[linux-block.git] / net / sched / act_skbedit.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 * Copyright (c) 2008, Intel Corporation.
 *
 * Author: Alexander Duyck <alexander.h.duyck@intel.com>
 */

#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 <net/ip.h>
#include <net/ipv6.h>
#include <net/dsfield.h>
#include <net/pkt_cls.h>
#include <net/tc_wrapper.h>

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

static struct tc_action_ops act_skbedit_ops;

static u16 tcf_skbedit_hash(struct tcf_skbedit_params *params,
			    struct sk_buff *skb)
{
	u16 queue_mapping = params->queue_mapping;

	if (params->flags & SKBEDIT_F_TXQ_SKBHASH) {
		u32 hash = skb_get_hash(skb);

		queue_mapping += hash % params->mapping_mod;
	}

	return netdev_cap_txqueue(skb->dev, queue_mapping);
}

TC_INDIRECT_SCOPE int tcf_skbedit_act(struct sk_buff *skb,
				      const struct tc_action *a,
				      struct tcf_result *res)
{
	struct tcf_skbedit *d = to_skbedit(a);
	struct tcf_skbedit_params *params;
	int action;

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

	params = rcu_dereference_bh(d->params);
	action = READ_ONCE(d->tcf_action);

	if (params->flags & SKBEDIT_F_PRIORITY)
		skb->priority = params->priority;
	if (params->flags & SKBEDIT_F_INHERITDSFIELD) {
		int wlen = skb_network_offset(skb);

		switch (skb_protocol(skb, true)) {
		case htons(ETH_P_IP):
			wlen += sizeof(struct iphdr);
			if (!pskb_may_pull(skb, wlen))
				goto err;
			skb->priority = ipv4_get_dsfield(ip_hdr(skb)) >> 2;
			break;

		case htons(ETH_P_IPV6):
			wlen += sizeof(struct ipv6hdr);
			if (!pskb_may_pull(skb, wlen))
				goto err;
			skb->priority = ipv6_get_dsfield(ipv6_hdr(skb)) >> 2;
			break;
		}
	}
	if (params->flags & SKBEDIT_F_QUEUE_MAPPING &&
	    skb->dev->real_num_tx_queues > params->queue_mapping) {
#ifdef CONFIG_NET_EGRESS
		netdev_xmit_skip_txqueue(true);
#endif
		skb_set_queue_mapping(skb, tcf_skbedit_hash(params, skb));
	}
	if (params->flags & SKBEDIT_F_MARK) {
		skb->mark &= ~params->mask;
		skb->mark |= params->mark & params->mask;
	}
	if (params->flags & SKBEDIT_F_PTYPE)
		skb->pkt_type = params->ptype;
	return action;

err:
	qstats_drop_inc(this_cpu_ptr(d->common.cpu_qstats));
	return TC_ACT_SHOT;
}

static void tcf_skbedit_stats_update(struct tc_action *a, u64 bytes,
				     u64 packets, u64 drops,
				     u64 lastuse, bool hw)
{
	struct tcf_skbedit *d = to_skbedit(a);
	struct tcf_t *tm = &d->tcf_tm;

	tcf_action_update_stats(a, bytes, packets, drops, hw);
	tm->lastuse = max_t(u64, tm->lastuse, lastuse);
}

static const struct nla_policy skbedit_policy[TCA_SKBEDIT_MAX + 1] = {
	[TCA_SKBEDIT_PARMS]		= { .len = sizeof(struct tc_skbedit) },
	[TCA_SKBEDIT_PRIORITY]		= { .len = sizeof(u32) },
	[TCA_SKBEDIT_QUEUE_MAPPING]	= { .len = sizeof(u16) },
	[TCA_SKBEDIT_MARK]		= { .len = sizeof(u32) },
	[TCA_SKBEDIT_PTYPE]		= { .len = sizeof(u16) },
	[TCA_SKBEDIT_MASK]		= { .len = sizeof(u32) },
	[TCA_SKBEDIT_FLAGS]		= { .len = sizeof(u64) },
	[TCA_SKBEDIT_QUEUE_MAPPING_MAX]	= { .len = sizeof(u16) },
};

static int tcf_skbedit_init(struct net *net, struct nlattr *nla,
			    struct nlattr *est, struct tc_action **a,
			    struct tcf_proto *tp, u32 act_flags,
			    struct netlink_ext_ack *extack)
{
	struct tc_action_net *tn = net_generic(net, act_skbedit_ops.net_id);
	bool bind = act_flags & TCA_ACT_FLAGS_BIND;
	struct tcf_skbedit_params *params_new;
	struct nlattr *tb[TCA_SKBEDIT_MAX + 1];
	struct tcf_chain *goto_ch = NULL;
	struct tc_skbedit *parm;
	struct tcf_skbedit *d;
	u32 flags = 0, *priority = NULL, *mark = NULL, *mask = NULL;
	u16 *queue_mapping = NULL, *ptype = NULL;
	u16 mapping_mod = 1;
	bool exists = false;
	int ret = 0, err;
	u32 index;

	if (nla == NULL)
		return -EINVAL;

	err = nla_parse_nested_deprecated(tb, TCA_SKBEDIT_MAX, nla,
					  skbedit_policy, NULL);
	if (err < 0)
		return err;

	if (tb[TCA_SKBEDIT_PARMS] == NULL)
		return -EINVAL;

	if (tb[TCA_SKBEDIT_PRIORITY] != NULL) {
		flags |= SKBEDIT_F_PRIORITY;
		priority = nla_data(tb[TCA_SKBEDIT_PRIORITY]);
	}

	if (tb[TCA_SKBEDIT_QUEUE_MAPPING] != NULL) {
		if (is_tcf_skbedit_ingress(act_flags) &&
		    !(act_flags & TCA_ACT_FLAGS_SKIP_SW)) {
			NL_SET_ERR_MSG_MOD(extack, "\"queue_mapping\" option on receive side is hardware only, use skip_sw");
			return -EOPNOTSUPP;
		}
		flags |= SKBEDIT_F_QUEUE_MAPPING;
		queue_mapping = nla_data(tb[TCA_SKBEDIT_QUEUE_MAPPING]);
	}

	if (tb[TCA_SKBEDIT_PTYPE] != NULL) {
		ptype = nla_data(tb[TCA_SKBEDIT_PTYPE]);
		if (!skb_pkt_type_ok(*ptype))
			return -EINVAL;
		flags |= SKBEDIT_F_PTYPE;
	}

	if (tb[TCA_SKBEDIT_MARK] != NULL) {
		flags |= SKBEDIT_F_MARK;
		mark = nla_data(tb[TCA_SKBEDIT_MARK]);
	}

	if (tb[TCA_SKBEDIT_MASK] != NULL) {
		flags |= SKBEDIT_F_MASK;
		mask = nla_data(tb[TCA_SKBEDIT_MASK]);
	}

	if (tb[TCA_SKBEDIT_FLAGS] != NULL) {
		u64 *pure_flags = nla_data(tb[TCA_SKBEDIT_FLAGS]);

		if (*pure_flags & SKBEDIT_F_TXQ_SKBHASH) {
			u16 *queue_mapping_max;

			if (!tb[TCA_SKBEDIT_QUEUE_MAPPING] ||
			    !tb[TCA_SKBEDIT_QUEUE_MAPPING_MAX]) {
				NL_SET_ERR_MSG_MOD(extack, "Missing required range of queue_mapping.");
				return -EINVAL;
			}

			queue_mapping_max =
				nla_data(tb[TCA_SKBEDIT_QUEUE_MAPPING_MAX]);
			if (*queue_mapping_max < *queue_mapping) {
				NL_SET_ERR_MSG_MOD(extack, "The range of queue_mapping is invalid, max < min.");
				return -EINVAL;
			}

			mapping_mod = *queue_mapping_max - *queue_mapping + 1;
			flags |= SKBEDIT_F_TXQ_SKBHASH;
		}
		if (*pure_flags & SKBEDIT_F_INHERITDSFIELD)
			flags |= SKBEDIT_F_INHERITDSFIELD;
	}

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

	if (!flags) {
		if (exists)
			tcf_idr_release(*a, bind);
		else
			tcf_idr_cleanup(tn, index);
		return -EINVAL;
	}

	if (!exists) {
		ret = tcf_idr_create(tn, index, est, a,
				     &act_skbedit_ops, bind, true, act_flags);
		if (ret) {
			tcf_idr_cleanup(tn, index);
			return ret;
		}

		d = to_skbedit(*a);
		ret = ACT_P_CREATED;
	} else {
		d = to_skbedit(*a);
		if (!(act_flags & TCA_ACT_FLAGS_REPLACE)) {
			tcf_idr_release(*a, bind);
			return -EEXIST;
		}
	}
	err = tcf_action_check_ctrlact(parm->action, tp, &goto_ch, extack);
	if (err < 0)
		goto release_idr;

	params_new = kzalloc(sizeof(*params_new), GFP_KERNEL);
	if (unlikely(!params_new)) {
		err = -ENOMEM;
		goto put_chain;
	}

	params_new->flags = flags;
	if (flags & SKBEDIT_F_PRIORITY)
		params_new->priority = *priority;
	if (flags & SKBEDIT_F_QUEUE_MAPPING) {
		params_new->queue_mapping = *queue_mapping;
		params_new->mapping_mod = mapping_mod;
	}
	if (flags & SKBEDIT_F_MARK)
		params_new->mark = *mark;
	if (flags & SKBEDIT_F_PTYPE)
		params_new->ptype = *ptype;
	/* default behaviour is to use all the bits */
	params_new->mask = 0xffffffff;
	if (flags & SKBEDIT_F_MASK)
		params_new->mask = *mask;

	spin_lock_bh(&d->tcf_lock);
	goto_ch = tcf_action_set_ctrlact(*a, parm->action, goto_ch);
	params_new = rcu_replace_pointer(d->params, params_new,
					 lockdep_is_held(&d->tcf_lock));
	spin_unlock_bh(&d->tcf_lock);
	if (params_new)
		kfree_rcu(params_new, rcu);
	if (goto_ch)
		tcf_chain_put_by_act(goto_ch);

	return ret;
put_chain:
	if (goto_ch)
		tcf_chain_put_by_act(goto_ch);
release_idr:
	tcf_idr_release(*a, bind);
	return err;
}

static int tcf_skbedit_dump(struct sk_buff *skb, struct tc_action *a,
			    int bind, int ref)
{
	unsigned char *b = skb_tail_pointer(skb);
	struct tcf_skbedit *d = to_skbedit(a);
	struct tcf_skbedit_params *params;
	struct tc_skbedit opt = {
		.index   = d->tcf_index,
		.refcnt  = refcount_read(&d->tcf_refcnt) - ref,
		.bindcnt = atomic_read(&d->tcf_bindcnt) - bind,
	};
	u64 pure_flags = 0;
	struct tcf_t t;

	spin_lock_bh(&d->tcf_lock);
	params = rcu_dereference_protected(d->params,
					   lockdep_is_held(&d->tcf_lock));
	opt.action = d->tcf_action;

	if (nla_put(skb, TCA_SKBEDIT_PARMS, sizeof(opt), &opt))
		goto nla_put_failure;
	if ((params->flags & SKBEDIT_F_PRIORITY) &&
	    nla_put_u32(skb, TCA_SKBEDIT_PRIORITY, params->priority))
		goto nla_put_failure;
	if ((params->flags & SKBEDIT_F_QUEUE_MAPPING) &&
	    nla_put_u16(skb, TCA_SKBEDIT_QUEUE_MAPPING, params->queue_mapping))
		goto nla_put_failure;
	if ((params->flags & SKBEDIT_F_MARK) &&
	    nla_put_u32(skb, TCA_SKBEDIT_MARK, params->mark))
		goto nla_put_failure;
	if ((params->flags & SKBEDIT_F_PTYPE) &&
	    nla_put_u16(skb, TCA_SKBEDIT_PTYPE, params->ptype))
		goto nla_put_failure;
	if ((params->flags & SKBEDIT_F_MASK) &&
	    nla_put_u32(skb, TCA_SKBEDIT_MASK, params->mask))
		goto nla_put_failure;
	if (params->flags & SKBEDIT_F_INHERITDSFIELD)
		pure_flags |= SKBEDIT_F_INHERITDSFIELD;
	if (params->flags & SKBEDIT_F_TXQ_SKBHASH) {
		if (nla_put_u16(skb, TCA_SKBEDIT_QUEUE_MAPPING_MAX,
				params->queue_mapping + params->mapping_mod - 1))
			goto nla_put_failure;

		pure_flags |= SKBEDIT_F_TXQ_SKBHASH;
	}
	if (pure_flags != 0 &&
	    nla_put(skb, TCA_SKBEDIT_FLAGS, sizeof(pure_flags), &pure_flags))
		goto nla_put_failure;

	tcf_tm_dump(&t, &d->tcf_tm);
	if (nla_put_64bit(skb, TCA_SKBEDIT_TM, sizeof(t), &t, TCA_SKBEDIT_PAD))
		goto nla_put_failure;
	spin_unlock_bh(&d->tcf_lock);

	return skb->len;

nla_put_failure:
	spin_unlock_bh(&d->tcf_lock);
	nlmsg_trim(skb, b);
	return -1;
}

static void tcf_skbedit_cleanup(struct tc_action *a)
{
	struct tcf_skbedit *d = to_skbedit(a);
	struct tcf_skbedit_params *params;

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

static size_t tcf_skbedit_get_fill_size(const struct tc_action *act)
{
	return nla_total_size(sizeof(struct tc_skbedit))
		+ nla_total_size(sizeof(u32)) /* TCA_SKBEDIT_PRIORITY */
		+ nla_total_size(sizeof(u16)) /* TCA_SKBEDIT_QUEUE_MAPPING */
		+ nla_total_size(sizeof(u16)) /* TCA_SKBEDIT_QUEUE_MAPPING_MAX */
		+ nla_total_size(sizeof(u32)) /* TCA_SKBEDIT_MARK */
		+ nla_total_size(sizeof(u16)) /* TCA_SKBEDIT_PTYPE */
		+ nla_total_size(sizeof(u32)) /* TCA_SKBEDIT_MASK */
		+ nla_total_size_64bit(sizeof(u64)); /* TCA_SKBEDIT_FLAGS */
}

static int tcf_skbedit_offload_act_setup(struct tc_action *act, void *entry_data,
					 u32 *index_inc, bool bind,
					 struct netlink_ext_ack *extack)
{
	if (bind) {
		struct flow_action_entry *entry = entry_data;

		if (is_tcf_skbedit_mark(act)) {
			entry->id = FLOW_ACTION_MARK;
			entry->mark = tcf_skbedit_mark(act);
		} else if (is_tcf_skbedit_ptype(act)) {
			entry->id = FLOW_ACTION_PTYPE;
			entry->ptype = tcf_skbedit_ptype(act);
		} else if (is_tcf_skbedit_priority(act)) {
			entry->id = FLOW_ACTION_PRIORITY;
			entry->priority = tcf_skbedit_priority(act);
		} else if (is_tcf_skbedit_tx_queue_mapping(act)) {
			NL_SET_ERR_MSG_MOD(extack, "Offload not supported when \"queue_mapping\" option is used on transmit side");
			return -EOPNOTSUPP;
		} else if (is_tcf_skbedit_rx_queue_mapping(act)) {
			entry->id = FLOW_ACTION_RX_QUEUE_MAPPING;
			entry->rx_queue = tcf_skbedit_rx_queue_mapping(act);
		} else if (is_tcf_skbedit_inheritdsfield(act)) {
			NL_SET_ERR_MSG_MOD(extack, "Offload not supported when \"inheritdsfield\" option is used");
			return -EOPNOTSUPP;
		} else {
			NL_SET_ERR_MSG_MOD(extack, "Unsupported skbedit option offload");
			return -EOPNOTSUPP;
		}
		*index_inc = 1;
	} else {
		struct flow_offload_action *fl_action = entry_data;

		if (is_tcf_skbedit_mark(act))
			fl_action->id = FLOW_ACTION_MARK;
		else if (is_tcf_skbedit_ptype(act))
			fl_action->id = FLOW_ACTION_PTYPE;
		else if (is_tcf_skbedit_priority(act))
			fl_action->id = FLOW_ACTION_PRIORITY;
		else if (is_tcf_skbedit_rx_queue_mapping(act))
			fl_action->id = FLOW_ACTION_RX_QUEUE_MAPPING;
		else
			return -EOPNOTSUPP;
	}

	return 0;
}

static struct tc_action_ops act_skbedit_ops = {
	.kind		=	"skbedit",
	.id		=	TCA_ID_SKBEDIT,
	.owner		=	THIS_MODULE,
	.act		=	tcf_skbedit_act,
	.stats_update	=	tcf_skbedit_stats_update,
	.dump		=	tcf_skbedit_dump,
	.init		=	tcf_skbedit_init,
	.cleanup	=	tcf_skbedit_cleanup,
	.get_fill_size	=	tcf_skbedit_get_fill_size,
	.offload_act_setup =	tcf_skbedit_offload_act_setup,
	.size		=	sizeof(struct tcf_skbedit),
};

static __net_init int skbedit_init_net(struct net *net)
{
	struct tc_action_net *tn = net_generic(net, act_skbedit_ops.net_id);

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

static void __net_exit skbedit_exit_net(struct list_head *net_list)
{
	tc_action_net_exit(net_list, act_skbedit_ops.net_id);
}

static struct pernet_operations skbedit_net_ops = {
	.init = skbedit_init_net,
	.exit_batch = skbedit_exit_net,
	.id   = &act_skbedit_ops.net_id,
	.size = sizeof(struct tc_action_net),
};

MODULE_AUTHOR("Alexander Duyck, <alexander.h.duyck@intel.com>");
MODULE_DESCRIPTION("SKB Editing");
MODULE_LICENSE("GPL");

static int __init skbedit_init_module(void)
{
	return tcf_register_action(&act_skbedit_ops, &skbedit_net_ops);
}

static void __exit skbedit_cleanup_module(void)
{
	tcf_unregister_action(&act_skbedit_ops, &skbedit_net_ops);
}

module_init(skbedit_init_module);
module_exit(skbedit_cleanup_module);
Commit	Line	Data
9952f691	1	// SPDX-License-Identifier: GPL-2.0-only
ca9b0e27 AD	2	/*
	3	* Copyright (c) 2008, Intel Corporation.
	4	*
ca9b0e27 AD	5	* Author: Alexander Duyck <alexander.h.duyck@intel.com>
	6	*/
	7
	8	#include <linux/module.h>
	9	#include <linux/init.h>
	10	#include <linux/kernel.h>
	11	#include <linux/skbuff.h>
	12	#include <linux/rtnetlink.h>
	13	#include <net/netlink.h>
	14	#include <net/pkt_sched.h>
e7e3728b QF	15	#include <net/ip.h>
	16	#include <net/ipv6.h>
	17	#include <net/dsfield.h>
ec7727bb	18	#include <net/pkt_cls.h>
871cf386	19	#include <net/tc_wrapper.h>
ca9b0e27 AD	20
	21	#include <linux/tc_act/tc_skbedit.h>
	22	#include <net/tc_act/tc_skbedit.h>
	23
a85a970a	24	static struct tc_action_ops act_skbedit_ops;
ddf97ccd	25
38a6f086 TZ	26	static u16 tcf_skbedit_hash(struct tcf_skbedit_params *params,
	27	struct sk_buff *skb)
	28	{
	29	u16 queue_mapping = params->queue_mapping;
	30
	31	if (params->flags & SKBEDIT_F_TXQ_SKBHASH) {
	32	u32 hash = skb_get_hash(skb);
	33
	34	queue_mapping += hash % params->mapping_mod;
	35	}
	36
	37	return netdev_cap_txqueue(skb->dev, queue_mapping);
	38	}
	39
871cf386 PT	40	TC_INDIRECT_SCOPE int tcf_skbedit_act(struct sk_buff *skb,
	41	const struct tc_action *a,
	42	struct tcf_result *res)
ca9b0e27	43	{
a85a970a	44	struct tcf_skbedit *d = to_skbedit(a);
c749cdda DC	45	struct tcf_skbedit_params *params;
c749cdda DC	46	int action;
ca9b0e27	47
9c4a4e48	48	tcf_lastuse_update(&d->tcf_tm);
50dc9a85	49	bstats_update(this_cpu_ptr(d->common.cpu_bstats), skb);
ca9b0e27	50
7fd4b288	51	params = rcu_dereference_bh(d->params);
c749cdda DC	52	action = READ_ONCE(d->tcf_action);
	53
	54	if (params->flags & SKBEDIT_F_PRIORITY)
	55	skb->priority = params->priority;
	56	if (params->flags & SKBEDIT_F_INHERITDSFIELD) {
e7e3728b QF	57	int wlen = skb_network_offset(skb);
e7e3728b QF	58
d7bf2ebe	59	switch (skb_protocol(skb, true)) {
e7e3728b QF	60	case htons(ETH_P_IP):
	61	wlen += sizeof(struct iphdr);
	62	if (!pskb_may_pull(skb, wlen))
	63	goto err;
	64	skb->priority = ipv4_get_dsfield(ip_hdr(skb)) >> 2;
	65	break;
	66
	67	case htons(ETH_P_IPV6):
	68	wlen += sizeof(struct ipv6hdr);
	69	if (!pskb_may_pull(skb, wlen))
	70	goto err;
	71	skb->priority = ipv6_get_dsfield(ipv6_hdr(skb)) >> 2;
	72	break;
	73	}
	74	}
c749cdda	75	if (params->flags & SKBEDIT_F_QUEUE_MAPPING &&
2f1e85b1 TZ	76	skb->dev->real_num_tx_queues > params->queue_mapping) {
	77	#ifdef CONFIG_NET_EGRESS
	78	netdev_xmit_skip_txqueue(true);
	79	#endif
38a6f086	80	skb_set_queue_mapping(skb, tcf_skbedit_hash(params, skb));
2f1e85b1	81	}
c749cdda DC	82	if (params->flags & SKBEDIT_F_MARK) {
	83	skb->mark &= ~params->mask;
	84	skb->mark \|= params->mark & params->mask;
4fe77d82	85	}
c749cdda DC	86	if (params->flags & SKBEDIT_F_PTYPE)
c749cdda DC	87	skb->pkt_type = params->ptype;
c749cdda	88	return action;
7fd4b288	89
e7e3728b	90	err:
6f3dfb0d	91	qstats_drop_inc(this_cpu_ptr(d->common.cpu_qstats));
7fd4b288	92	return TC_ACT_SHOT;
ca9b0e27 AD	93	}
ca9b0e27 AD	94
837cb17d	95	static void tcf_skbedit_stats_update(struct tc_action *a, u64 bytes,
4b61d3e8 PL	96	u64 packets, u64 drops,
4b61d3e8 PL	97	u64 lastuse, bool hw)
837cb17d PM	98	{
	99	struct tcf_skbedit *d = to_skbedit(a);
	100	struct tcf_t *tm = &d->tcf_tm;
	101
4b61d3e8	102	tcf_action_update_stats(a, bytes, packets, drops, hw);
837cb17d PM	103	tm->lastuse = max_t(u64, tm->lastuse, lastuse);
	104	}
	105
ca9b0e27 AD	106	static const struct nla_policy skbedit_policy[TCA_SKBEDIT_MAX + 1] = {
	107	[TCA_SKBEDIT_PARMS] = { .len = sizeof(struct tc_skbedit) },
	108	[TCA_SKBEDIT_PRIORITY] = { .len = sizeof(u32) },
	109	[TCA_SKBEDIT_QUEUE_MAPPING] = { .len = sizeof(u16) },
1c55d62e	110	[TCA_SKBEDIT_MARK] = { .len = sizeof(u32) },
ff202ee1	111	[TCA_SKBEDIT_PTYPE] = { .len = sizeof(u16) },
4fe77d82	112	[TCA_SKBEDIT_MASK] = { .len = sizeof(u32) },
e7e3728b	113	[TCA_SKBEDIT_FLAGS] = { .len = sizeof(u64) },
38a6f086	114	[TCA_SKBEDIT_QUEUE_MAPPING_MAX] = { .len = sizeof(u16) },
ca9b0e27 AD	115	};
ca9b0e27 AD	116
c1b52739	117	static int tcf_skbedit_init(struct net net, struct nlattr nla,
a85a970a	118	struct nlattr est, struct tc_action *a,
abbb0d33	119	struct tcf_proto *tp, u32 act_flags,
789871bb	120	struct netlink_ext_ack *extack)
ca9b0e27	121	{
acd0a7ab	122	struct tc_action_net *tn = net_generic(net, act_skbedit_ops.net_id);
695176bf	123	bool bind = act_flags & TCA_ACT_FLAGS_BIND;
6d7a8df6	124	struct tcf_skbedit_params *params_new;
ca9b0e27	125	struct nlattr *tb[TCA_SKBEDIT_MAX + 1];
ec7727bb	126	struct tcf_chain *goto_ch = NULL;
ca9b0e27 AD	127	struct tc_skbedit *parm;
ca9b0e27 AD	128	struct tcf_skbedit *d;
4fe77d82	129	u32 flags = 0, priority = NULL, mark = NULL, *mask = NULL;
ff202ee1	130	u16 queue_mapping = NULL, ptype = NULL;
38a6f086	131	u16 mapping_mod = 1;
b2313077 WC	132	bool exists = false;
b2313077 WC	133	int ret = 0, err;
7be8ef2c	134	u32 index;
ca9b0e27 AD	135
	136	if (nla == NULL)
	137	return -EINVAL;
	138
8cb08174 JB	139	err = nla_parse_nested_deprecated(tb, TCA_SKBEDIT_MAX, nla,
8cb08174 JB	140	skbedit_policy, NULL);
ca9b0e27 AD	141	if (err < 0)
	142	return err;
	143
	144	if (tb[TCA_SKBEDIT_PARMS] == NULL)
	145	return -EINVAL;
	146
	147	if (tb[TCA_SKBEDIT_PRIORITY] != NULL) {
	148	flags \|= SKBEDIT_F_PRIORITY;
	149	priority = nla_data(tb[TCA_SKBEDIT_PRIORITY]);
	150	}
	151
	152	if (tb[TCA_SKBEDIT_QUEUE_MAPPING] != NULL) {
4a6a676f AN	153	if (is_tcf_skbedit_ingress(act_flags) &&
	154	!(act_flags & TCA_ACT_FLAGS_SKIP_SW)) {
	155	NL_SET_ERR_MSG_MOD(extack, "\"queue_mapping\" option on receive side is hardware only, use skip_sw");
	156	return -EOPNOTSUPP;
	157	}
ca9b0e27 AD	158	flags \|= SKBEDIT_F_QUEUE_MAPPING;
	159	queue_mapping = nla_data(tb[TCA_SKBEDIT_QUEUE_MAPPING]);
	160	}
1c55d62e	161
ff202ee1 JHS	162	if (tb[TCA_SKBEDIT_PTYPE] != NULL) {
	163	ptype = nla_data(tb[TCA_SKBEDIT_PTYPE]);
	164	if (!skb_pkt_type_ok(*ptype))
	165	return -EINVAL;
	166	flags \|= SKBEDIT_F_PTYPE;
	167	}
	168
1c55d62e	169	if (tb[TCA_SKBEDIT_MARK] != NULL) {
	170	flags \|= SKBEDIT_F_MARK;
	171	mark = nla_data(tb[TCA_SKBEDIT_MARK]);
	172	}
	173
4fe77d82 AQ	174	if (tb[TCA_SKBEDIT_MASK] != NULL) {
	175	flags \|= SKBEDIT_F_MASK;
	176	mask = nla_data(tb[TCA_SKBEDIT_MASK]);
	177	}
	178
e7e3728b QF	179	if (tb[TCA_SKBEDIT_FLAGS] != NULL) {
	180	u64 *pure_flags = nla_data(tb[TCA_SKBEDIT_FLAGS]);
	181
38a6f086 TZ	182	if (*pure_flags & SKBEDIT_F_TXQ_SKBHASH) {
	183	u16 *queue_mapping_max;
	184
	185	if (!tb[TCA_SKBEDIT_QUEUE_MAPPING] \|\|
	186	!tb[TCA_SKBEDIT_QUEUE_MAPPING_MAX]) {
	187	NL_SET_ERR_MSG_MOD(extack, "Missing required range of queue_mapping.");
	188	return -EINVAL;
	189	}
	190
	191	queue_mapping_max =
	192	nla_data(tb[TCA_SKBEDIT_QUEUE_MAPPING_MAX]);
	193	if (queue_mapping_max < queue_mapping) {
	194	NL_SET_ERR_MSG_MOD(extack, "The range of queue_mapping is invalid, max < min.");
	195	return -EINVAL;
	196	}
	197
	198	mapping_mod = queue_mapping_max - queue_mapping + 1;
	199	flags \|= SKBEDIT_F_TXQ_SKBHASH;
	200	}
e7e3728b QF	201	if (*pure_flags & SKBEDIT_F_INHERITDSFIELD)
	202	flags \|= SKBEDIT_F_INHERITDSFIELD;
	203	}
	204
ca9b0e27	205	parm = nla_data(tb[TCA_SKBEDIT_PARMS]);
7be8ef2c DL	206	index = parm->index;
7be8ef2c DL	207	err = tcf_idr_check_alloc(tn, &index, a, bind);
0190c1d4 VB	208	if (err < 0)
	209	return err;
	210	exists = err;
5e1567ae JHS	211	if (exists && bind)
	212	return 0;
	213
	214	if (!flags) {
af5d0184 RM	215	if (exists)
af5d0184 RM	216	tcf_idr_release(*a, bind);
0190c1d4	217	else
7be8ef2c	218	tcf_idr_cleanup(tn, index);
5e1567ae JHS	219	return -EINVAL;
	220	}
	221
	222	if (!exists) {
7be8ef2c	223	ret = tcf_idr_create(tn, index, est, a,
40bd094d	224	&act_skbedit_ops, bind, true, act_flags);
0190c1d4	225	if (ret) {
7be8ef2c	226	tcf_idr_cleanup(tn, index);
86062033	227	return ret;
0190c1d4	228	}
ca9b0e27	229
a85a970a	230	d = to_skbedit(*a);
ca9b0e27 AD	231	ret = ACT_P_CREATED;
ca9b0e27 AD	232	} else {
a85a970a	233	d = to_skbedit(*a);
695176bf	234	if (!(act_flags & TCA_ACT_FLAGS_REPLACE)) {
4e8ddd7f	235	tcf_idr_release(*a, bind);
ca9b0e27	236	return -EEXIST;
4e8ddd7f	237	}
ca9b0e27	238	}
ec7727bb DC	239	err = tcf_action_check_ctrlact(parm->action, tp, &goto_ch, extack);
	240	if (err < 0)
	241	goto release_idr;
ca9b0e27	242
c749cdda DC	243	params_new = kzalloc(sizeof(*params_new), GFP_KERNEL);
c749cdda DC	244	if (unlikely(!params_new)) {
ec7727bb DC	245	err = -ENOMEM;
ec7727bb DC	246	goto put_chain;
c749cdda DC	247	}
	248
	249	params_new->flags = flags;
ca9b0e27	250	if (flags & SKBEDIT_F_PRIORITY)
c749cdda	251	params_new->priority = *priority;
38a6f086	252	if (flags & SKBEDIT_F_QUEUE_MAPPING) {
c749cdda	253	params_new->queue_mapping = *queue_mapping;
38a6f086 TZ	254	params_new->mapping_mod = mapping_mod;
38a6f086 TZ	255	}
1c55d62e	256	if (flags & SKBEDIT_F_MARK)
c749cdda	257	params_new->mark = *mark;
ff202ee1	258	if (flags & SKBEDIT_F_PTYPE)
c749cdda	259	params_new->ptype = *ptype;
4fe77d82	260	/* default behaviour is to use all the bits */
c749cdda	261	params_new->mask = 0xffffffff;
4fe77d82	262	if (flags & SKBEDIT_F_MASK)
c749cdda	263	params_new->mask = *mask;
1c55d62e	264
6d7a8df6	265	spin_lock_bh(&d->tcf_lock);
ec7727bb	266	goto_ch = tcf_action_set_ctrlact(*a, parm->action, goto_ch);
445d3749 PM	267	params_new = rcu_replace_pointer(d->params, params_new,
445d3749 PM	268	lockdep_is_held(&d->tcf_lock));
6d7a8df6 VB	269	spin_unlock_bh(&d->tcf_lock);
	270	if (params_new)
	271	kfree_rcu(params_new, rcu);
ec7727bb DC	272	if (goto_ch)
ec7727bb DC	273	tcf_chain_put_by_act(goto_ch);
ca9b0e27	274
ca9b0e27	275	return ret;
ec7727bb DC	276	put_chain:
	277	if (goto_ch)
	278	tcf_chain_put_by_act(goto_ch);
	279	release_idr:
	280	tcf_idr_release(*a, bind);
	281	return err;
ca9b0e27 AD	282	}
ca9b0e27 AD	283
cc7ec456 ED	284	static int tcf_skbedit_dump(struct sk_buff skb, struct tc_action a,
cc7ec456 ED	285	int bind, int ref)
ca9b0e27 AD	286	{
ca9b0e27 AD	287	unsigned char *b = skb_tail_pointer(skb);
a85a970a	288	struct tcf_skbedit *d = to_skbedit(a);
c749cdda	289	struct tcf_skbedit_params *params;
1c40be12 ED	290	struct tc_skbedit opt = {
1c40be12 ED	291	.index = d->tcf_index,
036bb443 VB	292	.refcnt = refcount_read(&d->tcf_refcnt) - ref,
036bb443 VB	293	.bindcnt = atomic_read(&d->tcf_bindcnt) - bind,
1c40be12	294	};
e7e3728b	295	u64 pure_flags = 0;
c749cdda DC	296	struct tcf_t t;
c749cdda DC	297
6d7a8df6 VB	298	spin_lock_bh(&d->tcf_lock);
	299	params = rcu_dereference_protected(d->params,
	300	lockdep_is_held(&d->tcf_lock));
	301	opt.action = d->tcf_action;
ca9b0e27	302
1b34ec43 DM	303	if (nla_put(skb, TCA_SKBEDIT_PARMS, sizeof(opt), &opt))
1b34ec43 DM	304	goto nla_put_failure;
c749cdda DC	305	if ((params->flags & SKBEDIT_F_PRIORITY) &&
c749cdda DC	306	nla_put_u32(skb, TCA_SKBEDIT_PRIORITY, params->priority))
1b34ec43	307	goto nla_put_failure;
c749cdda DC	308	if ((params->flags & SKBEDIT_F_QUEUE_MAPPING) &&
c749cdda DC	309	nla_put_u16(skb, TCA_SKBEDIT_QUEUE_MAPPING, params->queue_mapping))
1b34ec43	310	goto nla_put_failure;
c749cdda DC	311	if ((params->flags & SKBEDIT_F_MARK) &&
c749cdda DC	312	nla_put_u32(skb, TCA_SKBEDIT_MARK, params->mark))
1b34ec43	313	goto nla_put_failure;
c749cdda DC	314	if ((params->flags & SKBEDIT_F_PTYPE) &&
c749cdda DC	315	nla_put_u16(skb, TCA_SKBEDIT_PTYPE, params->ptype))
ff202ee1	316	goto nla_put_failure;
c749cdda DC	317	if ((params->flags & SKBEDIT_F_MASK) &&
c749cdda DC	318	nla_put_u32(skb, TCA_SKBEDIT_MASK, params->mask))
4fe77d82	319	goto nla_put_failure;
c749cdda	320	if (params->flags & SKBEDIT_F_INHERITDSFIELD)
e7e3728b	321	pure_flags \|= SKBEDIT_F_INHERITDSFIELD;
38a6f086 TZ	322	if (params->flags & SKBEDIT_F_TXQ_SKBHASH) {
	323	if (nla_put_u16(skb, TCA_SKBEDIT_QUEUE_MAPPING_MAX,
	324	params->queue_mapping + params->mapping_mod - 1))
	325	goto nla_put_failure;
	326
	327	pure_flags \|= SKBEDIT_F_TXQ_SKBHASH;
	328	}
e7e3728b QF	329	if (pure_flags != 0 &&
	330	nla_put(skb, TCA_SKBEDIT_FLAGS, sizeof(pure_flags), &pure_flags))
	331	goto nla_put_failure;
48d8ee16 JHS	332
48d8ee16 JHS	333	tcf_tm_dump(&t, &d->tcf_tm);
9854518e	334	if (nla_put_64bit(skb, TCA_SKBEDIT_TM, sizeof(t), &t, TCA_SKBEDIT_PAD))
1b34ec43	335	goto nla_put_failure;
6d7a8df6 VB	336	spin_unlock_bh(&d->tcf_lock);
6d7a8df6 VB	337
ca9b0e27 AD	338	return skb->len;
	339
	340	nla_put_failure:
6d7a8df6	341	spin_unlock_bh(&d->tcf_lock);
ca9b0e27 AD	342	nlmsg_trim(skb, b);
	343	return -1;
	344	}
	345
c749cdda DC	346	static void tcf_skbedit_cleanup(struct tc_action *a)
	347	{
	348	struct tcf_skbedit *d = to_skbedit(a);
	349	struct tcf_skbedit_params *params;
	350
	351	params = rcu_dereference_protected(d->params, 1);
	352	if (params)
	353	kfree_rcu(params, rcu);
	354	}
	355
e1fea322 RM	356	static size_t tcf_skbedit_get_fill_size(const struct tc_action *act)
	357	{
	358	return nla_total_size(sizeof(struct tc_skbedit))
	359	+ nla_total_size(sizeof(u32)) /* TCA_SKBEDIT_PRIORITY */
	360	+ nla_total_size(sizeof(u16)) /* TCA_SKBEDIT_QUEUE_MAPPING */
38a6f086	361	+ nla_total_size(sizeof(u16)) /* TCA_SKBEDIT_QUEUE_MAPPING_MAX */
e1fea322 RM	362	+ nla_total_size(sizeof(u32)) /* TCA_SKBEDIT_MARK */
	363	+ nla_total_size(sizeof(u16)) /* TCA_SKBEDIT_PTYPE */
	364	+ nla_total_size(sizeof(u32)) /* TCA_SKBEDIT_MASK */
	365	+ nla_total_size_64bit(sizeof(u64)); /* TCA_SKBEDIT_FLAGS */
	366	}
	367
c54e1d92	368	static int tcf_skbedit_offload_act_setup(struct tc_action act, void entry_data,
c2ccf84e IS	369	u32 *index_inc, bool bind,
c2ccf84e IS	370	struct netlink_ext_ack *extack)
c54e1d92 BZ	371	{
	372	if (bind) {
	373	struct flow_action_entry *entry = entry_data;
	374
	375	if (is_tcf_skbedit_mark(act)) {
	376	entry->id = FLOW_ACTION_MARK;
	377	entry->mark = tcf_skbedit_mark(act);
	378	} else if (is_tcf_skbedit_ptype(act)) {
	379	entry->id = FLOW_ACTION_PTYPE;
	380	entry->ptype = tcf_skbedit_ptype(act);
	381	} else if (is_tcf_skbedit_priority(act)) {
	382	entry->id = FLOW_ACTION_PRIORITY;
	383	entry->priority = tcf_skbedit_priority(act);
4a6a676f AN	384	} else if (is_tcf_skbedit_tx_queue_mapping(act)) {
4a6a676f AN	385	NL_SET_ERR_MSG_MOD(extack, "Offload not supported when \"queue_mapping\" option is used on transmit side");
a9c64939	386	return -EOPNOTSUPP;
4a6a676f AN	387	} else if (is_tcf_skbedit_rx_queue_mapping(act)) {
	388	entry->id = FLOW_ACTION_RX_QUEUE_MAPPING;
	389	entry->rx_queue = tcf_skbedit_rx_queue_mapping(act);
a9c64939 IS	390	} else if (is_tcf_skbedit_inheritdsfield(act)) {
	391	NL_SET_ERR_MSG_MOD(extack, "Offload not supported when \"inheritdsfield\" option is used");
	392	return -EOPNOTSUPP;
c54e1d92	393	} else {
a9c64939	394	NL_SET_ERR_MSG_MOD(extack, "Unsupported skbedit option offload");
c54e1d92 BZ	395	return -EOPNOTSUPP;
	396	}
	397	*index_inc = 1;
	398	} else {
8cbfe939 BZ	399	struct flow_offload_action *fl_action = entry_data;
	400
	401	if (is_tcf_skbedit_mark(act))
	402	fl_action->id = FLOW_ACTION_MARK;
	403	else if (is_tcf_skbedit_ptype(act))
	404	fl_action->id = FLOW_ACTION_PTYPE;
	405	else if (is_tcf_skbedit_priority(act))
	406	fl_action->id = FLOW_ACTION_PRIORITY;
4a6a676f AN	407	else if (is_tcf_skbedit_rx_queue_mapping(act))
4a6a676f AN	408	fl_action->id = FLOW_ACTION_RX_QUEUE_MAPPING;
8cbfe939 BZ	409	else
8cbfe939 BZ	410	return -EOPNOTSUPP;
c54e1d92 BZ	411	}
	412
	413	return 0;
	414	}
	415
ca9b0e27 AD	416	static struct tc_action_ops act_skbedit_ops = {
ca9b0e27 AD	417	.kind = "skbedit",
eddd2cf1	418	.id = TCA_ID_SKBEDIT,
ca9b0e27	419	.owner = THIS_MODULE,
45da1dac	420	.act = tcf_skbedit_act,
837cb17d	421	.stats_update = tcf_skbedit_stats_update,
ca9b0e27	422	.dump = tcf_skbedit_dump,
ca9b0e27	423	.init = tcf_skbedit_init,
c749cdda	424	.cleanup = tcf_skbedit_cleanup,
e1fea322	425	.get_fill_size = tcf_skbedit_get_fill_size,
c54e1d92	426	.offload_act_setup = tcf_skbedit_offload_act_setup,
a85a970a	427	.size = sizeof(struct tcf_skbedit),
ddf97ccd WC	428	};
	429
	430	static __net_init int skbedit_init_net(struct net *net)
	431	{
acd0a7ab	432	struct tc_action_net *tn = net_generic(net, act_skbedit_ops.net_id);
ddf97ccd	433
981471bd	434	return tc_action_net_init(net, tn, &act_skbedit_ops);
ddf97ccd WC	435	}
ddf97ccd WC	436
039af9c6	437	static void __net_exit skbedit_exit_net(struct list_head *net_list)
ddf97ccd	438	{
acd0a7ab	439	tc_action_net_exit(net_list, act_skbedit_ops.net_id);
ddf97ccd WC	440	}
	441
	442	static struct pernet_operations skbedit_net_ops = {
	443	.init = skbedit_init_net,
039af9c6	444	.exit_batch = skbedit_exit_net,
acd0a7ab	445	.id = &act_skbedit_ops.net_id,
ddf97ccd	446	.size = sizeof(struct tc_action_net),
ca9b0e27 AD	447	};
	448
	449	MODULE_AUTHOR("Alexander Duyck, <alexander.h.duyck@intel.com>");
	450	MODULE_DESCRIPTION("SKB Editing");
	451	MODULE_LICENSE("GPL");
	452
	453	static int __init skbedit_init_module(void)
	454	{
ddf97ccd	455	return tcf_register_action(&act_skbedit_ops, &skbedit_net_ops);
ca9b0e27 AD	456	}
	457
	458	static void __exit skbedit_cleanup_module(void)
	459	{
ddf97ccd	460	tcf_unregister_action(&act_skbedit_ops, &skbedit_net_ops);
ca9b0e27 AD	461	}
	462
	463	module_init(skbedit_init_module);
	464	module_exit(skbedit_cleanup_module);