diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 44 |
1 files changed, 40 insertions, 4 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index fbca0494f871..5533f6e4f4a4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -3972,7 +3972,7 @@ static void gfx_v9_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, struct amdgpu_job *job, struct amdgpu_ib *ib, - bool ctx_switch) + uint32_t flags) { unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 header, control = 0; @@ -4005,11 +4005,27 @@ static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, static void gfx_v9_0_ring_emit_ib_compute(struct amdgpu_ring *ring, struct amdgpu_job *job, struct amdgpu_ib *ib, - bool ctx_switch) + uint32_t flags) { unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24); + /* Currently, there is a high possibility to get wave ID mismatch + * between ME and GDS, leading to a hw deadlock, because ME generates + * different wave IDs than the GDS expects. This situation happens + * randomly when at least 5 compute pipes use GDS ordered append. + * The wave IDs generated by ME are also wrong after suspend/resume. + * Those are probably bugs somewhere else in the kernel driver. + * + * Writing GDS_COMPUTE_MAX_WAVE_ID resets wave ID counters in ME and + * GDS to 0 for this ring (me/pipe). + */ + if (ib->flags & AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID) { + amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + amdgpu_ring_write(ring, mmGDS_COMPUTE_MAX_WAVE_ID); + amdgpu_ring_write(ring, ring->adev->gds.gds_compute_max_wave_id); + } + amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2)); BUG_ON(ib->gpu_addr & 0x3); /* Dword align */ amdgpu_ring_write(ring, @@ -4729,7 +4745,7 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_compute = { SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 + 2 + /* gfx_v9_0_ring_emit_vm_flush */ 8 + 8 + 8, /* gfx_v9_0_ring_emit_fence x3 for user fence, vm fence */ - .emit_ib_size = 4, /* gfx_v9_0_ring_emit_ib_compute */ + .emit_ib_size = 7, /* gfx_v9_0_ring_emit_ib_compute */ .emit_ib = gfx_v9_0_ring_emit_ib_compute, .emit_fence = gfx_v9_0_ring_emit_fence, .emit_pipeline_sync = gfx_v9_0_ring_emit_pipeline_sync, @@ -4764,7 +4780,7 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_kiq = { SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 + 2 + /* gfx_v9_0_ring_emit_vm_flush */ 8 + 8 + 8, /* gfx_v9_0_ring_emit_fence_kiq x3 for user fence, vm fence */ - .emit_ib_size = 4, /* gfx_v9_0_ring_emit_ib_compute */ + .emit_ib_size = 7, /* gfx_v9_0_ring_emit_ib_compute */ .emit_fence = gfx_v9_0_ring_emit_fence_kiq, .test_ring = gfx_v9_0_ring_test_ring, .insert_nop = amdgpu_ring_insert_nop, @@ -4846,6 +4862,26 @@ static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev) break; } + switch (adev->asic_type) { + case CHIP_VEGA10: + case CHIP_VEGA20: + adev->gds.gds_compute_max_wave_id = 0x7ff; + break; + case CHIP_VEGA12: + adev->gds.gds_compute_max_wave_id = 0x27f; + break; + case CHIP_RAVEN: + if (adev->rev_id >= 0x8) + adev->gds.gds_compute_max_wave_id = 0x77; /* raven2 */ + else + adev->gds.gds_compute_max_wave_id = 0x15f; /* raven1 */ + break; + default: + /* this really depends on the chip */ + adev->gds.gds_compute_max_wave_id = 0x7ff; + break; + } + adev->gds.gws.total_size = 64; adev->gds.oa.total_size = 16; |