Re: SMP problems with -CURRENT? (and BETA5)

On Tue, 21 Sep 2004 06:11:33 +0200, Johann Hugo
<jhugo_at_icomtek.csir.co.za> wrote:
> On Tuesday 21 September 2004 02:26, Adam K Kirchhoff wrote:
> > What makes you think it's the cpu throttling that's causing the lockups
> > for me?  It does say "acpi_cpu: throttling enabled, 8 steps (100% to
> > 12.5%)"  but it continues on the same line with "currently 100%"  Maybe
> > I'm reading it wrong but that would indicate to me that it's still
> > running at 100%....
> >
> > In fact, I just checked and BETA4, which works fine, also displays the
> > same line about cpu throttling.
> >
> > Any other ideas?  I'd really like to make sure that this problem doesn't
> > continue beyond BETA5 :-)  Anything I can do to make sure this gets
> > resolved (before I go and send a problem report)?
> >
> > Adam
> 
> I had a similar problem. If you do a verbose boot you can see that it
> actually continues past  "acpi_cpu: throttling enabled ...." and it stops
> with "ata0: resetting done" . Looks like some problem with the apic.
> 
> I've made a new kernel without "device apic and options SMP " and now it boots
> fine.
> 
> Johann
> 
> 
> 
> 
> _______________________________________________
> freebsd-current_at_freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscribe_at_freebsd.org"
> 

That's disabling SMP and HTT support completely. Which means it's now
operating as a uniprocessor (sp?). In order for HTT to work it has to
have SMP + apic. One of the other things (from tests I've done here)
is that I would look into playing with the MPSAFE enable code.

(grep MP -r /usr/src/sys/i386/ | less)

As for the throttling issue, I've never seen it throttle down like
that and then back up on an init under 5.3-BETA2 or BETA3. I did start
seeing problems with it under BETA4 and then BETA5. It doesn't do it
under 4.10/4-STABLE either.

Now with the ata0 comment, that's it resetting the actual controller
(not the drives) Now that COULD indicate a problem with anything DMA
registration (not sure if that's the right word for it) to ACPI. Yes I
know this all raises more questions than it answers, however I'm
hoping some of the devels on the list will step forward and either
tell *me* I'm on crack and express the why of that, or explain in
better details than I can. (I'm not a developer, I manage systems,
nothing more.)

-- 
David D.W. Downey