blob: 2d4d006f40d1a11b7a46be276255259e35100cfb [file] [log] [blame]
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301/******************************************************************************
2 *
3 * Copyright (C) 2015 The Android Open Source Project
4 *
5 * Licensed under the Apache License, Version 2.0 (the "License");
6 * you may not use this file except in compliance with the License.
7 * You may obtain a copy of the License at:
8 *
9 * http://www.apache.org/licenses/LICENSE-2.0
10 *
11 * Unless required by applicable law or agreed to in writing, software
12 * distributed under the License is distributed on an "AS IS" BASIS,
13 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14 * See the License for the specific language governing permissions and
15 * limitations under the License.
16 *
17 *****************************************************************************
18 * Originally developed and contributed by Ittiam Systems Pvt. Ltd, Bangalore
19*/
20/*!
21 **************************************************************************
22 * \file ih264d_process_intra_mb.c
23 *
24 * \brief
25 * Contains routines that decode a I slice type
26 *
27 * Detailed_description
28 *
29 * \date
30 * 07/07/2003
31 *
32 * \author NS
33 **************************************************************************
34 */
35
36#include <string.h>
37#include "ih264d_bitstrm.h"
38#include "ih264d_defs.h"
39#include "ih264d_debug.h"
40#include "ih264d_tables.h"
41#include "ih264d_structs.h"
42#include "ih264d_defs.h"
43#include "ih264d_parse_cavlc.h"
44#include "ih264d_mb_utils.h"
45#include "ih264d_parse_slice.h"
46#include "ih264d_process_intra_mb.h"
47#include "ih264d_error_handler.h"
48#include "ih264d_quant_scaling.h"
49#include "ih264d_tables.h"
50
51/*!
52 **************************************************************************
53 * \if Function name : ih264d_itrans_recon_luma_dc \endif
54 *
55 * \brief
56 * This function does InvTransform, scaling and reconstruction of Luma DC.
57 *
58 * \return
59 * 0 on Success and Error code otherwise
60 **************************************************************************
61 */
62void ih264d_itrans_recon_luma_dc(dec_struct_t *ps_dec,
63 WORD16* pi2_src,
64 WORD16* pi2_coeff_block,
65 const UWORD16 *pu2_weigh_mat)
66{
67 WORD32 i;
68 WORD16 pi2_out[16];
69 WORD32 pi4_tmp[16];
70 WORD16 *pi2_out_ptr = &pi2_out[0];
71 PROFILE_DISABLE_IQ_IT_RECON_RETURN()
72 ps_dec->pf_ihadamard_scaling_4x4(pi2_src, pi2_out,
73 ps_dec->pu2_quant_scale_y, pu2_weigh_mat,
74 ps_dec->u1_qp_y_div6, pi4_tmp);
75 for(i = 0; i < 4; i++)
76 {
77 pi2_coeff_block[0] = pi2_out_ptr[0];
78 pi2_coeff_block[4 * 16] = pi2_out_ptr[4];
79 pi2_coeff_block[8 * 16] = pi2_out_ptr[8];
80 pi2_coeff_block[12 * 16] = pi2_out_ptr[12];
81
82 pi2_out_ptr++; /* Point to next column */
83 pi2_coeff_block += 16;
84 }
85}
86/*!
87 **************************************************************************
88 * \if Function name : ih264d_read_intra_pred_modes \endif
89 *
90 * \brief
91 * Reads the intra pred mode related values of I4x4 MB from bitstream.
92 *
93 * This function will read the prev intra pred mode flags and
94 * stores it in pu1_prev_intra4x4_pred_mode_flag. If the u4_flag
95 * indicates that most probable mode is not intra pred mode, then
96 * the rem_intra4x4_pred_mode is read and stored in
97 * pu1_rem_intra4x4_pred_mode array.
98 *
99 *
100 * \return
101 * 0 on success and Error code otherwise
102 *
103 **************************************************************************
104 */
105WORD32 ih264d_read_intra_pred_modes(dec_struct_t * ps_dec,
106 UWORD8 * pu1_prev_intra4x4_pred_mode_flag,
107 UWORD8 * pu1_rem_intra4x4_pred_mode,
108 UWORD32 u4_trans_form8x8)
109{
110 WORD32 i4x4_luma_blk_idx = 0, i8x8_luma_blk_idx = 0;
111
112 dec_bit_stream_t * ps_bitstrm = ps_dec->ps_bitstrm;
113
114 if(!u4_trans_form8x8)
115 {
116 for(i4x4_luma_blk_idx = 0; i4x4_luma_blk_idx < 16; ++i4x4_luma_blk_idx)
117 {
118 UWORD32 u4_temp;
119 SWITCHOFFTRACE;
120
121 GETBIT(u4_temp, ps_bitstrm->u4_ofst, ps_bitstrm->pu4_buffer);
122 *pu1_prev_intra4x4_pred_mode_flag = (UWORD8)u4_temp;
123 if(!(*pu1_prev_intra4x4_pred_mode_flag))
124 {
125 GETBITS(u4_temp, ps_bitstrm->u4_ofst, ps_bitstrm->pu4_buffer, 3);
126
127 *(pu1_rem_intra4x4_pred_mode) = (UWORD8)u4_temp;
128 }
129
130 pu1_prev_intra4x4_pred_mode_flag++;
131 pu1_rem_intra4x4_pred_mode++;
132 }
133 }
134 else
135 {
136 /**********************************************************************/
137 /* prev_intra4x4_pred_modes to be interpreted as */
138 /* prev_intra8x8_pred_modes in case of transform 8x8 */
139 /**********************************************************************/
140 for(i8x8_luma_blk_idx = 0; i8x8_luma_blk_idx < 4; i8x8_luma_blk_idx++)
141 {
142 UWORD32 u4_temp;
143 GETBIT(u4_temp, ps_bitstrm->u4_ofst, ps_bitstrm->pu4_buffer);
144 *pu1_prev_intra4x4_pred_mode_flag = (UWORD8)u4_temp;
145 if(!(*pu1_prev_intra4x4_pred_mode_flag))
146 {
147 GETBITS(u4_temp, ps_bitstrm->u4_ofst, ps_bitstrm->pu4_buffer, 3);
148
149 (*pu1_rem_intra4x4_pred_mode) = (UWORD8)u4_temp;
150 }
151 pu1_prev_intra4x4_pred_mode_flag++;
152 pu1_rem_intra4x4_pred_mode++;
153 }
154 }
155 return (0);
156}
157WORD32 ih264d_unpack_coeff4x4_4x4blk(dec_struct_t * ps_dec,
158 WORD16 *pi2_out_coeff_data,
159 UWORD8 *pu1_inv_scan)
160{
161 tu_sblk4x4_coeff_data_t *ps_tu_4x4 = (tu_sblk4x4_coeff_data_t *)ps_dec->pv_proc_tu_coeff_data;
162 UWORD16 u2_sig_coeff_map = ps_tu_4x4->u2_sig_coeff_map;
163 WORD32 idx = 0;
164 WORD16 *pi2_coeff_data = &ps_tu_4x4->ai2_level[0];
165 WORD32 dc_only_flag = 0;
166 WORD32 num_coeff = 0;
167
168 PROFILE_DISABLE_UNPACK_LUMA()
169 while(u2_sig_coeff_map)
170 {
171 idx = CLZ(u2_sig_coeff_map);
172
173 idx = 31 - idx;
174 RESET_BIT(u2_sig_coeff_map,idx);
175
176 idx = pu1_inv_scan[idx];
177 pi2_out_coeff_data[idx] = *pi2_coeff_data++;
178 num_coeff++;
179 }
180
181 if((num_coeff == 1) && (idx == 0))
182 {
183 dc_only_flag = 1;
184 }
185
186 {
187 WORD32 offset;
188 offset = (UWORD8 *)pi2_coeff_data - (UWORD8 *)ps_tu_4x4;
189 offset = ALIGN4(offset);
190 ps_dec->pv_proc_tu_coeff_data = (void *)((UWORD8 *)ps_dec->pv_proc_tu_coeff_data + offset);
191 }
192
193 return dc_only_flag;
194}
195
196UWORD32 ih264d_unpack_coeff4x4_8x8blk(dec_struct_t * ps_dec,
197 dec_mb_info_t * ps_cur_mb_info,
198 UWORD16 ui2_luma_csbp,
199 WORD16 *pi2_out_coeff_data)
200{
201 UWORD8 *pu1_inv_scan;
202 UWORD8 u1_mb_field_decoding_flag = ps_cur_mb_info->u1_mb_field_decodingflag;
203 UWORD8 u1_field_coding_flag = ps_cur_mb_info->ps_curmb->u1_mb_fld;
204 UWORD32 u4_luma_dc_only_csbp = 0;
205 WORD32 dc_only_flag = 0;
206
207 PROFILE_DISABLE_UNPACK_LUMA()
208 if(u1_field_coding_flag || u1_mb_field_decoding_flag)
209 {
210 pu1_inv_scan = (UWORD8 *)gau1_ih264d_inv_scan_fld;
211 }
212 else
213 {
214 pu1_inv_scan = (UWORD8 *)gau1_ih264d_inv_scan;
215 }
216
217 // sub 0
218 if(ui2_luma_csbp & 0x1)
219 {
220 memset(pi2_out_coeff_data,0,16*sizeof(WORD16));
221 dc_only_flag = ih264d_unpack_coeff4x4_4x4blk(ps_dec,
222 pi2_out_coeff_data,
223 pu1_inv_scan);
224
225 INSERT_BIT(u4_luma_dc_only_csbp, 0, dc_only_flag);
226 }
227
228 pi2_out_coeff_data += 16;
229 // sub 1
230 if(ui2_luma_csbp & 0x2)
231 {
232 memset(pi2_out_coeff_data,0,16*sizeof(WORD16));
233 dc_only_flag = ih264d_unpack_coeff4x4_4x4blk(ps_dec,
234 pi2_out_coeff_data,
235 pu1_inv_scan);
236 INSERT_BIT(u4_luma_dc_only_csbp, 1, dc_only_flag);
237 }
238
239 pi2_out_coeff_data += 16 + 32;
240 // sub 2
241 if(ui2_luma_csbp & 0x10)
242 {
243 memset(pi2_out_coeff_data,0,16*sizeof(WORD16));
244 dc_only_flag = ih264d_unpack_coeff4x4_4x4blk(ps_dec,
245 pi2_out_coeff_data,
246 pu1_inv_scan);
247 INSERT_BIT(u4_luma_dc_only_csbp, 4, dc_only_flag);
248 }
249
250 pi2_out_coeff_data += 16;
251 // sub 3
252 if(ui2_luma_csbp & 0x20)
253 {
254 memset(pi2_out_coeff_data,0,16*sizeof(WORD16));
255 dc_only_flag = ih264d_unpack_coeff4x4_4x4blk(ps_dec,
256 pi2_out_coeff_data,
257 pu1_inv_scan);
258 INSERT_BIT(u4_luma_dc_only_csbp, 5, dc_only_flag);
259 }
260 return u4_luma_dc_only_csbp;
261}
262WORD32 ih264d_unpack_coeff8x8_8x8blk_cavlc(dec_struct_t * ps_dec,
263 dec_mb_info_t * ps_cur_mb_info,
264 UWORD16 ui2_luma_csbp,
265 WORD16 *pi2_out_coeff_data)
266{
267 UWORD8 *pu1_inv_scan;
268 UWORD8 u1_mb_field_decoding_flag = ps_cur_mb_info->u1_mb_field_decodingflag;
269 UWORD8 u1_field_coding_flag = ps_cur_mb_info->ps_curmb->u1_mb_fld;
270 WORD32 dc_only_flag = 0;
271
272 PROFILE_DISABLE_UNPACK_LUMA()
273 if(ui2_luma_csbp & 0x33)
274 {
275 memset(pi2_out_coeff_data,0,64*sizeof(WORD16));
276 }
277
278 if(!u1_mb_field_decoding_flag)
279 {
280 pu1_inv_scan =
281 (UWORD8*)gau1_ih264d_inv_scan_prog8x8_cavlc[0];
282 }
283 else
284 {
285 pu1_inv_scan =
286 (UWORD8*)gau1_ih264d_inv_scan_int8x8_cavlc[0];
287 }
288 // sub 0
289 if(ui2_luma_csbp & 0x1)
290 {
291 dc_only_flag = ih264d_unpack_coeff4x4_4x4blk(ps_dec,
292 pi2_out_coeff_data,
293 pu1_inv_scan);
294 }
295
296 if(!u1_mb_field_decoding_flag)
297 {
298 pu1_inv_scan =
299 (UWORD8*)gau1_ih264d_inv_scan_prog8x8_cavlc[1];
300 }
301 else
302 {
303 pu1_inv_scan =
304 (UWORD8*)gau1_ih264d_inv_scan_int8x8_cavlc[1];
305 }
306 // sub 1
307 if(ui2_luma_csbp & 0x2)
308 {
309 dc_only_flag = 0;
310 ih264d_unpack_coeff4x4_4x4blk(ps_dec,
311 pi2_out_coeff_data,
312 pu1_inv_scan);
313 }
314
315 if(!u1_mb_field_decoding_flag)
316 {
317 pu1_inv_scan =
318 (UWORD8*)gau1_ih264d_inv_scan_prog8x8_cavlc[2];
319 }
320 else
321 {
322 pu1_inv_scan =
323 (UWORD8*)gau1_ih264d_inv_scan_int8x8_cavlc[2];
324 }
325 // sub 2
326 if(ui2_luma_csbp & 0x10)
327 {
328 dc_only_flag = 0;
329 ih264d_unpack_coeff4x4_4x4blk(ps_dec,
330 pi2_out_coeff_data,
331 pu1_inv_scan);
332 }
333
334 if(!u1_mb_field_decoding_flag)
335 {
336 pu1_inv_scan =
337 (UWORD8*)gau1_ih264d_inv_scan_prog8x8_cavlc[3];
338 }
339 else
340 {
341 pu1_inv_scan =
342 (UWORD8*)gau1_ih264d_inv_scan_int8x8_cavlc[3];
343 }
344 // sub 3
345 if(ui2_luma_csbp & 0x20)
346 {
347 dc_only_flag = 0;
348 ih264d_unpack_coeff4x4_4x4blk(ps_dec,
349 pi2_out_coeff_data,
350 pu1_inv_scan);
351 }
352 return dc_only_flag;
353}
354void ih264d_unpack_coeff4x4_8x8blk_chroma(dec_struct_t * ps_dec,
355 dec_mb_info_t * ps_cur_mb_info,
356 UWORD16 ui2_chroma_csbp,
357 WORD16 *pi2_out_coeff_data)
358{
359 UWORD8 *pu1_inv_scan;
360 UWORD8 u1_mb_field_decoding_flag = ps_cur_mb_info->u1_mb_field_decodingflag;
361 UWORD8 u1_field_coding_flag = ps_cur_mb_info->ps_curmb->u1_mb_fld;
362
363 PROFILE_DISABLE_UNPACK_CHROMA()
364 if(u1_field_coding_flag || u1_mb_field_decoding_flag)
365 {
366 pu1_inv_scan = (UWORD8 *)gau1_ih264d_inv_scan_fld;
367 }
368 else
369 {
370 pu1_inv_scan = (UWORD8 *)gau1_ih264d_inv_scan;
371 }
372
373 if(ui2_chroma_csbp & 0x1)
374 {
375 memset(pi2_out_coeff_data,0,16*sizeof(WORD16));
376 ih264d_unpack_coeff4x4_4x4blk(ps_dec,
377 pi2_out_coeff_data,
378 pu1_inv_scan);
379 }
380 pi2_out_coeff_data += 16;
381 if(ui2_chroma_csbp & 0x2)
382 {
383 memset(pi2_out_coeff_data,0,16*sizeof(WORD16));
384 ih264d_unpack_coeff4x4_4x4blk(ps_dec,
385 pi2_out_coeff_data,
386 pu1_inv_scan);
387 }
388
389 pi2_out_coeff_data += 16;
390 if(ui2_chroma_csbp & 0x4)
391 {
392 memset(pi2_out_coeff_data,0,16*sizeof(WORD16));
393 ih264d_unpack_coeff4x4_4x4blk(ps_dec,
394 pi2_out_coeff_data,
395 pu1_inv_scan);
396 }
397
398 pi2_out_coeff_data += 16;
399 if(ui2_chroma_csbp & 0x8)
400 {
401 memset(pi2_out_coeff_data,0,16*sizeof(WORD16));
402 ih264d_unpack_coeff4x4_4x4blk(ps_dec,
403 pi2_out_coeff_data,
404 pu1_inv_scan);
405 }
406}
407UWORD32 ih264d_unpack_luma_coeff4x4_mb(dec_struct_t * ps_dec,
408 dec_mb_info_t * ps_cur_mb_info,
409 UWORD8 intra_flag)
410{
411 UWORD8 u1_mb_type = ps_cur_mb_info->u1_mb_type;
412 UWORD16 ui2_luma_csbp = ps_cur_mb_info->u2_luma_csbp;
413 UWORD8 *pu1_inv_scan = ps_dec->pu1_inv_scan;
414 WORD16 *pi2_coeff_data = ps_dec->pi2_coeff_data;
415
416 PROFILE_DISABLE_UNPACK_LUMA()
417 if(!ps_cur_mb_info->u1_tran_form8x8)
418 {
419 UWORD32 u4_luma_dc_only_csbp = 0;
420 UWORD32 u4_temp = 0;
421 WORD16* pi2_dc_val = NULL;
422 /*
423 * Reserve the pointer to dc vals. The dc vals will be copied
424 * after unpacking of ac vals since memset to 0 inside.
425 */
426 if(intra_flag && (u1_mb_type != I_4x4_MB))
427 {
428 if(CHECKBIT(ps_cur_mb_info->u1_yuv_dc_block_flag,0))
429 {
430 pi2_dc_val = (WORD16 *)ps_dec->pv_proc_tu_coeff_data;
431
432 ps_dec->pv_proc_tu_coeff_data = (void *)(pi2_dc_val + 16);
433 }
434 }
435
436 if(ui2_luma_csbp)
437 {
438 pi2_coeff_data = ps_dec->pi2_coeff_data;
439 u4_temp = ih264d_unpack_coeff4x4_8x8blk(ps_dec,
440 ps_cur_mb_info,
441 ui2_luma_csbp,
442 pi2_coeff_data);
443 u4_luma_dc_only_csbp = u4_temp;
444
445 pi2_coeff_data += 32;
446
447 ui2_luma_csbp = ui2_luma_csbp >> 2;
448 u4_temp = ih264d_unpack_coeff4x4_8x8blk(ps_dec,
449 ps_cur_mb_info,
450 ui2_luma_csbp,
451 pi2_coeff_data);
452
453 u4_luma_dc_only_csbp |= (u4_temp << 2);
454
455 pi2_coeff_data += 32 + 64;
456
457 ui2_luma_csbp = ui2_luma_csbp >> 6;
458 u4_temp = ih264d_unpack_coeff4x4_8x8blk(ps_dec,
459 ps_cur_mb_info,
460 ui2_luma_csbp,
461 pi2_coeff_data);
462
463 u4_luma_dc_only_csbp |= (u4_temp << 8);
464
465 pi2_coeff_data += 32;
466
467 ui2_luma_csbp = ui2_luma_csbp >> 2;
468 u4_temp = ih264d_unpack_coeff4x4_8x8blk(ps_dec,
469 ps_cur_mb_info,
470 ui2_luma_csbp,
471 pi2_coeff_data);
472 u4_luma_dc_only_csbp |= (u4_temp << 10);
473 }
474
475 if(pi2_dc_val != NULL)
476 {
477 WORD32 i;
478 pi2_coeff_data = ps_dec->pi2_coeff_data;
479 for(i = 0; i < 4; i++)
480 {
481 pi2_coeff_data[0] = pi2_dc_val[0];
482 pi2_coeff_data[4 * 16] = pi2_dc_val[4];
483 pi2_coeff_data[8 * 16] = pi2_dc_val[8];
484 pi2_coeff_data[12 * 16] = pi2_dc_val[12];
485
486 pi2_dc_val++; /* Point to next column */
487 pi2_coeff_data += 16;
488 }
489 u4_luma_dc_only_csbp = ps_cur_mb_info->u2_luma_csbp ^ 0xFFFF;
490 }
491 return u4_luma_dc_only_csbp;
492 }
493 else
494 {
495 UWORD32 u4_luma_dc_only_cbp = 0;
496 WORD32 dc_only_flag;
497 if(ui2_luma_csbp)
498 {
499 pi2_coeff_data = ps_dec->pi2_coeff_data;
500 dc_only_flag = ih264d_unpack_coeff8x8_8x8blk_cavlc(ps_dec,
501 ps_cur_mb_info,
502 ui2_luma_csbp,
503 pi2_coeff_data);
504 INSERT_BIT(u4_luma_dc_only_cbp, 0, dc_only_flag);
505
506 pi2_coeff_data += 64;
507
508 ui2_luma_csbp = ui2_luma_csbp >> 2;
509 dc_only_flag = ih264d_unpack_coeff8x8_8x8blk_cavlc(ps_dec,
510 ps_cur_mb_info,
511 ui2_luma_csbp,
512 pi2_coeff_data);
513
514 INSERT_BIT(u4_luma_dc_only_cbp, 1, dc_only_flag);
515
516 pi2_coeff_data += 64;
517
518 ui2_luma_csbp = ui2_luma_csbp >> 6;
519 dc_only_flag = ih264d_unpack_coeff8x8_8x8blk_cavlc(ps_dec,
520 ps_cur_mb_info,
521 ui2_luma_csbp,
522 pi2_coeff_data);
523
524 INSERT_BIT(u4_luma_dc_only_cbp, 2, dc_only_flag);
525
526 pi2_coeff_data += 64;
527 ui2_luma_csbp = ui2_luma_csbp >> 2;
528 dc_only_flag = ih264d_unpack_coeff8x8_8x8blk_cavlc(ps_dec,
529 ps_cur_mb_info,
530 ui2_luma_csbp,
531 pi2_coeff_data);
532 INSERT_BIT(u4_luma_dc_only_cbp, 3, dc_only_flag);
533 }
534 return u4_luma_dc_only_cbp;
535 }
536
537}
538
539void ih264d_unpack_chroma_coeff4x4_mb(dec_struct_t * ps_dec,
540 dec_mb_info_t * ps_cur_mb_info)
541{
542 UWORD8 u1_mb_type = ps_cur_mb_info->u1_mb_type;
543 UWORD16 ui2_chroma_csbp = ps_cur_mb_info->u2_chroma_csbp;
544 UWORD8 *pu1_inv_scan = ps_dec->pu1_inv_scan;
545 WORD16 *pi2_coeff_data = ps_dec->pi2_coeff_data;
546 WORD32 i;
547 WORD16 *pi2_dc_val_u = NULL;
548 WORD16 *pi2_dc_val_v = NULL;
549
550 PROFILE_DISABLE_UNPACK_CHROMA()
551 if((ps_cur_mb_info->u1_cbp >> 4) == CBPC_ALLZERO)
552 return;
553
554 /*
555 * Reserve the pointers to dc vals. The dc vals will be copied
556 * after unpacking of ac vals since memset to 0 inside.
557 */
558 if(CHECKBIT(ps_cur_mb_info->u1_yuv_dc_block_flag,1))
559 {
560 pi2_dc_val_u = (WORD16 *)ps_dec->pv_proc_tu_coeff_data;
561
562 ps_dec->pv_proc_tu_coeff_data = (void *)(pi2_dc_val_u + 4);
563 }
564 if(CHECKBIT(ps_cur_mb_info->u1_yuv_dc_block_flag,2))
565 {
566 pi2_dc_val_v = (WORD16 *)ps_dec->pv_proc_tu_coeff_data;
567
568 ps_dec->pv_proc_tu_coeff_data = (void *)(pi2_dc_val_v + 4);
569 }
570
571 if((ps_cur_mb_info->u1_cbp >> 4) == CBPC_NONZERO)
572 {
573 pi2_coeff_data = ps_dec->pi2_coeff_data;
574 ih264d_unpack_coeff4x4_8x8blk_chroma(ps_dec,
575 ps_cur_mb_info,
576 ui2_chroma_csbp,
577 pi2_coeff_data);
578
579 pi2_coeff_data += 64;
580 ui2_chroma_csbp = ui2_chroma_csbp >> 4;
581 ih264d_unpack_coeff4x4_8x8blk_chroma(ps_dec,
582 ps_cur_mb_info,
583 ui2_chroma_csbp,
584 pi2_coeff_data);
585
586 }
587
588 pi2_coeff_data = ps_dec->pi2_coeff_data;
589 if(pi2_dc_val_u != NULL)
590 {
591 pi2_coeff_data[0] = *pi2_dc_val_u++;
592 pi2_coeff_data[1 * 16] = *pi2_dc_val_u++;
593 pi2_coeff_data[2 * 16] = *pi2_dc_val_u++;
594 pi2_coeff_data[3 * 16] = *pi2_dc_val_u++;
595 }
596 else
597 {
598 pi2_coeff_data[0] = 0;
599 pi2_coeff_data[1 * 16] = 0;
600 pi2_coeff_data[2 * 16] = 0;
601 pi2_coeff_data[3 * 16] = 0;
602 }
603 pi2_coeff_data += 64;
604 if(pi2_dc_val_v != NULL)
605 {
606 pi2_coeff_data[0] = *pi2_dc_val_v++;
607 pi2_coeff_data[1 * 16] = *pi2_dc_val_v++;
608 pi2_coeff_data[2 * 16] = *pi2_dc_val_v++;
609 pi2_coeff_data[3 * 16] = *pi2_dc_val_v++;
610 }
611 else
612 {
613 pi2_coeff_data[0] = 0;
614 pi2_coeff_data[1 * 16] = 0;
615 pi2_coeff_data[2 * 16] = 0;
616 pi2_coeff_data[3 * 16] = 0;
617 }
618}
619UWORD32 ih264d_unpack_luma_coeff8x8_mb(dec_struct_t * ps_dec,
620 dec_mb_info_t * ps_cur_mb_info)
621{
622 WORD32 blk_8x8_cnt;
623 WORD16 *pi2_out_coeff_data = ps_dec->pi2_coeff_data;
624 UWORD8 u1_field_coding_flag = ps_cur_mb_info->ps_curmb->u1_mb_fld;
625 UWORD8 *pu1_inv_scan;
626 UWORD32 u4_luma_dc_only_cbp = 0;
627
628 PROFILE_DISABLE_UNPACK_LUMA()
629 if(!u1_field_coding_flag)
630 {
631 /*******************************************************************/
632 /* initializing inverse scan matrices */
633 /*******************************************************************/
634 pu1_inv_scan = (UWORD8 *)gau1_ih264d_inv_scan_prog8x8_cabac;
635 }
636 else
637 {
638 /*******************************************************************/
639 /* initializing inverse scan matrices */
640 /*******************************************************************/
641 pu1_inv_scan = (UWORD8 *)gau1_ih264d_inv_scan_int8x8_cabac;
642 }
643
644 for(blk_8x8_cnt = 0; blk_8x8_cnt < 4; blk_8x8_cnt++)
645 {
646 if(CHECKBIT(ps_cur_mb_info->u1_cbp, blk_8x8_cnt))
647 {
648 tu_blk8x8_coeff_data_t *ps_tu_8x8 = (tu_blk8x8_coeff_data_t *)ps_dec->pv_proc_tu_coeff_data;
649 UWORD32 u4_sig_coeff_map;
650 WORD32 idx = 0;
651 WORD16 *pi2_coeff_data = &ps_tu_8x8->ai2_level[0];
652 WORD32 num_coeff = 0;
653
654 /* memset 64 coefficient to zero */
655 memset(pi2_out_coeff_data,0,64*sizeof(WORD16));
656
657 u4_sig_coeff_map = ps_tu_8x8->au4_sig_coeff_map[1];
658
659 while(u4_sig_coeff_map)
660 {
661 idx = CLZ(u4_sig_coeff_map);
662
663 idx = 31 - idx;
664 RESET_BIT(u4_sig_coeff_map,idx);
665
666 idx = pu1_inv_scan[idx + 32];
667 pi2_out_coeff_data[idx] = *pi2_coeff_data++;
668 num_coeff++;
669 }
670
671 u4_sig_coeff_map = ps_tu_8x8->au4_sig_coeff_map[0];
672 while(u4_sig_coeff_map)
673 {
674 idx = CLZ(u4_sig_coeff_map);
675
676 idx = 31 - idx;
677 RESET_BIT(u4_sig_coeff_map,idx);
678
679 idx = pu1_inv_scan[idx];
680 pi2_out_coeff_data[idx] = *pi2_coeff_data++;
681 num_coeff++;
682 }
683
684 if((num_coeff == 1) && (idx == 0))
685 {
686 SET_BIT(u4_luma_dc_only_cbp,blk_8x8_cnt);
687 }
688
689
690 {
691 WORD32 offset;
692 offset = (UWORD8 *)pi2_coeff_data - (UWORD8 *)ps_tu_8x8;
693 offset = ALIGN4(offset);
694 ps_dec->pv_proc_tu_coeff_data = (void *)((UWORD8 *)ps_dec->pv_proc_tu_coeff_data + offset);
695 }
696 }
697 pi2_out_coeff_data += 64;
698 }
699
700 return u4_luma_dc_only_cbp;
701}
702/*!
703 **************************************************************************
704 * \if Function name : ih264d_process_intra_mb \endif
705 *
706 * \brief
707 * This function decodes an I MB. Intraprediction is carried out followed
708 * by InvTramsform. Both IntraPrediction and Reconstrucion are carried out
709 * row buffer itself.
710 *
711 *
712 * \return
713 * 0 on Success and Error code otherwise
714 **************************************************************************
715 */
716WORD32 ih264d_process_intra_mb(dec_struct_t * ps_dec,
717 dec_mb_info_t * ps_cur_mb_info,
718 UWORD8 u1_mb_num)
719{
720 UWORD8 u1_mb_type = ps_cur_mb_info->u1_mb_type;
721 UWORD8 uc_temp = ps_cur_mb_info->u1_mb_ngbr_availablity;
722 UWORD8 u1_top_available = BOOLEAN(uc_temp & TOP_MB_AVAILABLE_MASK);
723 UWORD8 u1_left_available = BOOLEAN(uc_temp & LEFT_MB_AVAILABLE_MASK);
724 UWORD8 u1_use_top_right_mb = BOOLEAN(uc_temp & TOP_RIGHT_MB_AVAILABLE_MASK);
725 UWORD8 u1_use_top_left_mb = BOOLEAN(uc_temp & TOP_LEFT_MB_AVAILABLE_MASK);
726 UWORD8 uc_useTopMB = u1_top_available;
727 UWORD16 u2_use_left_mb = u1_left_available;
728 UWORD16 u2_use_left_mb_pack;
729 UWORD8 *pu1_luma_pred_buffer;
730 /* CHANGED CODE */
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -0700731 UWORD8 *pu1_luma_rec_buffer;
Hamsalekha S8d3d3032015-03-13 21:24:58 +0530732 UWORD8 *puc_top;
733
734 mb_neigbour_params_t *ps_left_mb;
735 mb_neigbour_params_t *ps_top_mb;
736 mb_neigbour_params_t *ps_top_right_mb;
737 mb_neigbour_params_t *ps_curmb;
738
739 UWORD16 u2_mbx = ps_cur_mb_info->u2_mbx;
740 UWORD32 ui_pred_width, ui_rec_width;
741 WORD16 *pi2_y_coeff;
742 UWORD8 u1_mbaff, u1_topmb, u1_mb_field_decoding_flag;
743 UWORD32 u4_num_pmbair;
744 UWORD16 ui2_luma_csbp = ps_cur_mb_info->u2_luma_csbp;
745 UWORD8 *pu1_yleft, *pu1_ytop_left;
746 /* Chroma variables*/
747 UWORD8 *pu1_top_u;
748 UWORD8 *pu1_uleft;
749 UWORD8 *pu1_u_top_left;
750 /* CHANGED CODE */
751 UWORD8 *pu1_mb_cb_rei1_buffer, *pu1_mb_cr_rei1_buffer;
752 UWORD32 u4_recwidth_cr;
753 /* CHANGED CODE */
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -0700754 tfr_ctxt_t *ps_frame_buf = ps_dec->ps_frame_buf_ip_recon;
Hamsalekha S8d3d3032015-03-13 21:24:58 +0530755 UWORD32 u4_luma_dc_only_csbp = 0;
756 UWORD32 u4_luma_dc_only_cbp = 0;
757
758 UWORD8 *pu1_prev_intra4x4_pred_mode_data = (UWORD8 *)ps_dec->pv_proc_tu_coeff_data; //Pointer to keep track of intra4x4_pred_mode data in pv_proc_tu_coeff_data buffer
759 u1_mbaff = ps_dec->ps_cur_slice->u1_mbaff_frame_flag;
760 u1_topmb = ps_cur_mb_info->u1_topmb;
761 u4_num_pmbair = (u1_mb_num >> u1_mbaff);
762
763
764 /*--------------------------------------------------------------------*/
765 /* Find the current MB's mb params */
766 /*--------------------------------------------------------------------*/
767 u1_mb_field_decoding_flag = ps_cur_mb_info->u1_mb_field_decodingflag;
768
769 ps_curmb = ps_cur_mb_info->ps_curmb;
770 ps_top_mb = ps_cur_mb_info->ps_top_mb;
771 ps_left_mb = ps_cur_mb_info->ps_left_mb;
772 ps_top_right_mb = ps_cur_mb_info->ps_top_right_mb;
773
774 /*--------------------------------------------------------------------*/
775 /* Check whether neighbouring MB is Inter MB and */
776 /* constrained intra pred is 1. */
777 /*--------------------------------------------------------------------*/
778 u2_use_left_mb_pack = (u2_use_left_mb << 8) + u2_use_left_mb;
779
780 if(ps_dec->ps_cur_pps->u1_constrained_intra_pred_flag)
781 {
782 UWORD8 u1_left = (UWORD8)u2_use_left_mb;
783
784 uc_useTopMB = uc_useTopMB
785 && ((ps_top_mb->u1_mb_type != P_MB)
786 && (ps_top_mb->u1_mb_type != B_MB));
787 u2_use_left_mb = u2_use_left_mb
788 && ((ps_left_mb->u1_mb_type != P_MB)
789 && (ps_left_mb->u1_mb_type != B_MB));
790
791 u2_use_left_mb_pack = (u2_use_left_mb << 8) + u2_use_left_mb;
792 if(u1_mbaff)
793 {
794 if(u1_mb_field_decoding_flag ^ ps_left_mb->u1_mb_fld)
795 {
796 u1_left = u1_left
797 && (((ps_left_mb + 1)->u1_mb_type != P_MB)
798 && ((ps_left_mb + 1)->u1_mb_type
799 != B_MB));
800 u2_use_left_mb = u2_use_left_mb && u1_left;
801 if(u1_mb_field_decoding_flag)
802 u2_use_left_mb_pack = (u1_left << 8)
803 + (u2_use_left_mb_pack & 0xff);
804 else
805 u2_use_left_mb_pack = (u2_use_left_mb << 8)
806 + (u2_use_left_mb);
807 }
808 }
809 u1_use_top_right_mb =
810 u1_use_top_right_mb
811 && ((ps_top_right_mb->u1_mb_type != P_MB)
812 && (ps_top_right_mb->u1_mb_type
813 != B_MB));
814
815 u1_use_top_left_mb =
816 u1_use_top_left_mb
817 && ((ps_cur_mb_info->u1_topleft_mbtype != P_MB)
818 && (ps_cur_mb_info->u1_topleft_mbtype
819 != B_MB));
820 }
821
822 /*********************Common pointer calculations *************************/
823 /* CHANGED CODE */
824 pu1_luma_pred_buffer = ps_dec->pu1_y;
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -0700825 pu1_luma_rec_buffer = ps_frame_buf->pu1_dest_y + (u4_num_pmbair << 4);
Hamsalekha S8d3d3032015-03-13 21:24:58 +0530826 pu1_mb_cb_rei1_buffer = ps_frame_buf->pu1_dest_u
827 + (u4_num_pmbair << 3) * YUV420SP_FACTOR;
828 pu1_mb_cr_rei1_buffer = ps_frame_buf->pu1_dest_v + (u4_num_pmbair << 3);
829 ui_pred_width = MB_SIZE;
830 ui_rec_width = ps_dec->u2_frm_wd_y << u1_mb_field_decoding_flag;
831 u4_recwidth_cr = ps_dec->u2_frm_wd_uv << u1_mb_field_decoding_flag;
832 /************* Current and top luma pointer *****************/
833
834 if(u1_mbaff)
835 {
836 if(u1_topmb == 0)
837 {
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -0700838 pu1_luma_rec_buffer += (
Hamsalekha S8d3d3032015-03-13 21:24:58 +0530839 u1_mb_field_decoding_flag ?
840 (ui_rec_width >> 1) :
841 (ui_rec_width << 4));
842 pu1_mb_cb_rei1_buffer += (
843 u1_mb_field_decoding_flag ?
844 (u4_recwidth_cr >> 1) :
845 (u4_recwidth_cr << 3));
846 pu1_mb_cr_rei1_buffer += (
847 u1_mb_field_decoding_flag ?
848 (u4_recwidth_cr >> 1) :
849 (u4_recwidth_cr << 3));
850 }
851 }
852
853 /* CHANGED CODE */
854 if(ps_dec->u4_use_intrapred_line_copy == 1)
855 {
856 puc_top = ps_dec->pu1_prev_y_intra_pred_line + (ps_cur_mb_info->u2_mbx << 4);
857 pu1_top_u = ps_dec->pu1_prev_u_intra_pred_line
858 + (ps_cur_mb_info->u2_mbx << 3) * YUV420SP_FACTOR;
859 }
860 else
861 {
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -0700862 puc_top = pu1_luma_rec_buffer - ui_rec_width;
Hamsalekha S8d3d3032015-03-13 21:24:58 +0530863 pu1_top_u = pu1_mb_cb_rei1_buffer - u4_recwidth_cr;
864 }
865 /* CHANGED CODE */
866
867 /************* Left pointer *****************/
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -0700868 pu1_yleft = pu1_luma_rec_buffer - 1;
Hamsalekha S8d3d3032015-03-13 21:24:58 +0530869 pu1_uleft = pu1_mb_cb_rei1_buffer - 1 * YUV420SP_FACTOR;
870
871 /**************Top Left pointer calculation**********/
872 pu1_ytop_left = puc_top - 1;
873 pu1_u_top_left = pu1_top_u - 1 * YUV420SP_FACTOR;
874
875 /* CHANGED CODE */
876 PROFILE_DISABLE_INTRA_PRED()
877 {
878 pu1_prev_intra4x4_pred_mode_data = (UWORD8 *)ps_dec->pv_proc_tu_coeff_data;
879 if(u1_mb_type == I_4x4_MB && ps_cur_mb_info->u1_tran_form8x8 == 0)
880 {
881 ps_dec->pv_proc_tu_coeff_data = (void *)((UWORD8 *)ps_dec->pv_proc_tu_coeff_data + 32);
882
883 }
884 else if (u1_mb_type == I_4x4_MB && ps_cur_mb_info->u1_tran_form8x8 == 1)
885 {
886 ps_dec->pv_proc_tu_coeff_data = (void *)((UWORD8 *)ps_dec->pv_proc_tu_coeff_data + 8);
887 }
888 }
889 if(!ps_cur_mb_info->u1_tran_form8x8)
890 {
891 u4_luma_dc_only_csbp = ih264d_unpack_luma_coeff4x4_mb(ps_dec,
892 ps_cur_mb_info,
893 1);
894 }
895 else
896 {
897 if(!ps_dec->ps_cur_pps->u1_entropy_coding_mode)
898 {
899 u4_luma_dc_only_cbp = ih264d_unpack_luma_coeff4x4_mb(ps_dec,
900 ps_cur_mb_info,
901 1);
902 }
903 else
904 {
905 u4_luma_dc_only_cbp = ih264d_unpack_luma_coeff8x8_mb(ps_dec,
906 ps_cur_mb_info);
907 }
908 }
909
910 pi2_y_coeff = ps_dec->pi2_coeff_data;
911
912 if(u1_mb_type != I_4x4_MB)
913 {
914 UWORD8 u1_intrapred_mode = MB_TYPE_TO_INTRA_16x16_MODE(u1_mb_type);
915 /*--------------------------------------------------------------------*/
916 /* 16x16 IntraPrediction */
917 /*--------------------------------------------------------------------*/
918 {
919 UWORD8 u1_packed_modes = (u1_top_available << 1)
920 + u1_left_available;
921 UWORD8 u1_err_code =
922 (u1_intrapred_mode & 1) ?
923 u1_intrapred_mode :
924 (u1_intrapred_mode ^ 2);
925
926 if((u1_err_code & u1_packed_modes) ^ u1_err_code)
Hamsalekha S9f0dcba2015-06-09 15:54:31 +0530927 {
928 u1_intrapred_mode = 0;
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -0700929 ps_dec->i4_error_code = ERROR_INTRAPRED;
Hamsalekha S9f0dcba2015-06-09 15:54:31 +0530930 }
Hamsalekha S8d3d3032015-03-13 21:24:58 +0530931 }
932 {
Harish Mahendrakard32eaf42016-04-18 16:38:54 +0530933 /* Align the size to multiple of 8, so that SIMD functions
934 can read 64 bits at a time. Only 33 bytes are actaully used */
935 UWORD8 au1_ngbr_pels[40];
Hamsalekha S8d3d3032015-03-13 21:24:58 +0530936 /* Get neighbour pixels */
937 /* left pels */
938 if(u2_use_left_mb)
939 {
940 WORD32 i;
941 for(i = 0; i < 16; i++)
942 au1_ngbr_pels[16 - 1 - i] = pu1_yleft[i * ui_rec_width];
943 }
944 else
945 {
946 memset(au1_ngbr_pels, 0, 16);
947 }
948
949 /* top left pels */
950 au1_ngbr_pels[16] = *pu1_ytop_left;
951
952 /* top pels */
953 if(uc_useTopMB)
954 {
955 memcpy(au1_ngbr_pels + 16 + 1, puc_top, 16);
956 }
957 else
958 {
959 memset(au1_ngbr_pels + 16 + 1, 0, 16);
960 }
961 PROFILE_DISABLE_INTRA_PRED()
962 ps_dec->apf_intra_pred_luma_16x16[u1_intrapred_mode](
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -0700963 au1_ngbr_pels, pu1_luma_rec_buffer, 1, ui_rec_width,
Hamsalekha S8d3d3032015-03-13 21:24:58 +0530964 ((uc_useTopMB << 2) | u2_use_left_mb));
965 }
966 {
967 UWORD32 i;
968 WORD16 ai2_tmp[16];
969 for(i = 0; i < 16; i++)
970 {
971 WORD16 *pi2_level = pi2_y_coeff + (i << 4);
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -0700972 UWORD8 *pu1_pred_sblk = pu1_luma_rec_buffer
Hamsalekha S8d3d3032015-03-13 21:24:58 +0530973 + ((i & 0x3) * BLK_SIZE)
974 + (i >> 2) * (ui_rec_width << 2);
975 PROFILE_DISABLE_IQ_IT_RECON()
976 {
977 if(CHECKBIT(ps_cur_mb_info->u2_luma_csbp, i))
978 {
979 ps_dec->pf_iquant_itrans_recon_luma_4x4(
980 pi2_level,
981 pu1_pred_sblk,
982 pu1_pred_sblk,
983 ui_rec_width,
984 ui_rec_width,
985 gau2_ih264_iquant_scale_4x4[ps_cur_mb_info->u1_qp_rem6],
986 (UWORD16 *)ps_dec->s_high_profile.i2_scalinglist4x4[0],
987 ps_cur_mb_info->u1_qp_div6, ai2_tmp, 1,
988 pi2_level);
989 }
990 else if((CHECKBIT(u4_luma_dc_only_csbp, i)) && pi2_level[0] != 0)
991 {
992 ps_dec->pf_iquant_itrans_recon_luma_4x4_dc(
993 pi2_level,
994 pu1_pred_sblk,
995 pu1_pred_sblk,
996 ui_rec_width,
997 ui_rec_width,
998 gau2_ih264_iquant_scale_4x4[ps_cur_mb_info->u1_qp_rem6],
999 (UWORD16 *)ps_dec->s_high_profile.i2_scalinglist4x4[0],
1000 ps_cur_mb_info->u1_qp_div6, ai2_tmp, 1,
1001 pi2_level);
1002 }
1003 }
1004 }
1005 }
1006 }
1007 else if(!ps_cur_mb_info->u1_tran_form8x8)
1008 {
1009 UWORD8 u1_is_left_sub_block, u1_is_top_sub_block = uc_useTopMB;
1010 UWORD8 u1_sub_blk_x, u1_sub_blk_y, u1_sub_mb_num;
1011 WORD8 i1_top_pred_mode;
1012 WORD8 i1_left_pred_mode;
1013 UWORD8 *pu1_top, *pu1_left, *pu1_top_left, *pu1_top_right;
1014 WORD8 *pi1_cur_pred_mode, *pi1_left_pred_mode, *pc_topPredMode;
1015 UWORD16 ui2_left_pred_buf_width = 0xffff;
1016 WORD8 i1_intra_pred;
1017 UWORD8 *pu1_prev_intra4x4_pred_mode_flag = pu1_prev_intra4x4_pred_mode_data;
1018 UWORD8 *pu1_rem_intra4x4_pred_mode = pu1_prev_intra4x4_pred_mode_data + 16;
1019 WORD16 *pi2_y_coeff1;
1020 UWORD8 u1_cur_sub_block;
1021 UWORD16 ui2_top_rt_mask;
1022
1023 /*--------------------------------------------------------------------*/
1024 /* 4x4 IntraPrediction */
1025 /*--------------------------------------------------------------------*/
1026 /* Calculation of Top Right subblock mask */
1027 /* */
1028 /* (a) Set it to default mask */
1029 /* [It has 0 for sublocks which will never have top-right sub block] */
1030 /* */
1031 /* (b) If top MB is not available */
1032 /* Clear the bits of the first row sub blocks */
1033 /* */
1034 /* (c) Set/Clear bit for top-right sublock of MB */
1035 /* [5 sub-block in decoding order] based on TOP RIGHT MB availablity */
1036 /*--------------------------------------------------------------------*/
1037
1038 pu1_top = puc_top;
1039
1040 ui2_top_rt_mask = (u1_use_top_right_mb << 3) | (0x5750);
1041 if(uc_useTopMB)
1042 ui2_top_rt_mask |= 0x7;
1043
1044 /*Top Related initialisations*/
1045
1046
1047 pi1_cur_pred_mode = ps_cur_mb_info->ps_curmb->pi1_intrapredmodes;
1048 pc_topPredMode = ps_cur_mb_info->ps_top_mb->pi1_intrapredmodes;
1049 /*--------------------------------------
1050 if(u1_mbaff)
1051 {
1052
1053 pi1_cur_pred_mode += (u2_mbx << 2);
1054 pc_topPredMode = pi1_cur_pred_mode + ps_cur_mb_info->i1_offset;
1055 pi1_cur_pred_mode += (u1_topmb) ? 0: 4;
1056 }*/
1057
1058 if(u1_top_available)
1059 {
1060 if(ps_top_mb->u1_mb_type == I_4x4_MB)
1061 *(WORD32*)pi1_cur_pred_mode = *(WORD32*)pc_topPredMode;
1062 else
1063 *(WORD32*)pi1_cur_pred_mode =
1064 (uc_useTopMB) ? DC_DC_DC_DC : NOT_VALID;
1065 }
1066 else
1067 *(WORD32*)pi1_cur_pred_mode = NOT_VALID;
1068 /* CHANGED CODE */
1069
1070 /* CHANGED CODE */
1071
1072 /*Left Related initialisations*/
1073 pi1_left_pred_mode = ps_dec->pi1_left_pred_mode;
1074 if(!u1_mbaff)
1075 {
1076
1077 if(u1_left_available)
1078 {
1079
1080 if(ps_left_mb->u1_mb_type != I_4x4_MB)
1081 *(WORD32*)pi1_left_pred_mode =
1082 (u2_use_left_mb_pack) ?
1083 DC_DC_DC_DC :
1084 NOT_VALID;
1085
1086 }
1087 else
1088 {
1089
1090 *(WORD32*)pi1_left_pred_mode = NOT_VALID;
1091 }
1092
1093 }
1094 else
1095 {
1096 UWORD8 u1_curMbfld = ps_cur_mb_info->u1_mb_field_decodingflag;
1097 UWORD8 u1_leftMbfld = ps_left_mb->u1_mb_fld;
1098
1099 if(u1_curMbfld ^ u1_leftMbfld)
1100 {
1101
1102 if(u1_topmb
1103 | ((u1_topmb == 0)
1104 && ((ps_curmb - 1)->u1_mb_type
1105 != I_4x4_MB)))
1106 {
1107 if(u1_left_available)
1108 {
1109 if(ps_left_mb->u1_mb_type != I_4x4_MB)
1110 {
1111 if(CHECKBIT(u2_use_left_mb_pack,0) == 0)
1112 *(WORD32*)pi1_left_pred_mode = NOT_VALID;
1113 else
1114 *(WORD32*)pi1_left_pred_mode = DC_DC_DC_DC;
1115 }
1116 }
1117 else
1118 *(WORD32*)pi1_left_pred_mode = NOT_VALID;
1119
1120 if(u1_curMbfld)
1121 {
1122 if(u1_left_available)
1123 {
1124 if((ps_left_mb + 1)->u1_mb_type != I_4x4_MB)
1125 {
1126 if(u2_use_left_mb_pack >> 8)
1127 *(WORD32*)(pi1_left_pred_mode + 4) =
1128 DC_DC_DC_DC;
1129 else
1130 *(WORD32*)(pi1_left_pred_mode + 4) =
1131 NOT_VALID;
1132 }
1133 }
1134 else
1135 *(WORD32*)(pi1_left_pred_mode + 4) = NOT_VALID;
1136 pi1_left_pred_mode[1] = pi1_left_pred_mode[2];
1137 pi1_left_pred_mode[2] = pi1_left_pred_mode[4];
1138 pi1_left_pred_mode[3] = pi1_left_pred_mode[6];
1139 *(WORD32*)(pi1_left_pred_mode + 4) =
1140 *(WORD32*)pi1_left_pred_mode;
1141 }
1142 else
1143 {
1144
1145 pi1_left_pred_mode[7] = pi1_left_pred_mode[3];
1146 pi1_left_pred_mode[6] = pi1_left_pred_mode[3];
1147 pi1_left_pred_mode[5] = pi1_left_pred_mode[2];
1148 pi1_left_pred_mode[4] = pi1_left_pred_mode[2];
1149 pi1_left_pred_mode[3] = pi1_left_pred_mode[1];
1150 pi1_left_pred_mode[2] = pi1_left_pred_mode[1];
1151 pi1_left_pred_mode[1] = pi1_left_pred_mode[0];
1152 }
1153 }
1154 pi1_left_pred_mode += (u1_topmb) ? 0 : 4;
1155 }
1156 else
1157 {
1158
1159 pi1_left_pred_mode += (u1_topmb) ? 0 : 4;
1160 if(u1_left_available)
1161 {
1162
1163 if(ps_left_mb->u1_mb_type != I_4x4_MB)
1164 *(WORD32*)pi1_left_pred_mode =
1165 (u2_use_left_mb_pack) ?
1166 DC_DC_DC_DC :
1167 NOT_VALID;
1168 }
1169 else
1170 *(WORD32*)pi1_left_pred_mode = NOT_VALID;
1171 }
1172 }
1173 /* One time pointer initialisations*/
1174 pi2_y_coeff1 = pi2_y_coeff;
1175 pu1_top_left = pu1_ytop_left;
1176
1177 /* Scan the sub-blocks in Raster Scan Order */
1178 for(u1_sub_mb_num = 0; u1_sub_mb_num < 16; u1_sub_mb_num++)
1179 {
Harish Mahendrakard32eaf42016-04-18 16:38:54 +05301180 /* Align the size to multiple of 8, so that SIMD functions
1181 can read 64 bits at a time. Only 13 bytes are actaully used */
1182 UWORD8 au1_ngbr_pels[16];
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301183
1184 u1_sub_blk_x = u1_sub_mb_num & 0x3;
1185 u1_sub_blk_y = u1_sub_mb_num >> 2;
1186 i1_top_pred_mode = pi1_cur_pred_mode[u1_sub_blk_x];
1187 i1_left_pred_mode = pi1_left_pred_mode[u1_sub_blk_y];
1188 u1_use_top_right_mb = (!!CHECKBIT(ui2_top_rt_mask, u1_sub_mb_num));
1189
1190 /*********** left subblock availability**********/
1191 if(u1_sub_blk_x)
1192 u1_is_left_sub_block = 1;
1193 else
1194 u1_is_left_sub_block =
1195 (u1_sub_blk_y < 2) ?
1196 (CHECKBIT(u2_use_left_mb_pack,
1197 0)) :
1198 (u2_use_left_mb_pack >> 8);
1199
1200 /* CHANGED CODE */
1201 if(u1_sub_blk_y)
1202 u1_is_top_sub_block = 1;
1203
1204 /* CHANGED CODE */
1205 /***************** Top *********************/
1206 if(ps_dec->u4_use_intrapred_line_copy == 1)
1207 {
1208
1209 if(u1_sub_blk_y)
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001210 pu1_top = pu1_luma_rec_buffer - ui_rec_width;
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301211 else
1212 pu1_top = puc_top + (u1_sub_blk_x << 2);
1213 }
1214 else
1215 {
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001216 pu1_top = pu1_luma_rec_buffer - ui_rec_width;
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301217 }
1218 /***************** Top Right *********************/
1219 pu1_top_right = pu1_top + 4;
1220 /***************** Top Left *********************/
1221 pu1_top_left = pu1_top - 1;
1222 /***************** Left *********************/
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001223 pu1_left = pu1_luma_rec_buffer - 1;
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301224 /* CHANGED CODE */
1225
1226 /*---------------------------------------------------------------*/
1227 /* Calculation of Intra prediction mode */
1228 /*---------------------------------------------------------------*/
1229 i1_intra_pred = ((i1_left_pred_mode < 0) | (i1_top_pred_mode < 0)) ?
1230 DC : MIN(i1_left_pred_mode, i1_top_pred_mode);
1231 {
1232 UWORD8 u1_packed_modes = (u1_is_top_sub_block << 1)
1233 + u1_is_left_sub_block;
1234 UWORD8 *pu1_intra_err_codes =
1235 (UWORD8 *)gau1_ih264d_intra_pred_err_code;
1236 UWORD8 uc_b2b0 = ((u1_sub_mb_num & 4) >> 1) | (u1_sub_mb_num & 1);
1237 UWORD8 uc_b3b1 = ((u1_sub_mb_num & 8) >> 2)
1238 | ((u1_sub_mb_num & 2) >> 1);
1239
1240 u1_cur_sub_block = (uc_b3b1 << 2) + uc_b2b0;
1241 PROFILE_DISABLE_INTRA_PRED()
1242 if(!pu1_prev_intra4x4_pred_mode_flag[u1_cur_sub_block])
1243 {
1244 i1_intra_pred =
1245 pu1_rem_intra4x4_pred_mode[u1_cur_sub_block]
1246 + (pu1_rem_intra4x4_pred_mode[u1_cur_sub_block]
1247 >= i1_intra_pred);
1248 }
Harish Mahendrakarf38d14c2016-04-20 16:06:51 +05301249 i1_intra_pred = CLIP3(0, 8, i1_intra_pred);
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301250 {
1251 UWORD8 u1_err_code = pu1_intra_err_codes[i1_intra_pred];
1252
Hamsalekha S9f0dcba2015-06-09 15:54:31 +05301253 if((u1_err_code & u1_packed_modes) ^ u1_err_code)
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301254 {
Hamsalekha S9f0dcba2015-06-09 15:54:31 +05301255 i1_intra_pred = 0;
1256 ps_dec->i4_error_code = ERROR_INTRAPRED;
1257 }
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301258
1259 }
1260 }
1261 {
1262 /* Get neighbour pixels */
1263 /* left pels */
1264 if(u1_is_left_sub_block)
1265 {
1266 WORD32 i;
1267 for(i = 0; i < 4; i++)
1268 au1_ngbr_pels[4 - 1 - i] = pu1_left[i * ui_rec_width];
1269 }
1270 else
1271 {
1272 memset(au1_ngbr_pels, 0, 4);
1273 }
1274
1275 /* top left pels */
1276 au1_ngbr_pels[4] = *pu1_top_left;
1277
1278 /* top pels */
1279 if(u1_is_top_sub_block)
1280 {
1281 memcpy(au1_ngbr_pels + 4 + 1, pu1_top, 4);
1282 }
1283 else
1284 {
1285 memset(au1_ngbr_pels + 4 + 1, 0, 4);
1286 }
1287
1288 /* top right pels */
1289 if(u1_use_top_right_mb)
1290 {
1291 memcpy(au1_ngbr_pels + 4 * 2 + 1, pu1_top_right, 4);
1292 }
1293 else if(u1_is_top_sub_block)
1294 {
1295 memset(au1_ngbr_pels + 4 * 2 + 1, au1_ngbr_pels[4 * 2], 4);
1296 }
1297 }
1298 PROFILE_DISABLE_INTRA_PRED()
1299 ps_dec->apf_intra_pred_luma_4x4[i1_intra_pred](
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001300 au1_ngbr_pels, pu1_luma_rec_buffer, 1,
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301301 ui_rec_width,
1302 ((u1_is_top_sub_block << 2) | u1_is_left_sub_block));
1303
1304 /* CHANGED CODE */
1305 if(CHECKBIT(ui2_luma_csbp, u1_sub_mb_num))
1306 {
1307 WORD16 ai2_tmp[16];
1308 PROFILE_DISABLE_IQ_IT_RECON()
1309 {
1310 if(CHECKBIT(u4_luma_dc_only_csbp, u1_sub_mb_num))
1311 {
1312 ps_dec->pf_iquant_itrans_recon_luma_4x4_dc(
1313 pi2_y_coeff1,
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001314 pu1_luma_rec_buffer,
1315 pu1_luma_rec_buffer,
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301316 ui_rec_width,
1317 ui_rec_width,
1318 gau2_ih264_iquant_scale_4x4[ps_cur_mb_info->u1_qp_rem6],
1319 (UWORD16 *)ps_dec->s_high_profile.i2_scalinglist4x4[0],
1320 ps_cur_mb_info->u1_qp_div6, ai2_tmp, 0,
1321 NULL);
1322 }
1323 else
1324 {
1325 ps_dec->pf_iquant_itrans_recon_luma_4x4(
1326 pi2_y_coeff1,
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001327 pu1_luma_rec_buffer,
1328 pu1_luma_rec_buffer,
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301329 ui_rec_width,
1330 ui_rec_width,
1331 gau2_ih264_iquant_scale_4x4[ps_cur_mb_info->u1_qp_rem6],
1332 (UWORD16 *)ps_dec->s_high_profile.i2_scalinglist4x4[0],
1333 ps_cur_mb_info->u1_qp_div6, ai2_tmp, 0,
1334 NULL);
1335 }
1336 }
1337
1338 }
1339
1340 /*---------------------------------------------------------------*/
1341 /* Update sub block number */
1342 /*---------------------------------------------------------------*/
1343 pi2_y_coeff1 += 16;
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001344 pu1_luma_rec_buffer +=
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301345 (u1_sub_blk_x == 3) ? (ui_rec_width << 2) - 12 : 4;
1346 pu1_luma_pred_buffer +=
1347 (u1_sub_blk_x == 3) ? (ui_pred_width << 2) - 12 : 4;
1348 /* CHANGED CODE */
1349 pi1_cur_pred_mode[u1_sub_blk_x] = i1_intra_pred;
1350 pi1_left_pred_mode[u1_sub_blk_y] = i1_intra_pred;
1351 }
1352 }
1353 else if((u1_mb_type == I_4x4_MB) && (ps_cur_mb_info->u1_tran_form8x8 == 1))
1354 {
1355 UWORD8 u1_is_left_sub_block, u1_is_top_sub_block = uc_useTopMB;
1356 UWORD8 u1_sub_blk_x, u1_sub_blk_y, u1_sub_mb_num;
1357 WORD8 i1_top_pred_mode;
1358 WORD8 i1_left_pred_mode;
1359 UWORD8 *pu1_top, *pu1_left, *pu1_top_left;
1360 WORD8 *pi1_cur_pred_mode, *pi1_left_pred_mode, *pc_topPredMode;
1361 UWORD16 ui2_left_pred_buf_width = 0xffff;
1362 WORD8 i1_intra_pred;
1363 UWORD8 *pu1_prev_intra4x4_pred_mode_flag = pu1_prev_intra4x4_pred_mode_data;
1364 UWORD8 *pu1_rem_intra4x4_pred_mode = pu1_prev_intra4x4_pred_mode_data + 4;
1365 WORD16 *pi2_y_coeff1;
1366 UWORD16 ui2_top_rt_mask;
1367 UWORD32 u4_4x4_left_offset = 0;
1368
1369 /*--------------------------------------------------------------------*/
1370 /* 8x8 IntraPrediction */
1371 /*--------------------------------------------------------------------*/
1372 /* Calculation of Top Right subblock mask */
1373 /* */
1374 /* (a) Set it to default mask */
1375 /* [It has 0 for sublocks which will never have top-right sub block] */
1376 /* */
1377 /* (b) If top MB is not available */
1378 /* Clear the bits of the first row sub blocks */
1379 /* */
1380 /* (c) Set/Clear bit for top-right sublock of MB */
1381 /* [5 sub-block in decoding order] based on TOP RIGHT MB availablity */
1382 /* */
1383 /* ui2_top_rt_mask: marks availibility of top right(neighbour) */
1384 /* in the 8x8 Block ordering */
1385 /* */
1386 /* tr0 tr1 */
1387 /* 0 1 tr3 */
1388 /* 2 3 */
1389 /* */
1390 /* Top rights for 0 is in top MB */
1391 /* top right of 1 will be in top right MB */
1392 /* top right of 3 in right MB and hence not available */
1393 /* This corresponds to ui2_top_rt_mask having default value 0x4 */
1394 /*--------------------------------------------------------------------*/
1395
1396 ui2_top_rt_mask = (u1_use_top_right_mb << 1) | (0x4);
1397
1398 if(uc_useTopMB)
1399 {
1400 ui2_top_rt_mask |= 0x1;
1401 }
1402
1403 /* Top Related initialisations */
1404 pi1_cur_pred_mode = ps_cur_mb_info->ps_curmb->pi1_intrapredmodes;
1405 pc_topPredMode = ps_cur_mb_info->ps_top_mb->pi1_intrapredmodes;
1406 /*
1407 if(u1_mbaff)
1408 {
1409 pi1_cur_pred_mode += (u2_mbx << 2);
1410 pc_topPredMode = pi1_cur_pred_mode + ps_cur_mb_info->i1_offset;
1411 pi1_cur_pred_mode += (u1_topmb) ? 0: 4;
1412 }
1413 */
1414 if(u1_top_available)
1415 {
1416 if(ps_top_mb->u1_mb_type == I_4x4_MB)
1417 {
1418 *(WORD32*)pi1_cur_pred_mode = *(WORD32*)pc_topPredMode;
1419 }
1420 else
1421 {
1422 *(WORD32*)pi1_cur_pred_mode =
1423 (uc_useTopMB) ? DC_DC_DC_DC : NOT_VALID;
1424 }
1425 }
1426 else
1427 {
1428 *(WORD32*)pi1_cur_pred_mode = NOT_VALID;
1429 }
1430
1431 pu1_top = puc_top - 8;
1432
1433 /*Left Related initialisations*/
1434 pi1_left_pred_mode = ps_dec->pi1_left_pred_mode;
1435
1436 if(!u1_mbaff)
1437 {
1438 if(u1_left_available)
1439 {
1440 if(ps_left_mb->u1_mb_type != I_4x4_MB)
1441 {
1442 *(WORD32*)pi1_left_pred_mode =
1443 (u2_use_left_mb_pack) ?
1444 DC_DC_DC_DC :
1445 NOT_VALID;
1446 }
1447 }
1448 else
1449 {
1450 *(WORD32*)pi1_left_pred_mode = NOT_VALID;
1451 }
1452 }
1453 else
1454 {
1455 UWORD8 u1_curMbfld = ps_cur_mb_info->u1_mb_field_decodingflag;
1456
1457 UWORD8 u1_leftMbfld = ps_left_mb->u1_mb_fld;
1458
1459 if((!u1_curMbfld) && (u1_leftMbfld))
1460 {
1461 u4_4x4_left_offset = 1;
1462 }
1463
1464 if(u1_curMbfld ^ u1_leftMbfld)
1465 {
1466
1467 if(u1_topmb
1468 | ((u1_topmb == 0)
1469 && ((ps_curmb - 1)->u1_mb_type
1470 != I_4x4_MB)))
1471
1472 {
1473 if(u1_left_available)
1474 {
1475 if(ps_left_mb->u1_mb_type != I_4x4_MB)
1476 {
1477 if(CHECKBIT(u2_use_left_mb_pack,0) == 0)
1478 {
1479 *(WORD32*)pi1_left_pred_mode = NOT_VALID;
1480 }
1481 else
1482 {
1483 *(WORD32*)pi1_left_pred_mode = DC_DC_DC_DC;
1484 }
1485 }
1486 }
1487 else
1488 {
1489 *(WORD32*)pi1_left_pred_mode = NOT_VALID;
1490 }
1491
1492 if(u1_curMbfld)
1493 {
1494 if(u1_left_available)
1495 {
1496 if((ps_left_mb + 1)->u1_mb_type != I_4x4_MB)
1497 {
1498 if(u2_use_left_mb_pack >> 8)
1499 {
1500 *(WORD32*)(pi1_left_pred_mode + 4) =
1501 DC_DC_DC_DC;
1502 }
1503 else
1504 {
1505 *(WORD32*)(pi1_left_pred_mode + 4) =
1506 NOT_VALID;
1507 }
1508 }
1509 }
1510 else
1511 {
1512 *(WORD32*)(pi1_left_pred_mode + 4) = NOT_VALID;
1513 }
1514
1515 pi1_left_pred_mode[1] = pi1_left_pred_mode[2];
1516 pi1_left_pred_mode[2] = pi1_left_pred_mode[4];
1517 pi1_left_pred_mode[3] = pi1_left_pred_mode[6];
1518 *(WORD32*)(pi1_left_pred_mode + 4) =
1519 *(WORD32*)pi1_left_pred_mode;
1520 }
1521 else
1522 {
1523 pi1_left_pred_mode[7] = pi1_left_pred_mode[3];
1524 pi1_left_pred_mode[6] = pi1_left_pred_mode[3];
1525 pi1_left_pred_mode[5] = pi1_left_pred_mode[2];
1526 pi1_left_pred_mode[4] = pi1_left_pred_mode[2];
1527 pi1_left_pred_mode[3] = pi1_left_pred_mode[1];
1528 pi1_left_pred_mode[2] = pi1_left_pred_mode[1];
1529 pi1_left_pred_mode[1] = pi1_left_pred_mode[0];
1530 }
1531 }
1532 pi1_left_pred_mode += (u1_topmb) ? 0 : 4;
1533 }
1534 else
1535 {
1536 pi1_left_pred_mode += (u1_topmb) ? 0 : 4;
1537
1538 if(u1_left_available)
1539 {
1540 if(ps_left_mb->u1_mb_type != I_4x4_MB)
1541 {
1542 *(WORD32*)pi1_left_pred_mode =
1543 (u2_use_left_mb_pack) ?
1544 DC_DC_DC_DC :
1545 NOT_VALID;
1546 }
1547 }
1548 else
1549 {
1550 *(WORD32*)pi1_left_pred_mode = NOT_VALID;
1551 }
1552 }
1553 }
1554
1555 /* One time pointer initialisations*/
1556 pi2_y_coeff1 = pi2_y_coeff;
1557
1558 if(u1_use_top_left_mb)
1559 {
1560 pu1_top_left = pu1_ytop_left;
1561 }
1562 else
1563 {
1564 pu1_top_left = NULL;
1565 }
1566
1567 /* Scan the sub-blocks in Raster Scan Order */
1568 for(u1_sub_mb_num = 0; u1_sub_mb_num < 4; u1_sub_mb_num++)
1569 {
1570 u1_sub_blk_x = (u1_sub_mb_num & 0x1);
1571 u1_sub_blk_y = (u1_sub_mb_num >> 1);
1572 i1_top_pred_mode = pi1_cur_pred_mode[u1_sub_blk_x << 1];
1573 i1_left_pred_mode = pi1_left_pred_mode[u1_sub_blk_y << 1];
1574
1575 if(2 == u1_sub_mb_num)
1576 {
1577 i1_left_pred_mode = pi1_left_pred_mode[(u1_sub_blk_y << 1)
1578 + u4_4x4_left_offset];
1579 }
1580
1581 u1_use_top_right_mb = (!!CHECKBIT(ui2_top_rt_mask, u1_sub_mb_num));
1582
1583 /*********** left subblock availability**********/
1584 if(u1_sub_blk_x)
1585 {
1586 u1_is_left_sub_block = 1;
1587 }
1588 else
1589 {
1590 u1_is_left_sub_block =
1591 (u1_sub_blk_y < 1) ?
1592 (CHECKBIT(u2_use_left_mb_pack,
1593 0)) :
1594 (u2_use_left_mb_pack >> 8);
1595 }
1596
1597 /***************** Top *********************/
1598 if(u1_sub_blk_y)
1599 {
1600 u1_is_top_sub_block = 1;
1601 // sushant
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001602 pu1_top = /*pu1_luma_pred_buffer*/pu1_luma_rec_buffer - ui_rec_width;
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301603 }
1604 else
1605 {
1606 pu1_top += 8;
1607 }
1608
1609 /***************** Left *********************/
1610 if((u1_sub_blk_x) | (u4_num_pmbair != 0))
1611 {
1612 // sushant
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001613 pu1_left = /*pu1_luma_pred_buffer*/pu1_luma_rec_buffer - 1;
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301614 ui2_left_pred_buf_width = ui_rec_width;
1615 }
1616 else
1617 {
1618 pu1_left = pu1_yleft;
1619 pu1_yleft += (ui_rec_width << 3);
1620 ui2_left_pred_buf_width = ui_rec_width;
1621 }
1622
1623 /***************** Top Left *********************/
1624 if(u1_sub_mb_num)
1625 {
1626 pu1_top_left = (u1_sub_blk_x) ?
1627 pu1_top - 1 : pu1_left - ui_rec_width;
1628
1629 if((u1_sub_blk_x && (!u1_is_top_sub_block))
1630 || ((!u1_sub_blk_x) && (!u1_is_left_sub_block)))
1631 {
1632 pu1_top_left = NULL;
1633 }
1634 }
1635
1636 /*---------------------------------------------------------------*/
1637 /* Calculation of Intra prediction mode */
1638 /*---------------------------------------------------------------*/
1639 i1_intra_pred = ((i1_left_pred_mode < 0) | (i1_top_pred_mode < 0)) ?
1640 DC : MIN(i1_left_pred_mode, i1_top_pred_mode);
1641 {
1642 UWORD8 u1_packed_modes = (u1_is_top_sub_block << 1)
1643 + u1_is_left_sub_block;
1644 UWORD8 *pu1_intra_err_codes =
1645 (UWORD8 *)gau1_ih264d_intra_pred_err_code;
1646
1647 /********************************************************************/
1648 /* Same intra4x4_pred_mode array is filled with intra4x4_pred_mode */
1649 /* for a MB with 8x8 intrapredicition */
1650 /********************************************************************/
1651 PROFILE_DISABLE_INTRA_PRED()
1652 if(!pu1_prev_intra4x4_pred_mode_flag[u1_sub_mb_num])
1653 {
1654 i1_intra_pred = pu1_rem_intra4x4_pred_mode[u1_sub_mb_num]
1655 + (pu1_rem_intra4x4_pred_mode[u1_sub_mb_num]
1656 >= i1_intra_pred);
1657 }
Harish Mahendrakarf38d14c2016-04-20 16:06:51 +05301658 i1_intra_pred = CLIP3(0, 8, i1_intra_pred);
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301659 {
1660 UWORD8 u1_err_code = pu1_intra_err_codes[i1_intra_pred];
1661
1662 if((u1_err_code & u1_packed_modes) ^ u1_err_code)
Hamsalekha S9f0dcba2015-06-09 15:54:31 +05301663 {
1664 i1_intra_pred = 0;
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001665 ps_dec->i4_error_code = ERROR_INTRAPRED;
Hamsalekha S9f0dcba2015-06-09 15:54:31 +05301666 }
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301667 }
1668 }
1669
1670 {
Harish Mahendrakard32eaf42016-04-18 16:38:54 +05301671 /* Align the size to multiple of 8, so that SIMD functions
1672 can read 64 bits at a time. Only 25 bytes are actaully used */
Thomala Srinivas2286dbb2018-10-09 18:29:54 +05301673 UWORD8 au1_ngbr_pels[32] = {0};
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301674 WORD32 ngbr_avail;
1675 ngbr_avail = u1_is_left_sub_block << 0;
1676 ngbr_avail |= u1_is_top_sub_block << 2;
1677
1678 if(pu1_top_left)
1679 ngbr_avail |= 1 << 1;
1680
1681 ngbr_avail |= u1_use_top_right_mb << 3;
1682 PROFILE_DISABLE_INTRA_PRED()
1683 {
1684 ps_dec->pf_intra_pred_ref_filtering(pu1_left, pu1_top_left,
1685 pu1_top, au1_ngbr_pels,
1686 ui2_left_pred_buf_width,
1687 ngbr_avail);
1688
1689 ps_dec->apf_intra_pred_luma_8x8[i1_intra_pred](
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001690 au1_ngbr_pels, pu1_luma_rec_buffer, 1,
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301691 ui_rec_width,
1692 ((u1_is_top_sub_block << 2) | u1_is_left_sub_block));
1693 }
1694 }
1695
1696 /* Inverse Transform and Reconstruction */
1697 if(CHECKBIT(ps_cur_mb_info->u1_cbp, u1_sub_mb_num))
1698 {
1699 WORD16 *pi2_scale_matrix_ptr;
1700 WORD16 ai2_tmp[64];
1701
1702 pi2_scale_matrix_ptr =
1703 ps_dec->s_high_profile.i2_scalinglist8x8[0];
1704 PROFILE_DISABLE_IQ_IT_RECON()
1705 {
1706 if(CHECKBIT(u4_luma_dc_only_cbp, u1_sub_mb_num))
1707 {
1708 ps_dec->pf_iquant_itrans_recon_luma_8x8_dc(
1709 pi2_y_coeff1,
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001710 pu1_luma_rec_buffer,
1711 pu1_luma_rec_buffer,
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301712 ui_rec_width,
1713 ui_rec_width,
1714 gau1_ih264d_dequant8x8_cavlc[ps_cur_mb_info->u1_qp_rem6],
1715 (UWORD16 *)pi2_scale_matrix_ptr,
1716 ps_cur_mb_info->u1_qp_div6, ai2_tmp, 0,
1717 NULL);
1718 }
1719 else
1720 {
1721 ps_dec->pf_iquant_itrans_recon_luma_8x8(
1722 pi2_y_coeff1,
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001723 pu1_luma_rec_buffer,
1724 pu1_luma_rec_buffer,
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301725 ui_rec_width,
1726 ui_rec_width,
1727 gau1_ih264d_dequant8x8_cavlc[ps_cur_mb_info->u1_qp_rem6],
1728 (UWORD16 *)pi2_scale_matrix_ptr,
1729 ps_cur_mb_info->u1_qp_div6, ai2_tmp, 0,
1730 NULL);
1731 }
1732 }
1733
1734 }
1735
1736 /*---------------------------------------------------------------*/
1737 /* Update sub block number */
1738 /*---------------------------------------------------------------*/
1739 pi2_y_coeff1 += 64;
1740
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001741 pu1_luma_rec_buffer +=
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301742 (u1_sub_blk_x == 1) ?
1743 (ui_rec_width << 3) - (8 * 1) : 8;
1744
1745 /*---------------------------------------------------------------*/
1746 /* Pred mode filled in terms of 4x4 block so replicated in 2 */
1747 /* locations. */
1748 /*---------------------------------------------------------------*/
1749 pi1_cur_pred_mode[u1_sub_blk_x << 1] = i1_intra_pred;
1750 pi1_cur_pred_mode[(u1_sub_blk_x << 1) + 1] = i1_intra_pred;
1751 pi1_left_pred_mode[u1_sub_blk_y << 1] = i1_intra_pred;
1752 pi1_left_pred_mode[(u1_sub_blk_y << 1) + 1] = i1_intra_pred;
1753 }
1754 }
1755 /* Decode Chroma Block */
1756 ih264d_unpack_chroma_coeff4x4_mb(ps_dec,
1757 ps_cur_mb_info);
1758 /*--------------------------------------------------------------------*/
1759 /* Chroma Blocks decoding */
1760 /*--------------------------------------------------------------------*/
1761 {
1762 UWORD8 u1_intra_chrom_pred_mode;
1763 UWORD8 u1_chroma_cbp = (UWORD8)(ps_cur_mb_info->u1_cbp >> 4);
1764
1765 /*--------------------------------------------------------------------*/
1766 /* Perform Chroma intra prediction */
1767 /*--------------------------------------------------------------------*/
1768
1769 u1_intra_chrom_pred_mode = CHROMA_TO_LUMA_INTRA_MODE(
1770 ps_cur_mb_info->u1_chroma_pred_mode);
1771
1772 {
1773 UWORD8 u1_packed_modes = (u1_top_available << 1)
1774 + u1_left_available;
1775 UWORD8 u1_err_code =
1776 (u1_intra_chrom_pred_mode & 1) ?
1777 u1_intra_chrom_pred_mode :
1778 (u1_intra_chrom_pred_mode ^ 2);
1779 if((u1_err_code & u1_packed_modes) ^ u1_err_code)
Hamsalekha S9f0dcba2015-06-09 15:54:31 +05301780 {
1781 u1_intra_chrom_pred_mode = 0;
Marco Nelissen8ef4c3f2015-06-03 07:26:32 -07001782 ps_dec->i4_error_code = ERROR_INTRAPRED;
Hamsalekha S9f0dcba2015-06-09 15:54:31 +05301783 }
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301784 }
1785
1786 /* CHANGED CODE */
1787 if(u1_chroma_cbp != CBPC_ALLZERO)
1788 {
1789 UWORD16 u2_chroma_csbp =
1790 (u1_chroma_cbp == CBPC_ACZERO) ?
1791 0 : ps_cur_mb_info->u2_chroma_csbp;
1792 UWORD32 u4_scale_u;
1793 UWORD32 u4_scale_v;
1794
1795 {
1796 UWORD16 au2_ngbr_pels[33];
1797 UWORD8 *pu1_ngbr_pels = (UWORD8 *)au2_ngbr_pels;
1798 UWORD16 *pu2_left_uv;
1799 UWORD16 *pu2_topleft_uv;
1800 WORD32 use_left1 = (u2_use_left_mb_pack & 0x0ff);
1801 WORD32 use_left2 = (u2_use_left_mb_pack & 0xff00) >> 8;
1802
1803 pu2_left_uv = (UWORD16 *)pu1_uleft;
1804 pu2_topleft_uv = (UWORD16 *)pu1_u_top_left;
1805 /* Get neighbour pixels */
1806 /* left pels */
1807 if(u2_use_left_mb_pack)
1808 {
1809 WORD32 i;
1810 if(use_left1)
1811 {
1812 for(i = 0; i < 4; i++)
1813 au2_ngbr_pels[8 - 1 - i] = pu2_left_uv[i
1814 * u4_recwidth_cr / YUV420SP_FACTOR];
1815 }
1816 else
1817 {
1818 memset(au2_ngbr_pels + 4, 0, 4 * sizeof(UWORD16));
1819 }
1820
1821 if(use_left2)
1822 {
1823 for(i = 4; i < 8; i++)
1824 au2_ngbr_pels[8 - 1 - i] = pu2_left_uv[i
1825 * u4_recwidth_cr / YUV420SP_FACTOR];
1826 }
1827 else
1828 {
1829 memset(au2_ngbr_pels, 0, 4 * sizeof(UWORD16));
1830 }
1831 }
1832 else
1833 {
1834 memset(au2_ngbr_pels, 0, 8 * sizeof(UWORD16));
1835 }
1836
1837 /* top left pels */
1838 au2_ngbr_pels[8] = *pu2_topleft_uv;
1839
1840 /* top pels */
1841 if(uc_useTopMB)
1842 {
1843 memcpy(au2_ngbr_pels + 8 + 1, pu1_top_u,
1844 8 * sizeof(UWORD16));
1845 }
1846 else
1847 {
1848 memset(au2_ngbr_pels + 8 + 1, 0, 8 * sizeof(UWORD16));
1849 }
1850
1851 PROFILE_DISABLE_INTRA_PRED()
1852 ps_dec->apf_intra_pred_chroma[u1_intra_chrom_pred_mode](
1853 pu1_ngbr_pels,
1854 pu1_mb_cb_rei1_buffer,
1855 1,
1856 u4_recwidth_cr,
1857 ((uc_useTopMB << 2) | (use_left2 << 4)
1858 | use_left1));
1859 }
1860 u4_scale_u = ps_cur_mb_info->u1_qpc_div6;
1861 u4_scale_v = ps_cur_mb_info->u1_qpcr_div6;
1862 pi2_y_coeff = ps_dec->pi2_coeff_data;
1863
1864 {
1865 UWORD32 i;
1866 WORD16 ai2_tmp[16];
1867 for(i = 0; i < 4; i++)
1868 {
1869 WORD16 *pi2_level = pi2_y_coeff + (i << 4);
1870 UWORD8 *pu1_pred_sblk = pu1_mb_cb_rei1_buffer
1871 + ((i & 0x1) * BLK_SIZE * YUV420SP_FACTOR)
1872 + (i >> 1) * (u4_recwidth_cr << 2);
1873 PROFILE_DISABLE_IQ_IT_RECON()
1874 {
1875 if(CHECKBIT(u2_chroma_csbp, i))
1876 {
1877 ps_dec->pf_iquant_itrans_recon_chroma_4x4(
1878 pi2_level,
1879 pu1_pred_sblk,
1880 pu1_pred_sblk,
1881 u4_recwidth_cr,
1882 u4_recwidth_cr,
1883 gau2_ih264_iquant_scale_4x4[ps_cur_mb_info->u1_qpc_rem6],
1884 (UWORD16 *)ps_dec->s_high_profile.i2_scalinglist4x4[1],
1885 u4_scale_u, ai2_tmp, pi2_level);
1886 }
1887 else if(pi2_level[0] != 0)
1888 {
1889 ps_dec->pf_iquant_itrans_recon_chroma_4x4_dc(
1890 pi2_level,
1891 pu1_pred_sblk,
1892 pu1_pred_sblk,
1893 u4_recwidth_cr,
1894 u4_recwidth_cr,
1895 gau2_ih264_iquant_scale_4x4[ps_cur_mb_info->u1_qpc_rem6],
1896 (UWORD16 *)ps_dec->s_high_profile.i2_scalinglist4x4[1],
1897 u4_scale_u, ai2_tmp, pi2_level);
1898 }
1899 }
1900
1901 }
1902 }
1903
1904 pi2_y_coeff += MB_CHROM_SIZE;
1905 u2_chroma_csbp = u2_chroma_csbp >> 4;
1906 {
1907 UWORD32 i;
1908 WORD16 ai2_tmp[16];
1909 for(i = 0; i < 4; i++)
1910 {
1911 WORD16 *pi2_level = pi2_y_coeff + (i << 4);
1912 UWORD8 *pu1_pred_sblk = pu1_mb_cb_rei1_buffer + 1
1913 + ((i & 0x1) * BLK_SIZE * YUV420SP_FACTOR)
1914 + (i >> 1) * (u4_recwidth_cr << 2);
1915 PROFILE_DISABLE_IQ_IT_RECON()
1916 {
1917 if(CHECKBIT(u2_chroma_csbp, i))
1918 {
1919 ps_dec->pf_iquant_itrans_recon_chroma_4x4(
1920 pi2_level,
1921 pu1_pred_sblk,
1922 pu1_pred_sblk,
1923 u4_recwidth_cr,
1924 u4_recwidth_cr,
1925 gau2_ih264_iquant_scale_4x4[ps_cur_mb_info->u1_qpcr_rem6],
1926 (UWORD16 *)ps_dec->s_high_profile.i2_scalinglist4x4[2],
1927 u4_scale_v, ai2_tmp, pi2_level);
1928 }
1929 else if(pi2_level[0] != 0)
1930 {
1931 ps_dec->pf_iquant_itrans_recon_chroma_4x4_dc(
1932 pi2_level,
1933 pu1_pred_sblk,
1934 pu1_pred_sblk,
1935 u4_recwidth_cr,
1936 u4_recwidth_cr,
1937 gau2_ih264_iquant_scale_4x4[ps_cur_mb_info->u1_qpcr_rem6],
1938 (UWORD16 *)ps_dec->s_high_profile.i2_scalinglist4x4[2],
1939 u4_scale_v, ai2_tmp, pi2_level);
1940 }
1941 }
1942 }
1943 }
1944
1945 }
1946 else
1947 {
1948 /* If no inverse transform is needed, pass recon buffer pointer */
1949 /* to Intraprediction module instead of pred buffer pointer */
1950 {
1951 UWORD16 au2_ngbr_pels[33];
1952 UWORD8 *pu1_ngbr_pels = (UWORD8 *)au2_ngbr_pels;
1953 UWORD16 *pu2_left_uv;
1954 UWORD16 *pu2_topleft_uv;
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301955 WORD32 use_left1 = (u2_use_left_mb_pack & 0x0ff);
1956 WORD32 use_left2 = (u2_use_left_mb_pack & 0xff00) >> 8;
1957
Martin Storsjo086dd8e2015-06-13 00:35:01 +03001958 pu2_topleft_uv = (UWORD16 *)pu1_u_top_left;
1959 pu2_left_uv = (UWORD16 *)pu1_uleft;
1960
Hamsalekha S8d3d3032015-03-13 21:24:58 +05301961 /* Get neighbour pixels */
1962 /* left pels */
1963 if(u2_use_left_mb_pack)
1964 {
1965 WORD32 i;
1966 if(use_left1)
1967 {
1968 for(i = 0; i < 4; i++)
1969 au2_ngbr_pels[8 - 1 - i] = pu2_left_uv[i
1970 * u4_recwidth_cr / YUV420SP_FACTOR];
1971 }
1972 else
1973 {
1974 memset(au2_ngbr_pels + 4, 0, 4 * sizeof(UWORD16));
1975 }
1976
1977 if(use_left2)
1978 {
1979 for(i = 4; i < 8; i++)
1980 au2_ngbr_pels[8 - 1 - i] = pu2_left_uv[i
1981 * u4_recwidth_cr / YUV420SP_FACTOR];
1982 }
1983 else
1984 {
1985 memset(au2_ngbr_pels, 0, 4 * sizeof(UWORD16));
1986 }
1987
1988 }
1989 else
1990 {
1991 memset(au2_ngbr_pels, 0, 8 * sizeof(UWORD16));
1992 }
1993
1994 /* top left pels */
1995 au2_ngbr_pels[8] = *pu2_topleft_uv;
1996
1997 /* top pels */
1998 if(uc_useTopMB)
1999 {
2000 memcpy(au2_ngbr_pels + 8 + 1, pu1_top_u,
2001 8 * sizeof(UWORD16));
2002 }
2003 else
2004 {
2005 memset(au2_ngbr_pels + 8 + 1, 0, 8 * sizeof(UWORD16));
2006 }
2007
2008 PROFILE_DISABLE_INTRA_PRED()
2009 ps_dec->apf_intra_pred_chroma[u1_intra_chrom_pred_mode](
2010 pu1_ngbr_pels,
2011 pu1_mb_cb_rei1_buffer,
2012 1,
2013 u4_recwidth_cr,
2014 ((uc_useTopMB << 2) | (use_left2 << 4)
2015 | use_left1));
2016 }
2017
2018 }
2019
2020 }
2021 return OK;
2022}