Index: lib/libc/powerpc64/string/bcopy.S
===================================================================
--- lib/libc/powerpc64/string/bcopy.S
+++ lib/libc/powerpc64/string/bcopy.S
@@ -34,6 +34,11 @@
 #define BLOCK_SIZE			(1 << BLOCK_SIZE_BITS)
 #define BLOCK_SIZE_MASK			(BLOCK_SIZE - 1)
 
+/* Minimum 8 byte alignment, to avoid cache-inhibited alignment faults.*/
+#ifndef ALIGN_MASK
+#define ALIGN_MASK			0x7
+#endif
+
 #define MULTI_PHASE_THRESHOLD		512
 
 #ifndef FN_NAME
@@ -67,8 +72,14 @@
 #endif
 
 	cmpldi	%r5, MULTI_PHASE_THRESHOLD
-	bge	.Lmulti_phase
-
+	blt	.Lavoid_vsx
+	/*check if aligned. if aligned, go to multi_phase*/
+	andi.	%r8, %r3, ALIGN_MASK
+	andi.	%r7, %r4, ALIGN_MASK
+	xor.	%r6, %r7, %r8 /*compare and store*/
+	beq	.Lmulti_phase
+
+.Lavoid_vsx:
 	/* align src */
 	cmpd	%r4, %r3		/* forward or backward copy? */
 	blt	.Lbackward_align
@@ -106,6 +117,10 @@
 	li	%r0, 1
 	li	%r8, 16
 	li	%r9, 0
+	/*compare again if src and dst were relatively aligned*/
+	cmpdi	%r6, 0
+	bne .Lunaligned
+	/*if they were, continue as normal*/
 	b	.Lsingle_phase
 
 .Lbackward_single_copy:
@@ -116,6 +131,10 @@
 	/* point src and dst to last byte */
 	addi	%r3, %r3, -1
 	addi	%r4, %r4, -1
+	/*compare again if src and dst were relatively aligned*/
+	cmpdi	%r6, 0
+	bne .Lunaligned
+	/*if they were, continue as normal*/
 
 .Lsingle_phase:
 	srdi.	%r6, %r5, 4		/* number of 16-bytes */
@@ -145,6 +164,11 @@
 	beq	.Ldone			/* 1-bytes == 0? done */
 
 	mtctr	%r6
+	b .Lsingle_1_loop
+
+.Lunaligned:
+	/*copy ever byte with single_1_loop*/
+	mtctr %r5
 	.align 5
 .Lsingle_1_loop:
 	lbz	%r6, 0(%r4)