Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 16 Jan 2009 16:50:30 -0800
From:      "Jason DiCioccio" <jd@ods.org>
To:        vwe@freebsd.org
Cc:        freebsd-net@freebsd.org, freebsd-bugs@freebsd.org, bugs-followup@freebsd.org
Subject:   Re: kern/130059: [panic] Leaking 50k mbufs/hour
Message-ID:  <f372a76b0901161650r3bbbad25raf1fedfb24d68063@mail.gmail.com>
In-Reply-To: <f372a76b0901021301k4a13d7aem980c3bc591bd44c4@mail.gmail.com>
References:  <200812311353.mBVDraLJ042040@freefall.freebsd.org> <f372a76b0812311041t35762cd2nd9b0bb7634001054@mail.gmail.com> <f372a76b0901021301k4a13d7aem980c3bc591bd44c4@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
I narrowed this bug down to the following patch to djb's ucspi-tcp
(adds ipv6 functionality):
http://www.fefe.de/ucspi/

I don't think that userland processes should be able to wreak that
much havoc on the network stack.  Another thing of note is that even
if I kill the processes causing the problem, the mbufs are never
reclaimed.  Seems like a permanent leak.

When I got rid of the ipv6 patch, the mbufs stopped building up and
everything has been fine..  Note that the ipv6 traffic for this
process was fairly minimal.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?f372a76b0901161650r3bbbad25raf1fedfb24d68063>