[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gpsd-dev] Problem with PPS detection - using Centos 7 and gpsd dev-
From: |
Mick Durkin |
Subject: |
Re: [gpsd-dev] Problem with PPS detection - using Centos 7 and gpsd dev-3.19a |
Date: |
Mon, 14 Jan 2019 13:27:20 +0000 (UTC) |
Hi Gary,
Here is a follow-up to your first mail.
1) The Perle driver.
I could attach this, but it is a largish source RPM, so it may be better if you
access it directly from Perle's website. I CAN forward it if you prefer. It is
available on https://www.perle.com. Look under the following:-
Support
Downloads
Serial & Parallel Cards
SPEED LE Serial & Parallel Cards
There you will see the Linux Source RPM version 3.92. I will compare this with
the standard driver and look at the references you gave me.
2) Compile gpsd without kpps.
This took a bit of doing, but eventually I was able to make it compile by
modifying SConstruct not to set HAVE_SYS_TIME and, as a result, pacifying
complaints in "timehint.c" and "gpsmon.c". I don't think this made any
difference to the main body of gpsd, as it ran nicely and detected my PPS via
the "ioctl (TIOCMIWAIT)" path.
This proves your conclusion that the Perle driver doesn't handle kernel PPS. I
attach a short file showing the changes I did to hack KPPS support out and then
a log of the successful run detecting the PPS via TIOCMIWAIT. I included the
messages from "NTP:" this time as it shows timevalues with -20 and -30
precision, so it looks like gpsd is shipping both sets of data to ntp.
I will look into the dirver code and if I can find a fix, I will come back to
you.
Thanks for the support.
Mick.
changes_to_suppress_kpps.txt
Description: Text document
pps_test_suppress_kpps.txt
Description: Text document