> On Mon, 1 Oct 2007, Yasunori Goto wrote:
quoted text
> 
> > +#ifdef CONFIG_MEMORY_HOTPLUG
> > +static void __slab_callback_offline(int nid)
> > +{
> > +	struct kmem_cache_node *n;
> > +	struct kmem_cache *s;
> > +
> > +	list_for_each_entry(s, &slab_caches, list) {
> > +		if (s->node[nid]) {
> > +			n = get_node(s, nid);
> > +			s->node[nid] = NULL;
> > +			kmem_cache_free(kmalloc_caches, n);
> > +		}
> > +	}
> > +}
> 
> I think we need to bug here if there are still objects on the node that 
> are in use. This will silently discard the objects.
> 
Here is just the rollback code for an allocation failure of
kmem_cache_node in halfway.
So, there is a case some of them are not allocated yet.
Any slabs don't use new kmem_cache_node before the new nodes page is
available --so far--.
But, in the future, here will be useful for node hot-unplug code,
and its check will be necessary.  Ok. I'll add its check.

Do you mean that just nr_slabs should be checked like followings?
I'm not sure this is enough.

    :
if (s->node[nid]) {
	n = get_node(s, nid);
	if (!atomic_read(&n->nr_slabs)) {
		s->node[nid] = NULL;
		kmem_cache_free(kmalloc_caches, n);
	}
}
    :
    :

Thanks.

-- 
Yasunori Goto 


-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/