Add generic arm non-neon memmove.

From OpenBSD.

Bug: http://b/63992911
Test: ran tests
Change-Id: If7d9166922776cdc9333ff04205f9c6312a812b3
diff --git a/libc/Android.bp b/libc/Android.bp
index b68eb7c..eac4414 100644
--- a/libc/Android.bp
+++ b/libc/Android.bp
@@ -546,7 +546,6 @@
 
         // May be overriden by per-arch optimized versions
         "upstream-openbsd/lib/libc/string/memchr.c",
-        "upstream-openbsd/lib/libc/string/memmove.c",
         "upstream-openbsd/lib/libc/string/memrchr.c",
         "upstream-openbsd/lib/libc/string/stpcpy.c",
         "upstream-openbsd/lib/libc/string/stpncpy.c",
@@ -566,7 +565,6 @@
             ],
             neon: {
                 exclude_srcs: [
-                    "upstream-openbsd/lib/libc/string/memmove.c",
                     "upstream-openbsd/lib/libc/string/stpcpy.c",
                     "upstream-openbsd/lib/libc/string/strcat.c",
                 ],
@@ -575,7 +573,6 @@
         arm64: {
             exclude_srcs: [
                 "upstream-openbsd/lib/libc/string/memchr.c",
-                "upstream-openbsd/lib/libc/string/memmove.c",
                 "upstream-openbsd/lib/libc/string/stpcpy.c",
                 "upstream-openbsd/lib/libc/string/strcpy.c",
                 "upstream-openbsd/lib/libc/string/strncmp.c",
@@ -584,7 +581,6 @@
         mips: {
             exclude_srcs: [
                 "upstream-openbsd/lib/libc/string/memchr.c",
-                "upstream-openbsd/lib/libc/string/memmove.c",
                 "upstream-openbsd/lib/libc/string/strcpy.c",
                 "upstream-openbsd/lib/libc/string/strncmp.c",
             ],
@@ -592,7 +588,6 @@
         mips64: {
             exclude_srcs: [
                 "upstream-openbsd/lib/libc/string/memchr.c",
-                "upstream-openbsd/lib/libc/string/memmove.c",
                 "upstream-openbsd/lib/libc/string/strcpy.c",
                 "upstream-openbsd/lib/libc/string/strncmp.c",
             ],
@@ -600,7 +595,6 @@
         x86: {
             exclude_srcs: [
                 "upstream-openbsd/lib/libc/string/memchr.c",
-                "upstream-openbsd/lib/libc/string/memmove.c",
                 "upstream-openbsd/lib/libc/string/memrchr.c",
                 "upstream-openbsd/lib/libc/string/stpcpy.c",
                 "upstream-openbsd/lib/libc/string/stpncpy.c",
@@ -620,7 +614,6 @@
 
         x86_64: {
             exclude_srcs: [
-                "upstream-openbsd/lib/libc/string/memmove.c",
                 "upstream-openbsd/lib/libc/string/stpcpy.c",
                 "upstream-openbsd/lib/libc/string/stpncpy.c",
                 "upstream-openbsd/lib/libc/string/strcat.c",
@@ -860,6 +853,7 @@
             srcs: [
                 "arch-arm/generic/bionic/memcmp.S",
                 "arch-arm/generic/bionic/memcpy.S",
+                "arch-arm/generic/bionic/memmove.S",
                 "arch-arm/generic/bionic/memset.S",
                 "arch-arm/generic/bionic/strcmp.S",
                 "arch-arm/generic/bionic/strcpy.S",
@@ -877,8 +871,8 @@
             ],
             cortex_a7: {
                 srcs: [
-                    "arch-arm/cortex-a7/bionic/memset.S",
                     "arch-arm/cortex-a7/bionic/memcpy.S",
+                    "arch-arm/cortex-a7/bionic/memset.S",
                 ],
                 exclude_srcs: [
                     "arch-arm/cortex-a15/bionic/memcpy.S",
@@ -897,8 +891,8 @@
                     "arch-arm/cortex-a9/bionic/strlen.S",
                 ],
                 exclude_srcs: [
-                    "arch-arm/cortex-a15/bionic/memset.S",
                     "arch-arm/cortex-a15/bionic/memcpy.S",
+                    "arch-arm/cortex-a15/bionic/memset.S",
 
                     "arch-arm/cortex-a15/bionic/stpcpy.S",
                     "arch-arm/cortex-a15/bionic/strcat.S",
@@ -1010,6 +1004,7 @@
                 ],
                 exclude_srcs: [
                     "arch-arm/generic/bionic/memcpy.S",
+                    "arch-arm/generic/bionic/memmove.S",
                     "arch-arm/generic/bionic/memset.S",
                     "arch-arm/generic/bionic/strcmp.S",
                     "arch-arm/generic/bionic/strcpy.S",
diff --git a/libc/NOTICE b/libc/NOTICE
index 35fc858..ae7da18 100644
--- a/libc/NOTICE
+++ b/libc/NOTICE
@@ -3101,6 +3101,35 @@
 
 -------------------------------------------------------------------
 
+Copyright (c) 1997 The NetBSD Foundation, Inc.
+All rights reserved.
+
+This code is derived from software contributed to The NetBSD Foundation
+by Neil A. Carson and Mark Brinicombe
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+1. Redistributions of source code must retain the above copyright
+   notice, this list of conditions and the following disclaimer.
+2. Redistributions in binary form must reproduce the above copyright
+   notice, this list of conditions and the following disclaimer in the
+   documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
+TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
+BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+POSSIBILITY OF SUCH DAMAGE.
+
+-------------------------------------------------------------------
+
 Copyright (c) 1997 Todd C. Miller <Todd.Miller@courtesan.com>
 
 Permission to use, copy, modify, and distribute this software for any
diff --git a/libc/arch-arm/generic/bionic/memmove.S b/libc/arch-arm/generic/bionic/memmove.S
new file mode 100644
index 0000000..d5ee99d
--- /dev/null
+++ b/libc/arch-arm/generic/bionic/memmove.S
@@ -0,0 +1,471 @@
+/*	$OpenBSD: _memcpy.S,v 1.6 2016/08/06 19:16:09 guenther Exp $	*/
+/*	$NetBSD: _memcpy.S,v 1.4 2003/04/05 23:08:52 bjh21 Exp $	*/
+
+/*-
+ * Copyright (c) 1997 The NetBSD Foundation, Inc.
+ * All rights reserved.
+ *
+ * This code is derived from software contributed to The NetBSD Foundation
+ * by Neil A. Carson and Mark Brinicombe
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
+ * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
+ * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
+ * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <private/bionic_asm.h>
+
+  .syntax unified
+
+/*
+ * This is one fun bit of code ...
+ * Some easy listening music is suggested while trying to understand this
+ * code e.g. Iron Maiden
+ *
+ * For anyone attempting to understand it :
+ *
+ * The core code is implemented here with simple stubs for memcpy()
+ * memmove() and bcopy().
+ *
+ * All local labels are prefixed with Lmemcpy_
+ * Following the prefix a label starting f is used in the forward copy code
+ * while a label using b is used in the backwards copy code
+ * The source and destination addresses determine whether a forward or
+ * backward copy is performed.
+ * Separate bits of code are used to deal with the following situations
+ * for both the forward and backwards copy.
+ * unaligned source address
+ * unaligned destination address
+ * Separate copy routines are used to produce an optimised result for each
+ * of these cases.
+ * The copy code will use LDM/STM instructions to copy up to 32 bytes at
+ * a time where possible.
+ *
+ * Note: r12 (aka ip) can be trashed during the function along with
+ * r0-r3 although r0-r2 have defined uses i.e. src, dest, len through out.
+ * Additional registers are preserved prior to use i.e. r4, r5 & lr
+ *
+ * Apologies for the state of the comments ;-)
+ */
+
+ENTRY_PRIVATE(bsd_safe_memcpy)
+	/* Determine copy direction */
+	cmp	r1, r0
+	bcc	.Lmemcpy_backwards
+
+	moveq	r0, #0			/* Quick abort for len=0 */
+	moveq	pc, lr
+
+	stmdb	sp!, {r0, lr}		/* memcpy() returns dest addr */
+	subs	r2, r2, #4
+	blt	.Lmemcpy_fl4		/* less than 4 bytes */
+	ands	r12, r0, #3
+	bne	.Lmemcpy_fdestul	/* oh unaligned destination addr */
+	ands	r12, r1, #3
+	bne	.Lmemcpy_fsrcul		/* oh unaligned source addr */
+
+.Lmemcpy_ft8:
+	/* We have aligned source and destination */
+	subs	r2, r2, #8
+	blt	.Lmemcpy_fl12		/* less than 12 bytes (4 from above) */
+	subs	r2, r2, #0x14         
+	blt	.Lmemcpy_fl32		/* less than 32 bytes (12 from above) */
+	stmdb	sp!, {r4}		/* borrow r4 */
+
+	/* blat 32 bytes at a time */
+	/* XXX for really big copies perhaps we should use more registers */
+.Lmemcpy_floop32:	
+	ldmia	r1!, {r3, r4, r12, lr}
+	stmia	r0!, {r3, r4, r12, lr}
+	ldmia	r1!, {r3, r4, r12, lr}
+	stmia	r0!, {r3, r4, r12, lr}
+	subs	r2, r2, #0x20         
+	bge	.Lmemcpy_floop32
+
+	cmn	r2, #0x10
+	ldmiage	r1!, {r3, r4, r12, lr}	/* blat a remaining 16 bytes */
+	stmiage	r0!, {r3, r4, r12, lr}
+	subge	r2, r2, #0x10         
+	ldmia	sp!, {r4}		/* return r4 */
+
+.Lmemcpy_fl32:
+	adds	r2, r2, #0x14         
+
+	/* blat 12 bytes at a time */
+.Lmemcpy_floop12:
+	ldmiage	r1!, {r3, r12, lr}
+	stmiage	r0!, {r3, r12, lr}
+	subsge	r2, r2, #0x0c         
+	bge	.Lmemcpy_floop12
+
+.Lmemcpy_fl12:
+	adds	r2, r2, #8
+	blt	.Lmemcpy_fl4
+
+	subs	r2, r2, #4
+	ldrlt	r3, [r1], #4
+	strlt	r3, [r0], #4
+	ldmiage	r1!, {r3, r12}
+	stmiage	r0!, {r3, r12}
+	subge	r2, r2, #4
+
+.Lmemcpy_fl4:
+	/* less than 4 bytes to go */
+	adds	r2, r2, #4
+	ldmiaeq	sp!, {r0, pc}		/* done */
+
+	/* copy the crud byte at a time */
+	cmp	r2, #2
+	ldrb	r3, [r1], #1
+	strb	r3, [r0], #1
+	ldrbge	r3, [r1], #1
+	strbge	r3, [r0], #1
+	ldrbgt	r3, [r1], #1
+	strbgt	r3, [r0], #1
+	ldmia	sp!, {r0, pc}
+
+	/* erg - unaligned destination */
+.Lmemcpy_fdestul:
+	rsb	r12, r12, #4
+	cmp	r12, #2
+
+	/* align destination with byte copies */
+	ldrb	r3, [r1], #1
+	strb	r3, [r0], #1
+	ldrbge	r3, [r1], #1
+	strbge	r3, [r0], #1
+	ldrbgt	r3, [r1], #1
+	strbgt	r3, [r0], #1
+	subs	r2, r2, r12
+	blt	.Lmemcpy_fl4		/* less the 4 bytes */
+
+	ands	r12, r1, #3
+	beq	.Lmemcpy_ft8		/* we have an aligned source */
+
+	/* erg - unaligned source */
+	/* This is where it gets nasty ... */
+.Lmemcpy_fsrcul:
+	bic	r1, r1, #3
+	ldr	lr, [r1], #4
+	cmp	r12, #2
+	bgt	.Lmemcpy_fsrcul3
+	beq	.Lmemcpy_fsrcul2
+	cmp	r2, #0x0c            
+	blt	.Lmemcpy_fsrcul1loop4
+	sub	r2, r2, #0x0c         
+	stmdb	sp!, {r4, r5}
+
+.Lmemcpy_fsrcul1loop16:
+	mov	r3, lr, lsr #8
+	ldmia	r1!, {r4, r5, r12, lr}
+	orr	r3, r3, r4, lsl #24
+	mov	r4, r4, lsr #8
+	orr	r4, r4, r5, lsl #24
+	mov	r5, r5, lsr #8
+	orr	r5, r5, r12, lsl #24
+	mov	r12, r12, lsr #8
+	orr	r12, r12, lr, lsl #24
+	stmia	r0!, {r3-r5, r12}
+	subs	r2, r2, #0x10         
+	bge	.Lmemcpy_fsrcul1loop16
+	ldmia	sp!, {r4, r5}
+	adds	r2, r2, #0x0c         
+	blt	.Lmemcpy_fsrcul1l4
+
+.Lmemcpy_fsrcul1loop4:
+	mov	r12, lr, lsr #8
+	ldr	lr, [r1], #4
+	orr	r12, r12, lr, lsl #24
+	str	r12, [r0], #4
+	subs	r2, r2, #4
+	bge	.Lmemcpy_fsrcul1loop4
+
+.Lmemcpy_fsrcul1l4:
+	sub	r1, r1, #3
+	b	.Lmemcpy_fl4
+
+.Lmemcpy_fsrcul2:
+	cmp	r2, #0x0c            
+	blt	.Lmemcpy_fsrcul2loop4
+	sub	r2, r2, #0x0c         
+	stmdb	sp!, {r4, r5}
+
+.Lmemcpy_fsrcul2loop16:
+	mov	r3, lr, lsr #16
+	ldmia	r1!, {r4, r5, r12, lr}
+	orr	r3, r3, r4, lsl #16
+	mov	r4, r4, lsr #16
+	orr	r4, r4, r5, lsl #16
+	mov	r5, r5, lsr #16
+	orr	r5, r5, r12, lsl #16
+	mov	r12, r12, lsr #16
+	orr	r12, r12, lr, lsl #16
+	stmia	r0!, {r3-r5, r12}
+	subs	r2, r2, #0x10         
+	bge	.Lmemcpy_fsrcul2loop16
+	ldmia	sp!, {r4, r5}
+	adds	r2, r2, #0x0c         
+	blt	.Lmemcpy_fsrcul2l4
+
+.Lmemcpy_fsrcul2loop4:
+	mov	r12, lr, lsr #16
+	ldr	lr, [r1], #4
+	orr	r12, r12, lr, lsl #16
+	str	r12, [r0], #4
+	subs	r2, r2, #4
+	bge	.Lmemcpy_fsrcul2loop4
+
+.Lmemcpy_fsrcul2l4:
+	sub	r1, r1, #2
+	b	.Lmemcpy_fl4
+
+.Lmemcpy_fsrcul3:
+	cmp	r2, #0x0c            
+	blt	.Lmemcpy_fsrcul3loop4
+	sub	r2, r2, #0x0c         
+	stmdb	sp!, {r4, r5}
+
+.Lmemcpy_fsrcul3loop16:
+	mov	r3, lr, lsr #24
+	ldmia	r1!, {r4, r5, r12, lr}
+	orr	r3, r3, r4, lsl #8
+	mov	r4, r4, lsr #24
+	orr	r4, r4, r5, lsl #8
+	mov	r5, r5, lsr #24
+	orr	r5, r5, r12, lsl #8
+	mov	r12, r12, lsr #24
+	orr	r12, r12, lr, lsl #8
+	stmia	r0!, {r3-r5, r12}
+	subs	r2, r2, #0x10         
+	bge	.Lmemcpy_fsrcul3loop16
+	ldmia	sp!, {r4, r5}
+	adds	r2, r2, #0x0c         
+	blt	.Lmemcpy_fsrcul3l4
+
+.Lmemcpy_fsrcul3loop4:
+	mov	r12, lr, lsr #24
+	ldr	lr, [r1], #4
+	orr	r12, r12, lr, lsl #8
+	str	r12, [r0], #4
+	subs	r2, r2, #4
+	bge	.Lmemcpy_fsrcul3loop4
+
+.Lmemcpy_fsrcul3l4:
+	sub	r1, r1, #1
+	b	.Lmemcpy_fl4
+
+.Lmemcpy_backwards:
+	add	r1, r1, r2
+	add	r0, r0, r2
+	subs	r2, r2, #4
+	blt	.Lmemcpy_bl4		/* less than 4 bytes */
+	ands	r12, r0, #3
+	bne	.Lmemcpy_bdestul	/* oh unaligned destination addr */
+	ands	r12, r1, #3
+	bne	.Lmemcpy_bsrcul		/* oh unaligned source addr */
+
+.Lmemcpy_bt8:
+	/* We have aligned source and destination */
+	subs	r2, r2, #8
+	blt	.Lmemcpy_bl12		/* less than 12 bytes (4 from above) */
+	stmdb	sp!, {r4, lr}
+	subs	r2, r2, #0x14		/* less than 32 bytes (12 from above) */
+	blt	.Lmemcpy_bl32
+
+	/* blat 32 bytes at a time */
+	/* XXX for really big copies perhaps we should use more registers */
+.Lmemcpy_bloop32:
+	ldmdb	r1!, {r3, r4, r12, lr}
+	stmdb	r0!, {r3, r4, r12, lr}
+	ldmdb	r1!, {r3, r4, r12, lr}
+	stmdb	r0!, {r3, r4, r12, lr}
+	subs	r2, r2, #0x20         
+	bge	.Lmemcpy_bloop32
+
+.Lmemcpy_bl32:
+	cmn	r2, #0x10            
+	ldmdbge	r1!, {r3, r4, r12, lr}	/* blat a remaining 16 bytes */
+	stmdbge	r0!, {r3, r4, r12, lr}
+	subge	r2, r2, #0x10         
+	adds	r2, r2, #0x14         
+	ldmdbge	r1!, {r3, r12, lr}	/* blat a remaining 12 bytes */
+	stmdbge	r0!, {r3, r12, lr}
+	subge	r2, r2, #0x0c         
+	ldmia	sp!, {r4, lr}
+
+.Lmemcpy_bl12:
+	adds	r2, r2, #8
+	blt	.Lmemcpy_bl4
+	subs	r2, r2, #4
+	ldrlt	r3, [r1, #-4]!
+	strlt	r3, [r0, #-4]!
+	ldmdbge	r1!, {r3, r12}
+	stmdbge	r0!, {r3, r12}
+	subge	r2, r2, #4
+
+.Lmemcpy_bl4:
+	/* less than 4 bytes to go */
+	adds	r2, r2, #4
+	moveq	pc, lr			/* done */
+
+	/* copy the crud byte at a time */
+	cmp	r2, #2
+	ldrb	r3, [r1, #-1]!
+	strb	r3, [r0, #-1]!
+	ldrbge	r3, [r1, #-1]!
+	strbge	r3, [r0, #-1]!
+	ldrbgt	r3, [r1, #-1]!
+	strbgt	r3, [r0, #-1]!
+	mov	pc, lr
+
+	/* erg - unaligned destination */
+.Lmemcpy_bdestul:
+	cmp	r12, #2
+
+	/* align destination with byte copies */
+	ldrb	r3, [r1, #-1]!
+	strb	r3, [r0, #-1]!
+	ldrbge	r3, [r1, #-1]!
+	strbge	r3, [r0, #-1]!
+	ldrbgt	r3, [r1, #-1]!
+	strbgt	r3, [r0, #-1]!
+	subs	r2, r2, r12
+	blt	.Lmemcpy_bl4		/* less than 4 bytes to go */
+	ands	r12, r1, #3
+	beq	.Lmemcpy_bt8		/* we have an aligned source */
+
+	/* erg - unaligned source */
+	/* This is where it gets nasty ... */
+.Lmemcpy_bsrcul:
+	bic	r1, r1, #3
+	ldr	r3, [r1, #0]
+	cmp	r12, #2
+	blt	.Lmemcpy_bsrcul1
+	beq	.Lmemcpy_bsrcul2
+	cmp	r2, #0x0c            
+	blt	.Lmemcpy_bsrcul3loop4
+	sub	r2, r2, #0x0c         
+	stmdb	sp!, {r4, r5, lr}
+
+.Lmemcpy_bsrcul3loop16:
+	mov	lr, r3, lsl #8
+	ldmdb	r1!, {r3-r5, r12}
+	orr	lr, lr, r12, lsr #24
+	mov	r12, r12, lsl #8
+	orr	r12, r12, r5, lsr #24
+	mov	r5, r5, lsl #8
+	orr	r5, r5, r4, lsr #24
+	mov	r4, r4, lsl #8
+	orr	r4, r4, r3, lsr #24
+	stmdb	r0!, {r4, r5, r12, lr}
+	subs	r2, r2, #0x10         
+	bge	.Lmemcpy_bsrcul3loop16
+	ldmia	sp!, {r4, r5, lr}
+	adds	r2, r2, #0x0c         
+	blt	.Lmemcpy_bsrcul3l4
+
+.Lmemcpy_bsrcul3loop4:
+	mov	r12, r3, lsl #8
+	ldr	r3, [r1, #-4]!
+	orr	r12, r12, r3, lsr #24
+	str	r12, [r0, #-4]!
+	subs	r2, r2, #4
+	bge	.Lmemcpy_bsrcul3loop4
+
+.Lmemcpy_bsrcul3l4:
+	add	r1, r1, #3
+	b	.Lmemcpy_bl4
+
+.Lmemcpy_bsrcul2:
+	cmp	r2, #0x0c            
+	blt	.Lmemcpy_bsrcul2loop4
+	sub	r2, r2, #0x0c         
+	stmdb	sp!, {r4, r5, lr}
+
+.Lmemcpy_bsrcul2loop16:
+	mov	lr, r3, lsl #16
+	ldmdb	r1!, {r3-r5, r12}
+	orr	lr, lr, r12, lsr #16
+	mov	r12, r12, lsl #16
+	orr	r12, r12, r5, lsr #16
+	mov	r5, r5, lsl #16
+	orr	r5, r5, r4, lsr #16
+	mov	r4, r4, lsl #16
+	orr	r4, r4, r3, lsr #16
+	stmdb	r0!, {r4, r5, r12, lr}
+	subs	r2, r2, #0x10         
+	bge	.Lmemcpy_bsrcul2loop16
+	ldmia	sp!, {r4, r5, lr}
+	adds	r2, r2, #0x0c         
+	blt	.Lmemcpy_bsrcul2l4
+
+.Lmemcpy_bsrcul2loop4:
+	mov	r12, r3, lsl #16
+	ldr	r3, [r1, #-4]!
+	orr	r12, r12, r3, lsr #16
+	str	r12, [r0, #-4]!
+	subs	r2, r2, #4
+	bge	.Lmemcpy_bsrcul2loop4
+
+.Lmemcpy_bsrcul2l4:
+	add	r1, r1, #2
+	b	.Lmemcpy_bl4
+
+.Lmemcpy_bsrcul1:
+	cmp	r2, #0x0c            
+	blt	.Lmemcpy_bsrcul1loop4
+	sub	r2, r2, #0x0c         
+	stmdb	sp!, {r4, r5, lr}
+
+.Lmemcpy_bsrcul1loop32:
+	mov	lr, r3, lsl #24
+	ldmdb	r1!, {r3-r5, r12}
+	orr	lr, lr, r12, lsr #8
+	mov	r12, r12, lsl #24
+	orr	r12, r12, r5, lsr #8
+	mov	r5, r5, lsl #24
+	orr	r5, r5, r4, lsr #8
+	mov	r4, r4, lsl #24
+	orr	r4, r4, r3, lsr #8
+	stmdb	r0!, {r4, r5, r12, lr}
+	subs	r2, r2, #0x10         
+	bge	.Lmemcpy_bsrcul1loop32
+	ldmia	sp!, {r4, r5, lr}
+	adds	r2, r2, #0x0c         
+	blt	.Lmemcpy_bsrcul1l4
+
+.Lmemcpy_bsrcul1loop4:
+	mov	r12, r3, lsl #24
+	ldr	r3, [r1, #-4]!
+	orr	r12, r12, r3, lsr #8
+	str	r12, [r0, #-4]!
+	subs	r2, r2, #4
+	bge	.Lmemcpy_bsrcul1loop4
+
+.Lmemcpy_bsrcul1l4:
+	add	r1, r1, #1
+	b	.Lmemcpy_bl4
+END(bsd_safe_memcpy)
+
+ENTRY(memmove)
+        stmfd   sp!, {r0, lr}
+        bl      bsd_safe_memcpy
+        ldmfd   sp!, {r0, pc}
+END(memmove)
diff --git a/libc/upstream-openbsd/lib/libc/string/memmove.c b/libc/upstream-openbsd/lib/libc/string/memmove.c
deleted file mode 100644
index 2f1deb2..0000000
--- a/libc/upstream-openbsd/lib/libc/string/memmove.c
+++ /dev/null
@@ -1,113 +0,0 @@
-/*	$OpenBSD: memmove.c,v 1.2 2015/08/31 02:53:57 guenther Exp $ */
-/*-
- * Copyright (c) 1990 The Regents of the University of California.
- * All rights reserved.
- *
- * This code is derived from software contributed to Berkeley by
- * Chris Torek.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * 1. Redistributions of source code must retain the above copyright
- *    notice, this list of conditions and the following disclaimer.
- * 2. Redistributions in binary form must reproduce the above copyright
- *    notice, this list of conditions and the following disclaimer in the
- *    documentation and/or other materials provided with the distribution.
- * 3. Neither the name of the University nor the names of its contributors
- *    may be used to endorse or promote products derived from this software
- *    without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
- * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
- * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
- * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
- * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
- * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
- * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
- * SUCH DAMAGE.
- */
-
-#include <string.h>
-
-/*
- * sizeof(word) MUST BE A POWER OF TWO
- * SO THAT wmask BELOW IS ALL ONES
- */
-typedef	long word;		/* "word" used for optimal copy speed */
-
-#define	wsize	sizeof(word)
-#define	wmask	(wsize - 1)
-
-/*
- * Copy a block of memory, handling overlap.
- */
-void *
-memmove(void *dst0, const void *src0, size_t length)
-{
-	char *dst = dst0;
-	const char *src = src0;
-	size_t t;
-
-	if (length == 0 || dst == src)		/* nothing to do */
-		goto done;
-
-	/*
-	 * Macros: loop-t-times; and loop-t-times, t>0
-	 */
-#define	TLOOP(s) if (t) TLOOP1(s)
-#define	TLOOP1(s) do { s; } while (--t)
-
-	if ((unsigned long)dst < (unsigned long)src) {
-		/*
-		 * Copy forward.
-		 */
-		t = (long)src;	/* only need low bits */
-		if ((t | (long)dst) & wmask) {
-			/*
-			 * Try to align operands.  This cannot be done
-			 * unless the low bits match.
-			 */
-			if ((t ^ (long)dst) & wmask || length < wsize)
-				t = length;
-			else
-				t = wsize - (t & wmask);
-			length -= t;
-			TLOOP1(*dst++ = *src++);
-		}
-		/*
-		 * Copy whole words, then mop up any trailing bytes.
-		 */
-		t = length / wsize;
-		TLOOP(*(word *)dst = *(word *)src; src += wsize; dst += wsize);
-		t = length & wmask;
-		TLOOP(*dst++ = *src++);
-	} else {
-		/*
-		 * Copy backwards.  Otherwise essentially the same.
-		 * Alignment works as before, except that it takes
-		 * (t&wmask) bytes to align, not wsize-(t&wmask).
-		 */
-		src += length;
-		dst += length;
-		t = (long)src;
-		if ((t | (long)dst) & wmask) {
-			if ((t ^ (long)dst) & wmask || length <= wsize)
-				t = length;
-			else
-				t &= wmask;
-			length -= t;
-			TLOOP1(*--dst = *--src);
-		}
-		t = length / wsize;
-		TLOOP(src -= wsize; dst -= wsize; *(word *)dst = *(word *)src);
-		t = length & wmask;
-		TLOOP(*--dst = *--src);
-	}
-done:
-	return (dst0);
-}
-DEF_STRONG(memmove);