[MPlayer-dev-eng] [PATCH] configure & MEMALIGN_HACK

Michael Niedermayer michaelni at gmx.at
Fri Jan 7 18:03:15 CET 2005


Hi

On Friday 07 January 2005 17:47, Reimar Döffinger wrote:
> Hi,
>
> On Fri, Jan 07, 2005 at 05:06:18PM +0100, Michael Niedermayer wrote:
> > On Friday 07 January 2005 15:26, Reimar Döffinger wrote:
> > [...]
> >
> > > > > > No how did that happen? Here should be the complete patch...
> > > > >
> > > > > Any comments? I intend to apply tomorrow.
> > > >
> > > > Seems to work. I got some
> > > > mencoder in free(): error: modified (chunk-) pointer
> > > > right now, so more of the same is needed.
> > >
> > > Try the attached patch.
> > > Btw. All those if (var) before av_free(var) aren't needed anymore as
> > > av_free does this check as well. I'll leave them there for now
> > > anyway...
> >
> > IMHO remove the unneeded if() and also look at av_freep() it could be
> > used to further simplify the code
>
> How about the attached version then? I decided to use av_freep
> everywhere.

looks ok


>
> > anyway, IMHO defining MEMALIGN_HACK by default is a bad idea, it will
> > cause many more problems, like breaking malloc debuggers
> > it also wont solve unaligned *alloc() outside libav* issues, and solving
> > these similarely will probably lead to more issues with things like
> > strdup()+free()
>
> But since it was applied there weren't any more problems with SSE2 on
> MinGW, or not?

well, i dont use windows ...


> Sure isn't the perfect solution, but still.
> Btw. And why will it break malloc debuggers?

because the allocated space is larger by 17 bytes accesses slightly over the 
end of the array wont be detected


[...]
-- 
Michael

"In any case, just because code is syntactically "valid" GNU C doesn't 
mean gcc can always compile it." -- justification to close a gcc bug




More information about the MPlayer-dev-eng mailing list