[linux-block.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,
		      enum flow_offload_tuple_dir dir)
{
	struct flow_offload_tuple *ft = &flow->tuplehash[dir].tuple;
	struct nf_conntrack_tuple *ctt = &flow->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, FLOW_OFFLOAD_DIR_ORIGINAL);
	flow_offload_fill_dir(flow, FLOW_OFFLOAD_DIR_REPLY);

	if (ct->status & IPS_SRC_NAT)
		__set_bit(NF_FLOW_SNAT, &flow->flags);
	if (ct->status & IPS_DST_NAT)
		__set_bit(NF_FLOW_DNAT, &flow->flags);

	return flow;

err_ct_refcnt:
	nf_ct_put(ct);

	return NULL;
}
EXPORT_SYMBOL_GPL(flow_offload_alloc);

static u32 flow_offload_dst_cookie(struct flow_offload_tuple *flow_tuple)
{
	const struct rt6_info *rt;

	if (flow_tuple->l3proto == NFPROTO_IPV6) {
		rt = (const struct rt6_info *)flow_tuple->dst_cache;
		return rt6_get_cookie(rt);
	}

	return 0;
}

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 *dst = route->tuple[dir].dst;
	int i, j = 0;

	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 = route->tuple[dir].in.ifindex;
	for (i = route->tuple[dir].in.num_encaps - 1; i >= 0; i--) {
		flow_tuple->encap[j].id = route->tuple[dir].in.encap[i].id;
		flow_tuple->encap[j].proto = route->tuple[dir].in.encap[i].proto;
		if (route->tuple[dir].in.ingress_vlans & BIT(i))
			flow_tuple->in_vlan_ingress |= BIT(j);
		j++;
	}
	flow_tuple->encap_num = route->tuple[dir].in.num_encaps;

	switch (route->tuple[dir].xmit_type) {
	case FLOW_OFFLOAD_XMIT_DIRECT:
		memcpy(flow_tuple->out.h_dest, route->tuple[dir].out.h_dest,
		       ETH_ALEN);
		memcpy(flow_tuple->out.h_source, route->tuple[dir].out.h_source,
		       ETH_ALEN);
		flow_tuple->out.ifidx = route->tuple[dir].out.ifindex;
		flow_tuple->out.hw_ifidx = route->tuple[dir].out.hw_ifindex;
		break;
	case FLOW_OFFLOAD_XMIT_XFRM:
	case FLOW_OFFLOAD_XMIT_NEIGH:
		if (!dst_hold_safe(route->tuple[dir].dst))
			return -1;

		flow_tuple->dst_cache = dst;
		flow_tuple->dst_cookie = flow_offload_dst_cookie(flow_tuple);
		break;
	default:
		WARN_ON_ONCE(1);
		break;
	}
	flow_tuple->xmit_type = route->tuple[dir].xmit_type;

	return 0;
}

static void nft_flow_dst_release(struct flow_offload *flow,
				 enum flow_offload_tuple_dir dir)
{
	if (flow->tuplehash[dir].tuple.xmit_type == FLOW_OFFLOAD_XMIT_NEIGH ||
	    flow->tuplehash[dir].tuple.xmit_type == FLOW_OFFLOAD_XMIT_XFRM)
		dst_release(flow->tuplehash[dir].tuple.dst_cache);
}

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:
	nft_flow_dst_release(flow, FLOW_OFFLOAD_DIR_ORIGINAL);

	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 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)
{
	nft_flow_dst_release(flow, FLOW_OFFLOAD_DIR_ORIGINAL);
	nft_flow_dst_release(flow, FLOW_OFFLOAD_DIR_REPLY);
}

void flow_offload_free(struct flow_offload *flow)
{
	switch (flow->type) {
	case NF_FLOW_OFFLOAD_ROUTE:
		flow_offload_route_release(flow);
		break;
	default:
		break;
	}
	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, __hash), 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, __hash), 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, __hash)))
		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 = nf_flowtable_time_stamp + 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;
	}

	nf_ct_offload_timeout(flow->ct);

	if (nf_flowtable_hw_offload(flow_table)) {
		__set_bit(NF_FLOW_HW, &flow->flags);
		nf_flow_offload_add(flow_table, flow);
	}

	return 0;
}
EXPORT_SYMBOL_GPL(flow_offload_add);

void flow_offload_refresh(struct nf_flowtable *flow_table,
			  struct flow_offload *flow)
{
	flow->timeout = nf_flowtable_time_stamp + NF_FLOW_TIMEOUT;

	if (likely(!nf_flowtable_hw_offload(flow_table) ||
		   !test_and_clear_bit(NF_FLOW_HW_REFRESH, &flow->flags)))
		return;

	nf_flow_offload_add(flow_table, flow);
}
EXPORT_SYMBOL_GPL(flow_offload_refresh);

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
		flow_offload_fixup_ct_timeout(flow->ct);

	flow_offload_free(flow);
}

void flow_offload_teardown(struct flow_offload *flow)
{
	set_bit(NF_FLOW_TEARDOWN, &flow->flags);

	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 (test_bit(NF_FLOW_TEARDOWN, &flow->flags))
		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 bool flow_offload_stale_dst(struct flow_offload_tuple *tuple)
{
	struct dst_entry *dst;

	if (tuple->xmit_type == FLOW_OFFLOAD_XMIT_NEIGH ||
	    tuple->xmit_type == FLOW_OFFLOAD_XMIT_XFRM) {
		dst = tuple->dst_cache;
		if (!dst_check(dst, tuple->dst_cookie))
			return true;
	}

	return false;
}

static bool nf_flow_has_stale_dst(struct flow_offload *flow)
{
	return flow_offload_stale_dst(&flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple) ||
	       flow_offload_stale_dst(&flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple);
}

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) ||
	    nf_flow_has_stale_dst(flow))
		set_bit(NF_FLOW_TEARDOWN, &flow->flags);

	if (test_bit(NF_FLOW_TEARDOWN, &flow->flags)) {
		if (test_bit(NF_FLOW_HW, &flow->flags)) {
			if (!test_bit(NF_FLOW_HW_DYING, &flow->flags))
				nf_flow_offload_del(flow_table, flow);
			else if (test_bit(NF_FLOW_HW_DEAD, &flow->flags))
				flow_offload_del(flow_table, flow);
		} else {
			flow_offload_del(flow_table, flow);
		}
	} else if (test_bit(NF_FLOW_HW, &flow->flags)) {
		nf_flow_offload_stats(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 void nf_flow_nat_port_tcp(struct sk_buff *skb, unsigned int thoff,
				 __be16 port, __be16 new_port)
{
	struct tcphdr *tcph;

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

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

	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, false);
		if (!udph->check)
			udph->check = CSUM_MANGLED_0;
	}
}

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

void 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;

	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;
	}

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

void 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;

	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;
	}

	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_DELAYED_WORK(&flowtable->gc_work, nf_flow_offload_work_gc);
	flow_block_init(&flowtable->flow_block);
	init_rwsem(&flowtable->flow_block_lock);

	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_teardown(flow);
}

void nf_flow_table_gc_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);
	nf_flow_table_offload_flush(flowtable);
}

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_gc_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);
	nf_flow_table_offload_flush(flow_table);
	if (nf_flowtable_hw_offload(flow_table))
		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);

static int __init nf_flow_table_module_init(void)
{
	return nf_flow_table_offload_init();
}

static void __exit nf_flow_table_module_exit(void)
{
	nf_flow_table_offload_exit();
}

module_init(nf_flow_table_module_init);
module_exit(nf_flow_table_module_exit);

MODULE_LICENSE("GPL");
MODULE_AUTHOR("Pablo Neira Ayuso <pablo@netfilter.org>");
MODULE_DESCRIPTION("Netfilter flow table module");
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	20	static void
458a1828	21	flow_offload_fill_dir(struct flow_offload *flow,
047b300e FF	22	enum flow_offload_tuple_dir dir)
	23	{
	24	struct flow_offload_tuple *ft = &flow->tuplehash[dir].tuple;
458a1828	25	struct nf_conntrack_tuple *ctt = &flow->ct->tuplehash[dir].tuple;
047b300e FF	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
458a1828	60	flow_offload_fill_dir(flow, FLOW_OFFLOAD_DIR_ORIGINAL);
458a1828	61	flow_offload_fill_dir(flow, FLOW_OFFLOAD_DIR_REPLY);
ac2a6666 PNA	62
ac2a6666 PNA	63	if (ct->status & IPS_SRC_NAT)
355a8b13	64	__set_bit(NF_FLOW_SNAT, &flow->flags);
df1e2025	65	if (ct->status & IPS_DST_NAT)
355a8b13	66	__set_bit(NF_FLOW_DNAT, &flow->flags);
ac2a6666 PNA	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
8b9229d1 PNA	77	static u32 flow_offload_dst_cookie(struct flow_offload_tuple *flow_tuple)
	78	{
	79	const struct rt6_info *rt;
	80
	81	if (flow_tuple->l3proto == NFPROTO_IPV6) {
	82	rt = (const struct rt6_info *)flow_tuple->dst_cache;
	83	return rt6_get_cookie(rt);
	84	}
	85
	86	return 0;
	87	}
	88
f1363e05 PNA	89	static int flow_offload_fill_route(struct flow_offload *flow,
	90	const struct nf_flow_route *route,
	91	enum flow_offload_tuple_dir dir)
	92	{
	93	struct flow_offload_tuple *flow_tuple = &flow->tuplehash[dir].tuple;
f1363e05	94	struct dst_entry *dst = route->tuple[dir].dst;
4cd91f7c	95	int i, j = 0;
f1363e05	96
f1363e05 PNA	97	switch (flow_tuple->l3proto) {
	98	case NFPROTO_IPV4:
	99	flow_tuple->mtu = ip_dst_mtu_maybe_forward(dst, true);
	100	break;
	101	case NFPROTO_IPV6:
	102	flow_tuple->mtu = ip6_dst_mtu_forward(dst);
	103	break;
	104	}
	105
c63a7cc4	106	flow_tuple->iifidx = route->tuple[dir].in.ifindex;
4cd91f7c PNA	107	for (i = route->tuple[dir].in.num_encaps - 1; i >= 0; i--) {
	108	flow_tuple->encap[j].id = route->tuple[dir].in.encap[i].id;
	109	flow_tuple->encap[j].proto = route->tuple[dir].in.encap[i].proto;
26267bf9 FF	110	if (route->tuple[dir].in.ingress_vlans & BIT(i))
26267bf9 FF	111	flow_tuple->in_vlan_ingress \|= BIT(j);
4cd91f7c PNA	112	j++;
	113	}
	114	flow_tuple->encap_num = route->tuple[dir].in.num_encaps;
7a27f6ab PNA	115
	116	switch (route->tuple[dir].xmit_type) {
	117	case FLOW_OFFLOAD_XMIT_DIRECT:
	118	memcpy(flow_tuple->out.h_dest, route->tuple[dir].out.h_dest,
	119	ETH_ALEN);
	120	memcpy(flow_tuple->out.h_source, route->tuple[dir].out.h_source,
	121	ETH_ALEN);
	122	flow_tuple->out.ifidx = route->tuple[dir].out.ifindex;
73f97025	123	flow_tuple->out.hw_ifidx = route->tuple[dir].out.hw_ifindex;
7a27f6ab PNA	124	break;
	125	case FLOW_OFFLOAD_XMIT_XFRM:
	126	case FLOW_OFFLOAD_XMIT_NEIGH:
	127	if (!dst_hold_safe(route->tuple[dir].dst))
	128	return -1;
	129
	130	flow_tuple->dst_cache = dst;
8b9229d1	131	flow_tuple->dst_cookie = flow_offload_dst_cookie(flow_tuple);
7a27f6ab	132	break;
78ed0a9b RD	133	default:
	134	WARN_ON_ONCE(1);
	135	break;
7a27f6ab	136	}
5139c0c0	137	flow_tuple->xmit_type = route->tuple[dir].xmit_type;
f1363e05 PNA	138
	139	return 0;
	140	}
	141
7a27f6ab PNA	142	static void nft_flow_dst_release(struct flow_offload *flow,
	143	enum flow_offload_tuple_dir dir)
	144	{
	145	if (flow->tuplehash[dir].tuple.xmit_type == FLOW_OFFLOAD_XMIT_NEIGH \|\|
	146	flow->tuplehash[dir].tuple.xmit_type == FLOW_OFFLOAD_XMIT_XFRM)
	147	dst_release(flow->tuplehash[dir].tuple.dst_cache);
	148	}
	149
f1363e05 PNA	150	int flow_offload_route_init(struct flow_offload *flow,
	151	const struct nf_flow_route *route)
	152	{
	153	int err;
	154
	155	err = flow_offload_fill_route(flow, route, FLOW_OFFLOAD_DIR_ORIGINAL);
	156	if (err < 0)
	157	return err;
	158
	159	err = flow_offload_fill_route(flow, route, FLOW_OFFLOAD_DIR_REPLY);
	160	if (err < 0)
	161	goto err_route_reply;
	162
	163	flow->type = NF_FLOW_OFFLOAD_ROUTE;
	164
	165	return 0;
	166
	167	err_route_reply:
7a27f6ab	168	nft_flow_dst_release(flow, FLOW_OFFLOAD_DIR_ORIGINAL);
f1363e05 PNA	169
	170	return err;
	171	}
	172	EXPORT_SYMBOL_GPL(flow_offload_route_init);
	173
da5984e5 FF	174	static void flow_offload_fixup_tcp(struct ip_ct_tcp *tcp)
	175	{
	176	tcp->state = TCP_CONNTRACK_ESTABLISHED;
	177	tcp->seen[0].td_maxwin = 0;
	178	tcp->seen[1].td_maxwin = 0;
	179	}
	180
e97d9404 FW	181	#define NF_FLOWTABLE_TCP_PICKUP_TIMEOUT (120 * HZ)
	182	#define NF_FLOWTABLE_UDP_PICKUP_TIMEOUT (30 * HZ)
	183
1e5b2471	184	static void flow_offload_fixup_ct_timeout(struct nf_conn *ct)
da5984e5 FF	185	{
da5984e5 FF	186	const struct nf_conntrack_l4proto *l4proto;
1e5b2471	187	int l4num = nf_ct_protonum(ct);
da5984e5	188	unsigned int timeout;
da5984e5	189
4a60dc74	190	l4proto = nf_ct_l4proto_find(l4num);
da5984e5 FF	191	if (!l4proto)
	192	return;
	193
da5984e5	194	if (l4num == IPPROTO_TCP)
e97d9404	195	timeout = NF_FLOWTABLE_TCP_PICKUP_TIMEOUT;
da5984e5	196	else if (l4num == IPPROTO_UDP)
e97d9404	197	timeout = NF_FLOWTABLE_UDP_PICKUP_TIMEOUT;
da5984e5 FF	198	else
	199	return;
	200
1e5b2471 PNA	201	if (nf_flow_timeout_delta(ct->timeout) > (__s32)timeout)
	202	ct->timeout = nfct_time_stamp + timeout;
	203	}
	204
	205	static void flow_offload_fixup_ct_state(struct nf_conn *ct)
	206	{
	207	if (nf_ct_protonum(ct) == IPPROTO_TCP)
	208	flow_offload_fixup_tcp(&ct->proto.tcp);
	209	}
	210
	211	static void flow_offload_fixup_ct(struct nf_conn *ct)
	212	{
	213	flow_offload_fixup_ct_state(ct);
	214	flow_offload_fixup_ct_timeout(ct);
da5984e5 FF	215	}
da5984e5 FF	216
f1363e05	217	static void flow_offload_route_release(struct flow_offload *flow)
ac2a6666	218	{
7a27f6ab PNA	219	nft_flow_dst_release(flow, FLOW_OFFLOAD_DIR_ORIGINAL);
7a27f6ab PNA	220	nft_flow_dst_release(flow, FLOW_OFFLOAD_DIR_REPLY);
f1363e05 PNA	221	}
	222
	223	void flow_offload_free(struct flow_offload *flow)
	224	{
	225	switch (flow->type) {
	226	case NF_FLOW_OFFLOAD_ROUTE:
	227	flow_offload_route_release(flow);
	228	break;
	229	default:
	230	break;
	231	}
b32d2f34	232	nf_ct_put(flow->ct);
62248df8	233	kfree_rcu(flow, rcu_head);
ac2a6666 PNA	234	}
	235	EXPORT_SYMBOL_GPL(flow_offload_free);
	236
a268de77 FF	237	static u32 flow_offload_hash(const void *data, u32 len, u32 seed)
	238	{
	239	const struct flow_offload_tuple *tuple = data;
	240
dbc859d9	241	return jhash(tuple, offsetof(struct flow_offload_tuple, __hash), seed);
a268de77 FF	242	}
	243
	244	static u32 flow_offload_hash_obj(const void *data, u32 len, u32 seed)
	245	{
	246	const struct flow_offload_tuple_rhash *tuplehash = data;
	247
dbc859d9	248	return jhash(&tuplehash->tuple, offsetof(struct flow_offload_tuple, __hash), seed);
a268de77 FF	249	}
	250
	251	static int flow_offload_hash_cmp(struct rhashtable_compare_arg *arg,
	252	const void *ptr)
	253	{
	254	const struct flow_offload_tuple *tuple = arg->key;
	255	const struct flow_offload_tuple_rhash *x = ptr;
	256
dbc859d9	257	if (memcmp(&x->tuple, tuple, offsetof(struct flow_offload_tuple, __hash)))
a268de77 FF	258	return 1;
	259
	260	return 0;
	261	}
	262
	263	static const struct rhashtable_params nf_flow_offload_rhash_params = {
	264	.head_offset = offsetof(struct flow_offload_tuple_rhash, node),
	265	.hashfn = flow_offload_hash,
	266	.obj_hashfn = flow_offload_hash_obj,
	267	.obj_cmpfn = flow_offload_hash_cmp,
	268	.automatic_shrinking = true,
	269	};
	270
ac2a6666 PNA	271	int flow_offload_add(struct nf_flowtable flow_table, struct flow_offload flow)
ac2a6666 PNA	272	{
43c8f131 TY	273	int err;
43c8f131 TY	274
fb46f1b7	275	flow->timeout = nf_flowtable_time_stamp + NF_FLOW_TIMEOUT;
daf61b02	276
43c8f131 TY	277	err = rhashtable_insert_fast(&flow_table->rhashtable,
	278	&flow->tuplehash[0].node,
	279	nf_flow_offload_rhash_params);
	280	if (err < 0)
	281	return err;
	282
	283	err = rhashtable_insert_fast(&flow_table->rhashtable,
	284	&flow->tuplehash[1].node,
	285	nf_flow_offload_rhash_params);
	286	if (err < 0) {
	287	rhashtable_remove_fast(&flow_table->rhashtable,
	288	&flow->tuplehash[0].node,
	289	nf_flow_offload_rhash_params);
	290	return err;
	291	}
ac2a6666	292
4203b19c RD	293	nf_ct_offload_timeout(flow->ct);
4203b19c RD	294
f698fe40 PNA	295	if (nf_flowtable_hw_offload(flow_table)) {
f698fe40 PNA	296	__set_bit(NF_FLOW_HW, &flow->flags);
c29f74e0	297	nf_flow_offload_add(flow_table, flow);
f698fe40	298	}
c29f74e0	299
ac2a6666 PNA	300	return 0;
	301	}
	302	EXPORT_SYMBOL_GPL(flow_offload_add);
	303
8b3646d6 PB	304	void flow_offload_refresh(struct nf_flowtable *flow_table,
	305	struct flow_offload *flow)
	306	{
	307	flow->timeout = nf_flowtable_time_stamp + NF_FLOW_TIMEOUT;
	308
	309	if (likely(!nf_flowtable_hw_offload(flow_table) \|\|
	310	!test_and_clear_bit(NF_FLOW_HW_REFRESH, &flow->flags)))
	311	return;
	312
	313	nf_flow_offload_add(flow_table, flow);
	314	}
	315	EXPORT_SYMBOL_GPL(flow_offload_refresh);
	316
3e68db2f PNA	317	static inline bool nf_flow_has_expired(const struct flow_offload *flow)
3e68db2f PNA	318	{
1e5b2471	319	return nf_flow_timeout_delta(flow->timeout) <= 0;
3e68db2f PNA	320	}
3e68db2f PNA	321
0ff90b6c FF	322	static void flow_offload_del(struct nf_flowtable *flow_table,
0ff90b6c FF	323	struct flow_offload *flow)
ac2a6666	324	{
ac2a6666 PNA	325	rhashtable_remove_fast(&flow_table->rhashtable,
ac2a6666 PNA	326	&flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].node,
a268de77	327	nf_flow_offload_rhash_params);
ac2a6666 PNA	328	rhashtable_remove_fast(&flow_table->rhashtable,
ac2a6666 PNA	329	&flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].node,
a268de77	330	nf_flow_offload_rhash_params);
ac2a6666	331
b32d2f34	332	clear_bit(IPS_OFFLOAD_BIT, &flow->ct->status);
da5984e5	333
3e68db2f	334	if (nf_flow_has_expired(flow))
b32d2f34	335	flow_offload_fixup_ct(flow->ct);
9ed81c8e	336	else
b32d2f34	337	flow_offload_fixup_ct_timeout(flow->ct);
3e68db2f	338
0ff90b6c	339	flow_offload_free(flow);
ac2a6666	340	}
ac2a6666	341
59c466dd FF	342	void flow_offload_teardown(struct flow_offload *flow)
59c466dd FF	343	{
355a8b13	344	set_bit(NF_FLOW_TEARDOWN, &flow->flags);
da5984e5	345
b32d2f34	346	flow_offload_fixup_ct_state(flow->ct);
59c466dd FF	347	}
	348	EXPORT_SYMBOL_GPL(flow_offload_teardown);
	349
ac2a6666 PNA	350	struct flow_offload_tuple_rhash *
	351	flow_offload_lookup(struct nf_flowtable *flow_table,
	352	struct flow_offload_tuple *tuple)
	353	{
ba03137f FF	354	struct flow_offload_tuple_rhash *tuplehash;
	355	struct flow_offload *flow;
	356	int dir;
	357
a2d88182 TY	358	tuplehash = rhashtable_lookup(&flow_table->rhashtable, tuple,
a2d88182 TY	359	nf_flow_offload_rhash_params);
ba03137f FF	360	if (!tuplehash)
	361	return NULL;
	362
	363	dir = tuplehash->tuple.dir;
	364	flow = container_of(tuplehash, struct flow_offload, tuplehash[dir]);
355a8b13	365	if (test_bit(NF_FLOW_TEARDOWN, &flow->flags))
ba03137f FF	366	return NULL;
ba03137f FF	367
b32d2f34	368	if (unlikely(nf_ct_is_dying(flow->ct)))
8cd2bc98 TY	369	return NULL;
8cd2bc98 TY	370
ba03137f	371	return tuplehash;
ac2a6666 PNA	372	}
	373	EXPORT_SYMBOL_GPL(flow_offload_lookup);
	374
49de9c09 TY	375	static int
	376	nf_flow_table_iterate(struct nf_flowtable *flow_table,
	377	void (iter)(struct flow_offload flow, void *data),
	378	void *data)
ac2a6666 PNA	379	{
	380	struct flow_offload_tuple_rhash *tuplehash;
	381	struct rhashtable_iter hti;
	382	struct flow_offload *flow;
0de22baa	383	int err = 0;
ac2a6666	384
0de22baa	385	rhashtable_walk_enter(&flow_table->rhashtable, &hti);
ac2a6666 PNA	386	rhashtable_walk_start(&hti);
	387
	388	while ((tuplehash = rhashtable_walk_next(&hti))) {
	389	if (IS_ERR(tuplehash)) {
0de22baa TY	390	if (PTR_ERR(tuplehash) != -EAGAIN) {
	391	err = PTR_ERR(tuplehash);
	392	break;
	393	}
ac2a6666 PNA	394	continue;
	395	}
	396	if (tuplehash->tuple.dir)
	397	continue;
	398
	399	flow = container_of(tuplehash, struct flow_offload, tuplehash[0]);
	400
	401	iter(flow, data);
	402	}
ac2a6666 PNA	403	rhashtable_walk_stop(&hti);
	404	rhashtable_walk_exit(&hti);
	405
	406	return err;
	407	}
ac2a6666	408
8b9229d1 PNA	409	static bool flow_offload_stale_dst(struct flow_offload_tuple *tuple)
	410	{
	411	struct dst_entry *dst;
	412
	413	if (tuple->xmit_type == FLOW_OFFLOAD_XMIT_NEIGH \|\|
	414	tuple->xmit_type == FLOW_OFFLOAD_XMIT_XFRM) {
	415	dst = tuple->dst_cache;
	416	if (!dst_check(dst, tuple->dst_cookie))
	417	return true;
	418	}
	419
	420	return false;
	421	}
	422
	423	static bool nf_flow_has_stale_dst(struct flow_offload *flow)
	424	{
	425	return flow_offload_stale_dst(&flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple) \|\|
	426	flow_offload_stale_dst(&flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple);
	427	}
	428
b9660987	429	static void nf_flow_offload_gc_step(struct flow_offload flow, void data)
ac2a6666	430	{
b9660987	431	struct nf_flowtable *flow_table = data;
ac2a6666	432
8b9229d1 PNA	433	if (nf_flow_has_expired(flow) \|\|
	434	nf_ct_is_dying(flow->ct) \|\|
	435	nf_flow_has_stale_dst(flow))
9ed81c8e PNA	436	set_bit(NF_FLOW_TEARDOWN, &flow->flags);
	437
	438	if (test_bit(NF_FLOW_TEARDOWN, &flow->flags)) {
355a8b13 PNA	439	if (test_bit(NF_FLOW_HW, &flow->flags)) {
355a8b13 PNA	440	if (!test_bit(NF_FLOW_HW_DYING, &flow->flags))
c29f74e0	441	nf_flow_offload_del(flow_table, flow);
355a8b13	442	else if (test_bit(NF_FLOW_HW_DEAD, &flow->flags))
c29f74e0 PNA	443	flow_offload_del(flow_table, flow);
	444	} else {
	445	flow_offload_del(flow_table, flow);
	446	}
355a8b13	447	} else if (test_bit(NF_FLOW_HW, &flow->flags)) {
79b9b685	448	nf_flow_offload_stats(flow_table, flow);
c29f74e0	449	}
b408c5b0 PNA	450	}
b408c5b0 PNA	451
a268de77	452	static void nf_flow_offload_work_gc(struct work_struct *work)
b408c5b0 PNA	453	{
	454	struct nf_flowtable *flow_table;
	455
	456	flow_table = container_of(work, struct nf_flowtable, gc_work.work);
b9660987	457	nf_flow_table_iterate(flow_table, nf_flow_offload_gc_step, flow_table);
ac2a6666 PNA	458	queue_delayed_work(system_power_efficient_wq, &flow_table->gc_work, HZ);
ac2a6666 PNA	459	}
ac2a6666	460
f4401262 PNA	461	static void nf_flow_nat_port_tcp(struct sk_buff *skb, unsigned int thoff,
f4401262 PNA	462	__be16 port, __be16 new_port)
ac2a6666 PNA	463	{
	464	struct tcphdr *tcph;
	465
ac2a6666	466	tcph = (void *)(skb_network_header(skb) + thoff);
8d6bca15	467	inet_proto_csum_replace2(&tcph->check, skb, port, new_port, false);
ac2a6666 PNA	468	}
ac2a6666 PNA	469
f4401262 PNA	470	static void nf_flow_nat_port_udp(struct sk_buff *skb, unsigned int thoff,
f4401262 PNA	471	__be16 port, __be16 new_port)
ac2a6666 PNA	472	{
	473	struct udphdr *udph;
	474
ac2a6666 PNA	475	udph = (void *)(skb_network_header(skb) + thoff);
	476	if (udph->check \|\| skb->ip_summed == CHECKSUM_PARTIAL) {
	477	inet_proto_csum_replace2(&udph->check, skb, port,
8d6bca15	478	new_port, false);
ac2a6666 PNA	479	if (!udph->check)
	480	udph->check = CSUM_MANGLED_0;
	481	}
ac2a6666 PNA	482	}
ac2a6666 PNA	483
f4401262 PNA	484	static void nf_flow_nat_port(struct sk_buff *skb, unsigned int thoff,
f4401262 PNA	485	u8 protocol, __be16 port, __be16 new_port)
ac2a6666 PNA	486	{
	487	switch (protocol) {
	488	case IPPROTO_TCP:
f4401262	489	nf_flow_nat_port_tcp(skb, thoff, port, new_port);
ac2a6666 PNA	490	break;
ac2a6666 PNA	491	case IPPROTO_UDP:
f4401262	492	nf_flow_nat_port_udp(skb, thoff, port, new_port);
ac2a6666 PNA	493	break;
ac2a6666 PNA	494	}
ac2a6666 PNA	495	}
ac2a6666 PNA	496
f4401262 PNA	497	void nf_flow_snat_port(const struct flow_offload *flow,
	498	struct sk_buff *skb, unsigned int thoff,
	499	u8 protocol, enum flow_offload_tuple_dir dir)
ac2a6666 PNA	500	{
	501	struct flow_ports *hdr;
	502	__be16 port, new_port;
	503
ac2a6666 PNA	504	hdr = (void *)(skb_network_header(skb) + thoff);
	505
	506	switch (dir) {
	507	case FLOW_OFFLOAD_DIR_ORIGINAL:
	508	port = hdr->source;
	509	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.dst_port;
	510	hdr->source = new_port;
	511	break;
	512	case FLOW_OFFLOAD_DIR_REPLY:
	513	port = hdr->dest;
	514	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.src_port;
	515	hdr->dest = new_port;
	516	break;
ac2a6666 PNA	517	}
ac2a6666 PNA	518
f4401262	519	nf_flow_nat_port(skb, thoff, protocol, port, new_port);
ac2a6666 PNA	520	}
	521	EXPORT_SYMBOL_GPL(nf_flow_snat_port);
	522
f4401262 PNA	523	void nf_flow_dnat_port(const struct flow_offload flow, struct sk_buff skb,
	524	unsigned int thoff, u8 protocol,
	525	enum flow_offload_tuple_dir dir)
ac2a6666 PNA	526	{
	527	struct flow_ports *hdr;
	528	__be16 port, new_port;
	529
ac2a6666 PNA	530	hdr = (void *)(skb_network_header(skb) + thoff);
	531
	532	switch (dir) {
	533	case FLOW_OFFLOAD_DIR_ORIGINAL:
	534	port = hdr->dest;
	535	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_REPLY].tuple.src_port;
	536	hdr->dest = new_port;
	537	break;
	538	case FLOW_OFFLOAD_DIR_REPLY:
	539	port = hdr->source;
	540	new_port = flow->tuplehash[FLOW_OFFLOAD_DIR_ORIGINAL].tuple.dst_port;
	541	hdr->source = new_port;
	542	break;
ac2a6666 PNA	543	}
ac2a6666 PNA	544
f4401262	545	nf_flow_nat_port(skb, thoff, protocol, port, new_port);
ac2a6666 PNA	546	}
	547	EXPORT_SYMBOL_GPL(nf_flow_dnat_port);
	548
a268de77 FF	549	int nf_flow_table_init(struct nf_flowtable *flowtable)
	550	{
	551	int err;
	552
740b486a	553	INIT_DELAYED_WORK(&flowtable->gc_work, nf_flow_offload_work_gc);
c29f74e0	554	flow_block_init(&flowtable->flow_block);
422c032a	555	init_rwsem(&flowtable->flow_block_lock);
a268de77 FF	556
	557	err = rhashtable_init(&flowtable->rhashtable,
	558	&nf_flow_offload_rhash_params);
	559	if (err < 0)
	560	return err;
	561
	562	queue_delayed_work(system_power_efficient_wq,
	563	&flowtable->gc_work, HZ);
	564
84453a90 FF	565	mutex_lock(&flowtable_lock);
	566	list_add(&flowtable->list, &flowtables);
	567	mutex_unlock(&flowtable_lock);
	568
a268de77 FF	569	return 0;
	570	}
	571	EXPORT_SYMBOL_GPL(nf_flow_table_init);
	572
c0ea1bcb PNA	573	static void nf_flow_table_do_cleanup(struct flow_offload flow, void data)
	574	{
	575	struct net_device *dev = data;
	576
59c466dd FF	577	if (!dev) {
59c466dd FF	578	flow_offload_teardown(flow);
c0ea1bcb	579	return;
59c466dd	580	}
b32d2f34 PNA	581
b32d2f34 PNA	582	if (net_eq(nf_ct_net(flow->ct), dev_net(dev)) &&
a3fb3698 TY	583	(flow->tuplehash[0].tuple.iifidx == dev->ifindex \|\|
a3fb3698 TY	584	flow->tuplehash[1].tuple.iifidx == dev->ifindex))
445db8d0	585	flow_offload_teardown(flow);
c0ea1bcb PNA	586	}
c0ea1bcb PNA	587
a8284c68 PNA	588	void nf_flow_table_gc_cleanup(struct nf_flowtable *flowtable,
a8284c68 PNA	589	struct net_device *dev)
c0ea1bcb	590	{
84453a90	591	nf_flow_table_iterate(flowtable, nf_flow_table_do_cleanup, dev);
c0ea1bcb	592	flush_delayed_work(&flowtable->gc_work);
91bfaa15	593	nf_flow_table_offload_flush(flowtable);
c0ea1bcb PNA	594	}
c0ea1bcb PNA	595
5f1be84a	596	void nf_flow_table_cleanup(struct net_device *dev)
c0ea1bcb	597	{
84453a90 FF	598	struct nf_flowtable *flowtable;
	599
	600	mutex_lock(&flowtable_lock);
	601	list_for_each_entry(flowtable, &flowtables, list)
a8284c68	602	nf_flow_table_gc_cleanup(flowtable, dev);
84453a90	603	mutex_unlock(&flowtable_lock);
c0ea1bcb PNA	604	}
	605	EXPORT_SYMBOL_GPL(nf_flow_table_cleanup);
	606
b408c5b0 PNA	607	void nf_flow_table_free(struct nf_flowtable *flow_table)
b408c5b0 PNA	608	{
84453a90 FF	609	mutex_lock(&flowtable_lock);
	610	list_del(&flow_table->list);
	611	mutex_unlock(&flowtable_lock);
978703f4	612
a268de77	613	cancel_delayed_work_sync(&flow_table->gc_work);
b408c5b0	614	nf_flow_table_iterate(flow_table, nf_flow_table_do_cleanup, NULL);
b9660987	615	nf_flow_table_iterate(flow_table, nf_flow_offload_gc_step, flow_table);
0f34f30a	616	nf_flow_table_offload_flush(flow_table);
c921ffe8 PB	617	if (nf_flowtable_hw_offload(flow_table))
	618	nf_flow_table_iterate(flow_table, nf_flow_offload_gc_step,
	619	flow_table);
a268de77	620	rhashtable_destroy(&flow_table->rhashtable);
b408c5b0 PNA	621	}
	622	EXPORT_SYMBOL_GPL(nf_flow_table_free);
	623
c29f74e0 PNA	624	static int __init nf_flow_table_module_init(void)
	625	{
	626	return nf_flow_table_offload_init();
	627	}
	628
	629	static void __exit nf_flow_table_module_exit(void)
	630	{
	631	nf_flow_table_offload_exit();
	632	}
	633
	634	module_init(nf_flow_table_module_init);
	635	module_exit(nf_flow_table_module_exit);
	636
ac2a6666 PNA	637	MODULE_LICENSE("GPL");
ac2a6666 PNA	638	MODULE_AUTHOR("Pablo Neira Ayuso <pablo@netfilter.org>");
4cacc395	639	MODULE_DESCRIPTION("Netfilter flow table module");