[ntp:questions] Meinberg NTP monitor, silly question

Dave Hart davehart at gmail.com
Mon Dec 21 15:43:19 UTC 2009

On Dec 21, 13:43 UTC, G8KBV wrote:
> That didn't change anything either, other than the registry setting type
> is now 'REG_EXPAND_SZ' like yours, the com port properties still show
> the original MS serial.sys file as the driver, and NTPD is still
> resolutely using usermode PPS timestamps as a result.

serialpps.sys is not a freestanding driver, and does not have its own
installer or GUI properties DLL.  Instead, it is a very lightly
modified serial.sys which is "installed" via a cheap trick of pointing
the serial.sys ImagePath registry entry at serialpps.sys instead of
serial.sys.  What you see is normal.

Further, despite the fact that this thread started out with you asking
what the 'o' meant in the ntpq peers billboard, which indicates
working PPSAPI (and hence working serialpps.sys), you've been
mistakenly assuming that because you see a message about user-mode
PPS, that means serialpps.sys is not working.  As long as one of your
refclocks has 'o' in the billboard, you know it is successfully
talking to serialpps.sys via PPSAPI.  The fact that the NMEA refclock
is using user-mode PPS timestamps does not imply the PPS/atom driver
is not.  In fact, the atom driver will only work at all on systems
with functional PPSAPI.  The "user-mode PPS" message _does_ mean you
could get a respectable result without using PPS(2)/atom/PPSAPI/

Dave Hart

More information about the questions mailing list