[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_l4proto.h>
#include <net/netfilter/nf_conntrack_tuple.h>

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

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

	ft->dir = dir;

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

struct flow_offload *flow_offload_alloc(struct nf_conn *ct)
{
	struct flow_offload *flow;

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

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

	flow->ct = ct;

	flow_offload_fill_dir(flow, ct, FLOW_OFFLOAD_DIR_ORIGINAL);
	flow_offload_fill_dir(flow, ct, 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_ct_refcnt:
	nf_ct_put(ct);

	return NULL;
}
EXPORT_SYMBOL_GPL(flow_offload_alloc);

static int flow_offload_fill_route(struct flow_offload *flow,
				   const struct nf_flow_route *route,
				   enum flow_offload_tuple_dir dir)
{
	struct flow_offload_tuple *flow_tuple = &flow->tuplehash[dir].tuple;
	struct dst_entry *other_dst = route->tuple[!dir].dst;
	struct dst_entry *dst = route->tuple[dir].dst;

	if (!dst_hold_safe(route->tuple[dir].dst))
		return -1;

	switch (flow_tuple->l3proto) {
	case NFPROTO_IPV4:
		flow_tuple->mtu = ip_dst_mtu_maybe_forward(dst, true);
		break;
	case NFPROTO_IPV6:
		flow_tuple->mtu = ip6_dst_mtu_forward(dst);
		break;
	}

	flow_tuple->iifidx = other_dst->dev->ifindex;
	flow_tuple->dst_cache = dst;

	return 0;
}

int flow_offload_route_init(struct flow_offload *flow,
			    const struct nf_flow_route *route)
{
	int err;

	err = flow_offload_fill_route(flow, route, FLOW_OFFLOAD_DIR_ORIGINAL);
	if (err < 0)
		return err;

	err = flow_offload_fill_route(flow, route, FLOW_OFFLOAD_DIR_REPLY);
	if (err < 0)
		goto err_route_reply;

	flow->type = NF_FLOW_OFFLOAD_ROUTE;

	return 0;

err_route_reply:
	dst_release(route->tuple[FLOW_OFFLOAD_DIR_ORIGINAL].dst);

	return err;
}
EXPORT_SYMBOL_GPL(flow_offload_route_init);

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 inline __s32 nf_flow_timeout_delta(unsigned int timeout)
{
	return (__s32)(timeout - (u32)jiffies);
}

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

	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;

	if (nf_flow_timeout_delta(ct->timeout) > (__s32)timeout)
		ct->timeout = nfct_time_stamp + timeout;
}

static void flow_offload_fixup_ct_state(struct nf_conn *ct)
{
	if (nf_ct_protonum(ct) == IPPROTO_TCP)
		flow_offload_fixup_tcp(&ct->proto.tcp);
}

static void flow_offload_fixup_ct(struct nf_conn *ct)
{
	flow_offload_fixup_ct_state(ct);
	flow_offload_fixup_ct_timeout(ct);
}

static void flow_offload_route_release(struct flow_offload *flow)
{
	dst_release(flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.dst_cache);
	dst_release(flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.dst_cache);
}

void flow_offload_free(struct flow_offload *flow)
{
	switch (flow->type) {
	case NF_FLOW_OFFLOAD_ROUTE:
		flow_offload_route_release(flow);
		break;
	default:
		break;
	}
	if (flow->flags & FLOW_OFFLOAD_DYING)
		nf_ct_delete(flow->ct, 0, 0);
	nf_ct_put(flow->ct);
	kfree_rcu(flow, 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;

	flow->timeout = (u32)jiffies + NF_FLOW_TIMEOUT;

	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;
	}

	return 0;
}
EXPORT_SYMBOL_GPL(flow_offload_add);

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

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

	clear_bit(IPS_OFFLOAD_BIT, &flow->ct->status);

	if (nf_flow_has_expired(flow))
		flow_offload_fixup_ct(flow->ct);
	else if (flow->flags & FLOW_OFFLOAD_TEARDOWN)
		flow_offload_fixup_ct_timeout(flow->ct);

	flow_offload_free(flow);
}

void flow_offload_teardown(struct flow_offload *flow)
{
	flow->flags |= FLOW_OFFLOAD_TEARDOWN;

	flow_offload_fixup_ct_state(flow->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(&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;

	if (unlikely(nf_ct_is_dying(flow->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;

	if (nf_flow_has_expired(flow) || nf_ct_is_dying(flow->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;

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

	if (net_eq(nf_ct_net(flow->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>
40d102cd	14	#include <net/netfilter/nf_conntrack_l4proto.h>
ac2a6666 PNA	15	#include <net/netfilter/nf_conntrack_tuple.h>
ac2a6666 PNA	16
84453a90 FF	17	static DEFINE_MUTEX(flowtable_lock);
	18	static LIST_HEAD(flowtables);
	19
047b300e FF	20	static void
047b300e FF	21	flow_offload_fill_dir(struct flow_offload flow, struct nf_conn ct,
047b300e FF	22	enum flow_offload_tuple_dir dir)
	23	{
	24	struct flow_offload_tuple *ft = &flow->tuplehash[dir].tuple;
	25	struct nf_conntrack_tuple *ctt = &ct->tuplehash[dir].tuple;
	26
	27	ft->dir = dir;
	28
	29	switch (ctt->src.l3num) {
	30	case NFPROTO_IPV4:
	31	ft->src_v4 = ctt->src.u3.in;
	32	ft->dst_v4 = ctt->dst.u3.in;
	33	break;
	34	case NFPROTO_IPV6:
	35	ft->src_v6 = ctt->src.u3.in6;
	36	ft->dst_v6 = ctt->dst.u3.in6;
	37	break;
	38	}
	39
	40	ft->l3proto = ctt->src.l3num;
	41	ft->l4proto = ctt->dst.protonum;
	42	ft->src_port = ctt->src.u.tcp.port;
	43	ft->dst_port = ctt->dst.u.tcp.port;
047b300e FF	44	}
047b300e FF	45
f1363e05	46	struct flow_offload flow_offload_alloc(struct nf_conn ct)
ac2a6666	47	{
ac2a6666 PNA	48	struct flow_offload *flow;
	49
	50	if (unlikely(nf_ct_is_dying(ct) \|\|
	51	!atomic_inc_not_zero(&ct->ct_general.use)))
	52	return NULL;
	53
62248df8 PNA	54	flow = kzalloc(sizeof(*flow), GFP_ATOMIC);
62248df8 PNA	55	if (!flow)
ac2a6666 PNA	56	goto err_ct_refcnt;
ac2a6666 PNA	57
b32d2f34	58	flow->ct = ct;
ac2a6666	59
f1363e05 PNA	60	flow_offload_fill_dir(flow, ct, FLOW_OFFLOAD_DIR_ORIGINAL);
f1363e05 PNA	61	flow_offload_fill_dir(flow, ct, FLOW_OFFLOAD_DIR_REPLY);
ac2a6666 PNA	62
	63	if (ct->status & IPS_SRC_NAT)
	64	flow->flags \|= FLOW_OFFLOAD_SNAT;
df1e2025	65	if (ct->status & IPS_DST_NAT)
ac2a6666 PNA	66	flow->flags \|= FLOW_OFFLOAD_DNAT;
	67
	68	return flow;
	69
ac2a6666 PNA	70	err_ct_refcnt:
	71	nf_ct_put(ct);
	72
	73	return NULL;
	74	}
	75	EXPORT_SYMBOL_GPL(flow_offload_alloc);
	76
f1363e05 PNA	77	static int flow_offload_fill_route(struct flow_offload *flow,
	78	const struct nf_flow_route *route,
	79	enum flow_offload_tuple_dir dir)
	80	{
	81	struct flow_offload_tuple *flow_tuple = &flow->tuplehash[dir].tuple;
	82	struct dst_entry *other_dst = route->tuple[!dir].dst;
	83	struct dst_entry *dst = route->tuple[dir].dst;
	84
	85	if (!dst_hold_safe(route->tuple[dir].dst))
	86	return -1;
	87
	88	switch (flow_tuple->l3proto) {
	89	case NFPROTO_IPV4:
	90	flow_tuple->mtu = ip_dst_mtu_maybe_forward(dst, true);
	91	break;
	92	case NFPROTO_IPV6:
	93	flow_tuple->mtu = ip6_dst_mtu_forward(dst);
	94	break;
	95	}
	96
	97	flow_tuple->iifidx = other_dst->dev->ifindex;
	98	flow_tuple->dst_cache = dst;
	99
	100	return 0;
	101	}
	102
	103	int flow_offload_route_init(struct flow_offload *flow,
	104	const struct nf_flow_route *route)
	105	{
	106	int err;
	107
	108	err = flow_offload_fill_route(flow, route, FLOW_OFFLOAD_DIR_ORIGINAL);
	109	if (err < 0)
	110	return err;
	111
	112	err = flow_offload_fill_route(flow, route, FLOW_OFFLOAD_DIR_REPLY);
	113	if (err < 0)
	114	goto err_route_reply;
	115
	116	flow->type = NF_FLOW_OFFLOAD_ROUTE;
	117
	118	return 0;
	119
	120	err_route_reply:
	121	dst_release(route->tuple[FLOW_OFFLOAD_DIR_ORIGINAL].dst);
	122
	123	return err;
	124	}
	125	EXPORT_SYMBOL_GPL(flow_offload_route_init);
	126
da5984e5 FF	127	static void flow_offload_fixup_tcp(struct ip_ct_tcp *tcp)
	128	{
	129	tcp->state = TCP_CONNTRACK_ESTABLISHED;
	130	tcp->seen[0].td_maxwin = 0;
	131	tcp->seen[1].td_maxwin = 0;
	132	}
	133
e97d9404 FW	134	#define NF_FLOWTABLE_TCP_PICKUP_TIMEOUT (120 * HZ)
	135	#define NF_FLOWTABLE_UDP_PICKUP_TIMEOUT (30 * HZ)
	136
1e5b2471 PNA	137	static inline __s32 nf_flow_timeout_delta(unsigned int timeout)
	138	{
	139	return (__s32)(timeout - (u32)jiffies);
	140	}
	141
	142	static void flow_offload_fixup_ct_timeout(struct nf_conn *ct)
da5984e5 FF	143	{
da5984e5 FF	144	const struct nf_conntrack_l4proto *l4proto;
1e5b2471	145	int l4num = nf_ct_protonum(ct);
da5984e5	146	unsigned int timeout;
da5984e5	147
4a60dc74	148	l4proto = nf_ct_l4proto_find(l4num);
da5984e5 FF	149	if (!l4proto)
	150	return;
	151
da5984e5	152	if (l4num == IPPROTO_TCP)
e97d9404	153	timeout = NF_FLOWTABLE_TCP_PICKUP_TIMEOUT;
da5984e5	154	else if (l4num == IPPROTO_UDP)
e97d9404	155	timeout = NF_FLOWTABLE_UDP_PICKUP_TIMEOUT;
da5984e5 FF	156	else
	157	return;
	158
1e5b2471 PNA	159	if (nf_flow_timeout_delta(ct->timeout) > (__s32)timeout)
	160	ct->timeout = nfct_time_stamp + timeout;
	161	}
	162
	163	static void flow_offload_fixup_ct_state(struct nf_conn *ct)
	164	{
	165	if (nf_ct_protonum(ct) == IPPROTO_TCP)
	166	flow_offload_fixup_tcp(&ct->proto.tcp);
	167	}
	168
	169	static void flow_offload_fixup_ct(struct nf_conn *ct)
	170	{
	171	flow_offload_fixup_ct_state(ct);
	172	flow_offload_fixup_ct_timeout(ct);
da5984e5 FF	173	}
da5984e5 FF	174
f1363e05	175	static void flow_offload_route_release(struct flow_offload *flow)
ac2a6666	176	{
ac2a6666 PNA	177	dst_release(flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.dst_cache);
ac2a6666 PNA	178	dst_release(flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.dst_cache);
f1363e05 PNA	179	}
	180
	181	void flow_offload_free(struct flow_offload *flow)
	182	{
	183	switch (flow->type) {
	184	case NF_FLOW_OFFLOAD_ROUTE:
	185	flow_offload_route_release(flow);
	186	break;
	187	default:
	188	break;
	189	}
da5984e5	190	if (flow->flags & FLOW_OFFLOAD_DYING)
b32d2f34 PNA	191	nf_ct_delete(flow->ct, 0, 0);
b32d2f34 PNA	192	nf_ct_put(flow->ct);
62248df8	193	kfree_rcu(flow, rcu_head);
ac2a6666 PNA	194	}
	195	EXPORT_SYMBOL_GPL(flow_offload_free);
	196
a268de77 FF	197	static u32 flow_offload_hash(const void *data, u32 len, u32 seed)
	198	{
	199	const struct flow_offload_tuple *tuple = data;
	200
	201	return jhash(tuple, offsetof(struct flow_offload_tuple, dir), seed);
	202	}
	203
	204	static u32 flow_offload_hash_obj(const void *data, u32 len, u32 seed)
	205	{
	206	const struct flow_offload_tuple_rhash *tuplehash = data;
	207
	208	return jhash(&tuplehash->tuple, offsetof(struct flow_offload_tuple, dir), seed);
	209	}
	210
	211	static int flow_offload_hash_cmp(struct rhashtable_compare_arg *arg,
	212	const void *ptr)
	213	{
	214	const struct flow_offload_tuple *tuple = arg->key;
	215	const struct flow_offload_tuple_rhash *x = ptr;
	216
	217	if (memcmp(&x->tuple, tuple, offsetof(struct flow_offload_tuple, dir)))
	218	return 1;
	219
	220	return 0;
	221	}
	222
	223	static const struct rhashtable_params nf_flow_offload_rhash_params = {
	224	.head_offset = offsetof(struct flow_offload_tuple_rhash, node),
	225	.hashfn = flow_offload_hash,
	226	.obj_hashfn = flow_offload_hash_obj,
	227	.obj_cmpfn = flow_offload_hash_cmp,
	228	.automatic_shrinking = true,
	229	};
	230
ac2a6666 PNA	231	int flow_offload_add(struct nf_flowtable flow_table, struct flow_offload flow)
ac2a6666 PNA	232	{
43c8f131 TY	233	int err;
43c8f131 TY	234
daf61b02 PNA	235	flow->timeout = (u32)jiffies + NF_FLOW_TIMEOUT;
daf61b02 PNA	236
43c8f131 TY	237	err = rhashtable_insert_fast(&flow_table->rhashtable,
	238	&flow->tuplehash[0].node,
	239	nf_flow_offload_rhash_params);
	240	if (err < 0)
	241	return err;
	242
	243	err = rhashtable_insert_fast(&flow_table->rhashtable,
	244	&flow->tuplehash[1].node,
	245	nf_flow_offload_rhash_params);
	246	if (err < 0) {
	247	rhashtable_remove_fast(&flow_table->rhashtable,
	248	&flow->tuplehash[0].node,
	249	nf_flow_offload_rhash_params);
	250	return err;
	251	}
ac2a6666	252
ac2a6666 PNA	253	return 0;
	254	}
	255	EXPORT_SYMBOL_GPL(flow_offload_add);
	256
3e68db2f PNA	257	static inline bool nf_flow_has_expired(const struct flow_offload *flow)
3e68db2f PNA	258	{
1e5b2471	259	return nf_flow_timeout_delta(flow->timeout) <= 0;
3e68db2f PNA	260	}
3e68db2f PNA	261
0ff90b6c FF	262	static void flow_offload_del(struct nf_flowtable *flow_table,
0ff90b6c FF	263	struct flow_offload *flow)
ac2a6666	264	{
ac2a6666 PNA	265	rhashtable_remove_fast(&flow_table->rhashtable,
ac2a6666 PNA	266	&flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].node,
a268de77	267	nf_flow_offload_rhash_params);
ac2a6666 PNA	268	rhashtable_remove_fast(&flow_table->rhashtable,
ac2a6666 PNA	269	&flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].node,
a268de77	270	nf_flow_offload_rhash_params);
ac2a6666	271
b32d2f34	272	clear_bit(IPS_OFFLOAD_BIT, &flow->ct->status);
da5984e5	273
3e68db2f	274	if (nf_flow_has_expired(flow))
b32d2f34	275	flow_offload_fixup_ct(flow->ct);
1e5b2471	276	else if (flow->flags & FLOW_OFFLOAD_TEARDOWN)
b32d2f34	277	flow_offload_fixup_ct_timeout(flow->ct);
3e68db2f	278
0ff90b6c	279	flow_offload_free(flow);
ac2a6666	280	}
ac2a6666	281
59c466dd FF	282	void flow_offload_teardown(struct flow_offload *flow)
	283	{
	284	flow->flags \|= FLOW_OFFLOAD_TEARDOWN;
da5984e5	285
b32d2f34	286	flow_offload_fixup_ct_state(flow->ct);
59c466dd FF	287	}
	288	EXPORT_SYMBOL_GPL(flow_offload_teardown);
	289
ac2a6666 PNA	290	struct flow_offload_tuple_rhash *
	291	flow_offload_lookup(struct nf_flowtable *flow_table,
	292	struct flow_offload_tuple *tuple)
	293	{
ba03137f FF	294	struct flow_offload_tuple_rhash *tuplehash;
	295	struct flow_offload *flow;
	296	int dir;
	297
a2d88182 TY	298	tuplehash = rhashtable_lookup(&flow_table->rhashtable, tuple,
a2d88182 TY	299	nf_flow_offload_rhash_params);
ba03137f FF	300	if (!tuplehash)
	301	return NULL;
	302
	303	dir = tuplehash->tuple.dir;
	304	flow = container_of(tuplehash, struct flow_offload, tuplehash[dir]);
	305	if (flow->flags & (FLOW_OFFLOAD_DYING \| FLOW_OFFLOAD_TEARDOWN))
	306	return NULL;
	307
b32d2f34	308	if (unlikely(nf_ct_is_dying(flow->ct)))
8cd2bc98 TY	309	return NULL;
8cd2bc98 TY	310
ba03137f	311	return tuplehash;
ac2a6666 PNA	312	}
	313	EXPORT_SYMBOL_GPL(flow_offload_lookup);
	314
49de9c09 TY	315	static int
	316	nf_flow_table_iterate(struct nf_flowtable *flow_table,
	317	void (iter)(struct flow_offload flow, void *data),
	318	void *data)
ac2a6666 PNA	319	{
	320	struct flow_offload_tuple_rhash *tuplehash;
	321	struct rhashtable_iter hti;
	322	struct flow_offload *flow;
0de22baa	323	int err = 0;
ac2a6666	324
0de22baa	325	rhashtable_walk_enter(&flow_table->rhashtable, &hti);
ac2a6666 PNA	326	rhashtable_walk_start(&hti);
	327
	328	while ((tuplehash = rhashtable_walk_next(&hti))) {
	329	if (IS_ERR(tuplehash)) {
0de22baa TY	330	if (PTR_ERR(tuplehash) != -EAGAIN) {
	331	err = PTR_ERR(tuplehash);
	332	break;
	333	}
ac2a6666 PNA	334	continue;
	335	}
	336	if (tuplehash->tuple.dir)
	337	continue;
	338
	339	flow = container_of(tuplehash, struct flow_offload, tuplehash[0]);
	340
	341	iter(flow, data);
	342	}
ac2a6666 PNA	343	rhashtable_walk_stop(&hti);
	344	rhashtable_walk_exit(&hti);
	345
	346	return err;
	347	}
ac2a6666	348
b9660987	349	static void nf_flow_offload_gc_step(struct flow_offload flow, void data)
ac2a6666	350	{
b9660987	351	struct nf_flowtable *flow_table = data;
ac2a6666	352
b32d2f34	353	if (nf_flow_has_expired(flow) \|\| nf_ct_is_dying(flow->ct) \|\|
b9660987 TY	354	(flow->flags & (FLOW_OFFLOAD_DYING \| FLOW_OFFLOAD_TEARDOWN)))
b9660987 TY	355	flow_offload_del(flow_table, flow);
b408c5b0 PNA	356	}
b408c5b0 PNA	357
a268de77	358	static void nf_flow_offload_work_gc(struct work_struct *work)
b408c5b0 PNA	359	{
	360	struct nf_flowtable *flow_table;
	361
	362	flow_table = container_of(work, struct nf_flowtable, gc_work.work);
b9660987	363	nf_flow_table_iterate(flow_table, nf_flow_offload_gc_step, flow_table);
ac2a6666 PNA	364	queue_delayed_work(system_power_efficient_wq, &flow_table->gc_work, HZ);
ac2a6666 PNA	365	}
ac2a6666 PNA	366
	367	static int nf_flow_nat_port_tcp(struct sk_buff *skb, unsigned int thoff,
	368	__be16 port, __be16 new_port)
	369	{
	370	struct tcphdr *tcph;
	371
	372	if (!pskb_may_pull(skb, thoff + sizeof(*tcph)) \|\|
	373	skb_try_make_writable(skb, thoff + sizeof(*tcph)))
	374	return -1;
	375
	376	tcph = (void *)(skb_network_header(skb) + thoff);
	377	inet_proto_csum_replace2(&tcph->check, skb, port, new_port, true);
	378
	379	return 0;
	380	}
	381
	382	static int nf_flow_nat_port_udp(struct sk_buff *skb, unsigned int thoff,
	383	__be16 port, __be16 new_port)
	384	{
	385	struct udphdr *udph;
	386
	387	if (!pskb_may_pull(skb, thoff + sizeof(*udph)) \|\|
	388	skb_try_make_writable(skb, thoff + sizeof(*udph)))
	389	return -1;
	390
	391	udph = (void *)(skb_network_header(skb) + thoff);
	392	if (udph->check \|\| skb->ip_summed == CHECKSUM_PARTIAL) {
	393	inet_proto_csum_replace2(&udph->check, skb, port,
	394	new_port, true);
	395	if (!udph->check)
	396	udph->check = CSUM_MANGLED_0;
	397	}
	398
	399	return 0;
	400	}
	401
	402	static int nf_flow_nat_port(struct sk_buff *skb, unsigned int thoff,
	403	u8 protocol, __be16 port, __be16 new_port)
	404	{
	405	switch (protocol) {
	406	case IPPROTO_TCP:
	407	if (nf_flow_nat_port_tcp(skb, thoff, port, new_port) < 0)
	408	return NF_DROP;
	409	break;
	410	case IPPROTO_UDP:
	411	if (nf_flow_nat_port_udp(skb, thoff, port, new_port) < 0)
	412	return NF_DROP;
	413	break;
	414	}
	415
	416	return 0;
	417	}
	418
	419	int nf_flow_snat_port(const struct flow_offload *flow,
	420	struct sk_buff *skb, unsigned int thoff,
	421	u8 protocol, enum flow_offload_tuple_dir dir)
	422	{
	423	struct flow_ports *hdr;
	424	__be16 port, new_port;
	425
	426	if (!pskb_may_pull(skb, thoff + sizeof(*hdr)) \|\|
	427	skb_try_make_writable(skb, thoff + sizeof(*hdr)))
	428	return -1;
	429
430	hdr = (void *)(skb_network_header(skb) + thoff);
431
432	switch (dir) {
433	case FLOW_OFFLOAD_DIR_ORIGINAL:
434	port = hdr->source;
435	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.dst_port;
436	hdr->source = new_port;
437	break;
438	case FLOW_OFFLOAD_DIR_REPLY:
439	port = hdr->dest;
440	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.src_port;
441	hdr->dest = new_port;
442	break;
443	default:
444	return -1;
445	}
446
447	return nf_flow_nat_port(skb, thoff, protocol, port, new_port);
448	}
449	EXPORT_SYMBOL_GPL(nf_flow_snat_port);
450
451	int nf_flow_dnat_port(const struct flow_offload *flow,
452	struct sk_buff *skb, unsigned int thoff,
453	u8 protocol, enum flow_offload_tuple_dir dir)
454	{
455	struct flow_ports *hdr;
456	__be16 port, new_port;
457
458	if (!pskb_may_pull(skb, thoff + sizeof(*hdr)) \|\|
459	skb_try_make_writable(skb, thoff + sizeof(*hdr)))
460	return -1;
461
462	hdr = (void *)(skb_network_header(skb) + thoff);
463
464	switch (dir) {
465	case FLOW_OFFLOAD_DIR_ORIGINAL:
466	port = hdr->dest;
467	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.src_port;
468	hdr->dest = new_port;
469	break;
470	case FLOW_OFFLOAD_DIR_REPLY:
471	port = hdr->source;
472	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.dst_port;
473	hdr->source = new_port;
474	break;
475	default:
476	return -1;
477	}
478
479	return nf_flow_nat_port(skb, thoff, protocol, port, new_port);
480	}
481	EXPORT_SYMBOL_GPL(nf_flow_dnat_port);
482
a268de77 FF	483	int nf_flow_table_init(struct nf_flowtable *flowtable)
	484	{
	485	int err;
	486
	487	INIT_DEFERRABLE_WORK(&flowtable->gc_work, nf_flow_offload_work_gc);
	488
	489	err = rhashtable_init(&flowtable->rhashtable,
	490	&nf_flow_offload_rhash_params);
	491	if (err < 0)
	492	return err;
	493
	494	queue_delayed_work(system_power_efficient_wq,
	495	&flowtable->gc_work, HZ);
	496
84453a90 FF	497	mutex_lock(&flowtable_lock);
	498	list_add(&flowtable->list, &flowtables);
	499	mutex_unlock(&flowtable_lock);
	500
a268de77 FF	501	return 0;
	502	}
	503	EXPORT_SYMBOL_GPL(nf_flow_table_init);
	504
c0ea1bcb PNA	505	static void nf_flow_table_do_cleanup(struct flow_offload flow, void data)
	506	{
	507	struct net_device *dev = data;
	508
59c466dd FF	509	if (!dev) {
59c466dd FF	510	flow_offload_teardown(flow);
c0ea1bcb	511	return;
59c466dd	512	}
b32d2f34 PNA	513
b32d2f34 PNA	514	if (net_eq(nf_ct_net(flow->ct), dev_net(dev)) &&
a3fb3698 TY	515	(flow->tuplehash[0].tuple.iifidx == dev->ifindex \|\|
a3fb3698 TY	516	flow->tuplehash[1].tuple.iifidx == dev->ifindex))
59c466dd	517	flow_offload_dead(flow);
c0ea1bcb PNA	518	}
	519
	520	static void nf_flow_table_iterate_cleanup(struct nf_flowtable *flowtable,
84453a90	521	struct net_device *dev)
c0ea1bcb	522	{
84453a90	523	nf_flow_table_iterate(flowtable, nf_flow_table_do_cleanup, dev);
c0ea1bcb PNA	524	flush_delayed_work(&flowtable->gc_work);
	525	}
	526
5f1be84a	527	void nf_flow_table_cleanup(struct net_device *dev)
c0ea1bcb	528	{
84453a90 FF	529	struct nf_flowtable *flowtable;
	530
	531	mutex_lock(&flowtable_lock);
	532	list_for_each_entry(flowtable, &flowtables, list)
	533	nf_flow_table_iterate_cleanup(flowtable, dev);
	534	mutex_unlock(&flowtable_lock);
c0ea1bcb PNA	535	}
	536	EXPORT_SYMBOL_GPL(nf_flow_table_cleanup);
	537
b408c5b0 PNA	538	void nf_flow_table_free(struct nf_flowtable *flow_table)
b408c5b0 PNA	539	{
84453a90 FF	540	mutex_lock(&flowtable_lock);
	541	list_del(&flow_table->list);
	542	mutex_unlock(&flowtable_lock);
a268de77	543	cancel_delayed_work_sync(&flow_table->gc_work);
b408c5b0	544	nf_flow_table_iterate(flow_table, nf_flow_table_do_cleanup, NULL);
b9660987	545	nf_flow_table_iterate(flow_table, nf_flow_offload_gc_step, flow_table);
a268de77	546	rhashtable_destroy(&flow_table->rhashtable);
b408c5b0 PNA	547	}
	548	EXPORT_SYMBOL_GPL(nf_flow_table_free);
	549
ac2a6666 PNA	550	MODULE_LICENSE("GPL");
ac2a6666 PNA	551	MODULE_AUTHOR("Pablo Neira Ayuso <pablo@netfilter.org>");