Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Wed, 13 Jul 2022 09:39:48 +0000
From:   Hyeonggon Yoo <42.hyeyoo@gmail.com>
To:     Christoph Lameter <cl@gentwo.de>
Cc:     Pekka Enberg <penberg@kernel.org>,
        David Rientjes <rientjes@google.com>,
        Joonsoo Kim <iamjoonsoo.kim@lge.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Vlastimil Babka <vbabka@suse.cz>,
        Roman Gushchin <roman.gushchin@linux.dev>,
        Joe Perches <joe@perches.com>,
        Vasily Averin <vasily.averin@linux.dev>,
        Matthew WilCox <willy@infradead.org>,
        linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: Re: [PATCH v3 1/15] mm/slab: move NUMA-related code to
 __do_cache_alloc()
Message-ID: <Ys6S5ALolh0z1JjJ@ip-172-31-24-42.ap-northeast-1.compute.internal>
References: <20220712133946.307181-1-42.hyeyoo@gmail.com>
 <20220712133946.307181-2-42.hyeyoo@gmail.com>
 <alpine.DEB.2.22.394.2207121626530.55992@gentwo.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <alpine.DEB.2.22.394.2207121626530.55992@gentwo.de>
Precedence: bulk

On Tue, Jul 12, 2022 at 04:29:10PM +0200, Christoph Lameter wrote:
> On Tue, 12 Jul 2022, Hyeonggon Yoo wrote:
> 
> > @@ -3241,31 +3219,46 @@ slab_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid, size_t orig_
> >  }
> >
> >  static __always_inline void *
> > -__do_cache_alloc(struct kmem_cache *cache, gfp_t flags)
> > +__do_cache_alloc(struct kmem_cache *cachep, gfp_t flags, int nodeid)
> >  {
> > -	void *objp;
> > +	void *objp = NULL;
> > +	int slab_node = numa_mem_id();
> >
> > -	if (current->mempolicy || cpuset_do_slab_mem_spread()) {
> > -		objp = alternate_node_alloc(cache, flags);
> > -		if (objp)
> > -			goto out;
> > +	if (nodeid == NUMA_NO_NODE) {
> > +		if (current->mempolicy || cpuset_do_slab_mem_spread()) {
> > +			objp = alternate_node_alloc(cachep, flags);
> > +			if (objp)
> > +				goto out;
> > +		}
> > +		/*
> > +		 * Use the locally cached objects if possible.
> > +		 * However ____cache_alloc does not allow fallback
> > +		 * to other nodes. It may fail while we still have
> > +		 * objects on other nodes available.
> > +		 */
> > +		objp = ____cache_alloc(cachep, flags);
> > +		nodeid = slab_node;
> > +	} else if (nodeid == slab_node) {
> > +		objp = ____cache_alloc(cachep, flags);
> > +	} else if (!get_node(cachep, nodeid)) {
> > +		/* Node not bootstrapped yet */
> > +		objp = fallback_alloc(cachep, flags);
> > +		goto out;
> >  	}
> > -	objp = ____cache_alloc(cache, flags);
> >
> >  	/*
> >  	 * We may just have run out of memory on the local node.
> >  	 * ____cache_alloc_node() knows how to locate memory on other nodes
> >  	 */
> >  	if (!objp)
> > -		objp = ____cache_alloc_node(cache, flags, numa_mem_id());
> > -
> > +		objp = ____cache_alloc_node(cachep, flags, nodeid);
> 
> 
> Does this preserve the original behavior? nodeid is the parameter passed
> to __do_cache_alloc(). numa_mem_id() is the nearest memory node.

Yes it does preserve the original behavior.

nodeid equals to value of numa_mem_id() when nodeid was NUMA_NO_NODE and
____cache_alloc() failed to allocate.