ipvs: correct usage/allocation of seqadj ext in ipvs
authorJesper Dangaard Brouer <brouer@redhat.com>
Mon, 16 Dec 2013 16:09:47 +0000 (17:09 +0100)
committerSimon Horman <horms@verge.net.au>
Fri, 27 Dec 2013 03:30:02 +0000 (12:30 +0900)
The IPVS FTP helper ip_vs_ftp could trigger an OOPS in nf_ct_seqadj_set,
after commit 41d73ec053d2 (netfilter: nf_conntrack: make sequence number
adjustments usuable without NAT).

This is because, the seqadj ext is now allocated dynamically, and the
IPVS code didn't handle this situation.  Fix this in the IPVS nfct
code by invoking the alloc function nfct_seqadj_ext_add().

Fixes: 41d73ec053d2 (netfilter: nf_conntrack: make sequence number adjustments usuable without NAT)
Suggested-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
net/netfilter/ipvs/ip_vs_nfct.c

index c8beafd401aa283f0befa0494e02649359bab83f..5a355a46d1dc4a70e82fadcd509ddd4fae34c3e1 100644 (file)
@@ -63,6 +63,7 @@
 #include <net/ip_vs.h>
 #include <net/netfilter/nf_conntrack_core.h>
 #include <net/netfilter/nf_conntrack_expect.h>
+#include <net/netfilter/nf_conntrack_seqadj.h>
 #include <net/netfilter/nf_conntrack_helper.h>
 #include <net/netfilter/nf_conntrack_zones.h>
 
@@ -97,6 +98,11 @@ ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp, int outin)
        if (CTINFO2DIR(ctinfo) != IP_CT_DIR_ORIGINAL)
                return;
 
+       /* Applications may adjust TCP seqs */
+       if (cp->app && nf_ct_protonum(ct) == IPPROTO_TCP &&
+           !nfct_seqadj(ct) && !nfct_seqadj_ext_add(ct))
+               return;
+
        /*
         * The connection is not yet in the hashtable, so we update it.
         * CIP->VIP will remain the same, so leave the tuple in