vxlan.c 81.6 KB
Newer Older
S
stephen hemminger 已提交
1
/*
R
Rami Rosen 已提交
2
 * VXLAN: Virtual eXtensible Local Area Network
S
stephen hemminger 已提交
3
 *
4
 * Copyright (c) 2012-2013 Vyatta Inc.
S
stephen hemminger 已提交
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */

#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

#include <linux/kernel.h>
#include <linux/types.h>
#include <linux/module.h>
#include <linux/errno.h>
#include <linux/slab.h>
#include <linux/skbuff.h>
#include <linux/rculist.h>
#include <linux/netdevice.h>
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/udp.h>
#include <linux/igmp.h>
#include <linux/etherdevice.h>
#include <linux/if_ether.h>
27
#include <linux/if_vlan.h>
S
stephen hemminger 已提交
28
#include <linux/hash.h>
Y
Yan Burman 已提交
29
#include <linux/ethtool.h>
D
David Stevens 已提交
30 31
#include <net/arp.h>
#include <net/ndisc.h>
S
stephen hemminger 已提交
32
#include <net/ip.h>
33
#include <net/ip_tunnels.h>
S
stephen hemminger 已提交
34 35
#include <net/icmp.h>
#include <net/udp.h>
T
Tom Herbert 已提交
36
#include <net/udp_tunnel.h>
S
stephen hemminger 已提交
37 38 39 40 41 42
#include <net/rtnetlink.h>
#include <net/route.h>
#include <net/dsfield.h>
#include <net/inet_ecn.h>
#include <net/net_namespace.h>
#include <net/netns/generic.h>
43
#include <net/vxlan.h>
44
#include <net/protocol.h>
45

C
Cong Wang 已提交
46 47 48 49
#if IS_ENABLED(CONFIG_IPV6)
#include <net/ipv6.h>
#include <net/addrconf.h>
#include <net/ip6_tunnel.h>
50
#include <net/ip6_checksum.h>
C
Cong Wang 已提交
51
#endif
T
Thomas Graf 已提交
52
#include <net/dst_metadata.h>
S
stephen hemminger 已提交
53 54 55

#define VXLAN_VERSION	"0.1"

56 57
#define PORT_HASH_BITS	8
#define PORT_HASH_SIZE  (1<<PORT_HASH_BITS)
S
stephen hemminger 已提交
58 59 60
#define FDB_AGE_DEFAULT 300 /* 5 min */
#define FDB_AGE_INTERVAL (10 * HZ)	/* rescan interval */

61 62
/* UDP port for VXLAN traffic.
 * The IANA assigned port is 4789, but the Linux default is 8472
S
Stephen Hemminger 已提交
63
 * for compatibility with early adopters.
64
 */
65 66
static unsigned short vxlan_port __read_mostly = 8472;
module_param_named(udp_port, vxlan_port, ushort, 0444);
S
stephen hemminger 已提交
67 68 69 70 71 72
MODULE_PARM_DESC(udp_port, "Destination UDP port");

static bool log_ecn_error = true;
module_param(log_ecn_error, bool, 0644);
MODULE_PARM_DESC(log_ecn_error, "Log packets received with corrupted ECN");

P
Pravin B Shelar 已提交
73
static int vxlan_net_id;
74
static struct rtnl_link_ops vxlan_link_ops;
75

76
static const u8 all_zeros_mac[ETH_ALEN + 2];
77

78
static int vxlan_sock_add(struct vxlan_dev *vxlan);
79

80 81 82 83
/* per-network namespace private data for this module */
struct vxlan_net {
	struct list_head  vxlan_list;
	struct hlist_head sock_list[PORT_HASH_SIZE];
84
	spinlock_t	  sock_lock;
85 86
};

S
stephen hemminger 已提交
87 88 89 90 91 92
/* Forwarding table entry */
struct vxlan_fdb {
	struct hlist_node hlist;	/* linked list of entries */
	struct rcu_head	  rcu;
	unsigned long	  updated;	/* jiffies */
	unsigned long	  used;
93
	struct list_head  remotes;
94
	u8		  eth_addr[ETH_ALEN];
S
stephen hemminger 已提交
95
	u16		  state;	/* see ndm_state */
96
	u8		  flags;	/* see ndm_flags */
S
stephen hemminger 已提交
97 98 99 100
};

/* salt for hash table */
static u32 vxlan_salt __read_mostly;
101
static struct workqueue_struct *vxlan_wq;
S
stephen hemminger 已提交
102

T
Thomas Graf 已提交
103 104
static inline bool vxlan_collect_metadata(struct vxlan_sock *vs)
{
105 106
	return vs->flags & VXLAN_F_COLLECT_METADATA ||
	       ip_tunnel_collect_metadata();
T
Thomas Graf 已提交
107 108
}

C
Cong Wang 已提交
109 110 111 112
#if IS_ENABLED(CONFIG_IPV6)
static inline
bool vxlan_addr_equal(const union vxlan_addr *a, const union vxlan_addr *b)
{
J
Jiri Benc 已提交
113 114 115 116 117 118
	if (a->sa.sa_family != b->sa.sa_family)
		return false;
	if (a->sa.sa_family == AF_INET6)
		return ipv6_addr_equal(&a->sin6.sin6_addr, &b->sin6.sin6_addr);
	else
		return a->sin.sin_addr.s_addr == b->sin.sin_addr.s_addr;
C
Cong Wang 已提交
119 120 121 122
}

static inline bool vxlan_addr_any(const union vxlan_addr *ipa)
{
J
Jiri Benc 已提交
123 124 125 126
	if (ipa->sa.sa_family == AF_INET6)
		return ipv6_addr_any(&ipa->sin6.sin6_addr);
	else
		return ipa->sin.sin_addr.s_addr == htonl(INADDR_ANY);
C
Cong Wang 已提交
127 128 129 130
}

static inline bool vxlan_addr_multicast(const union vxlan_addr *ipa)
{
J
Jiri Benc 已提交
131 132 133 134
	if (ipa->sa.sa_family == AF_INET6)
		return ipv6_addr_is_multicast(&ipa->sin6.sin6_addr);
	else
		return IN_MULTICAST(ntohl(ipa->sin.sin_addr.s_addr));
C
Cong Wang 已提交
135 136 137 138
}

static int vxlan_nla_get_addr(union vxlan_addr *ip, struct nlattr *nla)
{
J
Jiri Benc 已提交
139
	if (nla_len(nla) >= sizeof(struct in6_addr)) {
140
		ip->sin6.sin6_addr = nla_get_in6_addr(nla);
J
Jiri Benc 已提交
141 142 143
		ip->sa.sa_family = AF_INET6;
		return 0;
	} else if (nla_len(nla) >= sizeof(__be32)) {
144
		ip->sin.sin_addr.s_addr = nla_get_in_addr(nla);
J
Jiri Benc 已提交
145 146 147 148 149
		ip->sa.sa_family = AF_INET;
		return 0;
	} else {
		return -EAFNOSUPPORT;
	}
C
Cong Wang 已提交
150 151 152
}

static int vxlan_nla_put_addr(struct sk_buff *skb, int attr,
J
Jiri Benc 已提交
153
			      const union vxlan_addr *ip)
C
Cong Wang 已提交
154
{
J
Jiri Benc 已提交
155
	if (ip->sa.sa_family == AF_INET6)
156
		return nla_put_in6_addr(skb, attr, &ip->sin6.sin6_addr);
J
Jiri Benc 已提交
157
	else
158
		return nla_put_in_addr(skb, attr, ip->sin.sin_addr.s_addr);
C
Cong Wang 已提交
159 160 161 162 163 164 165
}

#else /* !CONFIG_IPV6 */

static inline
bool vxlan_addr_equal(const union vxlan_addr *a, const union vxlan_addr *b)
{
J
Jiri Benc 已提交
166
	return a->sin.sin_addr.s_addr == b->sin.sin_addr.s_addr;
C
Cong Wang 已提交
167 168 169 170
}

static inline bool vxlan_addr_any(const union vxlan_addr *ipa)
{
J
Jiri Benc 已提交
171
	return ipa->sin.sin_addr.s_addr == htonl(INADDR_ANY);
C
Cong Wang 已提交
172 173 174 175
}

static inline bool vxlan_addr_multicast(const union vxlan_addr *ipa)
{
J
Jiri Benc 已提交
176
	return IN_MULTICAST(ntohl(ipa->sin.sin_addr.s_addr));
C
Cong Wang 已提交
177 178 179 180
}

static int vxlan_nla_get_addr(union vxlan_addr *ip, struct nlattr *nla)
{
J
Jiri Benc 已提交
181 182 183
	if (nla_len(nla) >= sizeof(struct in6_addr)) {
		return -EAFNOSUPPORT;
	} else if (nla_len(nla) >= sizeof(__be32)) {
184
		ip->sin.sin_addr.s_addr = nla_get_in_addr(nla);
J
Jiri Benc 已提交
185 186 187 188 189
		ip->sa.sa_family = AF_INET;
		return 0;
	} else {
		return -EAFNOSUPPORT;
	}
C
Cong Wang 已提交
190 191 192
}

static int vxlan_nla_put_addr(struct sk_buff *skb, int attr,
J
Jiri Benc 已提交
193
			      const union vxlan_addr *ip)
C
Cong Wang 已提交
194
{
195
	return nla_put_in_addr(skb, attr, ip->sin.sin_addr.s_addr);
C
Cong Wang 已提交
196 197 198
}
#endif

199
/* Virtual Network hash table head */
200
static inline struct hlist_head *vni_head(struct vxlan_sock *vs, __be32 vni)
201
{
202
	return &vs->vni_list[hash_32((__force u32)vni, VNI_HASH_BITS)];
203 204 205 206
}

/* Socket hash table head */
static inline struct hlist_head *vs_head(struct net *net, __be16 port)
S
stephen hemminger 已提交
207 208 209
{
	struct vxlan_net *vn = net_generic(net, vxlan_net_id);

210 211 212
	return &vn->sock_list[hash_32(ntohs(port), PORT_HASH_BITS)];
}

213 214 215
/* First remote destination for a forwarding entry.
 * Guaranteed to be non-NULL because remotes are never deleted.
 */
216
static inline struct vxlan_rdst *first_remote_rcu(struct vxlan_fdb *fdb)
217
{
218 219 220 221 222 223
	return list_entry_rcu(fdb->remotes.next, struct vxlan_rdst, list);
}

static inline struct vxlan_rdst *first_remote_rtnl(struct vxlan_fdb *fdb)
{
	return list_first_entry(&fdb->remotes, struct vxlan_rdst, list);
224 225
}

226 227 228 229 230
/* Find VXLAN socket based on network namespace, address family and UDP port
 * and enabled unshareable flags.
 */
static struct vxlan_sock *vxlan_find_sock(struct net *net, sa_family_t family,
					  __be16 port, u32 flags)
231 232
{
	struct vxlan_sock *vs;
233 234

	flags &= VXLAN_F_RCV_FLAGS;
235 236

	hlist_for_each_entry_rcu(vs, vs_head(net, port), hlist) {
237
		if (inet_sk(vs->sock->sk)->inet_sport == port &&
238
		    vxlan_get_sk_family(vs) == family &&
239
		    vs->flags == flags)
240 241 242
			return vs;
	}
	return NULL;
S
stephen hemminger 已提交
243 244
}

245
static struct vxlan_dev *vxlan_vs_find_vni(struct vxlan_sock *vs, __be32 vni)
S
stephen hemminger 已提交
246 247 248
{
	struct vxlan_dev *vxlan;

249 250 251 252
	/* For flow based devices, map all packets to VNI 0 */
	if (vs->flags & VXLAN_F_COLLECT_METADATA)
		vni = 0;

253 254
	hlist_for_each_entry_rcu(vxlan, vni_head(vs, vni), hlist) {
		if (vxlan->default_dst.remote_vni == vni)
S
stephen hemminger 已提交
255 256 257 258 259 260
			return vxlan;
	}

	return NULL;
}

P
Pravin B Shelar 已提交
261
/* Look up VNI in a per net namespace table */
262
static struct vxlan_dev *vxlan_find_vni(struct net *net, __be32 vni,
263 264
					sa_family_t family, __be16 port,
					u32 flags)
P
Pravin B Shelar 已提交
265 266 267
{
	struct vxlan_sock *vs;

268
	vs = vxlan_find_sock(net, family, port, flags);
P
Pravin B Shelar 已提交
269 270 271
	if (!vs)
		return NULL;

272
	return vxlan_vs_find_vni(vs, vni);
P
Pravin B Shelar 已提交
273 274
}

S
stephen hemminger 已提交
275 276
/* Fill in neighbour message in skbuff. */
static int vxlan_fdb_info(struct sk_buff *skb, struct vxlan_dev *vxlan,
S
Stephen Hemminger 已提交
277 278 279
			  const struct vxlan_fdb *fdb,
			  u32 portid, u32 seq, int type, unsigned int flags,
			  const struct vxlan_rdst *rdst)
S
stephen hemminger 已提交
280 281 282 283 284
{
	unsigned long now = jiffies;
	struct nda_cacheinfo ci;
	struct nlmsghdr *nlh;
	struct ndmsg *ndm;
D
David Stevens 已提交
285
	bool send_ip, send_eth;
S
stephen hemminger 已提交
286 287 288 289 290 291 292

	nlh = nlmsg_put(skb, portid, seq, type, sizeof(*ndm), flags);
	if (nlh == NULL)
		return -EMSGSIZE;

	ndm = nlmsg_data(nlh);
	memset(ndm, 0, sizeof(*ndm));
D
David Stevens 已提交
293 294 295 296 297

	send_eth = send_ip = true;

	if (type == RTM_GETNEIGH) {
		ndm->ndm_family	= AF_INET;
C
Cong Wang 已提交
298
		send_ip = !vxlan_addr_any(&rdst->remote_ip);
D
David Stevens 已提交
299 300 301
		send_eth = !is_zero_ether_addr(fdb->eth_addr);
	} else
		ndm->ndm_family	= AF_BRIDGE;
S
stephen hemminger 已提交
302 303
	ndm->ndm_state = fdb->state;
	ndm->ndm_ifindex = vxlan->dev->ifindex;
304
	ndm->ndm_flags = fdb->flags;
305
	ndm->ndm_type = RTN_UNICAST;
S
stephen hemminger 已提交
306

307
	if (!net_eq(dev_net(vxlan->dev), vxlan->net) &&
308
	    nla_put_s32(skb, NDA_LINK_NETNSID,
309
			peernet2id_alloc(dev_net(vxlan->dev), vxlan->net)))
310 311
		goto nla_put_failure;

D
David Stevens 已提交
312
	if (send_eth && nla_put(skb, NDA_LLADDR, ETH_ALEN, &fdb->eth_addr))
S
stephen hemminger 已提交
313 314
		goto nla_put_failure;

C
Cong Wang 已提交
315
	if (send_ip && vxlan_nla_put_addr(skb, NDA_DST, &rdst->remote_ip))
316 317
		goto nla_put_failure;

318
	if (rdst->remote_port && rdst->remote_port != vxlan->cfg.dst_port &&
319 320
	    nla_put_be16(skb, NDA_PORT, rdst->remote_port))
		goto nla_put_failure;
321
	if (rdst->remote_vni != vxlan->default_dst.remote_vni &&
322
	    nla_put_u32(skb, NDA_VNI, be32_to_cpu(rdst->remote_vni)))
323 324 325
		goto nla_put_failure;
	if (rdst->remote_ifindex &&
	    nla_put_u32(skb, NDA_IFINDEX, rdst->remote_ifindex))
S
stephen hemminger 已提交
326 327 328 329 330 331 332 333 334 335
		goto nla_put_failure;

	ci.ndm_used	 = jiffies_to_clock_t(now - fdb->used);
	ci.ndm_confirmed = 0;
	ci.ndm_updated	 = jiffies_to_clock_t(now - fdb->updated);
	ci.ndm_refcnt	 = 0;

	if (nla_put(skb, NDA_CACHEINFO, sizeof(ci), &ci))
		goto nla_put_failure;

336 337
	nlmsg_end(skb, nlh);
	return 0;
S
stephen hemminger 已提交
338 339 340 341 342 343 344 345 346 347

nla_put_failure:
	nlmsg_cancel(skb, nlh);
	return -EMSGSIZE;
}

static inline size_t vxlan_nlmsg_size(void)
{
	return NLMSG_ALIGN(sizeof(struct ndmsg))
		+ nla_total_size(ETH_ALEN) /* NDA_LLADDR */
C
Cong Wang 已提交
348
		+ nla_total_size(sizeof(struct in6_addr)) /* NDA_DST */
349
		+ nla_total_size(sizeof(__be16)) /* NDA_PORT */
350 351
		+ nla_total_size(sizeof(__be32)) /* NDA_VNI */
		+ nla_total_size(sizeof(__u32)) /* NDA_IFINDEX */
352
		+ nla_total_size(sizeof(__s32)) /* NDA_LINK_NETNSID */
S
stephen hemminger 已提交
353 354 355
		+ nla_total_size(sizeof(struct nda_cacheinfo));
}

356 357
static void vxlan_fdb_notify(struct vxlan_dev *vxlan, struct vxlan_fdb *fdb,
			     struct vxlan_rdst *rd, int type)
S
stephen hemminger 已提交
358 359 360 361 362 363 364 365 366
{
	struct net *net = dev_net(vxlan->dev);
	struct sk_buff *skb;
	int err = -ENOBUFS;

	skb = nlmsg_new(vxlan_nlmsg_size(), GFP_ATOMIC);
	if (skb == NULL)
		goto errout;

367
	err = vxlan_fdb_info(skb, vxlan, fdb, 0, 0, type, 0, rd);
S
stephen hemminger 已提交
368 369 370 371 372 373 374 375 376 377 378 379 380 381
	if (err < 0) {
		/* -EMSGSIZE implies BUG in vxlan_nlmsg_size() */
		WARN_ON(err == -EMSGSIZE);
		kfree_skb(skb);
		goto errout;
	}

	rtnl_notify(skb, net, 0, RTNLGRP_NEIGH, NULL, GFP_ATOMIC);
	return;
errout:
	if (err < 0)
		rtnl_set_sk_err(net, RTNLGRP_NEIGH, err);
}

C
Cong Wang 已提交
382
static void vxlan_ip_miss(struct net_device *dev, union vxlan_addr *ipa)
D
David Stevens 已提交
383 384
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
385 386 387 388
	struct vxlan_fdb f = {
		.state = NUD_STALE,
	};
	struct vxlan_rdst remote = {
C
Cong Wang 已提交
389
		.remote_ip = *ipa, /* goes to NDA_DST */
390
		.remote_vni = cpu_to_be32(VXLAN_N_VID),
391
	};
392

393
	vxlan_fdb_notify(vxlan, &f, &remote, RTM_GETNEIGH);
D
David Stevens 已提交
394 395 396 397
}

static void vxlan_fdb_miss(struct vxlan_dev *vxlan, const u8 eth_addr[ETH_ALEN])
{
398 399 400
	struct vxlan_fdb f = {
		.state = NUD_STALE,
	};
401
	struct vxlan_rdst remote = { };
D
David Stevens 已提交
402 403 404

	memcpy(f.eth_addr, eth_addr, ETH_ALEN);

405
	vxlan_fdb_notify(vxlan, &f, &remote, RTM_GETNEIGH);
D
David Stevens 已提交
406 407
}

S
stephen hemminger 已提交
408 409 410 411 412 413 414 415
/* Hash Ethernet address */
static u32 eth_hash(const unsigned char *addr)
{
	u64 value = get_unaligned((u64 *)addr);

	/* only want 6 bytes */
#ifdef __BIG_ENDIAN
	value >>= 16;
416 417
#else
	value <<= 16;
S
stephen hemminger 已提交
418 419 420 421 422 423 424 425 426 427 428 429
#endif
	return hash_64(value, FDB_HASH_BITS);
}

/* Hash chain to use given mac address */
static inline struct hlist_head *vxlan_fdb_head(struct vxlan_dev *vxlan,
						const u8 *mac)
{
	return &vxlan->fdb_head[eth_hash(mac)];
}

/* Look up Ethernet address in forwarding table */
430
static struct vxlan_fdb *__vxlan_find_mac(struct vxlan_dev *vxlan,
S
stephen hemminger 已提交
431 432 433 434 435
					const u8 *mac)
{
	struct hlist_head *head = vxlan_fdb_head(vxlan, mac);
	struct vxlan_fdb *f;

436
	hlist_for_each_entry_rcu(f, head, hlist) {
437
		if (ether_addr_equal(mac, f->eth_addr))
S
stephen hemminger 已提交
438 439 440 441 442 443
			return f;
	}

	return NULL;
}

444 445 446 447 448 449 450 451 452 453 454 455
static struct vxlan_fdb *vxlan_find_mac(struct vxlan_dev *vxlan,
					const u8 *mac)
{
	struct vxlan_fdb *f;

	f = __vxlan_find_mac(vxlan, mac);
	if (f)
		f->used = jiffies;

	return f;
}

456 457
/* caller should hold vxlan->hash_lock */
static struct vxlan_rdst *vxlan_fdb_find_rdst(struct vxlan_fdb *f,
C
Cong Wang 已提交
458
					      union vxlan_addr *ip, __be16 port,
459
					      __be32 vni, __u32 ifindex)
460
{
461
	struct vxlan_rdst *rd;
462

463
	list_for_each_entry(rd, &f->remotes, list) {
C
Cong Wang 已提交
464
		if (vxlan_addr_equal(&rd->remote_ip, ip) &&
465 466 467
		    rd->remote_port == port &&
		    rd->remote_vni == vni &&
		    rd->remote_ifindex == ifindex)
468
			return rd;
469
	}
470

471 472 473
	return NULL;
}

474 475
/* Replace destination of unicast mac */
static int vxlan_fdb_replace(struct vxlan_fdb *f,
476 477
			     union vxlan_addr *ip, __be16 port, __be32 vni,
			     __u32 ifindex)
478 479 480 481 482 483 484 485 486 487
{
	struct vxlan_rdst *rd;

	rd = vxlan_fdb_find_rdst(f, ip, port, vni, ifindex);
	if (rd)
		return 0;

	rd = list_first_entry_or_null(&f->remotes, struct vxlan_rdst, list);
	if (!rd)
		return 0;
488 489

	dst_cache_reset(&rd->dst_cache);
C
Cong Wang 已提交
490
	rd->remote_ip = *ip;
491 492 493 494 495 496
	rd->remote_port = port;
	rd->remote_vni = vni;
	rd->remote_ifindex = ifindex;
	return 1;
}

497 498
/* Add/update destinations for multicast */
static int vxlan_fdb_append(struct vxlan_fdb *f,
499
			    union vxlan_addr *ip, __be16 port, __be32 vni,
500
			    __u32 ifindex, struct vxlan_rdst **rdp)
501 502 503 504 505 506 507
{
	struct vxlan_rdst *rd;

	rd = vxlan_fdb_find_rdst(f, ip, port, vni, ifindex);
	if (rd)
		return 0;

508 509 510
	rd = kmalloc(sizeof(*rd), GFP_ATOMIC);
	if (rd == NULL)
		return -ENOBUFS;
511 512 513 514 515 516

	if (dst_cache_init(&rd->dst_cache, GFP_ATOMIC)) {
		kfree(rd);
		return -ENOBUFS;
	}

C
Cong Wang 已提交
517
	rd->remote_ip = *ip;
518 519 520
	rd->remote_port = port;
	rd->remote_vni = vni;
	rd->remote_ifindex = ifindex;
521 522 523

	list_add_tail_rcu(&rd->list, &f->remotes);

524
	*rdp = rd;
525 526 527
	return 1;
}

T
Tom Herbert 已提交
528 529 530
static struct vxlanhdr *vxlan_gro_remcsum(struct sk_buff *skb,
					  unsigned int off,
					  struct vxlanhdr *vh, size_t hdrlen,
531 532
					  __be32 vni_field,
					  struct gro_remcsum *grc,
533
					  bool nopartial)
T
Tom Herbert 已提交
534
{
535
	size_t start, offset;
T
Tom Herbert 已提交
536 537

	if (skb->remcsum_offload)
538
		return vh;
T
Tom Herbert 已提交
539 540 541 542

	if (!NAPI_GRO_CB(skb)->csum_valid)
		return NULL;

543 544
	start = vxlan_rco_start(vni_field);
	offset = start + vxlan_rco_offset(vni_field);
T
Tom Herbert 已提交
545

546 547
	vh = skb_gro_remcsum_process(skb, (void *)vh, off, hdrlen,
				     start, offset, grc, nopartial);
T
Tom Herbert 已提交
548 549 550 551 552 553

	skb->remcsum_offload = 1;

	return vh;
}

554 555 556
static struct sk_buff **vxlan_gro_receive(struct sk_buff **head,
					  struct sk_buff *skb,
					  struct udp_offload *uoff)
557 558 559
{
	struct sk_buff *p, **pp = NULL;
	struct vxlanhdr *vh, *vh2;
560
	unsigned int hlen, off_vx;
561
	int flush = 1;
T
Tom Herbert 已提交
562 563
	struct vxlan_sock *vs = container_of(uoff, struct vxlan_sock,
					     udp_offloads);
564
	__be32 flags;
565 566 567
	struct gro_remcsum grc;

	skb_gro_remcsum_init(&grc);
568 569 570 571 572 573 574 575 576 577

	off_vx = skb_gro_offset(skb);
	hlen = off_vx + sizeof(*vh);
	vh   = skb_gro_header_fast(skb, off_vx);
	if (skb_gro_header_hard(skb, hlen)) {
		vh = skb_gro_header_slow(skb, hlen, off_vx);
		if (unlikely(!vh))
			goto out;
	}

T
Tom Herbert 已提交
578 579
	skb_gro_postpull_rcsum(skb, vh, sizeof(struct vxlanhdr));

580
	flags = vh->vx_flags;
T
Tom Herbert 已提交
581 582 583

	if ((flags & VXLAN_HF_RCO) && (vs->flags & VXLAN_F_REMCSUM_RX)) {
		vh = vxlan_gro_remcsum(skb, off_vx, vh, sizeof(struct vxlanhdr),
584
				       vh->vx_vni, &grc,
585 586
				       !!(vs->flags &
					  VXLAN_F_REMCSUM_NOPARTIAL));
T
Tom Herbert 已提交
587 588 589 590 591

		if (!vh)
			goto out;
	}

592 593
	skb_gro_pull(skb, sizeof(struct vxlanhdr)); /* pull vxlan header */

594 595 596 597 598
	for (p = *head; p; p = p->next) {
		if (!NAPI_GRO_CB(p)->same_flow)
			continue;

		vh2 = (struct vxlanhdr *)(p->data + off_vx);
T
Thomas Graf 已提交
599 600
		if (vh->vx_flags != vh2->vx_flags ||
		    vh->vx_vni != vh2->vx_vni) {
601 602 603 604 605
			NAPI_GRO_CB(p)->same_flow = 0;
			continue;
		}
	}

606
	pp = eth_gro_receive(head, skb);
607
	flush = 0;
608 609

out:
610
	skb_gro_remcsum_cleanup(skb, &grc);
611 612 613 614 615
	NAPI_GRO_CB(skb)->flush |= flush;

	return pp;
}

616 617
static int vxlan_gro_complete(struct sk_buff *skb, int nhoff,
			      struct udp_offload *uoff)
618
{
619 620
	udp_tunnel_gro_complete(skb, nhoff);

621
	return eth_gro_complete(skb, nhoff + sizeof(struct vxlanhdr));
622 623
}

624
/* Notify netdevs that UDP port started listening */
625
static void vxlan_notify_add_rx_port(struct vxlan_sock *vs)
626 627
{
	struct net_device *dev;
628
	struct sock *sk = vs->sock->sk;
629
	struct net *net = sock_net(sk);
630
	sa_family_t sa_family = vxlan_get_sk_family(vs);
J
Joseph Gasparakis 已提交
631
	__be16 port = inet_sk(sk)->inet_sport;
632 633 634
	int err;

	if (sa_family == AF_INET) {
635
		err = udp_add_offload(net, &vs->udp_offloads);
636 637 638
		if (err)
			pr_warn("vxlan: udp_add_offload failed with status %d\n", err);
	}
639 640 641 642 643 644 645 646 647 648 649

	rcu_read_lock();
	for_each_netdev_rcu(net, dev) {
		if (dev->netdev_ops->ndo_add_vxlan_port)
			dev->netdev_ops->ndo_add_vxlan_port(dev, sa_family,
							    port);
	}
	rcu_read_unlock();
}

/* Notify netdevs that UDP port is no more listening */
650
static void vxlan_notify_del_rx_port(struct vxlan_sock *vs)
651 652
{
	struct net_device *dev;
653
	struct sock *sk = vs->sock->sk;
654
	struct net *net = sock_net(sk);
655
	sa_family_t sa_family = vxlan_get_sk_family(vs);
J
Joseph Gasparakis 已提交
656
	__be16 port = inet_sk(sk)->inet_sport;
657 658 659 660 661 662 663 664

	rcu_read_lock();
	for_each_netdev_rcu(net, dev) {
		if (dev->netdev_ops->ndo_del_vxlan_port)
			dev->netdev_ops->ndo_del_vxlan_port(dev, sa_family,
							    port);
	}
	rcu_read_unlock();
665 666 667

	if (sa_family == AF_INET)
		udp_del_offload(&vs->udp_offloads);
668 669
}

S
stephen hemminger 已提交
670 671
/* Add new entry to forwarding table -- assumes lock held */
static int vxlan_fdb_create(struct vxlan_dev *vxlan,
C
Cong Wang 已提交
672
			    const u8 *mac, union vxlan_addr *ip,
673
			    __u16 state, __u16 flags,
674
			    __be16 port, __be32 vni, __u32 ifindex,
675
			    __u8 ndm_flags)
S
stephen hemminger 已提交
676
{
677
	struct vxlan_rdst *rd = NULL;
S
stephen hemminger 已提交
678 679 680
	struct vxlan_fdb *f;
	int notify = 0;

681
	f = __vxlan_find_mac(vxlan, mac);
S
stephen hemminger 已提交
682 683 684 685 686 687 688 689 690 691 692
	if (f) {
		if (flags & NLM_F_EXCL) {
			netdev_dbg(vxlan->dev,
				   "lost race to create %pM\n", mac);
			return -EEXIST;
		}
		if (f->state != state) {
			f->state = state;
			f->updated = jiffies;
			notify = 1;
		}
693 694 695 696 697
		if (f->flags != ndm_flags) {
			f->flags = ndm_flags;
			f->updated = jiffies;
			notify = 1;
		}
698 699 700 701
		if ((flags & NLM_F_REPLACE)) {
			/* Only change unicasts */
			if (!(is_multicast_ether_addr(f->eth_addr) ||
			     is_zero_ether_addr(f->eth_addr))) {
702
				notify |= vxlan_fdb_replace(f, ip, port, vni,
703 704 705 706
							   ifindex);
			} else
				return -EOPNOTSUPP;
		}
707
		if ((flags & NLM_F_APPEND) &&
708 709
		    (is_multicast_ether_addr(f->eth_addr) ||
		     is_zero_ether_addr(f->eth_addr))) {
710 711
			int rc = vxlan_fdb_append(f, ip, port, vni, ifindex,
						  &rd);
712 713 714 715 716

			if (rc < 0)
				return rc;
			notify |= rc;
		}
S
stephen hemminger 已提交
717 718 719 720
	} else {
		if (!(flags & NLM_F_CREATE))
			return -ENOENT;

721 722
		if (vxlan->cfg.addrmax &&
		    vxlan->addrcnt >= vxlan->cfg.addrmax)
S
stephen hemminger 已提交
723 724
			return -ENOSPC;

725 726 727 728 729
		/* Disallow replace to add a multicast entry */
		if ((flags & NLM_F_REPLACE) &&
		    (is_multicast_ether_addr(mac) || is_zero_ether_addr(mac)))
			return -EOPNOTSUPP;

C
Cong Wang 已提交
730
		netdev_dbg(vxlan->dev, "add %pM -> %pIS\n", mac, ip);
S
stephen hemminger 已提交
731 732 733 734 735 736
		f = kmalloc(sizeof(*f), GFP_ATOMIC);
		if (!f)
			return -ENOMEM;

		notify = 1;
		f->state = state;
737
		f->flags = ndm_flags;
S
stephen hemminger 已提交
738
		f->updated = f->used = jiffies;
739
		INIT_LIST_HEAD(&f->remotes);
S
stephen hemminger 已提交
740 741
		memcpy(f->eth_addr, mac, ETH_ALEN);

742
		vxlan_fdb_append(f, ip, port, vni, ifindex, &rd);
743

S
stephen hemminger 已提交
744 745 746 747 748
		++vxlan->addrcnt;
		hlist_add_head_rcu(&f->hlist,
				   vxlan_fdb_head(vxlan, mac));
	}

749 750 751 752 753
	if (notify) {
		if (rd == NULL)
			rd = first_remote_rtnl(f);
		vxlan_fdb_notify(vxlan, f, rd, RTM_NEWNEIGH);
	}
S
stephen hemminger 已提交
754 755 756 757

	return 0;
}

W
Wei Yongjun 已提交
758
static void vxlan_fdb_free(struct rcu_head *head)
759 760
{
	struct vxlan_fdb *f = container_of(head, struct vxlan_fdb, rcu);
761
	struct vxlan_rdst *rd, *nd;
762

763 764
	list_for_each_entry_safe(rd, nd, &f->remotes, list) {
		dst_cache_destroy(&rd->dst_cache);
765
		kfree(rd);
766
	}
767 768 769
	kfree(f);
}

S
stephen hemminger 已提交
770 771 772 773 774 775
static void vxlan_fdb_destroy(struct vxlan_dev *vxlan, struct vxlan_fdb *f)
{
	netdev_dbg(vxlan->dev,
		    "delete %pM\n", f->eth_addr);

	--vxlan->addrcnt;
776
	vxlan_fdb_notify(vxlan, f, first_remote_rtnl(f), RTM_DELNEIGH);
S
stephen hemminger 已提交
777 778

	hlist_del_rcu(&f->hlist);
779
	call_rcu(&f->rcu, vxlan_fdb_free);
S
stephen hemminger 已提交
780 781
}

M
Mike Rapoport 已提交
782
static int vxlan_fdb_parse(struct nlattr *tb[], struct vxlan_dev *vxlan,
783 784
			   union vxlan_addr *ip, __be16 *port, __be32 *vni,
			   u32 *ifindex)
S
stephen hemminger 已提交
785
{
786
	struct net *net = dev_net(vxlan->dev);
C
Cong Wang 已提交
787
	int err;
S
stephen hemminger 已提交
788

M
Mike Rapoport 已提交
789
	if (tb[NDA_DST]) {
C
Cong Wang 已提交
790 791 792
		err = vxlan_nla_get_addr(ip, tb[NDA_DST]);
		if (err)
			return err;
M
Mike Rapoport 已提交
793
	} else {
C
Cong Wang 已提交
794 795 796 797 798 799 800 801 802 803
		union vxlan_addr *remote = &vxlan->default_dst.remote_ip;
		if (remote->sa.sa_family == AF_INET) {
			ip->sin.sin_addr.s_addr = htonl(INADDR_ANY);
			ip->sa.sa_family = AF_INET;
#if IS_ENABLED(CONFIG_IPV6)
		} else {
			ip->sin6.sin6_addr = in6addr_any;
			ip->sa.sa_family = AF_INET6;
#endif
		}
M
Mike Rapoport 已提交
804
	}
S
stephen hemminger 已提交
805

806
	if (tb[NDA_PORT]) {
807
		if (nla_len(tb[NDA_PORT]) != sizeof(__be16))
808
			return -EINVAL;
M
Mike Rapoport 已提交
809 810
		*port = nla_get_be16(tb[NDA_PORT]);
	} else {
811
		*port = vxlan->cfg.dst_port;
M
Mike Rapoport 已提交
812
	}
813 814 815 816

	if (tb[NDA_VNI]) {
		if (nla_len(tb[NDA_VNI]) != sizeof(u32))
			return -EINVAL;
817
		*vni = cpu_to_be32(nla_get_u32(tb[NDA_VNI]));
M
Mike Rapoport 已提交
818 819 820
	} else {
		*vni = vxlan->default_dst.remote_vni;
	}
821 822

	if (tb[NDA_IFINDEX]) {
P
Pravin B Shelar 已提交
823
		struct net_device *tdev;
824 825 826

		if (nla_len(tb[NDA_IFINDEX]) != sizeof(u32))
			return -EINVAL;
M
Mike Rapoport 已提交
827
		*ifindex = nla_get_u32(tb[NDA_IFINDEX]);
828
		tdev = __dev_get_by_index(net, *ifindex);
P
Pravin B Shelar 已提交
829
		if (!tdev)
830
			return -EADDRNOTAVAIL;
M
Mike Rapoport 已提交
831 832 833 834 835 836 837 838 839 840
	} else {
		*ifindex = 0;
	}

	return 0;
}

/* Add static entry (via netlink) */
static int vxlan_fdb_add(struct ndmsg *ndm, struct nlattr *tb[],
			 struct net_device *dev,
841
			 const unsigned char *addr, u16 vid, u16 flags)
M
Mike Rapoport 已提交
842 843 844
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
	/* struct net *net = dev_net(vxlan->dev); */
C
Cong Wang 已提交
845
	union vxlan_addr ip;
M
Mike Rapoport 已提交
846
	__be16 port;
847 848
	__be32 vni;
	u32 ifindex;
M
Mike Rapoport 已提交
849 850 851 852 853 854 855 856 857 858 859 860 861 862
	int err;

	if (!(ndm->ndm_state & (NUD_PERMANENT|NUD_REACHABLE))) {
		pr_info("RTM_NEWNEIGH with invalid state %#x\n",
			ndm->ndm_state);
		return -EINVAL;
	}

	if (tb[NDA_DST] == NULL)
		return -EINVAL;

	err = vxlan_fdb_parse(tb, vxlan, &ip, &port, &vni, &ifindex);
	if (err)
		return err;
863

864 865 866
	if (vxlan->default_dst.remote_ip.sa.sa_family != ip.sa.sa_family)
		return -EAFNOSUPPORT;

S
stephen hemminger 已提交
867
	spin_lock_bh(&vxlan->hash_lock);
C
Cong Wang 已提交
868
	err = vxlan_fdb_create(vxlan, addr, &ip, ndm->ndm_state, flags,
869
			       port, vni, ifindex, ndm->ndm_flags);
S
stephen hemminger 已提交
870 871 872 873 874 875
	spin_unlock_bh(&vxlan->hash_lock);

	return err;
}

/* Delete entry (via netlink) */
876 877
static int vxlan_fdb_delete(struct ndmsg *ndm, struct nlattr *tb[],
			    struct net_device *dev,
878
			    const unsigned char *addr, u16 vid)
S
stephen hemminger 已提交
879 880 881
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
	struct vxlan_fdb *f;
882
	struct vxlan_rdst *rd = NULL;
C
Cong Wang 已提交
883
	union vxlan_addr ip;
884
	__be16 port;
885 886
	__be32 vni;
	u32 ifindex;
887 888 889 890 891 892 893
	int err;

	err = vxlan_fdb_parse(tb, vxlan, &ip, &port, &vni, &ifindex);
	if (err)
		return err;

	err = -ENOENT;
S
stephen hemminger 已提交
894 895 896

	spin_lock_bh(&vxlan->hash_lock);
	f = vxlan_find_mac(vxlan, addr);
897 898 899
	if (!f)
		goto out;

C
Cong Wang 已提交
900 901
	if (!vxlan_addr_any(&ip)) {
		rd = vxlan_fdb_find_rdst(f, &ip, port, vni, ifindex);
902 903 904 905 906 907 908 909 910 911 912
		if (!rd)
			goto out;
	}

	err = 0;

	/* remove a destination if it's not the only one on the list,
	 * otherwise destroy the fdb entry
	 */
	if (rd && !list_is_singular(&f->remotes)) {
		list_del_rcu(&rd->list);
913
		vxlan_fdb_notify(vxlan, f, rd, RTM_DELNEIGH);
914
		kfree_rcu(rd, rcu);
915
		goto out;
S
stephen hemminger 已提交
916
	}
917 918 919 920

	vxlan_fdb_destroy(vxlan, f);

out:
S
stephen hemminger 已提交
921 922 923 924 925 926 927
	spin_unlock_bh(&vxlan->hash_lock);

	return err;
}

/* Dump forwarding table */
static int vxlan_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb,
928 929
			  struct net_device *dev,
			  struct net_device *filter_dev, int idx)
S
stephen hemminger 已提交
930 931 932 933 934 935 936 937
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
	unsigned int h;

	for (h = 0; h < FDB_HASH_SIZE; ++h) {
		struct vxlan_fdb *f;
		int err;

938
		hlist_for_each_entry_rcu(f, &vxlan->fdb_head[h], hlist) {
939 940
			struct vxlan_rdst *rd;

941
			list_for_each_entry_rcu(rd, &f->remotes, list) {
942 943 944
				if (idx < cb->args[0])
					goto skip;

945 946 947 948 949
				err = vxlan_fdb_info(skb, vxlan, f,
						     NETLINK_CB(cb->skb).portid,
						     cb->nlh->nlmsg_seq,
						     RTM_NEWNEIGH,
						     NLM_F_MULTI, rd);
950 951
				if (err < 0) {
					cb->args[1] = err;
952
					goto out;
953
				}
954
skip:
955 956
				++idx;
			}
S
stephen hemminger 已提交
957 958
		}
	}
959
out:
S
stephen hemminger 已提交
960 961 962 963 964
	return idx;
}

/* Watch incoming packets to learn mapping between Ethernet address
 * and Tunnel endpoint.
965
 * Return true if packet is bogus and should be dropped.
S
stephen hemminger 已提交
966
 */
967
static bool vxlan_snoop(struct net_device *dev,
C
Cong Wang 已提交
968
			union vxlan_addr *src_ip, const u8 *src_mac)
S
stephen hemminger 已提交
969 970 971 972 973 974
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
	struct vxlan_fdb *f;

	f = vxlan_find_mac(vxlan, src_mac);
	if (likely(f)) {
975
		struct vxlan_rdst *rdst = first_remote_rcu(f);
976

C
Cong Wang 已提交
977
		if (likely(vxlan_addr_equal(&rdst->remote_ip, src_ip)))
978 979 980
			return false;

		/* Don't migrate static entries, drop packets */
981
		if (f->state & NUD_NOARP)
982
			return true;
S
stephen hemminger 已提交
983 984 985

		if (net_ratelimit())
			netdev_info(dev,
C
Cong Wang 已提交
986
				    "%pM migrated from %pIS to %pIS\n",
987
				    src_mac, &rdst->remote_ip.sa, &src_ip->sa);
S
stephen hemminger 已提交
988

C
Cong Wang 已提交
989
		rdst->remote_ip = *src_ip;
S
stephen hemminger 已提交
990
		f->updated = jiffies;
991
		vxlan_fdb_notify(vxlan, f, rdst, RTM_NEWNEIGH);
S
stephen hemminger 已提交
992 993 994
	} else {
		/* learned new entry */
		spin_lock(&vxlan->hash_lock);
995 996 997 998 999 1000

		/* close off race between vxlan_flush and incoming packets */
		if (netif_running(dev))
			vxlan_fdb_create(vxlan, src_mac, src_ip,
					 NUD_REACHABLE,
					 NLM_F_EXCL|NLM_F_CREATE,
1001
					 vxlan->cfg.dst_port,
1002 1003
					 vxlan->default_dst.remote_vni,
					 0, NTF_SELF);
S
stephen hemminger 已提交
1004 1005
		spin_unlock(&vxlan->hash_lock);
	}
1006 1007

	return false;
S
stephen hemminger 已提交
1008 1009 1010
}

/* See if multicast group is already in use by other ID */
1011
static bool vxlan_group_used(struct vxlan_net *vn, struct vxlan_dev *dev)
S
stephen hemminger 已提交
1012
{
1013
	struct vxlan_dev *vxlan;
1014
	unsigned short family = dev->default_dst.remote_ip.sa.sa_family;
S
stephen hemminger 已提交
1015

1016 1017 1018
	/* The vxlan_sock is only used by dev, leaving group has
	 * no effect on other vxlan devices.
	 */
1019 1020
	if (family == AF_INET && dev->vn4_sock &&
	    atomic_read(&dev->vn4_sock->refcnt) == 1)
1021
		return false;
1022 1023 1024 1025 1026
#if IS_ENABLED(CONFIG_IPV6)
	if (family == AF_INET6 && dev->vn6_sock &&
	    atomic_read(&dev->vn6_sock->refcnt) == 1)
		return false;
#endif
1027

1028
	list_for_each_entry(vxlan, &vn->vxlan_list, next) {
1029
		if (!netif_running(vxlan->dev) || vxlan == dev)
1030
			continue;
S
stephen hemminger 已提交
1031

1032
		if (family == AF_INET && vxlan->vn4_sock != dev->vn4_sock)
1033
			continue;
1034 1035 1036 1037
#if IS_ENABLED(CONFIG_IPV6)
		if (family == AF_INET6 && vxlan->vn6_sock != dev->vn6_sock)
			continue;
#endif
1038 1039 1040 1041 1042 1043 1044 1045 1046 1047

		if (!vxlan_addr_equal(&vxlan->default_dst.remote_ip,
				      &dev->default_dst.remote_ip))
			continue;

		if (vxlan->default_dst.remote_ifindex !=
		    dev->default_dst.remote_ifindex)
			continue;

		return true;
1048
	}
S
stephen hemminger 已提交
1049 1050 1051 1052

	return false;
}

1053
static void __vxlan_sock_release(struct vxlan_sock *vs)
1054
{
1055
	struct vxlan_net *vn;
1056

1057 1058
	if (!vs)
		return;
1059 1060
	if (!atomic_dec_and_test(&vs->refcnt))
		return;
S
stephen hemminger 已提交
1061

1062
	vn = net_generic(sock_net(vs->sock->sk), vxlan_net_id);
1063
	spin_lock(&vn->sock_lock);
1064
	hlist_del_rcu(&vs->hlist);
1065
	vxlan_notify_del_rx_port(vs);
1066 1067
	spin_unlock(&vn->sock_lock);

1068
	queue_work(vxlan_wq, &vs->del_work);
S
stephen hemminger 已提交
1069 1070
}

1071 1072 1073 1074 1075 1076 1077 1078
static void vxlan_sock_release(struct vxlan_dev *vxlan)
{
	__vxlan_sock_release(vxlan->vn4_sock);
#if IS_ENABLED(CONFIG_IPV6)
	__vxlan_sock_release(vxlan->vn6_sock);
#endif
}

1079
/* Update multicast group membership when first VNI on
1080
 * multicast address is brought up
1081
 */
1082
static int vxlan_igmp_join(struct vxlan_dev *vxlan)
S
stephen hemminger 已提交
1083
{
1084
	struct sock *sk;
C
Cong Wang 已提交
1085 1086
	union vxlan_addr *ip = &vxlan->default_dst.remote_ip;
	int ifindex = vxlan->default_dst.remote_ifindex;
1087
	int ret = -EINVAL;
S
stephen hemminger 已提交
1088

C
Cong Wang 已提交
1089 1090 1091 1092 1093 1094
	if (ip->sa.sa_family == AF_INET) {
		struct ip_mreqn mreq = {
			.imr_multiaddr.s_addr	= ip->sin.sin_addr.s_addr,
			.imr_ifindex		= ifindex,
		};

1095 1096
		sk = vxlan->vn4_sock->sock->sk;
		lock_sock(sk);
1097
		ret = ip_mc_join_group(sk, &mreq);
1098
		release_sock(sk);
C
Cong Wang 已提交
1099 1100
#if IS_ENABLED(CONFIG_IPV6)
	} else {
1101 1102
		sk = vxlan->vn6_sock->sock->sk;
		lock_sock(sk);
1103 1104
		ret = ipv6_stub->ipv6_sock_mc_join(sk, ifindex,
						   &ip->sin6.sin6_addr);
1105
		release_sock(sk);
C
Cong Wang 已提交
1106 1107
#endif
	}
S
stephen hemminger 已提交
1108

1109
	return ret;
S
stephen hemminger 已提交
1110 1111 1112
}

/* Inverse of vxlan_igmp_join when last VNI is brought down */
1113
static int vxlan_igmp_leave(struct vxlan_dev *vxlan)
S
stephen hemminger 已提交
1114
{
1115
	struct sock *sk;
C
Cong Wang 已提交
1116 1117
	union vxlan_addr *ip = &vxlan->default_dst.remote_ip;
	int ifindex = vxlan->default_dst.remote_ifindex;
1118
	int ret = -EINVAL;
S
stephen hemminger 已提交
1119

C
Cong Wang 已提交
1120 1121 1122 1123 1124 1125
	if (ip->sa.sa_family == AF_INET) {
		struct ip_mreqn mreq = {
			.imr_multiaddr.s_addr	= ip->sin.sin_addr.s_addr,
			.imr_ifindex		= ifindex,
		};

1126 1127
		sk = vxlan->vn4_sock->sock->sk;
		lock_sock(sk);
1128
		ret = ip_mc_leave_group(sk, &mreq);
1129
		release_sock(sk);
C
Cong Wang 已提交
1130 1131
#if IS_ENABLED(CONFIG_IPV6)
	} else {
1132 1133
		sk = vxlan->vn6_sock->sock->sk;
		lock_sock(sk);
1134 1135
		ret = ipv6_stub->ipv6_sock_mc_drop(sk, ifindex,
						   &ip->sin6.sin6_addr);
1136
		release_sock(sk);
C
Cong Wang 已提交
1137 1138
#endif
	}
S
stephen hemminger 已提交
1139

1140
	return ret;
S
stephen hemminger 已提交
1141 1142
}

1143 1144
static bool vxlan_remcsum(struct vxlanhdr *unparsed,
			  struct sk_buff *skb, u32 vxflags)
T
Tom Herbert 已提交
1145
{
1146
	size_t start, offset;
T
Tom Herbert 已提交
1147

1148 1149
	if (!(unparsed->vx_flags & VXLAN_HF_RCO) || skb->remcsum_offload)
		goto out;
1150

1151 1152
	start = vxlan_rco_start(unparsed->vx_vni);
	offset = start + vxlan_rco_offset(unparsed->vx_vni);
T
Tom Herbert 已提交
1153

1154
	if (!pskb_may_pull(skb, offset + sizeof(u16)))
J
Jiri Benc 已提交
1155
		return false;
T
Tom Herbert 已提交
1156

J
Jiri Benc 已提交
1157 1158
	skb_remcsum_process(skb, (void *)(vxlan_hdr(skb) + 1), start, offset,
			    !!(vxflags & VXLAN_F_REMCSUM_NOPARTIAL));
1159 1160 1161
out:
	unparsed->vx_flags &= ~VXLAN_HF_RCO;
	unparsed->vx_vni &= VXLAN_VNI_MASK;
J
Jiri Benc 已提交
1162
	return true;
T
Tom Herbert 已提交
1163 1164
}

1165
static void vxlan_parse_gbp_hdr(struct vxlanhdr *unparsed,
1166
				struct sk_buff *skb, u32 vxflags,
1167
				struct vxlan_metadata *md)
1168
{
1169
	struct vxlanhdr_gbp *gbp = (struct vxlanhdr_gbp *)unparsed;
1170
	struct metadata_dst *tun_dst;
1171 1172 1173

	if (!(unparsed->vx_flags & VXLAN_HF_GBP))
		goto out;
1174 1175 1176

	md->gbp = ntohs(gbp->policy_id);

1177
	tun_dst = (struct metadata_dst *)skb_dst(skb);
1178
	if (tun_dst) {
1179
		tun_dst->u.tun_info.key.tun_flags |= TUNNEL_VXLAN_OPT;
1180 1181
		tun_dst->u.tun_info.options_len = sizeof(*md);
	}
1182 1183 1184 1185 1186
	if (gbp->dont_learn)
		md->gbp |= VXLAN_GBP_DONT_LEARN;

	if (gbp->policy_applied)
		md->gbp |= VXLAN_GBP_POLICY_APPLIED;
1187

1188 1189 1190
	/* In flow-based mode, GBP is carried in dst_metadata */
	if (!(vxflags & VXLAN_F_COLLECT_METADATA))
		skb->mark = md->gbp;
1191 1192
out:
	unparsed->vx_flags &= ~VXLAN_GBP_USED_BITS;
1193 1194
}

1195 1196 1197
static bool vxlan_set_mac(struct vxlan_dev *vxlan,
			  struct vxlan_sock *vs,
			  struct sk_buff *skb)
1198 1199 1200 1201 1202 1203 1204 1205 1206
{
	union vxlan_addr saddr;

	skb_reset_mac_header(skb);
	skb->protocol = eth_type_trans(skb, vxlan->dev);
	skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);

	/* Ignore packet loops (and multicast echo) */
	if (ether_addr_equal(eth_hdr(skb)->h_source, vxlan->dev->dev_addr))
1207
		return false;
1208

1209
	/* Get address from the outer IP header */
1210
	if (vxlan_get_sk_family(vs) == AF_INET) {
1211
		saddr.sin.sin_addr.s_addr = ip_hdr(skb)->saddr;
1212 1213 1214
		saddr.sa.sa_family = AF_INET;
#if IS_ENABLED(CONFIG_IPV6)
	} else {
1215
		saddr.sin6.sin6_addr = ipv6_hdr(skb)->saddr;
1216 1217 1218 1219
		saddr.sa.sa_family = AF_INET6;
#endif
	}

1220 1221 1222 1223 1224 1225 1226
	if ((vxlan->flags & VXLAN_F_LEARN) &&
	    vxlan_snoop(skb->dev, &saddr, eth_hdr(skb)->h_source))
		return false;

	return true;
}

1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
static bool vxlan_ecn_decapsulate(struct vxlan_sock *vs, void *oiph,
				  struct sk_buff *skb)
{
	int err = 0;

	if (vxlan_get_sk_family(vs) == AF_INET)
		err = IP_ECN_decapsulate(oiph, skb);
#if IS_ENABLED(CONFIG_IPV6)
	else
		err = IP6_ECN_decapsulate(oiph, skb);
#endif

	if (unlikely(err) && log_ecn_error) {
		if (vxlan_get_sk_family(vs) == AF_INET)
			net_info_ratelimited("non-ECT from %pI4 with TOS=%#x\n",
					     &((struct iphdr *)oiph)->saddr,
					     ((struct iphdr *)oiph)->tos);
		else
			net_info_ratelimited("non-ECT from %pI6\n",
					     &((struct ipv6hdr *)oiph)->saddr);
	}
	return err <= 1;
}

S
stephen hemminger 已提交
1251
/* Callback from net/ipv4/udp.c to receive packets */
1252
static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
S
stephen hemminger 已提交
1253
{
1254
	struct pcpu_sw_netstats *stats;
1255
	struct vxlan_dev *vxlan;
P
Pravin B Shelar 已提交
1256
	struct vxlan_sock *vs;
1257
	struct vxlanhdr unparsed;
T
Thomas Graf 已提交
1258 1259
	struct vxlan_metadata _md;
	struct vxlan_metadata *md = &_md;
1260
	void *oiph;
S
stephen hemminger 已提交
1261 1262

	/* Need Vxlan and inner Ethernet header to be present */
1263
	if (!pskb_may_pull(skb, VXLAN_HLEN))
J
Jiri Benc 已提交
1264
		return 1;
S
stephen hemminger 已提交
1265

1266
	unparsed = *vxlan_hdr(skb);
J
Jiri Benc 已提交
1267 1268 1269 1270 1271 1272 1273
	/* VNI flag always required to be set */
	if (!(unparsed.vx_flags & VXLAN_HF_VNI)) {
		netdev_dbg(skb->dev, "invalid vxlan flags=%#x vni=%#x\n",
			   ntohl(vxlan_hdr(skb)->vx_flags),
			   ntohl(vxlan_hdr(skb)->vx_vni));
		/* Return non vxlan pkt */
		return 1;
S
stephen hemminger 已提交
1274
	}
J
Jiri Benc 已提交
1275 1276
	unparsed.vx_flags &= ~VXLAN_HF_VNI;
	unparsed.vx_vni &= ~VXLAN_VNI_MASK;
S
stephen hemminger 已提交
1277

1278
	vs = rcu_dereference_sk_user_data(sk);
P
Pravin B Shelar 已提交
1279
	if (!vs)
S
stephen hemminger 已提交
1280 1281
		goto drop;

1282 1283 1284 1285
	vxlan = vxlan_vs_find_vni(vs, vxlan_vni(vxlan_hdr(skb)->vx_vni));
	if (!vxlan)
		goto drop;

1286 1287
	if (iptunnel_pull_header(skb, VXLAN_HLEN, htons(ETH_P_TEB),
				 !net_eq(vxlan->net, dev_net(vxlan->dev))))
1288 1289
		goto drop;

T
Thomas Graf 已提交
1290
	if (vxlan_collect_metadata(vs)) {
J
Jiri Benc 已提交
1291
		__be32 vni = vxlan_vni(vxlan_hdr(skb)->vx_vni);
1292
		struct metadata_dst *tun_dst;
J
Jiri Benc 已提交
1293

1294
		tun_dst = udp_tun_rx_dst(skb, vxlan_get_sk_family(vs), TUNNEL_KEY,
J
Jiri Benc 已提交
1295
					 vxlan_vni_to_tun_id(vni), sizeof(*md));
1296

T
Thomas Graf 已提交
1297 1298 1299
		if (!tun_dst)
			goto drop;

1300
		md = ip_tunnel_info_opts(&tun_dst->u.tun_info);
1301 1302

		skb_dst_set(skb, (struct dst_entry *)tun_dst);
T
Thomas Graf 已提交
1303 1304 1305 1306
	} else {
		memset(md, 0, sizeof(*md));
	}

T
Thomas Graf 已提交
1307 1308 1309
	/* For backwards compatibility, only allow reserved fields to be
	 * used by VXLAN extensions if explicitly requested.
	 */
1310 1311 1312 1313
	if (vs->flags & VXLAN_F_REMCSUM_RX)
		if (!vxlan_remcsum(&unparsed, skb, vs->flags))
			goto drop;
	if (vs->flags & VXLAN_F_GBP)
1314
		vxlan_parse_gbp_hdr(&unparsed, skb, vs->flags, md);
T
Thomas Graf 已提交
1315

1316
	if (unparsed.vx_flags || unparsed.vx_vni) {
1317 1318 1319 1320
		/* If there are any unprocessed flags remaining treat
		 * this as a malformed packet. This behavior diverges from
		 * VXLAN RFC (RFC7348) which stipulates that bits in reserved
		 * in reserved fields are to be ignored. The approach here
1321
		 * maintains compatibility with previous stack code, and also
1322 1323 1324
		 * is more robust and provides a little more security in
		 * adding extensions to VXLAN.
		 */
J
Jiri Benc 已提交
1325
		goto drop;
1326 1327
	}

1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
	if (!vxlan_set_mac(vxlan, vs, skb))
		goto drop;

	oiph = skb_network_header(skb);
	skb_reset_network_header(skb);

	if (!vxlan_ecn_decapsulate(vs, oiph, skb)) {
		++vxlan->dev->stats.rx_frame_errors;
		++vxlan->dev->stats.rx_errors;
		goto drop;
	}

	stats = this_cpu_ptr(vxlan->dev->tstats);
	u64_stats_update_begin(&stats->syncp);
	stats->rx_packets++;
	stats->rx_bytes += skb->len;
	u64_stats_update_end(&stats->syncp);

	gro_cells_receive(&vxlan->gro_cells, skb);
P
Pravin B Shelar 已提交
1347 1348 1349
	return 0;

drop:
J
Jiri Benc 已提交
1350 1351 1352
	/* Consume bad packet */
	kfree_skb(skb);
	return 0;
P
Pravin B Shelar 已提交
1353 1354
}

D
David Stevens 已提交
1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402
static int arp_reduce(struct net_device *dev, struct sk_buff *skb)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
	struct arphdr *parp;
	u8 *arpptr, *sha;
	__be32 sip, tip;
	struct neighbour *n;

	if (dev->flags & IFF_NOARP)
		goto out;

	if (!pskb_may_pull(skb, arp_hdr_len(dev))) {
		dev->stats.tx_dropped++;
		goto out;
	}
	parp = arp_hdr(skb);

	if ((parp->ar_hrd != htons(ARPHRD_ETHER) &&
	     parp->ar_hrd != htons(ARPHRD_IEEE802)) ||
	    parp->ar_pro != htons(ETH_P_IP) ||
	    parp->ar_op != htons(ARPOP_REQUEST) ||
	    parp->ar_hln != dev->addr_len ||
	    parp->ar_pln != 4)
		goto out;
	arpptr = (u8 *)parp + sizeof(struct arphdr);
	sha = arpptr;
	arpptr += dev->addr_len;	/* sha */
	memcpy(&sip, arpptr, sizeof(sip));
	arpptr += sizeof(sip);
	arpptr += dev->addr_len;	/* tha */
	memcpy(&tip, arpptr, sizeof(tip));

	if (ipv4_is_loopback(tip) ||
	    ipv4_is_multicast(tip))
		goto out;

	n = neigh_lookup(&arp_tbl, &tip, dev);

	if (n) {
		struct vxlan_fdb *f;
		struct sk_buff	*reply;

		if (!(n->nud_state & NUD_CONNECTED)) {
			neigh_release(n);
			goto out;
		}

		f = vxlan_find_mac(vxlan, n->ha);
C
Cong Wang 已提交
1403
		if (f && vxlan_addr_any(&(first_remote_rcu(f)->remote_ip))) {
D
David Stevens 已提交
1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
			/* bridge-local neighbor */
			neigh_release(n);
			goto out;
		}

		reply = arp_create(ARPOP_REPLY, ETH_P_ARP, sip, dev, tip, sha,
				n->ha, sha);

		neigh_release(n);

1414 1415 1416
		if (reply == NULL)
			goto out;

D
David Stevens 已提交
1417 1418 1419 1420 1421 1422 1423
		skb_reset_mac_header(reply);
		__skb_pull(reply, skb_network_offset(reply));
		reply->ip_summed = CHECKSUM_UNNECESSARY;
		reply->pkt_type = PACKET_HOST;

		if (netif_rx_ni(reply) == NET_RX_DROP)
			dev->stats.rx_dropped++;
C
Cong Wang 已提交
1424 1425 1426
	} else if (vxlan->flags & VXLAN_F_L3MISS) {
		union vxlan_addr ipa = {
			.sin.sin_addr.s_addr = tip,
1427
			.sin.sin_family = AF_INET,
C
Cong Wang 已提交
1428 1429 1430 1431
		};

		vxlan_ip_miss(dev, &ipa);
	}
D
David Stevens 已提交
1432 1433 1434 1435 1436
out:
	consume_skb(skb);
	return NETDEV_TX_OK;
}

C
Cong Wang 已提交
1437
#if IS_ENABLED(CONFIG_IPV6)
1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462
static struct sk_buff *vxlan_na_create(struct sk_buff *request,
	struct neighbour *n, bool isrouter)
{
	struct net_device *dev = request->dev;
	struct sk_buff *reply;
	struct nd_msg *ns, *na;
	struct ipv6hdr *pip6;
	u8 *daddr;
	int na_olen = 8; /* opt hdr + ETH_ALEN for target */
	int ns_olen;
	int i, len;

	if (dev == NULL)
		return NULL;

	len = LL_RESERVED_SPACE(dev) + sizeof(struct ipv6hdr) +
		sizeof(*na) + na_olen + dev->needed_tailroom;
	reply = alloc_skb(len, GFP_ATOMIC);
	if (reply == NULL)
		return NULL;

	reply->protocol = htons(ETH_P_IPV6);
	reply->dev = dev;
	skb_reserve(reply, LL_RESERVED_SPACE(request->dev));
	skb_push(reply, sizeof(struct ethhdr));
1463
	skb_reset_mac_header(reply);
1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482

	ns = (struct nd_msg *)skb_transport_header(request);

	daddr = eth_hdr(request)->h_source;
	ns_olen = request->len - skb_transport_offset(request) - sizeof(*ns);
	for (i = 0; i < ns_olen-1; i += (ns->opt[i+1]<<3)) {
		if (ns->opt[i] == ND_OPT_SOURCE_LL_ADDR) {
			daddr = ns->opt + i + sizeof(struct nd_opt_hdr);
			break;
		}
	}

	/* Ethernet header */
	ether_addr_copy(eth_hdr(reply)->h_dest, daddr);
	ether_addr_copy(eth_hdr(reply)->h_source, n->ha);
	eth_hdr(reply)->h_proto = htons(ETH_P_IPV6);
	reply->protocol = htons(ETH_P_IPV6);

	skb_pull(reply, sizeof(struct ethhdr));
1483
	skb_reset_network_header(reply);
1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497
	skb_put(reply, sizeof(struct ipv6hdr));

	/* IPv6 header */

	pip6 = ipv6_hdr(reply);
	memset(pip6, 0, sizeof(struct ipv6hdr));
	pip6->version = 6;
	pip6->priority = ipv6_hdr(request)->priority;
	pip6->nexthdr = IPPROTO_ICMPV6;
	pip6->hop_limit = 255;
	pip6->daddr = ipv6_hdr(request)->saddr;
	pip6->saddr = *(struct in6_addr *)n->primary_key;

	skb_pull(reply, sizeof(struct ipv6hdr));
1498
	skb_reset_transport_header(reply);
1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525

	na = (struct nd_msg *)skb_put(reply, sizeof(*na) + na_olen);

	/* Neighbor Advertisement */
	memset(na, 0, sizeof(*na)+na_olen);
	na->icmph.icmp6_type = NDISC_NEIGHBOUR_ADVERTISEMENT;
	na->icmph.icmp6_router = isrouter;
	na->icmph.icmp6_override = 1;
	na->icmph.icmp6_solicited = 1;
	na->target = ns->target;
	ether_addr_copy(&na->opt[2], n->ha);
	na->opt[0] = ND_OPT_TARGET_LL_ADDR;
	na->opt[1] = na_olen >> 3;

	na->icmph.icmp6_cksum = csum_ipv6_magic(&pip6->saddr,
		&pip6->daddr, sizeof(*na)+na_olen, IPPROTO_ICMPV6,
		csum_partial(na, sizeof(*na)+na_olen, 0));

	pip6->payload_len = htons(sizeof(*na)+na_olen);

	skb_push(reply, sizeof(struct ipv6hdr));

	reply->ip_summed = CHECKSUM_UNNECESSARY;

	return reply;
}

C
Cong Wang 已提交
1526 1527 1528
static int neigh_reduce(struct net_device *dev, struct sk_buff *skb)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
1529
	struct nd_msg *msg;
C
Cong Wang 已提交
1530 1531
	const struct ipv6hdr *iphdr;
	const struct in6_addr *saddr, *daddr;
1532 1533
	struct neighbour *n;
	struct inet6_dev *in6_dev;
C
Cong Wang 已提交
1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547

	in6_dev = __in6_dev_get(dev);
	if (!in6_dev)
		goto out;

	iphdr = ipv6_hdr(skb);
	saddr = &iphdr->saddr;
	daddr = &iphdr->daddr;

	msg = (struct nd_msg *)skb_transport_header(skb);
	if (msg->icmph.icmp6_code != 0 ||
	    msg->icmph.icmp6_type != NDISC_NEIGHBOUR_SOLICITATION)
		goto out;

1548 1549 1550 1551 1552
	if (ipv6_addr_loopback(daddr) ||
	    ipv6_addr_is_multicast(&msg->target))
		goto out;

	n = neigh_lookup(ipv6_stub->nd_tbl, &msg->target, dev);
C
Cong Wang 已提交
1553 1554 1555

	if (n) {
		struct vxlan_fdb *f;
1556
		struct sk_buff *reply;
C
Cong Wang 已提交
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569

		if (!(n->nud_state & NUD_CONNECTED)) {
			neigh_release(n);
			goto out;
		}

		f = vxlan_find_mac(vxlan, n->ha);
		if (f && vxlan_addr_any(&(first_remote_rcu(f)->remote_ip))) {
			/* bridge-local neighbor */
			neigh_release(n);
			goto out;
		}

1570 1571 1572
		reply = vxlan_na_create(skb, n,
					!!(f ? f->flags & NTF_ROUTER : 0));

C
Cong Wang 已提交
1573
		neigh_release(n);
1574 1575 1576 1577 1578 1579 1580

		if (reply == NULL)
			goto out;

		if (netif_rx_ni(reply) == NET_RX_DROP)
			dev->stats.rx_dropped++;

C
Cong Wang 已提交
1581
	} else if (vxlan->flags & VXLAN_F_L3MISS) {
1582 1583
		union vxlan_addr ipa = {
			.sin6.sin6_addr = msg->target,
1584
			.sin6.sin6_family = AF_INET6,
1585 1586
		};

C
Cong Wang 已提交
1587 1588 1589 1590 1591 1592 1593 1594 1595
		vxlan_ip_miss(dev, &ipa);
	}

out:
	consume_skb(skb);
	return NETDEV_TX_OK;
}
#endif

D
David Stevens 已提交
1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606
static bool route_shortcircuit(struct net_device *dev, struct sk_buff *skb)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
	struct neighbour *n;

	if (is_multicast_ether_addr(eth_hdr(skb)->h_dest))
		return false;

	n = NULL;
	switch (ntohs(eth_hdr(skb)->h_proto)) {
	case ETH_P_IP:
1607 1608 1609
	{
		struct iphdr *pip;

D
David Stevens 已提交
1610 1611 1612 1613
		if (!pskb_may_pull(skb, sizeof(struct iphdr)))
			return false;
		pip = ip_hdr(skb);
		n = neigh_lookup(&arp_tbl, &pip->daddr, dev);
C
Cong Wang 已提交
1614 1615 1616
		if (!n && (vxlan->flags & VXLAN_F_L3MISS)) {
			union vxlan_addr ipa = {
				.sin.sin_addr.s_addr = pip->daddr,
1617
				.sin.sin_family = AF_INET,
C
Cong Wang 已提交
1618 1619 1620 1621 1622 1623
			};

			vxlan_ip_miss(dev, &ipa);
			return false;
		}

D
David Stevens 已提交
1624
		break;
1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637
	}
#if IS_ENABLED(CONFIG_IPV6)
	case ETH_P_IPV6:
	{
		struct ipv6hdr *pip6;

		if (!pskb_may_pull(skb, sizeof(struct ipv6hdr)))
			return false;
		pip6 = ipv6_hdr(skb);
		n = neigh_lookup(ipv6_stub->nd_tbl, &pip6->daddr, dev);
		if (!n && (vxlan->flags & VXLAN_F_L3MISS)) {
			union vxlan_addr ipa = {
				.sin6.sin6_addr = pip6->daddr,
1638
				.sin6.sin6_family = AF_INET6,
1639 1640 1641 1642 1643 1644 1645 1646 1647
			};

			vxlan_ip_miss(dev, &ipa);
			return false;
		}

		break;
	}
#endif
D
David Stevens 已提交
1648 1649 1650 1651 1652 1653 1654
	default:
		return false;
	}

	if (n) {
		bool diff;

1655
		diff = !ether_addr_equal(eth_hdr(skb)->h_dest, n->ha);
D
David Stevens 已提交
1656 1657 1658 1659 1660 1661 1662
		if (diff) {
			memcpy(eth_hdr(skb)->h_source, eth_hdr(skb)->h_dest,
				dev->addr_len);
			memcpy(eth_hdr(skb)->h_dest, n->ha, dev->addr_len);
		}
		neigh_release(n);
		return diff;
C
Cong Wang 已提交
1663 1664
	}

D
David Stevens 已提交
1665 1666 1667
	return false;
}

1668
static void vxlan_build_gbp_hdr(struct vxlanhdr *vxh, u32 vxflags,
T
Thomas Graf 已提交
1669 1670 1671 1672
				struct vxlan_metadata *md)
{
	struct vxlanhdr_gbp *gbp;

1673 1674 1675
	if (!md->gbp)
		return;

T
Thomas Graf 已提交
1676
	gbp = (struct vxlanhdr_gbp *)vxh;
1677
	vxh->vx_flags |= VXLAN_HF_GBP;
T
Thomas Graf 已提交
1678 1679 1680 1681 1682 1683 1684 1685 1686 1687

	if (md->gbp & VXLAN_GBP_DONT_LEARN)
		gbp->dont_learn = 1;

	if (md->gbp & VXLAN_GBP_POLICY_APPLIED)
		gbp->policy_applied = 1;

	gbp->policy_id = htons(md->gbp & VXLAN_GBP_ID_MASK);
}

1688 1689 1690
static int vxlan_build_skb(struct sk_buff *skb, struct dst_entry *dst,
			   int iphdr_len, __be32 vni,
			   struct vxlan_metadata *md, u32 vxflags,
1691
			   bool udp_sum)
C
Cong Wang 已提交
1692 1693 1694 1695
{
	struct vxlanhdr *vxh;
	int min_headroom;
	int err;
T
Tom Herbert 已提交
1696 1697
	int type = udp_sum ? SKB_GSO_UDP_TUNNEL_CSUM : SKB_GSO_UDP_TUNNEL;

1698
	if ((vxflags & VXLAN_F_REMCSUM_TX) &&
T
Tom Herbert 已提交
1699 1700 1701 1702 1703 1704
	    skb->ip_summed == CHECKSUM_PARTIAL) {
		int csum_start = skb_checksum_start_offset(skb);

		if (csum_start <= VXLAN_MAX_REMCSUM_START &&
		    !(csum_start & VXLAN_RCO_SHIFT_MASK) &&
		    (skb->csum_offset == offsetof(struct udphdr, check) ||
1705
		     skb->csum_offset == offsetof(struct tcphdr, check)))
T
Tom Herbert 已提交
1706 1707
			type |= SKB_GSO_TUNNEL_REMCSUM;
	}
C
Cong Wang 已提交
1708 1709

	min_headroom = LL_RESERVED_SPACE(dst->dev) + dst->header_len
1710
			+ VXLAN_HLEN + iphdr_len
1711
			+ (skb_vlan_tag_present(skb) ? VLAN_HLEN : 0);
1712 1713 1714

	/* Need space for new headers (invalidates iph ptr) */
	err = skb_cow_head(skb, min_headroom);
P
Pravin B Shelar 已提交
1715 1716
	if (unlikely(err)) {
		kfree_skb(skb);
1717
		return err;
P
Pravin B Shelar 已提交
1718
	}
1719

1720 1721 1722
	skb = vlan_hwaccel_push_inside(skb);
	if (WARN_ON(!skb))
		return -ENOMEM;
1723

1724
	skb = iptunnel_handle_offloads(skb, type);
1725 1726 1727
	if (IS_ERR(skb))
		return PTR_ERR(skb);

1728
	vxh = (struct vxlanhdr *) __skb_push(skb, sizeof(*vxh));
1729 1730
	vxh->vx_flags = VXLAN_HF_VNI;
	vxh->vx_vni = vxlan_vni_field(vni);
1731

T
Tom Herbert 已提交
1732
	if (type & SKB_GSO_TUNNEL_REMCSUM) {
1733
		unsigned int start;
T
Tom Herbert 已提交
1734

1735 1736 1737
		start = skb_checksum_start_offset(skb) - sizeof(struct vxlanhdr);
		vxh->vx_vni |= vxlan_compute_rco(start, skb->csum_offset);
		vxh->vx_flags |= VXLAN_HF_RCO;
T
Tom Herbert 已提交
1738 1739 1740 1741 1742 1743 1744

		if (!skb_is_gso(skb)) {
			skb->ip_summed = CHECKSUM_NONE;
			skb->encapsulation = 0;
		}
	}

1745 1746
	if (vxflags & VXLAN_F_GBP)
		vxlan_build_gbp_hdr(vxh, vxflags, md);
T
Thomas Graf 已提交
1747

1748
	skb_set_inner_protocol(skb, htons(ETH_P_TEB));
1749
	return 0;
1750 1751
}

1752 1753
static struct rtable *vxlan_get_route(struct vxlan_dev *vxlan,
				      struct sk_buff *skb, int oif, u8 tos,
1754 1755
				      __be32 daddr, __be32 *saddr,
				      struct dst_cache *dst_cache,
1756
				      const struct ip_tunnel_info *info)
1757
{
1758
	bool use_cache = ip_tunnel_dst_cache_usable(skb, info);
1759 1760 1761
	struct rtable *rt = NULL;
	struct flowi4 fl4;

1762 1763 1764
	if (tos && !info)
		use_cache = false;
	if (use_cache) {
1765 1766 1767 1768 1769
		rt = dst_cache_get_ip4(dst_cache, saddr);
		if (rt)
			return rt;
	}

1770 1771 1772 1773 1774 1775 1776 1777 1778
	memset(&fl4, 0, sizeof(fl4));
	fl4.flowi4_oif = oif;
	fl4.flowi4_tos = RT_TOS(tos);
	fl4.flowi4_mark = skb->mark;
	fl4.flowi4_proto = IPPROTO_UDP;
	fl4.daddr = daddr;
	fl4.saddr = vxlan->cfg.saddr.sin.sin_addr.s_addr;

	rt = ip_route_output_key(vxlan->net, &fl4);
1779
	if (!IS_ERR(rt)) {
1780
		*saddr = fl4.saddr;
1781 1782 1783
		if (use_cache)
			dst_cache_set_ip4(dst_cache, &rt->dst, fl4.saddr);
	}
1784 1785 1786
	return rt;
}

1787 1788
#if IS_ENABLED(CONFIG_IPV6)
static struct dst_entry *vxlan6_get_route(struct vxlan_dev *vxlan,
1789
					  struct sk_buff *skb, int oif, u8 tos,
1790
					  __be32 label,
1791
					  const struct in6_addr *daddr,
1792
					  struct in6_addr *saddr,
1793 1794
					  struct dst_cache *dst_cache,
					  const struct ip_tunnel_info *info)
1795
{
1796
	bool use_cache = ip_tunnel_dst_cache_usable(skb, info);
1797 1798 1799 1800
	struct dst_entry *ndst;
	struct flowi6 fl6;
	int err;

1801 1802
	if (tos && !info)
		use_cache = false;
1803
	if (use_cache) {
1804 1805 1806 1807 1808
		ndst = dst_cache_get_ip6(dst_cache, saddr);
		if (ndst)
			return ndst;
	}

1809 1810 1811 1812
	memset(&fl6, 0, sizeof(fl6));
	fl6.flowi6_oif = oif;
	fl6.daddr = *daddr;
	fl6.saddr = vxlan->cfg.saddr.sin6.sin6_addr;
1813
	fl6.flowlabel = ip6_make_flowinfo(RT_TOS(tos), label);
1814 1815 1816 1817 1818 1819 1820 1821 1822 1823
	fl6.flowi6_mark = skb->mark;
	fl6.flowi6_proto = IPPROTO_UDP;

	err = ipv6_stub->ipv6_dst_lookup(vxlan->net,
					 vxlan->vn6_sock->sock->sk,
					 &ndst, &fl6);
	if (err < 0)
		return ERR_PTR(err);

	*saddr = fl6.saddr;
1824
	if (use_cache)
1825
		dst_cache_set_ip6(dst_cache, ndst, saddr);
1826 1827 1828 1829
	return ndst;
}
#endif

1830 1831 1832 1833
/* Bypass encapsulation if the destination is local */
static void vxlan_encap_bypass(struct sk_buff *skb, struct vxlan_dev *src_vxlan,
			       struct vxlan_dev *dst_vxlan)
{
1834
	struct pcpu_sw_netstats *tx_stats, *rx_stats;
C
Cong Wang 已提交
1835 1836
	union vxlan_addr loopback;
	union vxlan_addr *remote_ip = &dst_vxlan->default_dst.remote_ip;
1837 1838
	struct net_device *dev = skb->dev;
	int len = skb->len;
1839

1840 1841
	tx_stats = this_cpu_ptr(src_vxlan->dev->tstats);
	rx_stats = this_cpu_ptr(dst_vxlan->dev->tstats);
1842 1843 1844 1845 1846
	skb->pkt_type = PACKET_HOST;
	skb->encapsulation = 0;
	skb->dev = dst_vxlan->dev;
	__skb_pull(skb, skb_network_offset(skb));

C
Cong Wang 已提交
1847 1848 1849 1850 1851 1852 1853 1854 1855 1856
	if (remote_ip->sa.sa_family == AF_INET) {
		loopback.sin.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
		loopback.sa.sa_family =  AF_INET;
#if IS_ENABLED(CONFIG_IPV6)
	} else {
		loopback.sin6.sin6_addr = in6addr_loopback;
		loopback.sa.sa_family =  AF_INET6;
#endif
	}

1857
	if (dst_vxlan->flags & VXLAN_F_LEARN)
C
Cong Wang 已提交
1858
		vxlan_snoop(skb->dev, &loopback, eth_hdr(skb)->h_source);
1859 1860 1861

	u64_stats_update_begin(&tx_stats->syncp);
	tx_stats->tx_packets++;
1862
	tx_stats->tx_bytes += len;
1863 1864 1865 1866 1867
	u64_stats_update_end(&tx_stats->syncp);

	if (netif_rx(skb) == NET_RX_SUCCESS) {
		u64_stats_update_begin(&rx_stats->syncp);
		rx_stats->rx_packets++;
1868
		rx_stats->rx_bytes += len;
1869 1870
		u64_stats_update_end(&rx_stats->syncp);
	} else {
1871
		dev->stats.rx_dropped++;
1872 1873 1874
	}
}

1875 1876
static void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
			   struct vxlan_rdst *rdst, bool did_rsc)
S
stephen hemminger 已提交
1877
{
1878
	struct dst_cache *dst_cache;
1879
	struct ip_tunnel_info *info;
S
stephen hemminger 已提交
1880
	struct vxlan_dev *vxlan = netdev_priv(dev);
1881
	struct sock *sk;
C
Cong Wang 已提交
1882
	struct rtable *rt = NULL;
S
stephen hemminger 已提交
1883
	const struct iphdr *old_iph;
C
Cong Wang 已提交
1884
	union vxlan_addr *dst;
T
Thomas Graf 已提交
1885 1886 1887
	union vxlan_addr remote_ip;
	struct vxlan_metadata _md;
	struct vxlan_metadata *md = &_md;
C
Cong Wang 已提交
1888
	__be16 src_port = 0, dst_port;
1889
	__be32 vni, label;
S
stephen hemminger 已提交
1890 1891
	__be16 df = 0;
	__u8 tos, ttl;
1892
	int err;
T
Thomas Graf 已提交
1893
	u32 flags = vxlan->flags;
1894
	bool udp_sum = false;
1895
	bool xnet = !net_eq(vxlan->net, dev_net(vxlan->dev));
S
stephen hemminger 已提交
1896

1897
	info = skb_tunnel_info(skb);
1898

T
Thomas Graf 已提交
1899
	if (rdst) {
1900
		dst_port = rdst->remote_port ? rdst->remote_port : vxlan->cfg.dst_port;
T
Thomas Graf 已提交
1901 1902
		vni = rdst->remote_vni;
		dst = &rdst->remote_ip;
1903
		dst_cache = &rdst->dst_cache;
T
Thomas Graf 已提交
1904 1905 1906 1907 1908 1909
	} else {
		if (!info) {
			WARN_ONCE(1, "%s: Missing encapsulation instructions\n",
				  dev->name);
			goto drop;
		}
1910
		dst_port = info->key.tp_dst ? : vxlan->cfg.dst_port;
1911
		vni = vxlan_tun_id_to_vni(info->key.tun_id);
1912 1913
		remote_ip.sa.sa_family = ip_tunnel_info_af(info);
		if (remote_ip.sa.sa_family == AF_INET)
1914 1915 1916
			remote_ip.sin.sin_addr.s_addr = info->key.u.ipv4.dst;
		else
			remote_ip.sin6.sin6_addr = info->key.u.ipv6.dst;
T
Thomas Graf 已提交
1917
		dst = &remote_ip;
1918
		dst_cache = &info->dst_cache;
T
Thomas Graf 已提交
1919
	}
D
David Stevens 已提交
1920

C
Cong Wang 已提交
1921
	if (vxlan_addr_any(dst)) {
D
David Stevens 已提交
1922 1923
		if (did_rsc) {
			/* short-circuited back to local bridge */
1924
			vxlan_encap_bypass(skb, vxlan, vxlan);
1925
			return;
D
David Stevens 已提交
1926
		}
1927
		goto drop;
D
David Stevens 已提交
1928
	}
1929

S
stephen hemminger 已提交
1930 1931
	old_iph = ip_hdr(skb);

1932
	ttl = vxlan->cfg.ttl;
C
Cong Wang 已提交
1933
	if (!ttl && vxlan_addr_multicast(dst))
S
stephen hemminger 已提交
1934 1935
		ttl = 1;

1936
	tos = vxlan->cfg.tos;
S
stephen hemminger 已提交
1937
	if (tos == 1)
1938
		tos = ip_tunnel_get_dsfield(old_iph, skb);
S
stephen hemminger 已提交
1939

1940
	label = vxlan->cfg.label;
1941 1942
	src_port = udp_flow_src_port(dev_net(dev), skb, vxlan->cfg.port_min,
				     vxlan->cfg.port_max, true);
S
stephen hemminger 已提交
1943

1944 1945 1946
	if (info) {
		ttl = info->key.ttl;
		tos = info->key.tos;
1947
		label = info->key.label;
1948
		udp_sum = !!(info->key.tun_flags & TUNNEL_CSUM);
1949 1950

		if (info->options_len)
1951
			md = ip_tunnel_info_opts(info);
1952 1953 1954 1955
	} else {
		md->gbp = skb->mark;
	}

C
Cong Wang 已提交
1956
	if (dst->sa.sa_family == AF_INET) {
1957 1958
		__be32 saddr;

1959 1960 1961 1962
		if (!vxlan->vn4_sock)
			goto drop;
		sk = vxlan->vn4_sock->sock->sk;

1963 1964
		rt = vxlan_get_route(vxlan, skb,
				     rdst ? rdst->remote_ifindex : 0, tos,
1965
				     dst->sin.sin_addr.s_addr, &saddr,
1966
				     dst_cache, info);
C
Cong Wang 已提交
1967 1968 1969 1970 1971 1972
		if (IS_ERR(rt)) {
			netdev_dbg(dev, "no route to %pI4\n",
				   &dst->sin.sin_addr.s_addr);
			dev->stats.tx_carrier_errors++;
			goto tx_error;
		}
S
stephen hemminger 已提交
1973

C
Cong Wang 已提交
1974 1975 1976 1977
		if (rt->dst.dev == dev) {
			netdev_dbg(dev, "circular route to %pI4\n",
				   &dst->sin.sin_addr.s_addr);
			dev->stats.collisions++;
1978
			goto rt_tx_error;
C
Cong Wang 已提交
1979 1980 1981 1982 1983 1984 1985 1986
		}

		/* Bypass encapsulation if the destination is local */
		if (rt->rt_flags & RTCF_LOCAL &&
		    !(rt->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))) {
			struct vxlan_dev *dst_vxlan;

			ip_rt_put(rt);
1987
			dst_vxlan = vxlan_find_vni(vxlan->net, vni,
1988 1989
						   dst->sa.sa_family, dst_port,
						   vxlan->flags);
C
Cong Wang 已提交
1990 1991 1992 1993 1994 1995
			if (!dst_vxlan)
				goto tx_error;
			vxlan_encap_bypass(skb, vxlan, dst_vxlan);
			return;
		}

1996 1997 1998 1999 2000
		if (!info)
			udp_sum = !(flags & VXLAN_F_UDP_ZERO_CSUM_TX);
		else if (info->key.tun_flags & TUNNEL_DONT_FRAGMENT)
			df = htons(IP_DF);

C
Cong Wang 已提交
2001 2002
		tos = ip_tunnel_ecn_encap(tos, old_iph, skb);
		ttl = ttl ? : ip4_dst_hoplimit(&rt->dst);
2003
		err = vxlan_build_skb(skb, &rt->dst, sizeof(struct iphdr),
2004
				      vni, md, flags, udp_sum);
2005 2006 2007 2008 2009 2010
		if (err < 0)
			goto xmit_tx_error;

		udp_tunnel_xmit_skb(rt, sk, skb, saddr,
				    dst->sin.sin_addr.s_addr, tos, ttl, df,
				    src_port, dst_port, xnet, !udp_sum);
C
Cong Wang 已提交
2011 2012 2013
#if IS_ENABLED(CONFIG_IPV6)
	} else {
		struct dst_entry *ndst;
2014
		struct in6_addr saddr;
J
Jiri Benc 已提交
2015
		u32 rt6i_flags;
C
Cong Wang 已提交
2016

2017 2018 2019 2020
		if (!vxlan->vn6_sock)
			goto drop;
		sk = vxlan->vn6_sock->sock->sk;

2021
		ndst = vxlan6_get_route(vxlan, skb,
2022
					rdst ? rdst->remote_ifindex : 0, tos,
2023
					label, &dst->sin6.sin6_addr, &saddr,
2024
					dst_cache, info);
2025
		if (IS_ERR(ndst)) {
C
Cong Wang 已提交
2026 2027 2028
			netdev_dbg(dev, "no route to %pI6\n",
				   &dst->sin6.sin6_addr);
			dev->stats.tx_carrier_errors++;
2029
			goto tx_error;
C
Cong Wang 已提交
2030
		}
S
stephen hemminger 已提交
2031

C
Cong Wang 已提交
2032 2033 2034 2035 2036 2037 2038
		if (ndst->dev == dev) {
			netdev_dbg(dev, "circular route to %pI6\n",
				   &dst->sin6.sin6_addr);
			dst_release(ndst);
			dev->stats.collisions++;
			goto tx_error;
		}
2039

C
Cong Wang 已提交
2040
		/* Bypass encapsulation if the destination is local */
J
Jiri Benc 已提交
2041 2042 2043
		rt6i_flags = ((struct rt6_info *)ndst)->rt6i_flags;
		if (rt6i_flags & RTF_LOCAL &&
		    !(rt6i_flags & (RTCF_BROADCAST | RTCF_MULTICAST))) {
C
Cong Wang 已提交
2044 2045 2046
			struct vxlan_dev *dst_vxlan;

			dst_release(ndst);
2047
			dst_vxlan = vxlan_find_vni(vxlan->net, vni,
2048 2049
						   dst->sa.sa_family, dst_port,
						   vxlan->flags);
C
Cong Wang 已提交
2050 2051 2052 2053 2054
			if (!dst_vxlan)
				goto tx_error;
			vxlan_encap_bypass(skb, vxlan, dst_vxlan);
			return;
		}
2055

2056 2057
		if (!info)
			udp_sum = !(flags & VXLAN_F_UDP_ZERO_CSUM6_TX);
2058

2059
		tos = ip_tunnel_ecn_encap(tos, old_iph, skb);
C
Cong Wang 已提交
2060
		ttl = ttl ? : ip6_dst_hoplimit(ndst);
2061 2062
		skb_scrub_packet(skb, xnet);
		err = vxlan_build_skb(skb, ndst, sizeof(struct ipv6hdr),
2063
				      vni, md, flags, udp_sum);
2064 2065 2066 2067 2068
		if (err < 0) {
			dst_release(ndst);
			return;
		}
		udp_tunnel6_xmit_skb(ndst, sk, skb, dev,
2069 2070
				     &saddr, &dst->sin6.sin6_addr, tos, ttl,
				     label, src_port, dst_port, !udp_sum);
C
Cong Wang 已提交
2071 2072
#endif
	}
2073

2074
	return;
S
stephen hemminger 已提交
2075 2076 2077 2078 2079

drop:
	dev->stats.tx_dropped++;
	goto tx_free;

2080 2081 2082
xmit_tx_error:
	/* skb is already freed. */
	skb = NULL;
2083 2084
rt_tx_error:
	ip_rt_put(rt);
S
stephen hemminger 已提交
2085 2086 2087 2088 2089 2090
tx_error:
	dev->stats.tx_errors++;
tx_free:
	dev_kfree_skb(skb);
}

2091 2092 2093 2094 2095 2096 2097 2098 2099
/* Transmit local packets over Vxlan
 *
 * Outer IP header inherits ECN and DF from inner header.
 * Outer UDP destination is the VXLAN assigned port.
 *           source port is based on hash of flow
 */
static netdev_tx_t vxlan_xmit(struct sk_buff *skb, struct net_device *dev)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
2100
	const struct ip_tunnel_info *info;
2101 2102
	struct ethhdr *eth;
	bool did_rsc = false;
2103
	struct vxlan_rdst *rdst, *fdst = NULL;
2104 2105
	struct vxlan_fdb *f;

2106
	info = skb_tunnel_info(skb);
2107

2108 2109
	skb_reset_mac_header(skb);

2110 2111 2112 2113 2114 2115 2116 2117 2118 2119
	if (vxlan->flags & VXLAN_F_COLLECT_METADATA) {
		if (info && info->mode & IP_TUNNEL_INFO_TX)
			vxlan_xmit_one(skb, dev, NULL, false);
		else
			kfree_skb(skb);
		return NETDEV_TX_OK;
	}

	if (vxlan->flags & VXLAN_F_PROXY) {
		eth = eth_hdr(skb);
C
Cong Wang 已提交
2120 2121 2122 2123
		if (ntohs(eth->h_proto) == ETH_P_ARP)
			return arp_reduce(dev, skb);
#if IS_ENABLED(CONFIG_IPV6)
		else if (ntohs(eth->h_proto) == ETH_P_IPV6 &&
2124 2125
			 pskb_may_pull(skb, sizeof(struct ipv6hdr)
				       + sizeof(struct nd_msg)) &&
C
Cong Wang 已提交
2126 2127 2128 2129 2130 2131 2132 2133 2134 2135
			 ipv6_hdr(skb)->nexthdr == IPPROTO_ICMPV6) {
				struct nd_msg *msg;

				msg = (struct nd_msg *)skb_transport_header(skb);
				if (msg->icmph.icmp6_code == 0 &&
				    msg->icmph.icmp6_type == NDISC_NEIGHBOUR_SOLICITATION)
					return neigh_reduce(dev, skb);
		}
#endif
	}
2136

2137
	eth = eth_hdr(skb);
2138
	f = vxlan_find_mac(vxlan, eth->h_dest);
2139 2140 2141
	did_rsc = false;

	if (f && (f->flags & NTF_ROUTER) && (vxlan->flags & VXLAN_F_RSC) &&
2142 2143
	    (ntohs(eth->h_proto) == ETH_P_IP ||
	     ntohs(eth->h_proto) == ETH_P_IPV6)) {
2144 2145 2146 2147 2148
		did_rsc = route_shortcircuit(dev, skb);
		if (did_rsc)
			f = vxlan_find_mac(vxlan, eth->h_dest);
	}

2149
	if (f == NULL) {
2150 2151 2152 2153 2154 2155 2156
		f = vxlan_find_mac(vxlan, all_zeros_mac);
		if (f == NULL) {
			if ((vxlan->flags & VXLAN_F_L2MISS) &&
			    !is_multicast_ether_addr(eth->h_dest))
				vxlan_fdb_miss(vxlan, eth->h_dest);

			dev->stats.tx_dropped++;
2157
			kfree_skb(skb);
2158 2159 2160
			return NETDEV_TX_OK;
		}
	}
2161

2162 2163
	list_for_each_entry_rcu(rdst, &f->remotes, list) {
		struct sk_buff *skb1;
2164

2165 2166 2167 2168
		if (!fdst) {
			fdst = rdst;
			continue;
		}
2169 2170 2171
		skb1 = skb_clone(skb, GFP_ATOMIC);
		if (skb1)
			vxlan_xmit_one(skb1, dev, rdst, did_rsc);
2172 2173
	}

2174 2175 2176 2177
	if (fdst)
		vxlan_xmit_one(skb, dev, fdst, did_rsc);
	else
		kfree_skb(skb);
2178
	return NETDEV_TX_OK;
2179 2180
}

S
stephen hemminger 已提交
2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192
/* Walk the forwarding table and purge stale entries */
static void vxlan_cleanup(unsigned long arg)
{
	struct vxlan_dev *vxlan = (struct vxlan_dev *) arg;
	unsigned long next_timer = jiffies + FDB_AGE_INTERVAL;
	unsigned int h;

	if (!netif_running(vxlan->dev))
		return;

	for (h = 0; h < FDB_HASH_SIZE; ++h) {
		struct hlist_node *p, *n;
2193 2194

		spin_lock_bh(&vxlan->hash_lock);
S
stephen hemminger 已提交
2195 2196 2197 2198 2199
		hlist_for_each_safe(p, n, &vxlan->fdb_head[h]) {
			struct vxlan_fdb *f
				= container_of(p, struct vxlan_fdb, hlist);
			unsigned long timeout;

2200
			if (f->state & NUD_PERMANENT)
S
stephen hemminger 已提交
2201 2202
				continue;

2203
			timeout = f->used + vxlan->cfg.age_interval * HZ;
S
stephen hemminger 已提交
2204 2205 2206 2207 2208 2209 2210 2211 2212
			if (time_before_eq(timeout, jiffies)) {
				netdev_dbg(vxlan->dev,
					   "garbage collect %pM\n",
					   f->eth_addr);
				f->state = NUD_STALE;
				vxlan_fdb_destroy(vxlan, f);
			} else if (time_before(timeout, next_timer))
				next_timer = timeout;
		}
2213
		spin_unlock_bh(&vxlan->hash_lock);
S
stephen hemminger 已提交
2214 2215 2216 2217 2218
	}

	mod_timer(&vxlan->age_timer, next_timer);
}

2219 2220
static void vxlan_vs_add_dev(struct vxlan_sock *vs, struct vxlan_dev *vxlan)
{
2221
	struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
2222
	__be32 vni = vxlan->default_dst.remote_vni;
2223

2224
	spin_lock(&vn->sock_lock);
2225
	hlist_add_head_rcu(&vxlan->hlist, vni_head(vs, vni));
2226
	spin_unlock(&vn->sock_lock);
2227 2228
}

S
stephen hemminger 已提交
2229 2230 2231
/* Setup stats when device is created */
static int vxlan_init(struct net_device *dev)
{
2232
	dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
2233
	if (!dev->tstats)
S
stephen hemminger 已提交
2234 2235 2236 2237 2238
		return -ENOMEM;

	return 0;
}

2239
static void vxlan_fdb_delete_default(struct vxlan_dev *vxlan)
2240 2241 2242 2243 2244 2245 2246 2247 2248 2249
{
	struct vxlan_fdb *f;

	spin_lock_bh(&vxlan->hash_lock);
	f = __vxlan_find_mac(vxlan, all_zeros_mac);
	if (f)
		vxlan_fdb_destroy(vxlan, f);
	spin_unlock_bh(&vxlan->hash_lock);
}

2250 2251 2252 2253
static void vxlan_uninit(struct net_device *dev)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);

2254
	vxlan_fdb_delete_default(vxlan);
2255

2256 2257 2258
	free_percpu(dev->tstats);
}

S
stephen hemminger 已提交
2259 2260 2261 2262
/* Start ageing timer and join group when device is brought up */
static int vxlan_open(struct net_device *dev)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
2263
	int ret;
2264

2265 2266 2267
	ret = vxlan_sock_add(vxlan);
	if (ret < 0)
		return ret;
S
stephen hemminger 已提交
2268

2269
	if (vxlan_addr_multicast(&vxlan->default_dst.remote_ip)) {
2270
		ret = vxlan_igmp_join(vxlan);
2271 2272
		if (ret == -EADDRINUSE)
			ret = 0;
2273
		if (ret) {
2274
			vxlan_sock_release(vxlan);
2275 2276
			return ret;
		}
S
stephen hemminger 已提交
2277 2278
	}

2279
	if (vxlan->cfg.age_interval)
S
stephen hemminger 已提交
2280 2281
		mod_timer(&vxlan->age_timer, jiffies + FDB_AGE_INTERVAL);

2282
	return ret;
S
stephen hemminger 已提交
2283 2284 2285 2286 2287
}

/* Purge the forwarding table */
static void vxlan_flush(struct vxlan_dev *vxlan)
{
2288
	unsigned int h;
S
stephen hemminger 已提交
2289 2290 2291 2292 2293 2294 2295

	spin_lock_bh(&vxlan->hash_lock);
	for (h = 0; h < FDB_HASH_SIZE; ++h) {
		struct hlist_node *p, *n;
		hlist_for_each_safe(p, n, &vxlan->fdb_head[h]) {
			struct vxlan_fdb *f
				= container_of(p, struct vxlan_fdb, hlist);
2296 2297 2298
			/* the all_zeros_mac entry is deleted at vxlan_uninit */
			if (!is_zero_ether_addr(f->eth_addr))
				vxlan_fdb_destroy(vxlan, f);
S
stephen hemminger 已提交
2299 2300 2301 2302 2303 2304 2305 2306 2307
		}
	}
	spin_unlock_bh(&vxlan->hash_lock);
}

/* Cleanup timer and forwarding table on shutdown */
static int vxlan_stop(struct net_device *dev)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
N
Nicolas Dichtel 已提交
2308
	struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
2309
	int ret = 0;
S
stephen hemminger 已提交
2310

2311
	if (vxlan_addr_multicast(&vxlan->default_dst.remote_ip) &&
2312
	    !vxlan_group_used(vn, vxlan))
2313
		ret = vxlan_igmp_leave(vxlan);
S
stephen hemminger 已提交
2314 2315 2316 2317

	del_timer_sync(&vxlan->age_timer);

	vxlan_flush(vxlan);
2318
	vxlan_sock_release(vxlan);
S
stephen hemminger 已提交
2319

2320
	return ret;
S
stephen hemminger 已提交
2321 2322 2323 2324 2325 2326 2327
}

/* Stub, nothing needs to be done. */
static void vxlan_set_multicast_list(struct net_device *dev)
{
}

D
David Wragg 已提交
2328 2329 2330
static int __vxlan_change_mtu(struct net_device *dev,
			      struct net_device *lowerdev,
			      struct vxlan_rdst *dst, int new_mtu, bool strict)
2331
{
D
David Wragg 已提交
2332
	int max_mtu = IP_MAX_MTU;
2333

D
David Wragg 已提交
2334 2335
	if (lowerdev)
		max_mtu = lowerdev->mtu;
2336 2337

	if (dst->remote_ip.sa.sa_family == AF_INET6)
D
David Wragg 已提交
2338
		max_mtu -= VXLAN6_HEADROOM;
2339
	else
D
David Wragg 已提交
2340
		max_mtu -= VXLAN_HEADROOM;
2341

D
David Wragg 已提交
2342
	if (new_mtu < 68)
2343 2344
		return -EINVAL;

D
David Wragg 已提交
2345 2346 2347 2348 2349 2350 2351
	if (new_mtu > max_mtu) {
		if (strict)
			return -EINVAL;

		new_mtu = max_mtu;
	}

2352 2353 2354 2355
	dev->mtu = new_mtu;
	return 0;
}

D
David Wragg 已提交
2356 2357 2358 2359 2360 2361 2362 2363 2364
static int vxlan_change_mtu(struct net_device *dev, int new_mtu)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
	struct vxlan_rdst *dst = &vxlan->default_dst;
	struct net_device *lowerdev = __dev_get_by_index(vxlan->net,
							 dst->remote_ifindex);
	return __vxlan_change_mtu(dev, lowerdev, dst, new_mtu, true);
}

2365 2366 2367 2368 2369 2370 2371 2372 2373 2374
static int vxlan_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
	struct ip_tunnel_info *info = skb_tunnel_info(skb);
	__be16 sport, dport;

	sport = udp_flow_src_port(dev_net(dev), skb, vxlan->cfg.port_min,
				  vxlan->cfg.port_max, true);
	dport = info->key.tp_dst ? : vxlan->cfg.dst_port;

2375
	if (ip_tunnel_info_af(info) == AF_INET) {
2376 2377
		struct rtable *rt;

2378 2379
		if (!vxlan->vn4_sock)
			return -EINVAL;
2380 2381
		rt = vxlan_get_route(vxlan, skb, 0, info->key.tos,
				     info->key.u.ipv4.dst,
2382
				     &info->key.u.ipv4.src, NULL, info);
2383 2384 2385
		if (IS_ERR(rt))
			return PTR_ERR(rt);
		ip_rt_put(rt);
2386 2387 2388 2389 2390 2391
	} else {
#if IS_ENABLED(CONFIG_IPV6)
		struct dst_entry *ndst;

		if (!vxlan->vn6_sock)
			return -EINVAL;
2392
		ndst = vxlan6_get_route(vxlan, skb, 0, info->key.tos,
2393
					info->key.label, &info->key.u.ipv6.dst,
2394
					&info->key.u.ipv6.src, NULL, info);
2395 2396 2397 2398 2399 2400 2401
		if (IS_ERR(ndst))
			return PTR_ERR(ndst);
		dst_release(ndst);
#else /* !CONFIG_IPV6 */
		return -EPFNOSUPPORT;
#endif
	}
2402 2403
	info->key.tp_src = sport;
	info->key.tp_dst = dport;
2404
	return 0;
2405 2406
}

2407
static const struct net_device_ops vxlan_netdev_ether_ops = {
S
stephen hemminger 已提交
2408
	.ndo_init		= vxlan_init,
2409
	.ndo_uninit		= vxlan_uninit,
S
stephen hemminger 已提交
2410 2411 2412
	.ndo_open		= vxlan_open,
	.ndo_stop		= vxlan_stop,
	.ndo_start_xmit		= vxlan_xmit,
2413
	.ndo_get_stats64	= ip_tunnel_get_stats64,
S
stephen hemminger 已提交
2414
	.ndo_set_rx_mode	= vxlan_set_multicast_list,
2415
	.ndo_change_mtu		= vxlan_change_mtu,
S
stephen hemminger 已提交
2416 2417 2418 2419 2420
	.ndo_validate_addr	= eth_validate_addr,
	.ndo_set_mac_address	= eth_mac_addr,
	.ndo_fdb_add		= vxlan_fdb_add,
	.ndo_fdb_del		= vxlan_fdb_delete,
	.ndo_fdb_dump		= vxlan_fdb_dump,
2421
	.ndo_fill_metadata_dst	= vxlan_fill_metadata_dst,
S
stephen hemminger 已提交
2422 2423 2424 2425 2426 2427 2428
};

/* Info for udev, that this is a virtual tunnel endpoint */
static struct device_type vxlan_type = {
	.name = "vxlan",
};

2429
/* Calls the ndo_add_vxlan_port of the caller in order to
J
Joseph Gasparakis 已提交
2430 2431
 * supply the listening VXLAN udp ports. Callers are expected
 * to implement the ndo_add_vxlan_port.
2432 2433 2434 2435 2436 2437 2438
 */
void vxlan_get_rx_port(struct net_device *dev)
{
	struct vxlan_sock *vs;
	struct net *net = dev_net(dev);
	struct vxlan_net *vn = net_generic(net, vxlan_net_id);
	sa_family_t sa_family;
J
Joseph Gasparakis 已提交
2439 2440
	__be16 port;
	unsigned int i;
2441 2442 2443

	spin_lock(&vn->sock_lock);
	for (i = 0; i < PORT_HASH_SIZE; ++i) {
J
Joseph Gasparakis 已提交
2444 2445
		hlist_for_each_entry_rcu(vs, &vn->sock_list[i], hlist) {
			port = inet_sk(vs->sock->sk)->inet_sport;
2446
			sa_family = vxlan_get_sk_family(vs);
2447 2448 2449 2450 2451 2452 2453 2454
			dev->netdev_ops->ndo_add_vxlan_port(dev, sa_family,
							    port);
		}
	}
	spin_unlock(&vn->sock_lock);
}
EXPORT_SYMBOL_GPL(vxlan_get_rx_port);

S
stephen hemminger 已提交
2455 2456 2457 2458
/* Initialize the device structure. */
static void vxlan_setup(struct net_device *dev)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
2459
	unsigned int h;
S
stephen hemminger 已提交
2460

2461
	dev->destructor = free_netdev;
S
stephen hemminger 已提交
2462 2463 2464
	SET_NETDEV_DEVTYPE(dev, &vxlan_type);

	dev->features	|= NETIF_F_LLTX;
2465
	dev->features	|= NETIF_F_SG | NETIF_F_HW_CSUM;
2466
	dev->features   |= NETIF_F_RXCSUM;
2467
	dev->features   |= NETIF_F_GSO_SOFTWARE;
2468

2469 2470
	dev->vlan_features = dev->features;
	dev->features |= NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_STAG_TX;
2471
	dev->hw_features |= NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_RXCSUM;
2472
	dev->hw_features |= NETIF_F_GSO_SOFTWARE;
2473
	dev->hw_features |= NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_STAG_TX;
2474
	netif_keep_dst(dev);
2475
	dev->priv_flags |= IFF_NO_QUEUE;
S
stephen hemminger 已提交
2476

2477
	INIT_LIST_HEAD(&vxlan->next);
S
stephen hemminger 已提交
2478 2479 2480 2481 2482 2483
	spin_lock_init(&vxlan->hash_lock);

	init_timer_deferrable(&vxlan->age_timer);
	vxlan->age_timer.function = vxlan_cleanup;
	vxlan->age_timer.data = (unsigned long) vxlan;

2484
	vxlan->cfg.dst_port = htons(vxlan_port);
2485

S
stephen hemminger 已提交
2486 2487
	vxlan->dev = dev;

T
Tom Herbert 已提交
2488 2489
	gro_cells_init(&vxlan->gro_cells, dev);

S
stephen hemminger 已提交
2490 2491 2492 2493
	for (h = 0; h < FDB_HASH_SIZE; ++h)
		INIT_HLIST_HEAD(&vxlan->fdb_head[h]);
}

2494 2495 2496 2497 2498 2499 2500 2501 2502
static void vxlan_ether_setup(struct net_device *dev)
{
	eth_hw_addr_random(dev);
	ether_setup(dev);
	dev->priv_flags &= ~IFF_TX_SKB_SHARING;
	dev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
	dev->netdev_ops = &vxlan_netdev_ether_ops;
}

S
stephen hemminger 已提交
2503 2504
static const struct nla_policy vxlan_policy[IFLA_VXLAN_MAX + 1] = {
	[IFLA_VXLAN_ID]		= { .type = NLA_U32 },
2505
	[IFLA_VXLAN_GROUP]	= { .len = FIELD_SIZEOF(struct iphdr, daddr) },
C
Cong Wang 已提交
2506
	[IFLA_VXLAN_GROUP6]	= { .len = sizeof(struct in6_addr) },
S
stephen hemminger 已提交
2507 2508
	[IFLA_VXLAN_LINK]	= { .type = NLA_U32 },
	[IFLA_VXLAN_LOCAL]	= { .len = FIELD_SIZEOF(struct iphdr, saddr) },
C
Cong Wang 已提交
2509
	[IFLA_VXLAN_LOCAL6]	= { .len = sizeof(struct in6_addr) },
S
stephen hemminger 已提交
2510 2511
	[IFLA_VXLAN_TOS]	= { .type = NLA_U8 },
	[IFLA_VXLAN_TTL]	= { .type = NLA_U8 },
2512
	[IFLA_VXLAN_LABEL]	= { .type = NLA_U32 },
S
stephen hemminger 已提交
2513 2514 2515
	[IFLA_VXLAN_LEARNING]	= { .type = NLA_U8 },
	[IFLA_VXLAN_AGEING]	= { .type = NLA_U32 },
	[IFLA_VXLAN_LIMIT]	= { .type = NLA_U32 },
2516
	[IFLA_VXLAN_PORT_RANGE] = { .len  = sizeof(struct ifla_vxlan_port_range) },
D
David Stevens 已提交
2517 2518 2519 2520
	[IFLA_VXLAN_PROXY]	= { .type = NLA_U8 },
	[IFLA_VXLAN_RSC]	= { .type = NLA_U8 },
	[IFLA_VXLAN_L2MISS]	= { .type = NLA_U8 },
	[IFLA_VXLAN_L3MISS]	= { .type = NLA_U8 },
2521
	[IFLA_VXLAN_COLLECT_METADATA]	= { .type = NLA_U8 },
2522
	[IFLA_VXLAN_PORT]	= { .type = NLA_U16 },
2523 2524 2525
	[IFLA_VXLAN_UDP_CSUM]	= { .type = NLA_U8 },
	[IFLA_VXLAN_UDP_ZERO_CSUM6_TX]	= { .type = NLA_U8 },
	[IFLA_VXLAN_UDP_ZERO_CSUM6_RX]	= { .type = NLA_U8 },
T
Tom Herbert 已提交
2526 2527
	[IFLA_VXLAN_REMCSUM_TX]	= { .type = NLA_U8 },
	[IFLA_VXLAN_REMCSUM_RX]	= { .type = NLA_U8 },
T
Thomas Graf 已提交
2528
	[IFLA_VXLAN_GBP]	= { .type = NLA_FLAG, },
2529
	[IFLA_VXLAN_REMCSUM_NOPARTIAL]	= { .type = NLA_FLAG },
S
stephen hemminger 已提交
2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554
};

static int vxlan_validate(struct nlattr *tb[], struct nlattr *data[])
{
	if (tb[IFLA_ADDRESS]) {
		if (nla_len(tb[IFLA_ADDRESS]) != ETH_ALEN) {
			pr_debug("invalid link address (not ethernet)\n");
			return -EINVAL;
		}

		if (!is_valid_ether_addr(nla_data(tb[IFLA_ADDRESS]))) {
			pr_debug("invalid all zero ethernet address\n");
			return -EADDRNOTAVAIL;
		}
	}

	if (!data)
		return -EINVAL;

	if (data[IFLA_VXLAN_ID]) {
		__u32 id = nla_get_u32(data[IFLA_VXLAN_ID]);
		if (id >= VXLAN_VID_MASK)
			return -ERANGE;
	}

2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565
	if (data[IFLA_VXLAN_PORT_RANGE]) {
		const struct ifla_vxlan_port_range *p
			= nla_data(data[IFLA_VXLAN_PORT_RANGE]);

		if (ntohs(p->high) < ntohs(p->low)) {
			pr_debug("port range %u .. %u not valid\n",
				 ntohs(p->low), ntohs(p->high));
			return -EINVAL;
		}
	}

S
stephen hemminger 已提交
2566 2567 2568
	return 0;
}

Y
Yan Burman 已提交
2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580
static void vxlan_get_drvinfo(struct net_device *netdev,
			      struct ethtool_drvinfo *drvinfo)
{
	strlcpy(drvinfo->version, VXLAN_VERSION, sizeof(drvinfo->version));
	strlcpy(drvinfo->driver, "vxlan", sizeof(drvinfo->driver));
}

static const struct ethtool_ops vxlan_ethtool_ops = {
	.get_drvinfo	= vxlan_get_drvinfo,
	.get_link	= ethtool_op_get_link,
};

2581 2582 2583
static void vxlan_del_work(struct work_struct *work)
{
	struct vxlan_sock *vs = container_of(work, struct vxlan_sock, del_work);
2584
	udp_tunnel_sock_release(vs->sock);
2585 2586 2587
	kfree_rcu(vs, rcu);
}

T
Tom Herbert 已提交
2588 2589
static struct socket *vxlan_create_sock(struct net *net, bool ipv6,
					__be16 port, u32 flags)
2590
{
C
Cong Wang 已提交
2591
	struct socket *sock;
T
Tom Herbert 已提交
2592 2593
	struct udp_port_cfg udp_conf;
	int err;
C
Cong Wang 已提交
2594

T
Tom Herbert 已提交
2595
	memset(&udp_conf, 0, sizeof(udp_conf));
C
Cong Wang 已提交
2596

T
Tom Herbert 已提交
2597 2598 2599
	if (ipv6) {
		udp_conf.family = AF_INET6;
		udp_conf.use_udp6_rx_checksums =
2600
		    !(flags & VXLAN_F_UDP_ZERO_CSUM6_RX);
2601
		udp_conf.ipv6_v6only = 1;
T
Tom Herbert 已提交
2602 2603
	} else {
		udp_conf.family = AF_INET;
C
Cong Wang 已提交
2604 2605
	}

T
Tom Herbert 已提交
2606
	udp_conf.local_udp_port = port;
2607

T
Tom Herbert 已提交
2608 2609 2610 2611
	/* Open UDP socket */
	err = udp_sock_create(net, &udp_conf, &sock);
	if (err < 0)
		return ERR_PTR(err);
C
Cong Wang 已提交
2612

Z
Zhi Yong Wu 已提交
2613
	return sock;
C
Cong Wang 已提交
2614 2615 2616
}

/* Create new listen socket if needed */
2617 2618
static struct vxlan_sock *vxlan_socket_create(struct net *net, bool ipv6,
					      __be16 port, u32 flags)
C
Cong Wang 已提交
2619 2620 2621 2622 2623
{
	struct vxlan_net *vn = net_generic(net, vxlan_net_id);
	struct vxlan_sock *vs;
	struct socket *sock;
	unsigned int h;
2624
	struct udp_tunnel_sock_cfg tunnel_cfg;
C
Cong Wang 已提交
2625

2626
	vs = kzalloc(sizeof(*vs), GFP_KERNEL);
C
Cong Wang 已提交
2627 2628 2629 2630 2631 2632 2633 2634
	if (!vs)
		return ERR_PTR(-ENOMEM);

	for (h = 0; h < VNI_HASH_SIZE; ++h)
		INIT_HLIST_HEAD(&vs->vni_list[h]);

	INIT_WORK(&vs->del_work, vxlan_del_work);

T
Tom Herbert 已提交
2635
	sock = vxlan_create_sock(net, ipv6, port, flags);
Z
Zhi Yong Wu 已提交
2636
	if (IS_ERR(sock)) {
2637 2638
		pr_info("Cannot bind port %d, err=%ld\n", ntohs(port),
			PTR_ERR(sock));
2639
		kfree(vs);
2640
		return ERR_CAST(sock);
2641
	}
C
Cong Wang 已提交
2642 2643

	vs->sock = sock;
2644
	atomic_set(&vs->refcnt, 1);
2645
	vs->flags = (flags & VXLAN_F_RCV_FLAGS);
2646

2647 2648 2649 2650 2651
	/* Initialize the vxlan udp offloads structure */
	vs->udp_offloads.port = port;
	vs->udp_offloads.callbacks.gro_receive  = vxlan_gro_receive;
	vs->udp_offloads.callbacks.gro_complete = vxlan_gro_complete;

2652 2653
	spin_lock(&vn->sock_lock);
	hlist_add_head_rcu(&vs->hlist, vs_head(net, port));
2654
	vxlan_notify_add_rx_port(vs);
2655
	spin_unlock(&vn->sock_lock);
2656 2657

	/* Mark socket as an encapsulation socket. */
2658 2659
	tunnel_cfg.sk_user_data = vs;
	tunnel_cfg.encap_type = 1;
2660
	tunnel_cfg.encap_rcv = vxlan_rcv;
2661 2662 2663
	tunnel_cfg.encap_destroy = NULL;

	setup_udp_tunnel_sock(net, sock, &tunnel_cfg);
C
Cong Wang 已提交
2664

2665 2666 2667
	return vs;
}

2668
static int __vxlan_sock_add(struct vxlan_dev *vxlan, bool ipv6)
2669
{
2670 2671
	struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
	struct vxlan_sock *vs = NULL;
2672

2673
	if (!vxlan->cfg.no_share) {
2674
		spin_lock(&vn->sock_lock);
2675 2676 2677
		vs = vxlan_find_sock(vxlan->net, ipv6 ? AF_INET6 : AF_INET,
				     vxlan->cfg.dst_port, vxlan->flags);
		if (vs && !atomic_add_unless(&vs->refcnt, 1, 0)) {
2678
			spin_unlock(&vn->sock_lock);
2679
			return -EBUSY;
2680 2681 2682
		}
		spin_unlock(&vn->sock_lock);
	}
2683
	if (!vs)
2684 2685
		vs = vxlan_socket_create(vxlan->net, ipv6,
					 vxlan->cfg.dst_port, vxlan->flags);
2686 2687
	if (IS_ERR(vs))
		return PTR_ERR(vs);
2688 2689 2690 2691 2692 2693
#if IS_ENABLED(CONFIG_IPV6)
	if (ipv6)
		vxlan->vn6_sock = vs;
	else
#endif
		vxlan->vn4_sock = vs;
2694 2695
	vxlan_vs_add_dev(vs, vxlan);
	return 0;
2696 2697
}

2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716
static int vxlan_sock_add(struct vxlan_dev *vxlan)
{
	bool ipv6 = vxlan->flags & VXLAN_F_IPV6;
	bool metadata = vxlan->flags & VXLAN_F_COLLECT_METADATA;
	int ret = 0;

	vxlan->vn4_sock = NULL;
#if IS_ENABLED(CONFIG_IPV6)
	vxlan->vn6_sock = NULL;
	if (ipv6 || metadata)
		ret = __vxlan_sock_add(vxlan, true);
#endif
	if (!ret && (!ipv6 || metadata))
		ret = __vxlan_sock_add(vxlan, false);
	if (ret < 0)
		vxlan_sock_release(vxlan);
	return ret;
}

2717 2718
static int vxlan_dev_configure(struct net *src_net, struct net_device *dev,
			       struct vxlan_config *conf)
S
stephen hemminger 已提交
2719
{
2720
	struct vxlan_net *vn = net_generic(src_net, vxlan_net_id);
2721
	struct vxlan_dev *vxlan = netdev_priv(dev), *tmp;
2722
	struct vxlan_rdst *dst = &vxlan->default_dst;
2723
	unsigned short needed_headroom = ETH_HLEN;
S
stephen hemminger 已提交
2724
	int err;
C
Cong Wang 已提交
2725
	bool use_ipv6 = false;
2726
	__be16 default_port = vxlan->cfg.dst_port;
2727
	struct net_device *lowerdev = NULL;
S
stephen hemminger 已提交
2728

2729 2730
	vxlan_ether_setup(dev);

2731
	vxlan->net = src_net;
N
Nicolas Dichtel 已提交
2732

2733
	dst->remote_vni = conf->vni;
S
stephen hemminger 已提交
2734

2735
	memcpy(&dst->remote_ip, &conf->remote_ip, sizeof(conf->remote_ip));
S
stephen hemminger 已提交
2736

2737 2738 2739
	/* Unless IPv6 is explicitly requested, assume IPv4 */
	if (!dst->remote_ip.sa.sa_family)
		dst->remote_ip.sa.sa_family = AF_INET;
C
Cong Wang 已提交
2740

2741
	if (dst->remote_ip.sa.sa_family == AF_INET6 ||
2742 2743 2744
	    vxlan->cfg.saddr.sa.sa_family == AF_INET6) {
		if (!IS_ENABLED(CONFIG_IPV6))
			return -EPFNOSUPPORT;
C
Cong Wang 已提交
2745
		use_ipv6 = true;
2746
		vxlan->flags |= VXLAN_F_IPV6;
2747
	}
S
stephen hemminger 已提交
2748

2749 2750 2751 2752 2753
	if (conf->label && !use_ipv6) {
		pr_info("label only supported in use with IPv6\n");
		return -EINVAL;
	}

2754
	if (conf->remote_ifindex) {
2755
		lowerdev = __dev_get_by_index(src_net, conf->remote_ifindex);
2756
		dst->remote_ifindex = conf->remote_ifindex;
2757 2758

		if (!lowerdev) {
2759
			pr_info("ifindex %d does not exist\n", dst->remote_ifindex);
2760 2761
			return -ENODEV;
		}
S
stephen hemminger 已提交
2762

C
Cong Wang 已提交
2763 2764 2765 2766 2767 2768 2769 2770 2771 2772
#if IS_ENABLED(CONFIG_IPV6)
		if (use_ipv6) {
			struct inet6_dev *idev = __in6_dev_get(lowerdev);
			if (idev && idev->cnf.disable_ipv6) {
				pr_info("IPv6 is disabled via sysctl\n");
				return -EPERM;
			}
		}
#endif

2773
		if (!conf->mtu)
C
Cong Wang 已提交
2774
			dev->mtu = lowerdev->mtu - (use_ipv6 ? VXLAN6_HEADROOM : VXLAN_HEADROOM);
2775

2776
		needed_headroom = lowerdev->hard_header_len;
J
Jiri Benc 已提交
2777
	}
S
stephen hemminger 已提交
2778

2779 2780 2781 2782 2783 2784
	if (conf->mtu) {
		err = __vxlan_change_mtu(dev, lowerdev, dst, conf->mtu, false);
		if (err)
			return err;
	}

2785 2786 2787 2788 2789 2790
	if (use_ipv6 || conf->flags & VXLAN_F_COLLECT_METADATA)
		needed_headroom += VXLAN6_HEADROOM;
	else
		needed_headroom += VXLAN_HEADROOM;
	dev->needed_headroom = needed_headroom;

2791 2792 2793 2794 2795 2796 2797 2798
	memcpy(&vxlan->cfg, conf, sizeof(*conf));
	if (!vxlan->cfg.dst_port)
		vxlan->cfg.dst_port = default_port;
	vxlan->flags |= conf->flags;

	if (!vxlan->cfg.age_interval)
		vxlan->cfg.age_interval = FDB_AGE_DEFAULT;

2799 2800 2801 2802 2803 2804 2805
	list_for_each_entry(tmp, &vn->vxlan_list, next) {
		if (tmp->cfg.vni == conf->vni &&
		    (tmp->default_dst.remote_ip.sa.sa_family == AF_INET6 ||
		     tmp->cfg.saddr.sa.sa_family == AF_INET6) == use_ipv6 &&
		    tmp->cfg.dst_port == vxlan->cfg.dst_port &&
		    (tmp->flags & VXLAN_F_RCV_FLAGS) ==
		    (vxlan->flags & VXLAN_F_RCV_FLAGS))
2806
		return -EEXIST;
2807
	}
2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866

	dev->ethtool_ops = &vxlan_ethtool_ops;

	/* create an fdb entry for a valid default destination */
	if (!vxlan_addr_any(&vxlan->default_dst.remote_ip)) {
		err = vxlan_fdb_create(vxlan, all_zeros_mac,
				       &vxlan->default_dst.remote_ip,
				       NUD_REACHABLE|NUD_PERMANENT,
				       NLM_F_EXCL|NLM_F_CREATE,
				       vxlan->cfg.dst_port,
				       vxlan->default_dst.remote_vni,
				       vxlan->default_dst.remote_ifindex,
				       NTF_SELF);
		if (err)
			return err;
	}

	err = register_netdevice(dev);
	if (err) {
		vxlan_fdb_delete_default(vxlan);
		return err;
	}

	list_add(&vxlan->next, &vn->vxlan_list);

	return 0;
}

struct net_device *vxlan_dev_create(struct net *net, const char *name,
				    u8 name_assign_type, struct vxlan_config *conf)
{
	struct nlattr *tb[IFLA_MAX+1];
	struct net_device *dev;
	int err;

	memset(&tb, 0, sizeof(tb));

	dev = rtnl_create_link(net, name, name_assign_type,
			       &vxlan_link_ops, tb);
	if (IS_ERR(dev))
		return dev;

	err = vxlan_dev_configure(net, dev, conf);
	if (err < 0) {
		free_netdev(dev);
		return ERR_PTR(err);
	}

	return dev;
}
EXPORT_SYMBOL_GPL(vxlan_dev_create);

static int vxlan_newlink(struct net *src_net, struct net_device *dev,
			 struct nlattr *tb[], struct nlattr *data[])
{
	struct vxlan_config conf;
	int err;

	memset(&conf, 0, sizeof(conf));
2867 2868

	if (data[IFLA_VXLAN_ID])
2869
		conf.vni = cpu_to_be32(nla_get_u32(data[IFLA_VXLAN_ID]));
2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895

	if (data[IFLA_VXLAN_GROUP]) {
		conf.remote_ip.sin.sin_addr.s_addr = nla_get_in_addr(data[IFLA_VXLAN_GROUP]);
	} else if (data[IFLA_VXLAN_GROUP6]) {
		if (!IS_ENABLED(CONFIG_IPV6))
			return -EPFNOSUPPORT;

		conf.remote_ip.sin6.sin6_addr = nla_get_in6_addr(data[IFLA_VXLAN_GROUP6]);
		conf.remote_ip.sa.sa_family = AF_INET6;
	}

	if (data[IFLA_VXLAN_LOCAL]) {
		conf.saddr.sin.sin_addr.s_addr = nla_get_in_addr(data[IFLA_VXLAN_LOCAL]);
		conf.saddr.sa.sa_family = AF_INET;
	} else if (data[IFLA_VXLAN_LOCAL6]) {
		if (!IS_ENABLED(CONFIG_IPV6))
			return -EPFNOSUPPORT;

		/* TODO: respect scope id */
		conf.saddr.sin6.sin6_addr = nla_get_in6_addr(data[IFLA_VXLAN_LOCAL6]);
		conf.saddr.sa.sa_family = AF_INET6;
	}

	if (data[IFLA_VXLAN_LINK])
		conf.remote_ifindex = nla_get_u32(data[IFLA_VXLAN_LINK]);

S
stephen hemminger 已提交
2896
	if (data[IFLA_VXLAN_TOS])
2897
		conf.tos  = nla_get_u8(data[IFLA_VXLAN_TOS]);
S
stephen hemminger 已提交
2898

2899
	if (data[IFLA_VXLAN_TTL])
2900
		conf.ttl = nla_get_u8(data[IFLA_VXLAN_TTL]);
2901

2902 2903 2904 2905
	if (data[IFLA_VXLAN_LABEL])
		conf.label = nla_get_be32(data[IFLA_VXLAN_LABEL]) &
			     IPV6_FLOWLABEL_MASK;

S
stephen hemminger 已提交
2906
	if (!data[IFLA_VXLAN_LEARNING] || nla_get_u8(data[IFLA_VXLAN_LEARNING]))
2907
		conf.flags |= VXLAN_F_LEARN;
S
stephen hemminger 已提交
2908 2909

	if (data[IFLA_VXLAN_AGEING])
2910
		conf.age_interval = nla_get_u32(data[IFLA_VXLAN_AGEING]);
S
stephen hemminger 已提交
2911

D
David Stevens 已提交
2912
	if (data[IFLA_VXLAN_PROXY] && nla_get_u8(data[IFLA_VXLAN_PROXY]))
2913
		conf.flags |= VXLAN_F_PROXY;
D
David Stevens 已提交
2914 2915

	if (data[IFLA_VXLAN_RSC] && nla_get_u8(data[IFLA_VXLAN_RSC]))
2916
		conf.flags |= VXLAN_F_RSC;
D
David Stevens 已提交
2917 2918

	if (data[IFLA_VXLAN_L2MISS] && nla_get_u8(data[IFLA_VXLAN_L2MISS]))
2919
		conf.flags |= VXLAN_F_L2MISS;
D
David Stevens 已提交
2920 2921

	if (data[IFLA_VXLAN_L3MISS] && nla_get_u8(data[IFLA_VXLAN_L3MISS]))
2922
		conf.flags |= VXLAN_F_L3MISS;
D
David Stevens 已提交
2923

S
stephen hemminger 已提交
2924
	if (data[IFLA_VXLAN_LIMIT])
2925
		conf.addrmax = nla_get_u32(data[IFLA_VXLAN_LIMIT]);
S
stephen hemminger 已提交
2926

2927 2928 2929 2930
	if (data[IFLA_VXLAN_COLLECT_METADATA] &&
	    nla_get_u8(data[IFLA_VXLAN_COLLECT_METADATA]))
		conf.flags |= VXLAN_F_COLLECT_METADATA;

2931 2932 2933
	if (data[IFLA_VXLAN_PORT_RANGE]) {
		const struct ifla_vxlan_port_range *p
			= nla_data(data[IFLA_VXLAN_PORT_RANGE]);
2934 2935
		conf.port_min = ntohs(p->low);
		conf.port_max = ntohs(p->high);
2936 2937
	}

2938
	if (data[IFLA_VXLAN_PORT])
2939
		conf.dst_port = nla_get_be16(data[IFLA_VXLAN_PORT]);
2940

2941 2942 2943
	if (data[IFLA_VXLAN_UDP_CSUM] &&
	    !nla_get_u8(data[IFLA_VXLAN_UDP_CSUM]))
		conf.flags |= VXLAN_F_UDP_ZERO_CSUM_TX;
2944 2945 2946

	if (data[IFLA_VXLAN_UDP_ZERO_CSUM6_TX] &&
	    nla_get_u8(data[IFLA_VXLAN_UDP_ZERO_CSUM6_TX]))
2947
		conf.flags |= VXLAN_F_UDP_ZERO_CSUM6_TX;
2948 2949 2950

	if (data[IFLA_VXLAN_UDP_ZERO_CSUM6_RX] &&
	    nla_get_u8(data[IFLA_VXLAN_UDP_ZERO_CSUM6_RX]))
2951
		conf.flags |= VXLAN_F_UDP_ZERO_CSUM6_RX;
2952

T
Tom Herbert 已提交
2953 2954
	if (data[IFLA_VXLAN_REMCSUM_TX] &&
	    nla_get_u8(data[IFLA_VXLAN_REMCSUM_TX]))
2955
		conf.flags |= VXLAN_F_REMCSUM_TX;
T
Tom Herbert 已提交
2956 2957 2958

	if (data[IFLA_VXLAN_REMCSUM_RX] &&
	    nla_get_u8(data[IFLA_VXLAN_REMCSUM_RX]))
2959
		conf.flags |= VXLAN_F_REMCSUM_RX;
T
Tom Herbert 已提交
2960

T
Thomas Graf 已提交
2961
	if (data[IFLA_VXLAN_GBP])
2962
		conf.flags |= VXLAN_F_GBP;
T
Thomas Graf 已提交
2963

2964
	if (data[IFLA_VXLAN_REMCSUM_NOPARTIAL])
2965
		conf.flags |= VXLAN_F_REMCSUM_NOPARTIAL;
2966

2967 2968 2969 2970 2971
	err = vxlan_dev_configure(src_net, dev, &conf);
	switch (err) {
	case -ENODEV:
		pr_info("ifindex %d does not exist\n", conf.remote_ifindex);
		break;
Y
Yan Burman 已提交
2972

2973 2974 2975
	case -EPERM:
		pr_info("IPv6 is disabled via sysctl\n");
		break;
S
stephen hemminger 已提交
2976

2977
	case -EEXIST:
2978
		pr_info("duplicate VNI %u\n", be32_to_cpu(conf.vni));
2979
		break;
2980 2981
	}

2982
	return err;
S
stephen hemminger 已提交
2983 2984 2985 2986 2987
}

static void vxlan_dellink(struct net_device *dev, struct list_head *head)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);
N
Nicolas Dichtel 已提交
2988
	struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
S
stephen hemminger 已提交
2989

2990
	spin_lock(&vn->sock_lock);
2991 2992
	if (!hlist_unhashed(&vxlan->hlist))
		hlist_del_rcu(&vxlan->hlist);
2993 2994
	spin_unlock(&vn->sock_lock);

T
Tom Herbert 已提交
2995
	gro_cells_destroy(&vxlan->gro_cells);
2996
	list_del(&vxlan->next);
S
stephen hemminger 已提交
2997 2998 2999 3000 3001 3002 3003
	unregister_netdevice_queue(dev, head);
}

static size_t vxlan_get_size(const struct net_device *dev)
{

	return nla_total_size(sizeof(__u32)) +	/* IFLA_VXLAN_ID */
C
Cong Wang 已提交
3004
		nla_total_size(sizeof(struct in6_addr)) + /* IFLA_VXLAN_GROUP{6} */
S
stephen hemminger 已提交
3005
		nla_total_size(sizeof(__u32)) +	/* IFLA_VXLAN_LINK */
C
Cong Wang 已提交
3006
		nla_total_size(sizeof(struct in6_addr)) + /* IFLA_VXLAN_LOCAL{6} */
S
stephen hemminger 已提交
3007 3008
		nla_total_size(sizeof(__u8)) +	/* IFLA_VXLAN_TTL */
		nla_total_size(sizeof(__u8)) +	/* IFLA_VXLAN_TOS */
3009
		nla_total_size(sizeof(__be32)) + /* IFLA_VXLAN_LABEL */
S
stephen hemminger 已提交
3010
		nla_total_size(sizeof(__u8)) +	/* IFLA_VXLAN_LEARNING */
D
David Stevens 已提交
3011 3012 3013 3014
		nla_total_size(sizeof(__u8)) +	/* IFLA_VXLAN_PROXY */
		nla_total_size(sizeof(__u8)) +	/* IFLA_VXLAN_RSC */
		nla_total_size(sizeof(__u8)) +	/* IFLA_VXLAN_L2MISS */
		nla_total_size(sizeof(__u8)) +	/* IFLA_VXLAN_L3MISS */
3015
		nla_total_size(sizeof(__u8)) +	/* IFLA_VXLAN_COLLECT_METADATA */
S
stephen hemminger 已提交
3016 3017
		nla_total_size(sizeof(__u32)) +	/* IFLA_VXLAN_AGEING */
		nla_total_size(sizeof(__u32)) +	/* IFLA_VXLAN_LIMIT */
3018
		nla_total_size(sizeof(struct ifla_vxlan_port_range)) +
3019 3020 3021 3022
		nla_total_size(sizeof(__be16)) + /* IFLA_VXLAN_PORT */
		nla_total_size(sizeof(__u8)) + /* IFLA_VXLAN_UDP_CSUM */
		nla_total_size(sizeof(__u8)) + /* IFLA_VXLAN_UDP_ZERO_CSUM6_TX */
		nla_total_size(sizeof(__u8)) + /* IFLA_VXLAN_UDP_ZERO_CSUM6_RX */
T
Tom Herbert 已提交
3023 3024
		nla_total_size(sizeof(__u8)) + /* IFLA_VXLAN_REMCSUM_TX */
		nla_total_size(sizeof(__u8)) + /* IFLA_VXLAN_REMCSUM_RX */
S
stephen hemminger 已提交
3025 3026 3027 3028 3029 3030
		0;
}

static int vxlan_fill_info(struct sk_buff *skb, const struct net_device *dev)
{
	const struct vxlan_dev *vxlan = netdev_priv(dev);
3031
	const struct vxlan_rdst *dst = &vxlan->default_dst;
3032
	struct ifla_vxlan_port_range ports = {
3033 3034
		.low =  htons(vxlan->cfg.port_min),
		.high = htons(vxlan->cfg.port_max),
3035
	};
S
stephen hemminger 已提交
3036

3037
	if (nla_put_u32(skb, IFLA_VXLAN_ID, be32_to_cpu(dst->remote_vni)))
S
stephen hemminger 已提交
3038 3039
		goto nla_put_failure;

C
Cong Wang 已提交
3040 3041
	if (!vxlan_addr_any(&dst->remote_ip)) {
		if (dst->remote_ip.sa.sa_family == AF_INET) {
3042 3043
			if (nla_put_in_addr(skb, IFLA_VXLAN_GROUP,
					    dst->remote_ip.sin.sin_addr.s_addr))
C
Cong Wang 已提交
3044 3045 3046
				goto nla_put_failure;
#if IS_ENABLED(CONFIG_IPV6)
		} else {
3047 3048
			if (nla_put_in6_addr(skb, IFLA_VXLAN_GROUP6,
					     &dst->remote_ip.sin6.sin6_addr))
C
Cong Wang 已提交
3049 3050 3051 3052
				goto nla_put_failure;
#endif
		}
	}
S
stephen hemminger 已提交
3053

3054
	if (dst->remote_ifindex && nla_put_u32(skb, IFLA_VXLAN_LINK, dst->remote_ifindex))
S
stephen hemminger 已提交
3055 3056
		goto nla_put_failure;

3057 3058
	if (!vxlan_addr_any(&vxlan->cfg.saddr)) {
		if (vxlan->cfg.saddr.sa.sa_family == AF_INET) {
3059
			if (nla_put_in_addr(skb, IFLA_VXLAN_LOCAL,
3060
					    vxlan->cfg.saddr.sin.sin_addr.s_addr))
C
Cong Wang 已提交
3061 3062 3063
				goto nla_put_failure;
#if IS_ENABLED(CONFIG_IPV6)
		} else {
3064
			if (nla_put_in6_addr(skb, IFLA_VXLAN_LOCAL6,
3065
					     &vxlan->cfg.saddr.sin6.sin6_addr))
C
Cong Wang 已提交
3066 3067 3068 3069
				goto nla_put_failure;
#endif
		}
	}
S
stephen hemminger 已提交
3070

3071 3072
	if (nla_put_u8(skb, IFLA_VXLAN_TTL, vxlan->cfg.ttl) ||
	    nla_put_u8(skb, IFLA_VXLAN_TOS, vxlan->cfg.tos) ||
3073
	    nla_put_be32(skb, IFLA_VXLAN_LABEL, vxlan->cfg.label) ||
D
David Stevens 已提交
3074 3075 3076 3077 3078 3079 3080 3081 3082
	    nla_put_u8(skb, IFLA_VXLAN_LEARNING,
			!!(vxlan->flags & VXLAN_F_LEARN)) ||
	    nla_put_u8(skb, IFLA_VXLAN_PROXY,
			!!(vxlan->flags & VXLAN_F_PROXY)) ||
	    nla_put_u8(skb, IFLA_VXLAN_RSC, !!(vxlan->flags & VXLAN_F_RSC)) ||
	    nla_put_u8(skb, IFLA_VXLAN_L2MISS,
			!!(vxlan->flags & VXLAN_F_L2MISS)) ||
	    nla_put_u8(skb, IFLA_VXLAN_L3MISS,
			!!(vxlan->flags & VXLAN_F_L3MISS)) ||
3083 3084
	    nla_put_u8(skb, IFLA_VXLAN_COLLECT_METADATA,
		       !!(vxlan->flags & VXLAN_F_COLLECT_METADATA)) ||
3085 3086 3087
	    nla_put_u32(skb, IFLA_VXLAN_AGEING, vxlan->cfg.age_interval) ||
	    nla_put_u32(skb, IFLA_VXLAN_LIMIT, vxlan->cfg.addrmax) ||
	    nla_put_be16(skb, IFLA_VXLAN_PORT, vxlan->cfg.dst_port) ||
3088
	    nla_put_u8(skb, IFLA_VXLAN_UDP_CSUM,
3089
			!(vxlan->flags & VXLAN_F_UDP_ZERO_CSUM_TX)) ||
3090 3091 3092
	    nla_put_u8(skb, IFLA_VXLAN_UDP_ZERO_CSUM6_TX,
			!!(vxlan->flags & VXLAN_F_UDP_ZERO_CSUM6_TX)) ||
	    nla_put_u8(skb, IFLA_VXLAN_UDP_ZERO_CSUM6_RX,
T
Tom Herbert 已提交
3093 3094 3095 3096 3097
			!!(vxlan->flags & VXLAN_F_UDP_ZERO_CSUM6_RX)) ||
	    nla_put_u8(skb, IFLA_VXLAN_REMCSUM_TX,
			!!(vxlan->flags & VXLAN_F_REMCSUM_TX)) ||
	    nla_put_u8(skb, IFLA_VXLAN_REMCSUM_RX,
			!!(vxlan->flags & VXLAN_F_REMCSUM_RX)))
S
stephen hemminger 已提交
3098 3099
		goto nla_put_failure;

3100 3101 3102
	if (nla_put(skb, IFLA_VXLAN_PORT_RANGE, sizeof(ports), &ports))
		goto nla_put_failure;

T
Thomas Graf 已提交
3103 3104 3105 3106
	if (vxlan->flags & VXLAN_F_GBP &&
	    nla_put_flag(skb, IFLA_VXLAN_GBP))
		goto nla_put_failure;

3107 3108 3109 3110
	if (vxlan->flags & VXLAN_F_REMCSUM_NOPARTIAL &&
	    nla_put_flag(skb, IFLA_VXLAN_REMCSUM_NOPARTIAL))
		goto nla_put_failure;

S
stephen hemminger 已提交
3111 3112 3113 3114 3115 3116
	return 0;

nla_put_failure:
	return -EMSGSIZE;
}

3117 3118 3119 3120 3121 3122 3123
static struct net *vxlan_get_link_net(const struct net_device *dev)
{
	struct vxlan_dev *vxlan = netdev_priv(dev);

	return vxlan->net;
}

S
stephen hemminger 已提交
3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134
static struct rtnl_link_ops vxlan_link_ops __read_mostly = {
	.kind		= "vxlan",
	.maxtype	= IFLA_VXLAN_MAX,
	.policy		= vxlan_policy,
	.priv_size	= sizeof(struct vxlan_dev),
	.setup		= vxlan_setup,
	.validate	= vxlan_validate,
	.newlink	= vxlan_newlink,
	.dellink	= vxlan_dellink,
	.get_size	= vxlan_get_size,
	.fill_info	= vxlan_fill_info,
3135
	.get_link_net	= vxlan_get_link_net,
S
stephen hemminger 已提交
3136 3137
};

3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163
static void vxlan_handle_lowerdev_unregister(struct vxlan_net *vn,
					     struct net_device *dev)
{
	struct vxlan_dev *vxlan, *next;
	LIST_HEAD(list_kill);

	list_for_each_entry_safe(vxlan, next, &vn->vxlan_list, next) {
		struct vxlan_rdst *dst = &vxlan->default_dst;

		/* In case we created vxlan device with carrier
		 * and we loose the carrier due to module unload
		 * we also need to remove vxlan device. In other
		 * cases, it's not necessary and remote_ifindex
		 * is 0 here, so no matches.
		 */
		if (dst->remote_ifindex == dev->ifindex)
			vxlan_dellink(vxlan->dev, &list_kill);
	}

	unregister_netdevice_many(&list_kill);
}

static int vxlan_lowerdev_event(struct notifier_block *unused,
				unsigned long event, void *ptr)
{
	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
3164
	struct vxlan_net *vn = net_generic(dev_net(dev), vxlan_net_id);
3165

3166
	if (event == NETDEV_UNREGISTER)
3167 3168 3169 3170 3171 3172 3173 3174 3175
		vxlan_handle_lowerdev_unregister(vn, dev);

	return NOTIFY_DONE;
}

static struct notifier_block vxlan_notifier_block __read_mostly = {
	.notifier_call = vxlan_lowerdev_event,
};

S
stephen hemminger 已提交
3176 3177 3178
static __net_init int vxlan_init_net(struct net *net)
{
	struct vxlan_net *vn = net_generic(net, vxlan_net_id);
3179
	unsigned int h;
S
stephen hemminger 已提交
3180

3181
	INIT_LIST_HEAD(&vn->vxlan_list);
3182
	spin_lock_init(&vn->sock_lock);
S
stephen hemminger 已提交
3183

3184 3185
	for (h = 0; h < PORT_HASH_SIZE; ++h)
		INIT_HLIST_HEAD(&vn->sock_list[h]);
S
stephen hemminger 已提交
3186 3187 3188 3189

	return 0;
}

N
Nicolas Dichtel 已提交
3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205
static void __net_exit vxlan_exit_net(struct net *net)
{
	struct vxlan_net *vn = net_generic(net, vxlan_net_id);
	struct vxlan_dev *vxlan, *next;
	struct net_device *dev, *aux;
	LIST_HEAD(list);

	rtnl_lock();
	for_each_netdev_safe(net, dev, aux)
		if (dev->rtnl_link_ops == &vxlan_link_ops)
			unregister_netdevice_queue(dev, &list);

	list_for_each_entry_safe(vxlan, next, &vn->vxlan_list, next) {
		/* If vxlan->dev is in the same netns, it has already been added
		 * to the list by the previous loop.
		 */
T
Tom Herbert 已提交
3206 3207
		if (!net_eq(dev_net(vxlan->dev), net)) {
			gro_cells_destroy(&vxlan->gro_cells);
3208
			unregister_netdevice_queue(vxlan->dev, &list);
T
Tom Herbert 已提交
3209
		}
N
Nicolas Dichtel 已提交
3210 3211 3212 3213 3214 3215
	}

	unregister_netdevice_many(&list);
	rtnl_unlock();
}

S
stephen hemminger 已提交
3216 3217
static struct pernet_operations vxlan_net_ops = {
	.init = vxlan_init_net,
N
Nicolas Dichtel 已提交
3218
	.exit = vxlan_exit_net,
S
stephen hemminger 已提交
3219 3220 3221 3222 3223 3224 3225 3226
	.id   = &vxlan_net_id,
	.size = sizeof(struct vxlan_net),
};

static int __init vxlan_init_module(void)
{
	int rc;

3227 3228 3229 3230
	vxlan_wq = alloc_workqueue("vxlan", 0, 0);
	if (!vxlan_wq)
		return -ENOMEM;

S
stephen hemminger 已提交
3231 3232
	get_random_bytes(&vxlan_salt, sizeof(vxlan_salt));

3233
	rc = register_pernet_subsys(&vxlan_net_ops);
S
stephen hemminger 已提交
3234 3235 3236
	if (rc)
		goto out1;

3237
	rc = register_netdevice_notifier(&vxlan_notifier_block);
S
stephen hemminger 已提交
3238 3239 3240
	if (rc)
		goto out2;

3241 3242 3243
	rc = rtnl_link_register(&vxlan_link_ops);
	if (rc)
		goto out3;
S
stephen hemminger 已提交
3244

3245 3246 3247
	return 0;
out3:
	unregister_netdevice_notifier(&vxlan_notifier_block);
S
stephen hemminger 已提交
3248
out2:
3249
	unregister_pernet_subsys(&vxlan_net_ops);
S
stephen hemminger 已提交
3250
out1:
3251
	destroy_workqueue(vxlan_wq);
S
stephen hemminger 已提交
3252 3253
	return rc;
}
3254
late_initcall(vxlan_init_module);
S
stephen hemminger 已提交
3255 3256 3257

static void __exit vxlan_cleanup_module(void)
{
3258
	rtnl_link_unregister(&vxlan_link_ops);
3259
	unregister_netdevice_notifier(&vxlan_notifier_block);
3260
	destroy_workqueue(vxlan_wq);
3261 3262
	unregister_pernet_subsys(&vxlan_net_ops);
	/* rcu_barrier() is called by netns */
S
stephen hemminger 已提交
3263 3264 3265 3266 3267
}
module_exit(vxlan_cleanup_module);

MODULE_LICENSE("GPL");
MODULE_VERSION(VXLAN_VERSION);
3268
MODULE_AUTHOR("Stephen Hemminger <stephen@networkplumber.org>");
J
Jesse Brandeburg 已提交
3269
MODULE_DESCRIPTION("Driver for VXLAN encapsulated traffic");
S
stephen hemminger 已提交
3270
MODULE_ALIAS_RTNL_LINK("vxlan");