It is possible for the DSA slave network device not to be part of a
bridge, yet have an upper device like a VLAN device be part of a bridge.
When that VLAN device is enslaved, since it does not define any
switchdev_ops, we will recurse down to the lower/physical port device,
call switchdev_port_obj_add() with a VLAN, and here we will check
br_vlan_enabled() on a NULL dp->bridge_dev, thus causing a NULL pointer
de-reference.

This is no longer a problem upstream after commit d17d9f5e5143
("switchdev: Replace port obj add/del SDO with a notification").

Fixes: 2ea7a679ca2a ("net: dsa: Don't add vlans when vlan filtering is 
disabled")
Reported-by: Frank Wunderlich <fran...@public-files.de>
Signed-off-by: Florian Fainelli <f.faine...@gmail.com>
---
David,

This affects v4.15..v4.19, but not v5.0-rc1 can you still queue this up
for -stable? Thanks!

 net/dsa/port.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/dsa/port.c b/net/dsa/port.c
index ed0595459df1..792a13068c50 100644
--- a/net/dsa/port.c
+++ b/net/dsa/port.c
@@ -255,7 +255,7 @@ int dsa_port_vlan_add(struct dsa_port *dp,
        if (netif_is_bridge_master(vlan->obj.orig_dev))
                return -EOPNOTSUPP;
 
-       if (br_vlan_enabled(dp->bridge_dev))
+       if (dp->bridge_dev && br_vlan_enabled(dp->bridge_dev))
                return dsa_port_notify(dp, DSA_NOTIFIER_VLAN_ADD, &info);
 
        return 0;
@@ -273,7 +273,7 @@ int dsa_port_vlan_del(struct dsa_port *dp,
        if (netif_is_bridge_master(vlan->obj.orig_dev))
                return -EOPNOTSUPP;
 
-       if (br_vlan_enabled(dp->bridge_dev))
+       if (dp->bridge_dev && br_vlan_enabled(dp->bridge_dev))
                return dsa_port_notify(dp, DSA_NOTIFIER_VLAN_DEL, &info);
 
        return 0;
-- 
2.19.1

Reply via email to