Re: [gpsd-dev] PPS in 3.11

[Miroslav Lichvar]

On Tue, Aug 26, 2014 at 02:35:21PM -0700, Gary E. Miller wrote:
> Uh, oh.  I see Miroslav's patch broke some unstated assumptions.

It restored the original behavior before PPS was broken by refactoring
in the commit 575444. It's sad to see every other release of gpsd with
broken PPS.

> His patch used the CALCULATED time of the last NMEA timestamp
> (last_fixtime_real).  Not the RECEIVED time of the last NMEA
> timestamp (last_fixtime_clock).

No, please check the code again. The delay between PPS and NMEA is
now calculated from the captured timestamps (last_fixtime_clock and
drift.clock variables). That's how it used to be when it worked
correctly.

As a test I recommend to configure chronyd or ntpd to use the PPS
source as the only source. Start with a larger initial offset, try it
in both directions (e.g. date -s '+ 10 sec' and date -s '- 10 sec')
and see if it can fix the clock.

-- 
Miroslav Lichvar