Return-Path: Received: from smtp1.linux-foundation.org ([140.211.169.13]:40764 "EHLO smtp1.linux-foundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753389Ab0LBBGi convert rfc822-to-8bit (ORCPT ); Wed, 1 Dec 2010 20:06:38 -0500 In-Reply-To: <20101201153608.78b331f9.akpm@linux-foundation.org> References: <1291217804-11257-1-git-send-email-Trond.Myklebust@netapp.com> <1291217804-11257-3-git-send-email-Trond.Myklebust@netapp.com> <1291217804-11257-4-git-send-email-Trond.Myklebust@netapp.com> <1291229669.6609.24.camel@heimdal.trondhjem.org> <1291234251.6609.39.camel@heimdal.trondhjem.org> <20101201123341.d12ef362.akpm@linux-foundation.org> <20101201133831.ea6ba10a.akpm@linux-foundation.org> <1291240272.6609.50.camel@heimdal.trondhjem.org> <20101201141351.8609140b.akpm@linux-foundation.org> <20101201143856.51f4f9d9.akpm@linux-foundation.org> <20101201153608.78b331f9.akpm@linux-foundation.org> From: Linus Torvalds Date: Wed, 1 Dec 2010 17:05:43 -0800 Message-ID: Subject: Re: [PATCH v2 3/3] NFS: Fix a memory leak in nfs_readdir To: Andrew Morton Cc: Trond Myklebust , Hugh Dickins , Nick Piggin , Nick Bowler , Linux Kernel Mailing List , linux-nfs@vger.kernel.org, Rik van Riel , Christoph Hellwig , Al Viro Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-nfs-owner@vger.kernel.org List-ID: MIME-Version: 1.0 On Wed, Dec 1, 2010 at 3:36 PM, Andrew Morton wrote: > On Wed, 1 Dec 2010 15:31:11 -0800 > Linus Torvalds wrote: >> >> Ergo: vmscan has a locked page leads to the filesystem being >> guaranteed to not be unmounted. ?And that, in turn, guarantees that >> the module won't be unloaded until the machine has gone through an >> idle cycle. > > The page isn't attached to the address_space any more: Did you even read the email? Here, let me quote the important parts: "module won't be unloaded until the machine has gone through an idle cycle" "This is pretty much how all the module races are handled. Doing module ref-counts per page (or per packet in flight for things like networking) would be prohibitively expensive." IOW, the whole "stop_machine()" part is fundamental. That whole "module unload won't happen until we've gone through an idle cycle" is EXACTLY why we don't need to have the page attached or ref-counted - we're still safe. Linus