From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 68E16C38145 for ; Thu, 8 Sep 2022 10:58:11 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229548AbiIHK6J (ORCPT ); Thu, 8 Sep 2022 06:58:09 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:53920 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229547AbiIHK6C (ORCPT ); Thu, 8 Sep 2022 06:58:02 -0400 Received: from dfw.source.kernel.org (dfw.source.kernel.org [139.178.84.217]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 134805FF68 for ; Thu, 8 Sep 2022 03:58:01 -0700 (PDT) Received: from smtp.kernel.org (relay.kernel.org [52.25.139.140]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dfw.source.kernel.org (Postfix) with ESMTPS id 94F1E61C3F for ; Thu, 8 Sep 2022 10:58:00 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id BBD07C433C1; Thu, 8 Sep 2022 10:57:57 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1662634680; bh=65Xz5LWNy5zeYP5z8NNG7GUzsoo/VQzp/ROjifsWCzA=; h=Date:Subject:To:Cc:References:From:In-Reply-To:From; b=IDwnaXP5fIZ7Xf2XT8Fzh2nuWq0c+kJgnV0PD6vR/DARLCQHayEe3RV59afe+g5j4 NKJiDl02U3NLIa3mW0IMjmy5fSFaPprefHC7eoLUpV/F1bvVOJDSHguVe+R96JJKMo 6b/Q53Bfx223ugJxWarNRaghJhoYxMc/c00ZXNt6hnVPd5eCnuTbfWP+aAJySgNIk8 VrU7AwwtSBqi/VzhT9UEVe25H9qh90ADobKaZ4tphMA0vusU9QbmIF6qNfxq9autYD XefVFUBwlKOf6dJVhiodNtgXBVXh5OugML9pviFrXOgGbQKVUA+THOkoahhc0Z0r80 x7AZpWqV2XVng== Message-ID: Date: Thu, 8 Sep 2022 18:57:53 +0800 MIME-Version: 1.0 User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:91.0) Gecko/20100101 Thunderbird/91.11.0 Subject: Re: [PATCH v2 -next] f2fs: fix wrong dirty page count when race between mmap and fallocate. Content-Language: en-US To: Shuqi Zhang , jaegeuk@kernel.org Cc: linux-f2fs-devel@lists.sourceforge.net, linux-kernel@vger.kernel.org References: <20220831022440.2985736-1-zhangshuqi3@huawei.com> From: Chao Yu In-Reply-To: <20220831022440.2985736-1-zhangshuqi3@huawei.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 2022/8/31 10:24, Shuqi Zhang wrote: > This is a BUG_ON issue as follows when running xfstest-generic-503: > WARNING: CPU: 21 PID: 1385 at fs/f2fs/inode.c:762 f2fs_evict_inode+0x847/0xaa0 > Modules linked in: > CPU: 21 PID: 1385 Comm: umount Not tainted 5.19.0-rc5+ #73 > Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-4.fc34 04/01/2014 > > Call Trace: > evict+0x129/0x2d0 > dispose_list+0x4f/0xb0 > evict_inodes+0x204/0x230 > generic_shutdown_super+0x5b/0x1e0 > kill_block_super+0x29/0x80 > kill_f2fs_super+0xe6/0x140 > deactivate_locked_super+0x44/0xc0 > deactivate_super+0x79/0x90 > cleanup_mnt+0x114/0x1a0 > __cleanup_mnt+0x16/0x20 > task_work_run+0x98/0x100 > exit_to_user_mode_prepare+0x3d0/0x3e0 > syscall_exit_to_user_mode+0x12/0x30 > do_syscall_64+0x42/0x80 > entry_SYSCALL_64_after_hwframe+0x46/0xb0 > > Function flow analysis when BUG occurs: > f2fs_fallocate mmap > do_page_fault > pte_spinlock // ---lock_pte > do_wp_page > wp_page_shared > pte_unmap_unlock // unlock_pte > do_page_mkwrite > f2fs_vm_page_mkwrite > down_read(invalidate_lock) > lock_page > if (PageMappedToDisk(page)) > goto out; > // set_page_dirty --NOT RUN > out: up_read(invalidate_lock); > finish_mkwrite_fault // unlock_pte > f2fs_collapse_range > down_write(i_mmap_sem) > truncate_pagecache > unmap_mapping_pages > i_mmap_lock_write // down_write(i_mmap_rwsem) > ...... > zap_pte_range > pte_offset_map_lock // ---lock_pte > set_page_dirty > f2fs_dirty_data_folio > if (!folio_test_dirty(folio)) { > fault_dirty_shared_page > set_page_dirty > f2fs_dirty_data_folio > if (!folio_test_dirty(folio)) { > filemap_dirty_folio > f2fs_update_dirty_folio // ++ > } > unlock_page > filemap_dirty_folio > f2fs_update_dirty_folio // page count++ > } > pte_unmap_unlock // --unlock_pte > i_mmap_unlock_write // up_write(i_mmap_rwsem) > truncate_inode_pages > up_write(i_mmap_sem) Will this race condition affact f2fs private field update? Thanks, > > When race happens between mmap-do_page_fault-wp_page_shared and > fallocate-truncate_pagecache-zap_pte_range, the zap_pte_range calls > function set_page_dirty without page lock. Besides, though > truncate_pagecache has immap and pte lock, wp_page_shared calls > fault_dirty_shared_page without any. In this case, two threads race > in f2fs_dirty_data_folio function. Page is set to dirty only ONCE, > but the count is added TWICE by calling filemap_dirty_folio. > Thus the count of dirty page cannot accord with the real dirty pages. > > Following is the solution to in case of race happens without any lock. > Since folio_test_set_dirty in filemap_dirty_folio is atomic, judge return > value will not be at risk of race. > > Signed-off-by: Shuqi Zhang > --- > fs/f2fs/checkpoint.c | 3 +-- > fs/f2fs/data.c | 3 +-- > fs/f2fs/node.c | 3 +-- > 3 files changed, 3 insertions(+), 6 deletions(-) > > diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c > index 8259e0fa97e1..3a3329bf1033 100644 > --- a/fs/f2fs/checkpoint.c > +++ b/fs/f2fs/checkpoint.c > @@ -448,8 +448,7 @@ static bool f2fs_dirty_meta_folio(struct address_space *mapping, > > if (!folio_test_uptodate(folio)) > folio_mark_uptodate(folio); > - if (!folio_test_dirty(folio)) { > - filemap_dirty_folio(mapping, folio); > + if (filemap_dirty_folio(mapping, folio)) { > inc_page_count(F2FS_M_SB(mapping), F2FS_DIRTY_META); > set_page_private_reference(&folio->page); > return true; > diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c > index aa3ccddfa037..16c39bd948a0 100644 > --- a/fs/f2fs/data.c > +++ b/fs/f2fs/data.c > @@ -3697,8 +3697,7 @@ static bool f2fs_dirty_data_folio(struct address_space *mapping, > folio_mark_uptodate(folio); > BUG_ON(folio_test_swapcache(folio)); > > - if (!folio_test_dirty(folio)) { > - filemap_dirty_folio(mapping, folio); > + if (filemap_dirty_folio(mapping, folio)) { > f2fs_update_dirty_folio(inode, folio); > return true; > } > diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c > index e06a0c478b39..3506ffcb31f8 100644 > --- a/fs/f2fs/node.c > +++ b/fs/f2fs/node.c > @@ -2147,8 +2147,7 @@ static bool f2fs_dirty_node_folio(struct address_space *mapping, > if (IS_INODE(&folio->page)) > f2fs_inode_chksum_set(F2FS_M_SB(mapping), &folio->page); > #endif > - if (!folio_test_dirty(folio)) { > - filemap_dirty_folio(mapping, folio); > + if (filemap_dirty_folio(mapping, folio)) { > inc_page_count(F2FS_M_SB(mapping), F2FS_DIRTY_NODES); > set_page_private_reference(&folio->page); > return true;