diff options
author | Aurelien Jarno <aurelien@aurel32.net> | 2022-10-03 23:16:46 +0200 |
---|---|---|
committer | Aurelien Jarno <aurelien@aurel32.net> | 2022-10-04 00:03:19 +0200 |
commit | 050d652900d5f88447af335b20b621344f16e192 (patch) | |
tree | 5da3367d4fcbae73ed3a0c235ede18c5f7624b55 | |
parent | 26d4359b448a39054e627ad072a7480dd24f0313 (diff) | |
download | glibc-050d652900d5f88447af335b20b621344f16e192.tar.gz glibc-050d652900d5f88447af335b20b621344f16e192.tar.xz glibc-050d652900d5f88447af335b20b621344f16e192.zip |
x86-64: Require BMI2 for AVX2 strncmp implementation
The AVX2 strncmp implementations uses the 'bzhi' instruction, which belongs to the BMI2 CPU feature. NB: It also uses the 'tzcnt' BMI1 instruction, but it is executed as BSF as BSF if the CPU doesn't support TZCNT, and produces the same result for non-zero input. Partially fixes: b77b06e0e296 ("x86: Optimize strcmp-avx2.S") Partially resolves: BZ #29611 Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com> (cherry picked from commit fc7de1d9b99ae1676bc626ddca422d7abee0eb48)
-rw-r--r-- | sysdeps/x86_64/multiarch/ifunc-impl-list.c | 7 | ||||
-rw-r--r-- | sysdeps/x86_64/multiarch/strncmp.c | 4 |
2 files changed, 7 insertions, 4 deletions
diff --git a/sysdeps/x86_64/multiarch/ifunc-impl-list.c b/sysdeps/x86_64/multiarch/ifunc-impl-list.c index 70931f1598..34d5f6efe5 100644 --- a/sysdeps/x86_64/multiarch/ifunc-impl-list.c +++ b/sysdeps/x86_64/multiarch/ifunc-impl-list.c @@ -1022,15 +1022,18 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array, /* Support sysdeps/x86_64/multiarch/strncmp.c. */ IFUNC_IMPL (i, name, strncmp, IFUNC_IMPL_ADD (array, i, strncmp, - CPU_FEATURE_USABLE (AVX2), + (CPU_FEATURE_USABLE (AVX2) + && CPU_FEATURE_USABLE (BMI2)), __strncmp_avx2) IFUNC_IMPL_ADD (array, i, strncmp, (CPU_FEATURE_USABLE (AVX2) + && CPU_FEATURE_USABLE (BMI2) && CPU_FEATURE_USABLE (RTM)), __strncmp_avx2_rtm) IFUNC_IMPL_ADD (array, i, strncmp, (CPU_FEATURE_USABLE (AVX512VL) - && CPU_FEATURE_USABLE (AVX512BW)), + && CPU_FEATURE_USABLE (AVX512BW) + && CPU_FEATURE_USABLE (BMI2)), __strncmp_evex) IFUNC_IMPL_ADD (array, i, strncmp, CPU_FEATURE_USABLE (SSE4_2), __strncmp_sse42) diff --git a/sysdeps/x86_64/multiarch/strncmp.c b/sysdeps/x86_64/multiarch/strncmp.c index f94a421784..7632d7b2dd 100644 --- a/sysdeps/x86_64/multiarch/strncmp.c +++ b/sysdeps/x86_64/multiarch/strncmp.c @@ -39,11 +39,11 @@ IFUNC_SELECTOR (void) const struct cpu_features* cpu_features = __get_cpu_features (); if (CPU_FEATURE_USABLE_P (cpu_features, AVX2) + && CPU_FEATURE_USABLE_P (cpu_features, BMI2) && CPU_FEATURES_ARCH_P (cpu_features, AVX_Fast_Unaligned_Load)) { if (CPU_FEATURE_USABLE_P (cpu_features, AVX512VL) - && CPU_FEATURE_USABLE_P (cpu_features, AVX512BW) - && CPU_FEATURE_USABLE_P (cpu_features, BMI2)) + && CPU_FEATURE_USABLE_P (cpu_features, AVX512BW)) return OPTIMIZE (evex); if (CPU_FEATURE_USABLE_P (cpu_features, RTM)) |