Enable AVX2 optimized memset only if -mavx2 works

	* config.h.in (HAVE_AVX2_SUPPORT): New #undef.
	* sysdeps/i386/configure.ac: Set HAVE_AVX2_SUPPORT and
	config-cflags-avx2.
	* sysdeps/x86_64/configure.ac: Likewise.
	* sysdeps/i386/configure: Regenerated.
	* sysdeps/x86_64/configure: Likewise.
	* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
	memset-avx2 only if config-cflags-avx2 is yes.
	* sysdeps/x86_64/multiarch/ifunc-impl-list.c (__libc_ifunc_impl_list):
	Tests for memset_chk and memset only if HAVE_AVX2_SUPPORT is
	defined.
	* sysdeps/x86_64/multiarch/memset.S: Define multiple versions
	only if HAVE_AVX2_SUPPORT is defined.
	* sysdeps/x86_64/multiarch/memset_chk.S: Likewise.

4 files changed, 21 insertions, 14 deletions

diff --git a/sysdeps/x86_64/multiarch/Makefile b/sysdeps/x86_64/multiarch/Makefile
index 42df96f636..3bb9702b95 100644
--- a/sysdeps/x86_64/multiarch/Makefile
+++ b/sysdeps/x86_64/multiarch/Makefile
@@ -17,8 +17,7 @@ sysdep_routines += strncat-c stpncpy-c strncpy-c strcmp-ssse3 \
 		   strcpy-sse2-unaligned strncpy-sse2-unaligned \
 		   stpcpy-sse2-unaligned stpncpy-sse2-unaligned \
 		   strcat-sse2-unaligned strncat-sse2-unaligned \
-		   strchr-sse2-no-bsf memcmp-ssse3 strstr-sse2-unaligned \
-		   memset-avx2
+		   strchr-sse2-no-bsf memcmp-ssse3 strstr-sse2-unaligned
 
 ifeq (yes,$(config-cflags-sse4))
 sysdep_routines += strcspn-c strpbrk-c strspn-c varshift
@@ -27,6 +26,10 @@ CFLAGS-strcspn-c.c += -msse4
 CFLAGS-strpbrk-c.c += -msse4
 CFLAGS-strspn-c.c += -msse4
 endif
+
+ifeq (yes,$(config-cflags-avx2))
+sysdep_routines += memset-avx2
+endif
 endif
 
 ifeq ($(subdir),wcsmbs)
diff --git a/sysdeps/x86_64/multiarch/ifunc-impl-list.c b/sysdeps/x86_64/multiarch/ifunc-impl-list.c
index f1593c5ea1..7e93e598db 100644
--- a/sysdeps/x86_64/multiarch/ifunc-impl-list.c
+++ b/sysdeps/x86_64/multiarch/ifunc-impl-list.c
@@ -61,6 +61,7 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array,
 			      __memmove_ssse3)
 	      IFUNC_IMPL_ADD (array, i, memmove, 1, __memmove_sse2))
 
+#ifdef HAVE_AVX2_SUPPORT
   /* Support sysdeps/x86_64/multiarch/memset_chk.S.  */
   IFUNC_IMPL (i, name, __memset_chk,
 	      IFUNC_IMPL_ADD (array, i, __memset_chk, 1, __memset_chk_sse2)
@@ -71,6 +72,7 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array,
   IFUNC_IMPL (i, name, memset,
 	      IFUNC_IMPL_ADD (array, i, memset, 1, __memset_sse2)
 	      IFUNC_IMPL_ADD (array, i, memset, HAS_AVX2, __memset_avx2))
+#endif
 
   /* Support sysdeps/x86_64/multiarch/stpncpy.S.  */
   IFUNC_IMPL (i, name, stpncpy,
diff --git a/sysdeps/x86_64/multiarch/memset.S b/sysdeps/x86_64/multiarch/memset.S
index 3113d1cbc0..00d46d12d0 100644
--- a/sysdeps/x86_64/multiarch/memset.S
+++ b/sysdeps/x86_64/multiarch/memset.S
@@ -17,12 +17,13 @@
    License along with the GNU C Library; if not, see
    <http://www.gnu.org/licenses/>.  */
 
+#ifdef HAVE_AVX2_SUPPORT
 #include <sysdep.h>
 #include <shlib-compat.h>
 #include <init-arch.h>
 
 /* Define multiple versions only for the definition in lib.  */
-#ifndef NOT_IN_libc
+# ifndef NOT_IN_libc
 ENTRY(memset)
 	.type	memset, @gnu_indirect_function
 	cmpl	$0, __cpu_features+KIND_OFFSET(%rip)
@@ -34,26 +35,27 @@ ENTRY(memset)
 	leaq	__memset_avx2(%rip), %rax
 2:	ret
 END(memset)
-#endif
+# endif
 
-#if !defined NOT_IN_libc
-# undef memset
-# define memset __memset_sse2
+# if !defined NOT_IN_libc
+#  undef memset
+#  define memset __memset_sse2
 
-# undef __memset_chk
-# define __memset_chk __memset_chk_sse2
+#  undef __memset_chk
+#  define __memset_chk __memset_chk_sse2
 
-# ifdef SHARED
+#  ifdef SHARED
 #  undef libc_hidden_builtin_def
 /* It doesn't make sense to send libc-internal memset calls through a PLT.
    The speedup we get from using GPR instruction is likely eaten away
    by the indirect call in the PLT.  */
 #  define libc_hidden_builtin_def(name) \
 	.globl __GI_memset; __GI_memset = __memset_sse2
-# endif
+#  endif
 
-# undef strong_alias
-# define strong_alias(original, alias)
+#  undef strong_alias
+#  define strong_alias(original, alias)
+# endif
 #endif
 
 #include "../memset.S"
diff --git a/sysdeps/x86_64/multiarch/memset_chk.S b/sysdeps/x86_64/multiarch/memset_chk.S
index 2182780822..8a607bd6b7 100644
--- a/sysdeps/x86_64/multiarch/memset_chk.S
+++ b/sysdeps/x86_64/multiarch/memset_chk.S
@@ -22,7 +22,7 @@
 
 /* Define multiple versions only for the definition in lib.  */
 #ifndef NOT_IN_libc
-# ifdef SHARED
+# if defined SHARED && defined HAVE_AVX2_SUPPORT
 ENTRY(__memset_chk)
 	.type	__memset_chk, @gnu_indirect_function
 	cmpl	$0, __cpu_features+KIND_OFFSET(%rip)