OSDN Git Service

net: openvswitch: ensure LSE is pullable before reading it
authorDavide Caratti <dcaratti@redhat.com>
Thu, 3 Dec 2020 09:46:06 +0000 (10:46 +0100)
committerJakub Kicinski <kuba@kernel.org>
Thu, 3 Dec 2020 19:13:29 +0000 (11:13 -0800)
when openvswitch is configured to mangle the LSE, the current value is
read from the packet dereferencing 4 bytes at mpls_hdr(): ensure that
the label is contained in the skb "linear" area.

Found by code inspection.

Fixes: d27cf5c59a12 ("net: core: add MPLS update core helper and use in OvS")
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Link: https://lore.kernel.org/r/aa099f245d93218b84b5c056b67b6058ccf81a66.1606987185.git.dcaratti@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
net/openvswitch/actions.c

index 5829a02..c3a6648 100644 (file)
@@ -199,6 +199,9 @@ static int set_mpls(struct sk_buff *skb, struct sw_flow_key *flow_key,
        __be32 lse;
        int err;
 
+       if (!pskb_may_pull(skb, skb_network_offset(skb) + MPLS_HLEN))
+               return -ENOMEM;
+
        stack = mpls_hdr(skb);
        lse = OVS_MASKED(stack->label_stack_entry, *mpls_lse, *mask);
        err = skb_mpls_update_lse(skb, lse);