[MPlayer-cvslog] r21385 - trunk/libvo/osd_template.c

[Reimar Döffinger]

Hello,
On Fri, Dec 01, 2006 at 11:23:02AM +0100, Diego Biurrun wrote:
> On Wed, Nov 29, 2006 at 03:46:51PM +0100, Reimar Döffinger wrote:
> > On Wed, Nov 29, 2006 at 03:30:25PM +0100, Diego Biurrun wrote:
> > > On Wed, Nov 29, 2006 at 02:49:48PM +0100, reimar wrote:
> > > > 
> > > > Modified:
> > > >    trunk/libvo/osd_template.c
> > > > 
> > > > --- trunk/libvo/osd_template.c	(original)
> > > > +++ trunk/libvo/osd_template.c	Wed Nov 29 14:49:48 2006
> > > > @@ -204,7 +204,7 @@
> > > >      for(y=0;y<h;y++){
> > > >          register unsigned char *dst = dstbase;
> > > >          register int x;
> > > > -#ifdef ARCH_X86
> > > > +#if defined(ARCH_X86) && (!defined(ARCH_X86_64) || defined(HAVE_MMX))
> > > >  #ifdef HAVE_MMX
> > > >      asm volatile(
> > > >  	PREFETCHW" %0\n\t"
> > > > @@ -277,7 +277,7 @@
> > > >  	    dst += 3;
> > > >          }
> > > >  #endif /* !HAVE_MMX */
> > > > -#else /*non x86 arch*/
> > > > +#else /*non x86 arch or x86_64 with MMX disabled */
> > > 
> > > This is backwards, same below.
> > 
> > Well, I always thought the way it is normally written is backwards :-).
> > Anyway, I just made my comment fit what was there before.
> 
> So can we make up a policy about this?  IMO the comments after #else and
> #endif should always contain exactly what the #if line contained.
> Everything else is either confusing or a pain to search backwards to...

Yes, of course, that's the only way that overall makes sense, which one
is "backwards" is another question though.
Just don't expect me to fix that kind of things in existing code.

Greetings,
Reimar Döffinger