Re: [gpsd-dev] What is our current bug state?

[Gary E. Miller]

Yo Hal!

On Sun, 01 Feb 2015 20:05:51 -0800
Hal Murray <address@hidden> wrote:

> 
> address@hidden said:
> > >This confirms Gary's fix to the PPS code.  I am no longer seeing a
> > >garbaged PPS offset. I'm still getting PPS from the Macx-1/GR-601W.
> 
> > Well, that was true for a short while.  Now I'm seeing a garbage
> > offset again. 
> 
> I'm using HEAD from a few minutes ago.
> 
> gpsmon to gpsd shows no PPS info.
> 
> gpsmon going direct to the hardware sees PPS.  Occasionally I see
> things like: PPS offset: -1422849089.00008
> I'm not sure what units that is using.

I've not looked in gpsmon, never use it myself.  But what you see is
obvious.

The units are seconds.  When you ran your test your current UNIX
time in seconds was 1422849089.  So it thought the PPS pulse came it at
UNIX time zero, and subtracted for the result.

> The log file shows most of the PPS lines after GPRMC lines.
> Occasionally it comes in the middle of a GPGSV burst.  I just tried
> again and the "occasionally" above matches the 1 out of 5 density of
> the GPGSV info.

The PPS lines are generated in a separate thread than the rest.  So
there will be only a casual relationship between them.
 
> I'm using a MR-350P running in NMEA mode at 4800.  I think it's a
> reasonably vanilla SiRF device.  The NMEA stuff is normally offset
> about 1/2 second. I'll get timings if that will help.

4800 is too slow.  You should bump that up.  No wonder you are missing
sentences, the serial speed is too slow for them all to fit.

And a quick look at gpsmon.c shows the ppsthread (only used in direct
mode) is just not handled properly.  Missing thread locking, missing
mutext init, last_fixtime is missing and likely more.  If it were me I
would just rip it out and fix the client mode.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97701
        address@hidden  Tel:+1(541)382-8588