[Bug 289734] panic tcp_usr_close while running mount command after configure NFS over TLS

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=289734

--- Comment #21 from Gleb Smirnoff <glebius@FreeBSD.org> ---
Please don't rush with reverting my code.  Let's first obtain a core or more
solid crash data before rushing with revertions.  The pthreads/fork side of the
change just can't be relevant, for kernel side it really doesn't matter if the
daemon forks or threads.  Conceptually I didn't change much wrt the design you
described in the comment #17, it still the same hack with a socket originally
owned by kernel handed to the daemon.  The number of arguments to the syscall
has changed, and the lookup is also slightly different. I agree that using
socket pointer value as a cookie looks suspicious.  We discussed that earlier
and I remember I was looking into if reuse is possible, and decided it is not.
I should look at this question once more and probably instrument the reuse
arificially to prove or debunk this theory.

-- 
You are receiving this mail because:
You are the assignee for the bug.