netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Eric Dumazet <dada1@cosmosbay.com>
To: minyard@acm.org
Cc: Linux Kernel <linux-kernel@vger.kernel.org>,
	netdev@vger.kernel.org, shemminger@vyatta.com,
	paulmck@linux.vnet.ibm.com
Subject: Re: [PATCH 3/3] Convert the UDP hash lock to RCU
Date: Mon, 06 Oct 2008 23:22:31 +0200	[thread overview]
Message-ID: <48EA8197.6080502@cosmosbay.com> (raw)
In-Reply-To: <20081006185026.GA10383@minyard.local>

Corey Minyard a écrit :
> Change the UDP hash lock from an rwlock to RCU.
> 
> Signed-off-by: Corey Minyard <cminyard@mvista.com>
> ---
>  include/net/udp.h |    9 +++++----
>  net/ipv4/udp.c    |   47 +++++++++++++++++++++++++++--------------------
>  net/ipv6/udp.c    |   17 +++++++++--------
>  3 files changed, 41 insertions(+), 32 deletions(-)
> 
> diff --git a/include/net/udp.h b/include/net/udp.h
> index addcdc6..35aa104 100644
> --- a/include/net/udp.h
> +++ b/include/net/udp.h
> @@ -51,7 +51,7 @@ struct udp_skb_cb {
>  #define UDP_SKB_CB(__skb)	((struct udp_skb_cb *)((__skb)->cb))
>  
>  extern struct hlist_head udp_hash[UDP_HTABLE_SIZE];
> -extern rwlock_t udp_hash_lock;
> +extern spinlock_t udp_hash_wlock;
>  
>  
>  /* Note: this must match 'valbool' in sock_setsockopt */
> @@ -112,12 +112,13 @@ static inline void udp_lib_hash(struct sock *sk)
>  
>  static inline void udp_lib_unhash(struct sock *sk)
>  {
> -	write_lock_bh(&udp_hash_lock);
> -	if (sk_del_node_init(sk)) {
> +	spin_lock_bh(&udp_hash_wlock);
> +	if (sk_del_node_init_rcu(sk)) {
>  		inet_sk(sk)->num = 0;
>  		sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
>  	}
> -	write_unlock_bh(&udp_hash_lock);
> +	spin_unlock_bh(&udp_hash_wlock);
> +	synchronize_rcu();

UDP central rwlock can hurt performance, because of cache line ping pong,
so your patch really makes sense.

Me wondering what impact this synchronize_rcu() can have on mono-threaded
VOIP applications using lot of UDP sockets. What is the maximum delay of
this function ?

For "struct file" freeing, we chose call_rcu() instead of synchronize_rcu()

Maybe we could add a generic rcu head to struct sock, and use call_rcu() in
sk_prot_free() for sockets needing RCU (udp after your patch is applied, maybe
tcp on future patches, while I believe previous work on the subject concluded
RCU was not giving good results for short lived http sessions) ?

Or just add SLAB_DESTROY_BY_RCU to slab creation in proto_register()
for "struct proto udp_prot/udpv6_prot" so that kmem_cache_free() 
done in sk_prot_free() can defer freeing to RCU...

  reply	other threads:[~2008-10-06 21:22 UTC|newest]

Thread overview: 134+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-10-06 18:50 [PATCH 3/3] Convert the UDP hash lock to RCU Corey Minyard
2008-10-06 21:22 ` Eric Dumazet [this message]
2008-10-06 21:40   ` David Miller
2008-10-06 23:08     ` Corey Minyard
2008-10-07  8:37       ` Evgeniy Polyakov
2008-10-07 14:16         ` Christoph Lameter
2008-10-07 14:29           ` Evgeniy Polyakov
2008-10-07 14:38             ` Christoph Lameter
2008-10-07 14:33           ` Paul E. McKenney
2008-10-07 14:45             ` Christoph Lameter
2008-10-07 15:07               ` Eric Dumazet
2008-10-07 15:07               ` Paul E. McKenney
2008-10-07  5:24     ` Eric Dumazet
2008-10-07  8:54       ` Benny Amorsen
2008-10-07 12:59         ` Eric Dumazet
2008-10-07 14:07           ` Stephen Hemminger
2008-10-07 20:55             ` David Miller
2008-10-07 21:20               ` Stephen Hemminger
2008-10-08 13:55               ` Eric Dumazet
2008-10-08 18:45                 ` David Miller
2008-10-28 20:37                   ` [PATCH 1/2] udp: introduce struct udp_table and multiple rwlocks Eric Dumazet
2008-10-28 21:23                     ` Christian Bell
2008-10-28 21:31                       ` Evgeniy Polyakov
2008-10-28 21:48                       ` Eric Dumazet
2008-10-28 21:28                     ` Evgeniy Polyakov
2008-10-28 20:42                   ` [PATCH 2/2] udp: RCU handling for Unicast packets Eric Dumazet
2008-10-28 22:45                     ` Eric Dumazet
2008-10-29  5:05                       ` David Miller
2008-10-29  8:23                         ` Eric Dumazet
2008-10-29  8:56                           ` David Miller
2008-10-29 10:19                             ` Eric Dumazet
2008-10-29 18:19                               ` David Miller
2008-10-29  9:04                           ` Eric Dumazet
2008-10-29  9:17                             ` David Miller
2008-10-29 13:17                             ` Corey Minyard
2008-10-29 14:36                               ` Eric Dumazet
2008-10-29 15:34                                 ` Corey Minyard
2008-10-29 16:09                                   ` Eric Dumazet
2008-10-29 16:37                                     ` Paul E. McKenney
2008-10-29 17:22                                       ` Corey Minyard
2008-10-29 17:45                                         ` Eric Dumazet
2008-10-29 18:28                                           ` Corey Minyard
2008-10-29 18:52                                             ` Paul E. McKenney
2008-10-29 20:00                                               ` Eric Dumazet
2008-10-29 20:17                                                 ` Paul E. McKenney
2008-10-29 21:29                                                   ` Corey Minyard
2008-10-29 21:57                                                     ` Eric Dumazet
2008-10-29 21:58                                                     ` Paul E. McKenney
2008-10-29 22:08                                                   ` Eric Dumazet
2008-10-30  3:22                                                     ` Corey Minyard
2008-10-30  5:50                                                       ` Eric Dumazet
2008-11-02  4:19                                                         ` David Miller
2008-10-30  5:40                                                     ` David Miller
2008-10-30  5:51                                                       ` Eric Dumazet
2008-10-30  7:04                                                         ` Eric Dumazet
2008-10-30  7:05                                                           ` David Miller
2008-10-30 15:40                                                     ` [PATCH] udp: Introduce special NULL pointers for hlist termination Eric Dumazet
2008-10-30 15:51                                                       ` Stephen Hemminger
2008-10-30 16:28                                                         ` Corey Minyard
2008-10-31 14:37                                                           ` Eric Dumazet
2008-10-31 14:55                                                             ` Pavel Emelyanov
2008-11-02  4:22                                                               ` David Miller
2008-10-30 17:12                                                         ` Eric Dumazet
2008-10-31  7:51                                                           ` David Miller
2008-10-30 16:01                                                       ` Peter Zijlstra
2008-10-31  0:14                                                       ` Keith Owens
2008-11-13 13:13                                                       ` [PATCH 0/3] net: RCU lookups for UDP, DCCP and TCP protocol Eric Dumazet
2008-11-13 17:20                                                         ` Andi Kleen
2008-11-17  3:41                                                         ` David Miller
2008-11-19 19:52                                                           ` Christoph Lameter
2008-11-13 13:14                                                       ` [PATCH 1/3] rcu: Introduce hlist_nulls variant of hlist Eric Dumazet
2008-11-13 13:29                                                         ` Peter Zijlstra
2008-11-13 13:44                                                           ` Eric Dumazet
2008-11-13 16:02                                                             ` [PATCH 4/3] rcu: documents rculist_nulls Eric Dumazet
2008-11-14 15:16                                                               ` Peter Zijlstra
2008-11-17  3:36                                                                 ` David Miller
2008-11-19 17:07                                                               ` Paul E. McKenney
2008-11-14 15:16                                                         ` [PATCH 1/3] rcu: Introduce hlist_nulls variant of hlist Peter Zijlstra
2008-11-19 17:01                                                         ` Paul E. McKenney
2008-11-19 17:53                                                           ` Eric Dumazet
2008-11-19 18:46                                                             ` Paul E. McKenney
2008-11-19 18:53                                                               ` Arnaldo Carvalho de Melo
2008-11-19 21:17                                                                 ` Paul E. McKenney
2008-11-19 20:39                                                               ` Eric Dumazet
2008-11-19 21:21                                                                 ` Paul E. McKenney
2008-11-13 13:15                                                       ` [PATCH 2/3] udp: Use hlist_nulls in UDP RCU code Eric Dumazet
2008-11-19 17:29                                                         ` Paul E. McKenney
2008-11-19 17:53                                                           ` Eric Dumazet
2008-11-13 13:15                                                       ` [PATCH 3/3] net: Convert TCP & DCCP hash tables to use RCU / hlist_nulls Eric Dumazet
2008-11-13 13:34                                                         ` Peter Zijlstra
2008-11-13 13:51                                                           ` Eric Dumazet
2008-11-13 14:08                                                             ` Christoph Lameter
2008-11-13 14:22                                                             ` Peter Zijlstra
2008-11-13 14:27                                                               ` Christoph Lameter
2008-11-19 17:53                                                         ` Paul E. McKenney
2008-11-23  9:33                                                         ` [PATCH] net: Convert TCP/DCCP listening hash tables to use RCU Eric Dumazet
2008-11-23 15:59                                                           ` Paul E. McKenney
2008-11-23 18:42                                                             ` Eric Dumazet
2008-11-23 19:17                                                               ` Paul E. McKenney
2008-11-23 20:18                                                                 ` Eric Dumazet
2008-11-23 22:33                                                                   ` Paul E. McKenney
2008-11-24  1:23                                                           ` David Miller
2008-10-30 11:04                                                 ` [PATCH 2/2] udp: RCU handling for Unicast packets Peter Zijlstra
2008-10-30 11:30                                                   ` Eric Dumazet
2008-10-30 18:25                                                     ` Paul E. McKenney
2008-10-31 16:40                                                       ` Eric Dumazet
2008-11-01  3:10                                                         ` Paul E. McKenney
2008-10-29 17:32                                       ` Eric Dumazet
2008-10-29 18:11                                         ` Paul E. McKenney
2008-10-29 18:29                                           ` David Miller
2008-10-29 18:38                                             ` Paul E. McKenney
2008-10-29 18:36                                           ` Eric Dumazet
2008-10-29 18:20                                 ` David Miller
2008-10-30 11:12                                 ` Peter Zijlstra
2008-10-30 11:29                                   ` Eric Dumazet
2008-10-28 20:37                 ` [PATCH 0/2] udp: Convert the UDP hash lock to RCU Eric Dumazet
2008-10-28 21:28                   ` Stephen Hemminger
2008-10-28 21:50                     ` Eric Dumazet
2008-10-07 16:43           ` [PATCH 3/3] " Corey Minyard
2008-10-07 18:26       ` David Miller
2008-10-08  8:35         ` Eric Dumazet
2008-10-08 16:38           ` David Miller
2008-10-07  8:31     ` Peter Zijlstra
2008-10-07 14:36       ` Paul E. McKenney
2008-10-07 18:29       ` David Miller
2008-10-06 22:07   ` Corey Minyard
2008-10-07  8:17   ` Peter Zijlstra
2008-10-07  9:24     ` Eric Dumazet
2008-10-07 14:15       ` Christoph Lameter
2008-10-07 14:38         ` Paul E. McKenney
2008-10-07 14:50         ` Eric Dumazet
2008-10-07 15:05           ` Paul E. McKenney
2008-10-07 15:09           ` Peter Zijlstra
2008-10-07 15:23           ` Christoph Lameter

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=48EA8197.6080502@cosmosbay.com \
    --to=dada1@cosmosbay.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=minyard@acm.org \
    --cc=netdev@vger.kernel.org \
    --cc=paulmck@linux.vnet.ibm.com \
    --cc=shemminger@vyatta.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).