[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>

#if 0 /* control */
#define DPRINTK(format,args...) printk(KERN_DEBUG format,##args)
#else
#define DPRINTK(format,args...)
#endif

/* The list of all installed classifier types */

static struct tcf_proto_ops *tcf_proto_base;

/* 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;
}

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;
}

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

	/* Find qdisc */
	if (!parent) {
		q = dev->qdisc_sleeping;
		parent = q->handle;
	} else if ((q = qdisc_lookup(dev, TC_H_MAJ(t->tcm_parent))) == 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;
		if ((tp = kzalloc(sizeof(*tp), GFP_KERNEL)) == 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;
		if ((err = tp_ops->init(tp)) != 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, 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
}


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;
}

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
}

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;
}

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;
}

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);

EXPORT_SYMBOL(register_tcf_proto_ops);
EXPORT_SYMBOL(unregister_tcf_proto_ops);
EXPORT_SYMBOL(tcf_exts_validate);
EXPORT_SYMBOL(tcf_exts_destroy);
EXPORT_SYMBOL(tcf_exts_change);
EXPORT_SYMBOL(tcf_exts_dump);
EXPORT_SYMBOL(tcf_exts_dump_stats);
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
	32	#if 0 /* control */
	33	#define DPRINTK(format,args...) printk(KERN_DEBUG format,##args)
	34	#else
	35	#define DPRINTK(format,args...)
	36	#endif
	37
	38	/* The list of all installed classifier types */
	39
	40	static struct tcf_proto_ops *tcf_proto_base;
	41
	42	/* Protects list of registered TC modules. It is pure SMP lock. */
	43	static DEFINE_RWLOCK(cls_mod_lock);
	44
	45	/* Find classifier type by string name */
	46
	47	static struct tcf_proto_ops * tcf_proto_lookup_ops(struct rtattr *kind)
	48	{
	49	struct tcf_proto_ops *t = NULL;
	50
	51	if (kind) {
	52	read_lock(&cls_mod_lock);
	53	for (t = tcf_proto_base; t; t = t->next) {
	54	if (rtattr_strcmp(kind, t->kind) == 0) {
	55	if (!try_module_get(t->owner))
	56	t = NULL;
	57	break;
	58	}
	59	}
	60	read_unlock(&cls_mod_lock);
	61	}
	62	return t;
	63	}
	64
	65	/* Register(unregister) new classifier type */
	66
	67	int register_tcf_proto_ops(struct tcf_proto_ops *ops)
	68	{
	69	struct tcf_proto_ops t, *tp;
	70	int rc = -EEXIST;
	71
	72	write_lock(&cls_mod_lock);
	73	for (tp = &tcf_proto_base; (t = *tp) != NULL; tp = &t->next)
	74	if (!strcmp(ops->kind, t->kind))
	75	goto out;
	76
	77	ops->next = NULL;
	78	*tp = ops;
	79	rc = 0;
	80	out:
	81	write_unlock(&cls_mod_lock);
	82	return rc;
	83	}
	84
	85	int unregister_tcf_proto_ops(struct tcf_proto_ops *ops)
	86	{
	87	struct tcf_proto_ops t, *tp;
	88	int rc = -ENOENT;
	89
	90	write_lock(&cls_mod_lock);
	91	for (tp = &tcf_proto_base; (t=*tp) != NULL; tp = &t->next)
	92	if (t == ops)
93	break;
94
95	if (!t)
96	goto out;
97	*tp = t->next;
98	rc = 0;
99	out:
100	write_unlock(&cls_mod_lock);
101	return rc;
102	}
103
104	static int tfilter_notify(struct sk_buff oskb, struct nlmsghdr n,
105	struct tcf_proto *tp, unsigned long fh, int event);
106
107
108	/* Select new prio value from the range, managed by kernel. */
109
110	static __inline__ u32 tcf_auto_prio(struct tcf_proto *tp)
111	{
112	u32 first = TC_H_MAKE(0xC0000000U,0U);
113
114	if (tp)
115	first = tp->prio-1;
116
117	return first;
118	}
119
120	/* Add/change/delete/get a filter node */
121
122	static int tc_ctl_tfilter(struct sk_buff skb, struct nlmsghdr n, void *arg)
123	{
b854272b	124	struct net *net = skb->sk->sk_net;
1da177e4 LT	125	struct rtattr **tca;
	126	struct tcmsg *t;
	127	u32 protocol;
	128	u32 prio;
	129	u32 nprio;
	130	u32 parent;
	131	struct net_device *dev;
	132	struct Qdisc *q;
	133	struct tcf_proto back, chain;
	134	struct tcf_proto *tp;
	135	struct tcf_proto_ops *tp_ops;
20fea08b	136	const struct Qdisc_class_ops *cops;
1da177e4 LT	137	unsigned long cl;
	138	unsigned long fh;
	139	int err;
	140
b854272b DL	141	if (net != &init_net)
	142	return -EINVAL;
	143
1da177e4 LT	144	replay:
	145	tca = arg;
	146	t = NLMSG_DATA(n);
	147	protocol = TC_H_MIN(t->tcm_info);
	148	prio = TC_H_MAJ(t->tcm_info);
	149	nprio = prio;
	150	parent = t->tcm_parent;
	151	cl = 0;
	152
	153	if (prio == 0) {
	154	/* If no priority is given, user wants we allocated it. */
	155	if (n->nlmsg_type != RTM_NEWTFILTER \|\| !(n->nlmsg_flags&NLM_F_CREATE))
	156	return -ENOENT;
	157	prio = TC_H_MAKE(0x80000000U,0U);
	158	}
	159
	160	/* Find head of filter chain. */
	161
	162	/* Find link */
881d966b	163	if ((dev = __dev_get_by_index(&init_net, t->tcm_ifindex)) == NULL)
1da177e4 LT	164	return -ENODEV;
	165
	166	/* Find qdisc */
	167	if (!parent) {
	168	q = dev->qdisc_sleeping;
	169	parent = q->handle;
	170	} else if ((q = qdisc_lookup(dev, TC_H_MAJ(t->tcm_parent))) == NULL)
	171	return -EINVAL;
	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;
c7b1b249	216	if ((tp = kzalloc(sizeof(*tp), GFP_KERNEL)) == NULL)
1da177e4 LT	217	goto errout;
	218	err = -EINVAL;
	219	tp_ops = tcf_proto_lookup_ops(tca[TCA_KIND-1]);
	220	if (tp_ops == NULL) {
	221	#ifdef CONFIG_KMOD
	222	struct rtattr *kind = tca[TCA_KIND-1];
	223	char name[IFNAMSIZ];
	224
	225	if (kind != NULL &&
	226	rtattr_strlcpy(name, kind, IFNAMSIZ) < IFNAMSIZ) {
	227	rtnl_unlock();
	228	request_module("cls_%s", name);
	229	rtnl_lock();
	230	tp_ops = tcf_proto_lookup_ops(kind);
	231	/* We dropped the RTNL semaphore in order to
	232	* perform the module load. So, even if we
	233	* succeeded in loading the module we have to
	234	* replay the request. We indicate this using
	235	* -EAGAIN.
	236	*/
	237	if (tp_ops != NULL) {
	238	module_put(tp_ops->owner);
	239	err = -EAGAIN;
	240	}
	241	}
	242	#endif
	243	kfree(tp);
	244	goto errout;
	245	}
1da177e4 LT	246	tp->ops = tp_ops;
	247	tp->protocol = protocol;
	248	tp->prio = nprio ? : tcf_auto_prio(*back);
	249	tp->q = q;
	250	tp->classify = tp_ops->classify;
	251	tp->classid = parent;
	252	if ((err = tp_ops->init(tp)) != 0) {
	253	module_put(tp_ops->owner);
	254	kfree(tp);
	255	goto errout;
	256	}
	257
	258	qdisc_lock_tree(dev);
	259	tp->next = *back;
	260	*back = tp;
	261	qdisc_unlock_tree(dev);
	262
	263	} else if (tca[TCA_KIND-1] && rtattr_strcmp(tca[TCA_KIND-1], tp->ops->kind))
	264	goto errout;
	265
	266	fh = tp->ops->get(tp, t->tcm_handle);
	267
	268	if (fh == 0) {
	269	if (n->nlmsg_type == RTM_DELTFILTER && t->tcm_handle == 0) {
	270	qdisc_lock_tree(dev);
	271	*back = tp->next;
	272	qdisc_unlock_tree(dev);
	273
	274	tfilter_notify(skb, n, tp, fh, RTM_DELTFILTER);
	275	tcf_destroy(tp);
	276	err = 0;
	277	goto errout;
	278	}
	279
	280	err = -ENOENT;
	281	if (n->nlmsg_type != RTM_NEWTFILTER \|\| !(n->nlmsg_flags&NLM_F_CREATE))
	282	goto errout;
	283	} else {
	284	switch (n->nlmsg_type) {
10297b99	285	case RTM_NEWTFILTER:
1da177e4 LT	286	err = -EEXIST;
	287	if (n->nlmsg_flags&NLM_F_EXCL)
	288	goto errout;
	289	break;
	290	case RTM_DELTFILTER:
	291	err = tp->ops->delete(tp, fh);
	292	if (err == 0)
	293	tfilter_notify(skb, n, tp, fh, RTM_DELTFILTER);
	294	goto errout;
	295	case RTM_GETTFILTER:
	296	err = tfilter_notify(skb, n, tp, fh, RTM_NEWTFILTER);
	297	goto errout;
	298	default:
	299	err = -EINVAL;
	300	goto errout;
	301	}
	302	}
	303
	304	err = tp->ops->change(tp, cl, t->tcm_handle, tca, &fh);
	305	if (err == 0)
	306	tfilter_notify(skb, n, tp, fh, RTM_NEWTFILTER);
	307
	308	errout:
	309	if (cl)
	310	cops->put(q, cl);
	311	if (err == -EAGAIN)
	312	/* Replay the request. */
	313	goto replay;
	314	return err;
	315	}
	316
	317	static int
	318	tcf_fill_node(struct sk_buff skb, struct tcf_proto tp, unsigned long fh,
e431b8c0	319	u32 pid, u32 seq, u16 flags, int event)
1da177e4 LT	320	{
	321	struct tcmsg *tcm;
	322	struct nlmsghdr *nlh;
27a884dc	323	unsigned char *b = skb_tail_pointer(skb);
1da177e4	324
e431b8c0	325	nlh = NLMSG_NEW(skb, pid, seq, event, sizeof(*tcm), flags);
1da177e4 LT	326	tcm = NLMSG_DATA(nlh);
1da177e4 LT	327	tcm->tcm_family = AF_UNSPEC;
9ef1d4c7 PM	328	tcm->tcm__pad1 = 0;
9ef1d4c7 PM	329	tcm->tcm__pad1 = 0;
1da177e4 LT	330	tcm->tcm_ifindex = tp->q->dev->ifindex;
	331	tcm->tcm_parent = tp->classid;
	332	tcm->tcm_info = TC_H_MAKE(tp->prio, tp->protocol);
	333	RTA_PUT(skb, TCA_KIND, IFNAMSIZ, tp->ops->kind);
	334	tcm->tcm_handle = fh;
	335	if (RTM_DELTFILTER != event) {
	336	tcm->tcm_handle = 0;
	337	if (tp->ops->dump && tp->ops->dump(tp, fh, skb, tcm) < 0)
	338	goto rtattr_failure;
	339	}
27a884dc	340	nlh->nlmsg_len = skb_tail_pointer(skb) - b;
1da177e4 LT	341	return skb->len;
	342
	343	nlmsg_failure:
	344	rtattr_failure:
dc5fc579	345	nlmsg_trim(skb, b);
1da177e4 LT	346	return -1;
	347	}
	348
	349	static int tfilter_notify(struct sk_buff oskb, struct nlmsghdr n,
	350	struct tcf_proto *tp, unsigned long fh, int event)
	351	{
	352	struct sk_buff *skb;
	353	u32 pid = oskb ? NETLINK_CB(oskb).pid : 0;
	354
	355	skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
	356	if (!skb)
	357	return -ENOBUFS;
	358
	359	if (tcf_fill_node(skb, tp, fh, pid, n->nlmsg_seq, 0, event) <= 0) {
	360	kfree_skb(skb);
	361	return -EINVAL;
	362	}
	363
ac6d439d	364	return rtnetlink_send(skb, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
1da177e4 LT	365	}
	366
	367	struct tcf_dump_args
	368	{
	369	struct tcf_walker w;
	370	struct sk_buff *skb;
	371	struct netlink_callback *cb;
	372	};
	373
	374	static int tcf_node_dump(struct tcf_proto tp, unsigned long n, struct tcf_walker arg)
	375	{
	376	struct tcf_dump_args a = (void)arg;
	377
	378	return tcf_fill_node(a->skb, tp, n, NETLINK_CB(a->cb->skb).pid,
	379	a->cb->nlh->nlmsg_seq, NLM_F_MULTI, RTM_NEWTFILTER);
	380	}
	381
	382	static int tc_dump_tfilter(struct sk_buff skb, struct netlink_callback cb)
	383	{
b854272b	384	struct net *net = skb->sk->sk_net;
1da177e4 LT	385	int t;
	386	int s_t;
	387	struct net_device *dev;
	388	struct Qdisc *q;
	389	struct tcf_proto tp, *chain;
	390	struct tcmsg tcm = (struct tcmsg)NLMSG_DATA(cb->nlh);
	391	unsigned long cl = 0;
20fea08b	392	const struct Qdisc_class_ops *cops;
1da177e4 LT	393	struct tcf_dump_args arg;
1da177e4 LT	394
b854272b DL	395	if (net != &init_net)
	396	return 0;
	397
1da177e4 LT	398	if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm)))
1da177e4 LT	399	return skb->len;
881d966b	400	if ((dev = dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
1da177e4 LT	401	return skb->len;
1da177e4 LT	402
1da177e4 LT	403	if (!tcm->tcm_parent)
	404	q = dev->qdisc_sleeping;
	405	else
	406	q = qdisc_lookup(dev, TC_H_MAJ(tcm->tcm_parent));
	407	if (!q)
	408	goto out;
	409	if ((cops = q->ops->cl_ops) == NULL)
	410	goto errout;
	411	if (TC_H_MIN(tcm->tcm_parent)) {
	412	cl = cops->get(q, tcm->tcm_parent);
	413	if (cl == 0)
	414	goto errout;
	415	}
	416	chain = cops->tcf_chain(q, cl);
	417	if (chain == NULL)
	418	goto errout;
	419
	420	s_t = cb->args[0];
	421
	422	for (tp=*chain, t=0; tp; tp = tp->next, t++) {
	423	if (t < s_t) continue;
	424	if (TC_H_MAJ(tcm->tcm_info) &&
	425	TC_H_MAJ(tcm->tcm_info) != tp->prio)
	426	continue;
	427	if (TC_H_MIN(tcm->tcm_info) &&
	428	TC_H_MIN(tcm->tcm_info) != tp->protocol)
	429	continue;
	430	if (t > s_t)
	431	memset(&cb->args[1], 0, sizeof(cb->args)-sizeof(cb->args[0]));
	432	if (cb->args[1] == 0) {
	433	if (tcf_fill_node(skb, tp, 0, NETLINK_CB(cb->skb).pid,
	434	cb->nlh->nlmsg_seq, NLM_F_MULTI, RTM_NEWTFILTER) <= 0) {
	435	break;
	436	}
	437	cb->args[1] = 1;
	438	}
	439	if (tp->ops->walk == NULL)
	440	continue;
	441	arg.w.fn = tcf_node_dump;
	442	arg.skb = skb;
	443	arg.cb = cb;
	444	arg.w.stop = 0;
	445	arg.w.skip = cb->args[1]-1;
	446	arg.w.count = 0;
	447	tp->ops->walk(tp, &arg.w);
	448	cb->args[1] = arg.w.count+1;
	449	if (arg.w.stop)
	450	break;
	451	}
	452
	453	cb->args[0] = t;
	454
	455	errout:
	456	if (cl)
	457	cops->put(q, cl);
	458	out:
1da177e4 LT	459	dev_put(dev);
	460	return skb->len;
	461	}
	462
	463	void
	464	tcf_exts_destroy(struct tcf_proto tp, struct tcf_exts exts)
	465	{
	466	#ifdef CONFIG_NET_CLS_ACT
	467	if (exts->action) {
	468	tcf_action_destroy(exts->action, TCA_ACT_UNBIND);
	469	exts->action = NULL;
	470	}
1da177e4 LT	471	#endif
	472	}
	473
	474
	475	int
	476	tcf_exts_validate(struct tcf_proto tp, struct rtattr *tb,
10297b99 YH	477	struct rtattr rate_tlv, struct tcf_exts exts,
10297b99 YH	478	struct tcf_ext_map *map)
1da177e4 LT	479	{
1da177e4 LT	480	memset(exts, 0, sizeof(*exts));
10297b99	481
1da177e4 LT	482	#ifdef CONFIG_NET_CLS_ACT
	483	{
	484	int err;
	485	struct tc_action *act;
	486
	487	if (map->police && tb[map->police-1]) {
	488	act = tcf_action_init_1(tb[map->police-1], rate_tlv, "police",
	489	TCA_ACT_NOREPLACE, TCA_ACT_BIND, &err);
	490	if (act == NULL)
	491	return err;
	492
	493	act->type = TCA_OLD_COMPAT;
	494	exts->action = act;
	495	} else if (map->action && tb[map->action-1]) {
	496	act = tcf_action_init(tb[map->action-1], rate_tlv, NULL,
	497	TCA_ACT_NOREPLACE, TCA_ACT_BIND, &err);
	498	if (act == NULL)
	499	return err;
	500
	501	exts->action = act;
	502	}
	503	}
1da177e4 LT	504	#else
	505	if ((map->action && tb[map->action-1]) \|\|
	506	(map->police && tb[map->police-1]))
	507	return -EOPNOTSUPP;
	508	#endif
	509
	510	return 0;
	511	}
	512
	513	void
	514	tcf_exts_change(struct tcf_proto tp, struct tcf_exts dst,
10297b99	515	struct tcf_exts *src)
1da177e4 LT	516	{
	517	#ifdef CONFIG_NET_CLS_ACT
	518	if (src->action) {
	519	struct tc_action *act;
	520	tcf_tree_lock(tp);
	521	act = xchg(&dst->action, src->action);
	522	tcf_tree_unlock(tp);
	523	if (act)
	524	tcf_action_destroy(act, TCA_ACT_UNBIND);
	525	}
1da177e4 LT	526	#endif
	527	}
	528
	529	int
	530	tcf_exts_dump(struct sk_buff skb, struct tcf_exts exts,
	531	struct tcf_ext_map *map)
	532	{
	533	#ifdef CONFIG_NET_CLS_ACT
	534	if (map->action && exts->action) {
	535	/*
	536	* again for backward compatible mode - we want
	537	* to work with both old and new modes of entering
	538	* tc data even if iproute2 was newer - jhs
	539	*/
27a884dc	540	struct rtattr p_rta = (struct rtattr )skb_tail_pointer(skb);
1da177e4 LT	541
	542	if (exts->action->type != TCA_OLD_COMPAT) {
	543	RTA_PUT(skb, map->action, 0, NULL);
	544	if (tcf_action_dump(skb, exts->action, 0, 0) < 0)
	545	goto rtattr_failure;
27a884dc	546	p_rta->rta_len = skb_tail_pointer(skb) - (u8 *)p_rta;
1da177e4 LT	547	} else if (map->police) {
	548	RTA_PUT(skb, map->police, 0, NULL);
	549	if (tcf_action_dump_old(skb, exts->action, 0, 0) < 0)
	550	goto rtattr_failure;
27a884dc	551	p_rta->rta_len = skb_tail_pointer(skb) - (u8 *)p_rta;
1da177e4 LT	552	}
1da177e4 LT	553	}
1da177e4 LT	554	#endif
	555	return 0;
	556	rtattr_failure: __attribute__ ((unused))
	557	return -1;
	558	}
	559
	560	int
	561	tcf_exts_dump_stats(struct sk_buff skb, struct tcf_exts exts,
10297b99	562	struct tcf_ext_map *map)
1da177e4 LT	563	{
	564	#ifdef CONFIG_NET_CLS_ACT
	565	if (exts->action)
	566	if (tcf_action_copy_stats(skb, exts->action, 1) < 0)
	567	goto rtattr_failure;
1da177e4 LT	568	#endif
	569	return 0;
	570	rtattr_failure: __attribute__ ((unused))
	571	return -1;
	572	}
	573
	574	static int __init tc_filter_init(void)
	575	{
82623c0d TG	576	rtnl_register(PF_UNSPEC, RTM_NEWTFILTER, tc_ctl_tfilter, NULL);
	577	rtnl_register(PF_UNSPEC, RTM_DELTFILTER, tc_ctl_tfilter, NULL);
	578	rtnl_register(PF_UNSPEC, RTM_GETTFILTER, tc_ctl_tfilter,
	579	tc_dump_tfilter);
1da177e4	580
1da177e4 LT	581	return 0;
	582	}
	583
	584	subsys_initcall(tc_filter_init);
	585
	586	EXPORT_SYMBOL(register_tcf_proto_ops);
	587	EXPORT_SYMBOL(unregister_tcf_proto_ops);
	588	EXPORT_SYMBOL(tcf_exts_validate);
	589	EXPORT_SYMBOL(tcf_exts_destroy);
	590	EXPORT_SYMBOL(tcf_exts_change);
	591	EXPORT_SYMBOL(tcf_exts_dump);
	592	EXPORT_SYMBOL(tcf_exts_dump_stats);