[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: squeeze: graphical environment broken

On Tue, Jun 23, 2009 at 08:19:26PM +1200, Michael Cree wrote:
> BTW, to run 
> the latest Xserver on Alpha one must compile their own 2.6.30 kernel (or 
> patch a 2.6.29 kernel).  The 2.6.30 kernel doesn't seem to have 
> propagated to Alpha on Debian unstable yet.

Is that xserver-xorg >= 7.4?  I'm waiting on that one to maybe fix the
gnome-settings-daemon issue, because I can't upgrade past the current
testing versions of gnome-settings-daemon and gnome-session with the
xserver-xorg 1:7.3+18 package.

I've had consistently good luck running kernels built from the stock
kernel.org source.  On those rare occasions where we (people who run
on Alphas) uncover something strange directly attributable to the kernel
code, the fix is delivered quickly.  I don't suppose it hurts that I'm
usually the canary in that particular coal mine :-).

> Can someone advise whether I should register a bug report of the  kernel 
> oops generated by the Xserver against kernel (which would have to go the 
> linux kernel list since the kernel hasn't arrived in Debian yet) or 
> against the Xserver?  I normally would report it against the kernel, but 
> I am guessing because the Xserver has low level access to graphics 
> hardware, and presumably some kernel data, that it might be better to 
> advise the X developers. 

The kernel folks would be interested from the standpoint of there
shouldn't be *anything* in userland that crashes the kernel.  However,
as you surmise, when an app has low-level access to hardware, the usual
rules don't apply.  For what it's worth, I'm currently seeing some list
traffic related to xkb causing segfaults.  Anyway, I'd probably report
the "oops" to both l-k and the X people.  If nothing else, I would
think the kernel people could provide a clue as to what got stomped on
and how.

Bob Tracy          |  "Talk is cheap because supply exceeds demand."
rct@frus.com       |   -- Bill Kennedy

Reply to: