Re: [Mesa3d-users] Possible tesselation bug in Mesa 6.2

[Braden M. <br...@en...>]

On Sun, 2005-01-23 at 11:51 +0000, Keith Whitwell wrote:
> Braden McDaniel wrote:
> > On Sun, 2004-12-05 at 00:47 -0500, Braden McDaniel wrote:
> > 
> >>I'm getting this crash in some tesselation code when using Mesa 6.2:
> >>
> >>        Mesa: _save_upgrade_vertex: dangling reference attr 8
> >>        *** glibc detected *** corrupted double-linked list: 0x08808ae8 ***
> >>        Aborted
> >>
> >>Now, I haven't totally ruled out a bug in my code; however, this code
> >>does not crash when run with Mesa 6.0.
> > 
> > 
> > Can someone give me more information on what this message from Mesa
> > suggests? I'm seeing it in a slightly different context now. Now I get a
> > crash when cleaning up the vertices I allocate in a
> > GLU_TESS_COMBINE_DATA callback. I'm pretty sure I'm not deleting these
> > twice; is it possible that Mesa is free'ing them?
> > 
> 
> I've just fixed a bug associated with the "dangling reference" message, 
> it's committed to the Mesa trunk.

Interestingly, I still get the "dangling reference" message from Mesa
when using the CVS sources. But I do not get the message from glibc and
the crash that came with it. :-)

> The glibc message is unknown - maybe there's a malloc/free bug that only 
> you are triggering?  I'd suggest valgrind for chasing those sorts of 
> problems down.

Yes... No luck with valgrind so far, though.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>