Re: [PATCH 1/2] tickless idle cpus: core patch - v2

[Srivatsa Vaddagiri <vatsa@in.ibm.com>]

On Wed, Apr 12, 2006 at 03:35:20AM +1000, Paul Mackerras wrote:
> It would be nice if we could arrange to call stop_hz_timer from the
> top-level cpu_idle() function rather than having to call it from the
> individual power_save() functions such as power4_idle().  Can you see
> a problem with doing that?

I had considered doing that, but one problem with it is - how do we ensure that 
start_hz_timer will be called before idle thread calls schedule? A problem 
scenario is when the power_save() function returns without taking an interrupt 
(as is possible in pseries_dedicated_idle_sleep?), since start_hz_timer is 
currently called from only an interrupt context.

Now we could contemplate calling start_hz_timer directly from cpu_idle
when power_save() function returns - but how do we get the register
context required as an argument in start_hz_timer()?

-- 
Regards,
vatsa