netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Cong Wang <xiyou.wangcong@gmail.com>
To: netdev@vger.kernel.org
Cc: Cong Wang <xiyou.wangcong@gmail.com>,
	Lucas Bates <lucasb@mojatatu.com>,
	Jamal Hadi Salim <jhs@mojatatu.com>,
	Jiri Pirko <jiri@resnulli.us>
Subject: [Patch net 02/13] net_sched: introduce tcf_exts_get_net() and tcf_exts_put_net()
Date: Mon,  6 Nov 2017 13:47:19 -0800	[thread overview]
Message-ID: <20171106214730.24421-3-xiyou.wangcong@gmail.com> (raw)
In-Reply-To: <20171106214730.24421-1-xiyou.wangcong@gmail.com>

Instead of holding netns refcnt in tc actions, we can minimize
the holding time by saving it in struct tcf_exts instead. This
means we can just hold netns refcnt right before call_rcu() and
release it after tcf_exts_destroy() is done.

However, because on netns cleanup path we call tcf_proto_destroy()
too, obviously we can not hold netns for a zero refcnt, in this
case we have to do cleanup synchronously. It is fine for RCU too,
the caller cleanup_net() already waits for a grace period.

For other cases, refcnt is non-zero and we can safely grab it as
normal and release it after we are done.

This patch provides two new API for each filter to use:
tcf_exts_get_net() and tcf_exts_put_net(). And all filters now can
use the following pattern:

void __destroy_filter() {
  tcf_exts_destroy();
  tcf_exts_put_net();  // <== release netns refcnt
  kfree();
}
void some_work() {
  rtnl_lock();
  __destroy_filter();
  rtnl_unlock();
}
void some_rcu_callback() {
  tcf_queue_work(some_work);
}

if (tcf_exts_get_net())  // <== hold netns refcnt
  call_rcu(some_rcu_callback);
else
  __destroy_filter();

Cc: Lucas Bates <lucasb@mojatatu.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
---
 include/net/pkt_cls.h | 24 ++++++++++++++++++++++++
 net/sched/cls_api.c   |  1 +
 2 files changed, 25 insertions(+)

diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h
index 70ca2437740e..8826747ef83e 100644
--- a/include/net/pkt_cls.h
+++ b/include/net/pkt_cls.h
@@ -94,6 +94,7 @@ struct tcf_exts {
 	__u32	type; /* for backward compat(TCA_OLD_COMPAT) */
 	int nr_actions;
 	struct tc_action **actions;
+	struct net *net;
 #endif
 	/* Map to export classifier specific extension TLV types to the
 	 * generic extensions API. Unsupported extensions must be set to 0.
@@ -107,6 +108,7 @@ static inline int tcf_exts_init(struct tcf_exts *exts, int action, int police)
 #ifdef CONFIG_NET_CLS_ACT
 	exts->type = 0;
 	exts->nr_actions = 0;
+	exts->net = NULL;
 	exts->actions = kcalloc(TCA_ACT_MAX_PRIO, sizeof(struct tc_action *),
 				GFP_KERNEL);
 	if (!exts->actions)
@@ -117,6 +119,28 @@ static inline int tcf_exts_init(struct tcf_exts *exts, int action, int police)
 	return 0;
 }
 
+/* Return false if the netns is being destroyed in cleanup_net(). Callers
+ * need to do cleanup synchronously in this case, otherwise may race with
+ * tc_action_net_exit(). Return true for other cases.
+ */
+static inline bool tcf_exts_get_net(struct tcf_exts *exts)
+{
+#ifdef CONFIG_NET_CLS_ACT
+	exts->net = maybe_get_net(exts->net);
+	return exts->net != NULL;
+#else
+	return true;
+#endif
+}
+
+static inline void tcf_exts_put_net(struct tcf_exts *exts)
+{
+#ifdef CONFIG_NET_CLS_ACT
+	if (exts->net)
+		put_net(exts->net);
+#endif
+}
+
 static inline void tcf_exts_to_list(const struct tcf_exts *exts,
 				    struct list_head *actions)
 {
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index b2d310745487..ecbb019efcbd 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -927,6 +927,7 @@ int tcf_exts_validate(struct net *net, struct tcf_proto *tp, struct nlattr **tb,
 				exts->actions[i++] = act;
 			exts->nr_actions = i;
 		}
+		exts->net = net;
 	}
 #else
 	if ((exts->action && tb[exts->action]) ||
-- 
2.13.0

  parent reply	other threads:[~2017-11-06 21:47 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-11-06 21:47 [Patch net 00/13] net_sched: close the race between call_rcu() and cleanup_net() Cong Wang
2017-11-06 21:47 ` [Patch net 01/13] Revert "net_sched: hold netns refcnt for each action" Cong Wang
2017-11-06 21:47 ` Cong Wang [this message]
2017-11-06 21:47 ` [Patch net 03/13] cls_basic: use tcf_exts_get_net() before call_rcu() Cong Wang
2017-11-06 21:47 ` [Patch net 04/13] cls_bpf: " Cong Wang
2017-11-06 21:47 ` [Patch net 05/13] cls_cgroup: " Cong Wang
2017-11-09 22:31   ` Roman Mashak
2017-11-09 23:52     ` Cong Wang
2017-11-06 21:47 ` [Patch net 06/13] cls_flow: " Cong Wang
2017-11-06 21:47 ` [Patch net 07/13] cls_flower: " Cong Wang
2017-11-06 21:47 ` [Patch net 08/13] cls_fw: " Cong Wang
2017-11-06 21:47 ` [Patch net 09/13] cls_matchall: " Cong Wang
2017-11-06 21:47 ` [Patch net 10/13] cls_route: " Cong Wang
2017-11-06 21:47 ` [Patch net 11/13] cls_rsvp: " Cong Wang
2017-11-06 21:47 ` [Patch net 12/13] cls_tcindex: " Cong Wang
2017-11-06 21:47 ` [Patch net 13/13] cls_u32: " Cong Wang
2017-11-09  2:26 ` [Patch net 00/13] net_sched: close the race between call_rcu() and cleanup_net() David Miller
2017-11-09  2:32   ` Cong Wang

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=20171106214730.24421-3-xiyou.wangcong@gmail.com \
    --to=xiyou.wangcong@gmail.com \
    --cc=jhs@mojatatu.com \
    --cc=jiri@resnulli.us \
    --cc=lucasb@mojatatu.com \
    --cc=netdev@vger.kernel.org \
    /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).