powerpc: POWER7 optimizations ibm/2.8/master

    Add optimizations for classification functions (32-bit and 64-bit) and
    string functions (32-bit and 64-bit).

    powerpc: Re-work the Implies structure

    This patch tries to organize the implies files for ppc, since there are
    a number of processors and most of them are compatible with each other
    (backwards compatible).

    Having in mind that we start the search for processor-specific files in
    the sysdeps/unix/sysv/linux tree
    (sysdeps/unix/sysv/linux/powerpc/powerpc[32|64]/[processor]/fpu to be
    exact), we would like to grab any linux-specific code from that tree
    prior to going through the other tree (sysdeps/powerpc/...).

    For that, i removed the Implies files that were originally inside the
    fpu directories and placed then in the non-fpu directories (still inside
    the unix/sysv/linux tree). If no processor-specific/linux-specific files
    could be found, we "imply" the other tree's (sysdeps/powerpc/...) fpu
    directory for that specific processor AND also the non-fpu directory for
    that same tree.

    If, again, no processor-specific code is found, we read another Implies
    file that will point to the most compatible processor that we should
    grab code from, and so on, until we reach the power4 processor.

    So, in summary, the Implies files will live inside these directories
    now:

    * sysdeps/unix/sysv/linux/powerpc/powerpc[32|64]/[processor]
    * sysdeps/powerpc/powerpc[32|64]/[processor]

    Practical example of the order we will use to pick power6-specific code
    with the new structure.

    sysdeps/unix/sysv/linux/powerpc/powerpc[32|64]/power6/fpu ->
    sysdeps/unix/sysv/linux/powerpc/powerpc[32|64]/power6 ->
    sysdeps/powerpc/powerpc[32|64]/power6/fpu ->
    sysdeps/powerpc/powerpc[32|64]/power6 ->
    sysdeps/powerpc/powerpc[32|64]/power5+/fpu ->
    sysdeps/powerpc/powerpc[32|64]/power5+ ->
    sysdeps/powerpc/powerpc[32|64]/power5/fpu ->
    sysdeps/powerpc/powerpc[32|64]/power5 ->
    sysdeps/powerpc/powerpc[32|64]/power4/fpu ->
    sysdeps/powerpc/powerpc[32|64]/power4 (from here, it'll go to the
    generic path as usual)

1 files changed, 988 insertions, 0 deletions

diff --git a/sysdeps/powerpc/powerpc32/power7/memcmp.S b/sysdeps/powerpc/powerpc32/power7/memcmp.S
new file mode 100644
index 0000000000..d529b492fc
--- /dev/null
+++ b/sysdeps/powerpc/powerpc32/power7/memcmp.S
@@ -0,0 +1,988 @@
+/* Optimized memcmp implementation for POWER7/PowerPC32.
+   Copyright (C) 2010 Free Software Foundation, Inc.
+   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., 51 Franklin Street, Fifth Floor, Boston MA
+   02110-1301 USA.  */
+
+#include <sysdep.h>
+#include <bp-sym.h>
+#include <bp-asm.h>
+
+/* int [r3] memcmp (const char *s1 [r3],
+		    const char *s2 [r4],
+		    size_t size [r5])  */
+
+	.machine power7
+EALIGN (BP_SYM(memcmp),4,0)
+	CALL_MCOUNT
+
+#define rTMP	r0
+#define rRTN	r3
+#define rSTR1	r3	/* first string arg */
+#define rSTR2	r4	/* second string arg */
+#define rN	r5	/* max string length */
+#define rWORD1	r6	/* current word in s1 */
+#define rWORD2	r7	/* current word in s2 */
+#define rWORD3	r8	/* next word in s1 */
+#define rWORD4	r9	/* next word in s2 */
+#define rWORD5	r10	/* next word in s1 */
+#define rWORD6	r11	/* next word in s2 */
+#define rBITDIF	r12	/* bits that differ in s1 & s2 words */
+#define rWORD7	r30	/* next word in s1 */
+#define rWORD8	r31	/* next word in s2 */
+
+	xor	rTMP,rSTR2,rSTR1
+	cmplwi	cr6,rN,0
+	cmplwi	cr1,rN,12
+	clrlwi.	rTMP,rTMP,30
+	clrlwi	rBITDIF,rSTR1,30
+	cmplwi	cr5,rBITDIF,0
+	beq-	cr6,L(zeroLength)
+	dcbt	0,rSTR1
+	dcbt	0,rSTR2
+
+	/* If less than 8 bytes or not aligned, use the unaligned
+	   byte loop.  */
+
+	blt	cr1,L(bytealigned)
+	stwu	1,-64(1)
+	cfi_adjust_cfa_offset(64)
+	stw	r31,48(1)
+	cfi_offset(31,(48-64))
+	stw	r30,44(1)
+	cfi_offset(30,(44-64))
+	bne	L(unaligned)
+/* At this point we know both strings have the same alignment and the
+   compare length is at least 8 bytes.  rBITDIF contains the low order
+   2 bits of rSTR1 and cr5 contains the result of the logical compare
+   of rBITDIF to 0.  If rBITDIF == 0 then we are already word
+   aligned and can perform the word aligned loop.
+
+   Otherwise we know the two strings have the same alignment (but not
+   yet word aligned).  So we force the string addresses to the next lower
+   word boundary and special case this first word using shift left to
+   eliminate bits preceeding the first byte.  Since we want to join the
+   normal (word aligned) compare loop, starting at the second word,
+   we need to adjust the length (rN) and special case the loop
+   versioning for the first word. This insures that the loop count is
+   correct and the first word (shifted) is in the expected register pair. */
+	.align	4
+L(samealignment):
+	clrrwi	rSTR1,rSTR1,2
+	clrrwi	rSTR2,rSTR2,2
+	beq	cr5,L(Waligned)
+	add	rN,rN,rBITDIF
+	slwi	r11,rBITDIF,3
+	srwi	rTMP,rN,4	/* Divide by 16 */
+	andi.	rBITDIF,rN,12	/* Get the word remainder */
+	lwz	rWORD1,0(rSTR1)
+	lwz	rWORD2,0(rSTR2)
+	cmplwi	cr1,rBITDIF,8
+	cmplwi	cr7,rN,16
+	clrlwi	rN,rN,30
+	beq	L(dPs4)
+	mtctr	rTMP
+	bgt	cr1,L(dPs3)
+	beq	cr1,L(dPs2)
+
+/* Remainder is 4 */
+	.align	3
+L(dsP1):
+	slw	rWORD5,rWORD1,r11
+	slw	rWORD6,rWORD2,r11
+	cmplw	cr5,rWORD5,rWORD6
+	blt	cr7,L(dP1x)
+/* Do something useful in this cycle since we have to branch anyway.  */
+	lwz	rWORD1,4(rSTR1)
+	lwz	rWORD2,4(rSTR2)
+	cmplw	cr0,rWORD1,rWORD2
+	b	L(dP1e)
+/* Remainder is 8 */
+	.align	4
+L(dPs2):
+	slw	rWORD5,rWORD1,r11
+	slw	rWORD6,rWORD2,r11
+	cmplw	cr6,rWORD5,rWORD6
+	blt	cr7,L(dP2x)
+/* Do something useful in this cycle since we have to branch anyway.  */
+	lwz	rWORD7,4(rSTR1)
+	lwz	rWORD8,4(rSTR2)
+	cmplw	cr5,rWORD7,rWORD8
+	b	L(dP2e)
+/* Remainder is 12 */
+	.align	4
+L(dPs3):
+	slw	rWORD3,rWORD1,r11
+	slw	rWORD4,rWORD2,r11
+	cmplw	cr1,rWORD3,rWORD4
+	b	L(dP3e)
+/* Count is a multiple of 16, remainder is 0 */
+	.align	4
+L(dPs4):
+	mtctr	rTMP
+	slw	rWORD1,rWORD1,r11
+	slw	rWORD2,rWORD2,r11
+	cmplw	cr0,rWORD1,rWORD2
+	b	L(dP4e)
+
+/* At this point we know both strings are word aligned and the
+   compare length is at least 8 bytes.  */
+	.align	4
+L(Waligned):
+	andi.	rBITDIF,rN,12	/* Get the word remainder */
+	srwi	rTMP,rN,4	/* Divide by 16 */
+	cmplwi	cr1,rBITDIF,8
+	cmplwi	cr7,rN,16
+	clrlwi	rN,rN,30
+	beq	L(dP4)
+	bgt	cr1,L(dP3)
+	beq	cr1,L(dP2)
+
+/* Remainder is 4 */
+	.align	4
+L(dP1):
+	mtctr	rTMP
+/* Normally we'd use rWORD7/rWORD8 here, but since we might exit early
+   (8-15 byte compare), we want to use only volatile registers.  This
+   means we can avoid restoring non-volatile registers since we did not
+   change any on the early exit path.  The key here is the non-early
+   exit path only cares about the condition code (cr5), not about which
+   register pair was used.  */
+	lwz	rWORD5,0(rSTR1)
+	lwz	rWORD6,0(rSTR2)
+	cmplw	cr5,rWORD5,rWORD6
+	blt	cr7,L(dP1x)
+	lwz	rWORD1,4(rSTR1)
+	lwz	rWORD2,4(rSTR2)
+	cmplw	cr0,rWORD1,rWORD2
+L(dP1e):
+	lwz	rWORD3,8(rSTR1)
+	lwz	rWORD4,8(rSTR2)
+	cmplw	cr1,rWORD3,rWORD4
+	lwz	rWORD5,12(rSTR1)
+	lwz	rWORD6,12(rSTR2)
+	cmplw	cr6,rWORD5,rWORD6
+	bne	cr5,L(dLcr5)
+	bne	cr0,L(dLcr0)
+
+	lwzu	rWORD7,16(rSTR1)
+	lwzu	rWORD8,16(rSTR2)
+	bne	cr1,L(dLcr1)
+	cmplw	cr5,rWORD7,rWORD8
+	bdnz	L(dLoop)
+	bne	cr6,L(dLcr6)
+	lwz	r30,44(1)
+	lwz	r31,48(1)
+	.align	3
+L(dP1x):
+	slwi.	r12,rN,3
+	bne	cr5,L(dLcr5)
+	subfic	rN,r12,32	/* Shift count is 32 - (rN * 8).  */
+	lwz	1,0(1)
+	bne	L(d00)
+	li	rRTN,0
+	blr
+
+/* Remainder is 8 */
+	.align	4
+L(dP2):
+	mtctr	rTMP
+	lwz	rWORD5,0(rSTR1)
+	lwz	rWORD6,0(rSTR2)
+	cmplw	cr6,rWORD5,rWORD6
+	blt	cr7,L(dP2x)
+	lwz	rWORD7,4(rSTR1)
+	lwz	rWORD8,4(rSTR2)
+	cmplw	cr5,rWORD7,rWORD8
+L(dP2e):
+	lwz	rWORD1,8(rSTR1)
+	lwz	rWORD2,8(rSTR2)
+	cmplw	cr0,rWORD1,rWORD2
+	lwz	rWORD3,12(rSTR1)
+	lwz	rWORD4,12(rSTR2)
+	cmplw	cr1,rWORD3,rWORD4
+	addi	rSTR1,rSTR1,4
+	addi	rSTR2,rSTR2,4
+	bne	cr6,L(dLcr6)
+	bne	cr5,L(dLcr5)
+	b	L(dLoop2)
+/* Again we are on a early exit path (16-23 byte compare), we want to
+   only use volatile registers and avoid restoring non-volatile
+   registers.  */
+	.align	4
+L(dP2x):
+	lwz	rWORD3,4(rSTR1)
+	lwz	rWORD4,4(rSTR2)
+	cmplw	cr5,rWORD3,rWORD4
+	slwi.	r12,rN,3
+	bne	cr6,L(dLcr6)
+	addi	rSTR1,rSTR1,4
+	addi	rSTR2,rSTR2,4
+	bne	cr5,L(dLcr5)
+	subfic	rN,r12,32	/* Shift count is 32 - (rN * 8).  */
+	lwz	1,0(1)
+	bne	L(d00)
+	li	rRTN,0
+	blr
+
+/* Remainder is 12 */
+	.align	4
+L(dP3):
+	mtctr	rTMP
+	lwz	rWORD3,0(rSTR1)
+	lwz	rWORD4,0(rSTR2)
+	cmplw	cr1,rWORD3,rWORD4
+L(dP3e):
+	lwz	rWORD5,4(rSTR1)
+	lwz	rWORD6,4(rSTR2)
+	cmplw	cr6,rWORD5,rWORD6
+	blt	cr7,L(dP3x)
+	lwz	rWORD7,8(rSTR1)
+	lwz	rWORD8,8(rSTR2)
+	cmplw	cr5,rWORD7,rWORD8
+	lwz	rWORD1,12(rSTR1)
+	lwz	rWORD2,12(rSTR2)
+	cmplw	cr0,rWORD1,rWORD2
+	addi	rSTR1,rSTR1,8
+	addi	rSTR2,rSTR2,8
+	bne	cr1,L(dLcr1)
+	bne	cr6,L(dLcr6)
+	b	L(dLoop1)
+/* Again we are on a early exit path (24-31 byte compare), we want to
+   only use volatile registers and avoid restoring non-volatile
+   registers.  */
+	.align	4
+L(dP3x):
+	lwz	rWORD1,8(rSTR1)
+	lwz	rWORD2,8(rSTR2)
+	cmplw	cr5,rWORD1,rWORD2
+	slwi.	r12,rN,3
+	bne	cr1,L(dLcr1)
+	addi	rSTR1,rSTR1,8
+	addi	rSTR2,rSTR2,8
+	bne	cr6,L(dLcr6)
+	subfic	rN,r12,32	/* Shift count is 32 - (rN * 8).  */
+	bne	cr5,L(dLcr5)
+	lwz	1,0(1)
+	bne	L(d00)
+	li	rRTN,0
+	blr
+
+/* Count is a multiple of 16, remainder is 0 */
+	.align	4
+L(dP4):
+	mtctr	rTMP
+	lwz	rWORD1,0(rSTR1)
+	lwz	rWORD2,0(rSTR2)
+	cmplw	cr0,rWORD1,rWORD2
+L(dP4e):
+	lwz	rWORD3,4(rSTR1)
+	lwz	rWORD4,4(rSTR2)
+	cmplw	cr1,rWORD3,rWORD4
+	lwz	rWORD5,8(rSTR1)
+	lwz	rWORD6,8(rSTR2)
+	cmplw	cr6,rWORD5,rWORD6
+	lwzu	rWORD7,12(rSTR1)
+	lwzu	rWORD8,12(rSTR2)
+	cmplw	cr5,rWORD7,rWORD8
+	bne	cr0,L(dLcr0)
+	bne	cr1,L(dLcr1)
+	bdz-	L(d24)		/* Adjust CTR as we start with +4 */
+/* This is the primary loop */
+	.align	4
+L(dLoop):
+	lwz	rWORD1,4(rSTR1)
+	lwz	rWORD2,4(rSTR2)
+	cmplw	cr1,rWORD3,rWORD4
+	bne	cr6,L(dLcr6)
+L(dLoop1):
+	lwz	rWORD3,8(rSTR1)
+	lwz	rWORD4,8(rSTR2)
+	cmplw	cr6,rWORD5,rWORD6
+	bne	cr5,L(dLcr5)
+L(dLoop2):
+	lwz	rWORD5,12(rSTR1)
+	lwz	rWORD6,12(rSTR2)
+	cmplw	cr5,rWORD7,rWORD8
+	bne	cr0,L(dLcr0)
+L(dLoop3):
+	lwzu	rWORD7,16(rSTR1)
+	lwzu	rWORD8,16(rSTR2)
+	bne	cr1,L(dLcr1)
+	cmplw	cr0,rWORD1,rWORD2
+	bdnz	L(dLoop)
+
+L(dL4):
+	cmplw	cr1,rWORD3,rWORD4
+	bne	cr6,L(dLcr6)
+	cmplw	cr6,rWORD5,rWORD6
+	bne	cr5,L(dLcr5)
+	cmplw	cr5,rWORD7,rWORD8
+L(d44):
+	bne	cr0,L(dLcr0)
+L(d34):
+	bne	cr1,L(dLcr1)
+L(d24):
+	bne	cr6,L(dLcr6)
+L(d14):
+	slwi.	r12,rN,3
+	bne	cr5,L(dLcr5)
+L(d04):
+	lwz	r30,44(1)
+	lwz	r31,48(1)
+	lwz	1,0(1)
+	subfic	rN,r12,32	/* Shift count is 32 - (rN * 8).  */
+	beq	L(zeroLength)
+/* At this point we have a remainder of 1 to 3 bytes to compare.  Since
+   we are aligned it is safe to load the whole word, and use
+   shift right to eliminate bits beyond the compare length. */
+L(d00):
+	lwz	rWORD1,4(rSTR1)
+	lwz	rWORD2,4(rSTR2)
+	srw	rWORD1,rWORD1,rN
+	srw	rWORD2,rWORD2,rN
+	cmplw	rWORD1,rWORD2
+	li	rRTN,0
+	beqlr
+	li	rRTN,1
+	bgtlr
+	li	rRTN,-1
+	blr
+
+	.align	4
+L(dLcr0):
+	lwz	r30,44(1)
+	lwz	r31,48(1)
+	li	rRTN,1
+	lwz	1,0(1)
+	bgtlr	cr0
+	li	rRTN,-1
+	blr
+	.align	4
+L(dLcr1):
+	lwz	r30,44(1)
+	lwz	r31,48(1)
+	li	rRTN,1
+	lwz	1,0(1)
+	bgtlr	cr1
+	li	rRTN,-1
+	blr
+	.align	4
+L(dLcr6):
+	lwz	r30,44(1)
+	lwz	r31,48(1)
+	li	rRTN,1
+	lwz	1,0(1)
+	bgtlr	cr6
+	li	rRTN,-1
+	blr
+	.align	4
+L(dLcr5):
+	lwz	r30,44(1)
+	lwz	r31,48(1)
+L(dLcr5x):
+	li	rRTN,1
+	lwz	1,0(1)
+	bgtlr	cr5
+	li	rRTN,-1
+	blr
+
+	.align	4
+L(bytealigned):
+	cfi_adjust_cfa_offset(-64)
+	mtctr	rN
+
+/* We need to prime this loop.  This loop is swing modulo scheduled
+   to avoid pipe delays.  The dependent instruction latencies (load to
+   compare to conditional branch) is 2 to 3 cycles.  In this loop each
+   dispatch group ends in a branch and takes 1 cycle.  Effectively
+   the first iteration of the loop only serves to load operands and
+   branches based on compares are delayed until the next loop.
+
+   So we must precondition some registers and condition codes so that
+   we don't exit the loop early on the first iteration.  */
+	lbz	rWORD1,0(rSTR1)
+	lbz	rWORD2,0(rSTR2)
+	bdz	L(b11)
+	cmplw	cr0,rWORD1,rWORD2
+	lbz	rWORD3,1(rSTR1)
+	lbz	rWORD4,1(rSTR2)
+	bdz	L(b12)
+	cmplw	cr1,rWORD3,rWORD4
+	lbzu	rWORD5,2(rSTR1)
+	lbzu	rWORD6,2(rSTR2)
+	bdz	L(b13)
+	.align	4
+L(bLoop):
+	lbzu	rWORD1,1(rSTR1)
+	lbzu	rWORD2,1(rSTR2)
+	bne	cr0,L(bLcr0)
+
+	cmplw	cr6,rWORD5,rWORD6
+	bdz	L(b3i)
+
+	lbzu	rWORD3,1(rSTR1)
+	lbzu	rWORD4,1(rSTR2)
+	bne	cr1,L(bLcr1)
+
+	cmplw	cr0,rWORD1,rWORD2
+	bdz	L(b2i)
+
+	lbzu	rWORD5,1(rSTR1)
+	lbzu	rWORD6,1(rSTR2)
+	bne	cr6,L(bLcr6)
+
+	cmplw	cr1,rWORD3,rWORD4
+	bdnz	L(bLoop)
+
+/* We speculatively loading bytes before we have tested the previous
+   bytes.  But we must avoid overrunning the length (in the ctr) to
+   prevent these speculative loads from causing a segfault.  In this
+   case the loop will exit early (before the all pending bytes are
+   tested.  In this case we must complete the pending operations
+   before returning.  */
+L(b1i):
+	bne	cr0,L(bLcr0)
+	bne	cr1,L(bLcr1)
+	b	L(bx56)
+	.align	4
+L(b2i):
+	bne	cr6,L(bLcr6)
+	bne	cr0,L(bLcr0)
+	b	L(bx34)
+	.align	4
+L(b3i):
+	bne	cr1,L(bLcr1)
+	bne	cr6,L(bLcr6)
+	b	L(bx12)
+	.align	4
+L(bLcr0):
+	li	rRTN,1
+	bgtlr	cr0
+	li	rRTN,-1
+	blr
+L(bLcr1):
+	li	rRTN,1
+	bgtlr	cr1
+	li	rRTN,-1
+	blr
+L(bLcr6):
+	li	rRTN,1
+	bgtlr	cr6
+	li	rRTN,-1
+	blr
+
+L(b13):
+	bne	cr0,L(bx12)
+	bne	cr1,L(bx34)
+L(bx56):
+	sub	rRTN,rWORD5,rWORD6
+	blr
+	nop
+L(b12):
+	bne	cr0,L(bx12)
+L(bx34):
+	sub	rRTN,rWORD3,rWORD4
+	blr
+
+L(b11):
+L(bx12):
+	sub	rRTN,rWORD1,rWORD2
+	blr
+
+	.align	4
+L(zeroLengthReturn):
+
+L(zeroLength):
+	li	rRTN,0
+	blr
+
+	cfi_adjust_cfa_offset(64)
+	.align	4
+/* At this point we know the strings have different alignment and the
+   compare length is at least 8 bytes.  rBITDIF contains the low order
+   2 bits of rSTR1 and cr5 contains the result of the logical compare
+   of rBITDIF to 0.  If rBITDIF == 0 then rStr1 is word aligned and can
+   perform the Wunaligned loop.
+
+   Otherwise we know that rSTR1 is not aready word aligned yet.
+   So we can force the string addresses to the next lower word
+   boundary and special case this first word using shift left to
+   eliminate bits preceeding the first byte.  Since we want to join the
+   normal (Wualigned) compare loop, starting at the second word,
+   we need to adjust the length (rN) and special case the loop
+   versioning for the first W. This insures that the loop count is
+   correct and the first W (shifted) is in the expected resister pair.  */
+#define rSHL		r29	/* Unaligned shift left count.  */
+#define rSHR		r28	/* Unaligned shift right count.  */
+#define rB		r27	/* Left rotation temp for rWORD2.  */
+#define rD		r26	/* Left rotation temp for rWORD4.  */
+#define rF		r25	/* Left rotation temp for rWORD6.  */
+#define rH		r24	/* Left rotation temp for rWORD8.  */
+#define rA		r0	/* Right rotation temp for rWORD2.  */
+#define rC		r12	/* Right rotation temp for rWORD4.  */
+#define rE		r0	/* Right rotation temp for rWORD6.  */
+#define rG		r12	/* Right rotation temp for rWORD8.  */
+L(unaligned):
+	stw	r29,40(r1)
+	cfi_offset(r29,(40-64))
+	clrlwi	rSHL,rSTR2,30
+	stw	r28,36(r1)
+	cfi_offset(r28,(36-64))
+	beq	cr5,L(Wunaligned)
+	stw	r27,32(r1)
+	cfi_offset(r27,(32-64))
+/* Adjust the logical start of rSTR2 to compensate for the extra bits
+   in the 1st rSTR1 W.  */
+	sub	r27,rSTR2,rBITDIF
+/* But do not attempt to address the W before that W that contains
+   the actual start of rSTR2.  */
+	clrrwi	rSTR2,rSTR2,2
+	stw	r26,28(r1)
+	cfi_offset(r26,(28-64))
+/* Compute the left/right shift counts for the unalign rSTR2,
+   compensating for the logical (W aligned) start of rSTR1.  */
+	clrlwi	rSHL,r27,30
+	clrrwi	rSTR1,rSTR1,2
+	stw	r25,24(r1)
+	cfi_offset(r25,(24-64))
+	slwi	rSHL,rSHL,3
+	cmplw	cr5,r27,rSTR2
+	add	rN,rN,rBITDIF
+	slwi	r11,rBITDIF,3
+	stw	r24,20(r1)
+	cfi_offset(r24,(20-64))
+	subfic	rSHR,rSHL,32
+	srwi	rTMP,rN,4	/* Divide by 16 */
+	andi.	rBITDIF,rN,12	/* Get the W remainder */
+/* We normally need to load 2 Ws to start the unaligned rSTR2, but in
+   this special case those bits may be discarded anyway.  Also we
+   must avoid loading a W where none of the bits are part of rSTR2 as
+   this may cross a page boundary and cause a page fault.  */
+	li	rWORD8,0
+	blt	cr5,L(dus0)
+	lwz	rWORD8,0(rSTR2)
+	la	rSTR2,4(rSTR2)
+	slw	rWORD8,rWORD8,rSHL
+
+L(dus0):
+	lwz	rWORD1,0(rSTR1)
+	lwz	rWORD2,0(rSTR2)
+	cmplwi	cr1,rBITDIF,8
+	cmplwi	cr7,rN,16
+	srw	rG,rWORD2,rSHR
+	clrlwi	rN,rN,30
+	beq	L(duPs4)
+	mtctr	rTMP
+	or	rWORD8,rG,rWORD8
+	bgt	cr1,L(duPs3)
+	beq	cr1,L(duPs2)
+
+/* Remainder is 4 */
+	.align	4
+L(dusP1):
+	slw	rB,rWORD2,rSHL
+	slw	rWORD7,rWORD1,r11
+	slw	rWORD8,rWORD8,r11
+	bge	cr7,L(duP1e)
+/* At this point we exit early with the first word compare
+   complete and remainder of 0 to 3 bytes.  See L(du14) for details on
+   how we handle the remaining bytes.  */
+	cmplw	cr5,rWORD7,rWORD8
+	slwi.	rN,rN,3
+	bne	cr5,L(duLcr5)
+	cmplw	cr7,rN,rSHR
+	beq	L(duZeroReturn)
+	li	rA,0
+	ble	cr7,L(dutrim)
+	lwz	rWORD2,4(rSTR2)
+	srw	rA,rWORD2,rSHR
+	b	L(dutrim)
+/* Remainder is 8 */
+	.align	4
+L(duPs2):
+	slw	rH,rWORD2,rSHL
+	slw	rWORD5,rWORD1,r11
+	slw	rWORD6,rWORD8,r11
+	b	L(duP2e)
+/* Remainder is 12 */
+	.align	4
+L(duPs3):
+	slw	rF,rWORD2,rSHL
+	slw	rWORD3,rWORD1,r11
+	slw	rWORD4,rWORD8,r11
+	b	L(duP3e)
+/* Count is a multiple of 16, remainder is 0 */
+	.align	4
+L(duPs4):
+	mtctr	rTMP
+	or	rWORD8,rG,rWORD8
+	slw	rD,rWORD2,rSHL
+	slw	rWORD1,rWORD1,r11
+	slw	rWORD2,rWORD8,r11
+	b	L(duP4e)
+
+/* At this point we know rSTR1 is word aligned and the
+   compare length is at least 8 bytes.  */
+	.align	4
+L(Wunaligned):
+	stw	r27,32(r1)
+	cfi_offset(r27,(32-64))
+	clrrwi	rSTR2,rSTR2,2
+	stw	r26,28(r1)
+	cfi_offset(r26,(28-64))
+	srwi	rTMP,rN,4	/* Divide by 16 */
+	stw	r25,24(r1)
+	cfi_offset(r25,(24-64))
+	andi.	rBITDIF,rN,12	/* Get the W remainder */
+	stw	r24,20(r1)
+	cfi_offset(r24,(24-64))
+	slwi	rSHL,rSHL,3
+	lwz	rWORD6,0(rSTR2)
+	lwzu	rWORD8,4(rSTR2)
+	cmplwi	cr1,rBITDIF,8
+	cmplwi	cr7,rN,16
+	clrlwi	rN,rN,30
+	subfic	rSHR,rSHL,32
+	slw	rH,rWORD6,rSHL
+	beq	L(duP4)
+	mtctr	rTMP
+	bgt	cr1,L(duP3)
+	beq	cr1,L(duP2)
+
+/* Remainder is 4 */
+	.align	4
+L(duP1):
+	srw	rG,rWORD8,rSHR
+	lwz	rWORD7,0(rSTR1)
+	slw	rB,rWORD8,rSHL
+	or	rWORD8,rG,rH
+	blt	cr7,L(duP1x)
+L(duP1e):
+	lwz	rWORD1,4(rSTR1)
+	lwz	rWORD2,4(rSTR2)
+	cmplw	cr5,rWORD7,rWORD8
+	srw	rA,rWORD2,rSHR
+	slw	rD,rWORD2,rSHL
+	or	rWORD2,rA,rB
+	lwz	rWORD3,8(rSTR1)
+	lwz	rWORD4,8(rSTR2)
+	cmplw	cr0,rWORD1,rWORD2
+	srw	rC,rWORD4,rSHR
+	slw	rF,rWORD4,rSHL
+	bne	cr5,L(duLcr5)
+	or	rWORD4,rC,rD
+	lwz	rWORD5,12(rSTR1)
+	lwz	rWORD6,12(rSTR2)
+	cmplw	cr1,rWORD3,rWORD4
+	srw	rE,rWORD6,rSHR
+	slw	rH,rWORD6,rSHL
+	bne	cr0,L(duLcr0)
+	or	rWORD6,rE,rF
+	cmplw	cr6,rWORD5,rWORD6
+	b	L(duLoop3)
+	.align	4
+/* At this point we exit early with the first word compare
+   complete and remainder of 0 to 3 bytes.  See L(du14) for details on
+   how we handle the remaining bytes.  */
+L(duP1x):
+	cmplw	cr5,rWORD7,rWORD8
+	slwi.	rN,rN,3
+	bne	cr5,L(duLcr5)
+	cmplw	cr7,rN,rSHR
+	beq	L(duZeroReturn)
+	li	rA,0
+	ble	cr7,L(dutrim)
+	ld	rWORD2,8(rSTR2)
+	srw	rA,rWORD2,rSHR
+	b	L(dutrim)
+/* Remainder is 8 */
+	.align	4
+L(duP2):
+	srw	rE,rWORD8,rSHR
+	lwz	rWORD5,0(rSTR1)
+	or	rWORD6,rE,rH
+	slw	rH,rWORD8,rSHL
+L(duP2e):
+	lwz	rWORD7,4(rSTR1)
+	lwz	rWORD8,4(rSTR2)
+	cmplw	cr6,rWORD5,rWORD6
+	srw	rG,rWORD8,rSHR
+	slw	rB,rWORD8,rSHL
+	or	rWORD8,rG,rH
+	blt	cr7,L(duP2x)
+	lwz	rWORD1,8(rSTR1)
+	lwz	rWORD2,8(rSTR2)
+	cmplw	cr5,rWORD7,rWORD8
+	bne	cr6,L(duLcr6)
+	srw	rA,rWORD2,rSHR
+	slw	rD,rWORD2,rSHL
+	or	rWORD2,rA,rB
+	lwz	rWORD3,12(rSTR1)
+	lwz	rWORD4,12(rSTR2)
+	cmplw	cr0,rWORD1,rWORD2
+	bne	cr5,L(duLcr5)
+	srw	rC,rWORD4,rSHR
+	slw	rF,rWORD4,rSHL
+	or	rWORD4,rC,rD
+	addi	rSTR1,rSTR1,4
+	addi	rSTR2,rSTR2,4
+	cmplw	cr1,rWORD3,rWORD4
+	b	L(duLoop2)
+	.align	4
+L(duP2x):
+	cmplw	cr5,rWORD7,rWORD8
+	addi	rSTR1,rSTR1,4
+	addi	rSTR2,rSTR2,4
+	bne	cr6,L(duLcr6)
+	slwi.	rN,rN,3
+	bne	cr5,L(duLcr5)
+	cmplw	cr7,rN,rSHR
+	beq	L(duZeroReturn)
+	li	rA,0
+	ble	cr7,L(dutrim)
+	lwz	rWORD2,4(rSTR2)
+	srw	rA,rWORD2,rSHR
+	b	L(dutrim)
+
+/* Remainder is 12 */
+	.align	4
+L(duP3):
+	srw	rC,rWORD8,rSHR
+	lwz	rWORD3,0(rSTR1)
+	slw	rF,rWORD8,rSHL
+	or	rWORD4,rC,rH
+L(duP3e):
+	lwz	rWORD5,4(rSTR1)
+	lwz	rWORD6,4(rSTR2)
+	cmplw	cr1,rWORD3,rWORD4
+	srw	rE,rWORD6,rSHR
+	slw	rH,rWORD6,rSHL
+	or	rWORD6,rE,rF
+	lwz	rWORD7,8(rSTR1)
+	lwz	rWORD8,8(rSTR2)
+	cmplw	cr6,rWORD5,rWORD6
+	bne	cr1,L(duLcr1)
+	srw	rG,rWORD8,rSHR
+	slw	rB,rWORD8,rSHL
+	or	rWORD8,rG,rH
+	blt	cr7,L(duP3x)
+	lwz	rWORD1,12(rSTR1)
+	lwz	rWORD2,12(rSTR2)
+	cmplw	cr5,rWORD7,rWORD8
+	bne	cr6,L(duLcr6)
+	srw	rA,rWORD2,rSHR
+	slw	rD,rWORD2,rSHL
+	or	rWORD2,rA,rB
+	addi	rSTR1,rSTR1,8
+	addi	rSTR2,rSTR2,8
+	cmplw	cr0,rWORD1,rWORD2
+	b	L(duLoop1)
+	.align	4
+L(duP3x):
+	addi	rSTR1,rSTR1,8
+	addi	rSTR2,rSTR2,8
+	bne	cr1,L(duLcr1)
+	cmplw	cr5,rWORD7,rWORD8
+	bne	cr6,L(duLcr6)
+	slwi.	rN,rN,3
+	bne	cr5,L(duLcr5)
+	cmplw	cr7,rN,rSHR
+	beq	L(duZeroReturn)
+	li	rA,0
+	ble	cr7,L(dutrim)
+	lwz	rWORD2,4(rSTR2)
+	srw	rA,rWORD2,rSHR
+	b	L(dutrim)
+
+/* Count is a multiple of 16, remainder is 0 */
+	.align	4
+L(duP4):
+	mtctr	rTMP
+	srw	rA,rWORD8,rSHR
+	lwz	rWORD1,0(rSTR1)
+	slw	rD,rWORD8,rSHL
+	or	rWORD2,rA,rH
+L(duP4e):
+	lwz	rWORD3,4(rSTR1)
+	lwz	rWORD4,4(rSTR2)
+	cmplw	cr0,rWORD1,rWORD2
+	srw	rC,rWORD4,rSHR
+	slw	rF,rWORD4,rSHL
+	or	rWORD4,rC,rD
+	lwz	rWORD5,8(rSTR1)
+	lwz	rWORD6,8(rSTR2)
+	cmplw	cr1,rWORD3,rWORD4
+	bne	cr0,L(duLcr0)
+	srw	rE,rWORD6,rSHR
+	slw	rH,rWORD6,rSHL
+	or	rWORD6,rE,rF
+	lwzu	rWORD7,12(rSTR1)
+	lwzu	rWORD8,12(rSTR2)
+	cmplw	cr6,rWORD5,rWORD6
+	bne	cr1,L(duLcr1)
+	srw	rG,rWORD8,rSHR
+	slw	rB,rWORD8,rSHL
+	or	rWORD8,rG,rH
+	cmplw	cr5,rWORD7,rWORD8
+	bdz	L(du24)		/* Adjust CTR as we start with +4 */
+/* This is the primary loop */
+	.align	4
+L(duLoop):
+	lwz	rWORD1,4(rSTR1)
+	lwz	rWORD2,4(rSTR2)
+	cmplw	cr1,rWORD3,rWORD4
+	bne	cr6,L(duLcr6)
+	srw	rA,rWORD2,rSHR
+	slw	rD,rWORD2,rSHL
+	or	rWORD2,rA,rB
+L(duLoop1):
+	lwz	rWORD3,8(rSTR1)
+	lwz	rWORD4,8(rSTR2)
+	cmplw	cr6,rWORD5,rWORD6
+	bne	cr5,L(duLcr5)
+	srw	rC,rWORD4,rSHR
+	slw	rF,rWORD4,rSHL
+	or	rWORD4,rC,rD
+L(duLoop2):
+	lwz	rWORD5,12(rSTR1)
+	lwz	rWORD6,12(rSTR2)
+	cmplw	cr5,rWORD7,rWORD8
+	bne	cr0,L(duLcr0)
+	srw	rE,rWORD6,rSHR
+	slw	rH,rWORD6,rSHL
+	or	rWORD6,rE,rF
+L(duLoop3):
+	lwzu	rWORD7,16(rSTR1)
+	lwzu	rWORD8,16(rSTR2)
+	cmplw	cr0,rWORD1,rWORD2
+	bne	cr1,L(duLcr1)
+	srw	rG,rWORD8,rSHR
+	slw	rB,rWORD8,rSHL
+	or	rWORD8,rG,rH
+	bdnz	L(duLoop)
+
+L(duL4):
+	bne	cr1,L(duLcr1)
+	cmplw	cr1,rWORD3,rWORD4
+	bne	cr6,L(duLcr6)
+	cmplw	cr6,rWORD5,rWORD6
+	bne	cr5,L(duLcr5)
+	cmplw	cr5,rWORD7,rWORD8
+L(du44):
+	bne	cr0,L(duLcr0)
+L(du34):
+	bne	cr1,L(duLcr1)
+L(du24):
+	bne	cr6,L(duLcr6)
+L(du14):
+	slwi.	rN,rN,3
+	bne	cr5,L(duLcr5)
+/* At this point we have a remainder of 1 to 3 bytes to compare.  We use
+   shift right to eliminate bits beyond the compare length.
+
+   However it may not be safe to load rWORD2 which may be beyond the
+   string length. So we compare the bit length of the remainder to
+   the right shift count (rSHR). If the bit count is less than or equal
+   we do not need to load rWORD2 (all significant bits are already in
+   rB).  */
+	cmplw	cr7,rN,rSHR
+	beq	L(duZeroReturn)
+	li	rA,0
+	ble	cr7,L(dutrim)
+	lwz	rWORD2,4(rSTR2)
+	srw	rA,rWORD2,rSHR
+	.align	4
+L(dutrim):
+	lwz	rWORD1,4(rSTR1)
+	lwz	r31,48(1)
+	subfic	rN,rN,32	/* Shift count is 32 - (rN * 8).  */
+	or	rWORD2,rA,rB
+	lwz	r30,44(1)
+	lwz	r29,40(r1)
+	srw	rWORD1,rWORD1,rN
+	srw	rWORD2,rWORD2,rN
+	lwz	r28,36(r1)
+	lwz	r27,32(r1)
+	cmplw	rWORD1,rWORD2
+	li	rRTN,0
+	beq	L(dureturn26)
+	li	rRTN,1
+	bgt	L(dureturn26)
+	li	rRTN,-1
+	b	L(dureturn26)
+	.align	4
+L(duLcr0):
+	lwz	r31,48(1)
+	lwz	r30,44(1)
+	li	rRTN,1
+	bgt	cr0,L(dureturn29)
+	lwz	r29,40(r1)
+	lwz	r28,36(r1)
+	li	rRTN,-1
+	b	L(dureturn27)
+	.align	4
+L(duLcr1):
+	lwz	r31,48(1)
+	lwz	r30,44(1)
+	li	rRTN,1
+	bgt	cr1,L(dureturn29)
+	lwz	r29,40(r1)
+	lwz	r28,36(r1)
+	li	rRTN,-1
+	b	L(dureturn27)
+	.align	4
+L(duLcr6):
+	lwz	r31,48(1)
+	lwz	r30,44(1)
+	li	rRTN,1
+	bgt	cr6,L(dureturn29)
+	lwz	r29,40(r1)
+	lwz	r28,36(r1)
+	li	rRTN,-1
+	b	L(dureturn27)
+	.align	4
+L(duLcr5):
+	lwz	r31,48(1)
+	lwz	r30,44(1)
+	li	rRTN,1
+	bgt	cr5,L(dureturn29)
+	lwz	r29,40(r1)
+	lwz	r28,36(r1)
+	li	rRTN,-1
+	b	L(dureturn27)
+	.align	3
+L(duZeroReturn):
+	li	rRTN,0
+	.align	4
+L(dureturn):
+	lwz	r31,48(1)
+	lwz	r30,44(1)
+L(dureturn29):
+	lwz	r29,40(r1)
+	lwz	r28,36(r1)
+L(dureturn27):
+	lwz	r27,32(r1)
+L(dureturn26):
+	lwz	r26,28(r1)
+L(dureturn25):
+	lwz	r25,24(r1)
+	lwz	r24,20(r1)
+	lwz	1,0(1)
+	blr
+END (BP_SYM (memcmp))
+libc_hidden_builtin_def (memcmp)
+weak_alias (memcmp,bcmp)