Merge tag 'ipsec-next-2024-03-06' of...

Merge tag 'ipsec-next-2024-03-06' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next

Steffen Klassert says:

====================
1) Introduce forwarding of ICMP Error messages. That is specified
   in RFC 4301 but was never implemented. From Antony Antony.

2) Use KMEM_CACHE instead of kmem_cache_create in xfrm6_tunnel_init()
   and xfrm_policy_init(). From Kunwu Chan.

3) Do not allocate stats in the xfrm interface driver, this can be done
   on net core now. From Breno Leitao.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

net/ipv6/xfrm6_tunnel.c

@@ -355,10 +355,7 @@ static int __init xfrm6_tunnel_init(void)
 {
 	int rv;
 
-	xfrm6_tunnel_spi_kmem = kmem_cache_create("xfrm6_tunnel_spi",
-						  sizeof(struct xfrm6_tunnel_spi),
-						  0, SLAB_HWCACHE_ALIGN,
-						  NULL);
+	xfrm6_tunnel_spi_kmem = KMEM_CACHE(xfrm6_tunnel_spi, SLAB_HWCACHE_ALIGN);
 	if (!xfrm6_tunnel_spi_kmem)
 		return -ENOMEM;
 	rv = register_pernet_subsys(&xfrm6_tunnel_net_ops);

net/xfrm/xfrm_interface_core.c

@@ -240,7 +240,6 @@ static void xfrmi_dev_free(struct net_device *dev)
 	struct xfrm_if *xi = netdev_priv(dev);
 
 	gro_cells_destroy(&xi->gro_cells);
-	free_percpu(dev->tstats);
 }
 
 static int xfrmi_create(struct net_device *dev)
@@ -749,6 +748,7 @@ static void xfrmi_dev_setup(struct net_device *dev)
 	dev->flags 		= IFF_NOARP;
 	dev->needs_free_netdev	= true;
 	dev->priv_destructor	= xfrmi_dev_free;
+	dev->pcpu_stat_type	= NETDEV_PCPU_STAT_TSTATS;
 	netif_keep_dst(dev);
 
 	eth_broadcast_addr(dev->broadcast);
@@ -765,15 +765,9 @@ static int xfrmi_dev_init(struct net_device *dev)
 	struct net_device *phydev = __dev_get_by_index(xi->net, xi->p.link);
 	int err;
 
-	dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
-	if (!dev->tstats)
-		return -ENOMEM;
-
 	err = gro_cells_init(&xi->gro_cells, dev);
-	if (err) {
-		free_percpu(dev->tstats);
+	if (err)
 		return err;
-	}
 
 	dev->features |= NETIF_F_LLTX;
 	dev->features |= XFRMI_FEATURES;

net/xfrm/xfrm_policy.c

@@ -29,6 +29,7 @@
 #include <linux/audit.h>
 #include <linux/rhashtable.h>
 #include <linux/if_tunnel.h>
+#include <linux/icmp.h>
 #include <net/dst.h>
 #include <net/flow.h>
 #include <net/inet_ecn.h>
@@ -3505,6 +3506,128 @@ static inline int secpath_has_nontransport(const struct sec_path *sp, int k, int
 	return 0;
 }
 
+static bool icmp_err_packet(const struct flowi *fl, unsigned short family)
+{
+	const struct flowi4 *fl4 = &fl->u.ip4;
+
+	if (family == AF_INET &&
+	    fl4->flowi4_proto == IPPROTO_ICMP &&
+	    (fl4->fl4_icmp_type == ICMP_DEST_UNREACH ||
+	     fl4->fl4_icmp_type == ICMP_TIME_EXCEEDED))
+		return true;
+
+#if IS_ENABLED(CONFIG_IPV6)
+	if (family == AF_INET6) {
+		const struct flowi6 *fl6 = &fl->u.ip6;
+
+		if (fl6->flowi6_proto == IPPROTO_ICMPV6 &&
+		    (fl6->fl6_icmp_type == ICMPV6_DEST_UNREACH ||
+		    fl6->fl6_icmp_type == ICMPV6_PKT_TOOBIG ||
+		    fl6->fl6_icmp_type == ICMPV6_TIME_EXCEED))
+			return true;
+	}
+#endif
+	return false;
+}
+
+static bool xfrm_icmp_flow_decode(struct sk_buff *skb, unsigned short family,
+				  const struct flowi *fl, struct flowi *fl1)
+{
+	bool ret = true;
+	struct sk_buff *newskb = skb_clone(skb, GFP_ATOMIC);
+	int hl = family == AF_INET ? (sizeof(struct iphdr) +  sizeof(struct icmphdr)) :
+		 (sizeof(struct ipv6hdr) + sizeof(struct icmp6hdr));
+
+	if (!newskb)
+		return true;
+
+	if (!pskb_pull(newskb, hl))
+		goto out;
+
+	skb_reset_network_header(newskb);
+
+	if (xfrm_decode_session_reverse(dev_net(skb->dev), newskb, fl1, family) < 0)
+		goto out;
+
+	fl1->flowi_oif = fl->flowi_oif;
+	fl1->flowi_mark = fl->flowi_mark;
+	fl1->flowi_tos = fl->flowi_tos;
+	nf_nat_decode_session(newskb, fl1, family);
+	ret = false;
+
+out:
+	consume_skb(newskb);
+	return ret;
+}
+
+static bool xfrm_selector_inner_icmp_match(struct sk_buff *skb, unsigned short family,
+					   const struct xfrm_selector *sel,
+					   const struct flowi *fl)
+{
+	bool ret = false;
+
+	if (icmp_err_packet(fl, family)) {
+		struct flowi fl1;
+
+		if (xfrm_icmp_flow_decode(skb, family, fl, &fl1))
+			return ret;
+
+		ret = xfrm_selector_match(sel, &fl1, family);
+	}
+
+	return ret;
+}
+
+static inline struct
+xfrm_policy *xfrm_in_fwd_icmp(struct sk_buff *skb,
+			      const struct flowi *fl, unsigned short family,
+			      u32 if_id)
+{
+	struct xfrm_policy *pol = NULL;
+
+	if (icmp_err_packet(fl, family)) {
+		struct flowi fl1;
+		struct net *net = dev_net(skb->dev);
+
+		if (xfrm_icmp_flow_decode(skb, family, fl, &fl1))
+			return pol;
+
+		pol = xfrm_policy_lookup(net, &fl1, family, XFRM_POLICY_FWD, if_id);
+	}
+
+	return pol;
+}
+
+static inline struct
+dst_entry *xfrm_out_fwd_icmp(struct sk_buff *skb, struct flowi *fl,
+			     unsigned short family, struct dst_entry *dst)
+{
+	if (icmp_err_packet(fl, family)) {
+		struct net *net = dev_net(skb->dev);
+		struct dst_entry *dst2;
+		struct flowi fl1;
+
+		if (xfrm_icmp_flow_decode(skb, family, fl, &fl1))
+			return dst;
+
+		dst_hold(dst);
+
+		dst2 = xfrm_lookup(net, dst, &fl1, NULL, (XFRM_LOOKUP_QUEUE | XFRM_LOOKUP_ICMP));
+
+		if (IS_ERR(dst2))
+			return dst;
+
+		if (dst2->xfrm) {
+			dst_release(dst);
+			dst = dst2;
+		} else {
+			dst_release(dst2);
+		}
+	}
+
+	return dst;
+}
+
 int __xfrm_policy_check(struct sock *sk, int dir, struct sk_buff *skb,
 			unsigned short family)
 {
@@ -3551,9 +3674,17 @@ int __xfrm_policy_check(struct sock *sk, int dir, struct sk_buff *skb,
 
 		for (i = sp->len - 1; i >= 0; i--) {
 			struct xfrm_state *x = sp->xvec[i];
+			int ret = 0;
+
 			if (!xfrm_selector_match(&x->sel, &fl, family)) {
-				XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEMISMATCH);
-				return 0;
+				ret = 1;
+				if (x->props.flags & XFRM_STATE_ICMP &&
+				    xfrm_selector_inner_icmp_match(skb, family, &x->sel, &fl))
+					ret = 0;
+				if (ret) {
+					XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEMISMATCH);
+					return 0;
+				}
 			}
 		}
 	}
@@ -3576,6 +3707,9 @@ int __xfrm_policy_check(struct sock *sk, int dir, struct sk_buff *skb,
 		return 0;
 	}
 
+	if (!pol && dir == XFRM_POLICY_FWD)
+		pol = xfrm_in_fwd_icmp(skb, &fl, family, if_id);
+
 	if (!pol) {
 		if (net->xfrm.policy_default[dir] == XFRM_USERPOLICY_BLOCK) {
 			XFRM_INC_STATS(net, LINUX_MIB_XFRMINNOPOLS);
@@ -3709,6 +3843,10 @@ int __xfrm_route_forward(struct sk_buff *skb, unsigned short family)
 		res = 0;
 		dst = NULL;
 	}
+
+	if (dst && !dst->xfrm)
+		dst = xfrm_out_fwd_icmp(skb, &fl, family, dst);
+
 	skb_dst_set(skb, dst);
 	return res;
 }
@@ -4027,10 +4165,7 @@ static int __net_init xfrm_policy_init(struct net *net)
 	int dir, err;
 
 	if (net_eq(net, &init_net)) {
-		xfrm_dst_cache = kmem_cache_create("xfrm_dst_cache",
-					   sizeof(struct xfrm_dst),
-					   0, SLAB_HWCACHE_ALIGN|SLAB_PANIC,
-					   NULL);
+		xfrm_dst_cache = KMEM_CACHE(xfrm_dst, SLAB_HWCACHE_ALIGN | SLAB_PANIC);
 		err = rhashtable_init(&xfrm_policy_inexact_table,
 				      &xfrm_pol_inexact_params);
 		BUG_ON(err);