1 /* SPDX-License-Identifier: GPL-2.0-only */ !! 1 /* memcpy.S: Sparc optimized memcpy and memmove code
2 /* !! 2 * Hand optimized from GNU libc's memcpy and memmove
3 * Copyright (c) 2012-2021, Arm Limited. !! 3 * Copyright (C) 1991,1996 Free Software Foundation
4 * !! 4 * Copyright (C) 1995 Linus Torvalds (Linus.Torvalds@helsinki.fi)
5 * Adapted from the original at: !! 5 * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
6 * https://github.com/ARM-software/optimized-r !! 6 * Copyright (C) 1996 Eddie C. Dost (ecd@skynet.be)
>> 7 * Copyright (C) 1996 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
7 */ 8 */
8 9
9 #include <linux/linkage.h> !! 10 #ifdef __KERNEL__
10 #include <asm/assembler.h> <<
11 11
12 /* Assumptions: !! 12 #define FUNC(x) \
13 * !! 13 .globl x; \
14 * ARMv8-a, AArch64, unaligned accesses. !! 14 .type x,@function; \
15 * !! 15 .align 4; \
>> 16 x:
>> 17
>> 18 #undef FASTER_REVERSE
>> 19 #undef FASTER_NONALIGNED
>> 20 #define FASTER_ALIGNED
>> 21
>> 22 /* In kernel these functions don't return a value.
>> 23 * One should use macros in asm/string.h for that purpose.
>> 24 * We return 0, so that bugs are more apparent.
16 */ 25 */
>> 26 #define SETUP_RETL
>> 27 #define RETL_INSN clr %o0
17 28
18 #define L(label) .L ## label !! 29 #else
19 30
20 #define dstin x0 !! 31 /* libc */
21 #define src x1 !! 32
22 #define count x2 !! 33 #include "DEFS.h"
23 #define dst x3 !! 34
24 #define srcend x4 !! 35 #define FASTER_REVERSE
25 #define dstend x5 !! 36 #define FASTER_NONALIGNED
26 #define A_l x6 !! 37 #define FASTER_ALIGNED
27 #define A_lw w6 !! 38
28 #define A_h x7 !! 39 #define SETUP_RETL mov %o0, %g6
29 #define B_l x8 !! 40 #define RETL_INSN mov %g6, %o0
30 #define B_lw w8 !! 41
31 #define B_h x9 !! 42 #endif
32 #define C_l x10 !! 43
33 #define C_lw w10 !! 44 /* Both these macros have to start with exactly the same insn */
34 #define C_h x11 !! 45 #define MOVE_BIGCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
35 #define D_l x12 !! 46 ldd [%src + (offset) + 0x00], %t0; \
36 #define D_h x13 !! 47 ldd [%src + (offset) + 0x08], %t2; \
37 #define E_l x14 !! 48 ldd [%src + (offset) + 0x10], %t4; \
38 #define E_h x15 !! 49 ldd [%src + (offset) + 0x18], %t6; \
39 #define F_l x16 !! 50 st %t0, [%dst + (offset) + 0x00]; \
40 #define F_h x17 !! 51 st %t1, [%dst + (offset) + 0x04]; \
41 #define G_l count !! 52 st %t2, [%dst + (offset) + 0x08]; \
42 #define G_h dst !! 53 st %t3, [%dst + (offset) + 0x0c]; \
43 #define H_l src !! 54 st %t4, [%dst + (offset) + 0x10]; \
44 #define H_h srcend !! 55 st %t5, [%dst + (offset) + 0x14]; \
45 #define tmp1 x14 !! 56 st %t6, [%dst + (offset) + 0x18]; \
46 !! 57 st %t7, [%dst + (offset) + 0x1c];
47 /* This implementation handles overlaps and su !! 58
48 from a single entry point. It uses unalign !! 59 #define MOVE_BIGALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
49 sequences to keep the code small, simple an !! 60 ldd [%src + (offset) + 0x00], %t0; \
50 !! 61 ldd [%src + (offset) + 0x08], %t2; \
51 Copies are split into 3 main cases: small c !! 62 ldd [%src + (offset) + 0x10], %t4; \
52 copies of up to 128 bytes, and large copies !! 63 ldd [%src + (offset) + 0x18], %t6; \
53 check is negligible since it is only requir !! 64 std %t0, [%dst + (offset) + 0x00]; \
54 !! 65 std %t2, [%dst + (offset) + 0x08]; \
55 Large copies use a software pipelined loop !! 66 std %t4, [%dst + (offset) + 0x10]; \
56 The destination pointer is 16-byte aligned !! 67 std %t6, [%dst + (offset) + 0x18];
57 The loop tail is handled by always copying !! 68
58 */ !! 69 #define MOVE_LASTCHUNK(src, dst, offset, t0, t1, t2, t3) \
59 !! 70 ldd [%src - (offset) - 0x10], %t0; \
60 SYM_FUNC_START(__pi_memcpy) !! 71 ldd [%src - (offset) - 0x08], %t2; \
61 add srcend, src, count !! 72 st %t0, [%dst - (offset) - 0x10]; \
62 add dstend, dstin, count !! 73 st %t1, [%dst - (offset) - 0x0c]; \
63 cmp count, 128 !! 74 st %t2, [%dst - (offset) - 0x08]; \
64 b.hi L(copy_long) !! 75 st %t3, [%dst - (offset) - 0x04];
65 cmp count, 32 !! 76
66 b.hi L(copy32_128) !! 77 #define MOVE_LASTALIGNCHUNK(src, dst, offset, t0, t1, t2, t3) \
67 !! 78 ldd [%src - (offset) - 0x10], %t0; \
68 /* Small copies: 0..32 bytes. */ !! 79 ldd [%src - (offset) - 0x08], %t2; \
69 cmp count, 16 !! 80 std %t0, [%dst - (offset) - 0x10]; \
70 b.lo L(copy16) !! 81 std %t2, [%dst - (offset) - 0x08];
71 ldp A_l, A_h, [src] !! 82
72 ldp D_l, D_h, [srcend, -16] !! 83 #define MOVE_SHORTCHUNK(src, dst, offset, t0, t1) \
73 stp A_l, A_h, [dstin] !! 84 ldub [%src - (offset) - 0x02], %t0; \
74 stp D_l, D_h, [dstend, -16] !! 85 ldub [%src - (offset) - 0x01], %t1; \
75 ret !! 86 stb %t0, [%dst - (offset) - 0x02]; \
76 !! 87 stb %t1, [%dst - (offset) - 0x01];
77 /* Copy 8-15 bytes. */ !! 88
78 L(copy16): !! 89 /* Both these macros have to start with exactly the same insn */
79 tbz count, 3, L(copy8) !! 90 #define RMOVE_BIGCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
80 ldr A_l, [src] !! 91 ldd [%src - (offset) - 0x20], %t0; \
81 ldr A_h, [srcend, -8] !! 92 ldd [%src - (offset) - 0x18], %t2; \
82 str A_l, [dstin] !! 93 ldd [%src - (offset) - 0x10], %t4; \
83 str A_h, [dstend, -8] !! 94 ldd [%src - (offset) - 0x08], %t6; \
84 ret !! 95 st %t0, [%dst - (offset) - 0x20]; \
85 !! 96 st %t1, [%dst - (offset) - 0x1c]; \
86 .p2align 3 !! 97 st %t2, [%dst - (offset) - 0x18]; \
87 /* Copy 4-7 bytes. */ !! 98 st %t3, [%dst - (offset) - 0x14]; \
88 L(copy8): !! 99 st %t4, [%dst - (offset) - 0x10]; \
89 tbz count, 2, L(copy4) !! 100 st %t5, [%dst - (offset) - 0x0c]; \
90 ldr A_lw, [src] !! 101 st %t6, [%dst - (offset) - 0x08]; \
91 ldr B_lw, [srcend, -4] !! 102 st %t7, [%dst - (offset) - 0x04];
92 str A_lw, [dstin] !! 103
93 str B_lw, [dstend, -4] !! 104 #define RMOVE_BIGALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
94 ret !! 105 ldd [%src - (offset) - 0x20], %t0; \
95 !! 106 ldd [%src - (offset) - 0x18], %t2; \
96 /* Copy 0..3 bytes using a branchless !! 107 ldd [%src - (offset) - 0x10], %t4; \
97 L(copy4): !! 108 ldd [%src - (offset) - 0x08], %t6; \
98 cbz count, L(copy0) !! 109 std %t0, [%dst - (offset) - 0x20]; \
99 lsr tmp1, count, 1 !! 110 std %t2, [%dst - (offset) - 0x18]; \
100 ldrb A_lw, [src] !! 111 std %t4, [%dst - (offset) - 0x10]; \
101 ldrb C_lw, [srcend, -1] !! 112 std %t6, [%dst - (offset) - 0x08];
102 ldrb B_lw, [src, tmp1] !! 113
103 strb A_lw, [dstin] !! 114 #define RMOVE_LASTCHUNK(src, dst, offset, t0, t1, t2, t3) \
104 strb B_lw, [dstin, tmp1] !! 115 ldd [%src + (offset) + 0x00], %t0; \
105 strb C_lw, [dstend, -1] !! 116 ldd [%src + (offset) + 0x08], %t2; \
106 L(copy0): !! 117 st %t0, [%dst + (offset) + 0x00]; \
107 ret !! 118 st %t1, [%dst + (offset) + 0x04]; \
108 !! 119 st %t2, [%dst + (offset) + 0x08]; \
109 .p2align 4 !! 120 st %t3, [%dst + (offset) + 0x0c];
110 /* Medium copies: 33..128 bytes. */ !! 121
111 L(copy32_128): !! 122 #define RMOVE_SHORTCHUNK(src, dst, offset, t0, t1) \
112 ldp A_l, A_h, [src] !! 123 ldub [%src + (offset) + 0x00], %t0; \
113 ldp B_l, B_h, [src, 16] !! 124 ldub [%src + (offset) + 0x01], %t1; \
114 ldp C_l, C_h, [srcend, -32] !! 125 stb %t0, [%dst + (offset) + 0x00]; \
115 ldp D_l, D_h, [srcend, -16] !! 126 stb %t1, [%dst + (offset) + 0x01];
116 cmp count, 64 !! 127
117 b.hi L(copy128) !! 128 #define SMOVE_CHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, prev, shil, shir, offset2) \
118 stp A_l, A_h, [dstin] !! 129 ldd [%src + (offset) + 0x00], %t0; \
119 stp B_l, B_h, [dstin, 16] !! 130 ldd [%src + (offset) + 0x08], %t2; \
120 stp C_l, C_h, [dstend, -32] !! 131 srl %t0, shir, %t5; \
121 stp D_l, D_h, [dstend, -16] !! 132 srl %t1, shir, %t6; \
122 ret !! 133 sll %t0, shil, %t0; \
123 !! 134 or %t5, %prev, %t5; \
124 .p2align 4 !! 135 sll %t1, shil, %prev; \
125 /* Copy 65..128 bytes. */ !! 136 or %t6, %t0, %t0; \
126 L(copy128): !! 137 srl %t2, shir, %t1; \
127 ldp E_l, E_h, [src, 32] !! 138 srl %t3, shir, %t6; \
128 ldp F_l, F_h, [src, 48] !! 139 sll %t2, shil, %t2; \
129 cmp count, 96 !! 140 or %t1, %prev, %t1; \
130 b.ls L(copy96) !! 141 std %t4, [%dst + (offset) + (offset2) - 0x04]; \
131 ldp G_l, G_h, [srcend, -64] !! 142 std %t0, [%dst + (offset) + (offset2) + 0x04]; \
132 ldp H_l, H_h, [srcend, -48] !! 143 sll %t3, shil, %prev; \
133 stp G_l, G_h, [dstend, -64] !! 144 or %t6, %t2, %t4;
134 stp H_l, H_h, [dstend, -48] !! 145
135 L(copy96): !! 146 #define SMOVE_ALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, prev, shil, shir, offset2) \
136 stp A_l, A_h, [dstin] !! 147 ldd [%src + (offset) + 0x00], %t0; \
137 stp B_l, B_h, [dstin, 16] !! 148 ldd [%src + (offset) + 0x08], %t2; \
138 stp E_l, E_h, [dstin, 32] !! 149 srl %t0, shir, %t4; \
139 stp F_l, F_h, [dstin, 48] !! 150 srl %t1, shir, %t5; \
140 stp C_l, C_h, [dstend, -32] !! 151 sll %t0, shil, %t6; \
141 stp D_l, D_h, [dstend, -16] !! 152 or %t4, %prev, %t0; \
142 ret !! 153 sll %t1, shil, %prev; \
143 !! 154 or %t5, %t6, %t1; \
144 .p2align 4 !! 155 srl %t2, shir, %t4; \
145 /* Copy more than 128 bytes. */ !! 156 srl %t3, shir, %t5; \
146 L(copy_long): !! 157 sll %t2, shil, %t6; \
147 /* Use backwards copy if there is an o !! 158 or %t4, %prev, %t2; \
148 sub tmp1, dstin, src !! 159 sll %t3, shil, %prev; \
149 cbz tmp1, L(copy0) !! 160 or %t5, %t6, %t3; \
150 cmp tmp1, count !! 161 std %t0, [%dst + (offset) + (offset2) + 0x00]; \
151 b.lo L(copy_long_backwards) !! 162 std %t2, [%dst + (offset) + (offset2) + 0x08];
152 !! 163
153 /* Copy 16 bytes and then align dst to !! 164 .text
154 !! 165 .align 4
155 ldp D_l, D_h, [src] !! 166
156 and tmp1, dstin, 15 !! 167 #ifdef FASTER_REVERSE
157 bic dst, dstin, 15 !! 168
158 sub src, src, tmp1 !! 169 70: /* rdword_align */
159 add count, count, tmp1 /* Cou !! 170
160 ldp A_l, A_h, [src, 16] !! 171 andcc %o1, 1, %g0
161 stp D_l, D_h, [dstin] !! 172 be 4f
162 ldp B_l, B_h, [src, 32] !! 173 andcc %o1, 2, %g0
163 ldp C_l, C_h, [src, 48] !! 174
164 ldp D_l, D_h, [src, 64]! !! 175 ldub [%o1 - 1], %g2
165 subs count, count, 128 + 16 /* Tes !! 176 sub %o1, 1, %o1
166 b.ls L(copy64_from_end) !! 177 stb %g2, [%o0 - 1]
167 !! 178 sub %o2, 1, %o2
168 L(loop64): !! 179 be 3f
169 stp A_l, A_h, [dst, 16] !! 180 sub %o0, 1, %o0
170 ldp A_l, A_h, [src, 16] !! 181 4:
171 stp B_l, B_h, [dst, 32] !! 182 lduh [%o1 - 2], %g2
172 ldp B_l, B_h, [src, 32] !! 183 sub %o1, 2, %o1
173 stp C_l, C_h, [dst, 48] !! 184 sth %g2, [%o0 - 2]
174 ldp C_l, C_h, [src, 48] !! 185 sub %o2, 2, %o2
175 stp D_l, D_h, [dst, 64]! !! 186 b 3f
176 ldp D_l, D_h, [src, 64]! !! 187 sub %o0, 2, %o0
177 subs count, count, 64 !! 188
178 b.hi L(loop64) !! 189 #endif /* FASTER_REVERSE */
179 !! 190
180 /* Write the last iteration and copy 6 !! 191 0:
181 L(copy64_from_end): !! 192 retl
182 ldp E_l, E_h, [srcend, -64] !! 193 nop ! Only bcopy returns here and it retuns void...
183 stp A_l, A_h, [dst, 16] !! 194
184 ldp A_l, A_h, [srcend, -48] !! 195 #ifdef __KERNEL__
185 stp B_l, B_h, [dst, 32] !! 196 FUNC(amemmove)
186 ldp B_l, B_h, [srcend, -32] !! 197 FUNC(__memmove)
187 stp C_l, C_h, [dst, 48] !! 198 #endif
188 ldp C_l, C_h, [srcend, -16] !! 199 FUNC(memmove)
189 stp D_l, D_h, [dst, 64] !! 200 cmp %o0, %o1
190 stp E_l, E_h, [dstend, -64] !! 201 SETUP_RETL
191 stp A_l, A_h, [dstend, -48] !! 202 bleu 9f
192 stp B_l, B_h, [dstend, -32] !! 203 sub %o0, %o1, %o4
193 stp C_l, C_h, [dstend, -16] !! 204
194 ret !! 205 add %o1, %o2, %o3
195 !! 206 cmp %o3, %o0
196 .p2align 4 !! 207 bleu 0f
197 !! 208 andcc %o4, 3, %o5
198 /* Large backwards copy for overlappin !! 209
199 Copy 16 bytes and then align dst to !! 210 #ifndef FASTER_REVERSE
200 L(copy_long_backwards): !! 211
201 ldp D_l, D_h, [srcend, -16] !! 212 add %o1, %o2, %o1
202 and tmp1, dstend, 15 !! 213 add %o0, %o2, %o0
203 sub srcend, srcend, tmp1 !! 214 sub %o1, 1, %o1
204 sub count, count, tmp1 !! 215 sub %o0, 1, %o0
205 ldp A_l, A_h, [srcend, -16] !! 216
206 stp D_l, D_h, [dstend, -16] !! 217 1: /* reverse_bytes */
207 ldp B_l, B_h, [srcend, -32] !! 218
208 ldp C_l, C_h, [srcend, -48] !! 219 ldub [%o1], %o4
209 ldp D_l, D_h, [srcend, -64]! !! 220 subcc %o2, 1, %o2
210 sub dstend, dstend, tmp1 !! 221 stb %o4, [%o0]
211 subs count, count, 128 !! 222 sub %o1, 1, %o1
212 b.ls L(copy64_from_start) !! 223 bne 1b
213 !! 224 sub %o0, 1, %o0
214 L(loop64_backwards): !! 225
215 stp A_l, A_h, [dstend, -16] !! 226 retl
216 ldp A_l, A_h, [srcend, -16] !! 227 RETL_INSN
217 stp B_l, B_h, [dstend, -32] !! 228
218 ldp B_l, B_h, [srcend, -32] !! 229 #else /* FASTER_REVERSE */
219 stp C_l, C_h, [dstend, -48] !! 230
220 ldp C_l, C_h, [srcend, -48] !! 231 add %o1, %o2, %o1
221 stp D_l, D_h, [dstend, -64]! !! 232 add %o0, %o2, %o0
222 ldp D_l, D_h, [srcend, -64]! !! 233 bne 77f
223 subs count, count, 64 !! 234 cmp %o2, 15
224 b.hi L(loop64_backwards) !! 235 bleu 91f
225 !! 236 andcc %o1, 3, %g0
226 /* Write the last iteration and copy 6 !! 237 bne 70b
227 L(copy64_from_start): !! 238 3:
228 ldp G_l, G_h, [src, 48] !! 239 andcc %o1, 4, %g0
229 stp A_l, A_h, [dstend, -16] !! 240
230 ldp A_l, A_h, [src, 32] !! 241 be 2f
231 stp B_l, B_h, [dstend, -32] !! 242 mov %o2, %g1
232 ldp B_l, B_h, [src, 16] !! 243
233 stp C_l, C_h, [dstend, -48] !! 244 ld [%o1 - 4], %o4
234 ldp C_l, C_h, [src] !! 245 sub %g1, 4, %g1
235 stp D_l, D_h, [dstend, -64] !! 246 st %o4, [%o0 - 4]
236 stp G_l, G_h, [dstin, 48] !! 247 sub %o1, 4, %o1
237 stp A_l, A_h, [dstin, 32] !! 248 sub %o0, 4, %o0
238 stp B_l, B_h, [dstin, 16] !! 249 2:
239 stp C_l, C_h, [dstin] !! 250 andcc %g1, 0xffffff80, %g7
240 ret !! 251 be 3f
241 SYM_FUNC_END(__pi_memcpy) !! 252 andcc %o0, 4, %g0
242 !! 253
243 SYM_FUNC_ALIAS(__memcpy, __pi_memcpy) !! 254 be 74f + 4
244 EXPORT_SYMBOL(__memcpy) !! 255 5:
245 SYM_FUNC_ALIAS_WEAK(memcpy, __memcpy) !! 256 RMOVE_BIGCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
246 EXPORT_SYMBOL(memcpy) !! 257 RMOVE_BIGCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
247 !! 258 RMOVE_BIGCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
248 SYM_FUNC_ALIAS(__pi_memmove, __pi_memcpy) !! 259 RMOVE_BIGCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
249 !! 260 subcc %g7, 128, %g7
250 SYM_FUNC_ALIAS(__memmove, __pi_memmove) !! 261 sub %o1, 128, %o1
251 EXPORT_SYMBOL(__memmove) !! 262 bne 5b
252 SYM_FUNC_ALIAS_WEAK(memmove, __memmove) !! 263 sub %o0, 128, %o0
253 EXPORT_SYMBOL(memmove) !! 264 3:
>> 265 andcc %g1, 0x70, %g7
>> 266 be 72f
>> 267 andcc %g1, 8, %g0
>> 268
>> 269 sethi %hi(72f), %o5
>> 270 srl %g7, 1, %o4
>> 271 add %g7, %o4, %o4
>> 272 sub %o1, %g7, %o1
>> 273 sub %o5, %o4, %o5
>> 274 jmpl %o5 + %lo(72f), %g0
>> 275 sub %o0, %g7, %o0
>> 276
>> 277 71: /* rmemcpy_table */
>> 278 RMOVE_LASTCHUNK(o1, o0, 0x60, g2, g3, g4, g5)
>> 279 RMOVE_LASTCHUNK(o1, o0, 0x50, g2, g3, g4, g5)
>> 280 RMOVE_LASTCHUNK(o1, o0, 0x40, g2, g3, g4, g5)
>> 281 RMOVE_LASTCHUNK(o1, o0, 0x30, g2, g3, g4, g5)
>> 282 RMOVE_LASTCHUNK(o1, o0, 0x20, g2, g3, g4, g5)
>> 283 RMOVE_LASTCHUNK(o1, o0, 0x10, g2, g3, g4, g5)
>> 284 RMOVE_LASTCHUNK(o1, o0, 0x00, g2, g3, g4, g5)
>> 285
>> 286 72: /* rmemcpy_table_end */
>> 287
>> 288 be 73f
>> 289 andcc %g1, 4, %g0
>> 290
>> 291 ldd [%o1 - 0x08], %g2
>> 292 sub %o0, 8, %o0
>> 293 sub %o1, 8, %o1
>> 294 st %g2, [%o0]
>> 295 st %g3, [%o0 + 0x04]
>> 296
>> 297 73: /* rmemcpy_last7 */
>> 298
>> 299 be 1f
>> 300 andcc %g1, 2, %g0
>> 301
>> 302 ld [%o1 - 4], %g2
>> 303 sub %o1, 4, %o1
>> 304 st %g2, [%o0 - 4]
>> 305 sub %o0, 4, %o0
>> 306 1:
>> 307 be 1f
>> 308 andcc %g1, 1, %g0
>> 309
>> 310 lduh [%o1 - 2], %g2
>> 311 sub %o1, 2, %o1
>> 312 sth %g2, [%o0 - 2]
>> 313 sub %o0, 2, %o0
>> 314 1:
>> 315 be 1f
>> 316 nop
>> 317
>> 318 ldub [%o1 - 1], %g2
>> 319 stb %g2, [%o0 - 1]
>> 320 1:
>> 321 retl
>> 322 RETL_INSN
>> 323
>> 324 74: /* rldd_std */
>> 325 RMOVE_BIGALIGNCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
>> 326 RMOVE_BIGALIGNCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
>> 327 RMOVE_BIGALIGNCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
>> 328 RMOVE_BIGALIGNCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
>> 329 subcc %g7, 128, %g7
>> 330 sub %o1, 128, %o1
>> 331 bne 74b
>> 332 sub %o0, 128, %o0
>> 333
>> 334 andcc %g1, 0x70, %g7
>> 335 be 72b
>> 336 andcc %g1, 8, %g0
>> 337
>> 338 sethi %hi(72b), %o5
>> 339 srl %g7, 1, %o4
>> 340 add %g7, %o4, %o4
>> 341 sub %o1, %g7, %o1
>> 342 sub %o5, %o4, %o5
>> 343 jmpl %o5 + %lo(72b), %g0
>> 344 sub %o0, %g7, %o0
>> 345
>> 346 75: /* rshort_end */
>> 347
>> 348 and %o2, 0xe, %o3
>> 349 2:
>> 350 sethi %hi(76f), %o5
>> 351 sll %o3, 3, %o4
>> 352 sub %o0, %o3, %o0
>> 353 sub %o5, %o4, %o5
>> 354 sub %o1, %o3, %o1
>> 355 jmpl %o5 + %lo(76f), %g0
>> 356 andcc %o2, 1, %g0
>> 357
>> 358 RMOVE_SHORTCHUNK(o1, o0, 0x0c, g2, g3)
>> 359 RMOVE_SHORTCHUNK(o1, o0, 0x0a, g2, g3)
>> 360 RMOVE_SHORTCHUNK(o1, o0, 0x08, g2, g3)
>> 361 RMOVE_SHORTCHUNK(o1, o0, 0x06, g2, g3)
>> 362 RMOVE_SHORTCHUNK(o1, o0, 0x04, g2, g3)
>> 363 RMOVE_SHORTCHUNK(o1, o0, 0x02, g2, g3)
>> 364 RMOVE_SHORTCHUNK(o1, o0, 0x00, g2, g3)
>> 365
>> 366 76: /* rshort_table_end */
>> 367
>> 368 be 1f
>> 369 nop
>> 370 ldub [%o1 - 1], %g2
>> 371 stb %g2, [%o0 - 1]
>> 372 1:
>> 373 retl
>> 374 RETL_INSN
>> 375
>> 376 91: /* rshort_aligned_end */
>> 377
>> 378 bne 75b
>> 379 andcc %o2, 8, %g0
>> 380
>> 381 be 1f
>> 382 andcc %o2, 4, %g0
>> 383
>> 384 ld [%o1 - 0x08], %g2
>> 385 ld [%o1 - 0x04], %g3
>> 386 sub %o1, 8, %o1
>> 387 st %g2, [%o0 - 0x08]
>> 388 st %g3, [%o0 - 0x04]
>> 389 sub %o0, 8, %o0
>> 390 1:
>> 391 b 73b
>> 392 mov %o2, %g1
>> 393
>> 394 77: /* rnon_aligned */
>> 395 cmp %o2, 15
>> 396 bleu 75b
>> 397 andcc %o0, 3, %g0
>> 398 be 64f
>> 399 andcc %o0, 1, %g0
>> 400 be 63f
>> 401 andcc %o0, 2, %g0
>> 402 ldub [%o1 - 1], %g5
>> 403 sub %o1, 1, %o1
>> 404 stb %g5, [%o0 - 1]
>> 405 sub %o0, 1, %o0
>> 406 be 64f
>> 407 sub %o2, 1, %o2
>> 408 63:
>> 409 ldub [%o1 - 1], %g5
>> 410 sub %o1, 2, %o1
>> 411 stb %g5, [%o0 - 1]
>> 412 sub %o0, 2, %o0
>> 413 ldub [%o1], %g5
>> 414 sub %o2, 2, %o2
>> 415 stb %g5, [%o0]
>> 416 64:
>> 417 and %o1, 3, %g2
>> 418 and %o1, -4, %o1
>> 419 and %o2, 0xc, %g3
>> 420 add %o1, 4, %o1
>> 421 cmp %g3, 4
>> 422 sll %g2, 3, %g4
>> 423 mov 32, %g2
>> 424 be 4f
>> 425 sub %g2, %g4, %g7
>> 426
>> 427 blu 3f
>> 428 cmp %g3, 8
>> 429
>> 430 be 2f
>> 431 srl %o2, 2, %g3
>> 432
>> 433 ld [%o1 - 4], %o3
>> 434 add %o0, -8, %o0
>> 435 ld [%o1 - 8], %o4
>> 436 add %o1, -16, %o1
>> 437 b 7f
>> 438 add %g3, 1, %g3
>> 439 2:
>> 440 ld [%o1 - 4], %o4
>> 441 add %o0, -4, %o0
>> 442 ld [%o1 - 8], %g1
>> 443 add %o1, -12, %o1
>> 444 b 8f
>> 445 add %g3, 2, %g3
>> 446 3:
>> 447 ld [%o1 - 4], %o5
>> 448 add %o0, -12, %o0
>> 449 ld [%o1 - 8], %o3
>> 450 add %o1, -20, %o1
>> 451 b 6f
>> 452 srl %o2, 2, %g3
>> 453 4:
>> 454 ld [%o1 - 4], %g1
>> 455 srl %o2, 2, %g3
>> 456 ld [%o1 - 8], %o5
>> 457 add %o1, -24, %o1
>> 458 add %o0, -16, %o0
>> 459 add %g3, -1, %g3
>> 460
>> 461 ld [%o1 + 12], %o3
>> 462 5:
>> 463 sll %o5, %g4, %g2
>> 464 srl %g1, %g7, %g5
>> 465 or %g2, %g5, %g2
>> 466 st %g2, [%o0 + 12]
>> 467 6:
>> 468 ld [%o1 + 8], %o4
>> 469 sll %o3, %g4, %g2
>> 470 srl %o5, %g7, %g5
>> 471 or %g2, %g5, %g2
>> 472 st %g2, [%o0 + 8]
>> 473 7:
>> 474 ld [%o1 + 4], %g1
>> 475 sll %o4, %g4, %g2
>> 476 srl %o3, %g7, %g5
>> 477 or %g2, %g5, %g2
>> 478 st %g2, [%o0 + 4]
>> 479 8:
>> 480 ld [%o1], %o5
>> 481 sll %g1, %g4, %g2
>> 482 srl %o4, %g7, %g5
>> 483 addcc %g3, -4, %g3
>> 484 or %g2, %g5, %g2
>> 485 add %o1, -16, %o1
>> 486 st %g2, [%o0]
>> 487 add %o0, -16, %o0
>> 488 bne,a 5b
>> 489 ld [%o1 + 12], %o3
>> 490 sll %o5, %g4, %g2
>> 491 srl %g1, %g7, %g5
>> 492 srl %g4, 3, %g3
>> 493 or %g2, %g5, %g2
>> 494 add %o1, %g3, %o1
>> 495 andcc %o2, 2, %g0
>> 496 st %g2, [%o0 + 12]
>> 497 be 1f
>> 498 andcc %o2, 1, %g0
>> 499
>> 500 ldub [%o1 + 15], %g5
>> 501 add %o1, -2, %o1
>> 502 stb %g5, [%o0 + 11]
>> 503 add %o0, -2, %o0
>> 504 ldub [%o1 + 16], %g5
>> 505 stb %g5, [%o0 + 12]
>> 506 1:
>> 507 be 1f
>> 508 nop
>> 509 ldub [%o1 + 15], %g5
>> 510 stb %g5, [%o0 + 11]
>> 511 1:
>> 512 retl
>> 513 RETL_INSN
>> 514
>> 515 #endif /* FASTER_REVERSE */
>> 516
>> 517 /* NOTE: This code is executed just for the cases,
>> 518 where %src (=%o1) & 3 is != 0.
>> 519 We need to align it to 4. So, for (%src & 3)
>> 520 1 we need to do ldub,lduh
>> 521 2 lduh
>> 522 3 just ldub
>> 523 so even if it looks weird, the branches
>> 524 are correct here. -jj
>> 525 */
>> 526 78: /* dword_align */
>> 527
>> 528 andcc %o1, 1, %g0
>> 529 be 4f
>> 530 andcc %o1, 2, %g0
>> 531
>> 532 ldub [%o1], %g2
>> 533 add %o1, 1, %o1
>> 534 stb %g2, [%o0]
>> 535 sub %o2, 1, %o2
>> 536 bne 3f
>> 537 add %o0, 1, %o0
>> 538 4:
>> 539 lduh [%o1], %g2
>> 540 add %o1, 2, %o1
>> 541 sth %g2, [%o0]
>> 542 sub %o2, 2, %o2
>> 543 b 3f
>> 544 add %o0, 2, %o0
>> 545
>> 546 #ifdef __KERNEL__
>> 547 FUNC(__memcpy)
>> 548 #endif
>> 549 FUNC(memcpy) /* %o0=dst %o1=src %o2=len */
>> 550
>> 551 sub %o0, %o1, %o4
>> 552 SETUP_RETL
>> 553 9:
>> 554 andcc %o4, 3, %o5
>> 555 0:
>> 556 bne 86f
>> 557 cmp %o2, 15
>> 558
>> 559 bleu 90f
>> 560 andcc %o1, 3, %g0
>> 561
>> 562 bne 78b
>> 563 3:
>> 564 andcc %o1, 4, %g0
>> 565
>> 566 be 2f
>> 567 mov %o2, %g1
>> 568
>> 569 ld [%o1], %o4
>> 570 sub %g1, 4, %g1
>> 571 st %o4, [%o0]
>> 572 add %o1, 4, %o1
>> 573 add %o0, 4, %o0
>> 574 2:
>> 575 andcc %g1, 0xffffff80, %g7
>> 576 be 3f
>> 577 andcc %o0, 4, %g0
>> 578
>> 579 be 82f + 4
>> 580 5:
>> 581 MOVE_BIGCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
>> 582 MOVE_BIGCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
>> 583 MOVE_BIGCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
>> 584 MOVE_BIGCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
>> 585 subcc %g7, 128, %g7
>> 586 add %o1, 128, %o1
>> 587 bne 5b
>> 588 add %o0, 128, %o0
>> 589 3:
>> 590 andcc %g1, 0x70, %g7
>> 591 be 80f
>> 592 andcc %g1, 8, %g0
>> 593
>> 594 sethi %hi(80f), %o5
>> 595 srl %g7, 1, %o4
>> 596 add %g7, %o4, %o4
>> 597 add %o1, %g7, %o1
>> 598 sub %o5, %o4, %o5
>> 599 jmpl %o5 + %lo(80f), %g0
>> 600 add %o0, %g7, %o0
>> 601
>> 602 79: /* memcpy_table */
>> 603
>> 604 MOVE_LASTCHUNK(o1, o0, 0x60, g2, g3, g4, g5)
>> 605 MOVE_LASTCHUNK(o1, o0, 0x50, g2, g3, g4, g5)
>> 606 MOVE_LASTCHUNK(o1, o0, 0x40, g2, g3, g4, g5)
>> 607 MOVE_LASTCHUNK(o1, o0, 0x30, g2, g3, g4, g5)
>> 608 MOVE_LASTCHUNK(o1, o0, 0x20, g2, g3, g4, g5)
>> 609 MOVE_LASTCHUNK(o1, o0, 0x10, g2, g3, g4, g5)
>> 610 MOVE_LASTCHUNK(o1, o0, 0x00, g2, g3, g4, g5)
>> 611
>> 612 80: /* memcpy_table_end */
>> 613 be 81f
>> 614 andcc %g1, 4, %g0
>> 615
>> 616 ldd [%o1], %g2
>> 617 add %o0, 8, %o0
>> 618 st %g2, [%o0 - 0x08]
>> 619 add %o1, 8, %o1
>> 620 st %g3, [%o0 - 0x04]
>> 621
>> 622 81: /* memcpy_last7 */
>> 623
>> 624 be 1f
>> 625 andcc %g1, 2, %g0
>> 626
>> 627 ld [%o1], %g2
>> 628 add %o1, 4, %o1
>> 629 st %g2, [%o0]
>> 630 add %o0, 4, %o0
>> 631 1:
>> 632 be 1f
>> 633 andcc %g1, 1, %g0
>> 634
>> 635 lduh [%o1], %g2
>> 636 add %o1, 2, %o1
>> 637 sth %g2, [%o0]
>> 638 add %o0, 2, %o0
>> 639 1:
>> 640 be 1f
>> 641 nop
>> 642
>> 643 ldub [%o1], %g2
>> 644 stb %g2, [%o0]
>> 645 1:
>> 646 retl
>> 647 RETL_INSN
>> 648
>> 649 82: /* ldd_std */
>> 650 MOVE_BIGALIGNCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
>> 651 MOVE_BIGALIGNCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
>> 652 MOVE_BIGALIGNCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
>> 653 MOVE_BIGALIGNCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
>> 654 subcc %g7, 128, %g7
>> 655 add %o1, 128, %o1
>> 656 bne 82b
>> 657 add %o0, 128, %o0
>> 658
>> 659 #ifndef FASTER_ALIGNED
>> 660
>> 661 andcc %g1, 0x70, %g7
>> 662 be 80b
>> 663 andcc %g1, 8, %g0
>> 664
>> 665 sethi %hi(80b), %o5
>> 666 srl %g7, 1, %o4
>> 667 add %g7, %o4, %o4
>> 668 add %o1, %g7, %o1
>> 669 sub %o5, %o4, %o5
>> 670 jmpl %o5 + %lo(80b), %g0
>> 671 add %o0, %g7, %o0
>> 672
>> 673 #else /* FASTER_ALIGNED */
>> 674
>> 675 andcc %g1, 0x70, %g7
>> 676 be 84f
>> 677 andcc %g1, 8, %g0
>> 678
>> 679 sethi %hi(84f), %o5
>> 680 add %o1, %g7, %o1
>> 681 sub %o5, %g7, %o5
>> 682 jmpl %o5 + %lo(84f), %g0
>> 683 add %o0, %g7, %o0
>> 684
>> 685 83: /* amemcpy_table */
>> 686
>> 687 MOVE_LASTALIGNCHUNK(o1, o0, 0x60, g2, g3, g4, g5)
>> 688 MOVE_LASTALIGNCHUNK(o1, o0, 0x50, g2, g3, g4, g5)
>> 689 MOVE_LASTALIGNCHUNK(o1, o0, 0x40, g2, g3, g4, g5)
>> 690 MOVE_LASTALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5)
>> 691 MOVE_LASTALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5)
>> 692 MOVE_LASTALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5)
>> 693 MOVE_LASTALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5)
>> 694
>> 695 84: /* amemcpy_table_end */
>> 696 be 85f
>> 697 andcc %g1, 4, %g0
>> 698
>> 699 ldd [%o1], %g2
>> 700 add %o0, 8, %o0
>> 701 std %g2, [%o0 - 0x08]
>> 702 add %o1, 8, %o1
>> 703 85: /* amemcpy_last7 */
>> 704 be 1f
>> 705 andcc %g1, 2, %g0
>> 706
>> 707 ld [%o1], %g2
>> 708 add %o1, 4, %o1
>> 709 st %g2, [%o0]
>> 710 add %o0, 4, %o0
>> 711 1:
>> 712 be 1f
>> 713 andcc %g1, 1, %g0
>> 714
>> 715 lduh [%o1], %g2
>> 716 add %o1, 2, %o1
>> 717 sth %g2, [%o0]
>> 718 add %o0, 2, %o0
>> 719 1:
>> 720 be 1f
>> 721 nop
>> 722
>> 723 ldub [%o1], %g2
>> 724 stb %g2, [%o0]
>> 725 1:
>> 726 retl
>> 727 RETL_INSN
>> 728
>> 729 #endif /* FASTER_ALIGNED */
>> 730
>> 731 86: /* non_aligned */
>> 732 cmp %o2, 6
>> 733 bleu 88f
>> 734
>> 735 #ifdef FASTER_NONALIGNED
>> 736
>> 737 cmp %o2, 256
>> 738 bcc 87f
>> 739
>> 740 #endif /* FASTER_NONALIGNED */
>> 741
>> 742 andcc %o0, 3, %g0
>> 743 be 61f
>> 744 andcc %o0, 1, %g0
>> 745 be 60f
>> 746 andcc %o0, 2, %g0
>> 747
>> 748 ldub [%o1], %g5
>> 749 add %o1, 1, %o1
>> 750 stb %g5, [%o0]
>> 751 sub %o2, 1, %o2
>> 752 bne 61f
>> 753 add %o0, 1, %o0
>> 754 60:
>> 755 ldub [%o1], %g3
>> 756 add %o1, 2, %o1
>> 757 stb %g3, [%o0]
>> 758 sub %o2, 2, %o2
>> 759 ldub [%o1 - 1], %g3
>> 760 add %o0, 2, %o0
>> 761 stb %g3, [%o0 - 1]
>> 762 61:
>> 763 and %o1, 3, %g2
>> 764 and %o2, 0xc, %g3
>> 765 and %o1, -4, %o1
>> 766 cmp %g3, 4
>> 767 sll %g2, 3, %g4
>> 768 mov 32, %g2
>> 769 be 4f
>> 770 sub %g2, %g4, %g7
>> 771
>> 772 blu 3f
>> 773 cmp %g3, 0x8
>> 774
>> 775 be 2f
>> 776 srl %o2, 2, %g3
>> 777
>> 778 ld [%o1], %o3
>> 779 add %o0, -8, %o0
>> 780 ld [%o1 + 4], %o4
>> 781 b 8f
>> 782 add %g3, 1, %g3
>> 783 2:
>> 784 ld [%o1], %o4
>> 785 add %o0, -12, %o0
>> 786 ld [%o1 + 4], %o5
>> 787 add %g3, 2, %g3
>> 788 b 9f
>> 789 add %o1, -4, %o1
>> 790 3:
>> 791 ld [%o1], %g1
>> 792 add %o0, -4, %o0
>> 793 ld [%o1 + 4], %o3
>> 794 srl %o2, 2, %g3
>> 795 b 7f
>> 796 add %o1, 4, %o1
>> 797 4:
>> 798 ld [%o1], %o5
>> 799 cmp %o2, 7
>> 800 ld [%o1 + 4], %g1
>> 801 srl %o2, 2, %g3
>> 802 bleu 10f
>> 803 add %o1, 8, %o1
>> 804
>> 805 ld [%o1], %o3
>> 806 add %g3, -1, %g3
>> 807 5:
>> 808 sll %o5, %g4, %g2
>> 809 srl %g1, %g7, %g5
>> 810 or %g2, %g5, %g2
>> 811 st %g2, [%o0]
>> 812 7:
>> 813 ld [%o1 + 4], %o4
>> 814 sll %g1, %g4, %g2
>> 815 srl %o3, %g7, %g5
>> 816 or %g2, %g5, %g2
>> 817 st %g2, [%o0 + 4]
>> 818 8:
>> 819 ld [%o1 + 8], %o5
>> 820 sll %o3, %g4, %g2
>> 821 srl %o4, %g7, %g5
>> 822 or %g2, %g5, %g2
>> 823 st %g2, [%o0 + 8]
>> 824 9:
>> 825 ld [%o1 + 12], %g1
>> 826 sll %o4, %g4, %g2
>> 827 srl %o5, %g7, %g5
>> 828 addcc %g3, -4, %g3
>> 829 or %g2, %g5, %g2
>> 830 add %o1, 16, %o1
>> 831 st %g2, [%o0 + 12]
>> 832 add %o0, 16, %o0
>> 833 bne,a 5b
>> 834 ld [%o1], %o3
>> 835 10:
>> 836 sll %o5, %g4, %g2
>> 837 srl %g1, %g7, %g5
>> 838 srl %g7, 3, %g3
>> 839 or %g2, %g5, %g2
>> 840 sub %o1, %g3, %o1
>> 841 andcc %o2, 2, %g0
>> 842 st %g2, [%o0]
>> 843 be 1f
>> 844 andcc %o2, 1, %g0
>> 845
>> 846 ldub [%o1], %g2
>> 847 add %o1, 2, %o1
>> 848 stb %g2, [%o0 + 4]
>> 849 add %o0, 2, %o0
>> 850 ldub [%o1 - 1], %g2
>> 851 stb %g2, [%o0 + 3]
>> 852 1:
>> 853 be 1f
>> 854 nop
>> 855 ldub [%o1], %g2
>> 856 stb %g2, [%o0 + 4]
>> 857 1:
>> 858 retl
>> 859 RETL_INSN
>> 860
>> 861 #ifdef FASTER_NONALIGNED
>> 862
>> 863 87: /* faster_nonaligned */
>> 864
>> 865 andcc %o1, 3, %g0
>> 866 be 3f
>> 867 andcc %o1, 1, %g0
>> 868
>> 869 be 4f
>> 870 andcc %o1, 2, %g0
>> 871
>> 872 ldub [%o1], %g2
>> 873 add %o1, 1, %o1
>> 874 stb %g2, [%o0]
>> 875 sub %o2, 1, %o2
>> 876 bne 3f
>> 877 add %o0, 1, %o0
>> 878 4:
>> 879 lduh [%o1], %g2
>> 880 add %o1, 2, %o1
>> 881 srl %g2, 8, %g3
>> 882 sub %o2, 2, %o2
>> 883 stb %g3, [%o0]
>> 884 add %o0, 2, %o0
>> 885 stb %g2, [%o0 - 1]
>> 886 3:
>> 887 andcc %o1, 4, %g0
>> 888
>> 889 bne 2f
>> 890 cmp %o5, 1
>> 891
>> 892 ld [%o1], %o4
>> 893 srl %o4, 24, %g2
>> 894 stb %g2, [%o0]
>> 895 srl %o4, 16, %g3
>> 896 stb %g3, [%o0 + 1]
>> 897 srl %o4, 8, %g2
>> 898 stb %g2, [%o0 + 2]
>> 899 sub %o2, 4, %o2
>> 900 stb %o4, [%o0 + 3]
>> 901 add %o1, 4, %o1
>> 902 add %o0, 4, %o0
>> 903 2:
>> 904 be 33f
>> 905 cmp %o5, 2
>> 906 be 32f
>> 907 sub %o2, 4, %o2
>> 908 31:
>> 909 ld [%o1], %g2
>> 910 add %o1, 4, %o1
>> 911 srl %g2, 24, %g3
>> 912 and %o0, 7, %g5
>> 913 stb %g3, [%o0]
>> 914 cmp %g5, 7
>> 915 sll %g2, 8, %g1
>> 916 add %o0, 4, %o0
>> 917 be 41f
>> 918 and %o2, 0xffffffc0, %o3
>> 919 ld [%o0 - 7], %o4
>> 920 4:
>> 921 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 922 SMOVE_CHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 923 SMOVE_CHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 924 SMOVE_CHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 925 subcc %o3, 64, %o3
>> 926 add %o1, 64, %o1
>> 927 bne 4b
>> 928 add %o0, 64, %o0
>> 929
>> 930 andcc %o2, 0x30, %o3
>> 931 be,a 1f
>> 932 srl %g1, 16, %g2
>> 933 4:
>> 934 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 935 subcc %o3, 16, %o3
>> 936 add %o1, 16, %o1
>> 937 bne 4b
>> 938 add %o0, 16, %o0
>> 939
>> 940 srl %g1, 16, %g2
>> 941 1:
>> 942 st %o4, [%o0 - 7]
>> 943 sth %g2, [%o0 - 3]
>> 944 srl %g1, 8, %g4
>> 945 b 88f
>> 946 stb %g4, [%o0 - 1]
>> 947 32:
>> 948 ld [%o1], %g2
>> 949 add %o1, 4, %o1
>> 950 srl %g2, 16, %g3
>> 951 and %o0, 7, %g5
>> 952 sth %g3, [%o0]
>> 953 cmp %g5, 6
>> 954 sll %g2, 16, %g1
>> 955 add %o0, 4, %o0
>> 956 be 42f
>> 957 and %o2, 0xffffffc0, %o3
>> 958 ld [%o0 - 6], %o4
>> 959 4:
>> 960 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 961 SMOVE_CHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 962 SMOVE_CHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 963 SMOVE_CHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 964 subcc %o3, 64, %o3
>> 965 add %o1, 64, %o1
>> 966 bne 4b
>> 967 add %o0, 64, %o0
>> 968
>> 969 andcc %o2, 0x30, %o3
>> 970 be,a 1f
>> 971 srl %g1, 16, %g2
>> 972 4:
>> 973 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 974 subcc %o3, 16, %o3
>> 975 add %o1, 16, %o1
>> 976 bne 4b
>> 977 add %o0, 16, %o0
>> 978
>> 979 srl %g1, 16, %g2
>> 980 1:
>> 981 st %o4, [%o0 - 6]
>> 982 b 88f
>> 983 sth %g2, [%o0 - 2]
>> 984 33:
>> 985 ld [%o1], %g2
>> 986 sub %o2, 4, %o2
>> 987 srl %g2, 24, %g3
>> 988 and %o0, 7, %g5
>> 989 stb %g3, [%o0]
>> 990 cmp %g5, 5
>> 991 srl %g2, 8, %g4
>> 992 sll %g2, 24, %g1
>> 993 sth %g4, [%o0 + 1]
>> 994 add %o1, 4, %o1
>> 995 be 43f
>> 996 and %o2, 0xffffffc0, %o3
>> 997
>> 998 ld [%o0 - 1], %o4
>> 999 add %o0, 4, %o0
>> 1000 4:
>> 1001 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, -1)
>> 1002 SMOVE_CHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, -1)
>> 1003 SMOVE_CHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, -1)
>> 1004 SMOVE_CHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, -1)
>> 1005 subcc %o3, 64, %o3
>> 1006 add %o1, 64, %o1
>> 1007 bne 4b
>> 1008 add %o0, 64, %o0
>> 1009
>> 1010 andcc %o2, 0x30, %o3
>> 1011 be,a 1f
>> 1012 srl %g1, 24, %g2
>> 1013 4:
>> 1014 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, -1)
>> 1015 subcc %o3, 16, %o3
>> 1016 add %o1, 16, %o1
>> 1017 bne 4b
>> 1018 add %o0, 16, %o0
>> 1019
>> 1020 srl %g1, 24, %g2
>> 1021 1:
>> 1022 st %o4, [%o0 - 5]
>> 1023 b 88f
>> 1024 stb %g2, [%o0 - 1]
>> 1025 41:
>> 1026 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 1027 SMOVE_ALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 1028 SMOVE_ALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 1029 SMOVE_ALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 1030 subcc %o3, 64, %o3
>> 1031 add %o1, 64, %o1
>> 1032 bne 41b
>> 1033 add %o0, 64, %o0
>> 1034
>> 1035 andcc %o2, 0x30, %o3
>> 1036 be,a 1f
>> 1037 srl %g1, 16, %g2
>> 1038 4:
>> 1039 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 1040 subcc %o3, 16, %o3
>> 1041 add %o1, 16, %o1
>> 1042 bne 4b
>> 1043 add %o0, 16, %o0
>> 1044
>> 1045 srl %g1, 16, %g2
>> 1046 1:
>> 1047 sth %g2, [%o0 - 3]
>> 1048 srl %g1, 8, %g4
>> 1049 b 88f
>> 1050 stb %g4, [%o0 - 1]
>> 1051 43:
>> 1052 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, 3)
>> 1053 SMOVE_ALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, 3)
>> 1054 SMOVE_ALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, 3)
>> 1055 SMOVE_ALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, 3)
>> 1056 subcc %o3, 64, %o3
>> 1057 add %o1, 64, %o1
>> 1058 bne 43b
>> 1059 add %o0, 64, %o0
>> 1060
>> 1061 andcc %o2, 0x30, %o3
>> 1062 be,a 1f
>> 1063 srl %g1, 24, %g2
>> 1064 4:
>> 1065 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, 3)
>> 1066 subcc %o3, 16, %o3
>> 1067 add %o1, 16, %o1
>> 1068 bne 4b
>> 1069 add %o0, 16, %o0
>> 1070
>> 1071 srl %g1, 24, %g2
>> 1072 1:
>> 1073 stb %g2, [%o0 + 3]
>> 1074 b 88f
>> 1075 add %o0, 4, %o0
>> 1076 42:
>> 1077 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 1078 SMOVE_ALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 1079 SMOVE_ALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 1080 SMOVE_ALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 1081 subcc %o3, 64, %o3
>> 1082 add %o1, 64, %o1
>> 1083 bne 42b
>> 1084 add %o0, 64, %o0
>> 1085
>> 1086 andcc %o2, 0x30, %o3
>> 1087 be,a 1f
>> 1088 srl %g1, 16, %g2
>> 1089 4:
>> 1090 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 1091 subcc %o3, 16, %o3
>> 1092 add %o1, 16, %o1
>> 1093 bne 4b
>> 1094 add %o0, 16, %o0
>> 1095
>> 1096 srl %g1, 16, %g2
>> 1097 1:
>> 1098 sth %g2, [%o0 - 2]
>> 1099
>> 1100 /* Fall through */
>> 1101
>> 1102 #endif /* FASTER_NONALIGNED */
>> 1103
>> 1104 88: /* short_end */
>> 1105
>> 1106 and %o2, 0xe, %o3
>> 1107 20:
>> 1108 sethi %hi(89f), %o5
>> 1109 sll %o3, 3, %o4
>> 1110 add %o0, %o3, %o0
>> 1111 sub %o5, %o4, %o5
>> 1112 add %o1, %o3, %o1
>> 1113 jmpl %o5 + %lo(89f), %g0
>> 1114 andcc %o2, 1, %g0
>> 1115
>> 1116 MOVE_SHORTCHUNK(o1, o0, 0x0c, g2, g3)
>> 1117 MOVE_SHORTCHUNK(o1, o0, 0x0a, g2, g3)
>> 1118 MOVE_SHORTCHUNK(o1, o0, 0x08, g2, g3)
>> 1119 MOVE_SHORTCHUNK(o1, o0, 0x06, g2, g3)
>> 1120 MOVE_SHORTCHUNK(o1, o0, 0x04, g2, g3)
>> 1121 MOVE_SHORTCHUNK(o1, o0, 0x02, g2, g3)
>> 1122 MOVE_SHORTCHUNK(o1, o0, 0x00, g2, g3)
>> 1123
>> 1124 89: /* short_table_end */
>> 1125
>> 1126 be 1f
>> 1127 nop
>> 1128
>> 1129 ldub [%o1], %g2
>> 1130 stb %g2, [%o0]
>> 1131 1:
>> 1132 retl
>> 1133 RETL_INSN
>> 1134
>> 1135 90: /* short_aligned_end */
>> 1136 bne 88b
>> 1137 andcc %o2, 8, %g0
>> 1138
>> 1139 be 1f
>> 1140 andcc %o2, 4, %g0
>> 1141
>> 1142 ld [%o1 + 0x00], %g2
>> 1143 ld [%o1 + 0x04], %g3
>> 1144 add %o1, 8, %o1
>> 1145 st %g2, [%o0 + 0x00]
>> 1146 st %g3, [%o0 + 0x04]
>> 1147 add %o0, 8, %o0
>> 1148 1:
>> 1149 b 81b
>> 1150 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.