[linux.git] / net / sched / cls_flow.c

/*
 * net/sched/cls_flow.c		Generic flow classifier
 *
 * Copyright (c) 2007, 2008 Patrick McHardy <[email protected]>
 *
 * 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 <net/pkt_cls.h>
#include <net/ip.h>
#include <net/route.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 const struct tcf_ext_map flow_ext_map = {
	.action	= TCA_FLOW_ACT,
	.police	= TCA_FLOW_POLICE,
};

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)
{
	switch (skb->protocol) {
	case htons(ETH_P_IP):
		return ntohl(ip_hdr(skb)->saddr);
	case htons(ETH_P_IPV6):
		return ntohl(ipv6_hdr(skb)->saddr.s6_addr32[3]);
	default:
		return addr_fold(skb->sk);
	}
}

static u32 flow_get_dst(const struct sk_buff *skb)
{
	switch (skb->protocol) {
	case htons(ETH_P_IP):
		return ntohl(ip_hdr(skb)->daddr);
	case htons(ETH_P_IPV6):
		return ntohl(ipv6_hdr(skb)->daddr.s6_addr32[3]);
	default:
		return addr_fold(skb->dst) ^ (__force u16)skb->protocol;
	}
}

static u32 flow_get_proto(const struct sk_buff *skb)
{
	switch (skb->protocol) {
	case htons(ETH_P_IP):
		return ip_hdr(skb)->protocol;
	case htons(ETH_P_IPV6):
		return ipv6_hdr(skb)->nexthdr;
	default:
		return 0;
	}
}

static int has_ports(u8 protocol)
{
	switch (protocol) {
	case IPPROTO_TCP:
	case IPPROTO_UDP:
	case IPPROTO_UDPLITE:
	case IPPROTO_SCTP:
	case IPPROTO_DCCP:
	case IPPROTO_ESP:
		return 1;
	default:
		return 0;
	}
}

static u32 flow_get_proto_src(const struct sk_buff *skb)
{
	u32 res = 0;

	switch (skb->protocol) {
	case htons(ETH_P_IP): {
		struct iphdr *iph = ip_hdr(skb);

		if (!(iph->frag_off&htons(IP_MF|IP_OFFSET)) &&
		    has_ports(iph->protocol))
			res = ntohs(*(__be16 *)((void *)iph + iph->ihl * 4));
		break;
	}
	case htons(ETH_P_IPV6): {
		struct ipv6hdr *iph = ipv6_hdr(skb);

		if (has_ports(iph->nexthdr))
			res = ntohs(*(__be16 *)&iph[1]);
		break;
	}
	default:
		res = addr_fold(skb->sk);
	}

	return res;
}

static u32 flow_get_proto_dst(const struct sk_buff *skb)
{
	u32 res = 0;

	switch (skb->protocol) {
	case htons(ETH_P_IP): {
		struct iphdr *iph = ip_hdr(skb);

		if (!(iph->frag_off&htons(IP_MF|IP_OFFSET)) &&
		    has_ports(iph->protocol))
			res = ntohs(*(__be16 *)((void *)iph + iph->ihl * 4 + 2));
		break;
	}
	case htons(ETH_P_IPV6): {
		struct ipv6hdr *iph = ipv6_hdr(skb);

		if (has_ports(iph->nexthdr))
			res = ntohs(*(__be16 *)((void *)&iph[1] + 2));
		break;
	}
	default:
		res = addr_fold(skb->dst) ^ (__force u16)skb->protocol;
	}

	return res;
}

static u32 flow_get_iif(const struct sk_buff *skb)
{
	return 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;					\
	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)
{
	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);
}

static u32 flow_get_nfct_dst(const struct sk_buff *skb)
{
	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);
}

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

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

static u32 flow_get_rtclassid(const struct sk_buff *skb)
{
#ifdef CONFIG_NET_CLS_ROUTE
	if (skb->dst)
		return skb->dst->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)
		return skb->sk->sk_socket->file->f_cred->fsuid;
	return 0;
}

static u32 flow_get_skgid(const struct sk_buff *skb)
{
	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file)
		return skb->sk->sk_socket->file->f_cred->fsgid;
	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_key_get(const struct sk_buff *skb, int key)
{
	switch (key) {
	case FLOW_KEY_SRC:
		return flow_get_src(skb);
	case FLOW_KEY_DST:
		return flow_get_dst(skb);
	case FLOW_KEY_PROTO:
		return flow_get_proto(skb);
	case FLOW_KEY_PROTO_SRC:
		return flow_get_proto_src(skb);
	case FLOW_KEY_PROTO_DST:
		return flow_get_proto_dst(skb);
	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);
	case FLOW_KEY_NFCT_DST:
		return flow_get_nfct_dst(skb);
	case FLOW_KEY_NFCT_PROTO_SRC:
		return flow_get_nfct_proto_src(skb);
	case FLOW_KEY_NFCT_PROTO_DST:
		return flow_get_nfct_proto_dst(skb);
	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);
	default:
		WARN_ON(1);
		return 0;
	}
}

static int flow_classify(struct sk_buff *skb, 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[f->nkeys];

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

		keymask = f->keymask;

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

		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 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;
	}

	err = tcf_exts_validate(tp, tb, tca[TCA_RATE], &e, &flow_ext_map);
	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;

		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)
{
	return;
}

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;

	NLA_PUT_U32(skb, TCA_FLOW_KEYS, f->keymask);
	NLA_PUT_U32(skb, TCA_FLOW_MODE, f->mode);

	if (f->mask != ~0 || f->xor != 0) {
		NLA_PUT_U32(skb, TCA_FLOW_MASK, f->mask);
		NLA_PUT_U32(skb, TCA_FLOW_XOR, f->xor);
	}
	if (f->rshift)
		NLA_PUT_U32(skb, TCA_FLOW_RSHIFT, f->rshift);
	if (f->addend)
		NLA_PUT_U32(skb, TCA_FLOW_ADDEND, f->addend);

	if (f->divisor)
		NLA_PUT_U32(skb, TCA_FLOW_DIVISOR, f->divisor);
	if (f->baseclass)
		NLA_PUT_U32(skb, TCA_FLOW_BASECLASS, f->baseclass);

	if (f->perturb_period)
		NLA_PUT_U32(skb, TCA_FLOW_PERTURB, f->perturb_period / HZ);

	if (tcf_exts_dump(skb, &f->exts, &flow_ext_map) < 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, &flow_ext_map) < 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 <[email protected]>");
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 <[email protected]>
	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>
e5dfb815 PM	23
	24	#include <net/pkt_cls.h>
	25	#include <net/ip.h>
	26	#include <net/route.h>
	27	#if defined(CONFIG_NF_CONNTRACK) \|\| defined(CONFIG_NF_CONNTRACK_MODULE)
	28	#include <net/netfilter/nf_conntrack.h>
	29	#endif
	30
	31	struct flow_head {
	32	struct list_head filters;
	33	};
	34
	35	struct flow_filter {
	36	struct list_head list;
	37	struct tcf_exts exts;
	38	struct tcf_ematch_tree ematches;
72d9794f PM	39	struct timer_list perturb_timer;
72d9794f PM	40	u32 perturb_period;
e5dfb815 PM	41	u32 handle;
	42
	43	u32 nkeys;
	44	u32 keymask;
	45	u32 mode;
	46	u32 mask;
	47	u32 xor;
	48	u32 rshift;
	49	u32 addend;
	50	u32 divisor;
	51	u32 baseclass;
72d9794f	52	u32 hashrnd;
e5dfb815 PM	53	};
e5dfb815 PM	54
e5dfb815 PM	55	static const struct tcf_ext_map flow_ext_map = {
	56	.action = TCA_FLOW_ACT,
	57	.police = TCA_FLOW_POLICE,
	58	};
	59
	60	static inline u32 addr_fold(void *addr)
	61	{
	62	unsigned long a = (unsigned long)addr;
	63
	64	return (a & 0xFFFFFFFF) ^ (BITS_PER_LONG > 32 ? a >> 32 : 0);
	65	}
	66
	67	static u32 flow_get_src(const struct sk_buff *skb)
	68	{
	69	switch (skb->protocol) {
60678040	70	case htons(ETH_P_IP):
e5dfb815	71	return ntohl(ip_hdr(skb)->saddr);
60678040	72	case htons(ETH_P_IPV6):
e5dfb815 PM	73	return ntohl(ipv6_hdr(skb)->saddr.s6_addr32[3]);
	74	default:
	75	return addr_fold(skb->sk);
	76	}
	77	}
	78
	79	static u32 flow_get_dst(const struct sk_buff *skb)
	80	{
	81	switch (skb->protocol) {
60678040	82	case htons(ETH_P_IP):
e5dfb815	83	return ntohl(ip_hdr(skb)->daddr);
60678040	84	case htons(ETH_P_IPV6):
e5dfb815 PM	85	return ntohl(ipv6_hdr(skb)->daddr.s6_addr32[3]);
	86	default:
	87	return addr_fold(skb->dst) ^ (__force u16)skb->protocol;
	88	}
	89	}
	90
	91	static u32 flow_get_proto(const struct sk_buff *skb)
	92	{
	93	switch (skb->protocol) {
60678040	94	case htons(ETH_P_IP):
e5dfb815	95	return ip_hdr(skb)->protocol;
60678040	96	case htons(ETH_P_IPV6):
e5dfb815 PM	97	return ipv6_hdr(skb)->nexthdr;
	98	default:
	99	return 0;
	100	}
	101	}
	102
	103	static int has_ports(u8 protocol)
	104	{
	105	switch (protocol) {
	106	case IPPROTO_TCP:
	107	case IPPROTO_UDP:
	108	case IPPROTO_UDPLITE:
	109	case IPPROTO_SCTP:
	110	case IPPROTO_DCCP:
	111	case IPPROTO_ESP:
	112	return 1;
	113	default:
	114	return 0;
	115	}
	116	}
	117
	118	static u32 flow_get_proto_src(const struct sk_buff *skb)
	119	{
	120	u32 res = 0;
	121
	122	switch (skb->protocol) {
60678040	123	case htons(ETH_P_IP): {
e5dfb815 PM	124	struct iphdr *iph = ip_hdr(skb);
	125
	126	if (!(iph->frag_off&htons(IP_MF\|IP_OFFSET)) &&
	127	has_ports(iph->protocol))
	128	res = ntohs((__be16 )((void )iph + iph->ihl 4));
	129	break;
	130	}
60678040	131	case htons(ETH_P_IPV6): {
e5dfb815 PM	132	struct ipv6hdr *iph = ipv6_hdr(skb);
	133
	134	if (has_ports(iph->nexthdr))
	135	res = ntohs((__be16 )&iph[1]);
	136	break;
	137	}
	138	default:
	139	res = addr_fold(skb->sk);
	140	}
	141
	142	return res;
	143	}
	144
	145	static u32 flow_get_proto_dst(const struct sk_buff *skb)
	146	{
	147	u32 res = 0;
	148
	149	switch (skb->protocol) {
60678040	150	case htons(ETH_P_IP): {
e5dfb815 PM	151	struct iphdr *iph = ip_hdr(skb);
	152
	153	if (!(iph->frag_off&htons(IP_MF\|IP_OFFSET)) &&
	154	has_ports(iph->protocol))
	155	res = ntohs((__be16 )((void )iph + iph->ihl 4 + 2));
	156	break;
	157	}
60678040	158	case htons(ETH_P_IPV6): {
e5dfb815 PM	159	struct ipv6hdr *iph = ipv6_hdr(skb);
	160
	161	if (has_ports(iph->nexthdr))
	162	res = ntohs((__be16 )((void *)&iph[1] + 2));
	163	break;
	164	}
	165	default:
	166	res = addr_fold(skb->dst) ^ (__force u16)skb->protocol;
	167	}
	168
	169	return res;
	170	}
	171
	172	static u32 flow_get_iif(const struct sk_buff *skb)
	173	{
	174	return skb->iif;
	175	}
	176
	177	static u32 flow_get_priority(const struct sk_buff *skb)
	178	{
	179	return skb->priority;
	180	}
	181
	182	static u32 flow_get_mark(const struct sk_buff *skb)
	183	{
	184	return skb->mark;
	185	}
	186
	187	static u32 flow_get_nfct(const struct sk_buff *skb)
	188	{
	189	#if defined(CONFIG_NF_CONNTRACK) \|\| defined(CONFIG_NF_CONNTRACK_MODULE)
	190	return addr_fold(skb->nfct);
	191	#else
	192	return 0;
	193	#endif
	194	}
	195
	196	#if defined(CONFIG_NF_CONNTRACK) \|\| defined(CONFIG_NF_CONNTRACK_MODULE)
	197	#define CTTUPLE(skb, member) \
	198	({ \
	199	enum ip_conntrack_info ctinfo; \
	200	struct nf_conn *ct = nf_ct_get(skb, &ctinfo); \
	201	if (ct == NULL) \
	202	goto fallback; \
	203	ct->tuplehash[CTINFO2DIR(ctinfo)].tuple.member; \
	204	})
	205	#else
	206	#define CTTUPLE(skb, member) \
	207	({ \
	208	goto fallback; \
	209	0; \
	210	})
	211	#endif
	212
	213	static u32 flow_get_nfct_src(const struct sk_buff *skb)
	214	{
	215	switch (skb->protocol) {
60678040	216	case htons(ETH_P_IP):
e5dfb815	217	return ntohl(CTTUPLE(skb, src.u3.ip));
60678040	218	case htons(ETH_P_IPV6):
e5dfb815 PM	219	return ntohl(CTTUPLE(skb, src.u3.ip6[3]));
	220	}
	221	fallback:
	222	return flow_get_src(skb);
	223	}
	224
	225	static u32 flow_get_nfct_dst(const struct sk_buff *skb)
	226	{
	227	switch (skb->protocol) {
60678040	228	case htons(ETH_P_IP):
e5dfb815	229	return ntohl(CTTUPLE(skb, dst.u3.ip));
60678040	230	case htons(ETH_P_IPV6):
e5dfb815 PM	231	return ntohl(CTTUPLE(skb, dst.u3.ip6[3]));
	232	}
	233	fallback:
	234	return flow_get_dst(skb);
	235	}
	236
	237	static u32 flow_get_nfct_proto_src(const struct sk_buff *skb)
	238	{
	239	return ntohs(CTTUPLE(skb, src.u.all));
	240	fallback:
	241	return flow_get_proto_src(skb);
	242	}
	243
	244	static u32 flow_get_nfct_proto_dst(const struct sk_buff *skb)
	245	{
	246	return ntohs(CTTUPLE(skb, dst.u.all));
	247	fallback:
	248	return flow_get_proto_dst(skb);
	249	}
	250
	251	static u32 flow_get_rtclassid(const struct sk_buff *skb)
	252	{
	253	#ifdef CONFIG_NET_CLS_ROUTE
	254	if (skb->dst)
	255	return skb->dst->tclassid;
	256	#endif
	257	return 0;
	258	}
	259
	260	static u32 flow_get_skuid(const struct sk_buff *skb)
	261	{
	262	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file)
d76b0d9b	263	return skb->sk->sk_socket->file->f_cred->fsuid;
e5dfb815 PM	264	return 0;
	265	}
	266
	267	static u32 flow_get_skgid(const struct sk_buff *skb)
	268	{
	269	if (skb->sk && skb->sk->sk_socket && skb->sk->sk_socket->file)
d76b0d9b	270	return skb->sk->sk_socket->file->f_cred->fsgid;
e5dfb815 PM	271	return 0;
	272	}
	273
9ec13810 PM	274	static u32 flow_get_vlan_tag(const struct sk_buff *skb)
	275	{
	276	u16 uninitialized_var(tag);
	277
	278	if (vlan_get_tag(skb, &tag) < 0)
	279	return 0;
	280	return tag & VLAN_VID_MASK;
	281	}
	282
e5dfb815 PM	283	static u32 flow_key_get(const struct sk_buff *skb, int key)
	284	{
	285	switch (key) {
	286	case FLOW_KEY_SRC:
	287	return flow_get_src(skb);
	288	case FLOW_KEY_DST:
	289	return flow_get_dst(skb);
	290	case FLOW_KEY_PROTO:
	291	return flow_get_proto(skb);
	292	case FLOW_KEY_PROTO_SRC:
	293	return flow_get_proto_src(skb);
	294	case FLOW_KEY_PROTO_DST:
	295	return flow_get_proto_dst(skb);
	296	case FLOW_KEY_IIF:
	297	return flow_get_iif(skb);
	298	case FLOW_KEY_PRIORITY:
	299	return flow_get_priority(skb);
	300	case FLOW_KEY_MARK:
	301	return flow_get_mark(skb);
	302	case FLOW_KEY_NFCT:
	303	return flow_get_nfct(skb);
	304	case FLOW_KEY_NFCT_SRC:
	305	return flow_get_nfct_src(skb);
	306	case FLOW_KEY_NFCT_DST:
	307	return flow_get_nfct_dst(skb);
	308	case FLOW_KEY_NFCT_PROTO_SRC:
	309	return flow_get_nfct_proto_src(skb);
	310	case FLOW_KEY_NFCT_PROTO_DST:
	311	return flow_get_nfct_proto_dst(skb);
	312	case FLOW_KEY_RTCLASSID:
	313	return flow_get_rtclassid(skb);
	314	case FLOW_KEY_SKUID:
	315	return flow_get_skuid(skb);
	316	case FLOW_KEY_SKGID:
	317	return flow_get_skgid(skb);
9ec13810 PM	318	case FLOW_KEY_VLAN_TAG:
9ec13810 PM	319	return flow_get_vlan_tag(skb);
e5dfb815 PM	320	default:
	321	WARN_ON(1);
	322	return 0;
	323	}
	324	}
	325
	326	static int flow_classify(struct sk_buff skb, struct tcf_proto tp,
	327	struct tcf_result *res)
	328	{
	329	struct flow_head *head = tp->root;
	330	struct flow_filter *f;
	331	u32 keymask;
	332	u32 classid;
	333	unsigned int n, key;
	334	int r;
	335
	336	list_for_each_entry(f, &head->filters, list) {
	337	u32 keys[f->nkeys];
	338
	339	if (!tcf_em_tree_match(skb, &f->ematches, NULL))
	340	continue;
	341
	342	keymask = f->keymask;
	343
	344	for (n = 0; n < f->nkeys; n++) {
	345	key = ffs(keymask) - 1;
	346	keymask &= ~(1 << key);
	347	keys[n] = flow_key_get(skb, key);
	348	}
	349
	350	if (f->mode == FLOW_MODE_HASH)
72d9794f	351	classid = jhash2(keys, f->nkeys, f->hashrnd);
e5dfb815 PM	352	else {
	353	classid = keys[0];
	354	classid = (classid & f->mask) ^ f->xor;
	355	classid = (classid >> f->rshift) + f->addend;
	356	}
	357
	358	if (f->divisor)
	359	classid %= f->divisor;
	360
	361	res->class = 0;
	362	res->classid = TC_H_MAKE(f->baseclass, f->baseclass + classid);
	363
	364	r = tcf_exts_exec(skb, &f->exts, res);
	365	if (r < 0)
	366	continue;
	367	return r;
	368	}
	369	return -1;
	370	}
	371
72d9794f PM	372	static void flow_perturbation(unsigned long arg)
	373	{
	374	struct flow_filter f = (struct flow_filter )arg;
	375
	376	get_random_bytes(&f->hashrnd, 4);
	377	if (f->perturb_period)
	378	mod_timer(&f->perturb_timer, jiffies + f->perturb_period);
	379	}
	380
e5dfb815 PM	381	static const struct nla_policy flow_policy[TCA_FLOW_MAX + 1] = {
	382	[TCA_FLOW_KEYS] = { .type = NLA_U32 },
	383	[TCA_FLOW_MODE] = { .type = NLA_U32 },
	384	[TCA_FLOW_BASECLASS] = { .type = NLA_U32 },
	385	[TCA_FLOW_RSHIFT] = { .type = NLA_U32 },
	386	[TCA_FLOW_ADDEND] = { .type = NLA_U32 },
	387	[TCA_FLOW_MASK] = { .type = NLA_U32 },
	388	[TCA_FLOW_XOR] = { .type = NLA_U32 },
	389	[TCA_FLOW_DIVISOR] = { .type = NLA_U32 },
	390	[TCA_FLOW_ACT] = { .type = NLA_NESTED },
	391	[TCA_FLOW_POLICE] = { .type = NLA_NESTED },
	392	[TCA_FLOW_EMATCHES] = { .type = NLA_NESTED },
72d9794f	393	[TCA_FLOW_PERTURB] = { .type = NLA_U32 },
e5dfb815 PM	394	};
	395
	396	static int flow_change(struct tcf_proto *tp, unsigned long base,
	397	u32 handle, struct nlattr **tca,
	398	unsigned long *arg)
	399	{
	400	struct flow_head *head = tp->root;
	401	struct flow_filter *f;
	402	struct nlattr *opt = tca[TCA_OPTIONS];
	403	struct nlattr *tb[TCA_FLOW_MAX + 1];
	404	struct tcf_exts e;
	405	struct tcf_ematch_tree t;
	406	unsigned int nkeys = 0;
72d9794f	407	unsigned int perturb_period = 0;
e5dfb815 PM	408	u32 baseclass = 0;
	409	u32 keymask = 0;
	410	u32 mode;
	411	int err;
	412
	413	if (opt == NULL)
	414	return -EINVAL;
	415
	416	err = nla_parse_nested(tb, TCA_FLOW_MAX, opt, flow_policy);
	417	if (err < 0)
	418	return err;
	419
	420	if (tb[TCA_FLOW_BASECLASS]) {
	421	baseclass = nla_get_u32(tb[TCA_FLOW_BASECLASS]);
	422	if (TC_H_MIN(baseclass) == 0)
	423	return -EINVAL;
	424	}
	425
	426	if (tb[TCA_FLOW_KEYS]) {
	427	keymask = nla_get_u32(tb[TCA_FLOW_KEYS]);
e5dfb815 PM	428
	429	nkeys = hweight32(keymask);
	430	if (nkeys == 0)
	431	return -EINVAL;
4f250491 PM	432
	433	if (fls(keymask) - 1 > FLOW_KEY_MAX)
	434	return -EOPNOTSUPP;
e5dfb815 PM	435	}
	436
	437	err = tcf_exts_validate(tp, tb, tca[TCA_RATE], &e, &flow_ext_map);
	438	if (err < 0)
	439	return err;
	440
	441	err = tcf_em_tree_validate(tp, tb[TCA_FLOW_EMATCHES], &t);
	442	if (err < 0)
	443	goto err1;
	444
	445	f = (struct flow_filter )arg;
	446	if (f != NULL) {
	447	err = -EINVAL;
	448	if (f->handle != handle && handle)
	449	goto err2;
	450
	451	mode = f->mode;
	452	if (tb[TCA_FLOW_MODE])
	453	mode = nla_get_u32(tb[TCA_FLOW_MODE]);
	454	if (mode != FLOW_MODE_HASH && nkeys > 1)
	455	goto err2;
72d9794f PM	456
	457	if (mode == FLOW_MODE_HASH)
	458	perturb_period = f->perturb_period;
	459	if (tb[TCA_FLOW_PERTURB]) {
	460	if (mode != FLOW_MODE_HASH)
	461	goto err2;
	462	perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
	463	}
e5dfb815 PM	464	} else {
	465	err = -EINVAL;
	466	if (!handle)
	467	goto err2;
	468	if (!tb[TCA_FLOW_KEYS])
	469	goto err2;
	470
	471	mode = FLOW_MODE_MAP;
	472	if (tb[TCA_FLOW_MODE])
	473	mode = nla_get_u32(tb[TCA_FLOW_MODE]);
	474	if (mode != FLOW_MODE_HASH && nkeys > 1)
	475	goto err2;
	476
72d9794f PM	477	if (tb[TCA_FLOW_PERTURB]) {
	478	if (mode != FLOW_MODE_HASH)
	479	goto err2;
	480	perturb_period = nla_get_u32(tb[TCA_FLOW_PERTURB]) * HZ;
	481	}
	482
e5dfb815 PM	483	if (TC_H_MAJ(baseclass) == 0)
	484	baseclass = TC_H_MAKE(tp->q->handle, baseclass);
	485	if (TC_H_MIN(baseclass) == 0)
	486	baseclass = TC_H_MAKE(baseclass, 1);
	487
	488	err = -ENOBUFS;
	489	f = kzalloc(sizeof(*f), GFP_KERNEL);
	490	if (f == NULL)
	491	goto err2;
	492
	493	f->handle = handle;
	494	f->mask = ~0U;
72d9794f PM	495
	496	get_random_bytes(&f->hashrnd, 4);
	497	f->perturb_timer.function = flow_perturbation;
	498	f->perturb_timer.data = (unsigned long)f;
	499	init_timer_deferrable(&f->perturb_timer);
e5dfb815 PM	500	}
	501
	502	tcf_exts_change(tp, &f->exts, &e);
	503	tcf_em_tree_change(tp, &f->ematches, &t);
	504
	505	tcf_tree_lock(tp);
	506
	507	if (tb[TCA_FLOW_KEYS]) {
	508	f->keymask = keymask;
	509	f->nkeys = nkeys;
	510	}
	511
	512	f->mode = mode;
	513
	514	if (tb[TCA_FLOW_MASK])
	515	f->mask = nla_get_u32(tb[TCA_FLOW_MASK]);
	516	if (tb[TCA_FLOW_XOR])
	517	f->xor = nla_get_u32(tb[TCA_FLOW_XOR]);
	518	if (tb[TCA_FLOW_RSHIFT])
	519	f->rshift = nla_get_u32(tb[TCA_FLOW_RSHIFT]);
	520	if (tb[TCA_FLOW_ADDEND])
	521	f->addend = nla_get_u32(tb[TCA_FLOW_ADDEND]);
	522
	523	if (tb[TCA_FLOW_DIVISOR])
	524	f->divisor = nla_get_u32(tb[TCA_FLOW_DIVISOR]);
	525	if (baseclass)
	526	f->baseclass = baseclass;
	527
72d9794f PM	528	f->perturb_period = perturb_period;
	529	del_timer(&f->perturb_timer);
	530	if (perturb_period)
	531	mod_timer(&f->perturb_timer, jiffies + perturb_period);
	532
e5dfb815 PM	533	if (*arg == 0)
	534	list_add_tail(&f->list, &head->filters);
	535
	536	tcf_tree_unlock(tp);
	537
	538	*arg = (unsigned long)f;
	539	return 0;
	540
	541	err2:
	542	tcf_em_tree_destroy(tp, &t);
	543	err1:
	544	tcf_exts_destroy(tp, &e);
	545	return err;
	546	}
	547
	548	static void flow_destroy_filter(struct tcf_proto tp, struct flow_filter f)
	549	{
72d9794f	550	del_timer_sync(&f->perturb_timer);
e5dfb815 PM	551	tcf_exts_destroy(tp, &f->exts);
	552	tcf_em_tree_destroy(tp, &f->ematches);
	553	kfree(f);
	554	}
	555
	556	static int flow_delete(struct tcf_proto *tp, unsigned long arg)
	557	{
	558	struct flow_filter f = (struct flow_filter )arg;
	559
	560	tcf_tree_lock(tp);
	561	list_del(&f->list);
	562	tcf_tree_unlock(tp);
	563	flow_destroy_filter(tp, f);
	564	return 0;
	565	}
	566
	567	static int flow_init(struct tcf_proto *tp)
	568	{
	569	struct flow_head *head;
	570
e5dfb815 PM	571	head = kzalloc(sizeof(*head), GFP_KERNEL);
	572	if (head == NULL)
	573	return -ENOBUFS;
	574	INIT_LIST_HEAD(&head->filters);
	575	tp->root = head;
	576	return 0;
	577	}
	578
	579	static void flow_destroy(struct tcf_proto *tp)
	580	{
	581	struct flow_head *head = tp->root;
	582	struct flow_filter f, next;
	583
	584	list_for_each_entry_safe(f, next, &head->filters, list) {
	585	list_del(&f->list);
	586	flow_destroy_filter(tp, f);
	587	}
	588	kfree(head);
	589	}
	590
	591	static unsigned long flow_get(struct tcf_proto *tp, u32 handle)
	592	{
	593	struct flow_head *head = tp->root;
	594	struct flow_filter *f;
	595
	596	list_for_each_entry(f, &head->filters, list)
	597	if (f->handle == handle)
	598	return (unsigned long)f;
	599	return 0;
	600	}
	601
	602	static void flow_put(struct tcf_proto *tp, unsigned long f)
	603	{
	604	return;
	605	}
	606
	607	static int flow_dump(struct tcf_proto *tp, unsigned long fh,
	608	struct sk_buff skb, struct tcmsg t)
	609	{
	610	struct flow_filter f = (struct flow_filter )fh;
	611	struct nlattr *nest;
	612
	613	if (f == NULL)
	614	return skb->len;
	615
	616	t->tcm_handle = f->handle;
	617
	618	nest = nla_nest_start(skb, TCA_OPTIONS);
	619	if (nest == NULL)
	620	goto nla_put_failure;
	621
	622	NLA_PUT_U32(skb, TCA_FLOW_KEYS, f->keymask);
	623	NLA_PUT_U32(skb, TCA_FLOW_MODE, f->mode);
	624
	625	if (f->mask != ~0 \|\| f->xor != 0) {
	626	NLA_PUT_U32(skb, TCA_FLOW_MASK, f->mask);
	627	NLA_PUT_U32(skb, TCA_FLOW_XOR, f->xor);
	628	}
	629	if (f->rshift)
	630	NLA_PUT_U32(skb, TCA_FLOW_RSHIFT, f->rshift);
	631	if (f->addend)
	632	NLA_PUT_U32(skb, TCA_FLOW_ADDEND, f->addend);
	633
	634	if (f->divisor)
635	NLA_PUT_U32(skb, TCA_FLOW_DIVISOR, f->divisor);
636	if (f->baseclass)
637	NLA_PUT_U32(skb, TCA_FLOW_BASECLASS, f->baseclass);
638
72d9794f PM	639	if (f->perturb_period)
	640	NLA_PUT_U32(skb, TCA_FLOW_PERTURB, f->perturb_period / HZ);
	641
e5dfb815 PM	642	if (tcf_exts_dump(skb, &f->exts, &flow_ext_map) < 0)
e5dfb815 PM	643	goto nla_put_failure;
0aead543	644	#ifdef CONFIG_NET_EMATCH
e5dfb815 PM	645	if (f->ematches.hdr.nmatches &&
	646	tcf_em_tree_dump(skb, &f->ematches, TCA_FLOW_EMATCHES) < 0)
	647	goto nla_put_failure;
0aead543	648	#endif
e5dfb815 PM	649	nla_nest_end(skb, nest);
	650
	651	if (tcf_exts_dump_stats(skb, &f->exts, &flow_ext_map) < 0)
	652	goto nla_put_failure;
	653
	654	return skb->len;
	655
	656	nla_put_failure:
	657	nlmsg_trim(skb, nest);
	658	return -1;
	659	}
	660
	661	static void flow_walk(struct tcf_proto tp, struct tcf_walker arg)
	662	{
	663	struct flow_head *head = tp->root;
	664	struct flow_filter *f;
	665
	666	list_for_each_entry(f, &head->filters, list) {
	667	if (arg->count < arg->skip)
	668	goto skip;
	669	if (arg->fn(tp, (unsigned long)f, arg) < 0) {
	670	arg->stop = 1;
	671	break;
	672	}
	673	skip:
	674	arg->count++;
	675	}
	676	}
	677
	678	static struct tcf_proto_ops cls_flow_ops __read_mostly = {
	679	.kind = "flow",
	680	.classify = flow_classify,
	681	.init = flow_init,
	682	.destroy = flow_destroy,
	683	.change = flow_change,
	684	.delete = flow_delete,
	685	.get = flow_get,
	686	.put = flow_put,
	687	.dump = flow_dump,
	688	.walk = flow_walk,
	689	.owner = THIS_MODULE,
	690	};
	691
	692	static int __init cls_flow_init(void)
	693	{
	694	return register_tcf_proto_ops(&cls_flow_ops);
	695	}
	696
	697	static void __exit cls_flow_exit(void)
	698	{
	699	unregister_tcf_proto_ops(&cls_flow_ops);
	700	}
	701
	702	module_init(cls_flow_init);
	703	module_exit(cls_flow_exit);
	704
	705	MODULE_LICENSE("GPL");
	706	MODULE_AUTHOR("Patrick McHardy <[email protected]>");
	707	MODULE_DESCRIPTION("TC flow classifier");