[X86] _mm256_permutevar8x32_ps should take an integer vector for its shuffle index input.

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@254270 91177308-0d34-0410-b5e6-96231b3b80d8

include/clang/Basic/BuiltinsX86.def

@@ -597,7 +597,7 @@ TARGET_BUILTIN(__builtin_ia32_psrlqi256, "V4LLiV4LLii", "", "avx2")
 TARGET_BUILTIN(__builtin_ia32_psrlq256, "V4LLiV4LLiV2LLi", "", "avx2")
 TARGET_BUILTIN(__builtin_ia32_movntdqa256, "V4LLiV4LLiC*", "", "avx2")
 TARGET_BUILTIN(__builtin_ia32_permvarsi256, "V8iV8iV8i", "", "avx2")
-TARGET_BUILTIN(__builtin_ia32_permvarsf256, "V8fV8fV8f", "", "avx2")
+TARGET_BUILTIN(__builtin_ia32_permvarsf256, "V8fV8fV8i", "", "avx2")
 TARGET_BUILTIN(__builtin_ia32_permti256, "V4LLiV4LLiV4LLiIc", "", "avx2")
 TARGET_BUILTIN(__builtin_ia32_maskloadd256, "V8iV8iC*V8i", "", "avx2")
 TARGET_BUILTIN(__builtin_ia32_maskloadq256, "V4LLiV4LLiC*V4LLi", "", "avx2")

lib/Headers/avx2intrin.h

@@ -867,9 +867,9 @@ _mm256_permutevar8x32_epi32(__m256i __a, __m256i __b)
                                    ((M) & 0x30) >> 4, ((M) & 0xc0) >> 6); })
 
 static __inline__ __m256 __DEFAULT_FN_ATTRS
-_mm256_permutevar8x32_ps(__m256 __a, __m256 __b)
+_mm256_permutevar8x32_ps(__m256 __a, __m256i __b)
 {
-  return (__m256)__builtin_ia32_permvarsf256((__v8sf)__a, (__v8sf)__b);
+  return (__m256)__builtin_ia32_permvarsf256((__v8sf)__a, (__v8si)__b);
 }
 
 #define _mm256_permute4x64_epi64(V, M) __extension__ ({ \

test/CodeGen/avx2-builtins.c

@@ -862,7 +862,7 @@ __m256d test_mm256_permute4x64_pd(__m256d a) {
   return _mm256_permute4x64_pd(a, 25);
 }
 
-__m256 test_mm256_permutevar8x32_ps(__m256 a, __m256 b) {
+__m256 test_mm256_permutevar8x32_ps(__m256 a, __m256i b) {
   // CHECK: @llvm.x86.avx2.permps
   // CHECK-ASM: vpermps %ymm{{.*}}, %ymm{{.*}}, %ymm{{.*}}
   return _mm256_permutevar8x32_ps(a, b);