From: Jason Ekstrand Date: Fri, 29 May 2015 16:40:03 +0000 (-0700) Subject: vk/device: Allow for starting a new surface state buffer X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=4ffbab5ae06531d088404245e860822b758970e3;p=mesa.git vk/device: Allow for starting a new surface state buffer This commit allows for us to create a whole new surface state buffer when the old one runs out of room. We simply re-emit the state base address for the new state, re-emit binding tables, and keep going. --- diff --git a/src/vulkan/aub.c b/src/vulkan/aub.c index dfe52213a40..e3f333a54aa 100644 --- a/src/vulkan/aub.c +++ b/src/vulkan/aub.c @@ -273,9 +273,14 @@ anv_cmd_buffer_dump(struct anv_cmd_buffer *cmd_buffer) } assert(first_bbo->prev_batch_bo == NULL); - relocate_bo(&cmd_buffer->surface_bo, - cmd_buffer->surface_relocs.relocs, - cmd_buffer->surface_relocs.num_relocs, aub_bos); + for (struct anv_batch_bo *bbo = cmd_buffer->surface_batch_bo; + bbo != NULL; bbo = bbo->prev_batch_bo) { + + /* Handle relocations for this surface state BO */ + relocate_bo(&bbo->bo, + &cmd_buffer->surface_relocs.relocs[bbo->first_reloc], + bbo->num_relocs, aub_bos); + } for (uint32_t i = 0; i < cmd_buffer->bo_count; i++) { bo = cmd_buffer->exec2_bos[i]; diff --git a/src/vulkan/device.c b/src/vulkan/device.c index 54e70d3a5de..27165e229b6 100644 --- a/src/vulkan/device.c +++ b/src/vulkan/device.c @@ -2285,16 +2285,23 @@ anv_cmd_buffer_destroy(struct anv_device *device, struct anv_batch_bo *bbo = cmd_buffer->last_batch_bo; while (bbo->prev_batch_bo) { struct anv_batch_bo *prev = bbo->prev_batch_bo; - anv_batch_bo_destroy(bbo, cmd_buffer->device); + anv_batch_bo_destroy(bbo, device); bbo = prev; } + anv_reloc_list_finish(&cmd_buffer->batch.relocs, device); - anv_bo_pool_free(&device->batch_bo_pool, &cmd_buffer->surface_bo); + /* Destroy all of the surface state buffers */ + bbo = cmd_buffer->surface_batch_bo; + while (bbo->prev_batch_bo) { + struct anv_batch_bo *prev = bbo->prev_batch_bo; + anv_batch_bo_destroy(bbo, device); + bbo = prev; + } anv_reloc_list_finish(&cmd_buffer->surface_relocs, device); + anv_state_stream_finish(&cmd_buffer->surface_state_stream); anv_state_stream_finish(&cmd_buffer->dynamic_state_stream); anv_state_stream_finish(&cmd_buffer->binding_table_state_stream); - anv_reloc_list_finish(&cmd_buffer->batch.relocs, device); anv_device_free(device, cmd_buffer->exec2_objects); anv_device_free(device, cmd_buffer->exec2_bos); anv_device_free(device, cmd_buffer); @@ -2377,13 +2384,17 @@ VkResult anv_CreateCommandBuffer( anv_batch_bo_start(cmd_buffer->last_batch_bo, &cmd_buffer->batch, GEN8_MI_BATCH_BUFFER_START_length * 4); - result = anv_bo_pool_alloc(&device->batch_bo_pool, &cmd_buffer->surface_bo); + result = anv_batch_bo_create(device, &cmd_buffer->surface_batch_bo); if (result != VK_SUCCESS) goto fail_batch_relocs; + cmd_buffer->surface_batch_bo->first_reloc = 0; + + result = anv_reloc_list_init(&cmd_buffer->surface_relocs, device); + if (result != VK_SUCCESS) + goto fail_ss_batch_bo; /* Start surface_next at 1 so surface offset 0 is invalid. */ cmd_buffer->surface_next = 1; - anv_reloc_list_init(&cmd_buffer->surface_relocs, device); cmd_buffer->exec2_objects = NULL; cmd_buffer->exec2_bos = NULL; @@ -2407,6 +2418,8 @@ VkResult anv_CreateCommandBuffer( return VK_SUCCESS; + fail_ss_batch_bo: + anv_batch_bo_destroy(cmd_buffer->surface_batch_bo, device); fail_batch_relocs: anv_reloc_list_finish(&cmd_buffer->batch.relocs, device); fail_batch_bo: @@ -2429,7 +2442,7 @@ anv_cmd_buffer_emit_state_base_address(struct anv_cmd_buffer *cmd_buffer) .GeneralStateBufferSize = 0xfffff, .GeneralStateBufferSizeModifyEnable = true, - .SurfaceStateBaseAddress = { &cmd_buffer->surface_bo, 0 }, + .SurfaceStateBaseAddress = { &cmd_buffer->surface_batch_bo->bo, 0 }, .SurfaceStateMemoryObjectControlState = GEN8_MOCS, .SurfaceStateBaseAddressModifyEnable = true, @@ -2599,6 +2612,9 @@ VkResult anv_EndCommandBuffer( anv_batch_emit(batch, GEN8_MI_NOOP); anv_batch_bo_finish(cmd_buffer->last_batch_bo, &cmd_buffer->batch); + cmd_buffer->surface_batch_bo->num_relocs = + cmd_buffer->surface_relocs.num_relocs - cmd_buffer->surface_batch_bo->first_reloc; + cmd_buffer->surface_batch_bo->length = cmd_buffer->surface_next; cmd_buffer->bo_count = 0; cmd_buffer->need_reloc = false; @@ -2606,10 +2622,13 @@ VkResult anv_EndCommandBuffer( /* Lock for access to bo->index. */ pthread_mutex_lock(&device->mutex); - /* Add block pool bos first so we can add them with their relocs. */ - anv_cmd_buffer_add_bo(cmd_buffer, &cmd_buffer->surface_bo, - cmd_buffer->surface_relocs.relocs, - cmd_buffer->surface_relocs.num_relocs); + /* Add surface state bos first so we can add them with their relocs. */ + for (struct anv_batch_bo *bbo = cmd_buffer->surface_batch_bo; + bbo != NULL; bbo = bbo->prev_batch_bo) { + anv_cmd_buffer_add_bo(cmd_buffer, &bbo->bo, + &cmd_buffer->surface_relocs.relocs[bbo->first_reloc], + bbo->num_relocs); + } /* Add all of the BOs referenced by surface state */ anv_cmd_buffer_add_validate_bos(cmd_buffer, &cmd_buffer->surface_relocs); @@ -2674,7 +2693,15 @@ VkResult anv_ResetCommandBuffer( anv_batch_bo_start(cmd_buffer->last_batch_bo, &cmd_buffer->batch, GEN8_MI_BATCH_BUFFER_START_length * 4); - cmd_buffer->surface_next = 0; + /* Delete all but the first batch bo */ + while (cmd_buffer->surface_batch_bo->prev_batch_bo) { + struct anv_batch_bo *prev = cmd_buffer->surface_batch_bo->prev_batch_bo; + anv_batch_bo_destroy(cmd_buffer->surface_batch_bo, cmd_buffer->device); + cmd_buffer->surface_batch_bo = prev; + } + assert(cmd_buffer->surface_batch_bo->prev_batch_bo == NULL); + + cmd_buffer->surface_next = 1; cmd_buffer->surface_relocs.num_relocs = 0; return VK_SUCCESS; @@ -2740,15 +2767,46 @@ anv_cmd_buffer_alloc_surface_state(struct anv_cmd_buffer *cmd_buffer, struct anv_state state; state.offset = ALIGN_U32(cmd_buffer->surface_next, alignment); - state.map = cmd_buffer->surface_bo.map + state.offset; + if (state.offset + size > cmd_buffer->surface_batch_bo->bo.size) + return (struct anv_state) { 0 }; + + state.map = cmd_buffer->surface_batch_bo->bo.map + state.offset; state.alloc_size = size; cmd_buffer->surface_next = state.offset + size; - assert(state.offset + size < cmd_buffer->surface_bo.size); + assert(state.offset + size <= cmd_buffer->surface_batch_bo->bo.size); return state; } +static VkResult +anv_cmd_buffer_new_surface_state_bo(struct anv_cmd_buffer *cmd_buffer) +{ + struct anv_batch_bo *new_bbo, *old_bbo = cmd_buffer->surface_batch_bo; + + /* Finish off the old buffer */ + old_bbo->num_relocs = + cmd_buffer->surface_relocs.num_relocs - old_bbo->first_reloc; + old_bbo->length = cmd_buffer->surface_next; + + VkResult result = anv_batch_bo_create(cmd_buffer->device, &new_bbo); + if (result != VK_SUCCESS) + return result; + + new_bbo->first_reloc = cmd_buffer->surface_relocs.num_relocs; + cmd_buffer->surface_next = 1; + + new_bbo->prev_batch_bo = old_bbo; + cmd_buffer->surface_batch_bo = new_bbo; + + /* Re-emit state base addresses so we get the new surface state base + * address before we start emitting binding tables etc. + */ + anv_cmd_buffer_emit_state_base_address(cmd_buffer); + + return VK_SUCCESS; +} + void anv_CmdBindDescriptorSets( VkCmdBuffer cmdBuffer, VkPipelineBindPoint pipelineBindPoint, @@ -2827,7 +2885,7 @@ void anv_CmdBindVertexBuffers( } } -static void +static VkResult cmd_buffer_emit_binding_table(struct anv_cmd_buffer *cmd_buffer, unsigned stage) { @@ -2849,12 +2907,15 @@ cmd_buffer_emit_binding_table(struct anv_cmd_buffer *cmd_buffer, uint32_t surface_count = layout ? layout->stage[stage].surface_count : 0; if (color_attachments + surface_count == 0) - return; + return VK_SUCCESS; size = (bias + surface_count) * sizeof(uint32_t); bt_state = anv_cmd_buffer_alloc_surface_state(cmd_buffer, size, 32); uint32_t *bt_map = bt_state.map; + if (bt_state.map == NULL) + return VK_ERROR_OUT_OF_DEVICE_MEMORY; + static const uint32_t binding_table_opcodes[] = { [VK_SHADER_STAGE_VERTEX] = 38, [VK_SHADER_STAGE_TESS_CONTROL] = 39, @@ -2876,6 +2937,9 @@ cmd_buffer_emit_binding_table(struct anv_cmd_buffer *cmd_buffer, struct anv_state state = anv_cmd_buffer_alloc_surface_state(cmd_buffer, 64, 64); + if (state.map == NULL) + return VK_ERROR_OUT_OF_DEVICE_MEMORY; + memcpy(state.map, view->surface_state.map, 64); /* The address goes in dwords 8 and 9 of the SURFACE_STATE */ @@ -2889,7 +2953,7 @@ cmd_buffer_emit_binding_table(struct anv_cmd_buffer *cmd_buffer, } if (layout == NULL) - return; + return VK_SUCCESS; for (uint32_t set = 0; set < layout->num_sets; set++) { struct anv_descriptor_set_binding *d = &cmd_buffer->descriptors[set]; @@ -2906,6 +2970,9 @@ cmd_buffer_emit_binding_table(struct anv_cmd_buffer *cmd_buffer, struct anv_state state = anv_cmd_buffer_alloc_surface_state(cmd_buffer, 64, 64); + if (state.map == NULL) + return VK_ERROR_OUT_OF_DEVICE_MEMORY; + uint32_t offset; if (surface_slots[b].dynamic_slot >= 0) { uint32_t dynamic_offset = @@ -2929,25 +2996,30 @@ cmd_buffer_emit_binding_table(struct anv_cmd_buffer *cmd_buffer, bt_map[start + b] = state.offset; } } + + return VK_SUCCESS; } -static void +static VkResult cmd_buffer_emit_samplers(struct anv_cmd_buffer *cmd_buffer, unsigned stage) { struct anv_pipeline_layout *layout = cmd_buffer->pipeline->layout; struct anv_state state; if (!layout) - return; + return VK_SUCCESS; uint32_t sampler_count = layout->stage[stage].sampler_count; if (sampler_count == 0) - return; + return VK_SUCCESS; uint32_t size = sampler_count * 16; state = anv_state_stream_alloc(&cmd_buffer->dynamic_state_stream, size, 32); + if (state.map == NULL) + return VK_ERROR_OUT_OF_DEVICE_MEMORY; + static const uint32_t sampler_state_opcodes[] = { [VK_SHADER_STAGE_VERTEX] = 43, [VK_SHADER_STAGE_TESS_CONTROL] = 44, /* HS */ @@ -2981,14 +3053,37 @@ cmd_buffer_emit_samplers(struct anv_cmd_buffer *cmd_buffer, unsigned stage) sampler->state, sizeof(sampler->state)); } } + + return VK_SUCCESS; } static void flush_descriptor_sets(struct anv_cmd_buffer *cmd_buffer) { + VkResult result; for (uint32_t s = 0; s < VK_NUM_SHADER_STAGE; s++) { - cmd_buffer_emit_binding_table(cmd_buffer, s); - cmd_buffer_emit_samplers(cmd_buffer, s); + result = cmd_buffer_emit_binding_table(cmd_buffer, s); + if (result != VK_SUCCESS) + break; + + result = cmd_buffer_emit_samplers(cmd_buffer, s); + if (result != VK_SUCCESS) + break; + } + + if (result != VK_SUCCESS) { + assert(result == VK_ERROR_OUT_OF_DEVICE_MEMORY); + + result = anv_cmd_buffer_new_surface_state_bo(cmd_buffer); + assert(result == VK_SUCCESS); + + for (uint32_t s = 0; s < VK_NUM_SHADER_STAGE; s++) { + result = cmd_buffer_emit_binding_table(cmd_buffer, s); + result = cmd_buffer_emit_samplers(cmd_buffer, s); + } + + /* It had better succeed this time */ + assert(result == VK_SUCCESS); } cmd_buffer->dirty &= ~ANV_CMD_BUFFER_DESCRIPTOR_SET_DIRTY; diff --git a/src/vulkan/private.h b/src/vulkan/private.h index 72a10e5c0a8..153cb22d245 100644 --- a/src/vulkan/private.h +++ b/src/vulkan/private.h @@ -609,7 +609,7 @@ struct anv_vertex_binding { struct anv_descriptor_set_binding { struct anv_descriptor_set * set; - uint32_t dynamic_offsets[256]; + uint32_t dynamic_offsets[128]; }; struct anv_cmd_buffer { @@ -626,7 +626,7 @@ struct anv_cmd_buffer { uint32_t bo_count; struct anv_batch batch; struct anv_batch_bo * last_batch_bo; - struct anv_bo surface_bo; + struct anv_batch_bo * surface_batch_bo; uint32_t surface_next; struct anv_reloc_list surface_relocs; struct anv_state_stream binding_table_state_stream;