11 Nov
2014
11 Nov
'14
5:30 p.m.
On Tue, Nov 11, 2014 at 08:22:55PM +0530, Shilpasri G Bhat wrote:
On 11/10/2014 07:19 PM, Peter Zijlstra wrote:
It it broken because who says rq->prev still exists?
'task_cumulative_load()' is handled by the cpufreq governor's kworker. And the kworker is queued only if there is task running on cpu which guarantees the existence of rq->prev in a running state.
Not so, there is no guarantee it will still be running by the time the kworker actually comes around to running.