Merge branch 'halasa-hdlc' of git://git.tuxdriver.com/git/netdev-jwl
[linux-2.6] / drivers / media / common / saa7146_hlp.c
1 #include <linux/kernel.h>
2 #include <media/saa7146_vv.h>
3
4 static void calculate_output_format_register(struct saa7146_dev* saa, u32 palette, u32* clip_format)
5 {
6         /* clear out the necessary bits */
7         *clip_format &= 0x0000ffff;
8         /* set these bits new */
9         *clip_format |=  (( ((palette&0xf00)>>8) << 30) | ((palette&0x00f) << 24) | (((palette&0x0f0)>>4) << 16));
10 }
11
12 static void calculate_hps_source_and_sync(struct saa7146_dev *dev, int source, int sync, u32* hps_ctrl)
13 {
14         *hps_ctrl &= ~(MASK_30 | MASK_31 | MASK_28);
15         *hps_ctrl |= (source << 30) | (sync << 28);
16 }
17
18 static void calculate_hxo_and_hyo(struct saa7146_vv *vv, u32* hps_h_scale, u32* hps_ctrl)
19 {
20         int hyo = 0, hxo = 0;
21
22         hyo = vv->standard->v_offset;
23         hxo = vv->standard->h_offset;
24
25         *hps_h_scale    &= ~(MASK_B0 | 0xf00);
26         *hps_h_scale    |= (hxo <<  0);
27
28         *hps_ctrl       &= ~(MASK_W0 | MASK_B2);
29         *hps_ctrl       |= (hyo << 12);
30 }
31
32 /* helper functions for the calculation of the horizontal- and vertical
33    scaling registers, clip-format-register etc ...
34    these functions take pointers to the (most-likely read-out
35    original-values) and manipulate them according to the requested
36    changes.
37 */
38
39 /* hps_coeff used for CXY and CXUV; scale 1/1 -> scale 1/64 */
40 static struct {
41         u16 hps_coeff;
42         u16 weight_sum;
43 } hps_h_coeff_tab [] = {
44         {0x00,   2}, {0x02,   4}, {0x00,   4}, {0x06,   8}, {0x02,   8},
45         {0x08,   8}, {0x00,   8}, {0x1E,  16}, {0x0E,   8}, {0x26,   8},
46         {0x06,   8}, {0x42,   8}, {0x02,   8}, {0x80,   8}, {0x00,   8},
47         {0xFE,  16}, {0xFE,   8}, {0x7E,   8}, {0x7E,   8}, {0x3E,   8},
48         {0x3E,   8}, {0x1E,   8}, {0x1E,   8}, {0x0E,   8}, {0x0E,   8},
49         {0x06,   8}, {0x06,   8}, {0x02,   8}, {0x02,   8}, {0x00,   8},
50         {0x00,   8}, {0xFE,  16}, {0xFE,   8}, {0xFE,   8}, {0xFE,   8},
51         {0xFE,   8}, {0xFE,   8}, {0xFE,   8}, {0xFE,   8}, {0xFE,   8},
52         {0xFE,   8}, {0xFE,   8}, {0xFE,   8}, {0xFE,   8}, {0xFE,   8},
53         {0xFE,   8}, {0xFE,   8}, {0xFE,   8}, {0xFE,   8}, {0x7E,   8},
54         {0x7E,   8}, {0x3E,   8}, {0x3E,   8}, {0x1E,   8}, {0x1E,   8},
55         {0x0E,   8}, {0x0E,   8}, {0x06,   8}, {0x06,   8}, {0x02,   8},
56         {0x02,   8}, {0x00,   8}, {0x00,   8}, {0xFE,  16}
57 };
58
59 /* table of attenuation values for horizontal scaling */
60 static u8 h_attenuation[] = { 1, 2, 4, 8, 2, 4, 8, 16, 0};
61
62 /* calculate horizontal scale registers */
63 static int calculate_h_scale_registers(struct saa7146_dev *dev,
64         int in_x, int out_x, int flip_lr,
65         u32* hps_ctrl, u32* hps_v_gain, u32* hps_h_prescale, u32* hps_h_scale)
66 {
67         /* horizontal prescaler */
68         u32 dcgx = 0, xpsc = 0, xacm = 0, cxy = 0, cxuv = 0;
69         /* horizontal scaler */
70         u32 xim = 0, xp = 0, xsci =0;
71         /* vertical scale & gain */
72         u32 pfuv = 0;
73
74         /* helper variables */
75         u32 h_atten = 0, i = 0;
76
77         if ( 0 == out_x ) {
78                 return -EINVAL;
79         }
80
81         /* mask out vanity-bit */
82         *hps_ctrl &= ~MASK_29;
83
84         /* calculate prescale-(xspc)-value:     [n   .. 1/2) : 1
85                                                 [1/2 .. 1/3) : 2
86                                                 [1/3 .. 1/4) : 3
87                                                 ...             */
88         if (in_x > out_x) {
89                 xpsc = in_x / out_x;
90         }
91         else {
92                 /* zooming */
93                 xpsc = 1;
94         }
95
96         /* if flip_lr-bit is set, number of pixels after
97            horizontal prescaling must be < 384 */
98         if ( 0 != flip_lr ) {
99
100                 /* set vanity bit */
101                 *hps_ctrl |= MASK_29;
102
103                 while (in_x / xpsc >= 384 )
104                         xpsc++;
105         }
106         /* if zooming is wanted, number of pixels after
107            horizontal prescaling must be < 768 */
108         else {
109                 while ( in_x / xpsc >= 768 )
110                         xpsc++;
111         }
112
113         /* maximum prescale is 64 (p.69) */
114         if ( xpsc > 64 )
115                 xpsc = 64;
116
117         /* keep xacm clear*/
118         xacm = 0;
119
120         /* set horizontal filter parameters (CXY = CXUV) */
121         cxy = hps_h_coeff_tab[( (xpsc - 1) < 63 ? (xpsc - 1) : 63 )].hps_coeff;
122         cxuv = cxy;
123
124         /* calculate and set horizontal fine scale (xsci) */
125
126         /* bypass the horizontal scaler ? */
127         if ( (in_x == out_x) && ( 1 == xpsc ) )
128                 xsci = 0x400;
129         else
130                 xsci = ( (1024 * in_x) / (out_x * xpsc) ) + xpsc;
131
132         /* set start phase for horizontal fine scale (xp) to 0 */
133         xp = 0;
134
135         /* set xim, if we bypass the horizontal scaler */
136         if ( 0x400 == xsci )
137                 xim = 1;
138         else
139                 xim = 0;
140
141         /* if the prescaler is bypassed, enable horizontal
142            accumulation mode (xacm) and clear dcgx */
143         if( 1 == xpsc ) {
144                 xacm = 1;
145                 dcgx = 0;
146         } else {
147                 xacm = 0;
148                 /* get best match in the table of attenuations
149                    for horizontal scaling */
150                 h_atten = hps_h_coeff_tab[( (xpsc - 1) < 63 ? (xpsc - 1) : 63 )].weight_sum;
151
152                 for (i = 0; h_attenuation[i] != 0; i++) {
153                         if (h_attenuation[i] >= h_atten)
154                                 break;
155                 }
156
157                 dcgx = i;
158         }
159
160         /* the horizontal scaling increment controls the UV filter
161            to reduce the bandwith to improve the display quality,
162            so set it ... */
163         if ( xsci == 0x400)
164                 pfuv = 0x00;
165         else if ( xsci < 0x600)
166                 pfuv = 0x01;
167         else if ( xsci < 0x680)
168                 pfuv = 0x11;
169         else if ( xsci < 0x700)
170                 pfuv = 0x22;
171         else
172                 pfuv = 0x33;
173
174
175         *hps_v_gain  &= MASK_W0|MASK_B2;
176         *hps_v_gain  |= (pfuv << 24);
177
178         *hps_h_scale    &= ~(MASK_W1 | 0xf000);
179         *hps_h_scale    |= (xim << 31) | (xp << 24) | (xsci << 12);
180
181         *hps_h_prescale |= (dcgx << 27) | ((xpsc-1) << 18) | (xacm << 17) | (cxy << 8) | (cxuv << 0);
182
183         return 0;
184 }
185
186 static struct {
187         u16 hps_coeff;
188         u16 weight_sum;
189 } hps_v_coeff_tab [] = {
190  {0x0100,   2},  {0x0102,   4},  {0x0300,   4},  {0x0106,   8},  {0x0502,   8},
191  {0x0708,   8},  {0x0F00,   8},  {0x011E,  16},  {0x110E,  16},  {0x1926,  16},
192  {0x3906,  16},  {0x3D42,  16},  {0x7D02,  16},  {0x7F80,  16},  {0xFF00,  16},
193  {0x01FE,  32},  {0x01FE,  32},  {0x817E,  32},  {0x817E,  32},  {0xC13E,  32},
194  {0xC13E,  32},  {0xE11E,  32},  {0xE11E,  32},  {0xF10E,  32},  {0xF10E,  32},
195  {0xF906,  32},  {0xF906,  32},  {0xFD02,  32},  {0xFD02,  32},  {0xFF00,  32},
196  {0xFF00,  32},  {0x01FE,  64},  {0x01FE,  64},  {0x01FE,  64},  {0x01FE,  64},
197  {0x01FE,  64},  {0x01FE,  64},  {0x01FE,  64},  {0x01FE,  64},  {0x01FE,  64},
198  {0x01FE,  64},  {0x01FE,  64},  {0x01FE,  64},  {0x01FE,  64},  {0x01FE,  64},
199  {0x01FE,  64},  {0x01FE,  64},  {0x01FE,  64},  {0x01FE,  64},  {0x817E,  64},
200  {0x817E,  64},  {0xC13E,  64},  {0xC13E,  64},  {0xE11E,  64},  {0xE11E,  64},
201  {0xF10E,  64},  {0xF10E,  64},  {0xF906,  64},  {0xF906,  64},  {0xFD02,  64},
202  {0xFD02,  64},  {0xFF00,  64},  {0xFF00,  64},  {0x01FE, 128}
203 };
204
205 /* table of attenuation values for vertical scaling */
206 static u16 v_attenuation[] = { 2, 4, 8, 16, 32, 64, 128, 256, 0};
207
208 /* calculate vertical scale registers */
209 static int calculate_v_scale_registers(struct saa7146_dev *dev, enum v4l2_field field,
210         int in_y, int out_y, u32* hps_v_scale, u32* hps_v_gain)
211 {
212         int lpi = 0;
213
214         /* vertical scaling */
215         u32 yacm = 0, ysci = 0, yacl = 0, ypo = 0, ype = 0;
216         /* vertical scale & gain */
217         u32 dcgy = 0, cya_cyb = 0;
218
219         /* helper variables */
220         u32 v_atten = 0, i = 0;
221
222         /* error, if vertical zooming */
223         if ( in_y < out_y ) {
224                 return -EINVAL;
225         }
226
227         /* linear phase interpolation may be used
228            if scaling is between 1 and 1/2 (both fields used)
229            or scaling is between 1/2 and 1/4 (if only one field is used) */
230
231         if (V4L2_FIELD_HAS_BOTH(field)) {
232                 if( 2*out_y >= in_y) {
233                         lpi = 1;
234                 }
235         } else if (field == V4L2_FIELD_TOP
236                 || field == V4L2_FIELD_ALTERNATE
237                 || field == V4L2_FIELD_BOTTOM) {
238                 if( 4*out_y >= in_y ) {
239                         lpi = 1;
240                 }
241                 out_y *= 2;
242         }
243         if( 0 != lpi ) {
244
245                 yacm = 0;
246                 yacl = 0;
247                 cya_cyb = 0x00ff;
248
249                 /* calculate scaling increment */
250                 if ( in_y > out_y )
251                         ysci = ((1024 * in_y) / (out_y + 1)) - 1024;
252                 else
253                         ysci = 0;
254
255                 dcgy = 0;
256
257                 /* calculate ype and ypo */
258                 ype = ysci / 16;
259                 ypo = ype + (ysci / 64);
260
261         } else {
262                 yacm = 1;
263
264                 /* calculate scaling increment */
265                 ysci = (((10 * 1024 * (in_y - out_y - 1)) / in_y) + 9) / 10;
266
267                 /* calculate ype and ypo */
268                 ypo = ype = ((ysci + 15) / 16);
269
270                 /* the sequence length interval (yacl) has to be set according
271                    to the prescale value, e.g.  [n   .. 1/2) : 0
272                                                 [1/2 .. 1/3) : 1
273                                                 [1/3 .. 1/4) : 2
274                                                 ... */
275                 if ( ysci < 512) {
276                         yacl = 0;
277                 } else {
278                         yacl = ( ysci / (1024 - ysci) );
279                 }
280
281                 /* get filter coefficients for cya, cyb from table hps_v_coeff_tab */
282                 cya_cyb = hps_v_coeff_tab[ (yacl < 63 ? yacl : 63 ) ].hps_coeff;
283
284                 /* get best match in the table of attenuations for vertical scaling */
285                 v_atten = hps_v_coeff_tab[ (yacl < 63 ? yacl : 63 ) ].weight_sum;
286
287                 for (i = 0; v_attenuation[i] != 0; i++) {
288                         if (v_attenuation[i] >= v_atten)
289                                 break;
290                 }
291
292                 dcgy = i;
293         }
294
295         /* ypo and ype swapped in spec ? */
296         *hps_v_scale    |= (yacm << 31) | (ysci << 21) | (yacl << 15) | (ypo << 8 ) | (ype << 1);
297
298         *hps_v_gain     &= ~(MASK_W0|MASK_B2);
299         *hps_v_gain     |= (dcgy << 16) | (cya_cyb << 0);
300
301         return 0;
302 }
303
304 /* simple bubble-sort algorithm with duplicate elimination */
305 static int sort_and_eliminate(u32* values, int* count)
306 {
307         int low = 0, high = 0, top = 0, temp = 0;
308         int cur = 0, next = 0;
309
310         /* sanity checks */
311         if( (0 > *count) || (NULL == values) ) {
312                 return -EINVAL;
313         }
314
315         /* bubble sort the first Â´count´ items of the array Â´values´ */
316         for( top = *count; top > 0; top--) {
317                 for( low = 0, high = 1; high < top; low++, high++) {
318                         if( values[low] > values[high] ) {
319                                 temp = values[low];
320                                 values[low] = values[high];
321                                 values[high] = temp;
322                         }
323                 }
324         }
325
326         /* remove duplicate items */
327         for( cur = 0, next = 1; next < *count; next++) {
328                 if( values[cur] != values[next])
329                         values[++cur] = values[next];
330         }
331
332         *count = cur + 1;
333
334         return 0;
335 }
336
337 static void calculate_clipping_registers_rect(struct saa7146_dev *dev, struct saa7146_fh *fh,
338         struct saa7146_video_dma *vdma2, u32* clip_format, u32* arbtr_ctrl, enum v4l2_field field)
339 {
340         struct saa7146_vv *vv = dev->vv_data;
341         u32 *clipping = vv->d_clipping.cpu_addr;
342
343         int width = fh->ov.win.w.width;
344         int height =  fh->ov.win.w.height;
345         int clipcount = fh->ov.nclips;
346
347         u32 line_list[32];
348         u32 pixel_list[32];
349         int numdwords = 0;
350
351         int i = 0, j = 0;
352         int cnt_line = 0, cnt_pixel = 0;
353
354         int x[32], y[32], w[32], h[32];
355
356         /* clear out memory */
357         memset(&line_list[0],  0x00, sizeof(u32)*32);
358         memset(&pixel_list[0], 0x00, sizeof(u32)*32);
359         memset(clipping,  0x00, SAA7146_CLIPPING_MEM);
360
361         /* fill the line and pixel-lists */
362         for(i = 0; i < clipcount; i++) {
363                 int l = 0, r = 0, t = 0, b = 0;
364
365                 x[i] = fh->ov.clips[i].c.left;
366                 y[i] = fh->ov.clips[i].c.top;
367                 w[i] = fh->ov.clips[i].c.width;
368                 h[i] = fh->ov.clips[i].c.height;
369
370                 if( w[i] < 0) {
371                         x[i] += w[i]; w[i] = -w[i];
372                 }
373                 if( h[i] < 0) {
374                         y[i] += h[i]; h[i] = -h[i];
375                 }
376                 if( x[i] < 0) {
377                         w[i] += x[i]; x[i] = 0;
378                 }
379                 if( y[i] < 0) {
380                         h[i] += y[i]; y[i] = 0;
381                 }
382                 if( 0 != vv->vflip ) {
383                         y[i] = height - y[i] - h[i];
384                 }
385
386                 l = x[i];
387                 r = x[i]+w[i];
388                 t = y[i];
389                 b = y[i]+h[i];
390
391                 /* insert left/right coordinates */
392                 pixel_list[ 2*i   ] = min_t(int, l, width);
393                 pixel_list[(2*i)+1] = min_t(int, r, width);
394                 /* insert top/bottom coordinates */
395                 line_list[ 2*i   ] = min_t(int, t, height);
396                 line_list[(2*i)+1] = min_t(int, b, height);
397         }
398
399         /* sort and eliminate lists */
400         cnt_line = cnt_pixel = 2*clipcount;
401         sort_and_eliminate( &pixel_list[0], &cnt_pixel );
402         sort_and_eliminate( &line_list[0], &cnt_line );
403
404         /* calculate the number of used u32s */
405         numdwords = max_t(int, (cnt_line+1), (cnt_pixel+1))*2;
406         numdwords = max_t(int, 4, numdwords);
407         numdwords = min_t(int, 64, numdwords);
408
409         /* fill up cliptable */
410         for(i = 0; i < cnt_pixel; i++) {
411                 clipping[2*i] |= cpu_to_le32(pixel_list[i] << 16);
412         }
413         for(i = 0; i < cnt_line; i++) {
414                 clipping[(2*i)+1] |= cpu_to_le32(line_list[i] << 16);
415         }
416
417         /* fill up cliptable with the display infos */
418         for(j = 0; j < clipcount; j++) {
419
420                 for(i = 0; i < cnt_pixel; i++) {
421
422                         if( x[j] < 0)
423                                 x[j] = 0;
424
425                         if( pixel_list[i] < (x[j] + w[j])) {
426
427                                 if ( pixel_list[i] >= x[j] ) {
428                                         clipping[2*i] |= cpu_to_le32(1 << j);
429                                 }
430                         }
431                 }
432                 for(i = 0; i < cnt_line; i++) {
433
434                         if( y[j] < 0)
435                                 y[j] = 0;
436
437                         if( line_list[i] < (y[j] + h[j]) ) {
438
439                                 if( line_list[i] >= y[j] ) {
440                                         clipping[(2*i)+1] |= cpu_to_le32(1 << j);
441                                 }
442                         }
443                 }
444         }
445
446         /* adjust arbitration control register */
447         *arbtr_ctrl &= 0xffff00ff;
448         *arbtr_ctrl |= 0x00001c00;
449
450         vdma2->base_even        = vv->d_clipping.dma_handle;
451         vdma2->base_odd         = vv->d_clipping.dma_handle;
452         vdma2->prot_addr        = vv->d_clipping.dma_handle+((sizeof(u32))*(numdwords));
453         vdma2->base_page        = 0x04;
454         vdma2->pitch            = 0x00;
455         vdma2->num_line_byte    = (0 << 16 | (sizeof(u32))*(numdwords-1) );
456
457         /* set clipping-mode. this depends on the field(s) used */
458         *clip_format &= 0xfffffff7;
459         if (V4L2_FIELD_HAS_BOTH(field)) {
460                 *clip_format |= 0x00000008;
461         } else {
462                 *clip_format |= 0x00000000;
463         }
464 }
465
466 /* disable clipping */
467 static void saa7146_disable_clipping(struct saa7146_dev *dev)
468 {
469         u32 clip_format = saa7146_read(dev, CLIP_FORMAT_CTRL);
470
471         /* mask out relevant bits (=lower word)*/
472         clip_format &= MASK_W1;
473
474         /* upload clipping-registers*/
475         saa7146_write(dev, CLIP_FORMAT_CTRL,clip_format);
476         saa7146_write(dev, MC2, (MASK_05 | MASK_21));
477
478         /* disable video dma2 */
479         saa7146_write(dev, MC1, MASK_21);
480 }
481
482 static void saa7146_set_clipping_rect(struct saa7146_fh *fh)
483 {
484         struct saa7146_dev *dev = fh->dev;
485         enum v4l2_field field = fh->ov.win.field;
486         struct  saa7146_video_dma vdma2;
487         u32 clip_format;
488         u32 arbtr_ctrl;
489
490         /* check clipcount, disable clipping if clipcount == 0*/
491         if( fh->ov.nclips == 0 ) {
492                 saa7146_disable_clipping(dev);
493                 return;
494         }
495
496         clip_format = saa7146_read(dev, CLIP_FORMAT_CTRL);
497         arbtr_ctrl = saa7146_read(dev, PCI_BT_V1);
498
499         calculate_clipping_registers_rect(dev, fh, &vdma2, &clip_format, &arbtr_ctrl, field);
500
501         /* set clipping format */
502         clip_format &= 0xffff0008;
503         clip_format |= (SAA7146_CLIPPING_RECT << 4);
504
505         /* prepare video dma2 */
506         saa7146_write(dev, BASE_EVEN2,          vdma2.base_even);
507         saa7146_write(dev, BASE_ODD2,           vdma2.base_odd);
508         saa7146_write(dev, PROT_ADDR2,          vdma2.prot_addr);
509         saa7146_write(dev, BASE_PAGE2,          vdma2.base_page);
510         saa7146_write(dev, PITCH2,              vdma2.pitch);
511         saa7146_write(dev, NUM_LINE_BYTE2,      vdma2.num_line_byte);
512
513         /* prepare the rest */
514         saa7146_write(dev, CLIP_FORMAT_CTRL,clip_format);
515         saa7146_write(dev, PCI_BT_V1, arbtr_ctrl);
516
517         /* upload clip_control-register, clipping-registers, enable video dma2 */
518         saa7146_write(dev, MC2, (MASK_05 | MASK_21 | MASK_03 | MASK_19));
519         saa7146_write(dev, MC1, (MASK_05 | MASK_21));
520 }
521
522 static void saa7146_set_window(struct saa7146_dev *dev, int width, int height, enum v4l2_field field)
523 {
524         struct saa7146_vv *vv = dev->vv_data;
525
526         int source = vv->current_hps_source;
527         int sync = vv->current_hps_sync;
528
529         u32 hps_v_scale = 0, hps_v_gain  = 0, hps_ctrl = 0, hps_h_prescale = 0, hps_h_scale = 0;
530
531         /* set vertical scale */
532         hps_v_scale = 0; /* all bits get set by the function-call */
533         hps_v_gain  = 0; /* fixme: saa7146_read(dev, HPS_V_GAIN);*/
534         calculate_v_scale_registers(dev, field, vv->standard->v_field*2, height, &hps_v_scale, &hps_v_gain);
535
536         /* set horizontal scale */
537         hps_ctrl        = 0;
538         hps_h_prescale  = 0; /* all bits get set in the function */
539         hps_h_scale     = 0;
540         calculate_h_scale_registers(dev, vv->standard->h_pixels, width, vv->hflip, &hps_ctrl, &hps_v_gain, &hps_h_prescale, &hps_h_scale);
541
542         /* set hyo and hxo */
543         calculate_hxo_and_hyo(vv, &hps_h_scale, &hps_ctrl);
544         calculate_hps_source_and_sync(dev, source, sync, &hps_ctrl);
545
546         /* write out new register contents */
547         saa7146_write(dev, HPS_V_SCALE, hps_v_scale);
548         saa7146_write(dev, HPS_V_GAIN,  hps_v_gain);
549         saa7146_write(dev, HPS_CTRL,    hps_ctrl);
550         saa7146_write(dev, HPS_H_PRESCALE,hps_h_prescale);
551         saa7146_write(dev, HPS_H_SCALE, hps_h_scale);
552
553         /* upload shadow-ram registers */
554         saa7146_write(dev, MC2, (MASK_05 | MASK_06 | MASK_21 | MASK_22) );
555 }
556
557 /* calculate the new memory offsets for a desired position */
558 static void saa7146_set_position(struct saa7146_dev *dev, int w_x, int w_y, int w_height, enum v4l2_field field, u32 pixelformat)
559 {
560         struct saa7146_vv *vv = dev->vv_data;
561         struct saa7146_format *sfmt = format_by_fourcc(dev, pixelformat);
562
563         int b_depth = vv->ov_fmt->depth;
564         int b_bpl = vv->ov_fb.fmt.bytesperline;
565         u32 base = (u32)vv->ov_fb.base;
566
567         struct  saa7146_video_dma vdma1;
568
569         /* calculate memory offsets for picture, look if we shall top-down-flip */
570         vdma1.pitch     = 2*b_bpl;
571         if ( 0 == vv->vflip ) {
572                 vdma1.base_even = (u32)base + (w_y * (vdma1.pitch/2)) + (w_x * (b_depth / 8));
573                 vdma1.base_odd  = vdma1.base_even + (vdma1.pitch / 2);
574                 vdma1.prot_addr = vdma1.base_even + (w_height * (vdma1.pitch / 2));
575         }
576         else {
577                 vdma1.base_even = (u32)base + ((w_y+w_height) * (vdma1.pitch/2)) + (w_x * (b_depth / 8));
578                 vdma1.base_odd  = vdma1.base_even - (vdma1.pitch / 2);
579                 vdma1.prot_addr = vdma1.base_odd - (w_height * (vdma1.pitch / 2));
580         }
581
582         if (V4L2_FIELD_HAS_BOTH(field)) {
583         } else if (field == V4L2_FIELD_ALTERNATE) {
584                 /* fixme */
585                 vdma1.base_odd = vdma1.prot_addr;
586                 vdma1.pitch /= 2;
587         } else if (field == V4L2_FIELD_TOP) {
588                 vdma1.base_odd = vdma1.prot_addr;
589                 vdma1.pitch /= 2;
590         } else if (field == V4L2_FIELD_BOTTOM) {
591                 vdma1.base_odd = vdma1.base_even;
592                 vdma1.base_even = vdma1.prot_addr;
593                 vdma1.pitch /= 2;
594         }
595
596         if ( 0 != vv->vflip ) {
597                 vdma1.pitch *= -1;
598         }
599
600         vdma1.base_page = sfmt->swap;
601         vdma1.num_line_byte = (vv->standard->v_field<<16)+vv->standard->h_pixels;
602
603         saa7146_write_out_dma(dev, 1, &vdma1);
604 }
605
606 static void saa7146_set_output_format(struct saa7146_dev *dev, unsigned long palette)
607 {
608         u32 clip_format = saa7146_read(dev, CLIP_FORMAT_CTRL);
609
610         /* call helper function */
611         calculate_output_format_register(dev,palette,&clip_format);
612
613         /* update the hps registers */
614         saa7146_write(dev, CLIP_FORMAT_CTRL, clip_format);
615         saa7146_write(dev, MC2, (MASK_05 | MASK_21));
616 }
617
618 /* select input-source */
619 void saa7146_set_hps_source_and_sync(struct saa7146_dev *dev, int source, int sync)
620 {
621         struct saa7146_vv *vv = dev->vv_data;
622         u32 hps_ctrl = 0;
623
624         /* read old state */
625         hps_ctrl = saa7146_read(dev, HPS_CTRL);
626
627         hps_ctrl &= ~( MASK_31 | MASK_30 | MASK_28 );
628         hps_ctrl |= (source << 30) | (sync << 28);
629
630         /* write back & upload register */
631         saa7146_write(dev, HPS_CTRL, hps_ctrl);
632         saa7146_write(dev, MC2, (MASK_05 | MASK_21));
633
634         vv->current_hps_source = source;
635         vv->current_hps_sync = sync;
636 }
637
638 int saa7146_enable_overlay(struct saa7146_fh *fh)
639 {
640         struct saa7146_dev *dev = fh->dev;
641         struct saa7146_vv *vv = dev->vv_data;
642
643         saa7146_set_window(dev, fh->ov.win.w.width, fh->ov.win.w.height, fh->ov.win.field);
644         saa7146_set_position(dev, fh->ov.win.w.left, fh->ov.win.w.top, fh->ov.win.w.height, fh->ov.win.field, vv->ov_fmt->pixelformat);
645         saa7146_set_output_format(dev, vv->ov_fmt->trans);
646         saa7146_set_clipping_rect(fh);
647
648         /* enable video dma1 */
649         saa7146_write(dev, MC1, (MASK_06 | MASK_22));
650         return 0;
651 }
652
653 void saa7146_disable_overlay(struct saa7146_fh *fh)
654 {
655         struct saa7146_dev *dev = fh->dev;
656
657         /* disable clipping + video dma1 */
658         saa7146_disable_clipping(dev);
659         saa7146_write(dev, MC1, MASK_22);
660 }
661
662 void saa7146_write_out_dma(struct saa7146_dev* dev, int which, struct saa7146_video_dma* vdma)
663 {
664         int where = 0;
665
666         if( which < 1 || which > 3) {
667                 return;
668         }
669
670         /* calculate starting address */
671         where  = (which-1)*0x18;
672
673         saa7146_write(dev, where,       vdma->base_odd);
674         saa7146_write(dev, where+0x04,  vdma->base_even);
675         saa7146_write(dev, where+0x08,  vdma->prot_addr);
676         saa7146_write(dev, where+0x0c,  vdma->pitch);
677         saa7146_write(dev, where+0x10,  vdma->base_page);
678         saa7146_write(dev, where+0x14,  vdma->num_line_byte);
679
680         /* upload */
681         saa7146_write(dev, MC2, (MASK_02<<(which-1))|(MASK_18<<(which-1)));
682 /*
683         printk("vdma%d.base_even:     0x%08x\n", which,vdma->base_even);
684         printk("vdma%d.base_odd:      0x%08x\n", which,vdma->base_odd);
685         printk("vdma%d.prot_addr:     0x%08x\n", which,vdma->prot_addr);
686         printk("vdma%d.base_page:     0x%08x\n", which,vdma->base_page);
687         printk("vdma%d.pitch:         0x%08x\n", which,vdma->pitch);
688         printk("vdma%d.num_line_byte: 0x%08x\n", which,vdma->num_line_byte);
689 */
690 }
691
692 static int calculate_video_dma_grab_packed(struct saa7146_dev* dev, struct saa7146_buf *buf)
693 {
694         struct saa7146_vv *vv = dev->vv_data;
695         struct saa7146_video_dma vdma1;
696
697         struct saa7146_format *sfmt = format_by_fourcc(dev,buf->fmt->pixelformat);
698
699         int width = buf->fmt->width;
700         int height = buf->fmt->height;
701         int bytesperline = buf->fmt->bytesperline;
702         enum v4l2_field field = buf->fmt->field;
703
704         int depth = sfmt->depth;
705
706         DEB_CAP(("[size=%dx%d,fields=%s]\n",
707                 width,height,v4l2_field_names[field]));
708
709         if( bytesperline != 0) {
710                 vdma1.pitch = bytesperline*2;
711         } else {
712                 vdma1.pitch = (width*depth*2)/8;
713         }
714         vdma1.num_line_byte     = ((vv->standard->v_field<<16) + vv->standard->h_pixels);
715         vdma1.base_page         = buf->pt[0].dma | ME1 | sfmt->swap;
716
717         if( 0 != vv->vflip ) {
718                 vdma1.prot_addr = buf->pt[0].offset;
719                 vdma1.base_even = buf->pt[0].offset+(vdma1.pitch/2)*height;
720                 vdma1.base_odd  = vdma1.base_even - (vdma1.pitch/2);
721         } else {
722                 vdma1.base_even = buf->pt[0].offset;
723                 vdma1.base_odd  = vdma1.base_even + (vdma1.pitch/2);
724                 vdma1.prot_addr = buf->pt[0].offset+(vdma1.pitch/2)*height;
725         }
726
727         if (V4L2_FIELD_HAS_BOTH(field)) {
728         } else if (field == V4L2_FIELD_ALTERNATE) {
729                 /* fixme */
730                 if ( vv->last_field == V4L2_FIELD_TOP ) {
731                         vdma1.base_odd  = vdma1.prot_addr;
732                         vdma1.pitch /= 2;
733                 } else if ( vv->last_field == V4L2_FIELD_BOTTOM ) {
734                         vdma1.base_odd  = vdma1.base_even;
735                         vdma1.base_even = vdma1.prot_addr;
736                         vdma1.pitch /= 2;
737                 }
738         } else if (field == V4L2_FIELD_TOP) {
739                 vdma1.base_odd  = vdma1.prot_addr;
740                 vdma1.pitch /= 2;
741         } else if (field == V4L2_FIELD_BOTTOM) {
742                 vdma1.base_odd  = vdma1.base_even;
743                 vdma1.base_even = vdma1.prot_addr;
744                 vdma1.pitch /= 2;
745         }
746
747         if( 0 != vv->vflip ) {
748                 vdma1.pitch *= -1;
749         }
750
751         saa7146_write_out_dma(dev, 1, &vdma1);
752         return 0;
753 }
754
755 static int calc_planar_422(struct saa7146_vv *vv, struct saa7146_buf *buf, struct saa7146_video_dma *vdma2, struct saa7146_video_dma *vdma3)
756 {
757         int height = buf->fmt->height;
758         int width = buf->fmt->width;
759
760         vdma2->pitch    = width;
761         vdma3->pitch    = width;
762
763         /* fixme: look at bytesperline! */
764
765         if( 0 != vv->vflip ) {
766                 vdma2->prot_addr        = buf->pt[1].offset;
767                 vdma2->base_even        = ((vdma2->pitch/2)*height)+buf->pt[1].offset;
768                 vdma2->base_odd         = vdma2->base_even - (vdma2->pitch/2);
769
770                 vdma3->prot_addr        = buf->pt[2].offset;
771                 vdma3->base_even        = ((vdma3->pitch/2)*height)+buf->pt[2].offset;
772                 vdma3->base_odd         = vdma3->base_even - (vdma3->pitch/2);
773         } else {
774                 vdma3->base_even        = buf->pt[2].offset;
775                 vdma3->base_odd         = vdma3->base_even + (vdma3->pitch/2);
776                 vdma3->prot_addr        = (vdma3->pitch/2)*height+buf->pt[2].offset;
777
778                 vdma2->base_even        = buf->pt[1].offset;
779                 vdma2->base_odd         = vdma2->base_even + (vdma2->pitch/2);
780                 vdma2->prot_addr        = (vdma2->pitch/2)*height+buf->pt[1].offset;
781         }
782
783         return 0;
784 }
785
786 static int calc_planar_420(struct saa7146_vv *vv, struct saa7146_buf *buf, struct saa7146_video_dma *vdma2, struct saa7146_video_dma *vdma3)
787 {
788         int height = buf->fmt->height;
789         int width = buf->fmt->width;
790
791         vdma2->pitch    = width/2;
792         vdma3->pitch    = width/2;
793
794         if( 0 != vv->vflip ) {
795                 vdma2->prot_addr        = buf->pt[2].offset;
796                 vdma2->base_even        = ((vdma2->pitch/2)*height)+buf->pt[2].offset;
797                 vdma2->base_odd         = vdma2->base_even - (vdma2->pitch/2);
798
799                 vdma3->prot_addr        = buf->pt[1].offset;
800                 vdma3->base_even        = ((vdma3->pitch/2)*height)+buf->pt[1].offset;
801                 vdma3->base_odd         = vdma3->base_even - (vdma3->pitch/2);
802
803         } else {
804                 vdma3->base_even        = buf->pt[2].offset;
805                 vdma3->base_odd         = vdma3->base_even + (vdma3->pitch);
806                 vdma3->prot_addr        = (vdma3->pitch/2)*height+buf->pt[2].offset;
807
808                 vdma2->base_even        = buf->pt[1].offset;
809                 vdma2->base_odd         = vdma2->base_even + (vdma2->pitch);
810                 vdma2->prot_addr        = (vdma2->pitch/2)*height+buf->pt[1].offset;
811         }
812         return 0;
813 }
814
815 static int calculate_video_dma_grab_planar(struct saa7146_dev* dev, struct saa7146_buf *buf)
816 {
817         struct saa7146_vv *vv = dev->vv_data;
818         struct saa7146_video_dma vdma1;
819         struct saa7146_video_dma vdma2;
820         struct saa7146_video_dma vdma3;
821
822         struct saa7146_format *sfmt = format_by_fourcc(dev,buf->fmt->pixelformat);
823
824         int width = buf->fmt->width;
825         int height = buf->fmt->height;
826         enum v4l2_field field = buf->fmt->field;
827
828         BUG_ON(0 == buf->pt[0].dma);
829         BUG_ON(0 == buf->pt[1].dma);
830         BUG_ON(0 == buf->pt[2].dma);
831
832         DEB_CAP(("[size=%dx%d,fields=%s]\n",
833                 width,height,v4l2_field_names[field]));
834
835         /* fixme: look at bytesperline! */
836
837         /* fixme: what happens for user space buffers here?. The offsets are
838            most likely wrong, this version here only works for page-aligned
839            buffers, modifications to the pagetable-functions are necessary...*/
840
841         vdma1.pitch             = width*2;
842         vdma1.num_line_byte     = ((vv->standard->v_field<<16) + vv->standard->h_pixels);
843         vdma1.base_page         = buf->pt[0].dma | ME1;
844
845         if( 0 != vv->vflip ) {
846                 vdma1.prot_addr = buf->pt[0].offset;
847                 vdma1.base_even = ((vdma1.pitch/2)*height)+buf->pt[0].offset;
848                 vdma1.base_odd  = vdma1.base_even - (vdma1.pitch/2);
849         } else {
850                 vdma1.base_even = buf->pt[0].offset;
851                 vdma1.base_odd  = vdma1.base_even + (vdma1.pitch/2);
852                 vdma1.prot_addr = (vdma1.pitch/2)*height+buf->pt[0].offset;
853         }
854
855         vdma2.num_line_byte     = 0; /* unused */
856         vdma2.base_page         = buf->pt[1].dma | ME1;
857
858         vdma3.num_line_byte     = 0; /* unused */
859         vdma3.base_page         = buf->pt[2].dma | ME1;
860
861         switch( sfmt->depth ) {
862                 case 12: {
863                         calc_planar_420(vv,buf,&vdma2,&vdma3);
864                         break;
865                 }
866                 case 16: {
867                         calc_planar_422(vv,buf,&vdma2,&vdma3);
868                         break;
869                 }
870                 default: {
871                         return -1;
872                 }
873         }
874
875         if (V4L2_FIELD_HAS_BOTH(field)) {
876         } else if (field == V4L2_FIELD_ALTERNATE) {
877                 /* fixme */
878                 vdma1.base_odd  = vdma1.prot_addr;
879                 vdma1.pitch /= 2;
880                 vdma2.base_odd  = vdma2.prot_addr;
881                 vdma2.pitch /= 2;
882                 vdma3.base_odd  = vdma3.prot_addr;
883                 vdma3.pitch /= 2;
884         } else if (field == V4L2_FIELD_TOP) {
885                 vdma1.base_odd  = vdma1.prot_addr;
886                 vdma1.pitch /= 2;
887                 vdma2.base_odd  = vdma2.prot_addr;
888                 vdma2.pitch /= 2;
889                 vdma3.base_odd  = vdma3.prot_addr;
890                 vdma3.pitch /= 2;
891         } else if (field == V4L2_FIELD_BOTTOM) {
892                 vdma1.base_odd  = vdma1.base_even;
893                 vdma1.base_even = vdma1.prot_addr;
894                 vdma1.pitch /= 2;
895                 vdma2.base_odd  = vdma2.base_even;
896                 vdma2.base_even = vdma2.prot_addr;
897                 vdma2.pitch /= 2;
898                 vdma3.base_odd  = vdma3.base_even;
899                 vdma3.base_even = vdma3.prot_addr;
900                 vdma3.pitch /= 2;
901         }
902
903         if( 0 != vv->vflip ) {
904                 vdma1.pitch *= -1;
905                 vdma2.pitch *= -1;
906                 vdma3.pitch *= -1;
907         }
908
909         saa7146_write_out_dma(dev, 1, &vdma1);
910         if( (sfmt->flags & FORMAT_BYTE_SWAP) != 0 ) {
911                 saa7146_write_out_dma(dev, 3, &vdma2);
912                 saa7146_write_out_dma(dev, 2, &vdma3);
913         } else {
914                 saa7146_write_out_dma(dev, 2, &vdma2);
915                 saa7146_write_out_dma(dev, 3, &vdma3);
916         }
917         return 0;
918 }
919
920 static void program_capture_engine(struct saa7146_dev *dev, int planar)
921 {
922         struct saa7146_vv *vv = dev->vv_data;
923         int count = 0;
924
925         unsigned long e_wait = vv->current_hps_sync == SAA7146_HPS_SYNC_PORT_A ? CMD_E_FID_A : CMD_E_FID_B;
926         unsigned long o_wait = vv->current_hps_sync == SAA7146_HPS_SYNC_PORT_A ? CMD_O_FID_A : CMD_O_FID_B;
927
928         /* wait for o_fid_a/b / e_fid_a/b toggle only if rps register 0 is not set*/
929         WRITE_RPS0(CMD_PAUSE | CMD_OAN | CMD_SIG0 | o_wait);
930         WRITE_RPS0(CMD_PAUSE | CMD_OAN | CMD_SIG0 | e_wait);
931
932         /* set rps register 0 */
933         WRITE_RPS0(CMD_WR_REG | (1 << 8) | (MC2/4));
934         WRITE_RPS0(MASK_27 | MASK_11);
935
936         /* turn on video-dma1 */
937         WRITE_RPS0(CMD_WR_REG_MASK | (MC1/4));
938         WRITE_RPS0(MASK_06 | MASK_22);                  /* => mask */
939         WRITE_RPS0(MASK_06 | MASK_22);                  /* => values */
940         if( 0 != planar ) {
941                 /* turn on video-dma2 */
942                 WRITE_RPS0(CMD_WR_REG_MASK | (MC1/4));
943                 WRITE_RPS0(MASK_05 | MASK_21);                  /* => mask */
944                 WRITE_RPS0(MASK_05 | MASK_21);                  /* => values */
945
946                 /* turn on video-dma3 */
947                 WRITE_RPS0(CMD_WR_REG_MASK | (MC1/4));
948                 WRITE_RPS0(MASK_04 | MASK_20);                  /* => mask */
949                 WRITE_RPS0(MASK_04 | MASK_20);                  /* => values */
950         }
951
952         /* wait for o_fid_a/b / e_fid_a/b toggle */
953         if ( vv->last_field == V4L2_FIELD_INTERLACED ) {
954                 WRITE_RPS0(CMD_PAUSE | o_wait);
955                 WRITE_RPS0(CMD_PAUSE | e_wait);
956         } else if ( vv->last_field == V4L2_FIELD_TOP ) {
957                 WRITE_RPS0(CMD_PAUSE | (vv->current_hps_sync == SAA7146_HPS_SYNC_PORT_A ? MASK_10 : MASK_09));
958                 WRITE_RPS0(CMD_PAUSE | o_wait);
959         } else if ( vv->last_field == V4L2_FIELD_BOTTOM ) {
960                 WRITE_RPS0(CMD_PAUSE | (vv->current_hps_sync == SAA7146_HPS_SYNC_PORT_A ? MASK_10 : MASK_09));
961                 WRITE_RPS0(CMD_PAUSE | e_wait);
962         }
963
964         /* turn off video-dma1 */
965         WRITE_RPS0(CMD_WR_REG_MASK | (MC1/4));
966         WRITE_RPS0(MASK_22 | MASK_06);                  /* => mask */
967         WRITE_RPS0(MASK_22);                            /* => values */
968         if( 0 != planar ) {
969                 /* turn off video-dma2 */
970                 WRITE_RPS0(CMD_WR_REG_MASK | (MC1/4));
971                 WRITE_RPS0(MASK_05 | MASK_21);                  /* => mask */
972                 WRITE_RPS0(MASK_21);                            /* => values */
973
974                 /* turn off video-dma3 */
975                 WRITE_RPS0(CMD_WR_REG_MASK | (MC1/4));
976                 WRITE_RPS0(MASK_04 | MASK_20);                  /* => mask */
977                 WRITE_RPS0(MASK_20);                            /* => values */
978         }
979
980         /* generate interrupt */
981         WRITE_RPS0(CMD_INTERRUPT);
982
983         /* stop */
984         WRITE_RPS0(CMD_STOP);
985 }
986
987 void saa7146_set_capture(struct saa7146_dev *dev, struct saa7146_buf *buf, struct saa7146_buf *next)
988 {
989         struct saa7146_format *sfmt = format_by_fourcc(dev,buf->fmt->pixelformat);
990         struct saa7146_vv *vv = dev->vv_data;
991         u32 vdma1_prot_addr;
992
993         DEB_CAP(("buf:%p, next:%p\n",buf,next));
994
995         vdma1_prot_addr = saa7146_read(dev, PROT_ADDR1);
996         if( 0 == vdma1_prot_addr ) {
997                 /* clear out beginning of streaming bit (rps register 0)*/
998                 DEB_CAP(("forcing sync to new frame\n"));
999                 saa7146_write(dev, MC2, MASK_27 );
1000         }
1001
1002         saa7146_set_window(dev, buf->fmt->width, buf->fmt->height, buf->fmt->field);
1003         saa7146_set_output_format(dev, sfmt->trans);
1004         saa7146_disable_clipping(dev);
1005
1006         if ( vv->last_field == V4L2_FIELD_INTERLACED ) {
1007         } else if ( vv->last_field == V4L2_FIELD_TOP ) {
1008                 vv->last_field = V4L2_FIELD_BOTTOM;
1009         } else if ( vv->last_field == V4L2_FIELD_BOTTOM ) {
1010                 vv->last_field = V4L2_FIELD_TOP;
1011         }
1012
1013         if( 0 != IS_PLANAR(sfmt->trans)) {
1014                 calculate_video_dma_grab_planar(dev, buf);
1015                 program_capture_engine(dev,1);
1016         } else {
1017                 calculate_video_dma_grab_packed(dev, buf);
1018                 program_capture_engine(dev,0);
1019         }
1020
1021 /*
1022         printk("vdma%d.base_even:     0x%08x\n", 1,saa7146_read(dev,BASE_EVEN1));
1023         printk("vdma%d.base_odd:      0x%08x\n", 1,saa7146_read(dev,BASE_ODD1));
1024         printk("vdma%d.prot_addr:     0x%08x\n", 1,saa7146_read(dev,PROT_ADDR1));
1025         printk("vdma%d.base_page:     0x%08x\n", 1,saa7146_read(dev,BASE_PAGE1));
1026         printk("vdma%d.pitch:         0x%08x\n", 1,saa7146_read(dev,PITCH1));
1027         printk("vdma%d.num_line_byte: 0x%08x\n", 1,saa7146_read(dev,NUM_LINE_BYTE1));
1028         printk("vdma%d => vptr      : 0x%08x\n", 1,saa7146_read(dev,PCI_VDP1));
1029 */
1030
1031         /* write the address of the rps-program */
1032         saa7146_write(dev, RPS_ADDR0, dev->d_rps0.dma_handle);
1033
1034         /* turn on rps */
1035         saa7146_write(dev, MC1, (MASK_12 | MASK_28));
1036 }