// SPDX-License-Identifier: GPL-2.0
// Copyright (c) 2020 Anton Protopopov
//
// Based on tcpconnect(8) from BCC by Brendan Gregg
#include <vmlinux.h>

#include <bpf/bpf_helpers.h>
#include <bpf/bpf_core_read.h>
#include <bpf/bpf_tracing.h>

#include "maps.bpf.h"
#include "tcpconnect.h"

SEC(".rodata") int filter_ports[MAX_PORTS];
const volatile int filter_ports_len = 0;
const volatile uid_t filter_uid = -1;
const volatile pid_t filter_pid = 0;
const volatile bool do_count = 0;

/* Define here, because there are conflicts with include files */
#define AF_INET		2
#define AF_INET6	10

struct {
	__uint(type, BPF_MAP_TYPE_HASH);
	__uint(max_entries, MAX_ENTRIES);
	__type(key, u32);
	__type(value, struct sock *);
	__uint(map_flags, BPF_F_NO_PREALLOC);
} sockets SEC(".maps");

struct {
	__uint(type, BPF_MAP_TYPE_HASH);
	__uint(max_entries, MAX_ENTRIES);
	__type(key, struct ipv4_flow_key);
	__type(value, u64);
	__uint(map_flags, BPF_F_NO_PREALLOC);
} ipv4_count SEC(".maps");

struct {
	__uint(type, BPF_MAP_TYPE_HASH);
	__uint(max_entries, MAX_ENTRIES);
	__type(key, struct ipv6_flow_key);
	__type(value, u64);
	__uint(map_flags, BPF_F_NO_PREALLOC);
} ipv6_count SEC(".maps");

struct {
	__uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
	__uint(key_size, sizeof(u32));
	__uint(value_size, sizeof(u32));
} events SEC(".maps");

static __always_inline bool filter_port(__u16 port)
{
	int i;

	if (filter_ports_len == 0)
		return false;

	for (i = 0; i < filter_ports_len; i++) {
		if (port == filter_ports[i])
			return false;
	}
	return true;
}

static __always_inline int
enter_tcp_connect(struct pt_regs *ctx, struct sock *sk)
{
	__u64 pid_tgid = bpf_get_current_pid_tgid();
	__u32 pid = pid_tgid >> 32;
	__u32 tid = pid_tgid;
	__u32 uid;

	if (filter_pid && pid != filter_pid)
		return 0;

	uid = bpf_get_current_uid_gid();
	if (filter_uid != (uid_t) -1 && uid != filter_uid)
		return 0;

	bpf_map_update_elem(&sockets, &tid, &sk, 0);
	return 0;
}

static  __always_inline void count_v4(struct sock *sk, __u16 dport)
{
	struct ipv4_flow_key key = {};
	static __u64 zero;
	__u64 *val;

	BPF_CORE_READ_INTO(&key.saddr, sk, __sk_common.skc_rcv_saddr);
	BPF_CORE_READ_INTO(&key.daddr, sk, __sk_common.skc_daddr);
	key.dport = dport;
	val = bpf_map_lookup_or_try_init(&ipv4_count, &key, &zero);
	if (val)
		__atomic_add_fetch(val, 1, __ATOMIC_RELAXED);
}

static __always_inline void count_v6(struct sock *sk, __u16 dport)
{
	struct ipv6_flow_key key = {};
	static const __u64 zero;
	__u64 *val;

	BPF_CORE_READ_INTO(&key.saddr, sk,
			   __sk_common.skc_v6_rcv_saddr.in6_u.u6_addr32);
	BPF_CORE_READ_INTO(&key.daddr, sk,
			   __sk_common.skc_v6_daddr.in6_u.u6_addr32);
	key.dport = dport;

	val = bpf_map_lookup_or_try_init(&ipv6_count, &key, &zero);
	if (val)
		__atomic_add_fetch(val, 1, __ATOMIC_RELAXED);
}

static __always_inline void
trace_v4(struct pt_regs *ctx, pid_t pid, struct sock *sk, __u16 dport)
{
	struct event event = {};

	event.af = AF_INET;
	event.pid = pid;
	event.uid = bpf_get_current_uid_gid();
	event.ts_us = bpf_ktime_get_ns() / 1000;
	BPF_CORE_READ_INTO(&event.saddr_v4, sk, __sk_common.skc_rcv_saddr);
	BPF_CORE_READ_INTO(&event.daddr_v4, sk, __sk_common.skc_daddr);
	event.dport = dport;
	bpf_get_current_comm(event.task, sizeof(event.task));

	bpf_perf_event_output(ctx, &events, BPF_F_CURRENT_CPU,
			      &event, sizeof(event));
}

static __always_inline void
trace_v6(struct pt_regs *ctx, pid_t pid, struct sock *sk, __u16 dport)
{
	struct event event = {};

	event.af = AF_INET6;
	event.pid = pid;
	event.uid = bpf_get_current_uid_gid();
	event.ts_us = bpf_ktime_get_ns() / 1000;
	BPF_CORE_READ_INTO(&event.saddr_v6, sk,
			   __sk_common.skc_v6_rcv_saddr.in6_u.u6_addr32);
	BPF_CORE_READ_INTO(&event.daddr_v6, sk,
			   __sk_common.skc_v6_daddr.in6_u.u6_addr32);
	event.dport = dport;
	bpf_get_current_comm(event.task, sizeof(event.task));

	bpf_perf_event_output(ctx, &events, BPF_F_CURRENT_CPU,
			      &event, sizeof(event));
}

static __always_inline int
exit_tcp_connect(struct pt_regs *ctx, int ret, int ip_ver)
{
	__u64 pid_tgid = bpf_get_current_pid_tgid();
	__u32 pid = pid_tgid >> 32;
	__u32 tid = pid_tgid;
	struct sock **skpp;
	struct sock *sk;
	__u16 dport;

	skpp = bpf_map_lookup_elem(&sockets, &tid);
	if (!skpp)
		return 0;

	if (ret)
		goto end;

	sk = *skpp;

	BPF_CORE_READ_INTO(&dport, sk, __sk_common.skc_dport);
	if (filter_port(dport))
		goto end;

	if (do_count) {
		if (ip_ver == 4)
			count_v4(sk, dport);
		else
			count_v6(sk, dport);
	} else {
		if (ip_ver == 4)
			trace_v4(ctx, pid, sk, dport);
		else
			trace_v6(ctx, pid, sk, dport);
	}

end:
	bpf_map_delete_elem(&sockets, &tid);
	return 0;
}

SEC("kprobe/tcp_v4_connect")
int BPF_KPROBE(tcp_v4_connect, struct sock *sk)
{
	return enter_tcp_connect(ctx, sk);
}

SEC("kretprobe/tcp_v4_connect")
int BPF_KRETPROBE(tcp_v4_connect_ret, int ret)
{
	return exit_tcp_connect(ctx, ret, 4);
}

SEC("kprobe/tcp_v6_connect")
int BPF_KPROBE(tcp_v6_connect, struct sock *sk)
{
	return enter_tcp_connect(ctx, sk);
}

SEC("kretprobe/tcp_v6_connect")
int BPF_KRETPROBE(tcp_v6_connect_ret, int ret)
{
	return exit_tcp_connect(ctx, ret, 6);
}

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