latency spikes, was Re: [LinuxPPS] slightly offtopic

[James Boddington]

Udo van den Heuvel wrote:
> Did somethign change in the kernel. I am runnign 2.6.18 since a few
> days. It looks like the PPS edge (200 ms in length) is part of this game
> now.
> flag2 doesn't matter, though, as a few short tests show.

Looking at http://gitweb.enneenne.com/?p=linuxpps;a=summary the most recent 
patch I have applied is the 'Timestamps are now collected at the beginning of 
"linuxpps_event()" function.' patch from 2 months ago. I don't have the 3 
commits since then.

I applied the patch I had to 2.6.18. The correct edge of the pulse is being 
used for me. Getting the expected offsets to my local machines and 'ntpdate -q 
0.au.pool.ntp.org' is showing the expected results.

I have noticed a different drift value compared to what I had with 2.6.17. 
Maybe around -41 instead of -68.5.

-- 
    James