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 500
  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 "rtp_internal.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 static void 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     register_dynamic_payload_handler(&mp4v_es_handler);
  61     register_dynamic_payload_handler(&mpeg4_generic_handler);
  62     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, rtp_payload_data_t *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     av_set_pts_info(s->st, 32, 1, 90000);
 285     if (!strcmp(ff_rtp_enc_name(payload_type), "MP2T")) {
 286         s->ts = mpegts_parse_open(s->ic);
 287         if (s->ts == NULL) {
 288             av_free(s);
 289             return NULL;
 290         }
 291     } else {
 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 static int rtp_parse_mp4_au(RTPDemuxContext *s, const uint8_t *buf)
 315 {
 316     int au_headers_length, au_header_size, i;
 317     GetBitContext getbitcontext;
 318     rtp_payload_data_t *infos;
 319
 320     infos = s->rtp_payload_data;
 321
 322     if (infos == NULL)
 323         return -1;
 324
 325     /* decode the first 2 bytes where the AUHeader sections are stored
 326        length in bits */
 327     au_headers_length = AV_RB16(buf);
 328
 329     if (au_headers_length > RTP_MAX_PACKET_LENGTH)
 330       return -1;
 331
 332     infos->au_headers_length_bytes = (au_headers_length + 7) / 8;
 333
 334     /* skip AU headers length section (2 bytes) */
 335     buf += 2;
 336
 337     init_get_bits(&getbitcontext, buf, infos->au_headers_length_bytes * 8);
 338
 339     /* XXX: Wrong if optionnal additional sections are present (cts, dts etc...) */
 340     au_header_size = infos->sizelength + infos->indexlength;
 341     if (au_header_size <= 0 || (au_headers_length % au_header_size != 0))
 342         return -1;
 343
 344     infos->nb_au_headers = au_headers_length / au_header_size;
 345     infos->au_headers = av_malloc(sizeof(struct AUHeaders) * infos->nb_au_headers);
 346
 347     /* XXX: We handle multiple AU Section as only one (need to fix this for interleaving)
 348        In my test, the FAAD decoder does not behave correctly when sending each AU one by one
 349        but does when sending the whole as one big packet...  */
 350     infos->au_headers[0].size = 0;
 351     infos->au_headers[0].index = 0;
 352     for (i = 0; i < infos->nb_au_headers; ++i) {
 353         infos->au_headers[0].size += get_bits_long(&getbitcontext, infos->sizelength);
 354         infos->au_headers[0].index = get_bits_long(&getbitcontext, infos->indexlength);
 355     }
 356
 357     infos->nb_au_headers = 1;
 358
 359     return 0;
 360 }
 361
 362 /**
 363  * This was the second switch in rtp_parse packet.  Normalizes time, if required, sets stream_index, etc.
 364  */
 365 static void finalize_packet(RTPDemuxContext *s, AVPacket *pkt, uint32_t timestamp)
 366 {
 367     if (s->last_rtcp_ntp_time != AV_NOPTS_VALUE) {
 368         int64_t addend;
 369         int delta_timestamp;
 370
 371         /* compute pts from timestamp with received ntp_time */
 372         delta_timestamp = timestamp - s->last_rtcp_timestamp;
 373         /* convert to the PTS timebase */
 374         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);
 375         pkt->pts = addend + delta_timestamp;
 376     }
 377     pkt->stream_index = s->st->index;
 378 }
 379
 380 /**
 381  * Parse an RTP or RTCP packet directly sent as a buffer.
 382  * @param s RTP parse context.
 383  * @param pkt returned packet
 384  * @param buf input buffer or NULL to read the next packets
 385  * @param len buffer len
 386  * @return 0 if a packet is returned, 1 if a packet is returned and more can follow
 387  * (use buf as NULL to read the next). -1 if no packet (error or no more packet).
 388  */
 389 int rtp_parse_packet(RTPDemuxContext *s, AVPacket *pkt,
 390                      const uint8_t *buf, int len)
 391 {
 392     unsigned int ssrc, h;
 393     int payload_type, seq, ret, flags = 0;
 394     AVStream *st;
 395     uint32_t timestamp;
 396     int rv= 0;
 397
 398     if (!buf) {
 399         /* return the next packets, if any */
 400         if(s->st && s->parse_packet) {
 401             timestamp= 0; ///< Should not be used if buf is NULL, but should be set to the timestamp of the packet returned....
 402             rv= s->parse_packet(s, pkt, &timestamp, NULL, 0, flags);
 403             finalize_packet(s, pkt, timestamp);
 404             return rv;
 405         } else {
 406             // TODO: Move to a dynamic packet handler (like above)
 407             if (s->read_buf_index >= s->read_buf_size)
 408                 return -1;
 409             ret = mpegts_parse_packet(s->ts, pkt, s->buf + s->read_buf_index,
 410                                       s->read_buf_size - s->read_buf_index);
 411             if (ret < 0)
 412                 return -1;
 413             s->read_buf_index += ret;
 414             if (s->read_buf_index < s->read_buf_size)
 415                 return 1;
 416             else
 417                 return 0;
 418         }
 419     }
 420
 421     if (len < 12)
 422         return -1;
 423
 424     if ((buf[0] & 0xc0) != (RTP_VERSION << 6))
 425         return -1;
 426     if (buf[1] >= 200 && buf[1] <= 204) {
 427         rtcp_parse_packet(s, buf, len);
 428         return -1;
 429     }
 430     payload_type = buf[1] & 0x7f;
 431     seq  = AV_RB16(buf + 2);
 432     timestamp = AV_RB32(buf + 4);
 433     ssrc = AV_RB32(buf + 8);
 434     /* store the ssrc in the RTPDemuxContext */
 435     s->ssrc = ssrc;
 436
 437     /* NOTE: we can handle only one payload type */
 438     if (s->payload_type != payload_type)
 439         return -1;
 440
 441     st = s->st;
 442     // only do something with this if all the rtp checks pass...
 443     if(!rtp_valid_packet_in_sequence(&s->statistics, seq))
 444     {
 445         av_log(st?st->codec:NULL, AV_LOG_ERROR, "RTP: PT=%02x: bad cseq %04x expected=%04x\n",
 446                payload_type, seq, ((s->seq + 1) & 0xffff));
 447         return -1;
 448     }
 449
 450     s->seq = seq;
 451     len -= 12;
 452     buf += 12;
 453
 454     if (!st) {
 455         /* specific MPEG2TS demux support */
 456         ret = mpegts_parse_packet(s->ts, pkt, buf, len);
 457         if (ret < 0)
 458             return -1;
 459         if (ret < len) {
 460             s->read_buf_size = len - ret;
 461             memcpy(s->buf, buf + ret, s->read_buf_size);
 462             s->read_buf_index = 0;
 463             return 1;
 464         }
 465     } else if (s->parse_packet) {
 466         rv = s->parse_packet(s, pkt, &timestamp, buf, len, flags);
 467     } else {
 468         // at this point, the RTP header has been stripped;  This is ASSUMING that there is only 1 CSRC, which in't wise.
 469         switch(st->codec->codec_id) {
 470         case CODEC_ID_MP2:
 471             /* better than nothing: skip mpeg audio RTP header */
 472             if (len <= 4)
 473                 return -1;
 474             h = AV_RB32(buf);
 475             len -= 4;
 476             buf += 4;
 477             av_new_packet(pkt, len);
 478             memcpy(pkt->data, buf, len);
 479             break;
 480         case CODEC_ID_MPEG1VIDEO:
 481         case CODEC_ID_MPEG2VIDEO:
 482             /* better than nothing: skip mpeg video RTP header */
 483             if (len <= 4)
 484                 return -1;
 485             h = AV_RB32(buf);
 486             buf += 4;
 487             len -= 4;
 488             if (h & (1 << 26)) {
 489                 /* mpeg2 */
 490                 if (len <= 4)
 491                     return -1;
 492                 buf += 4;
 493                 len -= 4;
 494             }
 495             av_new_packet(pkt, len);
 496             memcpy(pkt->data, buf, len);
 497             break;
 498             // moved from below, verbatim.  this is because this section handles packets, and the lower switch handles
 499             // timestamps.
 500             // TODO: Put this into a dynamic packet handler...
 501         case CODEC_ID_AAC:
 502             if (rtp_parse_mp4_au(s, buf))
 503                 return -1;
 504             {
 505                 rtp_payload_data_t *infos = s->rtp_payload_data;
 506                 if (infos == NULL)
 507                     return -1;
 508                 buf += infos->au_headers_length_bytes + 2;
 509                 len -= infos->au_headers_length_bytes + 2;
 510
 511                 /* XXX: Fixme we only handle the case where rtp_parse_mp4_au define
 512                     one au_header */
 513                 av_new_packet(pkt, infos->au_headers[0].size);
 514                 memcpy(pkt->data, buf, infos->au_headers[0].size);
 515                 buf += infos->au_headers[0].size;
 516                 len -= infos->au_headers[0].size;
 517             }
 518             s->read_buf_size = len;
 519             rv= 0;
 520             break;
 521         default:
 522             av_new_packet(pkt, len);
 523             memcpy(pkt->data, buf, len);
 524             break;
 525         }
 526
 527         // now perform timestamp things....
 528         finalize_packet(s, pkt, timestamp);
 529     }
 530     return rv;
 531 }
 532
 533 void rtp_parse_close(RTPDemuxContext *s)
 534 {
 535     // TODO: fold this into the protocol specific data fields.
 536     if (!strcmp(ff_rtp_enc_name(s->payload_type), "MP2T")) {
 537         mpegts_parse_close(s->ts);
 538     }
 539     av_free(s);
 540 }