[PATCH RESEND net-next 0/2] net_rwsem fixes

[PATCH RESEND net-next 0/2] net_rwsem fixes

[Kirill Tkhai]

(Forgot to CC netdev@ :)

Hi,

there is wext_netdev_notifier_call()->wireless_nlevent_flush()
netdevice notifier, which takes net_rwsem, so we can't take
net_rwsem in {,un}register_netdevice_notifier().

Since {,un}register_netdevice_notifier() is executed under
pernet_ops_rwsem, net_namespace_list can't change, while we
holding it, so there is no need net_rwsem in these functions [1/2].

The same is in [2/2]. We make callers of __rtnl_link_unregister()
take pernet_ops_rwsem, and close the race with setup_net()
and cleanup_net(), so __rtnl_link_unregister() does not need it.
This also fixes the problem of that __rtnl_link_unregister() does
not see initializing and exiting nets.

Thanks,
Kirill
---

Kirill Tkhai (2):
      net: Remove net_rwsem from {,un}register_netdevice_notifier()
      net: Do not take net_rwsem in __rtnl_link_unregister()


 drivers/net/dummy.c      |    2 ++
 drivers/net/ifb.c        |    2 ++
 net/core/dev.c           |    5 -----
 net/core/net_namespace.c |    1 +
 net/core/rtnetlink.c     |    6 +++---
 5 files changed, 8 insertions(+), 8 deletions(-)

[PATCH RESEND net-next 1/2] net: Remove net_rwsem from {, un}register_netdevice_notifier()

[Kirill Tkhai]

These functions take net_rwsem, while wireless_nlevent_flush()
also takes it. But down_read() can't be taken recursive,
because of rw_semaphore design, which prevents it to be occupied
by only readers forever.

Since we take pernet_ops_rwsem in {,un}register_netdevice_notifier(),
net list can't change, so these down_read()/up_read() can be removed.

Fixes: f0b07bb151b0 "net: Introduce net_rwsem to protect net_namespace_list"
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
---
 net/core/dev.c |    5 -----
 1 file changed, 5 deletions(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index 07da7add4845..8edb58829124 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1633,7 +1633,6 @@ int register_netdevice_notifier(struct notifier_block *nb)
 		goto unlock;
 	if (dev_boot_phase)
 		goto unlock;
-	down_read(&net_rwsem);
 	for_each_net(net) {
 		for_each_netdev(net, dev) {
 			err = call_netdevice_notifier(nb, NETDEV_REGISTER, dev);
@@ -1647,7 +1646,6 @@ int register_netdevice_notifier(struct notifier_block *nb)
 			call_netdevice_notifier(nb, NETDEV_UP, dev);
 		}
 	}
-	up_read(&net_rwsem);
 
 unlock:
 	rtnl_unlock();
@@ -1671,7 +1669,6 @@ int register_netdevice_notifier(struct notifier_block *nb)
 	}
 
 outroll:
-	up_read(&net_rwsem);
 	raw_notifier_chain_unregister(&netdev_chain, nb);
 	goto unlock;
 }
@@ -1704,7 +1701,6 @@ int unregister_netdevice_notifier(struct notifier_block *nb)
 	if (err)
 		goto unlock;
 
-	down_read(&net_rwsem);
 	for_each_net(net) {
 		for_each_netdev(net, dev) {
 			if (dev->flags & IFF_UP) {
@@ -1715,7 +1711,6 @@ int unregister_netdevice_notifier(struct notifier_block *nb)
 			call_netdevice_notifier(nb, NETDEV_UNREGISTER, dev);
 		}
 	}
-	up_read(&net_rwsem);
 unlock:
 	rtnl_unlock();
 	up_write(&pernet_ops_rwsem);

[PATCH RESEND net-next 2/2] net: Do not take net_rwsem in __rtnl_link_unregister()

[Kirill Tkhai]

This function calls call_netdevice_notifier(), which also
may take net_rwsem. So, we can't use net_rwsem here.

This patch makes callers of this functions take pernet_ops_rwsem,
like register_netdevice_notifier() does. This will protect
the modifications of net_namespace_list, and allows notifiers
to take it (they won't have to care about context).

Since __rtnl_link_unregister() is used on module load
and unload (which are not frequent operations), this looks
for me better, than make all call_netdevice_notifier()
always executing in "protected net_namespace_list" context.

Also, this fixes the problem we had a deal in 328fbe747ad4
"Close race between {un, }register_netdevice_notifier and ...",
and guarantees __rtnl_link_unregister() does not skip
exitting net.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
---
 drivers/net/dummy.c      |    2 ++
 drivers/net/ifb.c        |    2 ++
 net/core/net_namespace.c |    1 +
 net/core/rtnetlink.c     |    6 +++---
 4 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/drivers/net/dummy.c b/drivers/net/dummy.c
index 30b1c8512049..0d15a12a4560 100644
--- a/drivers/net/dummy.c
+++ b/drivers/net/dummy.c
@@ -219,6 +219,7 @@ static int __init dummy_init_module(void)
 {
 	int i, err = 0;
 
+	down_write(&pernet_ops_rwsem);
 	rtnl_lock();
 	err = __rtnl_link_register(&dummy_link_ops);
 	if (err < 0)
@@ -233,6 +234,7 @@ static int __init dummy_init_module(void)
 
 out:
 	rtnl_unlock();
+	up_write(&pernet_ops_rwsem);
 
 	return err;
 }
diff --git a/drivers/net/ifb.c b/drivers/net/ifb.c
index 0008da7e9d4c..5f2897ec0edc 100644
--- a/drivers/net/ifb.c
+++ b/drivers/net/ifb.c
@@ -330,6 +330,7 @@ static int __init ifb_init_module(void)
 {
 	int i, err;
 
+	down_write(&pernet_ops_rwsem);
 	rtnl_lock();
 	err = __rtnl_link_register(&ifb_link_ops);
 	if (err < 0)
@@ -344,6 +345,7 @@ static int __init ifb_init_module(void)
 
 out:
 	rtnl_unlock();
+	up_write(&pernet_ops_rwsem);
 
 	return err;
 }
diff --git a/net/core/net_namespace.c b/net/core/net_namespace.c
index 7fdf321d4997..a11e03f920d3 100644
--- a/net/core/net_namespace.c
+++ b/net/core/net_namespace.c
@@ -51,6 +51,7 @@ static bool init_net_initialized;
  * outside.
  */
 DECLARE_RWSEM(pernet_ops_rwsem);
+EXPORT_SYMBOL_GPL(pernet_ops_rwsem);
 
 #define MIN_PERNET_OPS_ID	\
 	((sizeof(struct net_generic) + sizeof(void *) - 1) / sizeof(void *))
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index e86b28482ca7..45936922d7e2 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -412,17 +412,17 @@ static void __rtnl_kill_links(struct net *net, struct rtnl_link_ops *ops)
  * __rtnl_link_unregister - Unregister rtnl_link_ops from rtnetlink.
  * @ops: struct rtnl_link_ops * to unregister
  *
- * The caller must hold the rtnl_mutex.
+ * The caller must hold the rtnl_mutex and guarantee net_namespace_list
+ * integrity (hold pernet_ops_rwsem for writing to close the race
+ * with setup_net() and cleanup_net()).
  */
 void __rtnl_link_unregister(struct rtnl_link_ops *ops)
 {
 	struct net *net;
 
-	down_read(&net_rwsem);
 	for_each_net(net) {
 		__rtnl_kill_links(net, ops);
 	}
-	up_read(&net_rwsem);
 	list_del(&ops->list);
 }
 EXPORT_SYMBOL_GPL(__rtnl_link_unregister);

Re: [PATCH RESEND net-next 0/2] net_rwsem fixes

[David Miller]

From: Kirill Tkhai <ktkhai@virtuozzo.com>
Date: Fri, 30 Mar 2018 19:38:16 +0300

> there is wext_netdev_notifier_call()->wireless_nlevent_flush()
> netdevice notifier, which takes net_rwsem, so we can't take
> net_rwsem in {,un}register_netdevice_notifier().
> 
> Since {,un}register_netdevice_notifier() is executed under
> pernet_ops_rwsem, net_namespace_list can't change, while we
> holding it, so there is no need net_rwsem in these functions [1/2].
> 
> The same is in [2/2]. We make callers of __rtnl_link_unregister()
> take pernet_ops_rwsem, and close the race with setup_net()
> and cleanup_net(), so __rtnl_link_unregister() does not need it.
> This also fixes the problem of that __rtnl_link_unregister() does
> not see initializing and exiting nets.

Series applied.