> From: linux-pm-bounces@xxxxxxxxxxxxxxxxxxxxxxxxxx [mailto:linux-pm- > bounces@xxxxxxxxxxxxxxxxxxxxxxxxxx] On Behalf Of Surinder P Singh > Sent: Tuesday, May 19, 2009 11:25 AM > I have a question about cpuidle C-state accounting in > cpuidle_idle_call() as implemented in kernel v2.6.29. > > I understand it is possible for the target state entering callback to > return 0, if for some reason it could not enter the requested C-state > as decided by the governor. In this case, shouldn't the usage count be > updated conditionally to reflect the fact that the state may or may > not actually have been entered ? essentially, something like this: If it fails state due to bus master check you can do like the reference acpi driver does. I validated the same on OMAP3 a while back. http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=addbad46ed0906cd584784423b9d0babc7476446 In code I've used I don't demote other states. But there are state aborts. In those cases residency is low or 0. Regards, Richard W. _______________________________________________ linux-pm mailing list linux-pm@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/linux-pm
