[FFmpeg-devel] [PATCH] libavformat/mpegtsenc.c -- correctly re-emit extradata ahead of IDR pictures

[Marton Balint]

On Fri, 27 Jan 2023, John Coiner wrote:

> From: John Coiner <jcoiner at google.com>
>
> This is the proposed fix for https://trac.ffmpeg.org/ticket/10148
>
> Very sorry to send the same patch 3 times; this time with `git send-email' so it should be legal :)
>
> ---
> libavformat/mpegtsenc.c | 18 +++++++++++-------
> 1 file changed, 11 insertions(+), 7 deletions(-)
>
> diff --git a/libavformat/mpegtsenc.c b/libavformat/mpegtsenc.c
> index 48d39e6a7d..6cb398f99c 100644
> --- a/libavformat/mpegtsenc.c
> +++ b/libavformat/mpegtsenc.c
> @@ -1886,17 +1886,21 @@ static int mpegts_write_packet_internal(AVFormatContext *s, AVPacket *pkt)
>         if (extradd && AV_RB24(st->codecpar->extradata) > 1)
>             extradd = 0;
>
> -        do {
> +        while (p < buf_end
> +               && extradd > 0
> +               && (state & 0x1f) != 5  // IDR picture
> +               && (state & 0x1f) != 1  // non-IDR picture
> +               ) {
>             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) != 5) {
> +            // Not an IDR picture
>             extradd = 0;
> -        if ((state & 0x1f) != 9) { // AUD NAL
> +        }
> +        if (extradd > 0) {

This does not look right, because AFAIK you always want to insert an AUD 
unless it is already there. I guess that is why current code works as it 
is, it assumes if an AUD is already present, then IDR-s are already 
prefixed with SPS/PPS, and no magic is needed.

I wonder if the issue is indeed caused by this:
https://patchwork.ffmpeg.org/project/ffmpeg/patch/tencent_EE0E40DE9A569FE5AB454E6E700A2DA79A08@qq.com/

Because if it is, then maybe nothing need to be fixed in the mpegts muxer.

Regards,
Marton