diff options
author | Noah Goldstein <goldstein.w.n@gmail.com> | 2022-07-13 16:33:00 -0700 |
---|---|---|
committer | Noah Goldstein <goldstein.w.n@gmail.com> | 2022-07-16 03:07:59 -0700 |
commit | 192979ee351315b84ff4277d0985f26243717cd7 (patch) | |
tree | a62f2737cc82add1047d89ef9e89984849a6de8d /sysdeps/x86_64/multiarch | |
parent | ceabdcd130ca7043b0fcf2676183d79431d10493 (diff) | |
download | glibc-192979ee351315b84ff4277d0985f26243717cd7.tar.gz glibc-192979ee351315b84ff4277d0985f26243717cd7.tar.xz glibc-192979ee351315b84ff4277d0985f26243717cd7.zip |
x86: Add support to build wcscpy with explicit ISA level
1. Add ISA level build guards to different implementations. - wcscpy-ssse3.S is used as ISA level 2/3/4. - wcscpy-generic.c is only used at ISA level 1 and will only build if compiled with ISA level == 1. Otherwise there is no reason to include it as we will always use wcscpy-ssse3.S 2. Refactor the ifunc selector and ifunc implementation list to use the ISA level aware wrapper macros that allow functions below the compiled ISA level (with a guranteed replacement) to be skipped. Tested with and without multiarch on x86_64 for ISA levels: {generic, x86-64-v2, x86-64-v3, x86-64-v4} And m32 with and without multiarch.
Diffstat (limited to 'sysdeps/x86_64/multiarch')
-rw-r--r-- | sysdeps/x86_64/multiarch/Makefile | 1 | ||||
-rw-r--r-- | sysdeps/x86_64/multiarch/ifunc-impl-list.c | 12 | ||||
-rw-r--r-- | sysdeps/x86_64/multiarch/wcscpy-generic.c | 9 | ||||
-rw-r--r-- | sysdeps/x86_64/multiarch/wcscpy-ssse3.S | 16 | ||||
-rw-r--r-- | sysdeps/x86_64/multiarch/wcscpy.c | 5 |
5 files changed, 31 insertions, 12 deletions
diff --git a/sysdeps/x86_64/multiarch/Makefile b/sysdeps/x86_64/multiarch/Makefile index ba29a65716..df4601c294 100644 --- a/sysdeps/x86_64/multiarch/Makefile +++ b/sysdeps/x86_64/multiarch/Makefile @@ -134,7 +134,6 @@ sysdep_routines += \ wcscmp-avx2-rtm \ wcscmp-evex \ wcscmp-sse2 \ - wcscpy-generic \ wcscpy-ssse3 \ wcslen-avx2 \ wcslen-avx2-rtm \ diff --git a/sysdeps/x86_64/multiarch/ifunc-impl-list.c b/sysdeps/x86_64/multiarch/ifunc-impl-list.c index 3b1df9b73c..9318e98cc8 100644 --- a/sysdeps/x86_64/multiarch/ifunc-impl-list.c +++ b/sysdeps/x86_64/multiarch/ifunc-impl-list.c @@ -791,9 +791,15 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array, /* Support sysdeps/x86_64/multiarch/wcscpy.c. */ IFUNC_IMPL (i, name, wcscpy, - IFUNC_IMPL_ADD (array, i, wcscpy, CPU_FEATURE_USABLE (SSSE3), - __wcscpy_ssse3) - IFUNC_IMPL_ADD (array, i, wcscpy, 1, __wcscpy_generic)) + /* ISA V4 wrapper for SSSE3 implementation because + the SSSE3 implementation is also used at ISA + level 3/4. */ + X86_IFUNC_IMPL_ADD_V4 (array, i, wcscpy, + CPU_FEATURE_USABLE (SSSE3), + __wcscpy_ssse3) + X86_IFUNC_IMPL_ADD_V1 (array, i, wcscpy, + 1, + __wcscpy_generic)) /* Support sysdeps/x86_64/multiarch/wcslen.c. */ IFUNC_IMPL (i, name, wcslen, diff --git a/sysdeps/x86_64/multiarch/wcscpy-generic.c b/sysdeps/x86_64/multiarch/wcscpy-generic.c index 5ea905f33f..93d314aaad 100644 --- a/sysdeps/x86_64/multiarch/wcscpy-generic.c +++ b/sysdeps/x86_64/multiarch/wcscpy-generic.c @@ -17,8 +17,11 @@ <https://www.gnu.org/licenses/>. */ -#if IS_IN (libc) +#include <isa-level.h> + +#if ISA_SHOULD_BUILD (1) + # define WCSCPY __wcscpy_generic -#endif +# include <wcsmbs/wcscpy.c> -#include <wcsmbs/wcscpy.c> +#endif diff --git a/sysdeps/x86_64/multiarch/wcscpy-ssse3.S b/sysdeps/x86_64/multiarch/wcscpy-ssse3.S index aa2b9d030f..62741071e6 100644 --- a/sysdeps/x86_64/multiarch/wcscpy-ssse3.S +++ b/sysdeps/x86_64/multiarch/wcscpy-ssse3.S @@ -16,11 +16,21 @@ License along with the GNU C Library; if not, see <https://www.gnu.org/licenses/>. */ -#if IS_IN (libc) +#include <isa-level.h> + +/* MINIMUM_X86_ISA_LEVEL <= 4 because there are not V3/V4 + implementations so we need this to build for ISA V3/V4 + builds. */ +#if ISA_SHOULD_BUILD (4) + +# ifndef WCSCPY +# define WCSCPY __wcscpy_ssse3 +# endif + # include <sysdep.h> .section .text.ssse3,"ax",@progbits -ENTRY (__wcscpy_ssse3) +ENTRY (WCSCPY) mov %rsi, %rcx mov %rdi, %rdx @@ -547,5 +557,5 @@ L(Exit16): mov %rdi, %rax ret -END(__wcscpy_ssse3) +END(WCSCPY) #endif diff --git a/sysdeps/x86_64/multiarch/wcscpy.c b/sysdeps/x86_64/multiarch/wcscpy.c index 53c3228dc2..92c917b6b4 100644 --- a/sysdeps/x86_64/multiarch/wcscpy.c +++ b/sysdeps/x86_64/multiarch/wcscpy.c @@ -26,15 +26,16 @@ # define SYMBOL_NAME wcscpy # include <init-arch.h> -extern __typeof (REDIRECT_NAME) OPTIMIZE (generic) attribute_hidden; extern __typeof (REDIRECT_NAME) OPTIMIZE (ssse3) attribute_hidden; +extern __typeof (REDIRECT_NAME) OPTIMIZE (generic) attribute_hidden; + static inline void * IFUNC_SELECTOR (void) { const struct cpu_features* cpu_features = __get_cpu_features (); - if (CPU_FEATURE_USABLE_P (cpu_features, SSSE3)) + if (X86_ISA_CPU_FEATURE_USABLE_P (cpu_features, SSSE3)) return OPTIMIZE (ssse3); return OPTIMIZE (generic); |