ip: validate header length on virtual device xmit

KMSAN detected read beyond end of buffer in vti and sit devices when
passing truncated packets with PF_PACKET. The issue affects additional
ip tunnel devices.

Extend commit 76c0ddd8c3a6 ("ip6_tunnel: be careful when accessing the
inner header") and commit ccfec9e5cb2d ("ip_tunnel: be careful when
accessing the inner header").

Move the check to a separate helper and call at the start of each
ndo_start_xmit function in net/ipv4 and net/ipv6.

Minor changes:
- convert dev_kfree_skb to kfree_skb on error path,
  as dev_kfree_skb calls consume_skb which is not for error paths.
- use pskb_network_may_pull even though that is pedantic here,
  as the same as pskb_may_pull for devices without llheaders.
- do not cache ipv6 hdrs if used only once
  (unsafe across pskb_may_pull, was more relevant to earlier patch)

Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

9 files changed, 66 insertions, 32 deletions

diff --git a/include/net/ip_tunnels.h b/include/net/ip_tunnels.h
index cbcf35ce1b14..34f019650941 100644
--- a/include/net/ip_tunnels.h
+++ b/include/net/ip_tunnels.h
@@ -308,6 +308,26 @@ int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops *op,
 int ip_tunnel_encap_setup(struct ip_tunnel *t,
                           struct ip_tunnel_encap *ipencap);
 
+static inline bool pskb_inet_may_pull(struct sk_buff *skb)
+{
+        int nhlen;
+
+        switch (skb->protocol) {
+#if IS_ENABLED(CONFIG_IPV6)
+        case htons(ETH_P_IPV6):
+                nhlen = sizeof(struct ipv6hdr);
+                break;
+#endif
+        case htons(ETH_P_IP):
+                nhlen = sizeof(struct iphdr);
+                break;
+        default:
+                nhlen = 0;
+        }
+
+        return pskb_network_may_pull(skb, nhlen);
+}
+
 static inline int ip_encap_hlen(struct ip_tunnel_encap *e)
 {
         const struct ip_tunnel_encap_ops *ops;
diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index c7a7bd58a23c..d1d09f3e5f9e 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -676,6 +676,9 @@ static netdev_tx_t ipgre_xmit(struct sk_buff *skb,
         struct ip_tunnel *tunnel = netdev_priv(dev);
         const struct iphdr *tnl_params;
 
+        if (!pskb_inet_may_pull(skb))
+                goto free_skb;
+
         if (tunnel->collect_md) {
                 gre_fb_xmit(skb, dev, skb->protocol);
                 return NETDEV_TX_OK;
@@ -719,6 +722,9 @@ static netdev_tx_t erspan_xmit(struct sk_buff *skb,
         struct ip_tunnel *tunnel = netdev_priv(dev);
         bool truncate = false;
 
+        if (!pskb_inet_may_pull(skb))
+                goto free_skb;
+
         if (tunnel->collect_md) {
                 erspan_fb_xmit(skb, dev, skb->protocol);
                 return NETDEV_TX_OK;
@@ -762,6 +768,9 @@ static netdev_tx_t gre_tap_xmit(struct sk_buff *skb,
 {
         struct ip_tunnel *tunnel = netdev_priv(dev);
 
+        if (!pskb_inet_may_pull(skb))
+                goto free_skb;
+
         if (tunnel->collect_md) {
                 gre_fb_xmit(skb, dev, htons(ETH_P_TEB));
                 return NETDEV_TX_OK;
diff --git a/net/ipv4/ip_tunnel.c b/net/ipv4/ip_tunnel.c
index 284a22154b4e..c4f5602308ed 100644
--- a/net/ipv4/ip_tunnel.c
+++ b/net/ipv4/ip_tunnel.c
@@ -627,7 +627,6 @@ void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
                     const struct iphdr *tnl_params, u8 protocol)
 {
         struct ip_tunnel *tunnel = netdev_priv(dev);
-        unsigned int inner_nhdr_len = 0;
         const struct iphdr *inner_iph;
         struct flowi4 fl4;
         u8     tos, ttl;
@@ -637,14 +636,6 @@ void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
         __be32 dst;
         bool connected;
 
-        /* ensure we can access the inner net header, for several users below */
-        if (skb->protocol == htons(ETH_P_IP))
-                inner_nhdr_len = sizeof(struct iphdr);
-        else if (skb->protocol == htons(ETH_P_IPV6))
-                inner_nhdr_len = sizeof(struct ipv6hdr);
-        if (unlikely(!pskb_may_pull(skb, inner_nhdr_len)))
-                goto tx_error;
-
         inner_iph = (const struct iphdr *)skb_inner_network_header(skb);
         connected = (tunnel->parms.iph.daddr != 0);
 
diff --git a/net/ipv4/ip_vti.c b/net/ipv4/ip_vti.c
index de31b302d69c..d7b43e700023 100644
--- a/net/ipv4/ip_vti.c
+++ b/net/ipv4/ip_vti.c
@@ -241,6 +241,9 @@ static netdev_tx_t vti_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
         struct ip_tunnel *tunnel = netdev_priv(dev);
         struct flowi fl;
 
+        if (!pskb_inet_may_pull(skb))
+                goto tx_err;
+
         memset(&fl, 0, sizeof(fl));
 
         switch (skb->protocol) {
@@ -253,15 +256,18 @@ static netdev_tx_t vti_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
                 memset(IP6CB(skb), 0, sizeof(*IP6CB(skb)));
                 break;
         default:
-                dev->stats.tx_errors++;
-                dev_kfree_skb(skb);
-                return NETDEV_TX_OK;
+                goto tx_err;
         }
 
         /* override mark with tunnel output key */
         fl.flowi_mark = be32_to_cpu(tunnel->parms.o_key);
 
         return vti_xmit(skb, dev, &fl);
+
+tx_err:
+        dev->stats.tx_errors++;
+        kfree_skb(skb);
+        return NETDEV_TX_OK;
 }
 
 static int vti4_err(struct sk_buff *skb, u32 info)
diff --git a/net/ipv6/ip6_gre.c b/net/ipv6/ip6_gre.c
index 229e55c99021..09d0826742f8 100644
--- a/net/ipv6/ip6_gre.c
+++ b/net/ipv6/ip6_gre.c
@@ -881,6 +881,9 @@ static netdev_tx_t ip6gre_tunnel_xmit(struct sk_buff *skb,
         struct net_device_stats *stats = &t->dev->stats;
         int ret;
 
+        if (!pskb_inet_may_pull(skb))
+                goto tx_err;
+
         if (!ip6_tnl_xmit_ctl(t, &t->parms.laddr, &t->parms.raddr))
                 goto tx_err;
 
@@ -923,6 +926,9 @@ static netdev_tx_t ip6erspan_tunnel_xmit(struct sk_buff *skb,
         int nhoff;
         int thoff;
 
+        if (!pskb_inet_may_pull(skb))
+                goto tx_err;
+
         if (!ip6_tnl_xmit_ctl(t, &t->parms.laddr, &t->parms.raddr))
                 goto tx_err;
 
@@ -995,8 +1001,6 @@ static netdev_tx_t ip6erspan_tunnel_xmit(struct sk_buff *skb,
                         goto tx_err;
                 }
         } else {
-                struct ipv6hdr *ipv6h = ipv6_hdr(skb);
-
                 switch (skb->protocol) {
                 case htons(ETH_P_IP):
                         memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
@@ -1004,7 +1008,7 @@ static netdev_tx_t ip6erspan_tunnel_xmit(struct sk_buff *skb,
                                                  &dsfield, &encap_limit);
                         break;
                 case htons(ETH_P_IPV6):
-                        if (ipv6_addr_equal(&t->parms.raddr, &ipv6h->saddr))
+                        if (ipv6_addr_equal(&t->parms.raddr, &ipv6_hdr(skb)->saddr))
                                 goto tx_err;
                         if (prepare_ip6gre_xmit_ipv6(skb, dev, &fl6,
                                                      &dsfield, &encap_limit))
diff --git a/net/ipv6/ip6_tunnel.c b/net/ipv6/ip6_tunnel.c
index 99179b9c8384..0c6403cf8b52 100644
--- a/net/ipv6/ip6_tunnel.c
+++ b/net/ipv6/ip6_tunnel.c
@@ -1243,10 +1243,6 @@ ip4ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
         u8 tproto;
         int err;
 
-        /* ensure we can access the full inner ip header */
-        if (!pskb_may_pull(skb, sizeof(struct iphdr)))
-                return -1;
-
         iph = ip_hdr(skb);
         memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
 
@@ -1321,9 +1317,6 @@ ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
         u8 tproto;
         int err;
 
-        if (unlikely(!pskb_may_pull(skb, sizeof(*ipv6h))))
-                return -1;
-
         ipv6h = ipv6_hdr(skb);
         tproto = READ_ONCE(t->parms.proto);
         if ((tproto != IPPROTO_IPV6 && tproto != 0) ||
@@ -1405,6 +1398,9 @@ ip6_tnl_start_xmit(struct sk_buff *skb, struct net_device *dev)
         struct net_device_stats *stats = &t->dev->stats;
         int ret;
 
+        if (!pskb_inet_may_pull(skb))
+                goto tx_err;
+
         switch (skb->protocol) {
         case htons(ETH_P_IP):
                 ret = ip4ip6_tnl_xmit(skb, dev);
diff --git a/net/ipv6/ip6_vti.c b/net/ipv6/ip6_vti.c
index 706fe42e4928..8b6eefff2f7e 100644
--- a/net/ipv6/ip6_vti.c
+++ b/net/ipv6/ip6_vti.c
@@ -522,18 +522,18 @@ vti6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
 {
         struct ip6_tnl *t = netdev_priv(dev);
         struct net_device_stats *stats = &t->dev->stats;
-        struct ipv6hdr *ipv6h;
         struct flowi fl;
         int ret;
 
+        if (!pskb_inet_may_pull(skb))
+                goto tx_err;
+
         memset(&fl, 0, sizeof(fl));
 
         switch (skb->protocol) {
         case htons(ETH_P_IPV6):
-                ipv6h = ipv6_hdr(skb);
-
                 if ((t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) ||
-                    vti6_addr_conflict(t, ipv6h))
+                    vti6_addr_conflict(t, ipv6_hdr(skb)))
                         goto tx_err;
 
                 xfrm_decode_session(skb, &fl, AF_INET6);
diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index 8276f1224f16..30337b38274b 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -51,6 +51,7 @@
 #include <linux/export.h>
 #include <net/ip6_checksum.h>
 #include <linux/netconf.h>
+#include <net/ip_tunnels.h>
 
 #include <linux/nospec.h>
 
@@ -599,13 +600,12 @@ static netdev_tx_t reg_vif_xmit(struct sk_buff *skb,
                 .flowi6_iif     = skb->skb_iif ? : LOOPBACK_IFINDEX,
                 .flowi6_mark    = skb->mark,
         };
-        int err;
 
-        err = ip6mr_fib_lookup(net, &fl6, &mrt);
-        if (err < 0) {
-                kfree_skb(skb);
-                return err;
-        }
+        if (!pskb_inet_may_pull(skb))
+                goto tx_err;
+
+        if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
+                goto tx_err;
 
         read_lock(&mrt_lock);
         dev->stats.tx_bytes += skb->len;
@@ -614,6 +614,11 @@ static netdev_tx_t reg_vif_xmit(struct sk_buff *skb,
         read_unlock(&mrt_lock);
         kfree_skb(skb);
         return NETDEV_TX_OK;
+
+tx_err:
+        dev->stats.tx_errors++;
+        kfree_skb(skb);
+        return NETDEV_TX_OK;
 }
 
 static int reg_vif_get_iflink(const struct net_device *dev)
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index 51c9f75f34b9..1e03305c0549 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -1021,6 +1021,9 @@ tx_error:
 static netdev_tx_t sit_tunnel_xmit(struct sk_buff *skb,
                                    struct net_device *dev)
 {
+        if (!pskb_inet_may_pull(skb))
+                goto tx_err;
+
         switch (skb->protocol) {
         case htons(ETH_P_IP):
                 sit_tunnel_xmit__(skb, dev, IPPROTO_IPIP);