[SGVLUG] CPU speed scaling

[Dustin]

On Sat, 26 Nov 2005, Mike Fedyk wrote:

> >better.  I suspect it was loading the wrong laptop driver.

> Don't let this go unreported to a BTS.  At the very least someone will 
> be able to find it there if they run into the problem.

Yeah, I suppose I should.  I suppose I should also report the #$$%^^%&$%^
eth1394 driver issue, huh?  I half suspect the latter is only triggered by
me personally, however, perhaps through some insidious wetware-detection
scheme in the kernel. :-)

What kills me about eth1394 is it has to be a very rarely-used driver (as
you noticed, nobody claimed to have ever used ethernet-over-firewire) and
I'd think anyone using it would very well know how to load it manually.

Back to the cpu scaling conflict, when I had no clue what was causing it I
was going to file a bug with the acpi_cpufreq maintainer, but since it
appears to be a driver loading conflict it could be hardware detection and
I should send it directly to the Gentoo BTS and let them decide whether to
kick it further upstream.

> If you can, do a binary search on the options you removed to find the 
> config option that triggered the problem.

It's just compilation time, I guess. :-)  I already have the compile down
to 16.5 minutes, though, so it ain't that bad.  (The default kernel took
22-23 minutes, as opposed to the stock Debian kernel's 50 min on a
supposedly faster machine--I should check and see if maybe the laptop is
actually that much faster than the desktop in actual use, due to better
memory bandwidth or the huge Pentium-M L2 cache or something.  I once had
a (much more thoroughly)  stripped-down Gentoo kernel compiling in 7
minutes on the desktop, though, so I can't see how it would be that much
slower.)

Dustin