>What I have is: NTPv4 working _with_ a clock line discipline for the
>cheap (U$120 or so) UltraLink M320 WWVB receiver on NetBSD 1.3.3, i386.

Hey wow, I'm looking at one of those as well :-)

I think the right thing to do for all of the Ultralink receivers is to
tie yourself to the PPS signal; that way you don't have to deal with things
like a FIFO timeout for various UARTs.  Unfortunately, the current M320
PPS output is TTL level instead of RS-232; fixing this isn't hard, but
it will involve a bit of additional circuitry.  This of course means
we need all of the code to handle PPS signals in the kernel, but we
should do that anyway :-)  (According to the guys at Ultralink, the
next version of the 320 will have a RS-232 PPS output).

Did you just write a new "parse" refclock module, or did you implement
a completely new refclock?

>I had planned to hunt down and kill (1), then talk to some of the NetBSD
>core people to get some advice about (2) before plunging into (3).  All
>of this will have to wait for two months or so unless someone else wants
>to try picking up where I've left off.  (You'll have to supply you own
>M320 though :).)

I'm willing to take a look at it; please email me your diffs.

--Ken