[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 rcu_head		rcu;
};

struct flow_filter {
	struct list_head	list;
	struct tcf_exts		exts;
	struct tcf_ematch_tree	ematches;
	struct tcf_proto	*tp;
	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;
	struct rcu_head		rcu;
};

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 = rcu_dereference_bh(tp->root);
	struct flow_filter *f;
	u32 keymask;
	u32 classid;
	unsigned int n, key;
	int r;

	list_for_each_entry_rcu(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 void flow_destroy_filter(struct rcu_head *head)
{
	struct flow_filter *f = container_of(head, struct flow_filter, rcu);

	del_timer_sync(&f->perturb_timer);
	tcf_exts_destroy(&f->exts);
	tcf_em_tree_destroy(&f->ematches);
	kfree(f);
}

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, bool ovr)
{
	struct flow_head *head = rtnl_dereference(tp->root);
	struct flow_filter *fold, *fnew;
	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, ovr);
	if (err < 0)
		return err;

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

	err = -ENOBUFS;
	fnew = kzalloc(sizeof(*fnew), GFP_KERNEL);
	if (!fnew)
		goto err2;

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

		/* Copy fold into fnew */
		fnew->handle = fold->handle;
		fnew->keymask = fold->keymask;
		fnew->tp = fold->tp;

		fnew->handle = fold->handle;
		fnew->nkeys = fold->nkeys;
		fnew->keymask = fold->keymask;
		fnew->mode = fold->mode;
		fnew->mask = fold->mask;
		fnew->xor = fold->xor;
		fnew->rshift = fold->rshift;
		fnew->addend = fold->addend;
		fnew->divisor = fold->divisor;
		fnew->baseclass = fold->baseclass;
		fnew->hashrnd = fold->hashrnd;

		mode = fold->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 = fold->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);

		fnew->handle = handle;
		fnew->mask  = ~0U;
		fnew->tp = tp;
		get_random_bytes(&fnew->hashrnd, 4);
		tcf_exts_init(&fnew->exts, TCA_FLOW_ACT, TCA_FLOW_POLICE);
	}

	fnew->perturb_timer.function = flow_perturbation;
	fnew->perturb_timer.data = (unsigned long)fnew;
	init_timer_deferrable(&fnew->perturb_timer);

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

	netif_keep_dst(qdisc_dev(tp->q));

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

	fnew->mode = mode;

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

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

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

	if (*arg == 0)
		list_add_tail_rcu(&fnew->list, &head->filters);
	else
		list_replace_rcu(&fnew->list, &fold->list);

	*arg = (unsigned long)fnew;

	if (fold)
		call_rcu(&fold->rcu, flow_destroy_filter);
	return 0;

err2:
	tcf_em_tree_destroy(&t);
	kfree(fnew);
err1:
	tcf_exts_destroy(&e);
	return err;
}

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

	list_del_rcu(&f->list);
	call_rcu(&f->rcu, flow_destroy_filter);
	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);
	rcu_assign_pointer(tp->root, head);
	return 0;
}

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

	list_for_each_entry_safe(f, next, &head->filters, list) {
		list_del_rcu(&f->list);
		call_rcu(&f->rcu, flow_destroy_filter);
	}
	RCU_INIT_POINTER(tp->root, NULL);
	kfree_rcu(head, rcu);
}

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

	list_for_each_entry_rcu(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 net *net, 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 = rtnl_dereference(tp->root);
	struct flow_filter *f;

	list_for_each_entry_rcu(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;
70da9f0b	37	struct rcu_head rcu;
e5dfb815 PM	38	};
	39
	40	struct flow_filter {
	41	struct list_head list;
	42	struct tcf_exts exts;
	43	struct tcf_ematch_tree ematches;
70da9f0b	44	struct tcf_proto *tp;
72d9794f PM	45	struct timer_list perturb_timer;
72d9794f PM	46	u32 perturb_period;
e5dfb815 PM	47	u32 handle;
	48
	49	u32 nkeys;
	50	u32 keymask;
	51	u32 mode;
	52	u32 mask;
	53	u32 xor;
	54	u32 rshift;
	55	u32 addend;
	56	u32 divisor;
	57	u32 baseclass;
72d9794f	58	u32 hashrnd;
70da9f0b	59	struct rcu_head rcu;
e5dfb815 PM	60	};
e5dfb815 PM	61
e5dfb815 PM	62	static inline u32 addr_fold(void *addr)
	63	{
	64	unsigned long a = (unsigned long)addr;
	65
	66	return (a & 0xFFFFFFFF) ^ (BITS_PER_LONG > 32 ? a >> 32 : 0);
	67	}
	68
6bd2a9af	69	static u32 flow_get_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	70	{
6bd2a9af ED	71	if (flow->src)
6bd2a9af ED	72	return ntohl(flow->src);
4b95c3d4	73	return addr_fold(skb->sk);
e5dfb815 PM	74	}
e5dfb815 PM	75
6bd2a9af	76	static u32 flow_get_dst(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	77	{
6bd2a9af ED	78	if (flow->dst)
6bd2a9af ED	79	return ntohl(flow->dst);
4b95c3d4	80	return addr_fold(skb_dst(skb)) ^ (__force u16)skb->protocol;
e5dfb815 PM	81	}
e5dfb815 PM	82
6bd2a9af	83	static u32 flow_get_proto(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	84	{
6bd2a9af	85	return flow->ip_proto;
e5dfb815 PM	86	}
e5dfb815 PM	87
6bd2a9af	88	static u32 flow_get_proto_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	89	{
6bd2a9af ED	90	if (flow->ports)
6bd2a9af ED	91	return ntohs(flow->port16[0]);
e5dfb815	92
859c2012 ED	93	return addr_fold(skb->sk);
	94	}
	95
6bd2a9af	96	static u32 flow_get_proto_dst(const struct sk_buff skb, const struct flow_keys flow)
859c2012	97	{
6bd2a9af ED	98	if (flow->ports)
6bd2a9af ED	99	return ntohs(flow->port16[1]);
e5dfb815	100
4b95c3d4	101	return addr_fold(skb_dst(skb)) ^ (__force u16)skb->protocol;
e5dfb815 PM	102	}
	103
	104	static u32 flow_get_iif(const struct sk_buff *skb)
	105	{
8964be4a	106	return skb->skb_iif;
e5dfb815 PM	107	}
	108
	109	static u32 flow_get_priority(const struct sk_buff *skb)
	110	{
	111	return skb->priority;
	112	}
	113
	114	static u32 flow_get_mark(const struct sk_buff *skb)
	115	{
	116	return skb->mark;
	117	}
	118
	119	static u32 flow_get_nfct(const struct sk_buff *skb)
	120	{
	121	#if defined(CONFIG_NF_CONNTRACK) \|\| defined(CONFIG_NF_CONNTRACK_MODULE)
	122	return addr_fold(skb->nfct);
	123	#else
	124	return 0;
	125	#endif
	126	}
	127
	128	#if defined(CONFIG_NF_CONNTRACK) \|\| defined(CONFIG_NF_CONNTRACK_MODULE)
	129	#define CTTUPLE(skb, member) \
	130	({ \
	131	enum ip_conntrack_info ctinfo; \
859c2012	132	const struct nf_conn *ct = nf_ct_get(skb, &ctinfo); \
e5dfb815 PM	133	if (ct == NULL) \
	134	goto fallback; \
	135	ct->tuplehash[CTINFO2DIR(ctinfo)].tuple.member; \
	136	})
	137	#else
	138	#define CTTUPLE(skb, member) \
	139	({ \
	140	goto fallback; \
	141	0; \
	142	})
	143	#endif
	144
6bd2a9af	145	static u32 flow_get_nfct_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815 PM	146	{
e5dfb815 PM	147	switch (skb->protocol) {
60678040	148	case htons(ETH_P_IP):
e5dfb815	149	return ntohl(CTTUPLE(skb, src.u3.ip));
60678040	150	case htons(ETH_P_IPV6):
e5dfb815 PM	151	return ntohl(CTTUPLE(skb, src.u3.ip6[3]));
	152	}
	153	fallback:
6bd2a9af	154	return flow_get_src(skb, flow);
e5dfb815 PM	155	}
e5dfb815 PM	156
6bd2a9af	157	static u32 flow_get_nfct_dst(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815 PM	158	{
e5dfb815 PM	159	switch (skb->protocol) {
60678040	160	case htons(ETH_P_IP):
e5dfb815	161	return ntohl(CTTUPLE(skb, dst.u3.ip));
60678040	162	case htons(ETH_P_IPV6):
e5dfb815 PM	163	return ntohl(CTTUPLE(skb, dst.u3.ip6[3]));
	164	}
	165	fallback:
6bd2a9af	166	return flow_get_dst(skb, flow);
e5dfb815 PM	167	}
e5dfb815 PM	168
6bd2a9af	169	static u32 flow_get_nfct_proto_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815 PM	170	{
	171	return ntohs(CTTUPLE(skb, src.u.all));
	172	fallback:
6bd2a9af	173	return flow_get_proto_src(skb, flow);
e5dfb815 PM	174	}
e5dfb815 PM	175
6bd2a9af	176	static u32 flow_get_nfct_proto_dst(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815 PM	177	{
	178	return ntohs(CTTUPLE(skb, dst.u.all));
	179	fallback:
6bd2a9af	180	return flow_get_proto_dst(skb, flow);
e5dfb815 PM	181	}
	182
	183	static u32 flow_get_rtclassid(const struct sk_buff *skb)
	184	{
c7066f70	185	#ifdef CONFIG_IP_ROUTE_CLASSID
adf30907 ED	186	if (skb_dst(skb))
adf30907 ED	187	return skb_dst(skb)->tclassid;
e5dfb815 PM	188	#endif
	189	return 0;
	190	}
	191
	192	static u32 flow_get_skuid(const struct sk_buff *skb)
	193	{
a6c6796c EB	194	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file) {
	195	kuid_t skuid = skb->sk->sk_socket->file->f_cred->fsuid;
	196	return from_kuid(&init_user_ns, skuid);
	197	}
e5dfb815 PM	198	return 0;
	199	}
	200
	201	static u32 flow_get_skgid(const struct sk_buff *skb)
	202	{
a6c6796c EB	203	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file) {
	204	kgid_t skgid = skb->sk->sk_socket->file->f_cred->fsgid;
	205	return from_kgid(&init_user_ns, skgid);
	206	}
e5dfb815 PM	207	return 0;
	208	}
	209
9ec13810 PM	210	static u32 flow_get_vlan_tag(const struct sk_buff *skb)
	211	{
	212	u16 uninitialized_var(tag);
	213
	214	if (vlan_get_tag(skb, &tag) < 0)
	215	return 0;
	216	return tag & VLAN_VID_MASK;
	217	}
	218
739a91ef CG	219	static u32 flow_get_rxhash(struct sk_buff *skb)
739a91ef CG	220	{
3958afa1	221	return skb_get_hash(skb);
739a91ef CG	222	}
739a91ef CG	223
6bd2a9af	224	static u32 flow_key_get(struct sk_buff skb, int key, struct flow_keys flow)
e5dfb815 PM	225	{
	226	switch (key) {
	227	case FLOW_KEY_SRC:
6bd2a9af	228	return flow_get_src(skb, flow);
e5dfb815	229	case FLOW_KEY_DST:
6bd2a9af	230	return flow_get_dst(skb, flow);
e5dfb815	231	case FLOW_KEY_PROTO:
6bd2a9af	232	return flow_get_proto(skb, flow);
e5dfb815	233	case FLOW_KEY_PROTO_SRC:
6bd2a9af	234	return flow_get_proto_src(skb, flow);
e5dfb815	235	case FLOW_KEY_PROTO_DST:
6bd2a9af	236	return flow_get_proto_dst(skb, flow);
e5dfb815 PM	237	case FLOW_KEY_IIF:
	238	return flow_get_iif(skb);
	239	case FLOW_KEY_PRIORITY:
	240	return flow_get_priority(skb);
	241	case FLOW_KEY_MARK:
	242	return flow_get_mark(skb);
	243	case FLOW_KEY_NFCT:
	244	return flow_get_nfct(skb);
	245	case FLOW_KEY_NFCT_SRC:
6bd2a9af	246	return flow_get_nfct_src(skb, flow);
e5dfb815	247	case FLOW_KEY_NFCT_DST:
6bd2a9af	248	return flow_get_nfct_dst(skb, flow);
e5dfb815	249	case FLOW_KEY_NFCT_PROTO_SRC:
6bd2a9af	250	return flow_get_nfct_proto_src(skb, flow);
e5dfb815	251	case FLOW_KEY_NFCT_PROTO_DST:
6bd2a9af	252	return flow_get_nfct_proto_dst(skb, flow);
e5dfb815 PM	253	case FLOW_KEY_RTCLASSID:
	254	return flow_get_rtclassid(skb);
	255	case FLOW_KEY_SKUID:
	256	return flow_get_skuid(skb);
	257	case FLOW_KEY_SKGID:
	258	return flow_get_skgid(skb);
9ec13810 PM	259	case FLOW_KEY_VLAN_TAG:
9ec13810 PM	260	return flow_get_vlan_tag(skb);
739a91ef CG	261	case FLOW_KEY_RXHASH:
739a91ef CG	262	return flow_get_rxhash(skb);
e5dfb815 PM	263	default:
	264	WARN_ON(1);
	265	return 0;
	266	}
	267	}
	268
6bd2a9af ED	269	#define FLOW_KEYS_NEEDED ((1 << FLOW_KEY_SRC) \| \
	270	(1 << FLOW_KEY_DST) \| \
	271	(1 << FLOW_KEY_PROTO) \| \
	272	(1 << FLOW_KEY_PROTO_SRC) \| \
	273	(1 << FLOW_KEY_PROTO_DST) \| \
	274	(1 << FLOW_KEY_NFCT_SRC) \| \
	275	(1 << FLOW_KEY_NFCT_DST) \| \
	276	(1 << FLOW_KEY_NFCT_PROTO_SRC) \| \
	277	(1 << FLOW_KEY_NFCT_PROTO_DST))
	278
dc7f9f6e	279	static int flow_classify(struct sk_buff skb, const struct tcf_proto tp,
e5dfb815 PM	280	struct tcf_result *res)
e5dfb815 PM	281	{
70da9f0b	282	struct flow_head *head = rcu_dereference_bh(tp->root);
e5dfb815 PM	283	struct flow_filter *f;
	284	u32 keymask;
	285	u32 classid;
	286	unsigned int n, key;
	287	int r;
	288
70da9f0b	289	list_for_each_entry_rcu(f, &head->filters, list) {
3a53943b	290	u32 keys[FLOW_KEY_MAX + 1];
6bd2a9af	291	struct flow_keys flow_keys;
e5dfb815 PM	292
	293	if (!tcf_em_tree_match(skb, &f->ematches, NULL))
	294	continue;
	295
	296	keymask = f->keymask;
6bd2a9af ED	297	if (keymask & FLOW_KEYS_NEEDED)
6bd2a9af ED	298	skb_flow_dissect(skb, &flow_keys);
e5dfb815 PM	299
	300	for (n = 0; n < f->nkeys; n++) {
	301	key = ffs(keymask) - 1;
	302	keymask &= ~(1 << key);
6bd2a9af	303	keys[n] = flow_key_get(skb, key, &flow_keys);
e5dfb815 PM	304	}
	305
	306	if (f->mode == FLOW_MODE_HASH)
72d9794f	307	classid = jhash2(keys, f->nkeys, f->hashrnd);
e5dfb815 PM	308	else {
	309	classid = keys[0];
	310	classid = (classid & f->mask) ^ f->xor;
	311	classid = (classid >> f->rshift) + f->addend;
	312	}
	313
	314	if (f->divisor)
	315	classid %= f->divisor;
	316
	317	res->class = 0;
	318	res->classid = TC_H_MAKE(f->baseclass, f->baseclass + classid);
	319
	320	r = tcf_exts_exec(skb, &f->exts, res);
	321	if (r < 0)
	322	continue;
	323	return r;
	324	}
	325	return -1;
	326	}
	327
72d9794f PM	328	static void flow_perturbation(unsigned long arg)
	329	{
	330	struct flow_filter f = (struct flow_filter )arg;
	331
	332	get_random_bytes(&f->hashrnd, 4);
	333	if (f->perturb_period)
	334	mod_timer(&f->perturb_timer, jiffies + f->perturb_period);
	335	}
	336
e5dfb815 PM	337	static const struct nla_policy flow_policy[TCA_FLOW_MAX + 1] = {
	338	[TCA_FLOW_KEYS] = { .type = NLA_U32 },
	339	[TCA_FLOW_MODE] = { .type = NLA_U32 },
	340	[TCA_FLOW_BASECLASS] = { .type = NLA_U32 },
	341	[TCA_FLOW_RSHIFT] = { .type = NLA_U32 },
	342	[TCA_FLOW_ADDEND] = { .type = NLA_U32 },
	343	[TCA_FLOW_MASK] = { .type = NLA_U32 },
	344	[TCA_FLOW_XOR] = { .type = NLA_U32 },
	345	[TCA_FLOW_DIVISOR] = { .type = NLA_U32 },
	346	[TCA_FLOW_ACT] = { .type = NLA_NESTED },
	347	[TCA_FLOW_POLICE] = { .type = NLA_NESTED },
	348	[TCA_FLOW_EMATCHES] = { .type = NLA_NESTED },
72d9794f	349	[TCA_FLOW_PERTURB] = { .type = NLA_U32 },
e5dfb815 PM	350	};
e5dfb815 PM	351
70da9f0b JF	352	static void flow_destroy_filter(struct rcu_head *head)
	353	{
	354	struct flow_filter *f = container_of(head, struct flow_filter, rcu);
	355
	356	del_timer_sync(&f->perturb_timer);
18d0264f	357	tcf_exts_destroy(&f->exts);
82a470f1	358	tcf_em_tree_destroy(&f->ematches);
70da9f0b JF	359	kfree(f);
	360	}
	361
c1b52739	362	static int flow_change(struct net net, struct sk_buff in_skb,
af4c6641	363	struct tcf_proto *tp, unsigned long base,
e5dfb815	364	u32 handle, struct nlattr **tca,
2f7ef2f8	365	unsigned long *arg, bool ovr)
e5dfb815	366	{
70da9f0b JF	367	struct flow_head *head = rtnl_dereference(tp->root);
70da9f0b JF	368	struct flow_filter fold, fnew;
e5dfb815 PM	369	struct nlattr *opt = tca[TCA_OPTIONS];
	370	struct nlattr *tb[TCA_FLOW_MAX + 1];
	371	struct tcf_exts e;
	372	struct tcf_ematch_tree t;
	373	unsigned int nkeys = 0;
72d9794f	374	unsigned int perturb_period = 0;
e5dfb815 PM	375	u32 baseclass = 0;
	376	u32 keymask = 0;
	377	u32 mode;
	378	int err;
	379
	380	if (opt == NULL)
	381	return -EINVAL;
	382
	383	err = nla_parse_nested(tb, TCA_FLOW_MAX, opt, flow_policy);
	384	if (err < 0)
	385	return err;
	386
	387	if (tb[TCA_FLOW_BASECLASS]) {
	388	baseclass = nla_get_u32(tb[TCA_FLOW_BASECLASS]);
	389	if (TC_H_MIN(baseclass) == 0)
	390	return -EINVAL;
	391	}
	392
	393	if (tb[TCA_FLOW_KEYS]) {
	394	keymask = nla_get_u32(tb[TCA_FLOW_KEYS]);
e5dfb815 PM	395
	396	nkeys = hweight32(keymask);
	397	if (nkeys == 0)
	398	return -EINVAL;
4f250491 PM	399
	400	if (fls(keymask) - 1 > FLOW_KEY_MAX)
	401	return -EOPNOTSUPP;
a6c6796c EB	402
a6c6796c EB	403	if ((keymask & (FLOW_KEY_SKUID\|FLOW_KEY_SKGID)) &&
e32123e5	404	sk_user_ns(NETLINK_CB(in_skb).sk) != &init_user_ns)
a6c6796c	405	return -EOPNOTSUPP;
e5dfb815 PM	406	}
e5dfb815 PM	407
5da57f42	408	tcf_exts_init(&e, TCA_FLOW_ACT, TCA_FLOW_POLICE);
2f7ef2f8	409	err = tcf_exts_validate(net, tp, tb, tca[TCA_RATE], &e, ovr);
e5dfb815 PM	410	if (err < 0)
	411	return err;
	412
	413	err = tcf_em_tree_validate(tp, tb[TCA_FLOW_EMATCHES], &t);
	414	if (err < 0)
	415	goto err1;
	416
70da9f0b JF	417	err = -ENOBUFS;
	418	fnew = kzalloc(sizeof(*fnew), GFP_KERNEL);
	419	if (!fnew)
	420	goto err2;
	421
	422	fold = (struct flow_filter )arg;
	423	if (fold) {
e5dfb815	424	err = -EINVAL;
70da9f0b	425	if (fold->handle != handle && handle)
e5dfb815 PM	426	goto err2;
e5dfb815 PM	427
70da9f0b JF	428	/* Copy fold into fnew */
	429	fnew->handle = fold->handle;
	430	fnew->keymask = fold->keymask;
	431	fnew->tp = fold->tp;
	432
	433	fnew->handle = fold->handle;
	434	fnew->nkeys = fold->nkeys;
	435	fnew->keymask = fold->keymask;
	436	fnew->mode = fold->mode;
	437	fnew->mask = fold->mask;
	438	fnew->xor = fold->xor;
	439	fnew->rshift = fold->rshift;
	440	fnew->addend = fold->addend;
	441	fnew->divisor = fold->divisor;
	442	fnew->baseclass = fold->baseclass;
	443	fnew->hashrnd = fold->hashrnd;
	444
	445	mode = fold->mode;
e5dfb815 PM	446	if (tb[TCA_FLOW_MODE])
	447	mode = nla_get_u32(tb[TCA_FLOW_MODE]);
	448	if (mode != FLOW_MODE_HASH && nkeys > 1)
	449	goto err2;
72d9794f PM	450
72d9794f PM	451	if (mode == FLOW_MODE_HASH)
70da9f0b	452	perturb_period = fold->perturb_period;
72d9794f PM	453	if (tb[TCA_FLOW_PERTURB]) {
	454	if (mode != FLOW_MODE_HASH)
	455	goto err2;
	456	perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
	457	}
e5dfb815 PM	458	} else {
	459	err = -EINVAL;
	460	if (!handle)
	461	goto err2;
	462	if (!tb[TCA_FLOW_KEYS])
	463	goto err2;
	464
	465	mode = FLOW_MODE_MAP;
	466	if (tb[TCA_FLOW_MODE])
	467	mode = nla_get_u32(tb[TCA_FLOW_MODE]);
	468	if (mode != FLOW_MODE_HASH && nkeys > 1)
	469	goto err2;
	470
72d9794f PM	471	if (tb[TCA_FLOW_PERTURB]) {
	472	if (mode != FLOW_MODE_HASH)
	473	goto err2;
	474	perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
	475	}
	476
e5dfb815 PM	477	if (TC_H_MAJ(baseclass) == 0)
	478	baseclass = TC_H_MAKE(tp->q->handle, baseclass);
	479	if (TC_H_MIN(baseclass) == 0)
	480	baseclass = TC_H_MAKE(baseclass, 1);
	481
70da9f0b JF	482	fnew->handle = handle;
	483	fnew->mask = ~0U;
	484	fnew->tp = tp;
	485	get_random_bytes(&fnew->hashrnd, 4);
	486	tcf_exts_init(&fnew->exts, TCA_FLOW_ACT, TCA_FLOW_POLICE);
e5dfb815 PM	487	}
e5dfb815 PM	488
70da9f0b JF	489	fnew->perturb_timer.function = flow_perturbation;
	490	fnew->perturb_timer.data = (unsigned long)fnew;
	491	init_timer_deferrable(&fnew->perturb_timer);
e5dfb815	492
70da9f0b JF	493	tcf_exts_change(tp, &fnew->exts, &e);
70da9f0b JF	494	tcf_em_tree_change(tp, &fnew->ematches, &t);
e5dfb815	495
02875878 ED	496	netif_keep_dst(qdisc_dev(tp->q));
02875878 ED	497
e5dfb815	498	if (tb[TCA_FLOW_KEYS]) {
70da9f0b JF	499	fnew->keymask = keymask;
70da9f0b JF	500	fnew->nkeys = nkeys;
e5dfb815 PM	501	}
e5dfb815 PM	502
70da9f0b	503	fnew->mode = mode;
e5dfb815 PM	504
e5dfb815 PM	505	if (tb[TCA_FLOW_MASK])
70da9f0b	506	fnew->mask = nla_get_u32(tb[TCA_FLOW_MASK]);
e5dfb815	507	if (tb[TCA_FLOW_XOR])
70da9f0b	508	fnew->xor = nla_get_u32(tb[TCA_FLOW_XOR]);
e5dfb815	509	if (tb[TCA_FLOW_RSHIFT])
70da9f0b	510	fnew->rshift = nla_get_u32(tb[TCA_FLOW_RSHIFT]);
e5dfb815	511	if (tb[TCA_FLOW_ADDEND])
70da9f0b	512	fnew->addend = nla_get_u32(tb[TCA_FLOW_ADDEND]);
e5dfb815 PM	513
e5dfb815 PM	514	if (tb[TCA_FLOW_DIVISOR])
70da9f0b	515	fnew->divisor = nla_get_u32(tb[TCA_FLOW_DIVISOR]);
e5dfb815	516	if (baseclass)
70da9f0b	517	fnew->baseclass = baseclass;
e5dfb815	518
70da9f0b	519	fnew->perturb_period = perturb_period;
72d9794f	520	if (perturb_period)
70da9f0b	521	mod_timer(&fnew->perturb_timer, jiffies + perturb_period);
72d9794f	522
e5dfb815	523	if (*arg == 0)
70da9f0b JF	524	list_add_tail_rcu(&fnew->list, &head->filters);
	525	else
	526	list_replace_rcu(&fnew->list, &fold->list);
e5dfb815	527
70da9f0b	528	*arg = (unsigned long)fnew;
e5dfb815	529
70da9f0b JF	530	if (fold)
70da9f0b JF	531	call_rcu(&fold->rcu, flow_destroy_filter);
e5dfb815 PM	532	return 0;
	533
	534	err2:
82a470f1	535	tcf_em_tree_destroy(&t);
70da9f0b	536	kfree(fnew);
e5dfb815	537	err1:
18d0264f	538	tcf_exts_destroy(&e);
e5dfb815 PM	539	return err;
	540	}
	541
e5dfb815 PM	542	static int flow_delete(struct tcf_proto *tp, unsigned long arg)
	543	{
	544	struct flow_filter f = (struct flow_filter )arg;
	545
70da9f0b JF	546	list_del_rcu(&f->list);
70da9f0b JF	547	call_rcu(&f->rcu, flow_destroy_filter);
e5dfb815 PM	548	return 0;
	549	}
	550
	551	static int flow_init(struct tcf_proto *tp)
	552	{
	553	struct flow_head *head;
	554
e5dfb815 PM	555	head = kzalloc(sizeof(*head), GFP_KERNEL);
	556	if (head == NULL)
	557	return -ENOBUFS;
	558	INIT_LIST_HEAD(&head->filters);
70da9f0b	559	rcu_assign_pointer(tp->root, head);
e5dfb815 PM	560	return 0;
	561	}
	562
	563	static void flow_destroy(struct tcf_proto *tp)
	564	{
70da9f0b	565	struct flow_head *head = rtnl_dereference(tp->root);
e5dfb815 PM	566	struct flow_filter f, next;
	567
	568	list_for_each_entry_safe(f, next, &head->filters, list) {
70da9f0b JF	569	list_del_rcu(&f->list);
70da9f0b JF	570	call_rcu(&f->rcu, flow_destroy_filter);
e5dfb815	571	}
70da9f0b JF	572	RCU_INIT_POINTER(tp->root, NULL);
70da9f0b JF	573	kfree_rcu(head, rcu);
e5dfb815 PM	574	}
	575
	576	static unsigned long flow_get(struct tcf_proto *tp, u32 handle)
	577	{
70da9f0b	578	struct flow_head *head = rtnl_dereference(tp->root);
e5dfb815 PM	579	struct flow_filter *f;
e5dfb815 PM	580
70da9f0b	581	list_for_each_entry_rcu(f, &head->filters, list)
e5dfb815 PM	582	if (f->handle == handle)
	583	return (unsigned long)f;
	584	return 0;
	585	}
	586
	587	static void flow_put(struct tcf_proto *tp, unsigned long f)
	588	{
e5dfb815 PM	589	}
e5dfb815 PM	590
832d1d5b	591	static int flow_dump(struct net net, struct tcf_proto tp, unsigned long fh,
e5dfb815 PM	592	struct sk_buff skb, struct tcmsg t)
	593	{
	594	struct flow_filter f = (struct flow_filter )fh;
	595	struct nlattr *nest;
	596
	597	if (f == NULL)
	598	return skb->len;
	599
	600	t->tcm_handle = f->handle;
	601
	602	nest = nla_nest_start(skb, TCA_OPTIONS);
	603	if (nest == NULL)
	604	goto nla_put_failure;
	605
1b34ec43 DM	606	if (nla_put_u32(skb, TCA_FLOW_KEYS, f->keymask) \|\|
	607	nla_put_u32(skb, TCA_FLOW_MODE, f->mode))
	608	goto nla_put_failure;
e5dfb815 PM	609
e5dfb815 PM	610	if (f->mask != ~0 \|\| f->xor != 0) {
1b34ec43 DM	611	if (nla_put_u32(skb, TCA_FLOW_MASK, f->mask) \|\|
	612	nla_put_u32(skb, TCA_FLOW_XOR, f->xor))
	613	goto nla_put_failure;
e5dfb815	614	}
1b34ec43 DM	615	if (f->rshift &&
	616	nla_put_u32(skb, TCA_FLOW_RSHIFT, f->rshift))
	617	goto nla_put_failure;
	618	if (f->addend &&
	619	nla_put_u32(skb, TCA_FLOW_ADDEND, f->addend))
	620	goto nla_put_failure;
e5dfb815	621
1b34ec43 DM	622	if (f->divisor &&
	623	nla_put_u32(skb, TCA_FLOW_DIVISOR, f->divisor))
	624	goto nla_put_failure;
	625	if (f->baseclass &&
	626	nla_put_u32(skb, TCA_FLOW_BASECLASS, f->baseclass))
	627	goto nla_put_failure;
e5dfb815	628
1b34ec43 DM	629	if (f->perturb_period &&
	630	nla_put_u32(skb, TCA_FLOW_PERTURB, f->perturb_period / HZ))
	631	goto nla_put_failure;
72d9794f	632
5da57f42	633	if (tcf_exts_dump(skb, &f->exts) < 0)
e5dfb815	634	goto nla_put_failure;
0aead543	635	#ifdef CONFIG_NET_EMATCH
e5dfb815 PM	636	if (f->ematches.hdr.nmatches &&
	637	tcf_em_tree_dump(skb, &f->ematches, TCA_FLOW_EMATCHES) < 0)
	638	goto nla_put_failure;
0aead543	639	#endif
e5dfb815 PM	640	nla_nest_end(skb, nest);
e5dfb815 PM	641
5da57f42	642	if (tcf_exts_dump_stats(skb, &f->exts) < 0)
e5dfb815 PM	643	goto nla_put_failure;
	644
	645	return skb->len;
	646
	647	nla_put_failure:
	648	nlmsg_trim(skb, nest);
	649	return -1;
	650	}
	651
	652	static void flow_walk(struct tcf_proto tp, struct tcf_walker arg)
	653	{
70da9f0b	654	struct flow_head *head = rtnl_dereference(tp->root);
e5dfb815 PM	655	struct flow_filter *f;
e5dfb815 PM	656
70da9f0b	657	list_for_each_entry_rcu(f, &head->filters, list) {
e5dfb815 PM	658	if (arg->count < arg->skip)
	659	goto skip;
	660	if (arg->fn(tp, (unsigned long)f, arg) < 0) {
	661	arg->stop = 1;
	662	break;
	663	}
	664	skip:
	665	arg->count++;
	666	}
	667	}
	668
	669	static struct tcf_proto_ops cls_flow_ops __read_mostly = {
	670	.kind = "flow",
	671	.classify = flow_classify,
	672	.init = flow_init,
	673	.destroy = flow_destroy,
	674	.change = flow_change,
	675	.delete = flow_delete,
	676	.get = flow_get,
	677	.put = flow_put,
	678	.dump = flow_dump,
	679	.walk = flow_walk,
	680	.owner = THIS_MODULE,
	681	};
	682
	683	static int __init cls_flow_init(void)
	684	{
	685	return register_tcf_proto_ops(&cls_flow_ops);
	686	}
	687
	688	static void __exit cls_flow_exit(void)
	689	{
	690	unregister_tcf_proto_ops(&cls_flow_ops);
	691	}
	692
	693	module_init(cls_flow_init);
	694	module_exit(cls_flow_exit);
	695
	696	MODULE_LICENSE("GPL");
	697	MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
	698	MODULE_DESCRIPTION("TC flow classifier");