|
103
by Will Newton
Split bionic reference code into A15 and A9 versions. |
1 |
/*
|
2 |
* Copyright (C) 2008 The Android Open Source Project
|
|
3 |
* All rights reserved.
|
|
4 |
*
|
|
5 |
* Redistribution and use in source and binary forms, with or without
|
|
6 |
* modification, are permitted provided that the following conditions
|
|
7 |
* are met:
|
|
8 |
* * Redistributions of source code must retain the above copyright
|
|
9 |
* notice, this list of conditions and the following disclaimer.
|
|
10 |
* * Redistributions in binary form must reproduce the above copyright
|
|
11 |
* notice, this list of conditions and the following disclaimer in
|
|
12 |
* the documentation and/or other materials provided with the
|
|
13 |
* distribution.
|
|
14 |
*
|
|
15 |
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
|
16 |
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
|
17 |
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
|
|
18 |
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
|
|
19 |
* COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
|
|
20 |
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
|
|
21 |
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
|
|
22 |
* OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
|
|
23 |
* AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
|
|
24 |
* OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
|
|
25 |
* OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
26 |
* SUCH DAMAGE.
|
|
27 |
*/
|
|
28 |
||
29 |
#include <string.h> |
|
30 |
#include <stdint.h> |
|
31 |
#undef strlen
|
|
32 |
||
33 |
#define __ARM_HAVE_PLD 1
|
|
34 |
||
35 |
size_t strlen(const char *s) |
|
36 |
{
|
|
37 |
__builtin_prefetch(s); |
|
38 |
__builtin_prefetch(s+32); |
|
39 |
||
40 |
union { |
|
41 |
const char *b; |
|
42 |
const uint32_t *w; |
|
43 |
uintptr_t i; |
|
44 |
} u; |
|
45 |
||
46 |
// these are some scratch variables for the asm code below
|
|
47 |
uint32_t v, t; |
|
48 |
||
49 |
// initialize the string length to zero
|
|
50 |
size_t l = 0; |
|
51 |
||
52 |
// align the pointer to a 32-bit word boundary
|
|
53 |
u.b = s; |
|
54 |
while (u.i & 0x3) { |
|
55 |
if (__builtin_expect(*u.b++ == 0, 0)) { |
|
56 |
goto done; |
|
57 |
}
|
|
58 |
l++; |
|
59 |
}
|
|
60 |
||
61 |
// loop for each word, testing if it contains a zero byte
|
|
62 |
// if so, exit the loop and update the length.
|
|
63 |
// We need to process 32 bytes per loop to schedule PLD properly
|
|
64 |
// and achieve the maximum bus speed.
|
|
65 |
asm( |
|
66 |
"ldr %[v], [%[s]], #4 \n" |
|
67 |
"sub %[l], %[l], %[s] \n" |
|
68 |
"0: \n" |
|
69 |
#if __ARM_HAVE_PLD
|
|
70 |
"pld [%[s], #64] \n" |
|
71 |
#endif
|
|
72 |
"sub %[t], %[v], %[mask], lsr #7\n" |
|
73 |
"and %[t], %[t], %[mask] \n" |
|
74 |
"bics %[t], %[t], %[v] \n" |
|
75 |
"ldreq %[v], [%[s]], #4 \n" |
|
76 |
#if !defined(__OPTIMIZE_SIZE__)
|
|
77 |
"bne 1f \n" |
|
78 |
"sub %[t], %[v], %[mask], lsr #7\n" |
|
79 |
"and %[t], %[t], %[mask] \n" |
|
80 |
"bics %[t], %[t], %[v] \n" |
|
81 |
"ldreq %[v], [%[s]], #4 \n" |
|
82 |
"bne 1f \n" |
|
83 |
"sub %[t], %[v], %[mask], lsr #7\n" |
|
84 |
"and %[t], %[t], %[mask] \n" |
|
85 |
"bics %[t], %[t], %[v] \n" |
|
86 |
"ldreq %[v], [%[s]], #4 \n" |
|
87 |
"bne 1f \n" |
|
88 |
"sub %[t], %[v], %[mask], lsr #7\n" |
|
89 |
"and %[t], %[t], %[mask] \n" |
|
90 |
"bics %[t], %[t], %[v] \n" |
|
91 |
"ldreq %[v], [%[s]], #4 \n" |
|
92 |
"bne 1f \n" |
|
93 |
"sub %[t], %[v], %[mask], lsr #7\n" |
|
94 |
"and %[t], %[t], %[mask] \n" |
|
95 |
"bics %[t], %[t], %[v] \n" |
|
96 |
"ldreq %[v], [%[s]], #4 \n" |
|
97 |
"bne 1f \n" |
|
98 |
"sub %[t], %[v], %[mask], lsr #7\n" |
|
99 |
"and %[t], %[t], %[mask] \n" |
|
100 |
"bics %[t], %[t], %[v] \n" |
|
101 |
"ldreq %[v], [%[s]], #4 \n" |
|
102 |
"bne 1f \n" |
|
103 |
"sub %[t], %[v], %[mask], lsr #7\n" |
|
104 |
"and %[t], %[t], %[mask] \n" |
|
105 |
"bics %[t], %[t], %[v] \n" |
|
106 |
"ldreq %[v], [%[s]] , #4 \n" |
|
107 |
"bne 1f \n" |
|
108 |
"sub %[t], %[v], %[mask], lsr #7\n" |
|
109 |
"and %[t], %[t], %[mask] \n" |
|
110 |
"bics %[t], %[t], %[v] \n" |
|
111 |
"ldreq %[v], [%[s]], #4 \n" |
|
112 |
#endif
|
|
113 |
"beq 0b \n" |
|
114 |
"1: \n" |
|
115 |
"add %[l], %[l], %[s] \n" |
|
116 |
"tst %[v], #0xFF \n" |
|
117 |
"beq 2f \n" |
|
118 |
"add %[l], %[l], #1 \n" |
|
119 |
"tst %[v], #0xFF00 \n" |
|
120 |
"beq 2f \n" |
|
121 |
"add %[l], %[l], #1 \n" |
|
122 |
"tst %[v], #0xFF0000 \n" |
|
123 |
"addne %[l], %[l], #1 \n" |
|
124 |
"2: \n" |
|
125 |
: [l]"=&r"(l), [v]"=&r"(v), [t]"=&r"(t), [s]"=&r"(u.b) |
|
126 |
: "%[l]"(l), "%[s]"(u.b), [mask]"r"(0x80808080UL) |
|
127 |
: "cc" |
|
128 |
);
|
|
129 |
||
130 |
done: |
|
131 |
return l; |
|
132 |
}
|