xfs: merge fsync and O_SYNC handling

The guarantees for O_SYNC are exactly the same as the ones we need to
make for an fsync call (and given that Linux O_SYNC is O_DSYNC the
equivalent is fdadatasync, but we treat both the same in XFS), except
with a range data writeout.  Jan Kara has started unifying these two
path for filesystems using the generic helpers, and I've started to
look at XFS.

The actual transaction commited by xfs_fsync and xfs_write_sync_logforce
has a different transaction number, but actually is exactly the same.
We'll only use the fsync transaction going forward.  One major difference
is that xfs_write_sync_logforce never issues a cache flush unless we
commit a transaction causing that as a side-effect, which is an obvious
bug in the O_SYNC handling.  Second all the locking and i_update_size
vs i_update_core changes from 978b7237123d007b9fa983af6e0e2fa8f97f9934
never made it to xfs_write_sync_logforce, so we add them back.

To make xfs_fsync easily usable from the O_SYNC path, the filemap_fdatawait
call is moved up to xfs_file_fsync, so that we don't wait on the whole
file after we already waited for our portion in xfs_write.

We'll also use a plain call to filemap_write_and_wait_range instead
of the previous sync_page_rang which did it in two steps including
an half-hearted inode write out that doesn't help us.

Once we're done with this also remove the now useless i_update_size
tracking.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Felix Blyakher <felixb@sgi.com>
Signed-off-by: Felix Blyakher <felixb@sgi.com>

1 files changed, 0 insertions, 84 deletions

diff --git a/fs/xfs/xfs_rw.c b/fs/xfs/xfs_rw.c
index fea68615ed23..3f816ad7ff19 100644
--- a/fs/xfs/xfs_rw.c
+++ b/fs/xfs/xfs_rw.c
@@ -88,90 +88,6 @@ xfs_write_clear_setuid(
 }
 
 /*
- * Handle logging requirements of various synchronous types of write.
- */
-int
-xfs_write_sync_logforce(
-        xfs_mount_t     *mp,
-        xfs_inode_t     *ip)
-{
-        int             error = 0;
-
-        /*
-         * If we're treating this as O_DSYNC and we have not updated the
-         * size, force the log.
-         */
-        if (!(mp->m_flags & XFS_MOUNT_OSYNCISOSYNC) &&
-            !(ip->i_update_size)) {
-                xfs_inode_log_item_t    *iip = ip->i_itemp;
-
-                /*
-                 * If an allocation transaction occurred
-                 * without extending the size, then we have to force
-                 * the log up the proper point to ensure that the
-                 * allocation is permanent.  We can't count on
-                 * the fact that buffered writes lock out direct I/O
-                 * writes - the direct I/O write could have extended
-                 * the size nontransactionally, then finished before
-                 * we started.  xfs_write_file will think that the file
-                 * didn't grow but the update isn't safe unless the
-                 * size change is logged.
-                 *
-                 * Force the log if we've committed a transaction
-                 * against the inode or if someone else has and
-                 * the commit record hasn't gone to disk (e.g.
-                 * the inode is pinned).  This guarantees that
-                 * all changes affecting the inode are permanent
-                 * when we return.
-                 */
-                if (iip && iip->ili_last_lsn) {
-                        error = _xfs_log_force(mp, iip->ili_last_lsn,
-                                        XFS_LOG_FORCE | XFS_LOG_SYNC, NULL);
-                } else if (xfs_ipincount(ip) > 0) {
-                        error = _xfs_log_force(mp, (xfs_lsn_t)0,
-                                        XFS_LOG_FORCE | XFS_LOG_SYNC, NULL);
-                }
-
-        } else {
-                xfs_trans_t     *tp;
-
-                /*
-                 * O_SYNC or O_DSYNC _with_ a size update are handled
-                 * the same way.
-                 *
-                 * If the write was synchronous then we need to make
-                 * sure that the inode modification time is permanent.
-                 * We'll have updated the timestamp above, so here
-                 * we use a synchronous transaction to log the inode.
-                 * It's not fast, but it's necessary.
-                 *
-                 * If this a dsync write and the size got changed
-                 * non-transactionally, then we need to ensure that
-                 * the size change gets logged in a synchronous
-                 * transaction.
-                 */
-                tp = xfs_trans_alloc(mp, XFS_TRANS_WRITE_SYNC);
-                if ((error = xfs_trans_reserve(tp, 0,
-                                                XFS_SWRITE_LOG_RES(mp),
-                                                0, 0, 0))) {
-                        /* Transaction reserve failed */
-                        xfs_trans_cancel(tp, 0);
-                } else {
-                        /* Transaction reserve successful */
-                        xfs_ilock(ip, XFS_ILOCK_EXCL);
-                        xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
-                        xfs_trans_ihold(tp, ip);
-                        xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
-                        xfs_trans_set_sync(tp);
-                        error = xfs_trans_commit(tp, 0);
-                        xfs_iunlock(ip, XFS_ILOCK_EXCL);
-                }
-        }
-
-        return error;
-}
-
-/*
  * Force a shutdown of the filesystem instantly while keeping
  * the filesystem consistent. We don't do an unmount here; just shutdown
  * the shop, make sure that absolutely nothing persistent happens to