[thirdparty/kernel/stable.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, bool rtnl_held,
			   struct netlink_ext_ack *extack)
{
	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;
	u32 psample_group_num;
	struct tcf_sample *s;
	bool exists = false;
	int ret, err;

	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]);

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

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

	psample_group_num = nla_get_u32(tb[TCA_SAMPLE_PSAMPLE_GROUP]);
	psample_group = psample_group_get(net, psample_group_num);
	if (!psample_group) {
		tcf_idr_release(*a, bind);
		return -ENOMEM;
	}

	s = to_sample(*a);

	spin_lock_bh(&s->tcf_lock);
	s->tcf_action = parm->action;
	s->rate = nla_get_u32(tb[TCA_SAMPLE_RATE]);
	s->psample_group_num = psample_group_num;
	RCU_INIT_POINTER(s->psample_group, psample_group);

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

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

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

	/* last reference to action, no need to lock */
	psample_group = rcu_dereference_protected(s->psample_group, 1);
	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);

	psample_group = rcu_dereference_bh(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);
	}

	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,
		.refcnt     = refcount_read(&s->tcf_refcnt) - ref,
		.bindcnt    = atomic_read(&s->tcf_bindcnt) - bind,
	};
	struct tcf_t t;

	spin_lock_bh(&s->tcf_lock);
	opt.action = s->tcf_action;
	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;
	spin_unlock_bh(&s->tcf_lock);

	return skb->len;

nla_put_failure:
	spin_unlock_bh(&s->tcf_lock);
	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 netlink_ext_ack *extack)
{
	struct tc_action_net *tn = net_generic(net, sample_net_id);

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

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",
	.id	  = TCA_ID_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(tn, &act_sample_ops);
}

static void __net_exit sample_exit_net(struct list_head *net_list)
{
	tc_action_net_exit(net_list, sample_net_id);
}

static struct pernet_operations sample_net_ops = {
	.init = sample_init_net,
	.exit_batch = 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,
789871bb VB	40	int bind, bool rtnl_held,
789871bb VB	41	struct netlink_ext_ack *extack)
5c5670fa YG	42	{
	43	struct tc_action_net *tn = net_generic(net, sample_net_id);
	44	struct nlattr *tb[TCA_SAMPLE_MAX + 1];
	45	struct psample_group *psample_group;
	46	struct tc_sample *parm;
653cd284	47	u32 psample_group_num;
5c5670fa YG	48	struct tcf_sample *s;
5c5670fa YG	49	bool exists = false;
0190c1d4	50	int ret, err;
5c5670fa YG	51
	52	if (!nla)
	53	return -EINVAL;
fceb6435	54	ret = nla_parse_nested(tb, TCA_SAMPLE_MAX, nla, sample_policy, NULL);
5c5670fa YG	55	if (ret < 0)
	56	return ret;
	57	if (!tb[TCA_SAMPLE_PARMS] \|\| !tb[TCA_SAMPLE_RATE] \|\|
	58	!tb[TCA_SAMPLE_PSAMPLE_GROUP])
	59	return -EINVAL;
	60
	61	parm = nla_data(tb[TCA_SAMPLE_PARMS]);
	62
0190c1d4 VB	63	err = tcf_idr_check_alloc(tn, &parm->index, a, bind);
	64	if (err < 0)
	65	return err;
	66	exists = err;
5c5670fa YG	67	if (exists && bind)
	68	return 0;
	69
	70	if (!exists) {
65a206c0	71	ret = tcf_idr_create(tn, parm->index, est, a,
34043d25	72	&act_sample_ops, bind, true);
0190c1d4 VB	73	if (ret) {
0190c1d4 VB	74	tcf_idr_cleanup(tn, parm->index);
5c5670fa	75	return ret;
0190c1d4	76	}
5c5670fa	77	ret = ACT_P_CREATED;
4e8ddd7f	78	} else if (!ovr) {
65a206c0	79	tcf_idr_release(*a, bind);
4e8ddd7f	80	return -EEXIST;
5c5670fa	81	}
5c5670fa	82
653cd284 VB	83	psample_group_num = nla_get_u32(tb[TCA_SAMPLE_PSAMPLE_GROUP]);
653cd284 VB	84	psample_group = psample_group_get(net, psample_group_num);
cadb9c9f	85	if (!psample_group) {
4e8ddd7f	86	tcf_idr_release(*a, bind);
5c5670fa	87	return -ENOMEM;
cadb9c9f	88	}
653cd284 VB	89
	90	s = to_sample(*a);
	91
	92	spin_lock_bh(&s->tcf_lock);
	93	s->tcf_action = parm->action;
	94	s->rate = nla_get_u32(tb[TCA_SAMPLE_RATE]);
	95	s->psample_group_num = psample_group_num;
5c5670fa YG	96	RCU_INIT_POINTER(s->psample_group, psample_group);
	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	}
653cd284	102	spin_unlock_bh(&s->tcf_lock);
5c5670fa YG	103
5c5670fa YG	104	if (ret == ACT_P_CREATED)
65a206c0	105	tcf_idr_insert(tn, *a);
5c5670fa YG	106	return ret;
	107	}
	108
9a63b255	109	static void tcf_sample_cleanup(struct tc_action *a)
5c5670fa	110	{
90a6ec85	111	struct tcf_sample *s = to_sample(a);
5c5670fa YG	112	struct psample_group *psample_group;
5c5670fa YG	113
d7728495 VB	114	/* last reference to action, no need to lock */
d7728495 VB	115	psample_group = rcu_dereference_protected(s->psample_group, 1);
5c5670fa	116	RCU_INIT_POINTER(s->psample_group, NULL);
1f110e7c DC	117	if (psample_group)
1f110e7c 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
7fd4b288	150	psample_group = rcu_dereference_bh(s->psample_group);
5c5670fa YG	151
	152	/* randomly sample packets according to rate */
	153	if (psample_group && (prandom_u32() % s->rate == 0)) {
	154	if (!skb_at_tc_ingress(skb)) {
	155	iif = skb->skb_iif;
	156	oif = skb->dev->ifindex;
	157	} else {
	158	iif = skb->dev->ifindex;
	159	oif = 0;
	160	}
	161
	162	/* on ingress, the mac header gets popped, so push it back */
	163	if (skb_at_tc_ingress(skb) && tcf_sample_dev_ok_push(skb->dev))
	164	skb_push(skb, skb->mac_len);
	165
	166	size = s->truncate ? s->trunc_size : skb->len;
	167	psample_sample_packet(psample_group, skb, size, iif, oif,
	168	s->rate);
	169
	170	if (skb_at_tc_ingress(skb) && tcf_sample_dev_ok_push(skb->dev))
	171	skb_pull(skb, skb->mac_len);
	172	}
	173
5c5670fa YG	174	return retval;
	175	}
	176
	177	static int tcf_sample_dump(struct sk_buff skb, struct tc_action a,
	178	int bind, int ref)
	179	{
	180	unsigned char *b = skb_tail_pointer(skb);
	181	struct tcf_sample *s = to_sample(a);
	182	struct tc_sample opt = {
	183	.index = s->tcf_index,
036bb443 VB	184	.refcnt = refcount_read(&s->tcf_refcnt) - ref,
036bb443 VB	185	.bindcnt = atomic_read(&s->tcf_bindcnt) - bind,
5c5670fa YG	186	};
	187	struct tcf_t t;
	188
653cd284	189	spin_lock_bh(&s->tcf_lock);
d7728495	190	opt.action = s->tcf_action;
5c5670fa YG	191	if (nla_put(skb, TCA_SAMPLE_PARMS, sizeof(opt), &opt))
	192	goto nla_put_failure;
	193
	194	tcf_tm_dump(&t, &s->tcf_tm);
	195	if (nla_put_64bit(skb, TCA_SAMPLE_TM, sizeof(t), &t, TCA_SAMPLE_PAD))
	196	goto nla_put_failure;
	197
	198	if (nla_put_u32(skb, TCA_SAMPLE_RATE, s->rate))
	199	goto nla_put_failure;
	200
	201	if (s->truncate)
	202	if (nla_put_u32(skb, TCA_SAMPLE_TRUNC_SIZE, s->trunc_size))
	203	goto nla_put_failure;
	204
	205	if (nla_put_u32(skb, TCA_SAMPLE_PSAMPLE_GROUP, s->psample_group_num))
	206	goto nla_put_failure;
653cd284	207	spin_unlock_bh(&s->tcf_lock);
d7728495	208
5c5670fa YG	209	return skb->len;
	210
	211	nla_put_failure:
653cd284	212	spin_unlock_bh(&s->tcf_lock);
5c5670fa YG	213	nlmsg_trim(skb, b);
	214	return -1;
	215	}
	216
	217	static int tcf_sample_walker(struct net net, struct sk_buff skb,
	218	struct netlink_callback *cb, int type,
41780105 AA	219	const struct tc_action_ops *ops,
41780105 AA	220	struct netlink_ext_ack *extack)
5c5670fa YG	221	{
	222	struct tc_action_net *tn = net_generic(net, sample_net_id);
	223
b3620145	224	return tcf_generic_walker(tn, skb, cb, type, ops, extack);
5c5670fa YG	225	}
5c5670fa YG	226
f061b48c	227	static int tcf_sample_search(struct net net, struct tc_action *a, u32 index)
5c5670fa YG	228	{
	229	struct tc_action_net *tn = net_generic(net, sample_net_id);
	230
65a206c0	231	return tcf_idr_search(tn, a, index);
5c5670fa YG	232	}
	233
	234	static struct tc_action_ops act_sample_ops = {
	235	.kind = "sample",
eddd2cf1	236	.id = TCA_ID_SAMPLE,
5c5670fa YG	237	.owner = THIS_MODULE,
	238	.act = tcf_sample_act,
	239	.dump = tcf_sample_dump,
	240	.init = tcf_sample_init,
	241	.cleanup = tcf_sample_cleanup,
	242	.walk = tcf_sample_walker,
	243	.lookup = tcf_sample_search,
	244	.size = sizeof(struct tcf_sample),
	245	};
	246
	247	static __net_init int sample_init_net(struct net *net)
	248	{
	249	struct tc_action_net *tn = net_generic(net, sample_net_id);
	250
c7e460ce	251	return tc_action_net_init(tn, &act_sample_ops);
5c5670fa YG	252	}
5c5670fa YG	253
039af9c6	254	static void __net_exit sample_exit_net(struct list_head *net_list)
5c5670fa	255	{
039af9c6	256	tc_action_net_exit(net_list, sample_net_id);
5c5670fa YG	257	}
	258
	259	static struct pernet_operations sample_net_ops = {
	260	.init = sample_init_net,
039af9c6	261	.exit_batch = sample_exit_net,
5c5670fa YG	262	.id = &sample_net_id,
	263	.size = sizeof(struct tc_action_net),
	264	};
	265
	266	static int __init sample_init_module(void)
	267	{
	268	return tcf_register_action(&act_sample_ops, &sample_net_ops);
	269	}
	270
	271	static void __exit sample_cleanup_module(void)
	272	{
	273	tcf_unregister_action(&act_sample_ops, &sample_net_ops);
	274	}
	275
	276	module_init(sample_init_module);
	277	module_exit(sample_cleanup_module);
	278
f1fd20c3	279	MODULE_AUTHOR("Yotam Gigi <yotam.gi@gmail.com>");
5c5670fa YG	280	MODULE_DESCRIPTION("Packet sampling action");
5c5670fa YG	281	MODULE_LICENSE("GPL v2");