diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 59 |
1 files changed, 31 insertions, 28 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 2932ade7dbd0..ee41d5592c51 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -1544,24 +1544,6 @@ static void gfx_v10_0_constants_init(struct amdgpu_device *adev) gfx_v10_0_init_compute_vmid(adev); - mutex_lock(&adev->grbm_idx_mutex); - /* - * making sure that the following register writes will be broadcasted - * to all the shaders - */ - gfx_v10_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); - - tmp = REG_SET_FIELD(0, PA_SC_FIFO_SIZE, SC_FRONTEND_PRIM_FIFO_SIZE, - adev->gfx.config.sc_prim_fifo_size_frontend); - tmp = REG_SET_FIELD(tmp, PA_SC_FIFO_SIZE, SC_BACKEND_PRIM_FIFO_SIZE, - adev->gfx.config.sc_prim_fifo_size_backend); - tmp = REG_SET_FIELD(tmp, PA_SC_FIFO_SIZE, SC_HIZ_TILE_FIFO_SIZE, - adev->gfx.config.sc_hiz_tile_fifo_size); - tmp = REG_SET_FIELD(tmp, PA_SC_FIFO_SIZE, SC_EARLYZ_TILE_FIFO_SIZE, - adev->gfx.config.sc_earlyz_tile_fifo_size); - WREG32_SOC15(GC, 0, mmPA_SC_FIFO_SIZE, tmp); - - mutex_unlock(&adev->grbm_idx_mutex); } static void gfx_v10_0_enable_gui_idle_interrupt(struct amdgpu_device *adev, @@ -4215,6 +4197,15 @@ static void gfx_v10_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 header, control = 0; + /* Prevent a hw deadlock due to a wave ID mismatch between ME and GDS. + * This resets the wave ID counters. (needed by transform feedback) + * TODO: This might only be needed on a VMID switch when we change + * the GDS OA mapping, not sure. + */ + amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + amdgpu_ring_write(ring, mmVGT_GS_MAX_WAVE_ID); + amdgpu_ring_write(ring, ring->adev->gds.vgt_gs_max_wave_id); + if (ib->flags & AMDGPU_IB_FLAG_CE) header = PACKET3(PACKET3_INDIRECT_BUFFER_CNST, 2); else @@ -4252,6 +4243,22 @@ static void gfx_v10_0_ring_emit_ib_compute(struct amdgpu_ring *ring, 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, @@ -4278,11 +4285,7 @@ static void gfx_v10_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 6)); amdgpu_ring_write(ring, (PACKET3_RELEASE_MEM_GCR_SEQ | PACKET3_RELEASE_MEM_GCR_GL2_WB | - PACKET3_RELEASE_MEM_GCR_GL2_INV | - PACKET3_RELEASE_MEM_GCR_GL2_US | - PACKET3_RELEASE_MEM_GCR_GL1_INV | - PACKET3_RELEASE_MEM_GCR_GLV_INV | - PACKET3_RELEASE_MEM_GCR_GLM_INV | + PACKET3_RELEASE_MEM_GCR_GLM_INV | /* must be set with GLM_WB */ PACKET3_RELEASE_MEM_GCR_GLM_WB | PACKET3_RELEASE_MEM_CACHE_POLICY(3) | PACKET3_RELEASE_MEM_EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) | @@ -4948,7 +4951,7 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_gfx = { 5 + /* HDP_INVL */ 8 + 8 + /* FENCE x2 */ 2, /* SWITCH_BUFFER */ - .emit_ib_size = 4, /* gfx_v10_0_ring_emit_ib_gfx */ + .emit_ib_size = 7, /* gfx_v10_0_ring_emit_ib_gfx */ .emit_ib = gfx_v10_0_ring_emit_ib_gfx, .emit_fence = gfx_v10_0_ring_emit_fence, .emit_pipeline_sync = gfx_v10_0_ring_emit_pipeline_sync, @@ -4987,7 +4990,7 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_compute = { SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 + 2 + /* gfx_v10_0_ring_emit_vm_flush */ 8 + 8 + 8, /* gfx_v10_0_ring_emit_fence x3 for user fence, vm fence */ - .emit_ib_size = 4, /* gfx_v10_0_ring_emit_ib_compute */ + .emit_ib_size = 7, /* gfx_v10_0_ring_emit_ib_compute */ .emit_ib = gfx_v10_0_ring_emit_ib_compute, .emit_fence = gfx_v10_0_ring_emit_fence, .emit_pipeline_sync = gfx_v10_0_ring_emit_pipeline_sync, @@ -5020,7 +5023,7 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_kiq = { SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 + 2 + /* gfx_v10_0_ring_emit_vm_flush */ 8 + 8 + 8, /* gfx_v10_0_ring_emit_fence_kiq x3 for user fence, vm fence */ - .emit_ib_size = 4, /* gfx_v10_0_ring_emit_ib_compute */ + .emit_ib_size = 7, /* gfx_v10_0_ring_emit_ib_compute */ .emit_ib = gfx_v10_0_ring_emit_ib_compute, .emit_fence = gfx_v10_0_ring_emit_fence_kiq, .test_ring = gfx_v10_0_ring_test_ring, @@ -5096,10 +5099,10 @@ static void gfx_v10_0_set_gds_init(struct amdgpu_device *adev) /* init asic gds info */ switch (adev->asic_type) { case CHIP_NAVI10: - adev->gds.gds_size = 0x10000; - break; default: adev->gds.gds_size = 0x10000; + adev->gds.gds_compute_max_wave_id = 0x4ff; + adev->gds.vgt_gs_max_wave_id = 0x3ff; break; } |