https://github.com/Ko496-glitch updated https://github.com/llvm/llvm-project/pull/199153
>From 7f3d9116f86c629003a5a415fb73c0301e59dea3 Mon Sep 17 00:00:00 2001 From: Jannick Kremer <[email protected]> Date: Mon, 8 Jun 2026 06:10:06 +0200 Subject: [PATCH 1/2] added more test --- .github/workflows/libclang-python-tests.yml | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/.github/workflows/libclang-python-tests.yml b/.github/workflows/libclang-python-tests.yml index b417cd74c61a9..ae68132965886 100644 --- a/.github/workflows/libclang-python-tests.yml +++ b/.github/workflows/libclang-python-tests.yml @@ -30,7 +30,7 @@ jobs: strategy: fail-fast: false matrix: - python-version: ["3.8", "3.13"] + python-version: ["3.8", "3.14"] steps: - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2 with: @@ -38,7 +38,6 @@ jobs: - name: Setup Python uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0 with: - persist-credentials: false python-version: ${{ matrix.python-version }} - name: Setup ccache uses: hendrikmuhs/ccache-action@33522472633dbd32578e909b315f5ee43ba878ce # v1.2.22 >From 2c20180ff511da67d6580c794a3f125a9792bc9b Mon Sep 17 00:00:00 2001 From: Kartik Ohlan <[email protected]> Date: Sun, 7 Jun 2026 21:49:06 -0700 Subject: [PATCH 2/2] restore the files --- .../lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp | 4 + clang/test/CodeGen/AArch64/neon-intrinsics.c | 50 --- clang/test/CodeGen/AArch64/neon/intrinsics.c | 317 ++++++++++++++++++ 3 files changed, 321 insertions(+), 50 deletions(-) diff --git a/clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp b/clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp index be906d0671e3a..854c30013553c 100644 --- a/clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp +++ b/clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp @@ -441,6 +441,10 @@ static mlir::Value emitCommonNeonSISDBuiltinExpr( case NEON::BI__builtin_neon_vsrid_n_u64: case NEON::BI__builtin_neon_vslid_n_s64: case NEON::BI__builtin_neon_vslid_n_u64: + case NEON::BI__builtin_neon_vqshlh_s16: + case NEON::BI__builtin_neon_vqshlh_u16: + case NEON::BI__builtin_neon_vqshlh_n_s16: + case NEON::BI__builtin_neon_vqshlh_n_u16: break; } diff --git a/clang/test/CodeGen/AArch64/neon-intrinsics.c b/clang/test/CodeGen/AArch64/neon-intrinsics.c index 1f6359bbe2c89..4145f2e7d3ffe 100644 --- a/clang/test/CodeGen/AArch64/neon-intrinsics.c +++ b/clang/test/CodeGen/AArch64/neon-intrinsics.c @@ -8677,19 +8677,6 @@ int8_t test_vqshlb_s8(int8_t a, int8_t b) { return vqshlb_s8(a, b); } -// CHECK-LABEL: define dso_local i16 @test_vqshlh_s16( -// CHECK-SAME: i16 noundef [[A:%.*]], i16 noundef [[B:%.*]]) #[[ATTR0]] { -// CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = insertelement <4 x i16> poison, i16 [[A]], i64 0 -// CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i16> poison, i16 [[B]], i64 0 -// CHECK-NEXT: [[VQSHLH_S16_I:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqshl.v4i16(<4 x i16> [[TMP0]], <4 x i16> [[TMP1]]) -// CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i16> [[VQSHLH_S16_I]], i64 0 -// CHECK-NEXT: ret i16 [[TMP2]] -// -int16_t test_vqshlh_s16(int16_t a, int16_t b) { - return vqshlh_s16(a, b); -} - // CHECK-LABEL: define dso_local i32 @test_vqshls_s32( // CHECK-SAME: i32 noundef [[A:%.*]], i32 noundef [[B:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] @@ -8723,19 +8710,6 @@ uint8_t test_vqshlb_u8(uint8_t a, int8_t b) { return vqshlb_u8(a, b); } -// CHECK-LABEL: define dso_local i16 @test_vqshlh_u16( -// CHECK-SAME: i16 noundef [[A:%.*]], i16 noundef [[B:%.*]]) #[[ATTR0]] { -// CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = insertelement <4 x i16> poison, i16 [[A]], i64 0 -// CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i16> poison, i16 [[B]], i64 0 -// CHECK-NEXT: [[VQSHLH_U16_I:%.*]] = call <4 x i16> @llvm.aarch64.neon.uqshl.v4i16(<4 x i16> [[TMP0]], <4 x i16> [[TMP1]]) -// CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i16> [[VQSHLH_U16_I]], i64 0 -// CHECK-NEXT: ret i16 [[TMP2]] -// -uint16_t test_vqshlh_u16(uint16_t a, int16_t b) { - return vqshlh_u16(a, b); -} - // CHECK-LABEL: define dso_local i32 @test_vqshls_u32( // CHECK-SAME: i32 noundef [[A:%.*]], i32 noundef [[B:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] @@ -14241,18 +14215,6 @@ int8_t test_vqshlb_n_s8(int8_t a) { return (int8_t)vqshlb_n_s8(a, 7); } -// CHECK-LABEL: define dso_local i16 @test_vqshlh_n_s16( -// CHECK-SAME: i16 noundef [[A:%.*]]) #[[ATTR0]] { -// CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = insertelement <4 x i16> poison, i16 [[A]], i64 0 -// CHECK-NEXT: [[VQSHLH_N_S16:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqshl.v4i16(<4 x i16> [[TMP0]], <4 x i16> <i16 15, i16 poison, i16 poison, i16 poison>) -// CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i16> [[VQSHLH_N_S16]], i64 0 -// CHECK-NEXT: ret i16 [[TMP1]] -// -int16_t test_vqshlh_n_s16(int16_t a) { - return (int16_t)vqshlh_n_s16(a, 15); -} - // CHECK-LABEL: define dso_local i32 @test_vqshls_n_s32( // CHECK-SAME: i32 noundef [[A:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] @@ -14447,18 +14409,6 @@ uint8_t test_vqshlb_n_u8(uint8_t a) { return (uint8_t)vqshlb_n_u8(a, 7); } -// CHECK-LABEL: define dso_local i16 @test_vqshlh_n_u16( -// CHECK-SAME: i16 noundef [[A:%.*]]) #[[ATTR0]] { -// CHECK-NEXT: [[ENTRY:.*:]] -// CHECK-NEXT: [[TMP0:%.*]] = insertelement <4 x i16> poison, i16 [[A]], i64 0 -// CHECK-NEXT: [[VQSHLH_N_U16:%.*]] = call <4 x i16> @llvm.aarch64.neon.uqshl.v4i16(<4 x i16> [[TMP0]], <4 x i16> <i16 15, i16 poison, i16 poison, i16 poison>) -// CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i16> [[VQSHLH_N_U16]], i64 0 -// CHECK-NEXT: ret i16 [[TMP1]] -// -uint16_t test_vqshlh_n_u16(uint16_t a) { - return (uint16_t)vqshlh_n_u16(a, 15); -} - // CHECK-LABEL: define dso_local i32 @test_vqshls_n_u32( // CHECK-SAME: i32 noundef [[A:%.*]]) #[[ATTR0]] { // CHECK-NEXT: [[ENTRY:.*:]] diff --git a/clang/test/CodeGen/AArch64/neon/intrinsics.c b/clang/test/CodeGen/AArch64/neon/intrinsics.c index aa3213efd1735..2a274bd9658af 100644 --- a/clang/test/CodeGen/AArch64/neon/intrinsics.c +++ b/clang/test/CodeGen/AArch64/neon/intrinsics.c @@ -494,6 +494,323 @@ float64_t test_vmaxvq_f64(float64x2_t a) { // TODO: Implement the remaining intrinsics from this group. //===------------------------------------------------------===// +<<<<<<< Updated upstream +======= +// ALL-LABEL: {{.*}}@test_vqshlh_s16( +int16_t test_vqshlh_s16(int16_t a, int16_t b) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshl" {{%.*}} + + // LLVM-SAME: i16 {{.*}}[[A:%.*]], i16 {{.*}}[[B:%.*]]) + // LLVM: [[TMP0:%.*]] = insertelement <4 x i16> poison, i16 [[A]], i64 0 + // LLVM: [[TMP1:%.*]] = insertelement <4 x i16> poison, i16 [[B]], i64 0 + // LLVM: [[VQSHLH_S16_I:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqshl.v4i16(<4 x i16> [[TMP0]], <4 x i16> [[TMP1]]) + // LLVM: [[TMP2:%.*]] = extractelement <4 x i16> [[VQSHLH_S16_I]], i64 0 + // LLVM: ret i16 [[TMP2]] + return vqshlh_s16(a, b); +} + +// ALL-LABEL: {{.*}}@test_vqshlh_u16( +uint16_t test_vqshlh_u16(uint16_t a, int16_t b) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.uqshl" {{%.*}} + + // LLVM-SAME: i16 {{.*}}[[A:%.*]], i16 {{.*}}[[B:%.*]]) + // LLVM: [[TMP0:%.*]] = insertelement <4 x i16> poison, i16 [[A]], i64 0 + // LLVM: [[TMP1:%.*]] = insertelement <4 x i16> poison, i16 [[B]], i64 0 + // LLVM: [[VQSHLH_U16_I:%.*]] = call <4 x i16> @llvm.aarch64.neon.uqshl.v4i16(<4 x i16> [[TMP0]], <4 x i16> [[TMP1]]) + // LLVM: [[TMP2:%.*]] = extractelement <4 x i16> [[VQSHLH_U16_I]], i64 0 + // LLVM: ret i16 [[TMP2]] + return vqshlh_u16(a, b); +} + +// ALL-LABEL: {{.*}}@test_vqshlh_n_s16( +int16_t test_vqshlh_n_s16(int16_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshl" {{%.*}} + + // LLVM-SAME: i16 {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = insertelement <4 x i16> poison, i16 [[A]], i64 0 + // LLVM: [[VQSHLH_N_S16:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqshl.v4i16(<4 x i16> [[TMP0]], <4 x i16> <i16 15, i16 poison, i16 poison, i16 poison>) + // LLVM: [[TMP1:%.*]] = extractelement <4 x i16> [[VQSHLH_N_S16]], i64 0 + // LLVM: ret i16 [[TMP1]] + return (int16_t)vqshlh_n_s16(a, 15); +} + +// ALL-LABEL: {{.*}}@test_vqshl_n_s32( +int32x2_t test_vqshl_n_s32(int32x2_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshl" + + // LLVM-SAME: <2 x i32> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <2 x i32> [[A]] to <8 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> + // LLVM: [[VQSHL_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.sqshl.v2i32(<2 x i32> [[VQSHL_N]], <2 x i32> zeroinitializer) + // LLVM: ret <2 x i32> [[VQSHL_N1]] + return vqshl_n_s32(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshl_n_s64( +int64x1_t test_vqshl_n_s64(int64x1_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshl" + + // LLVM-SAME: <1 x i64> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <1 x i64> [[A]] to <8 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64> + // LLVM: [[VQSHL_N1:%.*]] = call <1 x i64> @llvm.aarch64.neon.sqshl.v1i64(<1 x i64> [[VQSHL_N]], <1 x i64> {{.*}}) + // LLVM: ret <1 x i64> [[VQSHL_N1]] + return vqshl_n_s64(a, 1); +} + +// ALL-LABEL: {{.*}}@test_vqshlq_n_s8( +int8x16_t test_vqshlq_n_s8(int8x16_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshl" + + // LLVM-SAME: <16 x i8> {{.*}}[[A:%.*]]) + // LLVM: [[VQSHL_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.sqshl.v16i8(<16 x i8> [[A]], <16 x i8> zeroinitializer) + // LLVM: ret <16 x i8> [[VQSHL_N]] + return vqshlq_n_s8(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshlq_n_s16( +int16x8_t test_vqshlq_n_s16(int16x8_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshl" + + // LLVM-SAME: <8 x i16> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <8 x i16> [[A]] to <16 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> + // LLVM: [[VQSHL_N1:%.*]] = call <8 x i16> @llvm.aarch64.neon.sqshl.v8i16(<8 x i16> [[VQSHL_N]], <8 x i16> zeroinitializer) + // LLVM: ret <8 x i16> [[VQSHL_N1]] + return vqshlq_n_s16(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshlq_n_s32( +int32x4_t test_vqshlq_n_s32(int32x4_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshl" + + // LLVM-SAME: <4 x i32> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <4 x i32> [[A]] to <16 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> + // LLVM: [[VQSHL_N1:%.*]] = call <4 x i32> @llvm.aarch64.neon.sqshl.v4i32(<4 x i32> [[VQSHL_N]], <4 x i32> zeroinitializer) + // LLVM: ret <4 x i32> [[VQSHL_N1]] + return vqshlq_n_s32(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshlq_n_s64( +int64x2_t test_vqshlq_n_s64(int64x2_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshl" + + // LLVM-SAME: <2 x i64> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <2 x i64> [[A]] to <16 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> + // LLVM: [[VQSHL_N1:%.*]] = call <2 x i64> @llvm.aarch64.neon.sqshl.v2i64(<2 x i64> [[VQSHL_N]], <2 x i64> zeroinitializer) + // LLVM: ret <2 x i64> [[VQSHL_N1]] + return vqshlq_n_s64(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshl_n_u8( +uint8x8_t test_vqshl_n_u8(uint8x8_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.uqshl" + + // LLVM-SAME: <8 x i8> {{.*}}[[A:%.*]]) + // LLVM: [[VQSHL_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.uqshl.v8i8(<8 x i8> [[A]], <8 x i8> zeroinitializer) + // LLVM: ret <8 x i8> [[VQSHL_N]] + return vqshl_n_u8(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshl_n_u16( +uint16x4_t test_vqshl_n_u16(uint16x4_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.uqshl" + + // LLVM-SAME: <4 x i16> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <4 x i16> [[A]] to <8 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> + // LLVM: [[VQSHL_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.uqshl.v4i16(<4 x i16> [[VQSHL_N]], <4 x i16> zeroinitializer) + // LLVM: ret <4 x i16> [[VQSHL_N1]] + return vqshl_n_u16(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshl_n_u32( +uint32x2_t test_vqshl_n_u32(uint32x2_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.uqshl" + + // LLVM-SAME: <2 x i32> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <2 x i32> [[A]] to <8 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> + // LLVM: [[VQSHL_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.uqshl.v2i32(<2 x i32> [[VQSHL_N]], <2 x i32> zeroinitializer) + // LLVM: ret <2 x i32> [[VQSHL_N1]] + return vqshl_n_u32(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshl_n_u64( +uint64x1_t test_vqshl_n_u64(uint64x1_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.uqshl" + + // LLVM-SAME: <1 x i64> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <1 x i64> [[A]] to <8 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64> + // LLVM: [[VQSHL_N1:%.*]] = call <1 x i64> @llvm.aarch64.neon.uqshl.v1i64(<1 x i64> [[VQSHL_N]], <1 x i64> {{.*}}) + // LLVM: ret <1 x i64> [[VQSHL_N1]] + return vqshl_n_u64(a, 1); +} + +// ALL-LABEL: {{.*}}@test_vqshlq_n_u8( +uint8x16_t test_vqshlq_n_u8(uint8x16_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.uqshl" + + // LLVM-SAME: <16 x i8> {{.*}}[[A:%.*]]) + // LLVM: [[VQSHL_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.uqshl.v16i8(<16 x i8> [[A]], <16 x i8> zeroinitializer) + // LLVM: ret <16 x i8> [[VQSHL_N]] + return vqshlq_n_u8(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshlq_n_u16( +uint16x8_t test_vqshlq_n_u16(uint16x8_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.uqshl" + + // LLVM-SAME: <8 x i16> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <8 x i16> [[A]] to <16 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> + // LLVM: [[VQSHL_N1:%.*]] = call <8 x i16> @llvm.aarch64.neon.uqshl.v8i16(<8 x i16> [[VQSHL_N]], <8 x i16> zeroinitializer) + // LLVM: ret <8 x i16> [[VQSHL_N1]] + return vqshlq_n_u16(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshlq_n_u32( +uint32x4_t test_vqshlq_n_u32(uint32x4_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.uqshl" + + // LLVM-SAME: <4 x i32> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <4 x i32> [[A]] to <16 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> + // LLVM: [[VQSHL_N1:%.*]] = call <4 x i32> @llvm.aarch64.neon.uqshl.v4i32(<4 x i32> [[VQSHL_N]], <4 x i32> zeroinitializer) + // LLVM: ret <4 x i32> [[VQSHL_N1]] + return vqshlq_n_u32(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshlq_n_u64( +uint64x2_t test_vqshlq_n_u64(uint64x2_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.uqshl" + + // LLVM-SAME: <2 x i64> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <2 x i64> [[A]] to <16 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> + // LLVM: [[VQSHL_N1:%.*]] = call <2 x i64> @llvm.aarch64.neon.uqshl.v2i64(<2 x i64> [[VQSHL_N]], <2 x i64> zeroinitializer) + // LLVM: ret <2 x i64> [[VQSHL_N1]] + return vqshlq_n_u64(a, 0); +} + +// ALL-LABEL: {{.*}}@test_vqshlu_n_s8( +uint8x8_t test_vqshlu_n_s8(int8x8_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshlu" + + // LLVM-SAME: <8 x i8> {{.*}}[[A:%.*]]) + // LLVM: [[VQSHLU_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqshlu.v8i8(<8 x i8> [[A]], <8 x i8> {{.*}}) + // LLVM: ret <8 x i8> [[VQSHLU_N]] + return vqshlu_n_s8(a, 3); +} + +// ALL-LABEL: {{.*}}@test_vqshlu_n_s16( +uint16x4_t test_vqshlu_n_s16(int16x4_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshlu" + + // LLVM-SAME: <4 x i16> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <4 x i16> [[A]] to <8 x i8> + // LLVM: [[VQSHLU_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> + // LLVM: [[VQSHLU_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqshlu.v4i16(<4 x i16> [[VQSHLU_N]], <4 x i16> {{.*}}) + // LLVM: ret <4 x i16> [[VQSHLU_N1]] + return vqshlu_n_s16(a, 3); +} + +// ALL-LABEL: {{.*}}@test_vqshlu_n_s32( +uint32x2_t test_vqshlu_n_s32(int32x2_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshlu" + + // LLVM-SAME: <2 x i32> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <2 x i32> [[A]] to <8 x i8> + // LLVM: [[VQSHLU_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> + // LLVM: [[VQSHLU_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.sqshlu.v2i32(<2 x i32> [[VQSHLU_N]], <2 x i32> {{.*}}) + // LLVM: ret <2 x i32> [[VQSHLU_N1]] + return vqshlu_n_s32(a, 3); +} + +// ALL-LABEL: {{.*}}@test_vqshlu_n_s64( +uint64x1_t test_vqshlu_n_s64(int64x1_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshlu" + + // LLVM-SAME: <1 x i64> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <1 x i64> [[A]] to <8 x i8> + // LLVM: [[VQSHLU_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64> + // LLVM: [[VQSHLU_N1:%.*]] = call <1 x i64> @llvm.aarch64.neon.sqshlu.v1i64(<1 x i64> [[VQSHLU_N]], <1 x i64> {{.*}}) + // LLVM: ret <1 x i64> [[VQSHLU_N1]] + return vqshlu_n_s64(a, 1); +} + +// ALL-LABEL: {{.*}}@test_vqshluq_n_s8( +uint8x16_t test_vqshluq_n_s8(int8x16_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshlu" + + // LLVM-SAME: <16 x i8> {{.*}}[[A:%.*]]) + // LLVM: [[VQSHLU_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.sqshlu.v16i8(<16 x i8> [[A]], <16 x i8> splat (i8 3)) + // LLVM: ret <16 x i8> [[VQSHLU_N]] + return vqshluq_n_s8(a, 3); +} + +// ALL-LABEL: {{.*}}@test_vqshluq_n_s16( +uint16x8_t test_vqshluq_n_s16(int16x8_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshlu" + + // LLVM-SAME: <8 x i16> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <8 x i16> [[A]] to <16 x i8> + // LLVM: [[VQSHLU_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> + // LLVM: [[VQSHLU_N1:%.*]] = call <8 x i16> @llvm.aarch64.neon.sqshlu.v8i16(<8 x i16> [[VQSHLU_N]], <8 x i16> splat (i16 3)) + // LLVM: ret <8 x i16> [[VQSHLU_N1]] + return vqshluq_n_s16(a, 3); +} + +// ALL-LABEL: {{.*}}@test_vqshluq_n_s32( +uint32x4_t test_vqshluq_n_s32(int32x4_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshlu" + + // LLVM-SAME: <4 x i32> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <4 x i32> [[A]] to <16 x i8> + // LLVM: [[VQSHLU_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> + // LLVM: [[VQSHLU_N1:%.*]] = call <4 x i32> @llvm.aarch64.neon.sqshlu.v4i32(<4 x i32> [[VQSHLU_N]], <4 x i32> splat (i32 3)) + // LLVM: ret <4 x i32> [[VQSHLU_N1]] + return vqshluq_n_s32(a, 3); +} + +// ALL-LABEL: {{.*}}@test_vqshluq_n_s64( +uint64x2_t test_vqshluq_n_s64(int64x2_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshlu" + + // LLVM-SAME: <2 x i64> {{.*}}[[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <2 x i64> [[A]] to <16 x i8> + // LLVM: [[VQSHLU_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> + // LLVM: [[VQSHLU_N1:%.*]] = call <2 x i64> @llvm.aarch64.neon.sqshlu.v2i64(<2 x i64> [[VQSHLU_N]], <2 x i64> splat (i64 3)) + // LLVM: ret <2 x i64> [[VQSHLU_N1]] + return vqshluq_n_s64(a, 3); +} + +// ALL-LABEL: @test_vqshl_n_s8( +int8x8_t test_vqshl_n_s8(int8x8_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshl" + + // LLVM-SAME: <8 x i8> {{.*}} [[A:%.*]]) + // LLVM: [[VQSHL_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqshl.v8i8(<8 x i8> [[A]], <8 x i8> zeroinitializer) + // LLVM: ret <8 x i8> [[VQSHL_N]] + return vqshl_n_s8(a, 0); +} + +// ALL-LABEL: @test_vqshl_n_s16( +int16x4_t test_vqshl_n_s16(int16x4_t a) { + // CIR: cir.call_llvm_intrinsic "aarch64.neon.sqshl" + // + // LLVM-SAME: <4 x i16> {{.*}} [[A:%.*]]) + // LLVM: [[TMP0:%.*]] = bitcast <4 x i16> [[A]] to <8 x i8> + // LLVM: [[VQSHL_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> + // LLVM: [[VQSHL_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqshl.v4i16(<4 x i16> [[VQSHL_N]], <4 x i16> zeroinitializer) + // LLVM: ret <4 x i16> [[VQSHL_N1]] + return vqshl_n_s16(a, 0); +} + +>>>>>>> Stashed changes // ALL-LABEL: @test_vqshlud_n_s64( int64_t test_vqshlud_n_s64(int64_t a) { // CIR: [[CONST:%.*]] = cir.const #cir.int<63> : !s64i _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
