Merge branch 'main' into mat_bool_operator_tests

Author: farzonl

.github/workflows/pr-matrix.yaml

@@ -32,6 +32,7 @@ jobs:
       SKU: ${{ matrix.SKU }}
       TestTarget: ${{ matrix.TestTarget }}
       OffloadTest-branch: ${{ github.event.pull_request.head.sha }}
+      SplitBuild: true
       LLVM-ExtraCMakeArgs: -DCMAKE_C_COMPILER=clang-cl -DCMAKE_CXX_COMPILER=clang-cl -DOFFLOADTEST_USE_CLANG_TIDY=On
 
   Exec-Tests-Windows-Warp:
@@ -50,6 +51,7 @@ jobs:
       SKU: ${{ matrix.SKU }}
       TestTarget: ${{ matrix.TestTarget }}
       OffloadTest-branch: ${{ github.event.pull_request.head.sha }}
+      SplitBuild: true
       LLVM-ExtraCMakeArgs: -DCMAKE_C_COMPILER=clang-cl -DCMAKE_CXX_COMPILER=clang-cl -DOFFLOADTEST_USE_CLANG_TIDY=On
 
   Exec-Tests-Extra:
@@ -72,6 +74,47 @@ jobs:
       SKU: ${{ matrix.SKU }}
       TestTarget: ${{ matrix.TestTarget }}
       OffloadTest-branch: ${{ github.event.pull_request.head.sha }}
+      SplitBuild: true
+      LLVM-ExtraCMakeArgs: -DCMAKE_C_COMPILER=clang-cl -DCMAKE_CXX_COMPILER=clang-cl -DOFFLOADTEST_USE_CLANG_TIDY=On
+
+  Exec-Tests-Extra-DX-All:
+    permissions:
+      contents: read
+      checks: write
+    if: contains( github.event.pull_request.labels.*.name, 'test-dx-all')
+    strategy:
+      fail-fast: false
+      matrix:
+        SKU: [windows-amd, windows-qc]
+        TestTarget: [check-hlsl-d3d12, check-hlsl-clang-d3d12]
+
+    uses: ./.github/workflows/build-and-test-callable.yaml
+    with:
+      OS: windows
+      SKU: ${{ matrix.SKU }}
+      TestTarget: ${{ matrix.TestTarget }}
+      OffloadTest-branch: ${{ github.event.pull_request.head.sha }}
+      SplitBuild: true
+      LLVM-ExtraCMakeArgs: -DCMAKE_C_COMPILER=clang-cl -DCMAKE_CXX_COMPILER=clang-cl -DOFFLOADTEST_USE_CLANG_TIDY=On
+
+  Exec-Tests-Extra-VK-All:
+    permissions:
+      contents: read
+      checks: write
+    if: contains( github.event.pull_request.labels.*.name, 'test-vk-all')
+    strategy:
+      fail-fast: false
+      matrix:
+        SKU: [windows-amd, windows-nvidia, windows-qc]
+        TestTarget: [check-hlsl-vk, check-hlsl-clang-vk]
+
+    uses: ./.github/workflows/build-and-test-callable.yaml
+    with:
+      OS: windows
+      SKU: ${{ matrix.SKU }}
+      TestTarget: ${{ matrix.TestTarget }}
+      OffloadTest-branch: ${{ github.event.pull_request.head.sha }}
+      SplitBuild: true
       LLVM-ExtraCMakeArgs: -DCMAKE_C_COMPILER=clang-cl -DCMAKE_CXX_COMPILER=clang-cl -DOFFLOADTEST_USE_CLANG_TIDY=On
 
   Exec-Tests-MacOS:

include/API/AccelerationStructure.h

@@ -14,13 +14,26 @@
 #include "API/Resources.h"
 
 #include "llvm/ADT/ArrayRef.h"
+#include "llvm/ADT/BitmaskEnum.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/Support/Error.h"
 
 #include <cstdint>
 #include <variant>
 
 namespace offloadtest {
+LLVM_ENABLE_BITMASK_ENUMS_IN_NAMESPACE();
+
+// Bit values match D3D12_RAYTRACING_INSTANCE_FLAGS, VkGeometryInstanceFlagBits-
+// KHR, and MTLAccelerationStructureInstanceOptions so backends can pass the
+// value through unchanged.
+enum AccelerationStructureInstanceFlags : uint32_t {
+  InstanceFlagNone = 0,
+  TriangleCullDisable = 1 << 0,
+  TriangleFrontCounterclockwise = 1 << 1,
+  ForceOpaque = 1 << 2,
+  ForceNonOpaque = 1 << 3,
+};
 
 struct AccelerationStructureSizes {
   uint64_t ResultDataMaxSizeInBytes = 0;
@@ -57,6 +70,7 @@ struct AccelerationStructureInstance {
   uint8_t InstanceMask = 0xFF;
   // 24-bit; high bits are truncated by the backend to match DXR's bitfield.
   uint32_t InstanceContributionToHitGroupIndex = 0;
+  AccelerationStructureInstanceFlags Flags = InstanceFlagNone;
   AccelerationStructure *BLAS = nullptr;
 };
 
@@ -178,4 +192,9 @@ class AccelerationStructure {
 
 } // namespace offloadtest
 
+namespace llvm {
+LLVM_DECLARE_ENUM_AS_BITMASK(::offloadtest::AccelerationStructureInstanceFlags,
+                             ::offloadtest::ForceNonOpaque);
+} // namespace llvm
+
 #endif // OFFLOADTEST_API_ACCELERATIONSTRUCTURE_H

include/Support/Pipeline.h

@@ -13,6 +13,7 @@
 #ifndef OFFLOADTEST_SUPPORT_PIPELINE_H
 #define OFFLOADTEST_SUPPORT_PIPELINE_H
 
+#include "API/AccelerationStructure.h"
 #include "API/Enums.h"
 #include "API/Resources.h"
 #include "llvm/ADT/SmallVector.h"
@@ -574,6 +575,7 @@ struct InstanceDesc {
   uint32_t InstanceID = 0;
   uint8_t InstanceMask = 0xFF;
   uint32_t InstanceContributionToHitGroupIndex = 0;
+  AccelerationStructureInstanceFlags Flags = InstanceFlagNone;
 };
 
 struct TLASDesc {
@@ -850,6 +852,19 @@ template <> struct MappingTraits<offloadtest::ShaderBindingTableDesc> {
   static void mapping(IO &I, offloadtest::ShaderBindingTableDesc &S);
 };
 
+template <>
+struct ScalarBitSetTraits<offloadtest::AccelerationStructureInstanceFlags> {
+  static void bitset(IO &I,
+                     offloadtest::AccelerationStructureInstanceFlags &V) {
+#define BIT_CASE(Val) I.bitSetCase(V, #Val, offloadtest::Val)
+    BIT_CASE(TriangleCullDisable);
+    BIT_CASE(TriangleFrontCounterclockwise);
+    BIT_CASE(ForceOpaque);
+    BIT_CASE(ForceNonOpaque);
+#undef BIT_CASE
+  }
+};
+
 template <> struct ScalarEnumerationTraits<offloadtest::Rule> {
   static void enumeration(IO &I, offloadtest::Rule &V) {
 #define ENUM_CASE(Val) I.enumCase(V, #Val, offloadtest::Rule::Val)