net: add recursion limit to GRO

[ Upstream commit fcd91dd449867c6bfe56a81cabba76b829fd05cd ]

Currently, GRO can do unlimited recursion through the gro_receive
handlers.  This was fixed for tunneling protocols by limiting tunnel GRO
to one level with encap_mark, but both VLAN and TEB still have this
problem.  Thus, the kernel is vulnerable to a stack overflow, if we
receive a packet composed entirely of VLAN headers.

This patch adds a recursion counter to the GRO layer to prevent stack
overflow.  When a gro_receive function hits the recursion limit, GRO is
aborted for this skb and it is processed normally.  This recursion
counter is put in the GRO CB, but could be turned into a percpu counter
if we run out of space in the CB.

Thanks to Vladimír Beneš <vbenes@redhat.com> for the initial bug report.

Fixes: CVE-2016-7039
Fixes: 9b174d88c257 ("net: Add Transparent Ethernet Bridging GRO support.")
Fixes: 66e5133f19e9 ("vlan: Add GRO support for non hardware accelerated vlan")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Reviewed-by: Jiri Benc <jbenc@redhat.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

11 files changed, 51 insertions, 12 deletions

diff --git a/drivers/net/geneve.c b/drivers/net/geneve.c
index 69e31e2a68fc..4827c6987ac3 100644
--- a/drivers/net/geneve.c
+++ b/drivers/net/geneve.c
@@ -440,7 +440,7 @@ static struct sk_buff **geneve_gro_receive(struct sk_buff **head,
 
 	skb_gro_pull(skb, gh_len);
 	skb_gro_postpull_rcsum(skb, gh, gh_len);
-	pp = ptype->callbacks.gro_receive(head, skb);
+	pp = call_gro_receive(ptype->callbacks.gro_receive, head, skb);
 
 out_unlock:
 	rcu_read_unlock();
diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index 003780901628..6fa8e165878e 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -593,7 +593,7 @@ static struct sk_buff **vxlan_gro_receive(struct sk_buff **head,
 		}
 	}
 
-	pp = eth_gro_receive(head, skb);
+	pp = call_gro_receive(eth_gro_receive, head, skb);
 
 out:
 	skb_gro_remcsum_cleanup(skb, &grc);
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 12b4d54a8ffa..9d6025703f73 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -2003,7 +2003,10 @@ struct napi_gro_cb {
 	/* Used in foo-over-udp, set in udp[46]_gro_receive */
 	u8	is_ipv6:1;
 
-	/* 7 bit hole */
+	/* Number of gro_receive callbacks this packet already went through */
+	u8 recursion_counter:4;
+
+	/* 3 bit hole */
 
 	/* used to support CHECKSUM_COMPLETE for tunneling protocols */
 	__wsum	csum;
@@ -2014,6 +2017,25 @@ struct napi_gro_cb {
 
 #define NAPI_GRO_CB(skb) ((struct napi_gro_cb *)(skb)->cb)
 
+#define GRO_RECURSION_LIMIT 15
+static inline int gro_recursion_inc_test(struct sk_buff *skb)
+{
+	return ++NAPI_GRO_CB(skb)->recursion_counter == GRO_RECURSION_LIMIT;
+}
+
+typedef struct sk_buff **(*gro_receive_t)(struct sk_buff **, struct sk_buff *);
+static inline struct sk_buff **call_gro_receive(gro_receive_t cb,
+						struct sk_buff **head,
+						struct sk_buff *skb)
+{
+	if (unlikely(gro_recursion_inc_test(skb))) {
+		NAPI_GRO_CB(skb)->flush |= 1;
+		return NULL;
+	}
+
+	return cb(head, skb);
+}
+
 struct packet_type {
 	__be16			type;	/* This is really htons(ether_type). */
 	struct net_device	*dev;	/* NULL is wildcarded here	     */
@@ -2059,6 +2081,22 @@ struct udp_offload {
 	struct udp_offload_callbacks callbacks;
 };
 
+typedef struct sk_buff **(*gro_receive_udp_t)(struct sk_buff **,
+					      struct sk_buff *,
+					      struct udp_offload *);
+static inline struct sk_buff **call_gro_receive_udp(gro_receive_udp_t cb,
+						    struct sk_buff **head,
+						    struct sk_buff *skb,
+						    struct udp_offload *uoff)
+{
+	if (unlikely(gro_recursion_inc_test(skb))) {
+		NAPI_GRO_CB(skb)->flush |= 1;
+		return NULL;
+	}
+
+	return cb(head, skb, uoff);
+}
+
 /* often modified stats are per cpu, other are shared (netdev->stats) */
 struct pcpu_sw_netstats {
 	u64     rx_packets;
diff --git a/net/8021q/vlan.c b/net/8021q/vlan.c
index d2cd9de4b724..ad8d6e6b87ca 100644
--- a/net/8021q/vlan.c
+++ b/net/8021q/vlan.c
@@ -659,7 +659,7 @@ static struct sk_buff **vlan_gro_receive(struct sk_buff **head,
 
 	skb_gro_pull(skb, sizeof(*vhdr));
 	skb_gro_postpull_rcsum(skb, vhdr, sizeof(*vhdr));
-	pp = ptype->callbacks.gro_receive(head, skb);
+	pp = call_gro_receive(ptype->callbacks.gro_receive, head, skb);
 
 out_unlock:
 	rcu_read_unlock();
diff --git a/net/core/dev.c b/net/core/dev.c
index 5d9ec0458998..d200a7ccbde6 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -4240,6 +4240,7 @@ static enum gro_result dev_gro_receive(struct napi_struct *napi, struct sk_buff
 		NAPI_GRO_CB(skb)->flush = 0;
 		NAPI_GRO_CB(skb)->free = 0;
 		NAPI_GRO_CB(skb)->encap_mark = 0;
+		NAPI_GRO_CB(skb)->recursion_counter = 0;
 		NAPI_GRO_CB(skb)->gro_remcsum_start = 0;
 
 		/* Setup for GRO checksum validation */
diff --git a/net/ethernet/eth.c b/net/ethernet/eth.c
index 9e63f252a89e..de85d4e1cf43 100644
--- a/net/ethernet/eth.c
+++ b/net/ethernet/eth.c
@@ -436,7 +436,7 @@ struct sk_buff **eth_gro_receive(struct sk_buff **head,
 
 	skb_gro_pull(skb, sizeof(*eh));
 	skb_gro_postpull_rcsum(skb, eh, sizeof(*eh));
-	pp = ptype->callbacks.gro_receive(head, skb);
+	pp = call_gro_receive(ptype->callbacks.gro_receive, head, skb);
 
 out_unlock:
 	rcu_read_unlock();
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 1a5c1ca3ad3c..afc18e9ca94a 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -1372,7 +1372,7 @@ static struct sk_buff **inet_gro_receive(struct sk_buff **head,
 	skb_gro_pull(skb, sizeof(*iph));
 	skb_set_transport_header(skb, skb_gro_offset(skb));
 
-	pp = ops->callbacks.gro_receive(head, skb);
+	pp = call_gro_receive(ops->callbacks.gro_receive, head, skb);
 
 out_unlock:
 	rcu_read_unlock();
diff --git a/net/ipv4/fou.c b/net/ipv4/fou.c
index 08d7de55e57e..08d8ee124538 100644
--- a/net/ipv4/fou.c
+++ b/net/ipv4/fou.c
@@ -201,7 +201,7 @@ static struct sk_buff **fou_gro_receive(struct sk_buff **head,
 	if (!ops || !ops->callbacks.gro_receive)
 		goto out_unlock;
 
-	pp = ops->callbacks.gro_receive(head, skb);
+	pp = call_gro_receive(ops->callbacks.gro_receive, head, skb);
 
 out_unlock:
 	rcu_read_unlock();
@@ -360,7 +360,7 @@ static struct sk_buff **gue_gro_receive(struct sk_buff **head,
 	if (WARN_ON_ONCE(!ops || !ops->callbacks.gro_receive))
 		goto out_unlock;
 
-	pp = ops->callbacks.gro_receive(head, skb);
+	pp = call_gro_receive(ops->callbacks.gro_receive, head, skb);
 
 out_unlock:
 	rcu_read_unlock();
diff --git a/net/ipv4/gre_offload.c b/net/ipv4/gre_offload.c
index e603004c1af8..79ae0d7becbf 100644
--- a/net/ipv4/gre_offload.c
+++ b/net/ipv4/gre_offload.c
@@ -219,7 +219,7 @@ static struct sk_buff **gre_gro_receive(struct sk_buff **head,
 	/* Adjusted NAPI_GRO_CB(skb)->csum after skb_gro_pull()*/
 	skb_gro_postpull_rcsum(skb, greh, grehlen);
 
-	pp = ptype->callbacks.gro_receive(head, skb);
+	pp = call_gro_receive(ptype->callbacks.gro_receive, head, skb);
 
 out_unlock:
 	rcu_read_unlock();
diff --git a/net/ipv4/udp_offload.c b/net/ipv4/udp_offload.c
index 0e36e56dfd22..6396f1c80ae9 100644
--- a/net/ipv4/udp_offload.c
+++ b/net/ipv4/udp_offload.c
@@ -339,8 +339,8 @@ unflush:
 	skb_gro_pull(skb, sizeof(struct udphdr)); /* pull encapsulating udp header */
 	skb_gro_postpull_rcsum(skb, uh, sizeof(struct udphdr));
 	NAPI_GRO_CB(skb)->proto = uo_priv->offload->ipproto;
-	pp = uo_priv->offload->callbacks.gro_receive(head, skb,
-						     uo_priv->offload);
+	pp = call_gro_receive_udp(uo_priv->offload->callbacks.gro_receive,
+				  head, skb, uo_priv->offload);
 
 out_unlock:
 	rcu_read_unlock();
diff --git a/net/ipv6/ip6_offload.c b/net/ipv6/ip6_offload.c
index 82e9f3076028..efe6268b8bc3 100644
--- a/net/ipv6/ip6_offload.c
+++ b/net/ipv6/ip6_offload.c
@@ -247,7 +247,7 @@ static struct sk_buff **ipv6_gro_receive(struct sk_buff **head,
 
 	skb_gro_postpull_rcsum(skb, iph, nlen);
 
-	pp = ops->callbacks.gro_receive(head, skb);
+	pp = call_gro_receive(ops->callbacks.gro_receive, head, skb);
 
 out_unlock:
 	rcu_read_unlock();