x86p_16.asm
1
2
3
4
5
6
7
8
9
10
11
12
13
14
;
; x86 format converters for HERMES
; Copyright (c) 1998 Glenn Fielder (gaffer@gaffer.org)
; This source code is licensed under the GNU LGPL
;
; Please refer to the file COPYING.LIB contained in the distribution for
; licensing conditions
;
; Routines adjusted for Hermes by Christian Nentwich (brn@eleet.mcb.at)
; Used with permission.
;
BITS 32
15
%include "common.inc"
16
17
18
19
20
SDL_FUNC _ConvertX86p16_16BGR565
SDL_FUNC _ConvertX86p16_16RGB555
SDL_FUNC _ConvertX86p16_16BGR555
SDL_FUNC _ConvertX86p16_8RGB332
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
482
483
484
485
486
487
488
EXTERN _ConvertX86
EXTERN _x86return
SECTION .text
_ConvertX86p16_16BGR565:
; check short
cmp ecx,BYTE 16
ja .L3
.L1 ; short loop
mov al,[esi]
mov ah,[esi+1]
mov ebx,eax
mov edx,eax
shr eax,11
and eax,BYTE 11111b
and ebx,11111100000b
shl edx,11
add eax,ebx
add eax,edx
mov [edi],al
mov [edi+1],ah
add esi,BYTE 2
add edi,BYTE 2
dec ecx
jnz .L1
.L2
jmp _x86return
.L3 ; head
mov eax,edi
and eax,BYTE 11b
jz .L4
mov al,[esi]
mov ah,[esi+1]
mov ebx,eax
mov edx,eax
shr eax,11
and eax,BYTE 11111b
and ebx,11111100000b
shl edx,11
add eax,ebx
add eax,edx
mov [edi],al
mov [edi+1],ah
add esi,BYTE 2
add edi,BYTE 2
dec ecx
.L4 ; save count
push ecx
; unroll twice
shr ecx,1
; point arrays to end
lea esi,[esi+ecx*4]
lea edi,[edi+ecx*4]
; negative counter
neg ecx
jmp SHORT .L6
.L5 mov [edi+ecx*4-4],eax
.L6 mov eax,[esi+ecx*4]
mov ebx,[esi+ecx*4]
and eax,07E007E0h
mov edx,[esi+ecx*4]
and ebx,0F800F800h
shr ebx,11
and edx,001F001Fh
shl edx,11
add eax,ebx
add eax,edx
inc ecx
jnz .L5
mov [edi+ecx*4-4],eax
; tail
pop ecx
and ecx,BYTE 1
jz .L7
mov al,[esi]
mov ah,[esi+1]
mov ebx,eax
mov edx,eax
shr eax,11
and eax,BYTE 11111b
and ebx,11111100000b
shl edx,11
add eax,ebx
add eax,edx
mov [edi],al
mov [edi+1],ah
add esi,BYTE 2
add edi,BYTE 2
.L7
jmp _x86return
_ConvertX86p16_16RGB555:
; check short
cmp ecx,BYTE 32
ja .L3
.L1 ; short loop
mov al,[esi]
mov ah,[esi+1]
mov ebx,eax
shr ebx,1
and ebx, 0111111111100000b
and eax,BYTE 0000000000011111b
add eax,ebx
mov [edi],al
mov [edi+1],ah
add esi,BYTE 2
add edi,BYTE 2
dec ecx
jnz .L1
.L2
jmp _x86return
.L3 ; head
mov eax,edi
and eax,BYTE 11b
jz .L4
mov al,[esi]
mov ah,[esi+1]
mov ebx,eax
shr ebx,1
and ebx, 0111111111100000b
and eax,BYTE 0000000000011111b
add eax,ebx
mov [edi],al
mov [edi+1],ah
add esi,BYTE 2
add edi,BYTE 2
dec ecx
.L4 ; save ebp
push ebp
; save count
push ecx
; unroll four times
shr ecx,2
; point arrays to end
lea esi,[esi+ecx*8]
lea edi,[edi+ecx*8]
; negative counter
xor ebp,ebp
sub ebp,ecx
.L5 mov eax,[esi+ebp*8] ; agi?
mov ecx,[esi+ebp*8+4]
mov ebx,eax
mov edx,ecx
and eax,0FFC0FFC0h
and ecx,0FFC0FFC0h
shr eax,1
and ebx,001F001Fh
shr ecx,1
and edx,001F001Fh
add eax,ebx
add ecx,edx
mov [edi+ebp*8],eax
mov [edi+ebp*8+4],ecx
inc ebp
jnz .L5
; tail
pop ecx
.L6 and ecx,BYTE 11b
jz .L7
mov al,[esi]
mov ah,[esi+1]
mov ebx,eax
shr ebx,1
and ebx, 0111111111100000b
and eax,BYTE 0000000000011111b
add eax,ebx
mov [edi],al
mov [edi+1],ah
add esi,BYTE 2
add edi,BYTE 2
dec ecx
jmp SHORT .L6
.L7 pop ebp
jmp _x86return
_ConvertX86p16_16BGR555:
; check short
cmp ecx,BYTE 16
ja .L3
.L1 ; short loop
mov al,[esi]
mov ah,[esi+1]
mov ebx,eax
mov edx,eax
shr eax,11
and eax,BYTE 11111b
shr ebx,1
and ebx,1111100000b
shl edx,10
and edx,0111110000000000b
add eax,ebx
add eax,edx
mov [edi],al
mov [edi+1],ah
add esi,BYTE 2
add edi,BYTE 2
dec ecx
jnz .L1
.L2
jmp _x86return
.L3 ; head
mov eax,edi
and eax,BYTE 11b
jz .L4
mov al,[esi]
mov ah,[esi+1]
mov ebx,eax
mov edx,eax
shr eax,11
and eax,BYTE 11111b
shr ebx,1
and ebx,1111100000b
shl edx,10
and edx,0111110000000000b
add eax,ebx
add eax,edx
mov [edi],al
mov [edi+1],ah
add esi,BYTE 2
add edi,BYTE 2
dec ecx
.L4 ; save count
push ecx
; unroll twice
shr ecx,1
; point arrays to end
lea esi,[esi+ecx*4]
lea edi,[edi+ecx*4]
; negative counter
neg ecx
jmp SHORT .L6
.L5 mov [edi+ecx*4-4],eax
.L6 mov eax,[esi+ecx*4]
shr eax,1
mov ebx,[esi+ecx*4]
and eax,03E003E0h
mov edx,[esi+ecx*4]
and ebx,0F800F800h
shr ebx,11
and edx,001F001Fh
shl edx,10
add eax,ebx
add eax,edx
inc ecx
jnz .L5
mov [edi+ecx*4-4],eax
; tail
pop ecx
and ecx,BYTE 1
jz .L7
mov al,[esi]
mov ah,[esi+1]
mov ebx,eax
mov edx,eax
shr eax,11
and eax,BYTE 11111b
shr ebx,1
and ebx,1111100000b
shl edx,10
and edx,0111110000000000b
add eax,ebx
add eax,edx
mov [edi],al
mov [edi+1],ah
add esi,BYTE 2
add edi,BYTE 2
.L7
jmp _x86return
_ConvertX86p16_8RGB332:
; check short
cmp ecx,BYTE 16
ja .L3
.L1 ; short loop
mov al,[esi+0]
mov ah,[esi+1]
mov ebx,eax
mov edx,eax
and eax,BYTE 11000b ; blue
shr eax,3
and ebx,11100000000b ; green
shr ebx,6
and edx,1110000000000000b ; red
shr edx,8
add eax,ebx
add eax,edx
mov [edi],al
add esi,BYTE 2
inc edi
dec ecx
jnz .L1
.L2
jmp _x86return
.L3 mov eax,edi
and eax,BYTE 11b
jz .L4
mov al,[esi+0]
mov ah,[esi+1]
mov ebx,eax
mov edx,eax
and eax,BYTE 11000b ; blue
shr eax,3
and ebx,11100000000b ; green
shr ebx,6
and edx,1110000000000000b ; red
shr edx,8
add eax,ebx
add eax,edx
mov [edi],al
add esi,BYTE 2
inc edi
dec ecx
jmp SHORT .L3
.L4 ; save ebp
push ebp
; save count
push ecx
; unroll 4 times
shr ecx,2
; prestep
mov dl,[esi+0]
mov bl,[esi+1]
mov dh,[esi+2]
.L5 shl edx,16
mov bh,[esi+3]
shl ebx,16
mov dl,[esi+4]
mov dh,[esi+6]
mov bl,[esi+5]
and edx,00011000000110000001100000011000b
mov bh,[esi+7]
ror edx,16+3
mov eax,ebx ; setup eax for reds
and ebx,00000111000001110000011100000111b
and eax,11100000111000001110000011100000b ; reds
ror ebx,16-2
add esi,BYTE 8
ror eax,16
add edi,BYTE 4
add eax,ebx
mov bl,[esi+1] ; greens
add eax,edx
mov dl,[esi+0] ; blues
mov [edi-4],eax
mov dh,[esi+2]
dec ecx
jnz .L5
; check tail
pop ecx
and ecx,BYTE 11b
jz .L7
.L6 ; tail
mov al,[esi+0]
mov ah,[esi+1]
mov ebx,eax
mov edx,eax
and eax,BYTE 11000b ; blue
shr eax,3
and ebx,11100000000b ; green
shr ebx,6
and edx,1110000000000000b ; red
shr edx,8
add eax,ebx
add eax,edx
mov [edi],al
add esi,BYTE 2
inc edi
dec ecx
jnz .L6
.L7 pop ebp
jmp _x86return
489
490
491
%ifidn __OUTPUT_FORMAT__,elf
section .note.GNU-stack noalloc noexec nowrite progbits
%endif