
 	/* IPC Device -> Host, Async command completion */
 	VPU_JSM_MSG_ASYNC_CMD_DONE = 0x2200,
+	/**
+	 * IPC Device -> Host, engine reset complete
+	 * @see vpu_ipc_msg_payload_engine_reset_done
+	 */
 	VPU_JSM_MSG_ENGINE_RESET_DONE = VPU_JSM_MSG_ASYNC_CMD_DONE,
+	/**
+	 * Preempt complete message
+	 * @see vpu_ipc_msg_payload_engine_preempt_done
+	 */
 	VPU_JSM_MSG_ENGINE_PREEMPT_DONE = 0x2201,
 	VPU_JSM_MSG_REGISTER_DB_DONE = 0x2202,
 	VPU_JSM_MSG_UNREGISTER_DB_DONE = 0x2203,
+	/**
+	 * Response to query engine heartbeat.
+	 * @see vpu_ipc_msg_payload_query_engine_hb_done
+	 */
 	VPU_JSM_MSG_QUERY_ENGINE_HB_DONE = 0x2204,
 	VPU_JSM_MSG_GET_POWER_LEVEL_COUNT_DONE = 0x2205,
 	VPU_JSM_MSG_GET_POWER_LEVEL_DONE = 0x2206,
@@ -575,7 +672,10 @@ enum vpu_ipc_msg_type {
 	VPU_JSM_MSG_TRACE_GET_CAPABILITY_RSP = 0x220c,
 	/** Response to VPU_JSM_MSG_TRACE_GET_NAME. */
 	VPU_JSM_MSG_TRACE_GET_NAME_RSP = 0x220d,
-	/** Response to VPU_JSM_MSG_SSID_RELEASE. */
+	/**
+	 * Response to VPU_JSM_MSG_SSID_RELEASE.
+	 * @see vpu_ipc_msg_payload_ssid_release
+	 */
 	VPU_JSM_MSG_SSID_RELEASE_DONE = 0x220e,
 	/**
 	 * Response to VPU_JSM_MSG_METRIC_STREAMER_START.
@@ -605,37 +705,71 @@ enum vpu_ipc_msg_type {
 	/**
 	 * Asynchronous event sent from the VPU to the host either when the current
 	 * metric buffer is full or when the VPU has collected a multiple of
-	 * @notify_sample_count samples as indicated through the start command
-	 * (VPU_JSM_MSG_METRIC_STREAMER_START). Returns information about collected
-	 * metric data.
+	 * @ref vpu_jsm_metric_streamer_start::notify_sample_count samples as indicated
+	 * through the start command (VPU_JSM_MSG_METRIC_STREAMER_START). Returns
+	 * information about collected metric data.
 	 * @see vpu_jsm_metric_streamer_done
 	 */
 	VPU_JSM_MSG_METRIC_STREAMER_NOTIFICATION = 0x2213,
-	/** Response to control command: Priority band setup */
+	/**
+	 * Response to control command: Priority band setup
+	 * @see vpu_ipc_msg_payload_hws_priority_band_setup
+	 */
 	VPU_JSM_MSG_SET_PRIORITY_BAND_SETUP_RSP = 0x2214,
-	/** Response to control command: Create command queue */
+	/**
+	 * Response to control command: Create command queue
+	 * @see vpu_ipc_msg_payload_hws_create_cmdq_rsp
+	 */
 	VPU_JSM_MSG_CREATE_CMD_QUEUE_RSP = 0x2215,
-	/** Response to control command: Destroy command queue */
+	/**
+	 * Response to control command: Destroy command queue
+	 * @see vpu_ipc_msg_payload_hws_destroy_cmdq
+	 */
 	VPU_JSM_MSG_DESTROY_CMD_QUEUE_RSP = 0x2216,
-	/** Response to control command: Set context scheduling properties */
+	/**
+	 * Response to control command: Set context scheduling properties
+	 * @see vpu_ipc_msg_payload_hws_set_context_sched_properties
+	 */
 	VPU_JSM_MSG_SET_CONTEXT_SCHED_PROPERTIES_RSP = 0x2217,
-	/** Response to control command: Log buffer setting */
+	/**
+	 * Response to control command: Log buffer setting
+	 * @see vpu_ipc_msg_payload_hws_set_scheduling_log
+	 */
 	VPU_JSM_MSG_HWS_SET_SCHEDULING_LOG_RSP = 0x2218,
-	/* IPC Device -> Host, HWS notify index entry of log buffer written */
+	/**
+	 * IPC Device -> Host, HWS notify index entry of log buffer written
+	 * @see vpu_ipc_msg_payload_hws_scheduling_log_notification
+	 */
 	VPU_JSM_MSG_HWS_SCHEDULING_LOG_NOTIFICATION = 0x2219,
-	/* IPC Device -> Host, HWS completion of a context suspend request */
+	/**
+	 * IPC Device -> Host, HWS completion of a context suspend request
+	 * @see vpu_ipc_msg_payload_hws_suspend_cmdq
+	 */
 	VPU_JSM_MSG_HWS_SUSPEND_CMDQ_DONE = 0x221a,
-	/* Response to control command: Resume command queue */
+	/**
+	 * Response to control command: Resume command queue
+	 * @see vpu_ipc_msg_payload_hws_resume_cmdq
+	 */
 	VPU_JSM_MSG_HWS_RESUME_CMDQ_RSP = 0x221b,
-	/* Response to control command: Resume engine command response */
+	/**
+	 * Response to control command: Resume engine command response
+	 * @see vpu_ipc_msg_payload_hws_resume_engine
+	 */
 	VPU_JSM_MSG_HWS_RESUME_ENGINE_DONE = 0x221c,
-	/* Response to control command: Enable survivability/DCT mode */
+	/**
+	 * Response to control command: Enable survivability/DCT mode
+	 * This command has no payload
+	 */
 	VPU_JSM_MSG_DCT_ENABLE_DONE = 0x221d,
-	/* Response to control command: Disable survivability/DCT mode */
+	/**
+	 * Response to control command: Disable survivability/DCT mode
+	 * This command has no payload
+	 */
 	VPU_JSM_MSG_DCT_DISABLE_DONE = 0x221e,
 	/**
 	 * Response to state dump control command.
-	 * NOTE: Please introduce new ASYNC responses before this one. *
+	 * This command has no payload.
+	 * NOTE: Please introduce new ASYNC responses before this one.
 	 */
 	VPU_JSM_MSG_STATE_DUMP_RSP = 0x22FF,
 
@@ -653,57 +787,66 @@ enum vpu_ipc_msg_type {
 
 enum vpu_ipc_msg_status { VPU_JSM_MSG_FREE, VPU_JSM_MSG_ALLOCATED };
 
-/*
- * Host <-> LRT IPC message payload definitions
+/**
+ * Engine reset request payload
+ * @see VPU_JSM_MSG_ENGINE_RESET
  */
 struct vpu_ipc_msg_payload_engine_reset {
-	/* Engine to be reset. */
+	/** Engine to be reset. */
 	u32 engine_idx;
-	/* Reserved */
+	/** Reserved */
 	u32 reserved_0;
 };
 
+/**
+ * Engine preemption request struct
+ * @see VPU_JSM_MSG_ENGINE_PREEMPT
+ */
 struct vpu_ipc_msg_payload_engine_preempt {
-	/* Engine to be preempted. */
+	/** Engine to be preempted. */
 	u32 engine_idx;
-	/* ID of the preemption request. */
+	/** ID of the preemption request. */
 	u32 preempt_id;
 };
 
-/*
- * @brief Register doorbell command structure.
+/**
+ * Register doorbell command structure.
  * This structure supports doorbell registration for only OS scheduling.
  * @see VPU_JSM_MSG_REGISTER_DB
  */
 struct vpu_ipc_msg_payload_register_db {
-	/* Index of the doorbell to register. */
+	/** Index of the doorbell to register. */
 	u32 db_idx;
-	/* Reserved */
+	/** Reserved */
 	u32 reserved_0;
-	/* Virtual address in Global GTT pointing to the start of job queue. */
+	/** Virtual address in Global GTT pointing to the start of job queue. */
 	u64 jobq_base;
-	/* Size of the job queue in bytes. */
+	/** Size of the job queue in bytes. */
 	u32 jobq_size;
-	/* Host sub-stream ID for the context assigned to the doorbell. */
+	/** Host sub-stream ID for the context assigned to the doorbell. */
 	u32 host_ssid;
 };
 
 /**
- * @brief Unregister doorbell command structure.
+ * Unregister doorbell command structure.
  * Request structure to unregister a doorbell for both HW and OS scheduling.
  * @see VPU_JSM_MSG_UNREGISTER_DB
  */
 struct vpu_ipc_msg_payload_unregister_db {
-	/* Index of the doorbell to unregister. */
+	/** Index of the doorbell to unregister. */
 	u32 db_idx;
-	/* Reserved */
+	/** Reserved */
 	u32 reserved_0;
 };
 
+/**
+ * Heartbeat request structure
+ * @see VPU_JSM_MSG_QUERY_ENGINE_HB
+ */
 struct vpu_ipc_msg_payload_query_engine_hb {
-	/* Engine to return heartbeat value. */
+	/** Engine to return heartbeat value. */
 	u32 engine_idx;
-	/* Reserved */
+	/** Reserved */
 	u32 reserved_0;
 };
 
@@ -723,10 +866,14 @@ struct vpu_ipc_msg_payload_power_level {
 	u32 reserved_0;
 };
 
+/**
+ * Structure for requesting ssid release
+ * @see VPU_JSM_MSG_SSID_RELEASE
+ */
 struct vpu_ipc_msg_payload_ssid_release {
-	/* Host sub-stream ID for the context to be released. */
+	/** Host sub-stream ID for the context to be released. */
 	u32 host_ssid;
-	/* Reserved */
+	/** Reserved */
 	u32 reserved_0;
 };
 
@@ -752,7 +899,7 @@ struct vpu_jsm_metric_streamer_start {
 	u64 sampling_rate;
 	/**
 	 * If > 0 the VPU will send a VPU_JSM_MSG_METRIC_STREAMER_NOTIFICATION message
-	 * after every @notify_sample_count samples is collected or dropped by the VPU.
+	 * after every @ref notify_sample_count samples is collected or dropped by the VPU.
 	 * If set to UINT_MAX the VPU will only generate a notification when the metric
 	 * buffer is full. If set to 0 the VPU will never generate a notification.
 	 */
@@ -762,9 +909,9 @@ struct vpu_jsm_metric_streamer_start {
 	 * Address and size of the buffer where the VPU will write metric data. The
 	 * VPU writes all counters from enabled metric groups one after another. If
 	 * there is no space left to write data at the next sample period the VPU
-	 * will switch to the next buffer (@see next_buffer_addr) and will optionally
-	 * send a notification to the host driver if @notify_sample_count is non-zero.
-	 * If @next_buffer_addr is NULL the VPU will stop collecting metric data.
+	 * will switch to the next buffer (@ref next_buffer_addr) and will optionally
+	 * send a notification to the host driver if @ref notify_sample_count is non-zero.
+	 * If @ref next_buffer_addr is NULL the VPU will stop collecting metric data.
 	 */
 	u64 buffer_addr;
 	u64 buffer_size;
@@ -827,63 +974,80 @@ struct vpu_jsm_metric_streamer_update {
 	u64 next_buffer_size;
 };
 
+/**
+ * Device -> host job completion message.
+ * @see VPU_JSM_MSG_JOB_DONE
+ */
 struct vpu_ipc_msg_payload_job_done {
-	/* Engine to which the job was submitted. */
+	/** Engine to which the job was submitted. */
 	u32 engine_idx;
-	/* Index of the doorbell to which the job was submitted */
+	/** Index of the doorbell to which the job was submitted */
 	u32 db_idx;
-	/* ID of the completed job */
+	/** ID of the completed job */
 	u32 job_id;
-	/* Status of the completed job */
+	/** Status of the completed job */
 	u32 job_status;
-	/* Host SSID */
+	/** Host SSID */
 	u32 host_ssid;
-	/* Zero Padding */
+	/** Zero Padding */
 	u32 reserved_0;
-	/* Command queue id */
+	/** Command queue id */
 	u64 cmdq_id;
 };
 
-/*
+/**
  * Notification message upon native fence signalling.
  * @see VPU_JSM_MSG_NATIVE_FENCE_SIGNALLED
  */
 struct vpu_ipc_msg_payload_native_fence_signalled {
-	/* Engine ID. */
+	/** Engine ID. */
 	u32 engine_idx;
-	/* Host SSID. */
+	/** Host SSID. */
 	u32 host_ssid;
-	/* CMDQ ID */
+	/** CMDQ ID */
 	u64 cmdq_id;
-	/* Fence object handle. */
+	/** Fence object handle. */
 	u64 fence_handle;
 };
 
+/**
+ * vpu_ipc_msg_payload_engine_reset_done will contain an array of this structure
+ * which contains which queues caused reset if FW was able to detect any error.
+ * @see vpu_ipc_msg_payload_engine_reset_done
+ */
 struct vpu_jsm_engine_reset_context {
-	/* Host SSID */
+	/** Host SSID */
 	u32 host_ssid;
-	/* Zero Padding */
+	/** Zero Padding */
 	u32 reserved_0;
-	/* Command queue id */
+	/** Command queue id */
 	u64 cmdq_id;
-	/* See VPU_ENGINE_RESET_CONTEXT_* defines */
+	/** See VPU_ENGINE_RESET_CONTEXT_* defines */
 	u64 flags;
 };
 
+/**
+ * Engine reset response.
+ * @see VPU_JSM_MSG_ENGINE_RESET_DONE
+ */
 struct vpu_ipc_msg_payload_engine_reset_done {
-	/* Engine ordinal */
+	/** Engine ordinal */
 	u32 engine_idx;
-	/* Number of impacted contexts */
+	/** Number of impacted contexts */
 	u32 num_impacted_contexts;
-	/* Array of impacted command queue ids and their flags */
+	/** Array of impacted command queue ids and their flags */
 	struct vpu_jsm_engine_reset_context
 		impacted_contexts[VPU_MAX_ENGINE_RESET_IMPACTED_CONTEXTS];
 };
 
+/**
+ * Preemption response struct
+ * @see VPU_JSM_MSG_ENGINE_PREEMPT_DONE
+ */
 struct vpu_ipc_msg_payload_engine_preempt_done {
-	/* Engine preempted. */
+	/** Engine preempted. */
 	u32 engine_idx;
-	/* ID of the preemption request. */
+	/** ID of the preemption request. */
 	u32 preempt_id;
 };
 
@@ -912,12 +1076,16 @@ struct vpu_ipc_msg_payload_unregister_db_done {
 	u32 reserved_0;
 };
 
+/**
+ * Structure for heartbeat response
+ * @see VPU_JSM_MSG_QUERY_ENGINE_HB_DONE
+ */
 struct vpu_ipc_msg_payload_query_engine_hb_done {
-	/* Engine returning heartbeat value. */
+	/** Engine returning heartbeat value. */
 	u32 engine_idx;
-	/* Reserved */
+	/** Reserved */
 	u32 reserved_0;
-	/* Heartbeat value. */
+	/** Heartbeat value. */
 	u64 heartbeat;
 };
 
@@ -937,7 +1105,10 @@ struct vpu_ipc_msg_payload_get_power_level_count_done {
 	u8 power_limit[16];
 };
 
-/* HWS priority band setup request / response */
+/**
+ * HWS priority band setup request / response
+ * @see VPU_JSM_MSG_SET_PRIORITY_BAND_SETUP
+ */
 struct vpu_ipc_msg_payload_hws_priority_band_setup {
 	/*
 	 * Grace period in 100ns units when preempting another priority band for
@@ -964,15 +1135,23 @@ struct vpu_ipc_msg_payload_hws_priority_band_setup {
 	 * TDR timeout value in milliseconds. Default value of 0 meaning no timeout.
 	 */
 	u32 tdr_timeout;
+	/* Non-interactive queue timeout for no progress of heartbeat in milliseconds.
+	 * Default value of 0 meaning no timeout.
+	 */
+	u32 non_interactive_no_progress_timeout;
+	/*
+	 * Non-interactive queue upper limit timeout value in milliseconds. Default
+	 * value of 0 meaning no timeout.
+	 */
+	u32 non_interactive_timeout;
 };
 
-/*
+/**
  * @brief HWS create command queue request.
  * Host will create a command queue via this command.
  * Note: Cmdq group is a handle of an object which
  * may contain one or more command queues.
  * @see VPU_JSM_MSG_CREATE_CMD_QUEUE
- * @see VPU_JSM_MSG_CREATE_CMD_QUEUE_RSP
  */
 struct vpu_ipc_msg_payload_hws_create_cmdq {
 	/* Process id */
@@ -993,66 +1172,73 @@ struct vpu_ipc_msg_payload_hws_create_cmdq {
 	u32 reserved_0;
 };
 
-/*
- * @brief HWS create command queue response.
- * @see VPU_JSM_MSG_CREATE_CMD_QUEUE
+/**
+ * HWS create command queue response.
  * @see VPU_JSM_MSG_CREATE_CMD_QUEUE_RSP
  */
 struct vpu_ipc_msg_payload_hws_create_cmdq_rsp {
-	/* Process id */
+	/** Process id */
 	u64 process_id;
-	/* Host SSID */
+	/** Host SSID */
 	u32 host_ssid;
-	/* Engine for which queue is being created */
+	/** Engine for which queue is being created */
 	u32 engine_idx;
-	/* Command queue group */
+	/** Command queue group */
 	u64 cmdq_group;
-	/* Command queue id */
+	/** Command queue id */
 	u64 cmdq_id;
 };
 
-/* HWS destroy command queue request / response */
+/**
+ * HWS destroy command queue request / response
+ * @see VPU_JSM_MSG_DESTROY_CMD_QUEUE
+ * @see VPU_JSM_MSG_DESTROY_CMD_QUEUE_RSP
+ */
 struct vpu_ipc_msg_payload_hws_destroy_cmdq {
-	/* Host SSID */
+	/** Host SSID */
 	u32 host_ssid;
-	/* Zero Padding */
+	/** Zero Padding */
 	u32 reserved;
-	/* Command queue id */
+	/** Command queue id */
 	u64 cmdq_id;
 };
 
-/* HWS set context scheduling properties request / response */
+/**
+ * HWS set context scheduling properties request / response
+ * @see VPU_JSM_MSG_SET_CONTEXT_SCHED_PROPERTIES
+ * @see VPU_JSM_MSG_SET_CONTEXT_SCHED_PROPERTIES_RSP
+ */
 struct vpu_ipc_msg_payload_hws_set_context_sched_properties {
-	/* Host SSID */
+	/** Host SSID */
 	u32 host_ssid;
-	/* Zero Padding */
+	/** Zero Padding */
 	u32 reserved_0;
-	/* Command queue id */
+	/** Command queue id */
 	u64 cmdq_id;
-	/*
+	/**
 	 * Priority band to assign to work of this context.
 	 * Available priority bands: @see enum vpu_job_scheduling_priority_band
 	 */
 	u32 priority_band;
-	/* Inside realtime band assigns a further priority */
+	/** Inside realtime band assigns a further priority */
 	u32 realtime_priority_level;
-	/* Priority relative to other contexts in the same process */
+	/** Priority relative to other contexts in the same process */
 	s32 in_process_priority;
-	/* Zero padding / Reserved */
+	/** Zero padding / Reserved */
 	u32 reserved_1;
-	/*
+	/**
 	 * Context quantum relative to other contexts of same priority in the same process
 	 * Minimum value supported by NPU is 1ms (10000 in 100ns units).
 	 */
 	u64 context_quantum;
-	/* Grace period when preempting context of the same priority within the same process */
+	/** Grace period when preempting context of the same priority within the same process */
 	u64 grace_period_same_priority;
-	/* Grace period when preempting context of a lower priority within the same process */
+	/** Grace period when preempting context of a lower priority within the same process */
 	u64 grace_period_lower_priority;
 };
 
-/*
- * @brief Register doorbell command structure.
+/**
+ * Register doorbell command structure.
  * This structure supports doorbell registration for both HW and OS scheduling.
  * Note: Queue base and size are added here so that the same structure can be used for
  * OS scheduling and HW scheduling. For OS scheduling, cmdq_id will be ignored
@@ -1061,27 +1247,27 @@ struct vpu_ipc_msg_payload_hws_set_context_sched_properties {
  * @see VPU_JSM_MSG_HWS_REGISTER_DB
  */
 struct vpu_jsm_hws_register_db {
-	/* Index of the doorbell to register. */
+	/** Index of the doorbell to register. */
 	u32 db_id;
-	/* Host sub-stream ID for the context assigned to the doorbell. */
+	/** Host sub-stream ID for the context assigned to the doorbell. */
 	u32 host_ssid;
-	/* ID of the command queue associated with the doorbell. */
+	/** ID of the command queue associated with the doorbell. */
 	u64 cmdq_id;
-	/* Virtual address pointing to the start of command queue. */
+	/** Virtual address pointing to the start of command queue. */
 	u64 cmdq_base;
-	/* Size of the command queue in bytes. */
+	/** Size of the command queue in bytes. */
 	u64 cmdq_size;
 };
 
-/*
- * @brief Structure to set another buffer to be used for scheduling-related logging.
+/**
+ * Structure to set another buffer to be used for scheduling-related logging.
  * The size of the logging buffer and the number of entries is defined as part of the
  * buffer itself as described next.
  * The log buffer received from the host is made up of;
- *   - header:     32 bytes in size, as shown in 'struct vpu_hws_log_buffer_header'.
+ *   - header:     32 bytes in size, as shown in @ref vpu_hws_log_buffer_header.
  *                 The header contains the number of log entries in the buffer.
  *   - log entry:  0 to n-1, each log entry is 32 bytes in size, as shown in
- *                 'struct vpu_hws_log_buffer_entry'.
+ *                 @ref vpu_hws_log_buffer_entry.
  *                 The entry contains the VPU timestamp, operation type and data.
  * The host should provide the notify index value of log buffer to VPU. This is a
  * value defined within the log buffer and when written to will generate the
@@ -1095,30 +1281,30 @@ struct vpu_jsm_hws_register_db {
  * @see VPU_JSM_MSG_HWS_SCHEDULING_LOG_NOTIFICATION
  */
 struct vpu_ipc_msg_payload_hws_set_scheduling_log {
-	/* Engine ordinal */
+	/** Engine ordinal */
 	u32 engine_idx;
-	/* Host SSID */
+	/** Host SSID */
 	u32 host_ssid;
-	/*
+	/**
 	 * VPU log buffer virtual address.
 	 * Set to 0 to disable logging for this engine.
 	 */
 	u64 vpu_log_buffer_va;
-	/*
+	/**
 	 * Notify index of log buffer. VPU_JSM_MSG_HWS_SCHEDULING_LOG_NOTIFICATION
 	 * is generated when an event log is written to this index.
 	 */
 	u64 notify_index;
-	/*
+	/**
 	 * Field is now deprecated, will be removed when KMD is updated to support removal
 	 */
 	u32 enable_extra_events;
-	/* Zero Padding */
+	/** Zero Padding */
 	u32 reserved_0;
 };
 
-/*
- * @brief The scheduling log notification is generated by VPU when it writes
+/**
+ * The scheduling log notification is generated by VPU when it writes
  * an event into the log buffer at the notify_index. VPU notifies host with
  * VPU_JSM_MSG_HWS_SCHEDULING_LOG_NOTIFICATION. This is an asynchronous
  * message from VPU to host.
@@ -1126,14 +1312,14 @@ struct vpu_ipc_msg_payload_hws_set_scheduling_log {
  * @see VPU_JSM_MSG_HWS_SET_SCHEDULING_LOG
  */
 struct vpu_ipc_msg_payload_hws_scheduling_log_notification {
-	/* Engine ordinal */
+	/** Engine ordinal */
 	u32 engine_idx;
-	/* Zero Padding */
+	/** Zero Padding */
 	u32 reserved_0;
 };
 
-/*
- * @brief HWS suspend command queue request and done structure.
+/**
+ * HWS suspend command queue request and done structure.
  * Host will request the suspend of contexts and VPU will;
  *   - Suspend all work on this context
  *   - Preempt any running work
@@ -1152,21 +1338,21 @@ struct vpu_ipc_msg_payload_hws_scheduling_log_notification {
  * @see VPU_JSM_MSG_HWS_SUSPEND_CMDQ_DONE
  */
 struct vpu_ipc_msg_payload_hws_suspend_cmdq {
-	/* Host SSID */
+	/** Host SSID */
 	u32 host_ssid;
-	/* Zero Padding */
+	/** Zero Padding */
 	u32 reserved_0;
-	/* Command queue id */
+	/** Command queue id */
 	u64 cmdq_id;
-	/*
+	/**
 	 * Suspend fence value - reported by the VPU suspend context
 	 * completed once suspend is complete.
 	 */
 	u64 suspend_fence_value;
 };
 
-/*
- * @brief HWS Resume command queue request / response structure.
+/**
+ * HWS Resume command queue request / response structure.
  * Host will request the resume of a context;
  *  - VPU will resume all work on this context
  *  - Scheduler will allow this context to be scheduled
@@ -1174,25 +1360,25 @@ struct vpu_ipc_msg_payload_hws_suspend_cmdq {
  * @see VPU_JSM_MSG_HWS_RESUME_CMDQ_RSP
  */
 struct vpu_ipc_msg_payload_hws_resume_cmdq {
-	/* Host SSID */
+	/** Host SSID */
 	u32 host_ssid;
-	/* Zero Padding */
+	/** Zero Padding */
 	u32 reserved_0;
-	/* Command queue id */
+	/** Command queue id */
 	u64 cmdq_id;
 };
 
-/*
- * @brief HWS Resume engine request / response structure.
- * After a HWS engine reset, all scheduling is stopped on VPU until a engine resume.
+/**
+ * HWS Resume engine request / response structure.
+ * After a HWS engine reset, all scheduling is stopped on VPU until an engine resume.
  * Host shall send this command to resume scheduling of any valid queue.
- * @see VPU_JSM_MSG_HWS_RESUME_ENGINE
+ * @see VPU_JSM_MSG_HWS_ENGINE_RESUME
  * @see VPU_JSM_MSG_HWS_RESUME_ENGINE_DONE
  */
 struct vpu_ipc_msg_payload_hws_resume_engine {
-	/* Engine to be resumed */
+	/** Engine to be resumed */
 	u32 engine_idx;
-	/* Reserved */
+	/** Reserved */
 	u32 reserved_0;
 };
 
@@ -1326,7 +1512,7 @@ struct vpu_jsm_metric_streamer_done {
 /**
  * Metric group description placed in the metric buffer after successful completion
  * of the VPU_JSM_MSG_METRIC_STREAMER_INFO command. This is followed by one or more
- * @vpu_jsm_metric_counter_descriptor records.
+ * @ref vpu_jsm_metric_counter_descriptor records.
  * @see VPU_JSM_MSG_METRIC_STREAMER_INFO
  */
 struct vpu_jsm_metric_group_descriptor {
@@ -1413,29 +1599,24 @@ struct vpu_jsm_metric_counter_descriptor {
 };
 
 /**
- * Payload for VPU_JSM_MSG_DYNDBG_CONTROL requests.
+ * Payload for @ref VPU_JSM_MSG_DYNDBG_CONTROL requests.
  *
- * VPU_JSM_MSG_DYNDBG_CONTROL are used to control the VPU FW Dynamic Debug
- * feature, which allows developers to selectively enable / disable MVLOG_DEBUG
- * messages. This is equivalent to the Dynamic Debug functionality provided by
- * Linux
- * (https://www.kernel.org/doc/html/latest/admin-guide/dynamic-debug-howto.html)
- * The host can control Dynamic Debug behavior by sending dyndbg commands, which
- * have the same syntax as Linux
- * dyndbg commands.
+ * VPU_JSM_MSG_DYNDBG_CONTROL requests are used to control the VPU FW dynamic debug
+ * feature, which allows developers to selectively enable/disable code to obtain
+ * additional FW information. This is equivalent to the dynamic debug functionality
+ * provided by Linux. The host can control dynamic debug behavior by sending dyndbg
+ * commands, using the same syntax as for Linux dynamic debug commands.
  *
- * NOTE: in order for MVLOG_DEBUG messages to be actually printed, the host
- * still has to set the logging level to MVLOG_DEBUG, using the
- * VPU_JSM_MSG_TRACE_SET_CONFIG command.
+ * @see https://www.kernel.org/doc/html/latest/admin-guide/dynamic-debug-howto.html.
  *
- * The host can see the current dynamic debug configuration by executing a
- * special 'show' command. The dyndbg configuration will be printed to the
- * configured logging destination using MVLOG_INFO logging level.
+ * NOTE:
+ * As the dynamic debug feature uses MVLOG messages to provide information, the host
+ * must first set the logging level to MVLOG_DEBUG, using the @ref VPU_JSM_MSG_TRACE_SET_CONFIG
+ * command.
  */
 struct vpu_ipc_msg_payload_dyndbg_control {
 	/**
-	 * Dyndbg command (same format as Linux dyndbg); must be a NULL-terminated
-	 * string.
+	 * Dyndbg command to be executed.
 	 */
 	char dyndbg_cmd[VPU_DYNDBG_CMD_MAX_LEN];
 };
@@ -1456,7 +1637,7 @@ struct vpu_ipc_msg_payload_pwr_d0i3_enter {
 };
 
 /**
- * Payload for VPU_JSM_MSG_DCT_ENABLE message.
+ * Payload for @ref VPU_JSM_MSG_DCT_ENABLE message.
  *
  * Default values for DCT active/inactive times are 5.3ms and 30ms respectively,
  * corresponding to a 85% duty cycle. This payload allows the host to tune these
@@ -1513,28 +1694,28 @@ union vpu_ipc_msg_payload {
 	struct vpu_ipc_msg_payload_pwr_dct_control pwr_dct_control;
 };
 
-/*
- * Host <-> LRT IPC message base structure.
+/**
+ * Host <-> NPU IPC message base structure.
  *
  * NOTE: All instances of this object must be aligned on a 64B boundary
  * to allow proper handling of VPU cache operations.
  */
 struct vpu_jsm_msg {
-	/* Reserved */
+	/** Reserved */
 	u64 reserved_0;
-	/* Message type, see vpu_ipc_msg_type enum. */
+	/** Message type, see @ref vpu_ipc_msg_type. */
 	u32 type;
-	/* Buffer status, see vpu_ipc_msg_status enum. */
+	/** Buffer status, see @ref vpu_ipc_msg_status. */
 	u32 status;
-	/*
+	/**
 	 * Request ID, provided by the host in a request message and passed
 	 * back by VPU in the response message.
 	 */
 	u32 request_id;
-	/* Request return code set by the VPU, see VPU_JSM_STATUS_* defines. */
+	/** Request return code set by the VPU, see VPU_JSM_STATUS_* defines. */
 	u32 result;
 	u64 reserved_1;
-	/* Message payload depending on message type, see vpu_ipc_msg_payload union. */
+	/** Message payload depending on message type, see vpu_ipc_msg_payload union. */
 	union vpu_ipc_msg_payload payload;
 };
 
diff --git a/drivers/accel/qaic/Kconfig b/drivers/accel/qaic/Kconfig
index 5e405a19c157..116e42d152ca 100644
--- a/drivers/accel/qaic/Kconfig
+++ b/drivers/accel/qaic/Kconfig
@@ -9,6 +9,7 @@ config DRM_ACCEL_QAIC
 	depends on PCI && HAS_IOMEM
 	depends on MHI_BUS
 	select CRC32
+	select WANT_DEV_COREDUMP
 	help
 	  Enables driver for Qualcomm's Cloud AI accelerator PCIe cards that are
 	  designed to accelerate Deep Learning inference workloads.
diff --git a/drivers/accel/qaic/Makefile b/drivers/accel/qaic/Makefile
index 35e883515629..71f727b74da3 100644
--- a/drivers/accel/qaic/Makefile
+++ b/drivers/accel/qaic/Makefile
@@ -10,6 +10,9 @@ qaic-y := \
 	qaic_control.o \
 	qaic_data.o \
 	qaic_drv.o \
+	qaic_ras.o \
+	qaic_ssr.o \
+	qaic_sysfs.o \
 	qaic_timesync.o \
 	sahara.o
 
diff --git a/drivers/accel/qaic/qaic.h b/drivers/accel/qaic/qaic.h
index 0dbb8e32e4b9..fa7a8155658c 100644
--- a/drivers/accel/qaic/qaic.h
+++ b/drivers/accel/qaic/qaic.h
@@ -21,6 +21,7 @@
 
 #define QAIC_DBC_BASE		SZ_128K
 #define QAIC_DBC_SIZE		SZ_4K
+#define QAIC_SSR_DBC_SENTINEL	U32_MAX /* No ongoing SSR sentinel */
 
 #define QAIC_NO_PARTITION	-1
 
@@ -47,6 +48,22 @@ enum __packed dev_states {
 	QAIC_ONLINE,
 };
 
+enum dbc_states {
+	/* DBC is free and can be activated */
+	DBC_STATE_IDLE,
+	/* DBC is activated and a workload is running on device */
+	DBC_STATE_ASSIGNED,
+	/* Sub-system associated with this workload has crashed and it will shutdown soon */
+	DBC_STATE_BEFORE_SHUTDOWN,
+	/* Sub-system associated with this workload has crashed and it has shutdown */
+	DBC_STATE_AFTER_SHUTDOWN,
+	/* Sub-system associated with this workload is shutdown and it will be powered up soon */
+	DBC_STATE_BEFORE_POWER_UP,
+	/* Sub-system associated with this workload is now powered up */
+	DBC_STATE_AFTER_POWER_UP,
+	DBC_STATE_MAX,
+};
+
 extern bool datapath_polling;
 
 struct qaic_user {
@@ -97,6 +114,8 @@ struct dma_bridge_chan {
 	 * response queue's head and tail pointer of this DBC.
 	 */
 	void __iomem		*dbc_base;
+	/* Synchronizes access to Request queue's head and tail pointer */
+	struct mutex		req_lock;
 	/* Head of list where each node is a memory handle queued in request queue */
 	struct list_head	xfer_list;
 	/* Synchronizes DBC readers during cleanup */
@@ -112,6 +131,8 @@ struct dma_bridge_chan {
 	unsigned int		irq;
 	/* Polling work item to simulate interrupts */
 	struct work_struct	poll_work;
+	/* Represents various states of this DBC from enum dbc_states */
+	unsigned int		state;
 };
 
 struct qaic_device {
@@ -159,6 +180,8 @@ struct qaic_device {
 	struct mhi_device	*qts_ch;
 	/* Work queue for tasks related to MHI "QAIC_TIMESYNC" channel */
 	struct workqueue_struct	*qts_wq;
+	/* MHI "QAIC_TIMESYNC_PERIODIC" channel device */
+	struct mhi_device	*mqts_ch;
 	/* Head of list of page allocated by MHI bootlog device */
 	struct list_head        bootlog;
 	/* MHI bootlog channel device */
@@ -167,6 +190,22 @@ struct qaic_device {
 	struct workqueue_struct *bootlog_wq;
 	/* Synchronizes access of pages in MHI bootlog device */
 	struct mutex            bootlog_mutex;
+	/* MHI RAS channel device */
+	struct mhi_device	*ras_ch;
+	/* Correctable error count */
+	unsigned int		ce_count;
+	/* Un-correctable error count */
+	unsigned int		ue_count;
+	/* Un-correctable non-fatal error count */
+	unsigned int		ue_nf_count;
+	/* MHI SSR channel device */
+	struct mhi_device	*ssr_ch;
+	/* Work queue for tasks related to MHI SSR device */
+	struct workqueue_struct	*ssr_wq;
+	/* Buffer to collect SSR crashdump via SSR MHI channel */
+	void			*ssr_mhi_buf;
+	/* DBC which is under SSR. Sentinel U32_MAX would mean that no SSR in progress */
+	u32			ssr_dbc;
 };
 
 struct qaic_drm_device {
@@ -185,6 +224,8 @@ struct qaic_drm_device {
 	struct list_head	users;
 	/* Synchronizes access to users list */
 	struct mutex		users_mutex;
+	/* Pointer to array of DBC sysfs attributes */
+	void			*sysfs_attrs;
 };
 
 struct qaic_bo {
@@ -213,8 +254,6 @@ struct qaic_bo {
 	bool			sliced;
 	/* Request ID of this BO if it is queued for execution */
 	u16			req_id;
-	/* Handle assigned to this BO */
-	u32			handle;
 	/* Wait on this for completion of DMA transfer of this BO */
 	struct completion	xfer_done;
 	/*
@@ -309,6 +348,13 @@ int qaic_partial_execute_bo_ioctl(struct drm_device *dev, void *data, struct drm
 int qaic_wait_bo_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv);
 int qaic_perf_stats_bo_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv);
 int qaic_detach_slice_bo_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv);
-void irq_polling_work(struct work_struct *work);
+void qaic_irq_polling_work(struct work_struct *work);
+void qaic_dbc_enter_ssr(struct qaic_device *qdev, u32 dbc_id);
+void qaic_dbc_exit_ssr(struct qaic_device *qdev);
+
+/* qaic_sysfs.c */
+int qaic_sysfs_init(struct qaic_drm_device *qddev);
+void qaic_sysfs_remove(struct qaic_drm_device *qddev);
+void set_dbc_state(struct qaic_device *qdev, u32 dbc_id, unsigned int state);
 
 #endif /* _QAIC_H_ */
diff --git a/drivers/accel/qaic/qaic_control.c b/drivers/accel/qaic/qaic_control.c
index d8bdab69f800..428d8f65bff3 100644
--- a/drivers/accel/qaic/qaic_control.c
+++ b/drivers/accel/qaic/qaic_control.c
@@ -17,6 +17,7 @@
 #include <linux/overflow.h>
 #include <linux/pci.h>
 #include <linux/scatterlist.h>
+#include <linux/sched/signal.h>
 #include <linux/types.h>
 #include <linux/uaccess.h>
 #include <linux/workqueue.h>
@@ -30,7 +31,7 @@
 #define MANAGE_MAGIC_NUMBER		((__force __le32)0x43494151) /* "QAIC" in little endian */
 #define QAIC_DBC_Q_GAP			SZ_256
 #define QAIC_DBC_Q_BUF_ALIGN		SZ_4K
-#define QAIC_MANAGE_EXT_MSG_LENGTH	SZ_64K /* Max DMA message length */
+#define QAIC_MANAGE_WIRE_MSG_LENGTH	SZ_64K /* Max DMA message length */
 #define QAIC_WRAPPER_MAX_SIZE		SZ_4K
 #define QAIC_MHI_RETRY_WAIT_MS		100
 #define QAIC_MHI_RETRY_MAX		20
@@ -309,6 +310,7 @@ static void save_dbc_buf(struct qaic_device *qdev, struct ioctl_resources *resou
 		enable_dbc(qdev, dbc_id, usr);
 		qdev->dbc[dbc_id].in_use = true;
 		resources->buf = NULL;
+		set_dbc_state(qdev, dbc_id, DBC_STATE_ASSIGNED);
 	}
 }
 
@@ -367,7 +369,7 @@ static int encode_passthrough(struct qaic_device *qdev, void *trans, struct wrap
 	if (in_trans->hdr.len % 8 != 0)
 		return -EINVAL;
 
-	if (size_add(msg_hdr_len, in_trans->hdr.len) > QAIC_MANAGE_EXT_MSG_LENGTH)
+	if (size_add(msg_hdr_len, in_trans->hdr.len) > QAIC_MANAGE_WIRE_MSG_LENGTH)
 		return -ENOSPC;
 
 	trans_wrapper = add_wrapper(wrappers,
@@ -407,7 +409,7 @@ static int find_and_map_user_pages(struct qaic_device *qdev,
 		return -EINVAL;
 	remaining = in_trans->size - resources->xferred_dma_size;
 	if (remaining == 0)
-		return 0;
+		return -EINVAL;
 
 	if (check_add_overflow(xfer_start_addr, remaining, &end))
 		return -EINVAL;
@@ -495,7 +497,7 @@ static int encode_addr_size_pairs(struct dma_xfer *xfer, struct wrapper_list *wr
 
 	nents = sgt->nents;
 	nents_dma = nents;
-	*size = QAIC_MANAGE_EXT_MSG_LENGTH - msg_hdr_len - sizeof(**out_trans);
+	*size = QAIC_MANAGE_WIRE_MSG_LENGTH - msg_hdr_len - sizeof(**out_trans);
 	for_each_sgtable_dma_sg(sgt, sg, i) {
 		*size -= sizeof(*asp);
 		/* Save 1K for possible follow-up transactions. */
@@ -576,7 +578,7 @@ static int encode_dma(struct qaic_device *qdev, void *trans, struct wrapper_list
 
 	/* There should be enough space to hold at least one ASP entry. */
 	if (size_add(msg_hdr_len, sizeof(*out_trans) + sizeof(struct wire_addr_size_pair)) >
-	    QAIC_MANAGE_EXT_MSG_LENGTH)
+	    QAIC_MANAGE_WIRE_MSG_LENGTH)
 		return -ENOMEM;
 
 	xfer = kmalloc(sizeof(*xfer), GFP_KERNEL);
@@ -645,7 +647,7 @@ static int encode_activate(struct qaic_device *qdev, void *trans, struct wrapper
 	msg = &wrapper->msg;
 	msg_hdr_len = le32_to_cpu(msg->hdr.len);
 
-	if (size_add(msg_hdr_len, sizeof(*out_trans)) > QAIC_MANAGE_MAX_MSG_LENGTH)
+	if (size_add(msg_hdr_len, sizeof(*out_trans)) > QAIC_MANAGE_WIRE_MSG_LENGTH)
 		return -ENOSPC;
 
 	if (!in_trans->queue_size)
@@ -655,8 +657,9 @@ static int encode_activate(struct qaic_device *qdev, void *trans, struct wrapper
 		return -EINVAL;
 
 	nelem = in_trans->queue_size;
-	size = (get_dbc_req_elem_size() + get_dbc_rsp_elem_size()) * nelem;
-	if (size / nelem != get_dbc_req_elem_size() + get_dbc_rsp_elem_size())
+	if (check_mul_overflow((u32)(get_dbc_req_elem_size() + get_dbc_rsp_elem_size()),
+			       nelem,
+			       &size))
 		return -EINVAL;
 
 	if (size + QAIC_DBC_Q_GAP + QAIC_DBC_Q_BUF_ALIGN < size)
@@ -729,7 +732,7 @@ static int encode_status(struct qaic_device *qdev, void *trans, struct wrapper_l
 	msg = &wrapper->msg;
 	msg_hdr_len = le32_to_cpu(msg->hdr.len);
 
-	if (size_add(msg_hdr_len, in_trans->hdr.len) > QAIC_MANAGE_MAX_MSG_LENGTH)
+	if (size_add(msg_hdr_len, in_trans->hdr.len) > QAIC_MANAGE_WIRE_MSG_LENGTH)
 		return -ENOSPC;
 
 	trans_wrapper = add_wrapper(wrappers, sizeof(*trans_wrapper));
@@ -810,7 +813,7 @@ static int encode_message(struct qaic_device *qdev, struct manage_msg *user_msg,
 		}
 
 		if (ret)
-			break;
+			goto out;
 	}
 
 	if (user_len != user_msg->len)
@@ -921,6 +924,7 @@ static int decode_deactivate(struct qaic_device *qdev, void *trans, u32 *msg_len
 	}
 
 	release_dbc(qdev, dbc_id);
+	set_dbc_state(qdev, dbc_id, DBC_STATE_IDLE);
 	*msg_len += sizeof(*in_trans);
 
 	return 0;
@@ -1052,7 +1056,7 @@ static void *msg_xfer(struct qaic_device *qdev, struct wrapper_list *wrappers, u
 	init_completion(&elem.xfer_done);
 	if (likely(!qdev->cntl_lost_buf)) {
 		/*
-		 * The max size of request to device is QAIC_MANAGE_EXT_MSG_LENGTH.
+		 * The max size of request to device is QAIC_MANAGE_WIRE_MSG_LENGTH.
 		 * The max size of response from device is QAIC_MANAGE_MAX_MSG_LENGTH.
 		 */
 		out_buf = kmalloc(QAIC_MANAGE_MAX_MSG_LENGTH, GFP_KERNEL);
@@ -1079,7 +1083,6 @@ static void *msg_xfer(struct qaic_device *qdev, struct wrapper_list *wrappers, u
 
 	list_for_each_entry(w, &wrappers->list, list) {
 		kref_get(&w->ref_count);
-		retry_count = 0;
 		ret = mhi_queue_buf(qdev->cntl_ch, DMA_TO_DEVICE, &w->msg, w->len,
 				    list_is_last(&w->list, &wrappers->list) ? MHI_EOT : MHI_CHAIN);
 		if (ret) {
diff --git a/drivers/accel/qaic/qaic_data.c b/drivers/accel/qaic/qaic_data.c
index 1bce1af7c72c..60cb4d65d48e 100644
--- a/drivers/accel/qaic/qaic_data.c
+++ b/drivers/accel/qaic/qaic_data.c
@@ -18,6 +18,7 @@
 #include <linux/scatterlist.h>
 #include <linux/spinlock.h>
 #include <linux/srcu.h>
+#include <linux/string.h>
 #include <linux/types.h>
 #include <linux/uaccess.h>
 #include <linux/wait.h>
@@ -165,7 +166,7 @@ static void free_slice(struct kref *kref)
 	drm_gem_object_put(&slice->bo->base);
 	sg_free_table(slice->sgt);
 	kfree(slice->sgt);
-	kfree(slice->reqs);
+	kvfree(slice->reqs);
 	kfree(slice);
 }
 
@@ -404,7 +405,7 @@ static int qaic_map_one_slice(struct qaic_device *qdev, struct qaic_bo *bo,
 		goto free_sgt;
 	}
 
-	slice->reqs = kcalloc(sgt->nents, sizeof(*slice->reqs), GFP_KERNEL);
+	slice->reqs = kvcalloc(sgt->nents, sizeof(*slice->reqs), GFP_KERNEL);
 	if (!slice->reqs) {
 		ret = -ENOMEM;
 		goto free_slice;
@@ -430,7 +431,7 @@ static int qaic_map_one_slice(struct qaic_device *qdev, struct qaic_bo *bo,
 	return 0;
 
 free_req:
-	kfree(slice->reqs);
+	kvfree(slice->reqs);
 free_slice:
 	kfree(slice);
 free_sgt:
@@ -643,8 +644,36 @@ static void qaic_free_object(struct drm_gem_object *obj)
 	kfree(bo);
 }
 
+static struct sg_table *qaic_get_sg_table(struct drm_gem_object *obj)
+{
+	struct qaic_bo *bo = to_qaic_bo(obj);
+	struct scatterlist *sg, *sg_in;
+	struct sg_table *sgt, *sgt_in;
+	int i;
+
+	sgt_in = bo->sgt;
+
+	sgt = kmalloc(sizeof(*sgt), GFP_KERNEL);
+	if (!sgt)
+		return ERR_PTR(-ENOMEM);
+
+	if (sg_alloc_table(sgt, sgt_in->orig_nents, GFP_KERNEL)) {
+		kfree(sgt);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	sg = sgt->sgl;
+	for_each_sgtable_sg(sgt_in, sg_in, i) {
+		memcpy(sg, sg_in, sizeof(*sg));
+		sg = sg_next(sg);
+	}
+
+	return sgt;
+}
+
 static const struct drm_gem_object_funcs qaic_gem_funcs = {
 	.free = qaic_free_object,
+	.get_sg_table = qaic_get_sg_table,
 	.print_info = qaic_gem_print_info,
 	.mmap = qaic_gem_object_mmap,
 	.vm_ops = &drm_vm_ops,
@@ -731,7 +760,6 @@ int qaic_create_bo_ioctl(struct drm_device *dev, void *data, struct drm_file *fi
 	if (ret)
 		goto free_bo;
 
-	bo->handle = args->handle;
 	drm_gem_object_put(obj);
 	srcu_read_unlock(&qdev->dev_lock, qdev_rcu_id);
 	srcu_read_unlock(&usr->qddev_lock, usr_rcu_id);
@@ -954,8 +982,9 @@ int qaic_attach_slice_bo_ioctl(struct drm_device *dev, void *data, struct drm_fi
 	if (args->hdr.count == 0)
 		return -EINVAL;
 
-	arg_size = args->hdr.count * sizeof(*slice_ent);
-	if (arg_size / args->hdr.count != sizeof(*slice_ent))
+	if (check_mul_overflow((unsigned long)args->hdr.count,
+			       (unsigned long)sizeof(*slice_ent),
+			       &arg_size))
 		return -EINVAL;
 
 	if (!(args->hdr.dir == DMA_TO_DEVICE || args->hdr.dir == DMA_FROM_DEVICE))
@@ -985,18 +1014,12 @@ int qaic_attach_slice_bo_ioctl(struct drm_device *dev, void *data, struct drm_fi
 
 	user_data = u64_to_user_ptr(args->data);
 
-	slice_ent = kzalloc(arg_size, GFP_KERNEL);
-	if (!slice_ent) {
-		ret = -EINVAL;
+	slice_ent = memdup_user(user_data, arg_size);
+	if (IS_ERR(slice_ent)) {
+		ret = PTR_ERR(slice_ent);
 		goto unlock_dev_srcu;
 	}
 
-	ret = copy_from_user(slice_ent, user_data, arg_size);
-	if (ret) {
-		ret = -EFAULT;
-		goto free_slice_ent;
-	}
-
 	obj = drm_gem_object_lookup(file_priv, args->hdr.handle);
 	if (!obj) {
 		ret = -ENOENT;
@@ -1024,6 +1047,11 @@ int qaic_attach_slice_bo_ioctl(struct drm_device *dev, void *data, struct drm_fi
 		goto unlock_ch_srcu;
 	}
 
+	if (dbc->id == qdev->ssr_dbc) {
+		ret = -EPIPE;
+		goto unlock_ch_srcu;
+	}
+
 	ret = qaic_prepare_bo(qdev, bo, &args->hdr);
 	if (ret)
 		goto unlock_ch_srcu;
@@ -1301,8 +1329,6 @@ static int __qaic_execute_bo_ioctl(struct drm_device *dev, void *data, struct dr
 	int usr_rcu_id, qdev_rcu_id;
 	struct qaic_device *qdev;
 	struct qaic_user *usr;
-	u8 __user *user_data;
-	unsigned long n;
 	u64 received_ts;
 	u32 queue_level;
 	u64 submit_ts;
@@ -1315,20 +1341,12 @@ static int __qaic_execute_bo_ioctl(struct drm_device *dev, void *data, struct dr
 	received_ts = ktime_get_ns();
 
 	size = is_partial ? sizeof(struct qaic_partial_execute_entry) : sizeof(*exec);
-	n = (unsigned long)size * args->hdr.count;
-	if (args->hdr.count == 0 || n / args->hdr.count != size)
+	if (args->hdr.count == 0)
 		return -EINVAL;
 
-	user_data = u64_to_user_ptr(args->data);
-
-	exec = kcalloc(args->hdr.count, size, GFP_KERNEL);
-	if (!exec)
-		return -ENOMEM;
-
-	if (copy_from_user(exec, user_data, n)) {
-		ret = -EFAULT;
-		goto free_exec;
-	}
+	exec = memdup_array_user(u64_to_user_ptr(args->data), args->hdr.count, size);
+	if (IS_ERR(exec))
+		return PTR_ERR(exec);
 
 	usr = file_priv->driver_priv;
 	usr_rcu_id = srcu_read_lock(&usr->qddev_lock);
@@ -1357,13 +1375,22 @@ static int __qaic_execute_bo_ioctl(struct drm_device *dev, void *data, struct dr
 		goto release_ch_rcu;
 	}
 
+	if (dbc->id == qdev->ssr_dbc) {
+		ret = -EPIPE;
+		goto release_ch_rcu;
+	}
+
+	ret = mutex_lock_interruptible(&dbc->req_lock);
+	if (ret)
+		goto release_ch_rcu;
+
 	head = readl(dbc->dbc_base + REQHP_OFF);
 	tail = readl(dbc->dbc_base + REQTP_OFF);
 
 	if (head == U32_MAX || tail == U32_MAX) {
 		/* PCI link error */
 		ret = -ENODEV;
-		goto release_ch_rcu;
+		goto unlock_req_lock;
 	}
 
 	queue_level = head <= tail ? tail - head : dbc->nelem - (head - tail);
@@ -1371,11 +1398,12 @@ static int __qaic_execute_bo_ioctl(struct drm_device *dev, void *data, struct dr
 	ret = send_bo_list_to_device(qdev, file_priv, exec, args->hdr.count, is_partial, dbc,
 				     head, &tail);
 	if (ret)
-		goto release_ch_rcu;
+		goto unlock_req_lock;
 
 	/* Finalize commit to hardware */
 	submit_ts = ktime_get_ns();
 	writel(tail, dbc->dbc_base + REQTP_OFF);
+	mutex_unlock(&dbc->req_lock);
 
 	update_profiling_data(file_priv, exec, args->hdr.count, is_partial, received_ts,
 			      submit_ts, queue_level);
@@ -1383,13 +1411,15 @@ static int __qaic_execute_bo_ioctl(struct drm_device *dev, void *data, struct dr
 	if (datapath_polling)
 		schedule_work(&dbc->poll_work);
 
+unlock_req_lock:
+	if (ret)
+		mutex_unlock(&dbc->req_lock);
 release_ch_rcu:
 	srcu_read_unlock(&dbc->ch_lock, rcu_id);
 unlock_dev_srcu:
 	srcu_read_unlock(&qdev->dev_lock, qdev_rcu_id);
 unlock_usr_srcu:
 	srcu_read_unlock(&usr->qddev_lock, usr_rcu_id);
-free_exec:
 	kfree(exec);
 	return ret;
 }
@@ -1484,7 +1514,7 @@ irqreturn_t dbc_irq_handler(int irq, void *data)
 	return IRQ_WAKE_THREAD;
 }
 
-void irq_polling_work(struct work_struct *work)
+void qaic_irq_polling_work(struct work_struct *work)
 {
 	struct dma_bridge_chan *dbc = container_of(work, struct dma_bridge_chan,  poll_work);
 	unsigned long flags;
@@ -1702,6 +1732,11 @@ int qaic_wait_bo_ioctl(struct drm_device *dev, void *data, struct drm_file *file
 		goto unlock_ch_srcu;
 	}
 
+	if (dbc->id == qdev->ssr_dbc) {
+		ret = -EPIPE;
+		goto unlock_ch_srcu;
+	}
+
 	obj = drm_gem_object_lookup(file_priv, args->handle);
 	if (!obj) {
 		ret = -ENOENT;
@@ -1722,6 +1757,9 @@ int qaic_wait_bo_ioctl(struct drm_device *dev, void *data, struct drm_file *file
 	if (!dbc->usr)
 		ret = -EPERM;
 
+	if (dbc->id == qdev->ssr_dbc)
+		ret = -EPIPE;
+
 put_obj:
 	drm_gem_object_put(obj);
 unlock_ch_srcu:
@@ -1742,7 +1780,8 @@ int qaic_perf_stats_bo_ioctl(struct drm_device *dev, void *data, struct drm_file
 	struct qaic_device *qdev;
 	struct qaic_user *usr;
 	struct qaic_bo *bo;
-	int ret, i;
+	int ret = 0;
+	int i;
 
 	usr = file_priv->driver_priv;
 	usr_rcu_id = srcu_read_lock(&usr->qddev_lock);
@@ -1763,18 +1802,12 @@ int qaic_perf_stats_bo_ioctl(struct drm_device *dev, void *data, struct drm_file
 		goto unlock_dev_srcu;
 	}
 
-	ent = kcalloc(args->hdr.count, sizeof(*ent), GFP_KERNEL);
-	if (!ent) {
-		ret = -EINVAL;
+	ent = memdup_array_user(u64_to_user_ptr(args->data), args->hdr.count, sizeof(*ent));
+	if (IS_ERR(ent)) {
+		ret = PTR_ERR(ent);
 		goto unlock_dev_srcu;
 	}
 
-	ret = copy_from_user(ent, u64_to_user_ptr(args->data), args->hdr.count * sizeof(*ent));
-	if (ret) {
-		ret = -EFAULT;
-		goto free_ent;
-	}
-
 	for (i = 0; i < args->hdr.count; i++) {
 		obj = drm_gem_object_lookup(file_priv, ent[i].handle);
 		if (!obj) {
@@ -1782,6 +1815,16 @@ int qaic_perf_stats_bo_ioctl(struct drm_device *dev, void *data, struct drm_file
 			goto free_ent;
 		}
 		bo = to_qaic_bo(obj);
+		if (!bo->sliced) {
+			drm_gem_object_put(obj);
+			ret = -EINVAL;
+			goto free_ent;
+		}
+		if (bo->dbc->id != args->hdr.dbc_id) {
+			drm_gem_object_put(obj);
+			ret = -EINVAL;
+			goto free_ent;
+		}
 		/*
 		 * perf stats ioctl is called before wait ioctl is complete then
 		 * the latency information is invalid.
@@ -1920,6 +1963,17 @@ static void empty_xfer_list(struct qaic_device *qdev, struct dma_bridge_chan *db
 	spin_unlock_irqrestore(&dbc->xfer_lock, flags);
 }
 
+static void sync_empty_xfer_list(struct qaic_device *qdev, struct dma_bridge_chan *dbc)
+{
+	empty_xfer_list(qdev, dbc);
+	synchronize_srcu(&dbc->ch_lock);
+	/*
+	 * Threads holding channel lock, may add more elements in the xfer_list.
+	 * Flush out these elements from xfer_list.
+	 */
+	empty_xfer_list(qdev, dbc);
+}
+
 int disable_dbc(struct qaic_device *qdev, u32 dbc_id, struct qaic_user *usr)
 {
 	if (!qdev->dbc[dbc_id].usr || qdev->dbc[dbc_id].usr->handle != usr->handle)
@@ -1934,7 +1988,7 @@ int disable_dbc(struct qaic_device *qdev, u32 dbc_id, struct qaic_user *usr)
  * enable_dbc - Enable the DBC. DBCs are disabled by removing the context of
  * user. Add user context back to DBC to enable it. This function trusts the
  * DBC ID passed and expects the DBC to be disabled.
- * @qdev: Qranium device handle
+ * @qdev: qaic device handle
  * @dbc_id: ID of the DBC
  * @usr: User context
  */
@@ -1948,13 +2002,7 @@ void wakeup_dbc(struct qaic_device *qdev, u32 dbc_id)
 	struct dma_bridge_chan *dbc = &qdev->dbc[dbc_id];
 
 	dbc->usr = NULL;
-	empty_xfer_list(qdev, dbc);
-	synchronize_srcu(&dbc->ch_lock);
-	/*
-	 * Threads holding channel lock, may add more elements in the xfer_list.
-	 * Flush out these elements from xfer_list.
-	 */
-	empty_xfer_list(qdev, dbc);
+	sync_empty_xfer_list(qdev, dbc);
 }
 
 void release_dbc(struct qaic_device *qdev, u32 dbc_id)
@@ -1995,3 +2043,30 @@ void qaic_data_get_fifo_info(struct dma_bridge_chan *dbc, u32 *head, u32 *tail)
 	*head = readl(dbc->dbc_base + REQHP_OFF);
 	*tail = readl(dbc->dbc_base + REQTP_OFF);
 }
+
+/*
+ * qaic_dbc_enter_ssr - Prepare to enter in sub system reset(SSR) for given DBC ID.
+ * @qdev: qaic device handle
+ * @dbc_id: ID of the DBC which will enter SSR
+ *
+ * The device will automatically deactivate the workload as not
+ * all errors can be silently recovered. The user will be
+ * notified and will need to decide the required recovery
+ * action to take.
+ */
+void qaic_dbc_enter_ssr(struct qaic_device *qdev, u32 dbc_id)
+{
+	qdev->ssr_dbc = dbc_id;
+	release_dbc(qdev, dbc_id);
+}
+
+/*
+ * qaic_dbc_exit_ssr - Prepare to exit from sub system reset(SSR) for given DBC ID.
+ * @qdev: qaic device handle
+ *
+ * The DBC returns to an operational state and begins accepting work after exiting SSR.
+ */
+void qaic_dbc_exit_ssr(struct qaic_device *qdev)
+{
+	qdev->ssr_dbc = QAIC_SSR_DBC_SENTINEL;
+}
diff --git a/drivers/accel/qaic/qaic_debugfs.c b/drivers/accel/qaic/qaic_debugfs.c
index a991b8198dc4..8dc4fe5bb560 100644
--- a/drivers/accel/qaic/qaic_debugfs.c
+++ b/drivers/accel/qaic/qaic_debugfs.c
@@ -218,6 +218,9 @@ static int qaic_bootlog_mhi_probe(struct mhi_device *mhi_dev, const struct mhi_d
 	if (ret)
 		goto destroy_workqueue;
 
+	dev_set_drvdata(&mhi_dev->dev, qdev);
+	qdev->bootlog_ch = mhi_dev;
+
 	for (i = 0; i < BOOTLOG_POOL_SIZE; i++) {
 		msg = devm_kzalloc(&qdev->pdev->dev, sizeof(*msg), GFP_KERNEL);
 		if (!msg) {
@@ -233,8 +236,6 @@ static int qaic_bootlog_mhi_probe(struct mhi_device *mhi_dev, const struct mhi_d
 			goto mhi_unprepare;
 	}
 
-	dev_set_drvdata(&mhi_dev->dev, qdev);
-	qdev->bootlog_ch = mhi_dev;
 	return 0;
 
 mhi_unprepare:
diff --git a/drivers/accel/qaic/qaic_drv.c b/drivers/accel/qaic/qaic_drv.c
index 3b415e2c9431..4c70bd949d53 100644
--- a/drivers/accel/qaic/qaic_drv.c
+++ b/drivers/accel/qaic/qaic_drv.c
@@ -29,6 +29,8 @@
 #include "mhi_controller.h"
 #include "qaic.h"
 #include "qaic_debugfs.h"
+#include "qaic_ras.h"
+#include "qaic_ssr.h"
 #include "qaic_timesync.h"
 #include "sahara.h"
 
@@ -269,6 +271,13 @@ static int qaic_create_drm_device(struct qaic_device *qdev, s32 partition_id)
 		return ret;
 	}
 
+	ret = qaic_sysfs_init(qddev);
+	if (ret) {
+		drm_dev_unregister(drm);
+		pci_dbg(qdev->pdev, "qaic_sysfs_init failed %d\n", ret);
+		return ret;
+	}
+
 	qaic_debugfs_init(qddev);
 
 	return ret;
@@ -280,6 +289,7 @@ static void qaic_destroy_drm_device(struct qaic_device *qdev, s32 partition_id)
 	struct drm_device *drm = to_drm(qddev);
 	struct qaic_user *usr;
 
+	qaic_sysfs_remove(qddev);
 	drm_dev_unregister(drm);
 	qddev->partition_id = 0;
 	/*
@@ -381,6 +391,7 @@ void qaic_dev_reset_clean_local_state(struct qaic_device *qdev)
 	qaic_notify_reset(qdev);
 
 	/* start tearing things down */
+	qaic_clean_up_ssr(qdev);
 	for (i = 0; i < qdev->num_dbc; ++i)
 		release_dbc(qdev, i);
 }
@@ -430,11 +441,18 @@ static struct qaic_device *create_qdev(struct pci_dev *pdev,
 	qdev->qts_wq = qaicm_wq_init(drm, "qaic_ts");
 	if (IS_ERR(qdev->qts_wq))
 		return NULL;
+	qdev->ssr_wq = qaicm_wq_init(drm, "qaic_ssr");
+	if (IS_ERR(qdev->ssr_wq))
+		return NULL;
 
 	ret = qaicm_srcu_init(drm, &qdev->dev_lock);
 	if (ret)
 		return NULL;
 
+	ret = qaic_ssr_init(qdev, drm);
+	if (ret)
+		pci_info(pdev, "QAIC SSR crashdump collection not supported.\n");
+
 	qdev->qddev = qddev;
 	qdev->pdev = pdev;
 	qddev->qdev = qdev;
@@ -453,6 +471,9 @@ static struct qaic_device *create_qdev(struct pci_dev *pdev,
 			return NULL;
 		init_waitqueue_head(&qdev->dbc[i].dbc_release);
 		INIT_LIST_HEAD(&qdev->dbc[i].bo_lists);
+		ret = drmm_mutex_init(drm, &qdev->dbc[i].req_lock);
+		if (ret)
+			return NULL;
 	}
 
 	return qdev;
@@ -541,7 +562,7 @@ static int init_msi(struct qaic_device *qdev, struct pci_dev *pdev)
 			qdev->dbc[i].irq = pci_irq_vector(pdev, qdev->single_msi ? 0 : i + 1);
 			if (!qdev->single_msi)
 				disable_irq_nosync(qdev->dbc[i].irq);
-			INIT_WORK(&qdev->dbc[i].poll_work, irq_polling_work);
+			INIT_WORK(&qdev->dbc[i].poll_work, qaic_irq_polling_work);
 		}
 	}
 
@@ -656,6 +677,92 @@ static const struct pci_error_handlers qaic_pci_err_handler = {
 	.reset_done = qaic_pci_reset_done,
 };
 
+static bool qaic_is_under_reset(struct qaic_device *qdev)
+{
+	int rcu_id;
+	bool ret;
+
+	rcu_id = srcu_read_lock(&qdev->dev_lock);
+	ret = qdev->dev_state != QAIC_ONLINE;
+	srcu_read_unlock(&qdev->dev_lock, rcu_id);
+	return ret;
+}
+
+static bool qaic_data_path_busy(struct qaic_device *qdev)
+{
+	bool ret = false;
+	int dev_rcu_id;
+	int i;
+
+	dev_rcu_id = srcu_read_lock(&qdev->dev_lock);
+	if (qdev->dev_state != QAIC_ONLINE) {
+		srcu_read_unlock(&qdev->dev_lock, dev_rcu_id);
+		return false;
+	}
+	for (i = 0; i < qdev->num_dbc; i++) {
+		struct dma_bridge_chan *dbc = &qdev->dbc[i];
+		unsigned long flags;
+		int ch_rcu_id;
+
+		ch_rcu_id = srcu_read_lock(&dbc->ch_lock);
+		if (!dbc->usr || !dbc->in_use) {
+			srcu_read_unlock(&dbc->ch_lock, ch_rcu_id);
+			continue;
+		}
+		spin_lock_irqsave(&dbc->xfer_lock, flags);
+		ret = !list_empty(&dbc->xfer_list);
+		spin_unlock_irqrestore(&dbc->xfer_lock, flags);
+		srcu_read_unlock(&dbc->ch_lock, ch_rcu_id);
+		if (ret)
+			break;
+	}
+	srcu_read_unlock(&qdev->dev_lock, dev_rcu_id);
+	return ret;
+}
+
+static int qaic_pm_suspend(struct device *dev)
+{
+	struct qaic_device *qdev = pci_get_drvdata(to_pci_dev(dev));
+
+	dev_dbg(dev, "Suspending..\n");
+	if (qaic_data_path_busy(qdev)) {
+		dev_dbg(dev, "Device's datapath is busy. Aborting suspend..\n");
+		return -EBUSY;
+	}
+	if (qaic_is_under_reset(qdev)) {
+		dev_dbg(dev, "Device is under reset. Aborting suspend..\n");
+		return -EBUSY;
+	}
+	qaic_mqts_ch_stop_timer(qdev->mqts_ch);
+	qaic_pci_reset_prepare(qdev->pdev);
+	pci_save_state(qdev->pdev);
+	pci_disable_device(qdev->pdev);
+	pci_set_power_state(qdev->pdev, PCI_D3hot);
+	return 0;
+}
+
+static int qaic_pm_resume(struct device *dev)
+{
+	struct qaic_device *qdev = pci_get_drvdata(to_pci_dev(dev));
+	int ret;
+
+	dev_dbg(dev, "Resuming..\n");
+	pci_set_power_state(qdev->pdev, PCI_D0);
+	pci_restore_state(qdev->pdev);
+	ret = pci_enable_device(qdev->pdev);
+	if (ret) {
+		dev_err(dev, "pci_enable_device failed on resume %d\n", ret);
+		return ret;
+	}
+	pci_set_master(qdev->pdev);
+	qaic_pci_reset_done(qdev->pdev);
+	return 0;
+}
+
+static const struct dev_pm_ops qaic_pm_ops = {
+	SYSTEM_SLEEP_PM_OPS(qaic_pm_suspend, qaic_pm_resume)
+};
+
 static struct pci_driver qaic_pci_driver = {
 	.name = QAIC_NAME,
 	.id_table = qaic_ids,
@@ -663,6 +770,9 @@ static struct pci_driver qaic_pci_driver = {
 	.remove = qaic_pci_remove,
 	.shutdown = qaic_pci_shutdown,
 	.err_handler = &qaic_pci_err_handler,
+	.driver = {
+		.pm = pm_sleep_ptr(&qaic_pm_ops),
+	},
 };
 
 static int __init qaic_init(void)
@@ -695,8 +805,19 @@ static int __init qaic_init(void)
 	if (ret)
 		pr_debug("qaic: qaic_bootlog_register failed %d\n", ret);
 
+	ret = qaic_ras_register();
+	if (ret)
+		pr_debug("qaic: qaic_ras_register failed %d\n", ret);
+	ret = qaic_ssr_register();
+	if (ret) {
+		pr_debug("qaic: qaic_ssr_register failed %d\n", ret);
+		goto free_bootlog;
+	}
+
 	return 0;
 
+free_bootlog:
+	qaic_bootlog_unregister();
 free_mhi:
 	mhi_driver_unregister(&qaic_mhi_driver);
 free_pci:
@@ -722,6 +843,8 @@ static void __exit qaic_exit(void)
 	 * reinitializing the link_up state after the cleanup is done.
 	 */
 	link_up = true;
+	qaic_ssr_unregister();
+	qaic_ras_unregister();
 	qaic_bootlog_unregister();
 	qaic_timesync_deinit();
 	sahara_unregister();
diff --git a/drivers/accel/qaic/qaic_ras.c b/drivers/accel/qaic/qaic_ras.c
new file mode 100644
index 000000000000..f1d52a710136
--- /dev/null
+++ b/drivers/accel/qaic/qaic_ras.c
@@ -0,0 +1,642 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+/* Copyright (c) 2020-2021, The Linux Foundation. All rights reserved. */
+/* Copyright (c) 2022-2024 Qualcomm Innovation Center, Inc. All rights reserved. */
+/* Copyright (c) Qualcomm Technologies, Inc. and/or its subsidiaries. */
+
+#include <asm/byteorder.h>
+#include <linux/device.h>
+#include <linux/kernel.h>
+#include <linux/mhi.h>
+
+#include "qaic.h"
+#include "qaic_ras.h"
+
+#define MAGIC		0x55AA
+#define VERSION		0x2
+#define HDR_SZ		12
+#define NUM_TEMP_LVL	3
+#define POWER_BREAK	BIT(0)
+
+enum msg_type {
+	MSG_PUSH, /* async push from device */
+	MSG_REQ,  /* sync request to device */
+	MSG_RESP, /* sync response from device */
+};
+
+enum err_type {
+	CE,	/* correctable error */
+	UE,	/* uncorrectable error */
+	UE_NF,	/* uncorrectable error that is non-fatal, expect a disruption */
+	ERR_TYPE_MAX,
+};
+
+static const char * const err_type_str[] = {
+	[CE]    = "Correctable",
+	[UE]    = "Uncorrectable",
+	[UE_NF] = "Uncorrectable Non-Fatal",
+};
+
+static const char * const err_class_str[] = {
+	[CE]    = "Warning",
+	[UE]    = "Fatal",
+	[UE_NF] = "Warning",
+};
+
+enum err_source {
+	SOC_MEM,
+	PCIE,
+	DDR,
+	SYS_BUS1,
+	SYS_BUS2,
+	NSP_MEM,
+	TSENS,
+};
+
+static const char * const err_src_str[TSENS + 1] = {
+	[SOC_MEM]	= "SoC Memory",
+	[PCIE]		= "PCIE",
+	[DDR]		= "DDR",
+	[SYS_BUS1]	= "System Bus source 1",
+	[SYS_BUS2]	= "System Bus source 2",
+	[NSP_MEM]	= "NSP Memory",
+	[TSENS]		= "Temperature Sensors",
+};
+
+struct ras_data {
+	/* header start */
+	/* Magic number to validate the message */
+	u16 magic;
+	/* RAS version number */
+	u16 ver;
+	u32 seq_num;
+	/* RAS message type */
+	u8  type;
+	u8  id;
+	/* Size of RAS message without the header in byte */
+	u16 len;
+	/* header end */
+	s32 result;
+	/*
+	 * Error source
+	 * 0 : SoC Memory
+	 * 1 : PCIE
+	 * 2 : DDR
+	 * 3 : System Bus source 1
+	 * 4 : System Bus source 2
+	 * 5 : NSP Memory
+	 * 6 : Temperature Sensors
+	 */
+	u32 source;
+	/*
+	 * Stores the error type, there are three types of error in RAS
+	 * 0 : correctable error (CE)
+	 * 1 : uncorrectable error (UE)
+	 * 2 : uncorrectable error that is non-fatal (UE_NF)
+	 */
+	u32 err_type;
+	u32 err_threshold;
+	u32 ce_count;
+	u32 ue_count;
+	u32 intr_num;
+	/* Data specific to error source */
+	u8  syndrome[64];
+} __packed;
+
+struct soc_mem_syndrome {
+	u64 error_address[8];
+} __packed;
+
+struct nsp_mem_syndrome {
+	u32 error_address[8];
+	u8 nsp_id;
+} __packed;
+
+struct ddr_syndrome {
+	u32 count;
+	u32 irq_status;
+	u32 data_31_0[2];
+	u32 data_63_32[2];
+	u32 data_95_64[2];
+	u32 data_127_96[2];
+	u32 addr_lsb;
+	u16 addr_msb;
+	u16 parity_bits;
+	u16 instance;
+	u16 err_type;
+} __packed;
+
+struct tsens_syndrome {
+	u32 threshold_type;
+	s32 temp;
+} __packed;
+
+struct sysbus1_syndrome {
+	u32 slave;
+	u32 err_type;
+	u16 addr[8];
+	u8  instance;
+} __packed;
+
+struct sysbus2_syndrome {
+	u32 lsb3;
+	u32 msb3;
+	u32 lsb2;
+	u32 msb2;
+	u32 ext_id;
+	u16 path;
+	u16 op_type;
+	u16 len;
+	u16 redirect;
+	u8  valid;
+	u8  word_error;
+	u8  non_secure;
+	u8  opc;
+	u8  error_code;
+	u8  trans_type;
+	u8  addr_space;
+	u8  instance;
+} __packed;
+
+struct pcie_syndrome {
+	/* CE info */
+	u32 bad_tlp;
+	u32 bad_dllp;
+	u32 replay_rollover;
+	u32 replay_timeout;
+	u32 rx_err;
+	u32 internal_ce_count;
+	/* UE_NF info */
+	u32 fc_timeout;
+	u32 poison_tlp;
+	u32 ecrc_err;
+	u32 unsupported_req;
+	u32 completer_abort;
+	u32 completion_timeout;
+	/* UE info */
+	u32 addr;
+	u8  index;
+	/*
+	 * Flag to indicate specific event of PCIe
+	 * BIT(0): Power break (low power)
+	 * BIT(1) to BIT(7): Reserved
+	 */
+	u8 flag;
+} __packed;
+
+static const char * const threshold_type_str[NUM_TEMP_LVL] = {
+	[0] = "lower",
+	[1] = "upper",
+	[2] = "critical",
+};
+
+static void ras_msg_to_cpu(struct ras_data *msg)
+{
+	struct sysbus1_syndrome *sysbus1_syndrome = (struct sysbus1_syndrome *)&msg->syndrome[0];
+	struct sysbus2_syndrome *sysbus2_syndrome = (struct sysbus2_syndrome *)&msg->syndrome[0];
+	struct soc_mem_syndrome *soc_syndrome = (struct soc_mem_syndrome *)&msg->syndrome[0];
+	struct nsp_mem_syndrome *nsp_syndrome = (struct nsp_mem_syndrome *)&msg->syndrome[0];
+	struct tsens_syndrome *tsens_syndrome = (struct tsens_syndrome *)&msg->syndrome[0];
+	struct pcie_syndrome *pcie_syndrome = (struct pcie_syndrome *)&msg->syndrome[0];
+	struct ddr_syndrome *ddr_syndrome = (struct ddr_syndrome *)&msg->syndrome[0];
+	int i;
+
+	le16_to_cpus(&msg->magic);
+	le16_to_cpus(&msg->ver);
+	le32_to_cpus(&msg->seq_num);
+	le16_to_cpus(&msg->len);
+	le32_to_cpus(&msg->result);
+	le32_to_cpus(&msg->source);
+	le32_to_cpus(&msg->err_type);
+	le32_to_cpus(&msg->err_threshold);
+	le32_to_cpus(&msg->ce_count);
+	le32_to_cpus(&msg->ue_count);
+	le32_to_cpus(&msg->intr_num);
+
+	switch (msg->source) {
+	case SOC_MEM:
+		for (i = 0; i < 8; i++)
+			le64_to_cpus(&soc_syndrome->error_address[i]);
+		break;
+	case PCIE:
+		le32_to_cpus(&pcie_syndrome->bad_tlp);
+		le32_to_cpus(&pcie_syndrome->bad_dllp);
+		le32_to_cpus(&pcie_syndrome->replay_rollover);
+		le32_to_cpus(&pcie_syndrome->replay_timeout);
+		le32_to_cpus(&pcie_syndrome->rx_err);
+		le32_to_cpus(&pcie_syndrome->internal_ce_count);
+		le32_to_cpus(&pcie_syndrome->fc_timeout);
+		le32_to_cpus(&pcie_syndrome->poison_tlp);
+		le32_to_cpus(&pcie_syndrome->ecrc_err);
+		le32_to_cpus(&pcie_syndrome->unsupported_req);
+		le32_to_cpus(&pcie_syndrome->completer_abort);
+		le32_to_cpus(&pcie_syndrome->completion_timeout);
+		le32_to_cpus(&pcie_syndrome->addr);
+		break;
+	case DDR:
+		le16_to_cpus(&ddr_syndrome->instance);
+		le16_to_cpus(&ddr_syndrome->err_type);
+		le32_to_cpus(&ddr_syndrome->count);
+		le32_to_cpus(&ddr_syndrome->irq_status);
+		le32_to_cpus(&ddr_syndrome->data_31_0[0]);
+		le32_to_cpus(&ddr_syndrome->data_31_0[1]);
+		le32_to_cpus(&ddr_syndrome->data_63_32[0]);
+		le32_to_cpus(&ddr_syndrome->data_63_32[1]);
+		le32_to_cpus(&ddr_syndrome->data_95_64[0]);
+		le32_to_cpus(&ddr_syndrome->data_95_64[1]);
+		le32_to_cpus(&ddr_syndrome->data_127_96[0]);
+		le32_to_cpus(&ddr_syndrome->data_127_96[1]);
+		le16_to_cpus(&ddr_syndrome->parity_bits);
+		le16_to_cpus(&ddr_syndrome->addr_msb);
+		le32_to_cpus(&ddr_syndrome->addr_lsb);
+		break;
+	case SYS_BUS1:
+		le32_to_cpus(&sysbus1_syndrome->slave);
+		le32_to_cpus(&sysbus1_syndrome->err_type);
+		for (i = 0; i < 8; i++)
+			le16_to_cpus(&sysbus1_syndrome->addr[i]);
+		break;
+	case SYS_BUS2:
+		le16_to_cpus(&sysbus2_syndrome->op_type);
+		le16_to_cpus(&sysbus2_syndrome->len);
+		le16_to_cpus(&sysbus2_syndrome->redirect);
+		le16_to_cpus(&sysbus2_syndrome->path);
+		le32_to_cpus(&sysbus2_syndrome->ext_id);
+		le32_to_cpus(&sysbus2_syndrome->lsb2);
+		le32_to_cpus(&sysbus2_syndrome->msb2);
+		le32_to_cpus(&sysbus2_syndrome->lsb3);
+		le32_to_cpus(&sysbus2_syndrome->msb3);
+		break;
+	case NSP_MEM:
+		for (i = 0; i < 8; i++)
+			le32_to_cpus(&nsp_syndrome->error_address[i]);
+		break;
+	case TSENS:
+		le32_to_cpus(&tsens_syndrome->threshold_type);
+		le32_to_cpus(&tsens_syndrome->temp);
+		break;
+	}
+}
+
+static void decode_ras_msg(struct qaic_device *qdev, struct ras_data *msg)
+{
+	struct sysbus1_syndrome *sysbus1_syndrome = (struct sysbus1_syndrome *)&msg->syndrome[0];
+	struct sysbus2_syndrome *sysbus2_syndrome = (struct sysbus2_syndrome *)&msg->syndrome[0];
+	struct soc_mem_syndrome *soc_syndrome = (struct soc_mem_syndrome *)&msg->syndrome[0];
+	struct nsp_mem_syndrome *nsp_syndrome = (struct nsp_mem_syndrome *)&msg->syndrome[0];
+	struct tsens_syndrome *tsens_syndrome = (struct tsens_syndrome *)&msg->syndrome[0];
+	struct pcie_syndrome *pcie_syndrome = (struct pcie_syndrome *)&msg->syndrome[0];
+	struct ddr_syndrome *ddr_syndrome = (struct ddr_syndrome *)&msg->syndrome[0];
+	char *class;
+	char *level;
+
+	if (msg->magic != MAGIC) {
+		pci_warn(qdev->pdev, "Dropping RAS message with invalid magic %x\n", msg->magic);
+		return;
+	}
+
+	if (!msg->ver || msg->ver > VERSION) {
+		pci_warn(qdev->pdev, "Dropping RAS message with invalid version %d\n", msg->ver);
+		return;
+	}
+
+	if (msg->type != MSG_PUSH) {
+		pci_warn(qdev->pdev, "Dropping non-PUSH RAS message\n");
+		return;
+	}
+
+	if (msg->len != sizeof(*msg) - HDR_SZ) {
+		pci_warn(qdev->pdev, "Dropping RAS message with invalid len %d\n", msg->len);
+		return;
+	}
+
+	if (msg->err_type >= ERR_TYPE_MAX) {
+		pci_warn(qdev->pdev, "Dropping RAS message with err type %d\n", msg->err_type);
+		return;
+	}
+
+	if (msg->err_type == UE)
+		level = KERN_ERR;
+	else
+		level = KERN_WARNING;
+
+	switch (msg->source) {
+	case SOC_MEM:
+		dev_printk(level, &qdev->pdev->dev, "RAS event.\nClass:%s\nDescription:%s %s %s\nError Threshold for this report %d\nSyndrome:\n    0x%llx\n    0x%llx\n    0x%llx\n    0x%llx\n    0x%llx\n    0x%llx\n    0x%llx\n    0x%llx\n",
+			   err_class_str[msg->err_type],
+			   err_type_str[msg->err_type],
+			   "error from",
+			   err_src_str[msg->source],
+			   msg->err_threshold,
+			   soc_syndrome->error_address[0],
+			   soc_syndrome->error_address[1],
+			   soc_syndrome->error_address[2],
+			   soc_syndrome->error_address[3],
+			   soc_syndrome->error_address[4],
+			   soc_syndrome->error_address[5],
+			   soc_syndrome->error_address[6],
+			   soc_syndrome->error_address[7]);
+		break;
+	case PCIE:
+		dev_printk(level, &qdev->pdev->dev, "RAS event.\nClass:%s\nDescription:%s %s %s\nError Threshold for this report %d\n",
+			   err_class_str[msg->err_type],
+			   err_type_str[msg->err_type],
+			   "error from",
+			   err_src_str[msg->source],
+			   msg->err_threshold);
+
+		switch (msg->err_type) {
+		case CE:
+			/*
+			 * Modeled after AER prints. This continues the dev_printk() from a few
+			 * lines up. We reduce duplication of code, but also avoid re-printing the
+			 * PCI device info so that the end result looks uniform to the log user.
+			 */
+			printk(KERN_WARNING pr_fmt("Syndrome:\n    Bad TLP count %d\n    Bad DLLP count %d\n    Replay Rollover count %d\n    Replay Timeout count %d\n    Recv Error count %d\n    Internal CE count %d\n"),
+			       pcie_syndrome->bad_tlp,
+			       pcie_syndrome->bad_dllp,
+			       pcie_syndrome->replay_rollover,
+			       pcie_syndrome->replay_timeout,
+			       pcie_syndrome->rx_err,
+			       pcie_syndrome->internal_ce_count);
+			if (msg->ver > 0x1)
+				pr_warn("    Power break %s\n",
+					pcie_syndrome->flag & POWER_BREAK ? "ON" : "OFF");
+			break;
+		case UE:
+			printk(KERN_ERR pr_fmt("Syndrome:\n    Index %d\n    Address 0x%x\n"),
+			       pcie_syndrome->index, pcie_syndrome->addr);
+			break;
+		case UE_NF:
+			printk(KERN_WARNING pr_fmt("Syndrome:\n    FC timeout count %d\n    Poisoned TLP count %d\n    ECRC error count %d\n    Unsupported request count %d\n    Completer abort count %d\n    Completion timeout count %d\n"),
+			       pcie_syndrome->fc_timeout,
+			       pcie_syndrome->poison_tlp,
+			       pcie_syndrome->ecrc_err,
+			       pcie_syndrome->unsupported_req,
+			       pcie_syndrome->completer_abort,
+			       pcie_syndrome->completion_timeout);
+			break;
+		default:
+			break;
+		}
+		break;
+	case DDR:
+		dev_printk(level, &qdev->pdev->dev, "RAS event.\nClass:%s\nDescription:%s %s %s\nError Threshold for this report %d\nSyndrome:\n    Instance %d\n    Count %d\n    Data 31_0 0x%x 0x%x\n    Data 63_32 0x%x 0x%x\n    Data 95_64 0x%x 0x%x\n    Data 127_96 0x%x 0x%x\n    Parity bits 0x%x\n    Address msb 0x%x\n    Address lsb 0x%x\n",
+			   err_class_str[msg->err_type],
+			   err_type_str[msg->err_type],
+			   "error from",
+			   err_src_str[msg->source],
+			   msg->err_threshold,
+			   ddr_syndrome->instance,
+			   ddr_syndrome->count,
+			   ddr_syndrome->data_31_0[1],
+			   ddr_syndrome->data_31_0[0],
+			   ddr_syndrome->data_63_32[1],
+			   ddr_syndrome->data_63_32[0],
+			   ddr_syndrome->data_95_64[1],
+			   ddr_syndrome->data_95_64[0],
+			   ddr_syndrome->data_127_96[1],
+			   ddr_syndrome->data_127_96[0],
+			   ddr_syndrome->parity_bits,
+			   ddr_syndrome->addr_msb,
+			   ddr_syndrome->addr_lsb);
+		break;
+	case SYS_BUS1:
+		dev_printk(level, &qdev->pdev->dev, "RAS event.\nClass:%s\nDescription:%s %s %s\nError Threshold for this report %d\nSyndrome:\n    instance %d\n    %s\n    err_type %d\n    address0 0x%x\n    address1 0x%x\n    address2 0x%x\n    address3 0x%x\n    address4 0x%x\n    address5 0x%x\n    address6 0x%x\n    address7 0x%x\n",
+			   err_class_str[msg->err_type],
+			   err_type_str[msg->err_type],
+			   "error from",
+			   err_src_str[msg->source],
+			   msg->err_threshold,
+			   sysbus1_syndrome->instance,
+			   sysbus1_syndrome->slave ? "Slave" : "Master",
+			   sysbus1_syndrome->err_type,
+			   sysbus1_syndrome->addr[0],
+			   sysbus1_syndrome->addr[1],
+			   sysbus1_syndrome->addr[2],
+			   sysbus1_syndrome->addr[3],
+			   sysbus1_syndrome->addr[4],
+			   sysbus1_syndrome->addr[5],
+			   sysbus1_syndrome->addr[6],
+			   sysbus1_syndrome->addr[7]);
+		break;
+	case SYS_BUS2:
+		dev_printk(level, &qdev->pdev->dev, "RAS event.\nClass:%s\nDescription:%s %s %s\nError Threshold for this report %d\nSyndrome:\n    instance %d\n    valid %d\n    word error %d\n    non-secure %d\n    opc %d\n    error code %d\n    transaction type %d\n    address space %d\n    operation type %d\n    len %d\n    redirect %d\n    path %d\n    ext_id %d\n    lsb2 %d\n    msb2 %d\n    lsb3 %d\n    msb3 %d\n",
+			   err_class_str[msg->err_type],
+			   err_type_str[msg->err_type],
+			   "error from",
+			   err_src_str[msg->source],
+			   msg->err_threshold,
+			   sysbus2_syndrome->instance,
+			   sysbus2_syndrome->valid,
+			   sysbus2_syndrome->word_error,
+			   sysbus2_syndrome->non_secure,
+			   sysbus2_syndrome->opc,
+			   sysbus2_syndrome->error_code,
+			   sysbus2_syndrome->trans_type,
+			   sysbus2_syndrome->addr_space,
+			   sysbus2_syndrome->op_type,
+			   sysbus2_syndrome->len,
+			   sysbus2_syndrome->redirect,
+			   sysbus2_syndrome->path,
+			   sysbus2_syndrome->ext_id,
+			   sysbus2_syndrome->lsb2,
+			   sysbus2_syndrome->msb2,
+			   sysbus2_syndrome->lsb3,
+			   sysbus2_syndrome->msb3);
+		break;
+	case NSP_MEM:
+		dev_printk(level, &qdev->pdev->dev, "RAS event.\nClass:%s\nDescription:%s %s %s\nError Threshold for this report %d\nSyndrome:\n    NSP ID %d\n    0x%x\n    0x%x\n    0x%x\n    0x%x\n    0x%x\n    0x%x\n    0x%x\n    0x%x\n",
+			   err_class_str[msg->err_type],
+			   err_type_str[msg->err_type],
+			   "error from",
+			   err_src_str[msg->source],
+			   msg->err_threshold,
+			   nsp_syndrome->nsp_id,
+			   nsp_syndrome->error_address[0],
+			   nsp_syndrome->error_address[1],
+			   nsp_syndrome->error_address[2],
+			   nsp_syndrome->error_address[3],
+			   nsp_syndrome->error_address[4],
+			   nsp_syndrome->error_address[5],
+			   nsp_syndrome->error_address[6],
+			   nsp_syndrome->error_address[7]);
+		break;
+	case TSENS:
+		if (tsens_syndrome->threshold_type >= NUM_TEMP_LVL) {
+			pci_warn(qdev->pdev, "Dropping RAS message with invalid temp threshold %d\n",
+				 tsens_syndrome->threshold_type);
+			break;
+		}
+
+		if (msg->err_type)
+			class = "Fatal";
+		else if (tsens_syndrome->threshold_type)
+			class = "Critical";
+		else
+			class = "Warning";
+
+		dev_printk(level, &qdev->pdev->dev, "RAS event.\nClass:%s\nDescription:%s %s %s\nError Threshold for this report %d\nSyndrome:\n    %s threshold\n    %d deg C\n",
+			   class,
+			   err_type_str[msg->err_type],
+			   "error from",
+			   err_src_str[msg->source],
+			   msg->err_threshold,
+			   threshold_type_str[tsens_syndrome->threshold_type],
+			   tsens_syndrome->temp);
+		break;
+	}
+
+	/* Uncorrectable errors are fatal */
+	if (msg->err_type == UE)
+		mhi_soc_reset(qdev->mhi_cntrl);
+
+	switch (msg->err_type) {
+	case CE:
+		if (qdev->ce_count != UINT_MAX)
+			qdev->ce_count++;
+		break;
+	case UE:
+		if (qdev->ce_count != UINT_MAX)
+			qdev->ue_count++;
+		break;
+	case UE_NF:
+		if (qdev->ce_count != UINT_MAX)
+			qdev->ue_nf_count++;
+		break;
+	default:
+		/* not possible */
+		break;
+	}
+}
+
+static ssize_t ce_count_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	struct qaic_device *qdev = pci_get_drvdata(to_pci_dev(dev));
+
+	return sysfs_emit(buf, "%d\n", qdev->ce_count);
+}
+
+static ssize_t ue_count_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	struct qaic_device *qdev = pci_get_drvdata(to_pci_dev(dev));
+
+	return sysfs_emit(buf, "%d\n", qdev->ue_count);
+}
+
+static ssize_t ue_nonfatal_count_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	struct qaic_device *qdev = pci_get_drvdata(to_pci_dev(dev));
+
+	return sysfs_emit(buf, "%d\n", qdev->ue_nf_count);
+}
+
+static DEVICE_ATTR_RO(ce_count);
+static DEVICE_ATTR_RO(ue_count);
+static DEVICE_ATTR_RO(ue_nonfatal_count);
+
+static struct attribute *ras_attrs[] = {
+	&dev_attr_ce_count.attr,
+	&dev_attr_ue_count.attr,
+	&dev_attr_ue_nonfatal_count.attr,
+	NULL,
+};
+
+static struct attribute_group ras_group = {
+	.attrs = ras_attrs,
+};
+
+static int qaic_ras_mhi_probe(struct mhi_device *mhi_dev, const struct mhi_device_id *id)
+{
+	struct qaic_device *qdev = pci_get_drvdata(to_pci_dev(mhi_dev->mhi_cntrl->cntrl_dev));
+	struct ras_data *resp;
+	int ret;
+
+	ret = mhi_prepare_for_transfer(mhi_dev);
+	if (ret)
+		return ret;
+
+	resp = kzalloc(sizeof(*resp), GFP_KERNEL);
+	if (!resp) {
+		mhi_unprepare_from_transfer(mhi_dev);
+		return -ENOMEM;
+	}
+
+	ret = mhi_queue_buf(mhi_dev, DMA_FROM_DEVICE, resp, sizeof(*resp), MHI_EOT);
+	if (ret) {
+		kfree(resp);
+		mhi_unprepare_from_transfer(mhi_dev);
+		return ret;
+	}
+
+	ret = device_add_group(&qdev->pdev->dev, &ras_group);
+	if (ret) {
+		mhi_unprepare_from_transfer(mhi_dev);
+		pci_dbg(qdev->pdev, "ras add sysfs failed %d\n", ret);
+		return ret;
+	}
+
+	dev_set_drvdata(&mhi_dev->dev, qdev);
+	qdev->ras_ch = mhi_dev;
+
+	return ret;
+}
+
+static void qaic_ras_mhi_remove(struct mhi_device *mhi_dev)
+{
+	struct qaic_device *qdev;
+
+	qdev = dev_get_drvdata(&mhi_dev->dev);
+	qdev->ras_ch = NULL;
+	device_remove_group(&qdev->pdev->dev, &ras_group);
+	mhi_unprepare_from_transfer(mhi_dev);
+}
+
+static void qaic_ras_mhi_ul_xfer_cb(struct mhi_device *mhi_dev, struct mhi_result *mhi_result) {}
+
+static void qaic_ras_mhi_dl_xfer_cb(struct mhi_device *mhi_dev, struct mhi_result *mhi_result)
+{
+	struct qaic_device *qdev = dev_get_drvdata(&mhi_dev->dev);
+	struct ras_data *msg = mhi_result->buf_addr;
+	int ret;
+
+	if (mhi_result->transaction_status) {
+		kfree(msg);
+		return;
+	}
+
+	ras_msg_to_cpu(msg);
+	decode_ras_msg(qdev, msg);
+
+	ret = mhi_queue_buf(qdev->ras_ch, DMA_FROM_DEVICE, msg, sizeof(*msg), MHI_EOT);
+	if (ret) {
+		dev_err(&mhi_dev->dev, "Cannot requeue RAS recv buf %d\n", ret);
+		kfree(msg);
+	}
+}
+
+static const struct mhi_device_id qaic_ras_mhi_match_table[] = {
+	{ .chan = "QAIC_STATUS", },
+	{},
+};
+
+static struct mhi_driver qaic_ras_mhi_driver = {
+	.id_table = qaic_ras_mhi_match_table,
+	.remove = qaic_ras_mhi_remove,
+	.probe = qaic_ras_mhi_probe,
+	.ul_xfer_cb = qaic_ras_mhi_ul_xfer_cb,
+	.dl_xfer_cb = qaic_ras_mhi_dl_xfer_cb,
+	.driver = {
+		.name = "qaic_ras",
+	},
+};
+
+int qaic_ras_register(void)
+{
+	return mhi_driver_register(&qaic_ras_mhi_driver);
+}
+
+void qaic_ras_unregister(void)
+{
+	mhi_driver_unregister(&qaic_ras_mhi_driver);
+}
diff --git a/drivers/accel/qaic/qaic_ras.h b/drivers/accel/qaic/qaic_ras.h
new file mode 100644
index 000000000000..d44a4eeeb060
--- /dev/null
+++ b/drivers/accel/qaic/qaic_ras.h
@@ -0,0 +1,10 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright (c) 2020, The Linux Foundation. All rights reserved. */
+
+#ifndef __QAIC_RAS_H__
+#define __QAIC_RAS_H__
+
+int qaic_ras_register(void);
+void qaic_ras_unregister(void);
+
+#endif /* __QAIC_RAS_H__ */
diff --git a/drivers/accel/qaic/qaic_ssr.c b/drivers/accel/qaic/qaic_ssr.c
new file mode 100644
index 000000000000..9b662d690371
--- /dev/null
+++ b/drivers/accel/qaic/qaic_ssr.c
@@ -0,0 +1,815 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+/* Copyright (c) 2020-2021, The Linux Foundation. All rights reserved. */
+/* Copyright (c) 2021-2024 Qualcomm Innovation Center, Inc. All rights reserved. */
+
+#include <asm/byteorder.h>
+#include <drm/drm_file.h>
+#include <drm/drm_managed.h>
+#include <linux/devcoredump.h>
+#include <linux/device.h>
+#include <linux/kernel.h>
+#include <linux/mhi.h>
+#include <linux/workqueue.h>
+
+#include "qaic.h"
+#include "qaic_ssr.h"
+
+#define SSR_RESP_MSG_SZ 32
+#define SSR_MHI_BUF_SIZE SZ_64K
+#define SSR_MEM_READ_DATA_SIZE ((u64)SSR_MHI_BUF_SIZE - sizeof(struct ssr_crashdump))
+#define SSR_MEM_READ_CHUNK_SIZE ((u64)SSR_MEM_READ_DATA_SIZE - sizeof(struct ssr_memory_read_rsp))
+
+#define DEBUG_TRANSFER_INFO		BIT(0)
+#define DEBUG_TRANSFER_INFO_RSP		BIT(1)
+#define MEMORY_READ			BIT(2)
+#define MEMORY_READ_RSP			BIT(3)
+#define DEBUG_TRANSFER_DONE		BIT(4)
+#define DEBUG_TRANSFER_DONE_RSP		BIT(5)
+#define SSR_EVENT			BIT(8)
+#define SSR_EVENT_RSP			BIT(9)
+
+#define SSR_EVENT_NACK		BIT(0)
+#define BEFORE_SHUTDOWN		BIT(1)
+#define AFTER_SHUTDOWN		BIT(2)
+#define BEFORE_POWER_UP		BIT(3)
+#define AFTER_POWER_UP		BIT(4)
+
+struct debug_info_table {
+	/* Save preferences. Default is mandatory */
+	u64 save_perf;
+	/* Base address of the debug region */
+	u64 mem_base;
+	/* Size of debug region in bytes */
+	u64 len;
+	/* Description */
+	char desc[20];
+	/* Filename of debug region */
+	char filename[20];
+};
+
+struct _ssr_hdr {
+	__le32 cmd;
+	__le32 len;
+	__le32 dbc_id;
+};
+
+struct ssr_hdr {
+	u32 cmd;
+	u32 len;
+	u32 dbc_id;
+};
+
+struct ssr_debug_transfer_info {
+	struct ssr_hdr hdr;
+	u32 resv;
+	u64 tbl_addr;
+	u64 tbl_len;
+} __packed;
+
+struct ssr_debug_transfer_info_rsp {
+	struct _ssr_hdr hdr;
+	__le32 ret;
+} __packed;
+
+struct ssr_memory_read {
+	struct _ssr_hdr hdr;
+	__le32 resv;
+	__le64 addr;
+	__le64 len;
+} __packed;
+
+struct ssr_memory_read_rsp {
+	struct _ssr_hdr hdr;
+	__le32 resv;
+	u8 data[];
+} __packed;
+
+struct ssr_debug_transfer_done {
+	struct _ssr_hdr hdr;
+	__le32 resv;
+} __packed;
+
+struct ssr_debug_transfer_done_rsp {
+	struct _ssr_hdr hdr;
+	__le32 ret;
+} __packed;
+
+struct ssr_event {
+	struct ssr_hdr hdr;
+	u32 event;
+} __packed;
+
+struct ssr_event_rsp {
+	struct _ssr_hdr hdr;
+	__le32 event;
+} __packed;
+
+struct ssr_resp {
+	/* Work struct to schedule work coming on QAIC_SSR channel */
+	struct work_struct work;
+	/* Root struct of device, used to access device resources */
+	struct qaic_device *qdev;
+	/* Buffer used by MHI for transfer requests */
+	u8 data[] __aligned(8);
+};
+
+/* SSR crashdump book keeping structure */
+struct ssr_dump_info {
+	/* DBC associated with this SSR crashdump */
+	struct dma_bridge_chan *dbc;
+	/*
+	 * It will be used when we complete the crashdump download and switch
+	 * to waiting on SSR events
+	 */
+	struct ssr_resp *resp;
+	/* MEMORY READ request MHI buffer.*/
+	struct ssr_memory_read *read_buf_req;
+	/* TRUE: ->read_buf_req is queued for MHI transaction. FALSE: Otherwise */
+	bool read_buf_req_queued;
+	/* Address of table in host */
+	void *tbl_addr;
+	/* Total size of table */
+	u64 tbl_len;
+	/* Offset of table(->tbl_addr) where the new chunk will be dumped */
+	u64 tbl_off;
+	/* Address of table in device/target */
+	u64 tbl_addr_dev;
+	/* Ptr to the entire dump */
+	void *dump_addr;
+	/* Entire crashdump size */
+	u64 dump_sz;
+	/* Offset of crashdump(->dump_addr) where the new chunk will be dumped */
+	u64 dump_off;
+	/* Points to the table entry we are currently downloading */
+	struct debug_info_table *tbl_ent;
+	/* Offset in the current table entry(->tbl_ent) for next chuck */
+	u64 tbl_ent_off;
+};
+
+struct ssr_crashdump {
+	/*
+	 * Points to a book keeping struct maintained by MHI SSR device while
+	 * downloading a SSR crashdump. It is NULL when crashdump downloading
+	 * not in progress.
+	 */
+	struct ssr_dump_info *dump_info;
+	/* Work struct to schedule work coming on QAIC_SSR channel */
+	struct work_struct work;
+	/* Root struct of device, used to access device resources */
+	struct qaic_device *qdev;
+	/* Buffer used by MHI for transfer requests */
+	u8 data[];
+};
+
+#define QAIC_SSR_DUMP_V1_MAGIC 0x1234567890abcdef
+#define QAIC_SSR_DUMP_V1_VER   1
+struct dump_file_meta {
+	u64 magic;
+	u64 version;
+	u64 size;		/* Total size of the entire dump */
+	u64 tbl_len;		/* Length of the table in byte */
+};
+
+/*
+ * Layout of crashdump
+ *              +------------------------------------------+
+ *              |         Crashdump Meta structure         |
+ *              | type: struct dump_file_meta              |
+ *              +------------------------------------------+
+ *              |             Crashdump Table              |
+ *              | type: array of struct debug_info_table   |
+ *              |                                          |
+ *              |                                          |
+ *              |                                          |
+ *              +------------------------------------------+
+ *              |                Crashdump                 |
+ *              |                                          |
+ *              |                                          |
+ *              |                                          |
+ *              |                                          |
+ *              |                                          |
+ *              +------------------------------------------+
+ */
+
+static void free_ssr_dump_info(struct ssr_crashdump *ssr_crash)
+{
+	struct ssr_dump_info *dump_info = ssr_crash->dump_info;
+
+	ssr_crash->dump_info = NULL;
+	if (!dump_info)
+		return;
+	if (!dump_info->read_buf_req_queued)
+		kfree(dump_info->read_buf_req);
+	vfree(dump_info->tbl_addr);
+	vfree(dump_info->dump_addr);
+	kfree(dump_info);
+}
+
+void qaic_clean_up_ssr(struct qaic_device *qdev)
+{
+	struct ssr_crashdump *ssr_crash = qdev->ssr_mhi_buf;
+
+	if (!ssr_crash)
+		return;
+
+	qaic_dbc_exit_ssr(qdev);
+	free_ssr_dump_info(ssr_crash);
+}
+
+static int alloc_dump(struct ssr_dump_info *dump_info)
+{
+	struct debug_info_table *tbl_ent = dump_info->tbl_addr;
+	struct dump_file_meta *dump_meta;
+	u64 tbl_sz_lp = 0;
+	u64 dump_size = 0;
+
+	while (tbl_sz_lp < dump_info->tbl_len) {
+		le64_to_cpus(&tbl_ent->save_perf);
+		le64_to_cpus(&tbl_ent->mem_base);
+		le64_to_cpus(&tbl_ent->len);
+
+		if (tbl_ent->len == 0)
+			return -EINVAL;
+
+		dump_size += tbl_ent->len;
+		tbl_ent++;
+		tbl_sz_lp += sizeof(*tbl_ent);
+	}
+
+	dump_info->dump_sz = dump_size + dump_info->tbl_len + sizeof(*dump_meta);
+	dump_info->dump_addr = vzalloc(dump_info->dump_sz);
+	if (!dump_info->dump_addr)
+		return -ENOMEM;
+
+	/* Copy crashdump meta and table */
+	dump_meta = dump_info->dump_addr;
+	dump_meta->magic = QAIC_SSR_DUMP_V1_MAGIC;
+	dump_meta->version = QAIC_SSR_DUMP_V1_VER;
+	dump_meta->size = dump_info->dump_sz;
+	dump_meta->tbl_len = dump_info->tbl_len;
+	memcpy(dump_info->dump_addr + sizeof(*dump_meta), dump_info->tbl_addr, dump_info->tbl_len);
+	/* Offset by crashdump meta and table (copied above) */
+	dump_info->dump_off = dump_info->tbl_len + sizeof(*dump_meta);
+
+	return 0;
+}
+
+static int send_xfer_done(struct qaic_device *qdev, void *resp, u32 dbc_id)
+{
+	struct ssr_debug_transfer_done *xfer_done;
+	int ret;
+
+	xfer_done = kmalloc(sizeof(*xfer_done), GFP_KERNEL);
+	if (!xfer_done) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ret = mhi_queue_buf(qdev->ssr_ch, DMA_FROM_DEVICE, resp, SSR_RESP_MSG_SZ, MHI_EOT);
+	if (ret)
+		goto free_xfer_done;
+
+	xfer_done->hdr.cmd = cpu_to_le32(DEBUG_TRANSFER_DONE);
+	xfer_done->hdr.len = cpu_to_le32(sizeof(*xfer_done));
+	xfer_done->hdr.dbc_id = cpu_to_le32(dbc_id);
+
+	ret = mhi_queue_buf(qdev->ssr_ch, DMA_TO_DEVICE, xfer_done, sizeof(*xfer_done), MHI_EOT);
+	if (ret)
+		goto free_xfer_done;
+
+	return 0;
+
+free_xfer_done:
+	kfree(xfer_done);
+out:
+	return ret;
+}
+
+static int mem_read_req(struct qaic_device *qdev, u64 dest_addr, u64 dest_len)
+{
+	struct ssr_crashdump *ssr_crash = qdev->ssr_mhi_buf;
+	struct ssr_memory_read *read_buf_req;
+	struct ssr_dump_info *dump_info;
+	int ret;
+
+	dump_info = ssr_crash->dump_info;
+	ret = mhi_queue_buf(qdev->ssr_ch, DMA_FROM_DEVICE, ssr_crash->data, SSR_MEM_READ_DATA_SIZE,
+			    MHI_EOT);
+	if (ret)
+		goto out;
+
+	read_buf_req = dump_info->read_buf_req;
+	read_buf_req->hdr.cmd = cpu_to_le32(MEMORY_READ);
+	read_buf_req->hdr.len = cpu_to_le32(sizeof(*read_buf_req));
+	read_buf_req->hdr.dbc_id = cpu_to_le32(qdev->ssr_dbc);
+	read_buf_req->addr = cpu_to_le64(dest_addr);
+	read_buf_req->len = cpu_to_le64(dest_len);
+
+	ret = mhi_queue_buf(qdev->ssr_ch, DMA_TO_DEVICE, read_buf_req, sizeof(*read_buf_req),
+			    MHI_EOT);
+	if (!ret)
+		dump_info->read_buf_req_queued = true;
+
+out:
+	return ret;
+}
+
+static int ssr_copy_table(struct ssr_dump_info *dump_info, void *data, u64 len)
+{
+	if (len > dump_info->tbl_len - dump_info->tbl_off)
+		return -EINVAL;
+
+	memcpy(dump_info->tbl_addr + dump_info->tbl_off, data, len);
+	dump_info->tbl_off += len;
+
+	/* Entire table has been downloaded, alloc dump memory */
+	if (dump_info->tbl_off == dump_info->tbl_len) {
+		dump_info->tbl_ent = dump_info->tbl_addr;
+		return alloc_dump(dump_info);
+	}
+
+	return 0;
+}
+
+static int ssr_copy_dump(struct ssr_dump_info *dump_info, void *data, u64 len)
+{
+	struct debug_info_table *tbl_ent;
+
+	tbl_ent = dump_info->tbl_ent;
+
+	if (len > tbl_ent->len - dump_info->tbl_ent_off)
+		return -EINVAL;
+
+	memcpy(dump_info->dump_addr + dump_info->dump_off, data, len);
+	dump_info->dump_off += len;
+	dump_info->tbl_ent_off += len;
+
+	/*
+	 * Current segment (a entry in table) of the crashdump is complete,
+	 * move to next one
+	 */
+	if (tbl_ent->len == dump_info->tbl_ent_off) {
+		dump_info->tbl_ent++;
+		dump_info->tbl_ent_off = 0;
+	}
+
+	return 0;
+}
+
+static void ssr_dump_worker(struct work_struct *work)
+{
+	struct ssr_crashdump *ssr_crash = container_of(work, struct ssr_crashdump, work);
+	struct qaic_device *qdev = ssr_crash->qdev;
+	struct ssr_memory_read_rsp *mem_rd_resp;
+	struct debug_info_table *tbl_ent;
+	struct ssr_dump_info *dump_info;
+	u64 dest_addr, dest_len;
+	struct _ssr_hdr *_hdr;
+	struct ssr_hdr hdr;
+	u64 data_len;
+	int ret;
+
+	mem_rd_resp = (struct ssr_memory_read_rsp *)ssr_crash->data;
+	_hdr = &mem_rd_resp->hdr;
+	hdr.cmd = le32_to_cpu(_hdr->cmd);
+	hdr.len = le32_to_cpu(_hdr->len);
+	hdr.dbc_id = le32_to_cpu(_hdr->dbc_id);
+
+	if (hdr.dbc_id != qdev->ssr_dbc)
+		goto reset_device;
+
+	dump_info = ssr_crash->dump_info;
+	if (!dump_info)
+		goto reset_device;
+
+	if (hdr.cmd != MEMORY_READ_RSP)
+		goto free_dump_info;
+
+	if (hdr.len > SSR_MEM_READ_DATA_SIZE)
+		goto free_dump_info;
+
+	data_len = hdr.len - sizeof(*mem_rd_resp);
+
+	if (dump_info->tbl_off < dump_info->tbl_len) /* Chunk belongs to table */
+		ret = ssr_copy_table(dump_info, mem_rd_resp->data, data_len);
+	else /* Chunk belongs to crashdump */
+		ret = ssr_copy_dump(dump_info, mem_rd_resp->data, data_len);
+
+	if (ret)
+		goto free_dump_info;
+
+	if (dump_info->tbl_off < dump_info->tbl_len) {
+		/* Continue downloading table */
+		dest_addr = dump_info->tbl_addr_dev + dump_info->tbl_off;
+		dest_len = min(SSR_MEM_READ_CHUNK_SIZE, dump_info->tbl_len - dump_info->tbl_off);
+		ret = mem_read_req(qdev, dest_addr, dest_len);
+	} else if (dump_info->dump_off < dump_info->dump_sz) {
+		/* Continue downloading crashdump */
+		tbl_ent = dump_info->tbl_ent;
+		dest_addr = tbl_ent->mem_base + dump_info->tbl_ent_off;
+		dest_len = min(SSR_MEM_READ_CHUNK_SIZE, tbl_ent->len - dump_info->tbl_ent_off);
+		ret = mem_read_req(qdev, dest_addr, dest_len);
+	} else {
+		/* Crashdump download complete */
+		ret = send_xfer_done(qdev, dump_info->resp->data, hdr.dbc_id);
+	}
+
+	/* Most likely a MHI xfer has failed */
+	if (ret)
+		goto free_dump_info;
+
+	return;
+
+free_dump_info:
+	/* Free the allocated memory */
+	free_ssr_dump_info(ssr_crash);
+reset_device:
+	/*
+	 * After subsystem crashes in device crashdump collection begins but
+	 * something went wrong while collecting crashdump, now instead of
+	 * handling this error we just reset the device as the best effort has
+	 * been made
+	 */
+	mhi_soc_reset(qdev->mhi_cntrl);
+}
+
+static struct ssr_dump_info *alloc_dump_info(struct qaic_device *qdev,
+					     struct ssr_debug_transfer_info *debug_info)
+{
+	struct ssr_dump_info *dump_info;
+	int ret;
+
+	le64_to_cpus(&debug_info->tbl_len);
+	le64_to_cpus(&debug_info->tbl_addr);
+
+	if (debug_info->tbl_len == 0 ||
+	    debug_info->tbl_len % sizeof(struct debug_info_table) != 0) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	/* Allocate SSR crashdump book keeping structure */
+	dump_info = kzalloc(sizeof(*dump_info), GFP_KERNEL);
+	if (!dump_info) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	/* Buffer used to send MEMORY READ request to device via MHI */
+	dump_info->read_buf_req = kzalloc(sizeof(*dump_info->read_buf_req), GFP_KERNEL);
+	if (!dump_info->read_buf_req) {
+		ret = -ENOMEM;
+		goto free_dump_info;
+	}
+
+	/* Crashdump meta table buffer */
+	dump_info->tbl_addr = vzalloc(debug_info->tbl_len);
+	if (!dump_info->tbl_addr) {
+		ret = -ENOMEM;
+		goto free_read_buf_req;
+	}
+
+	dump_info->tbl_addr_dev = debug_info->tbl_addr;
+	dump_info->tbl_len = debug_info->tbl_len;
+
+	return dump_info;
+
+free_read_buf_req:
+	kfree(dump_info->read_buf_req);
+free_dump_info:
+	kfree(dump_info);
+out:
+	return ERR_PTR(ret);
+}
+
+static int dbg_xfer_info_rsp(struct qaic_device *qdev, struct dma_bridge_chan *dbc,
+			     struct ssr_debug_transfer_info *debug_info)
+{
+	struct ssr_debug_transfer_info_rsp *debug_rsp;
+	struct ssr_crashdump *ssr_crash = NULL;
+	int ret = 0, ret2;
+
+	debug_rsp = kmalloc(sizeof(*debug_rsp), GFP_KERNEL);
+	if (!debug_rsp)
+		return -ENOMEM;
+
+	if (!qdev->ssr_mhi_buf) {
+		ret = -ENOMEM;
+		goto send_rsp;
+	}
+
+	if (dbc->state != DBC_STATE_BEFORE_POWER_UP) {
+		ret = -EINVAL;
+		goto send_rsp;
+	}
+
+	ssr_crash = qdev->ssr_mhi_buf;
+	ssr_crash->dump_info = alloc_dump_info(qdev, debug_info);
+	if (IS_ERR(ssr_crash->dump_info)) {
+		ret = PTR_ERR(ssr_crash->dump_info);
+		ssr_crash->dump_info = NULL;
+	}
+
+send_rsp:
+	debug_rsp->hdr.cmd = cpu_to_le32(DEBUG_TRANSFER_INFO_RSP);
+	debug_rsp->hdr.len = cpu_to_le32(sizeof(*debug_rsp));
+	debug_rsp->hdr.dbc_id = cpu_to_le32(dbc->id);
+	/*
+	 * 0 = Return an ACK confirming the host is ready to download crashdump
+	 * 1 = Return an NACK confirming the host is not ready to download crashdump
+	 */
+	debug_rsp->ret = cpu_to_le32(ret ? 1 : 0);
+
+	ret2 = mhi_queue_buf(qdev->ssr_ch, DMA_TO_DEVICE, debug_rsp, sizeof(*debug_rsp), MHI_EOT);
+	if (ret2) {
+		free_ssr_dump_info(ssr_crash);
+		kfree(debug_rsp);
+		return ret2;
+	}
+
+	return ret;
+}
+
+static void dbg_xfer_done_rsp(struct qaic_device *qdev, struct dma_bridge_chan *dbc,
+			      struct ssr_debug_transfer_done_rsp *xfer_rsp)
+{
+	struct ssr_crashdump *ssr_crash = qdev->ssr_mhi_buf;
+	u32 status = le32_to_cpu(xfer_rsp->ret);
+	struct device *dev = &qdev->pdev->dev;
+	struct ssr_dump_info *dump_info;
+
+	dump_info = ssr_crash->dump_info;
+	if (!dump_info)
+		return;
+
+	if (status) {
+		free_ssr_dump_info(ssr_crash);
+		return;
+	}
+
+	dev_coredumpv(dev, dump_info->dump_addr, dump_info->dump_sz, GFP_KERNEL);
+	/* dev_coredumpv will free dump_info->dump_addr */
+	dump_info->dump_addr = NULL;
+	free_ssr_dump_info(ssr_crash);
+}
+
+static void ssr_worker(struct work_struct *work)
+{
+	struct ssr_resp *resp = container_of(work, struct ssr_resp, work);
+	struct ssr_hdr *hdr = (struct ssr_hdr *)resp->data;
+	struct ssr_dump_info *dump_info = NULL;
+	struct qaic_device *qdev = resp->qdev;
+	struct ssr_crashdump *ssr_crash;
+	struct ssr_event_rsp *event_rsp;
+	struct dma_bridge_chan *dbc;
+	struct ssr_event *event;
+	u32 ssr_event_ack;
+	int ret;
+
+	le32_to_cpus(&hdr->cmd);
+	le32_to_cpus(&hdr->len);
+	le32_to_cpus(&hdr->dbc_id);
+
+	if (hdr->len > SSR_RESP_MSG_SZ)
+		goto out;
+
+	if (hdr->dbc_id >= qdev->num_dbc)
+		goto out;
+
+	dbc = &qdev->dbc[hdr->dbc_id];
+
+	switch (hdr->cmd) {
+	case DEBUG_TRANSFER_INFO:
+		ret = dbg_xfer_info_rsp(qdev, dbc, (struct ssr_debug_transfer_info *)resp->data);
+		if (ret)
+			break;
+
+		ssr_crash = qdev->ssr_mhi_buf;
+		dump_info = ssr_crash->dump_info;
+		dump_info->dbc = dbc;
+		dump_info->resp = resp;
+
+		/* Start by downloading debug table */
+		ret = mem_read_req(qdev, dump_info->tbl_addr_dev,
+				   min(dump_info->tbl_len, SSR_MEM_READ_CHUNK_SIZE));
+		if (ret) {
+			free_ssr_dump_info(ssr_crash);
+			break;
+		}
+
+		/*
+		 * Till now everything went fine, which means that we will be
+		 * collecting crashdump chunk by chunk. Do not queue a response
+		 * buffer for SSR cmds till the crashdump is complete.
+		 */
+		return;
+	case SSR_EVENT:
+		event = (struct ssr_event *)hdr;
+		le32_to_cpus(&event->event);
+		ssr_event_ack = event->event;
+		ssr_crash = qdev->ssr_mhi_buf;
+
+		switch (event->event) {
+		case BEFORE_SHUTDOWN:
+			set_dbc_state(qdev, hdr->dbc_id, DBC_STATE_BEFORE_SHUTDOWN);
+			qaic_dbc_enter_ssr(qdev, hdr->dbc_id);
+			break;
+		case AFTER_SHUTDOWN:
+			set_dbc_state(qdev, hdr->dbc_id, DBC_STATE_AFTER_SHUTDOWN);
+			break;
+		case BEFORE_POWER_UP:
+			set_dbc_state(qdev, hdr->dbc_id, DBC_STATE_BEFORE_POWER_UP);
+			break;
+		case AFTER_POWER_UP:
+			/*
+			 * If dump info is a non NULL value it means that we
+			 * have received this SSR event while downloading a
+			 * crashdump for this DBC is still in progress. NACK
+			 * the SSR event
+			 */
+			if (ssr_crash && ssr_crash->dump_info) {
+				free_ssr_dump_info(ssr_crash);
+				ssr_event_ack = SSR_EVENT_NACK;
+				break;
+			}
+
+			set_dbc_state(qdev, hdr->dbc_id, DBC_STATE_AFTER_POWER_UP);
+			break;
+		default:
+			break;
+		}
+
+		event_rsp = kmalloc(sizeof(*event_rsp), GFP_KERNEL);
+		if (!event_rsp)
+			break;
+
+		event_rsp->hdr.cmd = cpu_to_le32(SSR_EVENT_RSP);
+		event_rsp->hdr.len = cpu_to_le32(sizeof(*event_rsp));
+		event_rsp->hdr.dbc_id = cpu_to_le32(hdr->dbc_id);
+		event_rsp->event = cpu_to_le32(ssr_event_ack);
+
+		ret = mhi_queue_buf(qdev->ssr_ch, DMA_TO_DEVICE, event_rsp, sizeof(*event_rsp),
+				    MHI_EOT);
+		if (ret)
+			kfree(event_rsp);
+
+		if (event->event == AFTER_POWER_UP && ssr_event_ack != SSR_EVENT_NACK) {
+			qaic_dbc_exit_ssr(qdev);
+			set_dbc_state(qdev, hdr->dbc_id, DBC_STATE_IDLE);
+		}
+
+		break;
+	case DEBUG_TRANSFER_DONE_RSP:
+		dbg_xfer_done_rsp(qdev, dbc, (struct ssr_debug_transfer_done_rsp *)hdr);
+		break;
+	default:
+		break;
+	}
+
+out:
+	ret = mhi_queue_buf(qdev->ssr_ch, DMA_FROM_DEVICE, resp->data, SSR_RESP_MSG_SZ, MHI_EOT);
+	if (ret)
+		kfree(resp);
+}
+
+static int qaic_ssr_mhi_probe(struct mhi_device *mhi_dev, const struct mhi_device_id *id)
+{
+	struct qaic_device *qdev = pci_get_drvdata(to_pci_dev(mhi_dev->mhi_cntrl->cntrl_dev));
+	struct ssr_resp *resp;
+	int ret;
+
+	ret = mhi_prepare_for_transfer(mhi_dev);
+	if (ret)
+		return ret;
+
+	resp = kzalloc(sizeof(*resp) + SSR_RESP_MSG_SZ, GFP_KERNEL);
+	if (!resp) {
+		mhi_unprepare_from_transfer(mhi_dev);
+		return -ENOMEM;
+	}
+
+	resp->qdev = qdev;
+	INIT_WORK(&resp->work, ssr_worker);
+
+	ret = mhi_queue_buf(mhi_dev, DMA_FROM_DEVICE, resp->data, SSR_RESP_MSG_SZ, MHI_EOT);
+	if (ret) {
+		kfree(resp);
+		mhi_unprepare_from_transfer(mhi_dev);
+		return ret;
+	}
+
+	dev_set_drvdata(&mhi_dev->dev, qdev);
+	qdev->ssr_ch = mhi_dev;
+
+	return 0;
+}
+
+static void qaic_ssr_mhi_remove(struct mhi_device *mhi_dev)
+{
+	struct qaic_device *qdev;
+
+	qdev = dev_get_drvdata(&mhi_dev->dev);
+	mhi_unprepare_from_transfer(qdev->ssr_ch);
+	qdev->ssr_ch = NULL;
+}
+
+static void qaic_ssr_mhi_ul_xfer_cb(struct mhi_device *mhi_dev, struct mhi_result *mhi_result)
+{
+	struct qaic_device *qdev = dev_get_drvdata(&mhi_dev->dev);
+	struct ssr_crashdump *ssr_crash = qdev->ssr_mhi_buf;
+	struct _ssr_hdr *hdr = mhi_result->buf_addr;
+	struct ssr_dump_info *dump_info;
+
+	if (mhi_result->transaction_status) {
+		kfree(mhi_result->buf_addr);
+		return;
+	}
+
+	/*
+	 * MEMORY READ is used to download crashdump. And crashdump is
+	 * downloaded chunk by chunk in a series of MEMORY READ SSR commands.
+	 * Hence to avoid too many kmalloc() and kfree() of the same MEMORY READ
+	 * request buffer, we allocate only one such buffer and free it only
+	 * once.
+	 */
+	if (le32_to_cpu(hdr->cmd) == MEMORY_READ) {
+		dump_info = ssr_crash->dump_info;
+		if (dump_info) {
+			dump_info->read_buf_req_queued = false;
+			return;
+		}
+	}
+
+	kfree(mhi_result->buf_addr);
+}
+
+static void qaic_ssr_mhi_dl_xfer_cb(struct mhi_device *mhi_dev, struct mhi_result *mhi_result)
+{
+	struct ssr_resp *resp = container_of(mhi_result->buf_addr, struct ssr_resp, data);
+	struct qaic_device *qdev = dev_get_drvdata(&mhi_dev->dev);
+	struct ssr_crashdump *ssr_crash = qdev->ssr_mhi_buf;
+	bool memory_read_rsp = false;
+
+	if (ssr_crash && ssr_crash->data == mhi_result->buf_addr)
+		memory_read_rsp = true;
+
+	if (mhi_result->transaction_status) {
+		/* Do not free SSR crashdump buffer as it allocated via managed APIs */
+		if (!memory_read_rsp)
+			kfree(resp);
+		return;
+	}
+
+	if (memory_read_rsp)
+		queue_work(qdev->ssr_wq, &ssr_crash->work);
+	else
+		queue_work(qdev->ssr_wq, &resp->work);
+}
+
+static const struct mhi_device_id qaic_ssr_mhi_match_table[] = {
+	{ .chan = "QAIC_SSR", },
+	{},
+};
+
+static struct mhi_driver qaic_ssr_mhi_driver = {
+	.id_table = qaic_ssr_mhi_match_table,
+	.remove = qaic_ssr_mhi_remove,
+	.probe = qaic_ssr_mhi_probe,
+	.ul_xfer_cb = qaic_ssr_mhi_ul_xfer_cb,
+	.dl_xfer_cb = qaic_ssr_mhi_dl_xfer_cb,
+	.driver = {
+		.name = "qaic_ssr",
+	},
+};
+
+int qaic_ssr_init(struct qaic_device *qdev, struct drm_device *drm)
+{
+	struct ssr_crashdump *ssr_crash;
+
+	qdev->ssr_dbc = QAIC_SSR_DBC_SENTINEL;
+
+	/*
+	 * Device requests only one SSR at a time. So allocating only one
+	 * buffer to download crashdump is good enough.
+	 */
+	ssr_crash = drmm_kzalloc(drm, SSR_MHI_BUF_SIZE, GFP_KERNEL);
+	if (!ssr_crash)
+		return -ENOMEM;
+
+	ssr_crash->qdev = qdev;
+	INIT_WORK(&ssr_crash->work, ssr_dump_worker);
+	qdev->ssr_mhi_buf = ssr_crash;
+
+	return 0;
+}
+
+int qaic_ssr_register(void)
+{
+	return mhi_driver_register(&qaic_ssr_mhi_driver);
+}
+
+void qaic_ssr_unregister(void)
+{
+	mhi_driver_unregister(&qaic_ssr_mhi_driver);
+}
diff --git a/drivers/accel/qaic/qaic_ssr.h b/drivers/accel/qaic/qaic_ssr.h
new file mode 100644
index 000000000000..97ccff305750
--- /dev/null
+++ b/drivers/accel/qaic/qaic_ssr.h
@@ -0,0 +1,17 @@
+/* SPDX-License-Identifier: GPL-2.0-only
+ *
+ * Copyright (c) 2020, The Linux Foundation. All rights reserved.
+ * Copyright (c) 2021, 2024 Qualcomm Innovation Center, Inc. All rights reserved.
+ */
+
+#ifndef __QAIC_SSR_H__
+#define __QAIC_SSR_H__
+
+struct drm_device;
+struct qaic_device;
+
+int qaic_ssr_register(void);
+void qaic_ssr_unregister(void);
+void qaic_clean_up_ssr(struct qaic_device *qdev);
+int qaic_ssr_init(struct qaic_device *qdev, struct drm_device *drm);
+#endif /* __QAIC_SSR_H__ */
diff --git a/drivers/accel/qaic/qaic_sysfs.c b/drivers/accel/qaic/qaic_sysfs.c
new file mode 100644
index 000000000000..e0afb0ffb589
--- /dev/null
+++ b/drivers/accel/qaic/qaic_sysfs.c
@@ -0,0 +1,109 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+/* Copyright (c) 2020-2025, The Linux Foundation. All rights reserved. */
+
+#include <drm/drm_file.h>
+#include <drm/drm_managed.h>
+#include <linux/device.h>
+#include <linux/kernel.h>
+#include <linux/kobject.h>
+#include <linux/mutex.h>
+#include <linux/sysfs.h>
+
+#include "qaic.h"
+
+#define NAME_LEN		14
+
+struct dbc_attribute {
+	struct device_attribute dev_attr;
+	u32 dbc_id;
+	char name[NAME_LEN];
+};
+
+static ssize_t dbc_state_show(struct device *dev, struct device_attribute *a, char *buf)
+{
+	struct dbc_attribute *dbc_attr = container_of(a, struct dbc_attribute, dev_attr);
+	struct drm_minor *minor = dev_get_drvdata(dev);
+	struct qaic_device *qdev;
+
+	qdev = to_qaic_device(minor->dev);
+	return sysfs_emit(buf, "%d\n", qdev->dbc[dbc_attr->dbc_id].state);
+}
+
+void set_dbc_state(struct qaic_device *qdev, u32 dbc_id, unsigned int state)
+{
+	struct device *kdev = to_accel_kdev(qdev->qddev);
+	char *envp[3] = {};
+	char state_str[16];
+	char id_str[12];
+
+	envp[0] = id_str;
+	envp[1] = state_str;
+
+	if (state >= DBC_STATE_MAX)
+		return;
+	if (dbc_id >= qdev->num_dbc)
+		return;
+	if (state == qdev->dbc[dbc_id].state)
+		return;
+
+	scnprintf(id_str, ARRAY_SIZE(id_str), "DBC_ID=%d", dbc_id);
+	scnprintf(state_str, ARRAY_SIZE(state_str), "DBC_STATE=%d", state);
+
+	qdev->dbc[dbc_id].state = state;
+	kobject_uevent_env(&kdev->kobj, KOBJ_CHANGE, envp);
+}
+
+int qaic_sysfs_init(struct qaic_drm_device *qddev)
+{
+	struct device *kdev = to_accel_kdev(qddev);
+	struct drm_device *drm = to_drm(qddev);
+	u32 num_dbc = qddev->qdev->num_dbc;
+	struct dbc_attribute *dbc_attrs;
+	int i, ret;
+
+	dbc_attrs = drmm_kcalloc(drm, num_dbc, sizeof(*dbc_attrs), GFP_KERNEL);
+	if (!dbc_attrs)
+		return -ENOMEM;
+
+	for (i = 0; i < num_dbc; ++i) {
+		struct dbc_attribute *dbc_attr = &dbc_attrs[i];
+
+		sysfs_attr_init(&dbc_attr->dev_attr.attr);
+		dbc_attr->dbc_id = i;
+		scnprintf(dbc_attr->name, NAME_LEN, "dbc%d_state", i);
+		dbc_attr->dev_attr.attr.name = dbc_attr->name;
+		dbc_attr->dev_attr.attr.mode = 0444;
+		dbc_attr->dev_attr.show = dbc_state_show;
+		ret = sysfs_create_file(&kdev->kobj, &dbc_attr->dev_attr.attr);
+		if (ret) {
+			int j;
+
+			for (j = 0; j < i; ++j) {
+				dbc_attr = &dbc_attrs[j];
+				sysfs_remove_file(&kdev->kobj, &dbc_attr->dev_attr.attr);
+			}
+			drmm_kfree(drm, dbc_attrs);
+			return ret;
+		}
+	}
+
+	qddev->sysfs_attrs = dbc_attrs;
+	return 0;
+}
+
+void qaic_sysfs_remove(struct qaic_drm_device *qddev)
+{
+	struct dbc_attribute *dbc_attrs = qddev->sysfs_attrs;
+	struct device *kdev = to_accel_kdev(qddev);
+	u32 num_dbc = qddev->qdev->num_dbc;
+	int i;
+
+	if (!dbc_attrs)
+		return;
+
+	qddev->sysfs_attrs = NULL;
+	for (i = 0; i < num_dbc; ++i)
+		sysfs_remove_file(&kdev->kobj, &dbc_attrs[i].dev_attr.attr);
+	drmm_kfree(to_drm(qddev), dbc_attrs);
+}
diff --git a/drivers/accel/qaic/qaic_timesync.c b/drivers/accel/qaic/qaic_timesync.c
index 3fac540f8e03..8af2475f4f36 100644
--- a/drivers/accel/qaic/qaic_timesync.c
+++ b/drivers/accel/qaic/qaic_timesync.c
@@ -171,6 +171,13 @@ mod_timer:
 		dev_err(mqtsdev->dev, "%s mod_timer error:%d\n", __func__, ret);
 }
 
+void qaic_mqts_ch_stop_timer(struct mhi_device *mhi_dev)
+{
+	struct mqts_dev *mqtsdev = dev_get_drvdata(&mhi_dev->dev);
+
+	timer_delete_sync(&mqtsdev->timer);
+}
+
 static int qaic_timesync_probe(struct mhi_device *mhi_dev, const struct mhi_device_id *id)
 {
 	struct qaic_device *qdev = pci_get_drvdata(to_pci_dev(mhi_dev->mhi_cntrl->cntrl_dev));
@@ -206,6 +213,7 @@ static int qaic_timesync_probe(struct mhi_device *mhi_dev, const struct mhi_devi
 	timer->expires = jiffies + msecs_to_jiffies(timesync_delay_ms);
 	add_timer(timer);
 	dev_set_drvdata(&mhi_dev->dev, mqtsdev);
+	qdev->mqts_ch = mhi_dev;
 
 	return 0;
 
@@ -221,6 +229,7 @@ static void qaic_timesync_remove(struct mhi_device *mhi_dev)
 {
 	struct mqts_dev *mqtsdev = dev_get_drvdata(&mhi_dev->dev);
 
+	mqtsdev->qdev->mqts_ch = NULL;
 	timer_delete_sync(&mqtsdev->timer);
 	mhi_unprepare_from_transfer(mqtsdev->mhi_dev);
 	kfree(mqtsdev->sync_msg);
diff --git a/drivers/accel/qaic/qaic_timesync.h b/drivers/accel/qaic/qaic_timesync.h
index 851b7acd43bb..77b9c2b55057 100644
--- a/drivers/accel/qaic/qaic_timesync.h
+++ b/drivers/accel/qaic/qaic_timesync.h
@@ -6,6 +6,9 @@
 #ifndef __QAIC_TIMESYNC_H__
 #define __QAIC_TIMESYNC_H__
 
+#include <linux/mhi.h>
+
 int qaic_timesync_init(void);
 void qaic_timesync_deinit(void);
+void qaic_mqts_ch_stop_timer(struct mhi_device *mhi_dev);
 #endif /* __QAIC_TIMESYNC_H__ */
diff --git a/drivers/accel/qaic/sahara.c b/drivers/accel/qaic/sahara.c
index 3ebcc1f7ff58..fd3c3b2d1fd3 100644
--- a/drivers/accel/qaic/sahara.c
+++ b/drivers/accel/qaic/sahara.c
@@ -159,6 +159,7 @@ struct sahara_context {
 	struct sahara_packet		*rx;
 	struct work_struct		fw_work;
 	struct work_struct		dump_work;
+	struct work_struct		read_data_work;
 	struct mhi_device		*mhi_dev;
 	const char * const		*image_table;
 	u32				table_size;
@@ -174,7 +175,10 @@ struct sahara_context {
 	u64				dump_image_offset;
 	void				*mem_dump_freespace;
 	u64				dump_images_left;
+	u32				read_data_offset;
+	u32				read_data_length;
 	bool				is_mem_dump_mode;
+	bool				non_streaming;
 };
 
 static const char * const aic100_image_table[] = {
@@ -194,6 +198,7 @@ static const char * const aic200_image_table[] = {
 	[23] = "qcom/aic200/aop.mbn",
 	[32] = "qcom/aic200/tz.mbn",
 	[33] = "qcom/aic200/hypvm.mbn",
+	[38] = "qcom/aic200/xbl_config.elf",
 	[39] = "qcom/aic200/aic200_abl.elf",
 	[40] = "qcom/aic200/apdp.mbn",
 	[41] = "qcom/aic200/devcfg.mbn",
@@ -202,6 +207,7 @@ static const char * const aic200_image_table[] = {
 	[49] = "qcom/aic200/shrm.elf",
 	[50] = "qcom/aic200/cpucp.elf",
 	[51] = "qcom/aic200/aop_devcfg.mbn",
+	[54] = "qcom/aic200/qupv3fw.elf",
 	[57] = "qcom/aic200/cpucp_dtbs.elf",
 	[62] = "qcom/aic200/uefi_dtbs.elf",
 	[63] = "qcom/aic200/xbl_ac_config.mbn",
@@ -213,9 +219,15 @@ static const char * const aic200_image_table[] = {
 	[69] = "qcom/aic200/dcd.mbn",
 	[73] = "qcom/aic200/gearvm.mbn",
 	[74] = "qcom/aic200/sti.bin",
-	[75] = "qcom/aic200/pvs.bin",
+	[76] = "qcom/aic200/tz_qti_config.mbn",
+	[78] = "qcom/aic200/pvs.bin",
 };
 
+static bool is_streaming(struct sahara_context *context)
+{
+	return !context->non_streaming;
+}
+
 static int sahara_find_image(struct sahara_context *context, u32 image_id)
 {
 	int ret;
@@ -265,6 +277,8 @@ static void sahara_send_reset(struct sahara_context *context)
 	int ret;
 
 	context->is_mem_dump_mode = false;
+	context->read_data_offset = 0;
+	context->read_data_length = 0;
 
 	context->tx[0]->cmd = cpu_to_le32(SAHARA_RESET_CMD);
 	context->tx[0]->length = cpu_to_le32(SAHARA_RESET_LENGTH);
@@ -319,9 +333,39 @@ static void sahara_hello(struct sahara_context *context)
 		dev_err(&context->mhi_dev->dev, "Unable to send hello response %d\n", ret);
 }
 
+static int read_data_helper(struct sahara_context *context, int buf_index)
+{
+	enum mhi_flags mhi_flag;
+	u32 pkt_data_len;
+	int ret;
+
+	pkt_data_len = min(context->read_data_length, SAHARA_PACKET_MAX_SIZE);
+
+	memcpy(context->tx[buf_index],
+	       &context->firmware->data[context->read_data_offset],
+	       pkt_data_len);
+
+	context->read_data_offset += pkt_data_len;
+	context->read_data_length -= pkt_data_len;
+
+	if (is_streaming(context) || !context->read_data_length)
+		mhi_flag = MHI_EOT;
+	else
+		mhi_flag = MHI_CHAIN;
+
+	ret = mhi_queue_buf(context->mhi_dev, DMA_TO_DEVICE,
+			    context->tx[buf_index], pkt_data_len, mhi_flag);
+	if (ret) {
+		dev_err(&context->mhi_dev->dev, "Unable to send read_data response %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
+
 static void sahara_read_data(struct sahara_context *context)
 {
-	u32 image_id, data_offset, data_len, pkt_data_len;
+	u32 image_id, data_offset, data_len;
 	int ret;
 	int i;
 
@@ -357,7 +401,7 @@ static void sahara_read_data(struct sahara_context *context)
 	 * and is not needed here on error.
 	 */
 
-	if (data_len > SAHARA_TRANSFER_MAX_SIZE) {
+	if (context->non_streaming && data_len > SAHARA_TRANSFER_MAX_SIZE) {
 		dev_err(&context->mhi_dev->dev, "Malformed read_data packet - data len %d exceeds max xfer size %d\n",
 			data_len, SAHARA_TRANSFER_MAX_SIZE);
 		sahara_send_reset(context);
@@ -378,22 +422,18 @@ static void sahara_read_data(struct sahara_context *context)
 		return;
 	}
 
-	for (i = 0; i < SAHARA_NUM_TX_BUF && data_len; ++i) {
-		pkt_data_len = min(data_len, SAHARA_PACKET_MAX_SIZE);
-
-		memcpy(context->tx[i], &context->firmware->data[data_offset], pkt_data_len);
+	context->read_data_offset = data_offset;
+	context->read_data_length = data_len;
 
-		data_offset += pkt_data_len;
-		data_len -= pkt_data_len;
+	if (is_streaming(context)) {
+		schedule_work(&context->read_data_work);
+		return;
+	}
 
-		ret = mhi_queue_buf(context->mhi_dev, DMA_TO_DEVICE,
-				    context->tx[i], pkt_data_len,
-				    !data_len ? MHI_EOT : MHI_CHAIN);
-		if (ret) {
-			dev_err(&context->mhi_dev->dev, "Unable to send read_data response %d\n",
-				ret);
-			return;
-		}
+	for (i = 0; i < SAHARA_NUM_TX_BUF && context->read_data_length; ++i) {
+		ret = read_data_helper(context, i);
+		if (ret)
+			break;
 	}
 }
 
@@ -538,6 +578,7 @@ static void sahara_parse_dump_table(struct sahara_context *context)
 	struct sahara_memory_dump_meta_v1 *dump_meta;
 	u64 table_nents;
 	u64 dump_length;
+	u64 mul_bytes;
 	int ret;
 	u64 i;
 
@@ -551,8 +592,9 @@ static void sahara_parse_dump_table(struct sahara_context *context)
 		dev_table[i].description[SAHARA_TABLE_ENTRY_STR_LEN - 1] = 0;
 		dev_table[i].filename[SAHARA_TABLE_ENTRY_STR_LEN - 1] = 0;
 
-		dump_length = size_add(dump_length, le64_to_cpu(dev_table[i].length));
-		if (dump_length == SIZE_MAX) {
+		if (check_add_overflow(dump_length,
+				       le64_to_cpu(dev_table[i].length),
+				       &dump_length)) {
 			/* Discard the dump */
 			sahara_send_reset(context);
 			return;
@@ -568,14 +610,17 @@ static void sahara_parse_dump_table(struct sahara_context *context)
 			dev_table[i].filename);
 	}
 
-	dump_length = size_add(dump_length, sizeof(*dump_meta));
-	if (dump_length == SIZE_MAX) {
+	if (check_add_overflow(dump_length, (u64)sizeof(*dump_meta), &dump_length)) {
 		/* Discard the dump */
 		sahara_send_reset(context);
 		return;
 	}
-	dump_length = size_add(dump_length, size_mul(sizeof(*image_out_table), table_nents));
-	if (dump_length == SIZE_MAX) {
+	if (check_mul_overflow((u64)sizeof(*image_out_table), table_nents, &mul_bytes)) {
+		/* Discard the dump */
+		sahara_send_reset(context);
+		return;
+	}
+	if (check_add_overflow(dump_length, mul_bytes, &dump_length)) {
 		/* Discard the dump */
 		sahara_send_reset(context);
 		return;
@@ -615,7 +660,7 @@ static void sahara_parse_dump_table(struct sahara_context *context)
 
 	/* Request the first chunk of the first image */
 	context->dump_image = &image_out_table[0];
-	dump_length = min(context->dump_image->length, SAHARA_READ_MAX_SIZE);
+	dump_length = min_t(u64, context->dump_image->length, SAHARA_READ_MAX_SIZE);
 	/* Avoid requesting EOI sized data so that we can identify errors */
 	if (dump_length == SAHARA_END_OF_IMAGE_LENGTH)
 		dump_length = SAHARA_END_OF_IMAGE_LENGTH / 2;
@@ -663,7 +708,7 @@ static void sahara_parse_dump_image(struct sahara_context *context)
 
 	/* Get next image chunk */
 	dump_length = context->dump_image->length - context->dump_image_offset;
-	dump_length = min(dump_length, SAHARA_READ_MAX_SIZE);
+	dump_length = min_t(u64, dump_length, SAHARA_READ_MAX_SIZE);
 	/* Avoid requesting EOI sized data so that we can identify errors */
 	if (dump_length == SAHARA_END_OF_IMAGE_LENGTH)
 		dump_length = SAHARA_END_OF_IMAGE_LENGTH / 2;
@@ -742,6 +787,13 @@ error:
 	sahara_send_reset(context);
 }
 
+static void sahara_read_data_processing(struct work_struct *work)
+{
+	struct sahara_context *context = container_of(work, struct sahara_context, read_data_work);
+
+	read_data_helper(context, 0);
+}
+
 static int sahara_mhi_probe(struct mhi_device *mhi_dev, const struct mhi_device_id *id)
 {
 	struct sahara_context *context;
@@ -756,34 +808,56 @@ static int sahara_mhi_probe(struct mhi_device *mhi_dev, const struct mhi_device_
 	if (!context->rx)
 		return -ENOMEM;
 
+	if (!strcmp(mhi_dev->mhi_cntrl->name, "AIC200")) {
+		context->image_table = aic200_image_table;
+		context->table_size = ARRAY_SIZE(aic200_image_table);
+	} else {
+		context->image_table = aic100_image_table;
+		context->table_size = ARRAY_SIZE(aic100_image_table);
+		context->non_streaming = true;
+	}
+
 	/*
-	 * AIC100 defines SAHARA_TRANSFER_MAX_SIZE as the largest value it
-	 * will request for READ_DATA. This is larger than
-	 * SAHARA_PACKET_MAX_SIZE, and we need 9x SAHARA_PACKET_MAX_SIZE to
-	 * cover SAHARA_TRANSFER_MAX_SIZE. When the remote side issues a
-	 * READ_DATA, it requires a transfer of the exact size requested. We
-	 * can use MHI_CHAIN to link multiple buffers into a single transfer
-	 * but the remote side will not consume the buffers until it sees an
-	 * EOT, thus we need to allocate enough buffers to put in the tx fifo
-	 * to cover an entire READ_DATA request of the max size.
+	 * There are two firmware implementations for READ_DATA handling.
+	 * The older "SBL" implementation defines a Sahara transfer size, and
+	 * expects that the response is a single transport transfer. If the
+	 * FW wants to transfer a file that is larger than the transfer size,
+	 * the FW will issue multiple READ_DATA commands. For this
+	 * implementation, we need to allocate enough buffers to contain the
+	 * entire Sahara transfer size.
+	 *
+	 * The newer "XBL" implementation does not define a maximum transfer
+	 * size and instead expects the data to be streamed over using the
+	 * transport level MTU. The FW will issue a single READ_DATA command
+	 * of whatever size, and consume multiple transport level transfers
+	 * until the expected amount of data is consumed. For this
+	 * implementation we only need a single buffer of the transport MTU
+	 * but we'll need to be able to use it multiple times for a single
+	 * READ_DATA request.
+	 *
+	 * AIC100 is the SBL implementation and defines SAHARA_TRANSFER_MAX_SIZE
+	 * and we need 9x SAHARA_PACKET_MAX_SIZE to cover that. We can use
+	 * MHI_CHAIN to link multiple buffers into a single transfer but the
+	 * remote side will not consume the buffers until it sees an EOT, thus
+	 * we need to allocate enough buffers to put in the tx fifo to cover an
+	 * entire READ_DATA request of the max size.
+	 *
+	 * AIC200 is the XBL implementation, and so a single buffer will work.
 	 */
 	for (i = 0; i < SAHARA_NUM_TX_BUF; ++i) {
-		context->tx[i] = devm_kzalloc(&mhi_dev->dev, SAHARA_PACKET_MAX_SIZE, GFP_KERNEL);
+		context->tx[i] = devm_kzalloc(&mhi_dev->dev,
+					      SAHARA_PACKET_MAX_SIZE,
+					      GFP_KERNEL);
 		if (!context->tx[i])
 			return -ENOMEM;
+		if (is_streaming(context))
+			break;
 	}
 
 	context->mhi_dev = mhi_dev;
 	INIT_WORK(&context->fw_work, sahara_processing);
 	INIT_WORK(&context->dump_work, sahara_dump_processing);
-
-	if (!strcmp(mhi_dev->mhi_cntrl->name, "AIC200")) {
-		context->image_table = aic200_image_table;
-		context->table_size = ARRAY_SIZE(aic200_image_table);
-	} else {
-		context->image_table = aic100_image_table;
-		context->table_size = ARRAY_SIZE(aic100_image_table);
-	}
+	INIT_WORK(&context->read_data_work, sahara_read_data_processing);
 
 	context->active_image_id = SAHARA_IMAGE_ID_NONE;
 	dev_set_drvdata(&mhi_dev->dev, context);
@@ -814,6 +888,10 @@ static void sahara_mhi_remove(struct mhi_device *mhi_dev)
 
 static void sahara_mhi_ul_xfer_cb(struct mhi_device *mhi_dev, struct mhi_result *mhi_result)
 {
+	struct sahara_context *context = dev_get_drvdata(&mhi_dev->dev);
+
+	if (!mhi_result->transaction_status && context->read_data_length && is_streaming(context))
+		schedule_work(&context->read_data_work);
 }
 
 static void sahara_mhi_dl_xfer_cb(struct mhi_device *mhi_dev, struct mhi_result *mhi_result)
diff --git a/drivers/accel/rocket/Kconfig b/drivers/accel/rocket/Kconfig
new file mode 100644
index 000000000000..16465abe0660
--- /dev/null
+++ b/drivers/accel/rocket/Kconfig
@@ -0,0 +1,24 @@
+# SPDX-License-Identifier: GPL-2.0-only
+
+config DRM_ACCEL_ROCKET
+	tristate "Rocket (support for Rockchip NPUs)"
+	depends on DRM_ACCEL
+	depends on (ARCH_ROCKCHIP && ARM64) || COMPILE_TEST
+	depends on ROCKCHIP_IOMMU || COMPILE_TEST
+	depends on MMU
+	select DRM_SCHED
+	select DRM_GEM_SHMEM_HELPER
+	help
+	  Choose this option if you have a Rockchip SoC that contains a
+	  compatible Neural Processing Unit (NPU), such as the RK3588. Called by
+	  Rockchip either RKNN or RKNPU, it accelerates inference of neural
+	  networks.
+
+	  The interface exposed to userspace is described in
+	  include/uapi/drm/rocket_accel.h and is used by the Rocket userspace
+	  driver in Mesa3D.
+
+	  If unsure, say N.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called rocket.
diff --git a/drivers/accel/rocket/Makefile b/drivers/accel/rocket/Makefile
new file mode 100644
index 000000000000..3713dfe223d6
--- /dev/null
+++ b/drivers/accel/rocket/Makefile
@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: GPL-2.0-only
+
+obj-$(CONFIG_DRM_ACCEL_ROCKET) := rocket.o
+
+rocket-y := \
+	rocket_core.o \
+	rocket_device.o \
+	rocket_drv.o \
+	rocket_gem.o \
+	rocket_job.o
diff --git a/drivers/accel/rocket/rocket_core.c b/drivers/accel/rocket/rocket_core.c
new file mode 100644
index 000000000000..abe7719c1db4
--- /dev/null
+++ b/drivers/accel/rocket/rocket_core.c
@@ -0,0 +1,110 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+
+#include <linux/clk.h>
+#include <linux/delay.h>
+#include <linux/dev_printk.h>
+#include <linux/dma-mapping.h>
+#include <linux/err.h>
+#include <linux/iommu.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+#include <linux/reset.h>
+
+#include "rocket_core.h"
+#include "rocket_job.h"
+
+int rocket_core_init(struct rocket_core *core)
+{
+	struct device *dev = core->dev;
+	struct platform_device *pdev = to_platform_device(dev);
+	u32 version;
+	int err = 0;
+
+	core->resets[0].id = "srst_a";
+	core->resets[1].id = "srst_h";
+	err = devm_reset_control_bulk_get_exclusive(&pdev->dev, ARRAY_SIZE(core->resets),
+						    core->resets);
+	if (err)
+		return dev_err_probe(dev, err, "failed to get resets for core %d\n", core->index);
+
+	err = devm_clk_bulk_get(dev, ARRAY_SIZE(core->clks), core->clks);
+	if (err)
+		return dev_err_probe(dev, err, "failed to get clocks for core %d\n", core->index);
+
+	core->pc_iomem = devm_platform_ioremap_resource_byname(pdev, "pc");
+	if (IS_ERR(core->pc_iomem)) {
+		dev_err(dev, "couldn't find PC registers %ld\n", PTR_ERR(core->pc_iomem));
+		return PTR_ERR(core->pc_iomem);
+	}
+
+	core->cna_iomem = devm_platform_ioremap_resource_byname(pdev, "cna");
+	if (IS_ERR(core->cna_iomem)) {
+		dev_err(dev, "couldn't find CNA registers %ld\n", PTR_ERR(core->cna_iomem));
+		return PTR_ERR(core->cna_iomem);
+	}
+
+	core->core_iomem = devm_platform_ioremap_resource_byname(pdev, "core");
+	if (IS_ERR(core->core_iomem)) {
+		dev_err(dev, "couldn't find CORE registers %ld\n", PTR_ERR(core->core_iomem));
+		return PTR_ERR(core->core_iomem);
+	}
+
+	dma_set_max_seg_size(dev, UINT_MAX);
+
+	err = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(40));
+	if (err)
+		return err;
+
+	core->iommu_group = iommu_group_get(dev);
+
+	err = rocket_job_init(core);
+	if (err)
+		return err;
+
+	pm_runtime_use_autosuspend(dev);
+
+	/*
+	 * As this NPU will be most often used as part of a media pipeline that
+	 * ends presenting in a display, choose 50 ms (~3 frames at 60Hz) as an
+	 * autosuspend delay as that will keep the device powered up while the
+	 * pipeline is running.
+	 */
+	pm_runtime_set_autosuspend_delay(dev, 50);
+
+	pm_runtime_enable(dev);
+
+	err = pm_runtime_resume_and_get(dev);
+	if (err) {
+		rocket_job_fini(core);
+		return err;
+	}
+
+	version = rocket_pc_readl(core, VERSION);
+	version += rocket_pc_readl(core, VERSION_NUM) & 0xffff;
+
+	pm_runtime_mark_last_busy(dev);
+	pm_runtime_put_autosuspend(dev);
+
+	dev_info(dev, "Rockchip NPU core %d version: %d\n", core->index, version);
+
+	return 0;
+}
+
+void rocket_core_fini(struct rocket_core *core)
+{
+	pm_runtime_dont_use_autosuspend(core->dev);
+	pm_runtime_disable(core->dev);
+	iommu_group_put(core->iommu_group);
+	core->iommu_group = NULL;
+	rocket_job_fini(core);
+}
+
+void rocket_core_reset(struct rocket_core *core)
+{
+	reset_control_bulk_assert(ARRAY_SIZE(core->resets), core->resets);
+
+	udelay(10);
+
+	reset_control_bulk_deassert(ARRAY_SIZE(core->resets), core->resets);
+}
diff --git a/drivers/accel/rocket/rocket_core.h b/drivers/accel/rocket/rocket_core.h
new file mode 100644
index 000000000000..f6d7382854ca
--- /dev/null
+++ b/drivers/accel/rocket/rocket_core.h
@@ -0,0 +1,64 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+
+#ifndef __ROCKET_CORE_H__
+#define __ROCKET_CORE_H__
+
+#include <drm/gpu_scheduler.h>
+#include <linux/clk.h>
+#include <linux/io.h>
+#include <linux/mutex_types.h>
+#include <linux/reset.h>
+
+#include "rocket_registers.h"
+
+#define rocket_pc_readl(core, reg) \
+	readl((core)->pc_iomem + (REG_PC_##reg))
+#define rocket_pc_writel(core, reg, value) \
+	writel(value, (core)->pc_iomem + (REG_PC_##reg))
+
+#define rocket_cna_readl(core, reg) \
+	readl((core)->cna_iomem + (REG_CNA_##reg) - REG_CNA_S_STATUS)
+#define rocket_cna_writel(core, reg, value) \
+	writel(value, (core)->cna_iomem + (REG_CNA_##reg) - REG_CNA_S_STATUS)
+
+#define rocket_core_readl(core, reg) \
+	readl((core)->core_iomem + (REG_CORE_##reg) - REG_CORE_S_STATUS)
+#define rocket_core_writel(core, reg, value) \
+	writel(value, (core)->core_iomem + (REG_CORE_##reg) - REG_CORE_S_STATUS)
+
+struct rocket_core {
+	struct device *dev;
+	struct rocket_device *rdev;
+	unsigned int index;
+
+	int irq;
+	void __iomem *pc_iomem;
+	void __iomem *cna_iomem;
+	void __iomem *core_iomem;
+	struct clk_bulk_data clks[4];
+	struct reset_control_bulk_data resets[2];
+
+	struct iommu_group *iommu_group;
+
+	struct mutex job_lock;
+	struct rocket_job *in_flight_job;
+
+	spinlock_t fence_lock;
+
+	struct {
+		struct workqueue_struct *wq;
+		struct work_struct work;
+		atomic_t pending;
+	} reset;
+
+	struct drm_gpu_scheduler sched;
+	u64 fence_context;
+	u64 emit_seqno;
+};
+
+int rocket_core_init(struct rocket_core *core);
+void rocket_core_fini(struct rocket_core *core);
+void rocket_core_reset(struct rocket_core *core);
+
+#endif
diff --git a/drivers/accel/rocket/rocket_device.c b/drivers/accel/rocket/rocket_device.c
new file mode 100644
index 000000000000..46e6ee1e72c5
--- /dev/null
+++ b/drivers/accel/rocket/rocket_device.c
@@ -0,0 +1,60 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+
+#include <drm/drm_drv.h>
+#include <linux/array_size.h>
+#include <linux/clk.h>
+#include <linux/dma-mapping.h>
+#include <linux/platform_device.h>
+#include <linux/of.h>
+
+#include "rocket_device.h"
+
+struct rocket_device *rocket_device_init(struct platform_device *pdev,
+					 const struct drm_driver *rocket_drm_driver)
+{
+	struct device *dev = &pdev->dev;
+	struct device_node *core_node;
+	struct rocket_device *rdev;
+	struct drm_device *ddev;
+	unsigned int num_cores = 0;
+	int err;
+
+	rdev = devm_drm_dev_alloc(dev, rocket_drm_driver, struct rocket_device, ddev);
+	if (IS_ERR(rdev))
+		return rdev;
+
+	ddev = &rdev->ddev;
+	dev_set_drvdata(dev, rdev);
+
+	for_each_compatible_node(core_node, NULL, "rockchip,rk3588-rknn-core")
+		if (of_device_is_available(core_node))
+			num_cores++;
+
+	rdev->cores = devm_kcalloc(dev, num_cores, sizeof(*rdev->cores), GFP_KERNEL);
+	if (!rdev->cores)
+		return ERR_PTR(-ENOMEM);
+
+	dma_set_max_seg_size(dev, UINT_MAX);
+
+	err = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(40));
+	if (err)
+		return ERR_PTR(err);
+
+	err = devm_mutex_init(dev, &rdev->sched_lock);
+	if (err)
+		return ERR_PTR(-ENOMEM);
+
+	err = drm_dev_register(ddev, 0);
+	if (err)
+		return ERR_PTR(err);
+
+	return rdev;
+}
+
+void rocket_device_fini(struct rocket_device *rdev)
+{
+	WARN_ON(rdev->num_cores > 0);
+
+	drm_dev_unregister(&rdev->ddev);
+}
diff --git a/drivers/accel/rocket/rocket_device.h b/drivers/accel/rocket/rocket_device.h
new file mode 100644
index 000000000000..ce662abc01d3
--- /dev/null
+++ b/drivers/accel/rocket/rocket_device.h
@@ -0,0 +1,30 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+
+#ifndef __ROCKET_DEVICE_H__
+#define __ROCKET_DEVICE_H__
+
+#include <drm/drm_device.h>
+#include <linux/clk.h>
+#include <linux/container_of.h>
+#include <linux/iommu.h>
+#include <linux/platform_device.h>
+
+#include "rocket_core.h"
+
+struct rocket_device {
+	struct drm_device ddev;
+
+	struct mutex sched_lock;
+
+	struct rocket_core *cores;
+	unsigned int num_cores;
+};
+
+struct rocket_device *rocket_device_init(struct platform_device *pdev,
+					 const struct drm_driver *rocket_drm_driver);
+void rocket_device_fini(struct rocket_device *rdev);
+#define to_rocket_device(drm_dev) \
+	((struct rocket_device *)(container_of((drm_dev), struct rocket_device, ddev)))
+
+#endif /* __ROCKET_DEVICE_H__ */
diff --git a/drivers/accel/rocket/rocket_drv.c b/drivers/accel/rocket/rocket_drv.c
new file mode 100644
index 000000000000..5c0b63f0a8f0
--- /dev/null
+++ b/drivers/accel/rocket/rocket_drv.c
@@ -0,0 +1,290 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+
+#include <drm/drm_accel.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_gem.h>
+#include <drm/drm_ioctl.h>
+#include <drm/rocket_accel.h>
+#include <linux/clk.h>
+#include <linux/err.h>
+#include <linux/iommu.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+
+#include "rocket_drv.h"
+#include "rocket_gem.h"
+#include "rocket_job.h"
+
+/*
+ * Facade device, used to expose a single DRM device to userspace, that
+ * schedules jobs to any RKNN cores in the system.
+ */
+static struct platform_device *drm_dev;
+static struct rocket_device *rdev;
+
+static void
+rocket_iommu_domain_destroy(struct kref *kref)
+{
+	struct rocket_iommu_domain *domain = container_of(kref, struct rocket_iommu_domain, kref);
+
+	iommu_domain_free(domain->domain);
+	domain->domain = NULL;
+	kfree(domain);
+}
+
+static struct rocket_iommu_domain*
+rocket_iommu_domain_create(struct device *dev)
+{
+	struct rocket_iommu_domain *domain = kmalloc(sizeof(*domain), GFP_KERNEL);
+	void *err;
+
+	if (!domain)
+		return ERR_PTR(-ENOMEM);
+
+	domain->domain = iommu_paging_domain_alloc(dev);
+	if (IS_ERR(domain->domain)) {
+		err = ERR_CAST(domain->domain);
+		kfree(domain);
+		return err;
+	}
+	kref_init(&domain->kref);
+
+	return domain;
+}
+
+struct rocket_iommu_domain *
+rocket_iommu_domain_get(struct rocket_file_priv *rocket_priv)
+{
+	kref_get(&rocket_priv->domain->kref);
+	return rocket_priv->domain;
+}
+
+void
+rocket_iommu_domain_put(struct rocket_iommu_domain *domain)
+{
+	kref_put(&domain->kref, rocket_iommu_domain_destroy);
+}
+
+static int
+rocket_open(struct drm_device *dev, struct drm_file *file)
+{
+	struct rocket_device *rdev = to_rocket_device(dev);
+	struct rocket_file_priv *rocket_priv;
+	u64 start, end;
+	int ret;
+
+	if (!try_module_get(THIS_MODULE))
+		return -EINVAL;
+
+	rocket_priv = kzalloc(sizeof(*rocket_priv), GFP_KERNEL);
+	if (!rocket_priv) {
+		ret = -ENOMEM;
+		goto err_put_mod;
+	}
+
+	rocket_priv->rdev = rdev;
+	rocket_priv->domain = rocket_iommu_domain_create(rdev->cores[0].dev);
+	if (IS_ERR(rocket_priv->domain)) {
+		ret = PTR_ERR(rocket_priv->domain);
+		goto err_free;
+	}
+
+	file->driver_priv = rocket_priv;
+
+	start = rocket_priv->domain->domain->geometry.aperture_start;
+	end = rocket_priv->domain->domain->geometry.aperture_end;
+	drm_mm_init(&rocket_priv->mm, start, end - start + 1);
+	mutex_init(&rocket_priv->mm_lock);
+
+	ret = rocket_job_open(rocket_priv);
+	if (ret)
+		goto err_mm_takedown;
+
+	return 0;
+
+err_mm_takedown:
+	mutex_destroy(&rocket_priv->mm_lock);
+	drm_mm_takedown(&rocket_priv->mm);
+	rocket_iommu_domain_put(rocket_priv->domain);
+err_free:
+	kfree(rocket_priv);
+err_put_mod:
+	module_put(THIS_MODULE);
+	return ret;
+}
+
+static void
+rocket_postclose(struct drm_device *dev, struct drm_file *file)
+{
+	struct rocket_file_priv *rocket_priv = file->driver_priv;
+
+	rocket_job_close(rocket_priv);
+	mutex_destroy(&rocket_priv->mm_lock);
+	drm_mm_takedown(&rocket_priv->mm);
+	rocket_iommu_domain_put(rocket_priv->domain);
+	kfree(rocket_priv);
+	module_put(THIS_MODULE);
+}
+
+static const struct drm_ioctl_desc rocket_drm_driver_ioctls[] = {
+#define ROCKET_IOCTL(n, func) \
+	DRM_IOCTL_DEF_DRV(ROCKET_##n, rocket_ioctl_##func, 0)
+
+	ROCKET_IOCTL(CREATE_BO, create_bo),
+	ROCKET_IOCTL(SUBMIT, submit),
+	ROCKET_IOCTL(PREP_BO, prep_bo),
+	ROCKET_IOCTL(FINI_BO, fini_bo),
+};
+
+DEFINE_DRM_ACCEL_FOPS(rocket_accel_driver_fops);
+
+/*
+ * Rocket driver version:
+ * - 1.0 - initial interface
+ */
+static const struct drm_driver rocket_drm_driver = {
+	.driver_features	= DRIVER_COMPUTE_ACCEL | DRIVER_GEM,
+	.open			= rocket_open,
+	.postclose		= rocket_postclose,
+	.gem_create_object	= rocket_gem_create_object,
+	.ioctls			= rocket_drm_driver_ioctls,
+	.num_ioctls		= ARRAY_SIZE(rocket_drm_driver_ioctls),
+	.fops			= &rocket_accel_driver_fops,
+	.name			= "rocket",
+	.desc			= "rocket DRM",
+};
+
+static int rocket_probe(struct platform_device *pdev)
+{
+	if (rdev == NULL) {
+		/* First core probing, initialize DRM device. */
+		rdev = rocket_device_init(drm_dev, &rocket_drm_driver);
+		if (IS_ERR(rdev)) {
+			dev_err(&pdev->dev, "failed to initialize rocket device\n");
+			return PTR_ERR(rdev);
+		}
+	}
+
+	unsigned int core = rdev->num_cores;
+
+	dev_set_drvdata(&pdev->dev, rdev);
+
+	rdev->cores[core].rdev = rdev;
+	rdev->cores[core].dev = &pdev->dev;
+	rdev->cores[core].index = core;
+
+	rdev->num_cores++;
+
+	return rocket_core_init(&rdev->cores[core]);
+}
+
+static void rocket_remove(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+
+	for (unsigned int core = 0; core < rdev->num_cores; core++) {
+		if (rdev->cores[core].dev == dev) {
+			rocket_core_fini(&rdev->cores[core]);
+			rdev->num_cores--;
+			break;
+		}
+	}
+
+	if (rdev->num_cores == 0) {
+		/* Last core removed, deinitialize DRM device. */
+		rocket_device_fini(rdev);
+		rdev = NULL;
+	}
+}
+
+static const struct of_device_id dt_match[] = {
+	{ .compatible = "rockchip,rk3588-rknn-core" },
+	{}
+};
+MODULE_DEVICE_TABLE(of, dt_match);
+
+static int find_core_for_dev(struct device *dev)
+{
+	struct rocket_device *rdev = dev_get_drvdata(dev);
+
+	for (unsigned int core = 0; core < rdev->num_cores; core++) {
+		if (dev == rdev->cores[core].dev)
+			return core;
+	}
+
+	return -1;
+}
+
+static int rocket_device_runtime_resume(struct device *dev)
+{
+	struct rocket_device *rdev = dev_get_drvdata(dev);
+	int core = find_core_for_dev(dev);
+	int err = 0;
+
+	if (core < 0)
+		return -ENODEV;
+
+	err = clk_bulk_prepare_enable(ARRAY_SIZE(rdev->cores[core].clks), rdev->cores[core].clks);
+	if (err) {
+		dev_err(dev, "failed to enable (%d) clocks for core %d\n", err, core);
+		return err;
+	}
+
+	return 0;
+}
+
+static int rocket_device_runtime_suspend(struct device *dev)
+{
+	struct rocket_device *rdev = dev_get_drvdata(dev);
+	int core = find_core_for_dev(dev);
+
+	if (core < 0)
+		return -ENODEV;
+
+	if (!rocket_job_is_idle(&rdev->cores[core]))
+		return -EBUSY;
+
+	clk_bulk_disable_unprepare(ARRAY_SIZE(rdev->cores[core].clks), rdev->cores[core].clks);
+
+	return 0;
+}
+
+EXPORT_GPL_DEV_PM_OPS(rocket_pm_ops) = {
+	RUNTIME_PM_OPS(rocket_device_runtime_suspend, rocket_device_runtime_resume, NULL)
+	SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend, pm_runtime_force_resume)
+};
+
+static struct platform_driver rocket_driver = {
+	.probe = rocket_probe,
+	.remove = rocket_remove,
+	.driver	 = {
+		.name = "rocket",
+		.pm = pm_ptr(&rocket_pm_ops),
+		.of_match_table = dt_match,
+	},
+};
+
+static int __init rocket_register(void)
+{
+	drm_dev = platform_device_register_simple("rknn", -1, NULL, 0);
+	if (IS_ERR(drm_dev))
+		return PTR_ERR(drm_dev);
+
+	return platform_driver_register(&rocket_driver);
+}
+
+static void __exit rocket_unregister(void)
+{
+	platform_driver_unregister(&rocket_driver);
+
+	platform_device_unregister(drm_dev);
+}
+
+module_init(rocket_register);
+module_exit(rocket_unregister);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("DRM driver for the Rockchip NPU IP");
+MODULE_AUTHOR("Tomeu Vizoso");
diff --git a/drivers/accel/rocket/rocket_drv.h b/drivers/accel/rocket/rocket_drv.h
new file mode 100644
index 000000000000..2c673bb99ccc
--- /dev/null
+++ b/drivers/accel/rocket/rocket_drv.h
@@ -0,0 +1,32 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+
+#ifndef __ROCKET_DRV_H__
+#define __ROCKET_DRV_H__
+
+#include <drm/drm_mm.h>
+#include <drm/gpu_scheduler.h>
+
+#include "rocket_device.h"
+
+extern const struct dev_pm_ops rocket_pm_ops;
+
+struct rocket_iommu_domain {
+	struct iommu_domain *domain;
+	struct kref kref;
+};
+
+struct rocket_file_priv {
+	struct rocket_device *rdev;
+
+	struct rocket_iommu_domain *domain;
+	struct drm_mm mm;
+	struct mutex mm_lock;
+
+	struct drm_sched_entity sched_entity;
+};
+
+struct rocket_iommu_domain *rocket_iommu_domain_get(struct rocket_file_priv *rocket_priv);
+void rocket_iommu_domain_put(struct rocket_iommu_domain *domain);
+
+#endif
diff --git a/drivers/accel/rocket/rocket_gem.c b/drivers/accel/rocket/rocket_gem.c
new file mode 100644
index 000000000000..624c4ecf5a34
--- /dev/null
+++ b/drivers/accel/rocket/rocket_gem.c
@@ -0,0 +1,182 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+
+#include <drm/drm_device.h>
+#include <drm/drm_print.h>
+#include <drm/drm_utils.h>
+#include <drm/rocket_accel.h>
+#include <linux/dma-mapping.h>
+#include <linux/iommu.h>
+
+#include "rocket_drv.h"
+#include "rocket_gem.h"
+
+static void rocket_gem_bo_free(struct drm_gem_object *obj)
+{
+	struct rocket_gem_object *bo = to_rocket_bo(obj);
+	struct rocket_file_priv *rocket_priv = bo->driver_priv;
+	size_t unmapped;
+
+	drm_WARN_ON(obj->dev, refcount_read(&bo->base.pages_use_count) > 1);
+
+	unmapped = iommu_unmap(bo->domain->domain, bo->mm.start, bo->size);
+	drm_WARN_ON(obj->dev, unmapped != bo->size);
+
+	mutex_lock(&rocket_priv->mm_lock);
+	drm_mm_remove_node(&bo->mm);
+	mutex_unlock(&rocket_priv->mm_lock);
+
+	rocket_iommu_domain_put(bo->domain);
+	bo->domain = NULL;
+
+	drm_gem_shmem_free(&bo->base);
+}
+
+static const struct drm_gem_object_funcs rocket_gem_funcs = {
+	.free = rocket_gem_bo_free,
+	.print_info = drm_gem_shmem_object_print_info,
+	.pin = drm_gem_shmem_object_pin,
+	.unpin = drm_gem_shmem_object_unpin,
+	.get_sg_table = drm_gem_shmem_object_get_sg_table,
+	.vmap = drm_gem_shmem_object_vmap,
+	.vunmap = drm_gem_shmem_object_vunmap,
+	.mmap = drm_gem_shmem_object_mmap,
+	.vm_ops = &drm_gem_shmem_vm_ops,
+};
+
+struct drm_gem_object *rocket_gem_create_object(struct drm_device *dev, size_t size)
+{
+	struct rocket_gem_object *obj;
+
+	obj = kzalloc(sizeof(*obj), GFP_KERNEL);
+	if (!obj)
+		return ERR_PTR(-ENOMEM);
+
+	obj->base.base.funcs = &rocket_gem_funcs;
+
+	return &obj->base.base;
+}
+
+int rocket_ioctl_create_bo(struct drm_device *dev, void *data, struct drm_file *file)
+{
+	struct rocket_file_priv *rocket_priv = file->driver_priv;
+	struct drm_rocket_create_bo *args = data;
+	struct drm_gem_shmem_object *shmem_obj;
+	struct rocket_gem_object *rkt_obj;
+	struct drm_gem_object *gem_obj;
+	struct sg_table *sgt;
+	int ret;
+
+	shmem_obj = drm_gem_shmem_create(dev, args->size);
+	if (IS_ERR(shmem_obj))
+		return PTR_ERR(shmem_obj);
+
+	gem_obj = &shmem_obj->base;
+	rkt_obj = to_rocket_bo(gem_obj);
+
+	rkt_obj->driver_priv = rocket_priv;
+	rkt_obj->domain = rocket_iommu_domain_get(rocket_priv);
+	rkt_obj->size = args->size;
+	rkt_obj->offset = 0;
+
+	ret = drm_gem_handle_create(file, gem_obj, &args->handle);
+	drm_gem_object_put(gem_obj);
+	if (ret)
+		goto err;
+
+	sgt = drm_gem_shmem_get_pages_sgt(shmem_obj);
+	if (IS_ERR(sgt)) {
+		ret = PTR_ERR(sgt);
+		goto err;
+	}
+
+	mutex_lock(&rocket_priv->mm_lock);
+	ret = drm_mm_insert_node_generic(&rocket_priv->mm, &rkt_obj->mm,
+					 rkt_obj->size, PAGE_SIZE,
+					 0, 0);
+	mutex_unlock(&rocket_priv->mm_lock);
+
+	ret = iommu_map_sgtable(rocket_priv->domain->domain,
+				rkt_obj->mm.start,
+				shmem_obj->sgt,
+				IOMMU_READ | IOMMU_WRITE);
+	if (ret < 0 || ret < args->size) {
+		drm_err(dev, "failed to map buffer: size=%d request_size=%u\n",
+			ret, args->size);
+		ret = -ENOMEM;
+		goto err_remove_node;
+	}
+
+	/* iommu_map_sgtable might have aligned the size */
+	rkt_obj->size = ret;
+	args->offset = drm_vma_node_offset_addr(&gem_obj->vma_node);
+	args->dma_address = rkt_obj->mm.start;
+
+	return 0;
+
+err_remove_node:
+	mutex_lock(&rocket_priv->mm_lock);
+	drm_mm_remove_node(&rkt_obj->mm);
+	mutex_unlock(&rocket_priv->mm_lock);
+
+err:
+	drm_gem_shmem_object_free(gem_obj);
+
+	return ret;
+}
+
+int rocket_ioctl_prep_bo(struct drm_device *dev, void *data, struct drm_file *file)
+{
+	struct drm_rocket_prep_bo *args = data;
+	unsigned long timeout = drm_timeout_abs_to_jiffies(args->timeout_ns);
+	struct drm_gem_object *gem_obj;
+	struct drm_gem_shmem_object *shmem_obj;
+	long ret = 0;
+
+	if (args->reserved != 0) {
+		drm_dbg(dev, "Reserved field in drm_rocket_prep_bo struct should be 0.\n");
+		return -EINVAL;
+	}
+
+	gem_obj = drm_gem_object_lookup(file, args->handle);
+	if (!gem_obj)
+		return -ENOENT;
+
+	ret = dma_resv_wait_timeout(gem_obj->resv, DMA_RESV_USAGE_WRITE, true, timeout);
+	if (!ret)
+		ret = timeout ? -ETIMEDOUT : -EBUSY;
+
+	shmem_obj = &to_rocket_bo(gem_obj)->base;
+
+	dma_sync_sgtable_for_cpu(dev->dev, shmem_obj->sgt, DMA_BIDIRECTIONAL);
+
+	drm_gem_object_put(gem_obj);
+
+	return ret;
+}
+
+int rocket_ioctl_fini_bo(struct drm_device *dev, void *data, struct drm_file *file)
+{
+	struct drm_rocket_fini_bo *args = data;
+	struct drm_gem_shmem_object *shmem_obj;
+	struct rocket_gem_object *rkt_obj;
+	struct drm_gem_object *gem_obj;
+
+	if (args->reserved != 0) {
+		drm_dbg(dev, "Reserved field in drm_rocket_fini_bo struct should be 0.\n");
+		return -EINVAL;
+	}
+
+	gem_obj = drm_gem_object_lookup(file, args->handle);
+	if (!gem_obj)
+		return -ENOENT;
+
+	rkt_obj = to_rocket_bo(gem_obj);
+	shmem_obj = &rkt_obj->base;
+
+	dma_sync_sgtable_for_device(dev->dev, shmem_obj->sgt, DMA_BIDIRECTIONAL);
+
+	drm_gem_object_put(gem_obj);
+
+	return 0;
+}
diff --git a/drivers/accel/rocket/rocket_gem.h b/drivers/accel/rocket/rocket_gem.h
new file mode 100644
index 000000000000..240430334509
--- /dev/null
+++ b/drivers/accel/rocket/rocket_gem.h
@@ -0,0 +1,34 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+
+#ifndef __ROCKET_GEM_H__
+#define __ROCKET_GEM_H__
+
+#include <drm/drm_gem_shmem_helper.h>
+
+struct rocket_gem_object {
+	struct drm_gem_shmem_object base;
+
+	struct rocket_file_priv *driver_priv;
+
+	struct rocket_iommu_domain *domain;
+	struct drm_mm_node mm;
+	size_t size;
+	u32 offset;
+};
+
+struct drm_gem_object *rocket_gem_create_object(struct drm_device *dev, size_t size);
+
+int rocket_ioctl_create_bo(struct drm_device *dev, void *data, struct drm_file *file);
+
+int rocket_ioctl_prep_bo(struct drm_device *dev, void *data, struct drm_file *file);
+
+int rocket_ioctl_fini_bo(struct drm_device *dev, void *data, struct drm_file *file);
+
+static inline
+struct  rocket_gem_object *to_rocket_bo(struct drm_gem_object *obj)
+{
+	return container_of(to_drm_gem_shmem_obj(obj), struct rocket_gem_object, base);
+}
+
+#endif
diff --git a/drivers/accel/rocket/rocket_job.c b/drivers/accel/rocket/rocket_job.c
new file mode 100644
index 000000000000..acd606160dc9
--- /dev/null
+++ b/drivers/accel/rocket/rocket_job.c
@@ -0,0 +1,637 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright 2019 Linaro, Ltd, Rob Herring <robh@kernel.org> */
+/* Copyright 2019 Collabora ltd. */
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+
+#include <drm/drm_print.h>
+#include <drm/drm_file.h>
+#include <drm/drm_gem.h>
+#include <drm/rocket_accel.h>
+#include <linux/interrupt.h>
+#include <linux/iommu.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+
+#include "rocket_core.h"
+#include "rocket_device.h"
+#include "rocket_drv.h"
+#include "rocket_job.h"
+#include "rocket_registers.h"
+
+#define JOB_TIMEOUT_MS 500
+
+static struct rocket_job *
+to_rocket_job(struct drm_sched_job *sched_job)
+{
+	return container_of(sched_job, struct rocket_job, base);
+}
+
+static const char *rocket_fence_get_driver_name(struct dma_fence *fence)
+{
+	return "rocket";
+}
+
+static const char *rocket_fence_get_timeline_name(struct dma_fence *fence)
+{
+	return "rockchip-npu";
+}
+
+static const struct dma_fence_ops rocket_fence_ops = {
+	.get_driver_name = rocket_fence_get_driver_name,
+	.get_timeline_name = rocket_fence_get_timeline_name,
+};
+
+static struct dma_fence *rocket_fence_create(struct rocket_core *core)
+{
+	struct dma_fence *fence;
+
+	fence = kzalloc(sizeof(*fence), GFP_KERNEL);
+	if (!fence)
+		return ERR_PTR(-ENOMEM);
+
+	dma_fence_init(fence, &rocket_fence_ops, &core->fence_lock,
+		       core->fence_context, ++core->emit_seqno);
+
+	return fence;
+}
+
+static int
+rocket_copy_tasks(struct drm_device *dev,
+		  struct drm_file *file_priv,
+		  struct drm_rocket_job *job,
+		  struct rocket_job *rjob)
+{
+	int ret = 0;
+
+	if (job->task_struct_size < sizeof(struct drm_rocket_task))
+		return -EINVAL;
+
+	rjob->task_count = job->task_count;
+
+	if (!rjob->task_count)
+		return 0;
+
+	rjob->tasks = kvmalloc_array(job->task_count, sizeof(*rjob->tasks), GFP_KERNEL);
+	if (!rjob->tasks) {
+		drm_dbg(dev, "Failed to allocate task array\n");
+		return -ENOMEM;
+	}
+
+	for (int i = 0; i < rjob->task_count; i++) {
+		struct drm_rocket_task task = {0};
+
+		if (copy_from_user(&task,
+				   u64_to_user_ptr(job->tasks) + i * job->task_struct_size,
+				   sizeof(task))) {
+			drm_dbg(dev, "Failed to copy incoming tasks\n");
+			ret = -EFAULT;
+			goto fail;
+		}
+
+		if (task.regcmd_count == 0) {
+			drm_dbg(dev, "regcmd_count field in drm_rocket_task should be > 0.\n");
+			ret = -EINVAL;
+			goto fail;
+		}
+
+		rjob->tasks[i].regcmd = task.regcmd;
+		rjob->tasks[i].regcmd_count = task.regcmd_count;
+	}
+
+	return 0;
+
+fail:
+	kvfree(rjob->tasks);
+	return ret;
+}
+
+static void rocket_job_hw_submit(struct rocket_core *core, struct rocket_job *job)
+{
+	struct rocket_task *task;
+	unsigned int extra_bit;
+
+	/* Don't queue the job if a reset is in progress */
+	if (atomic_read(&core->reset.pending))
+		return;
+
+	/* GO ! */
+
+	task = &job->tasks[job->next_task_idx];
+	job->next_task_idx++;
+
+	rocket_pc_writel(core, BASE_ADDRESS, 0x1);
+
+	 /* From rknpu, in the TRM this bit is marked as reserved */
+	extra_bit = 0x10000000 * core->index;
+	rocket_cna_writel(core, S_POINTER, CNA_S_POINTER_POINTER_PP_EN(1) |
+					   CNA_S_POINTER_EXECUTER_PP_EN(1) |
+					   CNA_S_POINTER_POINTER_PP_MODE(1) |
+					   extra_bit);
+
+	rocket_core_writel(core, S_POINTER, CORE_S_POINTER_POINTER_PP_EN(1) |
+					    CORE_S_POINTER_EXECUTER_PP_EN(1) |
+					    CORE_S_POINTER_POINTER_PP_MODE(1) |
+					    extra_bit);
+
+	rocket_pc_writel(core, BASE_ADDRESS, task->regcmd);
+	rocket_pc_writel(core, REGISTER_AMOUNTS,
+			 PC_REGISTER_AMOUNTS_PC_DATA_AMOUNT((task->regcmd_count + 1) / 2 - 1));
+
+	rocket_pc_writel(core, INTERRUPT_MASK, PC_INTERRUPT_MASK_DPU_0 | PC_INTERRUPT_MASK_DPU_1);
+	rocket_pc_writel(core, INTERRUPT_CLEAR, PC_INTERRUPT_CLEAR_DPU_0 | PC_INTERRUPT_CLEAR_DPU_1);
+
+	rocket_pc_writel(core, TASK_CON, PC_TASK_CON_RESERVED_0(1) |
+					 PC_TASK_CON_TASK_COUNT_CLEAR(1) |
+					 PC_TASK_CON_TASK_NUMBER(1) |
+					 PC_TASK_CON_TASK_PP_EN(1));
+
+	rocket_pc_writel(core, TASK_DMA_BASE_ADDR, PC_TASK_DMA_BASE_ADDR_DMA_BASE_ADDR(0x0));
+
+	rocket_pc_writel(core, OPERATION_ENABLE, PC_OPERATION_ENABLE_OP_EN(1));
+
+	dev_dbg(core->dev, "Submitted regcmd at 0x%llx to core %d", task->regcmd, core->index);
+}
+
+static int rocket_acquire_object_fences(struct drm_gem_object **bos,
+					int bo_count,
+					struct drm_sched_job *job,
+					bool is_write)
+{
+	int i, ret;
+
+	for (i = 0; i < bo_count; i++) {
+		ret = dma_resv_reserve_fences(bos[i]->resv, 1);
+		if (ret)
+			return ret;
+
+		ret = drm_sched_job_add_implicit_dependencies(job, bos[i],
+							      is_write);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static void rocket_attach_object_fences(struct drm_gem_object **bos,
+					int bo_count,
+					struct dma_fence *fence)
+{
+	int i;
+
+	for (i = 0; i < bo_count; i++)
+		dma_resv_add_fence(bos[i]->resv, fence, DMA_RESV_USAGE_WRITE);
+}
+
+static int rocket_job_push(struct rocket_job *job)
+{
+	struct rocket_device *rdev = job->rdev;
+	struct drm_gem_object **bos;
+	struct ww_acquire_ctx acquire_ctx;
+	int ret = 0;
+
+	bos = kvmalloc_array(job->in_bo_count + job->out_bo_count, sizeof(void *),
+			     GFP_KERNEL);
+	memcpy(bos, job->in_bos, job->in_bo_count * sizeof(void *));
+	memcpy(&bos[job->in_bo_count], job->out_bos, job->out_bo_count * sizeof(void *));
+
+	ret = drm_gem_lock_reservations(bos, job->in_bo_count + job->out_bo_count, &acquire_ctx);
+	if (ret)
+		goto err;
+
+	scoped_guard(mutex, &rdev->sched_lock) {
+		drm_sched_job_arm(&job->base);
+
+		job->inference_done_fence = dma_fence_get(&job->base.s_fence->finished);
+
+		ret = rocket_acquire_object_fences(job->in_bos, job->in_bo_count, &job->base, false);
+		if (ret)
+			goto err_unlock;
+
+		ret = rocket_acquire_object_fences(job->out_bos, job->out_bo_count, &job->base, true);
+		if (ret)
+			goto err_unlock;
+
+		kref_get(&job->refcount); /* put by scheduler job completion */
+
+		drm_sched_entity_push_job(&job->base);
+	}
+
+	rocket_attach_object_fences(job->out_bos, job->out_bo_count, job->inference_done_fence);
+
+err_unlock:
+	drm_gem_unlock_reservations(bos, job->in_bo_count + job->out_bo_count, &acquire_ctx);
+err:
+	kvfree(bos);
+
+	return ret;
+}
+
+static void rocket_job_cleanup(struct kref *ref)
+{
+	struct rocket_job *job = container_of(ref, struct rocket_job,
+						refcount);
+	unsigned int i;
+
+	rocket_iommu_domain_put(job->domain);
+
+	dma_fence_put(job->done_fence);
+	dma_fence_put(job->inference_done_fence);
+
+	if (job->in_bos) {
+		for (i = 0; i < job->in_bo_count; i++)
+			drm_gem_object_put(job->in_bos[i]);
+
+		kvfree(job->in_bos);
+	}
+
+	if (job->out_bos) {
+		for (i = 0; i < job->out_bo_count; i++)
+			drm_gem_object_put(job->out_bos[i]);
+
+		kvfree(job->out_bos);
+	}
+
+	kvfree(job->tasks);
+
+	kfree(job);
+}
+
+static void rocket_job_put(struct rocket_job *job)
+{
+	kref_put(&job->refcount, rocket_job_cleanup);
+}
+
+static void rocket_job_free(struct drm_sched_job *sched_job)
+{
+	struct rocket_job *job = to_rocket_job(sched_job);
+
+	drm_sched_job_cleanup(sched_job);
+
+	rocket_job_put(job);
+}
+
+static struct rocket_core *sched_to_core(struct rocket_device *rdev,
+					 struct drm_gpu_scheduler *sched)
+{
+	unsigned int core;
+
+	for (core = 0; core < rdev->num_cores; core++) {
+		if (&rdev->cores[core].sched == sched)
+			return &rdev->cores[core];
+	}
+
+	return NULL;
+}
+
+static struct dma_fence *rocket_job_run(struct drm_sched_job *sched_job)
+{
+	struct rocket_job *job = to_rocket_job(sched_job);
+	struct rocket_device *rdev = job->rdev;
+	struct rocket_core *core = sched_to_core(rdev, sched_job->sched);
+	struct dma_fence *fence = NULL;
+	int ret;
+
+	if (unlikely(job->base.s_fence->finished.error))
+		return NULL;
+
+	/*
+	 * Nothing to execute: can happen if the job has finished while
+	 * we were resetting the NPU.
+	 */
+	if (job->next_task_idx == job->task_count)
+		return NULL;
+
+	fence = rocket_fence_create(core);
+	if (IS_ERR(fence))
+		return fence;
+
+	if (job->done_fence)
+		dma_fence_put(job->done_fence);
+	job->done_fence = dma_fence_get(fence);
+
+	ret = pm_runtime_get_sync(core->dev);
+	if (ret < 0)
+		return fence;
+
+	ret = iommu_attach_group(job->domain->domain, core->iommu_group);
+	if (ret < 0)
+		return fence;
+
+	scoped_guard(mutex, &core->job_lock) {
+		core->in_flight_job = job;
+		rocket_job_hw_submit(core, job);
+	}
+
+	return fence;
+}
+
+static void rocket_job_handle_irq(struct rocket_core *core)
+{
+	pm_runtime_mark_last_busy(core->dev);
+
+	rocket_pc_writel(core, OPERATION_ENABLE, 0x0);
+	rocket_pc_writel(core, INTERRUPT_CLEAR, 0x1ffff);
+
+	scoped_guard(mutex, &core->job_lock)
+		if (core->in_flight_job) {
+			if (core->in_flight_job->next_task_idx < core->in_flight_job->task_count) {
+				rocket_job_hw_submit(core, core->in_flight_job);
+				return;
+			}
+
+			iommu_detach_group(NULL, iommu_group_get(core->dev));
+			dma_fence_signal(core->in_flight_job->done_fence);
+			pm_runtime_put_autosuspend(core->dev);
+			core->in_flight_job = NULL;
+		}
+}
+
+static void
+rocket_reset(struct rocket_core *core, struct drm_sched_job *bad)
+{
+	if (!atomic_read(&core->reset.pending))
+		return;
+
+	drm_sched_stop(&core->sched, bad);
+
+	/*
+	 * Remaining interrupts have been handled, but we might still have
+	 * stuck jobs. Let's make sure the PM counters stay balanced by
+	 * manually calling pm_runtime_put_noidle().
+	 */
+	scoped_guard(mutex, &core->job_lock) {
+		if (core->in_flight_job)
+			pm_runtime_put_noidle(core->dev);
+
+		iommu_detach_group(NULL, core->iommu_group);
+
+		core->in_flight_job = NULL;
+	}
+
+	/* Proceed with reset now. */
+	rocket_core_reset(core);
+
+	/* NPU has been reset, we can clear the reset pending bit. */
+	atomic_set(&core->reset.pending, 0);
+
+	/* Restart the scheduler */
+	drm_sched_start(&core->sched, 0);
+}
+
+static enum drm_gpu_sched_stat rocket_job_timedout(struct drm_sched_job *sched_job)
+{
+	struct rocket_job *job = to_rocket_job(sched_job);
+	struct rocket_device *rdev = job->rdev;
+	struct rocket_core *core = sched_to_core(rdev, sched_job->sched);
+
+	dev_err(core->dev, "NPU job timed out");
+
+	atomic_set(&core->reset.pending, 1);
+	rocket_reset(core, sched_job);
+
+	return DRM_GPU_SCHED_STAT_RESET;
+}
+
+static void rocket_reset_work(struct work_struct *work)
+{
+	struct rocket_core *core;
+
+	core = container_of(work, struct rocket_core, reset.work);
+	rocket_reset(core, NULL);
+}
+
+static const struct drm_sched_backend_ops rocket_sched_ops = {
+	.run_job = rocket_job_run,
+	.timedout_job = rocket_job_timedout,
+	.free_job = rocket_job_free
+};
+
+static irqreturn_t rocket_job_irq_handler_thread(int irq, void *data)
+{
+	struct rocket_core *core = data;
+
+	rocket_job_handle_irq(core);
+
+	return IRQ_HANDLED;
+}
+
+static irqreturn_t rocket_job_irq_handler(int irq, void *data)
+{
+	struct rocket_core *core = data;
+	u32 raw_status = rocket_pc_readl(core, INTERRUPT_RAW_STATUS);
+
+	WARN_ON(raw_status & PC_INTERRUPT_RAW_STATUS_DMA_READ_ERROR);
+	WARN_ON(raw_status & PC_INTERRUPT_RAW_STATUS_DMA_WRITE_ERROR);
+
+	if (!(raw_status & PC_INTERRUPT_RAW_STATUS_DPU_0 ||
+	      raw_status & PC_INTERRUPT_RAW_STATUS_DPU_1))
+		return IRQ_NONE;
+
+	rocket_pc_writel(core, INTERRUPT_MASK, 0x0);
+
+	return IRQ_WAKE_THREAD;
+}
+
+int rocket_job_init(struct rocket_core *core)
+{
+	struct drm_sched_init_args args = {
+		.ops = &rocket_sched_ops,
+		.num_rqs = DRM_SCHED_PRIORITY_COUNT,
+		.credit_limit = 1,
+		.timeout = msecs_to_jiffies(JOB_TIMEOUT_MS),
+		.name = dev_name(core->dev),
+		.dev = core->dev,
+	};
+	int ret;
+
+	INIT_WORK(&core->reset.work, rocket_reset_work);
+	spin_lock_init(&core->fence_lock);
+	mutex_init(&core->job_lock);
+
+	core->irq = platform_get_irq(to_platform_device(core->dev), 0);
+	if (core->irq < 0)
+		return core->irq;
+
+	ret = devm_request_threaded_irq(core->dev, core->irq,
+					rocket_job_irq_handler,
+					rocket_job_irq_handler_thread,
+					IRQF_SHARED, dev_name(core->dev),
+					core);
+	if (ret) {
+		dev_err(core->dev, "failed to request job irq");
+		return ret;
+	}
+
+	core->reset.wq = alloc_ordered_workqueue("rocket-reset-%d", 0, core->index);
+	if (!core->reset.wq)
+		return -ENOMEM;
+
+	core->fence_context = dma_fence_context_alloc(1);
+
+	args.timeout_wq = core->reset.wq;
+	ret = drm_sched_init(&core->sched, &args);
+	if (ret) {
+		dev_err(core->dev, "Failed to create scheduler: %d.", ret);
+		goto err_sched;
+	}
+
+	return 0;
+
+err_sched:
+	drm_sched_fini(&core->sched);
+
+	destroy_workqueue(core->reset.wq);
+	return ret;
+}
+
+void rocket_job_fini(struct rocket_core *core)
+{
+	drm_sched_fini(&core->sched);
+
+	cancel_work_sync(&core->reset.work);
+	destroy_workqueue(core->reset.wq);
+}
+
+int rocket_job_open(struct rocket_file_priv *rocket_priv)
+{
+	struct rocket_device *rdev = rocket_priv->rdev;
+	struct drm_gpu_scheduler **scheds = kmalloc_array(rdev->num_cores,
+							  sizeof(*scheds),
+							  GFP_KERNEL);
+	unsigned int core;
+	int ret;
+
+	for (core = 0; core < rdev->num_cores; core++)
+		scheds[core] = &rdev->cores[core].sched;
+
+	ret = drm_sched_entity_init(&rocket_priv->sched_entity,
+				    DRM_SCHED_PRIORITY_NORMAL,
+				    scheds,
+				    rdev->num_cores, NULL);
+	if (WARN_ON(ret))
+		return ret;
+
+	return 0;
+}
+
+void rocket_job_close(struct rocket_file_priv *rocket_priv)
+{
+	struct drm_sched_entity *entity = &rocket_priv->sched_entity;
+
+	kfree(entity->sched_list);
+	drm_sched_entity_destroy(entity);
+}
+
+int rocket_job_is_idle(struct rocket_core *core)
+{
+	/* If there are any jobs in this HW queue, we're not idle */
+	if (atomic_read(&core->sched.credit_count))
+		return false;
+
+	return true;
+}
+
+static int rocket_ioctl_submit_job(struct drm_device *dev, struct drm_file *file,
+				   struct drm_rocket_job *job)
+{
+	struct rocket_device *rdev = to_rocket_device(dev);
+	struct rocket_file_priv *file_priv = file->driver_priv;
+	struct rocket_job *rjob = NULL;
+	int ret = 0;
+
+	if (job->task_count == 0)
+		return -EINVAL;
+
+	rjob = kzalloc(sizeof(*rjob), GFP_KERNEL);
+	if (!rjob)
+		return -ENOMEM;
+
+	kref_init(&rjob->refcount);
+
+	rjob->rdev = rdev;
+
+	ret = drm_sched_job_init(&rjob->base,
+				 &file_priv->sched_entity,
+				 1, NULL, file->client_id);
+	if (ret)
+		goto out_put_job;
+
+	ret = rocket_copy_tasks(dev, file, job, rjob);
+	if (ret)
+		goto out_cleanup_job;
+
+	ret = drm_gem_objects_lookup(file, u64_to_user_ptr(job->in_bo_handles),
+				     job->in_bo_handle_count, &rjob->in_bos);
+	if (ret)
+		goto out_cleanup_job;
+
+	rjob->in_bo_count = job->in_bo_handle_count;
+
+	ret = drm_gem_objects_lookup(file, u64_to_user_ptr(job->out_bo_handles),
+				     job->out_bo_handle_count, &rjob->out_bos);
+	if (ret)
+		goto out_cleanup_job;
+
+	rjob->out_bo_count = job->out_bo_handle_count;
+
+	rjob->domain = rocket_iommu_domain_get(file_priv);
+
+	ret = rocket_job_push(rjob);
+	if (ret)
+		goto out_cleanup_job;
+
+out_cleanup_job:
+	if (ret)
+		drm_sched_job_cleanup(&rjob->base);
+out_put_job:
+	rocket_job_put(rjob);
+
+	return ret;
+}
+
+int rocket_ioctl_submit(struct drm_device *dev, void *data, struct drm_file *file)
+{
+	struct drm_rocket_submit *args = data;
+	struct drm_rocket_job *jobs;
+	int ret = 0;
+	unsigned int i = 0;
+
+	if (args->job_count == 0)
+		return 0;
+
+	if (args->job_struct_size < sizeof(struct drm_rocket_job)) {
+		drm_dbg(dev, "job_struct_size field in drm_rocket_submit struct is too small.\n");
+		return -EINVAL;
+	}
+
+	if (args->reserved != 0) {
+		drm_dbg(dev, "Reserved field in drm_rocket_submit struct should be 0.\n");
+		return -EINVAL;
+	}
+
+	jobs = kvmalloc_array(args->job_count, sizeof(*jobs), GFP_KERNEL);
+	if (!jobs) {
+		drm_dbg(dev, "Failed to allocate incoming job array\n");
+		return -ENOMEM;
+	}
+
+	for (i = 0; i < args->job_count; i++) {
+		if (copy_from_user(&jobs[i],
+				   u64_to_user_ptr(args->jobs) + i * args->job_struct_size,
+				   sizeof(*jobs))) {
+			ret = -EFAULT;
+			drm_dbg(dev, "Failed to copy incoming job array\n");
+			goto exit;
+		}
+	}
+
+
+	for (i = 0; i < args->job_count; i++)
+		rocket_ioctl_submit_job(dev, file, &jobs[i]);
+
+exit:
+	kvfree(jobs);
+
+	return ret;
+}
diff --git a/drivers/accel/rocket/rocket_job.h b/drivers/accel/rocket/rocket_job.h
new file mode 100644
index 000000000000..4ae00feec3b9
--- /dev/null
+++ b/drivers/accel/rocket/rocket_job.h
@@ -0,0 +1,52 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright 2024-2025 Tomeu Vizoso <tomeu@tomeuvizoso.net> */
+
+#ifndef __ROCKET_JOB_H__
+#define __ROCKET_JOB_H__
+
+#include <drm/drm_drv.h>
+#include <drm/gpu_scheduler.h>
+
+#include "rocket_core.h"
+#include "rocket_drv.h"
+
+struct rocket_task {
+	u64 regcmd;
+	u32 regcmd_count;
+};
+
+struct rocket_job {
+	struct drm_sched_job base;
+
+	struct rocket_device *rdev;
+
+	struct drm_gem_object **in_bos;
+	struct drm_gem_object **out_bos;
+
+	u32 in_bo_count;
+	u32 out_bo_count;
+
+	struct rocket_task *tasks;
+	u32 task_count;
+	u32 next_task_idx;
+
+	/* Fence to be signaled by drm-sched once its done with the job */
+	struct dma_fence *inference_done_fence;
+
+	/* Fence to be signaled by IRQ handler when the job is complete. */
+	struct dma_fence *done_fence;
+
+	struct rocket_iommu_domain *domain;
+
+	struct kref refcount;
+};
+
+int rocket_ioctl_submit(struct drm_device *dev, void *data, struct drm_file *file);
+
+int rocket_job_init(struct rocket_core *core);
+void rocket_job_fini(struct rocket_core *core);
+int rocket_job_open(struct rocket_file_priv *rocket_priv);
+void rocket_job_close(struct rocket_file_priv *rocket_priv);
+int rocket_job_is_idle(struct rocket_core *core);
+
+#endif
diff --git a/drivers/accel/rocket/rocket_registers.h b/drivers/accel/rocket/rocket_registers.h
new file mode 100644
index 000000000000..9aef614c3470
--- /dev/null
+++ b/drivers/accel/rocket/rocket_registers.h
@@ -0,0 +1,4404 @@
+/* SPDX-License-Identifier: GPL-2.0-only OR MIT */
+
+#ifndef __ROCKET_REGISTERS_XML__
+#define __ROCKET_REGISTERS_XML__
+
+/* Autogenerated file, DO NOT EDIT manually!
+
+This file was generated by the rules-ng-ng gen_header.py tool in this git repository:
+http://gitlab.freedesktop.org/mesa/mesa/
+git clone https://gitlab.freedesktop.org/mesa/mesa.git
+
+The rules-ng-ng source files this header was generated from are:
+
+- /home/tomeu/src/mesa/src/gallium/drivers/rocket/registers.xml (  60076 bytes, from Wed Jun 12 10:02:25 2024)
+
+Copyright (C) 2024-2025 by the following authors:
+- Tomeu Vizoso <tomeu@tomeuvizoso.net>
+*/
+
+#define REG_PC_VERSION						0x00000000
+#define PC_VERSION_VERSION__MASK				0xffffffff
+#define PC_VERSION_VERSION__SHIFT				0
+static inline uint32_t PC_VERSION_VERSION(uint32_t val)
+{
+	return ((val) << PC_VERSION_VERSION__SHIFT) & PC_VERSION_VERSION__MASK;
+}
+
+#define REG_PC_VERSION_NUM					0x00000004
+#define PC_VERSION_NUM_VERSION_NUM__MASK			0xffffffff
+#define PC_VERSION_NUM_VERSION_NUM__SHIFT			0
+static inline uint32_t PC_VERSION_NUM_VERSION_NUM(uint32_t val)
+{
+	return ((val) << PC_VERSION_NUM_VERSION_NUM__SHIFT) & PC_VERSION_NUM_VERSION_NUM__MASK;
+}
+
+#define REG_PC_OPERATION_ENABLE					0x00000008
+#define PC_OPERATION_ENABLE_RESERVED_0__MASK			0xfffffffe
+#define PC_OPERATION_ENABLE_RESERVED_0__SHIFT			1
+static inline uint32_t PC_OPERATION_ENABLE_RESERVED_0(uint32_t val)
+{
+	return ((val) << PC_OPERATION_ENABLE_RESERVED_0__SHIFT) & PC_OPERATION_ENABLE_RESERVED_0__MASK;
+}
+#define PC_OPERATION_ENABLE_OP_EN__MASK				0x00000001
+#define PC_OPERATION_ENABLE_OP_EN__SHIFT			0
+static inline uint32_t PC_OPERATION_ENABLE_OP_EN(uint32_t val)
+{
+	return ((val) << PC_OPERATION_ENABLE_OP_EN__SHIFT) & PC_OPERATION_ENABLE_OP_EN__MASK;
+}
+
+#define REG_PC_BASE_ADDRESS					0x00000010
+#define PC_BASE_ADDRESS_PC_SOURCE_ADDR__MASK			0xfffffff0
+#define PC_BASE_ADDRESS_PC_SOURCE_ADDR__SHIFT			4
+static inline uint32_t PC_BASE_ADDRESS_PC_SOURCE_ADDR(uint32_t val)
+{
+	return ((val) << PC_BASE_ADDRESS_PC_SOURCE_ADDR__SHIFT) & PC_BASE_ADDRESS_PC_SOURCE_ADDR__MASK;
+}
+#define PC_BASE_ADDRESS_RESERVED_0__MASK			0x0000000e
+#define PC_BASE_ADDRESS_RESERVED_0__SHIFT			1
+static inline uint32_t PC_BASE_ADDRESS_RESERVED_0(uint32_t val)
+{
+	return ((val) << PC_BASE_ADDRESS_RESERVED_0__SHIFT) & PC_BASE_ADDRESS_RESERVED_0__MASK;
+}
+#define PC_BASE_ADDRESS_PC_SEL__MASK				0x00000001
+#define PC_BASE_ADDRESS_PC_SEL__SHIFT				0
+static inline uint32_t PC_BASE_ADDRESS_PC_SEL(uint32_t val)
+{
+	return ((val) << PC_BASE_ADDRESS_PC_SEL__SHIFT) & PC_BASE_ADDRESS_PC_SEL__MASK;
+}
+
+#define REG_PC_REGISTER_AMOUNTS					0x00000014
+#define PC_REGISTER_AMOUNTS_RESERVED_0__MASK			0xffff0000
+#define PC_REGISTER_AMOUNTS_RESERVED_0__SHIFT			16
+static inline uint32_t PC_REGISTER_AMOUNTS_RESERVED_0(uint32_t val)
+{
+	return ((val) << PC_REGISTER_AMOUNTS_RESERVED_0__SHIFT) & PC_REGISTER_AMOUNTS_RESERVED_0__MASK;
+}
+#define PC_REGISTER_AMOUNTS_PC_DATA_AMOUNT__MASK		0x0000ffff
+#define PC_REGISTER_AMOUNTS_PC_DATA_AMOUNT__SHIFT		0
+static inline uint32_t PC_REGISTER_AMOUNTS_PC_DATA_AMOUNT(uint32_t val)
+{
+	return ((val) << PC_REGISTER_AMOUNTS_PC_DATA_AMOUNT__SHIFT) & PC_REGISTER_AMOUNTS_PC_DATA_AMOUNT__MASK;
+}
+
+#define REG_PC_INTERRUPT_MASK					0x00000020
+#define PC_INTERRUPT_MASK_RESERVED_0__MASK			0xffffc000
+#define PC_INTERRUPT_MASK_RESERVED_0__SHIFT			14
+static inline uint32_t PC_INTERRUPT_MASK_RESERVED_0(uint32_t val)
+{
+	return ((val) << PC_INTERRUPT_MASK_RESERVED_0__SHIFT) & PC_INTERRUPT_MASK_RESERVED_0__MASK;
+}
+#define PC_INTERRUPT_MASK_DMA_WRITE_ERROR			0x00002000
+#define PC_INTERRUPT_MASK_DMA_READ_ERROR			0x00001000
+#define PC_INTERRUPT_MASK_PPU_1					0x00000800
+#define PC_INTERRUPT_MASK_PPU_0					0x00000400
+#define PC_INTERRUPT_MASK_DPU_1					0x00000200
+#define PC_INTERRUPT_MASK_DPU_0					0x00000100
+#define PC_INTERRUPT_MASK_CORE_1				0x00000080
+#define PC_INTERRUPT_MASK_CORE_0				0x00000040
+#define PC_INTERRUPT_MASK_CNA_CSC_1				0x00000020
+#define PC_INTERRUPT_MASK_CNA_CSC_0				0x00000010
+#define PC_INTERRUPT_MASK_CNA_WEIGHT_1				0x00000008
+#define PC_INTERRUPT_MASK_CNA_WEIGHT_0				0x00000004
+#define PC_INTERRUPT_MASK_CNA_FEATURE_1				0x00000002
+#define PC_INTERRUPT_MASK_CNA_FEATURE_0				0x00000001
+
+#define REG_PC_INTERRUPT_CLEAR					0x00000024
+#define PC_INTERRUPT_CLEAR_RESERVED_0__MASK			0xffffc000
+#define PC_INTERRUPT_CLEAR_RESERVED_0__SHIFT			14
+static inline uint32_t PC_INTERRUPT_CLEAR_RESERVED_0(uint32_t val)
+{
+	return ((val) << PC_INTERRUPT_CLEAR_RESERVED_0__SHIFT) & PC_INTERRUPT_CLEAR_RESERVED_0__MASK;
+}
+#define PC_INTERRUPT_CLEAR_DMA_WRITE_ERROR			0x00002000
+#define PC_INTERRUPT_CLEAR_DMA_READ_ERROR			0x00001000
+#define PC_INTERRUPT_CLEAR_PPU_1				0x00000800
+#define PC_INTERRUPT_CLEAR_PPU_0				0x00000400
+#define PC_INTERRUPT_CLEAR_DPU_1				0x00000200
+#define PC_INTERRUPT_CLEAR_DPU_0				0x00000100
+#define PC_INTERRUPT_CLEAR_CORE_1				0x00000080
+#define PC_INTERRUPT_CLEAR_CORE_0				0x00000040
+#define PC_INTERRUPT_CLEAR_CNA_CSC_1				0x00000020
+#define PC_INTERRUPT_CLEAR_CNA_CSC_0				0x00000010
+#define PC_INTERRUPT_CLEAR_CNA_WEIGHT_1				0x00000008
+#define PC_INTERRUPT_CLEAR_CNA_WEIGHT_0				0x00000004
+#define PC_INTERRUPT_CLEAR_CNA_FEATURE_1			0x00000002
+#define PC_INTERRUPT_CLEAR_CNA_FEATURE_0			0x00000001
+
+#define REG_PC_INTERRUPT_STATUS					0x00000028
+#define PC_INTERRUPT_STATUS_RESERVED_0__MASK			0xffffc000
+#define PC_INTERRUPT_STATUS_RESERVED_0__SHIFT			14
+static inline uint32_t PC_INTERRUPT_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << PC_INTERRUPT_STATUS_RESERVED_0__SHIFT) & PC_INTERRUPT_STATUS_RESERVED_0__MASK;
+}
+#define PC_INTERRUPT_STATUS_DMA_WRITE_ERROR			0x00002000
+#define PC_INTERRUPT_STATUS_DMA_READ_ERROR			0x00001000
+#define PC_INTERRUPT_STATUS_PPU_1				0x00000800
+#define PC_INTERRUPT_STATUS_PPU_0				0x00000400
+#define PC_INTERRUPT_STATUS_DPU_1				0x00000200
+#define PC_INTERRUPT_STATUS_DPU_0				0x00000100
+#define PC_INTERRUPT_STATUS_CORE_1				0x00000080
+#define PC_INTERRUPT_STATUS_CORE_0				0x00000040
+#define PC_INTERRUPT_STATUS_CNA_CSC_1				0x00000020
+#define PC_INTERRUPT_STATUS_CNA_CSC_0				0x00000010
+#define PC_INTERRUPT_STATUS_CNA_WEIGHT_1			0x00000008
+#define PC_INTERRUPT_STATUS_CNA_WEIGHT_0			0x00000004
+#define PC_INTERRUPT_STATUS_CNA_FEATURE_1			0x00000002
+#define PC_INTERRUPT_STATUS_CNA_FEATURE_0			0x00000001
+
+#define REG_PC_INTERRUPT_RAW_STATUS				0x0000002c
+#define PC_INTERRUPT_RAW_STATUS_RESERVED_0__MASK		0xffffc000
+#define PC_INTERRUPT_RAW_STATUS_RESERVED_0__SHIFT		14
+static inline uint32_t PC_INTERRUPT_RAW_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << PC_INTERRUPT_RAW_STATUS_RESERVED_0__SHIFT) & PC_INTERRUPT_RAW_STATUS_RESERVED_0__MASK;
+}
+#define PC_INTERRUPT_RAW_STATUS_DMA_WRITE_ERROR			0x00002000
+#define PC_INTERRUPT_RAW_STATUS_DMA_READ_ERROR			0x00001000
+#define PC_INTERRUPT_RAW_STATUS_PPU_1				0x00000800
+#define PC_INTERRUPT_RAW_STATUS_PPU_0				0x00000400
+#define PC_INTERRUPT_RAW_STATUS_DPU_1				0x00000200
+#define PC_INTERRUPT_RAW_STATUS_DPU_0				0x00000100
+#define PC_INTERRUPT_RAW_STATUS_CORE_1				0x00000080
+#define PC_INTERRUPT_RAW_STATUS_CORE_0				0x00000040
+#define PC_INTERRUPT_RAW_STATUS_CNA_CSC_1			0x00000020
+#define PC_INTERRUPT_RAW_STATUS_CNA_CSC_0			0x00000010
+#define PC_INTERRUPT_RAW_STATUS_CNA_WEIGHT_1			0x00000008
+#define PC_INTERRUPT_RAW_STATUS_CNA_WEIGHT_0			0x00000004
+#define PC_INTERRUPT_RAW_STATUS_CNA_FEATURE_1			0x00000002
+#define PC_INTERRUPT_RAW_STATUS_CNA_FEATURE_0			0x00000001
+
+#define REG_PC_TASK_CON						0x00000030
+#define PC_TASK_CON_RESERVED_0__MASK				0xffffc000
+#define PC_TASK_CON_RESERVED_0__SHIFT				14
+static inline uint32_t PC_TASK_CON_RESERVED_0(uint32_t val)
+{
+	return ((val) << PC_TASK_CON_RESERVED_0__SHIFT) & PC_TASK_CON_RESERVED_0__MASK;
+}
+#define PC_TASK_CON_TASK_COUNT_CLEAR__MASK			0x00002000
+#define PC_TASK_CON_TASK_COUNT_CLEAR__SHIFT			13
+static inline uint32_t PC_TASK_CON_TASK_COUNT_CLEAR(uint32_t val)
+{
+	return ((val) << PC_TASK_CON_TASK_COUNT_CLEAR__SHIFT) & PC_TASK_CON_TASK_COUNT_CLEAR__MASK;
+}
+#define PC_TASK_CON_TASK_PP_EN__MASK				0x00001000
+#define PC_TASK_CON_TASK_PP_EN__SHIFT				12
+static inline uint32_t PC_TASK_CON_TASK_PP_EN(uint32_t val)
+{
+	return ((val) << PC_TASK_CON_TASK_PP_EN__SHIFT) & PC_TASK_CON_TASK_PP_EN__MASK;
+}
+#define PC_TASK_CON_TASK_NUMBER__MASK				0x00000fff
+#define PC_TASK_CON_TASK_NUMBER__SHIFT				0
+static inline uint32_t PC_TASK_CON_TASK_NUMBER(uint32_t val)
+{
+	return ((val) << PC_TASK_CON_TASK_NUMBER__SHIFT) & PC_TASK_CON_TASK_NUMBER__MASK;
+}
+
+#define REG_PC_TASK_DMA_BASE_ADDR				0x00000034
+#define PC_TASK_DMA_BASE_ADDR_DMA_BASE_ADDR__MASK		0xfffffff0
+#define PC_TASK_DMA_BASE_ADDR_DMA_BASE_ADDR__SHIFT		4
+static inline uint32_t PC_TASK_DMA_BASE_ADDR_DMA_BASE_ADDR(uint32_t val)
+{
+	return ((val) << PC_TASK_DMA_BASE_ADDR_DMA_BASE_ADDR__SHIFT) & PC_TASK_DMA_BASE_ADDR_DMA_BASE_ADDR__MASK;
+}
+#define PC_TASK_DMA_BASE_ADDR_RESERVED_0__MASK			0x0000000f
+#define PC_TASK_DMA_BASE_ADDR_RESERVED_0__SHIFT			0
+static inline uint32_t PC_TASK_DMA_BASE_ADDR_RESERVED_0(uint32_t val)
+{
+	return ((val) << PC_TASK_DMA_BASE_ADDR_RESERVED_0__SHIFT) & PC_TASK_DMA_BASE_ADDR_RESERVED_0__MASK;
+}
+
+#define REG_PC_TASK_STATUS					0x0000003c
+#define PC_TASK_STATUS_RESERVED_0__MASK				0xf0000000
+#define PC_TASK_STATUS_RESERVED_0__SHIFT			28
+static inline uint32_t PC_TASK_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << PC_TASK_STATUS_RESERVED_0__SHIFT) & PC_TASK_STATUS_RESERVED_0__MASK;
+}
+#define PC_TASK_STATUS_TASK_STATUS__MASK			0x0fffffff
+#define PC_TASK_STATUS_TASK_STATUS__SHIFT			0
+static inline uint32_t PC_TASK_STATUS_TASK_STATUS(uint32_t val)
+{
+	return ((val) << PC_TASK_STATUS_TASK_STATUS__SHIFT) & PC_TASK_STATUS_TASK_STATUS__MASK;
+}
+
+#define REG_CNA_S_STATUS					0x00001000
+#define CNA_S_STATUS_RESERVED_0__MASK				0xfffc0000
+#define CNA_S_STATUS_RESERVED_0__SHIFT				18
+static inline uint32_t CNA_S_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_S_STATUS_RESERVED_0__SHIFT) & CNA_S_STATUS_RESERVED_0__MASK;
+}
+#define CNA_S_STATUS_STATUS_1__MASK				0x00030000
+#define CNA_S_STATUS_STATUS_1__SHIFT				16
+static inline uint32_t CNA_S_STATUS_STATUS_1(uint32_t val)
+{
+	return ((val) << CNA_S_STATUS_STATUS_1__SHIFT) & CNA_S_STATUS_STATUS_1__MASK;
+}
+#define CNA_S_STATUS_RESERVED_1__MASK				0x0000fffc
+#define CNA_S_STATUS_RESERVED_1__SHIFT				2
+static inline uint32_t CNA_S_STATUS_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_S_STATUS_RESERVED_1__SHIFT) & CNA_S_STATUS_RESERVED_1__MASK;
+}
+#define CNA_S_STATUS_STATUS_0__MASK				0x00000003
+#define CNA_S_STATUS_STATUS_0__SHIFT				0
+static inline uint32_t CNA_S_STATUS_STATUS_0(uint32_t val)
+{
+	return ((val) << CNA_S_STATUS_STATUS_0__SHIFT) & CNA_S_STATUS_STATUS_0__MASK;
+}
+
+#define REG_CNA_S_POINTER					0x00001004
+#define CNA_S_POINTER_RESERVED_0__MASK				0xfffe0000
+#define CNA_S_POINTER_RESERVED_0__SHIFT				17
+static inline uint32_t CNA_S_POINTER_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_S_POINTER_RESERVED_0__SHIFT) & CNA_S_POINTER_RESERVED_0__MASK;
+}
+#define CNA_S_POINTER_EXECUTER__MASK				0x00010000
+#define CNA_S_POINTER_EXECUTER__SHIFT				16
+static inline uint32_t CNA_S_POINTER_EXECUTER(uint32_t val)
+{
+	return ((val) << CNA_S_POINTER_EXECUTER__SHIFT) & CNA_S_POINTER_EXECUTER__MASK;
+}
+#define CNA_S_POINTER_RESERVED_1__MASK				0x0000ffc0
+#define CNA_S_POINTER_RESERVED_1__SHIFT				6
+static inline uint32_t CNA_S_POINTER_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_S_POINTER_RESERVED_1__SHIFT) & CNA_S_POINTER_RESERVED_1__MASK;
+}
+#define CNA_S_POINTER_EXECUTER_PP_CLEAR__MASK			0x00000020
+#define CNA_S_POINTER_EXECUTER_PP_CLEAR__SHIFT			5
+static inline uint32_t CNA_S_POINTER_EXECUTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << CNA_S_POINTER_EXECUTER_PP_CLEAR__SHIFT) & CNA_S_POINTER_EXECUTER_PP_CLEAR__MASK;
+}
+#define CNA_S_POINTER_POINTER_PP_CLEAR__MASK			0x00000010
+#define CNA_S_POINTER_POINTER_PP_CLEAR__SHIFT			4
+static inline uint32_t CNA_S_POINTER_POINTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << CNA_S_POINTER_POINTER_PP_CLEAR__SHIFT) & CNA_S_POINTER_POINTER_PP_CLEAR__MASK;
+}
+#define CNA_S_POINTER_POINTER_PP_MODE__MASK			0x00000008
+#define CNA_S_POINTER_POINTER_PP_MODE__SHIFT			3
+static inline uint32_t CNA_S_POINTER_POINTER_PP_MODE(uint32_t val)
+{
+	return ((val) << CNA_S_POINTER_POINTER_PP_MODE__SHIFT) & CNA_S_POINTER_POINTER_PP_MODE__MASK;
+}
+#define CNA_S_POINTER_EXECUTER_PP_EN__MASK			0x00000004
+#define CNA_S_POINTER_EXECUTER_PP_EN__SHIFT			2
+static inline uint32_t CNA_S_POINTER_EXECUTER_PP_EN(uint32_t val)
+{
+	return ((val) << CNA_S_POINTER_EXECUTER_PP_EN__SHIFT) & CNA_S_POINTER_EXECUTER_PP_EN__MASK;
+}
+#define CNA_S_POINTER_POINTER_PP_EN__MASK			0x00000002
+#define CNA_S_POINTER_POINTER_PP_EN__SHIFT			1
+static inline uint32_t CNA_S_POINTER_POINTER_PP_EN(uint32_t val)
+{
+	return ((val) << CNA_S_POINTER_POINTER_PP_EN__SHIFT) & CNA_S_POINTER_POINTER_PP_EN__MASK;
+}
+#define CNA_S_POINTER_POINTER__MASK				0x00000001
+#define CNA_S_POINTER_POINTER__SHIFT				0
+static inline uint32_t CNA_S_POINTER_POINTER(uint32_t val)
+{
+	return ((val) << CNA_S_POINTER_POINTER__SHIFT) & CNA_S_POINTER_POINTER__MASK;
+}
+
+#define REG_CNA_OPERATION_ENABLE				0x00001008
+#define CNA_OPERATION_ENABLE_RESERVED_0__MASK			0xfffffffe
+#define CNA_OPERATION_ENABLE_RESERVED_0__SHIFT			1
+static inline uint32_t CNA_OPERATION_ENABLE_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_OPERATION_ENABLE_RESERVED_0__SHIFT) & CNA_OPERATION_ENABLE_RESERVED_0__MASK;
+}
+#define CNA_OPERATION_ENABLE_OP_EN__MASK			0x00000001
+#define CNA_OPERATION_ENABLE_OP_EN__SHIFT			0
+static inline uint32_t CNA_OPERATION_ENABLE_OP_EN(uint32_t val)
+{
+	return ((val) << CNA_OPERATION_ENABLE_OP_EN__SHIFT) & CNA_OPERATION_ENABLE_OP_EN__MASK;
+}
+
+#define REG_CNA_CONV_CON1					0x0000100c
+#define CNA_CONV_CON1_RESERVED_0__MASK				0x80000000
+#define CNA_CONV_CON1_RESERVED_0__SHIFT				31
+static inline uint32_t CNA_CONV_CON1_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON1_RESERVED_0__SHIFT) & CNA_CONV_CON1_RESERVED_0__MASK;
+}
+#define CNA_CONV_CON1_NONALIGN_DMA__MASK			0x40000000
+#define CNA_CONV_CON1_NONALIGN_DMA__SHIFT			30
+static inline uint32_t CNA_CONV_CON1_NONALIGN_DMA(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON1_NONALIGN_DMA__SHIFT) & CNA_CONV_CON1_NONALIGN_DMA__MASK;
+}
+#define CNA_CONV_CON1_GROUP_LINE_OFF__MASK			0x20000000
+#define CNA_CONV_CON1_GROUP_LINE_OFF__SHIFT			29
+static inline uint32_t CNA_CONV_CON1_GROUP_LINE_OFF(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON1_GROUP_LINE_OFF__SHIFT) & CNA_CONV_CON1_GROUP_LINE_OFF__MASK;
+}
+#define CNA_CONV_CON1_RESERVED_1__MASK				0x1ffe0000
+#define CNA_CONV_CON1_RESERVED_1__SHIFT				17
+static inline uint32_t CNA_CONV_CON1_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON1_RESERVED_1__SHIFT) & CNA_CONV_CON1_RESERVED_1__MASK;
+}
+#define CNA_CONV_CON1_DECONV__MASK				0x00010000
+#define CNA_CONV_CON1_DECONV__SHIFT				16
+static inline uint32_t CNA_CONV_CON1_DECONV(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON1_DECONV__SHIFT) & CNA_CONV_CON1_DECONV__MASK;
+}
+#define CNA_CONV_CON1_ARGB_IN__MASK				0x0000f000
+#define CNA_CONV_CON1_ARGB_IN__SHIFT				12
+static inline uint32_t CNA_CONV_CON1_ARGB_IN(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON1_ARGB_IN__SHIFT) & CNA_CONV_CON1_ARGB_IN__MASK;
+}
+#define CNA_CONV_CON1_RESERVED_2__MASK				0x00000c00
+#define CNA_CONV_CON1_RESERVED_2__SHIFT				10
+static inline uint32_t CNA_CONV_CON1_RESERVED_2(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON1_RESERVED_2__SHIFT) & CNA_CONV_CON1_RESERVED_2__MASK;
+}
+#define CNA_CONV_CON1_PROC_PRECISION__MASK			0x00000380
+#define CNA_CONV_CON1_PROC_PRECISION__SHIFT			7
+static inline uint32_t CNA_CONV_CON1_PROC_PRECISION(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON1_PROC_PRECISION__SHIFT) & CNA_CONV_CON1_PROC_PRECISION__MASK;
+}
+#define CNA_CONV_CON1_IN_PRECISION__MASK			0x00000070
+#define CNA_CONV_CON1_IN_PRECISION__SHIFT			4
+static inline uint32_t CNA_CONV_CON1_IN_PRECISION(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON1_IN_PRECISION__SHIFT) & CNA_CONV_CON1_IN_PRECISION__MASK;
+}
+#define CNA_CONV_CON1_CONV_MODE__MASK				0x0000000f
+#define CNA_CONV_CON1_CONV_MODE__SHIFT				0
+static inline uint32_t CNA_CONV_CON1_CONV_MODE(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON1_CONV_MODE__SHIFT) & CNA_CONV_CON1_CONV_MODE__MASK;
+}
+
+#define REG_CNA_CONV_CON2					0x00001010
+#define CNA_CONV_CON2_RESERVED_0__MASK				0xff000000
+#define CNA_CONV_CON2_RESERVED_0__SHIFT				24
+static inline uint32_t CNA_CONV_CON2_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON2_RESERVED_0__SHIFT) & CNA_CONV_CON2_RESERVED_0__MASK;
+}
+#define CNA_CONV_CON2_KERNEL_GROUP__MASK			0x00ff0000
+#define CNA_CONV_CON2_KERNEL_GROUP__SHIFT			16
+static inline uint32_t CNA_CONV_CON2_KERNEL_GROUP(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON2_KERNEL_GROUP__SHIFT) & CNA_CONV_CON2_KERNEL_GROUP__MASK;
+}
+#define CNA_CONV_CON2_RESERVED_1__MASK				0x0000c000
+#define CNA_CONV_CON2_RESERVED_1__SHIFT				14
+static inline uint32_t CNA_CONV_CON2_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON2_RESERVED_1__SHIFT) & CNA_CONV_CON2_RESERVED_1__MASK;
+}
+#define CNA_CONV_CON2_FEATURE_GRAINS__MASK			0x00003ff0
+#define CNA_CONV_CON2_FEATURE_GRAINS__SHIFT			4
+static inline uint32_t CNA_CONV_CON2_FEATURE_GRAINS(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON2_FEATURE_GRAINS__SHIFT) & CNA_CONV_CON2_FEATURE_GRAINS__MASK;
+}
+#define CNA_CONV_CON2_RESERVED_2__MASK				0x00000008
+#define CNA_CONV_CON2_RESERVED_2__SHIFT				3
+static inline uint32_t CNA_CONV_CON2_RESERVED_2(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON2_RESERVED_2__SHIFT) & CNA_CONV_CON2_RESERVED_2__MASK;
+}
+#define CNA_CONV_CON2_CSC_WO_EN__MASK				0x00000004
+#define CNA_CONV_CON2_CSC_WO_EN__SHIFT				2
+static inline uint32_t CNA_CONV_CON2_CSC_WO_EN(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON2_CSC_WO_EN__SHIFT) & CNA_CONV_CON2_CSC_WO_EN__MASK;
+}
+#define CNA_CONV_CON2_CSC_DO_EN__MASK				0x00000002
+#define CNA_CONV_CON2_CSC_DO_EN__SHIFT				1
+static inline uint32_t CNA_CONV_CON2_CSC_DO_EN(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON2_CSC_DO_EN__SHIFT) & CNA_CONV_CON2_CSC_DO_EN__MASK;
+}
+#define CNA_CONV_CON2_CMD_FIFO_SRST__MASK			0x00000001
+#define CNA_CONV_CON2_CMD_FIFO_SRST__SHIFT			0
+static inline uint32_t CNA_CONV_CON2_CMD_FIFO_SRST(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON2_CMD_FIFO_SRST__SHIFT) & CNA_CONV_CON2_CMD_FIFO_SRST__MASK;
+}
+
+#define REG_CNA_CONV_CON3					0x00001014
+#define CNA_CONV_CON3_RESERVED_0__MASK				0x80000000
+#define CNA_CONV_CON3_RESERVED_0__SHIFT				31
+static inline uint32_t CNA_CONV_CON3_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_RESERVED_0__SHIFT) & CNA_CONV_CON3_RESERVED_0__MASK;
+}
+#define CNA_CONV_CON3_NN_MODE__MASK				0x70000000
+#define CNA_CONV_CON3_NN_MODE__SHIFT				28
+static inline uint32_t CNA_CONV_CON3_NN_MODE(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_NN_MODE__SHIFT) & CNA_CONV_CON3_NN_MODE__MASK;
+}
+#define CNA_CONV_CON3_RESERVED_1__MASK				0x0c000000
+#define CNA_CONV_CON3_RESERVED_1__SHIFT				26
+static inline uint32_t CNA_CONV_CON3_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_RESERVED_1__SHIFT) & CNA_CONV_CON3_RESERVED_1__MASK;
+}
+#define CNA_CONV_CON3_ATROUS_Y_DILATION__MASK			0x03e00000
+#define CNA_CONV_CON3_ATROUS_Y_DILATION__SHIFT			21
+static inline uint32_t CNA_CONV_CON3_ATROUS_Y_DILATION(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_ATROUS_Y_DILATION__SHIFT) & CNA_CONV_CON3_ATROUS_Y_DILATION__MASK;
+}
+#define CNA_CONV_CON3_ATROUS_X_DILATION__MASK			0x001f0000
+#define CNA_CONV_CON3_ATROUS_X_DILATION__SHIFT			16
+static inline uint32_t CNA_CONV_CON3_ATROUS_X_DILATION(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_ATROUS_X_DILATION__SHIFT) & CNA_CONV_CON3_ATROUS_X_DILATION__MASK;
+}
+#define CNA_CONV_CON3_RESERVED_2__MASK				0x0000c000
+#define CNA_CONV_CON3_RESERVED_2__SHIFT				14
+static inline uint32_t CNA_CONV_CON3_RESERVED_2(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_RESERVED_2__SHIFT) & CNA_CONV_CON3_RESERVED_2__MASK;
+}
+#define CNA_CONV_CON3_DECONV_Y_STRIDE__MASK			0x00003800
+#define CNA_CONV_CON3_DECONV_Y_STRIDE__SHIFT			11
+static inline uint32_t CNA_CONV_CON3_DECONV_Y_STRIDE(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_DECONV_Y_STRIDE__SHIFT) & CNA_CONV_CON3_DECONV_Y_STRIDE__MASK;
+}
+#define CNA_CONV_CON3_DECONV_X_STRIDE__MASK			0x00000700
+#define CNA_CONV_CON3_DECONV_X_STRIDE__SHIFT			8
+static inline uint32_t CNA_CONV_CON3_DECONV_X_STRIDE(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_DECONV_X_STRIDE__SHIFT) & CNA_CONV_CON3_DECONV_X_STRIDE__MASK;
+}
+#define CNA_CONV_CON3_RESERVED_3__MASK				0x000000c0
+#define CNA_CONV_CON3_RESERVED_3__SHIFT				6
+static inline uint32_t CNA_CONV_CON3_RESERVED_3(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_RESERVED_3__SHIFT) & CNA_CONV_CON3_RESERVED_3__MASK;
+}
+#define CNA_CONV_CON3_CONV_Y_STRIDE__MASK			0x00000038
+#define CNA_CONV_CON3_CONV_Y_STRIDE__SHIFT			3
+static inline uint32_t CNA_CONV_CON3_CONV_Y_STRIDE(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_CONV_Y_STRIDE__SHIFT) & CNA_CONV_CON3_CONV_Y_STRIDE__MASK;
+}
+#define CNA_CONV_CON3_CONV_X_STRIDE__MASK			0x00000007
+#define CNA_CONV_CON3_CONV_X_STRIDE__SHIFT			0
+static inline uint32_t CNA_CONV_CON3_CONV_X_STRIDE(uint32_t val)
+{
+	return ((val) << CNA_CONV_CON3_CONV_X_STRIDE__SHIFT) & CNA_CONV_CON3_CONV_X_STRIDE__MASK;
+}
+
+#define REG_CNA_DATA_SIZE0					0x00001020
+#define CNA_DATA_SIZE0_RESERVED_0__MASK				0xf8000000
+#define CNA_DATA_SIZE0_RESERVED_0__SHIFT			27
+static inline uint32_t CNA_DATA_SIZE0_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE0_RESERVED_0__SHIFT) & CNA_DATA_SIZE0_RESERVED_0__MASK;
+}
+#define CNA_DATA_SIZE0_DATAIN_WIDTH__MASK			0x07ff0000
+#define CNA_DATA_SIZE0_DATAIN_WIDTH__SHIFT			16
+static inline uint32_t CNA_DATA_SIZE0_DATAIN_WIDTH(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE0_DATAIN_WIDTH__SHIFT) & CNA_DATA_SIZE0_DATAIN_WIDTH__MASK;
+}
+#define CNA_DATA_SIZE0_RESERVED_1__MASK				0x0000f800
+#define CNA_DATA_SIZE0_RESERVED_1__SHIFT			11
+static inline uint32_t CNA_DATA_SIZE0_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE0_RESERVED_1__SHIFT) & CNA_DATA_SIZE0_RESERVED_1__MASK;
+}
+#define CNA_DATA_SIZE0_DATAIN_HEIGHT__MASK			0x000007ff
+#define CNA_DATA_SIZE0_DATAIN_HEIGHT__SHIFT			0
+static inline uint32_t CNA_DATA_SIZE0_DATAIN_HEIGHT(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE0_DATAIN_HEIGHT__SHIFT) & CNA_DATA_SIZE0_DATAIN_HEIGHT__MASK;
+}
+
+#define REG_CNA_DATA_SIZE1					0x00001024
+#define CNA_DATA_SIZE1_RESERVED_0__MASK				0xc0000000
+#define CNA_DATA_SIZE1_RESERVED_0__SHIFT			30
+static inline uint32_t CNA_DATA_SIZE1_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE1_RESERVED_0__SHIFT) & CNA_DATA_SIZE1_RESERVED_0__MASK;
+}
+#define CNA_DATA_SIZE1_DATAIN_CHANNEL_REAL__MASK		0x3fff0000
+#define CNA_DATA_SIZE1_DATAIN_CHANNEL_REAL__SHIFT		16
+static inline uint32_t CNA_DATA_SIZE1_DATAIN_CHANNEL_REAL(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE1_DATAIN_CHANNEL_REAL__SHIFT) & CNA_DATA_SIZE1_DATAIN_CHANNEL_REAL__MASK;
+}
+#define CNA_DATA_SIZE1_DATAIN_CHANNEL__MASK			0x0000ffff
+#define CNA_DATA_SIZE1_DATAIN_CHANNEL__SHIFT			0
+static inline uint32_t CNA_DATA_SIZE1_DATAIN_CHANNEL(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE1_DATAIN_CHANNEL__SHIFT) & CNA_DATA_SIZE1_DATAIN_CHANNEL__MASK;
+}
+
+#define REG_CNA_DATA_SIZE2					0x00001028
+#define CNA_DATA_SIZE2_RESERVED_0__MASK				0xfffff800
+#define CNA_DATA_SIZE2_RESERVED_0__SHIFT			11
+static inline uint32_t CNA_DATA_SIZE2_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE2_RESERVED_0__SHIFT) & CNA_DATA_SIZE2_RESERVED_0__MASK;
+}
+#define CNA_DATA_SIZE2_DATAOUT_WIDTH__MASK			0x000007ff
+#define CNA_DATA_SIZE2_DATAOUT_WIDTH__SHIFT			0
+static inline uint32_t CNA_DATA_SIZE2_DATAOUT_WIDTH(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE2_DATAOUT_WIDTH__SHIFT) & CNA_DATA_SIZE2_DATAOUT_WIDTH__MASK;
+}
+
+#define REG_CNA_DATA_SIZE3					0x0000102c
+#define CNA_DATA_SIZE3_RESERVED_0__MASK				0xff000000
+#define CNA_DATA_SIZE3_RESERVED_0__SHIFT			24
+static inline uint32_t CNA_DATA_SIZE3_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE3_RESERVED_0__SHIFT) & CNA_DATA_SIZE3_RESERVED_0__MASK;
+}
+#define CNA_DATA_SIZE3_SURF_MODE__MASK				0x00c00000
+#define CNA_DATA_SIZE3_SURF_MODE__SHIFT				22
+static inline uint32_t CNA_DATA_SIZE3_SURF_MODE(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE3_SURF_MODE__SHIFT) & CNA_DATA_SIZE3_SURF_MODE__MASK;
+}
+#define CNA_DATA_SIZE3_DATAOUT_ATOMICS__MASK			0x003fffff
+#define CNA_DATA_SIZE3_DATAOUT_ATOMICS__SHIFT			0
+static inline uint32_t CNA_DATA_SIZE3_DATAOUT_ATOMICS(uint32_t val)
+{
+	return ((val) << CNA_DATA_SIZE3_DATAOUT_ATOMICS__SHIFT) & CNA_DATA_SIZE3_DATAOUT_ATOMICS__MASK;
+}
+
+#define REG_CNA_WEIGHT_SIZE0					0x00001030
+#define CNA_WEIGHT_SIZE0_WEIGHT_BYTES__MASK			0xffffffff
+#define CNA_WEIGHT_SIZE0_WEIGHT_BYTES__SHIFT			0
+static inline uint32_t CNA_WEIGHT_SIZE0_WEIGHT_BYTES(uint32_t val)
+{
+	return ((val) << CNA_WEIGHT_SIZE0_WEIGHT_BYTES__SHIFT) & CNA_WEIGHT_SIZE0_WEIGHT_BYTES__MASK;
+}
+
+#define REG_CNA_WEIGHT_SIZE1					0x00001034
+#define CNA_WEIGHT_SIZE1_RESERVED_0__MASK			0xfff80000
+#define CNA_WEIGHT_SIZE1_RESERVED_0__SHIFT			19
+static inline uint32_t CNA_WEIGHT_SIZE1_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_WEIGHT_SIZE1_RESERVED_0__SHIFT) & CNA_WEIGHT_SIZE1_RESERVED_0__MASK;
+}
+#define CNA_WEIGHT_SIZE1_WEIGHT_BYTES_PER_KERNEL__MASK		0x0007ffff
+#define CNA_WEIGHT_SIZE1_WEIGHT_BYTES_PER_KERNEL__SHIFT		0
+static inline uint32_t CNA_WEIGHT_SIZE1_WEIGHT_BYTES_PER_KERNEL(uint32_t val)
+{
+	return ((val) << CNA_WEIGHT_SIZE1_WEIGHT_BYTES_PER_KERNEL__SHIFT) & CNA_WEIGHT_SIZE1_WEIGHT_BYTES_PER_KERNEL__MASK;
+}
+
+#define REG_CNA_WEIGHT_SIZE2					0x00001038
+#define CNA_WEIGHT_SIZE2_RESERVED_0__MASK			0xe0000000
+#define CNA_WEIGHT_SIZE2_RESERVED_0__SHIFT			29
+static inline uint32_t CNA_WEIGHT_SIZE2_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_WEIGHT_SIZE2_RESERVED_0__SHIFT) & CNA_WEIGHT_SIZE2_RESERVED_0__MASK;
+}
+#define CNA_WEIGHT_SIZE2_WEIGHT_WIDTH__MASK			0x1f000000
+#define CNA_WEIGHT_SIZE2_WEIGHT_WIDTH__SHIFT			24
+static inline uint32_t CNA_WEIGHT_SIZE2_WEIGHT_WIDTH(uint32_t val)
+{
+	return ((val) << CNA_WEIGHT_SIZE2_WEIGHT_WIDTH__SHIFT) & CNA_WEIGHT_SIZE2_WEIGHT_WIDTH__MASK;
+}
+#define CNA_WEIGHT_SIZE2_RESERVED_1__MASK			0x00e00000
+#define CNA_WEIGHT_SIZE2_RESERVED_1__SHIFT			21
+static inline uint32_t CNA_WEIGHT_SIZE2_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_WEIGHT_SIZE2_RESERVED_1__SHIFT) & CNA_WEIGHT_SIZE2_RESERVED_1__MASK;
+}
+#define CNA_WEIGHT_SIZE2_WEIGHT_HEIGHT__MASK			0x001f0000
+#define CNA_WEIGHT_SIZE2_WEIGHT_HEIGHT__SHIFT			16
+static inline uint32_t CNA_WEIGHT_SIZE2_WEIGHT_HEIGHT(uint32_t val)
+{
+	return ((val) << CNA_WEIGHT_SIZE2_WEIGHT_HEIGHT__SHIFT) & CNA_WEIGHT_SIZE2_WEIGHT_HEIGHT__MASK;
+}
+#define CNA_WEIGHT_SIZE2_RESERVED_2__MASK			0x0000c000
+#define CNA_WEIGHT_SIZE2_RESERVED_2__SHIFT			14
+static inline uint32_t CNA_WEIGHT_SIZE2_RESERVED_2(uint32_t val)
+{
+	return ((val) << CNA_WEIGHT_SIZE2_RESERVED_2__SHIFT) & CNA_WEIGHT_SIZE2_RESERVED_2__MASK;
+}
+#define CNA_WEIGHT_SIZE2_WEIGHT_KERNELS__MASK			0x00003fff
+#define CNA_WEIGHT_SIZE2_WEIGHT_KERNELS__SHIFT			0
+static inline uint32_t CNA_WEIGHT_SIZE2_WEIGHT_KERNELS(uint32_t val)
+{
+	return ((val) << CNA_WEIGHT_SIZE2_WEIGHT_KERNELS__SHIFT) & CNA_WEIGHT_SIZE2_WEIGHT_KERNELS__MASK;
+}
+
+#define REG_CNA_CBUF_CON0					0x00001040
+#define CNA_CBUF_CON0_RESERVED_0__MASK				0xffffc000
+#define CNA_CBUF_CON0_RESERVED_0__SHIFT				14
+static inline uint32_t CNA_CBUF_CON0_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_CBUF_CON0_RESERVED_0__SHIFT) & CNA_CBUF_CON0_RESERVED_0__MASK;
+}
+#define CNA_CBUF_CON0_WEIGHT_REUSE__MASK			0x00002000
+#define CNA_CBUF_CON0_WEIGHT_REUSE__SHIFT			13
+static inline uint32_t CNA_CBUF_CON0_WEIGHT_REUSE(uint32_t val)
+{
+	return ((val) << CNA_CBUF_CON0_WEIGHT_REUSE__SHIFT) & CNA_CBUF_CON0_WEIGHT_REUSE__MASK;
+}
+#define CNA_CBUF_CON0_DATA_REUSE__MASK				0x00001000
+#define CNA_CBUF_CON0_DATA_REUSE__SHIFT				12
+static inline uint32_t CNA_CBUF_CON0_DATA_REUSE(uint32_t val)
+{
+	return ((val) << CNA_CBUF_CON0_DATA_REUSE__SHIFT) & CNA_CBUF_CON0_DATA_REUSE__MASK;
+}
+#define CNA_CBUF_CON0_RESERVED_1__MASK				0x00000800
+#define CNA_CBUF_CON0_RESERVED_1__SHIFT				11
+static inline uint32_t CNA_CBUF_CON0_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_CBUF_CON0_RESERVED_1__SHIFT) & CNA_CBUF_CON0_RESERVED_1__MASK;
+}
+#define CNA_CBUF_CON0_FC_DATA_BANK__MASK			0x00000700
+#define CNA_CBUF_CON0_FC_DATA_BANK__SHIFT			8
+static inline uint32_t CNA_CBUF_CON0_FC_DATA_BANK(uint32_t val)
+{
+	return ((val) << CNA_CBUF_CON0_FC_DATA_BANK__SHIFT) & CNA_CBUF_CON0_FC_DATA_BANK__MASK;
+}
+#define CNA_CBUF_CON0_WEIGHT_BANK__MASK				0x000000f0
+#define CNA_CBUF_CON0_WEIGHT_BANK__SHIFT			4
+static inline uint32_t CNA_CBUF_CON0_WEIGHT_BANK(uint32_t val)
+{
+	return ((val) << CNA_CBUF_CON0_WEIGHT_BANK__SHIFT) & CNA_CBUF_CON0_WEIGHT_BANK__MASK;
+}
+#define CNA_CBUF_CON0_DATA_BANK__MASK				0x0000000f
+#define CNA_CBUF_CON0_DATA_BANK__SHIFT				0
+static inline uint32_t CNA_CBUF_CON0_DATA_BANK(uint32_t val)
+{
+	return ((val) << CNA_CBUF_CON0_DATA_BANK__SHIFT) & CNA_CBUF_CON0_DATA_BANK__MASK;
+}
+
+#define REG_CNA_CBUF_CON1					0x00001044
+#define CNA_CBUF_CON1_RESERVED_0__MASK				0xffffc000
+#define CNA_CBUF_CON1_RESERVED_0__SHIFT				14
+static inline uint32_t CNA_CBUF_CON1_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_CBUF_CON1_RESERVED_0__SHIFT) & CNA_CBUF_CON1_RESERVED_0__MASK;
+}
+#define CNA_CBUF_CON1_DATA_ENTRIES__MASK			0x00003fff
+#define CNA_CBUF_CON1_DATA_ENTRIES__SHIFT			0
+static inline uint32_t CNA_CBUF_CON1_DATA_ENTRIES(uint32_t val)
+{
+	return ((val) << CNA_CBUF_CON1_DATA_ENTRIES__SHIFT) & CNA_CBUF_CON1_DATA_ENTRIES__MASK;
+}
+
+#define REG_CNA_CVT_CON0					0x0000104c
+#define CNA_CVT_CON0_RESERVED_0__MASK				0xf0000000
+#define CNA_CVT_CON0_RESERVED_0__SHIFT				28
+static inline uint32_t CNA_CVT_CON0_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON0_RESERVED_0__SHIFT) & CNA_CVT_CON0_RESERVED_0__MASK;
+}
+#define CNA_CVT_CON0_CVT_TRUNCATE_3__MASK			0x0fc00000
+#define CNA_CVT_CON0_CVT_TRUNCATE_3__SHIFT			22
+static inline uint32_t CNA_CVT_CON0_CVT_TRUNCATE_3(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON0_CVT_TRUNCATE_3__SHIFT) & CNA_CVT_CON0_CVT_TRUNCATE_3__MASK;
+}
+#define CNA_CVT_CON0_CVT_TRUNCATE_2__MASK			0x003f0000
+#define CNA_CVT_CON0_CVT_TRUNCATE_2__SHIFT			16
+static inline uint32_t CNA_CVT_CON0_CVT_TRUNCATE_2(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON0_CVT_TRUNCATE_2__SHIFT) & CNA_CVT_CON0_CVT_TRUNCATE_2__MASK;
+}
+#define CNA_CVT_CON0_CVT_TRUNCATE_1__MASK			0x0000fc00
+#define CNA_CVT_CON0_CVT_TRUNCATE_1__SHIFT			10
+static inline uint32_t CNA_CVT_CON0_CVT_TRUNCATE_1(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON0_CVT_TRUNCATE_1__SHIFT) & CNA_CVT_CON0_CVT_TRUNCATE_1__MASK;
+}
+#define CNA_CVT_CON0_CVT_TRUNCATE_0__MASK			0x000003f0
+#define CNA_CVT_CON0_CVT_TRUNCATE_0__SHIFT			4
+static inline uint32_t CNA_CVT_CON0_CVT_TRUNCATE_0(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON0_CVT_TRUNCATE_0__SHIFT) & CNA_CVT_CON0_CVT_TRUNCATE_0__MASK;
+}
+#define CNA_CVT_CON0_DATA_SIGN__MASK				0x00000008
+#define CNA_CVT_CON0_DATA_SIGN__SHIFT				3
+static inline uint32_t CNA_CVT_CON0_DATA_SIGN(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON0_DATA_SIGN__SHIFT) & CNA_CVT_CON0_DATA_SIGN__MASK;
+}
+#define CNA_CVT_CON0_ROUND_TYPE__MASK				0x00000004
+#define CNA_CVT_CON0_ROUND_TYPE__SHIFT				2
+static inline uint32_t CNA_CVT_CON0_ROUND_TYPE(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON0_ROUND_TYPE__SHIFT) & CNA_CVT_CON0_ROUND_TYPE__MASK;
+}
+#define CNA_CVT_CON0_CVT_TYPE__MASK				0x00000002
+#define CNA_CVT_CON0_CVT_TYPE__SHIFT				1
+static inline uint32_t CNA_CVT_CON0_CVT_TYPE(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON0_CVT_TYPE__SHIFT) & CNA_CVT_CON0_CVT_TYPE__MASK;
+}
+#define CNA_CVT_CON0_CVT_BYPASS__MASK				0x00000001
+#define CNA_CVT_CON0_CVT_BYPASS__SHIFT				0
+static inline uint32_t CNA_CVT_CON0_CVT_BYPASS(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON0_CVT_BYPASS__SHIFT) & CNA_CVT_CON0_CVT_BYPASS__MASK;
+}
+
+#define REG_CNA_CVT_CON1					0x00001050
+#define CNA_CVT_CON1_CVT_SCALE0__MASK				0xffff0000
+#define CNA_CVT_CON1_CVT_SCALE0__SHIFT				16
+static inline uint32_t CNA_CVT_CON1_CVT_SCALE0(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON1_CVT_SCALE0__SHIFT) & CNA_CVT_CON1_CVT_SCALE0__MASK;
+}
+#define CNA_CVT_CON1_CVT_OFFSET0__MASK				0x0000ffff
+#define CNA_CVT_CON1_CVT_OFFSET0__SHIFT				0
+static inline uint32_t CNA_CVT_CON1_CVT_OFFSET0(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON1_CVT_OFFSET0__SHIFT) & CNA_CVT_CON1_CVT_OFFSET0__MASK;
+}
+
+#define REG_CNA_CVT_CON2					0x00001054
+#define CNA_CVT_CON2_CVT_SCALE1__MASK				0xffff0000
+#define CNA_CVT_CON2_CVT_SCALE1__SHIFT				16
+static inline uint32_t CNA_CVT_CON2_CVT_SCALE1(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON2_CVT_SCALE1__SHIFT) & CNA_CVT_CON2_CVT_SCALE1__MASK;
+}
+#define CNA_CVT_CON2_CVT_OFFSET1__MASK				0x0000ffff
+#define CNA_CVT_CON2_CVT_OFFSET1__SHIFT				0
+static inline uint32_t CNA_CVT_CON2_CVT_OFFSET1(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON2_CVT_OFFSET1__SHIFT) & CNA_CVT_CON2_CVT_OFFSET1__MASK;
+}
+
+#define REG_CNA_CVT_CON3					0x00001058
+#define CNA_CVT_CON3_CVT_SCALE2__MASK				0xffff0000
+#define CNA_CVT_CON3_CVT_SCALE2__SHIFT				16
+static inline uint32_t CNA_CVT_CON3_CVT_SCALE2(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON3_CVT_SCALE2__SHIFT) & CNA_CVT_CON3_CVT_SCALE2__MASK;
+}
+#define CNA_CVT_CON3_CVT_OFFSET2__MASK				0x0000ffff
+#define CNA_CVT_CON3_CVT_OFFSET2__SHIFT				0
+static inline uint32_t CNA_CVT_CON3_CVT_OFFSET2(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON3_CVT_OFFSET2__SHIFT) & CNA_CVT_CON3_CVT_OFFSET2__MASK;
+}
+
+#define REG_CNA_CVT_CON4					0x0000105c
+#define CNA_CVT_CON4_CVT_SCALE3__MASK				0xffff0000
+#define CNA_CVT_CON4_CVT_SCALE3__SHIFT				16
+static inline uint32_t CNA_CVT_CON4_CVT_SCALE3(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON4_CVT_SCALE3__SHIFT) & CNA_CVT_CON4_CVT_SCALE3__MASK;
+}
+#define CNA_CVT_CON4_CVT_OFFSET3__MASK				0x0000ffff
+#define CNA_CVT_CON4_CVT_OFFSET3__SHIFT				0
+static inline uint32_t CNA_CVT_CON4_CVT_OFFSET3(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON4_CVT_OFFSET3__SHIFT) & CNA_CVT_CON4_CVT_OFFSET3__MASK;
+}
+
+#define REG_CNA_FC_CON0						0x00001060
+#define CNA_FC_CON0_FC_SKIP_DATA__MASK				0xffff0000
+#define CNA_FC_CON0_FC_SKIP_DATA__SHIFT				16
+static inline uint32_t CNA_FC_CON0_FC_SKIP_DATA(uint32_t val)
+{
+	return ((val) << CNA_FC_CON0_FC_SKIP_DATA__SHIFT) & CNA_FC_CON0_FC_SKIP_DATA__MASK;
+}
+#define CNA_FC_CON0_RESERVED_0__MASK				0x0000fffe
+#define CNA_FC_CON0_RESERVED_0__SHIFT				1
+static inline uint32_t CNA_FC_CON0_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_FC_CON0_RESERVED_0__SHIFT) & CNA_FC_CON0_RESERVED_0__MASK;
+}
+#define CNA_FC_CON0_FC_SKIP_EN__MASK				0x00000001
+#define CNA_FC_CON0_FC_SKIP_EN__SHIFT				0
+static inline uint32_t CNA_FC_CON0_FC_SKIP_EN(uint32_t val)
+{
+	return ((val) << CNA_FC_CON0_FC_SKIP_EN__SHIFT) & CNA_FC_CON0_FC_SKIP_EN__MASK;
+}
+
+#define REG_CNA_FC_CON1						0x00001064
+#define CNA_FC_CON1_RESERVED_0__MASK				0xfffe0000
+#define CNA_FC_CON1_RESERVED_0__SHIFT				17
+static inline uint32_t CNA_FC_CON1_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_FC_CON1_RESERVED_0__SHIFT) & CNA_FC_CON1_RESERVED_0__MASK;
+}
+#define CNA_FC_CON1_DATA_OFFSET__MASK				0x0001ffff
+#define CNA_FC_CON1_DATA_OFFSET__SHIFT				0
+static inline uint32_t CNA_FC_CON1_DATA_OFFSET(uint32_t val)
+{
+	return ((val) << CNA_FC_CON1_DATA_OFFSET__SHIFT) & CNA_FC_CON1_DATA_OFFSET__MASK;
+}
+
+#define REG_CNA_PAD_CON0					0x00001068
+#define CNA_PAD_CON0_RESERVED_0__MASK				0xffffff00
+#define CNA_PAD_CON0_RESERVED_0__SHIFT				8
+static inline uint32_t CNA_PAD_CON0_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_PAD_CON0_RESERVED_0__SHIFT) & CNA_PAD_CON0_RESERVED_0__MASK;
+}
+#define CNA_PAD_CON0_PAD_LEFT__MASK				0x000000f0
+#define CNA_PAD_CON0_PAD_LEFT__SHIFT				4
+static inline uint32_t CNA_PAD_CON0_PAD_LEFT(uint32_t val)
+{
+	return ((val) << CNA_PAD_CON0_PAD_LEFT__SHIFT) & CNA_PAD_CON0_PAD_LEFT__MASK;
+}
+#define CNA_PAD_CON0_PAD_TOP__MASK				0x0000000f
+#define CNA_PAD_CON0_PAD_TOP__SHIFT				0
+static inline uint32_t CNA_PAD_CON0_PAD_TOP(uint32_t val)
+{
+	return ((val) << CNA_PAD_CON0_PAD_TOP__SHIFT) & CNA_PAD_CON0_PAD_TOP__MASK;
+}
+
+#define REG_CNA_FEATURE_DATA_ADDR				0x00001070
+#define CNA_FEATURE_DATA_ADDR_FEATURE_BASE_ADDR__MASK		0xffffffff
+#define CNA_FEATURE_DATA_ADDR_FEATURE_BASE_ADDR__SHIFT		0
+static inline uint32_t CNA_FEATURE_DATA_ADDR_FEATURE_BASE_ADDR(uint32_t val)
+{
+	return ((val) << CNA_FEATURE_DATA_ADDR_FEATURE_BASE_ADDR__SHIFT) & CNA_FEATURE_DATA_ADDR_FEATURE_BASE_ADDR__MASK;
+}
+
+#define REG_CNA_FC_CON2						0x00001074
+#define CNA_FC_CON2_RESERVED_0__MASK				0xfffe0000
+#define CNA_FC_CON2_RESERVED_0__SHIFT				17
+static inline uint32_t CNA_FC_CON2_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_FC_CON2_RESERVED_0__SHIFT) & CNA_FC_CON2_RESERVED_0__MASK;
+}
+#define CNA_FC_CON2_WEIGHT_OFFSET__MASK				0x0001ffff
+#define CNA_FC_CON2_WEIGHT_OFFSET__SHIFT			0
+static inline uint32_t CNA_FC_CON2_WEIGHT_OFFSET(uint32_t val)
+{
+	return ((val) << CNA_FC_CON2_WEIGHT_OFFSET__SHIFT) & CNA_FC_CON2_WEIGHT_OFFSET__MASK;
+}
+
+#define REG_CNA_DMA_CON0					0x00001078
+#define CNA_DMA_CON0_OV4K_BYPASS__MASK				0x80000000
+#define CNA_DMA_CON0_OV4K_BYPASS__SHIFT				31
+static inline uint32_t CNA_DMA_CON0_OV4K_BYPASS(uint32_t val)
+{
+	return ((val) << CNA_DMA_CON0_OV4K_BYPASS__SHIFT) & CNA_DMA_CON0_OV4K_BYPASS__MASK;
+}
+#define CNA_DMA_CON0_RESERVED_0__MASK				0x7ff00000
+#define CNA_DMA_CON0_RESERVED_0__SHIFT				20
+static inline uint32_t CNA_DMA_CON0_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_DMA_CON0_RESERVED_0__SHIFT) & CNA_DMA_CON0_RESERVED_0__MASK;
+}
+#define CNA_DMA_CON0_WEIGHT_BURST_LEN__MASK			0x000f0000
+#define CNA_DMA_CON0_WEIGHT_BURST_LEN__SHIFT			16
+static inline uint32_t CNA_DMA_CON0_WEIGHT_BURST_LEN(uint32_t val)
+{
+	return ((val) << CNA_DMA_CON0_WEIGHT_BURST_LEN__SHIFT) & CNA_DMA_CON0_WEIGHT_BURST_LEN__MASK;
+}
+#define CNA_DMA_CON0_RESERVED_1__MASK				0x0000fff0
+#define CNA_DMA_CON0_RESERVED_1__SHIFT				4
+static inline uint32_t CNA_DMA_CON0_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_DMA_CON0_RESERVED_1__SHIFT) & CNA_DMA_CON0_RESERVED_1__MASK;
+}
+#define CNA_DMA_CON0_DATA_BURST_LEN__MASK			0x0000000f
+#define CNA_DMA_CON0_DATA_BURST_LEN__SHIFT			0
+static inline uint32_t CNA_DMA_CON0_DATA_BURST_LEN(uint32_t val)
+{
+	return ((val) << CNA_DMA_CON0_DATA_BURST_LEN__SHIFT) & CNA_DMA_CON0_DATA_BURST_LEN__MASK;
+}
+
+#define REG_CNA_DMA_CON1					0x0000107c
+#define CNA_DMA_CON1_RESERVED_0__MASK				0xf0000000
+#define CNA_DMA_CON1_RESERVED_0__SHIFT				28
+static inline uint32_t CNA_DMA_CON1_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_DMA_CON1_RESERVED_0__SHIFT) & CNA_DMA_CON1_RESERVED_0__MASK;
+}
+#define CNA_DMA_CON1_LINE_STRIDE__MASK				0x0fffffff
+#define CNA_DMA_CON1_LINE_STRIDE__SHIFT				0
+static inline uint32_t CNA_DMA_CON1_LINE_STRIDE(uint32_t val)
+{
+	return ((val) << CNA_DMA_CON1_LINE_STRIDE__SHIFT) & CNA_DMA_CON1_LINE_STRIDE__MASK;
+}
+
+#define REG_CNA_DMA_CON2					0x00001080
+#define CNA_DMA_CON2_RESERVED_0__MASK				0xf0000000
+#define CNA_DMA_CON2_RESERVED_0__SHIFT				28
+static inline uint32_t CNA_DMA_CON2_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_DMA_CON2_RESERVED_0__SHIFT) & CNA_DMA_CON2_RESERVED_0__MASK;
+}
+#define CNA_DMA_CON2_SURF_STRIDE__MASK				0x0fffffff
+#define CNA_DMA_CON2_SURF_STRIDE__SHIFT				0
+static inline uint32_t CNA_DMA_CON2_SURF_STRIDE(uint32_t val)
+{
+	return ((val) << CNA_DMA_CON2_SURF_STRIDE__SHIFT) & CNA_DMA_CON2_SURF_STRIDE__MASK;
+}
+
+#define REG_CNA_FC_DATA_SIZE0					0x00001084
+#define CNA_FC_DATA_SIZE0_RESERVED_0__MASK			0xc0000000
+#define CNA_FC_DATA_SIZE0_RESERVED_0__SHIFT			30
+static inline uint32_t CNA_FC_DATA_SIZE0_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_FC_DATA_SIZE0_RESERVED_0__SHIFT) & CNA_FC_DATA_SIZE0_RESERVED_0__MASK;
+}
+#define CNA_FC_DATA_SIZE0_DMA_WIDTH__MASK			0x3fff0000
+#define CNA_FC_DATA_SIZE0_DMA_WIDTH__SHIFT			16
+static inline uint32_t CNA_FC_DATA_SIZE0_DMA_WIDTH(uint32_t val)
+{
+	return ((val) << CNA_FC_DATA_SIZE0_DMA_WIDTH__SHIFT) & CNA_FC_DATA_SIZE0_DMA_WIDTH__MASK;
+}
+#define CNA_FC_DATA_SIZE0_RESERVED_1__MASK			0x0000f800
+#define CNA_FC_DATA_SIZE0_RESERVED_1__SHIFT			11
+static inline uint32_t CNA_FC_DATA_SIZE0_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_FC_DATA_SIZE0_RESERVED_1__SHIFT) & CNA_FC_DATA_SIZE0_RESERVED_1__MASK;
+}
+#define CNA_FC_DATA_SIZE0_DMA_HEIGHT__MASK			0x000007ff
+#define CNA_FC_DATA_SIZE0_DMA_HEIGHT__SHIFT			0
+static inline uint32_t CNA_FC_DATA_SIZE0_DMA_HEIGHT(uint32_t val)
+{
+	return ((val) << CNA_FC_DATA_SIZE0_DMA_HEIGHT__SHIFT) & CNA_FC_DATA_SIZE0_DMA_HEIGHT__MASK;
+}
+
+#define REG_CNA_FC_DATA_SIZE1					0x00001088
+#define CNA_FC_DATA_SIZE1_RESERVED_0__MASK			0xffff0000
+#define CNA_FC_DATA_SIZE1_RESERVED_0__SHIFT			16
+static inline uint32_t CNA_FC_DATA_SIZE1_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_FC_DATA_SIZE1_RESERVED_0__SHIFT) & CNA_FC_DATA_SIZE1_RESERVED_0__MASK;
+}
+#define CNA_FC_DATA_SIZE1_DMA_CHANNEL__MASK			0x0000ffff
+#define CNA_FC_DATA_SIZE1_DMA_CHANNEL__SHIFT			0
+static inline uint32_t CNA_FC_DATA_SIZE1_DMA_CHANNEL(uint32_t val)
+{
+	return ((val) << CNA_FC_DATA_SIZE1_DMA_CHANNEL__SHIFT) & CNA_FC_DATA_SIZE1_DMA_CHANNEL__MASK;
+}
+
+#define REG_CNA_CLK_GATE					0x00001090
+#define CNA_CLK_GATE_RESERVED_0__MASK				0xffffffe0
+#define CNA_CLK_GATE_RESERVED_0__SHIFT				5
+static inline uint32_t CNA_CLK_GATE_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_CLK_GATE_RESERVED_0__SHIFT) & CNA_CLK_GATE_RESERVED_0__MASK;
+}
+#define CNA_CLK_GATE_CBUF_CS_DISABLE_CLKGATE__MASK		0x00000010
+#define CNA_CLK_GATE_CBUF_CS_DISABLE_CLKGATE__SHIFT		4
+static inline uint32_t CNA_CLK_GATE_CBUF_CS_DISABLE_CLKGATE(uint32_t val)
+{
+	return ((val) << CNA_CLK_GATE_CBUF_CS_DISABLE_CLKGATE__SHIFT) & CNA_CLK_GATE_CBUF_CS_DISABLE_CLKGATE__MASK;
+}
+#define CNA_CLK_GATE_RESERVED_1__MASK				0x00000008
+#define CNA_CLK_GATE_RESERVED_1__SHIFT				3
+static inline uint32_t CNA_CLK_GATE_RESERVED_1(uint32_t val)
+{
+	return ((val) << CNA_CLK_GATE_RESERVED_1__SHIFT) & CNA_CLK_GATE_RESERVED_1__MASK;
+}
+#define CNA_CLK_GATE_CSC_DISABLE_CLKGATE__MASK			0x00000004
+#define CNA_CLK_GATE_CSC_DISABLE_CLKGATE__SHIFT			2
+static inline uint32_t CNA_CLK_GATE_CSC_DISABLE_CLKGATE(uint32_t val)
+{
+	return ((val) << CNA_CLK_GATE_CSC_DISABLE_CLKGATE__SHIFT) & CNA_CLK_GATE_CSC_DISABLE_CLKGATE__MASK;
+}
+#define CNA_CLK_GATE_CNA_WEIGHT_DISABLE_CLKGATE__MASK		0x00000002
+#define CNA_CLK_GATE_CNA_WEIGHT_DISABLE_CLKGATE__SHIFT		1
+static inline uint32_t CNA_CLK_GATE_CNA_WEIGHT_DISABLE_CLKGATE(uint32_t val)
+{
+	return ((val) << CNA_CLK_GATE_CNA_WEIGHT_DISABLE_CLKGATE__SHIFT) & CNA_CLK_GATE_CNA_WEIGHT_DISABLE_CLKGATE__MASK;
+}
+#define CNA_CLK_GATE_CNA_FEATURE_DISABLE_CLKGATE__MASK		0x00000001
+#define CNA_CLK_GATE_CNA_FEATURE_DISABLE_CLKGATE__SHIFT		0
+static inline uint32_t CNA_CLK_GATE_CNA_FEATURE_DISABLE_CLKGATE(uint32_t val)
+{
+	return ((val) << CNA_CLK_GATE_CNA_FEATURE_DISABLE_CLKGATE__SHIFT) & CNA_CLK_GATE_CNA_FEATURE_DISABLE_CLKGATE__MASK;
+}
+
+#define REG_CNA_DCOMP_CTRL					0x00001100
+#define CNA_DCOMP_CTRL_RESERVED_0__MASK				0xfffffff0
+#define CNA_DCOMP_CTRL_RESERVED_0__SHIFT			4
+static inline uint32_t CNA_DCOMP_CTRL_RESERVED_0(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_CTRL_RESERVED_0__SHIFT) & CNA_DCOMP_CTRL_RESERVED_0__MASK;
+}
+#define CNA_DCOMP_CTRL_WT_DEC_BYPASS__MASK			0x00000008
+#define CNA_DCOMP_CTRL_WT_DEC_BYPASS__SHIFT			3
+static inline uint32_t CNA_DCOMP_CTRL_WT_DEC_BYPASS(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_CTRL_WT_DEC_BYPASS__SHIFT) & CNA_DCOMP_CTRL_WT_DEC_BYPASS__MASK;
+}
+#define CNA_DCOMP_CTRL_DECOMP_CONTROL__MASK			0x00000007
+#define CNA_DCOMP_CTRL_DECOMP_CONTROL__SHIFT			0
+static inline uint32_t CNA_DCOMP_CTRL_DECOMP_CONTROL(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_CTRL_DECOMP_CONTROL__SHIFT) & CNA_DCOMP_CTRL_DECOMP_CONTROL__MASK;
+}
+
+#define REG_CNA_DCOMP_REGNUM					0x00001104
+#define CNA_DCOMP_REGNUM_DCOMP_REGNUM__MASK			0xffffffff
+#define CNA_DCOMP_REGNUM_DCOMP_REGNUM__SHIFT			0
+static inline uint32_t CNA_DCOMP_REGNUM_DCOMP_REGNUM(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_REGNUM_DCOMP_REGNUM__SHIFT) & CNA_DCOMP_REGNUM_DCOMP_REGNUM__MASK;
+}
+
+#define REG_CNA_DCOMP_ADDR0					0x00001110
+#define CNA_DCOMP_ADDR0_DECOMPRESS_ADDR0__MASK			0xffffffff
+#define CNA_DCOMP_ADDR0_DECOMPRESS_ADDR0__SHIFT			0
+static inline uint32_t CNA_DCOMP_ADDR0_DECOMPRESS_ADDR0(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_ADDR0_DECOMPRESS_ADDR0__SHIFT) & CNA_DCOMP_ADDR0_DECOMPRESS_ADDR0__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT0					0x00001140
+#define CNA_DCOMP_AMOUNT0_DCOMP_AMOUNT0__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT0_DCOMP_AMOUNT0__SHIFT			0
+static inline uint32_t CNA_DCOMP_AMOUNT0_DCOMP_AMOUNT0(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT0_DCOMP_AMOUNT0__SHIFT) & CNA_DCOMP_AMOUNT0_DCOMP_AMOUNT0__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT1					0x00001144
+#define CNA_DCOMP_AMOUNT1_DCOMP_AMOUNT1__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT1_DCOMP_AMOUNT1__SHIFT			0
+static inline uint32_t CNA_DCOMP_AMOUNT1_DCOMP_AMOUNT1(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT1_DCOMP_AMOUNT1__SHIFT) & CNA_DCOMP_AMOUNT1_DCOMP_AMOUNT1__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT2					0x00001148
+#define CNA_DCOMP_AMOUNT2_DCOMP_AMOUNT2__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT2_DCOMP_AMOUNT2__SHIFT			0
+static inline uint32_t CNA_DCOMP_AMOUNT2_DCOMP_AMOUNT2(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT2_DCOMP_AMOUNT2__SHIFT) & CNA_DCOMP_AMOUNT2_DCOMP_AMOUNT2__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT3					0x0000114c
+#define CNA_DCOMP_AMOUNT3_DCOMP_AMOUNT3__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT3_DCOMP_AMOUNT3__SHIFT			0
+static inline uint32_t CNA_DCOMP_AMOUNT3_DCOMP_AMOUNT3(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT3_DCOMP_AMOUNT3__SHIFT) & CNA_DCOMP_AMOUNT3_DCOMP_AMOUNT3__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT4					0x00001150
+#define CNA_DCOMP_AMOUNT4_DCOMP_AMOUNT4__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT4_DCOMP_AMOUNT4__SHIFT			0
+static inline uint32_t CNA_DCOMP_AMOUNT4_DCOMP_AMOUNT4(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT4_DCOMP_AMOUNT4__SHIFT) & CNA_DCOMP_AMOUNT4_DCOMP_AMOUNT4__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT5					0x00001154
+#define CNA_DCOMP_AMOUNT5_DCOMP_AMOUNT5__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT5_DCOMP_AMOUNT5__SHIFT			0
+static inline uint32_t CNA_DCOMP_AMOUNT5_DCOMP_AMOUNT5(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT5_DCOMP_AMOUNT5__SHIFT) & CNA_DCOMP_AMOUNT5_DCOMP_AMOUNT5__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT6					0x00001158
+#define CNA_DCOMP_AMOUNT6_DCOMP_AMOUNT6__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT6_DCOMP_AMOUNT6__SHIFT			0
+static inline uint32_t CNA_DCOMP_AMOUNT6_DCOMP_AMOUNT6(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT6_DCOMP_AMOUNT6__SHIFT) & CNA_DCOMP_AMOUNT6_DCOMP_AMOUNT6__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT7					0x0000115c
+#define CNA_DCOMP_AMOUNT7_DCOMP_AMOUNT7__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT7_DCOMP_AMOUNT7__SHIFT			0
+static inline uint32_t CNA_DCOMP_AMOUNT7_DCOMP_AMOUNT7(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT7_DCOMP_AMOUNT7__SHIFT) & CNA_DCOMP_AMOUNT7_DCOMP_AMOUNT7__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT8					0x00001160
+#define CNA_DCOMP_AMOUNT8_DCOMP_AMOUNT8__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT8_DCOMP_AMOUNT8__SHIFT			0
+static inline uint32_t CNA_DCOMP_AMOUNT8_DCOMP_AMOUNT8(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT8_DCOMP_AMOUNT8__SHIFT) & CNA_DCOMP_AMOUNT8_DCOMP_AMOUNT8__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT9					0x00001164
+#define CNA_DCOMP_AMOUNT9_DCOMP_AMOUNT9__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT9_DCOMP_AMOUNT9__SHIFT			0
+static inline uint32_t CNA_DCOMP_AMOUNT9_DCOMP_AMOUNT9(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT9_DCOMP_AMOUNT9__SHIFT) & CNA_DCOMP_AMOUNT9_DCOMP_AMOUNT9__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT10					0x00001168
+#define CNA_DCOMP_AMOUNT10_DCOMP_AMOUNT10__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT10_DCOMP_AMOUNT10__SHIFT		0
+static inline uint32_t CNA_DCOMP_AMOUNT10_DCOMP_AMOUNT10(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT10_DCOMP_AMOUNT10__SHIFT) & CNA_DCOMP_AMOUNT10_DCOMP_AMOUNT10__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT11					0x0000116c
+#define CNA_DCOMP_AMOUNT11_DCOMP_AMOUNT11__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT11_DCOMP_AMOUNT11__SHIFT		0
+static inline uint32_t CNA_DCOMP_AMOUNT11_DCOMP_AMOUNT11(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT11_DCOMP_AMOUNT11__SHIFT) & CNA_DCOMP_AMOUNT11_DCOMP_AMOUNT11__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT12					0x00001170
+#define CNA_DCOMP_AMOUNT12_DCOMP_AMOUNT12__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT12_DCOMP_AMOUNT12__SHIFT		0
+static inline uint32_t CNA_DCOMP_AMOUNT12_DCOMP_AMOUNT12(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT12_DCOMP_AMOUNT12__SHIFT) & CNA_DCOMP_AMOUNT12_DCOMP_AMOUNT12__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT13					0x00001174
+#define CNA_DCOMP_AMOUNT13_DCOMP_AMOUNT13__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT13_DCOMP_AMOUNT13__SHIFT		0
+static inline uint32_t CNA_DCOMP_AMOUNT13_DCOMP_AMOUNT13(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT13_DCOMP_AMOUNT13__SHIFT) & CNA_DCOMP_AMOUNT13_DCOMP_AMOUNT13__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT14					0x00001178
+#define CNA_DCOMP_AMOUNT14_DCOMP_AMOUNT14__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT14_DCOMP_AMOUNT14__SHIFT		0
+static inline uint32_t CNA_DCOMP_AMOUNT14_DCOMP_AMOUNT14(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT14_DCOMP_AMOUNT14__SHIFT) & CNA_DCOMP_AMOUNT14_DCOMP_AMOUNT14__MASK;
+}
+
+#define REG_CNA_DCOMP_AMOUNT15					0x0000117c
+#define CNA_DCOMP_AMOUNT15_DCOMP_AMOUNT15__MASK			0xffffffff
+#define CNA_DCOMP_AMOUNT15_DCOMP_AMOUNT15__SHIFT		0
+static inline uint32_t CNA_DCOMP_AMOUNT15_DCOMP_AMOUNT15(uint32_t val)
+{
+	return ((val) << CNA_DCOMP_AMOUNT15_DCOMP_AMOUNT15__SHIFT) & CNA_DCOMP_AMOUNT15_DCOMP_AMOUNT15__MASK;
+}
+
+#define REG_CNA_CVT_CON5					0x00001180
+#define CNA_CVT_CON5_PER_CHANNEL_CVT_EN__MASK			0xffffffff
+#define CNA_CVT_CON5_PER_CHANNEL_CVT_EN__SHIFT			0
+static inline uint32_t CNA_CVT_CON5_PER_CHANNEL_CVT_EN(uint32_t val)
+{
+	return ((val) << CNA_CVT_CON5_PER_CHANNEL_CVT_EN__SHIFT) & CNA_CVT_CON5_PER_CHANNEL_CVT_EN__MASK;
+}
+
+#define REG_CNA_PAD_CON1					0x00001184
+#define CNA_PAD_CON1_PAD_VALUE__MASK				0xffffffff
+#define CNA_PAD_CON1_PAD_VALUE__SHIFT				0
+static inline uint32_t CNA_PAD_CON1_PAD_VALUE(uint32_t val)
+{
+	return ((val) << CNA_PAD_CON1_PAD_VALUE__SHIFT) & CNA_PAD_CON1_PAD_VALUE__MASK;
+}
+
+#define REG_CORE_S_STATUS					0x00003000
+#define CORE_S_STATUS_RESERVED_0__MASK				0xfffc0000
+#define CORE_S_STATUS_RESERVED_0__SHIFT				18
+static inline uint32_t CORE_S_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << CORE_S_STATUS_RESERVED_0__SHIFT) & CORE_S_STATUS_RESERVED_0__MASK;
+}
+#define CORE_S_STATUS_STATUS_1__MASK				0x00030000
+#define CORE_S_STATUS_STATUS_1__SHIFT				16
+static inline uint32_t CORE_S_STATUS_STATUS_1(uint32_t val)
+{
+	return ((val) << CORE_S_STATUS_STATUS_1__SHIFT) & CORE_S_STATUS_STATUS_1__MASK;
+}
+#define CORE_S_STATUS_RESERVED_1__MASK				0x0000fffc
+#define CORE_S_STATUS_RESERVED_1__SHIFT				2
+static inline uint32_t CORE_S_STATUS_RESERVED_1(uint32_t val)
+{
+	return ((val) << CORE_S_STATUS_RESERVED_1__SHIFT) & CORE_S_STATUS_RESERVED_1__MASK;
+}
+#define CORE_S_STATUS_STATUS_0__MASK				0x00000003
+#define CORE_S_STATUS_STATUS_0__SHIFT				0
+static inline uint32_t CORE_S_STATUS_STATUS_0(uint32_t val)
+{
+	return ((val) << CORE_S_STATUS_STATUS_0__SHIFT) & CORE_S_STATUS_STATUS_0__MASK;
+}
+
+#define REG_CORE_S_POINTER					0x00003004
+#define CORE_S_POINTER_RESERVED_0__MASK				0xfffe0000
+#define CORE_S_POINTER_RESERVED_0__SHIFT			17
+static inline uint32_t CORE_S_POINTER_RESERVED_0(uint32_t val)
+{
+	return ((val) << CORE_S_POINTER_RESERVED_0__SHIFT) & CORE_S_POINTER_RESERVED_0__MASK;
+}
+#define CORE_S_POINTER_EXECUTER__MASK				0x00010000
+#define CORE_S_POINTER_EXECUTER__SHIFT				16
+static inline uint32_t CORE_S_POINTER_EXECUTER(uint32_t val)
+{
+	return ((val) << CORE_S_POINTER_EXECUTER__SHIFT) & CORE_S_POINTER_EXECUTER__MASK;
+}
+#define CORE_S_POINTER_RESERVED_1__MASK				0x0000ffc0
+#define CORE_S_POINTER_RESERVED_1__SHIFT			6
+static inline uint32_t CORE_S_POINTER_RESERVED_1(uint32_t val)
+{
+	return ((val) << CORE_S_POINTER_RESERVED_1__SHIFT) & CORE_S_POINTER_RESERVED_1__MASK;
+}
+#define CORE_S_POINTER_EXECUTER_PP_CLEAR__MASK			0x00000020
+#define CORE_S_POINTER_EXECUTER_PP_CLEAR__SHIFT			5
+static inline uint32_t CORE_S_POINTER_EXECUTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << CORE_S_POINTER_EXECUTER_PP_CLEAR__SHIFT) & CORE_S_POINTER_EXECUTER_PP_CLEAR__MASK;
+}
+#define CORE_S_POINTER_POINTER_PP_CLEAR__MASK			0x00000010
+#define CORE_S_POINTER_POINTER_PP_CLEAR__SHIFT			4
+static inline uint32_t CORE_S_POINTER_POINTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << CORE_S_POINTER_POINTER_PP_CLEAR__SHIFT) & CORE_S_POINTER_POINTER_PP_CLEAR__MASK;
+}
+#define CORE_S_POINTER_POINTER_PP_MODE__MASK			0x00000008
+#define CORE_S_POINTER_POINTER_PP_MODE__SHIFT			3
+static inline uint32_t CORE_S_POINTER_POINTER_PP_MODE(uint32_t val)
+{
+	return ((val) << CORE_S_POINTER_POINTER_PP_MODE__SHIFT) & CORE_S_POINTER_POINTER_PP_MODE__MASK;
+}
+#define CORE_S_POINTER_EXECUTER_PP_EN__MASK			0x00000004
+#define CORE_S_POINTER_EXECUTER_PP_EN__SHIFT			2
+static inline uint32_t CORE_S_POINTER_EXECUTER_PP_EN(uint32_t val)
+{
+	return ((val) << CORE_S_POINTER_EXECUTER_PP_EN__SHIFT) & CORE_S_POINTER_EXECUTER_PP_EN__MASK;
+}
+#define CORE_S_POINTER_POINTER_PP_EN__MASK			0x00000002
+#define CORE_S_POINTER_POINTER_PP_EN__SHIFT			1
+static inline uint32_t CORE_S_POINTER_POINTER_PP_EN(uint32_t val)
+{
+	return ((val) << CORE_S_POINTER_POINTER_PP_EN__SHIFT) & CORE_S_POINTER_POINTER_PP_EN__MASK;
+}
+#define CORE_S_POINTER_POINTER__MASK				0x00000001
+#define CORE_S_POINTER_POINTER__SHIFT				0
+static inline uint32_t CORE_S_POINTER_POINTER(uint32_t val)
+{
+	return ((val) << CORE_S_POINTER_POINTER__SHIFT) & CORE_S_POINTER_POINTER__MASK;
+}
+
+#define REG_CORE_OPERATION_ENABLE				0x00003008
+#define CORE_OPERATION_ENABLE_RESERVED_0__MASK			0xfffffffe
+#define CORE_OPERATION_ENABLE_RESERVED_0__SHIFT			1
+static inline uint32_t CORE_OPERATION_ENABLE_RESERVED_0(uint32_t val)
+{
+	return ((val) << CORE_OPERATION_ENABLE_RESERVED_0__SHIFT) & CORE_OPERATION_ENABLE_RESERVED_0__MASK;
+}
+#define CORE_OPERATION_ENABLE_OP_EN__MASK			0x00000001
+#define CORE_OPERATION_ENABLE_OP_EN__SHIFT			0
+static inline uint32_t CORE_OPERATION_ENABLE_OP_EN(uint32_t val)
+{
+	return ((val) << CORE_OPERATION_ENABLE_OP_EN__SHIFT) & CORE_OPERATION_ENABLE_OP_EN__MASK;
+}
+
+#define REG_CORE_MAC_GATING					0x0000300c
+#define CORE_MAC_GATING_RESERVED_0__MASK			0xf8000000
+#define CORE_MAC_GATING_RESERVED_0__SHIFT			27
+static inline uint32_t CORE_MAC_GATING_RESERVED_0(uint32_t val)
+{
+	return ((val) << CORE_MAC_GATING_RESERVED_0__SHIFT) & CORE_MAC_GATING_RESERVED_0__MASK;
+}
+#define CORE_MAC_GATING_SLCG_OP_EN__MASK			0x07ffffff
+#define CORE_MAC_GATING_SLCG_OP_EN__SHIFT			0
+static inline uint32_t CORE_MAC_GATING_SLCG_OP_EN(uint32_t val)
+{
+	return ((val) << CORE_MAC_GATING_SLCG_OP_EN__SHIFT) & CORE_MAC_GATING_SLCG_OP_EN__MASK;
+}
+
+#define REG_CORE_MISC_CFG					0x00003010
+#define CORE_MISC_CFG_RESERVED_0__MASK				0xfff00000
+#define CORE_MISC_CFG_RESERVED_0__SHIFT				20
+static inline uint32_t CORE_MISC_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << CORE_MISC_CFG_RESERVED_0__SHIFT) & CORE_MISC_CFG_RESERVED_0__MASK;
+}
+#define CORE_MISC_CFG_SOFT_GATING__MASK				0x000fc000
+#define CORE_MISC_CFG_SOFT_GATING__SHIFT			14
+static inline uint32_t CORE_MISC_CFG_SOFT_GATING(uint32_t val)
+{
+	return ((val) << CORE_MISC_CFG_SOFT_GATING__SHIFT) & CORE_MISC_CFG_SOFT_GATING__MASK;
+}
+#define CORE_MISC_CFG_RESERVED_1__MASK				0x00003800
+#define CORE_MISC_CFG_RESERVED_1__SHIFT				11
+static inline uint32_t CORE_MISC_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << CORE_MISC_CFG_RESERVED_1__SHIFT) & CORE_MISC_CFG_RESERVED_1__MASK;
+}
+#define CORE_MISC_CFG_PROC_PRECISION__MASK			0x00000700
+#define CORE_MISC_CFG_PROC_PRECISION__SHIFT			8
+static inline uint32_t CORE_MISC_CFG_PROC_PRECISION(uint32_t val)
+{
+	return ((val) << CORE_MISC_CFG_PROC_PRECISION__SHIFT) & CORE_MISC_CFG_PROC_PRECISION__MASK;
+}
+#define CORE_MISC_CFG_RESERVED_2__MASK				0x000000fc
+#define CORE_MISC_CFG_RESERVED_2__SHIFT				2
+static inline uint32_t CORE_MISC_CFG_RESERVED_2(uint32_t val)
+{
+	return ((val) << CORE_MISC_CFG_RESERVED_2__SHIFT) & CORE_MISC_CFG_RESERVED_2__MASK;
+}
+#define CORE_MISC_CFG_DW_EN__MASK				0x00000002
+#define CORE_MISC_CFG_DW_EN__SHIFT				1
+static inline uint32_t CORE_MISC_CFG_DW_EN(uint32_t val)
+{
+	return ((val) << CORE_MISC_CFG_DW_EN__SHIFT) & CORE_MISC_CFG_DW_EN__MASK;
+}
+#define CORE_MISC_CFG_QD_EN__MASK				0x00000001
+#define CORE_MISC_CFG_QD_EN__SHIFT				0
+static inline uint32_t CORE_MISC_CFG_QD_EN(uint32_t val)
+{
+	return ((val) << CORE_MISC_CFG_QD_EN__SHIFT) & CORE_MISC_CFG_QD_EN__MASK;
+}
+
+#define REG_CORE_DATAOUT_SIZE_0					0x00003014
+#define CORE_DATAOUT_SIZE_0_DATAOUT_HEIGHT__MASK		0xffff0000
+#define CORE_DATAOUT_SIZE_0_DATAOUT_HEIGHT__SHIFT		16
+static inline uint32_t CORE_DATAOUT_SIZE_0_DATAOUT_HEIGHT(uint32_t val)
+{
+	return ((val) << CORE_DATAOUT_SIZE_0_DATAOUT_HEIGHT__SHIFT) & CORE_DATAOUT_SIZE_0_DATAOUT_HEIGHT__MASK;
+}
+#define CORE_DATAOUT_SIZE_0_DATAOUT_WIDTH__MASK			0x0000ffff
+#define CORE_DATAOUT_SIZE_0_DATAOUT_WIDTH__SHIFT		0
+static inline uint32_t CORE_DATAOUT_SIZE_0_DATAOUT_WIDTH(uint32_t val)
+{
+	return ((val) << CORE_DATAOUT_SIZE_0_DATAOUT_WIDTH__SHIFT) & CORE_DATAOUT_SIZE_0_DATAOUT_WIDTH__MASK;
+}
+
+#define REG_CORE_DATAOUT_SIZE_1					0x00003018
+#define CORE_DATAOUT_SIZE_1_RESERVED_0__MASK			0xffff0000
+#define CORE_DATAOUT_SIZE_1_RESERVED_0__SHIFT			16
+static inline uint32_t CORE_DATAOUT_SIZE_1_RESERVED_0(uint32_t val)
+{
+	return ((val) << CORE_DATAOUT_SIZE_1_RESERVED_0__SHIFT) & CORE_DATAOUT_SIZE_1_RESERVED_0__MASK;
+}
+#define CORE_DATAOUT_SIZE_1_DATAOUT_CHANNEL__MASK		0x0000ffff
+#define CORE_DATAOUT_SIZE_1_DATAOUT_CHANNEL__SHIFT		0
+static inline uint32_t CORE_DATAOUT_SIZE_1_DATAOUT_CHANNEL(uint32_t val)
+{
+	return ((val) << CORE_DATAOUT_SIZE_1_DATAOUT_CHANNEL__SHIFT) & CORE_DATAOUT_SIZE_1_DATAOUT_CHANNEL__MASK;
+}
+
+#define REG_CORE_CLIP_TRUNCATE					0x0000301c
+#define CORE_CLIP_TRUNCATE_RESERVED_0__MASK			0xffffff80
+#define CORE_CLIP_TRUNCATE_RESERVED_0__SHIFT			7
+static inline uint32_t CORE_CLIP_TRUNCATE_RESERVED_0(uint32_t val)
+{
+	return ((val) << CORE_CLIP_TRUNCATE_RESERVED_0__SHIFT) & CORE_CLIP_TRUNCATE_RESERVED_0__MASK;
+}
+#define CORE_CLIP_TRUNCATE_ROUND_TYPE__MASK			0x00000040
+#define CORE_CLIP_TRUNCATE_ROUND_TYPE__SHIFT			6
+static inline uint32_t CORE_CLIP_TRUNCATE_ROUND_TYPE(uint32_t val)
+{
+	return ((val) << CORE_CLIP_TRUNCATE_ROUND_TYPE__SHIFT) & CORE_CLIP_TRUNCATE_ROUND_TYPE__MASK;
+}
+#define CORE_CLIP_TRUNCATE_RESERVED_1__MASK			0x00000020
+#define CORE_CLIP_TRUNCATE_RESERVED_1__SHIFT			5
+static inline uint32_t CORE_CLIP_TRUNCATE_RESERVED_1(uint32_t val)
+{
+	return ((val) << CORE_CLIP_TRUNCATE_RESERVED_1__SHIFT) & CORE_CLIP_TRUNCATE_RESERVED_1__MASK;
+}
+#define CORE_CLIP_TRUNCATE_CLIP_TRUNCATE__MASK			0x0000001f
+#define CORE_CLIP_TRUNCATE_CLIP_TRUNCATE__SHIFT			0
+static inline uint32_t CORE_CLIP_TRUNCATE_CLIP_TRUNCATE(uint32_t val)
+{
+	return ((val) << CORE_CLIP_TRUNCATE_CLIP_TRUNCATE__SHIFT) & CORE_CLIP_TRUNCATE_CLIP_TRUNCATE__MASK;
+}
+
+#define REG_DPU_S_STATUS					0x00004000
+#define DPU_S_STATUS_RESERVED_0__MASK				0xfffc0000
+#define DPU_S_STATUS_RESERVED_0__SHIFT				18
+static inline uint32_t DPU_S_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_S_STATUS_RESERVED_0__SHIFT) & DPU_S_STATUS_RESERVED_0__MASK;
+}
+#define DPU_S_STATUS_STATUS_1__MASK				0x00030000
+#define DPU_S_STATUS_STATUS_1__SHIFT				16
+static inline uint32_t DPU_S_STATUS_STATUS_1(uint32_t val)
+{
+	return ((val) << DPU_S_STATUS_STATUS_1__SHIFT) & DPU_S_STATUS_STATUS_1__MASK;
+}
+#define DPU_S_STATUS_RESERVED_1__MASK				0x0000fffc
+#define DPU_S_STATUS_RESERVED_1__SHIFT				2
+static inline uint32_t DPU_S_STATUS_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_S_STATUS_RESERVED_1__SHIFT) & DPU_S_STATUS_RESERVED_1__MASK;
+}
+#define DPU_S_STATUS_STATUS_0__MASK				0x00000003
+#define DPU_S_STATUS_STATUS_0__SHIFT				0
+static inline uint32_t DPU_S_STATUS_STATUS_0(uint32_t val)
+{
+	return ((val) << DPU_S_STATUS_STATUS_0__SHIFT) & DPU_S_STATUS_STATUS_0__MASK;
+}
+
+#define REG_DPU_S_POINTER					0x00004004
+#define DPU_S_POINTER_RESERVED_0__MASK				0xfffe0000
+#define DPU_S_POINTER_RESERVED_0__SHIFT				17
+static inline uint32_t DPU_S_POINTER_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_S_POINTER_RESERVED_0__SHIFT) & DPU_S_POINTER_RESERVED_0__MASK;
+}
+#define DPU_S_POINTER_EXECUTER__MASK				0x00010000
+#define DPU_S_POINTER_EXECUTER__SHIFT				16
+static inline uint32_t DPU_S_POINTER_EXECUTER(uint32_t val)
+{
+	return ((val) << DPU_S_POINTER_EXECUTER__SHIFT) & DPU_S_POINTER_EXECUTER__MASK;
+}
+#define DPU_S_POINTER_RESERVED_1__MASK				0x0000ffc0
+#define DPU_S_POINTER_RESERVED_1__SHIFT				6
+static inline uint32_t DPU_S_POINTER_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_S_POINTER_RESERVED_1__SHIFT) & DPU_S_POINTER_RESERVED_1__MASK;
+}
+#define DPU_S_POINTER_EXECUTER_PP_CLEAR__MASK			0x00000020
+#define DPU_S_POINTER_EXECUTER_PP_CLEAR__SHIFT			5
+static inline uint32_t DPU_S_POINTER_EXECUTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << DPU_S_POINTER_EXECUTER_PP_CLEAR__SHIFT) & DPU_S_POINTER_EXECUTER_PP_CLEAR__MASK;
+}
+#define DPU_S_POINTER_POINTER_PP_CLEAR__MASK			0x00000010
+#define DPU_S_POINTER_POINTER_PP_CLEAR__SHIFT			4
+static inline uint32_t DPU_S_POINTER_POINTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << DPU_S_POINTER_POINTER_PP_CLEAR__SHIFT) & DPU_S_POINTER_POINTER_PP_CLEAR__MASK;
+}
+#define DPU_S_POINTER_POINTER_PP_MODE__MASK			0x00000008
+#define DPU_S_POINTER_POINTER_PP_MODE__SHIFT			3
+static inline uint32_t DPU_S_POINTER_POINTER_PP_MODE(uint32_t val)
+{
+	return ((val) << DPU_S_POINTER_POINTER_PP_MODE__SHIFT) & DPU_S_POINTER_POINTER_PP_MODE__MASK;
+}
+#define DPU_S_POINTER_EXECUTER_PP_EN__MASK			0x00000004
+#define DPU_S_POINTER_EXECUTER_PP_EN__SHIFT			2
+static inline uint32_t DPU_S_POINTER_EXECUTER_PP_EN(uint32_t val)
+{
+	return ((val) << DPU_S_POINTER_EXECUTER_PP_EN__SHIFT) & DPU_S_POINTER_EXECUTER_PP_EN__MASK;
+}
+#define DPU_S_POINTER_POINTER_PP_EN__MASK			0x00000002
+#define DPU_S_POINTER_POINTER_PP_EN__SHIFT			1
+static inline uint32_t DPU_S_POINTER_POINTER_PP_EN(uint32_t val)
+{
+	return ((val) << DPU_S_POINTER_POINTER_PP_EN__SHIFT) & DPU_S_POINTER_POINTER_PP_EN__MASK;
+}
+#define DPU_S_POINTER_POINTER__MASK				0x00000001
+#define DPU_S_POINTER_POINTER__SHIFT				0
+static inline uint32_t DPU_S_POINTER_POINTER(uint32_t val)
+{
+	return ((val) << DPU_S_POINTER_POINTER__SHIFT) & DPU_S_POINTER_POINTER__MASK;
+}
+
+#define REG_DPU_OPERATION_ENABLE				0x00004008
+#define DPU_OPERATION_ENABLE_RESERVED_0__MASK			0xfffffffe
+#define DPU_OPERATION_ENABLE_RESERVED_0__SHIFT			1
+static inline uint32_t DPU_OPERATION_ENABLE_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_OPERATION_ENABLE_RESERVED_0__SHIFT) & DPU_OPERATION_ENABLE_RESERVED_0__MASK;
+}
+#define DPU_OPERATION_ENABLE_OP_EN__MASK			0x00000001
+#define DPU_OPERATION_ENABLE_OP_EN__SHIFT			0
+static inline uint32_t DPU_OPERATION_ENABLE_OP_EN(uint32_t val)
+{
+	return ((val) << DPU_OPERATION_ENABLE_OP_EN__SHIFT) & DPU_OPERATION_ENABLE_OP_EN__MASK;
+}
+
+#define REG_DPU_FEATURE_MODE_CFG				0x0000400c
+#define DPU_FEATURE_MODE_CFG_COMB_USE__MASK			0x80000000
+#define DPU_FEATURE_MODE_CFG_COMB_USE__SHIFT			31
+static inline uint32_t DPU_FEATURE_MODE_CFG_COMB_USE(uint32_t val)
+{
+	return ((val) << DPU_FEATURE_MODE_CFG_COMB_USE__SHIFT) & DPU_FEATURE_MODE_CFG_COMB_USE__MASK;
+}
+#define DPU_FEATURE_MODE_CFG_TP_EN__MASK			0x40000000
+#define DPU_FEATURE_MODE_CFG_TP_EN__SHIFT			30
+static inline uint32_t DPU_FEATURE_MODE_CFG_TP_EN(uint32_t val)
+{
+	return ((val) << DPU_FEATURE_MODE_CFG_TP_EN__SHIFT) & DPU_FEATURE_MODE_CFG_TP_EN__MASK;
+}
+#define DPU_FEATURE_MODE_CFG_RGP_TYPE__MASK			0x3c000000
+#define DPU_FEATURE_MODE_CFG_RGP_TYPE__SHIFT			26
+static inline uint32_t DPU_FEATURE_MODE_CFG_RGP_TYPE(uint32_t val)
+{
+	return ((val) << DPU_FEATURE_MODE_CFG_RGP_TYPE__SHIFT) & DPU_FEATURE_MODE_CFG_RGP_TYPE__MASK;
+}
+#define DPU_FEATURE_MODE_CFG_NONALIGN__MASK			0x02000000
+#define DPU_FEATURE_MODE_CFG_NONALIGN__SHIFT			25
+static inline uint32_t DPU_FEATURE_MODE_CFG_NONALIGN(uint32_t val)
+{
+	return ((val) << DPU_FEATURE_MODE_CFG_NONALIGN__SHIFT) & DPU_FEATURE_MODE_CFG_NONALIGN__MASK;
+}
+#define DPU_FEATURE_MODE_CFG_SURF_LEN__MASK			0x01fffe00
+#define DPU_FEATURE_MODE_CFG_SURF_LEN__SHIFT			9
+static inline uint32_t DPU_FEATURE_MODE_CFG_SURF_LEN(uint32_t val)
+{
+	return ((val) << DPU_FEATURE_MODE_CFG_SURF_LEN__SHIFT) & DPU_FEATURE_MODE_CFG_SURF_LEN__MASK;
+}
+#define DPU_FEATURE_MODE_CFG_BURST_LEN__MASK			0x000001e0
+#define DPU_FEATURE_MODE_CFG_BURST_LEN__SHIFT			5
+static inline uint32_t DPU_FEATURE_MODE_CFG_BURST_LEN(uint32_t val)
+{
+	return ((val) << DPU_FEATURE_MODE_CFG_BURST_LEN__SHIFT) & DPU_FEATURE_MODE_CFG_BURST_LEN__MASK;
+}
+#define DPU_FEATURE_MODE_CFG_CONV_MODE__MASK			0x00000018
+#define DPU_FEATURE_MODE_CFG_CONV_MODE__SHIFT			3
+static inline uint32_t DPU_FEATURE_MODE_CFG_CONV_MODE(uint32_t val)
+{
+	return ((val) << DPU_FEATURE_MODE_CFG_CONV_MODE__SHIFT) & DPU_FEATURE_MODE_CFG_CONV_MODE__MASK;
+}
+#define DPU_FEATURE_MODE_CFG_OUTPUT_MODE__MASK			0x00000006
+#define DPU_FEATURE_MODE_CFG_OUTPUT_MODE__SHIFT			1
+static inline uint32_t DPU_FEATURE_MODE_CFG_OUTPUT_MODE(uint32_t val)
+{
+	return ((val) << DPU_FEATURE_MODE_CFG_OUTPUT_MODE__SHIFT) & DPU_FEATURE_MODE_CFG_OUTPUT_MODE__MASK;
+}
+#define DPU_FEATURE_MODE_CFG_FLYING_MODE__MASK			0x00000001
+#define DPU_FEATURE_MODE_CFG_FLYING_MODE__SHIFT			0
+static inline uint32_t DPU_FEATURE_MODE_CFG_FLYING_MODE(uint32_t val)
+{
+	return ((val) << DPU_FEATURE_MODE_CFG_FLYING_MODE__SHIFT) & DPU_FEATURE_MODE_CFG_FLYING_MODE__MASK;
+}
+
+#define REG_DPU_DATA_FORMAT					0x00004010
+#define DPU_DATA_FORMAT_OUT_PRECISION__MASK			0xe0000000
+#define DPU_DATA_FORMAT_OUT_PRECISION__SHIFT			29
+static inline uint32_t DPU_DATA_FORMAT_OUT_PRECISION(uint32_t val)
+{
+	return ((val) << DPU_DATA_FORMAT_OUT_PRECISION__SHIFT) & DPU_DATA_FORMAT_OUT_PRECISION__MASK;
+}
+#define DPU_DATA_FORMAT_IN_PRECISION__MASK			0x1c000000
+#define DPU_DATA_FORMAT_IN_PRECISION__SHIFT			26
+static inline uint32_t DPU_DATA_FORMAT_IN_PRECISION(uint32_t val)
+{
+	return ((val) << DPU_DATA_FORMAT_IN_PRECISION__SHIFT) & DPU_DATA_FORMAT_IN_PRECISION__MASK;
+}
+#define DPU_DATA_FORMAT_EW_TRUNCATE_NEG__MASK			0x03ff0000
+#define DPU_DATA_FORMAT_EW_TRUNCATE_NEG__SHIFT			16
+static inline uint32_t DPU_DATA_FORMAT_EW_TRUNCATE_NEG(uint32_t val)
+{
+	return ((val) << DPU_DATA_FORMAT_EW_TRUNCATE_NEG__SHIFT) & DPU_DATA_FORMAT_EW_TRUNCATE_NEG__MASK;
+}
+#define DPU_DATA_FORMAT_BN_MUL_SHIFT_VALUE_NEG__MASK		0x0000fc00
+#define DPU_DATA_FORMAT_BN_MUL_SHIFT_VALUE_NEG__SHIFT		10
+static inline uint32_t DPU_DATA_FORMAT_BN_MUL_SHIFT_VALUE_NEG(uint32_t val)
+{
+	return ((val) << DPU_DATA_FORMAT_BN_MUL_SHIFT_VALUE_NEG__SHIFT) & DPU_DATA_FORMAT_BN_MUL_SHIFT_VALUE_NEG__MASK;
+}
+#define DPU_DATA_FORMAT_BS_MUL_SHIFT_VALUE_NEG__MASK		0x000003f0
+#define DPU_DATA_FORMAT_BS_MUL_SHIFT_VALUE_NEG__SHIFT		4
+static inline uint32_t DPU_DATA_FORMAT_BS_MUL_SHIFT_VALUE_NEG(uint32_t val)
+{
+	return ((val) << DPU_DATA_FORMAT_BS_MUL_SHIFT_VALUE_NEG__SHIFT) & DPU_DATA_FORMAT_BS_MUL_SHIFT_VALUE_NEG__MASK;
+}
+#define DPU_DATA_FORMAT_MC_SURF_OUT__MASK			0x00000008
+#define DPU_DATA_FORMAT_MC_SURF_OUT__SHIFT			3
+static inline uint32_t DPU_DATA_FORMAT_MC_SURF_OUT(uint32_t val)
+{
+	return ((val) << DPU_DATA_FORMAT_MC_SURF_OUT__SHIFT) & DPU_DATA_FORMAT_MC_SURF_OUT__MASK;
+}
+#define DPU_DATA_FORMAT_PROC_PRECISION__MASK			0x00000007
+#define DPU_DATA_FORMAT_PROC_PRECISION__SHIFT			0
+static inline uint32_t DPU_DATA_FORMAT_PROC_PRECISION(uint32_t val)
+{
+	return ((val) << DPU_DATA_FORMAT_PROC_PRECISION__SHIFT) & DPU_DATA_FORMAT_PROC_PRECISION__MASK;
+}
+
+#define REG_DPU_OFFSET_PEND					0x00004014
+#define DPU_OFFSET_PEND_RESERVED_0__MASK			0xffff0000
+#define DPU_OFFSET_PEND_RESERVED_0__SHIFT			16
+static inline uint32_t DPU_OFFSET_PEND_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_OFFSET_PEND_RESERVED_0__SHIFT) & DPU_OFFSET_PEND_RESERVED_0__MASK;
+}
+#define DPU_OFFSET_PEND_OFFSET_PEND__MASK			0x0000ffff
+#define DPU_OFFSET_PEND_OFFSET_PEND__SHIFT			0
+static inline uint32_t DPU_OFFSET_PEND_OFFSET_PEND(uint32_t val)
+{
+	return ((val) << DPU_OFFSET_PEND_OFFSET_PEND__SHIFT) & DPU_OFFSET_PEND_OFFSET_PEND__MASK;
+}
+
+#define REG_DPU_DST_BASE_ADDR					0x00004020
+#define DPU_DST_BASE_ADDR_DST_BASE_ADDR__MASK			0xffffffff
+#define DPU_DST_BASE_ADDR_DST_BASE_ADDR__SHIFT			0
+static inline uint32_t DPU_DST_BASE_ADDR_DST_BASE_ADDR(uint32_t val)
+{
+	return ((val) << DPU_DST_BASE_ADDR_DST_BASE_ADDR__SHIFT) & DPU_DST_BASE_ADDR_DST_BASE_ADDR__MASK;
+}
+
+#define REG_DPU_DST_SURF_STRIDE					0x00004024
+#define DPU_DST_SURF_STRIDE_DST_SURF_STRIDE__MASK		0xfffffff0
+#define DPU_DST_SURF_STRIDE_DST_SURF_STRIDE__SHIFT		4
+static inline uint32_t DPU_DST_SURF_STRIDE_DST_SURF_STRIDE(uint32_t val)
+{
+	return ((val) << DPU_DST_SURF_STRIDE_DST_SURF_STRIDE__SHIFT) & DPU_DST_SURF_STRIDE_DST_SURF_STRIDE__MASK;
+}
+#define DPU_DST_SURF_STRIDE_RESERVED_0__MASK			0x0000000f
+#define DPU_DST_SURF_STRIDE_RESERVED_0__SHIFT			0
+static inline uint32_t DPU_DST_SURF_STRIDE_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_DST_SURF_STRIDE_RESERVED_0__SHIFT) & DPU_DST_SURF_STRIDE_RESERVED_0__MASK;
+}
+
+#define REG_DPU_DATA_CUBE_WIDTH					0x00004030
+#define DPU_DATA_CUBE_WIDTH_RESERVED_0__MASK			0xffffe000
+#define DPU_DATA_CUBE_WIDTH_RESERVED_0__SHIFT			13
+static inline uint32_t DPU_DATA_CUBE_WIDTH_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_WIDTH_RESERVED_0__SHIFT) & DPU_DATA_CUBE_WIDTH_RESERVED_0__MASK;
+}
+#define DPU_DATA_CUBE_WIDTH_WIDTH__MASK				0x00001fff
+#define DPU_DATA_CUBE_WIDTH_WIDTH__SHIFT			0
+static inline uint32_t DPU_DATA_CUBE_WIDTH_WIDTH(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_WIDTH_WIDTH__SHIFT) & DPU_DATA_CUBE_WIDTH_WIDTH__MASK;
+}
+
+#define REG_DPU_DATA_CUBE_HEIGHT				0x00004034
+#define DPU_DATA_CUBE_HEIGHT_RESERVED_0__MASK			0xfe000000
+#define DPU_DATA_CUBE_HEIGHT_RESERVED_0__SHIFT			25
+static inline uint32_t DPU_DATA_CUBE_HEIGHT_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_HEIGHT_RESERVED_0__SHIFT) & DPU_DATA_CUBE_HEIGHT_RESERVED_0__MASK;
+}
+#define DPU_DATA_CUBE_HEIGHT_MINMAX_CTL__MASK			0x01c00000
+#define DPU_DATA_CUBE_HEIGHT_MINMAX_CTL__SHIFT			22
+static inline uint32_t DPU_DATA_CUBE_HEIGHT_MINMAX_CTL(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_HEIGHT_MINMAX_CTL__SHIFT) & DPU_DATA_CUBE_HEIGHT_MINMAX_CTL__MASK;
+}
+#define DPU_DATA_CUBE_HEIGHT_RESERVED_1__MASK			0x003fe000
+#define DPU_DATA_CUBE_HEIGHT_RESERVED_1__SHIFT			13
+static inline uint32_t DPU_DATA_CUBE_HEIGHT_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_HEIGHT_RESERVED_1__SHIFT) & DPU_DATA_CUBE_HEIGHT_RESERVED_1__MASK;
+}
+#define DPU_DATA_CUBE_HEIGHT_HEIGHT__MASK			0x00001fff
+#define DPU_DATA_CUBE_HEIGHT_HEIGHT__SHIFT			0
+static inline uint32_t DPU_DATA_CUBE_HEIGHT_HEIGHT(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_HEIGHT_HEIGHT__SHIFT) & DPU_DATA_CUBE_HEIGHT_HEIGHT__MASK;
+}
+
+#define REG_DPU_DATA_CUBE_NOTCH_ADDR				0x00004038
+#define DPU_DATA_CUBE_NOTCH_ADDR_RESERVED_0__MASK		0xe0000000
+#define DPU_DATA_CUBE_NOTCH_ADDR_RESERVED_0__SHIFT		29
+static inline uint32_t DPU_DATA_CUBE_NOTCH_ADDR_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_NOTCH_ADDR_RESERVED_0__SHIFT) & DPU_DATA_CUBE_NOTCH_ADDR_RESERVED_0__MASK;
+}
+#define DPU_DATA_CUBE_NOTCH_ADDR_NOTCH_ADDR_1__MASK		0x1fff0000
+#define DPU_DATA_CUBE_NOTCH_ADDR_NOTCH_ADDR_1__SHIFT		16
+static inline uint32_t DPU_DATA_CUBE_NOTCH_ADDR_NOTCH_ADDR_1(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_NOTCH_ADDR_NOTCH_ADDR_1__SHIFT) & DPU_DATA_CUBE_NOTCH_ADDR_NOTCH_ADDR_1__MASK;
+}
+#define DPU_DATA_CUBE_NOTCH_ADDR_RESERVED_1__MASK		0x0000e000
+#define DPU_DATA_CUBE_NOTCH_ADDR_RESERVED_1__SHIFT		13
+static inline uint32_t DPU_DATA_CUBE_NOTCH_ADDR_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_NOTCH_ADDR_RESERVED_1__SHIFT) & DPU_DATA_CUBE_NOTCH_ADDR_RESERVED_1__MASK;
+}
+#define DPU_DATA_CUBE_NOTCH_ADDR_NOTCH_ADDR_0__MASK		0x00001fff
+#define DPU_DATA_CUBE_NOTCH_ADDR_NOTCH_ADDR_0__SHIFT		0
+static inline uint32_t DPU_DATA_CUBE_NOTCH_ADDR_NOTCH_ADDR_0(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_NOTCH_ADDR_NOTCH_ADDR_0__SHIFT) & DPU_DATA_CUBE_NOTCH_ADDR_NOTCH_ADDR_0__MASK;
+}
+
+#define REG_DPU_DATA_CUBE_CHANNEL				0x0000403c
+#define DPU_DATA_CUBE_CHANNEL_RESERVED_0__MASK			0xe0000000
+#define DPU_DATA_CUBE_CHANNEL_RESERVED_0__SHIFT			29
+static inline uint32_t DPU_DATA_CUBE_CHANNEL_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_CHANNEL_RESERVED_0__SHIFT) & DPU_DATA_CUBE_CHANNEL_RESERVED_0__MASK;
+}
+#define DPU_DATA_CUBE_CHANNEL_ORIG_CHANNEL__MASK		0x1fff0000
+#define DPU_DATA_CUBE_CHANNEL_ORIG_CHANNEL__SHIFT		16
+static inline uint32_t DPU_DATA_CUBE_CHANNEL_ORIG_CHANNEL(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_CHANNEL_ORIG_CHANNEL__SHIFT) & DPU_DATA_CUBE_CHANNEL_ORIG_CHANNEL__MASK;
+}
+#define DPU_DATA_CUBE_CHANNEL_RESERVED_1__MASK			0x0000e000
+#define DPU_DATA_CUBE_CHANNEL_RESERVED_1__SHIFT			13
+static inline uint32_t DPU_DATA_CUBE_CHANNEL_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_CHANNEL_RESERVED_1__SHIFT) & DPU_DATA_CUBE_CHANNEL_RESERVED_1__MASK;
+}
+#define DPU_DATA_CUBE_CHANNEL_CHANNEL__MASK			0x00001fff
+#define DPU_DATA_CUBE_CHANNEL_CHANNEL__SHIFT			0
+static inline uint32_t DPU_DATA_CUBE_CHANNEL_CHANNEL(uint32_t val)
+{
+	return ((val) << DPU_DATA_CUBE_CHANNEL_CHANNEL__SHIFT) & DPU_DATA_CUBE_CHANNEL_CHANNEL__MASK;
+}
+
+#define REG_DPU_BS_CFG						0x00004040
+#define DPU_BS_CFG_RESERVED_0__MASK				0xfff00000
+#define DPU_BS_CFG_RESERVED_0__SHIFT				20
+static inline uint32_t DPU_BS_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_RESERVED_0__SHIFT) & DPU_BS_CFG_RESERVED_0__MASK;
+}
+#define DPU_BS_CFG_BS_ALU_ALGO__MASK				0x000f0000
+#define DPU_BS_CFG_BS_ALU_ALGO__SHIFT				16
+static inline uint32_t DPU_BS_CFG_BS_ALU_ALGO(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_BS_ALU_ALGO__SHIFT) & DPU_BS_CFG_BS_ALU_ALGO__MASK;
+}
+#define DPU_BS_CFG_RESERVED_1__MASK				0x0000fe00
+#define DPU_BS_CFG_RESERVED_1__SHIFT				9
+static inline uint32_t DPU_BS_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_RESERVED_1__SHIFT) & DPU_BS_CFG_RESERVED_1__MASK;
+}
+#define DPU_BS_CFG_BS_ALU_SRC__MASK				0x00000100
+#define DPU_BS_CFG_BS_ALU_SRC__SHIFT				8
+static inline uint32_t DPU_BS_CFG_BS_ALU_SRC(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_BS_ALU_SRC__SHIFT) & DPU_BS_CFG_BS_ALU_SRC__MASK;
+}
+#define DPU_BS_CFG_BS_RELUX_EN__MASK				0x00000080
+#define DPU_BS_CFG_BS_RELUX_EN__SHIFT				7
+static inline uint32_t DPU_BS_CFG_BS_RELUX_EN(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_BS_RELUX_EN__SHIFT) & DPU_BS_CFG_BS_RELUX_EN__MASK;
+}
+#define DPU_BS_CFG_BS_RELU_BYPASS__MASK				0x00000040
+#define DPU_BS_CFG_BS_RELU_BYPASS__SHIFT			6
+static inline uint32_t DPU_BS_CFG_BS_RELU_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_BS_RELU_BYPASS__SHIFT) & DPU_BS_CFG_BS_RELU_BYPASS__MASK;
+}
+#define DPU_BS_CFG_BS_MUL_PRELU__MASK				0x00000020
+#define DPU_BS_CFG_BS_MUL_PRELU__SHIFT				5
+static inline uint32_t DPU_BS_CFG_BS_MUL_PRELU(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_BS_MUL_PRELU__SHIFT) & DPU_BS_CFG_BS_MUL_PRELU__MASK;
+}
+#define DPU_BS_CFG_BS_MUL_BYPASS__MASK				0x00000010
+#define DPU_BS_CFG_BS_MUL_BYPASS__SHIFT				4
+static inline uint32_t DPU_BS_CFG_BS_MUL_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_BS_MUL_BYPASS__SHIFT) & DPU_BS_CFG_BS_MUL_BYPASS__MASK;
+}
+#define DPU_BS_CFG_RESERVED_2__MASK				0x0000000c
+#define DPU_BS_CFG_RESERVED_2__SHIFT				2
+static inline uint32_t DPU_BS_CFG_RESERVED_2(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_RESERVED_2__SHIFT) & DPU_BS_CFG_RESERVED_2__MASK;
+}
+#define DPU_BS_CFG_BS_ALU_BYPASS__MASK				0x00000002
+#define DPU_BS_CFG_BS_ALU_BYPASS__SHIFT				1
+static inline uint32_t DPU_BS_CFG_BS_ALU_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_BS_ALU_BYPASS__SHIFT) & DPU_BS_CFG_BS_ALU_BYPASS__MASK;
+}
+#define DPU_BS_CFG_BS_BYPASS__MASK				0x00000001
+#define DPU_BS_CFG_BS_BYPASS__SHIFT				0
+static inline uint32_t DPU_BS_CFG_BS_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_BS_CFG_BS_BYPASS__SHIFT) & DPU_BS_CFG_BS_BYPASS__MASK;
+}
+
+#define REG_DPU_BS_ALU_CFG					0x00004044
+#define DPU_BS_ALU_CFG_BS_ALU_OPERAND__MASK			0xffffffff
+#define DPU_BS_ALU_CFG_BS_ALU_OPERAND__SHIFT			0
+static inline uint32_t DPU_BS_ALU_CFG_BS_ALU_OPERAND(uint32_t val)
+{
+	return ((val) << DPU_BS_ALU_CFG_BS_ALU_OPERAND__SHIFT) & DPU_BS_ALU_CFG_BS_ALU_OPERAND__MASK;
+}
+
+#define REG_DPU_BS_MUL_CFG					0x00004048
+#define DPU_BS_MUL_CFG_BS_MUL_OPERAND__MASK			0xffff0000
+#define DPU_BS_MUL_CFG_BS_MUL_OPERAND__SHIFT			16
+static inline uint32_t DPU_BS_MUL_CFG_BS_MUL_OPERAND(uint32_t val)
+{
+	return ((val) << DPU_BS_MUL_CFG_BS_MUL_OPERAND__SHIFT) & DPU_BS_MUL_CFG_BS_MUL_OPERAND__MASK;
+}
+#define DPU_BS_MUL_CFG_RESERVED_0__MASK				0x0000c000
+#define DPU_BS_MUL_CFG_RESERVED_0__SHIFT			14
+static inline uint32_t DPU_BS_MUL_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_BS_MUL_CFG_RESERVED_0__SHIFT) & DPU_BS_MUL_CFG_RESERVED_0__MASK;
+}
+#define DPU_BS_MUL_CFG_BS_MUL_SHIFT_VALUE__MASK			0x00003f00
+#define DPU_BS_MUL_CFG_BS_MUL_SHIFT_VALUE__SHIFT		8
+static inline uint32_t DPU_BS_MUL_CFG_BS_MUL_SHIFT_VALUE(uint32_t val)
+{
+	return ((val) << DPU_BS_MUL_CFG_BS_MUL_SHIFT_VALUE__SHIFT) & DPU_BS_MUL_CFG_BS_MUL_SHIFT_VALUE__MASK;
+}
+#define DPU_BS_MUL_CFG_RESERVED_1__MASK				0x000000fc
+#define DPU_BS_MUL_CFG_RESERVED_1__SHIFT			2
+static inline uint32_t DPU_BS_MUL_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_BS_MUL_CFG_RESERVED_1__SHIFT) & DPU_BS_MUL_CFG_RESERVED_1__MASK;
+}
+#define DPU_BS_MUL_CFG_BS_TRUNCATE_SRC__MASK			0x00000002
+#define DPU_BS_MUL_CFG_BS_TRUNCATE_SRC__SHIFT			1
+static inline uint32_t DPU_BS_MUL_CFG_BS_TRUNCATE_SRC(uint32_t val)
+{
+	return ((val) << DPU_BS_MUL_CFG_BS_TRUNCATE_SRC__SHIFT) & DPU_BS_MUL_CFG_BS_TRUNCATE_SRC__MASK;
+}
+#define DPU_BS_MUL_CFG_BS_MUL_SRC__MASK				0x00000001
+#define DPU_BS_MUL_CFG_BS_MUL_SRC__SHIFT			0
+static inline uint32_t DPU_BS_MUL_CFG_BS_MUL_SRC(uint32_t val)
+{
+	return ((val) << DPU_BS_MUL_CFG_BS_MUL_SRC__SHIFT) & DPU_BS_MUL_CFG_BS_MUL_SRC__MASK;
+}
+
+#define REG_DPU_BS_RELUX_CMP_VALUE				0x0000404c
+#define DPU_BS_RELUX_CMP_VALUE_BS_RELUX_CMP_DAT__MASK		0xffffffff
+#define DPU_BS_RELUX_CMP_VALUE_BS_RELUX_CMP_DAT__SHIFT		0
+static inline uint32_t DPU_BS_RELUX_CMP_VALUE_BS_RELUX_CMP_DAT(uint32_t val)
+{
+	return ((val) << DPU_BS_RELUX_CMP_VALUE_BS_RELUX_CMP_DAT__SHIFT) & DPU_BS_RELUX_CMP_VALUE_BS_RELUX_CMP_DAT__MASK;
+}
+
+#define REG_DPU_BS_OW_CFG					0x00004050
+#define DPU_BS_OW_CFG_RGP_CNTER__MASK				0xf0000000
+#define DPU_BS_OW_CFG_RGP_CNTER__SHIFT				28
+static inline uint32_t DPU_BS_OW_CFG_RGP_CNTER(uint32_t val)
+{
+	return ((val) << DPU_BS_OW_CFG_RGP_CNTER__SHIFT) & DPU_BS_OW_CFG_RGP_CNTER__MASK;
+}
+#define DPU_BS_OW_CFG_TP_ORG_EN__MASK				0x08000000
+#define DPU_BS_OW_CFG_TP_ORG_EN__SHIFT				27
+static inline uint32_t DPU_BS_OW_CFG_TP_ORG_EN(uint32_t val)
+{
+	return ((val) << DPU_BS_OW_CFG_TP_ORG_EN__SHIFT) & DPU_BS_OW_CFG_TP_ORG_EN__MASK;
+}
+#define DPU_BS_OW_CFG_RESERVED_0__MASK				0x07fff800
+#define DPU_BS_OW_CFG_RESERVED_0__SHIFT				11
+static inline uint32_t DPU_BS_OW_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_BS_OW_CFG_RESERVED_0__SHIFT) & DPU_BS_OW_CFG_RESERVED_0__MASK;
+}
+#define DPU_BS_OW_CFG_SIZE_E_2__MASK				0x00000700
+#define DPU_BS_OW_CFG_SIZE_E_2__SHIFT				8
+static inline uint32_t DPU_BS_OW_CFG_SIZE_E_2(uint32_t val)
+{
+	return ((val) << DPU_BS_OW_CFG_SIZE_E_2__SHIFT) & DPU_BS_OW_CFG_SIZE_E_2__MASK;
+}
+#define DPU_BS_OW_CFG_SIZE_E_1__MASK				0x000000e0
+#define DPU_BS_OW_CFG_SIZE_E_1__SHIFT				5
+static inline uint32_t DPU_BS_OW_CFG_SIZE_E_1(uint32_t val)
+{
+	return ((val) << DPU_BS_OW_CFG_SIZE_E_1__SHIFT) & DPU_BS_OW_CFG_SIZE_E_1__MASK;
+}
+#define DPU_BS_OW_CFG_SIZE_E_0__MASK				0x0000001c
+#define DPU_BS_OW_CFG_SIZE_E_0__SHIFT				2
+static inline uint32_t DPU_BS_OW_CFG_SIZE_E_0(uint32_t val)
+{
+	return ((val) << DPU_BS_OW_CFG_SIZE_E_0__SHIFT) & DPU_BS_OW_CFG_SIZE_E_0__MASK;
+}
+#define DPU_BS_OW_CFG_OD_BYPASS__MASK				0x00000002
+#define DPU_BS_OW_CFG_OD_BYPASS__SHIFT				1
+static inline uint32_t DPU_BS_OW_CFG_OD_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_BS_OW_CFG_OD_BYPASS__SHIFT) & DPU_BS_OW_CFG_OD_BYPASS__MASK;
+}
+#define DPU_BS_OW_CFG_OW_SRC__MASK				0x00000001
+#define DPU_BS_OW_CFG_OW_SRC__SHIFT				0
+static inline uint32_t DPU_BS_OW_CFG_OW_SRC(uint32_t val)
+{
+	return ((val) << DPU_BS_OW_CFG_OW_SRC__SHIFT) & DPU_BS_OW_CFG_OW_SRC__MASK;
+}
+
+#define REG_DPU_BS_OW_OP					0x00004054
+#define DPU_BS_OW_OP_RESERVED_0__MASK				0xffff0000
+#define DPU_BS_OW_OP_RESERVED_0__SHIFT				16
+static inline uint32_t DPU_BS_OW_OP_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_BS_OW_OP_RESERVED_0__SHIFT) & DPU_BS_OW_OP_RESERVED_0__MASK;
+}
+#define DPU_BS_OW_OP_OW_OP__MASK				0x0000ffff
+#define DPU_BS_OW_OP_OW_OP__SHIFT				0
+static inline uint32_t DPU_BS_OW_OP_OW_OP(uint32_t val)
+{
+	return ((val) << DPU_BS_OW_OP_OW_OP__SHIFT) & DPU_BS_OW_OP_OW_OP__MASK;
+}
+
+#define REG_DPU_WDMA_SIZE_0					0x00004058
+#define DPU_WDMA_SIZE_0_RESERVED_0__MASK			0xf0000000
+#define DPU_WDMA_SIZE_0_RESERVED_0__SHIFT			28
+static inline uint32_t DPU_WDMA_SIZE_0_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_WDMA_SIZE_0_RESERVED_0__SHIFT) & DPU_WDMA_SIZE_0_RESERVED_0__MASK;
+}
+#define DPU_WDMA_SIZE_0_TP_PRECISION__MASK			0x08000000
+#define DPU_WDMA_SIZE_0_TP_PRECISION__SHIFT			27
+static inline uint32_t DPU_WDMA_SIZE_0_TP_PRECISION(uint32_t val)
+{
+	return ((val) << DPU_WDMA_SIZE_0_TP_PRECISION__SHIFT) & DPU_WDMA_SIZE_0_TP_PRECISION__MASK;
+}
+#define DPU_WDMA_SIZE_0_SIZE_C_WDMA__MASK			0x07ff0000
+#define DPU_WDMA_SIZE_0_SIZE_C_WDMA__SHIFT			16
+static inline uint32_t DPU_WDMA_SIZE_0_SIZE_C_WDMA(uint32_t val)
+{
+	return ((val) << DPU_WDMA_SIZE_0_SIZE_C_WDMA__SHIFT) & DPU_WDMA_SIZE_0_SIZE_C_WDMA__MASK;
+}
+#define DPU_WDMA_SIZE_0_RESERVED_1__MASK			0x0000e000
+#define DPU_WDMA_SIZE_0_RESERVED_1__SHIFT			13
+static inline uint32_t DPU_WDMA_SIZE_0_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_WDMA_SIZE_0_RESERVED_1__SHIFT) & DPU_WDMA_SIZE_0_RESERVED_1__MASK;
+}
+#define DPU_WDMA_SIZE_0_CHANNEL_WDMA__MASK			0x00001fff
+#define DPU_WDMA_SIZE_0_CHANNEL_WDMA__SHIFT			0
+static inline uint32_t DPU_WDMA_SIZE_0_CHANNEL_WDMA(uint32_t val)
+{
+	return ((val) << DPU_WDMA_SIZE_0_CHANNEL_WDMA__SHIFT) & DPU_WDMA_SIZE_0_CHANNEL_WDMA__MASK;
+}
+
+#define REG_DPU_WDMA_SIZE_1					0x0000405c
+#define DPU_WDMA_SIZE_1_RESERVED_0__MASK			0xe0000000
+#define DPU_WDMA_SIZE_1_RESERVED_0__SHIFT			29
+static inline uint32_t DPU_WDMA_SIZE_1_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_WDMA_SIZE_1_RESERVED_0__SHIFT) & DPU_WDMA_SIZE_1_RESERVED_0__MASK;
+}
+#define DPU_WDMA_SIZE_1_HEIGHT_WDMA__MASK			0x1fff0000
+#define DPU_WDMA_SIZE_1_HEIGHT_WDMA__SHIFT			16
+static inline uint32_t DPU_WDMA_SIZE_1_HEIGHT_WDMA(uint32_t val)
+{
+	return ((val) << DPU_WDMA_SIZE_1_HEIGHT_WDMA__SHIFT) & DPU_WDMA_SIZE_1_HEIGHT_WDMA__MASK;
+}
+#define DPU_WDMA_SIZE_1_RESERVED_1__MASK			0x0000e000
+#define DPU_WDMA_SIZE_1_RESERVED_1__SHIFT			13
+static inline uint32_t DPU_WDMA_SIZE_1_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_WDMA_SIZE_1_RESERVED_1__SHIFT) & DPU_WDMA_SIZE_1_RESERVED_1__MASK;
+}
+#define DPU_WDMA_SIZE_1_WIDTH_WDMA__MASK			0x00001fff
+#define DPU_WDMA_SIZE_1_WIDTH_WDMA__SHIFT			0
+static inline uint32_t DPU_WDMA_SIZE_1_WIDTH_WDMA(uint32_t val)
+{
+	return ((val) << DPU_WDMA_SIZE_1_WIDTH_WDMA__SHIFT) & DPU_WDMA_SIZE_1_WIDTH_WDMA__MASK;
+}
+
+#define REG_DPU_BN_CFG						0x00004060
+#define DPU_BN_CFG_RESERVED_0__MASK				0xfff00000
+#define DPU_BN_CFG_RESERVED_0__SHIFT				20
+static inline uint32_t DPU_BN_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_RESERVED_0__SHIFT) & DPU_BN_CFG_RESERVED_0__MASK;
+}
+#define DPU_BN_CFG_BN_ALU_ALGO__MASK				0x000f0000
+#define DPU_BN_CFG_BN_ALU_ALGO__SHIFT				16
+static inline uint32_t DPU_BN_CFG_BN_ALU_ALGO(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_BN_ALU_ALGO__SHIFT) & DPU_BN_CFG_BN_ALU_ALGO__MASK;
+}
+#define DPU_BN_CFG_RESERVED_1__MASK				0x0000fe00
+#define DPU_BN_CFG_RESERVED_1__SHIFT				9
+static inline uint32_t DPU_BN_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_RESERVED_1__SHIFT) & DPU_BN_CFG_RESERVED_1__MASK;
+}
+#define DPU_BN_CFG_BN_ALU_SRC__MASK				0x00000100
+#define DPU_BN_CFG_BN_ALU_SRC__SHIFT				8
+static inline uint32_t DPU_BN_CFG_BN_ALU_SRC(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_BN_ALU_SRC__SHIFT) & DPU_BN_CFG_BN_ALU_SRC__MASK;
+}
+#define DPU_BN_CFG_BN_RELUX_EN__MASK				0x00000080
+#define DPU_BN_CFG_BN_RELUX_EN__SHIFT				7
+static inline uint32_t DPU_BN_CFG_BN_RELUX_EN(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_BN_RELUX_EN__SHIFT) & DPU_BN_CFG_BN_RELUX_EN__MASK;
+}
+#define DPU_BN_CFG_BN_RELU_BYPASS__MASK				0x00000040
+#define DPU_BN_CFG_BN_RELU_BYPASS__SHIFT			6
+static inline uint32_t DPU_BN_CFG_BN_RELU_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_BN_RELU_BYPASS__SHIFT) & DPU_BN_CFG_BN_RELU_BYPASS__MASK;
+}
+#define DPU_BN_CFG_BN_MUL_PRELU__MASK				0x00000020
+#define DPU_BN_CFG_BN_MUL_PRELU__SHIFT				5
+static inline uint32_t DPU_BN_CFG_BN_MUL_PRELU(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_BN_MUL_PRELU__SHIFT) & DPU_BN_CFG_BN_MUL_PRELU__MASK;
+}
+#define DPU_BN_CFG_BN_MUL_BYPASS__MASK				0x00000010
+#define DPU_BN_CFG_BN_MUL_BYPASS__SHIFT				4
+static inline uint32_t DPU_BN_CFG_BN_MUL_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_BN_MUL_BYPASS__SHIFT) & DPU_BN_CFG_BN_MUL_BYPASS__MASK;
+}
+#define DPU_BN_CFG_RESERVED_2__MASK				0x0000000c
+#define DPU_BN_CFG_RESERVED_2__SHIFT				2
+static inline uint32_t DPU_BN_CFG_RESERVED_2(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_RESERVED_2__SHIFT) & DPU_BN_CFG_RESERVED_2__MASK;
+}
+#define DPU_BN_CFG_BN_ALU_BYPASS__MASK				0x00000002
+#define DPU_BN_CFG_BN_ALU_BYPASS__SHIFT				1
+static inline uint32_t DPU_BN_CFG_BN_ALU_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_BN_ALU_BYPASS__SHIFT) & DPU_BN_CFG_BN_ALU_BYPASS__MASK;
+}
+#define DPU_BN_CFG_BN_BYPASS__MASK				0x00000001
+#define DPU_BN_CFG_BN_BYPASS__SHIFT				0
+static inline uint32_t DPU_BN_CFG_BN_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_BN_CFG_BN_BYPASS__SHIFT) & DPU_BN_CFG_BN_BYPASS__MASK;
+}
+
+#define REG_DPU_BN_ALU_CFG					0x00004064
+#define DPU_BN_ALU_CFG_BN_ALU_OPERAND__MASK			0xffffffff
+#define DPU_BN_ALU_CFG_BN_ALU_OPERAND__SHIFT			0
+static inline uint32_t DPU_BN_ALU_CFG_BN_ALU_OPERAND(uint32_t val)
+{
+	return ((val) << DPU_BN_ALU_CFG_BN_ALU_OPERAND__SHIFT) & DPU_BN_ALU_CFG_BN_ALU_OPERAND__MASK;
+}
+
+#define REG_DPU_BN_MUL_CFG					0x00004068
+#define DPU_BN_MUL_CFG_BN_MUL_OPERAND__MASK			0xffff0000
+#define DPU_BN_MUL_CFG_BN_MUL_OPERAND__SHIFT			16
+static inline uint32_t DPU_BN_MUL_CFG_BN_MUL_OPERAND(uint32_t val)
+{
+	return ((val) << DPU_BN_MUL_CFG_BN_MUL_OPERAND__SHIFT) & DPU_BN_MUL_CFG_BN_MUL_OPERAND__MASK;
+}
+#define DPU_BN_MUL_CFG_RESERVED_0__MASK				0x0000c000
+#define DPU_BN_MUL_CFG_RESERVED_0__SHIFT			14
+static inline uint32_t DPU_BN_MUL_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_BN_MUL_CFG_RESERVED_0__SHIFT) & DPU_BN_MUL_CFG_RESERVED_0__MASK;
+}
+#define DPU_BN_MUL_CFG_BN_MUL_SHIFT_VALUE__MASK			0x00003f00
+#define DPU_BN_MUL_CFG_BN_MUL_SHIFT_VALUE__SHIFT		8
+static inline uint32_t DPU_BN_MUL_CFG_BN_MUL_SHIFT_VALUE(uint32_t val)
+{
+	return ((val) << DPU_BN_MUL_CFG_BN_MUL_SHIFT_VALUE__SHIFT) & DPU_BN_MUL_CFG_BN_MUL_SHIFT_VALUE__MASK;
+}
+#define DPU_BN_MUL_CFG_RESERVED_1__MASK				0x000000fc
+#define DPU_BN_MUL_CFG_RESERVED_1__SHIFT			2
+static inline uint32_t DPU_BN_MUL_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_BN_MUL_CFG_RESERVED_1__SHIFT) & DPU_BN_MUL_CFG_RESERVED_1__MASK;
+}
+#define DPU_BN_MUL_CFG_BN_TRUNCATE_SRC__MASK			0x00000002
+#define DPU_BN_MUL_CFG_BN_TRUNCATE_SRC__SHIFT			1
+static inline uint32_t DPU_BN_MUL_CFG_BN_TRUNCATE_SRC(uint32_t val)
+{
+	return ((val) << DPU_BN_MUL_CFG_BN_TRUNCATE_SRC__SHIFT) & DPU_BN_MUL_CFG_BN_TRUNCATE_SRC__MASK;
+}
+#define DPU_BN_MUL_CFG_BN_MUL_SRC__MASK				0x00000001
+#define DPU_BN_MUL_CFG_BN_MUL_SRC__SHIFT			0
+static inline uint32_t DPU_BN_MUL_CFG_BN_MUL_SRC(uint32_t val)
+{
+	return ((val) << DPU_BN_MUL_CFG_BN_MUL_SRC__SHIFT) & DPU_BN_MUL_CFG_BN_MUL_SRC__MASK;
+}
+
+#define REG_DPU_BN_RELUX_CMP_VALUE				0x0000406c
+#define DPU_BN_RELUX_CMP_VALUE_BN_RELUX_CMP_DAT__MASK		0xffffffff
+#define DPU_BN_RELUX_CMP_VALUE_BN_RELUX_CMP_DAT__SHIFT		0
+static inline uint32_t DPU_BN_RELUX_CMP_VALUE_BN_RELUX_CMP_DAT(uint32_t val)
+{
+	return ((val) << DPU_BN_RELUX_CMP_VALUE_BN_RELUX_CMP_DAT__SHIFT) & DPU_BN_RELUX_CMP_VALUE_BN_RELUX_CMP_DAT__MASK;
+}
+
+#define REG_DPU_EW_CFG						0x00004070
+#define DPU_EW_CFG_EW_CVT_TYPE__MASK				0x80000000
+#define DPU_EW_CFG_EW_CVT_TYPE__SHIFT				31
+static inline uint32_t DPU_EW_CFG_EW_CVT_TYPE(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_CVT_TYPE__SHIFT) & DPU_EW_CFG_EW_CVT_TYPE__MASK;
+}
+#define DPU_EW_CFG_EW_CVT_ROUND__MASK				0x40000000
+#define DPU_EW_CFG_EW_CVT_ROUND__SHIFT				30
+static inline uint32_t DPU_EW_CFG_EW_CVT_ROUND(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_CVT_ROUND__SHIFT) & DPU_EW_CFG_EW_CVT_ROUND__MASK;
+}
+#define DPU_EW_CFG_EW_DATA_MODE__MASK				0x30000000
+#define DPU_EW_CFG_EW_DATA_MODE__SHIFT				28
+static inline uint32_t DPU_EW_CFG_EW_DATA_MODE(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_DATA_MODE__SHIFT) & DPU_EW_CFG_EW_DATA_MODE__MASK;
+}
+#define DPU_EW_CFG_RESERVED_0__MASK				0x0f000000
+#define DPU_EW_CFG_RESERVED_0__SHIFT				24
+static inline uint32_t DPU_EW_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_RESERVED_0__SHIFT) & DPU_EW_CFG_RESERVED_0__MASK;
+}
+#define DPU_EW_CFG_EDATA_SIZE__MASK				0x00c00000
+#define DPU_EW_CFG_EDATA_SIZE__SHIFT				22
+static inline uint32_t DPU_EW_CFG_EDATA_SIZE(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EDATA_SIZE__SHIFT) & DPU_EW_CFG_EDATA_SIZE__MASK;
+}
+#define DPU_EW_CFG_EW_EQUAL_EN__MASK				0x00200000
+#define DPU_EW_CFG_EW_EQUAL_EN__SHIFT				21
+static inline uint32_t DPU_EW_CFG_EW_EQUAL_EN(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_EQUAL_EN__SHIFT) & DPU_EW_CFG_EW_EQUAL_EN__MASK;
+}
+#define DPU_EW_CFG_EW_BINARY_EN__MASK				0x00100000
+#define DPU_EW_CFG_EW_BINARY_EN__SHIFT				20
+static inline uint32_t DPU_EW_CFG_EW_BINARY_EN(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_BINARY_EN__SHIFT) & DPU_EW_CFG_EW_BINARY_EN__MASK;
+}
+#define DPU_EW_CFG_EW_ALU_ALGO__MASK				0x000f0000
+#define DPU_EW_CFG_EW_ALU_ALGO__SHIFT				16
+static inline uint32_t DPU_EW_CFG_EW_ALU_ALGO(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_ALU_ALGO__SHIFT) & DPU_EW_CFG_EW_ALU_ALGO__MASK;
+}
+#define DPU_EW_CFG_RESERVED_1__MASK				0x0000f800
+#define DPU_EW_CFG_RESERVED_1__SHIFT				11
+static inline uint32_t DPU_EW_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_RESERVED_1__SHIFT) & DPU_EW_CFG_RESERVED_1__MASK;
+}
+#define DPU_EW_CFG_EW_RELUX_EN__MASK				0x00000400
+#define DPU_EW_CFG_EW_RELUX_EN__SHIFT				10
+static inline uint32_t DPU_EW_CFG_EW_RELUX_EN(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_RELUX_EN__SHIFT) & DPU_EW_CFG_EW_RELUX_EN__MASK;
+}
+#define DPU_EW_CFG_EW_RELU_BYPASS__MASK				0x00000200
+#define DPU_EW_CFG_EW_RELU_BYPASS__SHIFT			9
+static inline uint32_t DPU_EW_CFG_EW_RELU_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_RELU_BYPASS__SHIFT) & DPU_EW_CFG_EW_RELU_BYPASS__MASK;
+}
+#define DPU_EW_CFG_EW_OP_CVT_BYPASS__MASK			0x00000100
+#define DPU_EW_CFG_EW_OP_CVT_BYPASS__SHIFT			8
+static inline uint32_t DPU_EW_CFG_EW_OP_CVT_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_OP_CVT_BYPASS__SHIFT) & DPU_EW_CFG_EW_OP_CVT_BYPASS__MASK;
+}
+#define DPU_EW_CFG_EW_LUT_BYPASS__MASK				0x00000080
+#define DPU_EW_CFG_EW_LUT_BYPASS__SHIFT				7
+static inline uint32_t DPU_EW_CFG_EW_LUT_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_LUT_BYPASS__SHIFT) & DPU_EW_CFG_EW_LUT_BYPASS__MASK;
+}
+#define DPU_EW_CFG_EW_OP_SRC__MASK				0x00000040
+#define DPU_EW_CFG_EW_OP_SRC__SHIFT				6
+static inline uint32_t DPU_EW_CFG_EW_OP_SRC(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_OP_SRC__SHIFT) & DPU_EW_CFG_EW_OP_SRC__MASK;
+}
+#define DPU_EW_CFG_EW_MUL_PRELU__MASK				0x00000020
+#define DPU_EW_CFG_EW_MUL_PRELU__SHIFT				5
+static inline uint32_t DPU_EW_CFG_EW_MUL_PRELU(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_MUL_PRELU__SHIFT) & DPU_EW_CFG_EW_MUL_PRELU__MASK;
+}
+#define DPU_EW_CFG_RESERVED_2__MASK				0x00000018
+#define DPU_EW_CFG_RESERVED_2__SHIFT				3
+static inline uint32_t DPU_EW_CFG_RESERVED_2(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_RESERVED_2__SHIFT) & DPU_EW_CFG_RESERVED_2__MASK;
+}
+#define DPU_EW_CFG_EW_OP_TYPE__MASK				0x00000004
+#define DPU_EW_CFG_EW_OP_TYPE__SHIFT				2
+static inline uint32_t DPU_EW_CFG_EW_OP_TYPE(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_OP_TYPE__SHIFT) & DPU_EW_CFG_EW_OP_TYPE__MASK;
+}
+#define DPU_EW_CFG_EW_OP_BYPASS__MASK				0x00000002
+#define DPU_EW_CFG_EW_OP_BYPASS__SHIFT				1
+static inline uint32_t DPU_EW_CFG_EW_OP_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_OP_BYPASS__SHIFT) & DPU_EW_CFG_EW_OP_BYPASS__MASK;
+}
+#define DPU_EW_CFG_EW_BYPASS__MASK				0x00000001
+#define DPU_EW_CFG_EW_BYPASS__SHIFT				0
+static inline uint32_t DPU_EW_CFG_EW_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_EW_CFG_EW_BYPASS__SHIFT) & DPU_EW_CFG_EW_BYPASS__MASK;
+}
+
+#define REG_DPU_EW_CVT_OFFSET_VALUE				0x00004074
+#define DPU_EW_CVT_OFFSET_VALUE_EW_OP_CVT_OFFSET__MASK		0xffffffff
+#define DPU_EW_CVT_OFFSET_VALUE_EW_OP_CVT_OFFSET__SHIFT		0
+static inline uint32_t DPU_EW_CVT_OFFSET_VALUE_EW_OP_CVT_OFFSET(uint32_t val)
+{
+	return ((val) << DPU_EW_CVT_OFFSET_VALUE_EW_OP_CVT_OFFSET__SHIFT) & DPU_EW_CVT_OFFSET_VALUE_EW_OP_CVT_OFFSET__MASK;
+}
+
+#define REG_DPU_EW_CVT_SCALE_VALUE				0x00004078
+#define DPU_EW_CVT_SCALE_VALUE_EW_TRUNCATE__MASK		0xffc00000
+#define DPU_EW_CVT_SCALE_VALUE_EW_TRUNCATE__SHIFT		22
+static inline uint32_t DPU_EW_CVT_SCALE_VALUE_EW_TRUNCATE(uint32_t val)
+{
+	return ((val) << DPU_EW_CVT_SCALE_VALUE_EW_TRUNCATE__SHIFT) & DPU_EW_CVT_SCALE_VALUE_EW_TRUNCATE__MASK;
+}
+#define DPU_EW_CVT_SCALE_VALUE_EW_OP_CVT_SHIFT__MASK		0x003f0000
+#define DPU_EW_CVT_SCALE_VALUE_EW_OP_CVT_SHIFT__SHIFT		16
+static inline uint32_t DPU_EW_CVT_SCALE_VALUE_EW_OP_CVT_SHIFT(uint32_t val)
+{
+	return ((val) << DPU_EW_CVT_SCALE_VALUE_EW_OP_CVT_SHIFT__SHIFT) & DPU_EW_CVT_SCALE_VALUE_EW_OP_CVT_SHIFT__MASK;
+}
+#define DPU_EW_CVT_SCALE_VALUE_EW_OP_CVT_SCALE__MASK		0x0000ffff
+#define DPU_EW_CVT_SCALE_VALUE_EW_OP_CVT_SCALE__SHIFT		0
+static inline uint32_t DPU_EW_CVT_SCALE_VALUE_EW_OP_CVT_SCALE(uint32_t val)
+{
+	return ((val) << DPU_EW_CVT_SCALE_VALUE_EW_OP_CVT_SCALE__SHIFT) & DPU_EW_CVT_SCALE_VALUE_EW_OP_CVT_SCALE__MASK;
+}
+
+#define REG_DPU_EW_RELUX_CMP_VALUE				0x0000407c
+#define DPU_EW_RELUX_CMP_VALUE_EW_RELUX_CMP_DAT__MASK		0xffffffff
+#define DPU_EW_RELUX_CMP_VALUE_EW_RELUX_CMP_DAT__SHIFT		0
+static inline uint32_t DPU_EW_RELUX_CMP_VALUE_EW_RELUX_CMP_DAT(uint32_t val)
+{
+	return ((val) << DPU_EW_RELUX_CMP_VALUE_EW_RELUX_CMP_DAT__SHIFT) & DPU_EW_RELUX_CMP_VALUE_EW_RELUX_CMP_DAT__MASK;
+}
+
+#define REG_DPU_OUT_CVT_OFFSET					0x00004080
+#define DPU_OUT_CVT_OFFSET_OUT_CVT_OFFSET__MASK			0xffffffff
+#define DPU_OUT_CVT_OFFSET_OUT_CVT_OFFSET__SHIFT		0
+static inline uint32_t DPU_OUT_CVT_OFFSET_OUT_CVT_OFFSET(uint32_t val)
+{
+	return ((val) << DPU_OUT_CVT_OFFSET_OUT_CVT_OFFSET__SHIFT) & DPU_OUT_CVT_OFFSET_OUT_CVT_OFFSET__MASK;
+}
+
+#define REG_DPU_OUT_CVT_SCALE					0x00004084
+#define DPU_OUT_CVT_SCALE_RESERVED_0__MASK			0xfffe0000
+#define DPU_OUT_CVT_SCALE_RESERVED_0__SHIFT			17
+static inline uint32_t DPU_OUT_CVT_SCALE_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_OUT_CVT_SCALE_RESERVED_0__SHIFT) & DPU_OUT_CVT_SCALE_RESERVED_0__MASK;
+}
+#define DPU_OUT_CVT_SCALE_FP32TOFP16_EN__MASK			0x00010000
+#define DPU_OUT_CVT_SCALE_FP32TOFP16_EN__SHIFT			16
+static inline uint32_t DPU_OUT_CVT_SCALE_FP32TOFP16_EN(uint32_t val)
+{
+	return ((val) << DPU_OUT_CVT_SCALE_FP32TOFP16_EN__SHIFT) & DPU_OUT_CVT_SCALE_FP32TOFP16_EN__MASK;
+}
+#define DPU_OUT_CVT_SCALE_OUT_CVT_SCALE__MASK			0x0000ffff
+#define DPU_OUT_CVT_SCALE_OUT_CVT_SCALE__SHIFT			0
+static inline uint32_t DPU_OUT_CVT_SCALE_OUT_CVT_SCALE(uint32_t val)
+{
+	return ((val) << DPU_OUT_CVT_SCALE_OUT_CVT_SCALE__SHIFT) & DPU_OUT_CVT_SCALE_OUT_CVT_SCALE__MASK;
+}
+
+#define REG_DPU_OUT_CVT_SHIFT					0x00004088
+#define DPU_OUT_CVT_SHIFT_CVT_TYPE__MASK			0x80000000
+#define DPU_OUT_CVT_SHIFT_CVT_TYPE__SHIFT			31
+static inline uint32_t DPU_OUT_CVT_SHIFT_CVT_TYPE(uint32_t val)
+{
+	return ((val) << DPU_OUT_CVT_SHIFT_CVT_TYPE__SHIFT) & DPU_OUT_CVT_SHIFT_CVT_TYPE__MASK;
+}
+#define DPU_OUT_CVT_SHIFT_CVT_ROUND__MASK			0x40000000
+#define DPU_OUT_CVT_SHIFT_CVT_ROUND__SHIFT			30
+static inline uint32_t DPU_OUT_CVT_SHIFT_CVT_ROUND(uint32_t val)
+{
+	return ((val) << DPU_OUT_CVT_SHIFT_CVT_ROUND__SHIFT) & DPU_OUT_CVT_SHIFT_CVT_ROUND__MASK;
+}
+#define DPU_OUT_CVT_SHIFT_RESERVED_0__MASK			0x3ff00000
+#define DPU_OUT_CVT_SHIFT_RESERVED_0__SHIFT			20
+static inline uint32_t DPU_OUT_CVT_SHIFT_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_OUT_CVT_SHIFT_RESERVED_0__SHIFT) & DPU_OUT_CVT_SHIFT_RESERVED_0__MASK;
+}
+#define DPU_OUT_CVT_SHIFT_MINUS_EXP__MASK			0x000ff000
+#define DPU_OUT_CVT_SHIFT_MINUS_EXP__SHIFT			12
+static inline uint32_t DPU_OUT_CVT_SHIFT_MINUS_EXP(uint32_t val)
+{
+	return ((val) << DPU_OUT_CVT_SHIFT_MINUS_EXP__SHIFT) & DPU_OUT_CVT_SHIFT_MINUS_EXP__MASK;
+}
+#define DPU_OUT_CVT_SHIFT_OUT_CVT_SHIFT__MASK			0x00000fff
+#define DPU_OUT_CVT_SHIFT_OUT_CVT_SHIFT__SHIFT			0
+static inline uint32_t DPU_OUT_CVT_SHIFT_OUT_CVT_SHIFT(uint32_t val)
+{
+	return ((val) << DPU_OUT_CVT_SHIFT_OUT_CVT_SHIFT__SHIFT) & DPU_OUT_CVT_SHIFT_OUT_CVT_SHIFT__MASK;
+}
+
+#define REG_DPU_EW_OP_VALUE_0					0x00004090
+#define DPU_EW_OP_VALUE_0_EW_OPERAND_0__MASK			0xffffffff
+#define DPU_EW_OP_VALUE_0_EW_OPERAND_0__SHIFT			0
+static inline uint32_t DPU_EW_OP_VALUE_0_EW_OPERAND_0(uint32_t val)
+{
+	return ((val) << DPU_EW_OP_VALUE_0_EW_OPERAND_0__SHIFT) & DPU_EW_OP_VALUE_0_EW_OPERAND_0__MASK;
+}
+
+#define REG_DPU_EW_OP_VALUE_1					0x00004094
+#define DPU_EW_OP_VALUE_1_EW_OPERAND_1__MASK			0xffffffff
+#define DPU_EW_OP_VALUE_1_EW_OPERAND_1__SHIFT			0
+static inline uint32_t DPU_EW_OP_VALUE_1_EW_OPERAND_1(uint32_t val)
+{
+	return ((val) << DPU_EW_OP_VALUE_1_EW_OPERAND_1__SHIFT) & DPU_EW_OP_VALUE_1_EW_OPERAND_1__MASK;
+}
+
+#define REG_DPU_EW_OP_VALUE_2					0x00004098
+#define DPU_EW_OP_VALUE_2_EW_OPERAND_2__MASK			0xffffffff
+#define DPU_EW_OP_VALUE_2_EW_OPERAND_2__SHIFT			0
+static inline uint32_t DPU_EW_OP_VALUE_2_EW_OPERAND_2(uint32_t val)
+{
+	return ((val) << DPU_EW_OP_VALUE_2_EW_OPERAND_2__SHIFT) & DPU_EW_OP_VALUE_2_EW_OPERAND_2__MASK;
+}
+
+#define REG_DPU_EW_OP_VALUE_3					0x0000409c
+#define DPU_EW_OP_VALUE_3_EW_OPERAND_3__MASK			0xffffffff
+#define DPU_EW_OP_VALUE_3_EW_OPERAND_3__SHIFT			0
+static inline uint32_t DPU_EW_OP_VALUE_3_EW_OPERAND_3(uint32_t val)
+{
+	return ((val) << DPU_EW_OP_VALUE_3_EW_OPERAND_3__SHIFT) & DPU_EW_OP_VALUE_3_EW_OPERAND_3__MASK;
+}
+
+#define REG_DPU_EW_OP_VALUE_4					0x000040a0
+#define DPU_EW_OP_VALUE_4_EW_OPERAND_4__MASK			0xffffffff
+#define DPU_EW_OP_VALUE_4_EW_OPERAND_4__SHIFT			0
+static inline uint32_t DPU_EW_OP_VALUE_4_EW_OPERAND_4(uint32_t val)
+{
+	return ((val) << DPU_EW_OP_VALUE_4_EW_OPERAND_4__SHIFT) & DPU_EW_OP_VALUE_4_EW_OPERAND_4__MASK;
+}
+
+#define REG_DPU_EW_OP_VALUE_5					0x000040a4
+#define DPU_EW_OP_VALUE_5_EW_OPERAND_5__MASK			0xffffffff
+#define DPU_EW_OP_VALUE_5_EW_OPERAND_5__SHIFT			0
+static inline uint32_t DPU_EW_OP_VALUE_5_EW_OPERAND_5(uint32_t val)
+{
+	return ((val) << DPU_EW_OP_VALUE_5_EW_OPERAND_5__SHIFT) & DPU_EW_OP_VALUE_5_EW_OPERAND_5__MASK;
+}
+
+#define REG_DPU_EW_OP_VALUE_6					0x000040a8
+#define DPU_EW_OP_VALUE_6_EW_OPERAND_6__MASK			0xffffffff
+#define DPU_EW_OP_VALUE_6_EW_OPERAND_6__SHIFT			0
+static inline uint32_t DPU_EW_OP_VALUE_6_EW_OPERAND_6(uint32_t val)
+{
+	return ((val) << DPU_EW_OP_VALUE_6_EW_OPERAND_6__SHIFT) & DPU_EW_OP_VALUE_6_EW_OPERAND_6__MASK;
+}
+
+#define REG_DPU_EW_OP_VALUE_7					0x000040ac
+#define DPU_EW_OP_VALUE_7_EW_OPERAND_7__MASK			0xffffffff
+#define DPU_EW_OP_VALUE_7_EW_OPERAND_7__SHIFT			0
+static inline uint32_t DPU_EW_OP_VALUE_7_EW_OPERAND_7(uint32_t val)
+{
+	return ((val) << DPU_EW_OP_VALUE_7_EW_OPERAND_7__SHIFT) & DPU_EW_OP_VALUE_7_EW_OPERAND_7__MASK;
+}
+
+#define REG_DPU_SURFACE_ADD					0x000040c0
+#define DPU_SURFACE_ADD_SURF_ADD__MASK				0xfffffff0
+#define DPU_SURFACE_ADD_SURF_ADD__SHIFT				4
+static inline uint32_t DPU_SURFACE_ADD_SURF_ADD(uint32_t val)
+{
+	return ((val) << DPU_SURFACE_ADD_SURF_ADD__SHIFT) & DPU_SURFACE_ADD_SURF_ADD__MASK;
+}
+#define DPU_SURFACE_ADD_RESERVED_0__MASK			0x0000000f
+#define DPU_SURFACE_ADD_RESERVED_0__SHIFT			0
+static inline uint32_t DPU_SURFACE_ADD_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_SURFACE_ADD_RESERVED_0__SHIFT) & DPU_SURFACE_ADD_RESERVED_0__MASK;
+}
+
+#define REG_DPU_LUT_ACCESS_CFG					0x00004100
+#define DPU_LUT_ACCESS_CFG_RESERVED_0__MASK			0xfffc0000
+#define DPU_LUT_ACCESS_CFG_RESERVED_0__SHIFT			18
+static inline uint32_t DPU_LUT_ACCESS_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_LUT_ACCESS_CFG_RESERVED_0__SHIFT) & DPU_LUT_ACCESS_CFG_RESERVED_0__MASK;
+}
+#define DPU_LUT_ACCESS_CFG_LUT_ACCESS_TYPE__MASK		0x00020000
+#define DPU_LUT_ACCESS_CFG_LUT_ACCESS_TYPE__SHIFT		17
+static inline uint32_t DPU_LUT_ACCESS_CFG_LUT_ACCESS_TYPE(uint32_t val)
+{
+	return ((val) << DPU_LUT_ACCESS_CFG_LUT_ACCESS_TYPE__SHIFT) & DPU_LUT_ACCESS_CFG_LUT_ACCESS_TYPE__MASK;
+}
+#define DPU_LUT_ACCESS_CFG_LUT_TABLE_ID__MASK			0x00010000
+#define DPU_LUT_ACCESS_CFG_LUT_TABLE_ID__SHIFT			16
+static inline uint32_t DPU_LUT_ACCESS_CFG_LUT_TABLE_ID(uint32_t val)
+{
+	return ((val) << DPU_LUT_ACCESS_CFG_LUT_TABLE_ID__SHIFT) & DPU_LUT_ACCESS_CFG_LUT_TABLE_ID__MASK;
+}
+#define DPU_LUT_ACCESS_CFG_RESERVED_1__MASK			0x0000fc00
+#define DPU_LUT_ACCESS_CFG_RESERVED_1__SHIFT			10
+static inline uint32_t DPU_LUT_ACCESS_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_LUT_ACCESS_CFG_RESERVED_1__SHIFT) & DPU_LUT_ACCESS_CFG_RESERVED_1__MASK;
+}
+#define DPU_LUT_ACCESS_CFG_LUT_ADDR__MASK			0x000003ff
+#define DPU_LUT_ACCESS_CFG_LUT_ADDR__SHIFT			0
+static inline uint32_t DPU_LUT_ACCESS_CFG_LUT_ADDR(uint32_t val)
+{
+	return ((val) << DPU_LUT_ACCESS_CFG_LUT_ADDR__SHIFT) & DPU_LUT_ACCESS_CFG_LUT_ADDR__MASK;
+}
+
+#define REG_DPU_LUT_ACCESS_DATA					0x00004104
+#define DPU_LUT_ACCESS_DATA_RESERVED_0__MASK			0xffff0000
+#define DPU_LUT_ACCESS_DATA_RESERVED_0__SHIFT			16
+static inline uint32_t DPU_LUT_ACCESS_DATA_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_LUT_ACCESS_DATA_RESERVED_0__SHIFT) & DPU_LUT_ACCESS_DATA_RESERVED_0__MASK;
+}
+#define DPU_LUT_ACCESS_DATA_LUT_ACCESS_DATA__MASK		0x0000ffff
+#define DPU_LUT_ACCESS_DATA_LUT_ACCESS_DATA__SHIFT		0
+static inline uint32_t DPU_LUT_ACCESS_DATA_LUT_ACCESS_DATA(uint32_t val)
+{
+	return ((val) << DPU_LUT_ACCESS_DATA_LUT_ACCESS_DATA__SHIFT) & DPU_LUT_ACCESS_DATA_LUT_ACCESS_DATA__MASK;
+}
+
+#define REG_DPU_LUT_CFG						0x00004108
+#define DPU_LUT_CFG_RESERVED_0__MASK				0xffffff00
+#define DPU_LUT_CFG_RESERVED_0__SHIFT				8
+static inline uint32_t DPU_LUT_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_LUT_CFG_RESERVED_0__SHIFT) & DPU_LUT_CFG_RESERVED_0__MASK;
+}
+#define DPU_LUT_CFG_LUT_CAL_SEL__MASK				0x00000080
+#define DPU_LUT_CFG_LUT_CAL_SEL__SHIFT				7
+static inline uint32_t DPU_LUT_CFG_LUT_CAL_SEL(uint32_t val)
+{
+	return ((val) << DPU_LUT_CFG_LUT_CAL_SEL__SHIFT) & DPU_LUT_CFG_LUT_CAL_SEL__MASK;
+}
+#define DPU_LUT_CFG_LUT_HYBRID_PRIORITY__MASK			0x00000040
+#define DPU_LUT_CFG_LUT_HYBRID_PRIORITY__SHIFT			6
+static inline uint32_t DPU_LUT_CFG_LUT_HYBRID_PRIORITY(uint32_t val)
+{
+	return ((val) << DPU_LUT_CFG_LUT_HYBRID_PRIORITY__SHIFT) & DPU_LUT_CFG_LUT_HYBRID_PRIORITY__MASK;
+}
+#define DPU_LUT_CFG_LUT_OFLOW_PRIORITY__MASK			0x00000020
+#define DPU_LUT_CFG_LUT_OFLOW_PRIORITY__SHIFT			5
+static inline uint32_t DPU_LUT_CFG_LUT_OFLOW_PRIORITY(uint32_t val)
+{
+	return ((val) << DPU_LUT_CFG_LUT_OFLOW_PRIORITY__SHIFT) & DPU_LUT_CFG_LUT_OFLOW_PRIORITY__MASK;
+}
+#define DPU_LUT_CFG_LUT_UFLOW_PRIORITY__MASK			0x00000010
+#define DPU_LUT_CFG_LUT_UFLOW_PRIORITY__SHIFT			4
+static inline uint32_t DPU_LUT_CFG_LUT_UFLOW_PRIORITY(uint32_t val)
+{
+	return ((val) << DPU_LUT_CFG_LUT_UFLOW_PRIORITY__SHIFT) & DPU_LUT_CFG_LUT_UFLOW_PRIORITY__MASK;
+}
+#define DPU_LUT_CFG_LUT_LO_LE_MUX__MASK				0x0000000c
+#define DPU_LUT_CFG_LUT_LO_LE_MUX__SHIFT			2
+static inline uint32_t DPU_LUT_CFG_LUT_LO_LE_MUX(uint32_t val)
+{
+	return ((val) << DPU_LUT_CFG_LUT_LO_LE_MUX__SHIFT) & DPU_LUT_CFG_LUT_LO_LE_MUX__MASK;
+}
+#define DPU_LUT_CFG_LUT_EXPAND_EN__MASK				0x00000002
+#define DPU_LUT_CFG_LUT_EXPAND_EN__SHIFT			1
+static inline uint32_t DPU_LUT_CFG_LUT_EXPAND_EN(uint32_t val)
+{
+	return ((val) << DPU_LUT_CFG_LUT_EXPAND_EN__SHIFT) & DPU_LUT_CFG_LUT_EXPAND_EN__MASK;
+}
+#define DPU_LUT_CFG_LUT_ROAD_SEL__MASK				0x00000001
+#define DPU_LUT_CFG_LUT_ROAD_SEL__SHIFT				0
+static inline uint32_t DPU_LUT_CFG_LUT_ROAD_SEL(uint32_t val)
+{
+	return ((val) << DPU_LUT_CFG_LUT_ROAD_SEL__SHIFT) & DPU_LUT_CFG_LUT_ROAD_SEL__MASK;
+}
+
+#define REG_DPU_LUT_INFO					0x0000410c
+#define DPU_LUT_INFO_RESERVED_0__MASK				0xff000000
+#define DPU_LUT_INFO_RESERVED_0__SHIFT				24
+static inline uint32_t DPU_LUT_INFO_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_LUT_INFO_RESERVED_0__SHIFT) & DPU_LUT_INFO_RESERVED_0__MASK;
+}
+#define DPU_LUT_INFO_LUT_LO_INDEX_SELECT__MASK			0x00ff0000
+#define DPU_LUT_INFO_LUT_LO_INDEX_SELECT__SHIFT			16
+static inline uint32_t DPU_LUT_INFO_LUT_LO_INDEX_SELECT(uint32_t val)
+{
+	return ((val) << DPU_LUT_INFO_LUT_LO_INDEX_SELECT__SHIFT) & DPU_LUT_INFO_LUT_LO_INDEX_SELECT__MASK;
+}
+#define DPU_LUT_INFO_LUT_LE_INDEX_SELECT__MASK			0x0000ff00
+#define DPU_LUT_INFO_LUT_LE_INDEX_SELECT__SHIFT			8
+static inline uint32_t DPU_LUT_INFO_LUT_LE_INDEX_SELECT(uint32_t val)
+{
+	return ((val) << DPU_LUT_INFO_LUT_LE_INDEX_SELECT__SHIFT) & DPU_LUT_INFO_LUT_LE_INDEX_SELECT__MASK;
+}
+#define DPU_LUT_INFO_RESERVED_1__MASK				0x000000ff
+#define DPU_LUT_INFO_RESERVED_1__SHIFT				0
+static inline uint32_t DPU_LUT_INFO_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_LUT_INFO_RESERVED_1__SHIFT) & DPU_LUT_INFO_RESERVED_1__MASK;
+}
+
+#define REG_DPU_LUT_LE_START					0x00004110
+#define DPU_LUT_LE_START_LUT_LE_START__MASK			0xffffffff
+#define DPU_LUT_LE_START_LUT_LE_START__SHIFT			0
+static inline uint32_t DPU_LUT_LE_START_LUT_LE_START(uint32_t val)
+{
+	return ((val) << DPU_LUT_LE_START_LUT_LE_START__SHIFT) & DPU_LUT_LE_START_LUT_LE_START__MASK;
+}
+
+#define REG_DPU_LUT_LE_END					0x00004114
+#define DPU_LUT_LE_END_LUT_LE_END__MASK				0xffffffff
+#define DPU_LUT_LE_END_LUT_LE_END__SHIFT			0
+static inline uint32_t DPU_LUT_LE_END_LUT_LE_END(uint32_t val)
+{
+	return ((val) << DPU_LUT_LE_END_LUT_LE_END__SHIFT) & DPU_LUT_LE_END_LUT_LE_END__MASK;
+}
+
+#define REG_DPU_LUT_LO_START					0x00004118
+#define DPU_LUT_LO_START_LUT_LO_START__MASK			0xffffffff
+#define DPU_LUT_LO_START_LUT_LO_START__SHIFT			0
+static inline uint32_t DPU_LUT_LO_START_LUT_LO_START(uint32_t val)
+{
+	return ((val) << DPU_LUT_LO_START_LUT_LO_START__SHIFT) & DPU_LUT_LO_START_LUT_LO_START__MASK;
+}
+
+#define REG_DPU_LUT_LO_END					0x0000411c
+#define DPU_LUT_LO_END_LUT_LO_END__MASK				0xffffffff
+#define DPU_LUT_LO_END_LUT_LO_END__SHIFT			0
+static inline uint32_t DPU_LUT_LO_END_LUT_LO_END(uint32_t val)
+{
+	return ((val) << DPU_LUT_LO_END_LUT_LO_END__SHIFT) & DPU_LUT_LO_END_LUT_LO_END__MASK;
+}
+
+#define REG_DPU_LUT_LE_SLOPE_SCALE				0x00004120
+#define DPU_LUT_LE_SLOPE_SCALE_LUT_LE_SLOPE_OFLOW_SCALE__MASK	0xffff0000
+#define DPU_LUT_LE_SLOPE_SCALE_LUT_LE_SLOPE_OFLOW_SCALE__SHIFT	16
+static inline uint32_t DPU_LUT_LE_SLOPE_SCALE_LUT_LE_SLOPE_OFLOW_SCALE(uint32_t val)
+{
+	return ((val) << DPU_LUT_LE_SLOPE_SCALE_LUT_LE_SLOPE_OFLOW_SCALE__SHIFT) & DPU_LUT_LE_SLOPE_SCALE_LUT_LE_SLOPE_OFLOW_SCALE__MASK;
+}
+#define DPU_LUT_LE_SLOPE_SCALE_LUT_LE_SLOPE_UFLOW_SCALE__MASK	0x0000ffff
+#define DPU_LUT_LE_SLOPE_SCALE_LUT_LE_SLOPE_UFLOW_SCALE__SHIFT	0
+static inline uint32_t DPU_LUT_LE_SLOPE_SCALE_LUT_LE_SLOPE_UFLOW_SCALE(uint32_t val)
+{
+	return ((val) << DPU_LUT_LE_SLOPE_SCALE_LUT_LE_SLOPE_UFLOW_SCALE__SHIFT) & DPU_LUT_LE_SLOPE_SCALE_LUT_LE_SLOPE_UFLOW_SCALE__MASK;
+}
+
+#define REG_DPU_LUT_LE_SLOPE_SHIFT				0x00004124
+#define DPU_LUT_LE_SLOPE_SHIFT_RESERVED_0__MASK			0xfffffc00
+#define DPU_LUT_LE_SLOPE_SHIFT_RESERVED_0__SHIFT		10
+static inline uint32_t DPU_LUT_LE_SLOPE_SHIFT_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_LUT_LE_SLOPE_SHIFT_RESERVED_0__SHIFT) & DPU_LUT_LE_SLOPE_SHIFT_RESERVED_0__MASK;
+}
+#define DPU_LUT_LE_SLOPE_SHIFT_LUT_LE_SLOPE_OFLOW_SHIFT__MASK	0x000003e0
+#define DPU_LUT_LE_SLOPE_SHIFT_LUT_LE_SLOPE_OFLOW_SHIFT__SHIFT	5
+static inline uint32_t DPU_LUT_LE_SLOPE_SHIFT_LUT_LE_SLOPE_OFLOW_SHIFT(uint32_t val)
+{
+	return ((val) << DPU_LUT_LE_SLOPE_SHIFT_LUT_LE_SLOPE_OFLOW_SHIFT__SHIFT) & DPU_LUT_LE_SLOPE_SHIFT_LUT_LE_SLOPE_OFLOW_SHIFT__MASK;
+}
+#define DPU_LUT_LE_SLOPE_SHIFT_LUT_LE_SLOPE_UFLOW_SHIFT__MASK	0x0000001f
+#define DPU_LUT_LE_SLOPE_SHIFT_LUT_LE_SLOPE_UFLOW_SHIFT__SHIFT	0
+static inline uint32_t DPU_LUT_LE_SLOPE_SHIFT_LUT_LE_SLOPE_UFLOW_SHIFT(uint32_t val)
+{
+	return ((val) << DPU_LUT_LE_SLOPE_SHIFT_LUT_LE_SLOPE_UFLOW_SHIFT__SHIFT) & DPU_LUT_LE_SLOPE_SHIFT_LUT_LE_SLOPE_UFLOW_SHIFT__MASK;
+}
+
+#define REG_DPU_LUT_LO_SLOPE_SCALE				0x00004128
+#define DPU_LUT_LO_SLOPE_SCALE_LUT_LO_SLOPE_OFLOW_SCALE__MASK	0xffff0000
+#define DPU_LUT_LO_SLOPE_SCALE_LUT_LO_SLOPE_OFLOW_SCALE__SHIFT	16
+static inline uint32_t DPU_LUT_LO_SLOPE_SCALE_LUT_LO_SLOPE_OFLOW_SCALE(uint32_t val)
+{
+	return ((val) << DPU_LUT_LO_SLOPE_SCALE_LUT_LO_SLOPE_OFLOW_SCALE__SHIFT) & DPU_LUT_LO_SLOPE_SCALE_LUT_LO_SLOPE_OFLOW_SCALE__MASK;
+}
+#define DPU_LUT_LO_SLOPE_SCALE_LUT_LO_SLOPE_UFLOW_SCALE__MASK	0x0000ffff
+#define DPU_LUT_LO_SLOPE_SCALE_LUT_LO_SLOPE_UFLOW_SCALE__SHIFT	0
+static inline uint32_t DPU_LUT_LO_SLOPE_SCALE_LUT_LO_SLOPE_UFLOW_SCALE(uint32_t val)
+{
+	return ((val) << DPU_LUT_LO_SLOPE_SCALE_LUT_LO_SLOPE_UFLOW_SCALE__SHIFT) & DPU_LUT_LO_SLOPE_SCALE_LUT_LO_SLOPE_UFLOW_SCALE__MASK;
+}
+
+#define REG_DPU_LUT_LO_SLOPE_SHIFT				0x0000412c
+#define DPU_LUT_LO_SLOPE_SHIFT_RESERVED_0__MASK			0xfffffc00
+#define DPU_LUT_LO_SLOPE_SHIFT_RESERVED_0__SHIFT		10
+static inline uint32_t DPU_LUT_LO_SLOPE_SHIFT_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_LUT_LO_SLOPE_SHIFT_RESERVED_0__SHIFT) & DPU_LUT_LO_SLOPE_SHIFT_RESERVED_0__MASK;
+}
+#define DPU_LUT_LO_SLOPE_SHIFT_LUT_LO_SLOPE_OFLOW_SHIFT__MASK	0x000003e0
+#define DPU_LUT_LO_SLOPE_SHIFT_LUT_LO_SLOPE_OFLOW_SHIFT__SHIFT	5
+static inline uint32_t DPU_LUT_LO_SLOPE_SHIFT_LUT_LO_SLOPE_OFLOW_SHIFT(uint32_t val)
+{
+	return ((val) << DPU_LUT_LO_SLOPE_SHIFT_LUT_LO_SLOPE_OFLOW_SHIFT__SHIFT) & DPU_LUT_LO_SLOPE_SHIFT_LUT_LO_SLOPE_OFLOW_SHIFT__MASK;
+}
+#define DPU_LUT_LO_SLOPE_SHIFT_LUT_LO_SLOPE_UFLOW_SHIFT__MASK	0x0000001f
+#define DPU_LUT_LO_SLOPE_SHIFT_LUT_LO_SLOPE_UFLOW_SHIFT__SHIFT	0
+static inline uint32_t DPU_LUT_LO_SLOPE_SHIFT_LUT_LO_SLOPE_UFLOW_SHIFT(uint32_t val)
+{
+	return ((val) << DPU_LUT_LO_SLOPE_SHIFT_LUT_LO_SLOPE_UFLOW_SHIFT__SHIFT) & DPU_LUT_LO_SLOPE_SHIFT_LUT_LO_SLOPE_UFLOW_SHIFT__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_S_STATUS				0x00005000
+#define DPU_RDMA_RDMA_S_STATUS_RESERVED_0__MASK			0xfffc0000
+#define DPU_RDMA_RDMA_S_STATUS_RESERVED_0__SHIFT		18
+static inline uint32_t DPU_RDMA_RDMA_S_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_STATUS_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_S_STATUS_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_S_STATUS_STATUS_1__MASK			0x00030000
+#define DPU_RDMA_RDMA_S_STATUS_STATUS_1__SHIFT			16
+static inline uint32_t DPU_RDMA_RDMA_S_STATUS_STATUS_1(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_STATUS_STATUS_1__SHIFT) & DPU_RDMA_RDMA_S_STATUS_STATUS_1__MASK;
+}
+#define DPU_RDMA_RDMA_S_STATUS_RESERVED_1__MASK			0x0000fffc
+#define DPU_RDMA_RDMA_S_STATUS_RESERVED_1__SHIFT		2
+static inline uint32_t DPU_RDMA_RDMA_S_STATUS_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_STATUS_RESERVED_1__SHIFT) & DPU_RDMA_RDMA_S_STATUS_RESERVED_1__MASK;
+}
+#define DPU_RDMA_RDMA_S_STATUS_STATUS_0__MASK			0x00000003
+#define DPU_RDMA_RDMA_S_STATUS_STATUS_0__SHIFT			0
+static inline uint32_t DPU_RDMA_RDMA_S_STATUS_STATUS_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_STATUS_STATUS_0__SHIFT) & DPU_RDMA_RDMA_S_STATUS_STATUS_0__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_S_POINTER				0x00005004
+#define DPU_RDMA_RDMA_S_POINTER_RESERVED_0__MASK		0xfffe0000
+#define DPU_RDMA_RDMA_S_POINTER_RESERVED_0__SHIFT		17
+static inline uint32_t DPU_RDMA_RDMA_S_POINTER_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_POINTER_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_S_POINTER_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_S_POINTER_EXECUTER__MASK			0x00010000
+#define DPU_RDMA_RDMA_S_POINTER_EXECUTER__SHIFT			16
+static inline uint32_t DPU_RDMA_RDMA_S_POINTER_EXECUTER(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_POINTER_EXECUTER__SHIFT) & DPU_RDMA_RDMA_S_POINTER_EXECUTER__MASK;
+}
+#define DPU_RDMA_RDMA_S_POINTER_RESERVED_1__MASK		0x0000ffc0
+#define DPU_RDMA_RDMA_S_POINTER_RESERVED_1__SHIFT		6
+static inline uint32_t DPU_RDMA_RDMA_S_POINTER_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_POINTER_RESERVED_1__SHIFT) & DPU_RDMA_RDMA_S_POINTER_RESERVED_1__MASK;
+}
+#define DPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_CLEAR__MASK		0x00000020
+#define DPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_CLEAR__SHIFT	5
+static inline uint32_t DPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_CLEAR__SHIFT) & DPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_CLEAR__MASK;
+}
+#define DPU_RDMA_RDMA_S_POINTER_POINTER_PP_CLEAR__MASK		0x00000010
+#define DPU_RDMA_RDMA_S_POINTER_POINTER_PP_CLEAR__SHIFT		4
+static inline uint32_t DPU_RDMA_RDMA_S_POINTER_POINTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_POINTER_POINTER_PP_CLEAR__SHIFT) & DPU_RDMA_RDMA_S_POINTER_POINTER_PP_CLEAR__MASK;
+}
+#define DPU_RDMA_RDMA_S_POINTER_POINTER_PP_MODE__MASK		0x00000008
+#define DPU_RDMA_RDMA_S_POINTER_POINTER_PP_MODE__SHIFT		3
+static inline uint32_t DPU_RDMA_RDMA_S_POINTER_POINTER_PP_MODE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_POINTER_POINTER_PP_MODE__SHIFT) & DPU_RDMA_RDMA_S_POINTER_POINTER_PP_MODE__MASK;
+}
+#define DPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_EN__MASK		0x00000004
+#define DPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_EN__SHIFT		2
+static inline uint32_t DPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_EN(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_EN__SHIFT) & DPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_EN__MASK;
+}
+#define DPU_RDMA_RDMA_S_POINTER_POINTER_PP_EN__MASK		0x00000002
+#define DPU_RDMA_RDMA_S_POINTER_POINTER_PP_EN__SHIFT		1
+static inline uint32_t DPU_RDMA_RDMA_S_POINTER_POINTER_PP_EN(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_POINTER_POINTER_PP_EN__SHIFT) & DPU_RDMA_RDMA_S_POINTER_POINTER_PP_EN__MASK;
+}
+#define DPU_RDMA_RDMA_S_POINTER_POINTER__MASK			0x00000001
+#define DPU_RDMA_RDMA_S_POINTER_POINTER__SHIFT			0
+static inline uint32_t DPU_RDMA_RDMA_S_POINTER_POINTER(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_S_POINTER_POINTER__SHIFT) & DPU_RDMA_RDMA_S_POINTER_POINTER__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_OPERATION_ENABLE			0x00005008
+#define DPU_RDMA_RDMA_OPERATION_ENABLE_RESERVED_0__MASK		0xfffffffe
+#define DPU_RDMA_RDMA_OPERATION_ENABLE_RESERVED_0__SHIFT	1
+static inline uint32_t DPU_RDMA_RDMA_OPERATION_ENABLE_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_OPERATION_ENABLE_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_OPERATION_ENABLE_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_OPERATION_ENABLE_OP_EN__MASK		0x00000001
+#define DPU_RDMA_RDMA_OPERATION_ENABLE_OP_EN__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_OPERATION_ENABLE_OP_EN(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_OPERATION_ENABLE_OP_EN__SHIFT) & DPU_RDMA_RDMA_OPERATION_ENABLE_OP_EN__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_DATA_CUBE_WIDTH			0x0000500c
+#define DPU_RDMA_RDMA_DATA_CUBE_WIDTH_RESERVED_0__MASK		0xffffe000
+#define DPU_RDMA_RDMA_DATA_CUBE_WIDTH_RESERVED_0__SHIFT		13
+static inline uint32_t DPU_RDMA_RDMA_DATA_CUBE_WIDTH_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_DATA_CUBE_WIDTH_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_DATA_CUBE_WIDTH_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_DATA_CUBE_WIDTH_WIDTH__MASK		0x00001fff
+#define DPU_RDMA_RDMA_DATA_CUBE_WIDTH_WIDTH__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_DATA_CUBE_WIDTH_WIDTH(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_DATA_CUBE_WIDTH_WIDTH__SHIFT) & DPU_RDMA_RDMA_DATA_CUBE_WIDTH_WIDTH__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_DATA_CUBE_HEIGHT			0x00005010
+#define DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_RESERVED_0__MASK		0xe0000000
+#define DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_RESERVED_0__SHIFT	29
+static inline uint32_t DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_EW_LINE_NOTCH_ADDR__MASK	0x1fff0000
+#define DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_EW_LINE_NOTCH_ADDR__SHIFT	16
+static inline uint32_t DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_EW_LINE_NOTCH_ADDR(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_EW_LINE_NOTCH_ADDR__SHIFT) & DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_EW_LINE_NOTCH_ADDR__MASK;
+}
+#define DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_RESERVED_1__MASK		0x0000e000
+#define DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_RESERVED_1__SHIFT	13
+static inline uint32_t DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_RESERVED_1__SHIFT) & DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_RESERVED_1__MASK;
+}
+#define DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_HEIGHT__MASK		0x00001fff
+#define DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_HEIGHT__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_HEIGHT(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_HEIGHT__SHIFT) & DPU_RDMA_RDMA_DATA_CUBE_HEIGHT_HEIGHT__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_DATA_CUBE_CHANNEL			0x00005014
+#define DPU_RDMA_RDMA_DATA_CUBE_CHANNEL_RESERVED_0__MASK	0xffffe000
+#define DPU_RDMA_RDMA_DATA_CUBE_CHANNEL_RESERVED_0__SHIFT	13
+static inline uint32_t DPU_RDMA_RDMA_DATA_CUBE_CHANNEL_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_DATA_CUBE_CHANNEL_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_DATA_CUBE_CHANNEL_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_DATA_CUBE_CHANNEL_CHANNEL__MASK		0x00001fff
+#define DPU_RDMA_RDMA_DATA_CUBE_CHANNEL_CHANNEL__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_DATA_CUBE_CHANNEL_CHANNEL(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_DATA_CUBE_CHANNEL_CHANNEL__SHIFT) & DPU_RDMA_RDMA_DATA_CUBE_CHANNEL_CHANNEL__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_SRC_BASE_ADDR				0x00005018
+#define DPU_RDMA_RDMA_SRC_BASE_ADDR_SRC_BASE_ADDR__MASK		0xffffffff
+#define DPU_RDMA_RDMA_SRC_BASE_ADDR_SRC_BASE_ADDR__SHIFT	0
+static inline uint32_t DPU_RDMA_RDMA_SRC_BASE_ADDR_SRC_BASE_ADDR(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SRC_BASE_ADDR_SRC_BASE_ADDR__SHIFT) & DPU_RDMA_RDMA_SRC_BASE_ADDR_SRC_BASE_ADDR__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_BRDMA_CFG				0x0000501c
+#define DPU_RDMA_RDMA_BRDMA_CFG_RESERVED_0__MASK		0xffffffe0
+#define DPU_RDMA_RDMA_BRDMA_CFG_RESERVED_0__SHIFT		5
+static inline uint32_t DPU_RDMA_RDMA_BRDMA_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_BRDMA_CFG_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_BRDMA_CFG_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_BRDMA_CFG_BRDMA_DATA_USE__MASK		0x0000001e
+#define DPU_RDMA_RDMA_BRDMA_CFG_BRDMA_DATA_USE__SHIFT		1
+static inline uint32_t DPU_RDMA_RDMA_BRDMA_CFG_BRDMA_DATA_USE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_BRDMA_CFG_BRDMA_DATA_USE__SHIFT) & DPU_RDMA_RDMA_BRDMA_CFG_BRDMA_DATA_USE__MASK;
+}
+#define DPU_RDMA_RDMA_BRDMA_CFG_RESERVED_1__MASK		0x00000001
+#define DPU_RDMA_RDMA_BRDMA_CFG_RESERVED_1__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_BRDMA_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_BRDMA_CFG_RESERVED_1__SHIFT) & DPU_RDMA_RDMA_BRDMA_CFG_RESERVED_1__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_BS_BASE_ADDR				0x00005020
+#define DPU_RDMA_RDMA_BS_BASE_ADDR_BS_BASE_ADDR__MASK		0xffffffff
+#define DPU_RDMA_RDMA_BS_BASE_ADDR_BS_BASE_ADDR__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_BS_BASE_ADDR_BS_BASE_ADDR(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_BS_BASE_ADDR_BS_BASE_ADDR__SHIFT) & DPU_RDMA_RDMA_BS_BASE_ADDR_BS_BASE_ADDR__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_NRDMA_CFG				0x00005028
+#define DPU_RDMA_RDMA_NRDMA_CFG_RESERVED_0__MASK		0xffffffe0
+#define DPU_RDMA_RDMA_NRDMA_CFG_RESERVED_0__SHIFT		5
+static inline uint32_t DPU_RDMA_RDMA_NRDMA_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_NRDMA_CFG_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_NRDMA_CFG_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_NRDMA_CFG_NRDMA_DATA_USE__MASK		0x0000001e
+#define DPU_RDMA_RDMA_NRDMA_CFG_NRDMA_DATA_USE__SHIFT		1
+static inline uint32_t DPU_RDMA_RDMA_NRDMA_CFG_NRDMA_DATA_USE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_NRDMA_CFG_NRDMA_DATA_USE__SHIFT) & DPU_RDMA_RDMA_NRDMA_CFG_NRDMA_DATA_USE__MASK;
+}
+#define DPU_RDMA_RDMA_NRDMA_CFG_RESERVED_1__MASK		0x00000001
+#define DPU_RDMA_RDMA_NRDMA_CFG_RESERVED_1__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_NRDMA_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_NRDMA_CFG_RESERVED_1__SHIFT) & DPU_RDMA_RDMA_NRDMA_CFG_RESERVED_1__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_BN_BASE_ADDR				0x0000502c
+#define DPU_RDMA_RDMA_BN_BASE_ADDR_BN_BASE_ADDR__MASK		0xffffffff
+#define DPU_RDMA_RDMA_BN_BASE_ADDR_BN_BASE_ADDR__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_BN_BASE_ADDR_BN_BASE_ADDR(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_BN_BASE_ADDR_BN_BASE_ADDR__SHIFT) & DPU_RDMA_RDMA_BN_BASE_ADDR_BN_BASE_ADDR__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_ERDMA_CFG				0x00005034
+#define DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DATA_MODE__MASK		0xc0000000
+#define DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DATA_MODE__SHIFT		30
+static inline uint32_t DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DATA_MODE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DATA_MODE__SHIFT) & DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DATA_MODE__MASK;
+}
+#define DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_SURF_MODE__MASK		0x20000000
+#define DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_SURF_MODE__SHIFT		29
+static inline uint32_t DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_SURF_MODE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_SURF_MODE__SHIFT) & DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_SURF_MODE__MASK;
+}
+#define DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_NONALIGN__MASK		0x10000000
+#define DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_NONALIGN__SHIFT		28
+static inline uint32_t DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_NONALIGN(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_NONALIGN__SHIFT) & DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_NONALIGN__MASK;
+}
+#define DPU_RDMA_RDMA_ERDMA_CFG_RESERVED_0__MASK		0x0ffffff0
+#define DPU_RDMA_RDMA_ERDMA_CFG_RESERVED_0__SHIFT		4
+static inline uint32_t DPU_RDMA_RDMA_ERDMA_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_ERDMA_CFG_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_ERDMA_CFG_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DATA_SIZE__MASK		0x0000000c
+#define DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DATA_SIZE__SHIFT		2
+static inline uint32_t DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DATA_SIZE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DATA_SIZE__SHIFT) & DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DATA_SIZE__MASK;
+}
+#define DPU_RDMA_RDMA_ERDMA_CFG_OV4K_BYPASS__MASK		0x00000002
+#define DPU_RDMA_RDMA_ERDMA_CFG_OV4K_BYPASS__SHIFT		1
+static inline uint32_t DPU_RDMA_RDMA_ERDMA_CFG_OV4K_BYPASS(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_ERDMA_CFG_OV4K_BYPASS__SHIFT) & DPU_RDMA_RDMA_ERDMA_CFG_OV4K_BYPASS__MASK;
+}
+#define DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DISABLE__MASK		0x00000001
+#define DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DISABLE__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DISABLE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DISABLE__SHIFT) & DPU_RDMA_RDMA_ERDMA_CFG_ERDMA_DISABLE__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_EW_BASE_ADDR				0x00005038
+#define DPU_RDMA_RDMA_EW_BASE_ADDR_EW_BASE_ADDR__MASK		0xffffffff
+#define DPU_RDMA_RDMA_EW_BASE_ADDR_EW_BASE_ADDR__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_EW_BASE_ADDR_EW_BASE_ADDR(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_EW_BASE_ADDR_EW_BASE_ADDR__SHIFT) & DPU_RDMA_RDMA_EW_BASE_ADDR_EW_BASE_ADDR__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_EW_SURF_STRIDE			0x00005040
+#define DPU_RDMA_RDMA_EW_SURF_STRIDE_EW_SURF_STRIDE__MASK	0xfffffff0
+#define DPU_RDMA_RDMA_EW_SURF_STRIDE_EW_SURF_STRIDE__SHIFT	4
+static inline uint32_t DPU_RDMA_RDMA_EW_SURF_STRIDE_EW_SURF_STRIDE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_EW_SURF_STRIDE_EW_SURF_STRIDE__SHIFT) & DPU_RDMA_RDMA_EW_SURF_STRIDE_EW_SURF_STRIDE__MASK;
+}
+#define DPU_RDMA_RDMA_EW_SURF_STRIDE_RESERVED_0__MASK		0x0000000f
+#define DPU_RDMA_RDMA_EW_SURF_STRIDE_RESERVED_0__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_EW_SURF_STRIDE_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_EW_SURF_STRIDE_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_EW_SURF_STRIDE_RESERVED_0__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_FEATURE_MODE_CFG			0x00005044
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_RESERVED_0__MASK		0xfffc0000
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_RESERVED_0__SHIFT	18
+static inline uint32_t DPU_RDMA_RDMA_FEATURE_MODE_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_FEATURE_MODE_CFG_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_FEATURE_MODE_CFG_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_IN_PRECISION__MASK	0x00038000
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_IN_PRECISION__SHIFT	15
+static inline uint32_t DPU_RDMA_RDMA_FEATURE_MODE_CFG_IN_PRECISION(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_FEATURE_MODE_CFG_IN_PRECISION__SHIFT) & DPU_RDMA_RDMA_FEATURE_MODE_CFG_IN_PRECISION__MASK;
+}
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_BURST_LEN__MASK		0x00007800
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_BURST_LEN__SHIFT		11
+static inline uint32_t DPU_RDMA_RDMA_FEATURE_MODE_CFG_BURST_LEN(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_FEATURE_MODE_CFG_BURST_LEN__SHIFT) & DPU_RDMA_RDMA_FEATURE_MODE_CFG_BURST_LEN__MASK;
+}
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_COMB_USE__MASK		0x00000700
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_COMB_USE__SHIFT		8
+static inline uint32_t DPU_RDMA_RDMA_FEATURE_MODE_CFG_COMB_USE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_FEATURE_MODE_CFG_COMB_USE__SHIFT) & DPU_RDMA_RDMA_FEATURE_MODE_CFG_COMB_USE__MASK;
+}
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_PROC_PRECISION__MASK	0x000000e0
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_PROC_PRECISION__SHIFT	5
+static inline uint32_t DPU_RDMA_RDMA_FEATURE_MODE_CFG_PROC_PRECISION(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_FEATURE_MODE_CFG_PROC_PRECISION__SHIFT) & DPU_RDMA_RDMA_FEATURE_MODE_CFG_PROC_PRECISION__MASK;
+}
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_MRDMA_DISABLE__MASK	0x00000010
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_MRDMA_DISABLE__SHIFT	4
+static inline uint32_t DPU_RDMA_RDMA_FEATURE_MODE_CFG_MRDMA_DISABLE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_FEATURE_MODE_CFG_MRDMA_DISABLE__SHIFT) & DPU_RDMA_RDMA_FEATURE_MODE_CFG_MRDMA_DISABLE__MASK;
+}
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_MRDMA_FP16TOFP32_EN__MASK	0x00000008
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_MRDMA_FP16TOFP32_EN__SHIFT	3
+static inline uint32_t DPU_RDMA_RDMA_FEATURE_MODE_CFG_MRDMA_FP16TOFP32_EN(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_FEATURE_MODE_CFG_MRDMA_FP16TOFP32_EN__SHIFT) & DPU_RDMA_RDMA_FEATURE_MODE_CFG_MRDMA_FP16TOFP32_EN__MASK;
+}
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_CONV_MODE__MASK		0x00000006
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_CONV_MODE__SHIFT		1
+static inline uint32_t DPU_RDMA_RDMA_FEATURE_MODE_CFG_CONV_MODE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_FEATURE_MODE_CFG_CONV_MODE__SHIFT) & DPU_RDMA_RDMA_FEATURE_MODE_CFG_CONV_MODE__MASK;
+}
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_FLYING_MODE__MASK	0x00000001
+#define DPU_RDMA_RDMA_FEATURE_MODE_CFG_FLYING_MODE__SHIFT	0
+static inline uint32_t DPU_RDMA_RDMA_FEATURE_MODE_CFG_FLYING_MODE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_FEATURE_MODE_CFG_FLYING_MODE__SHIFT) & DPU_RDMA_RDMA_FEATURE_MODE_CFG_FLYING_MODE__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_SRC_DMA_CFG				0x00005048
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_LINE_NOTCH_ADDR__MASK		0xfff80000
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_LINE_NOTCH_ADDR__SHIFT	19
+static inline uint32_t DPU_RDMA_RDMA_SRC_DMA_CFG_LINE_NOTCH_ADDR(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SRC_DMA_CFG_LINE_NOTCH_ADDR__SHIFT) & DPU_RDMA_RDMA_SRC_DMA_CFG_LINE_NOTCH_ADDR__MASK;
+}
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_RESERVED_0__MASK		0x0007c000
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_RESERVED_0__SHIFT		14
+static inline uint32_t DPU_RDMA_RDMA_SRC_DMA_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SRC_DMA_CFG_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_SRC_DMA_CFG_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_POOLING_METHOD__MASK		0x00002000
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_POOLING_METHOD__SHIFT		13
+static inline uint32_t DPU_RDMA_RDMA_SRC_DMA_CFG_POOLING_METHOD(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SRC_DMA_CFG_POOLING_METHOD__SHIFT) & DPU_RDMA_RDMA_SRC_DMA_CFG_POOLING_METHOD__MASK;
+}
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_UNPOOLING_EN__MASK		0x00001000
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_UNPOOLING_EN__SHIFT		12
+static inline uint32_t DPU_RDMA_RDMA_SRC_DMA_CFG_UNPOOLING_EN(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SRC_DMA_CFG_UNPOOLING_EN__SHIFT) & DPU_RDMA_RDMA_SRC_DMA_CFG_UNPOOLING_EN__MASK;
+}
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_STRIDE_HEIGHT__MASK	0x00000e00
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_STRIDE_HEIGHT__SHIFT	9
+static inline uint32_t DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_STRIDE_HEIGHT(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_STRIDE_HEIGHT__SHIFT) & DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_STRIDE_HEIGHT__MASK;
+}
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_STRIDE_WIDTH__MASK	0x000001c0
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_STRIDE_WIDTH__SHIFT	6
+static inline uint32_t DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_STRIDE_WIDTH(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_STRIDE_WIDTH__SHIFT) & DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_STRIDE_WIDTH__MASK;
+}
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_HEIGHT__MASK		0x00000038
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_HEIGHT__SHIFT		3
+static inline uint32_t DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_HEIGHT(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_HEIGHT__SHIFT) & DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_HEIGHT__MASK;
+}
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_WIDTH__MASK		0x00000007
+#define DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_WIDTH__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_WIDTH(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_WIDTH__SHIFT) & DPU_RDMA_RDMA_SRC_DMA_CFG_KERNEL_WIDTH__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_SURF_NOTCH				0x0000504c
+#define DPU_RDMA_RDMA_SURF_NOTCH_SURF_NOTCH_ADDR__MASK		0xfffffff0
+#define DPU_RDMA_RDMA_SURF_NOTCH_SURF_NOTCH_ADDR__SHIFT		4
+static inline uint32_t DPU_RDMA_RDMA_SURF_NOTCH_SURF_NOTCH_ADDR(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SURF_NOTCH_SURF_NOTCH_ADDR__SHIFT) & DPU_RDMA_RDMA_SURF_NOTCH_SURF_NOTCH_ADDR__MASK;
+}
+#define DPU_RDMA_RDMA_SURF_NOTCH_RESERVED_0__MASK		0x0000000f
+#define DPU_RDMA_RDMA_SURF_NOTCH_RESERVED_0__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_SURF_NOTCH_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_SURF_NOTCH_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_SURF_NOTCH_RESERVED_0__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_PAD_CFG				0x00005064
+#define DPU_RDMA_RDMA_PAD_CFG_PAD_VALUE__MASK			0xffff0000
+#define DPU_RDMA_RDMA_PAD_CFG_PAD_VALUE__SHIFT			16
+static inline uint32_t DPU_RDMA_RDMA_PAD_CFG_PAD_VALUE(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_PAD_CFG_PAD_VALUE__SHIFT) & DPU_RDMA_RDMA_PAD_CFG_PAD_VALUE__MASK;
+}
+#define DPU_RDMA_RDMA_PAD_CFG_RESERVED_0__MASK			0x0000ff80
+#define DPU_RDMA_RDMA_PAD_CFG_RESERVED_0__SHIFT			7
+static inline uint32_t DPU_RDMA_RDMA_PAD_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_PAD_CFG_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_PAD_CFG_RESERVED_0__MASK;
+}
+#define DPU_RDMA_RDMA_PAD_CFG_PAD_TOP__MASK			0x00000070
+#define DPU_RDMA_RDMA_PAD_CFG_PAD_TOP__SHIFT			4
+static inline uint32_t DPU_RDMA_RDMA_PAD_CFG_PAD_TOP(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_PAD_CFG_PAD_TOP__SHIFT) & DPU_RDMA_RDMA_PAD_CFG_PAD_TOP__MASK;
+}
+#define DPU_RDMA_RDMA_PAD_CFG_RESERVED_1__MASK			0x00000008
+#define DPU_RDMA_RDMA_PAD_CFG_RESERVED_1__SHIFT			3
+static inline uint32_t DPU_RDMA_RDMA_PAD_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_PAD_CFG_RESERVED_1__SHIFT) & DPU_RDMA_RDMA_PAD_CFG_RESERVED_1__MASK;
+}
+#define DPU_RDMA_RDMA_PAD_CFG_PAD_LEFT__MASK			0x00000007
+#define DPU_RDMA_RDMA_PAD_CFG_PAD_LEFT__SHIFT			0
+static inline uint32_t DPU_RDMA_RDMA_PAD_CFG_PAD_LEFT(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_PAD_CFG_PAD_LEFT__SHIFT) & DPU_RDMA_RDMA_PAD_CFG_PAD_LEFT__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_WEIGHT				0x00005068
+#define DPU_RDMA_RDMA_WEIGHT_E_WEIGHT__MASK			0xff000000
+#define DPU_RDMA_RDMA_WEIGHT_E_WEIGHT__SHIFT			24
+static inline uint32_t DPU_RDMA_RDMA_WEIGHT_E_WEIGHT(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_WEIGHT_E_WEIGHT__SHIFT) & DPU_RDMA_RDMA_WEIGHT_E_WEIGHT__MASK;
+}
+#define DPU_RDMA_RDMA_WEIGHT_N_WEIGHT__MASK			0x00ff0000
+#define DPU_RDMA_RDMA_WEIGHT_N_WEIGHT__SHIFT			16
+static inline uint32_t DPU_RDMA_RDMA_WEIGHT_N_WEIGHT(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_WEIGHT_N_WEIGHT__SHIFT) & DPU_RDMA_RDMA_WEIGHT_N_WEIGHT__MASK;
+}
+#define DPU_RDMA_RDMA_WEIGHT_B_WEIGHT__MASK			0x0000ff00
+#define DPU_RDMA_RDMA_WEIGHT_B_WEIGHT__SHIFT			8
+static inline uint32_t DPU_RDMA_RDMA_WEIGHT_B_WEIGHT(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_WEIGHT_B_WEIGHT__SHIFT) & DPU_RDMA_RDMA_WEIGHT_B_WEIGHT__MASK;
+}
+#define DPU_RDMA_RDMA_WEIGHT_M_WEIGHT__MASK			0x000000ff
+#define DPU_RDMA_RDMA_WEIGHT_M_WEIGHT__SHIFT			0
+static inline uint32_t DPU_RDMA_RDMA_WEIGHT_M_WEIGHT(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_WEIGHT_M_WEIGHT__SHIFT) & DPU_RDMA_RDMA_WEIGHT_M_WEIGHT__MASK;
+}
+
+#define REG_DPU_RDMA_RDMA_EW_SURF_NOTCH				0x0000506c
+#define DPU_RDMA_RDMA_EW_SURF_NOTCH_EW_SURF_NOTCH__MASK		0xfffffff0
+#define DPU_RDMA_RDMA_EW_SURF_NOTCH_EW_SURF_NOTCH__SHIFT	4
+static inline uint32_t DPU_RDMA_RDMA_EW_SURF_NOTCH_EW_SURF_NOTCH(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_EW_SURF_NOTCH_EW_SURF_NOTCH__SHIFT) & DPU_RDMA_RDMA_EW_SURF_NOTCH_EW_SURF_NOTCH__MASK;
+}
+#define DPU_RDMA_RDMA_EW_SURF_NOTCH_RESERVED_0__MASK		0x0000000f
+#define DPU_RDMA_RDMA_EW_SURF_NOTCH_RESERVED_0__SHIFT		0
+static inline uint32_t DPU_RDMA_RDMA_EW_SURF_NOTCH_RESERVED_0(uint32_t val)
+{
+	return ((val) << DPU_RDMA_RDMA_EW_SURF_NOTCH_RESERVED_0__SHIFT) & DPU_RDMA_RDMA_EW_SURF_NOTCH_RESERVED_0__MASK;
+}
+
+#define REG_PPU_S_STATUS					0x00006000
+#define PPU_S_STATUS_RESERVED_0__MASK				0xfffc0000
+#define PPU_S_STATUS_RESERVED_0__SHIFT				18
+static inline uint32_t PPU_S_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_S_STATUS_RESERVED_0__SHIFT) & PPU_S_STATUS_RESERVED_0__MASK;
+}
+#define PPU_S_STATUS_STATUS_1__MASK				0x00030000
+#define PPU_S_STATUS_STATUS_1__SHIFT				16
+static inline uint32_t PPU_S_STATUS_STATUS_1(uint32_t val)
+{
+	return ((val) << PPU_S_STATUS_STATUS_1__SHIFT) & PPU_S_STATUS_STATUS_1__MASK;
+}
+#define PPU_S_STATUS_RESERVED_1__MASK				0x0000fffc
+#define PPU_S_STATUS_RESERVED_1__SHIFT				2
+static inline uint32_t PPU_S_STATUS_RESERVED_1(uint32_t val)
+{
+	return ((val) << PPU_S_STATUS_RESERVED_1__SHIFT) & PPU_S_STATUS_RESERVED_1__MASK;
+}
+#define PPU_S_STATUS_STATUS_0__MASK				0x00000003
+#define PPU_S_STATUS_STATUS_0__SHIFT				0
+static inline uint32_t PPU_S_STATUS_STATUS_0(uint32_t val)
+{
+	return ((val) << PPU_S_STATUS_STATUS_0__SHIFT) & PPU_S_STATUS_STATUS_0__MASK;
+}
+
+#define REG_PPU_S_POINTER					0x00006004
+#define PPU_S_POINTER_RESERVED_0__MASK				0xfffe0000
+#define PPU_S_POINTER_RESERVED_0__SHIFT				17
+static inline uint32_t PPU_S_POINTER_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_S_POINTER_RESERVED_0__SHIFT) & PPU_S_POINTER_RESERVED_0__MASK;
+}
+#define PPU_S_POINTER_EXECUTER__MASK				0x00010000
+#define PPU_S_POINTER_EXECUTER__SHIFT				16
+static inline uint32_t PPU_S_POINTER_EXECUTER(uint32_t val)
+{
+	return ((val) << PPU_S_POINTER_EXECUTER__SHIFT) & PPU_S_POINTER_EXECUTER__MASK;
+}
+#define PPU_S_POINTER_RESERVED_1__MASK				0x0000ffc0
+#define PPU_S_POINTER_RESERVED_1__SHIFT				6
+static inline uint32_t PPU_S_POINTER_RESERVED_1(uint32_t val)
+{
+	return ((val) << PPU_S_POINTER_RESERVED_1__SHIFT) & PPU_S_POINTER_RESERVED_1__MASK;
+}
+#define PPU_S_POINTER_EXECUTER_PP_CLEAR__MASK			0x00000020
+#define PPU_S_POINTER_EXECUTER_PP_CLEAR__SHIFT			5
+static inline uint32_t PPU_S_POINTER_EXECUTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << PPU_S_POINTER_EXECUTER_PP_CLEAR__SHIFT) & PPU_S_POINTER_EXECUTER_PP_CLEAR__MASK;
+}
+#define PPU_S_POINTER_POINTER_PP_CLEAR__MASK			0x00000010
+#define PPU_S_POINTER_POINTER_PP_CLEAR__SHIFT			4
+static inline uint32_t PPU_S_POINTER_POINTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << PPU_S_POINTER_POINTER_PP_CLEAR__SHIFT) & PPU_S_POINTER_POINTER_PP_CLEAR__MASK;
+}
+#define PPU_S_POINTER_POINTER_PP_MODE__MASK			0x00000008
+#define PPU_S_POINTER_POINTER_PP_MODE__SHIFT			3
+static inline uint32_t PPU_S_POINTER_POINTER_PP_MODE(uint32_t val)
+{
+	return ((val) << PPU_S_POINTER_POINTER_PP_MODE__SHIFT) & PPU_S_POINTER_POINTER_PP_MODE__MASK;
+}
+#define PPU_S_POINTER_EXECUTER_PP_EN__MASK			0x00000004
+#define PPU_S_POINTER_EXECUTER_PP_EN__SHIFT			2
+static inline uint32_t PPU_S_POINTER_EXECUTER_PP_EN(uint32_t val)
+{
+	return ((val) << PPU_S_POINTER_EXECUTER_PP_EN__SHIFT) & PPU_S_POINTER_EXECUTER_PP_EN__MASK;
+}
+#define PPU_S_POINTER_POINTER_PP_EN__MASK			0x00000002
+#define PPU_S_POINTER_POINTER_PP_EN__SHIFT			1
+static inline uint32_t PPU_S_POINTER_POINTER_PP_EN(uint32_t val)
+{
+	return ((val) << PPU_S_POINTER_POINTER_PP_EN__SHIFT) & PPU_S_POINTER_POINTER_PP_EN__MASK;
+}
+#define PPU_S_POINTER_POINTER__MASK				0x00000001
+#define PPU_S_POINTER_POINTER__SHIFT				0
+static inline uint32_t PPU_S_POINTER_POINTER(uint32_t val)
+{
+	return ((val) << PPU_S_POINTER_POINTER__SHIFT) & PPU_S_POINTER_POINTER__MASK;
+}
+
+#define REG_PPU_OPERATION_ENABLE				0x00006008
+#define PPU_OPERATION_ENABLE_RESERVED_0__MASK			0xfffffffe
+#define PPU_OPERATION_ENABLE_RESERVED_0__SHIFT			1
+static inline uint32_t PPU_OPERATION_ENABLE_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_ENABLE_RESERVED_0__SHIFT) & PPU_OPERATION_ENABLE_RESERVED_0__MASK;
+}
+#define PPU_OPERATION_ENABLE_OP_EN__MASK			0x00000001
+#define PPU_OPERATION_ENABLE_OP_EN__SHIFT			0
+static inline uint32_t PPU_OPERATION_ENABLE_OP_EN(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_ENABLE_OP_EN__SHIFT) & PPU_OPERATION_ENABLE_OP_EN__MASK;
+}
+
+#define REG_PPU_DATA_CUBE_IN_WIDTH				0x0000600c
+#define PPU_DATA_CUBE_IN_WIDTH_RESERVED_0__MASK			0xffffe000
+#define PPU_DATA_CUBE_IN_WIDTH_RESERVED_0__SHIFT		13
+static inline uint32_t PPU_DATA_CUBE_IN_WIDTH_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_IN_WIDTH_RESERVED_0__SHIFT) & PPU_DATA_CUBE_IN_WIDTH_RESERVED_0__MASK;
+}
+#define PPU_DATA_CUBE_IN_WIDTH_CUBE_IN_WIDTH__MASK		0x00001fff
+#define PPU_DATA_CUBE_IN_WIDTH_CUBE_IN_WIDTH__SHIFT		0
+static inline uint32_t PPU_DATA_CUBE_IN_WIDTH_CUBE_IN_WIDTH(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_IN_WIDTH_CUBE_IN_WIDTH__SHIFT) & PPU_DATA_CUBE_IN_WIDTH_CUBE_IN_WIDTH__MASK;
+}
+
+#define REG_PPU_DATA_CUBE_IN_HEIGHT				0x00006010
+#define PPU_DATA_CUBE_IN_HEIGHT_RESERVED_0__MASK		0xffffe000
+#define PPU_DATA_CUBE_IN_HEIGHT_RESERVED_0__SHIFT		13
+static inline uint32_t PPU_DATA_CUBE_IN_HEIGHT_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_IN_HEIGHT_RESERVED_0__SHIFT) & PPU_DATA_CUBE_IN_HEIGHT_RESERVED_0__MASK;
+}
+#define PPU_DATA_CUBE_IN_HEIGHT_CUBE_IN_HEIGHT__MASK		0x00001fff
+#define PPU_DATA_CUBE_IN_HEIGHT_CUBE_IN_HEIGHT__SHIFT		0
+static inline uint32_t PPU_DATA_CUBE_IN_HEIGHT_CUBE_IN_HEIGHT(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_IN_HEIGHT_CUBE_IN_HEIGHT__SHIFT) & PPU_DATA_CUBE_IN_HEIGHT_CUBE_IN_HEIGHT__MASK;
+}
+
+#define REG_PPU_DATA_CUBE_IN_CHANNEL				0x00006014
+#define PPU_DATA_CUBE_IN_CHANNEL_RESERVED_0__MASK		0xffffe000
+#define PPU_DATA_CUBE_IN_CHANNEL_RESERVED_0__SHIFT		13
+static inline uint32_t PPU_DATA_CUBE_IN_CHANNEL_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_IN_CHANNEL_RESERVED_0__SHIFT) & PPU_DATA_CUBE_IN_CHANNEL_RESERVED_0__MASK;
+}
+#define PPU_DATA_CUBE_IN_CHANNEL_CUBE_IN_CHANNEL__MASK		0x00001fff
+#define PPU_DATA_CUBE_IN_CHANNEL_CUBE_IN_CHANNEL__SHIFT		0
+static inline uint32_t PPU_DATA_CUBE_IN_CHANNEL_CUBE_IN_CHANNEL(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_IN_CHANNEL_CUBE_IN_CHANNEL__SHIFT) & PPU_DATA_CUBE_IN_CHANNEL_CUBE_IN_CHANNEL__MASK;
+}
+
+#define REG_PPU_DATA_CUBE_OUT_WIDTH				0x00006018
+#define PPU_DATA_CUBE_OUT_WIDTH_RESERVED_0__MASK		0xffffe000
+#define PPU_DATA_CUBE_OUT_WIDTH_RESERVED_0__SHIFT		13
+static inline uint32_t PPU_DATA_CUBE_OUT_WIDTH_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_OUT_WIDTH_RESERVED_0__SHIFT) & PPU_DATA_CUBE_OUT_WIDTH_RESERVED_0__MASK;
+}
+#define PPU_DATA_CUBE_OUT_WIDTH_CUBE_OUT_WIDTH__MASK		0x00001fff
+#define PPU_DATA_CUBE_OUT_WIDTH_CUBE_OUT_WIDTH__SHIFT		0
+static inline uint32_t PPU_DATA_CUBE_OUT_WIDTH_CUBE_OUT_WIDTH(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_OUT_WIDTH_CUBE_OUT_WIDTH__SHIFT) & PPU_DATA_CUBE_OUT_WIDTH_CUBE_OUT_WIDTH__MASK;
+}
+
+#define REG_PPU_DATA_CUBE_OUT_HEIGHT				0x0000601c
+#define PPU_DATA_CUBE_OUT_HEIGHT_RESERVED_0__MASK		0xffffe000
+#define PPU_DATA_CUBE_OUT_HEIGHT_RESERVED_0__SHIFT		13
+static inline uint32_t PPU_DATA_CUBE_OUT_HEIGHT_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_OUT_HEIGHT_RESERVED_0__SHIFT) & PPU_DATA_CUBE_OUT_HEIGHT_RESERVED_0__MASK;
+}
+#define PPU_DATA_CUBE_OUT_HEIGHT_CUBE_OUT_HEIGHT__MASK		0x00001fff
+#define PPU_DATA_CUBE_OUT_HEIGHT_CUBE_OUT_HEIGHT__SHIFT		0
+static inline uint32_t PPU_DATA_CUBE_OUT_HEIGHT_CUBE_OUT_HEIGHT(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_OUT_HEIGHT_CUBE_OUT_HEIGHT__SHIFT) & PPU_DATA_CUBE_OUT_HEIGHT_CUBE_OUT_HEIGHT__MASK;
+}
+
+#define REG_PPU_DATA_CUBE_OUT_CHANNEL				0x00006020
+#define PPU_DATA_CUBE_OUT_CHANNEL_RESERVED_0__MASK		0xffffe000
+#define PPU_DATA_CUBE_OUT_CHANNEL_RESERVED_0__SHIFT		13
+static inline uint32_t PPU_DATA_CUBE_OUT_CHANNEL_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_OUT_CHANNEL_RESERVED_0__SHIFT) & PPU_DATA_CUBE_OUT_CHANNEL_RESERVED_0__MASK;
+}
+#define PPU_DATA_CUBE_OUT_CHANNEL_CUBE_OUT_CHANNEL__MASK	0x00001fff
+#define PPU_DATA_CUBE_OUT_CHANNEL_CUBE_OUT_CHANNEL__SHIFT	0
+static inline uint32_t PPU_DATA_CUBE_OUT_CHANNEL_CUBE_OUT_CHANNEL(uint32_t val)
+{
+	return ((val) << PPU_DATA_CUBE_OUT_CHANNEL_CUBE_OUT_CHANNEL__SHIFT) & PPU_DATA_CUBE_OUT_CHANNEL_CUBE_OUT_CHANNEL__MASK;
+}
+
+#define REG_PPU_OPERATION_MODE_CFG				0x00006024
+#define PPU_OPERATION_MODE_CFG_RESERVED_0__MASK			0x80000000
+#define PPU_OPERATION_MODE_CFG_RESERVED_0__SHIFT		31
+static inline uint32_t PPU_OPERATION_MODE_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_MODE_CFG_RESERVED_0__SHIFT) & PPU_OPERATION_MODE_CFG_RESERVED_0__MASK;
+}
+#define PPU_OPERATION_MODE_CFG_INDEX_EN__MASK			0x40000000
+#define PPU_OPERATION_MODE_CFG_INDEX_EN__SHIFT			30
+static inline uint32_t PPU_OPERATION_MODE_CFG_INDEX_EN(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_MODE_CFG_INDEX_EN__SHIFT) & PPU_OPERATION_MODE_CFG_INDEX_EN__MASK;
+}
+#define PPU_OPERATION_MODE_CFG_RESERVED_1__MASK			0x20000000
+#define PPU_OPERATION_MODE_CFG_RESERVED_1__SHIFT		29
+static inline uint32_t PPU_OPERATION_MODE_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_MODE_CFG_RESERVED_1__SHIFT) & PPU_OPERATION_MODE_CFG_RESERVED_1__MASK;
+}
+#define PPU_OPERATION_MODE_CFG_NOTCH_ADDR__MASK			0x1fff0000
+#define PPU_OPERATION_MODE_CFG_NOTCH_ADDR__SHIFT		16
+static inline uint32_t PPU_OPERATION_MODE_CFG_NOTCH_ADDR(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_MODE_CFG_NOTCH_ADDR__SHIFT) & PPU_OPERATION_MODE_CFG_NOTCH_ADDR__MASK;
+}
+#define PPU_OPERATION_MODE_CFG_RESERVED_2__MASK			0x0000ff00
+#define PPU_OPERATION_MODE_CFG_RESERVED_2__SHIFT		8
+static inline uint32_t PPU_OPERATION_MODE_CFG_RESERVED_2(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_MODE_CFG_RESERVED_2__SHIFT) & PPU_OPERATION_MODE_CFG_RESERVED_2__MASK;
+}
+#define PPU_OPERATION_MODE_CFG_USE_CNT__MASK			0x000000e0
+#define PPU_OPERATION_MODE_CFG_USE_CNT__SHIFT			5
+static inline uint32_t PPU_OPERATION_MODE_CFG_USE_CNT(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_MODE_CFG_USE_CNT__SHIFT) & PPU_OPERATION_MODE_CFG_USE_CNT__MASK;
+}
+#define PPU_OPERATION_MODE_CFG_FLYING_MODE__MASK		0x00000010
+#define PPU_OPERATION_MODE_CFG_FLYING_MODE__SHIFT		4
+static inline uint32_t PPU_OPERATION_MODE_CFG_FLYING_MODE(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_MODE_CFG_FLYING_MODE__SHIFT) & PPU_OPERATION_MODE_CFG_FLYING_MODE__MASK;
+}
+#define PPU_OPERATION_MODE_CFG_RESERVED_3__MASK			0x0000000c
+#define PPU_OPERATION_MODE_CFG_RESERVED_3__SHIFT		2
+static inline uint32_t PPU_OPERATION_MODE_CFG_RESERVED_3(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_MODE_CFG_RESERVED_3__SHIFT) & PPU_OPERATION_MODE_CFG_RESERVED_3__MASK;
+}
+#define PPU_OPERATION_MODE_CFG_POOLING_METHOD__MASK		0x00000003
+#define PPU_OPERATION_MODE_CFG_POOLING_METHOD__SHIFT		0
+static inline uint32_t PPU_OPERATION_MODE_CFG_POOLING_METHOD(uint32_t val)
+{
+	return ((val) << PPU_OPERATION_MODE_CFG_POOLING_METHOD__SHIFT) & PPU_OPERATION_MODE_CFG_POOLING_METHOD__MASK;
+}
+
+#define REG_PPU_POOLING_KERNEL_CFG				0x00006034
+#define PPU_POOLING_KERNEL_CFG_RESERVED_0__MASK			0xff000000
+#define PPU_POOLING_KERNEL_CFG_RESERVED_0__SHIFT		24
+static inline uint32_t PPU_POOLING_KERNEL_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_POOLING_KERNEL_CFG_RESERVED_0__SHIFT) & PPU_POOLING_KERNEL_CFG_RESERVED_0__MASK;
+}
+#define PPU_POOLING_KERNEL_CFG_KERNEL_STRIDE_HEIGHT__MASK	0x00f00000
+#define PPU_POOLING_KERNEL_CFG_KERNEL_STRIDE_HEIGHT__SHIFT	20
+static inline uint32_t PPU_POOLING_KERNEL_CFG_KERNEL_STRIDE_HEIGHT(uint32_t val)
+{
+	return ((val) << PPU_POOLING_KERNEL_CFG_KERNEL_STRIDE_HEIGHT__SHIFT) & PPU_POOLING_KERNEL_CFG_KERNEL_STRIDE_HEIGHT__MASK;
+}
+#define PPU_POOLING_KERNEL_CFG_KERNEL_STRIDE_WIDTH__MASK	0x000f0000
+#define PPU_POOLING_KERNEL_CFG_KERNEL_STRIDE_WIDTH__SHIFT	16
+static inline uint32_t PPU_POOLING_KERNEL_CFG_KERNEL_STRIDE_WIDTH(uint32_t val)
+{
+	return ((val) << PPU_POOLING_KERNEL_CFG_KERNEL_STRIDE_WIDTH__SHIFT) & PPU_POOLING_KERNEL_CFG_KERNEL_STRIDE_WIDTH__MASK;
+}
+#define PPU_POOLING_KERNEL_CFG_RESERVED_1__MASK			0x0000f000
+#define PPU_POOLING_KERNEL_CFG_RESERVED_1__SHIFT		12
+static inline uint32_t PPU_POOLING_KERNEL_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << PPU_POOLING_KERNEL_CFG_RESERVED_1__SHIFT) & PPU_POOLING_KERNEL_CFG_RESERVED_1__MASK;
+}
+#define PPU_POOLING_KERNEL_CFG_KERNEL_HEIGHT__MASK		0x00000f00
+#define PPU_POOLING_KERNEL_CFG_KERNEL_HEIGHT__SHIFT		8
+static inline uint32_t PPU_POOLING_KERNEL_CFG_KERNEL_HEIGHT(uint32_t val)
+{
+	return ((val) << PPU_POOLING_KERNEL_CFG_KERNEL_HEIGHT__SHIFT) & PPU_POOLING_KERNEL_CFG_KERNEL_HEIGHT__MASK;
+}
+#define PPU_POOLING_KERNEL_CFG_RESERVED_2__MASK			0x000000f0
+#define PPU_POOLING_KERNEL_CFG_RESERVED_2__SHIFT		4
+static inline uint32_t PPU_POOLING_KERNEL_CFG_RESERVED_2(uint32_t val)
+{
+	return ((val) << PPU_POOLING_KERNEL_CFG_RESERVED_2__SHIFT) & PPU_POOLING_KERNEL_CFG_RESERVED_2__MASK;
+}
+#define PPU_POOLING_KERNEL_CFG_KERNEL_WIDTH__MASK		0x0000000f
+#define PPU_POOLING_KERNEL_CFG_KERNEL_WIDTH__SHIFT		0
+static inline uint32_t PPU_POOLING_KERNEL_CFG_KERNEL_WIDTH(uint32_t val)
+{
+	return ((val) << PPU_POOLING_KERNEL_CFG_KERNEL_WIDTH__SHIFT) & PPU_POOLING_KERNEL_CFG_KERNEL_WIDTH__MASK;
+}
+
+#define REG_PPU_RECIP_KERNEL_WIDTH				0x00006038
+#define PPU_RECIP_KERNEL_WIDTH_RESERVED_0__MASK			0xfffe0000
+#define PPU_RECIP_KERNEL_WIDTH_RESERVED_0__SHIFT		17
+static inline uint32_t PPU_RECIP_KERNEL_WIDTH_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RECIP_KERNEL_WIDTH_RESERVED_0__SHIFT) & PPU_RECIP_KERNEL_WIDTH_RESERVED_0__MASK;
+}
+#define PPU_RECIP_KERNEL_WIDTH_RECIP_KERNEL_WIDTH__MASK		0x0001ffff
+#define PPU_RECIP_KERNEL_WIDTH_RECIP_KERNEL_WIDTH__SHIFT	0
+static inline uint32_t PPU_RECIP_KERNEL_WIDTH_RECIP_KERNEL_WIDTH(uint32_t val)
+{
+	return ((val) << PPU_RECIP_KERNEL_WIDTH_RECIP_KERNEL_WIDTH__SHIFT) & PPU_RECIP_KERNEL_WIDTH_RECIP_KERNEL_WIDTH__MASK;
+}
+
+#define REG_PPU_RECIP_KERNEL_HEIGHT				0x0000603c
+#define PPU_RECIP_KERNEL_HEIGHT_RESERVED_0__MASK		0xfffe0000
+#define PPU_RECIP_KERNEL_HEIGHT_RESERVED_0__SHIFT		17
+static inline uint32_t PPU_RECIP_KERNEL_HEIGHT_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RECIP_KERNEL_HEIGHT_RESERVED_0__SHIFT) & PPU_RECIP_KERNEL_HEIGHT_RESERVED_0__MASK;
+}
+#define PPU_RECIP_KERNEL_HEIGHT_RECIP_KERNEL_HEIGHT__MASK	0x0001ffff
+#define PPU_RECIP_KERNEL_HEIGHT_RECIP_KERNEL_HEIGHT__SHIFT	0
+static inline uint32_t PPU_RECIP_KERNEL_HEIGHT_RECIP_KERNEL_HEIGHT(uint32_t val)
+{
+	return ((val) << PPU_RECIP_KERNEL_HEIGHT_RECIP_KERNEL_HEIGHT__SHIFT) & PPU_RECIP_KERNEL_HEIGHT_RECIP_KERNEL_HEIGHT__MASK;
+}
+
+#define REG_PPU_POOLING_PADDING_CFG				0x00006040
+#define PPU_POOLING_PADDING_CFG_RESERVED_0__MASK		0xffff8000
+#define PPU_POOLING_PADDING_CFG_RESERVED_0__SHIFT		15
+static inline uint32_t PPU_POOLING_PADDING_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_POOLING_PADDING_CFG_RESERVED_0__SHIFT) & PPU_POOLING_PADDING_CFG_RESERVED_0__MASK;
+}
+#define PPU_POOLING_PADDING_CFG_PAD_BOTTOM__MASK		0x00007000
+#define PPU_POOLING_PADDING_CFG_PAD_BOTTOM__SHIFT		12
+static inline uint32_t PPU_POOLING_PADDING_CFG_PAD_BOTTOM(uint32_t val)
+{
+	return ((val) << PPU_POOLING_PADDING_CFG_PAD_BOTTOM__SHIFT) & PPU_POOLING_PADDING_CFG_PAD_BOTTOM__MASK;
+}
+#define PPU_POOLING_PADDING_CFG_RESERVED_1__MASK		0x00000800
+#define PPU_POOLING_PADDING_CFG_RESERVED_1__SHIFT		11
+static inline uint32_t PPU_POOLING_PADDING_CFG_RESERVED_1(uint32_t val)
+{
+	return ((val) << PPU_POOLING_PADDING_CFG_RESERVED_1__SHIFT) & PPU_POOLING_PADDING_CFG_RESERVED_1__MASK;
+}
+#define PPU_POOLING_PADDING_CFG_PAD_RIGHT__MASK			0x00000700
+#define PPU_POOLING_PADDING_CFG_PAD_RIGHT__SHIFT		8
+static inline uint32_t PPU_POOLING_PADDING_CFG_PAD_RIGHT(uint32_t val)
+{
+	return ((val) << PPU_POOLING_PADDING_CFG_PAD_RIGHT__SHIFT) & PPU_POOLING_PADDING_CFG_PAD_RIGHT__MASK;
+}
+#define PPU_POOLING_PADDING_CFG_RESERVED_2__MASK		0x00000080
+#define PPU_POOLING_PADDING_CFG_RESERVED_2__SHIFT		7
+static inline uint32_t PPU_POOLING_PADDING_CFG_RESERVED_2(uint32_t val)
+{
+	return ((val) << PPU_POOLING_PADDING_CFG_RESERVED_2__SHIFT) & PPU_POOLING_PADDING_CFG_RESERVED_2__MASK;
+}
+#define PPU_POOLING_PADDING_CFG_PAD_TOP__MASK			0x00000070
+#define PPU_POOLING_PADDING_CFG_PAD_TOP__SHIFT			4
+static inline uint32_t PPU_POOLING_PADDING_CFG_PAD_TOP(uint32_t val)
+{
+	return ((val) << PPU_POOLING_PADDING_CFG_PAD_TOP__SHIFT) & PPU_POOLING_PADDING_CFG_PAD_TOP__MASK;
+}
+#define PPU_POOLING_PADDING_CFG_RESERVED_3__MASK		0x00000008
+#define PPU_POOLING_PADDING_CFG_RESERVED_3__SHIFT		3
+static inline uint32_t PPU_POOLING_PADDING_CFG_RESERVED_3(uint32_t val)
+{
+	return ((val) << PPU_POOLING_PADDING_CFG_RESERVED_3__SHIFT) & PPU_POOLING_PADDING_CFG_RESERVED_3__MASK;
+}
+#define PPU_POOLING_PADDING_CFG_PAD_LEFT__MASK			0x00000007
+#define PPU_POOLING_PADDING_CFG_PAD_LEFT__SHIFT			0
+static inline uint32_t PPU_POOLING_PADDING_CFG_PAD_LEFT(uint32_t val)
+{
+	return ((val) << PPU_POOLING_PADDING_CFG_PAD_LEFT__SHIFT) & PPU_POOLING_PADDING_CFG_PAD_LEFT__MASK;
+}
+
+#define REG_PPU_PADDING_VALUE_1_CFG				0x00006044
+#define PPU_PADDING_VALUE_1_CFG_PAD_VALUE_0__MASK		0xffffffff
+#define PPU_PADDING_VALUE_1_CFG_PAD_VALUE_0__SHIFT		0
+static inline uint32_t PPU_PADDING_VALUE_1_CFG_PAD_VALUE_0(uint32_t val)
+{
+	return ((val) << PPU_PADDING_VALUE_1_CFG_PAD_VALUE_0__SHIFT) & PPU_PADDING_VALUE_1_CFG_PAD_VALUE_0__MASK;
+}
+
+#define REG_PPU_PADDING_VALUE_2_CFG				0x00006048
+#define PPU_PADDING_VALUE_2_CFG_RESERVED_0__MASK		0xfffffff8
+#define PPU_PADDING_VALUE_2_CFG_RESERVED_0__SHIFT		3
+static inline uint32_t PPU_PADDING_VALUE_2_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_PADDING_VALUE_2_CFG_RESERVED_0__SHIFT) & PPU_PADDING_VALUE_2_CFG_RESERVED_0__MASK;
+}
+#define PPU_PADDING_VALUE_2_CFG_PAD_VALUE_1__MASK		0x00000007
+#define PPU_PADDING_VALUE_2_CFG_PAD_VALUE_1__SHIFT		0
+static inline uint32_t PPU_PADDING_VALUE_2_CFG_PAD_VALUE_1(uint32_t val)
+{
+	return ((val) << PPU_PADDING_VALUE_2_CFG_PAD_VALUE_1__SHIFT) & PPU_PADDING_VALUE_2_CFG_PAD_VALUE_1__MASK;
+}
+
+#define REG_PPU_DST_BASE_ADDR					0x00006070
+#define PPU_DST_BASE_ADDR_DST_BASE_ADDR__MASK			0xfffffff0
+#define PPU_DST_BASE_ADDR_DST_BASE_ADDR__SHIFT			4
+static inline uint32_t PPU_DST_BASE_ADDR_DST_BASE_ADDR(uint32_t val)
+{
+	return ((val) << PPU_DST_BASE_ADDR_DST_BASE_ADDR__SHIFT) & PPU_DST_BASE_ADDR_DST_BASE_ADDR__MASK;
+}
+#define PPU_DST_BASE_ADDR_RESERVED_0__MASK			0x0000000f
+#define PPU_DST_BASE_ADDR_RESERVED_0__SHIFT			0
+static inline uint32_t PPU_DST_BASE_ADDR_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_DST_BASE_ADDR_RESERVED_0__SHIFT) & PPU_DST_BASE_ADDR_RESERVED_0__MASK;
+}
+
+#define REG_PPU_DST_SURF_STRIDE					0x0000607c
+#define PPU_DST_SURF_STRIDE_DST_SURF_STRIDE__MASK		0xfffffff0
+#define PPU_DST_SURF_STRIDE_DST_SURF_STRIDE__SHIFT		4
+static inline uint32_t PPU_DST_SURF_STRIDE_DST_SURF_STRIDE(uint32_t val)
+{
+	return ((val) << PPU_DST_SURF_STRIDE_DST_SURF_STRIDE__SHIFT) & PPU_DST_SURF_STRIDE_DST_SURF_STRIDE__MASK;
+}
+#define PPU_DST_SURF_STRIDE_RESERVED_0__MASK			0x0000000f
+#define PPU_DST_SURF_STRIDE_RESERVED_0__SHIFT			0
+static inline uint32_t PPU_DST_SURF_STRIDE_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_DST_SURF_STRIDE_RESERVED_0__SHIFT) & PPU_DST_SURF_STRIDE_RESERVED_0__MASK;
+}
+
+#define REG_PPU_DATA_FORMAT					0x00006084
+#define PPU_DATA_FORMAT_INDEX_ADD__MASK				0xfffffff0
+#define PPU_DATA_FORMAT_INDEX_ADD__SHIFT			4
+static inline uint32_t PPU_DATA_FORMAT_INDEX_ADD(uint32_t val)
+{
+	return ((val) << PPU_DATA_FORMAT_INDEX_ADD__SHIFT) & PPU_DATA_FORMAT_INDEX_ADD__MASK;
+}
+#define PPU_DATA_FORMAT_DPU_FLYIN__MASK				0x00000008
+#define PPU_DATA_FORMAT_DPU_FLYIN__SHIFT			3
+static inline uint32_t PPU_DATA_FORMAT_DPU_FLYIN(uint32_t val)
+{
+	return ((val) << PPU_DATA_FORMAT_DPU_FLYIN__SHIFT) & PPU_DATA_FORMAT_DPU_FLYIN__MASK;
+}
+#define PPU_DATA_FORMAT_PROC_PRECISION__MASK			0x00000007
+#define PPU_DATA_FORMAT_PROC_PRECISION__SHIFT			0
+static inline uint32_t PPU_DATA_FORMAT_PROC_PRECISION(uint32_t val)
+{
+	return ((val) << PPU_DATA_FORMAT_PROC_PRECISION__SHIFT) & PPU_DATA_FORMAT_PROC_PRECISION__MASK;
+}
+
+#define REG_PPU_MISC_CTRL					0x000060dc
+#define PPU_MISC_CTRL_SURF_LEN__MASK				0xffff0000
+#define PPU_MISC_CTRL_SURF_LEN__SHIFT				16
+static inline uint32_t PPU_MISC_CTRL_SURF_LEN(uint32_t val)
+{
+	return ((val) << PPU_MISC_CTRL_SURF_LEN__SHIFT) & PPU_MISC_CTRL_SURF_LEN__MASK;
+}
+#define PPU_MISC_CTRL_RESERVED_0__MASK				0x0000fe00
+#define PPU_MISC_CTRL_RESERVED_0__SHIFT				9
+static inline uint32_t PPU_MISC_CTRL_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_MISC_CTRL_RESERVED_0__SHIFT) & PPU_MISC_CTRL_RESERVED_0__MASK;
+}
+#define PPU_MISC_CTRL_MC_SURF_OUT__MASK				0x00000100
+#define PPU_MISC_CTRL_MC_SURF_OUT__SHIFT			8
+static inline uint32_t PPU_MISC_CTRL_MC_SURF_OUT(uint32_t val)
+{
+	return ((val) << PPU_MISC_CTRL_MC_SURF_OUT__SHIFT) & PPU_MISC_CTRL_MC_SURF_OUT__MASK;
+}
+#define PPU_MISC_CTRL_NONALIGN__MASK				0x00000080
+#define PPU_MISC_CTRL_NONALIGN__SHIFT				7
+static inline uint32_t PPU_MISC_CTRL_NONALIGN(uint32_t val)
+{
+	return ((val) << PPU_MISC_CTRL_NONALIGN__SHIFT) & PPU_MISC_CTRL_NONALIGN__MASK;
+}
+#define PPU_MISC_CTRL_RESERVED_1__MASK				0x00000070
+#define PPU_MISC_CTRL_RESERVED_1__SHIFT				4
+static inline uint32_t PPU_MISC_CTRL_RESERVED_1(uint32_t val)
+{
+	return ((val) << PPU_MISC_CTRL_RESERVED_1__SHIFT) & PPU_MISC_CTRL_RESERVED_1__MASK;
+}
+#define PPU_MISC_CTRL_BURST_LEN__MASK				0x0000000f
+#define PPU_MISC_CTRL_BURST_LEN__SHIFT				0
+static inline uint32_t PPU_MISC_CTRL_BURST_LEN(uint32_t val)
+{
+	return ((val) << PPU_MISC_CTRL_BURST_LEN__SHIFT) & PPU_MISC_CTRL_BURST_LEN__MASK;
+}
+
+#define REG_PPU_RDMA_RDMA_S_STATUS				0x00007000
+#define PPU_RDMA_RDMA_S_STATUS_RESERVED_0__MASK			0xfffc0000
+#define PPU_RDMA_RDMA_S_STATUS_RESERVED_0__SHIFT		18
+static inline uint32_t PPU_RDMA_RDMA_S_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_STATUS_RESERVED_0__SHIFT) & PPU_RDMA_RDMA_S_STATUS_RESERVED_0__MASK;
+}
+#define PPU_RDMA_RDMA_S_STATUS_STATUS_1__MASK			0x00030000
+#define PPU_RDMA_RDMA_S_STATUS_STATUS_1__SHIFT			16
+static inline uint32_t PPU_RDMA_RDMA_S_STATUS_STATUS_1(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_STATUS_STATUS_1__SHIFT) & PPU_RDMA_RDMA_S_STATUS_STATUS_1__MASK;
+}
+#define PPU_RDMA_RDMA_S_STATUS_RESERVED_1__MASK			0x0000fffc
+#define PPU_RDMA_RDMA_S_STATUS_RESERVED_1__SHIFT		2
+static inline uint32_t PPU_RDMA_RDMA_S_STATUS_RESERVED_1(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_STATUS_RESERVED_1__SHIFT) & PPU_RDMA_RDMA_S_STATUS_RESERVED_1__MASK;
+}
+#define PPU_RDMA_RDMA_S_STATUS_STATUS_0__MASK			0x00000003
+#define PPU_RDMA_RDMA_S_STATUS_STATUS_0__SHIFT			0
+static inline uint32_t PPU_RDMA_RDMA_S_STATUS_STATUS_0(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_STATUS_STATUS_0__SHIFT) & PPU_RDMA_RDMA_S_STATUS_STATUS_0__MASK;
+}
+
+#define REG_PPU_RDMA_RDMA_S_POINTER				0x00007004
+#define PPU_RDMA_RDMA_S_POINTER_RESERVED_0__MASK		0xfffe0000
+#define PPU_RDMA_RDMA_S_POINTER_RESERVED_0__SHIFT		17
+static inline uint32_t PPU_RDMA_RDMA_S_POINTER_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_POINTER_RESERVED_0__SHIFT) & PPU_RDMA_RDMA_S_POINTER_RESERVED_0__MASK;
+}
+#define PPU_RDMA_RDMA_S_POINTER_EXECUTER__MASK			0x00010000
+#define PPU_RDMA_RDMA_S_POINTER_EXECUTER__SHIFT			16
+static inline uint32_t PPU_RDMA_RDMA_S_POINTER_EXECUTER(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_POINTER_EXECUTER__SHIFT) & PPU_RDMA_RDMA_S_POINTER_EXECUTER__MASK;
+}
+#define PPU_RDMA_RDMA_S_POINTER_RESERVED_1__MASK		0x0000ffc0
+#define PPU_RDMA_RDMA_S_POINTER_RESERVED_1__SHIFT		6
+static inline uint32_t PPU_RDMA_RDMA_S_POINTER_RESERVED_1(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_POINTER_RESERVED_1__SHIFT) & PPU_RDMA_RDMA_S_POINTER_RESERVED_1__MASK;
+}
+#define PPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_CLEAR__MASK		0x00000020
+#define PPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_CLEAR__SHIFT	5
+static inline uint32_t PPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_CLEAR__SHIFT) & PPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_CLEAR__MASK;
+}
+#define PPU_RDMA_RDMA_S_POINTER_POINTER_PP_CLEAR__MASK		0x00000010
+#define PPU_RDMA_RDMA_S_POINTER_POINTER_PP_CLEAR__SHIFT		4
+static inline uint32_t PPU_RDMA_RDMA_S_POINTER_POINTER_PP_CLEAR(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_POINTER_POINTER_PP_CLEAR__SHIFT) & PPU_RDMA_RDMA_S_POINTER_POINTER_PP_CLEAR__MASK;
+}
+#define PPU_RDMA_RDMA_S_POINTER_POINTER_PP_MODE__MASK		0x00000008
+#define PPU_RDMA_RDMA_S_POINTER_POINTER_PP_MODE__SHIFT		3
+static inline uint32_t PPU_RDMA_RDMA_S_POINTER_POINTER_PP_MODE(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_POINTER_POINTER_PP_MODE__SHIFT) & PPU_RDMA_RDMA_S_POINTER_POINTER_PP_MODE__MASK;
+}
+#define PPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_EN__MASK		0x00000004
+#define PPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_EN__SHIFT		2
+static inline uint32_t PPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_EN(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_EN__SHIFT) & PPU_RDMA_RDMA_S_POINTER_EXECUTER_PP_EN__MASK;
+}
+#define PPU_RDMA_RDMA_S_POINTER_POINTER_PP_EN__MASK		0x00000002
+#define PPU_RDMA_RDMA_S_POINTER_POINTER_PP_EN__SHIFT		1
+static inline uint32_t PPU_RDMA_RDMA_S_POINTER_POINTER_PP_EN(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_POINTER_POINTER_PP_EN__SHIFT) & PPU_RDMA_RDMA_S_POINTER_POINTER_PP_EN__MASK;
+}
+#define PPU_RDMA_RDMA_S_POINTER_POINTER__MASK			0x00000001
+#define PPU_RDMA_RDMA_S_POINTER_POINTER__SHIFT			0
+static inline uint32_t PPU_RDMA_RDMA_S_POINTER_POINTER(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_S_POINTER_POINTER__SHIFT) & PPU_RDMA_RDMA_S_POINTER_POINTER__MASK;
+}
+
+#define REG_PPU_RDMA_RDMA_OPERATION_ENABLE			0x00007008
+#define PPU_RDMA_RDMA_OPERATION_ENABLE_RESERVED_0__MASK		0xfffffffe
+#define PPU_RDMA_RDMA_OPERATION_ENABLE_RESERVED_0__SHIFT	1
+static inline uint32_t PPU_RDMA_RDMA_OPERATION_ENABLE_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_OPERATION_ENABLE_RESERVED_0__SHIFT) & PPU_RDMA_RDMA_OPERATION_ENABLE_RESERVED_0__MASK;
+}
+#define PPU_RDMA_RDMA_OPERATION_ENABLE_OP_EN__MASK		0x00000001
+#define PPU_RDMA_RDMA_OPERATION_ENABLE_OP_EN__SHIFT		0
+static inline uint32_t PPU_RDMA_RDMA_OPERATION_ENABLE_OP_EN(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_OPERATION_ENABLE_OP_EN__SHIFT) & PPU_RDMA_RDMA_OPERATION_ENABLE_OP_EN__MASK;
+}
+
+#define REG_PPU_RDMA_RDMA_CUBE_IN_WIDTH				0x0000700c
+#define PPU_RDMA_RDMA_CUBE_IN_WIDTH_RESERVED_0__MASK		0xffffe000
+#define PPU_RDMA_RDMA_CUBE_IN_WIDTH_RESERVED_0__SHIFT		13
+static inline uint32_t PPU_RDMA_RDMA_CUBE_IN_WIDTH_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_CUBE_IN_WIDTH_RESERVED_0__SHIFT) & PPU_RDMA_RDMA_CUBE_IN_WIDTH_RESERVED_0__MASK;
+}
+#define PPU_RDMA_RDMA_CUBE_IN_WIDTH_CUBE_IN_WIDTH__MASK		0x00001fff
+#define PPU_RDMA_RDMA_CUBE_IN_WIDTH_CUBE_IN_WIDTH__SHIFT	0
+static inline uint32_t PPU_RDMA_RDMA_CUBE_IN_WIDTH_CUBE_IN_WIDTH(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_CUBE_IN_WIDTH_CUBE_IN_WIDTH__SHIFT) & PPU_RDMA_RDMA_CUBE_IN_WIDTH_CUBE_IN_WIDTH__MASK;
+}
+
+#define REG_PPU_RDMA_RDMA_CUBE_IN_HEIGHT			0x00007010
+#define PPU_RDMA_RDMA_CUBE_IN_HEIGHT_RESERVED_0__MASK		0xffffe000
+#define PPU_RDMA_RDMA_CUBE_IN_HEIGHT_RESERVED_0__SHIFT		13
+static inline uint32_t PPU_RDMA_RDMA_CUBE_IN_HEIGHT_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_CUBE_IN_HEIGHT_RESERVED_0__SHIFT) & PPU_RDMA_RDMA_CUBE_IN_HEIGHT_RESERVED_0__MASK;
+}
+#define PPU_RDMA_RDMA_CUBE_IN_HEIGHT_CUBE_IN_HEIGHT__MASK	0x00001fff
+#define PPU_RDMA_RDMA_CUBE_IN_HEIGHT_CUBE_IN_HEIGHT__SHIFT	0
+static inline uint32_t PPU_RDMA_RDMA_CUBE_IN_HEIGHT_CUBE_IN_HEIGHT(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_CUBE_IN_HEIGHT_CUBE_IN_HEIGHT__SHIFT) & PPU_RDMA_RDMA_CUBE_IN_HEIGHT_CUBE_IN_HEIGHT__MASK;
+}
+
+#define REG_PPU_RDMA_RDMA_CUBE_IN_CHANNEL			0x00007014
+#define PPU_RDMA_RDMA_CUBE_IN_CHANNEL_RESERVED_0__MASK		0xffffe000
+#define PPU_RDMA_RDMA_CUBE_IN_CHANNEL_RESERVED_0__SHIFT		13
+static inline uint32_t PPU_RDMA_RDMA_CUBE_IN_CHANNEL_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_CUBE_IN_CHANNEL_RESERVED_0__SHIFT) & PPU_RDMA_RDMA_CUBE_IN_CHANNEL_RESERVED_0__MASK;
+}
+#define PPU_RDMA_RDMA_CUBE_IN_CHANNEL_CUBE_IN_CHANNEL__MASK	0x00001fff
+#define PPU_RDMA_RDMA_CUBE_IN_CHANNEL_CUBE_IN_CHANNEL__SHIFT	0
+static inline uint32_t PPU_RDMA_RDMA_CUBE_IN_CHANNEL_CUBE_IN_CHANNEL(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_CUBE_IN_CHANNEL_CUBE_IN_CHANNEL__SHIFT) & PPU_RDMA_RDMA_CUBE_IN_CHANNEL_CUBE_IN_CHANNEL__MASK;
+}
+
+#define REG_PPU_RDMA_RDMA_SRC_BASE_ADDR				0x0000701c
+#define PPU_RDMA_RDMA_SRC_BASE_ADDR_SRC_BASE_ADDR__MASK		0xffffffff
+#define PPU_RDMA_RDMA_SRC_BASE_ADDR_SRC_BASE_ADDR__SHIFT	0
+static inline uint32_t PPU_RDMA_RDMA_SRC_BASE_ADDR_SRC_BASE_ADDR(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_SRC_BASE_ADDR_SRC_BASE_ADDR__SHIFT) & PPU_RDMA_RDMA_SRC_BASE_ADDR_SRC_BASE_ADDR__MASK;
+}
+
+#define REG_PPU_RDMA_RDMA_SRC_LINE_STRIDE			0x00007024
+#define PPU_RDMA_RDMA_SRC_LINE_STRIDE_SRC_LINE_STRIDE__MASK	0xfffffff0
+#define PPU_RDMA_RDMA_SRC_LINE_STRIDE_SRC_LINE_STRIDE__SHIFT	4
+static inline uint32_t PPU_RDMA_RDMA_SRC_LINE_STRIDE_SRC_LINE_STRIDE(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_SRC_LINE_STRIDE_SRC_LINE_STRIDE__SHIFT) & PPU_RDMA_RDMA_SRC_LINE_STRIDE_SRC_LINE_STRIDE__MASK;
+}
+#define PPU_RDMA_RDMA_SRC_LINE_STRIDE_RESERVED_0__MASK		0x0000000f
+#define PPU_RDMA_RDMA_SRC_LINE_STRIDE_RESERVED_0__SHIFT		0
+static inline uint32_t PPU_RDMA_RDMA_SRC_LINE_STRIDE_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_SRC_LINE_STRIDE_RESERVED_0__SHIFT) & PPU_RDMA_RDMA_SRC_LINE_STRIDE_RESERVED_0__MASK;
+}
+
+#define REG_PPU_RDMA_RDMA_SRC_SURF_STRIDE			0x00007028
+#define PPU_RDMA_RDMA_SRC_SURF_STRIDE_SRC_SURF_STRIDE__MASK	0xfffffff0
+#define PPU_RDMA_RDMA_SRC_SURF_STRIDE_SRC_SURF_STRIDE__SHIFT	4
+static inline uint32_t PPU_RDMA_RDMA_SRC_SURF_STRIDE_SRC_SURF_STRIDE(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_SRC_SURF_STRIDE_SRC_SURF_STRIDE__SHIFT) & PPU_RDMA_RDMA_SRC_SURF_STRIDE_SRC_SURF_STRIDE__MASK;
+}
+#define PPU_RDMA_RDMA_SRC_SURF_STRIDE_RESERVED_0__MASK		0x0000000f
+#define PPU_RDMA_RDMA_SRC_SURF_STRIDE_RESERVED_0__SHIFT		0
+static inline uint32_t PPU_RDMA_RDMA_SRC_SURF_STRIDE_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_SRC_SURF_STRIDE_RESERVED_0__SHIFT) & PPU_RDMA_RDMA_SRC_SURF_STRIDE_RESERVED_0__MASK;
+}
+
+#define REG_PPU_RDMA_RDMA_DATA_FORMAT				0x00007030
+#define PPU_RDMA_RDMA_DATA_FORMAT_RESERVED_0__MASK		0xfffffffc
+#define PPU_RDMA_RDMA_DATA_FORMAT_RESERVED_0__SHIFT		2
+static inline uint32_t PPU_RDMA_RDMA_DATA_FORMAT_RESERVED_0(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_DATA_FORMAT_RESERVED_0__SHIFT) & PPU_RDMA_RDMA_DATA_FORMAT_RESERVED_0__MASK;
+}
+#define PPU_RDMA_RDMA_DATA_FORMAT_IN_PRECISION__MASK		0x00000003
+#define PPU_RDMA_RDMA_DATA_FORMAT_IN_PRECISION__SHIFT		0
+static inline uint32_t PPU_RDMA_RDMA_DATA_FORMAT_IN_PRECISION(uint32_t val)
+{
+	return ((val) << PPU_RDMA_RDMA_DATA_FORMAT_IN_PRECISION__SHIFT) & PPU_RDMA_RDMA_DATA_FORMAT_IN_PRECISION__MASK;
+}
+
+#define REG_DDMA_CFG_OUTSTANDING				0x00008000
+#define DDMA_CFG_OUTSTANDING_RESERVED_0__MASK			0xffff0000
+#define DDMA_CFG_OUTSTANDING_RESERVED_0__SHIFT			16
+static inline uint32_t DDMA_CFG_OUTSTANDING_RESERVED_0(uint32_t val)
+{
+	return ((val) << DDMA_CFG_OUTSTANDING_RESERVED_0__SHIFT) & DDMA_CFG_OUTSTANDING_RESERVED_0__MASK;
+}
+#define DDMA_CFG_OUTSTANDING_WR_OS_CNT__MASK			0x0000ff00
+#define DDMA_CFG_OUTSTANDING_WR_OS_CNT__SHIFT			8
+static inline uint32_t DDMA_CFG_OUTSTANDING_WR_OS_CNT(uint32_t val)
+{
+	return ((val) << DDMA_CFG_OUTSTANDING_WR_OS_CNT__SHIFT) & DDMA_CFG_OUTSTANDING_WR_OS_CNT__MASK;
+}
+#define DDMA_CFG_OUTSTANDING_RD_OS_CNT__MASK			0x000000ff
+#define DDMA_CFG_OUTSTANDING_RD_OS_CNT__SHIFT			0
+static inline uint32_t DDMA_CFG_OUTSTANDING_RD_OS_CNT(uint32_t val)
+{
+	return ((val) << DDMA_CFG_OUTSTANDING_RD_OS_CNT__SHIFT) & DDMA_CFG_OUTSTANDING_RD_OS_CNT__MASK;
+}
+
+#define REG_DDMA_RD_WEIGHT_0					0x00008004
+#define DDMA_RD_WEIGHT_0_RD_WEIGHT_PDP__MASK			0xff000000
+#define DDMA_RD_WEIGHT_0_RD_WEIGHT_PDP__SHIFT			24
+static inline uint32_t DDMA_RD_WEIGHT_0_RD_WEIGHT_PDP(uint32_t val)
+{
+	return ((val) << DDMA_RD_WEIGHT_0_RD_WEIGHT_PDP__SHIFT) & DDMA_RD_WEIGHT_0_RD_WEIGHT_PDP__MASK;
+}
+#define DDMA_RD_WEIGHT_0_RD_WEIGHT_DPU__MASK			0x00ff0000
+#define DDMA_RD_WEIGHT_0_RD_WEIGHT_DPU__SHIFT			16
+static inline uint32_t DDMA_RD_WEIGHT_0_RD_WEIGHT_DPU(uint32_t val)
+{
+	return ((val) << DDMA_RD_WEIGHT_0_RD_WEIGHT_DPU__SHIFT) & DDMA_RD_WEIGHT_0_RD_WEIGHT_DPU__MASK;
+}
+#define DDMA_RD_WEIGHT_0_RD_WEIGHT_KERNEL__MASK			0x0000ff00
+#define DDMA_RD_WEIGHT_0_RD_WEIGHT_KERNEL__SHIFT		8
+static inline uint32_t DDMA_RD_WEIGHT_0_RD_WEIGHT_KERNEL(uint32_t val)
+{
+	return ((val) << DDMA_RD_WEIGHT_0_RD_WEIGHT_KERNEL__SHIFT) & DDMA_RD_WEIGHT_0_RD_WEIGHT_KERNEL__MASK;
+}
+#define DDMA_RD_WEIGHT_0_RD_WEIGHT_FEATURE__MASK		0x000000ff
+#define DDMA_RD_WEIGHT_0_RD_WEIGHT_FEATURE__SHIFT		0
+static inline uint32_t DDMA_RD_WEIGHT_0_RD_WEIGHT_FEATURE(uint32_t val)
+{
+	return ((val) << DDMA_RD_WEIGHT_0_RD_WEIGHT_FEATURE__SHIFT) & DDMA_RD_WEIGHT_0_RD_WEIGHT_FEATURE__MASK;
+}
+
+#define REG_DDMA_WR_WEIGHT_0					0x00008008
+#define DDMA_WR_WEIGHT_0_RESERVED_0__MASK			0xffff0000
+#define DDMA_WR_WEIGHT_0_RESERVED_0__SHIFT			16
+static inline uint32_t DDMA_WR_WEIGHT_0_RESERVED_0(uint32_t val)
+{
+	return ((val) << DDMA_WR_WEIGHT_0_RESERVED_0__SHIFT) & DDMA_WR_WEIGHT_0_RESERVED_0__MASK;
+}
+#define DDMA_WR_WEIGHT_0_WR_WEIGHT_PDP__MASK			0x0000ff00
+#define DDMA_WR_WEIGHT_0_WR_WEIGHT_PDP__SHIFT			8
+static inline uint32_t DDMA_WR_WEIGHT_0_WR_WEIGHT_PDP(uint32_t val)
+{
+	return ((val) << DDMA_WR_WEIGHT_0_WR_WEIGHT_PDP__SHIFT) & DDMA_WR_WEIGHT_0_WR_WEIGHT_PDP__MASK;
+}
+#define DDMA_WR_WEIGHT_0_WR_WEIGHT_DPU__MASK			0x000000ff
+#define DDMA_WR_WEIGHT_0_WR_WEIGHT_DPU__SHIFT			0
+static inline uint32_t DDMA_WR_WEIGHT_0_WR_WEIGHT_DPU(uint32_t val)
+{
+	return ((val) << DDMA_WR_WEIGHT_0_WR_WEIGHT_DPU__SHIFT) & DDMA_WR_WEIGHT_0_WR_WEIGHT_DPU__MASK;
+}
+
+#define REG_DDMA_CFG_ID_ERROR					0x0000800c
+#define DDMA_CFG_ID_ERROR_RESERVED_0__MASK			0xfffffc00
+#define DDMA_CFG_ID_ERROR_RESERVED_0__SHIFT			10
+static inline uint32_t DDMA_CFG_ID_ERROR_RESERVED_0(uint32_t val)
+{
+	return ((val) << DDMA_CFG_ID_ERROR_RESERVED_0__SHIFT) & DDMA_CFG_ID_ERROR_RESERVED_0__MASK;
+}
+#define DDMA_CFG_ID_ERROR_WR_RESP_ID__MASK			0x000003c0
+#define DDMA_CFG_ID_ERROR_WR_RESP_ID__SHIFT			6
+static inline uint32_t DDMA_CFG_ID_ERROR_WR_RESP_ID(uint32_t val)
+{
+	return ((val) << DDMA_CFG_ID_ERROR_WR_RESP_ID__SHIFT) & DDMA_CFG_ID_ERROR_WR_RESP_ID__MASK;
+}
+#define DDMA_CFG_ID_ERROR_RESERVED_1__MASK			0x00000020
+#define DDMA_CFG_ID_ERROR_RESERVED_1__SHIFT			5
+static inline uint32_t DDMA_CFG_ID_ERROR_RESERVED_1(uint32_t val)
+{
+	return ((val) << DDMA_CFG_ID_ERROR_RESERVED_1__SHIFT) & DDMA_CFG_ID_ERROR_RESERVED_1__MASK;
+}
+#define DDMA_CFG_ID_ERROR_RD_RESP_ID__MASK			0x0000001f
+#define DDMA_CFG_ID_ERROR_RD_RESP_ID__SHIFT			0
+static inline uint32_t DDMA_CFG_ID_ERROR_RD_RESP_ID(uint32_t val)
+{
+	return ((val) << DDMA_CFG_ID_ERROR_RD_RESP_ID__SHIFT) & DDMA_CFG_ID_ERROR_RD_RESP_ID__MASK;
+}
+
+#define REG_DDMA_RD_WEIGHT_1					0x00008010
+#define DDMA_RD_WEIGHT_1_RESERVED_0__MASK			0xffffff00
+#define DDMA_RD_WEIGHT_1_RESERVED_0__SHIFT			8
+static inline uint32_t DDMA_RD_WEIGHT_1_RESERVED_0(uint32_t val)
+{
+	return ((val) << DDMA_RD_WEIGHT_1_RESERVED_0__SHIFT) & DDMA_RD_WEIGHT_1_RESERVED_0__MASK;
+}
+#define DDMA_RD_WEIGHT_1_RD_WEIGHT_PC__MASK			0x000000ff
+#define DDMA_RD_WEIGHT_1_RD_WEIGHT_PC__SHIFT			0
+static inline uint32_t DDMA_RD_WEIGHT_1_RD_WEIGHT_PC(uint32_t val)
+{
+	return ((val) << DDMA_RD_WEIGHT_1_RD_WEIGHT_PC__SHIFT) & DDMA_RD_WEIGHT_1_RD_WEIGHT_PC__MASK;
+}
+
+#define REG_DDMA_CFG_DMA_FIFO_CLR				0x00008014
+#define DDMA_CFG_DMA_FIFO_CLR_RESERVED_0__MASK			0xfffffffe
+#define DDMA_CFG_DMA_FIFO_CLR_RESERVED_0__SHIFT			1
+static inline uint32_t DDMA_CFG_DMA_FIFO_CLR_RESERVED_0(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_FIFO_CLR_RESERVED_0__SHIFT) & DDMA_CFG_DMA_FIFO_CLR_RESERVED_0__MASK;
+}
+#define DDMA_CFG_DMA_FIFO_CLR_DMA_FIFO_CLR__MASK		0x00000001
+#define DDMA_CFG_DMA_FIFO_CLR_DMA_FIFO_CLR__SHIFT		0
+static inline uint32_t DDMA_CFG_DMA_FIFO_CLR_DMA_FIFO_CLR(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_FIFO_CLR_DMA_FIFO_CLR__SHIFT) & DDMA_CFG_DMA_FIFO_CLR_DMA_FIFO_CLR__MASK;
+}
+
+#define REG_DDMA_CFG_DMA_ARB					0x00008018
+#define DDMA_CFG_DMA_ARB_RESERVED_0__MASK			0xfffffc00
+#define DDMA_CFG_DMA_ARB_RESERVED_0__SHIFT			10
+static inline uint32_t DDMA_CFG_DMA_ARB_RESERVED_0(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_ARB_RESERVED_0__SHIFT) & DDMA_CFG_DMA_ARB_RESERVED_0__MASK;
+}
+#define DDMA_CFG_DMA_ARB_WR_ARBIT_MODEL__MASK			0x00000200
+#define DDMA_CFG_DMA_ARB_WR_ARBIT_MODEL__SHIFT			9
+static inline uint32_t DDMA_CFG_DMA_ARB_WR_ARBIT_MODEL(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_ARB_WR_ARBIT_MODEL__SHIFT) & DDMA_CFG_DMA_ARB_WR_ARBIT_MODEL__MASK;
+}
+#define DDMA_CFG_DMA_ARB_RD_ARBIT_MODEL__MASK			0x00000100
+#define DDMA_CFG_DMA_ARB_RD_ARBIT_MODEL__SHIFT			8
+static inline uint32_t DDMA_CFG_DMA_ARB_RD_ARBIT_MODEL(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_ARB_RD_ARBIT_MODEL__SHIFT) & DDMA_CFG_DMA_ARB_RD_ARBIT_MODEL__MASK;
+}
+#define DDMA_CFG_DMA_ARB_RESERVED_1__MASK			0x00000080
+#define DDMA_CFG_DMA_ARB_RESERVED_1__SHIFT			7
+static inline uint32_t DDMA_CFG_DMA_ARB_RESERVED_1(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_ARB_RESERVED_1__SHIFT) & DDMA_CFG_DMA_ARB_RESERVED_1__MASK;
+}
+#define DDMA_CFG_DMA_ARB_WR_FIX_ARB__MASK			0x00000070
+#define DDMA_CFG_DMA_ARB_WR_FIX_ARB__SHIFT			4
+static inline uint32_t DDMA_CFG_DMA_ARB_WR_FIX_ARB(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_ARB_WR_FIX_ARB__SHIFT) & DDMA_CFG_DMA_ARB_WR_FIX_ARB__MASK;
+}
+#define DDMA_CFG_DMA_ARB_RESERVED_2__MASK			0x00000008
+#define DDMA_CFG_DMA_ARB_RESERVED_2__SHIFT			3
+static inline uint32_t DDMA_CFG_DMA_ARB_RESERVED_2(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_ARB_RESERVED_2__SHIFT) & DDMA_CFG_DMA_ARB_RESERVED_2__MASK;
+}
+#define DDMA_CFG_DMA_ARB_RD_FIX_ARB__MASK			0x00000007
+#define DDMA_CFG_DMA_ARB_RD_FIX_ARB__SHIFT			0
+static inline uint32_t DDMA_CFG_DMA_ARB_RD_FIX_ARB(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_ARB_RD_FIX_ARB__SHIFT) & DDMA_CFG_DMA_ARB_RD_FIX_ARB__MASK;
+}
+
+#define REG_DDMA_CFG_DMA_RD_QOS					0x00008020
+#define DDMA_CFG_DMA_RD_QOS_RESERVED_0__MASK			0xfffffc00
+#define DDMA_CFG_DMA_RD_QOS_RESERVED_0__SHIFT			10
+static inline uint32_t DDMA_CFG_DMA_RD_QOS_RESERVED_0(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_QOS_RESERVED_0__SHIFT) & DDMA_CFG_DMA_RD_QOS_RESERVED_0__MASK;
+}
+#define DDMA_CFG_DMA_RD_QOS_RD_PC_QOS__MASK			0x00000300
+#define DDMA_CFG_DMA_RD_QOS_RD_PC_QOS__SHIFT			8
+static inline uint32_t DDMA_CFG_DMA_RD_QOS_RD_PC_QOS(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_QOS_RD_PC_QOS__SHIFT) & DDMA_CFG_DMA_RD_QOS_RD_PC_QOS__MASK;
+}
+#define DDMA_CFG_DMA_RD_QOS_RD_PPU_QOS__MASK			0x000000c0
+#define DDMA_CFG_DMA_RD_QOS_RD_PPU_QOS__SHIFT			6
+static inline uint32_t DDMA_CFG_DMA_RD_QOS_RD_PPU_QOS(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_QOS_RD_PPU_QOS__SHIFT) & DDMA_CFG_DMA_RD_QOS_RD_PPU_QOS__MASK;
+}
+#define DDMA_CFG_DMA_RD_QOS_RD_DPU_QOS__MASK			0x00000030
+#define DDMA_CFG_DMA_RD_QOS_RD_DPU_QOS__SHIFT			4
+static inline uint32_t DDMA_CFG_DMA_RD_QOS_RD_DPU_QOS(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_QOS_RD_DPU_QOS__SHIFT) & DDMA_CFG_DMA_RD_QOS_RD_DPU_QOS__MASK;
+}
+#define DDMA_CFG_DMA_RD_QOS_RD_KERNEL_QOS__MASK			0x0000000c
+#define DDMA_CFG_DMA_RD_QOS_RD_KERNEL_QOS__SHIFT		2
+static inline uint32_t DDMA_CFG_DMA_RD_QOS_RD_KERNEL_QOS(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_QOS_RD_KERNEL_QOS__SHIFT) & DDMA_CFG_DMA_RD_QOS_RD_KERNEL_QOS__MASK;
+}
+#define DDMA_CFG_DMA_RD_QOS_RD_FEATURE_QOS__MASK		0x00000003
+#define DDMA_CFG_DMA_RD_QOS_RD_FEATURE_QOS__SHIFT		0
+static inline uint32_t DDMA_CFG_DMA_RD_QOS_RD_FEATURE_QOS(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_QOS_RD_FEATURE_QOS__SHIFT) & DDMA_CFG_DMA_RD_QOS_RD_FEATURE_QOS__MASK;
+}
+
+#define REG_DDMA_CFG_DMA_RD_CFG					0x00008024
+#define DDMA_CFG_DMA_RD_CFG_RESERVED_0__MASK			0xffffe000
+#define DDMA_CFG_DMA_RD_CFG_RESERVED_0__SHIFT			13
+static inline uint32_t DDMA_CFG_DMA_RD_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_CFG_RESERVED_0__SHIFT) & DDMA_CFG_DMA_RD_CFG_RESERVED_0__MASK;
+}
+#define DDMA_CFG_DMA_RD_CFG_RD_ARLOCK__MASK			0x00001000
+#define DDMA_CFG_DMA_RD_CFG_RD_ARLOCK__SHIFT			12
+static inline uint32_t DDMA_CFG_DMA_RD_CFG_RD_ARLOCK(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_CFG_RD_ARLOCK__SHIFT) & DDMA_CFG_DMA_RD_CFG_RD_ARLOCK__MASK;
+}
+#define DDMA_CFG_DMA_RD_CFG_RD_ARCACHE__MASK			0x00000f00
+#define DDMA_CFG_DMA_RD_CFG_RD_ARCACHE__SHIFT			8
+static inline uint32_t DDMA_CFG_DMA_RD_CFG_RD_ARCACHE(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_CFG_RD_ARCACHE__SHIFT) & DDMA_CFG_DMA_RD_CFG_RD_ARCACHE__MASK;
+}
+#define DDMA_CFG_DMA_RD_CFG_RD_ARPROT__MASK			0x000000e0
+#define DDMA_CFG_DMA_RD_CFG_RD_ARPROT__SHIFT			5
+static inline uint32_t DDMA_CFG_DMA_RD_CFG_RD_ARPROT(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_CFG_RD_ARPROT__SHIFT) & DDMA_CFG_DMA_RD_CFG_RD_ARPROT__MASK;
+}
+#define DDMA_CFG_DMA_RD_CFG_RD_ARBURST__MASK			0x00000018
+#define DDMA_CFG_DMA_RD_CFG_RD_ARBURST__SHIFT			3
+static inline uint32_t DDMA_CFG_DMA_RD_CFG_RD_ARBURST(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_CFG_RD_ARBURST__SHIFT) & DDMA_CFG_DMA_RD_CFG_RD_ARBURST__MASK;
+}
+#define DDMA_CFG_DMA_RD_CFG_RD_ARSIZE__MASK			0x00000007
+#define DDMA_CFG_DMA_RD_CFG_RD_ARSIZE__SHIFT			0
+static inline uint32_t DDMA_CFG_DMA_RD_CFG_RD_ARSIZE(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_RD_CFG_RD_ARSIZE__SHIFT) & DDMA_CFG_DMA_RD_CFG_RD_ARSIZE__MASK;
+}
+
+#define REG_DDMA_CFG_DMA_WR_CFG					0x00008028
+#define DDMA_CFG_DMA_WR_CFG_RESERVED_0__MASK			0xffffe000
+#define DDMA_CFG_DMA_WR_CFG_RESERVED_0__SHIFT			13
+static inline uint32_t DDMA_CFG_DMA_WR_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_WR_CFG_RESERVED_0__SHIFT) & DDMA_CFG_DMA_WR_CFG_RESERVED_0__MASK;
+}
+#define DDMA_CFG_DMA_WR_CFG_WR_AWLOCK__MASK			0x00001000
+#define DDMA_CFG_DMA_WR_CFG_WR_AWLOCK__SHIFT			12
+static inline uint32_t DDMA_CFG_DMA_WR_CFG_WR_AWLOCK(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_WR_CFG_WR_AWLOCK__SHIFT) & DDMA_CFG_DMA_WR_CFG_WR_AWLOCK__MASK;
+}
+#define DDMA_CFG_DMA_WR_CFG_WR_AWCACHE__MASK			0x00000f00
+#define DDMA_CFG_DMA_WR_CFG_WR_AWCACHE__SHIFT			8
+static inline uint32_t DDMA_CFG_DMA_WR_CFG_WR_AWCACHE(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_WR_CFG_WR_AWCACHE__SHIFT) & DDMA_CFG_DMA_WR_CFG_WR_AWCACHE__MASK;
+}
+#define DDMA_CFG_DMA_WR_CFG_WR_AWPROT__MASK			0x000000e0
+#define DDMA_CFG_DMA_WR_CFG_WR_AWPROT__SHIFT			5
+static inline uint32_t DDMA_CFG_DMA_WR_CFG_WR_AWPROT(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_WR_CFG_WR_AWPROT__SHIFT) & DDMA_CFG_DMA_WR_CFG_WR_AWPROT__MASK;
+}
+#define DDMA_CFG_DMA_WR_CFG_WR_AWBURST__MASK			0x00000018
+#define DDMA_CFG_DMA_WR_CFG_WR_AWBURST__SHIFT			3
+static inline uint32_t DDMA_CFG_DMA_WR_CFG_WR_AWBURST(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_WR_CFG_WR_AWBURST__SHIFT) & DDMA_CFG_DMA_WR_CFG_WR_AWBURST__MASK;
+}
+#define DDMA_CFG_DMA_WR_CFG_WR_AWSIZE__MASK			0x00000007
+#define DDMA_CFG_DMA_WR_CFG_WR_AWSIZE__SHIFT			0
+static inline uint32_t DDMA_CFG_DMA_WR_CFG_WR_AWSIZE(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_WR_CFG_WR_AWSIZE__SHIFT) & DDMA_CFG_DMA_WR_CFG_WR_AWSIZE__MASK;
+}
+
+#define REG_DDMA_CFG_DMA_WSTRB					0x0000802c
+#define DDMA_CFG_DMA_WSTRB_WR_WSTRB__MASK			0xffffffff
+#define DDMA_CFG_DMA_WSTRB_WR_WSTRB__SHIFT			0
+static inline uint32_t DDMA_CFG_DMA_WSTRB_WR_WSTRB(uint32_t val)
+{
+	return ((val) << DDMA_CFG_DMA_WSTRB_WR_WSTRB__SHIFT) & DDMA_CFG_DMA_WSTRB_WR_WSTRB__MASK;
+}
+
+#define REG_DDMA_CFG_STATUS					0x00008030
+#define DDMA_CFG_STATUS_RESERVED_0__MASK			0xfffffe00
+#define DDMA_CFG_STATUS_RESERVED_0__SHIFT			9
+static inline uint32_t DDMA_CFG_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << DDMA_CFG_STATUS_RESERVED_0__SHIFT) & DDMA_CFG_STATUS_RESERVED_0__MASK;
+}
+#define DDMA_CFG_STATUS_IDEL__MASK				0x00000100
+#define DDMA_CFG_STATUS_IDEL__SHIFT				8
+static inline uint32_t DDMA_CFG_STATUS_IDEL(uint32_t val)
+{
+	return ((val) << DDMA_CFG_STATUS_IDEL__SHIFT) & DDMA_CFG_STATUS_IDEL__MASK;
+}
+#define DDMA_CFG_STATUS_RESERVED_1__MASK			0x000000ff
+#define DDMA_CFG_STATUS_RESERVED_1__SHIFT			0
+static inline uint32_t DDMA_CFG_STATUS_RESERVED_1(uint32_t val)
+{
+	return ((val) << DDMA_CFG_STATUS_RESERVED_1__SHIFT) & DDMA_CFG_STATUS_RESERVED_1__MASK;
+}
+
+#define REG_SDMA_CFG_OUTSTANDING				0x00009000
+#define SDMA_CFG_OUTSTANDING_RESERVED_0__MASK			0xffff0000
+#define SDMA_CFG_OUTSTANDING_RESERVED_0__SHIFT			16
+static inline uint32_t SDMA_CFG_OUTSTANDING_RESERVED_0(uint32_t val)
+{
+	return ((val) << SDMA_CFG_OUTSTANDING_RESERVED_0__SHIFT) & SDMA_CFG_OUTSTANDING_RESERVED_0__MASK;
+}
+#define SDMA_CFG_OUTSTANDING_WR_OS_CNT__MASK			0x0000ff00
+#define SDMA_CFG_OUTSTANDING_WR_OS_CNT__SHIFT			8
+static inline uint32_t SDMA_CFG_OUTSTANDING_WR_OS_CNT(uint32_t val)
+{
+	return ((val) << SDMA_CFG_OUTSTANDING_WR_OS_CNT__SHIFT) & SDMA_CFG_OUTSTANDING_WR_OS_CNT__MASK;
+}
+#define SDMA_CFG_OUTSTANDING_RD_OS_CNT__MASK			0x000000ff
+#define SDMA_CFG_OUTSTANDING_RD_OS_CNT__SHIFT			0
+static inline uint32_t SDMA_CFG_OUTSTANDING_RD_OS_CNT(uint32_t val)
+{
+	return ((val) << SDMA_CFG_OUTSTANDING_RD_OS_CNT__SHIFT) & SDMA_CFG_OUTSTANDING_RD_OS_CNT__MASK;
+}
+
+#define REG_SDMA_RD_WEIGHT_0					0x00009004
+#define SDMA_RD_WEIGHT_0_RD_WEIGHT_PDP__MASK			0xff000000
+#define SDMA_RD_WEIGHT_0_RD_WEIGHT_PDP__SHIFT			24
+static inline uint32_t SDMA_RD_WEIGHT_0_RD_WEIGHT_PDP(uint32_t val)
+{
+	return ((val) << SDMA_RD_WEIGHT_0_RD_WEIGHT_PDP__SHIFT) & SDMA_RD_WEIGHT_0_RD_WEIGHT_PDP__MASK;
+}
+#define SDMA_RD_WEIGHT_0_RD_WEIGHT_DPU__MASK			0x00ff0000
+#define SDMA_RD_WEIGHT_0_RD_WEIGHT_DPU__SHIFT			16
+static inline uint32_t SDMA_RD_WEIGHT_0_RD_WEIGHT_DPU(uint32_t val)
+{
+	return ((val) << SDMA_RD_WEIGHT_0_RD_WEIGHT_DPU__SHIFT) & SDMA_RD_WEIGHT_0_RD_WEIGHT_DPU__MASK;
+}
+#define SDMA_RD_WEIGHT_0_RD_WEIGHT_KERNEL__MASK			0x0000ff00
+#define SDMA_RD_WEIGHT_0_RD_WEIGHT_KERNEL__SHIFT		8
+static inline uint32_t SDMA_RD_WEIGHT_0_RD_WEIGHT_KERNEL(uint32_t val)
+{
+	return ((val) << SDMA_RD_WEIGHT_0_RD_WEIGHT_KERNEL__SHIFT) & SDMA_RD_WEIGHT_0_RD_WEIGHT_KERNEL__MASK;
+}
+#define SDMA_RD_WEIGHT_0_RD_WEIGHT_FEATURE__MASK		0x000000ff
+#define SDMA_RD_WEIGHT_0_RD_WEIGHT_FEATURE__SHIFT		0
+static inline uint32_t SDMA_RD_WEIGHT_0_RD_WEIGHT_FEATURE(uint32_t val)
+{
+	return ((val) << SDMA_RD_WEIGHT_0_RD_WEIGHT_FEATURE__SHIFT) & SDMA_RD_WEIGHT_0_RD_WEIGHT_FEATURE__MASK;
+}
+
+#define REG_SDMA_WR_WEIGHT_0					0x00009008
+#define SDMA_WR_WEIGHT_0_RESERVED_0__MASK			0xffff0000
+#define SDMA_WR_WEIGHT_0_RESERVED_0__SHIFT			16
+static inline uint32_t SDMA_WR_WEIGHT_0_RESERVED_0(uint32_t val)
+{
+	return ((val) << SDMA_WR_WEIGHT_0_RESERVED_0__SHIFT) & SDMA_WR_WEIGHT_0_RESERVED_0__MASK;
+}
+#define SDMA_WR_WEIGHT_0_WR_WEIGHT_PDP__MASK			0x0000ff00
+#define SDMA_WR_WEIGHT_0_WR_WEIGHT_PDP__SHIFT			8
+static inline uint32_t SDMA_WR_WEIGHT_0_WR_WEIGHT_PDP(uint32_t val)
+{
+	return ((val) << SDMA_WR_WEIGHT_0_WR_WEIGHT_PDP__SHIFT) & SDMA_WR_WEIGHT_0_WR_WEIGHT_PDP__MASK;
+}
+#define SDMA_WR_WEIGHT_0_WR_WEIGHT_DPU__MASK			0x000000ff
+#define SDMA_WR_WEIGHT_0_WR_WEIGHT_DPU__SHIFT			0
+static inline uint32_t SDMA_WR_WEIGHT_0_WR_WEIGHT_DPU(uint32_t val)
+{
+	return ((val) << SDMA_WR_WEIGHT_0_WR_WEIGHT_DPU__SHIFT) & SDMA_WR_WEIGHT_0_WR_WEIGHT_DPU__MASK;
+}
+
+#define REG_SDMA_CFG_ID_ERROR					0x0000900c
+#define SDMA_CFG_ID_ERROR_RESERVED_0__MASK			0xfffffc00
+#define SDMA_CFG_ID_ERROR_RESERVED_0__SHIFT			10
+static inline uint32_t SDMA_CFG_ID_ERROR_RESERVED_0(uint32_t val)
+{
+	return ((val) << SDMA_CFG_ID_ERROR_RESERVED_0__SHIFT) & SDMA_CFG_ID_ERROR_RESERVED_0__MASK;
+}
+#define SDMA_CFG_ID_ERROR_WR_RESP_ID__MASK			0x000003c0
+#define SDMA_CFG_ID_ERROR_WR_RESP_ID__SHIFT			6
+static inline uint32_t SDMA_CFG_ID_ERROR_WR_RESP_ID(uint32_t val)
+{
+	return ((val) << SDMA_CFG_ID_ERROR_WR_RESP_ID__SHIFT) & SDMA_CFG_ID_ERROR_WR_RESP_ID__MASK;
+}
+#define SDMA_CFG_ID_ERROR_RESERVED_1__MASK			0x00000020
+#define SDMA_CFG_ID_ERROR_RESERVED_1__SHIFT			5
+static inline uint32_t SDMA_CFG_ID_ERROR_RESERVED_1(uint32_t val)
+{
+	return ((val) << SDMA_CFG_ID_ERROR_RESERVED_1__SHIFT) & SDMA_CFG_ID_ERROR_RESERVED_1__MASK;
+}
+#define SDMA_CFG_ID_ERROR_RD_RESP_ID__MASK			0x0000001f
+#define SDMA_CFG_ID_ERROR_RD_RESP_ID__SHIFT			0
+static inline uint32_t SDMA_CFG_ID_ERROR_RD_RESP_ID(uint32_t val)
+{
+	return ((val) << SDMA_CFG_ID_ERROR_RD_RESP_ID__SHIFT) & SDMA_CFG_ID_ERROR_RD_RESP_ID__MASK;
+}
+
+#define REG_SDMA_RD_WEIGHT_1					0x00009010
+#define SDMA_RD_WEIGHT_1_RESERVED_0__MASK			0xffffff00
+#define SDMA_RD_WEIGHT_1_RESERVED_0__SHIFT			8
+static inline uint32_t SDMA_RD_WEIGHT_1_RESERVED_0(uint32_t val)
+{
+	return ((val) << SDMA_RD_WEIGHT_1_RESERVED_0__SHIFT) & SDMA_RD_WEIGHT_1_RESERVED_0__MASK;
+}
+#define SDMA_RD_WEIGHT_1_RD_WEIGHT_PC__MASK			0x000000ff
+#define SDMA_RD_WEIGHT_1_RD_WEIGHT_PC__SHIFT			0
+static inline uint32_t SDMA_RD_WEIGHT_1_RD_WEIGHT_PC(uint32_t val)
+{
+	return ((val) << SDMA_RD_WEIGHT_1_RD_WEIGHT_PC__SHIFT) & SDMA_RD_WEIGHT_1_RD_WEIGHT_PC__MASK;
+}
+
+#define REG_SDMA_CFG_DMA_FIFO_CLR				0x00009014
+#define SDMA_CFG_DMA_FIFO_CLR_RESERVED_0__MASK			0xfffffffe
+#define SDMA_CFG_DMA_FIFO_CLR_RESERVED_0__SHIFT			1
+static inline uint32_t SDMA_CFG_DMA_FIFO_CLR_RESERVED_0(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_FIFO_CLR_RESERVED_0__SHIFT) & SDMA_CFG_DMA_FIFO_CLR_RESERVED_0__MASK;
+}
+#define SDMA_CFG_DMA_FIFO_CLR_DMA_FIFO_CLR__MASK		0x00000001
+#define SDMA_CFG_DMA_FIFO_CLR_DMA_FIFO_CLR__SHIFT		0
+static inline uint32_t SDMA_CFG_DMA_FIFO_CLR_DMA_FIFO_CLR(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_FIFO_CLR_DMA_FIFO_CLR__SHIFT) & SDMA_CFG_DMA_FIFO_CLR_DMA_FIFO_CLR__MASK;
+}
+
+#define REG_SDMA_CFG_DMA_ARB					0x00009018
+#define SDMA_CFG_DMA_ARB_RESERVED_0__MASK			0xfffffc00
+#define SDMA_CFG_DMA_ARB_RESERVED_0__SHIFT			10
+static inline uint32_t SDMA_CFG_DMA_ARB_RESERVED_0(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_ARB_RESERVED_0__SHIFT) & SDMA_CFG_DMA_ARB_RESERVED_0__MASK;
+}
+#define SDMA_CFG_DMA_ARB_WR_ARBIT_MODEL__MASK			0x00000200
+#define SDMA_CFG_DMA_ARB_WR_ARBIT_MODEL__SHIFT			9
+static inline uint32_t SDMA_CFG_DMA_ARB_WR_ARBIT_MODEL(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_ARB_WR_ARBIT_MODEL__SHIFT) & SDMA_CFG_DMA_ARB_WR_ARBIT_MODEL__MASK;
+}
+#define SDMA_CFG_DMA_ARB_RD_ARBIT_MODEL__MASK			0x00000100
+#define SDMA_CFG_DMA_ARB_RD_ARBIT_MODEL__SHIFT			8
+static inline uint32_t SDMA_CFG_DMA_ARB_RD_ARBIT_MODEL(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_ARB_RD_ARBIT_MODEL__SHIFT) & SDMA_CFG_DMA_ARB_RD_ARBIT_MODEL__MASK;
+}
+#define SDMA_CFG_DMA_ARB_RESERVED_1__MASK			0x00000080
+#define SDMA_CFG_DMA_ARB_RESERVED_1__SHIFT			7
+static inline uint32_t SDMA_CFG_DMA_ARB_RESERVED_1(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_ARB_RESERVED_1__SHIFT) & SDMA_CFG_DMA_ARB_RESERVED_1__MASK;
+}
+#define SDMA_CFG_DMA_ARB_WR_FIX_ARB__MASK			0x00000070
+#define SDMA_CFG_DMA_ARB_WR_FIX_ARB__SHIFT			4
+static inline uint32_t SDMA_CFG_DMA_ARB_WR_FIX_ARB(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_ARB_WR_FIX_ARB__SHIFT) & SDMA_CFG_DMA_ARB_WR_FIX_ARB__MASK;
+}
+#define SDMA_CFG_DMA_ARB_RESERVED_2__MASK			0x00000008
+#define SDMA_CFG_DMA_ARB_RESERVED_2__SHIFT			3
+static inline uint32_t SDMA_CFG_DMA_ARB_RESERVED_2(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_ARB_RESERVED_2__SHIFT) & SDMA_CFG_DMA_ARB_RESERVED_2__MASK;
+}
+#define SDMA_CFG_DMA_ARB_RD_FIX_ARB__MASK			0x00000007
+#define SDMA_CFG_DMA_ARB_RD_FIX_ARB__SHIFT			0
+static inline uint32_t SDMA_CFG_DMA_ARB_RD_FIX_ARB(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_ARB_RD_FIX_ARB__SHIFT) & SDMA_CFG_DMA_ARB_RD_FIX_ARB__MASK;
+}
+
+#define REG_SDMA_CFG_DMA_RD_QOS					0x00009020
+#define SDMA_CFG_DMA_RD_QOS_RESERVED_0__MASK			0xfffffc00
+#define SDMA_CFG_DMA_RD_QOS_RESERVED_0__SHIFT			10
+static inline uint32_t SDMA_CFG_DMA_RD_QOS_RESERVED_0(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_QOS_RESERVED_0__SHIFT) & SDMA_CFG_DMA_RD_QOS_RESERVED_0__MASK;
+}
+#define SDMA_CFG_DMA_RD_QOS_RD_PC_QOS__MASK			0x00000300
+#define SDMA_CFG_DMA_RD_QOS_RD_PC_QOS__SHIFT			8
+static inline uint32_t SDMA_CFG_DMA_RD_QOS_RD_PC_QOS(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_QOS_RD_PC_QOS__SHIFT) & SDMA_CFG_DMA_RD_QOS_RD_PC_QOS__MASK;
+}
+#define SDMA_CFG_DMA_RD_QOS_RD_PPU_QOS__MASK			0x000000c0
+#define SDMA_CFG_DMA_RD_QOS_RD_PPU_QOS__SHIFT			6
+static inline uint32_t SDMA_CFG_DMA_RD_QOS_RD_PPU_QOS(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_QOS_RD_PPU_QOS__SHIFT) & SDMA_CFG_DMA_RD_QOS_RD_PPU_QOS__MASK;
+}
+#define SDMA_CFG_DMA_RD_QOS_RD_DPU_QOS__MASK			0x00000030
+#define SDMA_CFG_DMA_RD_QOS_RD_DPU_QOS__SHIFT			4
+static inline uint32_t SDMA_CFG_DMA_RD_QOS_RD_DPU_QOS(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_QOS_RD_DPU_QOS__SHIFT) & SDMA_CFG_DMA_RD_QOS_RD_DPU_QOS__MASK;
+}
+#define SDMA_CFG_DMA_RD_QOS_RD_KERNEL_QOS__MASK			0x0000000c
+#define SDMA_CFG_DMA_RD_QOS_RD_KERNEL_QOS__SHIFT		2
+static inline uint32_t SDMA_CFG_DMA_RD_QOS_RD_KERNEL_QOS(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_QOS_RD_KERNEL_QOS__SHIFT) & SDMA_CFG_DMA_RD_QOS_RD_KERNEL_QOS__MASK;
+}
+#define SDMA_CFG_DMA_RD_QOS_RD_FEATURE_QOS__MASK		0x00000003
+#define SDMA_CFG_DMA_RD_QOS_RD_FEATURE_QOS__SHIFT		0
+static inline uint32_t SDMA_CFG_DMA_RD_QOS_RD_FEATURE_QOS(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_QOS_RD_FEATURE_QOS__SHIFT) & SDMA_CFG_DMA_RD_QOS_RD_FEATURE_QOS__MASK;
+}
+
+#define REG_SDMA_CFG_DMA_RD_CFG					0x00009024
+#define SDMA_CFG_DMA_RD_CFG_RESERVED_0__MASK			0xffffe000
+#define SDMA_CFG_DMA_RD_CFG_RESERVED_0__SHIFT			13
+static inline uint32_t SDMA_CFG_DMA_RD_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_CFG_RESERVED_0__SHIFT) & SDMA_CFG_DMA_RD_CFG_RESERVED_0__MASK;
+}
+#define SDMA_CFG_DMA_RD_CFG_RD_ARLOCK__MASK			0x00001000
+#define SDMA_CFG_DMA_RD_CFG_RD_ARLOCK__SHIFT			12
+static inline uint32_t SDMA_CFG_DMA_RD_CFG_RD_ARLOCK(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_CFG_RD_ARLOCK__SHIFT) & SDMA_CFG_DMA_RD_CFG_RD_ARLOCK__MASK;
+}
+#define SDMA_CFG_DMA_RD_CFG_RD_ARCACHE__MASK			0x00000f00
+#define SDMA_CFG_DMA_RD_CFG_RD_ARCACHE__SHIFT			8
+static inline uint32_t SDMA_CFG_DMA_RD_CFG_RD_ARCACHE(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_CFG_RD_ARCACHE__SHIFT) & SDMA_CFG_DMA_RD_CFG_RD_ARCACHE__MASK;
+}
+#define SDMA_CFG_DMA_RD_CFG_RD_ARPROT__MASK			0x000000e0
+#define SDMA_CFG_DMA_RD_CFG_RD_ARPROT__SHIFT			5
+static inline uint32_t SDMA_CFG_DMA_RD_CFG_RD_ARPROT(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_CFG_RD_ARPROT__SHIFT) & SDMA_CFG_DMA_RD_CFG_RD_ARPROT__MASK;
+}
+#define SDMA_CFG_DMA_RD_CFG_RD_ARBURST__MASK			0x00000018
+#define SDMA_CFG_DMA_RD_CFG_RD_ARBURST__SHIFT			3
+static inline uint32_t SDMA_CFG_DMA_RD_CFG_RD_ARBURST(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_CFG_RD_ARBURST__SHIFT) & SDMA_CFG_DMA_RD_CFG_RD_ARBURST__MASK;
+}
+#define SDMA_CFG_DMA_RD_CFG_RD_ARSIZE__MASK			0x00000007
+#define SDMA_CFG_DMA_RD_CFG_RD_ARSIZE__SHIFT			0
+static inline uint32_t SDMA_CFG_DMA_RD_CFG_RD_ARSIZE(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_RD_CFG_RD_ARSIZE__SHIFT) & SDMA_CFG_DMA_RD_CFG_RD_ARSIZE__MASK;
+}
+
+#define REG_SDMA_CFG_DMA_WR_CFG					0x00009028
+#define SDMA_CFG_DMA_WR_CFG_RESERVED_0__MASK			0xffffe000
+#define SDMA_CFG_DMA_WR_CFG_RESERVED_0__SHIFT			13
+static inline uint32_t SDMA_CFG_DMA_WR_CFG_RESERVED_0(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_WR_CFG_RESERVED_0__SHIFT) & SDMA_CFG_DMA_WR_CFG_RESERVED_0__MASK;
+}
+#define SDMA_CFG_DMA_WR_CFG_WR_AWLOCK__MASK			0x00001000
+#define SDMA_CFG_DMA_WR_CFG_WR_AWLOCK__SHIFT			12
+static inline uint32_t SDMA_CFG_DMA_WR_CFG_WR_AWLOCK(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_WR_CFG_WR_AWLOCK__SHIFT) & SDMA_CFG_DMA_WR_CFG_WR_AWLOCK__MASK;
+}
+#define SDMA_CFG_DMA_WR_CFG_WR_AWCACHE__MASK			0x00000f00
+#define SDMA_CFG_DMA_WR_CFG_WR_AWCACHE__SHIFT			8
+static inline uint32_t SDMA_CFG_DMA_WR_CFG_WR_AWCACHE(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_WR_CFG_WR_AWCACHE__SHIFT) & SDMA_CFG_DMA_WR_CFG_WR_AWCACHE__MASK;
+}
+#define SDMA_CFG_DMA_WR_CFG_WR_AWPROT__MASK			0x000000e0
+#define SDMA_CFG_DMA_WR_CFG_WR_AWPROT__SHIFT			5
+static inline uint32_t SDMA_CFG_DMA_WR_CFG_WR_AWPROT(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_WR_CFG_WR_AWPROT__SHIFT) & SDMA_CFG_DMA_WR_CFG_WR_AWPROT__MASK;
+}
+#define SDMA_CFG_DMA_WR_CFG_WR_AWBURST__MASK			0x00000018
+#define SDMA_CFG_DMA_WR_CFG_WR_AWBURST__SHIFT			3
+static inline uint32_t SDMA_CFG_DMA_WR_CFG_WR_AWBURST(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_WR_CFG_WR_AWBURST__SHIFT) & SDMA_CFG_DMA_WR_CFG_WR_AWBURST__MASK;
+}
+#define SDMA_CFG_DMA_WR_CFG_WR_AWSIZE__MASK			0x00000007
+#define SDMA_CFG_DMA_WR_CFG_WR_AWSIZE__SHIFT			0
+static inline uint32_t SDMA_CFG_DMA_WR_CFG_WR_AWSIZE(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_WR_CFG_WR_AWSIZE__SHIFT) & SDMA_CFG_DMA_WR_CFG_WR_AWSIZE__MASK;
+}
+
+#define REG_SDMA_CFG_DMA_WSTRB					0x0000902c
+#define SDMA_CFG_DMA_WSTRB_WR_WSTRB__MASK			0xffffffff
+#define SDMA_CFG_DMA_WSTRB_WR_WSTRB__SHIFT			0
+static inline uint32_t SDMA_CFG_DMA_WSTRB_WR_WSTRB(uint32_t val)
+{
+	return ((val) << SDMA_CFG_DMA_WSTRB_WR_WSTRB__SHIFT) & SDMA_CFG_DMA_WSTRB_WR_WSTRB__MASK;
+}
+
+#define REG_SDMA_CFG_STATUS					0x00009030
+#define SDMA_CFG_STATUS_RESERVED_0__MASK			0xfffffe00
+#define SDMA_CFG_STATUS_RESERVED_0__SHIFT			9
+static inline uint32_t SDMA_CFG_STATUS_RESERVED_0(uint32_t val)
+{
+	return ((val) << SDMA_CFG_STATUS_RESERVED_0__SHIFT) & SDMA_CFG_STATUS_RESERVED_0__MASK;
+}
+#define SDMA_CFG_STATUS_IDEL__MASK				0x00000100
+#define SDMA_CFG_STATUS_IDEL__SHIFT				8
+static inline uint32_t SDMA_CFG_STATUS_IDEL(uint32_t val)
+{
+	return ((val) << SDMA_CFG_STATUS_IDEL__SHIFT) & SDMA_CFG_STATUS_IDEL__MASK;
+}
+#define SDMA_CFG_STATUS_RESERVED_1__MASK			0x000000ff
+#define SDMA_CFG_STATUS_RESERVED_1__SHIFT			0
+static inline uint32_t SDMA_CFG_STATUS_RESERVED_1(uint32_t val)
+{
+	return ((val) << SDMA_CFG_STATUS_RESERVED_1__SHIFT) & SDMA_CFG_STATUS_RESERVED_1__MASK;
+}
+
+#define REG_GLOBAL_OPERATION_ENABLE				0x0000f008
+#define GLOBAL_OPERATION_ENABLE_RESERVED_0__MASK		0xffffff80
+#define GLOBAL_OPERATION_ENABLE_RESERVED_0__SHIFT		7
+static inline uint32_t GLOBAL_OPERATION_ENABLE_RESERVED_0(uint32_t val)
+{
+	return ((val) << GLOBAL_OPERATION_ENABLE_RESERVED_0__SHIFT) & GLOBAL_OPERATION_ENABLE_RESERVED_0__MASK;
+}
+#define GLOBAL_OPERATION_ENABLE_PPU_RDMA_OP_EN__MASK		0x00000040
+#define GLOBAL_OPERATION_ENABLE_PPU_RDMA_OP_EN__SHIFT		6
+static inline uint32_t GLOBAL_OPERATION_ENABLE_PPU_RDMA_OP_EN(uint32_t val)
+{
+	return ((val) << GLOBAL_OPERATION_ENABLE_PPU_RDMA_OP_EN__SHIFT) & GLOBAL_OPERATION_ENABLE_PPU_RDMA_OP_EN__MASK;
+}
+#define GLOBAL_OPERATION_ENABLE_PPU_OP_EN__MASK			0x00000020
+#define GLOBAL_OPERATION_ENABLE_PPU_OP_EN__SHIFT		5
+static inline uint32_t GLOBAL_OPERATION_ENABLE_PPU_OP_EN(uint32_t val)
+{
+	return ((val) << GLOBAL_OPERATION_ENABLE_PPU_OP_EN__SHIFT) & GLOBAL_OPERATION_ENABLE_PPU_OP_EN__MASK;
+}
+#define GLOBAL_OPERATION_ENABLE_DPU_RDMA_OP_EN__MASK		0x00000010
+#define GLOBAL_OPERATION_ENABLE_DPU_RDMA_OP_EN__SHIFT		4
+static inline uint32_t GLOBAL_OPERATION_ENABLE_DPU_RDMA_OP_EN(uint32_t val)
+{
+	return ((val) << GLOBAL_OPERATION_ENABLE_DPU_RDMA_OP_EN__SHIFT) & GLOBAL_OPERATION_ENABLE_DPU_RDMA_OP_EN__MASK;
+}
+#define GLOBAL_OPERATION_ENABLE_DPU_OP_EN__MASK			0x00000008
+#define GLOBAL_OPERATION_ENABLE_DPU_OP_EN__SHIFT		3
+static inline uint32_t GLOBAL_OPERATION_ENABLE_DPU_OP_EN(uint32_t val)
+{
+	return ((val) << GLOBAL_OPERATION_ENABLE_DPU_OP_EN__SHIFT) & GLOBAL_OPERATION_ENABLE_DPU_OP_EN__MASK;
+}
+#define GLOBAL_OPERATION_ENABLE_CORE_OP_EN__MASK		0x00000004
+#define GLOBAL_OPERATION_ENABLE_CORE_OP_EN__SHIFT		2
+static inline uint32_t GLOBAL_OPERATION_ENABLE_CORE_OP_EN(uint32_t val)
+{
+	return ((val) << GLOBAL_OPERATION_ENABLE_CORE_OP_EN__SHIFT) & GLOBAL_OPERATION_ENABLE_CORE_OP_EN__MASK;
+}
+#define GLOBAL_OPERATION_ENABLE_RESERVED_1__MASK		0x00000002
+#define GLOBAL_OPERATION_ENABLE_RESERVED_1__SHIFT		1
+static inline uint32_t GLOBAL_OPERATION_ENABLE_RESERVED_1(uint32_t val)
+{
+	return ((val) << GLOBAL_OPERATION_ENABLE_RESERVED_1__SHIFT) & GLOBAL_OPERATION_ENABLE_RESERVED_1__MASK;
+}
+#define GLOBAL_OPERATION_ENABLE_CNA_OP_EN__MASK			0x00000001
+#define GLOBAL_OPERATION_ENABLE_CNA_OP_EN__SHIFT		0
+static inline uint32_t GLOBAL_OPERATION_ENABLE_CNA_OP_EN(uint32_t val)
+{
+	return ((val) << GLOBAL_OPERATION_ENABLE_CNA_OP_EN__SHIFT) & GLOBAL_OPERATION_ENABLE_CNA_OP_EN__MASK;
+}
+
+#endif /* __ROCKET_REGISTERS_XML__ */
diff --git a/drivers/acpi/Kconfig b/drivers/acpi/Kconfig
index 7bc40c2735ac..ca00a5dbcf75 100644
--- a/drivers/acpi/Kconfig
+++ b/drivers/acpi/Kconfig
@@ -394,6 +394,7 @@ config ACPI_TABLE_OVERRIDE_VIA_BUILTIN_INITRD
 
 config ACPI_DEBUG
 	bool "Debug Statements"
+	default y
 	help
 	  The ACPI subsystem can produce debug output.  Saying Y enables this
 	  output and increases the kernel size by around 50K.
@@ -460,7 +461,7 @@ config ACPI_HED
 
 config ACPI_BGRT
 	bool "Boottime Graphics Resource Table support"
-	depends on EFI && (X86 || ARM64 || LOONGARCH)
+	depends on EFI
 	help
 	  This driver adds support for exposing the ACPI Boottime Graphics
 	  Resource Table, which allows the operating system to obtain
@@ -546,6 +547,10 @@ if ARM64
 source "drivers/acpi/arm64/Kconfig"
 endif
 
+if RISCV
+source "drivers/acpi/riscv/Kconfig"
+endif
+
 config ACPI_PPTT
 	bool
 
diff --git a/drivers/acpi/acpi_dbg.c b/drivers/acpi/acpi_dbg.c
index d50261d05f3a..515b20d0b698 100644
--- a/drivers/acpi/acpi_dbg.c
+++ b/drivers/acpi/acpi_dbg.c
@@ -569,11 +569,11 @@ static int acpi_aml_release(struct inode *inode, struct file *file)
 	return 0;
 }
 
-static int acpi_aml_read_user(char __user *buf, int len)
+static ssize_t acpi_aml_read_user(char __user *buf, size_t len)
 {
-	int ret;
 	struct circ_buf *crc = &acpi_aml_io.out_crc;
-	int n;
+	ssize_t ret;
+	size_t n;
 	char *p;
 
 	ret = acpi_aml_lock_read(crc, ACPI_AML_OUT_USER);
@@ -582,7 +582,7 @@ static int acpi_aml_read_user(char __user *buf, int len)
 	/* sync head before removing logs */
 	smp_rmb();
 	p = &crc->buf[crc->tail];
-	n = min(len, circ_count_to_end(crc));
+	n = min_t(size_t, len, circ_count_to_end(crc));
 	if (copy_to_user(buf, p, n)) {
 		ret = -EFAULT;
 		goto out;
@@ -599,8 +599,8 @@ out:
 static ssize_t acpi_aml_read(struct file *file, char __user *buf,
 			     size_t count, loff_t *ppos)
 {
-	int ret = 0;
-	int size = 0;
+	ssize_t ret = 0;
+	ssize_t size = 0;
 
 	if (!count)
 		return 0;
@@ -639,11 +639,11 @@ again:
 	return size > 0 ? size : ret;
 }
 
-static int acpi_aml_write_user(const char __user *buf, int len)
+static ssize_t acpi_aml_write_user(const char __user *buf, size_t len)
 {
-	int ret;
 	struct circ_buf *crc = &acpi_aml_io.in_crc;
-	int n;
+	ssize_t ret;
+	size_t n;
 	char *p;
 
 	ret = acpi_aml_lock_write(crc, ACPI_AML_IN_USER);
@@ -652,7 +652,7 @@ static int acpi_aml_write_user(const char __user *buf, int len)
 	/* sync tail before inserting cmds */
 	smp_mb();
 	p = &crc->buf[crc->head];
-	n = min(len, circ_space_to_end(crc));
+	n = min_t(size_t, len, circ_space_to_end(crc));
 	if (copy_from_user(p, buf, n)) {
 		ret = -EFAULT;
 		goto out;
@@ -663,14 +663,14 @@ static int acpi_aml_write_user(const char __user *buf, int len)
 	ret = n;
 out:
 	acpi_aml_unlock_fifo(ACPI_AML_IN_USER, ret >= 0);
-	return n;
+	return ret;
 }
 
 static ssize_t acpi_aml_write(struct file *file, const char __user *buf,
 			      size_t count, loff_t *ppos)
 {
-	int ret = 0;
-	int size = 0;
+	ssize_t ret = 0;
+	ssize_t size = 0;
 
 	if (!count)
 		return 0;
diff --git a/drivers/acpi/acpi_mrrm.c b/drivers/acpi/acpi_mrrm.c
index 47ea3ccc2142..6d69554c940e 100644
--- a/drivers/acpi/acpi_mrrm.c
+++ b/drivers/acpi/acpi_mrrm.c
@@ -63,6 +63,9 @@ static __init int acpi_parse_mrrm(struct acpi_table_header *table)
 	if (!mrrm)
 		return -ENODEV;
 
+	if (mrrm->header.revision != 1)
+		return -EINVAL;
+
 	if (mrrm->flags & ACPI_MRRM_FLAGS_REGION_ASSIGNMENT_OS)
 		return -EOPNOTSUPP;
 
@@ -149,26 +152,49 @@ ATTRIBUTE_GROUPS(memory_range);
 
 static __init int add_boot_memory_ranges(void)
 {
-	struct kobject *pkobj, *kobj;
+	struct kobject *pkobj, *kobj, **kobjs;
 	int ret = -EINVAL;
-	char *name;
+	char name[16];
+	int i;
 
 	pkobj = kobject_create_and_add("memory_ranges", acpi_kobj);
+	if (!pkobj)
+		return -ENOMEM;
 
-	for (int i = 0; i < mrrm_mem_entry_num; i++) {
-		name = kasprintf(GFP_KERNEL, "range%d", i);
-		if (!name) {
-			ret = -ENOMEM;
-			break;
-		}
+	kobjs = kcalloc(mrrm_mem_entry_num, sizeof(*kobjs), GFP_KERNEL);
+	if (!kobjs) {
+		kobject_put(pkobj);
+		return -ENOMEM;
+	}
 
+	for (i = 0; i < mrrm_mem_entry_num; i++) {
+		scnprintf(name, sizeof(name), "range%d", i);
 		kobj = kobject_create_and_add(name, pkobj);
+		if (!kobj) {
+			ret = -ENOMEM;
+			goto cleanup;
+		}
 
 		ret = sysfs_create_groups(kobj, memory_range_groups);
-		if (ret)
-			return ret;
+		if (ret) {
+			kobject_put(kobj);
+			goto cleanup;
+		}
+		kobjs[i] = kobj;
 	}
 
+	kfree(kobjs);
+	return 0;
+
+cleanup:
+	for (int j = 0; j < i; j++) {
+		if (kobjs[j]) {
+			sysfs_remove_groups(kobjs[j], memory_range_groups);
+			kobject_put(kobjs[j]);
+		}
+	}
+	kfree(kobjs);
+	kobject_put(pkobj);
 	return ret;
 }
 
diff --git a/drivers/acpi/acpi_pad.c b/drivers/acpi/acpi_pad.c
index 6f8bbe1247a5..c9a0bcaba2e4 100644
--- a/drivers/acpi/acpi_pad.c
+++ b/drivers/acpi/acpi_pad.c
@@ -33,7 +33,7 @@
 static DEFINE_MUTEX(isolated_cpus_lock);
 static DEFINE_MUTEX(round_robin_lock);
 
-static unsigned long power_saving_mwait_eax;
+static unsigned int power_saving_mwait_eax;
 
 static unsigned char tsc_detected_unstable;
 static unsigned char tsc_marked_unstable;
diff --git a/drivers/acpi/acpi_processor.c b/drivers/acpi/acpi_processor.c
index 7cf6101cb4c7..7ec1dc04fd11 100644
--- a/drivers/acpi/acpi_processor.c
+++ b/drivers/acpi/acpi_processor.c
@@ -275,7 +275,7 @@ static inline int acpi_processor_hotadd_init(struct acpi_processor *pr,
 
 static int acpi_processor_get_info(struct acpi_device *device)
 {
-	union acpi_object object = { 0 };
+	union acpi_object object = { .processor = { 0 } };
 	struct acpi_buffer buffer = { sizeof(union acpi_object), &object };
 	struct acpi_processor *pr = acpi_driver_data(device);
 	int device_declaration = 0;
@@ -815,7 +815,7 @@ bool acpi_processor_claim_cst_control(void)
 	cst_control_claimed = true;
 	return true;
 }
-EXPORT_SYMBOL_GPL(acpi_processor_claim_cst_control);
+EXPORT_SYMBOL_NS_GPL(acpi_processor_claim_cst_control, "ACPI_PROCESSOR_IDLE");
 
 /**
  * acpi_processor_evaluate_cst - Evaluate the processor _CST control method.
@@ -994,5 +994,5 @@ end:
 
 	return ret;
 }
-EXPORT_SYMBOL_GPL(acpi_processor_evaluate_cst);
+EXPORT_SYMBOL_NS_GPL(acpi_processor_evaluate_cst, "ACPI_PROCESSOR_IDLE");
 #endif /* CONFIG_ACPI_PROCESSOR_CSTATE */
diff --git a/drivers/acpi/acpi_tad.c b/drivers/acpi/acpi_tad.c
index 825c2a8acea4..6d870d97ada6 100644
--- a/drivers/acpi/acpi_tad.c
+++ b/drivers/acpi/acpi_tad.c
@@ -90,19 +90,18 @@ static int acpi_tad_set_real_time(struct device *dev, struct acpi_tad_rt *rt)
 	args[0].buffer.pointer = (u8 *)rt;
 	args[0].buffer.length = sizeof(*rt);
 
-	pm_runtime_get_sync(dev);
+	PM_RUNTIME_ACQUIRE(dev, pm);
+	if (PM_RUNTIME_ACQUIRE_ERR(&pm))
+		return -ENXIO;
 
 	status = acpi_evaluate_integer(handle, "_SRT", &arg_list, &retval);
-
-	pm_runtime_put_sync(dev);
-
 	if (ACPI_FAILURE(status) || retval)
 		return -EIO;
 
 	return 0;
 }
 
-static int acpi_tad_get_real_time(struct device *dev, struct acpi_tad_rt *rt)
+static int acpi_tad_evaluate_grt(struct device *dev, struct acpi_tad_rt *rt)
 {
 	acpi_handle handle = ACPI_HANDLE(dev);
 	struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER };
@@ -111,12 +110,7 @@ static int acpi_tad_get_real_time(struct device *dev, struct acpi_tad_rt *rt)
 	acpi_status status;
 	int ret = -EIO;
 
-	pm_runtime_get_sync(dev);
-
 	status = acpi_evaluate_object(handle, "_GRT", NULL, &output);
-
-	pm_runtime_put_sync(dev);
-
 	if (ACPI_FAILURE(status))
 		goto out_free;
 
@@ -139,6 +133,21 @@ out_free:
 	return ret;
 }
 
+static int acpi_tad_get_real_time(struct device *dev, struct acpi_tad_rt *rt)
+{
+	int ret;
+
+	PM_RUNTIME_ACQUIRE(dev, pm);
+	if (PM_RUNTIME_ACQUIRE_ERR(&pm))
+		return -ENXIO;
+
+	ret = acpi_tad_evaluate_grt(dev, rt);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
 static char *acpi_tad_rt_next_field(char *s, int *val)
 {
 	char *p;
@@ -233,7 +242,7 @@ static ssize_t time_show(struct device *dev, struct device_attribute *attr,
 	if (ret)
 		return ret;
 
-	return sprintf(buf, "%u:%u:%u:%u:%u:%u:%d:%u\n",
+	return sysfs_emit(buf, "%u:%u:%u:%u:%u:%u:%d:%u\n",
 		       rt.year, rt.month, rt.day, rt.hour, rt.minute, rt.second,
 		       rt.tz, rt.daylight);
 }
@@ -266,12 +275,11 @@ static int acpi_tad_wake_set(struct device *dev, char *method, u32 timer_id,
 	args[0].integer.value = timer_id;
 	args[1].integer.value = value;
 
-	pm_runtime_get_sync(dev);
+	PM_RUNTIME_ACQUIRE(dev, pm);
+	if (PM_RUNTIME_ACQUIRE_ERR(&pm))
+		return -ENXIO;
 
 	status = acpi_evaluate_integer(handle, method, &arg_list, &retval);
-
-	pm_runtime_put_sync(dev);
-
 	if (ACPI_FAILURE(status) || retval)
 		return -EIO;
 
@@ -314,12 +322,11 @@ static ssize_t acpi_tad_wake_read(struct device *dev, char *buf, char *method,
 
 	args[0].integer.value = timer_id;
 
-	pm_runtime_get_sync(dev);
+	PM_RUNTIME_ACQUIRE(dev, pm);
+	if (PM_RUNTIME_ACQUIRE_ERR(&pm))
+		return -ENXIO;
 
 	status = acpi_evaluate_integer(handle, method, &arg_list, &retval);
-
-	pm_runtime_put_sync(dev);
-
 	if (ACPI_FAILURE(status))
 		return -EIO;
 
@@ -370,12 +377,11 @@ static int acpi_tad_clear_status(struct device *dev, u32 timer_id)
 
 	args[0].integer.value = timer_id;
 
-	pm_runtime_get_sync(dev);
+	PM_RUNTIME_ACQUIRE(dev, pm);
+	if (PM_RUNTIME_ACQUIRE_ERR(&pm))
+		return -ENXIO;
 
 	status = acpi_evaluate_integer(handle, "_CWS", &arg_list, &retval);
-
-	pm_runtime_put_sync(dev);
-
 	if (ACPI_FAILURE(status) || retval)
 		return -EIO;
 
@@ -411,12 +417,11 @@ static ssize_t acpi_tad_status_read(struct device *dev, char *buf, u32 timer_id)
 
 	args[0].integer.value = timer_id;
 
-	pm_runtime_get_sync(dev);
+	PM_RUNTIME_ACQUIRE(dev, pm);
+	if (PM_RUNTIME_ACQUIRE_ERR(&pm))
+		return -ENXIO;
 
 	status = acpi_evaluate_integer(handle, "_GWS", &arg_list, &retval);
-
-	pm_runtime_put_sync(dev);
-
 	if (ACPI_FAILURE(status))
 		return -EIO;
 
@@ -428,7 +433,7 @@ static ssize_t caps_show(struct device *dev, struct device_attribute *attr,
 {
 	struct acpi_tad_driver_data *dd = dev_get_drvdata(dev);
 
-	return sprintf(buf, "0x%02X\n", dd->capabilities);
+	return sysfs_emit(buf, "0x%02X\n", dd->capabilities);
 }
 
 static DEVICE_ATTR_RO(caps);
@@ -563,21 +568,24 @@ static void acpi_tad_remove(struct platform_device *pdev)
 
 	device_init_wakeup(dev, false);
 
-	pm_runtime_get_sync(dev);
+	if (dd->capabilities & ACPI_TAD_RT)
+		sysfs_remove_group(&dev->kobj, &acpi_tad_time_attr_group);
 
 	if (dd->capabilities & ACPI_TAD_DC_WAKE)
 		sysfs_remove_group(&dev->kobj, &acpi_tad_dc_attr_group);
 
 	sysfs_remove_group(&dev->kobj, &acpi_tad_attr_group);
 
-	acpi_tad_disable_timer(dev, ACPI_TAD_AC_TIMER);
-	acpi_tad_clear_status(dev, ACPI_TAD_AC_TIMER);
-	if (dd->capabilities & ACPI_TAD_DC_WAKE) {
-		acpi_tad_disable_timer(dev, ACPI_TAD_DC_TIMER);
-		acpi_tad_clear_status(dev, ACPI_TAD_DC_TIMER);
+	scoped_guard(pm_runtime_noresume, dev) {
+		acpi_tad_disable_timer(dev, ACPI_TAD_AC_TIMER);
+		acpi_tad_clear_status(dev, ACPI_TAD_AC_TIMER);
+		if (dd->capabilities & ACPI_TAD_DC_WAKE) {
+			acpi_tad_disable_timer(dev, ACPI_TAD_DC_TIMER);
+			acpi_tad_clear_status(dev, ACPI_TAD_DC_TIMER);
+		}
 	}
 
-	pm_runtime_put_sync(dev);
+	pm_runtime_suspend(dev);
 	pm_runtime_disable(dev);
 	acpi_remove_cmos_rtc_space_handler(handle);
 }
diff --git a/drivers/acpi/acpi_video.c b/drivers/acpi/acpi_video.c
index 103f29661576..be8e7e18abca 100644
--- a/drivers/acpi/acpi_video.c
+++ b/drivers/acpi/acpi_video.c
@@ -1959,8 +1959,10 @@ static void acpi_video_bus_remove_notify_handler(struct acpi_video_bus *video)
 	struct acpi_video_device *dev;
 
 	mutex_lock(&video->device_list_lock);
-	list_for_each_entry(dev, &video->video_device_list, entry)
+	list_for_each_entry(dev, &video->video_device_list, entry) {
 		acpi_video_dev_remove_notify_handler(dev);
+		cancel_delayed_work_sync(&dev->switch_brightness_work);
+	}
 	mutex_unlock(&video->device_list_lock);
 
 	acpi_video_bus_stop_devices(video);
diff --git a/drivers/acpi/acpica/acdebug.h b/drivers/acpi/acpica/acdebug.h
index fe6d38b43c9a..91241bd6917a 100644
--- a/drivers/acpi/acpica/acdebug.h
+++ b/drivers/acpi/acpica/acdebug.h
@@ -37,7 +37,7 @@ struct acpi_db_argument_info {
 struct acpi_db_execute_walk {
 	u32 count;
 	u32 max_count;
-	char name_seg[ACPI_NAMESEG_SIZE + 1] ACPI_NONSTRING;
+	char name_seg[ACPI_NAMESEG_SIZE + 1];
 };
 
 #define PARAM_LIST(pl)                  pl
diff --git a/drivers/acpi/acpica/aclocal.h b/drivers/acpi/acpica/aclocal.h
index 0c41f0097e8d..f98640086f4e 100644
--- a/drivers/acpi/acpica/aclocal.h
+++ b/drivers/acpi/acpica/aclocal.h
@@ -1141,7 +1141,7 @@ struct acpi_port_info {
 #define ACPI_RESOURCE_NAME_PIN_GROUP_FUNCTION   0x91
 #define ACPI_RESOURCE_NAME_PIN_GROUP_CONFIG     0x92
 #define ACPI_RESOURCE_NAME_CLOCK_INPUT          0x93
-#define ACPI_RESOURCE_NAME_LARGE_MAX            0x94
+#define ACPI_RESOURCE_NAME_LARGE_MAX            0x93
 
 /*****************************************************************************
  *
diff --git a/drivers/acpi/acpica/acpredef.h b/drivers/acpi/acpica/acpredef.h
index 76c5ed02e916..da2c45880cc7 100644
--- a/drivers/acpi/acpica/acpredef.h
+++ b/drivers/acpi/acpica/acpredef.h
@@ -450,7 +450,8 @@ const union acpi_predefined_info acpi_gbl_predefined_methods[] = {
 
 	{{"_DSM",
 	  METHOD_4ARGS(ACPI_TYPE_BUFFER, ACPI_TYPE_INTEGER, ACPI_TYPE_INTEGER,
-		       ACPI_TYPE_ANY) | ARG_COUNT_IS_MINIMUM,
+		       ACPI_TYPE_ANY | ACPI_TYPE_PACKAGE) |
+		       ARG_COUNT_IS_MINIMUM,
 	  METHOD_RETURNS(ACPI_RTYPE_ALL)}},	/* Must return a value, but it can be of any type */
 
 	{{"_DSS", METHOD_1ARGS(ACPI_TYPE_INTEGER),
diff --git a/drivers/acpi/acpica/dsmethod.c b/drivers/acpi/acpica/dsmethod.c
index c8f37f4e6626..45ec32e81903 100644
--- a/drivers/acpi/acpica/dsmethod.c
+++ b/drivers/acpi/acpica/dsmethod.c
@@ -462,7 +462,6 @@ acpi_ds_call_control_method(struct acpi_thread_state *thread,
 	struct acpi_walk_state *next_walk_state = NULL;
 	union acpi_operand_object *obj_desc;
 	struct acpi_evaluate_info *info;
-	u32 i;
 
 	ACPI_FUNCTION_TRACE_PTR(ds_call_control_method, this_walk_state);
 
@@ -483,6 +482,20 @@ acpi_ds_call_control_method(struct acpi_thread_state *thread,
 		return_ACPI_STATUS(AE_NULL_OBJECT);
 	}
 
+	if (this_walk_state->num_operands < obj_desc->method.param_count) {
+		ACPI_ERROR((AE_INFO, "Missing argument(s) for method [%4.4s]",
+			    acpi_ut_get_node_name(method_node)));
+
+		return_ACPI_STATUS(AE_AML_TOO_FEW_ARGUMENTS);
+	}
+
+	else if (this_walk_state->num_operands > obj_desc->method.param_count) {
+		ACPI_ERROR((AE_INFO, "Too many arguments for method [%4.4s]",
+			    acpi_ut_get_node_name(method_node)));
+
+		return_ACPI_STATUS(AE_AML_TOO_MANY_ARGUMENTS);
+	}
+
 	/* Init for new method, possibly wait on method mutex */
 
 	status =
@@ -539,14 +552,7 @@ acpi_ds_call_control_method(struct acpi_thread_state *thread,
 	 * Delete the operands on the previous walkstate operand stack
 	 * (they were copied to new objects)
 	 */
-	for (i = 0; i < obj_desc->method.param_count; i++) {
-		acpi_ut_remove_reference(this_walk_state->operands[i]);
-		this_walk_state->operands[i] = NULL;
-	}
-
-	/* Clear the operand stack */
-
-	this_walk_state->num_operands = 0;
+	acpi_ds_clear_operands(this_walk_state);
 
 	ACPI_DEBUG_PRINT((ACPI_DB_DISPATCH,
 			  "**** Begin nested execution of [%4.4s] **** WalkState=%p\n",
diff --git a/drivers/acpi/acpica/evglock.c b/drivers/acpi/acpica/evglock.c
index fa3e0d00d1ca..df2a4ab0e0da 100644
--- a/drivers/acpi/acpica/evglock.c
+++ b/drivers/acpi/acpica/evglock.c
@@ -42,6 +42,10 @@ acpi_status acpi_ev_init_global_lock_handler(void)
 		return_ACPI_STATUS(AE_OK);
 	}
 
+	if (!acpi_gbl_use_global_lock) {
+		return_ACPI_STATUS(AE_OK);
+	}
+
 	/* Attempt installation of the global lock handler */
 
 	status = acpi_install_fixed_event_handler(ACPI_EVENT_GLOBAL,
diff --git a/drivers/acpi/acpica/extrace.c b/drivers/acpi/acpica/extrace.c
index d34497f3576a..36934d4f26fb 100644
--- a/drivers/acpi/acpica/extrace.c
+++ b/drivers/acpi/acpica/extrace.c
@@ -136,9 +136,9 @@ acpi_ex_trace_point(acpi_trace_event_type type,
 
 	if (pathname) {
 		ACPI_DEBUG_PRINT((ACPI_DB_TRACE_POINT,
-				  "%s %s [0x%p:%s] execution.\n",
+				  "%s %s [%s] execution.\n",
 				  acpi_ex_get_trace_event_name(type),
-				  begin ? "Begin" : "End", aml, pathname));
+				  begin ? "Begin" : "End", pathname));
 	} else {
 		ACPI_DEBUG_PRINT((ACPI_DB_TRACE_POINT,
 				  "%s %s [0x%p] execution.\n",
diff --git a/drivers/acpi/acpica/nswalk.c b/drivers/acpi/acpica/nswalk.c
index a2ac06a26e92..5670ff5a43cd 100644
--- a/drivers/acpi/acpica/nswalk.c
+++ b/drivers/acpi/acpica/nswalk.c
@@ -169,9 +169,12 @@ acpi_ns_walk_namespace(acpi_object_type type,
 
 	if (start_node == ACPI_ROOT_OBJECT) {
 		start_node = acpi_gbl_root_node;
-		if (!start_node) {
-			return_ACPI_STATUS(AE_NO_NAMESPACE);
-		}
+	}
+
+	/* Avoid walking the namespace if the StartNode is NULL */
+
+	if (!start_node) {
+		return_ACPI_STATUS(AE_NO_NAMESPACE);
 	}
 
 	/* Null child means "get first node" */
diff --git a/drivers/acpi/acpica/psopinfo.c b/drivers/acpi/acpica/psopinfo.c
index 1c8044ffcb97..532ea307a675 100644
--- a/drivers/acpi/acpica/psopinfo.c
+++ b/drivers/acpi/acpica/psopinfo.c
@@ -34,7 +34,7 @@ static const u8 acpi_gbl_argument_count[] =
 
 const struct acpi_opcode_info *acpi_ps_get_opcode_info(u16 opcode)
 {
-#ifdef ACPI_DEBUG_OUTPUT
+#if defined ACPI_ASL_COMPILER && defined ACPI_DEBUG_OUTPUT
 	const char *opcode_name = "Unknown AML opcode";
 #endif
 
@@ -102,11 +102,11 @@ const struct acpi_opcode_info *acpi_ps_get_opcode_info(u16 opcode)
 	default:
 		break;
 	}
-#endif
 
 	/* Unknown AML opcode */
 
 	ACPI_DEBUG_PRINT((ACPI_DB_EXEC, "%s [%4.4X]\n", opcode_name, opcode));
+#endif
 
 	return (&acpi_gbl_aml_op_info[_UNK]);
 }
diff --git a/drivers/acpi/acpica/tbprint.c b/drivers/acpi/acpica/tbprint.c
index fd64460a2e26..e5631027f7f1 100644
--- a/drivers/acpi/acpica/tbprint.c
+++ b/drivers/acpi/acpica/tbprint.c
@@ -95,6 +95,11 @@ acpi_tb_print_table_header(acpi_physical_address address,
 {
 	struct acpi_table_header local_header;
 
+#pragma GCC diagnostic push
+#if defined(__GNUC__) && __GNUC__ >= 11
+#pragma GCC diagnostic ignored "-Wstringop-overread"
+#endif
+
 	if (ACPI_COMPARE_NAMESEG(header->signature, ACPI_SIG_FACS)) {
 
 		/* FACS only has signature and length fields */
@@ -121,6 +126,14 @@ acpi_tb_print_table_header(acpi_physical_address address,
 			   ACPI_CAST_PTR(struct acpi_table_rsdp,
 					 header)->revision,
 			   local_header.oem_id));
+	} else if (acpi_gbl_CDAT && !acpi_ut_valid_nameseg(header->signature)) {
+
+		/* CDAT does not use the common ACPI table header */
+
+		ACPI_INFO(("%-4.4s 0x%8.8X%8.8X %06X",
+			   ACPI_SIG_CDAT, ACPI_FORMAT_UINT64(address),
+			   ACPI_CAST_PTR(struct acpi_table_cdat,
+					 header)->length));
 	} else {
 		/* Standard ACPI table with full common header */
 
@@ -135,4 +148,5 @@ acpi_tb_print_table_header(acpi_physical_address address,
 			   local_header.asl_compiler_id,
 			   local_header.asl_compiler_revision));
 	}
+#pragma GCC diagnostic pop
 }
diff --git a/drivers/acpi/apei/apei-internal.h b/drivers/acpi/apei/apei-internal.h
index cd2766c69d78..77c10a7a7a9f 100644
--- a/drivers/acpi/apei/apei-internal.h
+++ b/drivers/acpi/apei/apei-internal.h
@@ -131,7 +131,7 @@ static inline u32 cper_estatus_len(struct acpi_hest_generic_status *estatus)
 
 int apei_osc_setup(void);
 
-int einj_get_available_error_type(u32 *type);
+int einj_get_available_error_type(u32 *type, int einj_action);
 int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2, u64 param3,
 		      u64 param4);
 int einj_cxl_rch_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
diff --git a/drivers/acpi/apei/einj-core.c b/drivers/acpi/apei/einj-core.c
index fea11a35eea3..305c240a303f 100644
--- a/drivers/acpi/apei/einj-core.c
+++ b/drivers/acpi/apei/einj-core.c
@@ -33,6 +33,8 @@
 #define SLEEP_UNIT_MAX		5000			/* 5ms */
 /* Firmware should respond within 1 seconds */
 #define FIRMWARE_TIMEOUT	(1 * USEC_PER_SEC)
+#define COMPONENT_LEN		16
+#define ACPI65_EINJV2_SUPP	BIT(30)
 #define ACPI5_VENDOR_BIT	BIT(31)
 #define MEM_ERROR_MASK		(ACPI_EINJ_MEMORY_CORRECTABLE | \
 				ACPI_EINJ_MEMORY_UNCORRECTABLE | \
@@ -49,6 +51,28 @@
  */
 static int acpi5;
 
+struct syndrome_array {
+	union {
+		u8	acpi_id[COMPONENT_LEN];
+		u8	device_id[COMPONENT_LEN];
+		u8	pcie_sbdf[COMPONENT_LEN];
+		u8	vendor_id[COMPONENT_LEN];
+	} comp_id;
+	union {
+		u8	proc_synd[COMPONENT_LEN];
+		u8	mem_synd[COMPONENT_LEN];
+		u8	pcie_synd[COMPONENT_LEN];
+		u8	vendor_synd[COMPONENT_LEN];
+	} comp_synd;
+};
+
+struct einjv2_extension_struct {
+	u32 length;
+	u16 revision;
+	u16 component_arr_count;
+	struct syndrome_array component_arr[] __counted_by(component_arr_count);
+};
+
 struct set_error_type_with_address {
 	u32	type;
 	u32	vendor_extension;
@@ -57,11 +81,13 @@ struct set_error_type_with_address {
 	u64	memory_address;
 	u64	memory_address_range;
 	u32	pcie_sbdf;
+	struct	einjv2_extension_struct einjv2_struct;
 };
 enum {
 	SETWA_FLAGS_APICID = 1,
 	SETWA_FLAGS_MEM = 2,
 	SETWA_FLAGS_PCIE_SBDF = 4,
+	SETWA_FLAGS_EINJV2 = 8,
 };
 
 /*
@@ -83,7 +109,10 @@ static struct debugfs_blob_wrapper vendor_blob;
 static struct debugfs_blob_wrapper vendor_errors;
 static char vendor_dev[64];
 
+static u32 max_nr_components;
 static u32 available_error_type;
+static u32 available_error_type_v2;
+static struct syndrome_array *syndrome_data;
 
 /*
  * Some BIOSes allow parameters to the SET_ERROR_TYPE entries in the
@@ -151,7 +180,10 @@ static DEFINE_MUTEX(einj_mutex);
  */
 bool einj_initialized __ro_after_init;
 
-static void *einj_param;
+static void __iomem *einj_param;
+static u32 v5param_size;
+static u32 v66param_size;
+static bool is_v2;
 
 static void einj_exec_ctx_init(struct apei_exec_context *ctx)
 {
@@ -159,13 +191,13 @@ static void einj_exec_ctx_init(struct apei_exec_context *ctx)
 			   EINJ_TAB_ENTRY(einj_tab), einj_tab->entries);
 }
 
-static int __einj_get_available_error_type(u32 *type)
+static int __einj_get_available_error_type(u32 *type, int einj_action)
 {
 	struct apei_exec_context ctx;
 	int rc;
 
 	einj_exec_ctx_init(&ctx);
-	rc = apei_exec_run(&ctx, ACPI_EINJ_GET_ERROR_TYPE);
+	rc = apei_exec_run(&ctx, einj_action);
 	if (rc)
 		return rc;
 	*type = apei_exec_ctx_get_output(&ctx);
@@ -174,17 +206,34 @@ static int __einj_get_available_error_type(u32 *type)
 }
 
 /* Get error injection capabilities of the platform */
-int einj_get_available_error_type(u32 *type)
+int einj_get_available_error_type(u32 *type, int einj_action)
 {
 	int rc;
 
 	mutex_lock(&einj_mutex);
-	rc = __einj_get_available_error_type(type);
+	rc = __einj_get_available_error_type(type, einj_action);
 	mutex_unlock(&einj_mutex);
 
 	return rc;
 }
 
+static int einj_get_available_error_types(u32 *type1, u32 *type2)
+{
+	int rc;
+
+	rc = einj_get_available_error_type(type1, ACPI_EINJ_GET_ERROR_TYPE);
+	if (rc)
+		return rc;
+	if (*type1 & ACPI65_EINJV2_SUPP) {
+		rc = einj_get_available_error_type(type2,
+						   ACPI_EINJV2_GET_ERROR_TYPE);
+		if (rc)
+			return rc;
+	}
+
+	return 0;
+}
+
 static int einj_timedout(u64 *t)
 {
 	if ((s64)*t < SLEEP_UNIT_MIN) {
@@ -216,24 +265,44 @@ static void check_vendor_extension(u64 paddr,
 				   struct set_error_type_with_address *v5param)
 {
 	int	offset = v5param->vendor_extension;
-	struct	vendor_error_type_extension *v;
+	struct	vendor_error_type_extension v;
+	struct vendor_error_type_extension __iomem *p;
 	u32	sbdf;
 
 	if (!offset)
 		return;
-	v = acpi_os_map_iomem(paddr + offset, sizeof(*v));
-	if (!v)
+	p = acpi_os_map_iomem(paddr + offset, sizeof(*p));
+	if (!p)
 		return;
-	get_oem_vendor_struct(paddr, offset, v);
-	sbdf = v->pcie_sbdf;
+	memcpy_fromio(&v, p, sizeof(v));
+	get_oem_vendor_struct(paddr, offset, &v);
+	sbdf = v.pcie_sbdf;
 	sprintf(vendor_dev, "%x:%x:%x.%x vendor_id=%x device_id=%x rev_id=%x\n",
 		sbdf >> 24, (sbdf >> 16) & 0xff,
 		(sbdf >> 11) & 0x1f, (sbdf >> 8) & 0x7,
-		 v->vendor_id, v->device_id, v->rev_id);
-	acpi_os_unmap_iomem(v, sizeof(*v));
+		 v.vendor_id, v.device_id, v.rev_id);
+	acpi_os_unmap_iomem(p, sizeof(v));
 }
 
-static void *einj_get_parameter_address(void)
+static u32 einjv2_init(struct einjv2_extension_struct *e)
+{
+	if (e->revision != 1) {
+		pr_info("Unknown v2 extension revision %u\n", e->revision);
+		return 0;
+	}
+	if (e->length < sizeof(*e) || e->length > PAGE_SIZE) {
+		pr_info(FW_BUG "Bad1 v2 extension length %u\n", e->length);
+		return 0;
+	}
+	if ((e->length - sizeof(*e)) % sizeof(e->component_arr[0])) {
+		pr_info(FW_BUG "Bad2 v2 extension length %u\n", e->length);
+		return 0;
+	}
+
+	return (e->length - sizeof(*e)) / sizeof(e->component_arr[0]);
+}
+
+static void __iomem *einj_get_parameter_address(void)
 {
 	int i;
 	u64 pa_v4 = 0, pa_v5 = 0;
@@ -254,26 +323,43 @@ static void *einj_get_parameter_address(void)
 		entry++;
 	}
 	if (pa_v5) {
-		struct set_error_type_with_address *v5param;
+		struct set_error_type_with_address v5param;
+		struct set_error_type_with_address __iomem *p;
 
-		v5param = acpi_os_map_iomem(pa_v5, sizeof(*v5param));
-		if (v5param) {
+		v5param_size = sizeof(v5param);
+		p = acpi_os_map_iomem(pa_v5, sizeof(*p));
+		if (p) {
+			memcpy_fromio(&v5param, p, v5param_size);
 			acpi5 = 1;
-			check_vendor_extension(pa_v5, v5param);
-			return v5param;
+			check_vendor_extension(pa_v5, &v5param);
+			if (available_error_type & ACPI65_EINJV2_SUPP) {
+				struct einjv2_extension_struct *e;
+
+				e = &v5param.einjv2_struct;
+				max_nr_components = einjv2_init(e);
+
+				/* remap including einjv2_extension_struct */
+				acpi_os_unmap_iomem(p, v5param_size);
+				v66param_size = v5param_size - sizeof(*e) + e->length;
+				p = acpi_os_map_iomem(pa_v5, v66param_size);
+			}
+
+			return p;
 		}
 	}
 	if (param_extension && pa_v4) {
-		struct einj_parameter *v4param;
+		struct einj_parameter v4param;
+		struct einj_parameter __iomem *p;
 
-		v4param = acpi_os_map_iomem(pa_v4, sizeof(*v4param));
-		if (!v4param)
+		p = acpi_os_map_iomem(pa_v4, sizeof(*p));
+		if (!p)
 			return NULL;
-		if (v4param->reserved1 || v4param->reserved2) {
-			acpi_os_unmap_iomem(v4param, sizeof(*v4param));
+		memcpy_fromio(&v4param, p, sizeof(v4param));
+		if (v4param.reserved1 || v4param.reserved2) {
+			acpi_os_unmap_iomem(p, sizeof(v4param));
 			return NULL;
 		}
-		return v4param;
+		return p;
 	}
 
 	return NULL;
@@ -319,7 +405,8 @@ static struct acpi_generic_address *einj_get_trigger_parameter_region(
 static int __einj_error_trigger(u64 trigger_paddr, u32 type,
 				u64 param1, u64 param2)
 {
-	struct acpi_einj_trigger *trigger_tab = NULL;
+	struct acpi_einj_trigger trigger_tab;
+	struct acpi_einj_trigger *full_trigger_tab;
 	struct apei_exec_context trigger_ctx;
 	struct apei_resources trigger_resources;
 	struct acpi_whea_header *trigger_entry;
@@ -327,54 +414,60 @@ static int __einj_error_trigger(u64 trigger_paddr, u32 type,
 	u32 table_size;
 	int rc = -EIO;
 	struct acpi_generic_address *trigger_param_region = NULL;
+	struct acpi_einj_trigger __iomem *p = NULL;
 
-	r = request_mem_region(trigger_paddr, sizeof(*trigger_tab),
+	r = request_mem_region(trigger_paddr, sizeof(trigger_tab),
 			       "APEI EINJ Trigger Table");
 	if (!r) {
 		pr_err("Can not request [mem %#010llx-%#010llx] for Trigger table\n",
 		       (unsigned long long)trigger_paddr,
 		       (unsigned long long)trigger_paddr +
-			    sizeof(*trigger_tab) - 1);
+			    sizeof(trigger_tab) - 1);
 		goto out;
 	}
-	trigger_tab = ioremap_cache(trigger_paddr, sizeof(*trigger_tab));
-	if (!trigger_tab) {
+	p = ioremap_cache(trigger_paddr, sizeof(*p));
+	if (!p) {
 		pr_err("Failed to map trigger table!\n");
 		goto out_rel_header;
 	}
-	rc = einj_check_trigger_header(trigger_tab);
+	memcpy_fromio(&trigger_tab, p, sizeof(trigger_tab));
+	rc = einj_check_trigger_header(&trigger_tab);
 	if (rc) {
 		pr_warn(FW_BUG "Invalid trigger error action table.\n");
 		goto out_rel_header;
 	}
 
 	/* No action structures in the TRIGGER_ERROR table, nothing to do */
-	if (!trigger_tab->entry_count)
+	if (!trigger_tab.entry_count)
 		goto out_rel_header;
 
 	rc = -EIO;
-	table_size = trigger_tab->table_size;
-	r = request_mem_region(trigger_paddr + sizeof(*trigger_tab),
-			       table_size - sizeof(*trigger_tab),
+	table_size = trigger_tab.table_size;
+	full_trigger_tab = kmalloc(table_size, GFP_KERNEL);
+	if (!full_trigger_tab)
+		goto out_rel_header;
+	r = request_mem_region(trigger_paddr + sizeof(trigger_tab),
+			       table_size - sizeof(trigger_tab),
 			       "APEI EINJ Trigger Table");
 	if (!r) {
 		pr_err("Can not request [mem %#010llx-%#010llx] for Trigger Table Entry\n",
-		       (unsigned long long)trigger_paddr + sizeof(*trigger_tab),
+		       (unsigned long long)trigger_paddr + sizeof(trigger_tab),
 		       (unsigned long long)trigger_paddr + table_size - 1);
-		goto out_rel_header;
+		goto out_free_trigger_tab;
 	}
-	iounmap(trigger_tab);
-	trigger_tab = ioremap_cache(trigger_paddr, table_size);
-	if (!trigger_tab) {
+	iounmap(p);
+	p = ioremap_cache(trigger_paddr, table_size);
+	if (!p) {
 		pr_err("Failed to map trigger table!\n");
 		goto out_rel_entry;
 	}
+	memcpy_fromio(full_trigger_tab, p, table_size);
 	trigger_entry = (struct acpi_whea_header *)
-		((char *)trigger_tab + sizeof(struct acpi_einj_trigger));
+		((char *)full_trigger_tab + sizeof(struct acpi_einj_trigger));
 	apei_resources_init(&trigger_resources);
 	apei_exec_ctx_init(&trigger_ctx, einj_ins_type,
 			   ARRAY_SIZE(einj_ins_type),
-			   trigger_entry, trigger_tab->entry_count);
+			   trigger_entry, trigger_tab.entry_count);
 	rc = apei_exec_collect_resources(&trigger_ctx, &trigger_resources);
 	if (rc)
 		goto out_fini;
@@ -392,7 +485,7 @@ static int __einj_error_trigger(u64 trigger_paddr, u32 type,
 
 		apei_resources_init(&addr_resources);
 		trigger_param_region = einj_get_trigger_parameter_region(
-			trigger_tab, param1, param2);
+			full_trigger_tab, param1, param2);
 		if (trigger_param_region) {
 			rc = apei_resources_add(&addr_resources,
 				trigger_param_region->address,
@@ -421,23 +514,34 @@ out_release:
 out_fini:
 	apei_resources_fini(&trigger_resources);
 out_rel_entry:
-	release_mem_region(trigger_paddr + sizeof(*trigger_tab),
-			   table_size - sizeof(*trigger_tab));
+	release_mem_region(trigger_paddr + sizeof(trigger_tab),
+			   table_size - sizeof(trigger_tab));
+out_free_trigger_tab:
+	kfree(full_trigger_tab);
 out_rel_header:
-	release_mem_region(trigger_paddr, sizeof(*trigger_tab));
+	release_mem_region(trigger_paddr, sizeof(trigger_tab));
 out:
-	if (trigger_tab)
-		iounmap(trigger_tab);
+	if (p)
+		iounmap(p);
 
 	return rc;
 }
 
+static bool is_end_of_list(u8 *val)
+{
+	for (int i = 0; i < COMPONENT_LEN; ++i) {
+		if (val[i] != 0xFF)
+			return false;
+	}
+	return true;
+}
 static int __einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 			       u64 param3, u64 param4)
 {
 	struct apei_exec_context ctx;
+	u32 param_size = is_v2 ? v66param_size : v5param_size;
 	u64 val, trigger_paddr, timeout = FIRMWARE_TIMEOUT;
-	int rc;
+	int i, rc;
 
 	einj_exec_ctx_init(&ctx);
 
@@ -446,8 +550,13 @@ static int __einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 		return rc;
 	apei_exec_ctx_set_input(&ctx, type);
 	if (acpi5) {
-		struct set_error_type_with_address *v5param = einj_param;
+		struct set_error_type_with_address *v5param;
 
+		v5param = kmalloc(param_size, GFP_KERNEL);
+		if (!v5param)
+			return -ENOMEM;
+
+		memcpy_fromio(v5param, einj_param, param_size);
 		v5param->type = type;
 		if (type & ACPI5_VENDOR_BIT) {
 			switch (vendor_flags) {
@@ -467,8 +576,21 @@ static int __einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 			v5param->flags = flags;
 			v5param->memory_address = param1;
 			v5param->memory_address_range = param2;
-			v5param->apicid = param3;
-			v5param->pcie_sbdf = param4;
+
+			if (is_v2) {
+				for (i = 0; i < max_nr_components; i++) {
+					if (is_end_of_list(syndrome_data[i].comp_id.acpi_id))
+						break;
+					v5param->einjv2_struct.component_arr[i].comp_id =
+						syndrome_data[i].comp_id;
+					v5param->einjv2_struct.component_arr[i].comp_synd =
+						syndrome_data[i].comp_synd;
+				}
+				v5param->einjv2_struct.component_arr_count = i;
+			} else {
+				v5param->apicid = param3;
+				v5param->pcie_sbdf = param4;
+			}
 		} else {
 			switch (type) {
 			case ACPI_EINJ_PROCESSOR_CORRECTABLE:
@@ -492,15 +614,19 @@ static int __einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 				break;
 			}
 		}
+		memcpy_toio(einj_param, v5param, param_size);
+		kfree(v5param);
 	} else {
 		rc = apei_exec_run(&ctx, ACPI_EINJ_SET_ERROR_TYPE);
 		if (rc)
 			return rc;
 		if (einj_param) {
-			struct einj_parameter *v4param = einj_param;
+			struct einj_parameter v4param;
 
-			v4param->param1 = param1;
-			v4param->param2 = param2;
+			memcpy_fromio(&v4param, einj_param, sizeof(v4param));
+			v4param.param1 = param1;
+			v4param.param2 = param2;
+			memcpy_toio(einj_param, &v4param, sizeof(v4param));
 		}
 	}
 	rc = apei_exec_run(&ctx, ACPI_EINJ_EXECUTE_OPERATION);
@@ -543,6 +669,43 @@ static int __einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 	return rc;
 }
 
+/* Allow almost all types of address except MMIO. */
+static bool is_allowed_range(u64 base_addr, u64 size)
+{
+	int i;
+	/*
+	 * MMIO region is usually claimed with IORESOURCE_MEM + IORES_DESC_NONE.
+	 * However, IORES_DESC_NONE is treated like a wildcard when we check if
+	 * region intersects with known resource. So do an allow list check for
+	 * IORES_DESCs that definitely or most likely not MMIO.
+	 */
+	int non_mmio_desc[] = {
+		IORES_DESC_CRASH_KERNEL,
+		IORES_DESC_ACPI_TABLES,
+		IORES_DESC_ACPI_NV_STORAGE,
+		IORES_DESC_PERSISTENT_MEMORY,
+		IORES_DESC_PERSISTENT_MEMORY_LEGACY,
+		/* Treat IORES_DESC_DEVICE_PRIVATE_MEMORY as MMIO. */
+		IORES_DESC_RESERVED,
+		IORES_DESC_SOFT_RESERVED,
+	};
+
+	if (region_intersects(base_addr, size, IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE)
+			      == REGION_INTERSECTS)
+		return true;
+
+	for (i = 0; i < ARRAY_SIZE(non_mmio_desc); ++i) {
+		if (region_intersects(base_addr, size, IORESOURCE_MEM, non_mmio_desc[i])
+				      == REGION_INTERSECTS)
+			return true;
+	}
+
+	if (arch_is_platform_page(base_addr))
+		return true;
+
+	return false;
+}
+
 /* Inject the specified hardware error */
 int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2, u64 param3,
 		      u64 param4)
@@ -551,10 +714,15 @@ int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2, u64 param3,
 	u64 base_addr, size;
 
 	/* If user manually set "flags", make sure it is legal */
-	if (flags && (flags &
-		~(SETWA_FLAGS_APICID|SETWA_FLAGS_MEM|SETWA_FLAGS_PCIE_SBDF)))
+	if (flags && (flags & ~(SETWA_FLAGS_APICID | SETWA_FLAGS_MEM |
+		      SETWA_FLAGS_PCIE_SBDF | SETWA_FLAGS_EINJV2)))
 		return -EINVAL;
 
+	/* check if type is a valid EINJv2 error type */
+	if (is_v2) {
+		if (!(type & available_error_type_v2))
+			return -EINVAL;
+	}
 	/*
 	 * We need extra sanity checks for memory errors.
 	 * Other types leap directly to injection.
@@ -584,19 +752,15 @@ int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2, u64 param3,
 	 * Disallow crazy address masks that give BIOS leeway to pick
 	 * injection address almost anywhere. Insist on page or
 	 * better granularity and that target address is normal RAM or
-	 * NVDIMM.
+	 * as long as is not MMIO.
 	 */
 	base_addr = param1 & param2;
 	size = ~param2 + 1;
 
-	if (((param2 & PAGE_MASK) != PAGE_MASK) ||
-	    ((region_intersects(base_addr, size, IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE)
-				!= REGION_INTERSECTS) &&
-	     (region_intersects(base_addr, size, IORESOURCE_MEM, IORES_DESC_PERSISTENT_MEMORY)
-				!= REGION_INTERSECTS) &&
-	     (region_intersects(base_addr, size, IORESOURCE_MEM, IORES_DESC_SOFT_RESERVED)
-				!= REGION_INTERSECTS) &&
-	     !arch_is_platform_page(base_addr)))
+	if ((param2 & PAGE_MASK) != PAGE_MASK)
+		return -EINVAL;
+
+	if (!is_allowed_range(base_addr, size))
 		return -EINVAL;
 
 	if (is_zero_pfn(base_addr >> PAGE_SHIFT))
@@ -632,6 +796,8 @@ static u64 error_param2;
 static u64 error_param3;
 static u64 error_param4;
 static struct dentry *einj_debug_dir;
+static char einj_buf[32];
+static bool einj_v2_enabled;
 static struct { u32 mask; const char *str; } const einj_error_type_string[] = {
 	{ BIT(0), "Processor Correctable" },
 	{ BIT(1), "Processor Uncorrectable non-fatal" },
@@ -648,6 +814,12 @@ static struct { u32 mask; const char *str; } const einj_error_type_string[] = {
 	{ BIT(31), "Vendor Defined Error Types" },
 };
 
+static struct { u32 mask; const char *str; } const einjv2_error_type_string[] = {
+	{ BIT(0), "EINJV2 Processor Error" },
+	{ BIT(1), "EINJV2 Memory Error" },
+	{ BIT(2), "EINJV2 PCI Express Error" },
+};
+
 static int available_error_type_show(struct seq_file *m, void *v)
 {
 
@@ -655,17 +827,22 @@ static int available_error_type_show(struct seq_file *m, void *v)
 		if (available_error_type & einj_error_type_string[pos].mask)
 			seq_printf(m, "0x%08x\t%s\n", einj_error_type_string[pos].mask,
 				   einj_error_type_string[pos].str);
-
+	if ((available_error_type & ACPI65_EINJV2_SUPP) && einj_v2_enabled) {
+		for (int pos = 0; pos < ARRAY_SIZE(einjv2_error_type_string); pos++) {
+			if (available_error_type_v2 & einjv2_error_type_string[pos].mask)
+				seq_printf(m, "V2_0x%08x\t%s\n", einjv2_error_type_string[pos].mask,
+					   einjv2_error_type_string[pos].str);
+		}
+	}
 	return 0;
 }
 
 DEFINE_SHOW_ATTRIBUTE(available_error_type);
 
-static int error_type_get(void *data, u64 *val)
+static ssize_t error_type_get(struct file *file, char __user *buf,
+				size_t count, loff_t *ppos)
 {
-	*val = error_type;
-
-	return 0;
+	return simple_read_from_buffer(buf, count, ppos, einj_buf, strlen(einj_buf));
 }
 
 bool einj_is_cxl_error_type(u64 type)
@@ -692,15 +869,35 @@ int einj_validate_error_type(u64 type)
 	if (tval & (tval - 1))
 		return -EINVAL;
 	if (!vendor)
-		if (!(type & available_error_type))
+		if (!(type & (available_error_type | available_error_type_v2)))
 			return -EINVAL;
 
 	return 0;
 }
 
-static int error_type_set(void *data, u64 val)
+static ssize_t error_type_set(struct file *file, const char __user *buf,
+				size_t count, loff_t *ppos)
 {
 	int rc;
+	u64 val;
+
+	/* Leave the last character for the NUL terminator */
+	if (count > sizeof(einj_buf) - 1)
+		return -EINVAL;
+
+	memset(einj_buf, 0, sizeof(einj_buf));
+	if (copy_from_user(einj_buf, buf, count))
+		return -EFAULT;
+
+	if (strncmp(einj_buf, "V2_", 3) == 0) {
+		if (!sscanf(einj_buf, "V2_%llx", &val))
+			return -EINVAL;
+		is_v2 = true;
+	} else {
+		if (!sscanf(einj_buf, "%llx", &val))
+			return -EINVAL;
+		is_v2 = false;
+	}
 
 	rc = einj_validate_error_type(val);
 	if (rc)
@@ -708,17 +905,24 @@ static int error_type_set(void *data, u64 val)
 
 	error_type = val;
 
-	return 0;
+	return count;
 }
 
-DEFINE_DEBUGFS_ATTRIBUTE(error_type_fops, error_type_get, error_type_set,
-			 "0x%llx\n");
+static const struct file_operations error_type_fops = {
+	.read		= error_type_get,
+	.write		= error_type_set,
+};
 
 static int error_inject_set(void *data, u64 val)
 {
 	if (!error_type)
 		return -EINVAL;
 
+	if (is_v2)
+		error_flags |= SETWA_FLAGS_EINJV2;
+	else
+		error_flags &= ~SETWA_FLAGS_EINJV2;
+
 	return einj_error_inject(error_type, error_flags, error_param1, error_param2,
 		error_param3, error_param4);
 }
@@ -741,6 +945,98 @@ static int einj_check_table(struct acpi_table_einj *einj_tab)
 	return 0;
 }
 
+static ssize_t u128_read(struct file *f, char __user *buf, size_t count, loff_t *off)
+{
+	char output[2 * COMPONENT_LEN + 1];
+	u8 *data = f->f_inode->i_private;
+	int i;
+
+	if (*off >= sizeof(output))
+		return 0;
+
+	for (i = 0; i < COMPONENT_LEN; i++)
+		sprintf(output + 2 * i, "%.02x", data[COMPONENT_LEN - i - 1]);
+	output[2 * COMPONENT_LEN] = '\n';
+
+	return simple_read_from_buffer(buf, count, off, output, sizeof(output));
+}
+
+static ssize_t u128_write(struct file *f, const char __user *buf, size_t count, loff_t *off)
+{
+	char input[2 + 2 * COMPONENT_LEN + 2];
+	u8 *save = f->f_inode->i_private;
+	u8 tmp[COMPONENT_LEN];
+	char byte[3] = {};
+	char *s, *e;
+	ssize_t c;
+	long val;
+	int i;
+
+	/* Require that user supply whole input line in one write(2) syscall */
+	if (*off)
+		return -EINVAL;
+
+	c = simple_write_to_buffer(input, sizeof(input), off, buf, count);
+	if (c < 0)
+		return c;
+
+	if (c < 1 || input[c - 1] != '\n')
+		return -EINVAL;
+
+	/* Empty line means invalidate this entry */
+	if (c == 1) {
+		memset(save, 0xff, COMPONENT_LEN);
+		return c;
+	}
+
+	if (input[0] == '0' && (input[1] == 'x' || input[1] == 'X'))
+		s = input + 2;
+	else
+		s = input;
+	e = input + c - 1;
+
+	for (i = 0; i < COMPONENT_LEN; i++) {
+		byte[1] = *--e;
+		byte[0] = e > s ? *--e : '0';
+		if (kstrtol(byte, 16, &val))
+			return -EINVAL;
+		tmp[i] = val;
+		if (e <= s)
+			break;
+	}
+	while (++i < COMPONENT_LEN)
+		tmp[i] = 0;
+
+	memcpy(save, tmp, COMPONENT_LEN);
+
+	return c;
+}
+
+static const struct file_operations u128_fops = {
+	.read	= u128_read,
+	.write	= u128_write,
+};
+
+static bool setup_einjv2_component_files(void)
+{
+	char name[32];
+
+	syndrome_data = kcalloc(max_nr_components, sizeof(syndrome_data[0]), GFP_KERNEL);
+	if (!syndrome_data)
+		return false;
+
+	for (int i = 0; i < max_nr_components; i++) {
+		sprintf(name, "component_id%d", i);
+		debugfs_create_file(name, 0600, einj_debug_dir,
+				    &syndrome_data[i].comp_id, &u128_fops);
+		sprintf(name, "component_syndrome%d", i);
+		debugfs_create_file(name, 0600, einj_debug_dir,
+				    &syndrome_data[i].comp_synd, &u128_fops);
+	}
+
+	return true;
+}
+
 static int __init einj_probe(struct faux_device *fdev)
 {
 	int rc;
@@ -764,7 +1060,7 @@ static int __init einj_probe(struct faux_device *fdev)
 		goto err_put_table;
 	}
 
-	rc = einj_get_available_error_type(&available_error_type);
+	rc = einj_get_available_error_types(&available_error_type, &available_error_type_v2);
 	if (rc)
 		goto err_put_table;
 
@@ -812,6 +1108,8 @@ static int __init einj_probe(struct faux_device *fdev)
 				   &error_param4);
 		debugfs_create_x32("notrigger", S_IRUSR | S_IWUSR,
 				   einj_debug_dir, &notrigger);
+		if (available_error_type & ACPI65_EINJV2_SUPP)
+			einj_v2_enabled = setup_einjv2_component_files();
 	}
 
 	if (vendor_dev[0]) {
@@ -842,14 +1140,19 @@ err_put_table:
 	return rc;
 }
 
-static void __exit einj_remove(struct faux_device *fdev)
+static void einj_remove(struct faux_device *fdev)
 {
 	struct apei_exec_context ctx;
 
 	if (einj_param) {
-		acpi_size size = (acpi5) ?
-			sizeof(struct set_error_type_with_address) :
-			sizeof(struct einj_parameter);
+		acpi_size size;
+
+		if (v66param_size)
+			size = v66param_size;
+		else if (acpi5)
+			size = v5param_size;
+		else
+			size = sizeof(struct einj_parameter);
 
 		acpi_os_unmap_iomem(einj_param, size);
 		if (vendor_errors.size)
@@ -860,19 +1163,14 @@ static void __exit einj_remove(struct faux_device *fdev)
 	apei_resources_release(&einj_resources);
 	apei_resources_fini(&einj_resources);
 	debugfs_remove_recursive(einj_debug_dir);
+	kfree(syndrome_data);
 	acpi_put_table((struct acpi_table_header *)einj_tab);
 }
 
 static struct faux_device *einj_dev;
-/*
- * einj_remove() lives in .exit.text. For drivers registered via
- * platform_driver_probe() this is ok because they cannot get unbound at
- * runtime. So mark the driver struct with __refdata to prevent modpost
- * triggering a section mismatch warning.
- */
-static struct faux_device_ops einj_device_ops __refdata = {
+static struct faux_device_ops einj_device_ops = {
 	.probe = einj_probe,
-	.remove = __exit_p(einj_remove),
+	.remove = einj_remove,
 };
 
 static int __init einj_init(void)
@@ -883,19 +1181,16 @@ static int __init einj_init(void)
 	}
 
 	einj_dev = faux_device_create("acpi-einj", NULL, &einj_device_ops);
-	if (!einj_dev)
-		return -ENODEV;
 
-	einj_initialized = true;
+	if (einj_dev)
+		einj_initialized = true;
 
 	return 0;
 }
 
 static void __exit einj_exit(void)
 {
-	if (einj_initialized)
-		faux_device_destroy(einj_dev);
-
+	faux_device_destroy(einj_dev);
 }
 
 module_init(einj_init);
diff --git a/drivers/acpi/apei/einj-cxl.c b/drivers/acpi/apei/einj-cxl.c
index 78da9ae543a2..e70a416ec925 100644
--- a/drivers/acpi/apei/einj-cxl.c
+++ b/drivers/acpi/apei/einj-cxl.c
@@ -30,7 +30,7 @@ int einj_cxl_available_error_type_show(struct seq_file *m, void *v)
 	int cxl_err, rc;
 	u32 available_error_type = 0;
 
-	rc = einj_get_available_error_type(&available_error_type);
+	rc = einj_get_available_error_type(&available_error_type, ACPI_EINJ_GET_ERROR_TYPE);
 	if (rc)
 		return rc;
 
diff --git a/drivers/acpi/apei/erst-dbg.c b/drivers/acpi/apei/erst-dbg.c
index 246076341e8c..ff0e8bf8e97a 100644
--- a/drivers/acpi/apei/erst-dbg.c
+++ b/drivers/acpi/apei/erst-dbg.c
@@ -60,9 +60,8 @@ static long erst_dbg_ioctl(struct file *f, unsigned int cmd, unsigned long arg)
 
 	switch (cmd) {
 	case APEI_ERST_CLEAR_RECORD:
-		rc = copy_from_user(&record_id, (void __user *)arg,
-				    sizeof(record_id));
-		if (rc)
+		if (copy_from_user(&record_id, (void __user *)arg,
+				   sizeof(record_id)))
 			return -EFAULT;
 		return erst_clear(record_id);
 	case APEI_ERST_GET_RECORD_COUNT:
@@ -175,8 +174,7 @@ static ssize_t erst_dbg_write(struct file *filp, const char __user *ubuf,
 		erst_dbg_buf = p;
 		erst_dbg_buf_len = usize;
 	}
-	rc = copy_from_user(erst_dbg_buf, ubuf, usize);
-	if (rc) {
+	if (copy_from_user(erst_dbg_buf, ubuf, usize)) {
 		rc = -EFAULT;
 		goto out;
 	}
diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index f0584ccad451..0dc767392a6c 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -22,6 +22,7 @@
 #include <linux/moduleparam.h>
 #include <linux/init.h>
 #include <linux/acpi.h>
+#include <linux/bitfield.h>
 #include <linux/io.h>
 #include <linux/interrupt.h>
 #include <linux/timer.h>
@@ -43,6 +44,7 @@
 #include <linux/uuid.h>
 #include <linux/ras.h>
 #include <linux/task_work.h>
+#include <linux/vmcore_info.h>
 
 #include <acpi/actbl1.h>
 #include <acpi/ghes.h>
@@ -464,39 +466,58 @@ static void ghes_clear_estatus(struct ghes *ghes,
 		ghes_ack_error(ghes->generic_v2);
 }
 
-/*
- * Called as task_work before returning to user-space.
- * Ensure any queued work has been done before we return to the context that
- * triggered the notification.
+/**
+ * struct ghes_task_work - for synchronous RAS event
+ *
+ * @twork:                callback_head for task work
+ * @pfn:                  page frame number of corrupted page
+ * @flags:                work control flags
+ *
+ * Structure to pass task work to be handled before
+ * returning to user-space via task_work_add().
  */
-static void ghes_kick_task_work(struct callback_head *head)
+struct ghes_task_work {
+	struct callback_head twork;
+	u64 pfn;
+	int flags;
+};
+
+static void memory_failure_cb(struct callback_head *twork)
 {
-	struct acpi_hest_generic_status *estatus;
-	struct ghes_estatus_node *estatus_node;
-	u32 node_len;
+	struct ghes_task_work *twcb = container_of(twork, struct ghes_task_work, twork);
+	int ret;
 
-	estatus_node = container_of(head, struct ghes_estatus_node, task_work);
-	if (IS_ENABLED(CONFIG_ACPI_APEI_MEMORY_FAILURE))
-		memory_failure_queue_kick(estatus_node->task_work_cpu);
+	ret = memory_failure(twcb->pfn, twcb->flags);
+	gen_pool_free(ghes_estatus_pool, (unsigned long)twcb, sizeof(*twcb));
 
-	estatus = GHES_ESTATUS_FROM_NODE(estatus_node);
-	node_len = GHES_ESTATUS_NODE_LEN(cper_estatus_len(estatus));
-	gen_pool_free(ghes_estatus_pool, (unsigned long)estatus_node, node_len);
+	if (!ret || ret == -EHWPOISON || ret == -EOPNOTSUPP)
+		return;
+
+	pr_err("%#llx: Sending SIGBUS to %s:%d due to hardware memory corruption\n",
+			twcb->pfn, current->comm, task_pid_nr(current));
+	force_sig(SIGBUS);
 }
 
 static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 {
+	struct ghes_task_work *twcb;
 	unsigned long pfn;
 
 	if (!IS_ENABLED(CONFIG_ACPI_APEI_MEMORY_FAILURE))
 		return false;
 
 	pfn = PHYS_PFN(physical_addr);
-	if (!pfn_valid(pfn) && !arch_is_platform_page(physical_addr)) {
-		pr_warn_ratelimited(FW_WARN GHES_PFX
-		"Invalid address in generic error data: %#llx\n",
-		physical_addr);
-		return false;
+
+	if (flags == MF_ACTION_REQUIRED && current->mm) {
+		twcb = (void *)gen_pool_alloc(ghes_estatus_pool, sizeof(*twcb));
+		if (!twcb)
+			return false;
+
+		twcb->pfn = pfn;
+		twcb->flags = flags;
+		init_task_work(&twcb->twork, memory_failure_cb);
+		task_work_add(current, &twcb->twork, TWA_RESUME);
+		return true;
 	}
 
 	memory_failure_queue(pfn, flags);
@@ -527,26 +548,25 @@ static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
 }
 
 static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata,
-				       int sev, bool sync)
+				     int sev, bool sync)
 {
 	struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
 	int flags = sync ? MF_ACTION_REQUIRED : 0;
+	char error_type[120];
 	bool queued = false;
 	int sec_sev, i;
 	char *p;
 
-	log_arm_hw_error(err);
-
 	sec_sev = ghes_severity(gdata->error_severity);
+	log_arm_hw_error(err, sec_sev);
 	if (sev != GHES_SEV_RECOVERABLE || sec_sev != GHES_SEV_RECOVERABLE)
 		return false;
 
 	p = (char *)(err + 1);
 	for (i = 0; i < err->err_info_num; i++) {
 		struct cper_arm_err_info *err_info = (struct cper_arm_err_info *)p;
-		bool is_cache = (err_info->type == CPER_ARM_CACHE_ERROR);
+		bool is_cache = err_info->type & CPER_ARM_CACHE_ERROR;
 		bool has_pa = (err_info->validation_bits & CPER_ARM_INFO_VALID_PHYSICAL_ADDR);
-		const char *error_type = "unknown error";
 
 		/*
 		 * The field (err_info->error_info & BIT(26)) is fixed to set to
@@ -560,12 +580,15 @@ static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata,
 			continue;
 		}
 
-		if (err_info->type < ARRAY_SIZE(cper_proc_error_type_strs))
-			error_type = cper_proc_error_type_strs[err_info->type];
+		cper_bits_to_str(error_type, sizeof(error_type),
+				 FIELD_GET(CPER_ARM_ERR_TYPE_MASK, err_info->type),
+				 cper_proc_error_type_strs,
+				 ARRAY_SIZE(cper_proc_error_type_strs));
 
 		pr_warn_ratelimited(FW_WARN GHES_PFX
-				    "Unhandled processor error type: %s\n",
-				    error_type);
+				    "Unhandled processor error type 0x%02x: %s%s\n",
+				    err_info->type, error_type,
+				    (err_info->type & ~CPER_ARM_ERR_TYPE_MASK) ? " with reserved bit(s)" : "");
 		p += err_info->length;
 	}
 
@@ -842,7 +865,41 @@ int cxl_cper_kfifo_get(struct cxl_cper_work_data *wd)
 }
 EXPORT_SYMBOL_NS_GPL(cxl_cper_kfifo_get, "CXL");
 
-static bool ghes_do_proc(struct ghes *ghes,
+static void ghes_log_hwerr(int sev, guid_t *sec_type)
+{
+	if (sev != CPER_SEV_RECOVERABLE)
+		return;
+
+	if (guid_equal(sec_type, &CPER_SEC_PROC_ARM) ||
+	    guid_equal(sec_type, &CPER_SEC_PROC_GENERIC) ||
+	    guid_equal(sec_type, &CPER_SEC_PROC_IA)) {
+		hwerr_log_error_type(HWERR_RECOV_CPU);
+		return;
+	}
+
+	if (guid_equal(sec_type, &CPER_SEC_CXL_PROT_ERR) ||
+	    guid_equal(sec_type, &CPER_SEC_CXL_GEN_MEDIA_GUID) ||
+	    guid_equal(sec_type, &CPER_SEC_CXL_DRAM_GUID) ||
+	    guid_equal(sec_type, &CPER_SEC_CXL_MEM_MODULE_GUID)) {
+		hwerr_log_error_type(HWERR_RECOV_CXL);
+		return;
+	}
+
+	if (guid_equal(sec_type, &CPER_SEC_PCIE) ||
+	    guid_equal(sec_type, &CPER_SEC_PCI_X_BUS)) {
+		hwerr_log_error_type(HWERR_RECOV_PCI);
+		return;
+	}
+
+	if (guid_equal(sec_type, &CPER_SEC_PLATFORM_MEM)) {
+		hwerr_log_error_type(HWERR_RECOV_MEMORY);
+		return;
+	}
+
+	hwerr_log_error_type(HWERR_RECOV_OTHERS);
+}
+
+static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
 	int sev, sec_sev;
@@ -863,6 +920,7 @@ static bool ghes_do_proc(struct ghes *ghes,
 		if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
 			fru_text = gdata->fru_text;
 
+		ghes_log_hwerr(sev, sec_type);
 		if (guid_equal(sec_type, &CPER_SEC_PLATFORM_MEM)) {
 			struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
@@ -870,11 +928,9 @@ static bool ghes_do_proc(struct ghes *ghes,
 
 			arch_apei_report_mem_error(sev, mem_err);
 			queued = ghes_handle_memory_failure(gdata, sev, sync);
-		}
-		else if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
+		} else if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
 			ghes_handle_aer(gdata);
-		}
-		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
+		} else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
 			queued = ghes_handle_arm_hw_error(gdata, sev, sync);
 		} else if (guid_equal(sec_type, &CPER_SEC_CXL_PROT_ERR)) {
 			struct cxl_cper_sec_prot_err *prot_err = acpi_hest_get_payload(gdata);
@@ -902,7 +958,16 @@ static bool ghes_do_proc(struct ghes *ghes,
 		}
 	}
 
-	return queued;
+	/*
+	 * If no memory failure work is queued for abnormal synchronous
+	 * errors, do a force kill.
+	 */
+	if (sync && !queued) {
+		dev_err(ghes->dev,
+			HW_ERR GHES_PFX "%s:%d: synchronous unrecoverable error (SIGBUS)\n",
+			current->comm, task_pid_nr(current));
+		force_sig(SIGBUS);
+	}
 }
 
 static void __ghes_print_estatus(const char *pfx,
@@ -1088,6 +1153,8 @@ static void __ghes_panic(struct ghes *ghes,
 
 	__ghes_print_estatus(KERN_EMERG, ghes->generic, estatus);
 
+	add_taint(TAINT_MACHINE_CHECK, LOCKDEP_STILL_OK);
+
 	ghes_clear_estatus(ghes, estatus, buf_paddr, fixmap_idx);
 
 	if (!panic_timeout)
@@ -1171,12 +1238,10 @@ static int ghes_notify_hed(struct notifier_block *this, unsigned long event,
 	int ret = NOTIFY_DONE;
 
 	spin_lock_irqsave(&ghes_notify_lock_irq, flags);
-	rcu_read_lock();
 	list_for_each_entry_rcu(ghes, &ghes_hed, list) {
 		if (!ghes_proc(ghes))
 			ret = NOTIFY_OK;
 	}
-	rcu_read_unlock();
 	spin_unlock_irqrestore(&ghes_notify_lock_irq, flags);
 
 	return ret;
@@ -1206,9 +1271,7 @@ static void ghes_proc_in_irq(struct irq_work *irq_work)
 	struct ghes_estatus_node *estatus_node;
 	struct acpi_hest_generic *generic;
 	struct acpi_hest_generic_status *estatus;
-	bool task_work_pending;
 	u32 len, node_len;
-	int ret;
 
 	llnode = llist_del_all(&ghes_estatus_llist);
 	/*
@@ -1223,25 +1286,16 @@ static void ghes_proc_in_irq(struct irq_work *irq_work)
 		estatus = GHES_ESTATUS_FROM_NODE(estatus_node);
 		len = cper_estatus_len(estatus);
 		node_len = GHES_ESTATUS_NODE_LEN(len);
-		task_work_pending = ghes_do_proc(estatus_node->ghes, estatus);
+
+		ghes_do_proc(estatus_node->ghes, estatus);
+
 		if (!ghes_estatus_cached(estatus)) {
 			generic = estatus_node->generic;
 			if (ghes_print_estatus(NULL, generic, estatus))
 				ghes_estatus_cache_add(generic, estatus);
 		}
-
-		if (task_work_pending && current->mm) {
-			estatus_node->task_work.func = ghes_kick_task_work;
-			estatus_node->task_work_cpu = smp_processor_id();
-			ret = task_work_add(current, &estatus_node->task_work,
-					    TWA_RESUME);
-			if (ret)
-				estatus_node->task_work.func = NULL;
-		}
-
-		if (!estatus_node->task_work.func)
-			gen_pool_free(ghes_estatus_pool,
-				      (unsigned long)estatus_node, node_len);
+		gen_pool_free(ghes_estatus_pool, (unsigned long)estatus_node,
+			      node_len);
 
 		llnode = next;
 	}
@@ -1302,7 +1356,6 @@ static int ghes_in_nmi_queue_one_entry(struct ghes *ghes,
 
 	estatus_node->ghes = ghes;
 	estatus_node->generic = ghes->generic;
-	estatus_node->task_work.func = NULL;
 	estatus = GHES_ESTATUS_FROM_NODE(estatus_node);
 
 	if (__ghes_read_estatus(estatus, buf_paddr, fixmap_idx, len)) {
diff --git a/drivers/acpi/arm64/Kconfig b/drivers/acpi/arm64/Kconfig
index b3ed6212244c..f2fd79f22e7d 100644
--- a/drivers/acpi/arm64/Kconfig
+++ b/drivers/acpi/arm64/Kconfig
@@ -21,3 +21,6 @@ config ACPI_AGDI
 
 config ACPI_APMT
 	bool
+
+config ACPI_MPAM
+	bool
diff --git a/drivers/acpi/arm64/Makefile b/drivers/acpi/arm64/Makefile
index 05ecde9eaabe..9390b57cb564 100644
--- a/drivers/acpi/arm64/Makefile
+++ b/drivers/acpi/arm64/Makefile
@@ -4,6 +4,7 @@ obj-$(CONFIG_ACPI_APMT) 	+= apmt.o
 obj-$(CONFIG_ACPI_FFH)		+= ffh.o
 obj-$(CONFIG_ACPI_GTDT) 	+= gtdt.o
 obj-$(CONFIG_ACPI_IORT) 	+= iort.o
+obj-$(CONFIG_ACPI_MPAM) 	+= mpam.o
 obj-$(CONFIG_ACPI_PROCESSOR_IDLE) += cpuidle.o
 obj-$(CONFIG_ARM_AMBA)		+= amba.o
 obj-y				+= dma.o init.o
diff --git a/drivers/acpi/arm64/gtdt.c b/drivers/acpi/arm64/gtdt.c
index 70f8290b659d..ffc867bac2d6 100644
--- a/drivers/acpi/arm64/gtdt.c
+++ b/drivers/acpi/arm64/gtdt.c
@@ -303,40 +303,6 @@ error:
 	return -EINVAL;
 }
 
-/**
- * acpi_arch_timer_mem_init() - Get the info of all GT blocks in GTDT table.
- * @timer_mem:	The pointer to the array of struct arch_timer_mem for returning
- *		the result of parsing. The element number of this array should
- *		be platform_timer_count(the total number of platform timers).
- * @timer_count: It points to a integer variable which is used for storing the
- *		number of GT blocks we have parsed.
- *
- * Return: 0 if success, -EINVAL/-ENODEV if error.
- */
-int __init acpi_arch_timer_mem_init(struct arch_timer_mem *timer_mem,
-				    int *timer_count)
-{
-	int ret;
-	void *platform_timer;
-
-	*timer_count = 0;
-	for_each_platform_timer(platform_timer) {
-		if (is_timer_block(platform_timer)) {
-			ret = gtdt_parse_timer_block(platform_timer, timer_mem);
-			if (ret)
-				return ret;
-			timer_mem++;
-			(*timer_count)++;
-		}
-	}
-
-	if (*timer_count)
-		pr_info("found %d memory-mapped timer block(s).\n",
-			*timer_count);
-
-	return 0;
-}
-
 /*
  * Initialize a SBSA generic Watchdog platform device info from GTDT
  */
@@ -388,11 +354,11 @@ static int __init gtdt_import_sbsa_gwdt(struct acpi_gtdt_watchdog *wd,
 	return 0;
 }
 
-static int __init gtdt_sbsa_gwdt_init(void)
+static int __init gtdt_platform_timer_init(void)
 {
 	void *platform_timer;
 	struct acpi_table_header *table;
-	int ret, timer_count, gwdt_count = 0;
+	int ret, timer_count, gwdt_count = 0, mmio_timer_count = 0;
 
 	if (acpi_disabled)
 		return 0;
@@ -414,20 +380,41 @@ static int __init gtdt_sbsa_gwdt_init(void)
 		goto out_put_gtdt;
 
 	for_each_platform_timer(platform_timer) {
+		ret = 0;
+
 		if (is_non_secure_watchdog(platform_timer)) {
 			ret = gtdt_import_sbsa_gwdt(platform_timer, gwdt_count);
 			if (ret)
-				break;
+				continue;
 			gwdt_count++;
+		} else 	if (is_timer_block(platform_timer)) {
+			struct arch_timer_mem atm = {};
+			struct platform_device *pdev;
+
+			ret = gtdt_parse_timer_block(platform_timer, &atm);
+			if (ret)
+				continue;
+
+			pdev = platform_device_register_data(NULL, "gtdt-arm-mmio-timer",
+							     mmio_timer_count, &atm,
+							     sizeof(atm));
+			if (IS_ERR(pdev)) {
+				pr_err("Can't register timer %d\n", mmio_timer_count);
+				continue;
+			}
+
+			mmio_timer_count++;
 		}
 	}
 
 	if (gwdt_count)
 		pr_info("found %d SBSA generic Watchdog(s).\n", gwdt_count);
+	if (mmio_timer_count)
+		pr_info("found %d Generic MMIO timer(s).\n", mmio_timer_count);
 
 out_put_gtdt:
 	acpi_put_table(table);
 	return ret;
 }
 
-device_initcall(gtdt_sbsa_gwdt_init);
+device_initcall(gtdt_platform_timer_init);
diff --git a/drivers/acpi/arm64/iort.c b/drivers/acpi/arm64/iort.c
index 98759d6199d3..65f0f56ad753 100644
--- a/drivers/acpi/arm64/iort.c
+++ b/drivers/acpi/arm64/iort.c
@@ -937,8 +937,10 @@ static u32 *iort_rmr_alloc_sids(u32 *sids, u32 count, u32 id_start,
 
 	new_sids = krealloc_array(sids, count + new_count,
 				  sizeof(*new_sids), GFP_KERNEL);
-	if (!new_sids)
+	if (!new_sids) {
+		kfree(sids);
 		return NULL;
+	}
 
 	for (i = count; i < total_count; i++)
 		new_sids[i] = id_start++;
diff --git a/drivers/acpi/arm64/mpam.c b/drivers/acpi/arm64/mpam.c
new file mode 100644
index 000000000000..84963a20c3e7
--- /dev/null
+++ b/drivers/acpi/arm64/mpam.c
@@ -0,0 +1,411 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (C) 2025 Arm Ltd.
+
+/* Parse the MPAM ACPI table feeding the discovered nodes into the driver */
+
+#define pr_fmt(fmt) "ACPI MPAM: " fmt
+
+#include <linux/acpi.h>
+#include <linux/arm_mpam.h>
+#include <linux/bits.h>
+#include <linux/cpu.h>
+#include <linux/cpumask.h>
+#include <linux/platform_device.h>
+
+#include <acpi/processor.h>
+
+/*
+ * Flags for acpi_table_mpam_msc.*_interrupt_flags.
+ * See 2.1.1 Interrupt Flags, Table 5, of DEN0065B_MPAM_ACPI_3.0-bet.
+ */
+#define ACPI_MPAM_MSC_IRQ_MODE                              BIT(0)
+#define ACPI_MPAM_MSC_IRQ_TYPE_MASK                         GENMASK(2, 1)
+#define ACPI_MPAM_MSC_IRQ_TYPE_WIRED                        0
+#define ACPI_MPAM_MSC_IRQ_AFFINITY_TYPE_MASK                BIT(3)
+#define ACPI_MPAM_MSC_IRQ_AFFINITY_TYPE_PROCESSOR           0
+#define ACPI_MPAM_MSC_IRQ_AFFINITY_TYPE_PROCESSOR_CONTAINER 1
+#define ACPI_MPAM_MSC_IRQ_AFFINITY_VALID                    BIT(4)
+
+/*
+ * Encodings for the MSC node body interface type field.
+ * See 2.1 MPAM MSC node, Table 4 of DEN0065B_MPAM_ACPI_3.0-bet.
+ */
+#define ACPI_MPAM_MSC_IFACE_MMIO   0x00
+#define ACPI_MPAM_MSC_IFACE_PCC    0x0a
+
+static bool _is_ppi_partition(u32 flags)
+{
+	u32 aff_type, is_ppi;
+	bool ret;
+
+	is_ppi = FIELD_GET(ACPI_MPAM_MSC_IRQ_AFFINITY_VALID, flags);
+	if (!is_ppi)
+		return false;
+
+	aff_type = FIELD_GET(ACPI_MPAM_MSC_IRQ_AFFINITY_TYPE_MASK, flags);
+	ret = (aff_type == ACPI_MPAM_MSC_IRQ_AFFINITY_TYPE_PROCESSOR_CONTAINER);
+	if (ret)
+		pr_err_once("Partitioned interrupts not supported\n");
+
+	return ret;
+}
+
+static int acpi_mpam_register_irq(struct platform_device *pdev,
+				  u32 intid, u32 flags)
+{
+	int irq;
+	u32 int_type;
+	int trigger;
+
+	if (!intid)
+		return -EINVAL;
+
+	if (_is_ppi_partition(flags))
+		return -EINVAL;
+
+	trigger = FIELD_GET(ACPI_MPAM_MSC_IRQ_MODE, flags);
+	int_type = FIELD_GET(ACPI_MPAM_MSC_IRQ_TYPE_MASK, flags);
+	if (int_type != ACPI_MPAM_MSC_IRQ_TYPE_WIRED)
+		return -EINVAL;
+
+	irq = acpi_register_gsi(&pdev->dev, intid, trigger, ACPI_ACTIVE_HIGH);
+	if (irq < 0)
+		pr_err_once("Failed to register interrupt 0x%x with ACPI\n", intid);
+
+	return irq;
+}
+
+static void acpi_mpam_parse_irqs(struct platform_device *pdev,
+				 struct acpi_mpam_msc_node *tbl_msc,
+				 struct resource *res, int *res_idx)
+{
+	u32 flags, intid;
+	int irq;
+
+	intid = tbl_msc->overflow_interrupt;
+	flags = tbl_msc->overflow_interrupt_flags;
+	irq = acpi_mpam_register_irq(pdev, intid, flags);
+	if (irq > 0)
+		res[(*res_idx)++] = DEFINE_RES_IRQ_NAMED(irq, "overflow");
+
+	intid = tbl_msc->error_interrupt;
+	flags = tbl_msc->error_interrupt_flags;
+	irq = acpi_mpam_register_irq(pdev, intid, flags);
+	if (irq > 0)
+		res[(*res_idx)++] = DEFINE_RES_IRQ_NAMED(irq, "error");
+}
+
+static int acpi_mpam_parse_resource(struct mpam_msc *msc,
+				    struct acpi_mpam_resource_node *res)
+{
+	int level, nid;
+	u32 cache_id;
+
+	switch (res->locator_type) {
+	case ACPI_MPAM_LOCATION_TYPE_PROCESSOR_CACHE:
+		cache_id = res->locator.cache_locator.cache_reference;
+		level = find_acpi_cache_level_from_id(cache_id);
+		if (level <= 0) {
+			pr_err_once("Bad level (%d) for cache with id %u\n", level, cache_id);
+			return -EINVAL;
+		}
+		return mpam_ris_create(msc, res->ris_index, MPAM_CLASS_CACHE,
+				       level, cache_id);
+	case ACPI_MPAM_LOCATION_TYPE_MEMORY:
+		nid = pxm_to_node(res->locator.memory_locator.proximity_domain);
+		if (nid == NUMA_NO_NODE) {
+			pr_debug("Bad proximity domain %lld, using node 0 instead\n",
+				 res->locator.memory_locator.proximity_domain);
+			nid = 0;
+		}
+		return mpam_ris_create(msc, res->ris_index, MPAM_CLASS_MEMORY,
+				       MPAM_CLASS_ID_DEFAULT, nid);
+	default:
+		/* These get discovered later and are treated as unknown */
+		return 0;
+	}
+}
+
+int acpi_mpam_parse_resources(struct mpam_msc *msc,
+			      struct acpi_mpam_msc_node *tbl_msc)
+{
+	int i, err;
+	char *ptr, *table_end;
+	struct acpi_mpam_resource_node *resource;
+
+	table_end = (char *)tbl_msc + tbl_msc->length;
+	ptr = (char *)(tbl_msc + 1);
+	for (i = 0; i < tbl_msc->num_resource_nodes; i++) {
+		u64 max_deps, remaining_table;
+
+		if (ptr + sizeof(*resource) > table_end)
+			return -EINVAL;
+
+		resource = (struct acpi_mpam_resource_node *)ptr;
+
+		remaining_table = table_end - ptr;
+		max_deps = remaining_table / sizeof(struct acpi_mpam_func_deps);
+		if (resource->num_functional_deps > max_deps) {
+			pr_debug("MSC has impossible number of functional dependencies\n");
+			return -EINVAL;
+		}
+
+		err = acpi_mpam_parse_resource(msc, resource);
+		if (err)
+			return err;
+
+		ptr += sizeof(*resource);
+		ptr += resource->num_functional_deps * sizeof(struct acpi_mpam_func_deps);
+	}
+
+	return 0;
+}
+
+/*
+ * Creates the device power management link and returns true if the
+ * acpi id is valid and usable for cpu affinity.  This is the case
+ * when the linked device is a processor or a processor container.
+ */
+static bool __init parse_msc_pm_link(struct acpi_mpam_msc_node *tbl_msc,
+				     struct platform_device *pdev,
+				     u32 *acpi_id)
+{
+	char hid[sizeof(tbl_msc->hardware_id_linked_device) + 1] = { 0 };
+	bool acpi_id_valid = false;
+	struct acpi_device *buddy;
+	char uid[11];
+	int len;
+
+	memcpy(hid, &tbl_msc->hardware_id_linked_device,
+	       sizeof(tbl_msc->hardware_id_linked_device));
+
+	if (!strcmp(hid, ACPI_PROCESSOR_CONTAINER_HID)) {
+		*acpi_id = tbl_msc->instance_id_linked_device;
+		acpi_id_valid = true;
+	}
+
+	len = snprintf(uid, sizeof(uid), "%u",
+		       tbl_msc->instance_id_linked_device);
+	if (len >= sizeof(uid)) {
+		pr_debug("Failed to convert uid of device for power management.");
+		return acpi_id_valid;
+	}
+
+	buddy = acpi_dev_get_first_match_dev(hid, uid, -1);
+	if (buddy) {
+		device_link_add(&pdev->dev, &buddy->dev, DL_FLAG_STATELESS);
+		acpi_dev_put(buddy);
+	}
+
+	return acpi_id_valid;
+}
+
+static int decode_interface_type(struct acpi_mpam_msc_node *tbl_msc,
+				 enum mpam_msc_iface *iface)
+{
+	switch (tbl_msc->interface_type) {
+	case ACPI_MPAM_MSC_IFACE_MMIO:
+		*iface = MPAM_IFACE_MMIO;
+		return 0;
+	case ACPI_MPAM_MSC_IFACE_PCC:
+		*iface = MPAM_IFACE_PCC;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static struct platform_device * __init acpi_mpam_parse_msc(struct acpi_mpam_msc_node *tbl_msc)
+{
+	struct platform_device *pdev __free(platform_device_put) =
+		platform_device_alloc("mpam_msc", tbl_msc->identifier);
+	int next_res = 0, next_prop = 0, err;
+	/* pcc, nrdy, affinity and a sentinel */
+	struct property_entry props[4] = { 0 };
+	/* mmio, 2xirq, no sentinel. */
+	struct resource res[3] = { 0 };
+	struct acpi_device *companion;
+	enum mpam_msc_iface iface;
+	char uid[16];
+	u32 acpi_id;
+
+	if (!pdev)
+		return ERR_PTR(-ENOMEM);
+
+	/* Some power management is described in the namespace: */
+	err = snprintf(uid, sizeof(uid), "%u", tbl_msc->identifier);
+	if (err > 0 && err < sizeof(uid)) {
+		companion = acpi_dev_get_first_match_dev("ARMHAA5C", uid, -1);
+		if (companion) {
+			ACPI_COMPANION_SET(&pdev->dev, companion);
+			acpi_dev_put(companion);
+		} else {
+			pr_debug("MSC.%u: missing namespace entry\n", tbl_msc->identifier);
+		}
+	}
+
+	if (decode_interface_type(tbl_msc, &iface)) {
+		pr_debug("MSC.%u: unknown interface type\n", tbl_msc->identifier);
+		return ERR_PTR(-EINVAL);
+	}
+
+	if (iface == MPAM_IFACE_MMIO) {
+		res[next_res++] = DEFINE_RES_MEM_NAMED(tbl_msc->base_address,
+						       tbl_msc->mmio_size,
+						       "MPAM:MSC");
+	} else if (iface == MPAM_IFACE_PCC) {
+		props[next_prop++] = PROPERTY_ENTRY_U32("pcc-channel",
+							tbl_msc->base_address);
+	}
+
+	acpi_mpam_parse_irqs(pdev, tbl_msc, res, &next_res);
+
+	WARN_ON_ONCE(next_res > ARRAY_SIZE(res));
+	err = platform_device_add_resources(pdev, res, next_res);
+	if (err)
+		return ERR_PTR(err);
+
+	props[next_prop++] = PROPERTY_ENTRY_U32("arm,not-ready-us",
+						tbl_msc->max_nrdy_usec);
+
+	/*
+	 * The MSC's CPU affinity is described via its linked power
+	 * management device, but only if it points at a Processor or
+	 * Processor Container.
+	 */
+	if (parse_msc_pm_link(tbl_msc, pdev, &acpi_id))
+		props[next_prop++] = PROPERTY_ENTRY_U32("cpu_affinity", acpi_id);
+
+	WARN_ON_ONCE(next_prop > ARRAY_SIZE(props) - 1);
+	err = device_create_managed_software_node(&pdev->dev, props, NULL);
+	if (err)
+		return ERR_PTR(err);
+
+	/*
+	 * Stash the table entry for acpi_mpam_parse_resources() to discover
+	 * what this MSC controls.
+	 */
+	err = platform_device_add_data(pdev, tbl_msc, tbl_msc->length);
+	if (err)
+		return ERR_PTR(err);
+
+	err = platform_device_add(pdev);
+	if (err)
+		return ERR_PTR(err);
+
+	return_ptr(pdev);
+}
+
+static int __init acpi_mpam_parse(void)
+{
+	char *table_end, *table_offset;
+	struct acpi_mpam_msc_node *tbl_msc;
+	struct platform_device *pdev;
+
+	if (acpi_disabled || !system_supports_mpam())
+		return 0;
+
+	struct acpi_table_header *table __free(acpi_put_table) =
+		acpi_get_table_pointer(ACPI_SIG_MPAM, 0);
+
+	if (IS_ERR(table))
+		return 0;
+
+	if (table->revision < 1) {
+		pr_debug("MPAM ACPI table revision %d not supported\n", table->revision);
+		return 0;
+	}
+
+	table_offset = (char *)(table + 1);
+	table_end = (char *)table + table->length;
+
+	while (table_offset < table_end) {
+		tbl_msc = (struct acpi_mpam_msc_node *)table_offset;
+		if (table_offset + sizeof(*tbl_msc) > table_end ||
+		    table_offset + tbl_msc->length > table_end) {
+			pr_err("MSC entry overlaps end of ACPI table\n");
+			return -EINVAL;
+		}
+		table_offset += tbl_msc->length;
+
+		/*
+		 * If any of the reserved fields are set, make no attempt to
+		 * parse the MSC structure. This MSC will still be counted by
+		 * acpi_mpam_count_msc(), meaning the MPAM driver can't probe
+		 * against all MSC, and will never be enabled. There is no way
+		 * to enable it safely, because we cannot determine safe
+		 * system-wide partid and pmg ranges in this situation.
+		 */
+		if (tbl_msc->reserved || tbl_msc->reserved1 || tbl_msc->reserved2) {
+			pr_err_once("Unrecognised MSC, MPAM not usable\n");
+			pr_debug("MSC.%u: reserved field set\n", tbl_msc->identifier);
+			continue;
+		}
+
+		if (!tbl_msc->mmio_size) {
+			pr_debug("MSC.%u: marked as disabled\n", tbl_msc->identifier);
+			continue;
+		}
+
+		pdev = acpi_mpam_parse_msc(tbl_msc);
+		if (IS_ERR(pdev))
+			return PTR_ERR(pdev);
+	}
+
+	return 0;
+}
+
+/**
+ * acpi_mpam_count_msc() - Count the number of MSC described by firmware.
+ *
+ * Returns the number of MSCs, or zero for an error.
+ *
+ * This can be called before or in parallel with acpi_mpam_parse().
+ */
+int acpi_mpam_count_msc(void)
+{
+	char *table_end, *table_offset;
+	struct acpi_mpam_msc_node *tbl_msc;
+	int count = 0;
+
+	if (acpi_disabled || !system_supports_mpam())
+		return 0;
+
+	struct acpi_table_header *table __free(acpi_put_table) =
+		acpi_get_table_pointer(ACPI_SIG_MPAM, 0);
+
+	if (IS_ERR(table))
+		return 0;
+
+	if (table->revision < 1)
+		return 0;
+
+	table_offset = (char *)(table + 1);
+	table_end = (char *)table + table->length;
+
+	while (table_offset < table_end) {
+		tbl_msc = (struct acpi_mpam_msc_node *)table_offset;
+
+		if (table_offset + sizeof(*tbl_msc) > table_end)
+			return -EINVAL;
+		if (tbl_msc->length < sizeof(*tbl_msc))
+			return -EINVAL;
+		if (tbl_msc->length > table_end - table_offset)
+			return -EINVAL;
+		table_offset += tbl_msc->length;
+
+		if (!tbl_msc->mmio_size)
+			continue;
+
+		count++;
+	}
+
+	return count;
+}
+
+/*
+ * Call after ACPI devices have been created, which happens behind acpi_scan_init()
+ * called from subsys_initcall(). PCC requires the mailbox driver, which is
+ * initialised from postcore_initcall().
+ */
+subsys_initcall_sync(acpi_mpam_parse);
diff --git a/drivers/acpi/battery.c b/drivers/acpi/battery.c
index 45593612a4db..34181fa52e93 100644
--- a/drivers/acpi/battery.c
+++ b/drivers/acpi/battery.c
@@ -91,8 +91,7 @@ enum {
 };
 
 struct acpi_battery {
-	struct mutex lock;
-	struct mutex sysfs_lock;
+	struct mutex update_lock;
 	struct power_supply *bat;
 	struct power_supply_desc bat_desc;
 	struct acpi_device *device;
@@ -243,23 +242,10 @@ static int acpi_battery_get_property(struct power_supply *psy,
 		break;
 	case POWER_SUPPLY_PROP_CURRENT_NOW:
 	case POWER_SUPPLY_PROP_POWER_NOW:
-		if (battery->rate_now == ACPI_BATTERY_VALUE_UNKNOWN) {
+		if (battery->rate_now == ACPI_BATTERY_VALUE_UNKNOWN)
 			ret = -ENODEV;
-			break;
-		}
-
-		val->intval = battery->rate_now * 1000;
-		/*
-		 * When discharging, the current should be reported as a
-		 * negative number as per the power supply class interface
-		 * definition.
-		 */
-		if (psp == POWER_SUPPLY_PROP_CURRENT_NOW &&
-		    (battery->state & ACPI_BATTERY_STATE_DISCHARGING) &&
-		    acpi_battery_handle_discharging(battery)
-				== POWER_SUPPLY_STATUS_DISCHARGING)
-			val->intval = -val->intval;
-
+		else
+			val->intval = battery->rate_now * 1000;
 		break;
 	case POWER_SUPPLY_PROP_CHARGE_FULL_DESIGN:
 	case POWER_SUPPLY_PROP_ENERGY_FULL_DESIGN:
@@ -548,11 +534,9 @@ static int acpi_battery_get_info(struct acpi_battery *battery)
 		struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
 		acpi_status status = AE_ERROR;
 
-		mutex_lock(&battery->lock);
 		status = acpi_evaluate_object(battery->device->handle,
 					      use_bix ? "_BIX":"_BIF",
 					      NULL, &buffer);
-		mutex_unlock(&battery->lock);
 
 		if (ACPI_FAILURE(status)) {
 			acpi_handle_info(battery->device->handle,
@@ -589,11 +573,8 @@ static int acpi_battery_get_state(struct acpi_battery *battery)
 			msecs_to_jiffies(cache_time)))
 		return 0;
 
-	mutex_lock(&battery->lock);
 	status = acpi_evaluate_object(battery->device->handle, "_BST",
 				      NULL, &buffer);
-	mutex_unlock(&battery->lock);
-
 	if (ACPI_FAILURE(status)) {
 		acpi_handle_info(battery->device->handle,
 				 "_BST evaluation failed: %s",
@@ -641,11 +622,8 @@ static int acpi_battery_set_alarm(struct acpi_battery *battery)
 	    !test_bit(ACPI_BATTERY_ALARM_PRESENT, &battery->flags))
 		return -ENODEV;
 
-	mutex_lock(&battery->lock);
 	status = acpi_execute_simple_method(battery->device->handle, "_BTP",
 					    battery->alarm);
-	mutex_unlock(&battery->lock);
-
 	if (ACPI_FAILURE(status))
 		return -ENODEV;
 
@@ -917,15 +895,12 @@ static int sysfs_add_battery(struct acpi_battery *battery)
 
 static void sysfs_remove_battery(struct acpi_battery *battery)
 {
-	mutex_lock(&battery->sysfs_lock);
-	if (!battery->bat) {
-		mutex_unlock(&battery->sysfs_lock);
+	if (!battery->bat)
 		return;
-	}
+
 	battery_hook_remove_battery(battery);
 	power_supply_unregister(battery->bat);
 	battery->bat = NULL;
-	mutex_unlock(&battery->sysfs_lock);
 }
 
 static void find_battery(const struct dmi_header *dm, void *private)
@@ -1085,6 +1060,9 @@ static void acpi_battery_notify(acpi_handle handle, u32 event, void *data)
 
 	if (!battery)
 		return;
+
+	guard(mutex)(&battery->update_lock);
+
 	old = battery->bat;
 	/*
 	 * On Acer Aspire V5-573G notifications are sometimes triggered too
@@ -1107,21 +1085,22 @@ static void acpi_battery_notify(acpi_handle handle, u32 event, void *data)
 }
 
 static int battery_notify(struct notifier_block *nb,
-			       unsigned long mode, void *_unused)
+			  unsigned long mode, void *_unused)
 {
 	struct acpi_battery *battery = container_of(nb, struct acpi_battery,
 						    pm_nb);
-	int result;
 
-	switch (mode) {
-	case PM_POST_HIBERNATION:
-	case PM_POST_SUSPEND:
+	if (mode == PM_POST_SUSPEND || mode == PM_POST_HIBERNATION) {
+		guard(mutex)(&battery->update_lock);
+
 		if (!acpi_battery_present(battery))
 			return 0;
 
 		if (battery->bat) {
 			acpi_battery_refresh(battery);
 		} else {
+			int result;
+
 			result = acpi_battery_get_info(battery);
 			if (result)
 				return result;
@@ -1133,7 +1112,6 @@ static int battery_notify(struct notifier_block *nb,
 
 		acpi_battery_init_alarm(battery);
 		acpi_battery_get_state(battery);
-		break;
 	}
 
 	return 0;
@@ -1211,6 +1189,8 @@ static int acpi_battery_update_retry(struct acpi_battery *battery)
 {
 	int retry, ret;
 
+	guard(mutex)(&battery->update_lock);
+
 	for (retry = 5; retry; retry--) {
 		ret = acpi_battery_update(battery, false);
 		if (!ret)
@@ -1221,6 +1201,13 @@ static int acpi_battery_update_retry(struct acpi_battery *battery)
 	return ret;
 }
 
+static void sysfs_battery_cleanup(struct acpi_battery *battery)
+{
+	guard(mutex)(&battery->update_lock);
+
+	sysfs_remove_battery(battery);
+}
+
 static int acpi_battery_add(struct acpi_device *device)
 {
 	int result = 0;
@@ -1239,11 +1226,8 @@ static int acpi_battery_add(struct acpi_device *device)
 	strscpy(acpi_device_name(device), ACPI_BATTERY_DEVICE_NAME);
 	strscpy(acpi_device_class(device), ACPI_BATTERY_CLASS);
 	device->driver_data = battery;
-	result = devm_mutex_init(&device->dev, &battery->lock);
-	if (result)
-		return result;
 
-	result = devm_mutex_init(&device->dev, &battery->sysfs_lock);
+	result = devm_mutex_init(&device->dev, &battery->update_lock);
 	if (result)
 		return result;
 
@@ -1275,7 +1259,7 @@ fail_pm:
 	device_init_wakeup(&device->dev, 0);
 	unregister_pm_notifier(&battery->pm_nb);
 fail:
-	sysfs_remove_battery(battery);
+	sysfs_battery_cleanup(battery);
 
 	return result;
 }
@@ -1294,6 +1278,9 @@ static void acpi_battery_remove(struct acpi_device *device)
 
 	device_init_wakeup(&device->dev, 0);
 	unregister_pm_notifier(&battery->pm_nb);
+
+	guard(mutex)(&battery->update_lock);
+
 	sysfs_remove_battery(battery);
 }
 
@@ -1310,6 +1297,9 @@ static int acpi_battery_resume(struct device *dev)
 		return -EINVAL;
 
 	battery->update_time = 0;
+
+	guard(mutex)(&battery->update_lock);
+
 	acpi_battery_update(battery, true);
 	return 0;
 }
diff --git a/drivers/acpi/bgrt.c b/drivers/acpi/bgrt.c
index 35ece8e9f15d..0fdd581ef96f 100644
--- a/drivers/acpi/bgrt.c
+++ b/drivers/acpi/bgrt.c
@@ -47,7 +47,7 @@ static const struct bin_attribute *const bgrt_bin_attributes[] = {
 
 static const struct attribute_group bgrt_attribute_group = {
 	.attrs = bgrt_attributes,
-	.bin_attrs_new = bgrt_bin_attributes,
+	.bin_attrs = bgrt_bin_attributes,
 };
 
 int __init acpi_parse_bgrt(struct acpi_table_header *table)
diff --git a/drivers/acpi/bus.c b/drivers/acpi/bus.c
index c2ab2783303f..a984ccd4a2a0 100644
--- a/drivers/acpi/bus.c
+++ b/drivers/acpi/bus.c
@@ -1406,7 +1406,7 @@ static int __init acpi_bus_init(void)
 		goto error1;
 
 	/*
-	 * Register the for all standard device notifications.
+	 * Register for all standard device notifications.
 	 */
 	status =
 	    acpi_install_notify_handler(ACPI_ROOT_OBJECT, ACPI_SYSTEM_NOTIFY,
diff --git a/drivers/acpi/button.c b/drivers/acpi/button.c
index 0a7026040188..3c6dd9b4ba0a 100644
--- a/drivers/acpi/button.c
+++ b/drivers/acpi/button.c
@@ -619,8 +619,10 @@ static int acpi_button_add(struct acpi_device *device)
 
 	input_set_drvdata(input, device);
 	error = input_register_device(input);
-	if (error)
+	if (error) {
+		input_free_device(input);
 		goto err_remove_fs;
+	}
 
 	switch (device->device_type) {
 	case ACPI_BUS_TYPE_POWER_BUTTON:
diff --git a/drivers/acpi/cppc_acpi.c b/drivers/acpi/cppc_acpi.c
index a9ae2fd62863..3bdeeee3414e 100644
--- a/drivers/acpi/cppc_acpi.c
+++ b/drivers/acpi/cppc_acpi.c
@@ -460,7 +460,7 @@ bool acpi_cpc_valid(void)
 	if (acpi_disabled)
 		return false;
 
-	for_each_present_cpu(cpu) {
+	for_each_online_cpu(cpu) {
 		cpc_ptr = per_cpu(cpc_desc_ptr, cpu);
 		if (!cpc_ptr)
 			return false;
@@ -476,7 +476,7 @@ bool cppc_allow_fast_switch(void)
 	struct cpc_desc *cpc_ptr;
 	int cpu;
 
-	for_each_possible_cpu(cpu) {
+	for_each_online_cpu(cpu) {
 		cpc_ptr = per_cpu(cpc_desc_ptr, cpu);
 		desired_reg = &cpc_ptr->cpc_regs[DESIRED_PERF];
 		if (!CPC_IN_SYSTEM_MEMORY(desired_reg) &&
@@ -750,7 +750,7 @@ int acpi_cppc_processor_probe(struct acpi_processor *pr)
 	}
 
 	/*
-	 * Disregard _CPC if the number of entries in the return pachage is not
+	 * Disregard _CPC if the number of entries in the return package is not
 	 * as expected, but support future revisions being proper supersets of
 	 * the v3 and only causing more entries to be returned by _CPC.
 	 */
@@ -1435,7 +1435,7 @@ bool cppc_perf_ctrs_in_pcc(void)
 {
 	int cpu;
 
-	for_each_present_cpu(cpu) {
+	for_each_online_cpu(cpu) {
 		struct cpc_register_resource *ref_perf_reg;
 		struct cpc_desc *cpc_desc;
 
@@ -1876,7 +1876,7 @@ EXPORT_SYMBOL_GPL(cppc_set_perf);
  * If desired_reg is in the SystemMemory or SystemIo ACPI address space,
  * then assume there is no latency.
  */
-unsigned int cppc_get_transition_latency(int cpu_num)
+int cppc_get_transition_latency(int cpu_num)
 {
 	/*
 	 * Expected transition latency is based on the PCCT timing values
@@ -1889,31 +1889,29 @@ unsigned int cppc_get_transition_latency(int cpu_num)
 	 *              completion of a command before issuing the next command,
 	 *              in microseconds.
 	 */
-	unsigned int latency_ns = 0;
 	struct cpc_desc *cpc_desc;
 	struct cpc_register_resource *desired_reg;
 	int pcc_ss_id = per_cpu(cpu_pcc_subspace_idx, cpu_num);
 	struct cppc_pcc_data *pcc_ss_data;
+	int latency_ns = 0;
 
 	cpc_desc = per_cpu(cpc_desc_ptr, cpu_num);
 	if (!cpc_desc)
-		return CPUFREQ_ETERNAL;
+		return -ENODATA;
 
 	desired_reg = &cpc_desc->cpc_regs[DESIRED_PERF];
 	if (CPC_IN_SYSTEM_MEMORY(desired_reg) || CPC_IN_SYSTEM_IO(desired_reg))
 		return 0;
-	else if (!CPC_IN_PCC(desired_reg))
-		return CPUFREQ_ETERNAL;
 
-	if (pcc_ss_id < 0)
-		return CPUFREQ_ETERNAL;
+	if (!CPC_IN_PCC(desired_reg) || pcc_ss_id < 0)
+		return -ENODATA;
 
 	pcc_ss_data = pcc_data[pcc_ss_id];
 	if (pcc_ss_data->pcc_mpar)
 		latency_ns = 60 * (1000 * 1000 * 1000 / pcc_ss_data->pcc_mpar);
 
-	latency_ns = max(latency_ns, pcc_ss_data->pcc_nominal * 1000);
-	latency_ns = max(latency_ns, pcc_ss_data->pcc_mrtt * 1000);
+	latency_ns = max_t(int, latency_ns, pcc_ss_data->pcc_nominal * 1000);
+	latency_ns = max_t(int, latency_ns, pcc_ss_data->pcc_mrtt * 1000);
 
 	return latency_ns;
 }
diff --git a/drivers/acpi/device_pm.c b/drivers/acpi/device_pm.c
index dbd4446025ec..4e0583274b8f 100644
--- a/drivers/acpi/device_pm.c
+++ b/drivers/acpi/device_pm.c
@@ -1119,6 +1119,8 @@ int acpi_subsys_prepare(struct device *dev)
 {
 	struct acpi_device *adev = ACPI_COMPANION(dev);
 
+	dev_pm_set_strict_midlayer(dev, true);
+
 	if (dev->driver && dev->driver->pm && dev->driver->pm->prepare) {
 		int ret = dev->driver->pm->prepare(dev);
 
@@ -1147,6 +1149,8 @@ void acpi_subsys_complete(struct device *dev)
 	 */
 	if (pm_runtime_suspended(dev) && pm_resume_via_firmware())
 		pm_request_resume(dev);
+
+	dev_pm_set_strict_midlayer(dev, false);
 }
 EXPORT_SYMBOL_GPL(acpi_subsys_complete);
 
@@ -1362,6 +1366,8 @@ static int acpi_subsys_poweroff_noirq(struct device *dev)
 }
 #endif /* CONFIG_PM_SLEEP */
 
+static void acpi_dev_pm_detach(struct device *dev, bool power_off);
+
 static struct dev_pm_domain acpi_general_pm_domain = {
 	.ops = {
 		.runtime_suspend = acpi_subsys_runtime_suspend,
@@ -1382,6 +1388,7 @@ static struct dev_pm_domain acpi_general_pm_domain = {
 		.restore_early = acpi_subsys_restore_early,
 #endif
 	},
+	.detach = acpi_dev_pm_detach,
 };
 
 /**
@@ -1465,7 +1472,6 @@ int acpi_dev_pm_attach(struct device *dev, bool power_on)
 		acpi_device_wakeup_disable(adev);
 	}
 
-	dev->pm_domain->detach = acpi_dev_pm_detach;
 	return 1;
 }
 EXPORT_SYMBOL_GPL(acpi_dev_pm_attach);
diff --git a/drivers/acpi/device_sysfs.c b/drivers/acpi/device_sysfs.c
index 3961fc47152c..cd199fbe4dc9 100644
--- a/drivers/acpi/device_sysfs.c
+++ b/drivers/acpi/device_sysfs.c
@@ -464,7 +464,7 @@ static ssize_t description_show(struct device *dev,
 
 	buf[result++] = '\n';
 
-	kfree(str_obj);
+	ACPI_FREE(str_obj);
 
 	return result;
 }
diff --git a/drivers/acpi/dptf/Makefile b/drivers/acpi/dptf/Makefile
index 297340682f66..e912a3be1d28 100644
--- a/drivers/acpi/dptf/Makefile
+++ b/drivers/acpi/dptf/Makefile
@@ -1,4 +1,3 @@
 # SPDX-License-Identifier: GPL-2.0-only
-obj-$(CONFIG_ACPI)             += int340x_thermal.o
 obj-$(CONFIG_DPTF_POWER)	+= dptf_power.o
 obj-$(CONFIG_DPTF_PCH_FIVR)	+= dptf_pch_fivr.o
diff --git a/drivers/acpi/dptf/dptf_pch_fivr.c b/drivers/acpi/dptf/dptf_pch_fivr.c
index 952216c67d58..8d7e555929d3 100644
--- a/drivers/acpi/dptf/dptf_pch_fivr.c
+++ b/drivers/acpi/dptf/dptf_pch_fivr.c
@@ -41,7 +41,7 @@ static int pch_fivr_read(acpi_handle handle, char *method, struct pch_fivr_resp
 	ret = 0;
 
 release_buffer:
-	kfree(buffer.pointer);
+	ACPI_FREE(buffer.pointer);
 	return ret;
 }
 
diff --git a/drivers/acpi/dptf/dptf_power.c b/drivers/acpi/dptf/dptf_power.c
index e8caf4106ff9..55ccbb8ddbe3 100644
--- a/drivers/acpi/dptf/dptf_power.c
+++ b/drivers/acpi/dptf/dptf_power.c
@@ -238,6 +238,10 @@ static const struct acpi_device_id int3407_device_ids[] = {
 	{"INTC10A5", 0},
 	{"INTC10D8", 0},
 	{"INTC10D9", 0},
+	{"INTC1100", 0},
+	{"INTC1101", 0},
+	{"INTC10F7", 0},
+	{"INTC10F8", 0},
 	{"", 0},
 };
 MODULE_DEVICE_TABLE(acpi, int3407_device_ids);
diff --git a/drivers/acpi/dptf/int340x_thermal.c b/drivers/acpi/dptf/int340x_thermal.c
deleted file mode 100644
index aef7aca2161d..000000000000
--- a/drivers/acpi/dptf/int340x_thermal.c
+++ /dev/null
@@ -1,87 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-only
-/*
- * ACPI support for int340x thermal drivers
- *
- * Copyright (C) 2014, Intel Corporation
- * Authors: Zhang Rui <rui.zhang@intel.com>
- */
-
-#include <linux/acpi.h>
-#include <linux/module.h>
-
-#include "../internal.h"
-
-#define INT3401_DEVICE 0X01
-static const struct acpi_device_id int340x_thermal_device_ids[] = {
-	{"INT3400"},
-	{"INT3401", INT3401_DEVICE},
-	{"INT3402"},
-	{"INT3403"},
-	{"INT3404"},
-	{"INT3406"},
-	{"INT3407"},
-	{"INT3408"},
-	{"INT3409"},
-	{"INT340A"},
-	{"INT340B"},
-	{"INT3532"},
-	{"INTC1040"},
-	{"INTC1041"},
-	{"INTC1042"},
-	{"INTC1043"},
-	{"INTC1044"},
-	{"INTC1045"},
-	{"INTC1046"},
-	{"INTC1047"},
-	{"INTC1048"},
-	{"INTC1049"},
-	{"INTC1050"},
-	{"INTC1060"},
-	{"INTC1061"},
-	{"INTC1062"},
-	{"INTC1063"},
-	{"INTC1064"},
-	{"INTC1065"},
-	{"INTC1066"},
-	{"INTC1068"},
-	{"INTC1069"},
-	{"INTC106A"},
-	{"INTC106B"},
-	{"INTC106C"},
-	{"INTC106D"},
-	{"INTC10A0"},
-	{"INTC10A1"},
-	{"INTC10A2"},
-	{"INTC10A3"},
-	{"INTC10A4"},
-	{"INTC10A5"},
-	{"INTC10D4"},
-	{"INTC10D5"},
-	{"INTC10D6"},
-	{"INTC10D7"},
-	{"INTC10D8"},
-	{"INTC10D9"},
-	{""},
-};
-
-static int int340x_thermal_handler_attach(struct acpi_device *adev,
-					const struct acpi_device_id *id)
-{
-	if (IS_ENABLED(CONFIG_INT340X_THERMAL))
-		acpi_create_platform_device(adev, NULL);
-	/* Intel SoC DTS thermal driver needs INT3401 to set IRQ descriptor */
-	else if (IS_ENABLED(CONFIG_INTEL_SOC_DTS_THERMAL) &&
-		 id->driver_data == INT3401_DEVICE)
-		acpi_create_platform_device(adev, NULL);
-	return 1;
-}
-
-static struct acpi_scan_handler int340x_thermal_handler = {
-	.ids = int340x_thermal_device_ids,
-	.attach = int340x_thermal_handler_attach,
-};
-
-void __init acpi_int340x_thermal_init(void)
-{
-	acpi_scan_add_handler(&int340x_thermal_handler);
-}
diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c
index 6f4203716b53..59b3d50ff01e 100644
--- a/drivers/acpi/ec.c
+++ b/drivers/acpi/ec.c
@@ -23,8 +23,10 @@
 #include <linux/delay.h>
 #include <linux/interrupt.h>
 #include <linux/list.h>
+#include <linux/printk.h>
 #include <linux/spinlock.h>
 #include <linux/slab.h>
+#include <linux/string.h>
 #include <linux/suspend.h>
 #include <linux/acpi.h>
 #include <linux/dmi.h>
@@ -2031,6 +2033,25 @@ void __init acpi_ec_ecdt_probe(void)
 		goto out;
 	}
 
+	if (!strlen(ecdt_ptr->id)) {
+		/*
+		 * The ECDT table on some MSI notebooks contains invalid data, together
+		 * with an empty ID string ("").
+		 *
+		 * Section 5.2.15 of the ACPI specification requires the ID string to be
+		 * a "fully qualified reference to the (...) embedded controller device",
+		 * so this string always has to start with a backslash.
+		 *
+		 * However some ThinkBook machines have a ECDT table with a valid EC
+		 * description but an invalid ID string ("_SB.PC00.LPCB.EC0").
+		 *
+		 * Because of this we only check if the ID string is empty in order to
+		 * avoid the obvious cases.
+		 */
+		pr_err(FW_BUG "Ignoring ECDT due to empty ID string\n");
+		goto out;
+	}
+
 	ec = acpi_ec_alloc();
 	if (!ec)
 		goto out;
@@ -2273,7 +2294,8 @@ static int acpi_ec_init_workqueues(void)
 		ec_wq = alloc_ordered_workqueue("kec", 0);
 
 	if (!ec_query_wq)
-		ec_query_wq = alloc_workqueue("kec_query", 0, ec_max_queries);
+		ec_query_wq = alloc_workqueue("kec_query", WQ_PERCPU,
+					      ec_max_queries);
 
 	if (!ec_wq || !ec_query_wq) {
 		acpi_ec_destroy_workqueues();
diff --git a/drivers/acpi/fan.h b/drivers/acpi/fan.h
index 15eba1c70e66..97ce3212edf3 100644
--- a/drivers/acpi/fan.h
+++ b/drivers/acpi/fan.h
@@ -11,6 +11,7 @@
 #define _ACPI_FAN_H_
 
 #include <linux/kconfig.h>
+#include <linux/limits.h>
 
 #define ACPI_FAN_DEVICE_IDS	\
 	{"INT3404", }, /* Fan */ \
@@ -20,6 +21,8 @@
 	{"INTC106A", }, /* Fan for Lunar Lake generation */ \
 	{"INTC10A2", }, /* Fan for Raptor Lake generation */ \
 	{"INTC10D6", }, /* Fan for Panther Lake generation */ \
+	{"INTC10FE", }, /* Fan for Wildcat Lake generation */ \
+	{"INTC10F5", }, /* Fan for Nova Lake generation */ \
 	{"PNP0C0B", } /* Generic ACPI fan */
 
 #define ACPI_FPS_NAME_LEN	20
@@ -48,24 +51,64 @@ struct acpi_fan_fst {
 };
 
 struct acpi_fan {
+	acpi_handle handle;
 	bool acpi4;
 	bool has_fst;
 	struct acpi_fan_fif fif;
 	struct acpi_fan_fps *fps;
 	int fps_count;
+	/* A value of 0 means that trippoint-related functions are not supported */
+	u32 fan_trip_granularity;
+#if IS_REACHABLE(CONFIG_HWMON)
+	struct device *hdev;
+#endif
 	struct thermal_cooling_device *cdev;
 	struct device_attribute fst_speed;
 	struct device_attribute fine_grain_control;
 };
 
-int acpi_fan_get_fst(struct acpi_device *device, struct acpi_fan_fst *fst);
+/**
+ * acpi_fan_speed_valid - Check if fan speed value is valid
+ * @speeed: Speed value returned by the ACPI firmware
+ *
+ * Check if the fan speed value returned by the ACPI firmware is valid. This function is
+ * necessary as ACPI firmware implementations can return 0xFFFFFFFF to signal that the
+ * ACPI fan does not support speed reporting. Additionally, some buggy ACPI firmware
+ * implementations return a value larger than the 32-bit integer value defined by
+ * the ACPI specification when using placeholder values. Such invalid values are also
+ * detected by this function.
+ *
+ * Returns: True if the fan speed value is valid, false otherwise.
+ */
+static inline bool acpi_fan_speed_valid(u64 speed)
+{
+	return speed < U32_MAX;
+}
+
+/**
+ * acpi_fan_power_valid - Check if fan power value is valid
+ * @power: Power value returned by the ACPI firmware
+ *
+ * Check if the fan power value returned by the ACPI firmware is valid.
+ * See acpi_fan_speed_valid() for details.
+ *
+ * Returns: True if the fan power value is valid, false otherwise.
+ */
+static inline bool acpi_fan_power_valid(u64 power)
+{
+	return power < U32_MAX;
+}
+
+int acpi_fan_get_fst(acpi_handle handle, struct acpi_fan_fst *fst);
 int acpi_fan_create_attributes(struct acpi_device *device);
 void acpi_fan_delete_attributes(struct acpi_device *device);
 
 #if IS_REACHABLE(CONFIG_HWMON)
-int devm_acpi_fan_create_hwmon(struct acpi_device *device);
+int devm_acpi_fan_create_hwmon(struct device *dev);
+void acpi_fan_notify_hwmon(struct device *dev);
 #else
-static inline int devm_acpi_fan_create_hwmon(struct acpi_device *device) { return 0; };
+static inline int devm_acpi_fan_create_hwmon(struct device *dev) { return 0; };
+static inline void acpi_fan_notify_hwmon(struct device *dev) { };
 #endif
 
 #endif
diff --git a/drivers/acpi/fan_attr.c b/drivers/acpi/fan_attr.c
index 22d29ac2447c..9b7fa52f3c2a 100644
--- a/drivers/acpi/fan_attr.c
+++ b/drivers/acpi/fan_attr.c
@@ -22,9 +22,9 @@ static ssize_t show_state(struct device *dev, struct device_attribute *attr, cha
 	int count;
 
 	if (fps->control == 0xFFFFFFFF || fps->control > 100)
-		count = scnprintf(buf, PAGE_SIZE, "not-defined:");
+		count = sysfs_emit(buf, "not-defined:");
 	else
-		count = scnprintf(buf, PAGE_SIZE, "%lld:", fps->control);
+		count = sysfs_emit(buf, "%lld:", fps->control);
 
 	if (fps->trip_point == 0xFFFFFFFF || fps->trip_point > 9)
 		count += sysfs_emit_at(buf, count, "not-defined:");
@@ -55,11 +55,11 @@ static ssize_t show_fan_speed(struct device *dev, struct device_attribute *attr,
 	struct acpi_fan_fst fst;
 	int status;
 
-	status = acpi_fan_get_fst(acpi_dev, &fst);
+	status = acpi_fan_get_fst(acpi_dev->handle, &fst);
 	if (status)
 		return status;
 
-	return sprintf(buf, "%lld\n", fst.speed);
+	return sysfs_emit(buf, "%lld\n", fst.speed);
 }
 
 static ssize_t show_fine_grain_control(struct device *dev, struct device_attribute *attr, char *buf)
@@ -67,7 +67,7 @@ static ssize_t show_fine_grain_control(struct device *dev, struct device_attribu
 	struct acpi_device *acpi_dev = container_of(dev, struct acpi_device, dev);
 	struct acpi_fan *fan = acpi_driver_data(acpi_dev);
 
-	return sprintf(buf, "%d\n", fan->fif.fine_grain_ctrl);
+	return sysfs_emit(buf, "%d\n", fan->fif.fine_grain_ctrl);
 }
 
 int acpi_fan_create_attributes(struct acpi_device *device)
diff --git a/drivers/acpi/fan_core.c b/drivers/acpi/fan_core.c
index 8ad12ad3aaaf..fb08b8549ed7 100644
--- a/drivers/acpi/fan_core.c
+++ b/drivers/acpi/fan_core.c
@@ -7,11 +7,16 @@
  *  Copyright (C) 2022 Intel Corporation. All rights reserved.
  */
 
+#include <linux/bits.h>
 #include <linux/kernel.h>
+#include <linux/limits.h>
+#include <linux/math.h>
+#include <linux/math64.h>
 #include <linux/module.h>
 #include <linux/init.h>
 #include <linux/types.h>
 #include <linux/uaccess.h>
+#include <linux/uuid.h>
 #include <linux/thermal.h>
 #include <linux/acpi.h>
 #include <linux/platform_device.h>
@@ -19,6 +24,26 @@
 
 #include "fan.h"
 
+#define ACPI_FAN_NOTIFY_STATE_CHANGED	0x80
+
+/*
+ * Defined inside the "Fan Noise Signal" section at
+ * https://learn.microsoft.com/en-us/windows-hardware/design/device-experiences/design-guide.
+ */
+static const guid_t acpi_fan_microsoft_guid = GUID_INIT(0xA7611840, 0x99FE, 0x41AE, 0xA4, 0x88,
+							0x35, 0xC7, 0x59, 0x26, 0xC8, 0xEB);
+#define ACPI_FAN_DSM_GET_TRIP_POINT_GRANULARITY 1
+#define ACPI_FAN_DSM_SET_TRIP_POINTS		2
+#define ACPI_FAN_DSM_GET_OPERATING_RANGES	3
+
+/*
+ * Ensures that fans with a very low trip point granularity
+ * do not send too many notifications.
+ */
+static uint min_trip_distance = 100;
+module_param(min_trip_distance, uint, 0);
+MODULE_PARM_DESC(min_trip_distance, "Minimum distance between fan speed trip points in RPM");
+
 static const struct acpi_device_id fan_device_ids[] = {
 	ACPI_FAN_DEVICE_IDS,
 	{"", 0},
@@ -44,25 +69,30 @@ static int fan_get_max_state(struct thermal_cooling_device *cdev, unsigned long
 	return 0;
 }
 
-int acpi_fan_get_fst(struct acpi_device *device, struct acpi_fan_fst *fst)
+int acpi_fan_get_fst(acpi_handle handle, struct acpi_fan_fst *fst)
 {
 	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
 	union acpi_object *obj;
 	acpi_status status;
 	int ret = 0;
 
-	status = acpi_evaluate_object(device->handle, "_FST", NULL, &buffer);
-	if (ACPI_FAILURE(status)) {
-		dev_err(&device->dev, "Get fan state failed\n");
-		return -ENODEV;
-	}
+	status = acpi_evaluate_object(handle, "_FST", NULL, &buffer);
+	if (ACPI_FAILURE(status))
+		return -EIO;
 
 	obj = buffer.pointer;
-	if (!obj || obj->type != ACPI_TYPE_PACKAGE ||
-	    obj->package.count != 3 ||
-	    obj->package.elements[1].type != ACPI_TYPE_INTEGER) {
-		dev_err(&device->dev, "Invalid _FST data\n");
-		ret = -EINVAL;
+	if (!obj)
+		return -ENODATA;
+
+	if (obj->type != ACPI_TYPE_PACKAGE || obj->package.count != 3) {
+		ret = -EPROTO;
+		goto err;
+	}
+
+	if (obj->package.elements[0].type != ACPI_TYPE_INTEGER ||
+	    obj->package.elements[1].type != ACPI_TYPE_INTEGER ||
+	    obj->package.elements[2].type != ACPI_TYPE_INTEGER) {
+		ret = -EPROTO;
 		goto err;
 	}
 
@@ -81,7 +111,7 @@ static int fan_get_state_acpi4(struct acpi_device *device, unsigned long *state)
 	struct acpi_fan_fst fst;
 	int status, i;
 
-	status = acpi_fan_get_fst(device, &fst);
+	status = acpi_fan_get_fst(device->handle, &fst);
 	if (status)
 		return status;
 
@@ -102,7 +132,7 @@ match_fps:
 			break;
 	}
 	if (i == fan->fps_count) {
-		dev_dbg(&device->dev, "Invalid control value returned\n");
+		dev_dbg(&device->dev, "No matching fps control value\n");
 		return -EINVAL;
 	}
 
@@ -203,18 +233,6 @@ static const struct thermal_cooling_device_ops fan_cooling_ops = {
  * --------------------------------------------------------------------------
 */
 
-static bool acpi_fan_has_fst(struct acpi_device *device)
-{
-	return acpi_has_method(device->handle, "_FST");
-}
-
-static bool acpi_fan_is_acpi4(struct acpi_device *device)
-{
-	return acpi_has_method(device->handle, "_FIF") &&
-	       acpi_has_method(device->handle, "_FPS") &&
-	       acpi_has_method(device->handle, "_FSL");
-}
-
 static int acpi_fan_get_fif(struct acpi_device *device)
 {
 	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
@@ -315,6 +333,182 @@ err:
 	return status;
 }
 
+static int acpi_fan_dsm_init(struct device *dev)
+{
+	union acpi_object dummy = {
+		.package = {
+			.type = ACPI_TYPE_PACKAGE,
+			.count = 0,
+			.elements = NULL,
+		},
+	};
+	struct acpi_fan *fan = dev_get_drvdata(dev);
+	union acpi_object *obj;
+	int ret = 0;
+
+	if (!acpi_check_dsm(fan->handle, &acpi_fan_microsoft_guid, 0,
+			    BIT(ACPI_FAN_DSM_GET_TRIP_POINT_GRANULARITY) |
+			    BIT(ACPI_FAN_DSM_SET_TRIP_POINTS)))
+		return 0;
+
+	dev_info(dev, "Using Microsoft fan extensions\n");
+
+	obj = acpi_evaluate_dsm_typed(fan->handle, &acpi_fan_microsoft_guid, 0,
+				      ACPI_FAN_DSM_GET_TRIP_POINT_GRANULARITY, &dummy,
+				      ACPI_TYPE_INTEGER);
+	if (!obj)
+		return -EIO;
+
+	if (obj->integer.value > U32_MAX)
+		ret = -EOVERFLOW;
+	else
+		fan->fan_trip_granularity = obj->integer.value;
+
+	kfree(obj);
+
+	return ret;
+}
+
+static int acpi_fan_dsm_set_trip_points(struct device *dev, u64 upper, u64 lower)
+{
+	union acpi_object args[2] = {
+		{
+			.integer = {
+				.type = ACPI_TYPE_INTEGER,
+				.value = lower,
+			},
+		},
+		{
+			.integer = {
+				.type = ACPI_TYPE_INTEGER,
+				.value = upper,
+			},
+		},
+	};
+	struct acpi_fan *fan = dev_get_drvdata(dev);
+	union acpi_object in = {
+		.package = {
+			.type = ACPI_TYPE_PACKAGE,
+			.count = ARRAY_SIZE(args),
+			.elements = args,
+		},
+	};
+	union acpi_object *obj;
+
+	obj = acpi_evaluate_dsm(fan->handle, &acpi_fan_microsoft_guid, 0,
+				ACPI_FAN_DSM_SET_TRIP_POINTS, &in);
+	kfree(obj);
+
+	return 0;
+}
+
+static int acpi_fan_dsm_start(struct device *dev)
+{
+	struct acpi_fan *fan = dev_get_drvdata(dev);
+	int ret;
+
+	if (!fan->fan_trip_granularity)
+		return 0;
+
+	/*
+	 * Some firmware implementations only update the values returned by the
+	 * _FST control method when a notification is received. This usually
+	 * works with Microsoft Windows as setting up trip points will keep
+	 * triggering said notifications, but will cause issues when using _FST
+	 * without the Microsoft-specific trip point extension.
+	 *
+	 * Because of this, an initial notification needs to be triggered to
+	 * start the cycle of trip points updates. This is achieved by setting
+	 * the trip points sequencially to two separate ranges. As by the
+	 * Microsoft specification the firmware should trigger a notification
+	 * immediately if the fan speed is outside the trip point range. This
+	 * _should_ result in at least one notification as both ranges do not
+	 * overlap, meaning that the current fan speed needs to be outside at
+	 * least one range.
+	 */
+	ret = acpi_fan_dsm_set_trip_points(dev, fan->fan_trip_granularity, 0);
+	if (ret < 0)
+		return ret;
+
+	return acpi_fan_dsm_set_trip_points(dev, fan->fan_trip_granularity * 3,
+					    fan->fan_trip_granularity * 2);
+}
+
+static int acpi_fan_dsm_update_trips_points(struct device *dev, struct acpi_fan_fst *fst)
+{
+	struct acpi_fan *fan = dev_get_drvdata(dev);
+	u64 upper, lower;
+
+	if (!fan->fan_trip_granularity)
+		return 0;
+
+	if (!acpi_fan_speed_valid(fst->speed))
+		return -EINVAL;
+
+	upper = roundup_u64(fst->speed + min_trip_distance, fan->fan_trip_granularity);
+	if (fst->speed <= min_trip_distance) {
+		lower = 0;
+	} else {
+		/*
+		 * Valid fan speed values cannot be larger than 32 bit, so
+		 * we can safely assume that no overflow will happen here.
+		 */
+		lower = rounddown((u32)fst->speed - min_trip_distance, fan->fan_trip_granularity);
+	}
+
+	return acpi_fan_dsm_set_trip_points(dev, upper, lower);
+}
+
+static void acpi_fan_notify_handler(acpi_handle handle, u32 event, void *context)
+{
+	struct device *dev = context;
+	struct acpi_fan_fst fst;
+	int ret;
+
+	switch (event) {
+	case ACPI_FAN_NOTIFY_STATE_CHANGED:
+		/*
+		 * The ACPI specification says that we must evaluate _FST when we
+		 * receive an ACPI event indicating that the fan state has changed.
+		 */
+		ret = acpi_fan_get_fst(handle, &fst);
+		if (ret < 0) {
+			dev_err(dev, "Error retrieving current fan status: %d\n", ret);
+		} else {
+			ret = acpi_fan_dsm_update_trips_points(dev, &fst);
+			if (ret < 0)
+				dev_err(dev, "Failed to update trip points: %d\n", ret);
+		}
+
+		acpi_fan_notify_hwmon(dev);
+		acpi_bus_generate_netlink_event("fan", dev_name(dev), event, 0);
+		break;
+	default:
+		dev_dbg(dev, "Unsupported ACPI notification 0x%x\n", event);
+		break;
+	}
+}
+
+static void acpi_fan_notify_remove(void *data)
+{
+	struct acpi_fan *fan = data;
+
+	acpi_remove_notify_handler(fan->handle, ACPI_DEVICE_NOTIFY, acpi_fan_notify_handler);
+}
+
+static int devm_acpi_fan_notify_init(struct device *dev)
+{
+	struct acpi_fan *fan = dev_get_drvdata(dev);
+	acpi_status status;
+
+	status = acpi_install_notify_handler(fan->handle, ACPI_DEVICE_NOTIFY,
+					     acpi_fan_notify_handler, dev);
+	if (ACPI_FAILURE(status))
+		return -EIO;
+
+	return devm_add_action_or_reset(dev, acpi_fan_notify_remove, fan);
+}
+
 static int acpi_fan_probe(struct platform_device *pdev)
 {
 	int result = 0;
@@ -323,17 +517,24 @@ static int acpi_fan_probe(struct platform_device *pdev)
 	struct acpi_device *device = ACPI_COMPANION(&pdev->dev);
 	char *name;
 
+	if (!device)
+		return -ENODEV;
+
 	fan = devm_kzalloc(&pdev->dev, sizeof(*fan), GFP_KERNEL);
 	if (!fan) {
 		dev_err(&device->dev, "No memory for fan\n");
 		return -ENOMEM;
 	}
+
+	fan->handle = device->handle;
 	device->driver_data = fan;
 	platform_set_drvdata(pdev, fan);
 
-	if (acpi_fan_has_fst(device)) {
+	if (acpi_has_method(device->handle, "_FST")) {
 		fan->has_fst = true;
-		fan->acpi4 = acpi_fan_is_acpi4(device);
+		fan->acpi4 = acpi_has_method(device->handle, "_FIF") &&
+				acpi_has_method(device->handle, "_FPS") &&
+				acpi_has_method(device->handle, "_FSL");
 	}
 
 	if (fan->acpi4) {
@@ -347,10 +548,24 @@ static int acpi_fan_probe(struct platform_device *pdev)
 	}
 
 	if (fan->has_fst) {
-		result = devm_acpi_fan_create_hwmon(device);
+		result = acpi_fan_dsm_init(&pdev->dev);
+		if (result)
+			return result;
+
+		result = devm_acpi_fan_create_hwmon(&pdev->dev);
 		if (result)
 			return result;
 
+		result = devm_acpi_fan_notify_init(&pdev->dev);
+		if (result)
+			return result;
+
+		result = acpi_fan_dsm_start(&pdev->dev);
+		if (result) {
+			dev_err(&pdev->dev, "Failed to start Microsoft fan extensions\n");
+			return result;
+		}
+
 		result = acpi_fan_create_attributes(device);
 		if (result)
 			return result;
@@ -436,8 +651,14 @@ static int acpi_fan_suspend(struct device *dev)
 
 static int acpi_fan_resume(struct device *dev)
 {
-	int result;
 	struct acpi_fan *fan = dev_get_drvdata(dev);
+	int result;
+
+	if (fan->has_fst) {
+		result = acpi_fan_dsm_start(dev);
+		if (result)
+			dev_err(dev, "Failed to start Microsoft fan extensions: %d\n", result);
+	}
 
 	if (fan->acpi4)
 		return 0;
diff --git a/drivers/acpi/fan_hwmon.c b/drivers/acpi/fan_hwmon.c
index e8d90605106e..d3374f8f524b 100644
--- a/drivers/acpi/fan_hwmon.c
+++ b/drivers/acpi/fan_hwmon.c
@@ -15,10 +15,6 @@
 
 #include "fan.h"
 
-/* Returned when the ACPI fan does not support speed reporting */
-#define FAN_SPEED_UNAVAILABLE	U32_MAX
-#define FAN_POWER_UNAVAILABLE	U32_MAX
-
 static struct acpi_fan_fps *acpi_fan_get_current_fps(struct acpi_fan *fan, u64 control)
 {
 	unsigned int i;
@@ -77,7 +73,7 @@ static umode_t acpi_fan_hwmon_is_visible(const void *drvdata, enum hwmon_sensor_
 			 * when the associated attribute should not be created.
 			 */
 			for (i = 0; i < fan->fps_count; i++) {
-				if (fan->fps[i].power != FAN_POWER_UNAVAILABLE)
+				if (acpi_fan_power_valid(fan->fps[i].power))
 					return 0444;
 			}
 
@@ -93,13 +89,12 @@ static umode_t acpi_fan_hwmon_is_visible(const void *drvdata, enum hwmon_sensor_
 static int acpi_fan_hwmon_read(struct device *dev, enum hwmon_sensor_types type, u32 attr,
 			       int channel, long *val)
 {
-	struct acpi_device *adev = to_acpi_device(dev->parent);
 	struct acpi_fan *fan = dev_get_drvdata(dev);
 	struct acpi_fan_fps *fps;
 	struct acpi_fan_fst fst;
 	int ret;
 
-	ret = acpi_fan_get_fst(adev, &fst);
+	ret = acpi_fan_get_fst(fan->handle, &fst);
 	if (ret < 0)
 		return ret;
 
@@ -107,7 +102,7 @@ static int acpi_fan_hwmon_read(struct device *dev, enum hwmon_sensor_types type,
 	case hwmon_fan:
 		switch (attr) {
 		case hwmon_fan_input:
-			if (fst.speed == FAN_SPEED_UNAVAILABLE)
+			if (!acpi_fan_speed_valid(fst.speed))
 				return -ENODEV;
 
 			if (fst.speed > LONG_MAX)
@@ -135,7 +130,7 @@ static int acpi_fan_hwmon_read(struct device *dev, enum hwmon_sensor_types type,
 			if (!fps)
 				return -EIO;
 
-			if (fps->power == FAN_POWER_UNAVAILABLE)
+			if (!acpi_fan_power_valid(fps->power))
 				return -ENODEV;
 
 			if (fps->power > LONG_MAX / MICROWATT_PER_MILLIWATT)
@@ -167,12 +162,19 @@ static const struct hwmon_chip_info acpi_fan_hwmon_chip_info = {
 	.info = acpi_fan_hwmon_info,
 };
 
-int devm_acpi_fan_create_hwmon(struct acpi_device *device)
+void acpi_fan_notify_hwmon(struct device *dev)
+{
+	struct acpi_fan *fan = dev_get_drvdata(dev);
+
+	hwmon_notify_event(fan->hdev, hwmon_fan, hwmon_fan_input, 0);
+}
+
+int devm_acpi_fan_create_hwmon(struct device *dev)
 {
-	struct acpi_fan *fan = acpi_driver_data(device);
-	struct device *hdev;
+	struct acpi_fan *fan = dev_get_drvdata(dev);
+
+	fan->hdev = devm_hwmon_device_register_with_info(dev, "acpi_fan", fan,
+							 &acpi_fan_hwmon_chip_info, NULL);
 
-	hdev = devm_hwmon_device_register_with_info(&device->dev, "acpi_fan", fan,
-						    &acpi_fan_hwmon_chip_info, NULL);
-	return PTR_ERR_OR_ZERO(hdev);
+	return PTR_ERR_OR_ZERO(fan->hdev);
 }
diff --git a/drivers/acpi/internal.h b/drivers/acpi/internal.h
index 00910ccd7eda..40f875b265a9 100644
--- a/drivers/acpi/internal.h
+++ b/drivers/acpi/internal.h
@@ -27,7 +27,6 @@ static inline void acpi_pci_link_init(void) {}
 void acpi_processor_init(void);
 void acpi_platform_init(void);
 void acpi_pnp_init(void);
-void acpi_int340x_thermal_init(void);
 int acpi_sysfs_init(void);
 void acpi_gpe_apply_masked_gpes(void);
 void acpi_container_init(void);
@@ -140,6 +139,7 @@ int __acpi_device_uevent_modalias(const struct acpi_device *adev,
 /* --------------------------------------------------------------------------
                                   Power Resource
    -------------------------------------------------------------------------- */
+void acpi_power_resources_init(void);
 void acpi_power_resources_list_free(struct list_head *list);
 int acpi_extract_power_resources(union acpi_object *package, unsigned int start,
 				 struct list_head *list);
@@ -175,6 +175,12 @@ bool processor_physically_present(acpi_handle handle);
 static inline void acpi_early_processor_control_setup(void) {}
 #endif
 
+#ifdef CONFIG_ACPI_PROCESSOR_CSTATE
+void acpi_idle_rescan_dead_smt_siblings(void);
+#else
+static inline void acpi_idle_rescan_dead_smt_siblings(void) {}
+#endif
+
 /* --------------------------------------------------------------------------
                                   Embedded Controller
    -------------------------------------------------------------------------- */
diff --git a/drivers/acpi/irq.c b/drivers/acpi/irq.c
index 76a856c32c4d..d1595156c86a 100644
--- a/drivers/acpi/irq.c
+++ b/drivers/acpi/irq.c
@@ -300,6 +300,25 @@ int acpi_irq_get(acpi_handle handle, unsigned int index, struct resource *res)
 }
 EXPORT_SYMBOL_GPL(acpi_irq_get);
 
+const struct cpumask *acpi_irq_get_affinity(acpi_handle handle,
+					    unsigned int index)
+{
+	struct irq_fwspec_info info;
+	struct irq_fwspec fwspec;
+	unsigned long flags;
+
+	if (acpi_irq_parse_one(handle, index, &fwspec, &flags))
+		return NULL;
+
+	if (irq_populate_fwspec_info(&fwspec, &info))
+		return NULL;
+
+	if (!(info.flags & IRQ_FWSPEC_INFO_AFFINITY_VALID))
+		return NULL;
+
+	return info.affinity;
+}
+
 /**
  * acpi_set_irq_model - Setup the GSI irqdomain information
  * @model: the value assigned to acpi_irq_model
diff --git a/drivers/acpi/nfit/core.c b/drivers/acpi/nfit/core.c
index ae035b93da08..3eb56b77cb6d 100644
--- a/drivers/acpi/nfit/core.c
+++ b/drivers/acpi/nfit/core.c
@@ -2637,7 +2637,7 @@ static int acpi_nfit_register_region(struct acpi_nfit_desc *acpi_desc,
 	if (ndr_desc->target_node == NUMA_NO_NODE) {
 		ndr_desc->target_node = phys_to_target_node(spa->address);
 		dev_info(acpi_desc->dev, "changing target node from %d to %d for nfit region [%pa-%pa]",
-			NUMA_NO_NODE, ndr_desc->numa_node, &res.start, &res.end);
+			NUMA_NO_NODE, ndr_desc->target_node, &res.start, &res.end);
 	}
 
 	/*
diff --git a/drivers/acpi/nfit/intel.c b/drivers/acpi/nfit/intel.c
index 3902759abcba..bce6f6a18426 100644
--- a/drivers/acpi/nfit/intel.c
+++ b/drivers/acpi/nfit/intel.c
@@ -55,10 +55,9 @@ static unsigned long intel_security_flags(struct nvdimm *nvdimm,
 {
 	struct nfit_mem *nfit_mem = nvdimm_provider_data(nvdimm);
 	unsigned long security_flags = 0;
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_get_security_state cmd;
-	} nd_cmd = {
+	) nd_cmd = {
 		.pkg = {
 			.nd_command = NVDIMM_INTEL_GET_SECURITY_STATE,
 			.nd_family = NVDIMM_FAMILY_INTEL,
@@ -120,10 +119,9 @@ static unsigned long intel_security_flags(struct nvdimm *nvdimm,
 static int intel_security_freeze(struct nvdimm *nvdimm)
 {
 	struct nfit_mem *nfit_mem = nvdimm_provider_data(nvdimm);
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_freeze_lock cmd;
-	} nd_cmd = {
+	) nd_cmd = {
 		.pkg = {
 			.nd_command = NVDIMM_INTEL_FREEZE_LOCK,
 			.nd_family = NVDIMM_FAMILY_INTEL,
@@ -153,10 +151,9 @@ static int intel_security_change_key(struct nvdimm *nvdimm,
 	unsigned int cmd = ptype == NVDIMM_MASTER ?
 		NVDIMM_INTEL_SET_MASTER_PASSPHRASE :
 		NVDIMM_INTEL_SET_PASSPHRASE;
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_set_passphrase cmd;
-	} nd_cmd = {
+	) nd_cmd = {
 		.pkg = {
 			.nd_family = NVDIMM_FAMILY_INTEL,
 			.nd_size_in = ND_INTEL_PASSPHRASE_SIZE * 2,
@@ -195,10 +192,9 @@ static int __maybe_unused intel_security_unlock(struct nvdimm *nvdimm,
 		const struct nvdimm_key_data *key_data)
 {
 	struct nfit_mem *nfit_mem = nvdimm_provider_data(nvdimm);
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_unlock_unit cmd;
-	} nd_cmd = {
+	) nd_cmd = {
 		.pkg = {
 			.nd_command = NVDIMM_INTEL_UNLOCK_UNIT,
 			.nd_family = NVDIMM_FAMILY_INTEL,
@@ -234,10 +230,9 @@ static int intel_security_disable(struct nvdimm *nvdimm,
 {
 	int rc;
 	struct nfit_mem *nfit_mem = nvdimm_provider_data(nvdimm);
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_disable_passphrase cmd;
-	} nd_cmd = {
+	) nd_cmd = {
 		.pkg = {
 			.nd_command = NVDIMM_INTEL_DISABLE_PASSPHRASE,
 			.nd_family = NVDIMM_FAMILY_INTEL,
@@ -277,10 +272,9 @@ static int __maybe_unused intel_security_erase(struct nvdimm *nvdimm,
 	struct nfit_mem *nfit_mem = nvdimm_provider_data(nvdimm);
 	unsigned int cmd = ptype == NVDIMM_MASTER ?
 		NVDIMM_INTEL_MASTER_SECURE_ERASE : NVDIMM_INTEL_SECURE_ERASE;
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_secure_erase cmd;
-	} nd_cmd = {
+	) nd_cmd = {
 		.pkg = {
 			.nd_family = NVDIMM_FAMILY_INTEL,
 			.nd_size_in = ND_INTEL_PASSPHRASE_SIZE,
@@ -318,10 +312,9 @@ static int __maybe_unused intel_security_query_overwrite(struct nvdimm *nvdimm)
 {
 	int rc;
 	struct nfit_mem *nfit_mem = nvdimm_provider_data(nvdimm);
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_query_overwrite cmd;
-	} nd_cmd = {
+	) nd_cmd = {
 		.pkg = {
 			.nd_command = NVDIMM_INTEL_QUERY_OVERWRITE,
 			.nd_family = NVDIMM_FAMILY_INTEL,
@@ -354,10 +347,9 @@ static int __maybe_unused intel_security_overwrite(struct nvdimm *nvdimm,
 {
 	int rc;
 	struct nfit_mem *nfit_mem = nvdimm_provider_data(nvdimm);
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_overwrite cmd;
-	} nd_cmd = {
+	) nd_cmd = {
 		.pkg = {
 			.nd_command = NVDIMM_INTEL_OVERWRITE,
 			.nd_family = NVDIMM_FAMILY_INTEL,
@@ -407,10 +399,9 @@ const struct nvdimm_security_ops *intel_security_ops = &__intel_security_ops;
 static int intel_bus_fwa_businfo(struct nvdimm_bus_descriptor *nd_desc,
 		struct nd_intel_bus_fw_activate_businfo *info)
 {
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_bus_fw_activate_businfo cmd;
-	} nd_cmd = {
+	) nd_cmd = {
 		.pkg = {
 			.nd_command = NVDIMM_BUS_INTEL_FW_ACTIVATE_BUSINFO,
 			.nd_family = NVDIMM_BUS_FAMILY_INTEL,
@@ -518,33 +509,31 @@ static enum nvdimm_fwa_capability intel_bus_fwa_capability(
 static int intel_bus_fwa_activate(struct nvdimm_bus_descriptor *nd_desc)
 {
 	struct acpi_nfit_desc *acpi_desc = to_acpi_desc(nd_desc);
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_bus_fw_activate cmd;
-	} nd_cmd = {
-		.pkg = {
-			.nd_command = NVDIMM_BUS_INTEL_FW_ACTIVATE,
-			.nd_family = NVDIMM_BUS_FAMILY_INTEL,
-			.nd_size_in = sizeof(nd_cmd.cmd.iodev_state),
-			.nd_size_out =
-				sizeof(struct nd_intel_bus_fw_activate),
-			.nd_fw_size =
-				sizeof(struct nd_intel_bus_fw_activate),
-		},
+	) nd_cmd;
+	int rc;
+
+	nd_cmd.pkg = (struct nd_cmd_pkg) {
+		.nd_command = NVDIMM_BUS_INTEL_FW_ACTIVATE,
+		.nd_family = NVDIMM_BUS_FAMILY_INTEL,
+		.nd_size_in = sizeof(nd_cmd.cmd.iodev_state),
+		.nd_size_out =
+			sizeof(struct nd_intel_bus_fw_activate),
+		.nd_fw_size =
+			sizeof(struct nd_intel_bus_fw_activate),
+	};
+	nd_cmd.cmd = (struct nd_intel_bus_fw_activate) {
 		/*
 		 * Even though activate is run from a suspended context,
 		 * for safety, still ask platform firmware to force
 		 * quiesce devices by default. Let a module
 		 * parameter override that policy.
 		 */
-		.cmd = {
-			.iodev_state = acpi_desc->fwa_noidle
-				? ND_INTEL_BUS_FWA_IODEV_OS_IDLE
-				: ND_INTEL_BUS_FWA_IODEV_FORCE_IDLE,
-		},
+		.iodev_state = acpi_desc->fwa_noidle
+			? ND_INTEL_BUS_FWA_IODEV_OS_IDLE
+			: ND_INTEL_BUS_FWA_IODEV_FORCE_IDLE,
 	};
-	int rc;
-
 	switch (intel_bus_fwa_state(nd_desc)) {
 	case NVDIMM_FWA_ARMED:
 	case NVDIMM_FWA_ARM_OVERFLOW:
@@ -582,10 +571,9 @@ const struct nvdimm_bus_fw_ops *intel_bus_fw_ops = &__intel_bus_fw_ops;
 static int intel_fwa_dimminfo(struct nvdimm *nvdimm,
 		struct nd_intel_fw_activate_dimminfo *info)
 {
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_fw_activate_dimminfo cmd;
-	} nd_cmd = {
+	) nd_cmd = {
 		.pkg = {
 			.nd_command = NVDIMM_INTEL_FW_ACTIVATE_DIMMINFO,
 			.nd_family = NVDIMM_FAMILY_INTEL,
@@ -688,27 +676,24 @@ static int intel_fwa_arm(struct nvdimm *nvdimm, enum nvdimm_fwa_trigger arm)
 {
 	struct nfit_mem *nfit_mem = nvdimm_provider_data(nvdimm);
 	struct acpi_nfit_desc *acpi_desc = nfit_mem->acpi_desc;
-	struct {
-		struct nd_cmd_pkg pkg;
+	TRAILING_OVERLAP(struct nd_cmd_pkg, pkg, nd_payload,
 		struct nd_intel_fw_activate_arm cmd;
-	} nd_cmd = {
-		.pkg = {
-			.nd_command = NVDIMM_INTEL_FW_ACTIVATE_ARM,
-			.nd_family = NVDIMM_FAMILY_INTEL,
-			.nd_size_in = sizeof(nd_cmd.cmd.activate_arm),
-			.nd_size_out =
-				sizeof(struct nd_intel_fw_activate_arm),
-			.nd_fw_size =
-				sizeof(struct nd_intel_fw_activate_arm),
-		},
-		.cmd = {
-			.activate_arm = arm == NVDIMM_FWA_ARM
-				? ND_INTEL_DIMM_FWA_ARM
-				: ND_INTEL_DIMM_FWA_DISARM,
-		},
-	};
+	) nd_cmd;
 	int rc;
 
+	nd_cmd.pkg = (struct nd_cmd_pkg) {
+		.nd_command = NVDIMM_INTEL_FW_ACTIVATE_ARM,
+		.nd_family = NVDIMM_FAMILY_INTEL,
+		.nd_size_in = sizeof(nd_cmd.cmd.activate_arm),
+		.nd_size_out = sizeof(struct nd_intel_fw_activate_arm),
+		.nd_fw_size = sizeof(struct nd_intel_fw_activate_arm),
+	};
+	nd_cmd.cmd = (struct nd_intel_fw_activate_arm) {
+		.activate_arm = arm == NVDIMM_FWA_ARM ?
+					ND_INTEL_DIMM_FWA_ARM :
+					ND_INTEL_DIMM_FWA_DISARM,
+	};
+
 	switch (intel_fwa_state(nvdimm)) {
 	case NVDIMM_FWA_INVALID:
 		return -ENXIO;
diff --git a/drivers/acpi/numa/hmat.c b/drivers/acpi/numa/hmat.c
index 9d9052258e92..77a81627aaef 100644
--- a/drivers/acpi/numa/hmat.c
+++ b/drivers/acpi/numa/hmat.c
@@ -74,7 +74,6 @@ struct memory_target {
 	struct node_cache_attrs cache_attrs;
 	u8 gen_port_device_handle[ACPI_SRAT_DEVICE_HANDLE_SIZE];
 	bool registered;
-	bool ext_updated;	/* externally updated */
 };
 
 struct memory_initiator {
@@ -368,35 +367,6 @@ static void hmat_update_target_access(struct memory_target *target,
 	}
 }
 
-int hmat_update_target_coordinates(int nid, struct access_coordinate *coord,
-				   enum access_coordinate_class access)
-{
-	struct memory_target *target;
-	int pxm;
-
-	if (nid == NUMA_NO_NODE)
-		return -EINVAL;
-
-	pxm = node_to_pxm(nid);
-	guard(mutex)(&target_lock);
-	target = find_mem_target(pxm);
-	if (!target)
-		return -ENODEV;
-
-	hmat_update_target_access(target, ACPI_HMAT_READ_LATENCY,
-				  coord->read_latency, access);
-	hmat_update_target_access(target, ACPI_HMAT_WRITE_LATENCY,
-				  coord->write_latency, access);
-	hmat_update_target_access(target, ACPI_HMAT_READ_BANDWIDTH,
-				  coord->read_bandwidth, access);
-	hmat_update_target_access(target, ACPI_HMAT_WRITE_BANDWIDTH,
-				  coord->write_bandwidth, access);
-	target->ext_updated = true;
-
-	return 0;
-}
-EXPORT_SYMBOL_GPL(hmat_update_target_coordinates);
-
 static __init void hmat_add_locality(struct acpi_hmat_locality *hmat_loc)
 {
 	struct memory_locality *loc;
@@ -773,10 +743,6 @@ static void hmat_update_target_attrs(struct memory_target *target,
 	u32 best = 0;
 	int i;
 
-	/* Don't update if an external agent has changed the data.  */
-	if (target->ext_updated)
-		return;
-
 	/* Don't update for generic port if there's no device handle */
 	if ((access == NODE_ACCESS_CLASS_GENPORT_SINK_LOCAL ||
 	     access == NODE_ACCESS_CLASS_GENPORT_SINK_CPU) &&
@@ -908,11 +874,33 @@ static void hmat_register_target_devices(struct memory_target *target)
 	}
 }
 
-static void hmat_register_target(struct memory_target *target)
+static void hmat_hotplug_target(struct memory_target *target)
 {
 	int nid = pxm_to_node(target->memory_pxm);
 
 	/*
+	 * Skip offline nodes. This can happen when memory marked EFI_MEMORY_SP,
+	 * "specific purpose", is applied to all the memory in a proximity
+	 * domain leading to * the node being marked offline / unplugged, or if
+	 * memory-only "hotplug" node is offline.
+	 */
+	if (nid == NUMA_NO_NODE || !node_online(nid))
+		return;
+
+	guard(mutex)(&target_lock);
+	if (target->registered)
+		return;
+
+	hmat_register_target_initiators(target);
+	hmat_register_target_cache(target);
+	hmat_register_target_perf(target, ACCESS_COORDINATE_LOCAL);
+	hmat_register_target_perf(target, ACCESS_COORDINATE_CPU);
+	target->registered = true;
+}
+
+static void hmat_register_target(struct memory_target *target)
+{
+	/*
 	 * Devices may belong to either an offline or online
 	 * node, so unconditionally add them.
 	 */
@@ -922,32 +910,15 @@ static void hmat_register_target(struct memory_target *target)
 	 * Register generic port perf numbers. The nid may not be
 	 * initialized and is still NUMA_NO_NODE.
 	 */
-	mutex_lock(&target_lock);
-	if (*(u16 *)target->gen_port_device_handle) {
-		hmat_update_generic_target(target);
-		target->registered = true;
+	scoped_guard(mutex, &target_lock) {
+		if (*(u16 *)target->gen_port_device_handle) {
+			hmat_update_generic_target(target);
+			target->registered = true;
+			return;
+		}
 	}
-	mutex_unlock(&target_lock);
-
-	/*
-	 * Skip offline nodes. This can happen when memory
-	 * marked EFI_MEMORY_SP, "specific purpose", is applied
-	 * to all the memory in a proximity domain leading to
-	 * the node being marked offline / unplugged, or if
-	 * memory-only "hotplug" node is offline.
-	 */
-	if (nid == NUMA_NO_NODE || !node_online(nid))
-		return;
 
-	mutex_lock(&target_lock);
-	if (!target->registered) {
-		hmat_register_target_initiators(target);
-		hmat_register_target_cache(target);
-		hmat_register_target_perf(target, ACCESS_COORDINATE_LOCAL);
-		hmat_register_target_perf(target, ACCESS_COORDINATE_CPU);
-		target->registered = true;
-	}
-	mutex_unlock(&target_lock);
+	hmat_hotplug_target(target);
 }
 
 static void hmat_register_targets(void)
@@ -962,10 +933,10 @@ static int hmat_callback(struct notifier_block *self,
 			 unsigned long action, void *arg)
 {
 	struct memory_target *target;
-	struct memory_notify *mnb = arg;
-	int pxm, nid = mnb->status_change_nid;
+	struct node_notify *nn = arg;
+	int pxm, nid = nn->nid;
 
-	if (nid == NUMA_NO_NODE || action != MEM_ONLINE)
+	if (action != NODE_ADDED_FIRST_MEMORY)
 		return NOTIFY_OK;
 
 	pxm = node_to_pxm(nid);
@@ -973,7 +944,7 @@ static int hmat_callback(struct notifier_block *self,
 	if (!target)
 		return NOTIFY_OK;
 
-	hmat_register_target(target);
+	hmat_hotplug_target(target);
 	return NOTIFY_OK;
 }
 
@@ -1118,7 +1089,7 @@ static __init int hmat_init(void)
 	hmat_register_targets();
 
 	/* Keep the table and structures if the notifier may use them */
-	if (hotplug_memory_notifier(hmat_callback, HMAT_CALLBACK_PRI))
+	if (hotplug_node_notifier(hmat_callback, HMAT_CALLBACK_PRI))
 		goto out_put;
 
 	if (!hmat_set_default_dram_perf())
diff --git a/drivers/acpi/numa/srat.c b/drivers/acpi/numa/srat.c
index 53816dfab645..aa87ee1583a4 100644
--- a/drivers/acpi/numa/srat.c
+++ b/drivers/acpi/numa/srat.c
@@ -237,7 +237,7 @@ acpi_table_print_srat_entry(struct acpi_subtable_header *header)
 		struct acpi_srat_generic_affinity *p =
 			(struct acpi_srat_generic_affinity *)header;
 
-		if (p->device_handle_type == 0) {
+		if (p->device_handle_type == 1) {
 			/*
 			 * For pci devices this may be the only place they
 			 * are assigned a proximity domain
diff --git a/drivers/acpi/osl.c b/drivers/acpi/osl.c
index 5ff343096ece..05393a7315fe 100644
--- a/drivers/acpi/osl.c
+++ b/drivers/acpi/osl.c
@@ -398,7 +398,7 @@ static void acpi_os_drop_map_ref(struct acpi_ioremap *map)
 	list_del_rcu(&map->list);
 
 	INIT_RCU_WORK(&map->track.rwork, acpi_os_map_remove);
-	queue_rcu_work(system_wq, &map->track.rwork);
+	queue_rcu_work(system_percpu_wq, &map->track.rwork);
 }
 
 /**
@@ -1694,8 +1694,8 @@ acpi_status __init acpi_os_initialize(void)
 
 acpi_status __init acpi_os_initialize1(void)
 {
-	kacpid_wq = alloc_workqueue("kacpid", 0, 1);
-	kacpi_notify_wq = alloc_workqueue("kacpi_notify", 0, 0);
+	kacpid_wq = alloc_workqueue("kacpid", WQ_PERCPU, 1);
+	kacpi_notify_wq = alloc_workqueue("kacpi_notify", WQ_PERCPU, 0);
 	kacpi_hotplug_wq = alloc_ordered_workqueue("kacpi_hotplug", 0);
 	BUG_ON(!kacpid_wq);
 	BUG_ON(!kacpi_notify_wq);
diff --git a/drivers/acpi/pci_irq.c b/drivers/acpi/pci_irq.c
index 630fe0a34bc6..ad81aa03fe2f 100644
--- a/drivers/acpi/pci_irq.c
+++ b/drivers/acpi/pci_irq.c
@@ -22,6 +22,7 @@
 #include <linux/acpi.h>
 #include <linux/slab.h>
 #include <linux/interrupt.h>
+#include <linux/string_choices.h>
 
 struct acpi_prt_entry {
 	struct acpi_pci_id	id;
@@ -468,7 +469,7 @@ int acpi_pci_irq_enable(struct pci_dev *dev)
 	dev_dbg(&dev->dev, "PCI INT %c%s -> GSI %u (%s, %s) -> IRQ %d\n",
 		pin_name(pin), link_desc, gsi,
 		(triggering == ACPI_LEVEL_SENSITIVE) ? "level" : "edge",
-		(polarity == ACPI_ACTIVE_LOW) ? "low" : "high", dev->irq);
+		str_low_high(polarity == ACPI_ACTIVE_LOW), dev->irq);
 
 	kfree(entry);
 	return 0;
diff --git a/drivers/acpi/pci_link.c b/drivers/acpi/pci_link.c
index 08e10b6226dc..bed7dc85612e 100644
--- a/drivers/acpi/pci_link.c
+++ b/drivers/acpi/pci_link.c
@@ -268,7 +268,7 @@ static int acpi_pci_link_get_current(struct acpi_pci_link *link)
 
 	link->irq.active = irq;
 
-	acpi_handle_debug(handle, "Link at IRQ %d \n", link->irq.active);
+	acpi_handle_debug(handle, "Link at IRQ %d\n", link->irq.active);
 
       end:
 	return result;
@@ -761,7 +761,7 @@ static int acpi_pci_link_resume(struct acpi_pci_link *link)
 	return 0;
 }
 
-static void irqrouter_resume(void)
+static void irqrouter_resume(void *data)
 {
 	struct acpi_pci_link *link;
 
@@ -888,10 +888,14 @@ static int __init acpi_irq_balance_set(char *str)
 
 __setup("acpi_irq_balance", acpi_irq_balance_set);
 
-static struct syscore_ops irqrouter_syscore_ops = {
+static const struct syscore_ops irqrouter_syscore_ops = {
 	.resume = irqrouter_resume,
 };
 
+static struct syscore irqrouter_syscore = {
+	.ops = &irqrouter_syscore_ops,
+};
+
 void __init acpi_pci_link_init(void)
 {
 	if (acpi_noirq)
@@ -904,6 +908,6 @@ void __init acpi_pci_link_init(void)
 		else
 			acpi_irq_balance = 0;
 	}
-	register_syscore_ops(&irqrouter_syscore_ops);
+	register_syscore(&irqrouter_syscore);
 	acpi_scan_add_handler(&pci_link_handler);
 }
diff --git a/drivers/acpi/pfr_update.c b/drivers/acpi/pfr_update.c
index 031d1ba81b86..11b1c2828005 100644
--- a/drivers/acpi/pfr_update.c
+++ b/drivers/acpi/pfr_update.c
@@ -127,8 +127,11 @@ static int query_capability(struct pfru_update_cap_info *cap_hdr,
 					  pfru_dev->rev_id,
 					  PFRU_FUNC_QUERY_UPDATE_CAP,
 					  NULL, ACPI_TYPE_PACKAGE);
-	if (!out_obj)
+	if (!out_obj) {
+		dev_dbg(pfru_dev->parent_dev,
+			"Query cap failed with no object\n");
 		return ret;
+	}
 
 	if (out_obj->package.count < CAP_NR_IDX ||
 	    out_obj->package.elements[CAP_STATUS_IDX].type != ACPI_TYPE_INTEGER ||
@@ -141,13 +144,17 @@ static int query_capability(struct pfru_update_cap_info *cap_hdr,
 	    out_obj->package.elements[CAP_DRV_SVN_IDX].type != ACPI_TYPE_INTEGER ||
 	    out_obj->package.elements[CAP_PLAT_ID_IDX].type != ACPI_TYPE_BUFFER ||
 	    out_obj->package.elements[CAP_OEM_ID_IDX].type != ACPI_TYPE_BUFFER ||
-	    out_obj->package.elements[CAP_OEM_INFO_IDX].type != ACPI_TYPE_BUFFER)
+	    out_obj->package.elements[CAP_OEM_INFO_IDX].type != ACPI_TYPE_BUFFER) {
+		dev_dbg(pfru_dev->parent_dev,
+			"Query cap failed with invalid package count/type\n");
 		goto free_acpi_buffer;
+	}
 
 	cap_hdr->status = out_obj->package.elements[CAP_STATUS_IDX].integer.value;
 	if (cap_hdr->status != DSM_SUCCEED) {
 		ret = -EBUSY;
-		dev_dbg(pfru_dev->parent_dev, "Error Status:%d\n", cap_hdr->status);
+		dev_dbg(pfru_dev->parent_dev, "Query cap Error Status:%d\n",
+			cap_hdr->status);
 		goto free_acpi_buffer;
 	}
 
@@ -193,24 +200,32 @@ static int query_buffer(struct pfru_com_buf_info *info,
 	out_obj = acpi_evaluate_dsm_typed(handle, &pfru_guid,
 					  pfru_dev->rev_id, PFRU_FUNC_QUERY_BUF,
 					  NULL, ACPI_TYPE_PACKAGE);
-	if (!out_obj)
+	if (!out_obj) {
+		dev_dbg(pfru_dev->parent_dev,
+			"Query buf failed with no object\n");
 		return ret;
+	}
 
 	if (out_obj->package.count < BUF_NR_IDX ||
 	    out_obj->package.elements[BUF_STATUS_IDX].type != ACPI_TYPE_INTEGER ||
 	    out_obj->package.elements[BUF_EXT_STATUS_IDX].type != ACPI_TYPE_INTEGER ||
 	    out_obj->package.elements[BUF_ADDR_LOW_IDX].type != ACPI_TYPE_INTEGER ||
 	    out_obj->package.elements[BUF_ADDR_HI_IDX].type != ACPI_TYPE_INTEGER ||
-	    out_obj->package.elements[BUF_SIZE_IDX].type != ACPI_TYPE_INTEGER)
+	    out_obj->package.elements[BUF_SIZE_IDX].type != ACPI_TYPE_INTEGER) {
+		dev_dbg(pfru_dev->parent_dev,
+			"Query buf failed with invalid package count/type\n");
 		goto free_acpi_buffer;
+	}
 
 	info->status = out_obj->package.elements[BUF_STATUS_IDX].integer.value;
 	info->ext_status =
 		out_obj->package.elements[BUF_EXT_STATUS_IDX].integer.value;
 	if (info->status != DSM_SUCCEED) {
 		ret = -EBUSY;
-		dev_dbg(pfru_dev->parent_dev, "Error Status:%d\n", info->status);
-		dev_dbg(pfru_dev->parent_dev, "Error Extended Status:%d\n", info->ext_status);
+		dev_dbg(pfru_dev->parent_dev,
+			"Query buf failed with Error Status:%d\n", info->status);
+		dev_dbg(pfru_dev->parent_dev,
+			"Query buf failed with Error Extended Status:%d\n", info->ext_status);
 
 		goto free_acpi_buffer;
 	}
@@ -295,12 +310,16 @@ static bool applicable_image(const void *data, struct pfru_update_cap_info *cap,
 	m_img_hdr = data + size;
 
 	type = get_image_type(m_img_hdr, pfru_dev);
-	if (type < 0)
+	if (type < 0) {
+		dev_dbg(pfru_dev->parent_dev, "Invalid image type\n");
 		return false;
+	}
 
 	size = adjust_efi_size(m_img_hdr, size);
-	if (size < 0)
+	if (size < 0) {
+		dev_dbg(pfru_dev->parent_dev, "Invalid image size\n");
 		return false;
+	}
 
 	auth = data + size;
 	size += sizeof(u64) + auth->auth_info.hdr.len;
@@ -310,7 +329,7 @@ static bool applicable_image(const void *data, struct pfru_update_cap_info *cap,
 	if (type == PFRU_CODE_INJECT_TYPE)
 		return payload_hdr->rt_ver >= cap->code_rt_version;
 
-	return payload_hdr->rt_ver >= cap->drv_rt_version;
+	return payload_hdr->svn_ver >= cap->drv_svn;
 }
 
 static void print_update_debug_info(struct pfru_updated_result *result,
@@ -346,8 +365,11 @@ static int start_update(int action, struct pfru_device *pfru_dev)
 	out_obj = acpi_evaluate_dsm_typed(handle, &pfru_guid,
 					  pfru_dev->rev_id, PFRU_FUNC_START,
 					  &in_obj, ACPI_TYPE_PACKAGE);
-	if (!out_obj)
+	if (!out_obj) {
+		dev_dbg(pfru_dev->parent_dev,
+			"Update failed to start with no object\n");
 		return ret;
+	}
 
 	if (out_obj->package.count < UPDATE_NR_IDX ||
 	    out_obj->package.elements[UPDATE_STATUS_IDX].type != ACPI_TYPE_INTEGER ||
@@ -355,8 +377,11 @@ static int start_update(int action, struct pfru_device *pfru_dev)
 	    out_obj->package.elements[UPDATE_AUTH_TIME_LOW_IDX].type != ACPI_TYPE_INTEGER ||
 	    out_obj->package.elements[UPDATE_AUTH_TIME_HI_IDX].type != ACPI_TYPE_INTEGER ||
 	    out_obj->package.elements[UPDATE_EXEC_TIME_LOW_IDX].type != ACPI_TYPE_INTEGER ||
-	    out_obj->package.elements[UPDATE_EXEC_TIME_HI_IDX].type != ACPI_TYPE_INTEGER)
+	    out_obj->package.elements[UPDATE_EXEC_TIME_HI_IDX].type != ACPI_TYPE_INTEGER) {
+		dev_dbg(pfru_dev->parent_dev,
+			"Update failed with invalid package count/type\n");
 		goto free_acpi_buffer;
+	}
 
 	update_result.status =
 		out_obj->package.elements[UPDATE_STATUS_IDX].integer.value;
@@ -365,8 +390,10 @@ static int start_update(int action, struct pfru_device *pfru_dev)
 
 	if (update_result.status != DSM_SUCCEED) {
 		ret = -EBUSY;
-		dev_dbg(pfru_dev->parent_dev, "Error Status:%d\n", update_result.status);
-		dev_dbg(pfru_dev->parent_dev, "Error Extended Status:%d\n",
+		dev_dbg(pfru_dev->parent_dev,
+			"Update failed with Error Status:%d\n", update_result.status);
+		dev_dbg(pfru_dev->parent_dev,
+			"Update failed with Error Extended Status:%d\n",
 			update_result.ext_status);
 
 		goto free_acpi_buffer;
@@ -450,8 +477,10 @@ static ssize_t pfru_write(struct file *file, const char __user *buf,
 	if (ret)
 		return ret;
 
-	if (len > buf_info.buf_size)
+	if (len > buf_info.buf_size) {
+		dev_dbg(pfru_dev->parent_dev, "Capsule image size too large\n");
 		return -EINVAL;
+	}
 
 	iov.iov_base = (void __user *)buf;
 	iov.iov_len = len;
@@ -460,10 +489,14 @@ static ssize_t pfru_write(struct file *file, const char __user *buf,
 	/* map the communication buffer */
 	phy_addr = (phys_addr_t)((buf_info.addr_hi << 32) | buf_info.addr_lo);
 	buf_ptr = memremap(phy_addr, buf_info.buf_size, MEMREMAP_WB);
-	if (!buf_ptr)
+	if (!buf_ptr) {
+		dev_dbg(pfru_dev->parent_dev, "Failed to remap the buffer\n");
 		return -ENOMEM;
+	}
 
 	if (!copy_from_iter_full(buf_ptr, len, &iter)) {
+		dev_dbg(pfru_dev->parent_dev,
+			"Failed to copy the data from the user space buffer\n");
 		ret = -EINVAL;
 		goto unmap;
 	}
diff --git a/drivers/acpi/platform_profile.c b/drivers/acpi/platform_profile.c
index b43f4459a4f6..ea04a8c69215 100644
--- a/drivers/acpi/platform_profile.c
+++ b/drivers/acpi/platform_profile.c
@@ -37,6 +37,7 @@ static const char * const profile_names[] = {
 	[PLATFORM_PROFILE_BALANCED] = "balanced",
 	[PLATFORM_PROFILE_BALANCED_PERFORMANCE] = "balanced-performance",
 	[PLATFORM_PROFILE_PERFORMANCE] = "performance",
+	[PLATFORM_PROFILE_MAX_POWER] = "max-power",
 	[PLATFORM_PROFILE_CUSTOM] = "custom",
 };
 static_assert(ARRAY_SIZE(profile_names) == PLATFORM_PROFILE_LAST);
@@ -506,7 +507,8 @@ int platform_profile_cycle(void)
 		if (err)
 			return err;
 
-		if (profile == PLATFORM_PROFILE_CUSTOM ||
+		if (profile == PLATFORM_PROFILE_MAX_POWER ||
+		    profile == PLATFORM_PROFILE_CUSTOM ||
 		    profile == PLATFORM_PROFILE_LAST)
 			return -EINVAL;
 
@@ -515,7 +517,8 @@ int platform_profile_cycle(void)
 		if (err)
 			return err;
 
-		/* never iterate into a custom if all drivers supported it */
+		/* never iterate into a custom or max power if all drivers supported it */
+		clear_bit(PLATFORM_PROFILE_MAX_POWER, data.aggregate);
 		clear_bit(PLATFORM_PROFILE_CUSTOM, data.aggregate);
 
 		next = find_next_bit_wrap(data.aggregate,
diff --git a/drivers/acpi/power.c b/drivers/acpi/power.c
index b7243d7563b1..361a7721a6a8 100644
--- a/drivers/acpi/power.c
+++ b/drivers/acpi/power.c
@@ -23,6 +23,7 @@
 
 #define pr_fmt(fmt) "ACPI: PM: " fmt
 
+#include <linux/delay.h>
 #include <linux/dmi.h>
 #include <linux/kernel.h>
 #include <linux/module.h>
@@ -63,6 +64,9 @@ struct acpi_power_resource_entry {
 	struct acpi_power_resource *resource;
 };
 
+static bool hp_eb_gp12pxp_quirk;
+static bool unused_power_resources_quirk;
+
 static LIST_HEAD(acpi_power_resource_list);
 static DEFINE_MUTEX(power_resource_list_lock);
 
@@ -992,6 +996,38 @@ struct acpi_device *acpi_add_power_resource(acpi_handle handle)
 }
 
 #ifdef CONFIG_ACPI_SLEEP
+static bool resource_is_gp12pxp(acpi_handle handle)
+{
+	const char *path;
+	bool ret;
+
+	path = acpi_handle_path(handle);
+	ret = path && strcmp(path, "\\_SB_.PCI0.GP12.PXP_") == 0;
+	kfree(path);
+
+	return ret;
+}
+
+static void acpi_resume_on_eb_gp12pxp(struct acpi_power_resource *resource)
+{
+	acpi_handle_notice(resource->device.handle,
+			   "HP EB quirk - turning OFF then ON\n");
+
+	__acpi_power_off(resource);
+	__acpi_power_on(resource);
+
+	/*
+	 * Use the same delay as DSDT uses in modem _RST method.
+	 *
+	 * Otherwise we get "Unable to change power state from unknown to D0,
+	 * device inaccessible" error for the modem PCI device after thaw.
+	 *
+	 * This power resource is normally being enabled only during thaw (once)
+	 * so this wait is not a performance issue.
+	 */
+	msleep(200);
+}
+
 void acpi_resume_power_resources(void)
 {
 	struct acpi_power_resource *resource;
@@ -1013,8 +1049,14 @@ void acpi_resume_power_resources(void)
 
 		if (state == ACPI_POWER_RESOURCE_STATE_OFF
 		    && resource->ref_count) {
-			acpi_handle_debug(resource->device.handle, "Turning ON\n");
-			__acpi_power_on(resource);
+			if (hp_eb_gp12pxp_quirk &&
+			    resource_is_gp12pxp(resource->device.handle)) {
+				acpi_resume_on_eb_gp12pxp(resource);
+			} else {
+				acpi_handle_debug(resource->device.handle,
+						  "Turning ON\n");
+				__acpi_power_on(resource);
+			}
 		}
 
 		mutex_unlock(&resource->resource_lock);
@@ -1024,6 +1066,41 @@ void acpi_resume_power_resources(void)
 }
 #endif
 
+static const struct dmi_system_id dmi_hp_elitebook_gp12pxp_quirk[] = {
+/*
+ * This laptop (and possibly similar models too) has power resource called
+ * "GP12.PXP_" for its WWAN modem.
+ *
+ * For this power resource to turn ON power for the modem it needs certain
+ * internal flag called "ONEN" to be set.
+ * This flag only gets set from this power resource "_OFF" method, while the
+ * actual modem power gets turned off during suspend by "GP12.PTS" method
+ * called from the global "_PTS" (Prepare To Sleep) method.
+ * On the other hand, this power resource "_OFF" method implementation just
+ * sets the aforementioned flag without actually doing anything else (it
+ * doesn't contain any code to actually turn off power).
+ *
+ * The above means that when upon hibernation finish we try to set this
+ * power resource back ON since its "_STA" method returns 0 (while the resource
+ * is still considered in use) its "_ON" method won't do anything since
+ * that "ONEN" flag is not set.
+ * Overall, this means the modem is dead until laptop is rebooted since its
+ * power has been cut by "_PTS" and its PCI configuration was lost and not able
+ * to be restored.
+ *
+ * The easiest way to workaround the issue is to call this power resource
+ * "_OFF" method before calling the "_ON" method to make sure the "ONEN"
+ * flag gets properly set.
+ */
+	{
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "HP"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "HP EliteBook 855 G7 Notebook PC"),
+		},
+	},
+	{}
+};
+
 static const struct dmi_system_id dmi_leave_unused_power_resources_on[] = {
 	{
 		/*
@@ -1046,7 +1123,7 @@ void acpi_turn_off_unused_power_resources(void)
 {
 	struct acpi_power_resource *resource;
 
-	if (dmi_check_system(dmi_leave_unused_power_resources_on))
+	if (unused_power_resources_quirk)
 		return;
 
 	mutex_lock(&power_resource_list_lock);
@@ -1065,3 +1142,10 @@ void acpi_turn_off_unused_power_resources(void)
 
 	mutex_unlock(&power_resource_list_lock);
 }
+
+void __init acpi_power_resources_init(void)
+{
+	hp_eb_gp12pxp_quirk = dmi_check_system(dmi_hp_elitebook_gp12pxp_quirk);
+	unused_power_resources_quirk =
+		dmi_check_system(dmi_leave_unused_power_resources_on);
+}
diff --git a/drivers/acpi/pptt.c b/drivers/acpi/pptt.c
index 54676e3d82dd..de5f8c018333 100644
--- a/drivers/acpi/pptt.c
+++ b/drivers/acpi/pptt.c
@@ -21,6 +21,25 @@
 #include <linux/cacheinfo.h>
 #include <acpi/processor.h>
 
+/*
+ * The acpi_pptt_cache_v1 in actbl2.h, which is imported from acpica,
+ * only contains the cache_id field rather than all the fields of the
+ * Cache Type Structure. Use this alternative structure until it is
+ * resolved in acpica.
+ */
+struct acpi_pptt_cache_v1_full {
+	struct acpi_subtable_header header;
+	u16 reserved;
+	u32 flags;
+	u32 next_level_of_cache;
+	u32 size;
+	u32 number_of_sets;
+	u8 associativity;
+	u8 attributes;
+	u16 line_size;
+	u32 cache_id;
+} __packed;
+
 static struct acpi_subtable_header *fetch_pptt_subtable(struct acpi_table_header *table_hdr,
 							u32 pptt_ref)
 {
@@ -56,6 +75,18 @@ static struct acpi_pptt_cache *fetch_pptt_cache(struct acpi_table_header *table_
 	return (struct acpi_pptt_cache *)fetch_pptt_subtable(table_hdr, pptt_ref);
 }
 
+static struct acpi_pptt_cache_v1_full *upgrade_pptt_cache(struct acpi_pptt_cache *cache)
+{
+	if (cache->header.length < sizeof(struct acpi_pptt_cache_v1_full))
+		return NULL;
+
+	/* No use for v1 if the only additional field is invalid */
+	if (!(cache->flags & ACPI_PPTT_CACHE_ID_VALID))
+		return NULL;
+
+	return (struct acpi_pptt_cache_v1_full *)cache;
+}
+
 static struct acpi_subtable_header *acpi_get_pptt_resource(struct acpi_table_header *table_hdr,
 							   struct acpi_pptt_processor *node,
 							   int resource)
@@ -177,14 +208,14 @@ acpi_find_cache_level(struct acpi_table_header *table_hdr,
 }
 
 /**
- * acpi_count_levels() - Given a PPTT table, and a CPU node, count the cache
- * levels and split cache levels (data/instruction).
+ * acpi_count_levels() - Given a PPTT table, and a CPU node, count the
+ * total number of levels and split cache levels (data/instruction).
  * @table_hdr: Pointer to the head of the PPTT table
  * @cpu_node: processor node we wish to count caches for
- * @levels: Number of levels if success.
  * @split_levels:	Number of split cache levels (data/instruction) if
  *			success. Can by NULL.
  *
+ * Return: number of levels.
  * Given a processor node containing a processing unit, walk into it and count
  * how many levels exist solely for it, and then walk up each level until we hit
  * the root node (ignore the package level because it may be possible to have
@@ -192,14 +223,18 @@ acpi_find_cache_level(struct acpi_table_header *table_hdr,
  * split cache levels (data/instruction) that exist at each level on the way
  * up.
  */
-static void acpi_count_levels(struct acpi_table_header *table_hdr,
-			      struct acpi_pptt_processor *cpu_node,
-			      unsigned int *levels, unsigned int *split_levels)
+static int acpi_count_levels(struct acpi_table_header *table_hdr,
+			     struct acpi_pptt_processor *cpu_node,
+			     unsigned int *split_levels)
 {
+	int current_level = 0;
+
 	do {
-		acpi_find_cache_level(table_hdr, cpu_node, levels, split_levels, 0, 0);
+		acpi_find_cache_level(table_hdr, cpu_node, &current_level, split_levels, 0, 0);
 		cpu_node = fetch_pptt_node(table_hdr, cpu_node->parent);
 	} while (cpu_node);
+
+	return current_level;
 }
 
 /**
@@ -351,7 +386,6 @@ static struct acpi_pptt_cache *acpi_find_cache_node(struct acpi_table_header *ta
  * @this_leaf: Kernel cache info structure being updated
  * @found_cache: The PPTT node describing this cache instance
  * @cpu_node: A unique reference to describe this cache instance
- * @revision: The revision of the PPTT table
  *
  * The ACPI spec implies that the fields in the cache structures are used to
  * extend and correct the information probed from the hardware. Lets only
@@ -361,10 +395,9 @@ static struct acpi_pptt_cache *acpi_find_cache_node(struct acpi_table_header *ta
  */
 static void update_cache_properties(struct cacheinfo *this_leaf,
 				    struct acpi_pptt_cache *found_cache,
-				    struct acpi_pptt_processor *cpu_node,
-				    u8 revision)
+				    struct acpi_pptt_processor *cpu_node)
 {
-	struct acpi_pptt_cache_v1* found_cache_v1;
+	struct acpi_pptt_cache_v1_full *found_cache_v1;
 
 	this_leaf->fw_token = cpu_node;
 	if (found_cache->flags & ACPI_PPTT_SIZE_PROPERTY_VALID)
@@ -414,9 +447,8 @@ static void update_cache_properties(struct cacheinfo *this_leaf,
 	    found_cache->flags & ACPI_PPTT_CACHE_TYPE_VALID)
 		this_leaf->type = CACHE_TYPE_UNIFIED;
 
-	if (revision >= 3 && (found_cache->flags & ACPI_PPTT_CACHE_ID_VALID)) {
-		found_cache_v1 = ACPI_ADD_PTR(struct acpi_pptt_cache_v1,
-	                                      found_cache, sizeof(struct acpi_pptt_cache));
+	found_cache_v1 = upgrade_pptt_cache(found_cache);
+	if (found_cache_v1) {
 		this_leaf->id = found_cache_v1->cache_id;
 		this_leaf->attributes |= CACHE_ID;
 	}
@@ -441,8 +473,7 @@ static void cache_setup_acpi_cpu(struct acpi_table_header *table,
 		pr_debug("found = %p %p\n", found_cache, cpu_node);
 		if (found_cache)
 			update_cache_properties(this_leaf, found_cache,
-						ACPI_TO_POINTER(ACPI_PTR_DIFF(cpu_node, table)),
-						table->revision);
+						ACPI_TO_POINTER(ACPI_PTR_DIFF(cpu_node, table)));
 
 		index++;
 	}
@@ -645,7 +676,7 @@ int acpi_get_cache_info(unsigned int cpu, unsigned int *levels,
 	if (!cpu_node)
 		return -ENOENT;
 
-	acpi_count_levels(table, cpu_node, levels, split_levels);
+	*levels = acpi_count_levels(table, cpu_node, split_levels);
 
 	pr_debug("Cache Setup: last_level=%d split_levels=%d\n",
 		 *levels, split_levels ? *split_levels : -1);
@@ -817,3 +848,218 @@ int find_acpi_cpu_topology_hetero_id(unsigned int cpu)
 	return find_acpi_cpu_topology_tag(cpu, PPTT_ABORT_PACKAGE,
 					  ACPI_PPTT_ACPI_IDENTICAL);
 }
+
+/**
+ * acpi_pptt_get_child_cpus() - Find all the CPUs below a PPTT
+ * processor hierarchy node
+ *
+ * @table_hdr:		A reference to the PPTT table
+ * @parent_node:	A pointer to the processor hierarchy node in the
+ *			table_hdr
+ * @cpus:		A cpumask to fill with the CPUs below @parent_node
+ *
+ * Walks up the PPTT from every possible CPU to find if the provided
+ * @parent_node is a parent of this CPU.
+ */
+static void acpi_pptt_get_child_cpus(struct acpi_table_header *table_hdr,
+				     struct acpi_pptt_processor *parent_node,
+				     cpumask_t *cpus)
+{
+	struct acpi_pptt_processor *cpu_node;
+	u32 acpi_id;
+	int cpu;
+
+	cpumask_clear(cpus);
+
+	for_each_possible_cpu(cpu) {
+		acpi_id = get_acpi_id_for_cpu(cpu);
+		cpu_node = acpi_find_processor_node(table_hdr, acpi_id);
+
+		while (cpu_node) {
+			if (cpu_node == parent_node) {
+				cpumask_set_cpu(cpu, cpus);
+				break;
+			}
+			cpu_node = fetch_pptt_node(table_hdr, cpu_node->parent);
+		}
+	}
+}
+
+/**
+ * acpi_pptt_get_cpus_from_container() - Populate a cpumask with all CPUs in a
+ *                                       processor container
+ * @acpi_cpu_id:	The UID of the processor container
+ * @cpus:		The resulting CPU mask
+ *
+ * Find the specified Processor Container, and fill @cpus with all the cpus
+ * below it.
+ *
+ * Not all 'Processor Hierarchy' entries in the PPTT are either a CPU
+ * or a Processor Container, they may exist purely to describe a
+ * Private resource. CPUs have to be leaves, so a Processor Container
+ * is a non-leaf that has the 'ACPI Processor ID valid' flag set.
+ */
+void acpi_pptt_get_cpus_from_container(u32 acpi_cpu_id, cpumask_t *cpus)
+{
+	struct acpi_table_header *table_hdr;
+	struct acpi_subtable_header *entry;
+	unsigned long table_end;
+	u32 proc_sz;
+
+	cpumask_clear(cpus);
+
+	table_hdr = acpi_get_pptt();
+	if (!table_hdr)
+		return;
+
+	table_end = (unsigned long)table_hdr + table_hdr->length;
+	entry = ACPI_ADD_PTR(struct acpi_subtable_header, table_hdr,
+			     sizeof(struct acpi_table_pptt));
+	proc_sz = sizeof(struct acpi_pptt_processor);
+	while ((unsigned long)entry + proc_sz <= table_end) {
+		if (entry->type == ACPI_PPTT_TYPE_PROCESSOR) {
+			struct acpi_pptt_processor *cpu_node;
+
+			cpu_node = (struct acpi_pptt_processor *)entry;
+			if (cpu_node->flags & ACPI_PPTT_ACPI_PROCESSOR_ID_VALID &&
+			    !acpi_pptt_leaf_node(table_hdr, cpu_node) &&
+			    cpu_node->acpi_processor_id == acpi_cpu_id) {
+				acpi_pptt_get_child_cpus(table_hdr, cpu_node, cpus);
+				break;
+			}
+		}
+		entry = ACPI_ADD_PTR(struct acpi_subtable_header, entry,
+				     entry->length);
+	}
+}
+
+/**
+ * find_acpi_cache_level_from_id() - Get the level of the specified cache
+ * @cache_id: The id field of the cache
+ *
+ * Determine the level relative to any CPU for the cache identified by
+ * cache_id. This allows the property to be found even if the CPUs are offline.
+ *
+ * The returned level can be used to group caches that are peers.
+ *
+ * The PPTT table must be rev 3 or later.
+ *
+ * If one CPU's L2 is shared with another CPU as L3, this function will return
+ * an unpredictable value.
+ *
+ * Return: -ENOENT if the PPTT doesn't exist, the revision isn't supported or
+ * the cache cannot be found.
+ * Otherwise returns a value which represents the level of the specified cache.
+ */
+int find_acpi_cache_level_from_id(u32 cache_id)
+{
+	int cpu;
+	struct acpi_table_header *table;
+
+	table = acpi_get_pptt();
+	if (!table)
+		return -ENOENT;
+
+	if (table->revision < 3)
+		return -ENOENT;
+
+	for_each_possible_cpu(cpu) {
+		bool empty;
+		int level = 1;
+		u32 acpi_cpu_id = get_acpi_id_for_cpu(cpu);
+		struct acpi_pptt_cache *cache;
+		struct acpi_pptt_processor *cpu_node;
+
+		cpu_node = acpi_find_processor_node(table, acpi_cpu_id);
+		if (!cpu_node)
+			continue;
+
+		do {
+			int cache_type[] = {CACHE_TYPE_INST, CACHE_TYPE_DATA, CACHE_TYPE_UNIFIED};
+
+			empty = true;
+			for (int i = 0; i < ARRAY_SIZE(cache_type); i++) {
+				struct acpi_pptt_cache_v1_full *cache_v1;
+
+				cache = acpi_find_cache_node(table, acpi_cpu_id, cache_type[i],
+							     level, &cpu_node);
+				if (!cache)
+					continue;
+
+				empty = false;
+
+				cache_v1 = upgrade_pptt_cache(cache);
+				if (cache_v1 && cache_v1->cache_id == cache_id)
+					return level;
+			}
+			level++;
+		} while (!empty);
+	}
+
+	return -ENOENT;
+}
+
+/**
+ * acpi_pptt_get_cpumask_from_cache_id() - Get the cpus associated with the
+ *					   specified cache
+ * @cache_id: The id field of the cache
+ * @cpus: Where to build the cpumask
+ *
+ * Determine which CPUs are below this cache in the PPTT. This allows the property
+ * to be found even if the CPUs are offline.
+ *
+ * The PPTT table must be rev 3 or later,
+ *
+ * Return: -ENOENT if the PPTT doesn't exist, or the cache cannot be found.
+ * Otherwise returns 0 and sets the cpus in the provided cpumask.
+ */
+int acpi_pptt_get_cpumask_from_cache_id(u32 cache_id, cpumask_t *cpus)
+{
+	int cpu;
+	struct acpi_table_header *table;
+
+	cpumask_clear(cpus);
+
+	table = acpi_get_pptt();
+	if (!table)
+		return -ENOENT;
+
+	if (table->revision < 3)
+		return -ENOENT;
+
+	for_each_possible_cpu(cpu) {
+		bool empty;
+		int level = 1;
+		u32 acpi_cpu_id = get_acpi_id_for_cpu(cpu);
+		struct acpi_pptt_cache *cache;
+		struct acpi_pptt_processor *cpu_node;
+
+		cpu_node = acpi_find_processor_node(table, acpi_cpu_id);
+		if (!cpu_node)
+			continue;
+
+		do {
+			int cache_type[] = {CACHE_TYPE_INST, CACHE_TYPE_DATA, CACHE_TYPE_UNIFIED};
+
+			empty = true;
+			for (int i = 0; i < ARRAY_SIZE(cache_type); i++) {
+				struct acpi_pptt_cache_v1_full *cache_v1;
+
+				cache = acpi_find_cache_node(table, acpi_cpu_id, cache_type[i],
+							     level, &cpu_node);
+
+				if (!cache)
+					continue;
+
+				empty = false;
+
+				cache_v1 = upgrade_pptt_cache(cache);
+				if (cache_v1 && cache_v1->cache_id == cache_id)
+					cpumask_set_cpu(cpu, cpus);
+			}
+			level++;
+		} while (!empty);
+	}
+
+	return 0;
+}
diff --git a/drivers/acpi/prmt.c b/drivers/acpi/prmt.c
index e549914a636c..7b8b5d2015ec 100644
--- a/drivers/acpi/prmt.c
+++ b/drivers/acpi/prmt.c
@@ -85,8 +85,6 @@ static u64 efi_pa_va_lookup(efi_guid_t *guid, u64 pa)
 		}
 	}
 
-	pr_warn("Failed to find VA for GUID: %pUL, PA: 0x%llx", guid, pa);
-
 	return 0;
 }
 
@@ -152,15 +150,52 @@ acpi_parse_prmt(union acpi_subtable_headers *header, const unsigned long end)
 		th = &tm->handlers[cur_handler];
 
 		guid_copy(&th->guid, (guid_t *)handler_info->handler_guid);
+
+		/*
+		 * Print an error message if handler_address is NULL, the parse of VA also
+		 * can be skipped.
+		 */
+		if (unlikely(!handler_info->handler_address)) {
+			pr_info("Skipping handler with NULL address for GUID: %pUL",
+					(guid_t *)handler_info->handler_guid);
+			continue;
+		}
+
 		th->handler_addr =
 			(void *)efi_pa_va_lookup(&th->guid, handler_info->handler_address);
+		/*
+		 * Print a warning message and skip the parse of VA if handler_addr is zero
+		 * which is not expected to ever happen.
+		 */
+		if (unlikely(!th->handler_addr)) {
+			pr_warn("Failed to find VA of handler for GUID: %pUL, PA: 0x%llx",
+				&th->guid, handler_info->handler_address);
+			continue;
+		}
 
 		th->static_data_buffer_addr =
 			efi_pa_va_lookup(&th->guid, handler_info->static_data_buffer_address);
+		/*
+		 * According to the PRM specification, static_data_buffer_address can be zero,
+		 * so avoid printing a warning message in that case.  Otherwise, if the
+		 * return value of efi_pa_va_lookup() is zero, print the message.
+		 */
+		if (unlikely(!th->static_data_buffer_addr && handler_info->static_data_buffer_address))
+			pr_warn("Failed to find VA of static data buffer for GUID: %pUL, PA: 0x%llx",
+				&th->guid, handler_info->static_data_buffer_address);
 
 		th->acpi_param_buffer_addr =
 			efi_pa_va_lookup(&th->guid, handler_info->acpi_param_buffer_address);
 
+		/*
+		 * According to the PRM specification, acpi_param_buffer_address can be zero,
+		 * so avoid printing a warning message in that case.  Otherwise, if the
+		 * return value of efi_pa_va_lookup() is zero, print the message.
+		 */
+		if (unlikely(!th->acpi_param_buffer_addr && handler_info->acpi_param_buffer_address))
+			pr_warn("Failed to find VA of acpi param buffer for GUID: %pUL, PA: 0x%llx",
+				&th->guid, handler_info->acpi_param_buffer_address);
+
 	} while (++cur_handler < tm->handler_count && (handler_info = get_next_handler(handler_info)));
 
 	return 0;
@@ -209,6 +244,12 @@ static struct prm_handler_info *find_prm_handler(const guid_t *guid)
 	return (struct prm_handler_info *) find_guid_info(guid, GET_HANDLER);
 }
 
+bool acpi_prm_handler_available(const guid_t *guid)
+{
+	return find_prm_handler(guid) && find_prm_module(guid);
+}
+EXPORT_SYMBOL_GPL(acpi_prm_handler_available);
+
 /* In-coming PRM commands */
 
 #define PRM_CMD_RUN_SERVICE		0
diff --git a/drivers/acpi/proc.c b/drivers/acpi/proc.c
index 4322f2da6d10..c08ead07252b 100644
--- a/drivers/acpi/proc.c
+++ b/drivers/acpi/proc.c
@@ -1,7 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 #include <linux/proc_fs.h>
 #include <linux/seq_file.h>
-#include <linux/export.h>
+#include <linux/string_choices.h>
 #include <linux/suspend.h>
 #include <linux/bcd.h>
 #include <linux/acpi.h>
@@ -30,17 +30,16 @@ acpi_system_wakeup_device_seq_show(struct seq_file *seq, void *offset)
 		if (!dev->wakeup.flags.valid)
 			continue;
 
-		seq_printf(seq, "%s\t  S%d\t",
+		seq_printf(seq, "%s\t  S%llu\t",
 			   dev->pnp.bus_id,
-			   (u32) dev->wakeup.sleep_state);
+			   dev->wakeup.sleep_state);
 
 		mutex_lock(&dev->physical_node_lock);
 
 		if (!dev->physical_node_count) {
 			seq_printf(seq, "%c%-8s\n",
 				dev->wakeup.flags.valid ? '*' : ' ',
-				device_may_wakeup(&dev->dev) ?
-					"enabled" : "disabled");
+				str_enabled_disabled(device_may_wakeup(&dev->dev)));
 		} else {
 			struct device *ldev;
 			list_for_each_entry(entry, &dev->physical_node_list,
@@ -55,9 +54,8 @@ acpi_system_wakeup_device_seq_show(struct seq_file *seq, void *offset)
 
 				seq_printf(seq, "%c%-8s  %s:%s\n",
 					dev->wakeup.flags.valid ? '*' : ' ',
-					(device_may_wakeup(&dev->dev) ||
-					device_may_wakeup(ldev)) ?
-					"enabled" : "disabled",
+					str_enabled_disabled(device_may_wakeup(ldev) ||
+							     device_may_wakeup(&dev->dev)),
 					ldev->bus ? ldev->bus->name :
 					"no-bus", dev_name(ldev));
 				put_device(ldev);
@@ -141,6 +139,5 @@ static const struct proc_ops acpi_system_wakeup_device_proc_ops = {
 void __init acpi_sleep_proc_init(void)
 {
 	/* 'wakeup device' [R/W] */
-	proc_create("wakeup", S_IFREG | S_IRUGO | S_IWUSR,
-		    acpi_root_dir, &acpi_system_wakeup_device_proc_ops);
+	proc_create("wakeup", 0644, acpi_root_dir, &acpi_system_wakeup_device_proc_ops);
 }
diff --git a/drivers/acpi/processor_core.c b/drivers/acpi/processor_core.c
index 9b6b71a2ffb5..a4498357bd16 100644
--- a/drivers/acpi/processor_core.c
+++ b/drivers/acpi/processor_core.c
@@ -54,7 +54,7 @@ static int map_x2apic_id(struct acpi_subtable_header *entry,
 	if (!(apic->lapic_flags & ACPI_MADT_ENABLED))
 		return -ENODEV;
 
-	if (device_declaration && (apic->uid == acpi_id)) {
+	if (apic->uid == acpi_id && (device_declaration || acpi_id < 255)) {
 		*apic_id = apic->local_apic_id;
 		return 0;
 	}
diff --git a/drivers/acpi/processor_driver.c b/drivers/acpi/processor_driver.c
index 3b281bc1e73c..65e779be64ff 100644
--- a/drivers/acpi/processor_driver.c
+++ b/drivers/acpi/processor_driver.c
@@ -279,6 +279,9 @@ static int __init acpi_processor_driver_init(void)
 	 * after acpi_cppc_processor_probe() has been called for all online CPUs
 	 */
 	acpi_processor_init_invariance_cppc();
+
+	acpi_idle_rescan_dead_smt_siblings();
+
 	return 0;
 err:
 	driver_unregister(&acpi_processor_driver);
diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
index e2febca2ec13..89f2f08b2554 100644
--- a/drivers/acpi/processor_idle.c
+++ b/drivers/acpi/processor_idle.c
@@ -24,6 +24,8 @@
 #include <acpi/processor.h>
 #include <linux/context_tracking.h>
 
+#include "internal.h"
+
 /*
  * Include the apic definitions for x86 to have the APIC timer related defines
  * available also for UP (on SMP it gets magically included via linux/smp.h).
@@ -55,6 +57,12 @@ struct cpuidle_driver acpi_idle_driver = {
 };
 
 #ifdef CONFIG_ACPI_PROCESSOR_CSTATE
+void acpi_idle_rescan_dead_smt_siblings(void)
+{
+	if (cpuidle_get_driver() == &acpi_idle_driver)
+		arch_cpu_rescan_dead_smt_siblings();
+}
+
 static
 DEFINE_PER_CPU(struct acpi_processor_cx * [CPUIDLE_STATE_MAX], acpi_cstate);
 
@@ -724,18 +732,16 @@ static int __cpuidle acpi_idle_enter_s2idle(struct cpuidle_device *dev,
 	return 0;
 }
 
-static int acpi_processor_setup_cpuidle_cx(struct acpi_processor *pr,
-					   struct cpuidle_device *dev)
+static void acpi_processor_setup_cpuidle_cx(struct acpi_processor *pr,
+					    struct cpuidle_device *dev)
 {
 	int i, count = ACPI_IDLE_STATE_START;
 	struct acpi_processor_cx *cx;
-	struct cpuidle_state *state;
 
 	if (max_cstate == 0)
 		max_cstate = 1;
 
 	for (i = 1; i < ACPI_PROCESSOR_MAX_POWER && i <= max_cstate; i++) {
-		state = &acpi_idle_driver.states[count];
 		cx = &pr->power.states[i];
 
 		if (!cx->valid)
@@ -743,27 +749,13 @@ static int acpi_processor_setup_cpuidle_cx(struct acpi_processor *pr,
 
 		per_cpu(acpi_cstate[count], dev->cpu) = cx;
 
-		if (lapic_timer_needs_broadcast(pr, cx))
-			state->flags |= CPUIDLE_FLAG_TIMER_STOP;
-
-		if (cx->type == ACPI_STATE_C3) {
-			state->flags |= CPUIDLE_FLAG_TLB_FLUSHED;
-			if (pr->flags.bm_check)
-				state->flags |= CPUIDLE_FLAG_RCU_IDLE;
-		}
-
 		count++;
 		if (count == CPUIDLE_STATE_MAX)
 			break;
 	}
-
-	if (!count)
-		return -EINVAL;
-
-	return 0;
 }
 
-static int acpi_processor_setup_cstates(struct acpi_processor *pr)
+static void acpi_processor_setup_cstates(struct acpi_processor *pr)
 {
 	int i, count;
 	struct acpi_processor_cx *cx;
@@ -810,17 +802,21 @@ static int acpi_processor_setup_cstates(struct acpi_processor *pr)
 		if (cx->type != ACPI_STATE_C1 && !acpi_idle_fallback_to_c1(pr))
 			state->enter_s2idle = acpi_idle_enter_s2idle;
 
+		if (lapic_timer_needs_broadcast(pr, cx))
+			state->flags |= CPUIDLE_FLAG_TIMER_STOP;
+
+		if (cx->type == ACPI_STATE_C3) {
+			state->flags |= CPUIDLE_FLAG_TLB_FLUSHED;
+			if (pr->flags.bm_check)
+				state->flags |= CPUIDLE_FLAG_RCU_IDLE;
+		}
+
 		count++;
 		if (count == CPUIDLE_STATE_MAX)
 			break;
 	}
 
 	drv->state_count = count;
-
-	if (!count)
-		return -EINVAL;
-
-	return 0;
 }
 
 static inline void acpi_processor_cstate_first_run_checks(void)
@@ -990,11 +986,6 @@ end:
 	return ret;
 }
 
-/*
- * flat_state_cnt - the number of composite LPI states after the process of flattening
- */
-static int flat_state_cnt;
-
 /**
  * combine_lpi_states - combine local and parent LPI states to form a composite LPI state
  *
@@ -1037,9 +1028,10 @@ static void stash_composite_state(struct acpi_lpi_states_array *curr_level,
 	curr_level->composite_states[curr_level->composite_states_size++] = t;
 }
 
-static int flatten_lpi_states(struct acpi_processor *pr,
-			      struct acpi_lpi_states_array *curr_level,
-			      struct acpi_lpi_states_array *prev_level)
+static unsigned int flatten_lpi_states(struct acpi_processor *pr,
+				       unsigned int flat_state_cnt,
+				       struct acpi_lpi_states_array *curr_level,
+				       struct acpi_lpi_states_array *prev_level)
 {
 	int i, j, state_count = curr_level->size;
 	struct acpi_lpi_state *p, *t = curr_level->entries;
@@ -1079,7 +1071,7 @@ static int flatten_lpi_states(struct acpi_processor *pr,
 	}
 
 	kfree(curr_level->entries);
-	return 0;
+	return flat_state_cnt;
 }
 
 int __weak acpi_processor_ffh_lpi_probe(unsigned int cpu)
@@ -1094,6 +1086,7 @@ static int acpi_processor_get_lpi_info(struct acpi_processor *pr)
 	acpi_handle handle = pr->handle, pr_ahandle;
 	struct acpi_device *d = NULL;
 	struct acpi_lpi_states_array info[2], *tmp, *prev, *curr;
+	unsigned int state_count;
 
 	/* make sure our architecture has support */
 	ret = acpi_processor_ffh_lpi_probe(pr->id);
@@ -1106,14 +1099,13 @@ static int acpi_processor_get_lpi_info(struct acpi_processor *pr)
 	if (!acpi_has_method(handle, "_LPI"))
 		return -EINVAL;
 
-	flat_state_cnt = 0;
 	prev = &info[0];
 	curr = &info[1];
 	handle = pr->handle;
 	ret = acpi_processor_evaluate_lpi(handle, prev);
 	if (ret)
 		return ret;
-	flatten_lpi_states(pr, prev, NULL);
+	state_count = flatten_lpi_states(pr, 0, prev, NULL);
 
 	status = acpi_get_parent(handle, &pr_ahandle);
 	while (ACPI_SUCCESS(status)) {
@@ -1135,18 +1127,19 @@ static int acpi_processor_get_lpi_info(struct acpi_processor *pr)
 			break;
 
 		/* flatten all the LPI states in this level of hierarchy */
-		flatten_lpi_states(pr, curr, prev);
+		state_count = flatten_lpi_states(pr, state_count, curr, prev);
 
 		tmp = prev, prev = curr, curr = tmp;
 
 		status = acpi_get_parent(handle, &pr_ahandle);
 	}
 
-	pr->power.count = flat_state_cnt;
 	/* reset the index after flattening */
-	for (i = 0; i < pr->power.count; i++)
+	for (i = 0; i < state_count; i++)
 		pr->power.lpi_states[i].index = i;
 
+	pr->power.count = state_count;
+
 	/* Tell driver that _LPI is supported. */
 	pr->flags.has_lpi = 1;
 	pr->flags.power = 1;
@@ -1238,7 +1231,8 @@ static int acpi_processor_setup_cpuidle_states(struct acpi_processor *pr)
 	if (pr->flags.has_lpi)
 		return acpi_processor_setup_lpi_states(pr);
 
-	return acpi_processor_setup_cstates(pr);
+	acpi_processor_setup_cstates(pr);
+	return 0;
 }
 
 /**
@@ -1258,7 +1252,8 @@ static int acpi_processor_setup_cpuidle_dev(struct acpi_processor *pr,
 	if (pr->flags.has_lpi)
 		return acpi_processor_ffh_lpi_probe(pr->id);
 
-	return acpi_processor_setup_cpuidle_cx(pr, dev);
+	acpi_processor_setup_cpuidle_cx(pr, dev);
+	return 0;
 }
 
 static int acpi_processor_get_power_info(struct acpi_processor *pr)
@@ -1397,6 +1392,9 @@ int acpi_processor_power_init(struct acpi_processor *pr)
 		if (retval) {
 			if (acpi_processor_registered == 0)
 				cpuidle_unregister_driver(&acpi_idle_driver);
+
+			per_cpu(acpi_cpuidle_device, pr->id) = NULL;
+			kfree(dev);
 			return retval;
 		}
 		acpi_processor_registered++;
@@ -1423,3 +1421,5 @@ int acpi_processor_power_exit(struct acpi_processor *pr)
 	pr->flags.power_setup_done = 0;
 	return 0;
 }
+
+MODULE_IMPORT_NS("ACPI_PROCESSOR_IDLE");
diff --git a/drivers/acpi/processor_perflib.c b/drivers/acpi/processor_perflib.c
index 64b8d1e19594..8972446b7162 100644
--- a/drivers/acpi/processor_perflib.c
+++ b/drivers/acpi/processor_perflib.c
@@ -173,6 +173,9 @@ void acpi_processor_ppc_init(struct cpufreq_policy *policy)
 {
 	unsigned int cpu;
 
+	if (ignore_ppc == 1)
+		return;
+
 	for_each_cpu(cpu, policy->related_cpus) {
 		struct acpi_processor *pr = per_cpu(processors, cpu);
 		int ret;
@@ -193,6 +196,14 @@ void acpi_processor_ppc_init(struct cpufreq_policy *policy)
 		if (ret < 0)
 			pr_err("Failed to add freq constraint for CPU%d (%d)\n",
 			       cpu, ret);
+
+		if (!pr->performance)
+			continue;
+
+		ret = acpi_processor_get_platform_limit(pr);
+		if (ret)
+			pr_err("Failed to update freq constraint for CPU%d (%d)\n",
+			       cpu, ret);
 	}
 }
 
diff --git a/drivers/acpi/processor_thermal.c b/drivers/acpi/processor_thermal.c
index 1219adb11ab9..c7b1dc5687ec 100644
--- a/drivers/acpi/processor_thermal.c
+++ b/drivers/acpi/processor_thermal.c
@@ -62,19 +62,14 @@ static int phys_package_first_cpu(int cpu)
 	return 0;
 }
 
-static int cpu_has_cpufreq(unsigned int cpu)
+static bool cpu_has_cpufreq(unsigned int cpu)
 {
-	struct cpufreq_policy *policy;
-
 	if (!acpi_processor_cpufreq_init)
 		return 0;
 
-	policy = cpufreq_cpu_get(cpu);
-	if (policy) {
-		cpufreq_cpu_put(policy);
-		return 1;
-	}
-	return 0;
+	struct cpufreq_policy *policy __free(put_cpufreq_policy) = cpufreq_cpu_get(cpu);
+
+	return policy != NULL;
 }
 
 static int cpufreq_get_max_state(unsigned int cpu)
@@ -93,12 +88,31 @@ static int cpufreq_get_cur_state(unsigned int cpu)
 	return reduction_step(cpu);
 }
 
+static bool cpufreq_update_thermal_limit(unsigned int cpu, struct acpi_processor *pr)
+{
+	unsigned long max_freq;
+	int ret;
+
+	struct cpufreq_policy *policy __free(put_cpufreq_policy) = cpufreq_cpu_get(cpu);
+	if (!policy)
+		return false;
+
+	max_freq = (policy->cpuinfo.max_freq *
+		(100 - reduction_step(cpu) * cpufreq_thermal_reduction_pctg)) / 100;
+
+	ret = freq_qos_update_request(&pr->thermal_req, max_freq);
+	if (ret < 0) {
+		pr_warn("Failed to update thermal freq constraint: CPU%d (%d)\n",
+			pr->id, ret);
+	}
+
+	return true;
+}
+
 static int cpufreq_set_cur_state(unsigned int cpu, int state)
 {
-	struct cpufreq_policy *policy;
 	struct acpi_processor *pr;
-	unsigned long max_freq;
-	int i, ret;
+	int i;
 
 	if (!cpu_has_cpufreq(cpu))
 		return 0;
@@ -120,20 +134,8 @@ static int cpufreq_set_cur_state(unsigned int cpu, int state)
 		if (unlikely(!freq_qos_request_active(&pr->thermal_req)))
 			continue;
 
-		policy = cpufreq_cpu_get(i);
-		if (!policy)
+		if (!cpufreq_update_thermal_limit(i, pr))
 			return -EINVAL;
-
-		max_freq = (policy->cpuinfo.max_freq *
-			    (100 - reduction_step(i) * cpufreq_thermal_reduction_pctg)) / 100;
-
-		cpufreq_cpu_put(policy);
-
-		ret = freq_qos_update_request(&pr->thermal_req, max_freq);
-		if (ret < 0) {
-			pr_warn("Failed to update thermal freq constraint: CPU%d (%d)\n",
-				pr->id, ret);
-		}
 	}
 	return 0;
 }
diff --git a/drivers/acpi/processor_throttling.c b/drivers/acpi/processor_throttling.c
index d1541a386fbc..f9c2bc1d4a3a 100644
--- a/drivers/acpi/processor_throttling.c
+++ b/drivers/acpi/processor_throttling.c
@@ -235,7 +235,7 @@ static int acpi_processor_throttling_notifier(unsigned long event, void *data)
 		if (pr->throttling_platform_limit > target_state)
 			target_state = pr->throttling_platform_limit;
 		if (target_state >= p_throttling->state_count) {
-			pr_warn("Exceed the limit of T-state \n");
+			pr_warn("Exceed the limit of T-state\n");
 			target_state = p_throttling->state_count - 1;
 		}
 		p_tstate->target_state = target_state;
diff --git a/drivers/acpi/property.c b/drivers/acpi/property.c
index 436019d96027..18e90067d567 100644
--- a/drivers/acpi/property.c
+++ b/drivers/acpi/property.c
@@ -83,6 +83,7 @@ static bool acpi_nondev_subnode_extract(union acpi_object *desc,
 					struct fwnode_handle *parent)
 {
 	struct acpi_data_node *dn;
+	acpi_handle scope = NULL;
 	bool result;
 
 	if (acpi_graph_ignore_port(handle))
@@ -98,59 +99,45 @@ static bool acpi_nondev_subnode_extract(union acpi_object *desc,
 	INIT_LIST_HEAD(&dn->data.properties);
 	INIT_LIST_HEAD(&dn->data.subnodes);
 
-	result = acpi_extract_properties(handle, desc, &dn->data);
-
-	if (handle) {
-		acpi_handle scope;
-		acpi_status status;
+	/*
+	 * The scope for the completion of relative pathname segments and
+	 * subnode object lookup is the one of the namespace node (device)
+	 * containing the object that has returned the package.  That is, it's
+	 * the scope of that object's parent device.
+	 */
+	if (handle)
+		acpi_get_parent(handle, &scope);
 
-		/*
-		 * The scope for the subnode object lookup is the one of the
-		 * namespace node (device) containing the object that has
-		 * returned the package.  That is, it's the scope of that
-		 * object's parent.
-		 */
-		status = acpi_get_parent(handle, &scope);
-		if (ACPI_SUCCESS(status)
-		    && acpi_enumerate_nondev_subnodes(scope, desc, &dn->data,
-						      &dn->fwnode))
-			result = true;
-	} else if (acpi_enumerate_nondev_subnodes(NULL, desc, &dn->data,
-						  &dn->fwnode)) {
+	/*
+	 * Extract properties from the _DSD-equivalent package pointed to by
+	 * desc and use scope (if not NULL) for the completion of relative
+	 * pathname segments.
+	 *
+	 * The extracted properties will be held in the new data node dn.
+	 */
+	result = acpi_extract_properties(scope, desc, &dn->data);
+	/*
+	 * Look for subnodes in the _DSD-equivalent package pointed to by desc
+	 * and create child nodes of dn if there are any.
+	 */
+	if (acpi_enumerate_nondev_subnodes(scope, desc, &dn->data, &dn->fwnode))
 		result = true;
-	}
-
-	if (result) {
-		dn->handle = handle;
-		dn->data.pointer = desc;
-		list_add_tail(&dn->sibling, list);
-		return true;
-	}
-
-	kfree(dn);
-	acpi_handle_debug(handle, "Invalid properties/subnodes data, skipping\n");
-	return false;
-}
-
-static bool acpi_nondev_subnode_data_ok(acpi_handle handle,
-					const union acpi_object *link,
-					struct list_head *list,
-					struct fwnode_handle *parent)
-{
-	struct acpi_buffer buf = { ACPI_ALLOCATE_BUFFER };
-	acpi_status status;
 
-	status = acpi_evaluate_object_typed(handle, NULL, NULL, &buf,
-					    ACPI_TYPE_PACKAGE);
-	if (ACPI_FAILURE(status))
+	if (!result) {
+		kfree(dn);
+		acpi_handle_debug(handle, "Invalid properties/subnodes data, skipping\n");
 		return false;
+	}
 
-	if (acpi_nondev_subnode_extract(buf.pointer, handle, link, list,
-					parent))
-		return true;
+	/*
+	 * This will be NULL if the desc package is embedded in an outer
+	 * _DSD-equivalent package and its scope cannot be determined.
+	 */
+	dn->handle = handle;
+	dn->data.pointer = desc;
+	list_add_tail(&dn->sibling, list);
 
-	ACPI_FREE(buf.pointer);
-	return false;
+	return true;
 }
 
 static bool acpi_nondev_subnode_ok(acpi_handle scope,
@@ -158,9 +145,16 @@ static bool acpi_nondev_subnode_ok(acpi_handle scope,
 				   struct list_head *list,
 				   struct fwnode_handle *parent)
 {
+	struct acpi_buffer buf = { ACPI_ALLOCATE_BUFFER };
 	acpi_handle handle;
 	acpi_status status;
 
+	/*
+	 * If the scope is unknown, the _DSD-equivalent package being parsed
+	 * was embedded in an outer _DSD-equivalent package as a result of
+	 * direct evaluation of an object pointed to by a reference.  In that
+	 * case, using a pathname as the target object pointer is invalid.
+	 */
 	if (!scope)
 		return false;
 
@@ -169,7 +163,17 @@ static bool acpi_nondev_subnode_ok(acpi_handle scope,
 	if (ACPI_FAILURE(status))
 		return false;
 
-	return acpi_nondev_subnode_data_ok(handle, link, list, parent);
+	status = acpi_evaluate_object_typed(handle, NULL, NULL, &buf,
+					    ACPI_TYPE_PACKAGE);
+	if (ACPI_FAILURE(status))
+		return false;
+
+	if (acpi_nondev_subnode_extract(buf.pointer, handle, link, list,
+					parent))
+		return true;
+
+	ACPI_FREE(buf.pointer);
+	return false;
 }
 
 static bool acpi_add_nondev_subnodes(acpi_handle scope,
@@ -180,9 +184,12 @@ static bool acpi_add_nondev_subnodes(acpi_handle scope,
 	bool ret = false;
 	int i;
 
+	/*
+	 * Every element in the links package is expected to represent a link
+	 * to a non-device node in a tree containing device-specific data.
+	 */
 	for (i = 0; i < links->package.count; i++) {
 		union acpi_object *link, *desc;
-		acpi_handle handle;
 		bool result;
 
 		link = &links->package.elements[i];
@@ -190,26 +197,53 @@ static bool acpi_add_nondev_subnodes(acpi_handle scope,
 		if (link->package.count != 2)
 			continue;
 
-		/* The first one must be a string. */
+		/* The first one (the key) must be a string. */
 		if (link->package.elements[0].type != ACPI_TYPE_STRING)
 			continue;
 
-		/* The second one may be a string, a reference or a package. */
+		/* The second one (the target) may be a string or a package. */
 		switch (link->package.elements[1].type) {
 		case ACPI_TYPE_STRING:
+			/*
+			 * The string is expected to be a full pathname or a
+			 * pathname segment relative to the given scope.  That
+			 * pathname is expected to point to an object returning
+			 * a package that contains _DSD-equivalent information.
+			 */
 			result = acpi_nondev_subnode_ok(scope, link, list,
 							 parent);
 			break;
-		case ACPI_TYPE_LOCAL_REFERENCE:
-			handle = link->package.elements[1].reference.handle;
-			result = acpi_nondev_subnode_data_ok(handle, link, list,
-							     parent);
-			break;
 		case ACPI_TYPE_PACKAGE:
+			/*
+			 * This happens when a reference is used in AML to
+			 * point to the target.  Since the target is expected
+			 * to be a named object, a reference to it will cause it
+			 * to be avaluated in place and its return package will
+			 * be embedded in the links package at the location of
+			 * the reference.
+			 *
+			 * The target package is expected to contain _DSD-
+			 * equivalent information, but the scope in which it
+			 * is located in the original AML is unknown.  Thus
+			 * it cannot contain pathname segments represented as
+			 * strings because there is no way to build full
+			 * pathnames out of them.
+			 */
+			acpi_handle_debug(scope, "subnode %s: Unknown scope\n",
+					  link->package.elements[0].string.pointer);
 			desc = &link->package.elements[1];
 			result = acpi_nondev_subnode_extract(desc, NULL, link,
 							     list, parent);
 			break;
+		case ACPI_TYPE_LOCAL_REFERENCE:
+			/*
+			 * It is not expected to see any local references in
+			 * the links package because referencing a named object
+			 * should cause it to be evaluated in place.
+			 */
+			acpi_handle_info(scope, "subnode %s: Unexpected reference\n",
+					 link->package.elements[0].string.pointer);
+			fallthrough;
 		default:
 			result = false;
 			break;
@@ -369,6 +403,9 @@ static void acpi_untie_nondev_subnodes(struct acpi_device_data *data)
 	struct acpi_data_node *dn;
 
 	list_for_each_entry(dn, &data->subnodes, sibling) {
+		if (!dn->handle)
+			continue;
+
 		acpi_detach_data(dn->handle, acpi_nondev_subnode_tag);
 
 		acpi_untie_nondev_subnodes(&dn->data);
@@ -383,6 +420,9 @@ static bool acpi_tie_nondev_subnodes(struct acpi_device_data *data)
 		acpi_status status;
 		bool ret;
 
+		if (!dn->handle)
+			continue;
+
 		status = acpi_attach_data(dn->handle, acpi_nondev_subnode_tag, dn);
 		if (ACPI_FAILURE(status) && status != AE_ALREADY_EXISTS) {
 			acpi_handle_err(dn->handle, "Can't tag data node\n");
@@ -804,13 +844,35 @@ acpi_fwnode_get_named_child_node(const struct fwnode_handle *fwnode,
 	return NULL;
 }
 
+static unsigned int acpi_fwnode_get_args_count(struct fwnode_handle *fwnode,
+					       const char *nargs_prop)
+{
+	const struct acpi_device_data *data;
+	const union acpi_object *obj;
+	int ret;
+
+	data = acpi_device_data_of_node(fwnode);
+	if (!data)
+		return 0;
+
+	ret = acpi_data_get_property(data, nargs_prop, ACPI_TYPE_INTEGER, &obj);
+	if (ret)
+		return 0;
+
+	return obj->integer.value;
+}
+
 static int acpi_get_ref_args(struct fwnode_reference_args *args,
 			     struct fwnode_handle *ref_fwnode,
+			     const char *nargs_prop,
 			     const union acpi_object **element,
 			     const union acpi_object *end, size_t num_args)
 {
 	u32 nargs = 0, i;
 
+	if (nargs_prop)
+		num_args = acpi_fwnode_get_args_count(ref_fwnode, nargs_prop);
+
 	/*
 	 * Assume the following integer elements are all args. Stop counting on
 	 * the first reference (possibly represented as a string) or end of the
@@ -882,45 +944,10 @@ static struct fwnode_handle *acpi_parse_string_ref(const struct fwnode_handle *f
 	return &dn->fwnode;
 }
 
-/**
- * __acpi_node_get_property_reference - returns handle to the referenced object
- * @fwnode: Firmware node to get the property from
- * @propname: Name of the property
- * @index: Index of the reference to return
- * @num_args: Maximum number of arguments after each reference
- * @args: Location to store the returned reference with optional arguments
- *	  (may be NULL)
- *
- * Find property with @name, verifify that it is a package containing at least
- * one object reference and if so, store the ACPI device object pointer to the
- * target object in @args->adev.  If the reference includes arguments, store
- * them in the @args->args[] array.
- *
- * If there's more than one reference in the property value package, @index is
- * used to select the one to return.
- *
- * It is possible to leave holes in the property value set like in the
- * example below:
- *
- * Package () {
- *     "cs-gpios",
- *     Package () {
- *        ^GPIO, 19, 0, 0,
- *        ^GPIO, 20, 0, 0,
- *        0,
- *        ^GPIO, 21, 0, 0,
- *     }
- * }
- *
- * Calling this function with index %2 or index %3 return %-ENOENT. If the
- * property does not contain any more values %-ENOENT is returned. The NULL
- * entry must be single integer and preferably contain value %0.
- *
- * Return: %0 on success, negative error code on failure.
- */
-int __acpi_node_get_property_reference(const struct fwnode_handle *fwnode,
-	const char *propname, size_t index, size_t num_args,
-	struct fwnode_reference_args *args)
+static int acpi_fwnode_get_reference_args(const struct fwnode_handle *fwnode,
+					  const char *propname, const char *nargs_prop,
+					  unsigned int args_count, unsigned int index,
+					  struct fwnode_reference_args *args)
 {
 	const union acpi_object *element, *end;
 	const union acpi_object *obj;
@@ -996,10 +1023,10 @@ int __acpi_node_get_property_reference(const struct fwnode_handle *fwnode,
 				return -EINVAL;
 
 			element++;
-
 			ret = acpi_get_ref_args(idx == index ? args : NULL,
 						acpi_fwnode_handle(device),
-						&element, end, num_args);
+						nargs_prop, &element, end,
+						args_count);
 			if (ret < 0)
 				return ret;
 
@@ -1014,10 +1041,9 @@ int __acpi_node_get_property_reference(const struct fwnode_handle *fwnode,
 				return -EINVAL;
 
 			element++;
-
 			ret = acpi_get_ref_args(idx == index ? args : NULL,
-						ref_fwnode, &element, end,
-						num_args);
+						ref_fwnode, nargs_prop, &element, end,
+						args_count);
 			if (ret < 0)
 				return ret;
 
@@ -1039,6 +1065,50 @@ int __acpi_node_get_property_reference(const struct fwnode_handle *fwnode,
 
 	return -ENOENT;
 }
+
+/**
+ * __acpi_node_get_property_reference - returns handle to the referenced object
+ * @fwnode: Firmware node to get the property from
+ * @propname: Name of the property
+ * @index: Index of the reference to return
+ * @num_args: Maximum number of arguments after each reference
+ * @args: Location to store the returned reference with optional arguments
+ *	  (may be NULL)
+ *
+ * Find property with @name, verifify that it is a package containing at least
+ * one object reference and if so, store the ACPI device object pointer to the
+ * target object in @args->adev.  If the reference includes arguments, store
+ * them in the @args->args[] array.
+ *
+ * If there's more than one reference in the property value package, @index is
+ * used to select the one to return.
+ *
+ * It is possible to leave holes in the property value set like in the
+ * example below:
+ *
+ * Package () {
+ *     "cs-gpios",
+ *     Package () {
+ *        ^GPIO, 19, 0, 0,
+ *        ^GPIO, 20, 0, 0,
+ *        0,
+ *        ^GPIO, 21, 0, 0,
+ *     }
+ * }
+ *
+ * Calling this function with index %2 or index %3 return %-ENOENT. If the
+ * property does not contain any more values %-ENOENT is returned. The NULL
+ * entry must be single integer and preferably contain value %0.
+ *
+ * Return: %0 on success, negative error code on failure.
+ */
+int __acpi_node_get_property_reference(const struct fwnode_handle *fwnode,
+				       const char *propname, size_t index,
+				       size_t num_args,
+				       struct fwnode_reference_args *args)
+{
+	return acpi_fwnode_get_reference_args(fwnode, propname, NULL, num_args, index, args);
+}
 EXPORT_SYMBOL_GPL(__acpi_node_get_property_reference);
 
 static int acpi_data_prop_read_single(const struct acpi_device_data *data,
@@ -1210,7 +1280,7 @@ static int acpi_data_prop_read(const struct acpi_device_data *data,
 		ret = acpi_copy_property_array_uint(items, (u64 *)val, nval);
 		break;
 	case DEV_PROP_STRING:
-		nval = min_t(u32, nval, obj->package.count);
+		nval = min(nval, obj->package.count);
 		if (nval == 0)
 			return -ENODATA;
 
@@ -1259,13 +1329,14 @@ static int stop_on_next(struct acpi_device *adev, void *data)
 	return 0;
 }
 
-/**
+/*
  * acpi_get_next_subnode - Return the next child node handle for a fwnode
  * @fwnode: Firmware node to find the next child node for.
  * @child: Handle to one of the device's child nodes or a null handle.
  */
-struct fwnode_handle *acpi_get_next_subnode(const struct fwnode_handle *fwnode,
-					    struct fwnode_handle *child)
+static struct fwnode_handle *
+acpi_get_next_subnode(const struct fwnode_handle *fwnode,
+		      struct fwnode_handle *child)
 {
 	struct acpi_device *adev = to_acpi_device_node(fwnode);
 
@@ -1318,6 +1389,28 @@ struct fwnode_handle *acpi_get_next_subnode(const struct fwnode_handle *fwnode,
 	return NULL;
 }
 
+/*
+ * acpi_get_next_present_subnode - Return the next present child node handle
+ * @fwnode: Firmware node to find the next child node for.
+ * @child: Handle to one of the device's child nodes or a null handle.
+ *
+ * Like acpi_get_next_subnode(), but the device nodes returned by
+ * acpi_get_next_present_subnode() are guaranteed to be present.
+ *
+ * Returns: The fwnode handle of the next present sub-node.
+ */
+static struct fwnode_handle *
+acpi_get_next_present_subnode(const struct fwnode_handle *fwnode,
+			      struct fwnode_handle *child)
+{
+	do {
+		child = acpi_get_next_subnode(fwnode, child);
+	} while (is_acpi_device_node(child) &&
+		 !acpi_device_is_present(to_acpi_device_node(child)));
+
+	return child;
+}
+
 /**
  * acpi_node_get_parent - Return parent fwnode of this fwnode
  * @fwnode: Firmware node whose parent to get
@@ -1380,7 +1473,7 @@ static struct fwnode_handle *acpi_graph_get_next_endpoint(
 
 	if (!prev) {
 		do {
-			port = fwnode_get_next_child_node(fwnode, port);
+			port = acpi_get_next_subnode(fwnode, port);
 			/*
 			 * The names of the port nodes begin with "port@"
 			 * followed by the number of the port node and they also
@@ -1398,14 +1491,17 @@ static struct fwnode_handle *acpi_graph_get_next_endpoint(
 	if (!port)
 		return NULL;
 
-	endpoint = fwnode_get_next_child_node(port, prev);
-	while (!endpoint) {
-		port = fwnode_get_next_child_node(fwnode, port);
-		if (!port)
+	do {
+		endpoint = acpi_get_next_subnode(port, prev);
+		if (endpoint)
 			break;
-		if (is_acpi_graph_node(port, "port"))
-			endpoint = fwnode_get_next_child_node(port, NULL);
-	}
+
+		prev = NULL;
+
+		do {
+			port = acpi_get_next_subnode(fwnode, port);
+		} while (port && !is_acpi_graph_node(port, "port"));
+	} while (port);
 
 	/*
 	 * The names of the endpoint nodes begin with "endpoint@" followed by
@@ -1558,16 +1654,6 @@ acpi_fwnode_property_read_string_array(const struct fwnode_handle *fwnode,
 				   val, nval);
 }
 
-static int
-acpi_fwnode_get_reference_args(const struct fwnode_handle *fwnode,
-			       const char *prop, const char *nargs_prop,
-			       unsigned int args_count, unsigned int index,
-			       struct fwnode_reference_args *args)
-{
-	return __acpi_node_get_property_reference(fwnode, prop, index,
-						  args_count, args);
-}
-
 static const char *acpi_fwnode_get_name(const struct fwnode_handle *fwnode)
 {
 	const struct acpi_device *adev;
@@ -1632,6 +1718,7 @@ static int acpi_fwnode_graph_parse_endpoint(const struct fwnode_handle *fwnode,
 	if (fwnode_property_read_u32(fwnode, "reg", &endpoint->id))
 		fwnode_property_read_u32(fwnode, "endpoint", &endpoint->id);
 
+	fwnode_handle_put(port_fwnode);
 	return 0;
 }
 
@@ -1662,7 +1749,7 @@ static int acpi_fwnode_irq_get(const struct fwnode_handle *fwnode,
 		.property_read_string_array =				\
 			acpi_fwnode_property_read_string_array,		\
 		.get_parent = acpi_node_get_parent,			\
-		.get_next_child_node = acpi_get_next_subnode,		\
+		.get_next_child_node = acpi_get_next_present_subnode,	\
 		.get_named_child_node = acpi_fwnode_get_named_child_node, \
 		.get_name = acpi_fwnode_get_name,			\
 		.get_name_prefix = acpi_fwnode_get_name_prefix,		\
diff --git a/drivers/acpi/resource.c b/drivers/acpi/resource.c
index 7d59c6c9185f..d16906f46484 100644
--- a/drivers/acpi/resource.c
+++ b/drivers/acpi/resource.c
@@ -17,6 +17,7 @@
 #include <linux/slab.h>
 #include <linux/irq.h>
 #include <linux/dmi.h>
+#include <linux/string_choices.h>
 
 #ifdef CONFIG_X86
 #define valid_IRQ(i) (((i) != 0) && ((i) != 2))
@@ -511,6 +512,13 @@ static const struct dmi_system_id irq1_level_low_skip_override[] = {
 		},
 	},
 	{
+		/* Asus Vivobook Pro N6506CU* */
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "ASUSTeK COMPUTER INC."),
+			DMI_MATCH(DMI_BOARD_NAME, "N6506CU"),
+		},
+	},
+	{
 		/* LG Electronics 17U70P */
 		.matches = {
 			DMI_MATCH(DMI_SYS_VENDOR, "LG Electronics"),
@@ -667,6 +675,13 @@ static const struct dmi_system_id irq1_edge_low_force_override[] = {
 		},
 	},
 	{
+		/* MACHENIKE L16P/L16P */
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "MACHENIKE"),
+			DMI_MATCH(DMI_BOARD_NAME, "L16P"),
+		},
+	},
+	{
 		/*
 		 * TongFang GM5HG0A in case of the SKIKK Vanaheim relabel the
 		 * board-name is changed, so check OEM strings instead. Note
@@ -766,7 +781,7 @@ static void acpi_dev_get_irqresource(struct resource *res, u32 gsi,
 			pr_warn("ACPI: IRQ %d override to %s%s, %s%s\n", gsi,
 				t ? "level" : "edge",
 				trig == triggering ? "" : "(!)",
-				p ? "low" : "high",
+				str_low_high(p),
 				pol == polarity ? "" : "(!)");
 			triggering = trig;
 			polarity = pol;
diff --git a/drivers/acpi/riscv/Kconfig b/drivers/acpi/riscv/Kconfig
new file mode 100644
index 000000000000..046296a18d00
--- /dev/null
+++ b/drivers/acpi/riscv/Kconfig
@@ -0,0 +1,7 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# ACPI Configuration for RISC-V
+#
+
+config ACPI_RIMT
+	bool
diff --git a/drivers/acpi/riscv/Makefile b/drivers/acpi/riscv/Makefile
index a96fdf1e2cb8..1284a076fa88 100644
--- a/drivers/acpi/riscv/Makefile
+++ b/drivers/acpi/riscv/Makefile
@@ -2,3 +2,4 @@
 obj-y					+= rhct.o init.o irq.o
 obj-$(CONFIG_ACPI_PROCESSOR_IDLE)	+= cpuidle.o
 obj-$(CONFIG_ACPI_CPPC_LIB)		+= cppc.o
+obj-$(CONFIG_ACPI_RIMT)			+= rimt.o
diff --git a/drivers/acpi/riscv/cppc.c b/drivers/acpi/riscv/cppc.c
index 4cdff387deff..42c1a9052470 100644
--- a/drivers/acpi/riscv/cppc.c
+++ b/drivers/acpi/riscv/cppc.c
@@ -37,10 +37,8 @@ static int __init sbi_cppc_init(void)
 {
 	if (sbi_spec_version >= sbi_mk_version(2, 0) &&
 	    sbi_probe_extension(SBI_EXT_CPPC) > 0) {
-		pr_info("SBI CPPC extension detected\n");
 		cppc_ext_present = true;
 	} else {
-		pr_info("SBI CPPC extension NOT detected!!\n");
 		cppc_ext_present = false;
 	}
 
@@ -121,7 +119,7 @@ int cpc_read_ffh(int cpu, struct cpc_reg *reg, u64 *val)
 
 		*val = data.ret.value;
 
-		return (data.ret.error) ? sbi_err_map_linux_errno(data.ret.error) : 0;
+		return data.ret.error;
 	}
 
 	return -EINVAL;
@@ -150,7 +148,7 @@ int cpc_write_ffh(int cpu, struct cpc_reg *reg, u64 val)
 
 		smp_call_function_single(cpu, cppc_ffh_csr_write, &data, 1);
 
-		return (data.ret.error) ? sbi_err_map_linux_errno(data.ret.error) : 0;
+		return data.ret.error;
 	}
 
 	return -EINVAL;
diff --git a/drivers/acpi/riscv/init.c b/drivers/acpi/riscv/init.c
index 673e4d5dd752..7c00f7995e86 100644
--- a/drivers/acpi/riscv/init.c
+++ b/drivers/acpi/riscv/init.c
@@ -10,4 +10,6 @@
 void __init acpi_arch_init(void)
 {
 	riscv_acpi_init_gsi_mapping();
+	if (IS_ENABLED(CONFIG_ACPI_RIMT))
+		riscv_acpi_rimt_init();
 }
diff --git a/drivers/acpi/riscv/init.h b/drivers/acpi/riscv/init.h
index 0b9a07e4031f..1680aa2aaf23 100644
--- a/drivers/acpi/riscv/init.h
+++ b/drivers/acpi/riscv/init.h
@@ -2,3 +2,4 @@
 #include <linux/init.h>
 
 void __init riscv_acpi_init_gsi_mapping(void);
+void __init riscv_acpi_rimt_init(void);
diff --git a/drivers/acpi/riscv/irq.c b/drivers/acpi/riscv/irq.c
index cced960c2aef..d9a2154d6c6a 100644
--- a/drivers/acpi/riscv/irq.c
+++ b/drivers/acpi/riscv/irq.c
@@ -10,6 +10,8 @@
 
 #include "init.h"
 
+#define RISCV_ACPI_INTC_FLAG_PENDING BIT(0)
+
 struct riscv_ext_intc_list {
 	acpi_handle		handle;
 	u32			gsi_base;
@@ -17,6 +19,7 @@ struct riscv_ext_intc_list {
 	u32			nr_idcs;
 	u32			id;
 	u32			type;
+	u32			flag;
 	struct list_head	list;
 };
 
@@ -69,6 +72,22 @@ static acpi_status riscv_acpi_update_gsi_handle(u32 gsi_base, acpi_handle handle
 	return AE_NOT_FOUND;
 }
 
+int riscv_acpi_update_gsi_range(u32 gsi_base, u32 nr_irqs)
+{
+	struct riscv_ext_intc_list *ext_intc_element;
+
+	list_for_each_entry(ext_intc_element, &ext_intc_list, list) {
+		if (gsi_base == ext_intc_element->gsi_base &&
+		    (ext_intc_element->flag & RISCV_ACPI_INTC_FLAG_PENDING)) {
+			ext_intc_element->nr_irqs = nr_irqs;
+			ext_intc_element->flag &= ~RISCV_ACPI_INTC_FLAG_PENDING;
+			return 0;
+		}
+	}
+
+	return -ENODEV;
+}
+
 int riscv_acpi_get_gsi_info(struct fwnode_handle *fwnode, u32 *gsi_base,
 			    u32 *id, u32 *nr_irqs, u32 *nr_idcs)
 {
@@ -115,20 +134,67 @@ struct fwnode_handle *riscv_acpi_get_gsi_domain_id(u32 gsi)
 static int __init riscv_acpi_register_ext_intc(u32 gsi_base, u32 nr_irqs, u32 nr_idcs,
 					       u32 id, u32 type)
 {
-	struct riscv_ext_intc_list *ext_intc_element;
+	struct riscv_ext_intc_list *ext_intc_element, *node, *prev;
 
 	ext_intc_element = kzalloc(sizeof(*ext_intc_element), GFP_KERNEL);
 	if (!ext_intc_element)
 		return -ENOMEM;
 
 	ext_intc_element->gsi_base = gsi_base;
-	ext_intc_element->nr_irqs = nr_irqs;
+
+	/* If nr_irqs is zero, indicate it in flag and set to max range possible */
+	if (nr_irqs) {
+		ext_intc_element->nr_irqs = nr_irqs;
+	} else {
+		ext_intc_element->flag |= RISCV_ACPI_INTC_FLAG_PENDING;
+		ext_intc_element->nr_irqs = U32_MAX - ext_intc_element->gsi_base;
+	}
+
 	ext_intc_element->nr_idcs = nr_idcs;
 	ext_intc_element->id = id;
-	list_add_tail(&ext_intc_element->list, &ext_intc_list);
+	list_for_each_entry(node, &ext_intc_list, list) {
+		if (node->gsi_base < ext_intc_element->gsi_base)
+			break;
+	}
+
+	/* Adjust the previous node's GSI range if that has pending registration */
+	prev = list_prev_entry(node, list);
+	if (!list_entry_is_head(prev, &ext_intc_list, list)) {
+		if (prev->flag & RISCV_ACPI_INTC_FLAG_PENDING)
+			prev->nr_irqs = ext_intc_element->gsi_base - prev->gsi_base;
+	}
+
+	list_add_tail(&ext_intc_element->list, &node->list);
 	return 0;
 }
 
+static acpi_status __init riscv_acpi_create_gsi_map_smsi(acpi_handle handle, u32 level,
+							 void *context, void **return_value)
+{
+	acpi_status status;
+	u64 gbase;
+
+	if (!acpi_has_method(handle, "_GSB")) {
+		acpi_handle_err(handle, "_GSB method not found\n");
+		return AE_ERROR;
+	}
+
+	status = acpi_evaluate_integer(handle, "_GSB", NULL, &gbase);
+	if (ACPI_FAILURE(status)) {
+		acpi_handle_err(handle, "failed to evaluate _GSB method\n");
+		return status;
+	}
+
+	riscv_acpi_register_ext_intc(gbase, 0, 0, 0, ACPI_RISCV_IRQCHIP_SMSI);
+	status = riscv_acpi_update_gsi_handle((u32)gbase, handle);
+	if (ACPI_FAILURE(status)) {
+		acpi_handle_err(handle, "failed to find the GSI mapping entry\n");
+		return status;
+	}
+
+	return AE_OK;
+}
+
 static acpi_status __init riscv_acpi_create_gsi_map(acpi_handle handle, u32 level,
 						    void *context, void **return_value)
 {
@@ -183,6 +249,9 @@ void __init riscv_acpi_init_gsi_mapping(void)
 
 	if (acpi_table_parse_madt(ACPI_MADT_TYPE_APLIC, riscv_acpi_aplic_parse_madt, 0) > 0)
 		acpi_get_devices("RSCV0002", riscv_acpi_create_gsi_map, NULL, NULL);
+
+	/* Unlike PLIC/APLIC, SYSMSI doesn't have MADT */
+	acpi_get_devices("RSCV0006", riscv_acpi_create_gsi_map_smsi, NULL, NULL);
 }
 
 static acpi_handle riscv_acpi_get_gsi_handle(u32 gsi)
diff --git a/drivers/acpi/riscv/rimt.c b/drivers/acpi/riscv/rimt.c
new file mode 100644
index 000000000000..7f423405e5ef
--- /dev/null
+++ b/drivers/acpi/riscv/rimt.c
@@ -0,0 +1,520 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2024-2025, Ventana Micro Systems Inc
+ *	Author: Sunil V L <sunilvl@ventanamicro.com>
+ *
+ */
+
+#define pr_fmt(fmt)	"ACPI: RIMT: " fmt
+
+#include <linux/acpi.h>
+#include <linux/acpi_rimt.h>
+#include <linux/iommu.h>
+#include <linux/list.h>
+#include <linux/pci.h>
+#include <linux/platform_device.h>
+#include "init.h"
+
+struct rimt_fwnode {
+	struct list_head list;
+	struct acpi_rimt_node *rimt_node;
+	struct fwnode_handle *fwnode;
+};
+
+static LIST_HEAD(rimt_fwnode_list);
+static DEFINE_SPINLOCK(rimt_fwnode_lock);
+
+#define RIMT_TYPE_MASK(type)	(1 << (type))
+#define RIMT_IOMMU_TYPE		BIT(0)
+
+/* Root pointer to the mapped RIMT table */
+static struct acpi_table_header *rimt_table;
+
+/**
+ * rimt_set_fwnode() - Create rimt_fwnode and use it to register
+ *		       iommu data in the rimt_fwnode_list
+ *
+ * @rimt_node: RIMT table node associated with the IOMMU
+ * @fwnode: fwnode associated with the RIMT node
+ *
+ * Returns: 0 on success
+ *          <0 on failure
+ */
+static int rimt_set_fwnode(struct acpi_rimt_node *rimt_node,
+			   struct fwnode_handle *fwnode)
+{
+	struct rimt_fwnode *np;
+
+	np = kzalloc(sizeof(*np), GFP_ATOMIC);
+
+	if (WARN_ON(!np))
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&np->list);
+	np->rimt_node = rimt_node;
+	np->fwnode = fwnode;
+
+	spin_lock(&rimt_fwnode_lock);
+	list_add_tail(&np->list, &rimt_fwnode_list);
+	spin_unlock(&rimt_fwnode_lock);
+
+	return 0;
+}
+
+static acpi_status rimt_match_node_callback(struct acpi_rimt_node *node,
+					    void *context)
+{
+	acpi_status status = AE_NOT_FOUND;
+	struct device *dev = context;
+
+	if (node->type == ACPI_RIMT_NODE_TYPE_IOMMU) {
+		struct acpi_rimt_iommu *iommu_node = (struct acpi_rimt_iommu *)&node->node_data;
+
+		if (dev_is_pci(dev)) {
+			struct pci_dev *pdev;
+			u16 bdf;
+
+			pdev = to_pci_dev(dev);
+			bdf = PCI_DEVID(pdev->bus->number, pdev->devfn);
+			if ((pci_domain_nr(pdev->bus) == iommu_node->pcie_segment_number) &&
+			    bdf == iommu_node->pcie_bdf) {
+				status = AE_OK;
+			} else {
+				status = AE_NOT_FOUND;
+			}
+		} else {
+			struct platform_device *pdev = to_platform_device(dev);
+			struct resource *res;
+
+			res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+			if (res && res->start == iommu_node->base_address)
+				status = AE_OK;
+			else
+				status = AE_NOT_FOUND;
+		}
+	} else if (node->type == ACPI_RIMT_NODE_TYPE_PCIE_ROOT_COMPLEX) {
+		struct acpi_rimt_pcie_rc *pci_rc;
+		struct pci_bus *bus;
+
+		bus = to_pci_bus(dev);
+		pci_rc = (struct acpi_rimt_pcie_rc *)node->node_data;
+
+		/*
+		 * It is assumed that PCI segment numbers maps one-to-one
+		 * with root complexes. Each segment number can represent only
+		 * one root complex.
+		 */
+		status = pci_rc->pcie_segment_number == pci_domain_nr(bus) ?
+							AE_OK : AE_NOT_FOUND;
+	} else if (node->type == ACPI_RIMT_NODE_TYPE_PLAT_DEVICE) {
+		struct acpi_buffer buf = { ACPI_ALLOCATE_BUFFER, NULL };
+		struct acpi_rimt_platform_device *ncomp;
+		struct device *plat_dev = dev;
+		struct acpi_device *adev;
+
+		/*
+		 * Walk the device tree to find a device with an
+		 * ACPI companion; there is no point in scanning
+		 * RIMT for a device matching a platform device if
+		 * the device does not have an ACPI companion to
+		 * start with.
+		 */
+		do {
+			adev = ACPI_COMPANION(plat_dev);
+			if (adev)
+				break;
+
+			plat_dev = plat_dev->parent;
+		} while (plat_dev);
+
+		if (!adev)
+			return status;
+
+		status = acpi_get_name(adev->handle, ACPI_FULL_PATHNAME, &buf);
+		if (ACPI_FAILURE(status)) {
+			dev_warn(plat_dev, "Can't get device full path name\n");
+			return status;
+		}
+
+		ncomp = (struct acpi_rimt_platform_device *)node->node_data;
+		status = !strcmp(ncomp->device_name, buf.pointer) ?
+							AE_OK : AE_NOT_FOUND;
+		acpi_os_free(buf.pointer);
+	}
+
+	return status;
+}
+
+static struct acpi_rimt_node *rimt_scan_node(enum acpi_rimt_node_type type,
+					     void *context)
+{
+	struct acpi_rimt_node *rimt_node, *rimt_end;
+	struct acpi_table_rimt *rimt;
+	int i;
+
+	if (!rimt_table)
+		return NULL;
+
+	/* Get the first RIMT node */
+	rimt = (struct acpi_table_rimt *)rimt_table;
+	rimt_node = ACPI_ADD_PTR(struct acpi_rimt_node, rimt,
+				 rimt->node_offset);
+	rimt_end = ACPI_ADD_PTR(struct acpi_rimt_node, rimt_table,
+				rimt_table->length);
+
+	for (i = 0; i < rimt->num_nodes; i++) {
+		if (WARN_TAINT(rimt_node >= rimt_end, TAINT_FIRMWARE_WORKAROUND,
+			       "RIMT node pointer overflows, bad table!\n"))
+			return NULL;
+
+		if (rimt_node->type == type &&
+		    ACPI_SUCCESS(rimt_match_node_callback(rimt_node, context)))
+			return rimt_node;
+
+		rimt_node = ACPI_ADD_PTR(struct acpi_rimt_node, rimt_node,
+					 rimt_node->length);
+	}
+
+	return NULL;
+}
+
+/*
+ * RISC-V supports IOMMU as a PCI device or a platform device.
+ * When it is a platform device, there should be a namespace device as
+ * well along with RIMT. To create the link between RIMT information and
+ * the platform device, the IOMMU driver should register itself with the
+ * RIMT module. This is true for PCI based IOMMU as well.
+ */
+int rimt_iommu_register(struct device *dev)
+{
+	struct fwnode_handle *rimt_fwnode;
+	struct acpi_rimt_node *node;
+
+	node = rimt_scan_node(ACPI_RIMT_NODE_TYPE_IOMMU, dev);
+	if (!node) {
+		pr_err("Could not find IOMMU node in RIMT\n");
+		return -ENODEV;
+	}
+
+	if (dev_is_pci(dev)) {
+		rimt_fwnode = acpi_alloc_fwnode_static();
+		if (!rimt_fwnode)
+			return -ENOMEM;
+
+		rimt_fwnode->dev = dev;
+		if (!dev->fwnode)
+			dev->fwnode = rimt_fwnode;
+
+		rimt_set_fwnode(node, rimt_fwnode);
+	} else {
+		rimt_set_fwnode(node, dev->fwnode);
+	}
+
+	return 0;
+}
+
+#ifdef CONFIG_IOMMU_API
+
+/**
+ * rimt_get_fwnode() - Retrieve fwnode associated with an RIMT node
+ *
+ * @node: RIMT table node to be looked-up
+ *
+ * Returns: fwnode_handle pointer on success, NULL on failure
+ */
+static struct fwnode_handle *rimt_get_fwnode(struct acpi_rimt_node *node)
+{
+	struct fwnode_handle *fwnode = NULL;
+	struct rimt_fwnode *curr;
+
+	spin_lock(&rimt_fwnode_lock);
+	list_for_each_entry(curr, &rimt_fwnode_list, list) {
+		if (curr->rimt_node == node) {
+			fwnode = curr->fwnode;
+			break;
+		}
+	}
+	spin_unlock(&rimt_fwnode_lock);
+
+	return fwnode;
+}
+
+static bool rimt_pcie_rc_supports_ats(struct acpi_rimt_node *node)
+{
+	struct acpi_rimt_pcie_rc *pci_rc;
+
+	pci_rc = (struct acpi_rimt_pcie_rc *)node->node_data;
+	return pci_rc->flags & ACPI_RIMT_PCIE_ATS_SUPPORTED;
+}
+
+static int rimt_iommu_xlate(struct device *dev, struct acpi_rimt_node *node, u32 deviceid)
+{
+	struct fwnode_handle *rimt_fwnode;
+
+	if (!node)
+		return -ENODEV;
+
+	rimt_fwnode = rimt_get_fwnode(node);
+
+	/*
+	 * The IOMMU drivers may not be probed yet.
+	 * Defer the IOMMU configuration
+	 */
+	if (!rimt_fwnode)
+		return -EPROBE_DEFER;
+
+	return acpi_iommu_fwspec_init(dev, deviceid, rimt_fwnode);
+}
+
+struct rimt_pci_alias_info {
+	struct device *dev;
+	struct acpi_rimt_node *node;
+	const struct iommu_ops *ops;
+};
+
+static int rimt_id_map(struct acpi_rimt_id_mapping *map, u8 type, u32 rid_in, u32 *rid_out)
+{
+	if (rid_in < map->source_id_base ||
+	    (rid_in > map->source_id_base + map->num_ids))
+		return -ENXIO;
+
+	*rid_out = map->dest_id_base + (rid_in - map->source_id_base);
+	return 0;
+}
+
+static struct acpi_rimt_node *rimt_node_get_id(struct acpi_rimt_node *node,
+					       u32 *id_out, int index)
+{
+	struct acpi_rimt_platform_device *plat_node;
+	u32 id_mapping_offset, num_id_mapping;
+	struct acpi_rimt_pcie_rc *pci_node;
+	struct acpi_rimt_id_mapping *map;
+	struct acpi_rimt_node *parent;
+
+	if (node->type == ACPI_RIMT_NODE_TYPE_PCIE_ROOT_COMPLEX) {
+		pci_node = (struct acpi_rimt_pcie_rc *)&node->node_data;
+		id_mapping_offset = pci_node->id_mapping_offset;
+		num_id_mapping = pci_node->num_id_mappings;
+	} else if (node->type == ACPI_RIMT_NODE_TYPE_PLAT_DEVICE) {
+		plat_node = (struct acpi_rimt_platform_device *)&node->node_data;
+		id_mapping_offset = plat_node->id_mapping_offset;
+		num_id_mapping = plat_node->num_id_mappings;
+	} else {
+		return NULL;
+	}
+
+	if (!id_mapping_offset || !num_id_mapping || index >= num_id_mapping)
+		return NULL;
+
+	map = ACPI_ADD_PTR(struct acpi_rimt_id_mapping, node,
+			   id_mapping_offset + index * sizeof(*map));
+
+	/* Firmware bug! */
+	if (!map->dest_offset) {
+		pr_err(FW_BUG "[node %p type %d] ID map has NULL parent reference\n",
+		       node, node->type);
+		return NULL;
+	}
+
+	parent = ACPI_ADD_PTR(struct acpi_rimt_node, rimt_table, map->dest_offset);
+
+	if (node->type == ACPI_RIMT_NODE_TYPE_PLAT_DEVICE ||
+	    node->type == ACPI_RIMT_NODE_TYPE_PCIE_ROOT_COMPLEX) {
+		*id_out = map->dest_id_base;
+		return parent;
+	}
+
+	return NULL;
+}
+
+static struct acpi_rimt_node *rimt_node_map_id(struct acpi_rimt_node *node,
+					       u32 id_in, u32 *id_out,
+					       u8 type_mask)
+{
+	struct acpi_rimt_platform_device *plat_node;
+	u32 id_mapping_offset, num_id_mapping;
+	struct acpi_rimt_pcie_rc *pci_node;
+	u32 id = id_in;
+
+	/* Parse the ID mapping tree to find specified node type */
+	while (node) {
+		struct acpi_rimt_id_mapping *map;
+		int i, rc = 0;
+		u32 map_id = id;
+
+		if (RIMT_TYPE_MASK(node->type) & type_mask) {
+			if (id_out)
+				*id_out = id;
+			return node;
+		}
+
+		if (node->type == ACPI_RIMT_NODE_TYPE_PCIE_ROOT_COMPLEX) {
+			pci_node = (struct acpi_rimt_pcie_rc *)&node->node_data;
+			id_mapping_offset = pci_node->id_mapping_offset;
+			num_id_mapping = pci_node->num_id_mappings;
+		} else if (node->type == ACPI_RIMT_NODE_TYPE_PLAT_DEVICE) {
+			plat_node = (struct acpi_rimt_platform_device *)&node->node_data;
+			id_mapping_offset = plat_node->id_mapping_offset;
+			num_id_mapping = plat_node->num_id_mappings;
+		} else {
+			goto fail_map;
+		}
+
+		if (!id_mapping_offset || !num_id_mapping)
+			goto fail_map;
+
+		map = ACPI_ADD_PTR(struct acpi_rimt_id_mapping, node,
+				   id_mapping_offset);
+
+		/* Firmware bug! */
+		if (!map->dest_offset) {
+			pr_err(FW_BUG "[node %p type %d] ID map has NULL parent reference\n",
+			       node, node->type);
+			goto fail_map;
+		}
+
+		/* Do the ID translation */
+		for (i = 0; i < num_id_mapping; i++, map++) {
+			rc = rimt_id_map(map, node->type, map_id, &id);
+			if (!rc)
+				break;
+		}
+
+		if (i == num_id_mapping)
+			goto fail_map;
+
+		node = ACPI_ADD_PTR(struct acpi_rimt_node, rimt_table,
+				    rc ? 0 : map->dest_offset);
+	}
+
+fail_map:
+	/* Map input ID to output ID unchanged on mapping failure */
+	if (id_out)
+		*id_out = id_in;
+
+	return NULL;
+}
+
+static struct acpi_rimt_node *rimt_node_map_platform_id(struct acpi_rimt_node *node, u32 *id_out,
+							u8 type_mask, int index)
+{
+	struct acpi_rimt_node *parent;
+	u32 id;
+
+	parent = rimt_node_get_id(node, &id, index);
+	if (!parent)
+		return NULL;
+
+	if (!(RIMT_TYPE_MASK(parent->type) & type_mask))
+		parent = rimt_node_map_id(parent, id, id_out, type_mask);
+	else
+		if (id_out)
+			*id_out = id;
+
+	return parent;
+}
+
+static int rimt_pci_iommu_init(struct pci_dev *pdev, u16 alias, void *data)
+{
+	struct rimt_pci_alias_info *info = data;
+	struct acpi_rimt_node *parent;
+	u32 deviceid;
+
+	parent = rimt_node_map_id(info->node, alias, &deviceid, RIMT_IOMMU_TYPE);
+	return rimt_iommu_xlate(info->dev, parent, deviceid);
+}
+
+static int rimt_plat_iommu_map(struct device *dev, struct acpi_rimt_node *node)
+{
+	struct acpi_rimt_node *parent;
+	int err = -ENODEV, i = 0;
+	u32 deviceid = 0;
+
+	do {
+		parent = rimt_node_map_platform_id(node, &deviceid,
+						   RIMT_IOMMU_TYPE,
+						   i++);
+
+		if (parent)
+			err = rimt_iommu_xlate(dev, parent, deviceid);
+	} while (parent && !err);
+
+	return err;
+}
+
+static int rimt_plat_iommu_map_id(struct device *dev,
+				  struct acpi_rimt_node *node,
+				  const u32 *in_id)
+{
+	struct acpi_rimt_node *parent;
+	u32 deviceid;
+
+	parent = rimt_node_map_id(node, *in_id, &deviceid, RIMT_IOMMU_TYPE);
+	if (parent)
+		return rimt_iommu_xlate(dev, parent, deviceid);
+
+	return -ENODEV;
+}
+
+/**
+ * rimt_iommu_configure_id - Set-up IOMMU configuration for a device.
+ *
+ * @dev: device to configure
+ * @id_in: optional input id const value pointer
+ *
+ * Returns: 0 on success, <0 on failure
+ */
+int rimt_iommu_configure_id(struct device *dev, const u32 *id_in)
+{
+	struct acpi_rimt_node *node;
+	int err = -ENODEV;
+
+	if (dev_is_pci(dev)) {
+		struct iommu_fwspec *fwspec;
+		struct pci_bus *bus = to_pci_dev(dev)->bus;
+		struct rimt_pci_alias_info info = { .dev = dev };
+
+		node = rimt_scan_node(ACPI_RIMT_NODE_TYPE_PCIE_ROOT_COMPLEX, &bus->dev);
+		if (!node)
+			return -ENODEV;
+
+		info.node = node;
+		err = pci_for_each_dma_alias(to_pci_dev(dev),
+					     rimt_pci_iommu_init, &info);
+
+		fwspec = dev_iommu_fwspec_get(dev);
+		if (fwspec && rimt_pcie_rc_supports_ats(node))
+			fwspec->flags |= IOMMU_FWSPEC_PCI_RC_ATS;
+	} else {
+		node = rimt_scan_node(ACPI_RIMT_NODE_TYPE_PLAT_DEVICE, dev);
+		if (!node)
+			return -ENODEV;
+
+		err = id_in ? rimt_plat_iommu_map_id(dev, node, id_in) :
+			      rimt_plat_iommu_map(dev, node);
+	}
+
+	return err;
+}
+
+#endif
+
+void __init riscv_acpi_rimt_init(void)
+{
+	acpi_status status;
+
+	/* rimt_table will be used at runtime after the rimt init,
+	 * so we don't need to call acpi_put_table() to release
+	 * the RIMT table mapping.
+	 */
+	status = acpi_get_table(ACPI_SIG_RIMT, 0, &rimt_table);
+	if (ACPI_FAILURE(status)) {
+		if (status != AE_NOT_FOUND) {
+			const char *msg = acpi_format_exception(status);
+
+			pr_err("Failed to get table, %s\n", msg);
+		}
+
+		return;
+	}
+}
diff --git a/drivers/acpi/sbs.c b/drivers/acpi/sbs.c
index a3f95a3fffde..d3edc3bcbf01 100644
--- a/drivers/acpi/sbs.c
+++ b/drivers/acpi/sbs.c
@@ -487,7 +487,7 @@ static int acpi_battery_read(struct acpi_battery *battery)
 		if (result)
 			return result;
 
-		battery->present = state & (1 << battery->id);
+		battery->present = !!(state & (1 << battery->id));
 		if (!battery->present)
 			return 0;
 
diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c
index fb1fe9f3b1a3..416d87f9bd10 100644
--- a/drivers/acpi/scan.c
+++ b/drivers/acpi/scan.c
@@ -11,6 +11,7 @@
 #include <linux/kernel.h>
 #include <linux/acpi.h>
 #include <linux/acpi_iort.h>
+#include <linux/acpi_rimt.h>
 #include <linux/acpi_viot.h>
 #include <linux/iommu.h>
 #include <linux/signal.h>
@@ -845,6 +846,8 @@ static bool acpi_info_matches_ids(struct acpi_device_info *info,
 static const char * const acpi_ignore_dep_ids[] = {
 	"PNP0D80", /* Windows-compatible System Power Management Controller */
 	"INT33BD", /* Intel Baytrail Mailbox Device */
+	"INTC10DE", /* Intel CVS LNL */
+	"INTC10E0", /* Intel CVS ARL */
 	"LATT2021", /* Lattice FW Update Client Driver */
 	NULL
 };
@@ -858,6 +861,8 @@ static const char * const acpi_honor_dep_ids[] = {
 	"INTC10CF", /* IVSC (MTL) driver must be loaded to allow i2c access to camera sensors */
 	"RSCV0001", /* RISC-V PLIC */
 	"RSCV0002", /* RISC-V APLIC */
+	"RSCV0005", /* RISC-V SBI MPXY MBOX */
+	"RSCV0006", /* RISC-V RPMI SYSMSI */
 	"PNP0C0F",  /* PCI Link Device */
 	NULL
 };
@@ -1629,7 +1634,10 @@ static int acpi_iommu_configure_id(struct device *dev, const u32 *id_in)
 
 	err = iort_iommu_configure_id(dev, id_in);
 	if (err && err != -EPROBE_DEFER)
+		err = rimt_iommu_configure_id(dev, id_in);
+	if (err && err != -EPROBE_DEFER)
 		err = viot_iommu_configure(dev);
+
 	mutex_unlock(&iommu_probe_device_lock);
 
 	return err;
@@ -2389,7 +2397,7 @@ static bool acpi_scan_clear_dep_queue(struct acpi_device *adev)
 	 * initial enumeration of devices is complete, put it into the unbound
 	 * workqueue.
 	 */
-	queue_work(system_unbound_wq, &cdw->work);
+	queue_work(system_dfl_wq, &cdw->work);
 
 	return true;
 }
@@ -2702,7 +2710,7 @@ void __init acpi_scan_init(void)
 	acpi_memory_hotplug_init();
 	acpi_watchdog_init();
 	acpi_pnp_init();
-	acpi_int340x_thermal_init();
+	acpi_power_resources_init();
 	acpi_init_lpit();
 
 	acpi_scan_add_handler(&generic_device_handler);
diff --git a/drivers/acpi/sleep.c b/drivers/acpi/sleep.c
index c8ee8e42b0f6..66ec81e306d4 100644
--- a/drivers/acpi/sleep.c
+++ b/drivers/acpi/sleep.c
@@ -642,7 +642,7 @@ static int acpi_suspend_enter(suspend_state_t pm_state)
 	/*
 	 * Disable all GPE and clear their status bits before interrupts are
 	 * enabled. Some GPEs (like wakeup GPEs) have no handlers and this can
-	 * prevent them from producing spurious interrups.
+	 * prevent them from producing spurious interrupts.
 	 *
 	 * acpi_leave_sleep_state() will reenable specific GPEs later.
 	 *
@@ -884,13 +884,13 @@ bool acpi_s2idle_wakeup(void)
 #ifdef CONFIG_PM_SLEEP
 static u32 saved_bm_rld;
 
-static int  acpi_save_bm_rld(void)
+static int  acpi_save_bm_rld(void *data)
 {
 	acpi_read_bit_register(ACPI_BITREG_BUS_MASTER_RLD, &saved_bm_rld);
 	return 0;
 }
 
-static void  acpi_restore_bm_rld(void)
+static void  acpi_restore_bm_rld(void *data)
 {
 	u32 resumed_bm_rld = 0;
 
@@ -901,14 +901,18 @@ static void  acpi_restore_bm_rld(void)
 	acpi_write_bit_register(ACPI_BITREG_BUS_MASTER_RLD, saved_bm_rld);
 }
 
-static struct syscore_ops acpi_sleep_syscore_ops = {
+static const struct syscore_ops acpi_sleep_syscore_ops = {
 	.suspend = acpi_save_bm_rld,
 	.resume = acpi_restore_bm_rld,
 };
 
+static struct syscore acpi_sleep_syscore = {
+	.ops = &acpi_sleep_syscore_ops,
+};
+
 static void acpi_sleep_syscore_init(void)
 {
-	register_syscore_ops(&acpi_sleep_syscore_ops);
+	register_syscore(&acpi_sleep_syscore);
 }
 #else
 static inline void acpi_sleep_syscore_init(void) {}
diff --git a/drivers/acpi/sleep.h b/drivers/acpi/sleep.h
index d960a238be4e..9c3cb109c5d2 100644
--- a/drivers/acpi/sleep.h
+++ b/drivers/acpi/sleep.h
@@ -17,10 +17,7 @@ static inline acpi_status acpi_set_waking_vector(u32 wakeup_address)
 
 extern int acpi_s2idle_begin(void);
 extern int acpi_s2idle_prepare(void);
-extern int acpi_s2idle_prepare_late(void);
-extern void acpi_s2idle_check(void);
 extern bool acpi_s2idle_wake(void);
-extern void acpi_s2idle_restore_early(void);
 extern void acpi_s2idle_restore(void);
 extern void acpi_s2idle_end(void);
 
diff --git a/drivers/acpi/spcr.c b/drivers/acpi/spcr.c
index cd36a97b0ea2..73cb933fdc89 100644
--- a/drivers/acpi/spcr.c
+++ b/drivers/acpi/spcr.c
@@ -141,12 +141,23 @@ int __init acpi_parse_spcr(bool enable_earlycon, bool enable_console)
 	case ACPI_DBG2_16550_NVIDIA:
 		uart = "uart";
 		break;
+	case ACPI_DBG2_RISCV_SBI_CON:
+		uart = "sbi";
+		break;
 	default:
 		err = -ENOENT;
 		goto done;
 	}
 
-	switch (table->baud_rate) {
+	/*
+	 * SPCR 1.09 defines Precise Baud Rate Filed contains a specific
+	 * non-zero baud rate which overrides the value of the Configured
+	 * Baud Rate field. If this field is zero or not present, Configured
+	 * Baud Rate is used.
+	 */
+	if (table->header.revision >= 4 && table->precise_baudrate)
+		baud_rate = table->precise_baudrate;
+	else switch (table->baud_rate) {
 	case 0:
 		/*
 		 * SPCR 1.04 defines 0 as a preconfigured state of UART.
diff --git a/drivers/acpi/sysfs.c b/drivers/acpi/sysfs.c
index a48ebbf768f9..e596224302f4 100644
--- a/drivers/acpi/sysfs.c
+++ b/drivers/acpi/sysfs.c
@@ -372,7 +372,7 @@ static int acpi_table_attr_init(struct kobject *tables_obj,
 	}
 
 	table_attr->attr.size = table_header->length;
-	table_attr->attr.read_new = acpi_table_show;
+	table_attr->attr.read = acpi_table_show;
 	table_attr->attr.attr.name = table_attr->filename;
 	table_attr->attr.attr.mode = 0400;
 
@@ -495,7 +495,7 @@ static int acpi_table_data_init(struct acpi_table_header *th)
 			if (!data_attr)
 				return -ENOMEM;
 			sysfs_attr_init(&data_attr->attr.attr);
-			data_attr->attr.read_new = acpi_data_show;
+			data_attr->attr.read = acpi_data_show;
 			data_attr->attr.attr.mode = 0400;
 			return acpi_data_objs[i].fn(th, data_attr);
 		}
diff --git a/drivers/acpi/tables.c b/drivers/acpi/tables.c
index fa9bb8c8ce95..4286e4af1092 100644
--- a/drivers/acpi/tables.c
+++ b/drivers/acpi/tables.c
@@ -408,7 +408,7 @@ static const char table_sigs[][ACPI_NAMESEG_SIZE] __nonstring_array __initconst
 	ACPI_SIG_PSDT, ACPI_SIG_RSDT, ACPI_SIG_XSDT, ACPI_SIG_SSDT,
 	ACPI_SIG_IORT, ACPI_SIG_NFIT, ACPI_SIG_HMAT, ACPI_SIG_PPTT,
 	ACPI_SIG_NHLT, ACPI_SIG_AEST, ACPI_SIG_CEDT, ACPI_SIG_AGDI,
-	ACPI_SIG_NBFT };
+	ACPI_SIG_NBFT, ACPI_SIG_SWFT, ACPI_SIG_MPAM};
 
 #define ACPI_HEADER_SIZE sizeof(struct acpi_table_header)
 
diff --git a/drivers/acpi/thermal.c b/drivers/acpi/thermal.c
index 5c2defe55898..a511f9ea0267 100644
--- a/drivers/acpi/thermal.c
+++ b/drivers/acpi/thermal.c
@@ -924,7 +924,7 @@ static int acpi_thermal_suspend(struct device *dev)
 static int acpi_thermal_resume(struct device *dev)
 {
 	struct acpi_thermal *tz;
-	int i, j, power_state;
+	int i, j;
 
 	if (!dev)
 		return -EINVAL;
@@ -939,10 +939,8 @@ static int acpi_thermal_resume(struct device *dev)
 		if (!acpi_thermal_trip_valid(acpi_trip))
 			break;
 
-		for (j = 0; j < acpi_trip->devices.count; j++) {
-			acpi_bus_update_power(acpi_trip->devices.handles[j],
-					      &power_state);
-		}
+		for (j = 0; j < acpi_trip->devices.count; j++)
+			acpi_bus_update_power(acpi_trip->devices.handles[j], NULL);
 	}
 
 	acpi_queue_thermal_check(tz);
@@ -1062,7 +1060,8 @@ static int __init acpi_thermal_init(void)
 	}
 
 	acpi_thermal_pm_queue = alloc_workqueue("acpi_thermal_pm",
-						WQ_HIGHPRI | WQ_MEM_RECLAIM, 0);
+						WQ_HIGHPRI | WQ_MEM_RECLAIM | WQ_PERCPU,
+						0);
 	if (!acpi_thermal_pm_queue)
 		return -ENODEV;
 
diff --git a/drivers/acpi/video_detect.c b/drivers/acpi/video_detect.c
index d507d5e08435..4cf74f173c78 100644
--- a/drivers/acpi/video_detect.c
+++ b/drivers/acpi/video_detect.c
@@ -948,6 +948,14 @@ static const struct dmi_system_id video_detect_dmi_table[] = {
 		DMI_MATCH(DMI_PRODUCT_NAME, "Mipad2"),
 		},
 	},
+	/* https://gitlab.freedesktop.org/drm/amd/-/issues/4512 */
+	{
+	 .callback = video_detect_force_native,
+	 .matches = {
+		DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"),
+		DMI_MATCH(DMI_PRODUCT_NAME, "82K8"),
+		},
+	},
 	{ },
 };
 
diff --git a/drivers/acpi/wakeup.c b/drivers/acpi/wakeup.c
index b02bf770aead..ff6dc957bc11 100644
--- a/drivers/acpi/wakeup.c
+++ b/drivers/acpi/wakeup.c
@@ -42,7 +42,7 @@ void acpi_enable_wakeup_devices(u8 sleep_state)
 	list_for_each_entry_safe(dev, tmp, &acpi_wakeup_device_list,
 				 wakeup_list) {
 		if (!dev->wakeup.flags.valid
-		    || sleep_state > (u32) dev->wakeup.sleep_state
+		    || sleep_state > dev->wakeup.sleep_state
 		    || !(device_may_wakeup(&dev->dev)
 			 || dev->wakeup.prepare_count))
 			continue;
@@ -67,7 +67,7 @@ void acpi_disable_wakeup_devices(u8 sleep_state)
 	list_for_each_entry_safe(dev, tmp, &acpi_wakeup_device_list,
 				 wakeup_list) {
 		if (!dev->wakeup.flags.valid
-		    || sleep_state > (u32) dev->wakeup.sleep_state
+		    || sleep_state > dev->wakeup.sleep_state
 		    || !(device_may_wakeup(&dev->dev)
 			 || dev->wakeup.prepare_count))
 			continue;
diff --git a/drivers/acpi/x86/lpss.c b/drivers/acpi/x86/lpss.c
index 258440b899a9..1dcb80ab0d23 100644
--- a/drivers/acpi/x86/lpss.c
+++ b/drivers/acpi/x86/lpss.c
@@ -181,7 +181,7 @@ static void byt_i2c_setup(struct lpss_private_data *pdata)
 	acpi_status status;
 	u64 uid;
 
-	/* Expected to always be successfull, but better safe then sorry */
+	/* Expected to always be successful, but better safe then sorry */
 	if (!acpi_dev_uid_to_integer(pdata->adev, &uid) && uid) {
 		/* Detect I2C bus shared with PUNIT and ignore its d3 status */
 		status = acpi_evaluate_integer(handle, "_SEM", NULL, &shared_host);
@@ -387,9 +387,6 @@ static const struct acpi_device_id acpi_lpss_device_ids[] = {
 	{ "INT3435", LPSS_ADDR(lpt_uart_dev_desc) },
 	{ "INT3436", LPSS_ADDR(lpt_sdio_dev_desc) },
 
-	/* Wildcat Point LPSS devices */
-	{ "INT3438", LPSS_ADDR(lpt_spi_dev_desc) },
-
 	{ }
 };
 
diff --git a/drivers/acpi/x86/s2idle.c b/drivers/acpi/x86/s2idle.c
index dd0b40b9bbe8..6d4d06236f61 100644
--- a/drivers/acpi/x86/s2idle.c
+++ b/drivers/acpi/x86/s2idle.c
@@ -299,34 +299,13 @@ free_acpi_buffer:
 	ACPI_FREE(out_obj);
 }
 
-/**
- * acpi_get_lps0_constraint - Get the LPS0 constraint for a device.
- * @adev: Device to get the constraint for.
- *
- * The LPS0 constraint is the shallowest (minimum) power state in which the
- * device can be so as to allow the platform as a whole to achieve additional
- * energy conservation by utilizing a system-wide low-power state.
- *
- * Returns:
- *  - ACPI power state value of the constraint for @adev on success.
- *  - Otherwise, ACPI_STATE_UNKNOWN.
- */
-int acpi_get_lps0_constraint(struct acpi_device *adev)
-{
-	struct lpi_constraints *entry;
-
-	for_each_lpi_constraint(entry) {
-		if (adev->handle == entry->handle)
-			return entry->min_dstate;
-	}
-
-	return ACPI_STATE_UNKNOWN;
-}
-
 static void lpi_check_constraints(void)
 {
 	struct lpi_constraints *entry;
 
+	if (IS_ERR_OR_NULL(lpi_constraints_table))
+		return;
+
 	for_each_lpi_constraint(entry) {
 		struct acpi_device *adev = acpi_fetch_acpi_dev(entry->handle);
 
@@ -508,11 +487,6 @@ static int lps0_device_attach(struct acpi_device *adev,
 
 	lps0_device_handle = adev->handle;
 
-	if (acpi_s2idle_vendor_amd())
-		lpi_device_get_constraints_amd();
-	else
-		lpi_device_get_constraints();
-
 	/*
 	 * Use suspend-to-idle by default if ACPI_FADT_LOW_POWER_S0 is set in
 	 * the FADT and the default suspend mode was not set from the command
@@ -539,7 +513,26 @@ static struct acpi_scan_handler lps0_handler = {
 	.attach = lps0_device_attach,
 };
 
-int acpi_s2idle_prepare_late(void)
+static int acpi_s2idle_begin_lps0(void)
+{
+	if (pm_debug_messages_on && !lpi_constraints_table) {
+		if (acpi_s2idle_vendor_amd())
+			lpi_device_get_constraints_amd();
+		else
+			lpi_device_get_constraints();
+
+		/*
+		 * Try to retrieve the constraints only once because failures
+		 * to do so usually are sticky.
+		 */
+		if (!lpi_constraints_table)
+			lpi_constraints_table = ERR_PTR(-ENODATA);
+	}
+
+	return acpi_s2idle_begin();
+}
+
+static int acpi_s2idle_prepare_late_lps0(void)
 {
 	struct acpi_s2idle_dev_ops *handler;
 
@@ -585,7 +578,7 @@ int acpi_s2idle_prepare_late(void)
 	return 0;
 }
 
-void acpi_s2idle_check(void)
+static void acpi_s2idle_check_lps0(void)
 {
 	struct acpi_s2idle_dev_ops *handler;
 
@@ -598,7 +591,7 @@ void acpi_s2idle_check(void)
 	}
 }
 
-void acpi_s2idle_restore_early(void)
+static void acpi_s2idle_restore_early_lps0(void)
 {
 	struct acpi_s2idle_dev_ops *handler;
 
@@ -636,12 +629,12 @@ void acpi_s2idle_restore_early(void)
 }
 
 static const struct platform_s2idle_ops acpi_s2idle_ops_lps0 = {
-	.begin = acpi_s2idle_begin,
+	.begin = acpi_s2idle_begin_lps0,
 	.prepare = acpi_s2idle_prepare,
-	.prepare_late = acpi_s2idle_prepare_late,
-	.check = acpi_s2idle_check,
+	.prepare_late = acpi_s2idle_prepare_late_lps0,
+	.check = acpi_s2idle_check_lps0,
 	.wake = acpi_s2idle_wake,
-	.restore_early = acpi_s2idle_restore_early,
+	.restore_early = acpi_s2idle_restore_early_lps0,
 	.restore = acpi_s2idle_restore,
 	.end = acpi_s2idle_end,
 };
diff --git a/drivers/amba/Kconfig b/drivers/amba/Kconfig
index fb6c7e0b4cce..14bb61ff801e 100644
--- a/drivers/amba/Kconfig
+++ b/drivers/amba/Kconfig
@@ -5,7 +5,7 @@ config ARM_AMBA
 if ARM_AMBA
 
 config TEGRA_AHB
-	bool
+	bool "Enable AHB driver for NVIDIA Tegra SoCs" if COMPILE_TEST
 	default y if ARCH_TEGRA
 	help
 	  Adds AHB configuration functionality for NVIDIA Tegra SoCs,
diff --git a/drivers/amba/bus.c b/drivers/amba/bus.c
index 71482d639a6d..952c45ca6e48 100644
--- a/drivers/amba/bus.c
+++ b/drivers/amba/bus.c
@@ -138,7 +138,7 @@ static int amba_read_periphid(struct amba_device *dev)
 	void __iomem *tmp;
 	int i, ret;
 
-	ret = dev_pm_domain_attach(&dev->dev, true);
+	ret = dev_pm_domain_attach(&dev->dev, PD_FLAG_ATTACH_POWER_ON);
 	if (ret) {
 		dev_dbg(&dev->dev, "can't get PM domain: %d\n", ret);
 		goto err_out;
@@ -291,15 +291,14 @@ static int amba_probe(struct device *dev)
 		if (ret < 0)
 			break;
 
-		ret = dev_pm_domain_attach(dev, true);
+		ret = dev_pm_domain_attach(dev, PD_FLAG_ATTACH_POWER_ON |
+						PD_FLAG_DETACH_POWER_OFF);
 		if (ret)
 			break;
 
 		ret = amba_get_enable_pclk(pcdev);
-		if (ret) {
-			dev_pm_domain_detach(dev, true);
+		if (ret)
 			break;
-		}
 
 		pm_runtime_get_noresume(dev);
 		pm_runtime_set_active(dev);
@@ -314,7 +313,6 @@ static int amba_probe(struct device *dev)
 		pm_runtime_put_noidle(dev);
 
 		amba_put_disable_pclk(pcdev);
-		dev_pm_domain_detach(dev, true);
 	} while (0);
 
 	return ret;
@@ -336,7 +334,6 @@ static void amba_remove(struct device *dev)
 	pm_runtime_put_noidle(dev);
 
 	amba_put_disable_pclk(pcdev);
-	dev_pm_domain_detach(dev, true);
 }
 
 static void amba_shutdown(struct device *dev)
diff --git a/drivers/amba/tegra-ahb.c b/drivers/amba/tegra-ahb.c
index c0e8b765522d..f23c3ed01810 100644
--- a/drivers/amba/tegra-ahb.c
+++ b/drivers/amba/tegra-ahb.c
@@ -144,6 +144,7 @@ int tegra_ahb_enable_smmu(struct device_node *dn)
 	if (!dev)
 		return -EPROBE_DEFER;
 	ahb = dev_get_drvdata(dev);
+	put_device(dev);
 	val = gizmo_readl(ahb, AHB_ARBITRATION_XBAR_CTRL);
 	val |= AHB_ARBITRATION_XBAR_CTRL_SMMU_INIT_DONE;
 	gizmo_writel(ahb, val, AHB_ARBITRATION_XBAR_CTRL);
diff --git a/drivers/android/Kconfig b/drivers/android/Kconfig
index 07aa8ae0a058..e2e402c9d175 100644
--- a/drivers/android/Kconfig
+++ b/drivers/android/Kconfig
@@ -4,6 +4,7 @@ menu "Android"
 config ANDROID_BINDER_IPC
 	bool "Android Binder IPC Driver"
 	depends on MMU
+	depends on NET
 	default n
 	help
 	  Binder is used in Android for both communication between processes,
@@ -13,6 +14,19 @@ config ANDROID_BINDER_IPC
 	  Android process, using Binder to identify, invoke and pass arguments
 	  between said processes.
 
+config ANDROID_BINDER_IPC_RUST
+	bool "Rust version of Android Binder IPC Driver"
+	depends on RUST && MMU && !ANDROID_BINDER_IPC
+	help
+	  This enables the Rust implementation of the Binder driver.
+
+	  Binder is used in Android for both communication between processes,
+	  and remote method invocation.
+
+	  This means one Android process can call a method/routine in another
+	  Android process, using Binder to identify, invoke and pass arguments
+	  between said processes.
+
 config ANDROID_BINDERFS
 	bool "Android Binderfs filesystem"
 	depends on ANDROID_BINDER_IPC
@@ -27,7 +41,7 @@ config ANDROID_BINDERFS
 
 config ANDROID_BINDER_DEVICES
 	string "Android Binder devices"
-	depends on ANDROID_BINDER_IPC
+	depends on ANDROID_BINDER_IPC || ANDROID_BINDER_IPC_RUST
 	default "binder,hwbinder,vndbinder"
 	help
 	  Default value for the binder.devices parameter.
@@ -37,14 +51,15 @@ config ANDROID_BINDER_DEVICES
 	  created. Each binder device has its own context manager, and is
 	  therefore logically separated from the other devices.
 
-config ANDROID_BINDER_IPC_SELFTEST
-	bool "Android Binder IPC Driver Selftest"
-	depends on ANDROID_BINDER_IPC
+config ANDROID_BINDER_ALLOC_KUNIT_TEST
+	tristate "KUnit Tests for Android Binder Alloc" if !KUNIT_ALL_TESTS
+	depends on ANDROID_BINDER_IPC && KUNIT
+	default KUNIT_ALL_TESTS
 	help
-	  This feature allows binder selftest to run.
+	  This feature builds the binder alloc KUnit tests.
 
-	  Binder selftest checks the allocation and free of binder buffers
-	  exhaustively with combinations of various buffer sizes and
-	  alignments.
+	  Each test case runs using a pared-down binder_alloc struct and
+	  test-specific freelist, which allows this KUnit module to be loaded
+	  for testing without interfering with a running system.
 
 endmenu
diff --git a/drivers/android/Makefile b/drivers/android/Makefile
index c9d3d0c99c25..e0c650d3898e 100644
--- a/drivers/android/Makefile
+++ b/drivers/android/Makefile
@@ -2,5 +2,6 @@
 ccflags-y += -I$(src)			# needed for trace events
 
 obj-$(CONFIG_ANDROID_BINDERFS)		+= binderfs.o
-obj-$(CONFIG_ANDROID_BINDER_IPC)	+= binder.o binder_alloc.o
-obj-$(CONFIG_ANDROID_BINDER_IPC_SELFTEST) += binder_alloc_selftest.o
+obj-$(CONFIG_ANDROID_BINDER_IPC)	+= binder.o binder_alloc.o binder_netlink.o
+obj-$(CONFIG_ANDROID_BINDER_ALLOC_KUNIT_TEST)	+= tests/
+obj-$(CONFIG_ANDROID_BINDER_IPC_RUST)	+= binder/
diff --git a/drivers/android/binder.c b/drivers/android/binder.c
index c463ca4a8fff..535fc881c8da 100644
--- a/drivers/android/binder.c
+++ b/drivers/android/binder.c
@@ -68,10 +68,13 @@
 #include <linux/sizes.h>
 #include <linux/ktime.h>
 
+#include <kunit/visibility.h>
+
 #include <uapi/linux/android/binder.h>
 
 #include <linux/cacheflush.h>
 
+#include "binder_netlink.h"
 #include "binder_internal.h"
 #include "binder_trace.h"
 
@@ -848,17 +851,8 @@ static int binder_inc_node_nilocked(struct binder_node *node, int strong,
 	} else {
 		if (!internal)
 			node->local_weak_refs++;
-		if (!node->has_weak_ref && list_empty(&node->work.entry)) {
-			if (target_list == NULL) {
-				pr_err("invalid inc weak node for %d\n",
-					node->debug_id);
-				return -EINVAL;
-			}
-			/*
-			 * See comment above
-			 */
+		if (!node->has_weak_ref && target_list && list_empty(&node->work.entry))
 			binder_enqueue_work_ilocked(&node->work, target_list);
-		}
 	}
 	return 0;
 }
@@ -1585,11 +1579,10 @@ static struct binder_thread *binder_get_txn_from(
 {
 	struct binder_thread *from;
 
-	spin_lock(&t->lock);
+	guard(spinlock)(&t->lock);
 	from = t->from;
 	if (from)
 		atomic_inc(&from->tmp_ref);
-	spin_unlock(&t->lock);
 	return from;
 }
 
@@ -2416,10 +2409,10 @@ err_fd_not_accepted:
 
 /**
  * struct binder_ptr_fixup - data to be fixed-up in target buffer
- * @offset	offset in target buffer to fixup
- * @skip_size	bytes to skip in copy (fixup will be written later)
- * @fixup_data	data to write at fixup offset
- * @node	list node
+ * @offset:      offset in target buffer to fixup
+ * @skip_size:   bytes to skip in copy (fixup will be written later)
+ * @fixup_data:  data to write at fixup offset
+ * @node:        list node
  *
  * This is used for the pointer fixup list (pf) which is created and consumed
  * during binder_transaction() and is only accessed locally. No
@@ -2436,10 +2429,10 @@ struct binder_ptr_fixup {
 
 /**
  * struct binder_sg_copy - scatter-gather data to be copied
- * @offset		offset in target buffer
- * @sender_uaddr	user address in source buffer
- * @length		bytes to copy
- * @node		list node
+ * @offset:        offset in target buffer
+ * @sender_uaddr:  user address in source buffer
+ * @length:        bytes to copy
+ * @node:          list node
  *
  * This is used for the sg copy list (sgc) which is created and consumed
  * during binder_transaction() and is only accessed locally. No
@@ -2992,6 +2985,69 @@ static void binder_set_txn_from_error(struct binder_transaction *t, int id,
 	binder_thread_dec_tmpref(from);
 }
 
+/**
+ * binder_netlink_report() - report a transaction failure via netlink
+ * @proc:	the binder proc sending the transaction
+ * @t:		the binder transaction that failed
+ * @data_size:	the user provided data size for the transaction
+ * @error:	enum binder_driver_return_protocol returned to sender
+ */
+static void binder_netlink_report(struct binder_proc *proc,
+				  struct binder_transaction *t,
+				  u32 data_size,
+				  u32 error)
+{
+	const char *context = proc->context->name;
+	struct sk_buff *skb;
+	void *hdr;
+
+	if (!genl_has_listeners(&binder_nl_family, &init_net,
+				BINDER_NLGRP_REPORT))
+		return;
+
+	trace_binder_netlink_report(context, t, data_size, error);
+
+	skb = genlmsg_new(GENLMSG_DEFAULT_SIZE, GFP_KERNEL);
+	if (!skb)
+		return;
+
+	hdr = genlmsg_put(skb, 0, 0, &binder_nl_family, 0, BINDER_CMD_REPORT);
+	if (!hdr)
+		goto free_skb;
+
+	if (nla_put_u32(skb, BINDER_A_REPORT_ERROR, error) ||
+	    nla_put_string(skb, BINDER_A_REPORT_CONTEXT, context) ||
+	    nla_put_u32(skb, BINDER_A_REPORT_FROM_PID, t->from_pid) ||
+	    nla_put_u32(skb, BINDER_A_REPORT_FROM_TID, t->from_tid))
+		goto cancel_skb;
+
+	if (t->to_proc &&
+	    nla_put_u32(skb, BINDER_A_REPORT_TO_PID, t->to_proc->pid))
+		goto cancel_skb;
+
+	if (t->to_thread &&
+	    nla_put_u32(skb, BINDER_A_REPORT_TO_TID, t->to_thread->pid))
+		goto cancel_skb;
+
+	if (t->is_reply && nla_put_flag(skb, BINDER_A_REPORT_IS_REPLY))
+		goto cancel_skb;
+
+	if (nla_put_u32(skb, BINDER_A_REPORT_FLAGS, t->flags) ||
+	    nla_put_u32(skb, BINDER_A_REPORT_CODE, t->code) ||
+	    nla_put_u32(skb, BINDER_A_REPORT_DATA_SIZE, data_size))
+		goto cancel_skb;
+
+	genlmsg_end(skb, hdr);
+	genlmsg_multicast(&binder_nl_family, skb, 0, BINDER_NLGRP_REPORT,
+			  GFP_KERNEL);
+	return;
+
+cancel_skb:
+	genlmsg_cancel(skb, hdr);
+free_skb:
+	nlmsg_free(skb);
+}
+
 static void binder_transaction(struct binder_proc *proc,
 			       struct binder_thread *thread,
 			       struct binder_transaction_data *tr, int reply,
@@ -3041,6 +3097,32 @@ static void binder_transaction(struct binder_proc *proc,
 	binder_set_extended_error(&thread->ee, t_debug_id, BR_OK, 0);
 	binder_inner_proc_unlock(proc);
 
+	t = kzalloc(sizeof(*t), GFP_KERNEL);
+	if (!t) {
+		binder_txn_error("%d:%d cannot allocate transaction\n",
+				 thread->pid, proc->pid);
+		return_error = BR_FAILED_REPLY;
+		return_error_param = -ENOMEM;
+		return_error_line = __LINE__;
+		goto err_alloc_t_failed;
+	}
+	INIT_LIST_HEAD(&t->fd_fixups);
+	binder_stats_created(BINDER_STAT_TRANSACTION);
+	spin_lock_init(&t->lock);
+	t->debug_id = t_debug_id;
+	t->start_time = t_start_time;
+	t->from_pid = proc->pid;
+	t->from_tid = thread->pid;
+	t->sender_euid = task_euid(proc->tsk);
+	t->code = tr->code;
+	t->flags = tr->flags;
+	t->priority = task_nice(current);
+	t->work.type = BINDER_WORK_TRANSACTION;
+	t->is_async = !reply && (tr->flags & TF_ONE_WAY);
+	t->is_reply = reply;
+	if (!reply && !(tr->flags & TF_ONE_WAY))
+		t->from = thread;
+
 	if (reply) {
 		binder_inner_proc_lock(proc);
 		in_reply_to = thread->transaction_stack;
@@ -3144,10 +3226,8 @@ static void binder_transaction(struct binder_proc *proc,
 		}
 		if (!target_node) {
 			binder_txn_error("%d:%d cannot find target node\n",
-				thread->pid, proc->pid);
-			/*
-			 * return_error is set above
-			 */
+					 proc->pid, thread->pid);
+			/* return_error is set above */
 			return_error_param = -EINVAL;
 			return_error_line = __LINE__;
 			goto err_dead_binder;
@@ -3229,24 +3309,13 @@ static void binder_transaction(struct binder_proc *proc,
 		}
 		binder_inner_proc_unlock(proc);
 	}
+
+	t->to_proc = target_proc;
+	t->to_thread = target_thread;
 	if (target_thread)
 		e->to_thread = target_thread->pid;
 	e->to_proc = target_proc->pid;
 
-	/* TODO: reuse incoming transaction for reply */
-	t = kzalloc(sizeof(*t), GFP_KERNEL);
-	if (t == NULL) {
-		binder_txn_error("%d:%d cannot allocate transaction\n",
-			thread->pid, proc->pid);
-		return_error = BR_FAILED_REPLY;
-		return_error_param = -ENOMEM;
-		return_error_line = __LINE__;
-		goto err_alloc_t_failed;
-	}
-	INIT_LIST_HEAD(&t->fd_fixups);
-	binder_stats_created(BINDER_STAT_TRANSACTION);
-	spin_lock_init(&t->lock);
-
 	tcomplete = kzalloc(sizeof(*tcomplete), GFP_KERNEL);
 	if (tcomplete == NULL) {
 		binder_txn_error("%d:%d cannot allocate work for transaction\n",
@@ -3258,9 +3327,6 @@ static void binder_transaction(struct binder_proc *proc,
 	}
 	binder_stats_created(BINDER_STAT_TRANSACTION_COMPLETE);
 
-	t->debug_id = t_debug_id;
-	t->start_time = t_start_time;
-
 	if (reply)
 		binder_debug(BINDER_DEBUG_TRANSACTION,
 			     "%d:%d BC_REPLY %d -> %d:%d, data size %lld-%lld-%lld\n",
@@ -3276,19 +3342,6 @@ static void binder_transaction(struct binder_proc *proc,
 			     (u64)tr->data_size, (u64)tr->offsets_size,
 			     (u64)extra_buffers_size);
 
-	if (!reply && !(tr->flags & TF_ONE_WAY))
-		t->from = thread;
-	else
-		t->from = NULL;
-	t->from_pid = proc->pid;
-	t->from_tid = thread->pid;
-	t->sender_euid = task_euid(proc->tsk);
-	t->to_proc = target_proc;
-	t->to_thread = target_thread;
-	t->code = tr->code;
-	t->flags = tr->flags;
-	t->priority = task_nice(current);
-
 	if (target_node && target_node->txn_security_ctx) {
 		u32 secid;
 		size_t added_size;
@@ -3681,11 +3734,13 @@ static void binder_transaction(struct binder_proc *proc,
 		return_error_line = __LINE__;
 		goto err_copy_data_failed;
 	}
-	if (t->buffer->oneway_spam_suspect)
+	if (t->buffer->oneway_spam_suspect) {
 		tcomplete->type = BINDER_WORK_TRANSACTION_ONEWAY_SPAM_SUSPECT;
-	else
+		binder_netlink_report(proc, t, tr->data_size,
+				      BR_ONEWAY_SPAM_SUSPECT);
+	} else {
 		tcomplete->type = BINDER_WORK_TRANSACTION_COMPLETE;
-	t->work.type = BINDER_WORK_TRANSACTION;
+	}
 
 	if (reply) {
 		binder_enqueue_thread_work(thread, tcomplete);
@@ -3713,7 +3768,6 @@ static void binder_transaction(struct binder_proc *proc,
 		 * the target replies (or there is an error).
 		 */
 		binder_enqueue_deferred_thread_work_ilocked(thread, tcomplete);
-		t->need_reply = 1;
 		t->from_parent = thread->transaction_stack;
 		thread->transaction_stack = t;
 		binder_inner_proc_unlock(proc);
@@ -3734,8 +3788,11 @@ static void binder_transaction(struct binder_proc *proc,
 		 * process and is put in a pending queue, waiting for the target
 		 * process to be unfrozen.
 		 */
-		if (return_error == BR_TRANSACTION_PENDING_FROZEN)
+		if (return_error == BR_TRANSACTION_PENDING_FROZEN) {
 			tcomplete->type = BINDER_WORK_TRANSACTION_PENDING;
+			binder_netlink_report(proc, t, tr->data_size,
+					      return_error);
+		}
 		binder_enqueue_thread_work(thread, tcomplete);
 		if (return_error &&
 		    return_error != BR_TRANSACTION_PENDING_FROZEN)
@@ -3784,9 +3841,6 @@ err_get_secctx_failed:
 err_alloc_tcomplete_failed:
 	if (trace_binder_txn_latency_free_enabled())
 		binder_txn_latency_free(t);
-	kfree(t);
-	binder_stats_deleted(BINDER_STAT_TRANSACTION);
-err_alloc_t_failed:
 err_bad_todo_list:
 err_bad_call_stack:
 err_empty_call_stack:
@@ -3797,6 +3851,11 @@ err_invalid_target_handle:
 		binder_dec_node_tmpref(target_node);
 	}
 
+	binder_netlink_report(proc, t, tr->data_size, return_error);
+	kfree(t);
+	binder_stats_deleted(BINDER_STAT_TRANSACTION);
+err_alloc_t_failed:
+
 	binder_debug(BINDER_DEBUG_FAILED_TRANSACTION,
 		     "%d:%d transaction %s to %d:%d failed %d/%d/%d, code %u size %lld-%lld line %d\n",
 		     proc->pid, thread->pid, reply ? "reply" :
@@ -3995,14 +4054,15 @@ binder_freeze_notification_done(struct binder_proc *proc,
 
 /**
  * binder_free_buf() - free the specified buffer
- * @proc:	binder proc that owns buffer
- * @buffer:	buffer to be freed
- * @is_failure:	failed to send transaction
+ * @proc:       binder proc that owns buffer
+ * @thread:     binder thread performing the buffer release
+ * @buffer:     buffer to be freed
+ * @is_failure: failed to send transaction
  *
- * If buffer for an async transaction, enqueue the next async
+ * If the buffer is for an async transaction, enqueue the next async
  * transaction from the node.
  *
- * Cleanup buffer and free it.
+ * Cleanup the buffer and free it.
  */
 static void
 binder_free_buf(struct binder_proc *proc,
@@ -4609,6 +4669,8 @@ static int binder_wait_for_work(struct binder_thread *thread,
  *
  * If we fail to allocate an fd, skip the install and release
  * any fds that have already been allocated.
+ *
+ * Return: 0 on success, a negative errno code on failure.
  */
 static int binder_apply_fd_fixups(struct binder_proc *proc,
 				  struct binder_transaction *t)
@@ -5384,10 +5446,9 @@ static int binder_ioctl_write_read(struct file *filp, unsigned long arg,
 	void __user *ubuf = (void __user *)arg;
 	struct binder_write_read bwr;
 
-	if (copy_from_user(&bwr, ubuf, sizeof(bwr))) {
-		ret = -EFAULT;
-		goto out;
-	}
+	if (copy_from_user(&bwr, ubuf, sizeof(bwr)))
+		return -EFAULT;
+
 	binder_debug(BINDER_DEBUG_READ_WRITE,
 		     "%d:%d write %lld at %016llx, read %lld at %016llx\n",
 		     proc->pid, thread->pid,
@@ -5402,8 +5463,6 @@ static int binder_ioctl_write_read(struct file *filp, unsigned long arg,
 		trace_binder_write_done(ret);
 		if (ret < 0) {
 			bwr.read_consumed = 0;
-			if (copy_to_user(ubuf, &bwr, sizeof(bwr)))
-				ret = -EFAULT;
 			goto out;
 		}
 	}
@@ -5417,22 +5476,17 @@ static int binder_ioctl_write_read(struct file *filp, unsigned long arg,
 		if (!binder_worklist_empty_ilocked(&proc->todo))
 			binder_wakeup_proc_ilocked(proc);
 		binder_inner_proc_unlock(proc);
-		if (ret < 0) {
-			if (copy_to_user(ubuf, &bwr, sizeof(bwr)))
-				ret = -EFAULT;
+		if (ret < 0)
 			goto out;
-		}
 	}
 	binder_debug(BINDER_DEBUG_READ_WRITE,
 		     "%d:%d wrote %lld of %lld, read return %lld of %lld\n",
 		     proc->pid, thread->pid,
 		     (u64)bwr.write_consumed, (u64)bwr.write_size,
 		     (u64)bwr.read_consumed, (u64)bwr.read_size);
-	if (copy_to_user(ubuf, &bwr, sizeof(bwr))) {
-		ret = -EFAULT;
-		goto out;
-	}
 out:
+	if (copy_to_user(ubuf, &bwr, sizeof(bwr)))
+		ret = -EFAULT;
 	return ret;
 }
 
@@ -5445,32 +5499,28 @@ static int binder_ioctl_set_ctx_mgr(struct file *filp,
 	struct binder_node *new_node;
 	kuid_t curr_euid = current_euid();
 
-	mutex_lock(&context->context_mgr_node_lock);
+	guard(mutex)(&context->context_mgr_node_lock);
 	if (context->binder_context_mgr_node) {
 		pr_err("BINDER_SET_CONTEXT_MGR already set\n");
-		ret = -EBUSY;
-		goto out;
+		return -EBUSY;
 	}
 	ret = security_binder_set_context_mgr(proc->cred);
 	if (ret < 0)
-		goto out;
+		return ret;
 	if (uid_valid(context->binder_context_mgr_uid)) {
 		if (!uid_eq(context->binder_context_mgr_uid, curr_euid)) {
 			pr_err("BINDER_SET_CONTEXT_MGR bad uid %d != %d\n",
 			       from_kuid(&init_user_ns, curr_euid),
 			       from_kuid(&init_user_ns,
 					 context->binder_context_mgr_uid));
-			ret = -EPERM;
-			goto out;
+			return -EPERM;
 		}
 	} else {
 		context->binder_context_mgr_uid = curr_euid;
 	}
 	new_node = binder_new_node(proc, fbo);
-	if (!new_node) {
-		ret = -ENOMEM;
-		goto out;
-	}
+	if (!new_node)
+		return -ENOMEM;
 	binder_node_lock(new_node);
 	new_node->local_weak_refs++;
 	new_node->local_strong_refs++;
@@ -5479,8 +5529,6 @@ static int binder_ioctl_set_ctx_mgr(struct file *filp,
 	context->binder_context_mgr_node = new_node;
 	binder_node_unlock(new_node);
 	binder_put_node(new_node);
-out:
-	mutex_unlock(&context->context_mgr_node_lock);
 	return ret;
 }
 
@@ -5716,11 +5764,6 @@ static long binder_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 	struct binder_thread *thread;
 	void __user *ubuf = (void __user *)arg;
 
-	/*pr_info("binder_ioctl: %d:%d %x %lx\n",
-			proc->pid, current->pid, cmd, arg);*/
-
-	binder_selftest_alloc(&proc->alloc);
-
 	trace_binder_ioctl(cmd, arg);
 
 	ret = wait_event_interruptible(binder_user_error_wait, binder_stop_on_user_error < 2);
@@ -5956,10 +5999,11 @@ static void binder_vma_close(struct vm_area_struct *vma)
 	binder_alloc_vma_close(&proc->alloc);
 }
 
-static vm_fault_t binder_vm_fault(struct vm_fault *vmf)
+VISIBLE_IF_KUNIT vm_fault_t binder_vm_fault(struct vm_fault *vmf)
 {
 	return VM_FAULT_SIGBUS;
 }
+EXPORT_SYMBOL_IF_KUNIT(binder_vm_fault);
 
 static const struct vm_operations_struct binder_vm_ops = {
 	.open = binder_vma_open,
@@ -6128,7 +6172,7 @@ static int binder_release(struct inode *nodp, struct file *filp)
 	debugfs_remove(proc->debugfs_entry);
 
 	if (proc->binderfs_entry) {
-		binderfs_remove_file(proc->binderfs_entry);
+		simple_recursive_removal(proc->binderfs_entry, NULL);
 		proc->binderfs_entry = NULL;
 	}
 
@@ -6322,14 +6366,13 @@ static DECLARE_WORK(binder_deferred_work, binder_deferred_func);
 static void
 binder_defer_work(struct binder_proc *proc, enum binder_deferred_state defer)
 {
-	mutex_lock(&binder_deferred_lock);
+	guard(mutex)(&binder_deferred_lock);
 	proc->deferred_work |= defer;
 	if (hlist_unhashed(&proc->deferred_work_node)) {
 		hlist_add_head(&proc->deferred_work_node,
 				&binder_deferred_list);
 		schedule_work(&binder_deferred_work);
 	}
-	mutex_unlock(&binder_deferred_lock);
 }
 
 static void print_binder_transaction_ilocked(struct seq_file *m,
@@ -6344,13 +6387,13 @@ static void print_binder_transaction_ilocked(struct seq_file *m,
 	spin_lock(&t->lock);
 	to_proc = t->to_proc;
 	seq_printf(m,
-		   "%s %d: %pK from %d:%d to %d:%d code %x flags %x pri %ld r%d elapsed %lldms",
+		   "%s %d: %pK from %d:%d to %d:%d code %x flags %x pri %ld a%d r%d elapsed %lldms",
 		   prefix, t->debug_id, t,
 		   t->from_pid,
 		   t->from_tid,
 		   to_proc ? to_proc->pid : 0,
 		   t->to_thread ? t->to_thread->pid : 0,
-		   t->code, t->flags, t->priority, t->need_reply,
+		   t->code, t->flags, t->priority, t->is_async, t->is_reply,
 		   ktime_ms_delta(current_time, t->start_time));
 	spin_unlock(&t->lock);
 
@@ -6871,14 +6914,13 @@ static int proc_show(struct seq_file *m, void *unused)
 	struct binder_proc *itr;
 	int pid = (unsigned long)m->private;
 
-	mutex_lock(&binder_procs_lock);
+	guard(mutex)(&binder_procs_lock);
 	hlist_for_each_entry(itr, &binder_procs, proc_node) {
 		if (itr->pid == pid) {
 			seq_puts(m, "binder proc state:\n");
 			print_binder_proc(m, itr, true, false);
 		}
 	}
-	mutex_unlock(&binder_procs_lock);
 
 	return 0;
 }
@@ -6996,16 +7038,14 @@ const struct binder_debugfs_entry binder_debugfs_entries[] = {
 
 void binder_add_device(struct binder_device *device)
 {
-	spin_lock(&binder_devices_lock);
+	guard(spinlock)(&binder_devices_lock);
 	hlist_add_head(&device->hlist, &binder_devices);
-	spin_unlock(&binder_devices_lock);
 }
 
 void binder_remove_device(struct binder_device *device)
 {
-	spin_lock(&binder_devices_lock);
+	guard(spinlock)(&binder_devices_lock);
 	hlist_del_init(&device->hlist);
-	spin_unlock(&binder_devices_lock);
 }
 
 static int __init init_binder_device(const char *name)
@@ -7085,12 +7125,19 @@ static int __init binder_init(void)
 		}
 	}
 
-	ret = init_binderfs();
+	ret = genl_register_family(&binder_nl_family);
 	if (ret)
 		goto err_init_binder_device_failed;
 
+	ret = init_binderfs();
+	if (ret)
+		goto err_init_binderfs_failed;
+
 	return ret;
 
+err_init_binderfs_failed:
+	genl_unregister_family(&binder_nl_family);
+
 err_init_binder_device_failed:
 	hlist_for_each_entry_safe(device, tmp, &binder_devices, hlist) {
 		misc_deregister(&device->miscdev);
@@ -7111,5 +7158,3 @@ device_initcall(binder_init);
 
 #define CREATE_TRACE_POINTS
 #include "binder_trace.h"
-
-MODULE_LICENSE("GPL v2");
diff --git a/drivers/android/binder/Makefile b/drivers/android/binder/Makefile
new file mode 100644
index 000000000000..09eabb527fa0
--- /dev/null
+++ b/drivers/android/binder/Makefile
@@ -0,0 +1,9 @@
+# SPDX-License-Identifier: GPL-2.0-only
+ccflags-y += -I$(src)	# needed for trace events
+
+obj-$(CONFIG_ANDROID_BINDER_IPC_RUST) += rust_binder.o
+rust_binder-y := \
+	rust_binder_main.o	\
+	rust_binderfs.o		\
+	rust_binder_events.o	\
+	page_range_helper.o
diff --git a/drivers/android/binder/allocation.rs b/drivers/android/binder/allocation.rs
new file mode 100644
index 000000000000..7f65a9c3a0e5
--- /dev/null
+++ b/drivers/android/binder/allocation.rs
@@ -0,0 +1,602 @@
+// SPDX-License-Identifier: GPL-2.0
+
+// Copyright (C) 2025 Google LLC.
+
+use core::mem::{size_of, size_of_val, MaybeUninit};
+use core::ops::Range;
+
+use kernel::{
+    bindings,
+    fs::file::{File, FileDescriptorReservation},
+    prelude::*,
+    sync::{aref::ARef, Arc},
+    transmute::{AsBytes, FromBytes},
+    uaccess::UserSliceReader,
+    uapi,
+};
+
+use crate::{
+    deferred_close::DeferredFdCloser,
+    defs::*,
+    node::{Node, NodeRef},
+    process::Process,
+    DArc,
+};
+
+#[derive(Default)]
+pub(crate) struct AllocationInfo {
+    /// Range within the allocation where we can find the offsets to the object descriptors.
+    pub(crate) offsets: Option<Range<usize>>,
+    /// The target node of the transaction this allocation is associated to.
+    /// Not set for replies.
+    pub(crate) target_node: Option<NodeRef>,
+    /// When this allocation is dropped, call `pending_oneway_finished` on the node.
+    ///
+    /// This is used to serialize oneway transaction on the same node. Binder guarantees that
+    /// oneway transactions to the same node are delivered sequentially in the order they are sent.
+    pub(crate) oneway_node: Option<DArc<Node>>,
+    /// Zero the data in the buffer on free.
+    pub(crate) clear_on_free: bool,
+    /// List of files embedded in this transaction.
+    file_list: FileList,
+}
+
+/// Represents an allocation that the kernel is currently using.
+///
+/// When allocations are idle, the range allocator holds the data related to them.
+///
+/// # Invariants
+///
+/// This allocation corresponds to an allocation in the range allocator, so the relevant pages are
+/// marked in use in the page range.
+pub(crate) struct Allocation {
+    pub(crate) offset: usize,
+    size: usize,
+    pub(crate) ptr: usize,
+    pub(crate) process: Arc<Process>,
+    allocation_info: Option<AllocationInfo>,
+    free_on_drop: bool,
+    pub(crate) oneway_spam_detected: bool,
+    #[allow(dead_code)]
+    pub(crate) debug_id: usize,
+}
+
+impl Allocation {
+    pub(crate) fn new(
+        process: Arc<Process>,
+        debug_id: usize,
+        offset: usize,
+        size: usize,
+        ptr: usize,
+        oneway_spam_detected: bool,
+    ) -> Self {
+        Self {
+            process,
+            offset,
+            size,
+            ptr,
+            debug_id,
+            oneway_spam_detected,
+            allocation_info: None,
+            free_on_drop: true,
+        }
+    }
+
+    fn size_check(&self, offset: usize, size: usize) -> Result {
+        let overflow_fail = offset.checked_add(size).is_none();
+        let cmp_size_fail = offset.wrapping_add(size) > self.size;
+        if overflow_fail || cmp_size_fail {
+            return Err(EFAULT);
+        }
+        Ok(())
+    }
+
+    pub(crate) fn copy_into(
+        &self,
+        reader: &mut UserSliceReader,
+        offset: usize,
+        size: usize,
+    ) -> Result {
+        self.size_check(offset, size)?;
+
+        // SAFETY: While this object exists, the range allocator will keep the range allocated, and
+        // in turn, the pages will be marked as in use.
+        unsafe {
+            self.process
+                .pages
+                .copy_from_user_slice(reader, self.offset + offset, size)
+        }
+    }
+
+    pub(crate) fn read<T: FromBytes>(&self, offset: usize) -> Result<T> {
+        self.size_check(offset, size_of::<T>())?;
+
+        // SAFETY: While this object exists, the range allocator will keep the range allocated, and
+        // in turn, the pages will be marked as in use.
+        unsafe { self.process.pages.read(self.offset + offset) }
+    }
+
+    pub(crate) fn write<T: ?Sized>(&self, offset: usize, obj: &T) -> Result {
+        self.size_check(offset, size_of_val::<T>(obj))?;
+
+        // SAFETY: While this object exists, the range allocator will keep the range allocated, and
+        // in turn, the pages will be marked as in use.
+        unsafe { self.process.pages.write(self.offset + offset, obj) }
+    }
+
+    pub(crate) fn fill_zero(&self) -> Result {
+        // SAFETY: While this object exists, the range allocator will keep the range allocated, and
+        // in turn, the pages will be marked as in use.
+        unsafe { self.process.pages.fill_zero(self.offset, self.size) }
+    }
+
+    pub(crate) fn keep_alive(mut self) {
+        self.process
+            .buffer_make_freeable(self.offset, self.allocation_info.take());
+        self.free_on_drop = false;
+    }
+
+    pub(crate) fn set_info(&mut self, info: AllocationInfo) {
+        self.allocation_info = Some(info);
+    }
+
+    pub(crate) fn get_or_init_info(&mut self) -> &mut AllocationInfo {
+        self.allocation_info.get_or_insert_with(Default::default)
+    }
+
+    pub(crate) fn set_info_offsets(&mut self, offsets: Range<usize>) {
+        self.get_or_init_info().offsets = Some(offsets);
+    }
+
+    pub(crate) fn set_info_oneway_node(&mut self, oneway_node: DArc<Node>) {
+        self.get_or_init_info().oneway_node = Some(oneway_node);
+    }
+
+    pub(crate) fn set_info_clear_on_drop(&mut self) {
+        self.get_or_init_info().clear_on_free = true;
+    }
+
+    pub(crate) fn set_info_target_node(&mut self, target_node: NodeRef) {
+        self.get_or_init_info().target_node = Some(target_node);
+    }
+
+    /// Reserve enough space to push at least `num_fds` fds.
+    pub(crate) fn info_add_fd_reserve(&mut self, num_fds: usize) -> Result {
+        self.get_or_init_info()
+            .file_list
+            .files_to_translate
+            .reserve(num_fds, GFP_KERNEL)?;
+
+        Ok(())
+    }
+
+    pub(crate) fn info_add_fd(
+        &mut self,
+        file: ARef<File>,
+        buffer_offset: usize,
+        close_on_free: bool,
+    ) -> Result {
+        self.get_or_init_info().file_list.files_to_translate.push(
+            FileEntry {
+                file,
+                buffer_offset,
+                close_on_free,
+            },
+            GFP_KERNEL,
+        )?;
+
+        Ok(())
+    }
+
+    pub(crate) fn set_info_close_on_free(&mut self, cof: FdsCloseOnFree) {
+        self.get_or_init_info().file_list.close_on_free = cof.0;
+    }
+
+    pub(crate) fn translate_fds(&mut self) -> Result<TranslatedFds> {
+        let file_list = match self.allocation_info.as_mut() {
+            Some(info) => &mut info.file_list,
+            None => return Ok(TranslatedFds::new()),
+        };
+
+        let files = core::mem::take(&mut file_list.files_to_translate);
+
+        let num_close_on_free = files.iter().filter(|entry| entry.close_on_free).count();
+        let mut close_on_free = KVec::with_capacity(num_close_on_free, GFP_KERNEL)?;
+
+        let mut reservations = KVec::with_capacity(files.len(), GFP_KERNEL)?;
+        for file_info in files {
+            let res = FileDescriptorReservation::get_unused_fd_flags(bindings::O_CLOEXEC)?;
+            let fd = res.reserved_fd();
+            self.write::<u32>(file_info.buffer_offset, &fd)?;
+
+            reservations.push(
+                Reservation {
+                    res,
+                    file: file_info.file,
+                },
+                GFP_KERNEL,
+            )?;
+            if file_info.close_on_free {
+                close_on_free.push(fd, GFP_KERNEL)?;
+            }
+        }
+
+        Ok(TranslatedFds {
+            reservations,
+            close_on_free: FdsCloseOnFree(close_on_free),
+        })
+    }
+
+    /// Should the looper return to userspace when freeing this allocation?
+    pub(crate) fn looper_need_return_on_free(&self) -> bool {
+        // Closing fds involves pushing task_work for execution when we return to userspace. Hence,
+        // we should return to userspace asap if we are closing fds.
+        match self.allocation_info {
+            Some(ref info) => !info.file_list.close_on_free.is_empty(),
+            None => false,
+        }
+    }
+}
+
+impl Drop for Allocation {
+    fn drop(&mut self) {
+        if !self.free_on_drop {
+            return;
+        }
+
+        if let Some(mut info) = self.allocation_info.take() {
+            if let Some(oneway_node) = info.oneway_node.as_ref() {
+                oneway_node.pending_oneway_finished();
+            }
+
+            info.target_node = None;
+
+            if let Some(offsets) = info.offsets.clone() {
+                let view = AllocationView::new(self, offsets.start);
+                for i in offsets.step_by(size_of::<usize>()) {
+                    if view.cleanup_object(i).is_err() {
+                        pr_warn!("Error cleaning up object at offset {}\n", i)
+                    }
+                }
+            }
+
+            for &fd in &info.file_list.close_on_free {
+                let closer = match DeferredFdCloser::new(GFP_KERNEL) {
+                    Ok(closer) => closer,
+                    Err(kernel::alloc::AllocError) => {
+                        // Ignore allocation failures.
+                        break;
+                    }
+                };
+
+                // Here, we ignore errors. The operation can fail if the fd is not valid, or if the
+                // method is called from a kthread. However, this is always called from a syscall,
+                // so the latter case cannot happen, and we don't care about the first case.
+                let _ = closer.close_fd(fd);
+            }
+
+            if info.clear_on_free {
+                if let Err(e) = self.fill_zero() {
+                    pr_warn!("Failed to clear data on free: {:?}", e);
+                }
+            }
+        }
+
+        self.process.buffer_raw_free(self.ptr);
+    }
+}
+
+/// A wrapper around `Allocation` that is being created.
+///
+/// If the allocation is destroyed while wrapped in this wrapper, then the allocation will be
+/// considered to be part of a failed transaction. Successful transactions avoid that by calling
+/// `success`, which skips the destructor.
+#[repr(transparent)]
+pub(crate) struct NewAllocation(pub(crate) Allocation);
+
+impl NewAllocation {
+    pub(crate) fn success(self) -> Allocation {
+        // This skips the destructor.
+        //
+        // SAFETY: This type is `#[repr(transparent)]`, so the layout matches.
+        unsafe { core::mem::transmute(self) }
+    }
+}
+
+impl core::ops::Deref for NewAllocation {
+    type Target = Allocation;
+    fn deref(&self) -> &Allocation {
+        &self.0
+    }
+}
+
+impl core::ops::DerefMut for NewAllocation {
+    fn deref_mut(&mut self) -> &mut Allocation {
+        &mut self.0
+    }
+}
+
+/// A view into the beginning of an allocation.
+///
+/// All attempts to read or write outside of the view will fail. To intentionally access outside of
+/// this view, use the `alloc` field of this struct directly.
+pub(crate) struct AllocationView<'a> {
+    pub(crate) alloc: &'a mut Allocation,
+    limit: usize,
+}
+
+impl<'a> AllocationView<'a> {
+    pub(crate) fn new(alloc: &'a mut Allocation, limit: usize) -> Self {
+        AllocationView { alloc, limit }
+    }
+
+    pub(crate) fn read<T: FromBytes>(&self, offset: usize) -> Result<T> {
+        if offset.checked_add(size_of::<T>()).ok_or(EINVAL)? > self.limit {
+            return Err(EINVAL);
+        }
+        self.alloc.read(offset)
+    }
+
+    pub(crate) fn write<T: AsBytes>(&self, offset: usize, obj: &T) -> Result {
+        if offset.checked_add(size_of::<T>()).ok_or(EINVAL)? > self.limit {
+            return Err(EINVAL);
+        }
+        self.alloc.write(offset, obj)
+    }
+
+    pub(crate) fn copy_into(
+        &self,
+        reader: &mut UserSliceReader,
+        offset: usize,
+        size: usize,
+    ) -> Result {
+        if offset.checked_add(size).ok_or(EINVAL)? > self.limit {
+            return Err(EINVAL);
+        }
+        self.alloc.copy_into(reader, offset, size)
+    }
+
+    pub(crate) fn transfer_binder_object(
+        &self,
+        offset: usize,
+        obj: &uapi::flat_binder_object,
+        strong: bool,
+        node_ref: NodeRef,
+    ) -> Result {
+        let mut newobj = FlatBinderObject::default();
+        let node = node_ref.node.clone();
+        if Arc::ptr_eq(&node_ref.node.owner, &self.alloc.process) {
+            // The receiving process is the owner of the node, so send it a binder object (instead
+            // of a handle).
+            let (ptr, cookie) = node.get_id();
+            newobj.hdr.type_ = if strong {
+                BINDER_TYPE_BINDER
+            } else {
+                BINDER_TYPE_WEAK_BINDER
+            };
+            newobj.flags = obj.flags;
+            newobj.__bindgen_anon_1.binder = ptr as _;
+            newobj.cookie = cookie as _;
+            self.write(offset, &newobj)?;
+            // Increment the user ref count on the node. It will be decremented as part of the
+            // destruction of the buffer, when we see a binder or weak-binder object.
+            node.update_refcount(true, 1, strong);
+        } else {
+            // The receiving process is different from the owner, so we need to insert a handle to
+            // the binder object.
+            let handle = self
+                .alloc
+                .process
+                .as_arc_borrow()
+                .insert_or_update_handle(node_ref, false)?;
+            newobj.hdr.type_ = if strong {
+                BINDER_TYPE_HANDLE
+            } else {
+                BINDER_TYPE_WEAK_HANDLE
+            };
+            newobj.flags = obj.flags;
+            newobj.__bindgen_anon_1.handle = handle;
+            if self.write(offset, &newobj).is_err() {
+                // Decrement ref count on the handle we just created.
+                let _ = self
+                    .alloc
+                    .process
+                    .as_arc_borrow()
+                    .update_ref(handle, false, strong);
+                return Err(EINVAL);
+            }
+        }
+
+        Ok(())
+    }
+
+    fn cleanup_object(&self, index_offset: usize) -> Result {
+        let offset = self.alloc.read(index_offset)?;
+        let header = self.read::<BinderObjectHeader>(offset)?;
+        match header.type_ {
+            BINDER_TYPE_WEAK_BINDER | BINDER_TYPE_BINDER => {
+                let obj = self.read::<FlatBinderObject>(offset)?;
+                let strong = header.type_ == BINDER_TYPE_BINDER;
+                // SAFETY: The type is `BINDER_TYPE_{WEAK_}BINDER`, so the `binder` field is
+                // populated.
+                let ptr = unsafe { obj.__bindgen_anon_1.binder };
+                let cookie = obj.cookie;
+                self.alloc.process.update_node(ptr, cookie, strong);
+                Ok(())
+            }
+            BINDER_TYPE_WEAK_HANDLE | BINDER_TYPE_HANDLE => {
+                let obj = self.read::<FlatBinderObject>(offset)?;
+                let strong = header.type_ == BINDER_TYPE_HANDLE;
+                // SAFETY: The type is `BINDER_TYPE_{WEAK_}HANDLE`, so the `handle` field is
+                // populated.
+                let handle = unsafe { obj.__bindgen_anon_1.handle };
+                self.alloc
+                    .process
+                    .as_arc_borrow()
+                    .update_ref(handle, false, strong)
+            }
+            _ => Ok(()),
+        }
+    }
+}
+
+/// A binder object as it is serialized.
+///
+/// # Invariants
+///
+/// All bytes must be initialized, and the value of `self.hdr.type_` must be one of the allowed
+/// types.
+#[repr(C)]
+pub(crate) union BinderObject {
+    hdr: uapi::binder_object_header,
+    fbo: uapi::flat_binder_object,
+    fdo: uapi::binder_fd_object,
+    bbo: uapi::binder_buffer_object,
+    fdao: uapi::binder_fd_array_object,
+}
+
+/// A view into a `BinderObject` that can be used in a match statement.
+pub(crate) enum BinderObjectRef<'a> {
+    Binder(&'a mut uapi::flat_binder_object),
+    Handle(&'a mut uapi::flat_binder_object),
+    Fd(&'a mut uapi::binder_fd_object),
+    Ptr(&'a mut uapi::binder_buffer_object),
+    Fda(&'a mut uapi::binder_fd_array_object),
+}
+
+impl BinderObject {
+    pub(crate) fn read_from(reader: &mut UserSliceReader) -> Result<BinderObject> {
+        let object = Self::read_from_inner(|slice| {
+            let read_len = usize::min(slice.len(), reader.len());
+            reader.clone_reader().read_slice(&mut slice[..read_len])?;
+            Ok(())
+        })?;
+
+        // If we used a object type smaller than the largest object size, then we've read more
+        // bytes than we needed to. However, we used `.clone_reader()` to avoid advancing the
+        // original reader. Now, we call `skip` so that the caller's reader is advanced by the
+        // right amount.
+        //
+        // The `skip` call fails if the reader doesn't have `size` bytes available. This could
+        // happen if the type header corresponds to an object type that is larger than the rest of
+        // the reader.
+        //
+        // Any extra bytes beyond the size of the object are inaccessible after this call, so
+        // reading them again from the `reader` later does not result in TOCTOU bugs.
+        reader.skip(object.size())?;
+
+        Ok(object)
+    }
+
+    /// Use the provided reader closure to construct a `BinderObject`.
+    ///
+    /// The closure should write the bytes for the object into the provided slice.
+    pub(crate) fn read_from_inner<R>(reader: R) -> Result<BinderObject>
+    where
+        R: FnOnce(&mut [u8; size_of::<BinderObject>()]) -> Result<()>,
+    {
+        let mut obj = MaybeUninit::<BinderObject>::zeroed();
+
+        // SAFETY: The lengths of `BinderObject` and `[u8; size_of::<BinderObject>()]` are equal,
+        // and the byte array has an alignment requirement of one, so the pointer cast is okay.
+        // Additionally, `obj` was initialized to zeros, so the byte array will not be
+        // uninitialized.
+        (reader)(unsafe { &mut *obj.as_mut_ptr().cast() })?;
+
+        // SAFETY: The entire object is initialized, so accessing this field is safe.
+        let type_ = unsafe { obj.assume_init_ref().hdr.type_ };
+        if Self::type_to_size(type_).is_none() {
+            // The value of `obj.hdr_type_` was invalid.
+            return Err(EINVAL);
+        }
+
+        // SAFETY: All bytes are initialized (since we zeroed them at the start) and we checked
+        // that `self.hdr.type_` is one of the allowed types, so the type invariants are satisfied.
+        unsafe { Ok(obj.assume_init()) }
+    }
+
+    pub(crate) fn as_ref(&mut self) -> BinderObjectRef<'_> {
+        use BinderObjectRef::*;
+        // SAFETY: The constructor ensures that all bytes of `self` are initialized, and all
+        // variants of this union accept all initialized bit patterns.
+        unsafe {
+            match self.hdr.type_ {
+                BINDER_TYPE_WEAK_BINDER | BINDER_TYPE_BINDER => Binder(&mut self.fbo),
+                BINDER_TYPE_WEAK_HANDLE | BINDER_TYPE_HANDLE => Handle(&mut self.fbo),
+                BINDER_TYPE_FD => Fd(&mut self.fdo),
+                BINDER_TYPE_PTR => Ptr(&mut self.bbo),
+                BINDER_TYPE_FDA => Fda(&mut self.fdao),
+                // SAFETY: By the type invariant, the value of `self.hdr.type_` cannot have any
+                // other value than the ones checked above.
+                _ => core::hint::unreachable_unchecked(),
+            }
+        }
+    }
+
+    pub(crate) fn size(&self) -> usize {
+        // SAFETY: The entire object is initialized, so accessing this field is safe.
+        let type_ = unsafe { self.hdr.type_ };
+
+        // SAFETY: The type invariants guarantee that the type field is correct.
+        unsafe { Self::type_to_size(type_).unwrap_unchecked() }
+    }
+
+    fn type_to_size(type_: u32) -> Option<usize> {
+        match type_ {
+            BINDER_TYPE_WEAK_BINDER => Some(size_of::<uapi::flat_binder_object>()),
+            BINDER_TYPE_BINDER => Some(size_of::<uapi::flat_binder_object>()),
+            BINDER_TYPE_WEAK_HANDLE => Some(size_of::<uapi::flat_binder_object>()),
+            BINDER_TYPE_HANDLE => Some(size_of::<uapi::flat_binder_object>()),
+            BINDER_TYPE_FD => Some(size_of::<uapi::binder_fd_object>()),
+            BINDER_TYPE_PTR => Some(size_of::<uapi::binder_buffer_object>()),
+            BINDER_TYPE_FDA => Some(size_of::<uapi::binder_fd_array_object>()),
+            _ => None,
+        }
+    }
+}
+
+#[derive(Default)]
+struct FileList {
+    files_to_translate: KVec<FileEntry>,
+    close_on_free: KVec<u32>,
+}
+
+struct FileEntry {
+    /// The file for which a descriptor will be created in the recipient process.
+    file: ARef<File>,
+    /// The offset in the buffer where the file descriptor is stored.
+    buffer_offset: usize,
+    /// Whether this fd should be closed when the allocation is freed.
+    close_on_free: bool,
+}
+
+pub(crate) struct TranslatedFds {
+    reservations: KVec<Reservation>,
+    /// If commit is called, then these fds should be closed. (If commit is not called, then they
+    /// shouldn't be closed.)
+    close_on_free: FdsCloseOnFree,
+}
+
+struct Reservation {
+    res: FileDescriptorReservation,
+    file: ARef<File>,
+}
+
+impl TranslatedFds {
+    pub(crate) fn new() -> Self {
+        Self {
+            reservations: KVec::new(),
+            close_on_free: FdsCloseOnFree(KVec::new()),
+        }
+    }
+
+    pub(crate) fn commit(self) -> FdsCloseOnFree {
+        for entry in self.reservations {
+            entry.res.fd_install(entry.file);
+        }
+
+        self.close_on_free
+    }
+}
+
+pub(crate) struct FdsCloseOnFree(KVec<u32>);
diff --git a/drivers/android/binder/context.rs b/drivers/android/binder/context.rs
new file mode 100644
index 000000000000..3d135ec03ca7
--- /dev/null
+++ b/drivers/android/binder/context.rs
@@ -0,0 +1,180 @@
+// SPDX-License-Identifier: GPL-2.0
+
+// Copyright (C) 2025 Google LLC.
+
+use kernel::{
+    error::Error,
+    list::{List, ListArc, ListLinks},
+    prelude::*,
+    security,
+    str::{CStr, CString},
+    sync::{Arc, Mutex},
+    task::Kuid,
+};
+
+use crate::{error::BinderError, node::NodeRef, process::Process};
+
+kernel::sync::global_lock! {
+    // SAFETY: We call `init` in the module initializer, so it's initialized before first use.
+    pub(crate) unsafe(uninit) static CONTEXTS: Mutex<ContextList> = ContextList {
+        list: List::new(),
+    };
+}
+
+pub(crate) struct ContextList {
+    list: List<Context>,
+}
+
+pub(crate) fn get_all_contexts() -> Result<KVec<Arc<Context>>> {
+    let lock = CONTEXTS.lock();
+
+    let count = lock.list.iter().count();
+
+    let mut ctxs = KVec::with_capacity(count, GFP_KERNEL)?;
+    for ctx in &lock.list {
+        ctxs.push(Arc::from(ctx), GFP_KERNEL)?;
+    }
+    Ok(ctxs)
+}
+
+/// This struct keeps track of the processes using this context, and which process is the context
+/// manager.
+struct Manager {
+    node: Option<NodeRef>,
+    uid: Option<Kuid>,
+    all_procs: List<Process>,
+}
+
+/// There is one context per binder file (/dev/binder, /dev/hwbinder, etc)
+#[pin_data]
+pub(crate) struct Context {
+    #[pin]
+    manager: Mutex<Manager>,
+    pub(crate) name: CString,
+    #[pin]
+    links: ListLinks,
+}
+
+kernel::list::impl_list_arc_safe! {
+    impl ListArcSafe<0> for Context { untracked; }
+}
+kernel::list::impl_list_item! {
+    impl ListItem<0> for Context {
+        using ListLinks { self.links };
+    }
+}
+
+impl Context {
+    pub(crate) fn new(name: &CStr) -> Result<Arc<Self>> {
+        let name = CString::try_from(name)?;
+        let list_ctx = ListArc::pin_init::<Error>(
+            try_pin_init!(Context {
+                name,
+                links <- ListLinks::new(),
+                manager <- kernel::new_mutex!(Manager {
+                    all_procs: List::new(),
+                    node: None,
+                    uid: None,
+                }, "Context::manager"),
+            }),
+            GFP_KERNEL,
+        )?;
+
+        let ctx = list_ctx.clone_arc();
+        CONTEXTS.lock().list.push_back(list_ctx);
+
+        Ok(ctx)
+    }
+
+    /// Called when the file for this context is unlinked.
+    ///
+    /// No-op if called twice.
+    pub(crate) fn deregister(&self) {
+        // SAFETY: We never add the context to any other linked list than this one, so it is either
+        // in this list, or not in any list.
+        unsafe { CONTEXTS.lock().list.remove(self) };
+    }
+
+    pub(crate) fn register_process(self: &Arc<Self>, proc: ListArc<Process>) {
+        if !Arc::ptr_eq(self, &proc.ctx) {
+            pr_err!("Context::register_process called on the wrong context.");
+            return;
+        }
+        self.manager.lock().all_procs.push_back(proc);
+    }
+
+    pub(crate) fn deregister_process(self: &Arc<Self>, proc: &Process) {
+        if !Arc::ptr_eq(self, &proc.ctx) {
+            pr_err!("Context::deregister_process called on the wrong context.");
+            return;
+        }
+        // SAFETY: We just checked that this is the right list.
+        unsafe { self.manager.lock().all_procs.remove(proc) };
+    }
+
+    pub(crate) fn set_manager_node(&self, node_ref: NodeRef) -> Result {
+        let mut manager = self.manager.lock();
+        if manager.node.is_some() {
+            pr_warn!("BINDER_SET_CONTEXT_MGR already set");
+            return Err(EBUSY);
+        }
+        security::binder_set_context_mgr(&node_ref.node.owner.cred)?;
+
+        // If the context manager has been set before, ensure that we use the same euid.
+        let caller_uid = Kuid::current_euid();
+        if let Some(ref uid) = manager.uid {
+            if *uid != caller_uid {
+                return Err(EPERM);
+            }
+        }
+
+        manager.node = Some(node_ref);
+        manager.uid = Some(caller_uid);
+        Ok(())
+    }
+
+    pub(crate) fn unset_manager_node(&self) {
+        let node_ref = self.manager.lock().node.take();
+        drop(node_ref);
+    }
+
+    pub(crate) fn get_manager_node(&self, strong: bool) -> Result<NodeRef, BinderError> {
+        self.manager
+            .lock()
+            .node
+            .as_ref()
+            .ok_or_else(BinderError::new_dead)?
+            .clone(strong)
+            .map_err(BinderError::from)
+    }
+
+    pub(crate) fn for_each_proc<F>(&self, mut func: F)
+    where
+        F: FnMut(&Process),
+    {
+        let lock = self.manager.lock();
+        for proc in &lock.all_procs {
+            func(&proc);
+        }
+    }
+
+    pub(crate) fn get_all_procs(&self) -> Result<KVec<Arc<Process>>> {
+        let lock = self.manager.lock();
+        let count = lock.all_procs.iter().count();
+
+        let mut procs = KVec::with_capacity(count, GFP_KERNEL)?;
+        for proc in &lock.all_procs {
+            procs.push(Arc::from(proc), GFP_KERNEL)?;
+        }
+        Ok(procs)
+    }
+
+    pub(crate) fn get_procs_with_pid(&self, pid: i32) -> Result<KVec<Arc<Process>>> {
+        let orig = self.get_all_procs()?;
+        let mut backing = KVec::with_capacity(orig.len(), GFP_KERNEL)?;
+        for proc in orig.into_iter().filter(|proc| proc.task.pid() == pid) {
+            backing.push(proc, GFP_KERNEL)?;
+        }
+        Ok(backing)
+    }
+}
diff --git a/drivers/android/binder/deferred_close.rs b/drivers/android/binder/deferred_close.rs
new file mode 100644
index 000000000000..ac895c04d0cb
--- /dev/null
+++ b/drivers/android/binder/deferred_close.rs
@@ -0,0 +1,204 @@
+// SPDX-License-Identifier: GPL-2.0
+
+// Copyright (C) 2025 Google LLC.
+
+//! Logic for closing files in a deferred manner.
+//!
+//! This file could make sense to have in `kernel::fs`, but it was rejected for being too
+//! Binder-specific.
+
+use core::mem::MaybeUninit;
+use kernel::{
+    alloc::{AllocError, Flags},
+    bindings,
+    prelude::*,
+};
+
+/// Helper used for closing file descriptors in a way that is safe even if the file is currently
+/// held using `fdget`.
+///
+/// Additional motivation can be found in commit 80cd795630d6 ("binder: fix use-after-free due to
+/// ksys_close() during fdget()") and in the comments on `binder_do_fd_close`.
+pub(crate) struct DeferredFdCloser {
+    inner: KBox<DeferredFdCloserInner>,
+}
+
+/// SAFETY: This just holds an allocation with no real content, so there's no safety issue with
+/// moving it across threads.
+unsafe impl Send for DeferredFdCloser {}
+/// SAFETY: This just holds an allocation with no real content, so there's no safety issue with
+/// moving it across threads.
+unsafe impl Sync for DeferredFdCloser {}
+
+/// # Invariants
+///
+/// If the `file` pointer is non-null, then it points at a `struct file` and owns a refcount to
+/// that file.
+#[repr(C)]
+struct DeferredFdCloserInner {
+    twork: MaybeUninit<bindings::callback_head>,
+    file: *mut bindings::file,
+}
+
+impl DeferredFdCloser {
+    /// Create a new [`DeferredFdCloser`].
+    pub(crate) fn new(flags: Flags) -> Result<Self, AllocError> {
+        Ok(Self {
+            // INVARIANT: The `file` pointer is null, so the type invariant does not apply.
+            inner: KBox::new(
+                DeferredFdCloserInner {
+                    twork: MaybeUninit::uninit(),
+                    file: core::ptr::null_mut(),
+                },
+                flags,
+            )?,
+        })
+    }
+
+    /// Schedule a task work that closes the file descriptor when this task returns to userspace.
+    ///
+    /// Fails if this is called from a context where we cannot run work when returning to
+    /// userspace. (E.g., from a kthread.)
+    pub(crate) fn close_fd(self, fd: u32) -> Result<(), DeferredFdCloseError> {
+        use bindings::task_work_notify_mode_TWA_RESUME as TWA_RESUME;
+
+        // In this method, we schedule the task work before closing the file. This is because
+        // scheduling a task work is fallible, and we need to know whether it will fail before we
+        // attempt to close the file.
+
+        // Task works are not available on kthreads.
+        let current = kernel::current!();
+
+        // Check if this is a kthread.
+        // SAFETY: Reading `flags` from a task is always okay.
+        if unsafe { ((*current.as_ptr()).flags & bindings::PF_KTHREAD) != 0 } {
+            return Err(DeferredFdCloseError::TaskWorkUnavailable);
+        }
+
+        // Transfer ownership of the box's allocation to a raw pointer. This disables the
+        // destructor, so we must manually convert it back to a KBox to drop it.
+        //
+        // Until we convert it back to a `KBox`, there are no aliasing requirements on this
+        // pointer.
+        let inner = KBox::into_raw(self.inner);
+
+        // The `callback_head` field is first in the struct, so this cast correctly gives us a
+        // pointer to the field.
+        let callback_head = inner.cast::<bindings::callback_head>();
+        // SAFETY: This pointer offset operation does not go out-of-bounds.
+        let file_field = unsafe { core::ptr::addr_of_mut!((*inner).file) };
+
+        let current = current.as_ptr();
+
+        // SAFETY: This function currently has exclusive access to the `DeferredFdCloserInner`, so
+        // it is okay for us to perform unsynchronized writes to its `callback_head` field.
+        unsafe { bindings::init_task_work(callback_head, Some(Self::do_close_fd)) };
+
+        // SAFETY: This inserts the `DeferredFdCloserInner` into the task workqueue for the current
+        // task. If this operation is successful, then this transfers exclusive ownership of the
+        // `callback_head` field to the C side until it calls `do_close_fd`, and we don't touch or
+        // invalidate the field during that time.
+        //
+        // When the C side calls `do_close_fd`, the safety requirements of that method are
+        // satisfied because when a task work is executed, the callback is given ownership of the
+        // pointer.
+        //
+        // The file pointer is currently null. If it is changed to be non-null before `do_close_fd`
+        // is called, then that change happens due to the write at the end of this function, and
+        // that write has a safety comment that explains why the refcount can be dropped when
+        // `do_close_fd` runs.
+        let res = unsafe { bindings::task_work_add(current, callback_head, TWA_RESUME) };
+
+        if res != 0 {
+            // SAFETY: Scheduling the task work failed, so we still have ownership of the box, so
+            // we may destroy it.
+            unsafe { drop(KBox::from_raw(inner)) };
+
+            return Err(DeferredFdCloseError::TaskWorkUnavailable);
+        }
+
+        // This removes the fd from the fd table in `current`. The file is not fully closed until
+        // `filp_close` is called. We are given ownership of one refcount to the file.
+        //
+        // SAFETY: This is safe no matter what `fd` is. If the `fd` is valid (that is, if the
+        // pointer is non-null), then we call `filp_close` on the returned pointer as required by
+        // `file_close_fd`.
+        let file = unsafe { bindings::file_close_fd(fd) };
+        if file.is_null() {
+            // We don't clean up the task work since that might be expensive if the task work queue
+            // is long. Just let it execute and let it clean up for itself.
+            return Err(DeferredFdCloseError::BadFd);
+        }
+
+        // Acquire a second refcount to the file.
+        //
+        // SAFETY: The `file` pointer points at a file with a non-zero refcount.
+        unsafe { bindings::get_file(file) };
+
+        // This method closes the fd, consuming one of our two refcounts. There could be active
+        // light refcounts created from that fd, so we must ensure that the file has a positive
+        // refcount for the duration of those active light refcounts. We do that by holding on to
+        // the second refcount until the current task returns to userspace.
+        //
+        // SAFETY: The `file` pointer is valid. Passing `current->files` as the file table to close
+        // it in is correct, since we just got the `fd` from `file_close_fd` which also uses
+        // `current->files`.
+        //
+        // Note: fl_owner_t is currently a void pointer.
+        unsafe { bindings::filp_close(file, (*current).files as bindings::fl_owner_t) };
+
+        // We update the file pointer that the task work is supposed to fput. This transfers
+        // ownership of our last refcount.
+        //
+        // INVARIANT: This changes the `file` field of a `DeferredFdCloserInner` from null to
+        // non-null. This doesn't break the type invariant for `DeferredFdCloserInner` because we
+        // still own a refcount to the file, so we can pass ownership of that refcount to the
+        // `DeferredFdCloserInner`.
+        //
+        // When `do_close_fd` runs, it must be safe for it to `fput` the refcount. However, this is
+        // the case because all light refcounts that are associated with the fd we closed
+        // previously must be dropped when `do_close_fd`, since light refcounts must be dropped
+        // before returning to userspace.
+        //
+        // SAFETY: Task works are executed on the current thread right before we return to
+        // userspace, so this write is guaranteed to happen before `do_close_fd` is called, which
+        // means that a race is not possible here.
+        unsafe { *file_field = file };
+
+        Ok(())
+    }
+
+    /// # Safety
+    ///
+    /// The provided pointer must point at the `twork` field of a `DeferredFdCloserInner` stored in
+    /// a `KBox`, and the caller must pass exclusive ownership of that `KBox`. Furthermore, if the
+    /// file pointer is non-null, then it must be okay to release the refcount by calling `fput`.
+    unsafe extern "C" fn do_close_fd(inner: *mut bindings::callback_head) {
+        // SAFETY: The caller just passed us ownership of this box.
+        let inner = unsafe { KBox::from_raw(inner.cast::<DeferredFdCloserInner>()) };
+        if !inner.file.is_null() {
+            // SAFETY: By the type invariants, we own a refcount to this file, and the caller
+            // guarantees that dropping the refcount now is okay.
+            unsafe { bindings::fput(inner.file) };
+        }
+        // The allocation is freed when `inner` goes out of scope.
+    }
+}
+
+/// Represents a failure to close an fd in a deferred manner.
+#[derive(Copy, Clone, Debug, Eq, PartialEq)]
+pub(crate) enum DeferredFdCloseError {
+    /// Closing the fd failed because we were unable to schedule a task work.
+    TaskWorkUnavailable,
+    /// Closing the fd failed because the fd does not exist.
+    BadFd,
+}
+
+impl From<DeferredFdCloseError> for Error {
+    fn from(err: DeferredFdCloseError) -> Error {
+        match err {
+            DeferredFdCloseError::TaskWorkUnavailable => ESRCH,
+            DeferredFdCloseError::BadFd => EBADF,
+        }
+    }
+}
diff --git a/drivers/android/binder/defs.rs b/drivers/android/binder/defs.rs
new file mode 100644
index 000000000000..33f51b4139c7
--- /dev/null
+++ b/drivers/android/binder/defs.rs
@@ -0,0 +1,182 @@
+// SPDX-License-Identifier: GPL-2.0
+
+// Copyright (C) 2025 Google LLC.
+
+use core::mem::MaybeUninit;
+use core::ops::{Deref, DerefMut};
+use kernel::{
+    transmute::{AsBytes, FromBytes},
+    uapi::{self, *},
+};
+
+macro_rules! pub_no_prefix {
+    ($prefix:ident, $($newname:ident),+ $(,)?) => {
+        $(pub(crate) const $newname: u32 = kernel::macros::concat_idents!($prefix, $newname);)+
+    };
+}
+
+pub_no_prefix!(
+    binder_driver_return_protocol_,
+    BR_TRANSACTION,
+    BR_TRANSACTION_SEC_CTX,
+    BR_REPLY,
+    BR_DEAD_REPLY,
+    BR_FAILED_REPLY,
+    BR_FROZEN_REPLY,
+    BR_NOOP,
+    BR_SPAWN_LOOPER,
+    BR_TRANSACTION_COMPLETE,
+    BR_TRANSACTION_PENDING_FROZEN,
+    BR_ONEWAY_SPAM_SUSPECT,
+    BR_OK,
+    BR_ERROR,
+    BR_INCREFS,
+    BR_ACQUIRE,
+    BR_RELEASE,
+    BR_DECREFS,
+    BR_DEAD_BINDER,
+    BR_CLEAR_DEATH_NOTIFICATION_DONE,
+    BR_FROZEN_BINDER,
+    BR_CLEAR_FREEZE_NOTIFICATION_DONE,
+);
+
+pub_no_prefix!(
+    binder_driver_command_protocol_,
+    BC_TRANSACTION,
+    BC_TRANSACTION_SG,
+    BC_REPLY,
+    BC_REPLY_SG,
+    BC_FREE_BUFFER,
+    BC_ENTER_LOOPER,
+    BC_EXIT_LOOPER,
+    BC_REGISTER_LOOPER,
+    BC_INCREFS,
+    BC_ACQUIRE,
+    BC_RELEASE,
+    BC_DECREFS,
+    BC_INCREFS_DONE,
+    BC_ACQUIRE_DONE,
+    BC_REQUEST_DEATH_NOTIFICATION,
+    BC_CLEAR_DEATH_NOTIFICATION,
+    BC_DEAD_BINDER_DONE,
+    BC_REQUEST_FREEZE_NOTIFICATION,
+    BC_CLEAR_FREEZE_NOTIFICATION,
+    BC_FREEZE_NOTIFICATION_DONE,
+);
+
+pub_no_prefix!(
+    flat_binder_object_flags_,
+    FLAT_BINDER_FLAG_ACCEPTS_FDS,
+    FLAT_BINDER_FLAG_TXN_SECURITY_CTX
+);
+
+pub_no_prefix!(
+    transaction_flags_,
+    TF_ONE_WAY,
+    TF_ACCEPT_FDS,
+    TF_CLEAR_BUF,
+    TF_UPDATE_TXN
+);
+
+pub(crate) use uapi::{
+    BINDER_TYPE_BINDER, BINDER_TYPE_FD, BINDER_TYPE_FDA, BINDER_TYPE_HANDLE, BINDER_TYPE_PTR,
+    BINDER_TYPE_WEAK_BINDER, BINDER_TYPE_WEAK_HANDLE,
+};
+
+macro_rules! decl_wrapper {
+    ($newname:ident, $wrapped:ty) => {
+        // Define a wrapper around the C type. Use `MaybeUninit` to enforce that the value of
+        // padding bytes must be preserved.
+        #[derive(Copy, Clone)]
+        #[repr(transparent)]
+        pub(crate) struct $newname(MaybeUninit<$wrapped>);
+
+        // SAFETY: This macro is only used with types where this is ok.
+        unsafe impl FromBytes for $newname {}
+        // SAFETY: This macro is only used with types where this is ok.
+        unsafe impl AsBytes for $newname {}
+
+        impl Deref for $newname {
+            type Target = $wrapped;
+            fn deref(&self) -> &Self::Target {
+                // SAFETY: We use `MaybeUninit` only to preserve padding. The value must still
+                // always be valid.
+                unsafe { self.0.assume_init_ref() }
+            }
+        }
+
+        impl DerefMut for $newname {
+            fn deref_mut(&mut self) -> &mut Self::Target {
+                // SAFETY: We use `MaybeUninit` only to preserve padding. The value must still
+                // always be valid.
+                unsafe { self.0.assume_init_mut() }
+            }
+        }
+
+        impl Default for $newname {
+            fn default() -> Self {
+                // Create a new value of this type where all bytes (including padding) are zeroed.
+                Self(MaybeUninit::zeroed())
+            }
+        }
+    };
+}
+
+decl_wrapper!(BinderNodeDebugInfo, uapi::binder_node_debug_info);
+decl_wrapper!(BinderNodeInfoForRef, uapi::binder_node_info_for_ref);
+decl_wrapper!(FlatBinderObject, uapi::flat_binder_object);
+decl_wrapper!(BinderFdObject, uapi::binder_fd_object);
+decl_wrapper!(BinderFdArrayObject, uapi::binder_fd_array_object);
+decl_wrapper!(BinderObjectHeader, uapi::binder_object_header);
+decl_wrapper!(BinderBufferObject, uapi::binder_buffer_object);
+decl_wrapper!(BinderTransactionData, uapi::binder_transaction_data);
+decl_wrapper!(
+    BinderTransactionDataSecctx,
+    uapi::binder_transaction_data_secctx
+);
+decl_wrapper!(BinderTransactionDataSg, uapi::binder_transaction_data_sg);
+decl_wrapper!(BinderWriteRead, uapi::binder_write_read);
+decl_wrapper!(BinderVersion, uapi::binder_version);
+decl_wrapper!(BinderFrozenStatusInfo, uapi::binder_frozen_status_info);
+decl_wrapper!(BinderFreezeInfo, uapi::binder_freeze_info);
+decl_wrapper!(BinderFrozenStateInfo, uapi::binder_frozen_state_info);
+decl_wrapper!(BinderHandleCookie, uapi::binder_handle_cookie);
+decl_wrapper!(ExtendedError, uapi::binder_extended_error);
+
+impl BinderVersion {
+    pub(crate) fn current() -> Self {
+        Self(MaybeUninit::new(uapi::binder_version {
+            protocol_version: BINDER_CURRENT_PROTOCOL_VERSION as _,
+        }))
+    }
+}
+
+impl BinderTransactionData {
+    pub(crate) fn with_buffers_size(self, buffers_size: u64) -> BinderTransactionDataSg {
+        BinderTransactionDataSg(MaybeUninit::new(uapi::binder_transaction_data_sg {
+            transaction_data: *self,
+            buffers_size,
+        }))
+    }
+}
+
+impl BinderTransactionDataSecctx {
+    /// View the inner data as wrapped in `BinderTransactionData`.
+    pub(crate) fn tr_data(&mut self) -> &mut BinderTransactionData {
+        // SAFETY: Transparent wrapper is safe to transmute.
+        unsafe {
+            &mut *(&mut self.transaction_data as *mut uapi::binder_transaction_data
+                as *mut BinderTransactionData)
+        }
+    }
+}
+
+impl ExtendedError {
+    pub(crate) fn new(id: u32, command: u32, param: i32) -> Self {
+        Self(MaybeUninit::new(uapi::binder_extended_error {
+            id,
+            command,
+            param,
+        }))
+    }
+}
diff --git a/drivers/android/binder/error.rs b/drivers/android/binder/error.rs
new file mode 100644
index 000000000000..b24497cfa292
--- /dev/null
+++ b/drivers/android/binder/error.rs
@@ -0,0 +1,100 @@
+// SPDX-License-Identifier: GPL-2.0
+
+// Copyright (C) 2025 Google LLC.
+
+use kernel::fmt;
+use kernel::prelude::*;
+
+use crate::defs::*;
+
+pub(crate) type BinderResult<T = ()> = core::result::Result<T, BinderError>;
+
+/// An error that will be returned to userspace via the `BINDER_WRITE_READ` ioctl rather than via
+/// errno.
+pub(crate) struct BinderError {
+    pub(crate) reply: u32,
+    source: Option<Error>,
+}
+
+impl BinderError {
+    pub(crate) fn new_dead() -> Self {
+        Self {
+            reply: BR_DEAD_REPLY,
+            source: None,
+        }
+    }
+
+    pub(crate) fn new_frozen() -> Self {
+        Self {
+            reply: BR_FROZEN_REPLY,
+            source: None,
+        }
+    }
+
+    pub(crate) fn new_frozen_oneway() -> Self {
+        Self {
+            reply: BR_TRANSACTION_PENDING_FROZEN,
+            source: None,
+        }
+    }
+
+    pub(crate) fn is_dead(&self) -> bool {
+        self.reply == BR_DEAD_REPLY
+    }
+
+    pub(crate) fn as_errno(&self) -> kernel::ffi::c_int {
+        self.source.unwrap_or(EINVAL).to_errno()
+    }
+
+    pub(crate) fn should_pr_warn(&self) -> bool {
+        self.source.is_some()
+    }
+}
+
+/// Convert an errno into a `BinderError` and store the errno used to construct it. The errno
+/// should be stored as the thread's extended error when given to userspace.
+impl From<Error> for BinderError {
+    fn from(source: Error) -> Self {
+        Self {
+            reply: BR_FAILED_REPLY,
+            source: Some(source),
+        }
+    }
+}
+
+impl From<kernel::fs::file::BadFdError> for BinderError {
+    fn from(source: kernel::fs::file::BadFdError) -> Self {
+        BinderError::from(Error::from(source))
+    }
+}
+
+impl From<kernel::alloc::AllocError> for BinderError {
+    fn from(_: kernel::alloc::AllocError) -> Self {
+        Self {
+            reply: BR_FAILED_REPLY,
+            source: Some(ENOMEM),
+        }
+    }
+}
+
+impl fmt::Debug for BinderError {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        match self.reply {
+            BR_FAILED_REPLY => match self.source.as_ref() {
+                Some(source) => f
+                    .debug_struct("BR_FAILED_REPLY")
+                    .field("source", source)
+                    .finish(),
+                None => f.pad("BR_FAILED_REPLY"),
+            },
+            BR_DEAD_REPLY => f.pad("BR_DEAD_REPLY"),
+            BR_FROZEN_REPLY => f.pad("BR_FROZEN_REPLY"),
+            BR_TRANSACTION_PENDING_FROZEN => f.pad("BR_TRANSACTION_PENDING_FROZEN"),
+            BR_TRANSACTION_COMPLETE => f.pad("BR_TRANSACTION_COMPLETE"),
+            _ => f
+                .debug_struct("BinderError")
+                .field("reply", &self.reply)
+                .finish(),
+        }
+    }
+}
diff --git a/drivers/android/binder/freeze.rs b/drivers/android/binder/freeze.rs
new file mode 100644
index 000000000000..53b60035639a
--- /dev/null
+++ b/drivers/android/binder/freeze.rs
@@ -0,0 +1,398 @@
+// SPDX-License-Identifier: GPL-2.0
+
+// Copyright (C) 2025 Google LLC.
+
+use kernel::{
+    alloc::AllocError,
+    list::ListArc,
+    prelude::*,
+    rbtree::{self, RBTreeNodeReservation},
+    seq_file::SeqFile,
+    seq_print,
+    sync::{Arc, UniqueArc},
+    uaccess::UserSliceReader,
+};
+
+use crate::{
+    defs::*, node::Node, process::Process, thread::Thread, BinderReturnWriter, DArc, DLArc,
+    DTRWrap, DeliverToRead,
+};
+
+#[derive(Clone, Copy, Eq, PartialEq, Ord, PartialOrd)]
+pub(crate) struct FreezeCookie(u64);
+
+/// Represents a listener for changes to the frozen state of a process.
+pub(crate) struct FreezeListener {
+    /// The node we are listening for.
+    pub(crate) node: DArc<Node>,
+    /// The cookie of this freeze listener.
+    cookie: FreezeCookie,
+    /// What value of `is_frozen` did we most recently tell userspace about?
+    last_is_frozen: Option<bool>,
+    /// We sent a `BR_FROZEN_BINDER` and we are waiting for `BC_FREEZE_NOTIFICATION_DONE` before
+    /// sending any other commands.
+    is_pending: bool,
+    /// Userspace sent `BC_CLEAR_FREEZE_NOTIFICATION` and we need to reply with
+    /// `BR_CLEAR_FREEZE_NOTIFICATION_DONE` as soon as possible. If `is_pending` is set, then we
+    /// must wait for it to be unset before we can reply.
+    is_clearing: bool,
+    /// Number of cleared duplicates that can't be deleted until userspace sends
+    /// `BC_FREEZE_NOTIFICATION_DONE`.
+    num_pending_duplicates: u64,
+    /// Number of cleared duplicates that can be deleted.
+    num_cleared_duplicates: u64,
+}
+
+impl FreezeListener {
+    /// Is it okay to create a new listener with the same cookie as this one for the provided node?
+    ///
+    /// Under some scenarios, userspace may delete a freeze listener and immediately recreate it
+    /// with the same cookie. This results in duplicate listeners. To avoid issues with ambiguity,
+    /// we allow this only if the new listener is for the same node, and we also require that the
+    /// old listener has already been cleared.
+    fn allow_duplicate(&self, node: &DArc<Node>) -> bool {
+        Arc::ptr_eq(&self.node, node) && self.is_clearing
+    }
+}
+
+type UninitFM = UniqueArc<core::mem::MaybeUninit<DTRWrap<FreezeMessage>>>;
+
+/// Represents a notification that the freeze state has changed.
+pub(crate) struct FreezeMessage {
+    cookie: FreezeCookie,
+}
+
+kernel::list::impl_list_arc_safe! {
+    impl ListArcSafe<0> for FreezeMessage {
+        untracked;
+    }
+}
+
+impl FreezeMessage {
+    fn new(flags: kernel::alloc::Flags) -> Result<UninitFM, AllocError> {
+        UniqueArc::new_uninit(flags)
+    }
+
+    fn init(ua: UninitFM, cookie: FreezeCookie) -> DLArc<FreezeMessage> {
+        match ua.pin_init_with(DTRWrap::new(FreezeMessage { cookie })) {
+            Ok(msg) => ListArc::from(msg),
+            Err(err) => match err {},
+        }
+    }
+}
+
+impl DeliverToRead for FreezeMessage {
+    fn do_work(
+        self: DArc<Self>,
+        thread: &Thread,
+        writer: &mut BinderReturnWriter<'_>,
+    ) -> Result<bool> {
+        let _removed_listener;
+        let mut node_refs = thread.process.node_refs.lock();
+        let Some(mut freeze_entry) = node_refs.freeze_listeners.find_mut(&self.cookie) else {
+            return Ok(true);
+        };
+        let freeze = freeze_entry.get_mut();
+
+        if freeze.num_cleared_duplicates > 0 {
+            freeze.num_cleared_duplicates -= 1;
+            drop(node_refs);
+            writer.write_code(BR_CLEAR_FREEZE_NOTIFICATION_DONE)?;
+            writer.write_payload(&self.cookie.0)?;
+            return Ok(true);
+        }
+
+        if freeze.is_pending {
+            return Ok(true);
+        }
+        if freeze.is_clearing {
+            kernel::warn_on!(freeze.num_cleared_duplicates != 0);
+            if freeze.num_pending_duplicates > 0 {
+                // The primary freeze listener was deleted, so convert a pending duplicate back
+                // into the primary one.
+                freeze.num_pending_duplicates -= 1;
+                freeze.is_pending = true;
+                freeze.is_clearing = true;
+            } else {
+                _removed_listener = freeze_entry.remove_node();
+            }
+            drop(node_refs);
+            writer.write_code(BR_CLEAR_FREEZE_NOTIFICATION_DONE)?;
+            writer.write_payload(&self.cookie.0)?;
+            Ok(true)
+        } else {
+            let is_frozen = freeze.node.owner.inner.lock().is_frozen.is_fully_frozen();
+            if freeze.last_is_frozen == Some(is_frozen) {
+                return Ok(true);
+            }
+
+            let mut state_info = BinderFrozenStateInfo::default();
+            state_info.is_frozen = is_frozen as u32;
+            state_info.cookie = freeze.cookie.0;
+            freeze.is_pending = true;
+            freeze.last_is_frozen = Some(is_frozen);
+            drop(node_refs);
+
+            writer.write_code(BR_FROZEN_BINDER)?;
+            writer.write_payload(&state_info)?;
+            // BR_FROZEN_BINDER notifications can cause transactions
+            Ok(false)
+        }
+    }
+
+    fn cancel(self: DArc<Self>) {}
+
+    fn should_sync_wakeup(&self) -> bool {
+        false
+    }
+
+    #[inline(never)]
+    fn debug_print(&self, m: &SeqFile, prefix: &str, _tprefix: &str) -> Result<()> {
+        seq_print!(m, "{}has frozen binder\n", prefix);
+        Ok(())
+    }
+}
+
+impl FreezeListener {
+    pub(crate) fn on_process_exit(&self, proc: &Arc<Process>) {
+        if !self.is_clearing {
+            self.node.remove_freeze_listener(proc);
+        }
+    }
+}
+
+impl Process {
+    pub(crate) fn request_freeze_notif(
+        self: &Arc<Self>,
+        reader: &mut UserSliceReader,
+    ) -> Result<()> {
+        let hc = reader.read::<BinderHandleCookie>()?;
+        let handle = hc.handle;
+        let cookie = FreezeCookie(hc.cookie);
+
+        let msg = FreezeMessage::new(GFP_KERNEL)?;
+        let alloc = RBTreeNodeReservation::new(GFP_KERNEL)?;
+
+        let mut node_refs_guard = self.node_refs.lock();
+        let node_refs = &mut *node_refs_guard;
+        let Some(info) = node_refs.by_handle.get_mut(&handle) else {
+            pr_warn!("BC_REQUEST_FREEZE_NOTIFICATION invalid ref {}\n", handle);
+            return Err(EINVAL);
+        };
+        if info.freeze().is_some() {
+            pr_warn!("BC_REQUEST_FREEZE_NOTIFICATION already set\n");
+            return Err(EINVAL);
+        }
+        let node_ref = info.node_ref();
+        let freeze_entry = node_refs.freeze_listeners.entry(cookie);
+
+        if let rbtree::Entry::Occupied(ref dupe) = freeze_entry {
+            if !dupe.get().allow_duplicate(&node_ref.node) {
+                pr_warn!("BC_REQUEST_FREEZE_NOTIFICATION duplicate cookie\n");
+                return Err(EINVAL);
+            }
+        }
+
+        // All failure paths must come before this call, and all modifications must come after this
+        // call.
+        node_ref.node.add_freeze_listener(self, GFP_KERNEL)?;
+
+        match freeze_entry {
+            rbtree::Entry::Vacant(entry) => {
+                entry.insert(
+                    FreezeListener {
+                        cookie,
+                        node: node_ref.node.clone(),
+                        last_is_frozen: None,
+                        is_pending: false,
+                        is_clearing: false,
+                        num_pending_duplicates: 0,
+                        num_cleared_duplicates: 0,
+                    },
+                    alloc,
+                );
+            }
+            rbtree::Entry::Occupied(mut dupe) => {
+                let dupe = dupe.get_mut();
+                if dupe.is_pending {
+                    dupe.num_pending_duplicates += 1;
+                } else {
+                    dupe.num_cleared_duplicates += 1;
+                }
+                dupe.last_is_frozen = None;
+                dupe.is_pending = false;
+                dupe.is_clearing = false;
+            }
+        }
+
+        *info.freeze() = Some(cookie);
+        let msg = FreezeMessage::init(msg, cookie);
+        drop(node_refs_guard);
+        let _ = self.push_work(msg);
+        Ok(())
+    }
+
+    pub(crate) fn freeze_notif_done(self: &Arc<Self>, reader: &mut UserSliceReader) -> Result<()> {
+        let cookie = FreezeCookie(reader.read()?);
+        let alloc = FreezeMessage::new(GFP_KERNEL)?;
+        let mut node_refs_guard = self.node_refs.lock();
+        let node_refs = &mut *node_refs_guard;
+        let Some(freeze) = node_refs.freeze_listeners.get_mut(&cookie) else {
+            pr_warn!("BC_FREEZE_NOTIFICATION_DONE {:016x} not found\n", cookie.0);
+            return Err(EINVAL);
+        };
+        let mut clear_msg = None;
+        if freeze.num_pending_duplicates > 0 {
+            clear_msg = Some(FreezeMessage::init(alloc, cookie));
+            freeze.num_pending_duplicates -= 1;
+            freeze.num_cleared_duplicates += 1;
+        } else {
+            if !freeze.is_pending {
+                pr_warn!(
+                    "BC_FREEZE_NOTIFICATION_DONE {:016x} not pending\n",
+                    cookie.0
+                );
+                return Err(EINVAL);
+            }
+            let is_frozen = freeze.node.owner.inner.lock().is_frozen.is_fully_frozen();
+            if freeze.is_clearing || freeze.last_is_frozen != Some(is_frozen) {
+                // Immediately send another FreezeMessage.
+                clear_msg = Some(FreezeMessage::init(alloc, cookie));
+            }
+            freeze.is_pending = false;
+        }
+        drop(node_refs_guard);
+        if let Some(clear_msg) = clear_msg {
+            let _ = self.push_work(clear_msg);
+        }
+        Ok(())
+    }
+
+    pub(crate) fn clear_freeze_notif(self: &Arc<Self>, reader: &mut UserSliceReader) -> Result<()> {
+        let hc = reader.read::<BinderHandleCookie>()?;
+        let handle = hc.handle;
+        let cookie = FreezeCookie(hc.cookie);
+
+        let alloc = FreezeMessage::new(GFP_KERNEL)?;
+        let mut node_refs_guard = self.node_refs.lock();
+        let node_refs = &mut *node_refs_guard;
+        let Some(info) = node_refs.by_handle.get_mut(&handle) else {
+            pr_warn!("BC_CLEAR_FREEZE_NOTIFICATION invalid ref {}\n", handle);
+            return Err(EINVAL);
+        };
+        let Some(info_cookie) = info.freeze() else {
+            pr_warn!("BC_CLEAR_FREEZE_NOTIFICATION freeze notification not active\n");
+            return Err(EINVAL);
+        };
+        if *info_cookie != cookie {
+            pr_warn!("BC_CLEAR_FREEZE_NOTIFICATION freeze notification cookie mismatch\n");
+            return Err(EINVAL);
+        }
+        let Some(listener) = node_refs.freeze_listeners.get_mut(&cookie) else {
+            pr_warn!("BC_CLEAR_FREEZE_NOTIFICATION invalid cookie {}\n", handle);
+            return Err(EINVAL);
+        };
+        listener.is_clearing = true;
+        listener.node.remove_freeze_listener(self);
+        *info.freeze() = None;
+        let mut msg = None;
+        if !listener.is_pending {
+            msg = Some(FreezeMessage::init(alloc, cookie));
+        }
+        drop(node_refs_guard);
+
+        if let Some(msg) = msg {
+            let _ = self.push_work(msg);
+        }
+        Ok(())
+    }
+
+    fn get_freeze_cookie(&self, node: &DArc<Node>) -> Option<FreezeCookie> {
+        let node_refs = &mut *self.node_refs.lock();
+        let handle = node_refs.by_node.get(&node.global_id())?;
+        let node_ref = node_refs.by_handle.get_mut(handle)?;
+        *node_ref.freeze()
+    }
+
+    /// Creates a vector of every freeze listener on this process.
+    ///
+    /// Returns pairs of the remote process listening for notifications and the local node it is
+    /// listening on.
+    #[expect(clippy::type_complexity)]
+    fn find_freeze_recipients(&self) -> Result<KVVec<(DArc<Node>, Arc<Process>)>, AllocError> {
+        // Defined before `inner` to drop after releasing spinlock if `push_within_capacity` fails.
+        let mut node_proc_pair;
+
+        // We pre-allocate space for up to 8 recipients before we take the spinlock. However, if
+        // the allocation fails, use a vector with a capacity of zero instead of failing. After
+        // all, there might not be any freeze listeners, in which case this operation could still
+        // succeed.
+        let mut recipients =
+            KVVec::with_capacity(8, GFP_KERNEL).unwrap_or_else(|_err| KVVec::new());
+
+        let mut inner = self.lock_with_nodes();
+        let mut curr = inner.nodes.cursor_front_mut();
+        while let Some(cursor) = curr {
+            let (key, node) = cursor.current();
+            let key = *key;
+            let list = node.freeze_list(&inner.inner);
+            let len = list.len();
+
+            if recipients.spare_capacity_mut().len() < len {
+                drop(inner);
+                recipients.reserve(len, GFP_KERNEL)?;
+                inner = self.lock_with_nodes();
+                // Find the node we were looking at and try again. If the set of nodes was changed,
+                // then just proceed to the next node. This is ok because we don't guarantee the
+                // inclusion of nodes that are added or removed in parallel with this operation.
+                curr = inner.nodes.cursor_lower_bound_mut(&key);
+                continue;
+            }
+
+            for proc in list {
+                node_proc_pair = (node.clone(), proc.clone());
+                recipients
+                    .push_within_capacity(node_proc_pair)
+                    .map_err(|_| {
+                        pr_err!(
+                            "push_within_capacity failed even though we checked the capacity\n"
+                        );
+                        AllocError
+                    })?;
+            }
+
+            curr = cursor.move_next();
+        }
+        Ok(recipients)
+    }
+
+    /// Prepare allocations for sending freeze messages.
+    pub(crate) fn prepare_freeze_messages(&self) -> Result<FreezeMessages, AllocError> {
+        let recipients = self.find_freeze_recipients()?;
+        let mut batch = KVVec::with_capacity(recipients.len(), GFP_KERNEL)?;
+        for (node, proc) in recipients {
+            let Some(cookie) = proc.get_freeze_cookie(&node) else {
+                // If the freeze listener was removed in the meantime, just discard the
+                // notification.
+                continue;
+            };
+            let msg_alloc = FreezeMessage::new(GFP_KERNEL)?;
+            let msg = FreezeMessage::init(msg_alloc, cookie);
+            batch.push((proc, msg), GFP_KERNEL)?;
+        }
+
+        Ok(FreezeMessages { batch })
+    }
+}
+
+pub(crate) struct FreezeMessages {
+    batch: KVVec<(Arc<Process>, DLArc<FreezeMessage>)>,
+}
+
+impl FreezeMessages {
+    pub(crate) fn send_messages(self) {
+        for (proc, msg) in self.batch {
+            let _ = proc.push_work(msg);
+        }
+    }
+}
diff --git a/drivers/android/binder/node.rs b/drivers/android/binder/node.rs
new file mode 100644
index 000000000000..c26d113ede96
--- /dev/null
+++ b/drivers/android/binder/node.rs
@@ -0,0 +1,1131 @@
+// SPDX-License-Identifier: GPL-2.0
+
+// Copyright (C) 2025 Google LLC.
+
+use kernel::{
+    list::{AtomicTracker, List, ListArc, ListLinks, TryNewListArc},
+    prelude::*,
+    seq_file::SeqFile,
+    seq_print,
+    sync::lock::{spinlock::SpinLockBackend, Guard},
+    sync::{Arc, LockedBy, SpinLock},
+};
+
+use crate::{
+    defs::*,
+    error::BinderError,
+    process::{NodeRefInfo, Process, ProcessInner},
+    thread::Thread,
+    transaction::Transaction,
+    BinderReturnWriter, DArc, DLArc, DTRWrap, DeliverToRead,
+};
+
+use core::mem;
+
+mod wrapper;
+pub(crate) use self::wrapper::CritIncrWrapper;
+
+#[derive(Debug)]
+pub(crate) struct CouldNotDeliverCriticalIncrement;
+
+/// Keeps track of how this node is scheduled.
+///
+/// There are two ways to schedule a node to a work list. Just schedule the node itself, or
+/// allocate a wrapper that references the node and schedule the wrapper. These wrappers exists to
+/// make it possible to "move" a node from one list to another - when `do_work` is called directly
+/// on the `Node`, then it's a no-op if there's also a pending wrapper.
+///
+/// Wrappers are generally only needed for zero-to-one refcount increments, and there are two cases
+/// of this: weak increments and strong increments. We call such increments "critical" because it
+/// is critical that they are delivered to the thread doing the increment. Some examples:
+///
+/// * One thread makes a zero-to-one strong increment, and another thread makes a zero-to-one weak
+///   increment. Delivering the node to the thread doing the weak increment is wrong, since the
+///   thread doing the strong increment may have ended a long time ago when the command is actually
+///   processed by userspace.
+///
+/// * We have a weak reference and are about to drop it on one thread. But then another thread does
+///   a zero-to-one strong increment. If the strong increment gets sent to the thread that was
+///   about to drop the weak reference, then the strong increment could be processed after the
+///   other thread has already exited, which would be too late.
+///
+/// Note that trying to create a `ListArc` to the node can succeed even if `has_normal_push` is
+/// set. This is because another thread might just have popped the node from a todo list, but not
+/// yet called `do_work`. However, if `has_normal_push` is false, then creating a `ListArc` should
+/// always succeed.
+///
+/// Like the other fields in `NodeInner`, the delivery state is protected by the process lock.
+struct DeliveryState {
+    /// Is the `Node` currently scheduled?
+    has_pushed_node: bool,
+
+    /// Is a wrapper currently scheduled?
+    ///
+    /// The wrapper is used only for strong zero2one increments.
+    has_pushed_wrapper: bool,
+
+    /// Is the currently scheduled `Node` scheduled due to a weak zero2one increment?
+    ///
+    /// Weak zero2one operations are always scheduled using the `Node`.
+    has_weak_zero2one: bool,
+
+    /// Is the currently scheduled wrapper/`Node` scheduled due to a strong zero2one increment?
+    ///
+    /// If `has_pushed_wrapper` is set, then the strong zero2one increment was scheduled using the
+    /// wrapper. Otherwise, `has_pushed_node` must be set and it was scheduled using the `Node`.
+    has_strong_zero2one: bool,
+}
+
+impl DeliveryState {
+    fn should_normal_push(&self) -> bool {
+        !self.has_pushed_node && !self.has_pushed_wrapper
+    }
+
+    fn did_normal_push(&mut self) {
+        assert!(self.should_normal_push());
+        self.has_pushed_node = true;
+    }
+
+    fn should_push_weak_zero2one(&self) -> bool {
+        !self.has_weak_zero2one && !self.has_strong_zero2one
+    }
+
+    fn can_push_weak_zero2one_normally(&self) -> bool {
+        !self.has_pushed_node
+    }
+
+    fn did_push_weak_zero2one(&mut self) {
+        assert!(self.should_push_weak_zero2one());
+        assert!(self.can_push_weak_zero2one_normally());
+        self.has_pushed_node = true;
+        self.has_weak_zero2one = true;
+    }
+
+    fn should_push_strong_zero2one(&self) -> bool {
+        !self.has_strong_zero2one
+    }
+
+    fn can_push_strong_zero2one_normally(&self) -> bool {
+        !self.has_pushed_node
+    }
+
+    fn did_push_strong_zero2one(&mut self) {
+        assert!(self.should_push_strong_zero2one());
+        assert!(self.can_push_strong_zero2one_normally());
+        self.has_pushed_node = true;
+        self.has_strong_zero2one = true;
+    }
+
+    fn did_push_strong_zero2one_wrapper(&mut self) {
+        assert!(self.should_push_strong_zero2one());
+        assert!(!self.can_push_strong_zero2one_normally());
+        self.has_pushed_wrapper = true;
+        self.has_strong_zero2one = true;
+    }
+}
+
+struct CountState {
+    /// The reference count.
+    count: usize,
+    /// Whether the process that owns this node thinks that we hold a refcount on it. (Note that
+    /// even if count is greater than one, we only increment it once in the owning process.)
+    has_count: bool,
+}
+
+impl CountState {
+    fn new() -> Self {
+        Self {
+            count: 0,
+            has_count: false,
+        }
+    }
+}
+
+struct NodeInner {
+    /// Strong refcounts held on this node by `NodeRef` objects.
+    strong: CountState,
+    /// Weak refcounts held on this node by `NodeRef` objects.
+    weak: CountState,
+    delivery_state: DeliveryState,
+    /// The binder driver guarantees that oneway transactions sent to the same node are serialized,
+    /// that is, userspace will not be given the next one until it has finished processing the
+    /// previous oneway transaction. This is done to avoid the case where two oneway transactions
+    /// arrive in opposite order from the order in which they were sent. (E.g., they could be
+    /// delivered to two different threads, which could appear as-if they were sent in opposite
+    /// order.)
+    ///
+    /// To fix that, we store pending oneway transactions in a separate list in the node, and don't
+    /// deliver the next oneway transaction until userspace signals that it has finished processing
+    /// the previous oneway transaction by calling the `BC_FREE_BUFFER` ioctl.
+    oneway_todo: List<DTRWrap<Transaction>>,
+    /// Keeps track of whether this node has a pending oneway transaction.
+    ///
+    /// When this is true, incoming oneway transactions are stored in `oneway_todo`, instead of
+    /// being delivered directly to the process.
+    has_oneway_transaction: bool,
+    /// List of processes to deliver a notification to when this node is destroyed (usually due to
+    /// the process dying).
+    death_list: List<DTRWrap<NodeDeath>, 1>,
+    /// List of processes to deliver freeze notifications to.
+    freeze_list: KVVec<Arc<Process>>,
+    /// The number of active BR_INCREFS or BR_ACQUIRE operations. (should be maximum two)
+    ///
+    /// If this is non-zero, then we postpone any BR_RELEASE or BR_DECREFS notifications until the
+    /// active operations have ended. This avoids the situation an increment and decrement get
+    /// reordered from userspace's perspective.
+    active_inc_refs: u8,
+    /// List of `NodeRefInfo` objects that reference this node.
+    refs: List<NodeRefInfo, { NodeRefInfo::LIST_NODE }>,
+}
+
+#[pin_data]
+pub(crate) struct Node {
+    pub(crate) debug_id: usize,
+    ptr: u64,
+    pub(crate) cookie: u64,
+    pub(crate) flags: u32,
+    pub(crate) owner: Arc<Process>,
+    inner: LockedBy<NodeInner, ProcessInner>,
+    #[pin]
+    links_track: AtomicTracker,
+}
+
+kernel::list::impl_list_arc_safe! {
+    impl ListArcSafe<0> for Node {
+        tracked_by links_track: AtomicTracker;
+    }
+}
+
+// Make `oneway_todo` work.
+kernel::list::impl_list_item! {
+    impl ListItem<0> for DTRWrap<Transaction> {
+        using ListLinks { self.links.inner };
+    }
+}
+
+impl Node {
+    pub(crate) fn new(
+        ptr: u64,
+        cookie: u64,
+        flags: u32,
+        owner: Arc<Process>,
+    ) -> impl PinInit<Self> {
+        pin_init!(Self {
+            inner: LockedBy::new(
+                &owner.inner,
+                NodeInner {
+                    strong: CountState::new(),
+                    weak: CountState::new(),
+                    delivery_state: DeliveryState {
+                        has_pushed_node: false,
+                        has_pushed_wrapper: false,
+                        has_weak_zero2one: false,
+                        has_strong_zero2one: false,
+                    },
+                    death_list: List::new(),
+                    oneway_todo: List::new(),
+                    freeze_list: KVVec::new(),
+                    has_oneway_transaction: false,
+                    active_inc_refs: 0,
+                    refs: List::new(),
+                },
+            ),
+            debug_id: super::next_debug_id(),
+            ptr,
+            cookie,
+            flags,
+            owner,
+            links_track <- AtomicTracker::new(),
+        })
+    }
+
+    pub(crate) fn has_oneway_transaction(&self, owner_inner: &mut ProcessInner) -> bool {
+        let inner = self.inner.access_mut(owner_inner);
+        inner.has_oneway_transaction
+    }
+
+    #[inline(never)]
+    pub(crate) fn full_debug_print(
+        &self,
+        m: &SeqFile,
+        owner_inner: &mut ProcessInner,
+    ) -> Result<()> {
+        let inner = self.inner.access_mut(owner_inner);
+        seq_print!(
+            m,
+            "  node {}: u{:016x} c{:016x} hs {} hw {} cs {} cw {}",
+            self.debug_id,
+            self.ptr,
+            self.cookie,
+            inner.strong.has_count,
+            inner.weak.has_count,
+            inner.strong.count,
+            inner.weak.count,
+        );
+        if !inner.refs.is_empty() {
+            seq_print!(m, " proc");
+            for node_ref in &inner.refs {
+                seq_print!(m, " {}", node_ref.process.task.pid());
+            }
+        }
+        seq_print!(m, "\n");
+        for t in &inner.oneway_todo {
+            t.debug_print_inner(m, "    pending async transaction ");
+        }
+        Ok(())
+    }
+
+    /// Insert the `NodeRef` into this `refs` list.
+    ///
+    /// # Safety
+    ///
+    /// It must be the case that `info.node_ref.node` is this node.
+    pub(crate) unsafe fn insert_node_info(
+        &self,
+        info: ListArc<NodeRefInfo, { NodeRefInfo::LIST_NODE }>,
+    ) {
+        self.inner
+            .access_mut(&mut self.owner.inner.lock())
+            .refs
+            .push_front(info);
+    }
+
+    /// Insert the `NodeRef` into this `refs` list.
+    ///
+    /// # Safety
+    ///
+    /// It must be the case that `info.node_ref.node` is this node.
+    pub(crate) unsafe fn remove_node_info(
+        &self,
+        info: &NodeRefInfo,
+    ) -> Option<ListArc<NodeRefInfo, { NodeRefInfo::LIST_NODE }>> {
+        // SAFETY: We always insert `NodeRefInfo` objects into the `refs` list of the node that it
+        // references in `info.node_ref.node`. That is this node, so `info` cannot possibly be in
+        // the `refs` list of another node.
+        unsafe {
+            self.inner
+                .access_mut(&mut self.owner.inner.lock())
+                .refs
+                .remove(info)
+        }
+    }
+
+    /// An id that is unique across all binder nodes on the system. Used as the key in the
+    /// `by_node` map.
+    pub(crate) fn global_id(&self) -> usize {
+        self as *const Node as usize
+    }
+
+    pub(crate) fn get_id(&self) -> (u64, u64) {
+        (self.ptr, self.cookie)
+    }
+
+    pub(crate) fn add_death(
+        &self,
+        death: ListArc<DTRWrap<NodeDeath>, 1>,
+        guard: &mut Guard<'_, ProcessInner, SpinLockBackend>,
+    ) {
+        self.inner.access_mut(guard).death_list.push_back(death);
+    }
+
+    pub(crate) fn inc_ref_done_locked(
+        self: &DArc<Node>,
+        _strong: bool,
+        owner_inner: &mut ProcessInner,
+    ) -> Option<DLArc<Node>> {
+        let inner = self.inner.access_mut(owner_inner);
+        if inner.active_inc_refs == 0 {
+            pr_err!("inc_ref_done called when no active inc_refs");
+            return None;
+        }
+
+        inner.active_inc_refs -= 1;
+        if inner.active_inc_refs == 0 {
+            // Having active inc_refs can inhibit dropping of ref-counts. Calculate whether we
+            // would send a refcount decrement, and if so, tell the caller to schedule us.
+            let strong = inner.strong.count > 0;
+            let has_strong = inner.strong.has_count;
+            let weak = strong || inner.weak.count > 0;
+            let has_weak = inner.weak.has_count;
+
+            let should_drop_weak = !weak && has_weak;
+            let should_drop_strong = !strong && has_strong;
+
+            // If we want to drop the ref-count again, tell the caller to schedule a work node for
+            // that.
+            let need_push = should_drop_weak || should_drop_strong;
+
+            if need_push && inner.delivery_state.should_normal_push() {
+                let list_arc = ListArc::try_from_arc(self.clone()).ok().unwrap();
+                inner.delivery_state.did_normal_push();
+                Some(list_arc)
+            } else {
+                None
+            }
+        } else {
+            None
+        }
+    }
+
+    pub(crate) fn update_refcount_locked(
+        self: &DArc<Node>,
+        inc: bool,
+        strong: bool,
+        count: usize,
+        owner_inner: &mut ProcessInner,
+    ) -> Option<DLArc<Node>> {
+        let is_dead = owner_inner.is_dead;
+        let inner = self.inner.access_mut(owner_inner);
+
+        // Get a reference to the state we'll update.
+        let state = if strong {
+            &mut inner.strong
+        } else {
+            &mut inner.weak
+        };
+
+        // Update the count and determine whether we need to push work.
+        let need_push = if inc {
+            state.count += count;
+            // TODO: This method shouldn't be used for zero-to-one increments.
+            !is_dead && !state.has_count
+        } else {
+            if state.count < count {
+                pr_err!("Failure: refcount underflow!");
+                return None;
+            }
+            state.count -= count;
+            !is_dead && state.count == 0 && state.has_count
+        };
+
+        if need_push && inner.delivery_state.should_normal_push() {
+            let list_arc = ListArc::try_from_arc(self.clone()).ok().unwrap();
+            inner.delivery_state.did_normal_push();
+            Some(list_arc)
+        } else {
+            None
+        }
+    }
+
+    pub(crate) fn incr_refcount_allow_zero2one(
+        self: &DArc<Self>,
+        strong: bool,
+        owner_inner: &mut ProcessInner,
+    ) -> Result<Option<DLArc<Node>>, CouldNotDeliverCriticalIncrement> {
+        let is_dead = owner_inner.is_dead;
+        let inner = self.inner.access_mut(owner_inner);
+
+        // Get a reference to the state we'll update.
+        let state = if strong {
+            &mut inner.strong
+        } else {
+            &mut inner.weak
+        };
+
+        // Update the count and determine whether we need to push work.
+        state.count += 1;
+        if is_dead || state.has_count {
+            return Ok(None);
+        }
+
+        // Userspace needs to be notified of this.
+        if !strong && inner.delivery_state.should_push_weak_zero2one() {
+            assert!(inner.delivery_state.can_push_weak_zero2one_normally());
+            let list_arc = ListArc::try_from_arc(self.clone()).ok().unwrap();
+            inner.delivery_state.did_push_weak_zero2one();
+            Ok(Some(list_arc))
+        } else if strong && inner.delivery_state.should_push_strong_zero2one() {
+            if inner.delivery_state.can_push_strong_zero2one_normally() {
+                let list_arc = ListArc::try_from_arc(self.clone()).ok().unwrap();
+                inner.delivery_state.did_push_strong_zero2one();
+                Ok(Some(list_arc))
+            } else {
+                state.count -= 1;
+                Err(CouldNotDeliverCriticalIncrement)
+            }
+        } else {
+            // Work is already pushed, and we don't need to push again.
+            Ok(None)
+        }
+    }
+
+    pub(crate) fn incr_refcount_allow_zero2one_with_wrapper(
+        self: &DArc<Self>,
+        strong: bool,
+        wrapper: CritIncrWrapper,
+        owner_inner: &mut ProcessInner,
+    ) -> Option<DLArc<dyn DeliverToRead>> {
+        match self.incr_refcount_allow_zero2one(strong, owner_inner) {
+            Ok(Some(node)) => Some(node as _),
+            Ok(None) => None,
+            Err(CouldNotDeliverCriticalIncrement) => {
+                assert!(strong);
+                let inner = self.inner.access_mut(owner_inner);
+                inner.strong.count += 1;
+                inner.delivery_state.did_push_strong_zero2one_wrapper();
+                Some(wrapper.init(self.clone()))
+            }
+        }
+    }
+
+    pub(crate) fn update_refcount(self: &DArc<Self>, inc: bool, count: usize, strong: bool) {
+        self.owner
+            .inner
+            .lock()
+            .update_node_refcount(self, inc, strong, count, None);
+    }
+
+    pub(crate) fn populate_counts(
+        &self,
+        out: &mut BinderNodeInfoForRef,
+        guard: &Guard<'_, ProcessInner, SpinLockBackend>,
+    ) {
+        let inner = self.inner.access(guard);
+        out.strong_count = inner.strong.count as _;
+        out.weak_count = inner.weak.count as _;
+    }
+
+    pub(crate) fn populate_debug_info(
+        &self,
+        out: &mut BinderNodeDebugInfo,
+        guard: &Guard<'_, ProcessInner, SpinLockBackend>,
+    ) {
+        out.ptr = self.ptr as _;
+        out.cookie = self.cookie as _;
+        let inner = self.inner.access(guard);
+        if inner.strong.has_count {
+            out.has_strong_ref = 1;
+        }
+        if inner.weak.has_count {
+            out.has_weak_ref = 1;
+        }
+    }
+
+    pub(crate) fn force_has_count(&self, guard: &mut Guard<'_, ProcessInner, SpinLockBackend>) {
+        let inner = self.inner.access_mut(guard);
+        inner.strong.has_count = true;
+        inner.weak.has_count = true;
+    }
+
+    fn write(&self, writer: &mut BinderReturnWriter<'_>, code: u32) -> Result {
+        writer.write_code(code)?;
+        writer.write_payload(&self.ptr)?;
+        writer.write_payload(&self.cookie)?;
+        Ok(())
+    }
+
+    pub(crate) fn submit_oneway(
+        &self,
+        transaction: DLArc<Transaction>,
+        guard: &mut Guard<'_, ProcessInner, SpinLockBackend>,
+    ) -> Result<(), (BinderError, DLArc<dyn DeliverToRead>)> {
+        if guard.is_dead {
+            return Err((BinderError::new_dead(), transaction));
+        }
+
+        let inner = self.inner.access_mut(guard);
+        if inner.has_oneway_transaction {
+            inner.oneway_todo.push_back(transaction);
+        } else {
+            inner.has_oneway_transaction = true;
+            guard.push_work(transaction)?;
+        }
+        Ok(())
+    }
+
+    pub(crate) fn release(&self) {
+        let mut guard = self.owner.inner.lock();
+        while let Some(work) = self.inner.access_mut(&mut guard).oneway_todo.pop_front() {
+            drop(guard);
+            work.into_arc().cancel();
+            guard = self.owner.inner.lock();
+        }
+
+        while let Some(death) = self.inner.access_mut(&mut guard).death_list.pop_front() {
+            drop(guard);
+            death.into_arc().set_dead();
+            guard = self.owner.inner.lock();
+        }
+    }
+
+    pub(crate) fn pending_oneway_finished(&self) {
+        let mut guard = self.owner.inner.lock();
+        if guard.is_dead {
+            // Cleanup will happen in `Process::deferred_release`.
+            return;
+        }
+
+        let inner = self.inner.access_mut(&mut guard);
+
+        let transaction = inner.oneway_todo.pop_front();
+        inner.has_oneway_transaction = transaction.is_some();
+        if let Some(transaction) = transaction {
+            match guard.push_work(transaction) {
+                Ok(()) => {}
+                Err((_err, work)) => {
+                    // Process is dead.
+                    // This shouldn't happen due to the `is_dead` check, but if it does, just drop
+                    // the transaction and return.
+                    drop(guard);
+                    drop(work);
+                }
+            }
+        }
+    }
+
+    /// Finds an outdated transaction that the given transaction can replace.
+    ///
+    /// If one is found, it is removed from the list and returned.
+    pub(crate) fn take_outdated_transaction(
+        &self,
+        new: &Transaction,
+        guard: &mut Guard<'_, ProcessInner, SpinLockBackend>,
+    ) -> Option<DLArc<Transaction>> {
+        let inner = self.inner.access_mut(guard);
+        let mut cursor = inner.oneway_todo.cursor_front();
+        while let Some(next) = cursor.peek_next() {
+            if new.can_replace(&next) {
+                return Some(next.remove());
+            }
+            cursor.move_next();
+        }
+        None
+    }
+
+    /// This is split into a separate function since it's called by both `Node::do_work` and
+    /// `NodeWrapper::do_work`.
+    fn do_work_locked(
+        &self,
+        writer: &mut BinderReturnWriter<'_>,
+        mut guard: Guard<'_, ProcessInner, SpinLockBackend>,
+    ) -> Result<bool> {
+        let inner = self.inner.access_mut(&mut guard);
+        let strong = inner.strong.count > 0;
+        let has_strong = inner.strong.has_count;
+        let weak = strong || inner.weak.count > 0;
+        let has_weak = inner.weak.has_count;
+
+        if weak && !has_weak {
+            inner.weak.has_count = true;
+            inner.active_inc_refs += 1;
+        }
+
+        if strong && !has_strong {
+            inner.strong.has_count = true;
+            inner.active_inc_refs += 1;
+        }
+
+        let no_active_inc_refs = inner.active_inc_refs == 0;
+        let should_drop_weak = no_active_inc_refs && (!weak && has_weak);
+        let should_drop_strong = no_active_inc_refs && (!strong && has_strong);
+        if should_drop_weak {
+            inner.weak.has_count = false;
+        }
+        if should_drop_strong {
+            inner.strong.has_count = false;
+        }
+        if no_active_inc_refs && !weak {
+            // Remove the node if there are no references to it.
+            guard.remove_node(self.ptr);
+        }
+        drop(guard);
+
+        if weak && !has_weak {
+            self.write(writer, BR_INCREFS)?;
+        }
+        if strong && !has_strong {
+            self.write(writer, BR_ACQUIRE)?;
+        }
+        if should_drop_strong {
+            self.write(writer, BR_RELEASE)?;
+        }
+        if should_drop_weak {
+            self.write(writer, BR_DECREFS)?;
+        }
+
+        Ok(true)
+    }
+
+    pub(crate) fn add_freeze_listener(
+        &self,
+        process: &Arc<Process>,
+        flags: kernel::alloc::Flags,
+    ) -> Result {
+        let mut vec_alloc = KVVec::<Arc<Process>>::new();
+        loop {
+            let mut guard = self.owner.inner.lock();
+            // Do not check for `guard.dead`. The `dead` flag that matters here is the owner of the
+            // listener, no the target.
+            let inner = self.inner.access_mut(&mut guard);
+            let len = inner.freeze_list.len();
+            if len >= inner.freeze_list.capacity() {
+                if len >= vec_alloc.capacity() {
+                    drop(guard);
+                    vec_alloc = KVVec::with_capacity((1 + len).next_power_of_two(), flags)?;
+                    continue;
+                }
+                mem::swap(&mut inner.freeze_list, &mut vec_alloc);
+                for elem in vec_alloc.drain_all() {
+                    inner.freeze_list.push_within_capacity(elem)?;
+                }
+            }
+            inner.freeze_list.push_within_capacity(process.clone())?;
+            return Ok(());
+        }
+    }
+
+    pub(crate) fn remove_freeze_listener(&self, p: &Arc<Process>) {
+        let _unused_capacity;
+        let mut guard = self.owner.inner.lock();
+        let inner = self.inner.access_mut(&mut guard);
+        let len = inner.freeze_list.len();
+        inner.freeze_list.retain(|proc| !Arc::ptr_eq(proc, p));
+        if len == inner.freeze_list.len() {
+            pr_warn!(
+                "Could not remove freeze listener for {}\n",
+                p.pid_in_current_ns()
+            );
+        }
+        if inner.freeze_list.is_empty() {
+            _unused_capacity = mem::take(&mut inner.freeze_list);
+        }
+    }
+
+    pub(crate) fn freeze_list<'a>(&'a self, guard: &'a ProcessInner) -> &'a [Arc<Process>] {
+        &self.inner.access(guard).freeze_list
+    }
+}
+
+impl DeliverToRead for Node {
+    fn do_work(
+        self: DArc<Self>,
+        _thread: &Thread,
+        writer: &mut BinderReturnWriter<'_>,
+    ) -> Result<bool> {
+        let mut owner_inner = self.owner.inner.lock();
+        let inner = self.inner.access_mut(&mut owner_inner);
+
+        assert!(inner.delivery_state.has_pushed_node);
+        if inner.delivery_state.has_pushed_wrapper {
+            // If the wrapper is scheduled, then we are either a normal push or weak zero2one
+            // increment, and the wrapper is a strong zero2one increment, so the wrapper always
+            // takes precedence over us.
+            assert!(inner.delivery_state.has_strong_zero2one);
+            inner.delivery_state.has_pushed_node = false;
+            inner.delivery_state.has_weak_zero2one = false;
+            return Ok(true);
+        }
+
+        inner.delivery_state.has_pushed_node = false;
+        inner.delivery_state.has_weak_zero2one = false;
+        inner.delivery_state.has_strong_zero2one = false;
+
+        self.do_work_locked(writer, owner_inner)
+    }
+
+    fn cancel(self: DArc<Self>) {}
+
+    fn should_sync_wakeup(&self) -> bool {
+        false
+    }
+
+    #[inline(never)]
+    fn debug_print(&self, m: &SeqFile, prefix: &str, _tprefix: &str) -> Result<()> {
+        seq_print!(
+            m,
+            "{}node work {}: u{:016x} c{:016x}\n",
+            prefix,
+            self.debug_id,
+            self.ptr,
+            self.cookie,
+        );
+        Ok(())
+    }
+}
+
+/// Represents something that holds one or more ref-counts to a `Node`.
+///
+/// Whenever process A holds a refcount to a node owned by a different process B, then process A
+/// will store a `NodeRef` that refers to the `Node` in process B. When process A releases the
+/// refcount, we destroy the NodeRef, which decrements the ref-count in process A.
+///
+/// This type is also used for some other cases. For example, a transaction allocation holds a
+/// refcount on the target node, and this is implemented by storing a `NodeRef` in the allocation
+/// so that the destructor of the allocation will drop a refcount of the `Node`.
+pub(crate) struct NodeRef {
+    pub(crate) node: DArc<Node>,
+    /// How many times does this NodeRef hold a refcount on the Node?
+    strong_node_count: usize,
+    weak_node_count: usize,
+    /// How many times does userspace hold a refcount on this NodeRef?
+    strong_count: usize,
+    weak_count: usize,
+}
+
+impl NodeRef {
+    pub(crate) fn new(node: DArc<Node>, strong_count: usize, weak_count: usize) -> Self {
+        Self {
+            node,
+            strong_node_count: strong_count,
+            weak_node_count: weak_count,
+            strong_count,
+            weak_count,
+        }
+    }
+
+    pub(crate) fn absorb(&mut self, mut other: Self) {
+        assert!(
+            Arc::ptr_eq(&self.node, &other.node),
+            "absorb called with differing nodes"
+        );
+        self.strong_node_count += other.strong_node_count;
+        self.weak_node_count += other.weak_node_count;
+        self.strong_count += other.strong_count;
+        self.weak_count += other.weak_count;
+        other.strong_count = 0;
+        other.weak_count = 0;
+        other.strong_node_count = 0;
+        other.weak_node_count = 0;
+
+        if self.strong_node_count >= 2 || self.weak_node_count >= 2 {
+            let mut guard = self.node.owner.inner.lock();
+            let inner = self.node.inner.access_mut(&mut guard);
+
+            if self.strong_node_count >= 2 {
+                inner.strong.count -= self.strong_node_count - 1;
+                self.strong_node_count = 1;
+                assert_ne!(inner.strong.count, 0);
+            }
+            if self.weak_node_count >= 2 {
+                inner.weak.count -= self.weak_node_count - 1;
+                self.weak_node_count = 1;
+                assert_ne!(inner.weak.count, 0);
+            }
+        }
+    }
+
+    pub(crate) fn get_count(&self) -> (usize, usize) {
+        (self.strong_count, self.weak_count)
+    }
+
+    pub(crate) fn clone(&self, strong: bool) -> Result<NodeRef> {
+        if strong && self.strong_count == 0 {
+            return Err(EINVAL);
+        }
+        Ok(self
+            .node
+            .owner
+            .inner
+            .lock()
+            .new_node_ref(self.node.clone(), strong, None))
+    }
+
+    /// Updates (increments or decrements) the number of references held against the node. If the
+    /// count being updated transitions from 0 to 1 or from 1 to 0, the node is notified by having
+    /// its `update_refcount` function called.
+    ///
+    /// Returns whether `self` should be removed (when both counts are zero).
+    pub(crate) fn update(&mut self, inc: bool, strong: bool) -> bool {
+        if strong && self.strong_count == 0 {
+            return false;
+        }
+        let (count, node_count, other_count) = if strong {
+            (
+                &mut self.strong_count,
+                &mut self.strong_node_count,
+                self.weak_count,
+            )
+        } else {
+            (
+                &mut self.weak_count,
+                &mut self.weak_node_count,
+                self.strong_count,
+            )
+        };
+        if inc {
+            if *count == 0 {
+                *node_count = 1;
+                self.node.update_refcount(true, 1, strong);
+            }
+            *count += 1;
+        } else {
+            if *count == 0 {
+                pr_warn!(
+                    "pid {} performed invalid decrement on ref\n",
+                    kernel::current!().pid()
+                );
+                return false;
+            }
+            *count -= 1;
+            if *count == 0 {
+                self.node.update_refcount(false, *node_count, strong);
+                *node_count = 0;
+                return other_count == 0;
+            }
+        }
+        false
+    }
+}
+
+impl Drop for NodeRef {
+    // This destructor is called conditionally from `Allocation::drop`. That branch is often
+    // mispredicted. Inlining this method call reduces the cost of those branch mispredictions.
+    #[inline(always)]
+    fn drop(&mut self) {
+        if self.strong_node_count > 0 {
+            self.node
+                .update_refcount(false, self.strong_node_count, true);
+        }
+        if self.weak_node_count > 0 {
+            self.node
+                .update_refcount(false, self.weak_node_count, false);
+        }
+    }
+}
+
+struct NodeDeathInner {
+    dead: bool,
+    cleared: bool,
+    notification_done: bool,
+    /// Indicates whether the normal flow was interrupted by removing the handle. In this case, we
+    /// need behave as if the death notification didn't exist (i.e., we don't deliver anything to
+    /// the user.
+    aborted: bool,
+}
+
+/// Used to deliver notifications when a process dies.
+///
+/// A process can request to be notified when a process dies using `BC_REQUEST_DEATH_NOTIFICATION`.
+/// This will make the driver send a `BR_DEAD_BINDER` to userspace when the process dies (or
+/// immediately if it is already dead). Userspace is supposed to respond with `BC_DEAD_BINDER_DONE`
+/// once it has processed the notification.
+///
+/// Userspace can unregister from death notifications using the `BC_CLEAR_DEATH_NOTIFICATION`
+/// command. In this case, the kernel will respond with `BR_CLEAR_DEATH_NOTIFICATION_DONE` once the
+/// notification has been removed. Note that if the remote process dies before the kernel has
+/// responded with `BR_CLEAR_DEATH_NOTIFICATION_DONE`, then the kernel will still send a
+/// `BR_DEAD_BINDER`, which userspace must be able to process. In this case, the kernel will wait
+/// for the `BC_DEAD_BINDER_DONE` command before it sends `BR_CLEAR_DEATH_NOTIFICATION_DONE`.
+///
+/// Note that even if the kernel sends a `BR_DEAD_BINDER`, this does not remove the death
+/// notification. Userspace must still remove it manually using `BC_CLEAR_DEATH_NOTIFICATION`.
+///
+/// If a process uses `BC_RELEASE` to destroy its last refcount on a node that has an active death
+/// registration, then the death registration is immediately deleted (we implement this using the
+/// `aborted` field). However, userspace is not supposed to delete a `NodeRef` without first
+/// deregistering death notifications, so this codepath is not executed under normal circumstances.
+#[pin_data]
+pub(crate) struct NodeDeath {
+    node: DArc<Node>,
+    process: Arc<Process>,
+    pub(crate) cookie: u64,
+    #[pin]
+    links_track: AtomicTracker<0>,
+    /// Used by the owner `Node` to store a list of registered death notifications.
+    ///
+    /// # Invariants
+    ///
+    /// Only ever used with the `death_list` list of `self.node`.
+    #[pin]
+    death_links: ListLinks<1>,
+    /// Used by the process to keep track of the death notifications for which we have sent a
+    /// `BR_DEAD_BINDER` but not yet received a `BC_DEAD_BINDER_DONE`.
+    ///
+    /// # Invariants
+    ///
+    /// Only ever used with the `delivered_deaths` list of `self.process`.
+    #[pin]
+    delivered_links: ListLinks<2>,
+    #[pin]
+    delivered_links_track: AtomicTracker<2>,
+    #[pin]
+    inner: SpinLock<NodeDeathInner>,
+}
+
+impl NodeDeath {
+    /// Constructs a new node death notification object.
+    pub(crate) fn new(
+        node: DArc<Node>,
+        process: Arc<Process>,
+        cookie: u64,
+    ) -> impl PinInit<DTRWrap<Self>> {
+        DTRWrap::new(pin_init!(
+            Self {
+                node,
+                process,
+                cookie,
+                links_track <- AtomicTracker::new(),
+                death_links <- ListLinks::new(),
+                delivered_links <- ListLinks::new(),
+                delivered_links_track <- AtomicTracker::new(),
+                inner <- kernel::new_spinlock!(NodeDeathInner {
+                    dead: false,
+                    cleared: false,
+                    notification_done: false,
+                    aborted: false,
+                }, "NodeDeath::inner"),
+            }
+        ))
+    }
+
+    /// Sets the cleared flag to `true`.
+    ///
+    /// It removes `self` from the node's death notification list if needed.
+    ///
+    /// Returns whether it needs to be queued.
+    pub(crate) fn set_cleared(self: &DArc<Self>, abort: bool) -> bool {
+        let (needs_removal, needs_queueing) = {
+            // Update state and determine if we need to queue a work item. We only need to do it
+            // when the node is not dead or if the user already completed the death notification.
+            let mut inner = self.inner.lock();
+            if abort {
+                inner.aborted = true;
+            }
+            if inner.cleared {
+                // Already cleared.
+                return false;
+            }
+            inner.cleared = true;
+            (!inner.dead, !inner.dead || inner.notification_done)
+        };
+
+        // Remove death notification from node.
+        if needs_removal {
+            let mut owner_inner = self.node.owner.inner.lock();
+            let node_inner = self.node.inner.access_mut(&mut owner_inner);
+            // SAFETY: A `NodeDeath` is never inserted into the death list of any node other than
+            // its owner, so it is either in this death list or in no death list.
+            unsafe { node_inner.death_list.remove(self) };
+        }
+        needs_queueing
+    }
+
+    /// Sets the 'notification done' flag to `true`.
+    pub(crate) fn set_notification_done(self: DArc<Self>, thread: &Thread) {
+        let needs_queueing = {
+            let mut inner = self.inner.lock();
+            inner.notification_done = true;
+            inner.cleared
+        };
+        if needs_queueing {
+            if let Some(death) = ListArc::try_from_arc_or_drop(self) {
+                let _ = thread.push_work_if_looper(death);
+            }
+        }
+    }
+
+    /// Sets the 'dead' flag to `true` and queues work item if needed.
+    pub(crate) fn set_dead(self: DArc<Self>) {
+        let needs_queueing = {
+            let mut inner = self.inner.lock();
+            if inner.cleared {
+                false
+            } else {
+                inner.dead = true;
+                true
+            }
+        };
+        if needs_queueing {
+            // Push the death notification to the target process. There is nothing else to do if
+            // it's already dead.
+            if let Some(death) = ListArc::try_from_arc_or_drop(self) {
+                let process = death.process.clone();
+                let _ = process.push_work(death);
+            }
+        }
+    }
+}
+
+kernel::list::impl_list_arc_safe! {
+    impl ListArcSafe<0> for NodeDeath {
+        tracked_by links_track: AtomicTracker;
+    }
+}
+
+kernel::list::impl_list_arc_safe! {
+    impl ListArcSafe<1> for DTRWrap<NodeDeath> { untracked; }
+}
+kernel::list::impl_list_item! {
+    impl ListItem<1> for DTRWrap<NodeDeath> {
+        using ListLinks { self.wrapped.death_links };
+    }
+}
+
+kernel::list::impl_list_arc_safe! {
+    impl ListArcSafe<2> for DTRWrap<NodeDeath> {
+        tracked_by wrapped: NodeDeath;
+    }
+}
+kernel::list::impl_list_arc_safe! {
+    impl ListArcSafe<2> for NodeDeath {
+        tracked_by delivered_links_track: AtomicTracker<2>;
+    }
+}
+kernel::list::impl_list_item! {
+    impl ListItem<2> for DTRWrap<NodeDeath> {
+        using ListLinks { self.wrapped.delivered_links };
+    }
+}
+
+impl DeliverToRead for NodeDeath {
+    fn do_work(
+        self: DArc<Self>,
+        _thread: &Thread,
+        writer: &mut BinderReturnWriter<'_>,
+    ) -> Result<bool> {
+        let done = {
+            let inner = self.inner.lock();
+            if inner.aborted {
+                return Ok(true);
+            }
+            inner.cleared && (!inner.dead || inner.notification_done)
+        };
+
+        let cookie = self.cookie;
+        let cmd = if done {
+            BR_CLEAR_DEATH_NOTIFICATION_DONE
+        } else {
+            let process = self.process.clone();
+            let mut process_inner = process.inner.lock();
+            let inner = self.inner.lock();
+            if inner.aborted {
+                return Ok(true);
+            }
+            // We're still holding the inner lock, so it cannot be aborted while we insert it into
+            // the delivered list.
+            process_inner.death_delivered(self.clone());
+            BR_DEAD_BINDER
+        };
+
+        writer.write_code(cmd)?;
+        writer.write_payload(&cookie)?;
+        // DEAD_BINDER notifications can cause transactions, so stop processing work items when we
+        // get to a death notification.
+        Ok(cmd != BR_DEAD_BINDER)
+    }
+
+    fn cancel(self: DArc<Self>) {}
+
+    fn should_sync_wakeup(&self) -> bool {
+        false
+    }
+
+    #[inline(never)]
+    fn debug_print(&self, m: &SeqFile, prefix: &str, _tprefix: &str) -> Result<()> {
+        let inner = self.inner.lock();
+
+        let dead_binder = inner.dead && !inner.notification_done;
+
+        if dead_binder {
+            if inner.cleared {
+                seq_print!(m, "{}has cleared dead binder\n", prefix);
+            } else {
+                seq_print!(m, "{}has dead binder\n", prefix);
+            }
+        } else {
+            seq_print!(m, "{}has cleared death notification\n", prefix);
+        }
+
+        Ok(())
+    }
+}
diff --git a/drivers/android/binder/node/wrapper.rs b/drivers/android/binder/node/wrapper.rs
new file mode 100644
index 000000000000..43294c050502
--- /dev/null
+++ b/drivers/android/binder/node/wrapper.rs
@@ -0,0 +1,78 @@
+// SPDX-License-Identifier: GPL-2.0
+
+// Copyright (C) 2025 Google LLC.
+
+use kernel::{list::ListArc, prelude::*, seq_file::SeqFile, seq_print, sync::UniqueArc};
+
+use crate::{node::Node, thread::Thread, BinderReturnWriter, DArc, DLArc, DTRWrap, DeliverToRead};
+
+use core::mem::MaybeUninit;
+
+pub(crate) struct CritIncrWrapper {
+    inner: UniqueArc<MaybeUninit<DTRWrap<NodeWrapper>>>,
+}
+
+impl CritIncrWrapper {
+    pub(crate) fn new() -> Result<Self> {
+        Ok(CritIncrWrapper {
+            inner: UniqueArc::new_uninit(GFP_KERNEL)?,
+        })
+    }
+
+    pub(super) fn init(self, node: DArc<Node>) -> DLArc<dyn DeliverToRead> {
+        match self.inner.pin_init_with(DTRWrap::new(NodeWrapper { node })) {
+            Ok(initialized) => ListArc::from(initialized) as _,
+            Err(err) => match err {},
+        }
+    }
+}
+
+struct NodeWrapper {
+    node: DArc<Node>,
+}
+
+kernel::list::impl_list_arc_safe! {
+    impl ListArcSafe<0> for NodeWrapper {
+        untracked;
+    }
+}
+
+impl DeliverToRead for NodeWrapper {
+    fn do_work(
+        self: DArc<Self>,
+        _thread: &Thread,
+        writer: &mut BinderReturnWriter<'_>,
+    ) -> Result<bool> {
+        let node = &self.node;
+        let mut owner_inner = node.owner.inner.lock();
+        let inner = node.inner.access_mut(&mut owner_inner);
+
+        let ds = &mut inner.delivery_state;
+
+        assert!(ds.has_pushed_wrapper);
+        assert!(ds.has_strong_zero2one);
+        ds.has_pushed_wrapper = false;
+        ds.has_strong_zero2one = false;
+
+        node.do_work_locked(writer, owner_inner)
+    }
+
+    fn cancel(self: DArc<Self>) {}
+
+    fn should_sync_wakeup(&self) -> bool {
+        false
+    }
+
+    #[inline(never)]
+    fn debug_print(&self, m: &SeqFile, prefix: &str, _tprefix: &str) -> Result<()> {
+        seq_print!(
+            m,
+            "{}node work {}: u{:016x} c{:016x}\n",
+            prefix,
+            self.node.debug_id,
+            self.node.ptr,
+            self.node.cookie,
+        );
+        Ok(())
+    }
+}
diff --git a/drivers/android/binder/page_range.rs b/drivers/android/binder/page_range.rs
new file mode 100644
index 000000000000..9379038f61f5
--- /dev/null
+++ b/drivers/android/binder/page_range.rs
@@ -0,0 +1,734 @@
+// SPDX-License-Identifier: GPL-2.0
+
+// Copyright (C) 2025 Google LLC.
+
+//! This module has utilities for managing a page range where unused pages may be reclaimed by a
+//! vma shrinker.
+
+// To avoid deadlocks, locks are taken in the order:
+//
+//  1. mmap lock
+//  2. spinlock
+//  3. lru spinlock
+//
+// The shrinker will use trylock methods because it locks them in a different order.
+
+use core::{
+    marker::PhantomPinned,
+    mem::{size_of, size_of_val, MaybeUninit},
+    ptr,
+};
+
+use kernel::{
+    bindings,
+    error::Result,
+    ffi::{c_ulong, c_void},
+    mm::{virt, Mm, MmWithUser},
+    new_mutex, new_spinlock,
+    page::{Page, PAGE_SHIFT, PAGE_SIZE},
+    prelude::*,
+    str::CStr,
+    sync::{aref::ARef, Mutex, SpinLock},
+    task::Pid,
+    transmute::FromBytes,
+    types::Opaque,
+    uaccess::UserSliceReader,
+};
+
+/// Represents a shrinker that can be registered with the kernel.
+///
+/// Each shrinker can be used by many `ShrinkablePageRange` objects.
+#[repr(C)]
+pub(crate) struct Shrinker {
+    inner: Opaque<*mut bindings::shrinker>,
+    list_lru: Opaque<bindings::list_lru>,
+}
+
+// SAFETY: The shrinker and list_lru are thread safe.
+unsafe impl Send for Shrinker {}
+// SAFETY: The shrinker and list_lru are thread safe.
+unsafe impl Sync for Shrinker {}
+
+impl Shrinker {
+    /// Create a new shrinker.
+    ///
+    /// # Safety
+    ///
+    /// Before using this shrinker with a `ShrinkablePageRange`, the `register` method must have
+    /// been called exactly once, and it must not have returned an error.
+    pub(crate) const unsafe fn new() -> Self {
+        Self {
+            inner: Opaque::uninit(),
+            list_lru: Opaque::uninit(),
+        }
+    }
+
+    /// Register this shrinker with the kernel.
+    pub(crate) fn register(&'static self, name: &CStr) -> Result<()> {
+        // SAFETY: These fields are not yet used, so it's okay to zero them.
+        unsafe {
+            self.inner.get().write(ptr::null_mut());
+            self.list_lru.get().write_bytes(0, 1);
+        }
+
+        // SAFETY: The field is not yet used, so we can initialize it.
+        let ret = unsafe { bindings::__list_lru_init(self.list_lru.get(), false, ptr::null_mut()) };
+        if ret != 0 {
+            return Err(Error::from_errno(ret));
+        }
+
+        // SAFETY: The `name` points at a valid c string.
+        let shrinker = unsafe { bindings::shrinker_alloc(0, name.as_char_ptr()) };
+        if shrinker.is_null() {
+            // SAFETY: We initialized it, so its okay to destroy it.
+            unsafe { bindings::list_lru_destroy(self.list_lru.get()) };
+            return Err(Error::from_errno(ret));
+        }
+
+        // SAFETY: We're about to register the shrinker, and these are the fields we need to
+        // initialize. (All other fields are already zeroed.)
+        unsafe {
+            (&raw mut (*shrinker).count_objects).write(Some(rust_shrink_count));
+            (&raw mut (*shrinker).scan_objects).write(Some(rust_shrink_scan));
+            (&raw mut (*shrinker).private_data).write(self.list_lru.get().cast());
+        }
+
+        // SAFETY: The new shrinker has been fully initialized, so we can register it.
+        unsafe { bindings::shrinker_register(shrinker) };
+
+        // SAFETY: This initializes the pointer to the shrinker so that we can use it.
+        unsafe { self.inner.get().write(shrinker) };
+
+        Ok(())
+    }
+}
+
+/// A container that manages a page range in a vma.
+///
+/// The pages can be thought of as an array of booleans of whether the pages are usable. The
+/// methods `use_range` and `stop_using_range` set all booleans in a range to true or false
+/// respectively. Initially, no pages are allocated. When a page is not used, it is not freed
+/// immediately. Instead, it is made available to the memory shrinker to free it if the device is
+/// under memory pressure.
+///
+/// It's okay for `use_range` and `stop_using_range` to race with each other, although there's no
+/// way to know whether an index ends up with true or false if a call to `use_range` races with
+/// another call to `stop_using_range` on a given index.
+///
+/// It's also okay for the two methods to race with themselves, e.g. if two threads call
+/// `use_range` on the same index, then that's fine and neither call will return until the page is
+/// allocated and mapped.
+///
+/// The methods that read or write to a range require that the page is marked as in use. So it is
+/// _not_ okay to call `stop_using_range` on a page that is in use by the methods that read or
+/// write to the page.
+#[pin_data(PinnedDrop)]
+pub(crate) struct ShrinkablePageRange {
+    /// Shrinker object registered with the kernel.
+    shrinker: &'static Shrinker,
+    /// Pid using this page range. Only used as debugging information.
+    pid: Pid,
+    /// The mm for the relevant process.
+    mm: ARef<Mm>,
+    /// Used to synchronize calls to `vm_insert_page` and `zap_page_range_single`.
+    #[pin]
+    mm_lock: Mutex<()>,
+    /// Spinlock protecting changes to pages.
+    #[pin]
+    lock: SpinLock<Inner>,
+
+    /// Must not move, since page info has pointers back.
+    #[pin]
+    _pin: PhantomPinned,
+}
+
+struct Inner {
+    /// Array of pages.
+    ///
+    /// Since this is also accessed by the shrinker, we can't use a `Box`, which asserts exclusive
+    /// ownership. To deal with that, we manage it using raw pointers.
+    pages: *mut PageInfo,
+    /// Length of the `pages` array.
+    size: usize,
+    /// The address of the vma to insert the pages into.
+    vma_addr: usize,
+}
+
+// SAFETY: proper locking is in place for `Inner`
+unsafe impl Send for Inner {}
+
+type StableMmGuard =
+    kernel::sync::lock::Guard<'static, (), kernel::sync::lock::mutex::MutexBackend>;
+
+/// An array element that describes the current state of a page.
+///
+/// There are three states:
+///
+///  * Free. The page is None. The `lru` element is not queued.
+///  * Available. The page is Some. The `lru` element is queued to the shrinker's lru.
+///  * Used. The page is Some. The `lru` element is not queued.
+///
+/// When an element is available, the shrinker is able to free the page.
+#[repr(C)]
+struct PageInfo {
+    lru: bindings::list_head,
+    page: Option<Page>,
+    range: *const ShrinkablePageRange,
+}
+
+impl PageInfo {
+    /// # Safety
+    ///
+    /// The caller ensures that writing to `me.page` is ok, and that the page is not currently set.
+    unsafe fn set_page(me: *mut PageInfo, page: Page) {
+        // SAFETY: This pointer offset is in bounds.
+        let ptr = unsafe { &raw mut (*me).page };
+
+        // SAFETY: The pointer is valid for writing, so also valid for reading.
+        if unsafe { (*ptr).is_some() } {
+            pr_err!("set_page called when there is already a page");
+            // SAFETY: We will initialize the page again below.
+            unsafe { ptr::drop_in_place(ptr) };
+        }
+
+        // SAFETY: The pointer is valid for writing.
+        unsafe { ptr::write(ptr, Some(page)) };
+    }
+
+    /// # Safety
+    ///
+    /// The caller ensures that reading from `me.page` is ok for the duration of 'a.
+    unsafe fn get_page<'a>(me: *const PageInfo) -> Option<&'a Page> {
+        // SAFETY: This pointer offset is in bounds.
+        let ptr = unsafe { &raw const (*me).page };
+
+        // SAFETY: The pointer is valid for reading.
+        unsafe { (*ptr).as_ref() }
+    }
+
+    /// # Safety
+    ///
+    /// The caller ensures that writing to `me.page` is ok for the duration of 'a.
+    unsafe fn take_page(me: *mut PageInfo) -> Option<Page> {
+        // SAFETY: This pointer offset is in bounds.
+        let ptr = unsafe { &raw mut (*me).page };
+
+        // SAFETY: The pointer is valid for reading.
+        unsafe { (*ptr).take() }
+    }
+
+    /// Add this page to the lru list, if not already in the list.
+    ///
+    /// # Safety
+    ///
+    /// The pointer must be valid, and it must be the right shrinker and nid.
+    unsafe fn list_lru_add(me: *mut PageInfo, nid: i32, shrinker: &'static Shrinker) {
+        // SAFETY: This pointer offset is in bounds.
+        let lru_ptr = unsafe { &raw mut (*me).lru };
+        // SAFETY: The lru pointer is valid, and we're not using it with any other lru list.
+        unsafe { bindings::list_lru_add(shrinker.list_lru.get(), lru_ptr, nid, ptr::null_mut()) };
+    }
+
+    /// Remove this page from the lru list, if it is in the list.
+    ///
+    /// # Safety
+    ///
+    /// The pointer must be valid, and it must be the right shrinker and nid.
+    unsafe fn list_lru_del(me: *mut PageInfo, nid: i32, shrinker: &'static Shrinker) {
+        // SAFETY: This pointer offset is in bounds.
+        let lru_ptr = unsafe { &raw mut (*me).lru };
+        // SAFETY: The lru pointer is valid, and we're not using it with any other lru list.
+        unsafe { bindings::list_lru_del(shrinker.list_lru.get(), lru_ptr, nid, ptr::null_mut()) };
+    }
+}
+
+impl ShrinkablePageRange {
+    /// Create a new `ShrinkablePageRange` using the given shrinker.
+    pub(crate) fn new(shrinker: &'static Shrinker) -> impl PinInit<Self, Error> {
+        try_pin_init!(Self {
+            shrinker,
+            pid: kernel::current!().pid(),
+            mm: ARef::from(&**kernel::current!().mm().ok_or(ESRCH)?),
+            mm_lock <- new_mutex!((), "ShrinkablePageRange::mm"),
+            lock <- new_spinlock!(Inner {
+                pages: ptr::null_mut(),
+                size: 0,
+                vma_addr: 0,
+            }, "ShrinkablePageRange"),
+            _pin: PhantomPinned,
+        })
+    }
+
+    pub(crate) fn stable_trylock_mm(&self) -> Option<StableMmGuard> {
+        // SAFETY: This extends the duration of the reference. Since this call happens before
+        // `mm_lock` is taken in the destructor of `ShrinkablePageRange`, the destructor will block
+        // until the returned guard is dropped. This ensures that the guard is valid until dropped.
+        let mm_lock = unsafe { &*ptr::from_ref(&self.mm_lock) };
+
+        mm_lock.try_lock()
+    }
+
+    /// Register a vma with this page range. Returns the size of the region.
+    pub(crate) fn register_with_vma(&self, vma: &virt::VmaNew) -> Result<usize> {
+        let num_bytes = usize::min(vma.end() - vma.start(), bindings::SZ_4M as usize);
+        let num_pages = num_bytes >> PAGE_SHIFT;
+
+        if !ptr::eq::<Mm>(&*self.mm, &**vma.mm()) {
+            pr_debug!("Failed to register with vma: invalid vma->vm_mm");
+            return Err(EINVAL);
+        }
+        if num_pages == 0 {
+            pr_debug!("Failed to register with vma: size zero");
+            return Err(EINVAL);
+        }
+
+        let mut pages = KVVec::<PageInfo>::with_capacity(num_pages, GFP_KERNEL)?;
+
+        // SAFETY: This just initializes the pages array.
+        unsafe {
+            let self_ptr = self as *const ShrinkablePageRange;
+            for i in 0..num_pages {
+                let info = pages.as_mut_ptr().add(i);
+                (&raw mut (*info).range).write(self_ptr);
+                (&raw mut (*info).page).write(None);
+                let lru = &raw mut (*info).lru;
+                (&raw mut (*lru).next).write(lru);
+                (&raw mut (*lru).prev).write(lru);
+            }
+        }
+
+        let mut inner = self.lock.lock();
+        if inner.size > 0 {
+            pr_debug!("Failed to register with vma: already registered");
+            drop(inner);
+            return Err(EBUSY);
+        }
+
+        inner.pages = pages.into_raw_parts().0;
+        inner.size = num_pages;
+        inner.vma_addr = vma.start();
+
+        Ok(num_pages)
+    }
+
+    /// Make sure that the given pages are allocated and mapped.
+    ///
+    /// Must not be called from an atomic context.
+    pub(crate) fn use_range(&self, start: usize, end: usize) -> Result<()> {
+        if start >= end {
+            return Ok(());
+        }
+        let mut inner = self.lock.lock();
+        assert!(end <= inner.size);
+
+        for i in start..end {
+            // SAFETY: This pointer offset is in bounds.
+            let page_info = unsafe { inner.pages.add(i) };
+
+            // SAFETY: The pointer is valid, and we hold the lock so reading from the page is okay.
+            if let Some(page) = unsafe { PageInfo::get_page(page_info) } {
+                // Since we're going to use the page, we should remove it from the lru list so that
+                // the shrinker will not free it.
+                //
+                // SAFETY: The pointer is valid, and this is the right shrinker.
+                //
+                // The shrinker can't free the page between the check and this call to
+                // `list_lru_del` because we hold the lock.
+                unsafe { PageInfo::list_lru_del(page_info, page.nid(), self.shrinker) };
+            } else {
+                // We have to allocate a new page. Use the slow path.
+                drop(inner);
+                // SAFETY: `i < end <= inner.size` so `i` is in bounds.
+                match unsafe { self.use_page_slow(i) } {
+                    Ok(()) => {}
+                    Err(err) => {
+                        pr_warn!("Error in use_page_slow: {:?}", err);
+                        return Err(err);
+                    }
+                }
+                inner = self.lock.lock();
+            }
+        }
+        Ok(())
+    }
+
+    /// Mark the given page as in use, slow path.
+    ///
+    /// Must not be called from an atomic context.
+    ///
+    /// # Safety
+    ///
+    /// Assumes that `i` is in bounds.
+    #[cold]
+    unsafe fn use_page_slow(&self, i: usize) -> Result<()> {
+        let new_page = Page::alloc_page(GFP_KERNEL | __GFP_HIGHMEM | __GFP_ZERO)?;
+
+        let mm_mutex = self.mm_lock.lock();
+        let inner = self.lock.lock();
+
+        // SAFETY: This pointer offset is in bounds.
+        let page_info = unsafe { inner.pages.add(i) };
+
+        // SAFETY: The pointer is valid, and we hold the lock so reading from the page is okay.
+        if let Some(page) = unsafe { PageInfo::get_page(page_info) } {
+            // The page was already there, or someone else added the page while we didn't hold the
+            // spinlock.
+            //
+            // SAFETY: The pointer is valid, and this is the right shrinker.
+            //
+            // The shrinker can't free the page between the check and this call to
+            // `list_lru_del` because we hold the lock.
+            unsafe { PageInfo::list_lru_del(page_info, page.nid(), self.shrinker) };
+            return Ok(());
+        }
+
+        let vma_addr = inner.vma_addr;
+        // Release the spinlock while we insert the page into the vma.
+        drop(inner);
+
+        // No overflow since we stay in bounds of the vma.
+        let user_page_addr = vma_addr + (i << PAGE_SHIFT);
+
+        // We use `mmput_async` when dropping the `mm` because `use_page_slow` is usually used from
+        // a remote process. If the call to `mmput` races with the process shutting down, then the
+        // caller of `use_page_slow` becomes responsible for cleaning up the `mm`, which doesn't
+        // happen until it returns to userspace. However, the caller might instead go to sleep and
+        // wait for the owner of the `mm` to wake it up, which doesn't happen because it's in the
+        // middle of a shutdown process that won't complete until the `mm` is dropped. This can
+        // amount to a deadlock.
+        //
+        // Using `mmput_async` avoids this, because then the `mm` cleanup is instead queued to a
+        // workqueue.
+        MmWithUser::into_mmput_async(self.mm.mmget_not_zero().ok_or(ESRCH)?)
+            .mmap_read_lock()
+            .vma_lookup(vma_addr)
+            .ok_or(ESRCH)?
+            .as_mixedmap_vma()
+            .ok_or(ESRCH)?
+            .vm_insert_page(user_page_addr, &new_page)
+            .inspect_err(|err| {
+                pr_warn!(
+                    "Failed to vm_insert_page({}): vma_addr:{} i:{} err:{:?}",
+                    user_page_addr,
+                    vma_addr,
+                    i,
+                    err
+                )
+            })?;
+
+        let inner = self.lock.lock();
+
+        // SAFETY: The `page_info` pointer is valid and currently does not have a page. The page
+        // can be written to since we hold the lock.
+        //
+        // We released and reacquired the spinlock since we checked that the page is null, but we
+        // always hold the mm_lock mutex when setting the page to a non-null value, so it's not
+        // possible for someone else to have changed it since our check.
+        unsafe { PageInfo::set_page(page_info, new_page) };
+
+        drop(inner);
+        drop(mm_mutex);
+
+        Ok(())
+    }
+
+    /// If the given page is in use, then mark it as available so that the shrinker can free it.
+    ///
+    /// May be called from an atomic context.
+    pub(crate) fn stop_using_range(&self, start: usize, end: usize) {
+        if start >= end {
+            return;
+        }
+        let inner = self.lock.lock();
+        assert!(end <= inner.size);
+
+        for i in (start..end).rev() {
+            // SAFETY: The pointer is in bounds.
+            let page_info = unsafe { inner.pages.add(i) };
+
+            // SAFETY: Okay for reading since we have the lock.
+            if let Some(page) = unsafe { PageInfo::get_page(page_info) } {
+                // SAFETY: The pointer is valid, and it's the right shrinker.
+                unsafe { PageInfo::list_lru_add(page_info, page.nid(), self.shrinker) };
+            }
+        }
+    }
+
+    /// Helper for reading or writing to a range of bytes that may overlap with several pages.
+    ///
+    /// # Safety
+    ///
+    /// All pages touched by this operation must be in use for the duration of this call.
+    unsafe fn iterate<T>(&self, mut offset: usize, mut size: usize, mut cb: T) -> Result
+    where
+        T: FnMut(&Page, usize, usize) -> Result,
+    {
+        if size == 0 {
+            return Ok(());
+        }
+
+        let (pages, num_pages) = {
+            let inner = self.lock.lock();
+            (inner.pages, inner.size)
+        };
+        let num_bytes = num_pages << PAGE_SHIFT;
+
+        // Check that the request is within the buffer.
+        if offset.checked_add(size).ok_or(EFAULT)? > num_bytes {
+            return Err(EFAULT);
+        }
+
+        let mut page_index = offset >> PAGE_SHIFT;
+        offset &= PAGE_SIZE - 1;
+        while size > 0 {
+            let available = usize::min(size, PAGE_SIZE - offset);
+            // SAFETY: The pointer is in bounds.
+            let page_info = unsafe { pages.add(page_index) };
+            // SAFETY: The caller guarantees that this page is in the "in use" state for the
+            // duration of this call to `iterate`, so nobody will change the page.
+            let page = unsafe { PageInfo::get_page(page_info) };
+            if page.is_none() {
+                pr_warn!("Page is null!");
+            }
+            let page = page.ok_or(EFAULT)?;
+            cb(page, offset, available)?;
+            size -= available;
+            page_index += 1;
+            offset = 0;
+        }
+        Ok(())
+    }
+
+    /// Copy from userspace into this page range.
+    ///
+    /// # Safety
+    ///
+    /// All pages touched by this operation must be in use for the duration of this call.
+    pub(crate) unsafe fn copy_from_user_slice(
+        &self,
+        reader: &mut UserSliceReader,
+        offset: usize,
+        size: usize,
+    ) -> Result {
+        // SAFETY: `self.iterate` has the same safety requirements as `copy_from_user_slice`.
+        unsafe {
+            self.iterate(offset, size, |page, offset, to_copy| {
+                page.copy_from_user_slice_raw(reader, offset, to_copy)
+            })
+        }
+    }
+
+    /// Copy from this page range into kernel space.
+    ///
+    /// # Safety
+    ///
+    /// All pages touched by this operation must be in use for the duration of this call.
+    pub(crate) unsafe fn read<T: FromBytes>(&self, offset: usize) -> Result<T> {
+        let mut out = MaybeUninit::<T>::uninit();
+        let mut out_offset = 0;
+        // SAFETY: `self.iterate` has the same safety requirements as `read`.
+        unsafe {
+            self.iterate(offset, size_of::<T>(), |page, offset, to_copy| {
+                // SAFETY: The sum of `offset` and `to_copy` is bounded by the size of T.
+                let obj_ptr = (out.as_mut_ptr() as *mut u8).add(out_offset);
+                // SAFETY: The pointer points is in-bounds of the `out` variable, so it is valid.
+                page.read_raw(obj_ptr, offset, to_copy)?;
+                out_offset += to_copy;
+                Ok(())
+            })?;
+        }
+        // SAFETY: We just initialised the data.
+        Ok(unsafe { out.assume_init() })
+    }
+
+    /// Copy from kernel space into this page range.
+    ///
+    /// # Safety
+    ///
+    /// All pages touched by this operation must be in use for the duration of this call.
+    pub(crate) unsafe fn write<T: ?Sized>(&self, offset: usize, obj: &T) -> Result {
+        let mut obj_offset = 0;
+        // SAFETY: `self.iterate` has the same safety requirements as `write`.
+        unsafe {
+            self.iterate(offset, size_of_val(obj), |page, offset, to_copy| {
+                // SAFETY: The sum of `offset` and `to_copy` is bounded by the size of T.
+                let obj_ptr = (obj as *const T as *const u8).add(obj_offset);
+                // SAFETY: We have a reference to the object, so the pointer is valid.
+                page.write_raw(obj_ptr, offset, to_copy)?;
+                obj_offset += to_copy;
+                Ok(())
+            })
+        }
+    }
+
+    /// Write zeroes to the given range.
+    ///
+    /// # Safety
+    ///
+    /// All pages touched by this operation must be in use for the duration of this call.
+    pub(crate) unsafe fn fill_zero(&self, offset: usize, size: usize) -> Result {
+        // SAFETY: `self.iterate` has the same safety requirements as `copy_into`.
+        unsafe {
+            self.iterate(offset, size, |page, offset, len| {
+                page.fill_zero_raw(offset, len)
+            })
+        }
+    }
+}
+
+#[pinned_drop]
+impl PinnedDrop for ShrinkablePageRange {
+    fn drop(self: Pin<&mut Self>) {
+        let (pages, size) = {
+            let lock = self.lock.lock();
+            (lock.pages, lock.size)
+        };
+
+        if size == 0 {
+            return;
+        }
+
+        // Note: This call is also necessary for the safety of `stable_trylock_mm`.
+        let mm_lock = self.mm_lock.lock();
+
+        // This is the destructor, so unlike the other methods, we only need to worry about races
+        // with the shrinker here. Since we hold the `mm_lock`, we also can't race with the
+        // shrinker, and after this loop, the shrinker will not access any of our pages since we
+        // removed them from the lru list.
+        for i in 0..size {
+            // SAFETY: Loop is in-bounds of the size.
+            let p_ptr = unsafe { pages.add(i) };
+            // SAFETY: No other readers, so we can read.
+            if let Some(p) = unsafe { PageInfo::get_page(p_ptr) } {
+                // SAFETY: The pointer is valid and it's the right shrinker.
+                unsafe { PageInfo::list_lru_del(p_ptr, p.nid(), self.shrinker) };
+            }
+        }
+
+        drop(mm_lock);
+
+        // SAFETY: `pages` was allocated as an `KVVec<PageInfo>` with capacity `size`. Furthermore,
+        // all `size` elements are initialized. Also, the array is no longer shared with the
+        // shrinker due to the above loop.
+        drop(unsafe { KVVec::from_raw_parts(pages, size, size) });
+    }
+}
+
+/// # Safety
+/// Called by the shrinker.
+#[no_mangle]
+unsafe extern "C" fn rust_shrink_count(
+    shrink: *mut bindings::shrinker,
+    _sc: *mut bindings::shrink_control,
+) -> c_ulong {
+    // SAFETY: We can access our own private data.
+    let list_lru = unsafe { (*shrink).private_data.cast::<bindings::list_lru>() };
+    // SAFETY: Accessing the lru list is okay. Just an FFI call.
+    unsafe { bindings::list_lru_count(list_lru) }
+}
+
+/// # Safety
+/// Called by the shrinker.
+#[no_mangle]
+unsafe extern "C" fn rust_shrink_scan(
+    shrink: *mut bindings::shrinker,
+    sc: *mut bindings::shrink_control,
+) -> c_ulong {
+    // SAFETY: We can access our own private data.
+    let list_lru = unsafe { (*shrink).private_data.cast::<bindings::list_lru>() };
+    // SAFETY: Caller guarantees that it is safe to read this field.
+    let nr_to_scan = unsafe { (*sc).nr_to_scan };
+    // SAFETY: Accessing the lru list is okay. Just an FFI call.
+    unsafe {
+        bindings::list_lru_walk(
+            list_lru,
+            Some(bindings::rust_shrink_free_page_wrap),
+            ptr::null_mut(),
+            nr_to_scan,
+        )
+    }
+}
+
+const LRU_SKIP: bindings::lru_status = bindings::lru_status_LRU_SKIP;
+const LRU_REMOVED_ENTRY: bindings::lru_status = bindings::lru_status_LRU_REMOVED_RETRY;
+
+/// # Safety
+/// Called by the shrinker.
+#[no_mangle]
+unsafe extern "C" fn rust_shrink_free_page(
+    item: *mut bindings::list_head,
+    lru: *mut bindings::list_lru_one,
+    _cb_arg: *mut c_void,
+) -> bindings::lru_status {
+    // Fields that should survive after unlocking the lru lock.
+    let page;
+    let page_index;
+    let mm;
+    let mmap_read;
+    let mm_mutex;
+    let vma_addr;
+
+    {
+        // CAST: The `list_head` field is first in `PageInfo`.
+        let info = item as *mut PageInfo;
+        // SAFETY: The `range` field of `PageInfo` is immutable.
+        let range = unsafe { &*((*info).range) };
+
+        mm = match range.mm.mmget_not_zero() {
+            Some(mm) => MmWithUser::into_mmput_async(mm),
+            None => return LRU_SKIP,
+        };
+
+        mm_mutex = match range.stable_trylock_mm() {
+            Some(guard) => guard,
+            None => return LRU_SKIP,
+        };
+
+        mmap_read = match mm.mmap_read_trylock() {
+            Some(guard) => guard,
+            None => return LRU_SKIP,
+        };
+
+        // We can't lock it normally here, since we hold the lru lock.
+        let inner = match range.lock.try_lock() {
+            Some(inner) => inner,
+            None => return LRU_SKIP,
+        };
+
+        // SAFETY: The item is in this lru list, so it's okay to remove it.
+        unsafe { bindings::list_lru_isolate(lru, item) };
+
+        // SAFETY: Both pointers are in bounds of the same allocation.
+        page_index = unsafe { info.offset_from(inner.pages) } as usize;
+
+        // SAFETY: We hold the spinlock, so we can take the page.
+        //
+        // This sets the page pointer to zero before we unmap it from the vma. However, we call
+        // `zap_page_range` before we release the mmap lock, so `use_page_slow` will not be able to
+        // insert a new page until after our call to `zap_page_range`.
+        page = unsafe { PageInfo::take_page(info) };
+        vma_addr = inner.vma_addr;
+
+        // From this point on, we don't access this PageInfo or ShrinkablePageRange again, because
+        // they can be freed at any point after we unlock `lru_lock`. This is with the exception of
+        // `mm_mutex` which is kept alive by holding the lock.
+    }
+
+    // SAFETY: The lru lock is locked when this method is called.
+    unsafe { bindings::spin_unlock(&raw mut (*lru).lock) };
+
+    if let Some(vma) = mmap_read.vma_lookup(vma_addr) {
+        let user_page_addr = vma_addr + (page_index << PAGE_SHIFT);
+        vma.zap_page_range_single(user_page_addr, PAGE_SIZE);
+    }
+
+    drop(mmap_read);
+    drop(mm_mutex);
+    drop(mm);
+    drop(page);
+
+    // SAFETY: We just unlocked the lru lock, but it should be locked when we return.
+    unsafe { bindings::spin_lock(&raw mut (*lru).lock) };
+
+    LRU_REMOVED_ENTRY
+}
diff --git a/drivers/android/binder/page_range_helper.c b/drivers/android/binder/page_range_helper.c
new file mode 100644
index 000000000000..496887723ee0
--- /dev/null
+++ b/drivers/android/binder/page_range_helper.c
@@ -0,0 +1,24 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/* C helper for page_range.rs to work around a CFI violation.
+ *
+ * Bindgen currently pretends that `enum lru_status` is the same as an integer.
+ * This assumption is fine ABI-wise, but once you add CFI to the mix, it
+ * triggers a CFI violation because `enum lru_status` gets a different CFI tag.
+ *
+ * This file contains a workaround until bindgen can be fixed.
+ *
+ * Copyright (C) 2025 Google LLC.
+ */
+#include "page_range_helper.h"
+
+unsigned int rust_shrink_free_page(struct list_head *item,
+				   struct list_lru_one *list,
+				   void *cb_arg);
+
+enum lru_status
+rust_shrink_free_page_wrap(struct list_head *item, struct list_lru_one *list,
+			   void *cb_arg)
+{
+	return rust_shrink_free_page(item, list, cb_arg);
+}
diff --git a/drivers/android/binder/page_range_helper.h b/drivers/android/binder/page_range_helper.h
new file mode 100644
index 000000000000..18dd2dd117b2
--- /dev/null
+++ b/drivers/android/binder/page_range_helper.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2025 Google, Inc.
+ */
+
+#ifndef _LINUX_PAGE_RANGE_HELPER_H
+#define _LINUX_PAGE_RANGE_HELPER_H
+
+#include <linux/list_lru.h>
+
+enum lru_status
+rust_shrink_free_page_wrap(struct list_head *item, struct list_lru_one *list,
+			   void *cb_arg);
+
+#endif /* _LINUX_PAGE_RANGE_HELPER_H */
diff --git a/drivers/android/binder/process.rs b/drivers/android/binder/process.rs
new file mode 100644
index 000000000000..132055b4790f
--- /dev/null
+++ b/drivers/android/binder/process.rs
@@ -0,0 +1,1745 @@
+// SPDX-License-Identifier: GPL-2.0
+
+// Copyright (C) 2025 Google LLC.
+
+//! This module defines the `Process` type, which represents a process using a particular binder
+//! context.
+//!
+//! The `Process` object keeps track of all of the resources that this process owns in the binder
+//! context.
+//!
+//! There is one `Process` object for each binder fd that a process has opened, so processes using
+//! several binder contexts have several `Process` objects. This ensures that the contexts are
+//! fully separated.
+
+use core::mem::take;
+
+use kernel::{
+    bindings,
+    cred::Credential,
+    error::Error,
+    fs::file::{self, File},
+    id_pool::IdPool,
+    list::{List, ListArc, ListArcField, ListLinks},
+    mm,
+    prelude::*,
+    rbtree::{self, RBTree, RBTreeNode, RBTreeNodeReservation},
+    seq_file::SeqFile,
+    seq_print,
+    sync::poll::PollTable,
+    sync::{
+        lock::{spinlock::SpinLockBackend, Guard},
+        Arc, ArcBorrow, CondVar, CondVarTimeoutResult, Mutex, SpinLock, UniqueArc,
+    },
+    task::Task,
+    types::ARef,
+    uaccess::{UserSlice, UserSliceReader},
+    uapi,
+    workqueue::{self, Work},
+};
+
+use crate::{
+    allocation::{Allocation, AllocationInfo, NewAllocation},
+    context::Context,
+    defs::*,
+    error::{BinderError, BinderResult},
+    node::{CouldNotDeliverCriticalIncrement, CritIncrWrapper, Node, NodeDeath, NodeRef},
+    page_range::ShrinkablePageRange,
+    range_alloc::{RangeAllocator, ReserveNew, ReserveNewArgs},
+    stats::BinderStats,
+    thread::{PushWorkRes, Thread},
+    BinderfsProcFile, DArc, DLArc, DTRWrap, DeliverToRead,
+};
+
+#[path = "freeze.rs"]
+mod freeze;
+use self::freeze::{FreezeCookie, FreezeListener};
+
+struct Mapping {
+    address: usize,
+    alloc: RangeAllocator<AllocationInfo>,
+}
+
+impl Mapping {
+    fn new(address: usize, size: usize) -> Self {
+        Self {
+            address,
+            alloc: RangeAllocator::new(size),
+        }
+    }
+}
+
+// bitflags for defer_work.
+const PROC_DEFER_FLUSH: u8 = 1;
+const PROC_DEFER_RELEASE: u8 = 2;
+
+#[derive(Copy, Clone)]
+pub(crate) enum IsFrozen {
+    Yes,
+    No,
+    InProgress,
+}
+
+impl IsFrozen {
+    /// Whether incoming transactions should be rejected due to freeze.
+    pub(crate) fn is_frozen(self) -> bool {
+        match self {
+            IsFrozen::Yes => true,
+            IsFrozen::No => false,
+            IsFrozen::InProgress => true,
+        }
+    }
+
+    /// Whether freeze notifications consider this process frozen.
+    pub(crate) fn is_fully_frozen(self) -> bool {
+        match self {
+            IsFrozen::Yes => true,
+            IsFrozen::No => false,
+            IsFrozen::InProgress => false,
+        }
+    }
+}
+
+/// The fields of `Process` protected by the spinlock.
+pub(crate) struct ProcessInner {
+    is_manager: bool,
+    pub(crate) is_dead: bool,
+    threads: RBTree<i32, Arc<Thread>>,
+    /// INVARIANT: Threads pushed to this list must be owned by this process.
+    ready_threads: List<Thread>,
+    nodes: RBTree<u64, DArc<Node>>,
+    mapping: Option<Mapping>,
+    work: List<DTRWrap<dyn DeliverToRead>>,
+    delivered_deaths: List<DTRWrap<NodeDeath>, 2>,
+
+    /// The number of requested threads that haven't registered yet.
+    requested_thread_count: u32,
+    /// The maximum number of threads used by the process thread pool.
+    max_threads: u32,
+    /// The number of threads the started and registered with the thread pool.
+    started_thread_count: u32,
+
+    /// Bitmap of deferred work to do.
+    defer_work: u8,
+
+    /// Number of transactions to be transmitted before processes in freeze_wait
+    /// are woken up.
+    outstanding_txns: u32,
+    /// Process is frozen and unable to service binder transactions.
+    pub(crate) is_frozen: IsFrozen,
+    /// Process received sync transactions since last frozen.
+    pub(crate) sync_recv: bool,
+    /// Process received async transactions since last frozen.
+    pub(crate) async_recv: bool,
+    pub(crate) binderfs_file: Option<BinderfsProcFile>,
+    /// Check for oneway spam
+    oneway_spam_detection_enabled: bool,
+}
+
+impl ProcessInner {
+    fn new() -> Self {
+        Self {
+            is_manager: false,
+            is_dead: false,
+            threads: RBTree::new(),
+            ready_threads: List::new(),
+            mapping: None,
+            nodes: RBTree::new(),
+            work: List::new(),
+            delivered_deaths: List::new(),
+            requested_thread_count: 0,
+            max_threads: 0,
+            started_thread_count: 0,
+            defer_work: 0,
+            outstanding_txns: 0,
+            is_frozen: IsFrozen::No,
+            sync_recv: false,
+            async_recv: false,
+            binderfs_file: None,
+            oneway_spam_detection_enabled: false,
+        }
+    }
+
+    /// Schedule the work item for execution on this process.
+    ///
+    /// If any threads are ready for work, then the work item is given directly to that thread and
+    /// it is woken up. Otherwise, it is pushed to the process work list.
+    ///
+    /// This call can fail only if the process is dead. In this case, the work item is returned to
+    /// the caller so that the caller can drop it after releasing the inner process lock. This is
+    /// necessary since the destructor of `Transaction` will take locks that can't necessarily be
+    /// taken while holding the inner process lock.
+    pub(crate) fn push_work(
+        &mut self,
+        work: DLArc<dyn DeliverToRead>,
+    ) -> Result<(), (BinderError, DLArc<dyn DeliverToRead>)> {
+        // Try to find a ready thread to which to push the work.
+        if let Some(thread) = self.ready_threads.pop_front() {
+            // Push to thread while holding state lock. This prevents the thread from giving up
+            // (for example, because of a signal) when we're about to deliver work.
+            match thread.push_work(work) {
+                PushWorkRes::Ok => Ok(()),
+                PushWorkRes::FailedDead(work) => Err((BinderError::new_dead(), work)),
+            }
+        } else if self.is_dead {
+            Err((BinderError::new_dead(), work))
+        } else {
+            let sync = work.should_sync_wakeup();
+
+            // Didn't find a thread waiting for proc work; this can happen
+            // in two scenarios:
+            // 1. All threads are busy handling transactions
+            //    In that case, one of those threads should call back into
+            //    the kernel driver soon and pick up this work.
+            // 2. Threads are using the (e)poll interface, in which case
+            //    they may be blocked on the waitqueue without having been
+            //    added to waiting_threads. For this case, we just iterate
+            //    over all threads not handling transaction work, and
+            //    wake them all up. We wake all because we don't know whether
+            //    a thread that called into (e)poll is handling non-binder
+            //    work currently.
+            self.work.push_back(work);
+
+            // Wake up polling threads, if any.
+            for thread in self.threads.values() {
+                thread.notify_if_poll_ready(sync);
+            }
+
+            Ok(())
+        }
+    }
+
+    pub(crate) fn remove_node(&mut self, ptr: u64) {
+        self.nodes.remove(&ptr);
+    }
+
+    /// Updates the reference count on the given node.
+    pub(crate) fn update_node_refcount(
+        &mut self,
+        node: &DArc<Node>,
+        inc: bool,
+        strong: bool,
+        count: usize,
+        othread: Option<&Thread>,
+    ) {
+        let push = node.update_refcount_locked(inc, strong, count, self);
+
+        // If we decided that we need to push work, push either to the process or to a thread if
+        // one is specified.