Received-SPF: pass (google.com: domain of linux-kernel+bounces-100694-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Tue, 12 Mar 2024 18:59:37 +0000
From: Matthew Wilcox <willy@infradead.org>
To: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Vlastimil Babka <vbabka@suse.cz>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Josh Poimboeuf <jpoimboe@kernel.org>,
	Jeff Layton <jlayton@kernel.org>,
	Chuck Lever <chuck.lever@oracle.com>, Kees Cook <kees@kernel.org>,
	Christoph Lameter <cl@linux.com>, Pekka Enberg <penberg@kernel.org>,
	David Rientjes <rientjes@google.com>,
	Joonsoo Kim <iamjoonsoo.kim@lge.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	Hyeonggon Yoo <42.hyeyoo@gmail.com>,
	Johannes Weiner <hannes@cmpxchg.org>,
	Michal Hocko <mhocko@kernel.org>,
	Shakeel Butt <shakeelb@google.com>,
	Muchun Song <muchun.song@linux.dev>,
	Alexander Viro <viro@zeniv.linux.org.uk>,
	Christian Brauner <brauner@kernel.org>, Jan Kara <jack@suse.cz>,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org,
	cgroups@vger.kernel.org, linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH RFC 1/4] mm, slab: move memcg charging to post-alloc hook
Message-ID: <ZfCmGYmvJlHAPiRz@casper.infradead.org>
References: <20240301-slab-memcg-v1-0-359328a46596@suse.cz>
 <20240301-slab-memcg-v1-1-359328a46596@suse.cz>
 <ZfCkfpogPQVMZnIG@P9FQF9L96D>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ZfCkfpogPQVMZnIG@P9FQF9L96D>

On Tue, Mar 12, 2024 at 11:52:46AM -0700, Roman Gushchin wrote:
> On Fri, Mar 01, 2024 at 06:07:08PM +0100, Vlastimil Babka wrote:
> > @@ -1926,71 +1939,51 @@ static bool __memcg_slab_pre_alloc_hook(struct kmem_cache *s,
> >  			return false;
> >  	}
> >  
> > -	if (obj_cgroup_charge(objcg, flags, objects * obj_full_size(s)))
> > +	if (obj_cgroup_charge(objcg, flags, size * obj_full_size(s)))
> >  		return false;
> >  
> > -	*objcgp = objcg;
> > +	for (i = 0; i < size; i++) {
> > +		slab = virt_to_slab(p[i]);
> 
> Not specific to this change, but I wonder if it makes sense to introduce virt_to_slab()
> variant without any extra checks for this and similar cases, where we know for sure
> that p resides on a slab page. What do you think?

You'd only save a single test_bit() ... is it really worth doing?
Cache misses are the expensive thing, not instructions.  And debugging
time: if somehow p[i] becomes not-on-a-slab-anymore, getting a NULL
pointer splat here before we go any further might be worth all the CPU
time wasted doing that test_bit().