datapath.c 57.1 KB
Newer Older
1
/*
2
 * Copyright (c) 2007-2014 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 47 48 49 50
 *
 * 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>
#include <linux/openvswitch.h>
#include <linux/rculist.h>
#include <linux/dmi.h>
#include <net/genetlink.h>
51 52
#include <net/net_namespace.h>
#include <net/netns/generic.h>
53 54 55

#include "datapath.h"
#include "flow.h"
56
#include "flow_table.h"
57
#include "flow_netlink.h"
58
#include "vport-internal_dev.h"
59
#include "vport-netdev.h"
60

61
int ovs_net_id __read_mostly;
62
EXPORT_SYMBOL_GPL(ovs_net_id);
63

64 65 66 67
static struct genl_family dp_packet_genl_family;
static struct genl_family dp_flow_genl_family;
static struct genl_family dp_datapath_genl_family;

68 69
static const struct nla_policy flow_policy[];

70 71
static const struct genl_multicast_group ovs_dp_flow_multicast_group = {
	.name = OVS_FLOW_MCGROUP,
72 73
};

74 75
static const struct genl_multicast_group ovs_dp_datapath_multicast_group = {
	.name = OVS_DATAPATH_MCGROUP,
76 77
};

78 79
static const struct genl_multicast_group ovs_dp_vport_multicast_group = {
	.name = OVS_VPORT_MCGROUP,
80 81
};

82 83
/* Check if need to build a reply message.
 * OVS userspace sets the NLM_F_ECHO flag if it needs the reply. */
84 85
static bool ovs_must_notify(struct genl_family *family, struct genl_info *info,
			    unsigned int group)
86 87
{
	return info->nlhdr->nlmsg_flags & NLM_F_ECHO ||
88
	       genl_has_listeners(family, genl_info_net(info), group);
89 90
}

91
static void ovs_notify(struct genl_family *family,
92
		       struct sk_buff *skb, struct genl_info *info)
93
{
94
	genl_notify(family, skb, genl_info_net(info), info->snd_portid,
95
		    0, info->nlhdr, GFP_KERNEL);
96 97
}

98 99 100
/**
 * DOC: Locking:
 *
101 102 103 104
 * 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.
105 106 107 108 109 110
 *
 * 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.
111 112
 *
 * The RTNL lock nests inside ovs_mutex.
113 114
 */

115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134
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;
}
135
EXPORT_SYMBOL_GPL(lockdep_ovsl_is_held);
136 137
#endif

138
static struct vport *new_vport(const struct vport_parms *);
139
static int queue_gso_packets(struct datapath *dp, struct sk_buff *,
140
			     const struct sw_flow_key *,
141
			     const struct dp_upcall_info *);
142
static int queue_userspace_packet(struct datapath *dp, struct sk_buff *,
143
				  const struct sw_flow_key *,
144 145
				  const struct dp_upcall_info *);

146 147
/* Must be called with rcu_read_lock. */
static struct datapath *get_dp_rcu(struct net *net, int dp_ifindex)
148
{
149
	struct net_device *dev = dev_get_by_index_rcu(net, dp_ifindex);
150 151 152 153

	if (dev) {
		struct vport *vport = ovs_internal_dev_get_vport(dev);
		if (vport)
154
			return vport->dp;
155
	}
156 157 158 159 160 161 162 163 164 165 166 167 168 169

	return NULL;
}

/* The caller must hold either ovs_mutex or rcu_read_lock to keep the
 * returned dp pointer valid.
 */
static inline struct datapath *get_dp(struct net *net, int dp_ifindex)
{
	struct datapath *dp;

	WARN_ON_ONCE(!rcu_read_lock_held() && !lockdep_ovsl_is_held());
	rcu_read_lock();
	dp = get_dp_rcu(net, dp_ifindex);
170 171 172 173 174
	rcu_read_unlock();

	return dp;
}

175
/* Must be called with rcu_read_lock or ovs_mutex. */
176
const char *ovs_dp_name(const struct datapath *dp)
177
{
178
	struct vport *vport = ovs_vport_ovsl_rcu(dp, OVSP_LOCAL);
179
	return ovs_vport_name(vport);
180 181
}

182
static int get_dpifindex(const struct datapath *dp)
183 184 185 186 187 188
{
	struct vport *local;
	int ifindex;

	rcu_read_lock();

189
	local = ovs_vport_rcu(dp, OVSP_LOCAL);
190
	if (local)
191
		ifindex = local->dev->ifindex;
192 193 194 195 196 197 198 199 200 201 202 203
	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);

204
	ovs_flow_tbl_destroy(&dp->table);
205
	free_percpu(dp->stats_percpu);
206
	kfree(dp->ports);
207 208 209
	kfree(dp);
}

210 211 212 213 214 215
static struct hlist_head *vport_hash_bucket(const struct datapath *dp,
					    u16 port_no)
{
	return &dp->ports[port_no & (DP_VPORT_HASH_BUCKETS - 1)];
}

216
/* Called with ovs_mutex or RCU read lock. */
217 218 219 220 221 222
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);
223
	hlist_for_each_entry_rcu(vport, head, dp_hash_node) {
224 225 226 227 228 229
		if (vport->port_no == port_no)
			return vport;
	}
	return NULL;
}

230
/* Called with ovs_mutex. */
231 232 233 234 235 236 237
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;
238
		struct hlist_head *head = vport_hash_bucket(dp, vport->port_no);
239

240
		hlist_add_head_rcu(&vport->dp_hash_node, head);
241 242 243 244 245 246
	}
	return vport;
}

void ovs_dp_detach_port(struct vport *p)
{
247
	ASSERT_OVSL();
248 249

	/* First drop references to device. */
250
	hlist_del_rcu(&p->dp_hash_node);
251 252 253 254 255 256

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

/* Must be called with rcu_read_lock. */
257
void ovs_dp_process_packet(struct sk_buff *skb, struct sw_flow_key *key)
258
{
259
	const struct vport *p = OVS_CB(skb)->input_vport;
260 261
	struct datapath *dp = p->dp;
	struct sw_flow *flow;
262
	struct sw_flow_actions *sf_acts;
263 264
	struct dp_stats_percpu *stats;
	u64 *stats_counter;
265
	u32 n_mask_hit;
266

267
	stats = this_cpu_ptr(dp->stats_percpu);
268 269

	/* Look up flow. */
270
	flow = ovs_flow_tbl_lookup_stats(&dp->table, key, &n_mask_hit);
271 272
	if (unlikely(!flow)) {
		struct dp_upcall_info upcall;
273
		int error;
274

275
		memset(&upcall, 0, sizeof(upcall));
276
		upcall.cmd = OVS_PACKET_CMD_MISS;
277
		upcall.portid = ovs_vport_find_upcall_portid(p, skb);
278
		upcall.mru = OVS_CB(skb)->mru;
279
		error = ovs_dp_upcall(dp, skb, key, &upcall);
280 281 282 283
		if (unlikely(error))
			kfree_skb(skb);
		else
			consume_skb(skb);
284 285 286 287
		stats_counter = &stats->n_missed;
		goto out;
	}

288 289 290
	ovs_flow_stats_update(flow, key->tp.flags, skb);
	sf_acts = rcu_dereference(flow->sf_acts);
	ovs_execute_actions(dp, skb, sf_acts, key);
291

292
	stats_counter = &stats->n_hit;
293 294 295

out:
	/* Update datapath statistics. */
296
	u64_stats_update_begin(&stats->syncp);
297
	(*stats_counter)++;
298
	stats->n_mask_hit += n_mask_hit;
299
	u64_stats_update_end(&stats->syncp);
300 301 302
}

int ovs_dp_upcall(struct datapath *dp, struct sk_buff *skb,
303
		  const struct sw_flow_key *key,
304
		  const struct dp_upcall_info *upcall_info)
305 306 307 308
{
	struct dp_stats_percpu *stats;
	int err;

309
	if (upcall_info->portid == 0) {
310 311 312 313 314
		err = -ENOTCONN;
		goto err;
	}

	if (!skb_is_gso(skb))
315
		err = queue_userspace_packet(dp, skb, key, upcall_info);
316
	else
317
		err = queue_gso_packets(dp, skb, key, upcall_info);
318 319 320 321 322 323
	if (err)
		goto err;

	return 0;

err:
324
	stats = this_cpu_ptr(dp->stats_percpu);
325

326
	u64_stats_update_begin(&stats->syncp);
327
	stats->n_lost++;
328
	u64_stats_update_end(&stats->syncp);
329 330 331 332

	return err;
}

333
static int queue_gso_packets(struct datapath *dp, struct sk_buff *skb,
334
			     const struct sw_flow_key *key,
335 336
			     const struct dp_upcall_info *upcall_info)
{
337
	unsigned short gso_type = skb_shinfo(skb)->gso_type;
338 339
	struct sw_flow_key later_key;
	struct sk_buff *segs, *nskb;
340
	struct ovs_skb_cb ovs_cb;
341 342
	int err;

343
	ovs_cb = *OVS_CB(skb);
344
	segs = __skb_gso_segment(skb, NETIF_F_SG, false);
345
	*OVS_CB(skb) = ovs_cb;
346 347
	if (IS_ERR(segs))
		return PTR_ERR(segs);
348 349
	if (segs == NULL)
		return -EINVAL;
350

351 352 353 354 355 356 357 358 359
	if (gso_type & SKB_GSO_UDP) {
		/* The initial flow key extracted by ovs_flow_key_extract()
		 * in this case is for a first fragment, so we need to
		 * properly mark later fragments.
		 */
		later_key = *key;
		later_key.ip.frag = OVS_FRAG_TYPE_LATER;
	}

360 361 362
	/* Queue all of the segments. */
	skb = segs;
	do {
363 364 365 366 367
		*OVS_CB(skb) = ovs_cb;
		if (gso_type & SKB_GSO_UDP && skb != segs)
			key = &later_key;

		err = queue_userspace_packet(dp, skb, key, upcall_info);
368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384
		if (err)
			break;

	} 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;
}

385
static size_t upcall_msg_size(const struct dp_upcall_info *upcall_info,
386
			      unsigned int hdrlen)
387 388
{
	size_t size = NLMSG_ALIGN(sizeof(struct ovs_header))
389
		+ nla_total_size(hdrlen) /* OVS_PACKET_ATTR_PACKET */
390
		+ nla_total_size(ovs_key_attr_size()); /* OVS_PACKET_ATTR_KEY */
391 392

	/* OVS_PACKET_ATTR_USERDATA */
393 394 395 396 397 398
	if (upcall_info->userdata)
		size += NLA_ALIGN(upcall_info->userdata->nla_len);

	/* OVS_PACKET_ATTR_EGRESS_TUN_KEY */
	if (upcall_info->egress_tun_info)
		size += nla_total_size(ovs_tun_key_attr_size());
399

400 401 402 403
	/* OVS_PACKET_ATTR_ACTIONS */
	if (upcall_info->actions_len)
		size += nla_total_size(upcall_info->actions_len);

404 405 406 407
	/* OVS_PACKET_ATTR_MRU */
	if (upcall_info->mru)
		size += nla_total_size(sizeof(upcall_info->mru));

408 409 410
	return size;
}

411 412 413 414 415 416 417 418 419 420
static void pad_packet(struct datapath *dp, struct sk_buff *skb)
{
	if (!(dp->user_features & OVS_DP_F_UNALIGNED)) {
		size_t plen = NLA_ALIGN(skb->len) - skb->len;

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

421
static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
422
				  const struct sw_flow_key *key,
423 424 425 426
				  const struct dp_upcall_info *upcall_info)
{
	struct ovs_header *upcall;
	struct sk_buff *nskb = NULL;
Li RongQing's avatar
Li RongQing committed
427
	struct sk_buff *user_skb = NULL; /* to be queued to userspace */
428
	struct nlattr *nla;
429
	struct genl_info info = {
430
		.dst_sk = ovs_dp_get_net(dp)->genl_sock,
431 432 433
		.snd_portid = upcall_info->portid,
	};
	size_t len;
434
	unsigned int hlen;
435 436 437 438 439
	int err, dp_ifindex;

	dp_ifindex = get_dpifindex(dp);
	if (!dp_ifindex)
		return -ENODEV;
440

441
	if (skb_vlan_tag_present(skb)) {
442 443 444 445
		nskb = skb_clone(skb, GFP_ATOMIC);
		if (!nskb)
			return -ENOMEM;

446
		nskb = __vlan_hwaccel_push_inside(nskb);
447
		if (!nskb)
448 449 450 451 452 453 454 455 456 457
			return -ENOMEM;

		skb = nskb;
	}

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

458 459 460 461 462 463 464 465 466 467 468 469 470 471
	/* 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;

472
	len = upcall_msg_size(upcall_info, hlen);
473
	user_skb = genlmsg_new_unicast(len, &info, GFP_ATOMIC);
474 475 476 477 478 479 480 481 482
	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;

483
	err = ovs_nla_put_key(key, key, OVS_PACKET_ATTR_KEY, false, user_skb);
484
	BUG_ON(err);
485 486

	if (upcall_info->userdata)
487 488 489
		__nla_put(user_skb, OVS_PACKET_ATTR_USERDATA,
			  nla_len(upcall_info->userdata),
			  nla_data(upcall_info->userdata));
490

491 492 493
	if (upcall_info->egress_tun_info) {
		nla = nla_nest_start(user_skb, OVS_PACKET_ATTR_EGRESS_TUN_KEY);
		err = ovs_nla_put_egress_tunnel_key(user_skb,
494 495
						    upcall_info->egress_tun_info,
						    upcall_info->egress_tun_opts);
496 497 498 499
		BUG_ON(err);
		nla_nest_end(user_skb, nla);
	}

500 501 502 503 504 505 506 507 508 509 510
	if (upcall_info->actions_len) {
		nla = nla_nest_start(user_skb, OVS_PACKET_ATTR_ACTIONS);
		err = ovs_nla_put_actions(upcall_info->actions,
					  upcall_info->actions_len,
					  user_skb);
		if (!err)
			nla_nest_end(user_skb, nla);
		else
			nla_nest_cancel(user_skb, nla);
	}

511 512 513 514 515 516 517 518 519 520
	/* Add OVS_PACKET_ATTR_MRU */
	if (upcall_info->mru) {
		if (nla_put_u16(user_skb, OVS_PACKET_ATTR_MRU,
				upcall_info->mru)) {
			err = -ENOBUFS;
			goto out;
		}
		pad_packet(dp, user_skb);
	}

521 522 523 524 525 526 527
	/* 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);
528

529 530 531
	err = skb_zerocopy(user_skb, skb, skb->len, hlen);
	if (err)
		goto out;
532

533
	/* Pad OVS_PACKET_ATTR_PACKET if linear copy was performed */
534
	pad_packet(dp, user_skb);
535

536
	((struct nlmsghdr *) user_skb->data)->nlmsg_len = user_skb->len;
537

538
	err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid);
Li RongQing's avatar
Li RongQing committed
539
	user_skb = NULL;
540
out:
541 542
	if (err)
		skb_tx_error(skb);
Li RongQing's avatar
Li RongQing committed
543
	kfree_skb(user_skb);
544 545 546 547 548 549 550
	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;
551
	struct net *net = sock_net(skb->sk);
552 553 554 555
	struct nlattr **a = info->attrs;
	struct sw_flow_actions *acts;
	struct sk_buff *packet;
	struct sw_flow *flow;
556
	struct sw_flow_actions *sf_acts;
557 558
	struct datapath *dp;
	struct ethhdr *eth;
559
	struct vport *input_vport;
560
	u16 mru = 0;
561 562
	int len;
	int err;
563
	bool log = !a[OVS_PACKET_ATTR_PROBE];
564 565 566

	err = -EINVAL;
	if (!a[OVS_PACKET_ATTR_PACKET] || !a[OVS_PACKET_ATTR_KEY] ||
567
	    !a[OVS_PACKET_ATTR_ACTIONS])
568 569 570 571 572 573 574 575 576
		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);

577
	nla_memcpy(__skb_put(packet, len), a[OVS_PACKET_ATTR_PACKET], len);
578 579 580 581 582 583 584

	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. */
585
	if (eth_proto_is_802_3(eth->h_proto))
586 587 588 589
		packet->protocol = eth->h_proto;
	else
		packet->protocol = htons(ETH_P_802_2);

590 591 592 593 594 595 596
	/* Set packet's mru */
	if (a[OVS_PACKET_ATTR_MRU]) {
		mru = nla_get_u16(a[OVS_PACKET_ATTR_MRU]);
		packet->ignore_df = 1;
	}
	OVS_CB(packet)->mru = mru;

597
	/* Build an sw_flow for sending this packet. */
598
	flow = ovs_flow_alloc();
599 600 601 602
	err = PTR_ERR(flow);
	if (IS_ERR(flow))
		goto err_kfree_skb;

603 604
	err = ovs_flow_key_extract_userspace(net, a[OVS_PACKET_ATTR_KEY],
					     packet, &flow->key, log);
605 606 607
	if (err)
		goto err_flow_free;

608
	err = ovs_nla_copy_actions(net, a[OVS_PACKET_ATTR_ACTIONS],
609
				   &flow->key, &acts, log);
610 611
	if (err)
		goto err_flow_free;
612

613
	rcu_assign_pointer(flow->sf_acts, acts);
614
	packet->priority = flow->key.phy.priority;
615
	packet->mark = flow->key.phy.skb_mark;
616 617

	rcu_read_lock();
618
	dp = get_dp_rcu(net, ovs_header->dp_ifindex);
619 620 621 622
	err = -ENODEV;
	if (!dp)
		goto err_unlock;

623 624 625 626 627 628 629
	input_vport = ovs_vport_rcu(dp, flow->key.phy.in_port);
	if (!input_vport)
		input_vport = ovs_vport_rcu(dp, OVSP_LOCAL);

	if (!input_vport)
		goto err_unlock;

630
	packet->dev = input_vport->dev;
631
	OVS_CB(packet)->input_vport = input_vport;
632
	sf_acts = rcu_dereference(flow->sf_acts);
633

634
	local_bh_disable();
635
	err = ovs_execute_actions(dp, packet, sf_acts, &flow->key);
636 637 638
	local_bh_enable();
	rcu_read_unlock();

639
	ovs_flow_free(flow, false);
640 641 642 643 644
	return err;

err_unlock:
	rcu_read_unlock();
err_flow_free:
645
	ovs_flow_free(flow, false);
646 647 648 649 650 651 652
err_kfree_skb:
	kfree_skb(packet);
err:
	return err;
}

static const struct nla_policy packet_policy[OVS_PACKET_ATTR_MAX + 1] = {
653
	[OVS_PACKET_ATTR_PACKET] = { .len = ETH_HLEN },
654 655
	[OVS_PACKET_ATTR_KEY] = { .type = NLA_NESTED },
	[OVS_PACKET_ATTR_ACTIONS] = { .type = NLA_NESTED },
656
	[OVS_PACKET_ATTR_PROBE] = { .type = NLA_FLAG },
657
	[OVS_PACKET_ATTR_MRU] = { .type = NLA_U16 },
658 659
};

660
static const struct genl_ops dp_packet_genl_ops[] = {
661 662 663 664 665 666 667
	{ .cmd = OVS_PACKET_CMD_EXECUTE,
	  .flags = GENL_ADMIN_PERM, /* Requires CAP_NET_ADMIN privilege. */
	  .policy = packet_policy,
	  .doit = ovs_packet_cmd_execute
	}
};

668 669 670 671 672 673 674 675 676 677 678 679
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,
	.maxattr = OVS_PACKET_ATTR_MAX,
	.netnsok = true,
	.parallel_ops = true,
	.ops = dp_packet_genl_ops,
	.n_ops = ARRAY_SIZE(dp_packet_genl_ops),
};

680
static void get_dp_stats(const struct datapath *dp, struct ovs_dp_stats *stats,
681
			 struct ovs_dp_megaflow_stats *mega_stats)
682 683 684
{
	int i;

685 686
	memset(mega_stats, 0, sizeof(*mega_stats));

687
	stats->n_flows = ovs_flow_tbl_count(&dp->table);
688
	mega_stats->n_masks = ovs_flow_tbl_num_masks(&dp->table);
689 690

	stats->n_hit = stats->n_missed = stats->n_lost = 0;
691

692 693 694 695 696 697 698 699
	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 {
700
			start = u64_stats_fetch_begin_irq(&percpu_stats->syncp);
701
			local_stats = *percpu_stats;
702
		} while (u64_stats_fetch_retry_irq(&percpu_stats->syncp, start));
703 704 705 706

		stats->n_hit += local_stats.n_hit;
		stats->n_missed += local_stats.n_missed;
		stats->n_lost += local_stats.n_lost;
707
		mega_stats->n_mask_hit += local_stats.n_mask_hit;
708 709 710
	}
}

711 712 713 714 715 716 717 718 719 720 721 722
static bool should_fill_key(const struct sw_flow_id *sfid, uint32_t ufid_flags)
{
	return ovs_identifier_is_ufid(sfid) &&
	       !(ufid_flags & OVS_UFID_F_OMIT_KEY);
}

static bool should_fill_mask(uint32_t ufid_flags)
{
	return !(ufid_flags & OVS_UFID_F_OMIT_MASK);
}

static bool should_fill_actions(uint32_t ufid_flags)
723
{
724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746
	return !(ufid_flags & OVS_UFID_F_OMIT_ACTIONS);
}

static size_t ovs_flow_cmd_msg_size(const struct sw_flow_actions *acts,
				    const struct sw_flow_id *sfid,
				    uint32_t ufid_flags)
{
	size_t len = NLMSG_ALIGN(sizeof(struct ovs_header));

	/* OVS_FLOW_ATTR_UFID */
	if (sfid && ovs_identifier_is_ufid(sfid))
		len += nla_total_size(sfid->ufid_len);

	/* OVS_FLOW_ATTR_KEY */
	if (!sfid || should_fill_key(sfid, ufid_flags))
		len += nla_total_size(ovs_key_attr_size());

	/* OVS_FLOW_ATTR_MASK */
	if (should_fill_mask(ufid_flags))
		len += nla_total_size(ovs_key_attr_size());

	/* OVS_FLOW_ATTR_ACTIONS */
	if (should_fill_actions(ufid_flags))
747
		len += nla_total_size(acts->orig_len);
748 749

	return len
750 751
		+ nla_total_size(sizeof(struct ovs_flow_stats)) /* OVS_FLOW_ATTR_STATS */
		+ nla_total_size(1) /* OVS_FLOW_ATTR_TCP_FLAGS */
752
		+ nla_total_size(8); /* OVS_FLOW_ATTR_USED */
753 754
}

755 756 757 758 759 760 761
/* Called with ovs_mutex or RCU read lock. */
static int ovs_flow_cmd_fill_stats(const struct sw_flow *flow,
				   struct sk_buff *skb)
{
	struct ovs_flow_stats stats;
	__be16 tcp_flags;
	unsigned long used;
762

763
	ovs_flow_stats_get(flow, &stats, &used, &tcp_flags);
764

765 766
	if (used &&
	    nla_put_u64(skb, OVS_FLOW_ATTR_USED, ovs_flow_used_time(used)))
767
		return -EMSGSIZE;
768

769
	if (stats.n_packets &&
770
	    nla_put(skb, OVS_FLOW_ATTR_STATS, sizeof(struct ovs_flow_stats), &stats))
771
		return -EMSGSIZE;
772

773 774
	if ((u8)ntohs(tcp_flags) &&
	     nla_put_u8(skb, OVS_FLOW_ATTR_TCP_FLAGS, (u8)ntohs(tcp_flags)))
775 776 777 778 779 780 781 782 783 784 785
		return -EMSGSIZE;

	return 0;
}

/* Called with ovs_mutex or RCU read lock. */
static int ovs_flow_cmd_fill_actions(const struct sw_flow *flow,
				     struct sk_buff *skb, int skb_orig_len)
{
	struct nlattr *start;
	int err;
786 787 788 789 790 791 792 793 794 795 796

	/* 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.
	 */
797 798
	start = nla_nest_start(skb, OVS_FLOW_ATTR_ACTIONS);
	if (start) {
799 800
		const struct sw_flow_actions *sf_acts;

801
		sf_acts = rcu_dereference_ovsl(flow->sf_acts);
802 803
		err = ovs_nla_put_actions(sf_acts->actions,
					  sf_acts->actions_len, skb);
804

805 806 807 808
		if (!err)
			nla_nest_end(skb, start);
		else {
			if (skb_orig_len)
809
				return err;
810 811 812

			nla_nest_cancel(skb, start);
		}
813 814 815 816 817 818 819 820 821 822
	} else if (skb_orig_len) {
		return -EMSGSIZE;
	}

	return 0;
}

/* Called with ovs_mutex or RCU read lock. */
static int ovs_flow_cmd_fill_info(const struct sw_flow *flow, int dp_ifindex,
				  struct sk_buff *skb, u32 portid,
823
				  u32 seq, u32 flags, u8 cmd, u32 ufid_flags)
824 825 826 827 828 829 830 831 832 833 834 835
{
	const int skb_orig_len = skb->len;
	struct ovs_header *ovs_header;
	int err;

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

	ovs_header->dp_ifindex = dp_ifindex;

836
	err = ovs_nla_put_identifier(flow, skb);
837 838 839
	if (err)
		goto error;

840 841 842 843 844 845 846 847 848 849 850
	if (should_fill_key(&flow->id, ufid_flags)) {
		err = ovs_nla_put_masked_key(flow, skb);
		if (err)
			goto error;
	}

	if (should_fill_mask(ufid_flags)) {
		err = ovs_nla_put_mask(flow, skb);
		if (err)
			goto error;
	}
851 852 853 854 855

	err = ovs_flow_cmd_fill_stats(flow, skb);
	if (err)
		goto error;

856 857 858 859 860
	if (should_fill_actions(ufid_flags)) {
		err = ovs_flow_cmd_fill_actions(flow, skb, skb_orig_len);
		if (err)
			goto error;
	}
861

862 863
	genlmsg_end(skb, ovs_header);
	return 0;
864 865 866 867 868 869

error:
	genlmsg_cancel(skb, ovs_header);
	return err;
}

870 871
/* May not be called with RCU read lock. */
static struct sk_buff *ovs_flow_cmd_alloc_info(const struct sw_flow_actions *acts,
872
					       const struct sw_flow_id *sfid,
873
					       struct genl_info *info,
874 875
					       bool always,
					       uint32_t ufid_flags)
876
{
877
	struct sk_buff *skb;
878
	size_t len;
879

880
	if (!always && !ovs_must_notify(&dp_flow_genl_family, info, 0))
881 882
		return NULL;

883 884
	len = ovs_flow_cmd_msg_size(acts, sfid, ufid_flags);
	skb = genlmsg_new_unicast(len, info, GFP_KERNEL);
885 886 887 888
	if (!skb)
		return ERR_PTR(-ENOMEM);

	return skb;
889 890
}

891 892 893 894
/* 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,
895
					       bool always, u32 ufid_flags)
896 897 898 899
{
	struct sk_buff *skb;
	int retval;

900 901
	skb = ovs_flow_cmd_alloc_info(ovsl_dereference(flow->sf_acts),
				      &flow->id, info, always, ufid_flags);
902
	if (IS_ERR_OR_NULL(skb))
903
		return skb;
904

905 906
	retval = ovs_flow_cmd_fill_info(flow, dp_ifindex, skb,
					info->snd_portid, info->snd_seq, 0,
907
					cmd, ufid_flags);
908 909 910 911
	BUG_ON(retval < 0);
	return skb;
}

912
static int ovs_flow_cmd_new(struct sk_buff *skb, struct genl_info *info)
913
{
914
	struct net *net = sock_net(skb->sk);
915 916
	struct nlattr **a = info->attrs;
	struct ovs_header *ovs_header = info->userhdr;
917
	struct sw_flow *flow = NULL, *new_flow;
918
	struct sw_flow_mask mask;
919 920
	struct sk_buff *reply;
	struct datapath *dp;
921
	struct sw_flow_key key;
922
	struct sw_flow_actions *acts;
923
	struct sw_flow_match match;
924
	u32 ufid_flags = ovs_nla_get_ufid_flags(a[OVS_FLOW_ATTR_UFID_FLAGS]);
925
	int error;
926
	bool log = !a[OVS_FLOW_ATTR_PROBE];
927

928
	/* Must have key and actions. */
929
	error = -EINVAL;
930
	if (!a[OVS_FLOW_ATTR_KEY]) {
931
		OVS_NLERR(log, "Flow key attr not present in new flow.");
932
		goto error;
933 934
	}
	if (!a[OVS_FLOW_ATTR_ACTIONS]) {
935
		OVS_NLERR(log, "Flow actions attr not present in new flow.");
936
		goto error;
937
	}
938

939 940 941 942 943 944 945 946 947 948
	/* Most of the time we need to allocate a new flow, do it before
	 * locking.
	 */
	new_flow = ovs_flow_alloc();
	if (IS_ERR(new_flow)) {
		error = PTR_ERR(new_flow);
		goto error;
	}

	/* Extract key. */
949
	ovs_match_init(&match, &key, &mask);
950
	error = ovs_nla_get_match(net, &match, a[OVS_FLOW_ATTR_KEY],
951
				  a[OVS_FLOW_ATTR_MASK], log);
Jesse Gross's avatar