AuroraMiddleware/glibc
1
0
Fork 0
mirror of https://sourceware.org/git/glibc.git synced 2024-12-11 13:50:06 +00:00
Code Issues Packages Projects Releases Wiki Activity
42675c6ff0
glibc/sysdeps/sparc/sparc64/memchr.S
David S. Miller 2daff75bab sparc: Fix v9 memchr with long buffer lengths. 
test-memchr fails on 32-bit sparc with v9 optimized memchr because
it uses signed branch comparisons on the buffer pointers.
2010-02-20 13:48:06 -08:00

264 lines
8.2 KiB
ArmAsm
Raw Blame History

/* memchr (str, ch, n) -- Return pointer to first occurrence of CH in STR less
than N.
For SPARC v9.
Copyright (C) 1998, 1999, 2000, 2003 Free Software Foundation, Inc.
This file is part of the GNU C Library.
Contributed by Jan Vondrak <jvon4518@ss1000.ms.mff.cuni.cz> and
Jakub Jelinek <jj@ultra.linux.cz>.
This version is developed using the same algorithm as the fast C
version which carries the following introduction:
Based on strlen implementation by Torbjorn Granlund (tege@sics.se),
with help from Dan Sahlin (dan@sics.se) and
commentary by Jim Blandy (jimb@ai.mit.edu);
adaptation to memchr suggested by Dick Karpinski (dick@cca.ucsf.edu),
and implemented by Roland McGrath (roland@ai.mit.edu).
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 <asm/asi.h>
#ifndef XCC
#define XCC xcc
#define USE_BPR
.register %g2, #scratch
.register %g3, #scratch
#endif
/* Normally, this uses
((xword - 0x0101010101010101) & 0x8080808080808080) test
to find out if any byte in xword could be zero. This is fast, but
also gives false alarm for any byte in range 0x81-0xff. It does
not matter for correctness, as if this test tells us there could
be some zero byte, we check it byte by byte, but if bytes with
high bits set are common in the strings, then this will give poor
performance. You can #define EIGHTBIT_NOT_RARE and the algorithm
will use one tick slower, but more precise test
((xword - 0x0101010101010101) & (~xword) & 0x8080808080808080),
which does not give any false alarms (but if some bits are set,
one cannot assume from it which bytes are zero and which are not).
It is yet to be measured, what is the correct default for glibc
in these days for an average user.
*/
.text
.align 32
ENTRY(__memchr)
and %o1, 0xff, %o1 /* IEU0 Group */
#ifdef USE_BPR
brz,pn %o2, 12f /* CTI+IEU1 */
#else
tst %o2 /* IEU1 */
be,pn %XCC, 12f /* CTI */
#endif
sll %o1, 8, %g3 /* IEU0 Group */
add %o0, %o2, %o2 /* IEU1 */
sethi %hi(0x01010101), %g1 /* IEU0 Group */
or %g3, %o1, %g3 /* IEU1 */
ldub [%o0], %o3 /* Load */
sllx %g3, 16, %g5 /* IEU0 Group */
or %g1, %lo(0x01010101), %g1 /* IEU1 */
sllx %g1, 32, %g2 /* IEU0 Group */
or %g3, %g5, %g3 /* IEU1 */
sllx %g3, 32, %g5 /* IEU0 Group */
cmp %o3, %o1 /* IEU1 */
be,pn %xcc, 13f /* CTI */
or %g1, %g2, %g1 /* IEU0 Group */
andcc %o0, 7, %g0 /* IEU1 */
bne,a,pn %icc, 21f /* CTI */
add %o0, 1, %o0 /* IEU0 Group */
ldx [%o0], %o3 /* Load Group */
sllx %g1, 7, %g2 /* IEU0 */
or %g3, %g5, %g3 /* IEU1 */
1: add %o0, 8, %o0 /* IEU0 Group */
xor %o3, %g3, %o4 /* IEU1 */
/* %g1 = 0101010101010101 *
* %g2 = 8080088080808080 *
* %g3 = c c c c c c c c *
* %o3 = value *
* %o4 = value XOR c */
2: cmp %o0, %o2 /* IEU1 Group */
bgu,pn %XCC, 11f /* CTI */
ldxa [%o0] ASI_PNF, %o3 /* Load */
sub %o4, %g1, %o5 /* IEU0 Group */
add %o0, 8, %o0 /* IEU1 */
#ifdef EIGHTBIT_NOT_RARE
andn %o5, %o4, %o5 /* IEU0 Group */
#endif
andcc %o5, %g2, %g0 /* IEU1 Group */
be,a,pt %xcc, 2b /* CTI */
xor %o3, %g3, %o4 /* IEU0 */
srlx %o4, 56, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 3f /* CTI */
srlx %o4, 48, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 4f /* CTI */
srlx %o4, 40, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 5f /* CTI */
srlx %o4, 32, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 6f /* CTI */
srlx %o4, 24, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 7f /* CTI */
srlx %o4, 16, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 8f /* CTI */
srlx %o4, 8, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 9f /* CTI */
andcc %o4, 0xff, %g0 /* IEU1 Group */
bne,pt %icc, 2b /* CTI */
xor %o3, %g3, %o4 /* IEU0 */
retl /* CTI+IEU1 Group */
add %o0, -9, %o0 /* IEU0 */
.align 16
3: retl /* CTI+IEU1 Group */
add %o0, -16, %o0 /* IEU0 */
4: retl /* CTI+IEU1 Group */
add %o0, -15, %o0 /* IEU0 */
5: retl /* CTI+IEU1 Group */
add %o0, -14, %o0 /* IEU0 */
6: retl /* CTI+IEU1 Group */
add %o0, -13, %o0 /* IEU0 */
7: retl /* CTI+IEU1 Group */
add %o0, -12, %o0 /* IEU0 */
8: retl /* CTI+IEU1 Group */
add %o0, -11, %o0 /* IEU0 */
9: retl /* CTI+IEU1 Group */
add %o0, -10, %o0 /* IEU0 */
11: sub %o4, %g1, %o5 /* IEU0 Group */
sub %o0, 8, %o0 /* IEU1 */
andcc %o5, %g2, %g0 /* IEU1 Group */
be,pt %xcc, 12f /* CTI */
sub %o2, %o0, %o2 /* IEU0 */
tst %o2 /* IEU1 Group */
be,pn %XCC, 12f /* CTI */
srlx %o4, 56, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 13f /* CTI */
cmp %o2, 1 /* IEU0 */
be,pn %XCC, 12f /* CTI Group */
srlx %o4, 48, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 14f /* CTI */
cmp %o2, 2 /* IEU1 Group */
be,pn %XCC, 12f /* CTI */
srlx %o4, 40, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 15f /* CTI */
cmp %o2, 3 /* IEU1 Group */
be,pn %XCC, 12f /* CTI */
srlx %o4, 32, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 16f /* CTI */
cmp %o2, 4 /* IEU1 Group */
be,pn %XCC, 12f /* CTI */
srlx %o4, 24, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 17f /* CTI */
cmp %o2, 5 /* IEU1 Group */
be,pn %XCC, 12f /* CTI */
srlx %o4, 16, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 18f /* CTI */
cmp %o2, 6 /* IEU1 Group */
be,pn %XCC, 12f /* CTI */
srlx %o4, 8, %g5 /* IEU0 */
andcc %g5, 0xff, %g0 /* IEU1 Group */
be,pn %icc, 19f /* CTI */
nop /* IEU0 */
12: retl /* CTI+IEU1 Group */
clr %o0 /* IEU0 */
nop /* Stub */
13: retl /* CTI+IEU1 Group */
nop /* IEU0 */
14: retl /* CTI+IEU1 Group */
add %o0, 1, %o0 /* IEU0 */
15: retl /* CTI+IEU1 Group */
add %o0, 2, %o0 /* IEU0 */
16: retl /* CTI+IEU1 Group */
add %o0, 3, %o0 /* IEU0 */
17: retl /* CTI+IEU1 Group */
add %o0, 4, %o0 /* IEU0 */
18: retl /* CTI+IEU1 Group */
add %o0, 5, %o0 /* IEU0 */
19: retl /* CTI+IEU1 Group */
add %o0, 6, %o0 /* IEU0 */
21: cmp %o0, %o2 /* IEU1 */
be,pn %XCC, 12b /* CTI */
sllx %g1, 7, %g2 /* IEU0 Group */
ldub [%o0], %o3 /* Load */
or %g3, %g5, %g3 /* IEU1 */
22: andcc %o0, 7, %g0 /* IEU1 Group */
be,a,pn %icc, 1b /* CTI */
ldx [%o0], %o3 /* Load */
cmp %o3, %o1 /* IEU1 Group */
be,pn %xcc, 23f /* CTI */
add %o0, 1, %o0 /* IEU0 */
cmp %o0, %o2 /* IEU1 Group */
bne,a,pt %XCC, 22b /* CTI */
ldub [%o0], %o3 /* Load */
retl /* CTI+IEU1 Group */
clr %o0 /* IEU0 */
23: retl /* CTI+IEU1 Group */
add %o0, -1, %o0 /* IEU0 */
END(__memchr)
weak_alias (__memchr, memchr)
#if !__BOUNDED_POINTERS__
weak_alias (__memchr, __ubp_memchr)
#endif
libc_hidden_builtin_def (memchr)
Reference in New Issue View Git Blame Copy Permalink