diff options
author | Thomas Martitz <kugel@rockbox.org> | 2012-01-07 19:56:09 +0100 |
---|---|---|
committer | Thomas Martitz <kugel@rockbox.org> | 2012-01-22 18:46:45 +0100 |
commit | a035261089403de259e74ce4dd196e2715138ed2 (patch) | |
tree | 440d75109a7ebfefef2a58076b38b1bfe9db8f63 /firmware/asm/arm/memcpy.S | |
parent | 8e8e978de6b6283b66a6829fa8c5e3b94674ce7d (diff) | |
download | rockbox-a035261089403de259e74ce4dd196e2715138ed2.tar.gz rockbox-a035261089403de259e74ce4dd196e2715138ed2.zip |
Move optimized memcpy and friends and strlen to firmware/asm,
using the new automatic-asm-picking infrastructure.
Diffstat (limited to 'firmware/asm/arm/memcpy.S')
-rw-r--r-- | firmware/asm/arm/memcpy.S | 176 |
1 files changed, 176 insertions, 0 deletions
diff --git a/firmware/asm/arm/memcpy.S b/firmware/asm/arm/memcpy.S new file mode 100644 index 0000000000..2a55fb5656 --- /dev/null +++ b/firmware/asm/arm/memcpy.S | |||
@@ -0,0 +1,176 @@ | |||
1 | /*************************************************************************** | ||
2 | * __________ __ ___. | ||
3 | * Open \______ \ ____ ____ | | _\_ |__ _______ ___ | ||
4 | * Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ / | ||
5 | * Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < < | ||
6 | * Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \ | ||
7 | * \/ \/ \/ \/ \/ | ||
8 | * $Id$ | ||
9 | * | ||
10 | * Copyright (C) 2006 Free Software Foundation, Inc. | ||
11 | * This file was originally part of the GNU C Library | ||
12 | * Contributed to glibc by MontaVista Software, Inc. (written by Nicolas Pitre) | ||
13 | * Adapted for Rockbox by Daniel Ankers | ||
14 | * | ||
15 | * This program is free software; you can redistribute it and/or | ||
16 | * modify it under the terms of the GNU General Public License | ||
17 | * as published by the Free Software Foundation; either version 2 | ||
18 | * of the License, or (at your option) any later version. | ||
19 | * | ||
20 | * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY | ||
21 | * KIND, either express or implied. | ||
22 | * | ||
23 | ****************************************************************************/ | ||
24 | |||
25 | #include "config.h" | ||
26 | |||
27 | /* | ||
28 | * Endian independent macros for shifting bytes within registers. | ||
29 | */ | ||
30 | #ifndef __ARMEB__ | ||
31 | #define pull lsr | ||
32 | #define push lsl | ||
33 | #else | ||
34 | #define pull lsl | ||
35 | #define push lsr | ||
36 | #endif | ||
37 | |||
38 | /* Prototype: void *memcpy(void *dest, const void *src, size_t n); */ | ||
39 | |||
40 | .section .icode,"ax",%progbits | ||
41 | |||
42 | .align 2 | ||
43 | .global memcpy | ||
44 | .type memcpy,%function | ||
45 | |||
46 | memcpy: | ||
47 | stmfd sp!, {r0, r4, lr} | ||
48 | |||
49 | subs r2, r2, #4 | ||
50 | blt 8f | ||
51 | ands ip, r0, #3 | ||
52 | bne 9f | ||
53 | ands ip, r1, #3 | ||
54 | bne 10f | ||
55 | |||
56 | 1: subs r2, r2, #(28) | ||
57 | stmfd sp!, {r5 - r8} | ||
58 | blt 5f | ||
59 | |||
60 | 2: | ||
61 | 3: | ||
62 | 4: ldmia r1!, {r3, r4, r5, r6, r7, r8, ip, lr} | ||
63 | subs r2, r2, #32 | ||
64 | stmia r0!, {r3, r4, r5, r6, r7, r8, ip, lr} | ||
65 | bge 3b | ||
66 | |||
67 | 5: ands ip, r2, #28 | ||
68 | rsb ip, ip, #32 | ||
69 | addne pc, pc, ip @ C is always clear here | ||
70 | b 7f | ||
71 | 6: nop | ||
72 | ldr r3, [r1], #4 | ||
73 | ldr r4, [r1], #4 | ||
74 | ldr r5, [r1], #4 | ||
75 | ldr r6, [r1], #4 | ||
76 | ldr r7, [r1], #4 | ||
77 | ldr r8, [r1], #4 | ||
78 | ldr lr, [r1], #4 | ||
79 | |||
80 | add pc, pc, ip | ||
81 | nop | ||
82 | nop | ||
83 | str r3, [r0], #4 | ||
84 | str r4, [r0], #4 | ||
85 | str r5, [r0], #4 | ||
86 | str r6, [r0], #4 | ||
87 | str r7, [r0], #4 | ||
88 | str r8, [r0], #4 | ||
89 | str lr, [r0], #4 | ||
90 | |||
91 | 7: ldmfd sp!, {r5 - r8} | ||
92 | |||
93 | 8: movs r2, r2, lsl #31 | ||
94 | ldrneb r3, [r1], #1 | ||
95 | ldrcsb r4, [r1], #1 | ||
96 | ldrcsb ip, [r1] | ||
97 | strneb r3, [r0], #1 | ||
98 | strcsb r4, [r0], #1 | ||
99 | strcsb ip, [r0] | ||
100 | |||
101 | ldmpc regs="r0, r4" | ||
102 | |||
103 | 9: rsb ip, ip, #4 | ||
104 | cmp ip, #2 | ||
105 | ldrgtb r3, [r1], #1 | ||
106 | ldrgeb r4, [r1], #1 | ||
107 | ldrb lr, [r1], #1 | ||
108 | strgtb r3, [r0], #1 | ||
109 | strgeb r4, [r0], #1 | ||
110 | subs r2, r2, ip | ||
111 | strb lr, [r0], #1 | ||
112 | blt 8b | ||
113 | ands ip, r1, #3 | ||
114 | beq 1b | ||
115 | |||
116 | 10: bic r1, r1, #3 | ||
117 | cmp ip, #2 | ||
118 | ldr lr, [r1], #4 | ||
119 | beq 17f | ||
120 | bgt 18f | ||
121 | |||
122 | |||
123 | .macro forward_copy_shift pull push | ||
124 | |||
125 | subs r2, r2, #28 | ||
126 | blt 14f | ||
127 | |||
128 | 11: stmfd sp!, {r5 - r9} | ||
129 | |||
130 | 12: | ||
131 | 13: ldmia r1!, {r4, r5, r6, r7} | ||
132 | mov r3, lr, pull #\pull | ||
133 | subs r2, r2, #32 | ||
134 | ldmia r1!, {r8, r9, ip, lr} | ||
135 | orr r3, r3, r4, push #\push | ||
136 | mov r4, r4, pull #\pull | ||
137 | orr r4, r4, r5, push #\push | ||
138 | mov r5, r5, pull #\pull | ||
139 | orr r5, r5, r6, push #\push | ||
140 | mov r6, r6, pull #\pull | ||
141 | orr r6, r6, r7, push #\push | ||
142 | mov r7, r7, pull #\pull | ||
143 | orr r7, r7, r8, push #\push | ||
144 | mov r8, r8, pull #\pull | ||
145 | orr r8, r8, r9, push #\push | ||
146 | mov r9, r9, pull #\pull | ||
147 | orr r9, r9, ip, push #\push | ||
148 | mov ip, ip, pull #\pull | ||
149 | orr ip, ip, lr, push #\push | ||
150 | stmia r0!, {r3, r4, r5, r6, r7, r8, r9, ip} | ||
151 | bge 12b | ||
152 | |||
153 | ldmfd sp!, {r5 - r9} | ||
154 | |||
155 | 14: ands ip, r2, #28 | ||
156 | beq 16f | ||
157 | |||
158 | 15: mov r3, lr, pull #\pull | ||
159 | ldr lr, [r1], #4 | ||
160 | subs ip, ip, #4 | ||
161 | orr r3, r3, lr, push #\push | ||
162 | str r3, [r0], #4 | ||
163 | bgt 15b | ||
164 | |||
165 | 16: sub r1, r1, #(\push / 8) | ||
166 | b 8b | ||
167 | |||
168 | .endm | ||
169 | |||
170 | |||
171 | forward_copy_shift pull=8 push=24 | ||
172 | |||
173 | 17: forward_copy_shift pull=16 push=16 | ||
174 | |||
175 | 18: forward_copy_shift pull=24 push=8 | ||
176 | |||