/* * Copyright © 2016 Red Hat. * Copyright © 2016 Bas Nieuwenhuizen * SPDX-License-Identifier: MIT * * based in part on anv driver which is: * Copyright © 2015 Intel Corporation */ #include "tu_cmd_buffer.h" #include "vk_common_entrypoints.h" #include "vk_log.h" #include "vk_render_pass.h" #include "vk_util.h" #include "tu_buffer.h" #include "tu_clear_blit.h" #include "tu_cs.h" #include "tu_event.h" #include "tu_image.h" #include "tu_knl.h" #include "tu_tracepoints.h" #include "common/freedreno_gpu_event.h" #include "common/freedreno_lrz.h" #include "common/freedreno_vrs.h" enum tu_cmd_buffer_status { TU_CMD_BUFFER_STATUS_IDLE = 0, TU_CMD_BUFFER_STATUS_ACTIVE = 1, }; static struct tu_bo * tu_cmd_buffer_setup_status_tracking(struct tu_device *device) { struct tu_bo *status_bo; VkResult result; result = tu_bo_init_new_explicit_iova( device, NULL, &status_bo, sizeof(enum tu_cmd_buffer_status), 0, VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT | VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT, TU_BO_ALLOC_INTERNAL_RESOURCE, NULL, "cmd_buffer_status"); if (result != VK_SUCCESS) return NULL; result = tu_bo_map(device, status_bo, NULL); if (result != VK_SUCCESS) return NULL; return status_bo; } static VkResult tu_cmd_buffer_status_check_idle(struct tu_cmd_buffer *cmd_buffer) { if (cmd_buffer->status_bo == NULL) return VK_SUCCESS; const enum tu_cmd_buffer_status status = *(enum tu_cmd_buffer_status *)cmd_buffer->status_bo->map; switch (status) { case TU_CMD_BUFFER_STATUS_IDLE: return VK_SUCCESS; case TU_CMD_BUFFER_STATUS_ACTIVE: mesa_loge("Trying to reset or destroy cmd_buffer %p while in use", cmd_buffer); return vk_errorf(cmd_buffer, VK_ERROR_UNKNOWN, "Trying to reset or destroy while being used"); default: mesa_loge("Something went wrong with cmd_buffer status tracking"); return vk_error(cmd_buffer, VK_ERROR_UNKNOWN); } } static inline void tu_cmd_buffer_status_gpu_write(struct tu_cmd_buffer *cmd_buffer, enum tu_cmd_buffer_status status) { struct tu_cs *cs = &cmd_buffer->cs; if (cmd_buffer->status_bo == NULL) return; static_assert(sizeof(uint32_t) == sizeof(status), "Code below needs adjusting"); tu_cs_emit_pkt7(cs, CP_MEM_WRITE, 3); tu_cs_emit_qw(cs, cmd_buffer->status_bo->iova); tu_cs_emit(cs, (uint32_t)status); } static void tu_clone_trace_range(struct tu_cmd_buffer *cmd, struct tu_cs *cs, struct u_trace *dst, struct u_trace_iterator begin, struct u_trace_iterator end) { if (u_trace_iterator_equal(begin, end)) return; tu_cs_emit_wfi(cs); tu_cs_emit_pkt7(cs, CP_WAIT_FOR_ME, 0); u_trace_clone_append(begin, end, &cmd->trace, cs, tu_copy_buffer); } static void tu_clone_trace(struct tu_cmd_buffer *cmd, struct tu_cs *cs, struct u_trace *dst, struct u_trace *src) { tu_clone_trace_range(cmd, cs, dst, u_trace_begin_iterator(src), u_trace_end_iterator(src)); } template void tu_emit_raw_event_write(struct tu_cmd_buffer *cmd, struct tu_cs *cs, enum vgt_event_type event, bool needs_seqno) { if (CHIP == A6XX) { tu_cs_emit_pkt7(cs, CP_EVENT_WRITE, needs_seqno ? 4 : 1); tu_cs_emit(cs, CP_EVENT_WRITE_0_EVENT(event)); } else { tu_cs_emit_pkt7(cs, CP_EVENT_WRITE7, needs_seqno ? 4 : 1); tu_cs_emit(cs, CP_EVENT_WRITE7_0(.event = event, .write_src = EV_WRITE_USER_32B, .write_dst = EV_DST_RAM, .write_enabled = needs_seqno).value); } if (needs_seqno) { tu_cs_emit_qw(cs, global_iova(cmd, seqno_dummy)); tu_cs_emit(cs, 0); } } TU_GENX(tu_emit_raw_event_write); template void tu_emit_event_write(struct tu_cmd_buffer *cmd, struct tu_cs *cs, enum fd_gpu_event event) { struct fd_gpu_event_info event_info = fd_gpu_events[event]; tu_emit_raw_event_write(cmd, cs, event_info.raw_event, event_info.needs_seqno); } TU_GENX(tu_emit_event_write); /* Emits the tessfactor address to the top-level CS if it hasn't been already. * Updating this register requires a WFI if outstanding drawing is using it, but * tu6_init_hardware() will have WFIed before we started and no other draws * could be using the tessfactor address yet since we only emit one per cmdbuf. */ template static void tu6_lazy_emit_tessfactor_addr(struct tu_cmd_buffer *cmd) { if (cmd->state.tessfactor_addr_set) return; tu_cs_emit_regs(&cmd->cs, PC_TESS_BASE(CHIP, .qword = cmd->device->tess_bo->iova)); /* Updating PC_TESS_BASE could race with the next draw which uses it. */ cmd->state.cache.flush_bits |= TU_CMD_FLAG_WAIT_FOR_IDLE; cmd->state.tessfactor_addr_set = true; } static void tu6_lazy_init_vsc(struct tu_cmd_buffer *cmd) { struct tu_device *dev = cmd->device; uint32_t num_vsc_pipes = dev->physical_device->info->num_vsc_pipes; /* VSC buffers: * use vsc pitches from the largest values used so far with this device * if there hasn't been overflow, there will already be a scratch bo * allocated for these sizes * * if overflow is detected, the stream size is increased by 2x */ mtx_lock(&dev->mutex); struct tu6_global *global = dev->global_bo_map; uint32_t vsc_draw_overflow = global->vsc_draw_overflow; uint32_t vsc_prim_overflow = global->vsc_prim_overflow; if (vsc_draw_overflow >= dev->vsc_draw_strm_pitch) dev->vsc_draw_strm_pitch = (dev->vsc_draw_strm_pitch - VSC_PAD) * 2 + VSC_PAD; if (vsc_prim_overflow >= dev->vsc_prim_strm_pitch) dev->vsc_prim_strm_pitch = (dev->vsc_prim_strm_pitch - VSC_PAD) * 2 + VSC_PAD; cmd->vsc_prim_strm_pitch = dev->vsc_prim_strm_pitch; cmd->vsc_draw_strm_pitch = dev->vsc_draw_strm_pitch; mtx_unlock(&dev->mutex); uint32_t prim_strm_size = cmd->vsc_prim_strm_pitch * num_vsc_pipes; uint32_t draw_strm_size = cmd->vsc_draw_strm_pitch * num_vsc_pipes; uint32_t draw_strm_size_size = 4 * num_vsc_pipes; uint32_t state_size = 4 * num_vsc_pipes; cmd->vsc_size = prim_strm_size + draw_strm_size + draw_strm_size_size + state_size; cmd->vsc_prim_strm_offset = 0; cmd->vsc_draw_strm_offset = prim_strm_size; cmd->vsc_draw_strm_size_offset = cmd->vsc_draw_strm_offset + draw_strm_size; cmd->vsc_state_offset = cmd->vsc_draw_strm_size_offset + draw_strm_size_size; } static void tu_emit_vis_stream_patchpoint(struct tu_cmd_buffer *cmd, struct tu_cs *cs, uint32_t offset) { struct tu_vis_stream_patchpoint patchpoint = { .render_pass_idx = cmd->state.tile_render_pass_count, .data = cs->cur, .iova = tu_cs_get_cur_iova(cs), .offset = offset, }; util_dynarray_append(&cmd->vis_stream_patchpoints, patchpoint); tu_cs_emit_qw(cs, offset); } template static void tu_emit_vsc(struct tu_cmd_buffer *cmd, struct tu_cs *cs) { if (CHIP == A6XX) { tu_cs_emit_pkt4(cs, REG_A6XX_VSC_SIZE_BASE, 2); tu_emit_vis_stream_patchpoint(cmd, cs, cmd->vsc_draw_strm_size_offset); tu_cs_emit_pkt4(cs, REG_A6XX_VSC_PIPE_DATA_PRIM_BASE, 2); tu_emit_vis_stream_patchpoint(cmd, cs, cmd->vsc_prim_strm_offset); tu_cs_emit_pkt4(cs, REG_A6XX_VSC_PIPE_DATA_DRAW_BASE, 2); tu_emit_vis_stream_patchpoint(cmd, cs, cmd->vsc_draw_strm_offset); } else { tu_cs_emit_pkt7(cs, CP_SET_PSEUDO_REG, 3 * 3); tu_cs_emit(cs, A6XX_CP_SET_PSEUDO_REG__0_PSEUDO_REG(VSC_PIPE_DATA_DRAW_BASE)); tu_emit_vis_stream_patchpoint(cmd, cs, cmd->vsc_draw_strm_offset); tu_cs_emit(cs, A6XX_CP_SET_PSEUDO_REG__0_PSEUDO_REG(VSC_SIZE_BASE)); tu_emit_vis_stream_patchpoint(cmd, cs, cmd->vsc_draw_strm_size_offset); tu_cs_emit(cs, A6XX_CP_SET_PSEUDO_REG__0_PSEUDO_REG(VSC_PIPE_DATA_PRIM_BASE)); tu_emit_vis_stream_patchpoint(cmd, cs, cmd->vsc_prim_strm_offset); } cmd->vsc_initialized = true; } /* This workaround, copied from the blob, seems to ensure that the BVH node * cache is invalidated so that we don't read stale values when multiple BVHs * share the same address. */ static void tu_emit_rt_workaround(struct tu_cmd_buffer *cmd, struct tu_cs *cs) { tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_RT_WA_START); tu_cs_emit_regs(cs, A7XX_SP_CS_UNKNOWN_A9BE(.dword = 0x10000)); tu_cs_emit_regs(cs, A7XX_SP_PS_UNKNOWN_A9AB(.dword = 0x10000)); tu_emit_event_write(cmd, cs, FD_DUMMY_EVENT); tu_cs_emit_regs(cs, A7XX_SP_CS_UNKNOWN_A9BE(.dword = 0)); tu_cs_emit_regs(cs, A7XX_SP_PS_UNKNOWN_A9AB(.dword = 0)); tu_emit_event_write(cmd, cs, FD_DUMMY_EVENT); tu_emit_event_write(cmd, cs, FD_DUMMY_EVENT); tu_emit_event_write(cmd, cs, FD_DUMMY_EVENT); tu_emit_event_write(cmd, cs, FD_DUMMY_EVENT); tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_RT_WA_END); } template static void tu6_emit_flushes(struct tu_cmd_buffer *cmd_buffer, struct tu_cs *cs, struct tu_cache_state *cache) { BITMASK_ENUM(tu_cmd_flush_bits) flushes = cache->flush_bits; cache->flush_bits = 0; if (TU_DEBUG(FLUSHALL)) flushes |= TU_CMD_FLAG_ALL_CLEAN | TU_CMD_FLAG_ALL_INVALIDATE; if (TU_DEBUG(SYNCDRAW)) flushes |= TU_CMD_FLAG_WAIT_MEM_WRITES | TU_CMD_FLAG_WAIT_FOR_IDLE | TU_CMD_FLAG_WAIT_FOR_ME; /* Experiments show that invalidating CCU while it still has data in it * doesn't work, so make sure to always flush before invalidating in case * any data remains that hasn't yet been made available through a barrier. * However it does seem to work for UCHE. */ if (flushes & (TU_CMD_FLAG_CCU_CLEAN_COLOR | TU_CMD_FLAG_CCU_INVALIDATE_COLOR)) tu_emit_event_write(cmd_buffer, cs, FD_CCU_CLEAN_COLOR); if (flushes & (TU_CMD_FLAG_CCU_CLEAN_DEPTH | TU_CMD_FLAG_CCU_INVALIDATE_DEPTH)) tu_emit_event_write(cmd_buffer, cs, FD_CCU_CLEAN_DEPTH); if (flushes & TU_CMD_FLAG_CCU_INVALIDATE_COLOR) tu_emit_event_write(cmd_buffer, cs, FD_CCU_INVALIDATE_COLOR); if (flushes & TU_CMD_FLAG_CCU_INVALIDATE_DEPTH) tu_emit_event_write(cmd_buffer, cs, FD_CCU_INVALIDATE_DEPTH); if (flushes & TU_CMD_FLAG_CACHE_CLEAN) tu_emit_event_write(cmd_buffer, cs, FD_CACHE_CLEAN); if (flushes & TU_CMD_FLAG_CACHE_INVALIDATE) tu_emit_event_write(cmd_buffer, cs, FD_CACHE_INVALIDATE); if (flushes & TU_CMD_FLAG_BINDLESS_DESCRIPTOR_INVALIDATE) { tu_cs_emit_regs(cs, SP_UPDATE_CNTL(CHIP, .cs_bindless = CHIP == A6XX ? 0x1f : 0xff, .gfx_bindless = CHIP == A6XX ? 0x1f : 0xff, )); } if (CHIP >= A7XX && flushes & TU_CMD_FLAG_BLIT_CACHE_CLEAN) /* On A7XX, blit cache flushes are required to ensure blit writes are visible * via UCHE. This isn't necessary on A6XX, all writes should be visible implictly. */ tu_emit_event_write(cmd_buffer, cs, FD_CCU_CLEAN_BLIT_CACHE); if (CHIP >= A7XX && (flushes & TU_CMD_FLAG_CCHE_INVALIDATE) && /* Invalidating UCHE seems to also invalidate CCHE */ !(flushes & TU_CMD_FLAG_CACHE_INVALIDATE)) tu_cs_emit_pkt7(cs, CP_CCHE_INVALIDATE, 0); if (CHIP >= A7XX && (flushes & TU_CMD_FLAG_RTU_INVALIDATE) && cmd_buffer->device->physical_device->info->a7xx.has_rt_workaround) tu_emit_rt_workaround(cmd_buffer, cs); if (flushes & TU_CMD_FLAG_WAIT_MEM_WRITES) tu_cs_emit_pkt7(cs, CP_WAIT_MEM_WRITES, 0); if (flushes & TU_CMD_FLAG_WAIT_FOR_IDLE) tu_cs_emit_wfi(cs); if (flushes & TU_CMD_FLAG_WAIT_FOR_ME) tu_cs_emit_pkt7(cs, CP_WAIT_FOR_ME, 0); } static void tu7_write_onchip_val(struct tu_cs *cs, enum tu_onchip_addr addr, uint32_t val) { tu_cs_emit_pkt7(cs, CP_EVENT_WRITE7, 4); tu_cs_emit(cs, CP_EVENT_WRITE7_0_WRITE_DST(EV_DST_ONCHIP) | CP_EVENT_WRITE7_0_WRITE_SRC(EV_WRITE_USER_32B) | CP_EVENT_WRITE7_0_EVENT(DUMMY_EVENT) | CP_EVENT_WRITE7_0_WRITE_ENABLED); tu_cs_emit_qw(cs, addr); tu_cs_emit(cs, val); } /* "Normal" cache flushes outside the renderpass, that don't require any special handling */ template void tu_emit_cache_flush(struct tu_cmd_buffer *cmd_buffer) { struct tu_cs *cs = &cmd_buffer->cs; struct tu_cache_state *cache = &cmd_buffer->state.cache; BITMASK_ENUM(tu_cmd_flush_bits) flushes = cache->flush_bits; tu6_emit_flushes(cmd_buffer, cs, cache); if ((flushes & TU_CMD_FLAG_WAIT_FOR_BR) && CHIP >= A7XX) { tu_cs_emit_pkt7(cs, CP_THREAD_CONTROL, 1); tu_cs_emit(cs, CP_THREAD_CONTROL_0_THREAD(CP_SET_THREAD_BOTH)); tu_cs_emit_pkt7(cs, CP_MODIFY_TIMESTAMP, 1); tu_cs_emit(cs, CP_MODIFY_TIMESTAMP_0_ADD(1) | CP_MODIFY_TIMESTAMP_0_OP(MODIFY_TIMESTAMP_ADD_LOCAL)); tu7_thread_control(cs, CP_SET_THREAD_BV); tu7_write_onchip_timestamp(cs, TU_ONCHIP_CB_BV_TIMESTAMP); tu7_thread_control(cs, CP_SET_THREAD_BR); /* Wait for the previous WAIT_FOR_BR to execute on BV and reset the wait * value. */ tu7_wait_onchip_timestamp(cs, TU_ONCHIP_CB_BV_TIMESTAMP); /* Signal the wait value. */ tu7_write_onchip_val(cs, TU_ONCHIP_BARRIER, 1); tu7_thread_control(cs, CP_SET_THREAD_BV); /* Wait for the value. Note that we must use CP_WAIT_REG_MEM due to a * firmware bug which makes CP_WAIT_TIMESTAMP on BV deadlock with * preemption when BV waits for BR. Without this bug the whole thing * would be much, much simpler. */ tu7_wait_onchip_val(cs, TU_ONCHIP_BARRIER, 1); /* Reset the wait value. */ tu7_write_onchip_val(cs, TU_ONCHIP_BARRIER, 0); /* Resetting the wait value happens asynchronously (since it's an * EVENT_WRITE), but waiting for it happens synchronously. We need to * prevent BV from racing ahead to the next wait before it's reset. */ tu7_wait_onchip_val(cs, TU_ONCHIP_BARRIER, 0); tu7_thread_control(cs, CP_SET_THREAD_BR); } } TU_GENX(tu_emit_cache_flush); /* Renderpass cache flushes inside the draw_cs */ template void tu_emit_cache_flush_renderpass(struct tu_cmd_buffer *cmd_buffer) { if (!cmd_buffer->state.renderpass_cache.flush_bits && likely(!tu_env.debug)) return; struct tu_cs *cs = &cmd_buffer->draw_cs; struct tu_cache_state *cache = &cmd_buffer->state.renderpass_cache; tu6_emit_flushes(cmd_buffer, cs, cache); if (cmd_buffer->state.renderpass_cache.flush_bits & TU_CMD_FLAG_BLIT_CACHE_CLEAN) { cmd_buffer->state.blit_cache_cleaned = true; } } TU_GENX(tu_emit_cache_flush_renderpass); template static void emit_vpc_attr_buf(struct tu_cs *cs, struct tu_device *dev, bool gmem) { if (!dev->physical_device->info->a7xx.has_gmem_vpc_attr_buf) return; tu_cs_emit_regs(cs, VPC_ATTR_BUF_GMEM_SIZE(CHIP, gmem ? dev->physical_device->vpc_attr_buf_size_gmem : dev->physical_device->vpc_attr_buf_size_bypass), VPC_ATTR_BUF_GMEM_BASE(CHIP, gmem ? dev->physical_device->vpc_attr_buf_offset_gmem : dev->physical_device->vpc_attr_buf_offset_bypass), ); tu_cs_emit_regs(cs, PC_ATTR_BUF_GMEM_SIZE(CHIP, gmem ? dev->physical_device->vpc_attr_buf_size_gmem : dev->physical_device->vpc_attr_buf_size_bypass), ); } template static void emit_rb_ccu_cntl(struct tu_cs *cs, struct tu_device *dev, bool gmem) { /* The CCUs are a cache that allocates memory from GMEM while facilitating * framebuffer caching for sysmem rendering. The CCU is split into two parts, * one for color and one for depth. The size and offset of these in GMEM can * be configured separately. * * The most common configuration for the CCU is to occupy as much as possible * of GMEM (CACHE_SIZE_FULL) during sysmem rendering as GMEM is unused. On * the other hand, when rendering to GMEM, the CCUs can be left enabled at * any configuration as they don't interfere with GMEM rendering and only * overwrite GMEM when sysmem operations are performed. * * The vast majority of GMEM rendering doesn't need any sysmem operations * but there are some cases where it is required. For example, when the * framebuffer isn't aligned to the tile size or with certain MSAA resolves. * * To correctly handle these cases, we need to be able to switch between * sysmem and GMEM rendering. We do this by allocating a carveout at the * end of GMEM for the color CCU (as none of these operations are depth) * which the color CCU offset is set to and the GMEM size available to the * GMEM layout calculations is adjusted accordingly. */ uint32_t color_offset = gmem ? dev->physical_device->ccu_offset_gmem : dev->physical_device->ccu_offset_bypass; uint32_t color_offset_hi = color_offset >> 21; color_offset &= 0x1fffff; uint32_t depth_offset = gmem ? 0 : dev->physical_device->ccu_depth_offset_bypass; uint32_t depth_offset_hi = depth_offset >> 21; depth_offset &= 0x1fffff; enum a6xx_ccu_cache_size color_cache_size = !gmem ? CCU_CACHE_SIZE_FULL : !gmem ? CCU_CACHE_SIZE_FULL : (a6xx_ccu_cache_size)(dev->physical_device->info->a6xx.gmem_ccu_color_cache_fraction); if (CHIP == A7XX) { tu_cs_emit_regs(cs, A7XX_RB_CCU_CACHE_CNTL( .depth_offset_hi = depth_offset_hi, .color_offset_hi = color_offset_hi, .depth_cache_size = CCU_CACHE_SIZE_FULL, .depth_offset = depth_offset, .color_cache_size = color_cache_size, .color_offset = color_offset )); } else { tu_cs_emit_regs(cs, RB_CCU_CNTL(CHIP, .gmem_fast_clear_disable = !dev->physical_device->info->a6xx.has_gmem_fast_clear, .concurrent_resolve = dev->physical_device->info->a6xx.concurrent_resolve, .depth_offset_hi = 0, .color_offset_hi = color_offset_hi, .depth_cache_size = CCU_CACHE_SIZE_FULL, .depth_offset = 0, .color_cache_size = color_cache_size, .color_offset = color_offset )); } } /* Cache flushes for things that use the color/depth read/write path (i.e. * blits and draws). This deals with changing CCU state as well as the usual * cache flushing. */ template void tu_emit_cache_flush_ccu(struct tu_cmd_buffer *cmd_buffer, struct tu_cs *cs, enum tu_cmd_ccu_state ccu_state) { assert(ccu_state != TU_CMD_CCU_UNKNOWN); /* It's unsafe to flush inside condition because we clear flush_bits */ assert(!cs->cond_stack_depth); /* Changing CCU state must involve invalidating the CCU. In sysmem mode, * the CCU may also contain data that we haven't flushed out yet, so we * also need to flush. Also, in order to program RB_CCU_CNTL, we need to * emit a WFI as it isn't pipelined. * * Note: On A7XX, with the introduction of RB_CCU_CACHE_CNTL, we no longer need * to emit a WFI when changing a subset of CCU state. */ if (ccu_state != cmd_buffer->state.ccu_state) { if (cmd_buffer->state.ccu_state != TU_CMD_CCU_GMEM) { cmd_buffer->state.cache.flush_bits |= TU_CMD_FLAG_CCU_CLEAN_COLOR | TU_CMD_FLAG_CCU_CLEAN_DEPTH; cmd_buffer->state.cache.pending_flush_bits &= ~( TU_CMD_FLAG_CCU_CLEAN_COLOR | TU_CMD_FLAG_CCU_CLEAN_DEPTH); } cmd_buffer->state.cache.flush_bits |= TU_CMD_FLAG_CCU_INVALIDATE_COLOR | TU_CMD_FLAG_CCU_INVALIDATE_DEPTH | (CHIP == A6XX ? TU_CMD_FLAG_WAIT_FOR_IDLE : 0); cmd_buffer->state.cache.pending_flush_bits &= ~( TU_CMD_FLAG_CCU_INVALIDATE_COLOR | TU_CMD_FLAG_CCU_INVALIDATE_DEPTH | (CHIP == A6XX ? TU_CMD_FLAG_WAIT_FOR_IDLE : 0)); } tu_emit_cache_flush(cmd_buffer); if (ccu_state != cmd_buffer->state.ccu_state) { emit_rb_ccu_cntl(cs, cmd_buffer->device, ccu_state == TU_CMD_CCU_GMEM); if (cmd_buffer->device->physical_device->info->a7xx.has_gmem_vpc_attr_buf) { tu7_thread_control(cs, CP_SET_THREAD_BOTH); emit_vpc_attr_buf(cs, cmd_buffer->device, ccu_state == TU_CMD_CCU_GMEM); tu7_thread_control(cs, CP_SET_THREAD_BR); } cmd_buffer->state.ccu_state = ccu_state; } } TU_GENX(tu_emit_cache_flush_ccu); template static void tu6_emit_zs(struct tu_cmd_buffer *cmd, const struct tu_subpass *subpass, struct tu_cs *cs) { const uint32_t a = subpass->depth_stencil_attachment.attachment; if (a == VK_ATTACHMENT_UNUSED) { tu_cs_emit_regs(cs, RB_DEPTH_BUFFER_INFO(CHIP, .depth_format = DEPTH6_NONE), A6XX_RB_DEPTH_BUFFER_PITCH(0), A6XX_RB_DEPTH_BUFFER_ARRAY_PITCH(0), A6XX_RB_DEPTH_BUFFER_BASE(0), A6XX_RB_DEPTH_GMEM_BASE(0)); tu_cs_emit_regs(cs, A6XX_GRAS_SU_DEPTH_BUFFER_INFO(.depth_format = DEPTH6_NONE)); tu_cs_emit_regs(cs, RB_STENCIL_BUFFER_INFO(CHIP, 0)); return; } const struct tu_image_view *iview = cmd->state.attachments[a]; const struct tu_render_pass_attachment *attachment = &cmd->state.pass->attachments[a]; enum a6xx_depth_format fmt = tu6_pipe2depth(attachment->format); tu_cs_emit_pkt4(cs, REG_A6XX_RB_DEPTH_BUFFER_INFO, 6); tu_cs_emit(cs, RB_DEPTH_BUFFER_INFO(CHIP, .depth_format = fmt, .tilemode = TILE6_3, .losslesscompen = iview->view.ubwc_enabled, ).value); if (attachment->format == VK_FORMAT_D32_SFLOAT_S8_UINT) tu_cs_image_depth_ref(cs, iview, 0); else tu_cs_image_ref(cs, &iview->view, 0); tu_cs_emit(cs, tu_attachment_gmem_offset(cmd, attachment, 0)); tu_cs_emit_regs(cs, A6XX_GRAS_SU_DEPTH_BUFFER_INFO(.depth_format = fmt)); tu_cs_emit_pkt4(cs, REG_A6XX_RB_DEPTH_FLAG_BUFFER_BASE, 3); tu_cs_image_flag_ref(cs, &iview->view, 0); if (attachment->format == VK_FORMAT_D32_SFLOAT_S8_UINT || attachment->format == VK_FORMAT_S8_UINT) { tu_cs_emit_pkt4(cs, REG_A6XX_RB_STENCIL_BUFFER_INFO, 6); tu_cs_emit(cs, RB_STENCIL_BUFFER_INFO(CHIP, .separate_stencil = true, .tilemode = TILE6_3, ).value); if (attachment->format == VK_FORMAT_D32_SFLOAT_S8_UINT) { tu_cs_image_stencil_ref(cs, iview, 0); tu_cs_emit(cs, tu_attachment_gmem_offset_stencil(cmd, attachment, 0)); } else { tu_cs_image_ref(cs, &iview->view, 0); tu_cs_emit(cs, tu_attachment_gmem_offset(cmd, attachment, 0)); } } else { tu_cs_emit_regs(cs, RB_STENCIL_BUFFER_INFO(CHIP, 0)); } } template static void tu6_emit_mrt(struct tu_cmd_buffer *cmd, const struct tu_subpass *subpass, struct tu_cs *cs) { const struct tu_framebuffer *fb = cmd->state.framebuffer; enum a6xx_format mrt0_format = FMT6_NONE; uint32_t written = 0; for (uint32_t i = 0; i < subpass->color_count; ++i) { uint32_t a = subpass->color_attachments[i].attachment; unsigned remapped = cmd->vk.dynamic_graphics_state.cal.color_map[i]; if (a == VK_ATTACHMENT_UNUSED || remapped == MESA_VK_ATTACHMENT_UNUSED) continue; const struct tu_image_view *iview = cmd->state.attachments[a]; tu_cs_emit_regs(cs, RB_MRT_BUF_INFO(CHIP, remapped, .dword = iview->view.RB_MRT_BUF_INFO), A6XX_RB_MRT_PITCH(remapped, iview->view.pitch), A6XX_RB_MRT_ARRAY_PITCH(remapped, iview->view.layer_size), A6XX_RB_MRT_BASE(remapped, .qword = tu_layer_address(&iview->view, 0)), A6XX_RB_MRT_BASE_GMEM(remapped, tu_attachment_gmem_offset(cmd, &cmd->state.pass->attachments[a], 0) ), ); tu_cs_emit_regs(cs, A6XX_SP_PS_MRT_REG(remapped, .dword = iview->view.SP_PS_MRT_REG)); tu_cs_emit_pkt4(cs, REG_A6XX_RB_COLOR_FLAG_BUFFER_ADDR(remapped), 3); tu_cs_image_flag_ref(cs, &iview->view, 0); if (remapped == 0) mrt0_format = (enum a6xx_format) (iview->view.SP_PS_MRT_REG & 0xff); written |= 1u << remapped; } u_foreach_bit (i, ~written) { if (i >= subpass->color_count) break; /* From the VkPipelineRenderingCreateInfo definition: * * Valid formats indicate that an attachment can be used - but it * is still valid to set the attachment to NULL when beginning * rendering. * * This means that with dynamic rendering, pipelines may write to * some attachments that are UNUSED here. Setting the format to 0 * here should prevent them from writing to anything. This also seems * to also be required for alpha-to-coverage which can use the alpha * value for an otherwise-unused attachment. */ tu_cs_emit_regs(cs, RB_MRT_BUF_INFO(CHIP, i), A6XX_RB_MRT_PITCH(i), A6XX_RB_MRT_ARRAY_PITCH(i), A6XX_RB_MRT_BASE(i), A6XX_RB_MRT_BASE_GMEM(i), ); tu_cs_emit_regs(cs, A6XX_SP_PS_MRT_REG(i, .dword = 0)); } tu_cs_emit_regs(cs, A6XX_GRAS_LRZ_MRT_BUFFER_INFO_0(.color_format = mrt0_format)); const bool dither = subpass->legacy_dithering_enabled; const uint32_t dither_cntl = A6XX_RB_DITHER_CNTL( .dither_mode_mrt0 = dither ? DITHER_ALWAYS : DITHER_DISABLE, .dither_mode_mrt1 = dither ? DITHER_ALWAYS : DITHER_DISABLE, .dither_mode_mrt2 = dither ? DITHER_ALWAYS : DITHER_DISABLE, .dither_mode_mrt3 = dither ? DITHER_ALWAYS : DITHER_DISABLE, .dither_mode_mrt4 = dither ? DITHER_ALWAYS : DITHER_DISABLE, .dither_mode_mrt5 = dither ? DITHER_ALWAYS : DITHER_DISABLE, .dither_mode_mrt6 = dither ? DITHER_ALWAYS : DITHER_DISABLE, .dither_mode_mrt7 = dither ? DITHER_ALWAYS : DITHER_DISABLE, ) .value; tu_cs_emit_regs(cs, A6XX_RB_DITHER_CNTL(.dword = dither_cntl)); if (CHIP >= A7XX) { tu_cs_emit_regs(cs, A7XX_SP_DITHER_CNTL(.dword = dither_cntl)); } tu_cs_emit_regs(cs, A6XX_RB_SRGB_CNTL(.dword = subpass->srgb_cntl)); tu_cs_emit_regs(cs, A6XX_SP_SRGB_CNTL(.dword = subpass->srgb_cntl)); unsigned layers = MAX2(fb->layers, util_logbase2(subpass->multiview_mask) + 1); tu_cs_emit_regs(cs, A6XX_GRAS_CL_ARRAY_SIZE(layers - 1)); } struct tu_bin_size_params { enum a6xx_render_mode render_mode; bool force_lrz_write_dis; enum a6xx_buffers_location buffers_location; enum a6xx_lrz_feedback_mask lrz_feedback_zmode_mask; bool force_lrz_dis; }; template static void tu6_emit_bin_size(struct tu_cs *cs, uint32_t bin_w, uint32_t bin_h, struct tu_bin_size_params &&p) { if (CHIP == A6XX) { tu_cs_emit_regs( cs, A6XX_GRAS_SC_BIN_CNTL(.binw = bin_w, .binh = bin_h, .render_mode = p.render_mode, .force_lrz_write_dis = p.force_lrz_write_dis, .buffers_location = p.buffers_location, .lrz_feedback_zmode_mask = p.lrz_feedback_zmode_mask, )); } else { tu_cs_emit_regs(cs, A6XX_GRAS_SC_BIN_CNTL(.binw = bin_w, .binh = bin_h, .render_mode = p.render_mode, .force_lrz_write_dis = p.force_lrz_write_dis, .lrz_feedback_zmode_mask = p.lrz_feedback_zmode_mask, .force_lrz_dis = p.force_lrz_dis)); } tu_cs_emit_regs(cs, RB_CNTL(CHIP, .binw = bin_w, .binh = bin_h, .render_mode = p.render_mode, .force_lrz_write_dis = p.force_lrz_write_dis, .buffers_location = p.buffers_location, .lrz_feedback_zmode_mask = p.lrz_feedback_zmode_mask, )); /* no flag for RB_RESOLVE_CNTL_3... */ tu_cs_emit_regs(cs, A6XX_RB_RESOLVE_CNTL_3(.binw = bin_w, .binh = bin_h)); } template static void tu6_emit_render_cntl(struct tu_cmd_buffer *cmd, const struct tu_subpass *subpass, struct tu_cs *cs, bool binning); template <> void tu6_emit_render_cntl(struct tu_cmd_buffer *cmd, const struct tu_subpass *subpass, struct tu_cs *cs, bool binning) { /* doesn't RB_RENDER_CNTL set differently for binning pass: */ bool no_track = !cmd->device->physical_device->info->a6xx.has_cp_reg_write; uint32_t cntl = 0; cntl |= A6XX_RB_RENDER_CNTL_CCUSINGLECACHELINESIZE(2); if (binning) { if (no_track) return; cntl |= A6XX_RB_RENDER_CNTL_FS_DISABLE; } else { uint32_t mrts_ubwc_enable = 0; for (uint32_t i = 0; i < subpass->color_count; ++i) { uint32_t a = subpass->color_attachments[i].attachment; unsigned remapped = cmd->vk.dynamic_graphics_state.cal.color_map[i]; if (a == VK_ATTACHMENT_UNUSED || remapped == MESA_VK_ATTACHMENT_UNUSED) continue; const struct tu_image_view *iview = cmd->state.attachments[a]; if (iview->view.ubwc_enabled) mrts_ubwc_enable |= 1 << remapped; } cntl |= A6XX_RB_RENDER_CNTL_FLAG_MRTS(mrts_ubwc_enable); const uint32_t a = subpass->depth_stencil_attachment.attachment; if (a != VK_ATTACHMENT_UNUSED) { const struct tu_image_view *iview = cmd->state.attachments[a]; if (iview->view.ubwc_enabled) cntl |= A6XX_RB_RENDER_CNTL_FLAG_DEPTH; } if (no_track) { tu_cs_emit_pkt4(cs, REG_A6XX_RB_RENDER_CNTL, 1); tu_cs_emit(cs, cntl); return; } /* In the !binning case, we need to set RB_RENDER_CNTL in the draw_cs * in order to set it correctly for the different subpasses. However, * that means the packets we're emitting also happen during binning. So * we need to guard the write on !BINNING at CP execution time. */ tu_cs_reserve(cs, 3 + 4); tu_cs_emit_pkt7(cs, CP_COND_REG_EXEC, 2); tu_cs_emit(cs, CP_COND_REG_EXEC_0_MODE(RENDER_MODE) | CP_COND_REG_EXEC_0_GMEM | CP_COND_REG_EXEC_0_SYSMEM); tu_cs_emit(cs, RENDER_MODE_CP_COND_REG_EXEC_1_DWORDS(4)); } tu_cs_emit_pkt7(cs, CP_REG_WRITE, 3); tu_cs_emit(cs, CP_REG_WRITE_0_TRACKER(TRACK_RENDER_CNTL)); tu_cs_emit(cs, REG_A6XX_RB_RENDER_CNTL); tu_cs_emit(cs, cntl); } template <> void tu6_emit_render_cntl(struct tu_cmd_buffer *cmd, const struct tu_subpass *subpass, struct tu_cs *cs, bool binning) { } static void tu6_emit_blit_scissor(struct tu_cmd_buffer *cmd, struct tu_cs *cs, bool align, bool used_by_sysmem) { struct tu_physical_device *phys_dev = cmd->device->physical_device; const VkRect2D *render_area = &cmd->state.render_area; /* Avoid assertion fails with an empty render area at (0, 0) where the * subtraction below wraps around. Empty render areas should be forced to * the sysmem path by use_sysmem_rendering(). It's not even clear whether * an empty scissor here works, and the blob seems to force sysmem too as * it sets something wrong (non-empty) for the scissor. */ if (render_area->extent.width == 0 || render_area->extent.height == 0) return; uint32_t x1 = render_area->offset.x; uint32_t y1 = render_area->offset.y; uint32_t x2 = x1 + render_area->extent.width - 1; uint32_t y2 = y1 + render_area->extent.height - 1; if (align) { x1 = x1 & ~(phys_dev->info->gmem_align_w - 1); y1 = y1 & ~(phys_dev->info->gmem_align_h - 1); x2 = ALIGN_POT(x2 + 1, phys_dev->info->gmem_align_w) - 1; y2 = ALIGN_POT(y2 + 1, phys_dev->info->gmem_align_h) - 1; } /* With FDM offset, bins are shifted to the right in GMEM space compared to * framebuffer space. We do not use RB_BLIT_SCISSOR_* for loads and stores * because those do not use the fast path, but we do use it for * LOAD_OP_CLEAR. Expand the render area so that GMEM clears work * correctly. We may over-clear but that's ok because the store is clipped * to the render area. */ if (tu_enable_fdm_offset(cmd)) { const struct tu_tiling_config *tiling = cmd->state.tiling; /* If this is a generic clear that's also used in sysmem mode then we * need to emit the unmodified render area in sysmem mode because * over-clearing is not allowed. */ if (used_by_sysmem) { tu_cs_emit_regs(cs, A6XX_RB_RESOLVE_CNTL_1(.x = x1, .y = y1), A6XX_RB_RESOLVE_CNTL_2(.x = x2, .y = y2)); tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(RENDER_MODE) | CP_COND_REG_EXEC_0_GMEM); } x2 += tiling->tile0.width; y2 += tiling->tile0.height; tu_cs_emit_regs(cs, A6XX_RB_RESOLVE_CNTL_1(.x = x1, .y = y1), A6XX_RB_RESOLVE_CNTL_2(.x = x2, .y = y2)); if (used_by_sysmem) { tu_cond_exec_end(cs); } } else { tu_cs_emit_regs(cs, A6XX_RB_RESOLVE_CNTL_1(.x = x1, .y = y1), A6XX_RB_RESOLVE_CNTL_2(.x = x2, .y = y2)); } } void tu6_emit_window_scissor(struct tu_cs *cs, uint32_t x1, uint32_t y1, uint32_t x2, uint32_t y2) { tu_cs_emit_regs(cs, A6XX_GRAS_SC_WINDOW_SCISSOR_TL(.x = x1, .y = y1), A6XX_GRAS_SC_WINDOW_SCISSOR_BR(.x = x2, .y = y2)); tu_cs_emit_regs(cs, A6XX_GRAS_A2D_SCISSOR_TL(.x = x1, .y = y1), A6XX_GRAS_A2D_SCISSOR_BR(.x = x2, .y = y2)); } template void tu6_emit_window_offset(struct tu_cs *cs, uint32_t x1, uint32_t y1) { tu_cs_emit_regs(cs, A6XX_RB_WINDOW_OFFSET(.x = x1, .y = y1)); tu_cs_emit_regs(cs, A6XX_RB_RESOLVE_WINDOW_OFFSET(.x = x1, .y = y1)); tu_cs_emit_regs(cs, SP_WINDOW_OFFSET(CHIP, .x = x1, .y = y1)); tu_cs_emit_regs(cs, A6XX_TPL1_WINDOW_OFFSET(.x = x1, .y = y1)); if (CHIP >= A7XX) { tu_cs_emit_regs(cs, A7XX_TPL1_A2D_WINDOW_OFFSET(.x = x1, .y = y1)); } } void tu6_apply_depth_bounds_workaround(struct tu_device *device, uint32_t *rb_depth_cntl) { if (!device->physical_device->info->a6xx.depth_bounds_require_depth_test_quirk) return; /* On some GPUs it is necessary to enable z test for depth bounds test when * UBWC is enabled. Otherwise, the GPU would hang. FUNC_ALWAYS is required to * pass z test. Relevant tests: * dEQP-VK.pipeline.extended_dynamic_state.two_draws_dynamic.depth_bounds_test_disable * dEQP-VK.dynamic_state.ds_state.depth_bounds_1 */ *rb_depth_cntl |= A6XX_RB_DEPTH_CNTL_Z_TEST_ENABLE | A6XX_RB_DEPTH_CNTL_ZFUNC(FUNC_ALWAYS); } static void tu_cs_emit_draw_state(struct tu_cs *cs, uint32_t id, struct tu_draw_state state) { uint32_t enable_mask; switch (id) { case TU_DRAW_STATE_VS: case TU_DRAW_STATE_FS: case TU_DRAW_STATE_VPC: /* The blob seems to not enable this (DESC_SETS_LOAD) for binning, even * when resources would actually be used in the binning shader. * Presumably the overhead of prefetching the resources isn't * worth it. */ case TU_DRAW_STATE_DESC_SETS_LOAD: enable_mask = CP_SET_DRAW_STATE__0_GMEM | CP_SET_DRAW_STATE__0_SYSMEM; break; case TU_DRAW_STATE_VS_BINNING: case TU_DRAW_STATE_GS_BINNING: enable_mask = CP_SET_DRAW_STATE__0_BINNING; break; case TU_DRAW_STATE_INPUT_ATTACHMENTS_GMEM: enable_mask = CP_SET_DRAW_STATE__0_GMEM; break; case TU_DRAW_STATE_PRIM_MODE_GMEM: /* On a7xx the prim mode is the same for gmem and sysmem, and it no * longer depends on dynamic state, so we reuse the gmem state for * everything: */ if (cs->device->physical_device->info->a6xx.has_coherent_ubwc_flag_caches) { enable_mask = CP_SET_DRAW_STATE__0_GMEM | CP_SET_DRAW_STATE__0_SYSMEM | CP_SET_DRAW_STATE__0_BINNING; } else { enable_mask = CP_SET_DRAW_STATE__0_GMEM; } break; case TU_DRAW_STATE_INPUT_ATTACHMENTS_SYSMEM: enable_mask = CP_SET_DRAW_STATE__0_SYSMEM; break; case TU_DRAW_STATE_DYNAMIC + TU_DYNAMIC_STATE_PRIM_MODE_SYSMEM: if (!cs->device->physical_device->info->a6xx.has_coherent_ubwc_flag_caches) { /* By also applying the state during binning we ensure that there * is no rotation applied, by previous A6XX_GRAS_SC_CNTL::rotation. */ enable_mask = CP_SET_DRAW_STATE__0_SYSMEM | CP_SET_DRAW_STATE__0_BINNING; } else { static_assert(TU_DYNAMIC_STATE_PRIM_MODE_SYSMEM == TU_DYNAMIC_STATE_A7XX_FRAGMENT_SHADING_RATE); enable_mask = CP_SET_DRAW_STATE__0_GMEM | CP_SET_DRAW_STATE__0_SYSMEM | CP_SET_DRAW_STATE__0_BINNING; } break; default: enable_mask = CP_SET_DRAW_STATE__0_GMEM | CP_SET_DRAW_STATE__0_SYSMEM | CP_SET_DRAW_STATE__0_BINNING; break; } STATIC_ASSERT(TU_DRAW_STATE_COUNT <= 32); /* We need to reload the descriptors every time the descriptor sets * change. However, the commands we send only depend on the pipeline * because the whole point is to cache descriptors which are used by the * pipeline. There's a problem here, in that the firmware has an * "optimization" which skips executing groups that are set to the same * value as the last draw. This means that if the descriptor sets change * but not the pipeline, we'd try to re-execute the same buffer which * the firmware would ignore and we wouldn't pre-load the new * descriptors. Set the DIRTY bit to avoid this optimization. * * We set the dirty bit for shader draw states because they contain * CP_LOAD_STATE packets that are invalidated by the PROGRAM_CONFIG draw * state, so if PROGRAM_CONFIG changes but one of the shaders stays the * same then we still need to re-emit everything. The GLES blob which * implements separate shader draw states does the same thing. * * We also need to set this bit for draw states which may be patched by the * GPU, because their underlying memory may change between setting the draw * state. */ if (id == TU_DRAW_STATE_DESC_SETS_LOAD || id == TU_DRAW_STATE_VS || id == TU_DRAW_STATE_VS_BINNING || id == TU_DRAW_STATE_HS || id == TU_DRAW_STATE_DS || id == TU_DRAW_STATE_GS || id == TU_DRAW_STATE_GS_BINNING || id == TU_DRAW_STATE_FS || state.writeable) enable_mask |= CP_SET_DRAW_STATE__0_DIRTY; tu_cs_emit(cs, CP_SET_DRAW_STATE__0_COUNT(state.size) | enable_mask | CP_SET_DRAW_STATE__0_GROUP_ID(id) | COND(!state.size || !state.iova, CP_SET_DRAW_STATE__0_DISABLE)); tu_cs_emit_qw(cs, state.iova); } void tu6_emit_msaa(struct tu_cs *cs, VkSampleCountFlagBits vk_samples, bool msaa_disable) { const enum a3xx_msaa_samples samples = tu_msaa_samples(vk_samples); msaa_disable |= (samples == MSAA_ONE); tu_cs_emit_regs(cs, A6XX_TPL1_RAS_MSAA_CNTL(samples), A6XX_TPL1_DEST_MSAA_CNTL(.samples = samples, .msaa_disable = msaa_disable)); tu_cs_emit_regs(cs, A6XX_GRAS_SC_RAS_MSAA_CNTL(samples), A6XX_GRAS_SC_DEST_MSAA_CNTL(.samples = samples, .msaa_disable = msaa_disable)); tu_cs_emit_regs(cs, A6XX_RB_RAS_MSAA_CNTL(samples), A6XX_RB_DEST_MSAA_CNTL(.samples = samples, .msaa_disable = msaa_disable)); } static void tu6_update_msaa(struct tu_cmd_buffer *cmd) { VkSampleCountFlagBits samples = cmd->vk.dynamic_graphics_state.ms.rasterization_samples;; /* The samples may not be set by the pipeline or dynamically if raster * discard is enabled. We can set any valid value, but don't set the * default invalid value of 0. */ if (samples == 0) samples = VK_SAMPLE_COUNT_1_BIT; tu6_emit_msaa(&cmd->draw_cs, samples, cmd->state.msaa_disable); } static void tu6_update_msaa_disable(struct tu_cmd_buffer *cmd) { VkPrimitiveTopology topology = (VkPrimitiveTopology)cmd->vk.dynamic_graphics_state.ia.primitive_topology; bool is_line = topology == VK_PRIMITIVE_TOPOLOGY_LINE_LIST || topology == VK_PRIMITIVE_TOPOLOGY_LINE_LIST_WITH_ADJACENCY || topology == VK_PRIMITIVE_TOPOLOGY_LINE_STRIP || topology == VK_PRIMITIVE_TOPOLOGY_LINE_STRIP_WITH_ADJACENCY || (topology == VK_PRIMITIVE_TOPOLOGY_PATCH_LIST && cmd->state.shaders[MESA_SHADER_TESS_EVAL] && cmd->state.shaders[MESA_SHADER_TESS_EVAL]->variant && cmd->state.shaders[MESA_SHADER_TESS_EVAL]->variant->key.tessellation == IR3_TESS_ISOLINES); bool msaa_disable = is_line && cmd->vk.dynamic_graphics_state.rs.line.mode == VK_LINE_RASTERIZATION_MODE_BRESENHAM_KHR; if (cmd->state.msaa_disable != msaa_disable) { cmd->state.msaa_disable = msaa_disable; tu6_update_msaa(cmd); } } static const struct tu_vsc_config * tu_vsc_config(struct tu_cmd_buffer *cmd, const struct tu_tiling_config *tiling) { if (tu_enable_fdm_offset(cmd)) return &tiling->fdm_offset_vsc; return &tiling->vsc; } static bool use_hw_binning(struct tu_cmd_buffer *cmd) { const struct tu_framebuffer *fb = cmd->state.framebuffer; const struct tu_tiling_config *tiling = &fb->tiling[cmd->state.gmem_layout]; const struct tu_vsc_config *vsc = tu_vsc_config(cmd, tiling); /* XFB commands are emitted for BINNING || SYSMEM, which makes it * incompatible with non-hw binning GMEM rendering. this is required because * some of the XFB commands need to only be executed once. * use_sysmem_rendering() should have made sure we only ended up here if no * XFB was used. */ if (cmd->state.rp.xfb_used) { assert(vsc->binning_possible); return true; } /* VK_QUERY_TYPE_PRIMITIVES_GENERATED_EXT emulates GL_PRIMITIVES_GENERATED, * which wasn't designed to care about tilers and expects the result not to * be multiplied by tile count. * See https://gitlab.khronos.org/vulkan/vulkan/-/issues/3131 */ if (cmd->state.rp.has_prim_generated_query_in_rp || cmd->state.prim_generated_query_running_before_rp) { assert(vsc->binning_possible); return true; } return vsc->binning; } static bool use_sysmem_rendering(struct tu_cmd_buffer *cmd, struct tu_renderpass_result **autotune_result) { if (TU_DEBUG(SYSMEM)) { cmd->state.rp.gmem_disable_reason = "TU_DEBUG(SYSMEM)"; return true; } /* can't fit attachments into gmem */ if (!cmd->state.tiling->possible) { cmd->state.rp.gmem_disable_reason = "Can't fit attachments into gmem"; return true; } /* Use sysmem for empty render areas */ if (cmd->state.render_area.extent.width == 0 || cmd->state.render_area.extent.height == 0) { cmd->state.rp.gmem_disable_reason = "Render area is empty"; return true; } if (cmd->state.rp.has_tess) { cmd->state.rp.gmem_disable_reason = "Uses tessellation shaders"; return true; } if (cmd->state.rp.disable_gmem) { /* gmem_disable_reason is set where disable_gmem is set. */ return true; } const struct tu_vsc_config *vsc = tu_vsc_config(cmd, cmd->state.tiling); /* XFB is incompatible with non-hw binning GMEM rendering, see use_hw_binning */ if (cmd->state.rp.xfb_used && !vsc->binning_possible) { cmd->state.rp.gmem_disable_reason = "XFB is incompatible with non-hw binning GMEM rendering"; return true; } /* QUERY_TYPE_PRIMITIVES_GENERATED is incompatible with non-hw binning * GMEM rendering, see use_hw_binning. */ if ((cmd->state.rp.has_prim_generated_query_in_rp || cmd->state.prim_generated_query_running_before_rp) && !vsc->binning_possible) { cmd->state.rp.gmem_disable_reason = "QUERY_TYPE_PRIMITIVES_GENERATED is incompatible with non-hw binning GMEM rendering"; return true; } if (TU_DEBUG(GMEM)) return false; bool use_sysmem = tu_autotune_use_bypass(&cmd->device->autotune, cmd, autotune_result); if (*autotune_result) { list_addtail(&(*autotune_result)->node, &cmd->renderpass_autotune_results); } if (use_sysmem) { cmd->state.rp.gmem_disable_reason = "Autotune selected sysmem"; } return use_sysmem; } /* Optimization: there is no reason to load gmem if there is no * geometry to process. COND_REG_EXEC predicate is set here, * but the actual skip happens in tu_load_gmem_attachment() and tile_store_cs, * for each blit separately. */ template static void tu6_emit_cond_for_load_stores(struct tu_cmd_buffer *cmd, struct tu_cs *cs, uint32_t pipe, uint32_t slot, bool skip_wfm) { const struct tu_vsc_config *vsc = tu_vsc_config(cmd, cmd->state.tiling); if (vsc->binning_possible && cmd->state.pass->has_cond_load_store) { if (CHIP >= A7XX) { tu_cs_emit_pkt7(cs, CP_REG_TEST, 1); tu_cs_emit(cs, A6XX_CP_REG_TEST_0_SCRATCH_MEM_OFFSET(pipe) | A6XX_CP_REG_TEST_0_SOURCE(SOURCE_SCRATCH_MEM) | A6XX_CP_REG_TEST_0_BIT(slot) | A6XX_CP_REG_TEST_0_SKIP_WAIT_FOR_ME); } else { tu_cs_emit_pkt7(cs, CP_REG_TEST, 1); tu_cs_emit(cs, A6XX_CP_REG_TEST_0_REG(REG_A6XX_VSC_CHANNEL_VISIBILITY(pipe)) | A6XX_CP_REG_TEST_0_BIT(slot) | COND(skip_wfm, A6XX_CP_REG_TEST_0_SKIP_WAIT_FOR_ME)); } } else { /* COND_REG_EXECs are not emitted in non-binning case */ } } struct tu_tile_config { VkOffset2D pos; uint32_t pipe; uint32_t slot_mask; VkExtent2D extent; VkExtent2D frag_areas[MAX_VIEWS]; }; /* For bin offsetting we want to do "Euclidean division," where the remainder * (i.e. the offset of the bin) is always positive. Unfortunately C/C++ * remainder and division don't do this, so we have to implement it ourselves. * * For example, we should have: * * euclid_rem(-3, 4) = 1 * euclid_rem(-4, 4) = 0 * euclid_rem(-4, 4) = 3 */ static int32_t euclid_rem(int32_t divisor, int32_t divisend) { if (divisor >= 0) return divisor % divisend; int32_t tmp = divisend - (-divisor % divisend); return tmp == divisend ? 0 : tmp; } /* Calculate how much the bins for a given view should be shifted to the left * and upwards, given the application-provided FDM offset. */ static VkOffset2D tu_bin_offset(VkOffset2D fdm_offset, const struct tu_tiling_config *tiling) { return (VkOffset2D) { euclid_rem(-fdm_offset.x, tiling->tile0.width), euclid_rem(-fdm_offset.y, tiling->tile0.height), }; } static uint32_t tu_fdm_num_layers(const struct tu_cmd_buffer *cmd) { return cmd->state.pass->num_views ? cmd->state.pass->num_views : (cmd->state.fdm_per_layer ? cmd->state.framebuffer->layers : 1); } template static void tu6_emit_tile_select(struct tu_cmd_buffer *cmd, struct tu_cs *cs, const struct tu_tile_config *tile, bool fdm, const VkOffset2D *fdm_offsets) { struct tu_physical_device *phys_dev = cmd->device->physical_device; const struct tu_tiling_config *tiling = cmd->state.tiling; const struct tu_framebuffer *fb = cmd->state.framebuffer; const struct tu_vsc_config *vsc = tu_vsc_config(cmd, tiling); bool hw_binning = use_hw_binning(cmd); tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_MODE(RM6_BIN_RENDER_START) | A6XX_CP_SET_MARKER_0_USES_GMEM); if (CHIP == A6XX && cmd->device->physical_device->has_preemption) { if (cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT) tu_cs_set_writeable(cs, true); tu_emit_vsc(cmd, &cmd->cs); if (cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT) tu_cs_set_writeable(cs, false); } unsigned views = tu_fdm_num_layers(cmd); bool bin_is_scaled = false; if (fdm) { for (unsigned i = 0; i < views; i++) { if (tile->frag_areas[i].width != 1 || tile->frag_areas[i].height != 1) { bin_is_scaled = true; break; } } } bool bin_scale_en = cmd->device->physical_device->info->a7xx.has_hw_bin_scaling && views <= MAX_HW_SCALED_VIEWS && !cmd->state.rp.shared_viewport && bin_is_scaled; /* We cannot support LRZ if we cannot use HW bin scaling and the bin is * scaled (i.e. less than full resolution) */ bool disable_lrz = bin_is_scaled && !bin_scale_en; /* We cannot support LRZ for the first row and column because the offset * required wouldn't be aligned to HW requirements. */ if (fdm_offsets && (tile->pos.x == 0 || tile->pos.y == 0)) disable_lrz = true; tu6_emit_bin_size( cs, tiling->tile0.width, tiling->tile0.height, { .render_mode = RENDERING_PASS, .force_lrz_write_dis = !phys_dev->info->a6xx.has_lrz_feedback, .buffers_location = BUFFERS_IN_GMEM, .lrz_feedback_zmode_mask = phys_dev->info->a6xx.has_lrz_feedback && !bin_is_scaled ? (hw_binning ? LRZ_FEEDBACK_EARLY_Z_OR_EARLY_Z_LATE_Z : LRZ_FEEDBACK_EARLY_Z_LATE_Z) : LRZ_FEEDBACK_NONE, .force_lrz_dis = CHIP >= A7XX && disable_lrz, }); tu_cs_emit_regs(cs, A6XX_VFD_RENDER_MODE(RENDERING_PASS)); const uint32_t x1 = tiling->tile0.width * tile->pos.x; const uint32_t y1 = tiling->tile0.height * tile->pos.y; const uint32_t x2 = MIN2(x1 + tiling->tile0.width, MAX_VIEWPORT_SIZE); const uint32_t y2 = MIN2(y1 + tiling->tile0.height, MAX_VIEWPORT_SIZE); if (bin_scale_en) { /* It seems that the window scissor happens *before* * GRAS_BIN_FOVEAT_OFFSET_* is applied to the fragment coordinates, * unlike the window offset which happens after it is applied. This * means that the window scissor cannot do its job and we have to * disable it by setting it to the entire FB size (plus an extra tile * size, in case GRAS_BIN_FOVEAT_OFFSET_* is not in use). With FDM it is * effectively replaced by the user's scissor anyway. */ uint32_t width = fb->width + tiling->tile0.width; uint32_t height = fb->height + tiling->tile0.height; tu6_emit_window_scissor(cs, 0, 0, width, height); } else { tu6_emit_window_scissor(cs, x1, y1, x2 - 1, y2 - 1); } tu6_emit_window_offset(cs, x1, y1); unsigned slot = ffs(tile->slot_mask) - 1; if (hw_binning) { bool abs_mask = cmd->device->physical_device->info->a7xx.has_abs_bin_mask; tu_cs_emit_pkt7(cs, CP_WAIT_FOR_ME, 0); tu_cs_emit_pkt7(cs, CP_SET_MODE, 1); tu_cs_emit(cs, 0x0); tu_cs_emit_pkt7(cs, CP_SET_BIN_DATA5_OFFSET, abs_mask ? 5 : 4); /* A702 also sets BIT(0) but that hangchecks */ tu_cs_emit(cs, vsc->pipe_sizes[tile->pipe] | CP_SET_BIN_DATA5_0_VSC_N(slot) | CP_SET_BIN_DATA5_0_VSC_MASK(tile->slot_mask >> slot) | COND(abs_mask, CP_SET_BIN_DATA5_0_ABS_MASK(ABS_MASK))); if (abs_mask) tu_cs_emit(cs, tile->slot_mask); tu_cs_emit(cs, tile->pipe * cmd->vsc_draw_strm_pitch); tu_cs_emit(cs, tile->pipe * 4); tu_cs_emit(cs, tile->pipe * cmd->vsc_prim_strm_pitch); } if (util_is_power_of_two_nonzero(tile->slot_mask)) tu6_emit_cond_for_load_stores(cmd, cs, tile->pipe, slot, hw_binning); tu_cs_emit_pkt7(cs, CP_SET_VISIBILITY_OVERRIDE, 1); tu_cs_emit(cs, !hw_binning); tu_cs_emit_pkt7(cs, CP_SET_MODE, 1); tu_cs_emit(cs, 0x0); if (fdm) { VkRect2D bin = { { x1, y1 }, { (x2 - x1) * tile->extent.width, (y2 - y1) * tile->extent.height } }; VkRect2D bins[views]; VkOffset2D frag_offsets[MAX_VIEWS]; for (unsigned i = 0; i < views; i++) { frag_offsets[i] = (VkOffset2D) { 0, 0 }; if (!fdm_offsets || cmd->state.rp.shared_viewport) { bins[i] = bin; continue; } VkOffset2D bin_offset = tu_bin_offset(fdm_offsets[i], tiling); bins[i].offset.x = MAX2(0, (int32_t)x1 - bin_offset.x); bins[i].offset.y = MAX2(0, (int32_t)y1 - bin_offset.y); bins[i].extent.width = MAX2(MIN2((int32_t)x1 + bin.extent.width - bin_offset.x, MAX_VIEWPORT_SIZE) - bins[i].offset.x, 0); bins[i].extent.height = MAX2(MIN2((int32_t)y1 + bin.extent.height - bin_offset.y, MAX_VIEWPORT_SIZE) - bins[i].offset.y, 0); } if (cmd->device->physical_device->info->a7xx.has_hw_bin_scaling) { if (bin_scale_en) { VkExtent2D frag_areas[MAX_HW_SCALED_VIEWS]; for (unsigned i = 0; i < MAX_HW_SCALED_VIEWS; i++) { if (i >= views) { /* Make sure unused views aren't garbage */ frag_areas[i] = (VkExtent2D) {1, 1}; frag_offsets[i] = (VkOffset2D) { 0, 0 }; continue; } frag_areas[i] = tile->frag_areas[i]; frag_offsets[i].x = x1 - x1 / tile->frag_areas[i].width; frag_offsets[i].y = y1 - y1 / tile->frag_areas[i].height; } tu_cs_emit_regs(cs, A7XX_GRAS_BIN_FOVEAT( .binscaleen = bin_scale_en, .xscale_0 = (enum a7xx_bin_scale)util_logbase2(frag_areas[0].width), .yscale_0 = (enum a7xx_bin_scale)util_logbase2(frag_areas[0].height), .xscale_1 = (enum a7xx_bin_scale)util_logbase2(frag_areas[1].width), .yscale_1 = (enum a7xx_bin_scale)util_logbase2(frag_areas[1].height), .xscale_2 = (enum a7xx_bin_scale)util_logbase2(frag_areas[2].width), .yscale_2 = (enum a7xx_bin_scale)util_logbase2(frag_areas[2].height), .xscale_3 = (enum a7xx_bin_scale)util_logbase2(frag_areas[3].width), .yscale_3 = (enum a7xx_bin_scale)util_logbase2(frag_areas[3].height), .xscale_4 = (enum a7xx_bin_scale)util_logbase2(frag_areas[4].width), .yscale_4 = (enum a7xx_bin_scale)util_logbase2(frag_areas[4].height), .xscale_5 = (enum a7xx_bin_scale)util_logbase2(frag_areas[5].width), .yscale_5 = (enum a7xx_bin_scale)util_logbase2(frag_areas[5].height)), A7XX_GRAS_BIN_FOVEAT_OFFSET_0( .xoffset_0 = frag_offsets[0].x, .xoffset_1 = frag_offsets[1].x, .xoffset_2 = frag_offsets[2].x), A7XX_GRAS_BIN_FOVEAT_OFFSET_1( .xoffset_3 = frag_offsets[3].x, .xoffset_4 = frag_offsets[4].x, .xoffset_5 = frag_offsets[5].x), A7XX_GRAS_BIN_FOVEAT_OFFSET_2( .yoffset_0 = frag_offsets[0].y, .yoffset_1 = frag_offsets[1].y, .yoffset_2 = frag_offsets[2].y), A7XX_GRAS_BIN_FOVEAT_OFFSET_3( .yoffset_3 = frag_offsets[3].y, .yoffset_4 = frag_offsets[4].y, .yoffset_5 = frag_offsets[5].y)); tu_cs_emit_regs(cs, A7XX_RB_BIN_FOVEAT( .binscaleen = bin_scale_en)); } else { tu_cs_emit_regs(cs, A7XX_GRAS_BIN_FOVEAT()); tu_cs_emit_regs(cs, A7XX_RB_BIN_FOVEAT()); } } util_dynarray_foreach (&cmd->fdm_bin_patchpoints, struct tu_fdm_bin_patchpoint, patch) { tu_cs_emit_pkt7(cs, CP_MEM_WRITE, 2 + patch->size); tu_cs_emit_qw(cs, patch->iova); patch->apply(cmd, cs, patch->data, (VkOffset2D) { x1, y1 }, frag_offsets, views, tile->frag_areas, bins); } /* Make the CP wait until the CP_MEM_WRITE's to the command buffers * land. When loading FS params via UBOs, we also need to invalidate * UCHE because the FS param patchpoint is read through UCHE. */ tu_cs_emit_pkt7(cs, CP_WAIT_MEM_WRITES, 0); if (cmd->device->compiler->load_shader_consts_via_preamble) { tu_emit_event_write(cmd, cs, FD_CACHE_INVALIDATE); tu_cs_emit_wfi(cs); } tu_cs_emit_pkt7(cs, CP_WAIT_FOR_ME, 0); } else if (cmd->device->physical_device->info->a7xx.has_hw_bin_scaling) { tu_cs_emit_regs(cs, A7XX_GRAS_BIN_FOVEAT()); tu_cs_emit_regs(cs, A7XX_RB_BIN_FOVEAT()); } } template static void tu6_emit_sysmem_resolve(struct tu_cmd_buffer *cmd, struct tu_cs *cs, uint32_t layer_mask, uint32_t a, uint32_t gmem_a) { const struct tu_framebuffer *fb = cmd->state.framebuffer; const struct tu_image_view *dst = cmd->state.attachments[a]; const struct tu_image_view *src = cmd->state.attachments[gmem_a]; tu_resolve_sysmem(cmd, cs, src, dst, layer_mask, fb->layers, &cmd->state.render_area); } template static void tu6_emit_sysmem_resolves(struct tu_cmd_buffer *cmd, struct tu_cs *cs, const struct tu_subpass *subpass) { if (subpass->resolve_attachments) { /* From the documentation for vkCmdNextSubpass, section 7.4 "Render Pass * Commands": * * End-of-subpass multisample resolves are treated as color * attachment writes for the purposes of synchronization. * This applies to resolve operations for both color and * depth/stencil attachments. That is, they are considered to * execute in the VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT * pipeline stage and their writes are synchronized with * VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT. Synchronization between * rendering within a subpass and any resolve operations at the end * of the subpass occurs automatically, without need for explicit * dependencies or pipeline barriers. However, if the resolve * attachment is also used in a different subpass, an explicit * dependency is needed. * * We use the CP_BLIT path for sysmem resolves, which is really a * transfer command, so we have to manually flush similar to the gmem * resolve case. However, a flush afterwards isn't needed because of the * last sentence and the fact that we're in sysmem mode. */ tu_emit_event_write(cmd, cs, FD_CCU_CLEAN_COLOR); if (subpass->resolve_depth_stencil) tu_emit_event_write(cmd, cs, FD_CCU_CLEAN_DEPTH); tu_emit_event_write(cmd, cs, FD_CACHE_INVALIDATE); /* Wait for the flushes to land before using the 2D engine */ tu_cs_emit_wfi(cs); for (unsigned i = 0; i < subpass->resolve_count; i++) { uint32_t a = subpass->resolve_attachments[i].attachment; if (a == VK_ATTACHMENT_UNUSED) continue; uint32_t gmem_a = tu_subpass_get_attachment_to_resolve(subpass, i); tu6_emit_sysmem_resolve(cmd, cs, subpass->multiview_mask, a, gmem_a); } } } template static void tu6_emit_sysmem_unresolve(struct tu_cmd_buffer *cmd, struct tu_cs *cs, uint32_t layer_mask, uint32_t a, uint32_t gmem_a) { const struct tu_framebuffer *fb = cmd->state.framebuffer; const struct tu_image_view *src = cmd->state.attachments[a]; const struct tu_image_view *dst = cmd->state.attachments[gmem_a]; tu_resolve_sysmem(cmd, cs, src, dst, layer_mask, fb->layers, &cmd->state.render_area); } template static void tu6_emit_sysmem_unresolves(struct tu_cmd_buffer *cmd, struct tu_cs *cs, const struct tu_subpass *subpass) { if (subpass->unresolve_count) { /* Similar to above, we need to explicitly flush afterwards to keep this * in sync with draw commands. However we also don't currently insert * dependencies when a resolve is followed by an unresolve so we also * need to manually flush for that case. */ tu_emit_event_write(cmd, cs, FD_CCU_CLEAN_COLOR); tu_emit_event_write(cmd, cs, FD_CACHE_INVALIDATE); /* Wait for the flushes to land before using the 2D engine */ tu_cs_emit_wfi(cs); bool unresolve_ds = false; for (unsigned i = 0; i < subpass->unresolve_count; i++) { uint32_t a = subpass->unresolve_attachments[i].attachment; if (a == VK_ATTACHMENT_UNUSED) continue; if (vk_format_is_depth_or_stencil(cmd->state.pass->attachments[a].format)) unresolve_ds = true; uint32_t gmem_a = tu_subpass_get_attachment_to_unresolve(subpass, i); tu6_emit_sysmem_unresolve(cmd, cs, subpass->multiview_mask, a, gmem_a); } tu_emit_event_write(cmd, cs, FD_CCU_CLEAN_COLOR); tu_emit_event_write(cmd, cs, FD_CCU_INVALIDATE_COLOR); if (unresolve_ds) { tu_emit_event_write(cmd, cs, FD_CCU_CLEAN_DEPTH); tu_emit_event_write(cmd, cs, FD_CCU_INVALIDATE_DEPTH); } tu_cs_emit_wfi(cs); } } template static void tu6_emit_gmem_resolves(struct tu_cmd_buffer *cmd, const struct tu_subpass *subpass, struct tu_resolve_group *resolve_group, struct tu_cs *cs) { const struct tu_render_pass *pass = cmd->state.pass; const struct tu_framebuffer *fb = cmd->state.framebuffer; if (subpass->resolve_attachments) { for (unsigned i = 0; i < subpass->resolve_count; i++) { uint32_t a = subpass->resolve_attachments[i].attachment; if (a == VK_ATTACHMENT_UNUSED) continue; uint32_t gmem_a = tu_subpass_get_attachment_to_resolve(subpass, i); tu_store_gmem_attachment(cmd, cs, resolve_group, a, gmem_a, fb->layers, subpass->multiview_mask, false); if (pass->attachments[a].gmem) { /* check if the resolved attachment is needed by later subpasses, * if it is, should be doing a GMEM->GMEM resolve instead of * GMEM->MEM->GMEM.. */ perf_debug(cmd->device, "TODO: missing GMEM->GMEM resolve path\n"); if (CHIP >= A7XX) tu_emit_event_write(cmd, cs, FD_CCU_CLEAN_BLIT_CACHE); tu_load_gmem_attachment(cmd, cs, resolve_group, a, a, false, true); } } } } /* Emits any tile stores at the end of a subpass. * * These are emitted into draw_cs for non-final subpasses, and tile_store_cs for * the final subpass. The draw_cs ones mean that we have to disable IB2 skipping * for the draw_cs so we don't exit before storing. The separate tile_store_cs * lets us leave IB2 skipping enabled in the common case of a single-subpass * renderpass (or dynamic rendering). * * To do better in the multi-subpass case, we'd need the individual CS entries * of draw_cs to have a flag for whether they can be skipped or not, and * interleave drawing cs entries with store cs entries. * * This is independent of cond_store_allowed, which is about "can we skip doing * the store if no other rendering happened in the tile?" We can only skip if * the cond that we set up at the start of the tile (or reset just before * calling tile_store_cs) is still in place. */ template static void tu6_emit_gmem_stores(struct tu_cmd_buffer *cmd, struct tu_cs *cs, struct tu_resolve_group *resolve_group, const struct tu_subpass *subpass) { const struct tu_render_pass *pass = cmd->state.pass; const struct tu_framebuffer *fb = cmd->state.framebuffer; const struct tu_vsc_config *vsc = tu_vsc_config(cmd, cmd->state.tiling); uint32_t subpass_idx = subpass - cmd->state.pass->subpasses; const bool cond_exec_allowed = vsc->binning_possible && cmd->state.pass->has_cond_load_store && (!cmd->state.rp.draw_cs_writes_to_cond_pred || cs != &cmd->draw_cs); bool scissor_emitted = false; /* Resolve should happen before store in case BLIT_EVENT_STORE_AND_CLEAR is * used for a store. * * Note that we're emitting the resolves into the tile store CS, which is * unconditionally executed (unlike draw_cs which depends on geometry having * been generated). a7xx has HW conditional resolve support that may skip * the resolve if geometry didn't cover it, anyway. */ if (subpass->resolve_attachments) { if (!scissor_emitted) { tu6_emit_blit_scissor(cmd, cs, true, false); scissor_emitted = true; } tu6_emit_gmem_resolves(cmd, subpass, resolve_group, cs); } for (uint32_t a = 0; a < pass->attachment_count; ++a) { const struct tu_render_pass_attachment *att = &pass->attachments[a]; /* Note: att->cond_store_allowed implies at least one of att->store_* set */ if (pass->attachments[a].gmem && att->last_subpass_idx == subpass_idx) { if (!scissor_emitted) { tu6_emit_blit_scissor(cmd, cs, true, false); scissor_emitted = true; } tu_store_gmem_attachment(cmd, cs, resolve_group, a, a, fb->layers, att->used_views, cond_exec_allowed); } } } template static void tu6_emit_tile_store_cs(struct tu_cmd_buffer *cmd, struct tu_cs *cs) { const struct tu_render_pass *pass = cmd->state.pass; uint32_t subpass_idx = pass->subpass_count - 1; const struct tu_subpass *subpass = &pass->subpasses[subpass_idx]; if (pass->has_fdm) tu_cs_set_writeable(cs, true); /* We believe setting the marker affects what state HW blocks save/restore * during preemption. So we only emit it before the stores at the end of the * last subpass, not other resolves. */ tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_MODE(RM6_BIN_RESOLVE) | A6XX_CP_SET_MARKER_0_USES_GMEM); struct tu_resolve_group resolve_group = {}; tu6_emit_gmem_stores(cmd, cs, &resolve_group, subpass); tu_emit_resolve_group(cmd, cs, &resolve_group); if (pass->has_fdm) tu_cs_set_writeable(cs, false); } void tu_disable_draw_states(struct tu_cmd_buffer *cmd, struct tu_cs *cs) { tu_cs_emit_pkt7(cs, CP_SET_DRAW_STATE, 3); tu_cs_emit(cs, CP_SET_DRAW_STATE__0_COUNT(0) | CP_SET_DRAW_STATE__0_DISABLE_ALL_GROUPS | CP_SET_DRAW_STATE__0_GROUP_ID(0)); tu_cs_emit(cs, CP_SET_DRAW_STATE__1_ADDR_LO(0)); tu_cs_emit(cs, CP_SET_DRAW_STATE__2_ADDR_HI(0)); cmd->state.dirty |= TU_CMD_DIRTY_DRAW_STATE; } template static void tu6_init_static_regs(struct tu_device *dev, struct tu_cs *cs) { const struct tu_physical_device *phys_dev = dev->physical_device; if (CHIP >= A7XX) { /* On A7XX, RB_CCU_CNTL was broken into two registers, RB_CCU_CNTL which has * static properties that can be set once, this requires a WFI to take effect. * While the newly introduced register RB_CCU_CACHE_CNTL has properties that may * change per-RP and don't require a WFI to take effect, only CCU inval/flush * events are required. */ enum a7xx_concurrent_resolve_mode resolve_mode = CONCURRENT_RESOLVE_MODE_2; if (TU_DEBUG(NO_CONCURRENT_RESOLVES)) resolve_mode = CONCURRENT_RESOLVE_MODE_DISABLED; enum a7xx_concurrent_unresolve_mode unresolve_mode = CONCURRENT_UNRESOLVE_MODE_FULL; if (TU_DEBUG(NO_CONCURRENT_UNRESOLVES)) unresolve_mode = CONCURRENT_UNRESOLVE_MODE_DISABLED; tu_cs_emit_regs(cs, RB_CCU_CNTL(A7XX, .gmem_fast_clear_disable = !dev->physical_device->info->a6xx.has_gmem_fast_clear, .concurrent_resolve_mode = resolve_mode, .concurrent_unresolve_mode = unresolve_mode, )); } for (size_t i = 0; i < ARRAY_SIZE(phys_dev->info->a6xx.magic_raw); i++) { auto magic_reg = phys_dev->info->a6xx.magic_raw[i]; if (!magic_reg.reg) break; uint32_t value = magic_reg.value; switch(magic_reg.reg) { case REG_A6XX_TPL1_DBG_ECO_CNTL1: value = (value & ~A6XX_TPL1_DBG_ECO_CNTL1_TP_UBWC_FLAG_HINT) | (phys_dev->info->a7xx.enable_tp_ubwc_flag_hint ? A6XX_TPL1_DBG_ECO_CNTL1_TP_UBWC_FLAG_HINT : 0); break; } tu_cs_emit_write_reg(cs, magic_reg.reg, value); } if (dev->physical_device->info->a6xx.has_attachment_shading_rate) { tu_cs_emit_write_reg(cs, REG_A7XX_GRAS_LRZ_QUALITY_LOOKUP_TABLE(0), fd_gras_shading_rate_lut(0)); tu_cs_emit_write_reg(cs, REG_A7XX_GRAS_LRZ_QUALITY_LOOKUP_TABLE(1), fd_gras_shading_rate_lut(1)); } tu_cs_emit_write_reg(cs, REG_A6XX_SP_NC_MODE_CNTL_2, 0); tu_cs_emit_write_reg(cs, REG_A6XX_SP_DBG_ECO_CNTL, phys_dev->info->a6xx.magic.SP_DBG_ECO_CNTL); tu_cs_emit_write_reg(cs, REG_A6XX_SP_PERFCTR_SHADER_MASK, 0x3f); if (CHIP == A6XX && !cs->device->physical_device->info->a6xx.is_a702) tu_cs_emit_write_reg(cs, REG_A6XX_TPL1_UNKNOWN_B605, 0x44); tu_cs_emit_write_reg(cs, REG_A6XX_TPL1_DBG_ECO_CNTL, phys_dev->info->a6xx.magic.TPL1_DBG_ECO_CNTL); if (CHIP == A6XX) { tu_cs_emit_write_reg(cs, REG_A6XX_HLSQ_UNKNOWN_BE00, 0x80); tu_cs_emit_write_reg(cs, REG_A6XX_HLSQ_UNKNOWN_BE01, 0); } tu_cs_emit_write_reg(cs, REG_A6XX_VPC_DBG_ECO_CNTL, phys_dev->info->a6xx.magic.VPC_DBG_ECO_CNTL); tu_cs_emit_write_reg(cs, REG_A6XX_GRAS_DBG_ECO_CNTL, phys_dev->info->a6xx.magic.GRAS_DBG_ECO_CNTL); if (CHIP == A6XX) { tu_cs_emit_write_reg(cs, REG_A6XX_HLSQ_DBG_ECO_CNTL, phys_dev->info->a6xx.magic.HLSQ_DBG_ECO_CNTL); } tu_cs_emit_write_reg(cs, REG_A6XX_SP_CHICKEN_BITS, phys_dev->info->a6xx.magic.SP_CHICKEN_BITS); tu_cs_emit_write_reg(cs, REG_A6XX_SP_GFX_USIZE, 0); // 2 on a740 ??? tu_cs_emit_write_reg(cs, REG_A6XX_TPL1_PS_ROTATION_CNTL, 0); if (CHIP == A6XX) tu_cs_emit_regs(cs, A6XX_HLSQ_SHARED_CONSTS(.enable = false)); tu_cs_emit_write_reg(cs, REG_A6XX_UCHE_UNKNOWN_0E12, phys_dev->info->a6xx.magic.UCHE_UNKNOWN_0E12); tu_cs_emit_write_reg(cs, REG_A6XX_UCHE_CLIENT_PF, phys_dev->info->a6xx.magic.UCHE_CLIENT_PF); tu_cs_emit_write_reg(cs, REG_A6XX_SP_UNKNOWN_A9A8, 0); tu_cs_emit_regs(cs, A6XX_SP_MODE_CNTL(.constant_demotion_enable = true, .isammode = ISAMMODE_GL, .shared_consts_enable = false)); tu_cs_emit_regs(cs, A6XX_VFD_MODE_CNTL(.vertex = true, .instance = true)); tu_cs_emit_write_reg(cs, REG_A6XX_RB_MODE_CNTL, 0x00000010); tu_cs_emit_write_reg(cs, REG_A6XX_PC_MODE_CNTL, phys_dev->info->a6xx.magic.PC_MODE_CNTL); tu_cs_emit_regs(cs, GRAS_MODE_CNTL(CHIP)); tu_cs_emit_write_reg(cs, REG_A6XX_RB_UNKNOWN_8818, 0); if (CHIP == A6XX) { tu_cs_emit_write_reg(cs, REG_A6XX_RB_UNKNOWN_8819, 0); tu_cs_emit_write_reg(cs, REG_A6XX_RB_UNKNOWN_881A, 0); tu_cs_emit_write_reg(cs, REG_A6XX_RB_UNKNOWN_881B, 0); tu_cs_emit_write_reg(cs, REG_A6XX_RB_UNKNOWN_881C, 0); tu_cs_emit_write_reg(cs, REG_A6XX_RB_UNKNOWN_881D, 0); tu_cs_emit_write_reg(cs, REG_A6XX_RB_UNKNOWN_881E, 0); } tu_cs_emit_write_reg(cs, REG_A6XX_RB_UNKNOWN_88F0, 0); tu_cs_emit_regs(cs, A6XX_VPC_REPLACE_MODE_CNTL(false)); tu_cs_emit_regs(cs, VPC_ROTATION_CNTL(CHIP)); tu_cs_emit_regs(cs, A6XX_VPC_SO_OVERRIDE(true)); tu_cs_emit_write_reg(cs, REG_A6XX_TPL1_PS_SWIZZLE_CNTL, 0); tu_cs_emit_regs(cs, GRAS_SC_SCREEN_SCISSOR_CNTL(CHIP)); if (CHIP == A6XX) { tu_cs_emit_write_reg(cs, REG_A6XX_GRAS_SU_CONSERVATIVE_RAS_CNTL, 0); tu_cs_emit_regs(cs, A6XX_PC_DGEN_SU_CONSERVATIVE_RAS_CNTL()); tu_cs_emit_write_reg(cs, REG_A6XX_VPC_UNKNOWN_9210, 0); tu_cs_emit_write_reg(cs, REG_A6XX_VPC_UNKNOWN_9211, 0); } tu_cs_emit_write_reg(cs, REG_A6XX_VPC_LB_MODE_CNTL, 0); tu_cs_emit_regs(cs, PC_CONTEXT_SWITCH_GFX_PREEMPTION_MODE(CHIP)); tu_cs_emit_regs(cs, A6XX_TPL1_MODE_CNTL(.isammode = ISAMMODE_GL, .texcoordroundmode = dev->instance->use_tex_coord_round_nearest_even_mode ? COORD_ROUND_NEAREST_EVEN : COORD_TRUNCATE, .nearestmipsnap = CLAMP_ROUND_TRUNCATE, .destdatatypeoverride = true)); tu_cs_emit_regs(cs, SP_REG_PROG_ID_3(CHIP, .dword = 0xfc)); tu_cs_emit_write_reg(cs, REG_A6XX_VFD_RENDER_MODE, 0x00000000); tu_cs_emit_write_reg(cs, REG_A6XX_PC_MODE_CNTL, phys_dev->info->a6xx.magic.PC_MODE_CNTL); tu_cs_emit_regs(cs, A6XX_RB_ALPHA_TEST_CNTL()); /* always disable alpha test */ tu_cs_emit_regs(cs, A6XX_TPL1_GFX_BORDER_COLOR_BASE(.bo = dev->global_bo, .bo_offset = gb_offset(bcolor))); tu_cs_emit_regs(cs, A6XX_TPL1_CS_BORDER_COLOR_BASE(.bo = dev->global_bo, .bo_offset = gb_offset(bcolor))); /* BR-only registers */ if (CHIP >= A7XX) tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(THREAD_MODE) | CP_COND_REG_EXEC_0_BR); tu_cs_emit_write_reg(cs, REG_A6XX_RB_DBG_ECO_CNTL, phys_dev->info->a6xx.magic.RB_DBG_ECO_CNTL); tu_cs_emit_write_reg(cs, REG_A6XX_RB_RBP_CNTL, phys_dev->info->a6xx.magic.RB_RBP_CNTL); if (CHIP >= A7XX) tu_cond_exec_end(cs); if (CHIP == A7XX) { tu_cs_emit_regs(cs, TPL1_BICUBIC_WEIGHTS_TABLE_REG(CHIP, 0, 0), TPL1_BICUBIC_WEIGHTS_TABLE_REG(CHIP, 1, 0x3fe05ff4), TPL1_BICUBIC_WEIGHTS_TABLE_REG(CHIP, 2, 0x3fa0ebee), TPL1_BICUBIC_WEIGHTS_TABLE_REG(CHIP, 3, 0x3f5193ed), TPL1_BICUBIC_WEIGHTS_TABLE_REG(CHIP, 4, 0x3f0243f0), ); } if (CHIP >= A7XX) { /* Blob sets these two per draw. */ tu_cs_emit_regs(cs, A7XX_PC_HS_BUFFER_SIZE(TU_TESS_PARAM_SIZE)); /* Blob adds a bit more space ({0x10, 0x20, 0x30, 0x40} bytes) * but the meaning of this additional space is not known, * so we play safe and don't add it. */ tu_cs_emit_regs(cs, A7XX_PC_TF_BUFFER_SIZE(TU_TESS_FACTOR_SIZE)); } /* There is an optimization to skip executing draw states for draws with no * instances. Instead of simply skipping the draw, internally the firmware * sets a bit in PC_DRAW_INITIATOR that seemingly skips the draw. However * there is a hardware bug where this bit does not always cause the FS * early preamble to be skipped. Because the draw states were skipped, * SP_PS_CNTL_0, SP_PS_BASE and so on are never updated and a * random FS preamble from the last draw is executed. If the last visible * draw is from the same submit, it shouldn't be a problem because we just * re-execute the same preamble and preambles don't have side effects, but * if it's from another process then we could execute a garbage preamble * leading to hangs and faults. To make sure this doesn't happen, we reset * SP_PS_CNTL_0 here, making sure that the EARLYPREAMBLE bit isn't set * so any leftover early preamble doesn't get executed. Other stages don't * seem to be affected. */ if (phys_dev->info->a6xx.has_early_preamble) { tu_cs_emit_regs(cs, A6XX_SP_PS_CNTL_0()); } /* Workaround for draw state with constlen not being applied for * zero-instance draw calls. See IR3_CONST_ALLOC_DRIVER_PARAMS allocation * for more info. */ tu_cs_emit_pkt4( cs, CHIP == A6XX ? REG_A6XX_SP_VS_CONST_CONFIG : REG_A7XX_SP_VS_CONST_CONFIG, 1); tu_cs_emit(cs, A6XX_SP_VS_CONST_CONFIG_CONSTLEN(8) | A6XX_SP_VS_CONST_CONFIG_ENABLED); } /* Set always-identical registers used specifically for GMEM */ static void tu7_emit_tile_render_begin_regs(struct tu_cs *cs) { tu_cs_emit_regs(cs, A7XX_RB_BUFFER_CNTL(0x0)); tu_cs_emit_regs(cs, A7XX_RB_CCU_DBG_ECO_CNTL(0x0)); tu_cs_emit_regs(cs, GRAS_LRZ_CB_CNTL(A7XX, 0x0)); tu_cs_emit_regs(cs, GRAS_MODE_CNTL(A7XX, 0x2)); tu_cs_emit_regs(cs, A7XX_RB_UNKNOWN_8E09(0x4)); tu_cs_emit_regs(cs, A7XX_RB_CLEAR_TARGET(.clear_mode = CLEAR_MODE_GMEM)); } /* Emit the bin restore preamble, which runs in between bins when L1 * preemption with skipsaverestore happens and we switch back to this context. * We need to restore static registers normally programmed at cmdbuf start * which weren't saved, and we need to program the CCU state which is normally * programmed before rendering the bins and isn't saved/restored by the CP * because it is always the same for GMEM render passes. */ template static void tu_emit_bin_preamble(struct tu_device *dev, struct tu_cs *cs, bool bv) { struct tu_physical_device *phys_dev = dev->physical_device; tu6_init_static_regs(dev, cs); if (!bv) emit_rb_ccu_cntl(cs, dev, true); emit_vpc_attr_buf(cs, dev, true); if (CHIP == A6XX) { tu_cs_emit_regs(cs, A6XX_PC_POWER_CNTL(phys_dev->info->a6xx.magic.PC_POWER_CNTL)); tu_cs_emit_regs(cs, A6XX_VFD_POWER_CNTL(phys_dev->info->a6xx.magic.PC_POWER_CNTL)); } if (CHIP == A7XX && !bv) { tu7_emit_tile_render_begin_regs(cs); } if (CHIP == A6XX) { tu_cs_emit_pkt7(cs, CP_MEM_TO_REG, 3); tu_cs_emit(cs, CP_MEM_TO_REG_0_REG(REG_A6XX_VSC_CHANNEL_VISIBILITY(0)) | CP_MEM_TO_REG_0_CNT(32)); tu_cs_emit_qw(cs, dev->global_bo->iova + gb_offset(vsc_state)); } } VkResult tu_init_bin_preamble(struct tu_device *device) { struct tu_cs preamble_cs; VkResult result = tu_cs_begin_sub_stream(&device->sub_cs, 256, &preamble_cs); if (result != VK_SUCCESS) return vk_startup_errorf(device->instance, result, "bin restore"); TU_CALLX(device, tu_emit_bin_preamble)(device, &preamble_cs, false); device->bin_preamble_entry = tu_cs_end_sub_stream(&device->sub_cs, &preamble_cs); if (device->physical_device->info->chip >= 7) { result = tu_cs_begin_sub_stream(&device->sub_cs, 256, &preamble_cs); if (result != VK_SUCCESS) return vk_startup_errorf(device->instance, result, "bin restore"); TU_CALLX(device, tu_emit_bin_preamble)(device, &preamble_cs, true); device->bin_preamble_bv_entry = tu_cs_end_sub_stream(&device->sub_cs, &preamble_cs); } return VK_SUCCESS; } template static void tu6_init_hw(struct tu_cmd_buffer *cmd, struct tu_cs *cs) { struct tu_device *dev = cmd->device; const struct tu_physical_device *phys_dev = dev->physical_device; if (CHIP == A6XX) { tu_emit_event_write(cmd, cs, FD_CACHE_INVALIDATE); } else { tu7_thread_control(cs, CP_SET_THREAD_BOTH); tu_emit_event_write(cmd, cs, FD_CCU_INVALIDATE_COLOR); tu_emit_event_write(cmd, cs, FD_CCU_INVALIDATE_DEPTH); tu_emit_event_write(cmd, cs, FD_LRZ_INVALIDATE); tu_emit_event_write(cmd, cs, FD_CACHE_INVALIDATE); tu_cs_emit_wfi(cs); } tu_cs_emit_regs(cs, SP_UPDATE_CNTL(CHIP, .vs_state = true, .hs_state = true, .ds_state = true, .gs_state = true, .fs_state = true, .cs_state = true, .cs_uav = true, .gfx_uav = true, .cs_shared_const = true, .gfx_shared_const = true, .cs_bindless = CHIP == A6XX ? 0x1f : 0xff, .gfx_bindless = CHIP == A6XX ? 0x1f : 0xff,)); tu_cs_emit_wfi(cs); if (dev->dbg_cmdbuf_stomp_cs) { tu_cs_emit_call(cs, dev->dbg_cmdbuf_stomp_cs); } cmd->state.cache.pending_flush_bits &= ~(TU_CMD_FLAG_WAIT_FOR_IDLE | TU_CMD_FLAG_CACHE_INVALIDATE); tu6_init_static_regs(cmd->device, cs); emit_rb_ccu_cntl(cs, cmd->device, false); emit_vpc_attr_buf(cs, cmd->device, false); cmd->state.ccu_state = TU_CMD_CCU_SYSMEM; tu_disable_draw_states(cmd, cs); if (phys_dev->info->a7xx.cmdbuf_start_a725_quirk) { tu_cs_reserve(cs, 3 + 4); tu_cs_emit_pkt7(cs, CP_COND_REG_EXEC, 2); tu_cs_emit(cs, CP_COND_REG_EXEC_0_MODE(THREAD_MODE) | CP_COND_REG_EXEC_0_BR | CP_COND_REG_EXEC_0_LPAC); tu_cs_emit(cs, RENDER_MODE_CP_COND_REG_EXEC_1_DWORDS(4)); tu_cs_emit_ib(cs, &dev->cmdbuf_start_a725_quirk_entry); } if (CHIP >= A7XX) { if (cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT) tu_cs_set_writeable(cs, true); /* This sets the amount BV is allowed to be ahead of BR when we do * BV_WAIT_FOR_BR. By setting it based on the vis stream count we * prevent write-after-read races with the vis stream. */ tu_cs_emit_pkt7(cs, CP_BV_BR_COUNT_OPS, 2); tu_cs_emit(cs, CP_BV_BR_COUNT_OPS_0_OP(PIPE_SET_BR_OFFSET)); struct tu_vis_stream_patchpoint *patchpoint = &cmd->vis_stream_count_patchpoint; patchpoint->data = cs->cur; patchpoint->iova = tu_cs_get_cur_iova(cs); tu_cs_emit(cs, 1); if (cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT) tu_cs_set_writeable(cs, false); tu7_thread_control(cs, CP_SET_THREAD_BR); } tu_cs_emit_pkt7(cs, CP_SET_AMBLE, 3); tu_cs_emit_qw(cs, cmd->device->bin_preamble_entry.bo->iova + cmd->device->bin_preamble_entry.offset); tu_cs_emit(cs, CP_SET_AMBLE_2_DWORDS(cmd->device->bin_preamble_entry.size / sizeof(uint32_t)) | CP_SET_AMBLE_2_TYPE(BIN_PREAMBLE_AMBLE_TYPE)); if (CHIP >= A7XX) { tu7_thread_control(cs, CP_SET_THREAD_BV); tu_cs_emit_pkt7(cs, CP_SET_AMBLE, 3); tu_cs_emit_qw(cs, cmd->device->bin_preamble_bv_entry.bo->iova + cmd->device->bin_preamble_bv_entry.offset); tu_cs_emit(cs, CP_SET_AMBLE_2_DWORDS(cmd->device->bin_preamble_bv_entry.size / sizeof(uint32_t)) | CP_SET_AMBLE_2_TYPE(BIN_PREAMBLE_AMBLE_TYPE)); tu7_thread_control(cs, CP_SET_THREAD_BOTH); tu7_set_pred_mask(cs, (1u << TU_PREDICATE_VTX_STATS_RUNNING) | (1u << TU_PREDICATE_VTX_STATS_NOT_RUNNING), (1u << TU_PREDICATE_VTX_STATS_NOT_RUNNING)); } tu_cs_emit_pkt7(cs, CP_SET_AMBLE, 3); tu_cs_emit_qw(cs, 0); tu_cs_emit(cs, CP_SET_AMBLE_2_TYPE(PREAMBLE_AMBLE_TYPE)); tu_cs_emit_pkt7(cs, CP_SET_AMBLE, 3); tu_cs_emit_qw(cs, 0); tu_cs_emit(cs, CP_SET_AMBLE_2_TYPE(POSTAMBLE_AMBLE_TYPE)); if (CHIP >= A7XX) { tu7_thread_control(cs, CP_SET_THREAD_BR); } tu_cs_sanity_check(cs); } bool tu_enable_fdm_offset(struct tu_cmd_buffer *cmd) { if (!cmd->state.pass) return false; if (!cmd->state.pass->has_fdm) return false; unsigned fdm_a = cmd->state.pass->fragment_density_map.attachment; if (fdm_a == VK_ATTACHMENT_UNUSED) return TU_DEBUG(FDM_OFFSET); const struct tu_image_view *fdm = cmd->state.attachments[fdm_a]; return fdm->image->vk.create_flags & VK_IMAGE_CREATE_FRAGMENT_DENSITY_MAP_OFFSET_BIT_EXT; } static void update_vsc_pipe(struct tu_cmd_buffer *cmd, struct tu_cs *cs, uint32_t num_vsc_pipes) { const struct tu_tiling_config *tiling = cmd->state.tiling; const struct tu_vsc_config *vsc = tu_vsc_config(cmd, tiling); tu_cs_emit_regs(cs, A6XX_VSC_BIN_SIZE(.width = tiling->tile0.width, .height = tiling->tile0.height)); tu_cs_emit_regs(cs, A6XX_VSC_EXPANDED_BIN_CNTL(.nx = vsc->tile_count.width, .ny = vsc->tile_count.height)); tu_cs_emit_pkt4(cs, REG_A6XX_VSC_PIPE_CONFIG_REG(0), num_vsc_pipes); tu_cs_emit_array(cs, vsc->pipe_config, num_vsc_pipes); tu_cs_emit_regs(cs, A6XX_VSC_PIPE_DATA_PRIM_STRIDE(cmd->vsc_prim_strm_pitch), A6XX_VSC_PIPE_DATA_PRIM_LENGTH(cmd->vsc_prim_strm_pitch - VSC_PAD)); tu_cs_emit_regs(cs, A6XX_VSC_PIPE_DATA_DRAW_STRIDE(cmd->vsc_draw_strm_pitch), A6XX_VSC_PIPE_DATA_DRAW_LENGTH(cmd->vsc_draw_strm_pitch - VSC_PAD)); tu_cs_emit_regs(cs, A7XX_VSC_UNKNOWN_0D08(0)); } static void emit_vsc_overflow_test(struct tu_cmd_buffer *cmd, struct tu_cs *cs) { const struct tu_tiling_config *tiling = cmd->state.tiling; const struct tu_vsc_config *vsc = tu_vsc_config(cmd, tiling); const uint32_t used_pipe_count = vsc->pipe_count.width * vsc->pipe_count.height; for (int i = 0; i < used_pipe_count; i++) { tu_cs_emit_pkt7(cs, CP_COND_WRITE5, 8); tu_cs_emit(cs, CP_COND_WRITE5_0_FUNCTION(WRITE_GE) | CP_COND_WRITE5_0_WRITE_MEMORY); tu_cs_emit(cs, REG_A6XX_VSC_PIPE_DATA_DRAW_SIZE(i)); tu_cs_emit(cs, 0); tu_cs_emit(cs, CP_COND_WRITE5_3_REF(cmd->vsc_draw_strm_pitch - VSC_PAD)); tu_cs_emit(cs, CP_COND_WRITE5_4_MASK(~0)); tu_cs_emit_qw(cs, global_iova(cmd, vsc_draw_overflow)); tu_cs_emit(cs, CP_COND_WRITE5_7_WRITE_DATA(cmd->vsc_draw_strm_pitch)); tu_cs_emit_pkt7(cs, CP_COND_WRITE5, 8); tu_cs_emit(cs, CP_COND_WRITE5_0_FUNCTION(WRITE_GE) | CP_COND_WRITE5_0_WRITE_MEMORY); tu_cs_emit(cs, REG_A6XX_VSC_PIPE_DATA_PRIM_SIZE(i)); tu_cs_emit(cs, 0); tu_cs_emit(cs, CP_COND_WRITE5_3_REF(cmd->vsc_prim_strm_pitch - VSC_PAD)); tu_cs_emit(cs, CP_COND_WRITE5_4_MASK(~0)); tu_cs_emit_qw(cs, global_iova(cmd, vsc_prim_overflow)); tu_cs_emit(cs, CP_COND_WRITE5_7_WRITE_DATA(cmd->vsc_prim_strm_pitch)); } tu_cs_emit_pkt7(cs, CP_WAIT_MEM_WRITES, 0); } template static void tu6_emit_binning_pass(struct tu_cmd_buffer *cmd, struct tu_cs *cs, const VkOffset2D *fdm_offsets, bool use_cb) { struct tu_physical_device *phys_dev = cmd->device->physical_device; const struct tu_framebuffer *fb = cmd->state.framebuffer; const struct tu_tiling_config *tiling = cmd->state.tiling; /* If this command buffer may be executed multiple times, then * viewports/scissor states may have been changed by previous executions * and we need to reset them before executing the binning IB. With FDM * offset the viewport also needs to be transformed during the binning * phase. */ if ((!(cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_ONE_TIME_SUBMIT_BIT) || fdm_offsets) && cmd->fdm_bin_patchpoints.size != 0) { unsigned num_views = tu_fdm_num_layers(cmd); VkExtent2D unscaled_frag_areas[num_views]; VkRect2D bins[num_views]; VkOffset2D frag_offsets[num_views]; for (unsigned i = 0; i < num_views; i++) { unscaled_frag_areas[i] = (VkExtent2D) { 1, 1 }; frag_offsets[i] = (VkOffset2D) { 0, 0 }; if (fdm_offsets && !cmd->state.rp.shared_viewport) { /* We need to shift over the viewport and scissor during the * binning pass to match the shift applied when rendering. The way * to do this is to make the per-view bin start negative. In the * actual rendering pass, the per-view bin start is shifted in a * negative direction but the first bin is clipped so that the bin * start is never negative, but we need to do this to avoid * clipping the user scissor to a non-zero common bin start. We * skip patching load/store below in order to avoid patching loads * and stores to a crazy negative-offset bin. The parts of the * framebuffer left or above the origin correspond to the * non-visible parts of the left or top bins that will be * discarded. The framebuffer still needs to extend to the * original bottom and right, to avoid incorrectly clipping the * user scissor, so we need to add to the width and height to * compensate. */ VkOffset2D bin_offset = tu_bin_offset(fdm_offsets[i], tiling); bins[i] = { { -bin_offset.x, -bin_offset.y }, { fb->width + bin_offset.x, fb->height + bin_offset.y }, }; } else { bins[i] = { { 0, 0 }, { fb->width, fb->height } }; } } util_dynarray_foreach (&cmd->fdm_bin_patchpoints, struct tu_fdm_bin_patchpoint, patch) { if (patch->flags & TU_FDM_SKIP_BINNING) continue; tu_cs_emit_pkt7(cs, CP_MEM_WRITE, 2 + patch->size); tu_cs_emit_qw(cs, patch->iova); patch->apply(cmd, cs, patch->data, (VkOffset2D) {0, 0}, frag_offsets, num_views, unscaled_frag_areas, bins); } tu_cs_emit_pkt7(cs, CP_WAIT_MEM_WRITES, 0); tu_cs_emit_pkt7(cs, CP_WAIT_FOR_ME, 0); } uint32_t width = fb->width + (fdm_offsets ? tiling->tile0.width : 0); uint32_t height = fb->height + (fdm_offsets ? tiling->tile0.height : 0); tu6_emit_window_scissor(cs, 0, 0, width - 1, height - 1); tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_MODE(RM6_BIN_VISIBILITY)); tu_cs_emit_pkt7(cs, CP_SET_VISIBILITY_OVERRIDE, 1); tu_cs_emit(cs, 0x1); tu_cs_emit_pkt7(cs, CP_SET_MODE, 1); tu_cs_emit(cs, 0x1); tu_cs_emit_wfi(cs); tu_cs_emit_regs(cs, A6XX_VFD_RENDER_MODE(.render_mode = BINNING_PASS)); update_vsc_pipe(cmd, cs, phys_dev->info->num_vsc_pipes); if (CHIP == A6XX) { tu_cs_emit_regs(cs, A6XX_PC_POWER_CNTL(phys_dev->info->a6xx.magic.PC_POWER_CNTL)); tu_cs_emit_regs(cs, A6XX_VFD_POWER_CNTL(phys_dev->info->a6xx.magic.PC_POWER_CNTL)); } tu_emit_event_write(cmd, cs, FD_VSC_BINNING_START); tu_cs_emit_regs(cs, A6XX_RB_WINDOW_OFFSET(.x = 0, .y = 0)); tu_cs_emit_regs(cs, A6XX_TPL1_WINDOW_OFFSET(.x = 0, .y = 0)); if (use_cb) trace_start_concurrent_binning_ib(&cmd->trace, cs, cmd); else trace_start_binning_ib(&cmd->trace, cs, cmd); /* emit IB to binning drawcmds: */ tu_cs_emit_call(cs, &cmd->draw_cs); if (use_cb) trace_end_concurrent_binning_ib(&cmd->trace, cs); else trace_end_binning_ib(&cmd->trace, cs); /* switching from binning pass to GMEM pass will cause a switch from * PROGRAM_BINNING to PROGRAM, which invalidates const state (XS_CONST states) * so make sure these states are re-emitted * (eventually these states shouldn't exist at all with shader prologue) * only VS and GS are invalidated, as FS isn't emitted in binning pass, * and we don't use HW binning when tesselation is used */ tu_cs_emit_pkt7(cs, CP_SET_DRAW_STATE, 3); tu_cs_emit(cs, CP_SET_DRAW_STATE__0_COUNT(0) | CP_SET_DRAW_STATE__0_DISABLE | CP_SET_DRAW_STATE__0_GROUP_ID(TU_DRAW_STATE_CONST)); tu_cs_emit(cs, CP_SET_DRAW_STATE__1_ADDR_LO(0)); tu_cs_emit(cs, CP_SET_DRAW_STATE__2_ADDR_HI(0)); tu_emit_event_write(cmd, cs, FD_VSC_BINNING_END); /* This flush is probably required because the VSC, which produces the * visibility stream, is a client of UCHE, whereas the CP needs to read the * visibility stream (without caching) to do draw skipping. The * WFI+WAIT_FOR_ME combination guarantees that the binning commands * submitted are finished before reading the VSC regs (in * emit_vsc_overflow_test) or the VSC_DATA buffer directly (implicitly as * part of draws). */ tu_emit_event_write(cmd, cs, FD_CACHE_CLEAN); tu_cs_emit_wfi(cs); tu_cs_emit_pkt7(cs, CP_WAIT_FOR_ME, 0); emit_vsc_overflow_test(cmd, cs); tu_cs_emit_pkt7(cs, CP_SET_VISIBILITY_OVERRIDE, 1); tu_cs_emit(cs, 0x0); tu_cs_emit_pkt7(cs, CP_SET_MODE, 1); tu_cs_emit(cs, 0x0); } static struct tu_draw_state tu_emit_input_attachments(struct tu_cmd_buffer *cmd, const struct tu_subpass *subpass, bool gmem) { const struct tu_tiling_config *tiling = cmd->state.tiling; /* note: we can probably emit input attachments just once for the whole * renderpass, this would avoid emitting both sysmem/gmem versions * * emit two texture descriptors for each input, as a workaround for * d24s8/d32s8, which can be sampled as both float (depth) and integer (stencil) * tu_shader lowers uint input attachment loads to use the 2nd descriptor * in the pair * TODO: a smarter workaround */ if (!subpass->input_count) return (struct tu_draw_state) {}; struct tu_cs_memory texture; VkResult result = tu_cs_alloc(&cmd->sub_cs, subpass->input_count * 2, A6XX_TEX_CONST_DWORDS, &texture); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return (struct tu_draw_state) {}; } for (unsigned i = 0; i < subpass->input_count * 2; i++) { uint32_t a = subpass->input_attachments[i / 2].attachment; if (a == VK_ATTACHMENT_UNUSED) continue; const struct tu_image_view *iview = cmd->state.attachments[a]; const struct tu_render_pass_attachment *att = &cmd->state.pass->attachments[a]; uint32_t dst[A6XX_TEX_CONST_DWORDS]; uint32_t gmem_offset = tu_attachment_gmem_offset(cmd, att, 0); uint32_t cpp = att->cpp; memcpy(dst, iview->view.descriptor, A6XX_TEX_CONST_DWORDS * 4); /* Cube descriptors require a different sampling instruction in shader, * however we don't know whether image is a cube or not until the start * of a renderpass. We have to patch the descriptor to make it compatible * with how it is sampled in shader. */ enum a6xx_tex_type tex_type = (enum a6xx_tex_type) pkt_field_get(A6XX_TEX_CONST_2_TYPE, dst[2]); if (tex_type == A6XX_TEX_CUBE) { dst[2] = pkt_field_set(A6XX_TEX_CONST_2_TYPE, dst[2], A6XX_TEX_2D); uint32_t depth = pkt_field_get(A6XX_TEX_CONST_5_DEPTH, dst[5]); dst[5] = pkt_field_set(A6XX_TEX_CONST_5_DEPTH, dst[5], depth * 6); } if (i % 2 == 1 && att->format == VK_FORMAT_D24_UNORM_S8_UINT) { /* note this works because spec says fb and input attachments * must use identity swizzle * * Also we clear swap to WZYX. This is because the view might have * picked XYZW to work better with border colors. */ dst[0] &= ~(A6XX_TEX_CONST_0_FMT__MASK | A6XX_TEX_CONST_0_SWAP__MASK | A6XX_TEX_CONST_0_SWIZ_X__MASK | A6XX_TEX_CONST_0_SWIZ_Y__MASK | A6XX_TEX_CONST_0_SWIZ_Z__MASK | A6XX_TEX_CONST_0_SWIZ_W__MASK); if (!cmd->device->physical_device->info->a6xx.has_z24uint_s8uint) { dst[0] |= A6XX_TEX_CONST_0_FMT(FMT6_8_8_8_8_UINT) | A6XX_TEX_CONST_0_SWIZ_X(A6XX_TEX_W) | A6XX_TEX_CONST_0_SWIZ_Y(A6XX_TEX_ZERO) | A6XX_TEX_CONST_0_SWIZ_Z(A6XX_TEX_ZERO) | A6XX_TEX_CONST_0_SWIZ_W(A6XX_TEX_ONE); } else { dst[0] |= A6XX_TEX_CONST_0_FMT(FMT6_Z24_UINT_S8_UINT) | A6XX_TEX_CONST_0_SWIZ_X(A6XX_TEX_Y) | A6XX_TEX_CONST_0_SWIZ_Y(A6XX_TEX_ZERO) | A6XX_TEX_CONST_0_SWIZ_Z(A6XX_TEX_ZERO) | A6XX_TEX_CONST_0_SWIZ_W(A6XX_TEX_ONE); } } if (i % 2 == 1 && att->format == VK_FORMAT_D32_SFLOAT_S8_UINT) { dst[0] = pkt_field_set(A6XX_TEX_CONST_0_FMT, dst[0], FMT6_8_UINT); dst[2] = pkt_field_set(A6XX_TEX_CONST_2_PITCHALIGN, dst[2], 0); dst[2] = pkt_field_set(A6XX_TEX_CONST_2_PITCH, dst[2], iview->stencil_pitch); dst[3] = 0; dst[4] = iview->stencil_base_addr; dst[5] = (dst[5] & 0xffff) | iview->stencil_base_addr >> 32; cpp = att->samples; gmem_offset = att->gmem_offset_stencil[cmd->state.gmem_layout]; } if (!gmem || !subpass->input_attachments[i / 2].patch_input_gmem) { memcpy(&texture.map[i * A6XX_TEX_CONST_DWORDS], dst, sizeof(dst)); continue; } /* patched for gmem */ dst[0] = pkt_field_set(A6XX_TEX_CONST_0_TILE_MODE, dst[0], TILE6_2); if (!iview->view.is_mutable) dst[0] = pkt_field_set(A6XX_TEX_CONST_0_SWAP, dst[0], WZYX); /* If FDM offset is used, the last row and column extend beyond the * framebuffer but are shifted over when storing. Expand the width and * height to account for that. */ if (tu_enable_fdm_offset(cmd)) { uint32_t width = pkt_field_get(A6XX_TEX_CONST_1_WIDTH, dst[1]); uint32_t height = pkt_field_get(A6XX_TEX_CONST_1_HEIGHT, dst[1]); width += cmd->state.tiling->tile0.width; height += cmd->state.tiling->tile0.height; dst[1] = pkt_field_set(A6XX_TEX_CONST_1_WIDTH, dst[1], width); dst[1] = pkt_field_set(A6XX_TEX_CONST_1_HEIGHT, dst[1], height); } dst[2] = A6XX_TEX_CONST_2_TYPE(A6XX_TEX_2D) | A6XX_TEX_CONST_2_PITCH(tiling->tile0.width * cpp); /* Note: it seems the HW implicitly calculates the array pitch with the * GMEM tiling, so we don't need to specify the pitch ourselves. */ dst[3] = 0; dst[4] = cmd->device->physical_device->gmem_base + gmem_offset; dst[5] &= A6XX_TEX_CONST_5_DEPTH__MASK; for (unsigned i = 6; i < A6XX_TEX_CONST_DWORDS; i++) dst[i] = 0; memcpy(&texture.map[i * A6XX_TEX_CONST_DWORDS], dst, sizeof(dst)); } struct tu_cs cs; struct tu_draw_state ds = tu_cs_draw_state(&cmd->sub_cs, &cs, 9); tu_cs_emit_pkt7(&cs, CP_LOAD_STATE6_FRAG, 3); tu_cs_emit(&cs, CP_LOAD_STATE6_0_DST_OFF(0) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_CONSTANTS) | CP_LOAD_STATE6_0_STATE_SRC(SS6_INDIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(SB6_FS_TEX) | CP_LOAD_STATE6_0_NUM_UNIT(subpass->input_count * 2)); tu_cs_emit_qw(&cs, texture.iova); tu_cs_emit_regs(&cs, A6XX_SP_PS_TEXMEMOBJ_BASE(.qword = texture.iova)); tu_cs_emit_regs(&cs, A6XX_SP_PS_TSIZE(subpass->input_count * 2)); assert(cs.cur == cs.end); /* validate draw state size */ return ds; } static void tu_set_input_attachments(struct tu_cmd_buffer *cmd, const struct tu_subpass *subpass) { struct tu_cs *cs = &cmd->draw_cs; tu_cs_emit_pkt7(cs, CP_SET_DRAW_STATE, 6); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_INPUT_ATTACHMENTS_GMEM, tu_emit_input_attachments(cmd, subpass, true)); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_INPUT_ATTACHMENTS_SYSMEM, tu_emit_input_attachments(cmd, subpass, false)); } static void tu_trace_start_render_pass(struct tu_cmd_buffer *cmd) { if (!u_trace_enabled(&cmd->device->trace_context)) return; uint32_t load_cpp = 0; uint32_t store_cpp = 0; uint32_t clear_cpp = 0; bool has_depth = false; char ubwc[MAX_RTS + 3]; for (uint32_t i = 0; i < cmd->state.pass->attachment_count; i++) { const struct tu_render_pass_attachment *attachment = &cmd->state.pass->attachments[i]; if (attachment->load) { load_cpp += attachment->cpp; } if (attachment->store) { store_cpp += attachment->cpp; } if (attachment->clear_mask) { clear_cpp += attachment->cpp; } has_depth |= vk_format_has_depth(attachment->format); } uint8_t ubwc_len = 0; const struct tu_subpass *subpass = &cmd->state.pass->subpasses[0]; for (uint32_t i = 0; i < subpass->color_count; i++) { uint32_t att = subpass->color_attachments[i].attachment; ubwc[ubwc_len++] = att == VK_ATTACHMENT_UNUSED ? '-' : cmd->state.attachments[att]->view.ubwc_enabled ? 'y' : 'n'; } if (subpass->depth_used) { ubwc[ubwc_len++] = '|'; ubwc[ubwc_len++] = cmd->state.attachments[subpass->depth_stencil_attachment.attachment] ->view.ubwc_enabled ? 'y' : 'n'; } ubwc[ubwc_len] = '\0'; uint32_t max_samples = 0; for (uint32_t i = 0; i < cmd->state.pass->subpass_count; i++) { max_samples = MAX2(max_samples, cmd->state.pass->subpasses[i].samples); } trace_start_render_pass(&cmd->trace, &cmd->cs, cmd, cmd->state.framebuffer, cmd->state.tiling, max_samples, clear_cpp, load_cpp, store_cpp, has_depth, ubwc); } template static void tu_trace_end_render_pass(struct tu_cmd_buffer *cmd, bool gmem) { if (!u_trace_enabled(&cmd->device->trace_context)) return; uint32_t avg_per_sample_bandwidth = cmd->state.rp.drawcall_bandwidth_per_sample_sum / MAX2(cmd->state.rp.drawcall_count, 1); struct u_trace_address addr = {}; if (cmd->state.lrz.image_view) { struct tu_image *image = cmd->state.lrz.image_view->image; addr.bo = image->mem->bo; addr.offset = (image->iova - image->mem->iova) + image->lrz_layout.lrz_fc_offset + offsetof(fd_lrzfc_layout, buffer[0].dir_track); } int32_t lrz_disabled_at_draw = cmd->state.rp.lrz_disabled_at_draw ? cmd->state.rp.lrz_disabled_at_draw : -1; int32_t lrz_write_disabled_at_draw = cmd->state.rp.lrz_write_disabled_at_draw ? cmd->state.rp.lrz_write_disabled_at_draw : -1; trace_end_render_pass( &cmd->trace, &cmd->cs, gmem, cmd->state.rp.gmem_disable_reason ? cmd->state.rp.gmem_disable_reason : "", cmd->state.rp.drawcall_count, avg_per_sample_bandwidth, cmd->state.lrz.valid, cmd->state.rp.lrz_disable_reason ? cmd->state.rp.lrz_disable_reason : "", lrz_disabled_at_draw, lrz_write_disabled_at_draw, addr); } static void tu_emit_renderpass_begin(struct tu_cmd_buffer *cmd) { /* We need to re-emit any draw states that are patched in order for them to * be correctly added to the per-renderpass patchpoint list, even if they * are the same as before. */ if (cmd->state.pass->has_fdm) cmd->state.dirty |= TU_CMD_DIRTY_FDM; /* We need to re-emit MSAA at the beginning of every renderpass because it * isn't part of a draw state that gets automatically re-emitted. */ BITSET_SET(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_MS_RASTERIZATION_SAMPLES); /* PC_CNTL isn't a part of a draw state and may be changed * by blits. */ BITSET_SET(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_IA_PRIMITIVE_RESTART_ENABLE); cmd->state.fdm_enabled = cmd->state.pass->has_fdm; } static bool tu7_emit_concurrent_binning(struct tu_cmd_buffer *cmd, struct tu_cs *cs, bool disable_cb) { if (disable_cb || /* LRZ can only be cleared via fast clear in BV. Disable CB if we can't * use it. */ !cmd->state.lrz.fast_clear || TU_DEBUG(NO_CONCURRENT_BINNING)) { tu_cs_emit_pkt7(cs, CP_THREAD_CONTROL, 1); tu_cs_emit(cs, CP_THREAD_CONTROL_0_THREAD(CP_SET_THREAD_BR) | CP_THREAD_CONTROL_0_CONCURRENT_BIN_DISABLE); tu7_set_pred_bit(cs, TU_PREDICATE_CB_ENABLED, false); return false; } tu7_thread_control(cs, CP_SET_THREAD_BOTH); /* Increment timestamp to make it unique in subsequent commands */ tu_cs_emit_pkt7(cs, CP_MODIFY_TIMESTAMP, 1); tu_cs_emit(cs, CP_MODIFY_TIMESTAMP_0_ADD(1) | CP_MODIFY_TIMESTAMP_0_OP(MODIFY_TIMESTAMP_ADD_LOCAL)); /* We initialize the "is concurrent binning enabled?" predicate to true and * disable it later if necessary. */ tu7_set_pred_bit(cs, TU_PREDICATE_CB_ENABLED, true); tu7_thread_control(cs, CP_SET_THREAD_BV); /* If there was an overflow in the BR resource table the register will be * set to 1 by CP_RESOURCE_LIST. Wait for it to clear here. */ tu7_wait_onchip_val(cs, TU_ONCHIP_CB_RESLIST_OVERFLOW, 0); tu_lrz_cb_begin(cmd, cs); return true; } template static void tu6_sysmem_render_begin(struct tu_cmd_buffer *cmd, struct tu_cs *cs, struct tu_renderpass_result *autotune_result) { const struct tu_framebuffer *fb = cmd->state.framebuffer; /* It seems that for sysmem render passes we have to use BV to clear LRZ * before the renderpass. Otherwise the clear doesn't become visible to * subsequent draws when LRZ has been flipped an odd number of times. * Presumably this works if concurrent binning is disabled, because the * blob relies on this, but that requires synchronizing BR and BV * unnecessarily, and we want BV to skip ahead across sysmem renderpasses. * * In the future, we may also support writing LRZ in BV. */ bool concurrent_binning = false; if (CHIP >= A7XX) { concurrent_binning = tu7_emit_concurrent_binning(cmd, cs, false); tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_MODE(RM6_BIN_VISIBILITY)); } tu_lrz_sysmem_begin(cmd, cs); if (concurrent_binning) { tu_lrz_after_bv(cmd, cs); tu7_write_onchip_timestamp(cs, TU_ONCHIP_CB_BV_TIMESTAMP); tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_MODE(RM7_BIN_VISIBILITY_END)); tu7_thread_control(cs, CP_SET_THREAD_BR); tu7_wait_onchip_timestamp(cs, TU_ONCHIP_CB_BV_TIMESTAMP); tu_lrz_before_sysmem_br(cmd, cs); } assert(fb->width > 0 && fb->height > 0); tu6_emit_window_scissor(cs, 0, 0, fb->width - 1, fb->height - 1); tu6_emit_window_offset(cs, 0, 0); tu6_emit_bin_size(cs, 0, 0, { .render_mode = RENDERING_PASS, .force_lrz_write_dis = !cmd->device->physical_device->info->a6xx.has_lrz_feedback, .buffers_location = BUFFERS_IN_SYSMEM, .lrz_feedback_zmode_mask = cmd->device->physical_device->info->a6xx.has_lrz_feedback ? LRZ_FEEDBACK_EARLY_Z_OR_EARLY_Z_LATE_Z : LRZ_FEEDBACK_NONE, }); if (CHIP == A7XX) { tu_cs_emit_regs(cs, RB_BUFFER_CNTL(CHIP, .z_sysmem = true, .s_sysmem = true, .rt0_sysmem = true, .rt1_sysmem = true, .rt2_sysmem = true, .rt3_sysmem = true, .rt4_sysmem = true, .rt5_sysmem = true, .rt6_sysmem = true, .rt7_sysmem = true, )); tu_cs_emit_regs(cs, A7XX_RB_CCU_DBG_ECO_CNTL(cmd->device->physical_device->info->a6xx.magic.RB_CCU_DBG_ECO_CNTL)); tu_cs_emit_regs(cs, GRAS_MODE_CNTL(A7XX, 0x2)); tu_cs_emit_regs(cs, A7XX_RB_UNKNOWN_8E09(0x4)); tu_cs_emit_regs(cs, A7XX_RB_CLEAR_TARGET(.clear_mode = CLEAR_MODE_SYSMEM)); } tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_MODE(RM6_DIRECT_RENDER)); /* A7XX TODO: blob doesn't use CP_SKIP_IB2_ENABLE_* */ tu_cs_emit_pkt7(cs, CP_SKIP_IB2_ENABLE_GLOBAL, 1); tu_cs_emit(cs, 0x0); tu_emit_cache_flush_ccu(cmd, cs, TU_CMD_CCU_SYSMEM); tu_cs_emit_pkt7(cs, CP_SET_VISIBILITY_OVERRIDE, 1); tu_cs_emit(cs, 0x1); tu_cs_emit_pkt7(cs, CP_SET_MODE, 1); tu_cs_emit(cs, 0x0); /* Reset bin scaling. */ if (cmd->device->physical_device->info->a7xx.has_hw_bin_scaling) { tu_cs_emit_regs(cs, A7XX_GRAS_BIN_FOVEAT()); tu_cs_emit_regs(cs, A7XX_RB_BIN_FOVEAT()); } tu_autotune_begin_renderpass(cmd, cs, autotune_result); tu_cs_sanity_check(cs); } template static void tu6_sysmem_render_end(struct tu_cmd_buffer *cmd, struct tu_cs *cs, struct tu_renderpass_result *autotune_result) { tu_autotune_end_renderpass(cmd, cs, autotune_result); /* Do any resolves of the last subpass. These are handled in the * tile_store_cs in the gmem path. */ tu6_emit_sysmem_resolves(cmd, cs, cmd->state.subpass); tu_cs_emit_call(cs, &cmd->draw_epilogue_cs); tu_cs_emit_pkt7(cs, CP_SKIP_IB2_ENABLE_GLOBAL, 1); tu_cs_emit(cs, 0x0); tu_lrz_sysmem_end(cmd, cs); /* Clear the resource list for any LRZ resources we emitted at the * beginning. */ if (CHIP >= A7XX) { tu_cs_emit_pkt7(cs, CP_EVENT_WRITE7, 4); tu_cs_emit(cs, CP_EVENT_WRITE7_0_EVENT(DUMMY_EVENT) | CP_EVENT_WRITE7_0_CLEAR_RENDER_RESOURCE | CP_EVENT_WRITE7_0_WRITE_DST(EV_DST_ONCHIP) | CP_EVENT_WRITE7_0_WRITE_SRC(EV_WRITE_USER_32B) | CP_EVENT_WRITE7_0_WRITE_ENABLED); tu_cs_emit_qw(cs, TU_ONCHIP_CB_RESLIST_OVERFLOW); tu_cs_emit(cs, 0); /* value */ } tu_cs_sanity_check(cs); } static void tu7_write_and_wait_onchip_timestamp(struct tu_cs *cs, enum tu_onchip_addr onchip_addr) { tu7_write_onchip_timestamp(cs, onchip_addr); tu7_wait_onchip_timestamp(cs, onchip_addr); } static bool tu7_emit_concurrent_binning_gmem(struct tu_cmd_buffer *cmd, struct tu_cs *cs, bool use_hw_binning) { /* xfb queries use data from the binning pass. If they are running outside * of a RP then we may have to deal with a mix of GMEM/sysmem renderpasses * where the counters increase on different processors. Just disable CB so * that everything happens on BR and we don't need difficult merging of BV * and BR results. In addition, RBBM primitive counters seem to not work * at all with concurrent binning, so disable if they are running before * the RP. */ bool disable_cb = cmd->state.xfb_query_running_before_rp || cmd->state.rp.has_prim_generated_query_in_rp || cmd->state.rp.has_vtx_stats_query_in_rp || cmd->state.prim_counters_running > 0; if (!tu7_emit_concurrent_binning(cmd, cs, disable_cb || !use_hw_binning)) return false; /* We want to disable concurrent binning if BV isn't far enough ahead of * BR. The core idea is to write a timestamp in BR and BV, and compare the * BR and BV timestamps for equality. if BR is fast enough, it will write * the timestamp ahead of BV and then when BV compares for equality it will * find them equal. BR cannot race too far ahead of BV because it must wait * for BV's determination to finish, which we do via another timestamp, so * either BV is ahead of BR or the timestamps are equal. * * We need to communicate the determination from BV to BR so they both * agree on whether concurrent binning is enabled or not. The easiest way * to do it is via a "when was concurrent binning last disabled" timestamp, * because we only have to set it when disabling concurrent binning. */ if (!TU_DEBUG(FORCE_CONCURRENT_BINNING)) { tu7_write_and_wait_onchip_timestamp(cs, TU_ONCHIP_CB_BV_TIMESTAMP); tu7_thread_control(cs, CP_SET_THREAD_BR); tu7_write_and_wait_onchip_timestamp(cs, TU_ONCHIP_CB_BR_TIMESTAMP); tu7_thread_control(cs, CP_SET_THREAD_BV); /* If in a secondary, dynamically disable CB if a vtx stats query is * running. */ if (cmd->vk.level == VK_COMMAND_BUFFER_LEVEL_SECONDARY) { tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(PRED_TEST) | CP_COND_REG_EXEC_0_PRED_BIT(TU_PREDICATE_VTX_STATS_RUNNING)); } const uint32_t bv_cond_dwords = 3 + 4 + 4; tu_cs_reserve(cs, 4 + bv_cond_dwords); tu_cs_emit_pkt7(cs, CP_COND_REG_EXEC, 3); tu_cs_emit(cs, CP_COND_REG_EXEC_0_MODE(REG_COMPARE) | CP_COND_REG_EXEC_0_REG0(TU_ONCHIP_CB_BR_TIMESTAMP) | CP_COND_REG_EXEC_0_ONCHIP_MEM); tu_cs_emit(cs, REG_COMPARE_CP_COND_REG_EXEC_1_REG1(TU_ONCHIP_CB_BV_TIMESTAMP) | REG_COMPARE_CP_COND_REG_EXEC_1_ONCHIP_MEM); tu_cs_emit(cs, bv_cond_dwords); if (cmd->vk.level == VK_COMMAND_BUFFER_LEVEL_SECONDARY) tu_cond_exec_end(cs); /* if (BR_TIMESTAMP == BV_TIMESTAMP) */ { tu7_write_and_wait_onchip_timestamp(cs, TU_ONCHIP_CB_BV_DISABLED_TIMESTAMP); tu7_set_pred_bit(cs, TU_PREDICATE_CB_ENABLED, false); } tu7_write_onchip_timestamp(cs, TU_ONCHIP_CB_BV_DETERMINATION_FINISHED_TIMESTAMP); tu7_thread_control(cs, CP_SET_THREAD_BR); tu7_wait_onchip_timestamp(cs, TU_ONCHIP_CB_BV_DETERMINATION_FINISHED_TIMESTAMP); const uint32_t br_cond_dwords = 4; tu_cs_reserve(cs, 4 + br_cond_dwords); tu_cs_emit_pkt7(cs, CP_COND_REG_EXEC, 3); tu_cs_emit(cs, CP_COND_REG_EXEC_0_MODE(REG_COMPARE) | CP_COND_REG_EXEC_0_REG0(TU_ONCHIP_CB_BR_TIMESTAMP) | CP_COND_REG_EXEC_0_ONCHIP_MEM); tu_cs_emit(cs, REG_COMPARE_CP_COND_REG_EXEC_1_REG1(TU_ONCHIP_CB_BV_DISABLED_TIMESTAMP) | REG_COMPARE_CP_COND_REG_EXEC_1_ONCHIP_MEM); tu_cs_emit(cs, br_cond_dwords); /* if (BR_TIMESTAMP == BV_DISABLED_TIMESTAMP) */ { tu7_set_pred_bit(cs, TU_PREDICATE_CB_ENABLED, false); } } /* At this point BV and BR are agreed on whether CB is enabled. If CB is * enabled, set the thread to BV for the binning pass, otherwise set BR and * disable concurrent binning. */ tu7_thread_control(cs, CP_SET_THREAD_BOTH); const uint32_t if_dwords = 5; const uint32_t else_dwords = 2; tu_cs_reserve(cs, 3 + if_dwords + else_dwords); tu_cs_emit_pkt7(cs, CP_COND_REG_EXEC, 2); tu_cs_emit(cs, CP_COND_REG_EXEC_0_MODE(PRED_TEST) | CP_COND_REG_EXEC_0_PRED_BIT(TU_PREDICATE_CB_ENABLED) | CP_COND_REG_EXEC_0_SKIP_WAIT_FOR_ME); tu_cs_emit(cs, if_dwords); /* if (CB is enabled) */ { tu7_thread_control(cs, CP_SET_THREAD_BV); /* Wait for BR vis stream reads to finish */ tu_cs_emit_pkt7(cs, CP_BV_BR_COUNT_OPS, 1); tu_cs_emit(cs, CP_BV_BR_COUNT_OPS_0_OP(PIPE_BV_WAIT_FOR_BR)); /* This is the NOP-as-else trick. If CB is disabled, this CP_NOP is * skipped and its body (the else) is executed. */ tu_cs_emit_pkt7(cs, CP_NOP, else_dwords); } /* else */ { tu_cs_emit_pkt7(cs, CP_THREAD_CONTROL, 1); tu_cs_emit(cs, CP_THREAD_CONTROL_0_THREAD(CP_SET_THREAD_BR) | CP_THREAD_CONTROL_0_CONCURRENT_BIN_DISABLE); } return true; } template static void tu6_tile_render_begin(struct tu_cmd_buffer *cmd, struct tu_cs *cs, struct tu_renderpass_result *autotune_result, const VkOffset2D *fdm_offsets) { struct tu_physical_device *phys_dev = cmd->device->physical_device; const struct tu_tiling_config *tiling = cmd->state.tiling; const struct tu_vsc_config *vsc = tu_vsc_config(cmd, tiling); const struct tu_render_pass *pass = cmd->state.pass; bool use_binning = use_hw_binning(cmd); /* User flushes should always be executed on BR. */ tu_emit_cache_flush_ccu(cmd, cs, TU_CMD_CCU_GMEM); bool use_cb = false; if (CHIP >= A7XX) { tu7_emit_tile_render_begin_regs(cs); use_cb = tu7_emit_concurrent_binning_gmem(cmd, cs, use_binning); } if (!use_cb) tu_trace_start_render_pass(cmd); tu_lrz_tiling_begin(cmd, cs); /* tu_lrz_tiling_begin() can accumulate additional flushes. If that happens * CB should be disabled, so it's safe to just emit them here. */ tu_emit_cache_flush_ccu(cmd, cs, TU_CMD_CCU_GMEM); tu_cs_emit_pkt7(cs, CP_SKIP_IB2_ENABLE_GLOBAL, 1); tu_cs_emit(cs, 0x0); /* Reset bin scaling. */ if (phys_dev->info->a7xx.has_hw_bin_scaling) { tu_cs_emit_regs(cs, A7XX_GRAS_BIN_FOVEAT()); tu_cs_emit_regs(cs, A7XX_RB_BIN_FOVEAT()); } if (use_binning) { if (!cmd->vsc_initialized) { tu6_lazy_init_vsc(cmd); } /* We always emit VSC before each renderpass, because due to * skipsaverestore the underlying VSC registers may have become * invalid. Normally we'd need to WFI before setting these non-context * registers, but we should be safe because we're only setting it to the * same value it had before. * * TODO: On a6xx, we have to emit this per-bin or make the amble include * these registers, because CP_SET_BIN_DATA5_OFFSET will use the * register instead of the pseudo register and its value won't survive * across preemptions. The blob seems to take the second approach and * emits the preamble lazily. We chose the per-bin approach but blob's * should be a better one. */ if (cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT) tu_cs_set_writeable(cs, true); tu_emit_vsc(cmd, cs); if (cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT) tu_cs_set_writeable(cs, false); tu6_emit_bin_size(cs, tiling->tile0.width, tiling->tile0.height, { .render_mode = BINNING_PASS, .buffers_location = BUFFERS_IN_GMEM, .lrz_feedback_zmode_mask = phys_dev->info->a6xx.has_lrz_feedback ? LRZ_FEEDBACK_EARLY_Z_LATE_Z : LRZ_FEEDBACK_NONE }); tu6_emit_render_cntl(cmd, cmd->state.subpass, cs, true); tu6_emit_binning_pass(cmd, cs, fdm_offsets, use_cb); if (CHIP == A6XX) { tu_cs_emit_regs(cs, A6XX_PC_POWER_CNTL(phys_dev->info->a6xx.magic.PC_POWER_CNTL)); tu_cs_emit_regs(cs, A6XX_VFD_POWER_CNTL(phys_dev->info->a6xx.magic.PC_POWER_CNTL)); } /* Enable early return from CP_INDIRECT_BUFFER once the visibility stream * is done. We don't enable this if there are stores in a non-final * subpass, because it's more important to be able to share gmem space * between attachments by storing early, than it is to do IB2 skipping * (which has an effect we struggle to even measure). */ if (pass->allow_ib2_skipping) { tu_cs_emit_pkt7(cs, CP_SKIP_IB2_ENABLE_GLOBAL, 1); tu_cs_emit(cs, 0x1); tu_cs_emit_pkt7(cs, CP_SKIP_IB2_ENABLE_LOCAL, 1); tu_cs_emit(cs, 0x1); } } else { if (vsc->binning_possible) { /* Mark all tiles as visible for tu6_emit_cond_for_load_stores(), since * the actual binner didn't run. */ int pipe_count = vsc->pipe_count.width * vsc->pipe_count.height; tu_cs_emit_pkt4(cs, REG_A6XX_VSC_CHANNEL_VISIBILITY(0), pipe_count); for (int i = 0; i < pipe_count; i++) tu_cs_emit(cs, ~0); } } if (vsc->binning_possible) { /* On a7xx we always need VSC allocated because the VSC state has to go * together with other stream data. We could allocate just the VSC state * if binning is disabled but it doesn't seem worth it. */ if (CHIP >= A7XX && !cmd->vsc_initialized) tu6_lazy_init_vsc(cmd); /* Upload state regs to memory to be restored on skipsaverestore * preemption. On a7xx this is considered part of the vis stream that * requires a patchpoint. */ if (CHIP >= A7XX && (cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT)) tu_cs_set_writeable(cs, true); tu_cs_emit_pkt7(cs, CP_REG_TO_MEM, 3); tu_cs_emit(cs, CP_REG_TO_MEM_0_REG(REG_A6XX_VSC_CHANNEL_VISIBILITY(0)) | CP_REG_TO_MEM_0_CNT(32)); if (CHIP >= A7XX) tu_emit_vis_stream_patchpoint(cmd, cs, cmd->vsc_state_offset); else tu_cs_emit_qw(cs, global_iova(cmd, vsc_state)); if (CHIP >= A7XX) { uint32_t num_vsc_pipes = phys_dev->info->num_vsc_pipes; tu_cs_emit_pkt7(cs, CP_WAIT_MEM_WRITES, 0); tu_cs_emit_pkt7(cs, CP_WAIT_FOR_ME, 0); if (use_binning) { tu_cs_emit_pkt7(cs, CP_THREAD_CONTROL, 1); tu_cs_emit(cs, CP_THREAD_CONTROL_0_THREAD(CP_SET_THREAD_BV)); tu_lrz_after_bv(cmd, cs); /* Signal that BV is done for this render pass. This always has to * be executed, even when CB is dynamically disabled, because we * need to keep BR and BV counts in sync with which visibility * streams are in use. */ tu_cs_emit_pkt7(cs, CP_EVENT_WRITE7, 1); tu_cs_emit(cs, CP_EVENT_WRITE7_0_EVENT(DUMMY_EVENT) | CP_EVENT_WRITE7_0_INC_BV_COUNT); /* This mode seems to be only used by BV and signals that a * simpler save/restore procedure can be used in between render * passes. */ tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_MODE(RM7_BIN_VISIBILITY_END)); } tu7_thread_control(cs, CP_SET_THREAD_BR); if (use_binning) { /* Wait for the BV to be done for this render pass. */ tu_cs_emit_pkt7(cs, CP_BV_BR_COUNT_OPS, 1); tu_cs_emit(cs, CP_BV_BR_COUNT_OPS_0_OP(PIPE_BR_WAIT_FOR_BV)); /* Emit vis stream on BR */ tu_emit_vsc(cmd, cs); } tu_cs_emit_pkt7(cs, CP_MEM_TO_SCRATCH_MEM, 4); tu_cs_emit(cs, num_vsc_pipes); /* count */ tu_cs_emit(cs, 0); /* offset */ tu_emit_vis_stream_patchpoint(cmd, cs, cmd->vsc_state_offset); } if (CHIP >= A7XX && (cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT)) tu_cs_set_writeable(cs, false); } else if (CHIP >= A7XX) { /* Earlier we disabled concurrent binning to make LRZ fast-clear work * with no HW binning, now re-enable it while staying on BR. */ tu7_thread_control(cs, CP_SET_THREAD_BR); } tu_lrz_before_tiles(cmd, cs, use_cb); if (use_cb) tu_trace_start_render_pass(cmd); tu_autotune_begin_renderpass(cmd, cs, autotune_result); tu_cs_sanity_check(cs); } template static void tu6_render_tile(struct tu_cmd_buffer *cmd, struct tu_cs *cs, const struct tu_tile_config *tile, bool fdm, const VkOffset2D *fdm_offsets) { tu6_emit_tile_select(cmd, &cmd->cs, tile, fdm, fdm_offsets); tu_lrz_before_tile(cmd, &cmd->cs); trace_start_draw_ib_gmem(&cmd->trace, &cmd->cs, cmd); /* Primitives that passed all tests are still counted in in each * tile even with HW binning beforehand. Do not permit it. */ if (cmd->state.prim_generated_query_running_before_rp) tu_emit_event_write(cmd, cs, FD_STOP_PRIMITIVE_CTRS); tu_cs_emit_call(cs, &cmd->draw_cs); if (cmd->state.prim_generated_query_running_before_rp) tu_emit_event_write(cmd, cs, FD_START_PRIMITIVE_CTRS); if (use_hw_binning(cmd)) { tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_MODE(RM6_BIN_END_OF_DRAWS) | A6XX_CP_SET_MARKER_0_USES_GMEM); } /* Predicate is changed in draw_cs so we have to re-emit it */ if (cmd->state.rp.draw_cs_writes_to_cond_pred && util_is_power_of_two_nonzero(tile->slot_mask)) { uint32_t slot = ffs(tile->slot_mask) - 1; tu6_emit_cond_for_load_stores(cmd, cs, tile->pipe, slot, false); } if (cmd->state.pass->allow_ib2_skipping) { /* Disable CP_INDIRECT_BUFFER/CP_DRAW skipping again at the end of the * pass -- tile_store_cs is for stores that can't be skipped based on * visibility. */ tu_cs_emit_pkt7(cs, CP_SKIP_IB2_ENABLE_GLOBAL, 1); tu_cs_emit(cs, 0x0); } tu_cs_emit_call(cs, &cmd->tile_store_cs); tu_clone_trace_range(cmd, cs, &cmd->trace, cmd->trace_renderpass_start, u_trace_end_iterator(&cmd->rp_trace)); tu_cs_emit_wfi(cs); tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_MODE(RM6_BIN_RENDER_END)); tu_cs_sanity_check(cs); trace_end_draw_ib_gmem(&cmd->trace, &cmd->cs); } template static void tu6_tile_render_end(struct tu_cmd_buffer *cmd, struct tu_cs *cs, struct tu_renderpass_result *autotune_result) { tu_autotune_end_renderpass(cmd, cs, autotune_result); tu_cs_emit_call(cs, &cmd->draw_epilogue_cs); tu_lrz_tiling_end(cmd, cs); bool hw_binning = use_hw_binning(cmd); if (hw_binning) { cmd->state.tile_render_pass_count++; } /* If we are using HW binning, signal that we are done with reading the vis * stream for this render pass by advancing the counter. Also clear render * resources, currently only used for LRZ, and reset the overflow onchip * register. */ if (CHIP >= A7XX) { tu_cs_emit_pkt7(cs, CP_EVENT_WRITE7, 4); tu_cs_emit(cs, CP_EVENT_WRITE7_0_EVENT(DUMMY_EVENT) | COND(hw_binning, CP_EVENT_WRITE7_0_INC_BR_COUNT) | CP_EVENT_WRITE7_0_CLEAR_RENDER_RESOURCE | CP_EVENT_WRITE7_0_WRITE_DST(EV_DST_ONCHIP) | CP_EVENT_WRITE7_0_WRITE_SRC(EV_WRITE_USER_32B) | CP_EVENT_WRITE7_0_WRITE_ENABLED); tu_cs_emit_qw(cs, TU_ONCHIP_CB_RESLIST_OVERFLOW); tu_cs_emit(cs, 0); /* value */ } tu_emit_event_write(cmd, cs, FD_CCU_CLEAN_BLIT_CACHE); tu_cs_sanity_check(cs); } static void tu_calc_frag_area(struct tu_cmd_buffer *cmd, struct tu_tile_config *tile, const struct tu_image_view *fdm, const VkOffset2D *fdm_offsets) { const struct tu_tiling_config *tiling = cmd->state.tiling; const uint32_t x1 = tiling->tile0.width * tile->pos.x; const uint32_t y1 = tiling->tile0.height * tile->pos.y; const uint32_t x2 = MIN2(x1 + tiling->tile0.width, MAX_VIEWPORT_SIZE); const uint32_t y2 = MIN2(y1 + tiling->tile0.height, MAX_VIEWPORT_SIZE); unsigned views = tu_fdm_num_layers(cmd); const struct tu_framebuffer *fb = cmd->state.framebuffer; struct tu_frag_area raw_areas[views]; if (fdm) { for (unsigned i = 0; i < views; i++) { VkOffset2D sample_pos = { 0, 0 }; /* Offsets less than a tile size are accomplished by sliding the * tiles. However once we shift a whole tile size then we reset the * tiles back to where they were at the beginning and we need to * adjust where each bin is sampling from: * * x offset = 0: * * ------------------------------------ * | * | * | * | (unused) | * ------------------------------------ * * x offset = 4: * * ------------------------- * | * | * | * | * | * ------------------------- * * x offset = 8: * * ------------------------------------ * | * | * | * | (unused) | * ------------------------------------ * * As the user's offset increases we slide the tiles to the right, * until we reach the whole tile size and reset the tile positions. * tu_bin_offset() returns an amount to shift to the left, negating * the offset. * * If we were forced to use a shared viewport, then we must not shift * over the tiles and instead must only shift when sampling because * we cannot shift the tiles differently per view. This disables * smooth transitions of the fragment density map and effectively * negates the extension. * * Note that we cannot clamp x2/y2 to the framebuffer size, as we * normally would do, because then tiles along the edge would * incorrectly nudge the sample_pos towards the center of the * framebuffer. If we shift one complete tile over towards the * center and reset the tiles as above, the sample_pos would * then shift back towards the edge and we could get a "pop" from * suddenly changing density due to the slight shift. */ if (fdm_offsets) { VkOffset2D offset = fdm_offsets[i]; if (!cmd->state.rp.shared_viewport) { VkOffset2D bin_offset = tu_bin_offset(fdm_offsets[i], tiling); offset.x += bin_offset.x; offset.y += bin_offset.y; } sample_pos.x = (x1 + x2) / 2 - offset.x; sample_pos.y = (y1 + y2) / 2 - offset.y; } else { sample_pos.x = (x1 + MIN2(x2, fb->width)) / 2; sample_pos.y = (y1 + MIN2(y2, fb->height)) / 2; } tu_fragment_density_map_sample(fdm, sample_pos.x, sample_pos.y, fb->width, fb->height, i, &raw_areas[i]); } } else { for (unsigned i = 0; i < views; i++) raw_areas[i].width = raw_areas[i].height = 1.0f; } for (unsigned i = 0; i < views; i++) { float floor_x, floor_y; float area = raw_areas[i].width * raw_areas[i].height; float frac_x = modff(raw_areas[i].width, &floor_x); float frac_y = modff(raw_areas[i].height, &floor_y); /* The spec allows rounding up one of the axes as long as the total * area is less than or equal to the original area. Take advantage of * this to try rounding up the number with the largest fraction. */ if ((frac_x > frac_y ? (floor_x + 1.f) * floor_y : floor_x * (floor_y + 1.f)) <= area) { if (frac_x > frac_y) floor_x += 1.f; else floor_y += 1.f; } uint32_t width = floor_x; uint32_t height = floor_y; /* Areas that aren't a power of two, especially large areas, can create * in floating-point rounding errors when dividing by the area in the * viewport that result in under-rendering. Round down to a power of two * to make sure all operations are exact. */ width = 1u << util_logbase2(width); height = 1u << util_logbase2(height); /* When FDM offset is enabled, the fragment area has to divide the * offset to make sure that we don't have tiles with partial fragments. * It would be bad to have the fragment area change as a function of the * offset, because we'd get "popping" as the resolution changes with the * offset, so just make sure it divides the offset granularity. This * should mean it always divides the offset for any possible offset. */ if (fdm_offsets) { width = MIN2(width, TU_FDM_OFFSET_GRANULARITY); height = MIN2(height, TU_FDM_OFFSET_GRANULARITY); } /* HW viewport scaling supports a maximum fragment width/height of 4. */ if (views <= MAX_HW_SCALED_VIEWS) { width = MIN2(width, 4); height = MIN2(height, 4); } /* Make sure that the width/height divides the tile width/height so * we don't have to do extra awkward clamping of the edges of each * bin when resolving. It also has to divide the fdm offset, if any. * Note that because the tile width is rounded to a multiple of 32 any * power of two 32 or less will work, and if there is an offset then it * must be a multiple of 4 so 2 or 4 will definitely work. * * TODO: Try to take advantage of the total area allowance here, too. */ while (tiling->tile0.width % width != 0) width /= 2; while (tiling->tile0.height % height != 0) height /= 2; tile->frag_areas[i].width = width; tile->frag_areas[i].height = height; } /* If at any point we were forced to use the same scaling for all * viewports, we need to make sure that any users *not* using shared * scaling, including loads/stores, also consistently share the scaling. */ if (cmd->state.rp.shared_viewport) { VkExtent2D frag_area = { UINT32_MAX, UINT32_MAX }; for (unsigned i = 0; i < views; i++) { frag_area.width = MIN2(frag_area.width, tile->frag_areas[i].width); frag_area.height = MIN2(frag_area.height, tile->frag_areas[i].height); } for (unsigned i = 0; i < views; i++) tile->frag_areas[i] = frag_area; } } static bool try_merge_tiles(struct tu_tile_config *dst, const struct tu_tile_config *src, unsigned views, bool has_abs_bin_mask) { uint32_t slot_mask = dst->slot_mask | src->slot_mask; /* The fragment areas must be the same. */ for (unsigned i = 0; i < views; i++) { if (dst->frag_areas[i].width != src->frag_areas[i].width || dst->frag_areas[i].height != src->frag_areas[i].height) return false; } /* The tiles must be vertically or horizontally adjacent and have the * compatible width/height. */ if (dst->pos.x == src->pos.x) { if (dst->extent.height != src->extent.height) return false; } else if (dst->pos.y == src->pos.y) { if (dst->extent.width != src->extent.width) return false; } else { return false; } if (!has_abs_bin_mask) { /* The mask of the combined tile has to fit in 16 bits */ uint32_t hw_mask = slot_mask >> (ffs(slot_mask) - 1); if ((hw_mask & 0xffff) != hw_mask) return false; } /* Note, this assumes that dst is below or to the right of src, which is * how we call this function below. */ VkExtent2D extent = { dst->extent.width + (dst->pos.x - src->pos.x), dst->extent.height + (dst->pos.y - src->pos.y), }; assert(dst->extent.height > 0); /* The common fragment areas must not be smaller than the combined bin * extent, so that the combined bin is not larger than the original * unscaled bin. */ for (unsigned i = 0; i < views; i++) { if (dst->frag_areas[i].width < extent.width || dst->frag_areas[i].height < extent.height) return false; } /* Ok, let's combine them. dst is below or to the right of src, so it takes * src's position. */ dst->extent = extent; dst->pos = src->pos; dst->slot_mask = slot_mask; return true; } template void tu_render_pipe_fdm(struct tu_cmd_buffer *cmd, uint32_t pipe, uint32_t tx1, uint32_t ty1, uint32_t tx2, uint32_t ty2, const struct tu_image_view *fdm, const VkOffset2D *fdm_offsets) { uint32_t width = tx2 - tx1; uint32_t height = ty2 - ty1; unsigned views = tu_fdm_num_layers(cmd); bool has_abs_mask = cmd->device->physical_device->info->a7xx.has_abs_bin_mask; struct tu_tile_config tiles[width * height]; /* Initialize tiles and sample fragment density map */ for (uint32_t y = 0; y < height; y++) { for (uint32_t x = 0; x < width; x++) { struct tu_tile_config *tile = &tiles[width * y + x]; tile->pos = { x + tx1, y + ty1 }; tile->extent = { 1, 1 }; tile->pipe = pipe; tile->slot_mask = 1u << (width * y + x); tu_calc_frag_area(cmd, tile, fdm, fdm_offsets); } } uint32_t merged_tiles = 0; /* Merge tiles */ for (uint32_t y = 0; y < height; y++) { for (uint32_t x = 0; x < width; x++) { struct tu_tile_config *tile = &tiles[width * y + x]; if (x > 0) { struct tu_tile_config *prev_x_tile = &tiles[width * y + x - 1]; if (try_merge_tiles(tile, prev_x_tile, views, has_abs_mask)) { merged_tiles |= prev_x_tile->slot_mask; } } if (y > 0) { unsigned prev_y_idx = width * (y - 1) + x; struct tu_tile_config *prev_y_tile = &tiles[prev_y_idx]; /* We can't merge prev_y_tile into tile if it's already been * merged horizontally into its neighbor in the previous row. */ if (!(merged_tiles & (1u << prev_y_idx)) && try_merge_tiles(tile, prev_y_tile, views, has_abs_mask)) { merged_tiles |= prev_y_tile->slot_mask; } } } } /* Finally, iterate over tiles and draw them */ for (uint32_t y = 0; y < height; y++) { for (uint32_t x = 0; x < width; x++) { uint32_t tx; if (y & 1) tx = width - 1 - x; else tx = x; unsigned tile_idx = y * width + tx; if (merged_tiles & (1u << tile_idx)) continue; tu6_render_tile(cmd, &cmd->cs, &tiles[tile_idx], true, fdm_offsets); } } } static VkResult tu_allocate_transient_attachments(struct tu_cmd_buffer *cmd, bool sysmem) { const struct tu_framebuffer *fb = cmd->state.framebuffer; const struct tu_render_pass *rp = cmd->state.pass; for (unsigned i = 0; i < fb->attachment_count; i++) { const struct tu_image_view *iview = cmd->state.attachments[i]; if (iview && !(iview->image->vk.create_flags & VK_IMAGE_CREATE_SPARSE_BINDING_BIT) && !iview->image->mem->bo && (sysmem || rp->attachments[i].load || rp->attachments[i].load_stencil || rp->attachments[i].store || rp->attachments[i].store_stencil)) { VkResult result = tu_allocate_lazy_memory(cmd->device, iview->image->mem); if (result != VK_SUCCESS) return result; } } return VK_SUCCESS; } template static void tu_cmd_render_tiles(struct tu_cmd_buffer *cmd, struct tu_renderpass_result *autotune_result, const VkOffset2D *fdm_offsets) { const struct tu_tiling_config *tiling = cmd->state.tiling; const struct tu_vsc_config *vsc = tu_vsc_config(cmd, tiling); const struct tu_image_view *fdm = NULL; VkResult result = tu_allocate_transient_attachments(cmd, false); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } if (cmd->state.pass->fragment_density_map.attachment != VK_ATTACHMENT_UNUSED) { fdm = cmd->state.attachments[cmd->state.pass->fragment_density_map.attachment]; } bool has_fdm = fdm || (TU_DEBUG(FDM) && cmd->state.pass->has_fdm); bool merge_tiles = has_fdm && !TU_DEBUG(NO_BIN_MERGING) && cmd->device->physical_device->info->a6xx.has_bin_mask; /* If not using FDM make sure not to accidentally apply the offsets */ if (!has_fdm) fdm_offsets = NULL; /* Create gmem stores now (at EndRenderPass time)) because they needed to * know whether to allow their conditional execution, which was tied to a * state that was known only at the end of the renderpass. They will be * called from tu6_render_tile(). */ tu_cs_begin(&cmd->tile_store_cs); tu6_emit_tile_store_cs(cmd, &cmd->tile_store_cs); tu_cs_end(&cmd->tile_store_cs); tu6_tile_render_begin(cmd, &cmd->cs, autotune_result, fdm_offsets); /* Note: we reverse the order of walking the pipes and tiles on every * other row, to improve texture cache locality compared to raster order. */ for (uint32_t py = 0; py < vsc->pipe_count.height; py++) { uint32_t pipe_row = py * vsc->pipe_count.width; for (uint32_t pipe_row_i = 0; pipe_row_i < vsc->pipe_count.width; pipe_row_i++) { uint32_t px; if (py & 1) px = vsc->pipe_count.width - 1 - pipe_row_i; else px = pipe_row_i; uint32_t pipe = pipe_row + px; uint32_t tx1 = px * vsc->pipe0.width; uint32_t ty1 = py * vsc->pipe0.height; uint32_t tx2 = MIN2(tx1 + vsc->pipe0.width, vsc->tile_count.width); uint32_t ty2 = MIN2(ty1 + vsc->pipe0.height, vsc->tile_count.height); if (merge_tiles) { tu_render_pipe_fdm(cmd, pipe, tx1, ty1, tx2, ty2, fdm, fdm_offsets); continue; } uint32_t tile_row_stride = tx2 - tx1; uint32_t slot_row = 0; for (uint32_t ty = ty1; ty < ty2; ty++) { for (uint32_t tile_row_i = 0; tile_row_i < tile_row_stride; tile_row_i++) { uint32_t tx; if (ty & 1) tx = tile_row_stride - 1 - tile_row_i; else tx = tile_row_i; struct tu_tile_config tile = { .pos = { tx1 + tx, ty }, .pipe = pipe, .slot_mask = 1u << (slot_row + tx), .extent = { 1, 1 }, }; if (has_fdm) tu_calc_frag_area(cmd, &tile, fdm, fdm_offsets); tu6_render_tile(cmd, &cmd->cs, &tile, has_fdm, fdm_offsets); } slot_row += tile_row_stride; } } } tu6_tile_render_end(cmd, &cmd->cs, autotune_result); tu_trace_end_render_pass(cmd, true); /* We have trashed the dynamically-emitted viewport, scissor, and FS params * via the patchpoints, so we need to re-emit them if they are reused for a * later render pass. */ if (cmd->state.pass->has_fdm) cmd->state.dirty |= TU_CMD_DIRTY_FDM; /* Reset the gmem store CS entry lists so that the next render pass * does its own stores. */ tu_cs_discard_entries(&cmd->tile_store_cs); } template static void tu_cmd_render_sysmem(struct tu_cmd_buffer *cmd, struct tu_renderpass_result *autotune_result) { VkResult result = tu_allocate_transient_attachments(cmd, true); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } tu_trace_start_render_pass(cmd); tu6_sysmem_render_begin(cmd, &cmd->cs, autotune_result); trace_start_draw_ib_sysmem(&cmd->trace, &cmd->cs, cmd); tu_cs_emit_call(&cmd->cs, &cmd->draw_cs); trace_end_draw_ib_sysmem(&cmd->trace, &cmd->cs); tu6_sysmem_render_end(cmd, &cmd->cs, autotune_result); tu_clone_trace_range(cmd, &cmd->cs, &cmd->trace, cmd->trace_renderpass_start, u_trace_end_iterator(&cmd->rp_trace)); tu_trace_end_render_pass(cmd, false); } template void tu_cmd_render(struct tu_cmd_buffer *cmd_buffer, const VkOffset2D *fdm_offsets) { if (cmd_buffer->state.rp.has_tess) tu6_lazy_emit_tessfactor_addr(cmd_buffer); struct tu_renderpass_result *autotune_result = NULL; if (use_sysmem_rendering(cmd_buffer, &autotune_result)) tu_cmd_render_sysmem(cmd_buffer, autotune_result); else tu_cmd_render_tiles(cmd_buffer, autotune_result, fdm_offsets); /* Outside of renderpasses we assume all draw states are disabled. We do * this outside the draw CS for the normal case where 3d gmem stores aren't * used. */ tu_disable_draw_states(cmd_buffer, &cmd_buffer->cs); } static void tu_reset_render_pass(struct tu_cmd_buffer *cmd_buffer) { /* discard draw_cs and draw_epilogue_cs entries now that the tiles are rendered */ tu_cs_discard_entries(&cmd_buffer->draw_cs); tu_cs_begin(&cmd_buffer->draw_cs); tu_cs_discard_entries(&cmd_buffer->draw_epilogue_cs); tu_cs_begin(&cmd_buffer->draw_epilogue_cs); cmd_buffer->state.pass = NULL; cmd_buffer->state.subpass = NULL; cmd_buffer->state.framebuffer = NULL; cmd_buffer->state.attachments = NULL; cmd_buffer->state.clear_values = NULL; cmd_buffer->state.gmem_layout = TU_GMEM_LAYOUT_COUNT; /* invalid value to prevent looking up gmem offsets */ memset(&cmd_buffer->state.rp, 0, sizeof(cmd_buffer->state.rp)); /* LRZ is not valid next time we use it */ cmd_buffer->state.lrz.valid = false; cmd_buffer->state.dirty |= TU_CMD_DIRTY_LRZ; /* Patchpoints have been executed */ util_dynarray_clear(&cmd_buffer->fdm_bin_patchpoints); ralloc_free(cmd_buffer->patchpoints_ctx); cmd_buffer->patchpoints_ctx = NULL; /* Discard RP trace contents */ u_trace_disable_event_range(cmd_buffer->trace_renderpass_start, u_trace_end_iterator(&cmd_buffer->rp_trace)); cmd_buffer->trace_renderpass_start = u_trace_end_iterator(&cmd_buffer->rp_trace); } static VkResult tu_create_cmd_buffer(struct vk_command_pool *pool, VkCommandBufferLevel level, struct vk_command_buffer **cmd_buffer_out) { struct tu_device *device = container_of(pool->base.device, struct tu_device, vk); struct tu_cmd_buffer *cmd_buffer; cmd_buffer = (struct tu_cmd_buffer *) vk_zalloc2( &device->vk.alloc, NULL, sizeof(*cmd_buffer), 8, VK_SYSTEM_ALLOCATION_SCOPE_OBJECT); if (cmd_buffer == NULL) return vk_error(device, VK_ERROR_OUT_OF_HOST_MEMORY); VkResult result = vk_command_buffer_init(pool, &cmd_buffer->vk, &tu_cmd_buffer_ops, level); if (result != VK_SUCCESS) { vk_free2(&device->vk.alloc, NULL, cmd_buffer); return result; } cmd_buffer->device = device; u_trace_init(&cmd_buffer->trace, &device->trace_context); u_trace_init(&cmd_buffer->rp_trace, &device->trace_context); cmd_buffer->trace_renderpass_start = u_trace_begin_iterator(&cmd_buffer->rp_trace); list_inithead(&cmd_buffer->renderpass_autotune_results); if (TU_DEBUG_START(CHECK_CMD_BUFFER_STATUS)) { cmd_buffer->status_bo = tu_cmd_buffer_setup_status_tracking(device); if (cmd_buffer->status_bo == NULL) { mesa_logw("Failed creating cmd_buffer status_bo. " "Won't track status for this cmd_buffer."); } } tu_cs_init(&cmd_buffer->cs, device, TU_CS_MODE_GROW, 4096, "cmd cs"); tu_cs_init(&cmd_buffer->draw_cs, device, TU_CS_MODE_GROW, 4096, "draw cs"); tu_cs_init(&cmd_buffer->tile_store_cs, device, TU_CS_MODE_GROW, 2048, "tile store cs"); tu_cs_init(&cmd_buffer->draw_epilogue_cs, device, TU_CS_MODE_GROW, 4096, "draw epilogue cs"); tu_cs_init(&cmd_buffer->sub_cs, device, TU_CS_MODE_SUB_STREAM, 2048, "draw sub cs"); tu_cs_init(&cmd_buffer->pre_chain.draw_cs, device, TU_CS_MODE_GROW, 4096, "prechain draw cs"); tu_cs_init(&cmd_buffer->pre_chain.draw_epilogue_cs, device, TU_CS_MODE_GROW, 4096, "prechain draw epiligoue cs"); for (unsigned i = 0; i < MAX_BIND_POINTS; i++) cmd_buffer->descriptors[i].push_set.base.type = VK_OBJECT_TYPE_DESCRIPTOR_SET; *cmd_buffer_out = &cmd_buffer->vk; return VK_SUCCESS; } static void tu_cmd_buffer_destroy(struct vk_command_buffer *vk_cmd_buffer) { struct tu_cmd_buffer *cmd_buffer = container_of(vk_cmd_buffer, struct tu_cmd_buffer, vk); tu_cs_finish(&cmd_buffer->cs); tu_cs_finish(&cmd_buffer->draw_cs); tu_cs_finish(&cmd_buffer->tile_store_cs); tu_cs_finish(&cmd_buffer->draw_epilogue_cs); tu_cs_finish(&cmd_buffer->sub_cs); tu_cs_finish(&cmd_buffer->pre_chain.draw_cs); tu_cs_finish(&cmd_buffer->pre_chain.draw_epilogue_cs); if (TU_DEBUG_START(CHECK_CMD_BUFFER_STATUS)) { tu_cmd_buffer_status_check_idle(cmd_buffer); tu_bo_unmap(cmd_buffer->device, cmd_buffer->status_bo, false); tu_bo_finish(cmd_buffer->device, cmd_buffer->status_bo); } u_trace_fini(&cmd_buffer->trace); u_trace_fini(&cmd_buffer->rp_trace); tu_autotune_free_results(cmd_buffer->device, &cmd_buffer->renderpass_autotune_results); for (unsigned i = 0; i < MAX_BIND_POINTS; i++) { if (cmd_buffer->descriptors[i].push_set.layout) vk_descriptor_set_layout_unref(&cmd_buffer->device->vk, &cmd_buffer->descriptors[i].push_set.layout->vk); vk_free(&cmd_buffer->device->vk.alloc, cmd_buffer->descriptors[i].push_set.mapped_ptr); } ralloc_free(cmd_buffer->patchpoints_ctx); ralloc_free(cmd_buffer->pre_chain.patchpoints_ctx); util_dynarray_fini(&cmd_buffer->fdm_bin_patchpoints); util_dynarray_fini(&cmd_buffer->pre_chain.fdm_bin_patchpoints); util_dynarray_fini(&cmd_buffer->vis_stream_patchpoints); util_dynarray_foreach (&cmd_buffer->vis_stream_bos, struct tu_bo *, bo) { tu_bo_finish(cmd_buffer->device, *bo); } mtx_lock(&cmd_buffer->device->vis_stream_suballocator_mtx); util_dynarray_foreach (&cmd_buffer->vis_stream_cs_bos, struct tu_vis_stream_patchpoint_cs, bo) { tu_suballoc_bo_free(&cmd_buffer->device->vis_stream_suballocator, &bo->cs_bo); tu_suballoc_bo_free(&cmd_buffer->device->vis_stream_suballocator, &bo->fence_bo); } mtx_unlock(&cmd_buffer->device->vis_stream_suballocator_mtx); util_dynarray_fini(&cmd_buffer->vis_stream_bos); util_dynarray_fini(&cmd_buffer->vis_stream_cs_bos); vk_command_buffer_finish(&cmd_buffer->vk); vk_free2(&cmd_buffer->device->vk.alloc, &cmd_buffer->vk.pool->alloc, cmd_buffer); } static void tu_reset_cmd_buffer(struct vk_command_buffer *vk_cmd_buffer, UNUSED VkCommandBufferResetFlags flags) { struct tu_cmd_buffer *cmd_buffer = container_of(vk_cmd_buffer, struct tu_cmd_buffer, vk); VkResult status_check_result = VK_SUCCESS; if (TU_DEBUG_START(CHECK_CMD_BUFFER_STATUS)) status_check_result = tu_cmd_buffer_status_check_idle(cmd_buffer); vk_command_buffer_reset(&cmd_buffer->vk); if (TU_DEBUG_START(CHECK_CMD_BUFFER_STATUS) && status_check_result != VK_SUCCESS) { cmd_buffer->vk.record_result = status_check_result; } tu_cs_reset(&cmd_buffer->cs); tu_cs_reset(&cmd_buffer->draw_cs); tu_cs_reset(&cmd_buffer->tile_store_cs); tu_cs_reset(&cmd_buffer->draw_epilogue_cs); tu_cs_reset(&cmd_buffer->sub_cs); tu_cs_reset(&cmd_buffer->pre_chain.draw_cs); tu_cs_reset(&cmd_buffer->pre_chain.draw_epilogue_cs); tu_autotune_free_results(cmd_buffer->device, &cmd_buffer->renderpass_autotune_results); for (unsigned i = 0; i < MAX_BIND_POINTS; i++) { memset(&cmd_buffer->descriptors[i].sets, 0, sizeof(cmd_buffer->descriptors[i].sets)); if (cmd_buffer->descriptors[i].push_set.layout) { vk_descriptor_set_layout_unref(&cmd_buffer->device->vk, &cmd_buffer->descriptors[i].push_set.layout->vk); } vk_free(&cmd_buffer->device->vk.alloc, cmd_buffer->descriptors[i].push_set.mapped_ptr); memset(&cmd_buffer->descriptors[i].push_set, 0, sizeof(cmd_buffer->descriptors[i].push_set)); cmd_buffer->descriptors[i].push_set.base.type = VK_OBJECT_TYPE_DESCRIPTOR_SET; cmd_buffer->descriptors[i].max_sets_bound = 0; cmd_buffer->descriptors[i].max_dynamic_offset_size = 0; } util_dynarray_foreach (&cmd_buffer->msrtss_color_temporaries, struct tu_device_memory *, mem) { tu_destroy_memory(cmd_buffer->device, *mem); } util_dynarray_clear(&cmd_buffer->msrtss_color_temporaries); util_dynarray_foreach (&cmd_buffer->msrtss_depth_temporaries, struct tu_device_memory *, mem) { tu_destroy_memory(cmd_buffer->device, *mem); } util_dynarray_clear(&cmd_buffer->msrtss_depth_temporaries); u_trace_fini(&cmd_buffer->trace); u_trace_init(&cmd_buffer->trace, &cmd_buffer->device->trace_context); u_trace_fini(&cmd_buffer->rp_trace); u_trace_init(&cmd_buffer->rp_trace, &cmd_buffer->device->trace_context); cmd_buffer->trace_renderpass_start = u_trace_begin_iterator(&cmd_buffer->rp_trace); cmd_buffer->state.max_vbs_bound = 0; cmd_buffer->vsc_initialized = false; cmd_buffer->prev_fsr_is_null = false; ralloc_free(cmd_buffer->patchpoints_ctx); ralloc_free(cmd_buffer->pre_chain.patchpoints_ctx); cmd_buffer->patchpoints_ctx = NULL; cmd_buffer->pre_chain.patchpoints_ctx = NULL; util_dynarray_clear(&cmd_buffer->fdm_bin_patchpoints); util_dynarray_clear(&cmd_buffer->pre_chain.fdm_bin_patchpoints); util_dynarray_clear(&cmd_buffer->vis_stream_patchpoints); util_dynarray_foreach (&cmd_buffer->vis_stream_bos, struct tu_bo *, bo) { tu_bo_finish(cmd_buffer->device, *bo); } mtx_lock(&cmd_buffer->device->vis_stream_suballocator_mtx); util_dynarray_foreach (&cmd_buffer->vis_stream_cs_bos, struct tu_vis_stream_patchpoint_cs, bo) { tu_suballoc_bo_free(&cmd_buffer->device->vis_stream_suballocator, &bo->cs_bo); tu_suballoc_bo_free(&cmd_buffer->device->vis_stream_suballocator, &bo->fence_bo); } mtx_unlock(&cmd_buffer->device->vis_stream_suballocator_mtx); util_dynarray_clear(&cmd_buffer->vis_stream_bos); util_dynarray_clear(&cmd_buffer->vis_stream_cs_bos); } const struct vk_command_buffer_ops tu_cmd_buffer_ops = { .create = tu_create_cmd_buffer, .reset = tu_reset_cmd_buffer, .destroy = tu_cmd_buffer_destroy, }; /* Initialize the cache, assuming all necessary flushes have happened but *not* * invalidations. */ static void tu_cache_init(struct tu_cache_state *cache) { cache->flush_bits = 0; cache->pending_flush_bits = TU_CMD_FLAG_ALL_INVALIDATE; } /* Unlike the public entrypoint, this doesn't handle cache tracking, and * tracking the CCU state. It's used for the driver to insert its own command * buffer in the middle of a submit. */ VkResult tu_cmd_buffer_begin(struct tu_cmd_buffer *cmd_buffer, const VkCommandBufferBeginInfo *pBeginInfo) { vk_command_buffer_begin(&cmd_buffer->vk, pBeginInfo); memset(&cmd_buffer->state, 0, sizeof(cmd_buffer->state)); vk_dynamic_graphics_state_init(&cmd_buffer->vk.dynamic_graphics_state); cmd_buffer->vk.dynamic_graphics_state.vi = &cmd_buffer->state.vi; cmd_buffer->vk.dynamic_graphics_state.ms.sample_locations = &cmd_buffer->state.sl; cmd_buffer->state.index_size = 0xff; /* dirty restart index */ cmd_buffer->state.gmem_layout = TU_GMEM_LAYOUT_COUNT; /* dirty value */ tu_cache_init(&cmd_buffer->state.cache); tu_cache_init(&cmd_buffer->state.renderpass_cache); cmd_buffer->usage_flags = pBeginInfo->flags; tu_cs_begin(&cmd_buffer->cs); tu_cs_begin(&cmd_buffer->draw_cs); tu_cs_begin(&cmd_buffer->draw_epilogue_cs); if (cmd_buffer->vk.level == VK_COMMAND_BUFFER_LEVEL_PRIMARY) { if (u_trace_enabled(&cmd_buffer->device->trace_context)) { trace_start_cmd_buffer(&cmd_buffer->trace, &cmd_buffer->cs, cmd_buffer, tu_env_debug_as_string(), ir3_shader_debug_as_string()); } } tu_cmd_buffer_status_gpu_write(cmd_buffer, TU_CMD_BUFFER_STATUS_ACTIVE); return VK_SUCCESS; } VKAPI_ATTR VkResult VKAPI_CALL tu_BeginCommandBuffer(VkCommandBuffer commandBuffer, const VkCommandBufferBeginInfo *pBeginInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd_buffer, commandBuffer); VkResult result = tu_cmd_buffer_begin(cmd_buffer, pBeginInfo); if (result != VK_SUCCESS) return result; /* setup initial configuration into command buffer */ if (cmd_buffer->vk.level == VK_COMMAND_BUFFER_LEVEL_PRIMARY) { switch (cmd_buffer->queue_family_index) { case TU_QUEUE_GENERAL: TU_CALLX(cmd_buffer->device, tu6_init_hw)(cmd_buffer, &cmd_buffer->cs); break; default: break; } } else if (cmd_buffer->vk.level == VK_COMMAND_BUFFER_LEVEL_SECONDARY) { const bool pass_continue = pBeginInfo->flags & VK_COMMAND_BUFFER_USAGE_RENDER_PASS_CONTINUE_BIT; if (u_trace_enabled(&cmd_buffer->device->trace_context)) { trace_start_secondary_cmd_buffer( pass_continue ? &cmd_buffer->rp_trace : &cmd_buffer->trace, pass_continue ? &cmd_buffer->draw_cs : &cmd_buffer->cs, cmd_buffer); } assert(pBeginInfo->pInheritanceInfo); cmd_buffer->inherited_pipeline_statistics = pBeginInfo->pInheritanceInfo->pipelineStatistics; cmd_buffer->state.occlusion_query_may_be_running = pBeginInfo->pInheritanceInfo->occlusionQueryEnable; vk_foreach_struct_const(ext, pBeginInfo->pInheritanceInfo) { switch (ext->sType) { case VK_STRUCTURE_TYPE_COMMAND_BUFFER_INHERITANCE_CONDITIONAL_RENDERING_INFO_EXT: { const VkCommandBufferInheritanceConditionalRenderingInfoEXT *cond_rend = (VkCommandBufferInheritanceConditionalRenderingInfoEXT *) ext; cmd_buffer->state.predication_active = cond_rend->conditionalRenderingEnable; break; } default: break; } } if (pass_continue) { const VkCommandBufferInheritanceRenderingInfo *rendering_info = vk_find_struct_const(pBeginInfo->pInheritanceInfo->pNext, COMMAND_BUFFER_INHERITANCE_RENDERING_INFO); if (TU_DEBUG(DYNAMIC)) { rendering_info = vk_get_command_buffer_inheritance_rendering_info(cmd_buffer->vk.level, pBeginInfo); } if (rendering_info) { tu_setup_dynamic_inheritance(cmd_buffer, rendering_info); cmd_buffer->state.pass = &cmd_buffer->dynamic_pass; cmd_buffer->state.subpass = &cmd_buffer->dynamic_subpass; const VkRenderingAttachmentLocationInfoKHR *location_info = vk_find_struct_const(pBeginInfo->pInheritanceInfo->pNext, RENDERING_ATTACHMENT_LOCATION_INFO_KHR); if (location_info) { vk_common_CmdSetRenderingAttachmentLocationsKHR(commandBuffer, location_info); } /* Unfortunately with dynamic renderpasses we get no indication * whether FDM is used in secondaries, so we have to assume it * always might be enabled. */ cmd_buffer->state.fdm_enabled = cmd_buffer->device->vk.enabled_features.fragmentDensityMap || TU_DEBUG(FDM); } else { cmd_buffer->state.pass = tu_render_pass_from_handle(pBeginInfo->pInheritanceInfo->renderPass); cmd_buffer->state.subpass = &cmd_buffer->state.pass->subpasses[pBeginInfo->pInheritanceInfo->subpass]; cmd_buffer->state.fdm_enabled = cmd_buffer->state.pass->has_fdm; } tu_fill_render_pass_state(&cmd_buffer->state.vk_rp, cmd_buffer->state.pass, cmd_buffer->state.subpass); vk_cmd_set_cb_attachment_count(&cmd_buffer->vk, cmd_buffer->state.subpass->color_count); cmd_buffer->state.dirty |= TU_CMD_DIRTY_SUBPASS; cmd_buffer->patchpoints_ctx = ralloc_context(NULL); /* We can't set the gmem layout here, because the state.pass only has * to be compatible (same formats/sample counts) with the primary's * renderpass, rather than exactly equal. */ tu_lrz_begin_secondary_cmdbuf(cmd_buffer); } else { /* When executing in the middle of another command buffer, the CCU * state is unknown. */ cmd_buffer->state.ccu_state = TU_CMD_CCU_UNKNOWN; } } return VK_SUCCESS; } static struct tu_cs tu_cmd_dynamic_state(struct tu_cmd_buffer *cmd, uint32_t id, uint32_t size) { struct tu_cs cs; assert(id < ARRAY_SIZE(cmd->state.dynamic_state)); cmd->state.dynamic_state[id] = tu_cs_draw_state(&cmd->sub_cs, &cs, size); /* note: this also avoids emitting draw states before renderpass clears, * which may use the 3D clear path (for MSAA cases) */ if (cmd->state.dirty & TU_CMD_DIRTY_DRAW_STATE) return cs; tu_cs_emit_pkt7(&cmd->draw_cs, CP_SET_DRAW_STATE, 3); tu_cs_emit_draw_state(&cmd->draw_cs, TU_DRAW_STATE_DYNAMIC + id, cmd->state.dynamic_state[id]); return cs; } static void tu_cmd_end_dynamic_state(struct tu_cmd_buffer *cmd, struct tu_cs *cs, uint32_t id) { assert(id < ARRAY_SIZE(cmd->state.dynamic_state)); cmd->state.dynamic_state[id] = tu_cs_end_draw_state(&cmd->sub_cs, cs); /* note: this also avoids emitting draw states before renderpass clears, * which may use the 3D clear path (for MSAA cases) */ if (cmd->state.dirty & TU_CMD_DIRTY_DRAW_STATE) return; tu_cs_emit_pkt7(&cmd->draw_cs, CP_SET_DRAW_STATE, 3); tu_cs_emit_draw_state(&cmd->draw_cs, TU_DRAW_STATE_DYNAMIC + id, cmd->state.dynamic_state[id]); } VKAPI_ATTR void VKAPI_CALL tu_CmdBindVertexBuffers2(VkCommandBuffer commandBuffer, uint32_t firstBinding, uint32_t bindingCount, const VkBuffer *pBuffers, const VkDeviceSize *pOffsets, const VkDeviceSize *pSizes, const VkDeviceSize *pStrides) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); struct tu_cs cs; cmd->state.max_vbs_bound = MAX2( cmd->state.max_vbs_bound, firstBinding + bindingCount); if (pStrides) { vk_cmd_set_vertex_binding_strides(&cmd->vk, firstBinding, bindingCount, pStrides); } cmd->state.vertex_buffers.iova = tu_cs_draw_state(&cmd->sub_cs, &cs, 4 * cmd->state.max_vbs_bound).iova; for (uint32_t i = 0; i < bindingCount; i++) { if (pBuffers[i] == VK_NULL_HANDLE) { cmd->state.vb[firstBinding + i].base = 0; cmd->state.vb[firstBinding + i].size = 0; } else { struct tu_buffer *buf = tu_buffer_from_handle(pBuffers[i]); cmd->state.vb[firstBinding + i].base = vk_buffer_address(&buf->vk, pOffsets[i]); cmd->state.vb[firstBinding + i].size = vk_buffer_range(&buf->vk, pOffsets[i], pSizes ? pSizes[i] : VK_WHOLE_SIZE); } } for (uint32_t i = 0; i < cmd->state.max_vbs_bound; i++) { tu_cs_emit_regs(&cs, A6XX_VFD_VERTEX_BUFFER_BASE(i, .qword = cmd->state.vb[i].base), A6XX_VFD_VERTEX_BUFFER_SIZE(i, cmd->state.vb[i].size)); } cmd->state.dirty |= TU_CMD_DIRTY_VERTEX_BUFFERS; } VKAPI_ATTR void VKAPI_CALL tu_CmdBindIndexBuffer2KHR(VkCommandBuffer commandBuffer, VkBuffer buffer, VkDeviceSize offset, VkDeviceSize size, VkIndexType indexType) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); VK_FROM_HANDLE(tu_buffer, buf, buffer); size = buf ? vk_buffer_range(&buf->vk, offset, size) : 0; uint32_t index_size, index_shift; uint32_t restart_index = vk_index_to_restart(indexType); switch (indexType) { case VK_INDEX_TYPE_UINT16: index_size = INDEX4_SIZE_16_BIT; index_shift = 1; break; case VK_INDEX_TYPE_UINT32: index_size = INDEX4_SIZE_32_BIT; index_shift = 2; break; case VK_INDEX_TYPE_UINT8_KHR: index_size = INDEX4_SIZE_8_BIT; index_shift = 0; break; default: UNREACHABLE("invalid VkIndexType"); } if (buf) { /* initialize/update the restart index */ if (cmd->state.index_size != index_size) tu_cs_emit_regs(&cmd->draw_cs, A6XX_PC_RESTART_INDEX(restart_index)); cmd->state.index_va = vk_buffer_address(&buf->vk, offset); cmd->state.max_index_count = size >> index_shift; cmd->state.index_size = index_size; } else { cmd->state.index_va = 0; cmd->state.max_index_count = 0; cmd->state.index_size = 0; } } template static void tu6_emit_descriptor_sets(struct tu_cmd_buffer *cmd, VkPipelineBindPoint bind_point) { struct tu_descriptor_state *descriptors_state = tu_get_descriptors_state(cmd, bind_point); uint32_t sp_bindless_base_reg, hlsq_bindless_base_reg; struct tu_cs *cs, state_cs; if (bind_point == VK_PIPELINE_BIND_POINT_GRAPHICS) { sp_bindless_base_reg = __SP_GFX_BINDLESS_BASE_DESCRIPTOR(0, {}).reg; hlsq_bindless_base_reg = REG_A6XX_HLSQ_BINDLESS_BASE(0); if (CHIP == A6XX) { cmd->state.desc_sets = tu_cs_draw_state(&cmd->sub_cs, &state_cs, 4 + 4 * descriptors_state->max_sets_bound + (descriptors_state->max_dynamic_offset_size ? 6 : 0)); } else { cmd->state.desc_sets = tu_cs_draw_state(&cmd->sub_cs, &state_cs, 3 + 2 * descriptors_state->max_sets_bound + (descriptors_state->max_dynamic_offset_size ? 3 : 0)); } cs = &state_cs; } else { assert(bind_point == VK_PIPELINE_BIND_POINT_COMPUTE); sp_bindless_base_reg = __SP_CS_BINDLESS_BASE_DESCRIPTOR(0, {}).reg; hlsq_bindless_base_reg = REG_A6XX_HLSQ_CS_BINDLESS_BASE(0); cs = &cmd->cs; } tu_cs_emit_pkt4(cs, sp_bindless_base_reg, 2 * descriptors_state->max_sets_bound); tu_cs_emit_array(cs, (const uint32_t*)descriptors_state->set_iova, 2 * descriptors_state->max_sets_bound); if (CHIP == A6XX) { tu_cs_emit_pkt4(cs, hlsq_bindless_base_reg, 2 * descriptors_state->max_sets_bound); tu_cs_emit_array(cs, (const uint32_t*)descriptors_state->set_iova, 2 * descriptors_state->max_sets_bound); } /* Dynamic descriptors get the reserved descriptor set. */ if (descriptors_state->max_dynamic_offset_size) { int reserved_set_idx = cmd->device->physical_device->reserved_set_idx; assert(reserved_set_idx >= 0); /* reserved set must be bound */ tu_cs_emit_pkt4(cs, sp_bindless_base_reg + reserved_set_idx * 2, 2); tu_cs_emit_qw(cs, descriptors_state->set_iova[reserved_set_idx]); if (CHIP == A6XX) { tu_cs_emit_pkt4(cs, hlsq_bindless_base_reg + reserved_set_idx * 2, 2); tu_cs_emit_qw(cs, descriptors_state->set_iova[reserved_set_idx]); } } tu_cs_emit_regs(cs, SP_UPDATE_CNTL(CHIP, .cs_bindless = bind_point == VK_PIPELINE_BIND_POINT_COMPUTE ? CHIP == A6XX ? 0x1f : 0xff : 0, .gfx_bindless = bind_point == VK_PIPELINE_BIND_POINT_GRAPHICS ? CHIP == A6XX ? 0x1f : 0xff : 0, )); if (bind_point == VK_PIPELINE_BIND_POINT_GRAPHICS) { assert(cs->cur == cs->end); /* validate draw state size */ /* note: this also avoids emitting draw states before renderpass clears, * which may use the 3D clear path (for MSAA cases) */ if (!(cmd->state.dirty & TU_CMD_DIRTY_DRAW_STATE)) { tu_cs_emit_pkt7(&cmd->draw_cs, CP_SET_DRAW_STATE, 3); tu_cs_emit_draw_state(&cmd->draw_cs, TU_DRAW_STATE_DESC_SETS, cmd->state.desc_sets); } } } /* We lazily emit the draw state for desciptor sets at draw time, so that we can * batch together multiple tu_CmdBindDescriptorSets() calls. ANGLE and zink * will often emit multiple bind calls in a draw. */ static void tu_dirty_desc_sets(struct tu_cmd_buffer *cmd, VkPipelineBindPoint pipelineBindPoint) { if (pipelineBindPoint == VK_PIPELINE_BIND_POINT_COMPUTE) { cmd->state.dirty |= TU_CMD_DIRTY_COMPUTE_DESC_SETS; } else { assert(pipelineBindPoint == VK_PIPELINE_BIND_POINT_GRAPHICS); cmd->state.dirty |= TU_CMD_DIRTY_DESC_SETS; } } static void tu_bind_descriptor_sets(struct tu_cmd_buffer *cmd, const VkBindDescriptorSetsInfoKHR *info, VkPipelineBindPoint bind_point) { VK_FROM_HANDLE(tu_pipeline_layout, layout, info->layout); unsigned dyn_idx = 0; struct tu_descriptor_state *descriptors_state = tu_get_descriptors_state(cmd, bind_point); descriptors_state->max_sets_bound = MAX2(descriptors_state->max_sets_bound, info->firstSet + info->descriptorSetCount); unsigned dynamic_offset_offset = 0; for (unsigned i = 0; i < info->firstSet; i++) { if (layout->set[i].layout) dynamic_offset_offset += layout->set[i].layout->dynamic_offset_size; } for (unsigned i = 0; i < info->descriptorSetCount; ++i) { unsigned idx = i + info->firstSet; VK_FROM_HANDLE(tu_descriptor_set, set, info->pDescriptorSets[i]); descriptors_state->sets[idx] = set; descriptors_state->set_iova[idx] = set ? (set->va | BINDLESS_DESCRIPTOR_64B) : 0; if (!set) continue; if (set->layout->has_inline_uniforms) cmd->state.dirty |= TU_CMD_DIRTY_SHADER_CONSTS; if (!set->layout->dynamic_offset_size) continue; uint32_t *src = set->dynamic_descriptors; uint32_t *dst = descriptors_state->dynamic_descriptors + dynamic_offset_offset / 4; for (unsigned j = 0; j < set->layout->binding_count; j++) { struct tu_descriptor_set_binding_layout *binding = &set->layout->binding[j]; if (vk_descriptor_type_is_dynamic(binding->type)) { for (unsigned k = 0; k < binding->array_size; k++, dyn_idx++) { assert(dyn_idx < info->dynamicOffsetCount); uint32_t offset = info->pDynamicOffsets[dyn_idx]; memcpy(dst, src, binding->size); if (binding->type == VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER_DYNAMIC) { /* Note: we can assume here that the addition won't roll * over and change the SIZE field. */ uint64_t va = src[0] | ((uint64_t)src[1] << 32); va += offset; dst[0] = va; dst[1] = va >> 32; } else { uint32_t *dst_desc = dst; for (unsigned i = 0; i < binding->size / (4 * A6XX_TEX_CONST_DWORDS); i++, dst_desc += A6XX_TEX_CONST_DWORDS) { /* Note: A6XX_TEX_CONST_5_DEPTH is always 0 */ uint64_t va = dst_desc[4] | ((uint64_t)dst_desc[5] << 32); uint32_t desc_offset = pkt_field_get( A6XX_TEX_CONST_2_STARTOFFSETTEXELS, dst_desc[2]); /* Use descriptor's format to determine the shift amount * that's to be used on the offset value. */ uint32_t format = pkt_field_get(A6XX_TEX_CONST_0_FMT, dst_desc[0]); unsigned offset_shift; switch (format) { case FMT6_16_UINT: offset_shift = 1; break; case FMT6_32_UINT: offset_shift = 2; break; case FMT6_8_UINT: default: offset_shift = 0; break; } va += desc_offset << offset_shift; va += offset; unsigned new_offset = (va & 0x3f) >> offset_shift; va &= ~0x3full; dst_desc[4] = va; dst_desc[5] = va >> 32; dst_desc[2] = pkt_field_set(A6XX_TEX_CONST_2_STARTOFFSETTEXELS, dst_desc[2], new_offset); } } dst += binding->size / 4; src += binding->size / 4; } } } if (layout->set[idx].layout) dynamic_offset_offset += layout->set[idx].layout->dynamic_offset_size; } assert(dyn_idx == info->dynamicOffsetCount); if (dynamic_offset_offset) { descriptors_state->max_dynamic_offset_size = MAX2(descriptors_state->max_dynamic_offset_size, dynamic_offset_offset); /* allocate and fill out dynamic descriptor set */ struct tu_cs_memory dynamic_desc_set; int reserved_set_idx = cmd->device->physical_device->reserved_set_idx; VkResult result = tu_cs_alloc(&cmd->sub_cs, descriptors_state->max_dynamic_offset_size / (4 * A6XX_TEX_CONST_DWORDS), A6XX_TEX_CONST_DWORDS, &dynamic_desc_set); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } memcpy(dynamic_desc_set.map, descriptors_state->dynamic_descriptors, descriptors_state->max_dynamic_offset_size); assert(reserved_set_idx >= 0); /* reserved set must be bound */ descriptors_state->set_iova[reserved_set_idx] = dynamic_desc_set.iova | BINDLESS_DESCRIPTOR_64B; } tu_dirty_desc_sets(cmd, bind_point); } VKAPI_ATTR void VKAPI_CALL tu_CmdBindDescriptorSets2KHR( VkCommandBuffer commandBuffer, const VkBindDescriptorSetsInfoKHR *pBindDescriptorSetsInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); if (pBindDescriptorSetsInfo->stageFlags & VK_SHADER_STAGE_COMPUTE_BIT) { tu_bind_descriptor_sets(cmd, pBindDescriptorSetsInfo, VK_PIPELINE_BIND_POINT_COMPUTE); } if (pBindDescriptorSetsInfo->stageFlags & VK_SHADER_STAGE_ALL_GRAPHICS) { tu_bind_descriptor_sets(cmd, pBindDescriptorSetsInfo, VK_PIPELINE_BIND_POINT_GRAPHICS); } } VKAPI_ATTR void VKAPI_CALL tu_CmdBindDescriptorBuffersEXT( VkCommandBuffer commandBuffer, uint32_t bufferCount, const VkDescriptorBufferBindingInfoEXT *pBindingInfos) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); for (unsigned i = 0; i < bufferCount; i++) cmd->state.descriptor_buffer_iova[i] = pBindingInfos[i].address; } static void tu_set_descriptor_buffer_offsets( struct tu_cmd_buffer *cmd, const VkSetDescriptorBufferOffsetsInfoEXT *info, VkPipelineBindPoint bind_point) { VK_FROM_HANDLE(tu_pipeline_layout, layout, info->layout); struct tu_descriptor_state *descriptors_state = tu_get_descriptors_state(cmd, bind_point); descriptors_state->max_sets_bound = MAX2(descriptors_state->max_sets_bound, info->firstSet + info->setCount); for (unsigned i = 0; i < info->setCount; ++i) { unsigned idx = i + info->firstSet; struct tu_descriptor_set_layout *set_layout = layout->set[idx].layout; descriptors_state->set_iova[idx] = (cmd->state.descriptor_buffer_iova[info->pBufferIndices[i]] + info->pOffsets[i]) | BINDLESS_DESCRIPTOR_64B; if (set_layout->has_inline_uniforms) cmd->state.dirty |= TU_CMD_DIRTY_SHADER_CONSTS; } tu_dirty_desc_sets(cmd, bind_point); } VKAPI_ATTR void VKAPI_CALL tu_CmdSetDescriptorBufferOffsets2EXT( VkCommandBuffer commandBuffer, const VkSetDescriptorBufferOffsetsInfoEXT *pSetDescriptorBufferOffsetsInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); if (pSetDescriptorBufferOffsetsInfo->stageFlags & VK_SHADER_STAGE_COMPUTE_BIT) { tu_set_descriptor_buffer_offsets(cmd, pSetDescriptorBufferOffsetsInfo, VK_PIPELINE_BIND_POINT_COMPUTE); } if (pSetDescriptorBufferOffsetsInfo->stageFlags & VK_SHADER_STAGE_ALL_GRAPHICS) { tu_set_descriptor_buffer_offsets(cmd, pSetDescriptorBufferOffsetsInfo, VK_PIPELINE_BIND_POINT_GRAPHICS); } } static void tu_bind_descriptor_buffer_embedded_samplers( struct tu_cmd_buffer *cmd, const VkBindDescriptorBufferEmbeddedSamplersInfoEXT *info, VkPipelineBindPoint bind_point) { VK_FROM_HANDLE(tu_pipeline_layout, layout, info->layout); struct tu_descriptor_set_layout *set_layout = layout->set[info->set].layout; struct tu_descriptor_state *descriptors_state = tu_get_descriptors_state(cmd, bind_point); descriptors_state->max_sets_bound = MAX2(descriptors_state->max_sets_bound, info->set + 1); descriptors_state->set_iova[info->set] = set_layout->embedded_samplers->iova | BINDLESS_DESCRIPTOR_64B; tu_dirty_desc_sets(cmd, bind_point); } VKAPI_ATTR void VKAPI_CALL tu_CmdBindDescriptorBufferEmbeddedSamplers2EXT( VkCommandBuffer commandBuffer, const VkBindDescriptorBufferEmbeddedSamplersInfoEXT *pBindDescriptorBufferEmbeddedSamplersInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); if (pBindDescriptorBufferEmbeddedSamplersInfo->stageFlags & VK_SHADER_STAGE_COMPUTE_BIT) { tu_bind_descriptor_buffer_embedded_samplers( cmd, pBindDescriptorBufferEmbeddedSamplersInfo, VK_PIPELINE_BIND_POINT_COMPUTE); } if (pBindDescriptorBufferEmbeddedSamplersInfo->stageFlags & VK_SHADER_STAGE_ALL_GRAPHICS) { tu_bind_descriptor_buffer_embedded_samplers( cmd, pBindDescriptorBufferEmbeddedSamplersInfo, VK_PIPELINE_BIND_POINT_GRAPHICS); } } static VkResult tu_push_descriptor_set_update_layout(struct tu_device *device, struct tu_descriptor_set *set, struct tu_descriptor_set_layout *layout) { if (set->layout == layout) return VK_SUCCESS; if (set->layout) vk_descriptor_set_layout_unref(&device->vk, &set->layout->vk); vk_descriptor_set_layout_ref(&layout->vk); set->layout = layout; if (set->host_size < layout->size) { void *new_buf = vk_realloc(&device->vk.alloc, set->mapped_ptr, layout->size, 8, VK_SYSTEM_ALLOCATION_SCOPE_OBJECT); if (!new_buf) return VK_ERROR_OUT_OF_HOST_MEMORY; set->mapped_ptr = (uint32_t *) new_buf; set->host_size = layout->size; } return VK_SUCCESS; } static void tu_push_descriptor_set(struct tu_cmd_buffer *cmd, const VkPushDescriptorSetInfoKHR *info, VkPipelineBindPoint bind_point) { VK_FROM_HANDLE(tu_pipeline_layout, pipe_layout, info->layout); struct tu_descriptor_set_layout *layout = pipe_layout->set[info->set].layout; struct tu_descriptor_set *set = &tu_get_descriptors_state(cmd, bind_point)->push_set; struct tu_cs_memory set_mem; VkResult result = tu_cs_alloc(&cmd->sub_cs, DIV_ROUND_UP(layout->size, A6XX_TEX_CONST_DWORDS * 4), A6XX_TEX_CONST_DWORDS, &set_mem); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } result = tu_push_descriptor_set_update_layout(cmd->device, set, layout); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } tu_update_descriptor_sets(cmd->device, tu_descriptor_set_to_handle(set), info->descriptorWriteCount, info->pDescriptorWrites, 0, NULL); memcpy(set_mem.map, set->mapped_ptr, layout->size); set->va = set_mem.iova; const VkDescriptorSet desc_set[] = { tu_descriptor_set_to_handle(set) }; vk_common_CmdBindDescriptorSets(tu_cmd_buffer_to_handle(cmd), bind_point, info->layout, info->set, 1, desc_set, 0, NULL); } VKAPI_ATTR void VKAPI_CALL tu_CmdPushDescriptorSet2KHR( VkCommandBuffer commandBuffer, const VkPushDescriptorSetInfoKHR *pPushDescriptorSetInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); if (pPushDescriptorSetInfo->stageFlags & VK_SHADER_STAGE_COMPUTE_BIT) { tu_push_descriptor_set(cmd, pPushDescriptorSetInfo, VK_PIPELINE_BIND_POINT_COMPUTE); } if (pPushDescriptorSetInfo->stageFlags & VK_SHADER_STAGE_ALL_GRAPHICS) { tu_push_descriptor_set(cmd, pPushDescriptorSetInfo, VK_PIPELINE_BIND_POINT_GRAPHICS); } } VKAPI_ATTR void VKAPI_CALL tu_CmdPushDescriptorSetWithTemplate2KHR( VkCommandBuffer commandBuffer, const VkPushDescriptorSetWithTemplateInfoKHR *pPushDescriptorSetWithTemplateInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); VK_FROM_HANDLE(tu_pipeline_layout, pipe_layout, pPushDescriptorSetWithTemplateInfo->layout); VK_FROM_HANDLE( tu_descriptor_update_template, templ, pPushDescriptorSetWithTemplateInfo->descriptorUpdateTemplate); struct tu_descriptor_set_layout *layout = pipe_layout->set[pPushDescriptorSetWithTemplateInfo->set].layout; struct tu_descriptor_set *set = &tu_get_descriptors_state(cmd, templ->bind_point)->push_set; struct tu_cs_memory set_mem; VkResult result = tu_cs_alloc(&cmd->sub_cs, DIV_ROUND_UP(layout->size, A6XX_TEX_CONST_DWORDS * 4), A6XX_TEX_CONST_DWORDS, &set_mem); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } result = tu_push_descriptor_set_update_layout(cmd->device, set, layout); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } tu_update_descriptor_set_with_template( cmd->device, set, pPushDescriptorSetWithTemplateInfo->descriptorUpdateTemplate, pPushDescriptorSetWithTemplateInfo->pData); memcpy(set_mem.map, set->mapped_ptr, layout->size); set->va = set_mem.iova; const VkDescriptorSet desc_set[] = { tu_descriptor_set_to_handle(set) }; vk_common_CmdBindDescriptorSets( tu_cmd_buffer_to_handle(cmd), templ->bind_point, pPushDescriptorSetWithTemplateInfo->layout, pPushDescriptorSetWithTemplateInfo->set, 1, desc_set, 0, NULL); } VKAPI_ATTR void VKAPI_CALL tu_CmdBindTransformFeedbackBuffersEXT(VkCommandBuffer commandBuffer, uint32_t firstBinding, uint32_t bindingCount, const VkBuffer *pBuffers, const VkDeviceSize *pOffsets, const VkDeviceSize *pSizes) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); struct tu_cs *cs = &cmd->draw_cs; /* using COND_REG_EXEC for xfb commands matches the blob behavior * presumably there isn't any benefit using a draw state when the * condition is (SYSMEM | BINNING) */ tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(RENDER_MODE) | CP_COND_REG_EXEC_0_SYSMEM | CP_COND_REG_EXEC_0_BINNING); for (uint32_t i = 0; i < bindingCount; i++) { VK_FROM_HANDLE(tu_buffer, buf, pBuffers[i]); uint64_t iova = vk_buffer_address(&buf->vk, pOffsets[i]); uint32_t size = vk_buffer_range(&buf->vk, pOffsets[i], pSizes ? pSizes[i] : VK_WHOLE_SIZE); uint32_t idx = i + firstBinding; /* BUFFER_BASE is 32-byte aligned, add remaining offset to BUFFER_OFFSET */ uint32_t offset = iova & 0x1f; iova &= ~(uint64_t) 0x1f; tu_cs_emit_pkt4(cs, REG_A6XX_VPC_SO_BUFFER_BASE(idx), 3); tu_cs_emit_qw(cs, iova); tu_cs_emit(cs, size + offset); cmd->state.streamout_offset[idx] = offset; } tu_cond_exec_end(cs); } VKAPI_ATTR void VKAPI_CALL tu_CmdBeginTransformFeedbackEXT(VkCommandBuffer commandBuffer, uint32_t firstCounterBuffer, uint32_t counterBufferCount, const VkBuffer *pCounterBuffers, const VkDeviceSize *pCounterBufferOffsets) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); struct tu_cs *cs = &cmd->draw_cs; tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(RENDER_MODE) | CP_COND_REG_EXEC_0_SYSMEM | CP_COND_REG_EXEC_0_BINNING); tu_cs_emit_regs(cs, A6XX_VPC_SO_OVERRIDE(false)); /* TODO: only update offset for active buffers */ for (uint32_t i = 0; i < IR3_MAX_SO_BUFFERS; i++) tu_cs_emit_regs(cs, A6XX_VPC_SO_BUFFER_OFFSET(i, cmd->state.streamout_offset[i])); for (uint32_t i = 0; i < (pCounterBuffers ? counterBufferCount : 0); i++) { uint32_t idx = firstCounterBuffer + i; uint32_t offset = cmd->state.streamout_offset[idx]; uint64_t counter_buffer_offset = pCounterBufferOffsets ? pCounterBufferOffsets[i] : 0u; if (!pCounterBuffers[i]) continue; VK_FROM_HANDLE(tu_buffer, buf, pCounterBuffers[i]); tu_cs_emit_pkt7(cs, CP_MEM_TO_REG, 3); tu_cs_emit(cs, CP_MEM_TO_REG_0_REG(REG_A6XX_VPC_SO_BUFFER_OFFSET(idx)) | CP_MEM_TO_REG_0_UNK31 | CP_MEM_TO_REG_0_CNT(1)); tu_cs_emit_qw(cs, vk_buffer_address(&buf->vk, counter_buffer_offset)); if (offset) { tu_cs_emit_pkt7(cs, CP_REG_RMW, 3); tu_cs_emit(cs, CP_REG_RMW_0_DST_REG(REG_A6XX_VPC_SO_BUFFER_OFFSET(idx)) | CP_REG_RMW_0_SRC1_ADD); tu_cs_emit(cs, 0xffffffff); tu_cs_emit(cs, offset); } } tu_cond_exec_end(cs); } template VKAPI_ATTR void VKAPI_CALL tu_CmdEndTransformFeedbackEXT(VkCommandBuffer commandBuffer, uint32_t firstCounterBuffer, uint32_t counterBufferCount, const VkBuffer *pCounterBuffers, const VkDeviceSize *pCounterBufferOffsets) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); struct tu_cs *cs = &cmd->draw_cs; tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(RENDER_MODE) | CP_COND_REG_EXEC_0_SYSMEM | CP_COND_REG_EXEC_0_BINNING); tu_cs_emit_regs(cs, A6XX_VPC_SO_OVERRIDE(true)); /* TODO: only flush buffers that need to be flushed */ for (uint32_t i = 0; i < IR3_MAX_SO_BUFFERS; i++) { /* note: FLUSH_BASE is always the same, so it could go in init_hw()? */ tu_cs_emit_pkt4(cs, REG_A6XX_VPC_SO_FLUSH_BASE(i), 2); tu_cs_emit_qw(cs, global_iova_arr(cmd, flush_base, i)); tu_emit_event_write(cmd, cs, (enum fd_gpu_event) (FD_FLUSH_SO_0 + i)); } for (uint32_t i = 0; i < (pCounterBuffers ? counterBufferCount : 0); i++) { uint32_t idx = firstCounterBuffer + i; uint32_t offset = cmd->state.streamout_offset[idx]; uint64_t counter_buffer_offset = pCounterBufferOffsets ? pCounterBufferOffsets[i] : 0u; if (!pCounterBuffers[i]) continue; VK_FROM_HANDLE(tu_buffer, buf, pCounterBuffers[i]); /* VPC_SO_FLUSH_BASE has dwords counter, but counter should be in bytes */ tu_cs_emit_pkt7(cs, CP_MEM_TO_REG, 3); tu_cs_emit(cs, CP_MEM_TO_REG_0_REG(REG_A6XX_CP_SCRATCH_REG(0)) | COND(CHIP == A6XX, CP_MEM_TO_REG_0_SHIFT_BY_2) | 0x40000 | /* ??? */ CP_MEM_TO_REG_0_UNK31 | CP_MEM_TO_REG_0_CNT(1)); tu_cs_emit_qw(cs, global_iova_arr(cmd, flush_base, idx)); if (offset) { tu_cs_emit_pkt7(cs, CP_REG_RMW, 3); tu_cs_emit(cs, CP_REG_RMW_0_DST_REG(REG_A6XX_CP_SCRATCH_REG(0)) | CP_REG_RMW_0_SRC1_ADD); tu_cs_emit(cs, 0xffffffff); tu_cs_emit(cs, -offset); } tu_cs_emit_pkt7(cs, CP_REG_TO_MEM, 3); tu_cs_emit(cs, CP_REG_TO_MEM_0_REG(REG_A6XX_CP_SCRATCH_REG(0)) | CP_REG_TO_MEM_0_CNT(1)); tu_cs_emit_qw(cs, vk_buffer_address(&buf->vk, counter_buffer_offset)); } tu_cond_exec_end(cs); cmd->state.rp.xfb_used = true; } TU_GENX(tu_CmdEndTransformFeedbackEXT); VKAPI_ATTR void VKAPI_CALL tu_CmdPushConstants2KHR(VkCommandBuffer commandBuffer, const VkPushConstantsInfoKHR *pPushConstantsInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); memcpy((char *) cmd->push_constants + pPushConstantsInfo->offset, pPushConstantsInfo->pValues, pPushConstantsInfo->size); cmd->state.dirty |= TU_CMD_DIRTY_SHADER_CONSTS; } /* Clean everything which has been made available but we haven't actually * cleaned yet. */ static void tu_clean_all_pending(struct tu_cache_state *cache) { cache->flush_bits |= cache->pending_flush_bits & TU_CMD_FLAG_ALL_CLEAN; cache->pending_flush_bits &= ~TU_CMD_FLAG_ALL_CLEAN; } template VKAPI_ATTR VkResult VKAPI_CALL tu_EndCommandBuffer(VkCommandBuffer commandBuffer) { VK_FROM_HANDLE(tu_cmd_buffer, cmd_buffer, commandBuffer); /* We currently flush CCU at the end of the command buffer, like * what the blob does. There's implicit synchronization around every * vkQueueSubmit, but the kernel only flushes the UCHE, and we don't * know yet if this command buffer will be the last in the submit so we * have to defensively flush everything else. * * TODO: We could definitely do better than this, since these flushes * aren't required by Vulkan, but we'd need kernel support to do that. * Ideally, we'd like the kernel to flush everything afterwards, so that we * wouldn't have to do any flushes here, and when submitting multiple * command buffers there wouldn't be any unnecessary flushes in between. */ if (cmd_buffer->state.pass) { tu_clean_all_pending(&cmd_buffer->state.renderpass_cache); tu_emit_cache_flush_renderpass(cmd_buffer); } else { tu_clean_all_pending(&cmd_buffer->state.cache); cmd_buffer->state.cache.flush_bits |= TU_CMD_FLAG_CCU_CLEAN_COLOR | TU_CMD_FLAG_CCU_CLEAN_DEPTH; tu_emit_cache_flush(cmd_buffer); } if (cmd_buffer->vk.level == VK_COMMAND_BUFFER_LEVEL_PRIMARY) { trace_end_cmd_buffer(&cmd_buffer->trace, &cmd_buffer->cs, cmd_buffer); } else { trace_end_secondary_cmd_buffer( cmd_buffer->state.pass ? &cmd_buffer->rp_trace : &cmd_buffer->trace, cmd_buffer->state.pass ? &cmd_buffer->draw_cs : &cmd_buffer->cs); } if (TU_DEBUG_START(CHECK_CMD_BUFFER_STATUS)) tu_cmd_buffer_status_gpu_write(cmd_buffer, TU_CMD_BUFFER_STATUS_IDLE); tu_cs_end(&cmd_buffer->cs); tu_cs_end(&cmd_buffer->draw_cs); tu_cs_end(&cmd_buffer->draw_epilogue_cs); return vk_command_buffer_end(&cmd_buffer->vk); } TU_GENX(tu_EndCommandBuffer); static void tu_bind_vs(struct tu_cmd_buffer *cmd, struct tu_shader *vs) { cmd->state.shaders[MESA_SHADER_VERTEX] = vs; } static void tu_bind_tcs(struct tu_cmd_buffer *cmd, struct tu_shader *tcs) { cmd->state.shaders[MESA_SHADER_TESS_CTRL] = tcs; } static void tu_bind_tes(struct tu_cmd_buffer *cmd, struct tu_shader *tes) { if (cmd->state.shaders[MESA_SHADER_TESS_EVAL] != tes) { cmd->state.shaders[MESA_SHADER_TESS_EVAL] = tes; cmd->state.dirty |= TU_CMD_DIRTY_TES; if (!cmd->state.tess_params.valid || cmd->state.tess_params.output_upper_left != tes->tes.tess_output_upper_left || cmd->state.tess_params.output_lower_left != tes->tes.tess_output_lower_left || cmd->state.tess_params.spacing != tes->tes.tess_spacing) { cmd->state.tess_params.output_upper_left = tes->tes.tess_output_upper_left; cmd->state.tess_params.output_lower_left = tes->tes.tess_output_lower_left; cmd->state.tess_params.spacing = tes->tes.tess_spacing; cmd->state.tess_params.valid = true; cmd->state.dirty |= TU_CMD_DIRTY_TESS_PARAMS; } } } static void tu_bind_gs(struct tu_cmd_buffer *cmd, struct tu_shader *gs) { cmd->state.shaders[MESA_SHADER_GEOMETRY] = gs; } static void tu_bind_fs(struct tu_cmd_buffer *cmd, struct tu_shader *fs) { if (cmd->state.shaders[MESA_SHADER_FRAGMENT] != fs) { cmd->state.shaders[MESA_SHADER_FRAGMENT] = fs; cmd->state.dirty |= TU_CMD_DIRTY_LRZ | TU_CMD_DIRTY_FS; } } /* We cannot do this only at pipeline bind time since pipeline * could have been bound at any time before current renderpass, * e.g. in the previous renderpass. */ static void tu_pipeline_update_rp_state(struct tu_cmd_state *cmd_state) { if (cmd_state->pipeline_disable_gmem && !cmd_state->rp.disable_gmem) { /* VK_EXT_attachment_feedback_loop_layout allows feedback loop to involve * not only input attachments but also sampled images or image resources. * But we cannot just patch gmem for image in the descriptors. * * At the moment, in context of DXVK, it is expected that only a few * drawcalls in a frame would use feedback loop and they would be wrapped * in their own renderpasses, so it should be ok to force sysmem. * * However, there are two further possible optimizations if need would * arise for other translation layer: * - Tiling could be enabled if we ensure that there is no barrier in * the renderpass; * - Check that both pipeline and attachments agree that feedback loop * is needed. */ perf_debug( cmd->device, "Disabling gmem due to VK_EXT_attachment_feedback_loop_layout"); cmd_state->rp.disable_gmem = true; cmd_state->rp.gmem_disable_reason = "VK_EXT_attachment_feedback_loop_layout may involve textures"; } if (cmd_state->pipeline_sysmem_single_prim_mode && !cmd_state->rp.sysmem_single_prim_mode) { perf_debug(cmd->device, "single_prim_mode due to pipeline settings"); cmd_state->rp.sysmem_single_prim_mode = true; } if (cmd_state->pipeline_has_tess) { cmd_state->rp.has_tess = true; } } VKAPI_ATTR void VKAPI_CALL tu_CmdBindPipeline(VkCommandBuffer commandBuffer, VkPipelineBindPoint pipelineBindPoint, VkPipeline _pipeline) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); VK_FROM_HANDLE(tu_pipeline, pipeline, _pipeline); if (pipelineBindPoint == VK_PIPELINE_BIND_POINT_COMPUTE) { cmd->state.shaders[MESA_SHADER_COMPUTE] = pipeline->shaders[MESA_SHADER_COMPUTE]; tu_cs_emit_state_ib(&cmd->cs, pipeline->shaders[MESA_SHADER_COMPUTE]->state); cmd->state.compute_load_state = pipeline->load_state; return; } assert(pipelineBindPoint == VK_PIPELINE_BIND_POINT_GRAPHICS); struct tu_graphics_pipeline *gfx_pipeline = tu_pipeline_to_graphics(pipeline); cmd->state.dirty |= TU_CMD_DIRTY_DESC_SETS | TU_CMD_DIRTY_SHADER_CONSTS | TU_CMD_DIRTY_VS_PARAMS | TU_CMD_DIRTY_PROGRAM; tu_bind_vs(cmd, pipeline->shaders[MESA_SHADER_VERTEX]); tu_bind_tcs(cmd, pipeline->shaders[MESA_SHADER_TESS_CTRL]); tu_bind_tes(cmd, pipeline->shaders[MESA_SHADER_TESS_EVAL]); tu_bind_gs(cmd, pipeline->shaders[MESA_SHADER_GEOMETRY]); tu_bind_fs(cmd, pipeline->shaders[MESA_SHADER_FRAGMENT]); /* We precompile static state and count it as dynamic, so we have to * manually clear bitset that tells which dynamic state is set, in order to * make sure that future dynamic state will be emitted. The issue is that * framework remembers only a past REAL dynamic state and compares a new * dynamic state against it, and not against our static state masquaraded * as dynamic. */ BITSET_ANDNOT(cmd->vk.dynamic_graphics_state.set, cmd->vk.dynamic_graphics_state.set, pipeline->static_state_mask); vk_cmd_set_dynamic_graphics_state(&cmd->vk, &gfx_pipeline->dynamic_state); cmd->state.program = pipeline->program; cmd->state.load_state = pipeline->load_state; cmd->state.prim_order_gmem = pipeline->prim_order.state_gmem; cmd->state.pipeline_sysmem_single_prim_mode = pipeline->prim_order.sysmem_single_prim_mode; cmd->state.pipeline_has_tess = pipeline->active_stages & VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT; cmd->state.pipeline_disable_gmem = gfx_pipeline->feedback_loop_may_involve_textures; tu_pipeline_update_rp_state(&cmd->state); if (pipeline->lrz_blend.valid) { if (cmd->state.lrz_blend_status != pipeline->lrz_blend.lrz_blend_status) { cmd->state.lrz_blend_status = pipeline->lrz_blend.lrz_blend_status; cmd->state.dirty |= TU_CMD_DIRTY_LRZ; } } cmd->state.pipeline_blend_lrz = pipeline->lrz_blend.valid; if (pipeline->disable_fs.valid) { if (cmd->state.disable_fs != pipeline->disable_fs.disable_fs) { cmd->state.disable_fs = pipeline->disable_fs.disable_fs; cmd->state.dirty |= TU_CMD_DIRTY_DISABLE_FS; } } cmd->state.pipeline_disable_fs = pipeline->disable_fs.valid; if (pipeline->bandwidth.valid) cmd->state.bandwidth = pipeline->bandwidth; cmd->state.pipeline_bandwidth = pipeline->bandwidth.valid; struct tu_cs *cs = &cmd->draw_cs; /* note: this also avoids emitting draw states before renderpass clears, * which may use the 3D clear path (for MSAA cases) */ if (!(cmd->state.dirty & TU_CMD_DIRTY_DRAW_STATE)) { uint32_t mask = pipeline->set_state_mask; tu_cs_emit_pkt7(cs, CP_SET_DRAW_STATE, 3 * (10 + util_bitcount(mask))); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_PROGRAM_CONFIG, pipeline->program.config_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VS, pipeline->program.vs_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VS_BINNING, pipeline->program.vs_binning_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_HS, pipeline->program.hs_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_DS, pipeline->program.ds_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_GS, pipeline->program.gs_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_GS_BINNING, pipeline->program.gs_binning_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_FS, pipeline->program.fs_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VPC, pipeline->program.vpc_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_PRIM_MODE_GMEM, pipeline->prim_order.state_gmem); u_foreach_bit(i, mask) tu_cs_emit_draw_state(cs, TU_DRAW_STATE_DYNAMIC + i, pipeline->dynamic_state[i]); } cmd->state.pipeline_draw_states = pipeline->set_state_mask; u_foreach_bit(i, pipeline->set_state_mask) cmd->state.dynamic_state[i] = pipeline->dynamic_state[i]; if (pipeline->shaders[MESA_SHADER_FRAGMENT]->fs.has_fdm != cmd->state.has_fdm) { cmd->state.dirty |= TU_CMD_DIRTY_FDM; cmd->state.has_fdm = pipeline->shaders[MESA_SHADER_FRAGMENT]->fs.has_fdm; } if (pipeline->program.per_layer_viewport != cmd->state.per_layer_viewport || pipeline->shaders[MESA_SHADER_FRAGMENT]->fs.max_fdm_layers != cmd->state.max_fdm_layers) { cmd->state.per_layer_viewport = pipeline->program.per_layer_viewport; cmd->state.max_fdm_layers = pipeline->shaders[MESA_SHADER_FRAGMENT]->fs.max_fdm_layers; cmd->state.dirty |= TU_CMD_DIRTY_FDM; } if (pipeline->program.per_view_viewport != cmd->state.per_view_viewport || pipeline->program.fake_single_viewport != cmd->state.fake_single_viewport) { cmd->state.per_view_viewport = pipeline->program.per_view_viewport; cmd->state.fake_single_viewport = pipeline->program.fake_single_viewport; cmd->state.dirty |= TU_CMD_DIRTY_PER_VIEW_VIEWPORT; } if (gfx_pipeline->feedback_loops != cmd->state.pipeline_feedback_loops) { cmd->state.pipeline_feedback_loops = gfx_pipeline->feedback_loops; cmd->state.dirty |= TU_CMD_DIRTY_FEEDBACK_LOOPS | TU_CMD_DIRTY_LRZ; } if (pipeline->program.writes_shading_rate != cmd->state.pipeline_writes_shading_rate || pipeline->program.reads_shading_rate != cmd->state.pipeline_reads_shading_rate) { cmd->state.pipeline_writes_shading_rate = pipeline->program.writes_shading_rate; cmd->state.pipeline_reads_shading_rate = pipeline->program.reads_shading_rate; cmd->state.dirty |= TU_CMD_DIRTY_SHADING_RATE; } bool raster_order_attachment_access = pipeline->output.raster_order_attachment_access || pipeline->ds.raster_order_attachment_access; if (!cmd->state.raster_order_attachment_access_valid || raster_order_attachment_access != cmd->state.raster_order_attachment_access) { cmd->state.raster_order_attachment_access = raster_order_attachment_access; cmd->state.dirty |= TU_CMD_DIRTY_RAST_ORDER; cmd->state.raster_order_attachment_access_valid = true; } } void tu_flush_for_access(struct tu_cache_state *cache, enum tu_cmd_access_mask src_mask, enum tu_cmd_access_mask dst_mask) { BITMASK_ENUM(tu_cmd_flush_bits) flush_bits = 0; if (src_mask & TU_ACCESS_SYSMEM_WRITE) { cache->pending_flush_bits |= TU_CMD_FLAG_ALL_INVALIDATE; } if (src_mask & TU_ACCESS_CP_WRITE) { /* Flush the CP write queue. */ cache->pending_flush_bits |= TU_CMD_FLAG_WAIT_MEM_WRITES | TU_CMD_FLAG_ALL_INVALIDATE; } #define SRC_FLUSH(domain, clean, invalidate) \ if (src_mask & TU_ACCESS_##domain##_WRITE) { \ cache->pending_flush_bits |= TU_CMD_FLAG_##clean | \ (TU_CMD_FLAG_ALL_INVALIDATE & ~TU_CMD_FLAG_##invalidate); \ } SRC_FLUSH(UCHE, CACHE_CLEAN, CACHE_INVALIDATE) SRC_FLUSH(CCU_COLOR, CCU_CLEAN_COLOR, CCU_INVALIDATE_COLOR) SRC_FLUSH(CCU_DEPTH, CCU_CLEAN_DEPTH, CCU_INVALIDATE_DEPTH) #undef SRC_FLUSH #define SRC_INCOHERENT_FLUSH(domain, clean, invalidate) \ if (src_mask & TU_ACCESS_##domain##_INCOHERENT_WRITE) { \ flush_bits |= TU_CMD_FLAG_##clean; \ cache->pending_flush_bits |= \ (TU_CMD_FLAG_ALL_INVALIDATE & ~TU_CMD_FLAG_##invalidate); \ } SRC_INCOHERENT_FLUSH(CCU_COLOR, CCU_CLEAN_COLOR, CCU_INVALIDATE_COLOR) SRC_INCOHERENT_FLUSH(CCU_DEPTH, CCU_CLEAN_DEPTH, CCU_INVALIDATE_DEPTH) SRC_INCOHERENT_FLUSH(UCHE, CACHE_CLEAN, CACHE_INVALIDATE) #undef SRC_INCOHERENT_FLUSH /* Treat host & sysmem write accesses the same, since the kernel implicitly * drains the queue before signalling completion to the host. */ if (dst_mask & (TU_ACCESS_SYSMEM_READ | TU_ACCESS_SYSMEM_WRITE)) { flush_bits |= cache->pending_flush_bits & TU_CMD_FLAG_ALL_CLEAN; } #define DST_FLUSH(domain, clean, invalidate) \ if (dst_mask & (TU_ACCESS_##domain##_READ | \ TU_ACCESS_##domain##_WRITE)) { \ flush_bits |= cache->pending_flush_bits & \ (TU_CMD_FLAG_##invalidate | \ (TU_CMD_FLAG_ALL_CLEAN & ~TU_CMD_FLAG_##clean)); \ } DST_FLUSH(UCHE, CACHE_CLEAN, CACHE_INVALIDATE) DST_FLUSH(CCU_COLOR, CCU_CLEAN_COLOR, CCU_INVALIDATE_COLOR) DST_FLUSH(CCU_DEPTH, CCU_CLEAN_DEPTH, CCU_INVALIDATE_DEPTH) #undef DST_FLUSH #define DST_INCOHERENT_FLUSH(domain, flush, invalidate) \ if (dst_mask & (TU_ACCESS_##domain##_INCOHERENT_READ | \ TU_ACCESS_##domain##_INCOHERENT_WRITE)) { \ flush_bits |= TU_CMD_FLAG_##invalidate | \ (cache->pending_flush_bits & \ (TU_CMD_FLAG_ALL_CLEAN & ~TU_CMD_FLAG_##flush)); \ } DST_INCOHERENT_FLUSH(CCU_COLOR, CCU_CLEAN_COLOR, CCU_INVALIDATE_COLOR) DST_INCOHERENT_FLUSH(CCU_DEPTH, CCU_CLEAN_DEPTH, CCU_INVALIDATE_DEPTH) DST_INCOHERENT_FLUSH(UCHE, CACHE_CLEAN, CACHE_INVALIDATE) if (dst_mask & TU_ACCESS_BINDLESS_DESCRIPTOR_READ) { flush_bits |= TU_CMD_FLAG_BINDLESS_DESCRIPTOR_INVALIDATE; } /* There are multiple incoherent copies of CCHE, so any read through it may * require invalidating it and we cannot optimize away invalidates. */ if (dst_mask & TU_ACCESS_CCHE_READ) { flush_bits |= TU_CMD_FLAG_CCHE_INVALIDATE; } /* The blit cache is a special case dependency between CP_EVENT_WRITE::BLIT * (from GMEM loads/clears) to any GMEM attachment reads done via the UCHE * (Eg: Input attachments/CP_BLIT) which needs an explicit BLIT_CACHE_CLEAN * for the event blit writes to land, it has the following properties: * - Set on reads rather than on writes, like flushes. * - Not executed automatically if pending, like invalidates. * - Pending bits passed through to secondary command buffers, if they're * continuing the render pass. */ if (src_mask & TU_ACCESS_BLIT_WRITE_GMEM) { cache->pending_flush_bits |= TU_CMD_FLAG_BLIT_CACHE_CLEAN; } if ((dst_mask & TU_ACCESS_UCHE_READ_GMEM) && (cache->pending_flush_bits & TU_CMD_FLAG_BLIT_CACHE_CLEAN)) { flush_bits |= TU_CMD_FLAG_BLIT_CACHE_CLEAN; } /* Nothing writes through the RTU cache so there's no point trying to * optimize this. Just always invalidate. */ if (dst_mask & TU_ACCESS_RTU_READ) flush_bits |= TU_CMD_FLAG_RTU_INVALIDATE; #undef DST_INCOHERENT_FLUSH cache->flush_bits |= flush_bits; cache->pending_flush_bits &= ~flush_bits; } /* When translating Vulkan access flags to which cache is accessed * (CCU/UCHE/sysmem), we should take into account both the access flags and * the stage so that accesses with MEMORY_READ_BIT/MEMORY_WRITE_BIT + a * specific stage return something sensible. The specification for * VK_KHR_synchronization2 says that we should do this: * * Additionally, scoping the pipeline stages into the barrier structs * allows the use of the MEMORY_READ and MEMORY_WRITE flags without * sacrificing precision. The per-stage access flags should be used to * disambiguate specific accesses in a given stage or set of stages - for * instance, between uniform reads and sampling operations. * * Note that while in all known cases the stage is actually enough, we should * still narrow things down based on the access flags to handle "old-style" * barriers that may specify a wider range of stages but more precise access * flags. These helpers allow us to do both. */ static bool filter_read_access(VkAccessFlags2 flags, VkPipelineStageFlags2 stages, VkAccessFlags2 tu_flags, VkPipelineStageFlags2 tu_stages) { return (flags & (tu_flags | VK_ACCESS_2_MEMORY_READ_BIT)) && (stages & (tu_stages | VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT)); } static bool filter_write_access(VkAccessFlags2 flags, VkPipelineStageFlags2 stages, VkAccessFlags2 tu_flags, VkPipelineStageFlags2 tu_stages) { return (flags & (tu_flags | VK_ACCESS_2_MEMORY_WRITE_BIT)) && (stages & (tu_stages | VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT)); } static bool gfx_read_access(VkAccessFlags2 flags, VkPipelineStageFlags2 stages, VkAccessFlags2 tu_flags, VkPipelineStageFlags2 tu_stages) { return filter_read_access(flags, stages, tu_flags, tu_stages | VK_PIPELINE_STAGE_2_ALL_GRAPHICS_BIT); } static bool gfx_write_access(VkAccessFlags2 flags, VkPipelineStageFlags2 stages, VkAccessFlags2 tu_flags, VkPipelineStageFlags2 tu_stages) { return filter_write_access(flags, stages, tu_flags, tu_stages | VK_PIPELINE_STAGE_2_ALL_GRAPHICS_BIT); } static enum tu_cmd_access_mask vk2tu_access(VkAccessFlags2 flags, VkAccessFlags3KHR flags2, VkPipelineStageFlags2 stages, bool image_only, bool gmem, bool sparse_aliasing) { BITMASK_ENUM(tu_cmd_access_mask) mask = 0; if (gfx_read_access(flags, stages, VK_ACCESS_2_INDIRECT_COMMAND_READ_BIT | VK_ACCESS_2_CONDITIONAL_RENDERING_READ_BIT_EXT | VK_ACCESS_2_TRANSFORM_FEEDBACK_COUNTER_READ_BIT_EXT | VK_ACCESS_2_HOST_READ_BIT, VK_PIPELINE_STAGE_2_DRAW_INDIRECT_BIT | VK_PIPELINE_STAGE_2_CONDITIONAL_RENDERING_BIT_EXT | VK_PIPELINE_STAGE_2_TRANSFORM_FEEDBACK_BIT_EXT | VK_PIPELINE_STAGE_2_HOST_BIT)) mask |= TU_ACCESS_SYSMEM_READ; if (gfx_write_access(flags, stages, VK_ACCESS_2_TRANSFORM_FEEDBACK_COUNTER_READ_BIT_EXT, VK_PIPELINE_STAGE_2_TRANSFORM_FEEDBACK_BIT_EXT)) mask |= TU_ACCESS_CP_WRITE; if (gfx_write_access(flags, stages, VK_ACCESS_2_HOST_WRITE_BIT, VK_PIPELINE_STAGE_2_HOST_BIT)) mask |= TU_ACCESS_SYSMEM_WRITE; #define SHADER_STAGES \ (VK_PIPELINE_STAGE_2_VERTEX_SHADER_BIT | \ VK_PIPELINE_STAGE_2_TESSELLATION_CONTROL_SHADER_BIT | \ VK_PIPELINE_STAGE_2_TESSELLATION_EVALUATION_SHADER_BIT | \ VK_PIPELINE_STAGE_2_GEOMETRY_SHADER_BIT | \ VK_PIPELINE_STAGE_2_PRE_RASTERIZATION_SHADERS_BIT | \ VK_PIPELINE_STAGE_2_FRAGMENT_SHADER_BIT | \ VK_PIPELINE_STAGE_2_COMPUTE_SHADER_BIT) if (gfx_read_access(flags, stages, VK_ACCESS_2_INDEX_READ_BIT | VK_ACCESS_2_VERTEX_ATTRIBUTE_READ_BIT | VK_ACCESS_2_UNIFORM_READ_BIT | VK_ACCESS_2_INPUT_ATTACHMENT_READ_BIT | VK_ACCESS_2_SHADER_READ_BIT | VK_ACCESS_2_SHADER_SAMPLED_READ_BIT | VK_ACCESS_2_SHADER_STORAGE_READ_BIT | VK_ACCESS_2_SHADER_BINDING_TABLE_READ_BIT_KHR | VK_ACCESS_2_ACCELERATION_STRUCTURE_READ_BIT_KHR, VK_PIPELINE_STAGE_2_INDEX_INPUT_BIT | VK_PIPELINE_STAGE_2_VERTEX_INPUT_BIT | VK_PIPELINE_STAGE_2_VERTEX_ATTRIBUTE_INPUT_BIT | VK_PIPELINE_STAGE_2_ACCELERATION_STRUCTURE_BUILD_BIT_KHR | VK_PIPELINE_STAGE_2_ACCELERATION_STRUCTURE_COPY_BIT_KHR | SHADER_STAGES)) { if (sparse_aliasing) mask |= TU_ACCESS_UCHE_INCOHERENT_READ; else mask |= TU_ACCESS_UCHE_READ; mask |= TU_ACCESS_CCHE_READ; } if (gfx_read_access(flags, stages, VK_ACCESS_2_ACCELERATION_STRUCTURE_READ_BIT_KHR, SHADER_STAGES)) mask |= TU_ACCESS_UCHE_READ | TU_ACCESS_CCHE_READ | TU_ACCESS_RTU_READ; /* Reading the AS for copying involves doing CmdDispatchIndirect with the * copy size as a parameter, so it's read by the CP as well as a shader. */ if (gfx_read_access(flags, stages, VK_ACCESS_2_ACCELERATION_STRUCTURE_READ_BIT_KHR, VK_PIPELINE_STAGE_2_ACCELERATION_STRUCTURE_BUILD_BIT_KHR | VK_PIPELINE_STAGE_2_ACCELERATION_STRUCTURE_COPY_BIT_KHR)) mask |= TU_ACCESS_SYSMEM_READ | TU_ACCESS_UCHE_READ | TU_ACCESS_CCHE_READ; if (gfx_read_access(flags, stages, VK_ACCESS_2_INPUT_ATTACHMENT_READ_BIT, SHADER_STAGES)) { mask |= TU_ACCESS_UCHE_READ_GMEM; if (sparse_aliasing) mask |= TU_ACCESS_UCHE_INCOHERENT_READ; } if (gfx_read_access(flags, stages, VK_ACCESS_2_DESCRIPTOR_BUFFER_READ_BIT_EXT, SHADER_STAGES)) { if (sparse_aliasing) mask |= TU_ACCESS_UCHE_INCOHERENT_READ; else mask |= TU_ACCESS_UCHE_READ; mask |= TU_ACCESS_BINDLESS_DESCRIPTOR_READ | TU_ACCESS_CCHE_READ; } if (gfx_write_access(flags, stages, VK_ACCESS_2_SHADER_WRITE_BIT | VK_ACCESS_2_SHADER_STORAGE_WRITE_BIT | VK_ACCESS_2_TRANSFORM_FEEDBACK_WRITE_BIT_EXT, VK_PIPELINE_STAGE_2_TRANSFORM_FEEDBACK_BIT_EXT | SHADER_STAGES)) { if (sparse_aliasing) mask |= TU_ACCESS_UCHE_INCOHERENT_WRITE; else mask |= TU_ACCESS_UCHE_WRITE; } if (gfx_write_access(flags, stages, VK_ACCESS_2_ACCELERATION_STRUCTURE_WRITE_BIT_KHR, VK_PIPELINE_STAGE_2_ACCELERATION_STRUCTURE_BUILD_BIT_KHR)) mask |= TU_ACCESS_UCHE_WRITE | TU_ACCESS_CP_WRITE; /* When using GMEM, the CCU is always flushed automatically to GMEM, and * then GMEM is flushed to sysmem. Furthermore, we already had to flush any * previous writes in sysmem mode when transitioning to GMEM. Therefore we * can ignore CCU and pretend that color attachments and transfers use * sysmem directly. */ if (gfx_read_access(flags, stages, VK_ACCESS_2_COLOR_ATTACHMENT_READ_BIT | VK_ACCESS_2_COLOR_ATTACHMENT_READ_NONCOHERENT_BIT_EXT, VK_PIPELINE_STAGE_2_COLOR_ATTACHMENT_OUTPUT_BIT)) { if (gmem) mask |= TU_ACCESS_SYSMEM_READ; else mask |= TU_ACCESS_CCU_COLOR_INCOHERENT_READ; } if (gfx_read_access(flags, stages, VK_ACCESS_2_DEPTH_STENCIL_ATTACHMENT_READ_BIT, VK_PIPELINE_STAGE_2_EARLY_FRAGMENT_TESTS_BIT | VK_PIPELINE_STAGE_2_LATE_FRAGMENT_TESTS_BIT)) { if (gmem) mask |= TU_ACCESS_SYSMEM_READ; else mask |= TU_ACCESS_CCU_DEPTH_INCOHERENT_READ; } if (gfx_write_access(flags, stages, VK_ACCESS_2_COLOR_ATTACHMENT_WRITE_BIT, VK_PIPELINE_STAGE_2_COLOR_ATTACHMENT_OUTPUT_BIT)) { if (gmem) { mask |= TU_ACCESS_SYSMEM_WRITE; } else { mask |= TU_ACCESS_CCU_COLOR_INCOHERENT_WRITE; } } if (gfx_write_access(flags, stages, VK_ACCESS_2_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT, VK_PIPELINE_STAGE_2_EARLY_FRAGMENT_TESTS_BIT | VK_PIPELINE_STAGE_2_LATE_FRAGMENT_TESTS_BIT)) { if (gmem) { mask |= TU_ACCESS_SYSMEM_WRITE; } else { mask |= TU_ACCESS_CCU_DEPTH_INCOHERENT_WRITE; } } if (filter_write_access(flags, stages, VK_ACCESS_2_TRANSFER_WRITE_BIT, VK_PIPELINE_STAGE_2_COPY_BIT | VK_PIPELINE_STAGE_2_BLIT_BIT | VK_PIPELINE_STAGE_2_CLEAR_BIT | VK_PIPELINE_STAGE_2_RESOLVE_BIT | VK_PIPELINE_STAGE_2_ALL_TRANSFER_BIT)) { if (gmem) { mask |= TU_ACCESS_SYSMEM_WRITE; } else if (image_only && !sparse_aliasing) { /* Because we always split up blits/copies of images involving * multiple layers, we always access each layer in the same way, with * the same base address, same format, etc. This means we can avoid * flushing between multiple writes to the same image. This elides * flushes between e.g. multiple blits to the same image. */ mask |= TU_ACCESS_CCU_COLOR_WRITE; } else { mask |= TU_ACCESS_CCU_COLOR_INCOHERENT_WRITE; } } if (filter_read_access(flags, stages, VK_ACCESS_2_TRANSFER_READ_BIT, VK_PIPELINE_STAGE_2_COPY_BIT | VK_PIPELINE_STAGE_2_BLIT_BIT | VK_PIPELINE_STAGE_2_RESOLVE_BIT | VK_PIPELINE_STAGE_2_ALL_TRANSFER_BIT)) { if (sparse_aliasing) mask |= TU_ACCESS_UCHE_INCOHERENT_READ; else mask |= TU_ACCESS_UCHE_READ; mask |= TU_ACCESS_CCHE_READ; } return mask; } /* These helpers deal with legacy BOTTOM_OF_PIPE/TOP_OF_PIPE stages. */ static VkPipelineStageFlags2 sanitize_src_stage(VkPipelineStageFlags2 stage_mask) { /* From the Vulkan spec: * * VK_PIPELINE_STAGE_2_TOP_OF_PIPE_BIT is ... equivalent to * VK_PIPELINE_STAGE_2_NONE in the first scope. * * VK_PIPELINE_STAGE_2_BOTTOM_OF_PIPE_BIT is equivalent to * VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT with VkAccessFlags2 set to 0 * when specified in the first synchronization scope, ... */ if (stage_mask & VK_PIPELINE_STAGE_2_BOTTOM_OF_PIPE_BIT) return VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT; return stage_mask & ~VK_PIPELINE_STAGE_2_TOP_OF_PIPE_BIT; } static VkPipelineStageFlags2 sanitize_dst_stage(VkPipelineStageFlags2 stage_mask) { /* From the Vulkan spec: * * VK_PIPELINE_STAGE_2_TOP_OF_PIPE_BIT is equivalent to * VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT with VkAccessFlags2 set to 0 * when specified in the second synchronization scope, ... * * VK_PIPELINE_STAGE_2_BOTTOM_OF_PIPE_BIT is ... equivalent to * VK_PIPELINE_STAGE_2_NONE in the second scope. * */ if (stage_mask & VK_PIPELINE_STAGE_2_TOP_OF_PIPE_BIT) return VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT; return stage_mask & ~VK_PIPELINE_STAGE_2_BOTTOM_OF_PIPE_BIT; } static enum tu_stage vk2tu_single_stage(struct tu_device *dev, VkPipelineStageFlags2 vk_stage, bool dst) { /* If the destination stage is executed on the CP, then the CP also has to * wait for any WFI's to finish. This is already done for draw calls, * including before indirect param reads, for the most part, so we just * need to WFI and can use TU_STAGE_GPU. * * However, some indirect draw opcodes, depending on firmware, don't have * implicit CP_WAIT_FOR_ME so we have to handle it manually. * * Transform feedback counters are read via CP_MEM_TO_REG, which implicitly * does CP_WAIT_FOR_ME, so we don't include them here. * * Currently we read the draw predicate using CP_MEM_TO_MEM, which * also implicitly does CP_WAIT_FOR_ME. However CP_DRAW_PRED_SET does *not* * implicitly do CP_WAIT_FOR_ME, it seems to only wait for counters to * complete since it's written for DX11 where you can only predicate on the * result of a query object. So if we implement 64-bit comparisons in the * future, or if CP_DRAW_PRED_SET grows the capability to do 32-bit * comparisons, then this will have to be dealt with. */ if (vk_stage == VK_PIPELINE_STAGE_2_DRAW_INDIRECT_BIT || vk_stage == VK_PIPELINE_STAGE_2_CONDITIONAL_RENDERING_BIT_EXT || vk_stage == VK_PIPELINE_STAGE_2_FRAGMENT_DENSITY_PROCESS_BIT_EXT) return TU_STAGE_BV_CP; if (vk_stage == VK_PIPELINE_STAGE_2_ALL_GRAPHICS_BIT || vk_stage == VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT) return dst ? TU_STAGE_BV_CP : TU_STAGE_BR; if (vk_stage == VK_PIPELINE_STAGE_2_HOST_BIT) return dst ? TU_STAGE_BOTTOM : TU_STAGE_BV_CP; if (dev->physical_device->info->chip >= 7) { if (vk_stage == VK_PIPELINE_STAGE_2_VERTEX_INPUT_BIT || vk_stage == VK_PIPELINE_STAGE_2_VERTEX_SHADER_BIT || vk_stage == VK_PIPELINE_STAGE_2_TESSELLATION_CONTROL_SHADER_BIT || vk_stage == VK_PIPELINE_STAGE_2_TESSELLATION_EVALUATION_SHADER_BIT || vk_stage == VK_PIPELINE_STAGE_2_GEOMETRY_SHADER_BIT || vk_stage == VK_PIPELINE_STAGE_2_INDEX_INPUT_BIT || vk_stage == VK_PIPELINE_STAGE_2_VERTEX_ATTRIBUTE_INPUT_BIT || vk_stage == VK_PIPELINE_STAGE_2_PRE_RASTERIZATION_SHADERS_BIT || vk_stage == VK_PIPELINE_STAGE_2_TRANSFORM_FEEDBACK_BIT_EXT || vk_stage == VK_PIPELINE_STAGE_2_CONDITIONAL_RENDERING_BIT_EXT) { return dst ? TU_STAGE_BV : TU_STAGE_BR; } } return TU_STAGE_BR; } static enum tu_stage vk2tu_src_stage(struct tu_device *dev, VkPipelineStageFlags2 vk_stages) { enum tu_stage stage = TU_STAGE_BV_CP; u_foreach_bit64 (bit, vk_stages) { enum tu_stage new_stage = vk2tu_single_stage(dev, 1ull << bit, false); stage = MAX2(stage, new_stage); } return stage; } static enum tu_stage vk2tu_dst_stage(struct tu_device *dev, VkPipelineStageFlags2 vk_stages) { enum tu_stage stage = TU_STAGE_BOTTOM; u_foreach_bit64 (bit, vk_stages) { enum tu_stage new_stage = vk2tu_single_stage(dev, 1ull << bit, true); stage = MIN2(stage, new_stage); } return stage; } static void tu_flush_for_stage(struct tu_cache_state *cache, enum tu_stage src_stage, enum tu_stage dst_stage) { /* Even if the source is the host or CP, the destination access could * generate invalidates that we have to wait to complete. */ if (src_stage < TU_STAGE_BR && (cache->flush_bits & TU_CMD_FLAG_ALL_INVALIDATE)) src_stage = TU_STAGE_BR; if (src_stage >= dst_stage) { cache->flush_bits |= TU_CMD_FLAG_WAIT_FOR_IDLE; if (dst_stage <= TU_STAGE_BV) { cache->flush_bits |= TU_CMD_FLAG_WAIT_FOR_BR; if (dst_stage == TU_STAGE_BV_CP) cache->pending_flush_bits |= TU_CMD_FLAG_WAIT_FOR_ME; } } } void tu_render_pass_state_merge(struct tu_render_pass_state *dst, const struct tu_render_pass_state *src) { dst->xfb_used |= src->xfb_used; dst->has_tess |= src->has_tess; dst->has_prim_generated_query_in_rp |= src->has_prim_generated_query_in_rp; dst->has_vtx_stats_query_in_rp |= src->has_vtx_stats_query_in_rp; dst->has_zpass_done_sample_count_write_in_rp |= src->has_zpass_done_sample_count_write_in_rp; dst->disable_gmem |= src->disable_gmem; dst->sysmem_single_prim_mode |= src->sysmem_single_prim_mode; dst->draw_cs_writes_to_cond_pred |= src->draw_cs_writes_to_cond_pred; dst->shared_viewport |= src->shared_viewport; dst->drawcall_count += src->drawcall_count; dst->drawcall_bandwidth_per_sample_sum += src->drawcall_bandwidth_per_sample_sum; if (!dst->lrz_disable_reason && src->lrz_disable_reason) { dst->lrz_disable_reason = src->lrz_disable_reason; dst->lrz_disabled_at_draw = dst->drawcall_count + src->lrz_disabled_at_draw; } if (!dst->lrz_write_disabled_at_draw && src->lrz_write_disabled_at_draw) { dst->lrz_write_disabled_at_draw = dst->drawcall_count + src->lrz_write_disabled_at_draw; } if (!dst->gmem_disable_reason && src->gmem_disable_reason) { dst->gmem_disable_reason = src->gmem_disable_reason; } } void tu_restore_suspended_pass(struct tu_cmd_buffer *cmd, struct tu_cmd_buffer *suspended) { cmd->state.pass = suspended->state.suspended_pass.pass; cmd->state.subpass = suspended->state.suspended_pass.subpass; cmd->state.framebuffer = suspended->state.suspended_pass.framebuffer; cmd->state.attachments = suspended->state.suspended_pass.attachments; cmd->state.clear_values = suspended->state.suspended_pass.clear_values; cmd->state.render_area = suspended->state.suspended_pass.render_area; cmd->state.gmem_layout = suspended->state.suspended_pass.gmem_layout; cmd->state.tiling = &cmd->state.framebuffer->tiling[cmd->state.gmem_layout]; cmd->state.lrz = suspended->state.suspended_pass.lrz; } /* Take the saved pre-chain in "secondary" and copy its commands to "cmd", * appending it after any saved-up commands in "cmd". */ void tu_append_pre_chain(struct tu_cmd_buffer *cmd, struct tu_cmd_buffer *secondary) { tu_cs_add_entries(&cmd->draw_cs, &secondary->pre_chain.draw_cs); tu_cs_add_entries(&cmd->draw_epilogue_cs, &secondary->pre_chain.draw_epilogue_cs); tu_render_pass_state_merge(&cmd->state.rp, &secondary->pre_chain.state); tu_clone_trace(cmd, &cmd->draw_cs, &cmd->rp_trace, &secondary->pre_chain.rp_trace); util_dynarray_append_dynarray(&cmd->fdm_bin_patchpoints, &secondary->pre_chain.fdm_bin_patchpoints); cmd->pre_chain.fdm_offset = secondary->pre_chain.fdm_offset; if (secondary->pre_chain.fdm_offset) { memcpy(cmd->pre_chain.fdm_offsets, secondary->pre_chain.fdm_offsets, sizeof(cmd->pre_chain.fdm_offsets)); } } /* Take the saved post-chain in "secondary" and copy it to "cmd". */ void tu_append_post_chain(struct tu_cmd_buffer *cmd, struct tu_cmd_buffer *secondary) { tu_cs_add_entries(&cmd->draw_cs, &secondary->draw_cs); tu_cs_add_entries(&cmd->draw_epilogue_cs, &secondary->draw_epilogue_cs); tu_clone_trace(cmd, &cmd->draw_cs, &cmd->rp_trace, &secondary->rp_trace); cmd->state.rp = secondary->state.rp; util_dynarray_append_dynarray(&cmd->fdm_bin_patchpoints, &secondary->fdm_bin_patchpoints); } /* Assuming "secondary" is just a sequence of suspended and resuming passes, * copy its state to "cmd". This also works instead of tu_append_post_chain(), * but it's a bit slower because we don't assume that the chain begins in * "secondary" and therefore have to care about the command buffer's * renderpass state. */ void tu_append_pre_post_chain(struct tu_cmd_buffer *cmd, struct tu_cmd_buffer *secondary) { tu_cs_add_entries(&cmd->draw_cs, &secondary->draw_cs); tu_cs_add_entries(&cmd->draw_epilogue_cs, &secondary->draw_epilogue_cs); tu_clone_trace(cmd, &cmd->draw_cs, &cmd->rp_trace, &secondary->rp_trace); tu_render_pass_state_merge(&cmd->state.rp, &secondary->state.rp); util_dynarray_append_dynarray(&cmd->fdm_bin_patchpoints, &secondary->fdm_bin_patchpoints); } /* Take the current render pass state and save it to "pre_chain" to be * combined later. */ static void tu_save_pre_chain(struct tu_cmd_buffer *cmd) { tu_cs_add_entries(&cmd->pre_chain.draw_cs, &cmd->draw_cs); tu_cs_add_entries(&cmd->pre_chain.draw_epilogue_cs, &cmd->draw_epilogue_cs); u_trace_move(&cmd->pre_chain.rp_trace, &cmd->rp_trace); cmd->pre_chain.state = cmd->state.rp; util_dynarray_append_dynarray(&cmd->pre_chain.fdm_bin_patchpoints, &cmd->fdm_bin_patchpoints); cmd->pre_chain.patchpoints_ctx = cmd->patchpoints_ctx; cmd->patchpoints_ctx = NULL; } VKAPI_ATTR void VKAPI_CALL tu_CmdExecuteCommands(VkCommandBuffer commandBuffer, uint32_t commandBufferCount, const VkCommandBuffer *pCmdBuffers) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); VkResult result; assert(commandBufferCount > 0); /* Emit any pending flushes. */ if (cmd->state.pass) { tu_clean_all_pending(&cmd->state.renderpass_cache); TU_CALLX(cmd->device, tu_emit_cache_flush_renderpass)(cmd); } else { tu_clean_all_pending(&cmd->state.cache); TU_CALLX(cmd->device, tu_emit_cache_flush)(cmd); } for (uint32_t i = 0; i < commandBufferCount; i++) { VK_FROM_HANDLE(tu_cmd_buffer, secondary, pCmdBuffers[i]); if (secondary->usage_flags & VK_COMMAND_BUFFER_USAGE_RENDER_PASS_CONTINUE_BIT) { assert(tu_cs_is_empty(&secondary->cs)); TU_CALLX(cmd->device, tu_lrz_flush_valid_during_renderpass) (cmd, &cmd->draw_cs); result = tu_cs_add_entries(&cmd->draw_cs, &secondary->draw_cs); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); break; } result = tu_cs_add_entries(&cmd->draw_epilogue_cs, &secondary->draw_epilogue_cs); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); break; } /* If LRZ was made invalid in secondary - we should disable * LRZ retroactively for the whole renderpass. */ if (!secondary->state.lrz.valid) cmd->state.lrz.valid = false; if (secondary->state.lrz.gpu_dir_set) cmd->state.lrz.gpu_dir_set = true; if (cmd->state.lrz.prev_direction == TU_LRZ_UNKNOWN && secondary->state.lrz.prev_direction != TU_LRZ_UNKNOWN) cmd->state.lrz.prev_direction = secondary->state.lrz.prev_direction; tu_clone_trace(cmd, &cmd->draw_cs, &cmd->rp_trace, &secondary->rp_trace); tu_render_pass_state_merge(&cmd->state.rp, &secondary->state.rp); util_dynarray_append_dynarray(&cmd->fdm_bin_patchpoints, &secondary->fdm_bin_patchpoints); } else { struct tu_cs *cs = &cmd->cs; /* If the secondary can be used multiple times, we have to set its * patchpoints on the GPU. Set them here, and create a new * patchpoint pointing to the CP_MEM_WRITE packet. Otherwise just * copy them over adjusting the index. */ bool simultaneous_use = secondary->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT; /* If this cmdbuf itself can be used multiple times in a submit then * its patchpoint will also be updated on the GPU. */ if (cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT) tu_cs_set_writeable(cs, true); util_dynarray_foreach (&secondary->vis_stream_patchpoints, struct tu_vis_stream_patchpoint, secondary_patchpoint) { struct tu_vis_stream_patchpoint patchpoint = *secondary_patchpoint; patchpoint.render_pass_idx += cmd->state.tile_render_pass_count; if (simultaneous_use) { tu_cs_reserve_space(cs, 5); tu_cs_emit_pkt7(cs, CP_MEM_WRITE, 4); tu_cs_emit_qw(cs, patchpoint.iova); patchpoint.iova = tu_cs_get_cur_iova(cs); patchpoint.data = cs->cur; tu_cs_emit_qw(cs, 0); } util_dynarray_append(&cmd->vis_stream_patchpoints, patchpoint); } if (cmd->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT) tu_cs_set_writeable(cs, false); if (simultaneous_use) { tu_cs_emit_pkt7(cs, CP_WAIT_MEM_WRITES, 0); tu_cs_emit_pkt7(cs, CP_WAIT_FOR_ME, 0); /* Make BV wait for updates on BR to land */ if (cmd->device->physical_device->info->chip >= 7) { tu_cs_emit_pkt7(cs, CP_THREAD_CONTROL, 1); tu_cs_emit(cs, CP_THREAD_CONTROL_0_THREAD(CP_SET_THREAD_BR) | CP_THREAD_CONTROL_0_SYNC_THREADS); } } cmd->state.tile_render_pass_count += secondary->state.tile_render_pass_count; cmd->vsc_size = MAX2(cmd->vsc_size, secondary->vsc_size); switch (secondary->state.suspend_resume) { case SR_NONE: assert(tu_cs_is_empty(&secondary->draw_cs)); assert(tu_cs_is_empty(&secondary->draw_epilogue_cs)); tu_cs_add_entries(&cmd->cs, &secondary->cs); tu_clone_trace(cmd, &cmd->cs, &cmd->trace, &secondary->trace); break; case SR_IN_PRE_CHAIN: /* cmd may be empty, which means that the chain begins before cmd * in which case we have to update its state. */ if (cmd->state.suspend_resume == SR_NONE) { cmd->state.suspend_resume = SR_IN_PRE_CHAIN; } /* The secondary is just a continuous suspend/resume chain so we * just have to append it to the the command buffer. */ assert(tu_cs_is_empty(&secondary->cs)); tu_append_pre_post_chain(cmd, secondary); break; case SR_AFTER_PRE_CHAIN: case SR_IN_CHAIN: case SR_IN_CHAIN_AFTER_PRE_CHAIN: if (secondary->state.suspend_resume == SR_AFTER_PRE_CHAIN || secondary->state.suspend_resume == SR_IN_CHAIN_AFTER_PRE_CHAIN) { tu_append_pre_chain(cmd, secondary); /* We're about to render, so we need to end the command stream * in case there were any extra commands generated by copying * the trace. */ tu_cs_end(&cmd->draw_cs); tu_cs_end(&cmd->draw_epilogue_cs); switch (cmd->state.suspend_resume) { case SR_NONE: case SR_IN_PRE_CHAIN: /* The renderpass chain ends in the secondary but isn't * started in the primary, so we have to move the state to * `pre_chain`. */ tu_save_pre_chain(cmd); cmd->state.suspend_resume = SR_AFTER_PRE_CHAIN; break; case SR_IN_CHAIN: case SR_IN_CHAIN_AFTER_PRE_CHAIN: { /* The renderpass ends in the secondary and starts somewhere * earlier in this primary. Since the last render pass in * the chain is in the secondary, we are technically outside * of a render pass. Fix that here by reusing the dynamic * render pass that was setup for the last suspended render * pass before the secondary. */ tu_restore_suspended_pass(cmd, cmd); const struct VkOffset2D *fdm_offsets = cmd->pre_chain.fdm_offset ? cmd->pre_chain.fdm_offsets : NULL; TU_CALLX(cmd->device, tu_cmd_render)(cmd, fdm_offsets); if (cmd->state.suspend_resume == SR_IN_CHAIN) cmd->state.suspend_resume = SR_NONE; else cmd->state.suspend_resume = SR_AFTER_PRE_CHAIN; break; } case SR_AFTER_PRE_CHAIN: UNREACHABLE("resuming render pass is not preceded by suspending one"); } tu_reset_render_pass(cmd); } tu_cs_add_entries(&cmd->cs, &secondary->cs); if (secondary->state.suspend_resume == SR_IN_CHAIN_AFTER_PRE_CHAIN || secondary->state.suspend_resume == SR_IN_CHAIN) { /* The secondary ends in a "post-chain" (the opposite of a * pre-chain) that we need to copy into the current command * buffer. */ tu_append_post_chain(cmd, secondary); cmd->state.suspended_pass = secondary->state.suspended_pass; switch (cmd->state.suspend_resume) { case SR_NONE: cmd->state.suspend_resume = SR_IN_CHAIN; break; case SR_AFTER_PRE_CHAIN: cmd->state.suspend_resume = SR_IN_CHAIN_AFTER_PRE_CHAIN; break; default: UNREACHABLE("suspending render pass is followed by a not resuming one"); } } } cmd->state.total_renderpasses += secondary->state.total_renderpasses; cmd->state.total_dispatches += secondary->state.total_dispatches; } cmd->state.index_size = secondary->state.index_size; /* for restart index update */ } cmd->state.dirty = ~0u; /* TODO: set dirty only what needs to be */ if (!cmd->state.lrz.gpu_dir_tracking && cmd->state.pass) { /* After a secondary command buffer is executed, LRZ is not valid * until it is cleared again. */ cmd->state.lrz.valid = false; } /* After executing secondary command buffers, there may have been arbitrary * flushes executed, so when we encounter a pipeline barrier with a * srcMask, we have to assume that we need to invalidate. Therefore we need * to re-initialize the cache with all pending invalidate bits set. */ if (cmd->state.pass) { struct tu_cache_state *cache = &cmd->state.renderpass_cache; BITMASK_ENUM(tu_cmd_flush_bits) retained_pending_flush_bits = cache->pending_flush_bits & TU_CMD_FLAG_BLIT_CACHE_CLEAN; tu_cache_init(cache); cache->pending_flush_bits |= retained_pending_flush_bits; } else { tu_cache_init(&cmd->state.cache); } } static void tu_subpass_barrier(struct tu_cmd_buffer *cmd_buffer, const struct tu_subpass_barrier *barrier, bool external) { /* Note: we don't know until the end of the subpass whether we'll use * sysmem, so assume sysmem here to be safe. */ struct tu_cache_state *cache = external ? &cmd_buffer->state.cache : &cmd_buffer->state.renderpass_cache; VkPipelineStageFlags2 src_stage_vk = sanitize_src_stage(barrier->src_stage_mask); VkPipelineStageFlags2 dst_stage_vk = sanitize_dst_stage(barrier->dst_stage_mask); BITMASK_ENUM(tu_cmd_access_mask) src_flags = vk2tu_access(barrier->src_access_mask, barrier->src_access_mask2, src_stage_vk, false, false, cmd_buffer->device->vk.enabled_features.sparseResidencyAliased); BITMASK_ENUM(tu_cmd_access_mask) dst_flags = vk2tu_access(barrier->dst_access_mask, barrier->dst_access_mask2, dst_stage_vk, false, false, cmd_buffer->device->vk.enabled_features.sparseResidencyAliased); if (barrier->incoherent_ccu_color) src_flags |= TU_ACCESS_CCU_COLOR_INCOHERENT_WRITE; if (barrier->incoherent_ccu_depth) src_flags |= TU_ACCESS_CCU_DEPTH_INCOHERENT_WRITE; tu_flush_for_access(cache, src_flags, dst_flags); enum tu_stage src_stage = vk2tu_src_stage(cmd_buffer->device, src_stage_vk); enum tu_stage dst_stage = vk2tu_dst_stage(cmd_buffer->device, dst_stage_vk); tu_flush_for_stage(cache, src_stage, dst_stage); } template static void tu_emit_subpass_begin_gmem(struct tu_cmd_buffer *cmd, struct tu_resolve_group *resolve_group) { struct tu_cs *cs = &cmd->draw_cs; uint32_t subpass_idx = cmd->state.subpass - cmd->state.pass->subpasses; const struct tu_vsc_config *vsc = tu_vsc_config(cmd, cmd->state.tiling); /* If we might choose to bin, then put the loads under a check for geometry * having been binned to this tile. If we don't choose to bin in the end, * then we will have manually set those registers to say geometry is present. * * However, if the draw CS has a write to the condition for some other reason * (perf queries), then we can't do this optimization since the * start-of-the-CS geometry condition will have been overwritten. */ bool cond_load_allowed = vsc->binning && cmd->state.pass->has_cond_load_store && !cmd->state.rp.draw_cs_writes_to_cond_pred; if (cmd->state.pass->has_fdm) tu_cs_set_writeable(cs, true); tu_cond_exec_start(cs, CP_COND_EXEC_0_RENDER_MODE_GMEM); /* This appears to be necessary when stores are followed by loads to the * same memory in GMEM, to prevent the loads from starting before the * stores have completed. See * dEQP-VK.pipeline.monolithic.multisample.multisampled_render_to_single_sampled.input_attachments.initialize.r8g8b8a8_unorm_r16g16b16a16_sfloat_r16g16b16a16_sint_d16_unorm.2x.ds_resolve_sample_zero.whole_framebuffer * for a testcase. * * TODO: why is this not necessary between the end of one tile and the * start of another? */ if (subpass_idx != 0) { tu_emit_event_write(cmd, cs, FD_CCU_CLEAN_BLIT_CACHE); } /* Emit gmem loads that are first used in this subpass. */ bool emitted_scissor = false; for (uint32_t i = 0; i < cmd->state.pass->attachment_count; ++i) { struct tu_render_pass_attachment *att = &cmd->state.pass->attachments[i]; if ((att->load || att->load_stencil) && att->first_subpass_idx == subpass_idx) { if (!emitted_scissor) { tu6_emit_blit_scissor(cmd, cs, true, false); emitted_scissor = true; } tu_load_gmem_attachment(cmd, cs, resolve_group, i, i, cond_load_allowed, false); } } /* Emit unresolves that replicate single-sampled attachments into * multisampled GMEM attachments. */ for (uint32_t i = 0; i < cmd->state.subpass->unresolve_count; ++i) { uint32_t a = cmd->state.subpass->unresolve_attachments[i].attachment; if (a == VK_ATTACHMENT_UNUSED) continue; uint32_t gmem_a = tu_subpass_get_attachment_to_unresolve(cmd->state.subpass, i); tu_load_gmem_attachment(cmd, cs, resolve_group, a, gmem_a, cond_load_allowed, true); } if (!cmd->device->physical_device->info->a7xx.has_generic_clear) { /* Emit gmem clears that are first used in this subpass. */ emitted_scissor = false; for (uint32_t i = 0; i < cmd->state.pass->attachment_count; ++i) { struct tu_render_pass_attachment *att = &cmd->state.pass->attachments[i]; if (att->clear_mask && att->first_subpass_idx == subpass_idx) { if (!emitted_scissor) { tu6_emit_blit_scissor(cmd, cs, false, false); emitted_scissor = true; } tu_clear_gmem_attachment(cmd, cs, resolve_group, i); } } } tu_cond_exec_end(cs); /* CP_COND_EXEC_0_RENDER_MODE_GMEM */ if (cmd->state.pass->has_fdm) tu_cs_set_writeable(cs, false); } /* Emits sysmem clears that are first used in this subpass. */ template static void tu_emit_subpass_begin_sysmem(struct tu_cmd_buffer *cmd) { if (cmd->device->physical_device->info->a7xx.has_generic_clear && !cmd->state.subpass->unresolve_count) return; struct tu_cs *cs = &cmd->draw_cs; uint32_t subpass_idx = cmd->state.subpass - cmd->state.pass->subpasses; tu_cond_exec_start(cs, CP_COND_EXEC_0_RENDER_MODE_SYSMEM); tu6_emit_sysmem_unresolves(cmd, cs, cmd->state.subpass); for (uint32_t i = 0; i < cmd->state.pass->attachment_count; ++i) { struct tu_render_pass_attachment *att = &cmd->state.pass->attachments[i]; if (att->clear_mask && att->first_subpass_idx == subpass_idx) tu_clear_sysmem_attachment(cmd, cs, i); } tu_cond_exec_end(cs); /* sysmem */ } static void tu7_emit_subpass_clear(struct tu_cmd_buffer *cmd, struct tu_resolve_group *resolve_group) { if (cmd->state.render_area.extent.width == 0 || cmd->state.render_area.extent.height == 0) return; struct tu_cs *cs = &cmd->draw_cs; uint32_t subpass_idx = cmd->state.subpass - cmd->state.pass->subpasses; tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(RENDER_MODE) | CP_COND_REG_EXEC_0_GMEM | CP_COND_REG_EXEC_0_SYSMEM); bool emitted_scissor = false; for (uint32_t i = 0; i < cmd->state.pass->attachment_count; ++i) { struct tu_render_pass_attachment *att = &cmd->state.pass->attachments[i]; if (att->clear_mask && att->first_subpass_idx == subpass_idx) { if (!emitted_scissor) { tu6_emit_blit_scissor(cmd, cs, false, true); emitted_scissor = true; } tu7_generic_clear_attachment(cmd, cs, resolve_group, i); } } tu_cond_exec_end(cs); } static void tu7_emit_subpass_shading_rate(struct tu_cmd_buffer *cmd, const struct tu_subpass *subpass, struct tu_cs *cs) { if (subpass->fsr_attachment == VK_ATTACHMENT_UNUSED) { tu_cs_emit_regs(cs, A7XX_GRAS_QUALITY_BUFFER_INFO(), A7XX_GRAS_QUALITY_BUFFER_DIMENSION()); tu_cs_emit_regs(cs, A7XX_GRAS_QUALITY_BUFFER_PITCH()); tu_cs_emit_regs(cs, A7XX_GRAS_QUALITY_BUFFER_BASE()); /* We need to invalidate cache when changing to NULL FSR attachment, but * only once. */ if (!cmd->prev_fsr_is_null) { tu_emit_raw_event_write(cmd, cs, LRZ_Q_CACHE_INVALIDATE, false); cmd->prev_fsr_is_null = true; } return; } const struct tu_image_view *iview = cmd->state.attachments[subpass->fsr_attachment]; assert(iview->vk.format == VK_FORMAT_R8_UINT); tu_cs_emit_regs( cs, A7XX_GRAS_QUALITY_BUFFER_INFO(.layered = true, .tile_mode = (a6xx_tile_mode) iview->image->layout[0] .tile_mode, ), A7XX_GRAS_QUALITY_BUFFER_DIMENSION(.width = iview->view.width, .height = iview->view.height)); tu_cs_emit_regs( cs, A7XX_GRAS_QUALITY_BUFFER_PITCH(.pitch = iview->view.pitch, .array_pitch = iview->view.layer_size)); tu_cs_emit_regs(cs, A7XX_GRAS_QUALITY_BUFFER_BASE(.qword = iview->view.base_addr)); tu_emit_raw_event_write(cmd, cs, LRZ_Q_CACHE_INVALIDATE, false); cmd->prev_fsr_is_null = false; } /* emit loads, clears, and mrt/zs/msaa/ubwc state for the subpass that is * starting (either at vkCmdBeginRenderPass2() or vkCmdNextSubpass2()) * * Clears and loads have to happen at this point, because with * VK_ATTACHMENT_DESCRIPTION_MAY_ALIAS_BIT the loads may depend on the output of * a previous aliased attachment's store. */ template static void tu_emit_subpass_begin(struct tu_cmd_buffer *cmd) { tu_fill_render_pass_state(&cmd->state.vk_rp, cmd->state.pass, cmd->state.subpass); struct tu_resolve_group resolve_group = {}; tu_emit_subpass_begin_gmem(cmd, &resolve_group); tu_emit_subpass_begin_sysmem(cmd); if (cmd->device->physical_device->info->a7xx.has_generic_clear) { tu7_emit_subpass_clear(cmd, &resolve_group); } tu_emit_resolve_group(cmd, &cmd->draw_cs, &resolve_group); tu6_emit_zs(cmd, cmd->state.subpass, &cmd->draw_cs); tu6_emit_mrt(cmd, cmd->state.subpass, &cmd->draw_cs); tu6_emit_render_cntl(cmd, cmd->state.subpass, &cmd->draw_cs, false); if (CHIP >= A7XX) { tu7_emit_subpass_shading_rate(cmd, cmd->state.subpass, &cmd->draw_cs); } tu_set_input_attachments(cmd, cmd->state.subpass); vk_cmd_set_cb_attachment_count(&cmd->vk, cmd->state.subpass->color_count); cmd->state.dirty |= TU_CMD_DIRTY_SUBPASS; } template VKAPI_ATTR void VKAPI_CALL tu_CmdBeginRenderPass2(VkCommandBuffer commandBuffer, const VkRenderPassBeginInfo *pRenderPassBegin, const VkSubpassBeginInfo *pSubpassBeginInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); if (TU_DEBUG(DYNAMIC)) { vk_common_CmdBeginRenderPass2(commandBuffer, pRenderPassBegin, pSubpassBeginInfo); return; } VK_FROM_HANDLE(tu_render_pass, pass, pRenderPassBegin->renderPass); VK_FROM_HANDLE(tu_framebuffer, fb, pRenderPassBegin->framebuffer); const struct VkRenderPassAttachmentBeginInfo *pAttachmentInfo = vk_find_struct_const(pRenderPassBegin->pNext, RENDER_PASS_ATTACHMENT_BEGIN_INFO); cmd->state.pass = pass; cmd->state.subpass = pass->subpasses; cmd->state.framebuffer = fb; cmd->state.render_area = pRenderPassBegin->renderArea; cmd->state.fdm_per_layer = pass->has_layered_fdm; if (pass->attachment_count > 0) { VK_MULTIALLOC(ma); vk_multialloc_add(&ma, &cmd->state.attachments, const struct tu_image_view *, pass->attachment_count); vk_multialloc_add(&ma, &cmd->state.clear_values, VkClearValue, pass->attachment_count); if (!vk_multialloc_alloc(&ma, &cmd->vk.pool->alloc, VK_SYSTEM_ALLOCATION_SCOPE_OBJECT)) { vk_command_buffer_set_error(&cmd->vk, VK_ERROR_OUT_OF_HOST_MEMORY); return; } } if (cmd->device->dbg_renderpass_stomp_cs) { tu_cs_emit_call(&cmd->cs, cmd->device->dbg_renderpass_stomp_cs); } for (unsigned i = 0; i < pass->user_attachment_count; i++) { cmd->state.attachments[i] = pAttachmentInfo ? tu_image_view_from_handle(pAttachmentInfo->pAttachments[i]) : cmd->state.framebuffer->attachments[i]; } for (unsigned i = 0; i < pass->attachment_count - pass->user_attachment_count; i++) { /* With imageless attachments, the only attachments in the framebuffer * are MSRTSS attachments. Without imageless attachments, they are after * the user's attachments. */ unsigned fb_idx = i + (pAttachmentInfo ? 0 : pass->user_attachment_count); cmd->state.attachments[i + pass->user_attachment_count] = cmd->state.framebuffer->attachments[fb_idx]; } if (pass->attachment_count) { for (unsigned i = 0; i < MIN2(pRenderPassBegin->clearValueCount, pass->user_attachment_count); i++) { struct tu_render_pass_attachment *att = &pass->attachments[i]; uint32_t idx = i; /* Clear values have to be remapped for MSRTSS, because they may be * moved to the multisample attachment. */ if (att->remapped_clear_att != VK_ATTACHMENT_UNUSED) idx = att->remapped_clear_att; cmd->state.clear_values[idx] = pRenderPassBegin->pClearValues[i]; } } tu_choose_gmem_layout(cmd); /* Note: because this is external, any flushes will happen before draw_cs * gets called. However deferred flushes could have to happen later as part * of the subpass. */ tu_subpass_barrier(cmd, &pass->subpasses[0].start_barrier, true); cmd->state.renderpass_cache.pending_flush_bits = cmd->state.cache.pending_flush_bits; cmd->state.renderpass_cache.flush_bits = 0; if (pass->subpasses[0].feedback_invalidate) { cmd->state.renderpass_cache.flush_bits |= TU_CMD_FLAG_CACHE_INVALIDATE | TU_CMD_FLAG_BLIT_CACHE_CLEAN | TU_CMD_FLAG_WAIT_FOR_IDLE; } tu_lrz_begin_renderpass(cmd); tu_emit_renderpass_begin(cmd); tu_emit_subpass_begin(cmd); cmd->patchpoints_ctx = ralloc_context(NULL); } TU_GENX(tu_CmdBeginRenderPass2); template VKAPI_ATTR void VKAPI_CALL tu_CmdBeginRendering(VkCommandBuffer commandBuffer, const VkRenderingInfo *pRenderingInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); tu_setup_dynamic_render_pass(cmd, pRenderingInfo); tu_setup_dynamic_framebuffer(cmd, pRenderingInfo); cmd->state.pass = &cmd->dynamic_pass; cmd->state.subpass = &cmd->dynamic_subpass; cmd->state.framebuffer = &cmd->dynamic_framebuffer; cmd->state.render_area = pRenderingInfo->renderArea; cmd->state.fdm_per_layer = pRenderingInfo->flags & VK_RENDERING_PER_LAYER_FRAGMENT_DENSITY_BIT_VALVE; cmd->state.blit_cache_cleaned = false; cmd->state.attachments = cmd->dynamic_attachments; cmd->state.clear_values = cmd->dynamic_clear_values; for (unsigned i = 0; i < pRenderingInfo->colorAttachmentCount; i++) { if (!pRenderingInfo->pColorAttachments[i].imageView) continue; uint32_t a = cmd->dynamic_subpass.color_attachments[i].attachment; cmd->state.clear_values[a] = pRenderingInfo->pColorAttachments[i].clearValue; /* With MSRTSS, the user's attachment corresponds to the * resolve/unresolve attachment, not the color attachment. The color * attachment is the transient multisample attachment. However the clear * happens on the multisample attachment, so we don't remap the * clear_values assignment above. */ bool msrtss = false; if (a >= cmd->dynamic_pass.user_attachment_count) { a = cmd->dynamic_pass.attachments[a].user_att; msrtss = true; } VK_FROM_HANDLE(tu_image_view, view, pRenderingInfo->pColorAttachments[i].imageView); cmd->state.attachments[a] = view; a = cmd->dynamic_subpass.resolve_attachments[i].attachment; if (!msrtss && a != VK_ATTACHMENT_UNUSED) { VK_FROM_HANDLE(tu_image_view, resolve_view, pRenderingInfo->pColorAttachments[i].resolveImageView); cmd->state.attachments[a] = resolve_view; } } uint32_t a = cmd->dynamic_subpass.depth_stencil_attachment.attachment; if (pRenderingInfo->pDepthAttachment || pRenderingInfo->pStencilAttachment) { const struct VkRenderingAttachmentInfo *common_info = (pRenderingInfo->pDepthAttachment && pRenderingInfo->pDepthAttachment->imageView != VK_NULL_HANDLE) ? pRenderingInfo->pDepthAttachment : pRenderingInfo->pStencilAttachment; if (common_info && common_info->imageView != VK_NULL_HANDLE) { VK_FROM_HANDLE(tu_image_view, view, common_info->imageView); if (pRenderingInfo->pDepthAttachment) { cmd->state.clear_values[a].depthStencil.depth = pRenderingInfo->pDepthAttachment->clearValue.depthStencil.depth; } if (pRenderingInfo->pStencilAttachment) { cmd->state.clear_values[a].depthStencil.stencil = pRenderingInfo->pStencilAttachment->clearValue.depthStencil.stencil; } bool msrtss = false; if (a >= cmd->dynamic_pass.user_attachment_count) { a = cmd->dynamic_pass.attachments[a].user_att; msrtss = true; } cmd->state.attachments[a] = view; if (!msrtss && cmd->dynamic_subpass.resolve_count > cmd->dynamic_subpass.color_count) { VK_FROM_HANDLE(tu_image_view, resolve_view, common_info->resolveImageView); a = cmd->dynamic_subpass.resolve_attachments[cmd->dynamic_subpass.color_count].attachment; cmd->state.attachments[a] = resolve_view; } } } a = cmd->dynamic_pass.fragment_density_map.attachment; if (a != VK_ATTACHMENT_UNUSED) { const VkRenderingFragmentDensityMapAttachmentInfoEXT *fdm_info = vk_find_struct_const(pRenderingInfo->pNext, RENDERING_FRAGMENT_DENSITY_MAP_ATTACHMENT_INFO_EXT); VK_FROM_HANDLE(tu_image_view, view, fdm_info->imageView); cmd->state.attachments[a] = view; } const VkRenderingAttachmentLocationInfoKHR ral_info = { .sType = VK_STRUCTURE_TYPE_RENDERING_ATTACHMENT_LOCATION_INFO_KHR, .colorAttachmentCount = pRenderingInfo->colorAttachmentCount, }; vk_cmd_set_rendering_attachment_locations(&cmd->vk, &ral_info); cmd->patchpoints_ctx = ralloc_context(NULL); a = cmd->dynamic_subpass.fsr_attachment; if (a != VK_ATTACHMENT_UNUSED) { const VkRenderingFragmentShadingRateAttachmentInfoKHR *fsr_info = vk_find_struct_const(pRenderingInfo->pNext, RENDERING_FRAGMENT_SHADING_RATE_ATTACHMENT_INFO_KHR); VK_FROM_HANDLE(tu_image_view, view, fsr_info->imageView); cmd->state.attachments[a] = view; } VkResult result = tu_setup_dynamic_msrtss(cmd); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } tu_choose_gmem_layout(cmd); cmd->state.renderpass_cache.pending_flush_bits = cmd->state.cache.pending_flush_bits; cmd->state.renderpass_cache.flush_bits = 0; bool resuming = pRenderingInfo->flags & VK_RENDERING_RESUMING_BIT; bool suspending = pRenderingInfo->flags & VK_RENDERING_SUSPENDING_BIT; cmd->state.suspending = suspending; cmd->state.resuming = resuming; if (!resuming && cmd->device->dbg_renderpass_stomp_cs) { tu_cs_emit_call(&cmd->cs, cmd->device->dbg_renderpass_stomp_cs); } /* We can't track LRZ across command buffer boundaries, so we have to * disable LRZ when resuming/suspending unless we can track on the GPU. */ if ((resuming || suspending) && !cmd->device->physical_device->info->a6xx.has_lrz_dir_tracking) { cmd->state.lrz.valid = false; } else { if (resuming) tu_lrz_begin_resumed_renderpass(cmd); else tu_lrz_begin_renderpass(cmd); } if (suspending) { cmd->state.suspended_pass.pass = cmd->state.pass; cmd->state.suspended_pass.subpass = cmd->state.subpass; cmd->state.suspended_pass.framebuffer = cmd->state.framebuffer; cmd->state.suspended_pass.render_area = cmd->state.render_area; cmd->state.suspended_pass.attachments = cmd->state.attachments; cmd->state.suspended_pass.clear_values = cmd->state.clear_values; cmd->state.suspended_pass.gmem_layout = cmd->state.gmem_layout; } if (!resuming) { tu_emit_renderpass_begin(cmd); tu_emit_subpass_begin(cmd); } if (suspending && !resuming) { /* entering a chain */ switch (cmd->state.suspend_resume) { case SR_NONE: cmd->state.suspend_resume = SR_IN_CHAIN; break; case SR_AFTER_PRE_CHAIN: cmd->state.suspend_resume = SR_IN_CHAIN_AFTER_PRE_CHAIN; break; case SR_IN_PRE_CHAIN: case SR_IN_CHAIN: case SR_IN_CHAIN_AFTER_PRE_CHAIN: UNREACHABLE("suspending render pass not followed by resuming pass"); break; } } if (resuming && cmd->state.suspend_resume == SR_NONE) cmd->state.suspend_resume = SR_IN_PRE_CHAIN; } TU_GENX(tu_CmdBeginRendering); template VKAPI_ATTR void VKAPI_CALL tu_CmdSetRenderingAttachmentLocationsKHR( VkCommandBuffer commandBuffer, const VkRenderingAttachmentLocationInfoKHR *pLocationInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); vk_common_CmdSetRenderingAttachmentLocationsKHR(commandBuffer, pLocationInfo); tu6_emit_mrt(cmd, cmd->state.subpass, &cmd->draw_cs); tu6_emit_render_cntl(cmd, cmd->state.subpass, &cmd->draw_cs, false); /* Same case as a drawcall not writing to some color attachments, but not * trying to make LRZ work in cases where we can prove that LRZ can work. */ if (cmd->state.lrz.valid) tu_lrz_disable_write_for_rp(cmd, "CmdSetRenderingAttachmentLocations"); /* Because this is just a remapping and not a different "reference", there * doesn't need to be a barrier between accesses to the same attachment * with a different index. This is different from "classic" renderpasses. * Before a7xx the CCU includes the render target ID in the cache location * calculation, so we need to manually flush/invalidate color CCU here * since the same render target/attachment may be in a different location. */ if (cmd->device->physical_device->info->chip == 6) { struct tu_cache_state *cache = &cmd->state.renderpass_cache; tu_flush_for_access(cache, TU_ACCESS_CCU_COLOR_INCOHERENT_WRITE, TU_ACCESS_CCU_COLOR_INCOHERENT_WRITE); cache->flush_bits |= TU_CMD_FLAG_WAIT_FOR_IDLE; } } TU_GENX(tu_CmdSetRenderingAttachmentLocationsKHR); VKAPI_ATTR void VKAPI_CALL tu_CmdSetRenderingInputAttachmentIndicesKHR( VkCommandBuffer commandBuffer, const VkRenderingInputAttachmentIndexInfoKHR *pLocationInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); vk_common_CmdSetRenderingInputAttachmentIndicesKHR(commandBuffer, pLocationInfo); const struct vk_input_attachment_location_state *ial = &cmd->vk.dynamic_graphics_state.ial; struct tu_subpass *subpass = &cmd->dynamic_subpass; for (unsigned i = 0; i < ARRAY_SIZE(cmd->dynamic_input_attachments); i++) { subpass->input_attachments[i].attachment = VK_ATTACHMENT_UNUSED; } unsigned input_count = 0; for (unsigned i = 0; i < subpass->color_count; i++) { if (ial->color_map[i] == MESA_VK_ATTACHMENT_UNUSED) continue; subpass->input_attachments[ial->color_map[i] + TU_DYN_INPUT_ATT_OFFSET].attachment = subpass->color_attachments[i].attachment; input_count = MAX2(input_count, ial->color_map[i] + TU_DYN_INPUT_ATT_OFFSET + 1); } if (ial->depth_att != MESA_VK_ATTACHMENT_UNUSED) { if (ial->depth_att == MESA_VK_ATTACHMENT_NO_INDEX) { subpass->input_attachments[0].attachment = subpass->depth_stencil_attachment.attachment; input_count = MAX2(input_count, 1); } else { subpass->input_attachments[ial->depth_att + TU_DYN_INPUT_ATT_OFFSET].attachment = subpass->depth_stencil_attachment.attachment; input_count = MAX2(input_count, ial->depth_att + TU_DYN_INPUT_ATT_OFFSET + 1); } } if (ial->stencil_att != MESA_VK_ATTACHMENT_UNUSED) { if (ial->stencil_att == MESA_VK_ATTACHMENT_NO_INDEX) { subpass->input_attachments[0].attachment = subpass->depth_stencil_attachment.attachment; input_count = MAX2(input_count, 1); } else { subpass->input_attachments[ial->stencil_att + TU_DYN_INPUT_ATT_OFFSET].attachment = subpass->depth_stencil_attachment.attachment; input_count = MAX2(input_count, ial->stencil_att + TU_DYN_INPUT_ATT_OFFSET + 1); } } subpass->input_count = input_count; tu_set_input_attachments(cmd, cmd->state.subpass); } template VKAPI_ATTR void VKAPI_CALL tu_CmdNextSubpass2(VkCommandBuffer commandBuffer, const VkSubpassBeginInfo *pSubpassBeginInfo, const VkSubpassEndInfo *pSubpassEndInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); if (TU_DEBUG(DYNAMIC)) { vk_common_CmdNextSubpass2(commandBuffer, pSubpassBeginInfo, pSubpassEndInfo); return; } struct tu_cs *cs = &cmd->draw_cs; const struct tu_subpass *subpass = cmd->state.subpass++; const struct tu_subpass *new_subpass = cmd->state.subpass; /* Track LRZ valid state * * TODO: Improve this tracking for keeping the state of the past depth/stencil images, * so if they become active again, we reuse its old state. */ if (new_subpass->depth_stencil_attachment.attachment != subpass->depth_stencil_attachment.attachment) { cmd->state.lrz.valid = false; cmd->state.dirty |= TU_CMD_DIRTY_LRZ; } if (cmd->state.tiling->possible) { if (cmd->state.pass->has_fdm) tu_cs_set_writeable(cs, true); tu_cond_exec_start(cs, CP_COND_EXEC_0_RENDER_MODE_GMEM); struct tu_resolve_group resolve_group = {}; tu6_emit_gmem_stores(cmd, &cmd->draw_cs, &resolve_group, subpass); tu_emit_resolve_group(cmd, cs, &resolve_group); tu_cond_exec_end(cs); if (cmd->state.pass->has_fdm) tu_cs_set_writeable(cs, false); tu_cond_exec_start(cs, CP_COND_EXEC_0_RENDER_MODE_SYSMEM); } tu6_emit_sysmem_resolves(cmd, cs, subpass); if (cmd->state.tiling->possible) tu_cond_exec_end(cs); /* Handle dependencies for the next subpass */ tu_subpass_barrier(cmd, &cmd->state.subpass->start_barrier, false); if (cmd->state.subpass->feedback_invalidate) { cmd->state.renderpass_cache.flush_bits |= TU_CMD_FLAG_CACHE_INVALIDATE | TU_CMD_FLAG_BLIT_CACHE_CLEAN | TU_CMD_FLAG_WAIT_FOR_IDLE; } tu_emit_subpass_begin(cmd); } TU_GENX(tu_CmdNextSubpass2); static uint32_t tu6_user_consts_size(const struct tu_const_state *const_state, bool ldgk, mesa_shader_stage type) { uint32_t dwords = 0; if (const_state->push_consts.type == IR3_PUSH_CONSTS_PER_STAGE) { unsigned num_units = const_state->push_consts.dwords; dwords += 4 + num_units; assert(num_units > 0); } if (ldgk) { dwords += 6 + (2 * const_state->num_inline_ubos + 4); } else { dwords += 8 * const_state->num_inline_ubos; } return dwords; } static void tu6_emit_per_stage_push_consts(struct tu_cs *cs, const struct tu_const_state *const_state, const struct ir3_const_state *ir_const_state, mesa_shader_stage type, uint32_t *push_constants) { if (const_state->push_consts.type == IR3_PUSH_CONSTS_PER_STAGE) { unsigned num_units = const_state->push_consts.dwords; unsigned offset_vec4 = ir_const_state->allocs.consts[IR3_CONST_ALLOC_PUSH_CONSTS] .offset_vec4; assert(num_units > 0); /* DST_OFF and NUM_UNIT requires vec4 units */ tu_cs_emit_pkt7(cs, tu6_stage2opcode(type), 3 + num_units); tu_cs_emit(cs, CP_LOAD_STATE6_0_DST_OFF(offset_vec4) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_CONSTANTS) | CP_LOAD_STATE6_0_STATE_SRC(SS6_DIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(tu6_stage2shadersb(type)) | CP_LOAD_STATE6_0_NUM_UNIT(num_units / 4)); tu_cs_emit(cs, 0); tu_cs_emit(cs, 0); unsigned lo = const_state->push_consts.lo_dwords; for (unsigned i = 0; i < num_units; i++) tu_cs_emit(cs, push_constants[i + lo]); } } static void tu6_emit_inline_ubo(struct tu_cs *cs, const struct tu_const_state *const_state, unsigned constlen, mesa_shader_stage type, struct tu_descriptor_state *descriptors) { assert(const_state->num_inline_ubos == 0 || !cs->device->physical_device->info->a7xx.load_shader_consts_via_preamble); /* Emit loads of inline uniforms. These load directly from the uniform's * storage space inside the descriptor set. */ for (unsigned i = 0; i < const_state->num_inline_ubos; i++) { const struct tu_inline_ubo *ubo = &const_state->ubos[i]; if (constlen <= ubo->const_offset_vec4) continue; uint64_t va = descriptors->set_iova[ubo->base] & ~0x3f; tu_cs_emit_pkt7(cs, tu6_stage2opcode(type), ubo->push_address ? 7 : 3); tu_cs_emit(cs, CP_LOAD_STATE6_0_DST_OFF(ubo->const_offset_vec4) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_CONSTANTS) | CP_LOAD_STATE6_0_STATE_SRC(ubo->push_address ? SS6_DIRECT : SS6_INDIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(tu6_stage2shadersb(type)) | CP_LOAD_STATE6_0_NUM_UNIT(MIN2(ubo->size_vec4, constlen - ubo->const_offset_vec4))); if (ubo->push_address) { tu_cs_emit(cs, 0); tu_cs_emit(cs, 0); tu_cs_emit_qw(cs, va + ubo->offset); tu_cs_emit(cs, 0); tu_cs_emit(cs, 0); } else { tu_cs_emit_qw(cs, va + ubo->offset); } } } static void tu7_emit_inline_ubo(struct tu_cs *cs, const struct tu_const_state *const_state, const struct ir3_const_state *ir_const_state, unsigned constlen, mesa_shader_stage type, struct tu_descriptor_state *descriptors) { uint64_t addresses[7] = {0}; unsigned offset = const_state->inline_uniforms_ubo.idx; if (offset == -1) return; for (unsigned i = 0; i < const_state->num_inline_ubos; i++) { const struct tu_inline_ubo *ubo = &const_state->ubos[i]; uint64_t va = descriptors->set_iova[ubo->base] & ~0x3f; addresses[i] = va + ubo->offset; } /* A7XX TODO: Emit data via sub_cs instead of NOP */ uint64_t iova = tu_cs_emit_data_nop(cs, (uint32_t *)addresses, const_state->num_inline_ubos * 2, 4); tu_cs_emit_pkt7(cs, tu6_stage2opcode(type), 5); tu_cs_emit(cs, CP_LOAD_STATE6_0_DST_OFF(offset) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_UBO) | CP_LOAD_STATE6_0_STATE_SRC(SS6_DIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(tu6_stage2shadersb(type)) | CP_LOAD_STATE6_0_NUM_UNIT(1)); tu_cs_emit(cs, CP_LOAD_STATE6_1_EXT_SRC_ADDR(0)); tu_cs_emit(cs, CP_LOAD_STATE6_2_EXT_SRC_ADDR_HI(0)); int size_vec4s = DIV_ROUND_UP(const_state->num_inline_ubos * 2, 4); tu_cs_emit_qw(cs, iova | ((uint64_t)A6XX_UBO_1_SIZE(size_vec4s) << 32)); } static void tu_emit_inline_ubo(struct tu_cs *cs, const struct tu_const_state *const_state, const struct ir3_const_state *ir_const_state, unsigned constlen, mesa_shader_stage type, struct tu_descriptor_state *descriptors) { if (!const_state->num_inline_ubos) return; if (cs->device->physical_device->info->a7xx.load_inline_uniforms_via_preamble_ldgk) { tu7_emit_inline_ubo(cs, const_state, ir_const_state, constlen, type, descriptors); } else { tu6_emit_inline_ubo(cs, const_state, constlen, type, descriptors); } } static void tu6_emit_shared_consts(struct tu_cs *cs, const struct tu_push_constant_range *shared_consts, uint32_t *push_constants, bool compute) { if (shared_consts->dwords > 0) { /* Offset and num_units for shared consts are in units of dwords. */ unsigned num_units = shared_consts->dwords; unsigned offset = shared_consts->lo_dwords; enum a6xx_state_type st = compute ? ST6_UBO : ST6_CONSTANTS; uint32_t cp_load_state = compute ? CP_LOAD_STATE6_FRAG : CP_LOAD_STATE6; tu_cs_emit_pkt7(cs, cp_load_state, 3 + num_units); tu_cs_emit(cs, CP_LOAD_STATE6_0_DST_OFF(offset) | CP_LOAD_STATE6_0_STATE_TYPE(st) | CP_LOAD_STATE6_0_STATE_SRC(SS6_DIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(SB6_UAV) | CP_LOAD_STATE6_0_NUM_UNIT(num_units)); tu_cs_emit(cs, 0); tu_cs_emit(cs, 0); for (unsigned i = 0; i < num_units; i++) tu_cs_emit(cs, push_constants[i + offset]); } } static void tu7_emit_shared_preamble_consts( struct tu_cs *cs, const struct tu_push_constant_range *shared_consts, uint32_t *push_constants) { tu_cs_emit_pkt4(cs, REG_A7XX_SP_SHARED_CONSTANT_GFX(shared_consts->lo_dwords), shared_consts->dwords); tu_cs_emit_array(cs, push_constants + shared_consts->lo_dwords, shared_consts->dwords); } static uint32_t tu6_const_size(struct tu_cmd_buffer *cmd, const struct tu_push_constant_range *shared_consts, bool compute) { uint32_t dwords = 0; if (shared_consts->type == IR3_PUSH_CONSTS_SHARED) { dwords += shared_consts->dwords + 4; } else if (shared_consts->type == IR3_PUSH_CONSTS_SHARED_PREAMBLE) { dwords += shared_consts->dwords + 1; } bool ldgk = cmd->device->physical_device->info->a7xx.load_inline_uniforms_via_preamble_ldgk; if (compute) { dwords += tu6_user_consts_size(&cmd->state.shaders[MESA_SHADER_COMPUTE]->const_state, ldgk, MESA_SHADER_COMPUTE); } else { for (uint32_t type = MESA_SHADER_VERTEX; type <= MESA_SHADER_FRAGMENT; type++) dwords += tu6_user_consts_size(&cmd->state.shaders[type]->const_state, ldgk, (mesa_shader_stage) type); } return dwords; } static struct tu_draw_state tu_emit_consts(struct tu_cmd_buffer *cmd, bool compute) { uint32_t dwords = 0; const struct tu_push_constant_range *shared_consts = compute ? &cmd->state.shaders[MESA_SHADER_COMPUTE]->const_state.push_consts : &cmd->state.program.shared_consts; dwords = tu6_const_size(cmd, shared_consts, compute); if (dwords == 0) return (struct tu_draw_state) {}; struct tu_cs cs; tu_cs_begin_sub_stream(&cmd->sub_cs, dwords, &cs); if (shared_consts->type == IR3_PUSH_CONSTS_SHARED) { tu6_emit_shared_consts(&cs, shared_consts, cmd->push_constants, compute); } else if (shared_consts->type == IR3_PUSH_CONSTS_SHARED_PREAMBLE) { tu7_emit_shared_preamble_consts(&cs, shared_consts, cmd->push_constants); } if (compute) { tu6_emit_per_stage_push_consts( &cs, &cmd->state.shaders[MESA_SHADER_COMPUTE]->const_state, cmd->state.shaders[MESA_SHADER_COMPUTE]->variant->const_state, MESA_SHADER_COMPUTE, cmd->push_constants); tu_emit_inline_ubo( &cs, &cmd->state.shaders[MESA_SHADER_COMPUTE]->const_state, cmd->state.shaders[MESA_SHADER_COMPUTE]->variant->const_state, cmd->state.shaders[MESA_SHADER_COMPUTE]->variant->constlen, MESA_SHADER_COMPUTE, tu_get_descriptors_state(cmd, VK_PIPELINE_BIND_POINT_COMPUTE)); } else { struct tu_descriptor_state *descriptors = tu_get_descriptors_state(cmd, VK_PIPELINE_BIND_POINT_GRAPHICS); for (uint32_t type = MESA_SHADER_VERTEX; type <= MESA_SHADER_FRAGMENT; type++) { const struct tu_program_descriptor_linkage *link = &cmd->state.program.link[type]; tu6_emit_per_stage_push_consts(&cs, &link->tu_const_state, &link->const_state, (mesa_shader_stage) type, cmd->push_constants); tu_emit_inline_ubo(&cs, &link->tu_const_state, &link->const_state, link->constlen, (mesa_shader_stage) type, descriptors); } } return tu_cs_end_draw_state(&cmd->sub_cs, &cs); } /* Returns true if stencil may be written when depth test fails. * This could be either from stencil written on depth test fail itself, * or stencil written on the stencil test failure where subsequent depth * test may also fail. */ static bool tu6_stencil_written_on_depth_fail( const struct vk_stencil_test_face_state *face) { switch (face->op.compare) { case VK_COMPARE_OP_ALWAYS: /* The stencil op always passes, no need to worry about failOp. */ return face->op.depth_fail != VK_STENCIL_OP_KEEP; case VK_COMPARE_OP_NEVER: /* The stencil op always fails, so failOp will always be used. */ return face->op.fail != VK_STENCIL_OP_KEEP; default: /* If the stencil test fails, depth may fail as well, so we can write * stencil when the depth fails if failOp is not VK_STENCIL_OP_KEEP. */ return face->op.fail != VK_STENCIL_OP_KEEP || face->op.depth_fail != VK_STENCIL_OP_KEEP; } } /* Returns true if the stencil write result may change based on the result of a * depth test. */ static bool tu6_stencil_written_based_on_depth_test( const struct vk_stencil_test_face_state *face) { switch (face->op.compare) { case VK_COMPARE_OP_ALWAYS: /* The stencil op always passes, no need to worry about failOp. */ return face->op.depth_fail != VK_STENCIL_OP_KEEP || face->op.pass != VK_STENCIL_OP_KEEP; case VK_COMPARE_OP_NEVER: /* The stencil op always fails, so failOp will always be used. */ return face->op.fail != VK_STENCIL_OP_KEEP; default: /* If the stencil test fails, depth may fail as well, so we can write * stencil when the depth fails if failOp is not VK_STENCIL_OP_KEEP. */ return face->op.fail != VK_STENCIL_OP_KEEP || face->op.pass != VK_STENCIL_OP_KEEP || face->op.depth_fail != VK_STENCIL_OP_KEEP; } } /* Various frontends (ANGLE, zink at least) will enable stencil testing with * what works out to be no-op writes. Simplify what they give us into flags * that LRZ can use. */ static void tu6_update_simplified_stencil_state(struct tu_cmd_buffer *cmd) { const struct vk_depth_stencil_state *ds = &cmd->vk.dynamic_graphics_state.ds; bool stencil_test_enable = ds->stencil.test_enable; if (!stencil_test_enable) { cmd->state.stencil_front_write = false; cmd->state.stencil_back_write = false; cmd->state.stencil_written_on_depth_fail = false; cmd->state.stencil_written_based_on_depth_test = false; return; } bool stencil_front_writemask = ds->stencil.front.write_mask; bool stencil_back_writemask = ds->stencil.back.write_mask; VkStencilOp front_fail_op = (VkStencilOp)ds->stencil.front.op.fail; VkStencilOp front_pass_op = (VkStencilOp)ds->stencil.front.op.pass; VkStencilOp front_depth_fail_op = (VkStencilOp)ds->stencil.front.op.depth_fail; VkStencilOp back_fail_op = (VkStencilOp)ds->stencil.back.op.fail; VkStencilOp back_pass_op = (VkStencilOp)ds->stencil.back.op.pass; VkStencilOp back_depth_fail_op = (VkStencilOp)ds->stencil.back.op.depth_fail; bool stencil_front_op_writes = front_pass_op != VK_STENCIL_OP_KEEP || front_fail_op != VK_STENCIL_OP_KEEP || front_depth_fail_op != VK_STENCIL_OP_KEEP; bool stencil_back_op_writes = back_pass_op != VK_STENCIL_OP_KEEP || back_fail_op != VK_STENCIL_OP_KEEP || back_depth_fail_op != VK_STENCIL_OP_KEEP; cmd->state.stencil_front_write = stencil_front_op_writes && stencil_front_writemask; cmd->state.stencil_back_write = stencil_back_op_writes && stencil_back_writemask; cmd->state.stencil_written_on_depth_fail = (cmd->state.stencil_front_write && tu6_stencil_written_on_depth_fail(&ds->stencil.front)) || (cmd->state.stencil_back_write && tu6_stencil_written_on_depth_fail(&ds->stencil.back)); cmd->state.stencil_written_based_on_depth_test = (cmd->state.stencil_front_write && tu6_stencil_written_based_on_depth_test(&ds->stencil.front)) || (cmd->state.stencil_back_write && tu6_stencil_written_based_on_depth_test(&ds->stencil.back)); } static bool tu6_writes_depth(struct tu_cmd_buffer *cmd, bool depth_test_enable) { bool depth_write_enable = cmd->vk.dynamic_graphics_state.ds.depth.write_enable; VkCompareOp depth_compare_op = (VkCompareOp) cmd->vk.dynamic_graphics_state.ds.depth.compare_op; bool depth_compare_op_writes = depth_compare_op != VK_COMPARE_OP_NEVER; return depth_test_enable && depth_write_enable && depth_compare_op_writes; } static bool tu6_writes_stencil(struct tu_cmd_buffer *cmd) { return cmd->state.stencil_front_write || cmd->state.stencil_back_write; } static bool tu_fs_reads_dynamic_ds_input_attachment(struct tu_cmd_buffer *cmd, const struct tu_shader *fs) { uint8_t depth_att = cmd->vk.dynamic_graphics_state.ial.depth_att; if (depth_att == MESA_VK_ATTACHMENT_UNUSED) return false; unsigned depth_idx = (depth_att == MESA_VK_ATTACHMENT_NO_INDEX) ? 0 : depth_att + 1; return fs->fs.dynamic_input_attachments_used & (1u << depth_idx); } static void tu6_build_depth_plane_z_mode(struct tu_cmd_buffer *cmd, struct tu_cs *cs) { enum a6xx_ztest_mode zmode = A6XX_EARLY_Z; bool depth_test_enable = cmd->vk.dynamic_graphics_state.ds.depth.test_enable; bool stencil_test_enable = cmd->vk.dynamic_graphics_state.ds.stencil.test_enable; bool ds_test_enable = depth_test_enable || stencil_test_enable; bool depth_write = tu6_writes_depth(cmd, depth_test_enable); bool stencil_write = tu6_writes_stencil(cmd); const struct tu_shader *fs = cmd->state.shaders[MESA_SHADER_FRAGMENT]; const struct tu_render_pass *pass = cmd->state.pass; const struct tu_subpass *subpass = cmd->state.subpass; VkFormat depth_format = VK_FORMAT_UNDEFINED; if (subpass->depth_stencil_attachment.attachment != VK_ATTACHMENT_UNUSED) depth_format = pass->attachments[subpass->depth_stencil_attachment.attachment].format; bool fs_kill_fragments = fs->variant->has_kill || /* EARLY_Z causes D/S to be written before FS but gl_SampleMask can * kill fragments, we cannot have EARLY_Z + gl_SampleMask + D/S writes. */ fs->variant->writes_smask || /* Alpha-to-coverage behaves like a discard. */ cmd->vk.dynamic_graphics_state.ms.alpha_to_coverage_enable; if ((fs_kill_fragments || (cmd->state.pipeline_feedback_loops & VK_IMAGE_ASPECT_DEPTH_BIT) || (cmd->vk.dynamic_graphics_state.feedback_loops & VK_IMAGE_ASPECT_DEPTH_BIT) || tu_fs_reads_dynamic_ds_input_attachment(cmd, fs)) && (depth_write || stencil_write)) { zmode = A6XX_EARLY_Z_LATE_Z; } /* If there is explicit depth direction in FS writing gl_FragDepth * may be compatible with LRZ test. */ if (cmd->state.lrz.enabled && fs->variant->writes_pos && zmode == A6XX_EARLY_Z) { zmode = A6XX_EARLY_Z_LATE_Z; } /* "EARLY_Z + discard" would yield incorrect occlusion query result, * since Vulkan expects occlusion query to happen after fragment shader. */ if (zmode == A6XX_EARLY_Z && fs_kill_fragments && cmd->state.occlusion_query_may_be_running) zmode = A6XX_EARLY_Z_LATE_Z; VkCompareOp compare_op = cmd->vk.dynamic_graphics_state.ds.depth.compare_op; /* This state combination wedges something in GPU causing hang. * Forcing A6XX_LATE_Z prevents it. Prop driver does the same. */ if (zmode == A6XX_EARLY_Z_LATE_Z && !depth_write && cmd->state.occlusion_query_may_be_running && (compare_op == VK_COMPARE_OP_ALWAYS || compare_op == VK_COMPARE_OP_NEVER)) { zmode = A6XX_LATE_Z; } if (zmode == A6XX_EARLY_Z_LATE_Z && (cmd->state.stencil_written_on_depth_fail || fs->fs.sample_shading || !vk_format_has_depth(depth_format) || !ds_test_enable)) { zmode = A6XX_LATE_Z; } if ((stencil_test_enable && depth_format == VK_FORMAT_S8_UINT) || (ds_test_enable && (fs->fs.lrz.force_late_z || cmd->state.lrz.force_late_z))) zmode = A6XX_LATE_Z; /* User defined early tests take precedence above all else */ if (fs->variant->fs.early_fragment_tests) zmode = A6XX_EARLY_Z; /* FS bypass requires early Z */ if (cmd->state.disable_fs) zmode = A6XX_EARLY_Z; tu_cs_emit_pkt4(cs, REG_A6XX_GRAS_SU_DEPTH_PLANE_CNTL, 1); tu_cs_emit(cs, A6XX_GRAS_SU_DEPTH_PLANE_CNTL_Z_MODE(zmode)); tu_cs_emit_pkt4(cs, REG_A6XX_RB_DEPTH_PLANE_CNTL, 1); tu_cs_emit(cs, A6XX_RB_DEPTH_PLANE_CNTL_Z_MODE(zmode)); } static uint32_t fs_params_offset(struct tu_cmd_buffer *cmd) { const struct tu_program_descriptor_linkage *link = &cmd->state.program.link[MESA_SHADER_FRAGMENT]; const struct ir3_const_state *const_state = &link->const_state; if (const_state->num_driver_params <= IR3_DP_FS_DYNAMIC) return 0; uint32_t param_offset = const_state->allocs.consts[IR3_CONST_ALLOC_DRIVER_PARAMS].offset_vec4; if (param_offset + IR3_DP_FS_DYNAMIC / 4 >= link->constlen) return 0; return param_offset + IR3_DP_FS_DYNAMIC / 4; } static uint32_t fs_params_size(struct tu_cmd_buffer *cmd) { const struct tu_program_descriptor_linkage *link = &cmd->state.program.link[MESA_SHADER_FRAGMENT]; const struct ir3_const_state *const_state = &link->const_state; return DIV_ROUND_UP(const_state->num_driver_params - IR3_DP_FS_DYNAMIC, 4); } struct apply_fs_params_state { unsigned num_consts; }; static void fdm_apply_fs_params(struct tu_cmd_buffer *cmd, struct tu_cs *cs, void *data, VkOffset2D common_bin_offset, const VkOffset2D *hw_viewport_offsets, unsigned views, const VkExtent2D *frag_areas, const VkRect2D *bins) { const struct apply_fs_params_state *state = (const struct apply_fs_params_state *)data; unsigned num_consts = state->num_consts; for (unsigned i = 0; i < num_consts; i++) { /* FDM per layer may be enabled in the shader but not in the renderpass, * in which case views will be 1 and we have to replicate the one view * to all of the layers. */ VkExtent2D area = frag_areas[MIN2(i, views - 1)]; VkRect2D bin = bins[MIN2(i, views - 1)]; VkOffset2D hw_viewport_offset = hw_viewport_offsets[MIN2(i, views - 1)]; VkOffset2D offset = tu_fdm_per_bin_offset(area, bin, common_bin_offset); offset.x -= hw_viewport_offset.x; offset.y -= hw_viewport_offset.y; tu_cs_emit(cs, area.width); tu_cs_emit(cs, area.height); tu_cs_emit(cs, fui(offset.x)); tu_cs_emit(cs, fui(offset.y)); } } static void tu_emit_fdm_params(struct tu_cmd_buffer *cmd, struct tu_cs *cs, struct tu_shader *fs, unsigned num_units) { STATIC_ASSERT(IR3_DP_FS(frag_invocation_count) == IR3_DP_FS_DYNAMIC); tu_cs_emit(cs, fs->fs.sample_shading ? cmd->vk.dynamic_graphics_state.ms.rasterization_samples : 1); tu_cs_emit(cs, 0); tu_cs_emit(cs, 0); tu_cs_emit(cs, 0); STATIC_ASSERT(IR3_DP_FS(frag_size) == IR3_DP_FS_DYNAMIC + 4); STATIC_ASSERT(IR3_DP_FS(frag_offset) == IR3_DP_FS_DYNAMIC + 6); if (num_units > 1) { if (fs->fs.has_fdm) { struct apply_fs_params_state state = { .num_consts = num_units - 1, }; tu_create_fdm_bin_patchpoint(cmd, cs, 4 * (num_units - 1), TU_FDM_SKIP_BINNING, fdm_apply_fs_params, state); } else { for (unsigned i = 1; i < num_units; i++) { tu_cs_emit(cs, 1); tu_cs_emit(cs, 1); tu_cs_emit(cs, fui(0.0f)); tu_cs_emit(cs, fui(0.0f)); } } } } static void tu6_emit_fs_params(struct tu_cmd_buffer *cmd) { uint32_t offset = fs_params_offset(cmd); if (offset == 0) { cmd->state.fs_params = (struct tu_draw_state) {}; return; } struct tu_shader *fs = cmd->state.shaders[MESA_SHADER_FRAGMENT]; unsigned num_units = fs_params_size(cmd); if (fs->fs.has_fdm) tu_cs_set_writeable(&cmd->sub_cs, true); struct tu_cs cs; VkResult result = tu_cs_begin_sub_stream(&cmd->sub_cs, 4 + 4 * num_units, &cs); if (result != VK_SUCCESS) { tu_cs_set_writeable(&cmd->sub_cs, false); vk_command_buffer_set_error(&cmd->vk, result); return; } tu_cs_emit_pkt7(&cs, CP_LOAD_STATE6_FRAG, 3 + 4 * num_units); tu_cs_emit(&cs, CP_LOAD_STATE6_0_DST_OFF(offset) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_CONSTANTS) | CP_LOAD_STATE6_0_STATE_SRC(SS6_DIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(SB6_FS_SHADER) | CP_LOAD_STATE6_0_NUM_UNIT(num_units)); tu_cs_emit(&cs, 0); tu_cs_emit(&cs, 0); tu_emit_fdm_params(cmd, &cs, fs, num_units); cmd->state.fs_params = tu_cs_end_draw_state(&cmd->sub_cs, &cs); if (fs->fs.has_fdm) tu_cs_set_writeable(&cmd->sub_cs, false); } static void tu7_emit_fs_params(struct tu_cmd_buffer *cmd) { struct tu_shader *fs = cmd->state.shaders[MESA_SHADER_FRAGMENT]; int ubo_offset = fs->const_state.fdm_ubo.idx; if (ubo_offset < 0) { cmd->state.fs_params = (struct tu_draw_state) {}; return; } unsigned num_units = DIV_ROUND_UP(fs->const_state.fdm_ubo.size, 4); if (fs->fs.has_fdm) tu_cs_set_writeable(&cmd->sub_cs, true); struct tu_cs cs; VkResult result = tu_cs_begin_sub_stream_aligned(&cmd->sub_cs, num_units, 4, &cs); if (result != VK_SUCCESS) { tu_cs_set_writeable(&cmd->sub_cs, false); vk_command_buffer_set_error(&cmd->vk, result); return; } tu_emit_fdm_params(cmd, &cs, fs, num_units); struct tu_draw_state fdm_ubo = tu_cs_end_draw_state(&cmd->sub_cs, &cs); if (fs->fs.has_fdm) tu_cs_set_writeable(&cmd->sub_cs, false); result = tu_cs_begin_sub_stream(&cmd->sub_cs, 6, &cs); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } tu_cs_emit_pkt7(&cs, CP_LOAD_STATE6_FRAG, 5); tu_cs_emit(&cs, CP_LOAD_STATE6_0_DST_OFF(ubo_offset) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_UBO)| CP_LOAD_STATE6_0_STATE_SRC(SS6_DIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(SB6_FS_SHADER) | CP_LOAD_STATE6_0_NUM_UNIT(1)); tu_cs_emit(&cs, CP_LOAD_STATE6_1_EXT_SRC_ADDR(0)); tu_cs_emit(&cs, CP_LOAD_STATE6_2_EXT_SRC_ADDR_HI(0)); tu_cs_emit_qw(&cs, fdm_ubo.iova | (uint64_t)A6XX_UBO_1_SIZE(num_units) << 32); cmd->state.fs_params = tu_cs_end_draw_state(&cmd->sub_cs, &cs); } static void tu_emit_fs_params(struct tu_cmd_buffer *cmd) { if (cmd->device->compiler->load_shader_consts_via_preamble) tu7_emit_fs_params(cmd); else tu6_emit_fs_params(cmd); } static void tu_flush_dynamic_input_attachments(struct tu_cmd_buffer *cmd) { struct tu_shader *fs = cmd->state.shaders[MESA_SHADER_FRAGMENT]; if (!fs->fs.dynamic_input_attachments_used) return; /* Input attachments may read data from a load op, so we have to invalidate * UCHE and force pending blits to complete unless we know it's already * been invalidated. This is the same as tu_subpass::feedback_invalidate * but for dynamic renderpasses. */ if (!cmd->state.blit_cache_cleaned) { cmd->state.renderpass_cache.flush_bits |= TU_CMD_FLAG_CACHE_INVALIDATE | TU_CMD_FLAG_BLIT_CACHE_CLEAN | TU_CMD_FLAG_WAIT_FOR_IDLE; } } template static VkResult tu6_draw_common(struct tu_cmd_buffer *cmd, struct tu_cs *cs, bool indexed, /* note: draw_count is 0 for indirect */ uint32_t draw_count) { const struct tu_program_state *program = &cmd->state.program; struct tu_render_pass_state *rp = &cmd->state.rp; trace_start_draw( &cmd->trace, &cmd->draw_cs, cmd, draw_count, cmd->state.program.stage_sha1[MESA_SHADER_VERTEX], cmd->state.program.stage_sha1[MESA_SHADER_TESS_CTRL], cmd->state.program.stage_sha1[MESA_SHADER_TESS_EVAL], cmd->state.program.stage_sha1[MESA_SHADER_GEOMETRY], cmd->state.program.stage_sha1[MESA_SHADER_FRAGMENT]); /* Emit state first, because it's needed for bandwidth calculations */ uint32_t dynamic_draw_state_dirty = 0; if (!BITSET_IS_EMPTY(cmd->vk.dynamic_graphics_state.dirty) || (cmd->state.dirty & ~TU_CMD_DIRTY_COMPUTE_DESC_SETS)) { dynamic_draw_state_dirty = tu_emit_draw_state(cmd); } /* Primitive restart value works in non-indexed draws, we have to disable * prim restart for such draws since we may read stale restart index. */ if (cmd->state.last_draw_indexed != indexed) { cmd->state.last_draw_indexed = indexed; BITSET_SET(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_IA_PRIMITIVE_RESTART_ENABLE); } /* Fill draw stats for autotuner */ rp->drawcall_count++; rp->drawcall_bandwidth_per_sample_sum += cmd->state.bandwidth.color_bandwidth_per_sample; /* add depth memory bandwidth cost */ const uint32_t depth_bandwidth = cmd->state.bandwidth.depth_cpp_per_sample; if (cmd->vk.dynamic_graphics_state.ds.depth.write_enable) rp->drawcall_bandwidth_per_sample_sum += depth_bandwidth; if (cmd->vk.dynamic_graphics_state.ds.depth.test_enable) rp->drawcall_bandwidth_per_sample_sum += depth_bandwidth; /* add stencil memory bandwidth cost */ const uint32_t stencil_bandwidth = cmd->state.bandwidth.stencil_cpp_per_sample; if (cmd->vk.dynamic_graphics_state.ds.stencil.test_enable) rp->drawcall_bandwidth_per_sample_sum += stencil_bandwidth * 2; if (cmd->state.dirty & TU_CMD_DIRTY_FS) tu_flush_dynamic_input_attachments(cmd); tu_emit_cache_flush_renderpass(cmd); if (BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_IA_PRIMITIVE_RESTART_ENABLE) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_RS_PROVOKING_VERTEX) || (cmd->state.dirty & TU_CMD_DIRTY_DRAW_STATE)) { bool primitive_restart_enabled = cmd->vk.dynamic_graphics_state.ia.primitive_restart_enable; bool primitive_restart = primitive_restart_enabled && indexed; bool provoking_vtx_last = cmd->vk.dynamic_graphics_state.rs.provoking_vertex == VK_PROVOKING_VERTEX_MODE_LAST_VERTEX_EXT; uint32_t primitive_cntl_0 = A6XX_PC_CNTL(.primitive_restart = primitive_restart, .provoking_vtx_last = provoking_vtx_last).value; tu_cs_emit_regs(cs, A6XX_PC_CNTL(.dword = primitive_cntl_0)); if (CHIP == A7XX) { tu_cs_emit_regs(cs, A7XX_VPC_PC_CNTL(.dword = primitive_cntl_0)); } } struct tu_tess_params *tess_params = &cmd->state.tess_params; if ((cmd->state.dirty & TU_CMD_DIRTY_TESS_PARAMS) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_TS_DOMAIN_ORIGIN) || (cmd->state.dirty & TU_CMD_DIRTY_DRAW_STATE)) { bool tess_upper_left_domain_origin = (VkTessellationDomainOrigin)cmd->vk.dynamic_graphics_state.ts.domain_origin == VK_TESSELLATION_DOMAIN_ORIGIN_UPPER_LEFT; tu_cs_emit_regs(cs, A6XX_PC_DS_PARAM( .spacing = tess_params->spacing, .output = tess_upper_left_domain_origin ? tess_params->output_upper_left : tess_params->output_lower_left)); } if (cmd->device->physical_device->info->a7xx.has_rt_workaround && cmd->state.program.uses_ray_intersection) { tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_SHADER_USES_RT); } /* Early exit if there is nothing to emit, saves CPU cycles */ uint32_t dirty = cmd->state.dirty; if (!dynamic_draw_state_dirty && !(dirty & ~TU_CMD_DIRTY_COMPUTE_DESC_SETS)) return VK_SUCCESS; bool dirty_lrz = (dirty & TU_CMD_DIRTY_LRZ) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_DS_DEPTH_TEST_ENABLE) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_DS_DEPTH_WRITE_ENABLE) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_DS_DEPTH_BOUNDS_TEST_ENABLE) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_DS_DEPTH_COMPARE_OP) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_DS_STENCIL_TEST_ENABLE) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_DS_STENCIL_OP) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_DS_STENCIL_WRITE_MASK) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_MS_ALPHA_TO_COVERAGE_ENABLE) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_ATTACHMENT_FEEDBACK_LOOP_ENABLE); if (dirty_lrz) { struct tu_cs cs; uint32_t size = 8 + (cmd->device->physical_device->info->a6xx.lrz_track_quirk ? 2 : 0) + (CHIP >= A7XX ? 2 : 0); // A7XX has extra packets from LRZ_CNTL2. cmd->state.lrz_and_depth_plane_state = tu_cs_draw_state(&cmd->sub_cs, &cs, size); tu6_update_simplified_stencil_state(cmd); tu6_emit_lrz(cmd, &cs); tu6_build_depth_plane_z_mode(cmd, &cs); } if (BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_ATTACHMENT_FEEDBACK_LOOP_ENABLE)) { if (cmd->vk.dynamic_graphics_state.feedback_loops && !cmd->state.rp.disable_gmem) { perf_debug( cmd->device, "Disabling gmem due to VK_EXT_attachment_feedback_loop_layout"); cmd->state.rp.disable_gmem = true; cmd->state.rp.gmem_disable_reason = "MESA_VK_DYNAMIC_ATTACHMENT_FEEDBACK_LOOP_ENABLE"; } } if (BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_VI_BINDINGS_VALID)) { cmd->state.vertex_buffers.size = util_last_bit(cmd->vk.dynamic_graphics_state.vi_bindings_valid) * 4; dirty |= TU_CMD_DIRTY_VERTEX_BUFFERS; } if (dirty & TU_CMD_DIRTY_SHADER_CONSTS) cmd->state.shader_const = tu_emit_consts(cmd, false); if (dirty & TU_CMD_DIRTY_DESC_SETS) tu6_emit_descriptor_sets(cmd, VK_PIPELINE_BIND_POINT_GRAPHICS); if (BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_MS_RASTERIZATION_SAMPLES) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_IA_PRIMITIVE_TOPOLOGY) || BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_RS_LINE_MODE) || (cmd->state.dirty & TU_CMD_DIRTY_TES) || (cmd->state.dirty & TU_CMD_DIRTY_DRAW_STATE)) { tu6_update_msaa_disable(cmd); } if (BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_MS_RASTERIZATION_SAMPLES) || (cmd->state.dirty & TU_CMD_DIRTY_DRAW_STATE)) { tu6_update_msaa(cmd); } bool dirty_fs_params = false; if (BITSET_TEST(cmd->vk.dynamic_graphics_state.dirty, MESA_VK_DYNAMIC_MS_RASTERIZATION_SAMPLES) || (cmd->state.dirty & (TU_CMD_DIRTY_PROGRAM | TU_CMD_DIRTY_FDM))) { tu_emit_fs_params(cmd); dirty_fs_params = true; } /* for the first draw in a renderpass, re-emit all the draw states * * and if a draw-state disabling path (CmdClearAttachments 3D fallback) was * used, then draw states must be re-emitted. note however this only happens * in the sysmem path, so this can be skipped this for the gmem path (TODO) * * the two input attachment states are excluded because secondary command * buffer doesn't have a state ib to restore it, and not re-emitting them * is OK since CmdClearAttachments won't disable/overwrite them */ if (dirty & TU_CMD_DIRTY_DRAW_STATE) { tu_pipeline_update_rp_state(&cmd->state); tu_cs_emit_pkt7(cs, CP_SET_DRAW_STATE, 3 * (TU_DRAW_STATE_COUNT - 2)); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_PROGRAM_CONFIG, program->config_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VS, program->vs_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VS_BINNING, program->vs_binning_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_HS, program->hs_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_DS, program->ds_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_GS, program->gs_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_GS_BINNING, program->gs_binning_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_FS, program->fs_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VPC, program->vpc_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_PRIM_MODE_GMEM, cmd->state.prim_order_gmem); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_CONST, cmd->state.shader_const); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_DESC_SETS, cmd->state.desc_sets); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_DESC_SETS_LOAD, cmd->state.load_state); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VB, cmd->state.vertex_buffers); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VS_PARAMS, cmd->state.vs_params); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_FS_PARAMS, cmd->state.fs_params); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_LRZ_AND_DEPTH_PLANE, cmd->state.lrz_and_depth_plane_state); for (uint32_t i = 0; i < ARRAY_SIZE(cmd->state.dynamic_state); i++) { tu_cs_emit_draw_state(cs, TU_DRAW_STATE_DYNAMIC + i, cmd->state.dynamic_state[i]); } } else { /* emit draw states that were just updated */ uint32_t draw_state_count = util_bitcount(dynamic_draw_state_dirty) + ((dirty & TU_CMD_DIRTY_SHADER_CONSTS) ? 1 : 0) + ((dirty & TU_CMD_DIRTY_DESC_SETS) ? 1 : 0) + ((dirty & TU_CMD_DIRTY_VERTEX_BUFFERS) ? 1 : 0) + ((dirty & TU_CMD_DIRTY_VS_PARAMS) ? 1 : 0) + (dirty_fs_params ? 1 : 0) + (dirty_lrz ? 1 : 0); if (draw_state_count > 0) tu_cs_emit_pkt7(cs, CP_SET_DRAW_STATE, 3 * draw_state_count); if (dirty & TU_CMD_DIRTY_SHADER_CONSTS) tu_cs_emit_draw_state(cs, TU_DRAW_STATE_CONST, cmd->state.shader_const); if (dirty & TU_CMD_DIRTY_DESC_SETS) { /* tu6_emit_descriptor_sets emitted the cmd->state.desc_sets draw state. */ tu_cs_emit_draw_state(cs, TU_DRAW_STATE_DESC_SETS_LOAD, cmd->state.load_state); } if (dirty & TU_CMD_DIRTY_VERTEX_BUFFERS) tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VB, cmd->state.vertex_buffers); u_foreach_bit (i, dynamic_draw_state_dirty) { tu_cs_emit_draw_state(cs, TU_DRAW_STATE_DYNAMIC + i, cmd->state.dynamic_state[i]); } if (dirty & TU_CMD_DIRTY_VS_PARAMS) tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VS_PARAMS, cmd->state.vs_params); if (dirty_fs_params) tu_cs_emit_draw_state(cs, TU_DRAW_STATE_FS_PARAMS, cmd->state.fs_params); if (dirty_lrz) { tu_cs_emit_draw_state(cs, TU_DRAW_STATE_LRZ_AND_DEPTH_PLANE, cmd->state.lrz_and_depth_plane_state); } } tu_cs_sanity_check(cs); /* There are too many graphics dirty bits to list here, so just list the * bits to preserve instead. The only things not emitted here are * compute-related state. */ cmd->state.dirty &= TU_CMD_DIRTY_COMPUTE_DESC_SETS; BITSET_ZERO(cmd->vk.dynamic_graphics_state.dirty); return VK_SUCCESS; } static uint32_t tu_draw_initiator(struct tu_cmd_buffer *cmd, enum pc_di_src_sel src_sel) { enum pc_di_primtype primtype = tu6_primtype((VkPrimitiveTopology)cmd->vk.dynamic_graphics_state.ia.primitive_topology); if (primtype == DI_PT_PATCHES0) primtype = (enum pc_di_primtype) (primtype + cmd->vk.dynamic_graphics_state.ts.patch_control_points); uint32_t initiator = CP_DRAW_INDX_OFFSET_0_PRIM_TYPE(primtype) | CP_DRAW_INDX_OFFSET_0_SOURCE_SELECT(src_sel) | CP_DRAW_INDX_OFFSET_0_INDEX_SIZE((enum a4xx_index_size) cmd->state.index_size) | CP_DRAW_INDX_OFFSET_0_VIS_CULL(USE_VISIBILITY); if (cmd->state.shaders[MESA_SHADER_GEOMETRY]->variant) initiator |= CP_DRAW_INDX_OFFSET_0_GS_ENABLE; const struct tu_shader *tes = cmd->state.shaders[MESA_SHADER_TESS_EVAL]; if (tes->variant) { switch (tes->variant->key.tessellation) { case IR3_TESS_TRIANGLES: initiator |= CP_DRAW_INDX_OFFSET_0_PATCH_TYPE(TESS_TRIANGLES) | CP_DRAW_INDX_OFFSET_0_TESS_ENABLE; break; case IR3_TESS_ISOLINES: initiator |= CP_DRAW_INDX_OFFSET_0_PATCH_TYPE(TESS_ISOLINES) | CP_DRAW_INDX_OFFSET_0_TESS_ENABLE; break; case IR3_TESS_QUADS: initiator |= CP_DRAW_INDX_OFFSET_0_PATCH_TYPE(TESS_QUADS) | CP_DRAW_INDX_OFFSET_0_TESS_ENABLE; break; } } return initiator; } static uint32_t vs_params_offset(struct tu_cmd_buffer *cmd) { const struct tu_program_descriptor_linkage *link = &cmd->state.program.link[MESA_SHADER_VERTEX]; const struct ir3_const_state *const_state = &link->const_state; uint32_t param_offset = const_state->allocs.consts[IR3_CONST_ALLOC_DRIVER_PARAMS].offset_vec4; if (!ir3_const_can_upload(&const_state->allocs, IR3_CONST_ALLOC_DRIVER_PARAMS, link->constlen)) return 0; /* this layout is required by CP_DRAW_INDIRECT_MULTI */ STATIC_ASSERT(IR3_DP_VS(draw_id) == 0); STATIC_ASSERT(IR3_DP_VS(vtxid_base) == 1); STATIC_ASSERT(IR3_DP_VS(instid_base) == 2); /* 0 means disabled for CP_DRAW_INDIRECT_MULTI */ assert(param_offset != 0); return param_offset; } template static void tu6_emit_empty_vs_params(struct tu_cmd_buffer *cmd) { if (cmd->state.last_vs_params.empty) return; if (cmd->device->physical_device->info->a7xx.load_shader_consts_via_preamble) { struct tu_cs cs; cmd->state.vs_params = tu_cs_draw_state(&cmd->sub_cs, &cs, 2); /* CP_LOAD_STATE6_GEOM from previous draws can override consts loaded for * indirect draws, causing problems like incorrect vertex index computation. * VS state invalidation avoids that. */ tu_cs_emit_regs(&cs, SP_UPDATE_CNTL(CHIP, .vs_state = true)); assert(cs.cur == cs.end); } else { cmd->state.vs_params = (struct tu_draw_state) {}; } cmd->state.dirty |= TU_CMD_DIRTY_VS_PARAMS; cmd->state.last_vs_params.empty = true; } static void tu6_emit_vs_params(struct tu_cmd_buffer *cmd, uint32_t draw_id, uint32_t vertex_offset, uint32_t first_instance) { uint32_t offset = vs_params_offset(cmd); /* Beside re-emitting params when they are changed, we should re-emit * them after constants are invalidated via SP_UPDATE_CNTL or after we * emit an empty vs params. */ if (!(cmd->state.dirty & (TU_CMD_DIRTY_DRAW_STATE | TU_CMD_DIRTY_VS_PARAMS | TU_CMD_DIRTY_PROGRAM)) && !cmd->state.last_vs_params.empty && (offset == 0 || draw_id == cmd->state.last_vs_params.draw_id) && vertex_offset == cmd->state.last_vs_params.vertex_offset && first_instance == cmd->state.last_vs_params.first_instance) { return; } uint64_t consts_iova = 0; if (offset) { struct tu_cs_memory consts; VkResult result = tu_cs_alloc(&cmd->sub_cs, 1, 4, &consts); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } consts.map[0] = draw_id; consts.map[1] = vertex_offset; consts.map[2] = first_instance; consts.map[3] = 0; consts_iova = consts.iova; } struct tu_cs cs; VkResult result = tu_cs_begin_sub_stream(&cmd->sub_cs, 3 + (offset ? 4 : 0), &cs); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } tu_cs_emit_regs(&cs, A6XX_VFD_INDEX_OFFSET(vertex_offset), A6XX_VFD_INSTANCE_START_OFFSET(first_instance)); /* It is implemented as INDIRECT load even on a750+ because with UBO * lowering it would be tricky to get const offset for to use in multidraw, * also we would need to ensure the offset is not 0. * TODO/A7XX: Rework vs params to use UBO lowering. */ if (offset) { tu_cs_emit_pkt7(&cs, CP_LOAD_STATE6_GEOM, 3); tu_cs_emit(&cs, CP_LOAD_STATE6_0_DST_OFF(offset) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_CONSTANTS) | CP_LOAD_STATE6_0_STATE_SRC(SS6_INDIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(SB6_VS_SHADER) | CP_LOAD_STATE6_0_NUM_UNIT(1)); tu_cs_emit_qw(&cs, consts_iova); } cmd->state.last_vs_params.vertex_offset = vertex_offset; cmd->state.last_vs_params.first_instance = first_instance; cmd->state.last_vs_params.draw_id = draw_id; cmd->state.last_vs_params.empty = false; struct tu_cs_entry entry = tu_cs_end_sub_stream(&cmd->sub_cs, &cs); cmd->state.vs_params = (struct tu_draw_state) {entry.bo->iova + entry.offset, entry.size / 4}; cmd->state.dirty |= TU_CMD_DIRTY_VS_PARAMS; } template VKAPI_ATTR void VKAPI_CALL tu_CmdDraw(VkCommandBuffer commandBuffer, uint32_t vertexCount, uint32_t instanceCount, uint32_t firstVertex, uint32_t firstInstance) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); struct tu_cs *cs = &cmd->draw_cs; tu6_emit_vs_params(cmd, 0, firstVertex, firstInstance); tu6_draw_common(cmd, cs, false, vertexCount); tu_cs_emit_pkt7(cs, CP_DRAW_INDX_OFFSET, 3); tu_cs_emit(cs, tu_draw_initiator(cmd, DI_SRC_SEL_AUTO_INDEX)); tu_cs_emit(cs, instanceCount); tu_cs_emit(cs, vertexCount); trace_end_draw(&cmd->trace, cs); } TU_GENX(tu_CmdDraw); template VKAPI_ATTR void VKAPI_CALL tu_CmdDrawMultiEXT(VkCommandBuffer commandBuffer, uint32_t drawCount, const VkMultiDrawInfoEXT *pVertexInfo, uint32_t instanceCount, uint32_t firstInstance, uint32_t stride) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); struct tu_cs *cs = &cmd->draw_cs; if (!drawCount) return; bool has_tess = cmd->state.shaders[MESA_SHADER_TESS_CTRL]->variant; uint32_t max_vertex_count = 0; if (has_tess) { uint32_t i = 0; vk_foreach_multi_draw(draw, i, pVertexInfo, drawCount, stride) { max_vertex_count = MAX2(max_vertex_count, draw->vertexCount); } } uint32_t i = 0; vk_foreach_multi_draw(draw, i, pVertexInfo, drawCount, stride) { tu6_emit_vs_params(cmd, i, draw->firstVertex, firstInstance); if (i == 0) tu6_draw_common(cmd, cs, false, max_vertex_count); if (cmd->state.dirty & TU_CMD_DIRTY_VS_PARAMS) { tu_cs_emit_pkt7(cs, CP_SET_DRAW_STATE, 3); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VS_PARAMS, cmd->state.vs_params); cmd->state.dirty &= ~TU_CMD_DIRTY_VS_PARAMS; } tu_cs_emit_pkt7(cs, CP_DRAW_INDX_OFFSET, 3); tu_cs_emit(cs, tu_draw_initiator(cmd, DI_SRC_SEL_AUTO_INDEX)); tu_cs_emit(cs, instanceCount); tu_cs_emit(cs, draw->vertexCount); } if (i != 0) trace_end_draw(&cmd->trace, cs); } TU_GENX(tu_CmdDrawMultiEXT); template VKAPI_ATTR void VKAPI_CALL tu_CmdDrawIndexed(VkCommandBuffer commandBuffer, uint32_t indexCount, uint32_t instanceCount, uint32_t firstIndex, int32_t vertexOffset, uint32_t firstInstance) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); struct tu_cs *cs = &cmd->draw_cs; tu6_emit_vs_params(cmd, 0, vertexOffset, firstInstance); tu6_draw_common(cmd, cs, true, indexCount); tu_cs_emit_pkt7(cs, CP_DRAW_INDX_OFFSET, 7); tu_cs_emit(cs, tu_draw_initiator(cmd, DI_SRC_SEL_DMA)); tu_cs_emit(cs, instanceCount); tu_cs_emit(cs, indexCount); tu_cs_emit(cs, firstIndex); tu_cs_emit_qw(cs, cmd->state.index_va); tu_cs_emit(cs, cmd->state.max_index_count); trace_end_draw(&cmd->trace, cs); } TU_GENX(tu_CmdDrawIndexed); template VKAPI_ATTR void VKAPI_CALL tu_CmdDrawMultiIndexedEXT(VkCommandBuffer commandBuffer, uint32_t drawCount, const VkMultiDrawIndexedInfoEXT *pIndexInfo, uint32_t instanceCount, uint32_t firstInstance, uint32_t stride, const int32_t *pVertexOffset) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); struct tu_cs *cs = &cmd->draw_cs; if (!drawCount) return; bool has_tess = cmd->state.shaders[MESA_SHADER_TESS_CTRL]->variant; uint32_t max_index_count = 0; if (has_tess) { uint32_t i = 0; vk_foreach_multi_draw_indexed(draw, i, pIndexInfo, drawCount, stride) { max_index_count = MAX2(max_index_count, draw->indexCount); } } uint32_t i = 0; vk_foreach_multi_draw_indexed(draw, i, pIndexInfo, drawCount, stride) { int32_t vertexOffset = pVertexOffset ? *pVertexOffset : draw->vertexOffset; tu6_emit_vs_params(cmd, i, vertexOffset, firstInstance); if (i == 0) tu6_draw_common(cmd, cs, true, max_index_count); if (cmd->state.dirty & TU_CMD_DIRTY_VS_PARAMS) { tu_cs_emit_pkt7(cs, CP_SET_DRAW_STATE, 3); tu_cs_emit_draw_state(cs, TU_DRAW_STATE_VS_PARAMS, cmd->state.vs_params); cmd->state.dirty &= ~TU_CMD_DIRTY_VS_PARAMS; } tu_cs_emit_pkt7(cs, CP_DRAW_INDX_OFFSET, 7); tu_cs_emit(cs, tu_draw_initiator(cmd, DI_SRC_SEL_DMA)); tu_cs_emit(cs, instanceCount); tu_cs_emit(cs, draw->indexCount); tu_cs_emit(cs, draw->firstIndex); tu_cs_emit_qw(cs, cmd->state.index_va); tu_cs_emit(cs, cmd->state.max_index_count); } if (i != 0) trace_end_draw(&cmd->trace, cs); } TU_GENX(tu_CmdDrawMultiIndexedEXT); /* Various firmware bugs/inconsistencies mean that some indirect draw opcodes * do not wait for WFI's to complete before executing. Add a WAIT_FOR_ME if * pending for these opcodes. This may result in a few extra WAIT_FOR_ME's * with these opcodes, but the alternative would add unnecessary WAIT_FOR_ME's * before draw opcodes that don't need it. */ static void draw_wfm(struct tu_cmd_buffer *cmd) { cmd->state.renderpass_cache.flush_bits |= cmd->state.renderpass_cache.pending_flush_bits & TU_CMD_FLAG_WAIT_FOR_ME; cmd->state.renderpass_cache.pending_flush_bits &= ~TU_CMD_FLAG_WAIT_FOR_ME; } template VKAPI_ATTR void VKAPI_CALL tu_CmdDrawIndirect(VkCommandBuffer commandBuffer, VkBuffer _buffer, VkDeviceSize offset, uint32_t drawCount, uint32_t stride) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); VK_FROM_HANDLE(tu_buffer, buf, _buffer); struct tu_cs *cs = &cmd->draw_cs; tu6_emit_empty_vs_params(cmd); if (cmd->device->physical_device->info->a6xx.indirect_draw_wfm_quirk) draw_wfm(cmd); tu6_draw_common(cmd, cs, false, 0); tu_cs_emit_pkt7(cs, CP_DRAW_INDIRECT_MULTI, 6); tu_cs_emit(cs, tu_draw_initiator(cmd, DI_SRC_SEL_AUTO_INDEX)); tu_cs_emit(cs, A6XX_CP_DRAW_INDIRECT_MULTI_1_OPCODE(INDIRECT_OP_NORMAL) | A6XX_CP_DRAW_INDIRECT_MULTI_1_DST_OFF(vs_params_offset(cmd))); tu_cs_emit(cs, drawCount); tu_cs_emit_qw(cs, vk_buffer_address(&buf->vk, offset)); tu_cs_emit(cs, stride); trace_end_draw(&cmd->trace, cs); } TU_GENX(tu_CmdDrawIndirect); template VKAPI_ATTR void VKAPI_CALL tu_CmdDrawIndexedIndirect(VkCommandBuffer commandBuffer, VkBuffer _buffer, VkDeviceSize offset, uint32_t drawCount, uint32_t stride) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); VK_FROM_HANDLE(tu_buffer, buf, _buffer); struct tu_cs *cs = &cmd->draw_cs; tu6_emit_empty_vs_params(cmd); if (cmd->device->physical_device->info->a6xx.indirect_draw_wfm_quirk) draw_wfm(cmd); tu6_draw_common(cmd, cs, true, 0); tu_cs_emit_pkt7(cs, CP_DRAW_INDIRECT_MULTI, 9); tu_cs_emit(cs, tu_draw_initiator(cmd, DI_SRC_SEL_DMA)); tu_cs_emit(cs, A6XX_CP_DRAW_INDIRECT_MULTI_1_OPCODE(INDIRECT_OP_INDEXED) | A6XX_CP_DRAW_INDIRECT_MULTI_1_DST_OFF(vs_params_offset(cmd))); tu_cs_emit(cs, drawCount); tu_cs_emit_qw(cs, cmd->state.index_va); tu_cs_emit(cs, cmd->state.max_index_count); tu_cs_emit_qw(cs, vk_buffer_address(&buf->vk, offset)); tu_cs_emit(cs, stride); trace_end_draw(&cmd->trace, cs); } TU_GENX(tu_CmdDrawIndexedIndirect); template VKAPI_ATTR void VKAPI_CALL tu_CmdDrawIndirectCount(VkCommandBuffer commandBuffer, VkBuffer _buffer, VkDeviceSize offset, VkBuffer countBuffer, VkDeviceSize countBufferOffset, uint32_t drawCount, uint32_t stride) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); VK_FROM_HANDLE(tu_buffer, buf, _buffer); VK_FROM_HANDLE(tu_buffer, count_buf, countBuffer); struct tu_cs *cs = &cmd->draw_cs; tu6_emit_empty_vs_params(cmd); /* It turns out that the firmware we have for a650 only partially fixed the * problem with CP_DRAW_INDIRECT_MULTI not waiting for WFI's to complete * before reading indirect parameters. It waits for WFI's before reading * the draw parameters, but after reading the indirect count :(. */ draw_wfm(cmd); tu6_draw_common(cmd, cs, false, 0); tu_cs_emit_pkt7(cs, CP_DRAW_INDIRECT_MULTI, 8); tu_cs_emit(cs, tu_draw_initiator(cmd, DI_SRC_SEL_AUTO_INDEX)); tu_cs_emit(cs, A6XX_CP_DRAW_INDIRECT_MULTI_1_OPCODE(INDIRECT_OP_INDIRECT_COUNT) | A6XX_CP_DRAW_INDIRECT_MULTI_1_DST_OFF(vs_params_offset(cmd))); tu_cs_emit(cs, drawCount); tu_cs_emit_qw(cs, vk_buffer_address(&buf->vk, offset)); tu_cs_emit_qw(cs, vk_buffer_address(&count_buf->vk, countBufferOffset)); tu_cs_emit(cs, stride); trace_end_draw(&cmd->trace, cs); } TU_GENX(tu_CmdDrawIndirectCount); template VKAPI_ATTR void VKAPI_CALL tu_CmdDrawIndexedIndirectCount(VkCommandBuffer commandBuffer, VkBuffer _buffer, VkDeviceSize offset, VkBuffer countBuffer, VkDeviceSize countBufferOffset, uint32_t drawCount, uint32_t stride) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); VK_FROM_HANDLE(tu_buffer, buf, _buffer); VK_FROM_HANDLE(tu_buffer, count_buf, countBuffer); struct tu_cs *cs = &cmd->draw_cs; tu6_emit_empty_vs_params(cmd); draw_wfm(cmd); tu6_draw_common(cmd, cs, true, 0); tu_cs_emit_pkt7(cs, CP_DRAW_INDIRECT_MULTI, 11); tu_cs_emit(cs, tu_draw_initiator(cmd, DI_SRC_SEL_DMA)); tu_cs_emit(cs, A6XX_CP_DRAW_INDIRECT_MULTI_1_OPCODE(INDIRECT_OP_INDIRECT_COUNT_INDEXED) | A6XX_CP_DRAW_INDIRECT_MULTI_1_DST_OFF(vs_params_offset(cmd))); tu_cs_emit(cs, drawCount); tu_cs_emit_qw(cs, cmd->state.index_va); tu_cs_emit(cs, cmd->state.max_index_count); tu_cs_emit_qw(cs, vk_buffer_address(&buf->vk, offset)); tu_cs_emit_qw(cs, vk_buffer_address(&count_buf->vk, countBufferOffset)); tu_cs_emit(cs, stride); trace_end_draw(&cmd->trace, cs); } TU_GENX(tu_CmdDrawIndexedIndirectCount); template VKAPI_ATTR void VKAPI_CALL tu_CmdDrawIndirectByteCountEXT(VkCommandBuffer commandBuffer, uint32_t instanceCount, uint32_t firstInstance, VkBuffer _counterBuffer, VkDeviceSize counterBufferOffset, uint32_t counterOffset, uint32_t vertexStride) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); VK_FROM_HANDLE(tu_buffer, buf, _counterBuffer); struct tu_cs *cs = &cmd->draw_cs; /* All known firmware versions do not wait for WFI's with CP_DRAW_AUTO. * Plus, for the common case where the counter buffer is written by * vkCmdEndTransformFeedback, we need to wait for the CP_WAIT_MEM_WRITES to * complete which means we need a WAIT_FOR_ME anyway. */ draw_wfm(cmd); tu6_emit_vs_params(cmd, 0, 0, firstInstance); tu6_draw_common(cmd, cs, false, 0); tu_cs_emit_pkt7(cs, CP_DRAW_AUTO, 6); tu_cs_emit(cs, tu_draw_initiator(cmd, DI_SRC_SEL_AUTO_XFB)); if (CHIP >= A7XX) { /* On a7xx the counter value and offset are shifted right by 2, so * the vertexStride should also be in units of dwords. */ vertexStride = vertexStride >> 2; } tu_cs_emit(cs, instanceCount); tu_cs_emit_qw(cs, vk_buffer_address(&buf->vk, counterBufferOffset)); tu_cs_emit(cs, counterOffset); tu_cs_emit(cs, vertexStride); trace_end_draw(&cmd->trace, cs); } TU_GENX(tu_CmdDrawIndirectByteCountEXT); struct tu_dispatch_info { /** * Determine the layout of the grid (in block units) to be used. */ uint32_t blocks[3]; /** * A starting offset for the grid. If unaligned is set, the offset * must still be aligned. */ uint32_t offsets[3]; /** * Whether it's an unaligned compute dispatch. */ bool unaligned; /** * Indirect compute parameters resource. */ VkDeviceAddress indirect; }; static inline struct ir3_driver_params_cs build_driver_params_cs(const struct ir3_shader_variant *variant, const struct tu_dispatch_info *info) { unsigned subgroup_size = variant->info.subgroup_size; unsigned subgroup_shift = util_logbase2(subgroup_size); return (struct ir3_driver_params_cs) { .num_work_groups_x = info->blocks[0], .num_work_groups_y = info->blocks[1], .num_work_groups_z = info->blocks[2], .work_dim = 0, .base_group_x = info->offsets[0], .base_group_y = info->offsets[1], .base_group_z = info->offsets[2], .subgroup_size = subgroup_size, .local_group_size_x = 0, .local_group_size_y = 0, .local_group_size_z = 0, .subgroup_id_shift = subgroup_shift, }; } template static void tu_emit_compute_driver_params(struct tu_cmd_buffer *cmd, struct tu_cs *cs, const struct tu_dispatch_info *info) { mesa_shader_stage type = MESA_SHADER_COMPUTE; const struct tu_shader *shader = cmd->state.shaders[MESA_SHADER_COMPUTE]; const struct ir3_shader_variant *variant = shader->variant; const struct ir3_const_state *const_state = variant->const_state; unsigned subgroup_size = variant->info.subgroup_size; unsigned subgroup_shift = util_logbase2(subgroup_size); if (cmd->device->physical_device->info->a7xx.load_shader_consts_via_preamble) { uint32_t num_consts = const_state->driver_params_ubo.size; if (num_consts == 0) return; bool direct_indirect_load = !(info->indirect & 0xf) && !(info->indirect && num_consts > IR3_DP_CS(base_group_x)); uint64_t iova = 0; if (!info->indirect) { struct ir3_driver_params_cs driver_params = build_driver_params_cs(variant, info); assert(num_consts <= dword_sizeof(driver_params)); struct tu_cs_memory consts; uint32_t consts_vec4 = DIV_ROUND_UP(num_consts, 4); VkResult result = tu_cs_alloc(&cmd->sub_cs, consts_vec4, 4, &consts); if (result != VK_SUCCESS) { vk_command_buffer_set_error(&cmd->vk, result); return; } memcpy(consts.map, &driver_params, num_consts * sizeof(uint32_t)); iova = consts.iova; } else if (direct_indirect_load) { iova = info->indirect; } else { /* Vulkan guarantees only 4 byte alignment for indirect_offset. * However, CP_LOAD_STATE.EXT_SRC_ADDR needs 16 byte alignment. */ uint64_t indirect_iova = info->indirect; /* Wait for any previous uses to finish. */ tu_cs_emit_wfi(cs); for (uint32_t i = 0; i < 3; i++) { tu_cs_emit_pkt7(cs, CP_MEM_TO_MEM, 5); tu_cs_emit(cs, 0); tu_cs_emit_qw(cs, global_iova_arr(cmd, cs_indirect_xyz, i)); tu_cs_emit_qw(cs, indirect_iova + i * sizeof(uint32_t)); } /* Fill out IR3_DP_CS_SUBGROUP_SIZE and IR3_DP_SUBGROUP_ID_SHIFT for * indirect dispatch. */ if (info->indirect && num_consts > IR3_DP_CS(base_group_x)) { uint32_t indirect_driver_params[8] = { 0, 0, 0, subgroup_size, 0, 0, 0, subgroup_shift, }; bool emit_local = num_consts > IR3_DP_CS(local_group_size_x); uint32_t emit_size = emit_local ? 8 : 4; tu_cs_emit_pkt7(cs, CP_MEM_WRITE, 2 + emit_size); tu_cs_emit_qw(cs, global_iova_arr(cmd, cs_indirect_xyz, 0) + 4 * sizeof(uint32_t)); for (uint32_t i = 0; i < emit_size; i++) { tu_cs_emit(cs, indirect_driver_params[i]); } } tu_cs_emit_pkt7(cs, CP_WAIT_MEM_WRITES, 0); tu_emit_event_write(cmd, cs, FD_CACHE_INVALIDATE); tu_cs_emit_wfi(cs); iova = global_iova(cmd, cs_indirect_xyz[0]); } tu_cs_emit_pkt7(cs, tu6_stage2opcode(type), 5); tu_cs_emit(cs, CP_LOAD_STATE6_0_DST_OFF(const_state->driver_params_ubo.idx) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_UBO) | CP_LOAD_STATE6_0_STATE_SRC(SS6_DIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(tu6_stage2shadersb(type)) | CP_LOAD_STATE6_0_NUM_UNIT(1)); tu_cs_emit(cs, CP_LOAD_STATE6_1_EXT_SRC_ADDR(0)); tu_cs_emit(cs, CP_LOAD_STATE6_2_EXT_SRC_ADDR_HI(0)); int size_vec4s = DIV_ROUND_UP(num_consts, 4); tu_cs_emit_qw(cs, iova | ((uint64_t)A6XX_UBO_1_SIZE(size_vec4s) << 32)); } else { uint32_t offset = const_state->allocs.consts[IR3_CONST_ALLOC_DRIVER_PARAMS].offset_vec4; if (!ir3_const_can_upload(&const_state->allocs, IR3_CONST_ALLOC_DRIVER_PARAMS, variant->constlen)) return; uint32_t num_consts = MIN2(const_state->num_driver_params, (variant->constlen - offset) * 4); if (!info->indirect) { struct ir3_driver_params_cs driver_params = build_driver_params_cs(variant, info); assert(num_consts <= dword_sizeof(driver_params)); /* push constants */ tu_cs_emit_pkt7(cs, tu6_stage2opcode(type), 3 + num_consts); tu_cs_emit(cs, CP_LOAD_STATE6_0_DST_OFF(offset) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_CONSTANTS) | CP_LOAD_STATE6_0_STATE_SRC(SS6_DIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(tu6_stage2shadersb(type)) | CP_LOAD_STATE6_0_NUM_UNIT(num_consts / 4)); tu_cs_emit(cs, 0); tu_cs_emit(cs, 0); tu_cs_emit_array(cs, (uint32_t *)&driver_params, num_consts); } else if (!(info->indirect & 0xf)) { tu_cs_emit_pkt7(cs, tu6_stage2opcode(type), 3); tu_cs_emit(cs, CP_LOAD_STATE6_0_DST_OFF(offset) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_CONSTANTS) | CP_LOAD_STATE6_0_STATE_SRC(SS6_INDIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(tu6_stage2shadersb(type)) | CP_LOAD_STATE6_0_NUM_UNIT(1)); tu_cs_emit_qw(cs, info->indirect); } else { /* Vulkan guarantees only 4 byte alignment for indirect_offset. * However, CP_LOAD_STATE.EXT_SRC_ADDR needs 16 byte alignment. */ /* Wait for any previous uses to finish. */ tu_cs_emit_wfi(cs); for (uint32_t i = 0; i < 3; i++) { tu_cs_emit_pkt7(cs, CP_MEM_TO_MEM, 5); tu_cs_emit(cs, 0); tu_cs_emit_qw(cs, global_iova_arr(cmd, cs_indirect_xyz, i)); tu_cs_emit_qw(cs, info->indirect + i * 4); } tu_cs_emit_pkt7(cs, CP_WAIT_MEM_WRITES, 0); tu_emit_event_write(cmd, cs, FD_CACHE_INVALIDATE); tu_cs_emit_wfi(cs); tu_cs_emit_pkt7(cs, tu6_stage2opcode(type), 3); tu_cs_emit(cs, CP_LOAD_STATE6_0_DST_OFF(offset) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_CONSTANTS) | CP_LOAD_STATE6_0_STATE_SRC(SS6_INDIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(tu6_stage2shadersb(type)) | CP_LOAD_STATE6_0_NUM_UNIT(1)); tu_cs_emit_qw(cs, global_iova(cmd, cs_indirect_xyz[0])); } /* Fill out IR3_DP_CS_SUBGROUP_SIZE and IR3_DP_SUBGROUP_ID_SHIFT for * indirect dispatch. */ if (info->indirect && num_consts > IR3_DP_CS(base_group_x)) { bool emit_local = num_consts > IR3_DP_CS(local_group_size_x); tu_cs_emit_pkt7(cs, tu6_stage2opcode(type), 7 + (emit_local ? 4 : 0)); tu_cs_emit(cs, CP_LOAD_STATE6_0_DST_OFF(offset + (IR3_DP_CS(base_group_x) / 4)) | CP_LOAD_STATE6_0_STATE_TYPE(ST6_CONSTANTS) | CP_LOAD_STATE6_0_STATE_SRC(SS6_DIRECT) | CP_LOAD_STATE6_0_STATE_BLOCK(tu6_stage2shadersb(type)) | CP_LOAD_STATE6_0_NUM_UNIT((num_consts - IR3_DP_CS(base_group_x)) / 4)); tu_cs_emit_qw(cs, 0); tu_cs_emit(cs, 0); /* BASE_GROUP_X */ tu_cs_emit(cs, 0); /* BASE_GROUP_Y */ tu_cs_emit(cs, 0); /* BASE_GROUP_Z */ tu_cs_emit(cs, subgroup_size); if (emit_local) { assert(num_consts == align(IR3_DP_CS(subgroup_id_shift), 4)); tu_cs_emit(cs, 0); /* LOCAL_GROUP_SIZE_X */ tu_cs_emit(cs, 0); /* LOCAL_GROUP_SIZE_Y */ tu_cs_emit(cs, 0); /* LOCAL_GROUP_SIZE_Z */ tu_cs_emit(cs, subgroup_shift); } } } } template static void tu_dispatch(struct tu_cmd_buffer *cmd, const struct tu_dispatch_info *info) { if (!info->indirect && (info->blocks[0] == 0 || info->blocks[1] == 0 || info->blocks[2] == 0)) return; struct tu_cs *cs = &cmd->cs; struct tu_shader *shader = cmd->state.shaders[MESA_SHADER_COMPUTE]; bool emit_instrlen_workaround = shader->variant->instrlen > cmd->device->physical_device->info->a6xx.instr_cache_size; /* We don't use draw states for dispatches, so the bound pipeline * could be overwritten by reg stomping in a renderpass or blit. */ if (cmd->device->dbg_renderpass_stomp_cs) { tu_cs_emit_state_ib(&cmd->cs, shader->state); } /* There appears to be a HW bug where in some rare circumstances it appears * to accidentally use the FS instrlen instead of the CS instrlen, which * affects all known gens. Based on various experiments it appears that the * issue is that when prefetching a branch destination and there is a cache * miss, when fetching from memory the HW bounds-checks the fetch against * SP_CS_INSTR_SIZE, except when one of the two register contexts is active * it accidentally fetches SP_PS_INSTR_SIZE from the other (inactive) * context. To workaround it we set the FS instrlen here and do a dummy * event to roll the context (because it fetches SP_PS_INSTR_SIZE from the * "wrong" context). Because the bug seems to involve cache misses, we * don't emit this if the entire CS program fits in cache, which will * hopefully be the majority of cases. * * See https://gitlab.freedesktop.org/mesa/mesa/-/issues/5892 */ if (emit_instrlen_workaround) { tu_cs_emit_regs(cs, A6XX_SP_PS_INSTR_SIZE(shader->variant->instrlen)); tu_emit_event_write(cmd, cs, FD_LABEL); } /* TODO: We could probably flush less if we add a compute_flush_bits * bitfield. */ tu_emit_cache_flush(cmd); /* note: no reason to have this in a separate IB */ tu_cs_emit_state_ib(cs, tu_emit_consts(cmd, true)); tu_emit_compute_driver_params(cmd, cs, info); if (cmd->state.dirty & TU_CMD_DIRTY_COMPUTE_DESC_SETS) { tu6_emit_descriptor_sets(cmd, VK_PIPELINE_BIND_POINT_COMPUTE); tu_cs_emit_state_ib(cs, cmd->state.compute_load_state); } cmd->state.dirty &= ~TU_CMD_DIRTY_COMPUTE_DESC_SETS; tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_MODE(RM6_COMPUTE)); const uint16_t *local_size = shader->variant->local_size; const uint32_t *num_groups = info->blocks; if (info->unaligned) { assert(CHIP >= A7XX); if (info->indirect) { /* This path is tailored for BVH building and currently only supports * 1-dimensional dispatches with a power-of-two local size. We use * CP_RUN_OPENCL instead of CP_EXEC_CS in order to dynamically set * SP_CS_KERNEL_GROUP_X, which is usually set implicitly by the * packet, to the number of workgroups. The registers for Y and Z * dimensions should be unused because we set the kernel dimension to * 1. */ assert(local_size[1] == 1 && local_size[2] == 1); assert(util_is_power_of_two_nonzero(local_size[0])); tu_cs_emit_regs(cs, SP_CS_NDRANGE_0(CHIP, .kerneldim = 1, .localsizex = local_size[0] - 1)); tu_cs_emit_regs(cs, SP_CS_NDRANGE_2(CHIP, .globaloff_x = 0)); /* This does: * - waits for pending cache flushes to finish * - CP_WAIT_FOR_ME * * In a sequence of indirect dispatches this shouldn't wait for the * previous dispatches to finish. */ tu_cs_emit_pkt7(cs, CP_MEM_TO_REG, 3); tu_cs_emit(cs, CP_MEM_TO_REG_0_REG(REG_A7XX_SP_CS_NDRANGE_1)); tu_cs_emit_qw(cs, info->indirect); tu_cs_emit_pkt7(cs, CP_SCRATCH_WRITE, 2); tu_cs_emit(cs, CP_SCRATCH_WRITE_0_SCRATCH(0)); tu_cs_emit(cs, ~0u); /* CP_REG_RMW and CP_REG_TO_SCRATCH implicitly do a CP_WAIT_FOR_IDLE * *and* CP_WAIT_FOR_ME, which is a full pipeline stall that we don't * want, so manually wait for the CP_MEM_TO_REG write to land and * then skip waiting below with SKIP_WAIT_FOR_ME. */ tu_cs_emit_pkt7(cs, CP_WAIT_FOR_ME, 0); /* scratch0 = ((scratch0 & CS_NDRANGE_1) + -1 * = ((~0 & CS_NDRANGE_1) + -1 * = CS_NDRANGE_1 - 1 */ tu_cs_emit_pkt7(cs, CP_REG_RMW, 3); tu_cs_emit(cs, CP_REG_RMW_0_DST_REG(0) | CP_REG_RMW_0_DST_SCRATCH | CP_REG_RMW_0_SKIP_WAIT_FOR_ME | CP_REG_RMW_0_SRC0_IS_REG | CP_REG_RMW_0_SRC1_ADD); tu_cs_emit(cs, REG_A7XX_SP_CS_NDRANGE_1); /* SRC0 */ tu_cs_emit(cs, -1); /* SRC1 */ /* scratch0 = ((scratch0 & (local_size - 1)) rot 2 * = ((scratch0 & (local_size - 1)) << 2 */ tu_cs_emit_pkt7(cs, CP_REG_RMW, 3); tu_cs_emit(cs, CP_REG_RMW_0_DST_REG(0) | CP_REG_RMW_0_DST_SCRATCH | CP_REG_RMW_0_SKIP_WAIT_FOR_ME | CP_REG_RMW_0_ROTATE(A7XX_SP_CS_NDRANGE_7_LOCALSIZEX__SHIFT)); tu_cs_emit(cs, local_size[0] - 1); /* SRC0 */ tu_cs_emit(cs, 0); /* SRC1 */ /* write scratch0 to SP_CS_NDRANGE_7 */ tu_cs_emit_pkt7(cs, CP_SCRATCH_TO_REG, 1); tu_cs_emit(cs, CP_SCRATCH_TO_REG_0_REG(REG_A7XX_SP_CS_NDRANGE_7) | CP_SCRATCH_TO_REG_0_SCRATCH(0)); tu_cs_emit_pkt7(cs, CP_SCRATCH_WRITE, 2); tu_cs_emit(cs, CP_SCRATCH_WRITE_0_SCRATCH(0)); tu_cs_emit(cs, ~0u); /* scratch0 = (scratch0 & CS_NDRANGE_1) + local_size - 1 * = (~0u & CS_NDRANGE_1) + local_size - 1 * = CS_NDRANGE_1 + local_size - 1 */ tu_cs_emit_pkt7(cs, CP_REG_RMW, 3); tu_cs_emit(cs, CP_REG_RMW_0_DST_REG(0) | CP_REG_RMW_0_DST_SCRATCH | CP_REG_RMW_0_SKIP_WAIT_FOR_ME | CP_REG_RMW_0_SRC0_IS_REG | CP_REG_RMW_0_SRC1_ADD); tu_cs_emit(cs, REG_A7XX_SP_CS_NDRANGE_1); /* SRC0 */ tu_cs_emit(cs, local_size[0] - 1); /* SRC1 */ unsigned local_size_log2 = util_logbase2(local_size[0]); /* scratch0 = (scratch0 & (~(local_size - 1)) rot (32 - log2(local_size)) * = scratch0 >> log2(local_size) * = scratch0 / local_size * = (CS_NDRANGE_1 + local_size - 1) / local_size */ tu_cs_emit_pkt7(cs, CP_REG_RMW, 3); tu_cs_emit(cs, CP_REG_RMW_0_DST_REG(0) | CP_REG_RMW_0_DST_SCRATCH | CP_REG_RMW_0_SKIP_WAIT_FOR_ME | CP_REG_RMW_0_ROTATE(32 - local_size_log2)); tu_cs_emit(cs, ~(local_size[0] - 1)); /* SRC0 */ tu_cs_emit(cs, 0); /* SRC1 */ /* write scratch0 to SP_CS_KERNEL_GROUP_X */ tu_cs_emit_pkt7(cs, CP_SCRATCH_TO_REG, 1); tu_cs_emit(cs, CP_SCRATCH_TO_REG_0_REG(REG_A7XX_SP_CS_KERNEL_GROUP_X) | CP_SCRATCH_TO_REG_0_SCRATCH(0)); } else { tu_cs_emit_regs(cs, SP_CS_NDRANGE_0(CHIP, .kerneldim = 3, .localsizex = local_size[0] - 1, .localsizey = local_size[1] - 1, .localsizez = local_size[2] - 1), SP_CS_NDRANGE_1(CHIP, .globalsize_x = num_groups[0]), SP_CS_NDRANGE_2(CHIP, .globaloff_x = 0), SP_CS_NDRANGE_3(CHIP, .globalsize_y = num_groups[1]), SP_CS_NDRANGE_4(CHIP, .globaloff_y = 0), SP_CS_NDRANGE_5(CHIP, .globalsize_z = num_groups[2]), SP_CS_NDRANGE_6(CHIP, .globaloff_z = 0)); uint32_t last_local_size[3]; for (unsigned i = 0; i < 3; i++) last_local_size[i] = ((num_groups[i] - 1) % local_size[i]) + 1; tu_cs_emit_regs(cs, A7XX_SP_CS_NDRANGE_7(.localsizex = last_local_size[0] - 1, .localsizey = last_local_size[1] - 1, .localsizez = last_local_size[2] - 1)); } } else { tu_cs_emit_regs(cs, SP_CS_NDRANGE_0(CHIP, .kerneldim = 3, .localsizex = local_size[0] - 1, .localsizey = local_size[1] - 1, .localsizez = local_size[2] - 1), SP_CS_NDRANGE_1(CHIP, .globalsize_x = local_size[0] * num_groups[0]), SP_CS_NDRANGE_2(CHIP, .globaloff_x = 0), SP_CS_NDRANGE_3(CHIP, .globalsize_y = local_size[1] * num_groups[1]), SP_CS_NDRANGE_4(CHIP, .globaloff_y = 0), SP_CS_NDRANGE_5(CHIP, .globalsize_z = local_size[2] * num_groups[2]), SP_CS_NDRANGE_6(CHIP, .globaloff_z = 0)); if (CHIP >= A7XX) { tu_cs_emit_regs(cs, A7XX_SP_CS_NDRANGE_7(.localsizex = local_size[0] - 1, .localsizey = local_size[1] - 1, .localsizez = local_size[2] - 1)); } } if (cmd->device->physical_device->info->a7xx.has_rt_workaround && shader->variant->info.uses_ray_intersection) { tu_cs_emit_pkt7(cs, CP_SET_MARKER, 1); tu_cs_emit(cs, A6XX_CP_SET_MARKER_0_SHADER_USES_RT); } if (info->indirect) { trace_start_compute_indirect(&cmd->trace, cs, cmd, info->unaligned, (char *)shader->variant->sha1_str); if (info->unaligned) { tu_cs_emit_pkt7(cs, CP_RUN_OPENCL, 1); tu_cs_emit(cs, 0x00000000); } else { tu_cs_emit_pkt7(cs, CP_EXEC_CS_INDIRECT, 4); tu_cs_emit(cs, 0x00000000); tu_cs_emit_qw(cs, info->indirect); tu_cs_emit(cs, A5XX_CP_EXEC_CS_INDIRECT_3_LOCALSIZEX(local_size[0] - 1) | A5XX_CP_EXEC_CS_INDIRECT_3_LOCALSIZEY(local_size[1] - 1) | A5XX_CP_EXEC_CS_INDIRECT_3_LOCALSIZEZ(local_size[2] - 1)); } trace_end_compute_indirect(&cmd->trace, cs, (struct u_trace_address) { .bo = NULL, .offset = info->indirect, }); } else { trace_start_compute(&cmd->trace, cs, cmd, info->indirect != 0, info->unaligned, local_size[0], local_size[1], local_size[2], info->blocks[0], info->blocks[1], info->blocks[2], (char *)shader->variant->sha1_str); if (info->unaligned) { tu_cs_emit_pkt7(cs, CP_EXEC_CS, 4); tu_cs_emit(cs, 0x00000000); tu_cs_emit(cs, CP_EXEC_CS_1_NGROUPS_X(DIV_ROUND_UP(info->blocks[0], local_size[0]))); tu_cs_emit(cs, CP_EXEC_CS_2_NGROUPS_Y(DIV_ROUND_UP(info->blocks[1], local_size[1]))); tu_cs_emit(cs, CP_EXEC_CS_3_NGROUPS_Z(DIV_ROUND_UP(info->blocks[2], local_size[2]))); } else { tu_cs_emit_pkt7(cs, CP_EXEC_CS, 4); tu_cs_emit(cs, 0x00000000); tu_cs_emit(cs, CP_EXEC_CS_1_NGROUPS_X(info->blocks[0])); tu_cs_emit(cs, CP_EXEC_CS_2_NGROUPS_Y(info->blocks[1])); tu_cs_emit(cs, CP_EXEC_CS_3_NGROUPS_Z(info->blocks[2])); } trace_end_compute(&cmd->trace, cs); } /* For the workaround above, because it's using the "wrong" context for * SP_PS_INSTR_SIZE we should emit another dummy event write to avoid a * potential race between writing the register and the CP_EXEC_CS we just * did. We don't need to reset the register because it will be re-emitted * anyway when the next renderpass starts. */ if (emit_instrlen_workaround) { tu_emit_event_write(cmd, cs, FD_LABEL); } cmd->state.total_dispatches++; } template VKAPI_ATTR void VKAPI_CALL tu_CmdDispatchBase(VkCommandBuffer commandBuffer, uint32_t base_x, uint32_t base_y, uint32_t base_z, uint32_t x, uint32_t y, uint32_t z) { VK_FROM_HANDLE(tu_cmd_buffer, cmd_buffer, commandBuffer); struct tu_dispatch_info info = {}; info.blocks[0] = x; info.blocks[1] = y; info.blocks[2] = z; info.offsets[0] = base_x; info.offsets[1] = base_y; info.offsets[2] = base_z; tu_dispatch(cmd_buffer, &info); } TU_GENX(tu_CmdDispatchBase); template VKAPI_ATTR void VKAPI_CALL tu_CmdDispatchIndirect(VkCommandBuffer commandBuffer, VkBuffer _buffer, VkDeviceSize offset) { VK_FROM_HANDLE(tu_cmd_buffer, cmd_buffer, commandBuffer); VK_FROM_HANDLE(tu_buffer, buffer, _buffer); struct tu_dispatch_info info = {}; info.indirect = vk_buffer_address(&buffer->vk, offset); tu_dispatch(cmd_buffer, &info); } TU_GENX(tu_CmdDispatchIndirect); void tu_dispatch_unaligned(VkCommandBuffer commandBuffer, uint32_t x, uint32_t y, uint32_t z) { VK_FROM_HANDLE(tu_cmd_buffer, cmd_buffer, commandBuffer); struct tu_dispatch_info info = {}; info.unaligned = true; info.blocks[0] = x; info.blocks[1] = y; info.blocks[2] = z; TU_CALLX(cmd_buffer->device, tu_dispatch)(cmd_buffer, &info); } void tu_dispatch_unaligned_indirect(VkCommandBuffer commandBuffer, VkDeviceAddress size_addr) { VK_FROM_HANDLE(tu_cmd_buffer, cmd_buffer, commandBuffer); struct tu_dispatch_info info = {}; info.unaligned = true; info.indirect = size_addr; TU_CALLX(cmd_buffer->device, tu_dispatch)(cmd_buffer, &info); } VKAPI_ATTR void VKAPI_CALL tu_CmdEndRenderPass2(VkCommandBuffer commandBuffer, const VkSubpassEndInfo *pSubpassEndInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd_buffer, commandBuffer); if (TU_DEBUG(DYNAMIC)) { vk_common_CmdEndRenderPass2(commandBuffer, pSubpassEndInfo); return; } const VkRenderPassFragmentDensityMapOffsetEndInfoEXT *fdm_offset_info = vk_find_struct_const(pSubpassEndInfo->pNext, RENDER_PASS_FRAGMENT_DENSITY_MAP_OFFSET_END_INFO_EXT); const VkOffset2D *fdm_offsets = (fdm_offset_info && fdm_offset_info->fragmentDensityOffsetCount > 0) ? fdm_offset_info->pFragmentDensityOffsets : NULL; VkOffset2D test_offsets[MAX_VIEWS]; if (TU_DEBUG(FDM) && TU_DEBUG(FDM_OFFSET)) { for (unsigned i = 0; i < tu_fdm_num_layers(cmd_buffer); i++) { test_offsets[i] = { 64, 64 }; } fdm_offsets = test_offsets; } tu_cs_end(&cmd_buffer->draw_cs); tu_cs_end(&cmd_buffer->draw_epilogue_cs); TU_CALLX(cmd_buffer->device, tu_cmd_render)(cmd_buffer, fdm_offsets); cmd_buffer->state.cache.pending_flush_bits |= cmd_buffer->state.renderpass_cache.pending_flush_bits; tu_subpass_barrier(cmd_buffer, &cmd_buffer->state.pass->end_barrier, true); vk_free(&cmd_buffer->vk.pool->alloc, cmd_buffer->state.attachments); tu_reset_render_pass(cmd_buffer); cmd_buffer->state.total_renderpasses++; } VKAPI_ATTR void VKAPI_CALL tu_CmdEndRendering2EXT(VkCommandBuffer commandBuffer, const VkRenderingEndInfoEXT *pRenderingEndInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd_buffer, commandBuffer); if (cmd_buffer->state.suspending) { cmd_buffer->state.suspended_pass.lrz = cmd_buffer->state.lrz; /* We cannot pass LRZ state to next resuming renderpass, so we have to * force disable it here. */ TU_CALLX(cmd_buffer->device, tu_lrz_flush_valid_during_renderpass) (cmd_buffer, &cmd_buffer->draw_cs); } const VkRenderPassFragmentDensityMapOffsetEndInfoEXT *fdm_offset_info = vk_find_struct_const(pRenderingEndInfo, RENDER_PASS_FRAGMENT_DENSITY_MAP_OFFSET_END_INFO_EXT); const VkOffset2D *fdm_offsets = (fdm_offset_info && fdm_offset_info->fragmentDensityOffsetCount > 0) ? fdm_offset_info->pFragmentDensityOffsets : NULL; VkOffset2D test_offsets[MAX_VIEWS]; if (TU_DEBUG(FDM) && TU_DEBUG(FDM_OFFSET)) { for (unsigned i = 0; i < tu_fdm_num_layers(cmd_buffer); i++) { test_offsets[i] = { 64, 64 }; } fdm_offsets = test_offsets; } if (!cmd_buffer->state.suspending) { tu_cs_end(&cmd_buffer->draw_cs); tu_cs_end(&cmd_buffer->draw_epilogue_cs); if (cmd_buffer->state.suspend_resume == SR_IN_PRE_CHAIN) { tu_save_pre_chain(cmd_buffer); cmd_buffer->pre_chain.fdm_offset = !!fdm_offsets; if (fdm_offsets) { memcpy(cmd_buffer->pre_chain.fdm_offsets, fdm_offsets, sizeof(VkOffset2D) * tu_fdm_num_layers(cmd_buffer)); } /* Even we don't call tu_cmd_render here, renderpass is finished * and draw states should be disabled. */ tu_disable_draw_states(cmd_buffer, &cmd_buffer->cs); } else { TU_CALLX(cmd_buffer->device, tu_cmd_render)(cmd_buffer, fdm_offsets); } tu_reset_render_pass(cmd_buffer); } if (cmd_buffer->state.resuming && !cmd_buffer->state.suspending) { /* exiting suspend/resume chain */ switch (cmd_buffer->state.suspend_resume) { case SR_IN_CHAIN: cmd_buffer->state.suspend_resume = SR_NONE; break; case SR_IN_PRE_CHAIN: case SR_IN_CHAIN_AFTER_PRE_CHAIN: cmd_buffer->state.suspend_resume = SR_AFTER_PRE_CHAIN; break; default: UNREACHABLE("suspending render pass not followed by resuming pass"); } } if (!cmd_buffer->state.suspending) { cmd_buffer->state.total_renderpasses++; } } void tu_barrier(struct tu_cmd_buffer *cmd, uint32_t dep_count, const VkDependencyInfo *dep_infos) { VkPipelineStageFlags2 srcStage = 0; VkPipelineStageFlags2 dstStage = 0; BITMASK_ENUM(tu_cmd_access_mask) src_flags = 0; BITMASK_ENUM(tu_cmd_access_mask) dst_flags = 0; /* Inside a renderpass, we don't know yet whether we'll be using sysmem * so we have to use the sysmem flushes. */ bool gmem = cmd->state.ccu_state == TU_CMD_CCU_GMEM && !cmd->state.pass; for (uint32_t dep_idx = 0; dep_idx < dep_count; dep_idx++) { const VkDependencyInfo *dep_info = &dep_infos[dep_idx]; for (uint32_t i = 0; i < dep_info->memoryBarrierCount; i++) { const VkMemoryBarrier2 *barrier = &dep_info->pMemoryBarriers[i]; VkPipelineStageFlags2 sanitized_src_stage = sanitize_src_stage(barrier->srcStageMask); VkPipelineStageFlags2 sanitized_dst_stage = sanitize_dst_stage(barrier->dstStageMask); VkAccessFlags3KHR src_access_mask2 = 0, dst_access_mask2 = 0; const VkMemoryBarrierAccessFlags3KHR *access3 = vk_find_struct_const(barrier->pNext, MEMORY_BARRIER_ACCESS_FLAGS_3_KHR); if (access3) { src_access_mask2 = access3->srcAccessMask3; dst_access_mask2 = access3->dstAccessMask3; } src_flags |= vk2tu_access(barrier->srcAccessMask, src_access_mask2, sanitized_src_stage, false, gmem, cmd->device->vk.enabled_features.sparseResidencyAliased); dst_flags |= vk2tu_access(barrier->dstAccessMask, dst_access_mask2, sanitized_dst_stage, false, gmem, cmd->device->vk.enabled_features.sparseResidencyAliased); srcStage |= sanitized_src_stage; dstStage |= sanitized_dst_stage; } for (uint32_t i = 0; i < dep_info->bufferMemoryBarrierCount; i++) { const VkBufferMemoryBarrier2 *barrier = &dep_info->pBufferMemoryBarriers[i]; VK_FROM_HANDLE(tu_buffer, buffer, barrier->buffer); bool sparse_aliasing = buffer->vk.create_flags & VK_BUFFER_CREATE_SPARSE_ALIASED_BIT; VkPipelineStageFlags2 sanitized_src_stage = sanitize_src_stage(barrier->srcStageMask); VkPipelineStageFlags2 sanitized_dst_stage = sanitize_dst_stage(barrier->dstStageMask); VkAccessFlags3KHR src_access_mask2 = 0, dst_access_mask2 = 0; const VkMemoryBarrierAccessFlags3KHR *access3 = vk_find_struct_const(barrier->pNext, MEMORY_BARRIER_ACCESS_FLAGS_3_KHR); if (access3) { src_access_mask2 = access3->srcAccessMask3; dst_access_mask2 = access3->dstAccessMask3; } src_flags |= vk2tu_access(barrier->srcAccessMask, src_access_mask2, sanitized_src_stage, false, gmem, sparse_aliasing); dst_flags |= vk2tu_access(barrier->dstAccessMask, dst_access_mask2, sanitized_dst_stage, false, gmem, sparse_aliasing); srcStage |= sanitized_src_stage; dstStage |= sanitized_dst_stage; } for (uint32_t i = 0; i < dep_info->imageMemoryBarrierCount; i++) { const VkImageMemoryBarrier2 *barrier = &dep_info->pImageMemoryBarriers[i]; VK_FROM_HANDLE(tu_image, image, barrier->image); VkImageLayout old_layout = barrier->oldLayout; bool sparse_aliasing = image->vk.create_flags & VK_BUFFER_CREATE_SPARSE_ALIASED_BIT; if (old_layout == VK_IMAGE_LAYOUT_UNDEFINED || old_layout == VK_IMAGE_LAYOUT_ZERO_INITIALIZED_EXT) { /* The underlying memory for this image may have been used earlier * within the same queue submission for a different image, which * means that there may be old, stale cache entries which are in the * "wrong" location, which could cause problems later after writing * to the image. We don't want these entries being flushed later and * overwriting the actual image, so we need to flush the CCU. */ if (vk_format_is_depth_or_stencil(image->vk.format)) { src_flags |= TU_ACCESS_CCU_DEPTH_INCOHERENT_WRITE; } else { src_flags |= TU_ACCESS_CCU_COLOR_INCOHERENT_WRITE; } } VkPipelineStageFlags2 sanitized_src_stage = sanitize_src_stage(barrier->srcStageMask); VkPipelineStageFlags2 sanitized_dst_stage = sanitize_dst_stage(barrier->dstStageMask); VkAccessFlags3KHR src_access_mask2 = 0, dst_access_mask2 = 0; const VkMemoryBarrierAccessFlags3KHR *access3 = vk_find_struct_const(barrier->pNext, MEMORY_BARRIER_ACCESS_FLAGS_3_KHR); if (access3) { src_access_mask2 = access3->srcAccessMask3; dst_access_mask2 = access3->dstAccessMask3; } src_flags |= vk2tu_access(barrier->srcAccessMask, src_access_mask2, sanitized_src_stage, true, gmem, sparse_aliasing); dst_flags |= vk2tu_access(barrier->dstAccessMask, dst_access_mask2, sanitized_dst_stage, true, gmem, sparse_aliasing); srcStage |= sanitized_src_stage; dstStage |= sanitized_dst_stage; } } if (cmd->state.pass) { const VkPipelineStageFlags framebuffer_space_stages = VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT | VK_PIPELINE_STAGE_EARLY_FRAGMENT_TESTS_BIT | VK_PIPELINE_STAGE_LATE_FRAGMENT_TESTS_BIT | VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT; /* We cannot have non-by-region "fb-space to fb-space" barriers. * * From the Vulkan 1.2.185 spec, section 7.6.1 "Subpass Self-dependency": * * If the source and destination stage masks both include * framebuffer-space stages, then dependencyFlags must include * VK_DEPENDENCY_BY_REGION_BIT. * [...] * Each of the synchronization scopes and access scopes of a * vkCmdPipelineBarrier2 or vkCmdPipelineBarrier command inside * a render pass instance must be a subset of the scopes of one of * the self-dependencies for the current subpass. * * If the self-dependency has VK_DEPENDENCY_BY_REGION_BIT or * VK_DEPENDENCY_VIEW_LOCAL_BIT set, then so must the pipeline barrier. * * By-region barriers are ok for gmem. All other barriers would involve * vtx stages which are NOT ok for gmem rendering. * See dep_invalid_for_gmem(). */ if ((srcStage & ~framebuffer_space_stages) || (dstStage & ~framebuffer_space_stages)) { cmd->state.rp.disable_gmem = true; cmd->state.rp.gmem_disable_reason = "Non-framebuffer-space barrier"; } } struct tu_cache_state *cache = cmd->state.pass ? &cmd->state.renderpass_cache : &cmd->state.cache; /* a750 has a HW bug where writing a UBWC compressed image with a compute * shader followed by reading it as a texture (or readonly image) requires * a CACHE_CLEAN event. Some notes about this bug: * - It only happens after a blit happens. * - It's fast-clear related, it happens when the image is fast cleared * before the write and the value read is (incorrectly) the fast clear * color. * - CACHE_FLUSH is supposed to be the same as CACHE_CLEAN + * CACHE_INVALIDATE, but it doesn't work whereas CACHE_CLEAN + * CACHE_INVALIDATE does. * * The srcAccess can be replaced by a OpMemoryBarrier(MakeAvailable), so * we can't use that to insert the flush. Instead we use the shader source * stage. */ if (cmd->device->physical_device->info->a7xx.ubwc_coherency_quirk && (srcStage & (VK_PIPELINE_STAGE_2_VERTEX_SHADER_BIT | VK_PIPELINE_STAGE_2_TESSELLATION_CONTROL_SHADER_BIT | VK_PIPELINE_STAGE_2_TESSELLATION_EVALUATION_SHADER_BIT | VK_PIPELINE_STAGE_2_GEOMETRY_SHADER_BIT | VK_PIPELINE_STAGE_2_FRAGMENT_SHADER_BIT | VK_PIPELINE_STAGE_2_COMPUTE_SHADER_BIT | VK_PIPELINE_STAGE_2_ALL_GRAPHICS_BIT | VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT))) { cache->flush_bits |= TU_CMD_FLAG_CACHE_CLEAN; cache->pending_flush_bits &= ~TU_CMD_FLAG_CACHE_CLEAN; } tu_flush_for_access(cache, src_flags, dst_flags); enum tu_stage src_stage = vk2tu_src_stage(cmd->device, srcStage); enum tu_stage dst_stage = vk2tu_dst_stage(cmd->device, dstStage); tu_flush_for_stage(cache, src_stage, dst_stage); } VKAPI_ATTR void VKAPI_CALL tu_CmdPipelineBarrier2(VkCommandBuffer commandBuffer, const VkDependencyInfo *pDependencyInfo) { VK_FROM_HANDLE(tu_cmd_buffer, cmd_buffer, commandBuffer); tu_barrier(cmd_buffer, 1, pDependencyInfo); } template void tu_write_event(struct tu_cmd_buffer *cmd, struct tu_event *event, VkPipelineStageFlags2 stageMask, unsigned value) { struct tu_cs *cs = &cmd->cs; /* vkCmdSetEvent/vkCmdResetEvent cannot be called inside a render pass */ assert(!cmd->state.pass); tu_emit_cache_flush(cmd); /* Flags that only require a top-of-pipe event. DrawIndirect parameters are * read by the CP, so the draw indirect stage counts as top-of-pipe too. */ VkPipelineStageFlags2 top_of_pipe_flags = VK_PIPELINE_STAGE_2_TOP_OF_PIPE_BIT | VK_PIPELINE_STAGE_2_DRAW_INDIRECT_BIT; if (!(stageMask & ~top_of_pipe_flags)) { tu_cs_emit_pkt7(cs, CP_MEM_WRITE, 3); tu_cs_emit_qw(cs, event->bo.iova); /* ADDR_LO/HI */ tu_cs_emit(cs, value); } else { /* Use a RB_DONE_TS event to wait for everything to complete. */ if (CHIP == A6XX) { tu_cs_emit_pkt7(cs, CP_EVENT_WRITE, 4); tu_cs_emit(cs, CP_EVENT_WRITE_0_EVENT(RB_DONE_TS)); } else { tu_cs_emit_pkt7(cs, CP_EVENT_WRITE7, 4); tu_cs_emit(cs, CP_EVENT_WRITE7_0(.event = RB_DONE_TS, .write_src = EV_WRITE_USER_32B, .write_dst = EV_DST_RAM, .write_enabled = true).value); } tu_cs_emit_qw(cs, event->bo.iova); tu_cs_emit(cs, value); } } TU_GENX(tu_write_event); template VKAPI_ATTR void VKAPI_CALL tu_CmdBeginConditionalRenderingEXT(VkCommandBuffer commandBuffer, const VkConditionalRenderingBeginInfoEXT *pConditionalRenderingBegin) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); cmd->state.predication_active = true; struct tu_cs *cs = cmd->state.pass ? &cmd->draw_cs : &cmd->cs; /* Wait for any writes to the predicate to land */ if (cmd->state.pass) tu_emit_cache_flush_renderpass(cmd); else tu_emit_cache_flush(cmd); VK_FROM_HANDLE(tu_buffer, buf, pConditionalRenderingBegin->buffer); uint64_t iova = vk_buffer_address(&buf->vk, pConditionalRenderingBegin->offset); /* qcom doesn't support 32-bit reference values, only 64-bit, but Vulkan * mandates 32-bit comparisons. Our workaround is to copy the the reference * value to the low 32-bits of a location where the high 32 bits are known * to be 0 and then compare that. * * BR and BV use separate predicate values so that setting the predicate * doesn't have to be synchronized between them. */ if (CHIP >= A7XX) { if (!cmd->state.pass) { tu_cs_emit_pkt7(cs, CP_THREAD_CONTROL, 1); tu_cs_emit(cs, CP_THREAD_CONTROL_0_THREAD(CP_SET_THREAD_BOTH)); } tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(THREAD_MODE) | CP_COND_REG_EXEC_0_BR); } tu_cs_emit_pkt7(cs, CP_MEM_TO_MEM, 5); tu_cs_emit(cs, 0); tu_cs_emit_qw(cs, global_iova(cmd, predicate)); tu_cs_emit_qw(cs, iova); if (CHIP >= A7XX) { tu_cond_exec_end(cs); tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(THREAD_MODE) | CP_COND_REG_EXEC_0_BV); tu_cs_emit_pkt7(cs, CP_MEM_TO_MEM, 5); tu_cs_emit(cs, 0); tu_cs_emit_qw(cs, global_iova(cmd, bv_predicate)); tu_cs_emit_qw(cs, iova); tu_cond_exec_end(cs); } tu_cs_emit_pkt7(cs, CP_WAIT_MEM_WRITES, 0); tu_cs_emit_pkt7(cs, CP_WAIT_FOR_ME, 0); tu_cs_emit_pkt7(cs, CP_DRAW_PRED_ENABLE_GLOBAL, 1); tu_cs_emit(cs, 1); bool inv = pConditionalRenderingBegin->flags & VK_CONDITIONAL_RENDERING_INVERTED_BIT_EXT; if (CHIP >= A7XX) { tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(THREAD_MODE) | CP_COND_REG_EXEC_0_BR); } tu_cs_emit_pkt7(cs, CP_DRAW_PRED_SET, 3); tu_cs_emit(cs, CP_DRAW_PRED_SET_0_SRC(PRED_SRC_MEM) | CP_DRAW_PRED_SET_0_TEST(inv ? EQ_0_PASS : NE_0_PASS)); tu_cs_emit_qw(cs, global_iova(cmd, predicate)); if (CHIP >= A7XX) { tu_cond_exec_end(cs); tu_cond_exec_start(cs, CP_COND_REG_EXEC_0_MODE(THREAD_MODE) | CP_COND_REG_EXEC_0_BV); tu_cs_emit_pkt7(cs, CP_DRAW_PRED_SET, 3); tu_cs_emit(cs, CP_DRAW_PRED_SET_0_SRC(PRED_SRC_MEM) | CP_DRAW_PRED_SET_0_TEST(inv ? EQ_0_PASS : NE_0_PASS)); tu_cs_emit_qw(cs, global_iova(cmd, bv_predicate)); tu_cond_exec_end(cs); } /* Restore original BR thread after setting BOTH */ if (CHIP >= A7XX && !cmd->state.pass) { tu_cs_emit_pkt7(cs, CP_THREAD_CONTROL, 1); tu_cs_emit(cs, CP_THREAD_CONTROL_0_THREAD(CP_SET_THREAD_BR)); } } TU_GENX(tu_CmdBeginConditionalRenderingEXT); template VKAPI_ATTR void VKAPI_CALL tu_CmdEndConditionalRenderingEXT(VkCommandBuffer commandBuffer) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); cmd->state.predication_active = false; struct tu_cs *cs = cmd->state.pass ? &cmd->draw_cs : &cmd->cs; if (CHIP >= A7XX && !cmd->state.pass) { tu_cs_emit_pkt7(cs, CP_THREAD_CONTROL, 1); tu_cs_emit(cs, CP_THREAD_CONTROL_0_THREAD(CP_SET_THREAD_BOTH)); } tu_cs_emit_pkt7(cs, CP_DRAW_PRED_ENABLE_GLOBAL, 1); tu_cs_emit(cs, 0); if (CHIP >= A7XX && !cmd->state.pass) { tu_cs_emit_pkt7(cs, CP_THREAD_CONTROL, 1); tu_cs_emit(cs, CP_THREAD_CONTROL_0_THREAD(CP_SET_THREAD_BR)); } } TU_GENX(tu_CmdEndConditionalRenderingEXT); template void tu_CmdWriteBufferMarker2AMD(VkCommandBuffer commandBuffer, VkPipelineStageFlagBits2 pipelineStage, VkBuffer dstBuffer, VkDeviceSize dstOffset, uint32_t marker) { /* Almost the same as tu_write_event, but also allowed in renderpass */ VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); VK_FROM_HANDLE(tu_buffer, buffer, dstBuffer); uint64_t va = vk_buffer_address(&buffer->vk, dstOffset); struct tu_cs *cs = cmd->state.pass ? &cmd->draw_cs : &cmd->cs; struct tu_cache_state *cache = cmd->state.pass ? &cmd->state.renderpass_cache : &cmd->state.cache; /* From the Vulkan 1.2.203 spec: * * The access scope for buffer marker writes falls under * the VK_ACCESS_TRANSFER_WRITE_BIT, and the pipeline stages for * identifying the synchronization scope must include both pipelineStage * and VK_PIPELINE_STAGE_TRANSFER_BIT. * * Transfer operations use CCU however here we write via CP. * Flush CCU in order to make the results of previous transfer * operation visible to CP. */ tu_flush_for_access(cache, TU_ACCESS_NONE, TU_ACCESS_SYSMEM_WRITE); /* Flags that only require a top-of-pipe event. DrawIndirect parameters are * read by the CP, so the draw indirect stage counts as top-of-pipe too. */ VkPipelineStageFlags2 top_of_pipe_flags = VK_PIPELINE_STAGE_2_TOP_OF_PIPE_BIT | VK_PIPELINE_STAGE_2_DRAW_INDIRECT_BIT; bool is_top_of_pipe = !(pipelineStage & ~top_of_pipe_flags); /* We have to WFI only if we flushed CCU here and are using CP_MEM_WRITE. * Otherwise: * - We do CP_EVENT_WRITE(RB_DONE_TS) which should wait for flushes; * - There was a barrier to synchronize other writes with WriteBufferMarkerAMD * and they had to include our pipelineStage which forces the WFI. */ if (cache->flush_bits && is_top_of_pipe) { cache->flush_bits |= TU_CMD_FLAG_WAIT_FOR_IDLE; } if (cmd->state.pass) { tu_emit_cache_flush_renderpass(cmd); } else { tu_emit_cache_flush(cmd); } if (is_top_of_pipe) { tu_cs_emit_pkt7(cs, CP_MEM_WRITE, 3); tu_cs_emit_qw(cs, va); /* ADDR_LO/HI */ tu_cs_emit(cs, marker); } else { /* Use a RB_DONE_TS event to wait for everything to complete. */ if (CHIP == A6XX) { tu_cs_emit_pkt7(cs, CP_EVENT_WRITE, 4); tu_cs_emit(cs, CP_EVENT_WRITE_0_EVENT(RB_DONE_TS)); } else { tu_cs_emit_pkt7(cs, CP_EVENT_WRITE7, 4); tu_cs_emit(cs, CP_EVENT_WRITE7_0(.event = RB_DONE_TS, .write_src = EV_WRITE_USER_32B, .write_dst = EV_DST_RAM, .write_enabled = true).value); } tu_cs_emit_qw(cs, va); tu_cs_emit(cs, marker); } /* Make sure the result of this write is visible to others. */ tu_flush_for_access(cache, TU_ACCESS_CP_WRITE, TU_ACCESS_NONE); } TU_GENX(tu_CmdWriteBufferMarker2AMD); void tu_write_buffer_cp(VkCommandBuffer commandBuffer, VkDeviceAddress addr, void *data, uint32_t size) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); TU_CALLX(cmd->device, tu_emit_cache_flush)(cmd); struct tu_cs *cs = &cmd->cs; tu_cs_emit_pkt7(cs, CP_MEM_WRITE, 2 + size / 4); tu_cs_emit_qw(cs, addr); tu_cs_emit_array(cs, (uint32_t *)data, size / 4); } void tu_flush_buffer_write_cp(VkCommandBuffer commandBuffer) { VK_FROM_HANDLE(tu_cmd_buffer, cmd, commandBuffer); struct tu_cache_state *cache = &cmd->state.cache; tu_flush_for_access(cache, TU_ACCESS_CP_WRITE, (enum tu_cmd_access_mask)0); }