On the accomplished day of Nov 9, Gary Thorpe said:

> --- "Bryan J. Phillippe (spamblock)" <bryan-spamtrap5@darkforest.org>
> wrote:
...
> > I recently posted about a problem I have on a PIII notebook running
> > NetBSD (*) where the laptop shuts off about two hours into a ./build of
> > current src.  After doing a bit more research, it appears this is the
> > standard behavior of the PIII when it begins to overheat - it will shut
> > off.  Most likely my notebook BIOS/mainboard detects this and powers
> > the whole system down.
>
> Does the laptop have a fan and does it turn on after heavy usage? The
> only way to prevent a power-off is to cool the system down either by
> throttling it back and/or turning on a fan. Usually the first will happen
> automatically before it shuts down, but I don't think NetBSD can detect
> clock speed changes at run time (nor Linux nor FreeBSD).

Yes, the fan is actually a multi-speed fan which is normally off
altogether.  I do a little work (launch mozilla and an xterm), it comes on.
Do a lot of work (start X, start a build), it comes on full.  I will run
some more experiments, if I have time this weekend, by trying additional
forced air and/or cooled air to see if that changes things.  I might even
try running it outside here; it's a bit chilly this time of year in the
Pacific Northwest USA. :-)

> > Is there not some kind of interrupt or event raised by the CPU when it
> > begins to overheat, that we can trap on and back off the scheduler or
> > something?
>
> Hardware monitors maybe? There is some sort of support in NetBSD I think,
> although I don't know the drivers can do/support some sort of
> notification or automated response.
>
> > (*) Tried latest current with ACPI patches.  The laptop is an HP
> > pavilion zt1230.  CPU reports as:
> >
> > cpu0 at mainbus0: (uniprocessor)
> > cpu0: Intel Pentium III (Tualatin) Pentium III Xeon (686-class),
> > 1333.20 MHz
>
> Maybe this is actually a bug in the cpu identification code (how many
> *Xeons* are shipped in laptops)? It probably doesn't relate to power
> issues, since both types of CPUs would be put in ACPI-capable systems.

Yeah, actually it is a bug.  It hasn't bothered me enough to say anything
yet, but the CPU is actually a Celeron.  Here is what XP says about it:

PROCESSOR_ARCHITECTURE	x86
PROCESSOR_IDENTIFIER	Family 6 Model 11 Stepping 1, GenuineIntel
PROCESSOR_LEVEL		6
PROCESSOR_REVISION	0b01

> > cpu0: features 383f9ff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,SEP,MTRR>
> > cpu0: features 383f9ff<PGE,MCA,CMOV,FGPAT,PSE36,MMX>
> > cpu0: features 383f9ff<FXSR,SSE>
> > cpu0: I-cache 16 KB 32b/line 4-way, D-cache 16 KB 32b/line 4-way
> > cpu0: L2 cache 256 KB 32b/line 8-way
> > cpu0: ITLB 32 4 KB entries 4-way, 2 4 MB entries fully associative
> > cpu0: DTLB 64 4 KB entries 4-way, 8 4 MB entries 4-way
> > cpu0: 8 page colors

-bp
--
# bryan_at_darkforest_dot_org
# Software Engineer