net: Fix high overhead of vlan sub-device teardown.

When a networking device is taken down that has a non-trivial number
of VLAN devices configured under it, we eat a full synchronize_net()
for every such VLAN device.

This is because of the call chain:

	NETDEV_DOWN notifier
	--> vlan_device_event()
		--> dev_change_flags()
		--> __dev_change_flags()
		--> __dev_close()
		--> __dev_close_many()
		--> dev_deactivate_many()
			--> synchronize_net()

This is kind of rediculous because we already have infrastructure for
batching doing operation X to a list of net devices so that we only
incur one sync.

So make use of that by exporting dev_close_many() and adjusting it's
interfaace so that the caller can fully manage the batch list.  Use
this in vlan_device_event() and all the overhead goes away.

Reported-by: Salam Noureddine <noureddine@arista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

2 files changed, 19 insertions, 7 deletions

diff --git a/net/8021q/vlan.c b/net/8021q/vlan.c
index 64c6bed4a3d3..98a30a5b8664 100644
--- a/net/8021q/vlan.c
+++ b/net/8021q/vlan.c
@@ -413,7 +413,10 @@ static int vlan_device_event(struct notifier_block *unused, unsigned long event,
 			vlan_transfer_features(dev, vlandev);
 		break;
 
-	case NETDEV_DOWN:
+	case NETDEV_DOWN: {
+		struct net_device *tmp;
+		LIST_HEAD(close_list);
+
 		if (dev->features & NETIF_F_HW_VLAN_CTAG_FILTER)
 			vlan_vid_del(dev, htons(ETH_P_8021Q), 0);
 
@@ -425,11 +428,18 @@ static int vlan_device_event(struct notifier_block *unused, unsigned long event,
 
 			vlan = vlan_dev_priv(vlandev);
 			if (!(vlan->flags & VLAN_FLAG_LOOSE_BINDING))
-				dev_change_flags(vlandev, flgs & ~IFF_UP);
+				list_add(&vlandev->close_list, &close_list);
+		}
+
+		dev_close_many(&close_list, false);
+
+		list_for_each_entry_safe(vlandev, tmp, &close_list, close_list) {
 			netif_stacked_transfer_operstate(dev, vlandev);
+			list_del_init(&vlandev->close_list);
 		}
+		list_del(&close_list);
 		break;
-
+	}
 	case NETDEV_UP:
 		/* Put all VLANs for this dev in the up state too.  */
 		vlan_group_for_each_dev(grp, i, vlandev) {
diff --git a/net/core/dev.c b/net/core/dev.c
index a1f24151db5b..5d43e010ef87 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1385,7 +1385,7 @@ static int __dev_close(struct net_device *dev)
 	return retval;
 }
 
-static int dev_close_many(struct list_head *head)
+int dev_close_many(struct list_head *head, bool unlink)
 {
 	struct net_device *dev, *tmp;
 
@@ -1399,11 +1399,13 @@ static int dev_close_many(struct list_head *head)
 	list_for_each_entry_safe(dev, tmp, head, close_list) {
 		rtmsg_ifinfo(RTM_NEWLINK, dev, IFF_UP|IFF_RUNNING, GFP_KERNEL);
 		call_netdevice_notifiers(NETDEV_DOWN, dev);
-		list_del_init(&dev->close_list);
+		if (unlink)
+			list_del_init(&dev->close_list);
 	}
 
 	return 0;
 }
+EXPORT_SYMBOL(dev_close_many);
 
 /**
  *	dev_close - shutdown an interface.
@@ -1420,7 +1422,7 @@ int dev_close(struct net_device *dev)
 		LIST_HEAD(single);
 
 		list_add(&dev->close_list, &single);
-		dev_close_many(&single);
+		dev_close_many(&single, true);
 		list_del(&single);
 	}
 	return 0;
@@ -5986,7 +5988,7 @@ static void rollback_registered_many(struct list_head *head)
 	/* If device is running, close it first. */
 	list_for_each_entry(dev, head, unreg_list)
 		list_add_tail(&dev->close_list, &close_head);
-	dev_close_many(&close_head);
+	dev_close_many(&close_head, true);
 
 	list_for_each_entry(dev, head, unreg_list) {
 		/* And unlink it from device chain. */