diff --git a/clang/include/clang/Basic/BuiltinsX86.td b/clang/include/clang/Basic/BuiltinsX86.td index 54b3ce08f5625..8332eac2a890c 100644 --- a/clang/include/clang/Basic/BuiltinsX86.td +++ b/clang/include/clang/Basic/BuiltinsX86.td @@ -123,13 +123,13 @@ let Attributes = [Const, NoThrow, RequiredVectorWidth<128>] in { } } - let Features = "ssse3" in { - def psignb128 : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>)">; - def psignw128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">; - def psignd128 : X86Builtin<"_Vector<4, int>(_Vector<4, int>, _Vector<4, int>)">; - } - let Features = "ssse3", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in { + def psignb128 + : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>)">; + def psignw128 + : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">; + def psignd128 + : X86Builtin<"_Vector<4, int>(_Vector<4, int>, _Vector<4, int>)">; def pmulhrsw128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">; def pmaddubsw128 : X86Builtin<"_Vector<8, short>(_Vector<16, char>, _Vector<16, char>)">; def pshufb128 : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>)">; @@ -603,10 +603,9 @@ let Features = "avx2", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] i def mpsadbw256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>, _Constant char)">; def palignr256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, " "_Vector<32, char>, _Constant int)">; - def psadbw256 : X86Builtin<"_Vector<4, long long int>(_Vector<32, char>, _Vector<32, char>)">; - def psignb256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>)">; - def psignw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">; - def psignd256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">; + def psadbw256 + : X86Builtin< + "_Vector<4, long long int>(_Vector<32, char>, _Vector<32, char>)">; def psllw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<8, short>)">; def pslldqi256_byteshift : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Constant int)">; def pslld256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<4, int>)">; @@ -677,7 +676,15 @@ let Features = "avx2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWi def phsubw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">; def phsubd256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">; def phsubsw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">; - + + def psignb256 + : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>)">; + def psignw256 + : X86Builtin< + "_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">; + def psignd256 + : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>)">; + def pshuflw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Constant int)">; def pshufhw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Constant int)">; def pshufd256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Constant int)">; diff --git a/clang/lib/AST/ByteCode/InterpBuiltin.cpp b/clang/lib/AST/ByteCode/InterpBuiltin.cpp index 39b991c68f0db..912db8e8109fa 100644 --- a/clang/lib/AST/ByteCode/InterpBuiltin.cpp +++ b/clang/lib/AST/ByteCode/InterpBuiltin.cpp @@ -3802,6 +3802,21 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call, return interp__builtin_ia32_movmsk_op(S, OpPC, Call); } + case X86::BI__builtin_ia32_psignb128: + case X86::BI__builtin_ia32_psignb256: + case X86::BI__builtin_ia32_psignw128: + case X86::BI__builtin_ia32_psignw256: + case X86::BI__builtin_ia32_psignd128: + case X86::BI__builtin_ia32_psignd256: + return interp__builtin_elementwise_int_binop( + S, OpPC, Call, [](const APInt &AElem, const APInt &BElem) { + if (BElem.isZero()) + return APInt::getZero(AElem.getBitWidth()); + if (BElem.isNegative()) + return -AElem; + return AElem; + }); + case clang::X86::BI__builtin_ia32_pavgb128: case clang::X86::BI__builtin_ia32_pavgw128: case clang::X86::BI__builtin_ia32_pavgb256: diff --git a/clang/lib/AST/ExprConstant.cpp b/clang/lib/AST/ExprConstant.cpp index 00aaaab957591..665a1c06b85ed 100644 --- a/clang/lib/AST/ExprConstant.cpp +++ b/clang/lib/AST/ExprConstant.cpp @@ -12312,6 +12312,21 @@ bool VectorExprEvaluator::VisitCallExpr(const CallExpr *E) { return Success(APValue(ResultElements.data(), ResultElements.size()), E); } + case X86::BI__builtin_ia32_psignb128: + case X86::BI__builtin_ia32_psignb256: + case X86::BI__builtin_ia32_psignw128: + case X86::BI__builtin_ia32_psignw256: + case X86::BI__builtin_ia32_psignd128: + case X86::BI__builtin_ia32_psignd256: + return EvaluateBinOpExpr( + [](const APInt &AElem, const APInt &BElem) -> APInt { + if (BElem.isZero()) + return APInt::getZero(AElem.getBitWidth()); + if (BElem.isNegative()) + return -AElem; + return AElem; + }); + case X86::BI__builtin_ia32_blendvpd: case X86::BI__builtin_ia32_blendvpd256: case X86::BI__builtin_ia32_blendvps: diff --git a/clang/lib/Headers/avx2intrin.h b/clang/lib/Headers/avx2intrin.h index fdb825fbbd134..3cbaaece7b38e 100644 --- a/clang/lib/Headers/avx2intrin.h +++ b/clang/lib/Headers/avx2intrin.h @@ -1975,10 +1975,9 @@ _mm256_shuffle_epi8(__m256i __a, __m256i __b) { /// \param __b /// A 256-bit integer vector]. /// \returns A 256-bit integer vector containing the result. -static __inline__ __m256i __DEFAULT_FN_ATTRS256 -_mm256_sign_epi8(__m256i __a, __m256i __b) -{ - return (__m256i)__builtin_ia32_psignb256((__v32qi)__a, (__v32qi)__b); +static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR +_mm256_sign_epi8(__m256i __a, __m256i __b) { + return (__m256i)__builtin_ia32_psignb256((__v32qi)__a, (__v32qi)__b); } /// Sets each element of the result to the corresponding element of the @@ -1996,10 +1995,9 @@ _mm256_sign_epi8(__m256i __a, __m256i __b) /// \param __b /// A 256-bit vector of [16 x i16]. /// \returns A 256-bit vector of [16 x i16] containing the result. -static __inline__ __m256i __DEFAULT_FN_ATTRS256 -_mm256_sign_epi16(__m256i __a, __m256i __b) -{ - return (__m256i)__builtin_ia32_psignw256((__v16hi)__a, (__v16hi)__b); +static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR +_mm256_sign_epi16(__m256i __a, __m256i __b) { + return (__m256i)__builtin_ia32_psignw256((__v16hi)__a, (__v16hi)__b); } /// Sets each element of the result to the corresponding element of the @@ -2017,10 +2015,9 @@ _mm256_sign_epi16(__m256i __a, __m256i __b) /// \param __b /// A 256-bit vector of [8 x i32]. /// \returns A 256-bit vector of [8 x i32] containing the result. -static __inline__ __m256i __DEFAULT_FN_ATTRS256 -_mm256_sign_epi32(__m256i __a, __m256i __b) -{ - return (__m256i)__builtin_ia32_psignd256((__v8si)__a, (__v8si)__b); +static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR +_mm256_sign_epi32(__m256i __a, __m256i __b) { + return (__m256i)__builtin_ia32_psignd256((__v8si)__a, (__v8si)__b); } /// Shifts each 128-bit half of the 256-bit integer vector \a a left by diff --git a/clang/lib/Headers/tmmintrin.h b/clang/lib/Headers/tmmintrin.h index 5d0f20f4d527d..95b32c0087ecb 100644 --- a/clang/lib/Headers/tmmintrin.h +++ b/clang/lib/Headers/tmmintrin.h @@ -641,10 +641,9 @@ _mm_shuffle_pi8(__m64 __a, __m64 __b) { /// A 128-bit integer vector containing control bytes corresponding to /// positions in the destination. /// \returns A 128-bit integer vector containing the resultant values. -static __inline__ __m128i __DEFAULT_FN_ATTRS -_mm_sign_epi8(__m128i __a, __m128i __b) -{ - return (__m128i)__builtin_ia32_psignb128((__v16qi)__a, (__v16qi)__b); +static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR +_mm_sign_epi8(__m128i __a, __m128i __b) { + return (__m128i)__builtin_ia32_psignb128((__v16qi)__a, (__v16qi)__b); } /// For each 16-bit integer in the first source operand, perform one of @@ -667,10 +666,9 @@ _mm_sign_epi8(__m128i __a, __m128i __b) /// A 128-bit integer vector containing control words corresponding to /// positions in the destination. /// \returns A 128-bit integer vector containing the resultant values. -static __inline__ __m128i __DEFAULT_FN_ATTRS -_mm_sign_epi16(__m128i __a, __m128i __b) -{ - return (__m128i)__builtin_ia32_psignw128((__v8hi)__a, (__v8hi)__b); +static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR +_mm_sign_epi16(__m128i __a, __m128i __b) { + return (__m128i)__builtin_ia32_psignw128((__v8hi)__a, (__v8hi)__b); } /// For each 32-bit integer in the first source operand, perform one of @@ -693,10 +691,9 @@ _mm_sign_epi16(__m128i __a, __m128i __b) /// A 128-bit integer vector containing control doublewords corresponding to /// positions in the destination. /// \returns A 128-bit integer vector containing the resultant values. -static __inline__ __m128i __DEFAULT_FN_ATTRS -_mm_sign_epi32(__m128i __a, __m128i __b) -{ - return (__m128i)__builtin_ia32_psignd128((__v4si)__a, (__v4si)__b); +static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR +_mm_sign_epi32(__m128i __a, __m128i __b) { + return (__m128i)__builtin_ia32_psignd128((__v4si)__a, (__v4si)__b); } /// For each 8-bit integer in the first source operand, perform one of @@ -719,11 +716,10 @@ _mm_sign_epi32(__m128i __a, __m128i __b) /// A 64-bit integer vector containing control bytes corresponding to /// positions in the destination. /// \returns A 64-bit integer vector containing the resultant values. -static __inline__ __m64 __DEFAULT_FN_ATTRS -_mm_sign_pi8(__m64 __a, __m64 __b) -{ - return __trunc64(__builtin_ia32_psignb128((__v16qi)__anyext128(__a), - (__v16qi)__anyext128(__b))); +static __inline__ __m64 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_sign_pi8(__m64 __a, + __m64 __b) { + return __trunc64(__builtin_ia32_psignb128((__v16qi)__zext128(__a), + (__v16qi)__zext128(__b))); } /// For each 16-bit integer in the first source operand, perform one of @@ -746,11 +742,10 @@ _mm_sign_pi8(__m64 __a, __m64 __b) /// A 64-bit integer vector containing control words corresponding to /// positions in the destination. /// \returns A 64-bit integer vector containing the resultant values. -static __inline__ __m64 __DEFAULT_FN_ATTRS -_mm_sign_pi16(__m64 __a, __m64 __b) -{ - return __trunc64(__builtin_ia32_psignw128((__v8hi)__anyext128(__a), - (__v8hi)__anyext128(__b))); +static __inline__ __m64 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_sign_pi16(__m64 __a, + __m64 __b) { + return __trunc64( + __builtin_ia32_psignw128((__v8hi)__zext128(__a), (__v8hi)__zext128(__b))); } /// For each 32-bit integer in the first source operand, perform one of @@ -773,11 +768,10 @@ _mm_sign_pi16(__m64 __a, __m64 __b) /// A 64-bit integer vector containing two control doublewords corresponding /// to positions in the destination. /// \returns A 64-bit integer vector containing the resultant values. -static __inline__ __m64 __DEFAULT_FN_ATTRS -_mm_sign_pi32(__m64 __a, __m64 __b) -{ - return __trunc64(__builtin_ia32_psignd128((__v4si)__anyext128(__a), - (__v4si)__anyext128(__b))); +static __inline__ __m64 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_sign_pi32(__m64 __a, + __m64 __b) { + return __trunc64( + __builtin_ia32_psignd128((__v4si)__zext128(__a), (__v4si)__zext128(__b))); } #undef __anyext128 diff --git a/clang/test/CodeGen/X86/avx2-builtins.c b/clang/test/CodeGen/X86/avx2-builtins.c index 03b1bdeb55d7c..5c52d84de56f5 100644 --- a/clang/test/CodeGen/X86/avx2-builtins.c +++ b/clang/test/CodeGen/X86/avx2-builtins.c @@ -1155,23 +1155,30 @@ __m256i test_mm256_shufflelo_epi16(__m256i a) { return _mm256_shufflelo_epi16(a, 83); } TEST_CONSTEXPR(match_v16hi(_mm256_shufflelo_epi16(((__m256i)(__v16hi){ 0,1,2,3, 4,5,6,7, 8,9,10,11, 12,13,14,15}), 83), 3,0,1,1, 4,5,6,7, 11,8,9,9, 12,13,14,15) ); + __m256i test_mm256_sign_epi8(__m256i a, __m256i b) { // CHECK-LABEL: test_mm256_sign_epi8 // CHECK: call <32 x i8> @llvm.x86.avx2.psign.b(<32 x i8> %{{.*}}, <32 x i8> %{{.*}}) return _mm256_sign_epi8(a, b); } +TEST_CONSTEXPR(match_v32qi(_mm256_sign_epi8( + (__m256i)(__v32qs){0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1}, + (__m256i)(__v32qs){0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,-1}), + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,-1)); __m256i test_mm256_sign_epi16(__m256i a, __m256i b) { // CHECK-LABEL: test_mm256_sign_epi16 // CHECK: call <16 x i16> @llvm.x86.avx2.psign.w(<16 x i16> %{{.*}}, <16 x i16> %{{.*}}) return _mm256_sign_epi16(a, b); } +TEST_CONSTEXPR(match_v16hi(_mm256_sign_epi16((__m256i)(__v16hi){0x77,0x77,0xbe,0xbe, -0x9,-0x9,-0x8,-0x8, 0,0,0,0, 0,0,0,0}, (__m256i)(__v16hi){-1,-256,1,256, -512,-1028,512,1028, -2048,-4096,0,0, 0,0,0,0}), -0x77,-0x77,0xbe,0xbe, 0x9,0x9,-0x8,-0x8, 0,0,0,0, 0,0,0,0)); __m256i test_mm256_sign_epi32(__m256i a, __m256i b) { // CHECK-LABEL: test_mm256_sign_epi32 // CHECK: call <8 x i32> @llvm.x86.avx2.psign.d(<8 x i32> %{{.*}}, <8 x i32> %{{.*}}) return _mm256_sign_epi32(a, b); } +TEST_CONSTEXPR(match_v8si(_mm256_sign_epi32((__m256i)(__v8si){0xbeef,0xfeed,0xbead,0xdeed, -1,2,-3,4}, (__m256i)(__v8si){0,0,0,0,-1,-1,-1,-1}), 0,0,0,0, 1,-2,3,-4)); __m256i test_mm256_slli_epi16(__m256i a) { // CHECK-LABEL: test_mm256_slli_epi16 diff --git a/clang/test/CodeGen/X86/mmx-builtins.c b/clang/test/CodeGen/X86/mmx-builtins.c index 767425577a130..273138063a1b1 100644 --- a/clang/test/CodeGen/X86/mmx-builtins.c +++ b/clang/test/CodeGen/X86/mmx-builtins.c @@ -602,23 +602,28 @@ __m64 test_mm_shuffle_pi16(__m64 a) { return _mm_shuffle_pi16(a, 3); } TEST_CONSTEXPR(match_v4hi(_mm_shuffle_pi16(((__m64)(__v4hi){0,1,2,3}), 3), 3,0,0,0)); + __m64 test_mm_sign_pi8(__m64 a, __m64 b) { // CHECK-LABEL: test_mm_sign_pi8 // CHECK: call <16 x i8> @llvm.x86.ssse3.psign.b.128( return _mm_sign_pi8(a, b); } +TEST_CONSTEXPR(match_v8qi(_mm_sign_pi8((__m64)(__v8qi){0,0,0,0, 0,0,0,0}, (__m64)(__v8qi){0,0,0,0, 0,0,0,0}), 0,0,0,0, 0,0,0,0)); +TEST_CONSTEXPR(match_v8qi(_mm_sign_pi8((__m64)(__v8qi){6,7,6,7, 6,7,6,7}, (__m64)(__v8qi){1,1,1,1, 0,0,0,0}), 6,7,6,7, 0,0,0,0)); __m64 test_mm_sign_pi16(__m64 a, __m64 b) { // CHECK-LABEL: test_mm_sign_pi16 // CHECK: call <8 x i16> @llvm.x86.ssse3.psign.w.128( return _mm_sign_pi16(a, b); } +TEST_CONSTEXPR(match_v4hi(_mm_sign_pi16((__m64)(__v4hi){-1,0,1,0}, (__m64)(__v4hi){1,0,-1,0}), -1,0,-1,0)); __m64 test_mm_sign_pi32(__m64 a, __m64 b) { // CHECK-LABEL: test_mm_sign_pi32 // CHECK: call <4 x i32> @llvm.x86.ssse3.psign.d.128( return _mm_sign_pi32(a, b); } +TEST_CONSTEXPR(match_v2si(_mm_sign_pi32((__m64)(__v2si){0x7FFF, -1}, (__m64)(__v2si){-1, 0x7FFF}), -0x7FFF, -1)); __m64 test_mm_sll_pi16(__m64 a, __m64 b) { // CHECK-LABEL: test_mm_sll_pi16 diff --git a/clang/test/CodeGen/X86/ssse3-builtins.c b/clang/test/CodeGen/X86/ssse3-builtins.c index f70afc01a1963..b7a4a2fe7ccd7 100644 --- a/clang/test/CodeGen/X86/ssse3-builtins.c +++ b/clang/test/CodeGen/X86/ssse3-builtins.c @@ -125,15 +125,18 @@ __m128i test_mm_sign_epi8(__m128i a, __m128i b) { // CHECK: call <16 x i8> @llvm.x86.ssse3.psign.b.128(<16 x i8> %{{.*}}, <16 x i8> %{{.*}}) return _mm_sign_epi8(a, b); } +TEST_CONSTEXPR(match_v16qi(_mm_sign_epi8((__m128i)(__v16qs){11,0,13,14, 0,16,17,18, 19,20,21,22, 23,24,25,26}, (__m128i)(__v16qs){0,1,0,1, -1,1,0,0, 0,0,1,1, -1,0,-1,0}), 0,0,0,14, 0,16,0,0, 0,0,21,22, -23,0,-25,0)); __m128i test_mm_sign_epi16(__m128i a, __m128i b) { // CHECK-LABEL: test_mm_sign_epi16 // CHECK: call <8 x i16> @llvm.x86.ssse3.psign.w.128(<8 x i16> %{{.*}}, <8 x i16> %{{.*}}) return _mm_sign_epi16(a, b); } +TEST_CONSTEXPR(match_v8hi(_mm_sign_epi16((__m128i)(__v8hi){0,-2,0,-4,0,-6,0,-8}, (__m128i)(__v8hi){-1,-2,-3,-4,-5,-6,7,-8}), 0,2,0,4,0,6,0,8)); __m128i test_mm_sign_epi32(__m128i a, __m128i b) { // CHECK-LABEL: test_mm_sign_epi32 // CHECK: call <4 x i32> @llvm.x86.ssse3.psign.d.128(<4 x i32> %{{.*}}, <4 x i32> %{{.*}}) return _mm_sign_epi32(a, b); } +TEST_CONSTEXPR(match_v4si(_mm_sign_epi32((__m128i)(__v4si){-1,-2,-3,-4}, (__m128i)(__v4si){-4,-3,-2,-1}), 1,2,3,4));