migrate_dead_tasks() vs sleep-after-exit_notify() problems?

[Oleg Nesterov]

On 05/12, Oleg Nesterov wrote:
> 
> 	exit_notify()   ->exit_state != 0, and perhaps the task disappears
> 	                  from global process list

This, btw, means that do_exit()->__free_pipe_info() is not cpu-hotplug friendly.

The task may sleep on mutex_lock(->i_mutex), and dequeued from rq->arrays[].

The parent reaps this task, or it was TASK_DEAD and reaped itself.

CPU_DEAD comes, migration_call(CPU_DEAD) can do nothing with this task:

	- we already did release_task()->__unhash_process(), so
	  migrate_live_tasks() can't see it

	- migrate_dead_tasks() can't see it because it was deactivated.

No?

probably __might_sleep() should also check __exit_state.

Oleg.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/