[LinuxPPS] Which patch to use?

[Jordan Hayes]

>> Next I did
>>
>> # ./ppsctl /dev/ttyS0 enable
>
> Now there is a device.

Here's what I see now:

# tree /sys/class/pps/
/sys/class/pps/
|-- pps0
|   |-- assert
|   |-- clear
|   |-- dev
|   |-- echo
|   |-- mode
|   |-- name
|   |-- path
|   |-- power
|   |   `-- wakeup
|   |-- subsystem -> ../../../class/pps
|   `-- uevent
`-- pps1
    |-- assert
    |-- clear
    |-- dev
    |-- device -> ../../../devices/pnp0/00:07
    |-- echo
    |-- mode
    |-- name
    |-- path
    |-- power
    |   `-- wakeup
    |-- subsystem -> ../../../class/pps
    `-- uevent

7 directories, 18 files

I still don't get a device on pps0 ... so should I be using /dev/pps1 
instead of /dev/pps0?

>> ntpd[22954]: refclock_open /dev/gps0: No such file or directory
>> ntpd[22954]: configuration of 127.127.20.0 failed
>>
>> Is /dev/gps0 supposed to be a symlink or something?
>
> # ls -l *ps*
> lrwxrwxrwx 1 root root      5 2008-05-02 13:58 gps0 -> ttyS0
> lrwxrwxrwx 1 root root      4 2008-05-02 14:04 gpspps0 -> pps0
> crw-rw---- 1 root tty  254, 0 2008-05-02 14:04 pps0

Ok, I have that now (and I see the mention in the wiki, but I don't see 
a mention of /dev/gpspps0 ... is that new?).

Now I get this at startup:

> refclock_nmea: found GPS source "/dev/gps0"
> refclock_nmea: try alternate PPS device "/dev/pps0"
> refclock_nmea: found PPS source "/dev/pps0"
> refclock_nmea: time_pps_kcbind failed: Operation not supported

Which seems pretty good.  But it's not quite there:

# ntpq -p
     remote   refid  st t when poll reach   delay   offset  jitter
==================================================================
 GPS_NMEA(0)  .GPS.   0 l    -   16    0    0.000    0.000   0.002

So reach is 0, which means it's not reading the device correctly?  I'm 
able to "cu -l /dev/ttyS0" and see $GPGGA and $GPRMC sentences from my 
GPS18 (when ntpd isn't running, of course).

/var/log/ntpstats/clockstats is filling up with $GPRMC lines, too.

/jordan