1 /* SPDX-License-Identifier: GPL-2.0-only */ !! 1 /* SPDX-License-Identifier: GPL-2.0 */
2 /* !! 2 /* memcpy.S: Sparc optimized memcpy and memmove code
3 * Copyright (c) 2012-2021, Arm Limited. !! 3 * Hand optimized from GNU libc's memcpy and memmove
4 * !! 4 * Copyright (C) 1991,1996 Free Software Foundation
5 * Adapted from the original at: !! 5 * Copyright (C) 1995 Linus Torvalds (Linus.Torvalds@helsinki.fi)
6 * https://github.com/ARM-software/optimized-r !! 6 * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
>> 7 * Copyright (C) 1996 Eddie C. Dost (ecd@skynet.be)
>> 8 * Copyright (C) 1996 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
7 */ 9 */
8 10
9 #include <linux/linkage.h> !! 11 #include <linux/export.h>
10 #include <asm/assembler.h> <<
11 12
12 /* Assumptions: !! 13 #define FUNC(x) \
13 * !! 14 .globl x; \
14 * ARMv8-a, AArch64, unaligned accesses. !! 15 .type x,@function; \
15 * !! 16 .align 4; \
16 */ !! 17 x:
17 !! 18
18 #define L(label) .L ## label !! 19 /* Both these macros have to start with exactly the same insn */
19 !! 20 #define MOVE_BIGCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
20 #define dstin x0 !! 21 ldd [%src + (offset) + 0x00], %t0; \
21 #define src x1 !! 22 ldd [%src + (offset) + 0x08], %t2; \
22 #define count x2 !! 23 ldd [%src + (offset) + 0x10], %t4; \
23 #define dst x3 !! 24 ldd [%src + (offset) + 0x18], %t6; \
24 #define srcend x4 !! 25 st %t0, [%dst + (offset) + 0x00]; \
25 #define dstend x5 !! 26 st %t1, [%dst + (offset) + 0x04]; \
26 #define A_l x6 !! 27 st %t2, [%dst + (offset) + 0x08]; \
27 #define A_lw w6 !! 28 st %t3, [%dst + (offset) + 0x0c]; \
28 #define A_h x7 !! 29 st %t4, [%dst + (offset) + 0x10]; \
29 #define B_l x8 !! 30 st %t5, [%dst + (offset) + 0x14]; \
30 #define B_lw w8 !! 31 st %t6, [%dst + (offset) + 0x18]; \
31 #define B_h x9 !! 32 st %t7, [%dst + (offset) + 0x1c];
32 #define C_l x10 !! 33
33 #define C_lw w10 !! 34 #define MOVE_BIGALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
34 #define C_h x11 !! 35 ldd [%src + (offset) + 0x00], %t0; \
35 #define D_l x12 !! 36 ldd [%src + (offset) + 0x08], %t2; \
36 #define D_h x13 !! 37 ldd [%src + (offset) + 0x10], %t4; \
37 #define E_l x14 !! 38 ldd [%src + (offset) + 0x18], %t6; \
38 #define E_h x15 !! 39 std %t0, [%dst + (offset) + 0x00]; \
39 #define F_l x16 !! 40 std %t2, [%dst + (offset) + 0x08]; \
40 #define F_h x17 !! 41 std %t4, [%dst + (offset) + 0x10]; \
41 #define G_l count !! 42 std %t6, [%dst + (offset) + 0x18];
42 #define G_h dst !! 43
43 #define H_l src !! 44 #define MOVE_LASTCHUNK(src, dst, offset, t0, t1, t2, t3) \
44 #define H_h srcend !! 45 ldd [%src - (offset) - 0x10], %t0; \
45 #define tmp1 x14 !! 46 ldd [%src - (offset) - 0x08], %t2; \
46 !! 47 st %t0, [%dst - (offset) - 0x10]; \
47 /* This implementation handles overlaps and su !! 48 st %t1, [%dst - (offset) - 0x0c]; \
48 from a single entry point. It uses unalign !! 49 st %t2, [%dst - (offset) - 0x08]; \
49 sequences to keep the code small, simple an !! 50 st %t3, [%dst - (offset) - 0x04];
50 !! 51
51 Copies are split into 3 main cases: small c !! 52 #define MOVE_LASTALIGNCHUNK(src, dst, offset, t0, t1, t2, t3) \
52 copies of up to 128 bytes, and large copies !! 53 ldd [%src - (offset) - 0x10], %t0; \
53 check is negligible since it is only requir !! 54 ldd [%src - (offset) - 0x08], %t2; \
54 !! 55 std %t0, [%dst - (offset) - 0x10]; \
55 Large copies use a software pipelined loop !! 56 std %t2, [%dst - (offset) - 0x08];
56 The destination pointer is 16-byte aligned !! 57
57 The loop tail is handled by always copying !! 58 #define MOVE_SHORTCHUNK(src, dst, offset, t0, t1) \
58 */ !! 59 ldub [%src - (offset) - 0x02], %t0; \
59 !! 60 ldub [%src - (offset) - 0x01], %t1; \
60 SYM_FUNC_START(__pi_memcpy) !! 61 stb %t0, [%dst - (offset) - 0x02]; \
61 add srcend, src, count !! 62 stb %t1, [%dst - (offset) - 0x01];
62 add dstend, dstin, count <<
63 cmp count, 128 <<
64 b.hi L(copy_long) <<
65 cmp count, 32 <<
66 b.hi L(copy32_128) <<
67 <<
68 /* Small copies: 0..32 bytes. */ <<
69 cmp count, 16 <<
70 b.lo L(copy16) <<
71 ldp A_l, A_h, [src] <<
72 ldp D_l, D_h, [srcend, -16] <<
73 stp A_l, A_h, [dstin] <<
74 stp D_l, D_h, [dstend, -16] <<
75 ret <<
76 <<
77 /* Copy 8-15 bytes. */ <<
78 L(copy16): <<
79 tbz count, 3, L(copy8) <<
80 ldr A_l, [src] <<
81 ldr A_h, [srcend, -8] <<
82 str A_l, [dstin] <<
83 str A_h, [dstend, -8] <<
84 ret <<
85 63
86 .p2align 3 !! 64 .text
87 /* Copy 4-7 bytes. */ !! 65 .align 4
88 L(copy8): <<
89 tbz count, 2, L(copy4) <<
90 ldr A_lw, [src] <<
91 ldr B_lw, [srcend, -4] <<
92 str A_lw, [dstin] <<
93 str B_lw, [dstend, -4] <<
94 ret <<
95 66
96 /* Copy 0..3 bytes using a branchless !! 67 FUNC(memmove)
97 L(copy4): !! 68 EXPORT_SYMBOL(memmove)
98 cbz count, L(copy0) !! 69 cmp %o0, %o1
99 lsr tmp1, count, 1 !! 70 mov %o0, %g7
100 ldrb A_lw, [src] !! 71 bleu 9f
101 ldrb C_lw, [srcend, -1] !! 72 sub %o0, %o1, %o4
102 ldrb B_lw, [src, tmp1] !! 73
103 strb A_lw, [dstin] !! 74 add %o1, %o2, %o3
104 strb B_lw, [dstin, tmp1] !! 75 cmp %o3, %o0
105 strb C_lw, [dstend, -1] !! 76 bleu 0f
106 L(copy0): !! 77 andcc %o4, 3, %o5
107 ret !! 78
108 !! 79 add %o1, %o2, %o1
109 .p2align 4 !! 80 add %o0, %o2, %o0
110 /* Medium copies: 33..128 bytes. */ !! 81 sub %o1, 1, %o1
111 L(copy32_128): !! 82 sub %o0, 1, %o0
112 ldp A_l, A_h, [src] !! 83
113 ldp B_l, B_h, [src, 16] !! 84 1: /* reverse_bytes */
114 ldp C_l, C_h, [srcend, -32] !! 85
115 ldp D_l, D_h, [srcend, -16] !! 86 ldub [%o1], %o4
116 cmp count, 64 !! 87 subcc %o2, 1, %o2
117 b.hi L(copy128) !! 88 stb %o4, [%o0]
118 stp A_l, A_h, [dstin] !! 89 sub %o1, 1, %o1
119 stp B_l, B_h, [dstin, 16] !! 90 bne 1b
120 stp C_l, C_h, [dstend, -32] !! 91 sub %o0, 1, %o0
121 stp D_l, D_h, [dstend, -16] !! 92
122 ret !! 93 retl
123 !! 94 mov %g7, %o0
124 .p2align 4 !! 95
125 /* Copy 65..128 bytes. */ !! 96 /* NOTE: This code is executed just for the cases,
126 L(copy128): !! 97 where %src (=%o1) & 3 is != 0.
127 ldp E_l, E_h, [src, 32] !! 98 We need to align it to 4. So, for (%src & 3)
128 ldp F_l, F_h, [src, 48] !! 99 1 we need to do ldub,lduh
129 cmp count, 96 !! 100 2 lduh
130 b.ls L(copy96) !! 101 3 just ldub
131 ldp G_l, G_h, [srcend, -64] !! 102 so even if it looks weird, the branches
132 ldp H_l, H_h, [srcend, -48] !! 103 are correct here. -jj
133 stp G_l, G_h, [dstend, -64] !! 104 */
134 stp H_l, H_h, [dstend, -48] !! 105 78: /* dword_align */
135 L(copy96): <<
136 stp A_l, A_h, [dstin] <<
137 stp B_l, B_h, [dstin, 16] <<
138 stp E_l, E_h, [dstin, 32] <<
139 stp F_l, F_h, [dstin, 48] <<
140 stp C_l, C_h, [dstend, -32] <<
141 stp D_l, D_h, [dstend, -16] <<
142 ret <<
143 106
144 .p2align 4 !! 107 andcc %o1, 1, %g0
145 /* Copy more than 128 bytes. */ !! 108 be 4f
146 L(copy_long): !! 109 andcc %o1, 2, %g0
147 /* Use backwards copy if there is an o !! 110
148 sub tmp1, dstin, src !! 111 ldub [%o1], %g2
149 cbz tmp1, L(copy0) !! 112 add %o1, 1, %o1
150 cmp tmp1, count !! 113 stb %g2, [%o0]
151 b.lo L(copy_long_backwards) !! 114 sub %o2, 1, %o2
152 !! 115 bne 3f
153 /* Copy 16 bytes and then align dst to !! 116 add %o0, 1, %o0
154 !! 117 4:
155 ldp D_l, D_h, [src] !! 118 lduh [%o1], %g2
156 and tmp1, dstin, 15 !! 119 add %o1, 2, %o1
157 bic dst, dstin, 15 !! 120 sth %g2, [%o0]
158 sub src, src, tmp1 !! 121 sub %o2, 2, %o2
159 add count, count, tmp1 /* Cou !! 122 b 3f
160 ldp A_l, A_h, [src, 16] !! 123 add %o0, 2, %o0
161 stp D_l, D_h, [dstin] <<
162 ldp B_l, B_h, [src, 32] <<
163 ldp C_l, C_h, [src, 48] <<
164 ldp D_l, D_h, [src, 64]! <<
165 subs count, count, 128 + 16 /* Tes <<
166 b.ls L(copy64_from_end) <<
167 <<
168 L(loop64): <<
169 stp A_l, A_h, [dst, 16] <<
170 ldp A_l, A_h, [src, 16] <<
171 stp B_l, B_h, [dst, 32] <<
172 ldp B_l, B_h, [src, 32] <<
173 stp C_l, C_h, [dst, 48] <<
174 ldp C_l, C_h, [src, 48] <<
175 stp D_l, D_h, [dst, 64]! <<
176 ldp D_l, D_h, [src, 64]! <<
177 subs count, count, 64 <<
178 b.hi L(loop64) <<
179 <<
180 /* Write the last iteration and copy 6 <<
181 L(copy64_from_end): <<
182 ldp E_l, E_h, [srcend, -64] <<
183 stp A_l, A_h, [dst, 16] <<
184 ldp A_l, A_h, [srcend, -48] <<
185 stp B_l, B_h, [dst, 32] <<
186 ldp B_l, B_h, [srcend, -32] <<
187 stp C_l, C_h, [dst, 48] <<
188 ldp C_l, C_h, [srcend, -16] <<
189 stp D_l, D_h, [dst, 64] <<
190 stp E_l, E_h, [dstend, -64] <<
191 stp A_l, A_h, [dstend, -48] <<
192 stp B_l, B_h, [dstend, -32] <<
193 stp C_l, C_h, [dstend, -16] <<
194 ret <<
195 124
196 .p2align 4 !! 125 FUNC(memcpy) /* %o0=dst %o1=src %o2=len */
>> 126 EXPORT_SYMBOL(memcpy)
197 127
198 /* Large backwards copy for overlappin !! 128 sub %o0, %o1, %o4
199 Copy 16 bytes and then align dst to !! 129 mov %o0, %g7
200 L(copy_long_backwards): !! 130 9:
201 ldp D_l, D_h, [srcend, -16] !! 131 andcc %o4, 3, %o5
202 and tmp1, dstend, 15 !! 132 0:
203 sub srcend, srcend, tmp1 !! 133 bne 86f
204 sub count, count, tmp1 !! 134 cmp %o2, 15
205 ldp A_l, A_h, [srcend, -16] !! 135
206 stp D_l, D_h, [dstend, -16] !! 136 bleu 90f
207 ldp B_l, B_h, [srcend, -32] !! 137 andcc %o1, 3, %g0
208 ldp C_l, C_h, [srcend, -48] !! 138
209 ldp D_l, D_h, [srcend, -64]! !! 139 bne 78b
210 sub dstend, dstend, tmp1 !! 140 3:
211 subs count, count, 128 !! 141 andcc %o1, 4, %g0
212 b.ls L(copy64_from_start) !! 142
213 !! 143 be 2f
214 L(loop64_backwards): !! 144 mov %o2, %g1
215 stp A_l, A_h, [dstend, -16] !! 145
216 ldp A_l, A_h, [srcend, -16] !! 146 ld [%o1], %o4
217 stp B_l, B_h, [dstend, -32] !! 147 sub %g1, 4, %g1
218 ldp B_l, B_h, [srcend, -32] !! 148 st %o4, [%o0]
219 stp C_l, C_h, [dstend, -48] !! 149 add %o1, 4, %o1
220 ldp C_l, C_h, [srcend, -48] !! 150 add %o0, 4, %o0
221 stp D_l, D_h, [dstend, -64]! !! 151 2:
222 ldp D_l, D_h, [srcend, -64]! !! 152 andcc %g1, 0xffffff80, %g0
223 subs count, count, 64 !! 153 be 3f
224 b.hi L(loop64_backwards) !! 154 andcc %o0, 4, %g0
225 !! 155
226 /* Write the last iteration and copy 6 !! 156 be 82f + 4
227 L(copy64_from_start): !! 157 5:
228 ldp G_l, G_h, [src, 48] !! 158 MOVE_BIGCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
229 stp A_l, A_h, [dstend, -16] !! 159 MOVE_BIGCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
230 ldp A_l, A_h, [src, 32] !! 160 MOVE_BIGCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
231 stp B_l, B_h, [dstend, -32] !! 161 MOVE_BIGCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
232 ldp B_l, B_h, [src, 16] !! 162 sub %g1, 128, %g1
233 stp C_l, C_h, [dstend, -48] !! 163 add %o1, 128, %o1
234 ldp C_l, C_h, [src] !! 164 cmp %g1, 128
235 stp D_l, D_h, [dstend, -64] !! 165 bge 5b
236 stp G_l, G_h, [dstin, 48] !! 166 add %o0, 128, %o0
237 stp A_l, A_h, [dstin, 32] !! 167 3:
238 stp B_l, B_h, [dstin, 16] !! 168 andcc %g1, 0x70, %g4
239 stp C_l, C_h, [dstin] !! 169 be 80f
>> 170 andcc %g1, 8, %g0
>> 171
>> 172 sethi %hi(80f), %o5
>> 173 srl %g4, 1, %o4
>> 174 add %g4, %o4, %o4
>> 175 add %o1, %g4, %o1
>> 176 sub %o5, %o4, %o5
>> 177 jmpl %o5 + %lo(80f), %g0
>> 178 add %o0, %g4, %o0
>> 179
>> 180 79: /* memcpy_table */
>> 181
>> 182 MOVE_LASTCHUNK(o1, o0, 0x60, g2, g3, g4, g5)
>> 183 MOVE_LASTCHUNK(o1, o0, 0x50, g2, g3, g4, g5)
>> 184 MOVE_LASTCHUNK(o1, o0, 0x40, g2, g3, g4, g5)
>> 185 MOVE_LASTCHUNK(o1, o0, 0x30, g2, g3, g4, g5)
>> 186 MOVE_LASTCHUNK(o1, o0, 0x20, g2, g3, g4, g5)
>> 187 MOVE_LASTCHUNK(o1, o0, 0x10, g2, g3, g4, g5)
>> 188 MOVE_LASTCHUNK(o1, o0, 0x00, g2, g3, g4, g5)
>> 189
>> 190 80: /* memcpy_table_end */
>> 191 be 81f
>> 192 andcc %g1, 4, %g0
>> 193
>> 194 ldd [%o1], %g2
>> 195 add %o0, 8, %o0
>> 196 st %g2, [%o0 - 0x08]
>> 197 add %o1, 8, %o1
>> 198 st %g3, [%o0 - 0x04]
>> 199
>> 200 81: /* memcpy_last7 */
>> 201
>> 202 be 1f
>> 203 andcc %g1, 2, %g0
>> 204
>> 205 ld [%o1], %g2
>> 206 add %o1, 4, %o1
>> 207 st %g2, [%o0]
>> 208 add %o0, 4, %o0
>> 209 1:
>> 210 be 1f
>> 211 andcc %g1, 1, %g0
>> 212
>> 213 lduh [%o1], %g2
>> 214 add %o1, 2, %o1
>> 215 sth %g2, [%o0]
>> 216 add %o0, 2, %o0
>> 217 1:
>> 218 be 1f
>> 219 nop
>> 220
>> 221 ldub [%o1], %g2
>> 222 stb %g2, [%o0]
>> 223 1:
>> 224 retl
>> 225 mov %g7, %o0
>> 226
>> 227 82: /* ldd_std */
>> 228 MOVE_BIGALIGNCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
>> 229 MOVE_BIGALIGNCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
>> 230 MOVE_BIGALIGNCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
>> 231 MOVE_BIGALIGNCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
>> 232 subcc %g1, 128, %g1
>> 233 add %o1, 128, %o1
>> 234 cmp %g1, 128
>> 235 bge 82b
>> 236 add %o0, 128, %o0
>> 237
>> 238 andcc %g1, 0x70, %g4
>> 239 be 84f
>> 240 andcc %g1, 8, %g0
>> 241
>> 242 sethi %hi(84f), %o5
>> 243 add %o1, %g4, %o1
>> 244 sub %o5, %g4, %o5
>> 245 jmpl %o5 + %lo(84f), %g0
>> 246 add %o0, %g4, %o0
>> 247
>> 248 83: /* amemcpy_table */
>> 249
>> 250 MOVE_LASTALIGNCHUNK(o1, o0, 0x60, g2, g3, g4, g5)
>> 251 MOVE_LASTALIGNCHUNK(o1, o0, 0x50, g2, g3, g4, g5)
>> 252 MOVE_LASTALIGNCHUNK(o1, o0, 0x40, g2, g3, g4, g5)
>> 253 MOVE_LASTALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5)
>> 254 MOVE_LASTALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5)
>> 255 MOVE_LASTALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5)
>> 256 MOVE_LASTALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5)
>> 257
>> 258 84: /* amemcpy_table_end */
>> 259 be 85f
>> 260 andcc %g1, 4, %g0
>> 261
>> 262 ldd [%o1], %g2
>> 263 add %o0, 8, %o0
>> 264 std %g2, [%o0 - 0x08]
>> 265 add %o1, 8, %o1
>> 266 85: /* amemcpy_last7 */
>> 267 be 1f
>> 268 andcc %g1, 2, %g0
>> 269
>> 270 ld [%o1], %g2
>> 271 add %o1, 4, %o1
>> 272 st %g2, [%o0]
>> 273 add %o0, 4, %o0
>> 274 1:
>> 275 be 1f
>> 276 andcc %g1, 1, %g0
>> 277
>> 278 lduh [%o1], %g2
>> 279 add %o1, 2, %o1
>> 280 sth %g2, [%o0]
>> 281 add %o0, 2, %o0
>> 282 1:
>> 283 be 1f
>> 284 nop
>> 285
>> 286 ldub [%o1], %g2
>> 287 stb %g2, [%o0]
>> 288 1:
>> 289 retl
>> 290 mov %g7, %o0
>> 291
>> 292 86: /* non_aligned */
>> 293 cmp %o2, 6
>> 294 bleu 88f
>> 295 nop
>> 296
>> 297 save %sp, -96, %sp
>> 298 andcc %i0, 3, %g0
>> 299 be 61f
>> 300 andcc %i0, 1, %g0
>> 301 be 60f
>> 302 andcc %i0, 2, %g0
>> 303
>> 304 ldub [%i1], %g5
>> 305 add %i1, 1, %i1
>> 306 stb %g5, [%i0]
>> 307 sub %i2, 1, %i2
>> 308 bne 61f
>> 309 add %i0, 1, %i0
>> 310 60:
>> 311 ldub [%i1], %g3
>> 312 add %i1, 2, %i1
>> 313 stb %g3, [%i0]
>> 314 sub %i2, 2, %i2
>> 315 ldub [%i1 - 1], %g3
>> 316 add %i0, 2, %i0
>> 317 stb %g3, [%i0 - 1]
>> 318 61:
>> 319 and %i1, 3, %g2
>> 320 and %i2, 0xc, %g3
>> 321 and %i1, -4, %i1
>> 322 cmp %g3, 4
>> 323 sll %g2, 3, %g4
>> 324 mov 32, %g2
>> 325 be 4f
>> 326 sub %g2, %g4, %l0
>> 327
>> 328 blu 3f
>> 329 cmp %g3, 0x8
>> 330
>> 331 be 2f
>> 332 srl %i2, 2, %g3
>> 333
>> 334 ld [%i1], %i3
>> 335 add %i0, -8, %i0
>> 336 ld [%i1 + 4], %i4
>> 337 b 8f
>> 338 add %g3, 1, %g3
>> 339 2:
>> 340 ld [%i1], %i4
>> 341 add %i0, -12, %i0
>> 342 ld [%i1 + 4], %i5
>> 343 add %g3, 2, %g3
>> 344 b 9f
>> 345 add %i1, -4, %i1
>> 346 3:
>> 347 ld [%i1], %g1
>> 348 add %i0, -4, %i0
>> 349 ld [%i1 + 4], %i3
>> 350 srl %i2, 2, %g3
>> 351 b 7f
>> 352 add %i1, 4, %i1
>> 353 4:
>> 354 ld [%i1], %i5
>> 355 cmp %i2, 7
>> 356 ld [%i1 + 4], %g1
>> 357 srl %i2, 2, %g3
>> 358 bleu 10f
>> 359 add %i1, 8, %i1
>> 360
>> 361 ld [%i1], %i3
>> 362 add %g3, -1, %g3
>> 363 5:
>> 364 sll %i5, %g4, %g2
>> 365 srl %g1, %l0, %g5
>> 366 or %g2, %g5, %g2
>> 367 st %g2, [%i0]
>> 368 7:
>> 369 ld [%i1 + 4], %i4
>> 370 sll %g1, %g4, %g2
>> 371 srl %i3, %l0, %g5
>> 372 or %g2, %g5, %g2
>> 373 st %g2, [%i0 + 4]
>> 374 8:
>> 375 ld [%i1 + 8], %i5
>> 376 sll %i3, %g4, %g2
>> 377 srl %i4, %l0, %g5
>> 378 or %g2, %g5, %g2
>> 379 st %g2, [%i0 + 8]
>> 380 9:
>> 381 ld [%i1 + 12], %g1
>> 382 sll %i4, %g4, %g2
>> 383 srl %i5, %l0, %g5
>> 384 addcc %g3, -4, %g3
>> 385 or %g2, %g5, %g2
>> 386 add %i1, 16, %i1
>> 387 st %g2, [%i0 + 12]
>> 388 add %i0, 16, %i0
>> 389 bne,a 5b
>> 390 ld [%i1], %i3
>> 391 10:
>> 392 sll %i5, %g4, %g2
>> 393 srl %g1, %l0, %g5
>> 394 srl %l0, 3, %g3
>> 395 or %g2, %g5, %g2
>> 396 sub %i1, %g3, %i1
>> 397 andcc %i2, 2, %g0
>> 398 st %g2, [%i0]
>> 399 be 1f
>> 400 andcc %i2, 1, %g0
>> 401
>> 402 ldub [%i1], %g2
>> 403 add %i1, 2, %i1
>> 404 stb %g2, [%i0 + 4]
>> 405 add %i0, 2, %i0
>> 406 ldub [%i1 - 1], %g2
>> 407 stb %g2, [%i0 + 3]
>> 408 1:
>> 409 be 1f
>> 410 nop
>> 411 ldub [%i1], %g2
>> 412 stb %g2, [%i0 + 4]
>> 413 1:
240 ret 414 ret
241 SYM_FUNC_END(__pi_memcpy) !! 415 restore %g7, %g0, %o0
242 <<
243 SYM_FUNC_ALIAS(__memcpy, __pi_memcpy) <<
244 EXPORT_SYMBOL(__memcpy) <<
245 SYM_FUNC_ALIAS_WEAK(memcpy, __memcpy) <<
246 EXPORT_SYMBOL(memcpy) <<
247 416
248 SYM_FUNC_ALIAS(__pi_memmove, __pi_memcpy) !! 417 88: /* short_end */
249 418
250 SYM_FUNC_ALIAS(__memmove, __pi_memmove) !! 419 and %o2, 0xe, %o3
251 EXPORT_SYMBOL(__memmove) !! 420 20:
252 SYM_FUNC_ALIAS_WEAK(memmove, __memmove) !! 421 sethi %hi(89f), %o5
253 EXPORT_SYMBOL(memmove) !! 422 sll %o3, 3, %o4
>> 423 add %o0, %o3, %o0
>> 424 sub %o5, %o4, %o5
>> 425 add %o1, %o3, %o1
>> 426 jmpl %o5 + %lo(89f), %g0
>> 427 andcc %o2, 1, %g0
>> 428
>> 429 MOVE_SHORTCHUNK(o1, o0, 0x0c, g2, g3)
>> 430 MOVE_SHORTCHUNK(o1, o0, 0x0a, g2, g3)
>> 431 MOVE_SHORTCHUNK(o1, o0, 0x08, g2, g3)
>> 432 MOVE_SHORTCHUNK(o1, o0, 0x06, g2, g3)
>> 433 MOVE_SHORTCHUNK(o1, o0, 0x04, g2, g3)
>> 434 MOVE_SHORTCHUNK(o1, o0, 0x02, g2, g3)
>> 435 MOVE_SHORTCHUNK(o1, o0, 0x00, g2, g3)
>> 436
>> 437 89: /* short_table_end */
>> 438
>> 439 be 1f
>> 440 nop
>> 441
>> 442 ldub [%o1], %g2
>> 443 stb %g2, [%o0]
>> 444 1:
>> 445 retl
>> 446 mov %g7, %o0
>> 447
>> 448 90: /* short_aligned_end */
>> 449 bne 88b
>> 450 andcc %o2, 8, %g0
>> 451
>> 452 be 1f
>> 453 andcc %o2, 4, %g0
>> 454
>> 455 ld [%o1 + 0x00], %g2
>> 456 ld [%o1 + 0x04], %g3
>> 457 add %o1, 8, %o1
>> 458 st %g2, [%o0 + 0x00]
>> 459 st %g3, [%o0 + 0x04]
>> 460 add %o0, 8, %o0
>> 461 1:
>> 462 b 81b
>> 463 mov %o2, %g1
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.