[net-next-2.6.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/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/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 dsmark_qdisc_data {
	struct Qdisc		*q;
	struct tcf_proto	*filter_list;
	u8			*mask;	/* "owns" the array */
	u8			*value;
	u16			indices;
	u32			default_index;	/* index range is 0...0xffff */
	int			set_tc_index;
};

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("dsmark_graft(sch %p,[qdisc %p],new %p,old %p)\n",
		sch, p, new, old);

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

	sch_tree_lock(sch);
	*old = xchg(&p->q, new);
	qdisc_tree_decrease_qlen(*old, (*old)->q.qlen);
	qdisc_reset(*old);
	sch_tree_unlock(sch);

	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("dsmark_get(sch %p,[qdisc %p],classid %x)\n",
		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;
	u8 mask = 0;

	pr_debug("dsmark_change(sch %p,[qdisc %p],classid %x,parent %x),"
		"arg 0x%lx\n", 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);
	if (err < 0)
		goto errout;

	if (tb[TCA_DSMARK_MASK])
		mask = nla_get_u8(tb[TCA_DSMARK_MASK]);

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

	if (tb[TCA_DSMARK_MASK])
		p->mask[*arg-1] = 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->mask[arg-1] = 0xff;
	p->value[arg-1] = 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("dsmark_walk(sch %p,[qdisc %p],walker %p)\n", sch, p, walker);

	if (walker->stop)
		return;

	for (i = 0; i < p->indices; i++) {
		if (p->mask[i] == 0xff && !p->value[i])
			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 **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 dsmark_qdisc_data *p = qdisc_priv(sch);
	int err;

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

	if (p->set_tc_index) {
		switch (skb->protocol) {
		case __constant_htons(ETH_P_IP):
			if (skb_cow_head(skb, sizeof(struct iphdr)))
				goto drop;

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

		case __constant_htons(ETH_P_IPV6):
			if (skb_cow_head(skb, sizeof(struct ipv6hdr)))
				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;
		int result = tc_classify(skb, p->filter_list, &res);

		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:
			kfree_skb(skb);
			return NET_XMIT_SUCCESS;

		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 = p->q->enqueue(skb, p->q);
	if (err != NET_XMIT_SUCCESS) {
		sch->qstats.drops++;
		return err;
	}

	sch->bstats.bytes += skb->len;
	sch->bstats.packets++;
	sch->q.qlen++;

	return NET_XMIT_SUCCESS;

drop:
	kfree_skb(skb);
	sch->qstats.drops++;
	return 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("dsmark_dequeue(sch %p,[qdisc %p])\n", sch, p);

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

	sch->q.qlen--;

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

	switch (skb->protocol) {
	case __constant_htons(ETH_P_IP):
		ipv4_change_dsfield(ip_hdr(skb), p->mask[index],
				    p->value[index]);
			break;
	case __constant_htons(ETH_P_IPV6):
		ipv6_change_dsfield(ipv6_hdr(skb), p->mask[index],
				    p->value[index]);
			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->mask[index] != 0xff || p->value[index])
			printk(KERN_WARNING
			       "dsmark_dequeue: unsupported protocol %d\n",
			       ntohs(skb->protocol));
		break;
	}

	return skb;
}

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

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

	err = p->q->ops->requeue(skb, p->q);
	if (err != NET_XMIT_SUCCESS) {
		sch->qstats.drops++;
		return err;
	}

	sch->q.qlen++;
	sch->qstats.requeues++;

	return NET_XMIT_SUCCESS;
}

static unsigned int dsmark_drop(struct Qdisc *sch)
{
	struct dsmark_qdisc_data *p = qdisc_priv(sch);
	unsigned int len;

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

	if (p->q->ops->drop == NULL)
		return 0;

	len = p->q->ops->drop(p->q);
	if (len)
		sch->q.qlen--;

	return len;
}

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;
	u8 *mask;

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

	if (!opt)
		goto errout;

	err = nla_parse_nested(tb, TCA_DSMARK_MAX, opt, dsmark_policy);
	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]);

	mask = kmalloc(indices * 2, GFP_KERNEL);
	if (mask == NULL) {
		err = -ENOMEM;
		goto errout;
	}

	p->mask = mask;
	memset(p->mask, 0xff, indices);

	p->value = p->mask + indices;
	memset(p->value, 0, indices);

	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, &pfifo_qdisc_ops, sch->handle);
	if (p->q == NULL)
		p->q = &noop_qdisc;

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

	err = 0;
errout:
	return err;
}

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

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

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

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

	tcf_destroy_chain(p->filter_list);
	qdisc_destroy(p->q);
	kfree(p->mask);
}

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("dsmark_dump_class(sch %p,[qdisc %p],class %ld\n", 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;
	NLA_PUT_U8(skb, TCA_DSMARK_MASK, p->mask[cl-1]);
	NLA_PUT_U8(skb, TCA_DSMARK_VALUE, p->value[cl-1]);

	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;
	NLA_PUT_U16(skb, TCA_DSMARK_INDICES, p->indices);

	if (p->default_index != NO_DEFAULT_INDEX)
		NLA_PUT_U16(skb, TCA_DSMARK_DEFAULT_INDEX, p->default_index);

	if (p->set_tc_index)
		NLA_PUT_FLAG(skb, TCA_DSMARK_SET_TC_INDEX);

	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,
	.requeue	=	dsmark_requeue,
	.drop		=	dsmark_drop,
	.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>
	7	#include <linux/init.h>
	8	#include <linux/types.h>
	9	#include <linux/string.h>
	10	#include <linux/errno.h>
	11	#include <linux/skbuff.h>
1da177e4	12	#include <linux/rtnetlink.h>
5b0ac72b	13	#include <linux/bitops.h>
1da177e4 LT	14	#include <net/pkt_sched.h>
	15	#include <net/dsfield.h>
	16	#include <net/inet_ecn.h>
	17	#include <asm/byteorder.h>
	18
1da177e4 LT	19	/*
	20	* classid class marking
	21	* ------- ----- -------
	22	* n/a 0 n/a
	23	* x:0 1 use entry [0]
	24	* ... ... ...
	25	* x:y y>0 y+1 use entry [y]
	26	* ... ... ...
	27	* x:indices-1 indices use entry [indices-1]
	28	* ... ... ...
	29	* x:y y+1 use entry [y & (indices-1)]
	30	* ... ... ...
	31	* 0xffff 0x10000 use entry [indices-1]
	32	*/
	33
	34
	35	#define NO_DEFAULT_INDEX (1 << 16)
	36
	37	struct dsmark_qdisc_data {
	38	struct Qdisc *q;
	39	struct tcf_proto *filter_list;
af0d1141 TG	40	u8 mask; / "owns" the array */
	41	u8 *value;
	42	u16 indices;
	43	u32 default_index; /* index range is 0...0xffff */
1da177e4 LT	44	int set_tc_index;
	45	};
	46
758cc43c TG	47	static inline int dsmark_valid_index(struct dsmark_qdisc_data *p, u16 index)
	48	{
	49	return (index <= p->indices && index > 0);
	50	}
1da177e4 LT	51
	52	/* ------------------------- Class/flow operations ------------------------- */
	53
af0d1141 TG	54	static int dsmark_graft(struct Qdisc *sch, unsigned long arg,
af0d1141 TG	55	struct Qdisc new, struct Qdisc *old)
1da177e4	56	{
81da99ed	57	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	58
81da99ed	59	pr_debug("dsmark_graft(sch %p,[qdisc %p],new %p,old %p)\n",
af0d1141	60	sch, p, new, old);
486b53e5 TG	61
486b53e5 TG	62	if (new == NULL) {
9f9afec4 PM	63	new = qdisc_create_dflt(sch->dev, &pfifo_qdisc_ops,
9f9afec4 PM	64	sch->handle);
486b53e5 TG	65	if (new == NULL)
	66	new = &noop_qdisc;
	67	}
	68
1da177e4	69	sch_tree_lock(sch);
af0d1141	70	*old = xchg(&p->q, new);
5e50da01	71	qdisc_tree_decrease_qlen(old, (old)->q.qlen);
af0d1141	72	qdisc_reset(*old);
af0d1141 TG	73	sch_tree_unlock(sch);
af0d1141 TG	74
10297b99	75	return 0;
1da177e4 LT	76	}
1da177e4 LT	77
1da177e4 LT	78	static struct Qdisc dsmark_leaf(struct Qdisc sch, unsigned long arg)
1da177e4 LT	79	{
81da99ed SH	80	struct dsmark_qdisc_data *p = qdisc_priv(sch);
81da99ed SH	81	return p->q;
1da177e4 LT	82	}
1da177e4 LT	83
af0d1141	84	static unsigned long dsmark_get(struct Qdisc *sch, u32 classid)
1da177e4	85	{
81da99ed SH	86	pr_debug("dsmark_get(sch %p,[qdisc %p],classid %x)\n",
81da99ed SH	87	sch, qdisc_priv(sch), classid);
1da177e4	88
af0d1141	89	return TC_H_MIN(classid) + 1;
1da177e4 LT	90	}
1da177e4 LT	91
1da177e4	92	static unsigned long dsmark_bind_filter(struct Qdisc *sch,
af0d1141	93	unsigned long parent, u32 classid)
1da177e4	94	{
af0d1141	95	return dsmark_get(sch, classid);
1da177e4 LT	96	}
1da177e4 LT	97
1da177e4 LT	98	static void dsmark_put(struct Qdisc *sch, unsigned long cl)
	99	{
	100	}
	101
27a3421e PM	102	static const struct nla_policy dsmark_policy[TCA_DSMARK_MAX + 1] = {
	103	[TCA_DSMARK_INDICES] = { .type = NLA_U16 },
	104	[TCA_DSMARK_DEFAULT_INDEX] = { .type = NLA_U16 },
	105	[TCA_DSMARK_SET_TC_INDEX] = { .type = NLA_FLAG },
	106	[TCA_DSMARK_MASK] = { .type = NLA_U8 },
	107	[TCA_DSMARK_VALUE] = { .type = NLA_U8 },
	108	};
	109
1da177e4	110	static int dsmark_change(struct Qdisc *sch, u32 classid, u32 parent,
1e90474c	111	struct nlattr *tca, unsigned long arg)
1da177e4	112	{
81da99ed	113	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1e90474c PM	114	struct nlattr *opt = tca[TCA_OPTIONS];
1e90474c PM	115	struct nlattr *tb[TCA_DSMARK_MAX + 1];
758cc43c TG	116	int err = -EINVAL;
758cc43c TG	117	u8 mask = 0;
1da177e4	118
81da99ed	119	pr_debug("dsmark_change(sch %p,[qdisc %p],classid %x,parent %x),"
758cc43c TG	120	"arg 0x%lx\n", sch, p, classid, parent, *arg);
	121
	122	if (!dsmark_valid_index(p, *arg)) {
	123	err = -ENOENT;
1e90474c	124	goto errout;
1da177e4	125	}
1da177e4	126
cee63723	127	if (!opt)
1e90474c	128	goto errout;
758cc43c	129
27a3421e	130	err = nla_parse_nested(tb, TCA_DSMARK_MAX, opt, dsmark_policy);
cee63723	131	if (err < 0)
27a3421e	132	goto errout;
cee63723	133
27a3421e	134	if (tb[TCA_DSMARK_MASK])
1e90474c	135	mask = nla_get_u8(tb[TCA_DSMARK_MASK]);
27a3421e PM	136
27a3421e PM	137	if (tb[TCA_DSMARK_VALUE])
1e90474c	138	p->value[*arg-1] = nla_get_u8(tb[TCA_DSMARK_VALUE]);
10297b99	139
1e90474c	140	if (tb[TCA_DSMARK_MASK])
758cc43c TG	141	p->mask[*arg-1] = mask;
	142
	143	err = 0;
	144
1e90474c	145	errout:
758cc43c TG	146	return err;
758cc43c TG	147	}
1da177e4	148
af0d1141	149	static int dsmark_delete(struct Qdisc *sch, unsigned long arg)
1da177e4	150	{
81da99ed	151	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	152
af0d1141	153	if (!dsmark_valid_index(p, arg))
1da177e4	154	return -EINVAL;
10297b99	155
1da177e4 LT	156	p->mask[arg-1] = 0xff;
1da177e4 LT	157	p->value[arg-1] = 0;
af0d1141	158
1da177e4 LT	159	return 0;
	160	}
	161
9d127fbd	162	static void dsmark_walk(struct Qdisc sch, struct qdisc_walker walker)
1da177e4	163	{
81da99ed	164	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4 LT	165	int i;
1da177e4 LT	166
81da99ed	167	pr_debug("dsmark_walk(sch %p,[qdisc %p],walker %p)\n", sch, p, walker);
af0d1141	168
1da177e4 LT	169	if (walker->stop)
1da177e4 LT	170	return;
af0d1141	171
1da177e4 LT	172	for (i = 0; i < p->indices; i++) {
1da177e4 LT	173	if (p->mask[i] == 0xff && !p->value[i])
0451eb07	174	goto ignore;
1da177e4 LT	175	if (walker->count >= walker->skip) {
	176	if (walker->fn(sch, i+1, walker) < 0) {
	177	walker->stop = 1;
	178	break;
	179	}
	180	}
10297b99	181	ignore:
0451eb07	182	walker->count++;
10297b99	183	}
1da177e4 LT	184	}
1da177e4 LT	185
9d127fbd SH	186	static inline struct tcf_proto *dsmark_find_tcf(struct Qdisc sch,
9d127fbd SH	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
9d127fbd	195	static int dsmark_enqueue(struct sk_buff skb, struct Qdisc sch)
1da177e4	196	{
81da99ed	197	struct dsmark_qdisc_data *p = qdisc_priv(sch);
af0d1141 TG	198	int err;
af0d1141 TG	199
81da99ed	200	pr_debug("dsmark_enqueue(skb %p,sch %p,[qdisc %p])\n", skb, sch, p);
1da177e4	201
1da177e4	202	if (p->set_tc_index) {
1da177e4	203	switch (skb->protocol) {
9d127fbd SH	204	case __constant_htons(ETH_P_IP):
	205	if (skb_cow_head(skb, sizeof(struct iphdr)))
	206	goto drop;
4c30719f	207
9d127fbd SH	208	skb->tc_index = ipv4_get_dsfield(ip_hdr(skb))
	209	& ~INET_ECN_MASK;
	210	break;
4c30719f	211
9d127fbd SH	212	case __constant_htons(ETH_P_IPV6):
	213	if (skb_cow_head(skb, sizeof(struct ipv6hdr)))
	214	goto drop;
4c30719f	215
9d127fbd SH	216	skb->tc_index = ipv6_get_dsfield(ipv6_hdr(skb))
	217	& ~INET_ECN_MASK;
	218	break;
	219	default:
	220	skb->tc_index = 0;
	221	break;
3ff50b79	222	}
1da177e4	223	}
af0d1141 TG	224
af0d1141 TG	225	if (TC_H_MAJ(skb->priority) == sch->handle)
1da177e4	226	skb->tc_index = TC_H_MIN(skb->priority);
af0d1141 TG	227	else {
	228	struct tcf_result res;
	229	int result = tc_classify(skb, p->filter_list, &res);
	230
81da99ed	231	pr_debug("result %d class 0x%04x\n", result, res.classid);
af0d1141	232
1da177e4	233	switch (result) {
f6853e2d PM	234	#ifdef CONFIG_NET_CLS_ACT
	235	case TC_ACT_QUEUED:
	236	case TC_ACT_STOLEN:
	237	kfree_skb(skb);
	238	return NET_XMIT_SUCCESS;
4c30719f	239
f6853e2d	240	case TC_ACT_SHOT:
4c30719f	241	goto drop;
1da177e4	242	#endif
c3bc7cff	243	case TC_ACT_OK:
f6853e2d PM	244	skb->tc_index = TC_H_MIN(res.classid);
f6853e2d PM	245	break;
4c30719f	246
f6853e2d PM	247	default:
	248	if (p->default_index != NO_DEFAULT_INDEX)
	249	skb->tc_index = p->default_index;
	250	break;
3ff50b79	251	}
1da177e4	252	}
1da177e4	253
9d127fbd	254	err = p->q->enqueue(skb, p->q);
af0d1141	255	if (err != NET_XMIT_SUCCESS) {
1da177e4	256	sch->qstats.drops++;
af0d1141	257	return err;
1da177e4	258	}
af0d1141	259
1da177e4 LT	260	sch->bstats.bytes += skb->len;
	261	sch->bstats.packets++;
	262	sch->q.qlen++;
1da177e4	263
af0d1141	264	return NET_XMIT_SUCCESS;
4c30719f SH	265
	266	drop:
	267	kfree_skb(skb);
	268	sch->qstats.drops++;
	269	return NET_XMIT_BYPASS;
af0d1141	270	}
1da177e4 LT	271
	272	static struct sk_buff dsmark_dequeue(struct Qdisc sch)
	273	{
81da99ed	274	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	275	struct sk_buff *skb;
af0d1141 TG	276	u32 index;
af0d1141 TG	277
81da99ed	278	pr_debug("dsmark_dequeue(sch %p,[qdisc %p])\n", sch, p);
1da177e4	279
1da177e4	280	skb = p->q->ops->dequeue(p->q);
af0d1141	281	if (skb == NULL)
1da177e4	282	return NULL;
af0d1141	283
1da177e4	284	sch->q.qlen--;
af0d1141 TG	285
af0d1141 TG	286	index = skb->tc_index & (p->indices - 1);
81da99ed	287	pr_debug("index %d->%d\n", skb->tc_index, index);
af0d1141	288
1da177e4	289	switch (skb->protocol) {
9d127fbd SH	290	case __constant_htons(ETH_P_IP):
	291	ipv4_change_dsfield(ip_hdr(skb), p->mask[index],
	292	p->value[index]);
1da177e4	293	break;
9d127fbd SH	294	case __constant_htons(ETH_P_IPV6):
	295	ipv6_change_dsfield(ipv6_hdr(skb), p->mask[index],
	296	p->value[index]);
1da177e4	297	break;
9d127fbd SH	298	default:
	299	/*
	300	* Only complain if a change was actually attempted.
	301	* This way, we can send non-IP traffic through dsmark
	302	* and don't need yet another qdisc as a bypass.
	303	*/
	304	if (p->mask[index] != 0xff \|\| p->value[index])
	305	printk(KERN_WARNING
	306	"dsmark_dequeue: unsupported protocol %d\n",
	307	ntohs(skb->protocol));
	308	break;
3ff50b79	309	}
af0d1141	310
1da177e4 LT	311	return skb;
	312	}
	313
9d127fbd	314	static int dsmark_requeue(struct sk_buff skb, struct Qdisc sch)
1da177e4	315	{
81da99ed	316	struct dsmark_qdisc_data *p = qdisc_priv(sch);
af0d1141	317	int err;
1da177e4	318
81da99ed	319	pr_debug("dsmark_requeue(skb %p,sch %p,[qdisc %p])\n", skb, sch, p);
af0d1141 TG	320
	321	err = p->q->ops->requeue(skb, p->q);
	322	if (err != NET_XMIT_SUCCESS) {
	323	sch->qstats.drops++;
	324	return err;
1da177e4	325	}
1da177e4	326
af0d1141 TG	327	sch->q.qlen++;
	328	sch->qstats.requeues++;
	329
	330	return NET_XMIT_SUCCESS;
	331	}
1da177e4 LT	332
	333	static unsigned int dsmark_drop(struct Qdisc *sch)
	334	{
81da99ed	335	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	336	unsigned int len;
10297b99	337
81da99ed	338	pr_debug("dsmark_reset(sch %p,[qdisc %p])\n", sch, p);
af0d1141 TG	339
af0d1141 TG	340	if (p->q->ops->drop == NULL)
1da177e4	341	return 0;
af0d1141 TG	342
	343	len = p->q->ops->drop(p->q);
	344	if (len)
	345	sch->q.qlen--;
	346
1da177e4 LT	347	return len;
	348	}
	349
1e90474c	350	static int dsmark_init(struct Qdisc sch, struct nlattr opt)
1da177e4	351	{
81da99ed	352	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1e90474c	353	struct nlattr *tb[TCA_DSMARK_MAX + 1];
758cc43c TG	354	int err = -EINVAL;
	355	u32 default_index = NO_DEFAULT_INDEX;
	356	u16 indices;
	357	u8 *mask;
	358
81da99ed	359	pr_debug("dsmark_init(sch %p,[qdisc %p],opt %p)\n", sch, p, opt);
758cc43c	360
cee63723 PM	361	if (!opt)
	362	goto errout;
	363
27a3421e	364	err = nla_parse_nested(tb, TCA_DSMARK_MAX, opt, dsmark_policy);
cee63723	365	if (err < 0)
758cc43c TG	366	goto errout;
758cc43c TG	367
cee63723	368	err = -EINVAL;
1e90474c	369	indices = nla_get_u16(tb[TCA_DSMARK_INDICES]);
5b0ac72b DM	370
5b0ac72b DM	371	if (hweight32(indices) != 1)
758cc43c TG	372	goto errout;
758cc43c TG	373
27a3421e	374	if (tb[TCA_DSMARK_DEFAULT_INDEX])
1e90474c	375	default_index = nla_get_u16(tb[TCA_DSMARK_DEFAULT_INDEX]);
758cc43c TG	376
	377	mask = kmalloc(indices * 2, GFP_KERNEL);
	378	if (mask == NULL) {
	379	err = -ENOMEM;
	380	goto errout;
1da177e4	381	}
758cc43c TG	382
	383	p->mask = mask;
	384	memset(p->mask, 0xff, indices);
	385
	386	p->value = p->mask + indices;
	387	memset(p->value, 0, indices);
	388
	389	p->indices = indices;
	390	p->default_index = default_index;
1e90474c	391	p->set_tc_index = nla_get_flag(tb[TCA_DSMARK_SET_TC_INDEX]);
758cc43c	392
9f9afec4	393	p->q = qdisc_create_dflt(sch->dev, &pfifo_qdisc_ops, sch->handle);
758cc43c	394	if (p->q == NULL)
1da177e4	395	p->q = &noop_qdisc;
758cc43c	396
81da99ed	397	pr_debug("dsmark_init: qdisc %p\n", p->q);
758cc43c TG	398
	399	err = 0;
	400	errout:
758cc43c	401	return err;
1da177e4 LT	402	}
1da177e4 LT	403
1da177e4 LT	404	static void dsmark_reset(struct Qdisc *sch)
1da177e4 LT	405	{
81da99ed	406	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	407
81da99ed	408	pr_debug("dsmark_reset(sch %p,[qdisc %p])\n", sch, p);
1da177e4 LT	409	qdisc_reset(p->q);
	410	sch->q.qlen = 0;
	411	}
	412
1da177e4 LT	413	static void dsmark_destroy(struct Qdisc *sch)
1da177e4 LT	414	{
81da99ed	415	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1da177e4	416
81da99ed	417	pr_debug("dsmark_destroy(sch %p,[qdisc %p])\n", sch, p);
af0d1141	418
a48b5a61	419	tcf_destroy_chain(p->filter_list);
1da177e4 LT	420	qdisc_destroy(p->q);
	421	kfree(p->mask);
	422	}
	423
1da177e4	424	static int dsmark_dump_class(struct Qdisc *sch, unsigned long cl,
02f23f09	425	struct sk_buff skb, struct tcmsg tcm)
1da177e4	426	{
81da99ed	427	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1e90474c	428	struct nlattr *opts = NULL;
1da177e4	429
81da99ed	430	pr_debug("dsmark_dump_class(sch %p,[qdisc %p],class %ld\n", sch, p, cl);
02f23f09 TG	431
02f23f09 TG	432	if (!dsmark_valid_index(p, cl))
1da177e4	433	return -EINVAL;
02f23f09 TG	434
02f23f09 TG	435	tcm->tcm_handle = TC_H_MAKE(TC_H_MAJ(sch->handle), cl-1);
cdc7f8e3	436	tcm->tcm_info = p->q->handle;
02f23f09	437
1e90474c PM	438	opts = nla_nest_start(skb, TCA_OPTIONS);
	439	if (opts == NULL)
	440	goto nla_put_failure;
	441	NLA_PUT_U8(skb, TCA_DSMARK_MASK, p->mask[cl-1]);
	442	NLA_PUT_U8(skb, TCA_DSMARK_VALUE, p->value[cl-1]);
02f23f09	443
1e90474c	444	return nla_nest_end(skb, opts);
1da177e4	445
1e90474c	446	nla_put_failure:
bc3ed28c TG	447	nla_nest_cancel(skb, opts);
bc3ed28c TG	448	return -EMSGSIZE;
1da177e4 LT	449	}
	450
	451	static int dsmark_dump(struct Qdisc sch, struct sk_buff skb)
	452	{
81da99ed	453	struct dsmark_qdisc_data *p = qdisc_priv(sch);
1e90474c	454	struct nlattr *opts = NULL;
1da177e4	455
1e90474c PM	456	opts = nla_nest_start(skb, TCA_OPTIONS);
	457	if (opts == NULL)
	458	goto nla_put_failure;
	459	NLA_PUT_U16(skb, TCA_DSMARK_INDICES, p->indices);
02f23f09 TG	460
02f23f09 TG	461	if (p->default_index != NO_DEFAULT_INDEX)
1e90474c	462	NLA_PUT_U16(skb, TCA_DSMARK_DEFAULT_INDEX, p->default_index);
1da177e4	463
1da177e4	464	if (p->set_tc_index)
1e90474c	465	NLA_PUT_FLAG(skb, TCA_DSMARK_SET_TC_INDEX);
02f23f09	466
1e90474c	467	return nla_nest_end(skb, opts);
1da177e4	468
1e90474c	469	nla_put_failure:
bc3ed28c TG	470	nla_nest_cancel(skb, opts);
bc3ed28c TG	471	return -EMSGSIZE;
1da177e4 LT	472	}
1da177e4 LT	473
20fea08b	474	static const struct Qdisc_class_ops dsmark_class_ops = {
1da177e4 LT	475	.graft = dsmark_graft,
	476	.leaf = dsmark_leaf,
	477	.get = dsmark_get,
	478	.put = dsmark_put,
	479	.change = dsmark_change,
	480	.delete = dsmark_delete,
	481	.walk = dsmark_walk,
	482	.tcf_chain = dsmark_find_tcf,
	483	.bind_tcf = dsmark_bind_filter,
	484	.unbind_tcf = dsmark_put,
	485	.dump = dsmark_dump_class,
	486	};
	487
20fea08b	488	static struct Qdisc_ops dsmark_qdisc_ops __read_mostly = {
1da177e4 LT	489	.next = NULL,
	490	.cl_ops = &dsmark_class_ops,
	491	.id = "dsmark",
	492	.priv_size = sizeof(struct dsmark_qdisc_data),
	493	.enqueue = dsmark_enqueue,
	494	.dequeue = dsmark_dequeue,
	495	.requeue = dsmark_requeue,
	496	.drop = dsmark_drop,
	497	.init = dsmark_init,
	498	.reset = dsmark_reset,
	499	.destroy = dsmark_destroy,
	500	.change = NULL,
	501	.dump = dsmark_dump,
	502	.owner = THIS_MODULE,
	503	};
	504
	505	static int __init dsmark_module_init(void)
	506	{
	507	return register_qdisc(&dsmark_qdisc_ops);
	508	}
af0d1141	509
10297b99	510	static void __exit dsmark_module_exit(void)
1da177e4 LT	511	{
	512	unregister_qdisc(&dsmark_qdisc_ops);
	513	}
af0d1141	514
1da177e4 LT	515	module_init(dsmark_module_init)
1da177e4 LT	516	module_exit(dsmark_module_exit)
af0d1141	517
1da177e4	518	MODULE_LICENSE("GPL");