[PATCH 00/14] replace call_rcu by kfree_rcu for simple kmem_cache_free callback

[Jason A. Donenfeld]

On Mon, Jun 17, 2024 at 03:50:56PM +0200, Uladzislau Rezki wrote:
> On Fri, Jun 14, 2024 at 09:33:45PM +0200, Jason A. Donenfeld wrote:
> > On Fri, Jun 14, 2024 at 02:35:33PM +0200, Uladzislau Rezki wrote:
> > > +	/* Should a destroy process be deferred? */
> > > +	if (s->flags & SLAB_DEFER_DESTROY) {
> > > +		list_move_tail(&s->list, &slab_caches_defer_destroy);
> > > +		schedule_delayed_work(&slab_caches_defer_destroy_work, HZ);
> > > +		goto out_unlock;
> > > +	}
> > 
> > Wouldn't it be smoother to have the actual kmem_cache_free() function
> > check to see if it's been marked for destruction and the refcount is
> > zero, rather than polling every one second? I mentioned this approach
> > in: https://lore.kernel.org/all/Zmo9-YGraiCj5-MI@zx2c4.com/ -
> > 
> >     I wonder if the right fix to this would be adding a `should_destroy`
> >     boolean to kmem_cache, which kmem_cache_destroy() sets to true. And
> >     then right after it checks `if (number_of_allocations == 0)
> >     actually_destroy()`, and likewise on each kmem_cache_free(), it
> >     could check `if (should_destroy && number_of_allocations == 0)
> >     actually_destroy()`. 
> > 
> I do not find pooling as bad way we can go with. But your proposal
> sounds reasonable to me also. We can combine both "prototypes" to
> one and offer.
> 
> Can you post a prototype here?

This is untested, but the simplest, shortest possible version would be:

diff --git a/mm/slab.h b/mm/slab.h
index 5f8f47c5bee0..907c0ea56c01 100644
--- a/mm/slab.h
+++ b/mm/slab.h
@@ -275,6 +275,7 @@ struct kmem_cache {
 	unsigned int inuse;		/* Offset to metadata */
 	unsigned int align;		/* Alignment */
 	unsigned int red_left_pad;	/* Left redzone padding size */
+	bool is_destroyed;		/* Destruction happens when no objects */
 	const char *name;		/* Name (only for display!) */
 	struct list_head list;		/* List of slab caches */
 #ifdef CONFIG_SYSFS
diff --git a/mm/slab_common.c b/mm/slab_common.c
index 1560a1546bb1..f700bed066d9 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -494,8 +494,8 @@ void kmem_cache_destroy(struct kmem_cache *s)
 		goto out_unlock;

 	err = shutdown_cache(s);
-	WARN(err, "%s %s: Slab cache still has objects when called from %pS",
-	     __func__, s->name, (void *)_RET_IP_);
+	if (err)
+		s->is_destroyed = true;
 out_unlock:
 	mutex_unlock(&slab_mutex);
 	cpus_read_unlock();
diff --git a/mm/slub.c b/mm/slub.c
index 1373ac365a46..7db8fe90a323 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -4510,6 +4510,8 @@ void kmem_cache_free(struct kmem_cache *s, void *x)
 		return;
 	trace_kmem_cache_free(_RET_IP_, x, s);
 	slab_free(s, virt_to_slab(x), x, _RET_IP_);
+	if (s->is_destroyed)
+		kmem_cache_destroy(s);
 }
 EXPORT_SYMBOL(kmem_cache_free);

@@ -5342,9 +5344,6 @@ static void free_partial(struct kmem_cache *s, struct kmem_cache_node *n)
 		if (!slab->inuse) {
 			remove_partial(n, slab);
 			list_add(&slab->slab_list, &discard);
-		} else {
-			list_slab_objects(s, slab,
-			  "Objects remaining in %s on __kmem_cache_shutdown()");
 		}
 	}
 	spin_unlock_irq(&n->list_lock);