[LinuxPPS] unofficial Test patch for linux v2.6.30-rc6

[William S. Brasher]

On Thu, 21 May 2009, Rodolfo Giometti wrote:

> On Wed, May 20, 2009 at 09:36:07AM -0500, William S. Brasher wrote:
> > On Wed, 20 May 2009, Rodolfo Giometti wrote:
> > 
> > > On Tue, May 19, 2009 at 02:13:09PM -0500, William S. Brasher wrote:
> > > > 
> > > > I have installed kernel 2.6.30-rc6 patched with ntp-pps-v2.6.30-rc5 and 
> > > > pps seems to be working fine with ntp-4.2.4p7 and glibc-2.9.
> > > > 
> > > > 
> > > > The first time I ran ppsldisc I got the following in the logs:
> > > 
> > > Did you got this error only once or each second?
> > > 
> > 
> > This happened one time, the first time I ran ppsldisc.  The warning has 
> > not repeated.  Shortly after I started ppsldisc I ran ppstest, and it 
> > reported pulses arriving as expected.
> > 
> > Ntp did start at that time because I had not yet added a rule to udev to 
> > link gps0 to /dev/ttyS0.
> > 
> > I also run setserial on the serial device, but I hadn't transfered a copy 
> > to the machine at the time I was testing pps.  That means that most likely 
> > ppsldisc was the first thing that requested access to the serial port.
> 
> Can you replicate the error condition?
> 
> Ciao,
> 
> Rodolfo
> 
> -- 
> 
> GNU/Linux Solutions                  e-mail: giometti at enneenne.com
> Linux Device Driver                          giometti at linux.it
> Embedded Systems                     phone:  +39 349 2432127
> UNIX programming                     skype:  rodolfo.giometti
> 
> 


The error happened one time, and one time only, the first time I ran 
ppsldisc.  The box had been shutdown for several days; I booted the box 
under an older non-pps kernel, transferred over the new kernel, shutdown 
and restarted the box.

Then I stopped the running, stock ntp and transferred over and installed a 
copy of the latest ntp patched with the nmea patch.  I also installed the 
new ppsldisc and ppstest binaries.  At that time I started ppsldisc, and 
then ran ppstest and noted the presence of pulse-per-second in the output.

About that time I got ready to start ntp, and noticed the warning message 
in the log that showed up the first time the serial port was accessed.  I 
haven't been able to replicate that.  Grrr.

The gps clock is a Garmin gps18lvc 5m sitting inside on a shelf, powered 
from the usb port on the Via.  The box had been powered down several days, 
so the gps had also been off for a few days.  Until ntp was started the 
serial port's speed was "set wrong" for the gps, so I couldn't read any 
nmea strings.

About that time I transferred over a setserial binary and installed a new 
udev rule, and started ntp.

Since then I've installed a new kernel with the latest pps patch and have 
rebooted the machine with and without ntp, and although ntp has started 
every time, and the serial port has recognized the pps signal every time- 
I have seen no failure of any kind.  Ntp just starts, and with SHIFT_PLL 
patched, syncs very quickly.