Commit | Line | Data |
---|---|---|
2ba45a60 DM |
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 | ||
25 | #include "avformat.h" | |
26 | #include "rtpdec.h" | |
27 | ||
28 | #define RTP_HEVC_PAYLOAD_HEADER_SIZE 2 | |
29 | #define RTP_HEVC_FU_HEADER_SIZE 1 | |
30 | #define RTP_HEVC_DONL_FIELD_SIZE 2 | |
31 | #define HEVC_SPECIFIED_NAL_UNIT_TYPES 48 | |
32 | ||
33 | /* SDP out-of-band signaling data */ | |
34 | struct PayloadContext { | |
35 | int using_donl_field; | |
36 | int profile_id; | |
37 | }; | |
38 | ||
39 | static const uint8_t start_sequence[] = { 0x00, 0x00, 0x00, 0x01 }; | |
40 | ||
41 | static av_cold PayloadContext *hevc_new_context(void) | |
42 | { | |
43 | return av_mallocz(sizeof(PayloadContext)); | |
44 | } | |
45 | ||
46 | static av_cold void hevc_free_context(PayloadContext *data) | |
47 | { | |
48 | av_free(data); | |
49 | } | |
50 | ||
51 | static av_cold int hevc_init(AVFormatContext *ctx, int st_index, | |
52 | PayloadContext *data) | |
53 | { | |
54 | av_dlog(ctx, "hevc_init() for stream %d\n", st_index); | |
55 | ||
56 | if (st_index < 0) | |
57 | return 0; | |
58 | ||
59 | ctx->streams[st_index]->need_parsing = AVSTREAM_PARSE_FULL; | |
60 | ||
61 | return 0; | |
62 | } | |
63 | ||
64 | static av_cold int hevc_sdp_parse_fmtp_config(AVFormatContext *s, | |
65 | AVStream *stream, | |
66 | PayloadContext *hevc_data, | |
67 | char *attr, char *value) | |
68 | { | |
69 | /* profile-space: 0-3 */ | |
70 | /* profile-id: 0-31 */ | |
71 | if (!strcmp(attr, "profile-id")) { | |
72 | hevc_data->profile_id = atoi(value); | |
73 | av_dlog(s, "SDP: found profile-id: %d\n", hevc_data->profile_id); | |
74 | } | |
75 | ||
76 | /* tier-flag: 0-1 */ | |
77 | /* level-id: 0-255 */ | |
78 | /* interop-constraints: [base16] */ | |
79 | /* profile-compatibility-indicator: [base16] */ | |
80 | /* sprop-sub-layer-id: 0-6, defines highest possible value for TID, default: 6 */ | |
81 | /* recv-sub-layer-id: 0-6 */ | |
82 | /* max-recv-level-id: 0-255 */ | |
83 | /* tx-mode: MSM,SSM */ | |
84 | /* sprop-vps: [base64] */ | |
85 | /* sprop-sps: [base64] */ | |
86 | /* sprop-pps: [base64] */ | |
87 | /* sprop-sei: [base64] */ | |
88 | /* max-lsr, max-lps, max-cpb, max-dpb, max-br, max-tr, max-tc */ | |
89 | /* max-fps */ | |
90 | ||
91 | /* sprop-max-don-diff: 0-32767 | |
92 | ||
93 | When the RTP stream depends on one or more other RTP | |
94 | streams (in this case tx-mode MUST be equal to "MSM" and | |
95 | MSM is in use), this parameter MUST be present and the | |
96 | value MUST be greater than 0. | |
97 | */ | |
98 | if (!strcmp(attr, "sprop-max-don-diff")) { | |
99 | if (atoi(value) > 0) | |
100 | hevc_data->using_donl_field = 1; | |
101 | av_dlog(s, "Found sprop-max-don-diff in SDP, DON field usage is: %d\n", | |
102 | hevc_data->using_donl_field); | |
103 | } | |
104 | ||
105 | /* sprop-depack-buf-nalus: 0-32767 */ | |
106 | if (!strcmp(attr, "sprop-depack-buf-nalus")) { | |
107 | if (atoi(value) > 0) | |
108 | hevc_data->using_donl_field = 1; | |
109 | av_dlog(s, "Found sprop-depack-buf-nalus in SDP, DON field usage is: %d\n", | |
110 | hevc_data->using_donl_field); | |
111 | } | |
112 | ||
113 | /* sprop-depack-buf-bytes: 0-4294967295 */ | |
114 | /* depack-buf-cap */ | |
115 | /* sprop-segmentation-id: 0-3 */ | |
116 | /* sprop-spatial-segmentation-idc: [base16] */ | |
117 | /* dec-parallel-ca: */ | |
118 | /* include-dph */ | |
119 | ||
120 | return 0; | |
121 | } | |
122 | ||
123 | static av_cold int hevc_parse_sdp_line(AVFormatContext *ctx, int st_index, | |
124 | PayloadContext *hevc_data, const char *line) | |
125 | { | |
126 | AVStream *current_stream; | |
127 | AVCodecContext *codec; | |
128 | const char *sdp_line_ptr = line; | |
129 | ||
130 | if (st_index < 0) | |
131 | return 0; | |
132 | ||
133 | current_stream = ctx->streams[st_index]; | |
134 | codec = current_stream->codec; | |
135 | ||
136 | if (av_strstart(sdp_line_ptr, "framesize:", &sdp_line_ptr)) { | |
137 | char str_video_width[50]; | |
138 | char *str_video_width_ptr = str_video_width; | |
139 | ||
140 | /* | |
141 | * parse "a=framesize:96 320-240" | |
142 | */ | |
143 | ||
144 | /* ignore spaces */ | |
145 | while (*sdp_line_ptr && *sdp_line_ptr == ' ') | |
146 | sdp_line_ptr++; | |
147 | /* ignore RTP payload ID */ | |
148 | while (*sdp_line_ptr && *sdp_line_ptr != ' ') | |
149 | sdp_line_ptr++; | |
150 | /* ignore spaces */ | |
151 | while (*sdp_line_ptr && *sdp_line_ptr == ' ') | |
152 | sdp_line_ptr++; | |
153 | /* extract the actual video resolution description */ | |
154 | while (*sdp_line_ptr && *sdp_line_ptr != '-' && | |
155 | (str_video_width_ptr - str_video_width) < sizeof(str_video_width) - 1) | |
156 | *str_video_width_ptr++ = *sdp_line_ptr++; | |
157 | /* add trailing zero byte */ | |
158 | *str_video_width_ptr = '\0'; | |
159 | ||
160 | /* determine the width value */ | |
161 | codec->width = atoi(str_video_width); | |
162 | /* jump beyond the "-" and determine the height value */ | |
163 | codec->height = atoi(sdp_line_ptr + 1); | |
164 | } else if (av_strstart(sdp_line_ptr, "fmtp:", &sdp_line_ptr)) { | |
165 | return ff_parse_fmtp(ctx, current_stream, hevc_data, sdp_line_ptr, | |
166 | hevc_sdp_parse_fmtp_config); | |
167 | } | |
168 | ||
169 | return 0; | |
170 | } | |
171 | ||
172 | static int hevc_handle_packet(AVFormatContext *ctx, PayloadContext *rtp_hevc_ctx, | |
173 | AVStream *st, AVPacket *pkt, uint32_t *timestamp, | |
174 | const uint8_t *buf, int len, uint16_t seq, | |
175 | int flags) | |
176 | { | |
177 | const uint8_t *rtp_pl = buf; | |
178 | int tid, lid, nal_type; | |
179 | int first_fragment, last_fragment, fu_type; | |
180 | uint8_t new_nal_header[2]; | |
181 | int res = 0; | |
182 | ||
183 | /* sanity check for size of input packet: 1 byte payload at least */ | |
184 | if (len < RTP_HEVC_PAYLOAD_HEADER_SIZE + 1) { | |
185 | av_log(ctx, AV_LOG_ERROR, "Too short RTP/HEVC packet, got %d bytes\n", len); | |
186 | return AVERROR_INVALIDDATA; | |
187 | } | |
188 | ||
189 | /* | |
190 | decode the HEVC payload header according to section 4 of draft version 6: | |
191 | ||
192 | 0 1 | |
193 | 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 | |
194 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | |
195 | |F| Type | LayerId | TID | | |
196 | +-------------+-----------------+ | |
197 | ||
198 | Forbidden zero (F): 1 bit | |
199 | NAL unit type (Type): 6 bits | |
200 | NUH layer ID (LayerId): 6 bits | |
201 | NUH temporal ID plus 1 (TID): 3 bits | |
202 | */ | |
203 | nal_type = (buf[0] >> 1) & 0x3f; | |
204 | lid = ((buf[0] << 5) & 0x20) | ((buf[1] >> 3) & 0x1f); | |
205 | tid = buf[1] & 0x07; | |
206 | ||
207 | /* sanity check for correct layer ID */ | |
208 | if (lid) { | |
209 | /* future scalable or 3D video coding extensions */ | |
210 | avpriv_report_missing_feature(ctx, "Multi-layer HEVC coding\n"); | |
211 | return AVERROR_PATCHWELCOME; | |
212 | } | |
213 | ||
214 | /* sanity check for correct temporal ID */ | |
215 | if (!tid) { | |
216 | av_log(ctx, AV_LOG_ERROR, "Illegal temporal ID in RTP/HEVC packet\n"); | |
217 | return AVERROR_INVALIDDATA; | |
218 | } | |
219 | ||
220 | /* sanity check for correct NAL unit type */ | |
221 | if (nal_type > 50) { | |
222 | av_log(ctx, AV_LOG_ERROR, "Unsupported (HEVC) NAL type (%d)\n", nal_type); | |
223 | return AVERROR_INVALIDDATA; | |
224 | } | |
225 | ||
226 | switch (nal_type) { | |
227 | /* aggregated packets (AP) */ | |
228 | case 48: | |
229 | /* pass the HEVC payload header */ | |
230 | buf += RTP_HEVC_PAYLOAD_HEADER_SIZE; | |
231 | len -= RTP_HEVC_PAYLOAD_HEADER_SIZE; | |
232 | ||
233 | /* pass the HEVC DONL field */ | |
234 | if (rtp_hevc_ctx->using_donl_field) { | |
235 | buf += RTP_HEVC_DONL_FIELD_SIZE; | |
236 | len -= RTP_HEVC_DONL_FIELD_SIZE; | |
237 | } | |
238 | ||
239 | /* fall-through */ | |
240 | /* video parameter set (VPS) */ | |
241 | case 32: | |
242 | /* sequence parameter set (SPS) */ | |
243 | case 33: | |
244 | /* picture parameter set (PPS) */ | |
245 | case 34: | |
246 | /* supplemental enhancement information (SEI) */ | |
247 | case 39: | |
248 | /* single NAL unit packet */ | |
249 | default: | |
250 | /* sanity check for size of input packet: 1 byte payload at least */ | |
251 | if (len < 1) { | |
252 | av_log(ctx, AV_LOG_ERROR, | |
253 | "Too short RTP/HEVC packet, got %d bytes of NAL unit type %d\n", | |
254 | len, nal_type); | |
255 | return AVERROR_INVALIDDATA; | |
256 | } | |
257 | ||
258 | /* create A/V packet */ | |
259 | if ((res = av_new_packet(pkt, sizeof(start_sequence) + len)) < 0) | |
260 | return res; | |
261 | /* A/V packet: copy start sequence */ | |
262 | memcpy(pkt->data, start_sequence, sizeof(start_sequence)); | |
263 | /* A/V packet: copy NAL unit data */ | |
264 | memcpy(pkt->data + sizeof(start_sequence), buf, len); | |
265 | ||
266 | break; | |
267 | /* fragmentation unit (FU) */ | |
268 | case 49: | |
269 | /* pass the HEVC payload header */ | |
270 | buf += RTP_HEVC_PAYLOAD_HEADER_SIZE; | |
271 | len -= RTP_HEVC_PAYLOAD_HEADER_SIZE; | |
272 | ||
273 | if (len < 1) | |
274 | return AVERROR_INVALIDDATA; | |
275 | /* | |
276 | decode the FU header | |
277 | ||
278 | 0 1 2 3 4 5 6 7 | |
279 | +-+-+-+-+-+-+-+-+ | |
280 | |S|E| FuType | | |
281 | +---------------+ | |
282 | ||
283 | Start fragment (S): 1 bit | |
284 | End fragment (E): 1 bit | |
285 | FuType: 6 bits | |
286 | */ | |
287 | first_fragment = buf[0] & 0x80; | |
288 | last_fragment = buf[0] & 0x40; | |
289 | fu_type = buf[0] & 0x3f; | |
290 | ||
291 | /* pass the HEVC FU header */ | |
292 | buf += RTP_HEVC_FU_HEADER_SIZE; | |
293 | len -= RTP_HEVC_FU_HEADER_SIZE; | |
294 | ||
295 | /* pass the HEVC DONL field */ | |
296 | if (rtp_hevc_ctx->using_donl_field) { | |
297 | buf += RTP_HEVC_DONL_FIELD_SIZE; | |
298 | len -= RTP_HEVC_DONL_FIELD_SIZE; | |
299 | } | |
300 | ||
301 | av_dlog(ctx, " FU type %d with %d bytes\n", fu_type, len); | |
302 | ||
303 | if (len > 0) { | |
304 | new_nal_header[0] = (rtp_pl[0] & 0x81) | (fu_type << 1); | |
305 | new_nal_header[1] = rtp_pl[1]; | |
306 | ||
307 | /* start fragment vs. subsequent fragments */ | |
308 | if (first_fragment) { | |
309 | if (!last_fragment) { | |
310 | /* create A/V packet which is big enough */ | |
311 | if ((res = av_new_packet(pkt, sizeof(start_sequence) + sizeof(new_nal_header) + len)) < 0) | |
312 | return res; | |
313 | /* A/V packet: copy start sequence */ | |
314 | memcpy(pkt->data, start_sequence, sizeof(start_sequence)); | |
315 | /* A/V packet: copy new NAL header */ | |
316 | memcpy(pkt->data + sizeof(start_sequence), new_nal_header, sizeof(new_nal_header)); | |
317 | /* A/V packet: copy NAL unit data */ | |
318 | memcpy(pkt->data + sizeof(start_sequence) + sizeof(new_nal_header), buf, len); | |
319 | } else { | |
320 | av_log(ctx, AV_LOG_ERROR, "Illegal combination of S and E bit in RTP/HEVC packet\n"); | |
321 | res = AVERROR_INVALIDDATA; | |
322 | } | |
323 | } else { | |
324 | /* create A/V packet */ | |
325 | if ((res = av_new_packet(pkt, len)) < 0) | |
326 | return res; | |
327 | /* A/V packet: copy NAL unit data */ | |
328 | memcpy(pkt->data, buf, len); | |
329 | } | |
330 | } else { | |
331 | /* sanity check for size of input packet: 1 byte payload at least */ | |
332 | av_log(ctx, AV_LOG_ERROR, | |
333 | "Too short RTP/HEVC packet, got %d bytes of NAL unit type %d\n", | |
334 | len, nal_type); | |
335 | res = AVERROR_INVALIDDATA; | |
336 | } | |
337 | ||
338 | break; | |
339 | /* PACI packet */ | |
340 | case 50: | |
341 | /* Temporal scalability control information (TSCI) */ | |
342 | avpriv_report_missing_feature(ctx, "PACI packets for RTP/HEVC\n"); | |
343 | res = AVERROR_PATCHWELCOME; | |
344 | break; | |
345 | } | |
346 | ||
347 | pkt->stream_index = st->index; | |
348 | ||
349 | return res; | |
350 | } | |
351 | ||
352 | RTPDynamicProtocolHandler ff_hevc_dynamic_handler = { | |
353 | .enc_name = "HEVC", | |
354 | .codec_type = AVMEDIA_TYPE_VIDEO, | |
355 | .codec_id = AV_CODEC_ID_HEVC, | |
356 | .init = hevc_init, | |
357 | .parse_sdp_a_line = hevc_parse_sdp_line, | |
358 | .alloc = hevc_new_context, | |
359 | .free = hevc_free_context, | |
360 | .parse_packet = hevc_handle_packet | |
361 | }; | |
362 | ||
363 | RTPDynamicProtocolHandler ff_h265_dynamic_handler = { | |
364 | .enc_name = "H265", | |
365 | .codec_type = AVMEDIA_TYPE_VIDEO, | |
366 | .codec_id = AV_CODEC_ID_HEVC, | |
367 | .init = hevc_init, | |
368 | .parse_sdp_a_line = hevc_parse_sdp_line, | |
369 | .alloc = hevc_new_context, | |
370 | .free = hevc_free_context, | |
371 | .parse_packet = hevc_handle_packet | |
372 | }; |