Re: Help needed to configure and compile gpsd to run on a Raspberry Pi 4 and NOT use systemd

[Mick Durkin]

Hi Gary & Mike,

We implemented Mike's suggestion about adding the "nohz=off" parameter
to the end of the single line in "cmdline.txt"and rebooted the Pi.

Having reviewed the "gpsd -nN -D 4 /dev/ttyAMA0" log from yesterday,
it looked like pps0 was sending data to SHM(1), so we issued "gpsd -n
/dev/ttyAMA0" and "ntpd -gN" commands (ntpd was a fresh install from
NTPsec's git) and the output from "ntpq -p" looks good(ish).

Here is a single shot view:-

[root@pihole:~]$ ntpq -p

     remote                                   refid      st t when
poll reach   delay   offset   jitter
=======================================================================================================
*SHM(1)                                  .PPS.            0 l    1
2  377   0.0000  -1.6773   0.0696
+ntp2.wup-de.hosts.301-moved.de          205.46.178.169   2 u    3
16  377  25.5152   1.5299   0.7698
-hc.gommels-bienen.de                    205.46.178.169   2 u    2
16  377  27.4352  -0.1648   2.7005
-server1a.meinberg.de                    131.188.3.222    2 u    1
16  377  28.1533   0.2503   0.9983
+time.cloudflare.com                     10.67.8.182      3 u    -
16  377  11.6901   1.1580   0.4343
-SHM(0)                                  .GPS.            3 l    7
8  377   0.0000 -130.338   3.2722
 LOCAL(0)                                .LOCL.           5 l    -
64    0   0.0000   0.0000   0.0000
[root@pihole:~]$

This was after a few minutes run, so maybe the offset and jitter
values for PPS will settle down. At least, ntpd is locking on to the
PPS signal and has a couple of externals with a '+' as well.

Maybe we can say that we have a good installation now?

As to Gary's query reference Pi3 and the "dtoverlay=disable-bt"
addition to /boot/firmware/config.txt, that was not any fault of the
gpsd documentation. That came straight from the original YouTube
tutorial follwed back before I got involved with trying to "use gpsd
without systemd and NTPsec instead of chrony". Again, that video might
have been recorded back pre-Pi4. Could it be tied in with changes
brought in with the Pi3 vs earlier devices or new-at-the-time distros?
Certainly, it is still relevant to the Pi4.

If it is considered that our setup is "valid", maybe some of the
pitfalls could be added to the gpsd site documents? We can easily
summarise what we did, both initially (h/w setup) and relevant changes
to the "config.txt" and "cmdline.txt" if it would help.

Can anyone (I reckon Mike is a likely candidate) running a similarly
configured Pi 4 (gpsd not under systemd with an NTPsec derived ntpd)
let us know what kind of offset and jitter  numbers they get from PPS?

BR

Mick