session setup bugfix and torture test

Jeremy Allison jra at
Mon Sep 23 20:46:15 CEST 2013

On Sat, Sep 21, 2013 at 03:05:22AM +0200, Michael Adam wrote:
> For now find attached an update to the patchset that
> addresses your comments below:

So I think this patchset is good but there's one more
thing I think we need on top, and that's a walk through
the sconn->smb2.requests list for "in-flight" requests,
and if they don't match the req pointer currently being
processed, they need to have do_encrypt and do_sign set
to false, and their req->session pointer set to NULL.

That will catch any long-lived but not yet cancelled
requests on the same session that's being torn
down. I know it's not the 100% correct fix as it's
not issuing cancels on them, but it will at least
prevent any crashes on completion.

I think we can do this as a separate patch on
top of this existing patchset, so I'm ok with
you adding my "Reviewed-by: Jeremy Allison <jra at>"
on this patchset and pushing to autobuild if
metze is also happy.

I'm on a plane for the next 18+ hours or so but
I'll try and catch up with this once I'm safely
in Europe.



More information about the samba-technical mailing list