Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Sun, 26 Jan 2020 14:44:31 -0800 (PST)
From:   David Rientjes <rientjes@google.com>
To:     Wei Yang <richardw.yang@linux.intel.com>
cc:     akpm@linux-foundation.org, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH] mm/vmscan.c: only adjust related kswapd cpu affinity
 when online cpu
In-Reply-To: <20200126132052.11921-1-richardw.yang@linux.intel.com>
Message-ID: <alpine.DEB.2.21.2001261443020.164983@chino.kir.corp.google.com>
References: <20200126132052.11921-1-richardw.yang@linux.intel.com>
User-Agent: Alpine 2.21 (DEB 202 2017-01-01)
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sun, 26 Jan 2020, Wei Yang wrote:

> When onlining a cpu, kswapd_cpu_online() is called to adjust kswapd cpu
> affinity.
> 
> Current routine does like this:
> 
>   * Iterate all the numa node
>   * Adjust cpu affinity when node has an online cpu
> 
> Actually we could improve this by:
> 
>   * Just adjust the numa node to which current online cpu belongs
> 
> Because we know which numa node the cpu belongs to and this cpu would
> not affect other node.
> 

Is there ever a situation where the cpu to be onlined would have appeared 
in the cpumask of another node and so a different kswapd's cpumask would 
now include an off-node cpu?

> Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
> ---
>  mm/vmscan.c | 19 ++++++++++---------
>  1 file changed, 10 insertions(+), 9 deletions(-)
> 
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 572fb17c6273..19c92d35045c 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -4049,18 +4049,19 @@ unsigned long shrink_all_memory(unsigned long nr_to_reclaim)
>     restore their cpu bindings. */
>  static int kswapd_cpu_online(unsigned int cpu)
>  {
> -	int nid;
> +	int nid = cpu_to_node(cpu);
> +	pg_data_t *pgdat;
> +	const struct cpumask *mask;
>  
> -	for_each_node_state(nid, N_MEMORY) {
> -		pg_data_t *pgdat = NODE_DATA(nid);
> -		const struct cpumask *mask;
> +	if (!node_state(nid, N_MEMORY))
> +		return 0;
>  
> -		mask = cpumask_of_node(pgdat->node_id);
> +	pgdat = NODE_DATA(nid);
> +	mask = cpumask_of_node(nid);
> +
> +	/* One of our CPUs online: restore mask */
> +	set_cpus_allowed_ptr(pgdat->kswapd, mask);
>  
> -		if (cpumask_any_and(cpu_online_mask, mask) < nr_cpu_ids)
> -			/* One of our CPUs online: restore mask */
> -			set_cpus_allowed_ptr(pgdat->kswapd, mask);
> -	}
>  	return 0;
>  }
>  
> -- 
> 2.17.1
> 
> 
>