Detecting the need for additional kernel options?

[Jarod Wilson]

Dave Jones wrote:
> On Sat, Feb 24, 2007 at 12:34:12PM -0500, Michel Salim wrote:
>  > Is Anaconda currently checking the machine configuration against a
>  > compatibility database? On some AMD64 laptops (including my HP L2000),
>  > the machine would lock up randomly (without any log message) if booted
>  > without "noapic" (I added "nosmp" for good measure).
>  > 
>  > Perhaps a user-maintainable database; we can consider tying it in with
>  > Smolt so that only users who own a particular model can submit
>  > required kernel options for it (to verify ownership, give the user a
>  > token that they must pass to Smolt, e.g. smoltSendProfile
>  > --key=AF32EBD4008)
>  > 
>  > Speaking of Smolt, the CPUspeed reported does not seem that useful,
>  > unless it interfaces with CPUspeed to first set the clockspeed to
>  > minimum, take a measurement, then set it to maximum, measure, and then
>  > restore the original setting? My 2GHz laptop was reported as a 1.6 ..
> 
> instead of faffing about changing speeds, it could just report
> the contents of /sys/devices/system/cpu/*/cpufreq/scaling_available_frequencies
> if present.

+1.

Or if its preferable to only have the max speed, just grab it from 
/sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq if present, and 
fall back to looking in /proc/cpuinfo scaling_max_freq isn't there.

-- 
Jarod Wilson
jwilson at redhat.com