diff options
| author | Erik de Castro Lopo <erikd@mega-nerd.com> | 2016-05-05 17:21:20 +1000 |
|---|---|---|
| committer | Erik de Castro Lopo <erikd@mega-nerd.com> | 2016-05-05 17:23:52 +1000 |
| commit | 94a61241b02064c7d9fe508f72a742f2a90b8492 (patch) | |
| tree | 63929d6dc3541cf2dc6500114f507c3dc155a156 | |
| parent | 387992bd6411411739014e633a80cfe639e4db36 (diff) | |
| download | flac-94a61241b02064c7d9fe508f72a742f2a90b8492.tar.gz | |
libFLAC: Add a workaround for a bug in MSVC2105 update2
MSVC2105 update2 compiles the C code:
abs_residual_partition_sums[partition] =
(FLAC__uint32)_mm_cvtsi128_si32(mm_sum);
into this:
movq QWORD PTR [rsi], xmm2
while it should be:
movd eax, xmm2
mov QWORD PTR [rsi], rax
With this patch, MSVC emits:
movq QWORD PTR [rsi], xmm2
mov DWORD PTR [rsi+4], r9d
so the price of this workaround is 1 extra write instruction per
partition.
Patch-from: lvqcl <lvqcl.mail@gmail.com>
| -rw-r--r-- | src/libFLAC/stream_encoder_intrin_avx2.c | 4 | ||||
| -rw-r--r-- | src/libFLAC/stream_encoder_intrin_sse2.c | 4 | ||||
| -rw-r--r-- | src/libFLAC/stream_encoder_intrin_ssse3.c | 4 |
3 files changed, 12 insertions, 0 deletions
diff --git a/src/libFLAC/stream_encoder_intrin_avx2.c b/src/libFLAC/stream_encoder_intrin_avx2.c index d0cd0e49..4730bf55 100644 --- a/src/libFLAC/stream_encoder_intrin_avx2.c +++ b/src/libFLAC/stream_encoder_intrin_avx2.c @@ -84,6 +84,10 @@ void FLAC__precompute_partition_info_sums_intrin_avx2(const FLAC__int32 residual sum128 = _mm_hadd_epi32(sum128, sum128); sum128 = _mm_hadd_epi32(sum128, sum128); abs_residual_partition_sums[partition] = (FLAC__uint32)_mm_cvtsi128_si32(sum128); +/* workaround for a bug in MSVC2015U2 - see https://connect.microsoft.com/VisualStudio/feedback/details/2659191/incorrect-code-generation-for-x86-64 */ +#if (defined _MSC_VER) && (_MSC_FULL_VER == 190023918) && (defined FLAC__CPU_X86_64) + abs_residual_partition_sums[partition] &= 0xFFFFFFFF; /**/ +#endif } } else { /* have to pessimistically use 64 bits for accumulator */ diff --git a/src/libFLAC/stream_encoder_intrin_sse2.c b/src/libFLAC/stream_encoder_intrin_sse2.c index ed01978c..55f76640 100644 --- a/src/libFLAC/stream_encoder_intrin_sse2.c +++ b/src/libFLAC/stream_encoder_intrin_sse2.c @@ -98,6 +98,10 @@ void FLAC__precompute_partition_info_sums_intrin_sse2(const FLAC__int32 residual mm_sum = _mm_add_epi32(mm_sum, _mm_srli_si128(mm_sum, 8)); mm_sum = _mm_add_epi32(mm_sum, _mm_srli_si128(mm_sum, 4)); abs_residual_partition_sums[partition] = (FLAC__uint32)_mm_cvtsi128_si32(mm_sum); +/* workaround for a bug in MSVC2015U2 - see https://connect.microsoft.com/VisualStudio/feedback/details/2659191/incorrect-code-generation-for-x86-64 */ +#if (defined _MSC_VER) && (_MSC_FULL_VER == 190023918) && (defined FLAC__CPU_X86_64) + abs_residual_partition_sums[partition] &= 0xFFFFFFFF; +#endif } } else { /* have to pessimistically use 64 bits for accumulator */ diff --git a/src/libFLAC/stream_encoder_intrin_ssse3.c b/src/libFLAC/stream_encoder_intrin_ssse3.c index 465950b2..551ffd31 100644 --- a/src/libFLAC/stream_encoder_intrin_ssse3.c +++ b/src/libFLAC/stream_encoder_intrin_ssse3.c @@ -87,6 +87,10 @@ void FLAC__precompute_partition_info_sums_intrin_ssse3(const FLAC__int32 residua mm_sum = _mm_hadd_epi32(mm_sum, mm_sum); mm_sum = _mm_hadd_epi32(mm_sum, mm_sum); abs_residual_partition_sums[partition] = (FLAC__uint32)_mm_cvtsi128_si32(mm_sum); +/* workaround for a bug in MSVC2015U2 - see https://connect.microsoft.com/VisualStudio/feedback/details/2659191/incorrect-code-generation-for-x86-64 */ +#if (defined _MSC_VER) && (_MSC_FULL_VER == 190023918) && (defined FLAC__CPU_X86_64) + abs_residual_partition_sums[partition] &= 0xFFFFFFFF; +#endif } } else { /* have to pessimistically use 64 bits for accumulator */ |