Re: early printk timings way off

[Jesper Juhl <jesper.juhl@gmail.com>]

On 9/16/05, Rogier Wolff <R.E.Wolff@bitwizard.nl> wrote:
> On Fri, Sep 16, 2005 at 01:00:45AM +0200, Jesper Juhl wrote:
> >
> > I'll just dig into it myself for now, but thank you, if I get really
> > stuck I may ask him.
> 
> The explanation: "jiffies starts at rollover minus a bit" seems to be
> spot-on: If jiffies are 32bit, and counting at 1000 per second, the
> 2^32 / 1000 works out.
> 
> I expect the kernel to run without turning on (timer) interrupts for a
> while during boot: It is still initializing things like memory and the
> processor. Without those, interrupts won't work. This means that the
> timer interrupt will not count in real-time.
> 
> A "jump" of 27 seconds seems unlikely, except if somehow the
> interrupts are somehow accounted. It could very well be that the
> kernel nowadays has a mechanism of measuring the fact that it missed a
> timer interrupt and corrects for that. This would mean that around the
> "jump", the kernel suddenly realized it missed around 27000 interrupts
> and added 27000 to "jiffies"....
> 
> I'd say: Would be nice to get the timings right, but not worth the
> trouble: There are good technical reasons for the observed facts.
> 
Thank you for that explanation Rogier.

-- 
Jesper Juhl <jesper.juhl@gmail.com>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html