On Fri, 17 May 2013 05:24:15 +0200, "Yuan Fu" <yuan.fu@xxxxxx> wrote: > Dear Dmitry Monakhov, > > I see a race condition, > __ext4_abort() > ... > EXT4_SB(sb)->s_mount_flags |= EXT4_MF_FS_ABORTED; > ... > smp_wmb() > > [if scheduled at this point ] > ... > sb->s_flags |= MS_RDONLY; > > > if schedule occur above point(in red). There comes race condition > the s_mount_flags set to EXT4_MF_FS_ABORTED. On the other hand > sb->s_flags is not set to MS_RDONLY. Now if ext4_fsync_file() is > called from some process, the check s_flags to MS_RDONLY will fail, > and it will flush unwritten io and not return -EORFS. If flush_unwritten_io() was called after fs was aborted it will return appropriate code (EROFS or EIO) so fsync(2) will fail as expected. > > thanks > > ----- Original Message ----- > From: Dmitry Monakhov > Sent: 05/16/13 05:58 PM > Subject: [PATCH 1/4] ext4: Fix fsync error handling after filesysteb abort. > If filesystem was aborted after inode's write back complete > but before it's metadata was updated we may return success > due to (sb->s_flags & MS_RDONLY) which is incorrect and > result in data loss. > In order to handle fs abort correctly we have to check > fs state once we discover that it is in MS_RDONLY state > > Test case: http://patchwork.ozlabs.org/patch/244297/ > > Signed-off-by: Dmitry Monakhov <dmonakhov@xxxxxxxxxx> > --- > fs/ext4/fsync.c | 8 ++++++-- > fs/ext4/super.c | 13 ++++++++++++- > 2 files changed, 18 insertions(+), 3 deletions(-) > > diff --git a/fs/ext4/fsync.c b/fs/ext4/fsync.c > index e0ba8a4..d7df2f1 100644 > --- a/fs/ext4/fsync.c > +++ b/fs/ext4/fsync.c > @@ -129,9 +129,13 @@ int ext4_sync_file(struct file *file, loff_t start, loff_t end, int datasync) > return ret; > mutex_lock(&inode->i_mutex); > > - if (inode->i_sb->s_flags & MS_RDONLY) > + if (inode->i_sb->s_flags & MS_RDONLY) { > + /* Make shure that we read updated s_mount_flags value */ > + smp_rmb(); > + if (EXT4_SB(inode->i_sb)->s_mount_flags & EXT4_MF_FS_ABORTED) > + ret = -EROFS; > goto out; > - > + } > ret = ext4_flush_unwritten_io(inode); > if (ret < 0) > goto out; > diff --git a/fs/ext4/super.c b/fs/ext4/super.c > index dbc7c09..6c91c8e 100644 > --- a/fs/ext4/super.c > +++ b/fs/ext4/super.c > @@ -398,6 +398,11 @@ static void ext4_handle_error(struct super_block *sb) > } > if (test_opt(sb, ERRORS_RO)) { > ext4_msg(sb, KERN_CRIT, "Remounting filesystem read-only"); > + /* > + * Make shure updated value of ->s_mount_flags will be visiable > + * before ->s_flags update > + */ > + smp_wmb(); > sb->s_flags |= MS_RDONLY; > } > if (test_opt(sb, ERRORS_PANIC)) > @@ -552,6 +557,7 @@ void __ext4_std_error(struct super_block *sb, const char *function, > * > * We unconditionally force the filesystem into an ABORT|READONLY state, > * unless the error response on the fs has been set to panic in which > + > * case we take the easy way out and panic immediately. > */ > > @@ -570,8 +576,13 @@ void __ext4_abort(struct super_block *sb, const char *function, > > if ((sb->s_flags & MS_RDONLY) == 0) { > ext4_msg(sb, KERN_CRIT, "Remounting filesystem read-only"); > - sb->s_flags |= MS_RDONLY; > EXT4_SB(sb)->s_mount_flags |= EXT4_MF_FS_ABORTED; > + /* > + * Make shure updated value of ->s_mount_flags will be visiable > + * before ->s_flags update > + */ > + smp_wmb(); > + sb->s_flags |= MS_RDONLY; > if (EXT4_SB(sb)->s_journal) > jbd2_journal_abort(EXT4_SB(sb)->s_journal, -EIO); > save_error_info(sb, function, line); > -- > 1.7.1 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-ext4" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html > > -- > To unsubscribe from this list: send the line "unsubscribe linux-ext4" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
