vxcan: Support per-netns device unregistration.
PATCH 1/1 vxcan: Support per-netns device unregistration.
Currently, vxcan_dellink() unregisters both local and peer devices synchronously under RTNL.
Once RTNL is removed, it can be called concurrently from different netns.
Let's use xchg() and unregister_netdevice_queue_net() to support per-netns device unregistration.
This way, each device is queued for destruction only once by the winner of the race.
Note that the extra netdev_hold() ensures that @peer obtained by the first xchg() is not freed during the subsequent access to netdev_priv(peer). The 2nd xchg() overwrites @dev to balance the refcount.
Tested:
1. Create two vxcan pairs (vxcan1-2, vxcan3-4) between two netns (ns1 & ns2).
# ip netns add ns1 # ip netns add ns2 # ip -n ns1 link add vxcan1 type vxcan peer vxcan2 netns ns2 # ip -n ns1 link add vxcan3 type vxcan peer vxcan4 netns ns2
2. Run bpftrace to check if the same process does NOT unregister the paired vxcan devices
# bpftrace -e '#include <linux/netdevice.h>
kprobe:free_netdev {
$dev = (struct net_device *)arg0;
printf("PID: %d | DEV: %s%s\n", pid, $dev->name, kstack());
}'3. Remove vxcan2 in ns2 and check bpftrace output
# ip -n ns2 link del vxcan2
PID: 1524 | DEV: vxcan2
free_netdev+5
netdev_run_todo+4798
rtnl_dellink+1507
rtnetlink_rcv_msg+1791
netlink_rcv_skb+504
...
PID: 453 | DEV: vxcan1
free_netdev+5
netdev_run_todo+4798
process_scheduled_works+2538
worker_thread+1906
kthread+806
ret_from_fork+805
ret_from_fork_asm+174. Remove ns2 (thus vxcan4) and check bpftrace output
# ip netns del ns2
PID: 12 | DEV: vxcan4
free_netdev+5
netdev_run_todo+4798
default_device_exit_batch+2271
ops_undo_list+993
cleanup_net+1122
process_scheduled_works+2538
worker_thread+1906
kthread+806
ret_from_fork+805
ret_from_fork_asm+17
...
PID: 462 | DEV: vxcan3
free_netdev+5
netdev_run_todo+4798
process_scheduled_works+2538
worker_thread+1906
kthread+806
ret_from_fork+805
ret_from_fork_asm+17---
This mirros commit d7fda2c776b2 ("veth: Support per-netns device unregistration.").
The patch is based on net-next.git because it seems both main/master of linux-can.git and linux-can-next.git are not synced with net-next. 71ac90cca6d8 is needed to verify the test output.
--- drivers/net/can/vxcan.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-)
drivers/net/can/vxcan.c +15 -11
@@ -33,6 +33,7 @@ MODULE_ALIAS_RTNL_LINK(DRV_NAME);struct vxcan_priv {struct net_device __rcu *peer; +netdevice_tracker peer_tracker; };static netdev_tx_t vxcan_xmit (struct sk_buff *oskb,struct net_device *dev)
@@ -268,9 +269,11 @@ static int vxcan_newlink(struct net_device *dev,/* cross link the device pair */ priv =netdev_priv (dev);rcu_assign_pointer (priv->peer, peer); +netdev_hold (peer, &priv->peer_tracker,GFP_KERNEL ); priv =netdev_priv (peer);rcu_assign_pointer (priv->peer, dev); +netdev_hold (dev, &priv->peer_tracker,GFP_KERNEL );return 0 ;
@@ -281,24 +284,25 @@ static int vxcan_newlink(struct net_device *dev,static void vxcan_dellink (struct net_device *dev,struct list_head *head) { +netdevice_tracker *peer_tracker;struct vxcan_priv *priv;struct net_device *peer; priv =netdev_priv (dev); - peer =rtnl_dereference (priv->peer); + peer_tracker = &priv->peer_tracker; + peer =unrcu_pointer (xchg (&priv->peer,NULL )); +if (!peer) +return ; -/* Note : dellink() is called from default_device_exit_batch(), -* before a rcu_synchronize() point. The devices are guaranteed -* not being freed before one RCU grace period. -*/ -RCU_INIT_POINTER (priv->peer,NULL );unregister_netdevice_queue (dev, head); -if (peer) { - priv =netdev_priv (peer); -RCU_INIT_POINTER (priv->peer,NULL ); -unregister_netdevice_queue (peer, head); - } + priv =netdev_priv (peer); + dev =unrcu_pointer (xchg (&priv->peer,NULL )); +if (dev) +unregister_netdevice_queue_net (dev_net (dev), peer, head); + +netdev_put (peer, peer_tracker); +netdev_put (dev, &priv->peer_tracker); }static const struct nla_policy vxcan_policy[VXCAN_INFO_MAX +1 ] = {
Signature · 1 line
2.55.0.795.g602f6c329a-goog
[email protected]
AI reviewThank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Low] The patch calls the undefined function `unregister_netdevice_queue_net()`, resulting in a…
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] The patch calls the undefined function unregister_netdevice_queue_net(), resulting in a compilation error. --
--- Patch [1]: [PATCH v1] vxcan: Support per-netns device unregistration. --- commit 882fcd31630d9ce6b0a57cc99738c2ad0579763f
Author: Kuniyuki Iwashima <[email protected]>
vxcan: Support per-netns device unregistration.
This commit updates vxcan_dellink() to unregister local and peer devices safely across different namespaces by using xchg(). The change aims to prevent double-queuing of device destruction when RTNL is removed.
Quoted 3 lines
drivers/net/can/vxcan.c +0
-0
[ ... ]
Quoted 4 lines
drivers/net/can/vxcan.c +0
-0@@ -281,24 +284,25 @@ static int vxcan_newlink(struct net_device *dev,static void vxcan_dellink (struct net_device *dev,struct list_head *head) {
[ ... ]
drivers/net/can/vxcan.c static int vxcan_newlink(struct net_device *dev, · line 302
Quoted 11 lines
drivers/net/can/vxcan.c +4
-5unregister_netdevice_queue (dev, head); -if (peer) { - priv =netdev_priv (peer); -RCU_INIT_POINTER (priv->peer,NULL ); -unregister_netdevice_queue (peer, head); - } + priv =netdev_priv (peer); + dev =unrcu_pointer (xchg (&priv->peer,NULL )); +if (dev) +unregister_netdevice_queue_net (dev_net (dev), peer, head);
[Severity: Low]
Is this missing a prerequisite patch? It appears that unregister_netdevice_queue_net() is not defined anywhere in the kernel tree, which leads to a compilation error when building vxcan.c.
Quoted 4 lines
drivers/net/can/vxcan.c +3
-0+ +netdev_put (peer, peer_tracker); +netdev_put (dev, &priv->peer_tracker); }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
Quoted 29 lines · On 12.07.26 23:34, [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] The patch calls the undefined functionunregister_netdevice_queue_net(), resulting in a compilation error. ----- Patch [1]: [PATCH v1] vxcan: Support per-netns device unregistration. --- commit
882fcd31630d9ce6b0a57cc99738c2ad0579763f
Author: Kuniyuki Iwashima <[email protected]>vxcan: Support per-netns device unregistration.
This commit updates
vxcan_dellink()to unregister local and peer devices safely across different namespaces by usingxchg(). The change aims to prevent double-queuing of device destruction when RTNL is removed.Quoted 3 lines
drivers/net/can/vxcan.c +0
-0[ ... ]
Quoted 4 lines
drivers/net/can/vxcan.c +0
-0@@ -281,24 +284,25 @@ static int vxcan_newlink(struct net_device *dev,static void vxcan_dellink (struct net_device *dev,struct list_head *head) {[ ... ]
Quoted 2 lines
unregister_netdevice_queue(dev, head);
Quoted 14 lines
drivers/net/can/vxcan.c static int vxcan_newlink(struct net_device *dev, · line 302
Quoted 9 lines
drivers/net/can/vxcan.c +4
-5-if (peer) { - priv =netdev_priv (peer); -RCU_INIT_POINTER (priv->peer,NULL ); -unregister_netdevice_queue (peer, head); - } + priv =netdev_priv (peer); + dev =unrcu_pointer (xchg (&priv->peer,NULL )); +if (dev) +unregister_netdevice_queue_net (dev_net (dev), peer, head);[Severity: Low]
Is this missing a prerequisite patch? It appears thatunregister_netdevice_queue_net()is not defined anywhere in the kernel tree, which leads to a compilation error when buildingvxcan.c.
The referenced veth.c patch was applied to net-next:
where also unregister_netdevice_queue_net() exists.
So this is a net-next resp. can-next patch.
Regards,
Oliver
Quoted 6 lines
Quoted 5 lines
drivers/net/can/vxcan.c +3
-0+ +netdev_put (peer, peer_tracker); +netdev_put (dev, &priv->peer_tracker); }
Quoted 45 lines · Oliver Hartkopp
Quoted 39 lines · On 12.07.26 23:34, [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] The patch calls the undefined functionunregister_netdevice_queue_net(), resulting in a compilation error. ----- Patch [1]: [PATCH v1] vxcan: Support per-netns device unregistration. --- commit
882fcd31630d9ce6b0a57cc99738c2ad0579763f
Author: Kuniyuki Iwashima <[email protected]>vxcan: Support per-netns device unregistration.
This commit updates
vxcan_dellink()to unregister local and peer devices safely across different namespaces by usingxchg(). The change aims to prevent double-queuing of device destruction when RTNL is removed.Quoted 3 lines
drivers/net/can/vxcan.c +0
-0[ ... ]
Quoted 4 lines
drivers/net/can/vxcan.c +0
-0@@ -281,24 +284,25 @@ static int vxcan_newlink(struct net_device *dev,static void vxcan_dellink (struct net_device *dev,struct list_head *head) {[ ... ]
drivers/net/can/vxcan.c static int vxcan_newlink(struct net_device *dev, · line 302
Quoted 11 lines
unregister_netdevice_queue(dev, head);drivers/net/can/vxcan.c +4
-5-if (peer) { - priv =netdev_priv (peer); -RCU_INIT_POINTER (priv->peer,NULL ); -unregister_netdevice_queue (peer, head); - } + priv =netdev_priv (peer); + dev =unrcu_pointer (xchg (&priv->peer,NULL )); +if (dev) +unregister_netdevice_queue_net (dev_net (dev), peer, head);Quoted 4 lines
[Severity: Low]
Is this missing a prerequisite patch? It appears thatunregister_netdevice_queue_net()is not defined anywhere in the kernel tree, which leads to a compilation error when buildingvxcan.c.
Quoted 7 lines
The referenced
veth.cpatch was applied to net-next:where also
unregister_netdevice_queue_net()exists.So this is a net-next resp. can-next patch.
Yes, I meant this is for can-next, but I was not sure how frequently the tree is synced against net-next and what's the best approach here.