Skip to contentSkip to conversation

net/tun: Call netdev notifiers

PATCH In review Martin Schiller 2 participants 4 messages Nov 18, 2020 – Nov 23, 2020

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH v5 1/1 net/tun: Call netdev notifiers

Call netdev notifiers before and after changing the device type.

---

Changes to v4:
* Fix copy'n'paste error

Changes to v3:
* Handle return value of call_netdevice_notifiers()

Changes to v2:
* Use subject_prefix 'net-next' to fix 'fixes_present' issue

Changes to v1:
* Fix 'subject_prefix' and 'checkpatch' warnings

---
 drivers/net/tun.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/net/tun.c +10 -1
index 3d45d56172cb..7c62d82c57db 100644
@@ -3071,10 +3071,19 @@ static long __tun_chr_ioctl(struct file *file, unsigned int cmd,
 				   "Linktype set failed because interface is up\n");
 			ret = -EBUSY;
 		} else {
+			ret = call_netdevice_notifiers(NETDEV_PRE_TYPE_CHANGE,
+						       tun->dev);
+			ret = notifier_to_errno(ret);
+			if (ret) {
+				netif_info(tun, drv, tun->dev,
+					   "Refused to change device type\n");
+				break;
+			}
 			tun->dev->type = (int) arg;
 			netif_info(tun, drv, tun->dev, "linktype set to %d\n",
 				   tun->dev->type);
-			ret = 0;
+			call_netdevice_notifiers(NETDEV_POST_TYPE_CHANGE,
+						 tun->dev);
 		}
 		break;
 
Signature · 1 line
2.20.1
maintainer
#1 lore raw
Quoted 3 lines · On Wed, 18 Nov 2020 07:39:19 +0100 Martin Schiller wrote:

Call netdev notifiers before and after changing the device type.

This is a fix, right? Can you give an example of something that goes wrong without this patch?

#2 lore raw
Quoted 7 lines · Jakub Kicinski
Quoted 3 lines · On Wed, 18 Nov 2020 07:39:19 +0100 Martin Schiller wrote:

Call netdev notifiers before and after changing the device type.

This is a fix, right? Can you give an example of something that goes wrong without this patch?

This change is related to my latest patches to the X.25 Subsystem:
https://patchwork.kernel.org/project/netdevbpf/list/?series=388087

I use a tun interface in a XoT (X.25 over TCP) application and use the TUNSETLINK ioctl to change the device type to ARPHRD_X25. As the default device type is ARPHRD_NONE the initial NETDEV_REGISTER event won't be catched by the X.25 Stack.

Therefore I have to use the NETDEV_POST_TYPE_CHANGE to make sure that the corresponding neighbour structure is created.

I could imagine that other protocols have similar requirements.

Whether this is a fix or a functional extension is hard to say.

Some time ago there was also a corresponding patch for the WAN/HDLC subsystem:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit?id=2f8364a291e8

maintainer
#3 lore raw
Quoted 20 lines · On Mon, 23 Nov 2020 07:18:07 +0100 Martin Schiller wrote:
Quoted 7 lines · Jakub Kicinski
Quoted 3 lines · On Wed, 18 Nov 2020 07:39:19 +0100 Martin Schiller wrote:

Call netdev notifiers before and after changing the device type.

This is a fix, right? Can you give an example of something that goes wrong without this patch?

This change is related to my latest patches to the X.25 Subsystem:
https://patchwork.kernel.org/project/netdevbpf/list/?series=388087

I use a tun interface in a XoT (X.25 over TCP) application and use the TUNSETLINK ioctl to change the device type to ARPHRD_X25. As the default device type is ARPHRD_NONE the initial NETDEV_REGISTER event won't be catched by the X.25 Stack.

Therefore I have to use the NETDEV_POST_TYPE_CHANGE to make sure that the corresponding neighbour structure is created.

Quoted 8 lines

I could imagine that other protocols have similar requirements.

Whether this is a fix or a functional extension is hard to say.

Some time ago there was also a corresponding patch for the WAN/HDLC subsystem:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit?id=2f8364a291e8

Thanks for this info, applied to net-next.