#define KBUILD_MODNAME "xdp_l3fwd"
#include <linux/bpf.h>
#include <linux/in.h>
#include <linux/if_ether.h>
#include <linux/if_packet.h>
#include <linux/if_vlan.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
#include <linux/udp.h>
#include <linux/tcp.h>

#include <bpf/bpf_helpers.h>

#define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF)

/* Forwarding ports */
struct bpf_map_def SEC("maps") xdp_l3fwd_ports = {
	.type = BPF_MAP_TYPE_DEVMAP,
	.key_size = sizeof(int),
	.value_size = sizeof(int),
	.max_entries = 512,
};

/* Stats key — identifies a connection or flow */
struct flow_key {
	__u8  proto;
	__u8  pad[3]; /* alignment */
	union {
		__u32 ipv4_src;
		__u8  ipv6_src[16];
	};
	union {
		__u32 ipv4_dst;
		__u8  ipv6_dst[16];
	};
	__u16 sport;
	__u16 dport;
};

/* Stats value — counts packets and bytes */
struct flow_stats {
	__u64 packets;
	__u64 bytes;
};

/* Hash map to track stats per flow */
struct bpf_map_def SEC("maps") xdp_flow_stats = {
	.type = BPF_MAP_TYPE_HASH,
	.key_size = sizeof(struct flow_key),
	.value_size = sizeof(struct flow_stats),
	.max_entries = 65536,
};

/* Decrease IPv4 TTL helper */
static __always_inline int ip_decrease_ttl(struct iphdr *iph)
{
	u32 check = (__force u32)iph->check;
	check += (__force u32)htons(0x0100);
	iph->check = (__force __sum16)(check + (check >= 0xFFFF));
	return --iph->ttl;
}

/* Record stats in the xdp_flow_stats map */
static __always_inline void record_stats(struct xdp_md *ctx, struct flow_key *key, __u64 bytes)
{
	struct flow_stats *stats;

	stats = bpf_map_lookup_elem(&xdp_flow_stats, key);
	if (stats) {
		__sync_fetch_and_add(&stats->packets, 1);
		__sync_fetch_and_add(&stats->bytes, bytes);
	} else {
		struct flow_stats new_stats = {
			.packets = 1,
			.bytes = bytes,
		};
		bpf_map_update_elem(&xdp_flow_stats, key, &new_stats, BPF_ANY);
	}
}

static __always_inline int xdp_l3fwd_flags(struct xdp_md *ctx, u32 flags)
{
	void *data_end = (void *)(long)ctx->data_end;
	void *data = (void *)(long)ctx->data;
	struct bpf_fib_lookup fib_params;
	struct ethhdr *eth = data;
	struct ipv6hdr *ip6h;
	struct iphdr *iph;
	u16 h_proto;
	u64 nh_off;
	int rc;

	nh_off = sizeof(*eth);
	if (data + nh_off > data_end)
		return XDP_DROP;

	__builtin_memset(&fib_params, 0, sizeof(fib_params));
	h_proto = eth->h_proto;

	struct flow_key key = {};
	__u64 bytes = data_end - data;

	if (h_proto == htons(ETH_P_IP)) {
		iph = data + nh_off;
		if ((void *)(iph + 1) > data_end)
			return XDP_DROP;

		key.proto = iph->protocol;
		key.ipv4_src = iph->saddr;
		key.ipv4_dst = iph->daddr;

		void *l4_hdr = (void *)iph + (iph->ihl * 4);
		if (l4_hdr + sizeof(struct udphdr) <= data_end) {
			if (iph->protocol == IPPROTO_TCP) {
				struct tcphdr *tcph = l4_hdr;
				key.sport = tcph->source;
				key.dport = tcph->dest;
			} else if (iph->protocol == IPPROTO_UDP) {
				struct udphdr *udph = l4_hdr;
				key.sport = udph->source;
				key.dport = udph->dest;
			}
		}

		/* Record stats before forwarding */
		record_stats(ctx, &key, bytes);

		if (iph->ttl <= 1)
			return XDP_PASS;

		fib_params.family = AF_INET;
		fib_params.tos = iph->tos;
		fib_params.l4_protocol = iph->protocol;
		fib_params.tot_len = ntohs(iph->tot_len);
		fib_params.ipv4_src = iph->saddr;
		fib_params.ipv4_dst = iph->daddr;

	} else if (h_proto == htons(ETH_P_IPV6)) {
		ip6h = data + nh_off;
		if ((void *)(ip6h + 1) > data_end)
			return XDP_DROP;

		key.proto = ip6h->nexthdr;
		__builtin_memcpy(key.ipv6_src, &ip6h->saddr, 16);
		__builtin_memcpy(key.ipv6_dst, &ip6h->daddr, 16);

		void *l4_hdr = (void *)(ip6h + 1);
		if (l4_hdr + sizeof(struct udphdr) <= data_end) {
			if (ip6h->nexthdr == IPPROTO_TCP) {
				struct tcphdr *tcph = l4_hdr;
				key.sport = tcph->source;
				key.dport = tcph->dest;
			} else if (ip6h->nexthdr == IPPROTO_UDP) {
				struct udphdr *udph = l4_hdr;
				key.sport = udph->source;
				key.dport = udph->dest;
			}
		}

		/* Record stats before forwarding */
		record_stats(ctx, &key, bytes);

		if (ip6h->hop_limit <= 1)
			return XDP_PASS;

		fib_params.family = AF_INET6;
		fib_params.flowinfo = *(__be32 *)ip6h & IPV6_FLOWINFO_MASK;
		fib_params.l4_protocol = ip6h->nexthdr;
		fib_params.tot_len = ntohs(ip6h->payload_len);
		__builtin_memcpy(fib_params.ipv6_src, &ip6h->saddr, 16);
		__builtin_memcpy(fib_params.ipv6_dst, &ip6h->daddr, 16);
	} else {
		return XDP_PASS;
	}

	fib_params.ifindex = ctx->ingress_ifindex;
	rc = bpf_fib_lookup(ctx, &fib_params, sizeof(fib_params), flags);
	if (rc == BPF_FIB_LKUP_RET_SUCCESS) {
		if (!bpf_map_lookup_elem(&xdp_l3fwd_ports, &fib_params.ifindex))
			return XDP_PASS;

		if (h_proto == htons(ETH_P_IP))
			ip_decrease_ttl(iph);
		else if (h_proto == htons(ETH_P_IPV6))
			ip6h->hop_limit--;

		__builtin_memcpy(eth->h_dest, fib_params.dmac, ETH_ALEN);
		__builtin_memcpy(eth->h_source, fib_params.smac, ETH_ALEN);
		return bpf_redirect_map(&xdp_l3fwd_ports, fib_params.ifindex, 0);
	}

	return XDP_PASS;
}

SEC("xdp_l3fwd")
int xdp_l3fwd_prog(struct xdp_md *ctx)
{
	return xdp_l3fwd_flags(ctx, 0);
}

SEC("xdp_l3fwd_direct")
int xdp_l3fwd_direct_prog(struct xdp_md *ctx)
{
	return xdp_l3fwd_flags(ctx, BPF_FIB_LOOKUP_DIRECT);
}

char _license[] SEC("license") = "GPL";