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 <luc...@mojatatu.com>
Cc: Jamal Hadi Salim <j...@mojatatu.com>
Cc: Jiri Pirko <j...@resnulli.us>
Signed-off-by: Cong Wang <xiyou.wangc...@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

Reply via email to