datapath.c 49.8 KB
Newer Older
1
/*
2
 * Copyright (c) 2007-2013 Nicira, Inc.
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of version 2 of the GNU General Public
 * License as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
 * 02110-1301, USA
 */

#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

#include <linux/init.h>
#include <linux/module.h>
#include <linux/if_arp.h>
#include <linux/if_vlan.h>
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/jhash.h>
#include <linux/delay.h>
#include <linux/time.h>
#include <linux/etherdevice.h>
#include <linux/genetlink.h>
#include <linux/kernel.h>
#include <linux/kthread.h>
#include <linux/mutex.h>
#include <linux/percpu.h>
#include <linux/rcupdate.h>
#include <linux/tcp.h>
#include <linux/udp.h>
#include <linux/ethtool.h>
#include <linux/wait.h>
#include <asm/div64.h>
#include <linux/highmem.h>
#include <linux/netfilter_bridge.h>
#include <linux/netfilter_ipv4.h>
#include <linux/inetdevice.h>
#include <linux/list.h>
47
#include <linux/lockdep.h>
48 49 50 51 52
#include <linux/openvswitch.h>
#include <linux/rculist.h>
#include <linux/dmi.h>
#include <linux/workqueue.h>
#include <net/genetlink.h>
53 54
#include <net/net_namespace.h>
#include <net/netns/generic.h>
55 56 57

#include "datapath.h"
#include "flow.h"
58
#include "flow_table.h"
59
#include "flow_netlink.h"
60
#include "vport-internal_dev.h"
61
#include "vport-netdev.h"
62

63 64
int ovs_net_id __read_mostly;

65 66 67 68 69 70 71 72 73
/* Check if need to build a reply message.
 * OVS userspace sets the NLM_F_ECHO flag if it needs the reply. */
static bool ovs_must_notify(struct genl_info *info,
			    const struct genl_multicast_group *grp)
{
	return info->nlhdr->nlmsg_flags & NLM_F_ECHO ||
		netlink_has_listeners(genl_info_net(info)->genl_sock, 0);
}

74
static void ovs_notify(struct genl_family *family,
75
		       struct sk_buff *skb, struct genl_info *info)
76
{
77
	genl_notify(family, skb, genl_info_net(info), info->snd_portid,
78
		    0, info->nlhdr, GFP_KERNEL);
79 80
}

81 82 83
/**
 * DOC: Locking:
 *
84 85 86 87
 * All writes e.g. Writes to device state (add/remove datapath, port, set
 * operations on vports, etc.), Writes to other state (flow table
 * modifications, set miscellaneous datapath parameters, etc.) are protected
 * by ovs_lock.
88 89 90 91 92 93
 *
 * Reads are protected by RCU.
 *
 * There are a few special cases (mostly stats) that have their own
 * synchronization but they nest under all of above and don't interact with
 * each other.
94 95
 *
 * The RTNL lock nests inside ovs_mutex.
96 97
 */

98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
static DEFINE_MUTEX(ovs_mutex);

void ovs_lock(void)
{
	mutex_lock(&ovs_mutex);
}

void ovs_unlock(void)
{
	mutex_unlock(&ovs_mutex);
}

#ifdef CONFIG_LOCKDEP
int lockdep_ovsl_is_held(void)
{
	if (debug_locks)
		return lockdep_is_held(&ovs_mutex);
	else
		return 1;
}
#endif

120
static struct vport *new_vport(const struct vport_parms *);
121
static int queue_gso_packets(struct datapath *dp, struct sk_buff *,
122
			     const struct dp_upcall_info *);
123
static int queue_userspace_packet(struct datapath *dp, struct sk_buff *,
124 125
				  const struct dp_upcall_info *);

126
/* Must be called with rcu_read_lock or ovs_mutex. */
127
static struct datapath *get_dp(struct net *net, int dp_ifindex)
128 129 130 131 132
{
	struct datapath *dp = NULL;
	struct net_device *dev;

	rcu_read_lock();
133
	dev = dev_get_by_index_rcu(net, dp_ifindex);
134 135 136 137 138 139 140 141 142 143
	if (dev) {
		struct vport *vport = ovs_internal_dev_get_vport(dev);
		if (vport)
			dp = vport->dp;
	}
	rcu_read_unlock();

	return dp;
}

144
/* Must be called with rcu_read_lock or ovs_mutex. */
Stephen Hemminger's avatar
Stephen Hemminger committed
145
static const char *ovs_dp_name(const struct datapath *dp)
146
{
147
	struct vport *vport = ovs_vport_ovsl_rcu(dp, OVSP_LOCAL);
148 149 150 151 152 153 154 155 156 157
	return vport->ops->get_name(vport);
}

static int get_dpifindex(struct datapath *dp)
{
	struct vport *local;
	int ifindex;

	rcu_read_lock();

158
	local = ovs_vport_rcu(dp, OVSP_LOCAL);
159
	if (local)
160
		ifindex = netdev_vport_priv(local)->dev->ifindex;
161 162 163 164 165 166 167 168 169 170 171 172 173
	else
		ifindex = 0;

	rcu_read_unlock();

	return ifindex;
}

static void destroy_dp_rcu(struct rcu_head *rcu)
{
	struct datapath *dp = container_of(rcu, struct datapath, rcu);

	free_percpu(dp->stats_percpu);
174
	release_net(ovs_dp_get_net(dp));
175
	kfree(dp->ports);
176 177 178
	kfree(dp);
}

179 180 181 182 183 184
static struct hlist_head *vport_hash_bucket(const struct datapath *dp,
					    u16 port_no)
{
	return &dp->ports[port_no & (DP_VPORT_HASH_BUCKETS - 1)];
}

185
/* Called with ovs_mutex or RCU read lock. */
186 187 188 189 190 191
struct vport *ovs_lookup_vport(const struct datapath *dp, u16 port_no)
{
	struct vport *vport;
	struct hlist_head *head;

	head = vport_hash_bucket(dp, port_no);
192
	hlist_for_each_entry_rcu(vport, head, dp_hash_node) {
193 194 195 196 197 198
		if (vport->port_no == port_no)
			return vport;
	}
	return NULL;
}

199
/* Called with ovs_mutex. */
200 201 202 203 204 205 206
static struct vport *new_vport(const struct vport_parms *parms)
{
	struct vport *vport;

	vport = ovs_vport_add(parms);
	if (!IS_ERR(vport)) {
		struct datapath *dp = parms->dp;
207
		struct hlist_head *head = vport_hash_bucket(dp, vport->port_no);
208

209
		hlist_add_head_rcu(&vport->dp_hash_node, head);
210 211 212 213 214 215
	}
	return vport;
}

void ovs_dp_detach_port(struct vport *p)
{
216
	ASSERT_OVSL();
217 218

	/* First drop references to device. */
219
	hlist_del_rcu(&p->dp_hash_node);
220 221 222 223 224 225 226 227 228 229 230 231 232

	/* Then destroy it. */
	ovs_vport_del(p);
}

/* Must be called with rcu_read_lock. */
void ovs_dp_process_received_packet(struct vport *p, struct sk_buff *skb)
{
	struct datapath *dp = p->dp;
	struct sw_flow *flow;
	struct dp_stats_percpu *stats;
	struct sw_flow_key key;
	u64 *stats_counter;
233
	u32 n_mask_hit;
234 235
	int error;

236
	stats = this_cpu_ptr(dp->stats_percpu);
237 238

	/* Extract flow from 'skb' into 'key'. */
239
	error = ovs_flow_extract(skb, p->port_no, &key);
240 241 242 243 244 245
	if (unlikely(error)) {
		kfree_skb(skb);
		return;
	}

	/* Look up flow. */
246
	flow = ovs_flow_tbl_lookup_stats(&dp->table, &key, &n_mask_hit);
247 248 249 250 251 252
	if (unlikely(!flow)) {
		struct dp_upcall_info upcall;

		upcall.cmd = OVS_PACKET_CMD_MISS;
		upcall.key = &key;
		upcall.userdata = NULL;
253
		upcall.portid = p->upcall_portid;
254 255 256 257 258 259 260
		ovs_dp_upcall(dp, skb, &upcall);
		consume_skb(skb);
		stats_counter = &stats->n_missed;
		goto out;
	}

	OVS_CB(skb)->flow = flow;
261
	OVS_CB(skb)->pkt_key = &key;
262

263
	ovs_flow_stats_update(OVS_CB(skb)->flow, skb);
264
	ovs_execute_actions(dp, skb);
265
	stats_counter = &stats->n_hit;
266 267 268

out:
	/* Update datapath statistics. */
269
	u64_stats_update_begin(&stats->syncp);
270
	(*stats_counter)++;
271
	stats->n_mask_hit += n_mask_hit;
272
	u64_stats_update_end(&stats->syncp);
273 274 275 276 277 278 279
}

static struct genl_family dp_packet_genl_family = {
	.id = GENL_ID_GENERATE,
	.hdrsize = sizeof(struct ovs_header),
	.name = OVS_PACKET_FAMILY,
	.version = OVS_PACKET_VERSION,
280
	.maxattr = OVS_PACKET_ATTR_MAX,
281 282
	.netnsok = true,
	.parallel_ops = true,
283 284 285
};

int ovs_dp_upcall(struct datapath *dp, struct sk_buff *skb,
286
		  const struct dp_upcall_info *upcall_info)
287 288 289 290
{
	struct dp_stats_percpu *stats;
	int err;

291
	if (upcall_info->portid == 0) {
292 293 294 295 296
		err = -ENOTCONN;
		goto err;
	}

	if (!skb_is_gso(skb))
297
		err = queue_userspace_packet(dp, skb, upcall_info);
298
	else
299
		err = queue_gso_packets(dp, skb, upcall_info);
300 301 302 303 304 305
	if (err)
		goto err;

	return 0;

err:
306
	stats = this_cpu_ptr(dp->stats_percpu);
307

308
	u64_stats_update_begin(&stats->syncp);
309
	stats->n_lost++;
310
	u64_stats_update_end(&stats->syncp);
311 312 313 314

	return err;
}

315
static int queue_gso_packets(struct datapath *dp, struct sk_buff *skb,
316 317
			     const struct dp_upcall_info *upcall_info)
{
318
	unsigned short gso_type = skb_shinfo(skb)->gso_type;
319 320 321 322 323
	struct dp_upcall_info later_info;
	struct sw_flow_key later_key;
	struct sk_buff *segs, *nskb;
	int err;

324
	segs = __skb_gso_segment(skb, NETIF_F_SG, false);
325 326
	if (IS_ERR(segs))
		return PTR_ERR(segs);
327 328 329 330

	/* Queue all of the segments. */
	skb = segs;
	do {
331
		err = queue_userspace_packet(dp, skb, upcall_info);
332 333 334
		if (err)
			break;

335
		if (skb == segs && gso_type & SKB_GSO_UDP) {
336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360
			/* The initial flow key extracted by ovs_flow_extract()
			 * in this case is for a first fragment, so we need to
			 * properly mark later fragments.
			 */
			later_key = *upcall_info->key;
			later_key.ip.frag = OVS_FRAG_TYPE_LATER;

			later_info = *upcall_info;
			later_info.key = &later_key;
			upcall_info = &later_info;
		}
	} while ((skb = skb->next));

	/* Free all of the segments. */
	skb = segs;
	do {
		nskb = skb->next;
		if (err)
			kfree_skb(skb);
		else
			consume_skb(skb);
	} while ((skb = nskb));
	return err;
}

361 362 363
static size_t key_attr_size(void)
{
	return    nla_total_size(4)   /* OVS_KEY_ATTR_PRIORITY */
364 365 366 367 368 369 370 371
		+ nla_total_size(0)   /* OVS_KEY_ATTR_TUNNEL */
		  + nla_total_size(8)   /* OVS_TUNNEL_KEY_ATTR_ID */
		  + nla_total_size(4)   /* OVS_TUNNEL_KEY_ATTR_IPV4_SRC */
		  + nla_total_size(4)   /* OVS_TUNNEL_KEY_ATTR_IPV4_DST */
		  + nla_total_size(1)   /* OVS_TUNNEL_KEY_ATTR_TOS */
		  + nla_total_size(1)   /* OVS_TUNNEL_KEY_ATTR_TTL */
		  + nla_total_size(0)   /* OVS_TUNNEL_KEY_ATTR_DONT_FRAGMENT */
		  + nla_total_size(0)   /* OVS_TUNNEL_KEY_ATTR_CSUM */
372 373 374 375 376 377 378 379 380 381 382 383
		+ nla_total_size(4)   /* OVS_KEY_ATTR_IN_PORT */
		+ nla_total_size(4)   /* OVS_KEY_ATTR_SKB_MARK */
		+ nla_total_size(12)  /* OVS_KEY_ATTR_ETHERNET */
		+ nla_total_size(2)   /* OVS_KEY_ATTR_ETHERTYPE */
		+ nla_total_size(4)   /* OVS_KEY_ATTR_8021Q */
		+ nla_total_size(0)   /* OVS_KEY_ATTR_ENCAP */
		+ nla_total_size(2)   /* OVS_KEY_ATTR_ETHERTYPE */
		+ nla_total_size(40)  /* OVS_KEY_ATTR_IPV6 */
		+ nla_total_size(2)   /* OVS_KEY_ATTR_ICMPV6 */
		+ nla_total_size(28); /* OVS_KEY_ATTR_ND */
}

384 385
static size_t upcall_msg_size(const struct nlattr *userdata,
			      unsigned int hdrlen)
386 387
{
	size_t size = NLMSG_ALIGN(sizeof(struct ovs_header))
388
		+ nla_total_size(hdrlen) /* OVS_PACKET_ATTR_PACKET */
389 390 391 392 393 394 395 396 397
		+ nla_total_size(key_attr_size()); /* OVS_PACKET_ATTR_KEY */

	/* OVS_PACKET_ATTR_USERDATA */
	if (userdata)
		size += NLA_ALIGN(userdata->nla_len);

	return size;
}

398
static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
399 400 401 402 403 404
				  const struct dp_upcall_info *upcall_info)
{
	struct ovs_header *upcall;
	struct sk_buff *nskb = NULL;
	struct sk_buff *user_skb; /* to be queued to userspace */
	struct nlattr *nla;
405
	struct genl_info info = {
406
		.dst_sk = ovs_dp_get_net(dp)->genl_sock,
407 408 409
		.snd_portid = upcall_info->portid,
	};
	size_t len;
410
	unsigned int hlen;
411 412 413 414 415
	int err, dp_ifindex;

	dp_ifindex = get_dpifindex(dp);
	if (!dp_ifindex)
		return -ENODEV;
416 417 418 419 420 421

	if (vlan_tx_tag_present(skb)) {
		nskb = skb_clone(skb, GFP_ATOMIC);
		if (!nskb)
			return -ENOMEM;

422
		nskb = __vlan_put_tag(nskb, nskb->vlan_proto, vlan_tx_tag_get(nskb));
423
		if (!nskb)
424 425 426 427 428 429 430 431 432 433 434
			return -ENOMEM;

		nskb->vlan_tci = 0;
		skb = nskb;
	}

	if (nla_attr_size(skb->len) > USHRT_MAX) {
		err = -EFBIG;
		goto out;
	}

435 436 437 438 439 440 441 442 443 444 445 446 447 448 449
	/* Complete checksum if needed */
	if (skb->ip_summed == CHECKSUM_PARTIAL &&
	    (err = skb_checksum_help(skb)))
		goto out;

	/* Older versions of OVS user space enforce alignment of the last
	 * Netlink attribute to NLA_ALIGNTO which would require extensive
	 * padding logic. Only perform zerocopy if padding is not required.
	 */
	if (dp->user_features & OVS_DP_F_UNALIGNED)
		hlen = skb_zerocopy_headlen(skb);
	else
		hlen = skb->len;

	len = upcall_msg_size(upcall_info->userdata, hlen);
450
	user_skb = genlmsg_new_unicast(len, &info, GFP_ATOMIC);
451 452 453 454 455 456 457 458 459 460
	if (!user_skb) {
		err = -ENOMEM;
		goto out;
	}

	upcall = genlmsg_put(user_skb, 0, 0, &dp_packet_genl_family,
			     0, upcall_info->cmd);
	upcall->dp_ifindex = dp_ifindex;

	nla = nla_nest_start(user_skb, OVS_PACKET_ATTR_KEY);
461
	ovs_nla_put_flow(upcall_info->key, upcall_info->key, user_skb);
462 463 464
	nla_nest_end(user_skb, nla);

	if (upcall_info->userdata)
465 466 467
		__nla_put(user_skb, OVS_PACKET_ATTR_USERDATA,
			  nla_len(upcall_info->userdata),
			  nla_data(upcall_info->userdata));
468

469 470 471 472 473 474 475
	/* Only reserve room for attribute header, packet data is added
	 * in skb_zerocopy() */
	if (!(nla = nla_reserve(user_skb, OVS_PACKET_ATTR_PACKET, 0))) {
		err = -ENOBUFS;
		goto out;
	}
	nla->nla_len = nla_attr_size(skb->len);
476

477 478 479
	err = skb_zerocopy(user_skb, skb, skb->len, hlen);
	if (err)
		goto out;
480

481 482 483 484 485 486 487 488
	/* Pad OVS_PACKET_ATTR_PACKET if linear copy was performed */
	if (!(dp->user_features & OVS_DP_F_UNALIGNED)) {
		size_t plen = NLA_ALIGN(user_skb->len) - user_skb->len;

		if (plen > 0)
			memset(skb_put(user_skb, plen), 0, plen);
	}

489
	((struct nlmsghdr *) user_skb->data)->nlmsg_len = user_skb->len;
490

491
	err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid);
492
out:
493 494
	if (err)
		skb_tx_error(skb);
495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512
	kfree_skb(nskb);
	return err;
}

static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
{
	struct ovs_header *ovs_header = info->userhdr;
	struct nlattr **a = info->attrs;
	struct sw_flow_actions *acts;
	struct sk_buff *packet;
	struct sw_flow *flow;
	struct datapath *dp;
	struct ethhdr *eth;
	int len;
	int err;

	err = -EINVAL;
	if (!a[OVS_PACKET_ATTR_PACKET] || !a[OVS_PACKET_ATTR_KEY] ||
513
	    !a[OVS_PACKET_ATTR_ACTIONS])
514 515 516 517 518 519 520 521 522
		goto err;

	len = nla_len(a[OVS_PACKET_ATTR_PACKET]);
	packet = __dev_alloc_skb(NET_IP_ALIGN + len, GFP_KERNEL);
	err = -ENOMEM;
	if (!packet)
		goto err;
	skb_reserve(packet, NET_IP_ALIGN);

523
	nla_memcpy(__skb_put(packet, len), a[OVS_PACKET_ATTR_PACKET], len);
524 525 526 527 528 529 530

	skb_reset_mac_header(packet);
	eth = eth_hdr(packet);

	/* Normally, setting the skb 'protocol' field would be handled by a
	 * call to eth_type_trans(), but it assumes there's a sending
	 * device, which we may not have. */
Simon Horman's avatar
Simon Horman committed
531
	if (ntohs(eth->h_proto) >= ETH_P_802_3_MIN)
532 533 534 535 536
		packet->protocol = eth->h_proto;
	else
		packet->protocol = htons(ETH_P_802_2);

	/* Build an sw_flow for sending this packet. */
537
	flow = ovs_flow_alloc();
538 539 540 541
	err = PTR_ERR(flow);
	if (IS_ERR(flow))
		goto err_kfree_skb;

542
	err = ovs_flow_extract(packet, -1, &flow->key);
543 544 545
	if (err)
		goto err_flow_free;

546
	err = ovs_nla_get_flow_metadata(flow, a[OVS_PACKET_ATTR_KEY]);
547 548
	if (err)
		goto err_flow_free;
549
	acts = ovs_nla_alloc_flow_actions(nla_len(a[OVS_PACKET_ATTR_ACTIONS]));
550 551 552
	err = PTR_ERR(acts);
	if (IS_ERR(acts))
		goto err_flow_free;
553

554 555
	err = ovs_nla_copy_actions(a[OVS_PACKET_ATTR_ACTIONS],
				   &flow->key, 0, &acts);
556
	rcu_assign_pointer(flow->sf_acts, acts);
557 558
	if (err)
		goto err_flow_free;
559 560

	OVS_CB(packet)->flow = flow;
561
	OVS_CB(packet)->pkt_key = &flow->key;
562
	packet->priority = flow->key.phy.priority;
563
	packet->mark = flow->key.phy.skb_mark;
564 565

	rcu_read_lock();
566
	dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
567 568 569 570 571 572 573 574 575
	err = -ENODEV;
	if (!dp)
		goto err_unlock;

	local_bh_disable();
	err = ovs_execute_actions(dp, packet);
	local_bh_enable();
	rcu_read_unlock();

576
	ovs_flow_free(flow, false);
577 578 579 580 581
	return err;

err_unlock:
	rcu_read_unlock();
err_flow_free:
582
	ovs_flow_free(flow, false);
583 584 585 586 587 588 589
err_kfree_skb:
	kfree_skb(packet);
err:
	return err;
}

static const struct nla_policy packet_policy[OVS_PACKET_ATTR_MAX + 1] = {
590
	[OVS_PACKET_ATTR_PACKET] = { .len = ETH_HLEN },
591 592 593 594
	[OVS_PACKET_ATTR_KEY] = { .type = NLA_NESTED },
	[OVS_PACKET_ATTR_ACTIONS] = { .type = NLA_NESTED },
};

595
static const struct genl_ops dp_packet_genl_ops[] = {
596 597 598 599 600 601 602
	{ .cmd = OVS_PACKET_CMD_EXECUTE,
	  .flags = GENL_ADMIN_PERM, /* Requires CAP_NET_ADMIN privilege. */
	  .policy = packet_policy,
	  .doit = ovs_packet_cmd_execute
	}
};

603 604
static void get_dp_stats(struct datapath *dp, struct ovs_dp_stats *stats,
			 struct ovs_dp_megaflow_stats *mega_stats)
605 606 607
{
	int i;

608 609
	memset(mega_stats, 0, sizeof(*mega_stats));

610
	stats->n_flows = ovs_flow_tbl_count(&dp->table);
611
	mega_stats->n_masks = ovs_flow_tbl_num_masks(&dp->table);
612 613

	stats->n_hit = stats->n_missed = stats->n_lost = 0;
614

615 616 617 618 619 620 621 622
	for_each_possible_cpu(i) {
		const struct dp_stats_percpu *percpu_stats;
		struct dp_stats_percpu local_stats;
		unsigned int start;

		percpu_stats = per_cpu_ptr(dp->stats_percpu, i);

		do {
623
			start = u64_stats_fetch_begin_irq(&percpu_stats->syncp);
624
			local_stats = *percpu_stats;
625
		} while (u64_stats_fetch_retry_irq(&percpu_stats->syncp, start));
626 627 628 629

		stats->n_hit += local_stats.n_hit;
		stats->n_missed += local_stats.n_missed;
		stats->n_lost += local_stats.n_lost;
630
		mega_stats->n_mask_hit += local_stats.n_mask_hit;
631 632 633 634 635 636 637 638 639 640 641 642 643 644
	}
}

static const struct nla_policy flow_policy[OVS_FLOW_ATTR_MAX + 1] = {
	[OVS_FLOW_ATTR_KEY] = { .type = NLA_NESTED },
	[OVS_FLOW_ATTR_ACTIONS] = { .type = NLA_NESTED },
	[OVS_FLOW_ATTR_CLEAR] = { .type = NLA_FLAG },
};

static struct genl_family dp_flow_genl_family = {
	.id = GENL_ID_GENERATE,
	.hdrsize = sizeof(struct ovs_header),
	.name = OVS_FLOW_FAMILY,
	.version = OVS_FLOW_VERSION,
645
	.maxattr = OVS_FLOW_ATTR_MAX,
646 647
	.netnsok = true,
	.parallel_ops = true,
648 649 650 651 652 653
};

static struct genl_multicast_group ovs_dp_flow_multicast_group = {
	.name = OVS_FLOW_MCGROUP
};

654 655 656 657
static size_t ovs_flow_cmd_msg_size(const struct sw_flow_actions *acts)
{
	return NLMSG_ALIGN(sizeof(struct ovs_header))
		+ nla_total_size(key_attr_size()) /* OVS_FLOW_ATTR_KEY */
658
		+ nla_total_size(key_attr_size()) /* OVS_FLOW_ATTR_MASK */
659 660 661 662 663 664
		+ nla_total_size(sizeof(struct ovs_flow_stats)) /* OVS_FLOW_ATTR_STATS */
		+ nla_total_size(1) /* OVS_FLOW_ATTR_TCP_FLAGS */
		+ nla_total_size(8) /* OVS_FLOW_ATTR_USED */
		+ nla_total_size(acts->actions_len); /* OVS_FLOW_ATTR_ACTIONS */
}

665
/* Called with ovs_mutex or RCU read lock. */
666
static int ovs_flow_cmd_fill_info(const struct sw_flow *flow, int dp_ifindex,
667
				  struct sk_buff *skb, u32 portid,
668 669 670
				  u32 seq, u32 flags, u8 cmd)
{
	const int skb_orig_len = skb->len;
671
	struct nlattr *start;
672
	struct ovs_flow_stats stats;
673 674
	__be16 tcp_flags;
	unsigned long used;
675 676 677 678
	struct ovs_header *ovs_header;
	struct nlattr *nla;
	int err;

679
	ovs_header = genlmsg_put(skb, portid, seq, &dp_flow_genl_family, flags, cmd);
680 681 682
	if (!ovs_header)
		return -EMSGSIZE;

683
	ovs_header->dp_ifindex = dp_ifindex;
684

685
	/* Fill flow key. */
686 687 688
	nla = nla_nest_start(skb, OVS_FLOW_ATTR_KEY);
	if (!nla)
		goto nla_put_failure;
689

690
	err = ovs_nla_put_flow(&flow->unmasked_key, &flow->unmasked_key, skb);
691 692 693 694 695 696 697 698
	if (err)
		goto error;
	nla_nest_end(skb, nla);

	nla = nla_nest_start(skb, OVS_FLOW_ATTR_MASK);
	if (!nla)
		goto nla_put_failure;

699
	err = ovs_nla_put_flow(&flow->key, &flow->mask->key, skb);
700 701
	if (err)
		goto error;
702

703 704
	nla_nest_end(skb, nla);

705
	ovs_flow_stats_get(flow, &stats, &used, &tcp_flags);
706

707 708 709
	if (used &&
	    nla_put_u64(skb, OVS_FLOW_ATTR_USED, ovs_flow_used_time(used)))
		goto nla_put_failure;
710

711
	if (stats.n_packets &&
712
	    nla_put(skb, OVS_FLOW_ATTR_STATS, sizeof(struct ovs_flow_stats), &stats))
713
		goto nla_put_failure;
714

715 716
	if ((u8)ntohs(tcp_flags) &&
	     nla_put_u8(skb, OVS_FLOW_ATTR_TCP_FLAGS, (u8)ntohs(tcp_flags)))
717
		goto nla_put_failure;
718 719 720 721 722 723 724 725 726 727 728

	/* If OVS_FLOW_ATTR_ACTIONS doesn't fit, skip dumping the actions if
	 * this is the first flow to be dumped into 'skb'.  This is unusual for
	 * Netlink but individual action lists can be longer than
	 * NLMSG_GOODSIZE and thus entirely undumpable if we didn't do this.
	 * The userspace caller can always fetch the actions separately if it
	 * really wants them.  (Most userspace callers in fact don't care.)
	 *
	 * This can only fail for dump operations because the skb is always
	 * properly sized for single flows.
	 */
729 730
	start = nla_nest_start(skb, OVS_FLOW_ATTR_ACTIONS);
	if (start) {
731 732
		const struct sw_flow_actions *sf_acts;

733
		sf_acts = rcu_dereference_ovsl(flow->sf_acts);
734 735
		err = ovs_nla_put_actions(sf_acts->actions,
					  sf_acts->actions_len, skb);
736

737 738 739 740 741 742 743 744 745 746
		if (!err)
			nla_nest_end(skb, start);
		else {
			if (skb_orig_len)
				goto error;

			nla_nest_cancel(skb, start);
		}
	} else if (skb_orig_len)
		goto nla_put_failure;
747 748 749 750 751 752 753 754 755 756

	return genlmsg_end(skb, ovs_header);

nla_put_failure:
	err = -EMSGSIZE;
error:
	genlmsg_cancel(skb, ovs_header);
	return err;
}

757 758
/* May not be called with RCU read lock. */
static struct sk_buff *ovs_flow_cmd_alloc_info(const struct sw_flow_actions *acts,
759 760
					       struct genl_info *info,
					       bool always)
761
{
762
	struct sk_buff *skb;
763

764 765 766
	if (!always && !ovs_must_notify(info, &ovs_dp_flow_multicast_group))
		return NULL;

767
	skb = genlmsg_new_unicast(ovs_flow_cmd_msg_size(acts), info, GFP_KERNEL);
768 769 770 771
	if (!skb)
		return ERR_PTR(-ENOMEM);

	return skb;
772 773
}

774 775 776 777 778
/* Called with ovs_mutex. */
static struct sk_buff *ovs_flow_cmd_build_info(const struct sw_flow *flow,
					       int dp_ifindex,
					       struct genl_info *info, u8 cmd,
					       bool always)
779 780 781 782
{
	struct sk_buff *skb;
	int retval;

783 784
	skb = ovs_flow_cmd_alloc_info(ovsl_dereference(flow->sf_acts), info,
				      always);
785 786
	if (!skb || IS_ERR(skb))
		return skb;
787

788 789 790
	retval = ovs_flow_cmd_fill_info(flow, dp_ifindex, skb,
					info->snd_portid, info->snd_seq, 0,
					cmd);
791 792 793 794
	BUG_ON(retval < 0);
	return skb;
}

795
static int ovs_flow_cmd_new(struct sk_buff *skb, struct genl_info *info)
796 797 798
{
	struct nlattr **a = info->attrs;
	struct ovs_header *ovs_header = info->userhdr;
799
	struct sw_flow_key key, masked_key;
800
	struct sw_flow *flow;
801
	struct sw_flow_mask mask;
802 803
	struct sk_buff *reply;
	struct datapath *dp;
804
	struct sw_flow_actions *acts;
805
	struct sw_flow_match match;
806 807 808 809 810 811
	int error;

	/* Extract key. */
	error = -EINVAL;
	if (!a[OVS_FLOW_ATTR_KEY])
		goto error;
812 813

	ovs_match_init(&match, &key, &mask);
814
	error = ovs_nla_get_match(&match,
815
				  a[OVS_FLOW_ATTR_KEY], a[OVS_FLOW_ATTR_MASK]);
816 817 818 819
	if (error)
		goto error;

	/* Validate actions. */
820 821 822
	error = -EINVAL;
	if (!a[OVS_FLOW_ATTR_ACTIONS])
		goto error;
823

824 825 826
	acts = ovs_nla_alloc_flow_actions(nla_len(a[OVS_FLOW_ATTR_ACTIONS]));
	error = PTR_ERR(acts);
	if (IS_ERR(acts))
827
		goto error;
828 829 830 831 832 833 834

	ovs_flow_mask_key(&masked_key, &key, &mask);
	error = ovs_nla_copy_actions(a[OVS_FLOW_ATTR_ACTIONS],
				     &masked_key, 0, &acts);
	if (error) {
		OVS_NLERR("Flow actions may not be safe on all matching packets.\n");
		goto err_kfree;
835 836
	}

837
	ovs_lock();
838
	dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
839 840
	error = -ENODEV;
	if (!dp)
841
		goto err_unlock_ovs;
842

843
	/* Check if this is a duplicate flow */
844
	flow = ovs_flow_tbl_lookup(&dp->table, &key);
845 846
	if (!flow) {
		/* Allocate flow. */
847
		flow = ovs_flow_alloc();
848 849
		if (IS_ERR(flow)) {
			error = PTR_ERR(flow);
850
			goto err_unlock_ovs;
851 852
		}

853 854
		flow->key = masked_key;
		flow->unmasked_key = key;
855 856 857
		rcu_assign_pointer(flow->sf_acts, acts);

		/* Put flow in bucket. */
858 859 860 861 862
		error = ovs_flow_tbl_insert(&dp->table, flow, &mask);
		if (error) {
			acts = NULL;
			goto err_flow_free;
		}
863
	} else {
864 865
		struct sw_flow_actions *old_acts;

866 867 868 869 870 871 872
		/* Bail out if we're not allowed to modify an existing flow.
		 * We accept NLM_F_CREATE in place of the intended NLM_F_EXCL
		 * because Generic Netlink treats the latter as a dump
		 * request.  We also accept NLM_F_EXCL in case that bug ever
		 * gets fixed.
		 */
		error = -EEXIST;
873
		if (info->nlhdr->nlmsg_flags & (NLM_F_CREATE | NLM_F_EXCL))
874
			goto err_unlock_ovs;
875

876
		/* The unmasked key has to be the same for flow updates. */
877
		if (!ovs_flow_cmp_unmasked_key(flow, &match))
878 879
			goto err_unlock_ovs;

880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907
		/* Update actions. */
		old_acts = ovsl_dereference(flow->sf_acts);
		rcu_assign_pointer(flow->sf_acts, acts);
		ovs_nla_free_flow_actions(old_acts);
	}

	reply = ovs_flow_cmd_build_info(flow, ovs_header->dp_ifindex,
					info, OVS_FLOW_CMD_NEW, false);
	ovs_unlock();

	if (reply) {
		if (!IS_ERR(reply))
			ovs_notify(&dp_flow_genl_family, reply, info);
		else
			netlink_set_err(sock_net(skb->sk)->genl_sock, 0, 0,
					PTR_ERR(reply));
	}
	return 0;

err_flow_free:
	ovs_flow_free(flow, false);
err_unlock_ovs:
	ovs_unlock();
err_kfree:
	kfree(acts);
error:
	return error;
}
908

909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945
static int ovs_flow_cmd_set(struct sk_buff *skb, struct genl_info *info)
{
	struct nlattr **a = info->attrs;
	struct ovs_header *ovs_header = info->userhdr;
	struct sw_flow_key key, masked_key;
	struct sw_flow *flow;
	struct sw_flow_mask mask;
	struct sk_buff *reply = NULL;
	struct datapath *dp;
	struct sw_flow_actions *acts = NULL;
	struct sw_flow_match match;
	int error;

	/* Extract key. */
	error = -EINVAL;
	if (!a[OVS_FLOW_ATTR_KEY])
		goto error;

	ovs_match_init(&match, &key, &mask);
	error = ovs_nla_get_match(&match,
				  a[OVS_FLOW_ATTR_KEY], a[OVS_FLOW_ATTR_MASK]);
	if (error)
		goto error;

	/* Validate actions. */
	if (a[OVS_FLOW_ATTR_ACTIONS]) {
		acts = ovs_nla_alloc_flow_actions(nla_len(a[OVS_FLOW_ATTR_ACTIONS]));
		error = PTR_ERR(acts);
		if (IS_ERR(acts))
			goto error;

		ovs_flow_mask_key(&masked_key, &key, &mask);
		error = ovs_nla_copy_actions(a[OVS_FLOW_ATTR_ACTIONS],
					     &masked_key, 0, &acts);
		if (error) {
			OVS_NLERR("Flow actions may not be safe on all matching packets.\n");
			goto err_kfree;
946
		}
947
	}
948

949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964
	ovs_lock();
	dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
	error = -ENODEV;
	if (!dp)
		goto err_unlock_ovs;

	/* Check that the flow exists. */
	flow = ovs_flow_tbl_lookup(&dp->table, &key);
	error = -ENOENT;
	if (!flow)
		goto err_unlock_ovs;

	/* The unmasked key has to be the same for flow updates. */
	error = -EEXIST;
	if (!ovs_flow_cmp_unmasked_key(flow, &match))
		goto err_unlock_ovs;
965

966 967 968 969 970 971 972
	/* Update actions, if present. */
	if (acts) {
		struct sw_flow_actions *old_acts;

		old_acts = ovsl_dereference(flow->sf_acts);
		rcu_assign_pointer(flow->sf_acts, acts);
		ovs_nla_free_flow_actions(old_acts);
973
	}
974 975 976 977 978 979

	reply = ovs_flow_cmd_build_info(flow, ovs_header->dp_ifindex,
					info, OVS_FLOW_CMD_NEW, false);
	/* Clear stats. */
	if (a[OVS_FLOW_ATTR_CLEAR])
		ovs_flow_stats_clear(flow);
980
	ovs_unlock();
981

982 983 984 985 986 987 988 989
	if (reply) {
		if (!IS_ERR(reply))
			ovs_notify(&dp_flow_genl_family, reply, info);
		else
			genl_set_err(&dp_flow_genl_family, sock_net(skb->sk), 0,
				     0, PTR_ERR(reply));
	}

990 991
	return 0;

992 993
err_unlock_ovs:
	ovs_unlock();
994 995
err_kfree:
	kfree(acts);
996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
error:
	return error;
}

static int ovs_flow_cmd_get(struct sk_buff *skb, struct genl_info *info)
{
	struct nlattr **a = info->attrs;
	struct ovs_header *ovs_header = info->userhdr;
	struct sw_flow_key key;
	struct sk_buff *reply;
	struct sw_flow *flow;
	struct datapath *dp;
1008
	struct sw_flow_match match;
1009 1010
	int err;

1011 1012
	if (!a[OVS_FLOW_ATTR_KEY]) {
		OVS_NLERR("Flow get message rejected, Key attribute missing.\n");
1013
		return -EINVAL;
1014 1015 1016
	}

	ovs_match_init(&match, &key, NULL);
1017
	err = ovs_nla_get_match(&match, a[OVS_FLOW_ATTR_KEY], NULL);
1018 1019 1020
	if (err)
		return err;

1021
	ovs_lock();
1022
	dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
1023 1024 1025 1026
	if (!dp) {
		err = -ENODEV;
		goto unlock;
	}
1027

1028
	flow = ovs_flow_tbl_lookup(&dp->table, &key);
1029
	if (!flow || !ovs_flow_cmp_unmasked_key(flow, &match)) {
1030 1031 1032
		err = -ENOENT;
		goto unlock;
	}
1033

1034 1035
	reply = ovs_flow_cmd_build_info(flow, ovs_header->dp_ifindex, info,
					OVS_FLOW_CMD_NEW, true);
1036 1037 1038 1039
	if (IS_ERR(reply)) {
		err = PTR_ERR(reply);
		goto unlock;
	}
1040

1041
	ovs_unlock();
1042
	return genlmsg_reply(reply, info);
1043 1044 1045
unlock:
	ovs_unlock();
	return err;
1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
}

static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
{
	struct nlattr **a = info->attrs;
	struct ovs_header *ovs_header = info->userhdr;
	struct sw_flow_key key;
	struct sk_buff *reply;
	struct sw_flow *flow;
	struct datapath *dp;
1056
	struct sw_flow_match match;
1057 1058
	int err;

1059 1060 1061 1062 1063 1064 1065
	if (likely(a[OVS_FLOW_ATTR_KEY])) {
		ovs_match_init(&match, &key, NULL);
		err = ovs_nla_get_match(&match, a[OVS_FLOW_ATTR_KEY], NULL);
		if (unlikely(err))
			return err;
	}

1066
	ovs_lock();
1067
	dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
1068
	if (unlikely(!dp)) {
1069 1070 1071
		err = -ENODEV;
		goto unlock;
	}
1072

1073
	if (unlikely(!a[OVS_FLOW_ATTR_KEY])) {
1074
		err = ovs_flow_tbl_flush(&dp->table);
1075 1076
		goto unlock;
	}
1077

1078
	flow = ovs_flow_tbl_lookup(&dp->table, &key);
1079
	if (unlikely(!flow || !ovs_flow_cmp_unmasked_key(flow, &match))) {
1080 1081