July 2006 - ffmpeg-cvslog - lists.ffmpeg.org

r5849 - trunk/libavformat/mxf.c
by subversion＠mplayerhq.hu 29 Jul '06

29 Jul '06

Author: bcoudurier Date: Sat Jul 29 15:18:24 2006 New Revision: 5849 Modified: trunk/libavformat/mxf.c Log: add some debug infos Modified: trunk/libavformat/mxf.c ============================================================================== --- trunk/libavformat/mxf.c (original) +++ trunk/libavformat/mxf.c Sat Jul 29 15:18:24 2006 @@ -221,6 +221,9 @@ av_log(s, AV_LOG_ERROR, "error reading KLV packet\n"); return -1; } +#ifdef DEBUG + PRINT_KEY(klv.key); +#endif if (IS_KLV_KEY(klv.key, mxf_essence_element_key)) { av_get_packet(&s->pb, pkt, klv.length); pkt->stream_index = mxf_get_stream_index(s, &klv); @@ -816,6 +819,9 @@ av_log(s, AV_LOG_ERROR, "error reading KLV packet\n"); return -1; } +#ifdef DEBUG + PRINT_KEY(klv.key); +#endif if (IS_KLV_KEY(klv.key, mxf_metadata_track_key)) ret = mxf_read_metadata_track(mxf, &klv); else if (IS_KLV_KEY(klv.key, mxf_metadata_static_track_key))

1 0

r5848 - trunk/libavformat/mxf.c
by subversion＠mplayerhq.hu 29 Jul '06

29 Jul '06

Author: bcoudurier Date: Sat Jul 29 15:13:08 2006 New Revision: 5848 Modified: trunk/libavformat/mxf.c Log: get rid of DECLARED_ALIGNED Modified: trunk/libavformat/mxf.c ============================================================================== --- trunk/libavformat/mxf.c (original) +++ trunk/libavformat/mxf.c Sat Jul 29 15:13:08 2006 @@ -45,9 +45,8 @@ //#define DEBUG #include "avformat.h" -#include "dsputil.h" -typedef DECLARE_ALIGNED_16(uint8_t, UID[16]); +typedef uint8_t UID[16]; enum MXFPackageType { MaterialPackage, @@ -83,7 +82,7 @@ MXFSequence *sequence; /* mandatory, and only one */ UID sequence_ref; int track_id; - int track_number; + uint8_t track_number[4]; AVRational edit_rate; } MXFTrack; @@ -160,7 +159,7 @@ /* partial keys to match */ static const uint8_t mxf_header_partition_pack_key[] = { 0x06,0x0e,0x2b,0x34,0x02,0x05,0x01,0x01,0x0d,0x01,0x02,0x01,0x01,0x02 }; -static const uint8_t mxf_essence_element_key[] = { 0x06,0x0e,0x2b,0x34,0x01,0x02,0x01 }; +static const uint8_t mxf_essence_element_key[] = { 0x06,0x0e,0x2b,0x34,0x01,0x02,0x01,0x01,0x0d,0x01,0x03,0x01 }; #define IS_KLV_KEY(x, y) (!memcmp(x, y, sizeof(y))) @@ -202,11 +201,12 @@ static int mxf_get_stream_index(AVFormatContext *s, KLVPacket *klv) { - int id = BE_32(klv->key + 12); /* SMPTE 379M 7.3 */ int i; for (i = 0; i < s->nb_streams; i++) { - if (s->streams[i]->id == id) + MXFTrack *track = s->streams[i]->priv_data; + /* SMPTE 379M 7.3 */ + if (!memcmp(klv->key + sizeof(mxf_essence_element_key), track->track_number, sizeof(track->track_number))) return i; } return -1; @@ -415,8 +415,7 @@ track->track_id = get_be32(pb); break; case 0x4804: - track->track_number = get_be32(pb); - dprintf("track number 0x%04X\n", track->track_number); + get_buffer(pb, track->track_number, 4); break; case 0x4B01: track->edit_rate.den = get_be32(pb); @@ -742,7 +741,8 @@ if (!source_track) continue; - st = av_new_stream(mxf->fc, source_track->track_number); + st = av_new_stream(mxf->fc, source_track->track_id); + st->priv_data = source_track; st->duration = component->duration; if (st->duration == -1) st->duration = AV_NOPTS_VALUE;

1 0

r5847 - trunk/libavformat/riff.c
by subversion＠mplayerhq.hu 29 Jul '06

29 Jul '06

Author: stefang Date: Sat Jul 29 10:50:29 2006 New Revision: 5847 Modified: trunk/libavformat/riff.c Log: add new tag 'CAVS' for Chinese AVS codec Modified: trunk/libavformat/riff.c ============================================================================== --- trunk/libavformat/riff.c (original) +++ trunk/libavformat/riff.c Sat Jul 29 10:50:29 2006 @@ -151,6 +151,7 @@ { CODEC_ID_CSCD, MKTAG('C', 'S', 'C', 'D') }, { CODEC_ID_ZMBV, MKTAG('Z', 'M', 'B', 'V') }, { CODEC_ID_KMVC, MKTAG('K', 'M', 'V', 'C') }, + { CODEC_ID_CAVS, MKTAG('C', 'A', 'V', 'S') }, { CODEC_ID_RAWVIDEO, 0 }, { CODEC_ID_NONE, 0 }, };

1 0

r5846 - in trunk/libavcodec: Makefile avcodec.h cavs.c cavsdsp.c i386/cavsdsp_mmx.c i386/dsputil_mmx.c
by subversion＠mplayerhq.hu 29 Jul '06

29 Jul '06

Author: stefang Date: Sat Jul 29 10:45:33 2006 New Revision: 5846 Added: trunk/libavcodec/i386/cavsdsp_mmx.c Modified: trunk/libavcodec/Makefile trunk/libavcodec/avcodec.h trunk/libavcodec/cavs.c trunk/libavcodec/cavsdsp.c trunk/libavcodec/i386/dsputil_mmx.c Log: some MMX optimizations for the CAVS decoder Modified: trunk/libavcodec/Makefile ============================================================================== --- trunk/libavcodec/Makefile (original) +++ trunk/libavcodec/Makefile Sat Jul 29 10:45:33 2006 @@ -288,6 +288,9 @@ ifeq ($(CONFIG_GPL),yes) OBJS += i386/idct_mmx_xvid.o endif +ifeq ($(CONFIG_CAVS_DECODER),yes) +OBJS += i386/cavsdsp_mmx.o +endif ifeq ($(TARGET_BUILTIN_VECTOR),yes) i386/fft_sse.o: CFLAGS+= -msse depend: CFLAGS+= -msse Modified: trunk/libavcodec/avcodec.h ============================================================================== --- trunk/libavcodec/avcodec.h (original) +++ trunk/libavcodec/avcodec.h Sat Jul 29 10:45:33 2006 @@ -1194,6 +1194,7 @@ #define FF_IDCT_VP3 12 #define FF_IDCT_IPP 13 #define FF_IDCT_XVIDMMX 14 +#define FF_IDCT_CAVS 15 /** * slice count. Modified: trunk/libavcodec/cavs.c ============================================================================== --- trunk/libavcodec/cavs.c (original) +++ trunk/libavcodec/cavs.c Sat Jul 29 10:45:33 2006 @@ -78,6 +78,7 @@ int qp; int qp_fixed; int cbp; + ScanTable scantable; /** intra prediction is done with un-deblocked samples they are saved here before deblocking the MB */ @@ -97,6 +98,7 @@ int scale_den[2]; ///< for scaling neighbouring MVs int got_keyframe; + DCTELEM *block; } AVSContext; /***************************************************************************** @@ -649,10 +651,9 @@ int dqm = dequant_mul[qp]; int dqs = dequant_shift[qp]; int dqa = 1 << (dqs - 1); - const uint8_t *scantab = ff_zigzag_direct; - DCTELEM block[64]; + const uint8_t *scantab = h->scantable.permutated; + DCTELEM *block = h->block; - memset(block,0,64*sizeof(DCTELEM)); for(i=0;i<65;i++) { level_code = get_ue_code(gb,r->golomb_order); if(level_code >= ESCAPE_CODE) { @@ -1135,8 +1136,10 @@ enum mb_t mb_type; if (!s->context_initialized) { + s->avctx->idct_algo = FF_IDCT_CAVS; if (MPV_common_init(s) < 0) return -1; + ff_init_scantable(s->dsp.idct_permutation,&h->scantable,ff_zigzag_direct); } get_bits(&s->gb,16);//bbv_dwlay if(h->stc == PIC_PB_START_CODE) { @@ -1281,6 +1284,7 @@ /* alloc space for co-located MVs and types */ h->col_mv = av_malloc( h->mb_width*h->mb_height*4*sizeof(vector_t)); h->col_type_base = av_malloc(h->mb_width*h->mb_height); + h->block = av_mallocz(64*sizeof(DCTELEM)); } static int decode_seq_header(AVSContext *h) { @@ -1478,6 +1482,7 @@ av_free(h->top_border_v); av_free(h->col_mv); av_free(h->col_type_base); + av_free(h->block); return 0; } Modified: trunk/libavcodec/cavsdsp.c ============================================================================== --- trunk/libavcodec/cavsdsp.c (original) +++ trunk/libavcodec/cavsdsp.c Sat Jul 29 10:45:33 2006 @@ -246,6 +246,7 @@ dst[i + 6*stride] = cm[ dst[i + 6*stride] + ((b1 - b5) >> 7)]; dst[i + 7*stride] = cm[ dst[i + 7*stride] + ((b0 - b4) >> 7)]; } + memset(block,0,64*sizeof(DCTELEM)); } /***************************************************************************** Added: trunk/libavcodec/i386/cavsdsp_mmx.c ============================================================================== --- (empty file) +++ trunk/libavcodec/i386/cavsdsp_mmx.c Sat Jul 29 10:45:33 2006 @@ -0,0 +1,516 @@ +/* + * Chinese AVS video (AVS1-P2, JiZhun profile) decoder. + * Copyright (c) 2006 Stefan Gehrer <stefan.gehrer at gmx.de> + * + * MMX optimised DSP functions, based on H.264 optimisations by + * Michael Niedermayer and Loren Merritt + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "../dsputil.h" +#include "common.h" + +DECLARE_ALIGNED_8(static const uint64_t,ff_pw_4 ) = 0x0004000400040004ULL; +DECLARE_ALIGNED_8(static const uint64_t,ff_pw_5 ) = 0x0005000500050005ULL; +DECLARE_ALIGNED_8(static const uint64_t,ff_pw_7 ) = 0x0007000700070007ULL; +DECLARE_ALIGNED_8(static const uint64_t,ff_pw_42) = 0x002A002A002A002AULL; +DECLARE_ALIGNED_8(static const uint64_t,ff_pw_64) = 0x0040004000400040ULL; +DECLARE_ALIGNED_8(static const uint64_t,ff_pw_96) = 0x0060006000600060ULL; + +/***************************************************************************** + * + * inverse transform + * + ****************************************************************************/ + +#define SUMSUB_BA( a, b ) \ + "paddw "#b", "#a" \n\t"\ + "paddw "#b", "#b" \n\t"\ + "psubw "#a", "#b" \n\t" + +#define SBUTTERFLY(a,b,t,n)\ + "movq " #a ", " #t " \n\t" /* abcd */\ + "punpckl" #n " " #b ", " #a " \n\t" /* aebf */\ + "punpckh" #n " " #b ", " #t " \n\t" /* cgdh */ + +#define TRANSPOSE4(a,b,c,d,t)\ + SBUTTERFLY(a,b,t,wd) /* a=aebf t=cgdh */\ + SBUTTERFLY(c,d,b,wd) /* c=imjn b=kolp */\ + SBUTTERFLY(a,c,d,dq) /* a=aeim d=bfjn */\ + SBUTTERFLY(t,b,c,dq) /* t=cgko c=dhlp */ + +static inline void cavs_idct8_1d(int16_t *block, uint64_t bias) +{ + asm volatile( + "movq 112(%0), %%mm4 \n\t" /* mm4 = src7 */ + "movq 16(%0), %%mm5 \n\t" /* mm5 = src1 */ + "movq 80(%0), %%mm2 \n\t" /* mm2 = src5 */ + "movq 48(%0), %%mm7 \n\t" /* mm7 = src3 */ + "movq %%mm4, %%mm0 \n\t" + "movq %%mm5, %%mm3 \n\t" + "movq %%mm2, %%mm6 \n\t" + "movq %%mm7, %%mm1 \n\t" + + "paddw %%mm4, %%mm4 \n\t" /* mm4 = 2*src7 */ + "paddw %%mm3, %%mm3 \n\t" /* mm3 = 2*src1 */ + "paddw %%mm6, %%mm6 \n\t" /* mm6 = 2*src5 */ + "paddw %%mm1, %%mm1 \n\t" /* mm1 = 2*src3 */ + "paddw %%mm4, %%mm0 \n\t" /* mm0 = 3*src7 */ + "paddw %%mm3, %%mm5 \n\t" /* mm5 = 3*src1 */ + "paddw %%mm6, %%mm2 \n\t" /* mm2 = 3*src5 */ + "paddw %%mm1, %%mm7 \n\t" /* mm7 = 3*src3 */ + "psubw %%mm4, %%mm5 \n\t" /* mm5 = 3*src1 - 2*src7 = a0 */ + "paddw %%mm6, %%mm7 \n\t" /* mm7 = 3*src3 + 2*src5 = a1 */ + "psubw %%mm2, %%mm1 \n\t" /* mm1 = 2*src3 - 3*src5 = a2 */ + "paddw %%mm0, %%mm3 \n\t" /* mm3 = 2*src1 + 3*src7 = a3 */ + + "movq %%mm5, %%mm4 \n\t" + "movq %%mm7, %%mm6 \n\t" + "movq %%mm3, %%mm0 \n\t" + "movq %%mm1, %%mm2 \n\t" + SUMSUB_BA( %%mm7, %%mm5 ) /* mm7 = a0 + a1 mm5 = a0 - a1 */ + "paddw %%mm3, %%mm7 \n\t" /* mm7 = a0 + a1 + a3 */ + "paddw %%mm1, %%mm5 \n\t" /* mm5 = a0 - a1 + a2 */ + "paddw %%mm7, %%mm7 \n\t" + "paddw %%mm5, %%mm5 \n\t" + "paddw %%mm6, %%mm7 \n\t" /* mm7 = b4 */ + "paddw %%mm4, %%mm5 \n\t" /* mm5 = b5 */ + + SUMSUB_BA( %%mm1, %%mm3 ) /* mm1 = a3 + a2 mm3 = a3 - a2 */ + "psubw %%mm1, %%mm4 \n\t" /* mm4 = a0 - a2 - a3 */ + "movq %%mm4, %%mm1 \n\t" /* mm1 = a0 - a2 - a3 */ + "psubw %%mm6, %%mm3 \n\t" /* mm3 = a3 - a2 - a1 */ + "paddw %%mm1, %%mm1 \n\t" + "paddw %%mm3, %%mm3 \n\t" + "psubw %%mm2, %%mm1 \n\t" /* mm1 = b7 */ + "paddw %%mm0, %%mm3 \n\t" /* mm3 = b6 */ + + "movq 32(%0), %%mm2 \n\t" /* mm2 = src2 */ + "movq 96(%0), %%mm6 \n\t" /* mm6 = src6 */ + "movq %%mm2, %%mm4 \n\t" + "movq %%mm6, %%mm0 \n\t" + "psllw $2, %%mm4 \n\t" /* mm4 = 4*src2 */ + "psllw $2, %%mm6 \n\t" /* mm6 = 4*src6 */ + "paddw %%mm4, %%mm2 \n\t" /* mm2 = 5*src2 */ + "paddw %%mm6, %%mm0 \n\t" /* mm0 = 5*src6 */ + "paddw %%mm2, %%mm2 \n\t" + "paddw %%mm0, %%mm0 \n\t" + "psubw %%mm0, %%mm4 \n\t" /* mm4 = 4*src2 - 10*src6 = a7 */ + "paddw %%mm2, %%mm6 \n\t" /* mm6 = 4*src6 + 10*src2 = a6 */ + + "movq (%0), %%mm2 \n\t" /* mm2 = src0 */ + "movq 64(%0), %%mm0 \n\t" /* mm0 = src4 */ + SUMSUB_BA( %%mm0, %%mm2 ) /* mm0 = src0+src4 mm2 = src0-src4 */ + "psllw $3, %%mm0 \n\t" + "psllw $3, %%mm2 \n\t" + "paddw %1, %%mm0 \n\t" /* add rounding bias */ + "paddw %1, %%mm2 \n\t" /* add rounding bias */ + + SUMSUB_BA( %%mm6, %%mm0 ) /* mm6 = a4 + a6 mm0 = a4 - a6 */ + SUMSUB_BA( %%mm4, %%mm2 ) /* mm4 = a5 + a7 mm2 = a5 - a7 */ + SUMSUB_BA( %%mm7, %%mm6 ) /* mm7 = dst0 mm6 = dst7 */ + SUMSUB_BA( %%mm5, %%mm4 ) /* mm5 = dst1 mm4 = dst6 */ + SUMSUB_BA( %%mm3, %%mm2 ) /* mm3 = dst2 mm2 = dst5 */ + SUMSUB_BA( %%mm1, %%mm0 ) /* mm1 = dst3 mm0 = dst4 */ + :: "r"(block), "m"(bias) + ); +} + +static void cavs_idct8_add_mmx(uint8_t *dst, int16_t *block, int stride) +{ + int i; + DECLARE_ALIGNED_8(int16_t, b2[64]); + + for(i=0; i<2; i++){ + DECLARE_ALIGNED_8(uint64_t, tmp); + + cavs_idct8_1d(block+4*i, ff_pw_4); + + asm volatile( + "psraw $3, %%mm7 \n\t" + "psraw $3, %%mm6 \n\t" + "psraw $3, %%mm5 \n\t" + "psraw $3, %%mm4 \n\t" + "psraw $3, %%mm3 \n\t" + "psraw $3, %%mm2 \n\t" + "psraw $3, %%mm1 \n\t" + "psraw $3, %%mm0 \n\t" + "movq %%mm7, %0 \n\t" + TRANSPOSE4( %%mm0, %%mm2, %%mm4, %%mm6, %%mm7 ) + "movq %%mm0, 8(%1) \n\t" + "movq %%mm6, 24(%1) \n\t" + "movq %%mm7, 40(%1) \n\t" + "movq %%mm4, 56(%1) \n\t" + "movq %0, %%mm7 \n\t" + TRANSPOSE4( %%mm7, %%mm5, %%mm3, %%mm1, %%mm0 ) + "movq %%mm7, (%1) \n\t" + "movq %%mm1, 16(%1) \n\t" + "movq %%mm0, 32(%1) \n\t" + "movq %%mm3, 48(%1) \n\t" + : "=m"(tmp) + : "r"(b2+32*i) + : "memory" + ); + } + + for(i=0; i<2; i++){ + cavs_idct8_1d(b2+4*i, ff_pw_64); + + asm volatile( + "psraw $7, %%mm7 \n\t" + "psraw $7, %%mm6 \n\t" + "psraw $7, %%mm5 \n\t" + "psraw $7, %%mm4 \n\t" + "psraw $7, %%mm3 \n\t" + "psraw $7, %%mm2 \n\t" + "psraw $7, %%mm1 \n\t" + "psraw $7, %%mm0 \n\t" + "movq %%mm7, (%0) \n\t" + "movq %%mm5, 16(%0) \n\t" + "movq %%mm3, 32(%0) \n\t" + "movq %%mm1, 48(%0) \n\t" + "movq %%mm0, 64(%0) \n\t" + "movq %%mm2, 80(%0) \n\t" + "movq %%mm4, 96(%0) \n\t" + "movq %%mm6, 112(%0) \n\t" + :: "r"(b2+4*i) + : "memory" + ); + } + + add_pixels_clamped_mmx(b2, dst, stride); + + /* clear block */ + asm volatile( + "pxor %%mm7, %%mm7 \n\t" + "movq %%mm7, (%0) \n\t" + "movq %%mm7, 8(%0) \n\t" + "movq %%mm7, 16(%0) \n\t" + "movq %%mm7, 24(%0) \n\t" + "movq %%mm7, 32(%0) \n\t" + "movq %%mm7, 40(%0) \n\t" + "movq %%mm7, 48(%0) \n\t" + "movq %%mm7, 56(%0) \n\t" + "movq %%mm7, 64(%0) \n\t" + "movq %%mm7, 72(%0) \n\t" + "movq %%mm7, 80(%0) \n\t" + "movq %%mm7, 88(%0) \n\t" + "movq %%mm7, 96(%0) \n\t" + "movq %%mm7, 104(%0) \n\t" + "movq %%mm7, 112(%0) \n\t" + "movq %%mm7, 120(%0) \n\t" + :: "r" (block) + ); +} + +/***************************************************************************** + * + * motion compensation + * + ****************************************************************************/ + +/* vertical filter [-1 -2 96 42 -7 0] */ +#define QPEL_CAVSV1(A,B,C,D,E,F,OP) \ + "movd (%0), "#F" \n\t"\ + "movq "#C", %%mm6 \n\t"\ + "pmullw %5, %%mm6 \n\t"\ + "movq "#D", %%mm7 \n\t"\ + "pmullw %6, %%mm7 \n\t"\ + "psllw $3, "#E" \n\t"\ + "psubw "#E", %%mm6 \n\t"\ + "psraw $3, "#E" \n\t"\ + "paddw %%mm7, %%mm6 \n\t"\ + "paddw "#E", %%mm6 \n\t"\ + "paddw "#B", "#B" \n\t"\ + "pxor %%mm7, %%mm7 \n\t"\ + "add %2, %0 \n\t"\ + "punpcklbw %%mm7, "#F" \n\t"\ + "psubw "#B", %%mm6 \n\t"\ + "psraw $1, "#B" \n\t"\ + "psubw "#A", %%mm6 \n\t"\ + "paddw %4, %%mm6 \n\t"\ + "psraw $7, %%mm6 \n\t"\ + "packuswb %%mm6, %%mm6 \n\t"\ + OP(%%mm6, (%1), A, d) \ + "add %3, %1 \n\t" + +/* vertical filter [ 0 -1 5 5 -1 0] */ +#define QPEL_CAVSV2(A,B,C,D,E,F,OP) \ + "movd (%0), "#F" \n\t"\ + "movq "#C", %%mm6 \n\t"\ + "paddw "#D", %%mm6 \n\t"\ + "pmullw %5, %%mm6 \n\t"\ + "add %2, %0 \n\t"\ + "punpcklbw %%mm7, "#F" \n\t"\ + "psubw "#B", %%mm6 \n\t"\ + "psubw "#E", %%mm6 \n\t"\ + "paddw %4, %%mm6 \n\t"\ + "psraw $3, %%mm6 \n\t"\ + "packuswb %%mm6, %%mm6 \n\t"\ + OP(%%mm6, (%1), A, d) \ + "add %3, %1 \n\t" + +/* vertical filter [ 0 -7 42 96 -2 -1] */ +#define QPEL_CAVSV3(A,B,C,D,E,F,OP) \ + "movd (%0), "#F" \n\t"\ + "movq "#C", %%mm6 \n\t"\ + "pmullw %6, %%mm6 \n\t"\ + "movq "#D", %%mm7 \n\t"\ + "pmullw %5, %%mm7 \n\t"\ + "psllw $3, "#B" \n\t"\ + "psubw "#B", %%mm6 \n\t"\ + "psraw $3, "#B" \n\t"\ + "paddw %%mm7, %%mm6 \n\t"\ + "paddw "#B", %%mm6 \n\t"\ + "paddw "#E", "#E" \n\t"\ + "pxor %%mm7, %%mm7 \n\t"\ + "add %2, %0 \n\t"\ + "punpcklbw %%mm7, "#F" \n\t"\ + "psubw "#E", %%mm6 \n\t"\ + "psraw $1, "#E" \n\t"\ + "psubw "#F", %%mm6 \n\t"\ + "paddw %4, %%mm6 \n\t"\ + "psraw $7, %%mm6 \n\t"\ + "packuswb %%mm6, %%mm6 \n\t"\ + OP(%%mm6, (%1), A, d) \ + "add %3, %1 \n\t" + + +#define QPEL_CAVSVNUM(VOP,OP,ADD,MUL1,MUL2)\ + int w= 2;\ + src -= 2*srcStride;\ + \ + while(w--){\ + asm volatile(\ + "pxor %%mm7, %%mm7 \n\t"\ + "movd (%0), %%mm0 \n\t"\ + "add %2, %0 \n\t"\ + "movd (%0), %%mm1 \n\t"\ + "add %2, %0 \n\t"\ + "movd (%0), %%mm2 \n\t"\ + "add %2, %0 \n\t"\ + "movd (%0), %%mm3 \n\t"\ + "add %2, %0 \n\t"\ + "movd (%0), %%mm4 \n\t"\ + "add %2, %0 \n\t"\ + "punpcklbw %%mm7, %%mm0 \n\t"\ + "punpcklbw %%mm7, %%mm1 \n\t"\ + "punpcklbw %%mm7, %%mm2 \n\t"\ + "punpcklbw %%mm7, %%mm3 \n\t"\ + "punpcklbw %%mm7, %%mm4 \n\t"\ + VOP(%%mm0, %%mm1, %%mm2, %%mm3, %%mm4, %%mm5, OP)\ + VOP(%%mm1, %%mm2, %%mm3, %%mm4, %%mm5, %%mm0, OP)\ + VOP(%%mm2, %%mm3, %%mm4, %%mm5, %%mm0, %%mm1, OP)\ + VOP(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, OP)\ + VOP(%%mm4, %%mm5, %%mm0, %%mm1, %%mm2, %%mm3, OP)\ + VOP(%%mm5, %%mm0, %%mm1, %%mm2, %%mm3, %%mm4, OP)\ + VOP(%%mm0, %%mm1, %%mm2, %%mm3, %%mm4, %%mm5, OP)\ + VOP(%%mm1, %%mm2, %%mm3, %%mm4, %%mm5, %%mm0, OP)\ + \ + : "+a"(src), "+c"(dst)\ + : "S"((long)srcStride), "D"((long)dstStride), "m"(ADD), "m"(MUL1), "m"(MUL2)\ + : "memory"\ + );\ + if(h==16){\ + asm volatile(\ + VOP(%%mm2, %%mm3, %%mm4, %%mm5, %%mm0, %%mm1, OP)\ + VOP(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, OP)\ + VOP(%%mm4, %%mm5, %%mm0, %%mm1, %%mm2, %%mm3, OP)\ + VOP(%%mm5, %%mm0, %%mm1, %%mm2, %%mm3, %%mm4, OP)\ + VOP(%%mm0, %%mm1, %%mm2, %%mm3, %%mm4, %%mm5, OP)\ + VOP(%%mm1, %%mm2, %%mm3, %%mm4, %%mm5, %%mm0, OP)\ + VOP(%%mm2, %%mm3, %%mm4, %%mm5, %%mm0, %%mm1, OP)\ + VOP(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, OP)\ + \ + : "+a"(src), "+c"(dst)\ + : "S"((long)srcStride), "D"((long)dstStride), "m"(ADD), "m"(MUL1), "m"(MUL2)\ + : "memory"\ + );\ + }\ + src += 4-(h+5)*srcStride;\ + dst += 4-h*dstStride;\ + } + +#define QPEL_CAVS(OPNAME, OP, MMX)\ +static void OPNAME ## cavs_qpel8_h_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ + int h=8;\ + asm volatile(\ + "pxor %%mm7, %%mm7 \n\t"\ + "movq %5, %%mm6 \n\t"\ + "1: \n\t"\ + "movq (%0), %%mm0 \n\t"\ + "movq 1(%0), %%mm2 \n\t"\ + "movq %%mm0, %%mm1 \n\t"\ + "movq %%mm2, %%mm3 \n\t"\ + "punpcklbw %%mm7, %%mm0 \n\t"\ + "punpckhbw %%mm7, %%mm1 \n\t"\ + "punpcklbw %%mm7, %%mm2 \n\t"\ + "punpckhbw %%mm7, %%mm3 \n\t"\ + "paddw %%mm2, %%mm0 \n\t"\ + "paddw %%mm3, %%mm1 \n\t"\ + "pmullw %%mm6, %%mm0 \n\t"\ + "pmullw %%mm6, %%mm1 \n\t"\ + "movq -1(%0), %%mm2 \n\t"\ + "movq 2(%0), %%mm4 \n\t"\ + "movq %%mm2, %%mm3 \n\t"\ + "movq %%mm4, %%mm5 \n\t"\ + "punpcklbw %%mm7, %%mm2 \n\t"\ + "punpckhbw %%mm7, %%mm3 \n\t"\ + "punpcklbw %%mm7, %%mm4 \n\t"\ + "punpckhbw %%mm7, %%mm5 \n\t"\ + "paddw %%mm4, %%mm2 \n\t"\ + "paddw %%mm3, %%mm5 \n\t"\ + "psubw %%mm2, %%mm0 \n\t"\ + "psubw %%mm5, %%mm1 \n\t"\ + "movq %6, %%mm5 \n\t"\ + "paddw %%mm5, %%mm0 \n\t"\ + "paddw %%mm5, %%mm1 \n\t"\ + "psraw $3, %%mm0 \n\t"\ + "psraw $3, %%mm1 \n\t"\ + "packuswb %%mm1, %%mm0 \n\t"\ + OP(%%mm0, (%1),%%mm5, q) \ + "add %3, %0 \n\t"\ + "add %4, %1 \n\t"\ + "decl %2 \n\t"\ + " jnz 1b \n\t"\ + : "+a"(src), "+c"(dst), "+m"(h)\ + : "d"((long)srcStride), "S"((long)dstStride), "m"(ff_pw_5), "m"(ff_pw_4)\ + : "memory"\ + );\ +}\ +\ +static inline void OPNAME ## cavs_qpel8or16_v1_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\ + QPEL_CAVSVNUM(QPEL_CAVSV1,OP,ff_pw_64,ff_pw_96,ff_pw_42) \ +}\ +\ +static inline void OPNAME ## cavs_qpel8or16_v2_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\ + QPEL_CAVSVNUM(QPEL_CAVSV2,OP,ff_pw_4,ff_pw_5,ff_pw_5) \ +}\ +\ +static inline void OPNAME ## cavs_qpel8or16_v3_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\ + QPEL_CAVSVNUM(QPEL_CAVSV3,OP,ff_pw_64,ff_pw_96,ff_pw_42) \ +}\ +\ +static void OPNAME ## cavs_qpel8_v1_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ + OPNAME ## cavs_qpel8or16_v1_ ## MMX(dst , src , dstStride, srcStride, 8);\ +}\ +static void OPNAME ## cavs_qpel16_v1_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ + OPNAME ## cavs_qpel8or16_v1_ ## MMX(dst , src , dstStride, srcStride, 16);\ + OPNAME ## cavs_qpel8or16_v1_ ## MMX(dst+8, src+8, dstStride, srcStride, 16);\ +}\ +\ +static void OPNAME ## cavs_qpel8_v2_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ + OPNAME ## cavs_qpel8or16_v2_ ## MMX(dst , src , dstStride, srcStride, 8);\ +}\ +static void OPNAME ## cavs_qpel16_v2_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ + OPNAME ## cavs_qpel8or16_v2_ ## MMX(dst , src , dstStride, srcStride, 16);\ + OPNAME ## cavs_qpel8or16_v2_ ## MMX(dst+8, src+8, dstStride, srcStride, 16);\ +}\ +\ +static void OPNAME ## cavs_qpel8_v3_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ + OPNAME ## cavs_qpel8or16_v3_ ## MMX(dst , src , dstStride, srcStride, 8);\ +}\ +static void OPNAME ## cavs_qpel16_v3_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ + OPNAME ## cavs_qpel8or16_v3_ ## MMX(dst , src , dstStride, srcStride, 16);\ + OPNAME ## cavs_qpel8or16_v3_ ## MMX(dst+8, src+8, dstStride, srcStride, 16);\ +}\ +\ +static void OPNAME ## cavs_qpel16_h_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ + OPNAME ## cavs_qpel8_h_ ## MMX(dst , src , dstStride, srcStride);\ + OPNAME ## cavs_qpel8_h_ ## MMX(dst+8, src+8, dstStride, srcStride);\ + src += 8*srcStride;\ + dst += 8*dstStride;\ + OPNAME ## cavs_qpel8_h_ ## MMX(dst , src , dstStride, srcStride);\ + OPNAME ## cavs_qpel8_h_ ## MMX(dst+8, src+8, dstStride, srcStride);\ +}\ + +#define CAVS_MC(OPNAME, SIZE, MMX) \ +static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc20_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\ + OPNAME ## cavs_qpel ## SIZE ## _h_ ## MMX(dst, src, stride, stride);\ +}\ +\ +static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc01_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\ + OPNAME ## cavs_qpel ## SIZE ## _v1_ ## MMX(dst, src, stride, stride);\ +}\ +\ +static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc02_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\ + OPNAME ## cavs_qpel ## SIZE ## _v2_ ## MMX(dst, src, stride, stride);\ +}\ +\ +static void ff_ ## OPNAME ## cavs_qpel ## SIZE ## _mc03_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\ + OPNAME ## cavs_qpel ## SIZE ## _v3_ ## MMX(dst, src, stride, stride);\ +}\ + +#define PUT_OP(a,b,temp, size) "mov" #size " " #a ", " #b " \n\t" +#define AVG_3DNOW_OP(a,b,temp, size) \ +"mov" #size " " #b ", " #temp " \n\t"\ +"pavgusb " #temp ", " #a " \n\t"\ +"mov" #size " " #a ", " #b " \n\t" +#define AVG_MMX2_OP(a,b,temp, size) \ +"mov" #size " " #b ", " #temp " \n\t"\ +"pavgb " #temp ", " #a " \n\t"\ +"mov" #size " " #a ", " #b " \n\t" + +QPEL_CAVS(put_, PUT_OP, 3dnow) +QPEL_CAVS(avg_, AVG_3DNOW_OP, 3dnow) +QPEL_CAVS(put_, PUT_OP, mmx2) +QPEL_CAVS(avg_, AVG_MMX2_OP, mmx2) + +CAVS_MC(put_, 8, 3dnow) +CAVS_MC(put_, 16,3dnow) +CAVS_MC(avg_, 8, 3dnow) +CAVS_MC(avg_, 16,3dnow) +CAVS_MC(put_, 8, mmx2) +CAVS_MC(put_, 16,mmx2) +CAVS_MC(avg_, 8, mmx2) +CAVS_MC(avg_, 16,mmx2) + +void ff_put_cavs_qpel8_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride); +void ff_avg_cavs_qpel8_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride); +void ff_put_cavs_qpel16_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride); +void ff_avg_cavs_qpel16_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride); + +void ff_cavsdsp_init_mmx2(DSPContext* c, AVCodecContext *avctx) { +#define dspfunc(PFX, IDX, NUM) \ + c->PFX ## _pixels_tab[IDX][ 0] = ff_ ## PFX ## NUM ## _mc00_mmx2; \ + c->PFX ## _pixels_tab[IDX][ 2] = ff_ ## PFX ## NUM ## _mc20_mmx2; \ + c->PFX ## _pixels_tab[IDX][ 4] = ff_ ## PFX ## NUM ## _mc01_mmx2; \ + c->PFX ## _pixels_tab[IDX][ 8] = ff_ ## PFX ## NUM ## _mc02_mmx2; \ + c->PFX ## _pixels_tab[IDX][12] = ff_ ## PFX ## NUM ## _mc03_mmx2; \ + + dspfunc(put_cavs_qpel, 0, 16); + dspfunc(put_cavs_qpel, 1, 8); + dspfunc(avg_cavs_qpel, 0, 16); + dspfunc(avg_cavs_qpel, 1, 8); +#undef dspfunc + c->cavs_idct8_add = cavs_idct8_add_mmx; +} + +void ff_cavsdsp_init_3dnow(DSPContext* c, AVCodecContext *avctx) { +#define dspfunc(PFX, IDX, NUM) \ + c->PFX ## _pixels_tab[IDX][ 0] = ff_ ## PFX ## NUM ## _mc00_mmx2; \ + c->PFX ## _pixels_tab[IDX][ 2] = ff_ ## PFX ## NUM ## _mc20_3dnow; \ + c->PFX ## _pixels_tab[IDX][ 4] = ff_ ## PFX ## NUM ## _mc01_3dnow; \ + c->PFX ## _pixels_tab[IDX][ 8] = ff_ ## PFX ## NUM ## _mc02_3dnow; \ + c->PFX ## _pixels_tab[IDX][12] = ff_ ## PFX ## NUM ## _mc03_3dnow; \ + + dspfunc(put_cavs_qpel, 0, 16); + dspfunc(put_cavs_qpel, 1, 8); + dspfunc(avg_cavs_qpel, 0, 16); + dspfunc(avg_cavs_qpel, 1, 8); +#undef dspfunc + c->cavs_idct8_add = cavs_idct8_add_mmx; +} Modified: trunk/libavcodec/i386/dsputil_mmx.c ============================================================================== --- trunk/libavcodec/i386/dsputil_mmx.c (original) +++ trunk/libavcodec/i386/dsputil_mmx.c Sat Jul 29 10:45:33 2006 @@ -2622,6 +2622,22 @@ #include "h264dsp_mmx.c" +/* AVS specific */ +void ff_cavsdsp_init_mmx2(DSPContext* c, AVCodecContext *avctx); + +void ff_put_cavs_qpel8_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride) { + put_pixels8_mmx(dst, src, stride, 8); +} +void ff_avg_cavs_qpel8_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride) { + avg_pixels8_mmx(dst, src, stride, 8); +} +void ff_put_cavs_qpel16_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride) { + put_pixels16_mmx(dst, src, stride, 16); +} +void ff_avg_cavs_qpel16_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride) { + avg_pixels16_mmx(dst, src, stride, 16); +} + /* external functions, from idct_mmx.c */ void ff_mmx_idct(DCTELEM *block); void ff_mmxext_idct(DCTELEM *block); @@ -2779,6 +2795,8 @@ c->idct_permutation_type= FF_PARTTRANS_IDCT_PERM; } #endif + }else if(idct_algo==FF_IDCT_CAVS){ + c->idct_permutation_type= FF_TRANSPOSE_IDCT_PERM; #ifdef CONFIG_GPL }else if(idct_algo==FF_IDCT_XVIDMMX){ if(mm_flags & MM_MMXEXT){ @@ -3012,6 +3030,10 @@ c->biweight_h264_pixels_tab[6]= ff_h264_biweight_4x4_mmx2; c->biweight_h264_pixels_tab[7]= ff_h264_biweight_4x2_mmx2; +#ifdef CONFIG_CAVS_DECODER + ff_cavsdsp_init_mmx2(c, avctx); +#endif + #ifdef CONFIG_ENCODERS c->sub_hfyu_median_prediction= sub_hfyu_median_prediction_mmx2; #endif //CONFIG_ENCODERS

1 0

r5845 - trunk/libavcodec/vc1.c
by subversion＠mplayerhq.hu 29 Jul '06

29 Jul '06

Author: kostya Date: Sat Jul 29 05:11:14 2006 New Revision: 5845 Modified: trunk/libavcodec/vc1.c Log: Set MpegEncContext->mspel flag (here it indicates that bicubic MC will be use) Modified: trunk/libavcodec/vc1.c ============================================================================== --- trunk/libavcodec/vc1.c (original) +++ trunk/libavcodec/vc1.c Sat Jul 29 05:11:14 2006 @@ -1474,6 +1474,7 @@ v->s.quarter_sample = 1; } else v->s.quarter_sample = 1; + v->s.mspel = !(v->mv_mode == MV_PMODE_1MV_HPEL_BILIN || (v->mv_mode == MV_PMODE_INTENSITY_COMP && v->mv_mode2 == MV_PMODE_1MV_HPEL_BILIN)); if ((v->mv_mode == MV_PMODE_INTENSITY_COMP && v->mv_mode2 == MV_PMODE_MIXED_MV) @@ -1523,6 +1524,7 @@ lowquant = (v->pq > 12) ? 0 : 1; v->mv_mode = get_bits1(gb) ? MV_PMODE_1MV : MV_PMODE_1MV_HPEL_BILIN; v->s.quarter_sample = (v->mv_mode == MV_PMODE_1MV); + v->s.mspel = v->s.quarter_sample; status = bitplane_decoding(v->direct_mb_plane, &v->dmb_is_raw, v); if (status < 0) return -1;

1 0

r5844 - trunk/libavcodec/vc1.c
by subversion＠mplayerhq.hu 29 Jul '06

29 Jul '06

Author: kostya Date: Sat Jul 29 05:07:30 2006 New Revision: 5844 Modified: trunk/libavcodec/vc1.c Log: Support range reduced frames Modified: trunk/libavcodec/vc1.c ============================================================================== --- trunk/libavcodec/vc1.c (original) +++ trunk/libavcodec/vc1.c Sat Jul 29 05:07:30 2006 @@ -880,6 +880,14 @@ int ys, us, vs; DSPContext *dsp = &v->s.dsp; + if(v->rangeredfrm) { + int i, j, k; + for(k = 0; k < 6; k++) + for(j = 0; j < 8; j++) + for(i = 0; i < 8; i++) + block[k][i + j*8] = ((block[k][i + j*8] - 128) << 1) + 128; + + } ys = v->s.current_picture.linesize[0]; us = v->s.current_picture.linesize[1]; vs = v->s.current_picture.linesize[2]; @@ -943,7 +951,7 @@ srcV = s->edge_emu_buffer + 18 * s->linesize; } - if((v->mv_mode == MV_PMODE_INTENSITY_COMP) + if(v->rangeredfrm || (v->mv_mode == MV_PMODE_INTENSITY_COMP) || (unsigned)src_x > s->h_edge_pos - (mx&3) - 16 || (unsigned)src_y > s->v_edge_pos - (my&3) - 16){ uint8_t *uvbuf= s->edge_emu_buffer + 18 * s->linesize; @@ -957,6 +965,26 @@ uvsrc_x, uvsrc_y, s->h_edge_pos >> 1, s->v_edge_pos >> 1); srcU = uvbuf; srcV = uvbuf + 16; + /* if we deal with range reduction we need to scale source blocks */ + if(v->rangeredfrm) { + int i, j; + uint8_t *src, *src2; + + src = srcY; + for(j = 0; j < 17; j++) { + for(i = 0; i < 17; i++) src[i] = ((src[i] - 128) >> 1) + 128; + src += s->linesize; + } + src = srcU; src2 = srcV; + for(j = 0; j < 9; j++) { + for(i = 0; i < 9; i++) { + src[i] = ((src[i] - 128) >> 1) + 128; + src2[i] = ((src2[i] - 128) >> 1) + 128; + } + src += s->uvlinesize; + src2 += s->uvlinesize; + } + } /* if we deal with intensity compensation we need to scale source blocks */ if(v->mv_mode == MV_PMODE_INTENSITY_COMP) { int i, j; @@ -1039,11 +1067,22 @@ srcY += src_y * s->linesize + src_x; - if((unsigned)src_x > s->h_edge_pos - (mx&3) - 16 + if(v->rangeredfrm || (unsigned)src_x > s->h_edge_pos - (mx&3) - 16 || (unsigned)src_y > s->v_edge_pos - (my&3) - 16){ ff_emulated_edge_mc(s->edge_emu_buffer, srcY, s->linesize, 16+1, 16+1, src_x, src_y, s->h_edge_pos, s->v_edge_pos); srcY = s->edge_emu_buffer; + /* if we deal with range reduction we need to scale source blocks */ + if(v->rangeredfrm) { + int i, j; + uint8_t *src; + + src = srcY; + for(j = 0; j < 17; j++) { + for(i = 0; i < 17; i++) src[i] = ((src[i] - 128) >> 1) + 128; + src += s->linesize; + } + } } if(!s->quarter_sample) { // hpel mc @@ -1141,7 +1180,7 @@ uvsrc_y = clip(uvsrc_y, -8, s->mb_height * 8); srcU = s->last_picture.data[1] + uvsrc_y * s->uvlinesize + uvsrc_x; srcV = s->last_picture.data[2] + uvsrc_y * s->uvlinesize + uvsrc_x; - if((unsigned)uvsrc_x > (s->h_edge_pos >> 1) - 9 + if(v->rangeredfrm || (unsigned)uvsrc_x > (s->h_edge_pos >> 1) - 9 || (unsigned)uvsrc_y > (s->v_edge_pos >> 1) - 9){ ff_emulated_edge_mc(s->edge_emu_buffer , srcU, s->uvlinesize, 8+1, 8+1, uvsrc_x, uvsrc_y, s->h_edge_pos >> 1, s->v_edge_pos >> 1); @@ -1149,6 +1188,22 @@ uvsrc_x, uvsrc_y, s->h_edge_pos >> 1, s->v_edge_pos >> 1); srcU = s->edge_emu_buffer; srcV = s->edge_emu_buffer + 16; + + /* if we deal with range reduction we need to scale source blocks */ + if(v->rangeredfrm) { + int i, j; + uint8_t *src, *src2; + + src = srcU; src2 = srcV; + for(j = 0; j < 9; j++) { + for(i = 0; i < 9; i++) { + src[i] = ((src[i] - 128) >> 1) + 128; + src2[i] = ((src2[i] - 128) >> 1) + 128; + } + src += s->uvlinesize; + src2 += s->uvlinesize; + } + } } if(v->fastuvmc) { @@ -2578,6 +2633,7 @@ vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i&4)?v->codingset2:v->codingset); if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue; vc1_inv_trans(s->block[i], 8, 8); + if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1; for(j = 0; j < 64; j++) s->block[i][j] += 128; s->dsp.put_pixels_clamped(s->block[i], s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); /* TODO: proper loop filtering */ @@ -2681,6 +2737,7 @@ vc1_decode_intra_block(v, s->block[i], i, is_coded[i], mquant, (i&4)?v->codingset2:v->codingset); if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue; vc1_inv_trans(s->block[i], 8, 8); + if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1; for(j = 0; j < 64; j++) s->block[i][j] += 128; s->dsp.put_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); /* TODO: proper loop filtering */ @@ -2846,6 +2903,7 @@ vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i&4)?v->codingset2:v->codingset); if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue; vc1_inv_trans(s->block[i], 8, 8); + if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1; for(j = 0; j < 64; j++) s->block[i][j] += 128; s->dsp.put_pixels_clamped(s->block[i], s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); /* TODO: proper loop filtering */

1 0

r5843 - trunk/libavcodec/vc1.c
by subversion＠mplayerhq.hu 29 Jul '06

29 Jul '06

Author: kostya Date: Sat Jul 29 05:05:29 2006 New Revision: 5843 Modified: trunk/libavcodec/vc1.c Log: Support grayscale decoding. Modified: trunk/libavcodec/vc1.c ============================================================================== --- trunk/libavcodec/vc1.c (original) +++ trunk/libavcodec/vc1.c Sat Jul 29 05:05:29 2006 @@ -891,8 +891,10 @@ dsp->put_pixels_clamped(block[2], Y, ys); dsp->put_pixels_clamped(block[3], Y + 8, ys); - dsp->put_pixels_clamped(block[4], v->s.dest[1], us); - dsp->put_pixels_clamped(block[5], v->s.dest[2], vs); + if(!(v->s.flags & CODEC_FLAG_GRAY)) { + dsp->put_pixels_clamped(block[4], v->s.dest[1], us); + dsp->put_pixels_clamped(block[5], v->s.dest[2], vs); + } } /** Do motion compensation over 1 macroblock @@ -935,6 +937,12 @@ srcU += uvsrc_y * s->uvlinesize + uvsrc_x; srcV += uvsrc_y * s->uvlinesize + uvsrc_x; + /* for grayscale we should not try to read from unknown area */ + if(s->flags & CODEC_FLAG_GRAY) { + srcU = s->edge_emu_buffer + 18 * s->linesize; + srcV = s->edge_emu_buffer + 18 * s->linesize; + } + if((v->mv_mode == MV_PMODE_INTENSITY_COMP) || (unsigned)src_x > s->h_edge_pos - (mx&3) - 16 || (unsigned)src_y > s->v_edge_pos - (my&3) - 16){ @@ -993,6 +1001,8 @@ else dsp->put_no_rnd_qpel_pixels_tab[0][dxy](s->dest[0], srcY, s->linesize); } + + if(s->flags & CODEC_FLAG_GRAY) return; /* Chroma MC always uses qpel blilinear */ uvdxy = ((uvmy & 3) << 2) | (uvmx & 3); if(!v->rnd){ @@ -1080,6 +1090,7 @@ static const int count[16] = { 0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4}; if(!v->s.last_picture.data[0])return; + if(s->flags & CODEC_FLAG_GRAY) return; for(i = 0; i < 4; i++) { mvx[i] = s->mv[0][i][0]; @@ -2565,6 +2576,7 @@ v->c_avail = v->mb_type[0][s->block_index[i] - 1]; vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i&4)?v->codingset2:v->codingset); + if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue; vc1_inv_trans(s->block[i], 8, 8); for(j = 0; j < 64; j++) s->block[i][j] += 128; s->dsp.put_pixels_clamped(s->block[i], s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); @@ -2579,7 +2591,8 @@ vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block); if(!v->ttmbf && ttmb < 8) ttmb = -1; first_block = 0; - s->dsp.add_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); + if((i<4) || !(s->flags & CODEC_FLAG_GRAY)) + s->dsp.add_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); } } } @@ -2666,6 +2679,7 @@ v->c_avail = v->mb_type[0][s->block_index[i] - 1]; vc1_decode_intra_block(v, s->block[i], i, is_coded[i], mquant, (i&4)?v->codingset2:v->codingset); + if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue; vc1_inv_trans(s->block[i], 8, 8); for(j = 0; j < 64; j++) s->block[i][j] += 128; s->dsp.put_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); @@ -2680,7 +2694,8 @@ status = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block); if(!v->ttmbf && ttmb < 8) ttmb = -1; first_block = 0; - s->dsp.add_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); + if((i<4) || !(s->flags & CODEC_FLAG_GRAY)) + s->dsp.add_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); } } return status; @@ -2829,6 +2844,7 @@ v->c_avail = v->mb_type[0][s->block_index[i] - 1]; vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i&4)?v->codingset2:v->codingset); + if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue; vc1_inv_trans(s->block[i], 8, 8); for(j = 0; j < 64; j++) s->block[i][j] += 128; s->dsp.put_pixels_clamped(s->block[i], s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); @@ -2843,7 +2859,8 @@ vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block); if(!v->ttmbf && ttmb < 8) ttmb = -1; first_block = 0; - s->dsp.add_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); + if((i<4) || !(s->flags & CODEC_FLAG_GRAY)) + s->dsp.add_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); } } } @@ -2928,16 +2945,20 @@ if(!s->first_slice_line) { vc1_v_overlap(s->dest[0], s->linesize, 0); vc1_v_overlap(s->dest[0] + 8, s->linesize, 0); - vc1_v_overlap(s->dest[1], s->uvlinesize, s->mb_y&1); - vc1_v_overlap(s->dest[2], s->uvlinesize, s->mb_y&1); + if(!(s->flags & CODEC_FLAG_GRAY)) { + vc1_v_overlap(s->dest[1], s->uvlinesize, s->mb_y&1); + vc1_v_overlap(s->dest[2], s->uvlinesize, s->mb_y&1); + } } vc1_v_overlap(s->dest[0] + 8 * s->linesize, s->linesize, 1); vc1_v_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize, 1); if(s->mb_x) { vc1_h_overlap(s->dest[0], s->linesize, 0); vc1_h_overlap(s->dest[0] + 8 * s->linesize, s->linesize, 0); - vc1_h_overlap(s->dest[1], s->uvlinesize, s->mb_x&1); - vc1_h_overlap(s->dest[2], s->uvlinesize, s->mb_x&1); + if(!(s->flags & CODEC_FLAG_GRAY)) { + vc1_h_overlap(s->dest[1], s->uvlinesize, s->mb_x&1); + vc1_h_overlap(s->dest[2], s->uvlinesize, s->mb_x&1); + } } vc1_h_overlap(s->dest[0] + 8, s->linesize, 1); vc1_h_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize, 1); @@ -3079,7 +3100,10 @@ GetBitContext gb; if (!avctx->extradata_size || !avctx->extradata) return -1; - avctx->pix_fmt = PIX_FMT_YUV420P; + if (!(avctx->flags & CODEC_FLAG_GRAY)) + avctx->pix_fmt = PIX_FMT_YUV420P; + else + avctx->pix_fmt = PIX_FMT_GRAY8; v->s.avctx = avctx; avctx->flags |= CODEC_FLAG_EMU_EDGE; v->s.flags |= CODEC_FLAG_EMU_EDGE;

1 0

r5842 - trunk/libavformat/mxf.c
by subversion＠mplayerhq.hu 28 Jul '06

28 Jul '06

Author: bcoudurier Date: Sat Jul 29 01:10:03 2006 New Revision: 5842 Modified: trunk/libavformat/mxf.c Log: cosmetics Modified: trunk/libavformat/mxf.c ============================================================================== --- trunk/libavformat/mxf.c (original) +++ trunk/libavformat/mxf.c Sat Jul 29 01:10:03 2006 @@ -144,23 +144,23 @@ enum CodecID id; } MXFCodecUL; -static const UID mxf_metadata_preface_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x2F, 0x00 }; -static const UID mxf_metadata_content_storage_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x18, 0x00 }; -static const UID mxf_metadata_source_package_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x37, 0x00 }; -static const UID mxf_metadata_material_package_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x36, 0x00 }; -static const UID mxf_metadata_sequence_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x0F, 0x00 }; -static const UID mxf_metadata_source_clip_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x11, 0x00 }; -static const UID mxf_metadata_multiple_descriptor_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x44, 0x00 }; -static const UID mxf_metadata_generic_sound_descriptor_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x42, 0x00 }; -static const UID mxf_metadata_cdci_descriptor_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x28, 0x00 }; -static const UID mxf_metadata_mpegvideo_descriptor_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x51, 0x00 }; -static const UID mxf_metadata_wave_descriptor_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x48, 0x00 }; -static const UID mxf_metadata_static_track_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x3A, 0x00 }; -static const UID mxf_metadata_track_key = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x53, 0x01, 0x01, 0x0d, 0x01, 0x01, 0x01, 0x01, 0x01, 0x3b, 0x00 }; +static const UID mxf_metadata_preface_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x2F,0x00 }; +static const UID mxf_metadata_content_storage_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x18,0x00 }; +static const UID mxf_metadata_source_package_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x37,0x00 }; +static const UID mxf_metadata_material_package_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x36,0x00 }; +static const UID mxf_metadata_sequence_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x0F,0x00 }; +static const UID mxf_metadata_source_clip_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x11,0x00 }; +static const UID mxf_metadata_multiple_descriptor_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x44,0x00 }; +static const UID mxf_metadata_generic_sound_descriptor_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x42,0x00 }; +static const UID mxf_metadata_cdci_descriptor_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x28,0x00 }; +static const UID mxf_metadata_mpegvideo_descriptor_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x51,0x00 }; +static const UID mxf_metadata_wave_descriptor_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x48,0x00 }; +static const UID mxf_metadata_static_track_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x3A,0x00 }; +static const UID mxf_metadata_track_key = { 0x06,0x0e,0x2b,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x3b,0x00 }; /* partial keys to match */ -static const uint8_t mxf_header_partition_pack_key[] = { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x05, 0x01, 0x01, 0x0d, 0x01, 0x02, 0x01, 0x01, 0x02 }; -static const uint8_t mxf_essence_element_key[] = { 0x06, 0x0e, 0x2b, 0x34, 0x01, 0x02, 0x01 }; +static const uint8_t mxf_header_partition_pack_key[] = { 0x06,0x0e,0x2b,0x34,0x02,0x05,0x01,0x01,0x0d,0x01,0x02,0x01,0x01,0x02 }; +static const uint8_t mxf_essence_element_key[] = { 0x06,0x0e,0x2b,0x34,0x01,0x02,0x01 }; #define IS_KLV_KEY(x, y) (!memcmp(x, y, sizeof(y)))

1 0

r5841 - trunk/libavformat/mxf.c
by subversion＠mplayerhq.hu 28 Jul '06

28 Jul '06

Author: bcoudurier Date: Sat Jul 29 01:07:03 2006 New Revision: 5841 Modified: trunk/libavformat/mxf.c Log: use ternary Modified: trunk/libavformat/mxf.c ============================================================================== --- trunk/libavformat/mxf.c (original) +++ trunk/libavformat/mxf.c Sat Jul 29 01:07:03 2006 @@ -197,10 +197,7 @@ klv->offset = url_ftell(pb); get_buffer(pb, klv->key, 16); klv->length = klv_decode_ber_length(pb); - if (klv->length == -1) - return -1; - else - return 0; + return klv->length == -1 ? -1 : 0; } static int mxf_get_stream_index(AVFormatContext *s, KLVPacket *klv) @@ -227,9 +224,7 @@ if (IS_KLV_KEY(klv.key, mxf_essence_element_key)) { av_get_packet(&s->pb, pkt, klv.length); pkt->stream_index = mxf_get_stream_index(s, &klv); - if (pkt->stream_index == -1) - return -1; - return 0; + return pkt->stream_index == -1 ? -1 : 0; } else url_fskip(&s->pb, klv.length); }

1 0

r5840 - trunk/libavformat/mxf.c
by subversion＠mplayerhq.hu 28 Jul '06

28 Jul '06

Author: bcoudurier Date: Sat Jul 29 01:03:00 2006 New Revision: 5840 Modified: trunk/libavformat/mxf.c Log: remove useless variables Modified: trunk/libavformat/mxf.c ============================================================================== --- trunk/libavformat/mxf.c (original) +++ trunk/libavformat/mxf.c Sat Jul 29 01:03:00 2006 @@ -91,7 +91,6 @@ UID uid; UID essence_container_ul; UID essence_codec_ul; - enum CodecType codec_type; AVRational sample_rate; AVRational aspect_ratio; int width; @@ -101,7 +100,6 @@ struct MXFDescriptor **sub_descriptors; UID *sub_descriptors_refs; int sub_descriptors_count; - int block_align; int linked_track_id; } MXFDescriptor; @@ -616,7 +614,6 @@ descriptor->linked_track_id = get_be32(pb); break; case 0x3201: /* PictureEssenceCoding */ - descriptor->codec_type = CODEC_TYPE_VIDEO; get_buffer(pb, descriptor->essence_codec_ul, 16); break; case 0x3203: @@ -629,15 +626,11 @@ descriptor->aspect_ratio.num = get_be32(pb); descriptor->aspect_ratio.den = get_be32(pb); break; - case 0x3D0A: - descriptor->block_align = get_be16(pb); - break; case 0x3D03: descriptor->sample_rate.num = get_be32(pb); descriptor->sample_rate.den = get_be32(pb); break; case 0x3D06: /* SoundEssenceCompression */ - descriptor->codec_type = CODEC_TYPE_AUDIO; get_buffer(pb, descriptor->essence_codec_ul, 16); break; case 0x3D07:

1 0