Re: [netmp] socket accesses

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

    Hmm.  One more thing.  The User side can set the flag but may still have
    to wakeup other user side readers that are blocked, and that might
    require a predicate message to be sent (all it would do is tell the
    protocol stack to wakeup any other readers).  All the predicate would
    do is sorwakeup() equivalent -- handle any pending netmsgs for blocking
    on read data.  And it could be optimized out and not sent at all if
    no other readers are blocked.

					-Matt
					Matthew Dillon 
					<dillon@backplane.com>