net: Do not unconditionally turn on carrier when turning off protodown

The protodown functionality allows user space to turn off the carrier of
a net device:

 # ip link add name dummy1 up type dummy
 # ip link add name macvlan1 up link dummy1 type macvlan mode bridge
 # ip link set dev macvlan1 protodown on
 $ ip -br link show dev macvlan1
 macvlan1@dummy1  DOWN           0a:5c:a3:05:c7:86 <NO-CARRIER,BROADCAST,MULTICAST,UP>

When protodown is turned off, the core unconditionally turns on the
carrier of the net device:

 # ip link set dev macvlan1 protodown off
 $ ip -br link show dev macvlan1
 macvlan1@dummy1  UP             0a:5c:a3:05:c7:86 <BROADCAST,MULTICAST,UP,LOWER_UP>

This is wrong as it means that a macvlan can end up with a carrier when
its lower device does not have a carrier:

 # ip link set dev dummy1 carrier off
 $ ip -br link show dev macvlan1
 macvlan1@dummy1  LOWERLAYERDOWN 0a:5c:a3:05:c7:86 <NO-CARRIER,BROADCAST,MULTICAST,UP>
 # ip link set dev macvlan1 protodown on
 # ip link set dev macvlan1 protodown off
 $ ip -br link show dev macvlan1
 macvlan1@dummy1  UP             0a:5c:a3:05:c7:86 <BROADCAST,MULTICAST,UP,LOWER_UP>

Solve this by resolving the linked net device and if one exists, inherit
its carrier state when protodown is turned off. Otherwise, if no linked
net device exists, as before, simply turn on the carrier.

Resolve the linked net device using a new helper and have it return the
device itself (in a similar fashion to dev_get_iflink()) if the device
does not implement both ndo_get_iflink() and get_link_net(). If the
latter is not implemented, it is unclear in which network namespace we
should look up the linked net device. Currently, this helper is only
used for net devices that support protodown (macvlan and vxlan) and for
both it returns the correct result.

Output with the patch:

 # ip link add name dummy1 up type dummy
 # ip link add name macvlan1 up link dummy1 type macvlan mode bridge
 # ip link set dev dummy1 carrier off
 $ ip -br link show dev macvlan1
 macvlan1@dummy1  LOWERLAYERDOWN 0a:5c:a3:05:c7:86 <NO-CARRIER,BROADCAST,MULTICAST,UP>
 # ip link set dev macvlan1 protodown on
 # ip link set dev macvlan1 protodown off
 $ ip -br link show dev macvlan1
 macvlan1@dummy1  LOWERLAYERDOWN 0a:5c:a3:05:c7:86 <NO-CARRIER,BROADCAST,MULTICAST,UP>
 # ip link set dev dummy1 carrier on
 $ ip -br link show dev macvlan1
 macvlan1@dummy1  UP             0a:5c:a3:05:c7:86 <BROADCAST,MULTICAST,UP,LOWER_UP>
 # ip link set dev macvlan1 protodown on
 # ip link set dev macvlan1 protodown off
 $ ip -br link show dev macvlan1
 macvlan1@dummy1  UP             0a:5c:a3:05:c7:86 <BROADCAST,MULTICAST,UP,LOWER_UP>

Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20260507105906.891817-4-idosch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

diff --git a/net/core/dev.c b/net/core/dev.c
index 02d16be3457b1bd8bd4fca3385a93dcf89c26960..b0691e03dd6b00c38ad929ba3f8ee3d2612b680a 100644 (file)
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -10141,16 +10141,35 @@ bool netdev_port_same_parent_id(struct net_device *a, struct net_device *b)
 }
 EXPORT_SYMBOL(netdev_port_same_parent_id);
 
+static struct net_device *dev_get_iflink_dev(struct net_device *dev)
+{
+       struct net *net;
+
+       ASSERT_RTNL();
+
+       if (!dev->netdev_ops->ndo_get_iflink || !dev->rtnl_link_ops ||
+           !dev->rtnl_link_ops->get_link_net)
+               return dev;
+
+       net = dev->rtnl_link_ops->get_link_net(dev);
+       return __dev_get_by_index(net, dev_get_iflink(dev));
+}
+
 int netif_change_proto_down(struct net_device *dev, bool proto_down)
 {
+       struct net_device *iflink_dev;
+
        if (!dev->change_proto_down)
                return -EOPNOTSUPP;
        if (!netif_device_present(dev))
                return -ENODEV;
+       iflink_dev = dev_get_iflink_dev(dev);
+       if (!iflink_dev)
+               return -ENODEV;
        WRITE_ONCE(dev->proto_down, proto_down);
        if (proto_down)
                netif_carrier_off(dev);
-       else
+       else if (dev == iflink_dev || netif_carrier_ok(iflink_dev))
                netif_carrier_on(dev);
        return 0;
 }