[mirror_ubuntu-jammy-kernel.git] / net / netfilter / nf_flow_table_core.c

// SPDX-License-Identifier: GPL-2.0-only
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/netfilter.h>
#include <linux/rhashtable.h>
#include <linux/netdevice.h>
#include <net/ip.h>
#include <net/ip6_route.h>
#include <net/netfilter/nf_tables.h>
#include <net/netfilter/nf_flow_table.h>
#include <net/netfilter/nf_conntrack.h>
#include <net/netfilter/nf_conntrack_core.h>
#include <net/netfilter/nf_conntrack_tuple.h>

struct flow_offload_entry {
	struct flow_offload	flow;
	struct nf_conn		*ct;
	struct rcu_head		rcu_head;
};

static DEFINE_MUTEX(flowtable_lock);
static LIST_HEAD(flowtables);

static void
flow_offload_fill_dir(struct flow_offload *flow, struct nf_conn *ct,
		      struct nf_flow_route *route,
		      enum flow_offload_tuple_dir dir)
{
	struct flow_offload_tuple *ft = &flow->tuplehash[dir].tuple;
	struct nf_conntrack_tuple *ctt = &ct->tuplehash[dir].tuple;
	struct dst_entry *other_dst = route->tuple[!dir].dst;
	struct dst_entry *dst = route->tuple[dir].dst;

	ft->dir = dir;

	switch (ctt->src.l3num) {
	case NFPROTO_IPV4:
		ft->src_v4 = ctt->src.u3.in;
		ft->dst_v4 = ctt->dst.u3.in;
		ft->mtu = ip_dst_mtu_maybe_forward(dst, true);
		break;
	case NFPROTO_IPV6:
		ft->src_v6 = ctt->src.u3.in6;
		ft->dst_v6 = ctt->dst.u3.in6;
		ft->mtu = ip6_dst_mtu_forward(dst);
		break;
	}

	ft->l3proto = ctt->src.l3num;
	ft->l4proto = ctt->dst.protonum;
	ft->src_port = ctt->src.u.tcp.port;
	ft->dst_port = ctt->dst.u.tcp.port;

	ft->iifidx = other_dst->dev->ifindex;
	ft->dst_cache = dst;
}

struct flow_offload *
flow_offload_alloc(struct nf_conn *ct, struct nf_flow_route *route)
{
	struct flow_offload_entry *entry;
	struct flow_offload *flow;

	if (unlikely(nf_ct_is_dying(ct) ||
	    !atomic_inc_not_zero(&ct->ct_general.use)))
		return NULL;

	entry = kzalloc(sizeof(*entry), GFP_ATOMIC);
	if (!entry)
		goto err_ct_refcnt;

	flow = &entry->flow;

	if (!dst_hold_safe(route->tuple[FLOW_OFFLOAD_DIR_ORIGINAL].dst))
		goto err_dst_cache_original;

	if (!dst_hold_safe(route->tuple[FLOW_OFFLOAD_DIR_REPLY].dst))
		goto err_dst_cache_reply;

	entry->ct = ct;

	flow_offload_fill_dir(flow, ct, route, FLOW_OFFLOAD_DIR_ORIGINAL);
	flow_offload_fill_dir(flow, ct, route, FLOW_OFFLOAD_DIR_REPLY);

	if (ct->status & IPS_SRC_NAT)
		flow->flags |= FLOW_OFFLOAD_SNAT;
	if (ct->status & IPS_DST_NAT)
		flow->flags |= FLOW_OFFLOAD_DNAT;

	return flow;

err_dst_cache_reply:
	dst_release(route->tuple[FLOW_OFFLOAD_DIR_ORIGINAL].dst);
err_dst_cache_original:
	kfree(entry);
err_ct_refcnt:
	nf_ct_put(ct);

	return NULL;
}
EXPORT_SYMBOL_GPL(flow_offload_alloc);

static void flow_offload_fixup_tcp(struct ip_ct_tcp *tcp)
{
	tcp->state = TCP_CONNTRACK_ESTABLISHED;
	tcp->seen[0].td_maxwin = 0;
	tcp->seen[1].td_maxwin = 0;
}

#define NF_FLOWTABLE_TCP_PICKUP_TIMEOUT	(120 * HZ)
#define NF_FLOWTABLE_UDP_PICKUP_TIMEOUT	(30 * HZ)

static void flow_offload_fixup_ct(struct nf_conn *ct)
{
	const struct nf_conntrack_l4proto *l4proto;
	unsigned int timeout;
	int l4num;

	l4num = nf_ct_protonum(ct);
	if (l4num == IPPROTO_TCP)
		flow_offload_fixup_tcp(&ct->proto.tcp);

	l4proto = nf_ct_l4proto_find(l4num);
	if (!l4proto)
		return;

	if (l4num == IPPROTO_TCP)
		timeout = NF_FLOWTABLE_TCP_PICKUP_TIMEOUT;
	else if (l4num == IPPROTO_UDP)
		timeout = NF_FLOWTABLE_UDP_PICKUP_TIMEOUT;
	else
		return;

	ct->timeout = nfct_time_stamp + timeout;
}

void flow_offload_free(struct flow_offload *flow)
{
	struct flow_offload_entry *e;

	dst_release(flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.dst_cache);
	dst_release(flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.dst_cache);
	e = container_of(flow, struct flow_offload_entry, flow);
	if (flow->flags & FLOW_OFFLOAD_DYING)
		nf_ct_delete(e->ct, 0, 0);
	nf_ct_put(e->ct);
	kfree_rcu(e, rcu_head);
}
EXPORT_SYMBOL_GPL(flow_offload_free);

static u32 flow_offload_hash(const void *data, u32 len, u32 seed)
{
	const struct flow_offload_tuple *tuple = data;

	return jhash(tuple, offsetof(struct flow_offload_tuple, dir), seed);
}

static u32 flow_offload_hash_obj(const void *data, u32 len, u32 seed)
{
	const struct flow_offload_tuple_rhash *tuplehash = data;

	return jhash(&tuplehash->tuple, offsetof(struct flow_offload_tuple, dir), seed);
}

static int flow_offload_hash_cmp(struct rhashtable_compare_arg *arg,
					const void *ptr)
{
	const struct flow_offload_tuple *tuple = arg->key;
	const struct flow_offload_tuple_rhash *x = ptr;

	if (memcmp(&x->tuple, tuple, offsetof(struct flow_offload_tuple, dir)))
		return 1;

	return 0;
}

static const struct rhashtable_params nf_flow_offload_rhash_params = {
	.head_offset		= offsetof(struct flow_offload_tuple_rhash, node),
	.hashfn			= flow_offload_hash,
	.obj_hashfn		= flow_offload_hash_obj,
	.obj_cmpfn		= flow_offload_hash_cmp,
	.automatic_shrinking	= true,
};

int flow_offload_add(struct nf_flowtable *flow_table, struct flow_offload *flow)
{
	int err;

	err = rhashtable_insert_fast(&flow_table->rhashtable,
				     &flow->tuplehash[0].node,
				     nf_flow_offload_rhash_params);
	if (err < 0)
		return err;

	err = rhashtable_insert_fast(&flow_table->rhashtable,
				     &flow->tuplehash[1].node,
				     nf_flow_offload_rhash_params);
	if (err < 0) {
		rhashtable_remove_fast(&flow_table->rhashtable,
				       &flow->tuplehash[0].node,
				       nf_flow_offload_rhash_params);
		return err;
	}

	flow->timeout = (u32)jiffies;
	return 0;
}
EXPORT_SYMBOL_GPL(flow_offload_add);

static inline bool nf_flow_has_expired(const struct flow_offload *flow)
{
	return (__s32)(flow->timeout - (u32)jiffies) <= 0;
}

static void flow_offload_del(struct nf_flowtable *flow_table,
			     struct flow_offload *flow)
{
	struct flow_offload_entry *e;

	rhashtable_remove_fast(&flow_table->rhashtable,
			       &flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].node,
			       nf_flow_offload_rhash_params);
	rhashtable_remove_fast(&flow_table->rhashtable,
			       &flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].node,
			       nf_flow_offload_rhash_params);

	e = container_of(flow, struct flow_offload_entry, flow);
	clear_bit(IPS_OFFLOAD_BIT, &e->ct->status);

	if (nf_flow_has_expired(flow))
		flow_offload_fixup_ct(e->ct);

	flow_offload_free(flow);
}

void flow_offload_teardown(struct flow_offload *flow)
{
	struct flow_offload_entry *e;

	flow->flags |= FLOW_OFFLOAD_TEARDOWN;

	e = container_of(flow, struct flow_offload_entry, flow);
	flow_offload_fixup_ct(e->ct);
}
EXPORT_SYMBOL_GPL(flow_offload_teardown);

struct flow_offload_tuple_rhash *
flow_offload_lookup(struct nf_flowtable *flow_table,
		    struct flow_offload_tuple *tuple)
{
	struct flow_offload_tuple_rhash *tuplehash;
	struct flow_offload *flow;
	struct flow_offload_entry *e;
	int dir;

	tuplehash = rhashtable_lookup(&flow_table->rhashtable, tuple,
				      nf_flow_offload_rhash_params);
	if (!tuplehash)
		return NULL;

	dir = tuplehash->tuple.dir;
	flow = container_of(tuplehash, struct flow_offload, tuplehash[dir]);
	if (flow->flags & (FLOW_OFFLOAD_DYING | FLOW_OFFLOAD_TEARDOWN))
		return NULL;

	e = container_of(flow, struct flow_offload_entry, flow);
	if (unlikely(nf_ct_is_dying(e->ct)))
		return NULL;

	return tuplehash;
}
EXPORT_SYMBOL_GPL(flow_offload_lookup);

static int
nf_flow_table_iterate(struct nf_flowtable *flow_table,
		      void (*iter)(struct flow_offload *flow, void *data),
		      void *data)
{
	struct flow_offload_tuple_rhash *tuplehash;
	struct rhashtable_iter hti;
	struct flow_offload *flow;
	int err = 0;

	rhashtable_walk_enter(&flow_table->rhashtable, &hti);
	rhashtable_walk_start(&hti);

	while ((tuplehash = rhashtable_walk_next(&hti))) {
		if (IS_ERR(tuplehash)) {
			if (PTR_ERR(tuplehash) != -EAGAIN) {
				err = PTR_ERR(tuplehash);
				break;
			}
			continue;
		}
		if (tuplehash->tuple.dir)
			continue;

		flow = container_of(tuplehash, struct flow_offload, tuplehash[0]);

		iter(flow, data);
	}
	rhashtable_walk_stop(&hti);
	rhashtable_walk_exit(&hti);

	return err;
}

static void nf_flow_offload_gc_step(struct flow_offload *flow, void *data)
{
	struct nf_flowtable *flow_table = data;
	struct flow_offload_entry *e;

	e = container_of(flow, struct flow_offload_entry, flow);
	if (nf_flow_has_expired(flow) || nf_ct_is_dying(e->ct) ||
	    (flow->flags & (FLOW_OFFLOAD_DYING | FLOW_OFFLOAD_TEARDOWN)))
		flow_offload_del(flow_table, flow);
}

static void nf_flow_offload_work_gc(struct work_struct *work)
{
	struct nf_flowtable *flow_table;

	flow_table = container_of(work, struct nf_flowtable, gc_work.work);
	nf_flow_table_iterate(flow_table, nf_flow_offload_gc_step, flow_table);
	queue_delayed_work(system_power_efficient_wq, &flow_table->gc_work, HZ);
}

static int nf_flow_nat_port_tcp(struct sk_buff *skb, unsigned int thoff,
				__be16 port, __be16 new_port)
{
	struct tcphdr *tcph;

	if (!pskb_may_pull(skb, thoff + sizeof(*tcph)) ||
	    skb_try_make_writable(skb, thoff + sizeof(*tcph)))
		return -1;

	tcph = (void *)(skb_network_header(skb) + thoff);
	inet_proto_csum_replace2(&tcph->check, skb, port, new_port, true);

	return 0;
}

static int nf_flow_nat_port_udp(struct sk_buff *skb, unsigned int thoff,
				__be16 port, __be16 new_port)
{
	struct udphdr *udph;

	if (!pskb_may_pull(skb, thoff + sizeof(*udph)) ||
	    skb_try_make_writable(skb, thoff + sizeof(*udph)))
		return -1;

	udph = (void *)(skb_network_header(skb) + thoff);
	if (udph->check || skb->ip_summed == CHECKSUM_PARTIAL) {
		inet_proto_csum_replace2(&udph->check, skb, port,
					 new_port, true);
		if (!udph->check)
			udph->check = CSUM_MANGLED_0;
	}

	return 0;
}

static int nf_flow_nat_port(struct sk_buff *skb, unsigned int thoff,
			    u8 protocol, __be16 port, __be16 new_port)
{
	switch (protocol) {
	case IPPROTO_TCP:
		if (nf_flow_nat_port_tcp(skb, thoff, port, new_port) < 0)
			return NF_DROP;
		break;
	case IPPROTO_UDP:
		if (nf_flow_nat_port_udp(skb, thoff, port, new_port) < 0)
			return NF_DROP;
		break;
	}

	return 0;
}

int nf_flow_snat_port(const struct flow_offload *flow,
		      struct sk_buff *skb, unsigned int thoff,
		      u8 protocol, enum flow_offload_tuple_dir dir)
{
	struct flow_ports *hdr;
	__be16 port, new_port;

	if (!pskb_may_pull(skb, thoff + sizeof(*hdr)) ||
	    skb_try_make_writable(skb, thoff + sizeof(*hdr)))
		return -1;

	hdr = (void *)(skb_network_header(skb) + thoff);

	switch (dir) {
	case FLOW_OFFLOAD_DIR_ORIGINAL:
		port = hdr->source;
		new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.dst_port;
		hdr->source = new_port;
		break;
	case FLOW_OFFLOAD_DIR_REPLY:
		port = hdr->dest;
		new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.src_port;
		hdr->dest = new_port;
		break;
	default:
		return -1;
	}

	return nf_flow_nat_port(skb, thoff, protocol, port, new_port);
}
EXPORT_SYMBOL_GPL(nf_flow_snat_port);

int nf_flow_dnat_port(const struct flow_offload *flow,
		      struct sk_buff *skb, unsigned int thoff,
		      u8 protocol, enum flow_offload_tuple_dir dir)
{
	struct flow_ports *hdr;
	__be16 port, new_port;

	if (!pskb_may_pull(skb, thoff + sizeof(*hdr)) ||
	    skb_try_make_writable(skb, thoff + sizeof(*hdr)))
		return -1;

	hdr = (void *)(skb_network_header(skb) + thoff);

	switch (dir) {
	case FLOW_OFFLOAD_DIR_ORIGINAL:
		port = hdr->dest;
		new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.src_port;
		hdr->dest = new_port;
		break;
	case FLOW_OFFLOAD_DIR_REPLY:
		port = hdr->source;
		new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.dst_port;
		hdr->source = new_port;
		break;
	default:
		return -1;
	}

	return nf_flow_nat_port(skb, thoff, protocol, port, new_port);
}
EXPORT_SYMBOL_GPL(nf_flow_dnat_port);

int nf_flow_table_init(struct nf_flowtable *flowtable)
{
	int err;

	INIT_DEFERRABLE_WORK(&flowtable->gc_work, nf_flow_offload_work_gc);

	err = rhashtable_init(&flowtable->rhashtable,
			      &nf_flow_offload_rhash_params);
	if (err < 0)
		return err;

	queue_delayed_work(system_power_efficient_wq,
			   &flowtable->gc_work, HZ);

	mutex_lock(&flowtable_lock);
	list_add(&flowtable->list, &flowtables);
	mutex_unlock(&flowtable_lock);

	return 0;
}
EXPORT_SYMBOL_GPL(nf_flow_table_init);

static void nf_flow_table_do_cleanup(struct flow_offload *flow, void *data)
{
	struct net_device *dev = data;
	struct flow_offload_entry *e;

	e = container_of(flow, struct flow_offload_entry, flow);

	if (!dev) {
		flow_offload_teardown(flow);
		return;
	}
	if (net_eq(nf_ct_net(e->ct), dev_net(dev)) &&
	    (flow->tuplehash[0].tuple.iifidx == dev->ifindex ||
	     flow->tuplehash[1].tuple.iifidx == dev->ifindex))
		flow_offload_dead(flow);
}

static void nf_flow_table_iterate_cleanup(struct nf_flowtable *flowtable,
					  struct net_device *dev)
{
	nf_flow_table_iterate(flowtable, nf_flow_table_do_cleanup, dev);
	flush_delayed_work(&flowtable->gc_work);
}

void nf_flow_table_cleanup(struct net_device *dev)
{
	struct nf_flowtable *flowtable;

	mutex_lock(&flowtable_lock);
	list_for_each_entry(flowtable, &flowtables, list)
		nf_flow_table_iterate_cleanup(flowtable, dev);
	mutex_unlock(&flowtable_lock);
}
EXPORT_SYMBOL_GPL(nf_flow_table_cleanup);

void nf_flow_table_free(struct nf_flowtable *flow_table)
{
	mutex_lock(&flowtable_lock);
	list_del(&flow_table->list);
	mutex_unlock(&flowtable_lock);
	cancel_delayed_work_sync(&flow_table->gc_work);
	nf_flow_table_iterate(flow_table, nf_flow_table_do_cleanup, NULL);
	nf_flow_table_iterate(flow_table, nf_flow_offload_gc_step, flow_table);
	rhashtable_destroy(&flow_table->rhashtable);
}
EXPORT_SYMBOL_GPL(nf_flow_table_free);

MODULE_LICENSE("GPL");
MODULE_AUTHOR("Pablo Neira Ayuso <pablo@netfilter.org>");
Commit	Line	Data
09c434b8	1	// SPDX-License-Identifier: GPL-2.0-only
ac2a6666 PNA	2	#include <linux/kernel.h>
	3	#include <linux/init.h>
	4	#include <linux/module.h>
	5	#include <linux/netfilter.h>
	6	#include <linux/rhashtable.h>
	7	#include <linux/netdevice.h>
4f3780c0 FF	8	#include <net/ip.h>
4f3780c0 FF	9	#include <net/ip6_route.h>
c0ea1bcb	10	#include <net/netfilter/nf_tables.h>
ac2a6666 PNA	11	#include <net/netfilter/nf_flow_table.h>
	12	#include <net/netfilter/nf_conntrack.h>
	13	#include <net/netfilter/nf_conntrack_core.h>
	14	#include <net/netfilter/nf_conntrack_tuple.h>
	15
	16	struct flow_offload_entry {
	17	struct flow_offload flow;
	18	struct nf_conn *ct;
	19	struct rcu_head rcu_head;
	20	};
	21
84453a90 FF	22	static DEFINE_MUTEX(flowtable_lock);
	23	static LIST_HEAD(flowtables);
	24
047b300e FF	25	static void
	26	flow_offload_fill_dir(struct flow_offload flow, struct nf_conn ct,
	27	struct nf_flow_route *route,
	28	enum flow_offload_tuple_dir dir)
	29	{
	30	struct flow_offload_tuple *ft = &flow->tuplehash[dir].tuple;
	31	struct nf_conntrack_tuple *ctt = &ct->tuplehash[dir].tuple;
10f4e765	32	struct dst_entry *other_dst = route->tuple[!dir].dst;
4f3780c0	33	struct dst_entry *dst = route->tuple[dir].dst;
047b300e FF	34
	35	ft->dir = dir;
	36
	37	switch (ctt->src.l3num) {
	38	case NFPROTO_IPV4:
	39	ft->src_v4 = ctt->src.u3.in;
	40	ft->dst_v4 = ctt->dst.u3.in;
4f3780c0	41	ft->mtu = ip_dst_mtu_maybe_forward(dst, true);
047b300e FF	42	break;
	43	case NFPROTO_IPV6:
	44	ft->src_v6 = ctt->src.u3.in6;
	45	ft->dst_v6 = ctt->dst.u3.in6;
4f3780c0	46	ft->mtu = ip6_dst_mtu_forward(dst);
047b300e FF	47	break;
	48	}
	49
	50	ft->l3proto = ctt->src.l3num;
	51	ft->l4proto = ctt->dst.protonum;
	52	ft->src_port = ctt->src.u.tcp.port;
	53	ft->dst_port = ctt->dst.u.tcp.port;
	54
10f4e765	55	ft->iifidx = other_dst->dev->ifindex;
4f3780c0	56	ft->dst_cache = dst;
047b300e FF	57	}
047b300e FF	58
ac2a6666 PNA	59	struct flow_offload *
	60	flow_offload_alloc(struct nf_conn ct, struct nf_flow_route route)
	61	{
	62	struct flow_offload_entry *entry;
	63	struct flow_offload *flow;
	64
	65	if (unlikely(nf_ct_is_dying(ct) \|\|
	66	!atomic_inc_not_zero(&ct->ct_general.use)))
	67	return NULL;
	68
	69	entry = kzalloc(sizeof(*entry), GFP_ATOMIC);
	70	if (!entry)
	71	goto err_ct_refcnt;
	72
	73	flow = &entry->flow;
	74
	75	if (!dst_hold_safe(route->tuple[FLOW_OFFLOAD_DIR_ORIGINAL].dst))
	76	goto err_dst_cache_original;
	77
	78	if (!dst_hold_safe(route->tuple[FLOW_OFFLOAD_DIR_REPLY].dst))
	79	goto err_dst_cache_reply;
	80
	81	entry->ct = ct;
	82
047b300e FF	83	flow_offload_fill_dir(flow, ct, route, FLOW_OFFLOAD_DIR_ORIGINAL);
047b300e FF	84	flow_offload_fill_dir(flow, ct, route, FLOW_OFFLOAD_DIR_REPLY);
ac2a6666 PNA	85
	86	if (ct->status & IPS_SRC_NAT)
	87	flow->flags \|= FLOW_OFFLOAD_SNAT;
df1e2025	88	if (ct->status & IPS_DST_NAT)
ac2a6666 PNA	89	flow->flags \|= FLOW_OFFLOAD_DNAT;
	90
	91	return flow;
	92
	93	err_dst_cache_reply:
	94	dst_release(route->tuple[FLOW_OFFLOAD_DIR_ORIGINAL].dst);
	95	err_dst_cache_original:
	96	kfree(entry);
	97	err_ct_refcnt:
	98	nf_ct_put(ct);
	99
	100	return NULL;
	101	}
	102	EXPORT_SYMBOL_GPL(flow_offload_alloc);
	103
da5984e5 FF	104	static void flow_offload_fixup_tcp(struct ip_ct_tcp *tcp)
	105	{
	106	tcp->state = TCP_CONNTRACK_ESTABLISHED;
	107	tcp->seen[0].td_maxwin = 0;
	108	tcp->seen[1].td_maxwin = 0;
	109	}
	110
e97d9404 FW	111	#define NF_FLOWTABLE_TCP_PICKUP_TIMEOUT (120 * HZ)
	112	#define NF_FLOWTABLE_UDP_PICKUP_TIMEOUT (30 * HZ)
	113
3e68db2f	114	static void flow_offload_fixup_ct(struct nf_conn *ct)
da5984e5 FF	115	{
da5984e5 FF	116	const struct nf_conntrack_l4proto *l4proto;
da5984e5 FF	117	unsigned int timeout;
	118	int l4num;
	119
	120	l4num = nf_ct_protonum(ct);
	121	if (l4num == IPPROTO_TCP)
	122	flow_offload_fixup_tcp(&ct->proto.tcp);
	123
4a60dc74	124	l4proto = nf_ct_l4proto_find(l4num);
da5984e5 FF	125	if (!l4proto)
	126	return;
	127
da5984e5	128	if (l4num == IPPROTO_TCP)
e97d9404	129	timeout = NF_FLOWTABLE_TCP_PICKUP_TIMEOUT;
da5984e5	130	else if (l4num == IPPROTO_UDP)
e97d9404	131	timeout = NF_FLOWTABLE_UDP_PICKUP_TIMEOUT;
da5984e5 FF	132	else
	133	return;
	134
	135	ct->timeout = nfct_time_stamp + timeout;
	136	}
	137
ac2a6666 PNA	138	void flow_offload_free(struct flow_offload *flow)
	139	{
	140	struct flow_offload_entry *e;
	141
	142	dst_release(flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.dst_cache);
	143	dst_release(flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.dst_cache);
	144	e = container_of(flow, struct flow_offload_entry, flow);
da5984e5 FF	145	if (flow->flags & FLOW_OFFLOAD_DYING)
da5984e5 FF	146	nf_ct_delete(e->ct, 0, 0);
0ff90b6c FF	147	nf_ct_put(e->ct);
0ff90b6c FF	148	kfree_rcu(e, rcu_head);
ac2a6666 PNA	149	}
	150	EXPORT_SYMBOL_GPL(flow_offload_free);
	151
a268de77 FF	152	static u32 flow_offload_hash(const void *data, u32 len, u32 seed)
	153	{
	154	const struct flow_offload_tuple *tuple = data;
	155
	156	return jhash(tuple, offsetof(struct flow_offload_tuple, dir), seed);
	157	}
	158
	159	static u32 flow_offload_hash_obj(const void *data, u32 len, u32 seed)
	160	{
	161	const struct flow_offload_tuple_rhash *tuplehash = data;
	162
	163	return jhash(&tuplehash->tuple, offsetof(struct flow_offload_tuple, dir), seed);
	164	}
	165
	166	static int flow_offload_hash_cmp(struct rhashtable_compare_arg *arg,
	167	const void *ptr)
	168	{
	169	const struct flow_offload_tuple *tuple = arg->key;
	170	const struct flow_offload_tuple_rhash *x = ptr;
	171
	172	if (memcmp(&x->tuple, tuple, offsetof(struct flow_offload_tuple, dir)))
	173	return 1;
	174
	175	return 0;
	176	}
	177
	178	static const struct rhashtable_params nf_flow_offload_rhash_params = {
	179	.head_offset = offsetof(struct flow_offload_tuple_rhash, node),
	180	.hashfn = flow_offload_hash,
	181	.obj_hashfn = flow_offload_hash_obj,
	182	.obj_cmpfn = flow_offload_hash_cmp,
	183	.automatic_shrinking = true,
	184	};
	185
ac2a6666 PNA	186	int flow_offload_add(struct nf_flowtable flow_table, struct flow_offload flow)
ac2a6666 PNA	187	{
43c8f131 TY	188	int err;
	189
	190	err = rhashtable_insert_fast(&flow_table->rhashtable,
	191	&flow->tuplehash[0].node,
	192	nf_flow_offload_rhash_params);
	193	if (err < 0)
	194	return err;
	195
	196	err = rhashtable_insert_fast(&flow_table->rhashtable,
	197	&flow->tuplehash[1].node,
	198	nf_flow_offload_rhash_params);
	199	if (err < 0) {
	200	rhashtable_remove_fast(&flow_table->rhashtable,
	201	&flow->tuplehash[0].node,
	202	nf_flow_offload_rhash_params);
	203	return err;
	204	}
ac2a6666	205
43c8f131	206	flow->timeout = (u32)jiffies;
ac2a6666 PNA	207	return 0;
	208	}
	209	EXPORT_SYMBOL_GPL(flow_offload_add);
	210
3e68db2f PNA	211	static inline bool nf_flow_has_expired(const struct flow_offload *flow)
	212	{
	213	return (__s32)(flow->timeout - (u32)jiffies) <= 0;
	214	}
	215
0ff90b6c FF	216	static void flow_offload_del(struct nf_flowtable *flow_table,
0ff90b6c FF	217	struct flow_offload *flow)
ac2a6666	218	{
da5984e5 FF	219	struct flow_offload_entry *e;
da5984e5 FF	220
ac2a6666 PNA	221	rhashtable_remove_fast(&flow_table->rhashtable,
ac2a6666 PNA	222	&flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].node,
a268de77	223	nf_flow_offload_rhash_params);
ac2a6666 PNA	224	rhashtable_remove_fast(&flow_table->rhashtable,
ac2a6666 PNA	225	&flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].node,
a268de77	226	nf_flow_offload_rhash_params);
ac2a6666	227
da5984e5 FF	228	e = container_of(flow, struct flow_offload_entry, flow);
	229	clear_bit(IPS_OFFLOAD_BIT, &e->ct->status);
	230
3e68db2f PNA	231	if (nf_flow_has_expired(flow))
	232	flow_offload_fixup_ct(e->ct);
	233
0ff90b6c	234	flow_offload_free(flow);
ac2a6666	235	}
ac2a6666	236
59c466dd FF	237	void flow_offload_teardown(struct flow_offload *flow)
59c466dd FF	238	{
da5984e5 FF	239	struct flow_offload_entry *e;
da5984e5 FF	240
59c466dd	241	flow->flags \|= FLOW_OFFLOAD_TEARDOWN;
da5984e5 FF	242
da5984e5 FF	243	e = container_of(flow, struct flow_offload_entry, flow);
3e68db2f	244	flow_offload_fixup_ct(e->ct);
59c466dd FF	245	}
	246	EXPORT_SYMBOL_GPL(flow_offload_teardown);
	247
ac2a6666 PNA	248	struct flow_offload_tuple_rhash *
	249	flow_offload_lookup(struct nf_flowtable *flow_table,
	250	struct flow_offload_tuple *tuple)
	251	{
ba03137f FF	252	struct flow_offload_tuple_rhash *tuplehash;
ba03137f FF	253	struct flow_offload *flow;
8cd2bc98	254	struct flow_offload_entry *e;
ba03137f FF	255	int dir;
ba03137f FF	256
a2d88182 TY	257	tuplehash = rhashtable_lookup(&flow_table->rhashtable, tuple,
a2d88182 TY	258	nf_flow_offload_rhash_params);
ba03137f FF	259	if (!tuplehash)
	260	return NULL;
	261
	262	dir = tuplehash->tuple.dir;
	263	flow = container_of(tuplehash, struct flow_offload, tuplehash[dir]);
	264	if (flow->flags & (FLOW_OFFLOAD_DYING \| FLOW_OFFLOAD_TEARDOWN))
	265	return NULL;
	266
8cd2bc98 TY	267	e = container_of(flow, struct flow_offload_entry, flow);
	268	if (unlikely(nf_ct_is_dying(e->ct)))
	269	return NULL;
	270
ba03137f	271	return tuplehash;
ac2a6666 PNA	272	}
	273	EXPORT_SYMBOL_GPL(flow_offload_lookup);
	274
49de9c09 TY	275	static int
	276	nf_flow_table_iterate(struct nf_flowtable *flow_table,
	277	void (iter)(struct flow_offload flow, void *data),
	278	void *data)
ac2a6666 PNA	279	{
	280	struct flow_offload_tuple_rhash *tuplehash;
	281	struct rhashtable_iter hti;
	282	struct flow_offload *flow;
0de22baa	283	int err = 0;
ac2a6666	284
0de22baa	285	rhashtable_walk_enter(&flow_table->rhashtable, &hti);
ac2a6666 PNA	286	rhashtable_walk_start(&hti);
	287
	288	while ((tuplehash = rhashtable_walk_next(&hti))) {
	289	if (IS_ERR(tuplehash)) {
0de22baa TY	290	if (PTR_ERR(tuplehash) != -EAGAIN) {
	291	err = PTR_ERR(tuplehash);
	292	break;
	293	}
ac2a6666 PNA	294	continue;
	295	}
	296	if (tuplehash->tuple.dir)
	297	continue;
	298
	299	flow = container_of(tuplehash, struct flow_offload, tuplehash[0]);
	300
	301	iter(flow, data);
	302	}
ac2a6666 PNA	303	rhashtable_walk_stop(&hti);
	304	rhashtable_walk_exit(&hti);
	305
	306	return err;
	307	}
ac2a6666	308
b9660987	309	static void nf_flow_offload_gc_step(struct flow_offload flow, void data)
ac2a6666	310	{
b9660987	311	struct nf_flowtable *flow_table = data;
8cd2bc98	312	struct flow_offload_entry *e;
ac2a6666	313
8cd2bc98 TY	314	e = container_of(flow, struct flow_offload_entry, flow);
8cd2bc98 TY	315	if (nf_flow_has_expired(flow) \|\| nf_ct_is_dying(e->ct) \|\|
b9660987 TY	316	(flow->flags & (FLOW_OFFLOAD_DYING \| FLOW_OFFLOAD_TEARDOWN)))
b9660987 TY	317	flow_offload_del(flow_table, flow);
b408c5b0 PNA	318	}
b408c5b0 PNA	319
a268de77	320	static void nf_flow_offload_work_gc(struct work_struct *work)
b408c5b0 PNA	321	{
	322	struct nf_flowtable *flow_table;
	323
	324	flow_table = container_of(work, struct nf_flowtable, gc_work.work);
b9660987	325	nf_flow_table_iterate(flow_table, nf_flow_offload_gc_step, flow_table);
ac2a6666 PNA	326	queue_delayed_work(system_power_efficient_wq, &flow_table->gc_work, HZ);
ac2a6666 PNA	327	}
ac2a6666 PNA	328
	329	static int nf_flow_nat_port_tcp(struct sk_buff *skb, unsigned int thoff,
	330	__be16 port, __be16 new_port)
	331	{
	332	struct tcphdr *tcph;
	333
	334	if (!pskb_may_pull(skb, thoff + sizeof(*tcph)) \|\|
	335	skb_try_make_writable(skb, thoff + sizeof(*tcph)))
	336	return -1;
	337
	338	tcph = (void *)(skb_network_header(skb) + thoff);
	339	inet_proto_csum_replace2(&tcph->check, skb, port, new_port, true);
	340
	341	return 0;
	342	}
	343
	344	static int nf_flow_nat_port_udp(struct sk_buff *skb, unsigned int thoff,
	345	__be16 port, __be16 new_port)
	346	{
	347	struct udphdr *udph;
	348
	349	if (!pskb_may_pull(skb, thoff + sizeof(*udph)) \|\|
	350	skb_try_make_writable(skb, thoff + sizeof(*udph)))
	351	return -1;
	352
	353	udph = (void *)(skb_network_header(skb) + thoff);
	354	if (udph->check \|\| skb->ip_summed == CHECKSUM_PARTIAL) {
	355	inet_proto_csum_replace2(&udph->check, skb, port,
	356	new_port, true);
	357	if (!udph->check)
	358	udph->check = CSUM_MANGLED_0;
	359	}
	360
	361	return 0;
	362	}
	363
	364	static int nf_flow_nat_port(struct sk_buff *skb, unsigned int thoff,
	365	u8 protocol, __be16 port, __be16 new_port)
	366	{
	367	switch (protocol) {
	368	case IPPROTO_TCP:
	369	if (nf_flow_nat_port_tcp(skb, thoff, port, new_port) < 0)
	370	return NF_DROP;
	371	break;
	372	case IPPROTO_UDP:
	373	if (nf_flow_nat_port_udp(skb, thoff, port, new_port) < 0)
	374	return NF_DROP;
	375	break;
	376	}
	377
	378	return 0;
	379	}
	380
	381	int nf_flow_snat_port(const struct flow_offload *flow,
	382	struct sk_buff *skb, unsigned int thoff,
	383	u8 protocol, enum flow_offload_tuple_dir dir)
	384	{
	385	struct flow_ports *hdr;
	386	__be16 port, new_port;
	387
	388	if (!pskb_may_pull(skb, thoff + sizeof(*hdr)) \|\|
	389	skb_try_make_writable(skb, thoff + sizeof(*hdr)))
	390	return -1;
	391
392	hdr = (void *)(skb_network_header(skb) + thoff);
393
394	switch (dir) {
395	case FLOW_OFFLOAD_DIR_ORIGINAL:
396	port = hdr->source;
397	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.dst_port;
398	hdr->source = new_port;
399	break;
400	case FLOW_OFFLOAD_DIR_REPLY:
401	port = hdr->dest;
402	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.src_port;
403	hdr->dest = new_port;
404	break;
405	default:
406	return -1;
407	}
408
409	return nf_flow_nat_port(skb, thoff, protocol, port, new_port);
410	}
411	EXPORT_SYMBOL_GPL(nf_flow_snat_port);
412
413	int nf_flow_dnat_port(const struct flow_offload *flow,
414	struct sk_buff *skb, unsigned int thoff,
415	u8 protocol, enum flow_offload_tuple_dir dir)
416	{
417	struct flow_ports *hdr;
418	__be16 port, new_port;
419
420	if (!pskb_may_pull(skb, thoff + sizeof(*hdr)) \|\|
421	skb_try_make_writable(skb, thoff + sizeof(*hdr)))
422	return -1;
423
424	hdr = (void *)(skb_network_header(skb) + thoff);
425
426	switch (dir) {
427	case FLOW_OFFLOAD_DIR_ORIGINAL:
428	port = hdr->dest;
429	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.src_port;
430	hdr->dest = new_port;
431	break;
432	case FLOW_OFFLOAD_DIR_REPLY:
433	port = hdr->source;
434	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.dst_port;
435	hdr->source = new_port;
436	break;
437	default:
438	return -1;
439	}
440
441	return nf_flow_nat_port(skb, thoff, protocol, port, new_port);
442	}
443	EXPORT_SYMBOL_GPL(nf_flow_dnat_port);
444
a268de77 FF	445	int nf_flow_table_init(struct nf_flowtable *flowtable)
	446	{
	447	int err;
	448
	449	INIT_DEFERRABLE_WORK(&flowtable->gc_work, nf_flow_offload_work_gc);
	450
	451	err = rhashtable_init(&flowtable->rhashtable,
	452	&nf_flow_offload_rhash_params);
	453	if (err < 0)
	454	return err;
	455
	456	queue_delayed_work(system_power_efficient_wq,
	457	&flowtable->gc_work, HZ);
	458
84453a90 FF	459	mutex_lock(&flowtable_lock);
	460	list_add(&flowtable->list, &flowtables);
	461	mutex_unlock(&flowtable_lock);
	462
a268de77 FF	463	return 0;
	464	}
	465	EXPORT_SYMBOL_GPL(nf_flow_table_init);
	466
c0ea1bcb PNA	467	static void nf_flow_table_do_cleanup(struct flow_offload flow, void data)
	468	{
	469	struct net_device *dev = data;
a3fb3698 TY	470	struct flow_offload_entry *e;
	471
	472	e = container_of(flow, struct flow_offload_entry, flow);
c0ea1bcb	473
59c466dd FF	474	if (!dev) {
59c466dd FF	475	flow_offload_teardown(flow);
c0ea1bcb	476	return;
59c466dd	477	}
a3fb3698 TY	478	if (net_eq(nf_ct_net(e->ct), dev_net(dev)) &&
	479	(flow->tuplehash[0].tuple.iifidx == dev->ifindex \|\|
	480	flow->tuplehash[1].tuple.iifidx == dev->ifindex))
59c466dd	481	flow_offload_dead(flow);
c0ea1bcb PNA	482	}
	483
	484	static void nf_flow_table_iterate_cleanup(struct nf_flowtable *flowtable,
84453a90	485	struct net_device *dev)
c0ea1bcb	486	{
84453a90	487	nf_flow_table_iterate(flowtable, nf_flow_table_do_cleanup, dev);
c0ea1bcb PNA	488	flush_delayed_work(&flowtable->gc_work);
	489	}
	490
5f1be84a	491	void nf_flow_table_cleanup(struct net_device *dev)
c0ea1bcb	492	{
84453a90 FF	493	struct nf_flowtable *flowtable;
	494
	495	mutex_lock(&flowtable_lock);
	496	list_for_each_entry(flowtable, &flowtables, list)
	497	nf_flow_table_iterate_cleanup(flowtable, dev);
	498	mutex_unlock(&flowtable_lock);
c0ea1bcb PNA	499	}
	500	EXPORT_SYMBOL_GPL(nf_flow_table_cleanup);
	501
b408c5b0 PNA	502	void nf_flow_table_free(struct nf_flowtable *flow_table)
b408c5b0 PNA	503	{
84453a90 FF	504	mutex_lock(&flowtable_lock);
	505	list_del(&flow_table->list);
	506	mutex_unlock(&flowtable_lock);
a268de77	507	cancel_delayed_work_sync(&flow_table->gc_work);
b408c5b0	508	nf_flow_table_iterate(flow_table, nf_flow_table_do_cleanup, NULL);
b9660987	509	nf_flow_table_iterate(flow_table, nf_flow_offload_gc_step, flow_table);
a268de77	510	rhashtable_destroy(&flow_table->rhashtable);
b408c5b0 PNA	511	}
	512	EXPORT_SYMBOL_GPL(nf_flow_table_free);
	513
ac2a6666 PNA	514	MODULE_LICENSE("GPL");
ac2a6666 PNA	515	MODULE_AUTHOR("Pablo Neira Ayuso <pablo@netfilter.org>");