Re: [gpsd-dev] Are we in good shape for a 3.11 release tomorrow?

[Gary E. Miller]

Yo Eric!

On Sun, 1 Dec 2013 16:36:42 -0500
"Eric S. Raymond" <address@hidden> wrote:

> Gary E. Miller <address@hidden>:
> >                                                             If I run
> > the serial speed on a GR-601W to 38,400 or more, gpsd sometimes get
> > the serial timestamp for the new second before the PPS is
> > detected.  Leading to an off by one error.  Easy workaround is set
> > the speed to 19,200 or less.
> 
> That is a very interesting bug!

It may not be quite that bad, another possibility is that the PPS
is seen first, but some sort of locking lets the serial time complete
first.

Here is an easy way to see the issue:

# gpsctl -s 9600 -f /dev/ttyUSB2
# stty speed 9600 < /dev/ttyUSB2
# gpsd -nND9  /dev/ttyUSB2 |& grep 'NTP' 

You will see pairs of reports, the first should be the PPS time, then
the serial time, the a one second pause, then repeat.  So here is the
output when ntpd is happy:

gpsd:RAW: NTP ntpshm_put(1) 1385936130.000000000 @ 1385936130.001522976
gpsd:RAW: NTP ntpshm_put(0) 1385936131.000000000 @ 1385936131.131315384
[pause]
gpsd:RAW: NTP ntpshm_put(1) 1385936131.000000000 @ 1385936131.001563636
gpsd:RAW: NTP ntpshm_put(0) 1385936132.000000000 @ 1385936132.132196645

But at high speeds it will may go wonky and they get inverted, PPS time
last:

# gpsctl -s 57600 -f /dev/ttyUSB2
# stty speed 57600 < /dev/ttyUSB2
# gpsd -nND9  /dev/ttyUSB2 |& grep 'NTP' 

gpsd:RAW: NTP ntpshm_put(0) 1385935441.000000000 @ 1385935441.108278479
gpsd:RAW: NTP ntpshm_put(1) 1385935441.000000000 @ 1385935440.116362834

Clearly something I'm missing...  Not a new bug, I'm just catching it
more.

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

signature.asc
Description: PGP signature