Instead of trying to unpin a pinned inode buffer by forcing the log in
xfs_iflush let the caller handle it. In case of AIL pushing we already
handle it more efficiently by forcing the log synchronously once after
cycling through all AIL entries.
Signed-off-by: Christoph Hellwig <hch@xxxxxx>
---
fs/xfs/xfs_inode.c | 7 -------
fs/xfs/xfs_sync.c | 7 +++++++
2 files changed, 7 insertions(+), 7 deletions(-)
Index: xfs/fs/xfs/xfs_inode.c
===================================================================
--- xfs.orig/fs/xfs/xfs_inode.c 2011-10-27 22:40:16.244688077 +0200
+++ xfs/fs/xfs/xfs_inode.c 2011-10-27 22:40:16.821171772 +0200
@@ -2468,13 +2468,6 @@ xfs_iflush(
goto corrupt_out;
/*
- * If the buffer is pinned then push on the log now so we won't
- * get stuck waiting in the write for too long.
- */
- if (xfs_buf_ispinned(bp))
- xfs_log_force(mp, 0);
-
- /*
* inode clustering:
* see if other inodes can be gathered into this write
*/
Index: xfs/fs/xfs/xfs_sync.c
===================================================================
--- xfs.orig/fs/xfs/xfs_sync.c 2011-10-27 22:40:16.257172598 +0200
+++ xfs/fs/xfs/xfs_sync.c 2011-10-27 22:40:16.821171772 +0200
@@ -724,6 +724,13 @@ restart:
delay(2);
goto restart;
} else if (!error) {
+ /*
+ * If the buffer is pinned then push on the log now so we
+ * won't get stuck waiting in the write for too long.
+ */
+ if (xfs_buf_ispinned(bp))
+ xfs_log_force(ip->i_mount, 0);
+
error = xfs_bwrite(bp);
xfs_buf_relse(bp);
}
_______________________________________________
xfs mailing list
xfs@xxxxxxxxxxx
http://oss.sgi.com/mailman/listinfo/xfs
