This repository has been archived by the owner on Feb 11, 2021. It is now read-only.
/
SDL_stretch.c
364 lines (333 loc) · 9.55 KB
1
2
/*
SDL - Simple DirectMedia Layer
3
Copyright (C) 1997-2010 Sam Lantinga
4
5
This library is free software; you can redistribute it and/or
6
modify it under the terms of the GNU Lesser General Public
7
License as published by the Free Software Foundation; either
8
version 2.1 of the License, or (at your option) any later version.
9
10
11
12
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13
Lesser General Public License for more details.
14
15
16
17
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
18
19
Sam Lantinga
20
slouken@libsdl.org
21
*/
22
#include "SDL_config.h"
23
24
25
26
27
28
29
30
31
32
33
34
35
36
/* This a stretch blit implementation based on ideas given to me by
Tomasz Cejner - thanks! :)
April 27, 2000 - Sam Lantinga
*/
#include "SDL_video.h"
#include "SDL_blit.h"
/* This isn't ready for general consumption yet - it should be folded
into the general blitting mechanism.
*/
37
#if ((defined(_MFC_VER) && defined(_M_IX86)/* && !defined(_WIN32_WCE) still needed? */) || \
38
defined(__WATCOMC__) || \
39
(defined(__GNUC__) && defined(__i386__))) && SDL_ASSEMBLY_ROUTINES
40
41
42
/* There's a bug with gcc 4.4.1 and -O2 where srcp doesn't get the correct
* value after the first scanline. FIXME? */
/*#define USE_ASM_STRETCH*/
43
44
45
46
#endif
#ifdef USE_ASM_STRETCH
47
48
49
50
51
52
53
54
55
56
#ifdef HAVE_MPROTECT
#include <sys/types.h>
#include <sys/mman.h>
#endif
#ifdef __GNUC__
#define PAGE_ALIGNED __attribute__((__aligned__(4096)))
#else
#define PAGE_ALIGNED
#endif
57
#if defined(_M_IX86) || defined(i386)
58
59
60
61
62
63
64
65
66
67
#define PREFIX16 0x66
#define STORE_BYTE 0xAA
#define STORE_WORD 0xAB
#define LOAD_BYTE 0xAC
#define LOAD_WORD 0xAD
#define RETURN 0xC3
#else
#error Need assembly opcodes for this architecture
#endif
68
static unsigned char copy_row[4096] PAGE_ALIGNED;
69
70
71
static int
generate_rowbytes(int src_w, int dst_w, int bpp)
72
{
73
74
75
76
77
static struct
{
int bpp;
int src_w;
int dst_w;
78
int status;
79
80
81
82
} last;
int i;
int pos, inc;
83
unsigned char *eip, *fence;
84
85
86
87
unsigned char load, store;
/* See if we need to regenerate the copy buffer */
if ((src_w == last.src_w) && (dst_w == last.dst_w) && (bpp == last.bpp)) {
88
return (last.status);
89
90
91
92
}
last.bpp = bpp;
last.src_w = src_w;
last.dst_w = dst_w;
93
last.status = -1;
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
switch (bpp) {
case 1:
load = LOAD_BYTE;
store = STORE_BYTE;
break;
case 2:
case 4:
load = LOAD_WORD;
store = STORE_WORD;
break;
default:
SDL_SetError("ASM stretch of %d bytes isn't supported\n", bpp);
return (-1);
}
109
110
111
112
113
114
115
#ifdef HAVE_MPROTECT
/* Make the code writeable */
if (mprotect(copy_row, sizeof(copy_row), PROT_READ | PROT_WRITE) < 0) {
SDL_SetError("Couldn't make copy buffer writeable");
return (-1);
}
#endif
116
117
118
pos = 0x10000;
inc = (src_w << 16) / dst_w;
eip = copy_row;
119
fence = copy_row + sizeof(copy_row)-2;
120
121
for (i = 0; i < dst_w; ++i) {
while (pos >= 0x10000L) {
122
123
124
if (eip == fence) {
return -1;
}
125
126
127
128
129
130
if (bpp == 2) {
*eip++ = PREFIX16;
}
*eip++ = load;
pos -= 0x10000L;
}
131
132
133
if (eip == fence) {
return -1;
}
134
135
136
137
138
139
140
141
if (bpp == 2) {
*eip++ = PREFIX16;
}
*eip++ = store;
pos += inc;
}
*eip++ = RETURN;
142
143
144
145
146
147
148
149
#ifdef HAVE_MPROTECT
/* Make the code executable but not writeable */
if (mprotect(copy_row, sizeof(copy_row), PROT_READ | PROT_EXEC) < 0) {
SDL_SetError("Couldn't make copy buffer executable");
return (-1);
}
#endif
last.status = 0;
150
return (0);
151
152
}
153
#endif /* USE_ASM_STRETCH */
154
155
#define DEFINE_COPY_ROW(name, type) \
156
static void name(type *src, int src_w, type *dst, int dst_w) \
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
{ \
int i; \
int pos, inc; \
type pixel = 0; \
\
pos = 0x10000; \
inc = (src_w << 16) / dst_w; \
for ( i=dst_w; i>0; --i ) { \
while ( pos >= 0x10000L ) { \
pixel = *src++; \
pos -= 0x10000L; \
} \
*dst++ = pixel; \
pos += inc; \
} \
172
}
173
/* *INDENT-OFF* */
174
DEFINE_COPY_ROW(copy_row1, Uint8)
175
176
177
DEFINE_COPY_ROW(copy_row2, Uint16)
DEFINE_COPY_ROW(copy_row4, Uint32)
/* *INDENT-ON* */
178
179
/* The ASM code doesn't handle 24-bpp stretch blits */
180
static void
181
copy_row3(Uint8 * src, int src_w, Uint8 * dst, int dst_w)
182
{
183
184
int i;
int pos, inc;
185
Uint8 pixel[3] = { 0, 0, 0 };
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
pos = 0x10000;
inc = (src_w << 16) / dst_w;
for (i = dst_w; i > 0; --i) {
while (pos >= 0x10000L) {
pixel[0] = *src++;
pixel[1] = *src++;
pixel[2] = *src++;
pos -= 0x10000L;
}
*dst++ = pixel[0];
*dst++ = pixel[1];
*dst++ = pixel[2];
pos += inc;
}
201
202
203
204
205
}
/* Perform a stretch blit between two surfaces of the same format.
NOTE: This function is not safe to call from multiple threads!
*/
206
int
207
208
SDL_SoftStretch(SDL_Surface * src, const SDL_Rect * srcrect,
SDL_Surface * dst, const SDL_Rect * dstrect)
209
{
210
211
212
213
214
215
216
217
218
219
int src_locked;
int dst_locked;
int pos, inc;
int dst_width;
int dst_maxrow;
int src_row, dst_row;
Uint8 *srcp = NULL;
Uint8 *dstp;
SDL_Rect full_src;
SDL_Rect full_dst;
220
221
222
#ifdef USE_ASM_STRETCH
SDL_bool use_asm = SDL_TRUE;
#ifdef __GNUC__
223
int u1, u2;
224
#endif
225
#endif /* USE_ASM_STRETCH */
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
const int bpp = dst->format->BytesPerPixel;
if (src->format->BitsPerPixel != dst->format->BitsPerPixel) {
SDL_SetError("Only works with same format surfaces");
return (-1);
}
/* Verify the blit rectangles */
if (srcrect) {
if ((srcrect->x < 0) || (srcrect->y < 0) ||
((srcrect->x + srcrect->w) > src->w) ||
((srcrect->y + srcrect->h) > src->h)) {
SDL_SetError("Invalid source blit rectangle");
return (-1);
}
} else {
full_src.x = 0;
full_src.y = 0;
full_src.w = src->w;
full_src.h = src->h;
srcrect = &full_src;
}
if (dstrect) {
if ((dstrect->x < 0) || (dstrect->y < 0) ||
((dstrect->x + dstrect->w) > dst->w) ||
((dstrect->y + dstrect->h) > dst->h)) {
SDL_SetError("Invalid destination blit rectangle");
return (-1);
}
} else {
full_dst.x = 0;
full_dst.y = 0;
full_dst.w = dst->w;
full_dst.h = dst->h;
dstrect = &full_dst;
}
/* Lock the destination if it's in hardware */
dst_locked = 0;
if (SDL_MUSTLOCK(dst)) {
if (SDL_LockSurface(dst) < 0) {
SDL_SetError("Unable to lock destination surface");
return (-1);
}
dst_locked = 1;
}
/* Lock the source if it's in hardware */
src_locked = 0;
if (SDL_MUSTLOCK(src)) {
if (SDL_LockSurface(src) < 0) {
if (dst_locked) {
SDL_UnlockSurface(dst);
}
SDL_SetError("Unable to lock source surface");
return (-1);
}
src_locked = 1;
}
/* Set up the data... */
pos = 0x10000;
inc = (srcrect->h << 16) / dstrect->h;
src_row = srcrect->y;
dst_row = dstrect->y;
dst_width = dstrect->w * bpp;
291
292
#ifdef USE_ASM_STRETCH
293
/* Write the opcodes for this stretch */
294
295
if ((bpp == 3) || (generate_rowbytes(srcrect->w, dstrect->w, bpp) < 0)) {
use_asm = SDL_FALSE;
296
}
297
298
#endif
299
300
301
302
303
304
305
306
307
308
/* Perform the stretch blit */
for (dst_maxrow = dst_row + dstrect->h; dst_row < dst_maxrow; ++dst_row) {
dstp = (Uint8 *) dst->pixels + (dst_row * dst->pitch)
+ (dstrect->x * bpp);
while (pos >= 0x10000L) {
srcp = (Uint8 *) src->pixels + (src_row * src->pitch)
+ (srcrect->x * bpp);
++src_row;
pos -= 0x10000L;
}
309
#ifdef USE_ASM_STRETCH
310
if (use_asm) {
311
#ifdef __GNUC__
312
313
314
__asm__ __volatile__("call *%4":"=&D"(u1), "=&S"(u2)
:"0"(dstp), "1"(srcp), "r"(copy_row)
:"memory");
315
#elif defined(_MSC_VER) || defined(__WATCOMC__)
316
/* *INDENT-OFF* */
317
318
319
{
void *code = copy_row;
__asm {
320
321
322
323
324
325
326
327
push edi
push esi
mov edi, dstp
mov esi, srcp
call dword ptr code
pop esi
pop edi
}
328
}
329
/* *INDENT-ON* */
330
331
332
#else
#error Need inline assembly for this compiler
#endif
333
} else
334
#endif
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
switch (bpp) {
case 1:
copy_row1(srcp, srcrect->w, dstp, dstrect->w);
break;
case 2:
copy_row2((Uint16 *) srcp, srcrect->w,
(Uint16 *) dstp, dstrect->w);
break;
case 3:
copy_row3(srcp, srcrect->w, dstp, dstrect->w);
break;
case 4:
copy_row4((Uint32 *) srcp, srcrect->w,
(Uint32 *) dstp, dstrect->w);
break;
}
351
352
353
354
355
356
357
358
359
360
361
pos += inc;
}
/* We need to unlock the surfaces if they're locked */
if (dst_locked) {
SDL_UnlockSurface(dst);
}
if (src_locked) {
SDL_UnlockSurface(src);
}
return (0);
362
363
}
364
/* vi: set ts=4 sw=4 expandtab: */