Bug#554450: not fixed by 003829072853546abd973266fe9b24d803f4f5cb
Hello,
> On Mon, Feb 01, 2010 at 10:00:35AM +0100, Uwe Kleine-König wrote:
> > Hello,
> >
> > On Sat, Jan 30, 2010 at 12:31:02PM +0100, Uwe Kleine-König wrote:
> > > reassign 554450 xserver-xorg-core 2:1.7.4-2
> > > thanks
> > >
> > > Hello,
> > >
> > > jcristau on irc guesses that this would be fixed by
> > >
> > > http://cgit.freedesktop.org/xorg/xserver/commit/?h=003829072853546abd973266fe9b24d803f4f5cb
> > >
> > > I will try to verify this later today.
> > This alone didn't help :-/
> >
> > I added some more tests for crtc being NULL but without success.
> Actually my problem was not that crtc is NULL but that the NULL pointer
> dereference already happend while calculating crtc. This didn't show up
> earlier as I just now compiled without -O2.
>
> (gdb) bt
> #0 0x081cdd92 in xf86GetGammaRampSize (pScreen=0x8b77f20)
> at ../../../../hw/xfree86/common/xf86cmap.c:1080
> #1 0x081c8b69 in VidModeGetGammaRampSize (scrnIndex=0)
> at ../../../../hw/xfree86/common/xf86VidMode.c:522
> #2 0xb740c88d in ProcXF86VidModeGetGammaRampSize (client=0x8e94250)
> at ../../../../../hw/xfree86/dixmods/extmod/xf86vmode.c:1606
> #3 0xb740cd3d in ProcXF86VidModeDispatch (client=0x8e94250)
> at ../../../../../hw/xfree86/dixmods/extmod/xf86vmode.c:1698
> #4 0x0806dd87 in Dispatch () at ../../dix/dispatch.c:439
> #5 0x08066b45 in main (argc=9, argv=0xbf8a8104, envp=0xbf8a812c)
> at ../../dix/main.c:285
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0xb7446720 (LWP 1489)]
> 0x081cdd92 in xf86GetGammaRampSize (pScreen=0x8b77f20)
> at ../../../../hw/xfree86/common/xf86cmap.c:1080
> 1080 RRCrtcPtr crtc = config->output[config->compat_output]->crtc->randr_crtc;
> (gdb) print config->output[config->compat_output]
> $1 = (xf86OutputPtr) 0x8b73668
> (gdb) print config->output[config->compat_output]->crtc
> $2 = (xf86CrtcPtr) 0x0
>
> This is too hard for me. So I let this for someone with more knowledge.
>
> While debugging I noticed that the problem isn't reliably triggered by
> gnome-screensaver-command -l. From the few tries I made I think that it
> doesn't happen if LVDS output is enabled or the resolution of VGA is
> 1024x768 (i.e. smaller than my default 1920x...).
In the following situations the bug occurs(-) or doesn't occur(+):
(-) After login when Gnome switched to VGA:1920x1200@60Hz+LVDS:off
(+) when I change display settings from above to e.g.
mirrored VGA+LVDS:1024x768@60Hz
side-by-side VGA+LVDS:1024x768@60Hz
(+) when I change to the text console (Ctrl+Alt+F1) and back
after login.
(Note, in this case the LVDS display shows a mouse pointer
in the center even though it's still "off" in the Gnome
Display Preferences. It doesn't move when the mouse is
moved.)
(+) After login when Gnome Display Preferences were switched to
VGA:1024x768@60Hz+LVDS:off (That is, VGA isn't changed as in
gdm the setting used is mirrored VGA+LVDS:1024x768@60Hz.)
This is still very annoying and I wonder if this qualifies to be
rc-critical.
Best regards
Uwe
--
Pengutronix e.K. | Uwe Kleine-König |
Industrial Linux Solutions | http://www.pengutronix.de/ |
Reply to: