[linux-2.6-block.git] / net / sched / cls_tcindex.c

/*
 * net/sched/cls_tcindex.c	Packet classifier for skb->tc_index
 *
 * Written 1998,1999 by Werner Almesberger, EPFL ICA
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/skbuff.h>
#include <linux/errno.h>
#include <linux/slab.h>
#include <net/act_api.h>
#include <net/netlink.h>
#include <net/pkt_cls.h>

/*
 * Passing parameters to the root seems to be done more awkwardly than really
 * necessary. At least, u32 doesn't seem to use such dirty hacks. To be
 * verified. FIXME.
 */

#define PERFECT_HASH_THRESHOLD	64	/* use perfect hash if not bigger */
#define DEFAULT_HASH_SIZE	64	/* optimized for diffserv */


struct tcindex_filter_result {
	struct tcf_exts		exts;
	struct tcf_result	res;
	struct rcu_head		rcu;
};

struct tcindex_filter {
	u16 key;
	struct tcindex_filter_result result;
	struct tcindex_filter __rcu *next;
	struct rcu_head rcu;
};


struct tcindex_data {
	struct tcindex_filter_result *perfect; /* perfect hash; NULL if none */
	struct tcindex_filter __rcu **h; /* imperfect hash; */
	struct tcf_proto *tp;
	u16 mask;		/* AND key with mask */
	u32 shift;		/* shift ANDed key to the right */
	u32 hash;		/* hash table size; 0 if undefined */
	u32 alloc_hash;		/* allocated size */
	u32 fall_through;	/* 0: only classify if explicit match */
	struct rcu_head rcu;
};

static inline int
tcindex_filter_is_set(struct tcindex_filter_result *r)
{
	return tcf_exts_is_predicative(&r->exts) || r->res.classid;
}

static struct tcindex_filter_result *
tcindex_lookup(struct tcindex_data *p, u16 key)
{
	if (p->perfect) {
		struct tcindex_filter_result *f = p->perfect + key;

		return tcindex_filter_is_set(f) ? f : NULL;
	} else if (p->h) {
		struct tcindex_filter __rcu **fp;
		struct tcindex_filter *f;

		fp = &p->h[key % p->hash];
		for (f = rcu_dereference_bh_rtnl(*fp);
		     f;
		     fp = &f->next, f = rcu_dereference_bh_rtnl(*fp))
			if (f->key == key)
				return &f->result;
	}

	return NULL;
}


static int tcindex_classify(struct sk_buff *skb, const struct tcf_proto *tp,
			    struct tcf_result *res)
{
	struct tcindex_data *p = rcu_dereference_bh(tp->root);
	struct tcindex_filter_result *f;
	int key = (skb->tc_index & p->mask) >> p->shift;

	pr_debug("tcindex_classify(skb %p,tp %p,res %p),p %p\n",
		 skb, tp, res, p);

	f = tcindex_lookup(p, key);
	if (!f) {
		if (!p->fall_through)
			return -1;
		res->classid = TC_H_MAKE(TC_H_MAJ(tp->q->handle), key);
		res->class = 0;
		pr_debug("alg 0x%x\n", res->classid);
		return 0;
	}
	*res = f->res;
	pr_debug("map 0x%x\n", res->classid);

	return tcf_exts_exec(skb, &f->exts, res);
}


static unsigned long tcindex_get(struct tcf_proto *tp, u32 handle)
{
	struct tcindex_data *p = rtnl_dereference(tp->root);
	struct tcindex_filter_result *r;

	pr_debug("tcindex_get(tp %p,handle 0x%08x)\n", tp, handle);
	if (p->perfect && handle >= p->alloc_hash)
		return 0;
	r = tcindex_lookup(p, handle);
	return r && tcindex_filter_is_set(r) ? (unsigned long) r : 0UL;
}

static int tcindex_init(struct tcf_proto *tp)
{
	struct tcindex_data *p;

	pr_debug("tcindex_init(tp %p)\n", tp);
	p = kzalloc(sizeof(struct tcindex_data), GFP_KERNEL);
	if (!p)
		return -ENOMEM;

	p->mask = 0xffff;
	p->hash = DEFAULT_HASH_SIZE;
	p->fall_through = 1;

	rcu_assign_pointer(tp->root, p);
	return 0;
}

static void tcindex_destroy_rexts(struct rcu_head *head)
{
	struct tcindex_filter_result *r;

	r = container_of(head, struct tcindex_filter_result, rcu);
	tcf_exts_destroy(&r->exts);
}

static void tcindex_destroy_fexts(struct rcu_head *head)
{
	struct tcindex_filter *f = container_of(head, struct tcindex_filter, rcu);

	tcf_exts_destroy(&f->result.exts);
	kfree(f);
}

static int tcindex_delete(struct tcf_proto *tp, unsigned long arg)
{
	struct tcindex_data *p = rtnl_dereference(tp->root);
	struct tcindex_filter_result *r = (struct tcindex_filter_result *) arg;
	struct tcindex_filter __rcu **walk;
	struct tcindex_filter *f = NULL;

	pr_debug("tcindex_delete(tp %p,arg 0x%lx),p %p\n", tp, arg, p);
	if (p->perfect) {
		if (!r->res.class)
			return -ENOENT;
	} else {
		int i;

		for (i = 0; i < p->hash; i++) {
			walk = p->h + i;
			for (f = rtnl_dereference(*walk); f;
			     walk = &f->next, f = rtnl_dereference(*walk)) {
				if (&f->result == r)
					goto found;
			}
		}
		return -ENOENT;

found:
		rcu_assign_pointer(*walk, rtnl_dereference(f->next));
	}
	tcf_unbind_filter(tp, &r->res);
	/* all classifiers are required to call tcf_exts_destroy() after rcu
	 * grace period, since converted-to-rcu actions are relying on that
	 * in cleanup() callback
	 */
	if (f)
		call_rcu(&f->rcu, tcindex_destroy_fexts);
	else
		call_rcu(&r->rcu, tcindex_destroy_rexts);
	return 0;
}

static int tcindex_destroy_element(struct tcf_proto *tp,
				   unsigned long arg,
				   struct tcf_walker *walker)
{
	return tcindex_delete(tp, arg);
}

static void __tcindex_destroy(struct rcu_head *head)
{
	struct tcindex_data *p = container_of(head, struct tcindex_data, rcu);

	kfree(p->perfect);
	kfree(p->h);
	kfree(p);
}

static inline int
valid_perfect_hash(struct tcindex_data *p)
{
	return  p->hash > (p->mask >> p->shift);
}

static const struct nla_policy tcindex_policy[TCA_TCINDEX_MAX + 1] = {
	[TCA_TCINDEX_HASH]		= { .type = NLA_U32 },
	[TCA_TCINDEX_MASK]		= { .type = NLA_U16 },
	[TCA_TCINDEX_SHIFT]		= { .type = NLA_U32 },
	[TCA_TCINDEX_FALL_THROUGH]	= { .type = NLA_U32 },
	[TCA_TCINDEX_CLASSID]		= { .type = NLA_U32 },
};

static void tcindex_filter_result_init(struct tcindex_filter_result *r)
{
	memset(r, 0, sizeof(*r));
	tcf_exts_init(&r->exts, TCA_TCINDEX_ACT, TCA_TCINDEX_POLICE);
}

static void __tcindex_partial_destroy(struct rcu_head *head)
{
	struct tcindex_data *p = container_of(head, struct tcindex_data, rcu);

	kfree(p->perfect);
	kfree(p);
}

static int
tcindex_set_parms(struct net *net, struct tcf_proto *tp, unsigned long base,
		  u32 handle, struct tcindex_data *p,
		  struct tcindex_filter_result *r, struct nlattr **tb,
		  struct nlattr *est, bool ovr)
{
	int err, balloc = 0;
	struct tcindex_filter_result new_filter_result, *old_r = r;
	struct tcindex_filter_result cr;
	struct tcindex_data *cp, *oldp;
	struct tcindex_filter *f = NULL; /* make gcc behave */
	struct tcf_exts e;

	tcf_exts_init(&e, TCA_TCINDEX_ACT, TCA_TCINDEX_POLICE);
	err = tcf_exts_validate(net, tp, tb, est, &e, ovr);
	if (err < 0)
		return err;

	err = -ENOMEM;
	/* tcindex_data attributes must look atomic to classifier/lookup so
	 * allocate new tcindex data and RCU assign it onto root. Keeping
	 * perfect hash and hash pointers from old data.
	 */
	cp = kzalloc(sizeof(*cp), GFP_KERNEL);
	if (!cp)
		goto errout;

	cp->mask = p->mask;
	cp->shift = p->shift;
	cp->hash = p->hash;
	cp->alloc_hash = p->alloc_hash;
	cp->fall_through = p->fall_through;
	cp->tp = tp;

	if (p->perfect) {
		int i;

		cp->perfect = kmemdup(p->perfect,
				      sizeof(*r) * cp->hash, GFP_KERNEL);
		if (!cp->perfect)
			goto errout;
		for (i = 0; i < cp->hash; i++)
			tcf_exts_init(&cp->perfect[i].exts,
				      TCA_TCINDEX_ACT, TCA_TCINDEX_POLICE);
		balloc = 1;
	}
	cp->h = p->h;

	tcindex_filter_result_init(&new_filter_result);
	tcindex_filter_result_init(&cr);
	if (old_r)
		cr.res = r->res;

	if (tb[TCA_TCINDEX_HASH])
		cp->hash = nla_get_u32(tb[TCA_TCINDEX_HASH]);

	if (tb[TCA_TCINDEX_MASK])
		cp->mask = nla_get_u16(tb[TCA_TCINDEX_MASK]);

	if (tb[TCA_TCINDEX_SHIFT])
		cp->shift = nla_get_u32(tb[TCA_TCINDEX_SHIFT]);

	err = -EBUSY;

	/* Hash already allocated, make sure that we still meet the
	 * requirements for the allocated hash.
	 */
	if (cp->perfect) {
		if (!valid_perfect_hash(cp) ||
		    cp->hash > cp->alloc_hash)
			goto errout_alloc;
	} else if (cp->h && cp->hash != cp->alloc_hash) {
		goto errout_alloc;
	}

	err = -EINVAL;
	if (tb[TCA_TCINDEX_FALL_THROUGH])
		cp->fall_through = nla_get_u32(tb[TCA_TCINDEX_FALL_THROUGH]);

	if (!cp->hash) {
		/* Hash not specified, use perfect hash if the upper limit
		 * of the hashing index is below the threshold.
		 */
		if ((cp->mask >> cp->shift) < PERFECT_HASH_THRESHOLD)
			cp->hash = (cp->mask >> cp->shift) + 1;
		else
			cp->hash = DEFAULT_HASH_SIZE;
	}

	if (!cp->perfect && !cp->h)
		cp->alloc_hash = cp->hash;

	/* Note: this could be as restrictive as if (handle & ~(mask >> shift))
	 * but then, we'd fail handles that may become valid after some future
	 * mask change. While this is extremely unlikely to ever matter,
	 * the check below is safer (and also more backwards-compatible).
	 */
	if (cp->perfect || valid_perfect_hash(cp))
		if (handle >= cp->alloc_hash)
			goto errout_alloc;


	err = -ENOMEM;
	if (!cp->perfect && !cp->h) {
		if (valid_perfect_hash(cp)) {
			int i;

			cp->perfect = kcalloc(cp->hash, sizeof(*r), GFP_KERNEL);
			if (!cp->perfect)
				goto errout_alloc;
			for (i = 0; i < cp->hash; i++)
				tcf_exts_init(&cp->perfect[i].exts,
					      TCA_TCINDEX_ACT,
					      TCA_TCINDEX_POLICE);
			balloc = 1;
		} else {
			struct tcindex_filter __rcu **hash;

			hash = kcalloc(cp->hash,
				       sizeof(struct tcindex_filter *),
				       GFP_KERNEL);

			if (!hash)
				goto errout_alloc;

			cp->h = hash;
			balloc = 2;
		}
	}

	if (cp->perfect)
		r = cp->perfect + handle;
	else
		r = tcindex_lookup(cp, handle) ? : &new_filter_result;

	if (r == &new_filter_result) {
		f = kzalloc(sizeof(*f), GFP_KERNEL);
		if (!f)
			goto errout_alloc;
		f->key = handle;
		tcindex_filter_result_init(&f->result);
		f->next = NULL;
	}

	if (tb[TCA_TCINDEX_CLASSID]) {
		cr.res.classid = nla_get_u32(tb[TCA_TCINDEX_CLASSID]);
		tcf_bind_filter(tp, &cr.res, base);
	}

	if (old_r)
		tcf_exts_change(tp, &r->exts, &e);
	else
		tcf_exts_change(tp, &cr.exts, &e);

	if (old_r && old_r != r)
		tcindex_filter_result_init(old_r);

	oldp = p;
	r->res = cr.res;
	rcu_assign_pointer(tp->root, cp);

	if (r == &new_filter_result) {
		struct tcindex_filter *nfp;
		struct tcindex_filter __rcu **fp;

		tcf_exts_change(tp, &f->result.exts, &r->exts);

		fp = cp->h + (handle % cp->hash);
		for (nfp = rtnl_dereference(*fp);
		     nfp;
		     fp = &nfp->next, nfp = rtnl_dereference(*fp))
				; /* nothing */

		rcu_assign_pointer(*fp, f);
	}

	if (oldp)
		call_rcu(&oldp->rcu, __tcindex_partial_destroy);
	return 0;

errout_alloc:
	if (balloc == 1)
		kfree(cp->perfect);
	else if (balloc == 2)
		kfree(cp->h);
errout:
	kfree(cp);
	tcf_exts_destroy(&e);
	return err;
}

static int
tcindex_change(struct net *net, struct sk_buff *in_skb,
	       struct tcf_proto *tp, unsigned long base, u32 handle,
	       struct nlattr **tca, unsigned long *arg, bool ovr)
{
	struct nlattr *opt = tca[TCA_OPTIONS];
	struct nlattr *tb[TCA_TCINDEX_MAX + 1];
	struct tcindex_data *p = rtnl_dereference(tp->root);
	struct tcindex_filter_result *r = (struct tcindex_filter_result *) *arg;
	int err;

	pr_debug("tcindex_change(tp %p,handle 0x%08x,tca %p,arg %p),opt %p,"
	    "p %p,r %p,*arg 0x%lx\n",
	    tp, handle, tca, arg, opt, p, r, arg ? *arg : 0L);

	if (!opt)
		return 0;

	err = nla_parse_nested(tb, TCA_TCINDEX_MAX, opt, tcindex_policy);
	if (err < 0)
		return err;

	return tcindex_set_parms(net, tp, base, handle, p, r, tb,
				 tca[TCA_RATE], ovr);
}

static void tcindex_walk(struct tcf_proto *tp, struct tcf_walker *walker)
{
	struct tcindex_data *p = rtnl_dereference(tp->root);
	struct tcindex_filter *f, *next;
	int i;

	pr_debug("tcindex_walk(tp %p,walker %p),p %p\n", tp, walker, p);
	if (p->perfect) {
		for (i = 0; i < p->hash; i++) {
			if (!p->perfect[i].res.class)
				continue;
			if (walker->count >= walker->skip) {
				if (walker->fn(tp,
				    (unsigned long) (p->perfect+i), walker)
				     < 0) {
					walker->stop = 1;
					return;
				}
			}
			walker->count++;
		}
	}
	if (!p->h)
		return;
	for (i = 0; i < p->hash; i++) {
		for (f = rtnl_dereference(p->h[i]); f; f = next) {
			next = rtnl_dereference(f->next);
			if (walker->count >= walker->skip) {
				if (walker->fn(tp, (unsigned long) &f->result,
				    walker) < 0) {
					walker->stop = 1;
					return;
				}
			}
			walker->count++;
		}
	}
}

static bool tcindex_destroy(struct tcf_proto *tp, bool force)
{
	struct tcindex_data *p = rtnl_dereference(tp->root);
	struct tcf_walker walker;

	if (!force)
		return false;

	pr_debug("tcindex_destroy(tp %p),p %p\n", tp, p);
	walker.count = 0;
	walker.skip = 0;
	walker.fn = tcindex_destroy_element;
	tcindex_walk(tp, &walker);

	RCU_INIT_POINTER(tp->root, NULL);
	call_rcu(&p->rcu, __tcindex_destroy);
	return true;
}


static int tcindex_dump(struct net *net, struct tcf_proto *tp, unsigned long fh,
    struct sk_buff *skb, struct tcmsg *t)
{
	struct tcindex_data *p = rtnl_dereference(tp->root);
	struct tcindex_filter_result *r = (struct tcindex_filter_result *) fh;
	struct nlattr *nest;

	pr_debug("tcindex_dump(tp %p,fh 0x%lx,skb %p,t %p),p %p,r %p\n",
		 tp, fh, skb, t, p, r);
	pr_debug("p->perfect %p p->h %p\n", p->perfect, p->h);

	nest = nla_nest_start(skb, TCA_OPTIONS);
	if (nest == NULL)
		goto nla_put_failure;

	if (!fh) {
		t->tcm_handle = ~0; /* whatever ... */
		if (nla_put_u32(skb, TCA_TCINDEX_HASH, p->hash) ||
		    nla_put_u16(skb, TCA_TCINDEX_MASK, p->mask) ||
		    nla_put_u32(skb, TCA_TCINDEX_SHIFT, p->shift) ||
		    nla_put_u32(skb, TCA_TCINDEX_FALL_THROUGH, p->fall_through))
			goto nla_put_failure;
		nla_nest_end(skb, nest);
	} else {
		if (p->perfect) {
			t->tcm_handle = r - p->perfect;
		} else {
			struct tcindex_filter *f;
			struct tcindex_filter __rcu **fp;
			int i;

			t->tcm_handle = 0;
			for (i = 0; !t->tcm_handle && i < p->hash; i++) {
				fp = &p->h[i];
				for (f = rtnl_dereference(*fp);
				     !t->tcm_handle && f;
				     fp = &f->next, f = rtnl_dereference(*fp)) {
					if (&f->result == r)
						t->tcm_handle = f->key;
				}
			}
		}
		pr_debug("handle = %d\n", t->tcm_handle);
		if (r->res.class &&
		    nla_put_u32(skb, TCA_TCINDEX_CLASSID, r->res.classid))
			goto nla_put_failure;

		if (tcf_exts_dump(skb, &r->exts) < 0)
			goto nla_put_failure;
		nla_nest_end(skb, nest);

		if (tcf_exts_dump_stats(skb, &r->exts) < 0)
			goto nla_put_failure;
	}

	return skb->len;

nla_put_failure:
	nla_nest_cancel(skb, nest);
	return -1;
}

static struct tcf_proto_ops cls_tcindex_ops __read_mostly = {
	.kind		=	"tcindex",
	.classify	=	tcindex_classify,
	.init		=	tcindex_init,
	.destroy	=	tcindex_destroy,
	.get		=	tcindex_get,
	.change		=	tcindex_change,
	.delete		=	tcindex_delete,
	.walk		=	tcindex_walk,
	.dump		=	tcindex_dump,
	.owner		=	THIS_MODULE,
};

static int __init init_tcindex(void)
{
	return register_tcf_proto_ops(&cls_tcindex_ops);
}

static void __exit exit_tcindex(void)
{
	unregister_tcf_proto_ops(&cls_tcindex_ops);
}

module_init(init_tcindex)
module_exit(exit_tcindex)
MODULE_LICENSE("GPL");
Commit	Line	Data
1da177e4 LT	1	/*
	2	* net/sched/cls_tcindex.c Packet classifier for skb->tc_index
	3	*
	4	* Written 1998,1999 by Werner Almesberger, EPFL ICA
	5	*/
	6
1da177e4 LT	7	#include <linux/module.h>
	8	#include <linux/types.h>
	9	#include <linux/kernel.h>
	10	#include <linux/skbuff.h>
	11	#include <linux/errno.h>
5a0e3ad6	12	#include <linux/slab.h>
1da177e4	13	#include <net/act_api.h>
dc5fc579	14	#include <net/netlink.h>
1da177e4	15	#include <net/pkt_cls.h>
1da177e4	16
1da177e4 LT	17	/*
	18	* Passing parameters to the root seems to be done more awkwardly than really
	19	* necessary. At least, u32 doesn't seem to use such dirty hacks. To be
	20	* verified. FIXME.
	21	*/
	22
	23	#define PERFECT_HASH_THRESHOLD 64 /* use perfect hash if not bigger */
	24	#define DEFAULT_HASH_SIZE 64 /* optimized for diffserv */
	25
	26
1da177e4 LT	27	struct tcindex_filter_result {
	28	struct tcf_exts exts;
	29	struct tcf_result res;
ed7aa879	30	struct rcu_head rcu;
1da177e4 LT	31	};
	32
	33	struct tcindex_filter {
	34	u16 key;
	35	struct tcindex_filter_result result;
331b7292 JF	36	struct tcindex_filter __rcu *next;
331b7292 JF	37	struct rcu_head rcu;
1da177e4 LT	38	};
	39
	40
	41	struct tcindex_data {
	42	struct tcindex_filter_result perfect; / perfect hash; NULL if none */
331b7292 JF	43	struct tcindex_filter __rcu *h; / imperfect hash; */
331b7292 JF	44	struct tcf_proto *tp;
1da177e4	45	u16 mask; /* AND key with mask */
331b7292 JF	46	u32 shift; /* shift ANDed key to the right */
	47	u32 hash; /* hash table size; 0 if undefined */
	48	u32 alloc_hash; /* allocated size */
	49	u32 fall_through; /* 0: only classify if explicit match */
	50	struct rcu_head rcu;
1da177e4 LT	51	};
1da177e4 LT	52
1da177e4 LT	53	static inline int
	54	tcindex_filter_is_set(struct tcindex_filter_result *r)
	55	{
	56	return tcf_exts_is_predicative(&r->exts) \|\| r->res.classid;
	57	}
	58
	59	static struct tcindex_filter_result *
	60	tcindex_lookup(struct tcindex_data *p, u16 key)
	61	{
331b7292 JF	62	if (p->perfect) {
	63	struct tcindex_filter_result *f = p->perfect + key;
	64
	65	return tcindex_filter_is_set(f) ? f : NULL;
	66	} else if (p->h) {
	67	struct tcindex_filter __rcu **fp;
	68	struct tcindex_filter *f;
1da177e4	69
331b7292 JF	70	fp = &p->h[key % p->hash];
	71	for (f = rcu_dereference_bh_rtnl(*fp);
	72	f;
	73	fp = &f->next, f = rcu_dereference_bh_rtnl(*fp))
1da177e4 LT	74	if (f->key == key)
	75	return &f->result;
	76	}
	77
	78	return NULL;
	79	}
	80
	81
dc7f9f6e	82	static int tcindex_classify(struct sk_buff skb, const struct tcf_proto tp,
1da177e4 LT	83	struct tcf_result *res)
1da177e4 LT	84	{
2f9a220e	85	struct tcindex_data *p = rcu_dereference_bh(tp->root);
1da177e4 LT	86	struct tcindex_filter_result *f;
	87	int key = (skb->tc_index & p->mask) >> p->shift;
	88
aa767bfe SH	89	pr_debug("tcindex_classify(skb %p,tp %p,res %p),p %p\n",
aa767bfe SH	90	skb, tp, res, p);
1da177e4 LT	91
	92	f = tcindex_lookup(p, key);
	93	if (!f) {
	94	if (!p->fall_through)
	95	return -1;
	96	res->classid = TC_H_MAKE(TC_H_MAJ(tp->q->handle), key);
	97	res->class = 0;
aa767bfe	98	pr_debug("alg 0x%x\n", res->classid);
1da177e4 LT	99	return 0;
	100	}
	101	*res = f->res;
aa767bfe	102	pr_debug("map 0x%x\n", res->classid);
1da177e4 LT	103
	104	return tcf_exts_exec(skb, &f->exts, res);
	105	}
	106
	107
	108	static unsigned long tcindex_get(struct tcf_proto *tp, u32 handle)
	109	{
331b7292	110	struct tcindex_data *p = rtnl_dereference(tp->root);
1da177e4 LT	111	struct tcindex_filter_result *r;
1da177e4 LT	112
aa767bfe	113	pr_debug("tcindex_get(tp %p,handle 0x%08x)\n", tp, handle);
1da177e4 LT	114	if (p->perfect && handle >= p->alloc_hash)
	115	return 0;
	116	r = tcindex_lookup(p, handle);
	117	return r && tcindex_filter_is_set(r) ? (unsigned long) r : 0UL;
	118	}
	119
1da177e4 LT	120	static int tcindex_init(struct tcf_proto *tp)
	121	{
	122	struct tcindex_data *p;
	123
aa767bfe SH	124	pr_debug("tcindex_init(tp %p)\n", tp);
aa767bfe SH	125	p = kzalloc(sizeof(struct tcindex_data), GFP_KERNEL);
1da177e4 LT	126	if (!p)
	127	return -ENOMEM;
	128
1da177e4 LT	129	p->mask = 0xffff;
	130	p->hash = DEFAULT_HASH_SIZE;
	131	p->fall_through = 1;
	132
331b7292	133	rcu_assign_pointer(tp->root, p);
1da177e4 LT	134	return 0;
	135	}
	136
ed7aa879 AS	137	static void tcindex_destroy_rexts(struct rcu_head *head)
	138	{
	139	struct tcindex_filter_result *r;
	140
	141	r = container_of(head, struct tcindex_filter_result, rcu);
	142	tcf_exts_destroy(&r->exts);
	143	}
	144
	145	static void tcindex_destroy_fexts(struct rcu_head *head)
	146	{
	147	struct tcindex_filter *f = container_of(head, struct tcindex_filter, rcu);
	148
	149	tcf_exts_destroy(&f->result.exts);
	150	kfree(f);
	151	}
	152
	153	static int tcindex_delete(struct tcf_proto *tp, unsigned long arg)
1da177e4	154	{
331b7292	155	struct tcindex_data *p = rtnl_dereference(tp->root);
1da177e4	156	struct tcindex_filter_result r = (struct tcindex_filter_result ) arg;
331b7292	157	struct tcindex_filter __rcu **walk;
1da177e4 LT	158	struct tcindex_filter *f = NULL;
1da177e4 LT	159
331b7292	160	pr_debug("tcindex_delete(tp %p,arg 0x%lx),p %p\n", tp, arg, p);
1da177e4 LT	161	if (p->perfect) {
	162	if (!r->res.class)
	163	return -ENOENT;
	164	} else {
	165	int i;
1da177e4	166
331b7292 JF	167	for (i = 0; i < p->hash; i++) {
	168	walk = p->h + i;
	169	for (f = rtnl_dereference(*walk); f;
	170	walk = &f->next, f = rtnl_dereference(*walk)) {
	171	if (&f->result == r)
1da177e4	172	goto found;
331b7292 JF	173	}
331b7292 JF	174	}
1da177e4 LT	175	return -ENOENT;
	176
	177	found:
331b7292	178	rcu_assign_pointer(*walk, rtnl_dereference(f->next));
1da177e4 LT	179	}
1da177e4 LT	180	tcf_unbind_filter(tp, &r->res);
ed7aa879 AS	181	/* all classifiers are required to call tcf_exts_destroy() after rcu
	182	* grace period, since converted-to-rcu actions are relying on that
	183	* in cleanup() callback
	184	*/
331b7292	185	if (f)
ed7aa879 AS	186	call_rcu(&f->rcu, tcindex_destroy_fexts);
	187	else
	188	call_rcu(&r->rcu, tcindex_destroy_rexts);
1da177e4 LT	189	return 0;
	190	}
	191
331b7292 JF	192	static int tcindex_destroy_element(struct tcf_proto *tp,
	193	unsigned long arg,
	194	struct tcf_walker *walker)
1da177e4	195	{
331b7292 JF	196	return tcindex_delete(tp, arg);
	197	}
	198
	199	static void __tcindex_destroy(struct rcu_head *head)
	200	{
	201	struct tcindex_data *p = container_of(head, struct tcindex_data, rcu);
	202
	203	kfree(p->perfect);
	204	kfree(p->h);
	205	kfree(p);
1da177e4 LT	206	}
	207
	208	static inline int
	209	valid_perfect_hash(struct tcindex_data *p)
	210	{
	211	return p->hash > (p->mask >> p->shift);
	212	}
	213
6fa8c014 PM	214	static const struct nla_policy tcindex_policy[TCA_TCINDEX_MAX + 1] = {
	215	[TCA_TCINDEX_HASH] = { .type = NLA_U32 },
	216	[TCA_TCINDEX_MASK] = { .type = NLA_U16 },
	217	[TCA_TCINDEX_SHIFT] = { .type = NLA_U32 },
	218	[TCA_TCINDEX_FALL_THROUGH] = { .type = NLA_U32 },
	219	[TCA_TCINDEX_CLASSID] = { .type = NLA_U32 },
	220	};
	221
bf63ac73 CW	222	static void tcindex_filter_result_init(struct tcindex_filter_result *r)
	223	{
	224	memset(r, 0, sizeof(*r));
	225	tcf_exts_init(&r->exts, TCA_TCINDEX_ACT, TCA_TCINDEX_POLICE);
	226	}
	227
331b7292 JF	228	static void __tcindex_partial_destroy(struct rcu_head *head)
	229	{
	230	struct tcindex_data *p = container_of(head, struct tcindex_data, rcu);
	231
	232	kfree(p->perfect);
	233	kfree(p);
	234	}
	235
1da177e4	236	static int
c1b52739 BL	237	tcindex_set_parms(struct net net, struct tcf_proto tp, unsigned long base,
	238	u32 handle, struct tcindex_data *p,
	239	struct tcindex_filter_result r, struct nlattr *tb,
2f7ef2f8	240	struct nlattr *est, bool ovr)
1da177e4 LT	241	{
	242	int err, balloc = 0;
	243	struct tcindex_filter_result new_filter_result, *old_r = r;
	244	struct tcindex_filter_result cr;
331b7292	245	struct tcindex_data cp, oldp;
1da177e4 LT	246	struct tcindex_filter f = NULL; / make gcc behave */
	247	struct tcf_exts e;
	248
5da57f42	249	tcf_exts_init(&e, TCA_TCINDEX_ACT, TCA_TCINDEX_POLICE);
2f7ef2f8	250	err = tcf_exts_validate(net, tp, tb, est, &e, ovr);
1da177e4 LT	251	if (err < 0)
1da177e4 LT	252	return err;
10297b99	253
02c5e844	254	err = -ENOMEM;
331b7292 JF	255	/* tcindex_data attributes must look atomic to classifier/lookup so
	256	* allocate new tcindex data and RCU assign it onto root. Keeping
	257	* perfect hash and hash pointers from old data.
	258	*/
a57a65ba	259	cp = kzalloc(sizeof(*cp), GFP_KERNEL);
02c5e844	260	if (!cp)
44b75e43	261	goto errout;
331b7292 JF	262
	263	cp->mask = p->mask;
	264	cp->shift = p->shift;
	265	cp->hash = p->hash;
	266	cp->alloc_hash = p->alloc_hash;
	267	cp->fall_through = p->fall_through;
	268	cp->tp = tp;
	269
	270	if (p->perfect) {
6e056569 WC	271	int i;
6e056569 WC	272
331b7292 JF	273	cp->perfect = kmemdup(p->perfect,
	274	sizeof(r) cp->hash, GFP_KERNEL);
	275	if (!cp->perfect)
	276	goto errout;
6e056569 WC	277	for (i = 0; i < cp->hash; i++)
	278	tcf_exts_init(&cp->perfect[i].exts,
	279	TCA_TCINDEX_ACT, TCA_TCINDEX_POLICE);
44b75e43	280	balloc = 1;
331b7292 JF	281	}
	282	cp->h = p->h;
	283
10ee1c34	284	tcindex_filter_result_init(&new_filter_result);
bf63ac73	285	tcindex_filter_result_init(&cr);
1da177e4	286	if (old_r)
bf63ac73	287	cr.res = r->res;
1da177e4	288
6fa8c014	289	if (tb[TCA_TCINDEX_HASH])
331b7292	290	cp->hash = nla_get_u32(tb[TCA_TCINDEX_HASH]);
1da177e4	291
6fa8c014	292	if (tb[TCA_TCINDEX_MASK])
331b7292	293	cp->mask = nla_get_u16(tb[TCA_TCINDEX_MASK]);
1da177e4	294
6fa8c014	295	if (tb[TCA_TCINDEX_SHIFT])
331b7292	296	cp->shift = nla_get_u32(tb[TCA_TCINDEX_SHIFT]);
1da177e4 LT	297
1da177e4 LT	298	err = -EBUSY;
331b7292	299
1da177e4 LT	300	/* Hash already allocated, make sure that we still meet the
	301	* requirements for the allocated hash.
	302	*/
331b7292 JF	303	if (cp->perfect) {
	304	if (!valid_perfect_hash(cp) \|\|
	305	cp->hash > cp->alloc_hash)
44b75e43	306	goto errout_alloc;
331b7292	307	} else if (cp->h && cp->hash != cp->alloc_hash) {
44b75e43	308	goto errout_alloc;
331b7292	309	}
1da177e4 LT	310
1da177e4 LT	311	err = -EINVAL;
6fa8c014	312	if (tb[TCA_TCINDEX_FALL_THROUGH])
331b7292	313	cp->fall_through = nla_get_u32(tb[TCA_TCINDEX_FALL_THROUGH]);
1da177e4	314
331b7292	315	if (!cp->hash) {
1da177e4 LT	316	/* Hash not specified, use perfect hash if the upper limit
	317	* of the hashing index is below the threshold.
	318	*/
331b7292 JF	319	if ((cp->mask >> cp->shift) < PERFECT_HASH_THRESHOLD)
331b7292 JF	320	cp->hash = (cp->mask >> cp->shift) + 1;
1da177e4	321	else
331b7292	322	cp->hash = DEFAULT_HASH_SIZE;
1da177e4 LT	323	}
1da177e4 LT	324
68f6a7c6	325	if (!cp->perfect && !cp->h)
331b7292	326	cp->alloc_hash = cp->hash;
1da177e4 LT	327
	328	/* Note: this could be as restrictive as if (handle & ~(mask >> shift))
	329	* but then, we'd fail handles that may become valid after some future
	330	* mask change. While this is extremely unlikely to ever matter,
	331	* the check below is safer (and also more backwards-compatible).
	332	*/
331b7292 JF	333	if (cp->perfect \|\| valid_perfect_hash(cp))
331b7292 JF	334	if (handle >= cp->alloc_hash)
44b75e43	335	goto errout_alloc;
1da177e4 LT	336
	337
	338	err = -ENOMEM;
331b7292 JF	339	if (!cp->perfect && !cp->h) {
331b7292 JF	340	if (valid_perfect_hash(cp)) {
bf63ac73 CW	341	int i;
bf63ac73 CW	342
331b7292 JF	343	cp->perfect = kcalloc(cp->hash, sizeof(*r), GFP_KERNEL);
331b7292 JF	344	if (!cp->perfect)
44b75e43	345	goto errout_alloc;
331b7292 JF	346	for (i = 0; i < cp->hash; i++)
	347	tcf_exts_init(&cp->perfect[i].exts,
	348	TCA_TCINDEX_ACT,
bf63ac73	349	TCA_TCINDEX_POLICE);
1da177e4 LT	350	balloc = 1;
1da177e4 LT	351	} else {
331b7292 JF	352	struct tcindex_filter __rcu **hash;
	353
	354	hash = kcalloc(cp->hash,
	355	sizeof(struct tcindex_filter *),
	356	GFP_KERNEL);
	357
	358	if (!hash)
44b75e43	359	goto errout_alloc;
331b7292 JF	360
331b7292 JF	361	cp->h = hash;
1da177e4 LT	362	balloc = 2;
	363	}
	364	}
	365
331b7292 JF	366	if (cp->perfect)
331b7292 JF	367	r = cp->perfect + handle;
1da177e4	368	else
331b7292	369	r = tcindex_lookup(cp, handle) ? : &new_filter_result;
1da177e4 LT	370
1da177e4 LT	371	if (r == &new_filter_result) {
0da974f4	372	f = kzalloc(sizeof(*f), GFP_KERNEL);
1da177e4 LT	373	if (!f)
1da177e4 LT	374	goto errout_alloc;
6e056569 WC	375	f->key = handle;
	376	tcindex_filter_result_init(&f->result);
	377	f->next = NULL;
10297b99	378	}
1da177e4	379
add93b61	380	if (tb[TCA_TCINDEX_CLASSID]) {
1587bac4	381	cr.res.classid = nla_get_u32(tb[TCA_TCINDEX_CLASSID]);
1da177e4	382	tcf_bind_filter(tp, &cr.res, base);
10297b99	383	}
1da177e4	384
bf63ac73 CW	385	if (old_r)
	386	tcf_exts_change(tp, &r->exts, &e);
	387	else
	388	tcf_exts_change(tp, &cr.exts, &e);
1da177e4	389
1da177e4	390	if (old_r && old_r != r)
bf63ac73	391	tcindex_filter_result_init(old_r);
1da177e4	392
331b7292	393	oldp = p;
bf63ac73	394	r->res = cr.res;
331b7292	395	rcu_assign_pointer(tp->root, cp);
1da177e4 LT	396
1da177e4 LT	397	if (r == &new_filter_result) {
331b7292 JF	398	struct tcindex_filter *nfp;
331b7292 JF	399	struct tcindex_filter __rcu **fp;
1da177e4	400
6e056569	401	tcf_exts_change(tp, &f->result.exts, &r->exts);
331b7292	402
69301eaa	403	fp = cp->h + (handle % cp->hash);
331b7292 JF	404	for (nfp = rtnl_dereference(*fp);
	405	nfp;
	406	fp = &nfp->next, nfp = rtnl_dereference(*fp))
	407	; /* nothing */
	408
	409	rcu_assign_pointer(*fp, f);
10297b99	410	}
1da177e4	411
331b7292 JF	412	if (oldp)
331b7292 JF	413	call_rcu(&oldp->rcu, __tcindex_partial_destroy);
1da177e4 LT	414	return 0;
	415
	416	errout_alloc:
	417	if (balloc == 1)
331b7292	418	kfree(cp->perfect);
1da177e4	419	else if (balloc == 2)
331b7292	420	kfree(cp->h);
1da177e4	421	errout:
331b7292	422	kfree(cp);
18d0264f	423	tcf_exts_destroy(&e);
1da177e4 LT	424	return err;
	425	}
	426
	427	static int
c1b52739	428	tcindex_change(struct net net, struct sk_buff in_skb,
af4c6641	429	struct tcf_proto *tp, unsigned long base, u32 handle,
2f7ef2f8	430	struct nlattr *tca, unsigned long arg, bool ovr)
1da177e4	431	{
add93b61 PM	432	struct nlattr *opt = tca[TCA_OPTIONS];
add93b61 PM	433	struct nlattr *tb[TCA_TCINDEX_MAX + 1];
331b7292	434	struct tcindex_data *p = rtnl_dereference(tp->root);
1da177e4	435	struct tcindex_filter_result r = (struct tcindex_filter_result ) *arg;
cee63723	436	int err;
1da177e4	437
aa767bfe	438	pr_debug("tcindex_change(tp %p,handle 0x%08x,tca %p,arg %p),opt %p,"
1da177e4 LT	439	"p %p,r %p,*arg 0x%lx\n",
	440	tp, handle, tca, arg, opt, p, r, arg ? *arg : 0L);
	441
	442	if (!opt)
	443	return 0;
	444
6fa8c014	445	err = nla_parse_nested(tb, TCA_TCINDEX_MAX, opt, tcindex_policy);
cee63723 PM	446	if (err < 0)
cee63723 PM	447	return err;
1da177e4	448
c1b52739	449	return tcindex_set_parms(net, tp, base, handle, p, r, tb,
2f7ef2f8	450	tca[TCA_RATE], ovr);
1da177e4 LT	451	}
1da177e4 LT	452
1da177e4 LT	453	static void tcindex_walk(struct tcf_proto tp, struct tcf_walker walker)
1da177e4 LT	454	{
331b7292	455	struct tcindex_data *p = rtnl_dereference(tp->root);
aa767bfe	456	struct tcindex_filter f, next;
1da177e4 LT	457	int i;
1da177e4 LT	458
aa767bfe	459	pr_debug("tcindex_walk(tp %p,walker %p),p %p\n", tp, walker, p);
1da177e4 LT	460	if (p->perfect) {
	461	for (i = 0; i < p->hash; i++) {
	462	if (!p->perfect[i].res.class)
	463	continue;
	464	if (walker->count >= walker->skip) {
	465	if (walker->fn(tp,
	466	(unsigned long) (p->perfect+i), walker)
	467	< 0) {
	468	walker->stop = 1;
	469	return;
	470	}
	471	}
	472	walker->count++;
	473	}
	474	}
	475	if (!p->h)
	476	return;
	477	for (i = 0; i < p->hash; i++) {
331b7292 JF	478	for (f = rtnl_dereference(p->h[i]); f; f = next) {
331b7292 JF	479	next = rtnl_dereference(f->next);
1da177e4	480	if (walker->count >= walker->skip) {
aa767bfe	481	if (walker->fn(tp, (unsigned long) &f->result,
1da177e4 LT	482	walker) < 0) {
	483	walker->stop = 1;
	484	return;
	485	}
	486	}
	487	walker->count++;
	488	}
	489	}
	490	}
	491
1e052be6	492	static bool tcindex_destroy(struct tcf_proto *tp, bool force)
1da177e4	493	{
331b7292	494	struct tcindex_data *p = rtnl_dereference(tp->root);
1da177e4 LT	495	struct tcf_walker walker;
1da177e4 LT	496
1e052be6 CW	497	if (!force)
	498	return false;
	499
aa767bfe	500	pr_debug("tcindex_destroy(tp %p),p %p\n", tp, p);
1da177e4 LT	501	walker.count = 0;
1da177e4 LT	502	walker.skip = 0;
e40f5c72	503	walker.fn = tcindex_destroy_element;
aa767bfe	504	tcindex_walk(tp, &walker);
331b7292 JF	505
	506	RCU_INIT_POINTER(tp->root, NULL);
	507	call_rcu(&p->rcu, __tcindex_destroy);
1e052be6	508	return true;
1da177e4 LT	509	}
	510
	511
832d1d5b	512	static int tcindex_dump(struct net net, struct tcf_proto tp, unsigned long fh,
1da177e4 LT	513	struct sk_buff skb, struct tcmsg t)
1da177e4 LT	514	{
331b7292	515	struct tcindex_data *p = rtnl_dereference(tp->root);
1da177e4	516	struct tcindex_filter_result r = (struct tcindex_filter_result ) fh;
4b3550ef	517	struct nlattr *nest;
1da177e4	518
6ea3b446 JP	519	pr_debug("tcindex_dump(tp %p,fh 0x%lx,skb %p,t %p),p %p,r %p\n",
6ea3b446 JP	520	tp, fh, skb, t, p, r);
aa767bfe	521	pr_debug("p->perfect %p p->h %p\n", p->perfect, p->h);
4b3550ef PM	522
	523	nest = nla_nest_start(skb, TCA_OPTIONS);
	524	if (nest == NULL)
	525	goto nla_put_failure;
	526
1da177e4 LT	527	if (!fh) {
1da177e4 LT	528	t->tcm_handle = ~0; /* whatever ... */
1b34ec43 DM	529	if (nla_put_u32(skb, TCA_TCINDEX_HASH, p->hash) \|\|
	530	nla_put_u16(skb, TCA_TCINDEX_MASK, p->mask) \|\|
	531	nla_put_u32(skb, TCA_TCINDEX_SHIFT, p->shift) \|\|
	532	nla_put_u32(skb, TCA_TCINDEX_FALL_THROUGH, p->fall_through))
	533	goto nla_put_failure;
4b3550ef	534	nla_nest_end(skb, nest);
1da177e4 LT	535	} else {
1da177e4 LT	536	if (p->perfect) {
331b7292	537	t->tcm_handle = r - p->perfect;
1da177e4 LT	538	} else {
1da177e4 LT	539	struct tcindex_filter *f;
331b7292	540	struct tcindex_filter __rcu **fp;
1da177e4 LT	541	int i;
	542
	543	t->tcm_handle = 0;
	544	for (i = 0; !t->tcm_handle && i < p->hash; i++) {
331b7292 JF	545	fp = &p->h[i];
	546	for (f = rtnl_dereference(*fp);
	547	!t->tcm_handle && f;
	548	fp = &f->next, f = rtnl_dereference(*fp)) {
1da177e4 LT	549	if (&f->result == r)
	550	t->tcm_handle = f->key;
	551	}
	552	}
	553	}
aa767bfe	554	pr_debug("handle = %d\n", t->tcm_handle);
1b34ec43 DM	555	if (r->res.class &&
	556	nla_put_u32(skb, TCA_TCINDEX_CLASSID, r->res.classid))
	557	goto nla_put_failure;
1da177e4	558
5da57f42	559	if (tcf_exts_dump(skb, &r->exts) < 0)
add93b61	560	goto nla_put_failure;
4b3550ef	561	nla_nest_end(skb, nest);
1da177e4	562
5da57f42	563	if (tcf_exts_dump_stats(skb, &r->exts) < 0)
add93b61	564	goto nla_put_failure;
1da177e4	565	}
10297b99	566
1da177e4 LT	567	return skb->len;
1da177e4 LT	568
add93b61	569	nla_put_failure:
6ea3b446	570	nla_nest_cancel(skb, nest);
1da177e4 LT	571	return -1;
	572	}
	573
2eb9d75c	574	static struct tcf_proto_ops cls_tcindex_ops __read_mostly = {
1da177e4 LT	575	.kind = "tcindex",
	576	.classify = tcindex_classify,
	577	.init = tcindex_init,
	578	.destroy = tcindex_destroy,
	579	.get = tcindex_get,
1da177e4 LT	580	.change = tcindex_change,
	581	.delete = tcindex_delete,
	582	.walk = tcindex_walk,
	583	.dump = tcindex_dump,
	584	.owner = THIS_MODULE,
	585	};
	586
	587	static int __init init_tcindex(void)
	588	{
	589	return register_tcf_proto_ops(&cls_tcindex_ops);
	590	}
	591
10297b99	592	static void __exit exit_tcindex(void)
1da177e4 LT	593	{
	594	unregister_tcf_proto_ops(&cls_tcindex_ops);
	595	}
	596
	597	module_init(init_tcindex)
	598	module_exit(exit_tcindex)
	599	MODULE_LICENSE("GPL");