[linux-2.6-block.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 <linux/err.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 nlattr *kind)
{
	struct tcf_proto_ops *t = NULL;

	if (kind) {
		read_lock(&cls_mod_lock);
		for (t = tcf_proto_base; t; t = t->next) {
			if (nla_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 = sock_net(skb->sk);
	struct nlattr *tca[TCA_MAX + 1];
	spinlock_t *root_lock;
	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:
	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;

	err = nlmsg_parse(n, sizeof(*t), tca, TCA_MAX, NULL);
	if (err < 0)
		return err;

	/* Find qdisc */
	if (!parent) {
		struct netdev_queue *dev_queue = netdev_get_tx_queue(dev, 0);
		q = dev_queue->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;
		}
	}

	root_lock = qdisc_root_sleeping_lock(q);

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

		if (tca[TCA_KIND] == 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 = -ENOENT;
		tp_ops = tcf_proto_lookup_ops(tca[TCA_KIND]);
		if (tp_ops == NULL) {
#ifdef CONFIG_KMOD
			struct nlattr *kind = tca[TCA_KIND];
			char name[IFNAMSIZ];

			if (kind != NULL &&
			    nla_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;
		}

		spin_lock_bh(root_lock);
		tp->next = *back;
		*back = tp;
		spin_unlock_bh(root_lock);

	} else if (tca[TCA_KIND] && nla_strcmp(tca[TCA_KIND], 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) {
			spin_lock_bh(root_lock);
			*back = tp->next;
			spin_unlock_bh(root_lock);

			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 = qdisc_dev(tp->q)->ifindex;
	tcm->tcm_parent = tp->classid;
	tcm->tcm_info = TC_H_MAKE(tp->prio, tp->protocol);
	NLA_PUT_STRING(skb, TCA_KIND, 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 nla_put_failure;
	}
	nlh->nlmsg_len = skb_tail_pointer(skb) - b;
	return skb->len;

nlmsg_failure:
nla_put_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 = sock_net(skb->sk);
	struct netdev_queue *dev_queue;
	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;

	dev_queue = netdev_get_tx_queue(dev, 0);
	if (!tcm->tcm_parent)
		q = dev_queue->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 nlattr **tb,
		  struct nlattr *rate_tlv, struct tcf_exts *exts,
		  const struct tcf_ext_map *map)
{
	memset(exts, 0, sizeof(*exts));

#ifdef CONFIG_NET_CLS_ACT
	{
		struct tc_action *act;

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

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

			exts->action = act;
		}
	}
#else
	if ((map->action && tb[map->action]) ||
	    (map->police && tb[map->police]))
		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,
		  const 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 nlattr *nest;

		if (exts->action->type != TCA_OLD_COMPAT) {
			nest = nla_nest_start(skb, map->action);
			if (nest == NULL)
				goto nla_put_failure;
			if (tcf_action_dump(skb, exts->action, 0, 0) < 0)
				goto nla_put_failure;
			nla_nest_end(skb, nest);
		} else if (map->police) {
			nest = nla_nest_start(skb, map->police);
			if (nest == NULL)
				goto nla_put_failure;
			if (tcf_action_dump_old(skb, exts->action, 0, 0) < 0)
				goto nla_put_failure;
			nla_nest_end(skb, nest);
		}
	}
#endif
	return 0;
nla_put_failure: __attribute__ ((unused))
	return -1;
}
EXPORT_SYMBOL(tcf_exts_dump);


int tcf_exts_dump_stats(struct sk_buff *skb, struct tcf_exts *exts,
			const struct tcf_ext_map *map)
{
#ifdef CONFIG_NET_CLS_ACT
	if (exts->action)
		if (tcf_action_copy_stats(skb, exts->action, 1) < 0)
			goto nla_put_failure;
#endif
	return 0;
nla_put_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>
ab27cfb8	26	#include <linux/err.h>
b854272b DL	27	#include <net/net_namespace.h>
b854272b DL	28	#include <net/sock.h>
dc5fc579	29	#include <net/netlink.h>
1da177e4 LT	30	#include <net/pkt_sched.h>
	31	#include <net/pkt_cls.h>
	32
1da177e4 LT	33	/* The list of all installed classifier types */
1da177e4 LT	34
2eb9d75c	35	static struct tcf_proto_ops *tcf_proto_base __read_mostly;
1da177e4 LT	36
	37	/* Protects list of registered TC modules. It is pure SMP lock. */
	38	static DEFINE_RWLOCK(cls_mod_lock);
	39
	40	/* Find classifier type by string name */
	41
add93b61	42	static struct tcf_proto_ops tcf_proto_lookup_ops(struct nlattr kind)
1da177e4 LT	43	{
	44	struct tcf_proto_ops *t = NULL;
	45
	46	if (kind) {
	47	read_lock(&cls_mod_lock);
	48	for (t = tcf_proto_base; t; t = t->next) {
add93b61	49	if (nla_strcmp(kind, t->kind) == 0) {
1da177e4 LT	50	if (!try_module_get(t->owner))
	51	t = NULL;
	52	break;
	53	}
	54	}
	55	read_unlock(&cls_mod_lock);
	56	}
	57	return t;
	58	}
	59
	60	/* Register(unregister) new classifier type */
	61
	62	int register_tcf_proto_ops(struct tcf_proto_ops *ops)
	63	{
	64	struct tcf_proto_ops t, *tp;
	65	int rc = -EEXIST;
	66
	67	write_lock(&cls_mod_lock);
	68	for (tp = &tcf_proto_base; (t = *tp) != NULL; tp = &t->next)
	69	if (!strcmp(ops->kind, t->kind))
	70	goto out;
	71
	72	ops->next = NULL;
	73	*tp = ops;
	74	rc = 0;
	75	out:
	76	write_unlock(&cls_mod_lock);
	77	return rc;
	78	}
aa767bfe	79	EXPORT_SYMBOL(register_tcf_proto_ops);
1da177e4 LT	80
	81	int unregister_tcf_proto_ops(struct tcf_proto_ops *ops)
	82	{
	83	struct tcf_proto_ops t, *tp;
	84	int rc = -ENOENT;
	85
	86	write_lock(&cls_mod_lock);
	87	for (tp = &tcf_proto_base; (t=*tp) != NULL; tp = &t->next)
	88	if (t == ops)
	89	break;
	90
	91	if (!t)
	92	goto out;
	93	*tp = t->next;
	94	rc = 0;
	95	out:
	96	write_unlock(&cls_mod_lock);
	97	return rc;
	98	}
aa767bfe	99	EXPORT_SYMBOL(unregister_tcf_proto_ops);
1da177e4 LT	100
	101	static int tfilter_notify(struct sk_buff oskb, struct nlmsghdr n,
	102	struct tcf_proto *tp, unsigned long fh, int event);
	103
	104
	105	/* Select new prio value from the range, managed by kernel. */
	106
aa767bfe	107	static inline u32 tcf_auto_prio(struct tcf_proto *tp)
1da177e4	108	{
aa767bfe	109	u32 first = TC_H_MAKE(0xC0000000U, 0U);
1da177e4 LT	110
	111	if (tp)
	112	first = tp->prio-1;
	113
	114	return first;
	115	}
	116
	117	/* Add/change/delete/get a filter node */
	118
	119	static int tc_ctl_tfilter(struct sk_buff skb, struct nlmsghdr n, void *arg)
	120	{
3b1e0a65	121	struct net *net = sock_net(skb->sk);
add93b61	122	struct nlattr *tca[TCA_MAX + 1];
55dbc640	123	spinlock_t *root_lock;
1da177e4 LT	124	struct tcmsg *t;
	125	u32 protocol;
	126	u32 prio;
	127	u32 nprio;
	128	u32 parent;
	129	struct net_device *dev;
	130	struct Qdisc *q;
	131	struct tcf_proto back, chain;
	132	struct tcf_proto *tp;
	133	struct tcf_proto_ops *tp_ops;
20fea08b	134	const struct Qdisc_class_ops *cops;
1da177e4 LT	135	unsigned long cl;
	136	unsigned long fh;
	137	int err;
	138
b854272b DL	139	if (net != &init_net)
	140	return -EINVAL;
	141
1da177e4	142	replay:
1da177e4 LT	143	t = NLMSG_DATA(n);
	144	protocol = TC_H_MIN(t->tcm_info);
	145	prio = TC_H_MAJ(t->tcm_info);
	146	nprio = prio;
	147	parent = t->tcm_parent;
	148	cl = 0;
	149
	150	if (prio == 0) {
	151	/* If no priority is given, user wants we allocated it. */
	152	if (n->nlmsg_type != RTM_NEWTFILTER \|\| !(n->nlmsg_flags&NLM_F_CREATE))
	153	return -ENOENT;
aa767bfe	154	prio = TC_H_MAKE(0x80000000U, 0U);
1da177e4 LT	155	}
	156
	157	/* Find head of filter chain. */
	158
	159	/* Find link */
aa767bfe SH	160	dev = __dev_get_by_index(&init_net, t->tcm_ifindex);
aa767bfe SH	161	if (dev == NULL)
1da177e4 LT	162	return -ENODEV;
1da177e4 LT	163
add93b61 PM	164	err = nlmsg_parse(n, sizeof(*t), tca, TCA_MAX, NULL);
	165	if (err < 0)
	166	return err;
	167
1da177e4 LT	168	/* Find qdisc */
1da177e4 LT	169	if (!parent) {
e8a0464c	170	struct netdev_queue *dev_queue = netdev_get_tx_queue(dev, 0);
b0e1e646	171	q = dev_queue->qdisc_sleeping;
1da177e4	172	parent = q->handle;
aa767bfe SH	173	} else {
	174	q = qdisc_lookup(dev, TC_H_MAJ(t->tcm_parent));
	175	if (q == NULL)
	176	return -EINVAL;
	177	}
1da177e4 LT	178
	179	/* Is it classful? */
	180	if ((cops = q->ops->cl_ops) == NULL)
	181	return -EINVAL;
	182
	183	/* Do we search for filter, attached to class? */
	184	if (TC_H_MIN(parent)) {
	185	cl = cops->get(q, parent);
	186	if (cl == 0)
	187	return -ENOENT;
	188	}
	189
	190	/* And the last stroke */
	191	chain = cops->tcf_chain(q, cl);
	192	err = -EINVAL;
	193	if (chain == NULL)
	194	goto errout;
	195
	196	/* Check the chain for existence of proto-tcf with this priority */
	197	for (back = chain; (tp=*back) != NULL; back = &tp->next) {
	198	if (tp->prio >= prio) {
	199	if (tp->prio == prio) {
	200	if (!nprio \|\| (tp->protocol != protocol && protocol))
	201	goto errout;
	202	} else
	203	tp = NULL;
	204	break;
	205	}
	206	}
	207
102396ae	208	root_lock = qdisc_root_sleeping_lock(q);
55dbc640	209
1da177e4 LT	210	if (tp == NULL) {
	211	/* Proto-tcf does not exist, create new one */
	212
add93b61	213	if (tca[TCA_KIND] == NULL \|\| !protocol)
1da177e4 LT	214	goto errout;
	215
	216	err = -ENOENT;
	217	if (n->nlmsg_type != RTM_NEWTFILTER \|\| !(n->nlmsg_flags&NLM_F_CREATE))
	218	goto errout;
	219
	220
	221	/* Create new proto tcf */
	222
	223	err = -ENOBUFS;
aa767bfe SH	224	tp = kzalloc(sizeof(*tp), GFP_KERNEL);
aa767bfe SH	225	if (tp == NULL)
1da177e4	226	goto errout;
f2df8249	227	err = -ENOENT;
add93b61	228	tp_ops = tcf_proto_lookup_ops(tca[TCA_KIND]);
1da177e4 LT	229	if (tp_ops == NULL) {
1da177e4 LT	230	#ifdef CONFIG_KMOD
add93b61	231	struct nlattr *kind = tca[TCA_KIND];
1da177e4 LT	232	char name[IFNAMSIZ];
	233
	234	if (kind != NULL &&
add93b61	235	nla_strlcpy(name, kind, IFNAMSIZ) < IFNAMSIZ) {
1da177e4 LT	236	rtnl_unlock();
	237	request_module("cls_%s", name);
	238	rtnl_lock();
	239	tp_ops = tcf_proto_lookup_ops(kind);
	240	/* We dropped the RTNL semaphore in order to
	241	* perform the module load. So, even if we
	242	* succeeded in loading the module we have to
	243	* replay the request. We indicate this using
	244	* -EAGAIN.
	245	*/
	246	if (tp_ops != NULL) {
	247	module_put(tp_ops->owner);
	248	err = -EAGAIN;
	249	}
	250	}
	251	#endif
	252	kfree(tp);
	253	goto errout;
	254	}
1da177e4 LT	255	tp->ops = tp_ops;
	256	tp->protocol = protocol;
	257	tp->prio = nprio ? : tcf_auto_prio(*back);
	258	tp->q = q;
	259	tp->classify = tp_ops->classify;
	260	tp->classid = parent;
aa767bfe SH	261
	262	err = tp_ops->init(tp);
	263	if (err != 0) {
1da177e4 LT	264	module_put(tp_ops->owner);
	265	kfree(tp);
	266	goto errout;
	267	}
	268
55dbc640	269	spin_lock_bh(root_lock);
1da177e4 LT	270	tp->next = *back;
1da177e4 LT	271	*back = tp;
55dbc640	272	spin_unlock_bh(root_lock);
1da177e4	273
add93b61	274	} else if (tca[TCA_KIND] && nla_strcmp(tca[TCA_KIND], tp->ops->kind))
1da177e4 LT	275	goto errout;
	276
	277	fh = tp->ops->get(tp, t->tcm_handle);
	278
	279	if (fh == 0) {
	280	if (n->nlmsg_type == RTM_DELTFILTER && t->tcm_handle == 0) {
55dbc640	281	spin_lock_bh(root_lock);
1da177e4	282	*back = tp->next;
323c0488	283	spin_unlock_bh(root_lock);
1da177e4 LT	284
	285	tfilter_notify(skb, n, tp, fh, RTM_DELTFILTER);
	286	tcf_destroy(tp);
	287	err = 0;
	288	goto errout;
	289	}
	290
	291	err = -ENOENT;
aa767bfe SH	292	if (n->nlmsg_type != RTM_NEWTFILTER \|\|
aa767bfe SH	293	!(n->nlmsg_flags & NLM_F_CREATE))
1da177e4 LT	294	goto errout;
	295	} else {
	296	switch (n->nlmsg_type) {
10297b99	297	case RTM_NEWTFILTER:
1da177e4	298	err = -EEXIST;
aa767bfe	299	if (n->nlmsg_flags & NLM_F_EXCL)
1da177e4 LT	300	goto errout;
	301	break;
	302	case RTM_DELTFILTER:
	303	err = tp->ops->delete(tp, fh);
	304	if (err == 0)
	305	tfilter_notify(skb, n, tp, fh, RTM_DELTFILTER);
	306	goto errout;
	307	case RTM_GETTFILTER:
	308	err = tfilter_notify(skb, n, tp, fh, RTM_NEWTFILTER);
	309	goto errout;
	310	default:
	311	err = -EINVAL;
	312	goto errout;
	313	}
	314	}
	315
	316	err = tp->ops->change(tp, cl, t->tcm_handle, tca, &fh);
	317	if (err == 0)
	318	tfilter_notify(skb, n, tp, fh, RTM_NEWTFILTER);
	319
	320	errout:
	321	if (cl)
	322	cops->put(q, cl);
	323	if (err == -EAGAIN)
	324	/* Replay the request. */
	325	goto replay;
	326	return err;
	327	}
	328
aa767bfe SH	329	static int tcf_fill_node(struct sk_buff skb, struct tcf_proto tp,
aa767bfe SH	330	unsigned long fh, u32 pid, u32 seq, u16 flags, int event)
1da177e4 LT	331	{
	332	struct tcmsg *tcm;
	333	struct nlmsghdr *nlh;
27a884dc	334	unsigned char *b = skb_tail_pointer(skb);
1da177e4	335
e431b8c0	336	nlh = NLMSG_NEW(skb, pid, seq, event, sizeof(*tcm), flags);
1da177e4 LT	337	tcm = NLMSG_DATA(nlh);
1da177e4 LT	338	tcm->tcm_family = AF_UNSPEC;
9ef1d4c7 PM	339	tcm->tcm__pad1 = 0;
9ef1d4c7 PM	340	tcm->tcm__pad1 = 0;
5ce2d488	341	tcm->tcm_ifindex = qdisc_dev(tp->q)->ifindex;
1da177e4 LT	342	tcm->tcm_parent = tp->classid;
1da177e4 LT	343	tcm->tcm_info = TC_H_MAKE(tp->prio, tp->protocol);
57e1c487	344	NLA_PUT_STRING(skb, TCA_KIND, tp->ops->kind);
1da177e4 LT	345	tcm->tcm_handle = fh;
	346	if (RTM_DELTFILTER != event) {
	347	tcm->tcm_handle = 0;
	348	if (tp->ops->dump && tp->ops->dump(tp, fh, skb, tcm) < 0)
add93b61	349	goto nla_put_failure;
1da177e4	350	}
27a884dc	351	nlh->nlmsg_len = skb_tail_pointer(skb) - b;
1da177e4 LT	352	return skb->len;
	353
	354	nlmsg_failure:
add93b61	355	nla_put_failure:
dc5fc579	356	nlmsg_trim(skb, b);
1da177e4 LT	357	return -1;
	358	}
	359
	360	static int tfilter_notify(struct sk_buff oskb, struct nlmsghdr n,
	361	struct tcf_proto *tp, unsigned long fh, int event)
	362	{
	363	struct sk_buff *skb;
	364	u32 pid = oskb ? NETLINK_CB(oskb).pid : 0;
	365
	366	skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
	367	if (!skb)
	368	return -ENOBUFS;
	369
	370	if (tcf_fill_node(skb, tp, fh, pid, n->nlmsg_seq, 0, event) <= 0) {
	371	kfree_skb(skb);
	372	return -EINVAL;
	373	}
	374
aa767bfe SH	375	return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC,
aa767bfe SH	376	n->nlmsg_flags & NLM_F_ECHO);
1da177e4 LT	377	}
1da177e4 LT	378
aa767bfe	379	struct tcf_dump_args {
1da177e4 LT	380	struct tcf_walker w;
	381	struct sk_buff *skb;
	382	struct netlink_callback *cb;
	383	};
	384
aa767bfe SH	385	static int tcf_node_dump(struct tcf_proto *tp, unsigned long n,
aa767bfe SH	386	struct tcf_walker *arg)
1da177e4	387	{
aa767bfe	388	struct tcf_dump_args a = (void )arg;
1da177e4 LT	389
	390	return tcf_fill_node(a->skb, tp, n, NETLINK_CB(a->cb->skb).pid,
	391	a->cb->nlh->nlmsg_seq, NLM_F_MULTI, RTM_NEWTFILTER);
	392	}
	393
	394	static int tc_dump_tfilter(struct sk_buff skb, struct netlink_callback cb)
	395	{
3b1e0a65	396	struct net *net = sock_net(skb->sk);
b0e1e646	397	struct netdev_queue *dev_queue;
1da177e4 LT	398	int t;
	399	int s_t;
	400	struct net_device *dev;
	401	struct Qdisc *q;
	402	struct tcf_proto tp, *chain;
aa767bfe	403	struct tcmsg tcm = (struct tcmsg )NLMSG_DATA(cb->nlh);
1da177e4	404	unsigned long cl = 0;
20fea08b	405	const struct Qdisc_class_ops *cops;
1da177e4 LT	406	struct tcf_dump_args arg;
1da177e4 LT	407
b854272b DL	408	if (net != &init_net)
	409	return 0;
	410
1da177e4 LT	411	if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm)))
1da177e4 LT	412	return skb->len;
881d966b	413	if ((dev = dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
1da177e4 LT	414	return skb->len;
1da177e4 LT	415
e8a0464c	416	dev_queue = netdev_get_tx_queue(dev, 0);
1da177e4	417	if (!tcm->tcm_parent)
b0e1e646	418	q = dev_queue->qdisc_sleeping;
1da177e4 LT	419	else
	420	q = qdisc_lookup(dev, TC_H_MAJ(tcm->tcm_parent));
	421	if (!q)
	422	goto out;
	423	if ((cops = q->ops->cl_ops) == NULL)
	424	goto errout;
	425	if (TC_H_MIN(tcm->tcm_parent)) {
	426	cl = cops->get(q, tcm->tcm_parent);
	427	if (cl == 0)
	428	goto errout;
	429	}
	430	chain = cops->tcf_chain(q, cl);
	431	if (chain == NULL)
	432	goto errout;
	433
	434	s_t = cb->args[0];
	435
	436	for (tp=*chain, t=0; tp; tp = tp->next, t++) {
	437	if (t < s_t) continue;
	438	if (TC_H_MAJ(tcm->tcm_info) &&
	439	TC_H_MAJ(tcm->tcm_info) != tp->prio)
	440	continue;
	441	if (TC_H_MIN(tcm->tcm_info) &&
	442	TC_H_MIN(tcm->tcm_info) != tp->protocol)
	443	continue;
	444	if (t > s_t)
	445	memset(&cb->args[1], 0, sizeof(cb->args)-sizeof(cb->args[0]));
	446	if (cb->args[1] == 0) {
	447	if (tcf_fill_node(skb, tp, 0, NETLINK_CB(cb->skb).pid,
aa767bfe SH	448	cb->nlh->nlmsg_seq, NLM_F_MULTI,
aa767bfe SH	449	RTM_NEWTFILTER) <= 0)
1da177e4	450	break;
aa767bfe	451
1da177e4 LT	452	cb->args[1] = 1;
	453	}
	454	if (tp->ops->walk == NULL)
	455	continue;
	456	arg.w.fn = tcf_node_dump;
	457	arg.skb = skb;
	458	arg.cb = cb;
	459	arg.w.stop = 0;
	460	arg.w.skip = cb->args[1]-1;
	461	arg.w.count = 0;
	462	tp->ops->walk(tp, &arg.w);
	463	cb->args[1] = arg.w.count+1;
	464	if (arg.w.stop)
	465	break;
	466	}
	467
	468	cb->args[0] = t;
	469
	470	errout:
	471	if (cl)
	472	cops->put(q, cl);
	473	out:
1da177e4 LT	474	dev_put(dev);
	475	return skb->len;
	476	}
	477
aa767bfe	478	void tcf_exts_destroy(struct tcf_proto tp, struct tcf_exts exts)
1da177e4 LT	479	{
	480	#ifdef CONFIG_NET_CLS_ACT
	481	if (exts->action) {
	482	tcf_action_destroy(exts->action, TCA_ACT_UNBIND);
	483	exts->action = NULL;
	484	}
1da177e4 LT	485	#endif
1da177e4 LT	486	}
aa767bfe	487	EXPORT_SYMBOL(tcf_exts_destroy);
1da177e4	488
add93b61 PM	489	int tcf_exts_validate(struct tcf_proto tp, struct nlattr *tb,
add93b61 PM	490	struct nlattr rate_tlv, struct tcf_exts exts,
5239008b	491	const struct tcf_ext_map *map)
1da177e4 LT	492	{
1da177e4 LT	493	memset(exts, 0, sizeof(*exts));
10297b99	494
1da177e4 LT	495	#ifdef CONFIG_NET_CLS_ACT
1da177e4 LT	496	{
1da177e4 LT	497	struct tc_action *act;
1da177e4 LT	498
add93b61	499	if (map->police && tb[map->police]) {
7ba699c6	500	act = tcf_action_init_1(tb[map->police], rate_tlv,
aa767bfe	501	"police", TCA_ACT_NOREPLACE,
ab27cfb8 PM	502	TCA_ACT_BIND);
	503	if (IS_ERR(act))
	504	return PTR_ERR(act);
1da177e4 LT	505
	506	act->type = TCA_OLD_COMPAT;
	507	exts->action = act;
add93b61	508	} else if (map->action && tb[map->action]) {
7ba699c6	509	act = tcf_action_init(tb[map->action], rate_tlv, NULL,
ab27cfb8 PM	510	TCA_ACT_NOREPLACE, TCA_ACT_BIND);
	511	if (IS_ERR(act))
	512	return PTR_ERR(act);
1da177e4 LT	513
	514	exts->action = act;
	515	}
	516	}
1da177e4	517	#else
add93b61 PM	518	if ((map->action && tb[map->action]) \|\|
add93b61 PM	519	(map->police && tb[map->police]))
1da177e4 LT	520	return -EOPNOTSUPP;
	521	#endif
	522
	523	return 0;
	524	}
aa767bfe	525	EXPORT_SYMBOL(tcf_exts_validate);
1da177e4	526
aa767bfe SH	527	void tcf_exts_change(struct tcf_proto tp, struct tcf_exts dst,
aa767bfe SH	528	struct tcf_exts *src)
1da177e4 LT	529	{
	530	#ifdef CONFIG_NET_CLS_ACT
	531	if (src->action) {
	532	struct tc_action *act;
	533	tcf_tree_lock(tp);
	534	act = xchg(&dst->action, src->action);
	535	tcf_tree_unlock(tp);
	536	if (act)
	537	tcf_action_destroy(act, TCA_ACT_UNBIND);
	538	}
1da177e4 LT	539	#endif
1da177e4 LT	540	}
aa767bfe	541	EXPORT_SYMBOL(tcf_exts_change);
1da177e4	542
aa767bfe	543	int tcf_exts_dump(struct sk_buff skb, struct tcf_exts exts,
5239008b	544	const struct tcf_ext_map *map)
1da177e4 LT	545	{
	546	#ifdef CONFIG_NET_CLS_ACT
	547	if (map->action && exts->action) {
	548	/*
	549	* again for backward compatible mode - we want
	550	* to work with both old and new modes of entering
	551	* tc data even if iproute2 was newer - jhs
	552	*/
4b3550ef	553	struct nlattr *nest;
1da177e4 LT	554
1da177e4 LT	555	if (exts->action->type != TCA_OLD_COMPAT) {
4b3550ef PM	556	nest = nla_nest_start(skb, map->action);
	557	if (nest == NULL)
	558	goto nla_put_failure;
1da177e4	559	if (tcf_action_dump(skb, exts->action, 0, 0) < 0)
add93b61	560	goto nla_put_failure;
4b3550ef	561	nla_nest_end(skb, nest);
1da177e4	562	} else if (map->police) {
4b3550ef PM	563	nest = nla_nest_start(skb, map->police);
	564	if (nest == NULL)
	565	goto nla_put_failure;
1da177e4	566	if (tcf_action_dump_old(skb, exts->action, 0, 0) < 0)
add93b61	567	goto nla_put_failure;
4b3550ef	568	nla_nest_end(skb, nest);
1da177e4 LT	569	}
1da177e4 LT	570	}
1da177e4 LT	571	#endif
1da177e4 LT	572	return 0;
add93b61	573	nla_put_failure: __attribute__ ((unused))
1da177e4 LT	574	return -1;
1da177e4 LT	575	}
aa767bfe	576	EXPORT_SYMBOL(tcf_exts_dump);
1da177e4	577
aa767bfe SH	578
aa767bfe SH	579	int tcf_exts_dump_stats(struct sk_buff skb, struct tcf_exts exts,
5239008b	580	const struct tcf_ext_map *map)
1da177e4 LT	581	{
	582	#ifdef CONFIG_NET_CLS_ACT
	583	if (exts->action)
	584	if (tcf_action_copy_stats(skb, exts->action, 1) < 0)
add93b61	585	goto nla_put_failure;
1da177e4 LT	586	#endif
1da177e4 LT	587	return 0;
add93b61	588	nla_put_failure: __attribute__ ((unused))
1da177e4 LT	589	return -1;
1da177e4 LT	590	}
aa767bfe	591	EXPORT_SYMBOL(tcf_exts_dump_stats);
1da177e4 LT	592
	593	static int __init tc_filter_init(void)
	594	{
82623c0d TG	595	rtnl_register(PF_UNSPEC, RTM_NEWTFILTER, tc_ctl_tfilter, NULL);
	596	rtnl_register(PF_UNSPEC, RTM_DELTFILTER, tc_ctl_tfilter, NULL);
	597	rtnl_register(PF_UNSPEC, RTM_GETTFILTER, tc_ctl_tfilter,
	598	tc_dump_tfilter);
1da177e4	599
1da177e4 LT	600	return 0;
	601	}
	602
	603	subsys_initcall(tc_filter_init);