Re: PPS_SYNC and NTP refclocks

[Frank Kardel <kardel%netbsd.org@localhost>]

On 08/08/12 00:52, Dave Tyson wrote:
> Is anyone successfully running a stratum 1 server using NetBSD current?
Yes. NetBSD 6.99.5 amd64 PPS_SYNC serial PPS via DCD.

     remote           refid      st t when poll reach   delay   offset  
jitter
==============================================================================
oGENERIC(2)      .GPS.            0 l   10   16  377    0.000    0.000   
0.000
 GENERIC(0)      .GPS.            0 l    1    8  377    0.000   
-0.027   0.000
+GENERIC(1)      .DCFa.           0 l   48   64  377    0.000   -0.600   
0.284

The PARSE driver can see PPS information from the serial port.

For GENERIC(0)
associd=38537 status=0040 , 4 events, clk_unspec,
device="Meinberg GPS16x receiver",
timecode="\x0209.08.12; 4; 19:35:22; +00:00;   S    ; yx.0484N   
zz.7878E  206m\x03",
poll=1114, noreply=0, badformat=0, baddata=1, fudgetime1=1.968,
stratum=0, refid=GPS, flags=1,
refclock_ppstime="d3ce8f7a.0001985f  Thu, Aug  9 2012 19:35:22.000",
refclock_time="d3ce8f7a.00000000  Thu, Aug  9 2012 19:35:22.000",
refclock_status="UTC DISPLAY; TIME CODE; PPS; POSITION; (LEAP 
INDICATION; PPS SIGNAL; POSITION)",
...

The difference of 27 usec to GENERIC(2) is that PPS information for 
GENERIC(0) goes
the serial driver PPS_SYNC way, while GENERIC(2) uses s direct driver 
for a PCIe card.

Did you try to run the  src/external/bsd/ntp/dist/util/pps-api.c program 
to check?

Frank