1 files changed, 146 insertions, 38 deletions

diff --git a/servers/rendering/rendering_device.cpp b/servers/rendering/rendering_device.cpp
index 962531c866..0227472d0e 100644
--- a/servers/rendering/rendering_device.cpp
+++ b/servers/rendering/rendering_device.cpp
@@ -671,9 +671,9 @@ RID RenderingDevice::texture_create(const TextureFormat &p_format, const Texture
 	TextureFormat format = p_format;
 
 	if (format.shareable_formats.size()) {
-		ERR_FAIL_COND_V_MSG(format.shareable_formats.find(format.format) == -1, RID(),
+		ERR_FAIL_COND_V_MSG(!format.shareable_formats.has(format.format), RID(),
 				"If supplied a list of shareable formats, the current format must be present in the list");
-		ERR_FAIL_COND_V_MSG(p_view.format_override != DATA_FORMAT_MAX && format.shareable_formats.find(p_view.format_override) == -1, RID(),
+		ERR_FAIL_COND_V_MSG(p_view.format_override != DATA_FORMAT_MAX && !format.shareable_formats.has(p_view.format_override), RID(),
 				"If supplied a list of shareable formats, the current view format override must be present in the list");
 	}
 
@@ -854,7 +854,7 @@ RID RenderingDevice::texture_create_shared(const TextureView &p_view, RID p_with
 	} else {
 		ERR_FAIL_INDEX_V(p_view.format_override, DATA_FORMAT_MAX, RID());
 
-		ERR_FAIL_COND_V_MSG(texture.allowed_shared_formats.find(p_view.format_override) == -1, RID(),
+		ERR_FAIL_COND_V_MSG(!texture.allowed_shared_formats.has(p_view.format_override), RID(),
 				"Format override is not in the list of allowed shareable formats for original texture.");
 		tv.format = p_view.format_override;
 	}
@@ -984,7 +984,7 @@ RID RenderingDevice::texture_create_shared_from_slice(const TextureView &p_view,
 	} else {
 		ERR_FAIL_INDEX_V(p_view.format_override, DATA_FORMAT_MAX, RID());
 
-		ERR_FAIL_COND_V_MSG(texture.allowed_shared_formats.find(p_view.format_override) == -1, RID(),
+		ERR_FAIL_COND_V_MSG(!texture.allowed_shared_formats.has(p_view.format_override), RID(),
 				"Format override is not in the list of allowed shareable formats for original texture.");
 		tv.format = p_view.format_override;
 	}
@@ -3154,9 +3154,6 @@ Error RenderingDevice::screen_create(DisplayServer::WindowID p_screen) {
 	RDD::SwapChainID swap_chain = driver->swap_chain_create(surface);
 	ERR_FAIL_COND_V_MSG(swap_chain.id == 0, ERR_CANT_CREATE, "Unable to create swap chain.");
 
-	Error err = driver->swap_chain_resize(main_queue, swap_chain, _get_swap_chain_desired_count());
-	ERR_FAIL_COND_V_MSG(err != OK, ERR_CANT_CREATE, "Unable to resize the new swap chain.");
-
 	screen_swap_chains[p_screen] = swap_chain;
 
 	return OK;
@@ -3335,12 +3332,15 @@ Error RenderingDevice::_draw_list_setup_framebuffer(Framebuffer *p_framebuffer,
 }
 
 Error RenderingDevice::_draw_list_render_pass_begin(Framebuffer *p_framebuffer, InitialAction p_initial_color_action, FinalAction p_final_color_action, InitialAction p_initial_depth_action, FinalAction p_final_depth_action, const Vector<Color> &p_clear_colors, float p_clear_depth, uint32_t p_clear_stencil, Point2i p_viewport_offset, Point2i p_viewport_size, RDD::FramebufferID p_framebuffer_driver_id, RDD::RenderPassID p_render_pass) {
-	LocalVector<RDD::RenderPassClearValue> clear_values;
-	LocalVector<RDG::ResourceTracker *> resource_trackers;
-	LocalVector<RDG::ResourceUsage> resource_usages;
+	thread_local LocalVector<RDD::RenderPassClearValue> clear_values;
+	thread_local LocalVector<RDG::ResourceTracker *> resource_trackers;
+	thread_local LocalVector<RDG::ResourceUsage> resource_usages;
 	bool uses_color = false;
 	bool uses_depth = false;
+	clear_values.clear();
 	clear_values.resize(p_framebuffer->texture_ids.size());
+	resource_trackers.clear();
+	resource_usages.clear();
 	int clear_values_count = 0;
 	{
 		int color_index = 0;
@@ -3451,9 +3451,9 @@ RenderingDevice::DrawListID RenderingDevice::draw_list_begin(RID p_framebuffer,
 	if (p_region != Rect2() && p_region != Rect2(Vector2(), viewport_size)) { // Check custom region.
 		Rect2i viewport(viewport_offset, viewport_size);
 		Rect2i regioni = p_region;
-		if (!(regioni.position.x >= viewport.position.x) && (regioni.position.y >= viewport.position.y) &&
-				((regioni.position.x + regioni.size.x) <= (viewport.position.x + viewport.size.x)) &&
-				((regioni.position.y + regioni.size.y) <= (viewport.position.y + viewport.size.y))) {
+		if (!((regioni.position.x >= viewport.position.x) && (regioni.position.y >= viewport.position.y) &&
+					((regioni.position.x + regioni.size.x) <= (viewport.position.x + viewport.size.x)) &&
+					((regioni.position.y + regioni.size.y) <= (viewport.position.y + viewport.size.y)))) {
 			ERR_FAIL_V_MSG(INVALID_ID, "When supplying a custom region, it must be contained within the framebuffer rectangle");
 		}
 
@@ -3774,13 +3774,13 @@ void RenderingDevice::draw_list_draw(DrawListID p_list, bool p_use_indices, uint
 
 #endif
 
-	// Bind descriptor sets.
-
+#ifdef DEBUG_ENABLED
 	for (uint32_t i = 0; i < dl->state.set_count; i++) {
 		if (dl->state.sets[i].pipeline_expected_format == 0) {
-			continue; // Nothing expected by this pipeline.
+			// Nothing expected by this pipeline.
+			continue;
 		}
-#ifdef DEBUG_ENABLED
+
 		if (dl->state.sets[i].pipeline_expected_format != dl->state.sets[i].uniform_set_format) {
 			if (dl->state.sets[i].uniform_set_format == 0) {
 				ERR_FAIL_MSG("Uniforms were never supplied for set (" + itos(i) + ") at the time of drawing, which are required by the pipeline.");
@@ -3791,9 +3791,22 @@ void RenderingDevice::draw_list_draw(DrawListID p_list, bool p_use_indices, uint
 				ERR_FAIL_MSG("Uniforms supplied for set (" + itos(i) + ", which was just freed) are not the same format as required by the pipeline shader. Pipeline shader requires the following bindings:\n" + _shader_uniform_debug(dl->state.pipeline_shader));
 			}
 		}
+	}
 #endif
-		draw_graph.add_draw_list_uniform_set_prepare_for_use(dl->state.pipeline_shader_driver_id, dl->state.sets[i].uniform_set_driver_id, i);
+
+	// Prepare descriptor sets if the API doesn't use pipeline barriers.
+	if (!driver->api_trait_get(RDD::API_TRAIT_HONORS_PIPELINE_BARRIERS)) {
+		for (uint32_t i = 0; i < dl->state.set_count; i++) {
+			if (dl->state.sets[i].pipeline_expected_format == 0) {
+				// Nothing expected by this pipeline.
+				continue;
+			}
+
+			draw_graph.add_draw_list_uniform_set_prepare_for_use(dl->state.pipeline_shader_driver_id, dl->state.sets[i].uniform_set_driver_id, i);
+		}
 	}
+
+	// Bind descriptor sets.
 	for (uint32_t i = 0; i < dl->state.set_count; i++) {
 		if (dl->state.sets[i].pipeline_expected_format == 0) {
 			continue; // Nothing expected by this pipeline.
@@ -3858,6 +3871,8 @@ void RenderingDevice::draw_list_draw(DrawListID p_list, bool p_use_indices, uint
 
 		draw_graph.add_draw_list_draw(to_draw, p_instances);
 	}
+
+	dl->state.draw_count++;
 }
 
 void RenderingDevice::draw_list_enable_scissor(DrawListID p_list, const Rect2 &p_rect) {
@@ -4165,13 +4180,13 @@ void RenderingDevice::compute_list_dispatch(ComputeListID p_list, uint32_t p_x_g
 
 #endif
 
-	// Bind descriptor sets.
-
+#ifdef DEBUG_ENABLED
 	for (uint32_t i = 0; i < cl->state.set_count; i++) {
 		if (cl->state.sets[i].pipeline_expected_format == 0) {
-			continue; // Nothing expected by this pipeline.
+			// Nothing expected by this pipeline.
+			continue;
 		}
-#ifdef DEBUG_ENABLED
+
 		if (cl->state.sets[i].pipeline_expected_format != cl->state.sets[i].uniform_set_format) {
 			if (cl->state.sets[i].uniform_set_format == 0) {
 				ERR_FAIL_MSG("Uniforms were never supplied for set (" + itos(i) + ") at the time of drawing, which are required by the pipeline.");
@@ -4182,9 +4197,22 @@ void RenderingDevice::compute_list_dispatch(ComputeListID p_list, uint32_t p_x_g
 				ERR_FAIL_MSG("Uniforms supplied for set (" + itos(i) + ", which was just freed) are not the same format as required by the pipeline shader. Pipeline shader requires the following bindings:\n" + _shader_uniform_debug(cl->state.pipeline_shader));
 			}
 		}
+	}
 #endif
-		draw_graph.add_compute_list_uniform_set_prepare_for_use(cl->state.pipeline_shader_driver_id, cl->state.sets[i].uniform_set_driver_id, i);
+
+	// Prepare descriptor sets if the API doesn't use pipeline barriers.
+	if (!driver->api_trait_get(RDD::API_TRAIT_HONORS_PIPELINE_BARRIERS)) {
+		for (uint32_t i = 0; i < cl->state.set_count; i++) {
+			if (cl->state.sets[i].pipeline_expected_format == 0) {
+				// Nothing expected by this pipeline.
+				continue;
+			}
+
+			draw_graph.add_compute_list_uniform_set_prepare_for_use(cl->state.pipeline_shader_driver_id, cl->state.sets[i].uniform_set_driver_id, i);
+		}
 	}
+
+	// Bind descriptor sets.
 	for (uint32_t i = 0; i < cl->state.set_count; i++) {
 		if (cl->state.sets[i].pipeline_expected_format == 0) {
 			continue; // Nothing expected by this pipeline.
@@ -4201,6 +4229,7 @@ void RenderingDevice::compute_list_dispatch(ComputeListID p_list, uint32_t p_x_g
 	}
 
 	draw_graph.add_compute_list_dispatch(p_x_groups, p_y_groups, p_z_groups);
+	cl->state.dispatch_count++;
 }
 
 void RenderingDevice::compute_list_dispatch_threads(ComputeListID p_list, uint32_t p_x_threads, uint32_t p_y_threads, uint32_t p_z_threads) {
@@ -4258,16 +4287,16 @@ void RenderingDevice::compute_list_dispatch_indirect(ComputeListID p_list, RID p
 
 #endif
 
-	// Bind descriptor sets.
-
+#ifdef DEBUG_ENABLED
 	for (uint32_t i = 0; i < cl->state.set_count; i++) {
 		if (cl->state.sets[i].pipeline_expected_format == 0) {
-			continue; // Nothing expected by this pipeline.
+			// Nothing expected by this pipeline.
+			continue;
 		}
-#ifdef DEBUG_ENABLED
+
 		if (cl->state.sets[i].pipeline_expected_format != cl->state.sets[i].uniform_set_format) {
 			if (cl->state.sets[i].uniform_set_format == 0) {
-				ERR_FAIL_MSG("Uniforms were never supplied for set (" + itos(i) + ") at the time of drawing, which are required by the pipeline");
+				ERR_FAIL_MSG("Uniforms were never supplied for set (" + itos(i) + ") at the time of drawing, which are required by the pipeline.");
 			} else if (uniform_set_owner.owns(cl->state.sets[i].uniform_set)) {
 				UniformSet *us = uniform_set_owner.get_or_null(cl->state.sets[i].uniform_set);
 				ERR_FAIL_MSG("Uniforms supplied for set (" + itos(i) + "):\n" + _shader_uniform_debug(us->shader_id, us->shader_set) + "\nare not the same format as required by the pipeline shader. Pipeline shader requires the following bindings:\n" + _shader_uniform_debug(cl->state.pipeline_shader));
@@ -4275,9 +4304,22 @@ void RenderingDevice::compute_list_dispatch_indirect(ComputeListID p_list, RID p
 				ERR_FAIL_MSG("Uniforms supplied for set (" + itos(i) + ", which was just freed) are not the same format as required by the pipeline shader. Pipeline shader requires the following bindings:\n" + _shader_uniform_debug(cl->state.pipeline_shader));
 			}
 		}
+	}
 #endif
-		draw_graph.add_compute_list_uniform_set_prepare_for_use(cl->state.pipeline_shader_driver_id, cl->state.sets[i].uniform_set_driver_id, i);
+
+	// Prepare descriptor sets if the API doesn't use pipeline barriers.
+	if (!driver->api_trait_get(RDD::API_TRAIT_HONORS_PIPELINE_BARRIERS)) {
+		for (uint32_t i = 0; i < cl->state.set_count; i++) {
+			if (cl->state.sets[i].pipeline_expected_format == 0) {
+				// Nothing expected by this pipeline.
+				continue;
+			}
+
+			draw_graph.add_compute_list_uniform_set_prepare_for_use(cl->state.pipeline_shader_driver_id, cl->state.sets[i].uniform_set_driver_id, i);
+		}
 	}
+
+	// Bind descriptor sets.
 	for (uint32_t i = 0; i < cl->state.set_count; i++) {
 		if (cl->state.sets[i].pipeline_expected_format == 0) {
 			continue; // Nothing expected by this pipeline.
@@ -4294,6 +4336,7 @@ void RenderingDevice::compute_list_dispatch_indirect(ComputeListID p_list, RID p
 	}
 
 	draw_graph.add_compute_list_dispatch_indirect(buffer->driver_id, p_offset);
+	cl->state.dispatch_count++;
 
 	if (buffer->draw_tracker != nullptr) {
 		draw_graph.add_compute_list_usage(buffer->draw_tracker, RDG::RESOURCE_USAGE_INDIRECT_BUFFER_READ);
@@ -4694,6 +4737,10 @@ String RenderingDevice::get_device_api_name() const {
 	return driver->get_api_name();
 }
 
+bool RenderingDevice::is_composite_alpha_supported() const {
+	return driver->is_composite_alpha_supported(main_queue);
+}
+
 String RenderingDevice::get_device_api_version() const {
 	return driver->get_api_version();
 }
@@ -4873,25 +4920,78 @@ void RenderingDevice::_end_frame() {
 		ERR_PRINT("Found open compute list at the end of the frame, this should never happen (further compute will likely not work).");
 	}
 
-	draw_graph.end(frames[frame].draw_command_buffer, RENDER_GRAPH_REORDER, RENDER_GRAPH_FULL_BARRIERS);
 	driver->command_buffer_end(frames[frame].setup_command_buffer);
-	driver->command_buffer_end(frames[frame].draw_command_buffer);
+
+	// The command buffer must be copied into a stack variable as the driver workarounds can change the command buffer in use.
+	RDD::CommandBufferID command_buffer = frames[frame].draw_command_buffer;
+	draw_graph.end(RENDER_GRAPH_REORDER, RENDER_GRAPH_FULL_BARRIERS, command_buffer, frames[frame].command_buffer_pool);
+	driver->command_buffer_end(command_buffer);
 	driver->end_segment();
 }
 
 void RenderingDevice::_execute_frame(bool p_present) {
+	// Check whether this frame should present the swap chains and in which queue.
 	const bool frame_can_present = p_present && !frames[frame].swap_chains_to_present.is_empty();
 	const bool separate_present_queue = main_queue != present_queue;
-	const VectorView<RDD::SemaphoreID> execute_draw_semaphore = frame_can_present && separate_present_queue ? frames[frame].draw_semaphore : VectorView<RDD::SemaphoreID>();
-	const VectorView<RDD::SwapChainID> execute_draw_swap_chains = frame_can_present && !separate_present_queue ? frames[frame].swap_chains_to_present : VectorView<RDD::SwapChainID>();
+	thread_local LocalVector<RDD::SwapChainID> swap_chains;
+	swap_chains.clear();
+
+	// Execute the setup command buffer.
 	driver->command_queue_execute_and_present(main_queue, {}, frames[frame].setup_command_buffer, frames[frame].setup_semaphore, {}, {});
-	driver->command_queue_execute_and_present(main_queue, frames[frame].setup_semaphore, frames[frame].draw_command_buffer, execute_draw_semaphore, frames[frame].draw_fence, execute_draw_swap_chains);
+
+	// Execute command buffers and use semaphores to wait on the execution of the previous one. Normally there's only one command buffer,
+	// but driver workarounds can force situations where there'll be more.
+	uint32_t command_buffer_count = 1;
+	RDG::CommandBufferPool &buffer_pool = frames[frame].command_buffer_pool;
+	if (buffer_pool.buffers_used > 0) {
+		command_buffer_count += buffer_pool.buffers_used;
+		buffer_pool.buffers_used = 0;
+	}
+
+	RDD::SemaphoreID wait_semaphore = frames[frame].setup_semaphore;
+	for (uint32_t i = 0; i < command_buffer_count; i++) {
+		RDD::CommandBufferID command_buffer;
+		RDD::SemaphoreID signal_semaphore;
+		RDD::FenceID signal_fence;
+		if (i > 0) {
+			command_buffer = buffer_pool.buffers[i - 1];
+			signal_semaphore = buffer_pool.semaphores[i - 1];
+		} else {
+			command_buffer = frames[frame].draw_command_buffer;
+			signal_semaphore = frames[frame].draw_semaphore;
+		}
+
+		bool signal_semaphore_valid;
+		if (i == (command_buffer_count - 1)) {
+			// This is the last command buffer, it should signal the fence.
+			signal_fence = frames[frame].draw_fence;
+			signal_semaphore_valid = false;
+
+			if (frame_can_present && separate_present_queue) {
+				// The semaphore is required if the frame can be presented and a separate present queue is used.
+				signal_semaphore_valid = true;
+			} else if (frame_can_present) {
+				// Just present the swap chains as part of the last command execution.
+				swap_chains = frames[frame].swap_chains_to_present;
+			}
+		} else {
+			// Semaphores always need to be signaled if it's not the last command buffer.
+			signal_semaphore_valid = true;
+		}
+
+		driver->command_queue_execute_and_present(main_queue, wait_semaphore, command_buffer, signal_semaphore_valid ? signal_semaphore : VectorView<RDD::SemaphoreID>(), signal_fence, swap_chains);
+
+		// Make the next command buffer wait on the semaphore signaled by this one.
+		wait_semaphore = signal_semaphore;
+	}
+
+	// Indicate the fence has been signaled so the next time the frame's contents need to be used, the CPU needs to wait on the work to be completed.
 	frames[frame].draw_fence_signaled = true;
 
 	if (frame_can_present) {
 		if (separate_present_queue) {
 			// Issue the presentation separately if the presentation queue is different from the main queue.
-			driver->command_queue_execute_and_present(present_queue, frames[frame].draw_semaphore, {}, {}, {}, frames[frame].swap_chains_to_present);
+			driver->command_queue_execute_and_present(present_queue, wait_semaphore, {}, {}, {}, frames[frame].swap_chains_to_present);
 		}
 
 		frames[frame].swap_chains_to_present.clear();
@@ -5040,6 +5140,9 @@ Error RenderingDevice::initialize(RenderingContextDriver *p_context, DisplayServ
 		frames[i].timestamp_cpu_result_values.resize(max_timestamp_query_elements);
 		frames[i].timestamp_result_values.resize(max_timestamp_query_elements);
 		frames[i].timestamp_result_count = 0;
+
+		// Assign the main queue family and command pool to the command buffer pool.
+		frames[i].command_buffer_pool.pool = frames[i].command_pool;
 	}
 
 	// Start from frame count, so everything else is immediately old.
@@ -5051,7 +5154,7 @@ Error RenderingDevice::initialize(RenderingContextDriver *p_context, DisplayServ
 	driver->command_buffer_begin(frames[0].draw_command_buffer);
 
 	// Create draw graph and start it initialized as well.
-	draw_graph.initialize(driver, frames.size(), main_queue_family, SECONDARY_COMMAND_BUFFERS_PER_FRAME);
+	draw_graph.initialize(driver, device, frames.size(), main_queue_family, SECONDARY_COMMAND_BUFFERS_PER_FRAME);
 	draw_graph.begin();
 
 	for (uint32_t i = 0; i < frames.size(); i++) {
@@ -5206,8 +5309,8 @@ void RenderingDevice::_free_rids(T &p_owner, const char *p_type) {
 }
 
 void RenderingDevice::capture_timestamp(const String &p_name) {
-	ERR_FAIL_COND_MSG(draw_list != nullptr, "Capturing timestamps during draw list creation is not allowed. Offending timestamp was: " + p_name);
-	ERR_FAIL_COND_MSG(compute_list != nullptr, "Capturing timestamps during compute list creation is not allowed. Offending timestamp was: " + p_name);
+	ERR_FAIL_COND_MSG(draw_list != nullptr && draw_list->state.draw_count > 0, "Capturing timestamps during draw list creation is not allowed. Offending timestamp was: " + p_name);
+	ERR_FAIL_COND_MSG(compute_list != nullptr && compute_list->state.dispatch_count > 0, "Capturing timestamps during compute list creation is not allowed. Offending timestamp was: " + p_name);
 	ERR_FAIL_COND(frames[frame].timestamp_count >= max_timestamp_query_elements);
 
 	draw_graph.add_capture_timestamp(frames[frame].timestamp_pool, frames[frame].timestamp_count);
@@ -5384,6 +5487,11 @@ void RenderingDevice::finalize() {
 		driver->semaphore_free(frames[i].setup_semaphore);
 		driver->semaphore_free(frames[i].draw_semaphore);
 		driver->fence_free(frames[i].draw_fence);
+
+		RDG::CommandBufferPool &buffer_pool = frames[i].command_buffer_pool;
+		for (uint32_t j = 0; j < buffer_pool.buffers.size(); j++) {
+			driver->semaphore_free(buffer_pool.semaphores[j]);
+		}
 	}
 
 	if (pipeline_cache_enabled) {