Merge pull request #2066 from billhollings/EXT_extended_dynamic_state3

Add support for VK_EXT_extended_dynamic_state3 extension.

Docs/MoltenVK_Runtime_UserGuide.md

@@ -375,6 +375,8 @@ In addition to core *Vulkan* functionality, **MoltenVK**  also supports the foll
   - *Requires Metal 3.1 for `VK_DYNAMIC_STATE_VERTEX_INPUT_BINDING_STRIDE`.*
 - `VK_EXT_extended_dynamic_state2`
   - *Primitive restart is always enabled, as Metal does not support disabling it (`VK_DYNAMIC_STATE_PRIMITIVE_RESTART_ENABLE_EXT`).*
+- `VK_EXT_extended_dynamic_state3`
+  - *Metal does not support `VK_POLYGON_MODE_POINT`*
 - `VK_EXT_external_memory_host`
 - `VK_EXT_fragment_shader_interlock`
   - *Requires Metal 2.0 and Raster Order Groups.*

Docs/Whats_New.md

@@ -18,6 +18,10 @@ MoltenVK 1.2.7
 
 Released TBD
 
+- Add support for extensions:
+	- `VK_EXT_extended_dynamic_state3` *(Metal does not support `VK_POLYGON_MODE_POINT`)*
+- Fix regression that broke `VK_POLYGON_MODE_LINE`.
+- Fix regression in marking rendering state dirty after `vkCmdClearAttachments()`.
 - Reduce disk space consumed after running `fetchDependencies` script by removing intermediate file caches.
 - Fix rare deadlock during launch via `dlopen()`.
 - Update to latest SPIRV-Cross:

MoltenVK/MoltenVK/Commands/MVKCmdDraw.mm

@@ -144,10 +144,9 @@
 
 void MVKCmdDraw::encode(MVKCommandEncoder* cmdEncoder) {
 
-    if (_vertexCount == 0 || _instanceCount == 0) {
-        // Nothing to do.
-        return;
-    }
+	if (_vertexCount == 0 || _instanceCount == 0) { return; }	// Nothing to do.
+
+	cmdEncoder->restartMetalRenderPassIfNeeded();
 
 	auto* pipeline = cmdEncoder->_graphicsPipelineState.getGraphicsPipeline();
 
@@ -172,7 +171,7 @@
 	} tessParams;
     uint32_t outControlPointCount = 0;
     if (pipeline->isTessellationPipeline()) {
-        tessParams.inControlPointCount = cmdEncoder->_graphicsPipelineState.getPatchControlPoints();
+        tessParams.inControlPointCount = cmdEncoder->_renderingState.getPatchControlPoints();
         outControlPointCount = pipeline->getOutputControlPointCount();
         tessParams.patchCount = mvkCeilingDivide(_vertexCount, tessParams.inControlPointCount) * _instanceCount;
     }
@@ -369,10 +368,9 @@
 
 void MVKCmdDrawIndexed::encode(MVKCommandEncoder* cmdEncoder) {
 
-    if (_indexCount == 0 || _instanceCount == 0) {
-        // Nothing to do.
-        return;
-    }
+	if (_indexCount == 0 || _instanceCount == 0) { return; }	// Nothing to do.
+
+	cmdEncoder->restartMetalRenderPassIfNeeded();
 
 	auto* pipeline = cmdEncoder->_graphicsPipelineState.getGraphicsPipeline();
 
@@ -401,7 +399,7 @@
 	} tessParams;
     uint32_t outControlPointCount = 0;
     if (pipeline->isTessellationPipeline()) {
-        tessParams.inControlPointCount = cmdEncoder->_graphicsPipelineState.getPatchControlPoints();
+        tessParams.inControlPointCount = cmdEncoder->_renderingState.getPatchControlPoints();
         outControlPointCount = pipeline->getOutputControlPointCount();
         tessParams.patchCount = mvkCeilingDivide(_indexCount, tessParams.inControlPointCount) * _instanceCount;
     }
@@ -649,6 +647,8 @@
 
 void MVKCmdDrawIndirect::encode(MVKCommandEncoder* cmdEncoder) {
 
+	cmdEncoder->restartMetalRenderPassIfNeeded();
+
 	auto* pipeline = cmdEncoder->_graphicsPipelineState.getGraphicsPipeline();
 
 	// Metal doesn't support triangle fans, so encode it as indexed indirect triangles instead.
@@ -686,7 +686,7 @@
         // encoding and execution. So we don't know how big to make the buffers.
         // We must assume an arbitrarily large number of vertices may be submitted.
         // But not too many, or we'll exhaust available VRAM.
-        inControlPointCount = cmdEncoder->_graphicsPipelineState.getPatchControlPoints();
+        inControlPointCount = cmdEncoder->_renderingState.getPatchControlPoints();
         outControlPointCount = pipeline->getOutputControlPointCount();
         vertexCount = kMVKMaxDrawIndirectVertexCount;
         patchCount = mvkCeilingDivide(vertexCount, inControlPointCount);
@@ -990,6 +990,7 @@
 }
 
 void MVKCmdDrawIndexedIndirect::encode(MVKCommandEncoder* cmdEncoder) {
+	cmdEncoder->restartMetalRenderPassIfNeeded();
 	encode(cmdEncoder, cmdEncoder->_graphicsResourcesState._mtlIndexBufferBinding);
 }
 
@@ -1034,7 +1035,7 @@
         // encoding and execution. So we don't know how big to make the buffers.
         // We must assume an arbitrarily large number of vertices may be submitted.
         // But not too many, or we'll exhaust available VRAM.
-        inControlPointCount = cmdEncoder->_graphicsPipelineState.getPatchControlPoints();
+        inControlPointCount = cmdEncoder->_renderingState.getPatchControlPoints();
         outControlPointCount = pipeline->getOutputControlPointCount();
         vertexCount = kMVKMaxDrawIndirectVertexCount;
         patchCount = mvkCeilingDivide(vertexCount, inControlPointCount);

MoltenVK/MoltenVK/Commands/MVKCmdRendering.h

@@ -46,7 +46,6 @@ class MVKCmdBeginRenderPassBase : public MVKCommand {
 
 protected:
 
-	MVKSmallVector<MVKSmallVector<MTLSamplePosition>> _subpassSamplePositions;
 	MVKRenderPass* _renderPass;
 	MVKFramebuffer* _framebuffer;
 	VkRect2D _renderArea;
@@ -203,7 +202,26 @@ class MVKCmdSetSampleLocations : public MVKCommand {
 protected:
 	MVKCommandTypePool<MVKCommand>* getTypePool(MVKCommandPool* cmdPool) override;
 
-	MVKSmallVector<MTLSamplePosition, 8> _samplePositions;
+	MVKSmallVector<VkSampleLocationEXT, kMVKMaxSampleCount> _sampleLocations;
+};
+
+
+#pragma mark -
+#pragma mark MVKCmdSetSampleLocationsEnable
+
+/** Vulkan command to dynamically enable custom sample locations. */
+class MVKCmdSetSampleLocationsEnable : public MVKCommand {
+
+public:
+	VkResult setContent(MVKCommandBuffer* cmdBuff,
+						VkBool32 sampleLocationsEnable);
+
+	void encode(MVKCommandEncoder* cmdEncoder) override;
+
+protected:
+	MVKCommandTypePool<MVKCommand>* getTypePool(MVKCommandPool* cmdPool) override;
+
+	VkBool32 _sampleLocationsEnable;
 };
 
 
@@ -366,6 +384,25 @@ class MVKCmdSetDepthWriteEnable : public MVKCommand {
 };
 
 
+#pragma mark -
+#pragma mark MVKCmdSetDepthClipEnable
+
+/** Vulkan command to dynamically enable depth clip. */
+class MVKCmdSetDepthClipEnable : public MVKCommand {
+
+public:
+	VkResult setContent(MVKCommandBuffer* cmdBuff,
+						VkBool32 depthClipEnable);
+
+	void encode(MVKCommandEncoder* cmdEncoder) override;
+
+protected:
+	MVKCommandTypePool<MVKCommand>* getTypePool(MVKCommandPool* cmdPool) override;
+
+	VkBool32 _depthClipEnable;
+};
+
+
 #pragma mark -
 #pragma mark MVKCmdSetDepthCompareOp
 
@@ -551,6 +588,25 @@ class MVKCmdSetPatchControlPoints : public MVKCommand {
 };
 
 
+#pragma mark -
+#pragma mark MVKCmdSetPolygonMode
+
+/** Vulkan command to dynamically set the polygon mode. */
+class MVKCmdSetPolygonMode : public MVKCommand {
+
+public:
+	VkResult setContent(MVKCommandBuffer* cmdBuff,
+						VkPolygonMode polygonMode);
+
+	void encode(MVKCommandEncoder* cmdEncoder) override;
+
+protected:
+	MVKCommandTypePool<MVKCommand>* getTypePool(MVKCommandPool* cmdPool) override;
+
+	VkPolygonMode _polygonMode;
+};
+
+
 #pragma mark -
 #pragma mark MVKCmdSetPrimitiveTopology
 

MoltenVK/MoltenVK/Commands/MVKCmdRendering.mm

@@ -36,30 +36,6 @@
 	_renderPass = (MVKRenderPass*)pRenderPassBegin->renderPass;
 	_framebuffer = (MVKFramebuffer*)pRenderPassBegin->framebuffer;
 	_renderArea = pRenderPassBegin->renderArea;
-	_subpassSamplePositions.clear();
-
-	for (const auto* next = (VkBaseInStructure*)pRenderPassBegin->pNext; next; next = next->pNext) {
-		switch (next->sType) {
-			case VK_STRUCTURE_TYPE_RENDER_PASS_SAMPLE_LOCATIONS_BEGIN_INFO_EXT: {
-				// Build an array of arrays, one array of sample positions for each subpass index.
-				// For subpasses not included in VkRenderPassSampleLocationsBeginInfoEXT, the resulting array of samples will be empty.
-				_subpassSamplePositions.resize(_renderPass->getSubpassCount());
-				auto* pRPSampLocnsInfo = (VkRenderPassSampleLocationsBeginInfoEXT*)next;
-				for (uint32_t spSLIdx = 0; spSLIdx < pRPSampLocnsInfo->postSubpassSampleLocationsCount; spSLIdx++) {
-					auto& spsl = pRPSampLocnsInfo->pPostSubpassSampleLocations[spSLIdx];
-					uint32_t spIdx = spsl.subpassIndex;
-					auto& spSampPosns = _subpassSamplePositions[spIdx];
-					for (uint32_t slIdx = 0; slIdx < spsl.sampleLocationsInfo.sampleLocationsCount; slIdx++) {
-						auto& sl = spsl.sampleLocationsInfo.pSampleLocations[slIdx];
-						spSampPosns.push_back(MTLSamplePositionMake(sl.x, sl.y));
-					}
-				}
-				break;
-			}
-			default:
-				break;
-		}
-	}
 
 	cmdBuff->_currentSubpassInfo.beginRenderpass(_renderPass);
 
@@ -86,23 +62,14 @@
 
 template <size_t N_CV, size_t N_A>
 void MVKCmdBeginRenderPass<N_CV, N_A>::encode(MVKCommandEncoder* cmdEncoder) {
-
-	// Convert the sample position array of arrays to an array of array-references,
-	// so that it can be passed to the command encoder.
-	size_t spSPCnt = _subpassSamplePositions.size();
-	MVKArrayRef<MTLSamplePosition> spSPRefs[spSPCnt];
-	for (uint32_t spSPIdx = 0; spSPIdx < spSPCnt; spSPIdx++) {
-		spSPRefs[spSPIdx] = _subpassSamplePositions[spSPIdx].contents();
-	}
-
 	cmdEncoder->beginRenderpass(this,
 								_contents,
 								_renderPass,
 								_framebuffer,
 								_renderArea,
 								_clearValues.contents(),
 								_attachments.contents(),
-								MVKArrayRef(spSPRefs, spSPCnt));
+								kMVKCommandUseBeginRenderPass);
 }
 
 template class MVKCmdBeginRenderPass<1, 0>;
@@ -217,17 +184,29 @@
 
 VkResult MVKCmdSetSampleLocations::setContent(MVKCommandBuffer* cmdBuff,
 											  const VkSampleLocationsInfoEXT* pSampleLocationsInfo) {
-
+	_sampleLocations.clear();
 	for (uint32_t slIdx = 0; slIdx < pSampleLocationsInfo->sampleLocationsCount; slIdx++) {
-		auto& sl = pSampleLocationsInfo->pSampleLocations[slIdx];
-		_samplePositions.push_back(MTLSamplePositionMake(sl.x, sl.y));
+		_sampleLocations.push_back(pSampleLocationsInfo->pSampleLocations[slIdx]);
 	}
-
 	return VK_SUCCESS;
 }
 
 void MVKCmdSetSampleLocations::encode(MVKCommandEncoder* cmdEncoder) {
-	cmdEncoder->setDynamicSamplePositions(_samplePositions.contents());
+	cmdEncoder->_renderingState.setSampleLocations(_sampleLocations.contents(), true);
+}
+
+
+#pragma mark -
+#pragma mark MVKCmdSetSampleLocationsEnable
+
+VkResult MVKCmdSetSampleLocationsEnable::setContent(MVKCommandBuffer* cmdBuff,
+													VkBool32 sampleLocationsEnable) {
+	_sampleLocationsEnable = sampleLocationsEnable;
+	return VK_SUCCESS;
+}
+
+void MVKCmdSetSampleLocationsEnable::encode(MVKCommandEncoder* cmdEncoder) {
+	cmdEncoder->_renderingState.setSampleLocationsEnable(_sampleLocationsEnable, true);
 }
 
 
@@ -240,7 +219,7 @@
 										  uint32_t viewportCount,
 										  const VkViewport* pViewports) {
 	_firstViewport = firstViewport;
-	_viewports.clear();	// Clear for reuse
+	_viewports.clear();
 	_viewports.reserve(viewportCount);
 	for (uint32_t vpIdx = 0; vpIdx < viewportCount; vpIdx++) {
 		_viewports.push_back(pViewports[vpIdx]);
@@ -267,7 +246,7 @@
 										 uint32_t scissorCount,
 										 const VkRect2D* pScissors) {
 	_firstScissor = firstScissor;
-	_scissors.clear();	// Clear for reuse
+	_scissors.clear();
 	_scissors.reserve(scissorCount);
 	for (uint32_t sIdx = 0; sIdx < scissorCount; sIdx++) {
 		_scissors.push_back(pScissors[sIdx]);
@@ -362,6 +341,20 @@
 }
 
 
+#pragma mark -
+#pragma mark MVKCmdSetDepthClipEnable
+
+VkResult MVKCmdSetDepthClipEnable::setContent(MVKCommandBuffer* cmdBuff,
+											  VkBool32 depthClipEnable) {
+	_depthClipEnable = depthClipEnable;
+	return VK_SUCCESS;
+}
+
+void MVKCmdSetDepthClipEnable::encode(MVKCommandEncoder* cmdEncoder) {
+	cmdEncoder->_renderingState.setDepthClipEnable(_depthClipEnable, true);
+}
+
+
 #pragma mark -
 #pragma mark MVKCmdSetDepthCompareOp
 
@@ -501,7 +494,21 @@
 }
 
 void MVKCmdSetPatchControlPoints::encode(MVKCommandEncoder* cmdEncoder) {
-	cmdEncoder->_graphicsPipelineState.setPatchControlPoints(_patchControlPoints);
+	cmdEncoder->_renderingState.setPatchControlPoints(_patchControlPoints, true);
+}
+
+
+#pragma mark -
+#pragma mark MVKCmdSetPolygonMode
+
+VkResult MVKCmdSetPolygonMode::setContent(MVKCommandBuffer* cmdBuff,
+										  VkPolygonMode polygonMode) {
+	_polygonMode = polygonMode;
+	return VK_SUCCESS;
+}
+
+void MVKCmdSetPolygonMode::encode(MVKCommandEncoder* cmdEncoder) {
+	cmdEncoder->_renderingState.setPolygonMode(_polygonMode, true);
 }
 
 

MoltenVK/MoltenVK/Commands/MVKCmdTransfer.mm

@@ -1506,8 +1506,10 @@ static inline MTLSize mvkClampMTLSize(MTLSize size, MTLOrigin origin, MTLSize ma
 
 	// Return to the previous rendering state on the next render activity
 	cmdEncoder->_graphicsPipelineState.markDirty();
+	cmdEncoder->_graphicsResourcesState.markDirty();
 	cmdEncoder->_depthStencilState.markDirty();
 	cmdEncoder->_renderingState.markDirty();
+	cmdEncoder->_occlusionQueryState.markDirty();
 }
 
 template <size_t N>