10.1-BETA2 possible kernel memory leak in routing table

[Alexander V. Chernikov]

On 01.10.2014 22:49, Rumen Telbizov wrote:
> Submitted PR with details at 
> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=194078
>
>
> On Wed, Oct 1, 2014 at 10:16 AM, Gleb Smirnoff <glebius at freebsd.org 
> <mailto:glebius at freebsd.org>> wrote:
>
>     On Wed, Oct 01, 2014 at 11:42:15AM -0400, Mike Tancsa wrote:
>     M> On 10/1/2014 9:51 AM, Gleb Smirnoff wrote:
>     M> > On Tue, Sep 30, 2014 at 04:56:00PM -0700, Rumen Telbizov wrote:
>     M> > R> Brian Somers and I are currently looking into the source
>     of PF in latest
>     M> > R> 10-STABLE and trying to figure out what is going on. We
>     were able to
>     M> > R> replicate this problem on a 11-CURRENT (Sep 12th) machine
>     as well. A simple
>     M> > R> PF ruleset with 1 rule and 1 table. Every few reloads of
>     the firewall
>     M> > R> and vmstat
>     M> > R> -m | grep routetbl shows increased memory usage.
>     M> >
>     M> > I plugged the easy leak, but there is also a hard one.
>     Actually, the
>     M> > entire pf_table.c needs a good shake. Right now I am out of
>     time for this.
>     M>
>     M>      Is that easy fix
>     M>
>     M>
>     http://lists.freebsd.org/pipermail/svn-src-head/2014-October/063178.html
>
>     Yes, it seems the leak slowed down.
>
>     M> Also, is there any work around to this ?  I tried a simple set
>     of pf
>     M> rules with no tables, hoping that was the cause of it, but
>     memory grows
>     M> with each pf reload.
>
>     No workaround available. Can you please file a PR for that? Once I
>     have
>     time, I will work on this.
>
Remaining leak is not related to pf.
It happens due to rn_detachhead() not properly freeing items inside it 
masks tree.
I'll try to fix this soon.
>
>
>     --
>     Totus tuus, Glebius.
>
>
>
>
> -- 
> Rumen Telbizov
> Unix Systems Administrator <http://telbizov.com>