Quoting Dave Hansen (dave@xxxxxxxxxxxxxxxxxx):
> On Wed, 2008-12-17 at 11:55 -0600, Serge E. Hallyn wrote:
> > -void free_ipc_ns(struct kref *kref)
> > +void put_ipc_ns(struct ipc_namespace *ns)
> > {
> > - struct ipc_namespace *ns;
> > + if (ns && atomic_dec_and_lock(&ns->count, &mq_lock)) {
> > + mq_clear_sbinfo(ns);
> > + spin_unlock(&mq_lock);
> > + mq_put_mnt(ns);
> > + free_ipc_ns(ns);
> > + }
> > +}
>
> OK, I see now why you wanted to remove the kref and add an atomic_t.
> That's probably worth a sentence in the patch description.
Good point, will do.
> I know that atomic_dec_and_lock() can be a nice logical thing to do, but
> is it necessary here? Basically, is there something that happens under
> mq_lock that can re-raise the ns->count?
A racing task from another ipc namespace can open a mq file through
the VFS and thereby bump the ipc's refcount. Will comment that here and
in the patch description.
thanks,
-serge
_______________________________________________
Containers mailing list
Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/containers
