svn commit: r219700 - head/sys/x86/x86

[Bruce Evans]

On Thu, 24 Mar 2011, Warner Losh wrote:

> On 03/23/2011 06:34, John Baldwin wrote:
>> On Thursday, March 17, 2011 5:06:09 pm Jung-uk Kim wrote:
>>> I really hate the idea of adjusting timecounter frequency from
>>> userland.  I guess "use ntpd(8)" is not a good answer for some
>>> people. :-(
>> Actually, that doesn't work well if your timecounter frequency is off by a
>> lot.  Having the timecounter frequency accurate improves the accuracy of
>> things like ntpd and ptpd.
>
> ntpd requires that the time counter be within 128ppm of true.  If the time 
> counter guess is off by more than that, you lose: ntpd won't work.

Is ntpd really that broken?  What does it do if the drift file has the
correct compensation to within 128 ppm?

I use an old version of ntpd in which I've observed a positive feedback
loop when -x is used to prevent steps and the slew wants to be more than
128 ppm due to a transient.

Bruce