Re: [PATCH 3/4] threadgroup: extend threadgroup_lock() to cover exit and exec

[Oleg Nesterov <oleg@redhat.com>]

On 09/18, Oleg Nesterov wrote:
>
> Of course I am in no position to ack the changes in this code, I do not
> fell I understand it enough. But afaics this series is fine.
  ^^^^
I do not feel ;)

> If I am right, afaics the only change 4/4 needs is that it should not add
> WARN_ON_ONCE(tsk->flags & PF_EXITING) into cgroup_task_migrate().

Hmm. Unless I missed something this WARN_ON_ONCE() is not right anyway,
because

> - * Call holding cgroup_mutex. May take task_lock of
> - * the task 'tsk' during call.
> + * Call with cgroup_mutex and threadgroup locked. May take task_lock of
> + * @tsk during call.
>   */
>  int cgroup_attach_task(struct cgroup *cgrp, struct task_struct *tsk)

this is not true, cgroup_attach_proc() has other callers.

cgroup_attach_task_all() is probably fine, tsk should be current.

but cpuset_do_move_task() doesn't look fine at first glance, we can race
with the exiting task.

And this leads to another question about the cgroup_mutex/threadgroup_lock
nesting...

Oleg.