[deliverable/linux.git] / net / sched / cls_api.c

/*
 * net/sched/cls_api.c	Packet classifier API.
 *
 *		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.
 *
 * Authors:	Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
 *
 * Changes:
 *
 * Eduardo J. Blanco <ejbs@netlabs.com.uy> :990222: kmod support
 *
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <linux/init.h>
#include <linux/kmod.h>
#include <linux/netlink.h>
#include <net/net_namespace.h>
#include <net/sock.h>
#include <net/netlink.h>
#include <net/pkt_sched.h>
#include <net/pkt_cls.h>

/* The list of all installed classifier types */

static struct tcf_proto_ops *tcf_proto_base __read_mostly;

/* Protects list of registered TC modules. It is pure SMP lock. */
static DEFINE_RWLOCK(cls_mod_lock);

/* Find classifier type by string name */

static struct tcf_proto_ops *tcf_proto_lookup_ops(struct rtattr *kind)
{
	struct tcf_proto_ops *t = NULL;

	if (kind) {
		read_lock(&cls_mod_lock);
		for (t = tcf_proto_base; t; t = t->next) {
			if (rtattr_strcmp(kind, t->kind) == 0) {
				if (!try_module_get(t->owner))
					t = NULL;
				break;
			}
		}
		read_unlock(&cls_mod_lock);
	}
	return t;
}

/* Register(unregister) new classifier type */

int register_tcf_proto_ops(struct tcf_proto_ops *ops)
{
	struct tcf_proto_ops *t, **tp;
	int rc = -EEXIST;

	write_lock(&cls_mod_lock);
	for (tp = &tcf_proto_base; (t = *tp) != NULL; tp = &t->next)
		if (!strcmp(ops->kind, t->kind))
			goto out;

	ops->next = NULL;
	*tp = ops;
	rc = 0;
out:
	write_unlock(&cls_mod_lock);
	return rc;
}
EXPORT_SYMBOL(register_tcf_proto_ops);

int unregister_tcf_proto_ops(struct tcf_proto_ops *ops)
{
	struct tcf_proto_ops *t, **tp;
	int rc = -ENOENT;

	write_lock(&cls_mod_lock);
	for (tp = &tcf_proto_base; (t=*tp) != NULL; tp = &t->next)
		if (t == ops)
			break;

	if (!t)
		goto out;
	*tp = t->next;
	rc = 0;
out:
	write_unlock(&cls_mod_lock);
	return rc;
}
EXPORT_SYMBOL(unregister_tcf_proto_ops);

static int tfilter_notify(struct sk_buff *oskb, struct nlmsghdr *n,
			  struct tcf_proto *tp, unsigned long fh, int event);


/* Select new prio value from the range, managed by kernel. */

static inline u32 tcf_auto_prio(struct tcf_proto *tp)
{
	u32 first = TC_H_MAKE(0xC0000000U, 0U);

	if (tp)
		first = tp->prio-1;

	return first;
}

/* Add/change/delete/get a filter node */

static int tc_ctl_tfilter(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
{
	struct net *net = skb->sk->sk_net;
	struct rtattr **tca;
	struct tcmsg *t;
	u32 protocol;
	u32 prio;
	u32 nprio;
	u32 parent;
	struct net_device *dev;
	struct Qdisc  *q;
	struct tcf_proto **back, **chain;
	struct tcf_proto *tp;
	struct tcf_proto_ops *tp_ops;
	const struct Qdisc_class_ops *cops;
	unsigned long cl;
	unsigned long fh;
	int err;

	if (net != &init_net)
		return -EINVAL;

replay:
	tca = arg;
	t = NLMSG_DATA(n);
	protocol = TC_H_MIN(t->tcm_info);
	prio = TC_H_MAJ(t->tcm_info);
	nprio = prio;
	parent = t->tcm_parent;
	cl = 0;

	if (prio == 0) {
		/* If no priority is given, user wants we allocated it. */
		if (n->nlmsg_type != RTM_NEWTFILTER || !(n->nlmsg_flags&NLM_F_CREATE))
			return -ENOENT;
		prio = TC_H_MAKE(0x80000000U, 0U);
	}

	/* Find head of filter chain. */

	/* Find link */
	dev = __dev_get_by_index(&init_net, t->tcm_ifindex);
	if (dev == NULL)
		return -ENODEV;

	/* Find qdisc */
	if (!parent) {
		q = dev->qdisc_sleeping;
		parent = q->handle;
	} else {
		q = qdisc_lookup(dev, TC_H_MAJ(t->tcm_parent));
		if (q == NULL)
			return -EINVAL;
	}

	/* Is it classful? */
	if ((cops = q->ops->cl_ops) == NULL)
		return -EINVAL;

	/* Do we search for filter, attached to class? */
	if (TC_H_MIN(parent)) {
		cl = cops->get(q, parent);
		if (cl == 0)
			return -ENOENT;
	}

	/* And the last stroke */
	chain = cops->tcf_chain(q, cl);
	err = -EINVAL;
	if (chain == NULL)
		goto errout;

	/* Check the chain for existence of proto-tcf with this priority */
	for (back = chain; (tp=*back) != NULL; back = &tp->next) {
		if (tp->prio >= prio) {
			if (tp->prio == prio) {
				if (!nprio || (tp->protocol != protocol && protocol))
					goto errout;
			} else
				tp = NULL;
			break;
		}
	}

	if (tp == NULL) {
		/* Proto-tcf does not exist, create new one */

		if (tca[TCA_KIND-1] == NULL || !protocol)
			goto errout;

		err = -ENOENT;
		if (n->nlmsg_type != RTM_NEWTFILTER || !(n->nlmsg_flags&NLM_F_CREATE))
			goto errout;


		/* Create new proto tcf */

		err = -ENOBUFS;
		tp = kzalloc(sizeof(*tp), GFP_KERNEL);
		if (tp == NULL)
			goto errout;
		err = -EINVAL;
		tp_ops = tcf_proto_lookup_ops(tca[TCA_KIND-1]);
		if (tp_ops == NULL) {
#ifdef CONFIG_KMOD
			struct rtattr *kind = tca[TCA_KIND-1];
			char name[IFNAMSIZ];

			if (kind != NULL &&
			    rtattr_strlcpy(name, kind, IFNAMSIZ) < IFNAMSIZ) {
				rtnl_unlock();
				request_module("cls_%s", name);
				rtnl_lock();
				tp_ops = tcf_proto_lookup_ops(kind);
				/* We dropped the RTNL semaphore in order to
				 * perform the module load.  So, even if we
				 * succeeded in loading the module we have to
				 * replay the request.  We indicate this using
				 * -EAGAIN.
				 */
				if (tp_ops != NULL) {
					module_put(tp_ops->owner);
					err = -EAGAIN;
				}
			}
#endif
			kfree(tp);
			goto errout;
		}
		tp->ops = tp_ops;
		tp->protocol = protocol;
		tp->prio = nprio ? : tcf_auto_prio(*back);
		tp->q = q;
		tp->classify = tp_ops->classify;
		tp->classid = parent;

		err = tp_ops->init(tp);
		if (err != 0) {
			module_put(tp_ops->owner);
			kfree(tp);
			goto errout;
		}

		qdisc_lock_tree(dev);
		tp->next = *back;
		*back = tp;
		qdisc_unlock_tree(dev);

	} else if (tca[TCA_KIND-1] && rtattr_strcmp(tca[TCA_KIND-1], tp->ops->kind))
		goto errout;

	fh = tp->ops->get(tp, t->tcm_handle);

	if (fh == 0) {
		if (n->nlmsg_type == RTM_DELTFILTER && t->tcm_handle == 0) {
			qdisc_lock_tree(dev);
			*back = tp->next;
			qdisc_unlock_tree(dev);

			tfilter_notify(skb, n, tp, fh, RTM_DELTFILTER);
			tcf_destroy(tp);
			err = 0;
			goto errout;
		}

		err = -ENOENT;
		if (n->nlmsg_type != RTM_NEWTFILTER ||
		    !(n->nlmsg_flags & NLM_F_CREATE))
			goto errout;
	} else {
		switch (n->nlmsg_type) {
		case RTM_NEWTFILTER:
			err = -EEXIST;
			if (n->nlmsg_flags & NLM_F_EXCL)
				goto errout;
			break;
		case RTM_DELTFILTER:
			err = tp->ops->delete(tp, fh);
			if (err == 0)
				tfilter_notify(skb, n, tp, fh, RTM_DELTFILTER);
			goto errout;
		case RTM_GETTFILTER:
			err = tfilter_notify(skb, n, tp, fh, RTM_NEWTFILTER);
			goto errout;
		default:
			err = -EINVAL;
			goto errout;
		}
	}

	err = tp->ops->change(tp, cl, t->tcm_handle, tca, &fh);
	if (err == 0)
		tfilter_notify(skb, n, tp, fh, RTM_NEWTFILTER);

errout:
	if (cl)
		cops->put(q, cl);
	if (err == -EAGAIN)
		/* Replay the request. */
		goto replay;
	return err;
}

static int tcf_fill_node(struct sk_buff *skb, struct tcf_proto *tp,
			 unsigned long fh, u32 pid, u32 seq, u16 flags, int event)
{
	struct tcmsg *tcm;
	struct nlmsghdr  *nlh;
	unsigned char *b = skb_tail_pointer(skb);

	nlh = NLMSG_NEW(skb, pid, seq, event, sizeof(*tcm), flags);
	tcm = NLMSG_DATA(nlh);
	tcm->tcm_family = AF_UNSPEC;
	tcm->tcm__pad1 = 0;
	tcm->tcm__pad1 = 0;
	tcm->tcm_ifindex = tp->q->dev->ifindex;
	tcm->tcm_parent = tp->classid;
	tcm->tcm_info = TC_H_MAKE(tp->prio, tp->protocol);
	RTA_PUT(skb, TCA_KIND, IFNAMSIZ, tp->ops->kind);
	tcm->tcm_handle = fh;
	if (RTM_DELTFILTER != event) {
		tcm->tcm_handle = 0;
		if (tp->ops->dump && tp->ops->dump(tp, fh, skb, tcm) < 0)
			goto rtattr_failure;
	}
	nlh->nlmsg_len = skb_tail_pointer(skb) - b;
	return skb->len;

nlmsg_failure:
rtattr_failure:
	nlmsg_trim(skb, b);
	return -1;
}

static int tfilter_notify(struct sk_buff *oskb, struct nlmsghdr *n,
			  struct tcf_proto *tp, unsigned long fh, int event)
{
	struct sk_buff *skb;
	u32 pid = oskb ? NETLINK_CB(oskb).pid : 0;

	skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
	if (!skb)
		return -ENOBUFS;

	if (tcf_fill_node(skb, tp, fh, pid, n->nlmsg_seq, 0, event) <= 0) {
		kfree_skb(skb);
		return -EINVAL;
	}

	return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC,
			      n->nlmsg_flags & NLM_F_ECHO);
}

struct tcf_dump_args {
	struct tcf_walker w;
	struct sk_buff *skb;
	struct netlink_callback *cb;
};

static int tcf_node_dump(struct tcf_proto *tp, unsigned long n,
			 struct tcf_walker *arg)
{
	struct tcf_dump_args *a = (void *)arg;

	return tcf_fill_node(a->skb, tp, n, NETLINK_CB(a->cb->skb).pid,
			     a->cb->nlh->nlmsg_seq, NLM_F_MULTI, RTM_NEWTFILTER);
}

static int tc_dump_tfilter(struct sk_buff *skb, struct netlink_callback *cb)
{
	struct net *net = skb->sk->sk_net;
	int t;
	int s_t;
	struct net_device *dev;
	struct Qdisc *q;
	struct tcf_proto *tp, **chain;
	struct tcmsg *tcm = (struct tcmsg *)NLMSG_DATA(cb->nlh);
	unsigned long cl = 0;
	const struct Qdisc_class_ops *cops;
	struct tcf_dump_args arg;

	if (net != &init_net)
		return 0;

	if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm)))
		return skb->len;
	if ((dev = dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
		return skb->len;

	if (!tcm->tcm_parent)
		q = dev->qdisc_sleeping;
	else
		q = qdisc_lookup(dev, TC_H_MAJ(tcm->tcm_parent));
	if (!q)
		goto out;
	if ((cops = q->ops->cl_ops) == NULL)
		goto errout;
	if (TC_H_MIN(tcm->tcm_parent)) {
		cl = cops->get(q, tcm->tcm_parent);
		if (cl == 0)
			goto errout;
	}
	chain = cops->tcf_chain(q, cl);
	if (chain == NULL)
		goto errout;

	s_t = cb->args[0];

	for (tp=*chain, t=0; tp; tp = tp->next, t++) {
		if (t < s_t) continue;
		if (TC_H_MAJ(tcm->tcm_info) &&
		    TC_H_MAJ(tcm->tcm_info) != tp->prio)
			continue;
		if (TC_H_MIN(tcm->tcm_info) &&
		    TC_H_MIN(tcm->tcm_info) != tp->protocol)
			continue;
		if (t > s_t)
			memset(&cb->args[1], 0, sizeof(cb->args)-sizeof(cb->args[0]));
		if (cb->args[1] == 0) {
			if (tcf_fill_node(skb, tp, 0, NETLINK_CB(cb->skb).pid,
					  cb->nlh->nlmsg_seq, NLM_F_MULTI,
					  RTM_NEWTFILTER) <= 0)
				break;

			cb->args[1] = 1;
		}
		if (tp->ops->walk == NULL)
			continue;
		arg.w.fn = tcf_node_dump;
		arg.skb = skb;
		arg.cb = cb;
		arg.w.stop = 0;
		arg.w.skip = cb->args[1]-1;
		arg.w.count = 0;
		tp->ops->walk(tp, &arg.w);
		cb->args[1] = arg.w.count+1;
		if (arg.w.stop)
			break;
	}

	cb->args[0] = t;

errout:
	if (cl)
		cops->put(q, cl);
out:
	dev_put(dev);
	return skb->len;
}

void tcf_exts_destroy(struct tcf_proto *tp, struct tcf_exts *exts)
{
#ifdef CONFIG_NET_CLS_ACT
	if (exts->action) {
		tcf_action_destroy(exts->action, TCA_ACT_UNBIND);
		exts->action = NULL;
	}
#endif
}
EXPORT_SYMBOL(tcf_exts_destroy);

int tcf_exts_validate(struct tcf_proto *tp, struct rtattr **tb,
		  struct rtattr *rate_tlv, struct tcf_exts *exts,
		  struct tcf_ext_map *map)
{
	memset(exts, 0, sizeof(*exts));

#ifdef CONFIG_NET_CLS_ACT
	{
		int err;
		struct tc_action *act;

		if (map->police && tb[map->police-1]) {
			act = tcf_action_init_1(tb[map->police-1], rate_tlv,
						"police", TCA_ACT_NOREPLACE,
						TCA_ACT_BIND, &err);
			if (act == NULL)
				return err;

			act->type = TCA_OLD_COMPAT;
			exts->action = act;
		} else if (map->action && tb[map->action-1]) {
			act = tcf_action_init(tb[map->action-1], rate_tlv, NULL,
				TCA_ACT_NOREPLACE, TCA_ACT_BIND, &err);
			if (act == NULL)
				return err;

			exts->action = act;
		}
	}
#else
	if ((map->action && tb[map->action-1]) ||
	    (map->police && tb[map->police-1]))
		return -EOPNOTSUPP;
#endif

	return 0;
}
EXPORT_SYMBOL(tcf_exts_validate);

void tcf_exts_change(struct tcf_proto *tp, struct tcf_exts *dst,
		     struct tcf_exts *src)
{
#ifdef CONFIG_NET_CLS_ACT
	if (src->action) {
		struct tc_action *act;
		tcf_tree_lock(tp);
		act = xchg(&dst->action, src->action);
		tcf_tree_unlock(tp);
		if (act)
			tcf_action_destroy(act, TCA_ACT_UNBIND);
	}
#endif
}
EXPORT_SYMBOL(tcf_exts_change);

int tcf_exts_dump(struct sk_buff *skb, struct tcf_exts *exts,
	      struct tcf_ext_map *map)
{
#ifdef CONFIG_NET_CLS_ACT
	if (map->action && exts->action) {
		/*
		 * again for backward compatible mode - we want
		 * to work with both old and new modes of entering
		 * tc data even if iproute2  was newer - jhs
		 */
		struct rtattr *p_rta = (struct rtattr *)skb_tail_pointer(skb);

		if (exts->action->type != TCA_OLD_COMPAT) {
			RTA_PUT(skb, map->action, 0, NULL);
			if (tcf_action_dump(skb, exts->action, 0, 0) < 0)
				goto rtattr_failure;
			p_rta->rta_len = skb_tail_pointer(skb) - (u8 *)p_rta;
		} else if (map->police) {
			RTA_PUT(skb, map->police, 0, NULL);
			if (tcf_action_dump_old(skb, exts->action, 0, 0) < 0)
				goto rtattr_failure;
			p_rta->rta_len = skb_tail_pointer(skb) - (u8 *)p_rta;
		}
	}
#endif
	return 0;
rtattr_failure: __attribute__ ((unused))
	return -1;
}
EXPORT_SYMBOL(tcf_exts_dump);


int tcf_exts_dump_stats(struct sk_buff *skb, struct tcf_exts *exts,
			struct tcf_ext_map *map)
{
#ifdef CONFIG_NET_CLS_ACT
	if (exts->action)
		if (tcf_action_copy_stats(skb, exts->action, 1) < 0)
			goto rtattr_failure;
#endif
	return 0;
rtattr_failure: __attribute__ ((unused))
	return -1;
}
EXPORT_SYMBOL(tcf_exts_dump_stats);

static int __init tc_filter_init(void)
{
	rtnl_register(PF_UNSPEC, RTM_NEWTFILTER, tc_ctl_tfilter, NULL);
	rtnl_register(PF_UNSPEC, RTM_DELTFILTER, tc_ctl_tfilter, NULL);
	rtnl_register(PF_UNSPEC, RTM_GETTFILTER, tc_ctl_tfilter,
						 tc_dump_tfilter);

	return 0;
}

subsys_initcall(tc_filter_init);
Commit	Line	Data
1da177e4 LT	1	/*
	2	* net/sched/cls_api.c Packet classifier API.
	3	*
	4	* This program is free software; you can redistribute it and/or
	5	* modify it under the terms of the GNU General Public License
	6	* as published by the Free Software Foundation; either version
	7	* 2 of the License, or (at your option) any later version.
	8	*
	9	* Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
	10	*
	11	* Changes:
	12	*
	13	* Eduardo J. Blanco <ejbs@netlabs.com.uy> :990222: kmod support
	14	*
	15	*/
	16
1da177e4 LT	17	#include <linux/module.h>
	18	#include <linux/types.h>
	19	#include <linux/kernel.h>
1da177e4	20	#include <linux/string.h>
1da177e4	21	#include <linux/errno.h>
1da177e4	22	#include <linux/skbuff.h>
1da177e4 LT	23	#include <linux/init.h>
1da177e4 LT	24	#include <linux/kmod.h>
dc5fc579	25	#include <linux/netlink.h>
b854272b DL	26	#include <net/net_namespace.h>
b854272b DL	27	#include <net/sock.h>
dc5fc579	28	#include <net/netlink.h>
1da177e4 LT	29	#include <net/pkt_sched.h>
	30	#include <net/pkt_cls.h>
	31
1da177e4 LT	32	/* The list of all installed classifier types */
1da177e4 LT	33
2eb9d75c	34	static struct tcf_proto_ops *tcf_proto_base __read_mostly;
1da177e4 LT	35
	36	/* Protects list of registered TC modules. It is pure SMP lock. */
	37	static DEFINE_RWLOCK(cls_mod_lock);
	38
	39	/* Find classifier type by string name */
	40
aa767bfe	41	static struct tcf_proto_ops tcf_proto_lookup_ops(struct rtattr kind)
1da177e4 LT	42	{
	43	struct tcf_proto_ops *t = NULL;
	44
	45	if (kind) {
	46	read_lock(&cls_mod_lock);
	47	for (t = tcf_proto_base; t; t = t->next) {
	48	if (rtattr_strcmp(kind, t->kind) == 0) {
	49	if (!try_module_get(t->owner))
	50	t = NULL;
	51	break;
	52	}
	53	}
	54	read_unlock(&cls_mod_lock);
	55	}
	56	return t;
	57	}
	58
	59	/* Register(unregister) new classifier type */
	60
	61	int register_tcf_proto_ops(struct tcf_proto_ops *ops)
	62	{
	63	struct tcf_proto_ops t, *tp;
	64	int rc = -EEXIST;
	65
	66	write_lock(&cls_mod_lock);
	67	for (tp = &tcf_proto_base; (t = *tp) != NULL; tp = &t->next)
	68	if (!strcmp(ops->kind, t->kind))
	69	goto out;
	70
	71	ops->next = NULL;
	72	*tp = ops;
	73	rc = 0;
	74	out:
	75	write_unlock(&cls_mod_lock);
	76	return rc;
	77	}
aa767bfe	78	EXPORT_SYMBOL(register_tcf_proto_ops);
1da177e4 LT	79
	80	int unregister_tcf_proto_ops(struct tcf_proto_ops *ops)
	81	{
	82	struct tcf_proto_ops t, *tp;
	83	int rc = -ENOENT;
	84
	85	write_lock(&cls_mod_lock);
	86	for (tp = &tcf_proto_base; (t=*tp) != NULL; tp = &t->next)
	87	if (t == ops)
	88	break;
	89
	90	if (!t)
	91	goto out;
	92	*tp = t->next;
	93	rc = 0;
	94	out:
	95	write_unlock(&cls_mod_lock);
	96	return rc;
	97	}
aa767bfe	98	EXPORT_SYMBOL(unregister_tcf_proto_ops);
1da177e4 LT	99
	100	static int tfilter_notify(struct sk_buff oskb, struct nlmsghdr n,
	101	struct tcf_proto *tp, unsigned long fh, int event);
	102
	103
	104	/* Select new prio value from the range, managed by kernel. */
	105
aa767bfe	106	static inline u32 tcf_auto_prio(struct tcf_proto *tp)
1da177e4	107	{
aa767bfe	108	u32 first = TC_H_MAKE(0xC0000000U, 0U);
1da177e4 LT	109
	110	if (tp)
	111	first = tp->prio-1;
	112
	113	return first;
	114	}
	115
	116	/* Add/change/delete/get a filter node */
	117
	118	static int tc_ctl_tfilter(struct sk_buff skb, struct nlmsghdr n, void *arg)
	119	{
b854272b	120	struct net *net = skb->sk->sk_net;
1da177e4 LT	121	struct rtattr **tca;
	122	struct tcmsg *t;
	123	u32 protocol;
	124	u32 prio;
	125	u32 nprio;
	126	u32 parent;
	127	struct net_device *dev;
	128	struct Qdisc *q;
	129	struct tcf_proto back, chain;
	130	struct tcf_proto *tp;
	131	struct tcf_proto_ops *tp_ops;
20fea08b	132	const struct Qdisc_class_ops *cops;
1da177e4 LT	133	unsigned long cl;
	134	unsigned long fh;
	135	int err;
	136
b854272b DL	137	if (net != &init_net)
	138	return -EINVAL;
	139
1da177e4 LT	140	replay:
	141	tca = arg;
	142	t = NLMSG_DATA(n);
	143	protocol = TC_H_MIN(t->tcm_info);
	144	prio = TC_H_MAJ(t->tcm_info);
	145	nprio = prio;
	146	parent = t->tcm_parent;
	147	cl = 0;
	148
	149	if (prio == 0) {
	150	/* If no priority is given, user wants we allocated it. */
	151	if (n->nlmsg_type != RTM_NEWTFILTER \|\| !(n->nlmsg_flags&NLM_F_CREATE))
	152	return -ENOENT;
aa767bfe	153	prio = TC_H_MAKE(0x80000000U, 0U);
1da177e4 LT	154	}
	155
	156	/* Find head of filter chain. */
	157
	158	/* Find link */
aa767bfe SH	159	dev = __dev_get_by_index(&init_net, t->tcm_ifindex);
aa767bfe SH	160	if (dev == NULL)
1da177e4 LT	161	return -ENODEV;
	162
	163	/* Find qdisc */
	164	if (!parent) {
	165	q = dev->qdisc_sleeping;
	166	parent = q->handle;
aa767bfe SH	167	} else {
	168	q = qdisc_lookup(dev, TC_H_MAJ(t->tcm_parent));
	169	if (q == NULL)
	170	return -EINVAL;
	171	}
1da177e4 LT	172
	173	/* Is it classful? */
	174	if ((cops = q->ops->cl_ops) == NULL)
	175	return -EINVAL;
	176
	177	/* Do we search for filter, attached to class? */
	178	if (TC_H_MIN(parent)) {
	179	cl = cops->get(q, parent);
	180	if (cl == 0)
	181	return -ENOENT;
	182	}
	183
	184	/* And the last stroke */
	185	chain = cops->tcf_chain(q, cl);
	186	err = -EINVAL;
	187	if (chain == NULL)
	188	goto errout;
	189
	190	/* Check the chain for existence of proto-tcf with this priority */
	191	for (back = chain; (tp=*back) != NULL; back = &tp->next) {
	192	if (tp->prio >= prio) {
	193	if (tp->prio == prio) {
	194	if (!nprio \|\| (tp->protocol != protocol && protocol))
	195	goto errout;
	196	} else
	197	tp = NULL;
	198	break;
	199	}
	200	}
	201
	202	if (tp == NULL) {
	203	/* Proto-tcf does not exist, create new one */
	204
	205	if (tca[TCA_KIND-1] == NULL \|\| !protocol)
	206	goto errout;
	207
	208	err = -ENOENT;
	209	if (n->nlmsg_type != RTM_NEWTFILTER \|\| !(n->nlmsg_flags&NLM_F_CREATE))
	210	goto errout;
	211
	212
	213	/* Create new proto tcf */
	214
	215	err = -ENOBUFS;
aa767bfe SH	216	tp = kzalloc(sizeof(*tp), GFP_KERNEL);
aa767bfe SH	217	if (tp == NULL)
1da177e4 LT	218	goto errout;
	219	err = -EINVAL;
	220	tp_ops = tcf_proto_lookup_ops(tca[TCA_KIND-1]);
	221	if (tp_ops == NULL) {
	222	#ifdef CONFIG_KMOD
	223	struct rtattr *kind = tca[TCA_KIND-1];
	224	char name[IFNAMSIZ];
	225
	226	if (kind != NULL &&
	227	rtattr_strlcpy(name, kind, IFNAMSIZ) < IFNAMSIZ) {
	228	rtnl_unlock();
	229	request_module("cls_%s", name);
	230	rtnl_lock();
	231	tp_ops = tcf_proto_lookup_ops(kind);
	232	/* We dropped the RTNL semaphore in order to
	233	* perform the module load. So, even if we
	234	* succeeded in loading the module we have to
	235	* replay the request. We indicate this using
	236	* -EAGAIN.
	237	*/
	238	if (tp_ops != NULL) {
	239	module_put(tp_ops->owner);
	240	err = -EAGAIN;
	241	}
	242	}
	243	#endif
	244	kfree(tp);
	245	goto errout;
	246	}
1da177e4 LT	247	tp->ops = tp_ops;
	248	tp->protocol = protocol;
	249	tp->prio = nprio ? : tcf_auto_prio(*back);
	250	tp->q = q;
	251	tp->classify = tp_ops->classify;
	252	tp->classid = parent;
aa767bfe SH	253
	254	err = tp_ops->init(tp);
	255	if (err != 0) {
1da177e4 LT	256	module_put(tp_ops->owner);
	257	kfree(tp);
	258	goto errout;
	259	}
	260
	261	qdisc_lock_tree(dev);
	262	tp->next = *back;
	263	*back = tp;
	264	qdisc_unlock_tree(dev);
	265
	266	} else if (tca[TCA_KIND-1] && rtattr_strcmp(tca[TCA_KIND-1], tp->ops->kind))
	267	goto errout;
	268
	269	fh = tp->ops->get(tp, t->tcm_handle);
	270
	271	if (fh == 0) {
	272	if (n->nlmsg_type == RTM_DELTFILTER && t->tcm_handle == 0) {
	273	qdisc_lock_tree(dev);
	274	*back = tp->next;
	275	qdisc_unlock_tree(dev);
	276
	277	tfilter_notify(skb, n, tp, fh, RTM_DELTFILTER);
	278	tcf_destroy(tp);
	279	err = 0;
	280	goto errout;
	281	}
	282
	283	err = -ENOENT;
aa767bfe SH	284	if (n->nlmsg_type != RTM_NEWTFILTER \|\|
aa767bfe SH	285	!(n->nlmsg_flags & NLM_F_CREATE))
1da177e4 LT	286	goto errout;
	287	} else {
	288	switch (n->nlmsg_type) {
10297b99	289	case RTM_NEWTFILTER:
1da177e4	290	err = -EEXIST;
aa767bfe	291	if (n->nlmsg_flags & NLM_F_EXCL)
1da177e4 LT	292	goto errout;
	293	break;
	294	case RTM_DELTFILTER:
	295	err = tp->ops->delete(tp, fh);
	296	if (err == 0)
	297	tfilter_notify(skb, n, tp, fh, RTM_DELTFILTER);
	298	goto errout;
	299	case RTM_GETTFILTER:
	300	err = tfilter_notify(skb, n, tp, fh, RTM_NEWTFILTER);
	301	goto errout;
	302	default:
	303	err = -EINVAL;
	304	goto errout;
	305	}
	306	}
	307
	308	err = tp->ops->change(tp, cl, t->tcm_handle, tca, &fh);
	309	if (err == 0)
	310	tfilter_notify(skb, n, tp, fh, RTM_NEWTFILTER);
	311
	312	errout:
	313	if (cl)
	314	cops->put(q, cl);
	315	if (err == -EAGAIN)
	316	/* Replay the request. */
	317	goto replay;
	318	return err;
	319	}
	320
aa767bfe SH	321	static int tcf_fill_node(struct sk_buff skb, struct tcf_proto tp,
aa767bfe SH	322	unsigned long fh, u32 pid, u32 seq, u16 flags, int event)
1da177e4 LT	323	{
	324	struct tcmsg *tcm;
	325	struct nlmsghdr *nlh;
27a884dc	326	unsigned char *b = skb_tail_pointer(skb);
1da177e4	327
e431b8c0	328	nlh = NLMSG_NEW(skb, pid, seq, event, sizeof(*tcm), flags);
1da177e4 LT	329	tcm = NLMSG_DATA(nlh);
1da177e4 LT	330	tcm->tcm_family = AF_UNSPEC;
9ef1d4c7 PM	331	tcm->tcm__pad1 = 0;
9ef1d4c7 PM	332	tcm->tcm__pad1 = 0;
1da177e4 LT	333	tcm->tcm_ifindex = tp->q->dev->ifindex;
	334	tcm->tcm_parent = tp->classid;
	335	tcm->tcm_info = TC_H_MAKE(tp->prio, tp->protocol);
	336	RTA_PUT(skb, TCA_KIND, IFNAMSIZ, tp->ops->kind);
	337	tcm->tcm_handle = fh;
	338	if (RTM_DELTFILTER != event) {
	339	tcm->tcm_handle = 0;
	340	if (tp->ops->dump && tp->ops->dump(tp, fh, skb, tcm) < 0)
	341	goto rtattr_failure;
	342	}
27a884dc	343	nlh->nlmsg_len = skb_tail_pointer(skb) - b;
1da177e4 LT	344	return skb->len;
	345
	346	nlmsg_failure:
	347	rtattr_failure:
dc5fc579	348	nlmsg_trim(skb, b);
1da177e4 LT	349	return -1;
	350	}
	351
	352	static int tfilter_notify(struct sk_buff oskb, struct nlmsghdr n,
	353	struct tcf_proto *tp, unsigned long fh, int event)
	354	{
	355	struct sk_buff *skb;
	356	u32 pid = oskb ? NETLINK_CB(oskb).pid : 0;
	357
	358	skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
	359	if (!skb)
	360	return -ENOBUFS;
	361
	362	if (tcf_fill_node(skb, tp, fh, pid, n->nlmsg_seq, 0, event) <= 0) {
	363	kfree_skb(skb);
	364	return -EINVAL;
	365	}
	366
aa767bfe SH	367	return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC,
aa767bfe SH	368	n->nlmsg_flags & NLM_F_ECHO);
1da177e4 LT	369	}
1da177e4 LT	370
aa767bfe	371	struct tcf_dump_args {
1da177e4 LT	372	struct tcf_walker w;
	373	struct sk_buff *skb;
	374	struct netlink_callback *cb;
	375	};
	376
aa767bfe SH	377	static int tcf_node_dump(struct tcf_proto *tp, unsigned long n,
aa767bfe SH	378	struct tcf_walker *arg)
1da177e4	379	{
aa767bfe	380	struct tcf_dump_args a = (void )arg;
1da177e4 LT	381
	382	return tcf_fill_node(a->skb, tp, n, NETLINK_CB(a->cb->skb).pid,
	383	a->cb->nlh->nlmsg_seq, NLM_F_MULTI, RTM_NEWTFILTER);
	384	}
	385
	386	static int tc_dump_tfilter(struct sk_buff skb, struct netlink_callback cb)
	387	{
b854272b	388	struct net *net = skb->sk->sk_net;
1da177e4 LT	389	int t;
	390	int s_t;
	391	struct net_device *dev;
	392	struct Qdisc *q;
	393	struct tcf_proto tp, *chain;
aa767bfe	394	struct tcmsg tcm = (struct tcmsg )NLMSG_DATA(cb->nlh);
1da177e4	395	unsigned long cl = 0;
20fea08b	396	const struct Qdisc_class_ops *cops;
1da177e4 LT	397	struct tcf_dump_args arg;
1da177e4 LT	398
b854272b DL	399	if (net != &init_net)
	400	return 0;
	401
1da177e4 LT	402	if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm)))
1da177e4 LT	403	return skb->len;
881d966b	404	if ((dev = dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
1da177e4 LT	405	return skb->len;
1da177e4 LT	406
1da177e4 LT	407	if (!tcm->tcm_parent)
	408	q = dev->qdisc_sleeping;
	409	else
	410	q = qdisc_lookup(dev, TC_H_MAJ(tcm->tcm_parent));
	411	if (!q)
	412	goto out;
	413	if ((cops = q->ops->cl_ops) == NULL)
	414	goto errout;
	415	if (TC_H_MIN(tcm->tcm_parent)) {
	416	cl = cops->get(q, tcm->tcm_parent);
	417	if (cl == 0)
	418	goto errout;
	419	}
	420	chain = cops->tcf_chain(q, cl);
	421	if (chain == NULL)
	422	goto errout;
	423
	424	s_t = cb->args[0];
	425
	426	for (tp=*chain, t=0; tp; tp = tp->next, t++) {
	427	if (t < s_t) continue;
	428	if (TC_H_MAJ(tcm->tcm_info) &&
	429	TC_H_MAJ(tcm->tcm_info) != tp->prio)
	430	continue;
	431	if (TC_H_MIN(tcm->tcm_info) &&
	432	TC_H_MIN(tcm->tcm_info) != tp->protocol)
	433	continue;
	434	if (t > s_t)
	435	memset(&cb->args[1], 0, sizeof(cb->args)-sizeof(cb->args[0]));
	436	if (cb->args[1] == 0) {
	437	if (tcf_fill_node(skb, tp, 0, NETLINK_CB(cb->skb).pid,
aa767bfe SH	438	cb->nlh->nlmsg_seq, NLM_F_MULTI,
aa767bfe SH	439	RTM_NEWTFILTER) <= 0)
1da177e4	440	break;
aa767bfe	441
1da177e4 LT	442	cb->args[1] = 1;
	443	}
	444	if (tp->ops->walk == NULL)
	445	continue;
	446	arg.w.fn = tcf_node_dump;
	447	arg.skb = skb;
	448	arg.cb = cb;
	449	arg.w.stop = 0;
	450	arg.w.skip = cb->args[1]-1;
	451	arg.w.count = 0;
	452	tp->ops->walk(tp, &arg.w);
	453	cb->args[1] = arg.w.count+1;
	454	if (arg.w.stop)
	455	break;
	456	}
	457
	458	cb->args[0] = t;
	459
	460	errout:
	461	if (cl)
	462	cops->put(q, cl);
	463	out:
1da177e4 LT	464	dev_put(dev);
	465	return skb->len;
	466	}
	467
aa767bfe	468	void tcf_exts_destroy(struct tcf_proto tp, struct tcf_exts exts)
1da177e4 LT	469	{
	470	#ifdef CONFIG_NET_CLS_ACT
	471	if (exts->action) {
	472	tcf_action_destroy(exts->action, TCA_ACT_UNBIND);
	473	exts->action = NULL;
	474	}
1da177e4 LT	475	#endif
1da177e4 LT	476	}
aa767bfe	477	EXPORT_SYMBOL(tcf_exts_destroy);
1da177e4	478
aa767bfe	479	int tcf_exts_validate(struct tcf_proto tp, struct rtattr *tb,
10297b99 YH	480	struct rtattr rate_tlv, struct tcf_exts exts,
10297b99 YH	481	struct tcf_ext_map *map)
1da177e4 LT	482	{
1da177e4 LT	483	memset(exts, 0, sizeof(*exts));
10297b99	484
1da177e4 LT	485	#ifdef CONFIG_NET_CLS_ACT
	486	{
	487	int err;
	488	struct tc_action *act;
	489
	490	if (map->police && tb[map->police-1]) {
aa767bfe SH	491	act = tcf_action_init_1(tb[map->police-1], rate_tlv,
	492	"police", TCA_ACT_NOREPLACE,
	493	TCA_ACT_BIND, &err);
1da177e4 LT	494	if (act == NULL)
	495	return err;
	496
	497	act->type = TCA_OLD_COMPAT;
	498	exts->action = act;
	499	} else if (map->action && tb[map->action-1]) {
	500	act = tcf_action_init(tb[map->action-1], rate_tlv, NULL,
	501	TCA_ACT_NOREPLACE, TCA_ACT_BIND, &err);
	502	if (act == NULL)
	503	return err;
	504
	505	exts->action = act;
	506	}
	507	}
1da177e4 LT	508	#else
	509	if ((map->action && tb[map->action-1]) \|\|
	510	(map->police && tb[map->police-1]))
	511	return -EOPNOTSUPP;
	512	#endif
	513
	514	return 0;
	515	}
aa767bfe	516	EXPORT_SYMBOL(tcf_exts_validate);
1da177e4	517
aa767bfe SH	518	void tcf_exts_change(struct tcf_proto tp, struct tcf_exts dst,
aa767bfe SH	519	struct tcf_exts *src)
1da177e4 LT	520	{
	521	#ifdef CONFIG_NET_CLS_ACT
	522	if (src->action) {
	523	struct tc_action *act;
	524	tcf_tree_lock(tp);
	525	act = xchg(&dst->action, src->action);
	526	tcf_tree_unlock(tp);
	527	if (act)
	528	tcf_action_destroy(act, TCA_ACT_UNBIND);
	529	}
1da177e4 LT	530	#endif
1da177e4 LT	531	}
aa767bfe	532	EXPORT_SYMBOL(tcf_exts_change);
1da177e4	533
aa767bfe	534	int tcf_exts_dump(struct sk_buff skb, struct tcf_exts exts,
1da177e4 LT	535	struct tcf_ext_map *map)
	536	{
	537	#ifdef CONFIG_NET_CLS_ACT
	538	if (map->action && exts->action) {
	539	/*
	540	* again for backward compatible mode - we want
	541	* to work with both old and new modes of entering
	542	* tc data even if iproute2 was newer - jhs
	543	*/
27a884dc	544	struct rtattr p_rta = (struct rtattr )skb_tail_pointer(skb);
1da177e4 LT	545
	546	if (exts->action->type != TCA_OLD_COMPAT) {
	547	RTA_PUT(skb, map->action, 0, NULL);
	548	if (tcf_action_dump(skb, exts->action, 0, 0) < 0)
	549	goto rtattr_failure;
27a884dc	550	p_rta->rta_len = skb_tail_pointer(skb) - (u8 *)p_rta;
1da177e4 LT	551	} else if (map->police) {
	552	RTA_PUT(skb, map->police, 0, NULL);
	553	if (tcf_action_dump_old(skb, exts->action, 0, 0) < 0)
	554	goto rtattr_failure;
27a884dc	555	p_rta->rta_len = skb_tail_pointer(skb) - (u8 *)p_rta;
1da177e4 LT	556	}
1da177e4 LT	557	}
1da177e4 LT	558	#endif
	559	return 0;
	560	rtattr_failure: __attribute__ ((unused))
	561	return -1;
	562	}
aa767bfe	563	EXPORT_SYMBOL(tcf_exts_dump);
1da177e4	564
aa767bfe SH	565
	566	int tcf_exts_dump_stats(struct sk_buff skb, struct tcf_exts exts,
	567	struct tcf_ext_map *map)
1da177e4 LT	568	{
	569	#ifdef CONFIG_NET_CLS_ACT
	570	if (exts->action)
	571	if (tcf_action_copy_stats(skb, exts->action, 1) < 0)
	572	goto rtattr_failure;
1da177e4 LT	573	#endif
	574	return 0;
	575	rtattr_failure: __attribute__ ((unused))
	576	return -1;
	577	}
aa767bfe	578	EXPORT_SYMBOL(tcf_exts_dump_stats);
1da177e4 LT	579
	580	static int __init tc_filter_init(void)
	581	{
82623c0d TG	582	rtnl_register(PF_UNSPEC, RTM_NEWTFILTER, tc_ctl_tfilter, NULL);
	583	rtnl_register(PF_UNSPEC, RTM_DELTFILTER, tc_ctl_tfilter, NULL);
	584	rtnl_register(PF_UNSPEC, RTM_GETTFILTER, tc_ctl_tfilter,
	585	tc_dump_tfilter);
1da177e4	586
1da177e4 LT	587	return 0;
	588	}
	589
	590	subsys_initcall(tc_filter_init);