[PATCH] NO_IDLE_HZ implementation for PPC64

[Srivatsa Vaddagiri]

On Tue, Oct 04, 2005 at 08:23:26AM +1000, Benjamin Herrenschmidt wrote:
> Why not just call the decrementer interrupt ? It should replay
> already...

The only concern about that is wrt update_process_times - it may 
account all skipped ticks towards irq context rather than
to idle context. That is possible if an external irq (do_IRQ)
wakes the CPU up (leading to irq_enter being called twice ..)
This could lead to, say, wrong idle statistics in vmstat?


-- 


Thanks and Regards,
Srivatsa Vaddagiri,
Linux Technology Center,
IBM Software Labs,
Bangalore, INDIA - 560017