>Perhaps I've just missed something, but is there still possibly a problem
>with keepalives, or has some default changed?
>I peripherally followed the discussion (last week?) about the bug(s) in
>tcp_subr.c causing keepalives not to register, and telnet sessions timing
>out after 2h 12m.
>What I'm seeing is my idle ssh sessions timing out after almost exactly 2
>hours.

	Maybe I'm misunderstanding something.
	Is there any possibility that tcp_respond() get called with NULL
	template?  If so I need to update tcp_respond() so that it can guess
	address family without looking into the IP version field in template.

itojun