vmware 的后门

       Vmware 后门是 vmware 和 vmware tools 通信的一个接口。例如，vmware-checkvm 程序就是利用这个后门检测自己是否运行在 vmware 里。

       这个后门开在 IO 端口 0x5658。利用这个后门时，必需：

l EAX = 0x564D5868 ( “VMXh” )

l EBX 为参数，一般不用。

l ECX 低 16 位为功能号。其实是一个函数数组的索引。Vmware 调用对应的函数处理后门请求。 这个函数数组共有 

l 36 个元素，但某些没有定义。ECX 的高 16 位为功能参数。

l EDX = 0x5658 ( “VX” )，为 IO 端口号。

       通过读端口 (in) 命令调用后门。

后门详细描述

0 未定义

1 getMhz 得到 CPU 速率

2 APM 函数族

3 getDiskGeo

4 getPtrLocation

5 setPtrLocation

6 得到宿主机剪贴板数据长度

7 读宿主机剪贴板数据

8 设置宿主机剪贴板数据长度

9 向宿主机剪贴板写数据

10 得到 vmware 版本

11 取设备信息

12 连接或断开设备

13 取 GUI 配置信息

14 设置 GUI 配置信息

15 取宿主机屏幕分辨率

16 未定义

17 未定义

18 osNotFound, vmware 提示插入引导盘

19 GetBiosUUID

20 取虚拟机内存大小

21 未定义

22 OS2 系统用到的一个函数

23 getTime，取宿主机时间

24 stopCatchup

25 未定义

26 未定义

27 未定义

28 initScsiIoprom

29 未定义

30 Message，通道函数族

31 rsvd0

32 rsvd1

33 rsvd2

34 ACPID 函数

35 未定义

应用

vmcall.s

       vmcall.s 为 vm.c 提供 vmcall(uint32_t out[4],int cmd,uint32_t param) 函数，实现调用后门功能

        .text

        .align 2

.globl vmcall

        .type vmcall,@function

vmcall:

        pushl %ebp

        movl %esp, %ebp

        movl 0x8(%ebp),%eax

        push %edi

        push %ebx

        push %ecx

        push %edx

        mov %eax,%edi

        mov $0x564d5868,%edx

        mov %edx,%eax

        mov 0xc(%ebp),%edx

        mov 0x10(%ebp),%ebx

        mov %edx,%ecx

        mov $0x5658,%edx

        in (%dx),%eax

        mov %eax,0x0(%edi)

        mov %ebx,0x4(%edi)

        mov %ecx,0x8(%edi)

        pop %edx

奇怪，被截断了。

        pop %ecx

        pop %ebx

        pop %edi

        leave

        ret

vm.c

       vm.c 为主程序。命令行参数为后门功能号

#include

#include

#include

extern uint32_t vmcall(uint32_t buf[4],int func,uint32_t arg);

void segfault(int seg)

{

        fprintf(stderr,"vmcall failed/n");

        _exit(1);

}

int main(int argc,char **argv)

{

        int i;

        uint32_t buf[4];

        int cmd;

        if (argc == 1)

        {

                puts("arg.......");

                return 1;

        }

        cmd = atoi(argv[1]);

        signal(SIGSEGV,segfault);

        memset(buf,0,sizeof(buf));

        vmcall(buf,cmd,0);

        printf("%x: %x-%x-%x-%x/n",cmd,buf[0],buf[1],buf[2],buf[3]);

         return 0;

}

举例

       编译： gcc –g –o vm vm.c vmcall.s

1. 1 号调用取 CPU 速率

$ ./vm 1

1: 69f-0-1-0

69f 十进制 1695，CPU 速率为 1695Mhz

2. 10 号调用取 vmware 版本，也用来判断是否运行在 vmware 里。

$ ./vm 10

a: 6-564d5868-4-0

6 是 vmware 版本号。注意这和 about 里看到的不同。4 表示是 vmware workstation，其它可能取值有：

2 ESX Server

1 Express

3 GSX Server

3. 15 号调用取宿主机屏幕分辨率

$ ./vm 15

f: 4000300-0-f-0

分辨率是 0x400 * 0x300，即 1024 * 768

4. 28 号调用 vmware 的实现有问题，导致我的 vmware workstation 4.0.0.4460 立刻崩溃。原因Vmware 试图去读 0x14 的内存地址，不过因为没有用户输入，无法利用。

附录：分析工具和环境

      Vmware 4.0.0.4460 和 vmware 4.0.5.6030。宿主机 Windows 2003 Standard。

      GuestOS RedHat 8 (Kernel 2.6.2, gcc 3.2)。

       IDApro 用于静态分析，OllyDBG 用于动态调试。

       静态分析 vmware-checkvm 程序发现这个后门。

       静态分析 + 动态调试得到各功能信息。 

How To Detect VMM Using (Almost) One CPU Instruction

Added by: A^C^E 

Date: 18.11.04 

Time: 08:52:13 

Category: Article 

Source: http://www.securiteam.com/securityreviews/6Z00H20BQS.html 

Summary 

The attached short (4 lines of code, that generate almost a single CPU instruction) exploit code can be used to detect whether the code is executed under a VMM or under a real environment. In addition to the exploit code, a detailed explanation of how this was found and why it works are also provided.

Details 

Swallowing the Red Pill is more or less equivalent to the following code (returns non zero when in Matrix):

int swallow_redpill () {

unsigned char m[2+4], rpill[] = "/x0f/x01/x0d/x00/x00/x00/x00/xc3";

*((unsigned*)&rpill[3]) = (unsigned)m;

((void(*)())&rpill)();

return (m[5]>0xd0) ? 1 : 0;

}

The heart of this code is actually the SIDT instruction (encoded as 0F010D[addr]), which stores the contents of the interrupt descriptor table register (IDTR) in the destination operand, which is actually a memory location. What is special and interesting about SIDT instruction is that, it can be executed in non privileged mode (ring3) but it returns the contents of the sensitive register, used internally by operating system.

Because there is only one IDTR register, but there are at least two OS running concurrently (i.e. the host and the guest OS), VMM needs to relocate the guest's IDTR in a safe place, so that it will not conflict with a host's one. Unfortunately, VMM cannot know if (and when) the process running in guest OS executes SIDT instruction, since it is not privileged (and it doesn't generate exception). Thus the process gets the relocated address of IDT table. It was observed that on VMWare, the relocated address of IDT is at address 0xffXXXXXX, whereas on Virtual PC it is 0xe8XXXXXX. This was tested on VMWare Workstation 4 and Virtual PC 2004, both running on Windows XP host OS.

Joanna Rutkowska came across this strange behavior of SIDT instruction a few years ago, when Joanna Rutkowska was testing Suckit rootkit on VMWare. Joanna Rutkowska noticed that it failed to load on VMWare whereas it seemed to work fine on the same distribution ran outside VM. After spending many hours Joanna Rutkowska figured out that the problematic instruction was actually SIDT, which was used by Suckit to get the address of the IDT table, and to hook its 0x80 entry through /dev/kmem device.

However, Joanna Rutkowska was not the first one who discovered this trick. Shortly after her adventure with Suckit Joanna Rutkowska found a very good USENIX paper about problems when implementing Virtual Machines on Intel processors, discussing of course SIDT problem, as well as many others.

So now, here is the simple code, written in C, which should compile on any all Intel based OS. Just in case you don't have the C compiler for Windows, there is also a binary version attached.

NOTE: This program will fail on systems with PAX/X^W/grsecurity, protection (as it was pointed out by Brad Spengler) since the rpill variable is not marked as executable. To make it run in such systems, mprotect() should be used to mark rpill with PROT_EXEC attribute. Another solution would be to just use asm() keyword instead of shellcode-like buffer. However, this program should be rather considered as a skeleton to build into your own shellcode, rather then standalone production class tool. In addition, Joanna's goal was to make it as simple and portable as possible. That's why Joanna didn't use asm() nor mprotect() since they are system or compiler dependent.

Joanna is also aware of another implementation of this technique, as well as some other tricks to fingerprint VMWare, which can be found at http://www.trapkit.de/.