serverid patchset ?

[Jeremy Allison]

On Tue, May 22, 2012 at 11:41:23AM -0700, Jeremy Allison wrote:
> 
> The only place where we could get zero valid share modes and
> a delete on close token is where the client sets delete on
> close and then the *server* crashes (or is terminated without
> clean shutdown). This is not a case we can test via experiment
> against Windows as it's essentially server dependent whether
> we delete the file or not after server crash.

Just to make it clear, even sending a TCP reset instead of
TCP fin from the client won't reproduce this state. We would
still treat it as SHUTDOWN_CLOSE in the server code.

We have to crash (or explicitly be aborted) in order to
reproduce this situation.

Jeremy.