Return-Path: Received: from smtp-out.google.com ([74.125.121.35]:37579 "EHLO smtp-out.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756745Ab0LBDeZ (ORCPT ); Wed, 1 Dec 2010 22:34:25 -0500 Received: from kpbe19.cbf.corp.google.com (kpbe19.cbf.corp.google.com [172.25.105.83]) by smtp-out.google.com with ESMTP id oB23YN37009067 for ; Wed, 1 Dec 2010 19:34:23 -0800 Received: from pvc22 (pvc22.prod.google.com [10.241.209.150]) by kpbe19.cbf.corp.google.com with ESMTP id oB23YLir013160 for ; Wed, 1 Dec 2010 19:34:22 -0800 Received: by pvc22 with SMTP id 22so4585307pvc.27 for ; Wed, 01 Dec 2010 19:34:21 -0800 (PST) Date: Wed, 1 Dec 2010 19:34:11 -0800 (PST) From: Hugh Dickins To: Trond Myklebust cc: Linus Torvalds , Nick Bowler , Linux Kernel Mailing List , linux-nfs@vger.kernel.org, Andrew Morton , Nick Piggin , Rik van Riel , Christoph Hellwig , Al Viro Subject: Re: [PATCH v3 2/3] Call the filesystem back whenever a page is removed from the page cache In-Reply-To: <1291259285-26803-3-git-send-email-Trond.Myklebust@netapp.com> Message-ID: References: <1291259285-26803-1-git-send-email-Trond.Myklebust@netapp.com> <1291259285-26803-2-git-send-email-Trond.Myklebust@netapp.com> <1291259285-26803-3-git-send-email-Trond.Myklebust@netapp.com> Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-nfs-owner@vger.kernel.org List-ID: MIME-Version: 1.0 On Wed, 1 Dec 2010, Trond Myklebust wrote: > > --- a/mm/truncate.c > +++ b/mm/truncate.c > @@ -108,6 +108,10 @@ truncate_complete_page(struct address_space *mapping, struct page *page) > clear_page_mlock(page); > remove_from_page_cache(page); > ClearPageMappedToDisk(page); > + > + if (mapping->a_ops->freepage) > + mapping->a_ops->freepage(page); > + > page_cache_release(page); /* pagecache ref */ > return 0; > } I think Linus recommended that one be done in remove_from_page_cache() to catch all instances: did that get overruled later for some reason? > --- a/mm/vmscan.c > +++ b/mm/vmscan.c > @@ -454,6 +454,7 @@ static int __remove_mapping(struct address_space *mapping, struct page *page) > BUG_ON(!PageLocked(page)); > BUG_ON(mapping != page_mapping(page)); > > + preempt_disable(); > spin_lock_irq(&mapping->tree_lock); > /* > * The non racy check for a busy page. > @@ -492,10 +493,19 @@ static int __remove_mapping(struct address_space *mapping, struct page *page) > swp_entry_t swap = { .val = page_private(page) }; > __delete_from_swap_cache(page); > spin_unlock_irq(&mapping->tree_lock); > + preempt_enable(); > swapcache_free(swap, page); > } else { > + void (*freepage)(struct page *); > + > + freepage = mapping->a_ops->freepage; > + > __remove_from_page_cache(page); > spin_unlock_irq(&mapping->tree_lock); > + if (freepage != NULL) > + freepage(page); > + preempt_enable(); > + > mem_cgroup_uncharge_cache_page(page); > } > > @@ -503,6 +513,7 @@ static int __remove_mapping(struct address_space *mapping, struct page *page) > > cannot_free: > spin_unlock_irq(&mapping->tree_lock); > + preempt_enable(); > return 0; > } I took his "stop_machine()" explanation ("an idle period for everything. And just a preemption reschedule isn't enough for that") to imply that there's no need for your preempt_disable/preempt_enable there: they don't add anything to the module unload case, and they don't help the spin_unlock_irq issue (and you're already being rightly careful to note freepage in advance). But maybe I misunderstood. Hugh