If IFF_LINK0 is set, do IP-in-IP instead of Ethernet-in-IP. This will

be used to implement overlay networks and more flexible road-warrior
support.

diff --git a/sys/net/if_enc.c b/sys/net/if_enc.c
index 60a6d83..367428c 100644 (file)
--- a/sys/net/if_enc.c
+++ b/sys/net/if_enc.c
@@ -1,4 +1,4 @@
-/*     $OpenBSD: if_enc.c,v 1.21 2000/03/17 10:25:21 angelos Exp $     */
+/*     $OpenBSD: if_enc.c,v 1.22 2000/04/08 16:55:58 angelos Exp $     */
 
 /*
  * The authors of this code are John Ioannidis (ji@tla.org),
@@ -244,8 +244,11 @@ struct ifnet *ifp;
        if (m == NULL) /* Empty queue */
          return;
 
-       /* First, we encapsulate in etherip */
-       err = etherip_output(m, tdb, &mp, 0, 0); /* Last 2 args not used */
+       /* Encapsulate in etherip or ip-in-ip, depending on interface flag */
+       if (ifp->if_flags & IFF_LINK0)
+         err = ipip_output(m, tdb, &mp, 0, 0); /* Last 2 args not used */
+       else
+         err = etherip_output(m, tdb, &mp, 0, 0); /* Last 2 args not used */
        if ((mp == NULL) || err)
        {
            /* Just skip this frame */
@@ -265,7 +268,11 @@ struct ifnet *ifp;
        /* IPsec packet processing -- skip encapsulation */
        ipsp_process_packet(m, tdb, protoflag, 1);
 
-       /* XXX Should find a way to avoid bridging-loops, some mbuf flag ? */
+       /*
+        * XXX
+        * Should find a way to avoid bridging/routing-loops,
+        * perhaps use some mbuf flag ?
+        */
     }
 #endif /* IPSEC */
 }