[OpenAFS-port-darwin] Different kernel panic that's probably AFS-related

[Derrick J Brashear]

On Wed, 17 Mar 2004, Mattias Amnefelt wrote:

> What carbon applications do when renaming files is that they first
> rename the file, then try to rename the resource-fork (which is called
> ._filename). In my test-cases this resource-fork doesn't exist, so the
> syscall fails after the first call to vop_lookup. They then try to
> remove the target resource-fork by calling the special delete
> systemcall. This is basicly a normal unlink but with the changed
> semantics. Delete doesn't unlink a file if it is busy, but returns an
> error instead. This call also fails after a vop_lookup.

I assume you have one of fakestat or fakestat-all on the system.

Until just now my test system had neither, and i wonder if perhaps that's
related.

I'd hoped /Developer/Tools/MvMac might have behaved the same way,
incidentally, but it appears not to based on ktrace.