1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
|
/*
* Copyright (C) 2016 Michael Brown <mbrown@fensystems.co.uk>.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License as
* published by the Free Software Foundation; either version 2 of the
* License, or any later version.
*
* This program 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
* General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
* 02110-1301, USA.
*
* You can also choose to distribute this program under the terms of
* the Unmodified Binary Distribution Licence (as given in the file
* COPYING.UBDL), provided that you have satisfied its requirements.
*/
/** @file
*
* Optimised string operations
*
*/
FILE_LICENCE ( GPL2_OR_LATER_OR_UBDL );
#include <string.h>
/** Block size (for "ldp"/"stp") */
#define ARM64_STRING_BLKSZ 16
/**
* Copy memory area
*
* @v dest Destination address
* @v src Source address
* @v len Length
* @ret dest Destination address
*/
void arm64_memcpy ( void *dest, const void *src, size_t len ) {
size_t len_pre;
size_t len_mid;
size_t len_post;
unsigned long discard_data;
unsigned long discard_low;
unsigned long discard_high;
unsigned long discard_len;
/* Calculate pre-aligned, aligned, and post-aligned lengths.
* (Align on the destination address, on the assumption that
* misaligned stores are likely to be more expensive than
* misaligned loads.)
*/
len_pre = ( ( ARM64_STRING_BLKSZ - ( ( intptr_t ) dest ) ) &
( ARM64_STRING_BLKSZ - 1 ) );
if ( len_pre > len )
len_pre = len;
len -= len_pre;
len_mid = ( len & ~( ARM64_STRING_BLKSZ - 1 ) );
len -= len_mid;
len_post = len;
/* Copy pre-aligned section */
__asm__ __volatile__ ( "cbz %2, 2f\n\t"
"\n1:\n\t"
"ldrb %w3, [%1], #1\n\t"
"strb %w3, [%0], #1\n\t"
"sub %2, %2, #1\n\t"
"cbnz %2, 1b\n\t"
"\n2:\n\t"
: "+r" ( dest ), "+r" ( src ),
"=&r" ( discard_len ),
"=&r" ( discard_data )
: "2" ( len_pre )
: "memory" );
/* Copy aligned section */
__asm__ __volatile__ ( "cbz %2, 2f\n\t"
"\n1:\n\t"
"ldp %3, %4, [%1], #16\n\t"
"stp %3, %4, [%0], #16\n\t"
"sub %2, %2, #16\n\t"
"cbnz %2, 1b\n\t"
"\n2:\n\t"
: "+r" ( dest ), "+r" ( src ),
"=&r" ( discard_len ),
"=&r" ( discard_low ),
"=&r" ( discard_high )
: "2" ( len_mid )
: "memory" );
/* Copy post-aligned section */
__asm__ __volatile__ ( "cbz %2, 2f\n\t"
"\n1:\n\t"
"ldrb %w3, [%1], #1\n\t"
"strb %w3, [%0], #1\n\t"
"sub %2, %2, #1\n\t"
"cbnz %2, 1b\n\t"
"\n2:\n\t"
: "+r" ( dest ), "+r" ( src ),
"=&r" ( discard_len ),
"=&r" ( discard_data )
: "2" ( len_post )
: "memory" );
}
/**
* Zero memory region
*
* @v dest Destination region
* @v len Length
*/
void arm64_bzero ( void *dest, size_t len ) {
size_t len_pre;
size_t len_mid;
size_t len_post;
unsigned long discard_len;
/* Calculate pre-aligned, aligned, and post-aligned lengths */
len_pre = ( ( ARM64_STRING_BLKSZ - ( ( intptr_t ) dest ) ) &
( ARM64_STRING_BLKSZ - 1 ) );
if ( len_pre > len )
len_pre = len;
len -= len_pre;
len_mid = ( len & ~( ARM64_STRING_BLKSZ - 1 ) );
len -= len_mid;
len_post = len;
/* Zero pre-aligned section */
__asm__ __volatile__ ( "cbz %1, 2f\n\t"
"\n1:\n\t"
"strb wzr, [%0], #1\n\t"
"sub %1, %1, #1\n\t"
"cbnz %1, 1b\n\t"
"\n2:\n\t"
: "+r" ( dest ),
"=&r" ( discard_len )
: "1" ( len_pre )
: "memory" );
/* Zero aligned section */
__asm__ __volatile__ ( "cbz %1, 2f\n\t"
"\n1:\n\t"
"stp xzr, xzr, [%0], #16\n\t"
"sub %1, %1, #16\n\t"
"cbnz %1, 1b\n\t"
"\n2:\n\t"
: "+r" ( dest ),
"=&r" ( discard_len )
: "1" ( len_mid )
: "memory" );
/* Zero post-aligned section */
__asm__ __volatile__ ( "cbz %1, 2f\n\t"
"\n1:\n\t"
"strb wzr, [%0], #1\n\t"
"sub %1, %1, #1\n\t"
"cbnz %1, 1b\n\t"
"\n2:\n\t"
: "+r" ( dest ),
"=&r" ( discard_len )
: "1" ( len_post )
: "memory" );
}
/**
* Fill memory region
*
* @v dest Destination region
* @v len Length
* @v character Fill character
*
* The unusual parameter order is to allow for more efficient
* tail-calling to arm64_memset() when zeroing a region.
*/
void arm64_memset ( void *dest, size_t len, int character ) {
size_t discard_offset;
/* Use optimised zeroing code if applicable */
if ( character == 0 ) {
arm64_bzero ( dest, len );
return;
}
/* Fill one byte at a time. Calling memset() with a non-zero
* value is relatively rare and unlikely to be
* performance-critical.
*/
__asm__ __volatile__ ( "cbz %0, 2f\n\t"
"\n1:\n\t"
"sub %0, %0, #1\n\t"
"strb %w2, [%1, %0]\n\t"
"cbnz %0, 1b\n\t"
"\n2:\n\t"
: "=&r" ( discard_offset )
: "r" ( dest ), "r" ( character ), "0" ( len )
: "memory" );
}
/**
* Copy (possibly overlapping) memory region forwards
*
* @v dest Destination region
* @v src Source region
* @v len Length
*/
void arm64_memmove_forwards ( void *dest, const void *src, size_t len ) {
void *discard_dest;
const void *discard_src;
unsigned long discard_data;
/* Assume memmove() is not performance-critical, and perform a
* bytewise copy for simplicity.
*/
__asm__ __volatile__ ( "b 2f\n\t"
"\n1:\n\t"
"ldrb %w2, [%1], #1\n\t"
"strb %w2, [%0], #1\n\t"
"\n2:\n\t"
"cmp %0, %3\n\t"
"bne 1b\n\t"
: "=&r" ( discard_dest ),
"=&r" ( discard_src ),
"=&r" ( discard_data )
: "r" ( dest + len ), "0" ( dest ), "1" ( src )
: "memory" );
}
/**
* Copy (possibly overlapping) memory region backwards
*
* @v dest Destination region
* @v src Source region
* @v len Length
*/
void arm64_memmove_backwards ( void *dest, const void *src, size_t len ) {
size_t discard_offset;
unsigned long discard_data;
/* Assume memmove() is not performance-critical, and perform a
* bytewise copy for simplicity.
*/
__asm__ __volatile__ ( "cbz %0, 2f\n\t"
"\n1:\n\t"
"sub %0, %0, #1\n\t"
"ldrb %w1, [%3, %0]\n\t"
"strb %w1, [%2, %0]\n\t"
"cbnz %0, 1b\n\t"
"\n2:\n\t"
: "=&r" ( discard_offset ),
"=&r" ( discard_data )
: "r" ( dest ), "r" ( src ), "0" ( len )
: "memory" );
}
/**
* Copy (possibly overlapping) memory region
*
* @v dest Destination region
* @v src Source region
* @v len Length
*/
void arm64_memmove ( void *dest, const void *src, size_t len ) {
if ( dest <= src ) {
arm64_memmove_forwards ( dest, src, len );
} else {
arm64_memmove_backwards ( dest, src, len );
}
}
|
