1 /* Copyright (C) 2011-2014 Free Software Foundation, Inc.
2 This file is part of the GNU C Library.
3 Code contributed by Dave Gilbert <david.gilbert@linaro.org>
5 The GNU C Library is free software; you can redistribute it and/or
6 modify it under the terms of the GNU Lesser General Public
7 License as published by the Free Software Foundation; either
8 version 2.1 of the License, or (at your option) any later version.
10 The GNU C Library is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 Lesser General Public License for more details.
15 You should have received a copy of the GNU Lesser General Public
16 License along with the GNU C Library. If not, see
17 <http://www.gnu.org/licenses/>. */
21 @ This memchr routine is optimised on a Cortex-A9 and should work on all ARMv7
22 @ and ARMv6T2 processors. It has a fast path for short sizes, and has an
23 @ optimised path for large data sets; the worst case is finding the match early
24 @ in a large data set.
25 @ Note: The use of cbz/cbnz means it's Thumb only
27 @ 2011-07-15 david.gilbert@linaro.org
28 @ Copy from Cortex strings release 21 and change license
29 @ http://bazaar.launchpad.net/~linaro-toolchain-dev/cortex-strings/trunk/view/head:/src/linaro-a9/memchr.S
30 @ Change function declarations/entry/exit
31 @ 2011-12-01 david.gilbert@linaro.org
32 @ Add some fixes from comments received (including use of ldrd instead ldm)
33 @ 2011-12-07 david.gilbert@linaro.org
34 @ Removed cbz from align loop - can't be taken
36 @ this lets us check a flag in a 00/ff byte easily in either endianness
38 #define CHARTSTMASK(c) 1<<(31-(c*8))
40 #define CHARTSTMASK(c) 1<<(c*8)
52 .type memchr,%function
54 @ r0 = start of memory to scan
55 @ r1 = character to look for
57 @ returns r0 = pointer to character or NULL if not found
58 and r1,r1,#0xff @ Don't think we can trust the caller to actually pass a char
60 cmp r2,#16 @ If it's short don't bother with anything clever
63 tst r0, #7 @ If it's already aligned skip the next bit
66 @ Work up to an aligned point
72 beq 50f @ If it matches exit found
74 bne 5b @ If not aligned yet then do next byte
77 @ At this point, we are aligned, we know we have at least 8 bytes to work with
79 cfi_adjust_cfa_offset (16)
80 cfi_rel_offset (r4, 0)
81 cfi_rel_offset (r5, 4)
82 cfi_rel_offset (r6, 8)
83 cfi_rel_offset (r7, 12)
87 orr r1, r1, r1, lsl #8 @ expand the match word across to all bytes
88 orr r1, r1, r1, lsl #16
89 bic r6, r2, #7 @ Number of double words to work with * 8
99 eor r4,r4, r1 @ Get it so that r4,r5 have 00's where the bytes match the target
101 uadd8 r4, r4, r7 @ Parallel add 0xff - sets the GE bits for anything that wasn't 0
102 sel r4, r3, r7 @ bytes are 00 for none-00 bytes, or ff for 00 bytes - NOTE INVERSION
103 uadd8 r5, r5, r7 @ Parallel add 0xff - sets the GE bits for anything that wasn't 0
104 sel r5, r4, r7 @ chained....bytes are 00 for none-00 bytes, or ff for 00 bytes - NOTE INVERSION
112 bne 15b @ (Flags from the subs above) If not run out of bytes then go around again
115 cfi_adjust_cfa_offset (-16)
121 and r1,r1,#0xff @ Get r1 back to a single character from the expansion above
122 and r2,r2,#7 @ Leave the count remaining as the number after the double words have been done
126 cbz r2, 40f @ 0 length or hit the end already then not found
132 21: @ Post aligned section, or just a short call
137 eor r3,r3,r1 @ r3 = 0 if match - doesn't break flags from sub
144 bne 21b @ on r2 flags
147 movs r0,#0 @ not found
151 subs r0,r0,#1 @ found
154 60: @ We're here because the fast path found a hit - now we have to track down exactly which word it was
155 @ r0 points to the start of the double word after the one that was tested
156 @ r4 has the 00/ff pattern for the first word, r5 has the chained value
160 moveq r4, r5 @ the end is in the 2nd word
161 subeq r0,r0,#3 @ Points to 2nd byte of 2nd word
162 subne r0,r0,#7 @ or 2nd byte of 1st word
164 @ r0 currently points to the 2nd byte of the word containing the hit
165 tst r4, # CHARTSTMASK(0) @ 1st character
168 tst r4, # CHARTSTMASK(1) @ 2nd character
171 tsteq r4, # (3<<15) @ 2nd & 3rd character
172 @ If not the 3rd must be the last one
177 cfi_adjust_cfa_offset (-16)
187 libc_hidden_builtin_def (memchr)