Re: clocksource mutex deadlock, cat current_clocksource (2.6.33-rc6/7)

[Andreas Mohr <andi@lisas.de>]

Hi,

On Tue, Feb 09, 2010 at 11:02:02AM +0100, Thomas Gleixner wrote:
> On Mon, 8 Feb 2010, Andreas Mohr wrote:
> 
> > On Mon, Feb 08, 2010 at 09:51:05PM +0100, Andreas Mohr wrote:
> > > Looks like it:
> > > - another bootup also had lockup message
> > > - all /var/log/dmesg* have lockup message, oldest is:
> > > 2010-02-07 20:00 dmesg.4.gz
> > > 
> > > Linux version 2.6.33-rc6 (root@note) (gcc version 4.3.4 (Debian 4.3.4-6)) #3 Sun Jan 31 23:47:51 CET 2010
> > 
> > -rc4 and 2.6.32.3 don't show lockup message, instant bootup without any
> > visible delay.
> 
> Just verified, that there is no timekeeping related patch after
> 2.6.33-rc4 except the kgdb lockup prevention patch, which is
> not touching any code which is relevant  at boot time.
> 
> So bisecting between rc4 and now is probably the best thing to do.

Frankly I'm now through so many different attempts
(cold-boot test, 4.3 compile, switch both lines to -O0 in Makefile,
use unmodified .config, restart with cleanly extracted tree,
revert several low-level compiler libraries, check for other
relevant package upgrades - it's not a LILO upgrade or
parameter problem either -, ...) without any improvement
that I'm about to give up (remember: 1+ hours per rebuild).
I simply don't have any idea what else could have broken it.
The only thing remaining is either some low-level compiler support library
that I didn't downgrade or actual hardware damage.

The one thing that might be the cause for my reboot-on-boot issues
is http://marc.info/?l=linux-kernel&m=126611098225127&w=2
(blank boot on Dell machines due to video mode issues - maybe this
increases chances of a reboot on other Dells such as mine).
Will do some verification via vga=ask or so.

At the moment I'm very tempted to simple deafly ignore the issue
until some or another enlightenment comes my way or future upgrades
of gcc/kernel happen to fix it.

Note that after reinstalling old custom kernel packages it does boot properly
(all recent build tests done using -rc4, and with old package reinstalled
-rc4 then did boot).


The lockup warning by NMI watchdog does seem to be more persistent
than my recent reboot issue, however,
but I might have easily missed that in earlier kernel packages.
----> we're dearly missing a suitable user-space kernel warning
notification layer that goes all the way to the desktop.
There's a netlink userspace thingy or so but it's just not
well-established.

Andreas Mohr