eBPF BCC 实现UNIX socket抓包

在之前，我写了一个《eBPF bpftrace 实现个UNIX socket抓包试试》，但是很受限啊，不能完整打印包数据信息，今天又写了一个BCC的，感觉没问题了。

不多说，直接上代码：

#!/usr/bin/python
# @lint-avoid-python-3-compatibility-imports
# 
# undump        Dump UNIX socket packets.
#               For Linux, uses BCC, eBPF. Embedded C.
# USAGE: undump [-h] [-t] [-p PID]
#
# This uses dynamic tracing of kernel functions, and will need to be updated
# to match kernel changes.
#
# Copyright (c) 2021 Rong Tao.
# Licensed under the GPL License, Version 2.0
#
# 27-Aug-2021   Rong Tao   Created this.
#
from __future__ import print_function
from bcc import BPF
from bcc.containers import filter_by_containers
from bcc.utils import printb
import argparse
from socket import inet_ntop, ntohs, AF_INET, AF_INET6
from struct import pack
from time import sleep
from datetime import datetime
import sys

# arguments
examples = """examples:
    ./undump           # trace/dump all UNIX packets
    ./undump -t        # include timestamps
    ./undump -p 181    # only trace/dump PID 181
"""
parser = argparse.ArgumentParser(
    description="Dump UNIX socket packets",
    formatter_class=argparse.RawDescriptionHelpFormatter,
    epilog=examples)
    
parser.add_argument("-t", "--timestamp", 
        action="store_true", help="include timestamp on output")
parser.add_argument("-p", "--pid",
        help="trace this PID only")
args = parser.parse_args()

# define BPF program
bpf_text = """
#include <uapi/linux/ptrace.h>
#include <net/sock.h>
#include <bcc/proto.h>
#include <linux/aio.h>
#include <linux/socket.h>
#include <linux/net.h>
#include <linux/fs.h>
#include <linux/mount.h>
#include <linux/module.h>
#include <net/sock.h>
#include <net/af_unix.h>


// separate data structs for ipv4 and ipv6
struct stream_data_t {
    u64 ts_us;
    u32 pid;
    u32 uid;
    u32 sock_state;
    u32 sock_type;  //type of socket[STREAM|DRGMA]
    u64 sock_flags;
    char task[TASK_COMM_LEN];
    char *unix_sock_path;
    int msg_namelen;
};
BPF_PERF_OUTPUT(stream_recvmsg_events);

#define MAX_PKT 512
struct recv_data_t {
    u32 recv_len;
    u8  pkt[MAX_PKT];
};

// single element per-cpu array to hold the current event off the stack
BPF_PERCPU_ARRAY(unix_data, struct recv_data_t,1);

BPF_PERF_OUTPUT(unix_recv_events);


//static int unix_stream_recvmsg(struct socket *sock, struct msghdr *msg,
//			       size_t size, int flags)
int trace_stream_entry(struct pt_regs *ctx)
{
    int ret = PT_REGS_RC(ctx);
    
    u64 pid_tgid = bpf_get_current_pid_tgid();
    u32 pid = pid_tgid >> 32;
    u32 tid = pid_tgid;

    FILTER_PID

    struct stream_data_t data4 = {.pid = pid,};
    data4.uid = bpf_get_current_uid_gid();
    data4.ts_us = bpf_ktime_get_ns() / 1000;

    struct socket *sock = (struct socket *)PT_REGS_PARM1(ctx);   
    struct msghdr *msg = (struct msghdr *)PT_REGS_PARM2(ctx);    

    data4.sock_state = sock->state;
    data4.sock_type = sock->type;
    data4.sock_flags = sock->flags;

    data4.msg_namelen = msg->msg_namelen;
    
    bpf_get_current_comm(&data4.task, sizeof(data4.task));
    
    struct unix_sock *unsock = (struct unix_sock *)sock->sk;
    data4.unix_sock_path = (char *)unsock->path.dentry->d_name.name;
    
    stream_recvmsg_events.perf_submit(ctx, &data4, sizeof(data4));

    
    return 0;
};


int trace_unix_stream_read_actor(struct pt_regs *ctx)
{
    u32 zero = 0;
    int ret = PT_REGS_RC(ctx);
    u64 pid_tgid = bpf_get_current_pid_tgid();
    u32 pid = pid_tgid >> 32;
    u32 tid = pid_tgid;

    FILTER_PID

	struct sk_buff *skb = (struct sk_buff *)PT_REGS_PARM1(ctx);

    struct recv_data_t *data = unix_data.lookup(&zero);
    if (!data) 
        return 0;

    unsigned int data_len = skb->len;
    if(data_len > MAX_PKT)
        return 0;
        
    void *iodata = (void *)skb->data;
    data->recv_len = data_len;
    
    bpf_probe_read(data->pkt, data_len, iodata);
    unix_recv_events.perf_submit(ctx, data, data_len+sizeof(u32));
    
    return 0;
}

"""

if args.pid:
    bpf_text = bpf_text.replace('FILTER_PID',
        'if (pid != %s) { return 0; }' % args.pid)

bpf_text = bpf_text.replace('FILTER_PID', '')
    


# process event
def print_stream_event(cpu, data, size):
    event = b["stream_recvmsg_events"].event(data)
    global start_ts
    if args.timestamp:
        if start_ts == 0:
            start_ts = event.ts_us
        printb(b"%-9.3f" % ((float(event.ts_us) - start_ts) / 1000000), nl="")
    printb(b"%-6s %-12s" % (event.pid, event.task))

# process event
def print_recv_pkg(cpu, data, size):
    event = b["unix_recv_events"].event(data)
    print("----------------", end="")
    for i in range(0, event.recv_len):
        print("%02x " % event.pkt[i], end="")
        sys.stdout.flush()
        if (i+1)%16 == 0:
            print("")
            print("----------------", end="")
    print("\n----------------recv %d bytes" % event.recv_len)
    

# initialize BPF
b = BPF(text=bpf_text)
b.attach_kprobe(event="unix_stream_recvmsg", fn_name="trace_stream_entry")
b.attach_kprobe(event="unix_stream_read_actor", fn_name="trace_unix_stream_read_actor")


print("Tracing UNIX socket packets ... Hit Ctrl-C to end")

# header
if args.timestamp:
    print("%-9s" % ("TIME(s)"), end="")


print("%-6s %-12s" % ("PID", "COMM"), end="")
print()


print()
start_ts = 0

# read events
b["stream_recvmsg_events"].open_perf_buffer(print_stream_event)
b["unix_recv_events"].open_perf_buffer(print_recv_pkg)

while True:
    try:
        b.perf_buffer_poll()
    except KeyboardInterrupt:
        exit()

help信息

[root@localhost study]# ./undump.py -h
usage: undump.py [-h] [-t] [-p PID]

Dump UNIX socket packets

optional arguments:
  -h, --help         show this help message and exit
  -t, --timestamp    include timestamp on output
  -p PID, --pid PID  trace this PID only

examples:
    ./undump           # trace/dump all UNIX packets
    ./undump -t        # include timestamps
    ./undump -p 181    # only trace/dump PID 181

效果：

abcdef前面的字段是我的消息头

 

struct MsgHdr {
    int src;
    int dst;
    int id;
    char data[];
}__attribute__((packed));