From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932222AbXBNLjB (ORCPT ); Wed, 14 Feb 2007 06:39:01 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S932224AbXBNLjB (ORCPT ); Wed, 14 Feb 2007 06:39:01 -0500 Received: from rere.qmqm.pl ([86.63.132.164]:57732 "EHLO rere.qmqm.pl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932222AbXBNLjA (ORCPT ); Wed, 14 Feb 2007 06:39:00 -0500 Date: Wed, 14 Feb 2007 12:38:41 +0100 From: =?iso-8859-2?Q?Micha=B3_Miros=B3aw?= To: netfilter-devel@lists.netfilter.org Cc: linux-kernel@vger.kernel.org Subject: Re: [PATCH 2.6.20 13/14] nfnetlink_log: fix reference counting Message-ID: <20070214113841.GA3005@rere.qmqm.pl> References: <20070212003738.GA8262@rere.qmqm.pl> <20070212202052.GA28704@rere.qmqm.pl> <20070212202255.GD28704@rere.qmqm.pl> <45D1B5FA.5050704@trash.net> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <45D1B5FA.5050704@trash.net> User-Agent: Mutt/1.5.9i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Tue, Feb 13, 2007 at 01:58:34PM +0100, Patrick McHardy wrote: > Micha³ Miros³aw wrote: > > Fix reference counting (memory leak) problem in __nfulnl_send() and callers > > related to packet queueing. > > > > Signed-off-by: Michał Mirosław > > > > --- linux-2.6.20/net/netfilter/nfnetlink_log.c.11 2007-02-12 17:35:50.000000000 +0100 > > +++ linux-2.6.20/net/netfilter/nfnetlink_log.c 2007-02-12 17:58:01.000000000 +0100 > > @@ -223,6 +223,11 @@ _instance_destroy2(struct nfulnl_instanc > > > > spin_lock_bh(&inst->lock); > > if (inst->skb) { > > + /* timer "holds" one reference (we have one more) */ > > + if (timer_pending(&inst->timer)) { > > + del_timer(&inst->timer); > > + instance_put(inst); > This should be done outside of the locked section and using > del_timer_sync to make sure the timer is not already active > and waiting for the lock. I found another solution, as this won't work (I explained the case in my reply to your comment on 07/10). I noticed that we should just check what del_timer() returns instead of timer_pending(). Heres the patch: Fix reference counting (memory leak) problem in __nfulnl_send() and callers related to packet queueing. Signed-off-by: Michał Mirosław --- linux-2.6.20/net/netfilter/nfnetlink_log.c.11 2007-02-12 17:35:50.000000000 +0100 +++ linux-2.6.20/net/netfilter/nfnetlink_log.c.12 2007-02-14 12:27:09.000000000 +0100 @@ -223,6 +223,9 @@ _instance_destroy2(struct nfulnl_instanc spin_lock_bh(&inst->lock); if (inst->skb) { + /* timer "holds" one reference (we have one more) */ + if (del_timer(&inst->timer)) + instance_put(inst); if (inst->qlen) __nfulnl_send(inst); if (inst->skb) { @@ -370,9 +373,6 @@ __nfulnl_send(struct nfulnl_instance *in { int status; - if (timer_pending(&inst->timer)) - del_timer(&inst->timer); - if (!inst->skb) return 0; @@ -683,6 +683,9 @@ nfulnl_log_packet(unsigned int pf, * enough room in the skb left. flush to userspace. */ UDEBUG("flushing old skb\n"); + /* timer "holds" one reference (we have another one) */ + if (del_timer(&inst->timer)) + instance_put(inst); __nfulnl_send(inst); }