Date: Sat, 22 Oct 2005 20:47:16 +0200 From: Stefan Ehmann <shoesoft@gmx.net> To: Gleb Smirnoff <glebius@FreeBSD.org> Cc: current@FreeBSD.org Subject: ipfw uid rules cause lockup (was Re: PREEMPTION still unusable with 6.0-RC1) Message-ID: <1130006836.50495.13.camel@taxman.pepperland> In-Reply-To: <20051022154537.GL59364@cell.sick.ru> References: <1129879049.771.6.camel@localhost> <20051022154537.GL59364@cell.sick.ru>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 2005-10-22 at 19:45 +0400, Gleb Smirnoff wrote: > On Fri, Oct 21, 2005 at 09:17:29AM +0200, Stefan Ehmann wrote: > S> I recently upgraded my 5.4-RELEASE machine to 6.0-RC1. Runs fine so far > S> if I disable PREEMPTION. > S> > S> With PREEMPTION enabled, I still get the same issues as described here: > S> http://lists.freebsd.org/pipermail/freebsd-current/2004-September/037949.html > S> > S> kernel config: > S> http://stud4.tuwien.ac.at/~e0125637/fbsd/kernconf-6.0-RC1 > S> > S> /var/log/messages output: > S> http://stud4.tuwien.ac.at/~e0125637/fbsd/messages-6.0-RC1 > S> > S> Any one else still experiencing these problems or has any idea what > S> causes this? > S> > S> BTW, are 6.0 discussions supposed to go to stable@ once 6.0 is released? > > AFAIK, the uid/gid rules in ipfw(8) require debug.mpsafenet=0. So > your deadlock isn't a surprise. I wonder why this limitation isn't > documented. A quick google search revealed http://www.freebsd.org/releases/5.3R/errata.html (31 Oct 2004) When the user/group rule clauses in pf(4) and ipfw(4) are used, the loader tunable debug.mpsafenet must be set to 0 (this is 1 by default). But this is supposed to be fixed in 5.4 (OTOH 5.4 doesn't have PREEMPTION enabled by default). I'll try tomorrow if debug.mpsafenet=0 will help.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1130006836.50495.13.camel>