[linux-2.6-block.git] / net / sched / cls_flow.c

/*
 * net/sched/cls_flow.c		Generic flow classifier
 *
 * Copyright (c) 2007, 2008 Patrick McHardy <kaber@trash.net>
 *
 * 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.
 */

#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/list.h>
#include <linux/jhash.h>
#include <linux/random.h>
#include <linux/pkt_cls.h>
#include <linux/skbuff.h>
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
#include <linux/if_vlan.h>
#include <linux/slab.h>
#include <linux/module.h>

#include <net/pkt_cls.h>
#include <net/ip.h>
#include <net/route.h>
#include <net/flow_keys.h>

#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
#include <net/netfilter/nf_conntrack.h>
#endif

struct flow_head {
	struct list_head	filters;
};

struct flow_filter {
	struct list_head	list;
	struct tcf_exts		exts;
	struct tcf_ematch_tree	ematches;
	struct timer_list	perturb_timer;
	u32			perturb_period;
	u32			handle;

	u32			nkeys;
	u32			keymask;
	u32			mode;
	u32			mask;
	u32			xor;
	u32			rshift;
	u32			addend;
	u32			divisor;
	u32			baseclass;
	u32			hashrnd;
};

static inline u32 addr_fold(void *addr)
{
	unsigned long a = (unsigned long)addr;

	return (a & 0xFFFFFFFF) ^ (BITS_PER_LONG > 32 ? a >> 32 : 0);
}

static u32 flow_get_src(const struct sk_buff *skb, const struct flow_keys *flow)
{
	if (flow->src)
		return ntohl(flow->src);
	return addr_fold(skb->sk);
}

static u32 flow_get_dst(const struct sk_buff *skb, const struct flow_keys *flow)
{
	if (flow->dst)
		return ntohl(flow->dst);
	return addr_fold(skb_dst(skb)) ^ (__force u16)skb->protocol;
}

static u32 flow_get_proto(const struct sk_buff *skb, const struct flow_keys *flow)
{
	return flow->ip_proto;
}

static u32 flow_get_proto_src(const struct sk_buff *skb, const struct flow_keys *flow)
{
	if (flow->ports)
		return ntohs(flow->port16[0]);

	return addr_fold(skb->sk);
}

static u32 flow_get_proto_dst(const struct sk_buff *skb, const struct flow_keys *flow)
{
	if (flow->ports)
		return ntohs(flow->port16[1]);

	return addr_fold(skb_dst(skb)) ^ (__force u16)skb->protocol;
}

static u32 flow_get_iif(const struct sk_buff *skb)
{
	return skb->skb_iif;
}

static u32 flow_get_priority(const struct sk_buff *skb)
{
	return skb->priority;
}

static u32 flow_get_mark(const struct sk_buff *skb)
{
	return skb->mark;
}

static u32 flow_get_nfct(const struct sk_buff *skb)
{
#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
	return addr_fold(skb->nfct);
#else
	return 0;
#endif
}

#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
#define CTTUPLE(skb, member)						\
({									\
	enum ip_conntrack_info ctinfo;					\
	const struct nf_conn *ct = nf_ct_get(skb, &ctinfo);		\
	if (ct == NULL)							\
		goto fallback;						\
	ct->tuplehash[CTINFO2DIR(ctinfo)].tuple.member;			\
})
#else
#define CTTUPLE(skb, member)						\
({									\
	goto fallback;							\
	0;								\
})
#endif

static u32 flow_get_nfct_src(const struct sk_buff *skb, const struct flow_keys *flow)
{
	switch (skb->protocol) {
	case htons(ETH_P_IP):
		return ntohl(CTTUPLE(skb, src.u3.ip));
	case htons(ETH_P_IPV6):
		return ntohl(CTTUPLE(skb, src.u3.ip6[3]));
	}
fallback:
	return flow_get_src(skb, flow);
}

static u32 flow_get_nfct_dst(const struct sk_buff *skb, const struct flow_keys *flow)
{
	switch (skb->protocol) {
	case htons(ETH_P_IP):
		return ntohl(CTTUPLE(skb, dst.u3.ip));
	case htons(ETH_P_IPV6):
		return ntohl(CTTUPLE(skb, dst.u3.ip6[3]));
	}
fallback:
	return flow_get_dst(skb, flow);
}

static u32 flow_get_nfct_proto_src(const struct sk_buff *skb, const struct flow_keys *flow)
{
	return ntohs(CTTUPLE(skb, src.u.all));
fallback:
	return flow_get_proto_src(skb, flow);
}

static u32 flow_get_nfct_proto_dst(const struct sk_buff *skb, const struct flow_keys *flow)
{
	return ntohs(CTTUPLE(skb, dst.u.all));
fallback:
	return flow_get_proto_dst(skb, flow);
}

static u32 flow_get_rtclassid(const struct sk_buff *skb)
{
#ifdef CONFIG_IP_ROUTE_CLASSID
	if (skb_dst(skb))
		return skb_dst(skb)->tclassid;
#endif
	return 0;
}

static u32 flow_get_skuid(const struct sk_buff *skb)
{
	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file) {
		kuid_t skuid = skb->sk->sk_socket->file->f_cred->fsuid;
		return from_kuid(&init_user_ns, skuid);
	}
	return 0;
}

static u32 flow_get_skgid(const struct sk_buff *skb)
{
	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file) {
		kgid_t skgid = skb->sk->sk_socket->file->f_cred->fsgid;
		return from_kgid(&init_user_ns, skgid);
	}
	return 0;
}

static u32 flow_get_vlan_tag(const struct sk_buff *skb)
{
	u16 uninitialized_var(tag);

	if (vlan_get_tag(skb, &tag) < 0)
		return 0;
	return tag & VLAN_VID_MASK;
}

static u32 flow_get_rxhash(struct sk_buff *skb)
{
	return skb_get_hash(skb);
}

static u32 flow_key_get(struct sk_buff *skb, int key, struct flow_keys *flow)
{
	switch (key) {
	case FLOW_KEY_SRC:
		return flow_get_src(skb, flow);
	case FLOW_KEY_DST:
		return flow_get_dst(skb, flow);
	case FLOW_KEY_PROTO:
		return flow_get_proto(skb, flow);
	case FLOW_KEY_PROTO_SRC:
		return flow_get_proto_src(skb, flow);
	case FLOW_KEY_PROTO_DST:
		return flow_get_proto_dst(skb, flow);
	case FLOW_KEY_IIF:
		return flow_get_iif(skb);
	case FLOW_KEY_PRIORITY:
		return flow_get_priority(skb);
	case FLOW_KEY_MARK:
		return flow_get_mark(skb);
	case FLOW_KEY_NFCT:
		return flow_get_nfct(skb);
	case FLOW_KEY_NFCT_SRC:
		return flow_get_nfct_src(skb, flow);
	case FLOW_KEY_NFCT_DST:
		return flow_get_nfct_dst(skb, flow);
	case FLOW_KEY_NFCT_PROTO_SRC:
		return flow_get_nfct_proto_src(skb, flow);
	case FLOW_KEY_NFCT_PROTO_DST:
		return flow_get_nfct_proto_dst(skb, flow);
	case FLOW_KEY_RTCLASSID:
		return flow_get_rtclassid(skb);
	case FLOW_KEY_SKUID:
		return flow_get_skuid(skb);
	case FLOW_KEY_SKGID:
		return flow_get_skgid(skb);
	case FLOW_KEY_VLAN_TAG:
		return flow_get_vlan_tag(skb);
	case FLOW_KEY_RXHASH:
		return flow_get_rxhash(skb);
	default:
		WARN_ON(1);
		return 0;
	}
}

#define FLOW_KEYS_NEEDED ((1 << FLOW_KEY_SRC) | 		\
			  (1 << FLOW_KEY_DST) |			\
			  (1 << FLOW_KEY_PROTO) |		\
			  (1 << FLOW_KEY_PROTO_SRC) |		\
			  (1 << FLOW_KEY_PROTO_DST) | 		\
			  (1 << FLOW_KEY_NFCT_SRC) |		\
			  (1 << FLOW_KEY_NFCT_DST) |		\
			  (1 << FLOW_KEY_NFCT_PROTO_SRC) |	\
			  (1 << FLOW_KEY_NFCT_PROTO_DST))

static int flow_classify(struct sk_buff *skb, const struct tcf_proto *tp,
			 struct tcf_result *res)
{
	struct flow_head *head = tp->root;
	struct flow_filter *f;
	u32 keymask;
	u32 classid;
	unsigned int n, key;
	int r;

	list_for_each_entry(f, &head->filters, list) {
		u32 keys[FLOW_KEY_MAX + 1];
		struct flow_keys flow_keys;

		if (!tcf_em_tree_match(skb, &f->ematches, NULL))
			continue;

		keymask = f->keymask;
		if (keymask & FLOW_KEYS_NEEDED)
			skb_flow_dissect(skb, &flow_keys);

		for (n = 0; n < f->nkeys; n++) {
			key = ffs(keymask) - 1;
			keymask &= ~(1 << key);
			keys[n] = flow_key_get(skb, key, &flow_keys);
		}

		if (f->mode == FLOW_MODE_HASH)
			classid = jhash2(keys, f->nkeys, f->hashrnd);
		else {
			classid = keys[0];
			classid = (classid & f->mask) ^ f->xor;
			classid = (classid >> f->rshift) + f->addend;
		}

		if (f->divisor)
			classid %= f->divisor;

		res->class   = 0;
		res->classid = TC_H_MAKE(f->baseclass, f->baseclass + classid);

		r = tcf_exts_exec(skb, &f->exts, res);
		if (r < 0)
			continue;
		return r;
	}
	return -1;
}

static void flow_perturbation(unsigned long arg)
{
	struct flow_filter *f = (struct flow_filter *)arg;

	get_random_bytes(&f->hashrnd, 4);
	if (f->perturb_period)
		mod_timer(&f->perturb_timer, jiffies + f->perturb_period);
}

static const struct nla_policy flow_policy[TCA_FLOW_MAX + 1] = {
	[TCA_FLOW_KEYS]		= { .type = NLA_U32 },
	[TCA_FLOW_MODE]		= { .type = NLA_U32 },
	[TCA_FLOW_BASECLASS]	= { .type = NLA_U32 },
	[TCA_FLOW_RSHIFT]	= { .type = NLA_U32 },
	[TCA_FLOW_ADDEND]	= { .type = NLA_U32 },
	[TCA_FLOW_MASK]		= { .type = NLA_U32 },
	[TCA_FLOW_XOR]		= { .type = NLA_U32 },
	[TCA_FLOW_DIVISOR]	= { .type = NLA_U32 },
	[TCA_FLOW_ACT]		= { .type = NLA_NESTED },
	[TCA_FLOW_POLICE]	= { .type = NLA_NESTED },
	[TCA_FLOW_EMATCHES]	= { .type = NLA_NESTED },
	[TCA_FLOW_PERTURB]	= { .type = NLA_U32 },
};

static int flow_change(struct net *net, struct sk_buff *in_skb,
		       struct tcf_proto *tp, unsigned long base,
		       u32 handle, struct nlattr **tca,
		       unsigned long *arg)
{
	struct flow_head *head = tp->root;
	struct flow_filter *f;
	struct nlattr *opt = tca[TCA_OPTIONS];
	struct nlattr *tb[TCA_FLOW_MAX + 1];
	struct tcf_exts e;
	struct tcf_ematch_tree t;
	unsigned int nkeys = 0;
	unsigned int perturb_period = 0;
	u32 baseclass = 0;
	u32 keymask = 0;
	u32 mode;
	int err;

	if (opt == NULL)
		return -EINVAL;

	err = nla_parse_nested(tb, TCA_FLOW_MAX, opt, flow_policy);
	if (err < 0)
		return err;

	if (tb[TCA_FLOW_BASECLASS]) {
		baseclass = nla_get_u32(tb[TCA_FLOW_BASECLASS]);
		if (TC_H_MIN(baseclass) == 0)
			return -EINVAL;
	}

	if (tb[TCA_FLOW_KEYS]) {
		keymask = nla_get_u32(tb[TCA_FLOW_KEYS]);

		nkeys = hweight32(keymask);
		if (nkeys == 0)
			return -EINVAL;

		if (fls(keymask) - 1 > FLOW_KEY_MAX)
			return -EOPNOTSUPP;

		if ((keymask & (FLOW_KEY_SKUID|FLOW_KEY_SKGID)) &&
		    sk_user_ns(NETLINK_CB(in_skb).sk) != &init_user_ns)
			return -EOPNOTSUPP;
	}

	tcf_exts_init(&e, TCA_FLOW_ACT, TCA_FLOW_POLICE);
	err = tcf_exts_validate(net, tp, tb, tca[TCA_RATE], &e);
	if (err < 0)
		return err;

	err = tcf_em_tree_validate(tp, tb[TCA_FLOW_EMATCHES], &t);
	if (err < 0)
		goto err1;

	f = (struct flow_filter *)*arg;
	if (f != NULL) {
		err = -EINVAL;
		if (f->handle != handle && handle)
			goto err2;

		mode = f->mode;
		if (tb[TCA_FLOW_MODE])
			mode = nla_get_u32(tb[TCA_FLOW_MODE]);
		if (mode != FLOW_MODE_HASH && nkeys > 1)
			goto err2;

		if (mode == FLOW_MODE_HASH)
			perturb_period = f->perturb_period;
		if (tb[TCA_FLOW_PERTURB]) {
			if (mode != FLOW_MODE_HASH)
				goto err2;
			perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
		}
	} else {
		err = -EINVAL;
		if (!handle)
			goto err2;
		if (!tb[TCA_FLOW_KEYS])
			goto err2;

		mode = FLOW_MODE_MAP;
		if (tb[TCA_FLOW_MODE])
			mode = nla_get_u32(tb[TCA_FLOW_MODE]);
		if (mode != FLOW_MODE_HASH && nkeys > 1)
			goto err2;

		if (tb[TCA_FLOW_PERTURB]) {
			if (mode != FLOW_MODE_HASH)
				goto err2;
			perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
		}

		if (TC_H_MAJ(baseclass) == 0)
			baseclass = TC_H_MAKE(tp->q->handle, baseclass);
		if (TC_H_MIN(baseclass) == 0)
			baseclass = TC_H_MAKE(baseclass, 1);

		err = -ENOBUFS;
		f = kzalloc(sizeof(*f), GFP_KERNEL);
		if (f == NULL)
			goto err2;

		f->handle = handle;
		f->mask	  = ~0U;
		tcf_exts_init(&f->exts, TCA_FLOW_ACT, TCA_FLOW_POLICE);

		get_random_bytes(&f->hashrnd, 4);
		f->perturb_timer.function = flow_perturbation;
		f->perturb_timer.data = (unsigned long)f;
		init_timer_deferrable(&f->perturb_timer);
	}

	tcf_exts_change(tp, &f->exts, &e);
	tcf_em_tree_change(tp, &f->ematches, &t);

	tcf_tree_lock(tp);

	if (tb[TCA_FLOW_KEYS]) {
		f->keymask = keymask;
		f->nkeys   = nkeys;
	}

	f->mode = mode;

	if (tb[TCA_FLOW_MASK])
		f->mask = nla_get_u32(tb[TCA_FLOW_MASK]);
	if (tb[TCA_FLOW_XOR])
		f->xor = nla_get_u32(tb[TCA_FLOW_XOR]);
	if (tb[TCA_FLOW_RSHIFT])
		f->rshift = nla_get_u32(tb[TCA_FLOW_RSHIFT]);
	if (tb[TCA_FLOW_ADDEND])
		f->addend = nla_get_u32(tb[TCA_FLOW_ADDEND]);

	if (tb[TCA_FLOW_DIVISOR])
		f->divisor = nla_get_u32(tb[TCA_FLOW_DIVISOR]);
	if (baseclass)
		f->baseclass = baseclass;

	f->perturb_period = perturb_period;
	del_timer(&f->perturb_timer);
	if (perturb_period)
		mod_timer(&f->perturb_timer, jiffies + perturb_period);

	if (*arg == 0)
		list_add_tail(&f->list, &head->filters);

	tcf_tree_unlock(tp);

	*arg = (unsigned long)f;
	return 0;

err2:
	tcf_em_tree_destroy(tp, &t);
err1:
	tcf_exts_destroy(tp, &e);
	return err;
}

static void flow_destroy_filter(struct tcf_proto *tp, struct flow_filter *f)
{
	del_timer_sync(&f->perturb_timer);
	tcf_exts_destroy(tp, &f->exts);
	tcf_em_tree_destroy(tp, &f->ematches);
	kfree(f);
}

static int flow_delete(struct tcf_proto *tp, unsigned long arg)
{
	struct flow_filter *f = (struct flow_filter *)arg;

	tcf_tree_lock(tp);
	list_del(&f->list);
	tcf_tree_unlock(tp);
	flow_destroy_filter(tp, f);
	return 0;
}

static int flow_init(struct tcf_proto *tp)
{
	struct flow_head *head;

	head = kzalloc(sizeof(*head), GFP_KERNEL);
	if (head == NULL)
		return -ENOBUFS;
	INIT_LIST_HEAD(&head->filters);
	tp->root = head;
	return 0;
}

static void flow_destroy(struct tcf_proto *tp)
{
	struct flow_head *head = tp->root;
	struct flow_filter *f, *next;

	list_for_each_entry_safe(f, next, &head->filters, list) {
		list_del(&f->list);
		flow_destroy_filter(tp, f);
	}
	kfree(head);
}

static unsigned long flow_get(struct tcf_proto *tp, u32 handle)
{
	struct flow_head *head = tp->root;
	struct flow_filter *f;

	list_for_each_entry(f, &head->filters, list)
		if (f->handle == handle)
			return (unsigned long)f;
	return 0;
}

static void flow_put(struct tcf_proto *tp, unsigned long f)
{
}

static int flow_dump(struct tcf_proto *tp, unsigned long fh,
		     struct sk_buff *skb, struct tcmsg *t)
{
	struct flow_filter *f = (struct flow_filter *)fh;
	struct nlattr *nest;

	if (f == NULL)
		return skb->len;

	t->tcm_handle = f->handle;

	nest = nla_nest_start(skb, TCA_OPTIONS);
	if (nest == NULL)
		goto nla_put_failure;

	if (nla_put_u32(skb, TCA_FLOW_KEYS, f->keymask) ||
	    nla_put_u32(skb, TCA_FLOW_MODE, f->mode))
		goto nla_put_failure;

	if (f->mask != ~0 || f->xor != 0) {
		if (nla_put_u32(skb, TCA_FLOW_MASK, f->mask) ||
		    nla_put_u32(skb, TCA_FLOW_XOR, f->xor))
			goto nla_put_failure;
	}
	if (f->rshift &&
	    nla_put_u32(skb, TCA_FLOW_RSHIFT, f->rshift))
		goto nla_put_failure;
	if (f->addend &&
	    nla_put_u32(skb, TCA_FLOW_ADDEND, f->addend))
		goto nla_put_failure;

	if (f->divisor &&
	    nla_put_u32(skb, TCA_FLOW_DIVISOR, f->divisor))
		goto nla_put_failure;
	if (f->baseclass &&
	    nla_put_u32(skb, TCA_FLOW_BASECLASS, f->baseclass))
		goto nla_put_failure;

	if (f->perturb_period &&
	    nla_put_u32(skb, TCA_FLOW_PERTURB, f->perturb_period / HZ))
		goto nla_put_failure;

	if (tcf_exts_dump(skb, &f->exts) < 0)
		goto nla_put_failure;
#ifdef CONFIG_NET_EMATCH
	if (f->ematches.hdr.nmatches &&
	    tcf_em_tree_dump(skb, &f->ematches, TCA_FLOW_EMATCHES) < 0)
		goto nla_put_failure;
#endif
	nla_nest_end(skb, nest);

	if (tcf_exts_dump_stats(skb, &f->exts) < 0)
		goto nla_put_failure;

	return skb->len;

nla_put_failure:
	nlmsg_trim(skb, nest);
	return -1;
}

static void flow_walk(struct tcf_proto *tp, struct tcf_walker *arg)
{
	struct flow_head *head = tp->root;
	struct flow_filter *f;

	list_for_each_entry(f, &head->filters, list) {
		if (arg->count < arg->skip)
			goto skip;
		if (arg->fn(tp, (unsigned long)f, arg) < 0) {
			arg->stop = 1;
			break;
		}
skip:
		arg->count++;
	}
}

static struct tcf_proto_ops cls_flow_ops __read_mostly = {
	.kind		= "flow",
	.classify	= flow_classify,
	.init		= flow_init,
	.destroy	= flow_destroy,
	.change		= flow_change,
	.delete		= flow_delete,
	.get		= flow_get,
	.put		= flow_put,
	.dump		= flow_dump,
	.walk		= flow_walk,
	.owner		= THIS_MODULE,
};

static int __init cls_flow_init(void)
{
	return register_tcf_proto_ops(&cls_flow_ops);
}

static void __exit cls_flow_exit(void)
{
	unregister_tcf_proto_ops(&cls_flow_ops);
}

module_init(cls_flow_init);
module_exit(cls_flow_exit);

MODULE_LICENSE("GPL");
MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
MODULE_DESCRIPTION("TC flow classifier");
Commit	Line	Data
e5dfb815 PM	1	/*
	2	* net/sched/cls_flow.c Generic flow classifier
	3	*
	4	* Copyright (c) 2007, 2008 Patrick McHardy <kaber@trash.net>
	5	*
	6	* This program is free software; you can redistribute it and/or
	7	* modify it under the terms of the GNU General Public License
	8	* as published by the Free Software Foundation; either version 2
	9	* of the License, or (at your option) any later version.
	10	*/
	11
	12	#include <linux/kernel.h>
	13	#include <linux/init.h>
	14	#include <linux/list.h>
	15	#include <linux/jhash.h>
	16	#include <linux/random.h>
	17	#include <linux/pkt_cls.h>
	18	#include <linux/skbuff.h>
	19	#include <linux/in.h>
	20	#include <linux/ip.h>
	21	#include <linux/ipv6.h>
9ec13810	22	#include <linux/if_vlan.h>
5a0e3ad6	23	#include <linux/slab.h>
3a9a231d	24	#include <linux/module.h>
e5dfb815 PM	25
	26	#include <net/pkt_cls.h>
	27	#include <net/ip.h>
	28	#include <net/route.h>
6bd2a9af ED	29	#include <net/flow_keys.h>
6bd2a9af ED	30
e5dfb815 PM	31	#if defined(CONFIG_NF_CONNTRACK) \|\| defined(CONFIG_NF_CONNTRACK_MODULE)
	32	#include <net/netfilter/nf_conntrack.h>
	33	#endif
	34
	35	struct flow_head {
	36	struct list_head filters;
	37	};
	38
	39	struct flow_filter {
	40	struct list_head list;
	41	struct tcf_exts exts;
	42	struct tcf_ematch_tree ematches;
72d9794f PM	43	struct timer_list perturb_timer;
72d9794f PM	44	u32 perturb_period;
e5dfb815 PM	45	u32 handle;
	46
	47	u32 nkeys;
	48	u32 keymask;
	49	u32 mode;
	50	u32 mask;
	51	u32 xor;
	52	u32 rshift;
	53	u32 addend;
	54	u32 divisor;
	55	u32 baseclass;
72d9794f	56	u32 hashrnd;
e5dfb815 PM	57	};
e5dfb815 PM	58
e5dfb815 PM	59	static inline u32 addr_fold(void *addr)
	60	{
	61	unsigned long a = (unsigned long)addr;
	62
	63	return (a & 0xFFFFFFFF) ^ (BITS_PER_LONG > 32 ? a >> 32 : 0);
	64	}
	65
6bd2a9af	66	static u32 flow_get_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	67	{
6bd2a9af ED	68	if (flow->src)
6bd2a9af ED	69	return ntohl(flow->src);
4b95c3d4	70	return addr_fold(skb->sk);
e5dfb815 PM	71	}
e5dfb815 PM	72
6bd2a9af	73	static u32 flow_get_dst(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	74	{
6bd2a9af ED	75	if (flow->dst)
6bd2a9af ED	76	return ntohl(flow->dst);
4b95c3d4	77	return addr_fold(skb_dst(skb)) ^ (__force u16)skb->protocol;
e5dfb815 PM	78	}
e5dfb815 PM	79
6bd2a9af	80	static u32 flow_get_proto(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	81	{
6bd2a9af	82	return flow->ip_proto;
e5dfb815 PM	83	}
e5dfb815 PM	84
6bd2a9af	85	static u32 flow_get_proto_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	86	{
6bd2a9af ED	87	if (flow->ports)
6bd2a9af ED	88	return ntohs(flow->port16[0]);
e5dfb815	89
859c2012 ED	90	return addr_fold(skb->sk);
	91	}
	92
6bd2a9af	93	static u32 flow_get_proto_dst(const struct sk_buff skb, const struct flow_keys flow)
859c2012	94	{
6bd2a9af ED	95	if (flow->ports)
6bd2a9af ED	96	return ntohs(flow->port16[1]);
e5dfb815	97
4b95c3d4	98	return addr_fold(skb_dst(skb)) ^ (__force u16)skb->protocol;
e5dfb815 PM	99	}
	100
	101	static u32 flow_get_iif(const struct sk_buff *skb)
	102	{
8964be4a	103	return skb->skb_iif;
e5dfb815 PM	104	}
	105
	106	static u32 flow_get_priority(const struct sk_buff *skb)
	107	{
	108	return skb->priority;
	109	}
	110
	111	static u32 flow_get_mark(const struct sk_buff *skb)
	112	{
	113	return skb->mark;
	114	}
	115
	116	static u32 flow_get_nfct(const struct sk_buff *skb)
	117	{
	118	#if defined(CONFIG_NF_CONNTRACK) \|\| defined(CONFIG_NF_CONNTRACK_MODULE)
	119	return addr_fold(skb->nfct);
	120	#else
	121	return 0;
	122	#endif
	123	}
	124
	125	#if defined(CONFIG_NF_CONNTRACK) \|\| defined(CONFIG_NF_CONNTRACK_MODULE)
	126	#define CTTUPLE(skb, member) \
	127	({ \
	128	enum ip_conntrack_info ctinfo; \
859c2012	129	const struct nf_conn *ct = nf_ct_get(skb, &ctinfo); \
e5dfb815 PM	130	if (ct == NULL) \
	131	goto fallback; \
	132	ct->tuplehash[CTINFO2DIR(ctinfo)].tuple.member; \
	133	})
	134	#else
	135	#define CTTUPLE(skb, member) \
	136	({ \
	137	goto fallback; \
	138	0; \
	139	})
	140	#endif
	141
6bd2a9af	142	static u32 flow_get_nfct_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815 PM	143	{
e5dfb815 PM	144	switch (skb->protocol) {
60678040	145	case htons(ETH_P_IP):
e5dfb815	146	return ntohl(CTTUPLE(skb, src.u3.ip));
60678040	147	case htons(ETH_P_IPV6):
e5dfb815 PM	148	return ntohl(CTTUPLE(skb, src.u3.ip6[3]));
	149	}
	150	fallback:
6bd2a9af	151	return flow_get_src(skb, flow);
e5dfb815 PM	152	}
e5dfb815 PM	153
6bd2a9af	154	static u32 flow_get_nfct_dst(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815 PM	155	{
e5dfb815 PM	156	switch (skb->protocol) {
60678040	157	case htons(ETH_P_IP):
e5dfb815	158	return ntohl(CTTUPLE(skb, dst.u3.ip));
60678040	159	case htons(ETH_P_IPV6):
e5dfb815 PM	160	return ntohl(CTTUPLE(skb, dst.u3.ip6[3]));
	161	}
	162	fallback:
6bd2a9af	163	return flow_get_dst(skb, flow);
e5dfb815 PM	164	}
e5dfb815 PM	165
6bd2a9af	166	static u32 flow_get_nfct_proto_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815 PM	167	{
	168	return ntohs(CTTUPLE(skb, src.u.all));
	169	fallback:
6bd2a9af	170	return flow_get_proto_src(skb, flow);
e5dfb815 PM	171	}
e5dfb815 PM	172
6bd2a9af	173	static u32 flow_get_nfct_proto_dst(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815 PM	174	{
	175	return ntohs(CTTUPLE(skb, dst.u.all));
	176	fallback:
6bd2a9af	177	return flow_get_proto_dst(skb, flow);
e5dfb815 PM	178	}
	179
	180	static u32 flow_get_rtclassid(const struct sk_buff *skb)
	181	{
c7066f70	182	#ifdef CONFIG_IP_ROUTE_CLASSID
adf30907 ED	183	if (skb_dst(skb))
adf30907 ED	184	return skb_dst(skb)->tclassid;
e5dfb815 PM	185	#endif
	186	return 0;
	187	}
	188
	189	static u32 flow_get_skuid(const struct sk_buff *skb)
	190	{
a6c6796c EB	191	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file) {
	192	kuid_t skuid = skb->sk->sk_socket->file->f_cred->fsuid;
	193	return from_kuid(&init_user_ns, skuid);
	194	}
e5dfb815 PM	195	return 0;
	196	}
	197
	198	static u32 flow_get_skgid(const struct sk_buff *skb)
	199	{
a6c6796c EB	200	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file) {
	201	kgid_t skgid = skb->sk->sk_socket->file->f_cred->fsgid;
	202	return from_kgid(&init_user_ns, skgid);
	203	}
e5dfb815 PM	204	return 0;
	205	}
	206
9ec13810 PM	207	static u32 flow_get_vlan_tag(const struct sk_buff *skb)
	208	{
	209	u16 uninitialized_var(tag);
	210
	211	if (vlan_get_tag(skb, &tag) < 0)
	212	return 0;
	213	return tag & VLAN_VID_MASK;
	214	}
	215
739a91ef CG	216	static u32 flow_get_rxhash(struct sk_buff *skb)
739a91ef CG	217	{
3958afa1	218	return skb_get_hash(skb);
739a91ef CG	219	}
739a91ef CG	220
6bd2a9af	221	static u32 flow_key_get(struct sk_buff skb, int key, struct flow_keys flow)
e5dfb815 PM	222	{
	223	switch (key) {
	224	case FLOW_KEY_SRC:
6bd2a9af	225	return flow_get_src(skb, flow);
e5dfb815	226	case FLOW_KEY_DST:
6bd2a9af	227	return flow_get_dst(skb, flow);
e5dfb815	228	case FLOW_KEY_PROTO:
6bd2a9af	229	return flow_get_proto(skb, flow);
e5dfb815	230	case FLOW_KEY_PROTO_SRC:
6bd2a9af	231	return flow_get_proto_src(skb, flow);
e5dfb815	232	case FLOW_KEY_PROTO_DST:
6bd2a9af	233	return flow_get_proto_dst(skb, flow);
e5dfb815 PM	234	case FLOW_KEY_IIF:
	235	return flow_get_iif(skb);
	236	case FLOW_KEY_PRIORITY:
	237	return flow_get_priority(skb);
	238	case FLOW_KEY_MARK:
	239	return flow_get_mark(skb);
	240	case FLOW_KEY_NFCT:
	241	return flow_get_nfct(skb);
	242	case FLOW_KEY_NFCT_SRC:
6bd2a9af	243	return flow_get_nfct_src(skb, flow);
e5dfb815	244	case FLOW_KEY_NFCT_DST:
6bd2a9af	245	return flow_get_nfct_dst(skb, flow);
e5dfb815	246	case FLOW_KEY_NFCT_PROTO_SRC:
6bd2a9af	247	return flow_get_nfct_proto_src(skb, flow);
e5dfb815	248	case FLOW_KEY_NFCT_PROTO_DST:
6bd2a9af	249	return flow_get_nfct_proto_dst(skb, flow);
e5dfb815 PM	250	case FLOW_KEY_RTCLASSID:
	251	return flow_get_rtclassid(skb);
	252	case FLOW_KEY_SKUID:
	253	return flow_get_skuid(skb);
	254	case FLOW_KEY_SKGID:
	255	return flow_get_skgid(skb);
9ec13810 PM	256	case FLOW_KEY_VLAN_TAG:
9ec13810 PM	257	return flow_get_vlan_tag(skb);
739a91ef CG	258	case FLOW_KEY_RXHASH:
739a91ef CG	259	return flow_get_rxhash(skb);
e5dfb815 PM	260	default:
	261	WARN_ON(1);
	262	return 0;
	263	}
	264	}
	265
6bd2a9af ED	266	#define FLOW_KEYS_NEEDED ((1 << FLOW_KEY_SRC) \| \
	267	(1 << FLOW_KEY_DST) \| \
	268	(1 << FLOW_KEY_PROTO) \| \
	269	(1 << FLOW_KEY_PROTO_SRC) \| \
	270	(1 << FLOW_KEY_PROTO_DST) \| \
	271	(1 << FLOW_KEY_NFCT_SRC) \| \
	272	(1 << FLOW_KEY_NFCT_DST) \| \
	273	(1 << FLOW_KEY_NFCT_PROTO_SRC) \| \
	274	(1 << FLOW_KEY_NFCT_PROTO_DST))
	275
dc7f9f6e	276	static int flow_classify(struct sk_buff skb, const struct tcf_proto tp,
e5dfb815 PM	277	struct tcf_result *res)
	278	{
	279	struct flow_head *head = tp->root;
	280	struct flow_filter *f;
	281	u32 keymask;
	282	u32 classid;
	283	unsigned int n, key;
	284	int r;
	285
	286	list_for_each_entry(f, &head->filters, list) {
3a53943b	287	u32 keys[FLOW_KEY_MAX + 1];
6bd2a9af	288	struct flow_keys flow_keys;
e5dfb815 PM	289
	290	if (!tcf_em_tree_match(skb, &f->ematches, NULL))
	291	continue;
	292
	293	keymask = f->keymask;
6bd2a9af ED	294	if (keymask & FLOW_KEYS_NEEDED)
6bd2a9af ED	295	skb_flow_dissect(skb, &flow_keys);
e5dfb815 PM	296
	297	for (n = 0; n < f->nkeys; n++) {
	298	key = ffs(keymask) - 1;
	299	keymask &= ~(1 << key);
6bd2a9af	300	keys[n] = flow_key_get(skb, key, &flow_keys);
e5dfb815 PM	301	}
	302
	303	if (f->mode == FLOW_MODE_HASH)
72d9794f	304	classid = jhash2(keys, f->nkeys, f->hashrnd);
e5dfb815 PM	305	else {
	306	classid = keys[0];
	307	classid = (classid & f->mask) ^ f->xor;
	308	classid = (classid >> f->rshift) + f->addend;
	309	}
	310
	311	if (f->divisor)
	312	classid %= f->divisor;
	313
	314	res->class = 0;
	315	res->classid = TC_H_MAKE(f->baseclass, f->baseclass + classid);
	316
	317	r = tcf_exts_exec(skb, &f->exts, res);
	318	if (r < 0)
	319	continue;
	320	return r;
	321	}
	322	return -1;
	323	}
	324
72d9794f PM	325	static void flow_perturbation(unsigned long arg)
	326	{
	327	struct flow_filter f = (struct flow_filter )arg;
	328
	329	get_random_bytes(&f->hashrnd, 4);
	330	if (f->perturb_period)
	331	mod_timer(&f->perturb_timer, jiffies + f->perturb_period);
	332	}
	333
e5dfb815 PM	334	static const struct nla_policy flow_policy[TCA_FLOW_MAX + 1] = {
	335	[TCA_FLOW_KEYS] = { .type = NLA_U32 },
	336	[TCA_FLOW_MODE] = { .type = NLA_U32 },
	337	[TCA_FLOW_BASECLASS] = { .type = NLA_U32 },
	338	[TCA_FLOW_RSHIFT] = { .type = NLA_U32 },
	339	[TCA_FLOW_ADDEND] = { .type = NLA_U32 },
	340	[TCA_FLOW_MASK] = { .type = NLA_U32 },
	341	[TCA_FLOW_XOR] = { .type = NLA_U32 },
	342	[TCA_FLOW_DIVISOR] = { .type = NLA_U32 },
	343	[TCA_FLOW_ACT] = { .type = NLA_NESTED },
	344	[TCA_FLOW_POLICE] = { .type = NLA_NESTED },
	345	[TCA_FLOW_EMATCHES] = { .type = NLA_NESTED },
72d9794f	346	[TCA_FLOW_PERTURB] = { .type = NLA_U32 },
e5dfb815 PM	347	};
e5dfb815 PM	348
c1b52739	349	static int flow_change(struct net net, struct sk_buff in_skb,
af4c6641	350	struct tcf_proto *tp, unsigned long base,
e5dfb815 PM	351	u32 handle, struct nlattr **tca,
	352	unsigned long *arg)
	353	{
	354	struct flow_head *head = tp->root;
	355	struct flow_filter *f;
	356	struct nlattr *opt = tca[TCA_OPTIONS];
	357	struct nlattr *tb[TCA_FLOW_MAX + 1];
	358	struct tcf_exts e;
	359	struct tcf_ematch_tree t;
	360	unsigned int nkeys = 0;
72d9794f	361	unsigned int perturb_period = 0;
e5dfb815 PM	362	u32 baseclass = 0;
	363	u32 keymask = 0;
	364	u32 mode;
	365	int err;
	366
	367	if (opt == NULL)
	368	return -EINVAL;
	369
	370	err = nla_parse_nested(tb, TCA_FLOW_MAX, opt, flow_policy);
	371	if (err < 0)
	372	return err;
	373
	374	if (tb[TCA_FLOW_BASECLASS]) {
	375	baseclass = nla_get_u32(tb[TCA_FLOW_BASECLASS]);
	376	if (TC_H_MIN(baseclass) == 0)
	377	return -EINVAL;
	378	}
	379
	380	if (tb[TCA_FLOW_KEYS]) {
	381	keymask = nla_get_u32(tb[TCA_FLOW_KEYS]);
e5dfb815 PM	382
	383	nkeys = hweight32(keymask);
	384	if (nkeys == 0)
	385	return -EINVAL;
4f250491 PM	386
	387	if (fls(keymask) - 1 > FLOW_KEY_MAX)
	388	return -EOPNOTSUPP;
a6c6796c EB	389
a6c6796c EB	390	if ((keymask & (FLOW_KEY_SKUID\|FLOW_KEY_SKGID)) &&
e32123e5	391	sk_user_ns(NETLINK_CB(in_skb).sk) != &init_user_ns)
a6c6796c	392	return -EOPNOTSUPP;
e5dfb815 PM	393	}
e5dfb815 PM	394
5da57f42 WC	395	tcf_exts_init(&e, TCA_FLOW_ACT, TCA_FLOW_POLICE);
5da57f42 WC	396	err = tcf_exts_validate(net, tp, tb, tca[TCA_RATE], &e);
e5dfb815 PM	397	if (err < 0)
	398	return err;
	399
	400	err = tcf_em_tree_validate(tp, tb[TCA_FLOW_EMATCHES], &t);
	401	if (err < 0)
	402	goto err1;
	403
	404	f = (struct flow_filter )arg;
	405	if (f != NULL) {
	406	err = -EINVAL;
	407	if (f->handle != handle && handle)
	408	goto err2;
	409
	410	mode = f->mode;
	411	if (tb[TCA_FLOW_MODE])
	412	mode = nla_get_u32(tb[TCA_FLOW_MODE]);
	413	if (mode != FLOW_MODE_HASH && nkeys > 1)
	414	goto err2;
72d9794f PM	415
	416	if (mode == FLOW_MODE_HASH)
	417	perturb_period = f->perturb_period;
	418	if (tb[TCA_FLOW_PERTURB]) {
	419	if (mode != FLOW_MODE_HASH)
	420	goto err2;
	421	perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
	422	}
e5dfb815 PM	423	} else {
	424	err = -EINVAL;
	425	if (!handle)
	426	goto err2;
	427	if (!tb[TCA_FLOW_KEYS])
	428	goto err2;
	429
	430	mode = FLOW_MODE_MAP;
	431	if (tb[TCA_FLOW_MODE])
	432	mode = nla_get_u32(tb[TCA_FLOW_MODE]);
	433	if (mode != FLOW_MODE_HASH && nkeys > 1)
	434	goto err2;
	435
72d9794f PM	436	if (tb[TCA_FLOW_PERTURB]) {
	437	if (mode != FLOW_MODE_HASH)
	438	goto err2;
	439	perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
	440	}
	441
e5dfb815 PM	442	if (TC_H_MAJ(baseclass) == 0)
	443	baseclass = TC_H_MAKE(tp->q->handle, baseclass);
	444	if (TC_H_MIN(baseclass) == 0)
	445	baseclass = TC_H_MAKE(baseclass, 1);
	446
	447	err = -ENOBUFS;
	448	f = kzalloc(sizeof(*f), GFP_KERNEL);
	449	if (f == NULL)
	450	goto err2;
	451
	452	f->handle = handle;
	453	f->mask = ~0U;
5da57f42	454	tcf_exts_init(&f->exts, TCA_FLOW_ACT, TCA_FLOW_POLICE);
72d9794f PM	455
	456	get_random_bytes(&f->hashrnd, 4);
	457	f->perturb_timer.function = flow_perturbation;
	458	f->perturb_timer.data = (unsigned long)f;
	459	init_timer_deferrable(&f->perturb_timer);
e5dfb815 PM	460	}
	461
	462	tcf_exts_change(tp, &f->exts, &e);
	463	tcf_em_tree_change(tp, &f->ematches, &t);
	464
	465	tcf_tree_lock(tp);
	466
	467	if (tb[TCA_FLOW_KEYS]) {
	468	f->keymask = keymask;
	469	f->nkeys = nkeys;
	470	}
	471
	472	f->mode = mode;
	473
	474	if (tb[TCA_FLOW_MASK])
	475	f->mask = nla_get_u32(tb[TCA_FLOW_MASK]);
	476	if (tb[TCA_FLOW_XOR])
	477	f->xor = nla_get_u32(tb[TCA_FLOW_XOR]);
	478	if (tb[TCA_FLOW_RSHIFT])
	479	f->rshift = nla_get_u32(tb[TCA_FLOW_RSHIFT]);
	480	if (tb[TCA_FLOW_ADDEND])
	481	f->addend = nla_get_u32(tb[TCA_FLOW_ADDEND]);
	482
	483	if (tb[TCA_FLOW_DIVISOR])
	484	f->divisor = nla_get_u32(tb[TCA_FLOW_DIVISOR]);
	485	if (baseclass)
	486	f->baseclass = baseclass;
	487
72d9794f PM	488	f->perturb_period = perturb_period;
	489	del_timer(&f->perturb_timer);
	490	if (perturb_period)
	491	mod_timer(&f->perturb_timer, jiffies + perturb_period);
	492
e5dfb815 PM	493	if (*arg == 0)
	494	list_add_tail(&f->list, &head->filters);
	495
	496	tcf_tree_unlock(tp);
	497
	498	*arg = (unsigned long)f;
	499	return 0;
	500
	501	err2:
	502	tcf_em_tree_destroy(tp, &t);
	503	err1:
	504	tcf_exts_destroy(tp, &e);
	505	return err;
	506	}
	507
	508	static void flow_destroy_filter(struct tcf_proto tp, struct flow_filter f)
	509	{
72d9794f	510	del_timer_sync(&f->perturb_timer);
e5dfb815 PM	511	tcf_exts_destroy(tp, &f->exts);
	512	tcf_em_tree_destroy(tp, &f->ematches);
	513	kfree(f);
	514	}
	515
	516	static int flow_delete(struct tcf_proto *tp, unsigned long arg)
	517	{
	518	struct flow_filter f = (struct flow_filter )arg;
	519
	520	tcf_tree_lock(tp);
	521	list_del(&f->list);
	522	tcf_tree_unlock(tp);
	523	flow_destroy_filter(tp, f);
	524	return 0;
	525	}
	526
	527	static int flow_init(struct tcf_proto *tp)
	528	{
	529	struct flow_head *head;
	530
e5dfb815 PM	531	head = kzalloc(sizeof(*head), GFP_KERNEL);
	532	if (head == NULL)
	533	return -ENOBUFS;
	534	INIT_LIST_HEAD(&head->filters);
	535	tp->root = head;
	536	return 0;
	537	}
	538
	539	static void flow_destroy(struct tcf_proto *tp)
	540	{
	541	struct flow_head *head = tp->root;
	542	struct flow_filter f, next;
	543
	544	list_for_each_entry_safe(f, next, &head->filters, list) {
	545	list_del(&f->list);
	546	flow_destroy_filter(tp, f);
	547	}
	548	kfree(head);
	549	}
	550
	551	static unsigned long flow_get(struct tcf_proto *tp, u32 handle)
	552	{
	553	struct flow_head *head = tp->root;
	554	struct flow_filter *f;
	555
	556	list_for_each_entry(f, &head->filters, list)
	557	if (f->handle == handle)
	558	return (unsigned long)f;
	559	return 0;
	560	}
	561
	562	static void flow_put(struct tcf_proto *tp, unsigned long f)
	563	{
e5dfb815 PM	564	}
	565
	566	static int flow_dump(struct tcf_proto *tp, unsigned long fh,
	567	struct sk_buff skb, struct tcmsg t)
	568	{
	569	struct flow_filter f = (struct flow_filter )fh;
	570	struct nlattr *nest;
	571
	572	if (f == NULL)
	573	return skb->len;
	574
	575	t->tcm_handle = f->handle;
	576
	577	nest = nla_nest_start(skb, TCA_OPTIONS);
	578	if (nest == NULL)
	579	goto nla_put_failure;
	580
1b34ec43 DM	581	if (nla_put_u32(skb, TCA_FLOW_KEYS, f->keymask) \|\|
	582	nla_put_u32(skb, TCA_FLOW_MODE, f->mode))
	583	goto nla_put_failure;
e5dfb815 PM	584
e5dfb815 PM	585	if (f->mask != ~0 \|\| f->xor != 0) {
1b34ec43 DM	586	if (nla_put_u32(skb, TCA_FLOW_MASK, f->mask) \|\|
	587	nla_put_u32(skb, TCA_FLOW_XOR, f->xor))
	588	goto nla_put_failure;
e5dfb815	589	}
1b34ec43 DM	590	if (f->rshift &&
	591	nla_put_u32(skb, TCA_FLOW_RSHIFT, f->rshift))
	592	goto nla_put_failure;
	593	if (f->addend &&
	594	nla_put_u32(skb, TCA_FLOW_ADDEND, f->addend))
	595	goto nla_put_failure;
e5dfb815	596
1b34ec43 DM	597	if (f->divisor &&
	598	nla_put_u32(skb, TCA_FLOW_DIVISOR, f->divisor))
	599	goto nla_put_failure;
	600	if (f->baseclass &&
	601	nla_put_u32(skb, TCA_FLOW_BASECLASS, f->baseclass))
	602	goto nla_put_failure;
e5dfb815	603
1b34ec43 DM	604	if (f->perturb_period &&
	605	nla_put_u32(skb, TCA_FLOW_PERTURB, f->perturb_period / HZ))
	606	goto nla_put_failure;
72d9794f	607
5da57f42	608	if (tcf_exts_dump(skb, &f->exts) < 0)
e5dfb815	609	goto nla_put_failure;
0aead543	610	#ifdef CONFIG_NET_EMATCH
e5dfb815 PM	611	if (f->ematches.hdr.nmatches &&
	612	tcf_em_tree_dump(skb, &f->ematches, TCA_FLOW_EMATCHES) < 0)
	613	goto nla_put_failure;
0aead543	614	#endif
e5dfb815 PM	615	nla_nest_end(skb, nest);
e5dfb815 PM	616
5da57f42	617	if (tcf_exts_dump_stats(skb, &f->exts) < 0)
e5dfb815 PM	618	goto nla_put_failure;
	619
	620	return skb->len;
	621
	622	nla_put_failure:
	623	nlmsg_trim(skb, nest);
	624	return -1;
	625	}
	626
	627	static void flow_walk(struct tcf_proto tp, struct tcf_walker arg)
	628	{
	629	struct flow_head *head = tp->root;
	630	struct flow_filter *f;
	631
	632	list_for_each_entry(f, &head->filters, list) {
	633	if (arg->count < arg->skip)
	634	goto skip;
	635	if (arg->fn(tp, (unsigned long)f, arg) < 0) {
	636	arg->stop = 1;
	637	break;
	638	}
	639	skip:
	640	arg->count++;
	641	}
	642	}
	643
	644	static struct tcf_proto_ops cls_flow_ops __read_mostly = {
	645	.kind = "flow",
	646	.classify = flow_classify,
	647	.init = flow_init,
	648	.destroy = flow_destroy,
	649	.change = flow_change,
	650	.delete = flow_delete,
	651	.get = flow_get,
	652	.put = flow_put,
	653	.dump = flow_dump,
	654	.walk = flow_walk,
	655	.owner = THIS_MODULE,
	656	};
	657
	658	static int __init cls_flow_init(void)
	659	{
	660	return register_tcf_proto_ops(&cls_flow_ops);
	661	}
	662
	663	static void __exit cls_flow_exit(void)
	664	{
	665	unregister_tcf_proto_ops(&cls_flow_ops);
	666	}
	667
	668	module_init(cls_flow_init);
	669	module_exit(cls_flow_exit);
	670
	671	MODULE_LICENSE("GPL");
	672	MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
	673	MODULE_DESCRIPTION("TC flow classifier");