[lm-sensors] [PATCH 5/5] Package Level Thermal Control and Power Limit Notification: pkgtemp doc

[Fenghua Yu]

On Fri, Aug 20, 2010 at 01:33:56AM -0700, Jean Delvare wrote:
> Fenghua,
> 
> On Thu, 19 Aug 2010 13:51:20 -0700, Fenghua Yu wrote:
> > On Thu, Aug 19, 2010 at 09:27:19AM -0700, Guenter Roeck wrote:
> > > I might spend some time rewriting the coretemp driver as described above,
> > > unless someone else picks it up, and unless there is opposition. 
> > > Obviously, that won't include the package sensor since there is now
> > > a separate driver for it.
> >
> > I agree with this method too. On a multiple socket system, the current coretemp
> > output will cause confusion since it only outputs core# without package#.
> 
> Good point.
> 
> > If it's ok for you, I can rewrite this part to have hwmon device per CPU with
> > both core and package thermal info and send out RFC patch soon.
> 
> Yes, please! If you have time to work on this, it would be very great.
> I am really curious to see how the driver would look like if we go with
> this approach. I can test the code, too (although I understand you
> won't have any difficulties getting your hands on recent Intel
> systems ;)
> 
> Also see my reply in the other thread about the handling of removed
> siblings. I suspect it will be very easy to add to the new design.
> 
> Side question: is it safe to assume a maximum of 2 siblings per core on
> Intel x86 CPUs?
I think architecturally it's not safe to assume 2 siblings per core on x86
although so far HT implementations have been having 2 siblings per core.

Linux kernel doesn't assume 2 siblings per core during initialization (please
check arch/x86/kernel/smpboot.c). This is right way to handle potential non 2
sibling case in the future.

Thanks.

-Fenghua