Merge branch 'net-API-and-initial-implementation-for-nexthop-objects'

David Ahern says:

====================
net: API and initial implementation for nexthop objects

This set contains the API and initial implementation for nexthops as
standalone objects.

Patch 1 contains the UAPI and updates to selinux struct.

Patch 2 contains the barebones code for nexthop commands, rbtree
maintenance and notifications.

Patch 3 then adds support for IPv4 gateways along with handling of
netdev events.

Patch 4 adds support for IPv6 gateways.

Patch 5 has the implementation of the encap attributes.

Patch 6 adds support for nexthop groups.

At the end of this set, nexthop objects can be created and deleted and
userspace can monitor nexthop events, but ipv4 and ipv6 routes can not
use them yet. Once the nexthop struct is defined, follow on sets add it
to fib{6}_info and handle it within the respective code before routes
can be inserted using them.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

include/net/net_namespace.h

@@ -19,6 +19,7 @@
 #include <net/netns/packet.h>
 #include <net/netns/ipv4.h>
 #include <net/netns/ipv6.h>
+#include <net/netns/nexthop.h>
 #include <net/netns/ieee802154_6lowpan.h>
 #include <net/netns/sctp.h>
 #include <net/netns/dccp.h>
@@ -108,6 +109,7 @@ struct net {
 	struct netns_mib	mib;
 	struct netns_packet	packet;
 	struct netns_unix	unx;
+	struct netns_nexthop	nexthop;
 	struct netns_ipv4	ipv4;
 #if IS_ENABLED(CONFIG_IPV6)
 	struct netns_ipv6	ipv6;

include/net/netns/nexthop.h

+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * nexthops in net namespaces
+ */
+
+#ifndef __NETNS_NEXTHOP_H__
+#define __NETNS_NEXTHOP_H__
+
+#include <linux/rbtree.h>
+
+struct netns_nexthop {
+	struct rb_root		rb_root;	/* tree of nexthops by id */
+	struct hlist_head	*devhash;	/* nexthops by device */
+
+	unsigned int		seq;		/* protected by rtnl_mutex */
+	u32			last_id_allocated;
+};
+#endif

include/net/nexthop.h

+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Generic nexthop implementation
+ *
+ * Copyright (c) 2017-19 Cumulus Networks
+ * Copyright (c) 2017-19 David Ahern <dsa@cumulusnetworks.com>
+ */
+
+#ifndef __LINUX_NEXTHOP_H
+#define __LINUX_NEXTHOP_H
+
+#include <linux/netdevice.h>
+#include <linux/types.h>
+#include <net/ip_fib.h>
+#include <net/ip6_fib.h>
+#include <net/netlink.h>
+
+#define NEXTHOP_VALID_USER_FLAGS RTNH_F_ONLINK
+
+struct nexthop;
+
+struct nh_config {
+	u32		nh_id;
+
+	u8		nh_family;
+	u8		nh_protocol;
+	u8		nh_blackhole;
+	u32		nh_flags;
+
+	int		nh_ifindex;
+	struct net_device *dev;
+
+	union {
+		__be32		ipv4;
+		struct in6_addr	ipv6;
+	} gw;
+
+	struct nlattr	*nh_grp;
+	u16		nh_grp_type;
+
+	struct nlattr	*nh_encap;
+	u16		nh_encap_type;
+
+	u32		nlflags;
+	struct nl_info	nlinfo;
+};
+
+struct nh_info {
+	struct hlist_node	dev_hash;    /* entry on netns devhash */
+	struct nexthop		*nh_parent;
+
+	u8			family;
+	bool			reject_nh;
+
+	union {
+		struct fib_nh_common	fib_nhc;
+		struct fib_nh		fib_nh;
+		struct fib6_nh		fib6_nh;
+	};
+};
+
+struct nh_grp_entry {
+	struct nexthop	*nh;
+	u8		weight;
+	atomic_t	upper_bound;
+
+	struct list_head nh_list;
+	struct nexthop	*nh_parent;  /* nexthop of group with this entry */
+};
+
+struct nh_group {
+	u16			num_nh;
+	bool			mpath;
+	bool			has_v4;
+	struct nh_grp_entry	nh_entries[0];
+};
+
+struct nexthop {
+	struct rb_node		rb_node;    /* entry on netns rbtree */
+	struct list_head	grp_list;   /* nh group entries using this nh */
+	struct net		*net;
+
+	u32			id;
+
+	u8			protocol;   /* app managing this nh */
+	u8			nh_flags;
+	bool			is_group;
+
+	refcount_t		refcnt;
+	struct rcu_head		rcu;
+
+	union {
+		struct nh_info	__rcu *nh_info;
+		struct nh_group __rcu *nh_grp;
+	};
+};
+
+/* caller is holding rcu or rtnl; no reference taken to nexthop */
+struct nexthop *nexthop_find_by_id(struct net *net, u32 id);
+void nexthop_free_rcu(struct rcu_head *head);
+
+static inline bool nexthop_get(struct nexthop *nh)
+{
+	return refcount_inc_not_zero(&nh->refcnt);
+}
+
+static inline void nexthop_put(struct nexthop *nh)
+{
+	if (refcount_dec_and_test(&nh->refcnt))
+		call_rcu(&nh->rcu, nexthop_free_rcu);
+}
+
+static inline bool nexthop_is_multipath(const struct nexthop *nh)
+{
+	if (nh->is_group) {
+		struct nh_group *nh_grp;
+
+		nh_grp = rcu_dereference_rtnl(nh->nh_grp);
+		return nh_grp->mpath;
+	}
+	return false;
+}
+
+struct nexthop *nexthop_select_path(struct nexthop *nh, int hash);
+
+static inline unsigned int nexthop_num_path(const struct nexthop *nh)
+{
+	unsigned int rc = 1;
+
+	if (nexthop_is_multipath(nh)) {
+		struct nh_group *nh_grp;
+
+		nh_grp = rcu_dereference_rtnl(nh->nh_grp);
+		rc = nh_grp->num_nh;
+	} else {
+		const struct nh_info *nhi;
+
+		nhi = rcu_dereference_rtnl(nh->nh_info);
+		if (nhi->reject_nh)
+			rc = 0;
+	}
+
+	return rc;
+}
+
+static inline
+struct nexthop *nexthop_mpath_select(const struct nexthop *nh, int nhsel)
+{
+	const struct nh_group *nhg = rcu_dereference_rtnl(nh->nh_grp);
+
+	/* for_nexthops macros in fib_semantics.c grabs a pointer to
+	 * the nexthop before checking nhsel
+	 */
+	if (nhsel > nhg->num_nh)
+		return NULL;
+
+	return nhg->nh_entries[nhsel].nh;
+}
+
+static inline
+int nexthop_mpath_fill_node(struct sk_buff *skb, struct nexthop *nh)
+{
+	struct nh_group *nhg = rtnl_dereference(nh->nh_grp);
+	int i;
+
+	for (i = 0; i < nhg->num_nh; i++) {
+		struct nexthop *nhe = nhg->nh_entries[i].nh;
+		struct nh_info *nhi = rcu_dereference_rtnl(nhe->nh_info);
+		struct fib_nh_common *nhc = &nhi->fib_nhc;
+		int weight = nhg->nh_entries[i].weight;
+
+		if (fib_add_nexthop(skb, nhc, weight) < 0)
+			return -EMSGSIZE;
+	}
+
+	return 0;
+}
+
+/* called with rcu lock */
+static inline bool nexthop_is_blackhole(const struct nexthop *nh)
+{
+	const struct nh_info *nhi;
+
+	if (nexthop_is_multipath(nh)) {
+		if (nexthop_num_path(nh) > 1)
+			return false;
+		nh = nexthop_mpath_select(nh, 0);
+		if (!nh)
+			return false;
+	}
+
+	nhi = rcu_dereference_rtnl(nh->nh_info);
+	return nhi->reject_nh;
+}
+#endif

include/uapi/linux/nexthop.h

+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+#ifndef _UAPI_LINUX_NEXTHOP_H
+#define _UAPI_LINUX_NEXTHOP_H
+
+#include <linux/types.h>
+
+struct nhmsg {
+	unsigned char	nh_family;
+	unsigned char	nh_scope;     /* return only */
+	unsigned char	nh_protocol;  /* Routing protocol that installed nh */
+	unsigned char	resvd;
+	unsigned int	nh_flags;     /* RTNH_F flags */
+};
+
+/* entry in a nexthop group */
+struct nexthop_grp {
+	__u32	id;	  /* nexthop id - must exist */
+	__u8	weight;   /* weight of this nexthop */
+	__u8	resvd1;
+	__u16	resvd2;
+};
+
+enum {
+	NEXTHOP_GRP_TYPE_MPATH,  /* default type if not specified */
+	__NEXTHOP_GRP_TYPE_MAX,
+};
+
+#define NEXTHOP_GRP_TYPE_MAX (__NEXTHOP_GRP_TYPE_MAX - 1)
+
+enum {
+	NHA_UNSPEC,
+	NHA_ID,		/* u32; id for nexthop. id == 0 means auto-assign */
+
+	NHA_GROUP,	/* array of nexthop_grp */
+	NHA_GROUP_TYPE,	/* u16 one of NEXTHOP_GRP_TYPE */
+	/* if NHA_GROUP attribute is added, no other attributes can be set */
+
+	NHA_BLACKHOLE,	/* flag; nexthop used to blackhole packets */
+	/* if NHA_BLACKHOLE is added, OIF, GATEWAY, ENCAP can not be set */
+
+	NHA_OIF,	/* u32; nexthop device */
+	NHA_GATEWAY,	/* be32 (IPv4) or in6_addr (IPv6) gw address */
+	NHA_ENCAP_TYPE, /* u16; lwt encap type */
+	NHA_ENCAP,	/* lwt encap data */
+
+	/* NHA_OIF can be appended to dump request to return only
+	 * nexthops using given device
+	 */
+	NHA_GROUPS,	/* flag; only return nexthop groups in dump */
+	NHA_MASTER,	/* u32;  only return nexthops with given master dev */
+
+	__NHA_MAX,
+};
+
+#define NHA_MAX	(__NHA_MAX - 1)
+#endif

include/uapi/linux/rtnetlink.h

@@ -157,6 +157,13 @@ enum {
 	RTM_GETCHAIN,
 #define RTM_GETCHAIN RTM_GETCHAIN
 
+	RTM_NEWNEXTHOP = 104,
+#define RTM_NEWNEXTHOP	RTM_NEWNEXTHOP
+	RTM_DELNEXTHOP,
+#define RTM_DELNEXTHOP	RTM_DELNEXTHOP
+	RTM_GETNEXTHOP,
+#define RTM_GETNEXTHOP	RTM_GETNEXTHOP
+
 	__RTM_MAX,
 #define RTM_MAX		(((__RTM_MAX + 3) & ~3) - 1)
 };
@@ -342,6 +349,7 @@ enum rtattr_type_t {
 	RTA_IP_PROTO,
 	RTA_SPORT,
 	RTA_DPORT,
+	RTA_NH_ID,
 	__RTA_MAX
 };
 
@@ -704,6 +712,8 @@ enum rtnetlink_groups {
 #define RTNLGRP_IPV4_MROUTE_R	RTNLGRP_IPV4_MROUTE_R
 	RTNLGRP_IPV6_MROUTE_R,
 #define RTNLGRP_IPV6_MROUTE_R	RTNLGRP_IPV6_MROUTE_R
+	RTNLGRP_NEXTHOP,
+#define RTNLGRP_NEXTHOP		RTNLGRP_NEXTHOP
 	__RTNLGRP_MAX
 };
 #define RTNLGRP_MAX	(__RTNLGRP_MAX - 1)

net/ipv4/Makefile

@@ -14,7 +14,7 @@ obj-y     := route.o inetpeer.o protocol.o \
 	     udp_offload.o arp.o icmp.o devinet.o af_inet.o igmp.o \
 	     fib_frontend.o fib_semantics.o fib_trie.o fib_notifier.o \
 	     inet_fragment.o ping.o ip_tunnel_core.o gre_offload.o \
-	     metrics.o netlink.o
+	     metrics.o netlink.o nexthop.o
 
 obj-$(CONFIG_BPFILTER) += bpfilter/
 

net/ipv4/nexthop.c

+// SPDX-License-Identifier: GPL-2.0
+/* Generic nexthop implementation
+ *
+ * Copyright (c) 2017-19 Cumulus Networks
+ * Copyright (c) 2017-19 David Ahern <dsa@cumulusnetworks.com>
+ */
+
+#include <linux/nexthop.h>
+#include <linux/rtnetlink.h>
+#include <linux/slab.h>
+#include <net/arp.h>
+#include <net/ipv6_stubs.h>
+#include <net/lwtunnel.h>
+#include <net/ndisc.h>
+#include <net/nexthop.h>
+#include <net/route.h>
+#include <net/sock.h>
+
+static void remove_nexthop(struct net *net, struct nexthop *nh,
+			   struct nl_info *nlinfo);
+
+#define NH_DEV_HASHBITS  8
+#define NH_DEV_HASHSIZE (1U << NH_DEV_HASHBITS)
+
+static const struct nla_policy rtm_nh_policy[NHA_MAX + 1] = {
+	[NHA_UNSPEC]		= { .strict_start_type = NHA_UNSPEC + 1 },
+	[NHA_ID]		= { .type = NLA_U32 },
+	[NHA_GROUP]		= { .type = NLA_BINARY },
+	[NHA_GROUP_TYPE]	= { .type = NLA_U16 },
+	[NHA_BLACKHOLE]		= { .type = NLA_FLAG },
+	[NHA_OIF]		= { .type = NLA_U32 },
+	[NHA_GATEWAY]		= { .type = NLA_BINARY },
+	[NHA_ENCAP_TYPE]	= { .type = NLA_U16 },
+	[NHA_ENCAP]		= { .type = NLA_NESTED },
+	[NHA_GROUPS]		= { .type = NLA_FLAG },
+	[NHA_MASTER]		= { .type = NLA_U32 },
+};
+
+static unsigned int nh_dev_hashfn(unsigned int val)
+{
+	unsigned int mask = NH_DEV_HASHSIZE - 1;
+
+	return (val ^
+		(val >> NH_DEV_HASHBITS) ^
+		(val >> (NH_DEV_HASHBITS * 2))) & mask;
+}
+
+static void nexthop_devhash_add(struct net *net, struct nh_info *nhi)
+{
+	struct net_device *dev = nhi->fib_nhc.nhc_dev;
+	struct hlist_head *head;
+	unsigned int hash;
+
+	WARN_ON(!dev);
+
+	hash = nh_dev_hashfn(dev->ifindex);
+	head = &net->nexthop.devhash[hash];
+	hlist_add_head(&nhi->dev_hash, head);
+}
+
+static void nexthop_free_mpath(struct nexthop *nh)
+{
+	struct nh_group *nhg;
+	int i;
+
+	nhg = rcu_dereference_raw(nh->nh_grp);
+	for (i = 0; i < nhg->num_nh; ++i)
+		WARN_ON(nhg->nh_entries[i].nh);
+
+	kfree(nhg);
+}
+
+static void nexthop_free_single(struct nexthop *nh)
+{
+	struct nh_info *nhi;
+
+	nhi = rcu_dereference_raw(nh->nh_info);
+	switch (nhi->family) {
+	case AF_INET:
+		fib_nh_release(nh->net, &nhi->fib_nh);
+		break;
+	case AF_INET6:
+		ipv6_stub->fib6_nh_release(&nhi->fib6_nh);
+		break;
+	}
+	kfree(nhi);
+}
+
+void nexthop_free_rcu(struct rcu_head *head)
+{
+	struct nexthop *nh = container_of(head, struct nexthop, rcu);
+
+	if (nh->is_group)
+		nexthop_free_mpath(nh);
+	else
+		nexthop_free_single(nh);
+
+	kfree(nh);
+}
+EXPORT_SYMBOL_GPL(nexthop_free_rcu);
+
+static struct nexthop *nexthop_alloc(void)
+{
+	struct nexthop *nh;
+
+	nh = kzalloc(sizeof(struct nexthop), GFP_KERNEL);
+	if (nh) {
+		INIT_LIST_HEAD(&nh->grp_list);
+	}
+	return nh;
+}
+
+static struct nh_group *nexthop_grp_alloc(u16 num_nh)
+{
+	size_t sz = offsetof(struct nexthop, nh_grp)
+		    + sizeof(struct nh_group)
+		    + sizeof(struct nh_grp_entry) * num_nh;
+	struct nh_group *nhg;
+
+	nhg = kzalloc(sz, GFP_KERNEL);
+	if (nhg)
+		nhg->num_nh = num_nh;
+
+	return nhg;
+}
+
+static void nh_base_seq_inc(struct net *net)
+{
+	while (++net->nexthop.seq == 0)
+		;
+}
+
+/* no reference taken; rcu lock or rtnl must be held */
+struct nexthop *nexthop_find_by_id(struct net *net, u32 id)
+{
+	struct rb_node **pp, *parent = NULL, *next;
+
+	pp = &net->nexthop.rb_root.rb_node;
+	while (1) {
+		struct nexthop *nh;
+
+		next = rcu_dereference_raw(*pp);
+		if (!next)
+			break;
+		parent = next;
+
+		nh = rb_entry(parent, struct nexthop, rb_node);
+		if (id < nh->id)
+			pp = &next->rb_left;
+		else if (id > nh->id)
+			pp = &next->rb_right;
+		else
+			return nh;
+	}
+	return NULL;
+}
+EXPORT_SYMBOL_GPL(nexthop_find_by_id);
+
+/* used for auto id allocation; called with rtnl held */
+static u32 nh_find_unused_id(struct net *net)
+{
+	u32 id_start = net->nexthop.last_id_allocated;
+
+	while (1) {
+		net->nexthop.last_id_allocated++;
+		if (net->nexthop.last_id_allocated == id_start)
+			break;
+
+		if (!nexthop_find_by_id(net, net->nexthop.last_id_allocated))
+			return net->nexthop.last_id_allocated;
+	}
+	return 0;
+}
+
+static int nla_put_nh_group(struct sk_buff *skb, struct nh_group *nhg)
+{
+	struct nexthop_grp *p;
+	size_t len = nhg->num_nh * sizeof(*p);
+	struct nlattr *nla;
+	u16 group_type = 0;
+	int i;
+
+	if (nhg->mpath)
+		group_type = NEXTHOP_GRP_TYPE_MPATH;
+
+	if (nla_put_u16(skb, NHA_GROUP_TYPE, group_type))
+		goto nla_put_failure;
+
+	nla = nla_reserve(skb, NHA_GROUP, len);
+	if (!nla)
+		goto nla_put_failure;
+
+	p = nla_data(nla);
+	for (i = 0; i < nhg->num_nh; ++i) {
+		p->id = nhg->nh_entries[i].nh->id;
+		p->weight = nhg->nh_entries[i].weight - 1;
+		p += 1;
+	}
+
+	return 0;
+
+nla_put_failure:
+	return -EMSGSIZE;
+}
+
+static int nh_fill_node(struct sk_buff *skb, struct nexthop *nh,
+			int event, u32 portid, u32 seq, unsigned int nlflags)
+{
+	struct fib6_nh *fib6_nh;
+	struct fib_nh *fib_nh;
+	struct nlmsghdr *nlh;
+	struct nh_info *nhi;
+	struct nhmsg *nhm;
+
+	nlh = nlmsg_put(skb, portid, seq, event, sizeof(*nhm), nlflags);
+	if (!nlh)
+		return -EMSGSIZE;
+
+	nhm = nlmsg_data(nlh);
+	nhm->nh_family = AF_UNSPEC;
+	nhm->nh_flags = nh->nh_flags;
+	nhm->nh_protocol = nh->protocol;
+	nhm->nh_scope = 0;
+	nhm->resvd = 0;
+
+	if (nla_put_u32(skb, NHA_ID, nh->id))
+		goto nla_put_failure;
+
+	if (nh->is_group) {
+		struct nh_group *nhg = rtnl_dereference(nh->nh_grp);
+
+		if (nla_put_nh_group(skb, nhg))
+			goto nla_put_failure;
+		goto out;
+	}
+
+	nhi = rtnl_dereference(nh->nh_info);
+	nhm->nh_family = nhi->family;
+	if (nhi->reject_nh) {
+		if (nla_put_flag(skb, NHA_BLACKHOLE))
+			goto nla_put_failure;
+		goto out;
+	} else {
+		const struct net_device *dev;
+
+		dev = nhi->fib_nhc.nhc_dev;
+		if (dev && nla_put_u32(skb, NHA_OIF, dev->ifindex))
+			goto nla_put_failure;
+	}
+
+	nhm->nh_scope = nhi->fib_nhc.nhc_scope;
+	switch (nhi->family) {
+	case AF_INET:
+		fib_nh = &nhi->fib_nh;
+		if (fib_nh->fib_nh_gw_family &&
+		    nla_put_u32(skb, NHA_GATEWAY, fib_nh->fib_nh_gw4))
+			goto nla_put_failure;
+		break;
+
+	case AF_INET6:
+		fib6_nh = &nhi->fib6_nh;
+		if (fib6_nh->fib_nh_gw_family &&
+		    nla_put_in6_addr(skb, NHA_GATEWAY, &fib6_nh->fib_nh_gw6))
+			goto nla_put_failure;
+		break;
+	}
+
+	if (nhi->fib_nhc.nhc_lwtstate &&
+	    lwtunnel_fill_encap(skb, nhi->fib_nhc.nhc_lwtstate,
+				NHA_ENCAP, NHA_ENCAP_TYPE) < 0)
+		goto nla_put_failure;
+
+out:
+	nlmsg_end(skb, nlh);
+	return 0;
+
+nla_put_failure:
+	return -EMSGSIZE;
+}
+
+static size_t nh_nlmsg_size_grp(struct nexthop *nh)
+{
+	struct nh_group *nhg = rtnl_dereference(nh->nh_grp);
+	size_t sz = sizeof(struct nexthop_grp) * nhg->num_nh;
+
+	return nla_total_size(sz) +
+	       nla_total_size(2);  /* NHA_GROUP_TYPE */
+}
+
+static size_t nh_nlmsg_size_single(struct nexthop *nh)
+{
+	struct nh_info *nhi = rtnl_dereference(nh->nh_info);
+	size_t sz;
+
+	/* covers NHA_BLACKHOLE since NHA_OIF and BLACKHOLE
+	 * are mutually exclusive
+	 */
+	sz = nla_total_size(4);  /* NHA_OIF */
+
+	switch (nhi->family) {
+	case AF_INET:
+		if (nhi->fib_nh.fib_nh_gw_family)
+			sz += nla_total_size(4);  /* NHA_GATEWAY */
+		break;
+
+	case AF_INET6:
+		/* NHA_GATEWAY */
+		if (nhi->fib6_nh.fib_nh_gw_family)
+			sz += nla_total_size(sizeof(const struct in6_addr));
+		break;
+	}
+
+	if (nhi->fib_nhc.nhc_lwtstate) {
+		sz += lwtunnel_get_encap_size(nhi->fib_nhc.nhc_lwtstate);
+		sz += nla_total_size(2);  /* NHA_ENCAP_TYPE */
+	}
+
+	return sz;
+}
+
+static size_t nh_nlmsg_size(struct nexthop *nh)
+{
+	size_t sz = nla_total_size(4);    /* NHA_ID */
+
+	if (nh->is_group)
+		sz += nh_nlmsg_size_grp(nh);
+	else
+		sz += nh_nlmsg_size_single(nh);
+
+	return sz;
+}
+
+static void nexthop_notify(int event, struct nexthop *nh, struct nl_info *info)
+{
+	unsigned int nlflags = info->nlh ? info->nlh->nlmsg_flags : 0;
+	u32 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
+	struct sk_buff *skb;
+	int err = -ENOBUFS;
+
+	skb = nlmsg_new(nh_nlmsg_size(nh), gfp_any());
+	if (!skb)
+		goto errout;
+
+	err = nh_fill_node(skb, nh, event, info->portid, seq, nlflags);
+	if (err < 0) {
+		/* -EMSGSIZE implies BUG in nh_nlmsg_size() */
+		WARN_ON(err == -EMSGSIZE);
+		kfree_skb(skb);
+		goto errout;
+	}
+
+	rtnl_notify(skb, info->nl_net, info->portid, RTNLGRP_NEXTHOP,
+		    info->nlh, gfp_any());
+	return;
+errout:
+	if (err < 0)
+		rtnl_set_sk_err(info->nl_net, RTNLGRP_NEXTHOP, err);
+}
+
+static bool valid_group_nh(struct nexthop *nh, unsigned int npaths,
+			   struct netlink_ext_ack *extack)
+{
+	if (nh->is_group) {
+		struct nh_group *nhg = rtnl_dereference(nh->nh_grp);
+
+		/* nested multipath (group within a group) is not
+		 * supported
+		 */
+		if (nhg->mpath) {
+			NL_SET_ERR_MSG(extack,
+				       "Multipath group can not be a nexthop within a group");
+			return false;
+		}
+	} else {
+		struct nh_info *nhi = rtnl_dereference(nh->nh_info);
+
+		if (nhi->reject_nh && npaths > 1) {
+			NL_SET_ERR_MSG(extack,
+				       "Blackhole nexthop can not be used in a group with more than 1 path");
+			return false;
+		}
+	}
+
+	return true;
+}
+
+static int nh_check_attr_group(struct net *net, struct nlattr *tb[],
+			       struct netlink_ext_ack *extack)
+{
+	unsigned int len = nla_len(tb[NHA_GROUP]);
+	struct nexthop_grp *nhg;
+	unsigned int i, j;
+
+	if (len & (sizeof(struct nexthop_grp) - 1)) {
+		NL_SET_ERR_MSG(extack,
+			       "Invalid length for nexthop group attribute");
+		return -EINVAL;
+	}
+
+	/* convert len to number of nexthop ids */
+	len /= sizeof(*nhg);
+
+	nhg = nla_data(tb[NHA_GROUP]);
+	for (i = 0; i < len; ++i) {
+		if (nhg[i].resvd1 || nhg[i].resvd2) {
+			NL_SET_ERR_MSG(extack, "Reserved fields in nexthop_grp must be 0");
+			return -EINVAL;
+		}
+		if (nhg[i].weight > 254) {
+			NL_SET_ERR_MSG(extack, "Invalid value for weight");
+			return -EINVAL;
+		}
+		for (j = i + 1; j < len; ++j) {
+			if (nhg[i].id == nhg[j].id) {
+				NL_SET_ERR_MSG(extack, "Nexthop id can not be used twice in a group");
+				return -EINVAL;
+			}
+		}
+	}
+
+	nhg = nla_data(tb[NHA_GROUP]);
+	for (i = 0; i < len; ++i) {
+		struct nexthop *nh;
+
+		nh = nexthop_find_by_id(net, nhg[i].id);
+		if (!nh) {
+			NL_SET_ERR_MSG(extack, "Invalid nexthop id");
+			return -EINVAL;
+		}
+		if (!valid_group_nh(nh, len, extack))
+			return -EINVAL;
+	}
+	for (i = NHA_GROUP + 1; i < __NHA_MAX; ++i) {
+		if (!tb[i])
+			continue;
+
+		NL_SET_ERR_MSG(extack,
+			       "No other attributes can be set in nexthop groups");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static bool ipv6_good_nh(const struct fib6_nh *nh)
+{
+	int state = NUD_REACHABLE;
+	struct neighbour *n;
+
+	rcu_read_lock_bh();
+
+	n = __ipv6_neigh_lookup_noref_stub(nh->fib_nh_dev, &nh->fib_nh_gw6);
+	if (n)
+		state = n->nud_state;
+
+	rcu_read_unlock_bh();
+
+	return !!(state & NUD_VALID);
+}
+
+static bool ipv4_good_nh(const struct fib_nh *nh)
+{
+	int state = NUD_REACHABLE;
+	struct neighbour *n;
+
+	rcu_read_lock_bh();
+
+	n = __ipv4_neigh_lookup_noref(nh->fib_nh_dev,
+				      (__force u32)nh->fib_nh_gw4);
+	if (n)
+		state = n->nud_state;
+
+	rcu_read_unlock_bh();
+
+	return !!(state & NUD_VALID);
+}
+
+struct nexthop *nexthop_select_path(struct nexthop *nh, int hash)
+{
+	struct nexthop *rc = NULL;
+	struct nh_group *nhg;
+	int i;
+
+	if (!nh->is_group)
+		return nh;
+
+	nhg = rcu_dereference(nh->nh_grp);
+	for (i = 0; i < nhg->num_nh; ++i) {
+		struct nh_grp_entry *nhge = &nhg->nh_entries[i];
+		struct nh_info *nhi;
+
+		if (hash > atomic_read(&nhge->upper_bound))
+			continue;
+
+		/* nexthops always check if it is good and does
+		 * not rely on a sysctl for this behavior
+		 */
+		nhi = rcu_dereference(nhge->nh->nh_info);
+		switch (nhi->family) {
+		case AF_INET:
+			if (ipv4_good_nh(&nhi->fib_nh))
+				return nhge->nh;
+			break;
+		case AF_INET6:
+			if (ipv6_good_nh(&nhi->fib6_nh))
+				return nhge->nh;
+			break;
+		}
+
+		if (!rc)
+			rc = nhge->nh;
+	}
+
+	return rc;
+}
+EXPORT_SYMBOL_GPL(nexthop_select_path);
+
+static void nh_group_rebalance(struct nh_group *nhg)
+{
+	int total = 0;
+	int w = 0;
+	int i;
+
+	for (i = 0; i < nhg->num_nh; ++i)
+		total += nhg->nh_entries[i].weight;
+
+	for (i = 0; i < nhg->num_nh; ++i) {
+		struct nh_grp_entry *nhge = &nhg->nh_entries[i];
+		int upper_bound;
+
+		w += nhge->weight;
+		upper_bound = DIV_ROUND_CLOSEST_ULL((u64)w << 31, total) - 1;
+		atomic_set(&nhge->upper_bound, upper_bound);
+	}
+}
+
+static void remove_nh_grp_entry(struct nh_grp_entry *nhge,
+				struct nh_group *nhg,
+				struct nl_info *nlinfo)
+{
+	struct nexthop *nh = nhge->nh;
+	struct nh_grp_entry *nhges;
+	bool found = false;
+	int i;
+
+	WARN_ON(!nh);
+
+	nhges = nhg->nh_entries;
+	for (i = 0; i < nhg->num_nh; ++i) {
+		if (found) {
+			nhges[i-1].nh = nhges[i].nh;
+			nhges[i-1].weight = nhges[i].weight;
+			list_del(&nhges[i].nh_list);
+			list_add(&nhges[i-1].nh_list, &nhges[i-1].nh->grp_list);
+		} else if (nhg->nh_entries[i].nh == nh) {
+			found = true;
+		}
+	}
+
+	if (WARN_ON(!found))
+		return;
+
+	nhg->num_nh--;
+	nhg->nh_entries[nhg->num_nh].nh = NULL;
+
+	nh_group_rebalance(nhg);
+
+	nexthop_put(nh);
+
+	if (nlinfo)
+		nexthop_notify(RTM_NEWNEXTHOP, nhge->nh_parent, nlinfo);
+}
+
+static void remove_nexthop_from_groups(struct net *net, struct nexthop *nh,
+				       struct nl_info *nlinfo)
+{
+	struct nh_grp_entry *nhge, *tmp;
+
+	list_for_each_entry_safe(nhge, tmp, &nh->grp_list, nh_list) {
+		struct nh_group *nhg;
+
+		list_del(&nhge->nh_list);
+		nhg = rtnl_dereference(nhge->nh_parent->nh_grp);
+		remove_nh_grp_entry(nhge, nhg, nlinfo);
+
+		/* if this group has no more entries then remove it */
+		if (!nhg->num_nh)
+			remove_nexthop(net, nhge->nh_parent, nlinfo);
+	}
+}
+
+static void remove_nexthop_group(struct nexthop *nh, struct nl_info *nlinfo)
+{
+	struct nh_group *nhg = rcu_dereference_rtnl(nh->nh_grp);
+	int i, num_nh = nhg->num_nh;
+
+	for (i = 0; i < num_nh; ++i) {
+		struct nh_grp_entry *nhge = &nhg->nh_entries[i];
+
+		if (WARN_ON(!nhge->nh))
+			continue;
+
+		list_del(&nhge->nh_list);
+		nexthop_put(nhge->nh);
+		nhge->nh = NULL;
+		nhg->num_nh--;
+	}
+}
+
+static void __remove_nexthop(struct net *net, struct nexthop *nh,
+			     struct nl_info *nlinfo)
+{
+	if (nh->is_group) {
+		remove_nexthop_group(nh, nlinfo);
+	} else {
+		struct nh_info *nhi;
+
+		nhi = rtnl_dereference(nh->nh_info);
+		if (nhi->fib_nhc.nhc_dev)
+			hlist_del(&nhi->dev_hash);
+
+		remove_nexthop_from_groups(net, nh, nlinfo);
+	}
+}
+
+static void remove_nexthop(struct net *net, struct nexthop *nh,
+			   struct nl_info *nlinfo)
+{
+	/* remove from the tree */
+	rb_erase(&nh->rb_node, &net->nexthop.rb_root);
+
+	if (nlinfo)
+		nexthop_notify(RTM_DELNEXTHOP, nh, nlinfo);
+
+	__remove_nexthop(net, nh, nlinfo);
+	nh_base_seq_inc(net);
+
+	nexthop_put(nh);
+}
+
+static int replace_nexthop(struct net *net, struct nexthop *old,
+			   struct nexthop *new, struct netlink_ext_ack *extack)
+{
+	return -EEXIST;
+}
+
+/* called with rtnl_lock held */
+static int insert_nexthop(struct net *net, struct nexthop *new_nh,
+			  struct nh_config *cfg, struct netlink_ext_ack *extack)
+{
+	struct rb_node **pp, *parent = NULL, *next;
+	struct rb_root *root = &net->nexthop.rb_root;
+	bool replace = !!(cfg->nlflags & NLM_F_REPLACE);
+	bool create = !!(cfg->nlflags & NLM_F_CREATE);
+	u32 new_id = new_nh->id;
+	int rc = -EEXIST;
+
+	pp = &root->rb_node;
+	while (1) {
+		struct nexthop *nh;
+
+		next = rtnl_dereference(*pp);
+		if (!next)
+			break;
+
+		parent = next;
+
+		nh = rb_entry(parent, struct nexthop, rb_node);
+		if (new_id < nh->id) {
+			pp = &next->rb_left;
+		} else if (new_id > nh->id) {
+			pp = &next->rb_right;
+		} else if (replace) {
+			rc = replace_nexthop(net, nh, new_nh, extack);
+			if (!rc)
+				new_nh = nh; /* send notification with old nh */
+			goto out;
+		} else {
+			/* id already exists and not a replace */
+			goto out;
+		}
+	}
+
+	if (replace && !create) {
+		NL_SET_ERR_MSG(extack, "Replace specified without create and no entry exists");
+		rc = -ENOENT;
+		goto out;
+	}
+
+	rb_link_node_rcu(&new_nh->rb_node, parent, pp);
+	rb_insert_color(&new_nh->rb_node, root);
+	rc = 0;
+out:
+	if (!rc) {
+		nh_base_seq_inc(net);
+		nexthop_notify(RTM_NEWNEXTHOP, new_nh, &cfg->nlinfo);
+	}
+
+	return rc;
+}
+
+/* rtnl */
+/* remove all nexthops tied to a device being deleted */
+static void nexthop_flush_dev(struct net_device *dev)
+{
+	unsigned int hash = nh_dev_hashfn(dev->ifindex);
+	struct net *net = dev_net(dev);
+	struct hlist_head *head = &net->nexthop.devhash[hash];
+	struct hlist_node *n;
+	struct nh_info *nhi;
+
+	hlist_for_each_entry_safe(nhi, n, head, dev_hash) {
+		if (nhi->fib_nhc.nhc_dev != dev)
+			continue;
+
+		remove_nexthop(net, nhi->nh_parent, NULL);
+	}
+}
+
+/* rtnl; called when net namespace is deleted */
+static void flush_all_nexthops(struct net *net)
+{
+	struct rb_root *root = &net->nexthop.rb_root;
+	struct rb_node *node;
+	struct nexthop *nh;
+
+	while ((node = rb_first(root))) {
+		nh = rb_entry(node, struct nexthop, rb_node);
+		remove_nexthop(net, nh, NULL);
+		cond_resched();
+	}
+}
+
+static struct nexthop *nexthop_create_group(struct net *net,
+					    struct nh_config *cfg)
+{
+	struct nlattr *grps_attr = cfg->nh_grp;
+	struct nexthop_grp *entry = nla_data(grps_attr);
+	struct nh_group *nhg;
+	struct nexthop *nh;
+	int i;
+
+	nh = nexthop_alloc();
+	if (!nh)
+		return ERR_PTR(-ENOMEM);
+
+	nh->is_group = 1;
+
+	nhg = nexthop_grp_alloc(nla_len(grps_attr) / sizeof(*entry));
+	if (!nhg) {
+		kfree(nh);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	for (i = 0; i < nhg->num_nh; ++i) {
+		struct nexthop *nhe;
+		struct nh_info *nhi;
+
+		nhe = nexthop_find_by_id(net, entry[i].id);
+		if (!nexthop_get(nhe))
+			goto out_no_nh;
+
+		nhi = rtnl_dereference(nhe->nh_info);
+		if (nhi->family == AF_INET)
+			nhg->has_v4 = true;
+
+		nhg->nh_entries[i].nh = nhe;
+		nhg->nh_entries[i].weight = entry[i].weight + 1;
+		list_add(&nhg->nh_entries[i].nh_list, &nhe->grp_list);
+		nhg->nh_entries[i].nh_parent = nh;
+	}
+
+	if (cfg->nh_grp_type == NEXTHOP_GRP_TYPE_MPATH) {
+		nhg->mpath = 1;
+		nh_group_rebalance(nhg);
+	}
+
+	rcu_assign_pointer(nh->nh_grp, nhg);
+
+	return nh;
+
+out_no_nh:
+	for (; i >= 0; --i)
+		nexthop_put(nhg->nh_entries[i].nh);
+
+	kfree(nhg);
+	kfree(nh);
+
+	return ERR_PTR(-ENOENT);
+}
+
+static int nh_create_ipv4(struct net *net, struct nexthop *nh,
+			  struct nh_info *nhi, struct nh_config *cfg,
+			  struct netlink_ext_ack *extack)
+{
+	struct fib_nh *fib_nh = &nhi->fib_nh;
+	struct fib_config fib_cfg = {
+		.fc_oif   = cfg->nh_ifindex,
+		.fc_gw4   = cfg->gw.ipv4,
+		.fc_gw_family = cfg->gw.ipv4 ? AF_INET : 0,
+		.fc_flags = cfg->nh_flags,
+		.fc_encap = cfg->nh_encap,
+		.fc_encap_type = cfg->nh_encap_type,
+	};
+	u32 tb_id = l3mdev_fib_table(cfg->dev);
+	int err = -EINVAL;
+
+	err = fib_nh_init(net, fib_nh, &fib_cfg, 1, extack);
+	if (err) {
+		fib_nh_release(net, fib_nh);
+		goto out;
+	}
+
+	/* sets nh_dev if successful */
+	err = fib_check_nh(net, fib_nh, tb_id, 0, extack);
+	if (!err) {
+		nh->nh_flags = fib_nh->fib_nh_flags;
+		fib_info_update_nh_saddr(net, fib_nh, fib_nh->fib_nh_scope);
+	} else {
+		fib_nh_release(net, fib_nh);
+	}
+out:
+	return err;
+}
+
+static int nh_create_ipv6(struct net *net,  struct nexthop *nh,
+			  struct nh_info *nhi, struct nh_config *cfg,
+			  struct netlink_ext_ack *extack)
+{
+	struct fib6_nh *fib6_nh = &nhi->fib6_nh;
+	struct fib6_config fib6_cfg = {
+		.fc_table = l3mdev_fib_table(cfg->dev),
+		.fc_ifindex = cfg->nh_ifindex,
+		.fc_gateway = cfg->gw.ipv6,
+		.fc_flags = cfg->nh_flags,
+		.fc_encap = cfg->nh_encap,
+		.fc_encap_type = cfg->nh_encap_type,
+	};
+	int err = -EINVAL;
+
+	if (!ipv6_addr_any(&cfg->gw.ipv6))
+		fib6_cfg.fc_flags |= RTF_GATEWAY;
+
+	/* sets nh_dev if successful */
+	err = ipv6_stub->fib6_nh_init(net, fib6_nh, &fib6_cfg, GFP_KERNEL,
+				      extack);
+	if (err)
+		ipv6_stub->fib6_nh_release(fib6_nh);
+	else
+		nh->nh_flags = fib6_nh->fib_nh_flags;
+
+	return err;
+}
+
+static struct nexthop *nexthop_create(struct net *net, struct nh_config *cfg,
+				      struct netlink_ext_ack *extack)
+{
+	struct nh_info *nhi;
+	struct nexthop *nh;
+	int err = 0;
+
+	nh = nexthop_alloc();
+	if (!nh)
+		return ERR_PTR(-ENOMEM);
+
+	nhi = kzalloc(sizeof(*nhi), GFP_KERNEL);
+	if (!nhi) {
+		kfree(nh);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	nh->nh_flags = cfg->nh_flags;
+	nh->net = net;
+
+	nhi->nh_parent = nh;
+	nhi->family = cfg->nh_family;
+	nhi->fib_nhc.nhc_scope = RT_SCOPE_LINK;
+
+	if (cfg->nh_blackhole) {
+		nhi->reject_nh = 1;
+		cfg->nh_ifindex = net->loopback_dev->ifindex;
+	}
+
+	switch (cfg->nh_family) {
+	case AF_INET:
+		err = nh_create_ipv4(net, nh, nhi, cfg, extack);
+		break;
+	case AF_INET6:
+		err = nh_create_ipv6(net, nh, nhi, cfg, extack);
+		break;
+	}
+
+	if (err) {
+		kfree(nhi);
+		kfree(nh);
+		return ERR_PTR(err);
+	}
+
+	/* add the entry to the device based hash */
+	nexthop_devhash_add(net, nhi);
+
+	rcu_assign_pointer(nh->nh_info, nhi);
+
+	return nh;
+}
+
+/* called with rtnl lock held */
+static struct nexthop *nexthop_add(struct net *net, struct nh_config *cfg,
+				   struct netlink_ext_ack *extack)
+{
+	struct nexthop *nh;
+	int err;
+
+	if (cfg->nlflags & NLM_F_REPLACE && !cfg->nh_id) {
+		NL_SET_ERR_MSG(extack, "Replace requires nexthop id");
+		return ERR_PTR(-EINVAL);
+	}
+
+	if (!cfg->nh_id) {
+		cfg->nh_id = nh_find_unused_id(net);
+		if (!cfg->nh_id) {
+			NL_SET_ERR_MSG(extack, "No unused id");
+			return ERR_PTR(-EINVAL);
+		}
+	}
+
+	if (cfg->nh_grp)
+		nh = nexthop_create_group(net, cfg);
+	else
+		nh = nexthop_create(net, cfg, extack);
+
+	if (IS_ERR(nh))
+		return nh;
+
+	refcount_set(&nh->refcnt, 1);
+	nh->id = cfg->nh_id;
+	nh->protocol = cfg->nh_protocol;
+	nh->net = net;
+
+	err = insert_nexthop(net, nh, cfg, extack);
+	if (err) {
+		__remove_nexthop(net, nh, NULL);
+		nexthop_put(nh);
+		nh = ERR_PTR(err);
+	}
+
+	return nh;
+}
+
+static int rtm_to_nh_config(struct net *net, struct sk_buff *skb,
+			    struct nlmsghdr *nlh, struct nh_config *cfg,
+			    struct netlink_ext_ack *extack)
+{
+	struct nhmsg *nhm = nlmsg_data(nlh);
+	struct nlattr *tb[NHA_MAX + 1];
+	int err;
+
+	err = nlmsg_parse(nlh, sizeof(*nhm), tb, NHA_MAX, rtm_nh_policy,
+			  extack);
+	if (err < 0)
+		return err;
+
+	err = -EINVAL;
+	if (nhm->resvd || nhm->nh_scope) {
+		NL_SET_ERR_MSG(extack, "Invalid values in ancillary header");
+		goto out;
+	}
+	if (nhm->nh_flags & ~NEXTHOP_VALID_USER_FLAGS) {
+		NL_SET_ERR_MSG(extack, "Invalid nexthop flags in ancillary header");
+		goto out;
+	}
+
+	switch (nhm->nh_family) {
+	case AF_INET:
+	case AF_INET6:
+		break;
+	case AF_UNSPEC:
+		if (tb[NHA_GROUP])
+			break;
+		/* fallthrough */
+	default:
+		NL_SET_ERR_MSG(extack, "Invalid address family");
+		goto out;
+	}
+
+	if (tb[NHA_GROUPS] || tb[NHA_MASTER]) {
+		NL_SET_ERR_MSG(extack, "Invalid attributes in request");
+		goto out;
+	}
+
+	memset(cfg, 0, sizeof(*cfg));
+	cfg->nlflags = nlh->nlmsg_flags;
+	cfg->nlinfo.portid = NETLINK_CB(skb).portid;
+	cfg->nlinfo.nlh = nlh;
+	cfg->nlinfo.nl_net = net;
+
+	cfg->nh_family = nhm->nh_family;
+	cfg->nh_protocol = nhm->nh_protocol;
+	cfg->nh_flags = nhm->nh_flags;
+
+	if (tb[NHA_ID])
+		cfg->nh_id = nla_get_u32(tb[NHA_ID]);
+
+	if (tb[NHA_GROUP]) {
+		if (nhm->nh_family != AF_UNSPEC) {
+			NL_SET_ERR_MSG(extack, "Invalid family for group");
+			goto out;
+		}
+		cfg->nh_grp = tb[NHA_GROUP];
+
+		cfg->nh_grp_type = NEXTHOP_GRP_TYPE_MPATH;
+		if (tb[NHA_GROUP_TYPE])
+			cfg->nh_grp_type = nla_get_u16(tb[NHA_GROUP_TYPE]);
+
+		if (cfg->nh_grp_type > NEXTHOP_GRP_TYPE_MAX) {
+			NL_SET_ERR_MSG(extack, "Invalid group type");
+			goto out;
+		}
+		err = nh_check_attr_group(net, tb, extack);
+
+		/* no other attributes should be set */
+		goto out;
+	}
+
+	if (tb[NHA_BLACKHOLE]) {
+		if (tb[NHA_GATEWAY] || tb[NHA_OIF] ||
+		    tb[NHA_ENCAP]   || tb[NHA_ENCAP_TYPE]) {
+			NL_SET_ERR_MSG(extack, "Blackhole attribute can not be used with gateway or oif");
+			goto out;
+		}
+
+		cfg->nh_blackhole = 1;
+		err = 0;
+		goto out;
+	}
+
+	if (!tb[NHA_OIF]) {
+		NL_SET_ERR_MSG(extack, "Device attribute required for non-blackhole nexthops");
+		goto out;
+	}
+
+	cfg->nh_ifindex = nla_get_u32(tb[NHA_OIF]);
+	if (cfg->nh_ifindex)
+		cfg->dev = __dev_get_by_index(net, cfg->nh_ifindex);
+
+	if (!cfg->dev) {
+		NL_SET_ERR_MSG(extack, "Invalid device index");
+		goto out;
+	} else if (!(cfg->dev->flags & IFF_UP)) {
+		NL_SET_ERR_MSG(extack, "Nexthop device is not up");
+		err = -ENETDOWN;
+		goto out;
+	} else if (!netif_carrier_ok(cfg->dev)) {
+		NL_SET_ERR_MSG(extack, "Carrier for nexthop device is down");
+		err = -ENETDOWN;
+		goto out;
+	}
+
+	err = -EINVAL;
+	if (tb[NHA_GATEWAY]) {
+		struct nlattr *gwa = tb[NHA_GATEWAY];
+
+		switch (cfg->nh_family) {
+		case AF_INET:
+			if (nla_len(gwa) != sizeof(u32)) {
+				NL_SET_ERR_MSG(extack, "Invalid gateway");
+				goto out;
+			}
+			cfg->gw.ipv4 = nla_get_be32(gwa);
+			break;
+		case AF_INET6:
+			if (nla_len(gwa) != sizeof(struct in6_addr)) {
+				NL_SET_ERR_MSG(extack, "Invalid gateway");
+				goto out;
+			}
+			cfg->gw.ipv6 = nla_get_in6_addr(gwa);
+			break;
+		default:
+			NL_SET_ERR_MSG(extack,
+				       "Unknown address family for gateway");
+			goto out;
+		}
+	} else {
+		/* device only nexthop (no gateway) */
+		if (cfg->nh_flags & RTNH_F_ONLINK) {
+			NL_SET_ERR_MSG(extack,
+				       "ONLINK flag can not be set for nexthop without a gateway");
+			goto out;
+		}
+	}
+
+	if (tb[NHA_ENCAP]) {
+		cfg->nh_encap = tb[NHA_ENCAP];
+
+		if (!tb[NHA_ENCAP_TYPE]) {
+			NL_SET_ERR_MSG(extack, "LWT encapsulation type is missing");
+			goto out;
+		}
+
+		cfg->nh_encap_type = nla_get_u16(tb[NHA_ENCAP_TYPE]);
+		err = lwtunnel_valid_encap_type(cfg->nh_encap_type, extack);
+		if (err < 0)
+			goto out;
+
+	} else if (tb[NHA_ENCAP_TYPE]) {
+		NL_SET_ERR_MSG(extack, "LWT encapsulation attribute is missing");
+		goto out;
+	}
+
+
+	err = 0;
+out:
+	return err;
+}
+
+/* rtnl */
+static int rtm_new_nexthop(struct sk_buff *skb, struct nlmsghdr *nlh,
+			   struct netlink_ext_ack *extack)
+{
+	struct net *net = sock_net(skb->sk);
+	struct nh_config cfg;
+	struct nexthop *nh;
+	int err;
+
+	err = rtm_to_nh_config(net, skb, nlh, &cfg, extack);
+	if (!err) {
+		nh = nexthop_add(net, &cfg, extack);
+		if (IS_ERR(nh))
+			err = PTR_ERR(nh);
+	}
+
+	return err;
+}
+
+static int nh_valid_get_del_req(struct nlmsghdr *nlh, u32 *id,
+				struct netlink_ext_ack *extack)
+{
+	struct nhmsg *nhm = nlmsg_data(nlh);
+	struct nlattr *tb[NHA_MAX + 1];
+	int err, i;
+
+	err = nlmsg_parse(nlh, sizeof(*nhm), tb, NHA_MAX, rtm_nh_policy,
+			  extack);
+	if (err < 0)
+		return err;
+
+	err = -EINVAL;
+	for (i = 0; i < __NHA_MAX; ++i) {
+		if (!tb[i])
+			continue;
+
+		switch (i) {
+		case NHA_ID:
+			break;
+		default:
+			NL_SET_ERR_MSG_ATTR(extack, tb[i],
+					    "Unexpected attribute in request");
+			goto out;
+		}
+	}
+	if (nhm->nh_protocol || nhm->resvd || nhm->nh_scope || nhm->nh_flags) {
+		NL_SET_ERR_MSG(extack, "Invalid values in header");
+		goto out;
+	}
+
+	if (!tb[NHA_ID]) {
+		NL_SET_ERR_MSG(extack, "Nexthop id is missing");
+		goto out;
+	}
+
+	*id = nla_get_u32(tb[NHA_ID]);
+	if (!(*id))
+		NL_SET_ERR_MSG(extack, "Invalid nexthop id");
+	else
+		err = 0;
+out:
+	return err;
+}
+
+/* rtnl */
+static int rtm_del_nexthop(struct sk_buff *skb, struct nlmsghdr *nlh,
+			   struct netlink_ext_ack *extack)
+{
+	struct net *net = sock_net(skb->sk);
+	struct nl_info nlinfo = {
+		.nlh = nlh,
+		.nl_net = net,
+		.portid = NETLINK_CB(skb).portid,
+	};
+	struct nexthop *nh;
+	int err;
+	u32 id;
+
+	err = nh_valid_get_del_req(nlh, &id, extack);
+	if (err)
+		return err;
+
+	nh = nexthop_find_by_id(net, id);
+	if (!nh)
+		return -ENOENT;
+
+	remove_nexthop(net, nh, &nlinfo);
+
+	return 0;
+}
+
+/* rtnl */
+static int rtm_get_nexthop(struct sk_buff *in_skb, struct nlmsghdr *nlh,
+			   struct netlink_ext_ack *extack)
+{
+	struct net *net = sock_net(in_skb->sk);
+	struct sk_buff *skb = NULL;
+	struct nexthop *nh;
+	int err;
+	u32 id;
+
+	err = nh_valid_get_del_req(nlh, &id, extack);
+	if (err)
+		return err;
+
+	err = -ENOBUFS;
+	skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
+	if (!skb)
+		goto out;
+
+	err = -ENOENT;
+	nh = nexthop_find_by_id(net, id);
+	if (!nh)
+		goto errout_free;
+
+	err = nh_fill_node(skb, nh, RTM_NEWNEXTHOP, NETLINK_CB(in_skb).portid,
+			   nlh->nlmsg_seq, 0);
+	if (err < 0) {
+		WARN_ON(err == -EMSGSIZE);
+		goto errout_free;
+	}
+
+	err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
+out:
+	return err;
+errout_free:
+	kfree_skb(skb);
+	goto out;
+}
+
+static bool nh_dump_filtered(struct nexthop *nh, int dev_idx, int master_idx,
+			     bool group_filter, u8 family)
+{
+	const struct net_device *dev;
+	const struct nh_info *nhi;
+
+	if (group_filter && !nh->is_group)
+		return true;
+
+	if (!dev_idx && !master_idx && !family)
+		return false;
+
+	if (nh->is_group)
+		return true;
+
+	nhi = rtnl_dereference(nh->nh_info);
+	if (family && nhi->family != family)
+		return true;
+
+	dev = nhi->fib_nhc.nhc_dev;
+	if (dev_idx && (!dev || dev->ifindex != dev_idx))
+		return true;
+
+	if (master_idx) {
+		struct net_device *master;
+
+		if (!dev)
+			return true;
+
+		master = netdev_master_upper_dev_get((struct net_device *)dev);
+		if (!master || master->ifindex != master_idx)
+			return true;
+	}
+
+	return false;
+}
+
+static int nh_valid_dump_req(const struct nlmsghdr *nlh, int *dev_idx,
+			     int *master_idx, bool *group_filter,
+			     struct netlink_callback *cb)
+{
+	struct netlink_ext_ack *extack = cb->extack;
+	struct nlattr *tb[NHA_MAX + 1];
+	struct nhmsg *nhm;
+	int err, i;
+	u32 idx;
+
+	err = nlmsg_parse(nlh, sizeof(*nhm), tb, NHA_MAX, rtm_nh_policy,
+			  NULL);
+	if (err < 0)
+		return err;
+
+	for (i = 0; i <= NHA_MAX; ++i) {
+		if (!tb[i])
+			continue;
+
+		switch (i) {
+		case NHA_OIF:
+			idx = nla_get_u32(tb[i]);
+			if (idx > INT_MAX) {
+				NL_SET_ERR_MSG(extack, "Invalid device index");
+				return -EINVAL;
+			}
+			*dev_idx = idx;
+			break;
+		case NHA_MASTER:
+			idx = nla_get_u32(tb[i]);
+			if (idx > INT_MAX) {
+				NL_SET_ERR_MSG(extack, "Invalid master device index");
+				return -EINVAL;
+			}
+			*master_idx = idx;
+			break;
+		case NHA_GROUPS:
+			*group_filter = true;
+			break;
+		default:
+			NL_SET_ERR_MSG(extack, "Unsupported attribute in dump request");
+			return -EINVAL;
+		}
+	}
+
+	nhm = nlmsg_data(nlh);
+	if (nhm->nh_protocol || nhm->resvd || nhm->nh_scope || nhm->nh_flags) {
+		NL_SET_ERR_MSG(extack, "Invalid values in header for nexthop dump request");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/* rtnl */
+static int rtm_dump_nexthop(struct sk_buff *skb, struct netlink_callback *cb)
+{
+	struct nhmsg *nhm = nlmsg_data(cb->nlh);
+	int dev_filter_idx = 0, master_idx = 0;
+	struct net *net = sock_net(skb->sk);
+	struct rb_root *root = &net->nexthop.rb_root;
+	bool group_filter = false;
+	struct rb_node *node;
+	int idx = 0, s_idx;
+	int err;
+
+	err = nh_valid_dump_req(cb->nlh, &dev_filter_idx, &master_idx,
+				&group_filter, cb);
+	if (err < 0)
+		return err;
+
+	s_idx = cb->args[0];
+	for (node = rb_first(root); node; node = rb_next(node)) {
+		struct nexthop *nh;
+
+		if (idx < s_idx)
+			goto cont;
+
+		nh = rb_entry(node, struct nexthop, rb_node);
+		if (nh_dump_filtered(nh, dev_filter_idx, master_idx,
+				     group_filter, nhm->nh_family))
+			goto cont;
+
+		err = nh_fill_node(skb, nh, RTM_NEWNEXTHOP,
+				   NETLINK_CB(cb->skb).portid,
+				   cb->nlh->nlmsg_seq, NLM_F_MULTI);
+		if (err < 0) {
+			if (likely(skb->len))
+				goto out;
+
+			goto out_err;
+		}
+cont:
+		idx++;
+	}
+
+out:
+	err = skb->len;
+out_err:
+	cb->args[0] = idx;
+	cb->seq = net->nexthop.seq;
+	nl_dump_check_consistent(cb, nlmsg_hdr(skb));
+
+	return err;
+}
+
+static void nexthop_sync_mtu(struct net_device *dev, u32 orig_mtu)
+{
+	unsigned int hash = nh_dev_hashfn(dev->ifindex);
+	struct net *net = dev_net(dev);
+	struct hlist_head *head = &net->nexthop.devhash[hash];
+	struct hlist_node *n;
+	struct nh_info *nhi;
+
+	hlist_for_each_entry_safe(nhi, n, head, dev_hash) {
+		if (nhi->fib_nhc.nhc_dev == dev) {
+			if (nhi->family == AF_INET)
+				fib_nhc_update_mtu(&nhi->fib_nhc, dev->mtu,
+						   orig_mtu);
+		}
+	}
+}
+
+/* rtnl */
+static int nh_netdev_event(struct notifier_block *this,
+			   unsigned long event, void *ptr)
+{
+	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
+	struct netdev_notifier_info_ext *info_ext;
+
+	switch (event) {
+	case NETDEV_DOWN:
+	case NETDEV_UNREGISTER:
+		nexthop_flush_dev(dev);
+		break;
+	case NETDEV_CHANGE:
+		if (!(dev_get_flags(dev) & (IFF_RUNNING | IFF_LOWER_UP)))
+			nexthop_flush_dev(dev);
+		break;
+	case NETDEV_CHANGEMTU:
+		info_ext = ptr;
+		nexthop_sync_mtu(dev, info_ext->ext.mtu);
+		rt_cache_flush(dev_net(dev));
+		break;
+	}
+	return NOTIFY_DONE;
+}
+
+static struct notifier_block nh_netdev_notifier = {
+	.notifier_call = nh_netdev_event,
+};
+
+static void __net_exit nexthop_net_exit(struct net *net)
+{
+	rtnl_lock();
+	flush_all_nexthops(net);
+	rtnl_unlock();
+	kfree(net->nexthop.devhash);
+}
+
+static int __net_init nexthop_net_init(struct net *net)
+{
+	size_t sz = sizeof(struct hlist_head) * NH_DEV_HASHSIZE;
+
+	net->nexthop.rb_root = RB_ROOT;
+	net->nexthop.devhash = kzalloc(sz, GFP_KERNEL);
+	if (!net->nexthop.devhash)
+		return -ENOMEM;
+
+	return 0;
+}
+
+static struct pernet_operations nexthop_net_ops = {
+	.init = nexthop_net_init,
+	.exit = nexthop_net_exit,
+};
+
+static int __init nexthop_init(void)
+{
+	register_pernet_subsys(&nexthop_net_ops);
+
+	register_netdevice_notifier(&nh_netdev_notifier);
+
+	rtnl_register(PF_UNSPEC, RTM_NEWNEXTHOP, rtm_new_nexthop, NULL, 0);
+	rtnl_register(PF_UNSPEC, RTM_DELNEXTHOP, rtm_del_nexthop, NULL, 0);
+	rtnl_register(PF_UNSPEC, RTM_GETNEXTHOP, rtm_get_nexthop,
+		      rtm_dump_nexthop, 0);
+
+	rtnl_register(PF_INET, RTM_NEWNEXTHOP, rtm_new_nexthop, NULL, 0);
+	rtnl_register(PF_INET, RTM_GETNEXTHOP, NULL, rtm_dump_nexthop, 0);
+
+	rtnl_register(PF_INET6, RTM_NEWNEXTHOP, rtm_new_nexthop, NULL, 0);
+	rtnl_register(PF_INET6, RTM_GETNEXTHOP, NULL, rtm_dump_nexthop, 0);
+
+	return 0;
+}
+subsys_initcall(nexthop_init);

security/selinux/nlmsgtab.c

@@ -83,6 +83,9 @@ static const struct nlmsg_perm nlmsg_route_perms[] =
 	{ RTM_NEWCHAIN,		NETLINK_ROUTE_SOCKET__NLMSG_WRITE },
 	{ RTM_DELCHAIN,		NETLINK_ROUTE_SOCKET__NLMSG_WRITE },
 	{ RTM_GETCHAIN,		NETLINK_ROUTE_SOCKET__NLMSG_READ  },
+	{ RTM_NEWNEXTHOP,	NETLINK_ROUTE_SOCKET__NLMSG_WRITE },
+	{ RTM_DELNEXTHOP,	NETLINK_ROUTE_SOCKET__NLMSG_WRITE },
+	{ RTM_GETNEXTHOP,	NETLINK_ROUTE_SOCKET__NLMSG_READ  },
 };
 
 static const struct nlmsg_perm nlmsg_tcpdiag_perms[] =
@@ -166,7 +169,7 @@ int selinux_nlmsg_lookup(u16 sclass, u16 nlmsg_type, u32 *perm)
 		 * structures at the top of this file with the new mappings
 		 * before updating the BUILD_BUG_ON() macro!
 		 */
-		BUILD_BUG_ON(RTM_MAX != (RTM_NEWCHAIN + 3));
+		BUILD_BUG_ON(RTM_MAX != (RTM_NEWNEXTHOP + 3));
 		err = nlmsg_perm(nlmsg_type, perm, nlmsg_route_perms,
 				 sizeof(nlmsg_route_perms));
 		break;