Re: system hang (Bug ID #0034213543 ;-) )

From: Amon Ott <ao@rsbac.org>
Subject: Re: system hang (Bug ID #0034213543 ;-) )
Date: Tue, 3 Jul 2001 12:02:17 +0200

Next Article (by Date): New Bugfixes for 1.1.1 Amon Ott
Previous Article (by Date):
Top of Thread: system hang (Bug ID #0034213543 ;-) ) Bencsath Boldizsar
Articles sorted by: [Date] [Author] [Subject]

On Mon, 25 Jun 2001 Bencsath Boldizsar wrote:
> System: more systems
> Kernel: 2.4.5 + VLan 1.0.1 patch + freeswan snapshot patch
> rsbac: 1.1.2pre5 (for 2.4.5)
> Symptom:
> 
> After booting new kernel, the system hangs (with or without oops), or
> simply resets. This happens at the time where the system mounts the hard
> drives.

OK, I saw rare mount hangs before. There is some kind of race with locking in
the mount code, which I will try to find and avoid.
  
> From 4 computers 3 died with that situation, and it was reproducible. The
> 1 which didn't die had rsbac installed before this kernel, 1 also had
> which hanged, 2 were clean (without rsbac). 
> 
> Specifications of the computers: Various. One Pentium Pro (hanged) 1 Duron
> 700 (no hang), 1 thunderbird (hangs), some with more ethernet cards.
> 
> Influence of other kernel patches: The system doesn't hang without RSBAC
> support, hangs with maintenance and normal rsbac kernels. The system also
> hangs if the IPSEC (freeswan) is turned off (but the patch exists). 
> 
> Trace:
> 
> At one system meltdown the kernel wrote some oops message, but it was too
> long for the screen. The last (20) lines were filled with
> <C01CB82C>. (stack?!) System map shows C01CB158-C01CC024 is rsbac_get_attr
> function.

In the hang case, the system loops in a spinlock, a typical deadlock. If
currently nested too deep, the stack might be exhausted and such errors can
occur.

I will rework through the mount code.

Amon.
-
To unsubscribe from the rsbac list, send a mail to
majordomo@rsbac.org with
unsubscribe rsbac
as single line in the body.

Go to Compuniverse LWGate Home Page.