[mirror_ubuntu-bionic-kernel.git] / net / sched / act_sample.c

/*
 * net/sched/act_sample.c - Packet sampling tc action
 * Copyright (c) 2017 Yotam Gigi <yotamg@mellanox.com>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */

#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <linux/rtnetlink.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/gfp.h>
#include <net/net_namespace.h>
#include <net/netlink.h>
#include <net/pkt_sched.h>
#include <linux/tc_act/tc_sample.h>
#include <net/tc_act/tc_sample.h>
#include <net/psample.h>

#include <linux/if_arp.h>

static unsigned int sample_net_id;
static struct tc_action_ops act_sample_ops;

static const struct nla_policy sample_policy[TCA_SAMPLE_MAX + 1] = {
	[TCA_SAMPLE_PARMS]		= { .len = sizeof(struct tc_sample) },
	[TCA_SAMPLE_RATE]		= { .type = NLA_U32 },
	[TCA_SAMPLE_TRUNC_SIZE]		= { .type = NLA_U32 },
	[TCA_SAMPLE_PSAMPLE_GROUP]	= { .type = NLA_U32 },
};

static int tcf_sample_init(struct net *net, struct nlattr *nla,
			   struct nlattr *est, struct tc_action **a, int ovr,
			   int bind)
{
	struct tc_action_net *tn = net_generic(net, sample_net_id);
	struct nlattr *tb[TCA_SAMPLE_MAX + 1];
	struct psample_group *psample_group;
	struct tc_sample *parm;
	struct tcf_sample *s;
	bool exists = false;
	u32 rate;
	int ret;

	if (!nla)
		return -EINVAL;
	ret = nla_parse_nested(tb, TCA_SAMPLE_MAX, nla, sample_policy, NULL);
	if (ret < 0)
		return ret;
	if (!tb[TCA_SAMPLE_PARMS] || !tb[TCA_SAMPLE_RATE] ||
	    !tb[TCA_SAMPLE_PSAMPLE_GROUP])
		return -EINVAL;

	parm = nla_data(tb[TCA_SAMPLE_PARMS]);

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

	if (!exists) {
		ret = tcf_idr_create(tn, parm->index, est, a,
				     &act_sample_ops, bind, true);
		if (ret)
			return ret;
		ret = ACT_P_CREATED;
	} else {
		tcf_idr_release(*a, bind);
		if (!ovr)
			return -EEXIST;
	}

	rate = nla_get_u32(tb[TCA_SAMPLE_RATE]);
	if (!rate) {
		tcf_idr_release(*a, bind);
		return -EINVAL;
	}

	s = to_sample(*a);
	s->tcf_action = parm->action;
	s->rate = nla_get_u32(tb[TCA_SAMPLE_RATE]);
	s->rate = rate;
	s->psample_group_num = nla_get_u32(tb[TCA_SAMPLE_PSAMPLE_GROUP]);
	psample_group = psample_group_get(net, s->psample_group_num);
	if (!psample_group) {
		if (ret == ACT_P_CREATED)
			tcf_idr_release(*a, bind);
		return -ENOMEM;
	}
	rcu_swap_protected(s->psample_group, psample_group,
			   lockdep_is_held(&s->tcf_lock));

	if (tb[TCA_SAMPLE_TRUNC_SIZE]) {
		s->truncate = true;
		s->trunc_size = nla_get_u32(tb[TCA_SAMPLE_TRUNC_SIZE]);
	}

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

static void tcf_sample_cleanup(struct tc_action *a, int bind)
{
	struct tcf_sample *s = to_sample(a);
	struct psample_group *psample_group;

	psample_group = rtnl_dereference(s->psample_group);
	RCU_INIT_POINTER(s->psample_group, NULL);
	if (psample_group)
		psample_group_put(psample_group);
}

static bool tcf_sample_dev_ok_push(struct net_device *dev)
{
	switch (dev->type) {
	case ARPHRD_TUNNEL:
	case ARPHRD_TUNNEL6:
	case ARPHRD_SIT:
	case ARPHRD_IPGRE:
	case ARPHRD_VOID:
	case ARPHRD_NONE:
		return false;
	default:
		return true;
	}
}

static int tcf_sample_act(struct sk_buff *skb, const struct tc_action *a,
			  struct tcf_result *res)
{
	struct tcf_sample *s = to_sample(a);
	struct psample_group *psample_group;
	int retval;
	int size;
	int iif;
	int oif;

	tcf_lastuse_update(&s->tcf_tm);
	bstats_cpu_update(this_cpu_ptr(s->common.cpu_bstats), skb);
	retval = READ_ONCE(s->tcf_action);

	rcu_read_lock();
	psample_group = rcu_dereference(s->psample_group);

	/* randomly sample packets according to rate */
	if (psample_group && (prandom_u32() % s->rate == 0)) {
		if (!skb_at_tc_ingress(skb)) {
			iif = skb->skb_iif;
			oif = skb->dev->ifindex;
		} else {
			iif = skb->dev->ifindex;
			oif = 0;
		}

		/* on ingress, the mac header gets popped, so push it back */
		if (skb_at_tc_ingress(skb) && tcf_sample_dev_ok_push(skb->dev))
			skb_push(skb, skb->mac_len);

		size = s->truncate ? s->trunc_size : skb->len;
		psample_sample_packet(psample_group, skb, size, iif, oif,
				      s->rate);

		if (skb_at_tc_ingress(skb) && tcf_sample_dev_ok_push(skb->dev))
			skb_pull(skb, skb->mac_len);
	}

	rcu_read_unlock();
	return retval;
}

static int tcf_sample_dump(struct sk_buff *skb, struct tc_action *a,
			   int bind, int ref)
{
	unsigned char *b = skb_tail_pointer(skb);
	struct tcf_sample *s = to_sample(a);
	struct tc_sample opt = {
		.index      = s->tcf_index,
		.action     = s->tcf_action,
		.refcnt     = s->tcf_refcnt - ref,
		.bindcnt    = s->tcf_bindcnt - bind,
	};
	struct tcf_t t;

	if (nla_put(skb, TCA_SAMPLE_PARMS, sizeof(opt), &opt))
		goto nla_put_failure;

	tcf_tm_dump(&t, &s->tcf_tm);
	if (nla_put_64bit(skb, TCA_SAMPLE_TM, sizeof(t), &t, TCA_SAMPLE_PAD))
		goto nla_put_failure;

	if (nla_put_u32(skb, TCA_SAMPLE_RATE, s->rate))
		goto nla_put_failure;

	if (s->truncate)
		if (nla_put_u32(skb, TCA_SAMPLE_TRUNC_SIZE, s->trunc_size))
			goto nla_put_failure;

	if (nla_put_u32(skb, TCA_SAMPLE_PSAMPLE_GROUP, s->psample_group_num))
		goto nla_put_failure;
	return skb->len;

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

static int tcf_sample_walker(struct net *net, struct sk_buff *skb,
			     struct netlink_callback *cb, int type,
			     const struct tc_action_ops *ops)
{
	struct tc_action_net *tn = net_generic(net, sample_net_id);

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

static int tcf_sample_search(struct net *net, struct tc_action **a, u32 index)
{
	struct tc_action_net *tn = net_generic(net, sample_net_id);

	return tcf_idr_search(tn, a, index);
}

static struct tc_action_ops act_sample_ops = {
	.kind	  = "sample",
	.type	  = TCA_ACT_SAMPLE,
	.owner	  = THIS_MODULE,
	.act	  = tcf_sample_act,
	.dump	  = tcf_sample_dump,
	.init	  = tcf_sample_init,
	.cleanup  = tcf_sample_cleanup,
	.walk	  = tcf_sample_walker,
	.lookup	  = tcf_sample_search,
	.size	  = sizeof(struct tcf_sample),
};

static __net_init int sample_init_net(struct net *net)
{
	struct tc_action_net *tn = net_generic(net, sample_net_id);

	return tc_action_net_init(net, tn, &act_sample_ops);
}

static void __net_exit sample_exit_net(struct net *net)
{
	struct tc_action_net *tn = net_generic(net, sample_net_id);

	tc_action_net_exit(tn);
}

static struct pernet_operations sample_net_ops = {
	.init = sample_init_net,
	.exit = sample_exit_net,
	.id   = &sample_net_id,
	.size = sizeof(struct tc_action_net),
};

static int __init sample_init_module(void)
{
	return tcf_register_action(&act_sample_ops, &sample_net_ops);
}

static void __exit sample_cleanup_module(void)
{
	tcf_unregister_action(&act_sample_ops, &sample_net_ops);
}

module_init(sample_init_module);
module_exit(sample_cleanup_module);

MODULE_AUTHOR("Yotam Gigi <yotam.gi@gmail.com>");
MODULE_DESCRIPTION("Packet sampling action");
MODULE_LICENSE("GPL v2");
Commit	Line	Data
5c5670fa YG	1	/*
	2	* net/sched/act_sample.c - Packet sampling tc action
	3	* Copyright (c) 2017 Yotam Gigi <yotamg@mellanox.com>
	4	*
	5	* This program is free software; you can redistribute it and/or modify
	6	* it under the terms of the GNU General Public License version 2 as
	7	* published by the Free Software Foundation.
	8	*/
	9
	10	#include <linux/types.h>
	11	#include <linux/kernel.h>
	12	#include <linux/string.h>
	13	#include <linux/errno.h>
	14	#include <linux/skbuff.h>
	15	#include <linux/rtnetlink.h>
	16	#include <linux/module.h>
	17	#include <linux/init.h>
	18	#include <linux/gfp.h>
	19	#include <net/net_namespace.h>
	20	#include <net/netlink.h>
	21	#include <net/pkt_sched.h>
	22	#include <linux/tc_act/tc_sample.h>
	23	#include <net/tc_act/tc_sample.h>
	24	#include <net/psample.h>
	25
	26	#include <linux/if_arp.h>
	27
5c5670fa YG	28	static unsigned int sample_net_id;
	29	static struct tc_action_ops act_sample_ops;
	30
	31	static const struct nla_policy sample_policy[TCA_SAMPLE_MAX + 1] = {
	32	[TCA_SAMPLE_PARMS] = { .len = sizeof(struct tc_sample) },
	33	[TCA_SAMPLE_RATE] = { .type = NLA_U32 },
	34	[TCA_SAMPLE_TRUNC_SIZE] = { .type = NLA_U32 },
	35	[TCA_SAMPLE_PSAMPLE_GROUP] = { .type = NLA_U32 },
	36	};
	37
	38	static int tcf_sample_init(struct net net, struct nlattr nla,
	39	struct nlattr est, struct tc_action *a, int ovr,
	40	int bind)
	41	{
	42	struct tc_action_net *tn = net_generic(net, sample_net_id);
	43	struct nlattr *tb[TCA_SAMPLE_MAX + 1];
	44	struct psample_group *psample_group;
	45	struct tc_sample *parm;
	46	struct tcf_sample *s;
	47	bool exists = false;
4aa971ec	48	u32 rate;
5c5670fa YG	49	int ret;
	50
	51	if (!nla)
	52	return -EINVAL;
fceb6435	53	ret = nla_parse_nested(tb, TCA_SAMPLE_MAX, nla, sample_policy, NULL);
5c5670fa YG	54	if (ret < 0)
	55	return ret;
	56	if (!tb[TCA_SAMPLE_PARMS] \|\| !tb[TCA_SAMPLE_RATE] \|\|
	57	!tb[TCA_SAMPLE_PSAMPLE_GROUP])
	58	return -EINVAL;
	59
	60	parm = nla_data(tb[TCA_SAMPLE_PARMS]);
	61
65a206c0	62	exists = tcf_idr_check(tn, parm->index, a, bind);
5c5670fa YG	63	if (exists && bind)
	64	return 0;
	65
	66	if (!exists) {
65a206c0	67	ret = tcf_idr_create(tn, parm->index, est, a,
511b1e5c	68	&act_sample_ops, bind, true);
5c5670fa YG	69	if (ret)
	70	return ret;
	71	ret = ACT_P_CREATED;
	72	} else {
65a206c0	73	tcf_idr_release(*a, bind);
5c5670fa YG	74	if (!ovr)
	75	return -EEXIST;
	76	}
5c5670fa	77
4aa971ec DC	78	rate = nla_get_u32(tb[TCA_SAMPLE_RATE]);
	79	if (!rate) {
	80	tcf_idr_release(*a, bind);
	81	return -EINVAL;
	82	}
	83
	84	s = to_sample(*a);
5c5670fa YG	85	s->tcf_action = parm->action;
5c5670fa YG	86	s->rate = nla_get_u32(tb[TCA_SAMPLE_RATE]);
4aa971ec	87	s->rate = rate;
5c5670fa YG	88	s->psample_group_num = nla_get_u32(tb[TCA_SAMPLE_PSAMPLE_GROUP]);
5c5670fa YG	89	psample_group = psample_group_get(net, s->psample_group_num);
cadb9c9f YG	90	if (!psample_group) {
cadb9c9f YG	91	if (ret == ACT_P_CREATED)
65a206c0	92	tcf_idr_release(*a, bind);
5c5670fa	93	return -ENOMEM;
cadb9c9f	94	}
a0c82d24 VB	95	rcu_swap_protected(s->psample_group, psample_group,
a0c82d24 VB	96	lockdep_is_held(&s->tcf_lock));
5c5670fa YG	97
	98	if (tb[TCA_SAMPLE_TRUNC_SIZE]) {
	99	s->truncate = true;
	100	s->trunc_size = nla_get_u32(tb[TCA_SAMPLE_TRUNC_SIZE]);
	101	}
	102
a0c82d24 VB	103	if (psample_group)
a0c82d24 VB	104	psample_group_put(psample_group);
5c5670fa	105	if (ret == ACT_P_CREATED)
65a206c0	106	tcf_idr_insert(tn, *a);
5c5670fa YG	107	return ret;
	108	}
	109
90a6ec85	110	static void tcf_sample_cleanup(struct tc_action *a, int bind)
5c5670fa	111	{
90a6ec85	112	struct tcf_sample *s = to_sample(a);
5c5670fa YG	113	struct psample_group *psample_group;
5c5670fa YG	114
90a6ec85	115	psample_group = rtnl_dereference(s->psample_group);
5c5670fa	116	RCU_INIT_POINTER(s->psample_group, NULL);
c65dc67b DC	117	if (psample_group)
c65dc67b DC	118	psample_group_put(psample_group);
5c5670fa YG	119	}
5c5670fa YG	120
5c5670fa YG	121	static bool tcf_sample_dev_ok_push(struct net_device *dev)
	122	{
	123	switch (dev->type) {
	124	case ARPHRD_TUNNEL:
	125	case ARPHRD_TUNNEL6:
	126	case ARPHRD_SIT:
	127	case ARPHRD_IPGRE:
	128	case ARPHRD_VOID:
	129	case ARPHRD_NONE:
	130	return false;
	131	default:
	132	return true;
	133	}
	134	}
	135
	136	static int tcf_sample_act(struct sk_buff skb, const struct tc_action a,
	137	struct tcf_result *res)
	138	{
	139	struct tcf_sample *s = to_sample(a);
	140	struct psample_group *psample_group;
	141	int retval;
	142	int size;
	143	int iif;
	144	int oif;
	145
	146	tcf_lastuse_update(&s->tcf_tm);
	147	bstats_cpu_update(this_cpu_ptr(s->common.cpu_bstats), skb);
	148	retval = READ_ONCE(s->tcf_action);
	149
	150	rcu_read_lock();
	151	psample_group = rcu_dereference(s->psample_group);
	152
	153	/* randomly sample packets according to rate */
	154	if (psample_group && (prandom_u32() % s->rate == 0)) {
	155	if (!skb_at_tc_ingress(skb)) {
	156	iif = skb->skb_iif;
	157	oif = skb->dev->ifindex;
	158	} else {
	159	iif = skb->dev->ifindex;
	160	oif = 0;
	161	}
	162
	163	/* on ingress, the mac header gets popped, so push it back */
	164	if (skb_at_tc_ingress(skb) && tcf_sample_dev_ok_push(skb->dev))
	165	skb_push(skb, skb->mac_len);
	166
	167	size = s->truncate ? s->trunc_size : skb->len;
	168	psample_sample_packet(psample_group, skb, size, iif, oif,
	169	s->rate);
	170
	171	if (skb_at_tc_ingress(skb) && tcf_sample_dev_ok_push(skb->dev))
	172	skb_pull(skb, skb->mac_len);
	173	}
	174
	175	rcu_read_unlock();
	176	return retval;
	177	}
	178
	179	static int tcf_sample_dump(struct sk_buff skb, struct tc_action a,
	180	int bind, int ref)
	181	{
	182	unsigned char *b = skb_tail_pointer(skb);
	183	struct tcf_sample *s = to_sample(a);
	184	struct tc_sample opt = {
185	.index = s->tcf_index,
186	.action = s->tcf_action,
187	.refcnt = s->tcf_refcnt - ref,
188	.bindcnt = s->tcf_bindcnt - bind,
189	};
190	struct tcf_t t;
191
192	if (nla_put(skb, TCA_SAMPLE_PARMS, sizeof(opt), &opt))
193	goto nla_put_failure;
194
195	tcf_tm_dump(&t, &s->tcf_tm);
196	if (nla_put_64bit(skb, TCA_SAMPLE_TM, sizeof(t), &t, TCA_SAMPLE_PAD))
197	goto nla_put_failure;
198
199	if (nla_put_u32(skb, TCA_SAMPLE_RATE, s->rate))
200	goto nla_put_failure;
201
202	if (s->truncate)
203	if (nla_put_u32(skb, TCA_SAMPLE_TRUNC_SIZE, s->trunc_size))
204	goto nla_put_failure;
205
206	if (nla_put_u32(skb, TCA_SAMPLE_PSAMPLE_GROUP, s->psample_group_num))
207	goto nla_put_failure;
208	return skb->len;
209
210	nla_put_failure:
211	nlmsg_trim(skb, b);
212	return -1;
213	}
214
215	static int tcf_sample_walker(struct net net, struct sk_buff skb,
216	struct netlink_callback *cb, int type,
217	const struct tc_action_ops *ops)
218	{
219	struct tc_action_net *tn = net_generic(net, sample_net_id);
220
221	return tcf_generic_walker(tn, skb, cb, type, ops);
222	}
223
224	static int tcf_sample_search(struct net net, struct tc_action *a, u32 index)
225	{
226	struct tc_action_net *tn = net_generic(net, sample_net_id);
227
65a206c0	228	return tcf_idr_search(tn, a, index);
5c5670fa YG	229	}
	230
	231	static struct tc_action_ops act_sample_ops = {
	232	.kind = "sample",
	233	.type = TCA_ACT_SAMPLE,
	234	.owner = THIS_MODULE,
	235	.act = tcf_sample_act,
	236	.dump = tcf_sample_dump,
	237	.init = tcf_sample_init,
	238	.cleanup = tcf_sample_cleanup,
	239	.walk = tcf_sample_walker,
	240	.lookup = tcf_sample_search,
	241	.size = sizeof(struct tcf_sample),
	242	};
	243
	244	static __net_init int sample_init_net(struct net *net)
	245	{
	246	struct tc_action_net *tn = net_generic(net, sample_net_id);
	247
8f4d0354	248	return tc_action_net_init(net, tn, &act_sample_ops);
5c5670fa YG	249	}
	250
	251	static void __net_exit sample_exit_net(struct net *net)
	252	{
	253	struct tc_action_net *tn = net_generic(net, sample_net_id);
	254
	255	tc_action_net_exit(tn);
	256	}
	257
	258	static struct pernet_operations sample_net_ops = {
	259	.init = sample_init_net,
	260	.exit = sample_exit_net,
	261	.id = &sample_net_id,
	262	.size = sizeof(struct tc_action_net),
	263	};
	264
	265	static int __init sample_init_module(void)
	266	{
	267	return tcf_register_action(&act_sample_ops, &sample_net_ops);
	268	}
	269
	270	static void __exit sample_cleanup_module(void)
	271	{
	272	tcf_unregister_action(&act_sample_ops, &sample_net_ops);
	273	}
	274
	275	module_init(sample_init_module);
	276	module_exit(sample_cleanup_module);
	277
f1fd20c3	278	MODULE_AUTHOR("Yotam Gigi <yotam.gi@gmail.com>");
5c5670fa YG	279	MODULE_DESCRIPTION("Packet sampling action");
5c5670fa YG	280	MODULE_LICENSE("GPL v2");