From patchwork Wed Oct 3 18:43:27 2012 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Dmitry Monakhov X-Patchwork-Id: 188896 Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id 1B6EE2C00B1 for ; Thu, 4 Oct 2012 06:06:00 +1000 (EST) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964998Ab2JCSnc (ORCPT ); Wed, 3 Oct 2012 14:43:32 -0400 Received: from mail-lb0-f174.google.com ([209.85.217.174]:48210 "EHLO mail-lb0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S964997Ab2JCSnb (ORCPT ); Wed, 3 Oct 2012 14:43:31 -0400 Received: by lbon3 with SMTP id n3so6499992lbo.19 for ; Wed, 03 Oct 2012 11:43:30 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=sender:from:to:cc:subject:date:message-id:x-mailer; bh=r/4IVzYqxk2XYQXL43XABJHjHtn25HicfJJZf4JN82Q=; b=Dh1mhK/EUcFMFZQZyQoIX1Nen8uDZY6G9MZRf2kfaJdeFx5hOcKeJu2Qb1gY4JZ8eQ hS/gS5von0muCk85/Gl1TKK73pIqmfD+vfQju7y9c/yTdw4gl+qrFtgVmtJUZTDphnZE /1rGJo13UO7vp5KLeGA4Va5C39+93g4lSvt9wf3fk8CTjssWS97f6WzzOIHe2jcfhQNU lGGQAK4nYlSEFd1I0Xg9ORpnc2nRUJaR/Xo3dyNluFq4wfloWDt6IV6lW+CfjFtbxYFm aZxOv46G+qcdHu/EdljMHLeFjP6z5NX4erTkQUUDvetm2UUDQd7WticGQoSj9t+XoZV1 xE9Q== Received: by 10.152.105.236 with SMTP id gp12mr2406815lab.35.1349289810176; Wed, 03 Oct 2012 11:43:30 -0700 (PDT) Received: from smtp.gmail.com (swsoft-msk-nat.sw.ru. [195.214.232.10]) by mx.google.com with ESMTPS id d7sm1632391lbn.3.2012.10.03.11.43.29 (version=TLSv1/SSLv3 cipher=OTHER); Wed, 03 Oct 2012 11:43:29 -0700 (PDT) From: Dmitry Monakhov To: linux-ext4@vger.kernel.org Cc: tytso@mit.edu, jack@suse.cz, Dmitry Monakhov Subject: [PATCH] ext4: fix ext4_flush_completed_IO wait semantics Date: Wed, 3 Oct 2012 22:43:27 +0400 Message-Id: <1349289807-18811-1-git-send-email-dmonakhov@openvz.org> X-Mailer: git-send-email 1.7.7.6 Sender: linux-ext4-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-ext4@vger.kernel.org BUG #1) All places where we call ext4_flush_completed_IO are broken because buffered io and DIO/AIO goes through three stages 1) submitted io, 2) completed io (in i_completed_io_list) conversion pended 3) finished io (conversion done) And by calling ext4_flush_completed_IO we will flush only requests which were in (2) stage, which is wrong because: 1) punch_hole and truncate _must_ wait for all outstanding unwritten io regardless to it's state. 2) fsync and nolock_dio_read should also wait because there is a time window between end_page_writeback() and ext4_add_complete_io() As result integrity fsync is broken in case of buffered write to fallocated region: fsync blkdev_completion ->filemap_write_and_wait_range ->ext4_end_bio ->end_page_writeback <-- filemap_write_and_wait_range return ->ext4_flush_completed_IO sees empty i_completed_io_list but pended conversion still exist ->ext4_add_complete_io BUG #2) Race window becomes wider due to 'ext4: completed_io locking cleanup V4' This patch make following changes: 1) ext4_flush_completed_io() now first try to flush completed io and when wait for any outstanding unwritten io via ext4_unwritten_wait() 2) Rename function to more appropriate name. 3) Assert that all callers of ext4_flush_unwritten_io should hold i_mutex to prevent endless wait Signed-off-by: Dmitry Monakhov Reviewed-by: Jan Kara --- fs/ext4/ext4.h | 3 ++- fs/ext4/extents.c | 6 +++--- fs/ext4/file.c | 2 +- fs/ext4/fsync.c | 2 +- fs/ext4/indirect.c | 8 +++++--- fs/ext4/page-io.c | 10 ++++++---- 6 files changed, 18 insertions(+), 13 deletions(-) diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 1be2b44..3ab2539 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -1947,7 +1947,7 @@ extern void ext4_htree_free_dir_info(struct dir_private_info *p); /* fsync.c */ extern int ext4_sync_file(struct file *, loff_t, loff_t, int); -extern int ext4_flush_completed_IO(struct inode *); +extern int ext4_flush_unwritten_io(struct inode *); /* hash.c */ extern int ext4fs_dirhash(const char *name, int len, struct @@ -2371,6 +2371,7 @@ extern const struct file_operations ext4_dir_operations; extern const struct inode_operations ext4_file_inode_operations; extern const struct file_operations ext4_file_operations; extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); +extern void ext4_unwritten_wait(struct inode *inode); /* namei.c */ extern const struct inode_operations ext4_dir_inode_operations; diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index c1fcf48..1c94cca 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -4268,7 +4268,7 @@ void ext4_ext_truncate(struct inode *inode) * finish any pending end_io work so we won't run the risk of * converting any truncated blocks to initialized later */ - ext4_flush_completed_IO(inode); + ext4_flush_unwritten_io(inode); /* * probably first extent we're gonna free will be last in block @@ -4847,10 +4847,10 @@ int ext4_ext_punch_hole(struct file *file, loff_t offset, loff_t length) /* Wait all existing dio workers, newcomers will block on i_mutex */ ext4_inode_block_unlocked_dio(inode); - inode_dio_wait(inode); - err = ext4_flush_completed_IO(inode); + err = ext4_flush_unwritten_io(inode); if (err) goto out_dio; + inode_dio_wait(inode); credits = ext4_writepage_trans_blocks(inode); handle = ext4_journal_start(inode, credits); diff --git a/fs/ext4/file.c b/fs/ext4/file.c index 39335bd..ca6f07a 100644 --- a/fs/ext4/file.c +++ b/fs/ext4/file.c @@ -55,7 +55,7 @@ static int ext4_release_file(struct inode *inode, struct file *filp) return 0; } -static void ext4_unwritten_wait(struct inode *inode) +void ext4_unwritten_wait(struct inode *inode) { wait_queue_head_t *wq = ext4_ioend_wq(inode); diff --git a/fs/ext4/fsync.c b/fs/ext4/fsync.c index 4600008..be1d89f 100644 --- a/fs/ext4/fsync.c +++ b/fs/ext4/fsync.c @@ -138,7 +138,7 @@ int ext4_sync_file(struct file *file, loff_t start, loff_t end, int datasync) if (inode->i_sb->s_flags & MS_RDONLY) goto out; - ret = ext4_flush_completed_IO(inode); + ret = ext4_flush_unwritten_io(inode); if (ret < 0) goto out; diff --git a/fs/ext4/indirect.c b/fs/ext4/indirect.c index 8d849da..37cd5a4 100644 --- a/fs/ext4/indirect.c +++ b/fs/ext4/indirect.c @@ -807,9 +807,11 @@ ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb, retry: if (rw == READ && ext4_should_dioread_nolock(inode)) { - if (unlikely(!list_empty(&ei->i_completed_io_list))) - ext4_flush_completed_IO(inode); - + if (unlikely(!atomic_read(&EXT4_I(inode)->i_unwritten))) { + mutex_lock(&inode->i_mutex); + ext4_flush_unwritten_io(inode); + mutex_unlock(&inode->i_mutex); + } /* * Nolock dioread optimization may be dynamically disabled * via ext4_inode_block_unlocked_dio(). Check inode's state diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c index 5b24c40..1f5df21 100644 --- a/fs/ext4/page-io.c +++ b/fs/ext4/page-io.c @@ -189,8 +189,6 @@ static int ext4_do_flush_completed_IO(struct inode *inode, list_add_tail(&io->list, &complete); } - /* It is important to update all flags for all end_io in one shot w/o - * dropping the lock.*/ spin_lock_irqsave(&ei->i_completed_io_lock, flags); while (!list_empty(&complete)) { io = list_entry(complete.next, ext4_io_end_t, list); @@ -228,9 +226,13 @@ static void ext4_end_io_work(struct work_struct *work) ext4_do_flush_completed_IO(io->inode, io); } -int ext4_flush_completed_IO(struct inode *inode) +int ext4_flush_unwritten_io(struct inode *inode) { - return ext4_do_flush_completed_IO(inode, NULL); + int ret; + WARN_ON_ONCE(!mutex_is_locked(&inode->i_mutex)); + ret = ext4_do_flush_completed_IO(inode, NULL); + ext4_unwritten_wait(inode); + return ret; } ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags)