2014-01-01 12:03:15 +01:00
|
|
|
/* Copyright (C) 1996-2014 Free Software Foundation, Inc.
|
1996-11-06 05:24:11 +01:00
|
|
|
Contributed by Richard Henderson (rth@tamu.edu)
|
|
|
|
This file is part of the GNU C Library.
|
|
|
|
|
|
|
|
The GNU C Library is free software; you can redistribute it and/or
|
2001-07-06 06:56:23 +02:00
|
|
|
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.
|
1996-11-06 05:24:11 +01:00
|
|
|
|
|
|
|
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
|
2001-07-06 06:56:23 +02:00
|
|
|
Lesser General Public License for more details.
|
1996-11-06 05:24:11 +01:00
|
|
|
|
2001-07-06 06:56:23 +02:00
|
|
|
You should have received a copy of the GNU Lesser General Public
|
2012-03-10 00:56:38 +01:00
|
|
|
License along with the GNU C Library. If not, see
|
|
|
|
<http://www.gnu.org/licenses/>. */
|
1996-11-06 05:24:11 +01:00
|
|
|
|
|
|
|
/* Bytewise compare two null-terminated strings. */
|
|
|
|
|
|
|
|
#include <sysdep.h>
|
|
|
|
|
|
|
|
.set noat
|
|
|
|
.set noreorder
|
|
|
|
|
|
|
|
.text
|
|
|
|
|
|
|
|
ENTRY(strcmp)
|
|
|
|
#ifdef PROF
|
|
|
|
ldgp gp, 0(pv)
|
|
|
|
lda AT, _mcount
|
|
|
|
jmp AT, (AT), _mcount
|
|
|
|
.prologue 1
|
|
|
|
#else
|
|
|
|
.prologue 0
|
|
|
|
#endif
|
|
|
|
|
|
|
|
ldq_u t0, 0(a0) # e0 : give cache time to catch up
|
|
|
|
xor a0, a1, t2 # .. e1 : are s1 and s2 co-aligned?
|
|
|
|
ldq_u t1, 0(a1) # e0 :
|
|
|
|
and t2, 7, t2 # .. e1 :
|
|
|
|
lda t3, -1 # e0 :
|
|
|
|
bne t2, $unaligned # .. e1 :
|
|
|
|
|
|
|
|
/* On entry to this basic block:
|
|
|
|
t0 == the first destination word for masking back in
|
|
|
|
t1 == the first source word.
|
|
|
|
t3 == -1. */
|
|
|
|
|
|
|
|
$aligned:
|
|
|
|
mskqh t3, a0, t3 # e0 :
|
|
|
|
nop # .. e1 :
|
|
|
|
ornot t1, t3, t1 # e0 :
|
|
|
|
ornot t0, t3, t0 # .. e1 :
|
|
|
|
cmpbge zero, t1, t7 # e0 : bits set iff null found
|
|
|
|
bne t7, $eos # e1 (zdb)
|
|
|
|
|
|
|
|
/* Aligned compare main loop.
|
|
|
|
On entry to this basic block:
|
|
|
|
t0 == an s1 word.
|
|
|
|
t1 == an s2 word not containing a null. */
|
|
|
|
|
|
|
|
$a_loop:
|
|
|
|
xor t0, t1, t2 # e0 :
|
|
|
|
bne t2, $wordcmp # .. e1 (zdb)
|
|
|
|
ldq_u t1, 8(a1) # e0 :
|
|
|
|
ldq_u t0, 8(a0) # .. e1 :
|
|
|
|
addq a1, 8, a1 # e0 :
|
|
|
|
addq a0, 8, a0 # .. e1 :
|
|
|
|
cmpbge zero, t1, t7 # e0 :
|
|
|
|
beq t7, $a_loop # .. e1 (zdb)
|
|
|
|
br $eos # e1 :
|
|
|
|
|
|
|
|
/* The two strings are not co-aligned. Align s1 and cope. */
|
|
|
|
|
|
|
|
$unaligned:
|
|
|
|
and a0, 7, t4 # e0 : find s1 misalignment
|
|
|
|
and a1, 7, t5 # .. e1 : find s2 misalignment
|
|
|
|
subq a1, t4, a1 # e0 :
|
|
|
|
|
|
|
|
/* If s2 misalignment is larger than s2 misalignment, we need
|
|
|
|
extra startup checks to avoid SEGV. */
|
|
|
|
|
|
|
|
cmplt t4, t5, t8 # .. e1 :
|
|
|
|
beq t8, $u_head # e1 :
|
|
|
|
|
|
|
|
mskqh t3, t5, t3 # e0 :
|
|
|
|
ornot t1, t3, t3 # e0 :
|
|
|
|
cmpbge zero, t3, t7 # e1 : is there a zero?
|
|
|
|
beq t7, $u_head # e1 :
|
|
|
|
|
|
|
|
/* We've found a zero in the first partial word of s2. Align
|
|
|
|
our current s1 and s2 words and compare what we've got. */
|
|
|
|
|
|
|
|
extql t1, t5, t1 # e0 :
|
|
|
|
extql t0, a0, t0 # e0 :
|
|
|
|
cmpbge zero, t1, t7 # .. e1 : find that zero again
|
|
|
|
br $eos # e1 : and finish up
|
|
|
|
|
|
|
|
.align 3
|
|
|
|
$u_head:
|
|
|
|
/* We know just enough now to be able to assemble the first
|
|
|
|
full word of s2. We can still find a zero at the end of it.
|
|
|
|
|
|
|
|
On entry to this basic block:
|
|
|
|
t0 == first word of s1
|
|
|
|
t1 == first partial word of s2. */
|
|
|
|
|
|
|
|
ldq_u t2, 8(a1) # e0 : load second partial s2 word
|
|
|
|
lda t3, -1 # .. e1 : create leading garbage mask
|
|
|
|
extql t1, a1, t1 # e0 : create first s2 word
|
|
|
|
mskqh t3, a0, t3 # e0 :
|
|
|
|
extqh t2, a1, t4 # e0 :
|
|
|
|
ornot t0, t3, t0 # .. e1 : kill s1 garbage
|
|
|
|
or t1, t4, t1 # e0 : s2 word now complete
|
|
|
|
cmpbge zero, t0, t7 # .. e1 : find zero in first s1 word
|
|
|
|
ornot t1, t3, t1 # e0 : kill s2 garbage
|
|
|
|
lda t3, -1 # .. e1 :
|
|
|
|
mskql t3, a1, t3 # e0 : mask for s2[1] bits we have seen
|
|
|
|
bne t7, $eos # .. e1 :
|
|
|
|
xor t0, t1, t4 # e0 : compare aligned words
|
|
|
|
bne t4, $wordcmp # .. e1 (zdb)
|
|
|
|
or t2, t3, t3 # e0 :
|
|
|
|
cmpbge zero, t3, t7 # e1 :
|
|
|
|
bne t7, $u_final # e1 :
|
|
|
|
|
|
|
|
/* Unaligned copy main loop. In order to avoid reading too much,
|
|
|
|
the loop is structured to detect zeros in aligned words from s2.
|
|
|
|
This has, unfortunately, effectively pulled half of a loop
|
|
|
|
iteration out into the head and half into the tail, but it does
|
|
|
|
prevent nastiness from accumulating in the very thing we want
|
|
|
|
to run as fast as possible.
|
|
|
|
|
|
|
|
On entry to this basic block:
|
|
|
|
t2 == the unshifted low-bits from the next s2 word. */
|
|
|
|
|
|
|
|
.align 3
|
|
|
|
$u_loop:
|
|
|
|
extql t2, a1, t3 # e0 :
|
|
|
|
ldq_u t2, 16(a1) # .. e1 : load next s2 high bits
|
|
|
|
ldq_u t0, 8(a0) # e0 : load next s1 word
|
|
|
|
addq a1, 8, a1 # .. e1 :
|
|
|
|
addq a0, 8, a0 # e0 :
|
|
|
|
nop # .. e1 :
|
|
|
|
extqh t2, a1, t1 # e0 :
|
|
|
|
cmpbge zero, t0, t7 # .. e1 : find zero in current s1 word
|
|
|
|
or t1, t3, t1 # e0 :
|
|
|
|
bne t7, $eos # .. e1 :
|
|
|
|
xor t0, t1, t4 # e0 : compare the words
|
|
|
|
bne t4, $wordcmp # .. e1 (zdb)
|
|
|
|
cmpbge zero, t2, t4 # e0 : find zero in next low bits
|
|
|
|
beq t4, $u_loop # .. e1 (zdb)
|
|
|
|
|
|
|
|
/* We've found a zero in the low bits of the last s2 word. Get
|
|
|
|
the next s1 word and align them. */
|
|
|
|
$u_final:
|
|
|
|
ldq_u t0, 8(a0) # e1 :
|
|
|
|
extql t2, a1, t1 # .. e0 :
|
|
|
|
cmpbge zero, t1, t7 # e0 :
|
|
|
|
|
|
|
|
/* We've found a zero somewhere in a word we just read.
|
|
|
|
On entry to this basic block:
|
|
|
|
t0 == s1 word
|
|
|
|
t1 == s2 word
|
|
|
|
t7 == cmpbge mask containing the zero. */
|
|
|
|
|
|
|
|
.align 3
|
|
|
|
$eos:
|
|
|
|
negq t7, t6 # e0 : create bytemask of valid data
|
|
|
|
and t6, t7, t8 # e1 :
|
|
|
|
subq t8, 1, t6 # e0 :
|
|
|
|
or t6, t8, t7 # e1 :
|
|
|
|
zapnot t0, t7, t0 # e0 : kill the garbage
|
|
|
|
zapnot t1, t7, t1 # .. e1 :
|
|
|
|
xor t0, t1, v0 # e0 : and compare
|
|
|
|
beq v0, $done # .. e1 :
|
|
|
|
|
|
|
|
/* Here we have two differing co-aligned words in t0 & t1.
|
1997-06-27 00:16:52 +02:00
|
|
|
Bytewise compare them and return (t0 > t1 ? 1 : -1). */
|
1996-11-06 05:24:11 +01:00
|
|
|
$wordcmp:
|
|
|
|
cmpbge t0, t1, t2 # e0 : comparison yields bit mask of ge
|
|
|
|
cmpbge t1, t0, t3 # .. e1 :
|
|
|
|
xor t2, t3, t0 # e0 : bits set iff t0/t1 bytes differ
|
|
|
|
negq t0, t1 # e1 : clear all but least bit
|
|
|
|
and t0, t1, t0 # e0 :
|
|
|
|
lda v0, -1 # .. e1 :
|
|
|
|
and t0, t2, t1 # e0 : was bit set in t0 > t1?
|
|
|
|
cmovne t1, 1, v0 # .. e1 (zdb)
|
|
|
|
|
|
|
|
$done:
|
|
|
|
ret # e1 :
|
|
|
|
|
|
|
|
END(strcmp)
|
2003-04-30 00:47:20 +02:00
|
|
|
libc_hidden_builtin_def (strcmp)
|