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
|
/*****************************************************************************
** Copyright (c) 2000 John Adcock, Tom Barry, Steve Grimm All rights reserved.
** port copyright (c) 2003 Miguel Freitas
******************************************************************************
**
** This file is subject to the terms of the GNU General Public License as
** published by the Free Software Foundation. A copy of this license is
** included with this software distribution in the file COPYING. If you
** do not have a copy, you may obtain a copy by writing to the Free
** Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
**
** This software 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 General Public License for more details
******************************************************************************
** CVS Log
**
** Revision 1.10 2006/12/21 09:54:45 dgp85
** Apply the textrel patch from Gentoo, thanks to PaX team for providing it. The patch was applied and tested for a while in Gentoo and Pardus, and solves also Debian's problems with non-PIC code. If problems will arise, they'll be debugged.
**
** Revision 1.9 2006/02/04 14:06:29 miguelfreitas
** Enable AMD64 mmx/sse support in some plugins (tvtime, libmpeg2, goom...)
** patch by dani3l
**
** Revision 1.8 2005/06/05 16:00:06 miguelfreitas
** quite some hacks for gcc 2.95 compatibility
**
** Revision 1.7 2004/04/09 02:57:06 miguelfreitas
** tvtime deinterlacing algorithms assumed top_field_first=1
** top_field_first=0 (aka bottom_field_first) should now work as expected
**
** Revision 1.6 2004/02/12 20:53:31 mroi
** my gcc (partly 3.4 already) optimizes these away, because they are only used
** inside inline assembler (which the compiler does not recognize); so actually
** the code is wrong (the asm parts should list these as inputs), but telling
** the compiler to keep them is the easier fix
**
** Revision 1.5 2004/01/05 12:15:55 siggi
** wonder why Mike isn't complaining about C++ style comments, any more...
**
** Revision 1.4 2004/01/05 01:47:26 tmmm
** DOS/Win CRs are forbidden, verboten, interdit
**
** Revision 1.3 2004/01/02 20:53:43 miguelfreitas
** better MANGLE from ffmpeg
**
** Revision 1.2 2004/01/02 20:47:03 miguelfreitas
** my small contribution to the cygwin port ;-)
**
** Revision 1.1 2003/06/22 17:30:03 miguelfreitas
** use our own port of greedy2frame (tvtime port is currently broken)
**
** Revision 1.8 2001/11/23 17:18:54 adcockj
** Fixed silly and/or confusion
**
** Revision 1.7 2001/11/22 22:27:00 adcockj
** Bug Fixes
**
** Revision 1.6 2001/11/21 15:21:40 adcockj
** Renamed DEINTERLACE_INFO to TDeinterlaceInfo in line with standards
** Changed TDeinterlaceInfo structure to have history of pictures.
**
** Revision 1.5 2001/07/31 06:48:33 adcockj
** Fixed index bug spotted by Peter Gubanov
**
** Revision 1.4 2001/07/13 16:13:33 adcockj
** Added CVS tags and removed tabs
**
*****************************************************************************/
/*
* This is the implementation of the Greedy 2-frame deinterlace algorithm
* described in DI_Greedy2Frame.c. It's in a separate file so we can compile
* variants for different CPU types; most of the code is the same in the
* different variants.
*/
/****************************************************************************
** Field 1 | Field 2 | Field 3 | Field 4 |
** T0 | | T1 | |
** | M0 | | M1 |
** B0 | | B1 | |
*/
/* debugging feature */
/* output the value of mm4 at this point which is pink where we will weave */
/* and green were we are going to bob */
/* uncomment next line to see this */
/* #define CHECK_BOBWEAVE */
#if !defined(MASKS_DEFINED)
#define MASKS_DEFINED
static const int64_t __attribute__((__used__)) YMask = 0x00ff00ff00ff00ffll;
static const int64_t __attribute__((__used__)) Mask = 0x7f7f7f7f7f7f7f7fll;
static const int64_t __attribute__((__used__)) DwordOne = 0x0000000100000001ll;
static const int64_t __attribute__((__used__)) DwordTwo = 0x0000000200000002ll;
static int64_t qwGreedyTwoFrameThreshold;
#endif
#include <mangle.h>
#if defined(IS_SSE)
static void DeinterlaceGreedy2Frame_SSE(uint8_t *output, int outstride,
deinterlace_frame_data_t *data,
int bottom_field, int second_field, int width, int height )
#elif defined(IS_3DNOW)
static void DeinterlaceGreedy2Frame_3DNOW(uint8_t *output, int outstride,
deinterlace_frame_data_t *data,
int bottom_field, int second_field, int width, int height )
#else
static void DeinterlaceGreedy2Frame_MMX(uint8_t *output, int outstride,
deinterlace_frame_data_t *data,
int bottom_field, int second_field, int width, int height )
#endif
{
#if defined(ARCH_X86) || defined(ARCH_X86_64)
int Line;
int stride = width * 2;
register uint8_t* M1;
register uint8_t* M0;
register uint8_t* T0;
register uint8_t* T1;
register uint8_t* B1;
register uint8_t* B0;
uint8_t* Dest = output;
register uint8_t* Dest2;
register int count;
uint32_t Pitch = stride*2;
uint32_t LineLength = stride;
uint32_t PitchRest = Pitch - (LineLength >> 3)*8;
qwGreedyTwoFrameThreshold = GreedyTwoFrameThreshold;
qwGreedyTwoFrameThreshold += (GreedyTwoFrameThreshold2 << 8);
qwGreedyTwoFrameThreshold += (qwGreedyTwoFrameThreshold << 48) +
(qwGreedyTwoFrameThreshold << 32) +
(qwGreedyTwoFrameThreshold << 16);
if( second_field ) {
M1 = data->f0;
T1 = data->f0;
M0 = data->f1;
T0 = data->f1;
} else {
M1 = data->f0;
T1 = data->f1;
M0 = data->f1;
T0 = data->f2;
}
if( bottom_field ) {
M1 += stride;
T1 += 0;
B1 = T1 + Pitch;
M0 += stride;
T0 += 0;
B0 = T0 + Pitch;
} else {
M1 += Pitch;
T1 += stride;
B1 = T1 + Pitch;
M0 += Pitch;
T0 += stride;
B0 = T0 + Pitch;
xine_fast_memcpy(Dest, M1, LineLength);
Dest += outstride;
}
for (Line = 0; Line < (height / 2) - 1; ++Line)
{
/* Always use the most recent data verbatim. By definition it's correct
* (it'd be shown on an interlaced display) and our job is to fill in
* the spaces between the new lines.
*/
xine_fast_memcpy(Dest, T1, stride);
Dest += outstride;
Dest2 = Dest;
count = LineLength >> 3;
do {
asm volatile(
/* Figure out what to do with the scanline above the one we just copied.
* See above for a description of the algorithm.
*/
".align 8 \n\t"
"movq %4, %%mm6 \n\t"
"movq %0, %%mm1 \n\t" // T1
"movq %1, %%mm0 \n\t" // M1
"movq %2, %%mm3 \n\t" // B1
"movq %3, %%mm2 \n\t" // M0
: /* no output */
: "m" (*T1), "m" (*M1),
"m" (*B1), "m" (*M0), "m" (Mask) );
asm volatile(
/* Figure out what to do with the scanline above the one we just copied.
* See above for a description of the algorithm.
* Average T1 and B1 so we can do interpolated bobbing if we bob onto T1
*/
"movq %%mm3, %%mm7 \n\t" /* mm7 = B1 */
#if defined(IS_SSE)
"pavgb %%mm1, %%mm7 \n\t"
#elif defined(IS_3DNOW)
"pavgusb %%mm1, %%mm7 \n\t"
#else
"movq %%mm1, %%mm5 \n\t" /* mm5 = T1 */
"psrlw $1, %%mm7 \n\t" /* mm7 = B1 / 2 */
"pand %%mm6, %%mm7 \n\t" /* mask off lower bits */
"psrlw $1, %%mm5 \n\t" /* mm5 = T1 / 2 */
"pand %%mm6, %%mm5 \n\t" /* mask off lower bits */
"paddw %%mm5, %%mm7 \n\t" /* mm7 = (T1 + B1) / 2 */
#endif
/* calculate |M1-M0| put result in mm4 need to keep mm0 intact
* if we have a good processor then make mm0 the average of M1 and M0
* which should make weave look better when there is small amounts of
* movement
*/
#if defined(IS_SSE)
"movq %%mm0, %%mm4 \n\t"
"movq %%mm2, %%mm5 \n\t"
"psubusb %%mm2, %%mm4 \n\t"
"psubusb %%mm0, %%mm5 \n\t"
"por %%mm5, %%mm4 \n\t"
"psrlw $1, %%mm4 \n\t"
"pavgb %%mm2, %%mm0 \n\t"
"pand %%mm6, %%mm4 \n\t"
#elif defined(IS_3DNOW)
"movq %%mm0, %%mm4 \n\t"
"movq %%mm2, %%mm5 \n\t"
"psubusb %%mm2, %%mm4 \n\t"
"psubusb %%mm0, %%mm5 \n\t"
"por %%mm5, %%mm4 \n\t"
"psrlw $1, %%mm4 \n\t"
"pavgusb %%mm2, %%mm0 \n\t"
"pand %%mm6, %%mm4 \n\t"
#else
"movq %%mm0, %%mm4 \n\t"
"psubusb %%mm2, %%mm4 \n\t"
"psubusb %%mm0, %%mm2 \n\t"
"por %%mm2, %%mm4 \n\t"
"psrlw $1, %%mm4 \n\t"
"pand %%mm6, %%mm4 \n\t"
#endif
/* if |M1-M0| > Threshold we want dword worth of twos */
"pcmpgtb %3, %%mm4 \n\t"
"pand %4, %%mm4 \n\t" /* get rid of sign bit */
"pcmpgtd %5, %%mm4 \n\t" /* do we want to bob */
"pandn %6, %%mm4 \n\t"
"movq %1, %%mm2 \n\t" /* mm2 = T0 */
/* calculate |T1-T0| put result in mm5 */
"movq %%mm2, %%mm5 \n\t"
"psubusb %%mm1, %%mm5 \n\t"
"psubusb %%mm2, %%mm1 \n\t"
"por %%mm1, %%mm5 \n\t"
"psrlw $1, %%mm5 \n\t"
"pand %%mm6, %%mm5 \n\t"
/* if |T1-T0| > Threshold we want dword worth of ones */
"pcmpgtb %3, %%mm5 \n\t"
"pand %%mm6, %%mm5 \n\t" /* get rid of sign bit */
"pcmpgtd %5, %%mm5 \n\t"
"pandn %5, %%mm5 \n\t"
"paddd %%mm5, %%mm4 \n\t"
"movq %2, %%mm2 \n\t" /* B0 */
/* calculate |B1-B0| put result in mm5 */
"movq %%mm2, %%mm5 \n\t"
"psubusb %%mm3, %%mm5 \n\t"
"psubusb %%mm2, %%mm3 \n\t"
"por %%mm3, %%mm5 \n\t"
"psrlw $1, %%mm5 \n\t"
"pand %%mm6, %%mm5 \n\t"
/* if |B1-B0| > Threshold we want dword worth of ones */
"pcmpgtb %3, %%mm5 \n\t"
"pand %%mm6, %%mm5 \n\t" /* get rid of any sign bit */
"pcmpgtd %5, %%mm5 \n\t"
"pandn %5, %%mm5 \n\t"
"paddd %%mm5, %%mm4 \n\t"
"pcmpgtd %6, %%mm4 \n\t"
/* debugging feature
* output the value of mm4 at this point which is pink where we will weave
* and green were we are going to bob */
#ifdef CHECK_BOBWEAVE
#ifdef IS_SSE
"movntq %%mm4, %0 \n\t"
#else
"movq %%mm4, %0 \n\t"
#endif
#else
"movq %%mm4, %%mm5 \n\t"
/* mm4 now is 1 where we want to weave and 0 where we want to bob */
"pand %%mm0, %%mm4 \n\t"
"pandn %%mm7, %%mm5 \n\t"
"por %%mm5, %%mm4 \n\t"
#ifdef IS_SSE
"movntq %%mm4, %0 \n\t"
#else
"movq %%mm4, %0 \n\t"
#endif
#endif
: "=m" (*Dest2)
: "m" (*T0), "m" (*B0), "m" (qwGreedyTwoFrameThreshold), "m" (Mask), "m" (DwordOne), "m" (DwordTwo) );
/* Advance to the next set of pixels. */
T1 += 8;
M1 += 8;
B1 += 8;
M0 += 8;
T0 += 8;
B0 += 8;
Dest2 += 8;
} while( --count );
Dest += outstride;
M1 += PitchRest;
T1 += PitchRest;
B1 += PitchRest;
M0 += PitchRest;
T0 += PitchRest;
B0 += PitchRest;
}
#ifdef IS_SSE
asm("sfence\n\t");
#endif
if( bottom_field )
{
xine_fast_memcpy(Dest, T1, stride);
Dest += outstride;
xine_fast_memcpy(Dest, M1, stride);
}
else
{
xine_fast_memcpy(Dest, T1, stride);
}
/* clear out the MMX registers ready for doing floating point again */
asm("emms\n\t");
#endif
}
|
