[FFmpeg-devel] [PATCH 1/2] h2645_parse: Propagate NAL header parsing errors

[James Almer]

On 3/18/2019 12:46 PM, Derek Buitenhuis wrote:
> If we don't propagate these errors, h264dec and hevcdec can get up to all sorts of
> weirdness, especially threaded, while trying to continue on with things they shouldn't.
> Can cause stuff like:
> 
>     [hevc @ 0x5555577107c0] get_buffer() cannot be called after ff_thread_finish_setup()
>     [hevc @ 0x5555577107c0] thread_get_buffer() failed
>     [hevc @ 0x5555577107c0] Error parsing NAL unit #5.
>     Error while decoding stream #0:0: Cannot allocate memory
>     <deadlock>
> 
> Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
> ---
>  libavcodec/h2645_parse.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/libavcodec/h2645_parse.c b/libavcodec/h2645_parse.c
> index 942f2c5d71..175c986c71 100644
> --- a/libavcodec/h2645_parse.c
> +++ b/libavcodec/h2645_parse.c
> @@ -503,6 +503,8 @@ int ff_h2645_packet_split(H2645Packet *pkt, const uint8_t *buf, int length,
>                         nal->type);
>              }
>              pkt->nb_nals--;
> +            if (ret < 0)
> +                return ret;

This will abort the splitting process when it's meant to only discard
the faulty NAL. Even the log message stats it's skipping it.