Re: Oops in kswapd, 2.4.19 kernel and before

From: Hugo Mills (hugo-lkml@carfax.org.uk)
Date: Mon Oct 28 2002 - 11:45:40 EST


On Mon, Oct 28, 2002 at 01:29:01PM +0100, Andrea Arcangeli wrote:
> On Mon, Oct 28, 2002 at 10:24:39AM +0000, Hugo Mills wrote:
> > I'm getting regular oopsen in kswapd on my 2.4.19 kernel. They
> > generally appear to happen while running Amanda (a tape backup
>
> if it only happens while or after running Amanda, it may be a tape
> driver bug.

   I may have seen it (once?) before without touching the tape drive,
although I'm not certain. I shall see if I can reproduce without use
of the tape.

> > Decoded oopsen are below (they _are_ decoded with the right system
> > maps, despite ksymoops's concerns). If there's anything else that's
> > needed in order to track this down, please let me know.
>
> the oopses shows some inode was corrupted, it doesn't tell us who is
> corrupting them but most likely it is not a piece of common code (a driver
> or a non mainstream feature or we should be able to reproduce it) You
> should try to localize the bug to a piece of code, by for example making
> 100% sure that it triggers as soon as you start amanda.

   It's not certain. I appear to have triggered it this morning on the
_third_ consecutive run of amflush. Again, I'll test more carefully.

> Then you can try to backup using another device (not tape) and see
> if you can still reproduce. finally you can try to use older or
> newer 2.4 drivers for the tape and see if there's any change that
> fixes the problem in the old/new drivers. Of course it isn't certain
> at all that it is the tape, I'm just guessing because you said it
> happens while backing up to the tape.

   I've definitely seen the problem throughout the 2.4 series. I don't
recall what the first 2.4 kernel I used was, but it was definitely
there in all mainstream kernels (and those -ac kernels I tried) from
about 2.4.14 onwards. I'll try 2.4.20-preX and report on that as well.

   Thanks for your help. It may be a week or two before I can get all
these tests completed, but I shall definitely report back when I'm
done.

   Hugo.

-- 
=== Hugo Mills: hugo@... carfax.org.uk | darksatanic.net | lug.org.uk ===
 PGP: 1024D/1C335860 from wwwkeys.eu.pgp.net or www.carfax.nildram.co.uk
   --- Anyone who claims their cryptographic protocol is secure is ---   
         either a genius or a fool.  Given the genius/fool ratio         
                 for our species,  the odds aren't good.                 


- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Thu Oct 31 2002 - 22:00:37 EST