ipv6: add key length check into rt6_select()
authorWei Wang <weiwan@google.com>
Fri, 6 Oct 2017 19:06:09 +0000 (12:06 -0700)
committerDavid S. Miller <davem@davemloft.net>
Sat, 7 Oct 2017 20:22:58 +0000 (21:22 +0100)
commit17ecf590b3cba19d9ecb410e340aa78128382abb
tree4e7110008039b7e656e34f8b098ee0650c831af7
parent8d1040e808bb2a5aeb4f0791b32bc7356a01ab11
ipv6: add key length check into rt6_select()

After rwlock is replaced with rcu and spinlock, fib6_lookup() could
potentially return an intermediate node if other thread is doing
fib6_del() on a route which is the only route on the node so that
fib6_repair_tree() will be called on this node and potentially assigns
fn->leaf to the its child's fn->leaf.

In order to detect this situation in rt6_select(), we have to check if
fn->fn_bit is consistent with the key length stored in the route. And
depending on if the fn is in the subtree or not, the key is either
rt->rt6i_dst or rt->rt6i_src.
If any inconsistency is found, that means the node no longer holds valid
routes in it. So net->ipv6.ip6_null_entry is returned.

Signed-off-by: Wei Wang <weiwan@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv6/route.c