netdev - [PATCH] appletalk: Pass IP-over-DDP packets through when 'ipddp0' interface is not present

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Tue, 23 Feb 2016 09:19:13 -0500
From:	Adam Seering <adam@...ring.org>
To:	Arnaldo Carvalho de Melo <acme@...stprotocols.net>,
	netdev@...r.kernel.org
Subject: [PATCH] appletalk: Pass IP-over-DDP packets through when 'ipddp0'
 interface is not present

Let userspace programs transmit and receive raw IP-over-DDP packets
with a kernel where "ipddp" was compiled as a module but is not loaded
(so no "ipddp0" network interface is exposed).  This makes the "module
is not loaded" behavior match the "module was never compiled" behavior.

Signed-off-by: Adam Seering <adam@...ring.org>

---

[edit] Re-sending at hopefully a more-appropriate time.

This is a small proposed change to the ddp code.  It's also my first
attempt at a kernel patch; I'd appreciate any feedback or discussion. 
I'm working on the Linux "macipgw" port; it duplicates some
functionality of the "ipddp" module, but the latter depends on
userspace utilities which I can no longer find up-to-date working
copies of?


diff --git a/net/appletalk/ddp.c b/net/appletalk/ddp.c
index d5871ac..d30e55f 100644
--- a/net/appletalk/ddp.c
+++ b/net/appletalk/ddp.c
@@ -1284,17 +1284,10 @@ static __inline__ int is_ip_over_ddp(struct sk_buff *skb)
        return skb->data[12] == 22;
 }
 
-static int handle_ip_over_ddp(struct sk_buff *skb)
+static int handle_ip_over_ddp(struct sk_buff *skb, struct net_device *dev)
 {
-       struct net_device *dev = __dev_get_by_name(&init_net, "ipddp0");
        struct net_device_stats *stats;
 
-       /* This needs to be able to handle ipddp"N" devices */
-       if (!dev) {
-               kfree_skb(skb);
-               return NET_RX_DROP;
-       }
-
        skb->protocol = htons(ETH_P_IP);
        skb_pull(skb, 13);
        skb->dev   = dev;
@@ -1308,7 +1301,7 @@ static int handle_ip_over_ddp(struct sk_buff *skb)
 #else
 /* make it easy for gcc to optimize this test out, i.e. kill the code */
 #define is_ip_over_ddp(skb) 0
-#define handle_ip_over_ddp(skb) 0
+#define handle_ip_over_ddp(skb, dev) 0
 #endif
 
 static int atalk_route_packet(struct sk_buff *skb, struct net_device *dev,
@@ -1418,6 +1411,8 @@ static int atalk_rcv(struct sk_buff *skb, struct net_device *dev,
        struct sock *sock;
        struct atalk_iface *atif;
        struct sockaddr_at tosat;
+       struct net_device *ipddp_dev;
+
        int origlen;
        __u16 len_hops;
 
@@ -1473,9 +1468,14 @@ static int atalk_rcv(struct sk_buff *skb, struct net_device *dev,
                return atalk_route_packet(skb, dev, ddp, len_hops, origlen);
        }
 
-       /* if IP over DDP is not selected this code will be optimized out */
-       if (is_ip_over_ddp(skb))
-               return handle_ip_over_ddp(skb);
+       /* if IP over DDP is not selected this code should be optimized out */
+       if (is_ip_over_ddp(skb)) {
+               ipddp_dev = __dev_get_by_name(&init_net, "ipddp0");
+
+               /* This needs to be able to handle ipddp"N" devices */
+               if (ipddp_dev)
+                       return handle_ip_over_ddp(skb, ipddp_dev);
+       }
        /*
         * Which socket - atalk_search_socket() looks for a *full match*
         * of the <net, node, port> tuple.