[PATCH ipsec-next 5/5] xfrm: check for PMTU in tunnel mode for packet offload

[Leon Romanovsky <leon@kernel.org>]

From: Leon Romanovsky <leonro@nvidia.com>

In tunnel mode, for the packet offload, there were no PMTU signaling
to the upper level about need to fragment the packet. As a solution,
call to already existing xfrm[4|6]_tunnel_check_size() to perform that.

Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 include/net/xfrm.h     |  9 +++++++++
 net/xfrm/xfrm_device.c | 10 ++++++++--
 net/xfrm/xfrm_output.c |  6 ++++--
 3 files changed, 21 insertions(+), 4 deletions(-)

diff --git a/include/net/xfrm.h b/include/net/xfrm.h
index 58f8f7661ec4..519ab1209e4c 100644
--- a/include/net/xfrm.h
+++ b/include/net/xfrm.h
@@ -1782,6 +1782,15 @@ int xfrm_trans_queue(struct sk_buff *skb,
 				   struct sk_buff *));
 int xfrm_output_resume(struct sock *sk, struct sk_buff *skb, int err);
 int xfrm_output(struct sock *sk, struct sk_buff *skb);
+int xfrm4_tunnel_check_size(struct sk_buff *skb);
+#if IS_ENABLED(CONFIG_IPV6)
+int xfrm6_tunnel_check_size(struct sk_buff *skb);
+#else
+static inline int xfrm6_tunnel_check_size(struct sk_buff *skb)
+{
+	return -EMSGSIZE;
+}
+#endif
 
 #if IS_ENABLED(CONFIG_NET_PKTGEN)
 int pktgen_xfrm_outer_mode_output(struct xfrm_state *x, struct sk_buff *skb);
diff --git a/net/xfrm/xfrm_device.c b/net/xfrm/xfrm_device.c
index 056df0e69d73..9ad1f85b0a27 100644
--- a/net/xfrm/xfrm_device.c
+++ b/net/xfrm/xfrm_device.c
@@ -419,12 +419,12 @@ bool xfrm_dev_offload_ok(struct sk_buff *skb, struct xfrm_state *x)
 	struct dst_entry *dst = skb_dst(skb);
 	struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
 	struct net_device *dev = x->xso.dev;
+	bool check_tunnel_size;
 
 	if (x->xso.type == XFRM_DEV_OFFLOAD_UNSPECIFIED)
 		return false;
 
-	if (x->xso.type == XFRM_DEV_OFFLOAD_PACKET ||
-	    ((dev == xfrm_dst_path(dst)->dev) && !xdst->child->xfrm)) {
+	if ((dev == xfrm_dst_path(dst)->dev) && !xdst->child->xfrm) {
 		mtu = xfrm_state_mtu(x, xdst->child_mtu_cached);
 		if (skb->len <= mtu)
 			goto ok;
@@ -436,16 +436,22 @@ bool xfrm_dev_offload_ok(struct sk_buff *skb, struct xfrm_state *x)
 	return false;
 
 ok:
+	check_tunnel_size = x->xso.type == XFRM_DEV_OFFLOAD_PACKET &&
+			    x->props.mode == XFRM_MODE_TUNNEL;
 	switch (x->props.family) {
 	case AF_INET:
 		/* Check for IPv4 options */
 		if (ip_hdr(skb)->ihl != 5)
 			return false;
+		if (check_tunnel_size && xfrm4_tunnel_check_size(skb))
+			return false;
 		break;
 	case AF_INET6:
 		/* Check for IPv6 extensions */
 		if (ipv6_ext_hdr(ipv6_hdr(skb)->nexthdr))
 			return false;
+		if (check_tunnel_size && xfrm6_tunnel_check_size(skb))
+			return false;
 		break;
 	default:
 		break;
diff --git a/net/xfrm/xfrm_output.c b/net/xfrm/xfrm_output.c
index f7abd42c077d..34c8e266641c 100644
--- a/net/xfrm/xfrm_output.c
+++ b/net/xfrm/xfrm_output.c
@@ -786,7 +786,7 @@ int xfrm_output(struct sock *sk, struct sk_buff *skb)
 }
 EXPORT_SYMBOL_GPL(xfrm_output);
 
-static int xfrm4_tunnel_check_size(struct sk_buff *skb)
+int xfrm4_tunnel_check_size(struct sk_buff *skb)
 {
 	int mtu, ret = 0;
 
@@ -812,6 +812,7 @@ static int xfrm4_tunnel_check_size(struct sk_buff *skb)
 out:
 	return ret;
 }
+EXPORT_SYMBOL_GPL(xfrm4_tunnel_check_size);
 
 static int xfrm4_extract_output(struct xfrm_state *x, struct sk_buff *skb)
 {
@@ -834,7 +835,7 @@ static int xfrm4_extract_output(struct xfrm_state *x, struct sk_buff *skb)
 }
 
 #if IS_ENABLED(CONFIG_IPV6)
-static int xfrm6_tunnel_check_size(struct sk_buff *skb)
+int xfrm6_tunnel_check_size(struct sk_buff *skb)
 {
 	int mtu, ret = 0;
 	struct dst_entry *dst = skb_dst(skb);
@@ -864,6 +865,7 @@ static int xfrm6_tunnel_check_size(struct sk_buff *skb)
 out:
 	return ret;
 }
+EXPORT_SYMBOL_GPL(xfrm6_tunnel_check_size);
 #endif
 
 static int xfrm6_extract_output(struct xfrm_state *x, struct sk_buff *skb)
-- 
2.48.1