Intrinsics: fix extract & insert when index is out of bound.

Now, all extract & insert intrinsics should have the correct and operation
to ignore higher bits.

rdar://15250497

llvm-svn: 193267
This commit is contained in:
Manman Ren 2013-10-23 20:33:14 +00:00
parent 1e6d8ac59b
commit c94122e05b
4 changed files with 53 additions and 11 deletions

View File

@ -435,21 +435,21 @@ static __inline int __attribute__((__always_inline__, __nodebug__))
_mm256_extract_epi32(__m256i __a, int const __imm)
{
__v8si __b = (__v8si)__a;
return __b[__imm];
return __b[__imm & 7];
}
static __inline int __attribute__((__always_inline__, __nodebug__))
_mm256_extract_epi16(__m256i __a, int const __imm)
{
__v16hi __b = (__v16hi)__a;
return __b[__imm];
return __b[__imm & 15];
}
static __inline int __attribute__((__always_inline__, __nodebug__))
_mm256_extract_epi8(__m256i __a, int const __imm)
{
__v32qi __b = (__v32qi)__a;
return __b[__imm];
return __b[__imm & 31];
}
#ifdef __x86_64__
@ -457,7 +457,7 @@ static __inline long long __attribute__((__always_inline__, __nodebug__))
_mm256_extract_epi64(__m256i __a, const int __imm)
{
__v4di __b = (__v4di)__a;
return __b[__imm];
return __b[__imm & 3];
}
#endif

View File

@ -197,7 +197,7 @@ _mm_max_epu32 (__m128i __V1, __m128i __V2)
#define _mm_extract_ps(X, N) (__extension__ \
({ union { int __i; float __f; } __t; \
__v4sf __a = (__v4sf)(X); \
__t.__f = __a[N]; \
__t.__f = __a[(N) & 3]; \
__t.__i;}))
/* Miscellaneous insert and extract macros. */
@ -215,14 +215,14 @@ _mm_max_epu32 (__m128i __V1, __m128i __V2)
/* Insert int into packed integer array at index. */
#define _mm_insert_epi8(X, I, N) (__extension__ ({ __v16qi __a = (__v16qi)(X); \
__a[(N)] = (I); \
__a[(N) & 15] = (I); \
__a;}))
#define _mm_insert_epi32(X, I, N) (__extension__ ({ __v4si __a = (__v4si)(X); \
__a[(N)] = (I); \
__a[(N) & 3] = (I); \
__a;}))
#ifdef __x86_64__
#define _mm_insert_epi64(X, I, N) (__extension__ ({ __v2di __a = (__v2di)(X); \
__a[(N)] = (I); \
__a[(N) & 1] = (I); \
__a;}))
#endif /* __x86_64__ */
@ -231,12 +231,12 @@ _mm_max_epu32 (__m128i __V1, __m128i __V2)
*/
#define _mm_extract_epi8(X, N) (__extension__ ({ __v16qi __a = (__v16qi)(X); \
(int)(unsigned char) \
__a[(N)];}))
__a[(N) & 15];}))
#define _mm_extract_epi32(X, N) (__extension__ ({ __v4si __a = (__v4si)(X); \
__a[(N)];}))
__a[(N) & 3];}))
#ifdef __x86_64__
#define _mm_extract_epi64(X, N) (__extension__ ({ __v2di __a = (__v2di)(X); \
__a[(N)];}))
__a[(N) & 1];}))
#endif /* __x86_64 */
/* SSE4 128-bit Packed Integer Comparisons. */

View File

@ -93,3 +93,21 @@ int test_mm_cmpistrz(__m128i A, __m128i B) {
// CHECK: @llvm.x86.sse42.pcmpistriz128
return _mm_cmpistrz(A, B, 7);
}
int test_extract_epi32(__m256i __a) {
// CHECK-LABEL: @test_extract_epi32
// CHECK: extractelement <8 x i32> %{{.*}}, i32 0
return _mm256_extract_epi32(__a, 8);
}
int test_extract_epi16(__m256i __a) {
// CHECK-LABEL: @test_extract_epi16
// CHECK: extractelement <16 x i16> %{{.*}}, i32 0
return _mm256_extract_epi16(__a, 16);
}
int test_extract_epi8(__m256i __a) {
// CHECK-LABEL: @test_extract_epi8
// CHECK: extractelement <32 x i8> %{{.*}}, i32 0
return _mm256_extract_epi8(__a, 32);
}

View File

@ -213,3 +213,27 @@ void test_extract_epi16(__m128i __a) {
// CHECK: extractelement <8 x i16> %{{.*}}, i32 [[x]]
_mm_extract_epi16(__a, 8);
}
int test_extract_ps(__m128i __a) {
// CHECK-LABEL: @test_extract_ps
// CHECK: extractelement <4 x float> %{{.*}}, i32 0
return _mm_extract_ps(__a, 4);
}
int test_extract_epi8(__m128i __a) {
// CHECK-LABEL: @test_extract_epi8
// CHECK: extractelement <16 x i8> %{{.*}}, i32 0
return _mm_extract_epi8(__a, 16);
}
int test_extract_epi32(__m128i __a) {
// CHECK-LABEL: @test_extract_epi32
// CHECK: extractelement <4 x i32> %{{.*}}, i32 0
return _mm_extract_epi32(__a, 4);
}
void test_insert_epi32(__m128i __a, int b) {
// CHECK-LABEL: @test_insert_epi32
// CHECK: insertelement <4 x i32> %{{.*}}, i32 %{{.*}}, i32 0
_mm_insert_epi32(__a, b, 4);
}