Re: [PATCH] fix nr_unused accounting, and avoid recursing in iput with I_WILL_FREE set

[Andrea Arcangeli <andrea@suse.de>]

On Mon, Oct 24, 2005 at 10:21:02PM -0400, Chris Mason wrote:
> The short version is that no additional patches should be needed for
> mainline.

This one may be needed too. Perhaps it's unnecessary for the MS_ACTIVE
case (I would assume in that case by design nobody can find the inode
while MS_ACTIVE is not set), but the unhashed case sounds more
interesting. At the moment I'm unsure who is using the unhashed
last-iput feature to get rid of the inode but the below looks needed at
the light of that feature.

Signed-off-by: Andrea Arcangeli <andrea@suse.de>

diff -r 2c7561cc445e fs/inode.c
--- a/fs/inode.c	Mon Oct 24 00:24:54 2005 +0011
+++ b/fs/inode.c	Tue Oct 25 16:06:25 2005 +0200
@@ -1088,6 +1088,7 @@
 	if (inode->i_data.nrpages)
 		truncate_inode_pages(&inode->i_data, 0);
 	clear_inode(inode);
+	wake_up_inode(inode);
 	destroy_inode(inode);
 }