On Fri, 12 Oct 2007 04:34:23 -0400, Dave B <spam@y2007.dberg.net> wrote:
> On Thu, Oct 11, 2007 at 07:03:07PM -0700, Scott Ellis wrote:
> ...
>> There's no mention of "SpeedStep" there, even though EST and TM2 are in
>> the features2 list.  After boot, there's no machdep.est.* branch.
>>
>> I've tried enabling and disabling EIST in the BIOS (as well as the silly
> ...
> 
>   My laptop's NetBSD4/i386 installation was also refusing to offer
> me the machdep.est.* branch--but only when it was running on
> battery.  When plugged in to AC instead, est.* items were avail-
> able.

Interesting.  When you booted on AC, do you recall if the feature was
masked
out of the features2 list, or if you got a note on boot?  From my reading
of
identcpu.c it looks like if the EST feature is listed and it's an Intel
CPU,
then you should at least get a warning that the BIOS has disabled
SpeedStep:

#ifdef ENHANCED_SPEEDSTEP
        if ((vendor == CPUVENDOR_INTEL) &&
            (ci->ci_feature2_flags & CPUID2_EST)) {
                if (rdmsr(MSR_MISC_ENABLE) & (1 << 16))
                        est_init(CPUVENDOR_INTEL);
                else
                        aprint_normal("%s: Enhanced SpeedStep disabled by "
                            "BIOS\n", device_xname(ci->ci_dev));
        }
#endif

I like your data-point, as I want to make sure the BIOS is correct (I can
live 
without EIST for awhile, so long as I don't need to change the BIOS later) 
since the machine runs headless and doesn't have serial BIOS redirection. 
Maybe 
I'll sprinkle some aprint_debug() messages in there to see how far it gets.

     ScottE