>You're proposing we test the line discipline in the hard-interrupt
>service routine, and for ``NTP'' disciplines, handling the ldisc stuff
>immediately rather than deferring?

Something like that (maybe do something similar to the way kgdb
setup is done now, by setting a magical bit in the driver softc
structure).

>I think the real question is, can this NTP optimization be done
>without adversely affecting the quality of the serial driver(s)
>concerned.  I'd run this by whoever maintains the relevant serial
>drivers.

My thinking was that this would be #ifdef'd so it wouldn't affect
people who didn't want it.

>I understand this is "for cheap", but spending the $$$ to get either a
>PPS signal or a faster CPU (or both) is going to make a *huge*
>difference to NTP quality.  Five or ten years ago this would be an
>obvious win, but with lowend GPS handhelds as low as $99 and sure to
>drop further, I really wonder if this is worth it.

Well, $99 is 5 times the current total monetary outlay of my CHU
receiver, shortwave radio, and NTP server PC (I did a _whole_ lot of
scavaging).  My idea of "cheap" I suspect is different that most
NetBSDer's, since I'm a) married and b) own a house (I mean no offense
to those NetBSDer's that are married or own homes; it's just that out
of all the NetBSDer's I've met, few fall into that category).  When you
get right down to it, a NTP time source at home doesn't have _much_
practical use, other than it would be really really cool.  My wife can
accept me doing something as geeky as getting a NTP time source at
home, as long as it doesn't cost too much :-)  But seriously ... when I
say cheap, I mean REALLY cheap (programming time is considered "free"
when it comes to the household budget :-) ).  And I must confess that
building a CHU receiver was kinda fun :-)

But, getting back to those cheap GPS receivers ... I _did_ look at
them, but the ones I found all had around 100ms latency, and no PPS
output (or a PPS output that wasn't very good).  To get to the 1 ms
range or better took more money.  If you know of a cheap GPS receiver
that has a good time output, please let me know!  The cheapest of the
lot that I could find was the Totally Accurate Clock kit plus a GPS OEM
board, but that still ends up costing a couple hundred dollars.

>Umm,  I just realized: are you asking about for  your own machine,
>or as a change to be pulled into the NetBSD tree?

Well, if it's clean enough, I'd like to have it put into the tree.
I don't know if it would get much use, but hey, you never know ...

--Ken