[mirror_ubuntu-jammy-kernel.git] / net / core / lwt_bpf.c

// SPDX-License-Identifier: GPL-2.0-only
/* Copyright (c) 2016 Thomas Graf <tgraf@tgraf.ch>
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/skbuff.h>
#include <linux/types.h>
#include <linux/bpf.h>
#include <net/lwtunnel.h>
#include <net/gre.h>
#include <net/ip6_route.h>
#include <net/ipv6_stubs.h>

struct bpf_lwt_prog {
	struct bpf_prog *prog;
	char *name;
};

struct bpf_lwt {
	struct bpf_lwt_prog in;
	struct bpf_lwt_prog out;
	struct bpf_lwt_prog xmit;
	int family;
};

#define MAX_PROG_NAME 256

static inline struct bpf_lwt *bpf_lwt_lwtunnel(struct lwtunnel_state *lwt)
{
	return (struct bpf_lwt *)lwt->data;
}

#define NO_REDIRECT false
#define CAN_REDIRECT true

static int run_lwt_bpf(struct sk_buff *skb, struct bpf_lwt_prog *lwt,
		       struct dst_entry *dst, bool can_redirect)
{
	int ret;

	/* Preempt disable is needed to protect per-cpu redirect_info between
	 * BPF prog and skb_do_redirect(). The call_rcu in bpf_prog_put() and
	 * access to maps strictly require a rcu_read_lock() for protection,
	 * mixing with BH RCU lock doesn't work.
	 */
	preempt_disable();
	bpf_compute_data_pointers(skb);
	ret = bpf_prog_run_save_cb(lwt->prog, skb);

	switch (ret) {
	case BPF_OK:
	case BPF_LWT_REROUTE:
		break;

	case BPF_REDIRECT:
		if (unlikely(!can_redirect)) {
			pr_warn_once("Illegal redirect return code in prog %s\n",
				     lwt->name ? : "<unknown>");
			ret = BPF_OK;
		} else {
			skb_reset_mac_header(skb);
			ret = skb_do_redirect(skb);
			if (ret == 0)
				ret = BPF_REDIRECT;
		}
		break;

	case BPF_DROP:
		kfree_skb(skb);
		ret = -EPERM;
		break;

	default:
		pr_warn_once("bpf-lwt: Illegal return value %u, expect packet loss\n", ret);
		kfree_skb(skb);
		ret = -EINVAL;
		break;
	}

	preempt_enable();

	return ret;
}

static int bpf_lwt_input_reroute(struct sk_buff *skb)
{
	int err = -EINVAL;

	if (skb->protocol == htons(ETH_P_IP)) {
		struct iphdr *iph = ip_hdr(skb);

		err = ip_route_input_noref(skb, iph->daddr, iph->saddr,
					   iph->tos, skb_dst(skb)->dev);
	} else if (skb->protocol == htons(ETH_P_IPV6)) {
		err = ipv6_stub->ipv6_route_input(skb);
	} else {
		err = -EAFNOSUPPORT;
	}

	if (err)
		goto err;
	return dst_input(skb);

err:
	kfree_skb(skb);
	return err;
}

static int bpf_input(struct sk_buff *skb)
{
	struct dst_entry *dst = skb_dst(skb);
	struct bpf_lwt *bpf;
	int ret;

	bpf = bpf_lwt_lwtunnel(dst->lwtstate);
	if (bpf->in.prog) {
		ret = run_lwt_bpf(skb, &bpf->in, dst, NO_REDIRECT);
		if (ret < 0)
			return ret;
		if (ret == BPF_LWT_REROUTE)
			return bpf_lwt_input_reroute(skb);
	}

	if (unlikely(!dst->lwtstate->orig_input)) {
		kfree_skb(skb);
		return -EINVAL;
	}

	return dst->lwtstate->orig_input(skb);
}

static int bpf_output(struct net *net, struct sock *sk, struct sk_buff *skb)
{
	struct dst_entry *dst = skb_dst(skb);
	struct bpf_lwt *bpf;
	int ret;

	bpf = bpf_lwt_lwtunnel(dst->lwtstate);
	if (bpf->out.prog) {
		ret = run_lwt_bpf(skb, &bpf->out, dst, NO_REDIRECT);
		if (ret < 0)
			return ret;
	}

	if (unlikely(!dst->lwtstate->orig_output)) {
		pr_warn_once("orig_output not set on dst for prog %s\n",
			     bpf->out.name);
		kfree_skb(skb);
		return -EINVAL;
	}

	return dst->lwtstate->orig_output(net, sk, skb);
}

static int xmit_check_hhlen(struct sk_buff *skb)
{
	int hh_len = skb_dst(skb)->dev->hard_header_len;

	if (skb_headroom(skb) < hh_len) {
		int nhead = HH_DATA_ALIGN(hh_len - skb_headroom(skb));

		if (pskb_expand_head(skb, nhead, 0, GFP_ATOMIC))
			return -ENOMEM;
	}

	return 0;
}

static int bpf_lwt_xmit_reroute(struct sk_buff *skb)
{
	struct net_device *l3mdev = l3mdev_master_dev_rcu(skb_dst(skb)->dev);
	int oif = l3mdev ? l3mdev->ifindex : 0;
	struct dst_entry *dst = NULL;
	int err = -EAFNOSUPPORT;
	struct sock *sk;
	struct net *net;
	bool ipv4;

	if (skb->protocol == htons(ETH_P_IP))
		ipv4 = true;
	else if (skb->protocol == htons(ETH_P_IPV6))
		ipv4 = false;
	else
		goto err;

	sk = sk_to_full_sk(skb->sk);
	if (sk) {
		if (sk->sk_bound_dev_if)
			oif = sk->sk_bound_dev_if;
		net = sock_net(sk);
	} else {
		net = dev_net(skb_dst(skb)->dev);
	}

	if (ipv4) {
		struct iphdr *iph = ip_hdr(skb);
		struct flowi4 fl4 = {};
		struct rtable *rt;

		fl4.flowi4_oif = oif;
		fl4.flowi4_mark = skb->mark;
		fl4.flowi4_uid = sock_net_uid(net, sk);
		fl4.flowi4_tos = RT_TOS(iph->tos);
		fl4.flowi4_flags = FLOWI_FLAG_ANYSRC;
		fl4.flowi4_proto = iph->protocol;
		fl4.daddr = iph->daddr;
		fl4.saddr = iph->saddr;

		rt = ip_route_output_key(net, &fl4);
		if (IS_ERR(rt)) {
			err = PTR_ERR(rt);
			goto err;
		}
		dst = &rt->dst;
	} else {
		struct ipv6hdr *iph6 = ipv6_hdr(skb);
		struct flowi6 fl6 = {};

		fl6.flowi6_oif = oif;
		fl6.flowi6_mark = skb->mark;
		fl6.flowi6_uid = sock_net_uid(net, sk);
		fl6.flowlabel = ip6_flowinfo(iph6);
		fl6.flowi6_proto = iph6->nexthdr;
		fl6.daddr = iph6->daddr;
		fl6.saddr = iph6->saddr;

		err = ipv6_stub->ipv6_dst_lookup(net, skb->sk, &dst, &fl6);
		if (unlikely(err))
			goto err;
		if (IS_ERR(dst)) {
			err = PTR_ERR(dst);
			goto err;
		}
	}
	if (unlikely(dst->error)) {
		err = dst->error;
		dst_release(dst);
		goto err;
	}

	/* Although skb header was reserved in bpf_lwt_push_ip_encap(), it
	 * was done for the previous dst, so we are doing it here again, in
	 * case the new dst needs much more space. The call below is a noop
	 * if there is enough header space in skb.
	 */
	err = skb_cow_head(skb, LL_RESERVED_SPACE(dst->dev));
	if (unlikely(err))
		goto err;

	skb_dst_drop(skb);
	skb_dst_set(skb, dst);

	err = dst_output(dev_net(skb_dst(skb)->dev), skb->sk, skb);
	if (unlikely(err))
		return err;

	/* ip[6]_finish_output2 understand LWTUNNEL_XMIT_DONE */
	return LWTUNNEL_XMIT_DONE;

err:
	kfree_skb(skb);
	return err;
}

static int bpf_xmit(struct sk_buff *skb)
{
	struct dst_entry *dst = skb_dst(skb);
	struct bpf_lwt *bpf;

	bpf = bpf_lwt_lwtunnel(dst->lwtstate);
	if (bpf->xmit.prog) {
		__be16 proto = skb->protocol;
		int ret;

		ret = run_lwt_bpf(skb, &bpf->xmit, dst, CAN_REDIRECT);
		switch (ret) {
		case BPF_OK:
			/* If the header changed, e.g. via bpf_lwt_push_encap,
			 * BPF_LWT_REROUTE below should have been used if the
			 * protocol was also changed.
			 */
			if (skb->protocol != proto) {
				kfree_skb(skb);
				return -EINVAL;
			}
			/* If the header was expanded, headroom might be too
			 * small for L2 header to come, expand as needed.
			 */
			ret = xmit_check_hhlen(skb);
			if (unlikely(ret))
				return ret;

			return LWTUNNEL_XMIT_CONTINUE;
		case BPF_REDIRECT:
			return LWTUNNEL_XMIT_DONE;
		case BPF_LWT_REROUTE:
			return bpf_lwt_xmit_reroute(skb);
		default:
			return ret;
		}
	}

	return LWTUNNEL_XMIT_CONTINUE;
}

static void bpf_lwt_prog_destroy(struct bpf_lwt_prog *prog)
{
	if (prog->prog)
		bpf_prog_put(prog->prog);

	kfree(prog->name);
}

static void bpf_destroy_state(struct lwtunnel_state *lwt)
{
	struct bpf_lwt *bpf = bpf_lwt_lwtunnel(lwt);

	bpf_lwt_prog_destroy(&bpf->in);
	bpf_lwt_prog_destroy(&bpf->out);
	bpf_lwt_prog_destroy(&bpf->xmit);
}

static const struct nla_policy bpf_prog_policy[LWT_BPF_PROG_MAX + 1] = {
	[LWT_BPF_PROG_FD]   = { .type = NLA_U32, },
	[LWT_BPF_PROG_NAME] = { .type = NLA_NUL_STRING,
				.len = MAX_PROG_NAME },
};

static int bpf_parse_prog(struct nlattr *attr, struct bpf_lwt_prog *prog,
			  enum bpf_prog_type type)
{
	struct nlattr *tb[LWT_BPF_PROG_MAX + 1];
	struct bpf_prog *p;
	int ret;
	u32 fd;

	ret = nla_parse_nested_deprecated(tb, LWT_BPF_PROG_MAX, attr,
					  bpf_prog_policy, NULL);
	if (ret < 0)
		return ret;

	if (!tb[LWT_BPF_PROG_FD] || !tb[LWT_BPF_PROG_NAME])
		return -EINVAL;

	prog->name = nla_memdup(tb[LWT_BPF_PROG_NAME], GFP_ATOMIC);
	if (!prog->name)
		return -ENOMEM;

	fd = nla_get_u32(tb[LWT_BPF_PROG_FD]);
	p = bpf_prog_get_type(fd, type);
	if (IS_ERR(p))
		return PTR_ERR(p);

	prog->prog = p;

	return 0;
}

static const struct nla_policy bpf_nl_policy[LWT_BPF_MAX + 1] = {
	[LWT_BPF_IN]		= { .type = NLA_NESTED, },
	[LWT_BPF_OUT]		= { .type = NLA_NESTED, },
	[LWT_BPF_XMIT]		= { .type = NLA_NESTED, },
	[LWT_BPF_XMIT_HEADROOM]	= { .type = NLA_U32 },
};

static int bpf_build_state(struct nlattr *nla,
			   unsigned int family, const void *cfg,
			   struct lwtunnel_state **ts,
			   struct netlink_ext_ack *extack)
{
	struct nlattr *tb[LWT_BPF_MAX + 1];
	struct lwtunnel_state *newts;
	struct bpf_lwt *bpf;
	int ret;

	if (family != AF_INET && family != AF_INET6)
		return -EAFNOSUPPORT;

	ret = nla_parse_nested_deprecated(tb, LWT_BPF_MAX, nla, bpf_nl_policy,
					  extack);
	if (ret < 0)
		return ret;

	if (!tb[LWT_BPF_IN] && !tb[LWT_BPF_OUT] && !tb[LWT_BPF_XMIT])
		return -EINVAL;

	newts = lwtunnel_state_alloc(sizeof(*bpf));
	if (!newts)
		return -ENOMEM;

	newts->type = LWTUNNEL_ENCAP_BPF;
	bpf = bpf_lwt_lwtunnel(newts);

	if (tb[LWT_BPF_IN]) {
		newts->flags |= LWTUNNEL_STATE_INPUT_REDIRECT;
		ret = bpf_parse_prog(tb[LWT_BPF_IN], &bpf->in,
				     BPF_PROG_TYPE_LWT_IN);
		if (ret  < 0)
			goto errout;
	}

	if (tb[LWT_BPF_OUT]) {
		newts->flags |= LWTUNNEL_STATE_OUTPUT_REDIRECT;
		ret = bpf_parse_prog(tb[LWT_BPF_OUT], &bpf->out,
				     BPF_PROG_TYPE_LWT_OUT);
		if (ret < 0)
			goto errout;
	}

	if (tb[LWT_BPF_XMIT]) {
		newts->flags |= LWTUNNEL_STATE_XMIT_REDIRECT;
		ret = bpf_parse_prog(tb[LWT_BPF_XMIT], &bpf->xmit,
				     BPF_PROG_TYPE_LWT_XMIT);
		if (ret < 0)
			goto errout;
	}

	if (tb[LWT_BPF_XMIT_HEADROOM]) {
		u32 headroom = nla_get_u32(tb[LWT_BPF_XMIT_HEADROOM]);

		if (headroom > LWT_BPF_MAX_HEADROOM) {
			ret = -ERANGE;
			goto errout;
		}

		newts->headroom = headroom;
	}

	bpf->family = family;
	*ts = newts;

	return 0;

errout:
	bpf_destroy_state(newts);
	kfree(newts);
	return ret;
}

static int bpf_fill_lwt_prog(struct sk_buff *skb, int attr,
			     struct bpf_lwt_prog *prog)
{
	struct nlattr *nest;

	if (!prog->prog)
		return 0;

	nest = nla_nest_start_noflag(skb, attr);
	if (!nest)
		return -EMSGSIZE;

	if (prog->name &&
	    nla_put_string(skb, LWT_BPF_PROG_NAME, prog->name))
		return -EMSGSIZE;

	return nla_nest_end(skb, nest);
}

static int bpf_fill_encap_info(struct sk_buff *skb, struct lwtunnel_state *lwt)
{
	struct bpf_lwt *bpf = bpf_lwt_lwtunnel(lwt);

	if (bpf_fill_lwt_prog(skb, LWT_BPF_IN, &bpf->in) < 0 ||
	    bpf_fill_lwt_prog(skb, LWT_BPF_OUT, &bpf->out) < 0 ||
	    bpf_fill_lwt_prog(skb, LWT_BPF_XMIT, &bpf->xmit) < 0)
		return -EMSGSIZE;

	return 0;
}

static int bpf_encap_nlsize(struct lwtunnel_state *lwtstate)
{
	int nest_len = nla_total_size(sizeof(struct nlattr)) +
		       nla_total_size(MAX_PROG_NAME) + /* LWT_BPF_PROG_NAME */
		       0;

	return nest_len + /* LWT_BPF_IN */
	       nest_len + /* LWT_BPF_OUT */
	       nest_len + /* LWT_BPF_XMIT */
	       0;
}

static int bpf_lwt_prog_cmp(struct bpf_lwt_prog *a, struct bpf_lwt_prog *b)
{
	/* FIXME:
	 * The LWT state is currently rebuilt for delete requests which
	 * results in a new bpf_prog instance. Comparing names for now.
	 */
	if (!a->name && !b->name)
		return 0;

	if (!a->name || !b->name)
		return 1;

	return strcmp(a->name, b->name);
}

static int bpf_encap_cmp(struct lwtunnel_state *a, struct lwtunnel_state *b)
{
	struct bpf_lwt *a_bpf = bpf_lwt_lwtunnel(a);
	struct bpf_lwt *b_bpf = bpf_lwt_lwtunnel(b);

	return bpf_lwt_prog_cmp(&a_bpf->in, &b_bpf->in) ||
	       bpf_lwt_prog_cmp(&a_bpf->out, &b_bpf->out) ||
	       bpf_lwt_prog_cmp(&a_bpf->xmit, &b_bpf->xmit);
}

static const struct lwtunnel_encap_ops bpf_encap_ops = {
	.build_state	= bpf_build_state,
	.destroy_state	= bpf_destroy_state,
	.input		= bpf_input,
	.output		= bpf_output,
	.xmit		= bpf_xmit,
	.fill_encap	= bpf_fill_encap_info,
	.get_encap_size = bpf_encap_nlsize,
	.cmp_encap	= bpf_encap_cmp,
	.owner		= THIS_MODULE,
};

static int handle_gso_type(struct sk_buff *skb, unsigned int gso_type,
			   int encap_len)
{
	struct skb_shared_info *shinfo = skb_shinfo(skb);

	gso_type |= SKB_GSO_DODGY;
	shinfo->gso_type |= gso_type;
	skb_decrease_gso_size(shinfo, encap_len);
	shinfo->gso_segs = 0;
	return 0;
}

static int handle_gso_encap(struct sk_buff *skb, bool ipv4, int encap_len)
{
	int next_hdr_offset;
	void *next_hdr;
	__u8 protocol;

	/* SCTP and UDP_L4 gso need more nuanced handling than what
	 * handle_gso_type() does above: skb_decrease_gso_size() is not enough.
	 * So at the moment only TCP GSO packets are let through.
	 */
	if (!(skb_shinfo(skb)->gso_type & (SKB_GSO_TCPV4 | SKB_GSO_TCPV6)))
		return -ENOTSUPP;

	if (ipv4) {
		protocol = ip_hdr(skb)->protocol;
		next_hdr_offset = sizeof(struct iphdr);
		next_hdr = skb_network_header(skb) + next_hdr_offset;
	} else {
		protocol = ipv6_hdr(skb)->nexthdr;
		next_hdr_offset = sizeof(struct ipv6hdr);
		next_hdr = skb_network_header(skb) + next_hdr_offset;
	}

	switch (protocol) {
	case IPPROTO_GRE:
		next_hdr_offset += sizeof(struct gre_base_hdr);
		if (next_hdr_offset > encap_len)
			return -EINVAL;

		if (((struct gre_base_hdr *)next_hdr)->flags & GRE_CSUM)
			return handle_gso_type(skb, SKB_GSO_GRE_CSUM,
					       encap_len);
		return handle_gso_type(skb, SKB_GSO_GRE, encap_len);

	case IPPROTO_UDP:
		next_hdr_offset += sizeof(struct udphdr);
		if (next_hdr_offset > encap_len)
			return -EINVAL;

		if (((struct udphdr *)next_hdr)->check)
			return handle_gso_type(skb, SKB_GSO_UDP_TUNNEL_CSUM,
					       encap_len);
		return handle_gso_type(skb, SKB_GSO_UDP_TUNNEL, encap_len);

	case IPPROTO_IP:
	case IPPROTO_IPV6:
		if (ipv4)
			return handle_gso_type(skb, SKB_GSO_IPXIP4, encap_len);
		else
			return handle_gso_type(skb, SKB_GSO_IPXIP6, encap_len);

	default:
		return -EPROTONOSUPPORT;
	}
}

int bpf_lwt_push_ip_encap(struct sk_buff *skb, void *hdr, u32 len, bool ingress)
{
	struct iphdr *iph;
	bool ipv4;
	int err;

	if (unlikely(len < sizeof(struct iphdr) || len > LWT_BPF_MAX_HEADROOM))
		return -EINVAL;

	/* validate protocol and length */
	iph = (struct iphdr *)hdr;
	if (iph->version == 4) {
		ipv4 = true;
		if (unlikely(len < iph->ihl * 4))
			return -EINVAL;
	} else if (iph->version == 6) {
		ipv4 = false;
		if (unlikely(len < sizeof(struct ipv6hdr)))
			return -EINVAL;
	} else {
		return -EINVAL;
	}

	if (ingress)
		err = skb_cow_head(skb, len + skb->mac_len);
	else
		err = skb_cow_head(skb,
				   len + LL_RESERVED_SPACE(skb_dst(skb)->dev));
	if (unlikely(err))
		return err;

	/* push the encap headers and fix pointers */
	skb_reset_inner_headers(skb);
	skb_reset_inner_mac_header(skb);  /* mac header is not yet set */
	skb_set_inner_protocol(skb, skb->protocol);
	skb->encapsulation = 1;
	skb_push(skb, len);
	if (ingress)
		skb_postpush_rcsum(skb, iph, len);
	skb_reset_network_header(skb);
	memcpy(skb_network_header(skb), hdr, len);
	bpf_compute_data_pointers(skb);
	skb_clear_hash(skb);

	if (ipv4) {
		skb->protocol = htons(ETH_P_IP);
		iph = ip_hdr(skb);

		if (!iph->check)
			iph->check = ip_fast_csum((unsigned char *)iph,
						  iph->ihl);
	} else {
		skb->protocol = htons(ETH_P_IPV6);
	}

	if (skb_is_gso(skb))
		return handle_gso_encap(skb, ipv4, len);

	return 0;
}

static int __init bpf_lwt_init(void)
{
	return lwtunnel_encap_add_ops(&bpf_encap_ops, LWTUNNEL_ENCAP_BPF);
}

subsys_initcall(bpf_lwt_init)
Commit	Line	Data
5b497af4	1	// SPDX-License-Identifier: GPL-2.0-only
3a0af8fd	2	/* Copyright (c) 2016 Thomas Graf <tgraf@tgraf.ch>
3a0af8fd TG	3	*/
	4
	5	#include <linux/kernel.h>
	6	#include <linux/module.h>
	7	#include <linux/skbuff.h>
	8	#include <linux/types.h>
	9	#include <linux/bpf.h>
	10	#include <net/lwtunnel.h>
ca78801a	11	#include <net/gre.h>
3bd0b152	12	#include <net/ip6_route.h>
3616d08b	13	#include <net/ipv6_stubs.h>
3a0af8fd TG	14
	15	struct bpf_lwt_prog {
	16	struct bpf_prog *prog;
	17	char *name;
	18	};
	19
	20	struct bpf_lwt {
	21	struct bpf_lwt_prog in;
	22	struct bpf_lwt_prog out;
	23	struct bpf_lwt_prog xmit;
	24	int family;
	25	};
	26
	27	#define MAX_PROG_NAME 256
	28
	29	static inline struct bpf_lwt bpf_lwt_lwtunnel(struct lwtunnel_state lwt)
	30	{
	31	return (struct bpf_lwt *)lwt->data;
	32	}
	33
	34	#define NO_REDIRECT false
	35	#define CAN_REDIRECT true
	36
	37	static int run_lwt_bpf(struct sk_buff skb, struct bpf_lwt_prog lwt,
	38	struct dst_entry *dst, bool can_redirect)
	39	{
	40	int ret;
	41
	42	/* Preempt disable is needed to protect per-cpu redirect_info between
	43	* BPF prog and skb_do_redirect(). The call_rcu in bpf_prog_put() and
	44	* access to maps strictly require a rcu_read_lock() for protection,
	45	* mixing with BH RCU lock doesn't work.
	46	*/
	47	preempt_disable();
6aaae2b6	48	bpf_compute_data_pointers(skb);
3a0af8fd	49	ret = bpf_prog_run_save_cb(lwt->prog, skb);
3a0af8fd TG	50
	51	switch (ret) {
	52	case BPF_OK:
3bd0b152	53	case BPF_LWT_REROUTE:
3a0af8fd TG	54	break;
	55
	56	case BPF_REDIRECT:
	57	if (unlikely(!can_redirect)) {
	58	pr_warn_once("Illegal redirect return code in prog %s\n",
	59	lwt->name ? : "<unknown>");
	60	ret = BPF_OK;
	61	} else {
e7c87bd6	62	skb_reset_mac_header(skb);
3a0af8fd TG	63	ret = skb_do_redirect(skb);
	64	if (ret == 0)
	65	ret = BPF_REDIRECT;
	66	}
	67	break;
	68
	69	case BPF_DROP:
	70	kfree_skb(skb);
	71	ret = -EPERM;
	72	break;
	73
	74	default:
	75	pr_warn_once("bpf-lwt: Illegal return value %u, expect packet loss\n", ret);
	76	kfree_skb(skb);
	77	ret = -EINVAL;
	78	break;
	79	}
	80
	81	preempt_enable();
	82
	83	return ret;
	84	}
	85
3bd0b152 PO	86	static int bpf_lwt_input_reroute(struct sk_buff *skb)
	87	{
	88	int err = -EINVAL;
	89
	90	if (skb->protocol == htons(ETH_P_IP)) {
	91	struct iphdr *iph = ip_hdr(skb);
	92
	93	err = ip_route_input_noref(skb, iph->daddr, iph->saddr,
	94	iph->tos, skb_dst(skb)->dev);
	95	} else if (skb->protocol == htons(ETH_P_IPV6)) {
	96	err = ipv6_stub->ipv6_route_input(skb);
	97	} else {
	98	err = -EAFNOSUPPORT;
	99	}
	100
	101	if (err)
	102	goto err;
	103	return dst_input(skb);
	104
	105	err:
	106	kfree_skb(skb);
	107	return err;
	108	}
	109
3a0af8fd TG	110	static int bpf_input(struct sk_buff *skb)
	111	{
	112	struct dst_entry *dst = skb_dst(skb);
	113	struct bpf_lwt *bpf;
	114	int ret;
	115
	116	bpf = bpf_lwt_lwtunnel(dst->lwtstate);
	117	if (bpf->in.prog) {
	118	ret = run_lwt_bpf(skb, &bpf->in, dst, NO_REDIRECT);
	119	if (ret < 0)
	120	return ret;
3bd0b152 PO	121	if (ret == BPF_LWT_REROUTE)
3bd0b152 PO	122	return bpf_lwt_input_reroute(skb);
3a0af8fd TG	123	}
	124
	125	if (unlikely(!dst->lwtstate->orig_input)) {
3a0af8fd TG	126	kfree_skb(skb);
	127	return -EINVAL;
	128	}
	129
	130	return dst->lwtstate->orig_input(skb);
	131	}
	132
	133	static int bpf_output(struct net net, struct sock sk, struct sk_buff *skb)
	134	{
	135	struct dst_entry *dst = skb_dst(skb);
	136	struct bpf_lwt *bpf;
	137	int ret;
	138
	139	bpf = bpf_lwt_lwtunnel(dst->lwtstate);
	140	if (bpf->out.prog) {
	141	ret = run_lwt_bpf(skb, &bpf->out, dst, NO_REDIRECT);
	142	if (ret < 0)
	143	return ret;
	144	}
	145
	146	if (unlikely(!dst->lwtstate->orig_output)) {
	147	pr_warn_once("orig_output not set on dst for prog %s\n",
	148	bpf->out.name);
	149	kfree_skb(skb);
	150	return -EINVAL;
	151	}
	152
	153	return dst->lwtstate->orig_output(net, sk, skb);
	154	}
	155
	156	static int xmit_check_hhlen(struct sk_buff *skb)
	157	{
	158	int hh_len = skb_dst(skb)->dev->hard_header_len;
	159
	160	if (skb_headroom(skb) < hh_len) {
	161	int nhead = HH_DATA_ALIGN(hh_len - skb_headroom(skb));
	162
	163	if (pskb_expand_head(skb, nhead, 0, GFP_ATOMIC))
	164	return -ENOMEM;
	165	}
	166
	167	return 0;
	168	}
	169
3bd0b152 PO	170	static int bpf_lwt_xmit_reroute(struct sk_buff *skb)
	171	{
	172	struct net_device *l3mdev = l3mdev_master_dev_rcu(skb_dst(skb)->dev);
	173	int oif = l3mdev ? l3mdev->ifindex : 0;
	174	struct dst_entry *dst = NULL;
fb405883	175	int err = -EAFNOSUPPORT;
3bd0b152 PO	176	struct sock *sk;
	177	struct net *net;
	178	bool ipv4;
3bd0b152 PO	179
	180	if (skb->protocol == htons(ETH_P_IP))
	181	ipv4 = true;
	182	else if (skb->protocol == htons(ETH_P_IPV6))
	183	ipv4 = false;
	184	else
fb405883	185	goto err;
3bd0b152 PO	186
	187	sk = sk_to_full_sk(skb->sk);
	188	if (sk) {
	189	if (sk->sk_bound_dev_if)
	190	oif = sk->sk_bound_dev_if;
	191	net = sock_net(sk);
	192	} else {
	193	net = dev_net(skb_dst(skb)->dev);
	194	}
	195
	196	if (ipv4) {
	197	struct iphdr *iph = ip_hdr(skb);
	198	struct flowi4 fl4 = {};
	199	struct rtable *rt;
	200
	201	fl4.flowi4_oif = oif;
	202	fl4.flowi4_mark = skb->mark;
	203	fl4.flowi4_uid = sock_net_uid(net, sk);
	204	fl4.flowi4_tos = RT_TOS(iph->tos);
	205	fl4.flowi4_flags = FLOWI_FLAG_ANYSRC;
	206	fl4.flowi4_proto = iph->protocol;
	207	fl4.daddr = iph->daddr;
	208	fl4.saddr = iph->saddr;
	209
	210	rt = ip_route_output_key(net, &fl4);
fb405883 PO	211	if (IS_ERR(rt)) {
	212	err = PTR_ERR(rt);
	213	goto err;
	214	}
3bd0b152 PO	215	dst = &rt->dst;
	216	} else {
	217	struct ipv6hdr *iph6 = ipv6_hdr(skb);
	218	struct flowi6 fl6 = {};
	219
	220	fl6.flowi6_oif = oif;
	221	fl6.flowi6_mark = skb->mark;
	222	fl6.flowi6_uid = sock_net_uid(net, sk);
	223	fl6.flowlabel = ip6_flowinfo(iph6);
	224	fl6.flowi6_proto = iph6->nexthdr;
	225	fl6.daddr = iph6->daddr;
	226	fl6.saddr = iph6->saddr;
	227
	228	err = ipv6_stub->ipv6_dst_lookup(net, skb->sk, &dst, &fl6);
fb405883 PO	229	if (unlikely(err))
	230	goto err;
	231	if (IS_ERR(dst)) {
	232	err = PTR_ERR(dst);
	233	goto err;
	234	}
3bd0b152 PO	235	}
3bd0b152 PO	236	if (unlikely(dst->error)) {
fb405883	237	err = dst->error;
3bd0b152	238	dst_release(dst);
fb405883	239	goto err;
3bd0b152 PO	240	}
	241
	242	/* Although skb header was reserved in bpf_lwt_push_ip_encap(), it
	243	* was done for the previous dst, so we are doing it here again, in
	244	* case the new dst needs much more space. The call below is a noop
	245	* if there is enough header space in skb.
	246	*/
	247	err = skb_cow_head(skb, LL_RESERVED_SPACE(dst->dev));
	248	if (unlikely(err))
fb405883	249	goto err;
3bd0b152 PO	250
	251	skb_dst_drop(skb);
	252	skb_dst_set(skb, dst);
	253
	254	err = dst_output(dev_net(skb_dst(skb)->dev), skb->sk, skb);
	255	if (unlikely(err))
bd16693f	256	return err;
3bd0b152 PO	257
	258	/* ip[6]_finish_output2 understand LWTUNNEL_XMIT_DONE */
	259	return LWTUNNEL_XMIT_DONE;
fb405883 PO	260
	261	err:
	262	kfree_skb(skb);
	263	return err;
3bd0b152 PO	264	}
3bd0b152 PO	265
3a0af8fd TG	266	static int bpf_xmit(struct sk_buff *skb)
	267	{
	268	struct dst_entry *dst = skb_dst(skb);
	269	struct bpf_lwt *bpf;
	270
	271	bpf = bpf_lwt_lwtunnel(dst->lwtstate);
	272	if (bpf->xmit.prog) {
3bd0b152	273	__be16 proto = skb->protocol;
3a0af8fd TG	274	int ret;
	275
	276	ret = run_lwt_bpf(skb, &bpf->xmit, dst, CAN_REDIRECT);
	277	switch (ret) {
	278	case BPF_OK:
3bd0b152 PO	279	/* If the header changed, e.g. via bpf_lwt_push_encap,
	280	* BPF_LWT_REROUTE below should have been used if the
	281	* protocol was also changed.
	282	*/
	283	if (skb->protocol != proto) {
	284	kfree_skb(skb);
	285	return -EINVAL;
	286	}
3a0af8fd TG	287	/* If the header was expanded, headroom might be too
	288	* small for L2 header to come, expand as needed.
	289	*/
	290	ret = xmit_check_hhlen(skb);
	291	if (unlikely(ret))
	292	return ret;
	293
	294	return LWTUNNEL_XMIT_CONTINUE;
	295	case BPF_REDIRECT:
	296	return LWTUNNEL_XMIT_DONE;
3bd0b152 PO	297	case BPF_LWT_REROUTE:
3bd0b152 PO	298	return bpf_lwt_xmit_reroute(skb);
3a0af8fd TG	299	default:
	300	return ret;
	301	}
	302	}
	303
	304	return LWTUNNEL_XMIT_CONTINUE;
	305	}
	306
	307	static void bpf_lwt_prog_destroy(struct bpf_lwt_prog *prog)
	308	{
	309	if (prog->prog)
	310	bpf_prog_put(prog->prog);
	311
	312	kfree(prog->name);
	313	}
	314
	315	static void bpf_destroy_state(struct lwtunnel_state *lwt)
	316	{
	317	struct bpf_lwt *bpf = bpf_lwt_lwtunnel(lwt);
	318
	319	bpf_lwt_prog_destroy(&bpf->in);
	320	bpf_lwt_prog_destroy(&bpf->out);
	321	bpf_lwt_prog_destroy(&bpf->xmit);
	322	}
	323
	324	static const struct nla_policy bpf_prog_policy[LWT_BPF_PROG_MAX + 1] = {
	325	[LWT_BPF_PROG_FD] = { .type = NLA_U32, },
	326	[LWT_BPF_PROG_NAME] = { .type = NLA_NUL_STRING,
	327	.len = MAX_PROG_NAME },
	328	};
	329
	330	static int bpf_parse_prog(struct nlattr attr, struct bpf_lwt_prog prog,
	331	enum bpf_prog_type type)
	332	{
	333	struct nlattr *tb[LWT_BPF_PROG_MAX + 1];
	334	struct bpf_prog *p;
	335	int ret;
	336	u32 fd;
	337
8cb08174 JB	338	ret = nla_parse_nested_deprecated(tb, LWT_BPF_PROG_MAX, attr,
8cb08174 JB	339	bpf_prog_policy, NULL);
3a0af8fd TG	340	if (ret < 0)
	341	return ret;
	342
	343	if (!tb[LWT_BPF_PROG_FD] \|\| !tb[LWT_BPF_PROG_NAME])
	344	return -EINVAL;
	345
71eb5255	346	prog->name = nla_memdup(tb[LWT_BPF_PROG_NAME], GFP_ATOMIC);
3a0af8fd TG	347	if (!prog->name)
	348	return -ENOMEM;
	349
	350	fd = nla_get_u32(tb[LWT_BPF_PROG_FD]);
	351	p = bpf_prog_get_type(fd, type);
	352	if (IS_ERR(p))
	353	return PTR_ERR(p);
	354
	355	prog->prog = p;
	356
	357	return 0;
	358	}
	359
	360	static const struct nla_policy bpf_nl_policy[LWT_BPF_MAX + 1] = {
	361	[LWT_BPF_IN] = { .type = NLA_NESTED, },
	362	[LWT_BPF_OUT] = { .type = NLA_NESTED, },
	363	[LWT_BPF_XMIT] = { .type = NLA_NESTED, },
	364	[LWT_BPF_XMIT_HEADROOM] = { .type = NLA_U32 },
	365	};
	366
30357d7d	367	static int bpf_build_state(struct nlattr *nla,
3a0af8fd	368	unsigned int family, const void *cfg,
9ae28727 DA	369	struct lwtunnel_state **ts,
9ae28727 DA	370	struct netlink_ext_ack *extack)
3a0af8fd TG	371	{
	372	struct nlattr *tb[LWT_BPF_MAX + 1];
	373	struct lwtunnel_state *newts;
	374	struct bpf_lwt *bpf;
	375	int ret;
	376
	377	if (family != AF_INET && family != AF_INET6)
	378	return -EAFNOSUPPORT;
	379
8cb08174 JB	380	ret = nla_parse_nested_deprecated(tb, LWT_BPF_MAX, nla, bpf_nl_policy,
8cb08174 JB	381	extack);
3a0af8fd TG	382	if (ret < 0)
	383	return ret;
	384
	385	if (!tb[LWT_BPF_IN] && !tb[LWT_BPF_OUT] && !tb[LWT_BPF_XMIT])
	386	return -EINVAL;
	387
	388	newts = lwtunnel_state_alloc(sizeof(*bpf));
	389	if (!newts)
	390	return -ENOMEM;
	391
	392	newts->type = LWTUNNEL_ENCAP_BPF;
	393	bpf = bpf_lwt_lwtunnel(newts);
	394
	395	if (tb[LWT_BPF_IN]) {
	396	newts->flags \|= LWTUNNEL_STATE_INPUT_REDIRECT;
	397	ret = bpf_parse_prog(tb[LWT_BPF_IN], &bpf->in,
	398	BPF_PROG_TYPE_LWT_IN);
	399	if (ret < 0)
	400	goto errout;
	401	}
	402
	403	if (tb[LWT_BPF_OUT]) {
	404	newts->flags \|= LWTUNNEL_STATE_OUTPUT_REDIRECT;
	405	ret = bpf_parse_prog(tb[LWT_BPF_OUT], &bpf->out,
	406	BPF_PROG_TYPE_LWT_OUT);
	407	if (ret < 0)
	408	goto errout;
	409	}
	410
	411	if (tb[LWT_BPF_XMIT]) {
	412	newts->flags \|= LWTUNNEL_STATE_XMIT_REDIRECT;
	413	ret = bpf_parse_prog(tb[LWT_BPF_XMIT], &bpf->xmit,
	414	BPF_PROG_TYPE_LWT_XMIT);
	415	if (ret < 0)
	416	goto errout;
	417	}
	418
	419	if (tb[LWT_BPF_XMIT_HEADROOM]) {
	420	u32 headroom = nla_get_u32(tb[LWT_BPF_XMIT_HEADROOM]);
	421
	422	if (headroom > LWT_BPF_MAX_HEADROOM) {
	423	ret = -ERANGE;
	424	goto errout;
	425	}
	426
	427	newts->headroom = headroom;
	428	}
	429
	430	bpf->family = family;
	431	*ts = newts;
	432
	433	return 0;
	434
	435	errout:
	436	bpf_destroy_state(newts);
	437	kfree(newts);
	438	return ret;
	439	}
	440
	441	static int bpf_fill_lwt_prog(struct sk_buff *skb, int attr,
	442	struct bpf_lwt_prog *prog)
	443	{
	444	struct nlattr *nest;
	445
446	if (!prog->prog)
447	return 0;
448
ae0be8de	449	nest = nla_nest_start_noflag(skb, attr);
3a0af8fd TG	450	if (!nest)
	451	return -EMSGSIZE;
	452
	453	if (prog->name &&
	454	nla_put_string(skb, LWT_BPF_PROG_NAME, prog->name))
	455	return -EMSGSIZE;
	456
	457	return nla_nest_end(skb, nest);
	458	}
	459
	460	static int bpf_fill_encap_info(struct sk_buff skb, struct lwtunnel_state lwt)
	461	{
	462	struct bpf_lwt *bpf = bpf_lwt_lwtunnel(lwt);
	463
	464	if (bpf_fill_lwt_prog(skb, LWT_BPF_IN, &bpf->in) < 0 \|\|
	465	bpf_fill_lwt_prog(skb, LWT_BPF_OUT, &bpf->out) < 0 \|\|
	466	bpf_fill_lwt_prog(skb, LWT_BPF_XMIT, &bpf->xmit) < 0)
	467	return -EMSGSIZE;
	468
	469	return 0;
	470	}
	471
	472	static int bpf_encap_nlsize(struct lwtunnel_state *lwtstate)
	473	{
	474	int nest_len = nla_total_size(sizeof(struct nlattr)) +
	475	nla_total_size(MAX_PROG_NAME) + /* LWT_BPF_PROG_NAME */
	476	0;
	477
	478	return nest_len + /* LWT_BPF_IN */
	479	nest_len + /* LWT_BPF_OUT */
	480	nest_len + /* LWT_BPF_XMIT */
	481	0;
	482	}
	483
79471b10	484	static int bpf_lwt_prog_cmp(struct bpf_lwt_prog a, struct bpf_lwt_prog b)
3a0af8fd TG	485	{
	486	/* FIXME:
	487	* The LWT state is currently rebuilt for delete requests which
	488	* results in a new bpf_prog instance. Comparing names for now.
	489	*/
	490	if (!a->name && !b->name)
	491	return 0;
	492
	493	if (!a->name \|\| !b->name)
	494	return 1;
	495
	496	return strcmp(a->name, b->name);
	497	}
	498
	499	static int bpf_encap_cmp(struct lwtunnel_state a, struct lwtunnel_state b)
	500	{
	501	struct bpf_lwt *a_bpf = bpf_lwt_lwtunnel(a);
	502	struct bpf_lwt *b_bpf = bpf_lwt_lwtunnel(b);
	503
	504	return bpf_lwt_prog_cmp(&a_bpf->in, &b_bpf->in) \|\|
	505	bpf_lwt_prog_cmp(&a_bpf->out, &b_bpf->out) \|\|
	506	bpf_lwt_prog_cmp(&a_bpf->xmit, &b_bpf->xmit);
	507	}
	508
	509	static const struct lwtunnel_encap_ops bpf_encap_ops = {
	510	.build_state = bpf_build_state,
	511	.destroy_state = bpf_destroy_state,
	512	.input = bpf_input,
	513	.output = bpf_output,
	514	.xmit = bpf_xmit,
	515	.fill_encap = bpf_fill_encap_info,
	516	.get_encap_size = bpf_encap_nlsize,
	517	.cmp_encap = bpf_encap_cmp,
88ff7334	518	.owner = THIS_MODULE,
3a0af8fd TG	519	};
3a0af8fd TG	520
ca78801a PO	521	static int handle_gso_type(struct sk_buff *skb, unsigned int gso_type,
	522	int encap_len)
	523	{
	524	struct skb_shared_info *shinfo = skb_shinfo(skb);
	525
	526	gso_type \|= SKB_GSO_DODGY;
	527	shinfo->gso_type \|= gso_type;
	528	skb_decrease_gso_size(shinfo, encap_len);
	529	shinfo->gso_segs = 0;
	530	return 0;
	531	}
	532
52f27877 PO	533	static int handle_gso_encap(struct sk_buff *skb, bool ipv4, int encap_len)
52f27877 PO	534	{
ca78801a PO	535	int next_hdr_offset;
	536	void *next_hdr;
	537	__u8 protocol;
	538
	539	/* SCTP and UDP_L4 gso need more nuanced handling than what
	540	* handle_gso_type() does above: skb_decrease_gso_size() is not enough.
	541	* So at the moment only TCP GSO packets are let through.
	542	*/
	543	if (!(skb_shinfo(skb)->gso_type & (SKB_GSO_TCPV4 \| SKB_GSO_TCPV6)))
	544	return -ENOTSUPP;
	545
	546	if (ipv4) {
	547	protocol = ip_hdr(skb)->protocol;
	548	next_hdr_offset = sizeof(struct iphdr);
	549	next_hdr = skb_network_header(skb) + next_hdr_offset;
	550	} else {
	551	protocol = ipv6_hdr(skb)->nexthdr;
	552	next_hdr_offset = sizeof(struct ipv6hdr);
	553	next_hdr = skb_network_header(skb) + next_hdr_offset;
	554	}
	555
	556	switch (protocol) {
	557	case IPPROTO_GRE:
	558	next_hdr_offset += sizeof(struct gre_base_hdr);
	559	if (next_hdr_offset > encap_len)
	560	return -EINVAL;
	561
	562	if (((struct gre_base_hdr *)next_hdr)->flags & GRE_CSUM)
	563	return handle_gso_type(skb, SKB_GSO_GRE_CSUM,
	564	encap_len);
	565	return handle_gso_type(skb, SKB_GSO_GRE, encap_len);
	566
	567	case IPPROTO_UDP:
	568	next_hdr_offset += sizeof(struct udphdr);
	569	if (next_hdr_offset > encap_len)
	570	return -EINVAL;
	571
	572	if (((struct udphdr *)next_hdr)->check)
	573	return handle_gso_type(skb, SKB_GSO_UDP_TUNNEL_CSUM,
	574	encap_len);
	575	return handle_gso_type(skb, SKB_GSO_UDP_TUNNEL, encap_len);
	576
	577	case IPPROTO_IP:
	578	case IPPROTO_IPV6:
	579	if (ipv4)
	580	return handle_gso_type(skb, SKB_GSO_IPXIP4, encap_len);
	581	else
	582	return handle_gso_type(skb, SKB_GSO_IPXIP6, encap_len);
	583
	584	default:
	585	return -EPROTONOSUPPORT;
	586	}
52f27877 PO	587	}
	588
	589	int bpf_lwt_push_ip_encap(struct sk_buff skb, void hdr, u32 len, bool ingress)
	590	{
	591	struct iphdr *iph;
	592	bool ipv4;
	593	int err;
	594
	595	if (unlikely(len < sizeof(struct iphdr) \|\| len > LWT_BPF_MAX_HEADROOM))
	596	return -EINVAL;
	597
	598	/* validate protocol and length */
	599	iph = (struct iphdr *)hdr;
	600	if (iph->version == 4) {
	601	ipv4 = true;
	602	if (unlikely(len < iph->ihl * 4))
	603	return -EINVAL;
	604	} else if (iph->version == 6) {
	605	ipv4 = false;
	606	if (unlikely(len < sizeof(struct ipv6hdr)))
	607	return -EINVAL;
	608	} else {
	609	return -EINVAL;
	610	}
	611
	612	if (ingress)
	613	err = skb_cow_head(skb, len + skb->mac_len);
	614	else
	615	err = skb_cow_head(skb,
	616	len + LL_RESERVED_SPACE(skb_dst(skb)->dev));
	617	if (unlikely(err))
	618	return err;
	619
	620	/* push the encap headers and fix pointers */
	621	skb_reset_inner_headers(skb);
ea0371f7 PO	622	skb_reset_inner_mac_header(skb); /* mac header is not yet set */
ea0371f7 PO	623	skb_set_inner_protocol(skb, skb->protocol);
52f27877 PO	624	skb->encapsulation = 1;
	625	skb_push(skb, len);
	626	if (ingress)
	627	skb_postpush_rcsum(skb, iph, len);
	628	skb_reset_network_header(skb);
	629	memcpy(skb_network_header(skb), hdr, len);
	630	bpf_compute_data_pointers(skb);
	631	skb_clear_hash(skb);
	632
	633	if (ipv4) {
	634	skb->protocol = htons(ETH_P_IP);
	635	iph = ip_hdr(skb);
	636
	637	if (!iph->check)
	638	iph->check = ip_fast_csum((unsigned char *)iph,
	639	iph->ihl);
	640	} else {
	641	skb->protocol = htons(ETH_P_IPV6);
	642	}
	643
	644	if (skb_is_gso(skb))
	645	return handle_gso_encap(skb, ipv4, len);
	646
	647	return 0;
	648	}
	649
3a0af8fd TG	650	static int __init bpf_lwt_init(void)
	651	{
	652	return lwtunnel_encap_add_ops(&bpf_encap_ops, LWTUNNEL_ENCAP_BPF);
	653	}
	654
	655	subsys_initcall(bpf_lwt_init)