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

#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 *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 = route->tuple[dir].ifindex;
	ft->oifidx = route->tuple[!dir].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_state(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)
{
	flow->timeout = (u32)jiffies;

	rhashtable_insert_fast(&flow_table->rhashtable,
			       &flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].node,
			       nf_flow_offload_rhash_params);
	rhashtable_insert_fast(&flow_table->rhashtable,
			       &flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].node,
			       nf_flow_offload_rhash_params);
	return 0;
}
EXPORT_SYMBOL_GPL(flow_offload_add);

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);

	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_state(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;
	int dir;

	tuplehash = rhashtable_lookup_fast(&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;

	return tuplehash;
}
EXPORT_SYMBOL_GPL(flow_offload_lookup);

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;

	err = rhashtable_walk_init(&flow_table->rhashtable, &hti, GFP_KERNEL);
	if (err)
		return err;

	rhashtable_walk_start(&hti);

	while ((tuplehash = rhashtable_walk_next(&hti))) {
		if (IS_ERR(tuplehash)) {
			err = PTR_ERR(tuplehash);
			if (err != -EAGAIN)
				goto out;

			continue;
		}
		if (tuplehash->tuple.dir)
			continue;

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

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

	return err;
}
EXPORT_SYMBOL_GPL(nf_flow_table_iterate);

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

static int nf_flow_offload_gc_step(struct nf_flowtable *flow_table)
{
	struct flow_offload_tuple_rhash *tuplehash;
	struct rhashtable_iter hti;
	struct flow_offload *flow;
	int err;

	err = rhashtable_walk_init(&flow_table->rhashtable, &hti, GFP_KERNEL);
	if (err)
		return 0;

	rhashtable_walk_start(&hti);

	while ((tuplehash = rhashtable_walk_next(&hti))) {
		if (IS_ERR(tuplehash)) {
			err = PTR_ERR(tuplehash);
			if (err != -EAGAIN)
				goto out;

			continue;
		}
		if (tuplehash->tuple.dir)
			continue;

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

		if (nf_flow_has_expired(flow) ||
		    (flow->flags & (FLOW_OFFLOAD_DYING |
				    FLOW_OFFLOAD_TEARDOWN)))
			flow_offload_del(flow_table, flow);
	}
out:
	rhashtable_walk_stop(&hti);
	rhashtable_walk_exit(&hti);

	return 1;
}

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_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;

	if (!dev) {
		flow_offload_teardown(flow);
		return;
	}

	if (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 *net, 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);
	WARN_ON(!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
ac2a6666 PNA	1	#include <linux/kernel.h>
	2	#include <linux/init.h>
	3	#include <linux/module.h>
	4	#include <linux/netfilter.h>
	5	#include <linux/rhashtable.h>
	6	#include <linux/netdevice.h>
4f3780c0 FF	7	#include <net/ip.h>
4f3780c0 FF	8	#include <net/ip6_route.h>
c0ea1bcb	9	#include <net/netfilter/nf_tables.h>
ac2a6666 PNA	10	#include <net/netfilter/nf_flow_table.h>
	11	#include <net/netfilter/nf_conntrack.h>
	12	#include <net/netfilter/nf_conntrack_core.h>
	13	#include <net/netfilter/nf_conntrack_tuple.h>
	14
	15	struct flow_offload_entry {
	16	struct flow_offload flow;
	17	struct nf_conn *ct;
	18	struct rcu_head rcu_head;
	19	};
	20
84453a90 FF	21	static DEFINE_MUTEX(flowtable_lock);
	22	static LIST_HEAD(flowtables);
	23
047b300e FF	24	static void
	25	flow_offload_fill_dir(struct flow_offload flow, struct nf_conn ct,
	26	struct nf_flow_route *route,
	27	enum flow_offload_tuple_dir dir)
	28	{
	29	struct flow_offload_tuple *ft = &flow->tuplehash[dir].tuple;
	30	struct nf_conntrack_tuple *ctt = &ct->tuplehash[dir].tuple;
4f3780c0	31	struct dst_entry *dst = route->tuple[dir].dst;
047b300e FF	32
	33	ft->dir = dir;
	34
	35	switch (ctt->src.l3num) {
	36	case NFPROTO_IPV4:
	37	ft->src_v4 = ctt->src.u3.in;
	38	ft->dst_v4 = ctt->dst.u3.in;
4f3780c0	39	ft->mtu = ip_dst_mtu_maybe_forward(dst, true);
047b300e FF	40	break;
	41	case NFPROTO_IPV6:
	42	ft->src_v6 = ctt->src.u3.in6;
	43	ft->dst_v6 = ctt->dst.u3.in6;
4f3780c0	44	ft->mtu = ip6_dst_mtu_forward(dst);
047b300e FF	45	break;
	46	}
	47
	48	ft->l3proto = ctt->src.l3num;
	49	ft->l4proto = ctt->dst.protonum;
	50	ft->src_port = ctt->src.u.tcp.port;
	51	ft->dst_port = ctt->dst.u.tcp.port;
	52
	53	ft->iifidx = route->tuple[dir].ifindex;
	54	ft->oifidx = route->tuple[!dir].ifindex;
4f3780c0	55	ft->dst_cache = dst;
047b300e FF	56	}
047b300e FF	57
ac2a6666 PNA	58	struct flow_offload *
	59	flow_offload_alloc(struct nf_conn ct, struct nf_flow_route route)
	60	{
	61	struct flow_offload_entry *entry;
	62	struct flow_offload *flow;
	63
	64	if (unlikely(nf_ct_is_dying(ct) \|\|
	65	!atomic_inc_not_zero(&ct->ct_general.use)))
	66	return NULL;
	67
	68	entry = kzalloc(sizeof(*entry), GFP_ATOMIC);
	69	if (!entry)
	70	goto err_ct_refcnt;
	71
	72	flow = &entry->flow;
	73
	74	if (!dst_hold_safe(route->tuple[FLOW_OFFLOAD_DIR_ORIGINAL].dst))
	75	goto err_dst_cache_original;
	76
	77	if (!dst_hold_safe(route->tuple[FLOW_OFFLOAD_DIR_REPLY].dst))
	78	goto err_dst_cache_reply;
	79
	80	entry->ct = ct;
	81
047b300e FF	82	flow_offload_fill_dir(flow, ct, route, FLOW_OFFLOAD_DIR_ORIGINAL);
047b300e FF	83	flow_offload_fill_dir(flow, ct, route, FLOW_OFFLOAD_DIR_REPLY);
ac2a6666 PNA	84
	85	if (ct->status & IPS_SRC_NAT)
	86	flow->flags \|= FLOW_OFFLOAD_SNAT;
df1e2025	87	if (ct->status & IPS_DST_NAT)
ac2a6666 PNA	88	flow->flags \|= FLOW_OFFLOAD_DNAT;
	89
	90	return flow;
	91
	92	err_dst_cache_reply:
	93	dst_release(route->tuple[FLOW_OFFLOAD_DIR_ORIGINAL].dst);
	94	err_dst_cache_original:
	95	kfree(entry);
	96	err_ct_refcnt:
	97	nf_ct_put(ct);
	98
	99	return NULL;
	100	}
	101	EXPORT_SYMBOL_GPL(flow_offload_alloc);
	102
da5984e5 FF	103	static void flow_offload_fixup_tcp(struct ip_ct_tcp *tcp)
	104	{
	105	tcp->state = TCP_CONNTRACK_ESTABLISHED;
	106	tcp->seen[0].td_maxwin = 0;
	107	tcp->seen[1].td_maxwin = 0;
	108	}
	109
e97d9404 FW	110	#define NF_FLOWTABLE_TCP_PICKUP_TIMEOUT (120 * HZ)
	111	#define NF_FLOWTABLE_UDP_PICKUP_TIMEOUT (30 * HZ)
	112
da5984e5 FF	113	static void flow_offload_fixup_ct_state(struct nf_conn *ct)
	114	{
	115	const struct nf_conntrack_l4proto *l4proto;
da5984e5 FF	116	unsigned int timeout;
	117	int l4num;
	118
	119	l4num = nf_ct_protonum(ct);
	120	if (l4num == IPPROTO_TCP)
	121	flow_offload_fixup_tcp(&ct->proto.tcp);
	122
dd2934a9	123	l4proto = __nf_ct_l4proto_find(l4num);
da5984e5 FF	124	if (!l4proto)
	125	return;
	126
da5984e5	127	if (l4num == IPPROTO_TCP)
e97d9404	128	timeout = NF_FLOWTABLE_TCP_PICKUP_TIMEOUT;
da5984e5	129	else if (l4num == IPPROTO_UDP)
e97d9404	130	timeout = NF_FLOWTABLE_UDP_PICKUP_TIMEOUT;
da5984e5 FF	131	else
	132	return;
	133
	134	ct->timeout = nfct_time_stamp + timeout;
	135	}
	136
ac2a6666 PNA	137	void flow_offload_free(struct flow_offload *flow)
	138	{
	139	struct flow_offload_entry *e;
	140
	141	dst_release(flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.dst_cache);
	142	dst_release(flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.dst_cache);
	143	e = container_of(flow, struct flow_offload_entry, flow);
da5984e5 FF	144	if (flow->flags & FLOW_OFFLOAD_DYING)
da5984e5 FF	145	nf_ct_delete(e->ct, 0, 0);
0ff90b6c FF	146	nf_ct_put(e->ct);
0ff90b6c FF	147	kfree_rcu(e, rcu_head);
ac2a6666 PNA	148	}
	149	EXPORT_SYMBOL_GPL(flow_offload_free);
	150
a268de77 FF	151	static u32 flow_offload_hash(const void *data, u32 len, u32 seed)
	152	{
	153	const struct flow_offload_tuple *tuple = data;
	154
	155	return jhash(tuple, offsetof(struct flow_offload_tuple, dir), seed);
	156	}
	157
	158	static u32 flow_offload_hash_obj(const void *data, u32 len, u32 seed)
	159	{
	160	const struct flow_offload_tuple_rhash *tuplehash = data;
	161
	162	return jhash(&tuplehash->tuple, offsetof(struct flow_offload_tuple, dir), seed);
	163	}
	164
	165	static int flow_offload_hash_cmp(struct rhashtable_compare_arg *arg,
	166	const void *ptr)
	167	{
	168	const struct flow_offload_tuple *tuple = arg->key;
	169	const struct flow_offload_tuple_rhash *x = ptr;
	170
	171	if (memcmp(&x->tuple, tuple, offsetof(struct flow_offload_tuple, dir)))
	172	return 1;
	173
	174	return 0;
	175	}
	176
	177	static const struct rhashtable_params nf_flow_offload_rhash_params = {
	178	.head_offset = offsetof(struct flow_offload_tuple_rhash, node),
	179	.hashfn = flow_offload_hash,
	180	.obj_hashfn = flow_offload_hash_obj,
	181	.obj_cmpfn = flow_offload_hash_cmp,
	182	.automatic_shrinking = true,
	183	};
	184
ac2a6666 PNA	185	int flow_offload_add(struct nf_flowtable flow_table, struct flow_offload flow)
	186	{
	187	flow->timeout = (u32)jiffies;
	188
	189	rhashtable_insert_fast(&flow_table->rhashtable,
	190	&flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].node,
a268de77	191	nf_flow_offload_rhash_params);
ac2a6666 PNA	192	rhashtable_insert_fast(&flow_table->rhashtable,
ac2a6666 PNA	193	&flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].node,
a268de77	194	nf_flow_offload_rhash_params);
ac2a6666 PNA	195	return 0;
	196	}
	197	EXPORT_SYMBOL_GPL(flow_offload_add);
	198
0ff90b6c FF	199	static void flow_offload_del(struct nf_flowtable *flow_table,
0ff90b6c FF	200	struct flow_offload *flow)
ac2a6666	201	{
da5984e5 FF	202	struct flow_offload_entry *e;
da5984e5 FF	203
ac2a6666 PNA	204	rhashtable_remove_fast(&flow_table->rhashtable,
ac2a6666 PNA	205	&flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].node,
a268de77	206	nf_flow_offload_rhash_params);
ac2a6666 PNA	207	rhashtable_remove_fast(&flow_table->rhashtable,
ac2a6666 PNA	208	&flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].node,
a268de77	209	nf_flow_offload_rhash_params);
ac2a6666	210
da5984e5 FF	211	e = container_of(flow, struct flow_offload_entry, flow);
	212	clear_bit(IPS_OFFLOAD_BIT, &e->ct->status);
	213
0ff90b6c	214	flow_offload_free(flow);
ac2a6666	215	}
ac2a6666	216
59c466dd FF	217	void flow_offload_teardown(struct flow_offload *flow)
59c466dd FF	218	{
da5984e5 FF	219	struct flow_offload_entry *e;
da5984e5 FF	220
59c466dd	221	flow->flags \|= FLOW_OFFLOAD_TEARDOWN;
da5984e5 FF	222
	223	e = container_of(flow, struct flow_offload_entry, flow);
	224	flow_offload_fixup_ct_state(e->ct);
59c466dd FF	225	}
	226	EXPORT_SYMBOL_GPL(flow_offload_teardown);
	227
ac2a6666 PNA	228	struct flow_offload_tuple_rhash *
	229	flow_offload_lookup(struct nf_flowtable *flow_table,
	230	struct flow_offload_tuple *tuple)
	231	{
ba03137f FF	232	struct flow_offload_tuple_rhash *tuplehash;
	233	struct flow_offload *flow;
	234	int dir;
	235
	236	tuplehash = rhashtable_lookup_fast(&flow_table->rhashtable, tuple,
	237	nf_flow_offload_rhash_params);
	238	if (!tuplehash)
	239	return NULL;
	240
	241	dir = tuplehash->tuple.dir;
	242	flow = container_of(tuplehash, struct flow_offload, tuplehash[dir]);
	243	if (flow->flags & (FLOW_OFFLOAD_DYING \| FLOW_OFFLOAD_TEARDOWN))
	244	return NULL;
	245
	246	return tuplehash;
ac2a6666 PNA	247	}
	248	EXPORT_SYMBOL_GPL(flow_offload_lookup);
	249
ac2a6666 PNA	250	int nf_flow_table_iterate(struct nf_flowtable *flow_table,
	251	void (iter)(struct flow_offload flow, void *data),
	252	void *data)
	253	{
	254	struct flow_offload_tuple_rhash *tuplehash;
	255	struct rhashtable_iter hti;
	256	struct flow_offload *flow;
	257	int err;
	258
	259	err = rhashtable_walk_init(&flow_table->rhashtable, &hti, GFP_KERNEL);
	260	if (err)
	261	return err;
	262
	263	rhashtable_walk_start(&hti);
	264
	265	while ((tuplehash = rhashtable_walk_next(&hti))) {
	266	if (IS_ERR(tuplehash)) {
	267	err = PTR_ERR(tuplehash);
	268	if (err != -EAGAIN)
	269	goto out;
	270
	271	continue;
	272	}
	273	if (tuplehash->tuple.dir)
	274	continue;
	275
	276	flow = container_of(tuplehash, struct flow_offload, tuplehash[0]);
	277
	278	iter(flow, data);
	279	}
	280	out:
	281	rhashtable_walk_stop(&hti);
	282	rhashtable_walk_exit(&hti);
	283
	284	return err;
	285	}
	286	EXPORT_SYMBOL_GPL(nf_flow_table_iterate);
	287
	288	static inline bool nf_flow_has_expired(const struct flow_offload *flow)
	289	{
	290	return (__s32)(flow->timeout - (u32)jiffies) <= 0;
	291	}
	292
b408c5b0	293	static int nf_flow_offload_gc_step(struct nf_flowtable *flow_table)
ac2a6666 PNA	294	{
ac2a6666 PNA	295	struct flow_offload_tuple_rhash *tuplehash;
ac2a6666 PNA	296	struct rhashtable_iter hti;
	297	struct flow_offload *flow;
	298	int err;
	299
ac2a6666 PNA	300	err = rhashtable_walk_init(&flow_table->rhashtable, &hti, GFP_KERNEL);
ac2a6666 PNA	301	if (err)
b408c5b0	302	return 0;
ac2a6666 PNA	303
	304	rhashtable_walk_start(&hti);
	305
	306	while ((tuplehash = rhashtable_walk_next(&hti))) {
	307	if (IS_ERR(tuplehash)) {
	308	err = PTR_ERR(tuplehash);
	309	if (err != -EAGAIN)
	310	goto out;
	311
	312	continue;
	313	}
	314	if (tuplehash->tuple.dir)
	315	continue;
	316
	317	flow = container_of(tuplehash, struct flow_offload, tuplehash[0]);
	318
	319	if (nf_flow_has_expired(flow) \|\|
59c466dd FF	320	(flow->flags & (FLOW_OFFLOAD_DYING \|
59c466dd FF	321	FLOW_OFFLOAD_TEARDOWN)))
ac2a6666	322	flow_offload_del(flow_table, flow);
ac2a6666 PNA	323	}
	324	out:
	325	rhashtable_walk_stop(&hti);
	326	rhashtable_walk_exit(&hti);
b408c5b0 PNA	327
	328	return 1;
	329	}
	330
a268de77	331	static void nf_flow_offload_work_gc(struct work_struct *work)
b408c5b0 PNA	332	{
	333	struct nf_flowtable *flow_table;
	334
	335	flow_table = container_of(work, struct nf_flowtable, gc_work.work);
	336	nf_flow_offload_gc_step(flow_table);
ac2a6666 PNA	337	queue_delayed_work(system_power_efficient_wq, &flow_table->gc_work, HZ);
ac2a6666 PNA	338	}
ac2a6666 PNA	339
	340	static int nf_flow_nat_port_tcp(struct sk_buff *skb, unsigned int thoff,
	341	__be16 port, __be16 new_port)
	342	{
	343	struct tcphdr *tcph;
	344
	345	if (!pskb_may_pull(skb, thoff + sizeof(*tcph)) \|\|
	346	skb_try_make_writable(skb, thoff + sizeof(*tcph)))
	347	return -1;
	348
	349	tcph = (void *)(skb_network_header(skb) + thoff);
	350	inet_proto_csum_replace2(&tcph->check, skb, port, new_port, true);
	351
	352	return 0;
	353	}
	354
	355	static int nf_flow_nat_port_udp(struct sk_buff *skb, unsigned int thoff,
	356	__be16 port, __be16 new_port)
	357	{
	358	struct udphdr *udph;
	359
	360	if (!pskb_may_pull(skb, thoff + sizeof(*udph)) \|\|
	361	skb_try_make_writable(skb, thoff + sizeof(*udph)))
	362	return -1;
	363
	364	udph = (void *)(skb_network_header(skb) + thoff);
	365	if (udph->check \|\| skb->ip_summed == CHECKSUM_PARTIAL) {
	366	inet_proto_csum_replace2(&udph->check, skb, port,
	367	new_port, true);
	368	if (!udph->check)
	369	udph->check = CSUM_MANGLED_0;
	370	}
	371
	372	return 0;
	373	}
	374
	375	static int nf_flow_nat_port(struct sk_buff *skb, unsigned int thoff,
	376	u8 protocol, __be16 port, __be16 new_port)
	377	{
	378	switch (protocol) {
	379	case IPPROTO_TCP:
	380	if (nf_flow_nat_port_tcp(skb, thoff, port, new_port) < 0)
	381	return NF_DROP;
	382	break;
	383	case IPPROTO_UDP:
	384	if (nf_flow_nat_port_udp(skb, thoff, port, new_port) < 0)
	385	return NF_DROP;
	386	break;
	387	}
	388
	389	return 0;
	390	}
	391
	392	int nf_flow_snat_port(const struct flow_offload *flow,
	393	struct sk_buff *skb, unsigned int thoff,
	394	u8 protocol, enum flow_offload_tuple_dir dir)
	395	{
	396	struct flow_ports *hdr;
	397	__be16 port, new_port;
	398
	399	if (!pskb_may_pull(skb, thoff + sizeof(*hdr)) \|\|
	400	skb_try_make_writable(skb, thoff + sizeof(*hdr)))
	401	return -1;
	402
403	hdr = (void *)(skb_network_header(skb) + thoff);
404
405	switch (dir) {
406	case FLOW_OFFLOAD_DIR_ORIGINAL:
407	port = hdr->source;
408	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.dst_port;
409	hdr->source = new_port;
410	break;
411	case FLOW_OFFLOAD_DIR_REPLY:
412	port = hdr->dest;
413	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.src_port;
414	hdr->dest = new_port;
415	break;
416	default:
417	return -1;
418	}
419
420	return nf_flow_nat_port(skb, thoff, protocol, port, new_port);
421	}
422	EXPORT_SYMBOL_GPL(nf_flow_snat_port);
423
424	int nf_flow_dnat_port(const struct flow_offload *flow,
425	struct sk_buff *skb, unsigned int thoff,
426	u8 protocol, enum flow_offload_tuple_dir dir)
427	{
428	struct flow_ports *hdr;
429	__be16 port, new_port;
430
431	if (!pskb_may_pull(skb, thoff + sizeof(*hdr)) \|\|
432	skb_try_make_writable(skb, thoff + sizeof(*hdr)))
433	return -1;
434
435	hdr = (void *)(skb_network_header(skb) + thoff);
436
437	switch (dir) {
438	case FLOW_OFFLOAD_DIR_ORIGINAL:
439	port = hdr->dest;
440	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.src_port;
441	hdr->dest = new_port;
442	break;
443	case FLOW_OFFLOAD_DIR_REPLY:
444	port = hdr->source;
445	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.dst_port;
446	hdr->source = new_port;
447	break;
448	default:
449	return -1;
450	}
451
452	return nf_flow_nat_port(skb, thoff, protocol, port, new_port);
453	}
454	EXPORT_SYMBOL_GPL(nf_flow_dnat_port);
455
a268de77 FF	456	int nf_flow_table_init(struct nf_flowtable *flowtable)
	457	{
	458	int err;
	459
	460	INIT_DEFERRABLE_WORK(&flowtable->gc_work, nf_flow_offload_work_gc);
	461
	462	err = rhashtable_init(&flowtable->rhashtable,
	463	&nf_flow_offload_rhash_params);
	464	if (err < 0)
	465	return err;
	466
	467	queue_delayed_work(system_power_efficient_wq,
	468	&flowtable->gc_work, HZ);
	469
84453a90 FF	470	mutex_lock(&flowtable_lock);
	471	list_add(&flowtable->list, &flowtables);
	472	mutex_unlock(&flowtable_lock);
	473
a268de77 FF	474	return 0;
	475	}
	476	EXPORT_SYMBOL_GPL(nf_flow_table_init);
	477
c0ea1bcb PNA	478	static void nf_flow_table_do_cleanup(struct flow_offload flow, void data)
	479	{
	480	struct net_device *dev = data;
	481
59c466dd FF	482	if (!dev) {
59c466dd FF	483	flow_offload_teardown(flow);
c0ea1bcb	484	return;
59c466dd	485	}
c0ea1bcb	486
59c466dd FF	487	if (flow->tuplehash[0].tuple.iifidx == dev->ifindex \|\|
	488	flow->tuplehash[1].tuple.iifidx == dev->ifindex)
	489	flow_offload_dead(flow);
c0ea1bcb PNA	490	}
	491
	492	static void nf_flow_table_iterate_cleanup(struct nf_flowtable *flowtable,
84453a90	493	struct net_device *dev)
c0ea1bcb	494	{
84453a90	495	nf_flow_table_iterate(flowtable, nf_flow_table_do_cleanup, dev);
c0ea1bcb PNA	496	flush_delayed_work(&flowtable->gc_work);
	497	}
	498
	499	void nf_flow_table_cleanup(struct net net, struct net_device dev)
	500	{
84453a90 FF	501	struct nf_flowtable *flowtable;
	502
	503	mutex_lock(&flowtable_lock);
	504	list_for_each_entry(flowtable, &flowtables, list)
	505	nf_flow_table_iterate_cleanup(flowtable, dev);
	506	mutex_unlock(&flowtable_lock);
c0ea1bcb PNA	507	}
	508	EXPORT_SYMBOL_GPL(nf_flow_table_cleanup);
	509
b408c5b0 PNA	510	void nf_flow_table_free(struct nf_flowtable *flow_table)
b408c5b0 PNA	511	{
84453a90 FF	512	mutex_lock(&flowtable_lock);
	513	list_del(&flow_table->list);
	514	mutex_unlock(&flowtable_lock);
a268de77	515	cancel_delayed_work_sync(&flow_table->gc_work);
b408c5b0 PNA	516	nf_flow_table_iterate(flow_table, nf_flow_table_do_cleanup, NULL);
b408c5b0 PNA	517	WARN_ON(!nf_flow_offload_gc_step(flow_table));
a268de77	518	rhashtable_destroy(&flow_table->rhashtable);
b408c5b0 PNA	519	}
	520	EXPORT_SYMBOL_GPL(nf_flow_table_free);
	521
ac2a6666 PNA	522	MODULE_LICENSE("GPL");
ac2a6666 PNA	523	MODULE_AUTHOR("Pablo Neira Ayuso <pablo@netfilter.org>");