From: Frank Min <Frank.Min@xxxxxxx>
1. Add dcc buffer flag for copy buffer
2. Add sdma 7.0 support copy dcc buffer
Signed-off-by: Likun Gao <Likun.Gao@xxxxxxx>
Signed-off-by: Frank Min <Frank.Min@xxxxxxx>
Reviewed-by: Alex Deucher <alexander.deucher@xxxxxxx>
Reviewed-by: Christian König <christian.koenig@xxxxxxx>
Signed-off-by: Alex Deucher <alexander.deucher@xxxxxxx>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 8 +++++++-
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h | 3 ++-
drivers/gpu/drm/amd/amdgpu/sdma_v6_0_0_pkt_open.h | 8 ++++++++
drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c | 12 ++++++++++--
4 files changed, 27 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index 0364a7bb5893..7c0907cd8b1a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -295,8 +295,8 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
struct amdgpu_res_cursor src_mm, dst_mm;
struct dma_fence *fence = NULL;
int r = 0;
-
uint32_t copy_flags = 0;
+ struct amdgpu_bo *abo_src, *abo_dst;
if (!adev->mman.buffer_funcs_enabled) {
DRM_ERROR("Trying to move memory with ring turned off.\n");
@@ -325,8 +325,14 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
if (r)
goto error;
+ abo_src = ttm_to_amdgpu_bo(src->bo);
+ abo_dst = ttm_to_amdgpu_bo(dst->bo);
if (tmz)
copy_flags |= AMDGPU_COPY_FLAGS_TMZ;
+ if (abo_src->flags & AMDGPU_GEM_CREATE_GFX12_DCC)
+ copy_flags |= AMDGPU_COPY_FLAGS_READ_DECOMPRESSED;
+ if (abo_dst->flags & AMDGPU_GEM_CREATE_GFX12_DCC)
+ copy_flags |= AMDGPU_COPY_FLAGS_WRITE_COMPRESSED;
r = amdgpu_copy_buffer(ring, from, to, cur_size, resv,
&next, false, true, copy_flags);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
index 53d5a5990c31..7c903a6c9ddb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
@@ -112,6 +112,8 @@ struct amdgpu_copy_mem {
};
#define AMDGPU_COPY_FLAGS_TMZ (1 << 0)
+#define AMDGPU_COPY_FLAGS_READ_DECOMPRESSED (1 << 1)
+#define AMDGPU_COPY_FLAGS_WRITE_COMPRESSED (1 << 2)
int amdgpu_gtt_mgr_init(struct amdgpu_device *adev, uint64_t gtt_size);
void amdgpu_gtt_mgr_fini(struct amdgpu_device *adev);
@@ -148,7 +150,6 @@ int amdgpu_ttm_init(struct amdgpu_device *adev);
void amdgpu_ttm_fini(struct amdgpu_device *adev);
void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev,
bool enable);
-
int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
uint64_t dst_offset, uint32_t byte_count,
struct dma_resv *resv,
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0_0_pkt_open.h b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0_0_pkt_open.h
index 6af23e7888ca..d8cf830916b9 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0_0_pkt_open.h
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0_0_pkt_open.h
@@ -91,6 +91,14 @@
#define SDMA_GCR_GLM_WB (1 << 4)
#define SDMA_GCR_GL1_RANGE(x) (((x) & 0x3) << 2)
#define SDMA_GCR_GLI_INV(x) (((x) & 0x3) << 0)
+
+#define SDMA_DCC_DATA_FORMAT(x) ((x) & 0x3f)
+#define SDMA_DCC_NUM_TYPE(x) (((x) & 0x7) << 9)
+#define SDMA_DCC_READ_CM(x) (((x) & 0x3) << 16)
+#define SDMA_DCC_WRITE_CM(x) (((x) & 0x3) << 18)
+#define SDMA_DCC_MAX_COM(x) (((x) & 0x3) << 24)
+#define SDMA_DCC_MAX_UCOM(x) (((x) & 0x1) << 26)
+
/*
** Definitions for SDMA_PKT_COPY_LINEAR packet
*/
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c
index ab1dea77be6e..96514fd77e35 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c
@@ -1568,13 +1568,22 @@ static void sdma_v7_0_emit_copy_buffer(struct amdgpu_ib *ib,
{
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((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0);
+ SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0) |
+ SDMA_PKT_COPY_LINEAR_HEADER_CPV((copy_flags &
+ (AMDGPU_COPY_FLAGS_READ_DECOMPRESSED | AMDGPU_COPY_FLAGS_WRITE_COMPRESSED)) ? 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);
ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+
+ if ((copy_flags & (AMDGPU_COPY_FLAGS_READ_DECOMPRESSED | AMDGPU_COPY_FLAGS_WRITE_COMPRESSED)))
+ ib->ptr[ib->length_dw++] = SDMA_DCC_DATA_FORMAT(4) | SDMA_DCC_NUM_TYPE(4) |
+ ((copy_flags & AMDGPU_COPY_FLAGS_READ_DECOMPRESSED) ? SDMA_DCC_READ_CM(2) : 0) |
+ ((copy_flags & AMDGPU_COPY_FLAGS_WRITE_COMPRESSED) ? SDMA_DCC_WRITE_CM(1) : 0) |
+ SDMA_DCC_MAX_COM(1) | SDMA_DCC_MAX_UCOM(1);
}
/**
@@ -1603,7 +1612,6 @@ static const struct amdgpu_buffer_funcs sdma_v7_0_buffer_funcs = {
.copy_max_bytes = 0x400000,
.copy_num_dw = 7,
.emit_copy_buffer = sdma_v7_0_emit_copy_buffer,
-
.fill_max_bytes = 0x400000,
.fill_num_dw = 5,
.emit_fill_buffer = sdma_v7_0_emit_fill_buffer,
--
2.45.1
