ipmr.c 63.8 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3
/*
 *	IP multicast routing support for mrouted 3.6/3.8
 *
4
 *		(c) 1995 Alan Cox, <alan@lxorguk.ukuu.org.uk>
L
Linus Torvalds 已提交
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
 *	  Linux Consultancy and Custom Driver Development
 *
 *	This program is free software; you can redistribute it and/or
 *	modify it under the terms of the GNU General Public License
 *	as published by the Free Software Foundation; either version
 *	2 of the License, or (at your option) any later version.
 *
 *	Fixes:
 *	Michael Chastain	:	Incorrect size of copying.
 *	Alan Cox		:	Added the cache manager code
 *	Alan Cox		:	Fixed the clone/copy bug and device race.
 *	Mike McLagan		:	Routing by source
 *	Malcolm Beattie		:	Buffer handling fixes.
 *	Alexey Kuznetsov	:	Double buffer free and other fixes.
 *	SVR Anand		:	Fixed several multicast bugs and problems.
 *	Alexey Kuznetsov	:	Status, optimisations and more.
 *	Brad Parker		:	Better behaviour on mrouted upcall
 *					overflow.
 *      Carlos Picoto           :       PIMv1 Support
 *	Pavlin Ivanov Radoslavov:	PIMv2 Registers must checksum only PIM header
25
 *					Relax this requirement to work with older peers.
L
Linus Torvalds 已提交
26 27 28 29 30
 *
 */

#include <asm/uaccess.h>
#include <linux/types.h>
31
#include <linux/capability.h>
L
Linus Torvalds 已提交
32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
#include <linux/errno.h>
#include <linux/timer.h>
#include <linux/mm.h>
#include <linux/kernel.h>
#include <linux/fcntl.h>
#include <linux/stat.h>
#include <linux/socket.h>
#include <linux/in.h>
#include <linux/inet.h>
#include <linux/netdevice.h>
#include <linux/inetdevice.h>
#include <linux/igmp.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/mroute.h>
#include <linux/init.h>
48
#include <linux/if_ether.h>
49
#include <linux/slab.h>
50
#include <net/net_namespace.h>
L
Linus Torvalds 已提交
51 52 53
#include <net/ip.h>
#include <net/protocol.h>
#include <linux/skbuff.h>
54
#include <net/route.h>
L
Linus Torvalds 已提交
55 56 57 58 59 60 61
#include <net/sock.h>
#include <net/icmp.h>
#include <net/udp.h>
#include <net/raw.h>
#include <linux/notifier.h>
#include <linux/if_arp.h>
#include <linux/netfilter_ipv4.h>
62
#include <linux/compat.h>
63
#include <linux/export.h>
64
#include <net/ip_tunnels.h>
L
Linus Torvalds 已提交
65
#include <net/checksum.h>
66
#include <net/netlink.h>
67
#include <net/fib_rules.h>
68
#include <linux/netconf.h>
L
Linus Torvalds 已提交
69 70 71 72 73

#if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2)
#define CONFIG_IP_PIMSM	1
#endif

74
struct mr_table {
75
	struct list_head	list;
76 77 78
#ifdef CONFIG_NET_NS
	struct net		*net;
#endif
79
	u32			id;
E
Eric Dumazet 已提交
80
	struct sock __rcu	*mroute_sk;
81 82 83 84 85 86
	struct timer_list	ipmr_expire_timer;
	struct list_head	mfc_unres_queue;
	struct list_head	mfc_cache_array[MFC_LINES];
	struct vif_device	vif_table[MAXVIFS];
	int			maxvif;
	atomic_t		cache_resolve_queue_len;
87 88
	bool			mroute_do_assert;
	bool			mroute_do_pim;
89 90 91 92 93
#if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2)
	int			mroute_reg_vif_num;
#endif
};

94 95 96 97 98 99 100 101
struct ipmr_rule {
	struct fib_rule		common;
};

struct ipmr_result {
	struct mr_table		*mrt;
};

L
Linus Torvalds 已提交
102
/* Big lock, protecting vif table, mrt cache and mroute socket state.
E
Eric Dumazet 已提交
103
 * Note that the changes are semaphored via rtnl_lock.
L
Linus Torvalds 已提交
104 105 106 107 108 109 110 111
 */

static DEFINE_RWLOCK(mrt_lock);

/*
 *	Multicast router control variables
 */

112
#define VIF_EXISTS(_mrt, _idx) ((_mrt)->vif_table[_idx].dev != NULL)
L
Linus Torvalds 已提交
113 114 115 116 117

/* Special spinlock for queue of unresolved entries */
static DEFINE_SPINLOCK(mfc_unres_lock);

/* We return to original Alan's scheme. Hash table of resolved
E
Eric Dumazet 已提交
118 119 120 121 122
 * entries is changed only in process context and protected
 * with weak lock mrt_lock. Queue of unresolved entries is protected
 * with strong spinlock mfc_unres_lock.
 *
 * In this case data path is free of exclusive locks at all.
L
Linus Torvalds 已提交
123 124
 */

125
static struct kmem_cache *mrt_cachep __read_mostly;
L
Linus Torvalds 已提交
126

127
static struct mr_table *ipmr_new_table(struct net *net, u32 id);
128 129
static void ipmr_free_table(struct mr_table *mrt);

130 131 132
static void ip_mr_forward(struct net *net, struct mr_table *mrt,
			  struct sk_buff *skb, struct mfc_cache *cache,
			  int local);
133
static int ipmr_cache_report(struct mr_table *mrt,
134
			     struct sk_buff *pkt, vifi_t vifi, int assert);
135 136
static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
			      struct mfc_cache *c, struct rtmsg *rtm);
137 138
static void mroute_netlink_event(struct mr_table *mrt, struct mfc_cache *mfc,
				 int cmd);
139
static void mroute_clean_tables(struct mr_table *mrt);
140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156
static void ipmr_expire_process(unsigned long arg);

#ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES
#define ipmr_for_each_table(mrt, net) \
	list_for_each_entry_rcu(mrt, &net->ipv4.mr_tables, list)

static struct mr_table *ipmr_get_table(struct net *net, u32 id)
{
	struct mr_table *mrt;

	ipmr_for_each_table(mrt, net) {
		if (mrt->id == id)
			return mrt;
	}
	return NULL;
}

D
David S. Miller 已提交
157
static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4,
158 159 160
			   struct mr_table **mrt)
{
	int err;
161 162 163 164 165
	struct ipmr_result res;
	struct fib_lookup_arg arg = {
		.result = &res,
		.flags = FIB_LOOKUP_NOREF,
	};
166

D
David S. Miller 已提交
167 168
	err = fib_rules_lookup(net->ipv4.mr_rules_ops,
			       flowi4_to_flowi(flp4), 0, &arg);
169 170 171 172 173 174 175 176 177 178 179
	if (err < 0)
		return err;
	*mrt = res.mrt;
	return 0;
}

static int ipmr_rule_action(struct fib_rule *rule, struct flowi *flp,
			    int flags, struct fib_lookup_arg *arg)
{
	struct ipmr_result *res = arg->result;
	struct mr_table *mrt;
L
Linus Torvalds 已提交
180

181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
	switch (rule->action) {
	case FR_ACT_TO_TBL:
		break;
	case FR_ACT_UNREACHABLE:
		return -ENETUNREACH;
	case FR_ACT_PROHIBIT:
		return -EACCES;
	case FR_ACT_BLACKHOLE:
	default:
		return -EINVAL;
	}

	mrt = ipmr_get_table(rule->fr_net, rule->table);
	if (mrt == NULL)
		return -EAGAIN;
	res->mrt = mrt;
	return 0;
}

static int ipmr_rule_match(struct fib_rule *rule, struct flowi *fl, int flags)
{
	return 1;
}

static const struct nla_policy ipmr_rule_policy[FRA_MAX + 1] = {
	FRA_GENERIC_POLICY,
};

static int ipmr_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
			       struct fib_rule_hdr *frh, struct nlattr **tb)
{
	return 0;
}

static int ipmr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh,
			     struct nlattr **tb)
{
	return 1;
}

static int ipmr_rule_fill(struct fib_rule *rule, struct sk_buff *skb,
			  struct fib_rule_hdr *frh)
{
	frh->dst_len = 0;
	frh->src_len = 0;
	frh->tos     = 0;
	return 0;
}

230
static const struct fib_rules_ops __net_initconst ipmr_rules_ops_template = {
231
	.family		= RTNL_FAMILY_IPMR,
232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
	.rule_size	= sizeof(struct ipmr_rule),
	.addr_size	= sizeof(u32),
	.action		= ipmr_rule_action,
	.match		= ipmr_rule_match,
	.configure	= ipmr_rule_configure,
	.compare	= ipmr_rule_compare,
	.default_pref	= fib_default_rule_pref,
	.fill		= ipmr_rule_fill,
	.nlgroup	= RTNLGRP_IPV4_RULE,
	.policy		= ipmr_rule_policy,
	.owner		= THIS_MODULE,
};

static int __net_init ipmr_rules_init(struct net *net)
{
	struct fib_rules_ops *ops;
	struct mr_table *mrt;
	int err;

	ops = fib_rules_register(&ipmr_rules_ops_template, net);
	if (IS_ERR(ops))
		return PTR_ERR(ops);

	INIT_LIST_HEAD(&net->ipv4.mr_tables);

	mrt = ipmr_new_table(net, RT_TABLE_DEFAULT);
	if (mrt == NULL) {
		err = -ENOMEM;
		goto err1;
	}

	err = fib_default_rule_add(ops, 0x7fff, RT_TABLE_DEFAULT, 0);
	if (err < 0)
		goto err2;

	net->ipv4.mr_rules_ops = ops;
	return 0;

err2:
271
	ipmr_free_table(mrt);
272 273 274 275 276 277 278 279 280
err1:
	fib_rules_unregister(ops);
	return err;
}

static void __net_exit ipmr_rules_exit(struct net *net)
{
	struct mr_table *mrt, *next;

E
Eric Dumazet 已提交
281 282
	list_for_each_entry_safe(mrt, next, &net->ipv4.mr_tables, list) {
		list_del(&mrt->list);
283
		ipmr_free_table(mrt);
E
Eric Dumazet 已提交
284
	}
285 286 287 288 289 290 291 292 293 294 295
	fib_rules_unregister(net->ipv4.mr_rules_ops);
}
#else
#define ipmr_for_each_table(mrt, net) \
	for (mrt = net->ipv4.mrt; mrt; mrt = NULL)

static struct mr_table *ipmr_get_table(struct net *net, u32 id)
{
	return net->ipv4.mrt;
}

D
David S. Miller 已提交
296
static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4,
297 298 299 300 301 302 303 304 305 306 307 308 309 310
			   struct mr_table **mrt)
{
	*mrt = net->ipv4.mrt;
	return 0;
}

static int __net_init ipmr_rules_init(struct net *net)
{
	net->ipv4.mrt = ipmr_new_table(net, RT_TABLE_DEFAULT);
	return net->ipv4.mrt ? 0 : -ENOMEM;
}

static void __net_exit ipmr_rules_exit(struct net *net)
{
311
	ipmr_free_table(net->ipv4.mrt);
312 313 314 315 316 317 318
}
#endif

static struct mr_table *ipmr_new_table(struct net *net, u32 id)
{
	struct mr_table *mrt;
	unsigned int i;
L
Linus Torvalds 已提交
319

320 321 322 323 324 325 326
	mrt = ipmr_get_table(net, id);
	if (mrt != NULL)
		return mrt;

	mrt = kzalloc(sizeof(*mrt), GFP_KERNEL);
	if (mrt == NULL)
		return NULL;
327
	write_pnet(&mrt->net, net);
328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346
	mrt->id = id;

	/* Forwarding cache */
	for (i = 0; i < MFC_LINES; i++)
		INIT_LIST_HEAD(&mrt->mfc_cache_array[i]);

	INIT_LIST_HEAD(&mrt->mfc_unres_queue);

	setup_timer(&mrt->ipmr_expire_timer, ipmr_expire_process,
		    (unsigned long)mrt);

#ifdef CONFIG_IP_PIMSM
	mrt->mroute_reg_vif_num = -1;
#endif
#ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES
	list_add_tail_rcu(&mrt->list, &net->ipv4.mr_tables);
#endif
	return mrt;
}
L
Linus Torvalds 已提交
347

348 349 350 351 352 353 354
static void ipmr_free_table(struct mr_table *mrt)
{
	del_timer_sync(&mrt->ipmr_expire_timer);
	mroute_clean_tables(mrt);
	kfree(mrt);
}

L
Linus Torvalds 已提交
355 356
/* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */

357 358
static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v)
{
359 360
	struct net *net = dev_net(dev);

361 362
	dev_close(dev);

363
	dev = __dev_get_by_name(net, "tunl0");
364
	if (dev) {
365
		const struct net_device_ops *ops = dev->netdev_ops;
366 367 368 369 370 371 372 373 374 375 376 377
		struct ifreq ifr;
		struct ip_tunnel_parm p;

		memset(&p, 0, sizeof(p));
		p.iph.daddr = v->vifc_rmt_addr.s_addr;
		p.iph.saddr = v->vifc_lcl_addr.s_addr;
		p.iph.version = 4;
		p.iph.ihl = 5;
		p.iph.protocol = IPPROTO_IPIP;
		sprintf(p.name, "dvmrp%d", v->vifc_vifi);
		ifr.ifr_ifru.ifru_data = (__force void __user *)&p;

378 379 380 381 382 383 384
		if (ops->ndo_do_ioctl) {
			mm_segment_t oldfs = get_fs();

			set_fs(KERNEL_DS);
			ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL);
			set_fs(oldfs);
		}
385 386 387
	}
}

L
Linus Torvalds 已提交
388
static
389
struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v)
L
Linus Torvalds 已提交
390 391 392
{
	struct net_device  *dev;

393
	dev = __dev_get_by_name(net, "tunl0");
L
Linus Torvalds 已提交
394 395

	if (dev) {
396
		const struct net_device_ops *ops = dev->netdev_ops;
L
Linus Torvalds 已提交
397 398 399 400 401 402 403 404 405 406 407 408
		int err;
		struct ifreq ifr;
		struct ip_tunnel_parm p;
		struct in_device  *in_dev;

		memset(&p, 0, sizeof(p));
		p.iph.daddr = v->vifc_rmt_addr.s_addr;
		p.iph.saddr = v->vifc_lcl_addr.s_addr;
		p.iph.version = 4;
		p.iph.ihl = 5;
		p.iph.protocol = IPPROTO_IPIP;
		sprintf(p.name, "dvmrp%d", v->vifc_vifi);
S
Stephen Hemminger 已提交
409
		ifr.ifr_ifru.ifru_data = (__force void __user *)&p;
L
Linus Torvalds 已提交
410

411 412 413 414 415 416
		if (ops->ndo_do_ioctl) {
			mm_segment_t oldfs = get_fs();

			set_fs(KERNEL_DS);
			err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL);
			set_fs(oldfs);
E
Eric Dumazet 已提交
417
		} else {
418
			err = -EOPNOTSUPP;
E
Eric Dumazet 已提交
419
		}
L
Linus Torvalds 已提交
420 421
		dev = NULL;

422 423
		if (err == 0 &&
		    (dev = __dev_get_by_name(net, p.name)) != NULL) {
L
Linus Torvalds 已提交
424 425
			dev->flags |= IFF_MULTICAST;

426
			in_dev = __in_dev_get_rtnl(dev);
427
			if (in_dev == NULL)
L
Linus Torvalds 已提交
428
				goto failure;
429 430

			ipv4_devconf_setall(in_dev);
431
			neigh_parms_data_state_setall(in_dev->arp_parms);
432
			IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0;
L
Linus Torvalds 已提交
433 434 435

			if (dev_open(dev))
				goto failure;
436
			dev_hold(dev);
L
Linus Torvalds 已提交
437 438 439 440 441 442 443 444 445 446 447 448 449 450 451
		}
	}
	return dev;

failure:
	/* allow the register to be completed before unregistering. */
	rtnl_unlock();
	rtnl_lock();

	unregister_netdevice(dev);
	return NULL;
}

#ifdef CONFIG_IP_PIMSM

452
static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, struct net_device *dev)
L
Linus Torvalds 已提交
453
{
454
	struct net *net = dev_net(dev);
455
	struct mr_table *mrt;
D
David S. Miller 已提交
456 457
	struct flowi4 fl4 = {
		.flowi4_oif	= dev->ifindex,
458
		.flowi4_iif	= skb->skb_iif ? : LOOPBACK_IFINDEX,
D
David S. Miller 已提交
459
		.flowi4_mark	= skb->mark,
460 461 462
	};
	int err;

D
David S. Miller 已提交
463
	err = ipmr_fib_lookup(net, &fl4, &mrt);
464 465
	if (err < 0) {
		kfree_skb(skb);
466
		return err;
467
	}
468

L
Linus Torvalds 已提交
469
	read_lock(&mrt_lock);
470 471
	dev->stats.tx_bytes += skb->len;
	dev->stats.tx_packets++;
472
	ipmr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, IGMPMSG_WHOLEPKT);
L
Linus Torvalds 已提交
473 474
	read_unlock(&mrt_lock);
	kfree_skb(skb);
475
	return NETDEV_TX_OK;
L
Linus Torvalds 已提交
476 477
}

478 479 480 481
static const struct net_device_ops reg_vif_netdev_ops = {
	.ndo_start_xmit	= reg_vif_xmit,
};

L
Linus Torvalds 已提交
482 483 484
static void reg_vif_setup(struct net_device *dev)
{
	dev->type		= ARPHRD_PIMREG;
485
	dev->mtu		= ETH_DATA_LEN - sizeof(struct iphdr) - 8;
L
Linus Torvalds 已提交
486
	dev->flags		= IFF_NOARP;
487
	dev->netdev_ops		= &reg_vif_netdev_ops;
L
Linus Torvalds 已提交
488
	dev->destructor		= free_netdev;
T
Tom Goff 已提交
489
	dev->features		|= NETIF_F_NETNS_LOCAL;
L
Linus Torvalds 已提交
490 491
}

492
static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt)
L
Linus Torvalds 已提交
493 494 495
{
	struct net_device *dev;
	struct in_device *in_dev;
496
	char name[IFNAMSIZ];
L
Linus Torvalds 已提交
497

498 499 500 501
	if (mrt->id == RT_TABLE_DEFAULT)
		sprintf(name, "pimreg");
	else
		sprintf(name, "pimreg%u", mrt->id);
L
Linus Torvalds 已提交
502

503
	dev = alloc_netdev(0, name, NET_NAME_UNKNOWN, reg_vif_setup);
L
Linus Torvalds 已提交
504 505 506 507

	if (dev == NULL)
		return NULL;

T
Tom Goff 已提交
508 509
	dev_net_set(dev, net);

L
Linus Torvalds 已提交
510 511 512 513 514 515
	if (register_netdevice(dev)) {
		free_netdev(dev);
		return NULL;
	}
	dev->iflink = 0;

516
	rcu_read_lock();
E
Eric Dumazet 已提交
517 518
	in_dev = __in_dev_get_rcu(dev);
	if (!in_dev) {
519
		rcu_read_unlock();
L
Linus Torvalds 已提交
520
		goto failure;
521
	}
L
Linus Torvalds 已提交
522

523
	ipv4_devconf_setall(in_dev);
524
	neigh_parms_data_state_setall(in_dev->arp_parms);
525 526
	IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0;
	rcu_read_unlock();
L
Linus Torvalds 已提交
527 528 529 530

	if (dev_open(dev))
		goto failure;

531 532
	dev_hold(dev);

L
Linus Torvalds 已提交
533 534 535 536 537 538 539 540 541 542 543 544
	return dev;

failure:
	/* allow the register to be completed before unregistering. */
	rtnl_unlock();
	rtnl_lock();

	unregister_netdevice(dev);
	return NULL;
}
#endif

545 546
/**
 *	vif_delete - Delete a VIF entry
547
 *	@notify: Set to 1, if the caller is a notifier_call
L
Linus Torvalds 已提交
548
 */
549

550
static int vif_delete(struct mr_table *mrt, int vifi, int notify,
551
		      struct list_head *head)
L
Linus Torvalds 已提交
552 553 554 555 556
{
	struct vif_device *v;
	struct net_device *dev;
	struct in_device *in_dev;

557
	if (vifi < 0 || vifi >= mrt->maxvif)
L
Linus Torvalds 已提交
558 559
		return -EADDRNOTAVAIL;

560
	v = &mrt->vif_table[vifi];
L
Linus Torvalds 已提交
561 562 563 564 565 566 567 568 569 570 571

	write_lock_bh(&mrt_lock);
	dev = v->dev;
	v->dev = NULL;

	if (!dev) {
		write_unlock_bh(&mrt_lock);
		return -EADDRNOTAVAIL;
	}

#ifdef CONFIG_IP_PIMSM
572 573
	if (vifi == mrt->mroute_reg_vif_num)
		mrt->mroute_reg_vif_num = -1;
L
Linus Torvalds 已提交
574 575
#endif

E
Eric Dumazet 已提交
576
	if (vifi + 1 == mrt->maxvif) {
L
Linus Torvalds 已提交
577
		int tmp;
E
Eric Dumazet 已提交
578 579

		for (tmp = vifi - 1; tmp >= 0; tmp--) {
580
			if (VIF_EXISTS(mrt, tmp))
L
Linus Torvalds 已提交
581 582
				break;
		}
583
		mrt->maxvif = tmp+1;
L
Linus Torvalds 已提交
584 585 586 587 588 589
	}

	write_unlock_bh(&mrt_lock);

	dev_set_allmulti(dev, -1);

E
Eric Dumazet 已提交
590 591
	in_dev = __in_dev_get_rtnl(dev);
	if (in_dev) {
592
		IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--;
593 594 595
		inet_netconf_notify_devconf(dev_net(dev),
					    NETCONFA_MC_FORWARDING,
					    dev->ifindex, &in_dev->cnf);
L
Linus Torvalds 已提交
596 597 598
		ip_rt_multicast_event(in_dev);
	}

E
Eric Dumazet 已提交
599
	if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER) && !notify)
600
		unregister_netdevice_queue(dev, head);
L
Linus Torvalds 已提交
601 602 603 604 605

	dev_put(dev);
	return 0;
}

606
static void ipmr_cache_free_rcu(struct rcu_head *head)
607
{
608 609
	struct mfc_cache *c = container_of(head, struct mfc_cache, rcu);

610 611 612
	kmem_cache_free(mrt_cachep, c);
}

613 614 615 616 617
static inline void ipmr_cache_free(struct mfc_cache *c)
{
	call_rcu(&c->rcu, ipmr_cache_free_rcu);
}

L
Linus Torvalds 已提交
618
/* Destroy an unresolved cache entry, killing queued skbs
E
Eric Dumazet 已提交
619
 * and reporting error to netlink readers.
L
Linus Torvalds 已提交
620 621
 */

622
static void ipmr_destroy_unres(struct mr_table *mrt, struct mfc_cache *c)
L
Linus Torvalds 已提交
623
{
624
	struct net *net = read_pnet(&mrt->net);
L
Linus Torvalds 已提交
625
	struct sk_buff *skb;
626
	struct nlmsgerr *e;
L
Linus Torvalds 已提交
627

628
	atomic_dec(&mrt->cache_resolve_queue_len);
L
Linus Torvalds 已提交
629

J
Jianjun Kong 已提交
630
	while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) {
631
		if (ip_hdr(skb)->version == 0) {
L
Linus Torvalds 已提交
632 633
			struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr));
			nlh->nlmsg_type = NLMSG_ERROR;
634
			nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
L
Linus Torvalds 已提交
635
			skb_trim(skb, nlh->nlmsg_len);
636
			e = nlmsg_data(nlh);
637 638
			e->error = -ETIMEDOUT;
			memset(&e->msg, 0, sizeof(e->msg));
639

640
			rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
E
Eric Dumazet 已提交
641
		} else {
L
Linus Torvalds 已提交
642
			kfree_skb(skb);
E
Eric Dumazet 已提交
643
		}
L
Linus Torvalds 已提交
644 645
	}

646
	ipmr_cache_free(c);
L
Linus Torvalds 已提交
647 648 649
}


650
/* Timer process for the unresolved queue. */
L
Linus Torvalds 已提交
651

652
static void ipmr_expire_process(unsigned long arg)
L
Linus Torvalds 已提交
653
{
654
	struct mr_table *mrt = (struct mr_table *)arg;
L
Linus Torvalds 已提交
655 656
	unsigned long now;
	unsigned long expires;
657
	struct mfc_cache *c, *next;
L
Linus Torvalds 已提交
658 659

	if (!spin_trylock(&mfc_unres_lock)) {
660
		mod_timer(&mrt->ipmr_expire_timer, jiffies+HZ/10);
L
Linus Torvalds 已提交
661 662 663
		return;
	}

664
	if (list_empty(&mrt->mfc_unres_queue))
L
Linus Torvalds 已提交
665 666 667 668 669
		goto out;

	now = jiffies;
	expires = 10*HZ;

670
	list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) {
L
Linus Torvalds 已提交
671 672 673 674 675 676 677
		if (time_after(c->mfc_un.unres.expires, now)) {
			unsigned long interval = c->mfc_un.unres.expires - now;
			if (interval < expires)
				expires = interval;
			continue;
		}

678
		list_del(&c->list);
679
		mroute_netlink_event(mrt, c, RTM_DELROUTE);
680
		ipmr_destroy_unres(mrt, c);
L
Linus Torvalds 已提交
681 682
	}

683 684
	if (!list_empty(&mrt->mfc_unres_queue))
		mod_timer(&mrt->ipmr_expire_timer, jiffies + expires);
L
Linus Torvalds 已提交
685 686 687 688 689 690 691

out:
	spin_unlock(&mfc_unres_lock);
}

/* Fill oifs list. It is called under write locked mrt_lock. */

692
static void ipmr_update_thresholds(struct mr_table *mrt, struct mfc_cache *cache,
693
				   unsigned char *ttls)
L
Linus Torvalds 已提交
694 695 696 697 698 699 700
{
	int vifi;

	cache->mfc_un.res.minvif = MAXVIFS;
	cache->mfc_un.res.maxvif = 0;
	memset(cache->mfc_un.res.ttls, 255, MAXVIFS);

701 702
	for (vifi = 0; vifi < mrt->maxvif; vifi++) {
		if (VIF_EXISTS(mrt, vifi) &&
703
		    ttls[vifi] && ttls[vifi] < 255) {
L
Linus Torvalds 已提交
704 705 706 707 708 709 710 711 712
			cache->mfc_un.res.ttls[vifi] = ttls[vifi];
			if (cache->mfc_un.res.minvif > vifi)
				cache->mfc_un.res.minvif = vifi;
			if (cache->mfc_un.res.maxvif <= vifi)
				cache->mfc_un.res.maxvif = vifi + 1;
		}
	}
}

713 714
static int vif_add(struct net *net, struct mr_table *mrt,
		   struct vifctl *vifc, int mrtsock)
L
Linus Torvalds 已提交
715 716
{
	int vifi = vifc->vifc_vifi;
717
	struct vif_device *v = &mrt->vif_table[vifi];
L
Linus Torvalds 已提交
718 719
	struct net_device *dev;
	struct in_device *in_dev;
720
	int err;
L
Linus Torvalds 已提交
721 722

	/* Is vif busy ? */
723
	if (VIF_EXISTS(mrt, vifi))
L
Linus Torvalds 已提交
724 725 726 727 728 729 730 731 732
		return -EADDRINUSE;

	switch (vifc->vifc_flags) {
#ifdef CONFIG_IP_PIMSM
	case VIFF_REGISTER:
		/*
		 * Special Purpose VIF in PIM
		 * All the packets will be sent to the daemon
		 */
733
		if (mrt->mroute_reg_vif_num >= 0)
L
Linus Torvalds 已提交
734
			return -EADDRINUSE;
735
		dev = ipmr_reg_vif(net, mrt);
L
Linus Torvalds 已提交
736 737
		if (!dev)
			return -ENOBUFS;
738 739 740
		err = dev_set_allmulti(dev, 1);
		if (err) {
			unregister_netdevice(dev);
741
			dev_put(dev);
742 743
			return err;
		}
L
Linus Torvalds 已提交
744 745
		break;
#endif
746
	case VIFF_TUNNEL:
747
		dev = ipmr_new_tunnel(net, vifc);
L
Linus Torvalds 已提交
748 749
		if (!dev)
			return -ENOBUFS;
750 751 752
		err = dev_set_allmulti(dev, 1);
		if (err) {
			ipmr_del_tunnel(dev, vifc);
753
			dev_put(dev);
754 755
			return err;
		}
L
Linus Torvalds 已提交
756
		break;
757 758

	case VIFF_USE_IFINDEX:
L
Linus Torvalds 已提交
759
	case 0:
760 761
		if (vifc->vifc_flags == VIFF_USE_IFINDEX) {
			dev = dev_get_by_index(net, vifc->vifc_lcl_ifindex);
E
Eric Dumazet 已提交
762
			if (dev && __in_dev_get_rtnl(dev) == NULL) {
763 764 765
				dev_put(dev);
				return -EADDRNOTAVAIL;
			}
E
Eric Dumazet 已提交
766
		} else {
767
			dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr);
E
Eric Dumazet 已提交
768
		}
L
Linus Torvalds 已提交
769 770
		if (!dev)
			return -EADDRNOTAVAIL;
771
		err = dev_set_allmulti(dev, 1);
772 773
		if (err) {
			dev_put(dev);
774
			return err;
775
		}
L
Linus Torvalds 已提交
776 777 778 779 780
		break;
	default:
		return -EINVAL;
	}

E
Eric Dumazet 已提交
781 782
	in_dev = __in_dev_get_rtnl(dev);
	if (!in_dev) {
783
		dev_put(dev);
L
Linus Torvalds 已提交
784
		return -EADDRNOTAVAIL;
785
	}
786
	IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++;
787 788
	inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, dev->ifindex,
				    &in_dev->cnf);
L
Linus Torvalds 已提交
789 790
	ip_rt_multicast_event(in_dev);

E
Eric Dumazet 已提交
791 792
	/* Fill in the VIF structures */

J
Jianjun Kong 已提交
793 794 795 796
	v->rate_limit = vifc->vifc_rate_limit;
	v->local = vifc->vifc_lcl_addr.s_addr;
	v->remote = vifc->vifc_rmt_addr.s_addr;
	v->flags = vifc->vifc_flags;
L
Linus Torvalds 已提交
797 798
	if (!mrtsock)
		v->flags |= VIFF_STATIC;
J
Jianjun Kong 已提交
799
	v->threshold = vifc->vifc_threshold;
L
Linus Torvalds 已提交
800 801 802 803 804
	v->bytes_in = 0;
	v->bytes_out = 0;
	v->pkt_in = 0;
	v->pkt_out = 0;
	v->link = dev->ifindex;
E
Eric Dumazet 已提交
805
	if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER))
L
Linus Torvalds 已提交
806 807 808 809
		v->link = dev->iflink;

	/* And finish update writing critical data */
	write_lock_bh(&mrt_lock);
J
Jianjun Kong 已提交
810
	v->dev = dev;
L
Linus Torvalds 已提交
811
#ifdef CONFIG_IP_PIMSM
E
Eric Dumazet 已提交
812
	if (v->flags & VIFF_REGISTER)
813
		mrt->mroute_reg_vif_num = vifi;
L
Linus Torvalds 已提交
814
#endif
815 816
	if (vifi+1 > mrt->maxvif)
		mrt->maxvif = vifi+1;
L
Linus Torvalds 已提交
817 818 819 820
	write_unlock_bh(&mrt_lock);
	return 0;
}

821
/* called with rcu_read_lock() */
822
static struct mfc_cache *ipmr_cache_find(struct mr_table *mrt,
823 824
					 __be32 origin,
					 __be32 mcastgrp)
L
Linus Torvalds 已提交
825
{
J
Jianjun Kong 已提交
826
	int line = MFC_HASH(mcastgrp, origin);
L
Linus Torvalds 已提交
827 828
	struct mfc_cache *c;

829
	list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list) {
830 831
		if (c->mfc_origin == origin && c->mfc_mcastgrp == mcastgrp)
			return c;
L
Linus Torvalds 已提交
832
	}
833
	return NULL;
L
Linus Torvalds 已提交
834 835
}

836 837 838 839
/* Look for a (*,*,oif) entry */
static struct mfc_cache *ipmr_cache_find_any_parent(struct mr_table *mrt,
						    int vifi)
{
840
	int line = MFC_HASH(htonl(INADDR_ANY), htonl(INADDR_ANY));
841 842 843
	struct mfc_cache *c;

	list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list)
844 845
		if (c->mfc_origin == htonl(INADDR_ANY) &&
		    c->mfc_mcastgrp == htonl(INADDR_ANY) &&
846 847 848 849 850 851 852 853 854 855
		    c->mfc_un.res.ttls[vifi] < 255)
			return c;

	return NULL;
}

/* Look for a (*,G) entry */
static struct mfc_cache *ipmr_cache_find_any(struct mr_table *mrt,
					     __be32 mcastgrp, int vifi)
{
856
	int line = MFC_HASH(mcastgrp, htonl(INADDR_ANY));
857 858
	struct mfc_cache *c, *proxy;

859
	if (mcastgrp == htonl(INADDR_ANY))
860 861 862
		goto skip;

	list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list)
863
		if (c->mfc_origin == htonl(INADDR_ANY) &&
864 865 866 867 868 869 870 871 872 873 874 875 876 877 878
		    c->mfc_mcastgrp == mcastgrp) {
			if (c->mfc_un.res.ttls[vifi] < 255)
				return c;

			/* It's ok if the vifi is part of the static tree */
			proxy = ipmr_cache_find_any_parent(mrt,
							   c->mfc_parent);
			if (proxy && proxy->mfc_un.res.ttls[vifi] < 255)
				return c;
		}

skip:
	return ipmr_cache_find_any_parent(mrt, vifi);
}

L
Linus Torvalds 已提交
879 880 881
/*
 *	Allocate a multicast cache entry
 */
882
static struct mfc_cache *ipmr_cache_alloc(void)
L
Linus Torvalds 已提交
883
{
J
Jianjun Kong 已提交
884
	struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL);
885 886 887

	if (c)
		c->mfc_un.res.minvif = MAXVIFS;
L
Linus Torvalds 已提交
888 889 890
	return c;
}

891
static struct mfc_cache *ipmr_cache_alloc_unres(void)
L
Linus Torvalds 已提交
892
{
J
Jianjun Kong 已提交
893
	struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC);
894 895 896 897 898

	if (c) {
		skb_queue_head_init(&c->mfc_un.unres.unresolved);
		c->mfc_un.unres.expires = jiffies + 10*HZ;
	}
L
Linus Torvalds 已提交
899 900 901 902 903 904
	return c;
}

/*
 *	A cache entry has gone into a resolved state from queued
 */
905

906 907
static void ipmr_cache_resolve(struct net *net, struct mr_table *mrt,
			       struct mfc_cache *uc, struct mfc_cache *c)
L
Linus Torvalds 已提交
908 909
{
	struct sk_buff *skb;
910
	struct nlmsgerr *e;
L
Linus Torvalds 已提交
911

E
Eric Dumazet 已提交
912
	/* Play the pending entries through our router */
L
Linus Torvalds 已提交
913

J
Jianjun Kong 已提交
914
	while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) {
915
		if (ip_hdr(skb)->version == 0) {
L
Linus Torvalds 已提交
916 917
			struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr));

918
			if (__ipmr_fill_mroute(mrt, skb, c, nlmsg_data(nlh)) > 0) {
E
Eric Dumazet 已提交
919 920
				nlh->nlmsg_len = skb_tail_pointer(skb) -
						 (u8 *)nlh;
L
Linus Torvalds 已提交
921 922
			} else {
				nlh->nlmsg_type = NLMSG_ERROR;
923
				nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
L
Linus Torvalds 已提交
924
				skb_trim(skb, nlh->nlmsg_len);
925
				e = nlmsg_data(nlh);
926 927
				e->error = -EMSGSIZE;
				memset(&e->msg, 0, sizeof(e->msg));
L
Linus Torvalds 已提交
928
			}
929

930
			rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
E
Eric Dumazet 已提交
931
		} else {
932
			ip_mr_forward(net, mrt, skb, c, 0);
E
Eric Dumazet 已提交
933
		}
L
Linus Torvalds 已提交
934 935 936 937 938 939 940 941 942
	}
}

/*
 *	Bounce a cache query up to mrouted. We could use netlink for this but mrouted
 *	expects the following bizarre scheme.
 *
 *	Called under mrt_lock.
 */
943

944
static int ipmr_cache_report(struct mr_table *mrt,
945
			     struct sk_buff *pkt, vifi_t vifi, int assert)
L
Linus Torvalds 已提交
946 947
{
	struct sk_buff *skb;
948
	const int ihl = ip_hdrlen(pkt);
L
Linus Torvalds 已提交
949 950
	struct igmphdr *igmp;
	struct igmpmsg *msg;
E
Eric Dumazet 已提交
951
	struct sock *mroute_sk;
L
Linus Torvalds 已提交
952 953 954 955 956 957 958 959 960
	int ret;

#ifdef CONFIG_IP_PIMSM
	if (assert == IGMPMSG_WHOLEPKT)
		skb = skb_realloc_headroom(pkt, sizeof(struct iphdr));
	else
#endif
		skb = alloc_skb(128, GFP_ATOMIC);

S
Stephen Hemminger 已提交
961
	if (!skb)
L
Linus Torvalds 已提交
962 963 964 965 966
		return -ENOBUFS;

#ifdef CONFIG_IP_PIMSM
	if (assert == IGMPMSG_WHOLEPKT) {
		/* Ugly, but we have no choice with this interface.
E
Eric Dumazet 已提交
967 968 969
		 * Duplicate old header, fix ihl, length etc.
		 * And all this only to mangle msg->im_msgtype and
		 * to set msg->im_mbz to "mbz" :-)
L
Linus Torvalds 已提交
970
		 */
971 972
		skb_push(skb, sizeof(struct iphdr));
		skb_reset_network_header(skb);
973
		skb_reset_transport_header(skb);
974
		msg = (struct igmpmsg *)skb_network_header(skb);
975
		memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr));
L
Linus Torvalds 已提交
976 977
		msg->im_msgtype = IGMPMSG_WHOLEPKT;
		msg->im_mbz = 0;
978
		msg->im_vif = mrt->mroute_reg_vif_num;
979 980 981
		ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2;
		ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) +
					     sizeof(struct iphdr));
982
	} else
L
Linus Torvalds 已提交
983
#endif
984 985
	{

E
Eric Dumazet 已提交
986
	/* Copy the IP header */
L
Linus Torvalds 已提交
987

988
	skb_set_network_header(skb, skb->len);
989
	skb_put(skb, ihl);
990
	skb_copy_to_linear_data(skb, pkt->data, ihl);
E
Eric Dumazet 已提交
991
	ip_hdr(skb)->protocol = 0;	/* Flag to the kernel this is a route add */
992
	msg = (struct igmpmsg *)skb_network_header(skb);
L
Linus Torvalds 已提交
993
	msg->im_vif = vifi;
E
Eric Dumazet 已提交
994
	skb_dst_set(skb, dst_clone(skb_dst(pkt)));
L
Linus Torvalds 已提交
995

E
Eric Dumazet 已提交
996
	/* Add our header */
L
Linus Torvalds 已提交
997

E
Eric Dumazet 已提交
998
	igmp = (struct igmphdr *)skb_put(skb, sizeof(struct igmphdr));
L
Linus Torvalds 已提交
999 1000
	igmp->type	=
	msg->im_msgtype = assert;
E
Eric Dumazet 已提交
1001 1002
	igmp->code	= 0;
	ip_hdr(skb)->tot_len = htons(skb->len);		/* Fix the length */
1003
	skb->transport_header = skb->network_header;
1004
	}
L
Linus Torvalds 已提交
1005

E
Eric Dumazet 已提交
1006 1007 1008 1009
	rcu_read_lock();
	mroute_sk = rcu_dereference(mrt->mroute_sk);
	if (mroute_sk == NULL) {
		rcu_read_unlock();
L
Linus Torvalds 已提交
1010 1011 1012 1013
		kfree_skb(skb);
		return -EINVAL;
	}

E
Eric Dumazet 已提交
1014 1015
	/* Deliver to mrouted */

E
Eric Dumazet 已提交
1016 1017
	ret = sock_queue_rcv_skb(mroute_sk, skb);
	rcu_read_unlock();
1018
	if (ret < 0) {
1019
		net_warn_ratelimited("mroute: pending queue full, dropping entries\n");
L
Linus Torvalds 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028
		kfree_skb(skb);
	}

	return ret;
}

/*
 *	Queue a packet for resolution. It gets locked cache entry!
 */
1029

L
Linus Torvalds 已提交
1030
static int
1031
ipmr_cache_unresolved(struct mr_table *mrt, vifi_t vifi, struct sk_buff *skb)
L
Linus Torvalds 已提交
1032
{
1033
	bool found = false;
L
Linus Torvalds 已提交
1034 1035
	int err;
	struct mfc_cache *c;
1036
	const struct iphdr *iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1037 1038

	spin_lock_bh(&mfc_unres_lock);
1039
	list_for_each_entry(c, &mrt->mfc_unres_queue, list) {
1040
		if (c->mfc_mcastgrp == iph->daddr &&
1041 1042
		    c->mfc_origin == iph->saddr) {
			found = true;
L
Linus Torvalds 已提交
1043
			break;
1044
		}
L
Linus Torvalds 已提交
1045 1046
	}

1047
	if (!found) {
E
Eric Dumazet 已提交
1048
		/* Create a new entry if allowable */
L
Linus Torvalds 已提交
1049

1050
		if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 ||
1051
		    (c = ipmr_cache_alloc_unres()) == NULL) {
L
Linus Torvalds 已提交
1052 1053 1054 1055 1056 1057
			spin_unlock_bh(&mfc_unres_lock);

			kfree_skb(skb);
			return -ENOBUFS;
		}

E
Eric Dumazet 已提交
1058 1059
		/* Fill in the new cache entry */

1060 1061 1062
		c->mfc_parent	= -1;
		c->mfc_origin	= iph->saddr;
		c->mfc_mcastgrp	= iph->daddr;
L
Linus Torvalds 已提交
1063

E
Eric Dumazet 已提交
1064 1065
		/* Reflect first query at mrouted. */

1066
		err = ipmr_cache_report(mrt, skb, vifi, IGMPMSG_NOCACHE);
1067
		if (err < 0) {
1068
			/* If the report failed throw the cache entry
L
Linus Torvalds 已提交
1069 1070 1071 1072
			   out - Brad Parker
			 */
			spin_unlock_bh(&mfc_unres_lock);

1073
			ipmr_cache_free(c);
L
Linus Torvalds 已提交
1074 1075 1076 1077
			kfree_skb(skb);
			return err;
		}

1078 1079
		atomic_inc(&mrt->cache_resolve_queue_len);
		list_add(&c->list, &mrt->mfc_unres_queue);
1080
		mroute_netlink_event(mrt, c, RTM_NEWROUTE);
L
Linus Torvalds 已提交
1081

1082 1083
		if (atomic_read(&mrt->cache_resolve_queue_len) == 1)
			mod_timer(&mrt->ipmr_expire_timer, c->mfc_un.unres.expires);
L
Linus Torvalds 已提交
1084 1085
	}

E
Eric Dumazet 已提交
1086 1087 1088
	/* See if we can append the packet */

	if (c->mfc_un.unres.unresolved.qlen > 3) {
L
Linus Torvalds 已提交
1089 1090 1091
		kfree_skb(skb);
		err = -ENOBUFS;
	} else {
J
Jianjun Kong 已提交
1092
		skb_queue_tail(&c->mfc_un.unres.unresolved, skb);
L
Linus Torvalds 已提交
1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
		err = 0;
	}

	spin_unlock_bh(&mfc_unres_lock);
	return err;
}

/*
 *	MFC cache manipulation by user space mroute daemon
 */

1104
static int ipmr_mfc_delete(struct mr_table *mrt, struct mfcctl *mfc, int parent)
L
Linus Torvalds 已提交
1105 1106
{
	int line;
1107
	struct mfc_cache *c, *next;
L
Linus Torvalds 已提交
1108

J
Jianjun Kong 已提交
1109
	line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr);
L
Linus Torvalds 已提交
1110

1111
	list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[line], list) {
L
Linus Torvalds 已提交
1112
		if (c->mfc_origin == mfc->mfcc_origin.s_addr &&
1113 1114
		    c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr &&
		    (parent == -1 || parent == c->mfc_parent)) {
1115
			list_del_rcu(&c->list);
1116
			mroute_netlink_event(mrt, c, RTM_DELROUTE);
1117
			ipmr_cache_free(c);
L
Linus Torvalds 已提交
1118 1119 1120 1121 1122 1123
			return 0;
		}
	}
	return -ENOENT;
}

1124
static int ipmr_mfc_add(struct net *net, struct mr_table *mrt,
1125
			struct mfcctl *mfc, int mrtsock, int parent)
L
Linus Torvalds 已提交
1126
{
1127
	bool found = false;
L
Linus Torvalds 已提交
1128
	int line;
1129
	struct mfc_cache *uc, *c;
L
Linus Torvalds 已提交
1130

1131 1132 1133
	if (mfc->mfcc_parent >= MAXVIFS)
		return -ENFILE;

J
Jianjun Kong 已提交
1134
	line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr);
L
Linus Torvalds 已提交
1135

1136
	list_for_each_entry(c, &mrt->mfc_cache_array[line], list) {
L
Linus Torvalds 已提交
1137
		if (c->mfc_origin == mfc->mfcc_origin.s_addr &&
1138 1139
		    c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr &&
		    (parent == -1 || parent == c->mfc_parent)) {
1140
			found = true;
L
Linus Torvalds 已提交
1141
			break;
1142
		}
L
Linus Torvalds 已提交
1143 1144
	}

1145
	if (found) {
L
Linus Torvalds 已提交
1146 1147
		write_lock_bh(&mrt_lock);
		c->mfc_parent = mfc->mfcc_parent;
1148
		ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls);
L
Linus Torvalds 已提交
1149 1150 1151
		if (!mrtsock)
			c->mfc_flags |= MFC_STATIC;
		write_unlock_bh(&mrt_lock);
1152
		mroute_netlink_event(mrt, c, RTM_NEWROUTE);
L
Linus Torvalds 已提交
1153 1154 1155
		return 0;
	}

1156
	if (mfc->mfcc_mcastgrp.s_addr != htonl(INADDR_ANY) &&
1157
	    !ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr))
L
Linus Torvalds 已提交
1158 1159
		return -EINVAL;

1160
	c = ipmr_cache_alloc();
J
Jianjun Kong 已提交
1161
	if (c == NULL)
L
Linus Torvalds 已提交
1162 1163
		return -ENOMEM;

J
Jianjun Kong 已提交
1164 1165 1166
	c->mfc_origin = mfc->mfcc_origin.s_addr;
	c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr;
	c->mfc_parent = mfc->mfcc_parent;
1167
	ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls);
L
Linus Torvalds 已提交
1168 1169 1170
	if (!mrtsock)
		c->mfc_flags |= MFC_STATIC;

1171
	list_add_rcu(&c->list, &mrt->mfc_cache_array[line]);
L
Linus Torvalds 已提交
1172 1173 1174 1175 1176

	/*
	 *	Check to see if we resolved a queued list. If so we
	 *	need to send on the frames and tidy up.
	 */
1177
	found = false;
L
Linus Torvalds 已提交
1178
	spin_lock_bh(&mfc_unres_lock);
1179
	list_for_each_entry(uc, &mrt->mfc_unres_queue, list) {
1180
		if (uc->mfc_origin == c->mfc_origin &&
L
Linus Torvalds 已提交
1181
		    uc->mfc_mcastgrp == c->mfc_mcastgrp) {
1182
			list_del(&uc->list);
1183
			atomic_dec(&mrt->cache_resolve_queue_len);
1184
			found = true;
L
Linus Torvalds 已提交
1185 1186 1187
			break;
		}
	}
1188 1189
	if (list_empty(&mrt->mfc_unres_queue))
		del_timer(&mrt->ipmr_expire_timer);
L
Linus Torvalds 已提交
1190 1191
	spin_unlock_bh(&mfc_unres_lock);

1192
	if (found) {
1193
		ipmr_cache_resolve(net, mrt, uc, c);
1194
		ipmr_cache_free(uc);
L
Linus Torvalds 已提交
1195
	}
1196
	mroute_netlink_event(mrt, c, RTM_NEWROUTE);
L
Linus Torvalds 已提交
1197 1198 1199 1200 1201 1202
	return 0;
}

/*
 *	Close the multicast socket, and clear the vif tables etc
 */
1203

1204
static void mroute_clean_tables(struct mr_table *mrt)
L
Linus Torvalds 已提交
1205 1206
{
	int i;
1207
	LIST_HEAD(list);
1208
	struct mfc_cache *c, *next;
1209

E
Eric Dumazet 已提交
1210 1211
	/* Shut down all active vif entries */

1212
	for (i = 0; i < mrt->maxvif; i++) {
E
Eric Dumazet 已提交
1213
		if (!(mrt->vif_table[i].flags & VIFF_STATIC))
1214
			vif_delete(mrt, i, 0, &list);
L
Linus Torvalds 已提交
1215
	}
1216
	unregister_netdevice_many(&list);
L
Linus Torvalds 已提交
1217

E
Eric Dumazet 已提交
1218 1219
	/* Wipe the cache */

1220
	for (i = 0; i < MFC_LINES; i++) {
1221
		list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[i], list) {
1222
			if (c->mfc_flags & MFC_STATIC)
L
Linus Torvalds 已提交
1223
				continue;
1224
			list_del_rcu(&c->list);
1225
			mroute_netlink_event(mrt, c, RTM_DELROUTE);
1226
			ipmr_cache_free(c);
L
Linus Torvalds 已提交
1227 1228 1229
		}
	}

1230
	if (atomic_read(&mrt->cache_resolve_queue_len) != 0) {
L
Linus Torvalds 已提交
1231
		spin_lock_bh(&mfc_unres_lock);
1232
		list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) {
1233
			list_del(&c->list);
1234
			mroute_netlink_event(mrt, c, RTM_DELROUTE);
1235
			ipmr_destroy_unres(mrt, c);
L
Linus Torvalds 已提交
1236 1237 1238 1239 1240
		}
		spin_unlock_bh(&mfc_unres_lock);
	}
}

E
Eric Dumazet 已提交
1241 1242 1243
/* called from ip_ra_control(), before an RCU grace period,
 * we dont need to call synchronize_rcu() here
 */
L
Linus Torvalds 已提交
1244 1245
static void mrtsock_destruct(struct sock *sk)
{
1246
	struct net *net = sock_net(sk);
1247
	struct mr_table *mrt;
1248

L
Linus Torvalds 已提交
1249
	rtnl_lock();
1250
	ipmr_for_each_table(mrt, net) {
E
Eric Dumazet 已提交
1251
		if (sk == rtnl_dereference(mrt->mroute_sk)) {
1252
			IPV4_DEVCONF_ALL(net, MC_FORWARDING)--;
1253 1254 1255
			inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING,
						    NETCONFA_IFINDEX_ALL,
						    net->ipv4.devconf_all);
1256
			RCU_INIT_POINTER(mrt->mroute_sk, NULL);
1257 1258
			mroute_clean_tables(mrt);
		}
L
Linus Torvalds 已提交
1259 1260 1261 1262 1263 1264 1265 1266 1267 1268
	}
	rtnl_unlock();
}

/*
 *	Socket options and virtual interface manipulation. The whole
 *	virtual interface system is a complete heap, but unfortunately
 *	that's how BSD mrouted happens to think. Maybe one day with a proper
 *	MOSPF/PIM router set up we can clean this up.
 */
1269

1270
int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
1271
{
1272
	int ret, parent = 0;
L
Linus Torvalds 已提交
1273 1274
	struct vifctl vif;
	struct mfcctl mfc;
1275
	struct net *net = sock_net(sk);
1276 1277
	struct mr_table *mrt;

1278 1279 1280 1281
	if (sk->sk_type != SOCK_RAW ||
	    inet_sk(sk)->inet_num != IPPROTO_IGMP)
		return -EOPNOTSUPP;

1282 1283 1284
	mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
	if (mrt == NULL)
		return -ENOENT;
1285

S
Stephen Hemminger 已提交
1286
	if (optname != MRT_INIT) {
1287
		if (sk != rcu_access_pointer(mrt->mroute_sk) &&
1288
		    !ns_capable(net->user_ns, CAP_NET_ADMIN))
L
Linus Torvalds 已提交
1289 1290 1291
			return -EACCES;
	}

S
Stephen Hemminger 已提交
1292 1293
	switch (optname) {
	case MRT_INIT:
J
Jianjun Kong 已提交
1294
		if (optlen != sizeof(int))
1295
			return -EINVAL;
L
Linus Torvalds 已提交
1296

S
Stephen Hemminger 已提交
1297
		rtnl_lock();
E
Eric Dumazet 已提交
1298
		if (rtnl_dereference(mrt->mroute_sk)) {
L
Linus Torvalds 已提交
1299
			rtnl_unlock();
S
Stephen Hemminger 已提交
1300 1301 1302 1303 1304
			return -EADDRINUSE;
		}

		ret = ip_ra_control(sk, 1, mrtsock_destruct);
		if (ret == 0) {
1305
			rcu_assign_pointer(mrt->mroute_sk, sk);
1306
			IPV4_DEVCONF_ALL(net, MC_FORWARDING)++;
1307 1308 1309
			inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING,
						    NETCONFA_IFINDEX_ALL,
						    net->ipv4.devconf_all);
S
Stephen Hemminger 已提交
1310 1311 1312 1313
		}
		rtnl_unlock();
		return ret;
	case MRT_DONE:
1314
		if (sk != rcu_access_pointer(mrt->mroute_sk))
S
Stephen Hemminger 已提交
1315 1316 1317 1318
			return -EACCES;
		return ip_ra_control(sk, 0, NULL);
	case MRT_ADD_VIF:
	case MRT_DEL_VIF:
J
Jianjun Kong 已提交
1319
		if (optlen != sizeof(vif))
S
Stephen Hemminger 已提交
1320
			return -EINVAL;
J
Jianjun Kong 已提交
1321
		if (copy_from_user(&vif, optval, sizeof(vif)))
S
Stephen Hemminger 已提交
1322 1323 1324 1325
			return -EFAULT;
		if (vif.vifc_vifi >= MAXVIFS)
			return -ENFILE;
		rtnl_lock();
J
Jianjun Kong 已提交
1326
		if (optname == MRT_ADD_VIF) {
E
Eric Dumazet 已提交
1327 1328
			ret = vif_add(net, mrt, &vif,
				      sk == rtnl_dereference(mrt->mroute_sk));
S
Stephen Hemminger 已提交
1329
		} else {
1330
			ret = vif_delete(mrt, vif.vifc_vifi, 0, NULL);
S
Stephen Hemminger 已提交
1331 1332 1333
		}
		rtnl_unlock();
		return ret;
L
Linus Torvalds 已提交
1334 1335 1336 1337 1338

		/*
		 *	Manipulate the forwarding caches. These live
		 *	in a sort of kernel/user symbiosis.
		 */
S
Stephen Hemminger 已提交
1339 1340
	case MRT_ADD_MFC:
	case MRT_DEL_MFC:
1341 1342 1343
		parent = -1;
	case MRT_ADD_MFC_PROXY:
	case MRT_DEL_MFC_PROXY:
J
Jianjun Kong 已提交
1344
		if (optlen != sizeof(mfc))
S
Stephen Hemminger 已提交
1345
			return -EINVAL;
J
Jianjun Kong 已提交
1346
		if (copy_from_user(&mfc, optval, sizeof(mfc)))
S
Stephen Hemminger 已提交
1347
			return -EFAULT;
1348 1349
		if (parent == 0)
			parent = mfc.mfcc_parent;
S
Stephen Hemminger 已提交
1350
		rtnl_lock();
1351 1352
		if (optname == MRT_DEL_MFC || optname == MRT_DEL_MFC_PROXY)
			ret = ipmr_mfc_delete(mrt, &mfc, parent);
S
Stephen Hemminger 已提交
1353
		else
E
Eric Dumazet 已提交
1354
			ret = ipmr_mfc_add(net, mrt, &mfc,
1355 1356
					   sk == rtnl_dereference(mrt->mroute_sk),
					   parent);
S
Stephen Hemminger 已提交
1357 1358
		rtnl_unlock();
		return ret;
L
Linus Torvalds 已提交
1359 1360 1361
		/*
		 *	Control PIM assert.
		 */
S
Stephen Hemminger 已提交
1362 1363 1364
	case MRT_ASSERT:
	{
		int v;
1365 1366
		if (optlen != sizeof(v))
			return -EINVAL;
E
Eric Dumazet 已提交
1367
		if (get_user(v, (int __user *)optval))
S
Stephen Hemminger 已提交
1368
			return -EFAULT;
1369
		mrt->mroute_do_assert = v;
S
Stephen Hemminger 已提交
1370 1371
		return 0;
	}
L
Linus Torvalds 已提交
1372
#ifdef CONFIG_IP_PIMSM
S
Stephen Hemminger 已提交
1373 1374
	case MRT_PIM:
	{
S
Stephen Hemminger 已提交
1375 1376
		int v;

1377 1378
		if (optlen != sizeof(v))
			return -EINVAL;
E
Eric Dumazet 已提交
1379
		if (get_user(v, (int __user *)optval))
S
Stephen Hemminger 已提交
1380
			return -EFAULT;
1381
		v = !!v;
S
Stephen Hemminger 已提交
1382

S
Stephen Hemminger 已提交
1383 1384
		rtnl_lock();
		ret = 0;
1385 1386 1387
		if (v != mrt->mroute_do_pim) {
			mrt->mroute_do_pim = v;
			mrt->mroute_do_assert = v;
L
Linus Torvalds 已提交
1388
		}
S
Stephen Hemminger 已提交
1389 1390 1391
		rtnl_unlock();
		return ret;
	}
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402
#endif
#ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES
	case MRT_TABLE:
	{
		u32 v;

		if (optlen != sizeof(u32))
			return -EINVAL;
		if (get_user(v, (u32 __user *)optval))
			return -EFAULT;

1403 1404 1405 1406
		/* "pimreg%u" should not exceed 16 bytes (IFNAMSIZ) */
		if (v != RT_TABLE_DEFAULT && v >= 1000000000)
			return -EINVAL;

1407 1408
		rtnl_lock();
		ret = 0;
E
Eric Dumazet 已提交
1409 1410 1411 1412 1413
		if (sk == rtnl_dereference(mrt->mroute_sk)) {
			ret = -EBUSY;
		} else {
			if (!ipmr_new_table(net, v))
				ret = -ENOMEM;
1414 1415
			else
				raw_sk(sk)->ipmr_table = v;
E
Eric Dumazet 已提交
1416
		}
1417 1418 1419
		rtnl_unlock();
		return ret;
	}
L
Linus Torvalds 已提交
1420
#endif
S
Stephen Hemminger 已提交
1421 1422 1423 1424 1425 1426
	/*
	 *	Spurious command, or MRT_VERSION which you cannot
	 *	set.
	 */
	default:
		return -ENOPROTOOPT;
L
Linus Torvalds 已提交
1427 1428 1429 1430 1431 1432
	}
}

/*
 *	Getsock opt support for the multicast routing system.
 */
1433

J
Jianjun Kong 已提交
1434
int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen)
L
Linus Torvalds 已提交
1435 1436 1437
{
	int olr;
	int val;
1438
	struct net *net = sock_net(sk);
1439 1440
	struct mr_table *mrt;

1441 1442 1443 1444
	if (sk->sk_type != SOCK_RAW ||
	    inet_sk(sk)->inet_num != IPPROTO_IGMP)
		return -EOPNOTSUPP;

1445 1446 1447
	mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
	if (mrt == NULL)
		return -ENOENT;
L
Linus Torvalds 已提交
1448

J
Jianjun Kong 已提交
1449
	if (optname != MRT_VERSION &&
L
Linus Torvalds 已提交
1450
#ifdef CONFIG_IP_PIMSM
E
Eric Dumazet 已提交
1451
	   optname != MRT_PIM &&
L
Linus Torvalds 已提交
1452
#endif
E
Eric Dumazet 已提交
1453
	   optname != MRT_ASSERT)
L
Linus Torvalds 已提交
1454 1455 1456 1457 1458 1459 1460 1461
		return -ENOPROTOOPT;

	if (get_user(olr, optlen))
		return -EFAULT;

	olr = min_t(unsigned int, olr, sizeof(int));
	if (olr < 0)
		return -EINVAL;
1462

J
Jianjun Kong 已提交
1463
	if (put_user(olr, optlen))
L
Linus Torvalds 已提交
1464
		return -EFAULT;
J
Jianjun Kong 已提交
1465 1466
	if (optname == MRT_VERSION)
		val = 0x0305;
L
Linus Torvalds 已提交
1467
#ifdef CONFIG_IP_PIMSM
J
Jianjun Kong 已提交
1468
	else if (optname == MRT_PIM)
1469
		val = mrt->mroute_do_pim;
L
Linus Torvalds 已提交
1470 1471
#endif
	else
1472
		val = mrt->mroute_do_assert;
J
Jianjun Kong 已提交
1473
	if (copy_to_user(optval, &val, olr))
L
Linus Torvalds 已提交
1474 1475 1476 1477 1478 1479 1480
		return -EFAULT;
	return 0;
}

/*
 *	The IP multicast ioctl support routines.
 */
1481

L
Linus Torvalds 已提交
1482 1483 1484 1485 1486 1487
int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg)
{
	struct sioc_sg_req sr;
	struct sioc_vif_req vr;
	struct vif_device *vif;
	struct mfc_cache *c;
1488
	struct net *net = sock_net(sk);
1489 1490 1491 1492 1493
	struct mr_table *mrt;

	mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
	if (mrt == NULL)
		return -ENOENT;
1494

S
Stephen Hemminger 已提交
1495 1496
	switch (cmd) {
	case SIOCGETVIFCNT:
J
Jianjun Kong 已提交
1497
		if (copy_from_user(&vr, arg, sizeof(vr)))
S
Stephen Hemminger 已提交
1498
			return -EFAULT;
1499
		if (vr.vifi >= mrt->maxvif)
S
Stephen Hemminger 已提交
1500 1501
			return -EINVAL;
		read_lock(&mrt_lock);
1502 1503
		vif = &mrt->vif_table[vr.vifi];
		if (VIF_EXISTS(mrt, vr.vifi)) {
J
Jianjun Kong 已提交
1504 1505 1506 1507
			vr.icount = vif->pkt_in;
			vr.ocount = vif->pkt_out;
			vr.ibytes = vif->bytes_in;
			vr.obytes = vif->bytes_out;
L
Linus Torvalds 已提交
1508 1509
			read_unlock(&mrt_lock);

J
Jianjun Kong 已提交
1510
			if (copy_to_user(arg, &vr, sizeof(vr)))
S
Stephen Hemminger 已提交
1511 1512 1513 1514 1515 1516
				return -EFAULT;
			return 0;
		}
		read_unlock(&mrt_lock);
		return -EADDRNOTAVAIL;
	case SIOCGETSGCNT:
J
Jianjun Kong 已提交
1517
		if (copy_from_user(&sr, arg, sizeof(sr)))
S
Stephen Hemminger 已提交
1518 1519
			return -EFAULT;

1520
		rcu_read_lock();
1521
		c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr);
S
Stephen Hemminger 已提交
1522 1523 1524 1525
		if (c) {
			sr.pktcnt = c->mfc_un.res.pkt;
			sr.bytecnt = c->mfc_un.res.bytes;
			sr.wrong_if = c->mfc_un.res.wrong_if;
1526
			rcu_read_unlock();
S
Stephen Hemminger 已提交
1527

J
Jianjun Kong 已提交
1528
			if (copy_to_user(arg, &sr, sizeof(sr)))
S
Stephen Hemminger 已提交
1529 1530 1531
				return -EFAULT;
			return 0;
		}
1532
		rcu_read_unlock();
S
Stephen Hemminger 已提交
1533 1534 1535
		return -EADDRNOTAVAIL;
	default:
		return -ENOIOCTLCMD;
L
Linus Torvalds 已提交
1536 1537 1538
	}
}

1539 1540 1541 1542 1543 1544 1545 1546 1547
#ifdef CONFIG_COMPAT
struct compat_sioc_sg_req {
	struct in_addr src;
	struct in_addr grp;
	compat_ulong_t pktcnt;
	compat_ulong_t bytecnt;
	compat_ulong_t wrong_if;
};

1548 1549 1550 1551 1552 1553 1554 1555
struct compat_sioc_vif_req {
	vifi_t	vifi;		/* Which iface */
	compat_ulong_t icount;
	compat_ulong_t ocount;
	compat_ulong_t ibytes;
	compat_ulong_t obytes;
};

1556 1557
int ipmr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg)
{
1558
	struct compat_sioc_sg_req sr;
1559 1560
	struct compat_sioc_vif_req vr;
	struct vif_device *vif;
1561 1562 1563 1564 1565 1566 1567 1568 1569
	struct mfc_cache *c;
	struct net *net = sock_net(sk);
	struct mr_table *mrt;

	mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
	if (mrt == NULL)
		return -ENOENT;

	switch (cmd) {
1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589
	case SIOCGETVIFCNT:
		if (copy_from_user(&vr, arg, sizeof(vr)))
			return -EFAULT;
		if (vr.vifi >= mrt->maxvif)
			return -EINVAL;
		read_lock(&mrt_lock);
		vif = &mrt->vif_table[vr.vifi];
		if (VIF_EXISTS(mrt, vr.vifi)) {
			vr.icount = vif->pkt_in;
			vr.ocount = vif->pkt_out;
			vr.ibytes = vif->bytes_in;
			vr.obytes = vif->bytes_out;
			read_unlock(&mrt_lock);

			if (copy_to_user(arg, &vr, sizeof(vr)))
				return -EFAULT;
			return 0;
		}
		read_unlock(&mrt_lock);
		return -EADDRNOTAVAIL;
1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613
	case SIOCGETSGCNT:
		if (copy_from_user(&sr, arg, sizeof(sr)))
			return -EFAULT;

		rcu_read_lock();
		c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr);
		if (c) {
			sr.pktcnt = c->mfc_un.res.pkt;
			sr.bytecnt = c->mfc_un.res.bytes;
			sr.wrong_if = c->mfc_un.res.wrong_if;
			rcu_read_unlock();

			if (copy_to_user(arg, &sr, sizeof(sr)))
				return -EFAULT;
			return 0;
		}
		rcu_read_unlock();
		return -EADDRNOTAVAIL;
	default:
		return -ENOIOCTLCMD;
	}
}
#endif

L
Linus Torvalds 已提交
1614 1615 1616

static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr)
{
1617
	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
1618
	struct net *net = dev_net(dev);
1619
	struct mr_table *mrt;
L
Linus Torvalds 已提交
1620 1621
	struct vif_device *v;
	int ct;
1622

L
Linus Torvalds 已提交
1623 1624
	if (event != NETDEV_UNREGISTER)
		return NOTIFY_DONE;
1625 1626 1627 1628 1629

	ipmr_for_each_table(mrt, net) {
		v = &mrt->vif_table[0];
		for (ct = 0; ct < mrt->maxvif; ct++, v++) {
			if (v->dev == dev)
1630
				vif_delete(mrt, ct, 1, NULL);
1631
		}
L
Linus Torvalds 已提交
1632 1633 1634 1635 1636
	}
	return NOTIFY_DONE;
}


J
Jianjun Kong 已提交
1637
static struct notifier_block ip_mr_notifier = {
L
Linus Torvalds 已提交
1638 1639 1640 1641
	.notifier_call = ipmr_device_event,
};

/*
E
Eric Dumazet 已提交
1642
 *	Encapsulate a packet by attaching a valid IPIP header to it.
L
Linus Torvalds 已提交
1643 1644 1645
 *	This avoids tunnel drivers and other mess and gives us the speed so
 *	important for multicast video.
 */
1646

A
Al Viro 已提交
1647
static void ip_encap(struct sk_buff *skb, __be32 saddr, __be32 daddr)
L
Linus Torvalds 已提交
1648
{
1649
	struct iphdr *iph;
1650
	const struct iphdr *old_iph = ip_hdr(skb);
1651 1652

	skb_push(skb, sizeof(struct iphdr));
1653
	skb->transport_header = skb->network_header;
1654
	skb_reset_network_header(skb);
1655
	iph = ip_hdr(skb);
L
Linus Torvalds 已提交
1656

E
Eric Dumazet 已提交
1657
	iph->version	=	4;
1658 1659
	iph->tos	=	old_iph->tos;
	iph->ttl	=	old_iph->ttl;
L
Linus Torvalds 已提交
1660 1661 1662 1663 1664 1665
	iph->frag_off	=	0;
	iph->daddr	=	daddr;
	iph->saddr	=	saddr;
	iph->protocol	=	IPPROTO_IPIP;
	iph->ihl	=	5;
	iph->tot_len	=	htons(skb->len);
E
Eric Dumazet 已提交
1666
	ip_select_ident(skb, NULL);
L
Linus Torvalds 已提交
1667 1668 1669 1670 1671 1672 1673 1674
	ip_send_check(iph);

	memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
	nf_reset(skb);
}

static inline int ipmr_forward_finish(struct sk_buff *skb)
{
E
Eric Dumazet 已提交
1675
	struct ip_options *opt = &(IPCB(skb)->opt);
L
Linus Torvalds 已提交
1676

E
Eric Dumazet 已提交
1677
	IP_INC_STATS_BH(dev_net(skb_dst(skb)->dev), IPSTATS_MIB_OUTFORWDATAGRAMS);
1678
	IP_ADD_STATS_BH(dev_net(skb_dst(skb)->dev), IPSTATS_MIB_OUTOCTETS, skb->len);
L
Linus Torvalds 已提交
1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689

	if (unlikely(opt->optlen))
		ip_forward_options(skb);

	return dst_output(skb);
}

/*
 *	Processing handlers for ipmr_forward
 */

1690 1691
static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
			    struct sk_buff *skb, struct mfc_cache *c, int vifi)
L
Linus Torvalds 已提交
1692
{
1693
	const struct iphdr *iph = ip_hdr(skb);
1694
	struct vif_device *vif = &mrt->vif_table[vifi];
L
Linus Torvalds 已提交
1695 1696
	struct net_device *dev;
	struct rtable *rt;
1697
	struct flowi4 fl4;
L
Linus Torvalds 已提交
1698 1699 1700 1701 1702 1703 1704 1705
	int    encap = 0;

	if (vif->dev == NULL)
		goto out_free;

#ifdef CONFIG_IP_PIMSM
	if (vif->flags & VIFF_REGISTER) {
		vif->pkt_out++;
J
Jianjun Kong 已提交
1706
		vif->bytes_out += skb->len;
1707 1708
		vif->dev->stats.tx_bytes += skb->len;
		vif->dev->stats.tx_packets++;
1709
		ipmr_cache_report(mrt, skb, vifi, IGMPMSG_WHOLEPKT);
1710
		goto out_free;
L
Linus Torvalds 已提交
1711 1712 1713
	}
#endif

E
Eric Dumazet 已提交
1714
	if (vif->flags & VIFF_TUNNEL) {
1715
		rt = ip_route_output_ports(net, &fl4, NULL,
1716 1717 1718 1719
					   vif->remote, vif->local,
					   0, 0,
					   IPPROTO_IPIP,
					   RT_TOS(iph->tos), vif->link);
1720
		if (IS_ERR(rt))
L
Linus Torvalds 已提交
1721 1722 1723
			goto out_free;
		encap = sizeof(struct iphdr);
	} else {
1724
		rt = ip_route_output_ports(net, &fl4, NULL, iph->daddr, 0,
1725 1726 1727
					   0, 0,
					   IPPROTO_IPIP,
					   RT_TOS(iph->tos), vif->link);
1728
		if (IS_ERR(rt))
L
Linus Torvalds 已提交
1729 1730 1731
			goto out_free;
	}

1732
	dev = rt->dst.dev;
L
Linus Torvalds 已提交
1733

1734
	if (skb->len+encap > dst_mtu(&rt->dst) && (ntohs(iph->frag_off) & IP_DF)) {
L
Linus Torvalds 已提交
1735
		/* Do not fragment multicasts. Alas, IPv4 does not
E
Eric Dumazet 已提交
1736 1737
		 * allow to send ICMP, so that packets will disappear
		 * to blackhole.
L
Linus Torvalds 已提交
1738 1739
		 */

1740
		IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_FRAGFAILS);
L
Linus Torvalds 已提交
1741 1742 1743 1744
		ip_rt_put(rt);
		goto out_free;
	}

1745
	encap += LL_RESERVED_SPACE(dev) + rt->dst.header_len;
L
Linus Torvalds 已提交
1746 1747

	if (skb_cow(skb, encap)) {
1748
		ip_rt_put(rt);
L
Linus Torvalds 已提交
1749 1750 1751 1752
		goto out_free;
	}

	vif->pkt_out++;
J
Jianjun Kong 已提交
1753
	vif->bytes_out += skb->len;
L
Linus Torvalds 已提交
1754

E
Eric Dumazet 已提交
1755
	skb_dst_drop(skb);
1756
	skb_dst_set(skb, &rt->dst);
1757
	ip_decrease_ttl(ip_hdr(skb));
L
Linus Torvalds 已提交
1758 1759

	/* FIXME: forward and output firewalls used to be called here.
E
Eric Dumazet 已提交
1760 1761
	 * What do we do with netfilter? -- RR
	 */
L
Linus Torvalds 已提交
1762 1763 1764
	if (vif->flags & VIFF_TUNNEL) {
		ip_encap(skb, vif->local, vif->remote);
		/* FIXME: extra output firewall step used to be here. --RR */
1765 1766
		vif->dev->stats.tx_packets++;
		vif->dev->stats.tx_bytes += skb->len;
L
Linus Torvalds 已提交
1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781
	}

	IPCB(skb)->flags |= IPSKB_FORWARDED;

	/*
	 * RFC1584 teaches, that DVMRP/PIM router must deliver packets locally
	 * not only before forwarding, but after forwarding on all output
	 * interfaces. It is clear, if mrouter runs a multicasting
	 * program, it should receive packets not depending to what interface
	 * program is joined.
	 * If we will not make it, the program will have to join on all
	 * interfaces. On the other hand, multihoming host (or router, but
	 * not mrouter) cannot join to more than one interface - it will
	 * result in receiving multiple packets.
	 */
1782
	NF_HOOK(NFPROTO_IPV4, NF_INET_FORWARD, skb, skb->dev, dev,
L
Linus Torvalds 已提交
1783 1784 1785 1786 1787 1788 1789
		ipmr_forward_finish);
	return;

out_free:
	kfree_skb(skb);
}

1790
static int ipmr_find_vif(struct mr_table *mrt, struct net_device *dev)
L
Linus Torvalds 已提交
1791 1792
{
	int ct;
1793 1794 1795

	for (ct = mrt->maxvif-1; ct >= 0; ct--) {
		if (mrt->vif_table[ct].dev == dev)
L
Linus Torvalds 已提交
1796 1797 1798 1799 1800 1801 1802
			break;
	}
	return ct;
}

/* "local" means that we should preserve one skb (for local delivery) */

1803 1804 1805
static void ip_mr_forward(struct net *net, struct mr_table *mrt,
			  struct sk_buff *skb, struct mfc_cache *cache,
			  int local)
L
Linus Torvalds 已提交
1806 1807 1808
{
	int psend = -1;
	int vif, ct;
1809
	int true_vifi = ipmr_find_vif(mrt, skb->dev);
L
Linus Torvalds 已提交
1810 1811 1812 1813 1814

	vif = cache->mfc_parent;
	cache->mfc_un.res.pkt++;
	cache->mfc_un.res.bytes += skb->len;

1815
	if (cache->mfc_origin == htonl(INADDR_ANY) && true_vifi >= 0) {
1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
		struct mfc_cache *cache_proxy;

		/* For an (*,G) entry, we only check that the incomming
		 * interface is part of the static tree.
		 */
		cache_proxy = ipmr_cache_find_any_parent(mrt, vif);
		if (cache_proxy &&
		    cache_proxy->mfc_un.res.ttls[true_vifi] < 255)
			goto forward;
	}

L
Linus Torvalds 已提交
1827 1828 1829
	/*
	 * Wrong interface: drop packet and (maybe) send PIM assert.
	 */
1830
	if (mrt->vif_table[vif].dev != skb->dev) {
1831
		if (rt_is_output_route(skb_rtable(skb))) {
L
Linus Torvalds 已提交
1832
			/* It is our own packet, looped back.
E
Eric Dumazet 已提交
1833 1834 1835 1836 1837 1838 1839 1840 1841
			 * Very complicated situation...
			 *
			 * The best workaround until routing daemons will be
			 * fixed is not to redistribute packet, if it was
			 * send through wrong interface. It means, that
			 * multicast applications WILL NOT work for
			 * (S,G), which have default multicast route pointing
			 * to wrong oif. In any case, it is not a good
			 * idea to use multicasting applications on router.
L
Linus Torvalds 已提交
1842 1843 1844 1845 1846 1847
			 */
			goto dont_forward;
		}

		cache->mfc_un.res.wrong_if++;

1848
		if (true_vifi >= 0 && mrt->mroute_do_assert &&
L
Linus Torvalds 已提交
1849
		    /* pimsm uses asserts, when switching from RPT to SPT,
E
Eric Dumazet 已提交
1850 1851 1852
		     * so that we cannot check that packet arrived on an oif.
		     * It is bad, but otherwise we would need to move pretty
		     * large chunk of pimd to kernel. Ough... --ANK
L
Linus Torvalds 已提交
1853
		     */
1854
		    (mrt->mroute_do_pim ||
1855
		     cache->mfc_un.res.ttls[true_vifi] < 255) &&
1856
		    time_after(jiffies,
L
Linus Torvalds 已提交
1857 1858
			       cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) {
			cache->mfc_un.res.last_assert = jiffies;
1859
			ipmr_cache_report(mrt, skb, true_vifi, IGMPMSG_WRONGVIF);
L
Linus Torvalds 已提交
1860 1861 1862 1863
		}
		goto dont_forward;
	}

1864
forward:
1865 1866
	mrt->vif_table[vif].pkt_in++;
	mrt->vif_table[vif].bytes_in += skb->len;
L
Linus Torvalds 已提交
1867 1868 1869 1870

	/*
	 *	Forward the frame
	 */
1871 1872
	if (cache->mfc_origin == htonl(INADDR_ANY) &&
	    cache->mfc_mcastgrp == htonl(INADDR_ANY)) {
1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885
		if (true_vifi >= 0 &&
		    true_vifi != cache->mfc_parent &&
		    ip_hdr(skb)->ttl >
				cache->mfc_un.res.ttls[cache->mfc_parent]) {
			/* It's an (*,*) entry and the packet is not coming from
			 * the upstream: forward the packet to the upstream
			 * only.
			 */
			psend = cache->mfc_parent;
			goto last_forward;
		}
		goto dont_forward;
	}
E
Eric Dumazet 已提交
1886 1887
	for (ct = cache->mfc_un.res.maxvif - 1;
	     ct >= cache->mfc_un.res.minvif; ct--) {
1888
		/* For (*,G) entry, don't forward to the incoming interface */
1889 1890
		if ((cache->mfc_origin != htonl(INADDR_ANY) ||
		     ct != true_vifi) &&
1891
		    ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) {
L
Linus Torvalds 已提交
1892 1893
			if (psend != -1) {
				struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
E
Eric Dumazet 已提交
1894

L
Linus Torvalds 已提交
1895
				if (skb2)
1896 1897
					ipmr_queue_xmit(net, mrt, skb2, cache,
							psend);
L
Linus Torvalds 已提交
1898
			}
J
Jianjun Kong 已提交
1899
			psend = ct;
L
Linus Torvalds 已提交
1900 1901
		}
	}
1902
last_forward:
L
Linus Torvalds 已提交
1903 1904 1905
	if (psend != -1) {
		if (local) {
			struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
E
Eric Dumazet 已提交
1906

L
Linus Torvalds 已提交
1907
			if (skb2)
1908
				ipmr_queue_xmit(net, mrt, skb2, cache, psend);
L
Linus Torvalds 已提交
1909
		} else {
1910
			ipmr_queue_xmit(net, mrt, skb, cache, psend);
1911
			return;
L
Linus Torvalds 已提交
1912 1913 1914 1915 1916 1917 1918 1919
		}
	}

dont_forward:
	if (!local)
		kfree_skb(skb);
}

1920
static struct mr_table *ipmr_rt_fib_lookup(struct net *net, struct sk_buff *skb)
1921
{
1922 1923
	struct rtable *rt = skb_rtable(skb);
	struct iphdr *iph = ip_hdr(skb);
D
David S. Miller 已提交
1924
	struct flowi4 fl4 = {
1925 1926
		.daddr = iph->daddr,
		.saddr = iph->saddr,
1927
		.flowi4_tos = RT_TOS(iph->tos),
D
David S. Miller 已提交
1928 1929 1930
		.flowi4_oif = (rt_is_output_route(rt) ?
			       skb->dev->ifindex : 0),
		.flowi4_iif = (rt_is_output_route(rt) ?
1931
			       LOOPBACK_IFINDEX :
D
David S. Miller 已提交
1932
			       skb->dev->ifindex),
1933
		.flowi4_mark = skb->mark,
1934 1935 1936 1937
	};
	struct mr_table *mrt;
	int err;

D
David S. Miller 已提交
1938
	err = ipmr_fib_lookup(net, &fl4, &mrt);
1939 1940 1941 1942
	if (err)
		return ERR_PTR(err);
	return mrt;
}
L
Linus Torvalds 已提交
1943 1944 1945

/*
 *	Multicast packets for forwarding arrive here
E
Eric Dumazet 已提交
1946
 *	Called with rcu_read_lock();
L
Linus Torvalds 已提交
1947 1948 1949 1950 1951
 */

int ip_mr_input(struct sk_buff *skb)
{
	struct mfc_cache *cache;
1952
	struct net *net = dev_net(skb->dev);
E
Eric Dumazet 已提交
1953
	int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL;
1954
	struct mr_table *mrt;
L
Linus Torvalds 已提交
1955 1956

	/* Packet is looped back after forward, it should not be
E
Eric Dumazet 已提交
1957
	 * forwarded second time, but still can be delivered locally.
L
Linus Torvalds 已提交
1958
	 */
E
Eric Dumazet 已提交
1959
	if (IPCB(skb)->flags & IPSKB_FORWARDED)
L
Linus Torvalds 已提交
1960 1961
		goto dont_forward;

1962
	mrt = ipmr_rt_fib_lookup(net, skb);
1963 1964 1965
	if (IS_ERR(mrt)) {
		kfree_skb(skb);
		return PTR_ERR(mrt);
1966
	}
L
Linus Torvalds 已提交
1967
	if (!local) {
E
Eric Dumazet 已提交
1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985
		if (IPCB(skb)->opt.router_alert) {
			if (ip_call_ra_chain(skb))
				return 0;
		} else if (ip_hdr(skb)->protocol == IPPROTO_IGMP) {
			/* IGMPv1 (and broken IGMPv2 implementations sort of
			 * Cisco IOS <= 11.2(8)) do not put router alert
			 * option to IGMP packets destined to routable
			 * groups. It is very bad, because it means
			 * that we can forward NO IGMP messages.
			 */
			struct sock *mroute_sk;

			mroute_sk = rcu_dereference(mrt->mroute_sk);
			if (mroute_sk) {
				nf_reset(skb);
				raw_rcv(mroute_sk, skb);
				return 0;
			}
L
Linus Torvalds 已提交
1986 1987 1988
		    }
	}

1989
	/* already under rcu_read_lock() */
1990
	cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr);
1991 1992 1993 1994 1995 1996 1997
	if (cache == NULL) {
		int vif = ipmr_find_vif(mrt, skb->dev);

		if (vif >= 0)
			cache = ipmr_cache_find_any(mrt, ip_hdr(skb)->daddr,
						    vif);
	}
L
Linus Torvalds 已提交
1998 1999 2000 2001

	/*
	 *	No usable cache entry
	 */
J
Jianjun Kong 已提交
2002
	if (cache == NULL) {
L
Linus Torvalds 已提交
2003 2004 2005 2006 2007
		int vif;

		if (local) {
			struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
			ip_local_deliver(skb);
2008
			if (skb2 == NULL)
L
Linus Torvalds 已提交
2009 2010 2011 2012
				return -ENOBUFS;
			skb = skb2;
		}

2013
		read_lock(&mrt_lock);
2014
		vif = ipmr_find_vif(mrt, skb->dev);
L
Linus Torvalds 已提交
2015
		if (vif >= 0) {
2016
			int err2 = ipmr_cache_unresolved(mrt, vif, skb);
L
Linus Torvalds 已提交
2017 2018
			read_unlock(&mrt_lock);

2019
			return err2;
L
Linus Torvalds 已提交
2020 2021 2022 2023 2024 2025
		}
		read_unlock(&mrt_lock);
		kfree_skb(skb);
		return -ENODEV;
	}

2026
	read_lock(&mrt_lock);
2027
	ip_mr_forward(net, mrt, skb, cache, local);
L
Linus Torvalds 已提交
2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041
	read_unlock(&mrt_lock);

	if (local)
		return ip_local_deliver(skb);

	return 0;

dont_forward:
	if (local)
		return ip_local_deliver(skb);
	kfree_skb(skb);
	return 0;
}

I
Ilpo Järvinen 已提交
2042
#ifdef CONFIG_IP_PIMSM
2043
/* called with rcu_read_lock() */
2044 2045
static int __pim_rcv(struct mr_table *mrt, struct sk_buff *skb,
		     unsigned int pimlen)
L
Linus Torvalds 已提交
2046
{
I
Ilpo Järvinen 已提交
2047 2048
	struct net_device *reg_dev = NULL;
	struct iphdr *encap;
L
Linus Torvalds 已提交
2049

I
Ilpo Järvinen 已提交
2050
	encap = (struct iphdr *)(skb_transport_header(skb) + pimlen);
L
Linus Torvalds 已提交
2051
	/*
E
Eric Dumazet 已提交
2052 2053 2054 2055
	 * Check that:
	 * a. packet is really sent to a multicast group
	 * b. packet is not a NULL-REGISTER
	 * c. packet is not truncated
L
Linus Torvalds 已提交
2056
	 */
2057
	if (!ipv4_is_multicast(encap->daddr) ||
L
Linus Torvalds 已提交
2058
	    encap->tot_len == 0 ||
I
Ilpo Järvinen 已提交
2059 2060
	    ntohs(encap->tot_len) + pimlen > skb->len)
		return 1;
L
Linus Torvalds 已提交
2061 2062

	read_lock(&mrt_lock);
2063 2064
	if (mrt->mroute_reg_vif_num >= 0)
		reg_dev = mrt->vif_table[mrt->mroute_reg_vif_num].dev;
L
Linus Torvalds 已提交
2065 2066
	read_unlock(&mrt_lock);

2067
	if (reg_dev == NULL)
I
Ilpo Järvinen 已提交
2068
		return 1;
L
Linus Torvalds 已提交
2069

2070
	skb->mac_header = skb->network_header;
2071
	skb_pull(skb, (u8 *)encap - skb->data);
2072
	skb_reset_network_header(skb);
L
Linus Torvalds 已提交
2073
	skb->protocol = htons(ETH_P_IP);
2074
	skb->ip_summed = CHECKSUM_NONE;
2075

2076
	skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev));
2077

L
Linus Torvalds 已提交
2078
	netif_rx(skb);
I
Ilpo Järvinen 已提交
2079

2080
	return NET_RX_SUCCESS;
I
Ilpo Järvinen 已提交
2081 2082 2083 2084 2085 2086 2087 2088
}
#endif

#ifdef CONFIG_IP_PIMSM_V1
/*
 * Handle IGMP messages of PIMv1
 */

E
Eric Dumazet 已提交
2089
int pim_rcv_v1(struct sk_buff *skb)
I
Ilpo Järvinen 已提交
2090 2091
{
	struct igmphdr *pim;
2092
	struct net *net = dev_net(skb->dev);
2093
	struct mr_table *mrt;
I
Ilpo Järvinen 已提交
2094 2095 2096 2097 2098 2099

	if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr)))
		goto drop;

	pim = igmp_hdr(skb);

2100
	mrt = ipmr_rt_fib_lookup(net, skb);
2101 2102
	if (IS_ERR(mrt))
		goto drop;
2103
	if (!mrt->mroute_do_pim ||
I
Ilpo Järvinen 已提交
2104 2105 2106
	    pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER)
		goto drop;

2107
	if (__pim_rcv(mrt, skb, sizeof(*pim))) {
I
Ilpo Järvinen 已提交
2108 2109 2110
drop:
		kfree_skb(skb);
	}
L
Linus Torvalds 已提交
2111 2112 2113 2114 2115
	return 0;
}
#endif

#ifdef CONFIG_IP_PIMSM_V2
E
Eric Dumazet 已提交
2116
static int pim_rcv(struct sk_buff *skb)
L
Linus Torvalds 已提交
2117 2118
{
	struct pimreghdr *pim;
2119 2120
	struct net *net = dev_net(skb->dev);
	struct mr_table *mrt;
L
Linus Torvalds 已提交
2121

I
Ilpo Järvinen 已提交
2122
	if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr)))
L
Linus Torvalds 已提交
2123 2124
		goto drop;

2125
	pim = (struct pimreghdr *)skb_transport_header(skb);
E
Eric Dumazet 已提交
2126 2127
	if (pim->type != ((PIM_VERSION << 4) | (PIM_REGISTER)) ||
	    (pim->flags & PIM_NULL_REGISTER) ||
2128
	    (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 &&
2129
	     csum_fold(skb_checksum(skb, 0, skb->len, 0))))
L
Linus Torvalds 已提交
2130 2131
		goto drop;

2132
	mrt = ipmr_rt_fib_lookup(net, skb);
2133 2134
	if (IS_ERR(mrt))
		goto drop;
2135
	if (__pim_rcv(mrt, skb, sizeof(*pim))) {
I
Ilpo Järvinen 已提交
2136 2137 2138
drop:
		kfree_skb(skb);
	}
L
Linus Torvalds 已提交
2139 2140 2141 2142
	return 0;
}
#endif

2143 2144
static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
			      struct mfc_cache *c, struct rtmsg *rtm)
L
Linus Torvalds 已提交
2145 2146 2147
{
	int ct;
	struct rtnexthop *nhp;
T
Thomas Graf 已提交
2148
	struct nlattr *mp_attr;
2149
	struct rta_mfc_stats mfcs;
L
Linus Torvalds 已提交
2150

2151
	/* If cache is unresolved, don't try to parse IIF and OIF */
2152
	if (c->mfc_parent >= MAXVIFS)
2153 2154
		return -ENOENT;

T
Thomas Graf 已提交
2155 2156 2157
	if (VIF_EXISTS(mrt, c->mfc_parent) &&
	    nla_put_u32(skb, RTA_IIF, mrt->vif_table[c->mfc_parent].dev->ifindex) < 0)
		return -EMSGSIZE;
L
Linus Torvalds 已提交
2158

T
Thomas Graf 已提交
2159 2160
	if (!(mp_attr = nla_nest_start(skb, RTA_MULTIPATH)))
		return -EMSGSIZE;
L
Linus Torvalds 已提交
2161 2162

	for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) {
2163
		if (VIF_EXISTS(mrt, ct) && c->mfc_un.res.ttls[ct] < 255) {
T
Thomas Graf 已提交
2164 2165 2166 2167 2168
			if (!(nhp = nla_reserve_nohdr(skb, sizeof(*nhp)))) {
				nla_nest_cancel(skb, mp_attr);
				return -EMSGSIZE;
			}

L
Linus Torvalds 已提交
2169 2170
			nhp->rtnh_flags = 0;
			nhp->rtnh_hops = c->mfc_un.res.ttls[ct];
2171
			nhp->rtnh_ifindex = mrt->vif_table[ct].dev->ifindex;
L
Linus Torvalds 已提交
2172 2173 2174
			nhp->rtnh_len = sizeof(*nhp);
		}
	}
T
Thomas Graf 已提交
2175 2176 2177

	nla_nest_end(skb, mp_attr);

2178 2179 2180 2181 2182 2183
	mfcs.mfcs_packets = c->mfc_un.res.pkt;
	mfcs.mfcs_bytes = c->mfc_un.res.bytes;
	mfcs.mfcs_wrong_if = c->mfc_un.res.wrong_if;
	if (nla_put(skb, RTA_MFC_STATS, sizeof(mfcs), &mfcs) < 0)
		return -EMSGSIZE;

L
Linus Torvalds 已提交
2184 2185 2186 2187
	rtm->rtm_type = RTN_MULTICAST;
	return 1;
}

2188 2189 2190
int ipmr_get_route(struct net *net, struct sk_buff *skb,
		   __be32 saddr, __be32 daddr,
		   struct rtmsg *rtm, int nowait)
L
Linus Torvalds 已提交
2191 2192
{
	struct mfc_cache *cache;
2193 2194
	struct mr_table *mrt;
	int err;
L
Linus Torvalds 已提交
2195

2196 2197 2198 2199
	mrt = ipmr_get_table(net, RT_TABLE_DEFAULT);
	if (mrt == NULL)
		return -ENOENT;

2200
	rcu_read_lock();
2201
	cache = ipmr_cache_find(mrt, saddr, daddr);
2202 2203
	if (cache == NULL && skb->dev) {
		int vif = ipmr_find_vif(mrt, skb->dev);
L
Linus Torvalds 已提交
2204

2205 2206 2207
		if (vif >= 0)
			cache = ipmr_cache_find_any(mrt, daddr, vif);
	}
J
Jianjun Kong 已提交
2208
	if (cache == NULL) {
2209
		struct sk_buff *skb2;
2210
		struct iphdr *iph;
L
Linus Torvalds 已提交
2211
		struct net_device *dev;
E
Eric Dumazet 已提交
2212
		int vif = -1;
L
Linus Torvalds 已提交
2213 2214

		if (nowait) {
2215
			rcu_read_unlock();
L
Linus Torvalds 已提交
2216 2217 2218 2219
			return -EAGAIN;
		}

		dev = skb->dev;
2220
		read_lock(&mrt_lock);
E
Eric Dumazet 已提交
2221 2222 2223
		if (dev)
			vif = ipmr_find_vif(mrt, dev);
		if (vif < 0) {
L
Linus Torvalds 已提交
2224
			read_unlock(&mrt_lock);
2225
			rcu_read_unlock();
L
Linus Torvalds 已提交
2226 2227
			return -ENODEV;
		}
2228 2229 2230
		skb2 = skb_clone(skb, GFP_ATOMIC);
		if (!skb2) {
			read_unlock(&mrt_lock);
2231
			rcu_read_unlock();
2232 2233 2234
			return -ENOMEM;
		}

2235 2236
		skb_push(skb2, sizeof(struct iphdr));
		skb_reset_network_header(skb2);
2237 2238
		iph = ip_hdr(skb2);
		iph->ihl = sizeof(struct iphdr) >> 2;
2239 2240
		iph->saddr = saddr;
		iph->daddr = daddr;
2241
		iph->version = 0;
2242
		err = ipmr_cache_unresolved(mrt, vif, skb2);
L
Linus Torvalds 已提交
2243
		read_unlock(&mrt_lock);
2244
		rcu_read_unlock();
L
Linus Torvalds 已提交
2245 2246 2247
		return err;
	}

2248 2249
	read_lock(&mrt_lock);
	if (!nowait && (rtm->rtm_flags & RTM_F_NOTIFY))
L
Linus Torvalds 已提交
2250
		cache->mfc_flags |= MFC_NOTIFY;
2251
	err = __ipmr_fill_mroute(mrt, skb, cache, rtm);
L
Linus Torvalds 已提交
2252
	read_unlock(&mrt_lock);
2253
	rcu_read_unlock();
L
Linus Torvalds 已提交
2254 2255 2256
	return err;
}

2257
static int ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
2258 2259
			    u32 portid, u32 seq, struct mfc_cache *c, int cmd,
			    int flags)
2260 2261 2262
{
	struct nlmsghdr *nlh;
	struct rtmsg *rtm;
2263
	int err;
2264

2265
	nlh = nlmsg_put(skb, portid, seq, cmd, sizeof(*rtm), flags);
2266 2267 2268 2269 2270 2271 2272 2273 2274
	if (nlh == NULL)
		return -EMSGSIZE;

	rtm = nlmsg_data(nlh);
	rtm->rtm_family   = RTNL_FAMILY_IPMR;
	rtm->rtm_dst_len  = 32;
	rtm->rtm_src_len  = 32;
	rtm->rtm_tos      = 0;
	rtm->rtm_table    = mrt->id;
D
David S. Miller 已提交
2275 2276
	if (nla_put_u32(skb, RTA_TABLE, mrt->id))
		goto nla_put_failure;
2277 2278
	rtm->rtm_type     = RTN_MULTICAST;
	rtm->rtm_scope    = RT_SCOPE_UNIVERSE;
2279 2280 2281 2282
	if (c->mfc_flags & MFC_STATIC)
		rtm->rtm_protocol = RTPROT_STATIC;
	else
		rtm->rtm_protocol = RTPROT_MROUTED;
2283 2284
	rtm->rtm_flags    = 0;

D
David S. Miller 已提交
2285 2286 2287
	if (nla_put_be32(skb, RTA_SRC, c->mfc_origin) ||
	    nla_put_be32(skb, RTA_DST, c->mfc_mcastgrp))
		goto nla_put_failure;
2288 2289 2290
	err = __ipmr_fill_mroute(mrt, skb, c, rtm);
	/* do not break the dump if cache is unresolved */
	if (err < 0 && err != -ENOENT)
2291 2292
		goto nla_put_failure;

2293 2294
	nlmsg_end(skb, nlh);
	return 0;
2295 2296 2297 2298 2299 2300

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

2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333
static size_t mroute_msgsize(bool unresolved, int maxvif)
{
	size_t len =
		NLMSG_ALIGN(sizeof(struct rtmsg))
		+ nla_total_size(4)	/* RTA_TABLE */
		+ nla_total_size(4)	/* RTA_SRC */
		+ nla_total_size(4)	/* RTA_DST */
		;

	if (!unresolved)
		len = len
		      + nla_total_size(4)	/* RTA_IIF */
		      + nla_total_size(0)	/* RTA_MULTIPATH */
		      + maxvif * NLA_ALIGN(sizeof(struct rtnexthop))
						/* RTA_MFC_STATS */
		      + nla_total_size(sizeof(struct rta_mfc_stats))
		;

	return len;
}

static void mroute_netlink_event(struct mr_table *mrt, struct mfc_cache *mfc,
				 int cmd)
{
	struct net *net = read_pnet(&mrt->net);
	struct sk_buff *skb;
	int err = -ENOBUFS;

	skb = nlmsg_new(mroute_msgsize(mfc->mfc_parent >= MAXVIFS, mrt->maxvif),
			GFP_ATOMIC);
	if (skb == NULL)
		goto errout;

2334
	err = ipmr_fill_mroute(mrt, skb, 0, 0, mfc, cmd, 0);
2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346
	if (err < 0)
		goto errout;

	rtnl_notify(skb, net, 0, RTNLGRP_IPV4_MROUTE, NULL, GFP_ATOMIC);
	return;

errout:
	kfree_skb(skb);
	if (err < 0)
		rtnl_set_sk_err(net, RTNLGRP_IPV4_MROUTE, err);
}

2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359
static int ipmr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb)
{
	struct net *net = sock_net(skb->sk);
	struct mr_table *mrt;
	struct mfc_cache *mfc;
	unsigned int t = 0, s_t;
	unsigned int h = 0, s_h;
	unsigned int e = 0, s_e;

	s_t = cb->args[0];
	s_h = cb->args[1];
	s_e = cb->args[2];

2360
	rcu_read_lock();
2361 2362 2363 2364 2365 2366
	ipmr_for_each_table(mrt, net) {
		if (t < s_t)
			goto next_table;
		if (t > s_t)
			s_h = 0;
		for (h = s_h; h < MFC_LINES; h++) {
2367
			list_for_each_entry_rcu(mfc, &mrt->mfc_cache_array[h], list) {
2368 2369 2370
				if (e < s_e)
					goto next_entry;
				if (ipmr_fill_mroute(mrt, skb,
2371
						     NETLINK_CB(cb->skb).portid,
2372
						     cb->nlh->nlmsg_seq,
2373 2374
						     mfc, RTM_NEWROUTE,
						     NLM_F_MULTI) < 0)
2375 2376 2377 2378 2379 2380
					goto done;
next_entry:
				e++;
			}
			e = s_e = 0;
		}
2381 2382 2383 2384 2385 2386 2387
		spin_lock_bh(&mfc_unres_lock);
		list_for_each_entry(mfc, &mrt->mfc_unres_queue, list) {
			if (e < s_e)
				goto next_entry2;
			if (ipmr_fill_mroute(mrt, skb,
					     NETLINK_CB(cb->skb).portid,
					     cb->nlh->nlmsg_seq,
2388 2389
					     mfc, RTM_NEWROUTE,
					     NLM_F_MULTI) < 0) {
2390 2391 2392 2393 2394 2395 2396 2397
				spin_unlock_bh(&mfc_unres_lock);
				goto done;
			}
next_entry2:
			e++;
		}
		spin_unlock_bh(&mfc_unres_lock);
		e = s_e = 0;
2398 2399 2400 2401 2402
		s_h = 0;
next_table:
		t++;
	}
done:
2403
	rcu_read_unlock();
2404 2405 2406 2407 2408 2409 2410 2411

	cb->args[2] = e;
	cb->args[1] = h;
	cb->args[0] = t;

	return skb->len;
}

2412
#ifdef CONFIG_PROC_FS
L
Linus Torvalds 已提交
2413
/*
E
Eric Dumazet 已提交
2414 2415
 *	The /proc interfaces to multicast routing :
 *	/proc/net/ip_mr_cache & /proc/net/ip_mr_vif
L
Linus Torvalds 已提交
2416 2417
 */
struct ipmr_vif_iter {
2418
	struct seq_net_private p;
2419
	struct mr_table *mrt;
L
Linus Torvalds 已提交
2420 2421 2422
	int ct;
};

2423 2424
static struct vif_device *ipmr_vif_seq_idx(struct net *net,
					   struct ipmr_vif_iter *iter,
L
Linus Torvalds 已提交
2425 2426
					   loff_t pos)
{
2427
	struct mr_table *mrt = iter->mrt;
2428 2429 2430

	for (iter->ct = 0; iter->ct < mrt->maxvif; ++iter->ct) {
		if (!VIF_EXISTS(mrt, iter->ct))
L
Linus Torvalds 已提交
2431
			continue;
2432
		if (pos-- == 0)
2433
			return &mrt->vif_table[iter->ct];
L
Linus Torvalds 已提交
2434 2435 2436 2437 2438
	}
	return NULL;
}

static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos)
S
Stephen Hemminger 已提交
2439
	__acquires(mrt_lock)
L
Linus Torvalds 已提交
2440
{
2441
	struct ipmr_vif_iter *iter = seq->private;
2442
	struct net *net = seq_file_net(seq);
2443 2444 2445 2446 2447 2448 2449
	struct mr_table *mrt;

	mrt = ipmr_get_table(net, RT_TABLE_DEFAULT);
	if (mrt == NULL)
		return ERR_PTR(-ENOENT);

	iter->mrt = mrt;
2450

L
Linus Torvalds 已提交
2451
	read_lock(&mrt_lock);
2452
	return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1)
L
Linus Torvalds 已提交
2453 2454 2455 2456 2457 2458
		: SEQ_START_TOKEN;
}

static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	struct ipmr_vif_iter *iter = seq->private;
2459
	struct net *net = seq_file_net(seq);
2460
	struct mr_table *mrt = iter->mrt;
L
Linus Torvalds 已提交
2461 2462 2463

	++*pos;
	if (v == SEQ_START_TOKEN)
2464
		return ipmr_vif_seq_idx(net, iter, 0);
2465

2466 2467
	while (++iter->ct < mrt->maxvif) {
		if (!VIF_EXISTS(mrt, iter->ct))
L
Linus Torvalds 已提交
2468
			continue;
2469
		return &mrt->vif_table[iter->ct];
L
Linus Torvalds 已提交
2470 2471 2472 2473 2474
	}
	return NULL;
}

static void ipmr_vif_seq_stop(struct seq_file *seq, void *v)
S
Stephen Hemminger 已提交
2475
	__releases(mrt_lock)
L
Linus Torvalds 已提交
2476 2477 2478 2479 2480 2481
{
	read_unlock(&mrt_lock);
}

static int ipmr_vif_seq_show(struct seq_file *seq, void *v)
{
2482 2483
	struct ipmr_vif_iter *iter = seq->private;
	struct mr_table *mrt = iter->mrt;
2484

L
Linus Torvalds 已提交
2485
	if (v == SEQ_START_TOKEN) {
2486
		seq_puts(seq,
L
Linus Torvalds 已提交
2487 2488 2489 2490 2491 2492 2493
			 "Interface      BytesIn  PktsIn  BytesOut PktsOut Flags Local    Remote\n");
	} else {
		const struct vif_device *vif = v;
		const char *name =  vif->dev ? vif->dev->name : "none";

		seq_printf(seq,
			   "%2Zd %-10s %8ld %7ld  %8ld %7ld %05X %08X %08X\n",
2494
			   vif - mrt->vif_table,
2495
			   name, vif->bytes_in, vif->pkt_in,
L
Linus Torvalds 已提交
2496 2497 2498 2499 2500 2501
			   vif->bytes_out, vif->pkt_out,
			   vif->flags, vif->local, vif->remote);
	}
	return 0;
}

2502
static const struct seq_operations ipmr_vif_seq_ops = {
L
Linus Torvalds 已提交
2503 2504 2505 2506 2507 2508 2509 2510
	.start = ipmr_vif_seq_start,
	.next  = ipmr_vif_seq_next,
	.stop  = ipmr_vif_seq_stop,
	.show  = ipmr_vif_seq_show,
};

static int ipmr_vif_open(struct inode *inode, struct file *file)
{
2511 2512
	return seq_open_net(inode, file, &ipmr_vif_seq_ops,
			    sizeof(struct ipmr_vif_iter));
L
Linus Torvalds 已提交
2513 2514
}

2515
static const struct file_operations ipmr_vif_fops = {
L
Linus Torvalds 已提交
2516 2517 2518 2519
	.owner	 = THIS_MODULE,
	.open    = ipmr_vif_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
2520
	.release = seq_release_net,
L
Linus Torvalds 已提交
2521 2522 2523
};

struct ipmr_mfc_iter {
2524
	struct seq_net_private p;
2525
	struct mr_table *mrt;
2526
	struct list_head *cache;
L
Linus Torvalds 已提交
2527 2528 2529 2530
	int ct;
};


2531 2532
static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net,
					  struct ipmr_mfc_iter *it, loff_t pos)
L
Linus Torvalds 已提交
2533
{
2534
	struct mr_table *mrt = it->mrt;
L
Linus Torvalds 已提交
2535 2536
	struct mfc_cache *mfc;

2537
	rcu_read_lock();
2538
	for (it->ct = 0; it->ct < MFC_LINES; it->ct++) {
2539
		it->cache = &mrt->mfc_cache_array[it->ct];
2540
		list_for_each_entry_rcu(mfc, it->cache, list)
2541
			if (pos-- == 0)
L
Linus Torvalds 已提交
2542
				return mfc;
2543
	}
2544
	rcu_read_unlock();
L
Linus Torvalds 已提交
2545 2546

	spin_lock_bh(&mfc_unres_lock);
2547
	it->cache = &mrt->mfc_unres_queue;
2548
	list_for_each_entry(mfc, it->cache, list)
2549
		if (pos-- == 0)
L
Linus Torvalds 已提交
2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560
			return mfc;
	spin_unlock_bh(&mfc_unres_lock);

	it->cache = NULL;
	return NULL;
}


static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos)
{
	struct ipmr_mfc_iter *it = seq->private;
2561
	struct net *net = seq_file_net(seq);
2562
	struct mr_table *mrt;
2563

2564 2565 2566
	mrt = ipmr_get_table(net, RT_TABLE_DEFAULT);
	if (mrt == NULL)
		return ERR_PTR(-ENOENT);
2567

2568
	it->mrt = mrt;
L
Linus Torvalds 已提交
2569 2570
	it->cache = NULL;
	it->ct = 0;
2571
	return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1)
L
Linus Torvalds 已提交
2572 2573 2574 2575 2576 2577 2578
		: SEQ_START_TOKEN;
}

static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	struct mfc_cache *mfc = v;
	struct ipmr_mfc_iter *it = seq->private;
2579
	struct net *net = seq_file_net(seq);
2580
	struct mr_table *mrt = it->mrt;
L
Linus Torvalds 已提交
2581 2582 2583 2584

	++*pos;

	if (v == SEQ_START_TOKEN)
2585
		return ipmr_mfc_seq_idx(net, seq->private, 0);
L
Linus Torvalds 已提交
2586

2587 2588
	if (mfc->list.next != it->cache)
		return list_entry(mfc->list.next, struct mfc_cache, list);
2589

2590
	if (it->cache == &mrt->mfc_unres_queue)
L
Linus Torvalds 已提交
2591 2592
		goto end_of_list;

2593
	BUG_ON(it->cache != &mrt->mfc_cache_array[it->ct]);
L
Linus Torvalds 已提交
2594 2595

	while (++it->ct < MFC_LINES) {
2596
		it->cache = &mrt->mfc_cache_array[it->ct];
2597 2598 2599
		if (list_empty(it->cache))
			continue;
		return list_first_entry(it->cache, struct mfc_cache, list);
L
Linus Torvalds 已提交
2600 2601 2602
	}

	/* exhausted cache_array, show unresolved */
2603
	rcu_read_unlock();
2604
	it->cache = &mrt->mfc_unres_queue;
L
Linus Torvalds 已提交
2605
	it->ct = 0;
2606

L
Linus Torvalds 已提交
2607
	spin_lock_bh(&mfc_unres_lock);
2608 2609
	if (!list_empty(it->cache))
		return list_first_entry(it->cache, struct mfc_cache, list);
L
Linus Torvalds 已提交
2610

E
Eric Dumazet 已提交
2611
end_of_list:
L
Linus Torvalds 已提交
2612 2613 2614 2615 2616 2617 2618 2619 2620
	spin_unlock_bh(&mfc_unres_lock);
	it->cache = NULL;

	return NULL;
}

static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v)
{
	struct ipmr_mfc_iter *it = seq->private;
2621
	struct mr_table *mrt = it->mrt;
L
Linus Torvalds 已提交
2622

2623
	if (it->cache == &mrt->mfc_unres_queue)
L
Linus Torvalds 已提交
2624
		spin_unlock_bh(&mfc_unres_lock);
2625
	else if (it->cache == &mrt->mfc_cache_array[it->ct])
2626
		rcu_read_unlock();
L
Linus Torvalds 已提交
2627 2628 2629 2630 2631 2632 2633
}

static int ipmr_mfc_seq_show(struct seq_file *seq, void *v)
{
	int n;

	if (v == SEQ_START_TOKEN) {
2634
		seq_puts(seq,
L
Linus Torvalds 已提交
2635 2636 2637 2638
		 "Group    Origin   Iif     Pkts    Bytes    Wrong Oifs\n");
	} else {
		const struct mfc_cache *mfc = v;
		const struct ipmr_mfc_iter *it = seq->private;
2639
		const struct mr_table *mrt = it->mrt;
2640

2641 2642 2643
		seq_printf(seq, "%08X %08X %-3hd",
			   (__force u32) mfc->mfc_mcastgrp,
			   (__force u32) mfc->mfc_origin,
2644
			   mfc->mfc_parent);
L
Linus Torvalds 已提交
2645

2646
		if (it->cache != &mrt->mfc_unres_queue) {
2647 2648 2649 2650
			seq_printf(seq, " %8lu %8lu %8lu",
				   mfc->mfc_un.res.pkt,
				   mfc->mfc_un.res.bytes,
				   mfc->mfc_un.res.wrong_if);
S
Stephen Hemminger 已提交
2651
			for (n = mfc->mfc_un.res.minvif;
E
Eric Dumazet 已提交
2652
			     n < mfc->mfc_un.res.maxvif; n++) {
2653
				if (VIF_EXISTS(mrt, n) &&
2654 2655
				    mfc->mfc_un.res.ttls[n] < 255)
					seq_printf(seq,
2656
					   " %2d:%-3d",
L
Linus Torvalds 已提交
2657 2658
					   n, mfc->mfc_un.res.ttls[n]);
			}
2659 2660 2661 2662 2663
		} else {
			/* unresolved mfc_caches don't contain
			 * pkt, bytes and wrong_if values
			 */
			seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul);
L
Linus Torvalds 已提交
2664 2665 2666 2667 2668 2669
		}
		seq_putc(seq, '\n');
	}
	return 0;
}

2670
static const struct seq_operations ipmr_mfc_seq_ops = {
L
Linus Torvalds 已提交
2671 2672 2673 2674 2675 2676 2677 2678
	.start = ipmr_mfc_seq_start,
	.next  = ipmr_mfc_seq_next,
	.stop  = ipmr_mfc_seq_stop,
	.show  = ipmr_mfc_seq_show,
};

static int ipmr_mfc_open(struct inode *inode, struct file *file)
{
2679 2680
	return seq_open_net(inode, file, &ipmr_mfc_seq_ops,
			    sizeof(struct ipmr_mfc_iter));
L
Linus Torvalds 已提交
2681 2682
}

2683
static const struct file_operations ipmr_mfc_fops = {
L
Linus Torvalds 已提交
2684 2685 2686 2687
	.owner	 = THIS_MODULE,
	.open    = ipmr_mfc_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
2688
	.release = seq_release_net,
L
Linus Torvalds 已提交
2689
};
2690
#endif
L
Linus Torvalds 已提交
2691 2692

#ifdef CONFIG_IP_PIMSM_V2
2693
static const struct net_protocol pim_protocol = {
L
Linus Torvalds 已提交
2694
	.handler	=	pim_rcv,
T
Tom Goff 已提交
2695
	.netns_ok	=	1,
L
Linus Torvalds 已提交
2696 2697 2698 2699 2700 2701 2702
};
#endif


/*
 *	Setup for IP multicast routing
 */
2703 2704
static int __net_init ipmr_net_init(struct net *net)
{
2705
	int err;
2706

2707 2708
	err = ipmr_rules_init(net);
	if (err < 0)
2709
		goto fail;
2710 2711 2712

#ifdef CONFIG_PROC_FS
	err = -ENOMEM;
2713
	if (!proc_create("ip_mr_vif", 0, net->proc_net, &ipmr_vif_fops))
2714
		goto proc_vif_fail;
2715
	if (!proc_create("ip_mr_cache", 0, net->proc_net, &ipmr_mfc_fops))
2716 2717
		goto proc_cache_fail;
#endif
2718 2719
	return 0;

2720 2721
#ifdef CONFIG_PROC_FS
proc_cache_fail:
2722
	remove_proc_entry("ip_mr_vif", net->proc_net);
2723
proc_vif_fail:
2724
	ipmr_rules_exit(net);
2725
#endif
2726 2727 2728 2729 2730 2731
fail:
	return err;
}

static void __net_exit ipmr_net_exit(struct net *net)
{
2732
#ifdef CONFIG_PROC_FS
2733 2734
	remove_proc_entry("ip_mr_cache", net->proc_net);
	remove_proc_entry("ip_mr_vif", net->proc_net);
2735
#endif
2736
	ipmr_rules_exit(net);
2737 2738 2739 2740 2741 2742
}

static struct pernet_operations ipmr_net_ops = {
	.init = ipmr_net_init,
	.exit = ipmr_net_exit,
};
2743

W
Wang Chen 已提交
2744
int __init ip_mr_init(void)
L
Linus Torvalds 已提交
2745
{
W
Wang Chen 已提交
2746 2747
	int err;

L
Linus Torvalds 已提交
2748 2749
	mrt_cachep = kmem_cache_create("ip_mrt_cache",
				       sizeof(struct mfc_cache),
2750
				       0, SLAB_HWCACHE_ALIGN | SLAB_PANIC,
2751
				       NULL);
W
Wang Chen 已提交
2752 2753 2754
	if (!mrt_cachep)
		return -ENOMEM;

2755 2756 2757 2758
	err = register_pernet_subsys(&ipmr_net_ops);
	if (err)
		goto reg_pernet_fail;

W
Wang Chen 已提交
2759 2760 2761
	err = register_netdevice_notifier(&ip_mr_notifier);
	if (err)
		goto reg_notif_fail;
T
Tom Goff 已提交
2762 2763
#ifdef CONFIG_IP_PIMSM_V2
	if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) {
J
Joe Perches 已提交
2764
		pr_err("%s: can't add PIM protocol\n", __func__);
T
Tom Goff 已提交
2765 2766 2767 2768
		err = -EAGAIN;
		goto add_proto_fail;
	}
#endif
2769 2770
	rtnl_register(RTNL_FAMILY_IPMR, RTM_GETROUTE,
		      NULL, ipmr_rtm_dumproute, NULL);
W
Wang Chen 已提交
2771
	return 0;
2772

T
Tom Goff 已提交
2773 2774 2775 2776
#ifdef CONFIG_IP_PIMSM_V2
add_proto_fail:
	unregister_netdevice_notifier(&ip_mr_notifier);
#endif
B
Benjamin Thery 已提交
2777
reg_notif_fail:
2778 2779
	unregister_pernet_subsys(&ipmr_net_ops);
reg_pernet_fail:
B
Benjamin Thery 已提交
2780
	kmem_cache_destroy(mrt_cachep);
W
Wang Chen 已提交
2781
	return err;
L
Linus Torvalds 已提交
2782
}