if (unlikely(page == NULL)) {
if (flags & XBF_READ_AHEAD) {
bp->b_page_count = i;
+ for (i = 0; i < bp->b_page_count; i++)
+ unlock_page(bp->b_pages[i]);
return -ENOMEM;
}
ASSERT(!PagePrivate(page));
if (!PageUptodate(page)) {
page_count--;
- if (blocksize < PAGE_CACHE_SIZE && !PagePrivate(page)) {
+ if (blocksize >= PAGE_CACHE_SIZE) {
+ if (flags & XBF_READ)
+ bp->b_flags |= _XBF_PAGE_LOCKED;
+ } else if (!PagePrivate(page)) {
if (test_page_region(page, offset, nbytes))
page_count++;
}
}
- unlock_page(page);
bp->b_pages[i] = page;
offset = 0;
}
+ if (!(bp->b_flags & _XBF_PAGE_LOCKED)) {
+ for (i = 0; i < bp->b_page_count; i++)
+ unlock_page(bp->b_pages[i]);
+ }
+
if (page_count == bp->b_page_count)
bp->b_flags |= XBF_DONE;
bp->b_count_desired = len;
bp->b_buffer_length = buflen;
bp->b_flags |= XBF_MAPPED;
+ bp->b_flags &= ~_XBF_PAGE_LOCKED;
return 0;
}
xfs_buf_t *bp,
int schedule)
{
- if (atomic_dec_and_test(&bp->b_io_remaining) == 1)
+ if (atomic_dec_and_test(&bp->b_io_remaining) == 1) {
+ bp->b_flags &= ~_XBF_PAGE_LOCKED;
xfs_buf_ioend(bp, schedule);
+ }
}
STATIC void
if (--bvec >= bio->bi_io_vec)
prefetchw(&bvec->bv_page->flags);
+
+ if (bp->b_flags & _XBF_PAGE_LOCKED)
+ unlock_page(page);
} while (bvec >= bio->bi_io_vec);
_xfs_buf_ioend(bp, 1);
* filesystem block size is not smaller than the page size.
*/
if ((bp->b_buffer_length < PAGE_CACHE_SIZE) &&
- (bp->b_flags & XBF_READ) &&
+ ((bp->b_flags & (XBF_READ|_XBF_PAGE_LOCKED)) ==
+ (XBF_READ|_XBF_PAGE_LOCKED)) &&
(blocksize >= PAGE_CACHE_SIZE)) {
bio = bio_alloc(GFP_NOIO, 1);
/*
* xfs_fsync
*
- * This is called to sync the inode and its data out to disk.
- * We need to hold the I/O lock while flushing the data, and
- * the inode lock while flushing the inode. The inode lock CANNOT
- * be held while flushing the data, so acquire after we're done
- * with that.
+ * This is called to sync the inode and its data out to disk. We need to hold
+ * the I/O lock while flushing the data, and the inode lock while flushing the
+ * inode. The inode lock CANNOT be held while flushing the data, so acquire
+ * after we're done with that.
*/
int
xfs_fsync(
- xfs_inode_t *ip,
- int flag,
- xfs_off_t start,
- xfs_off_t stop)
+ xfs_inode_t *ip)
{
xfs_trans_t *tp;
int error;
xfs_itrace_entry(ip);
- ASSERT(start >= 0 && stop >= -1);
-
if (XFS_FORCED_SHUTDOWN(ip->i_mount))
return XFS_ERROR(EIO);
- if (flag & FSYNC_DATA)
- filemap_fdatawait(vn_to_inode(XFS_ITOV(ip))->i_mapping);
+ /* capture size updates in I/O completion before writing the inode. */
+ error = filemap_fdatawait(vn_to_inode(XFS_ITOV(ip))->i_mapping);
+ if (error)
+ return XFS_ERROR(error);
/*
- * We always need to make sure that the required inode state
- * is safe on disk. The vnode might be clean but because
- * of committed transactions that haven't hit the disk yet.
- * Likewise, there could be unflushed non-transactional
- * changes to the inode core that have to go to disk.
+ * We always need to make sure that the required inode state is safe on
+ * disk. The vnode might be clean but we still might need to force the
+ * log because of committed transactions that haven't hit the disk yet.
+ * Likewise, there could be unflushed non-transactional changes to the
+ * inode core that have to go to disk and this requires us to issue
+ * a synchronous transaction to capture these changes correctly.
*
- * The following code depends on one assumption: that
- * any transaction that changes an inode logs the core
- * because it has to change some field in the inode core
- * (typically nextents or nblocks). That assumption
- * implies that any transactions against an inode will
- * catch any non-transactional updates. If inode-altering
- * transactions exist that violate this assumption, the
- * code breaks. Right now, it figures that if the involved
- * update_* field is clear and the inode is unpinned, the
- * inode is clean. Either it's been flushed or it's been
- * committed and the commit has hit the disk unpinning the inode.
- * (Note that xfs_inode_item_format() called at commit clears
- * the update_* fields.)
+ * This code relies on the assumption that if the update_* fields
+ * of the inode are clear and the inode is unpinned then it is clean
+ * and no action is required.
*/
xfs_ilock(ip, XFS_ILOCK_SHARED);
- /* If we are flushing data then we care about update_size
- * being set, otherwise we care about update_core
- */
- if ((flag & FSYNC_DATA) ?
- (ip->i_update_size == 0) :
- (ip->i_update_core == 0)) {
+ if (!(ip->i_update_size || ip->i_update_core)) {
/*
- * Timestamps/size haven't changed since last inode
- * flush or inode transaction commit. That means
- * either nothing got written or a transaction
- * committed which caught the updates. If the
- * latter happened and the transaction hasn't
- * hit the disk yet, the inode will be still
- * be pinned. If it is, force the log.
+ * Timestamps/size haven't changed since last inode flush or
+ * inode transaction commit. That means either nothing got
+ * written or a transaction committed which caught the updates.
+ * If the latter happened and the transaction hasn't hit the
+ * disk yet, the inode will be still be pinned. If it is,
+ * force the log.
*/
xfs_iunlock(ip, XFS_ILOCK_SHARED);
if (xfs_ipincount(ip)) {
- _xfs_log_force(ip->i_mount, (xfs_lsn_t)0,
- XFS_LOG_FORCE |
- ((flag & FSYNC_WAIT)
- ? XFS_LOG_SYNC : 0),
+ error = _xfs_log_force(ip->i_mount, (xfs_lsn_t)0,
+ XFS_LOG_FORCE | XFS_LOG_SYNC,
&log_flushed);
} else {
/*
- * If the inode is not pinned and nothing
- * has changed we don't need to flush the
- * cache.
+ * If the inode is not pinned and nothing has changed
+ * we don't need to flush the cache.
*/
changed = 0;
}
- error = 0;
} else {
/*
- * Kick off a transaction to log the inode
- * core to get the updates. Make it
- * sync if FSYNC_WAIT is passed in (which
- * is done by everybody but specfs). The
- * sync transaction will also force the log.
+ * Kick off a transaction to log the inode core to get the
+ * updates. The sync transaction will also force the log.
*/
xfs_iunlock(ip, XFS_ILOCK_SHARED);
tp = xfs_trans_alloc(ip->i_mount, XFS_TRANS_FSYNC_TS);
- if ((error = xfs_trans_reserve(tp, 0,
- XFS_FSYNC_TS_LOG_RES(ip->i_mount),
- 0, 0, 0))) {
+ error = xfs_trans_reserve(tp, 0,
+ XFS_FSYNC_TS_LOG_RES(ip->i_mount), 0, 0, 0);
+ if (error) {
xfs_trans_cancel(tp, 0);
return error;
}
xfs_ilock(ip, XFS_ILOCK_EXCL);
/*
- * Note - it's possible that we might have pushed
- * ourselves out of the way during trans_reserve
- * which would flush the inode. But there's no
- * guarantee that the inode buffer has actually
- * gone out yet (it's delwri). Plus the buffer
- * could be pinned anyway if it's part of an
- * inode in another recent transaction. So we
- * play it safe and fire off the transaction anyway.
+ * Note - it's possible that we might have pushed ourselves out
+ * of the way during trans_reserve which would flush the inode.
+ * But there's no guarantee that the inode buffer has actually
+ * gone out yet (it's delwri). Plus the buffer could be pinned
+ * anyway if it's part of an inode in another recent
+ * transaction. So we play it safe and fire off the
+ * transaction anyway.
*/
xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
xfs_trans_ihold(tp, ip);
xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
- if (flag & FSYNC_WAIT)
- xfs_trans_set_sync(tp);
+ xfs_trans_set_sync(tp);
error = _xfs_trans_commit(tp, 0, &log_flushed);
xfs_iunlock(ip, XFS_ILOCK_EXCL);
projects / cascardo / linux.git / commitdiff