| 1 | /* |
| 2 | * RTP parser for HEVC/H.265 payload format (draft version 6) |
| 3 | * Copyright (c) 2014 Thomas Volkert <thomas@homer-conferencing.com> |
| 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 | |
| 23 | #include "libavutil/avstring.h" |
| 24 | #include "libavutil/base64.h" |
| 25 | |
| 26 | #include "avformat.h" |
| 27 | #include "rtpdec.h" |
| 28 | |
| 29 | #define RTP_HEVC_PAYLOAD_HEADER_SIZE 2 |
| 30 | #define RTP_HEVC_FU_HEADER_SIZE 1 |
| 31 | #define RTP_HEVC_DONL_FIELD_SIZE 2 |
| 32 | #define HEVC_SPECIFIED_NAL_UNIT_TYPES 48 |
| 33 | |
| 34 | /* SDP out-of-band signaling data */ |
| 35 | struct PayloadContext { |
| 36 | int using_donl_field; |
| 37 | int profile_id; |
| 38 | uint8_t *sps, *pps, *vps, *sei; |
| 39 | int sps_size, pps_size, vps_size, sei_size; |
| 40 | }; |
| 41 | |
| 42 | static const uint8_t start_sequence[] = { 0x00, 0x00, 0x00, 0x01 }; |
| 43 | |
| 44 | static av_cold PayloadContext *hevc_new_context(void) |
| 45 | { |
| 46 | return av_mallocz(sizeof(PayloadContext)); |
| 47 | } |
| 48 | |
| 49 | static av_cold void hevc_free_context(PayloadContext *data) |
| 50 | { |
| 51 | av_free(data); |
| 52 | } |
| 53 | |
| 54 | static av_cold int hevc_init(AVFormatContext *ctx, int st_index, |
| 55 | PayloadContext *data) |
| 56 | { |
| 57 | av_dlog(ctx, "hevc_init() for stream %d\n", st_index); |
| 58 | |
| 59 | if (st_index < 0) |
| 60 | return 0; |
| 61 | |
| 62 | ctx->streams[st_index]->need_parsing = AVSTREAM_PARSE_FULL; |
| 63 | |
| 64 | return 0; |
| 65 | } |
| 66 | |
| 67 | static av_cold int hevc_sdp_parse_fmtp_config(AVFormatContext *s, |
| 68 | AVStream *stream, |
| 69 | PayloadContext *hevc_data, |
| 70 | char *attr, char *value) |
| 71 | { |
| 72 | /* profile-space: 0-3 */ |
| 73 | /* profile-id: 0-31 */ |
| 74 | if (!strcmp(attr, "profile-id")) { |
| 75 | hevc_data->profile_id = atoi(value); |
| 76 | av_dlog(s, "SDP: found profile-id: %d\n", hevc_data->profile_id); |
| 77 | } |
| 78 | |
| 79 | /* tier-flag: 0-1 */ |
| 80 | /* level-id: 0-255 */ |
| 81 | /* interop-constraints: [base16] */ |
| 82 | /* profile-compatibility-indicator: [base16] */ |
| 83 | /* sprop-sub-layer-id: 0-6, defines highest possible value for TID, default: 6 */ |
| 84 | /* recv-sub-layer-id: 0-6 */ |
| 85 | /* max-recv-level-id: 0-255 */ |
| 86 | /* tx-mode: MSM,SSM */ |
| 87 | /* sprop-vps: [base64] */ |
| 88 | /* sprop-sps: [base64] */ |
| 89 | /* sprop-pps: [base64] */ |
| 90 | /* sprop-sei: [base64] */ |
| 91 | if (!strcmp(attr, "sprop-vps") || !strcmp(attr, "sprop-sps") || |
| 92 | !strcmp(attr, "sprop-pps") || !strcmp(attr, "sprop-sei")) { |
| 93 | uint8_t **data_ptr; |
| 94 | int *size_ptr; |
| 95 | if (!strcmp(attr, "sprop-vps")) { |
| 96 | data_ptr = &hevc_data->vps; |
| 97 | size_ptr = &hevc_data->vps_size; |
| 98 | } else if (!strcmp(attr, "sprop-sps")) { |
| 99 | data_ptr = &hevc_data->sps; |
| 100 | size_ptr = &hevc_data->sps_size; |
| 101 | } else if (!strcmp(attr, "sprop-pps")) { |
| 102 | data_ptr = &hevc_data->pps; |
| 103 | size_ptr = &hevc_data->pps_size; |
| 104 | } else if (!strcmp(attr, "sprop-sei")) { |
| 105 | data_ptr = &hevc_data->sei; |
| 106 | size_ptr = &hevc_data->sei_size; |
| 107 | } |
| 108 | |
| 109 | while (*value) { |
| 110 | char base64packet[1024]; |
| 111 | uint8_t decoded_packet[1024]; |
| 112 | int decoded_packet_size; |
| 113 | char *dst = base64packet; |
| 114 | |
| 115 | while (*value && *value != ',' && |
| 116 | (dst - base64packet) < sizeof(base64packet) - 1) { |
| 117 | *dst++ = *value++; |
| 118 | } |
| 119 | *dst++ = '\0'; |
| 120 | |
| 121 | if (*value == ',') |
| 122 | value++; |
| 123 | |
| 124 | decoded_packet_size = av_base64_decode(decoded_packet, base64packet, |
| 125 | sizeof(decoded_packet)); |
| 126 | if (decoded_packet_size > 0) { |
| 127 | uint8_t *tmp = av_realloc(*data_ptr, decoded_packet_size + |
| 128 | sizeof(start_sequence) + *size_ptr); |
| 129 | if (!tmp) { |
| 130 | av_log(s, AV_LOG_ERROR, |
| 131 | "Unable to allocate memory for extradata!\n"); |
| 132 | return AVERROR(ENOMEM); |
| 133 | } |
| 134 | *data_ptr = tmp; |
| 135 | |
| 136 | memcpy(*data_ptr + *size_ptr, start_sequence, |
| 137 | sizeof(start_sequence)); |
| 138 | memcpy(*data_ptr + *size_ptr + sizeof(start_sequence), |
| 139 | decoded_packet, decoded_packet_size); |
| 140 | |
| 141 | *size_ptr += sizeof(start_sequence) + decoded_packet_size; |
| 142 | } |
| 143 | } |
| 144 | } |
| 145 | |
| 146 | /* max-lsr, max-lps, max-cpb, max-dpb, max-br, max-tr, max-tc */ |
| 147 | /* max-fps */ |
| 148 | |
| 149 | /* sprop-max-don-diff: 0-32767 |
| 150 | |
| 151 | When the RTP stream depends on one or more other RTP |
| 152 | streams (in this case tx-mode MUST be equal to "MSM" and |
| 153 | MSM is in use), this parameter MUST be present and the |
| 154 | value MUST be greater than 0. |
| 155 | */ |
| 156 | if (!strcmp(attr, "sprop-max-don-diff")) { |
| 157 | if (atoi(value) > 0) |
| 158 | hevc_data->using_donl_field = 1; |
| 159 | av_dlog(s, "Found sprop-max-don-diff in SDP, DON field usage is: %d\n", |
| 160 | hevc_data->using_donl_field); |
| 161 | } |
| 162 | |
| 163 | /* sprop-depack-buf-nalus: 0-32767 */ |
| 164 | if (!strcmp(attr, "sprop-depack-buf-nalus")) { |
| 165 | if (atoi(value) > 0) |
| 166 | hevc_data->using_donl_field = 1; |
| 167 | av_dlog(s, "Found sprop-depack-buf-nalus in SDP, DON field usage is: %d\n", |
| 168 | hevc_data->using_donl_field); |
| 169 | } |
| 170 | |
| 171 | /* sprop-depack-buf-bytes: 0-4294967295 */ |
| 172 | /* depack-buf-cap */ |
| 173 | /* sprop-segmentation-id: 0-3 */ |
| 174 | /* sprop-spatial-segmentation-idc: [base16] */ |
| 175 | /* dec-parallel-ca: */ |
| 176 | /* include-dph */ |
| 177 | |
| 178 | return 0; |
| 179 | } |
| 180 | |
| 181 | static av_cold int hevc_parse_sdp_line(AVFormatContext *ctx, int st_index, |
| 182 | PayloadContext *hevc_data, const char *line) |
| 183 | { |
| 184 | AVStream *current_stream; |
| 185 | AVCodecContext *codec; |
| 186 | const char *sdp_line_ptr = line; |
| 187 | |
| 188 | if (st_index < 0) |
| 189 | return 0; |
| 190 | |
| 191 | current_stream = ctx->streams[st_index]; |
| 192 | codec = current_stream->codec; |
| 193 | |
| 194 | if (av_strstart(sdp_line_ptr, "framesize:", &sdp_line_ptr)) { |
| 195 | char str_video_width[50]; |
| 196 | char *str_video_width_ptr = str_video_width; |
| 197 | |
| 198 | /* |
| 199 | * parse "a=framesize:96 320-240" |
| 200 | */ |
| 201 | |
| 202 | /* ignore spaces */ |
| 203 | while (*sdp_line_ptr && *sdp_line_ptr == ' ') |
| 204 | sdp_line_ptr++; |
| 205 | /* ignore RTP payload ID */ |
| 206 | while (*sdp_line_ptr && *sdp_line_ptr != ' ') |
| 207 | sdp_line_ptr++; |
| 208 | /* ignore spaces */ |
| 209 | while (*sdp_line_ptr && *sdp_line_ptr == ' ') |
| 210 | sdp_line_ptr++; |
| 211 | /* extract the actual video resolution description */ |
| 212 | while (*sdp_line_ptr && *sdp_line_ptr != '-' && |
| 213 | (str_video_width_ptr - str_video_width) < sizeof(str_video_width) - 1) |
| 214 | *str_video_width_ptr++ = *sdp_line_ptr++; |
| 215 | /* add trailing zero byte */ |
| 216 | *str_video_width_ptr = '\0'; |
| 217 | |
| 218 | /* determine the width value */ |
| 219 | codec->width = atoi(str_video_width); |
| 220 | /* jump beyond the "-" and determine the height value */ |
| 221 | codec->height = atoi(sdp_line_ptr + 1); |
| 222 | } else if (av_strstart(sdp_line_ptr, "fmtp:", &sdp_line_ptr)) { |
| 223 | int ret = ff_parse_fmtp(ctx, current_stream, hevc_data, sdp_line_ptr, |
| 224 | hevc_sdp_parse_fmtp_config); |
| 225 | if (hevc_data->vps_size || hevc_data->sps_size || |
| 226 | hevc_data->pps_size || hevc_data->sei_size) { |
| 227 | av_freep(&codec->extradata); |
| 228 | codec->extradata_size = hevc_data->vps_size + hevc_data->sps_size + |
| 229 | hevc_data->pps_size + hevc_data->sei_size; |
| 230 | codec->extradata = av_malloc(codec->extradata_size + |
| 231 | FF_INPUT_BUFFER_PADDING_SIZE); |
| 232 | if (!codec->extradata) { |
| 233 | ret = AVERROR(ENOMEM); |
| 234 | codec->extradata_size = 0; |
| 235 | } else { |
| 236 | int pos = 0; |
| 237 | memcpy(codec->extradata + pos, hevc_data->vps, hevc_data->vps_size); |
| 238 | pos += hevc_data->vps_size; |
| 239 | memcpy(codec->extradata + pos, hevc_data->sps, hevc_data->sps_size); |
| 240 | pos += hevc_data->sps_size; |
| 241 | memcpy(codec->extradata + pos, hevc_data->pps, hevc_data->pps_size); |
| 242 | pos += hevc_data->pps_size; |
| 243 | memcpy(codec->extradata + pos, hevc_data->sei, hevc_data->sei_size); |
| 244 | pos += hevc_data->sei_size; |
| 245 | memset(codec->extradata + pos, 0, FF_INPUT_BUFFER_PADDING_SIZE); |
| 246 | } |
| 247 | |
| 248 | av_freep(&hevc_data->vps); |
| 249 | av_freep(&hevc_data->sps); |
| 250 | av_freep(&hevc_data->pps); |
| 251 | av_freep(&hevc_data->sei); |
| 252 | hevc_data->vps_size = 0; |
| 253 | hevc_data->sps_size = 0; |
| 254 | hevc_data->pps_size = 0; |
| 255 | hevc_data->sei_size = 0; |
| 256 | } |
| 257 | return ret; |
| 258 | } |
| 259 | |
| 260 | return 0; |
| 261 | } |
| 262 | |
| 263 | static int hevc_handle_packet(AVFormatContext *ctx, PayloadContext *rtp_hevc_ctx, |
| 264 | AVStream *st, AVPacket *pkt, uint32_t *timestamp, |
| 265 | const uint8_t *buf, int len, uint16_t seq, |
| 266 | int flags) |
| 267 | { |
| 268 | const uint8_t *rtp_pl = buf; |
| 269 | int tid, lid, nal_type; |
| 270 | int first_fragment, last_fragment, fu_type; |
| 271 | uint8_t new_nal_header[2]; |
| 272 | int res = 0; |
| 273 | |
| 274 | /* sanity check for size of input packet: 1 byte payload at least */ |
| 275 | if (len < RTP_HEVC_PAYLOAD_HEADER_SIZE + 1) { |
| 276 | av_log(ctx, AV_LOG_ERROR, "Too short RTP/HEVC packet, got %d bytes\n", len); |
| 277 | return AVERROR_INVALIDDATA; |
| 278 | } |
| 279 | |
| 280 | /* |
| 281 | decode the HEVC payload header according to section 4 of draft version 6: |
| 282 | |
| 283 | 0 1 |
| 284 | 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 |
| 285 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |
| 286 | |F| Type | LayerId | TID | |
| 287 | +-------------+-----------------+ |
| 288 | |
| 289 | Forbidden zero (F): 1 bit |
| 290 | NAL unit type (Type): 6 bits |
| 291 | NUH layer ID (LayerId): 6 bits |
| 292 | NUH temporal ID plus 1 (TID): 3 bits |
| 293 | */ |
| 294 | nal_type = (buf[0] >> 1) & 0x3f; |
| 295 | lid = ((buf[0] << 5) & 0x20) | ((buf[1] >> 3) & 0x1f); |
| 296 | tid = buf[1] & 0x07; |
| 297 | |
| 298 | /* sanity check for correct layer ID */ |
| 299 | if (lid) { |
| 300 | /* future scalable or 3D video coding extensions */ |
| 301 | avpriv_report_missing_feature(ctx, "Multi-layer HEVC coding\n"); |
| 302 | return AVERROR_PATCHWELCOME; |
| 303 | } |
| 304 | |
| 305 | /* sanity check for correct temporal ID */ |
| 306 | if (!tid) { |
| 307 | av_log(ctx, AV_LOG_ERROR, "Illegal temporal ID in RTP/HEVC packet\n"); |
| 308 | return AVERROR_INVALIDDATA; |
| 309 | } |
| 310 | |
| 311 | /* sanity check for correct NAL unit type */ |
| 312 | if (nal_type > 50) { |
| 313 | av_log(ctx, AV_LOG_ERROR, "Unsupported (HEVC) NAL type (%d)\n", nal_type); |
| 314 | return AVERROR_INVALIDDATA; |
| 315 | } |
| 316 | |
| 317 | switch (nal_type) { |
| 318 | /* aggregated packets (AP) */ |
| 319 | case 48: |
| 320 | /* pass the HEVC payload header */ |
| 321 | buf += RTP_HEVC_PAYLOAD_HEADER_SIZE; |
| 322 | len -= RTP_HEVC_PAYLOAD_HEADER_SIZE; |
| 323 | |
| 324 | /* pass the HEVC DONL field */ |
| 325 | if (rtp_hevc_ctx->using_donl_field) { |
| 326 | buf += RTP_HEVC_DONL_FIELD_SIZE; |
| 327 | len -= RTP_HEVC_DONL_FIELD_SIZE; |
| 328 | } |
| 329 | |
| 330 | /* fall-through */ |
| 331 | /* video parameter set (VPS) */ |
| 332 | case 32: |
| 333 | /* sequence parameter set (SPS) */ |
| 334 | case 33: |
| 335 | /* picture parameter set (PPS) */ |
| 336 | case 34: |
| 337 | /* supplemental enhancement information (SEI) */ |
| 338 | case 39: |
| 339 | /* single NAL unit packet */ |
| 340 | default: |
| 341 | /* sanity check for size of input packet: 1 byte payload at least */ |
| 342 | if (len < 1) { |
| 343 | av_log(ctx, AV_LOG_ERROR, |
| 344 | "Too short RTP/HEVC packet, got %d bytes of NAL unit type %d\n", |
| 345 | len, nal_type); |
| 346 | return AVERROR_INVALIDDATA; |
| 347 | } |
| 348 | |
| 349 | /* create A/V packet */ |
| 350 | if ((res = av_new_packet(pkt, sizeof(start_sequence) + len)) < 0) |
| 351 | return res; |
| 352 | /* A/V packet: copy start sequence */ |
| 353 | memcpy(pkt->data, start_sequence, sizeof(start_sequence)); |
| 354 | /* A/V packet: copy NAL unit data */ |
| 355 | memcpy(pkt->data + sizeof(start_sequence), buf, len); |
| 356 | |
| 357 | break; |
| 358 | /* fragmentation unit (FU) */ |
| 359 | case 49: |
| 360 | /* pass the HEVC payload header */ |
| 361 | buf += RTP_HEVC_PAYLOAD_HEADER_SIZE; |
| 362 | len -= RTP_HEVC_PAYLOAD_HEADER_SIZE; |
| 363 | |
| 364 | /* |
| 365 | decode the FU header |
| 366 | |
| 367 | 0 1 2 3 4 5 6 7 |
| 368 | +-+-+-+-+-+-+-+-+ |
| 369 | |S|E| FuType | |
| 370 | +---------------+ |
| 371 | |
| 372 | Start fragment (S): 1 bit |
| 373 | End fragment (E): 1 bit |
| 374 | FuType: 6 bits |
| 375 | */ |
| 376 | first_fragment = buf[0] & 0x80; |
| 377 | last_fragment = buf[0] & 0x40; |
| 378 | fu_type = buf[0] & 0x3f; |
| 379 | |
| 380 | /* pass the HEVC FU header */ |
| 381 | buf += RTP_HEVC_FU_HEADER_SIZE; |
| 382 | len -= RTP_HEVC_FU_HEADER_SIZE; |
| 383 | |
| 384 | /* pass the HEVC DONL field */ |
| 385 | if (rtp_hevc_ctx->using_donl_field) { |
| 386 | buf += RTP_HEVC_DONL_FIELD_SIZE; |
| 387 | len -= RTP_HEVC_DONL_FIELD_SIZE; |
| 388 | } |
| 389 | |
| 390 | av_dlog(ctx, " FU type %d with %d bytes\n", fu_type, len); |
| 391 | |
| 392 | /* sanity check for size of input packet: 1 byte payload at least */ |
| 393 | if (len > 0) { |
| 394 | new_nal_header[0] = (rtp_pl[0] & 0x81) | (fu_type << 1); |
| 395 | new_nal_header[1] = rtp_pl[1]; |
| 396 | |
| 397 | /* start fragment vs. subsequent fragments */ |
| 398 | if (first_fragment) { |
| 399 | if (!last_fragment) { |
| 400 | /* create A/V packet which is big enough */ |
| 401 | if ((res = av_new_packet(pkt, sizeof(start_sequence) + sizeof(new_nal_header) + len)) < 0) |
| 402 | return res; |
| 403 | /* A/V packet: copy start sequence */ |
| 404 | memcpy(pkt->data, start_sequence, sizeof(start_sequence)); |
| 405 | /* A/V packet: copy new NAL header */ |
| 406 | memcpy(pkt->data + sizeof(start_sequence), new_nal_header, sizeof(new_nal_header)); |
| 407 | /* A/V packet: copy NAL unit data */ |
| 408 | memcpy(pkt->data + sizeof(start_sequence) + sizeof(new_nal_header), buf, len); |
| 409 | } else { |
| 410 | av_log(ctx, AV_LOG_ERROR, "Illegal combination of S and E bit in RTP/HEVC packet\n"); |
| 411 | res = AVERROR_INVALIDDATA; |
| 412 | } |
| 413 | } else { |
| 414 | /* create A/V packet */ |
| 415 | if ((res = av_new_packet(pkt, len)) < 0) |
| 416 | return res; |
| 417 | /* A/V packet: copy NAL unit data */ |
| 418 | memcpy(pkt->data, buf, len); |
| 419 | } |
| 420 | } else { |
| 421 | if (len < 0) { |
| 422 | av_log(ctx, AV_LOG_ERROR, |
| 423 | "Too short RTP/HEVC packet, got %d bytes of NAL unit type %d\n", |
| 424 | len, nal_type); |
| 425 | res = AVERROR_INVALIDDATA; |
| 426 | } else { |
| 427 | res = AVERROR(EAGAIN); |
| 428 | } |
| 429 | } |
| 430 | |
| 431 | break; |
| 432 | /* PACI packet */ |
| 433 | case 50: |
| 434 | /* Temporal scalability control information (TSCI) */ |
| 435 | avpriv_report_missing_feature(ctx, "PACI packets for RTP/HEVC\n"); |
| 436 | res = AVERROR_PATCHWELCOME; |
| 437 | break; |
| 438 | } |
| 439 | |
| 440 | pkt->stream_index = st->index; |
| 441 | |
| 442 | return res; |
| 443 | } |
| 444 | |
| 445 | RTPDynamicProtocolHandler ff_hevc_dynamic_handler = { |
| 446 | .enc_name = "H265", |
| 447 | .codec_type = AVMEDIA_TYPE_VIDEO, |
| 448 | .codec_id = AV_CODEC_ID_HEVC, |
| 449 | .init = hevc_init, |
| 450 | .parse_sdp_a_line = hevc_parse_sdp_line, |
| 451 | .alloc = hevc_new_context, |
| 452 | .free = hevc_free_context, |
| 453 | .parse_packet = hevc_handle_packet |
| 454 | }; |