https://github.com/farzonl created https://github.com/llvm/llvm-project/pull/179861
fixed https://github.com/llvm/llvm-project/issues/179859 For matrix types we need to check the language mode so we can change the matrix memory layout to arrays of vectors. To make this play nice with how the rest of clang treats matrices we need to modify the MaybeConvertMatrixAddress and the CreateMemTemp function to know how to reconstruct a flattened vector. Rest of changes is just test updates. >From 16a78cbc4a7ce64412fc50001bdfb9b714f02c49 Mon Sep 17 00:00:00 2001 From: Farzon Lotfi <[email protected]> Date: Wed, 4 Feb 2026 23:54:39 -0500 Subject: [PATCH] [HLSL] Represent Matrix as arrays of vectors in memory fixed https://github.com/llvm/llvm-project/issues/179859 For matrix types we need to check the language mode so we can change the matrix memory layout to arrays of vectors. To make this play nice with how the rest of clang treats matrices we need to modify the MaybeConvertMatrixAddress and the CreateMemTemp function to know how to reconstruct a flattened vector. Rest of changes is just test updates. --- clang/lib/CodeGen/CGExpr.cpp | 20 +++++++++--- clang/lib/CodeGen/CodeGenTypes.cpp | 9 ++++-- .../BasicFeatures/MatrixElementTypeCast.hlsl | 28 ++++++++-------- .../MatrixExplicitTruncation.hlsl | 32 +++++++++---------- .../MatrixImplicitTruncation.hlsl | 30 ++++++++--------- .../MatrixSingleSubscriptConstSwizzle.hlsl | 4 +-- .../MatrixSingleSubscriptDynamicSwizzle.hlsl | 6 ++-- .../MatrixSingleSubscriptGetter.hlsl | 20 ++++++------ .../MatrixSingleSubscriptSetter.hlsl | 2 +- .../BasicFeatures/MatrixSplat.hlsl | 20 ++++++------ clang/test/CodeGenHLSL/BoolMatrix.hlsl | 20 ++++++------ clang/test/CodeGenHLSL/basic_types.hlsl | 32 +++++++++---------- 12 files changed, 120 insertions(+), 103 deletions(-) diff --git a/clang/lib/CodeGen/CGExpr.cpp b/clang/lib/CodeGen/CGExpr.cpp index 69b616af20c88..b1106284af82f 100644 --- a/clang/lib/CodeGen/CGExpr.cpp +++ b/clang/lib/CodeGen/CGExpr.cpp @@ -201,8 +201,14 @@ RawAddress CodeGenFunction::CreateMemTemp(QualType Ty, CharUnits Align, if (Ty->isConstantMatrixType()) { auto *ArrayTy = cast<llvm::ArrayType>(Result.getElementType()); - auto *VectorTy = llvm::FixedVectorType::get(ArrayTy->getElementType(), - ArrayTy->getNumElements()); + auto *ArrayElementTy = ArrayTy->getElementType(); + auto ArrayElements = ArrayTy->getNumElements(); + if (getContext().getLangOpts().HLSL) { + auto *VectorTy = cast<llvm::FixedVectorType>(ArrayElementTy); + ArrayElementTy = VectorTy->getElementType(); + ArrayElements *= VectorTy->getNumElements(); + } + auto *VectorTy = llvm::FixedVectorType::get(ArrayElementTy, ArrayElements); Result = Address(Result.getPointer(), VectorTy, Result.getAlignment(), KnownNonNull); @@ -2280,8 +2286,14 @@ static RawAddress MaybeConvertMatrixAddress(RawAddress Addr, bool IsVector = true) { auto *ArrayTy = dyn_cast<llvm::ArrayType>(Addr.getElementType()); if (ArrayTy && IsVector) { - auto *VectorTy = llvm::FixedVectorType::get(ArrayTy->getElementType(), - ArrayTy->getNumElements()); + auto ArrayElements = ArrayTy->getNumElements(); + auto *ArrayElementTy = ArrayTy->getElementType(); + if (CGF.getContext().getLangOpts().HLSL) { + auto *VectorTy = cast<llvm::FixedVectorType>(ArrayElementTy); + ArrayElementTy = VectorTy->getElementType(); + ArrayElements *= VectorTy->getNumElements(); + } + auto *VectorTy = llvm::FixedVectorType::get(ArrayElementTy, ArrayElements); return Addr.withElementType(VectorTy); } diff --git a/clang/lib/CodeGen/CodeGenTypes.cpp b/clang/lib/CodeGen/CodeGenTypes.cpp index 0e1131d586433..fd7a8929a9be9 100644 --- a/clang/lib/CodeGen/CodeGenTypes.cpp +++ b/clang/lib/CodeGen/CodeGenTypes.cpp @@ -105,8 +105,13 @@ llvm::Type *CodeGenTypes::ConvertTypeForMem(QualType T) { const Type *Ty = Context.getCanonicalType(T).getTypePtr(); const ConstantMatrixType *MT = cast<ConstantMatrixType>(Ty); llvm::Type *IRElemTy = ConvertType(MT->getElementType()); - if (Context.getLangOpts().HLSL && T->isConstantMatrixBoolType()) - IRElemTy = ConvertTypeForMem(Context.BoolTy); + if (Context.getLangOpts().HLSL) { + if (T->isConstantMatrixBoolType()) + IRElemTy = ConvertTypeForMem(Context.BoolTy); + llvm::Type *VecTy = + llvm::FixedVectorType::get(IRElemTy, MT->getNumColumns()); + return llvm::ArrayType::get(VecTy, MT->getNumRows()); + } return llvm::ArrayType::get(IRElemTy, MT->getNumElementsFlattened()); } diff --git a/clang/test/CodeGenHLSL/BasicFeatures/MatrixElementTypeCast.hlsl b/clang/test/CodeGenHLSL/BasicFeatures/MatrixElementTypeCast.hlsl index 3bd7636212862..f48edc19b86f7 100644 --- a/clang/test/CodeGenHLSL/BasicFeatures/MatrixElementTypeCast.hlsl +++ b/clang/test/CodeGenHLSL/BasicFeatures/MatrixElementTypeCast.hlsl @@ -5,8 +5,8 @@ // CHECK-LABEL: define hidden noundef <6 x i32> @_Z22elementwise_type_cast0u11matrix_typeILm3ELm2EfE( // CHECK-SAME: <6 x float> noundef nofpclass(nan inf) [[F32:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca [6 x float], align 4 -// CHECK-NEXT: [[I32:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca [3 x <2 x float>], align 4 +// CHECK-NEXT: [[I32:%.*]] = alloca [3 x <2 x i32>], align 4 // CHECK-NEXT: store <6 x float> [[F32]], ptr [[F32_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <6 x float>, ptr [[F32_ADDR]], align 4 // CHECK-NEXT: [[CONV:%.*]] = fptosi <6 x float> [[TMP0]] to <6 x i32> @@ -22,8 +22,8 @@ int3x2 elementwise_type_cast0(float3x2 f32) { // CHECK-LABEL: define hidden noundef <6 x i32> @_Z22elementwise_type_cast1u11matrix_typeILm3ELm2EsE( // CHECK-SAME: <6 x i16> noundef [[I16_32:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I16_32_ADDR:%.*]] = alloca [6 x i16], align 2 -// CHECK-NEXT: [[I32:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[I16_32_ADDR:%.*]] = alloca [3 x <2 x i16>], align 2 +// CHECK-NEXT: [[I32:%.*]] = alloca [3 x <2 x i32>], align 4 // CHECK-NEXT: store <6 x i16> [[I16_32]], ptr [[I16_32_ADDR]], align 2 // CHECK-NEXT: [[TMP0:%.*]] = load <6 x i16>, ptr [[I16_32_ADDR]], align 2 // CHECK-NEXT: [[CONV:%.*]] = sext <6 x i16> [[TMP0]] to <6 x i32> @@ -39,8 +39,8 @@ int3x2 elementwise_type_cast1(int16_t3x2 i16_32) { // CHECK-LABEL: define hidden noundef <6 x i32> @_Z22elementwise_type_cast2u11matrix_typeILm3ELm2ElE( // CHECK-SAME: <6 x i64> noundef [[I64_32:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I64_32_ADDR:%.*]] = alloca [6 x i64], align 8 -// CHECK-NEXT: [[I32:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[I64_32_ADDR:%.*]] = alloca [3 x <2 x i64>], align 8 +// CHECK-NEXT: [[I32:%.*]] = alloca [3 x <2 x i32>], align 4 // CHECK-NEXT: store <6 x i64> [[I64_32]], ptr [[I64_32_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load <6 x i64>, ptr [[I64_32_ADDR]], align 8 // CHECK-NEXT: [[CONV:%.*]] = trunc <6 x i64> [[TMP0]] to <6 x i32> @@ -56,8 +56,8 @@ int3x2 elementwise_type_cast2(int64_t3x2 i64_32) { // CHECK-LABEL: define hidden noundef <6 x i16> @_Z22elementwise_type_cast3u11matrix_typeILm2ELm3EDhE( // CHECK-SAME: <6 x half> noundef nofpclass(nan inf) [[H23:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[H23_ADDR:%.*]] = alloca [6 x half], align 2 -// CHECK-NEXT: [[I23:%.*]] = alloca [6 x i16], align 2 +// CHECK-NEXT: [[H23_ADDR:%.*]] = alloca [2 x <3 x half>], align 2 +// CHECK-NEXT: [[I23:%.*]] = alloca [2 x <3 x i16>], align 2 // CHECK-NEXT: store <6 x half> [[H23]], ptr [[H23_ADDR]], align 2 // CHECK-NEXT: [[TMP0:%.*]] = load <6 x half>, ptr [[H23_ADDR]], align 2 // CHECK-NEXT: [[CONV:%.*]] = fptosi <6 x half> [[TMP0]] to <6 x i16> @@ -73,8 +73,8 @@ int16_t2x3 elementwise_type_cast3(half2x3 h23) { // CHECK-LABEL: define hidden noundef <6 x i32> @_Z22elementwise_type_cast4u11matrix_typeILm3ELm2EdE( // CHECK-SAME: <6 x double> noundef nofpclass(nan inf) [[D32:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[D32_ADDR:%.*]] = alloca [6 x double], align 8 -// CHECK-NEXT: [[I32:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[D32_ADDR:%.*]] = alloca [3 x <2 x double>], align 8 +// CHECK-NEXT: [[I32:%.*]] = alloca [3 x <2 x i32>], align 4 // CHECK-NEXT: store <6 x double> [[D32]], ptr [[D32_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load <6 x double>, ptr [[D32_ADDR]], align 8 // CHECK-NEXT: [[CONV:%.*]] = fptosi <6 x double> [[TMP0]] to <6 x i32> @@ -91,7 +91,7 @@ int3x2 elementwise_type_cast4(double3x2 d32) { // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[A:%.*]] = alloca [2 x [1 x i32]], align 4 -// CHECK-NEXT: [[B:%.*]] = alloca [2 x i32], align 4 +// CHECK-NEXT: [[B:%.*]] = alloca [2 x <1 x i32>], align 4 // CHECK-NEXT: [[AGG_TEMP:%.*]] = alloca [2 x [1 x i32]], align 4 // CHECK-NEXT: [[FLATCAST_TMP:%.*]] = alloca <2 x i32>, align 4 // CHECK-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[A]], ptr align 4 @__const._Z5call2v.A, i32 8, i1 false) @@ -120,7 +120,7 @@ struct S { // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 1 -// CHECK-NEXT: [[A:%.*]] = alloca [2 x i32], align 4 +// CHECK-NEXT: [[A:%.*]] = alloca [2 x <1 x i32>], align 4 // CHECK-NEXT: [[AGG_TEMP:%.*]] = alloca [[STRUCT_S]], align 1 // CHECK-NEXT: [[FLATCAST_TMP:%.*]] = alloca <2 x i32>, align 4 // CHECK-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[S]], ptr align 1 @__const._Z5call3v.s, i32 8, i1 false) @@ -155,7 +155,7 @@ struct Derived : BFields { // CHECK-LABEL: define hidden void @_Z5call47Derived( // CHECK-SAME: ptr noundef byval([[STRUCT_DERIVED:%.*]]) align 1 [[D:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[A:%.*]] = alloca [4 x i32], align 4 +// CHECK-NEXT: [[A:%.*]] = alloca [2 x <2 x i32>], align 4 // CHECK-NEXT: [[AGG_TEMP:%.*]] = alloca [[STRUCT_DERIVED]], align 1 // CHECK-NEXT: [[FLATCAST_TMP:%.*]] = alloca <4 x i32>, align 4 // CHECK-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[AGG_TEMP]], ptr align 1 [[D]], i32 19, i1 false) @@ -189,7 +189,7 @@ void call4(Derived D) { // CHECK-SAME: <4 x float> noundef nofpclass(nan inf) [[V:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[V_ADDR:%.*]] = alloca <4 x float>, align 16 -// CHECK-NEXT: [[M:%.*]] = alloca [4 x float], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [2 x <2 x float>], align 4 // CHECK-NEXT: [[HLSL_EWCAST_SRC:%.*]] = alloca <4 x float>, align 16 // CHECK-NEXT: [[FLATCAST_TMP:%.*]] = alloca <4 x float>, align 4 // CHECK-NEXT: store <4 x float> [[V]], ptr [[V_ADDR]], align 16 diff --git a/clang/test/CodeGenHLSL/BasicFeatures/MatrixExplicitTruncation.hlsl b/clang/test/CodeGenHLSL/BasicFeatures/MatrixExplicitTruncation.hlsl index f3c4bc496d5a4..56f816806d63f 100644 --- a/clang/test/CodeGenHLSL/BasicFeatures/MatrixExplicitTruncation.hlsl +++ b/clang/test/CodeGenHLSL/BasicFeatures/MatrixExplicitTruncation.hlsl @@ -4,8 +4,8 @@ // CHECK-LABEL: define hidden noundef <12 x i32> @_Z10trunc_castu11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I34:%.*]] = alloca [12 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I34:%.*]] = alloca [3 x <4 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <12 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11> @@ -21,8 +21,8 @@ // CHECK-LABEL: define hidden noundef <12 x i32> @_Z11trunc_cast0u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I43:%.*]] = alloca [12 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I43:%.*]] = alloca [4 x <3 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <12 x i32> <i32 0, i32 1, i32 2, i32 4, i32 5, i32 6, i32 8, i32 9, i32 10, i32 12, i32 13, i32 14> @@ -38,8 +38,8 @@ // CHECK-LABEL: define hidden noundef <9 x i32> @_Z11trunc_cast1u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I33:%.*]] = alloca [9 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I33:%.*]] = alloca [3 x <3 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <9 x i32> <i32 0, i32 1, i32 2, i32 4, i32 5, i32 6, i32 8, i32 9, i32 10> @@ -55,8 +55,8 @@ // CHECK-LABEL: define hidden noundef <6 x i32> @_Z11trunc_cast2u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I32:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I32:%.*]] = alloca [3 x <2 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <6 x i32> <i32 0, i32 1, i32 4, i32 5, i32 8, i32 9> @@ -72,8 +72,8 @@ // CHECK-LABEL: define hidden noundef <6 x i32> @_Z11trunc_cast3u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I23:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I23:%.*]] = alloca [2 x <3 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <6 x i32> <i32 0, i32 1, i32 2, i32 4, i32 5, i32 6> @@ -89,8 +89,8 @@ // CHECK-LABEL: define hidden noundef <4 x i32> @_Z11trunc_cast4u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I22:%.*]] = alloca [4 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I22:%.*]] = alloca [2 x <2 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <4 x i32> <i32 0, i32 1, i32 4, i32 5> @@ -106,8 +106,8 @@ // CHECK-LABEL: define hidden noundef <2 x i32> @_Z11trunc_cast5u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I21:%.*]] = alloca [2 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I21:%.*]] = alloca [2 x <1 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <2 x i32> <i32 0, i32 4> @@ -123,7 +123,7 @@ // CHECK-LABEL: define hidden noundef i32 @_Z11trunc_cast6u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 // CHECK-NEXT: [[I1:%.*]] = alloca i32, align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 @@ -140,7 +140,7 @@ // CHECK-LABEL: define hidden noundef i32 @_Z16trunc_multi_castu11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 // CHECK-NEXT: [[I1:%.*]] = alloca i32, align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 diff --git a/clang/test/CodeGenHLSL/BasicFeatures/MatrixImplicitTruncation.hlsl b/clang/test/CodeGenHLSL/BasicFeatures/MatrixImplicitTruncation.hlsl index e621f68623bd1..b58f567eb51d3 100644 --- a/clang/test/CodeGenHLSL/BasicFeatures/MatrixImplicitTruncation.hlsl +++ b/clang/test/CodeGenHLSL/BasicFeatures/MatrixImplicitTruncation.hlsl @@ -4,8 +4,8 @@ // CHECK-LABEL: define hidden noundef <12 x i32> @_Z10trunc_castu11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I34:%.*]] = alloca [12 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I34:%.*]] = alloca [3 x <4 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <12 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11> @@ -21,8 +21,8 @@ // CHECK-LABEL: define hidden noundef <12 x i32> @_Z11trunc_cast0u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I43:%.*]] = alloca [12 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I43:%.*]] = alloca [4 x <3 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <12 x i32> <i32 0, i32 1, i32 2, i32 4, i32 5, i32 6, i32 8, i32 9, i32 10, i32 12, i32 13, i32 14> @@ -38,8 +38,8 @@ // CHECK-LABEL: define hidden noundef <9 x i32> @_Z11trunc_cast1u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I33:%.*]] = alloca [9 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I33:%.*]] = alloca [3 x <3 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <9 x i32> <i32 0, i32 1, i32 2, i32 4, i32 5, i32 6, i32 8, i32 9, i32 10> @@ -55,8 +55,8 @@ // CHECK-LABEL: define hidden noundef <6 x i32> @_Z11trunc_cast2u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I32:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I32:%.*]] = alloca [3 x <2 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <6 x i32> <i32 0, i32 1, i32 4, i32 5, i32 8, i32 9> @@ -72,8 +72,8 @@ // CHECK-LABEL: define hidden noundef <6 x i32> @_Z11trunc_cast3u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I23:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I23:%.*]] = alloca [2 x <3 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <6 x i32> <i32 0, i32 1, i32 2, i32 4, i32 5, i32 6> @@ -89,8 +89,8 @@ // CHECK-LABEL: define hidden noundef <4 x i32> @_Z11trunc_cast4u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I22:%.*]] = alloca [4 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I22:%.*]] = alloca [2 x <2 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <4 x i32> <i32 0, i32 1, i32 4, i32 5> @@ -106,8 +106,8 @@ // CHECK-LABEL: define hidden noundef <2 x i32> @_Z11trunc_cast5u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 -// CHECK-NEXT: [[I21:%.*]] = alloca [2 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 +// CHECK-NEXT: [[I21:%.*]] = alloca [2 x <1 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TRUNC:%.*]] = shufflevector <16 x i32> [[TMP0]], <16 x i32> poison, <2 x i32> <i32 0, i32 4> @@ -123,7 +123,7 @@ // CHECK-LABEL: define hidden noundef i32 @_Z11trunc_cast6u11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[I44:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [16 x i32], align 4 +// CHECK-NEXT: [[I44_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 // CHECK-NEXT: [[I1:%.*]] = alloca i32, align 4 // CHECK-NEXT: store <16 x i32> [[I44]], ptr [[I44_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[I44_ADDR]], align 4 diff --git a/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptConstSwizzle.hlsl b/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptConstSwizzle.hlsl index 02885d153697a..2b950d8a51a38 100644 --- a/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptConstSwizzle.hlsl +++ b/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptConstSwizzle.hlsl @@ -108,7 +108,7 @@ void setVectorOnMatrixSwizzle(out int2x3 M, int3 V) { // CHECK-SAME: ptr noalias noundef nonnull align 4 dereferenceable(24) [[M:%.*]], <6 x i32> noundef [[N:%.*]], i32 noundef [[MINDEX:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[M_ADDR:%.*]] = alloca ptr, align 4 -// CHECK-NEXT: [[N_ADDR:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[N_ADDR:%.*]] = alloca [2 x <3 x i32>], align 4 // CHECK-NEXT: [[MINDEX_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: store ptr [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: store <6 x i32> [[N]], ptr [[N_ADDR]], align 4 @@ -139,7 +139,7 @@ void setMatrixFromMatrix(out int2x3 M, int2x3 N, int MIndex) { // CHECK-SAME: ptr noalias noundef nonnull align 4 dereferenceable(24) [[M:%.*]], <6 x i32> noundef [[N:%.*]], i32 noundef [[NINDEX:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[M_ADDR:%.*]] = alloca ptr, align 4 -// CHECK-NEXT: [[N_ADDR:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[N_ADDR:%.*]] = alloca [2 x <3 x i32>], align 4 // CHECK-NEXT: [[NINDEX_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: store ptr [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: store <6 x i32> [[N]], ptr [[N_ADDR]], align 4 diff --git a/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptDynamicSwizzle.hlsl b/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptDynamicSwizzle.hlsl index 97ce63f545cff..7190b6e1148a5 100644 --- a/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptDynamicSwizzle.hlsl +++ b/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptDynamicSwizzle.hlsl @@ -38,7 +38,7 @@ void setMatrix(out float4x4 M, int index, float4 V) { // CHECK-LABEL: define hidden noundef nofpclass(nan inf) <3 x float> @_Z9getMatrixu11matrix_typeILm4ELm4EfEi( // CHECK-SAME: <16 x float> noundef nofpclass(nan inf) [[M:%.*]], i32 noundef [[INDEX:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [16 x float], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [4 x <4 x float>], align 4 // CHECK-NEXT: [[INDEX_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: store <16 x float> [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: store i32 [[INDEX]], ptr [[INDEX_ADDR]], align 4 @@ -62,7 +62,7 @@ float3 getMatrix(float4x4 M, int index) { // CHECK-LABEL: define hidden noundef nofpclass(nan inf) <4 x float> @_Z9getMatrixu11matrix_typeILm3ELm3EfEi( // CHECK-SAME: <9 x float> noundef nofpclass(nan inf) [[M:%.*]], i32 noundef [[INDEX:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [9 x float], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [3 x <3 x float>], align 4 // CHECK-NEXT: [[INDEX_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: store <9 x float> [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: store i32 [[INDEX]], ptr [[INDEX_ADDR]], align 4 @@ -115,7 +115,7 @@ int3 getMatrixSwizzle2x3(out int2x3 M, int index) { // CHECK-SAME: ptr noalias noundef nonnull align 4 dereferenceable(24) [[M:%.*]], <6 x i32> noundef [[N:%.*]], i32 noundef [[INDEX:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[M_ADDR:%.*]] = alloca ptr, align 4 -// CHECK-NEXT: [[N_ADDR:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[N_ADDR:%.*]] = alloca [2 x <3 x i32>], align 4 // CHECK-NEXT: [[INDEX_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: store ptr [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: store <6 x i32> [[N]], ptr [[N_ADDR]], align 4 diff --git a/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptGetter.hlsl b/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptGetter.hlsl index df724d217fe6b..efa9381b515af 100644 --- a/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptGetter.hlsl +++ b/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptGetter.hlsl @@ -4,7 +4,7 @@ // CHECK-LABEL: define hidden noundef nofpclass(nan inf) <4 x float> @_Z24getFloatVecMatrixDynamicu11matrix_typeILm4ELm4EfEi( // CHECK-SAME: <16 x float> noundef nofpclass(nan inf) [[M:%.*]], i32 noundef [[INDEX:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [16 x float], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [4 x <4 x float>], align 4 // CHECK-NEXT: [[INDEX_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: store <16 x float> [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: store i32 [[INDEX]], ptr [[INDEX_ADDR]], align 4 @@ -31,7 +31,7 @@ float4 getFloatVecMatrixDynamic(float4x4 M, int index) { // CHECK-LABEL: define hidden noundef nofpclass(nan inf) float @_Z27getFloatScalarMatrixDynamicu11matrix_typeILm2ELm1EfEi( // CHECK-SAME: <2 x float> noundef nofpclass(nan inf) [[M:%.*]], i32 noundef [[INDEX:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [2 x float], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [2 x <1 x float>], align 4 // CHECK-NEXT: [[INDEX_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: store <2 x float> [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: store i32 [[INDEX]], ptr [[INDEX_ADDR]], align 4 @@ -50,7 +50,7 @@ float getFloatScalarMatrixDynamic(float2x1 M, int index) { // CHECK-LABEL: define hidden noundef nofpclass(nan inf) float @_Z28getFloatScalarMatrixConstantu11matrix_typeILm2ELm1EfE( // CHECK-SAME: <2 x float> noundef nofpclass(nan inf) [[M:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [2 x float], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [2 x <1 x float>], align 4 // CHECK-NEXT: store <2 x float> [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <2 x float>, ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[MATRIX_ELEM:%.*]] = extractelement <2 x float> [[TMP0]], i32 0 @@ -65,7 +65,7 @@ float getFloatScalarMatrixConstant(float2x1 M) { // CHECK-LABEL: define hidden noundef nofpclass(nan inf) float @_Z29getFloatScalarMatrixConstant2u11matrix_typeILm2ELm1EfE( // CHECK-SAME: <2 x float> noundef nofpclass(nan inf) [[M:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [2 x float], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [2 x <1 x float>], align 4 // CHECK-NEXT: store <2 x float> [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <2 x float>, ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[MATRIX_ELEM:%.*]] = extractelement <2 x float> [[TMP0]], i32 1 @@ -80,7 +80,7 @@ float getFloatScalarMatrixConstant2(float2x1 M) { // CHECK-LABEL: define hidden noundef <4 x i32> @_Z19getIntMatrixDynamicu11matrix_typeILm4ELm4EiEi( // CHECK-SAME: <16 x i32> noundef [[M:%.*]], i32 noundef [[INDEX:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [16 x i32], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 // CHECK-NEXT: [[INDEX_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: store <16 x i32> [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: store i32 [[INDEX]], ptr [[INDEX_ADDR]], align 4 @@ -107,7 +107,7 @@ int4 getIntMatrixDynamic(int4x4 M, int index) { // CHECK-LABEL: define hidden noundef nofpclass(nan inf) <4 x float> @_Z22AddFloatMatrixConstantu11matrix_typeILm4ELm4EfE( // CHECK-SAME: <16 x float> noundef nofpclass(nan inf) [[M:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [16 x float], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [4 x <4 x float>], align 4 // CHECK-NEXT: store <16 x float> [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x float>, ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[MATRIX_ELEM:%.*]] = extractelement <16 x float> [[TMP0]], i32 0 @@ -157,7 +157,7 @@ float4 AddFloatMatrixConstant(float4x4 M) { // CHECK-LABEL: define hidden noundef <4 x i32> @_Z20AddIntMatrixConstantu11matrix_typeILm4ELm4EiE( // CHECK-SAME: <16 x i32> noundef [[M:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [16 x i32], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 // CHECK-NEXT: store <16 x i32> [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[MATRIX_ELEM:%.*]] = extractelement <16 x i32> [[TMP0]], i32 0 @@ -207,7 +207,7 @@ int4 AddIntMatrixConstant(int4x4 M) { // CHECK-LABEL: define hidden noundef <3 x i1> @_Z23getBoolVecMatrixDynamicu11matrix_typeILm2ELm3EbEi( // CHECK-SAME: <6 x i1> noundef [[M:%.*]], i32 noundef [[INDEX:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [2 x <3 x i32>], align 4 // CHECK-NEXT: [[INDEX_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: [[TMP0:%.*]] = zext <6 x i1> [[M]] to <6 x i32> // CHECK-NEXT: store <6 x i32> [[TMP0]], ptr [[M_ADDR]], align 4 @@ -233,7 +233,7 @@ bool3 getBoolVecMatrixDynamic(bool2x3 M, int index) { // CHECK-LABEL: define hidden noundef <4 x i1> @_Z24getBoolVecMatrixConstantu11matrix_typeILm4ELm4EbE( // CHECK-SAME: <16 x i1> noundef [[M:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [16 x i32], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 // CHECK-NEXT: [[TMP0:%.*]] = zext <16 x i1> [[M]] to <16 x i32> // CHECK-NEXT: store <16 x i32> [[TMP0]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[TMP1:%.*]] = load <16 x i32>, ptr [[M_ADDR]], align 4 @@ -255,7 +255,7 @@ bool4 getBoolVecMatrixConstant(bool4x4 M) { // CHECK-LABEL: define hidden noundef i1 @_Z27getBoolScalarMatrixConstantu11matrix_typeILm3ELm1EbE( // CHECK-SAME: <3 x i1> noundef [[M:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [3 x i32], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [3 x <1 x i32>], align 4 // CHECK-NEXT: [[TMP0:%.*]] = zext <3 x i1> [[M]] to <3 x i32> // CHECK-NEXT: store <3 x i32> [[TMP0]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[TMP1:%.*]] = load <3 x i32>, ptr [[M_ADDR]], align 4 diff --git a/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptSetter.hlsl b/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptSetter.hlsl index 15861b3211606..ec362aa269986 100644 --- a/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptSetter.hlsl +++ b/clang/test/CodeGenHLSL/BasicFeatures/MatrixSingleSubscriptSetter.hlsl @@ -127,7 +127,7 @@ void setBoolMatrixScalar(out bool2x1 M, int index, bool S) { // CHECK-SAME: ptr noalias noundef nonnull align 4 dereferenceable(64) [[M:%.*]], <16 x i32> noundef [[N:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[M_ADDR:%.*]] = alloca ptr, align 4 -// CHECK-NEXT: [[N_ADDR:%.*]] = alloca [16 x i32], align 4 +// CHECK-NEXT: [[N_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 // CHECK-NEXT: store ptr [[M]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: store <16 x i32> [[N]], ptr [[N_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[N_ADDR]], align 4 diff --git a/clang/test/CodeGenHLSL/BasicFeatures/MatrixSplat.hlsl b/clang/test/CodeGenHLSL/BasicFeatures/MatrixSplat.hlsl index 9b9538e0afdd1..768c1b8e02bea 100644 --- a/clang/test/CodeGenHLSL/BasicFeatures/MatrixSplat.hlsl +++ b/clang/test/CodeGenHLSL/BasicFeatures/MatrixSplat.hlsl @@ -4,7 +4,7 @@ // CHECK-LABEL: define hidden void @_Z13ConstantSplatv( // CHECK-SAME: ) #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M:%.*]] = alloca [16 x i32], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [4 x <4 x i32>], align 4 // CHECK-NEXT: store <16 x i32> splat (i32 1), ptr [[M]], align 4 // CHECK-NEXT: ret void // @@ -15,7 +15,7 @@ void ConstantSplat() { // CHECK-LABEL: define hidden void @_Z18ConstantFloatSplatv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M:%.*]] = alloca [4 x float], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [2 x <2 x float>], align 4 // CHECK-NEXT: store <4 x float> splat (float 3.250000e+00), ptr [[M]], align 4 // CHECK-NEXT: ret void // @@ -26,7 +26,7 @@ void ConstantFloatSplat() { // CHECK-LABEL: define hidden void @_Z21ConstantTrueBoolSplatv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M:%.*]] = alloca [9 x i32], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [3 x <3 x i32>], align 4 // CHECK-NEXT: store <9 x i32> splat (i32 1), ptr [[M]], align 4 // CHECK-NEXT: ret void // @@ -37,7 +37,7 @@ void ConstantTrueBoolSplat() { // CHECK-LABEL: define hidden void @_Z22ConstantFalseBoolSplatv( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M:%.*]] = alloca [9 x i32], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [3 x <3 x i32>], align 4 // CHECK-NEXT: store <9 x i32> zeroinitializer, ptr [[M]], align 4 // CHECK-NEXT: ret void // @@ -49,7 +49,7 @@ void ConstantFalseBoolSplat() { // CHECK-SAME: float noundef nofpclass(nan inf) [[VALUE:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[VALUE_ADDR:%.*]] = alloca float, align 4 -// CHECK-NEXT: [[M:%.*]] = alloca [9 x float], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [3 x <3 x float>], align 4 // CHECK-NEXT: store float [[VALUE]], ptr [[VALUE_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[VALUE_ADDR]], align 4 // CHECK-NEXT: [[SPLAT_SPLATINSERT:%.*]] = insertelement <9 x float> poison, float [[TMP0]], i64 0 @@ -65,7 +65,7 @@ void DynamicSplat(float Value) { // CHECK-SAME: i1 noundef [[VALUE:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[VALUE_ADDR:%.*]] = alloca i32, align 4 -// CHECK-NEXT: [[M:%.*]] = alloca [16 x i32], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [4 x <4 x i32>], align 4 // CHECK-NEXT: [[STOREDV:%.*]] = zext i1 [[VALUE]] to i32 // CHECK-NEXT: store i32 [[STOREDV]], ptr [[VALUE_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[VALUE_ADDR]], align 4 @@ -84,7 +84,7 @@ void DynamicBoolSplat(bool Value) { // CHECK-SAME: <4 x float> noundef nofpclass(nan inf) [[VALUE:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[VALUE_ADDR:%.*]] = alloca <4 x float>, align 16 -// CHECK-NEXT: [[M:%.*]] = alloca [9 x float], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [3 x <3 x float>], align 4 // CHECK-NEXT: store <4 x float> [[VALUE]], ptr [[VALUE_ADDR]], align 16 // CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[VALUE_ADDR]], align 16 // CHECK-NEXT: [[CAST_VTRUNC:%.*]] = extractelement <4 x float> [[TMP0]], i32 0 @@ -101,7 +101,7 @@ void CastThenSplat(float4 Value) { // CHECK-SAME: <3 x i32> noundef [[VALUE:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[VALUE_ADDR:%.*]] = alloca <3 x i32>, align 16 -// CHECK-NEXT: [[M:%.*]] = alloca [4 x i32], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [2 x <2 x i32>], align 4 // CHECK-NEXT: store <3 x i32> [[VALUE]], ptr [[VALUE_ADDR]], align 16 // CHECK-NEXT: [[TMP0:%.*]] = load <3 x i32>, ptr [[VALUE_ADDR]], align 16 // CHECK-NEXT: [[TOBOOL:%.*]] = icmp ne <3 x i32> [[TMP0]], zeroinitializer @@ -120,7 +120,7 @@ void ExplicitIntToBoolCastThenSplat(int3 Value) { // CHECK-SAME: <2 x float> noundef nofpclass(nan inf) [[VALUE:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[VALUE_ADDR:%.*]] = alloca <2 x float>, align 8 -// CHECK-NEXT: [[M:%.*]] = alloca [6 x i32], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [2 x <3 x i32>], align 4 // CHECK-NEXT: store <2 x float> [[VALUE]], ptr [[VALUE_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load <2 x float>, ptr [[VALUE_ADDR]], align 8 // CHECK-NEXT: [[TOBOOL:%.*]] = fcmp reassoc nnan ninf nsz arcp afn une <2 x float> [[TMP0]], zeroinitializer @@ -139,7 +139,7 @@ void ExplicitFloatToBoolCastThenSplat(float2 Value) { // CHECK-SAME: i1 noundef [[VALUE:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[VALUE_ADDR:%.*]] = alloca i32, align 4 -// CHECK-NEXT: [[M:%.*]] = alloca [6 x float], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [3 x <2 x float>], align 4 // CHECK-NEXT: [[STOREDV:%.*]] = zext i1 [[VALUE]] to i32 // CHECK-NEXT: store i32 [[STOREDV]], ptr [[VALUE_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[VALUE_ADDR]], align 4 diff --git a/clang/test/CodeGenHLSL/BoolMatrix.hlsl b/clang/test/CodeGenHLSL/BoolMatrix.hlsl index 824b9656e6848..d6ac50c978405 100644 --- a/clang/test/CodeGenHLSL/BoolMatrix.hlsl +++ b/clang/test/CodeGenHLSL/BoolMatrix.hlsl @@ -11,7 +11,7 @@ struct S { // CHECK-SAME: ) #[[ATTR0:[0-9]+]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[RETVAL:%.*]] = alloca i1, align 4 -// CHECK-NEXT: [[B:%.*]] = alloca [4 x i32], align 4 +// CHECK-NEXT: [[B:%.*]] = alloca [2 x <2 x i32>], align 4 // CHECK-NEXT: store <4 x i32> splat (i32 1), ptr [[B]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load <4 x i32>, ptr [[B]], align 4 // CHECK-NEXT: [[MATRIXEXT:%.*]] = extractelement <4 x i32> [[TMP0]], i32 0 @@ -29,7 +29,7 @@ bool fn1() { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x i1>, align 4 // CHECK-NEXT: [[V_ADDR:%.*]] = alloca i32, align 4 -// CHECK-NEXT: [[A:%.*]] = alloca [4 x i32], align 4 +// CHECK-NEXT: [[A:%.*]] = alloca [2 x <2 x i32>], align 4 // CHECK-NEXT: [[STOREDV:%.*]] = zext i1 [[V]] to i32 // CHECK-NEXT: store i32 [[STOREDV]], ptr [[V_ADDR]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[V_ADDR]], align 4 @@ -77,11 +77,11 @@ bool fn3() { // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[RETVAL:%.*]] = alloca i1, align 4 -// CHECK-NEXT: [[ARR:%.*]] = alloca [2 x [4 x i32]], align 4 +// CHECK-NEXT: [[ARR:%.*]] = alloca [2 x [2 x <2 x i32>]], align 4 // CHECK-NEXT: store <4 x i32> splat (i32 1), ptr [[ARR]], align 4 -// CHECK-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [4 x i32], ptr [[ARR]], i32 1 +// CHECK-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [2 x <2 x i32>], ptr [[ARR]], i32 1 // CHECK-NEXT: store <4 x i32> zeroinitializer, ptr [[ARRAYINIT_ELEMENT]], align 4 -// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x [4 x i32]], ptr [[ARR]], i32 0, i32 0 +// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x [2 x <2 x i32>]], ptr [[ARR]], i32 0, i32 0 // CHECK-NEXT: [[TMP0:%.*]] = load <4 x i32>, ptr [[ARRAYIDX]], align 4 // CHECK-NEXT: [[MATRIXEXT:%.*]] = extractelement <4 x i32> [[TMP0]], i32 1 // CHECK-NEXT: store i32 [[MATRIXEXT]], ptr [[RETVAL]], align 4 @@ -96,7 +96,7 @@ bool fn4() { // CHECK-LABEL: define hidden void @_Z3fn5v( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[M:%.*]] = alloca [4 x i32], align 4 +// CHECK-NEXT: [[M:%.*]] = alloca [2 x <2 x i32>], align 4 // CHECK-NEXT: store <4 x i32> splat (i32 1), ptr [[M]], align 4 // CHECK-NEXT: [[TMP0:%.*]] = getelementptr <4 x i32>, ptr [[M]], i32 0, i32 3 // CHECK-NEXT: store i32 0, ptr [[TMP0]], align 4 @@ -134,11 +134,11 @@ void fn6() { // CHECK-LABEL: define hidden void @_Z3fn7v( // CHECK-SAME: ) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[ARR:%.*]] = alloca [2 x [4 x i32]], align 4 +// CHECK-NEXT: [[ARR:%.*]] = alloca [2 x [2 x <2 x i32>]], align 4 // CHECK-NEXT: store <4 x i32> splat (i32 1), ptr [[ARR]], align 4 -// CHECK-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [4 x i32], ptr [[ARR]], i32 1 +// CHECK-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [2 x <2 x i32>], ptr [[ARR]], i32 1 // CHECK-NEXT: store <4 x i32> zeroinitializer, ptr [[ARRAYINIT_ELEMENT]], align 4 -// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x [4 x i32]], ptr [[ARR]], i32 0, i32 0 +// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x [2 x <2 x i32>]], ptr [[ARR]], i32 0, i32 0 // CHECK-NEXT: [[TMP0:%.*]] = getelementptr <4 x i32>, ptr [[ARRAYIDX]], i32 0, i32 1 // CHECK-NEXT: store i32 0, ptr [[TMP0]], align 4 // CHECK-NEXT: ret void @@ -152,7 +152,7 @@ void fn7() { // CHECK-SAME: <16 x i1> noundef [[M:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] // CHECK-NEXT: [[RETVAL:%.*]] = alloca <16 x i1>, align 4 -// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [16 x i32], align 4 +// CHECK-NEXT: [[M_ADDR:%.*]] = alloca [4 x <4 x i32>], align 4 // CHECK-NEXT: [[TMP0:%.*]] = zext <16 x i1> [[M]] to <16 x i32> // CHECK-NEXT: store <16 x i32> [[TMP0]], ptr [[M_ADDR]], align 4 // CHECK-NEXT: [[TMP1:%.*]] = load <16 x i32>, ptr [[M_ADDR]], align 4 diff --git a/clang/test/CodeGenHLSL/basic_types.hlsl b/clang/test/CodeGenHLSL/basic_types.hlsl index 677a9a8f5d1de..0aaf7a1b77797 100644 --- a/clang/test/CodeGenHLSL/basic_types.hlsl +++ b/clang/test/CodeGenHLSL/basic_types.hlsl @@ -38,22 +38,22 @@ // CHECK: @double2_Val = external hidden addrspace(2) global <2 x double>, align 16 // CHECK: @double3_Val = external hidden addrspace(2) global <3 x double>, align 32 // CHECK: @double4_Val = external hidden addrspace(2) global <4 x double>, align 32 -// CHECK: @bool1x1_Val = external hidden addrspace(2) global [1 x i32], align 4 -// CHECK: @bool1x2_Val = external hidden addrspace(2) global [2 x i32], align 4 -// CHECK: @bool1x3_Val = external hidden addrspace(2) global [3 x i32], align 4 -// CHECK: @bool1x4_Val = external hidden addrspace(2) global [4 x i32], align 4 -// CHECK: @bool2x1_Val = external hidden addrspace(2) global [2 x i32], align 4 -// CHECK: @bool2x2_Val = external hidden addrspace(2) global [4 x i32], align 4 -// CHECK: @bool2x3_Val = external hidden addrspace(2) global [6 x i32], align 4 -// CHECK: @bool2x4_Val = external hidden addrspace(2) global [8 x i32], align 4 -// CHECK: @bool3x1_Val = external hidden addrspace(2) global [3 x i32], align 4 -// CHECK: @bool3x2_Val = external hidden addrspace(2) global [6 x i32], align 4 -// CHECK: @bool3x3_Val = external hidden addrspace(2) global [9 x i32], align 4 -// CHECK: @bool3x4_Val = external hidden addrspace(2) global [12 x i32], align 4 -// CHECK: @bool4x1_Val = external hidden addrspace(2) global [4 x i32], align 4 -// CHECK: @bool4x2_Val = external hidden addrspace(2) global [8 x i32], align 4 -// CHECK: @bool4x3_Val = external hidden addrspace(2) global [12 x i32], align 4 -// CHECK: @bool4x4_Val = external hidden addrspace(2) global [16 x i32], align 4 +// CHECK: @bool1x1_Val = external hidden addrspace(2) global [1 x <1 x i32>], align 4 +// CHECK: @bool1x2_Val = external hidden addrspace(2) global [1 x <2 x i32>], align 4 +// CHECK: @bool1x3_Val = external hidden addrspace(2) global [1 x <3 x i32>], align 4 +// CHECK: @bool1x4_Val = external hidden addrspace(2) global [1 x <4 x i32>], align 4 +// CHECK: @bool2x1_Val = external hidden addrspace(2) global [2 x <1 x i32>], align 4 +// CHECK: @bool2x2_Val = external hidden addrspace(2) global [2 x <2 x i32>], align 4 +// CHECK: @bool2x3_Val = external hidden addrspace(2) global [2 x <3 x i32>], align 4 +// CHECK: @bool2x4_Val = external hidden addrspace(2) global [2 x <4 x i32>], align 4 +// CHECK: @bool3x1_Val = external hidden addrspace(2) global [3 x <1 x i32>], align 4 +// CHECK: @bool3x2_Val = external hidden addrspace(2) global [3 x <2 x i32>], align 4 +// CHECK: @bool3x3_Val = external hidden addrspace(2) global [3 x <3 x i32>], align 4 +// CHECK: @bool3x4_Val = external hidden addrspace(2) global [3 x <4 x i32>], align 4 +// CHECK: @bool4x1_Val = external hidden addrspace(2) global [4 x <1 x i32>], align 4 +// CHECK: @bool4x2_Val = external hidden addrspace(2) global [4 x <2 x i32>], align 4 +// CHECK: @bool4x3_Val = external hidden addrspace(2) global [4 x <3 x i32>], align 4 +// CHECK: @bool4x4_Val = external hidden addrspace(2) global [4 x <4 x i32>], align 4 #ifdef NAMESPACED #define TYPE_DECL(T) hlsl::T T##_Val _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
