iv/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

drm/amdgpu: replace tmz flag into buffer flag

Browse Source 
Replace tmz flag into buffer flag to make it easier to understand
and extend

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Frank Min <Frank.Min@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Frank Min 2024-04-10 21:13:25 +08:00 committed by Alex Deucher
parent 92ed1e9cd5
commit ea9238a81b
15 changed files with 42 additions and 36 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
View File

@ -39,7 +39,7 @@ static int amdgpu_benchmark_do_move(struct amdgpu_device *adev, unsigned size,
for (i = 0; i < n; i++) {
struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
r = amdgpu_copy_buffer(ring, saddr, daddr, size, NULL, &fence,
false, false, false);
false, false, 0);
if (r)
goto exit_do_move;
r = dma_fence_wait(fence, false);

2
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
View File

@ -765,7 +765,7 @@ int amdgpu_bo_restore_shadow(struct amdgpu_bo *shadow, struct dma_fence **fence)
return amdgpu_copy_buffer(ring, shadow_addr, parent_addr,
amdgpu_bo_size(shadow), NULL, fence,
true, false, false);
true, false, 0);
}
/**

2
drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h
View File

@ -132,7 +132,7 @@ struct amdgpu_buffer_funcs {
uint64_t dst_offset,
/* number of byte to transfer */
uint32_t byte_count,
bool tmz);
uint32_t copy_flags);
/* maximum bytes in a single operation */
uint32_t fill_max_bytes;

18
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
View File

@ -236,7 +236,7 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
dst_addr = amdgpu_bo_gpu_offset(adev->gart.bo);
dst_addr += window * AMDGPU_GTT_MAX_TRANSFER_SIZE * 8;
amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr,
dst_addr, num_bytes, false);
dst_addr, num_bytes, 0);
amdgpu_ring_pad_ib(ring, &job->ibs[0]);
WARN_ON(job->ibs[0].length_dw > num_dw);
@ -296,6 +296,8 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
struct dma_fence *fence = NULL;
int r = 0;
uint32_t copy_flags = 0;
if (!adev->mman.buffer_funcs_enabled) {
DRM_ERROR("Trying to move memory with ring turned off.\n");
return -EINVAL;
@ -323,8 +325,11 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
if (r)
goto error;
r = amdgpu_copy_buffer(ring, from, to, cur_size,
resv, &next, false, true, tmz);
if (tmz)
copy_flags |= AMDGPU_COPY_FLAGS_TMZ;
r = amdgpu_copy_buffer(ring, from, to, cur_size, resv,
&next, false, true, copy_flags);
if (r)
goto error;
@ -1488,7 +1493,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
swap(src_addr, dst_addr);
amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr, dst_addr,
PAGE_SIZE, false);
PAGE_SIZE, 0);
amdgpu_ring_pad_ib(adev->mman.buffer_funcs_ring, &job->ibs[0]);
WARN_ON(job->ibs[0].length_dw > num_dw);
@ -2139,7 +2144,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
uint64_t dst_offset, uint32_t byte_count,
struct dma_resv *resv,
struct dma_fence **fence, bool direct_submit,
bool vm_needs_flush, bool tmz)
bool vm_needs_flush, uint32_t copy_flags)
{
struct amdgpu_device *adev = ring->adev;
unsigned int num_loops, num_dw;
@ -2165,8 +2170,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
uint32_t cur_size_in_bytes = min(byte_count, max_bytes);
amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_offset,
dst_offset, cur_size_in_bytes, tmz);
dst_offset, cur_size_in_bytes, copy_flags);
src_offset += cur_size_in_bytes;
dst_offset += cur_size_in_bytes;
byte_count -= cur_size_in_bytes;

4
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
View File

@ -111,6 +111,8 @@ struct amdgpu_copy_mem {
unsigned long offset;
};
#define AMDGPU_COPY_FLAGS_TMZ (1 << 0)
int amdgpu_gtt_mgr_init(struct amdgpu_device *adev, uint64_t gtt_size);
void amdgpu_gtt_mgr_fini(struct amdgpu_device *adev);
int amdgpu_preempt_mgr_init(struct amdgpu_device *adev);
@ -151,7 +153,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
uint64_t dst_offset, uint32_t byte_count,
struct dma_resv *resv,
struct dma_fence **fence, bool direct_submit,
bool vm_needs_flush, bool tmz);
bool vm_needs_flush, uint32_t copy_flags);
int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
const struct amdgpu_copy_mem *src,
const struct amdgpu_copy_mem *dst,

4
drivers/gpu/drm/amd/amdgpu/cik_sdma.c
View File

@ -1290,7 +1290,7 @@ static void cik_sdma_set_irq_funcs(struct amdgpu_device *adev)
* @src_offset: src GPU address
* @dst_offset: dst GPU address
* @byte_count: number of bytes to xfer
* @tmz: is this a secure operation
* @copy_flags: unused
*
* Copy GPU buffers using the DMA engine (CIK).
* Used by the amdgpu ttm implementation to move pages if
@ -1300,7 +1300,7 @@ static void cik_sdma_emit_copy_buffer(struct amdgpu_ib *ib,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
bool tmz)
uint32_t copy_flags)
{
ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY, SDMA_COPY_SUB_OPCODE_LINEAR, 0);
ib->ptr[ib->length_dw++] = byte_count;

4
drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
View File

@ -1176,7 +1176,7 @@ static void sdma_v2_4_set_irq_funcs(struct amdgpu_device *adev)
* @src_offset: src GPU address
* @dst_offset: dst GPU address
* @byte_count: number of bytes to xfer
* @tmz: unused
* @copy_flags: unused
*
* Copy GPU buffers using the DMA engine (VI).
* Used by the amdgpu ttm implementation to move pages if
@ -1186,7 +1186,7 @@ static void sdma_v2_4_emit_copy_buffer(struct amdgpu_ib *ib,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
bool tmz)
uint32_t copy_flags)
{
ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);

4
drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
View File

@ -1616,7 +1616,7 @@ static void sdma_v3_0_set_irq_funcs(struct amdgpu_device *adev)
* @src_offset: src GPU address
* @dst_offset: dst GPU address
* @byte_count: number of bytes to xfer
* @tmz: unused
* @copy_flags: unused
*
* Copy GPU buffers using the DMA engine (VI).
* Used by the amdgpu ttm implementation to move pages if
@ -1626,7 +1626,7 @@ static void sdma_v3_0_emit_copy_buffer(struct amdgpu_ib *ib,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
bool tmz)
uint32_t copy_flags)
{
ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);

6
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
View File

@ -2448,7 +2448,7 @@ static void sdma_v4_0_set_irq_funcs(struct amdgpu_device *adev)
* @src_offset: src GPU address
* @dst_offset: dst GPU address
* @byte_count: number of bytes to xfer
* @tmz: if a secure copy should be used
* @copy_flags: copy flags for the buffers
*
* Copy GPU buffers using the DMA engine (VEGA10/12).
* Used by the amdgpu ttm implementation to move pages if
@ -2458,11 +2458,11 @@ static void sdma_v4_0_emit_copy_buffer(struct amdgpu_ib *ib,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
bool tmz)
uint32_t copy_flags)
{
ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0);
SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
ib->ptr[ib->length_dw++] = byte_count - 1;
ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);

6
drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
View File

@ -1945,7 +1945,7 @@ static void sdma_v4_4_2_set_irq_funcs(struct amdgpu_device *adev)
* @src_offset: src GPU address
* @dst_offset: dst GPU address
* @byte_count: number of bytes to xfer
* @tmz: if a secure copy should be used
* @copy_flags: copy flags for the buffers
*
* Copy GPU buffers using the DMA engine.
* Used by the amdgpu ttm implementation to move pages if
@ -1955,11 +1955,11 @@ static void sdma_v4_4_2_emit_copy_buffer(struct amdgpu_ib *ib,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
bool tmz)
uint32_t copy_flags)
{
ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0);
SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
ib->ptr[ib->length_dw++] = byte_count - 1;
ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);

6
drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
View File

@ -1805,7 +1805,7 @@ static void sdma_v5_0_set_irq_funcs(struct amdgpu_device *adev)
* @src_offset: src GPU address
* @dst_offset: dst GPU address
* @byte_count: number of bytes to xfer
* @tmz: if a secure copy should be used
* @copy_flags: copy flags for the buffers
*
* Copy GPU buffers using the DMA engine (NAVI10).
* Used by the amdgpu ttm implementation to move pages if
@ -1815,11 +1815,11 @@ static void sdma_v5_0_emit_copy_buffer(struct amdgpu_ib *ib,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
bool tmz)
uint32_t copy_flags)
{
ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0);
SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
ib->ptr[ib->length_dw++] = byte_count - 1;
ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);

6
drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
View File

@ -1747,7 +1747,7 @@ static void sdma_v5_2_set_irq_funcs(struct amdgpu_device *adev)
* @src_offset: src GPU address
* @dst_offset: dst GPU address
* @byte_count: number of bytes to xfer
* @tmz: if a secure copy should be used
* @copy_flags: copy flags for the buffers
*
* Copy GPU buffers using the DMA engine.
* Used by the amdgpu ttm implementation to move pages if
@ -1757,11 +1757,11 @@ static void sdma_v5_2_emit_copy_buffer(struct amdgpu_ib *ib,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
bool tmz)
uint32_t copy_flags)
{
ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0);
SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
ib->ptr[ib->length_dw++] = byte_count - 1;
ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);

6
drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
View File

@ -1574,7 +1574,7 @@ static void sdma_v6_0_set_irq_funcs(struct amdgpu_device *adev)
* @src_offset: src GPU address
* @dst_offset: dst GPU address
* @byte_count: number of bytes to xfer
* @tmz: if a secure copy should be used
* @copy_flags: copy flags for the buffers
*
* Copy GPU buffers using the DMA engine.
* Used by the amdgpu ttm implementation to move pages if
@ -1584,11 +1584,11 @@ static void sdma_v6_0_emit_copy_buffer(struct amdgpu_ib *ib,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
bool tmz)
uint32_t copy_flags)
{
ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_COPY) |
SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) |
SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0);
SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
ib->ptr[ib->length_dw++] = byte_count - 1;
ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);

4
drivers/gpu/drm/amd/amdgpu/si_dma.c
View File

@ -761,7 +761,7 @@ static void si_dma_set_irq_funcs(struct amdgpu_device *adev)
* @src_offset: src GPU address
* @dst_offset: dst GPU address
* @byte_count: number of bytes to xfer
* @tmz: is this a secure operation
* @copy_flags: unused
*
* Copy GPU buffers using the DMA engine (VI).
* Used by the amdgpu ttm implementation to move pages if
@ -771,7 +771,7 @@ static void si_dma_emit_copy_buffer(struct amdgpu_ib *ib,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
bool tmz)
uint32_t copy_flags)
{
ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_COPY,
1, 0, 0, byte_count);

4
drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
View File

@ -77,7 +77,7 @@ svm_migrate_gart_map(struct amdgpu_ring *ring, uint64_t npages,
dst_addr = amdgpu_bo_gpu_offset(adev->gart.bo);
amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr,
dst_addr, num_bytes, false);
dst_addr, num_bytes, 0);
amdgpu_ring_pad_ib(ring, &job->ibs[0]);
WARN_ON(job->ibs[0].length_dw > num_dw);
@ -153,7 +153,7 @@ svm_migrate_copy_memory_gart(struct amdgpu_device *adev, dma_addr_t *sys,
}
r = amdgpu_copy_buffer(ring, gart_s, gart_d, size * PAGE_SIZE,
NULL, &next, false, true, false);
NULL, &next, false, true, 0);
if (r) {
dev_err(adev->dev, "fail %d to copy memory\n", r);
goto out_unlock;