It has been reported that a Linux NFSv4.1 client mount to a
FreeBSD server can result in the FreeBSD end of the TCP
socket being stuck in CLOSE_WAIT.
I think this can happen because the soclose() for the socket
cannot happen until the back channel is re-assigned to another
This patch adds soshutdown(..SHUT_WR) calls where the socket
is no longer usable.
The patch is simple, but the question is "does soshutdown(..SHUT_WR)
sound like the correct thing to do to get the TCP connection past