On 08/21/2014 04:53 AM, Janne Grunau wrote:
> Hi,
>
> this is a little more thourough and less ridicolous patch as
> alternative for your WIP patch 1. I think this is safer alternative.
> There are probably similar frame threading decoding dead locks in other
> mpegenccontext based decoders. So your pthread_frame fix is probably
> still a good idea.
> For release branches I prefer this though.
>
> Janne
>
> ---8<---
>
> MpegEncContext based decoders are only fully initialized after the
> ff_thread_get_buffer() call. If an decoding error occurs before the
> first ff_thread_get_buffer() call the decoder is left in state which
> causes dead locks in frame threaded decoding.
>
> CC: [email protected]
> ---
>  libavcodec/rv34.c | 16 +++++++++-------
>  1 file changed, 9 insertions(+), 7 deletions(-)
>
> diff --git a/libavcodec/rv34.c b/libavcodec/rv34.c
> index 0c36348..4dd154c 100644
> --- a/libavcodec/rv34.c
> +++ b/libavcodec/rv34.c
> @@ -1635,22 +1635,24 @@ int ff_rv34_decode_frame(AVCodecContext *avctx,
>      }else
>          slice_count = avctx->slice_count;
>  
> +#define ERR_INVALIDDATA(msg)                          \
> +    av_log(avctx, AV_LOG_ERROR, msg);                 \
> +    if (!s->linesize) s->context_initialized = 0;     \
> +    return AVERROR_INVALIDDATA;
> +
>      //parse first slice header to check whether this frame can be decoded
>      if(get_slice_offset(avctx, slices_hdr, 0) < 0 ||
>         get_slice_offset(avctx, slices_hdr, 0) > buf_size){
> -        av_log(avctx, AV_LOG_ERROR, "Slice offset is invalid\n");
> -        return AVERROR_INVALIDDATA;
> +        ERR_INVALIDDATA("Slice offset is invalid\n");
>      }
>      init_get_bits(&s->gb, buf+get_slice_offset(avctx, slices_hdr, 0), 
> (buf_size-get_slice_offset(avctx, slices_hdr, 0))*8);
>      if(r->parse_slice_header(r, &r->s.gb, &si) < 0 || si.start){
> -        av_log(avctx, AV_LOG_ERROR, "First slice header is incorrect\n");
> -        return AVERROR_INVALIDDATA;
> +        ERR_INVALIDDATA("First slice header is incorrect\n");
>      }
>      if ((!s->last_picture_ptr || !s->last_picture_ptr->f->data[0]) &&
>          si.type == AV_PICTURE_TYPE_B) {
> -        av_log(avctx, AV_LOG_ERROR, "Invalid decoder state: B-frame without "
> -               "reference data.\n");
> -        return AVERROR_INVALIDDATA;
> +        ERR_INVALIDDATA("Invalid decoder state: B-frame without reference "
> +                        "data.\n");
>      }
>      if(   (avctx->skip_frame >= AVDISCARD_NONREF && 
> si.type==AV_PICTURE_TYPE_B)
>         || (avctx->skip_frame >= AVDISCARD_NONKEY && 
> si.type!=AV_PICTURE_TYPE_I)

Nice. This does solve the problem.  Is there anything I can do to help check 
for leaks?  I've run it in valgrind and it
does show leaks, but I don't know if these are just normal things that are left 
unclean when avplay exits.

-- 
John      GnuPG fingerprint: D0EC B3DB C372 D1F1 0B01  83F0 49F1 D7B2 60D4 D0F7


Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to