Re: Deleted parent node in PUFFS lookup

[David Holland <dholland-tech%netbsd.org@localhost>]

On Thu, Nov 24, 2011 at 02:14:30PM +0100, Emmanuel Dreyfus wrote:
 > > When calling lookup, the parent (directory) is supposed to be locked
 > > to prevent exactly this sort of thing from happening. You and/or
 > > perfuse and/or puffs ignore this at your peril, because doing so
 > > violates atomicitiy guarantees and assumptions made by upper-level
 > > code.
 > 
 > I don't violate the need for a locked vnode during lookup, the problem
 > is with distributed filesystem, where another machine delete the
 > directory.

It's not just the vnode that's supposed to be locked; it's the
*directory*, IOW the contents. If it's not, open with O_CREAT|O_EXCL
will probably not work reliably, which is in turn a security hazard.

Although it's perhaps no worse than nfs then...

 > > However, if it's really disappeared, why not just fail with ENOENT? 
 > 
 > ENOENT tells the kernel that the entry it looked up does not exists. It
 > does not tell anything about the parent vnode being now stale.

What does NFS do for ESTALE cases? Whatever it is, there's probably a
way to do it via fuse, and therefore also via perfuse.

-- 
David A. Holland
dholland%netbsd.org@localhost