Re: [PATCH 08/57] microblaze_v7: Interrupt handling, timer support, selfmod code

[Michal Simek <monstr@monstr.eu>]

Thomas Gleixner wrote:
> On Sat, 21 Mar 2009, Michal Simek wrote:
>> CONFIG_HZ_100=y
>> # CONFIG_HZ_250 is not set
>> # CONFIG_HZ_300 is not set
>> # CONFIG_HZ_1000 is not set
>> CONFIG_HZ=100
>>
>> For NO_HZ val I shouldn't use HZ value because of NO_HZ and HZ values shouldn't
>> be in .config file. Am I right?
> 
> No. NO_HZ does not remove the HZ value from the kernel. There is way
> too much code depending on it. What NO_HZ does is to supress the
> periodic tick when the system goes idle and the next timer expiry is
> farther away than 1 HZ. When the system is non idle then the periodic
> tick runs again. NO_HZ is merily for power saving.
> 
> In a normal kernel when the machine is idle then we wake up every HZ
> to find out that there is no work to do and we go idle again. But if
> you think about it then this is stupid, because the kernel knows when
> the next timer is due to expire. So if the next timer expiry is
> e.g. 100ms away, then we can sleep for full 100ms instead of sleeping
> 10 * 10ms (HZ=100) or 100 * 1ms (HZ=1000). This allows systems to go
> into deeper power saving states.
> 
> You should be able to observe the difference by monitoring
> /proc/interrupts. A nohz enabled kernel should have significantly less
> timer interrupts on an idle system.

Yes, I see it. I think I sent my log in any previous email in this thread too +
I see it on my heartbeat led too.

Thanks for explanation,
Michal
> 
> Thanks,
> 
> 	tglx


-- 
Michal Simek, Ing. (M.Eng)
w: www.monstr.eu p: +42-0-721842854