From 0569936773c861c791f10bba5e2f4cac5fbb4e78 Mon Sep 17 00:00:00 2001 From: "H.J. Lu" Date: Tue, 2 Oct 2012 05:05:17 -0700 Subject: [PATCH] Define HAS_FMA with bit_FMA_Usable --- ChangeLog | 12 ++++++++++++ NEWS | 2 +- sysdeps/x86_64/multiarch/init-arch.c | 3 +++ sysdeps/x86_64/multiarch/init-arch.h | 9 +++++++-- 4 files changed, 23 insertions(+), 3 deletions(-) diff --git a/ChangeLog b/ChangeLog index f5183fac7a..d121446ac1 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,15 @@ +2012-10-02 H.J. Lu + + [BZ #14648] + * sysdeps/x86_64/multiarch/init-arch.c (__init_cpu_features): + Set bit_FMA_Usable if FMA is supported. + * sysdeps/x86_64/multiarch/init-arch.h (bit_FMA_Usable): New + macro. + (bit_FMA4_Usable): Updated. + (index_FMA_Usable): New macro. + (CPUID_FMA): Likewise + (HAS_FMA): Defined with bit_FMA_Usable. + 2012-10-01 Roland McGrath * bits/types.h (__swblk_t): Type removed. diff --git a/NEWS b/NEWS index fe7c78d4f6..3d62725666 100644 --- a/NEWS +++ b/NEWS @@ -15,7 +15,7 @@ Version 2.17 14195, 14237, 14252, 14283, 14298, 14303, 14307, 14328, 14331, 14336, 14337, 14347, 14349, 14376, 14459, 14476, 14477, 14505, 14510, 14516, 14518, 14519, 14530, 14532, 14538, 14543, 14544, 14545, 14562, 14576, - 14579, 14583, 14587, 14621, 14638, 14645. + 14579, 14583, 14587, 14621, 14638, 14645, 14648. * Support for STT_GNU_IFUNC symbols added for s390 and s390x. Optimized versions of memcpy, memset, and memcmp added for System z10 and diff --git a/sysdeps/x86_64/multiarch/init-arch.c b/sysdeps/x86_64/multiarch/init-arch.c index fb44dcfcf6..fed5ab8982 100644 --- a/sysdeps/x86_64/multiarch/init-arch.c +++ b/sysdeps/x86_64/multiarch/init-arch.c @@ -156,6 +156,9 @@ __init_cpu_features (void) /* Determine if AVX is usable. */ if (CPUID_AVX) __cpu_features.feature[index_AVX_Usable] |= bit_AVX_Usable; + /* Determine if FMA is usable. */ + if (CPUID_FMA) + __cpu_features.feature[index_FMA_Usable] |= bit_FMA_Usable; /* Determine if FMA4 is usable. */ if (CPUID_FMA4) __cpu_features.feature[index_FMA4_Usable] |= bit_FMA4_Usable; diff --git a/sysdeps/x86_64/multiarch/init-arch.h b/sysdeps/x86_64/multiarch/init-arch.h index 45e2651678..f33f1c81e5 100644 --- a/sysdeps/x86_64/multiarch/init-arch.h +++ b/sysdeps/x86_64/multiarch/init-arch.h @@ -22,7 +22,8 @@ #define bit_Fast_Unaligned_Load (1 << 4) #define bit_Prefer_PMINUB_for_stringop (1 << 5) #define bit_AVX_Usable (1 << 6) -#define bit_FMA4_Usable (1 << 7) +#define bit_FMA_Usable (1 << 7) +#define bit_FMA4_Usable (1 << 8) /* CPUID Feature flags. */ #define bit_SSE2 (1 << 26) @@ -56,6 +57,7 @@ # define index_Fast_Unaligned_Load FEATURE_INDEX_1*FEATURE_SIZE # define index_Prefer_PMINUB_for_stringop FEATURE_INDEX_1*FEATURE_SIZE # define index_AVX_Usable FEATURE_INDEX_1*FEATURE_SIZE +# define index_FMA_Usable FEATURE_INDEX_1*FEATURE_SIZE # define index_FMA4_Usable FEATURE_INDEX_1*FEATURE_SIZE #else /* __ASSEMBLER__ */ @@ -131,6 +133,8 @@ extern const struct cpu_features *__get_cpu_features (void) HAS_CPUID_FLAG (COMMON_CPUID_INDEX_1, ecx, bit_OSXSAVE) # define CPUID_AVX \ HAS_CPUID_FLAG (COMMON_CPUID_INDEX_1, ecx, bit_AVX) +# define CPUID_FMA \ + HAS_CPUID_FLAG (COMMON_CPUID_INDEX_1, ecx, bit_FMA) # define CPUID_FMA4 \ HAS_CPUID_FLAG (COMMON_CPUID_INDEX_80000001, ecx, bit_FMA4) @@ -140,7 +144,6 @@ extern const struct cpu_features *__get_cpu_features (void) # define HAS_SSSE3 HAS_CPU_FEATURE (COMMON_CPUID_INDEX_1, ecx, bit_SSSE3) # define HAS_SSE4_1 HAS_CPU_FEATURE (COMMON_CPUID_INDEX_1, ecx, bit_SSE4_1) # define HAS_SSE4_2 HAS_CPU_FEATURE (COMMON_CPUID_INDEX_1, ecx, bit_SSE4_2) -# define HAS_FMA HAS_CPU_FEATURE (COMMON_CPUID_INDEX_1, ecx, bit_FMA) # define index_Fast_Rep_String FEATURE_INDEX_1 # define index_Fast_Copy_Backward FEATURE_INDEX_1 @@ -148,6 +151,7 @@ extern const struct cpu_features *__get_cpu_features (void) # define index_Prefer_SSE_for_memop FEATURE_INDEX_1 # define index_Fast_Unaligned_Load FEATURE_INDEX_1 # define index_AVX_Usable FEATURE_INDEX_1 +# define index_FMA_Usable FEATURE_INDEX_1 # define index_FMA4_Usable FEATURE_INDEX_1 # define HAS_ARCH_FEATURE(name) \ @@ -159,6 +163,7 @@ extern const struct cpu_features *__get_cpu_features (void) # define HAS_PREFER_SSE_FOR_MEMOP HAS_ARCH_FEATURE (Prefer_SSE_for_memop) # define HAS_FAST_UNALIGNED_LOAD HAS_ARCH_FEATURE (Fast_Unaligned_Load) # define HAS_AVX HAS_ARCH_FEATURE (AVX_Usable) +# define HAS_FMA HAS_ARCH_FEATURE (FMA_Usable) # define HAS_FMA4 HAS_ARCH_FEATURE (FMA4_Usable) #endif /* __ASSEMBLER__ */ -- 2.11.4.GIT