From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757348AbXIRGjR (ORCPT ); Tue, 18 Sep 2007 02:39:17 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754819AbXIRGjD (ORCPT ); Tue, 18 Sep 2007 02:39:03 -0400 Received: from sacred.ru ([62.205.161.221]:55232 "EHLO sacred.ru" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754077AbXIRGjB (ORCPT ); Tue, 18 Sep 2007 02:39:01 -0400 Message-ID: <46EF71F0.9010606@openvz.org> Date: Tue, 18 Sep 2007 10:36:32 +0400 From: Pavel Emelyanov User-Agent: Thunderbird 2.0.0.6 (X11/20070728) MIME-Version: 1.0 To: "J. Bruce Fields" CC: Linux Kernel Mailing List , devel@openvz.org Subject: Re: [PATCH] Wake up mandatory locks waiter on chmod References: <46E94993.3090506@openvz.org> <20070916194135.GC31801@fieldses.org> <46EE20C4.1060405@openvz.org> <20070917145934.GA4957@fieldses.org> In-Reply-To: <20070917145934.GA4957@fieldses.org> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Greylist: Sender succeeded SMTP AUTH authentication, not delayed by milter-greylist-3.0 (sacred.ru [62.205.161.221]); Tue, 18 Sep 2007 10:38:51 +0400 (MSD) Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org J. Bruce Fields wrote: > On Mon, Sep 17, 2007 at 10:37:56AM +0400, Pavel Emelyanov wrote: >> J. Bruce Fields wrote: >>> Is there a small chance that a lock may be applied after this check: >>> >>>> + mandatory = (inode->i_flock && MANDATORY_LOCK(inode)); >>>> + >>> but early enough that someone can still block on the lock while the file >>> is still marked for mandatory locking? (And is the inode->i_flock check >>> there really necessary?) >> There is, but as you have noticed: > > OK, but why not just remove the inode->i_flock check there? I can't see > how it helps anyway. > >>> Well, there are probably worse races in the mandatory locking code. >> ...there are. The inode->i_lock is protected with lock_kernel() only >> and is not in sync with any other checks for inodes. This is sad :( >> but a good locking for locks is to be done... > > I would also prefer a locking scheme that didn't rely on the BKL. That > said, except for this race: I would as well :) But I don't know the locking code good enough to start fixing. Besides, even if I send a patch series that handles this, I don't think that anyone will accept it, due to "this changes too much code", "can you prove you fixed all the places" and so on... >>> (For example, my impression is that a mandatory lock can be applied just >>> after the locks_mandatory_area() checks but before the io actually >>> completes.) > > ... I'm not aware of other races in the existing file-locking code. It > sounds like you might be. Could you give specific examples? Well, there's a long standing BUG in leases code - when we made all the checks in inserting lease, we call the locks_alloc_lock() and may fall asleep. Bu after the wakeup nobody re-checks for the things to change. I suspect there are other bad places. > --b. >