[MPlayer-dev-eng] [PATCH] bigendian fix for hwac3

[elupus]

"reimar" <subversion at mplayerhq.hu> wrote in message 
news:20070914125213.AB4153B613 at natsuki.mplayerhq.hu...
> Author: reimar
> Date: Fri Sep 14 14:52:13 2007
> New Revision: 24456
>
> Log:
> Fix off-by-one error if fsize is odd (does handling that case even make 
> sense?)
> and remove a TODO comment that no longer applies.
>
>
> Modified:
>   trunk/libmpcodecs/ad_hwac3.c
>
> Modified: trunk/libmpcodecs/ad_hwac3.c
> ==============================================================================
> --- trunk/libmpcodecs/ad_hwac3.c (original)
> +++ trunk/libmpcodecs/ad_hwac3.c Fri Sep 14 14:52:13 2007
> @@ -360,10 +360,9 @@ static int decode_audio_dts(unsigned cha
> #ifdef WORDS_BIGENDIAN
>   memcpy(&buf[8], indata_ptr, fsize);
> #else
> -  //TODO if fzise is odd, swab doesn't copy the last byte
>   swab(indata_ptr, &buf[8], fsize);
>   if (fsize & 1)
> -    buf[8+fsize] = indata_ptr[fsize];
> +    buf[8+fsize-1] = indata_ptr[fsize-1];
> #endif
>   memset(&buf[fsize + 8], 0, nr_samples * 2 * 2 - (fsize + 8));


Yea it does. It does, i have a sample somewhere that has the issue. Think it 
was dts thou, but still the same. In essence the solution was to extend the 
data with a zero for swab to use. This how we handle it in xbmc's dvdplayer.

  if( size & 0x1 )
  {
    swab(src, dst, size-1);
    dst+=size-1;
    src+=size-1;

    dst[0] = 0x0;
    dst[1] = src[0];
    return size+1;
  }
  else
  {
    swab(src, dst, size);
    return size;
  }

cheers