Re: NSLU2 clock drift
On Fri, May 18, 2007 at 03:58:51PM -0600, Gordon Farquharson wrote:
> One had to use 'tickadj 10101' with kernel version 2.6.17 on the
> NSLU2 because it did not contain the clocksource subsystem.
The first part of your sentence might be true, but the second part
isn't quite. I.e. merging of the clocksource stuff wasn't the
necessary step and only possible way for the NSLU2 to have accurate
timekeeping.
(Some time ago, someone suggested changing the ixp4xx time keeping
code to do the exact same thing that the time keeping code on other
ARM platforms does (where those other ARM platforms already supported
run-time variable timer tick rates), but that idea got shot down by
the NSLU2 people.)
Reply to: