[prev in list] [next in list] [prev in thread] [next in thread] 

List:       netbsd-port-i386
Subject:    Re: RTC and clocks
From:       seebs () plethora ! net (Peter Seebach)
Date:       2006-09-19 1:57:34
Message-ID: 200609190157.k8J1vYNS025893 () guild ! plethora ! net
[Download RAW message or body]

In message <4F2C572A-9DE4-41D8-B80E-94D6AD7EB5E0@invisible.ca>, "Jared D. McNei
ll" writes:
>Are you running estd (from pkgsrc)? Or, was the 35:60 from when you  
>had booted running from battery maybe?

No estd.

The odd thing is that it would change over time.  (Note that "35" is an
approximation; I saw everything from about 32 to 60.)  It acted, in short,
exactly the way I'd expect an unreliable clock with a failing battery to act,
up to and including working better right after a reset, and then gradually
falling apart.

>In any case, you should use an alternate timer if you are using  
>speedstep -- TSC is known not to handle changes in CPU speed  
>properly. This can be done by, eg:

>	# sysctl -w kern.timecounter.hardware=i8254

>You probably want to use the ACPI timer for finer granularity, if  
>possible.

Hmm.  I may have speedstep disabled or something; certainly, it isn't getting
used while I'm plugged in.

I have never enabled ACPI on this system, because my experiences with ACPI
(admittedly, years back) were so consistently negative.  Maybe I should try
that.

-s
[prev in list] [next in list] [prev in thread] [next in thread]