We currently use the negative value in the conntrack code to encode
the packet verdict in the error. As NF_DROP is equal to 0, inverting
NF_DROP makes no sense and, as a result, no packets are ever dropped.
Signed-off-by: Christoph Paasch <christoph.paasch@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Patrick McHardy <kaber@trash.net>
        NF_CT_ASSERT(skb->nfct);
 
        ret = l4proto->packet(ct, skb, dataoff, ctinfo, pf, hooknum);
-       if (ret < 0) {
+       if (ret <= 0) {
                /* Invalid: inverse of the return code tells
                 * the netfilter core what to do */
                pr_debug("nf_conntrack_in: Can't track with proto module\n");
 
                         */
                        if (nf_ct_kill(ct))
                                return -NF_REPEAT;
-                       return -NF_DROP;
+                       return NF_DROP;
                }
                /* Fall through */
        case TCP_CONNTRACK_IGNORE:
                                nf_log_packet(pf, 0, skb, NULL, NULL, NULL,
                                          "nf_ct_tcp: killing out of sync session ");
                        nf_ct_kill(ct);
-                       return -NF_DROP;
+                       return NF_DROP;
                }
                ct->proto.tcp.last_index = index;
                ct->proto.tcp.last_dir = dir;