> With a configuration which I used extensively
> with never a panic in the past I now get a pretty much guaranteed
> panic with "union: missing upper layer in write".
> 
> I suppose I should put in a send-pr, but then I know unionfs is generally
> only considered half working anyway. 

If you don't report the bugs, there's no way it will ever be fixed.

> What has changed to make it worse?

Well, there were a number of changes to unionfs in 1.4; some of them
were due to fixes for bugs which I encountered in my use of the
filesystem (e.g., panics from the NFS server when exporting a
filesystem which had a uniom mount over it).

Part of the problem here is that there are many different ways to use
unionfs.  I've fixed the bugs which have gotten in my way, but we
don't have a good unionfs regression test suite.

By all means, *please* send-pr this, including, if at all possible:
   - backtrace from the panic
   - exact mount commands/fstab setup used to mount the filesystems.
   - how to provoke the panic.

Thanks.
					- Bill