From mboxrd@z Thu Jan  1 00:00:00 1970
To: 9fans@cse.psu.edu
Subject: Re: [9fans] Support for Mac OS X Leopard
From: "Russ Cox" <rsc@swtch.com>
Date: Sun, 30 Sep 2007 17:09:09 -0400
In-Reply-To: <14ec7b180709301127i4bdae0bcs276f771355322816@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Message-Id: <20070930210912.CB17D1E8C1C@holo.morphisms.net>
Topicbox-Message-UUID: c87ab7d4-ead2-11e9-9d60-3106f5b1d025

>> the right thing to do is to fix timesync to notice
>> such jumps and write them off as suspend/resume
>> instead of assuming that the cpu has gotten *very* slow.
>
> timesync still needs to set the correct time (a system which was
> suspended for, say, 10 hours will live 10 hours in the past). perhaps
> devcons.c:^writebintime should be the one to skip updating the clock
> frequency?

timesync is in charge of figuring out the
correspondence between cpu frequency
and real time.  the kernel does what timesync
says.  instead of the kernel second-guessing
timesync, timesync should just be fixed.

sometimes clock speed *does* vary w.r.t. real time.
if i switch from ac to battery on a laptop,
then timesync helps navigate the change
in cpu frequency.

russ