sysdeps/x86_64/multiarch/memcpy-ssse3.S  |   48 +++++++++++++++---------------
 sysdeps/x86_64/multiarch/memmove-ssse3.S |    8 ++--
 2 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/sysdeps/x86_64/multiarch/memcpy-ssse3.S b/sysdeps/x86_64/multiarch/memcpy-ssse3.S
index 9a878d3..747d2ef 100644
--- a/sysdeps/x86_64/multiarch/memcpy-ssse3.S
+++ b/sysdeps/x86_64/multiarch/memcpy-ssse3.S
@@ -61,18 +61,15 @@ ENTRY (MEMCPY)
 #ifdef USE_AS_MEMPCPY
 	add	%rdx, %rax
 #endif
-
-#ifdef USE_AS_MEMMOVE
-	cmp	%rsi, %rdi
-	jb	L(copy_forward)
-	je	L(write_0bytes)
-	cmp	$79, %rdx
-	jbe	L(copy_forward)
-	jmp	L(copy_backward)
-L(copy_forward):
-#endif
-	cmp	$79, %rdx
-	lea     L(table_less_80bytes)(%rip), %r11
+	/*
+	 * The small cases we can do without checking for any
+	 * overlap at all, since we do them as all loads followed
+	 * by all stores.
+	 *
+	 * So just jump through the less-than-80bytes table.
+	 */
+	cmp	$79,%rdx
+	lea	L(table_less_80bytes)(%rip), %r11
 	ja	L(80bytesormore)
 	movslq	(%r11, %rdx, 4), %r9
 	add	%rdx, %rsi
@@ -81,13 +78,24 @@ L(copy_forward):
 	jmp	*%r9
 	ud2
 
-	ALIGN (4)
+	/*
+	 * For the 80+ byte cases we need to check overlap
+	 */
 L(80bytesormore):
-#ifndef USE_AS_MEMMOVE
+	lea	(%rsi,%rdx),%r9
+	lea	(%rdi,%rdx),%r11
+	cmp	%rdi,%r9		/* dest start >= source end */
+	jae	L(nonoverlap)		/*  -> nonoverlapping */
+	cmp	%rsi,%r11		/* source start >= destination end */
+	jae	L(nonoverlap)		/*  -> nonoverlapping */
+	cmp	%rsi, %rdi		/* overlap: */
+	jb	L(copy_forward)		/* source < dest: forward copy */
+	je	L(write_0bytes)		/* source == dest: no copy */
+	jmp	L(copy_backward)	/* source > dest: backward copy */
+L(nonoverlap):
 	cmp	%dil, %sil
 	jle	L(copy_backward)
-#endif
-
+L(copy_forward):
 	movdqu	(%rsi), %xmm0
 	mov	%rdi, %rcx
 	and	$-16, %rdi
@@ -2805,7 +2813,6 @@ L(large_page_fwd):
 	movntdq	%xmm1, (%rdi)
 	lea	16(%rdi), %rdi
 	lea	-0x90(%rdx), %rdx
-#ifdef USE_AS_MEMMOVE
 	mov	%rsi, %r9
 	sub	%rdi, %r9
 	cmp	%rdx, %r9
@@ -2814,7 +2821,6 @@ L(large_page_fwd):
 	cmp	%rcx, %rdx
 	jb	L(ll_cache_copy_fwd_start)
 L(memmove_is_memcpy_fwd):
-#endif
 L(large_page_loop):
 	movdqu	(%rsi), %xmm0
 	movdqu	0x10(%rsi), %xmm1
@@ -2859,7 +2865,6 @@ L(large_page_less_64bytes):
 	sfence
 	BRANCH_TO_JMPTBL_ENTRY (L(table_less_80bytes), %rdx, 4)
 
-#ifdef USE_AS_MEMMOVE
 	ALIGN (4)
 L(ll_cache_copy_fwd_start):
 	prefetcht0 0x1c0(%rsi)
@@ -2906,7 +2911,6 @@ L(large_page_ll_less_fwd_64bytes):
 	add	%rdx, %rdi
 	BRANCH_TO_JMPTBL_ENTRY (L(table_less_80bytes), %rdx, 4)
 
-#endif
 	ALIGN (4)
 L(large_page_bwd):
 	movdqu	-0x10(%rsi), %xmm1
@@ -2915,7 +2919,6 @@ L(large_page_bwd):
 	movdqa	%xmm1, -0x10(%rdi)
 	lea	-16(%rdi), %rdi
 	lea	-0x90(%rdx), %rdx
-#ifdef USE_AS_MEMMOVE
 	mov	%rdi, %r9
 	sub	%rsi, %r9
 	cmp	%rdx, %r9
@@ -2923,7 +2926,6 @@ L(large_page_bwd):
 	cmp	%rcx, %r9
 	jb	L(ll_cache_copy_bwd_start)
 L(memmove_is_memcpy_bwd):
-#endif
 L(large_page_bwd_loop):
 	movdqu	-0x10(%rsi), %xmm0
 	movdqu	-0x20(%rsi), %xmm1
@@ -2966,7 +2968,6 @@ L(large_page_less_bwd_64bytes):
 	sfence
 	BRANCH_TO_JMPTBL_ENTRY (L(table_less_80bytes), %rdx, 4)
 
-#ifdef USE_AS_MEMMOVE
 	ALIGN (4)
 L(ll_cache_copy_bwd_start):
 	prefetcht0 -0x1c0(%rsi)
@@ -3010,7 +3011,6 @@ L(ll_cache_copy_bwd_start):
 	sub	$0x40, %rdx
 L(large_page_ll_less_bwd_64bytes):
 	BRANCH_TO_JMPTBL_ENTRY (L(table_less_80bytes), %rdx, 4)
-#endif
 
 END (MEMCPY)
 
diff --git a/sysdeps/x86_64/multiarch/memmove-ssse3.S b/sysdeps/x86_64/multiarch/memmove-ssse3.S
index 295430b..a38d445 100644
--- a/sysdeps/x86_64/multiarch/memmove-ssse3.S
+++ b/sysdeps/x86_64/multiarch/memmove-ssse3.S
@@ -1,4 +1,4 @@
-#define USE_AS_MEMMOVE
-#define MEMCPY		__memmove_ssse3
-#define MEMCPY_CHK	__memmove_chk_ssse3
-#include "memcpy-ssse3.S"
+#include <sysdep.h>
+
+strong_alias(__memmove_ssse3, __memcpy_ssse3);
+strong_alias(__memmove_chk_ssse3, __memcpy_chk_ssse3);