[FFmpeg-devel] questions about VP8 decode, we found that there seems some buggs of code

[Wei Gao]

2013/4/18 Ronald S. Bultje <rsbultje at gmail.com>

> Hi,
>
> On Tue, Apr 16, 2013 at 7:32 PM, Wei Gao <highgod0401 at gmail.com> wrote:
>
> > Hi,
> > the detail is as follows:
> > in the dsputil_template.c function: ff_emulated_edge_mc()
> > this function will be used to copy a rectangular area of samples to a
> > temporary buffer and replicate the
> > border samples.
> > line 165 to 170:    // copy existing part    for(; y<end_y; y++){
> > memcpy(buf, src, w*sizeof(pixel));        src += linesize;        buf +=
> > linesize;    }
> > the numbers of the loop is the end_y and when the (src + end_y *
> linesize)
> > was out of the border of the memory
> > of the src, the data of the buf will be chaotic and unsafe.
> > for example:the vp8_mc_luma() function in the vp8.c will call the
> > ff_emulated_edge_mc() function. when the input video is
> > 1080p, in the last line the end_y = 19 and the (src + 19 * linesize) will
> > be out of the border of the memory
> > of the src.
> >
>
> Do you have a sample that gives valgrind/asan warnings, different output
> than the reference decoder (libvpx/dixie) or otherwise can help us verify
> the bug?
>
 We want use GPU to decode VP8, but the output is a little different from C
code. So we reference the C code and find there are some bits overflowed.

>
> Ronald
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>