* [Patch net-next] vxlan: do not use vxlan_net before checking event type
@ 2014-01-17 4:20 Cong Wang
2014-01-17 4:28 ` Fan Du
0 siblings, 1 reply; 3+ messages in thread
From: Cong Wang @ 2014-01-17 4:20 UTC (permalink / raw)
To: netdev; +Cc: Daniel Borkmann, David S. Miller, Cong Wang
When cloning a netns, loopback device will be registered
and therefore an event will be notified. Of course
vxlan doesn't care about it, therefore should check if it
is NETDEV_UNREGISTER before getting the vxlan_net struct.
Otherwise, vxlan_net is probably not initialized yet at
this point.
Fixes: commit acaf4e70997ff5ef3588f5a
Reported-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Cc: Daniel Borkmann <dborkman@redhat.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
---
diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index a2dee80..2812559 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -2655,36 +2655,30 @@ static struct rtnl_link_ops vxlan_link_ops __read_mostly = {
.fill_info = vxlan_fill_info,
};
-static void vxlan_handle_lowerdev_unregister(struct vxlan_net *vn,
- struct net_device *dev)
-{
- struct vxlan_dev *vxlan, *next;
- LIST_HEAD(list_kill);
-
- list_for_each_entry_safe(vxlan, next, &vn->vxlan_list, next) {
- struct vxlan_rdst *dst = &vxlan->default_dst;
-
- /* In case we created vxlan device with carrier
- * and we loose the carrier due to module unload
- * we also need to remove vxlan device. In other
- * cases, it's not necessary and remote_ifindex
- * is 0 here, so no matches.
- */
- if (dst->remote_ifindex == dev->ifindex)
- vxlan_dellink(vxlan->dev, &list_kill);
- }
-
- unregister_netdevice_many(&list_kill);
-}
-
static int vxlan_lowerdev_event(struct notifier_block *unused,
unsigned long event, void *ptr)
{
- struct net_device *dev = netdev_notifier_info_to_dev(ptr);
- struct vxlan_net *vn = net_generic(dev_net(dev), vxlan_net_id);
+ if (event == NETDEV_UNREGISTER) {
+ struct net_device *dev = netdev_notifier_info_to_dev(ptr);
+ struct vxlan_net *vn = net_generic(dev_net(dev), vxlan_net_id);
+ struct vxlan_dev *vxlan, *next;
+ LIST_HEAD(list_kill);
+
+ list_for_each_entry_safe(vxlan, next, &vn->vxlan_list, next) {
+ struct vxlan_rdst *dst = &vxlan->default_dst;
+
+ /* In case we created vxlan device with carrier
+ * and we loose the carrier due to module unload
+ * we also need to remove vxlan device. In other
+ * cases, it's not necessary and remote_ifindex
+ * is 0 here, so no matches.
+ */
+ if (dst->remote_ifindex == dev->ifindex)
+ vxlan_dellink(vxlan->dev, &list_kill);
+ }
- if (event == NETDEV_UNREGISTER)
- vxlan_handle_lowerdev_unregister(vn, dev);
+ unregister_netdevice_many(&list_kill);
+ }
return NOTIFY_DONE;
}
^ permalink raw reply related [flat|nested] 3+ messages in thread* Re: [Patch net-next] vxlan: do not use vxlan_net before checking event type
2014-01-17 4:20 [Patch net-next] vxlan: do not use vxlan_net before checking event type Cong Wang
@ 2014-01-17 4:28 ` Fan Du
2014-01-17 8:39 ` Daniel Borkmann
0 siblings, 1 reply; 3+ messages in thread
From: Fan Du @ 2014-01-17 4:28 UTC (permalink / raw)
To: Cong Wang; +Cc: netdev, Daniel Borkmann, David S. Miller
Hi, Cong
On 2014年01月17日 12:20, Cong Wang wrote:
> When cloning a netns, loopback device will be registered
> and therefore an event will be notified. Of course
> vxlan doesn't care about it, therefore should check if it
> is NETDEV_UNREGISTER before getting the vxlan_net struct.
> Otherwise, vxlan_net is probably not initialized yet at
> this point.
I'm bit new to vxlan, but in vxlan_init_module
register_pernet_device is called before register_netdevice_notifier.
By my understanding once register_pernet_device is called,
then subsequent vxlan_notifier_block callback see a valid vxlan_net_id.
I mean execute vxlan_notifier_block callback indicates a valid vxlan_net_id,
or I miss somewhere else.
--
浮沉随浪只记今朝笑
--fan
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [Patch net-next] vxlan: do not use vxlan_net before checking event type
2014-01-17 4:28 ` Fan Du
@ 2014-01-17 8:39 ` Daniel Borkmann
0 siblings, 0 replies; 3+ messages in thread
From: Daniel Borkmann @ 2014-01-17 8:39 UTC (permalink / raw)
To: Fan Du; +Cc: Cong Wang, netdev, David S. Miller
On 01/17/2014 05:28 AM, Fan Du wrote:
> Hi, Cong
>
> On 2014年01月17日 12:20, Cong Wang wrote:
>> When cloning a netns, loopback device will be registered
>> and therefore an event will be notified. Of course
>> vxlan doesn't care about it, therefore should check if it
>> is NETDEV_UNREGISTER before getting the vxlan_net struct.
>> Otherwise, vxlan_net is probably not initialized yet at
>> this point.
>
> I'm bit new to vxlan, but in vxlan_init_module
>
> register_pernet_device is called before register_netdevice_notifier.
> By my understanding once register_pernet_device is called,
> then subsequent vxlan_notifier_block callback see a valid vxlan_net_id.
> I mean execute vxlan_notifier_block callback indicates a valid vxlan_net_id,
> or I miss somewhere else.
That was also my understanding, but apparently I missed that.
Very sorry. Then, it would have been enough to just provide this:
diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index a2dee80..d6ec71f 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -2681,10 +2681,12 @@ static int vxlan_lowerdev_event(struct notifier_block *unused,
unsigned long event, void *ptr)
{
struct net_device *dev = netdev_notifier_info_to_dev(ptr);
- struct vxlan_net *vn = net_generic(dev_net(dev), vxlan_net_id);
+ struct vxlan_net *vn;
- if (event == NETDEV_UNREGISTER)
+ if (event == NETDEV_UNREGISTER) {
+ vn = net_generic(dev_net(dev), vxlan_net_id);
vxlan_handle_lowerdev_unregister(vn, dev);
+ }
return NOTIFY_DONE;
}
^ permalink raw reply related [flat|nested] 3+ messages in thread
end of thread, other threads:[~2014-01-17 8:39 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2014-01-17 4:20 [Patch net-next] vxlan: do not use vxlan_net before checking event type Cong Wang
2014-01-17 4:28 ` Fan Du
2014-01-17 8:39 ` Daniel Borkmann
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).