Re: [patch] slab: add memory hotplug support

[Nick Piggin <npiggin@suse.de>]

On Mon, Mar 08, 2010 at 03:19:48PM -0800, David Rientjes wrote:
> On Fri, 5 Mar 2010, Nick Piggin wrote:
> 
> > > +#if defined(CONFIG_NUMA) && defined(CONFIG_MEMORY_HOTPLUG)
> > > +/*
> > > + * Drains and frees nodelists for a node on each slab cache, used for memory
> > > + * hotplug.  Returns -EBUSY if all objects cannot be drained on memory
> > > + * hot-remove so that the node is not removed.  When used because memory
> > > + * hot-add is canceled, the only result is the freed kmem_list3.
> > > + *
> > > + * Must hold cache_chain_mutex.
> > > + */
> > > +static int __meminit free_cache_nodelists_node(int node)
> > > +{
> > > +	struct kmem_cache *cachep;
> > > +	int ret = 0;
> > > +
> > > +	list_for_each_entry(cachep, &cache_chain, next) {
> > > +		struct array_cache *shared;
> > > +		struct array_cache **alien;
> > > +		struct kmem_list3 *l3;
> > > +
> > > +		l3 = cachep->nodelists[node];
> > > +		if (!l3)
> > > +			continue;
> > > +
> > > +		spin_lock_irq(&l3->list_lock);
> > > +		shared = l3->shared;
> > > +		if (shared) {
> > > +			free_block(cachep, shared->entry, shared->avail, node);
> > > +			l3->shared = NULL;
> > > +		}
> > > +		alien = l3->alien;
> > > +		l3->alien = NULL;
> > > +		spin_unlock_irq(&l3->list_lock);
> > > +
> > > +		if (alien) {
> > > +			drain_alien_cache(cachep, alien);
> > > +			free_alien_cache(alien);
> > > +		}
> > > +		kfree(shared);
> > > +
> > > +		drain_freelist(cachep, l3, l3->free_objects);
> > > +		if (!list_empty(&l3->slabs_full) ||
> > > +					!list_empty(&l3->slabs_partial)) {
> > > +			/*
> > > +			 * Continue to iterate through each slab cache to free
> > > +			 * as many nodelists as possible even though the
> > > +			 * offline will be canceled.
> > > +			 */
> > > +			ret = -EBUSY;
> > > +			continue;
> > > +		}
> > > +		kfree(l3);
> > > +		cachep->nodelists[node] = NULL;
> > 
> > What's stopping races of other CPUs trying to access l3 and array
> > caches while they're being freed?
> > 
> 
> numa_node_id() will not return an offlined nodeid and cache_alloc_node() 
> already does a fallback to other onlined nodes in case a nodeid is passed 
> to kmalloc_node() that does not have a nodelist.  l3->shared and l3->alien 
> cannot be accessed without l3->list_lock (drain, cache_alloc_refill, 
> cache_flusharray) or cache_chain_mutex (kmem_cache_destroy, cache_reap).

Yeah, but can't it _have_ a nodelist (ie. before it is set to NULL here)
while it is being accessed by another CPU and concurrently being freed
on this one? 


> > > +	}
> > > +	return ret;
> > > +}
> > > +
> > > +/*
> > > + * Onlines nid either as the result of memory hot-add or canceled hot-remove.
> > > + */
> > > +static int __meminit slab_node_online(int nid)
> > > +{
> > > +	int ret;
> > > +	mutex_lock(&cache_chain_mutex);
> > > +	ret = init_cache_nodelists_node(nid);
> > > +	mutex_unlock(&cache_chain_mutex);
> > > +	return ret;
> > > +}
> > > +
> > > +/*
> > > + * Offlines nid either as the result of memory hot-remove or canceled hot-add.
> > > + */
> > > +static int __meminit slab_node_offline(int nid)
> > > +{
> > > +	int ret;
> > > +	mutex_lock(&cache_chain_mutex);
> > > +	ret = free_cache_nodelists_node(nid);
> > > +	mutex_unlock(&cache_chain_mutex);
> > > +	return ret;
> > > +}
> > > +
> > > +static int __meminit slab_memory_callback(struct notifier_block *self,
> > > +					unsigned long action, void *arg)
> > > +{
> > > +	struct memory_notify *mnb = arg;
> > > +	int ret = 0;
> > > +	int nid;
> > > +
> > > +	nid = mnb->status_change_nid;
> > > +	if (nid < 0)
> > > +		goto out;
> > > +
> > > +	switch (action) {
> > > +	case MEM_GOING_ONLINE:
> > > +	case MEM_CANCEL_OFFLINE:
> > > +		ret = slab_node_online(nid);
> > > +		break;
> > 
> > This would explode if CANCEL_OFFLINE fails. Call it theoretical and
> > put a panic() in here and I don't mind. Otherwise you get corruption
> > somewhere in the slab code.
> > 
> 
> MEM_CANCEL_ONLINE would only fail here if a struct kmem_list3 couldn't be 
> allocated anywhere on the system and if that happens then the node simply 
> couldn't be allocated from (numa_node_id() would never return it as the 
> cpu's node, so it's possible to fallback in this scenario).

Why would it never return the CPU's node? It's CANCEL_OFFLINE that is
the problem.


> Instead of doing this all at MEM_GOING_OFFLINE, we could delay freeing of 
> the array caches and the nodelist until MEM_OFFLINE.  We're guaranteed 
> that all pages are freed at that point so there are no existing objects 
> that we need to track and then if the offline fails from a different 
> callback it would be possible to reset the l3->nodelists[node] pointers 
> since they haven't been freed yet.