Weird problem using latest kernel.

Weird problem using latest kernel.

[Kurt Roeckx]

I've been using 2.4.21-rc2, without patch, for a while without
problems.  Then I switched to 2.4.21-rc7 with the patch for rc6
found on http://www.coker.com.au/selinux/kern/

Using that, the first night something must have gone wrong using
updatedb, it was still running when I got back home that night,
which is ore than 12 hours after it starts.  You could hear the
hard disk going, the load was around 8, it was very unresponsive.
I just rebooted it.

The next day I had no problems, but the day after that I went to
see before I went to work and it was again still running.  This
time I started vmstat to have an idea of what's going on, and it
seems to be very heavily using the swap for some reason.

I killed updatedb and all it's children, but that didn't seem to
have helped, so I rebooted again.

After asking other people who run rc7, and searching list, nobody
else seems to have a problem, so I assumed it had to do something
with the lsm patch.  So I removed the patch and made a new
kernel.

I have no problems for 2 nights now with plain 2.4.21-rc7, didn't
have any problem with plain rc2 either, but things really don't
work very well with the lsm patched.


Kurt


--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.

Re: Weird problem using latest kernel.

[Russell Coker]

On Sun, 8 Jun 2003 21:46, Kurt Roeckx wrote:
> I've been using 2.4.21-rc2, without patch, for a while without
> problems.  Then I switched to 2.4.21-rc7 with the patch for rc6
> found on http://www.coker.com.au/selinux/kern/
>
> Using that, the first night something must have gone wrong using
> updatedb, it was still running when I got back home that night,
> which is ore than 12 hours after it starts.  You could hear the
> hard disk going, the load was around 8, it was very unresponsive.
> I just rebooted it.

I am having similar problems with 2.4.21-rc6 and also with rc3.

It seems that there is some sort of memory leak in the kernel.  After a while 
the machine starts thrashing even when there is nothing much happening.

One example is a machine running KDE3 and Mozilla became extremely slow.  It 
was then changed to run FVWM and Mozilla and was still extremely slow (even 
though FVWM uses much less memory than KDE).  I then rebooted the machine and 
ran KDE3 with Mozilla in the same manner as before, swap was not used at all!

I conclude that there is a memory leak in the kernel which requires a periodic 
reboot.

I don't think that this is a SE Linux related issue as my patch was a very 
simple forward-port of the 2.4.20 patch (which worked for a long time without 
such issues).  If you have evidence to the contrary then please describe it.

I believe that the bug is in the main kernel code.  I have not reported this 
to the linux-kernel list because I don't have enough hard evidence, and I 
expect that lots of other people have discovered the same bug.  If you choose 
to post to linux-kernel then please feel free to quote my message or provide 
the URL for the SE Linux list archives.

-- 
http://www.coker.com.au/selinux/   My NSA Security Enhanced Linux packages
http://www.coker.com.au/bonnie++/  Bonnie++ hard drive benchmark
http://www.coker.com.au/postal/    Postal SMTP/POP benchmark
http://www.coker.com.au/~russell/  My home page


--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.

Re: Weird problem using latest kernel.

[Kurt Roeckx]

On Sun, Jun 08, 2003 at 11:25:13PM +1000, Russell Coker wrote:
> On Sun, 8 Jun 2003 21:46, Kurt Roeckx wrote:
> > I've been using 2.4.21-rc2, without patch, for a while without
> > problems.  Then I switched to 2.4.21-rc7 with the patch for rc6
> > found on http://www.coker.com.au/selinux/kern/
> >
> > Using that, the first night something must have gone wrong using
> > updatedb, it was still running when I got back home that night,
> > which is ore than 12 hours after it starts.  You could hear the
> > hard disk going, the load was around 8, it was very unresponsive.
> > I just rebooted it.
> 
> I believe that the bug is in the main kernel code.  I have not reported this 
> to the linux-kernel list because I don't have enough hard evidence, and I 
> expect that lots of other people have discovered the same bug.  If you choose 
> to post to linux-kernel then please feel free to quote my message or provide 
> the URL for the SE Linux list archives.

Just for update, the current uptime is just over 4 days now
without any problems at all running on a plain 2.4.21-rc7.

I really believe the problem is in the patch.


Kurt


--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.