mirror of
https://sourceware.org/git/glibc.git
synced 2024-11-26 06:50:07 +00:00
5a82c74822
Also, change sources.redhat.com to sourceware.org. This patch was automatically generated by running the following shell script, which uses GNU sed, and which avoids modifying files imported from upstream: sed -ri ' s,(http|ftp)(://(.*\.)?(gnu|fsf|sourceware)\.org($|[^.]|\.[^a-z])),https\2,g s,(http|ftp)(://(.*\.)?)sources\.redhat\.com($|[^.]|\.[^a-z]),https\2sourceware.org\4,g ' \ $(find $(git ls-files) -prune -type f \ ! -name '*.po' \ ! -name 'ChangeLog*' \ ! -path COPYING ! -path COPYING.LIB \ ! -path manual/fdl-1.3.texi ! -path manual/lgpl-2.1.texi \ ! -path manual/texinfo.tex ! -path scripts/config.guess \ ! -path scripts/config.sub ! -path scripts/install-sh \ ! -path scripts/mkinstalldirs ! -path scripts/move-if-change \ ! -path INSTALL ! -path locale/programs/charmap-kw.h \ ! -path po/libc.pot ! -path sysdeps/gnu/errlist.c \ ! '(' -name configure \ -execdir test -f configure.ac -o -f configure.in ';' ')' \ ! '(' -name preconfigure \ -execdir test -f preconfigure.ac ';' ')' \ -print) and then by running 'make dist-prepare' to regenerate files built from the altered files, and then executing the following to cleanup: chmod a+x sysdeps/unix/sysv/linux/riscv/configure # Omit irrelevant whitespace and comment-only changes, # perhaps from a slightly-different Autoconf version. git checkout -f \ sysdeps/csky/configure \ sysdeps/hppa/configure \ sysdeps/riscv/configure \ sysdeps/unix/sysv/linux/csky/configure # Omit changes that caused a pre-commit check to fail like this: # remote: *** error: sysdeps/powerpc/powerpc64/ppc-mcount.S: trailing lines git checkout -f \ sysdeps/powerpc/powerpc64/ppc-mcount.S \ sysdeps/unix/sysv/linux/s390/s390-64/syscall.S # Omit change that caused a pre-commit check to fail like this: # remote: *** error: sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S: last line does not end in newline git checkout -f sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S
156 lines
4.4 KiB
ArmAsm
156 lines
4.4 KiB
ArmAsm
/* Optimized strchr implementation for PowerPC64.
|
|
Copyright (C) 1997-2019 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, see
|
|
<https://www.gnu.org/licenses/>. */
|
|
|
|
#include <sysdep.h>
|
|
|
|
/* See strlen.s for comments on how this works. */
|
|
|
|
/* char * [r3] strchr (const char *s [r3] , int c [r4] ) */
|
|
|
|
#ifndef STRCHR
|
|
# define STRCHR strchr
|
|
#endif
|
|
|
|
ENTRY_TOCLESS (STRCHR)
|
|
CALL_MCOUNT 2
|
|
|
|
#define rTMP1 r0
|
|
#define rRTN r3 /* outgoing result */
|
|
#define rSTR r8 /* current word pointer */
|
|
#define rCHR r4 /* byte we're looking for, spread over the whole word */
|
|
#define rWORD r5 /* the current word */
|
|
#define rCLZB rCHR /* leading zero byte count */
|
|
#define rFEFE r6 /* constant 0xfefefefefefefeff (-0x0101010101010101) */
|
|
#define r7F7F r7 /* constant 0x7f7f7f7f7f7f7f7f */
|
|
#define rTMP2 r9
|
|
#define rIGN r10 /* number of bits we should ignore in the first word */
|
|
#define rMASK r11 /* mask with the bits to ignore set to 0 */
|
|
#define rTMP3 r12
|
|
#define rTMP4 rIGN
|
|
#define rTMP5 rMASK
|
|
|
|
dcbt 0,rRTN
|
|
insrdi rCHR, rCHR, 8, 48
|
|
li rMASK, -1
|
|
insrdi rCHR, rCHR, 16, 32
|
|
rlwinm rIGN, rRTN, 3, 26, 28
|
|
insrdi rCHR, rCHR, 32, 0
|
|
lis rFEFE, -0x101
|
|
lis r7F7F, 0x7f7f
|
|
clrrdi rSTR, rRTN, 3
|
|
addi rFEFE, rFEFE, -0x101
|
|
addi r7F7F, r7F7F, 0x7f7f
|
|
sldi rTMP1, rFEFE, 32
|
|
insrdi r7F7F, r7F7F, 32, 0
|
|
add rFEFE, rFEFE, rTMP1
|
|
/* Test the first (partial?) word. */
|
|
ld rWORD, 0(rSTR)
|
|
#ifdef __LITTLE_ENDIAN__
|
|
sld rMASK, rMASK, rIGN
|
|
#else
|
|
srd rMASK, rMASK, rIGN
|
|
#endif
|
|
orc rWORD, rWORD, rMASK
|
|
add rTMP1, rFEFE, rWORD
|
|
nor rTMP2, r7F7F, rWORD
|
|
and. rTMP4, rTMP1, rTMP2
|
|
xor rTMP3, rCHR, rWORD
|
|
orc rTMP3, rTMP3, rMASK
|
|
b L(loopentry)
|
|
|
|
/* The loop. */
|
|
|
|
L(loop):
|
|
ldu rWORD, 8(rSTR)
|
|
and. rTMP5, rTMP1, rTMP2
|
|
/* Test for 0. */
|
|
add rTMP1, rFEFE, rWORD /* x - 0x01010101. */
|
|
nor rTMP2, r7F7F, rWORD /* ~(x | 0x7f7f7f7f) == ~x & 0x80808080. */
|
|
bne L(foundit)
|
|
and. rTMP4, rTMP1, rTMP2 /* (x - 0x01010101) & ~x & 0x80808080. */
|
|
/* Start test for the bytes we're looking for. */
|
|
xor rTMP3, rCHR, rWORD
|
|
L(loopentry):
|
|
add rTMP1, rFEFE, rTMP3
|
|
nor rTMP2, r7F7F, rTMP3
|
|
beq L(loop)
|
|
|
|
/* There is a zero byte in the word, but may also be a matching byte (either
|
|
before or after the zero byte). In fact, we may be looking for a
|
|
zero byte, in which case we return a match. */
|
|
and. rTMP5, rTMP1, rTMP2
|
|
li rRTN, 0
|
|
beqlr
|
|
/* At this point:
|
|
rTMP5 bytes are 0x80 for each match of c, 0 otherwise.
|
|
rTMP4 bytes are 0x80 for each match of 0, 0 otherwise.
|
|
But there may be false matches in the next most significant byte from
|
|
a true match due to carries. This means we need to recalculate the
|
|
matches using a longer method for big-endian. */
|
|
#ifdef __LITTLE_ENDIAN__
|
|
addi rTMP1, rTMP5, -1
|
|
andc rTMP1, rTMP1, rTMP5
|
|
cntlzd rCLZB, rTMP1
|
|
addi rTMP2, rTMP4, -1
|
|
andc rTMP2, rTMP2, rTMP4
|
|
cmpld rTMP1, rTMP2
|
|
bgtlr
|
|
subfic rCLZB, rCLZB, 64-7
|
|
#else
|
|
/* I think we could reduce this by two instructions by keeping the "nor"
|
|
results from the loop for reuse here. See strlen.S tail. Similarly
|
|
one instruction could be pruned from L(foundit). */
|
|
and rFEFE, r7F7F, rWORD
|
|
or rTMP5, r7F7F, rWORD
|
|
and rTMP1, r7F7F, rTMP3
|
|
or rTMP4, r7F7F, rTMP3
|
|
add rFEFE, rFEFE, r7F7F
|
|
add rTMP1, rTMP1, r7F7F
|
|
nor rWORD, rTMP5, rFEFE
|
|
nor rTMP2, rTMP4, rTMP1
|
|
cntlzd rCLZB, rTMP2
|
|
cmpld rWORD, rTMP2
|
|
bgtlr
|
|
#endif
|
|
srdi rCLZB, rCLZB, 3
|
|
add rRTN, rSTR, rCLZB
|
|
blr
|
|
|
|
L(foundit):
|
|
#ifdef __LITTLE_ENDIAN__
|
|
addi rTMP1, rTMP5, -1
|
|
andc rTMP1, rTMP1, rTMP5
|
|
cntlzd rCLZB, rTMP1
|
|
subfic rCLZB, rCLZB, 64-7-64
|
|
sradi rCLZB, rCLZB, 3
|
|
#else
|
|
and rTMP1, r7F7F, rTMP3
|
|
or rTMP4, r7F7F, rTMP3
|
|
add rTMP1, rTMP1, r7F7F
|
|
nor rTMP2, rTMP4, rTMP1
|
|
cntlzd rCLZB, rTMP2
|
|
subi rSTR, rSTR, 8
|
|
srdi rCLZB, rCLZB, 3
|
|
#endif
|
|
add rRTN, rSTR, rCLZB
|
|
blr
|
|
END (STRCHR)
|
|
|
|
weak_alias (strchr, index)
|
|
libc_hidden_builtin_def (strchr)
|