Open jlelli opened 6 years ago
As done @ https://github.com/jlelli/linux/blob/af3e79d29555b97dd096e2f8e36a0f50213808a8/kernel/sched/deadline.c#L1522 we instantaneously remove a task's running and total bandwidth contribution when the task is migrated between CPUs (see also push/pull functions), same also happens when setscheduler/affinity are called (the task in that case is dequeued and enqueued back with different scheduling/affinity parameters).
@lucabe72 shouldn't we wait the 0-lag time?
As done @ https://github.com/jlelli/linux/blob/af3e79d29555b97dd096e2f8e36a0f50213808a8/kernel/sched/deadline.c#L1522 we instantaneously remove a task's running and total bandwidth contribution when the task is migrated between CPUs (see also push/pull functions), same also happens when setscheduler/affinity are called (the task in that case is dequeued and enqueued back with different scheduling/affinity parameters).
@lucabe72 shouldn't we wait the 0-lag time?