[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_dissector.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)
{
	__be32 src = flow_get_u32_src(flow);

	if (src)
		return ntohl(src);

	return addr_fold(skb->sk);
}

static u32 flow_get_dst(const struct sk_buff *skb, const struct flow_keys *flow)
{
	__be32 dst = flow_get_u32_dst(flow);

	if (dst)
		return ntohl(dst);

	return addr_fold(skb_dst(skb)) ^ (__force u16) tc_skb_protocol(skb);
}

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

static u32 flow_get_proto_src(const struct sk_buff *skb, const struct flow_keys *flow)
{
	if (flow->ports.ports)
		return ntohs(flow->ports.src);

	return addr_fold(skb->sk);
}

static u32 flow_get_proto_dst(const struct sk_buff *skb, const struct flow_keys *flow)
{
	if (flow->ports.ports)
		return ntohs(flow->ports.dst);

	return addr_fold(skb_dst(skb)) ^ (__force u16) tc_skb_protocol(skb);
}

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 (tc_skb_protocol(skb)) {
	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 (tc_skb_protocol(skb)) {
	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_flow_keys(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;

	tcf_exts_init(&fnew->exts, TCA_FLOW_ACT, TCA_FLOW_POLICE);

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

		/* Copy fold into fnew */
		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);
	}

	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(&fold->list, &fnew->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 bool flow_destroy(struct tcf_proto *tp, bool force)
{
	struct flow_head *head = rtnl_dereference(tp->root);
	struct flow_filter *f, *next;

	if (!force && !list_empty(&head->filters))
		return false;

	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);
	return true;
}

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

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:
	nla_nest_cancel(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(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,
	.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>
1bd758eb	29	#include <net/flow_dissector.h>
6bd2a9af	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	{
c3f83241 TH	71	__be32 src = flow_get_u32_src(flow);
	72
	73	if (src)
	74	return ntohl(src);
	75
4b95c3d4	76	return addr_fold(skb->sk);
e5dfb815 PM	77	}
e5dfb815 PM	78
6bd2a9af	79	static u32 flow_get_dst(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	80	{
c3f83241 TH	81	__be32 dst = flow_get_u32_dst(flow);
	82
	83	if (dst)
	84	return ntohl(dst);
	85
d8b9605d	86	return addr_fold(skb_dst(skb)) ^ (__force u16) tc_skb_protocol(skb);
e5dfb815 PM	87	}
e5dfb815 PM	88
6bd2a9af	89	static u32 flow_get_proto(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	90	{
06635a35	91	return flow->basic.ip_proto;
e5dfb815 PM	92	}
e5dfb815 PM	93
6bd2a9af	94	static u32 flow_get_proto_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	95	{
06635a35	96	if (flow->ports.ports)
59346afe	97	return ntohs(flow->ports.src);
e5dfb815	98
859c2012 ED	99	return addr_fold(skb->sk);
	100	}
	101
6bd2a9af	102	static u32 flow_get_proto_dst(const struct sk_buff skb, const struct flow_keys flow)
859c2012	103	{
06635a35	104	if (flow->ports.ports)
59346afe	105	return ntohs(flow->ports.dst);
e5dfb815	106
d8b9605d	107	return addr_fold(skb_dst(skb)) ^ (__force u16) tc_skb_protocol(skb);
e5dfb815 PM	108	}
	109
	110	static u32 flow_get_iif(const struct sk_buff *skb)
	111	{
8964be4a	112	return skb->skb_iif;
e5dfb815 PM	113	}
	114
	115	static u32 flow_get_priority(const struct sk_buff *skb)
	116	{
	117	return skb->priority;
	118	}
	119
	120	static u32 flow_get_mark(const struct sk_buff *skb)
	121	{
	122	return skb->mark;
	123	}
	124
	125	static u32 flow_get_nfct(const struct sk_buff *skb)
	126	{
	127	#if defined(CONFIG_NF_CONNTRACK) \|\| defined(CONFIG_NF_CONNTRACK_MODULE)
	128	return addr_fold(skb->nfct);
	129	#else
	130	return 0;
	131	#endif
	132	}
	133
	134	#if defined(CONFIG_NF_CONNTRACK) \|\| defined(CONFIG_NF_CONNTRACK_MODULE)
	135	#define CTTUPLE(skb, member) \
	136	({ \
	137	enum ip_conntrack_info ctinfo; \
859c2012	138	const struct nf_conn *ct = nf_ct_get(skb, &ctinfo); \
e5dfb815 PM	139	if (ct == NULL) \
	140	goto fallback; \
	141	ct->tuplehash[CTINFO2DIR(ctinfo)].tuple.member; \
	142	})
	143	#else
	144	#define CTTUPLE(skb, member) \
	145	({ \
	146	goto fallback; \
	147	0; \
	148	})
	149	#endif
	150
6bd2a9af	151	static u32 flow_get_nfct_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	152	{
d8b9605d	153	switch (tc_skb_protocol(skb)) {
60678040	154	case htons(ETH_P_IP):
e5dfb815	155	return ntohl(CTTUPLE(skb, src.u3.ip));
60678040	156	case htons(ETH_P_IPV6):
e5dfb815 PM	157	return ntohl(CTTUPLE(skb, src.u3.ip6[3]));
	158	}
	159	fallback:
6bd2a9af	160	return flow_get_src(skb, flow);
e5dfb815 PM	161	}
e5dfb815 PM	162
6bd2a9af	163	static u32 flow_get_nfct_dst(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815	164	{
d8b9605d	165	switch (tc_skb_protocol(skb)) {
60678040	166	case htons(ETH_P_IP):
e5dfb815	167	return ntohl(CTTUPLE(skb, dst.u3.ip));
60678040	168	case htons(ETH_P_IPV6):
e5dfb815 PM	169	return ntohl(CTTUPLE(skb, dst.u3.ip6[3]));
	170	}
	171	fallback:
6bd2a9af	172	return flow_get_dst(skb, flow);
e5dfb815 PM	173	}
e5dfb815 PM	174
6bd2a9af	175	static u32 flow_get_nfct_proto_src(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815 PM	176	{
	177	return ntohs(CTTUPLE(skb, src.u.all));
	178	fallback:
6bd2a9af	179	return flow_get_proto_src(skb, flow);
e5dfb815 PM	180	}
e5dfb815 PM	181
6bd2a9af	182	static u32 flow_get_nfct_proto_dst(const struct sk_buff skb, const struct flow_keys flow)
e5dfb815 PM	183	{
	184	return ntohs(CTTUPLE(skb, dst.u.all));
	185	fallback:
6bd2a9af	186	return flow_get_proto_dst(skb, flow);
e5dfb815 PM	187	}
	188
	189	static u32 flow_get_rtclassid(const struct sk_buff *skb)
	190	{
c7066f70	191	#ifdef CONFIG_IP_ROUTE_CLASSID
adf30907 ED	192	if (skb_dst(skb))
adf30907 ED	193	return skb_dst(skb)->tclassid;
e5dfb815 PM	194	#endif
	195	return 0;
	196	}
	197
	198	static u32 flow_get_skuid(const struct sk_buff *skb)
	199	{
a6c6796c EB	200	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file) {
	201	kuid_t skuid = skb->sk->sk_socket->file->f_cred->fsuid;
	202	return from_kuid(&init_user_ns, skuid);
	203	}
e5dfb815 PM	204	return 0;
	205	}
	206
	207	static u32 flow_get_skgid(const struct sk_buff *skb)
	208	{
a6c6796c EB	209	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file) {
	210	kgid_t skgid = skb->sk->sk_socket->file->f_cred->fsgid;
	211	return from_kgid(&init_user_ns, skgid);
	212	}
e5dfb815 PM	213	return 0;
	214	}
	215
9ec13810 PM	216	static u32 flow_get_vlan_tag(const struct sk_buff *skb)
	217	{
	218	u16 uninitialized_var(tag);
	219
	220	if (vlan_get_tag(skb, &tag) < 0)
	221	return 0;
	222	return tag & VLAN_VID_MASK;
	223	}
	224
739a91ef CG	225	static u32 flow_get_rxhash(struct sk_buff *skb)
739a91ef CG	226	{
3958afa1	227	return skb_get_hash(skb);
739a91ef CG	228	}
739a91ef CG	229
6bd2a9af	230	static u32 flow_key_get(struct sk_buff skb, int key, struct flow_keys flow)
e5dfb815 PM	231	{
	232	switch (key) {
	233	case FLOW_KEY_SRC:
6bd2a9af	234	return flow_get_src(skb, flow);
e5dfb815	235	case FLOW_KEY_DST:
6bd2a9af	236	return flow_get_dst(skb, flow);
e5dfb815	237	case FLOW_KEY_PROTO:
6bd2a9af	238	return flow_get_proto(skb, flow);
e5dfb815	239	case FLOW_KEY_PROTO_SRC:
6bd2a9af	240	return flow_get_proto_src(skb, flow);
e5dfb815	241	case FLOW_KEY_PROTO_DST:
6bd2a9af	242	return flow_get_proto_dst(skb, flow);
e5dfb815 PM	243	case FLOW_KEY_IIF:
	244	return flow_get_iif(skb);
	245	case FLOW_KEY_PRIORITY:
	246	return flow_get_priority(skb);
	247	case FLOW_KEY_MARK:
	248	return flow_get_mark(skb);
	249	case FLOW_KEY_NFCT:
	250	return flow_get_nfct(skb);
	251	case FLOW_KEY_NFCT_SRC:
6bd2a9af	252	return flow_get_nfct_src(skb, flow);
e5dfb815	253	case FLOW_KEY_NFCT_DST:
6bd2a9af	254	return flow_get_nfct_dst(skb, flow);
e5dfb815	255	case FLOW_KEY_NFCT_PROTO_SRC:
6bd2a9af	256	return flow_get_nfct_proto_src(skb, flow);
e5dfb815	257	case FLOW_KEY_NFCT_PROTO_DST:
6bd2a9af	258	return flow_get_nfct_proto_dst(skb, flow);
e5dfb815 PM	259	case FLOW_KEY_RTCLASSID:
	260	return flow_get_rtclassid(skb);
	261	case FLOW_KEY_SKUID:
	262	return flow_get_skuid(skb);
	263	case FLOW_KEY_SKGID:
	264	return flow_get_skgid(skb);
9ec13810 PM	265	case FLOW_KEY_VLAN_TAG:
9ec13810 PM	266	return flow_get_vlan_tag(skb);
739a91ef CG	267	case FLOW_KEY_RXHASH:
739a91ef CG	268	return flow_get_rxhash(skb);
e5dfb815 PM	269	default:
	270	WARN_ON(1);
	271	return 0;
	272	}
	273	}
	274
6bd2a9af ED	275	#define FLOW_KEYS_NEEDED ((1 << FLOW_KEY_SRC) \| \
	276	(1 << FLOW_KEY_DST) \| \
	277	(1 << FLOW_KEY_PROTO) \| \
	278	(1 << FLOW_KEY_PROTO_SRC) \| \
	279	(1 << FLOW_KEY_PROTO_DST) \| \
	280	(1 << FLOW_KEY_NFCT_SRC) \| \
	281	(1 << FLOW_KEY_NFCT_DST) \| \
	282	(1 << FLOW_KEY_NFCT_PROTO_SRC) \| \
	283	(1 << FLOW_KEY_NFCT_PROTO_DST))
	284
dc7f9f6e	285	static int flow_classify(struct sk_buff skb, const struct tcf_proto tp,
e5dfb815 PM	286	struct tcf_result *res)
e5dfb815 PM	287	{
70da9f0b	288	struct flow_head *head = rcu_dereference_bh(tp->root);
e5dfb815 PM	289	struct flow_filter *f;
	290	u32 keymask;
	291	u32 classid;
	292	unsigned int n, key;
	293	int r;
	294
70da9f0b	295	list_for_each_entry_rcu(f, &head->filters, list) {
3a53943b	296	u32 keys[FLOW_KEY_MAX + 1];
6bd2a9af	297	struct flow_keys flow_keys;
e5dfb815 PM	298
	299	if (!tcf_em_tree_match(skb, &f->ematches, NULL))
	300	continue;
	301
	302	keymask = f->keymask;
6bd2a9af	303	if (keymask & FLOW_KEYS_NEEDED)
06635a35	304	skb_flow_dissect_flow_keys(skb, &flow_keys);
e5dfb815 PM	305
	306	for (n = 0; n < f->nkeys; n++) {
	307	key = ffs(keymask) - 1;
	308	keymask &= ~(1 << key);
6bd2a9af	309	keys[n] = flow_key_get(skb, key, &flow_keys);
e5dfb815 PM	310	}
	311
	312	if (f->mode == FLOW_MODE_HASH)
72d9794f	313	classid = jhash2(keys, f->nkeys, f->hashrnd);
e5dfb815 PM	314	else {
	315	classid = keys[0];
	316	classid = (classid & f->mask) ^ f->xor;
	317	classid = (classid >> f->rshift) + f->addend;
	318	}
	319
	320	if (f->divisor)
	321	classid %= f->divisor;
	322
	323	res->class = 0;
	324	res->classid = TC_H_MAKE(f->baseclass, f->baseclass + classid);
	325
	326	r = tcf_exts_exec(skb, &f->exts, res);
	327	if (r < 0)
	328	continue;
	329	return r;
	330	}
	331	return -1;
	332	}
	333
72d9794f PM	334	static void flow_perturbation(unsigned long arg)
	335	{
	336	struct flow_filter f = (struct flow_filter )arg;
	337
	338	get_random_bytes(&f->hashrnd, 4);
	339	if (f->perturb_period)
	340	mod_timer(&f->perturb_timer, jiffies + f->perturb_period);
	341	}
	342
e5dfb815 PM	343	static const struct nla_policy flow_policy[TCA_FLOW_MAX + 1] = {
	344	[TCA_FLOW_KEYS] = { .type = NLA_U32 },
	345	[TCA_FLOW_MODE] = { .type = NLA_U32 },
	346	[TCA_FLOW_BASECLASS] = { .type = NLA_U32 },
	347	[TCA_FLOW_RSHIFT] = { .type = NLA_U32 },
	348	[TCA_FLOW_ADDEND] = { .type = NLA_U32 },
	349	[TCA_FLOW_MASK] = { .type = NLA_U32 },
	350	[TCA_FLOW_XOR] = { .type = NLA_U32 },
	351	[TCA_FLOW_DIVISOR] = { .type = NLA_U32 },
	352	[TCA_FLOW_ACT] = { .type = NLA_NESTED },
	353	[TCA_FLOW_POLICE] = { .type = NLA_NESTED },
	354	[TCA_FLOW_EMATCHES] = { .type = NLA_NESTED },
72d9794f	355	[TCA_FLOW_PERTURB] = { .type = NLA_U32 },
e5dfb815 PM	356	};
e5dfb815 PM	357
70da9f0b JF	358	static void flow_destroy_filter(struct rcu_head *head)
	359	{
	360	struct flow_filter *f = container_of(head, struct flow_filter, rcu);
	361
	362	del_timer_sync(&f->perturb_timer);
18d0264f	363	tcf_exts_destroy(&f->exts);
82a470f1	364	tcf_em_tree_destroy(&f->ematches);
70da9f0b JF	365	kfree(f);
	366	}
	367
c1b52739	368	static int flow_change(struct net net, struct sk_buff in_skb,
af4c6641	369	struct tcf_proto *tp, unsigned long base,
e5dfb815	370	u32 handle, struct nlattr **tca,
2f7ef2f8	371	unsigned long *arg, bool ovr)
e5dfb815	372	{
70da9f0b JF	373	struct flow_head *head = rtnl_dereference(tp->root);
70da9f0b JF	374	struct flow_filter fold, fnew;
e5dfb815 PM	375	struct nlattr *opt = tca[TCA_OPTIONS];
	376	struct nlattr *tb[TCA_FLOW_MAX + 1];
	377	struct tcf_exts e;
	378	struct tcf_ematch_tree t;
	379	unsigned int nkeys = 0;
72d9794f	380	unsigned int perturb_period = 0;
e5dfb815 PM	381	u32 baseclass = 0;
	382	u32 keymask = 0;
	383	u32 mode;
	384	int err;
	385
	386	if (opt == NULL)
	387	return -EINVAL;
	388
	389	err = nla_parse_nested(tb, TCA_FLOW_MAX, opt, flow_policy);
	390	if (err < 0)
	391	return err;
	392
	393	if (tb[TCA_FLOW_BASECLASS]) {
	394	baseclass = nla_get_u32(tb[TCA_FLOW_BASECLASS]);
	395	if (TC_H_MIN(baseclass) == 0)
	396	return -EINVAL;
	397	}
	398
	399	if (tb[TCA_FLOW_KEYS]) {
	400	keymask = nla_get_u32(tb[TCA_FLOW_KEYS]);
e5dfb815 PM	401
	402	nkeys = hweight32(keymask);
	403	if (nkeys == 0)
	404	return -EINVAL;
4f250491 PM	405
	406	if (fls(keymask) - 1 > FLOW_KEY_MAX)
	407	return -EOPNOTSUPP;
a6c6796c EB	408
a6c6796c EB	409	if ((keymask & (FLOW_KEY_SKUID\|FLOW_KEY_SKGID)) &&
e32123e5	410	sk_user_ns(NETLINK_CB(in_skb).sk) != &init_user_ns)
a6c6796c	411	return -EOPNOTSUPP;
e5dfb815 PM	412	}
e5dfb815 PM	413
5da57f42	414	tcf_exts_init(&e, TCA_FLOW_ACT, TCA_FLOW_POLICE);
2f7ef2f8	415	err = tcf_exts_validate(net, tp, tb, tca[TCA_RATE], &e, ovr);
e5dfb815 PM	416	if (err < 0)
	417	return err;
	418
	419	err = tcf_em_tree_validate(tp, tb[TCA_FLOW_EMATCHES], &t);
	420	if (err < 0)
	421	goto err1;
	422
70da9f0b JF	423	err = -ENOBUFS;
	424	fnew = kzalloc(sizeof(*fnew), GFP_KERNEL);
	425	if (!fnew)
	426	goto err2;
	427
32b2f4b1 DB	428	tcf_exts_init(&fnew->exts, TCA_FLOW_ACT, TCA_FLOW_POLICE);
32b2f4b1 DB	429
70da9f0b JF	430	fold = (struct flow_filter )arg;
70da9f0b JF	431	if (fold) {
e5dfb815	432	err = -EINVAL;
70da9f0b	433	if (fold->handle != handle && handle)
e5dfb815 PM	434	goto err2;
e5dfb815 PM	435
70da9f0b	436	/* Copy fold into fnew */
70da9f0b	437	fnew->tp = fold->tp;
70da9f0b JF	438	fnew->handle = fold->handle;
	439	fnew->nkeys = fold->nkeys;
	440	fnew->keymask = fold->keymask;
	441	fnew->mode = fold->mode;
	442	fnew->mask = fold->mask;
	443	fnew->xor = fold->xor;
	444	fnew->rshift = fold->rshift;
	445	fnew->addend = fold->addend;
	446	fnew->divisor = fold->divisor;
	447	fnew->baseclass = fold->baseclass;
	448	fnew->hashrnd = fold->hashrnd;
	449
	450	mode = fold->mode;
e5dfb815 PM	451	if (tb[TCA_FLOW_MODE])
	452	mode = nla_get_u32(tb[TCA_FLOW_MODE]);
	453	if (mode != FLOW_MODE_HASH && nkeys > 1)
	454	goto err2;
72d9794f PM	455
72d9794f PM	456	if (mode == FLOW_MODE_HASH)
70da9f0b	457	perturb_period = fold->perturb_period;
72d9794f PM	458	if (tb[TCA_FLOW_PERTURB]) {
	459	if (mode != FLOW_MODE_HASH)
	460	goto err2;
	461	perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
	462	}
e5dfb815 PM	463	} else {
	464	err = -EINVAL;
	465	if (!handle)
	466	goto err2;
	467	if (!tb[TCA_FLOW_KEYS])
	468	goto err2;
	469
	470	mode = FLOW_MODE_MAP;
	471	if (tb[TCA_FLOW_MODE])
	472	mode = nla_get_u32(tb[TCA_FLOW_MODE]);
	473	if (mode != FLOW_MODE_HASH && nkeys > 1)
	474	goto err2;
	475
72d9794f PM	476	if (tb[TCA_FLOW_PERTURB]) {
	477	if (mode != FLOW_MODE_HASH)
	478	goto err2;
	479	perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
	480	}
	481
e5dfb815 PM	482	if (TC_H_MAJ(baseclass) == 0)
	483	baseclass = TC_H_MAKE(tp->q->handle, baseclass);
	484	if (TC_H_MIN(baseclass) == 0)
	485	baseclass = TC_H_MAKE(baseclass, 1);
	486
70da9f0b JF	487	fnew->handle = handle;
	488	fnew->mask = ~0U;
	489	fnew->tp = tp;
	490	get_random_bytes(&fnew->hashrnd, 4);
e5dfb815 PM	491	}
e5dfb815 PM	492
70da9f0b JF	493	fnew->perturb_timer.function = flow_perturbation;
	494	fnew->perturb_timer.data = (unsigned long)fnew;
	495	init_timer_deferrable(&fnew->perturb_timer);
e5dfb815	496
70da9f0b JF	497	tcf_exts_change(tp, &fnew->exts, &e);
70da9f0b JF	498	tcf_em_tree_change(tp, &fnew->ematches, &t);
e5dfb815	499
02875878 ED	500	netif_keep_dst(qdisc_dev(tp->q));
02875878 ED	501
e5dfb815	502	if (tb[TCA_FLOW_KEYS]) {
70da9f0b JF	503	fnew->keymask = keymask;
70da9f0b JF	504	fnew->nkeys = nkeys;
e5dfb815 PM	505	}
e5dfb815 PM	506
70da9f0b	507	fnew->mode = mode;
e5dfb815 PM	508
e5dfb815 PM	509	if (tb[TCA_FLOW_MASK])
70da9f0b	510	fnew->mask = nla_get_u32(tb[TCA_FLOW_MASK]);
e5dfb815	511	if (tb[TCA_FLOW_XOR])
70da9f0b	512	fnew->xor = nla_get_u32(tb[TCA_FLOW_XOR]);
e5dfb815	513	if (tb[TCA_FLOW_RSHIFT])
70da9f0b	514	fnew->rshift = nla_get_u32(tb[TCA_FLOW_RSHIFT]);
e5dfb815	515	if (tb[TCA_FLOW_ADDEND])
70da9f0b	516	fnew->addend = nla_get_u32(tb[TCA_FLOW_ADDEND]);
e5dfb815 PM	517
e5dfb815 PM	518	if (tb[TCA_FLOW_DIVISOR])
70da9f0b	519	fnew->divisor = nla_get_u32(tb[TCA_FLOW_DIVISOR]);
e5dfb815	520	if (baseclass)
70da9f0b	521	fnew->baseclass = baseclass;
e5dfb815	522
70da9f0b	523	fnew->perturb_period = perturb_period;
72d9794f	524	if (perturb_period)
70da9f0b	525	mod_timer(&fnew->perturb_timer, jiffies + perturb_period);
72d9794f	526
e5dfb815	527	if (*arg == 0)
70da9f0b JF	528	list_add_tail_rcu(&fnew->list, &head->filters);
70da9f0b JF	529	else
32b2f4b1	530	list_replace_rcu(&fold->list, &fnew->list);
e5dfb815	531
70da9f0b	532	*arg = (unsigned long)fnew;
e5dfb815	533
70da9f0b JF	534	if (fold)
70da9f0b JF	535	call_rcu(&fold->rcu, flow_destroy_filter);
e5dfb815 PM	536	return 0;
	537
	538	err2:
82a470f1	539	tcf_em_tree_destroy(&t);
70da9f0b	540	kfree(fnew);
e5dfb815	541	err1:
18d0264f	542	tcf_exts_destroy(&e);
e5dfb815 PM	543	return err;
	544	}
	545
e5dfb815 PM	546	static int flow_delete(struct tcf_proto *tp, unsigned long arg)
	547	{
	548	struct flow_filter f = (struct flow_filter )arg;
	549
70da9f0b JF	550	list_del_rcu(&f->list);
70da9f0b JF	551	call_rcu(&f->rcu, flow_destroy_filter);
e5dfb815 PM	552	return 0;
	553	}
	554
	555	static int flow_init(struct tcf_proto *tp)
	556	{
	557	struct flow_head *head;
	558
e5dfb815 PM	559	head = kzalloc(sizeof(*head), GFP_KERNEL);
	560	if (head == NULL)
	561	return -ENOBUFS;
	562	INIT_LIST_HEAD(&head->filters);
70da9f0b	563	rcu_assign_pointer(tp->root, head);
e5dfb815 PM	564	return 0;
	565	}
	566
1e052be6	567	static bool flow_destroy(struct tcf_proto *tp, bool force)
e5dfb815	568	{
70da9f0b	569	struct flow_head *head = rtnl_dereference(tp->root);
e5dfb815 PM	570	struct flow_filter f, next;
e5dfb815 PM	571
1e052be6 CW	572	if (!force && !list_empty(&head->filters))
	573	return false;
	574
e5dfb815	575	list_for_each_entry_safe(f, next, &head->filters, list) {
70da9f0b JF	576	list_del_rcu(&f->list);
70da9f0b JF	577	call_rcu(&f->rcu, flow_destroy_filter);
e5dfb815	578	}
70da9f0b JF	579	RCU_INIT_POINTER(tp->root, NULL);
70da9f0b JF	580	kfree_rcu(head, rcu);
1e052be6	581	return true;
e5dfb815 PM	582	}
	583
	584	static unsigned long flow_get(struct tcf_proto *tp, u32 handle)
	585	{
70da9f0b	586	struct flow_head *head = rtnl_dereference(tp->root);
e5dfb815 PM	587	struct flow_filter *f;
e5dfb815 PM	588
6a659cd0	589	list_for_each_entry(f, &head->filters, list)
e5dfb815 PM	590	if (f->handle == handle)
	591	return (unsigned long)f;
	592	return 0;
	593	}
	594
832d1d5b	595	static int flow_dump(struct net net, struct tcf_proto tp, unsigned long fh,
e5dfb815 PM	596	struct sk_buff skb, struct tcmsg t)
	597	{
	598	struct flow_filter f = (struct flow_filter )fh;
	599	struct nlattr *nest;
	600
	601	if (f == NULL)
	602	return skb->len;
	603
	604	t->tcm_handle = f->handle;
	605
	606	nest = nla_nest_start(skb, TCA_OPTIONS);
	607	if (nest == NULL)
	608	goto nla_put_failure;
	609
1b34ec43 DM	610	if (nla_put_u32(skb, TCA_FLOW_KEYS, f->keymask) \|\|
	611	nla_put_u32(skb, TCA_FLOW_MODE, f->mode))
	612	goto nla_put_failure;
e5dfb815 PM	613
e5dfb815 PM	614	if (f->mask != ~0 \|\| f->xor != 0) {
1b34ec43 DM	615	if (nla_put_u32(skb, TCA_FLOW_MASK, f->mask) \|\|
	616	nla_put_u32(skb, TCA_FLOW_XOR, f->xor))
	617	goto nla_put_failure;
e5dfb815	618	}
1b34ec43 DM	619	if (f->rshift &&
	620	nla_put_u32(skb, TCA_FLOW_RSHIFT, f->rshift))
	621	goto nla_put_failure;
	622	if (f->addend &&
	623	nla_put_u32(skb, TCA_FLOW_ADDEND, f->addend))
	624	goto nla_put_failure;
e5dfb815	625
1b34ec43 DM	626	if (f->divisor &&
	627	nla_put_u32(skb, TCA_FLOW_DIVISOR, f->divisor))
	628	goto nla_put_failure;
	629	if (f->baseclass &&
	630	nla_put_u32(skb, TCA_FLOW_BASECLASS, f->baseclass))
	631	goto nla_put_failure;
e5dfb815	632
1b34ec43 DM	633	if (f->perturb_period &&
	634	nla_put_u32(skb, TCA_FLOW_PERTURB, f->perturb_period / HZ))
	635	goto nla_put_failure;
72d9794f	636
5da57f42	637	if (tcf_exts_dump(skb, &f->exts) < 0)
e5dfb815	638	goto nla_put_failure;
0aead543	639	#ifdef CONFIG_NET_EMATCH
e5dfb815 PM	640	if (f->ematches.hdr.nmatches &&
	641	tcf_em_tree_dump(skb, &f->ematches, TCA_FLOW_EMATCHES) < 0)
	642	goto nla_put_failure;
0aead543	643	#endif
e5dfb815 PM	644	nla_nest_end(skb, nest);
e5dfb815 PM	645
5da57f42	646	if (tcf_exts_dump_stats(skb, &f->exts) < 0)
e5dfb815 PM	647	goto nla_put_failure;
	648
	649	return skb->len;
	650
	651	nla_put_failure:
6ea3b446	652	nla_nest_cancel(skb, nest);
e5dfb815 PM	653	return -1;
	654	}
	655
	656	static void flow_walk(struct tcf_proto tp, struct tcf_walker arg)
	657	{
70da9f0b	658	struct flow_head *head = rtnl_dereference(tp->root);
e5dfb815 PM	659	struct flow_filter *f;
e5dfb815 PM	660
6a659cd0	661	list_for_each_entry(f, &head->filters, list) {
e5dfb815 PM	662	if (arg->count < arg->skip)
	663	goto skip;
	664	if (arg->fn(tp, (unsigned long)f, arg) < 0) {
	665	arg->stop = 1;
	666	break;
	667	}
	668	skip:
	669	arg->count++;
	670	}
	671	}
	672
	673	static struct tcf_proto_ops cls_flow_ops __read_mostly = {
	674	.kind = "flow",
	675	.classify = flow_classify,
	676	.init = flow_init,
	677	.destroy = flow_destroy,
	678	.change = flow_change,
	679	.delete = flow_delete,
	680	.get = flow_get,
e5dfb815 PM	681	.dump = flow_dump,
	682	.walk = flow_walk,
	683	.owner = THIS_MODULE,
	684	};
	685
	686	static int __init cls_flow_init(void)
	687	{
	688	return register_tcf_proto_ops(&cls_flow_ops);
	689	}
	690
	691	static void __exit cls_flow_exit(void)
	692	{
	693	unregister_tcf_proto_ops(&cls_flow_ops);
	694	}
	695
	696	module_init(cls_flow_init);
	697	module_exit(cls_flow_exit);
	698
	699	MODULE_LICENSE("GPL");
	700	MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
	701	MODULE_DESCRIPTION("TC flow classifier");