[mirror_ubuntu-artful-kernel.git] / net / sched / sch_dsmark.c

/* net/sched/sch_dsmark.c - Differentiated Services field marker */

/* Written 1998-2000 by Werner Almesberger, EPFL ICA */


#include <linux/module.h>
#include <linux/init.h>
#include <linux/slab.h>
#include <linux/types.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <linux/rtnetlink.h>
#include <linux/bitops.h>
#include <net/pkt_sched.h>
#include <net/pkt_cls.h>
#include <net/dsfield.h>
#include <net/inet_ecn.h>
#include <asm/byteorder.h>

/*
 * classid	class		marking
 * -------	-----		-------
 *   n/a	  0		n/a
 *   x:0	  1		use entry [0]
 *   ...	 ...		...
 *   x:y y>0	 y+1		use entry [y]
 *   ...	 ...		...
 * x:indices-1	indices		use entry [indices-1]
 *   ...	 ...		...
 *   x:y	 y+1		use entry [y & (indices-1)]
 *   ...	 ...		...
 * 0xffff	0x10000		use entry [indices-1]
 */


#define NO_DEFAULT_INDEX	(1 << 16)

struct mask_value {
	u8			mask;
	u8			value;
};

struct dsmark_qdisc_data {
	struct Qdisc		*q;
	struct tcf_proto __rcu	*filter_list;
	struct mask_value	*mv;
	u16			indices;
	u8			set_tc_index;
	u32			default_index;	/* index range is 0...0xffff */
#define DSMARK_EMBEDDED_SZ	16
	struct mask_value	embedded[DSMARK_EMBEDDED_SZ];
};

static inline int dsmark_valid_index(struct dsmark_qdisc_data *p, u16 index)
{
	return index <= p->indices && index > 0;
}

/* ------------------------- Class/flow operations ------------------------- */

static int dsmark_graft(struct Qdisc *sch, unsigned long arg,
			struct Qdisc *new, struct Qdisc **old)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);

	pr_debug("%s(sch %p,[qdisc %p],new %p,old %p)\n",
		 __func__, sch, p, new, old);

	if (new == NULL) {
		new = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops,
					sch->handle);
		if (new == NULL)
			new = &noop_qdisc;
	}

	*old = qdisc_replace(sch, new, &p->q);
	return 0;
}

static struct Qdisc *dsmark_leaf(struct Qdisc *sch, unsigned long arg)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	return p->q;
}

static unsigned long dsmark_get(struct Qdisc *sch, u32 classid)
{
	pr_debug("%s(sch %p,[qdisc %p],classid %x)\n",
		 __func__, sch, qdisc_priv(sch), classid);

	return TC_H_MIN(classid) + 1;
}

static unsigned long dsmark_bind_filter(struct Qdisc *sch,
					unsigned long parent, u32 classid)
{
	return dsmark_get(sch, classid);
}

static void dsmark_put(struct Qdisc *sch, unsigned long cl)
{
}

static const struct nla_policy dsmark_policy[TCA_DSMARK_MAX + 1] = {
	[TCA_DSMARK_INDICES]		= { .type = NLA_U16 },
	[TCA_DSMARK_DEFAULT_INDEX]	= { .type = NLA_U16 },
	[TCA_DSMARK_SET_TC_INDEX]	= { .type = NLA_FLAG },
	[TCA_DSMARK_MASK]		= { .type = NLA_U8 },
	[TCA_DSMARK_VALUE]		= { .type = NLA_U8 },
};

static int dsmark_change(struct Qdisc *sch, u32 classid, u32 parent,
			 struct nlattr **tca, unsigned long *arg)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	struct nlattr *opt = tca[TCA_OPTIONS];
	struct nlattr *tb[TCA_DSMARK_MAX + 1];
	int err = -EINVAL;

	pr_debug("%s(sch %p,[qdisc %p],classid %x,parent %x), arg 0x%lx\n",
		 __func__, sch, p, classid, parent, *arg);

	if (!dsmark_valid_index(p, *arg)) {
		err = -ENOENT;
		goto errout;
	}

	if (!opt)
		goto errout;

	err = nla_parse_nested(tb, TCA_DSMARK_MAX, opt, dsmark_policy, NULL);
	if (err < 0)
		goto errout;

	if (tb[TCA_DSMARK_VALUE])
		p->mv[*arg - 1].value = nla_get_u8(tb[TCA_DSMARK_VALUE]);

	if (tb[TCA_DSMARK_MASK])
		p->mv[*arg - 1].mask = nla_get_u8(tb[TCA_DSMARK_MASK]);

	err = 0;

errout:
	return err;
}

static int dsmark_delete(struct Qdisc *sch, unsigned long arg)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);

	if (!dsmark_valid_index(p, arg))
		return -EINVAL;

	p->mv[arg - 1].mask = 0xff;
	p->mv[arg - 1].value = 0;

	return 0;
}

static void dsmark_walk(struct Qdisc *sch, struct qdisc_walker *walker)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	int i;

	pr_debug("%s(sch %p,[qdisc %p],walker %p)\n",
		 __func__, sch, p, walker);

	if (walker->stop)
		return;

	for (i = 0; i < p->indices; i++) {
		if (p->mv[i].mask == 0xff && !p->mv[i].value)
			goto ignore;
		if (walker->count >= walker->skip) {
			if (walker->fn(sch, i + 1, walker) < 0) {
				walker->stop = 1;
				break;
			}
		}
ignore:
		walker->count++;
	}
}

static inline struct tcf_proto __rcu **dsmark_find_tcf(struct Qdisc *sch,
						       unsigned long cl)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	return &p->filter_list;
}

/* --------------------------- Qdisc operations ---------------------------- */

static int dsmark_enqueue(struct sk_buff *skb, struct Qdisc *sch,
			  struct sk_buff **to_free)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	int err;

	pr_debug("%s(skb %p,sch %p,[qdisc %p])\n", __func__, skb, sch, p);

	if (p->set_tc_index) {
		int wlen = skb_network_offset(skb);

		switch (tc_skb_protocol(skb)) {
		case htons(ETH_P_IP):
			wlen += sizeof(struct iphdr);
			if (!pskb_may_pull(skb, wlen) ||
			    skb_try_make_writable(skb, wlen))
				goto drop;

			skb->tc_index = ipv4_get_dsfield(ip_hdr(skb))
				& ~INET_ECN_MASK;
			break;

		case htons(ETH_P_IPV6):
			wlen += sizeof(struct ipv6hdr);
			if (!pskb_may_pull(skb, wlen) ||
			    skb_try_make_writable(skb, wlen))
				goto drop;

			skb->tc_index = ipv6_get_dsfield(ipv6_hdr(skb))
				& ~INET_ECN_MASK;
			break;
		default:
			skb->tc_index = 0;
			break;
		}
	}

	if (TC_H_MAJ(skb->priority) == sch->handle)
		skb->tc_index = TC_H_MIN(skb->priority);
	else {
		struct tcf_result res;
		struct tcf_proto *fl = rcu_dereference_bh(p->filter_list);
		int result = tc_classify(skb, fl, &res, false);

		pr_debug("result %d class 0x%04x\n", result, res.classid);

		switch (result) {
#ifdef CONFIG_NET_CLS_ACT
		case TC_ACT_QUEUED:
		case TC_ACT_STOLEN:
			__qdisc_drop(skb, to_free);
			return NET_XMIT_SUCCESS | __NET_XMIT_STOLEN;

		case TC_ACT_SHOT:
			goto drop;
#endif
		case TC_ACT_OK:
			skb->tc_index = TC_H_MIN(res.classid);
			break;

		default:
			if (p->default_index != NO_DEFAULT_INDEX)
				skb->tc_index = p->default_index;
			break;
		}
	}

	err = qdisc_enqueue(skb, p->q, to_free);
	if (err != NET_XMIT_SUCCESS) {
		if (net_xmit_drop_count(err))
			qdisc_qstats_drop(sch);
		return err;
	}

	qdisc_qstats_backlog_inc(sch, skb);
	sch->q.qlen++;

	return NET_XMIT_SUCCESS;

drop:
	qdisc_drop(skb, sch, to_free);
	return NET_XMIT_SUCCESS | __NET_XMIT_BYPASS;
}

static struct sk_buff *dsmark_dequeue(struct Qdisc *sch)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	struct sk_buff *skb;
	u32 index;

	pr_debug("%s(sch %p,[qdisc %p])\n", __func__, sch, p);

	skb = qdisc_dequeue_peeked(p->q);
	if (skb == NULL)
		return NULL;

	qdisc_bstats_update(sch, skb);
	qdisc_qstats_backlog_dec(sch, skb);
	sch->q.qlen--;

	index = skb->tc_index & (p->indices - 1);
	pr_debug("index %d->%d\n", skb->tc_index, index);

	switch (tc_skb_protocol(skb)) {
	case htons(ETH_P_IP):
		ipv4_change_dsfield(ip_hdr(skb), p->mv[index].mask,
				    p->mv[index].value);
			break;
	case htons(ETH_P_IPV6):
		ipv6_change_dsfield(ipv6_hdr(skb), p->mv[index].mask,
				    p->mv[index].value);
			break;
	default:
		/*
		 * Only complain if a change was actually attempted.
		 * This way, we can send non-IP traffic through dsmark
		 * and don't need yet another qdisc as a bypass.
		 */
		if (p->mv[index].mask != 0xff || p->mv[index].value)
			pr_warn("%s: unsupported protocol %d\n",
				__func__, ntohs(tc_skb_protocol(skb)));
		break;
	}

	return skb;
}

static struct sk_buff *dsmark_peek(struct Qdisc *sch)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);

	pr_debug("%s(sch %p,[qdisc %p])\n", __func__, sch, p);

	return p->q->ops->peek(p->q);
}

static int dsmark_init(struct Qdisc *sch, struct nlattr *opt)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	struct nlattr *tb[TCA_DSMARK_MAX + 1];
	int err = -EINVAL;
	u32 default_index = NO_DEFAULT_INDEX;
	u16 indices;
	int i;

	pr_debug("%s(sch %p,[qdisc %p],opt %p)\n", __func__, sch, p, opt);

	if (!opt)
		goto errout;

	err = nla_parse_nested(tb, TCA_DSMARK_MAX, opt, dsmark_policy, NULL);
	if (err < 0)
		goto errout;

	err = -EINVAL;
	indices = nla_get_u16(tb[TCA_DSMARK_INDICES]);

	if (hweight32(indices) != 1)
		goto errout;

	if (tb[TCA_DSMARK_DEFAULT_INDEX])
		default_index = nla_get_u16(tb[TCA_DSMARK_DEFAULT_INDEX]);

	if (indices <= DSMARK_EMBEDDED_SZ)
		p->mv = p->embedded;
	else
		p->mv = kmalloc_array(indices, sizeof(*p->mv), GFP_KERNEL);
	if (!p->mv) {
		err = -ENOMEM;
		goto errout;
	}
	for (i = 0; i < indices; i++) {
		p->mv[i].mask = 0xff;
		p->mv[i].value = 0;
	}
	p->indices = indices;
	p->default_index = default_index;
	p->set_tc_index = nla_get_flag(tb[TCA_DSMARK_SET_TC_INDEX]);

	p->q = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops, sch->handle);
	if (p->q == NULL)
		p->q = &noop_qdisc;
	else
		qdisc_hash_add(p->q, true);

	pr_debug("%s: qdisc %p\n", __func__, p->q);

	err = 0;
errout:
	return err;
}

static void dsmark_reset(struct Qdisc *sch)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);

	pr_debug("%s(sch %p,[qdisc %p])\n", __func__, sch, p);
	qdisc_reset(p->q);
	sch->qstats.backlog = 0;
	sch->q.qlen = 0;
}

static void dsmark_destroy(struct Qdisc *sch)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);

	pr_debug("%s(sch %p,[qdisc %p])\n", __func__, sch, p);

	tcf_destroy_chain(&p->filter_list);
	qdisc_destroy(p->q);
	if (p->mv != p->embedded)
		kfree(p->mv);
}

static int dsmark_dump_class(struct Qdisc *sch, unsigned long cl,
			     struct sk_buff *skb, struct tcmsg *tcm)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	struct nlattr *opts = NULL;

	pr_debug("%s(sch %p,[qdisc %p],class %ld\n", __func__, sch, p, cl);

	if (!dsmark_valid_index(p, cl))
		return -EINVAL;

	tcm->tcm_handle = TC_H_MAKE(TC_H_MAJ(sch->handle), cl - 1);
	tcm->tcm_info = p->q->handle;

	opts = nla_nest_start(skb, TCA_OPTIONS);
	if (opts == NULL)
		goto nla_put_failure;
	if (nla_put_u8(skb, TCA_DSMARK_MASK, p->mv[cl - 1].mask) ||
	    nla_put_u8(skb, TCA_DSMARK_VALUE, p->mv[cl - 1].value))
		goto nla_put_failure;

	return nla_nest_end(skb, opts);

nla_put_failure:
	nla_nest_cancel(skb, opts);
	return -EMSGSIZE;
}

static int dsmark_dump(struct Qdisc *sch, struct sk_buff *skb)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	struct nlattr *opts = NULL;

	opts = nla_nest_start(skb, TCA_OPTIONS);
	if (opts == NULL)
		goto nla_put_failure;
	if (nla_put_u16(skb, TCA_DSMARK_INDICES, p->indices))
		goto nla_put_failure;

	if (p->default_index != NO_DEFAULT_INDEX &&
	    nla_put_u16(skb, TCA_DSMARK_DEFAULT_INDEX, p->default_index))
		goto nla_put_failure;

	if (p->set_tc_index &&
	    nla_put_flag(skb, TCA_DSMARK_SET_TC_INDEX))
		goto nla_put_failure;

	return nla_nest_end(skb, opts);

nla_put_failure:
	nla_nest_cancel(skb, opts);
	return -EMSGSIZE;
}

static const struct Qdisc_class_ops dsmark_class_ops = {
	.graft		=	dsmark_graft,
	.leaf		=	dsmark_leaf,
	.get		=	dsmark_get,
	.put		=	dsmark_put,
	.change		=	dsmark_change,
	.delete		=	dsmark_delete,
	.walk		=	dsmark_walk,
	.tcf_chain	=	dsmark_find_tcf,
	.bind_tcf	=	dsmark_bind_filter,
	.unbind_tcf	=	dsmark_put,
	.dump		=	dsmark_dump_class,
};

static struct Qdisc_ops dsmark_qdisc_ops __read_mostly = {
	.next		=	NULL,
	.cl_ops		=	&dsmark_class_ops,
	.id		=	"dsmark",
	.priv_size	=	sizeof(struct dsmark_qdisc_data),
	.enqueue	=	dsmark_enqueue,
	.dequeue	=	dsmark_dequeue,
	.peek		=	dsmark_peek,
	.init		=	dsmark_init,
	.reset		=	dsmark_reset,
	.destroy	=	dsmark_destroy,
	.change		=	NULL,
	.dump		=	dsmark_dump,
	.owner		=	THIS_MODULE,
};

static int __init dsmark_module_init(void)
{
	return register_qdisc(&dsmark_qdisc_ops);
}

static void __exit dsmark_module_exit(void)
{
	unregister_qdisc(&dsmark_qdisc_ops);
}

module_init(dsmark_module_init)
module_exit(dsmark_module_exit)

MODULE_LICENSE("GPL");
Commit	Line	Data
1da177e4 LT	1	/* net/sched/sch_dsmark.c - Differentiated Services field marker */
	2
	3	/* Written 1998-2000 by Werner Almesberger, EPFL ICA */
	4
	5
1da177e4 LT	6	#include <linux/module.h>
1da177e4 LT	7	#include <linux/init.h>
5a0e3ad6	8	#include <linux/slab.h>
1da177e4 LT	9	#include <linux/types.h>
	10	#include <linux/string.h>
	11	#include <linux/errno.h>
	12	#include <linux/skbuff.h>
1da177e4	13	#include <linux/rtnetlink.h>
5b0ac72b	14	#include <linux/bitops.h>
1da177e4	15	#include <net/pkt_sched.h>
cf1facda	16	#include <net/pkt_cls.h>
1da177e4 LT	17	#include <net/dsfield.h>
	18	#include <net/inet_ecn.h>
	19	#include <asm/byteorder.h>
	20
1da177e4 LT	21	/*
	22	* classid class marking
	23	* ------- ----- -------
	24	* n/a 0 n/a
	25	* x:0 1 use entry [0]
	26	* ... ... ...
	27	* x:y y>0 y+1 use entry [y]
	28	* ... ... ...
	29	* x:indices-1 indices use entry [indices-1]
	30	* ... ... ...
	31	* x:y y+1 use entry [y & (indices-1)]
	32	* ... ... ...
	33	* 0xffff 0x10000 use entry [indices-1]
	34	*/
	35
	36
	37	#define NO_DEFAULT_INDEX (1 << 16)
	38
47bbbb30 ED	39	struct mask_value {
	40	u8 mask;
	41	u8 value;
	42	};
	43
1da177e4 LT	44	struct dsmark_qdisc_data {
1da177e4 LT	45	struct Qdisc *q;
25d8c0d5	46	struct tcf_proto __rcu *filter_list;
47bbbb30	47	struct mask_value *mv;
af0d1141	48	u16 indices;
47bbbb30	49	u8 set_tc_index;
af0d1141	50	u32 default_index; /* index range is 0...0xffff */
47bbbb30 ED	51	#define DSMARK_EMBEDDED_SZ 16
47bbbb30 ED	52	struct mask_value embedded[DSMARK_EMBEDDED_SZ];
1da177e4 LT	53	};
1da177e4 LT	54
758cc43c TG	55	static inline int dsmark_valid_index(struct dsmark_qdisc_data *p, u16 index)
758cc43c TG	56	{
17569fae	57	return index <= p->indices && index > 0;
758cc43c	58	}
1da177e4 LT	59
	60	/* ------------------------- Class/flow operations ------------------------- */
	61
af0d1141 TG	62	static int dsmark_graft(struct Qdisc *sch, unsigned long arg,
af0d1141 TG	63	struct Qdisc new, struct Qdisc *old)
1da177e4	64	{
81da99ed	65	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	66
c76f2a2c YY	67	pr_debug("%s(sch %p,[qdisc %p],new %p,old %p)\n",
c76f2a2c YY	68	__func__, sch, p, new, old);
486b53e5 TG	69
486b53e5 TG	70	if (new == NULL) {
3511c913	71	new = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops,
9f9afec4	72	sch->handle);
486b53e5 TG	73	if (new == NULL)
	74	new = &noop_qdisc;
	75	}
	76
86a7996c	77	*old = qdisc_replace(sch, new, &p->q);
10297b99	78	return 0;
1da177e4 LT	79	}
1da177e4 LT	80
1da177e4 LT	81	static struct Qdisc dsmark_leaf(struct Qdisc sch, unsigned long arg)
1da177e4 LT	82	{
81da99ed SH	83	struct dsmark_qdisc_data *p = qdisc_priv(sch);
81da99ed SH	84	return p->q;
1da177e4 LT	85	}
1da177e4 LT	86
af0d1141	87	static unsigned long dsmark_get(struct Qdisc *sch, u32 classid)
1da177e4	88	{
c76f2a2c YY	89	pr_debug("%s(sch %p,[qdisc %p],classid %x)\n",
c76f2a2c YY	90	__func__, sch, qdisc_priv(sch), classid);
1da177e4	91
af0d1141	92	return TC_H_MIN(classid) + 1;
1da177e4 LT	93	}
1da177e4 LT	94
1da177e4	95	static unsigned long dsmark_bind_filter(struct Qdisc *sch,
af0d1141	96	unsigned long parent, u32 classid)
1da177e4	97	{
af0d1141	98	return dsmark_get(sch, classid);
1da177e4 LT	99	}
1da177e4 LT	100
1da177e4 LT	101	static void dsmark_put(struct Qdisc *sch, unsigned long cl)
	102	{
	103	}
	104
27a3421e PM	105	static const struct nla_policy dsmark_policy[TCA_DSMARK_MAX + 1] = {
	106	[TCA_DSMARK_INDICES] = { .type = NLA_U16 },
	107	[TCA_DSMARK_DEFAULT_INDEX] = { .type = NLA_U16 },
	108	[TCA_DSMARK_SET_TC_INDEX] = { .type = NLA_FLAG },
	109	[TCA_DSMARK_MASK] = { .type = NLA_U8 },
	110	[TCA_DSMARK_VALUE] = { .type = NLA_U8 },
	111	};
	112
1da177e4	113	static int dsmark_change(struct Qdisc *sch, u32 classid, u32 parent,
1e90474c	114	struct nlattr *tca, unsigned long arg)
1da177e4	115	{
81da99ed	116	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1e90474c PM	117	struct nlattr *opt = tca[TCA_OPTIONS];
1e90474c PM	118	struct nlattr *tb[TCA_DSMARK_MAX + 1];
758cc43c	119	int err = -EINVAL;
1da177e4	120
c76f2a2c YY	121	pr_debug("%s(sch %p,[qdisc %p],classid %x,parent %x), arg 0x%lx\n",
c76f2a2c YY	122	__func__, sch, p, classid, parent, *arg);
758cc43c TG	123
	124	if (!dsmark_valid_index(p, *arg)) {
	125	err = -ENOENT;
1e90474c	126	goto errout;
1da177e4	127	}
1da177e4	128
cee63723	129	if (!opt)
1e90474c	130	goto errout;
758cc43c	131
fceb6435	132	err = nla_parse_nested(tb, TCA_DSMARK_MAX, opt, dsmark_policy, NULL);
cee63723	133	if (err < 0)
27a3421e	134	goto errout;
cee63723	135
27a3421e	136	if (tb[TCA_DSMARK_VALUE])
47bbbb30	137	p->mv[*arg - 1].value = nla_get_u8(tb[TCA_DSMARK_VALUE]);
10297b99	138
1e90474c	139	if (tb[TCA_DSMARK_MASK])
47bbbb30	140	p->mv[*arg - 1].mask = nla_get_u8(tb[TCA_DSMARK_MASK]);
758cc43c TG	141
	142	err = 0;
	143
1e90474c	144	errout:
758cc43c TG	145	return err;
758cc43c TG	146	}
1da177e4	147
af0d1141	148	static int dsmark_delete(struct Qdisc *sch, unsigned long arg)
1da177e4	149	{
81da99ed	150	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	151
af0d1141	152	if (!dsmark_valid_index(p, arg))
1da177e4	153	return -EINVAL;
10297b99	154
47bbbb30 ED	155	p->mv[arg - 1].mask = 0xff;
47bbbb30 ED	156	p->mv[arg - 1].value = 0;
af0d1141	157
1da177e4 LT	158	return 0;
	159	}
	160
9d127fbd	161	static void dsmark_walk(struct Qdisc sch, struct qdisc_walker walker)
1da177e4	162	{
81da99ed	163	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4 LT	164	int i;
1da177e4 LT	165
c76f2a2c YY	166	pr_debug("%s(sch %p,[qdisc %p],walker %p)\n",
c76f2a2c YY	167	__func__, sch, p, walker);
af0d1141	168
1da177e4 LT	169	if (walker->stop)
1da177e4 LT	170	return;
af0d1141	171
1da177e4	172	for (i = 0; i < p->indices; i++) {
47bbbb30	173	if (p->mv[i].mask == 0xff && !p->mv[i].value)
0451eb07	174	goto ignore;
1da177e4	175	if (walker->count >= walker->skip) {
cc7ec456	176	if (walker->fn(sch, i + 1, walker) < 0) {
1da177e4 LT	177	walker->stop = 1;
	178	break;
	179	}
	180	}
10297b99	181	ignore:
0451eb07	182	walker->count++;
10297b99	183	}
1da177e4 LT	184	}
1da177e4 LT	185
25d8c0d5 JF	186	static inline struct tcf_proto __rcu *dsmark_find_tcf(struct Qdisc sch,
25d8c0d5 JF	187	unsigned long cl)
1da177e4	188	{
81da99ed SH	189	struct dsmark_qdisc_data *p = qdisc_priv(sch);
81da99ed SH	190	return &p->filter_list;
1da177e4 LT	191	}
1da177e4 LT	192
1da177e4 LT	193	/* --------------------------- Qdisc operations ---------------------------- */
1da177e4 LT	194
520ac30f ED	195	static int dsmark_enqueue(struct sk_buff skb, struct Qdisc sch,
520ac30f ED	196	struct sk_buff **to_free)
1da177e4	197	{
81da99ed	198	struct dsmark_qdisc_data *p = qdisc_priv(sch);
af0d1141 TG	199	int err;
af0d1141 TG	200
c76f2a2c	201	pr_debug("%s(skb %p,sch %p,[qdisc %p])\n", __func__, skb, sch, p);
1da177e4	202
1da177e4	203	if (p->set_tc_index) {
aea92fb2 ED	204	int wlen = skb_network_offset(skb);
aea92fb2 ED	205
d8b9605d	206	switch (tc_skb_protocol(skb)) {
60678040	207	case htons(ETH_P_IP):
aea92fb2 ED	208	wlen += sizeof(struct iphdr);
	209	if (!pskb_may_pull(skb, wlen) \|\|
	210	skb_try_make_writable(skb, wlen))
9d127fbd	211	goto drop;
4c30719f	212
9d127fbd SH	213	skb->tc_index = ipv4_get_dsfield(ip_hdr(skb))
	214	& ~INET_ECN_MASK;
	215	break;
4c30719f	216
60678040	217	case htons(ETH_P_IPV6):
aea92fb2 ED	218	wlen += sizeof(struct ipv6hdr);
	219	if (!pskb_may_pull(skb, wlen) \|\|
	220	skb_try_make_writable(skb, wlen))
9d127fbd	221	goto drop;
4c30719f	222
9d127fbd SH	223	skb->tc_index = ipv6_get_dsfield(ipv6_hdr(skb))
	224	& ~INET_ECN_MASK;
	225	break;
	226	default:
	227	skb->tc_index = 0;
	228	break;
3ff50b79	229	}
1da177e4	230	}
af0d1141 TG	231
af0d1141 TG	232	if (TC_H_MAJ(skb->priority) == sch->handle)
1da177e4	233	skb->tc_index = TC_H_MIN(skb->priority);
af0d1141 TG	234	else {
af0d1141 TG	235	struct tcf_result res;
25d8c0d5	236	struct tcf_proto *fl = rcu_dereference_bh(p->filter_list);
3b3ae880	237	int result = tc_classify(skb, fl, &res, false);
af0d1141	238
81da99ed	239	pr_debug("result %d class 0x%04x\n", result, res.classid);
af0d1141	240
1da177e4	241	switch (result) {
f6853e2d PM	242	#ifdef CONFIG_NET_CLS_ACT
	243	case TC_ACT_QUEUED:
	244	case TC_ACT_STOLEN:
520ac30f	245	__qdisc_drop(skb, to_free);
378a2f09	246	return NET_XMIT_SUCCESS \| __NET_XMIT_STOLEN;
4c30719f	247
f6853e2d	248	case TC_ACT_SHOT:
4c30719f	249	goto drop;
1da177e4	250	#endif
c3bc7cff	251	case TC_ACT_OK:
f6853e2d PM	252	skb->tc_index = TC_H_MIN(res.classid);
f6853e2d PM	253	break;
4c30719f	254
f6853e2d PM	255	default:
	256	if (p->default_index != NO_DEFAULT_INDEX)
	257	skb->tc_index = p->default_index;
	258	break;
3ff50b79	259	}
1da177e4	260	}
1da177e4	261
520ac30f	262	err = qdisc_enqueue(skb, p->q, to_free);
af0d1141	263	if (err != NET_XMIT_SUCCESS) {
378a2f09	264	if (net_xmit_drop_count(err))
25331d6c	265	qdisc_qstats_drop(sch);
af0d1141	266	return err;
1da177e4	267	}
af0d1141	268
bdf17661	269	qdisc_qstats_backlog_inc(sch, skb);
1da177e4	270	sch->q.qlen++;
1da177e4	271
af0d1141	272	return NET_XMIT_SUCCESS;
4c30719f SH	273
4c30719f SH	274	drop:
520ac30f	275	qdisc_drop(skb, sch, to_free);
c27f339a	276	return NET_XMIT_SUCCESS \| __NET_XMIT_BYPASS;
af0d1141	277	}
1da177e4 LT	278
	279	static struct sk_buff dsmark_dequeue(struct Qdisc sch)
	280	{
81da99ed	281	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	282	struct sk_buff *skb;
af0d1141 TG	283	u32 index;
af0d1141 TG	284
c76f2a2c	285	pr_debug("%s(sch %p,[qdisc %p])\n", __func__, sch, p);
1da177e4	286
f8b33d8e	287	skb = qdisc_dequeue_peeked(p->q);
af0d1141	288	if (skb == NULL)
1da177e4	289	return NULL;
af0d1141	290
9190b3b3	291	qdisc_bstats_update(sch, skb);
bdf17661	292	qdisc_qstats_backlog_dec(sch, skb);
1da177e4	293	sch->q.qlen--;
af0d1141 TG	294
af0d1141 TG	295	index = skb->tc_index & (p->indices - 1);
81da99ed	296	pr_debug("index %d->%d\n", skb->tc_index, index);
af0d1141	297
d8b9605d	298	switch (tc_skb_protocol(skb)) {
60678040	299	case htons(ETH_P_IP):
47bbbb30 ED	300	ipv4_change_dsfield(ip_hdr(skb), p->mv[index].mask,
47bbbb30 ED	301	p->mv[index].value);
1da177e4	302	break;
60678040	303	case htons(ETH_P_IPV6):
47bbbb30 ED	304	ipv6_change_dsfield(ipv6_hdr(skb), p->mv[index].mask,
47bbbb30 ED	305	p->mv[index].value);
1da177e4	306	break;
9d127fbd SH	307	default:
	308	/*
	309	* Only complain if a change was actually attempted.
	310	* This way, we can send non-IP traffic through dsmark
	311	* and don't need yet another qdisc as a bypass.
	312	*/
47bbbb30	313	if (p->mv[index].mask != 0xff \|\| p->mv[index].value)
c76f2a2c	314	pr_warn("%s: unsupported protocol %d\n",
d8b9605d	315	__func__, ntohs(tc_skb_protocol(skb)));
9d127fbd	316	break;
3ff50b79	317	}
af0d1141	318
1da177e4 LT	319	return skb;
	320	}
	321
8e3af978 JP	322	static struct sk_buff dsmark_peek(struct Qdisc sch)
	323	{
	324	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	325
c76f2a2c	326	pr_debug("%s(sch %p,[qdisc %p])\n", __func__, sch, p);
8e3af978 JP	327
	328	return p->q->ops->peek(p->q);
	329	}
	330
1e90474c	331	static int dsmark_init(struct Qdisc sch, struct nlattr opt)
1da177e4	332	{
81da99ed	333	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1e90474c	334	struct nlattr *tb[TCA_DSMARK_MAX + 1];
758cc43c TG	335	int err = -EINVAL;
	336	u32 default_index = NO_DEFAULT_INDEX;
	337	u16 indices;
47bbbb30	338	int i;
758cc43c	339
c76f2a2c	340	pr_debug("%s(sch %p,[qdisc %p],opt %p)\n", __func__, sch, p, opt);
758cc43c	341
cee63723 PM	342	if (!opt)
	343	goto errout;
	344
fceb6435	345	err = nla_parse_nested(tb, TCA_DSMARK_MAX, opt, dsmark_policy, NULL);
cee63723	346	if (err < 0)
758cc43c TG	347	goto errout;
758cc43c TG	348
cee63723	349	err = -EINVAL;
1e90474c	350	indices = nla_get_u16(tb[TCA_DSMARK_INDICES]);
5b0ac72b DM	351
5b0ac72b DM	352	if (hweight32(indices) != 1)
758cc43c TG	353	goto errout;
758cc43c TG	354
27a3421e	355	if (tb[TCA_DSMARK_DEFAULT_INDEX])
1e90474c	356	default_index = nla_get_u16(tb[TCA_DSMARK_DEFAULT_INDEX]);
758cc43c	357
47bbbb30 ED	358	if (indices <= DSMARK_EMBEDDED_SZ)
	359	p->mv = p->embedded;
	360	else
	361	p->mv = kmalloc_array(indices, sizeof(*p->mv), GFP_KERNEL);
	362	if (!p->mv) {
758cc43c TG	363	err = -ENOMEM;
758cc43c TG	364	goto errout;
1da177e4	365	}
47bbbb30 ED	366	for (i = 0; i < indices; i++) {
	367	p->mv[i].mask = 0xff;
	368	p->mv[i].value = 0;
	369	}
758cc43c TG	370	p->indices = indices;
758cc43c TG	371	p->default_index = default_index;
1e90474c	372	p->set_tc_index = nla_get_flag(tb[TCA_DSMARK_SET_TC_INDEX]);
758cc43c	373
3511c913	374	p->q = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops, sch->handle);
758cc43c	375	if (p->q == NULL)
1da177e4	376	p->q = &noop_qdisc;
49b49971 JK	377	else
49b49971 JK	378	qdisc_hash_add(p->q, true);
758cc43c	379
c76f2a2c	380	pr_debug("%s: qdisc %p\n", __func__, p->q);
758cc43c TG	381
	382	err = 0;
	383	errout:
758cc43c	384	return err;
1da177e4 LT	385	}
1da177e4 LT	386
1da177e4 LT	387	static void dsmark_reset(struct Qdisc *sch)
1da177e4 LT	388	{
81da99ed	389	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	390
c76f2a2c	391	pr_debug("%s(sch %p,[qdisc %p])\n", __func__, sch, p);
1da177e4	392	qdisc_reset(p->q);
bdf17661	393	sch->qstats.backlog = 0;
1da177e4 LT	394	sch->q.qlen = 0;
	395	}
	396
1da177e4 LT	397	static void dsmark_destroy(struct Qdisc *sch)
1da177e4 LT	398	{
81da99ed	399	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	400
c76f2a2c	401	pr_debug("%s(sch %p,[qdisc %p])\n", __func__, sch, p);
af0d1141	402
ff31ab56	403	tcf_destroy_chain(&p->filter_list);
1da177e4	404	qdisc_destroy(p->q);
47bbbb30 ED	405	if (p->mv != p->embedded)
47bbbb30 ED	406	kfree(p->mv);
1da177e4 LT	407	}
1da177e4 LT	408
1da177e4	409	static int dsmark_dump_class(struct Qdisc *sch, unsigned long cl,
02f23f09	410	struct sk_buff skb, struct tcmsg tcm)
1da177e4	411	{
81da99ed	412	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1e90474c	413	struct nlattr *opts = NULL;
1da177e4	414
c76f2a2c	415	pr_debug("%s(sch %p,[qdisc %p],class %ld\n", __func__, sch, p, cl);
02f23f09 TG	416
02f23f09 TG	417	if (!dsmark_valid_index(p, cl))
1da177e4	418	return -EINVAL;
02f23f09	419
cc7ec456	420	tcm->tcm_handle = TC_H_MAKE(TC_H_MAJ(sch->handle), cl - 1);
cdc7f8e3	421	tcm->tcm_info = p->q->handle;
02f23f09	422
1e90474c PM	423	opts = nla_nest_start(skb, TCA_OPTIONS);
	424	if (opts == NULL)
	425	goto nla_put_failure;
47bbbb30 ED	426	if (nla_put_u8(skb, TCA_DSMARK_MASK, p->mv[cl - 1].mask) \|\|
47bbbb30 ED	427	nla_put_u8(skb, TCA_DSMARK_VALUE, p->mv[cl - 1].value))
1b34ec43	428	goto nla_put_failure;
02f23f09	429
1e90474c	430	return nla_nest_end(skb, opts);
1da177e4	431
1e90474c	432	nla_put_failure:
bc3ed28c TG	433	nla_nest_cancel(skb, opts);
bc3ed28c TG	434	return -EMSGSIZE;
1da177e4 LT	435	}
	436
	437	static int dsmark_dump(struct Qdisc sch, struct sk_buff skb)
	438	{
81da99ed	439	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1e90474c	440	struct nlattr *opts = NULL;
1da177e4	441
1e90474c PM	442	opts = nla_nest_start(skb, TCA_OPTIONS);
	443	if (opts == NULL)
	444	goto nla_put_failure;
1b34ec43 DM	445	if (nla_put_u16(skb, TCA_DSMARK_INDICES, p->indices))
1b34ec43 DM	446	goto nla_put_failure;
02f23f09	447
1b34ec43 DM	448	if (p->default_index != NO_DEFAULT_INDEX &&
	449	nla_put_u16(skb, TCA_DSMARK_DEFAULT_INDEX, p->default_index))
	450	goto nla_put_failure;
1da177e4	451
1b34ec43 DM	452	if (p->set_tc_index &&
	453	nla_put_flag(skb, TCA_DSMARK_SET_TC_INDEX))
	454	goto nla_put_failure;
02f23f09	455
1e90474c	456	return nla_nest_end(skb, opts);
1da177e4	457
1e90474c	458	nla_put_failure:
bc3ed28c TG	459	nla_nest_cancel(skb, opts);
bc3ed28c TG	460	return -EMSGSIZE;
1da177e4 LT	461	}
1da177e4 LT	462
20fea08b	463	static const struct Qdisc_class_ops dsmark_class_ops = {
1da177e4 LT	464	.graft = dsmark_graft,
	465	.leaf = dsmark_leaf,
	466	.get = dsmark_get,
	467	.put = dsmark_put,
	468	.change = dsmark_change,
	469	.delete = dsmark_delete,
	470	.walk = dsmark_walk,
	471	.tcf_chain = dsmark_find_tcf,
	472	.bind_tcf = dsmark_bind_filter,
	473	.unbind_tcf = dsmark_put,
	474	.dump = dsmark_dump_class,
	475	};
	476
20fea08b	477	static struct Qdisc_ops dsmark_qdisc_ops __read_mostly = {
1da177e4 LT	478	.next = NULL,
	479	.cl_ops = &dsmark_class_ops,
	480	.id = "dsmark",
	481	.priv_size = sizeof(struct dsmark_qdisc_data),
	482	.enqueue = dsmark_enqueue,
	483	.dequeue = dsmark_dequeue,
8e3af978	484	.peek = dsmark_peek,
1da177e4 LT	485	.init = dsmark_init,
	486	.reset = dsmark_reset,
	487	.destroy = dsmark_destroy,
	488	.change = NULL,
	489	.dump = dsmark_dump,
	490	.owner = THIS_MODULE,
	491	};
	492
	493	static int __init dsmark_module_init(void)
	494	{
	495	return register_qdisc(&dsmark_qdisc_ops);
	496	}
af0d1141	497
10297b99	498	static void __exit dsmark_module_exit(void)
1da177e4 LT	499	{
	500	unregister_qdisc(&dsmark_qdisc_ops);
	501	}
af0d1141	502
1da177e4 LT	503	module_init(dsmark_module_init)
1da177e4 LT	504	module_exit(dsmark_module_exit)
af0d1141	505
1da177e4	506	MODULE_LICENSE("GPL");