[ntp:questions] strange time behaviour

Terje Mathisen "terje.mathisen at tmsw.no" at ntp.org
Tue May 26 13:06:31 UTC 2009

vhfmeyer at t-online.de wrote:
> That was my first thought, too. But 2^32 milliseconds give 49 days and
> 17 hours. Where are those 28 hours gone? Since the system is ntp-
> controlled, imprecision can't be the reason. I have some amount of CAN-
> bus and ethernet interrupts too, so maybe the total interrupt count
> could have been 2^32. But there is no total interrupt counter inside
> Linux. So where are those 28 hours gone?

Could this be a counter which is based off the CMOS clock?

The basic frequency of this chip is 2^15 Hz, and it can interrupt at any 
power of two fraction of this, i.e. 1024 Hz is common.

2^32/(1024*60*60*24) = 48.545, or 48 days 13:05 hours. The cmos clock 
frequency is not modified by ntp, so you could see at least some minutes 
more or less.


- <Terje.Mathisen at tmsw.no>
"almost all programming can be viewed as an exercise in caching"

More information about the questions mailing list