[CFT][PATCH] net: Delay default_device_exit_batch until no devices are unregistering

[ebiederm@xmission.com (Eric W. Biederman)]

The implementation is a little rough but the logic should be right.

Device registration and unregistration is serialized with the rtnl_lock.
The final pieces of device unregistration do not happen under the
rtnl_lock resulting in the possibility that while we wait for the
refcount of a device to drop to zero the network namespace is
unregistered while no locks are held.

Prevent that by keeping a count of the network devices that are being
unregistered and before we make the final pass through a network
namespace to flush out all of the network devices, wait for the count of
network devices being unregistered to drop to zero.

Reported-by: Francesco Ruggeri <fruggeri@aristanetworks.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
---

Francesco could you take a look at this.  I am about 99% certain this is
right but I am starting to fade.  So it is entirely possible I missed
something.

 net/core/dev.c |   12 ++++++++++++
 1 files changed, 12 insertions(+), 0 deletions(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index 5d702fe..c25e6f3 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -5002,10 +5002,13 @@ static int dev_new_index(struct net *net)
 
 /* Delayed registration/unregisteration */
 static LIST_HEAD(net_todo_list);
+static atomic_t netdev_unregistering = ATOMIC_INIT(0);
+static DECLARE_WAIT_QUEUE_HEAD(netdev_unregistering_wait);
 
 static void net_set_todo(struct net_device *dev)
 {
 	list_add_tail(&dev->todo_list, &net_todo_list);
+	atomic_inc(&netdev_unregistering);
 }
 
 static void rollback_registered_many(struct list_head *head)
@@ -5673,6 +5676,9 @@ void netdev_run_todo(void)
 		if (dev->destructor)
 			dev->destructor(dev);
 
+		if (atomic_dec_and_test(&netdev_unregistering))
+			wake_up(&netdev_unregistering_wait);
+
 		/* Free network device */
 		kobject_put(&dev->dev.kobj);
 	}
@@ -6369,7 +6375,13 @@ static void __net_exit default_device_exit_batch(struct list_head *net_list)
 	struct net *net;
 	LIST_HEAD(dev_kill_list);
 
+retry:
+	wait_event(netdev_unregistering_wait, (atomic_read(&netdev_unregistering) == 0));
 	rtnl_lock();
+	if (atomic_read(&netdev_unregistering) != 0) {
+		__rtnl_unlock();
+		goto retry;
+	}
 	list_for_each_entry(net, net_list, exit_list) {
 		for_each_netdev_reverse(net, dev) {
 			if (dev->rtnl_link_ops)
-- 
1.7.5.4