[linux-2.6-block.git] / net / core / lwt_bpf.c

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

#include <linux/filter.h>
#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;

	/* Migration disable and BH disable are needed to protect per-cpu
	 * redirect_info between BPF prog and skb_do_redirect().
	 */
	migrate_disable();
	local_bh_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;
	}

	local_bh_enable();
	migrate_enable();

	return ret;
}

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

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

		dev_hold(dev);
		skb_dst_drop(skb);
		err = ip_route_input_noref(skb, iph->daddr, iph->saddr,
					   iph->tos, dev);
		dev_put(dev);
	} else if (skb->protocol == htons(ETH_P_IPV6)) {
		skb_dst_drop(skb);
		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)
{
	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;

		dst = ipv6_stub->ipv6_dst_lookup_flow(net, skb->sk, &fl6, NULL);
		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) {
		int hh_len = dst->dev->hard_header_len;
		__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, hh_len);
			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 net *net, 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	*/
3a0af8fd TG	4
b6459415	5	#include <linux/filter.h>
3a0af8fd TG	6	#include <linux/kernel.h>
	7	#include <linux/module.h>
	8	#include <linux/skbuff.h>
	9	#include <linux/types.h>
	10	#include <linux/bpf.h>
	11	#include <net/lwtunnel.h>
ca78801a	12	#include <net/gre.h>
3bd0b152	13	#include <net/ip6_route.h>
3616d08b	14	#include <net/ipv6_stubs.h>
3a0af8fd TG	15
	16	struct bpf_lwt_prog {
	17	struct bpf_prog *prog;
	18	char *name;
	19	};
	20
	21	struct bpf_lwt {
	22	struct bpf_lwt_prog in;
	23	struct bpf_lwt_prog out;
	24	struct bpf_lwt_prog xmit;
	25	int family;
	26	};
	27
	28	#define MAX_PROG_NAME 256
	29
	30	static inline struct bpf_lwt bpf_lwt_lwtunnel(struct lwtunnel_state lwt)
	31	{
	32	return (struct bpf_lwt *)lwt->data;
	33	}
	34
	35	#define NO_REDIRECT false
	36	#define CAN_REDIRECT true
	37
	38	static int run_lwt_bpf(struct sk_buff skb, struct bpf_lwt_prog lwt,
	39	struct dst_entry *dst, bool can_redirect)
	40	{
	41	int ret;
	42
e3366884	43	/* Migration disable and BH disable are needed to protect per-cpu
d9054a1f	44	* redirect_info between BPF prog and skb_do_redirect().
3a0af8fd	45	*/
e3366884	46	migrate_disable();
d9054a1f	47	local_bh_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
d9054a1f	81	local_bh_enable();
e3366884	82	migrate_enable();
3a0af8fd TG	83
	84	return ret;
	85	}
	86
3bd0b152 PO	87	static int bpf_lwt_input_reroute(struct sk_buff *skb)
	88	{
	89	int err = -EINVAL;
	90
	91	if (skb->protocol == htons(ETH_P_IP)) {
9e8acd9c	92	struct net_device *dev = skb_dst(skb)->dev;
3bd0b152 PO	93	struct iphdr *iph = ip_hdr(skb);
3bd0b152 PO	94
9e8acd9c JB	95	dev_hold(dev);
9e8acd9c JB	96	skb_dst_drop(skb);
3bd0b152	97	err = ip_route_input_noref(skb, iph->daddr, iph->saddr,
9e8acd9c JB	98	iph->tos, dev);
9e8acd9c JB	99	dev_put(dev);
3bd0b152	100	} else if (skb->protocol == htons(ETH_P_IPV6)) {
9e8acd9c	101	skb_dst_drop(skb);
3bd0b152 PO	102	err = ipv6_stub->ipv6_route_input(skb);
	103	} else {
	104	err = -EAFNOSUPPORT;
	105	}
	106
	107	if (err)
	108	goto err;
	109	return dst_input(skb);
	110
	111	err:
	112	kfree_skb(skb);
	113	return err;
	114	}
	115
3a0af8fd TG	116	static int bpf_input(struct sk_buff *skb)
	117	{
	118	struct dst_entry *dst = skb_dst(skb);
	119	struct bpf_lwt *bpf;
	120	int ret;
	121
	122	bpf = bpf_lwt_lwtunnel(dst->lwtstate);
	123	if (bpf->in.prog) {
	124	ret = run_lwt_bpf(skb, &bpf->in, dst, NO_REDIRECT);
	125	if (ret < 0)
	126	return ret;
3bd0b152 PO	127	if (ret == BPF_LWT_REROUTE)
3bd0b152 PO	128	return bpf_lwt_input_reroute(skb);
3a0af8fd TG	129	}
	130
	131	if (unlikely(!dst->lwtstate->orig_input)) {
3a0af8fd TG	132	kfree_skb(skb);
	133	return -EINVAL;
	134	}
	135
	136	return dst->lwtstate->orig_input(skb);
	137	}
	138
	139	static int bpf_output(struct net net, struct sock sk, struct sk_buff *skb)
	140	{
	141	struct dst_entry *dst = skb_dst(skb);
	142	struct bpf_lwt *bpf;
	143	int ret;
	144
	145	bpf = bpf_lwt_lwtunnel(dst->lwtstate);
	146	if (bpf->out.prog) {
	147	ret = run_lwt_bpf(skb, &bpf->out, dst, NO_REDIRECT);
	148	if (ret < 0)
	149	return ret;
	150	}
	151
	152	if (unlikely(!dst->lwtstate->orig_output)) {
	153	pr_warn_once("orig_output not set on dst for prog %s\n",
	154	bpf->out.name);
	155	kfree_skb(skb);
	156	return -EINVAL;
	157	}
	158
	159	return dst->lwtstate->orig_output(net, sk, skb);
	160	}
	161
b02d196c	162	static int xmit_check_hhlen(struct sk_buff *skb, int hh_len)
3a0af8fd	163	{
3a0af8fd TG	164	if (skb_headroom(skb) < hh_len) {
	165	int nhead = HH_DATA_ALIGN(hh_len - skb_headroom(skb));
	166
	167	if (pskb_expand_head(skb, nhead, 0, GFP_ATOMIC))
	168	return -ENOMEM;
	169	}
	170
	171	return 0;
	172	}
	173
3bd0b152 PO	174	static int bpf_lwt_xmit_reroute(struct sk_buff *skb)
	175	{
	176	struct net_device *l3mdev = l3mdev_master_dev_rcu(skb_dst(skb)->dev);
	177	int oif = l3mdev ? l3mdev->ifindex : 0;
	178	struct dst_entry *dst = NULL;
fb405883	179	int err = -EAFNOSUPPORT;
3bd0b152 PO	180	struct sock *sk;
	181	struct net *net;
	182	bool ipv4;
3bd0b152 PO	183
	184	if (skb->protocol == htons(ETH_P_IP))
	185	ipv4 = true;
	186	else if (skb->protocol == htons(ETH_P_IPV6))
	187	ipv4 = false;
	188	else
fb405883	189	goto err;
3bd0b152 PO	190
	191	sk = sk_to_full_sk(skb->sk);
	192	if (sk) {
	193	if (sk->sk_bound_dev_if)
	194	oif = sk->sk_bound_dev_if;
	195	net = sock_net(sk);
	196	} else {
	197	net = dev_net(skb_dst(skb)->dev);
	198	}
	199
	200	if (ipv4) {
	201	struct iphdr *iph = ip_hdr(skb);
	202	struct flowi4 fl4 = {};
	203	struct rtable *rt;
	204
	205	fl4.flowi4_oif = oif;
	206	fl4.flowi4_mark = skb->mark;
	207	fl4.flowi4_uid = sock_net_uid(net, sk);
	208	fl4.flowi4_tos = RT_TOS(iph->tos);
	209	fl4.flowi4_flags = FLOWI_FLAG_ANYSRC;
	210	fl4.flowi4_proto = iph->protocol;
	211	fl4.daddr = iph->daddr;
	212	fl4.saddr = iph->saddr;
	213
	214	rt = ip_route_output_key(net, &fl4);
fb405883 PO	215	if (IS_ERR(rt)) {
	216	err = PTR_ERR(rt);
	217	goto err;
	218	}
3bd0b152 PO	219	dst = &rt->dst;
	220	} else {
	221	struct ipv6hdr *iph6 = ipv6_hdr(skb);
	222	struct flowi6 fl6 = {};
	223
	224	fl6.flowi6_oif = oif;
	225	fl6.flowi6_mark = skb->mark;
	226	fl6.flowi6_uid = sock_net_uid(net, sk);
	227	fl6.flowlabel = ip6_flowinfo(iph6);
	228	fl6.flowi6_proto = iph6->nexthdr;
	229	fl6.daddr = iph6->daddr;
	230	fl6.saddr = iph6->saddr;
	231
6c8991f4	232	dst = ipv6_stub->ipv6_dst_lookup_flow(net, skb->sk, &fl6, NULL);
fb405883 PO	233	if (IS_ERR(dst)) {
	234	err = PTR_ERR(dst);
	235	goto err;
	236	}
3bd0b152 PO	237	}
3bd0b152 PO	238	if (unlikely(dst->error)) {
fb405883	239	err = dst->error;
3bd0b152	240	dst_release(dst);
fb405883	241	goto err;
3bd0b152 PO	242	}
	243
	244	/* Although skb header was reserved in bpf_lwt_push_ip_encap(), it
	245	* was done for the previous dst, so we are doing it here again, in
	246	* case the new dst needs much more space. The call below is a noop
	247	* if there is enough header space in skb.
	248	*/
	249	err = skb_cow_head(skb, LL_RESERVED_SPACE(dst->dev));
	250	if (unlikely(err))
fb405883	251	goto err;
3bd0b152 PO	252
	253	skb_dst_drop(skb);
	254	skb_dst_set(skb, dst);
	255
	256	err = dst_output(dev_net(skb_dst(skb)->dev), skb->sk, skb);
	257	if (unlikely(err))
bd16693f	258	return err;
3bd0b152 PO	259
	260	/* ip[6]_finish_output2 understand LWTUNNEL_XMIT_DONE */
	261	return LWTUNNEL_XMIT_DONE;
fb405883 PO	262
	263	err:
	264	kfree_skb(skb);
	265	return err;
3bd0b152 PO	266	}
3bd0b152 PO	267
3a0af8fd TG	268	static int bpf_xmit(struct sk_buff *skb)
	269	{
	270	struct dst_entry *dst = skb_dst(skb);
	271	struct bpf_lwt *bpf;
	272
	273	bpf = bpf_lwt_lwtunnel(dst->lwtstate);
	274	if (bpf->xmit.prog) {
b02d196c	275	int hh_len = dst->dev->hard_header_len;
3bd0b152	276	__be16 proto = skb->protocol;
3a0af8fd TG	277	int ret;
	278
	279	ret = run_lwt_bpf(skb, &bpf->xmit, dst, CAN_REDIRECT);
	280	switch (ret) {
	281	case BPF_OK:
3bd0b152 PO	282	/* If the header changed, e.g. via bpf_lwt_push_encap,
	283	* BPF_LWT_REROUTE below should have been used if the
	284	* protocol was also changed.
	285	*/
	286	if (skb->protocol != proto) {
	287	kfree_skb(skb);
	288	return -EINVAL;
	289	}
3a0af8fd TG	290	/* If the header was expanded, headroom might be too
	291	* small for L2 header to come, expand as needed.
	292	*/
b02d196c	293	ret = xmit_check_hhlen(skb, hh_len);
3a0af8fd TG	294	if (unlikely(ret))
	295	return ret;
	296
	297	return LWTUNNEL_XMIT_CONTINUE;
	298	case BPF_REDIRECT:
	299	return LWTUNNEL_XMIT_DONE;
3bd0b152 PO	300	case BPF_LWT_REROUTE:
3bd0b152 PO	301	return bpf_lwt_xmit_reroute(skb);
3a0af8fd TG	302	default:
	303	return ret;
	304	}
	305	}
	306
	307	return LWTUNNEL_XMIT_CONTINUE;
	308	}
	309
	310	static void bpf_lwt_prog_destroy(struct bpf_lwt_prog *prog)
	311	{
	312	if (prog->prog)
	313	bpf_prog_put(prog->prog);
	314
	315	kfree(prog->name);
	316	}
	317
	318	static void bpf_destroy_state(struct lwtunnel_state *lwt)
	319	{
	320	struct bpf_lwt *bpf = bpf_lwt_lwtunnel(lwt);
	321
	322	bpf_lwt_prog_destroy(&bpf->in);
	323	bpf_lwt_prog_destroy(&bpf->out);
	324	bpf_lwt_prog_destroy(&bpf->xmit);
	325	}
	326
	327	static const struct nla_policy bpf_prog_policy[LWT_BPF_PROG_MAX + 1] = {
	328	[LWT_BPF_PROG_FD] = { .type = NLA_U32, },
	329	[LWT_BPF_PROG_NAME] = { .type = NLA_NUL_STRING,
	330	.len = MAX_PROG_NAME },
	331	};
	332
	333	static int bpf_parse_prog(struct nlattr attr, struct bpf_lwt_prog prog,
	334	enum bpf_prog_type type)
	335	{
	336	struct nlattr *tb[LWT_BPF_PROG_MAX + 1];
	337	struct bpf_prog *p;
	338	int ret;
	339	u32 fd;
	340
8cb08174 JB	341	ret = nla_parse_nested_deprecated(tb, LWT_BPF_PROG_MAX, attr,
8cb08174 JB	342	bpf_prog_policy, NULL);
3a0af8fd TG	343	if (ret < 0)
	344	return ret;
	345
	346	if (!tb[LWT_BPF_PROG_FD] \|\| !tb[LWT_BPF_PROG_NAME])
	347	return -EINVAL;
	348
71eb5255	349	prog->name = nla_memdup(tb[LWT_BPF_PROG_NAME], GFP_ATOMIC);
3a0af8fd TG	350	if (!prog->name)
	351	return -ENOMEM;
	352
	353	fd = nla_get_u32(tb[LWT_BPF_PROG_FD]);
	354	p = bpf_prog_get_type(fd, type);
	355	if (IS_ERR(p))
	356	return PTR_ERR(p);
	357
	358	prog->prog = p;
	359
	360	return 0;
	361	}
	362
	363	static const struct nla_policy bpf_nl_policy[LWT_BPF_MAX + 1] = {
	364	[LWT_BPF_IN] = { .type = NLA_NESTED, },
	365	[LWT_BPF_OUT] = { .type = NLA_NESTED, },
	366	[LWT_BPF_XMIT] = { .type = NLA_NESTED, },
	367	[LWT_BPF_XMIT_HEADROOM] = { .type = NLA_U32 },
	368	};
	369
faee6769	370	static int bpf_build_state(struct net net, struct nlattr nla,
3a0af8fd	371	unsigned int family, const void *cfg,
9ae28727 DA	372	struct lwtunnel_state **ts,
9ae28727 DA	373	struct netlink_ext_ack *extack)
3a0af8fd TG	374	{
	375	struct nlattr *tb[LWT_BPF_MAX + 1];
	376	struct lwtunnel_state *newts;
	377	struct bpf_lwt *bpf;
	378	int ret;
	379
	380	if (family != AF_INET && family != AF_INET6)
	381	return -EAFNOSUPPORT;
	382
8cb08174 JB	383	ret = nla_parse_nested_deprecated(tb, LWT_BPF_MAX, nla, bpf_nl_policy,
8cb08174 JB	384	extack);
3a0af8fd TG	385	if (ret < 0)
	386	return ret;
	387
	388	if (!tb[LWT_BPF_IN] && !tb[LWT_BPF_OUT] && !tb[LWT_BPF_XMIT])
	389	return -EINVAL;
	390
	391	newts = lwtunnel_state_alloc(sizeof(*bpf));
	392	if (!newts)
	393	return -ENOMEM;
	394
	395	newts->type = LWTUNNEL_ENCAP_BPF;
	396	bpf = bpf_lwt_lwtunnel(newts);
	397
	398	if (tb[LWT_BPF_IN]) {
	399	newts->flags \|= LWTUNNEL_STATE_INPUT_REDIRECT;
	400	ret = bpf_parse_prog(tb[LWT_BPF_IN], &bpf->in,
	401	BPF_PROG_TYPE_LWT_IN);
	402	if (ret < 0)
	403	goto errout;
	404	}
	405
	406	if (tb[LWT_BPF_OUT]) {
	407	newts->flags \|= LWTUNNEL_STATE_OUTPUT_REDIRECT;
	408	ret = bpf_parse_prog(tb[LWT_BPF_OUT], &bpf->out,
	409	BPF_PROG_TYPE_LWT_OUT);
	410	if (ret < 0)
	411	goto errout;
	412	}
	413
	414	if (tb[LWT_BPF_XMIT]) {
	415	newts->flags \|= LWTUNNEL_STATE_XMIT_REDIRECT;
	416	ret = bpf_parse_prog(tb[LWT_BPF_XMIT], &bpf->xmit,
	417	BPF_PROG_TYPE_LWT_XMIT);
	418	if (ret < 0)
	419	goto errout;
	420	}
	421
	422	if (tb[LWT_BPF_XMIT_HEADROOM]) {
	423	u32 headroom = nla_get_u32(tb[LWT_BPF_XMIT_HEADROOM]);
	424
	425	if (headroom > LWT_BPF_MAX_HEADROOM) {
	426	ret = -ERANGE;
	427	goto errout;
	428	}
	429
	430	newts->headroom = headroom;
	431	}
	432
	433	bpf->family = family;
	434	*ts = newts;
	435
	436	return 0;
	437
	438	errout:
	439	bpf_destroy_state(newts);
	440	kfree(newts);
	441	return ret;
	442	}
	443
	444	static int bpf_fill_lwt_prog(struct sk_buff *skb, int attr,
	445	struct bpf_lwt_prog *prog)
	446	{
	447	struct nlattr *nest;
	448
449	if (!prog->prog)
450	return 0;
451
ae0be8de	452	nest = nla_nest_start_noflag(skb, attr);
3a0af8fd TG	453	if (!nest)
	454	return -EMSGSIZE;
	455
	456	if (prog->name &&
	457	nla_put_string(skb, LWT_BPF_PROG_NAME, prog->name))
	458	return -EMSGSIZE;
	459
	460	return nla_nest_end(skb, nest);
	461	}
	462
	463	static int bpf_fill_encap_info(struct sk_buff skb, struct lwtunnel_state lwt)
	464	{
	465	struct bpf_lwt *bpf = bpf_lwt_lwtunnel(lwt);
	466
	467	if (bpf_fill_lwt_prog(skb, LWT_BPF_IN, &bpf->in) < 0 \|\|
	468	bpf_fill_lwt_prog(skb, LWT_BPF_OUT, &bpf->out) < 0 \|\|
	469	bpf_fill_lwt_prog(skb, LWT_BPF_XMIT, &bpf->xmit) < 0)
	470	return -EMSGSIZE;
	471
	472	return 0;
	473	}
	474
	475	static int bpf_encap_nlsize(struct lwtunnel_state *lwtstate)
	476	{
	477	int nest_len = nla_total_size(sizeof(struct nlattr)) +
	478	nla_total_size(MAX_PROG_NAME) + /* LWT_BPF_PROG_NAME */
	479	0;
	480
	481	return nest_len + /* LWT_BPF_IN */
	482	nest_len + /* LWT_BPF_OUT */
	483	nest_len + /* LWT_BPF_XMIT */
	484	0;
	485	}
	486
79471b10	487	static int bpf_lwt_prog_cmp(struct bpf_lwt_prog a, struct bpf_lwt_prog b)
3a0af8fd TG	488	{
	489	/* FIXME:
	490	* The LWT state is currently rebuilt for delete requests which
	491	* results in a new bpf_prog instance. Comparing names for now.
	492	*/
	493	if (!a->name && !b->name)
	494	return 0;
	495
	496	if (!a->name \|\| !b->name)
	497	return 1;
	498
	499	return strcmp(a->name, b->name);
	500	}
	501
	502	static int bpf_encap_cmp(struct lwtunnel_state a, struct lwtunnel_state b)
	503	{
	504	struct bpf_lwt *a_bpf = bpf_lwt_lwtunnel(a);
	505	struct bpf_lwt *b_bpf = bpf_lwt_lwtunnel(b);
	506
	507	return bpf_lwt_prog_cmp(&a_bpf->in, &b_bpf->in) \|\|
	508	bpf_lwt_prog_cmp(&a_bpf->out, &b_bpf->out) \|\|
	509	bpf_lwt_prog_cmp(&a_bpf->xmit, &b_bpf->xmit);
	510	}
	511
	512	static const struct lwtunnel_encap_ops bpf_encap_ops = {
	513	.build_state = bpf_build_state,
	514	.destroy_state = bpf_destroy_state,
	515	.input = bpf_input,
	516	.output = bpf_output,
	517	.xmit = bpf_xmit,
	518	.fill_encap = bpf_fill_encap_info,
	519	.get_encap_size = bpf_encap_nlsize,
	520	.cmp_encap = bpf_encap_cmp,
88ff7334	521	.owner = THIS_MODULE,
3a0af8fd TG	522	};
3a0af8fd TG	523
ca78801a PO	524	static int handle_gso_type(struct sk_buff *skb, unsigned int gso_type,
	525	int encap_len)
	526	{
	527	struct skb_shared_info *shinfo = skb_shinfo(skb);
	528
	529	gso_type \|= SKB_GSO_DODGY;
	530	shinfo->gso_type \|= gso_type;
	531	skb_decrease_gso_size(shinfo, encap_len);
	532	shinfo->gso_segs = 0;
	533	return 0;
	534	}
	535
52f27877 PO	536	static int handle_gso_encap(struct sk_buff *skb, bool ipv4, int encap_len)
52f27877 PO	537	{
ca78801a PO	538	int next_hdr_offset;
	539	void *next_hdr;
	540	__u8 protocol;
	541
	542	/* SCTP and UDP_L4 gso need more nuanced handling than what
	543	* handle_gso_type() does above: skb_decrease_gso_size() is not enough.
	544	* So at the moment only TCP GSO packets are let through.
	545	*/
	546	if (!(skb_shinfo(skb)->gso_type & (SKB_GSO_TCPV4 \| SKB_GSO_TCPV6)))
	547	return -ENOTSUPP;
	548
	549	if (ipv4) {
	550	protocol = ip_hdr(skb)->protocol;
	551	next_hdr_offset = sizeof(struct iphdr);
	552	next_hdr = skb_network_header(skb) + next_hdr_offset;
	553	} else {
	554	protocol = ipv6_hdr(skb)->nexthdr;
	555	next_hdr_offset = sizeof(struct ipv6hdr);
	556	next_hdr = skb_network_header(skb) + next_hdr_offset;
	557	}
	558
	559	switch (protocol) {
	560	case IPPROTO_GRE:
	561	next_hdr_offset += sizeof(struct gre_base_hdr);
	562	if (next_hdr_offset > encap_len)
	563	return -EINVAL;
	564
	565	if (((struct gre_base_hdr *)next_hdr)->flags & GRE_CSUM)
	566	return handle_gso_type(skb, SKB_GSO_GRE_CSUM,
	567	encap_len);
	568	return handle_gso_type(skb, SKB_GSO_GRE, encap_len);
	569
	570	case IPPROTO_UDP:
	571	next_hdr_offset += sizeof(struct udphdr);
	572	if (next_hdr_offset > encap_len)
	573	return -EINVAL;
	574
	575	if (((struct udphdr *)next_hdr)->check)
	576	return handle_gso_type(skb, SKB_GSO_UDP_TUNNEL_CSUM,
	577	encap_len);
	578	return handle_gso_type(skb, SKB_GSO_UDP_TUNNEL, encap_len);
	579
	580	case IPPROTO_IP:
	581	case IPPROTO_IPV6:
	582	if (ipv4)
	583	return handle_gso_type(skb, SKB_GSO_IPXIP4, encap_len);
	584	else
	585	return handle_gso_type(skb, SKB_GSO_IPXIP6, encap_len);
	586
	587	default:
	588	return -EPROTONOSUPPORT;
	589	}
52f27877 PO	590	}
	591
	592	int bpf_lwt_push_ip_encap(struct sk_buff skb, void hdr, u32 len, bool ingress)
	593	{
	594	struct iphdr *iph;
	595	bool ipv4;
	596	int err;
	597
	598	if (unlikely(len < sizeof(struct iphdr) \|\| len > LWT_BPF_MAX_HEADROOM))
	599	return -EINVAL;
	600
	601	/* validate protocol and length */
	602	iph = (struct iphdr *)hdr;
	603	if (iph->version == 4) {
	604	ipv4 = true;
	605	if (unlikely(len < iph->ihl * 4))
	606	return -EINVAL;
	607	} else if (iph->version == 6) {
	608	ipv4 = false;
	609	if (unlikely(len < sizeof(struct ipv6hdr)))
	610	return -EINVAL;
	611	} else {
	612	return -EINVAL;
	613	}
	614
	615	if (ingress)
	616	err = skb_cow_head(skb, len + skb->mac_len);
	617	else
	618	err = skb_cow_head(skb,
	619	len + LL_RESERVED_SPACE(skb_dst(skb)->dev));
	620	if (unlikely(err))
	621	return err;
	622
	623	/* push the encap headers and fix pointers */
	624	skb_reset_inner_headers(skb);
ea0371f7 PO	625	skb_reset_inner_mac_header(skb); /* mac header is not yet set */
ea0371f7 PO	626	skb_set_inner_protocol(skb, skb->protocol);
52f27877 PO	627	skb->encapsulation = 1;
	628	skb_push(skb, len);
	629	if (ingress)
	630	skb_postpush_rcsum(skb, iph, len);
	631	skb_reset_network_header(skb);
	632	memcpy(skb_network_header(skb), hdr, len);
	633	bpf_compute_data_pointers(skb);
	634	skb_clear_hash(skb);
	635
	636	if (ipv4) {
	637	skb->protocol = htons(ETH_P_IP);
	638	iph = ip_hdr(skb);
	639
	640	if (!iph->check)
	641	iph->check = ip_fast_csum((unsigned char *)iph,
	642	iph->ihl);
	643	} else {
	644	skb->protocol = htons(ETH_P_IPV6);
	645	}
	646
	647	if (skb_is_gso(skb))
	648	return handle_gso_encap(skb, ipv4, len);
	649
	650	return 0;
	651	}
	652
3a0af8fd TG	653	static int __init bpf_lwt_init(void)
	654	{
	655	return lwtunnel_encap_add_ops(&bpf_encap_ops, LWTUNNEL_ENCAP_BPF);
	656	}
	657
	658	subsys_initcall(bpf_lwt_init)