Assertion failure wth ext3 on standard Redhat 7.2 kernel

Assertion failure wth ext3 on standard Redhat 7.2 kernel

[Robert Lowery]

Hi,

I have recently upgraded my machine to redhat 7.2 (from redhat 6.2) and I am
constantly (eg < 1 hour of uptime) getting kernel crashes which I believe
are relating to ext3.  I know I should submit this to redhat's bugzilla, but
before I do I was hoping someone on the list might want to comment as I am
sure it is probably something I am doing wrong.

The machine has previously been rock solid, and the only hardware change I
have made recently is to add an extra 64M of RAM (Total 128MB).  I have run
memtest86 on it for 10 hours with no errors reported.

The last error I get is
Assertion failure in __journal_file_buffer() at transaction.c:1953:
"jh->b_jlist < 9".  I can still switch between consoles, but cannot do much
else other than press the reset button :(  

The machine is a Pentium233MMX with a 430FX (Trition I) motherboard.

Please let me know if I can gather any more information to help track this
down.

-Robert

Re: Assertion failure wth ext3 on standard Redhat 7.2 kernel

[Andrew Morton]

Robert Lowery wrote:
> 
> Hi,
> 
> I have recently upgraded my machine to redhat 7.2 (from redhat 6.2) and I am
> constantly (eg < 1 hour of uptime) getting kernel crashes which I believe
> are relating to ext3.  I know I should submit this to redhat's bugzilla, but
> before I do I was hoping someone on the list might want to comment as I am
> sure it is probably something I am doing wrong.
> 
> The machine has previously been rock solid, and the only hardware change I
> have made recently is to add an extra 64M of RAM (Total 128MB).  I have run
> memtest86 on it for 10 hours with no errors reported.
> 
> The last error I get is
> Assertion failure in __journal_file_buffer() at transaction.c:1953:
> "jh->b_jlist < 9".  I can still switch between consoles, but cannot do much
> else other than press the reset button :(
> 

ext3 downloads are currently running at 1,200 per day plus
an unknown number of Red Hat users, and you're the first to report
this one.   So it's going to be something odd.  It _could_ be bad
hardware, but if it's always failing in the same way, that sounds
unlikely.

Could you please force a `fsck' against the fs, let us know the
outcome?

Also, a ksymoops trace of the oops output would be most useful.

It looks like memory corruption of some form - a structure
member has an impossible value.  Are you using any less-than-mainstream
device drivers in that box?

-

Re: Assertion failure wth ext3 on standard Redhat 7.2 kernel

[Stephen Tweedie]

Hi,

On Fri, Nov 09, 2001 at 05:20:58PM +1100, Robert Lowery wrote:
 
> The last error I get is
> Assertion failure in __journal_file_buffer() at transaction.c:1953:

It's usually the _first_, not the last, error which is most revealing.
Did /var/log/messages capture anything more about the earlier errors?

Cheers,
 Stephen

Re: Assertion failure wth ext3 on standard Redhat 7.2 kernel

[Robert Lowery]

Hi Andrew,


>ext3 downloads are currently running at 1,200 per day plus
>an unknown number of Red Hat users, and you're the first to report
>this one. So it's going to be something odd. It _could_ be bad
>hardware, but if it's always failing in the same way, that sounds
>unlikely.

I too thought bad hardware.  I will try removing the extra 64M I added
recently and see if it still happens.


>Could you please force a `fsck' against the fs, let us know the
>outcome?
After a crash I say Y (within 5 seconds) on reboot to run an fsck and there
are usually corrupted files from what I was doing when it crashed.  Eg I was
running rpm -Uvh kernel-sources... and some files in /usr/src/linux... were
corrupted.

>Also, a ksymoops trace of the oops output would be most useful.

How do I do this when the box has crashed?  I can manually write down the
oops, but then what do I do?  Can I manually look up System.map to get what
you need?


>It looks like memory corruption of some form - a structure
>member has an impossible value. Are you using any less-than-mainstream
>device drivers in that box?
I agree.  Everything is pointing to the new memory, even though I
successfully ran memtest86 for 10 hours.

Thanks

-Robert

Re: Assertion failure wth ext3 on standard Redhat 7.2 kernel

[Robert Lowery]

> >It looks like memory corruption of some form - a structure
> >member has an impossible value. Are you using any less-than-mainstream
> >device drivers in that box?
P.S. The problem occurs on a completely virgin Redhat 7.2 install as well as
after I have applied all available updates. (after a few crashes and reboots
while applying them).

-Robert

Re: Assertion failure wth ext3 on standard Redhat 7.2 kernel

[Robert Lowery]

This problem has gone away now that I have removed the extra 64M RAM that I
added recently.

Funny that memtest86 did not pick up any problems.  Must be stray gamma rays
;)

Thanks

-Robert
----- Original Message -----
From: "Robert Lowery" <cangela@bigpond.net.au>
To: <akpm@zip.com.au>
Cc: <linux-kernel@vger.kernel.org>
Sent: Friday, November 09, 2001 10:45 PM
Subject: Re: Assertion failure wth ext3 on standard Redhat 7.2 kernel


> > >It looks like memory corruption of some form - a structure
> > >member has an impossible value. Are you using any less-than-mainstream
> > >device drivers in that box?
> P.S. The problem occurs on a completely virgin Redhat 7.2 install as well
as
> after I have applied all available updates. (after a few crashes and
reboots
> while applying them).
>
> -Robert
>