[deliverable/linux.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 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 void tcindex_put(struct tcf_proto *tp, unsigned long f)
{
	pr_debug("tcindex_put(tp %p,f 0x%lx)\n", tp, f);
}


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 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);
	tcf_exts_destroy(&r->exts);
	if (f)
		kfree_rcu(f, rcu);
	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 void tcindex_destroy(struct tcf_proto *tp)
{
	struct tcindex_data *p = rtnl_dereference(tp->root);
	struct tcf_walker walker;

	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);
}


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;
	unsigned char *b = skb_tail_pointer(skb);
	struct nlattr *nest;

	pr_debug("tcindex_dump(tp %p,fh 0x%lx,skb %p,t %p),p %p,r %p,b %p\n",
		 tp, fh, skb, t, p, r, b);
	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:
	nlmsg_trim(skb, b);
	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,
	.put		=	tcindex_put,
	.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;
	30	};
	31
	32	struct tcindex_filter {
	33	u16 key;
	34	struct tcindex_filter_result result;
331b7292 JF	35	struct tcindex_filter __rcu *next;
331b7292 JF	36	struct rcu_head rcu;
1da177e4 LT	37	};
	38
	39
	40	struct tcindex_data {
	41	struct tcindex_filter_result perfect; / perfect hash; NULL if none */
331b7292 JF	42	struct tcindex_filter __rcu *h; / imperfect hash; */
331b7292 JF	43	struct tcf_proto *tp;
1da177e4	44	u16 mask; /* AND key with mask */
331b7292 JF	45	u32 shift; /* shift ANDed key to the right */
	46	u32 hash; /* hash table size; 0 if undefined */
	47	u32 alloc_hash; /* allocated size */
	48	u32 fall_through; /* 0: only classify if explicit match */
	49	struct rcu_head rcu;
1da177e4 LT	50	};
1da177e4 LT	51
1da177e4 LT	52	static inline int
	53	tcindex_filter_is_set(struct tcindex_filter_result *r)
	54	{
	55	return tcf_exts_is_predicative(&r->exts) \|\| r->res.classid;
	56	}
	57
	58	static struct tcindex_filter_result *
	59	tcindex_lookup(struct tcindex_data *p, u16 key)
	60	{
331b7292 JF	61	if (p->perfect) {
	62	struct tcindex_filter_result *f = p->perfect + key;
	63
	64	return tcindex_filter_is_set(f) ? f : NULL;
	65	} else if (p->h) {
	66	struct tcindex_filter __rcu **fp;
	67	struct tcindex_filter *f;
1da177e4	68
331b7292 JF	69	fp = &p->h[key % p->hash];
	70	for (f = rcu_dereference_bh_rtnl(*fp);
	71	f;
	72	fp = &f->next, f = rcu_dereference_bh_rtnl(*fp))
1da177e4 LT	73	if (f->key == key)
	74	return &f->result;
	75	}
	76
	77	return NULL;
	78	}
	79
	80
dc7f9f6e	81	static int tcindex_classify(struct sk_buff skb, const struct tcf_proto tp,
1da177e4 LT	82	struct tcf_result *res)
1da177e4 LT	83	{
2f9a220e	84	struct tcindex_data *p = rcu_dereference_bh(tp->root);
1da177e4 LT	85	struct tcindex_filter_result *f;
	86	int key = (skb->tc_index & p->mask) >> p->shift;
	87
aa767bfe SH	88	pr_debug("tcindex_classify(skb %p,tp %p,res %p),p %p\n",
aa767bfe SH	89	skb, tp, res, p);
1da177e4 LT	90
	91	f = tcindex_lookup(p, key);
	92	if (!f) {
	93	if (!p->fall_through)
	94	return -1;
	95	res->classid = TC_H_MAKE(TC_H_MAJ(tp->q->handle), key);
	96	res->class = 0;
aa767bfe	97	pr_debug("alg 0x%x\n", res->classid);
1da177e4 LT	98	return 0;
	99	}
	100	*res = f->res;
aa767bfe	101	pr_debug("map 0x%x\n", res->classid);
1da177e4 LT	102
	103	return tcf_exts_exec(skb, &f->exts, res);
	104	}
	105
	106
	107	static unsigned long tcindex_get(struct tcf_proto *tp, u32 handle)
	108	{
331b7292	109	struct tcindex_data *p = rtnl_dereference(tp->root);
1da177e4 LT	110	struct tcindex_filter_result *r;
1da177e4 LT	111
aa767bfe	112	pr_debug("tcindex_get(tp %p,handle 0x%08x)\n", tp, handle);
1da177e4 LT	113	if (p->perfect && handle >= p->alloc_hash)
	114	return 0;
	115	r = tcindex_lookup(p, handle);
	116	return r && tcindex_filter_is_set(r) ? (unsigned long) r : 0UL;
	117	}
	118
	119
	120	static void tcindex_put(struct tcf_proto *tp, unsigned long f)
	121	{
aa767bfe	122	pr_debug("tcindex_put(tp %p,f 0x%lx)\n", tp, f);
1da177e4 LT	123	}
	124
	125
	126	static int tcindex_init(struct tcf_proto *tp)
	127	{
	128	struct tcindex_data *p;
	129
aa767bfe SH	130	pr_debug("tcindex_init(tp %p)\n", tp);
aa767bfe SH	131	p = kzalloc(sizeof(struct tcindex_data), GFP_KERNEL);
1da177e4 LT	132	if (!p)
	133	return -ENOMEM;
	134
1da177e4 LT	135	p->mask = 0xffff;
	136	p->hash = DEFAULT_HASH_SIZE;
	137	p->fall_through = 1;
	138
331b7292	139	rcu_assign_pointer(tp->root, p);
1da177e4 LT	140	return 0;
	141	}
	142
1da177e4	143	static int
331b7292	144	tcindex_delete(struct tcf_proto *tp, unsigned long arg)
1da177e4	145	{
331b7292	146	struct tcindex_data *p = rtnl_dereference(tp->root);
1da177e4	147	struct tcindex_filter_result r = (struct tcindex_filter_result ) arg;
331b7292	148	struct tcindex_filter __rcu **walk;
1da177e4 LT	149	struct tcindex_filter *f = NULL;
1da177e4 LT	150
331b7292	151	pr_debug("tcindex_delete(tp %p,arg 0x%lx),p %p\n", tp, arg, p);
1da177e4 LT	152	if (p->perfect) {
	153	if (!r->res.class)
	154	return -ENOENT;
	155	} else {
	156	int i;
1da177e4	157
331b7292 JF	158	for (i = 0; i < p->hash; i++) {
	159	walk = p->h + i;
	160	for (f = rtnl_dereference(*walk); f;
	161	walk = &f->next, f = rtnl_dereference(*walk)) {
	162	if (&f->result == r)
1da177e4	163	goto found;
331b7292 JF	164	}
331b7292 JF	165	}
1da177e4 LT	166	return -ENOENT;
	167
	168	found:
331b7292	169	rcu_assign_pointer(*walk, rtnl_dereference(f->next));
1da177e4 LT	170	}
1da177e4 LT	171	tcf_unbind_filter(tp, &r->res);
18d0264f	172	tcf_exts_destroy(&r->exts);
331b7292 JF	173	if (f)
331b7292 JF	174	kfree_rcu(f, rcu);
1da177e4 LT	175	return 0;
	176	}
	177
331b7292 JF	178	static int tcindex_destroy_element(struct tcf_proto *tp,
	179	unsigned long arg,
	180	struct tcf_walker *walker)
1da177e4	181	{
331b7292 JF	182	return tcindex_delete(tp, arg);
	183	}
	184
	185	static void __tcindex_destroy(struct rcu_head *head)
	186	{
	187	struct tcindex_data *p = container_of(head, struct tcindex_data, rcu);
	188
	189	kfree(p->perfect);
	190	kfree(p->h);
	191	kfree(p);
1da177e4 LT	192	}
	193
	194	static inline int
	195	valid_perfect_hash(struct tcindex_data *p)
	196	{
	197	return p->hash > (p->mask >> p->shift);
	198	}
	199
6fa8c014 PM	200	static const struct nla_policy tcindex_policy[TCA_TCINDEX_MAX + 1] = {
	201	[TCA_TCINDEX_HASH] = { .type = NLA_U32 },
	202	[TCA_TCINDEX_MASK] = { .type = NLA_U16 },
	203	[TCA_TCINDEX_SHIFT] = { .type = NLA_U32 },
	204	[TCA_TCINDEX_FALL_THROUGH] = { .type = NLA_U32 },
	205	[TCA_TCINDEX_CLASSID] = { .type = NLA_U32 },
	206	};
	207
bf63ac73 CW	208	static void tcindex_filter_result_init(struct tcindex_filter_result *r)
	209	{
	210	memset(r, 0, sizeof(*r));
	211	tcf_exts_init(&r->exts, TCA_TCINDEX_ACT, TCA_TCINDEX_POLICE);
	212	}
	213
331b7292 JF	214	static void __tcindex_partial_destroy(struct rcu_head *head)
	215	{
	216	struct tcindex_data *p = container_of(head, struct tcindex_data, rcu);
	217
	218	kfree(p->perfect);
	219	kfree(p);
	220	}
	221
1da177e4	222	static int
c1b52739 BL	223	tcindex_set_parms(struct net net, struct tcf_proto tp, unsigned long base,
	224	u32 handle, struct tcindex_data *p,
	225	struct tcindex_filter_result r, struct nlattr *tb,
2f7ef2f8	226	struct nlattr *est, bool ovr)
1da177e4 LT	227	{
	228	int err, balloc = 0;
	229	struct tcindex_filter_result new_filter_result, *old_r = r;
	230	struct tcindex_filter_result cr;
331b7292	231	struct tcindex_data cp, oldp;
1da177e4 LT	232	struct tcindex_filter f = NULL; / make gcc behave */
	233	struct tcf_exts e;
	234
5da57f42	235	tcf_exts_init(&e, TCA_TCINDEX_ACT, TCA_TCINDEX_POLICE);
2f7ef2f8	236	err = tcf_exts_validate(net, tp, tb, est, &e, ovr);
1da177e4 LT	237	if (err < 0)
1da177e4 LT	238	return err;
10297b99	239
02c5e844	240	err = -ENOMEM;
331b7292 JF	241	/* tcindex_data attributes must look atomic to classifier/lookup so
	242	* allocate new tcindex data and RCU assign it onto root. Keeping
	243	* perfect hash and hash pointers from old data.
	244	*/
a57a65ba	245	cp = kzalloc(sizeof(*cp), GFP_KERNEL);
02c5e844	246	if (!cp)
44b75e43	247	goto errout;
331b7292 JF	248
	249	cp->mask = p->mask;
	250	cp->shift = p->shift;
	251	cp->hash = p->hash;
	252	cp->alloc_hash = p->alloc_hash;
	253	cp->fall_through = p->fall_through;
	254	cp->tp = tp;
	255
	256	if (p->perfect) {
6e056569 WC	257	int i;
6e056569 WC	258
331b7292 JF	259	cp->perfect = kmemdup(p->perfect,
	260	sizeof(r) cp->hash, GFP_KERNEL);
	261	if (!cp->perfect)
	262	goto errout;
6e056569 WC	263	for (i = 0; i < cp->hash; i++)
	264	tcf_exts_init(&cp->perfect[i].exts,
	265	TCA_TCINDEX_ACT, TCA_TCINDEX_POLICE);
44b75e43	266	balloc = 1;
331b7292 JF	267	}
	268	cp->h = p->h;
	269
10ee1c34	270	tcindex_filter_result_init(&new_filter_result);
bf63ac73	271	tcindex_filter_result_init(&cr);
1da177e4	272	if (old_r)
bf63ac73	273	cr.res = r->res;
1da177e4	274
6fa8c014	275	if (tb[TCA_TCINDEX_HASH])
331b7292	276	cp->hash = nla_get_u32(tb[TCA_TCINDEX_HASH]);
1da177e4	277
6fa8c014	278	if (tb[TCA_TCINDEX_MASK])
331b7292	279	cp->mask = nla_get_u16(tb[TCA_TCINDEX_MASK]);
1da177e4	280
6fa8c014	281	if (tb[TCA_TCINDEX_SHIFT])
331b7292	282	cp->shift = nla_get_u32(tb[TCA_TCINDEX_SHIFT]);
1da177e4 LT	283
1da177e4 LT	284	err = -EBUSY;
331b7292	285
1da177e4 LT	286	/* Hash already allocated, make sure that we still meet the
	287	* requirements for the allocated hash.
	288	*/
331b7292 JF	289	if (cp->perfect) {
	290	if (!valid_perfect_hash(cp) \|\|
	291	cp->hash > cp->alloc_hash)
44b75e43	292	goto errout_alloc;
331b7292	293	} else if (cp->h && cp->hash != cp->alloc_hash) {
44b75e43	294	goto errout_alloc;
331b7292	295	}
1da177e4 LT	296
1da177e4 LT	297	err = -EINVAL;
6fa8c014	298	if (tb[TCA_TCINDEX_FALL_THROUGH])
331b7292	299	cp->fall_through = nla_get_u32(tb[TCA_TCINDEX_FALL_THROUGH]);
1da177e4	300
331b7292	301	if (!cp->hash) {
1da177e4 LT	302	/* Hash not specified, use perfect hash if the upper limit
	303	* of the hashing index is below the threshold.
	304	*/
331b7292 JF	305	if ((cp->mask >> cp->shift) < PERFECT_HASH_THRESHOLD)
331b7292 JF	306	cp->hash = (cp->mask >> cp->shift) + 1;
1da177e4	307	else
331b7292	308	cp->hash = DEFAULT_HASH_SIZE;
1da177e4 LT	309	}
1da177e4 LT	310
68f6a7c6	311	if (!cp->perfect && !cp->h)
331b7292	312	cp->alloc_hash = cp->hash;
1da177e4 LT	313
	314	/* Note: this could be as restrictive as if (handle & ~(mask >> shift))
	315	* but then, we'd fail handles that may become valid after some future
	316	* mask change. While this is extremely unlikely to ever matter,
	317	* the check below is safer (and also more backwards-compatible).
	318	*/
331b7292 JF	319	if (cp->perfect \|\| valid_perfect_hash(cp))
331b7292 JF	320	if (handle >= cp->alloc_hash)
44b75e43	321	goto errout_alloc;
1da177e4 LT	322
	323
	324	err = -ENOMEM;
331b7292 JF	325	if (!cp->perfect && !cp->h) {
331b7292 JF	326	if (valid_perfect_hash(cp)) {
bf63ac73 CW	327	int i;
bf63ac73 CW	328
331b7292 JF	329	cp->perfect = kcalloc(cp->hash, sizeof(*r), GFP_KERNEL);
331b7292 JF	330	if (!cp->perfect)
44b75e43	331	goto errout_alloc;
331b7292 JF	332	for (i = 0; i < cp->hash; i++)
	333	tcf_exts_init(&cp->perfect[i].exts,
	334	TCA_TCINDEX_ACT,
bf63ac73	335	TCA_TCINDEX_POLICE);
1da177e4 LT	336	balloc = 1;
1da177e4 LT	337	} else {
331b7292 JF	338	struct tcindex_filter __rcu **hash;
	339
	340	hash = kcalloc(cp->hash,
	341	sizeof(struct tcindex_filter *),
	342	GFP_KERNEL);
	343
	344	if (!hash)
44b75e43	345	goto errout_alloc;
331b7292 JF	346
331b7292 JF	347	cp->h = hash;
1da177e4 LT	348	balloc = 2;
	349	}
	350	}
	351
331b7292 JF	352	if (cp->perfect)
331b7292 JF	353	r = cp->perfect + handle;
1da177e4	354	else
331b7292	355	r = tcindex_lookup(cp, handle) ? : &new_filter_result;
1da177e4 LT	356
1da177e4 LT	357	if (r == &new_filter_result) {
0da974f4	358	f = kzalloc(sizeof(*f), GFP_KERNEL);
1da177e4 LT	359	if (!f)
1da177e4 LT	360	goto errout_alloc;
6e056569 WC	361	f->key = handle;
	362	tcindex_filter_result_init(&f->result);
	363	f->next = NULL;
10297b99	364	}
1da177e4	365
add93b61	366	if (tb[TCA_TCINDEX_CLASSID]) {
1587bac4	367	cr.res.classid = nla_get_u32(tb[TCA_TCINDEX_CLASSID]);
1da177e4	368	tcf_bind_filter(tp, &cr.res, base);
10297b99	369	}
1da177e4	370
bf63ac73 CW	371	if (old_r)
	372	tcf_exts_change(tp, &r->exts, &e);
	373	else
	374	tcf_exts_change(tp, &cr.exts, &e);
1da177e4	375
1da177e4	376	if (old_r && old_r != r)
bf63ac73	377	tcindex_filter_result_init(old_r);
1da177e4	378
331b7292	379	oldp = p;
bf63ac73	380	r->res = cr.res;
331b7292	381	rcu_assign_pointer(tp->root, cp);
1da177e4 LT	382
1da177e4 LT	383	if (r == &new_filter_result) {
331b7292 JF	384	struct tcindex_filter *nfp;
331b7292 JF	385	struct tcindex_filter __rcu **fp;
1da177e4	386
6e056569	387	tcf_exts_change(tp, &f->result.exts, &r->exts);
331b7292	388
69301eaa	389	fp = cp->h + (handle % cp->hash);
331b7292 JF	390	for (nfp = rtnl_dereference(*fp);
	391	nfp;
	392	fp = &nfp->next, nfp = rtnl_dereference(*fp))
	393	; /* nothing */
	394
	395	rcu_assign_pointer(*fp, f);
10297b99	396	}
1da177e4	397
331b7292 JF	398	if (oldp)
331b7292 JF	399	call_rcu(&oldp->rcu, __tcindex_partial_destroy);
1da177e4 LT	400	return 0;
	401
	402	errout_alloc:
	403	if (balloc == 1)
331b7292	404	kfree(cp->perfect);
1da177e4	405	else if (balloc == 2)
331b7292	406	kfree(cp->h);
1da177e4	407	errout:
331b7292	408	kfree(cp);
18d0264f	409	tcf_exts_destroy(&e);
1da177e4 LT	410	return err;
	411	}
	412
	413	static int
c1b52739	414	tcindex_change(struct net net, struct sk_buff in_skb,
af4c6641	415	struct tcf_proto *tp, unsigned long base, u32 handle,
2f7ef2f8	416	struct nlattr *tca, unsigned long arg, bool ovr)
1da177e4	417	{
add93b61 PM	418	struct nlattr *opt = tca[TCA_OPTIONS];
add93b61 PM	419	struct nlattr *tb[TCA_TCINDEX_MAX + 1];
331b7292	420	struct tcindex_data *p = rtnl_dereference(tp->root);
1da177e4	421	struct tcindex_filter_result r = (struct tcindex_filter_result ) *arg;
cee63723	422	int err;
1da177e4	423
aa767bfe	424	pr_debug("tcindex_change(tp %p,handle 0x%08x,tca %p,arg %p),opt %p,"
1da177e4 LT	425	"p %p,r %p,*arg 0x%lx\n",
	426	tp, handle, tca, arg, opt, p, r, arg ? *arg : 0L);
	427
	428	if (!opt)
	429	return 0;
	430
6fa8c014	431	err = nla_parse_nested(tb, TCA_TCINDEX_MAX, opt, tcindex_policy);
cee63723 PM	432	if (err < 0)
cee63723 PM	433	return err;
1da177e4	434
c1b52739	435	return tcindex_set_parms(net, tp, base, handle, p, r, tb,
2f7ef2f8	436	tca[TCA_RATE], ovr);
1da177e4 LT	437	}
1da177e4 LT	438
1da177e4 LT	439	static void tcindex_walk(struct tcf_proto tp, struct tcf_walker walker)
1da177e4 LT	440	{
331b7292	441	struct tcindex_data *p = rtnl_dereference(tp->root);
aa767bfe	442	struct tcindex_filter f, next;
1da177e4 LT	443	int i;
1da177e4 LT	444
aa767bfe	445	pr_debug("tcindex_walk(tp %p,walker %p),p %p\n", tp, walker, p);
1da177e4 LT	446	if (p->perfect) {
	447	for (i = 0; i < p->hash; i++) {
	448	if (!p->perfect[i].res.class)
	449	continue;
	450	if (walker->count >= walker->skip) {
	451	if (walker->fn(tp,
	452	(unsigned long) (p->perfect+i), walker)
	453	< 0) {
	454	walker->stop = 1;
	455	return;
	456	}
	457	}
	458	walker->count++;
	459	}
	460	}
	461	if (!p->h)
	462	return;
	463	for (i = 0; i < p->hash; i++) {
331b7292 JF	464	for (f = rtnl_dereference(p->h[i]); f; f = next) {
331b7292 JF	465	next = rtnl_dereference(f->next);
1da177e4	466	if (walker->count >= walker->skip) {
aa767bfe	467	if (walker->fn(tp, (unsigned long) &f->result,
1da177e4 LT	468	walker) < 0) {
	469	walker->stop = 1;
	470	return;
	471	}
	472	}
	473	walker->count++;
	474	}
	475	}
	476	}
	477
1da177e4 LT	478	static void tcindex_destroy(struct tcf_proto *tp)
1da177e4 LT	479	{
331b7292	480	struct tcindex_data *p = rtnl_dereference(tp->root);
1da177e4 LT	481	struct tcf_walker walker;
1da177e4 LT	482
aa767bfe	483	pr_debug("tcindex_destroy(tp %p),p %p\n", tp, p);
1da177e4 LT	484	walker.count = 0;
1da177e4 LT	485	walker.skip = 0;
e40f5c72	486	walker.fn = tcindex_destroy_element;
aa767bfe	487	tcindex_walk(tp, &walker);
331b7292 JF	488
	489	RCU_INIT_POINTER(tp->root, NULL);
	490	call_rcu(&p->rcu, __tcindex_destroy);
1da177e4 LT	491	}
	492
	493
832d1d5b	494	static int tcindex_dump(struct net net, struct tcf_proto tp, unsigned long fh,
1da177e4 LT	495	struct sk_buff skb, struct tcmsg t)
1da177e4 LT	496	{
331b7292	497	struct tcindex_data *p = rtnl_dereference(tp->root);
1da177e4	498	struct tcindex_filter_result r = (struct tcindex_filter_result ) fh;
27a884dc	499	unsigned char *b = skb_tail_pointer(skb);
4b3550ef	500	struct nlattr *nest;
1da177e4	501
aa767bfe SH	502	pr_debug("tcindex_dump(tp %p,fh 0x%lx,skb %p,t %p),p %p,r %p,b %p\n",
	503	tp, fh, skb, t, p, r, b);
	504	pr_debug("p->perfect %p p->h %p\n", p->perfect, p->h);
4b3550ef PM	505
	506	nest = nla_nest_start(skb, TCA_OPTIONS);
	507	if (nest == NULL)
	508	goto nla_put_failure;
	509
1da177e4 LT	510	if (!fh) {
1da177e4 LT	511	t->tcm_handle = ~0; /* whatever ... */
1b34ec43 DM	512	if (nla_put_u32(skb, TCA_TCINDEX_HASH, p->hash) \|\|
	513	nla_put_u16(skb, TCA_TCINDEX_MASK, p->mask) \|\|
	514	nla_put_u32(skb, TCA_TCINDEX_SHIFT, p->shift) \|\|
	515	nla_put_u32(skb, TCA_TCINDEX_FALL_THROUGH, p->fall_through))
	516	goto nla_put_failure;
4b3550ef	517	nla_nest_end(skb, nest);
1da177e4 LT	518	} else {
1da177e4 LT	519	if (p->perfect) {
331b7292	520	t->tcm_handle = r - p->perfect;
1da177e4 LT	521	} else {
1da177e4 LT	522	struct tcindex_filter *f;
331b7292	523	struct tcindex_filter __rcu **fp;
1da177e4 LT	524	int i;
	525
	526	t->tcm_handle = 0;
	527	for (i = 0; !t->tcm_handle && i < p->hash; i++) {
331b7292 JF	528	fp = &p->h[i];
	529	for (f = rtnl_dereference(*fp);
	530	!t->tcm_handle && f;
	531	fp = &f->next, f = rtnl_dereference(*fp)) {
1da177e4 LT	532	if (&f->result == r)
	533	t->tcm_handle = f->key;
	534	}
	535	}
	536	}
aa767bfe	537	pr_debug("handle = %d\n", t->tcm_handle);
1b34ec43 DM	538	if (r->res.class &&
	539	nla_put_u32(skb, TCA_TCINDEX_CLASSID, r->res.classid))
	540	goto nla_put_failure;
1da177e4	541
5da57f42	542	if (tcf_exts_dump(skb, &r->exts) < 0)
add93b61	543	goto nla_put_failure;
4b3550ef	544	nla_nest_end(skb, nest);
1da177e4	545
5da57f42	546	if (tcf_exts_dump_stats(skb, &r->exts) < 0)
add93b61	547	goto nla_put_failure;
1da177e4	548	}
10297b99	549
1da177e4 LT	550	return skb->len;
1da177e4 LT	551
add93b61	552	nla_put_failure:
dc5fc579	553	nlmsg_trim(skb, b);
1da177e4 LT	554	return -1;
	555	}
	556
2eb9d75c	557	static struct tcf_proto_ops cls_tcindex_ops __read_mostly = {
1da177e4 LT	558	.kind = "tcindex",
	559	.classify = tcindex_classify,
	560	.init = tcindex_init,
	561	.destroy = tcindex_destroy,
	562	.get = tcindex_get,
	563	.put = tcindex_put,
	564	.change = tcindex_change,
	565	.delete = tcindex_delete,
	566	.walk = tcindex_walk,
	567	.dump = tcindex_dump,
	568	.owner = THIS_MODULE,
	569	};
	570
	571	static int __init init_tcindex(void)
	572	{
	573	return register_tcf_proto_ops(&cls_tcindex_ops);
	574	}
	575
10297b99	576	static void __exit exit_tcindex(void)
1da177e4 LT	577	{
	578	unregister_tcf_proto_ops(&cls_tcindex_ops);
	579	}
	580
	581	module_init(init_tcindex)
	582	module_exit(exit_tcindex)
	583	MODULE_LICENSE("GPL");