[Mesa-users] Segmentation fault in "Fake_glXUseXFont" (Mesa 10.4.3)

[Brian Paul]

On 02/26/2015 03:41 AM, Martin Knoblauch wrote:
> On Wed, Feb 25, 2015 at 9:17 PM, Brian Paul <brianp at vmware.com
> <mailto:brianp at vmware.com>> wrote:
>
>     On 02/25/2015 09:59 AM, Martin Knoblauch wrote:
>
>
>
>
>
>         Hi Brian,
>
>            so I fixed the configure option, applied the patch and
>         rebuilt. The
>         program is now failing even quicker in:
>
>         #0  XMesaGetCurrentDrawable () at xm_api.c:1324
>
>         This is the only useful part of the stack and I need to omit the
>         rest to
>         hide the innocent and incompetent ...
>
>
>     I really don't see how we'd be failing at line 1324 unless the
>     'front' pointer was garbage (which it should never be).  Can you use
>     a debug build and gdb to try to narrow this down a little?
>
>     I also hacked the Mesa xfont.c demo and can't reproduce a crash in
>     XMesaGetCurrentDrawable().
>
>     -Brian
>
>
>
> Hi Brian,
>
>   this is what gdb tells me:
>
> 1324             return front->drawable;
> (gdb) list
> 1319    {
> 1320       XMesaBuffer xmbuffer = XMesaGetCurrentBuffer();
> 1321       if (xmbuffer) {
> 1322          struct xmesa_renderbuffer *front = xmbuffer->frontxrb;
> 1323          if (front)
> 1324             return front->drawable;
> 1325       }
> 1326       return 0;
> 1327    }
> 1328
> (gdb) print front
> $1 = (struct xmesa_renderbuffer *) 0x72c3800000030
> (gdb) print front->drawable
> Cannot access memory at address 0x72c38000000c0
> (gdb) print *front
> Cannot access memory at address 0x72c3800000030
> (gdb)
>
> Seems the "front" pointer is not-null, but not readable.
>
> Any ideas ?

Not off-hand.

Could you possibly try to hack a Mesa demo to reproduce the issue?  The 
Mesa xfont.c demo is what I was using.  I added a few calls to 
glXGetCurrentDrawable() but it didn't crash.

-Brian