xfrm: Add an IPsec hardware offloading API

This patch adds all the bits that are needed to do
IPsec hardware offload for IPsec states and ESP packets.
We add xfrmdev_ops to the net_device. xfrmdev_ops has
function pointers that are needed to manage the xfrm
states in the hardware and to do a per packet
offloading decision.

Joint work with:
Ilan Tayari <[email protected]>
Guy Shapiro <[email protected]>
Yossi Kuperman <[email protected]>

Signed-off-by: Guy Shapiro <[email protected]>
Signed-off-by: Ilan Tayari <[email protected]>
Signed-off-by: Yossi Kuperman <[email protected]>
Signed-off-by: Steffen Klassert <[email protected]>

Author: klassert

include/linux/netdevice.h

@@ -824,6 +824,16 @@ struct netdev_xdp {
 	};
 };
 
+#ifdef CONFIG_XFRM_OFFLOAD
+struct xfrmdev_ops {
+	int	(*xdo_dev_state_add) (struct xfrm_state *x);
+	void	(*xdo_dev_state_delete) (struct xfrm_state *x);
+	void	(*xdo_dev_state_free) (struct xfrm_state *x);
+	bool	(*xdo_dev_offload_ok) (struct sk_buff *skb,
+				       struct xfrm_state *x);
+};
+#endif
+
 /*
  * This structure defines the management hooks for network devices.
  * The following hooks can be defined; unless noted otherwise, they are
@@ -1697,6 +1707,10 @@ struct net_device {
 	const struct ndisc_ops *ndisc_ops;
 #endif
 
+#ifdef CONFIG_XFRM
+	const struct xfrmdev_ops *xfrmdev_ops;
+#endif
+
 	const struct header_ops *header_ops;
 
 	unsigned int		flags;

include/net/xfrm.h

@@ -120,6 +120,13 @@ struct xfrm_state_walk {
 	struct xfrm_address_filter *filter;
 };
 
+struct xfrm_state_offload {
+	struct net_device	*dev;
+	unsigned long		offload_handle;
+	unsigned int		num_exthdrs;
+	u8			flags;
+};
+
 /* Full description of state of transformer. */
 struct xfrm_state {
 	possible_net_t		xs_net;
@@ -207,6 +214,8 @@ struct xfrm_state {
 	struct xfrm_lifetime_cur curlft;
 	struct tasklet_hrtimer	mtimer;
 
+	struct xfrm_state_offload xso;
+
 	/* used to fix curlft->add_time when changing date */
 	long		saved_tmo;
 
@@ -1453,7 +1462,6 @@ struct xfrm6_tunnel {
 void xfrm_init(void);
 void xfrm4_init(void);
 int xfrm_state_init(struct net *net);
-void xfrm_dev_init(void);
 void xfrm_state_fini(struct net *net);
 void xfrm4_state_init(void);
 void xfrm4_protocol_init(void);
@@ -1559,6 +1567,7 @@ struct xfrmk_spdinfo {
 struct xfrm_state *xfrm_find_acq_byseq(struct net *net, u32 mark, u32 seq);
 int xfrm_state_delete(struct xfrm_state *x);
 int xfrm_state_flush(struct net *net, u8 proto, bool task_valid);
+int xfrm_dev_state_flush(struct net *net, struct net_device *dev, bool task_valid);
 void xfrm_sad_getinfo(struct net *net, struct xfrmk_sadinfo *si);
 void xfrm_spd_getinfo(struct net *net, struct xfrmk_spdinfo *si);
 u32 xfrm_replay_seqhi(struct xfrm_state *x, __be32 net_seq);
@@ -1641,6 +1650,11 @@ static inline int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb)
 }
 #endif
 
+struct dst_entry *__xfrm_dst_lookup(struct net *net, int tos, int oif,
+				    const xfrm_address_t *saddr,
+				    const xfrm_address_t *daddr,
+				    int family);
+
 struct xfrm_policy *xfrm_policy_alloc(struct net *net, gfp_t gfp);
 
 void xfrm_policy_walk_init(struct xfrm_policy_walk *walk, u8 type);
@@ -1846,6 +1860,55 @@ static inline struct xfrm_offload *xfrm_offload(struct sk_buff *skb)
 }
 #endif
 
+#ifdef CONFIG_XFRM_OFFLOAD
+void __net_init xfrm_dev_init(void);
+int xfrm_dev_state_add(struct net *net, struct xfrm_state *x,
+		       struct xfrm_user_offload *xuo);
+bool xfrm_dev_offload_ok(struct sk_buff *skb, struct xfrm_state *x);
+
+static inline void xfrm_dev_state_delete(struct xfrm_state *x)
+{
+	struct xfrm_state_offload *xso = &x->xso;
+
+	if (xso->dev)
+		xso->dev->xfrmdev_ops->xdo_dev_state_delete(x);
+}
+
+static inline void xfrm_dev_state_free(struct xfrm_state *x)
+{
+	struct xfrm_state_offload *xso = &x->xso;
+	 struct net_device *dev = xso->dev;
+
+	if (dev && dev->xfrmdev_ops) {
+		dev->xfrmdev_ops->xdo_dev_state_free(x);
+		xso->dev = NULL;
+		dev_put(dev);
+	}
+}
+#else
+static inline void __net_init xfrm_dev_init(void)
+{
+}
+
+static inline int xfrm_dev_state_add(struct net *net, struct xfrm_state *x, struct xfrm_user_offload *xuo)
+{
+	return 0;
+}
+
+static inline void xfrm_dev_state_delete(struct xfrm_state *x)
+{
+}
+
+static inline void xfrm_dev_state_free(struct xfrm_state *x)
+{
+}
+
+static inline bool xfrm_dev_offload_ok(struct sk_buff *skb, struct xfrm_state *x)
+{
+	return false;
+}
+#endif
+
 static inline int xfrm_mark_get(struct nlattr **attrs, struct xfrm_mark *m)
 {
 	if (attrs[XFRMA_MARK])

include/uapi/linux/xfrm.h

@@ -303,6 +303,7 @@ enum xfrm_attr_type_t {
 	XFRMA_PROTO,		/* __u8 */
 	XFRMA_ADDRESS_FILTER,	/* struct xfrm_address_filter */
 	XFRMA_PAD,
+	XFRMA_OFFLOAD_DEV,	/* struct xfrm_state_offload */
 	__XFRMA_MAX
 
 #define XFRMA_MAX (__XFRMA_MAX - 1)
@@ -494,6 +495,13 @@ struct xfrm_address_filter {
 	__u8				dplen;
 };
 
+struct xfrm_user_offload {
+	int				ifindex;
+	__u8				flags;
+};
+#define XFRM_OFFLOAD_IPV6	1
+#define XFRM_OFFLOAD_INBOUND	2
+
 #ifndef __KERNEL__
 /* backwards compatibility for userspace */
 #define XFRMGRP_ACQUIRE		1

net/ipv4/esp4.c

@@ -435,9 +435,6 @@ static int esp_output(struct xfrm_state *x, struct sk_buff *skb)
 	aead_request_set_crypt(req, sg, dsg, ivlen + clen, iv);
 	aead_request_set_ad(req, assoclen);
 
-	seqno = cpu_to_be64(XFRM_SKB_CB(skb)->seq.output.low +
-			    ((u64)XFRM_SKB_CB(skb)->seq.output.hi << 32));
-
 	memset(iv, 0, ivlen);
 	memcpy(iv + ivlen - min(ivlen, 8), (u8 *)&seqno + 8 - min(ivlen, 8),
 	       min(ivlen, 8));
@@ -470,6 +467,7 @@ static int esp_input_done2(struct sk_buff *skb, int err)
 {
 	const struct iphdr *iph;
 	struct xfrm_state *x = xfrm_input_state(skb);
+	struct xfrm_offload *xo = xfrm_offload(skb);
 	struct crypto_aead *aead = x->data;
 	int alen = crypto_aead_authsize(aead);
 	int hlen = sizeof(struct ip_esp_hdr) + crypto_aead_ivsize(aead);
@@ -478,7 +476,8 @@ static int esp_input_done2(struct sk_buff *skb, int err)
 	u8 nexthdr[2];
 	int padlen;
 
-	kfree(ESP_SKB_CB(skb)->tmp);
+	if (!xo || (xo && !(xo->flags & CRYPTO_DONE)))
+		kfree(ESP_SKB_CB(skb)->tmp);
 
 	if (unlikely(err))
 		goto out;

net/ipv4/xfrm4_output.c

@@ -29,7 +29,8 @@ static int xfrm4_tunnel_check_size(struct sk_buff *skb)
 		goto out;
 
 	mtu = dst_mtu(skb_dst(skb));
-	if (skb->len > mtu) {
+	if ((!skb_is_gso(skb) && skb->len > mtu) ||
+	    (skb_is_gso(skb) && skb_gso_network_seglen(skb) > ip_skb_dst_mtu(skb->sk, skb))) {
 		skb->protocol = htons(ETH_P_IP);
 
 		if (skb->sk)

net/ipv6/esp6.c

@@ -450,6 +450,7 @@ static int esp6_output(struct xfrm_state *x, struct sk_buff *skb)
 static int esp_input_done2(struct sk_buff *skb, int err)
 {
 	struct xfrm_state *x = xfrm_input_state(skb);
+	struct xfrm_offload *xo = xfrm_offload(skb);
 	struct crypto_aead *aead = x->data;
 	int alen = crypto_aead_authsize(aead);
 	int hlen = sizeof(struct ip_esp_hdr) + crypto_aead_ivsize(aead);
@@ -458,7 +459,8 @@ static int esp_input_done2(struct sk_buff *skb, int err)
 	int padlen;
 	u8 nexthdr[2];
 
-	kfree(ESP_SKB_CB(skb)->tmp);
+	if (!xo || (xo && !(xo->flags & CRYPTO_DONE)))
+		kfree(ESP_SKB_CB(skb)->tmp);
 
 	if (unlikely(err))
 		goto out;

net/ipv6/xfrm6_output.c

@@ -73,11 +73,16 @@ static int xfrm6_tunnel_check_size(struct sk_buff *skb)
 	int mtu, ret = 0;
 	struct dst_entry *dst = skb_dst(skb);
 
+	if (skb->ignore_df)
+		goto out;
+
 	mtu = dst_mtu(dst);
 	if (mtu < IPV6_MIN_MTU)
 		mtu = IPV6_MIN_MTU;
 
-	if (!skb->ignore_df && skb->len > mtu) {
+	if ((!skb_is_gso(skb) && skb->len > mtu) ||
+	    (skb_is_gso(skb) &&
+	     skb_gso_network_seglen(skb) > ip6_skb_dst_mtu(skb))) {
 		skb->dev = dst->dev;
 		skb->protocol = htons(ETH_P_IPV6);
 
@@ -89,7 +94,7 @@ static int xfrm6_tunnel_check_size(struct sk_buff *skb)
 			icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
 		ret = -EMSGSIZE;
 	}
-
+out:
 	return ret;
 }
 

net/xfrm/Makefile

@@ -4,7 +4,8 @@
 
 obj-$(CONFIG_XFRM) := xfrm_policy.o xfrm_state.o xfrm_hash.o \
 		      xfrm_input.o xfrm_output.o \
-		      xfrm_sysctl.o xfrm_replay.o xfrm_device.o
+		      xfrm_sysctl.o xfrm_replay.o
+obj-$(CONFIG_XFRM_OFFLOAD) += xfrm_device.o
 obj-$(CONFIG_XFRM_STATISTICS) += xfrm_proc.o
 obj-$(CONFIG_XFRM_ALGO) += xfrm_algo.o
 obj-$(CONFIG_XFRM_USER) += xfrm_user.o