From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755180AbZAEVFY (ORCPT ); Mon, 5 Jan 2009 16:05:24 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1753271AbZAEVFL (ORCPT ); Mon, 5 Jan 2009 16:05:11 -0500 Received: from casper.infradead.org ([85.118.1.10]:32993 "EHLO casper.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753212AbZAEVFK (ORCPT ); Mon, 5 Jan 2009 16:05:10 -0500 Subject: Re: [patch] mm: fix lockless pagecache reordering bug (was Re: From: Peter Zijlstra To: paulmck@linux.vnet.ibm.com Cc: Linus Torvalds , Nick Piggin , Peter Klotz , stable@kernel.org, Linux Memory Management List , Christoph Hellwig , Roman Kononov , linux-kernel@vger.kernel.org, xfs@oss.sgi.com, Andrew Morton In-Reply-To: <20090105201258.GN6959@linux.vnet.ibm.com> References: <20081230042333.GC27679@wotan.suse.de> <20090103214443.GA6612@infradead.org> <20090105014821.GA367@wotan.suse.de> <20090105041959.GC367@wotan.suse.de> <20090105064838.GA5209@wotan.suse.de> <49623384.2070801@aon.at> <20090105164135.GC32675@wotan.suse.de> <20090105180008.GE32675@wotan.suse.de> <20090105201258.GN6959@linux.vnet.ibm.com> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Mon, 05 Jan 2009 22:04:51 +0100 Message-Id: <1231189491.11687.22.camel@twins> Mime-Version: 1.0 X-Mailer: Evolution 2.24.2 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, 2009-01-05 at 12:12 -0800, Paul E. McKenney wrote: > On Mon, Jan 05, 2009 at 10:44:27AM -0800, Linus Torvalds wrote: > > On Mon, 5 Jan 2009, Nick Piggin wrote: > > > On Mon, Jan 05, 2009 at 09:30:55AM -0800, Linus Torvalds wrote: > > > Putting an rcu_dereference there might work, but I think it misses a > > > subtlety of this code. > > > > No, _you_ miss the subtlety of something that can change under you. > > > > Look at radix_tree_deref_slot(), and realize that without the > > rcu_dereference(), the compiler would actually be allowed to think that it > > can re-load anything from *pslot several times. So without my one-liner > > patch, the compiler can actually do this: > > > > register = load_from_memory(pslot) > > if (radix_tree_is_indirect_ptr(register)) > > goto fail: > > return load_from_memory(pslot); > > > > fail: > > return RADIX_TREE_RETRY; > > My guess is that Nick believes that the value in *pslot cannot change > in such as way as to cause radix_tree_is_indirect_ptr()'s return value > to change within a given RCU grace period, and that Linus disagrees. Nick's belief would indeed be true IFF all modifying ops including all uses of radix_tree_replace_slot() are serialized wrt. each other. However, since radix_tree_deref_slot() is the counterpart of radix_tree_replace_slot(), one would indeed expect rcu_dereference() therein, much like Linus suggests. While what Nick says is true, the lifetime management of the data objects is arranged externally from the radix tree -- I still think we need the rcu_dereference() even for that argument, as we want to support RCU lifetime management as well. From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from cuda.sgi.com (cuda2.sgi.com [192.48.176.25]) by oss.sgi.com (8.12.11.20060308/8.12.11/SuSE Linux 0.7) with ESMTP id n05L58Mc019339 for ; Mon, 5 Jan 2009 15:05:08 -0600 Received: from casper.infradead.org (localhost [127.0.0.1]) by cuda.sgi.com (Spam Firewall) with ESMTP id 30C9C62AD6 for ; Mon, 5 Jan 2009 13:05:02 -0800 (PST) Received: from casper.infradead.org (casper.infradead.org [85.118.1.10]) by cuda.sgi.com with ESMTP id 6CzFDagGm3XErTF1 for ; Mon, 05 Jan 2009 13:05:02 -0800 (PST) Subject: Re: [patch] mm: fix lockless pagecache reordering bug (was Re: From: Peter Zijlstra In-Reply-To: <20090105201258.GN6959@linux.vnet.ibm.com> References: <20081230042333.GC27679@wotan.suse.de> <20090103214443.GA6612@infradead.org> <20090105014821.GA367@wotan.suse.de> <20090105041959.GC367@wotan.suse.de> <20090105064838.GA5209@wotan.suse.de> <49623384.2070801@aon.at> <20090105164135.GC32675@wotan.suse.de> <20090105180008.GE32675@wotan.suse.de> <20090105201258.GN6959@linux.vnet.ibm.com> Date: Mon, 05 Jan 2009 22:04:51 +0100 Message-Id: <1231189491.11687.22.camel@twins> Mime-Version: 1.0 List-Id: XFS Filesystem from SGI List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: xfs-bounces@oss.sgi.com Errors-To: xfs-bounces@oss.sgi.com To: paulmck@linux.vnet.ibm.com Cc: Nick Piggin , linux-kernel@vger.kernel.org, Roman Kononov , xfs@oss.sgi.com, Christoph Hellwig , Linux Memory Management List , Andrew Morton , Peter Klotz , Linus Torvalds , stable@kernel.org On Mon, 2009-01-05 at 12:12 -0800, Paul E. McKenney wrote: > On Mon, Jan 05, 2009 at 10:44:27AM -0800, Linus Torvalds wrote: > > On Mon, 5 Jan 2009, Nick Piggin wrote: > > > On Mon, Jan 05, 2009 at 09:30:55AM -0800, Linus Torvalds wrote: > > > Putting an rcu_dereference there might work, but I think it misses a > > > subtlety of this code. > > > > No, _you_ miss the subtlety of something that can change under you. > > > > Look at radix_tree_deref_slot(), and realize that without the > > rcu_dereference(), the compiler would actually be allowed to think that it > > can re-load anything from *pslot several times. So without my one-liner > > patch, the compiler can actually do this: > > > > register = load_from_memory(pslot) > > if (radix_tree_is_indirect_ptr(register)) > > goto fail: > > return load_from_memory(pslot); > > > > fail: > > return RADIX_TREE_RETRY; > > My guess is that Nick believes that the value in *pslot cannot change > in such as way as to cause radix_tree_is_indirect_ptr()'s return value > to change within a given RCU grace period, and that Linus disagrees. Nick's belief would indeed be true IFF all modifying ops including all uses of radix_tree_replace_slot() are serialized wrt. each other. However, since radix_tree_deref_slot() is the counterpart of radix_tree_replace_slot(), one would indeed expect rcu_dereference() therein, much like Linus suggests. While what Nick says is true, the lifetime management of the data objects is arranged externally from the radix tree -- I still think we need the rcu_dereference() even for that argument, as we want to support RCU lifetime management as well. _______________________________________________ xfs mailing list xfs@oss.sgi.com http://oss.sgi.com/mailman/listinfo/xfs From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail172.messagelabs.com (mail172.messagelabs.com [216.82.254.3]) by kanga.kvack.org (Postfix) with ESMTP id 34E3F6B00C0 for ; Mon, 5 Jan 2009 16:05:01 -0500 (EST) Subject: Re: [patch] mm: fix lockless pagecache reordering bug (was Re: From: Peter Zijlstra In-Reply-To: <20090105201258.GN6959@linux.vnet.ibm.com> References: <20081230042333.GC27679@wotan.suse.de> <20090103214443.GA6612@infradead.org> <20090105014821.GA367@wotan.suse.de> <20090105041959.GC367@wotan.suse.de> <20090105064838.GA5209@wotan.suse.de> <49623384.2070801@aon.at> <20090105164135.GC32675@wotan.suse.de> <20090105180008.GE32675@wotan.suse.de> <20090105201258.GN6959@linux.vnet.ibm.com> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Mon, 05 Jan 2009 22:04:51 +0100 Message-Id: <1231189491.11687.22.camel@twins> Mime-Version: 1.0 Sender: owner-linux-mm@kvack.org To: paulmck@linux.vnet.ibm.com Cc: Linus Torvalds , Nick Piggin , Peter Klotz , stable@kernel.org, Linux Memory Management List , Christoph Hellwig , Roman Kononov , linux-kernel@vger.kernel.org, xfs@oss.sgi.com, Andrew Morton List-ID: On Mon, 2009-01-05 at 12:12 -0800, Paul E. McKenney wrote: > On Mon, Jan 05, 2009 at 10:44:27AM -0800, Linus Torvalds wrote: > > On Mon, 5 Jan 2009, Nick Piggin wrote: > > > On Mon, Jan 05, 2009 at 09:30:55AM -0800, Linus Torvalds wrote: > > > Putting an rcu_dereference there might work, but I think it misses a > > > subtlety of this code. > > > > No, _you_ miss the subtlety of something that can change under you. > > > > Look at radix_tree_deref_slot(), and realize that without the > > rcu_dereference(), the compiler would actually be allowed to think that it > > can re-load anything from *pslot several times. So without my one-liner > > patch, the compiler can actually do this: > > > > register = load_from_memory(pslot) > > if (radix_tree_is_indirect_ptr(register)) > > goto fail: > > return load_from_memory(pslot); > > > > fail: > > return RADIX_TREE_RETRY; > > My guess is that Nick believes that the value in *pslot cannot change > in such as way as to cause radix_tree_is_indirect_ptr()'s return value > to change within a given RCU grace period, and that Linus disagrees. Nick's belief would indeed be true IFF all modifying ops including all uses of radix_tree_replace_slot() are serialized wrt. each other. However, since radix_tree_deref_slot() is the counterpart of radix_tree_replace_slot(), one would indeed expect rcu_dereference() therein, much like Linus suggests. While what Nick says is true, the lifetime management of the data objects is arranged externally from the radix tree -- I still think we need the rcu_dereference() even for that argument, as we want to support RCU lifetime management as well. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org