1 /*
2 * Copyright (c) 2008-2009 Intel Corporation. All Rights Reserved.
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the
6 * "Software"), to deal in the Software without restriction, including
7 * without limitation the rights to use, copy, modify, merge, publish,
8 * distribute, sub license, and/or sell copies of the Software, and to
9 * permit persons to whom the Software is furnished to do so, subject to
10 * the following conditions:
11 *
12 * The above copyright notice and this permission notice (including the
13 * next paragraph) shall be included in all copies or substantial portions
14 * of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
17 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
18 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
19 * IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR
20 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
21 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
22 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
23 */
24 #include "loadsurface_yuv.h"
25
scale_2dimage(unsigned char * src_img,int src_imgw,int src_imgh,unsigned char * dst_img,int dst_imgw,int dst_imgh)26 static int scale_2dimage(unsigned char *src_img, int src_imgw, int src_imgh,
27 unsigned char *dst_img, int dst_imgw, int dst_imgh)
28 {
29 int row=0, col=0;
30
31 for (row=0; row<dst_imgh; row++) {
32 for (col=0; col<dst_imgw; col++) {
33 *(dst_img + row * dst_imgw + col) = *(src_img + (row * src_imgh/dst_imgh) * src_imgw + col * src_imgw/dst_imgw);
34 }
35 }
36
37 return 0;
38 }
39
40
YUV_blend_with_pic(int width,int height,unsigned char * Y_start,int Y_pitch,unsigned char * U_start,int U_pitch,unsigned char * V_start,int V_pitch,unsigned int fourcc,int fixed_alpha)41 static int YUV_blend_with_pic(int width, int height,
42 unsigned char *Y_start, int Y_pitch,
43 unsigned char *U_start, int U_pitch,
44 unsigned char *V_start, int V_pitch,
45 unsigned int fourcc, int fixed_alpha)
46 {
47 /* PIC YUV format */
48 unsigned char *pic_y_old = yuvga_pic;
49 unsigned char *pic_u_old = pic_y_old + 640*480;
50 unsigned char *pic_v_old = pic_u_old + 640*480/4;
51 unsigned char *pic_y, *pic_u, *pic_v;
52
53 int alpha_values[] = {100,90,80,70,60,50,40,30,20,30,40,50,60,70,80,90};
54
55 static int alpha_idx = 0;
56 int alpha;
57 int allocated = 0;
58
59 int row, col;
60
61 if (fixed_alpha == 0) {
62 alpha = alpha_values[alpha_idx % 16 ];
63 alpha_idx ++;
64 } else
65 alpha = fixed_alpha;
66
67 //alpha = 0;
68
69 pic_y = pic_y_old;
70 pic_u = pic_u_old;
71 pic_v = pic_v_old;
72
73 if (width != 640 || height != 480) { /* need to scale the pic */
74 pic_y = (unsigned char *)malloc(width * height);
75 pic_u = (unsigned char *)malloc(width * height/4);
76 pic_v = (unsigned char *)malloc(width * height/4);
77
78 allocated = 1;
79
80 scale_2dimage(pic_y_old, 640, 480,
81 pic_y, width, height);
82 scale_2dimage(pic_u_old, 320, 240,
83 pic_u, width/2, height/2);
84 scale_2dimage(pic_v_old, 320, 240,
85 pic_v, width/2, height/2);
86 }
87
88 /* begin blend */
89
90 /* Y plane */
91 int Y_pixel_stride = 1;
92 if (fourcc == VA_FOURCC_YUY2)
93 Y_pixel_stride = 2;
94
95 for (row=0; row<height; row++) {
96 unsigned char *p = Y_start + row * Y_pitch;
97 unsigned char *q = pic_y + row * width;
98 for (col=0; col<width; col++, q++) {
99 *p = *p * (100 - alpha) / 100 + *q * alpha/100;
100 p += Y_pixel_stride;
101 }
102 }
103
104 /* U/V plane */
105 int U_pixel_stride = 0, V_pixel_stride = 0;
106 int v_factor_to_nv12 = 1;
107 switch (fourcc) {
108 case VA_FOURCC_YV12:
109 U_pixel_stride = V_pixel_stride = 1;
110 break;
111 case VA_FOURCC_NV12:
112 U_pixel_stride = V_pixel_stride = 2;
113 break;
114 case VA_FOURCC_YUY2:
115 U_pixel_stride = V_pixel_stride = 4;
116 v_factor_to_nv12 = 2;
117 break;
118 default:
119 break;
120 }
121 for (row=0; row<height/2*v_factor_to_nv12; row++) {
122 unsigned char *pU = U_start + row * U_pitch;
123 unsigned char *pV = V_start + row * V_pitch;
124 unsigned char *qU = pic_u + row/v_factor_to_nv12 * width/2;
125 unsigned char *qV = pic_v + row/v_factor_to_nv12 * width/2;
126
127 for (col=0; col<width/2; col++, qU++, qV++) {
128 *pU = *pU * (100 - alpha) / 100 + *qU * alpha/100;
129 *pV = *pV * (100 - alpha) / 100 + *qV * alpha/100;
130
131 pU += U_pixel_stride;
132 pV += V_pixel_stride;
133 }
134 }
135
136
137 if (allocated) {
138 free(pic_y);
139 free(pic_u);
140 free(pic_v);
141 }
142
143 return 0;
144 }
145
yuvgen_planar(int width,int height,unsigned char * Y_start,int Y_pitch,unsigned char * U_start,int U_pitch,unsigned char * V_start,int V_pitch,unsigned int fourcc,int box_width,int row_shift,int field)146 static int yuvgen_planar(int width, int height,
147 unsigned char *Y_start, int Y_pitch,
148 unsigned char *U_start, int U_pitch,
149 unsigned char *V_start, int V_pitch,
150 unsigned int fourcc, int box_width, int row_shift,
151 int field)
152 {
153 int row, alpha;
154 unsigned char uv_value = 0x80;
155
156 /* copy Y plane */
157 int y_factor = 1;
158 if (fourcc == VA_FOURCC_YUY2) y_factor = 2;
159 for (row=0;row<height;row++) {
160 unsigned char *Y_row = Y_start + row * Y_pitch;
161 int jj, xpos, ypos;
162
163 ypos = (row / box_width) & 0x1;
164
165 /* fill garbage data into the other field */
166 if (((field == VA_TOP_FIELD) && (row &1))
167 || ((field == VA_BOTTOM_FIELD) && ((row &1)==0))) {
168 memset(Y_row, 0xff, width);
169 continue;
170 }
171
172 for (jj=0; jj<width; jj++) {
173 xpos = ((row_shift + jj) / box_width) & 0x1;
174 if (xpos == ypos)
175 Y_row[jj*y_factor] = 0xeb;
176 else
177 Y_row[jj*y_factor] = 0x10;
178
179 if (fourcc == VA_FOURCC_YUY2) {
180 Y_row[jj*y_factor+1] = uv_value; // it is for UV
181 }
182 }
183 }
184
185 /* copy UV data */
186 for( row =0; row < height/2; row++) {
187
188 /* fill garbage data into the other field */
189 if (((field == VA_TOP_FIELD) && (row &1))
190 || ((field == VA_BOTTOM_FIELD) && ((row &1)==0))) {
191 uv_value = 0xff;
192 }
193
194 unsigned char *U_row = U_start + row * U_pitch;
195 unsigned char *V_row = V_start + row * V_pitch;
196 switch (fourcc) {
197 case VA_FOURCC_NV12:
198 memset(U_row, uv_value, width);
199 break;
200 case VA_FOURCC_YV12:
201 memset (U_row,uv_value,width/2);
202 memset (V_row,uv_value,width/2);
203 break;
204 case VA_FOURCC_YUY2:
205 // see above. it is set with Y update.
206 break;
207 default:
208 printf("unsupported fourcc in loadsurface.h\n");
209 assert(0);
210 }
211 }
212
213 if (getenv("AUTO_UV") == 0)
214 return 0;
215
216 if (getenv("AUTO_ALPHA"))
217 alpha = 0;
218 else
219 alpha = 70;
220
221 YUV_blend_with_pic(width,height,
222 Y_start, Y_pitch,
223 U_start, U_pitch,
224 V_start, V_pitch,
225 fourcc, alpha);
226
227 return 0;
228 }
229
upload_surface(VADisplay va_dpy,VASurfaceID surface_id,int box_width,int row_shift,int field)230 static int upload_surface(VADisplay va_dpy, VASurfaceID surface_id,
231 int box_width, int row_shift,
232 int field)
233 {
234 VAImage surface_image;
235 void *surface_p=NULL, *U_start = NULL,*V_start = NULL;
236 VAStatus va_status;
237 unsigned int pitches[3]={0,0,0};
238
239 va_status = vaDeriveImage(va_dpy,surface_id,&surface_image);
240 CHECK_VASTATUS(va_status,"vaDeriveImage");
241
242 vaMapBuffer(va_dpy,surface_image.buf,&surface_p);
243 assert(VA_STATUS_SUCCESS == va_status);
244
245 pitches[0] = surface_image.pitches[0];
246 switch (surface_image.format.fourcc) {
247 case VA_FOURCC_NV12:
248 U_start = (char *)surface_p + surface_image.offsets[1];
249 V_start = (char *)U_start + 1;
250 pitches[1] = surface_image.pitches[1];
251 pitches[2] = surface_image.pitches[1];
252 break;
253 case VA_FOURCC_IYUV:
254 U_start = (char *)surface_p + surface_image.offsets[1];
255 V_start = (char *)surface_p + surface_image.offsets[2];
256 pitches[1] = surface_image.pitches[1];
257 pitches[2] = surface_image.pitches[2];
258 break;
259 case VA_FOURCC_YV12:
260 U_start = (char *)surface_p + surface_image.offsets[2];
261 V_start = (char *)surface_p + surface_image.offsets[1];
262 pitches[1] = surface_image.pitches[2];
263 pitches[2] = surface_image.pitches[1];
264 break;
265 case VA_FOURCC_YUY2:
266 U_start = (char *)surface_p + 1;
267 V_start = (char *)surface_p + 3;
268 pitches[1] = surface_image.pitches[0];
269 pitches[2] = surface_image.pitches[0];
270 break;
271 default:
272 assert(0);
273 }
274
275 /* assume surface is planar format */
276 yuvgen_planar(surface_image.width, surface_image.height,
277 (unsigned char *)surface_p, pitches[0],
278 (unsigned char *)U_start, pitches[1],
279 (unsigned char *)V_start, pitches[2],
280 surface_image.format.fourcc,
281 box_width, row_shift, field);
282
283 vaUnmapBuffer(va_dpy,surface_image.buf);
284
285 vaDestroyImage(va_dpy,surface_image.image_id);
286
287 return 0;
288 }
289
290 /*
291 * Upload YUV data from memory into a surface
292 * if src_fourcc == NV12, assume the buffer pointed by src_U
293 * is UV interleaved (src_V is ignored)
294 */
upload_surface_yuv(VADisplay va_dpy,VASurfaceID surface_id,int src_fourcc,int src_width,int src_height,unsigned char * src_Y,unsigned char * src_U,unsigned char * src_V)295 static int upload_surface_yuv(VADisplay va_dpy, VASurfaceID surface_id,
296 int src_fourcc, int src_width, int src_height,
297 unsigned char *src_Y, unsigned char *src_U, unsigned char *src_V)
298 {
299 VAImage surface_image;
300 unsigned char *surface_p=NULL, *Y_start=NULL, *U_start=NULL, *V_start=NULL;
301 int Y_pitch=0, U_pitch=0, V_pitch=0, row;
302 VAStatus va_status;
303
304 va_status = vaDeriveImage(va_dpy,surface_id, &surface_image);
305 CHECK_VASTATUS(va_status,"vaDeriveImage");
306
307 vaMapBuffer(va_dpy,surface_image.buf,(void **)&surface_p);
308 assert(VA_STATUS_SUCCESS == va_status);
309
310 Y_start = surface_p;
311 Y_pitch = surface_image.pitches[0];
312 switch (surface_image.format.fourcc) {
313 case VA_FOURCC_NV12:
314 U_start = (unsigned char *)surface_p + surface_image.offsets[1];
315 V_start = U_start + 1;
316 U_pitch = surface_image.pitches[1];
317 V_pitch = surface_image.pitches[1];
318 break;
319 case VA_FOURCC_IYUV:
320 U_start = (unsigned char *)surface_p + surface_image.offsets[1];
321 V_start = (unsigned char *)surface_p + surface_image.offsets[2];
322 U_pitch = surface_image.pitches[1];
323 V_pitch = surface_image.pitches[2];
324 break;
325 case VA_FOURCC_YV12:
326 U_start = (unsigned char *)surface_p + surface_image.offsets[2];
327 V_start = (unsigned char *)surface_p + surface_image.offsets[1];
328 U_pitch = surface_image.pitches[2];
329 V_pitch = surface_image.pitches[1];
330 break;
331 case VA_FOURCC_YUY2:
332 U_start = surface_p + 1;
333 V_start = surface_p + 3;
334 U_pitch = surface_image.pitches[0];
335 V_pitch = surface_image.pitches[0];
336 break;
337 default:
338 assert(0);
339 }
340
341 /* copy Y plane */
342 for (row=0;row<src_height;row++) {
343 unsigned char *Y_row = Y_start + row * Y_pitch;
344 memcpy(Y_row, src_Y + row*src_width, src_width);
345 }
346
347 for (row =0; row < src_height/2; row++) {
348 unsigned char *U_row = U_start + row * U_pitch;
349 unsigned char *u_ptr = NULL, *v_ptr=NULL;
350 int j;
351 switch (surface_image.format.fourcc) {
352 case VA_FOURCC_NV12:
353 if (src_fourcc == VA_FOURCC_NV12) {
354 memcpy(U_row, src_U + row * src_width, src_width);
355 break;
356 } else if (src_fourcc == VA_FOURCC_IYUV) {
357 u_ptr = src_U + row * (src_width/2);
358 v_ptr = src_V + row * (src_width/2);
359 } else if (src_fourcc == VA_FOURCC_YV12) {
360 v_ptr = src_U + row * (src_width/2);
361 u_ptr = src_V + row * (src_width/2);
362 }
363 for(j = 0; j < src_width/2; j++) {
364 U_row[2*j] = u_ptr[j];
365 U_row[2*j+1] = v_ptr[j];
366 }
367 break;
368 case VA_FOURCC_IYUV:
369 case VA_FOURCC_YV12:
370 case VA_FOURCC_YUY2:
371 default:
372 printf("unsupported fourcc in load_surface_yuv\n");
373 assert(0);
374 }
375 }
376
377 vaUnmapBuffer(va_dpy,surface_image.buf);
378
379 vaDestroyImage(va_dpy,surface_image.image_id);
380
381 return 0;
382 }
383
384 /*
385 * Download YUV data from a surface into memory
386 * Some hardward doesn't have a aperture for linear access of
387 * tiled surface, thus use vaGetImage to expect the implemnetion
388 * to do tile to linear convert
389 *
390 * if dst_fourcc == NV12, assume the buffer pointed by dst_U
391 * is UV interleaved (src_V is ignored)
392 */
download_surface_yuv(VADisplay va_dpy,VASurfaceID surface_id,int dst_fourcc,int dst_width,int dst_height,unsigned char * dst_Y,unsigned char * dst_U,unsigned char * dst_V)393 static int download_surface_yuv(VADisplay va_dpy, VASurfaceID surface_id,
394 int dst_fourcc, int dst_width, int dst_height,
395 unsigned char *dst_Y, unsigned char *dst_U, unsigned char *dst_V)
396 {
397 VAImage surface_image;
398 unsigned char *surface_p=NULL, *Y_start=NULL, *U_start=NULL,*V_start=NULL;
399 int Y_pitch=0, U_pitch=0, V_pitch=0, row;
400 VAStatus va_status;
401
402 va_status = vaDeriveImage(va_dpy,surface_id, &surface_image);
403 CHECK_VASTATUS(va_status,"vaDeriveImage");
404
405 vaMapBuffer(va_dpy,surface_image.buf,(void **)&surface_p);
406 assert(VA_STATUS_SUCCESS == va_status);
407
408 Y_start = surface_p;
409 Y_pitch = surface_image.pitches[0];
410 switch (surface_image.format.fourcc) {
411 case VA_FOURCC_NV12:
412 U_start = (unsigned char *)surface_p + surface_image.offsets[1];
413 V_start = U_start + 1;
414 U_pitch = surface_image.pitches[1];
415 V_pitch = surface_image.pitches[1];
416 break;
417 case VA_FOURCC_IYUV:
418 U_start = (unsigned char *)surface_p + surface_image.offsets[1];
419 V_start = (unsigned char *)surface_p + surface_image.offsets[2];
420 U_pitch = surface_image.pitches[1];
421 V_pitch = surface_image.pitches[2];
422 break;
423 case VA_FOURCC_YV12:
424 U_start = (unsigned char *)surface_p + surface_image.offsets[2];
425 V_start = (unsigned char *)surface_p + surface_image.offsets[1];
426 U_pitch = surface_image.pitches[2];
427 V_pitch = surface_image.pitches[1];
428 break;
429 case VA_FOURCC_YUY2:
430 U_start = surface_p + 1;
431 V_start = surface_p + 3;
432 U_pitch = surface_image.pitches[0];
433 V_pitch = surface_image.pitches[0];
434 break;
435 default:
436 assert(0);
437 }
438
439 /* copy Y plane */
440 for (row=0;row<dst_height;row++) {
441 unsigned char *Y_row = Y_start + row * Y_pitch;
442 memcpy(dst_Y + row*dst_width, Y_row, dst_width);
443 }
444
445 for (row =0; row < dst_height/2; row++) {
446 unsigned char *U_row = U_start + row * U_pitch;
447 unsigned char *u_ptr = NULL, *v_ptr = NULL;
448 int j;
449 switch (surface_image.format.fourcc) {
450 case VA_FOURCC_NV12:
451 if (dst_fourcc == VA_FOURCC_NV12) {
452 memcpy(dst_U + row * dst_width, U_row, dst_width);
453 break;
454 } else if (dst_fourcc == VA_FOURCC_IYUV) {
455 u_ptr = dst_U + row * (dst_width/2);
456 v_ptr = dst_V + row * (dst_width/2);
457 } else if (dst_fourcc == VA_FOURCC_YV12) {
458 v_ptr = dst_U + row * (dst_width/2);
459 u_ptr = dst_V + row * (dst_width/2);
460 }
461 for(j = 0; j < dst_width/2; j++) {
462 u_ptr[j] = U_row[2*j];
463 v_ptr[j] = U_row[2*j+1];
464 }
465 break;
466 case VA_FOURCC_IYUV:
467 case VA_FOURCC_YV12:
468 case VA_FOURCC_YUY2:
469 default:
470 printf("unsupported fourcc in load_surface_yuv\n");
471 assert(0);
472 }
473 }
474
475 vaUnmapBuffer(va_dpy,surface_image.buf);
476
477 vaDestroyImage(va_dpy,surface_image.image_id);
478
479 return 0;
480 }
481