summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu/si_dma.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/si_dma.c')
-rw-r--r--drivers/gpu/drm/amd/amdgpu/si_dma.c122
1 files changed, 69 insertions, 53 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/si_dma.c b/drivers/gpu/drm/amd/amdgpu/si_dma.c
index dbd78d5345a4..7f18e4875287 100644
--- a/drivers/gpu/drm/amd/amdgpu/si_dma.c
+++ b/drivers/gpu/drm/amd/amdgpu/si_dma.c
@@ -27,6 +27,8 @@
#include "si.h"
#include "sid.h"
+#include "oss/oss_1_0_d.h"
+#include "oss/oss_1_0_sh_mask.h"
const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
{
DMA0_REGISTER_OFFSET,
@@ -38,17 +40,31 @@ static void si_dma_set_buffer_funcs(struct amdgpu_device *adev);
static void si_dma_set_vm_pte_funcs(struct amdgpu_device *adev);
static void si_dma_set_irq_funcs(struct amdgpu_device *adev);
+/**
+ * si_dma_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (SI).
+ */
static uint64_t si_dma_ring_get_rptr(struct amdgpu_ring *ring)
{
return *ring->rptr_cpu_addr;
}
+/**
+ * si_dma_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (SI).
+ */
static uint64_t si_dma_ring_get_wptr(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
u32 me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
- return (RREG32(DMA_RB_WPTR + sdma_offsets[me]) & 0x3fffc) >> 2;
+ return (RREG32(mmDMA_GFX_RB_WPTR + sdma_offsets[me]) & 0x3fffc) >> 2;
}
static void si_dma_ring_set_wptr(struct amdgpu_ring *ring)
@@ -56,7 +72,7 @@ static void si_dma_ring_set_wptr(struct amdgpu_ring *ring)
struct amdgpu_device *adev = ring->adev;
u32 me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
- WREG32(DMA_RB_WPTR + sdma_offsets[me], (ring->wptr << 2) & 0x3fffc);
+ WREG32(mmDMA_GFX_RB_WPTR + sdma_offsets[me], (ring->wptr << 2) & 0x3fffc);
}
static void si_dma_ring_emit_ib(struct amdgpu_ring *ring,
@@ -117,9 +133,9 @@ static void si_dma_stop(struct amdgpu_device *adev)
for (i = 0; i < adev->sdma.num_instances; i++) {
/* dma0 */
- rb_cntl = RREG32(DMA_RB_CNTL + sdma_offsets[i]);
- rb_cntl &= ~DMA_RB_ENABLE;
- WREG32(DMA_RB_CNTL + sdma_offsets[i], rb_cntl);
+ rb_cntl = RREG32(mmDMA_GFX_RB_CNTL + sdma_offsets[i]);
+ rb_cntl &= ~DMA_GFX_RB_CNTL__RB_ENABLE_MASK;
+ WREG32(mmDMA_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
}
}
@@ -133,44 +149,44 @@ static int si_dma_start(struct amdgpu_device *adev)
for (i = 0; i < adev->sdma.num_instances; i++) {
ring = &adev->sdma.instance[i].ring;
- WREG32(DMA_SEM_INCOMPLETE_TIMER_CNTL + sdma_offsets[i], 0);
- WREG32(DMA_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
+ WREG32(mmDMA_SEM_INCOMPLETE_TIMER_CNTL + sdma_offsets[i], 0);
+ WREG32(mmDMA_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
/* Set ring buffer size in dwords */
rb_bufsz = order_base_2(ring->ring_size / 4);
rb_cntl = rb_bufsz << 1;
#ifdef __BIG_ENDIAN
- rb_cntl |= DMA_RB_SWAP_ENABLE | DMA_RPTR_WRITEBACK_SWAP_ENABLE;
+ rb_cntl |= DMA_GFX_RB_CNTL__RB_SWAP_ENABLE_MASK | DMA_GFX_RB_CNTL__RPTR_WRITEBACK_SWAP_ENABLE_MASK;
#endif
- WREG32(DMA_RB_CNTL + sdma_offsets[i], rb_cntl);
+ WREG32(mmDMA_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
/* Initialize the ring buffer's read and write pointers */
- WREG32(DMA_RB_RPTR + sdma_offsets[i], 0);
- WREG32(DMA_RB_WPTR + sdma_offsets[i], 0);
+ WREG32(mmDMA_GFX_RB_RPTR + sdma_offsets[i], 0);
+ WREG32(mmDMA_GFX_RB_WPTR + sdma_offsets[i], 0);
rptr_addr = ring->rptr_gpu_addr;
- WREG32(DMA_RB_RPTR_ADDR_LO + sdma_offsets[i], lower_32_bits(rptr_addr));
- WREG32(DMA_RB_RPTR_ADDR_HI + sdma_offsets[i], upper_32_bits(rptr_addr) & 0xFF);
+ WREG32(mmDMA_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i], lower_32_bits(rptr_addr));
+ WREG32(mmDMA_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i], upper_32_bits(rptr_addr) & 0xFF);
- rb_cntl |= DMA_RPTR_WRITEBACK_ENABLE;
+ rb_cntl |= DMA_GFX_RB_CNTL__RPTR_WRITEBACK_ENABLE_MASK;
- WREG32(DMA_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
+ WREG32(mmDMA_GFX_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
/* enable DMA IBs */
- ib_cntl = DMA_IB_ENABLE | CMD_VMID_FORCE;
+ ib_cntl = DMA_GFX_IB_CNTL__IB_ENABLE_MASK | DMA_GFX_IB_CNTL__CMD_VMID_FORCE_MASK;
#ifdef __BIG_ENDIAN
- ib_cntl |= DMA_IB_SWAP_ENABLE;
+ ib_cntl |= DMA_GFX_IB_CNTL__IB_SWAP_ENABLE_MASK;
#endif
- WREG32(DMA_IB_CNTL + sdma_offsets[i], ib_cntl);
+ WREG32(mmDMA_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
- dma_cntl = RREG32(DMA_CNTL + sdma_offsets[i]);
- dma_cntl &= ~CTXEMPTY_INT_ENABLE;
- WREG32(DMA_CNTL + sdma_offsets[i], dma_cntl);
+ dma_cntl = RREG32(mmDMA_CNTL + sdma_offsets[i]);
+ dma_cntl &= ~DMA_CNTL__CTXEMPTY_INT_ENABLE_MASK;
+ WREG32(mmDMA_CNTL + sdma_offsets[i], dma_cntl);
ring->wptr = 0;
- WREG32(DMA_RB_WPTR + sdma_offsets[i], ring->wptr << 2);
- WREG32(DMA_RB_CNTL + sdma_offsets[i], rb_cntl | DMA_RB_ENABLE);
+ WREG32(mmDMA_GFX_RB_WPTR + sdma_offsets[i], ring->wptr << 2);
+ WREG32(mmDMA_GFX_RB_CNTL + sdma_offsets[i], rb_cntl | DMA_GFX_RB_CNTL__RB_ENABLE_MASK);
r = amdgpu_ring_test_helper(ring);
if (r)
@@ -461,7 +477,7 @@ static int si_dma_early_init(struct amdgpu_ip_block *ip_block)
{
struct amdgpu_device *adev = ip_block->adev;
- adev->sdma.num_instances = 2;
+ adev->sdma.num_instances = SDMA_MAX_INSTANCE;
si_dma_set_ring_funcs(adev);
si_dma_set_buffer_funcs(adev);
@@ -541,13 +557,13 @@ static int si_dma_resume(struct amdgpu_ip_block *ip_block)
return si_dma_hw_init(ip_block);
}
-static bool si_dma_is_idle(void *handle)
+static bool si_dma_is_idle(struct amdgpu_ip_block *ip_block)
{
- struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ struct amdgpu_device *adev = ip_block->adev;
- u32 tmp = RREG32(SRBM_STATUS2);
+ u32 tmp = RREG32(mmSRBM_STATUS2);
- if (tmp & (DMA_BUSY_MASK | DMA1_BUSY_MASK))
+ if (tmp & (SRBM_STATUS2__DMA_BUSY_MASK | SRBM_STATUS2__DMA1_BUSY_MASK))
return false;
return true;
@@ -559,7 +575,7 @@ static int si_dma_wait_for_idle(struct amdgpu_ip_block *ip_block)
struct amdgpu_device *adev = ip_block->adev;
for (i = 0; i < adev->usec_timeout; i++) {
- if (si_dma_is_idle(adev))
+ if (si_dma_is_idle(ip_block))
return 0;
udelay(1);
}
@@ -583,14 +599,14 @@ static int si_dma_set_trap_irq_state(struct amdgpu_device *adev,
case AMDGPU_SDMA_IRQ_INSTANCE0:
switch (state) {
case AMDGPU_IRQ_STATE_DISABLE:
- sdma_cntl = RREG32(DMA_CNTL + DMA0_REGISTER_OFFSET);
- sdma_cntl &= ~TRAP_ENABLE;
- WREG32(DMA_CNTL + DMA0_REGISTER_OFFSET, sdma_cntl);
+ sdma_cntl = RREG32(mmDMA_CNTL + DMA0_REGISTER_OFFSET);
+ sdma_cntl &= ~DMA_CNTL__TRAP_ENABLE_MASK;
+ WREG32(mmDMA_CNTL + DMA0_REGISTER_OFFSET, sdma_cntl);
break;
case AMDGPU_IRQ_STATE_ENABLE:
- sdma_cntl = RREG32(DMA_CNTL + DMA0_REGISTER_OFFSET);
- sdma_cntl |= TRAP_ENABLE;
- WREG32(DMA_CNTL + DMA0_REGISTER_OFFSET, sdma_cntl);
+ sdma_cntl = RREG32(mmDMA_CNTL + DMA0_REGISTER_OFFSET);
+ sdma_cntl |= DMA_CNTL__TRAP_ENABLE_MASK;
+ WREG32(mmDMA_CNTL + DMA0_REGISTER_OFFSET, sdma_cntl);
break;
default:
break;
@@ -599,14 +615,14 @@ static int si_dma_set_trap_irq_state(struct amdgpu_device *adev,
case AMDGPU_SDMA_IRQ_INSTANCE1:
switch (state) {
case AMDGPU_IRQ_STATE_DISABLE:
- sdma_cntl = RREG32(DMA_CNTL + DMA1_REGISTER_OFFSET);
- sdma_cntl &= ~TRAP_ENABLE;
- WREG32(DMA_CNTL + DMA1_REGISTER_OFFSET, sdma_cntl);
+ sdma_cntl = RREG32(mmDMA_CNTL + DMA1_REGISTER_OFFSET);
+ sdma_cntl &= ~DMA_CNTL__TRAP_ENABLE_MASK;
+ WREG32(mmDMA_CNTL + DMA1_REGISTER_OFFSET, sdma_cntl);
break;
case AMDGPU_IRQ_STATE_ENABLE:
- sdma_cntl = RREG32(DMA_CNTL + DMA1_REGISTER_OFFSET);
- sdma_cntl |= TRAP_ENABLE;
- WREG32(DMA_CNTL + DMA1_REGISTER_OFFSET, sdma_cntl);
+ sdma_cntl = RREG32(mmDMA_CNTL + DMA1_REGISTER_OFFSET);
+ sdma_cntl |= DMA_CNTL__TRAP_ENABLE_MASK;
+ WREG32(mmDMA_CNTL + DMA1_REGISTER_OFFSET, sdma_cntl);
break;
default:
break;
@@ -645,11 +661,11 @@ static int si_dma_set_clockgating_state(struct amdgpu_ip_block *ip_block,
offset = DMA0_REGISTER_OFFSET;
else
offset = DMA1_REGISTER_OFFSET;
- orig = data = RREG32(DMA_POWER_CNTL + offset);
- data &= ~MEM_POWER_OVERRIDE;
+ orig = data = RREG32(mmDMA_POWER_CNTL + offset);
+ data &= ~DMA_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
if (data != orig)
- WREG32(DMA_POWER_CNTL + offset, data);
- WREG32(DMA_CLK_CTRL + offset, 0x00000100);
+ WREG32(mmDMA_POWER_CNTL + offset, data);
+ WREG32(mmDMA_CLK_CTRL + offset, 0x00000100);
}
} else {
for (i = 0; i < adev->sdma.num_instances; i++) {
@@ -657,15 +673,15 @@ static int si_dma_set_clockgating_state(struct amdgpu_ip_block *ip_block,
offset = DMA0_REGISTER_OFFSET;
else
offset = DMA1_REGISTER_OFFSET;
- orig = data = RREG32(DMA_POWER_CNTL + offset);
- data |= MEM_POWER_OVERRIDE;
+ orig = data = RREG32(mmDMA_POWER_CNTL + offset);
+ data |= DMA_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
if (data != orig)
- WREG32(DMA_POWER_CNTL + offset, data);
+ WREG32(mmDMA_POWER_CNTL + offset, data);
- orig = data = RREG32(DMA_CLK_CTRL + offset);
+ orig = data = RREG32(mmDMA_CLK_CTRL + offset);
data = 0xff000000;
if (data != orig)
- WREG32(DMA_CLK_CTRL + offset, data);
+ WREG32(mmDMA_CLK_CTRL + offset, data);
}
}
@@ -679,11 +695,11 @@ static int si_dma_set_powergating_state(struct amdgpu_ip_block *ip_block,
struct amdgpu_device *adev = ip_block->adev;
- WREG32(DMA_PGFSM_WRITE, 0x00002000);
- WREG32(DMA_PGFSM_CONFIG, 0x100010ff);
+ WREG32(mmDMA_PGFSM_WRITE, 0x00002000);
+ WREG32(mmDMA_PGFSM_CONFIG, 0x100010ff);
for (tmp = 0; tmp < 5; tmp++)
- WREG32(DMA_PGFSM_WRITE, 0);
+ WREG32(mmDMA_PGFSM_WRITE, 0);
return 0;
}
generated by cgit (git 2.34.1) at 2025-05-29 02:12:18 +0000