Spontaneous reboot after change ip_pkt_list_tot - deb

Spontaneous reboot after change ip_pkt_list_tot - deb

[M. Arts - Rootnet]

I'm working on some virtual servers running debian with kernel version:
2.6.18.8.xs5.5.0.13.442.

iptables version v1.3.6.

I've been running some tests on an test server which frankly doesn't get
much traffic. I've ran the recent module with a modified maximum
hitcount which has been running without problems for several weeks.
The file I edited for this
was /sys/module/ipt_recent/parameters/ip_pkt_list_tot


However once I changed it on a live environment it soon rebooted itself
without any real indication as to why. Even changing it at all resulted
in the same problem. Now I couldn't find much of a reason for this so
I'm hoping one of you could fill me in on this or could give me a
recommendation or tip.

I've tried looking up info but frankly couldn't find any real
information other than people recommending to change it, which at this
point seems like a curious recommendation.

Thanks in advance.

Re: Spontaneous reboot after change ip_pkt_list_tot - deb

[Patrick McHardy]

M. Arts - Rootnet wrote:
> I'm working on some virtual servers running debian with kernel version:
> 2.6.18.8.xs5.5.0.13.442.
> 
> iptables version v1.3.6.
> 
> I've been running some tests on an test server which frankly doesn't get
> much traffic. I've ran the recent module with a modified maximum
> hitcount which has been running without problems for several weeks.
> The file I edited for this
> was /sys/module/ipt_recent/parameters/ip_pkt_list_tot
> 
> 
> However once I changed it on a live environment it soon rebooted itself
> without any real indication as to why. Even changing it at all resulted
> in the same problem. Now I couldn't find much of a reason for this so
> I'm hoping one of you could fill me in on this or could give me a
> recommendation or tip.
> 
> I've tried looking up info but frankly couldn't find any real
> information other than people recommending to change it, which at this
> point seems like a curious recommendation.

There's a NULL pointer dereference in the seq file handling in
that version which can happen under memory pressure.

If its reproducable every time, thats unlikely to be the cause
though, in that case please try to capture an oops message (or
try a more recent version).