Re: [PATCH] task: Make task list manipulations RCU safe.

[Oleg Nesterov <oleg@tv-sign.ru>]

Some questions.

first_tgid:
	...
	for (; pos && pid_alive(pos); pos = next_task(pos))

I think this patch makes this 'pid_alive(pos)' unneeded?

next_tgid:
	rcu_read_lock();
	pos = start;
	if (pid_alive(start))
		pos = next_task(start);
	if (pid_alive(pos) && (pos != &init_task)) {
		get_task_struct(pos);
		goto done;
	}

The first 'pid_alive()' check is quite understandable.
What about the second one? I beleive, now it is unneeded
as well. The same for first_tid/next_tid.

Also, first_tid() does 'task_lock(leader)' while reading
->signal->count. Why? ->signal is protected by ->siglock,
but we don't need any locks because ->signal is rcu safe.
Same for proc_task_getattr(), s/task_lock/rcu_read_lock/.

Oleg.