Re: btrfs: could not do orphan cleanup -116

[Josef Bacik <josef@redhat.com>]

On Wed, Nov 09, 2011 at 03:43:37PM +0000, Maciej Marcin Piechotka wrote:
> On Mon, 2011-11-07 at 14:01 -0500, Josef Bacik wrote:
> > On Mon, Nov 07, 2011 at 06:41:44PM +0000, Maciej Marcin Piechotka wrote:
> > > On Mon, 2011-11-07 at 10:29 -0500, Josef Bacik wrote:
> > > > On Mon, Nov 07, 2011 at 01:17:04PM +0000, Maciej Marcin Piechotka wrote:
> > > > > Hello,
> > > > > 
> > > > > When I booted my machine (after clean powerdown) the following message
> > > > > appeared:
> > > > > [   32.757913] device fsid XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX devid 1
> > > > > transid 40864 /dev/mapper/XXXXX-XXXXX
> > > > > [   32.758466] btrfs: use lzo compression
> > > > > [   32.758475] btrfs: enabling disk space caching
> > > > > [   32.758483] btrfs: enabling inode map caching
> > > > > [   32.758490] btrfs: enabling auto defrag
> > > > > [   32.758497] btrfs: thread pool 2
> > > > > [   34.024359] btrfs: could not do orphan cleanup -116
> > > > > [   63.173382] btrfs: open_ctree failed
> > > > > 
> > > > > btrfsck has shown no errors.
> > > > > 
> > > > > Afterwards when I tried to mount filesystem it have shown:
> > > > > 
> > > > > [  116.012528] btrfs: enabling inode map caching
> > > > > [  116.012534] btrfs: enabling auto defrag
> > > > > [  116.012542] btrfs: thread pool 2
> > > > > 
> > > > > It looked like everything is working but to be sure I've run scrub:
> > > > > 
> > > > > scrub status for fsid XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX
> > > > >         scrub started at Mon Nov  7 13:05:50 2011 and finished after 606
> > > > > seconds
> > > > >         total bytes scrubbed: 32.37GB with 0 errors
> > > > > 
> > > > > PS. 
> > > > > # find include | xargs grep 116
> > > > > (...)
> > > > > include/asm-generic/errno.h:#define     ESTALE          116     /* Stale
> > > > > NFS file handle */
> > > > > (...)
> > > > > 
> > > > > I don't use NFS - I use LVM on LUKS. In any case - it have been done
> > > > > concurrently with mounting of other fs which had no such problems.
> > > > 
> > > > Yeah that's a bug with an earlier kernel, if you upgrade to 3.1 it should go
> > > > away.  Thanks,
> > > > 
> > > > Josef
> > > 
> > > 
> > > I'm using 'newer' kernel then 3.1 - I'm following Linus' tree (this
> > > kernel is few days old).
> > > 
> > 
> > Do you have the commit
> > 
> > a8c9e5769718d47e87cce40c9b84cab421804797
> > Btrfs: fix orphan cleanup regression
> > 
> > If not then the kernel isn't new enough.  Thanks,
> > 
> > Josef
> 
> It have been merged in master for 3.2-rc1 (shortly before tag).
> 
> Regards
> 

Does that mean you have that patch and are still hitting the problem?  If so try
this debug patch so I can see where it's coming from.  Thanks,

Josef


diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index faf8a60..d72225a 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -2153,14 +2173,17 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
 		found_key.offset = 0;
 		inode = btrfs_iget(root->fs_info->sb, &found_key, root, NULL);
 		ret = PTR_RET(inode);
-		if (ret && ret != -ESTALE)
+		if (ret && ret != -ESTALE) {
+			printk(KERN_ERR "ret is %d\n");
 			goto out;
+		}
 
 		/*
 		 * Inode is already gone but the orphan item is still there,
 		 * kill the orphan item.
 		 */
 		if (ret == -ESTALE) {
+			printk(KERN_ERR "bad inode, deleting orphan item\n");
 			trans = btrfs_start_transaction(root, 1);
 			if (IS_ERR(trans)) {
 				ret = PTR_ERR(trans);
@@ -2170,6 +2193,7 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
 						    found_key.objectid);
 			BUG_ON(ret);
 			btrfs_end_transaction(trans, root);
+			printk(KERN_ERR "successfully deleted orphan item\n");
 			continue;
 		}