From patchwork Thu Jul 21 10:17:50 2011 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Patrick McHardy X-Patchwork-Id: 106021 X-Patchwork-Delegate: davem@davemloft.net Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id EA75BB6F7E for ; Thu, 21 Jul 2011 20:18:16 +1000 (EST) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752240Ab1GUKSE (ORCPT ); Thu, 21 Jul 2011 06:18:04 -0400 Received: from stinky.trash.net ([213.144.137.162]:34510 "EHLO stinky.trash.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751564Ab1GUKSB (ORCPT ); Thu, 21 Jul 2011 06:18:01 -0400 Received: from x2.localnet. (localhost [127.0.0.1]) by stinky.trash.net (Postfix) with ESMTP id 31FAEB2C4D; Thu, 21 Jul 2011 12:17:59 +0200 (MEST) From: kaber@trash.net To: davem@davemloft.net Cc: netfilter-devel@vger.kernel.org, netdev@vger.kernel.org Subject: [PATCH 2/8] netfilter: nfnetlink: add RCU in nfnetlink_rcv_msg() Date: Thu, 21 Jul 2011 12:17:50 +0200 Message-Id: <1311243476-18236-3-git-send-email-kaber@trash.net> X-Mailer: git-send-email 1.7.2.3 In-Reply-To: <1311243476-18236-1-git-send-email-kaber@trash.net> References: <1311243476-18236-1-git-send-email-kaber@trash.net> Sender: netdev-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org From: Eric Dumazet Goal of this patch is to permit nfnetlink providers not mandate nfnl_mutex being held while nfnetlink_rcv_msg() calls them. If struct nfnl_callback contains a non NULL call_rcu(), then nfnetlink_rcv_msg() will use it instead of call() field, holding rcu_read_lock instead of nfnl_mutex Signed-off-by: Eric Dumazet CC: Florian Westphal CC: Eric Leblond Signed-off-by: Patrick McHardy --- include/linux/netfilter/nfnetlink.h | 3 ++ net/netfilter/nfnetlink.c | 40 ++++++++++++++++++++++++++-------- 2 files changed, 33 insertions(+), 10 deletions(-) diff --git a/include/linux/netfilter/nfnetlink.h b/include/linux/netfilter/nfnetlink.h index 2b11fc1..74d3386 100644 --- a/include/linux/netfilter/nfnetlink.h +++ b/include/linux/netfilter/nfnetlink.h @@ -60,6 +60,9 @@ struct nfnl_callback { int (*call)(struct sock *nl, struct sk_buff *skb, const struct nlmsghdr *nlh, const struct nlattr * const cda[]); + int (*call_rcu)(struct sock *nl, struct sk_buff *skb, + const struct nlmsghdr *nlh, + const struct nlattr * const cda[]); const struct nla_policy *policy; /* netlink attribute policy */ const u_int16_t attr_count; /* number of nlattr's */ }; diff --git a/net/netfilter/nfnetlink.c b/net/netfilter/nfnetlink.c index b4a4532..1905976 100644 --- a/net/netfilter/nfnetlink.c +++ b/net/netfilter/nfnetlink.c @@ -37,7 +37,7 @@ MODULE_ALIAS_NET_PF_PROTO(PF_NETLINK, NETLINK_NETFILTER); static char __initdata nfversion[] = "0.30"; -static const struct nfnetlink_subsystem *subsys_table[NFNL_SUBSYS_COUNT]; +static const struct nfnetlink_subsystem __rcu *subsys_table[NFNL_SUBSYS_COUNT]; static DEFINE_MUTEX(nfnl_mutex); void nfnl_lock(void) @@ -59,7 +59,7 @@ int nfnetlink_subsys_register(const struct nfnetlink_subsystem *n) nfnl_unlock(); return -EBUSY; } - subsys_table[n->subsys_id] = n; + rcu_assign_pointer(subsys_table[n->subsys_id], n); nfnl_unlock(); return 0; @@ -71,7 +71,7 @@ int nfnetlink_subsys_unregister(const struct nfnetlink_subsystem *n) nfnl_lock(); subsys_table[n->subsys_id] = NULL; nfnl_unlock(); - + synchronize_rcu(); return 0; } EXPORT_SYMBOL_GPL(nfnetlink_subsys_unregister); @@ -83,7 +83,7 @@ static inline const struct nfnetlink_subsystem *nfnetlink_get_subsys(u_int16_t t if (subsys_id >= NFNL_SUBSYS_COUNT) return NULL; - return subsys_table[subsys_id]; + return rcu_dereference(subsys_table[subsys_id]); } static inline const struct nfnl_callback * @@ -139,21 +139,27 @@ static int nfnetlink_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh) type = nlh->nlmsg_type; replay: + rcu_read_lock(); ss = nfnetlink_get_subsys(type); if (!ss) { #ifdef CONFIG_MODULES - nfnl_unlock(); + rcu_read_unlock(); request_module("nfnetlink-subsys-%d", NFNL_SUBSYS_ID(type)); - nfnl_lock(); + rcu_read_lock(); ss = nfnetlink_get_subsys(type); if (!ss) #endif + { + rcu_read_unlock(); return -EINVAL; + } } nc = nfnetlink_find_client(type, ss); - if (!nc) + if (!nc) { + rcu_read_unlock(); return -EINVAL; + } { int min_len = NLMSG_SPACE(sizeof(struct nfgenmsg)); @@ -167,7 +173,23 @@ replay: if (err < 0) return err; - err = nc->call(net->nfnl, skb, nlh, (const struct nlattr **)cda); + if (nc->call_rcu) { + err = nc->call_rcu(net->nfnl, skb, nlh, + (const struct nlattr **)cda); + rcu_read_unlock(); + } else { + rcu_read_unlock(); + nfnl_lock(); + if (rcu_dereference_protected( + subsys_table[NFNL_SUBSYS_ID(type)], + lockdep_is_held(&nfnl_mutex)) != ss || + nfnetlink_find_client(type, ss) != nc) + err = -EAGAIN; + else + err = nc->call(net->nfnl, skb, nlh, + (const struct nlattr **)cda); + nfnl_unlock(); + } if (err == -EAGAIN) goto replay; return err; @@ -176,9 +198,7 @@ replay: static void nfnetlink_rcv(struct sk_buff *skb) { - nfnl_lock(); netlink_rcv_skb(skb, &nfnetlink_rcv_msg); - nfnl_unlock(); } static int __net_init nfnetlink_net_init(struct net *net)