about summary refs log tree commit diff
path: root/sysdeps/sparc/sparc64/multiarch
diff options
context:
space:
mode:
authorDavid S. Miller <davem@davemloft.net>2010-02-25 23:52:07 -0800
committerDavid S. Miller <davem@davemloft.net>2010-02-25 23:58:51 -0800
commit3afd5a3b555694361181f854f658f669105d0ad6 (patch)
tree6b4df892f2d5be5ff3f777ed186153ff6765b173 /sysdeps/sparc/sparc64/multiarch
parente44f6f89da537816aff71d926a0b774a77545b48 (diff)
downloadglibc-3afd5a3b555694361181f854f658f669105d0ad6.tar.gz
glibc-3afd5a3b555694361181f854f658f669105d0ad6.tar.xz
glibc-3afd5a3b555694361181f854f658f669105d0ad6.zip
sparc: Add multiarch support for memset/bzero/memcpy. davem/sparc
2010-02-25  David S. Miller  <davem@davemloft.net>

	* sysdeps/sparc/elf/rtld-global-offsets.sym: New file.
	* sysdeps/sparc/Makefile (csu): Add rtld-global-offsets.sym to
	gen-as-const-headers.
	* sysdeps/sparc/sparc32/sparcv9/multiarch/Makefile: New file.
	* sysdeps/sparc/sparc32/sparcv9/multiarch/memcpy.S: New file.
	* sysdeps/sparc/sparc32/sparcv9/multiarch/memset.S: New file.
	* sysdeps/sparc/sparc32/sparcv9/sparcv9b/memcpy.S: Move to...
	* sysdeps/sparc/sparc32/sparcv9/multiarch/memcpy-ultra3.S: ...here.
	* sysdeps/sparc/sparc32/sparcv9/sparcv9v/memcpy.S: Move to...
	* sysdeps/sparc/sparc32/sparcv9/multiarch/memcpy-niagara1.S: ...here.
	* sysdeps/sparc/sparc32/sparcv9/sparcv9v/memset.S: Move to...
	* sysdeps/sparc/sparc32/sparcv9/multiarch/memset-niagara1.S: ...here.
	* sysdeps/sparc/sparc32/sparcv9/sparcv9v2/memcpy.S: Move to...
	* sysdeps/sparc/sparc32/sparcv9/multiarch/memcpy-niagara2.S: ...here.
	* sysdeps/sparc/sparc32/sparcv9/sparcv9v2/memset.S: Removed.
	* sysdeps/sparc/sparc64/multiarch/Makefile: New file.
	* sysdeps/sparc/sparc64/sparcv9v/memcpy.S: Move to...
	* sysdeps/sparc/sparc64/multiarch/memcpy-niagara1.S: ...here.
	* sysdeps/sparc/sparc64/sparcv9v2/memcpy.S: Move to...
	* sysdeps/sparc/sparc64/multiarch/memcpy-niagara2.S: ...here.
	* sysdeps/sparc/sparc64/sparcv9b/memcpy.S: Move to...
	* sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S: ...here.
	* sysdeps/sparc/sparc64/sparcv9v/memset.S: Move to...
	* sysdeps/sparc/sparc64/multiarch/memset-niagara1.S: ...here.
	* sysdeps/sparc/sparc64/sparcv9v2/memset.S: Removed.
	* sysdeps/sparc/sparc64/multiarch/memcpy.S: New file.
	* sysdeps/sparc/sparc64/multiarch/memset.S: New file.
Diffstat (limited to 'sysdeps/sparc/sparc64/multiarch')
-rw-r--r--sysdeps/sparc/sparc64/multiarch/Makefile4
-rw-r--r--sysdeps/sparc/sparc64/multiarch/memcpy-niagara1.S342
-rw-r--r--sysdeps/sparc/sparc64/multiarch/memcpy-niagara2.S492
-rw-r--r--sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S320
-rw-r--r--sysdeps/sparc/sparc64/multiarch/memcpy.S107
-rw-r--r--sysdeps/sparc/sparc64/multiarch/memset-niagara1.S132
-rw-r--r--sysdeps/sparc/sparc64/multiarch/memset.S145
7 files changed, 1542 insertions, 0 deletions
diff --git a/sysdeps/sparc/sparc64/multiarch/Makefile b/sysdeps/sparc/sparc64/multiarch/Makefile
new file mode 100644
index 0000000000..4d45042a95
--- /dev/null
+++ b/sysdeps/sparc/sparc64/multiarch/Makefile
@@ -0,0 +1,4 @@
+ifeq ($(subdir),string)
+sysdep_routines += memcpy-ultra3 memcpy-niagara1 memcpy-niagara2 \
+		   memset-niagara1
+endif
diff --git a/sysdeps/sparc/sparc64/multiarch/memcpy-niagara1.S b/sysdeps/sparc/sparc64/multiarch/memcpy-niagara1.S
new file mode 100644
index 0000000000..6a78295e81
--- /dev/null
+++ b/sysdeps/sparc/sparc64/multiarch/memcpy-niagara1.S
@@ -0,0 +1,342 @@
+/* Copy SIZE bytes from SRC to DEST.  For SUN4V Niagara.
+   Copyright (C) 2006, 2008 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by David S. Miller (davem@davemloft.net)
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+#define ASI_BLK_INIT_QUAD_LDD_P	0xe2
+#define ASI_P			0x80
+#define ASI_PNF			0x82
+
+#define LOAD(type,addr,dest)	type##a [addr] ASI_P, dest
+#define LOAD_TWIN(addr_reg,dest0,dest1)	\
+	ldda [addr_reg] ASI_BLK_INIT_QUAD_LDD_P, dest0
+
+#define STORE(type,src,addr)	type src, [addr]
+#define STORE_INIT(src,addr)	stxa src, [addr] %asi
+
+#ifndef XCC
+#define USE_BPR
+#define XCC xcc
+#endif
+
+#if !defined NOT_IN_libc
+
+	.register	%g2,#scratch
+	.register	%g3,#scratch
+	.register	%g6,#scratch
+
+	.text
+
+	.align		32
+ENTRY(__memcpy_niagara1)
+# ifndef USE_BPR
+	srl		%o2, 0, %o2
+# endif
+100:	/* %o0=dst, %o1=src, %o2=len */
+	mov		%o0, %g5
+	cmp		%o2, 0
+	be,pn		%XCC, 85f
+218:	 or		%o0, %o1, %o3
+	cmp		%o2, 16
+	blu,a,pn	%XCC, 80f
+	 or		%o3, %o2, %o3
+
+	/* 2 blocks (128 bytes) is the minimum we can do the block
+	 * copy with.  We need to ensure that we'll iterate at least
+	 * once in the block copy loop.  At worst we'll need to align
+	 * the destination to a 64-byte boundary which can chew up
+	 * to (64 - 1) bytes from the length before we perform the
+	 * block copy loop.
+	 */
+	cmp		%o2, (2 * 64)
+	blu,pt		%XCC, 70f
+	 andcc		%o3, 0x7, %g0
+
+	/* %o0:	dst
+	 * %o1:	src
+	 * %o2:	len  (known to be >= 128)
+	 *
+	 * The block copy loops will use %o4/%o5,%g2/%g3 as
+	 * temporaries while copying the data.
+	 */
+
+	LOAD(prefetch, %o1, #one_read)
+	wr		%g0, ASI_BLK_INIT_QUAD_LDD_P, %asi
+
+	/* Align destination on 64-byte boundary.  */
+	andcc		%o0, (64 - 1), %o4
+	be,pt		%XCC, 2f
+	 sub		%o4, 64, %o4
+	sub		%g0, %o4, %o4	! bytes to align dst
+	sub		%o2, %o4, %o2
+1:	subcc		%o4, 1, %o4
+	LOAD(ldub, %o1, %g1)
+	STORE(stb, %g1, %o0)
+	add		%o1, 1, %o1
+	bne,pt		%XCC, 1b
+	add		%o0, 1, %o0
+
+	/* If the source is on a 16-byte boundary we can do
+	 * the direct block copy loop.  If it is 8-byte aligned
+	 * we can do the 16-byte loads offset by -8 bytes and the
+	 * init stores offset by one register.
+	 *
+	 * If the source is not even 8-byte aligned, we need to do
+	 * shifting and masking (basically integer faligndata).
+	 *
+	 * The careful bit with init stores is that if we store
+	 * to any part of the cache line we have to store the whole
+	 * cacheline else we can end up with corrupt L2 cache line
+	 * contents.  Since the loop works on 64-bytes of 64-byte
+	 * aligned store data at a time, this is easy to ensure.
+	 */
+2:
+	andcc		%o1, (16 - 1), %o4
+	andn		%o2, (64 - 1), %g1	! block copy loop iterator
+	sub		%o2, %g1, %o2		! final sub-block copy bytes
+	be,pt		%XCC, 50f
+	 cmp		%o4, 8
+	be,a,pt		%XCC, 10f
+	 sub		%o1, 0x8, %o1
+
+	/* Neither 8-byte nor 16-byte aligned, shift and mask.  */
+	mov		%g1, %o4
+	and		%o1, 0x7, %g1
+	sll		%g1, 3, %g1
+	mov		64, %o3
+	andn		%o1, 0x7, %o1
+	LOAD(ldx, %o1, %g2)
+	sub		%o3, %g1, %o3
+	sllx		%g2, %g1, %g2
+
+#define SWIVEL_ONE_DWORD(SRC, TMP1, TMP2, PRE_VAL, PRE_SHIFT, POST_SHIFT, DST)\
+	LOAD(ldx, SRC, TMP1); \
+	srlx		TMP1, PRE_SHIFT, TMP2; \
+	or		TMP2, PRE_VAL, TMP2; \
+	STORE_INIT(TMP2, DST); \
+	sllx		TMP1, POST_SHIFT, PRE_VAL;
+
+1:	add		%o1, 0x8, %o1
+	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x00)
+	add		%o1, 0x8, %o1
+	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x08)
+	add		%o1, 0x8, %o1
+	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x10)
+	add		%o1, 0x8, %o1
+	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x18)
+	add		%o1, 32, %o1
+	LOAD(prefetch, %o1, #one_read)
+	sub		%o1, 32 - 8, %o1
+	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x20)
+	add		%o1, 8, %o1
+	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x28)
+	add		%o1, 8, %o1
+	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x30)
+	add		%o1, 8, %o1
+	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x38)
+	subcc		%o4, 64, %o4
+	bne,pt		%XCC, 1b
+	 add		%o0, 64, %o0
+
+#undef SWIVEL_ONE_DWORD
+
+	srl		%g1, 3, %g1
+	ba,pt		%XCC, 60f
+	 add		%o1, %g1, %o1
+
+10:	/* Destination is 64-byte aligned, source was only 8-byte
+	 * aligned but it has been subtracted by 8 and we perform
+	 * one twin load ahead, then add 8 back into source when
+	 * we finish the loop.
+	 */
+	LOAD_TWIN(%o1, %o4, %o5)
+1:	add		%o1, 16, %o1
+	LOAD_TWIN(%o1, %g2, %g3)
+	add		%o1, 16 + 32, %o1
+	LOAD(prefetch, %o1, #one_read)
+	sub		%o1, 32, %o1
+	STORE_INIT(%o5, %o0 + 0x00)		! initializes cache line
+	STORE_INIT(%g2, %o0 + 0x08)
+	LOAD_TWIN(%o1, %o4, %o5)
+	add		%o1, 16, %o1
+	STORE_INIT(%g3, %o0 + 0x10)
+	STORE_INIT(%o4, %o0 + 0x18)
+	LOAD_TWIN(%o1, %g2, %g3)
+	add		%o1, 16, %o1
+	STORE_INIT(%o5, %o0 + 0x20)
+	STORE_INIT(%g2, %o0 + 0x28)
+	LOAD_TWIN(%o1, %o4, %o5)
+	STORE_INIT(%g3, %o0 + 0x30)
+	STORE_INIT(%o4, %o0 + 0x38)
+	subcc		%g1, 64, %g1
+	bne,pt		%XCC, 1b
+	 add		%o0, 64, %o0
+
+	ba,pt		%XCC, 60f
+	 add		%o1, 0x8, %o1
+
+50:	/* Destination is 64-byte aligned, and source is 16-byte
+	 * aligned.
+	 */
+1:	LOAD_TWIN(%o1, %o4, %o5)
+	add	%o1, 16, %o1
+	LOAD_TWIN(%o1, %g2, %g3)
+	add	%o1, 16 + 32, %o1
+	LOAD(prefetch, %o1, #one_read)
+	sub	%o1, 32, %o1
+	STORE_INIT(%o4, %o0 + 0x00)		! initializes cache line
+	STORE_INIT(%o5, %o0 + 0x08)
+	LOAD_TWIN(%o1, %o4, %o5)
+	add	%o1, 16, %o1
+	STORE_INIT(%g2, %o0 + 0x10)
+	STORE_INIT(%g3, %o0 + 0x18)
+	LOAD_TWIN(%o1, %g2, %g3)
+	add	%o1, 16, %o1
+	STORE_INIT(%o4, %o0 + 0x20)
+	STORE_INIT(%o5, %o0 + 0x28)
+	STORE_INIT(%g2, %o0 + 0x30)
+	STORE_INIT(%g3, %o0 + 0x38)
+	subcc	%g1, 64, %g1
+	bne,pt	%XCC, 1b
+	 add	%o0, 64, %o0
+	/* fall through */
+
+60:
+	/* %o2 contains any final bytes still needed to be copied
+	 * over. If anything is left, we copy it one byte at a time.
+	 */
+	wr		%g0, ASI_PNF, %asi
+	brz,pt		%o2, 85f
+	 sub		%o0, %o1, %o3
+	ba,a,pt		%XCC, 90f
+
+	.align		64
+70: /* 16 < len <= 64 */
+	bne,pn		%XCC, 75f
+	 sub		%o0, %o1, %o3
+
+72:
+	andn		%o2, 0xf, %o4
+	and		%o2, 0xf, %o2
+1:	subcc		%o4, 0x10, %o4
+	LOAD(ldx, %o1, %o5)
+	add		%o1, 0x08, %o1
+	LOAD(ldx, %o1, %g1)
+	sub		%o1, 0x08, %o1
+	STORE(stx, %o5, %o1 + %o3)
+	add		%o1, 0x8, %o1
+	STORE(stx, %g1, %o1 + %o3)
+	bgu,pt		%XCC, 1b
+	 add		%o1, 0x8, %o1
+73:	andcc		%o2, 0x8, %g0
+	be,pt		%XCC, 1f
+	 nop
+	sub		%o2, 0x8, %o2
+	LOAD(ldx, %o1, %o5)
+	STORE(stx, %o5, %o1 + %o3)
+	add		%o1, 0x8, %o1
+1:	andcc		%o2, 0x4, %g0
+	be,pt		%XCC, 1f
+	 nop
+	sub		%o2, 0x4, %o2
+	LOAD(lduw, %o1, %o5)
+	STORE(stw, %o5, %o1 + %o3)
+	add		%o1, 0x4, %o1
+1:	cmp		%o2, 0
+	be,pt		%XCC, 85f
+	 nop
+	ba,pt		%XCC, 90f
+	 nop
+
+75:
+	andcc		%o0, 0x7, %g1
+	sub		%g1, 0x8, %g1
+	be,pn		%icc, 2f
+	 sub		%g0, %g1, %g1
+	sub		%o2, %g1, %o2
+
+1:	subcc		%g1, 1, %g1
+	LOAD(ldub, %o1, %o5)
+	STORE(stb, %o5, %o1 + %o3)
+	bgu,pt		%icc, 1b
+	 add		%o1, 1, %o1
+
+2:	add		%o1, %o3, %o0
+	andcc		%o1, 0x7, %g1
+	bne,pt		%icc, 8f
+	 sll		%g1, 3, %g1
+
+	cmp		%o2, 16
+	bgeu,pt		%icc, 72b
+	 nop
+	ba,a,pt		%XCC, 73b
+
+8:	mov		64, %o3
+	andn		%o1, 0x7, %o1
+	LOAD(ldx, %o1, %g2)
+	sub		%o3, %g1, %o3
+	andn		%o2, 0x7, %o4
+	sllx		%g2, %g1, %g2
+1:	add		%o1, 0x8, %o1
+	LOAD(ldx, %o1, %g3)
+	subcc		%o4, 0x8, %o4
+	srlx		%g3, %o3, %o5
+	or		%o5, %g2, %o5
+	STORE(stx, %o5, %o0)
+	add		%o0, 0x8, %o0
+	bgu,pt		%icc, 1b
+	 sllx		%g3, %g1, %g2
+
+	srl		%g1, 3, %g1
+	andcc		%o2, 0x7, %o2
+	be,pn		%icc, 85f
+	 add		%o1, %g1, %o1
+	ba,pt		%XCC, 90f
+	 sub		%o0, %o1, %o3
+
+	.align		64
+80: /* 0 < len <= 16 */
+	andcc		%o3, 0x3, %g0
+	bne,pn		%XCC, 90f
+	 sub		%o0, %o1, %o3
+
+1:
+	subcc		%o2, 4, %o2
+	LOAD(lduw, %o1, %g1)
+	STORE(stw, %g1, %o1 + %o3)
+	bgu,pt		%XCC, 1b
+	 add		%o1, 4, %o1
+
+85:	retl
+	 mov		%g5, %o0
+
+	.align		32
+90:
+	subcc		%o2, 1, %o2
+	LOAD(ldub, %o1, %g1)
+	STORE(stb, %g1, %o1 + %o3)
+	bgu,pt		%XCC, 90b
+	 add		%o1, 1, %o1
+	retl
+	 mov		%g5, %o0
+
+END(__memcpy_niagara1)
+
+#endif
diff --git a/sysdeps/sparc/sparc64/multiarch/memcpy-niagara2.S b/sysdeps/sparc/sparc64/multiarch/memcpy-niagara2.S
new file mode 100644
index 0000000000..35f6989aca
--- /dev/null
+++ b/sysdeps/sparc/sparc64/multiarch/memcpy-niagara2.S
@@ -0,0 +1,492 @@
+/* Copy SIZE bytes from SRC to DEST.  For SUN4V Niagara-2.
+   Copyright (C) 2007, 2008 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by David S. Miller (davem@davemloft.net)
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+#define ASI_BLK_INIT_QUAD_LDD_P	0xe2
+#define ASI_BLK_P		0xf0
+#define ASI_P			0x80
+#define ASI_PNF			0x82
+
+#define FPRS_FEF		0x04
+
+#define VISEntryHalf			\
+	rd	%fprs, %o5;		\
+	wr	%g0, FPRS_FEF, %fprs
+
+#define VISExitHalf			\
+	and	%o5, FPRS_FEF, %o5;	\
+	wr	%o5, 0x0, %fprs
+
+#define STORE_ASI		ASI_BLK_INIT_QUAD_LDD_P
+
+#define LOAD(type,addr,dest)	type [addr], dest
+#define LOAD_BLK(addr,dest)	ldda [addr] ASI_BLK_P, dest
+#define STORE(type,src,addr)	type src, [addr]
+#define STORE_BLK(src,addr)	stda src, [addr] ASI_BLK_P
+#define STORE_INIT(src,addr)	stxa src, [addr] STORE_ASI
+
+#ifndef XCC
+#define USE_BPR
+#define XCC xcc
+#endif
+
+#define FREG_FROB(x0, x1, x2, x3, x4, x5, x6, x7, x8) \
+	faligndata	%x0, %x1, %f0; \
+	faligndata	%x1, %x2, %f2; \
+	faligndata	%x2, %x3, %f4; \
+	faligndata	%x3, %x4, %f6; \
+	faligndata	%x4, %x5, %f8; \
+	faligndata	%x5, %x6, %f10; \
+	faligndata	%x6, %x7, %f12; \
+	faligndata	%x7, %x8, %f14;
+
+#define FREG_MOVE_1(x0) \
+	fmovd		%x0, %f0;
+#define FREG_MOVE_2(x0, x1) \
+	fmovd		%x0, %f0; \
+	fmovd		%x1, %f2;
+#define FREG_MOVE_3(x0, x1, x2) \
+	fmovd		%x0, %f0; \
+	fmovd		%x1, %f2; \
+	fmovd		%x2, %f4;
+#define FREG_MOVE_4(x0, x1, x2, x3) \
+	fmovd		%x0, %f0; \
+	fmovd		%x1, %f2; \
+	fmovd		%x2, %f4; \
+	fmovd		%x3, %f6;
+#define FREG_MOVE_5(x0, x1, x2, x3, x4) \
+	fmovd		%x0, %f0; \
+	fmovd		%x1, %f2; \
+	fmovd		%x2, %f4; \
+	fmovd		%x3, %f6; \
+	fmovd		%x4, %f8;
+#define FREG_MOVE_6(x0, x1, x2, x3, x4, x5) \
+	fmovd		%x0, %f0; \
+	fmovd		%x1, %f2; \
+	fmovd		%x2, %f4; \
+	fmovd		%x3, %f6; \
+	fmovd		%x4, %f8; \
+	fmovd		%x5, %f10;
+#define FREG_MOVE_7(x0, x1, x2, x3, x4, x5, x6) \
+	fmovd		%x0, %f0; \
+	fmovd		%x1, %f2; \
+	fmovd		%x2, %f4; \
+	fmovd		%x3, %f6; \
+	fmovd		%x4, %f8; \
+	fmovd		%x5, %f10; \
+	fmovd		%x6, %f12;
+#define FREG_MOVE_8(x0, x1, x2, x3, x4, x5, x6, x7) \
+	fmovd		%x0, %f0; \
+	fmovd		%x1, %f2; \
+	fmovd		%x2, %f4; \
+	fmovd		%x3, %f6; \
+	fmovd		%x4, %f8; \
+	fmovd		%x5, %f10; \
+	fmovd		%x6, %f12; \
+	fmovd		%x7, %f14;
+#define FREG_LOAD_1(base, x0) \
+	LOAD(ldd, base + 0x00, %x0)
+#define FREG_LOAD_2(base, x0, x1) \
+	LOAD(ldd, base + 0x00, %x0); \
+	LOAD(ldd, base + 0x08, %x1);
+#define FREG_LOAD_3(base, x0, x1, x2) \
+	LOAD(ldd, base + 0x00, %x0); \
+	LOAD(ldd, base + 0x08, %x1); \
+	LOAD(ldd, base + 0x10, %x2);
+#define FREG_LOAD_4(base, x0, x1, x2, x3) \
+	LOAD(ldd, base + 0x00, %x0); \
+	LOAD(ldd, base + 0x08, %x1); \
+	LOAD(ldd, base + 0x10, %x2); \
+	LOAD(ldd, base + 0x18, %x3);
+#define FREG_LOAD_5(base, x0, x1, x2, x3, x4) \
+	LOAD(ldd, base + 0x00, %x0); \
+	LOAD(ldd, base + 0x08, %x1); \
+	LOAD(ldd, base + 0x10, %x2); \
+	LOAD(ldd, base + 0x18, %x3); \
+	LOAD(ldd, base + 0x20, %x4);
+#define FREG_LOAD_6(base, x0, x1, x2, x3, x4, x5) \
+	LOAD(ldd, base + 0x00, %x0); \
+	LOAD(ldd, base + 0x08, %x1); \
+	LOAD(ldd, base + 0x10, %x2); \
+	LOAD(ldd, base + 0x18, %x3); \
+	LOAD(ldd, base + 0x20, %x4); \
+	LOAD(ldd, base + 0x28, %x5);
+#define FREG_LOAD_7(base, x0, x1, x2, x3, x4, x5, x6) \
+	LOAD(ldd, base + 0x00, %x0); \
+	LOAD(ldd, base + 0x08, %x1); \
+	LOAD(ldd, base + 0x10, %x2); \
+	LOAD(ldd, base + 0x18, %x3); \
+	LOAD(ldd, base + 0x20, %x4); \
+	LOAD(ldd, base + 0x28, %x5); \
+	LOAD(ldd, base + 0x30, %x6);
+
+#if !defined NOT_IN_libc
+
+	.register	%g2,#scratch
+	.register	%g3,#scratch
+	.register	%g6,#scratch
+
+	.text
+
+	.align		32
+ENTRY(__memcpy_niagara2)
+# ifndef USE_BPR
+	srl		%o2, 0, %o2
+# endif
+100:	/* %o0=dst, %o1=src, %o2=len */
+	mov		%o0, %g5
+	cmp		%o2, 0
+	be,pn		%XCC, 85f
+218:	 or		%o0, %o1, %o3
+	cmp		%o2, 16
+	blu,a,pn	%XCC, 80f
+	 or		%o3, %o2, %o3
+
+	/* 2 blocks (128 bytes) is the minimum we can do the block
+	 * copy with.  We need to ensure that we'll iterate at least
+	 * once in the block copy loop.  At worst we'll need to align
+	 * the destination to a 64-byte boundary which can chew up
+	 * to (64 - 1) bytes from the length before we perform the
+	 * block copy loop.
+	 *
+	 * However, the cut-off point, performance wise, is around
+	 * 4 64-byte blocks.
+	 */
+	cmp		%o2, (4 * 64)
+	blu,pt		%XCC, 75f
+	 andcc		%o3, 0x7, %g0
+
+	/* %o0:	dst
+	 * %o1:	src
+	 * %o2:	len  (known to be >= 128)
+	 *
+	 * The block copy loops can use %o4, %g2, %g3 as
+	 * temporaries while copying the data.  %o5 must
+	 * be preserved between VISEntryHalf and VISExitHalf
+	 */
+
+	LOAD(prefetch, %o1 + 0x000, #one_read)
+	LOAD(prefetch, %o1 + 0x040, #one_read)
+	LOAD(prefetch, %o1 + 0x080, #one_read)
+
+	/* Align destination on 64-byte boundary.  */
+	andcc		%o0, (64 - 1), %o4
+	be,pt		%XCC, 2f
+	 sub		%o4, 64, %o4
+	sub		%g0, %o4, %o4	! bytes to align dst
+	sub		%o2, %o4, %o2
+1:	subcc		%o4, 1, %o4
+	LOAD(ldub, %o1, %g1)
+	STORE(stb, %g1, %o0)
+	add		%o1, 1, %o1
+	bne,pt		%XCC, 1b
+	add		%o0, 1, %o0
+
+2:
+	/* Clobbers o5/g1/g2/g3/g7/icc/xcc.  We must preserve
+	 * o5 from here until we hit VISExitHalf.
+	 */
+	VISEntryHalf
+
+	alignaddr	%o1, %g0, %g0
+
+	add		%o1, (64 - 1), %o4
+	andn		%o4, (64 - 1), %o4
+	andn		%o2, (64 - 1), %g1
+	sub		%o2, %g1, %o2
+
+	and		%o1, (64 - 1), %g2
+	add		%o1, %g1, %o1
+	sub		%o0, %o4, %g3
+	brz,pt		%g2, 190f
+	 cmp		%g2, 32
+	blu,a		5f
+	 cmp		%g2, 16
+	cmp		%g2, 48
+	blu,a		4f
+	 cmp		%g2, 40
+	cmp		%g2, 56
+	blu		170f
+	 nop
+	ba,a,pt		%xcc, 180f
+
+4:	/* 32 <= low bits < 48 */
+	blu		150f
+	 nop
+	ba,a,pt		%xcc, 160f
+5:	/* 0 < low bits < 32 */
+	blu,a		6f
+	 cmp		%g2, 8
+	cmp		%g2, 24
+	blu		130f
+	 nop
+	ba,a,pt		%xcc, 140f
+6:	/* 0 < low bits < 16 */
+	bgeu		120f
+	 nop
+	/* fall through for 0 < low bits < 8 */
+110:	sub		%o4, 64, %g2
+	LOAD_BLK(%g2, %f0)
+1:	STORE_INIT(%g0, %o4 + %g3)
+	LOAD_BLK(%o4, %f16)
+	FREG_FROB(f0, f2, f4, f6, f8, f10, f12, f14, f16)
+	STORE_BLK(%f0, %o4 + %g3)
+	FREG_MOVE_8(f16, f18, f20, f22, f24, f26, f28, f30)
+	subcc		%g1, 64, %g1
+	add		%o4, 64, %o4
+	bne,pt		%XCC, 1b
+	 LOAD(prefetch, %o4 + 64, #one_read)
+	ba,pt		%xcc, 195f
+	 nop
+
+120:	sub		%o4, 56, %g2
+	FREG_LOAD_7(%g2, f0, f2, f4, f6, f8, f10, f12)
+1:	STORE_INIT(%g0, %o4 + %g3)
+	LOAD_BLK(%o4, %f16)
+	FREG_FROB(f0, f2, f4, f6, f8, f10, f12, f16, f18)
+	STORE_BLK(%f0, %o4 + %g3)
+	FREG_MOVE_7(f18, f20, f22, f24, f26, f28, f30)
+	subcc		%g1, 64, %g1
+	add		%o4, 64, %o4
+	bne,pt		%XCC, 1b
+	 LOAD(prefetch, %o4 + 64, #one_read)
+	ba,pt		%xcc, 195f
+	 nop
+
+130:	sub		%o4, 48, %g2
+	FREG_LOAD_6(%g2, f0, f2, f4, f6, f8, f10)
+1:	STORE_INIT(%g0, %o4 + %g3)
+	LOAD_BLK(%o4, %f16)
+	FREG_FROB(f0, f2, f4, f6, f8, f10, f16, f18, f20)
+	STORE_BLK(%f0, %o4 + %g3)
+	FREG_MOVE_6(f20, f22, f24, f26, f28, f30)
+	subcc		%g1, 64, %g1
+	add		%o4, 64, %o4
+	bne,pt		%XCC, 1b
+	 LOAD(prefetch, %o4 + 64, #one_read)
+	ba,pt		%xcc, 195f
+	 nop
+
+140:	sub		%o4, 40, %g2
+	FREG_LOAD_5(%g2, f0, f2, f4, f6, f8)
+1:	STORE_INIT(%g0, %o4 + %g3)
+	LOAD_BLK(%o4, %f16)
+	FREG_FROB(f0, f2, f4, f6, f8, f16, f18, f20, f22)
+	STORE_BLK(%f0, %o4 + %g3)
+	FREG_MOVE_5(f22, f24, f26, f28, f30)
+	subcc		%g1, 64, %g1
+	add		%o4, 64, %o4
+	bne,pt		%XCC, 1b
+	 LOAD(prefetch, %o4 + 64, #one_read)
+	ba,pt		%xcc, 195f
+	 nop
+
+150:	sub		%o4, 32, %g2
+	FREG_LOAD_4(%g2, f0, f2, f4, f6)
+1:	STORE_INIT(%g0, %o4 + %g3)
+	LOAD_BLK(%o4, %f16)
+	FREG_FROB(f0, f2, f4, f6, f16, f18, f20, f22, f24)
+	STORE_BLK(%f0, %o4 + %g3)
+	FREG_MOVE_4(f24, f26, f28, f30)
+	subcc		%g1, 64, %g1
+	add		%o4, 64, %o4
+	bne,pt		%XCC, 1b
+	 LOAD(prefetch, %o4 + 64, #one_read)
+	ba,pt		%xcc, 195f
+	 nop
+
+160:	sub		%o4, 24, %g2
+	FREG_LOAD_3(%g2, f0, f2, f4)
+1:	STORE_INIT(%g0, %o4 + %g3)
+	LOAD_BLK(%o4, %f16)
+	FREG_FROB(f0, f2, f4, f16, f18, f20, f22, f24, f26)
+	STORE_BLK(%f0, %o4 + %g3)
+	FREG_MOVE_3(f26, f28, f30)
+	subcc		%g1, 64, %g1
+	add		%o4, 64, %o4
+	bne,pt		%XCC, 1b
+	 LOAD(prefetch, %o4 + 64, #one_read)
+	ba,pt		%xcc, 195f
+	 nop
+
+170:	sub		%o4, 16, %g2
+	FREG_LOAD_2(%g2, f0, f2)
+1:	STORE_INIT(%g0, %o4 + %g3)
+	LOAD_BLK(%o4, %f16)
+	FREG_FROB(f0, f2, f16, f18, f20, f22, f24, f26, f28)
+	STORE_BLK(%f0, %o4 + %g3)
+	FREG_MOVE_2(f28, f30)
+	subcc		%g1, 64, %g1
+	add		%o4, 64, %o4
+	bne,pt		%XCC, 1b
+	 LOAD(prefetch, %o4 + 64, #one_read)
+	ba,pt		%xcc, 195f
+	 nop
+
+180:	sub		%o4, 8, %g2
+	FREG_LOAD_1(%g2, f0)
+1:	STORE_INIT(%g0, %o4 + %g3)
+	LOAD_BLK(%o4, %f16)
+	FREG_FROB(f0, f16, f18, f20, f22, f24, f26, f28, f30)
+	STORE_BLK(%f0, %o4 + %g3)
+	FREG_MOVE_1(f30)
+	subcc		%g1, 64, %g1
+	add		%o4, 64, %o4
+	bne,pt		%XCC, 1b
+	 LOAD(prefetch, %o4 + 64, #one_read)
+	ba,pt		%xcc, 195f
+	 nop
+
+190:
+1:	STORE_INIT(%g0, %o4 + %g3)
+	subcc		%g1, 64, %g1
+	LOAD_BLK(%o4, %f0)
+	STORE_BLK(%f0, %o4 + %g3)
+	add		%o4, 64, %o4
+	bne,pt		%XCC, 1b
+	 LOAD(prefetch, %o4 + 64, #one_read)
+
+195:
+	add		%o4, %g3, %o0
+	membar		#Sync
+
+	VISExitHalf
+
+	/* %o2 contains any final bytes still needed to be copied
+	 * over. If anything is left, we copy it one byte at a time.
+	 */
+	brz,pt		%o2, 85f
+	 sub		%o0, %o1, %o3
+	ba,a,pt		%XCC, 90f
+
+	.align		64
+75: /* 16 < len <= 64 */
+	bne,pn		%XCC, 75f
+	 sub		%o0, %o1, %o3
+
+72:
+	andn		%o2, 0xf, %o4
+	and		%o2, 0xf, %o2
+1:	subcc		%o4, 0x10, %o4
+	LOAD(ldx, %o1, %o5)
+	add		%o1, 0x08, %o1
+	LOAD(ldx, %o1, %g1)
+	sub		%o1, 0x08, %o1
+	STORE(stx, %o5, %o1 + %o3)
+	add		%o1, 0x8, %o1
+	STORE(stx, %g1, %o1 + %o3)
+	bgu,pt		%XCC, 1b
+	 add		%o1, 0x8, %o1
+73:	andcc		%o2, 0x8, %g0
+	be,pt		%XCC, 1f
+	 nop
+	sub		%o2, 0x8, %o2
+	LOAD(ldx, %o1, %o5)
+	STORE(stx, %o5, %o1 + %o3)
+	add		%o1, 0x8, %o1
+1:	andcc		%o2, 0x4, %g0
+	be,pt		%XCC, 1f
+	 nop
+	sub		%o2, 0x4, %o2
+	LOAD(lduw, %o1, %o5)
+	STORE(stw, %o5, %o1 + %o3)
+	add		%o1, 0x4, %o1
+1:	cmp		%o2, 0
+	be,pt		%XCC, 85f
+	 nop
+	ba,pt		%xcc, 90f
+	 nop
+
+75:
+	andcc		%o0, 0x7, %g1
+	sub		%g1, 0x8, %g1
+	be,pn		%icc, 2f
+	 sub		%g0, %g1, %g1
+	sub		%o2, %g1, %o2
+
+1:	subcc		%g1, 1, %g1
+	LOAD(ldub, %o1, %o5)
+	STORE(stb, %o5, %o1 + %o3)
+	bgu,pt		%icc, 1b
+	 add		%o1, 1, %o1
+
+2:	add		%o1, %o3, %o0
+	andcc		%o1, 0x7, %g1
+	bne,pt		%icc, 8f
+	 sll		%g1, 3, %g1
+
+	cmp		%o2, 16
+	bgeu,pt		%icc, 72b
+	 nop
+	ba,a,pt		%xcc, 73b
+
+8:	mov		64, %o3
+	andn		%o1, 0x7, %o1
+	LOAD(ldx, %o1, %g2)
+	sub		%o3, %g1, %o3
+	andn		%o2, 0x7, %o4
+	sllx		%g2, %g1, %g2
+1:	add		%o1, 0x8, %o1
+	LOAD(ldx, %o1, %g3)
+	subcc		%o4, 0x8, %o4
+	srlx		%g3, %o3, %o5
+	or		%o5, %g2, %o5
+	STORE(stx, %o5, %o0)
+	add		%o0, 0x8, %o0
+	bgu,pt		%icc, 1b
+	 sllx		%g3, %g1, %g2
+
+	srl		%g1, 3, %g1
+	andcc		%o2, 0x7, %o2
+	be,pn		%icc, 85f
+	 add		%o1, %g1, %o1
+	ba,pt		%xcc, 90f
+	 sub		%o0, %o1, %o3
+
+	.align		64
+80: /* 0 < len <= 16 */
+	andcc		%o3, 0x3, %g0
+	bne,pn		%XCC, 90f
+	 sub		%o0, %o1, %o3
+
+1:
+	subcc		%o2, 4, %o2
+	LOAD(lduw, %o1, %g1)
+	STORE(stw, %g1, %o1 + %o3)
+	bgu,pt		%XCC, 1b
+	 add		%o1, 4, %o1
+
+85:	retl
+	 mov		%g5, %o0
+
+	.align		32
+90:
+	subcc		%o2, 1, %o2
+	LOAD(ldub, %o1, %g1)
+	STORE(stb, %g1, %o1 + %o3)
+	bgu,pt		%XCC, 90b
+	 add		%o1, 1, %o1
+	retl
+	 mov		%g5, %o0
+
+END(__memcpy_niagara2)
+
+#endif
diff --git a/sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S b/sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S
new file mode 100644
index 0000000000..34ca089f93
--- /dev/null
+++ b/sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S
@@ -0,0 +1,320 @@
+/* Copy SIZE bytes from SRC to DEST.
+   For UltraSPARC-III.
+   Copyright (C) 2001, 2003 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by David S. Miller (davem@redhat.com)
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+#define ASI_BLK_P 0xf0
+#define FPRS_FEF  0x04
+#define VISEntryHalf rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs
+#define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
+
+#ifndef XCC
+#define USE_BPR
+#define XCC xcc
+#endif
+
+#if !defined NOT_IN_libc
+
+	.register	%g2,#scratch
+	.register	%g3,#scratch
+	.register	%g6,#scratch
+
+	.text
+
+	/* Special/non-trivial issues of this code:
+	 *
+	 * 1) %o5 is preserved from VISEntryHalf to VISExitHalf
+	 * 2) Only low 32 FPU registers are used so that only the
+	 *    lower half of the FPU register set is dirtied by this
+	 *    code.  This is especially important in the kernel.
+	 * 3) This code never prefetches cachelines past the end
+	 *    of the source buffer.
+	 *
+	 * The cheetah's flexible spine, oversized liver, enlarged heart,
+	 * slender muscular body, and claws make it the swiftest hunter
+	 * in Africa and the fastest animal on land.  Can reach speeds
+	 * of up to 2.4GB per second.
+	 */
+	.align		32
+ENTRY(__memcpy_ultra3)
+
+100: /* %o0=dst, %o1=src, %o2=len */
+	mov		%o0, %g5
+	cmp		%o2, 0
+	be,pn		%XCC, out
+218:	 or		%o0, %o1, %o3
+	cmp		%o2, 16
+	bleu,a,pn	%XCC, small_copy
+	 or		%o3, %o2, %o3
+
+	cmp		%o2, 256
+	blu,pt		%XCC, medium_copy
+	 andcc		%o3, 0x7, %g0
+
+	ba,pt		%xcc, enter
+	 andcc		%o0, 0x3f, %g2
+
+	/* Here len >= 256 and condition codes reflect execution
+	 * of "andcc %o0, 0x7, %g2", done by caller.
+	 */
+	.align		64
+enter:
+	/* Is 'dst' already aligned on an 64-byte boundary? */
+	be,pt		%XCC, 2f
+
+	/* Compute abs((dst & 0x3f) - 0x40) into %g2.  This is the number
+	 * of bytes to copy to make 'dst' 64-byte aligned.  We pre-
+	 * subtract this from 'len'.
+	 */
+	 sub		%g2, 0x40, %g2
+	sub		%g0, %g2, %g2
+	sub		%o2, %g2, %o2
+
+	/* Copy %g2 bytes from src to dst, one byte at a time. */
+1:	ldub		[%o1 + 0x00], %o3
+	add		%o1, 0x1, %o1
+	add		%o0, 0x1, %o0
+	subcc		%g2, 0x1, %g2
+
+	bg,pt		%XCC, 1b
+	 stb		%o3, [%o0 + -1]
+
+2:	VISEntryHalf
+	and		%o1, 0x7, %g1
+	ba,pt		%xcc, begin
+	 alignaddr	%o1, %g0, %o1
+
+	.align		64
+begin:
+	prefetch	[%o1 + 0x000], #one_read
+	prefetch	[%o1 + 0x040], #one_read
+	andn		%o2, (0x40 - 1), %o4
+	prefetch	[%o1 + 0x080], #one_read
+	prefetch	[%o1 + 0x0c0], #one_read
+	ldd		[%o1 + 0x000], %f0
+	prefetch	[%o1 + 0x100], #one_read
+	ldd		[%o1 + 0x008], %f2
+	prefetch	[%o1 + 0x140], #one_read
+	ldd		[%o1 + 0x010], %f4
+	prefetch	[%o1 + 0x180], #one_read
+	faligndata	%f0, %f2, %f16
+	ldd		[%o1 + 0x018], %f6
+	faligndata	%f2, %f4, %f18
+	ldd		[%o1 + 0x020], %f8
+	faligndata	%f4, %f6, %f20
+	ldd		[%o1 + 0x028], %f10
+	faligndata	%f6, %f8, %f22
+
+	ldd		[%o1 + 0x030], %f12
+	faligndata	%f8, %f10, %f24
+	ldd		[%o1 + 0x038], %f14
+	faligndata	%f10, %f12, %f26
+	ldd		[%o1 + 0x040], %f0
+
+	sub		%o4, 0x80, %o4
+	add		%o1, 0x40, %o1
+	ba,pt		%xcc, loop
+	 srl		%o4, 6, %o3
+
+	.align		64
+loop:
+	ldd		[%o1 + 0x008], %f2
+	faligndata	%f12, %f14, %f28
+	ldd		[%o1 + 0x010], %f4
+	faligndata	%f14, %f0, %f30
+	stda		%f16, [%o0] ASI_BLK_P
+	ldd		[%o1 + 0x018], %f6
+	faligndata	%f0, %f2, %f16
+
+	ldd		[%o1 + 0x020], %f8
+	faligndata	%f2, %f4, %f18
+	ldd		[%o1 + 0x028], %f10
+	faligndata	%f4, %f6, %f20
+	ldd		[%o1 + 0x030], %f12
+	faligndata	%f6, %f8, %f22
+	ldd		[%o1 + 0x038], %f14
+	faligndata	%f8, %f10, %f24
+
+	ldd		[%o1 + 0x040], %f0
+	prefetch	[%o1 + 0x180], #one_read
+	faligndata	%f10, %f12, %f26
+	subcc		%o3, 0x01, %o3
+	add		%o1, 0x40, %o1
+	bg,pt		%XCC, loop
+	 add		%o0, 0x40, %o0
+
+	/* Finally we copy the last full 64-byte block. */
+loopfini:
+	ldd		[%o1 + 0x008], %f2
+	faligndata	%f12, %f14, %f28
+	ldd		[%o1 + 0x010], %f4
+	faligndata	%f14, %f0, %f30
+	stda		%f16, [%o0] ASI_BLK_P
+	ldd		[%o1 + 0x018], %f6
+	faligndata	%f0, %f2, %f16
+	ldd		[%o1 + 0x020], %f8
+	faligndata	%f2, %f4, %f18
+	ldd		[%o1 + 0x028], %f10
+	faligndata	%f4, %f6, %f20
+	ldd		[%o1 + 0x030], %f12
+	faligndata	%f6, %f8, %f22
+	ldd		[%o1 + 0x038], %f14
+	faligndata	%f8, %f10, %f24
+	cmp		%g1, 0
+	be,pt		%XCC, 1f
+	 add		%o0, 0x40, %o0
+	ldd		[%o1 + 0x040], %f0
+1:	faligndata	%f10, %f12, %f26
+	faligndata	%f12, %f14, %f28
+	faligndata	%f14, %f0, %f30
+	stda		%f16, [%o0] ASI_BLK_P
+	add		%o0, 0x40, %o0
+	add		%o1, 0x40, %o1
+	membar		#Sync
+
+	/* Now we copy the (len modulo 64) bytes at the end.
+	 * Note how we borrow the %f0 loaded above.
+	 *
+	 * Also notice how this code is careful not to perform a
+	 * load past the end of the src buffer.
+	 */
+loopend:
+	and		%o2, 0x3f, %o2
+	andcc		%o2, 0x38, %g2
+	be,pn		%XCC, endcruft
+	 subcc		%g2, 0x8, %g2
+	be,pn		%XCC, endcruft
+	 cmp		%g1, 0
+
+	be,a,pt		%XCC, 1f
+	 ldd		[%o1 + 0x00], %f0
+
+1:	ldd		[%o1 + 0x08], %f2
+	add		%o1, 0x8, %o1
+	sub		%o2, 0x8, %o2
+	subcc		%g2, 0x8, %g2
+	faligndata	%f0, %f2, %f8
+	std		%f8, [%o0 + 0x00]
+	be,pn		%XCC, endcruft
+	 add		%o0, 0x8, %o0
+	ldd		[%o1 + 0x08], %f0
+	add		%o1, 0x8, %o1
+	sub		%o2, 0x8, %o2
+	subcc		%g2, 0x8, %g2
+	faligndata	%f2, %f0, %f8
+	std		%f8, [%o0 + 0x00]
+	bne,pn		%XCC, 1b
+	 add		%o0, 0x8, %o0
+
+	/* If anything is left, we copy it one byte at a time.
+	 * Note that %g1 is (src & 0x3) saved above before the
+	 * alignaddr was performed.
+	 */
+endcruft:
+	cmp		%o2, 0
+	add		%o1, %g1, %o1
+	VISExitHalf
+	be,pn		%XCC, out
+	 sub		%o0, %o1, %o3
+
+	andcc		%g1, 0x7, %g0
+	bne,pn		%icc, small_copy_unaligned
+	 andcc		%o2, 0x8, %g0
+	be,pt		%icc, 1f
+	 nop
+	ldx		[%o1], %o5
+	stx		%o5, [%o1 + %o3]
+	add		%o1, 0x8, %o1
+
+1:	andcc		%o2, 0x4, %g0
+	be,pt		%icc, 1f
+	 nop
+	lduw		[%o1], %o5
+	stw		%o5, [%o1 + %o3]
+	add		%o1, 0x4, %o1
+
+1:	andcc		%o2, 0x2, %g0
+	be,pt		%icc, 1f
+	 nop
+	lduh		[%o1], %o5
+	sth		%o5, [%o1 + %o3]
+	add		%o1, 0x2, %o1
+
+1:	andcc		%o2, 0x1, %g0
+	be,pt		%icc, out
+	 nop
+	ldub		[%o1], %o5
+	ba,pt		%xcc, out
+	 stb		%o5, [%o1 + %o3]
+
+medium_copy: /* 16 < len <= 64 */
+	bne,pn		%XCC, small_copy_unaligned
+	 sub		%o0, %o1, %o3
+
+medium_copy_aligned:
+	andn		%o2, 0x7, %o4
+	and		%o2, 0x7, %o2
+1:	subcc		%o4, 0x8, %o4
+	ldx		[%o1], %o5
+	stx		%o5, [%o1 + %o3]
+	bgu,pt		%XCC, 1b
+	 add		%o1, 0x8, %o1
+	andcc		%o2, 0x4, %g0
+	be,pt		%XCC, 1f
+	 nop
+	sub		%o2, 0x4, %o2
+	lduw		[%o1], %o5
+	stw		%o5, [%o1 + %o3]
+	add		%o1, 0x4, %o1
+1:	cmp		%o2, 0
+	be,pt		%XCC, out
+	 nop
+	ba,pt		%xcc, small_copy_unaligned
+	 nop
+
+small_copy: /* 0 < len <= 16 */
+	andcc		%o3, 0x3, %g0
+	bne,pn		%XCC, small_copy_unaligned
+	 sub		%o0, %o1, %o3
+
+small_copy_aligned:
+	subcc		%o2, 4, %o2
+	lduw		[%o1], %g1
+	stw		%g1, [%o1 + %o3]
+	bgu,pt		%XCC, small_copy_aligned
+	 add		%o1, 4, %o1
+
+out:	retl
+	 mov		%g5, %o0
+
+	.align	32
+small_copy_unaligned:
+	subcc		%o2, 1, %o2
+	ldub		[%o1], %g1
+	stb		%g1, [%o1 + %o3]
+	bgu,pt		%XCC, small_copy_unaligned
+	 add		%o1, 1, %o1
+	retl
+	 mov		%g5, %o0
+
+END(__memcpy_ultra3)
+
+#endif
\ No newline at end of file
diff --git a/sysdeps/sparc/sparc64/multiarch/memcpy.S b/sysdeps/sparc/sparc64/multiarch/memcpy.S
new file mode 100644
index 0000000000..a708de10e2
--- /dev/null
+++ b/sysdeps/sparc/sparc64/multiarch/memcpy.S
@@ -0,0 +1,107 @@
+/* Multiple versions of memcpy
+   Copyright (C) 2010 Free Software Foundation, Inc.
+   Contributed by David S. Miller (davem@davemloft.net)
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <rtld-global-offsets.h>
+
+#if !defined NOT_IN_libc
+	.text
+ENTRY(memcpy)
+	.type	memcpy, @gnu_indirect_function
+# ifdef SHARED
+	mov	%o7, %o5
+	sethi	%hi(_GLOBAL_OFFSET_TABLE_-4), %o3
+	call	1f
+	 or	%o3, %lo(_GLOBAL_OFFSET_TABLE_+4), %o3
+1:	add	%o7, %o3, %o3
+	mov	%o5, %o7
+	sethi	%hi(_rtld_global_ro), %o2
+	or	%o2, %lo(_rtld_global_ro), %o2
+#  ifdef __arch64__
+	ldx	[%o3 + %o2], %o2
+	ldx	[%o2 + RTLD_GLOBAL_RO_DL_HWCAP_OFFSET], %o2
+#  else
+	ld	[%o3 + %o2], %o2
+	ld	[%o2 + RTLD_GLOBAL_RO_DL_HWCAP_OFFSET + 4], %o2
+#  endif
+# else
+	set	_dl_hwcap, %o3
+#  ifdef __arch64__
+	ldx	[%o3], %o2
+#  else
+	ld	[%o3 + 4], %o2
+#  endif
+# endif
+	andcc	%o2, 0x80, %g0	! HWCAP_SPARC_N2
+	be	1f
+	 andcc	%o2, 0x40, %g0	! HWCAP_SPARC_BLKINIT
+# ifdef SHARED
+	sethi	%gdop_hix22(__memcpy_niagara2), %o1
+	xor	%o1, %gdop_lox10(__memcpy_niagara2), %o1
+# else
+	set	__memcpy_niagara2, %o1
+# endif
+	ba	10f
+	 nop
+1:	be	1f
+	 andcc	%o2, 0x20, %g0	! HWCAP_SPARC_ULTRA3
+# ifdef SHARED
+	sethi	%gdop_hix22(__memcpy_niagara1), %o1
+	xor	%o1, %gdop_lox10(__memcpy_niagara1), %o1
+# else
+	set	__memcpy_niagara1, %o1
+# endif
+	ba	10f
+	 nop
+1:	be	9f
+	 nop
+# ifdef SHARED
+	sethi	%gdop_hix22(__memcpy_ultra3), %o1
+	xor	%o1, %gdop_lox10(__memcpy_ultra3), %o1
+# else
+	set	__memcpy_ultra3, %o1
+# endif
+	ba	10f
+	 nop
+9:
+# ifdef SHARED
+	sethi	%gdop_hix22(__memcpy_ultra1), %o1
+	xor	%o1, %gdop_lox10(__memcpy_ultra1), %o1
+# else
+	set	__memcpy_ultra1, %o1
+# endif
+10:
+# ifdef SHARED
+	add	%o3, %o1, %o1
+# endif
+	retl
+	 mov	%o1, %o0
+END(memcpy)
+
+# undef libc_hidden_builtin_def
+/* IFUNC doesn't work with the hidden functions in a shared library.  */
+# define libc_hidden_builtin_def(name) \
+	.globl __GI_memcpy; __GI_memcpy = __memcpy_ultra1
+
+#define memcpy __memcpy_ultra1
+
+#endif
+
+#include "../memcpy.S"
diff --git a/sysdeps/sparc/sparc64/multiarch/memset-niagara1.S b/sysdeps/sparc/sparc64/multiarch/memset-niagara1.S
new file mode 100644
index 0000000000..20ea056216
--- /dev/null
+++ b/sysdeps/sparc/sparc64/multiarch/memset-niagara1.S
@@ -0,0 +1,132 @@
+/* Set a block of memory to some byte value.  For SUN4V Niagara.
+   Copyright (C) 2006, 2008 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by David S. Miller (davem@davemloft.net)
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+#define ASI_BLK_INIT_QUAD_LDD_P	0xe2
+#define ASI_P			0x80
+#define ASI_PNF			0x82
+
+#ifndef XCC
+#define USE_BPR
+#define XCC xcc
+#endif
+
+#if !defined NOT_IN_libc
+
+	.register	%g2,#scratch
+
+	.text
+	.align		32
+
+ENTRY(__memset_niagara1)
+	/* %o0=buf, %o1=pat, %o2=len */
+	and		%o1, 0xff, %o3
+	mov		%o2, %o1
+	sllx		%o3, 8, %g1
+	or		%g1, %o3, %o2
+	sllx		%o2, 16, %g1
+	or		%g1, %o2, %o2
+	sllx		%o2, 32, %g1
+	ba,pt		%XCC, 1f
+	 or		%g1, %o2, %o2
+END(__memset_niagara1)
+
+ENTRY(__bzero_niagara1)
+	clr		%o2
+1:
+# ifndef USE_BRP
+	srl		%o1, 0, %o1
+# endif
+	brz,pn		%o1, 90f
+	 mov		%o0, %o3
+
+	wr		%g0, ASI_P, %asi
+
+	cmp		%o1, 15
+	bl,pn		%icc, 70f
+	 andcc		%o0, 0x7, %g1
+	be,pt		%XCC, 2f
+	 mov		8, %g2
+	sub		%g2, %g1, %g1
+	sub		%o1, %g1, %o1
+1:	stba		%o2, [%o0 + 0x00] %asi
+	subcc		%g1, 1, %g1
+	bne,pt		%XCC, 1b
+	 add		%o0, 1, %o0
+2:	cmp		%o1, 128
+	bl,pn		%icc, 60f
+	 andcc		%o0, (64 - 1), %g1
+	be,pt		%XCC, 40f
+	 mov		64, %g2
+	sub		%g2, %g1, %g1
+	sub		%o1, %g1, %o1
+1:	stxa		%o2, [%o0 + 0x00] %asi
+	subcc		%g1, 8, %g1
+	bne,pt		%XCC, 1b
+	 add		%o0, 8, %o0
+
+40:
+	wr		%g0, ASI_BLK_INIT_QUAD_LDD_P, %asi
+	andn		%o1, (64 - 1), %g1
+	sub		%o1, %g1, %o1
+50:
+	stxa		%o2, [%o0 + 0x00] %asi
+	stxa		%o2, [%o0 + 0x08] %asi
+	stxa		%o2, [%o0 + 0x10] %asi
+	stxa		%o2, [%o0 + 0x18] %asi
+	stxa		%o2, [%o0 + 0x20] %asi
+	stxa		%o2, [%o0 + 0x28] %asi
+	stxa		%o2, [%o0 + 0x30] %asi
+	stxa		%o2, [%o0 + 0x38] %asi
+	subcc		%g1, 64, %g1
+	bne,pt		%XCC, 50b
+	 add		%o0, 64, %o0
+
+	wr		%g0, ASI_P, %asi
+	brz,pn		%o1, 80f
+60:
+	 andncc		%o1, 0x7, %g1
+	be,pn		%XCC, 2f
+	 sub		%o1, %g1, %o1
+1:	stxa		%o2, [%o0 + 0x00] %asi
+	subcc		%g1, 8, %g1
+	bne,pt		%XCC, 1b
+	 add		%o0, 8, %o0
+2:	brz,pt		%o1, 80f
+	 nop
+
+70:
+1:	stba		%o2, [%o0 + 0x00] %asi
+	subcc		%o1, 1, %o1
+	bne,pt		%icc, 1b
+	 add		%o0, 1, %o0
+
+	/* fallthrough */
+
+80:
+	wr		%g0, ASI_PNF, %asi
+
+90:
+	retl
+	 mov		%o3, %o0
+END(__bzero_niagara1)
+
+#endif
diff --git a/sysdeps/sparc/sparc64/multiarch/memset.S b/sysdeps/sparc/sparc64/multiarch/memset.S
new file mode 100644
index 0000000000..23e513f18f
--- /dev/null
+++ b/sysdeps/sparc/sparc64/multiarch/memset.S
@@ -0,0 +1,145 @@
+/* Multiple versions of memset and bzero
+   Copyright (C) 2010 Free Software Foundation, Inc.
+   Contributed by David S. Miller (davem@davemloft.net)
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <rtld-global-offsets.h>
+
+#if !defined NOT_IN_libc
+	.text
+ENTRY(memset)
+	.type	memset, @gnu_indirect_function
+# ifdef SHARED
+	mov	%o7, %o5
+	sethi	%hi(_GLOBAL_OFFSET_TABLE_-4), %o3
+	call	1f
+	 or	%o3, %lo(_GLOBAL_OFFSET_TABLE_+4), %o3
+1:	add	%o7, %o3, %o3
+	mov	%o5, %o7
+	sethi	%hi(_rtld_global_ro), %o2
+	or	%o2, %lo(_rtld_global_ro), %o2
+#  ifdef __arch64__
+	ldx	[%o3 + %o2], %o2
+	ldx	[%o2 + RTLD_GLOBAL_RO_DL_HWCAP_OFFSET], %o2
+#  else
+	ld	[%o3 + %o2], %o2
+	ld	[%o2 + RTLD_GLOBAL_RO_DL_HWCAP_OFFSET + 4], %o2
+#  endif
+# else
+	set	_dl_hwcap, %o3
+#  ifdef __arch64__
+	ldx	[%o3], %o2
+#  else
+	ld	[%o3 + 4], %o2
+#  endif
+# endif
+	andcc	%o2, 0x40, %g0	! HWCAP_SPARC_BLKINIT
+	be	9f
+	 nop
+# ifdef SHARED
+	sethi	%gdop_hix22(__memset_niagara1), %o1
+	xor	%o1, %gdop_lox10(__memset_niagara1), %o1
+# else
+	set	__memset_niagara1, %o1
+# endif
+	ba	10f
+	 nop
+9:
+# ifdef SHARED
+	sethi	%gdop_hix22(__memset_ultra1), %o1
+	xor	%o1, %gdop_lox10(__memset_ultra1), %o1
+# else
+	set	__memset_ultra1, %o1
+# endif
+10:
+# ifdef SHARED
+	add	%o3, %o1, %o1
+# endif
+	retl
+	 mov	%o1, %o0
+END(memset)
+
+ENTRY(__bzero)
+	.type	bzero, @gnu_indirect_function
+# ifdef SHARED
+	mov	%o7, %o5
+	sethi	%hi(_GLOBAL_OFFSET_TABLE_-4), %o3
+	call	1f
+	 or	%o3, %lo(_GLOBAL_OFFSET_TABLE_+4), %o3
+1:	add	%o7, %o3, %o3
+	mov	%o5, %o7
+	sethi	%hi(_rtld_global_ro), %o2
+	or	%o2, %lo(_rtld_global_ro), %o2
+#  ifdef __arch64__
+	ldx	[%o3 + %o2], %o2
+	ldx	[%o2 + RTLD_GLOBAL_RO_DL_HWCAP_OFFSET], %o2
+#  else
+	ld	[%o3 + %o2], %o2
+	ld	[%o2 + RTLD_GLOBAL_RO_DL_HWCAP_OFFSET + 4], %o2
+#  endif
+# else
+	set	_dl_hwcap, %o3
+#  ifdef __arch64__
+	ldx	[%o3], %o2
+#  else
+	ld	[%o3 + 4], %o2
+#  endif
+# endif
+	andcc	%o2, 0x40, %g0	! HWCAP_SPARC_BLKINIT
+	be	9f
+	 nop
+# ifdef SHARED
+	sethi	%gdop_hix22(__bzero_niagara1), %o1
+	xor	%o1, %gdop_lox10(__bzero_niagara1), %o1
+# else
+	set	__bzero_niagara1, %o1
+# endif
+	ba	10f
+	 nop
+9:
+# ifdef SHARED
+	sethi	%gdop_hix22(__memset_ultra1), %o1
+	xor	%o1, %gdop_lox10(__memset_ultra1), %o1
+# else
+	set	__bzero_ultra1, %o1
+# endif
+10:
+# ifdef SHARED
+	add	%o3, %o1, %o1
+# endif
+	retl
+	 mov	%o1, %o0
+END(__bzero)
+
+weak_alias (__bzero, bzero)
+
+# undef weak_alias
+# define weak_alias(a, b)
+
+# undef libc_hidden_builtin_def
+/* IFUNC doesn't work with the hidden functions in a shared library.  */
+# define libc_hidden_builtin_def(name) \
+	.globl __GI_memset; __GI_memset = __memset_ultra1
+
+#define memset __memset_ultra1
+#define __bzero __bzero_ultra1
+
+#endif
+
+#include "../memset.S"