diff options
| author | Chad Rosier <mcrosier@apple.com> | 2011-12-16 21:07:34 +0000 |
|---|---|---|
| committer | Chad Rosier <mcrosier@apple.com> | 2011-12-16 21:07:34 +0000 |
| commit | c5cda1121e270548ecf258d0ed72919a5211a94e (patch) | |
| tree | 363934e9af94bbb9204ebbe419fb8b24371cde46 | |
| parent | 80d39bf55f962278867f54696a033f1aa546d5a0 (diff) | |
Fix vperm2f128_* AVX intrinsics to use "I" (ICE) markings. Fix avxintrin.h to
take them into account.
rdar://10576962
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@146757 91177308-0d34-0410-b5e6-96231b3b80d8
| -rw-r--r-- | include/clang/Basic/BuiltinsX86.def | 6 | ||||
| -rw-r--r-- | lib/Headers/avxintrin.h | 27 |
2 files changed, 15 insertions, 18 deletions
diff --git a/include/clang/Basic/BuiltinsX86.def b/include/clang/Basic/BuiltinsX86.def index cfe4e6f5d1..2eb8b09b47 100644 --- a/include/clang/Basic/BuiltinsX86.def +++ b/include/clang/Basic/BuiltinsX86.def @@ -422,9 +422,9 @@ BUILTIN(__builtin_ia32_cvtps2pd256, "V4dV4f", "") BUILTIN(__builtin_ia32_cvttpd2dq256, "V4iV4d", "") BUILTIN(__builtin_ia32_cvtpd2dq256, "V4iV4d", "") BUILTIN(__builtin_ia32_cvttps2dq256, "V8iV8f", "") -BUILTIN(__builtin_ia32_vperm2f128_pd256, "V4dV4dV4dc", "") -BUILTIN(__builtin_ia32_vperm2f128_ps256, "V8fV8fV8fc", "") -BUILTIN(__builtin_ia32_vperm2f128_si256, "V8iV8iV8ic", "") +BUILTIN(__builtin_ia32_vperm2f128_pd256, "V4dV4dV4dIc", "") +BUILTIN(__builtin_ia32_vperm2f128_ps256, "V8fV8fV8fIc", "") +BUILTIN(__builtin_ia32_vperm2f128_si256, "V8iV8iV8iIc", "") BUILTIN(__builtin_ia32_vpermilpd, "V2dV2dc", "") BUILTIN(__builtin_ia32_vpermilps, "V4fV4fc", "") BUILTIN(__builtin_ia32_vpermilpd256, "V4dV4dc", "") diff --git a/lib/Headers/avxintrin.h b/lib/Headers/avxintrin.h index 85a2c747c2..3ae5dc76fc 100644 --- a/lib/Headers/avxintrin.h +++ b/lib/Headers/avxintrin.h @@ -286,23 +286,20 @@ _mm256_permute_ps(__m256 a, const int c) return (__m256)__builtin_ia32_vpermilps256((__v8sf)a, c); } -static __inline __m256d __attribute__((__always_inline__, __nodebug__)) -_mm256_permute2f128_pd(__m256d a, __m256d b, const int c) -{ - return (__m256d)__builtin_ia32_vperm2f128_pd256((__v4df)a, (__v4df)b, c); -} +#define _mm256_permute2f128_pd(V1, V2, M) __extension__ ({ \ + __m256d __V1 = (V1); \ + __m256d __V2 = (V2); \ + (__m256d)__builtin_ia32_vperm2f128_pd256((__v4df)__V1, (__v4df)__V2, M); }) -static __inline __m256 __attribute__((__always_inline__, __nodebug__)) -_mm256_permute2f128_ps(__m256 a, __m256 b, const int c) -{ - return (__m256)__builtin_ia32_vperm2f128_ps256((__v8sf)a, (__v8sf)b, c); -} +#define _mm256_permute2f128_ps(V1, V2, M) __extension__ ({ \ + __m256 __V1 = (V1); \ + __m256 __V2 = (V2); \ + (__m256)__builtin_ia32_vperm2f128_ps256((__v8sf)__V1, (__v8sf)__V2, M); }) -static __inline __m256i __attribute__((__always_inline__, __nodebug__)) -_mm256_permute2f128_si256(__m256i a, __m256i b, const int c) -{ - return (__m256i)__builtin_ia32_vperm2f128_si256((__v8si)a, (__v8si)b, c); -} +#define _mm256_permute2f128_si256(V1, V2, M) __extension__ ({ \ + __m256i __V1 = (V1); \ + __m256i __V2 = (V2); \ + (__m256i)__builtin_ia32_vperm2f128_si256((__v8si)__V1, (__v8si)__V2, M); }) /* Vector Blend */ #define _mm256_blend_pd(V1, V2, M) __extension__ ({ \ |