[FFmpeg-devel] [PATCH] EA TGQ video fixes

[Reimar Döffinger]

On Fri, Apr 17, 2009 at 11:02:41AM +0200, Vitor Sessak wrote:
> Reimar D?ffinger wrote:
> >> And this one should be correct but makes no difference:
> >>
> >> --- eatgq.c     2009-04-16 17:20:15.000000000 +0000
> >> +++ eatgq.c.bak 2009-04-16 17:19:11.000000000 +0000
> >> @@ -42,6 +42,7 @@
> >>      int width,height;
> >>      ScanTable scantable;
> >>      int qtable[64];
> >> +    DECLARE_ALIGNED_16(DCTELEM, block[6][64]);
> >>  } TgqContext;
> >>  
> >>  static av_cold int tgq_decode_init(AVCodecContext *avctx){
> >> @@ -144,7 +145,6 @@
> >>      int mode;
> >>      int i;
> >>      int8_t dc[6];
> >> -    DCTELEM block[6][64];
> >>  
> >>      mode = bytestream_get_byte((const uint8_t**)bs);
> >>      if (mode>buf_end-*bs) {
> >> @@ -156,8 +156,8 @@
> >>          GetBitContext gb;
> >>          init_get_bits(&gb, *bs, mode*8);
> >>          for(i=0; i<6; i++)
> >> -            tgq_decode_block(s, block[i], &gb);
> >> -        tgq_idct_put_mb(s, block, mb_x, mb_y);
> >> +            tgq_decode_block(s, s->block[i], &gb);
> >> +        tgq_idct_put_mb(s, s->block, mb_x, mb_y);
> >>      }else{
> >>          if (mode==3) {
> >>              memset(dc, (*bs)[0], 4);
> > 
> > Waiting to see what the FATE results look like now before doing anything
> > else.
> 
> Why keep it working by sheer luck if we can just fix it for good?

Because it is generally a good idea to know what you are doing -> I want
to wait and see how FATE looks now so we can see if and which things
applying this fixes.