Daniel Lezcano wrote:
On 01/10/2013 11:33 PM, amit daniel kachhap wrote:
On Thu, Jan 10, 2013 at 1:32 PM, Daniel Lezcano daniel.lezcano@free.fr
wrote:
On 01/10/2013 09:07 PM, amit daniel kachhap wrote:
Hi Daniel,
Hi Amit Daniel,
This hotplug noifiers looks fine. I suppose it should add extra state C1 in cpu0. If it is done like below than for normal cases(when all cpu's are online) there wont be any statistics for C0 state
I guess you meant state 0 which is WFI, right ? C0 state is the intel semantic for cpu fully turned on.
Yes I meant C0 as wfi
also which is required. Other patches look good.
Ok, that makes sense to have statistics even if they are only doing WFI.
Then the patch 4/5 is not ok, no ?
yes I suppose patch 4 and patch 5 are related and depends how you frame patch 5. I think it is better to create C0/C1 sysfs and other things in the beginning because it is a filesystem call and may increase the cpu hotplug time which is not worth. May be if cpuidle framework exposes some API to enable/disable states then it is better.
For patch 1,2 and 3, Acked-by: Amit Daniel Kachhap amit.daniel@samsung.com
Hi Kukjin,
is it possible to take these patches [1-3/5] ?
Looks OK to me, I will apply with Amit's ack.
The patches [3-4/5] could be ignored.
Probably, you mean [4-5/5] :-)
Thanks.
- Kukjin