Re: Funky "Blue screen" issue while rebooting from X with 2.6.18-git21

Re: Funky "Blue screen" issue while rebooting from X with 2.6.18-git21

[Adrian Bunk]

On Wed, Oct 04, 2006 at 10:16:41PM +0200, Jesper Juhl wrote:
> I have a strange "problem" with 2.6.18-git21 that I've never had with
> any previous kernel. If I open up an xterm in X, su to root and
> 'reboot' (or 'shutdown -r now') I instantly get a blue screen that
> persists until the box actually reboots.

Pavel, is this a known issue or should Jesper bisect?

> With previous kernels (all that I can remember, latest tested being
> 2.6.18-git15 (then I jumped from -git15 to -git21)) what happened was
> that X would die and I would be returned to text mode so that I could
> actually see all the shutdown messages from my init scripts (which is
> very nice).
> 
> I don't really know what info would be relevant to provide regarding
> this issue, so please ask for any info you need.
> 
> I can start testing kernels between 2.6.18-git15 and 2.6.18-git21
> and/or  doing git bisects if anyone thinks it will be useful. If you
> want that, please speak up, since I would rather not build and
> test-boot a lot of kernels for no reason if nobody wants the info. But
> if it will be useful I'll be happy to do it.
> 
> Anyway, there's a bug somewhere, let's squash it ;-)

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

Re: Funky "Blue screen" issue while rebooting from X with 2.6.18-git21

[Pavel Machek]

Hi!

On Sun 08-10-06 19:47:59, Adrian Bunk wrote:
> On Wed, Oct 04, 2006 at 10:16:41PM +0200, Jesper Juhl wrote:
> > I have a strange "problem" with 2.6.18-git21 that I've never had with
> > any previous kernel. If I open up an xterm in X, su to root and
> > 'reboot' (or 'shutdown -r now') I instantly get a blue screen that
> > persists until the box actually reboots.
> 
> Pavel, is this a known issue or should Jesper bisect?

Jesper should show it is kernel problem and not userland race.

If userspace does kill -15 -1; kill -9 -1, and X fails to shut down in
time, it is userland problem ('should wait for X to shut down').

-- 
Thanks for all the (sleeping) penguins.

Re: Funky "Blue screen" issue while rebooting from X with 2.6.18-git21

[Jesper Juhl]

On 08/10/06, Pavel Machek <pavel@ucw.cz> wrote:
> Hi!
>
> On Sun 08-10-06 19:47:59, Adrian Bunk wrote:
> > On Wed, Oct 04, 2006 at 10:16:41PM +0200, Jesper Juhl wrote:
> > > I have a strange "problem" with 2.6.18-git21 that I've never had with
> > > any previous kernel. If I open up an xterm in X, su to root and
> > > 'reboot' (or 'shutdown -r now') I instantly get a blue screen that
> > > persists until the box actually reboots.
> >
> > Pavel, is this a known issue or should Jesper bisect?
>
> Jesper should show it is kernel problem and not userland race.
>

Jesper will try to do that ;-)


> If userspace does kill -15 -1; kill -9 -1, and X fails to shut down in
> time, it is userland problem ('should wait for X to shut down').
>

Well, I just checked my initscript that is run when going into
runlevels 0 & 6, and it does this :

(...)

# Kill all processes.
# INIT is supposed to handle this entirely now, but this didn't always
# work correctly without this second pass at killing off the processes.
# Since INIT already notified the user that processes were being killed,
# we'll avoid echoing this info this time around.
if [ ! "$1" = "fast" ]; then # shutdown did not already kill all processes
  /sbin/killall5 -15
  /bin/sleep 5
  /sbin/killall5 -9
fi

(...)

For the complete script, see here :
ftp://ftp.kernel.org/pub/linux/kernel/people/juhl/rc.0


I'm trying to prove if it's a kernel or userspace problem, but it's
proving hard to reproduce reliably. But, what makes me suspect a
kernel problem is that I've never seen this with 2.6.17 and earlier
kernels, but since somewhere in the 2.6.18-rc series I've experienced
this "blue screen" problem once in a while and I've also had a problem
with the screen going all white when switching from X to a plain tty
and back (once it goes white it stays that way permanently until I
reboot) - I *never* see those issues when running 2.6.17.x and
earlier.

-- 
Jesper Juhl <jesper.juhl@gmail.com>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html

Re: Funky "Blue screen" issue while rebooting from X with 2.6.18-git21

[Pavel Machek]

Hi!

> >> > I have a strange "problem" with 2.6.18-git21 that I've never had with
> >> > any previous kernel. If I open up an xterm in X, su to root and
> >> > 'reboot' (or 'shutdown -r now') I instantly get a blue screen that
> >> > persists until the box actually reboots.
> >>
> >> Pavel, is this a known issue or should Jesper bisect?
> >
> >Jesper should show it is kernel problem and not userland race.
> >
> 
> Jesper will try to do that ;-)
> 
> 
> >If userspace does kill -15 -1; kill -9 -1, and X fails to shut down in
> >time, it is userland problem ('should wait for X to shut down').
> >
> 
> Well, I just checked my initscript that is run when going into
> runlevels 0 & 6, and it does this :
> 
> (...)
> 
> # Kill all processes.
> # INIT is supposed to handle this entirely now, but this didn't always
> # work correctly without this second pass at killing off the processes.
> # Since INIT already notified the user that processes were being killed,
> # we'll avoid echoing this info this time around.
> if [ ! "$1" = "fast" ]; then # shutdown did not already kill all processes
>  /sbin/killall5 -15
>  /bin/sleep 5
>  /sbin/killall5 -9
> fi

...so, if X takes more than five seconds to shut down, you kill it
with -9, resulting in blue screen. Too bad, and not an kernel problem.

Try inserting something like

	while ps -aux | grep myXserver;
		sleep 1;
	done

alternatively, remove/shorten the sleep and you should experience blue
screen in 2.6.17.

> kernels, but since somewhere in the 2.6.18-rc series I've experienced
> this "blue screen" problem once in a while and I've also had a problem
> with the screen going all white when switching from X to a plain tty
> and back (once it goes white it stays that way permanently until I
> reboot) - I *never* see those issues when running 2.6.17.x and
> earlier.

Maybe something got slower in 2.6.18?
								Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: Funky "Blue screen" issue while rebooting from X with 2.6.18-git21

[Jesper Juhl]

On 08/10/06, Pavel Machek <pavel@ucw.cz> wrote:
> Hi!
>
> > >> > I have a strange "problem" with 2.6.18-git21 that I've never had with
> > >> > any previous kernel. If I open up an xterm in X, su to root and
> > >> > 'reboot' (or 'shutdown -r now') I instantly get a blue screen that
> > >> > persists until the box actually reboots.
> > >>
> > >> Pavel, is this a known issue or should Jesper bisect?
> > >
> > >Jesper should show it is kernel problem and not userland race.
> > >
> >
> > Jesper will try to do that ;-)
> >
> >
> > >If userspace does kill -15 -1; kill -9 -1, and X fails to shut down in
> > >time, it is userland problem ('should wait for X to shut down').
> > >
> >
> > Well, I just checked my initscript that is run when going into
> > runlevels 0 & 6, and it does this :
> >
> > (...)
> >
> > # Kill all processes.
> > # INIT is supposed to handle this entirely now, but this didn't always
> > # work correctly without this second pass at killing off the processes.
> > # Since INIT already notified the user that processes were being killed,
> > # we'll avoid echoing this info this time around.
> > if [ ! "$1" = "fast" ]; then # shutdown did not already kill all processes
> >  /sbin/killall5 -15
> >  /bin/sleep 5
> >  /sbin/killall5 -9
> > fi
>
> ...so, if X takes more than five seconds to shut down, you kill it
> with -9, resulting in blue screen. Too bad, and not an kernel problem.
>
> Try inserting something like
>
>         while ps -aux | grep myXserver;
>                 sleep 1;
>         done
>
> alternatively, remove/shorten the sleep and you should experience blue
> screen in 2.6.17.
>

I'll try and test these things and report back in a few days when I've
gathered some data.


> > kernels, but since somewhere in the 2.6.18-rc series I've experienced
> > this "blue screen" problem once in a while and I've also had a problem
> > with the screen going all white when switching from X to a plain tty
> > and back (once it goes white it stays that way permanently until I
> > reboot) - I *never* see those issues when running 2.6.17.x and
> > earlier.
>
> Maybe something got slower in 2.6.18?
>                                                                 Pavel
>
Perhaps...

-- 
Jesper Juhl <jesper.juhl@gmail.com>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html