S3 experience on Thinkpad 570E

[John Baldwin]

On Tuesday 28 December 2004 06:20 pm, Andrea Campi wrote:
> On Tue, Dec 28, 2004 at 10:27:57AM -0500, John Baldwin wrote:
> > > > In a nutshell, acpiconf -s3 powers down; on powerup however the
> > > > machine is deadly slow. I replicated this with a stripped down
> > > > kernel; the only thing that was evidently wrong is that the clock
> > > > slowed down from 1000 to around 250 interrupts per second.
> > >
> > > It sounds like the clock interrupt source is not getting saved/restored
> > > properly if it slows after a resume.  I am not sure how to solve this.
> > > Perhaps John has something to add.
> >
> > Currently on i386 the clocks are not real new-bus devices and I'm not
> > sure if they have proper resume support.  I think we do have some sort of
> > hardcoded call to the clock code on i386 to resume the clocks but I'm not
> > sure.
>
> OK, thanks. Just for completeness, I'll check and report what
> happens with a different value of HZ (100, for instance ;-)), just in
> case this makes any difference--as I would expect if this is a case
> of resuming to some hardcoded value.

Actually, device pmtimer will ensure that the clocks are reprogrammed 
correctly.  You didn't leave that out of your kernel config did you?  (I 
think we should just make that standard like it is in 4.x rather than 
optional.)

-- 
John Baldwin <jhb at FreeBSD.org>  <><  http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve"  =  http://www.FreeBSD.org