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
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
|
#define __ASSEMBLY__
#include <linux/linkage.h>
#ifndef SYMBOL_NAME
#define SYMBOL_NAME(name) _ ## name
#endif
#ifndef ENTRY
#define ENTRY(name) .align 4; .globl _ ## name ## ; _ ## name ## :
#endif
.file "mem.S"
/* This file contains unrolled memcpy functions. */
/* Prototype: memcpy4to3( void *dest, void *src, int n ) */
/* Copies pixels from 4-byte-per-pixel screen to 3-byte-per-pixel screen,
*/
/* discarding the last byte of each pixel. */
/* Only uses 32-bit aligned word accesses. */
/* Instructions have been shuffled a bit for possible avoidance of */
/* pipeline hazards. */
.text
ENTRY(memcpy4to3)
pushl %ebp
movl %esp,%ebp
pushl %edi
pushl %esi
pushl %ebx
pushl %ecx
movl 8(%ebp),%edi /* destination address */
movl 12(%ebp),%esi /* source address */
movl 16(%ebp),%ecx /* number of pixels */
/* Handle chunks of 8 pixels. */
1: cmpl $8,%ecx
jl 2f
movl (%esi),%eax /* pixel 0 */
movl 4(%esi),%ebx /* pixel 1 */
shll $8,%eax /* BGR0 in 8-31 */
shrd $8,%ebx,%eax /* BGR0 in 0-23, B1 in 24-31 */
movl %eax,(%edi) /* write word */
shll $8,%ebx /* GR1 in 16-31 */
movl 8(%esi),%eax /* pixel 2 */
shrd $16,%eax,%ebx /* GR1 in 0-15, BG2 in 16-31 */
movl %ebx,4(%edi) /* write word */
shll $8,%eax /* move R2 into 24-31 */
movl 12(%esi),%ebx /* pixel 3 */
shrd $24,%ebx,%eax /* R2 in 0-7, BGR3 in 8-31 */
movl %eax,8(%edi) /* write word */
movl 16(%esi),%eax /* pixel 4 */
shll $8,%eax /* BGR4 in 8-31 */
movl 20(%esi),%ebx /* pixel 5 */
shrd $8,%ebx,%eax /* BGR4 in 0-23, B5 in 24-31 */
movl %eax,12(%edi) /* write word */
shll $8,%ebx /* GR5 in 16-31 */
movl 24(%esi),%eax /* pixel 6 */
shrd $16,%eax,%ebx /* GR5 in 0-15, BG6 in 16-31 */
movl %ebx,16(%edi) /* write word */
subl $8,%ecx /* blended end-of-loop instruction */
shll $8,%eax /* move R6 into 24-31 */
movl 28(%esi),%ebx /* pixel 7 */
shrd $24,%ebx,%eax /* R6 in 0-7, BGR7 in 8-31 */
addl $32,%esi /* blended end-of-loop instruction */
movl %eax,20(%edi) /* write word */
addl $24,%edi
jmp 1b
2: /* Do the remaining pixels. */
andl %ecx,%ecx
jz 4f /* none left */
3: movl (%esi),%eax
movw %ax,(%edi)
shrl $16,%eax
movb %al,2(%edi)
addl $4,%esi
addl $3,%edi
decl %ecx
jnz 3b
4:
popl %ecx
popl %ebx
popl %esi
popl %edi
popl %ebp
ret
/* Prototype: memcpy32shift8( void *dest, void *src, int n ) */
/* Copies pixels from 4-byte-per-pixel screen organized as BGR0 to */
/* 0BGR 4-byte-per-pixel screen. */
/* Used by copyscreen for ATI mach32 32-bit truecolor modes. */
.text
ENTRY(memcpy32shift8)
pushl %ebp
movl %esp,%ebp
pushl %edi
pushl %esi
pushl %ecx
pushl %ebx
movl 8(%ebp),%edi /* destination address */
movl 12(%ebp),%esi /* source address */
movl 16(%ebp),%ecx /* number of pixels */
/* Handle chunks of 8 pixels. */
1: cmpl $8,%ecx
jl 2f
movl (%esi),%eax
shll $8,%eax
movl %eax,(%edi)
movl 4(%esi),%edx
shll $8,%edx
movl %edx,4(%edi)
movl 8(%esi),%eax
shll $8,%eax
movl %eax,8(%edi)
movl 12(%esi),%edx
shll $8,%edx
movl %edx,12(%edi)
movl 16(%esi),%eax
shll $8,%eax
movl %eax,16(%edi)
movl 20(%esi),%edx
shll $8,%edx
movl %edx,20(%edi)
movl 24(%esi),%eax
subl $8,%ecx
shll $8,%eax
movl %eax,24(%edi)
movl 28(%esi),%edx
addl $32,%esi
shll $8,%edx
movl %edx,28(%edi)
addl $32,%edi
jmp 1b
2: andl %ecx,%ecx
jz 4f
3: movl (%esi),%eax
shll $8,%eax
movl %eax,(%edi)
addl $4,%esi
addl $4,%edi
decl %ecx
jnz 3b
4:
popl %ebx
popl %ecx
popl %esi
popl %edi
popl %ebp
ret
/* Optimized memcpy. */
/* Performance on par with inlined 32-bit aligned rep movsl on slow */
/* motherboard. */
/* Hypothesized to be fast on motherboards that handle writes efficiently */
/* and suffer with slow rep movsl microcode in 486/Pentium. */
/* (esp. Cyrix 486DX WB, Headland HTK 486 chipset, Pentium). */
/* Arguments passed in registers: */
/* destination address in %esi */
/* source address in %edx */
/* count in %ecx */
#define MOVEBYTE(n) movb n(%edx),%al; movb %al,n(%esi)
#define MOVESHORT(n) movw n(%edx),%ax; movw %ax,n(%esi)
#define MOVEWORD(n) movl n(%edx),%eax; movl %eax,n(%esi)
ENTRY(_memcpy_jumptable)
.long copy0
.long copy1, copy2, copy3, copy4
.long copy5, copy6, copy7, copy8
.long copy9, copy10, copy11, copy12
.long copy13, copy14, copy15, copy16
.long copy17, copy18, copy19, copy20
.long copy21, copy22, copy23, copy24
.long copy25, copy26, copy27, copy28
.long copy29, copy30, copy31, copy32
jumptable2:
.long align0, align1, align2, align3
ENTRY(_memcpyasm_regargs)
/* This is only valid if nu_bytes >= 3. */
/* Align destination to 32-bit boundary */
movl %esi,%eax
andl $3,%eax
jmp *jumptable2(,%eax,4)
align1: MOVESHORT(0)
MOVEBYTE(2)
addl $3,%edx
addl $3,%esi
subl $3,%ecx
jmp copyaligned
align3: MOVEBYTE(0)
incl %edx
incl %esi
decl %ecx
jmp copyaligned
align2: MOVESHORT(0)
addl $2,%edx
addl $2,%esi
subl $2,%ecx
align0:
copyaligned:
cmpl $32,%ecx
ja copyunrolled
/* <= 32 bytes. */
/* Copy remaining bytes (0-32). */
jmp *SYMBOL_NAME(_memcpy_jumptable)(,%ecx,4)
.align 4,0x90
/* memcpyasm_regargs_aligned is only called if nu_bytes > 32. */
ENTRY(_memcpyasm_regargs_aligned)
copyunrolled:
/* Copy chunks of 32 bytes. */
/* End-of-loop increment instructions blended in. */
addl $32,%esi /*P ok */
movl (%edx),%eax
movl %eax,(0-32)(%esi) /*P ok */
movl 4(%edx),%eax
movl %eax,(4-32)(%esi) /*P ok */
movl 8(%edx),%eax
movl %eax,(8-32)(%esi) /*P ok */
movl 12(%edx),%eax
movl %eax,(12-32)(%esi) /*P ok */
movl 16(%edx),%eax
addl $32,%edx /*P ok */
movl %eax,(16-32)(%esi)
subl $32,%ecx /*P ok */
movl (20-32)(%edx),%eax
movl %eax,(20-32)(%esi) /*P ok */
movl (24-32)(%edx),%eax
movl %eax,(24-32)(%esi) /*P ok */
movl (28-32)(%edx),%eax
movl %eax,(28-32)(%esi) /*P ok */
cmpl $32,%ecx
jge copyunrolled /*P fail */
/* Copy remaining bytes (less than 32). */
jmp *SYMBOL_NAME(_memcpy_jumptable)(,%ecx,4)
#define END ret
copy0: END
copy1: MOVEBYTE(0)
END
copy2: MOVESHORT(0)
END
copy3: MOVESHORT(0)
MOVEBYTE(2)
END
copy4: MOVEWORD(0)
END
copy5: MOVEWORD(0)
MOVEBYTE(4)
END
copy6: MOVEWORD(0)
MOVESHORT(4)
END
copy7: MOVEWORD(0)
MOVESHORT(4)
MOVEBYTE(6)
END
copy8: MOVEWORD(0)
MOVEWORD(4)
END
copy9: MOVEWORD(0)
MOVEWORD(4)
MOVEBYTE(8)
END
copy10: MOVEWORD(0)
MOVEWORD(4)
MOVESHORT(8)
END
copy11: MOVEWORD(0)
MOVEWORD(4)
MOVESHORT(8)
MOVEBYTE(10)
END
copy12: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
END
copy13: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEBYTE(12)
END
copy14: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVESHORT(12)
END
copy15: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVESHORT(12)
MOVEBYTE(14)
END
copy16: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
END
copy17: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEBYTE(16)
END
copy18: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVESHORT(16)
END
copy19: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVESHORT(16)
MOVEBYTE(18)
END
copy20: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
END
copy21: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVEBYTE(20)
END
copy22: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVESHORT(20)
END
copy23: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVESHORT(20)
MOVEBYTE(22)
END
copy24: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVEWORD(20)
END
copy25: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVEWORD(20)
MOVEBYTE(24)
END
copy26: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVEWORD(20)
MOVESHORT(24)
END
copy27: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVEWORD(20)
MOVESHORT(24)
MOVEBYTE(26)
END
copy28: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVEWORD(20)
MOVEWORD(24)
END
copy29: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVEWORD(20)
MOVEWORD(24)
MOVEBYTE(28)
END
copy30: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVEWORD(20)
MOVEWORD(24)
MOVESHORT(28)
END
copy31: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVEWORD(20)
MOVEWORD(24)
MOVESHORT(28)
MOVEBYTE(30)
END
copy32: MOVEWORD(0)
MOVEWORD(4)
MOVEWORD(8)
MOVEWORD(12)
MOVEWORD(16)
MOVEWORD(20)
MOVEWORD(24)
MOVEWORD(28)
END
|