diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c | 385 |
1 files changed, 342 insertions, 43 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index 425413fcaf02..c4e14015ec5b 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -1645,6 +1645,147 @@ static u32 gfx_v7_0_get_rb_active_bitmap(struct amdgpu_device *adev) return (~data) & mask; } +static void +gfx_v7_0_raster_config(struct amdgpu_device *adev, u32 *rconf, u32 *rconf1) +{ + switch (adev->asic_type) { + case CHIP_BONAIRE: + *rconf |= RB_MAP_PKR0(2) | RB_XSEL2(1) | SE_MAP(2) | + SE_XSEL(1) | SE_YSEL(1); + *rconf1 |= 0x0; + break; + case CHIP_HAWAII: + *rconf |= RB_MAP_PKR0(2) | RB_MAP_PKR1(2) | + RB_XSEL2(1) | PKR_MAP(2) | PKR_XSEL(1) | + PKR_YSEL(1) | SE_MAP(2) | SE_XSEL(2) | + SE_YSEL(3); + *rconf1 |= SE_PAIR_MAP(2) | SE_PAIR_XSEL(3) | + SE_PAIR_YSEL(2); + break; + case CHIP_KAVERI: + *rconf |= RB_MAP_PKR0(2); + *rconf1 |= 0x0; + break; + case CHIP_KABINI: + case CHIP_MULLINS: + *rconf |= 0x0; + *rconf1 |= 0x0; + break; + default: + DRM_ERROR("unknown asic: 0x%x\n", adev->asic_type); + break; + } +} + +static void +gfx_v7_0_write_harvested_raster_configs(struct amdgpu_device *adev, + u32 raster_config, u32 raster_config_1, + unsigned rb_mask, unsigned num_rb) +{ + unsigned sh_per_se = max_t(unsigned, adev->gfx.config.max_sh_per_se, 1); + unsigned num_se = max_t(unsigned, adev->gfx.config.max_shader_engines, 1); + unsigned rb_per_pkr = min_t(unsigned, num_rb / num_se / sh_per_se, 2); + unsigned rb_per_se = num_rb / num_se; + unsigned se_mask[4]; + unsigned se; + + se_mask[0] = ((1 << rb_per_se) - 1) & rb_mask; + se_mask[1] = (se_mask[0] << rb_per_se) & rb_mask; + se_mask[2] = (se_mask[1] << rb_per_se) & rb_mask; + se_mask[3] = (se_mask[2] << rb_per_se) & rb_mask; + + WARN_ON(!(num_se == 1 || num_se == 2 || num_se == 4)); + WARN_ON(!(sh_per_se == 1 || sh_per_se == 2)); + WARN_ON(!(rb_per_pkr == 1 || rb_per_pkr == 2)); + + if ((num_se > 2) && ((!se_mask[0] && !se_mask[1]) || + (!se_mask[2] && !se_mask[3]))) { + raster_config_1 &= ~SE_PAIR_MAP_MASK; + + if (!se_mask[0] && !se_mask[1]) { + raster_config_1 |= + SE_PAIR_MAP(RASTER_CONFIG_SE_PAIR_MAP_3); + } else { + raster_config_1 |= + SE_PAIR_MAP(RASTER_CONFIG_SE_PAIR_MAP_0); + } + } + + for (se = 0; se < num_se; se++) { + unsigned raster_config_se = raster_config; + unsigned pkr0_mask = ((1 << rb_per_pkr) - 1) << (se * rb_per_se); + unsigned pkr1_mask = pkr0_mask << rb_per_pkr; + int idx = (se / 2) * 2; + + if ((num_se > 1) && (!se_mask[idx] || !se_mask[idx + 1])) { + raster_config_se &= ~SE_MAP_MASK; + + if (!se_mask[idx]) { + raster_config_se |= SE_MAP(RASTER_CONFIG_SE_MAP_3); + } else { + raster_config_se |= SE_MAP(RASTER_CONFIG_SE_MAP_0); + } + } + + pkr0_mask &= rb_mask; + pkr1_mask &= rb_mask; + if (rb_per_se > 2 && (!pkr0_mask || !pkr1_mask)) { + raster_config_se &= ~PKR_MAP_MASK; + + if (!pkr0_mask) { + raster_config_se |= PKR_MAP(RASTER_CONFIG_PKR_MAP_3); + } else { + raster_config_se |= PKR_MAP(RASTER_CONFIG_PKR_MAP_0); + } + } + + if (rb_per_se >= 2) { + unsigned rb0_mask = 1 << (se * rb_per_se); + unsigned rb1_mask = rb0_mask << 1; + + rb0_mask &= rb_mask; + rb1_mask &= rb_mask; + if (!rb0_mask || !rb1_mask) { + raster_config_se &= ~RB_MAP_PKR0_MASK; + + if (!rb0_mask) { + raster_config_se |= + RB_MAP_PKR0(RASTER_CONFIG_RB_MAP_3); + } else { + raster_config_se |= + RB_MAP_PKR0(RASTER_CONFIG_RB_MAP_0); + } + } + + if (rb_per_se > 2) { + rb0_mask = 1 << (se * rb_per_se + rb_per_pkr); + rb1_mask = rb0_mask << 1; + rb0_mask &= rb_mask; + rb1_mask &= rb_mask; + if (!rb0_mask || !rb1_mask) { + raster_config_se &= ~RB_MAP_PKR1_MASK; + + if (!rb0_mask) { + raster_config_se |= + RB_MAP_PKR1(RASTER_CONFIG_RB_MAP_3); + } else { + raster_config_se |= + RB_MAP_PKR1(RASTER_CONFIG_RB_MAP_0); + } + } + } + } + + /* GRBM_GFX_INDEX has a different offset on CI+ */ + gfx_v7_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff); + WREG32(mmPA_SC_RASTER_CONFIG, raster_config_se); + WREG32(mmPA_SC_RASTER_CONFIG_1, raster_config_1); + } + + /* GRBM_GFX_INDEX has a different offset on CI+ */ + gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); +} + /** * gfx_v7_0_setup_rb - setup the RBs on the asic * @@ -1658,9 +1799,11 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev) { int i, j; u32 data; + u32 raster_config = 0, raster_config_1 = 0; u32 active_rbs = 0; u32 rb_bitmap_width_per_sh = adev->gfx.config.max_backends_per_se / adev->gfx.config.max_sh_per_se; + unsigned num_rb_pipes; mutex_lock(&adev->grbm_idx_mutex); for (i = 0; i < adev->gfx.config.max_shader_engines; i++) { @@ -1672,10 +1815,25 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev) } } gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); - mutex_unlock(&adev->grbm_idx_mutex); adev->gfx.config.backend_enable_mask = active_rbs; adev->gfx.config.num_rbs = hweight32(active_rbs); + + num_rb_pipes = min_t(unsigned, adev->gfx.config.max_backends_per_se * + adev->gfx.config.max_shader_engines, 16); + + gfx_v7_0_raster_config(adev, &raster_config, &raster_config_1); + + if (!adev->gfx.config.backend_enable_mask || + adev->gfx.config.num_rbs >= num_rb_pipes) { + WREG32(mmPA_SC_RASTER_CONFIG, raster_config); + WREG32(mmPA_SC_RASTER_CONFIG_1, raster_config_1); + } else { + gfx_v7_0_write_harvested_raster_configs(adev, raster_config, raster_config_1, + adev->gfx.config.backend_enable_mask, + num_rb_pipes); + } + mutex_unlock(&adev->grbm_idx_mutex); } /** @@ -1919,9 +2077,9 @@ static int gfx_v7_0_ring_test_ring(struct amdgpu_ring *ring) static void gfx_v7_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) { u32 ref_and_mask; - int usepfp = ring->type == AMDGPU_RING_TYPE_COMPUTE ? 0 : 1; + int usepfp = ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE ? 0 : 1; - if (ring->type == AMDGPU_RING_TYPE_COMPUTE) { + if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) { switch (ring->me) { case 1: ref_and_mask = GPU_HDP_FLUSH_DONE__CP2_MASK << ring->pipe; @@ -1947,6 +2105,18 @@ static void gfx_v7_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) amdgpu_ring_write(ring, 0x20); /* poll interval */ } +static void gfx_v7_0_ring_emit_vgt_flush(struct amdgpu_ring *ring) +{ + amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE, 0)); + amdgpu_ring_write(ring, EVENT_TYPE(VS_PARTIAL_FLUSH) | + EVENT_INDEX(4)); + + amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE, 0)); + amdgpu_ring_write(ring, EVENT_TYPE(VGT_FLUSH) | + EVENT_INDEX(0)); +} + + /** * gfx_v7_0_ring_emit_hdp_invalidate - emit an hdp invalidate on the cp * @@ -2096,6 +2266,26 @@ static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring, amdgpu_ring_write(ring, control); } +static void gfx_v7_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags) +{ + uint32_t dw2 = 0; + + dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */ + if (flags & AMDGPU_HAVE_CTX_SWITCH) { + gfx_v7_0_ring_emit_vgt_flush(ring); + /* set load_global_config & load_global_uconfig */ + dw2 |= 0x8001; + /* set load_cs_sh_regs */ + dw2 |= 0x01000000; + /* set load_per_context_state & load_gfx_sh_regs */ + dw2 |= 0x10002; + } + + amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1)); + amdgpu_ring_write(ring, dw2); + amdgpu_ring_write(ring, 0); +} + /** * gfx_v7_0_ring_test_ib - basic ring IB test * @@ -2109,7 +2299,7 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) { struct amdgpu_device *adev = ring->adev; struct amdgpu_ib ib; - struct fence *f = NULL; + struct dma_fence *f = NULL; uint32_t scratch; uint32_t tmp = 0; long r; @@ -2135,7 +2325,7 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) if (r) goto err2; - r = fence_wait_timeout(f, false, timeout); + r = dma_fence_wait_timeout(f, false, timeout); if (r == 0) { DRM_ERROR("amdgpu: IB test timed out\n"); r = -ETIMEDOUT; @@ -2156,7 +2346,7 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) err2: amdgpu_ib_free(adev, &ib, NULL); - fence_put(f); + dma_fence_put(f); err1: amdgpu_gfx_scratch_free(adev, scratch); return r; @@ -2443,7 +2633,7 @@ static int gfx_v7_0_cp_gfx_resume(struct amdgpu_device *adev) return 0; } -static u32 gfx_v7_0_ring_get_rptr_gfx(struct amdgpu_ring *ring) +static u32 gfx_v7_0_ring_get_rptr(struct amdgpu_ring *ring) { return ring->adev->wb.wb[ring->rptr_offs]; } @@ -2463,11 +2653,6 @@ static void gfx_v7_0_ring_set_wptr_gfx(struct amdgpu_ring *ring) (void)RREG32(mmCP_RB0_WPTR); } -static u32 gfx_v7_0_ring_get_rptr_compute(struct amdgpu_ring *ring) -{ - return ring->adev->wb.wb[ring->rptr_offs]; -} - static u32 gfx_v7_0_ring_get_wptr_compute(struct amdgpu_ring *ring) { /* XXX check if swapping is necessary on BE */ @@ -3050,7 +3235,7 @@ static int gfx_v7_0_cp_resume(struct amdgpu_device *adev) */ static void gfx_v7_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring) { - int usepfp = (ring->type == AMDGPU_RING_TYPE_GFX); + int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX); uint32_t seq = ring->fence_drv.sync_seq; uint64_t addr = ring->fence_drv.gpu_addr; @@ -3090,7 +3275,7 @@ static void gfx_v7_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring) static void gfx_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring, unsigned vm_id, uint64_t pd_addr) { - int usepfp = (ring->type == AMDGPU_RING_TYPE_GFX); + int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX); amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3)); amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) | @@ -3219,7 +3404,8 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev) if (adev->gfx.rlc.save_restore_obj == NULL) { r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED, + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED | + AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS, NULL, NULL, &adev->gfx.rlc.save_restore_obj); if (r) { @@ -3263,7 +3449,8 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev) if (adev->gfx.rlc.clear_state_obj == NULL) { r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED, + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED | + AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS, NULL, NULL, &adev->gfx.rlc.clear_state_obj); if (r) { @@ -3303,7 +3490,8 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev) if (adev->gfx.rlc.cp_table_obj == NULL) { r = amdgpu_bo_create(adev, adev->gfx.rlc.cp_table_size, PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED, + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED | + AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS, NULL, NULL, &adev->gfx.rlc.cp_table_obj); if (r) { @@ -4182,9 +4370,69 @@ static void gfx_v7_0_ring_emit_gds_switch(struct amdgpu_ring *ring, amdgpu_ring_write(ring, (1 << (oa_size + oa_base)) - (1 << oa_base)); } +static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address) +{ + WREG32(mmSQ_IND_INDEX, + (wave << SQ_IND_INDEX__WAVE_ID__SHIFT) | + (simd << SQ_IND_INDEX__SIMD_ID__SHIFT) | + (address << SQ_IND_INDEX__INDEX__SHIFT) | + (SQ_IND_INDEX__FORCE_READ_MASK)); + return RREG32(mmSQ_IND_DATA); +} + +static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd, + uint32_t wave, uint32_t thread, + uint32_t regno, uint32_t num, uint32_t *out) +{ + WREG32(mmSQ_IND_INDEX, + (wave << SQ_IND_INDEX__WAVE_ID__SHIFT) | + (simd << SQ_IND_INDEX__SIMD_ID__SHIFT) | + (regno << SQ_IND_INDEX__INDEX__SHIFT) | + (thread << SQ_IND_INDEX__THREAD_ID__SHIFT) | + (SQ_IND_INDEX__FORCE_READ_MASK) | + (SQ_IND_INDEX__AUTO_INCR_MASK)); + while (num--) + *(out++) = RREG32(mmSQ_IND_DATA); +} + +static void gfx_v7_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields) +{ + /* type 0 wave data */ + dst[(*no_fields)++] = 0; + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_STATUS); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_LO); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_HI); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_LO); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_HI); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_HW_ID); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW0); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW1); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_GPR_ALLOC); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_LDS_ALLOC); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TRAPSTS); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_STS); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TBA_LO); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TBA_HI); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_LO); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_HI); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_DBG0); + dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_M0); +} + +static void gfx_v7_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd, + uint32_t wave, uint32_t start, + uint32_t size, uint32_t *dst) +{ + wave_read_regs( + adev, simd, wave, 0, + start + SQIND_WAVE_SGPRS_OFFSET, size, dst); +} + static const struct amdgpu_gfx_funcs gfx_v7_0_gfx_funcs = { .get_gpu_clock_counter = &gfx_v7_0_get_gpu_clock_counter, .select_se_sh = &gfx_v7_0_select_se_sh, + .read_wave_data = &gfx_v7_0_read_wave_data, + .read_wave_sgprs = &gfx_v7_0_read_wave_sgprs, }; static const struct amdgpu_rlc_funcs gfx_v7_0_rlc_funcs = { @@ -4436,9 +4684,7 @@ static int gfx_v7_0_sw_init(void *handle) ring->ring_obj = NULL; sprintf(ring->name, "gfx"); r = amdgpu_ring_init(adev, ring, 1024, - PACKET3(PACKET3_NOP, 0x3FFF), 0xf, - &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP, - AMDGPU_RING_TYPE_GFX); + &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP); if (r) return r; } @@ -4463,32 +4709,27 @@ static int gfx_v7_0_sw_init(void *handle) irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP + ring->pipe; /* type-2 packets are deprecated on MEC, use type-3 instead */ r = amdgpu_ring_init(adev, ring, 1024, - PACKET3(PACKET3_NOP, 0x3FFF), 0xf, - &adev->gfx.eop_irq, irq_type, - AMDGPU_RING_TYPE_COMPUTE); + &adev->gfx.eop_irq, irq_type); if (r) return r; } /* reserve GDS, GWS and OA resource for gfx */ - r = amdgpu_bo_create(adev, adev->gds.mem.gfx_partition_size, - PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_GDS, 0, - NULL, NULL, &adev->gds.gds_gfx_bo); + r = amdgpu_bo_create_kernel(adev, adev->gds.mem.gfx_partition_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_GDS, + &adev->gds.gds_gfx_bo, NULL, NULL); if (r) return r; - r = amdgpu_bo_create(adev, adev->gds.gws.gfx_partition_size, - PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_GWS, 0, - NULL, NULL, &adev->gds.gws_gfx_bo); + r = amdgpu_bo_create_kernel(adev, adev->gds.gws.gfx_partition_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_GWS, + &adev->gds.gws_gfx_bo, NULL, NULL); if (r) return r; - r = amdgpu_bo_create(adev, adev->gds.oa.gfx_partition_size, - PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_OA, 0, - NULL, NULL, &adev->gds.oa_gfx_bo); + r = amdgpu_bo_create_kernel(adev, adev->gds.oa.gfx_partition_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_OA, + &adev->gds.oa_gfx_bo, NULL, NULL); if (r) return r; @@ -4504,9 +4745,9 @@ static int gfx_v7_0_sw_fini(void *handle) int i; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - amdgpu_bo_unref(&adev->gds.oa_gfx_bo); - amdgpu_bo_unref(&adev->gds.gws_gfx_bo); - amdgpu_bo_unref(&adev->gds.gds_gfx_bo); + amdgpu_bo_free_kernel(&adev->gds.oa_gfx_bo, NULL, NULL); + amdgpu_bo_free_kernel(&adev->gds.gws_gfx_bo, NULL, NULL); + amdgpu_bo_free_kernel(&adev->gds.gds_gfx_bo, NULL, NULL); for (i = 0; i < adev->gfx.num_gfx_rings; i++) amdgpu_ring_fini(&adev->gfx.gfx_ring[i]); @@ -4919,7 +5160,7 @@ static int gfx_v7_0_set_powergating_state(void *handle, return 0; } -const struct amd_ip_funcs gfx_v7_0_ip_funcs = { +static const struct amd_ip_funcs gfx_v7_0_ip_funcs = { .name = "gfx_v7_0", .early_init = gfx_v7_0_early_init, .late_init = gfx_v7_0_late_init, @@ -4937,10 +5178,21 @@ const struct amd_ip_funcs gfx_v7_0_ip_funcs = { }; static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = { - .get_rptr = gfx_v7_0_ring_get_rptr_gfx, + .type = AMDGPU_RING_TYPE_GFX, + .align_mask = 0xff, + .nop = PACKET3(PACKET3_NOP, 0x3FFF), + .get_rptr = gfx_v7_0_ring_get_rptr, .get_wptr = gfx_v7_0_ring_get_wptr_gfx, .set_wptr = gfx_v7_0_ring_set_wptr_gfx, - .parse_cs = NULL, + .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 */ + 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 */ + 17 + 6 + /* gfx_v7_0_ring_emit_vm_flush */ + 3 + 4, /* gfx_v7_ring_emit_cntxcntl including vgt flush*/ + .emit_ib_size = 4, /* gfx_v7_0_ring_emit_ib_gfx */ .emit_ib = gfx_v7_0_ring_emit_ib_gfx, .emit_fence = gfx_v7_0_ring_emit_fence_gfx, .emit_pipeline_sync = gfx_v7_0_ring_emit_pipeline_sync, @@ -4952,13 +5204,24 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = { .test_ib = gfx_v7_0_ring_test_ib, .insert_nop = amdgpu_ring_insert_nop, .pad_ib = amdgpu_ring_generic_pad_ib, + .emit_cntxcntl = gfx_v7_ring_emit_cntxcntl, }; static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = { - .get_rptr = gfx_v7_0_ring_get_rptr_compute, + .type = AMDGPU_RING_TYPE_COMPUTE, + .align_mask = 0xff, + .nop = PACKET3(PACKET3_NOP, 0x3FFF), + .get_rptr = gfx_v7_0_ring_get_rptr, .get_wptr = gfx_v7_0_ring_get_wptr_compute, .set_wptr = gfx_v7_0_ring_set_wptr_compute, - .parse_cs = NULL, + .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 */ + 7 + /* gfx_v7_0_ring_emit_pipeline_sync */ + 17 + /* gfx_v7_0_ring_emit_vm_flush */ + 7 + 7 + 7, /* gfx_v7_0_ring_emit_fence_compute x3 for user fence, vm fence */ + .emit_ib_size = 4, /* gfx_v7_0_ring_emit_ib_compute */ .emit_ib = gfx_v7_0_ring_emit_ib_compute, .emit_fence = gfx_v7_0_ring_emit_fence_compute, .emit_pipeline_sync = gfx_v7_0_ring_emit_pipeline_sync, @@ -5080,3 +5343,39 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev) cu_info->number = active_cu_number; cu_info->ao_cu_mask = ao_cu_mask; } + +const struct amdgpu_ip_block_version gfx_v7_0_ip_block = +{ + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &gfx_v7_0_ip_funcs, +}; + +const struct amdgpu_ip_block_version gfx_v7_1_ip_block = +{ + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 7, + .minor = 1, + .rev = 0, + .funcs = &gfx_v7_0_ip_funcs, +}; + +const struct amdgpu_ip_block_version gfx_v7_2_ip_block = +{ + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 7, + .minor = 2, + .rev = 0, + .funcs = &gfx_v7_0_ip_funcs, +}; + +const struct amdgpu_ip_block_version gfx_v7_3_ip_block = +{ + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 7, + .minor = 3, + .rev = 0, + .funcs = &gfx_v7_0_ip_funcs, +}; |