wzshiming/socks5

Socks5/Socks5h server and client. Full TCP/Bind/UDP and IPv4/IPv6 support

socks5

Socks5/Socks5h server and client

This project is to add protocol support for the Bridge, or it can be used alone

The following is the implementation of other proxy protocols

• Socks4
• HTTP Proxy
• Shadow Socks
• SSH Proxy
• Any Proxy
• Emux

Usage

API Documentation

Example

• Support for the CONNECT command
• Support for the BIND command
• Support for the ASSOCIATE command

License

Licensed under the MIT License. See LICENSE for the full license text.

socks5's Issues

Use Socks5

Please tell me. Can I use your Socks5 solution to replace the default Socks5 FRP plugin or Socks5.apk (which I currently use) on the tunnel client side to provide UDP support in my connection scheme? And can you help me organize my wiring diagram with your tools?

Does not work with Microsoft Edge Win10.

尝试通过 Socks5 代理建立双向 QUIC 连接失败

demo仓库：https://github.com/WankkoRee/go-http3-proxy

---

我希望通过 QUIC 的双向连接特性来实现端口穿透，即：客户端在主动访问服务端后，服务端通过已经建立起的 QUIC 连接，可以访问客户端上的服务。

未使用代理，本机直连，测试情况符合预期：

未使用代理，客户端位于容器内（模拟NAT网络环境），测试情况符合预期：

使用代理，本机直连，测试发现会变成单向连接：

我现在的需求是使用 Socks5 代理完成双向 QUIC 连接，我应该怎么改动我的代码？

---

我尝试过复写Dialer.ProxyPacketDial直接返回客户端服务的PacketConn，但是发现会报错连接已存在。

handleAssociate

hi,
nice package! thx alot!

there is a problem in: handleAssociate

seems like it returns ipv6 address and port 0

bind: address already in use

im getting this error after killing and restarting the socks5 server

$ ./cmd/socks5/socks5
[socks5] 2023/11/29 17:40:38 listen tcp :1080: bind: address already in use

maybe the server is missing some cleanup code, to close all connections

after killing and restarting the socks5 server

im killing the server with control-c (sigint)
so the server should have enough time for cleanup

run.sh

#!/usr/bin/env bash

set -e

netstat_out="$(netstat -nap 2>/dev/null | grep -w 1080 || true)"
if [ -n "$netstat_out" ]; then
  echo "error: address already in use"
  echo "$netstat_out"
  if true; then
    exit 1
  else
    echo "killing worker procs using port 1080"
    netstat -nap 2>/dev/null | grep -w 1080 | awk '{ print $7 }' | cut -d/ -f1 | xargs kill
  fi
fi

cd cmd/socks5/ &&
go build &&
cd ../.. &&
./cmd/socks5/socks5 &
socks5_pid=$!;
echo socks5_pid: $socks5_pid;
sleep 1;
curl --proxy socks5h://localhost:1080 https://httpbin.org/ip;
kill $socks5_pid

$ ./run.sh 
socks5_pid: 371297
{
  "origin": "xxxxxxxxxxxxx"
}

$ ./run.sh 
error: address already in use
tcp6       0      0 :::1080                 :::*                    LISTEN      371359/./cmd/socks5 
tcp6       0      0 ::1:60936               ::1:1080                TIME_WAIT   -                   

problem: the worker process 371359 keeps running
after then main process 371297 was killed