s390/qeth: fix L3 next-hop in xmit qeth hdr
authorJulian Wiedmann <jwi@linux.vnet.ibm.com>
Mon, 7 Aug 2017 11:28:39 +0000 (13:28 +0200)
committerDavid S. Miller <davem@davemloft.net>
Mon, 7 Aug 2017 18:24:37 +0000 (11:24 -0700)
On L3, the qeth_hdr struct needs to be filled with the next-hop
IP address.
The current code accesses rtable->rt_gateway without checking that
rtable is a valid address. The accidental access to a lowcore area
results in a random next-hop address in the qeth_hdr.
rtable (or more precisely, skb_dst(skb)) can be NULL in rare cases
(for instance together with AF_PACKET sockets).
This patch adds the missing NULL-ptr checks.

Signed-off-by: Julian Wiedmann <jwi@linux.vnet.ibm.com>
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Fixes: 87e7597b5a3 qeth: Move away from using neighbour entries in qeth_l3_fill_header()
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/s390/net/qeth_l3_main.c

index 8975cd32139047cf03afe9a93544cc919eaed661..d42e758518ed92e33dfd6c6aaf2a4ee2ec198ff0 100644 (file)
@@ -2512,7 +2512,7 @@ static void qeth_l3_fill_header(struct qeth_card *card, struct qeth_hdr *hdr,
                struct rtable *rt = (struct rtable *) dst;
                __be32 *pkey = &ip_hdr(skb)->daddr;
 
-               if (rt->rt_gateway)
+               if (rt && rt->rt_gateway)
                        pkey = &rt->rt_gateway;
 
                /* IPv4 */
@@ -2523,7 +2523,7 @@ static void qeth_l3_fill_header(struct qeth_card *card, struct qeth_hdr *hdr,
                struct rt6_info *rt = (struct rt6_info *) dst;
                struct in6_addr *pkey = &ipv6_hdr(skb)->daddr;
 
-               if (!ipv6_addr_any(&rt->rt6i_gateway))
+               if (rt && !ipv6_addr_any(&rt->rt6i_gateway))
                        pkey = &rt->rt6i_gateway;
 
                /* IPv6 */