On 05/16/2012 02:04 AM, David Rientjes wrote:

On Fri, 11 May 2012, Glauber Costa wrote:

quoted

diff --git a/mm/slab.c b/mm/slab.c
index e901a36..91b9c13 100644
--- a/mm/slab.c
+++ b/mm/slab.c

@@ -2118,6 +2118,7 @@ static void __kmem_cache_destroy(struct kmem_cache *cachep)
  			kfree(l3);
  		}
  	}
+	kfree(cachep->name);
  	kmem_cache_free(&cache_cache, cachep);
  }

@@ -2526,7 +2527,7 @@ kmem_cache_create (const char *name, size_t size, size_t align,
  		BUG_ON(ZERO_OR_NULL_PTR(cachep->slabp_cache));
  	}
  	cachep->ctor = ctor;
-	cachep->name = name;
+	cachep->name = kstrdup(name, GFP_KERNEL);

  	if (setup_cpu_cache(cachep, gfp)) {
  		__kmem_cache_destroy(cachep);

Couple problems:

  - allocating memory for a string of an unknown, unchecked size, and

  - could potentially return NULL which I suspect will cause problems
    later.

Well, this is what slub does.

I sent already two patches for it: One removing this from the slub, one 
adding this to the slab.

Right now I am comfortable with this one, because it makes it slightly 
easier in the latest patches of my series.

But note the word: slightest.

I am comfortable with any, provided slub and slab start behaving the same.

So whatever you guys decide between yourselves is fine, provided there 
is a decision.

Thanks for your review, David!