[LinuxPPS] refclock_nmea patch

[James Boddington]

Bernhard Schiffner wrote:
> Somesimes it happens, sometime not.
> In general now ntp is slower then it was before. I have had a look at the 
> sources but couldn't chatch the reason / difference. 

Plots of mine are at http://aiken.dnsalias.org/cgi-bin/linuxpps

Using minpoll 4 and default maxpoll on the refclocks. The convergence is slow 
enough that the poll interval mostly can not ramp up for the internet based 
servers. I saw a poll interval of 1024 about 3 of hours ago and 128 earlier but 
poll interval mostly stays at 64.

Compared to freebsd http://www.users.on.net/~boddingt/ntp/freebsd/ on the same 
hardware linux is performing quite badly. With freebsd I used the default 
minpoll maxpoll.

I skipped 2.6.21 and 2.6.22. Ran freebsd instead. The last linux I used that 
could change it's clock quickly enough was 2.6.17. From 2.6.18 to 2.6.21-rcX 
using minpoll 6 maxpoll 10 on the refclocks meant the time was at least +/- 
300us when freebsd would be within 4us on the same hardware. Still have to find 
out what the current kernel will do.

I dislike the current convergence rate when using a refclock but it did work 
well when last I tried it over dialup.

-- 
    James