At 3:30 PM -0500 11/20/02, Michael R. Zucca wrote:
>Aaron J. Grier wrote:
>
>>the problem with time slippage on the macs is that it's highly dependent
>>on disk usage and thus more random than systematic.  ntp has a really
>>hard time coping when random exceeds systematic error.
>>
>>basically the SCSI drivers have interrupts turned off so long that clock
>>interrupts are getting lost...
>
>
>I know I must be making horse-paste saying this, but is there a free 
>running counter register somewhere on the 68k Macs that we could use 
>to adjust for lost clock ticks?

Anyone look at crosswiring a serial port?  Connect the Tx clock to 
the Rx data and program it to divide down to a 1 PPS signal. 
(There's a 1PPS serial driver for NTP already.)

I played with this idea conceptually once, but I didn't know enough 
about kernel hacking to make a go of it.  I've also thrown away my 
Z8530 manuals.  It could get you a check on the existing clock, but I 
don't see a solution to the general problem unless the serial ports 
are higher priority than the clock (which they just might be, come to 
think of it).
-- 
The opinions expressed in this message are mine,
not those of Caltech, JPL, NASA, or the US Government.
Henry.B.Hotz@jpl.nasa.gov, or hbhotz@oxy.edu