diff options
author | Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com> | 2017-12-08 11:25:13 +0000 |
---|---|---|
committer | Damjan Marion <dmarion.lists@gmail.com> | 2017-12-08 15:18:09 +0000 |
commit | 20ec7165007bd38359ea516fe5389e9e828a9d62 (patch) | |
tree | 740807f77695c480fe3da5eca00c52650a9b12ac /src/vppinfra | |
parent | b716308ca558e8fcc1297903bf78c1673af44154 (diff) |
vppinfra: fix issues depending on compiler
It looks like different compiler versions produce different results for
expressions like "(cast) ptr + inc".
Use parenthesis to avoid such issues.
Change-Id: I93a9883bf5fc05ae462df5b004817775f0739405
Signed-off-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
Diffstat (limited to 'src/vppinfra')
-rw-r--r-- | src/vppinfra/string.h | 68 |
1 files changed, 34 insertions, 34 deletions
diff --git a/src/vppinfra/string.h b/src/vppinfra/string.h index 4d2ff7875a2..0e0627d088d 100644 --- a/src/vppinfra/string.h +++ b/src/vppinfra/string.h @@ -95,46 +95,46 @@ clib_memcpy64_x4 (void *d0, void *d1, void *d2, void *d3, void *s) _mm512_storeu_si512 (d3, r0); #elif defined (__AVX2__) - __m256i r0 = _mm256_loadu_si256 ((__m256i *) s + 0 * 32); - __m256i r1 = _mm256_loadu_si256 ((__m256i *) s + 1 * 32); + __m256i r0 = _mm256_loadu_si256 ((__m256i *) (s + 0 * 32)); + __m256i r1 = _mm256_loadu_si256 ((__m256i *) (s + 1 * 32)); - _mm256_storeu_si256 ((__m256i *) d0 + 0 * 32, r0); - _mm256_storeu_si256 ((__m256i *) d0 + 1 * 32, r1); + _mm256_storeu_si256 ((__m256i *) (d0 + 0 * 32), r0); + _mm256_storeu_si256 ((__m256i *) (d0 + 1 * 32), r1); - _mm256_storeu_si256 ((__m256i *) d1 + 0 * 32, r0); - _mm256_storeu_si256 ((__m256i *) d1 + 1 * 32, r1); + _mm256_storeu_si256 ((__m256i *) (d1 + 0 * 32), r0); + _mm256_storeu_si256 ((__m256i *) (d1 + 1 * 32), r1); - _mm256_storeu_si256 ((__m256i *) d2 + 0 * 32, r0); - _mm256_storeu_si256 ((__m256i *) d2 + 1 * 32, r1); + _mm256_storeu_si256 ((__m256i *) (d2 + 0 * 32), r0); + _mm256_storeu_si256 ((__m256i *) (d2 + 1 * 32), r1); - _mm256_storeu_si256 ((__m256i *) d3 + 0 * 32, r0); - _mm256_storeu_si256 ((__m256i *) d3 + 1 * 32, r1); + _mm256_storeu_si256 ((__m256i *) (d3 + 0 * 32), r0); + _mm256_storeu_si256 ((__m256i *) (d3 + 1 * 32), r1); #elif defined (__SSSE3__) - __m128i r0 = _mm_loadu_si128 ((__m128i *) s + 0 * 16); - __m128i r1 = _mm_loadu_si128 ((__m128i *) s + 1 * 16); - __m128i r2 = _mm_loadu_si128 ((__m128i *) s + 2 * 16); - __m128i r3 = _mm_loadu_si128 ((__m128i *) s + 3 * 16); - - _mm_storeu_si128 ((__m128i *) d0 + 0 * 16, r0); - _mm_storeu_si128 ((__m128i *) d0 + 1 * 16, r1); - _mm_storeu_si128 ((__m128i *) d0 + 2 * 16, r2); - _mm_storeu_si128 ((__m128i *) d0 + 3 * 16, r3); - - _mm_storeu_si128 ((__m128i *) d1 + 0 * 16, r0); - _mm_storeu_si128 ((__m128i *) d1 + 1 * 16, r1); - _mm_storeu_si128 ((__m128i *) d1 + 2 * 16, r2); - _mm_storeu_si128 ((__m128i *) d1 + 3 * 16, r3); - - _mm_storeu_si128 ((__m128i *) d2 + 0 * 16, r0); - _mm_storeu_si128 ((__m128i *) d2 + 1 * 16, r1); - _mm_storeu_si128 ((__m128i *) d2 + 2 * 16, r2); - _mm_storeu_si128 ((__m128i *) d2 + 3 * 16, r3); - - _mm_storeu_si128 ((__m128i *) d3 + 0 * 16, r0); - _mm_storeu_si128 ((__m128i *) d3 + 1 * 16, r1); - _mm_storeu_si128 ((__m128i *) d3 + 2 * 16, r2); - _mm_storeu_si128 ((__m128i *) d3 + 3 * 16, r3); + __m128i r0 = _mm_loadu_si128 ((__m128i *) (s + 0 * 16)); + __m128i r1 = _mm_loadu_si128 ((__m128i *) (s + 1 * 16)); + __m128i r2 = _mm_loadu_si128 ((__m128i *) (s + 2 * 16)); + __m128i r3 = _mm_loadu_si128 ((__m128i *) (s + 3 * 16)); + + _mm_storeu_si128 ((__m128i *) (d0 + 0 * 16), r0); + _mm_storeu_si128 ((__m128i *) (d0 + 1 * 16), r1); + _mm_storeu_si128 ((__m128i *) (d0 + 2 * 16), r2); + _mm_storeu_si128 ((__m128i *) (d0 + 3 * 16), r3); + + _mm_storeu_si128 ((__m128i *) (d1 + 0 * 16), r0); + _mm_storeu_si128 ((__m128i *) (d1 + 1 * 16), r1); + _mm_storeu_si128 ((__m128i *) (d1 + 2 * 16), r2); + _mm_storeu_si128 ((__m128i *) (d1 + 3 * 16), r3); + + _mm_storeu_si128 ((__m128i *) (d2 + 0 * 16), r0); + _mm_storeu_si128 ((__m128i *) (d2 + 1 * 16), r1); + _mm_storeu_si128 ((__m128i *) (d2 + 2 * 16), r2); + _mm_storeu_si128 ((__m128i *) (d2 + 3 * 16), r3); + + _mm_storeu_si128 ((__m128i *) (d3 + 0 * 16), r0); + _mm_storeu_si128 ((__m128i *) (d3 + 1 * 16), r1); + _mm_storeu_si128 ((__m128i *) (d3 + 2 * 16), r2); + _mm_storeu_si128 ((__m128i *) (d3 + 3 * 16), r3); #else clib_memcpy (d0, s, 64); |