On Sun, 12 Feb 2023, John Coiner wrote:
This is v2 of the patch for https://trac.ffmpeg.org/ticket/10148.
It implements the handling described at
http://ffmpeg.org/pipermail/ffmpeg-devel/2023-February/306542.html, that is:
* If we receive [AUD][IDR], we emit [AUD][SPS][PPS][IDR]
* If we receive [SEI][AUD][IDR], we emit [AUD][SPS][PPS][SEI][IDR]
This is speculative; it would be good to hear from Marton or others about
whether this is the right handling.
Looks fine to me, but I am no expert in H264 either.
Can you check and test the attached patch? It contins cosmetic changes to
your work. If that looks OK to you as well, than I will apply it in a few
days.
Thanks,
Marton
The other possible handling would be to simply prepend the extradata to the
existing frame's bitstream without relocating the AUD to be in front. So if we
received:
[AUD][IDR]
We would emit:
[SPS][PPS][AUD][IDR]
That's not quite compliant with the H.264 spec, which says an AUD shall be the
first NAL in a frame when it's present. I doubt real decoders care and it would
be simpler to implement. Let me know if you'd prefer the simpler handling.
---
libavformat/mpegtsenc.c | 55 ++++++++++++++++++++++++++++++++++++-----
1 file changed, 49 insertions(+), 6 deletions(-)
diff --git a/libavformat/mpegtsenc.c b/libavformat/mpegtsenc.c
index 00ad426086..a4a2d8cdaf 100644
--- a/libavformat/mpegtsenc.c
+++ b/libavformat/mpegtsenc.c
@@ -1835,6 +1835,13 @@ static int opus_get_packet_samples(AVFormatContext *s,
AVPacket *pkt)
return duration;
}
+// Copies `size_bytes` from `source` to `dest`.
+// Returns a pointer to the next destination address after the range written.
+static uint8_t* append(uint8_t* dest, const uint8_t* source, int size_bytes) {
+ memcpy(dest, source, size_bytes);
+ return dest + size_bytes;
+}
+
static int mpegts_write_packet_internal(AVFormatContext *s, AVPacket *pkt)
{
AVStream *st = s->streams[pkt->stream_index];
@@ -1877,6 +1884,7 @@ static int mpegts_write_packet_internal(AVFormatContext
*s, AVPacket *pkt)
if (st->codecpar->codec_id == AV_CODEC_ID_H264) {
const uint8_t *p = buf, *buf_end = p + size;
+ const uint8_t *found_aud = NULL, *found_aud_end = NULL;
uint32_t state = -1;
int extradd = (pkt->flags & AV_PKT_FLAG_KEY) ?
st->codecpar->extradata_size : 0;
int ret = ff_check_h264_startcode(s, st, pkt);
@@ -1886,17 +1894,34 @@ static int mpegts_write_packet_internal(AVFormatContext
*s, AVPacket *pkt)
if (extradd && AV_RB24(st->codecpar->extradata) > 1)
extradd = 0;
+ // Ensure that all pictures are prefixed with an AUD, and that
+ // IDR pictures are also prefixed with SPS and PPS. SPS and PPS
+ // are assumed to be available in 'extradata' if not found in-band.
do {
p = avpriv_find_start_code(p, buf_end, &state);
av_log(s, AV_LOG_TRACE, "nal %"PRId32"\n", state & 0x1f);
- if ((state & 0x1f) == 7)
+ if ((state & 0x1f) == 7) // SPS NAL
extradd = 0;
- } while (p < buf_end && (state & 0x1f) != 9 &&
- (state & 0x1f) != 5 && (state & 0x1f) != 1);
-
- if ((state & 0x1f) != 5)
+ if ((state & 0x1f) == 9) { // AUD NAL
+ found_aud = p - 4; // start of the 0x000001 start code.
+ found_aud_end = p + 1; // first byte past the AUD.
+
+ if (found_aud < buf)
+ found_aud = buf;
+ if (buf_end < found_aud_end)
+ found_aud_end = buf_end;
+ }
+ } while (p < buf_end
+ && (state & 0x1f) != 5 // IDR picture
+ && (state & 0x1f) != 1 // non-IDR picture
+ && (extradd > 0 || !found_aud));
+ if ((state & 0x1f) != 5) {
+ // Did not find an IDR picture; do not emit extradata.
extradd = 0;
- if ((state & 0x1f) != 9) { // AUD NAL
+ }
+
+ if (!found_aud) {
+ // Prefix 'buf' with the missing AUD, and extradata if needed.
data = av_malloc(pkt->size + 6 + extradd);
if (!data)
return AVERROR(ENOMEM);
@@ -1907,6 +1932,24 @@ static int mpegts_write_packet_internal(AVFormatContext
*s, AVPacket *pkt)
data[5] = 0xf0; // any slice type (0xe) + rbsp stop one bit
buf = data;
size = pkt->size + 6 + extradd;
+ } else if (extradd != 0) {
+ // Move the AUD up to the beginning of the frame, where the H.264
+ // spec requires it to appear. Emit the extradata after it.
+ const int aud_size = found_aud_end - found_aud;
+ const int new_pkt_size = pkt->size + 1 + extradd;
+ uint8_t *pos;
+ data = av_malloc(new_pkt_size);
+ if (!data)
+ return AVERROR(ENOMEM);
+ AV_WB8(data, 0x0);
+ pos = data + 1;
+ pos = append(pos, found_aud, aud_size);
+ pos = append(pos, st->codecpar->extradata, extradd);
+ pos = append(pos, pkt->data, found_aud - pkt->data);
+ pos = append(pos, found_aud_end, buf_end - found_aud_end);
+ av_assert0(data + new_pkt_size == pos);
+ buf = data;
+ size = new_pkt_size;
}
} else if (st->codecpar->codec_id == AV_CODEC_ID_AAC) {
if (pkt->size < 2) {
--
2.39.1.581.gbfd45094c4-goog
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
From fb042a8962c4f9c53e892ecae85b5531f19ca7de Mon Sep 17 00:00:00 2001
From: John Coiner <jpcoi...@gmail.com>
Date: Sun, 12 Feb 2023 18:06:07 -0500
Subject: [PATCH] avformat/mpegtsenc: re-emit extradata ahead of IDR pictures
even if AUD is already present
Current mpegtsenc code only inserts SPS/PPS from extradata before IDR frames if
AUD is also inserted.
Unfortunately some encoders may preface a key frame with an AUD, but no
SPS/PPS. In that case current code does not repeat the "extradata" and the
resulting HLS stream may become noncompliant and unjoinable.
Fix this by always inserting SPS/PPS and moving AUD to the beginning of the
packet if it is already present.
Fixes ticket #10148.
Signed-off-by: Marton Balint <c...@passwd.hu>
---
libavformat/mpegtsenc.c | 48 +++++++++++++++++++++++++++++++++++------
1 file changed, 41 insertions(+), 7 deletions(-)
diff --git a/libavformat/mpegtsenc.c b/libavformat/mpegtsenc.c
index 00ad426086..ceed089587 100644
--- a/libavformat/mpegtsenc.c
+++ b/libavformat/mpegtsenc.c
@@ -29,6 +29,8 @@
#include "libavcodec/ac3_parser_internal.h"
#include "libavcodec/avcodec.h"
+#include "libavcodec/bytestream.h"
+#include "libavcodec/h264.h"
#include "libavcodec/startcode.h"
#include "avformat.h"
@@ -1877,6 +1879,7 @@ static int mpegts_write_packet_internal(AVFormatContext *s, AVPacket *pkt)
if (st->codecpar->codec_id == AV_CODEC_ID_H264) {
const uint8_t *p = buf, *buf_end = p + size;
+ const uint8_t *found_aud = NULL, *found_aud_end = NULL;
uint32_t state = -1;
int extradd = (pkt->flags & AV_PKT_FLAG_KEY) ? st->codecpar->extradata_size : 0;
int ret = ff_check_h264_startcode(s, st, pkt);
@@ -1886,27 +1889,58 @@ static int mpegts_write_packet_internal(AVFormatContext *s, AVPacket *pkt)
if (extradd && AV_RB24(st->codecpar->extradata) > 1)
extradd = 0;
+ /* Ensure that all pictures are prefixed with an AUD, and that
+ * IDR pictures are also prefixed with SPS and PPS. SPS and PPS
+ * are assumed to be available in 'extradata' if not found in-band. */
do {
p = avpriv_find_start_code(p, buf_end, &state);
av_log(s, AV_LOG_TRACE, "nal %"PRId32"\n", state & 0x1f);
- if ((state & 0x1f) == 7)
+ if ((state & 0x1f) == H264_NAL_SPS)
extradd = 0;
- } while (p < buf_end && (state & 0x1f) != 9 &&
- (state & 0x1f) != 5 && (state & 0x1f) != 1);
-
- if ((state & 0x1f) != 5)
+ if ((state & 0x1f) == H264_NAL_AUD) {
+ found_aud = p - 4; // start of the 0x000001 start code.
+ found_aud_end = p + 1; // first byte past the AUD.
+ if (found_aud < buf)
+ found_aud = buf;
+ if (buf_end < found_aud_end)
+ found_aud_end = buf_end;
+ }
+ } while (p < buf_end
+ && (state & 0x1f) != H264_NAL_IDR_SLICE
+ && (state & 0x1f) != H264_NAL_SLICE
+ && (extradd > 0 || !found_aud));
+ if ((state & 0x1f) != H264_NAL_IDR_SLICE)
extradd = 0;
- if ((state & 0x1f) != 9) { // AUD NAL
+
+ if (!found_aud) {
+ /* Prefix 'buf' with the missing AUD, and extradata if needed. */
data = av_malloc(pkt->size + 6 + extradd);
if (!data)
return AVERROR(ENOMEM);
memcpy(data + 6, st->codecpar->extradata, extradd);
memcpy(data + 6 + extradd, pkt->data, pkt->size);
AV_WB32(data, 0x00000001);
- data[4] = 0x09;
+ data[4] = H264_NAL_AUD;
data[5] = 0xf0; // any slice type (0xe) + rbsp stop one bit
buf = data;
size = pkt->size + 6 + extradd;
+ } else if (extradd != 0) {
+ /* Move the AUD up to the beginning of the frame, where the H.264
+ * spec requires it to appear. Emit the extradata after it. */
+ PutByteContext pb;
+ const int new_pkt_size = pkt->size + 1 + extradd;
+ data = av_malloc(new_pkt_size);
+ if (!data)
+ return AVERROR(ENOMEM);
+ bytestream2_init_writer(&pb, data, new_pkt_size);
+ bytestream2_put_byte(&pb, 0x00);
+ bytestream2_put_buffer(&pb, found_aud, found_aud_end - found_aud);
+ bytestream2_put_buffer(&pb, st->codecpar->extradata, extradd);
+ bytestream2_put_buffer(&pb, pkt->data, found_aud - pkt->data);
+ bytestream2_put_buffer(&pb, found_aud_end, buf_end - found_aud_end);
+ av_assert0(new_pkt_size == bytestream2_tell_p(&pb));
+ buf = data;
+ size = new_pkt_size;
}
} else if (st->codecpar->codec_id == AV_CODEC_ID_AAC) {
if (pkt->size < 2) {
--
2.35.3
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
