From mboxrd@z Thu Jan 1 00:00:00 1970 From: Pablo Neira Ayuso Subject: Re: [PATCH 0/4] conntrack event subsystem updates for 2.6.31 Date: Wed, 06 May 2009 15:36:06 +0200 Message-ID: <4A019246.4070001@netfilter.org> References: <20090502141522.8067.436.stgit@Decadence> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Cc: netfilter-devel@vger.kernel.org To: kaber@trash.net Return-path: Received: from mail.us.es ([193.147.175.20]:48080 "EHLO mail.us.es" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754540AbZEFNg0 (ORCPT ); Wed, 6 May 2009 09:36:26 -0400 In-Reply-To: <20090502141522.8067.436.stgit@Decadence> Sender: netfilter-devel-owner@vger.kernel.org List-ID: Pablo Neira Ayuso wrote: > Hi Patrick, > > This is the first bunch of patches oriented to the event caching > system, I still have a couple more here but I'm still getting them > into shape. I noticed a "scheduling while atomic" problem in the event caching "replace notify chain by function pointer" patch. It seems that gfp_any(), which is used by nfnetlink_send(), returns GFP_KERNEL inside a RCU read-side lock section, that's invalid. Moreover, this triggers a backtrace when using very small buffers and making lots of requests from user-context (we hit schedule() due to __GFP_WAIT in the netlink code). Patrick, just to let you know in case that you look at these patches. I have fixed this here. I'll resend these patches once nf-next-2.6 is open. -- "Los honestos son inadaptados sociales" -- Les Luthiers