Re: [RFC PATCH 2/2] slab: implement bulk free in SLAB allocator

[Jesper Dangaard Brouer <brouer@redhat.com>]

On Fri, 4 Dec 2015 11:17:02 -0600 (CST)
Christoph Lameter <cl@linux.com> wrote:

> On Thu, 3 Dec 2015, Jesper Dangaard Brouer wrote:
> 
> > +void kmem_cache_free_bulk(struct kmem_cache *orig_s, size_t size, void **p)
> 
> orig_s? Thats strange
> 
> > +{
> > +	struct kmem_cache *s;
> 
> s?

The "s" comes from the slub.c code uses "struct kmem_cache *s" everywhere.

> > +	size_t i;
> > +
> > +	local_irq_disable();
> > +	for (i = 0; i < size; i++) {
> > +		void *objp = p[i];
> > +
> > +		s = cache_from_obj(orig_s, objp);
> 
> Does this support freeing objects from a set of different caches?

This is for supporting memcg (CONFIG_MEMCG_KMEM).

Quoting from commit 033745189b1b ("slub: add missing kmem cgroup
support to kmem_cache_free_bulk"):

   Incoming bulk free objects can belong to different kmem cgroups, and
   object free call can happen at a later point outside memcg context.  Thus,
   we need to keep the orig kmem_cache, to correctly verify if a memcg object
   match against its "root_cache" (s->memcg_params.root_cache).
 

> > +
> > +		debug_check_no_locks_freed(objp, s->object_size);
> > +		if (!(s->flags & SLAB_DEBUG_OBJECTS))
> > +			debug_check_no_obj_freed(objp, s->object_size);
> > +
> > +		__cache_free(s, objp, _RET_IP_);
> 
> The function could be further optimized if you take the code from
> __cache_free() and move stuff outside of the loop. The alien cache check
> f.e. and the Pfmemalloc checking may be moved out. The call to
> virt_to_head page may also be avoided if the objects are on the same
> page  as the last. So you may be able to function calls for the
> fastpath in the inner loop which may accelerate frees significantly.

Interesting! Maybe we can do a followup patch to pullout last
optimization's.  Right now I'm mostly interested in correctness and
clean code.  And we are already looking at a 80% speedup with these
patches ;-)

-- 
Best regards,
  Jesper Dangaard Brouer
  MSc.CS, Principal Kernel Engineer at Red Hat
  Author of http://www.iptv-analyzer.org
  LinkedIn: http://www.linkedin.com/in/brouer

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>