CPUFreq HAL [PSARC/2007/679 FastTrack timeout 12/12/2007]

[Randy Fishel]

> 
> Randy Fishel wrote:
> > On Wed, 19 Dec 2007, Darren J Moffat wrote:
> > 
> >> In addition to what Gary has said about the auth names I don't think a new 
top
> >> level node in the auth hierarchy is needed here.
> >>
> >> I would have expected it see something like: solaris.system.power.cpu.
> >>
> > 
> >   There is a proposal in the works for a 'solaris.powermanagement', 
> > and cpupm would (effectively) be the first element (followed later by 
> > other power management related items).  If the advice is that it is 
> > better suited at solaris.system.powermanagement (or even shortened to 
> > power), then the project team would be agreeable.
> 
> solaris.system.power. would be my preference.
> 
> I personally think that the extra word "management" is redundant since 
> there this is an authorisation you are implicitly managing something.

  I am OK with this, and it fits into the schemes.

> 
> On "cpu" vs "cpupm" what does the pm suffix add ?  Is there likely to be 
> other cpu<foo> things under this hierarchy ?
> 

  To some extent, the 'pm' part is also redundant, but was changed as
'freq' seemed too specific.  However, there is also some engineering
going into disabling cores, so there is the possibility that there
could be frequency control and core/chip disabling in the future.
However, it could also be argued that they both fit into the
system.power.cpu category.

  So again, I am OK with it being solaris.system.power.cpu, as this is
the current expected individual set, and if/when another set comes up
that requires something different, it can be revisited then.


	---- Randy