libavformat/rtpdec.c

   1 /*
   2  * RTP input format
   3  * Copyright (c) 2002 Fabrice Bellard
   4  *
   5  * This file is part of FFmpeg.
   6  *
   7  * FFmpeg is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2.1 of the License, or (at your option) any later version.
  11  *
  12  * FFmpeg is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with FFmpeg; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21
  22 /* needed for gethostname() */
  23 #define _XOPEN_SOURCE 600
  24
  25 #include "libavcodec/bitstream.h"
  26 #include "avformat.h"
  27 #include "mpegts.h"
  28
  29 #include <unistd.h>
  30 #include "network.h"
  31
  32 #include "rtpdec.h"
  33 #include "rtp_h264.h"
  34
  35 //#define DEBUG
  36
  37 /* TODO: - add RTCP statistics reporting (should be optional).
  38
  39          - add support for h263/mpeg4 packetized output : IDEA: send a
  40          buffer to 'rtp_write_packet' contains all the packets for ONE
  41          frame. Each packet should have a four byte header containing
  42          the length in big endian format (same trick as
  43          'url_open_dyn_packet_buf')
  44 */
  45
  46 /* statistics functions */
  47 RTPDynamicProtocolHandler *RTPFirstDynamicPayloadHandler= NULL;
  48
  49 static RTPDynamicProtocolHandler mp4v_es_handler= {"MP4V-ES", CODEC_TYPE_VIDEO, CODEC_ID_MPEG4};
  50 static RTPDynamicProtocolHandler mpeg4_generic_handler= {"mpeg4-generic", CODEC_TYPE_AUDIO, CODEC_ID_AAC};
  51
  52 void ff_register_dynamic_payload_handler(RTPDynamicProtocolHandler *handler)
  53 {
  54     handler->next= RTPFirstDynamicPayloadHandler;
  55     RTPFirstDynamicPayloadHandler= handler;
  56 }
  57
  58 void av_register_rtp_dynamic_payload_handlers(void)
  59 {
  60     ff_register_dynamic_payload_handler(&mp4v_es_handler);
  61     ff_register_dynamic_payload_handler(&mpeg4_generic_handler);
  62     ff_register_dynamic_payload_handler(&ff_h264_dynamic_handler);
  63 }
  64
  65 static int rtcp_parse_packet(RTPDemuxContext *s, const unsigned char *buf, int len)
  66 {
  67     if (buf[1] != 200)
  68         return -1;
  69     s->last_rtcp_ntp_time = AV_RB64(buf + 8);
  70     if (s->first_rtcp_ntp_time == AV_NOPTS_VALUE)
  71         s->first_rtcp_ntp_time = s->last_rtcp_ntp_time;
  72     s->last_rtcp_timestamp = AV_RB32(buf + 16);
  73     return 0;
  74 }
  75
  76 #define RTP_SEQ_MOD (1<<16)
  77
  78 /**
  79 * called on parse open packet
  80 */
  81 static void rtp_init_statistics(RTPStatistics *s, uint16_t base_sequence) // called on parse open packet.
  82 {
  83     memset(s, 0, sizeof(RTPStatistics));
  84     s->max_seq= base_sequence;
  85     s->probation= 1;
  86 }
  87
  88 /**
  89 * called whenever there is a large jump in sequence numbers, or when they get out of probation...
  90 */
  91 static void rtp_init_sequence(RTPStatistics *s, uint16_t seq)
  92 {
  93     s->max_seq= seq;
  94     s->cycles= 0;
  95     s->base_seq= seq -1;
  96     s->bad_seq= RTP_SEQ_MOD + 1;
  97     s->received= 0;
  98     s->expected_prior= 0;
  99     s->received_prior= 0;
 100     s->jitter= 0;
 101     s->transit= 0;
 102 }
 103
 104 /**
 105 * returns 1 if we should handle this packet.
 106 */
 107 static int rtp_valid_packet_in_sequence(RTPStatistics *s, uint16_t seq)
 108 {
 109     uint16_t udelta= seq - s->max_seq;
 110     const int MAX_DROPOUT= 3000;
 111     const int MAX_MISORDER = 100;
 112     const int MIN_SEQUENTIAL = 2;
 113
 114     /* source not valid until MIN_SEQUENTIAL packets with sequence seq. numbers have been received */
 115     if(s->probation)
 116     {
 117         if(seq==s->max_seq + 1) {
 118             s->probation--;
 119             s->max_seq= seq;
 120             if(s->probation==0) {
 121                 rtp_init_sequence(s, seq);
 122                 s->received++;
 123                 return 1;
 124             }
 125         } else {
 126             s->probation= MIN_SEQUENTIAL - 1;
 127             s->max_seq = seq;
 128         }
 129     } else if (udelta < MAX_DROPOUT) {
 130         // in order, with permissible gap
 131         if(seq < s->max_seq) {
 132             //sequence number wrapped; count antother 64k cycles
 133             s->cycles += RTP_SEQ_MOD;
 134         }
 135         s->max_seq= seq;
 136     } else if (udelta <= RTP_SEQ_MOD - MAX_MISORDER) {
 137         // sequence made a large jump...
 138         if(seq==s->bad_seq) {
 139             // two sequential packets-- assume that the other side restarted without telling us; just resync.
 140             rtp_init_sequence(s, seq);
 141         } else {
 142             s->bad_seq= (seq + 1) & (RTP_SEQ_MOD-1);
 143             return 0;
 144         }
 145     } else {
 146         // duplicate or reordered packet...
 147     }
 148     s->received++;
 149     return 1;
 150 }
 151
 152 #if 0
 153 /**
 154 * This function is currently unused; without a valid local ntp time, I don't see how we could calculate the
 155 * difference between the arrival and sent timestamp.  As a result, the jitter and transit statistics values
 156 * never change.  I left this in in case someone else can see a way. (rdm)
 157 */
 158 static void rtcp_update_jitter(RTPStatistics *s, uint32_t sent_timestamp, uint32_t arrival_timestamp)
 159 {
 160     uint32_t transit= arrival_timestamp - sent_timestamp;
 161     int d;
 162     s->transit= transit;
 163     d= FFABS(transit - s->transit);
 164     s->jitter += d - ((s->jitter + 8)>>4);
 165 }
 166 #endif
 167
 168 int rtp_check_and_send_back_rr(RTPDemuxContext *s, int count)
 169 {
 170     ByteIOContext *pb;
 171     uint8_t *buf;
 172     int len;
 173     int rtcp_bytes;
 174     RTPStatistics *stats= &s->statistics;
 175     uint32_t lost;
 176     uint32_t extended_max;
 177     uint32_t expected_interval;
 178     uint32_t received_interval;
 179     uint32_t lost_interval;
 180     uint32_t expected;
 181     uint32_t fraction;
 182     uint64_t ntp_time= s->last_rtcp_ntp_time; // TODO: Get local ntp time?
 183
 184     if (!s->rtp_ctx || (count < 1))
 185         return -1;
 186
 187     /* TODO: I think this is way too often; RFC 1889 has algorithm for this */
 188     /* XXX: mpeg pts hardcoded. RTCP send every 0.5 seconds */
 189     s->octet_count += count;
 190     rtcp_bytes = ((s->octet_count - s->last_octet_count) * RTCP_TX_RATIO_NUM) /
 191         RTCP_TX_RATIO_DEN;
 192     rtcp_bytes /= 50; // mmu_man: that's enough for me... VLC sends much less btw !?
 193     if (rtcp_bytes < 28)
 194         return -1;
 195     s->last_octet_count = s->octet_count;
 196
 197     if (url_open_dyn_buf(&pb) < 0)
 198         return -1;
 199
 200     // Receiver Report
 201     put_byte(pb, (RTP_VERSION << 6) + 1); /* 1 report block */
 202     put_byte(pb, 201);
 203     put_be16(pb, 7); /* length in words - 1 */
 204     put_be32(pb, s->ssrc); // our own SSRC
 205     put_be32(pb, s->ssrc); // XXX: should be the server's here!
 206     // some placeholders we should really fill...
 207     // RFC 1889/p64
 208     extended_max= stats->cycles + stats->max_seq;
 209     expected= extended_max - stats->base_seq + 1;
 210     lost= expected - stats->received;
 211     lost= FFMIN(lost, 0xffffff); // clamp it since it's only 24 bits...
 212     expected_interval= expected - stats->expected_prior;
 213     stats->expected_prior= expected;
 214     received_interval= stats->received - stats->received_prior;
 215     stats->received_prior= stats->received;
 216     lost_interval= expected_interval - received_interval;
 217     if (expected_interval==0 || lost_interval<=0) fraction= 0;
 218     else fraction = (lost_interval<<8)/expected_interval;
 219
 220     fraction= (fraction<<24) | lost;
 221
 222     put_be32(pb, fraction); /* 8 bits of fraction, 24 bits of total packets lost */
 223     put_be32(pb, extended_max); /* max sequence received */
 224     put_be32(pb, stats->jitter>>4); /* jitter */
 225
 226     if(s->last_rtcp_ntp_time==AV_NOPTS_VALUE)
 227     {
 228         put_be32(pb, 0); /* last SR timestamp */
 229         put_be32(pb, 0); /* delay since last SR */
 230     } else {
 231         uint32_t middle_32_bits= s->last_rtcp_ntp_time>>16; // this is valid, right? do we need to handle 64 bit values special?
 232         uint32_t delay_since_last= ntp_time - s->last_rtcp_ntp_time;
 233
 234         put_be32(pb, middle_32_bits); /* last SR timestamp */
 235         put_be32(pb, delay_since_last); /* delay since last SR */
 236     }
 237
 238     // CNAME
 239     put_byte(pb, (RTP_VERSION << 6) + 1); /* 1 report block */
 240     put_byte(pb, 202);
 241     len = strlen(s->hostname);
 242     put_be16(pb, (6 + len + 3) / 4); /* length in words - 1 */
 243     put_be32(pb, s->ssrc);
 244     put_byte(pb, 0x01);
 245     put_byte(pb, len);
 246     put_buffer(pb, s->hostname, len);
 247     // padding
 248     for (len = (6 + len) % 4; len % 4; len++) {
 249         put_byte(pb, 0);
 250     }
 251
 252     put_flush_packet(pb);
 253     len = url_close_dyn_buf(pb, &buf);
 254     if ((len > 0) && buf) {
 255         int result;
 256         dprintf(s->ic, "sending %d bytes of RR\n", len);
 257         result= url_write(s->rtp_ctx, buf, len);
 258         dprintf(s->ic, "result from url_write: %d\n", result);
 259         av_free(buf);
 260     }
 261     return 0;
 262 }
 263
 264 /**
 265  * open a new RTP parse context for stream 'st'. 'st' can be NULL for
 266  * MPEG2TS streams to indicate that they should be demuxed inside the
 267  * rtp demux (otherwise CODEC_ID_MPEG2TS packets are returned)
 268  * TODO: change this to not take rtp_payload data, and use the new dynamic payload system.
 269  */
 270 RTPDemuxContext *rtp_parse_open(AVFormatContext *s1, AVStream *st, URLContext *rtpc, int payload_type, RTPPayloadData *rtp_payload_data)
 271 {
 272     RTPDemuxContext *s;
 273
 274     s = av_mallocz(sizeof(RTPDemuxContext));
 275     if (!s)
 276         return NULL;
 277     s->payload_type = payload_type;
 278     s->last_rtcp_ntp_time = AV_NOPTS_VALUE;
 279     s->first_rtcp_ntp_time = AV_NOPTS_VALUE;
 280     s->ic = s1;
 281     s->st = st;
 282     s->rtp_payload_data = rtp_payload_data;
 283     rtp_init_statistics(&s->statistics, 0); // do we know the initial sequence from sdp?
 284     if (!strcmp(ff_rtp_enc_name(payload_type), "MP2T")) {
 285         s->ts = mpegts_parse_open(s->ic);
 286         if (s->ts == NULL) {
 287             av_free(s);
 288             return NULL;
 289         }
 290     } else {
 291         av_set_pts_info(st, 32, 1, 90000);
 292         switch(st->codec->codec_id) {
 293         case CODEC_ID_MPEG1VIDEO:
 294         case CODEC_ID_MPEG2VIDEO:
 295         case CODEC_ID_MP2:
 296         case CODEC_ID_MP3:
 297         case CODEC_ID_MPEG4:
 298         case CODEC_ID_H264:
 299             st->need_parsing = AVSTREAM_PARSE_FULL;
 300             break;
 301         default:
 302             if (st->codec->codec_type == CODEC_TYPE_AUDIO) {
 303                 av_set_pts_info(st, 32, 1, st->codec->sample_rate);
 304             }
 305             break;
 306         }
 307     }
 308     // needed to send back RTCP RR in RTSP sessions
 309     s->rtp_ctx = rtpc;
 310     gethostname(s->hostname, sizeof(s->hostname));
 311     return s;
 312 }
 313
 314 void
 315 rtp_parse_set_dynamic_protocol(RTPDemuxContext *s, PayloadContext *ctx,
 316                                RTPDynamicProtocolHandler *handler)
 317 {
 318     s->dynamic_protocol_context = ctx;
 319     s->parse_packet = handler->parse_packet;
 320 }
 321
 322 static int rtp_parse_mp4_au(RTPDemuxContext *s, const uint8_t *buf)
 323 {
 324     int au_headers_length, au_header_size, i;
 325     GetBitContext getbitcontext;
 326     RTPPayloadData *infos;
 327
 328     infos = s->rtp_payload_data;
 329
 330     if (infos == NULL)
 331         return -1;
 332
 333     /* decode the first 2 bytes where the AUHeader sections are stored
 334        length in bits */
 335     au_headers_length = AV_RB16(buf);
 336
 337     if (au_headers_length > RTP_MAX_PACKET_LENGTH)
 338       return -1;
 339
 340     infos->au_headers_length_bytes = (au_headers_length + 7) / 8;
 341
 342     /* skip AU headers length section (2 bytes) */
 343     buf += 2;
 344
 345     init_get_bits(&getbitcontext, buf, infos->au_headers_length_bytes * 8);
 346
 347     /* XXX: Wrong if optionnal additional sections are present (cts, dts etc...) */
 348     au_header_size = infos->sizelength + infos->indexlength;
 349     if (au_header_size <= 0 || (au_headers_length % au_header_size != 0))
 350         return -1;
 351
 352     infos->nb_au_headers = au_headers_length / au_header_size;
 353     infos->au_headers = av_malloc(sizeof(struct AUHeaders) * infos->nb_au_headers);
 354
 355     /* XXX: We handle multiple AU Section as only one (need to fix this for interleaving)
 356        In my test, the FAAD decoder does not behave correctly when sending each AU one by one
 357        but does when sending the whole as one big packet...  */
 358     infos->au_headers[0].size = 0;
 359     infos->au_headers[0].index = 0;
 360     for (i = 0; i < infos->nb_au_headers; ++i) {
 361         infos->au_headers[0].size += get_bits_long(&getbitcontext, infos->sizelength);
 362         infos->au_headers[0].index = get_bits_long(&getbitcontext, infos->indexlength);
 363     }
 364
 365     infos->nb_au_headers = 1;
 366
 367     return 0;
 368 }
 369
 370 /**
 371  * This was the second switch in rtp_parse packet.  Normalizes time, if required, sets stream_index, etc.
 372  */
 373 static void finalize_packet(RTPDemuxContext *s, AVPacket *pkt, uint32_t timestamp)
 374 {
 375     if (s->last_rtcp_ntp_time != AV_NOPTS_VALUE) {
 376         int64_t addend;
 377         int delta_timestamp;
 378
 379         /* compute pts from timestamp with received ntp_time */
 380         delta_timestamp = timestamp - s->last_rtcp_timestamp;
 381         /* convert to the PTS timebase */
 382         addend = av_rescale(s->last_rtcp_ntp_time - s->first_rtcp_ntp_time, s->st->time_base.den, (uint64_t)s->st->time_base.num << 32);
 383         pkt->pts = addend + delta_timestamp;
 384     }
 385     pkt->stream_index = s->st->index;
 386 }
 387
 388 /**
 389  * Parse an RTP or RTCP packet directly sent as a buffer.
 390  * @param s RTP parse context.
 391  * @param pkt returned packet
 392  * @param buf input buffer or NULL to read the next packets
 393  * @param len buffer len
 394  * @return 0 if a packet is returned, 1 if a packet is returned and more can follow
 395  * (use buf as NULL to read the next). -1 if no packet (error or no more packet).
 396  */
 397 int rtp_parse_packet(RTPDemuxContext *s, AVPacket *pkt,
 398                      const uint8_t *buf, int len)
 399 {
 400     unsigned int ssrc, h;
 401     int payload_type, seq, ret, flags = 0;
 402     AVStream *st;
 403     uint32_t timestamp;
 404     int rv= 0;
 405
 406     if (!buf) {
 407         /* return the next packets, if any */
 408         if(s->st && s->parse_packet) {
 409             timestamp= 0; ///< Should not be used if buf is NULL, but should be set to the timestamp of the packet returned....
 410             rv= s->parse_packet(s->ic, s->dynamic_protocol_context,
 411                                 s->st, pkt, &timestamp, NULL, 0, flags);
 412             finalize_packet(s, pkt, timestamp);
 413             return rv;
 414         } else {
 415             // TODO: Move to a dynamic packet handler (like above)
 416             if (s->read_buf_index >= s->read_buf_size)
 417                 return -1;
 418             ret = mpegts_parse_packet(s->ts, pkt, s->buf + s->read_buf_index,
 419                                       s->read_buf_size - s->read_buf_index);
 420             if (ret < 0)
 421                 return -1;
 422             s->read_buf_index += ret;
 423             if (s->read_buf_index < s->read_buf_size)
 424                 return 1;
 425             else
 426                 return 0;
 427         }
 428     }
 429
 430     if (len < 12)
 431         return -1;
 432
 433     if ((buf[0] & 0xc0) != (RTP_VERSION << 6))
 434         return -1;
 435     if (buf[1] >= 200 && buf[1] <= 204) {
 436         rtcp_parse_packet(s, buf, len);
 437         return -1;
 438     }
 439     payload_type = buf[1] & 0x7f;
 440     if (buf[1] & 0x80)
 441         flags |= RTP_FLAG_MARKER;
 442     seq  = AV_RB16(buf + 2);
 443     timestamp = AV_RB32(buf + 4);
 444     ssrc = AV_RB32(buf + 8);
 445     /* store the ssrc in the RTPDemuxContext */
 446     s->ssrc = ssrc;
 447
 448     /* NOTE: we can handle only one payload type */
 449     if (s->payload_type != payload_type)
 450         return -1;
 451
 452     st = s->st;
 453     // only do something with this if all the rtp checks pass...
 454     if(!rtp_valid_packet_in_sequence(&s->statistics, seq))
 455     {
 456         av_log(st?st->codec:NULL, AV_LOG_ERROR, "RTP: PT=%02x: bad cseq %04x expected=%04x\n",
 457                payload_type, seq, ((s->seq + 1) & 0xffff));
 458         return -1;
 459     }
 460
 461     s->seq = seq;
 462     len -= 12;
 463     buf += 12;
 464
 465     if (!st) {
 466         /* specific MPEG2TS demux support */
 467         ret = mpegts_parse_packet(s->ts, pkt, buf, len);
 468         if (ret < 0)
 469             return -1;
 470         if (ret < len) {
 471             s->read_buf_size = len - ret;
 472             memcpy(s->buf, buf + ret, s->read_buf_size);
 473             s->read_buf_index = 0;
 474             return 1;
 475         }
 476         return 0;
 477     } else if (s->parse_packet) {
 478         rv = s->parse_packet(s->ic, s->dynamic_protocol_context,
 479                              s->st, pkt, &timestamp, buf, len, flags);
 480     } else {
 481         // at this point, the RTP header has been stripped;  This is ASSUMING that there is only 1 CSRC, which in't wise.
 482         switch(st->codec->codec_id) {
 483         case CODEC_ID_MP2:
 484             /* better than nothing: skip mpeg audio RTP header */
 485             if (len <= 4)
 486                 return -1;
 487             h = AV_RB32(buf);
 488             len -= 4;
 489             buf += 4;
 490             av_new_packet(pkt, len);
 491             memcpy(pkt->data, buf, len);
 492             break;
 493         case CODEC_ID_MPEG1VIDEO:
 494         case CODEC_ID_MPEG2VIDEO:
 495             /* better than nothing: skip mpeg video RTP header */
 496             if (len <= 4)
 497                 return -1;
 498             h = AV_RB32(buf);
 499             buf += 4;
 500             len -= 4;
 501             if (h & (1 << 26)) {
 502                 /* mpeg2 */
 503                 if (len <= 4)
 504                     return -1;
 505                 buf += 4;
 506                 len -= 4;
 507             }
 508             av_new_packet(pkt, len);
 509             memcpy(pkt->data, buf, len);
 510             break;
 511             // moved from below, verbatim.  this is because this section handles packets, and the lower switch handles
 512             // timestamps.
 513             // TODO: Put this into a dynamic packet handler...
 514         case CODEC_ID_AAC:
 515             if (rtp_parse_mp4_au(s, buf))
 516                 return -1;
 517             {
 518                 RTPPayloadData *infos = s->rtp_payload_data;
 519                 if (infos == NULL)
 520                     return -1;
 521                 buf += infos->au_headers_length_bytes + 2;
 522                 len -= infos->au_headers_length_bytes + 2;
 523
 524                 /* XXX: Fixme we only handle the case where rtp_parse_mp4_au define
 525                     one au_header */
 526                 av_new_packet(pkt, infos->au_headers[0].size);
 527                 memcpy(pkt->data, buf, infos->au_headers[0].size);
 528                 buf += infos->au_headers[0].size;
 529                 len -= infos->au_headers[0].size;
 530             }
 531             s->read_buf_size = len;
 532             rv= 0;
 533             break;
 534         default:
 535             av_new_packet(pkt, len);
 536             memcpy(pkt->data, buf, len);
 537             break;
 538         }
 539     }
 540
 541     // now perform timestamp things....
 542     finalize_packet(s, pkt, timestamp);
 543
 544     return rv;
 545 }
 546
 547 void rtp_parse_close(RTPDemuxContext *s)
 548 {
 549     // TODO: fold this into the protocol specific data fields.
 550     if (!strcmp(ff_rtp_enc_name(s->payload_type), "MP2T")) {
 551         mpegts_parse_close(s->ts);
 552     }
 553     av_free(s);
 554 }