[linux-block.git] / net / netfilter / nf_conntrack_bpf.c

// SPDX-License-Identifier: GPL-2.0-only
/* Unstable Conntrack Helpers for XDP and TC-BPF hook
 *
 * These are called from the XDP and SCHED_CLS BPF programs. Note that it is
 * allowed to break compatibility for these functions since the interface they
 * are exposed through to BPF programs is explicitly unstable.
 */

#include <linux/bpf_verifier.h>
#include <linux/bpf.h>
#include <linux/btf.h>
#include <linux/filter.h>
#include <linux/mutex.h>
#include <linux/types.h>
#include <linux/btf_ids.h>
#include <linux/net_namespace.h>
#include <net/netfilter/nf_conntrack_bpf.h>
#include <net/netfilter/nf_conntrack_core.h>

/* bpf_ct_opts - Options for CT lookup helpers
 *
 * Members:
 * @netns_id   - Specify the network namespace for lookup
 *		 Values:
 *		   BPF_F_CURRENT_NETNS (-1)
 *		     Use namespace associated with ctx (xdp_md, __sk_buff)
 *		   [0, S32_MAX]
 *		     Network Namespace ID
 * @error      - Out parameter, set for any errors encountered
 *		 Values:
 *		   -EINVAL - Passed NULL for bpf_tuple pointer
 *		   -EINVAL - opts->reserved is not 0
 *		   -EINVAL - netns_id is less than -1
 *		   -EINVAL - opts__sz isn't NF_BPF_CT_OPTS_SZ (12)
 *		   -EPROTO - l4proto isn't one of IPPROTO_TCP or IPPROTO_UDP
 *		   -ENONET - No network namespace found for netns_id
 *		   -ENOENT - Conntrack lookup could not find entry for tuple
 *		   -EAFNOSUPPORT - tuple__sz isn't one of sizeof(tuple->ipv4)
 *				   or sizeof(tuple->ipv6)
 * @l4proto    - Layer 4 protocol
 *		 Values:
 *		   IPPROTO_TCP, IPPROTO_UDP
 * @dir:       - connection tracking tuple direction.
 * @reserved   - Reserved member, will be reused for more options in future
 *		 Values:
 *		   0
 */
struct bpf_ct_opts {
	s32 netns_id;
	s32 error;
	u8 l4proto;
	u8 dir;
	u8 reserved[2];
};

enum {
	NF_BPF_CT_OPTS_SZ = 12,
};

static int bpf_nf_ct_tuple_parse(struct bpf_sock_tuple *bpf_tuple,
				 u32 tuple_len, u8 protonum, u8 dir,
				 struct nf_conntrack_tuple *tuple)
{
	union nf_inet_addr *src = dir ? &tuple->dst.u3 : &tuple->src.u3;
	union nf_inet_addr *dst = dir ? &tuple->src.u3 : &tuple->dst.u3;
	union nf_conntrack_man_proto *sport = dir ? (void *)&tuple->dst.u
						  : &tuple->src.u;
	union nf_conntrack_man_proto *dport = dir ? &tuple->src.u
						  : (void *)&tuple->dst.u;

	if (unlikely(protonum != IPPROTO_TCP && protonum != IPPROTO_UDP))
		return -EPROTO;

	memset(tuple, 0, sizeof(*tuple));

	switch (tuple_len) {
	case sizeof(bpf_tuple->ipv4):
		tuple->src.l3num = AF_INET;
		src->ip = bpf_tuple->ipv4.saddr;
		sport->tcp.port = bpf_tuple->ipv4.sport;
		dst->ip = bpf_tuple->ipv4.daddr;
		dport->tcp.port = bpf_tuple->ipv4.dport;
		break;
	case sizeof(bpf_tuple->ipv6):
		tuple->src.l3num = AF_INET6;
		memcpy(src->ip6, bpf_tuple->ipv6.saddr, sizeof(bpf_tuple->ipv6.saddr));
		sport->tcp.port = bpf_tuple->ipv6.sport;
		memcpy(dst->ip6, bpf_tuple->ipv6.daddr, sizeof(bpf_tuple->ipv6.daddr));
		dport->tcp.port = bpf_tuple->ipv6.dport;
		break;
	default:
		return -EAFNOSUPPORT;
	}
	tuple->dst.protonum = protonum;
	tuple->dst.dir = dir;

	return 0;
}

static struct nf_conn *
__bpf_nf_ct_alloc_entry(struct net *net, struct bpf_sock_tuple *bpf_tuple,
			u32 tuple_len, struct bpf_ct_opts *opts, u32 opts_len,
			u32 timeout)
{
	struct nf_conntrack_tuple otuple, rtuple;
	struct nf_conn *ct;
	int err;

	if (!opts || !bpf_tuple || opts->reserved[0] || opts->reserved[1] ||
	    opts_len != NF_BPF_CT_OPTS_SZ)
		return ERR_PTR(-EINVAL);

	if (unlikely(opts->netns_id < BPF_F_CURRENT_NETNS))
		return ERR_PTR(-EINVAL);

	err = bpf_nf_ct_tuple_parse(bpf_tuple, tuple_len, opts->l4proto,
				    IP_CT_DIR_ORIGINAL, &otuple);
	if (err < 0)
		return ERR_PTR(err);

	err = bpf_nf_ct_tuple_parse(bpf_tuple, tuple_len, opts->l4proto,
				    IP_CT_DIR_REPLY, &rtuple);
	if (err < 0)
		return ERR_PTR(err);

	if (opts->netns_id >= 0) {
		net = get_net_ns_by_id(net, opts->netns_id);
		if (unlikely(!net))
			return ERR_PTR(-ENONET);
	}

	ct = nf_conntrack_alloc(net, &nf_ct_zone_dflt, &otuple, &rtuple,
				GFP_ATOMIC);
	if (IS_ERR(ct))
		goto out;

	memset(&ct->proto, 0, sizeof(ct->proto));
	__nf_ct_set_timeout(ct, timeout * HZ);

out:
	if (opts->netns_id >= 0)
		put_net(net);

	return ct;
}

static struct nf_conn *__bpf_nf_ct_lookup(struct net *net,
					  struct bpf_sock_tuple *bpf_tuple,
					  u32 tuple_len, struct bpf_ct_opts *opts,
					  u32 opts_len)
{
	struct nf_conntrack_tuple_hash *hash;
	struct nf_conntrack_tuple tuple;
	struct nf_conn *ct;
	int err;

	if (!opts || !bpf_tuple || opts->reserved[0] || opts->reserved[1] ||
	    opts_len != NF_BPF_CT_OPTS_SZ)
		return ERR_PTR(-EINVAL);
	if (unlikely(opts->l4proto != IPPROTO_TCP && opts->l4proto != IPPROTO_UDP))
		return ERR_PTR(-EPROTO);
	if (unlikely(opts->netns_id < BPF_F_CURRENT_NETNS))
		return ERR_PTR(-EINVAL);

	err = bpf_nf_ct_tuple_parse(bpf_tuple, tuple_len, opts->l4proto,
				    IP_CT_DIR_ORIGINAL, &tuple);
	if (err < 0)
		return ERR_PTR(err);

	if (opts->netns_id >= 0) {
		net = get_net_ns_by_id(net, opts->netns_id);
		if (unlikely(!net))
			return ERR_PTR(-ENONET);
	}

	hash = nf_conntrack_find_get(net, &nf_ct_zone_dflt, &tuple);
	if (opts->netns_id >= 0)
		put_net(net);
	if (!hash)
		return ERR_PTR(-ENOENT);

	ct = nf_ct_tuplehash_to_ctrack(hash);
	opts->dir = NF_CT_DIRECTION(hash);

	return ct;
}

BTF_ID_LIST(btf_nf_conn_ids)
BTF_ID(struct, nf_conn)
BTF_ID(struct, nf_conn___init)

/* Check writes into `struct nf_conn` */
static int _nf_conntrack_btf_struct_access(struct bpf_verifier_log *log,
					   const struct btf *btf,
					   const struct btf_type *t, int off,
					   int size, enum bpf_access_type atype,
					   u32 *next_btf_id,
					   enum bpf_type_flag *flag)
{
	const struct btf_type *ncit;
	const struct btf_type *nct;
	size_t end;

	ncit = btf_type_by_id(btf, btf_nf_conn_ids[1]);
	nct = btf_type_by_id(btf, btf_nf_conn_ids[0]);

	if (t != nct && t != ncit) {
		bpf_log(log, "only read is supported\n");
		return -EACCES;
	}

	/* `struct nf_conn` and `struct nf_conn___init` have the same layout
	 * so we are safe to simply merge offset checks here
	 */
	switch (off) {
#if defined(CONFIG_NF_CONNTRACK_MARK)
	case offsetof(struct nf_conn, mark):
		end = offsetofend(struct nf_conn, mark);
		break;
#endif
	default:
		bpf_log(log, "no write support to nf_conn at off %d\n", off);
		return -EACCES;
	}

	if (off + size > end) {
		bpf_log(log,
			"write access at off %d with size %d beyond the member of nf_conn ended at %zu\n",
			off, size, end);
		return -EACCES;
	}

	return 0;
}

__diag_push();
__diag_ignore_all("-Wmissing-prototypes",
		  "Global functions as their definitions will be in nf_conntrack BTF");

/* bpf_xdp_ct_alloc - Allocate a new CT entry
 *
 * Parameters:
 * @xdp_ctx	- Pointer to ctx (xdp_md) in XDP program
 *		    Cannot be NULL
 * @bpf_tuple	- Pointer to memory representing the tuple to look up
 *		    Cannot be NULL
 * @tuple__sz	- Length of the tuple structure
 *		    Must be one of sizeof(bpf_tuple->ipv4) or
 *		    sizeof(bpf_tuple->ipv6)
 * @opts	- Additional options for allocation (documented above)
 *		    Cannot be NULL
 * @opts__sz	- Length of the bpf_ct_opts structure
 *		    Must be NF_BPF_CT_OPTS_SZ (12)
 */
struct nf_conn___init *
bpf_xdp_ct_alloc(struct xdp_md *xdp_ctx, struct bpf_sock_tuple *bpf_tuple,
		 u32 tuple__sz, struct bpf_ct_opts *opts, u32 opts__sz)
{
	struct xdp_buff *ctx = (struct xdp_buff *)xdp_ctx;
	struct nf_conn *nfct;

	nfct = __bpf_nf_ct_alloc_entry(dev_net(ctx->rxq->dev), bpf_tuple, tuple__sz,
				       opts, opts__sz, 10);
	if (IS_ERR(nfct)) {
		if (opts)
			opts->error = PTR_ERR(nfct);
		return NULL;
	}

	return (struct nf_conn___init *)nfct;
}

/* bpf_xdp_ct_lookup - Lookup CT entry for the given tuple, and acquire a
 *		       reference to it
 *
 * Parameters:
 * @xdp_ctx	- Pointer to ctx (xdp_md) in XDP program
 *		    Cannot be NULL
 * @bpf_tuple	- Pointer to memory representing the tuple to look up
 *		    Cannot be NULL
 * @tuple__sz	- Length of the tuple structure
 *		    Must be one of sizeof(bpf_tuple->ipv4) or
 *		    sizeof(bpf_tuple->ipv6)
 * @opts	- Additional options for lookup (documented above)
 *		    Cannot be NULL
 * @opts__sz	- Length of the bpf_ct_opts structure
 *		    Must be NF_BPF_CT_OPTS_SZ (12)
 */
struct nf_conn *
bpf_xdp_ct_lookup(struct xdp_md *xdp_ctx, struct bpf_sock_tuple *bpf_tuple,
		  u32 tuple__sz, struct bpf_ct_opts *opts, u32 opts__sz)
{
	struct xdp_buff *ctx = (struct xdp_buff *)xdp_ctx;
	struct net *caller_net;
	struct nf_conn *nfct;

	caller_net = dev_net(ctx->rxq->dev);
	nfct = __bpf_nf_ct_lookup(caller_net, bpf_tuple, tuple__sz, opts, opts__sz);
	if (IS_ERR(nfct)) {
		if (opts)
			opts->error = PTR_ERR(nfct);
		return NULL;
	}
	return nfct;
}

/* bpf_skb_ct_alloc - Allocate a new CT entry
 *
 * Parameters:
 * @skb_ctx	- Pointer to ctx (__sk_buff) in TC program
 *		    Cannot be NULL
 * @bpf_tuple	- Pointer to memory representing the tuple to look up
 *		    Cannot be NULL
 * @tuple__sz	- Length of the tuple structure
 *		    Must be one of sizeof(bpf_tuple->ipv4) or
 *		    sizeof(bpf_tuple->ipv6)
 * @opts	- Additional options for allocation (documented above)
 *		    Cannot be NULL
 * @opts__sz	- Length of the bpf_ct_opts structure
 *		    Must be NF_BPF_CT_OPTS_SZ (12)
 */
struct nf_conn___init *
bpf_skb_ct_alloc(struct __sk_buff *skb_ctx, struct bpf_sock_tuple *bpf_tuple,
		 u32 tuple__sz, struct bpf_ct_opts *opts, u32 opts__sz)
{
	struct sk_buff *skb = (struct sk_buff *)skb_ctx;
	struct nf_conn *nfct;
	struct net *net;

	net = skb->dev ? dev_net(skb->dev) : sock_net(skb->sk);
	nfct = __bpf_nf_ct_alloc_entry(net, bpf_tuple, tuple__sz, opts, opts__sz, 10);
	if (IS_ERR(nfct)) {
		if (opts)
			opts->error = PTR_ERR(nfct);
		return NULL;
	}

	return (struct nf_conn___init *)nfct;
}

/* bpf_skb_ct_lookup - Lookup CT entry for the given tuple, and acquire a
 *		       reference to it
 *
 * Parameters:
 * @skb_ctx	- Pointer to ctx (__sk_buff) in TC program
 *		    Cannot be NULL
 * @bpf_tuple	- Pointer to memory representing the tuple to look up
 *		    Cannot be NULL
 * @tuple__sz	- Length of the tuple structure
 *		    Must be one of sizeof(bpf_tuple->ipv4) or
 *		    sizeof(bpf_tuple->ipv6)
 * @opts	- Additional options for lookup (documented above)
 *		    Cannot be NULL
 * @opts__sz	- Length of the bpf_ct_opts structure
 *		    Must be NF_BPF_CT_OPTS_SZ (12)
 */
struct nf_conn *
bpf_skb_ct_lookup(struct __sk_buff *skb_ctx, struct bpf_sock_tuple *bpf_tuple,
		  u32 tuple__sz, struct bpf_ct_opts *opts, u32 opts__sz)
{
	struct sk_buff *skb = (struct sk_buff *)skb_ctx;
	struct net *caller_net;
	struct nf_conn *nfct;

	caller_net = skb->dev ? dev_net(skb->dev) : sock_net(skb->sk);
	nfct = __bpf_nf_ct_lookup(caller_net, bpf_tuple, tuple__sz, opts, opts__sz);
	if (IS_ERR(nfct)) {
		if (opts)
			opts->error = PTR_ERR(nfct);
		return NULL;
	}
	return nfct;
}

/* bpf_ct_insert_entry - Add the provided entry into a CT map
 *
 * This must be invoked for referenced PTR_TO_BTF_ID.
 *
 * @nfct	 - Pointer to referenced nf_conn___init object, obtained
 *		   using bpf_xdp_ct_alloc or bpf_skb_ct_alloc.
 */
struct nf_conn *bpf_ct_insert_entry(struct nf_conn___init *nfct_i)
{
	struct nf_conn *nfct = (struct nf_conn *)nfct_i;
	int err;

	nfct->status |= IPS_CONFIRMED;
	err = nf_conntrack_hash_check_insert(nfct);
	if (err < 0) {
		nf_conntrack_free(nfct);
		return NULL;
	}
	return nfct;
}

/* bpf_ct_release - Release acquired nf_conn object
 *
 * This must be invoked for referenced PTR_TO_BTF_ID, and the verifier rejects
 * the program if any references remain in the program in all of the explored
 * states.
 *
 * Parameters:
 * @nf_conn	 - Pointer to referenced nf_conn object, obtained using
 *		   bpf_xdp_ct_lookup or bpf_skb_ct_lookup.
 */
void bpf_ct_release(struct nf_conn *nfct)
{
	if (!nfct)
		return;
	nf_ct_put(nfct);
}

/* bpf_ct_set_timeout - Set timeout of allocated nf_conn
 *
 * Sets the default timeout of newly allocated nf_conn before insertion.
 * This helper must be invoked for refcounted pointer to nf_conn___init.
 *
 * Parameters:
 * @nfct	 - Pointer to referenced nf_conn object, obtained using
 *                 bpf_xdp_ct_alloc or bpf_skb_ct_alloc.
 * @timeout      - Timeout in msecs.
 */
void bpf_ct_set_timeout(struct nf_conn___init *nfct, u32 timeout)
{
	__nf_ct_set_timeout((struct nf_conn *)nfct, msecs_to_jiffies(timeout));
}

/* bpf_ct_change_timeout - Change timeout of inserted nf_conn
 *
 * Change timeout associated of the inserted or looked up nf_conn.
 * This helper must be invoked for refcounted pointer to nf_conn.
 *
 * Parameters:
 * @nfct	 - Pointer to referenced nf_conn object, obtained using
 *		   bpf_ct_insert_entry, bpf_xdp_ct_lookup, or bpf_skb_ct_lookup.
 * @timeout      - New timeout in msecs.
 */
int bpf_ct_change_timeout(struct nf_conn *nfct, u32 timeout)
{
	return __nf_ct_change_timeout(nfct, msecs_to_jiffies(timeout));
}

/* bpf_ct_set_status - Set status field of allocated nf_conn
 *
 * Set the status field of the newly allocated nf_conn before insertion.
 * This must be invoked for referenced PTR_TO_BTF_ID to nf_conn___init.
 *
 * Parameters:
 * @nfct	 - Pointer to referenced nf_conn object, obtained using
 *		   bpf_xdp_ct_alloc or bpf_skb_ct_alloc.
 * @status       - New status value.
 */
int bpf_ct_set_status(const struct nf_conn___init *nfct, u32 status)
{
	return nf_ct_change_status_common((struct nf_conn *)nfct, status);
}

/* bpf_ct_change_status - Change status of inserted nf_conn
 *
 * Change the status field of the provided connection tracking entry.
 * This must be invoked for referenced PTR_TO_BTF_ID to nf_conn.
 *
 * Parameters:
 * @nfct	 - Pointer to referenced nf_conn object, obtained using
 *		   bpf_ct_insert_entry, bpf_xdp_ct_lookup or bpf_skb_ct_lookup.
 * @status       - New status value.
 */
int bpf_ct_change_status(struct nf_conn *nfct, u32 status)
{
	return nf_ct_change_status_common(nfct, status);
}

__diag_pop()

BTF_SET8_START(nf_ct_kfunc_set)
BTF_ID_FLAGS(func, bpf_xdp_ct_alloc, KF_ACQUIRE | KF_RET_NULL)
BTF_ID_FLAGS(func, bpf_xdp_ct_lookup, KF_ACQUIRE | KF_RET_NULL)
BTF_ID_FLAGS(func, bpf_skb_ct_alloc, KF_ACQUIRE | KF_RET_NULL)
BTF_ID_FLAGS(func, bpf_skb_ct_lookup, KF_ACQUIRE | KF_RET_NULL)
BTF_ID_FLAGS(func, bpf_ct_insert_entry, KF_ACQUIRE | KF_RET_NULL | KF_RELEASE)
BTF_ID_FLAGS(func, bpf_ct_release, KF_RELEASE)
BTF_ID_FLAGS(func, bpf_ct_set_timeout, KF_TRUSTED_ARGS)
BTF_ID_FLAGS(func, bpf_ct_change_timeout, KF_TRUSTED_ARGS)
BTF_ID_FLAGS(func, bpf_ct_set_status, KF_TRUSTED_ARGS)
BTF_ID_FLAGS(func, bpf_ct_change_status, KF_TRUSTED_ARGS)
BTF_SET8_END(nf_ct_kfunc_set)

static const struct btf_kfunc_id_set nf_conntrack_kfunc_set = {
	.owner = THIS_MODULE,
	.set   = &nf_ct_kfunc_set,
};

int register_nf_conntrack_bpf(void)
{
	int ret;

	ret = register_btf_kfunc_id_set(BPF_PROG_TYPE_XDP, &nf_conntrack_kfunc_set);
	ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_SCHED_CLS, &nf_conntrack_kfunc_set);
	if (!ret) {
		mutex_lock(&nf_conn_btf_access_lock);
		nfct_btf_struct_access = _nf_conntrack_btf_struct_access;
		mutex_unlock(&nf_conn_btf_access_lock);
	}

	return ret;
}

void cleanup_nf_conntrack_bpf(void)
{
	mutex_lock(&nf_conn_btf_access_lock);
	nfct_btf_struct_access = NULL;
	mutex_unlock(&nf_conn_btf_access_lock);
}
Commit	Line	Data
b4c2b959 KKD	1	// SPDX-License-Identifier: GPL-2.0-only
	2	/* Unstable Conntrack Helpers for XDP and TC-BPF hook
	3	*
	4	* These are called from the XDP and SCHED_CLS BPF programs. Note that it is
	5	* allowed to break compatibility for these functions since the interface they
	6	* are exposed through to BPF programs is explicitly unstable.
	7	*/
	8
864b656f	9	#include <linux/bpf_verifier.h>
b4c2b959 KKD	10	#include <linux/bpf.h>
b4c2b959 KKD	11	#include <linux/btf.h>
fdf21497	12	#include <linux/filter.h>
864b656f	13	#include <linux/mutex.h>
b4c2b959 KKD	14	#include <linux/types.h>
	15	#include <linux/btf_ids.h>
	16	#include <linux/net_namespace.h>
0b206c6d	17	#include <net/netfilter/nf_conntrack_bpf.h>
b4c2b959 KKD	18	#include <net/netfilter/nf_conntrack_core.h>
	19
	20	/* bpf_ct_opts - Options for CT lookup helpers
	21	*
	22	* Members:
	23	* @netns_id - Specify the network namespace for lookup
	24	* Values:
	25	* BPF_F_CURRENT_NETNS (-1)
	26	* Use namespace associated with ctx (xdp_md, __sk_buff)
	27	* [0, S32_MAX]
	28	* Network Namespace ID
	29	* @error - Out parameter, set for any errors encountered
	30	* Values:
	31	* -EINVAL - Passed NULL for bpf_tuple pointer
	32	* -EINVAL - opts->reserved is not 0
	33	* -EINVAL - netns_id is less than -1
	34	* -EINVAL - opts__sz isn't NF_BPF_CT_OPTS_SZ (12)
	35	* -EPROTO - l4proto isn't one of IPPROTO_TCP or IPPROTO_UDP
	36	* -ENONET - No network namespace found for netns_id
	37	* -ENOENT - Conntrack lookup could not find entry for tuple
	38	* -EAFNOSUPPORT - tuple__sz isn't one of sizeof(tuple->ipv4)
	39	* or sizeof(tuple->ipv6)
	40	* @l4proto - Layer 4 protocol
	41	* Values:
	42	* IPPROTO_TCP, IPPROTO_UDP
1963c740	43	* @dir: - connection tracking tuple direction.
b4c2b959 KKD	44	* @reserved - Reserved member, will be reused for more options in future
	45	* Values:
	46	* 0
	47	*/
	48	struct bpf_ct_opts {
	49	s32 netns_id;
	50	s32 error;
	51	u8 l4proto;
1963c740 LB	52	u8 dir;
1963c740 LB	53	u8 reserved[2];
b4c2b959 KKD	54	};
	55
	56	enum {
	57	NF_BPF_CT_OPTS_SZ = 12,
	58	};
	59
d7e79c97 LB	60	static int bpf_nf_ct_tuple_parse(struct bpf_sock_tuple *bpf_tuple,
	61	u32 tuple_len, u8 protonum, u8 dir,
	62	struct nf_conntrack_tuple *tuple)
	63	{
	64	union nf_inet_addr *src = dir ? &tuple->dst.u3 : &tuple->src.u3;
	65	union nf_inet_addr *dst = dir ? &tuple->src.u3 : &tuple->dst.u3;
	66	union nf_conntrack_man_proto sport = dir ? (void )&tuple->dst.u
	67	: &tuple->src.u;
	68	union nf_conntrack_man_proto *dport = dir ? &tuple->src.u
	69	: (void *)&tuple->dst.u;
	70
	71	if (unlikely(protonum != IPPROTO_TCP && protonum != IPPROTO_UDP))
	72	return -EPROTO;
	73
	74	memset(tuple, 0, sizeof(*tuple));
	75
	76	switch (tuple_len) {
	77	case sizeof(bpf_tuple->ipv4):
	78	tuple->src.l3num = AF_INET;
	79	src->ip = bpf_tuple->ipv4.saddr;
	80	sport->tcp.port = bpf_tuple->ipv4.sport;
	81	dst->ip = bpf_tuple->ipv4.daddr;
	82	dport->tcp.port = bpf_tuple->ipv4.dport;
	83	break;
	84	case sizeof(bpf_tuple->ipv6):
	85	tuple->src.l3num = AF_INET6;
	86	memcpy(src->ip6, bpf_tuple->ipv6.saddr, sizeof(bpf_tuple->ipv6.saddr));
	87	sport->tcp.port = bpf_tuple->ipv6.sport;
	88	memcpy(dst->ip6, bpf_tuple->ipv6.daddr, sizeof(bpf_tuple->ipv6.daddr));
	89	dport->tcp.port = bpf_tuple->ipv6.dport;
	90	break;
	91	default:
	92	return -EAFNOSUPPORT;
	93	}
	94	tuple->dst.protonum = protonum;
	95	tuple->dst.dir = dir;
	96
	97	return 0;
	98	}
	99
	100	static struct nf_conn *
	101	__bpf_nf_ct_alloc_entry(struct net net, struct bpf_sock_tuple bpf_tuple,
	102	u32 tuple_len, struct bpf_ct_opts *opts, u32 opts_len,
	103	u32 timeout)
	104	{
	105	struct nf_conntrack_tuple otuple, rtuple;
	106	struct nf_conn *ct;
	107	int err;
	108
	109	if (!opts \|\| !bpf_tuple \|\| opts->reserved[0] \|\| opts->reserved[1] \|\|
	110	opts_len != NF_BPF_CT_OPTS_SZ)
	111	return ERR_PTR(-EINVAL);
	112
	113	if (unlikely(opts->netns_id < BPF_F_CURRENT_NETNS))
	114	return ERR_PTR(-EINVAL);
	115
	116	err = bpf_nf_ct_tuple_parse(bpf_tuple, tuple_len, opts->l4proto,
	117	IP_CT_DIR_ORIGINAL, &otuple);
	118	if (err < 0)
	119	return ERR_PTR(err);
	120
	121	err = bpf_nf_ct_tuple_parse(bpf_tuple, tuple_len, opts->l4proto,
	122	IP_CT_DIR_REPLY, &rtuple);
	123	if (err < 0)
124	return ERR_PTR(err);
125
126	if (opts->netns_id >= 0) {
127	net = get_net_ns_by_id(net, opts->netns_id);
128	if (unlikely(!net))
129	return ERR_PTR(-ENONET);
130	}
131
132	ct = nf_conntrack_alloc(net, &nf_ct_zone_dflt, &otuple, &rtuple,
133	GFP_ATOMIC);
134	if (IS_ERR(ct))
135	goto out;
136
137	memset(&ct->proto, 0, sizeof(ct->proto));
138	__nf_ct_set_timeout(ct, timeout * HZ);
d7e79c97 LB	139
	140	out:
	141	if (opts->netns_id >= 0)
	142	put_net(net);
	143
	144	return ct;
	145	}
	146
b4c2b959 KKD	147	static struct nf_conn __bpf_nf_ct_lookup(struct net net,
b4c2b959 KKD	148	struct bpf_sock_tuple *bpf_tuple,
aed8ee7f KKD	149	u32 tuple_len, struct bpf_ct_opts *opts,
aed8ee7f KKD	150	u32 opts_len)
b4c2b959 KKD	151	{
	152	struct nf_conntrack_tuple_hash *hash;
	153	struct nf_conntrack_tuple tuple;
1963c740	154	struct nf_conn *ct;
d7e79c97	155	int err;
b4c2b959	156
aed8ee7f KKD	157	if (!opts \|\| !bpf_tuple \|\| opts->reserved[0] \|\| opts->reserved[1] \|\|
	158	opts_len != NF_BPF_CT_OPTS_SZ)
	159	return ERR_PTR(-EINVAL);
	160	if (unlikely(opts->l4proto != IPPROTO_TCP && opts->l4proto != IPPROTO_UDP))
b4c2b959	161	return ERR_PTR(-EPROTO);
aed8ee7f	162	if (unlikely(opts->netns_id < BPF_F_CURRENT_NETNS))
b4c2b959 KKD	163	return ERR_PTR(-EINVAL);
b4c2b959 KKD	164
d7e79c97 LB	165	err = bpf_nf_ct_tuple_parse(bpf_tuple, tuple_len, opts->l4proto,
	166	IP_CT_DIR_ORIGINAL, &tuple);
	167	if (err < 0)
	168	return ERR_PTR(err);
b4c2b959	169
aed8ee7f KKD	170	if (opts->netns_id >= 0) {
aed8ee7f KKD	171	net = get_net_ns_by_id(net, opts->netns_id);
b4c2b959 KKD	172	if (unlikely(!net))
	173	return ERR_PTR(-ENONET);
	174	}
	175
	176	hash = nf_conntrack_find_get(net, &nf_ct_zone_dflt, &tuple);
aed8ee7f	177	if (opts->netns_id >= 0)
b4c2b959 KKD	178	put_net(net);
	179	if (!hash)
	180	return ERR_PTR(-ENOENT);
1963c740 LB	181
1963c740 LB	182	ct = nf_ct_tuplehash_to_ctrack(hash);
aed8ee7f	183	opts->dir = NF_CT_DIRECTION(hash);
1963c740 LB	184
1963c740 LB	185	return ct;
b4c2b959 KKD	186	}
b4c2b959 KKD	187
864b656f DX	188	BTF_ID_LIST(btf_nf_conn_ids)
	189	BTF_ID(struct, nf_conn)
	190	BTF_ID(struct, nf_conn___init)
	191
	192	/* Check writes into `struct nf_conn` */
	193	static int _nf_conntrack_btf_struct_access(struct bpf_verifier_log *log,
	194	const struct btf *btf,
	195	const struct btf_type *t, int off,
	196	int size, enum bpf_access_type atype,
	197	u32 *next_btf_id,
	198	enum bpf_type_flag *flag)
	199	{
	200	const struct btf_type *ncit;
	201	const struct btf_type *nct;
	202	size_t end;
	203
	204	ncit = btf_type_by_id(btf, btf_nf_conn_ids[1]);
	205	nct = btf_type_by_id(btf, btf_nf_conn_ids[0]);
	206
	207	if (t != nct && t != ncit) {
	208	bpf_log(log, "only read is supported\n");
	209	return -EACCES;
	210	}
	211
	212	/* `struct nf_conn` and `struct nf_conn___init` have the same layout
	213	* so we are safe to simply merge offset checks here
	214	*/
	215	switch (off) {
	216	#if defined(CONFIG_NF_CONNTRACK_MARK)
	217	case offsetof(struct nf_conn, mark):
	218	end = offsetofend(struct nf_conn, mark);
	219	break;
	220	#endif
	221	default:
	222	bpf_log(log, "no write support to nf_conn at off %d\n", off);
	223	return -EACCES;
	224	}
	225
	226	if (off + size > end) {
	227	bpf_log(log,
	228	"write access at off %d with size %d beyond the member of nf_conn ended at %zu\n",
	229	off, size, end);
	230	return -EACCES;
	231	}
	232
	233	return 0;
	234	}
	235
b4c2b959	236	__diag_push();
0b206c6d KKD	237	__diag_ignore_all("-Wmissing-prototypes",
0b206c6d KKD	238	"Global functions as their definitions will be in nf_conntrack BTF");
b4c2b959	239
d7e79c97 LB	240	/* bpf_xdp_ct_alloc - Allocate a new CT entry
	241	*
	242	* Parameters:
	243	* @xdp_ctx - Pointer to ctx (xdp_md) in XDP program
	244	* Cannot be NULL
	245	* @bpf_tuple - Pointer to memory representing the tuple to look up
	246	* Cannot be NULL
	247	* @tuple__sz - Length of the tuple structure
	248	* Must be one of sizeof(bpf_tuple->ipv4) or
	249	* sizeof(bpf_tuple->ipv6)
	250	* @opts - Additional options for allocation (documented above)
	251	* Cannot be NULL
	252	* @opts__sz - Length of the bpf_ct_opts structure
	253	* Must be NF_BPF_CT_OPTS_SZ (12)
	254	*/
	255	struct nf_conn___init *
	256	bpf_xdp_ct_alloc(struct xdp_md xdp_ctx, struct bpf_sock_tuple bpf_tuple,
	257	u32 tuple__sz, struct bpf_ct_opts *opts, u32 opts__sz)
	258	{
	259	struct xdp_buff ctx = (struct xdp_buff )xdp_ctx;
	260	struct nf_conn *nfct;
	261
	262	nfct = __bpf_nf_ct_alloc_entry(dev_net(ctx->rxq->dev), bpf_tuple, tuple__sz,
	263	opts, opts__sz, 10);
	264	if (IS_ERR(nfct)) {
	265	if (opts)
	266	opts->error = PTR_ERR(nfct);
	267	return NULL;
	268	}
	269
	270	return (struct nf_conn___init *)nfct;
	271	}
	272
b4c2b959 KKD	273	/* bpf_xdp_ct_lookup - Lookup CT entry for the given tuple, and acquire a
	274	* reference to it
	275	*
	276	* Parameters:
	277	* @xdp_ctx - Pointer to ctx (xdp_md) in XDP program
	278	* Cannot be NULL
	279	* @bpf_tuple - Pointer to memory representing the tuple to look up
	280	* Cannot be NULL
	281	* @tuple__sz - Length of the tuple structure
	282	* Must be one of sizeof(bpf_tuple->ipv4) or
	283	* sizeof(bpf_tuple->ipv6)
	284	* @opts - Additional options for lookup (documented above)
	285	* Cannot be NULL
	286	* @opts__sz - Length of the bpf_ct_opts structure
	287	* Must be NF_BPF_CT_OPTS_SZ (12)
	288	*/
	289	struct nf_conn *
	290	bpf_xdp_ct_lookup(struct xdp_md xdp_ctx, struct bpf_sock_tuple bpf_tuple,
	291	u32 tuple__sz, struct bpf_ct_opts *opts, u32 opts__sz)
	292	{
	293	struct xdp_buff ctx = (struct xdp_buff )xdp_ctx;
	294	struct net *caller_net;
	295	struct nf_conn *nfct;
	296
b4c2b959	297	caller_net = dev_net(ctx->rxq->dev);
aed8ee7f	298	nfct = __bpf_nf_ct_lookup(caller_net, bpf_tuple, tuple__sz, opts, opts__sz);
b4c2b959	299	if (IS_ERR(nfct)) {
aed8ee7f KKD	300	if (opts)
aed8ee7f KKD	301	opts->error = PTR_ERR(nfct);
b4c2b959 KKD	302	return NULL;
	303	}
	304	return nfct;
	305	}
	306
d7e79c97 LB	307	/* bpf_skb_ct_alloc - Allocate a new CT entry
	308	*
	309	* Parameters:
	310	* @skb_ctx - Pointer to ctx (__sk_buff) in TC program
	311	* Cannot be NULL
	312	* @bpf_tuple - Pointer to memory representing the tuple to look up
	313	* Cannot be NULL
	314	* @tuple__sz - Length of the tuple structure
	315	* Must be one of sizeof(bpf_tuple->ipv4) or
	316	* sizeof(bpf_tuple->ipv6)
	317	* @opts - Additional options for allocation (documented above)
	318	* Cannot be NULL
	319	* @opts__sz - Length of the bpf_ct_opts structure
	320	* Must be NF_BPF_CT_OPTS_SZ (12)
	321	*/
	322	struct nf_conn___init *
	323	bpf_skb_ct_alloc(struct __sk_buff skb_ctx, struct bpf_sock_tuple bpf_tuple,
	324	u32 tuple__sz, struct bpf_ct_opts *opts, u32 opts__sz)
	325	{
	326	struct sk_buff skb = (struct sk_buff )skb_ctx;
	327	struct nf_conn *nfct;
	328	struct net *net;
	329
	330	net = skb->dev ? dev_net(skb->dev) : sock_net(skb->sk);
	331	nfct = __bpf_nf_ct_alloc_entry(net, bpf_tuple, tuple__sz, opts, opts__sz, 10);
	332	if (IS_ERR(nfct)) {
	333	if (opts)
	334	opts->error = PTR_ERR(nfct);
	335	return NULL;
	336	}
	337
	338	return (struct nf_conn___init *)nfct;
	339	}
	340
b4c2b959 KKD	341	/* bpf_skb_ct_lookup - Lookup CT entry for the given tuple, and acquire a
	342	* reference to it
	343	*
	344	* Parameters:
	345	* @skb_ctx - Pointer to ctx (__sk_buff) in TC program
	346	* Cannot be NULL
	347	* @bpf_tuple - Pointer to memory representing the tuple to look up
	348	* Cannot be NULL
	349	* @tuple__sz - Length of the tuple structure
	350	* Must be one of sizeof(bpf_tuple->ipv4) or
	351	* sizeof(bpf_tuple->ipv6)
	352	* @opts - Additional options for lookup (documented above)
	353	* Cannot be NULL
	354	* @opts__sz - Length of the bpf_ct_opts structure
	355	* Must be NF_BPF_CT_OPTS_SZ (12)
	356	*/
	357	struct nf_conn *
	358	bpf_skb_ct_lookup(struct __sk_buff skb_ctx, struct bpf_sock_tuple bpf_tuple,
	359	u32 tuple__sz, struct bpf_ct_opts *opts, u32 opts__sz)
	360	{
	361	struct sk_buff skb = (struct sk_buff )skb_ctx;
	362	struct net *caller_net;
	363	struct nf_conn *nfct;
	364
b4c2b959	365	caller_net = skb->dev ? dev_net(skb->dev) : sock_net(skb->sk);
aed8ee7f	366	nfct = __bpf_nf_ct_lookup(caller_net, bpf_tuple, tuple__sz, opts, opts__sz);
b4c2b959	367	if (IS_ERR(nfct)) {
aed8ee7f KKD	368	if (opts)
aed8ee7f KKD	369	opts->error = PTR_ERR(nfct);
b4c2b959 KKD	370	return NULL;
	371	}
	372	return nfct;
	373	}
	374
d7e79c97 LB	375	/* bpf_ct_insert_entry - Add the provided entry into a CT map
	376	*
	377	* This must be invoked for referenced PTR_TO_BTF_ID.
	378	*
0b389236	379	* @nfct - Pointer to referenced nf_conn___init object, obtained
d7e79c97 LB	380	* using bpf_xdp_ct_alloc or bpf_skb_ct_alloc.
d7e79c97 LB	381	*/
0b389236	382	struct nf_conn bpf_ct_insert_entry(struct nf_conn___init nfct_i)
d7e79c97	383	{
0b389236	384	struct nf_conn nfct = (struct nf_conn )nfct_i;
d7e79c97 LB	385	int err;
d7e79c97 LB	386
0fabd2aa	387	nfct->status \|= IPS_CONFIRMED;
d7e79c97 LB	388	err = nf_conntrack_hash_check_insert(nfct);
	389	if (err < 0) {
	390	nf_conntrack_free(nfct);
	391	return NULL;
	392	}
	393	return nfct;
	394	}
	395
b4c2b959 KKD	396	/* bpf_ct_release - Release acquired nf_conn object
	397	*
	398	* This must be invoked for referenced PTR_TO_BTF_ID, and the verifier rejects
	399	* the program if any references remain in the program in all of the explored
	400	* states.
	401	*
	402	* Parameters:
	403	* @nf_conn - Pointer to referenced nf_conn object, obtained using
	404	* bpf_xdp_ct_lookup or bpf_skb_ct_lookup.
	405	*/
	406	void bpf_ct_release(struct nf_conn *nfct)
	407	{
	408	if (!nfct)
	409	return;
	410	nf_ct_put(nfct);
	411	}
	412
0b389236 KKD	413	/* bpf_ct_set_timeout - Set timeout of allocated nf_conn
	414	*
	415	* Sets the default timeout of newly allocated nf_conn before insertion.
	416	* This helper must be invoked for refcounted pointer to nf_conn___init.
	417	*
	418	* Parameters:
	419	* @nfct - Pointer to referenced nf_conn object, obtained using
	420	* bpf_xdp_ct_alloc or bpf_skb_ct_alloc.
	421	* @timeout - Timeout in msecs.
	422	*/
	423	void bpf_ct_set_timeout(struct nf_conn___init *nfct, u32 timeout)
	424	{
	425	__nf_ct_set_timeout((struct nf_conn *)nfct, msecs_to_jiffies(timeout));
	426	}
	427
	428	/* bpf_ct_change_timeout - Change timeout of inserted nf_conn
	429	*
	430	* Change timeout associated of the inserted or looked up nf_conn.
	431	* This helper must be invoked for refcounted pointer to nf_conn.
	432	*
	433	* Parameters:
	434	* @nfct - Pointer to referenced nf_conn object, obtained using
	435	* bpf_ct_insert_entry, bpf_xdp_ct_lookup, or bpf_skb_ct_lookup.
	436	* @timeout - New timeout in msecs.
	437	*/
	438	int bpf_ct_change_timeout(struct nf_conn *nfct, u32 timeout)
	439	{
	440	return __nf_ct_change_timeout(nfct, msecs_to_jiffies(timeout));
	441	}
	442
ef69aa3a LB	443	/* bpf_ct_set_status - Set status field of allocated nf_conn
	444	*
	445	* Set the status field of the newly allocated nf_conn before insertion.
	446	* This must be invoked for referenced PTR_TO_BTF_ID to nf_conn___init.
	447	*
	448	* Parameters:
	449	* @nfct - Pointer to referenced nf_conn object, obtained using
	450	* bpf_xdp_ct_alloc or bpf_skb_ct_alloc.
	451	* @status - New status value.
	452	*/
	453	int bpf_ct_set_status(const struct nf_conn___init *nfct, u32 status)
	454	{
	455	return nf_ct_change_status_common((struct nf_conn *)nfct, status);
	456	}
	457
	458	/* bpf_ct_change_status - Change status of inserted nf_conn
	459	*
	460	* Change the status field of the provided connection tracking entry.
	461	* This must be invoked for referenced PTR_TO_BTF_ID to nf_conn.
	462	*
	463	* Parameters:
	464	* @nfct - Pointer to referenced nf_conn object, obtained using
	465	* bpf_ct_insert_entry, bpf_xdp_ct_lookup or bpf_skb_ct_lookup.
	466	* @status - New status value.
	467	*/
	468	int bpf_ct_change_status(struct nf_conn *nfct, u32 status)
	469	{
	470	return nf_ct_change_status_common(nfct, status);
	471	}
	472
b4c2b959 KKD	473	__diag_pop()
b4c2b959 KKD	474
a4703e31	475	BTF_SET8_START(nf_ct_kfunc_set)
d7e79c97	476	BTF_ID_FLAGS(func, bpf_xdp_ct_alloc, KF_ACQUIRE \| KF_RET_NULL)
a4703e31	477	BTF_ID_FLAGS(func, bpf_xdp_ct_lookup, KF_ACQUIRE \| KF_RET_NULL)
d7e79c97	478	BTF_ID_FLAGS(func, bpf_skb_ct_alloc, KF_ACQUIRE \| KF_RET_NULL)
a4703e31	479	BTF_ID_FLAGS(func, bpf_skb_ct_lookup, KF_ACQUIRE \| KF_RET_NULL)
d7e79c97	480	BTF_ID_FLAGS(func, bpf_ct_insert_entry, KF_ACQUIRE \| KF_RET_NULL \| KF_RELEASE)
a4703e31	481	BTF_ID_FLAGS(func, bpf_ct_release, KF_RELEASE)
0b389236 KKD	482	BTF_ID_FLAGS(func, bpf_ct_set_timeout, KF_TRUSTED_ARGS)
0b389236 KKD	483	BTF_ID_FLAGS(func, bpf_ct_change_timeout, KF_TRUSTED_ARGS)
ef69aa3a LB	484	BTF_ID_FLAGS(func, bpf_ct_set_status, KF_TRUSTED_ARGS)
ef69aa3a LB	485	BTF_ID_FLAGS(func, bpf_ct_change_status, KF_TRUSTED_ARGS)
a4703e31 KKD	486	BTF_SET8_END(nf_ct_kfunc_set)
	487
	488	static const struct btf_kfunc_id_set nf_conntrack_kfunc_set = {
	489	.owner = THIS_MODULE,
	490	.set = &nf_ct_kfunc_set,
b4c2b959 KKD	491	};
	492
	493	int register_nf_conntrack_bpf(void)
	494	{
	495	int ret;
	496
a4703e31	497	ret = register_btf_kfunc_id_set(BPF_PROG_TYPE_XDP, &nf_conntrack_kfunc_set);
864b656f DX	498	ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_SCHED_CLS, &nf_conntrack_kfunc_set);
	499	if (!ret) {
	500	mutex_lock(&nf_conn_btf_access_lock);
5a090aa3	501	nfct_btf_struct_access = _nf_conntrack_btf_struct_access;
864b656f DX	502	mutex_unlock(&nf_conn_btf_access_lock);
	503	}
	504
	505	return ret;
	506	}
	507
	508	void cleanup_nf_conntrack_bpf(void)
	509	{
	510	mutex_lock(&nf_conn_btf_access_lock);
5a090aa3	511	nfct_btf_struct_access = NULL;
864b656f	512	mutex_unlock(&nf_conn_btf_access_lock);
b4c2b959	513	}