116 lines
2.6 KiB
ArmAsm
116 lines
2.6 KiB
ArmAsm
/* Optimized strcmp implementation for PowerPC.
|
|
Copyright (C) 1997, 1999 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 Library General Public License as
|
|
published by the Free Software Foundation; either version 2 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
|
|
Library General Public License for more details.
|
|
|
|
You should have received a copy of the GNU Library General Public
|
|
License along with the GNU C Library; see the file COPYING.LIB. If not,
|
|
write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
|
|
Boston, MA 02111-1307, USA. */
|
|
|
|
#include <sysdep.h>
|
|
|
|
/* See strlen.s for comments on how the end-of-string testing works. */
|
|
|
|
EALIGN(strcmp,4,0)
|
|
/* int [r3] strcmp (const char *p1 [r3], const char *p2 [r4]) */
|
|
|
|
/* General register assignments:
|
|
r0: temporary
|
|
r3: pointer to previous word in s1
|
|
r4: pointer to previous word in s2
|
|
r5: current word from s1
|
|
r6: current word from s2
|
|
r7: 0xfefefeff
|
|
r8: 0x7f7f7f7f
|
|
r9: ~(word in s1 | 0x7f7f7f7f) */
|
|
|
|
/* Register assignments in the prologue:
|
|
r10: low 2 bits of p2-p1
|
|
r11: mask to orc with r5/r6 */
|
|
|
|
or r0,r4,r3
|
|
clrlwi. r0,r0,30
|
|
lis r7,0xfeff
|
|
bne L(unaligned)
|
|
|
|
lwz r5,0(r3)
|
|
lwz r6,0(r4)
|
|
lis r8,0x7f7f
|
|
addi r7,r7,-0x101
|
|
addi r8,r8,0x7f7f
|
|
b 1f
|
|
|
|
0: lwzu r5,4(r3)
|
|
bne cr1,L(different)
|
|
lwzu r6,4(r4)
|
|
1: add r0,r7,r5
|
|
nor r9,r8,r5
|
|
and. r0,r0,r9
|
|
cmpw cr1,r5,r6
|
|
beq+ 0b
|
|
L(endstring):
|
|
/* OK. We've hit the end of the string. We need to be careful that
|
|
we don't compare two strings as different because of gunk beyond
|
|
the end of the strings... */
|
|
and r0,r8,r5
|
|
beq cr1,L(equal)
|
|
add r0,r0,r8
|
|
xor. r10,r5,r6
|
|
andc r9,r9,r0
|
|
blt- L(highbit)
|
|
cntlzw r10,r10
|
|
cntlzw r9,r9
|
|
addi r9,r9,7
|
|
cmpw cr1,r9,r10
|
|
sub r3,r5,r6
|
|
bgelr+ cr1
|
|
L(equal):
|
|
li r3,0
|
|
blr
|
|
|
|
L(different):
|
|
lwz r5,-4(r3)
|
|
xor. r10,r5,r6
|
|
sub r3,r5,r6
|
|
bgelr+
|
|
L(highbit):
|
|
ori r3,r6,1
|
|
blr
|
|
|
|
|
|
/* Oh well. In this case, we just do a byte-by-byte comparison. */
|
|
.align 4
|
|
L(unaligned):
|
|
lbz r5,0(r3)
|
|
lbz r6,0(r4)
|
|
b 1f
|
|
|
|
0: lbzu r5,1(r3)
|
|
bne- 4f
|
|
lbzu r6,1(r4)
|
|
1: cmpwi cr1,r5,0
|
|
beq- cr1,3f
|
|
cmpw r5,r6
|
|
bne- 3f
|
|
lbzu r5,1(r3)
|
|
lbzu r6,1(r4)
|
|
cmpwi cr1,r5,0
|
|
cmpw r5,r6
|
|
bne+ cr1,0b
|
|
3: sub r3,r5,r6
|
|
blr
|
|
4: lbz r5,-1(r3)
|
|
sub r3,r5,r6
|
|
blr
|
|
END(strcmp)
|