[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ipfilter
Subject:    Re: FreeBSD 2.2.2 / ipf 3.2alpha9: page fault
From:       weave () navi ! net (Terrance Hodgins)
Date:       1997-07-25 18:02:55
[Download RAW message or body]

RE:
>Hello,
>
>when I try to delete the current NAT entries using "ipnat -F" I get the
>message
>"page fault while in kernel mode" and some additional informations; then
>the system reboots.
>
>Any solution? Thanks in advance.
>
>
>
>Stefan Witzel                                 switzel@uni-goettingen.de
>Universitaet Goettingen / Stabsstelle DV      -------------------------
>Gosslerstrasse 5-7                            fon: +49 551 394160
>37073 Goettingen                              fax: +49 551 399612
>Germany       
>
>
I don't have an answer, but can verify the problem.  I have been
running ip-filter since 3.2alpha4, and have had the same problem
with all of the following versions. I'm running Solaris 2.4 on
a Sparcstation 1.  "ipnat -C" works, but "ipnat -C -F" or just
"ipnat -F" is almost guaranteed to crash the system: "BAD TRAP,
Bad data", and the system panics and dumps core and reboots.

Interestingly, I have seen it actually work a couple of times,
but it is about 95% certain to crash the system.  So I have
just been refraining from using that command.

Which leads to a problem I ran into yesterday: Every time I get
disconnected, I am left with a list of dead connections that may
take days to finally get cleared out.  "ipnat -l" shows them.
Yesterday, the list got long enough to apparently break ipnat
or ipfilter. The system would just suddenly stop receiving.
Not even pinging my ISP would receive an answer.  So I tried
ipnat -F, and of course, crash and reboot.  But then the system
went back to working properly, as we started over with a nice,
clean, empty list.

By the way, haven't tested that on version 3.2beta1 yet, because
beta1 isn't working at all well for me.  It will run anywhere from
5 minutes to 45 minutes, and then the whole system freezes up,
silently, without any panics or core dumps.  But it is so dead
that it won't even respond to L1-a (or "Stop-a", depending on
your Sun keyboard labelling). Nothing but power-cycling the machine
will bring it back to life.  So I went back to version beta8,
and it's okay, as long as I don't use "ipnat -F".

*         Terrance Hodgins          *
*      Willamette Web Weavers       *
*          weave@navi.net           *
*    http://www.navi.net/~weave/    *


[prev in list] [next in list] [prev in thread] [next in thread]