linux/fs/ext3/fsync.c
Dmitry Monakhov 3df3219651 ext3: Fix fsync error handling after filesystem abort.
If filesystem was aborted we will return success
due to (sb->s_flags & MS_RDONLY) which is incorrect and
results 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/
Changes from V1:
 - fix spelling
 - fix smp_rmb()/debug order

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jan Kara <jack@suse.cz>
2013-07-04 19:22:55 +02:00

110 lines
3.2 KiB
C

/*
* linux/fs/ext3/fsync.c
*
* Copyright (C) 1993 Stephen Tweedie (sct@redhat.com)
* from
* Copyright (C) 1992 Remy Card (card@masi.ibp.fr)
* Laboratoire MASI - Institut Blaise Pascal
* Universite Pierre et Marie Curie (Paris VI)
* from
* linux/fs/minix/truncate.c Copyright (C) 1991, 1992 Linus Torvalds
*
* ext3fs fsync primitive
*
* Big-endian to little-endian byte-swapping/bitmaps by
* David S. Miller (davem@caip.rutgers.edu), 1995
*
* Removed unnecessary code duplication for little endian machines
* and excessive __inline__s.
* Andi Kleen, 1997
*
* Major simplications and cleanup - we only need to do the metadata, because
* we can depend on generic_block_fdatasync() to sync the data blocks.
*/
#include <linux/blkdev.h>
#include <linux/writeback.h>
#include "ext3.h"
/*
* akpm: A new design for ext3_sync_file().
*
* This is only called from sys_fsync(), sys_fdatasync() and sys_msync().
* There cannot be a transaction open by this task.
* Another task could have dirtied this inode. Its data can be in any
* state in the journalling system.
*
* What we do is just kick off a commit and wait on it. This will snapshot the
* inode to disk.
*/
int ext3_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
{
struct inode *inode = file->f_mapping->host;
struct ext3_inode_info *ei = EXT3_I(inode);
journal_t *journal = EXT3_SB(inode->i_sb)->s_journal;
int ret, needs_barrier = 0;
tid_t commit_tid;
trace_ext3_sync_file_enter(file, datasync);
if (inode->i_sb->s_flags & MS_RDONLY) {
/* Make sure that we read updated state */
smp_rmb();
if (EXT3_SB(inode->i_sb)->s_mount_state & EXT3_ERROR_FS)
return -EROFS;
return 0;
}
ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
if (ret)
goto out;
J_ASSERT(ext3_journal_current_handle() == NULL);
/*
* data=writeback,ordered:
* The caller's filemap_fdatawrite()/wait will sync the data.
* Metadata is in the journal, we wait for a proper transaction
* to commit here.
*
* data=journal:
* filemap_fdatawrite won't do anything (the buffers are clean).
* ext3_force_commit will write the file data into the journal and
* will wait on that.
* filemap_fdatawait() will encounter a ton of newly-dirtied pages
* (they were dirtied by commit). But that's OK - the blocks are
* safe in-journal, which is all fsync() needs to ensure.
*/
if (ext3_should_journal_data(inode)) {
ret = ext3_force_commit(inode->i_sb);
goto out;
}
if (datasync)
commit_tid = atomic_read(&ei->i_datasync_tid);
else
commit_tid = atomic_read(&ei->i_sync_tid);
if (test_opt(inode->i_sb, BARRIER) &&
!journal_trans_will_send_data_barrier(journal, commit_tid))
needs_barrier = 1;
log_start_commit(journal, commit_tid);
ret = log_wait_commit(journal, commit_tid);
/*
* In case we didn't commit a transaction, we have to flush
* disk caches manually so that data really is on persistent
* storage
*/
if (needs_barrier) {
int err;
err = blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL, NULL);
if (!ret)
ret = err;
}
out:
trace_ext3_sync_file_exit(inode, ret);
return ret;
}