From mboxrd@z Thu Jan  1 00:00:00 1970
From: Rick Jones <rick.jones2@hpe.com>
Subject: Re: [RFC net-next 2/2] udp: No longer use SLAB_DESTROY_BY_RCU
Date: Mon, 28 Mar 2016 11:44:00 -0700
Message-ID: <56F97B70.1000904@hpe.com>
References: <1458944964-12890-1-git-send-email-edumazet@google.com>
 <1458944964-12890-3-git-send-email-edumazet@google.com>
 <56F95884.8010901@hpe.com>
 <1459184434.6473.104.camel@edumazet-glaptop3.roam.corp.google.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Cc: Eric Dumazet <edumazet@google.com>,
	"David S . Miller" <davem@davemloft.net>,
	netdev <netdev@vger.kernel.org>,
	Tom Herbert <tom@herbertland.com>
To: Eric Dumazet <eric.dumazet@gmail.com>
Return-path: <netdev-owner@vger.kernel.org>
Received: from g4t3427.houston.hp.com ([15.201.208.55]:42877 "EHLO
	g4t3427.houston.hp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1751005AbcC1SoD (ORCPT
	<rfc822;netdev@vger.kernel.org>); Mon, 28 Mar 2016 14:44:03 -0400
In-Reply-To: <1459184434.6473.104.camel@edumazet-glaptop3.roam.corp.google.com>
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

On 03/28/2016 10:00 AM, Eric Dumazet wrote:
> On Mon, 2016-03-28 at 09:15 -0700, Rick Jones wrote:
>> On 03/25/2016 03:29 PM, Eric Dumazet wrote:
>>> UDP sockets are not short lived in the high usage case, so the added
>>> cost of call_rcu() should not be a concern.
>>
>> Even a busy DNS resolver?
>
> If you mean that a busy DNS resolver spends _most_ of its time doing :
>
> fd = socket()
> bind(fd  port=0)
>    < send and receive one frame >
> close(fd)

Yes.  Although it has been a long time, I thought that say the likes of 
a caching named in the middle between hosts and the rest of the DNS 
would behave that way as it was looking-up names on behalf those who 
asked it.

rick

>
> (If this is the case, may I suggest doing something different, and use
> some kind of caches ? It will be way faster.)
>
> Then the result for 10,000,000 loops of <socket()+bind()+close()> are
>
> Before patch :
>
> real	0m13.665s
> user	0m0.548s
> sys	0m12.372s
>
> After patch :
>
> real	0m20.599s
> user	0m0.465s
> sys	0m17.965s
>
> So the worst overhead is 700 ns
>
> This is roughly the cost for bringing 960 bytes from memory, or 15 cache
> lines (on x86_64)
>
> # grep UDP /proc/slabinfo
> UDPLITEv6              0      0   1088    7    2 : tunables   24   12    8 : slabdata      0      0      0
> UDPv6                 24     49   1088    7    2 : tunables   24   12    8 : slabdata      7      7      0
> UDP-Lite               0      0    960    4    1 : tunables   54   27    8 : slabdata      0      0      0
> UDP                   30     36    960    4    1 : tunables   54   27    8 : slabdata      9      9      2
>
> In reality, chances that UDP sockets are re-opened right after being
> freed and their 15 cache lines are very hot in cpu caches is quite
> small, so I would not worry at all about this rather stupid benchmark.
>
> int main(int argc, char *argv[]) {
> 	struct sockaddr_in addr;
> 	int i, fd, loops = 10000000;
>
> 	for (i = 0; i < loops; i++) {
> 		fd = socket(AF_INET, SOCK_DGRAM, 0);
> 		if (fd == -1) {
> 			perror("socket");
> 			break;
> 		}
> 		memset(&addr, 0, sizeof(addr));
> 		addr.sin_family = AF_INET;
> 		if (bind(fd, (const struct sockaddr *)&addr, sizeof(addr)) == -1) {
> 			perror("bind");
> 			break;
> 		}
> 		close(fd);
> 	}
> 	return 0;
> }
>