[linux.git] / net / sched / act_skbmod.c

/*
 * net/sched/act_skbmod.c  skb data modifier
 *
 * Copyright (c) 2016 Jamal Hadi Salim <[email protected]>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
*/

#include <linux/module.h>
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/skbuff.h>
#include <linux/rtnetlink.h>
#include <net/netlink.h>
#include <net/pkt_sched.h>

#include <linux/tc_act/tc_skbmod.h>
#include <net/tc_act/tc_skbmod.h>

static unsigned int skbmod_net_id;
static struct tc_action_ops act_skbmod_ops;

#define MAX_EDIT_LEN ETH_HLEN
static int tcf_skbmod_run(struct sk_buff *skb, const struct tc_action *a,
			  struct tcf_result *res)
{
	struct tcf_skbmod *d = to_skbmod(a);
	int action;
	struct tcf_skbmod_params *p;
	u64 flags;
	int err;

	tcf_lastuse_update(&d->tcf_tm);
	bstats_cpu_update(this_cpu_ptr(d->common.cpu_bstats), skb);

	/* XXX: if you are going to edit more fields beyond ethernet header
	 * (example when you add IP header replacement or vlan swap)
	 * then MAX_EDIT_LEN needs to change appropriately
	*/
	err = skb_ensure_writable(skb, MAX_EDIT_LEN);
	if (unlikely(err)) { /* best policy is to drop on the floor */
		qstats_overlimit_inc(this_cpu_ptr(d->common.cpu_qstats));
		return TC_ACT_SHOT;
	}

	rcu_read_lock();
	action = READ_ONCE(d->tcf_action);
	if (unlikely(action == TC_ACT_SHOT)) {
		qstats_overlimit_inc(this_cpu_ptr(d->common.cpu_qstats));
		rcu_read_unlock();
		return action;
	}

	p = rcu_dereference(d->skbmod_p);
	flags = p->flags;
	if (flags & SKBMOD_F_DMAC)
		ether_addr_copy(eth_hdr(skb)->h_dest, p->eth_dst);
	if (flags & SKBMOD_F_SMAC)
		ether_addr_copy(eth_hdr(skb)->h_source, p->eth_src);
	if (flags & SKBMOD_F_ETYPE)
		eth_hdr(skb)->h_proto = p->eth_type;
	rcu_read_unlock();

	if (flags & SKBMOD_F_SWAPMAC) {
		u16 tmpaddr[ETH_ALEN / 2]; /* ether_addr_copy() requirement */
		/*XXX: I am sure we can come up with more efficient swapping*/
		ether_addr_copy((u8 *)tmpaddr, eth_hdr(skb)->h_dest);
		ether_addr_copy(eth_hdr(skb)->h_dest, eth_hdr(skb)->h_source);
		ether_addr_copy(eth_hdr(skb)->h_source, (u8 *)tmpaddr);
	}

	return action;
}

static const struct nla_policy skbmod_policy[TCA_SKBMOD_MAX + 1] = {
	[TCA_SKBMOD_PARMS]		= { .len = sizeof(struct tc_skbmod) },
	[TCA_SKBMOD_DMAC]		= { .len = ETH_ALEN },
	[TCA_SKBMOD_SMAC]		= { .len = ETH_ALEN },
	[TCA_SKBMOD_ETYPE]		= { .type = NLA_U16 },
};

static int tcf_skbmod_init(struct net *net, struct nlattr *nla,
			   struct nlattr *est, struct tc_action **a,
			   int ovr, int bind, bool rtnl_held,
			   struct netlink_ext_ack *extack)
{
	struct tc_action_net *tn = net_generic(net, skbmod_net_id);
	struct nlattr *tb[TCA_SKBMOD_MAX + 1];
	struct tcf_skbmod_params *p, *p_old;
	struct tc_skbmod *parm;
	struct tcf_skbmod *d;
	bool exists = false;
	u8 *daddr = NULL;
	u8 *saddr = NULL;
	u16 eth_type = 0;
	u32 lflags = 0;
	int ret = 0, err;

	if (!nla)
		return -EINVAL;

	err = nla_parse_nested(tb, TCA_SKBMOD_MAX, nla, skbmod_policy, NULL);
	if (err < 0)
		return err;

	if (!tb[TCA_SKBMOD_PARMS])
		return -EINVAL;

	if (tb[TCA_SKBMOD_DMAC]) {
		daddr = nla_data(tb[TCA_SKBMOD_DMAC]);
		lflags |= SKBMOD_F_DMAC;
	}

	if (tb[TCA_SKBMOD_SMAC]) {
		saddr = nla_data(tb[TCA_SKBMOD_SMAC]);
		lflags |= SKBMOD_F_SMAC;
	}

	if (tb[TCA_SKBMOD_ETYPE]) {
		eth_type = nla_get_u16(tb[TCA_SKBMOD_ETYPE]);
		lflags |= SKBMOD_F_ETYPE;
	}

	parm = nla_data(tb[TCA_SKBMOD_PARMS]);
	if (parm->flags & SKBMOD_F_SWAPMAC)
		lflags = SKBMOD_F_SWAPMAC;

	exists = tcf_idr_check(tn, parm->index, a, bind);
	if (exists && bind)
		return 0;

	if (!lflags) {
		if (exists)
			tcf_idr_release(*a, bind);
		return -EINVAL;
	}

	if (!exists) {
		ret = tcf_idr_create(tn, parm->index, est, a,
				     &act_skbmod_ops, bind, true);
		if (ret)
			return ret;

		ret = ACT_P_CREATED;
	} else {
		tcf_idr_release(*a, bind);
		if (!ovr)
			return -EEXIST;
	}

	d = to_skbmod(*a);

	ASSERT_RTNL();
	p = kzalloc(sizeof(struct tcf_skbmod_params), GFP_KERNEL);
	if (unlikely(!p)) {
		if (ret == ACT_P_CREATED)
			tcf_idr_release(*a, bind);
		return -ENOMEM;
	}

	p->flags = lflags;
	d->tcf_action = parm->action;

	p_old = rtnl_dereference(d->skbmod_p);

	if (ovr)
		spin_lock_bh(&d->tcf_lock);

	if (lflags & SKBMOD_F_DMAC)
		ether_addr_copy(p->eth_dst, daddr);
	if (lflags & SKBMOD_F_SMAC)
		ether_addr_copy(p->eth_src, saddr);
	if (lflags & SKBMOD_F_ETYPE)
		p->eth_type = htons(eth_type);

	rcu_assign_pointer(d->skbmod_p, p);
	if (ovr)
		spin_unlock_bh(&d->tcf_lock);

	if (p_old)
		kfree_rcu(p_old, rcu);

	if (ret == ACT_P_CREATED)
		tcf_idr_insert(tn, *a);
	return ret;
}

static void tcf_skbmod_cleanup(struct tc_action *a)
{
	struct tcf_skbmod *d = to_skbmod(a);
	struct tcf_skbmod_params  *p;

	p = rcu_dereference_protected(d->skbmod_p, 1);
	if (p)
		kfree_rcu(p, rcu);
}

static int tcf_skbmod_dump(struct sk_buff *skb, struct tc_action *a,
			   int bind, int ref)
{
	struct tcf_skbmod *d = to_skbmod(a);
	unsigned char *b = skb_tail_pointer(skb);
	struct tcf_skbmod_params  *p = rtnl_dereference(d->skbmod_p);
	struct tc_skbmod opt = {
		.index   = d->tcf_index,
		.refcnt  = refcount_read(&d->tcf_refcnt) - ref,
		.bindcnt = atomic_read(&d->tcf_bindcnt) - bind,
		.action  = d->tcf_action,
	};
	struct tcf_t t;

	opt.flags  = p->flags;
	if (nla_put(skb, TCA_SKBMOD_PARMS, sizeof(opt), &opt))
		goto nla_put_failure;
	if ((p->flags & SKBMOD_F_DMAC) &&
	    nla_put(skb, TCA_SKBMOD_DMAC, ETH_ALEN, p->eth_dst))
		goto nla_put_failure;
	if ((p->flags & SKBMOD_F_SMAC) &&
	    nla_put(skb, TCA_SKBMOD_SMAC, ETH_ALEN, p->eth_src))
		goto nla_put_failure;
	if ((p->flags & SKBMOD_F_ETYPE) &&
	    nla_put_u16(skb, TCA_SKBMOD_ETYPE, ntohs(p->eth_type)))
		goto nla_put_failure;

	tcf_tm_dump(&t, &d->tcf_tm);
	if (nla_put_64bit(skb, TCA_SKBMOD_TM, sizeof(t), &t, TCA_SKBMOD_PAD))
		goto nla_put_failure;

	return skb->len;
nla_put_failure:
	nlmsg_trim(skb, b);
	return -1;
}

static int tcf_skbmod_walker(struct net *net, struct sk_buff *skb,
			     struct netlink_callback *cb, int type,
			     const struct tc_action_ops *ops,
			     struct netlink_ext_ack *extack)
{
	struct tc_action_net *tn = net_generic(net, skbmod_net_id);

	return tcf_generic_walker(tn, skb, cb, type, ops, extack);
}

static int tcf_skbmod_search(struct net *net, struct tc_action **a, u32 index,
			     struct netlink_ext_ack *extack)
{
	struct tc_action_net *tn = net_generic(net, skbmod_net_id);

	return tcf_idr_search(tn, a, index);
}

static struct tc_action_ops act_skbmod_ops = {
	.kind		=	"skbmod",
	.type		=	TCA_ACT_SKBMOD,
	.owner		=	THIS_MODULE,
	.act		=	tcf_skbmod_run,
	.dump		=	tcf_skbmod_dump,
	.init		=	tcf_skbmod_init,
	.cleanup	=	tcf_skbmod_cleanup,
	.walk		=	tcf_skbmod_walker,
	.lookup		=	tcf_skbmod_search,
	.size		=	sizeof(struct tcf_skbmod),
};

static __net_init int skbmod_init_net(struct net *net)
{
	struct tc_action_net *tn = net_generic(net, skbmod_net_id);

	return tc_action_net_init(tn, &act_skbmod_ops);
}

static void __net_exit skbmod_exit_net(struct list_head *net_list)
{
	tc_action_net_exit(net_list, skbmod_net_id);
}

static struct pernet_operations skbmod_net_ops = {
	.init = skbmod_init_net,
	.exit_batch = skbmod_exit_net,
	.id   = &skbmod_net_id,
	.size = sizeof(struct tc_action_net),
};

MODULE_AUTHOR("Jamal Hadi Salim, <[email protected]>");
MODULE_DESCRIPTION("SKB data mod-ing");
MODULE_LICENSE("GPL");

static int __init skbmod_init_module(void)
{
	return tcf_register_action(&act_skbmod_ops, &skbmod_net_ops);
}

static void __exit skbmod_cleanup_module(void)
{
	tcf_unregister_action(&act_skbmod_ops, &skbmod_net_ops);
}

module_init(skbmod_init_module);
module_exit(skbmod_cleanup_module);
Commit	Line	Data
86da71b5 JHS	1	/*
	2	* net/sched/act_skbmod.c skb data modifier
	3	*
	4	* Copyright (c) 2016 Jamal Hadi Salim <[email protected]>
	5	*
	6	* This program is free software; you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation; either version 2 of the License, or
	9	* (at your option) any later version.
	10	*/
	11
	12	#include <linux/module.h>
	13	#include <linux/init.h>
	14	#include <linux/kernel.h>
	15	#include <linux/skbuff.h>
	16	#include <linux/rtnetlink.h>
	17	#include <net/netlink.h>
	18	#include <net/pkt_sched.h>
	19
	20	#include <linux/tc_act/tc_skbmod.h>
	21	#include <net/tc_act/tc_skbmod.h>
	22
c7d03a00	23	static unsigned int skbmod_net_id;
86da71b5 JHS	24	static struct tc_action_ops act_skbmod_ops;
	25
	26	#define MAX_EDIT_LEN ETH_HLEN
	27	static int tcf_skbmod_run(struct sk_buff skb, const struct tc_action a,
	28	struct tcf_result *res)
	29	{
	30	struct tcf_skbmod *d = to_skbmod(a);
	31	int action;
	32	struct tcf_skbmod_params *p;
	33	u64 flags;
	34	int err;
	35
	36	tcf_lastuse_update(&d->tcf_tm);
	37	bstats_cpu_update(this_cpu_ptr(d->common.cpu_bstats), skb);
	38
	39	/* XXX: if you are going to edit more fields beyond ethernet header
	40	* (example when you add IP header replacement or vlan swap)
	41	* then MAX_EDIT_LEN needs to change appropriately
	42	*/
	43	err = skb_ensure_writable(skb, MAX_EDIT_LEN);
	44	if (unlikely(err)) { /* best policy is to drop on the floor */
	45	qstats_overlimit_inc(this_cpu_ptr(d->common.cpu_qstats));
	46	return TC_ACT_SHOT;
	47	}
	48
	49	rcu_read_lock();
	50	action = READ_ONCE(d->tcf_action);
	51	if (unlikely(action == TC_ACT_SHOT)) {
	52	qstats_overlimit_inc(this_cpu_ptr(d->common.cpu_qstats));
	53	rcu_read_unlock();
	54	return action;
	55	}
	56
	57	p = rcu_dereference(d->skbmod_p);
	58	flags = p->flags;
	59	if (flags & SKBMOD_F_DMAC)
	60	ether_addr_copy(eth_hdr(skb)->h_dest, p->eth_dst);
	61	if (flags & SKBMOD_F_SMAC)
	62	ether_addr_copy(eth_hdr(skb)->h_source, p->eth_src);
	63	if (flags & SKBMOD_F_ETYPE)
	64	eth_hdr(skb)->h_proto = p->eth_type;
	65	rcu_read_unlock();
	66
	67	if (flags & SKBMOD_F_SWAPMAC) {
	68	u16 tmpaddr[ETH_ALEN / 2]; /* ether_addr_copy() requirement */
	69	/XXX: I am sure we can come up with more efficient swapping/
	70	ether_addr_copy((u8 *)tmpaddr, eth_hdr(skb)->h_dest);
	71	ether_addr_copy(eth_hdr(skb)->h_dest, eth_hdr(skb)->h_source);
	72	ether_addr_copy(eth_hdr(skb)->h_source, (u8 *)tmpaddr);
	73	}
	74
	75	return action;
	76	}
	77
	78	static const struct nla_policy skbmod_policy[TCA_SKBMOD_MAX + 1] = {
	79	[TCA_SKBMOD_PARMS] = { .len = sizeof(struct tc_skbmod) },
	80	[TCA_SKBMOD_DMAC] = { .len = ETH_ALEN },
	81	[TCA_SKBMOD_SMAC] = { .len = ETH_ALEN },
	82	[TCA_SKBMOD_ETYPE] = { .type = NLA_U16 },
	83	};
	84
	85	static int tcf_skbmod_init(struct net net, struct nlattr nla,
	86	struct nlattr est, struct tc_action *a,
789871bb VB	87	int ovr, int bind, bool rtnl_held,
789871bb VB	88	struct netlink_ext_ack *extack)
86da71b5 JHS	89	{
	90	struct tc_action_net *tn = net_generic(net, skbmod_net_id);
	91	struct nlattr *tb[TCA_SKBMOD_MAX + 1];
	92	struct tcf_skbmod_params p, p_old;
	93	struct tc_skbmod *parm;
	94	struct tcf_skbmod *d;
	95	bool exists = false;
	96	u8 *daddr = NULL;
	97	u8 *saddr = NULL;
	98	u16 eth_type = 0;
	99	u32 lflags = 0;
	100	int ret = 0, err;
	101
	102	if (!nla)
	103	return -EINVAL;
	104
fceb6435	105	err = nla_parse_nested(tb, TCA_SKBMOD_MAX, nla, skbmod_policy, NULL);
86da71b5 JHS	106	if (err < 0)
	107	return err;
	108
	109	if (!tb[TCA_SKBMOD_PARMS])
	110	return -EINVAL;
	111
	112	if (tb[TCA_SKBMOD_DMAC]) {
	113	daddr = nla_data(tb[TCA_SKBMOD_DMAC]);
	114	lflags \|= SKBMOD_F_DMAC;
	115	}
	116
	117	if (tb[TCA_SKBMOD_SMAC]) {
	118	saddr = nla_data(tb[TCA_SKBMOD_SMAC]);
	119	lflags \|= SKBMOD_F_SMAC;
	120	}
	121
	122	if (tb[TCA_SKBMOD_ETYPE]) {
	123	eth_type = nla_get_u16(tb[TCA_SKBMOD_ETYPE]);
	124	lflags \|= SKBMOD_F_ETYPE;
	125	}
	126
	127	parm = nla_data(tb[TCA_SKBMOD_PARMS]);
	128	if (parm->flags & SKBMOD_F_SWAPMAC)
	129	lflags = SKBMOD_F_SWAPMAC;
	130
65a206c0	131	exists = tcf_idr_check(tn, parm->index, a, bind);
86da71b5 JHS	132	if (exists && bind)
	133	return 0;
	134
a52956df RM	135	if (!lflags) {
	136	if (exists)
	137	tcf_idr_release(*a, bind);
86da71b5	138	return -EINVAL;
a52956df	139	}
86da71b5 JHS	140
86da71b5 JHS	141	if (!exists) {
65a206c0 CM	142	ret = tcf_idr_create(tn, parm->index, est, a,
65a206c0 CM	143	&act_skbmod_ops, bind, true);
86da71b5 JHS	144	if (ret)
	145	return ret;
	146
	147	ret = ACT_P_CREATED;
	148	} else {
65a206c0	149	tcf_idr_release(*a, bind);
86da71b5 JHS	150	if (!ovr)
	151	return -EEXIST;
	152	}
	153
	154	d = to_skbmod(*a);
	155
	156	ASSERT_RTNL();
	157	p = kzalloc(sizeof(struct tcf_skbmod_params), GFP_KERNEL);
	158	if (unlikely(!p)) {
f29cdfbe	159	if (ret == ACT_P_CREATED)
65a206c0	160	tcf_idr_release(*a, bind);
86da71b5 JHS	161	return -ENOMEM;
	162	}
	163
	164	p->flags = lflags;
	165	d->tcf_action = parm->action;
	166
	167	p_old = rtnl_dereference(d->skbmod_p);
	168
	169	if (ovr)
	170	spin_lock_bh(&d->tcf_lock);
	171
	172	if (lflags & SKBMOD_F_DMAC)
	173	ether_addr_copy(p->eth_dst, daddr);
	174	if (lflags & SKBMOD_F_SMAC)
	175	ether_addr_copy(p->eth_src, saddr);
	176	if (lflags & SKBMOD_F_ETYPE)
	177	p->eth_type = htons(eth_type);
	178
	179	rcu_assign_pointer(d->skbmod_p, p);
	180	if (ovr)
	181	spin_unlock_bh(&d->tcf_lock);
	182
	183	if (p_old)
	184	kfree_rcu(p_old, rcu);
	185
	186	if (ret == ACT_P_CREATED)
65a206c0	187	tcf_idr_insert(tn, *a);
86da71b5 JHS	188	return ret;
	189	}
	190
9a63b255	191	static void tcf_skbmod_cleanup(struct tc_action *a)
86da71b5 JHS	192	{
	193	struct tcf_skbmod *d = to_skbmod(a);
	194	struct tcf_skbmod_params *p;
	195
	196	p = rcu_dereference_protected(d->skbmod_p, 1);
2d433610 DC	197	if (p)
2d433610 DC	198	kfree_rcu(p, rcu);
86da71b5 JHS	199	}
	200
	201	static int tcf_skbmod_dump(struct sk_buff skb, struct tc_action a,
	202	int bind, int ref)
	203	{
	204	struct tcf_skbmod *d = to_skbmod(a);
	205	unsigned char *b = skb_tail_pointer(skb);
	206	struct tcf_skbmod_params *p = rtnl_dereference(d->skbmod_p);
	207	struct tc_skbmod opt = {
	208	.index = d->tcf_index,
036bb443 VB	209	.refcnt = refcount_read(&d->tcf_refcnt) - ref,
036bb443 VB	210	.bindcnt = atomic_read(&d->tcf_bindcnt) - bind,
86da71b5 JHS	211	.action = d->tcf_action,
	212	};
	213	struct tcf_t t;
	214
	215	opt.flags = p->flags;
	216	if (nla_put(skb, TCA_SKBMOD_PARMS, sizeof(opt), &opt))
	217	goto nla_put_failure;
	218	if ((p->flags & SKBMOD_F_DMAC) &&
	219	nla_put(skb, TCA_SKBMOD_DMAC, ETH_ALEN, p->eth_dst))
	220	goto nla_put_failure;
	221	if ((p->flags & SKBMOD_F_SMAC) &&
	222	nla_put(skb, TCA_SKBMOD_SMAC, ETH_ALEN, p->eth_src))
	223	goto nla_put_failure;
	224	if ((p->flags & SKBMOD_F_ETYPE) &&
	225	nla_put_u16(skb, TCA_SKBMOD_ETYPE, ntohs(p->eth_type)))
	226	goto nla_put_failure;
	227
	228	tcf_tm_dump(&t, &d->tcf_tm);
	229	if (nla_put_64bit(skb, TCA_SKBMOD_TM, sizeof(t), &t, TCA_SKBMOD_PAD))
	230	goto nla_put_failure;
	231
	232	return skb->len;
	233	nla_put_failure:
86da71b5 JHS	234	nlmsg_trim(skb, b);
	235	return -1;
	236	}
	237
	238	static int tcf_skbmod_walker(struct net net, struct sk_buff skb,
	239	struct netlink_callback *cb, int type,
41780105 AA	240	const struct tc_action_ops *ops,
41780105 AA	241	struct netlink_ext_ack *extack)
86da71b5 JHS	242	{
	243	struct tc_action_net *tn = net_generic(net, skbmod_net_id);
	244
b3620145	245	return tcf_generic_walker(tn, skb, cb, type, ops, extack);
86da71b5 JHS	246	}
86da71b5 JHS	247
331a9295 AA	248	static int tcf_skbmod_search(struct net net, struct tc_action *a, u32 index,
331a9295 AA	249	struct netlink_ext_ack *extack)
86da71b5 JHS	250	{
	251	struct tc_action_net *tn = net_generic(net, skbmod_net_id);
	252
65a206c0	253	return tcf_idr_search(tn, a, index);
86da71b5 JHS	254	}
	255
	256	static struct tc_action_ops act_skbmod_ops = {
	257	.kind = "skbmod",
	258	.type = TCA_ACT_SKBMOD,
	259	.owner = THIS_MODULE,
	260	.act = tcf_skbmod_run,
	261	.dump = tcf_skbmod_dump,
	262	.init = tcf_skbmod_init,
	263	.cleanup = tcf_skbmod_cleanup,
	264	.walk = tcf_skbmod_walker,
	265	.lookup = tcf_skbmod_search,
	266	.size = sizeof(struct tcf_skbmod),
	267	};
	268
	269	static __net_init int skbmod_init_net(struct net *net)
	270	{
	271	struct tc_action_net *tn = net_generic(net, skbmod_net_id);
	272
c7e460ce	273	return tc_action_net_init(tn, &act_skbmod_ops);
86da71b5 JHS	274	}
86da71b5 JHS	275
039af9c6	276	static void __net_exit skbmod_exit_net(struct list_head *net_list)
86da71b5	277	{
039af9c6	278	tc_action_net_exit(net_list, skbmod_net_id);
86da71b5 JHS	279	}
	280
	281	static struct pernet_operations skbmod_net_ops = {
	282	.init = skbmod_init_net,
039af9c6	283	.exit_batch = skbmod_exit_net,
86da71b5 JHS	284	.id = &skbmod_net_id,
	285	.size = sizeof(struct tc_action_net),
	286	};
	287
	288	MODULE_AUTHOR("Jamal Hadi Salim, <[email protected]>");
	289	MODULE_DESCRIPTION("SKB data mod-ing");
	290	MODULE_LICENSE("GPL");
	291
	292	static int __init skbmod_init_module(void)
	293	{
	294	return tcf_register_action(&act_skbmod_ops, &skbmod_net_ops);
	295	}
	296
	297	static void __exit skbmod_cleanup_module(void)
	298	{
	299	tcf_unregister_action(&act_skbmod_ops, &skbmod_net_ops);
	300	}
	301
	302	module_init(skbmod_init_module);
	303	module_exit(skbmod_cleanup_module);