[PATCH] NETFILTER: {ip,nf}_conntrack: fix use-after-free in helper destroy callback invocation

When the helper module is removed for a master connection that has a
fulfilled expectation, but has already timed out and got removed from
the hash tables, nf_conntrack_helper_unregister can't find the master
connection to unset the helper, causing a use-after-free when the
expected connection is destroyed and releases the last reference to
the master.

The helper destroy callback was introduced for the PPtP helper to clean
up expectations and expected connections when the master connection
times out, but doing this from destroy_conntrack only works for
unfulfilled expectations since expected connections hold a reference
to the master, preventing its destruction. Move the destroy callback to
the timeout function, which fixes both problems.

Reported/tested by Gabor Burjan <buga@buvoshetes.hu>.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>

net/ipv4/netfilter/ip_conntrack_core.c

@@ -302,7 +302,6 @@ destroy_conntrack(struct nf_conntrack *nfct)
 {
 	struct ip_conntrack *ct = (struct ip_conntrack *)nfct;
 	struct ip_conntrack_protocol *proto;
-	struct ip_conntrack_helper *helper;
 	typeof(ip_conntrack_destroyed) destroyed;
 
 	DEBUGP("destroy_conntrack(%p)\n", ct);
@@ -312,10 +311,6 @@ destroy_conntrack(struct nf_conntrack *nfct)
 	ip_conntrack_event(IPCT_DESTROY, ct);
 	set_bit(IPS_DYING_BIT, &ct->status);
 
-	helper = ct->helper;
-	if (helper && helper->destroy)
-		helper->destroy(ct);
-
 	/* To make sure we don't get any weird locking issues here:
 	 * destroy_conntrack() MUST NOT be called with a write lock
 	 * to ip_conntrack_lock!!! -HW */
@@ -356,6 +351,11 @@ destroy_conntrack(struct nf_conntrack *nfct)
 static void death_by_timeout(unsigned long ul_conntrack)
 {
 	struct ip_conntrack *ct = (void *)ul_conntrack;
+	struct ip_conntrack_helper *helper;
+
+	helper = ct->helper;
+	if (helper && helper->destroy)
+		helper->destroy(ct);
 
 	write_lock_bh(&ip_conntrack_lock);
 	/* Inside lock so preempt is disabled on module removal path.

net/netfilter/nf_conntrack_core.c

@@ -315,7 +315,6 @@ static void
 destroy_conntrack(struct nf_conntrack *nfct)
 {
 	struct nf_conn *ct = (struct nf_conn *)nfct;
-	struct nf_conn_help *help = nfct_help(ct);
 	struct nf_conntrack_l3proto *l3proto;
 	struct nf_conntrack_l4proto *l4proto;
 	typeof(nf_conntrack_destroyed) destroyed;
@@ -327,9 +326,6 @@ destroy_conntrack(struct nf_conntrack *nfct)
 	nf_conntrack_event(IPCT_DESTROY, ct);
 	set_bit(IPS_DYING_BIT, &ct->status);
 
-	if (help && help->helper && help->helper->destroy)
-		help->helper->destroy(ct);
-
 	/* To make sure we don't get any weird locking issues here:
 	 * destroy_conntrack() MUST NOT be called with a write lock
 	 * to nf_conntrack_lock!!! -HW */
@@ -375,6 +371,10 @@ destroy_conntrack(struct nf_conntrack *nfct)
 static void death_by_timeout(unsigned long ul_conntrack)
 {
 	struct nf_conn *ct = (void *)ul_conntrack;
+	struct nf_conn_help *help = nfct_help(ct);
+
+	if (help && help->helper && help->helper->destroy)
+		help->helper->destroy(ct);
 
 	write_lock_bh(&nf_conntrack_lock);
 	/* Inside lock so preempt is disabled on module removal path.