drm/amdgpu: remove now superflous *_hdp operation

All HDP invalidation and most flush can now be replaced by the generic
ASIC function.

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Christian König 2018-01-19 15:19:16 +01:00 committed by Alex Deucher
parent 1b9d17db1c
commit 2ee150cda7
17 changed files with 18 additions and 302 deletions

View File

@ -1797,7 +1797,6 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
#define amdgpu_ring_emit_fence(r, addr, seq, flags) (r)->funcs->emit_fence((r), (addr), (seq), (flags))
#define amdgpu_ring_emit_gds_switch(r, v, db, ds, wb, ws, ab, as) (r)->funcs->emit_gds_switch((r), (v), (db), (ds), (wb), (ws), (ab), (as))
#define amdgpu_ring_emit_hdp_flush(r) (r)->funcs->emit_hdp_flush((r))
#define amdgpu_ring_emit_hdp_invalidate(r) (r)->funcs->emit_hdp_invalidate((r))
#define amdgpu_ring_emit_switch_buffer(r) (r)->funcs->emit_switch_buffer((r))
#define amdgpu_ring_emit_cntxcntl(r, d) (r)->funcs->emit_cntxcntl((r), (d))
#define amdgpu_ring_emit_rreg(r, d) (r)->funcs->emit_rreg((r), (d))

View File

@ -225,12 +225,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
#ifdef CONFIG_X86_64
if (!(adev->flags & AMD_IS_APU))
#endif
{
if (ring->funcs->emit_hdp_invalidate)
amdgpu_ring_emit_hdp_invalidate(ring);
else
amdgpu_asic_invalidate_hdp(adev, ring);
}
amdgpu_asic_invalidate_hdp(adev, ring);
r = amdgpu_fence_emit(ring, f);
if (r) {

View File

@ -128,7 +128,6 @@ struct amdgpu_ring_funcs {
void (*emit_vm_flush)(struct amdgpu_ring *ring, unsigned vmid,
unsigned pasid, uint64_t pd_addr);
void (*emit_hdp_flush)(struct amdgpu_ring *ring);
void (*emit_hdp_invalidate)(struct amdgpu_ring *ring);
void (*emit_gds_switch)(struct amdgpu_ring *ring, uint32_t vmid,
uint32_t gds_base, uint32_t gds_size,
uint32_t gws_base, uint32_t gws_size,

View File

@ -261,13 +261,6 @@ static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring)
amdgpu_ring_write(ring, (0xfff << 16) | 10); /* retry count, poll interval */
}
static void cik_sdma_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
amdgpu_ring_write(ring, mmHDP_DEBUG0);
amdgpu_ring_write(ring, 1);
}
/**
* cik_sdma_ring_emit_fence - emit a fence on the DMA ring
*
@ -1277,7 +1270,7 @@ static const struct amdgpu_ring_funcs cik_sdma_ring_funcs = {
.set_wptr = cik_sdma_ring_set_wptr,
.emit_frame_size =
6 + /* cik_sdma_ring_emit_hdp_flush */
3 + /* cik_sdma_ring_emit_hdp_invalidate */
3 + /* hdp invalidate */
6 + /* cik_sdma_ring_emit_pipeline_sync */
CIK_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* cik_sdma_ring_emit_vm_flush */
9 + 9 + 9, /* cik_sdma_ring_emit_fence x3 for user fence, vm fence */
@ -1287,7 +1280,6 @@ static const struct amdgpu_ring_funcs cik_sdma_ring_funcs = {
.emit_pipeline_sync = cik_sdma_ring_emit_pipeline_sync,
.emit_vm_flush = cik_sdma_ring_emit_vm_flush,
.emit_hdp_flush = cik_sdma_ring_emit_hdp_flush,
.emit_hdp_invalidate = cik_sdma_ring_emit_hdp_invalidate,
.test_ring = cik_sdma_ring_test_ring,
.test_ib = cik_sdma_ring_test_ib,
.insert_nop = cik_sdma_ring_insert_nop,

View File

@ -1809,17 +1809,6 @@ static int gfx_v6_0_ring_test_ring(struct amdgpu_ring *ring)
return r;
}
static void gfx_v6_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
{
/* flush hdp cache */
amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) |
WRITE_DATA_DST_SEL(0)));
amdgpu_ring_write(ring, mmHDP_MEM_COHERENCY_FLUSH_CNTL);
amdgpu_ring_write(ring, 0);
amdgpu_ring_write(ring, 0x1);
}
static void gfx_v6_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE, 0));
@ -1827,24 +1816,6 @@ static void gfx_v6_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
EVENT_INDEX(0));
}
/**
* gfx_v6_0_ring_emit_hdp_invalidate - emit an hdp invalidate on the cp
*
* @adev: amdgpu_device pointer
* @ridx: amdgpu ring index
*
* Emits an hdp invalidate on the cp.
*/
static void gfx_v6_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) |
WRITE_DATA_DST_SEL(0)));
amdgpu_ring_write(ring, mmHDP_DEBUG0);
amdgpu_ring_write(ring, 0);
amdgpu_ring_write(ring, 0x1);
}
static void gfx_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
u64 seq, unsigned flags)
{
@ -3507,8 +3478,7 @@ static const struct amdgpu_ring_funcs gfx_v6_0_ring_funcs_gfx = {
.get_wptr = gfx_v6_0_ring_get_wptr,
.set_wptr = gfx_v6_0_ring_set_wptr_gfx,
.emit_frame_size =
5 + /* gfx_v6_0_ring_emit_hdp_flush */
5 + /* gfx_v6_0_ring_emit_hdp_invalidate */
5 + 5 + /* hdp flush / invalidate */
14 + 14 + 14 + /* gfx_v6_0_ring_emit_fence x3 for user fence, vm fence */
7 + 4 + /* gfx_v6_0_ring_emit_pipeline_sync */
SI_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + 6 + /* gfx_v6_0_ring_emit_vm_flush */
@ -3518,8 +3488,6 @@ static const struct amdgpu_ring_funcs gfx_v6_0_ring_funcs_gfx = {
.emit_fence = gfx_v6_0_ring_emit_fence,
.emit_pipeline_sync = gfx_v6_0_ring_emit_pipeline_sync,
.emit_vm_flush = gfx_v6_0_ring_emit_vm_flush,
.emit_hdp_flush = gfx_v6_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = gfx_v6_0_ring_emit_hdp_invalidate,
.test_ring = gfx_v6_0_ring_test_ring,
.test_ib = gfx_v6_0_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,
@ -3535,8 +3503,7 @@ static const struct amdgpu_ring_funcs gfx_v6_0_ring_funcs_compute = {
.get_wptr = gfx_v6_0_ring_get_wptr,
.set_wptr = gfx_v6_0_ring_set_wptr_compute,
.emit_frame_size =
5 + /* gfx_v6_0_ring_emit_hdp_flush */
5 + /* gfx_v6_0_ring_emit_hdp_invalidate */
5 + 5 + /* hdp flush / invalidate */
7 + /* gfx_v6_0_ring_emit_pipeline_sync */
SI_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v6_0_ring_emit_vm_flush */
14 + 14 + 14, /* gfx_v6_0_ring_emit_fence x3 for user fence, vm fence */
@ -3545,8 +3512,6 @@ static const struct amdgpu_ring_funcs gfx_v6_0_ring_funcs_compute = {
.emit_fence = gfx_v6_0_ring_emit_fence,
.emit_pipeline_sync = gfx_v6_0_ring_emit_pipeline_sync,
.emit_vm_flush = gfx_v6_0_ring_emit_vm_flush,
.emit_hdp_flush = gfx_v6_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = gfx_v6_0_ring_emit_hdp_invalidate,
.test_ring = gfx_v6_0_ring_test_ring,
.test_ib = gfx_v6_0_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,

View File

@ -2147,26 +2147,6 @@ static void gfx_v7_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
EVENT_INDEX(0));
}
/**
* gfx_v7_0_ring_emit_hdp_invalidate - emit an hdp invalidate on the cp
*
* @adev: amdgpu_device pointer
* @ridx: amdgpu ring index
*
* Emits an hdp invalidate on the cp.
*/
static void gfx_v7_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
WRITE_DATA_DST_SEL(0) |
WR_CONFIRM));
amdgpu_ring_write(ring, mmHDP_DEBUG0);
amdgpu_ring_write(ring, 0);
amdgpu_ring_write(ring, 1);
}
/**
* gfx_v7_0_ring_emit_fence_gfx - emit a fence on the gfx ring
*
@ -5110,7 +5090,7 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = {
.emit_frame_size =
20 + /* gfx_v7_0_ring_emit_gds_switch */
7 + /* gfx_v7_0_ring_emit_hdp_flush */
5 + /* gfx_v7_0_ring_emit_hdp_invalidate */
5 + /* hdp invalidate */
12 + 12 + 12 + /* gfx_v7_0_ring_emit_fence_gfx x3 for user fence, vm fence */
7 + 4 + /* gfx_v7_0_ring_emit_pipeline_sync */
CIK_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + 6 + /* gfx_v7_0_ring_emit_vm_flush */
@ -5122,7 +5102,6 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = {
.emit_vm_flush = gfx_v7_0_ring_emit_vm_flush,
.emit_gds_switch = gfx_v7_0_ring_emit_gds_switch,
.emit_hdp_flush = gfx_v7_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = gfx_v7_0_ring_emit_hdp_invalidate,
.test_ring = gfx_v7_0_ring_test_ring,
.test_ib = gfx_v7_0_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,
@ -5142,7 +5121,7 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
.emit_frame_size =
20 + /* gfx_v7_0_ring_emit_gds_switch */
7 + /* gfx_v7_0_ring_emit_hdp_flush */
5 + /* gfx_v7_0_ring_emit_hdp_invalidate */
5 + /* hdp invalidate */
7 + /* gfx_v7_0_ring_emit_pipeline_sync */
CIK_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v7_0_ring_emit_vm_flush */
7 + 7 + 7, /* gfx_v7_0_ring_emit_fence_compute x3 for user fence, vm fence */
@ -5153,7 +5132,6 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
.emit_vm_flush = gfx_v7_0_ring_emit_vm_flush,
.emit_gds_switch = gfx_v7_0_ring_emit_gds_switch,
.emit_hdp_flush = gfx_v7_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = gfx_v7_0_ring_emit_hdp_invalidate,
.test_ring = gfx_v7_0_ring_test_ring,
.test_ib = gfx_v7_0_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,

View File

@ -6230,19 +6230,6 @@ static void gfx_v8_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
EVENT_INDEX(0));
}
static void gfx_v8_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
WRITE_DATA_DST_SEL(0) |
WR_CONFIRM));
amdgpu_ring_write(ring, mmHDP_DEBUG0);
amdgpu_ring_write(ring, 0);
amdgpu_ring_write(ring, 1);
}
static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
struct amdgpu_ib *ib,
unsigned vmid, bool ctx_switch)
@ -6887,7 +6874,6 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
.emit_vm_flush = gfx_v8_0_ring_emit_vm_flush,
.emit_gds_switch = gfx_v8_0_ring_emit_gds_switch,
.emit_hdp_flush = gfx_v8_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = gfx_v8_0_ring_emit_hdp_invalidate,
.test_ring = gfx_v8_0_ring_test_ring,
.test_ib = gfx_v8_0_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,
@ -6910,7 +6896,7 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
.emit_frame_size =
20 + /* gfx_v8_0_ring_emit_gds_switch */
7 + /* gfx_v8_0_ring_emit_hdp_flush */
5 + /* gfx_v8_0_ring_emit_hdp_invalidate */
5 + /* hdp_invalidate */
7 + /* gfx_v8_0_ring_emit_pipeline_sync */
VI_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v8_0_ring_emit_vm_flush */
7 + 7 + 7, /* gfx_v8_0_ring_emit_fence_compute x3 for user fence, vm fence */
@ -6921,7 +6907,6 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
.emit_vm_flush = gfx_v8_0_ring_emit_vm_flush,
.emit_gds_switch = gfx_v8_0_ring_emit_gds_switch,
.emit_hdp_flush = gfx_v8_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = gfx_v8_0_ring_emit_hdp_invalidate,
.test_ring = gfx_v8_0_ring_test_ring,
.test_ib = gfx_v8_0_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,
@ -6941,7 +6926,7 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_kiq = {
.emit_frame_size =
20 + /* gfx_v8_0_ring_emit_gds_switch */
7 + /* gfx_v8_0_ring_emit_hdp_flush */
5 + /* gfx_v8_0_ring_emit_hdp_invalidate */
5 + /* hdp_invalidate */
7 + /* gfx_v8_0_ring_emit_pipeline_sync */
17 + /* gfx_v8_0_ring_emit_vm_flush */
7 + 7 + 7, /* gfx_v8_0_ring_emit_fence_kiq x3 for user fence, vm fence */

View File

@ -3585,14 +3585,6 @@ static void gfx_v9_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
ref_and_mask, ref_and_mask, 0x20);
}
static void gfx_v9_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
gfx_v9_0_write_data_to_reg(ring, 0, true,
SOC15_REG_OFFSET(HDP, 0, mmHDP_READ_CACHE_INVALIDATE), 1);
}
static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
struct amdgpu_ib *ib,
unsigned vmid, bool ctx_switch)
@ -4319,7 +4311,6 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = {
.emit_vm_flush = gfx_v9_0_ring_emit_vm_flush,
.emit_gds_switch = gfx_v9_0_ring_emit_gds_switch,
.emit_hdp_flush = gfx_v9_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = gfx_v9_0_ring_emit_hdp_invalidate,
.test_ring = gfx_v9_0_ring_test_ring,
.test_ib = gfx_v9_0_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,
@ -4344,7 +4335,7 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_compute = {
.emit_frame_size =
20 + /* gfx_v9_0_ring_emit_gds_switch */
7 + /* gfx_v9_0_ring_emit_hdp_flush */
5 + /* gfx_v9_0_ring_emit_hdp_invalidate */
5 + /* hdp invalidate */
7 + /* gfx_v9_0_ring_emit_pipeline_sync */
SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 + 9 + /* gfx_v9_0_ring_emit_vm_flush */
8 + 8 + 8, /* gfx_v9_0_ring_emit_fence x3 for user fence, vm fence */
@ -4355,7 +4346,6 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_compute = {
.emit_vm_flush = gfx_v9_0_ring_emit_vm_flush,
.emit_gds_switch = gfx_v9_0_ring_emit_gds_switch,
.emit_hdp_flush = gfx_v9_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = gfx_v9_0_ring_emit_hdp_invalidate,
.test_ring = gfx_v9_0_ring_test_ring,
.test_ib = gfx_v9_0_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,
@ -4376,7 +4366,7 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_kiq = {
.emit_frame_size =
20 + /* gfx_v9_0_ring_emit_gds_switch */
7 + /* gfx_v9_0_ring_emit_hdp_flush */
5 + /* gfx_v9_0_ring_emit_hdp_invalidate */
5 + /* hdp invalidate */
7 + /* gfx_v9_0_ring_emit_pipeline_sync */
SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 + 9 + /* gfx_v9_0_ring_emit_vm_flush */
8 + 8 + 8, /* gfx_v9_0_ring_emit_fence_kiq x3 for user fence, vm fence */

View File

@ -289,13 +289,6 @@ static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring)
SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
}
static void sdma_v2_4_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
amdgpu_ring_write(ring, mmHDP_DEBUG0);
amdgpu_ring_write(ring, 1);
}
/**
* sdma_v2_4_ring_emit_fence - emit a fence on the DMA ring
*
@ -1200,7 +1193,7 @@ static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = {
.set_wptr = sdma_v2_4_ring_set_wptr,
.emit_frame_size =
6 + /* sdma_v2_4_ring_emit_hdp_flush */
3 + /* sdma_v2_4_ring_emit_hdp_invalidate */
3 + /* hdp invalidate */
6 + /* sdma_v2_4_ring_emit_pipeline_sync */
VI_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* sdma_v2_4_ring_emit_vm_flush */
10 + 10 + 10, /* sdma_v2_4_ring_emit_fence x3 for user fence, vm fence */
@ -1210,7 +1203,6 @@ static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = {
.emit_pipeline_sync = sdma_v2_4_ring_emit_pipeline_sync,
.emit_vm_flush = sdma_v2_4_ring_emit_vm_flush,
.emit_hdp_flush = sdma_v2_4_ring_emit_hdp_flush,
.emit_hdp_invalidate = sdma_v2_4_ring_emit_hdp_invalidate,
.test_ring = sdma_v2_4_ring_test_ring,
.test_ib = sdma_v2_4_ring_test_ib,
.insert_nop = sdma_v2_4_ring_insert_nop,

View File

@ -460,14 +460,6 @@ static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
}
static void sdma_v3_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
amdgpu_ring_write(ring, mmHDP_DEBUG0);
amdgpu_ring_write(ring, 1);
}
/**
* sdma_v3_0_ring_emit_fence - emit a fence on the DMA ring
*
@ -1634,7 +1626,7 @@ static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = {
.set_wptr = sdma_v3_0_ring_set_wptr,
.emit_frame_size =
6 + /* sdma_v3_0_ring_emit_hdp_flush */
3 + /* sdma_v3_0_ring_emit_hdp_invalidate */
3 + /* hdp invalidate */
6 + /* sdma_v3_0_ring_emit_pipeline_sync */
12 + /* sdma_v3_0_ring_emit_vm_flush */
10 + 10 + 10, /* sdma_v3_0_ring_emit_fence x3 for user fence, vm fence */
@ -1644,7 +1636,6 @@ static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = {
.emit_pipeline_sync = sdma_v3_0_ring_emit_pipeline_sync,
.emit_vm_flush = sdma_v3_0_ring_emit_vm_flush,
.emit_hdp_flush = sdma_v3_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = sdma_v3_0_ring_emit_hdp_invalidate,
.test_ring = sdma_v3_0_ring_test_ring,
.test_ib = sdma_v3_0_ring_test_ib,
.insert_nop = sdma_v3_0_ring_insert_nop,

View File

@ -375,16 +375,6 @@ static void sdma_v4_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
}
static void sdma_v4_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
amdgpu_ring_write(ring, SOC15_REG_OFFSET(HDP, 0, mmHDP_READ_CACHE_INVALIDATE));
amdgpu_ring_write(ring, 1);
}
/**
* sdma_v4_0_ring_emit_fence - emit a fence on the DMA ring
*
@ -1583,7 +1573,7 @@ static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = {
.set_wptr = sdma_v4_0_ring_set_wptr,
.emit_frame_size =
6 + /* sdma_v4_0_ring_emit_hdp_flush */
3 + /* sdma_v4_0_ring_emit_hdp_invalidate */
3 + /* hdp invalidate */
6 + /* sdma_v4_0_ring_emit_pipeline_sync */
SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* sdma_v4_0_ring_emit_vm_flush */
10 + 10 + 10, /* sdma_v4_0_ring_emit_fence x3 for user fence, vm fence */
@ -1593,7 +1583,6 @@ static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = {
.emit_pipeline_sync = sdma_v4_0_ring_emit_pipeline_sync,
.emit_vm_flush = sdma_v4_0_ring_emit_vm_flush,
.emit_hdp_flush = sdma_v4_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = sdma_v4_0_ring_emit_hdp_invalidate,
.test_ring = sdma_v4_0_ring_test_ring,
.test_ib = sdma_v4_0_ring_test_ib,
.insert_nop = sdma_v4_0_ring_insert_nop,

View File

@ -75,20 +75,6 @@ static void si_dma_ring_emit_ib(struct amdgpu_ring *ring,
}
static void si_dma_ring_emit_hdp_flush(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0));
amdgpu_ring_write(ring, (0xf << 16) | (HDP_MEM_COHERENCY_FLUSH_CNTL));
amdgpu_ring_write(ring, 1);
}
static void si_dma_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0));
amdgpu_ring_write(ring, (0xf << 16) | (HDP_DEBUG0));
amdgpu_ring_write(ring, 1);
}
/**
* si_dma_ring_emit_fence - emit a fence on the DMA ring
*
@ -772,8 +758,7 @@ static const struct amdgpu_ring_funcs si_dma_ring_funcs = {
.get_wptr = si_dma_ring_get_wptr,
.set_wptr = si_dma_ring_set_wptr,
.emit_frame_size =
3 + /* si_dma_ring_emit_hdp_flush */
3 + /* si_dma_ring_emit_hdp_invalidate */
3 + 3 + /* hdp flush / invalidate */
6 + /* si_dma_ring_emit_pipeline_sync */
SI_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* si_dma_ring_emit_vm_flush */
9 + 9 + 9, /* si_dma_ring_emit_fence x3 for user fence, vm fence */
@ -782,8 +767,6 @@ static const struct amdgpu_ring_funcs si_dma_ring_funcs = {
.emit_fence = si_dma_ring_emit_fence,
.emit_pipeline_sync = si_dma_ring_emit_pipeline_sync,
.emit_vm_flush = si_dma_ring_emit_vm_flush,
.emit_hdp_flush = si_dma_ring_emit_hdp_flush,
.emit_hdp_invalidate = si_dma_ring_emit_hdp_invalidate,
.test_ring = si_dma_ring_test_ring,
.test_ib = si_dma_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,

View File

@ -463,32 +463,6 @@ static void uvd_v4_2_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
amdgpu_ring_write(ring, 2);
}
/**
* uvd_v4_2_ring_emit_hdp_flush - emit an hdp flush
*
* @ring: amdgpu_ring pointer
*
* Emits an hdp flush.
*/
static void uvd_v4_2_ring_emit_hdp_flush(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, PACKET0(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0));
amdgpu_ring_write(ring, 0);
}
/**
* uvd_v4_2_ring_hdp_invalidate - emit an hdp invalidate
*
* @ring: amdgpu_ring pointer
*
* Emits an hdp invalidate.
*/
static void uvd_v4_2_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, PACKET0(mmHDP_DEBUG0, 0));
amdgpu_ring_write(ring, 1);
}
/**
* uvd_v4_2_ring_test_ring - register write test
*
@ -765,14 +739,10 @@ static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = {
.set_wptr = uvd_v4_2_ring_set_wptr,
.parse_cs = amdgpu_uvd_ring_parse_cs,
.emit_frame_size =
2 + /* uvd_v4_2_ring_emit_hdp_flush */
2 + /* uvd_v4_2_ring_emit_hdp_invalidate */
14, /* uvd_v4_2_ring_emit_fence x1 no user fence */
.emit_ib_size = 4, /* uvd_v4_2_ring_emit_ib */
.emit_ib = uvd_v4_2_ring_emit_ib,
.emit_fence = uvd_v4_2_ring_emit_fence,
.emit_hdp_flush = uvd_v4_2_ring_emit_hdp_flush,
.emit_hdp_invalidate = uvd_v4_2_ring_emit_hdp_invalidate,
.test_ring = uvd_v4_2_ring_test_ring,
.test_ib = amdgpu_uvd_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,

View File

@ -478,32 +478,6 @@ static void uvd_v5_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
amdgpu_ring_write(ring, 2);
}
/**
* uvd_v5_0_ring_emit_hdp_flush - emit an hdp flush
*
* @ring: amdgpu_ring pointer
*
* Emits an hdp flush.
*/
static void uvd_v5_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, PACKET0(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0));
amdgpu_ring_write(ring, 0);
}
/**
* uvd_v5_0_ring_hdp_invalidate - emit an hdp invalidate
*
* @ring: amdgpu_ring pointer
*
* Emits an hdp invalidate.
*/
static void uvd_v5_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, PACKET0(mmHDP_DEBUG0, 0));
amdgpu_ring_write(ring, 1);
}
/**
* uvd_v5_0_ring_test_ring - register write test
*
@ -873,14 +847,10 @@ static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = {
.set_wptr = uvd_v5_0_ring_set_wptr,
.parse_cs = amdgpu_uvd_ring_parse_cs,
.emit_frame_size =
2 + /* uvd_v5_0_ring_emit_hdp_flush */
2 + /* uvd_v5_0_ring_emit_hdp_invalidate */
14, /* uvd_v5_0_ring_emit_fence x1 no user fence */
.emit_ib_size = 6, /* uvd_v5_0_ring_emit_ib */
.emit_ib = uvd_v5_0_ring_emit_ib,
.emit_fence = uvd_v5_0_ring_emit_fence,
.emit_hdp_flush = uvd_v5_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = uvd_v5_0_ring_emit_hdp_invalidate,
.test_ring = uvd_v5_0_ring_test_ring,
.test_ib = amdgpu_uvd_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,

View File

@ -963,32 +963,6 @@ static void uvd_v6_0_enc_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
amdgpu_ring_write(ring, HEVC_ENC_CMD_TRAP);
}
/**
* uvd_v6_0_ring_emit_hdp_flush - emit an hdp flush
*
* @ring: amdgpu_ring pointer
*
* Emits an hdp flush.
*/
static void uvd_v6_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, PACKET0(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0));
amdgpu_ring_write(ring, 0);
}
/**
* uvd_v6_0_ring_hdp_invalidate - emit an hdp invalidate
*
* @ring: amdgpu_ring pointer
*
* Emits an hdp invalidate.
*/
static void uvd_v6_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
amdgpu_ring_write(ring, PACKET0(mmHDP_DEBUG0, 0));
amdgpu_ring_write(ring, 1);
}
/**
* uvd_v6_0_ring_test_ring - register write test
*
@ -1556,15 +1530,11 @@ static const struct amdgpu_ring_funcs uvd_v6_0_ring_phys_funcs = {
.set_wptr = uvd_v6_0_ring_set_wptr,
.parse_cs = amdgpu_uvd_ring_parse_cs,
.emit_frame_size =
2 + /* uvd_v6_0_ring_emit_hdp_flush */
2 + /* uvd_v6_0_ring_emit_hdp_invalidate */
10 + /* uvd_v6_0_ring_emit_pipeline_sync */
14, /* uvd_v6_0_ring_emit_fence x1 no user fence */
.emit_ib_size = 8, /* uvd_v6_0_ring_emit_ib */
.emit_ib = uvd_v6_0_ring_emit_ib,
.emit_fence = uvd_v6_0_ring_emit_fence,
.emit_hdp_flush = uvd_v6_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = uvd_v6_0_ring_emit_hdp_invalidate,
.test_ring = uvd_v6_0_ring_test_ring,
.test_ib = amdgpu_uvd_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,
@ -1582,8 +1552,7 @@ static const struct amdgpu_ring_funcs uvd_v6_0_ring_vm_funcs = {
.get_wptr = uvd_v6_0_ring_get_wptr,
.set_wptr = uvd_v6_0_ring_set_wptr,
.emit_frame_size =
2 + /* uvd_v6_0_ring_emit_hdp_flush */
2 + /* uvd_v6_0_ring_emit_hdp_invalidate */
6 + 6 + /* hdp flush / invalidate */
10 + /* uvd_v6_0_ring_emit_pipeline_sync */
20 + /* uvd_v6_0_ring_emit_vm_flush */
14 + 14, /* uvd_v6_0_ring_emit_fence x2 vm fence */
@ -1592,8 +1561,6 @@ static const struct amdgpu_ring_funcs uvd_v6_0_ring_vm_funcs = {
.emit_fence = uvd_v6_0_ring_emit_fence,
.emit_vm_flush = uvd_v6_0_ring_emit_vm_flush,
.emit_pipeline_sync = uvd_v6_0_ring_emit_pipeline_sync,
.emit_hdp_flush = uvd_v6_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = uvd_v6_0_ring_emit_hdp_invalidate,
.test_ring = uvd_v6_0_ring_test_ring,
.test_ib = amdgpu_uvd_ring_test_ib,
.insert_nop = amdgpu_ring_insert_nop,

View File

@ -1135,37 +1135,6 @@ static void uvd_v7_0_enc_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
amdgpu_ring_write(ring, HEVC_ENC_CMD_TRAP);
}
/**
* uvd_v7_0_ring_emit_hdp_flush - emit an hdp flush
*
* @ring: amdgpu_ring pointer
*
* Emits an hdp flush.
*/
static void uvd_v7_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(NBIF, 0,
mmHDP_MEM_COHERENCY_FLUSH_CNTL), 0));
amdgpu_ring_write(ring, 0);
}
/**
* uvd_v7_0_ring_hdp_invalidate - emit an hdp invalidate
*
* @ring: amdgpu_ring pointer
*
* Emits an hdp invalidate.
*/
static void uvd_v7_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(HDP, 0, mmHDP_READ_CACHE_INVALIDATE), 0));
amdgpu_ring_write(ring, 1);
}
/**
* uvd_v7_0_ring_test_ring - register write test
*
@ -1693,16 +1662,13 @@ static const struct amdgpu_ring_funcs uvd_v7_0_ring_vm_funcs = {
.get_wptr = uvd_v7_0_ring_get_wptr,
.set_wptr = uvd_v7_0_ring_set_wptr,
.emit_frame_size =
2 + /* uvd_v7_0_ring_emit_hdp_flush */
2 + /* uvd_v7_0_ring_emit_hdp_invalidate */
6 + 6 + /* hdp flush / invalidate */
SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 + 16 + /* uvd_v7_0_ring_emit_vm_flush */
14 + 14, /* uvd_v7_0_ring_emit_fence x2 vm fence */
.emit_ib_size = 8, /* uvd_v7_0_ring_emit_ib */
.emit_ib = uvd_v7_0_ring_emit_ib,
.emit_fence = uvd_v7_0_ring_emit_fence,
.emit_vm_flush = uvd_v7_0_ring_emit_vm_flush,
.emit_hdp_flush = uvd_v7_0_ring_emit_hdp_flush,
.emit_hdp_invalidate = uvd_v7_0_ring_emit_hdp_invalidate,
.test_ring = uvd_v7_0_ring_test_ring,
.test_ib = amdgpu_uvd_ring_test_ib,
.insert_nop = uvd_v7_0_ring_insert_nop,
@ -1722,6 +1688,7 @@ static const struct amdgpu_ring_funcs uvd_v7_0_enc_ring_vm_funcs = {
.get_wptr = uvd_v7_0_enc_ring_get_wptr,
.set_wptr = uvd_v7_0_enc_ring_set_wptr,
.emit_frame_size =
3 + 3 + /* hdp flush / invalidate */
SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 + 8 + /* uvd_v7_0_enc_ring_emit_vm_flush */
5 + 5 + /* uvd_v7_0_enc_ring_emit_fence x2 vm fence */
1, /* uvd_v7_0_enc_ring_insert_end */

View File

@ -809,21 +809,6 @@ static void vcn_v1_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64
amdgpu_ring_write(ring, VCN_DEC_CMD_TRAP << 1);
}
/**
* vcn_v1_0_dec_ring_hdp_invalidate - emit an hdp invalidate
*
* @ring: amdgpu_ring pointer
*
* Emits an hdp invalidate.
*/
static void vcn_v1_0_dec_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(HDP, 0, mmHDP_READ_CACHE_INVALIDATE), 0));
amdgpu_ring_write(ring, 1);
}
/**
* vcn_v1_0_dec_ring_emit_ib - execute indirect buffer
*
@ -1096,7 +1081,7 @@ static const struct amdgpu_ring_funcs vcn_v1_0_dec_ring_vm_funcs = {
.get_wptr = vcn_v1_0_dec_ring_get_wptr,
.set_wptr = vcn_v1_0_dec_ring_set_wptr,
.emit_frame_size =
2 + /* vcn_v1_0_dec_ring_emit_hdp_invalidate */
6 + 6 + /* hdp invalidate / flush */
SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 + 16 + /* vcn_v1_0_dec_ring_emit_vm_flush */
14 + 14 + /* vcn_v1_0_dec_ring_emit_fence x2 vm fence */
6,
@ -1104,7 +1089,6 @@ static const struct amdgpu_ring_funcs vcn_v1_0_dec_ring_vm_funcs = {
.emit_ib = vcn_v1_0_dec_ring_emit_ib,
.emit_fence = vcn_v1_0_dec_ring_emit_fence,
.emit_vm_flush = vcn_v1_0_dec_ring_emit_vm_flush,
.emit_hdp_invalidate = vcn_v1_0_dec_ring_emit_hdp_invalidate,
.test_ring = amdgpu_vcn_dec_ring_test_ring,
.test_ib = amdgpu_vcn_dec_ring_test_ib,
.insert_nop = vcn_v1_0_ring_insert_nop,