On Tue, 16 Jun 2009, Trent Piepho wrote:
> On Tue, 16 Jun 2009, Mauro Carvalho Chehab wrote:
> > Em Tue, 16 Jun 2009 17:57:20 +0200 (CEST)
> > Guennadi Liakhovetski <g.liakhovetski@xxxxxx> escreveu:
> > > On Sat, 30 May 2009, Trent Piepho wrote:
> > > + if (walign + halign < salign) {
> > > + /* Max walign where there is still a valid width */
> > > + unsigned int wmaxa = __fls(wmax ^ (wmin - 1));
> > >
> > > I cannot follow correctness of the above, sorry. Take a simple example:
> > > wmax=0x7f, wmin=7, wmaxa = __fls(0x7f ^ 6) = __fls(0x79) = 0. And the
> > > comment says it's the "maximum walign where there is still a valid width."
> > > What am I missing?
> > >
> > > +
> > > + /* up the smaller alignment until we have enough */
> > > + do {
> > > + if (walign <= halign && walign < wmaxa) {
> > >
> >
> > As I'm still cooking the patches, I prefer to postpone the align ones until we
> > are comfortable with them.
> >
> > Trent,
> >
> > Could you please take a look on the above comments
>
> There is no bug with the wmaxa code, but there is a different bug
> elsewhere.
>
> Please pull from http://linuxtv.org/hg/~tap/fix
>
> for the following changeset:
>
> 01/01: v4l2: Fix flaw in alignment code
> http://linuxtv.org/hg/~tap/fix?cmd=changeset;node=4ef7fb102b6c
/* up the smaller alignment until we have enough */
do {
- if (walign <= halign && walign < wmaxa) {
+ if (halign >= hmaxa ||
+ (walign <= halign && walign < wmaxa)) {
Do I understand it right now, that as soon as your halign now reaches
hmaxa, you'll stop incrementing halign and then keep incrementing walign
even beyond wmaxa?...
*w = clamp_align(*w, wmin, wmax, walign + 1);
walign = __ffs(*w);
} else {
Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
