Re: [PATCH] fs, proc: Introduce the /proc/<pid>/children entry v2

[Oleg Nesterov <oleg@redhat.com>]

On 12/09, Cyrill Gorcunov wrote:
>
> On Fri, Dec 09, 2011 at 04:30:09PM +0100, Oleg Nesterov wrote:
> ...
> > >
> > > It is a potential problem, from the lock-hold point of view and
> > > also it can cause large scheduling latencies.  What's involved in
> > > making ->children an rcu-protected list?
> >
> > At first glance, this doesn't look trivial... forget_original_parent()
> > abuses ->sibling.
> >
>
> But wait, forget_original_parent may move task out of the original
> list and put it in dead_children list, right? Then release_task
> may call delayed_put_task_struct under as call_rcu which should
> wait until we finish reading in our rcu section, isn't it?
>
> Even if we get inconsistent picture of children (ie we get pid
> of shildren which just getting dead) I think it's fine.

The problen is (one of the problems, in fact), list_move() changes
->next.

To simplify, let's talk about children_seq_start() your patch adds.
Suppose that we make ->children/sibling "rcu-safe" (we replace
__unhash_process()->list_del_init() with list_del_rcu, and so on).

Now, children_seq_start() does:

	rcu_read_lock();

	list_for_each(child, task->children)
		...;

Suppose that this task exits and reparents the child we are looking at.
Once reparent_leader() moves it into another list, list_for_each()
can never stop.

In short. forget_original_parent() changes the _head_ of the list,
in some sense.

> > But yes, it is not really nice to hold tasklist_lock here. May be
> > we can change this code so that every iteration records the reported
> > task_struct and then tries to continue. This means we should verify
> > that ->real_parent is still the same under tasklist, but at least
> > this way we do not hold it throughout.
> >
>
> And if real_parent is changed,

... or if list_empty(->sibling) == T

> I should simply skip such task and
> continue, right?

I think you should restart in this (hopefully unlikely) case. Yes, this
means the potentional data loss, but I guess we can't avoid this in any
case. For example, with the current patch children_seq_start() can miss
the _live_ thread if the already reported process was reaped.

> > Personally I'd even prefer /proc/pid/children/ directory (like
> > /proc/pid/task), but I guess this needs much more complications.
> >
>
> I fear so. Oleg, if it possible I would like to bring in as minimum
> code as I can ;)

Yes, yes, I agree.

Oleg.