I agree that this can be closed.
The problem in this case was that glXMakeCurrent failed (and that came
from mesa), and it caused kinfocenter to execute a very rare code
path, that called glXDestroyContext twice.
So I guess it is fixed on the two sides now: that path should not
execute anymore, and if in the future it does kinfocenter will not
call the destroy function twice.
Ivo
On Wed, Mar 19, 2008 at 7:14 PM, Raúl Sánchez Siles <[EMAIL PROTECTED]> wrote:
> Hello:
> El Domingo, 16 de Marzo de 2008, Ivo Anjo escribió:
>
>
> > Hello. I've been trying to close kinfocenter bugs upstream, and I've
> > tried to track down this one. I think the problem is on opengl.cpp,
> > around line 582:
> >
> > if (glXMakeCurrent(dpy, win, ctx)) {
> > // some code
> > }
> > else {
> > kDebug() << "Error: glXMakeCurrent failed\n";
> > glXDestroyContext(dpy, ctx);
> > }
> >
> > glXDestroyContext(dpy, ctx); <<--- crash
> >
> > This path should very rarely be hit, and that should explain why it's
> > hard to reproduce. I'll commit the fix to kde trunk, but I don't know
> > what the rules are for backporting for 3.5.x (or if a new 3.5.x
> > release is ever going out). Can anyone else confirm that with this the
> > bug goes away?
> >
> > Ivo Anjo
>
> Jon: Which version solves it?
> Ivo: Do you agree with Jon? Could we close this one?
>
> Thanks.
>
> --
> Raúl Sánchez Siles
> ----->Proud Debian user<-----
> Linux registered user #416098
>
