[FFmpeg-soc] [PATCH] Move format from AVFilterBuffer to AVFilterPicRef

[Stefano Sabatini]

On date Friday 2010-08-06 10:21:42 -0700, S.N. Hemanth Meenakshisundaram encoded:
> 
> > > On Sun, Aug 01, 2010 at 11:41:09PM -0700, S.N. Hemanth
> Meenakshisundaram wrote:
> > > >  ffplay.c                  |    8 ++++----
> > > >  libavfilter/avfilter.h    |    2 +-
> > > >  libavfilter/defaults.c    |    8 ++++----
> > > >  libavfilter/vf_overlay.c  |   12 ++++++------
> > > >  libavfilter/vsrc_buffer.c |    2 +-
> > > >  libavfilter/vsrc_movie.c  |    2 +-
> > > >  6 files changed, 17 insertions(+), 17 deletions(-)
> > > > 415376236550b0dd86fe180594d1f65f9d9d077f 
> 0001-Move-format-from-AVFilterBuffer-to-AVFilterPicRef.patch
> > >
> > > looks ok if tested
> 
> > This is causing regressions here.
> 
> > Hemanth, can you check it?
> 
> Hi Stefano,
> 
> The regression was being caused by my previous patch to resize
> AVFilterBuffer's data and linesize arrays to 8. Strangely, the errors only
> show up in make test after applying the current patch.
> 
> Applying the attached fix and then the current patch will solve the issue
> and make test passes. I had only tested with make lavfitest and so missed
> this. Will test with 'make test' too in the future.
> 
> Also, this bug fix will no longer be needed once the later patch to resize
> AVFilterBufferRef itself is applied.
> 
> Regards,
> Hemanth

> --- a/libavfilter/defaults.c
> +++ b/libavfilter/defaults.c
> @@ -59,8 +59,8 @@
>                                      // SIMD-friendly
>      av_fill_image_pointers(pic->data, pic->format, ref->h, buf, pic->linesize);
>  
> -    memcpy(ref->data,     pic->data,     sizeof(pic->data));
> -    memcpy(ref->linesize, pic->linesize, sizeof(pic->linesize));
> +    memcpy(ref->data,     pic->data,     4*sizeof(pic->data[0]));
> +    memcpy(ref->linesize, pic->linesize, 4*sizeof(pic->linesize[0]));
>  
>      return ref;

Both applied.