From mboxrd@z Thu Jan 1 00:00:00 1970 From: Li Zefan Subject: Re: [PATCH v1 3/5] cgroup: Protect access to task_cls_classid() when built as module Date: Mon, 20 Aug 2012 08:59:37 +0800 Message-ID: <50318BF9.2080803@huawei.com> References: <1345215494-9181-1-git-send-email-wagi@monom.org> <1345215494-9181-4-git-send-email-wagi@monom.org> <20120817182855.GA11607@hmsreliant.think-freely.org> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Daniel Wagner , , , Daniel Wagner , "David S. Miller" , Gao feng , Jamal Hadi Salim , John Fastabend , Tejun Heo To: Neil Horman Return-path: Received: from szxga02-in.huawei.com ([119.145.14.65]:46703 "EHLO szxga02-in.huawei.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751217Ab2HTBAy (ORCPT ); Sun, 19 Aug 2012 21:00:54 -0400 In-Reply-To: <20120817182855.GA11607@hmsreliant.think-freely.org> Sender: netdev-owner@vger.kernel.org List-ID: =E4=BA=8E 2012/8/18 2:28, Neil Horman =E5=86=99=E9=81=93: > On Fri, Aug 17, 2012 at 04:58:12PM +0200, Daniel Wagner wrote: >> From: Daniel Wagner >> >> The module version of task_cls_classid() checks if net_cls_sbusys_id >> is valid to indentify when it is okay to access the controller. >> >> Instead relying on the subusys_id to be set, make it explicit >> with a jump label. >> >> Signed-off-by: Daniel Wagner >> Cc: "David S. Miller" >> Cc: Gao feng >> Cc: Jamal Hadi Salim >> Cc: John Fastabend >> Cc: Li Zefan >> Cc: Neil Horman >> Cc: Tejun Heo >> Cc: netdev@vger.kernel.org >> Cc: cgroups@vger.kernel.org >> --- >> include/net/cls_cgroup.h | 5 ++++- >> net/core/sock.c | 5 +++++ >> net/sched/cls_cgroup.c | 9 +++++++++ >> 3 files changed, 18 insertions(+), 1 deletion(-) >> >> diff --git a/include/net/cls_cgroup.h b/include/net/cls_cgroup.h >> index 401672c..bbbd957 100644 >> --- a/include/net/cls_cgroup.h >> +++ b/include/net/cls_cgroup.h >> @@ -16,6 +16,7 @@ >> #include >> #include >> #include >> +#include >> =20 >> #ifdef CONFIG_CGROUPS >> struct cgroup_cls_state >> @@ -44,6 +45,8 @@ static inline u32 task_cls_classid(struct task_str= uct *p) >> } >> =20 >> #elif IS_MODULE(CONFIG_NET_CLS_CGROUP) >> +extern struct static_key cgroup_cls_enabled; >> +#define clscg_enabled static_key_false(&cgroup_cls_enabled) >> =20 >> extern int net_cls_subsys_id; >> =20 >> @@ -52,7 +55,7 @@ static inline u32 task_cls_classid(struct task_str= uct *p) >> int id; >> u32 classid =3D 0; >> =20 >> - if (in_interrupt()) >> + if (!clscg_enabled || in_interrupt()) >> return 0; >> =20 >> rcu_read_lock(); >> diff --git a/net/core/sock.c b/net/core/sock.c >> index 8f67ced..8106e77 100644 >> --- a/net/core/sock.c >> +++ b/net/core/sock.c >> @@ -327,6 +327,11 @@ int __sk_backlog_rcv(struct sock *sk, struct sk= _buff *skb) >> EXPORT_SYMBOL(__sk_backlog_rcv); >> =20 >> #if defined(CONFIG_CGROUPS) >> +#if IS_MODULE(CONFIG_NET_CLS_CGROUP) >> +struct static_key cgroup_cls_enabled =3D STATIC_KEY_INIT_FALSE; >> +EXPORT_SYMBOL_GPL(cgroup_cls_enabled); >> +#endif >> + >> #if !defined(CONFIG_NET_CLS_CGROUP) >> int net_cls_subsys_id =3D -1; >> EXPORT_SYMBOL_GPL(net_cls_subsys_id); >> diff --git a/net/sched/cls_cgroup.c b/net/sched/cls_cgroup.c >> index 7743ea8..0635894 100644 >> --- a/net/sched/cls_cgroup.c >> +++ b/net/sched/cls_cgroup.c >> @@ -44,12 +44,21 @@ static struct cgroup_subsys_state *cgrp_create(s= truct cgroup *cgrp) >> =20 >> if (cgrp->parent) >> cs->classid =3D cgrp_cls_state(cgrp->parent)->classid; >> +#if IS_MODULE(CONFIG_NET_CLS_CGROUP) >> + else if (!clscg_enabled) >> + static_key_slow_inc(&cgroup_cls_enabled); > This is racy I think. The read of the static key is atomic with othe= r reads, > but the entire conditional is not atomic. If two cpus were creating = cgroups in > parallel, it would be possible for both to read the static key as bei= ng zero > (the second cpu would read the key before the first cpu could increme= nt it). static_key_slow_inc() is called only when we're creating the root cgrou= p, and that's module loading. so it's safe. >> +#endif >> =20 >> return &cs->css; >> } >> =20 >> static void cgrp_destroy(struct cgroup *cgrp) >> { >> +#if IS_MODULE(CONFIG_NET_CLS_CGROUP) >> + if (!cgrp->parent && clscg_enabled) >> + static_key_slow_dec(&cgroup_cls_enabled); > Ditto here with the race above. I think what you want is one of: and this is called only when we're destroying the root cgroup, and that= 's module unload.