2016-08-11 23:14 GMT+08:00 Juri Lelli [off-list ref]:

On 11/08/16 16:45, Peter Zijlstra wrote:

quoted

On Thu, Aug 04, 2016 at 05:51:20PM +0800, Wanpeng Li wrote:

quoted

From: Wanpeng Li <redacted>

The dl task will be replenished after dl task timer fire and start a new
period. It will be enqueued and to re-evaluate its dependency on the tick
in order to restart it. However, if cpu is hot-unplug, irq_work_queue will
splash since the target cpu is offline.

As a result:

    WARNING: CPU: 2 PID: 0 at kernel/irq_work.c:69 irq_work_queue_on+0xad/0xe0
    Call Trace:
     dump_stack+0x99/0xd0
     __warn+0xd1/0xf0
     warn_slowpath_null+0x1d/0x20
     irq_work_queue_on+0xad/0xe0
     tick_nohz_full_kick_cpu+0x44/0x50
     tick_nohz_dep_set_cpu+0x74/0xb0
     enqueue_task_dl+0x226/0x480
     activate_task+0x5c/0xa0
     dl_task_timer+0x19b/0x2c0
     ? push_dl_task.part.31+0x190/0x190

Hurm, so this is after hot-unplug succeeded. We get a timer (which is
also already migrated), but we enqueue the dl task on the offline CPU,
because we need to do replenish because start_dl_timer() -- see the
comment in dl_task_timer() at #ifdef CONFIG_SMP.

Then, once we've enqueued the task on the offline cpu, do we migrate it.

Bit icky that, but I don't immediately see a better way.

[...]

quoted

Juri, any opinions?

So, we would need to do something like calling replenish_dl_entity()
directly, instead of enqueue_task_dl(). pi_se shouldn't be a problem as
the task shouldn't be boosted if it was throttled.

dl_task_offline_migration() will still need to deactivate_task(), but
that should be fine as we check RB_EMPTY_NODE() in __dequeue_dl_entity()
and dequeue_pushable_dl_task().

Thanks, I will do it today. :)

Regards,
Wanpeng Li