1 // Copyright 2022 The Go Authors. All rights reserved.
2 // Use of this source code is governed by a BSD-style
3 // license that can be found in the LICENSE file.
4
5 #include "textflag.h"
6
7 // bool cas(uint32 *ptr, uint32 old, uint32 new)
8 // Atomically:
9 // if(*ptr == old){
10 // *ptr = new;
11 // return 1;
12 // } else
13 // return 0;
14 TEXT ·Cas(SB), NOSPLIT, $0-17
15 MOVV ptr+0(FP), R4
16 MOVW old+8(FP), R5
17 MOVW new+12(FP), R6
18 DBAR
19 cas_again:
20 MOVV R6, R7
21 LL (R4), R8
22 BNE R5, R8, cas_fail
23 SC R7, (R4)
24 BEQ R7, cas_again
25 MOVV $1, R4
26 MOVB R4, ret+16(FP)
27 DBAR
28 RET
29 cas_fail:
30 MOVV $0, R4
31 JMP -4(PC)
32
33 // bool cas64(uint64 *ptr, uint64 old, uint64 new)
34 // Atomically:
35 // if(*ptr == old){
36 // *ptr = new;
37 // return 1;
38 // } else {
39 // return 0;
40 // }
41 TEXT ·Cas64(SB), NOSPLIT, $0-25
42 MOVV ptr+0(FP), R4
43 MOVV old+8(FP), R5
44 MOVV new+16(FP), R6
45 DBAR
46 cas64_again:
47 MOVV R6, R7
48 LLV (R4), R8
49 BNE R5, R8, cas64_fail
50 SCV R7, (R4)
51 BEQ R7, cas64_again
52 MOVV $1, R4
53 MOVB R4, ret+24(FP)
54 DBAR
55 RET
56 cas64_fail:
57 MOVV $0, R4
58 JMP -4(PC)
59
60 TEXT ·Casuintptr(SB), NOSPLIT, $0-25
61 JMP ·Cas64(SB)
62
63 TEXT ·CasRel(SB), NOSPLIT, $0-17
64 JMP ·Cas(SB)
65
66 TEXT ·Loaduintptr(SB), NOSPLIT|NOFRAME, $0-16
67 JMP ·Load64(SB)
68
69 TEXT ·Loaduint(SB), NOSPLIT|NOFRAME, $0-16
70 JMP ·Load64(SB)
71
72 TEXT ·Storeuintptr(SB), NOSPLIT, $0-16
73 JMP ·Store64(SB)
74
75 TEXT ·Xadduintptr(SB), NOSPLIT, $0-24
76 JMP ·Xadd64(SB)
77
78 TEXT ·Loadint64(SB), NOSPLIT, $0-16
79 JMP ·Load64(SB)
80
81 TEXT ·Xaddint64(SB), NOSPLIT, $0-24
82 JMP ·Xadd64(SB)
83
84 // bool casp(void **val, void *old, void *new)
85 // Atomically:
86 // if(*val == old){
87 // *val = new;
88 // return 1;
89 // } else
90 // return 0;
91 TEXT ·Casp1(SB), NOSPLIT, $0-25
92 JMP ·Cas64(SB)
93
94 // uint32 xadd(uint32 volatile *ptr, int32 delta)
95 // Atomically:
96 // *val += delta;
97 // return *val;
98 TEXT ·Xadd(SB), NOSPLIT, $0-20
99 MOVV ptr+0(FP), R4
100 MOVW delta+8(FP), R5
101 DBAR
102 LL (R4), R6
103 ADDU R6, R5, R7
104 MOVV R7, R6
105 SC R7, (R4)
106 BEQ R7, -4(PC)
107 MOVW R6, ret+16(FP)
108 DBAR
109 RET
110
111 TEXT ·Xadd64(SB), NOSPLIT, $0-24
112 MOVV ptr+0(FP), R4
113 MOVV delta+8(FP), R5
114 DBAR
115 LLV (R4), R6
116 ADDVU R6, R5, R7
117 MOVV R7, R6
118 SCV R7, (R4)
119 BEQ R7, -4(PC)
120 MOVV R6, ret+16(FP)
121 DBAR
122 RET
123
124 TEXT ·Xchg(SB), NOSPLIT, $0-20
125 MOVV ptr+0(FP), R4
126 MOVW new+8(FP), R5
127
128 DBAR
129 MOVV R5, R6
130 LL (R4), R7
131 SC R6, (R4)
132 BEQ R6, -3(PC)
133 MOVW R7, ret+16(FP)
134 DBAR
135 RET
136
137 TEXT ·Xchg64(SB), NOSPLIT, $0-24
138 MOVV ptr+0(FP), R4
139 MOVV new+8(FP), R5
140
141 DBAR
142 MOVV R5, R6
143 LLV (R4), R7
144 SCV R6, (R4)
145 BEQ R6, -3(PC)
146 MOVV R7, ret+16(FP)
147 DBAR
148 RET
149
150 TEXT ·Xchguintptr(SB), NOSPLIT, $0-24
151 JMP ·Xchg64(SB)
152
153 TEXT ·StorepNoWB(SB), NOSPLIT, $0-16
154 JMP ·Store64(SB)
155
156 TEXT ·StoreRel(SB), NOSPLIT, $0-12
157 JMP ·Store(SB)
158
159 TEXT ·StoreRel64(SB), NOSPLIT, $0-16
160 JMP ·Store64(SB)
161
162 TEXT ·StoreReluintptr(SB), NOSPLIT, $0-16
163 JMP ·Store64(SB)
164
165 TEXT ·Store(SB), NOSPLIT, $0-12
166 MOVV ptr+0(FP), R4
167 MOVW val+8(FP), R5
168 DBAR
169 MOVW R5, 0(R4)
170 DBAR
171 RET
172
173 TEXT ·Store8(SB), NOSPLIT, $0-9
174 MOVV ptr+0(FP), R4
175 MOVB val+8(FP), R5
176 DBAR
177 MOVB R5, 0(R4)
178 DBAR
179 RET
180
181 TEXT ·Store64(SB), NOSPLIT, $0-16
182 MOVV ptr+0(FP), R4
183 MOVV val+8(FP), R5
184 DBAR
185 MOVV R5, 0(R4)
186 DBAR
187 RET
188
189 // void Or8(byte volatile*, byte);
190 TEXT ·Or8(SB), NOSPLIT, $0-9
191 MOVV ptr+0(FP), R4
192 MOVBU val+8(FP), R5
193 // Align ptr down to 4 bytes so we can use 32-bit load/store.
194 MOVV $~3, R6
195 AND R4, R6
196 // R7 = ((ptr & 3) * 8)
197 AND $3, R4, R7
198 SLLV $3, R7
199 // Shift val for aligned ptr. R5 = val << R4
200 SLLV R7, R5
201
202 DBAR
203 LL (R6), R7
204 OR R5, R7
205 SC R7, (R6)
206 BEQ R7, -4(PC)
207 DBAR
208 RET
209
210 // void And8(byte volatile*, byte);
211 TEXT ·And8(SB), NOSPLIT, $0-9
212 MOVV ptr+0(FP), R4
213 MOVBU val+8(FP), R5
214 // Align ptr down to 4 bytes so we can use 32-bit load/store.
215 MOVV $~3, R6
216 AND R4, R6
217 // R7 = ((ptr & 3) * 8)
218 AND $3, R4, R7
219 SLLV $3, R7
220 // Shift val for aligned ptr. R5 = val << R7 | ^(0xFF << R7)
221 MOVV $0xFF, R8
222 SLLV R7, R5
223 SLLV R7, R8
224 NOR R0, R8
225 OR R8, R5
226
227 DBAR
228 LL (R6), R7
229 AND R5, R7
230 SC R7, (R6)
231 BEQ R7, -4(PC)
232 DBAR
233 RET
234
235 // func Or(addr *uint32, v uint32)
236 TEXT ·Or(SB), NOSPLIT, $0-12
237 MOVV ptr+0(FP), R4
238 MOVW val+8(FP), R5
239 DBAR
240 LL (R4), R6
241 OR R5, R6
242 SC R6, (R4)
243 BEQ R6, -4(PC)
244 DBAR
245 RET
246
247 // func And(addr *uint32, v uint32)
248 TEXT ·And(SB), NOSPLIT, $0-12
249 MOVV ptr+0(FP), R4
250 MOVW val+8(FP), R5
251 DBAR
252 LL (R4), R6
253 AND R5, R6
254 SC R6, (R4)
255 BEQ R6, -4(PC)
256 DBAR
257 RET
258
259 // uint32 runtime∕internal∕atomic·Load(uint32 volatile* ptr)
260 TEXT ·Load(SB),NOSPLIT|NOFRAME,$0-12
261 MOVV ptr+0(FP), R19
262 DBAR
263 MOVWU 0(R19), R19
264 DBAR
265 MOVW R19, ret+8(FP)
266 RET
267
268 // uint8 runtime∕internal∕atomic·Load8(uint8 volatile* ptr)
269 TEXT ·Load8(SB),NOSPLIT|NOFRAME,$0-9
270 MOVV ptr+0(FP), R19
271 DBAR
272 MOVBU 0(R19), R19
273 DBAR
274 MOVB R19, ret+8(FP)
275 RET
276
277 // uint64 runtime∕internal∕atomic·Load64(uint64 volatile* ptr)
278 TEXT ·Load64(SB),NOSPLIT|NOFRAME,$0-16
279 MOVV ptr+0(FP), R19
280 DBAR
281 MOVV 0(R19), R19
282 DBAR
283 MOVV R19, ret+8(FP)
284 RET
285
286 // void *runtime∕internal∕atomic·Loadp(void *volatile *ptr)
287 TEXT ·Loadp(SB),NOSPLIT|NOFRAME,$0-16
288 MOVV ptr+0(FP), R19
289 DBAR
290 MOVV 0(R19), R19
291 DBAR
292 MOVV R19, ret+8(FP)
293 RET
294
295 // uint32 runtime∕internal∕atomic·LoadAcq(uint32 volatile* ptr)
296 TEXT ·LoadAcq(SB),NOSPLIT|NOFRAME,$0-12
297 JMP ·Load(SB)
298
299 // uint64 ·LoadAcq64(uint64 volatile* ptr)
300 TEXT ·LoadAcq64(SB),NOSPLIT|NOFRAME,$0-16
301 JMP ·Load64(SB)
302
303 // uintptr ·LoadAcquintptr(uintptr volatile* ptr)
304 TEXT ·LoadAcquintptr(SB),NOSPLIT|NOFRAME,$0-16
305 JMP ·Load64(SB)
306
307
View as plain text