mirrors/flac
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

libFLAC: Add a workaround for a bug in MSVC2105 update2

Browse Source 
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>
This commit is contained in:
Erik de Castro Lopo 2016-05-05 17:21:20 +10:00
parent 387992bd64
commit 94a61241b0
3 changed files with 12 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/libFLAC/stream_encoder_intrin_avx2.c
View File

@ -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);
sum128 = _mm_hadd_epi32(sum128, sum128); sum128 = _mm_hadd_epi32(sum128, sum128);
abs_residual_partition_sums[partition] = (FLAC__uint32)_mm_cvtsi128_si32(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 */ else { /* have to pessimistically use 64 bits for accumulator */

4
src/libFLAC/stream_encoder_intrin_sse2.c
View File

@ -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, 8));
mm_sum = _mm_add_epi32(mm_sum, _mm_srli_si128(mm_sum, 4)); 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); 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 */ else { /* have to pessimistically use 64 bits for accumulator */

4
src/libFLAC/stream_encoder_intrin_ssse3.c
View File

@ -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);
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); 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 */ else { /* have to pessimistically use 64 bits for accumulator */