[Devel] Re: [PATCH 2/3] Pid ns helpers for signals

[sukadev at us.ibm.com]

Oleg Nesterov [oleg at tv-sign.ru] wrote:
| On 08/31, sukadev at us.ibm.com wrote:
| >
| > Define some helper functions that will be used to implement signal semantics
| > with multiple pid namespaces. 
| > 
| > 	is_current_in_ancestor_pid_ns(task)
| > 
| > 		TRUE iff active pid namespace of 'current' is an ancestor of
| > 		active pid namespace of @task.
| > 
| > 	is_current_in_same_or_ancestor_pid_ns(task)
| > 
| > 		TRUE iff active pid namespace of 'current' is either same as 
| > 		or an ancestor of active pid namespace of @task.
| 
| These names are awfull :) Yes, yes, it was me who suggested them... No, I can't
| suggest something better.

I agree :-) I tried smaller names like task_ancestor_pid_ns() and passing in
'current' as a parameter so its not in the name :-) but the functionality was
not obvious from the names. 

| 
| > + * Caller must hold a reference to @pid.
| > + */
| > +static inline struct pid_namespace *pid_active_ns(struct pid *pid)
| > +{
| > +	if (!pid)
| > +		return NULL;
| > +
| > +	return pid->numbers[pid->level].ns;
| > +}
| 
| Well, the comment is a bit misleading. Yes, my previous comment was not very
| clear. Yes, the function itself is not safe unless you know what are you doing,
| like, for example, get_pid(). I think it is better to just kill the comment.
| Please see below.

Ok. will remove the comment.
| 
| > +static struct pid_namespace *get_task_pid_ns(struct task_struct *tsk)
| > +{
| > +	struct pid *pid;
| > +	struct pid_namespace *ns;
| > +
| > +	pid = get_task_pid(tsk, PIDTYPE_PID);
| > +	ns = get_pid_ns(pid_active_ns(pid));
| > +	put_pid(pid);
| > +
| > +	return ns;
| > +}
| 
| Hmm. Firstly, we don't need this for the "current", but all users of this func
| also do get_task_pid_ns(current).
| 
| Also, we don't need get/put_pid. rcu locks are enough,
| 
| 	rcu_read_lock();
| 	ns = get_pid_ns(pid_active_ns(task_pid(tks)));
| 	rcu_read_unlock();
| 

Ok.

| However, do we really need this complications right now? Currently, we use
| this "compare namespaces" helpers only when we know that "struct pid" is
| stable. We are sending the signal to that task, it must be pid_alive(), and
| we either locked the task itself, or we hold tasklist.

My concern was that the task could detach and free its pid which in turn
would drop the last reference on a pid namespace and free it.

By trying to keep is_current_in_ancestor*() general, I guess it is more
complicated than it needs to be right now.

Would holding the rcu_read_lock() be enough or since our callers hold
it now, can we just drop that too ?

is_current_in_ancstor_pid_ns(tsk)

	rcu_read_lock();
	my_ns = pid_active_ns(current);
	tsk_ns = pid_active_ns(tsk)
	rc = is_ancestor_ns(my_ns, tsk_ns)
	rcu_read_unlock();

	return rc;

Thanks for the comments,

Suka
_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers