OpenCores
URL https://opencores.org/ocsvn/openrisc/openrisc/trunk

Subversion Repositories openrisc

[/] [openrisc/] [trunk/] [gnu-old/] [gcc-4.2.2/] [gcc/] [config/] [score/] [mul-div.S] - Blame information for rev 825

Go to most recent revision | Details | Compare with Previous | View Log

Line No. Rev Author Line
1 38 julius
/* Copyright (C) 2005, 2007 Free Software Foundation, Inc.
2
   Contributed by Sunnorth
3
 
4
   This file is part of GCC.
5
 
6
   GCC is free software; you can redistribute it and/or modify it
7
   under the terms of the GNU General Public License as published
8
   by the Free Software Foundation; either version 3, or (at your
9
   option) any later version.
10
 
11
   GCC is distributed in the hope that it will be useful, but WITHOUT
12
   ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
13
   or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
14
   License for more details.
15
 
16
   You should have received a copy of the GNU General Public License
17
        along with GCC; see the file COPYING3.  If not see
18
.  */
19
 
20
#define ra r3
21
#define a0 r4
22
#define a1 r5
23
#define a2 r6
24
#define a3 r7
25
#define v0 r23
26
 
27
#define t0 r8
28
#define t1 r9
29
#define t2 r10
30
#define t3 r11
31
#define t4 r22
32
 
33
#ifndef __pic__
34
#if !defined(L_mulsi3) && !defined(L_divsi3)
35
       .text
36
       .global _flush_cache
37
_flush_cache:
38
        srli    r9, r5, 4
39
        mv      r8, r4
40
        mtsr    r9, sr0
41
1:
42
        cache   0xe, [r8, 0]            # write back invalid dcache
43
        addi    r8, 16
44
        bcnz    1b
45
        mfcr    r8, cr4
46
        bittst! r8, 0x3                 # if LDM is enable, write back LDM
47
        beq!    6f
48
        ldi     r10, 0
49
        cache   0xc, [r10, 0]
50
6:
51
        bittst! r8, 0x2                 # if LIM is enable, refill it
52
        beq!    7f
53
        cache   0x4, [r10, 0]
54
7:
55
        #nop!
56
        #nop!
57
        #nop!
58
        #nop!
59
        #nop!
60
        mv      r8, r4
61
        mtsr    r9, sr0
62
2:
63
        cache   0x2, [r8, 0]            # invalid unlock icache
64
        #nop!
65
        #nop!
66
        #nop!
67
        #nop!
68
        #nop!
69
        addi    r8, 16
70
        bcnz    2b
71
        br      r3
72
#endif
73
 
74
/* FUNCTION
75
   (U) INT32 v0 = __mulsi3 ((U) INT32 a0, (U) INT32 a1);
76
   REGISTERS:
77
        use     t0
78
        modify  a0
79
        a1      -> become 0
80
   NOTE:
81
   this seems to give better performance to just rotate and add.  */
82
 
83
#ifdef L_mulsi3
84
        .text
85
        .global __umulsi3
86
        .global __mulsi3
87
        /* signed multiplication (32x32)  */
88
        .ent    __mulsi3
89
__umulsi3:
90
__mulsi3:
91
        li      t1, 0
92
__mulsi3_loop:
93
        andri.c t0, a1, 1               # t0 = multiplier[0]
94
        srli    a1, a1, 1               # a1 /= 2
95
        beq     __mulsi3_loop2          # skip if (t0 == 0)
96
        add     t1, t1, a0              # add multiplicand
97
__mulsi3_loop2:
98
        slli    a0, a0, 1               # multiplicand mul 2
99
        cmpi.c  a1, 0
100
        bne     __mulsi3_loop
101
        mv      r4, t1
102
        br      ra
103
        .end    __mulsi3
104
#endif /* L_mulsi3 */
105
 
106
/* FUNCTION
107
   UINT32 (v0) = __udivsi3 (UINT32 (a0), UINT32 (a1));
108
   INT32 (v0) = __divsi3 (INT32 (a0),  INT32 (a1));
109
   UINT32 (v0) = __umodsi3 (UINT32 (a0), UINT32 (a1));
110
   INT32 (v0) = __modsi3 (INT32 (a0),  INT32 (a1));
111
   DESCRIPTION
112
   performs 32-bit division/modulo.
113
   REGISTERS
114
   used t0      bit-index
115
        t1
116
   modify a0    becomes remainer  */
117
#ifdef L_divsi3
118
        .text
119
        .global __udivsi3
120
        .global __umodsi3
121
        .global __divsi3
122
        .global __modsi3
123
 
124
        /* unsigned division  */
125
        .ent    __udivsi3
126
__udivsi3:
127
        li      t4, 0
128
        cmpi.c  a1, 0
129
        beq     __uds_exit
130
        li      t0, 1
131
        blt     __uds_ok
132
__uds_normalize:
133
        cmp.c   a0, a1
134
        bcc     __uds_ok
135
        slli    a1, a1, 1
136
        slli    t0, t0, 1
137
        cmpi.c  a1, 0
138
        bge     __uds_normalize
139
__uds_ok:
140
__uds_loop2:
141
        cmp.c   a0, a1
142
        bcc     __uds_loop3
143
        sub     a0, a0, a1
144
        or      t4, t4, t0
145
__uds_loop3:
146
        srli    t0, t0, 1
147
        srli    a1, a1, 1
148
        cmpi.c  t0, 0
149
        bne     __uds_loop2
150
__uds_exit:
151
        mv      a1, a0
152
        mv      r4, t4
153
        br      ra
154
        .end    __udivsi3
155
 
156
        /* unsigned modulus  */
157
        .ent    __umodsi3
158
__umodsi3:
159
        mv      t3, ra
160
        jl      __udivsi3
161
        mv      r4, a1
162
        br      t3
163
        .end    __umodsi3
164
 
165
        /* abs and div  */
166
        .ent    __orgsi3
167
__orgsi3:
168
        cmpi.c  a0, 0
169
        bge     __orgsi3_a0p
170
        neg     a0, a0
171
__orgsi3_a0p:
172
        cmpi.c  a1, 0
173
        bge     __udivsi3
174
        neg     a1, a1
175
        b       __udivsi3               # goto udivsi3
176
        .end    __orgsi3
177
 
178
        /* signed division  */
179
        .ent    __divsi3
180
__divsi3:
181
        mv      t3, ra
182
        xor     t2, a0, a1
183
        jl      __orgsi3
184
__divsi3_adjust:
185
        cmpi.c  t2, 0
186
        bge     __divsi3_exit
187
        neg     r4, r4
188
__divsi3_exit:
189
        br      t3
190
        .end    __divsi3
191
 
192
        /* signed modulus  */
193
        .ent    __modsi3
194
__modsi3:
195
        mv      t3, ra
196
        mv      t2, a0
197
        jl      __orgsi3
198
        mv      r4, a1
199
        b       __divsi3_adjust
200
        .end    __modsi3
201
 
202
#endif /* L_divsi3 */
203
#else /* -fPIC */
204
#if !defined(L_mulsi3) && !defined(L_divsi3)
205
        .set pic
206
        .text
207
        .global _flush_cache
208
_flush_cache:
209
        addi    r0, -8                  # pic used
210
        .cpload r29                     # pic used
211
        srli    r9, r5, 4
212
        mv      r8, r4
213
        mtsr    r9, sr0
214
1:
215
        cache   0xe, [r8, 0]            # write back invalid dcache
216
        addi    r8, 16
217
        bcnz    1b
218
        mfcr    r8, cr4
219
        bittst! r8, 0x3                 # if LDM is enable, write back LDM
220
        beq!    6f
221
        ldi     r10, 0
222
        cache   0xc, [r10, 0]
223
6:
224
        bittst! r8, 0x2                 # if LIM is enable, refill it
225
        beq!    7f
226
        cache   0x4, [r10, 0]
227
7:
228
        #nop!
229
        #nop!
230
        #nop!
231
        #nop!
232
        #nop!
233
        mv      r8, r4
234
        mtsr    r9, sr0
235
2:
236
        cache   0x2, [r8, 0]            # invalid unlock icache
237
        #nop!
238
        #nop!
239
        #nop!
240
        #nop!
241
        #nop!
242
        addi    r8, 16
243
        bcnz    2b
244
        .cprestore r0, 12               # pic used
245
        addi    r0, 8                   # pic used
246
        br      r3
247
#endif
248
 
249
/* FUNCTION
250
   (U) INT32 v0 = __mulsi3 ((U) INT32 a0, (U) INT32 a1);
251
   REGISTERS:
252
        use     t0
253
        modify  a0
254
        a1      -> become 0
255
   NOTE:
256
   this seems to give better performance to just rotate and add.  */
257
 
258
#ifdef L_mulsi3
259
        .set pic
260
        .text
261
        .global __umulsi3
262
        .global __mulsi3
263
        /* signed multiplication (32x32)  */
264
        .ent    __mulsi3
265
__umulsi3:
266
__mulsi3:
267
        addi    r0, -8                  # pic used
268
        .cpload r29                     # pic used
269
        li      t1, 0
270
__mulsi3_loop:
271
        andri.c t0, a1, 1               # t0 = multiplier[0]
272
        srli    a1, a1, 1               # a1 /= 2
273
        beq     __mulsi3_loop2          # skip if (t0 == 0)
274
        add     t1, t1, a0              # add multiplicand
275
__mulsi3_loop2:
276
        slli    a0, a0, 1               # multiplicand mul 2
277
        cmpi.c  a1, 0
278
        bne     __mulsi3_loop
279
        mv      r4, t1
280
        .cprestore r0, 12               # pic used
281
        addi    r0, 8                   # pic used
282
        br      ra
283
        .end    __mulsi3
284
#endif /* L_mulsi3 */
285
 
286
/* FUNCTION
287
   UINT32 (v0) = __udivsi3 (UINT32 (a0), UINT32 (a1));
288
   INT32 (v0) = __divsi3 (INT32 (a0),  INT32 (a1));
289
   UINT32 (v0) = __umodsi3 (UINT32 (a0), UINT32 (a1));
290
   INT32 (v0) = __modsi3 (INT32 (a0),  INT32 (a1));
291
   DESCRIPTION
292
   performs 32-bit division/modulo.
293
   REGISTERS
294
   used t0      bit-index
295
        t1
296
   modify a0    becomes remainer  */
297
#ifdef L_divsi3
298
        .set pic
299
        .text
300
        .global __udivsi3
301
        .global __umodsi3
302
        .global __divsi3
303
        .global __modsi3
304
 
305
        /* unsigned division  */
306
        .ent    __udivsi3
307
__udivsi3:
308
        addi    r0, -8                  # pic used
309
        .cpload r29                     # pic used
310
        li      t4, 0
311
        cmpi.c  a1, 0
312
        beq     __uds_exit
313
        li      t0, 1
314
        blt     __uds_ok
315
__uds_normalize:
316
        cmp.c   a0, a1
317
        bcc     __uds_ok
318
        slli    a1, a1, 1
319
        slli    t0, t0, 1
320
        cmpi.c  a1, 0
321
        bge     __uds_normalize
322
__uds_ok:
323
__uds_loop2:
324
        cmp.c   a0, a1
325
        bcc     __uds_loop3
326
        sub     a0, a0, a1
327
        or      t4, t4, t0
328
__uds_loop3:
329
        srli    t0, t0, 1
330
        srli    a1, a1, 1
331
        cmpi.c  t0, 0
332
        bne     __uds_loop2
333
__uds_exit:
334
        mv      a1, a0
335
        mv      r4, t4
336
        .cprestore r0, 12               # pic used
337
        addi    r0, 8                   # pic used
338
        br      ra
339
        .end    __udivsi3
340
 
341
        /* unsigned modulus  */
342
        .ent    __umodsi3
343
__umodsi3:
344
        addi    r0, -8                  # pic used
345
        .cpload r29                     # pic used
346
        li      t1, 0
347
        mv      t3, ra
348
#       jl      __udivsi3
349
        la      r29, __udivsi3
350
        brl     r29
351
        mv      r4, a1
352
        .cprestore r0, 12               # pic used
353
        addi    r0, 8                   # pic used
354
        br      t3
355
        .end    __umodsi3
356
 
357
        /* abs and div  */
358
        .ent    __orgsi3
359
__orgsi3:
360
        cmpi.c  a0, 0
361
        bge     __orgsi3_a0p
362
        neg     a0, a0
363
__orgsi3_a0p:
364
        cmpi.c  a1, 0
365
        bge     __udivsi3
366
        neg     a1, a1
367
        b       __udivsi3               # goto udivsi3
368
        .end    __orgsi3
369
 
370
        /* signed division  */
371
        .ent    __divsi3
372
__divsi3:
373
        addi    r0, -8                  # pic used
374
        .cpload r29                     # pic used
375
        mv      t3, ra
376
        xor     t2, a0, a1
377
#       jl      __orgsi3
378
        la      r29, __orgsi3
379
        brl     r29
380
__divsi3_adjust:
381
        cmpi.c  t2, 0
382
        bge     __divsi3_exit
383
        neg     r4, r4
384
__divsi3_exit:
385
        .cprestore r0, 12               # pic used
386
        addi    r0, 8                   # pic used
387
        br      t3
388
        .end    __divsi3
389
 
390
        /* signed modulus  */
391
        .ent    __modsi3
392
__modsi3:
393
        addi    r0, -8                  # pic used
394
        .cpload r29                     # pic used
395
        mv      t3, ra
396
        mv      t2, a0
397
#       jl      __orgsi3
398
        la      r29, __orgsi3
399
        brl     r29
400
        mv      r4, a1
401
        b       __divsi3_adjust
402
        .end    __modsi3
403
 
404
#endif /*L_divsi3 */
405
#endif

powered by: WebSVN 2.1.0

© copyright 1999-2024 OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.