Avoid a kernel driver / X server crash situation.

Avoid a kernel driver / X server crash situation.

Change the order of our calls to xrandr to mask a situation that can cause
the GPU to hang, i915 driver to give up, and the X server to crash.  This
crash is not recoverable and requires a reboot.

BUG=chromium-os:6701
TEST=manual: [Un]Plug VGA, then hit Ctrl-F4.  Repeat...

This workaround changes the order of xrandr calls from:

xrandr --output LVDS1 --mode 1280x800
xrandr --output VGA1 --off
xrandr --fb 1280x800

to:

xrandr --output VGA1 --off
xrandr --fb 1280x800
xrandr --output LVDS1 --mode 1280x800

In my experimentation, turning on LVDS later makes a big difference.

But it is not perfect.  I have still seen the failure, but it is much
harder to reproduce.

Change-Id: I300887ef2c65173bafeccd8afdf657f91f71c736

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=e6cd2a5

[Daniel Erat, 2010-11-16 00:16:13]

Interesting. :-)  LGTM!

[marcheu1, 2010-11-16 01:58:04]

On Mon, Nov 15, 2010 at 16:16, <derat@chromium.org> wrote:

> Interesting. :-)  LGTM!
>
>
> http://codereview.chromium.org/5024002/
>

Hi,

I think I have a driver-level fix for this bug, I just need to get around to
imaging a device and testing it.

Would you guys mind leaving that to me for now? We can always apply a
workaround later if it doesn't work...

Thanks,
Stéphane

[bfreed, 2010-11-17 21:53:28]

Please share your driver patch.
I am very interested in testing it.

I am leaning toward checking in this workaround anyway.  It cleans up a bit of a
snap seen when returning from presentation mode.

On 2010/11/16 01:58:04, marcheu_google.com wrote:
> On Mon, Nov 15, 2010 at 16:16, <mailto:derat@chromium.org> wrote:
> 
> > Interesting. :-)  LGTM!
> >
> >
> > http://codereview.chromium.org/5024002/
> >
> 
> Hi,
> 
> I think I have a driver-level fix for this bug, I just need to get around to
> imaging a device and testing it.
> 
> Would you guys mind leaving that to me for now? We can always apply a
> workaround later if it doesn't work...
> 
> Thanks,
> Stéphane

[marcheu1, 2010-11-18 22:16:55]

Hi,

Okay, the issue here is more complex than I thought... I have a number of
fixes for the intel driver but they don't fix this one. I'll investigate
further, as I can reproduce here. In the meantime, I think you should push
the workaround.

Stéphane