[people/arne_f/kernel.git] / net / sched / act_bpf.c

/*
 * Copyright (c) 2015 Jiri Pirko <jiri@resnulli.us>
 *
 * 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 <linux/filter.h>
#include <linux/bpf.h>

#include <net/netlink.h>
#include <net/pkt_sched.h>

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

#define BPF_TAB_MASK		15
#define ACT_BPF_NAME_LEN	256

struct tcf_bpf_cfg {
	struct bpf_prog *filter;
	struct sock_filter *bpf_ops;
	const char *bpf_name;
	u32 bpf_fd;
	u16 bpf_num_ops;
	bool is_ebpf;
};

static int tcf_bpf(struct sk_buff *skb, const struct tc_action *act,
		   struct tcf_result *res)
{
	struct tcf_bpf *prog = act->priv;
	int action, filter_res;
	bool at_ingress = G_TC_AT(skb->tc_verd) & AT_INGRESS;

	if (unlikely(!skb_mac_header_was_set(skb)))
		return TC_ACT_UNSPEC;

	spin_lock(&prog->tcf_lock);

	prog->tcf_tm.lastuse = jiffies;
	bstats_update(&prog->tcf_bstats, skb);

	/* Needed here for accessing maps. */
	rcu_read_lock();
	if (at_ingress) {
		__skb_push(skb, skb->mac_len);
		filter_res = BPF_PROG_RUN(prog->filter, skb);
		__skb_pull(skb, skb->mac_len);
	} else {
		filter_res = BPF_PROG_RUN(prog->filter, skb);
	}
	rcu_read_unlock();

	/* A BPF program may overwrite the default action opcode.
	 * Similarly as in cls_bpf, if filter_res == -1 we use the
	 * default action specified from tc.
	 *
	 * In case a different well-known TC_ACT opcode has been
	 * returned, it will overwrite the default one.
	 *
	 * For everything else that is unkown, TC_ACT_UNSPEC is
	 * returned.
	 */
	switch (filter_res) {
	case TC_ACT_PIPE:
	case TC_ACT_RECLASSIFY:
	case TC_ACT_OK:
		action = filter_res;
		break;
	case TC_ACT_SHOT:
		action = filter_res;
		prog->tcf_qstats.drops++;
		break;
	case TC_ACT_UNSPEC:
		action = prog->tcf_action;
		break;
	default:
		action = TC_ACT_UNSPEC;
		break;
	}

	spin_unlock(&prog->tcf_lock);
	return action;
}

static bool tcf_bpf_is_ebpf(const struct tcf_bpf *prog)
{
	return !prog->bpf_ops;
}

static int tcf_bpf_dump_bpf_info(const struct tcf_bpf *prog,
				 struct sk_buff *skb)
{
	struct nlattr *nla;

	if (nla_put_u16(skb, TCA_ACT_BPF_OPS_LEN, prog->bpf_num_ops))
		return -EMSGSIZE;

	nla = nla_reserve(skb, TCA_ACT_BPF_OPS, prog->bpf_num_ops *
			  sizeof(struct sock_filter));
	if (nla == NULL)
		return -EMSGSIZE;

	memcpy(nla_data(nla), prog->bpf_ops, nla_len(nla));

	return 0;
}

static int tcf_bpf_dump_ebpf_info(const struct tcf_bpf *prog,
				  struct sk_buff *skb)
{
	if (nla_put_u32(skb, TCA_ACT_BPF_FD, prog->bpf_fd))
		return -EMSGSIZE;

	if (prog->bpf_name &&
	    nla_put_string(skb, TCA_ACT_BPF_NAME, prog->bpf_name))
		return -EMSGSIZE;

	return 0;
}

static int tcf_bpf_dump(struct sk_buff *skb, struct tc_action *act,
			int bind, int ref)
{
	unsigned char *tp = skb_tail_pointer(skb);
	struct tcf_bpf *prog = act->priv;
	struct tc_act_bpf opt = {
		.index   = prog->tcf_index,
		.refcnt  = prog->tcf_refcnt - ref,
		.bindcnt = prog->tcf_bindcnt - bind,
		.action  = prog->tcf_action,
	};
	struct tcf_t tm;
	int ret;

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

	if (tcf_bpf_is_ebpf(prog))
		ret = tcf_bpf_dump_ebpf_info(prog, skb);
	else
		ret = tcf_bpf_dump_bpf_info(prog, skb);
	if (ret)
		goto nla_put_failure;

	tm.install = jiffies_to_clock_t(jiffies - prog->tcf_tm.install);
	tm.lastuse = jiffies_to_clock_t(jiffies - prog->tcf_tm.lastuse);
	tm.expires = jiffies_to_clock_t(prog->tcf_tm.expires);

	if (nla_put(skb, TCA_ACT_BPF_TM, sizeof(tm), &tm))
		goto nla_put_failure;

	return skb->len;

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

static const struct nla_policy act_bpf_policy[TCA_ACT_BPF_MAX + 1] = {
	[TCA_ACT_BPF_PARMS]	= { .len = sizeof(struct tc_act_bpf) },
	[TCA_ACT_BPF_FD]	= { .type = NLA_U32 },
	[TCA_ACT_BPF_NAME]	= { .type = NLA_NUL_STRING, .len = ACT_BPF_NAME_LEN },
	[TCA_ACT_BPF_OPS_LEN]	= { .type = NLA_U16 },
	[TCA_ACT_BPF_OPS]	= { .type = NLA_BINARY,
				    .len = sizeof(struct sock_filter) * BPF_MAXINSNS },
};

static int tcf_bpf_init_from_ops(struct nlattr **tb, struct tcf_bpf_cfg *cfg)
{
	struct sock_filter *bpf_ops;
	struct sock_fprog_kern fprog_tmp;
	struct bpf_prog *fp;
	u16 bpf_size, bpf_num_ops;
	int ret;

	bpf_num_ops = nla_get_u16(tb[TCA_ACT_BPF_OPS_LEN]);
	if (bpf_num_ops	> BPF_MAXINSNS || bpf_num_ops == 0)
		return -EINVAL;

	bpf_size = bpf_num_ops * sizeof(*bpf_ops);
	if (bpf_size != nla_len(tb[TCA_ACT_BPF_OPS]))
		return -EINVAL;

	bpf_ops = kzalloc(bpf_size, GFP_KERNEL);
	if (bpf_ops == NULL)
		return -ENOMEM;

	memcpy(bpf_ops, nla_data(tb[TCA_ACT_BPF_OPS]), bpf_size);

	fprog_tmp.len = bpf_num_ops;
	fprog_tmp.filter = bpf_ops;

	ret = bpf_prog_create(&fp, &fprog_tmp);
	if (ret < 0) {
		kfree(bpf_ops);
		return ret;
	}

	cfg->bpf_ops = bpf_ops;
	cfg->bpf_num_ops = bpf_num_ops;
	cfg->filter = fp;
	cfg->is_ebpf = false;

	return 0;
}

static int tcf_bpf_init_from_efd(struct nlattr **tb, struct tcf_bpf_cfg *cfg)
{
	struct bpf_prog *fp;
	char *name = NULL;
	u32 bpf_fd;

	bpf_fd = nla_get_u32(tb[TCA_ACT_BPF_FD]);

	fp = bpf_prog_get(bpf_fd);
	if (IS_ERR(fp))
		return PTR_ERR(fp);

	if (fp->type != BPF_PROG_TYPE_SCHED_ACT) {
		bpf_prog_put(fp);
		return -EINVAL;
	}

	if (tb[TCA_ACT_BPF_NAME]) {
		name = kmemdup(nla_data(tb[TCA_ACT_BPF_NAME]),
			       nla_len(tb[TCA_ACT_BPF_NAME]),
			       GFP_KERNEL);
		if (!name) {
			bpf_prog_put(fp);
			return -ENOMEM;
		}
	}

	cfg->bpf_fd = bpf_fd;
	cfg->bpf_name = name;
	cfg->filter = fp;
	cfg->is_ebpf = true;

	return 0;
}

static void tcf_bpf_cfg_cleanup(const struct tcf_bpf_cfg *cfg)
{
	if (cfg->is_ebpf)
		bpf_prog_put(cfg->filter);
	else
		bpf_prog_destroy(cfg->filter);

	kfree(cfg->bpf_ops);
	kfree(cfg->bpf_name);
}

static void tcf_bpf_prog_fill_cfg(const struct tcf_bpf *prog,
				  struct tcf_bpf_cfg *cfg)
{
	cfg->is_ebpf = tcf_bpf_is_ebpf(prog);
	cfg->filter = prog->filter;

	cfg->bpf_ops = prog->bpf_ops;
	cfg->bpf_name = prog->bpf_name;
}

static int tcf_bpf_init(struct net *net, struct nlattr *nla,
			struct nlattr *est, struct tc_action *act,
			int replace, int bind)
{
	struct nlattr *tb[TCA_ACT_BPF_MAX + 1];
	struct tcf_bpf_cfg cfg, old;
	struct tc_act_bpf *parm;
	struct tcf_bpf *prog;
	bool is_bpf, is_ebpf;
	int ret;

	if (!nla)
		return -EINVAL;

	ret = nla_parse_nested(tb, TCA_ACT_BPF_MAX, nla, act_bpf_policy);
	if (ret < 0)
		return ret;

	is_bpf = tb[TCA_ACT_BPF_OPS_LEN] && tb[TCA_ACT_BPF_OPS];
	is_ebpf = tb[TCA_ACT_BPF_FD];

	if ((!is_bpf && !is_ebpf) || (is_bpf && is_ebpf) ||
	    !tb[TCA_ACT_BPF_PARMS])
		return -EINVAL;

	parm = nla_data(tb[TCA_ACT_BPF_PARMS]);

	memset(&cfg, 0, sizeof(cfg));

	ret = is_bpf ? tcf_bpf_init_from_ops(tb, &cfg) :
		       tcf_bpf_init_from_efd(tb, &cfg);
	if (ret < 0)
		return ret;

	if (!tcf_hash_check(parm->index, act, bind)) {
		ret = tcf_hash_create(parm->index, est, act,
				      sizeof(*prog), bind, false);
		if (ret < 0)
			goto destroy_fp;

		ret = ACT_P_CREATED;
	} else {
		/* Don't override defaults. */
		if (bind)
			goto destroy_fp;

		tcf_hash_release(act, bind);
		if (!replace) {
			ret = -EEXIST;
			goto destroy_fp;
		}
	}

	prog = to_bpf(act);
	spin_lock_bh(&prog->tcf_lock);

	if (ret != ACT_P_CREATED)
		tcf_bpf_prog_fill_cfg(prog, &old);

	prog->bpf_ops = cfg.bpf_ops;
	prog->bpf_name = cfg.bpf_name;

	if (cfg.bpf_num_ops)
		prog->bpf_num_ops = cfg.bpf_num_ops;
	if (cfg.bpf_fd)
		prog->bpf_fd = cfg.bpf_fd;

	prog->tcf_action = parm->action;
	prog->filter = cfg.filter;

	spin_unlock_bh(&prog->tcf_lock);

	if (ret == ACT_P_CREATED)
		tcf_hash_insert(act);
	else
		tcf_bpf_cfg_cleanup(&old);

	return ret;

destroy_fp:
	tcf_bpf_cfg_cleanup(&cfg);
	return ret;
}

static void tcf_bpf_cleanup(struct tc_action *act, int bind)
{
	struct tcf_bpf_cfg tmp;

	tcf_bpf_prog_fill_cfg(act->priv, &tmp);
	tcf_bpf_cfg_cleanup(&tmp);
}

static struct tc_action_ops act_bpf_ops __read_mostly = {
	.kind		=	"bpf",
	.type		=	TCA_ACT_BPF,
	.owner		=	THIS_MODULE,
	.act		=	tcf_bpf,
	.dump		=	tcf_bpf_dump,
	.cleanup	=	tcf_bpf_cleanup,
	.init		=	tcf_bpf_init,
};

static int __init bpf_init_module(void)
{
	return tcf_register_action(&act_bpf_ops, BPF_TAB_MASK);
}

static void __exit bpf_cleanup_module(void)
{
	tcf_unregister_action(&act_bpf_ops);
}

module_init(bpf_init_module);
module_exit(bpf_cleanup_module);

MODULE_AUTHOR("Jiri Pirko <jiri@resnulli.us>");
MODULE_DESCRIPTION("TC BPF based action");
MODULE_LICENSE("GPL v2");
Commit	Line	Data
d23b8ad8 JP	1	/*
	2	* Copyright (c) 2015 Jiri Pirko <jiri@resnulli.us>
	3	*
	4	* This program is free software; you can redistribute it and/or modify
	5	* it under the terms of the GNU General Public License as published by
	6	* the Free Software Foundation; either version 2 of the License, or
	7	* (at your option) any later version.
	8	*/
	9
	10	#include <linux/module.h>
	11	#include <linux/init.h>
	12	#include <linux/kernel.h>
	13	#include <linux/skbuff.h>
	14	#include <linux/rtnetlink.h>
	15	#include <linux/filter.h>
a8cb5f55 DB	16	#include <linux/bpf.h>
a8cb5f55 DB	17
d23b8ad8 JP	18	#include <net/netlink.h>
	19	#include <net/pkt_sched.h>
	20
	21	#include <linux/tc_act/tc_bpf.h>
	22	#include <net/tc_act/tc_bpf.h>
	23
a8cb5f55 DB	24	#define BPF_TAB_MASK 15
	25	#define ACT_BPF_NAME_LEN 256
	26
	27	struct tcf_bpf_cfg {
	28	struct bpf_prog *filter;
	29	struct sock_filter *bpf_ops;
f4eaed28	30	const char *bpf_name;
a8cb5f55 DB	31	u32 bpf_fd;
a8cb5f55 DB	32	u16 bpf_num_ops;
f4eaed28	33	bool is_ebpf;
a8cb5f55	34	};
d23b8ad8	35
a8cb5f55	36	static int tcf_bpf(struct sk_buff skb, const struct tc_action act,
d23b8ad8 JP	37	struct tcf_result *res)
d23b8ad8 JP	38	{
a8cb5f55	39	struct tcf_bpf *prog = act->priv;
ced585c8	40	int action, filter_res;
3431205e	41	bool at_ingress = G_TC_AT(skb->tc_verd) & AT_INGRESS;
d23b8ad8	42
a166151c AS	43	if (unlikely(!skb_mac_header_was_set(skb)))
	44	return TC_ACT_UNSPEC;
	45
a8cb5f55	46	spin_lock(&prog->tcf_lock);
ced585c8	47
a8cb5f55 DB	48	prog->tcf_tm.lastuse = jiffies;
a8cb5f55 DB	49	bstats_update(&prog->tcf_bstats, skb);
d23b8ad8	50
a8cb5f55 DB	51	/* Needed here for accessing maps. */
a8cb5f55 DB	52	rcu_read_lock();
3431205e AS	53	if (at_ingress) {
	54	__skb_push(skb, skb->mac_len);
	55	filter_res = BPF_PROG_RUN(prog->filter, skb);
	56	__skb_pull(skb, skb->mac_len);
	57	} else {
	58	filter_res = BPF_PROG_RUN(prog->filter, skb);
	59	}
a8cb5f55	60	rcu_read_unlock();
ced585c8 DB	61
	62	/* A BPF program may overwrite the default action opcode.
	63	* Similarly as in cls_bpf, if filter_res == -1 we use the
	64	* default action specified from tc.
	65	*
	66	* In case a different well-known TC_ACT opcode has been
	67	* returned, it will overwrite the default one.
	68	*
	69	* For everything else that is unkown, TC_ACT_UNSPEC is
	70	* returned.
	71	*/
	72	switch (filter_res) {
	73	case TC_ACT_PIPE:
	74	case TC_ACT_RECLASSIFY:
	75	case TC_ACT_OK:
	76	action = filter_res;
	77	break;
	78	case TC_ACT_SHOT:
	79	action = filter_res;
a8cb5f55	80	prog->tcf_qstats.drops++;
ced585c8 DB	81	break;
ced585c8 DB	82	case TC_ACT_UNSPEC:
a8cb5f55	83	action = prog->tcf_action;
ced585c8 DB	84	break;
	85	default:
	86	action = TC_ACT_UNSPEC;
	87	break;
d23b8ad8 JP	88	}
d23b8ad8 JP	89
a8cb5f55	90	spin_unlock(&prog->tcf_lock);
d23b8ad8 JP	91	return action;
	92	}
	93
a8cb5f55 DB	94	static bool tcf_bpf_is_ebpf(const struct tcf_bpf *prog)
	95	{
	96	return !prog->bpf_ops;
	97	}
	98
	99	static int tcf_bpf_dump_bpf_info(const struct tcf_bpf *prog,
	100	struct sk_buff *skb)
	101	{
	102	struct nlattr *nla;
	103
	104	if (nla_put_u16(skb, TCA_ACT_BPF_OPS_LEN, prog->bpf_num_ops))
	105	return -EMSGSIZE;
	106
	107	nla = nla_reserve(skb, TCA_ACT_BPF_OPS, prog->bpf_num_ops *
	108	sizeof(struct sock_filter));
	109	if (nla == NULL)
	110	return -EMSGSIZE;
	111
	112	memcpy(nla_data(nla), prog->bpf_ops, nla_len(nla));
	113
	114	return 0;
	115	}
	116
	117	static int tcf_bpf_dump_ebpf_info(const struct tcf_bpf *prog,
	118	struct sk_buff *skb)
	119	{
	120	if (nla_put_u32(skb, TCA_ACT_BPF_FD, prog->bpf_fd))
	121	return -EMSGSIZE;
	122
	123	if (prog->bpf_name &&
	124	nla_put_string(skb, TCA_ACT_BPF_NAME, prog->bpf_name))
	125	return -EMSGSIZE;
	126
	127	return 0;
	128	}
	129
	130	static int tcf_bpf_dump(struct sk_buff skb, struct tc_action act,
d23b8ad8 JP	131	int bind, int ref)
	132	{
	133	unsigned char *tp = skb_tail_pointer(skb);
a8cb5f55	134	struct tcf_bpf *prog = act->priv;
d23b8ad8	135	struct tc_act_bpf opt = {
a8cb5f55 DB	136	.index = prog->tcf_index,
	137	.refcnt = prog->tcf_refcnt - ref,
	138	.bindcnt = prog->tcf_bindcnt - bind,
	139	.action = prog->tcf_action,
d23b8ad8	140	};
a8cb5f55 DB	141	struct tcf_t tm;
a8cb5f55 DB	142	int ret;
d23b8ad8 JP	143
	144	if (nla_put(skb, TCA_ACT_BPF_PARMS, sizeof(opt), &opt))
	145	goto nla_put_failure;
	146
a8cb5f55 DB	147	if (tcf_bpf_is_ebpf(prog))
	148	ret = tcf_bpf_dump_ebpf_info(prog, skb);
	149	else
	150	ret = tcf_bpf_dump_bpf_info(prog, skb);
	151	if (ret)
d23b8ad8 JP	152	goto nla_put_failure;
d23b8ad8 JP	153
a8cb5f55 DB	154	tm.install = jiffies_to_clock_t(jiffies - prog->tcf_tm.install);
	155	tm.lastuse = jiffies_to_clock_t(jiffies - prog->tcf_tm.lastuse);
	156	tm.expires = jiffies_to_clock_t(prog->tcf_tm.expires);
d23b8ad8	157
a8cb5f55	158	if (nla_put(skb, TCA_ACT_BPF_TM, sizeof(tm), &tm))
d23b8ad8	159	goto nla_put_failure;
a8cb5f55	160
d23b8ad8 JP	161	return skb->len;
	162
	163	nla_put_failure:
	164	nlmsg_trim(skb, tp);
	165	return -1;
	166	}
	167
	168	static const struct nla_policy act_bpf_policy[TCA_ACT_BPF_MAX + 1] = {
	169	[TCA_ACT_BPF_PARMS] = { .len = sizeof(struct tc_act_bpf) },
a8cb5f55 DB	170	[TCA_ACT_BPF_FD] = { .type = NLA_U32 },
a8cb5f55 DB	171	[TCA_ACT_BPF_NAME] = { .type = NLA_NUL_STRING, .len = ACT_BPF_NAME_LEN },
d23b8ad8 JP	172	[TCA_ACT_BPF_OPS_LEN] = { .type = NLA_U16 },
	173	[TCA_ACT_BPF_OPS] = { .type = NLA_BINARY,
	174	.len = sizeof(struct sock_filter) * BPF_MAXINSNS },
	175	};
	176
a8cb5f55	177	static int tcf_bpf_init_from_ops(struct nlattr *tb, struct tcf_bpf_cfg cfg)
d23b8ad8	178	{
d23b8ad8	179	struct sock_filter *bpf_ops;
a8cb5f55	180	struct sock_fprog_kern fprog_tmp;
d23b8ad8	181	struct bpf_prog *fp;
a8cb5f55	182	u16 bpf_size, bpf_num_ops;
d23b8ad8 JP	183	int ret;
d23b8ad8 JP	184
d23b8ad8 JP	185	bpf_num_ops = nla_get_u16(tb[TCA_ACT_BPF_OPS_LEN]);
	186	if (bpf_num_ops > BPF_MAXINSNS \|\| bpf_num_ops == 0)
	187	return -EINVAL;
	188
	189	bpf_size = bpf_num_ops * sizeof(*bpf_ops);
fd3e646c DB	190	if (bpf_size != nla_len(tb[TCA_ACT_BPF_OPS]))
	191	return -EINVAL;
	192
d23b8ad8	193	bpf_ops = kzalloc(bpf_size, GFP_KERNEL);
a8cb5f55	194	if (bpf_ops == NULL)
d23b8ad8 JP	195	return -ENOMEM;
	196
	197	memcpy(bpf_ops, nla_data(tb[TCA_ACT_BPF_OPS]), bpf_size);
	198
a8cb5f55 DB	199	fprog_tmp.len = bpf_num_ops;
a8cb5f55 DB	200	fprog_tmp.filter = bpf_ops;
d23b8ad8	201
a8cb5f55 DB	202	ret = bpf_prog_create(&fp, &fprog_tmp);
	203	if (ret < 0) {
	204	kfree(bpf_ops);
	205	return ret;
	206	}
d23b8ad8	207
a8cb5f55 DB	208	cfg->bpf_ops = bpf_ops;
	209	cfg->bpf_num_ops = bpf_num_ops;
	210	cfg->filter = fp;
f4eaed28	211	cfg->is_ebpf = false;
a8cb5f55 DB	212
	213	return 0;
	214	}
	215
	216	static int tcf_bpf_init_from_efd(struct nlattr *tb, struct tcf_bpf_cfg cfg)
	217	{
	218	struct bpf_prog *fp;
	219	char *name = NULL;
	220	u32 bpf_fd;
	221
	222	bpf_fd = nla_get_u32(tb[TCA_ACT_BPF_FD]);
	223
	224	fp = bpf_prog_get(bpf_fd);
	225	if (IS_ERR(fp))
	226	return PTR_ERR(fp);
	227
	228	if (fp->type != BPF_PROG_TYPE_SCHED_ACT) {
	229	bpf_prog_put(fp);
	230	return -EINVAL;
	231	}
	232
	233	if (tb[TCA_ACT_BPF_NAME]) {
	234	name = kmemdup(nla_data(tb[TCA_ACT_BPF_NAME]),
	235	nla_len(tb[TCA_ACT_BPF_NAME]),
	236	GFP_KERNEL);
	237	if (!name) {
	238	bpf_prog_put(fp);
	239	return -ENOMEM;
	240	}
	241	}
	242
	243	cfg->bpf_fd = bpf_fd;
	244	cfg->bpf_name = name;
	245	cfg->filter = fp;
f4eaed28	246	cfg->is_ebpf = true;
a8cb5f55 DB	247
	248	return 0;
	249	}
	250
f4eaed28 DB	251	static void tcf_bpf_cfg_cleanup(const struct tcf_bpf_cfg *cfg)
	252	{
	253	if (cfg->is_ebpf)
	254	bpf_prog_put(cfg->filter);
	255	else
	256	bpf_prog_destroy(cfg->filter);
	257
	258	kfree(cfg->bpf_ops);
	259	kfree(cfg->bpf_name);
	260	}
	261
	262	static void tcf_bpf_prog_fill_cfg(const struct tcf_bpf *prog,
	263	struct tcf_bpf_cfg *cfg)
	264	{
	265	cfg->is_ebpf = tcf_bpf_is_ebpf(prog);
	266	cfg->filter = prog->filter;
	267
	268	cfg->bpf_ops = prog->bpf_ops;
	269	cfg->bpf_name = prog->bpf_name;
	270	}
	271
a8cb5f55 DB	272	static int tcf_bpf_init(struct net net, struct nlattr nla,
	273	struct nlattr est, struct tc_action act,
	274	int replace, int bind)
	275	{
	276	struct nlattr *tb[TCA_ACT_BPF_MAX + 1];
f4eaed28	277	struct tcf_bpf_cfg cfg, old;
a8cb5f55 DB	278	struct tc_act_bpf *parm;
a8cb5f55 DB	279	struct tcf_bpf *prog;
a8cb5f55 DB	280	bool is_bpf, is_ebpf;
	281	int ret;
	282
	283	if (!nla)
	284	return -EINVAL;
	285
	286	ret = nla_parse_nested(tb, TCA_ACT_BPF_MAX, nla, act_bpf_policy);
	287	if (ret < 0)
	288	return ret;
	289
	290	is_bpf = tb[TCA_ACT_BPF_OPS_LEN] && tb[TCA_ACT_BPF_OPS];
	291	is_ebpf = tb[TCA_ACT_BPF_FD];
	292
	293	if ((!is_bpf && !is_ebpf) \|\| (is_bpf && is_ebpf) \|\|
	294	!tb[TCA_ACT_BPF_PARMS])
	295	return -EINVAL;
	296
	297	parm = nla_data(tb[TCA_ACT_BPF_PARMS]);
	298
	299	memset(&cfg, 0, sizeof(cfg));
	300
	301	ret = is_bpf ? tcf_bpf_init_from_ops(tb, &cfg) :
	302	tcf_bpf_init_from_efd(tb, &cfg);
	303	if (ret < 0)
	304	return ret;
	305
	306	if (!tcf_hash_check(parm->index, act, bind)) {
	307	ret = tcf_hash_create(parm->index, est, act,
519c818e	308	sizeof(*prog), bind, false);
a8cb5f55	309	if (ret < 0)
d23b8ad8 JP	310	goto destroy_fp;
	311
	312	ret = ACT_P_CREATED;
	313	} else {
a8cb5f55	314	/* Don't override defaults. */
d23b8ad8 JP	315	if (bind)
d23b8ad8 JP	316	goto destroy_fp;
a8cb5f55 DB	317
	318	tcf_hash_release(act, bind);
	319	if (!replace) {
d23b8ad8 JP	320	ret = -EEXIST;
	321	goto destroy_fp;
	322	}
	323	}
	324
a8cb5f55 DB	325	prog = to_bpf(act);
	326	spin_lock_bh(&prog->tcf_lock);
	327
f4eaed28 DB	328	if (ret != ACT_P_CREATED)
	329	tcf_bpf_prog_fill_cfg(prog, &old);
	330
a8cb5f55 DB	331	prog->bpf_ops = cfg.bpf_ops;
	332	prog->bpf_name = cfg.bpf_name;
	333
	334	if (cfg.bpf_num_ops)
	335	prog->bpf_num_ops = cfg.bpf_num_ops;
	336	if (cfg.bpf_fd)
	337	prog->bpf_fd = cfg.bpf_fd;
	338
	339	prog->tcf_action = parm->action;
	340	prog->filter = cfg.filter;
	341
	342	spin_unlock_bh(&prog->tcf_lock);
d23b8ad8 JP	343
d23b8ad8 JP	344	if (ret == ACT_P_CREATED)
a8cb5f55	345	tcf_hash_insert(act);
f4eaed28 DB	346	else
f4eaed28 DB	347	tcf_bpf_cfg_cleanup(&old);
a8cb5f55	348
d23b8ad8 JP	349	return ret;
	350
	351	destroy_fp:
f4eaed28	352	tcf_bpf_cfg_cleanup(&cfg);
d23b8ad8 JP	353	return ret;
	354	}
	355
a8cb5f55	356	static void tcf_bpf_cleanup(struct tc_action *act, int bind)
d23b8ad8	357	{
f4eaed28	358	struct tcf_bpf_cfg tmp;
ddf06c1e	359
f4eaed28 DB	360	tcf_bpf_prog_fill_cfg(act->priv, &tmp);
f4eaed28 DB	361	tcf_bpf_cfg_cleanup(&tmp);
d23b8ad8 JP	362	}
d23b8ad8 JP	363
a8cb5f55 DB	364	static struct tc_action_ops act_bpf_ops __read_mostly = {
	365	.kind = "bpf",
	366	.type = TCA_ACT_BPF,
	367	.owner = THIS_MODULE,
	368	.act = tcf_bpf,
	369	.dump = tcf_bpf_dump,
	370	.cleanup = tcf_bpf_cleanup,
	371	.init = tcf_bpf_init,
d23b8ad8 JP	372	};
	373
	374	static int __init bpf_init_module(void)
	375	{
	376	return tcf_register_action(&act_bpf_ops, BPF_TAB_MASK);
	377	}
	378
	379	static void __exit bpf_cleanup_module(void)
	380	{
	381	tcf_unregister_action(&act_bpf_ops);
	382	}
	383
	384	module_init(bpf_init_module);
	385	module_exit(bpf_cleanup_module);
	386
	387	MODULE_AUTHOR("Jiri Pirko <jiri@resnulli.us>");
	388	MODULE_DESCRIPTION("TC BPF based action");
	389	MODULE_LICENSE("GPL v2");