[ntp:hackers] NEMA refclock, PPS and Solaris.

David Mills mills at udel.edu
Sat May 2 00:21:15 UTC 2009


Currently, Spectracom WWVB and GPS, Arbiter GPS, Audio IRIG, WWV and 
CHU, Traconex WWV, Hopf GPS. Those that worked in the past in Solaris, 
but are no longer available include TrueTime WWVB and GOES, Heath WWV, 
KSI/Odetics IRIG and Austron GPS. I think the ost important one now is 
the NMEA driver, but it should be furbished with the real or equivalent 
functionality of the new interface to look for satellite or antenna 
failures and PPS failures. It should also be equipped/tested for leap 
second insertion, if that is possible.

The Garmin OEM receivers are really cheap and I would expect them to 
become very popular.


Brian Utterback wrote:

> I'll give that a try. Can someone tell me which Solaris systems at 
> UDEL have refclocks. I need to make a list of refclocks known to work 
> with Solaris/OpenSolaris and late model NTP 4.2.5. Can I log into the 
> mentioned systems with my account from pogo?
> Dave Hart wrote:
>> On Fri, May 1, 2009 at 6:38 PM, Brian Utterback 
>> <brian.utterback at sun.com> wrote:
>>> May  1 10:10:34...refclock_nmea: time_pps_setparams failed: Invalid
>>> argument
>>> Looking at the code for time_pps_setparams, it appears that it is
>>> complaining either about the PPSAPI version not being 1, or that the
>>> offset isn't in the right units.
>> Looking at refclock_nmea.c, it does not set the version number before
>> calling time_pps_setparams.  While there are plenty of Solaris
>> machines with reference clocks at udel, I don't think any of them are
>> using refclock_nmea.  If one were this bug would have been noticed a
>> long time ago.  Add a line to set the version after the memset and you
>> should be good.  Hopefully this PPSAPI code will be leaving
>> refclock_nmea.c before long in favor of Dave Mills' new centralized
>> PPSAPI capability for non-atom refclocks.
>> Cheers,
>> Dave Hart

More information about the hackers mailing list