netfilter: ctnetlink: fix conntrack creation race

Conntrack creation through ctnetlink has two races:

- the timer may expire and free the conntrack concurrently, causing an
  invalid memory access when attempting to put it in the hash tables

- an identical conntrack entry may be created in the packet processing
  path in the time between the lookup and hash insertion

Hold the conntrack lock between the lookup and insertion to avoid this.

Reported-by: Zoltan Borbely <bozo@andrews.hu>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 3 insertions, 2 deletions

diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c
index a040d46f85d6..3b009a3e8549 100644
--- a/net/netfilter/nf_conntrack_netlink.c
+++ b/net/netfilter/nf_conntrack_netlink.c
@@ -1090,7 +1090,7 @@ ctnetlink_create_conntrack(struct nlattr *cda[],
 	struct nf_conn_help *help;
 	struct nf_conntrack_helper *helper;
 
-	ct = nf_conntrack_alloc(&init_net, otuple, rtuple, GFP_KERNEL);
+	ct = nf_conntrack_alloc(&init_net, otuple, rtuple, GFP_ATOMIC);
 	if (ct == NULL || IS_ERR(ct))
 		return -ENOMEM;
 
@@ -1212,13 +1212,14 @@ ctnetlink_new_conntrack(struct sock *ctnl, struct sk_buff *skb,
 			atomic_inc(&master_ct->ct_general.use);
 		}
 
-		spin_unlock_bh(&nf_conntrack_lock);
 		err = -ENOENT;
 		if (nlh->nlmsg_flags & NLM_F_CREATE)
 			err = ctnetlink_create_conntrack(cda,
 							 &otuple,
 							 &rtuple,
 							 master_ct);
+		spin_unlock_bh(&nf_conntrack_lock);
+
 		if (err < 0 && master_ct)
 			nf_ct_put(master_ct);