Re: [gpsd-dev] ntpshmmon

[Nuno Gonçalves]

On Tue, Jun 14, 2016 at 7:12 PM, Gary E. Miller <address@hidden> wrote:
> Yo Nuno!
>
> That looks good to me.  Or at least ntpshmmon is doing the right thing.
> ntpshmmon is showing exactly what is in the SHMs.
>
> That is a bug in gpsd.  I've seen a few other reports of it.
>
> The root problem is gpsd is double sending the NMEA tix time.  That
> got unhinged in the not too distant past.  You need to chase down why
> the gpsd is not locking out the second time.

I think theoretically you could be right, but the example doesn't show
it, but it clearly shows that there is another problem (the one I
explained and provided a fix patch!).

But, if the examples I provided are not clear for you, I apply this
simple patch:

-               if ( (cycle * 500000000LL) <= diff) {
+               if ( (cycle * 100000000LL) <= diff) {

Now let's see how it works:

$./ntpshmmon
ntpshmmon version 1
#      Name   Seen@                Clock                Real
    L Prec
sample NTP0 1465935409.233540853 1465935409.177047007 1465935409.000000000 0  -1
sample NTP0 1465935409.334640004 1465935409.177047007 1465935409.000000000 0  -1
sample NTP0 1465935409.434941240 1465935409.177047007 1465935409.000000000 0  -1
sample NTP0 1465935409.535388837 1465935409.177047007 1465935409.000000000 0  -1
sample NTP0 1465935409.636092791 1465935409.177047007 1465935409.000000000 0  -1
sample NTP0 1465935409.736102470 1465935409.177047007 1465935409.000000000 0  -1
sample NTP0 1465935409.836649010 1465935409.177047007 1465935409.000000000 0  -1
sample NTP0 1465935409.936914260 1465935409.177047007 1465935409.000000000 0  -1
sample NTP0 1465935410.037366757 1465935409.177047007 1465935409.000000000 0  -1
sample NTP0 1465935410.137673610 1465935409.177047007 1465935409.000000000 0  -1
sample NTP0 1465935410.238716935 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935410.339080843 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935410.439727097 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935410.539866472 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935410.640053545 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935410.741046998 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935410.841539104 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935410.942597959 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935411.043270085 1465935410.140829438 1465935410.000000000 0  -1
^C

But yes, GPSD could be sending 10 messages by second, theoretically,
so the best test is just to kill it:


sample NTP0 1465935549.465783808 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935549.566315674 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935549.667419954 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935549.768098164 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935549.868700994 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935549.969222271 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935550.069682243 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935550.169999878 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935550.270526450 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935550.371449272 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935550.472083610 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935550.572688321 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935550.673621804 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935550.774257915 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935550.874902762 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935550.975897652 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935551.076914745 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935551.177021023 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935551.277184099 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935551.377813697 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935551.477992982 1465935410.140829438 1465935410.000000000 0  -1
sample NTP0 1465935551.578657977 1465935410.140829438 1465935410.000000000 0  -1

So GPSD is dead and the messages continue to "arrive".

So, whatever is happening at GPSD is not the problem that I'm
reporting. I think you reached the same conclusion on this thread a
few months ago (date:Tue, Oct 27, 2015 at 6:14 PM). Sorry for not
providing the patch immediately while your memory was fresh.

Thanks,
Nuno