[LinuxPPS] Re2: some problems with linuxPPS + GPS system

[Hal V. Engel]

On Wednesday 10 September 2008 02:15:40 am david wrote:
snip
> > First you need to confirm that your pps signal is actaully working.  
> > Have you tried?
>
> Yes it seems to work
>
> >  ppstest /dev/pps0
>
> trying PPS source "/dev/pps0"
> found PPS source "/dev/pps0"
> ok, found 1 source(s), now start fetching data...
> source 0 - assert 1221037543.828813020, sequence: 88131 - clear
> 1221037543.028815792, sequence: 88130
> source 0 - assert 1221037543.828813020, sequence: 88131 - clear
> 1221037544.028833694, sequence: 88131
> source 0 - assert 1221037544.828845730, sequence: 88132 - clear
> 1221037544.028833694, sequence: 88131
> source 0 - assert 1221037544.828845730, sequence: 88132 - clear
> 1221037545.028867802, sequence: 88132
> source 0 - assert 1221037545.828853296, sequence: 88133 - clear
> 1221037545.028867802, sequence: 88132
> source 0 - assert 1221037545.828853296, sequence: 88133 - clear
> 1221037546.028870339, sequence: 88133
>

This looks good and it is the GPS PPS signal.  Notice how the assert happens 
about 200 milliseconds before the clear for the same sequence.

>
> Can you explain the difference between pps0 and pps1 ?
>
> sonde2:/usr/src/linux/Documentation/pps# ./ppstest /dev/pps1
> trying PPS source "/dev/pps1"
> found PPS source "/dev/pps1"
> ok, found 1 source(s), now start fetching data...
> source 0 - assert 1221037584.290399089, sequence: 88293 - clear
> 0.000000000, sequence: 0
> source 0 - assert 1221037585.290096541, sequence: 88294 - clear
> 0.000000000, sequence: 0
> source 0 - assert 1221037586.289778347, sequence: 88295 - clear
> 0.000000000, sequence: 0
> source 0 - assert 1221037587.289464624, sequence: 88296 - clear
> 0.000000000, sequence: 0
> source 0 - assert 1221037588.289143358, sequence: 88297 - clear
> 0.000000000, sequence: 0

This is the fake ktimer PPS signals.  Note that there is no clear events 
(clear 0.000000000, sequence: 0) for the interupt since the ktimer module only 
has asert events and not clear events.  Since you have a working refclock PPS 
signal you should unload ktimer and stop using it.

snip

> > If you are not getting the correct results for ppstest then we can help
> > you get that working.  Once that is working the rest is fairly easy.
>
> this is the reason why I'm afraid !

You didn't give us any information on how you built ntp other than that you 
patched it and followed the wiki.  You do have to pass it additional configure 
parms to make it build with PPS support for your refclock and this is covered 
in the wiki.  What configure command did you use for NTP?  Did you review the 
output from the configure command to make sure that it found what was needed 
(not covered in WIKI)?  In addition you need to make sure that it is picking 
up the right kernel and sys/timex.h headers during the build.  What other 
things did you do to setup the build?

I am working on a LinuxPPS howto for this stuff but it is not complete at this 
point.  It currently covers everthing the wiki does for those who are patching 
an existing kernel (IE. it does not cover the git stuff) but only for kernels 
starting with 2.6.26 and later.  In addition, this material has some changes 
compared to the wiki that reflect some of the things that list members have 
discovered as well information from other sources such as the OnCore setup 
stuff from the Remco blog.  I currently have about 8 pages in the howto but I 
need to spend some more time going through the emails on this list to flesh 
out things like setting up udev, testing ntpd and ntpd startup scripts and to 
see if there is additional information that needs to be added to or changed in 
the existing sections.   Plus I need to spend time cleaning up the document.

This howto should make it easier for those, like you, who are trying to get 
this working for the first time since I think what I have is a little clearer 
and simpler (where this is possible) and goes through things in a step by step 
fashion.  If you would like I can send you a copy off list with the 
understanding that it is far form complete, likely has some significant 
ommissions and maybe even some stuff that is completely wrong.  I started 
working on this about two weeks ago but then got side tracked.  But I am going 
to start working on it again today and I hope to have something I can share 
with the list later this week.

Hal
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://ml.enneenne.com/pipermail/linuxpps/attachments/20080910/a48879f2/attachment.htm