首页
社区
课程
招聘
[推荐][原创] 强网杯2024 ez_vm 手撕VM + DFA Attack Whitebox AES
发表于: 3天前 2128

[推荐][原创] 强网杯2024 ez_vm 手撕VM + DFA Attack Whitebox AES

3天前
2128

我已参透了符文!。虽迟但到,现在是12月1号,星期日的晚上。本来在中午的时候,就已经破解最后一关。但是下午刚好有钢琴课,只做完DFA的一组,就匆匆下线去上课。最终,上课回来之后也是马不停蹄的接着前面的思路继续做,完成了整个‘破解’过程。这也是我第一次完整分析复杂VM的题目,特此写下此篇,来做个总结。

首先这道题是一个栈虚拟机+JIT,在栈虚拟机中完成白盒AES加密过程。所以做出这道题需要:

总之,要工程量有工程量,要难度有工程量

一个exe程序 接收一个参数(flag)

flag正确输出:)

flag错误输出:(

首先在sub_14009A760中,发现了一个巨大的swicth case,在sub_14009A760调用的前有寄存器保存,后有寄存器恢复,显然sub_14009A760就是vm_dispatcher,在switch中的case就是不同的handler了。

bytecode就相当于VM中的二进制代码一样,而handler就是bytecode的解释器,可以理解为vcpu

下面拿case 0来剖析bytecode的格式

可以看到大switch中还有个小switch,细看小switch中每个case的逻辑都相同只不过操作数的大小不同,其中push_16的代码为:

可以看出来是内存模拟push的操作,最后写入内存的是2字节的数据,所以是push_16其他以此类推。

你可能疑问,为什么没有push8,这里我的理解为:本来x84_64架构中,只有16位/32位/64位,即使在16位下,push 1,实际上也是压入2个字节,即压栈时会根据CPU模式扩展到对应位数

调试查看case 0 + case 3 bytecode的内容为:

op opsize [operand]

在这个例子中 就存在operand 为 20 00 00 00 00 00 00 00,长度为8 byte,与opsize=08对应

存在operand2的例子会在后面讲到,这里先埋个坑。

根据前面的分析。总之,这就是一个push imm的指令,注意这是在栈虚拟机的实现,在虚拟机中有大量的push,pop,这些数push进去后面又pop出来拿来用,翻译x64等效指令就是:

为什么等效?mov reg, imm 其实就是占用一个reg存放imm(reg_index+1),对应于栈虚拟机的栈空间。相应的pop时候,释放栈空间,翻译时reg_index需要减1。(reg_index的动态加减其实就像模拟了rsp的移动)

可以看出时load8/16/32/64的操作,对应的操作可以简化为push(*pop()),所以对应的x64指令为:

case 3,5,6都为store操作,只有细微的差别,这里拿case 5来分析。

操作可以简化为*push() = pop(),翻译为x64指令:

case 7,8,0xb,0xd,0xe,0xf 分别对应add,sub,div,mul,and,or,逻辑类似

这里拿case 7分析,可以看到除了push(pop() + pop())的操作外,还有一系列其他的操作,其实这里是在设置eflags, *(_QWORD *)(a1 + 0x210)存储了虚拟机的eflags。翻译为:

sub,div,mul,and,or 以此类推

乍一看和sub没什么区别,其实不如在case0x12中,v105 = v104 - v103后,并没有push(v105),但是保存了改变的标志位,符合cmp的特征,翻译为:

case 0x15: jmp

可以看到这里会判断不同标志位,进行不同的跳转,这里分析下jmp 的bytecode格式

所以在ture_jmp:中,就是在计算当前的bytecode 的pc与offset的运算,即newpc = pc-offset

在false_jmp:中,直接就是更新pc到下一个bytecode的位置,相当于fall through。

case 0x1b: JIT

到这里会发现代码特别长,有很多陌生函数call,点进去还有一堆call,人工分析难度太大了。这里我是猜出来的,首先看下case 0x1b的bytecode:

然后观察前两行:

结合bytecode的内容,v15 = 3,v16指向了C1所在的地址,然后往后看这两个变量的引用:

发现在memcpy中被使用,所以可以断定v15指定了v16的长度,然后被复制到了内存的某个位置,这里我尝试使用capstone反汇编了一下,发现可以反汇编出来,所以就猜定是JIT了。

ret一目了然

到这里所有handler都知道如何翻译成x64汇编指令,所以可以着手编写parser了,下面是我的parser实现:

有一些要点需要注意:

将产生的parse.s汇编产生的目标文件放入ida中分析:

这里可以看到经典的AES shfit_rows tales(查表法),并且还有9轮大运算,最后是一个循环将密文拷贝到buf2上。

我们需要找到在VM中state[]所在的内存地址,这里shiftRows我们已经逆出来了,定位哪个是shiftRows其实不难,可以发现这个load操作:

但是知道了光知道state[] = a1->rbp + 22没用,因为我们最终还是要在虚拟机中修改,我们需要知道

a1->rbp + 22在虚拟机中的地址。

这里我采取的找到这一行load的汇编:

所以只需要在load处下断点,但是并不是所有load都需要断下,所以在ida中加入条件断点,就在bytecode存放地址的0x7ed处,经过多次调试的经验,发现bytecode的地址是固定,所以索性将

condition写死:

最后state的地址就是rdi中的值

最佳的patch时机应该在第九次列混淆之前,所以我们需要找到这个时机,完成patch

同样的思路,我们注意到了九次循环,如果我们能断在每个循环上,那么我就可以在循环中寻找patch时机,所以找到这个for循环的汇编的cmp指令,作为每次for循环的检查点:

写入ida 断点的condition:

这样知道patch地址和patch时机就可以开始patch了,如果patch了之后发现改变的字节为1个,那就是晚了一轮,如果改变的字节早了一轮那就是早了,总之在倒数第几轮里确定时机,我没记错的话实在地八次到达断点时,patch就是在第九次列混淆之前,patch之后的结果简直完美,只要两组4种group就可以会恢复出last round key。

接下来,写脚本来进行DFA 攻击

根据上一步恢复出来的last round key,恢复初始密钥:

最后提取密文解密

此题得解。

用了整整一个星期,终于把这道题硬生生啃下来了,其中遇到了很多的问题,如理解reg_index和stack的关系,实现过程中cmp的reg_index少减了一个1导致汇编出来的放入ida中的逻辑对不上,reg中一开始也使用rsp加入parse中但是到ida分析中导致一些信息丢失,因为这个问题出现时cmp的错误还没修复,所以我不是很确定是rsp的问题还是我cmp的问题。汇编出来放入ida反编译提示too big function,一度心灰意冷,搜索解决思路还好容易解决,不然卡在最后一步太难受了。还有就是DFA那里,还算顺利,总之就是耐心,细心。

感谢一些前人做的工作,如AES DFA攻击讲解很好,还有优秀的parser辅助我,导致这篇文章可以问世。

chal.exe 3766323862633565396633663134393532356365646630626636363036636630
:)
chal.exe 3766323862633565396633663134393532356365646630626636363036636630
:)
switch ( (char)_RAX )
    {
      case 0:
        __asm { tzcnt   eax, ebx; jumptable 000000014009A7F8 case 0 }
        v86 = _RAX;
        switch ( v86 )
        {
          case 0LL:
            v87 = *(unsigned __int8 *)v10;
            v88 = a1;
            goto LABEL_260;
          case 1LL:
            v87 = *(_WORD *)v10;
            v88 = a1;
LABEL_260:
            push_16(v88, v87);
            break;
          case 2LL:
            push_32(a1, *v10);
            break;
          case 3LL:
            push_64(a1, *(_QWORD *)v10);
            break;
          case 4LL:
            JUMPOUT(0x14009A190LL);
        }
switch ( (char)_RAX )
    {
      case 0:
        __asm { tzcnt   eax, ebx; jumptable 000000014009A7F8 case 0 }
        v86 = _RAX;
        switch ( v86 )
        {
          case 0LL:
            v87 = *(unsigned __int8 *)v10;
            v88 = a1;
            goto LABEL_260;
          case 1LL:
            v87 = *(_WORD *)v10;
            v88 = a1;
LABEL_260:
            push_16(v88, v87);
            break;
          case 2LL:
            push_32(a1, *v10);
            break;
          case 3LL:
            push_64(a1, *(_QWORD *)v10);
            break;
          case 4LL:
            JUMPOUT(0x14009A190LL);
        }
__int64 __fastcall sub_14009A6B0(__int64 a1, __int16 a2)
{
  __int64 result; // rax
 
  result = *(_QWORD *)(a1 + 8); // vrsp
  if ( result == *(_QWORD *)(a1 + 536) )
    BUG();
  *(_QWORD *)(a1 + 8) = result - 2;
  *(_WORD *)(result - 2) = a2;
  return result;
}
__int64 __fastcall sub_14009A6B0(__int64 a1, __int16 a2)
{
  __int64 result; // rax
 
  result = *(_QWORD *)(a1 + 8); // vrsp
  if ( result == *(_QWORD *)(a1 + 536) )
    BUG();
  *(_QWORD *)(a1 + 8) = result - 2;
  *(_WORD *)(result - 2) = a2;
  return result;
}
寄存器指向 bytecode格式 内存大小 注释
rdi→ op 1 byte 大swicth
opsize 1 byte 小switch
[rsi]→ operand 取决于op和opsize if exist
operand2 operand决定 if exist
case 1:
        v89 = (__int16 *)pop_64(a1);
        __asm { tzcnt   ecx, ebx }
        v91 = _RCX;
        switch ( v91 )
        {
          case 0LL:
            v13 = *(unsigned __int8 *)v89;
            goto LABEL_5;
          case 1LL:
            v13 = *v89;
            v8 = a1;
            goto LABEL_6;
          case 2LL:
            v58 = *(_DWORD *)v89;
            goto LABEL_289;
          case 3LL:
            v138 = *(_QWORD *)v89;
            goto LABEL_317;
        }
case 1:
        v89 = (__int16 *)pop_64(a1);
        __asm { tzcnt   ecx, ebx }
        v91 = _RCX;
        switch ( v91 )
        {
          case 0LL:
            v13 = *(unsigned __int8 *)v89;
            goto LABEL_5;
          case 1LL:
            v13 = *v89;
            v8 = a1;
            goto LABEL_6;
          case 2LL:
            v58 = *(_DWORD *)v89;
            goto LABEL_289;
          case 3LL:
            v138 = *(_QWORD *)v89;
            goto LABEL_317;
        }
bytecode format value length
op 0x15 1 byte
opsize 8 1 byte
jmp_condition(operand) 0-9 1 byte
offset(operand2) unknow 8 byte
case 27:
        v15 = (unsigned __int8)v4[2];
        v16 = v4 + 3;
        *(_QWORD *)a1 = v16;
        v2 = 0LL;
        memset(v443, 0, sizeof(v443));
        v444 = 0LL;
        v17 = 0LL;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0x9000000101uLL, 0);
        LOBYTE(v18) = 1;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0xA000000101uLL, v18);
        LOBYTE(v19) = 2;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0xB000000101uLL, v19);
        LOBYTE(v20) = 3;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0xC000000101uLL, v20);
        LOBYTE(v21) = 4;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0xD000000101uLL, v21);
        LOBYTE(v22) = 5;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0xE000000101uLL, v22);
        LOBYTE(v23) = 6;
        // ....
case 27:
        v15 = (unsigned __int8)v4[2];
        v16 = v4 + 3;
        *(_QWORD *)a1 = v16;
        v2 = 0LL;
        memset(v443, 0, sizeof(v443));
        v444 = 0LL;
        v17 = 0LL;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0x9000000101uLL, 0);
        LOBYTE(v18) = 1;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0xA000000101uLL, v18);
        LOBYTE(v19) = 2;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0xB000000101uLL, v19);
        LOBYTE(v20) = 3;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0xC000000101uLL, v20);
        LOBYTE(v21) = 4;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0xD000000101uLL, v21);
        LOBYTE(v22) = 5;
        sub_7FF68C2DCEF0(v7, byte_7FF68C2DA000, 0xE000000101uLL, v22);
        LOBYTE(v23) = 6;
        // ....
1B 08 03 C1 E1 02
1B 08 03 C1 E1 02
case 27:
        v15 = (unsigned __int8)v4[2];// v15 = 3
        v16 = v4 + 3;                // v16 -〉c1 e1 02
case 27:
        v15 = (unsigned __int8)v4[2];// v15 = 3
        v16 = v4 + 3;                // v16 -〉c1 e1 02
memcpy_1(v87 + *(_QWORD *)(a1 + 0x230), v16, v15);
memcpy_1(v87 + *(_QWORD *)(a1 + 0x230), v16, v15);
case 28:
        return a1;
case 28:
        return a1;
from capstone import *
 
md = Cs(CS_ARCH_X86, CS_MODE_64)
 
with open('chal.exe','rb') as f:
    vm_opcode = f.read()[0x97200:0x97200+0x15b8c]
 
print(vm_opcode[:16])
 
pc_max = len(vm_opcode)
pc = 0
reg_index = -1
 
def get_reg():
    reg_name = [
        # 'rax', used
        # 'rbx',
        # 'rcx',
        # 'rdx',
        'rdi',
        'rsi',
        # 'rsp',
        'rbp',
        'r8',
        'r9',
        'r10',
        'r11',
        'r12',
        'r13',
        'r14',
        'r15',
    ]
    assert reg_index >= 0 , "reg_index_error"
    assert reg_index < len(reg_name) , "reg_index_error"
    return reg_name[reg_index]
 
def get_reg_size():
    reg_name_size = [
        # ['al','ax','eax','rax'], used
        # ['bl','bx','ebx','rbx'],
        # ['cl','cx','ecx','rcx'],
        # ['dl','dx','edx','rdx'],
        ['dil','di','edi','rdi'],
        ['sil','si','esi','rsi'],
        # ['spl','sp','esp','rsp'],
        ['bpl','bp','ebp','rbp'],
        ['r8b','r8w','r8d','r8'],
        ['r9b','r9w','r9d','r9'],
        ['r10b','r10w','r10d','r10'],
        ['r11b','r11w','r11d','r11'],
        ['r12b','r12w','r12d','r12'],
        ['r13b','r13w','r13d','r13'],
        ['r14b','r14w','r14d','r14'],
        ['r15b','r15w','r15d','r15'],
    ]
    assert reg_index >= 0 , "reg_index_error"
    assert reg_index < len(reg_name_size) , "reg_index_error"
    return reg_name_size[reg_index][opsize.bit_length()-1]
 
opsize_arr = [1,2,4,8]
x64_asm = []
need_label = set()
pc_infor = []
 
while pc < pc_max:
    opcode = vm_opcode[pc]
    opsize = vm_opcode[pc+1]
    pc_infor.append(pc)
    x64_asm.append(f'lable_{hex(pc)}:')
    assert opsize in opsize_arr, "opsize error"
    if opcode == 0:
        imm = int.from_bytes(vm_opcode[pc+2:pc+2+opsize],'little')
        if opsize == 1:
            print(f"push16 {imm}")
        if opsize == 2:
            print(f"push16 {imm}")
        if opsize == 4:
            print(f"push32 {imm}")
        if opsize == 8:
            print(f"push64 {imm}")
        pc += 2+opsize
        reg_index += 1
        dst_reg = get_reg()
        asm = f'mov %s, {imm}' %  (dst_reg)
        x64_asm.append(asm)
 
    elif opcode == 1:
        if opsize == 1:
            print(f"load16")
        if opsize == 2:
            print(f"load16")
        if opsize == 4:
            print(f"load32")
        if opsize == 8:
            print(f"load64")
        pc += 2
        src_reg = get_reg()
        dst_reg = get_reg_size()
        asm = 'mov %s, [%s]' % (dst_reg, src_reg)
        if opsize < 4: #not support 32->64
            asm += '\nmovzx %s, %s' % (src_reg, dst_reg)
        x64_asm.append(asm)
 
    elif opcode == 2:
        print(f"{opcode} not impl")
        break
 
    elif opcode == 3:
        if opsize == 1:
            print(f"store16")
        if opsize == 2:
            print(f"store16")
        if opsize == 4:
            print(f"store32")
        if opsize == 8:
            print(f"store64")
        pc += 2
        dst_reg = get_reg()
        reg_index -= 1
        src_reg = get_reg_size()
        reg_index -= 1
        asm = "mov [%s], %s" % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 4:
        print(f"{opcode} not impl")
        break
 
    elif opcode == 5:
        if opsize == 1:
            print(f"store8")
        if opsize == 2:
            print(f"store16")
        if opsize == 4:
            print(f"store32u")
        if opsize == 8:
            print(f"store64")
        pc += 2
        dst_reg = get_reg()
        reg_index -= 1
        src_reg = get_reg_size()
        reg_index -= 1
        asm = "mov [%s], %s" % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 6:
        if opsize == 1:
            print(f"store8")
        if opsize == 2:
            print(f"store16")
        pc += 2
        dst_reg = get_reg()
        reg_index -= 1
        src_reg = get_reg_size()
        reg_index -= 1
        asm = "mov [%s], %s" % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 7:
        if opsize == 1:
            print(f"add16")
        if opsize == 2:
            print(f"add16")
        if opsize == 4:
            print(f"add32")
        if opsize == 8:
            print(f"add64")
        pc += 2
        src_reg = get_reg_size()
        reg_index -= 1
        dst_reg = get_reg_size()
        asm = 'add %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
     
 
    elif opcode == 8:
        if opsize == 1:
            print(f"sub16")
        if opsize == 2:
            print(f"sub16")
        if opsize == 4:
            print(f"sub32")
        if opsize == 8:
            print(f"sub64")
        pc += 2
        src_reg = get_reg_size()
        reg_index -= 1
        dst_reg = get_reg_size()
        asm = 'sub %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 9 or opcode == 0xa:
        print(f"{opcode} not impl")
        break
 
    elif opcode == 0x0b:
        if opsize == 1:
            print(f"div16")
        if opsize == 2:
            print(f"div16")
        if opsize == 4:
            print(f"div32")
        if opsize == 8:
            print(f"div64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        asm = 'xor rdx, rdx'
        asm += '\nmov rax, %s' % dst_reg
        asm += '\nmov rcx, %s' % src_reg
        asm += '\ndiv rcx'
        asm += '\nmov %s, rax'% dst_reg
        x64_asm.append(asm)
 
    elif opcode == 0x0c:
        print(f"{opcode} not impl")
        break
 
    elif opcode == 0x0d:
        if opsize == 1:
            print(f"imul16")
        if opsize == 2:
            print(f"imul16")
        if opsize == 4:
            print(f"imul32")
        if opsize == 8:
            print(f"imul64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        asm = 'imul %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 0x0e:
        if opsize == 1:
            print(f"and16")
        if opsize == 2:
            print(f"and16")
        if opsize == 4:
            print(f"and32")
        if opsize == 8:
            print(f"and64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        asm = 'and %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 0x0f:
        if opsize == 1:
            print(f"or16")
        if opsize == 2:
            print(f"or16")
        if opsize == 4:
            print(f"or32")
        if opsize == 8:
            print(f"or64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        asm = 'or %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 0x10:
        if opsize == 1:
            print(f"xor16")
        if opsize == 2:
            print(f"xor16")
        if opsize == 4:
            print(f"xor32")
        if opsize == 8:
            print(f"xor64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        asm = 'xor %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 0x11:
        if opsize == 1:
            print(f"~16")
        if opsize == 2:
            print(f"~16")
        if opsize == 4:
            print(f"~32")
        if opsize == 8:
            print(f"~64")
        pc += 2
        src_reg = get_reg()
        asm = 'not %s' % (src_reg)
        x64_asm.append(asm)
 
    elif opcode == 0x12:
        if opsize == 1:
            print(f"CMP16")
        if opsize == 2:
            print(f"CMP16")
        if opsize == 4:
            print(f"CMP32")
        if opsize == 8:
            print(f"CMP64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        reg_index -= 1
        asm = 'cmp %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 0x13 | opcode == 0x14:
        print(f"{opcode} not impl")
        break
 
    elif opcode == 0x15:
        jmp_condition = vm_opcode[pc+2]
        offset = int.from_bytes(vm_opcode[pc+3:pc+3+8],'little')
        jmp_pc = pc - offset & 2**64 - 1
        target = hex(jmp_pc)
        lable = "lable_%s" % target
        if jmp_condition == 0:
            print("jmp")
            asm = f'jmp {lable}'
            need_label.add(jmp_pc)
        if jmp_condition == 1:
            print("jz")
            asm = f'jz {lable}'
            need_label.add(jmp_pc)
        if jmp_condition == 2:
            print("jnz")
            asm = f'jnz {lable}'
            need_label.add(jmp_pc)
        if jmp_condition == 3:
            print("jbe")
            asm = f'jbe {lable}'
            need_label.add(jmp_pc)
        if jmp_condition == 4:
            print("ja")
            asm = f'ja {lable}'
            need_label.add(jmp_pc)
        if jmp_condition == 5:
            print("jae")
            asm = f'jae {lable}'
            need_label.add(jmp_pc)
        if jmp_condition == 6:
            print("jle")
            asm = f'jle {lable}'
            need_label.add(jmp_pc)
        if jmp_condition == 7:
            print("jg")
            asm = f'jg {lable}'
            need_label.add(jmp_pc)
        if jmp_condition == 8:
            print("jg"
            asm = f'jg {lable}'
            need_label.add(jmp_pc)
        pc += 11
        x64_asm.append(asm)
 
    elif opcode == 0x16:
        print("pushVM")
        reg_index += 1
        dst_reg = get_reg()
        asm = "mov %s ,rbx" % dst_reg
        pc += 2
        x64_asm.append(asm)
 
    elif opcode == 0x17:
        print("add64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        asm = 'add %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 0x18:
        print("imul64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        asm = 'imul %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 0x19:
        print("sub64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        asm = 'sub %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 0x1a:
        base = hex(int.from_bytes(vm_opcode[pc+2:pc+2+opsize],'little'))
        print(f"rebase {base}")
        pc += 2 + opsize
        x64_asm.append('nop')
 
    elif opcode == 0x1b:
        shellcode_len = vm_opcode[pc+2]
        print("JIT")
        print(f"shellcode {shellcode_len}")
        shellcode_byte = vm_opcode[pc+3:pc+3+shellcode_len]
        asm = f'JIT_{pc}:\n'
        for i in md.disasm(shellcode_byte,0):
            asm += f"{i.mnemonic} {i.op_str}\n"
            print(asm)
        pc += 3 + shellcode_len
        x64_asm.append(asm)
 
    elif opcode == 0x1c:
        print('return')
        asm = 'mov rax, rbx\nret'
        x64_asm.append(asm)
        break
 
with open("parse.s",'w') as f:
    f.write('''
.intel_syntax noprefix
.code64
.section .text
    .global _start
_start:
''')
    f.write('mov rbx, rcx\n')
    # for index,asm in enumerate(x64_asm):
    # #     if pc_infor[index] in need_label:
    #     lable = 'lable_' + hex(pc_infor[index])+":"
    #     f.write(lable)
    f.write('\n'.join(x64_asm))
from capstone import *
 
md = Cs(CS_ARCH_X86, CS_MODE_64)
 
with open('chal.exe','rb') as f:
    vm_opcode = f.read()[0x97200:0x97200+0x15b8c]
 
print(vm_opcode[:16])
 
pc_max = len(vm_opcode)
pc = 0
reg_index = -1
 
def get_reg():
    reg_name = [
        # 'rax', used
        # 'rbx',
        # 'rcx',
        # 'rdx',
        'rdi',
        'rsi',
        # 'rsp',
        'rbp',
        'r8',
        'r9',
        'r10',
        'r11',
        'r12',
        'r13',
        'r14',
        'r15',
    ]
    assert reg_index >= 0 , "reg_index_error"
    assert reg_index < len(reg_name) , "reg_index_error"
    return reg_name[reg_index]
 
def get_reg_size():
    reg_name_size = [
        # ['al','ax','eax','rax'], used
        # ['bl','bx','ebx','rbx'],
        # ['cl','cx','ecx','rcx'],
        # ['dl','dx','edx','rdx'],
        ['dil','di','edi','rdi'],
        ['sil','si','esi','rsi'],
        # ['spl','sp','esp','rsp'],
        ['bpl','bp','ebp','rbp'],
        ['r8b','r8w','r8d','r8'],
        ['r9b','r9w','r9d','r9'],
        ['r10b','r10w','r10d','r10'],
        ['r11b','r11w','r11d','r11'],
        ['r12b','r12w','r12d','r12'],
        ['r13b','r13w','r13d','r13'],
        ['r14b','r14w','r14d','r14'],
        ['r15b','r15w','r15d','r15'],
    ]
    assert reg_index >= 0 , "reg_index_error"
    assert reg_index < len(reg_name_size) , "reg_index_error"
    return reg_name_size[reg_index][opsize.bit_length()-1]
 
opsize_arr = [1,2,4,8]
x64_asm = []
need_label = set()
pc_infor = []
 
while pc < pc_max:
    opcode = vm_opcode[pc]
    opsize = vm_opcode[pc+1]
    pc_infor.append(pc)
    x64_asm.append(f'lable_{hex(pc)}:')
    assert opsize in opsize_arr, "opsize error"
    if opcode == 0:
        imm = int.from_bytes(vm_opcode[pc+2:pc+2+opsize],'little')
        if opsize == 1:
            print(f"push16 {imm}")
        if opsize == 2:
            print(f"push16 {imm}")
        if opsize == 4:
            print(f"push32 {imm}")
        if opsize == 8:
            print(f"push64 {imm}")
        pc += 2+opsize
        reg_index += 1
        dst_reg = get_reg()
        asm = f'mov %s, {imm}' %  (dst_reg)
        x64_asm.append(asm)
 
    elif opcode == 1:
        if opsize == 1:
            print(f"load16")
        if opsize == 2:
            print(f"load16")
        if opsize == 4:
            print(f"load32")
        if opsize == 8:
            print(f"load64")
        pc += 2
        src_reg = get_reg()
        dst_reg = get_reg_size()
        asm = 'mov %s, [%s]' % (dst_reg, src_reg)
        if opsize < 4: #not support 32->64
            asm += '\nmovzx %s, %s' % (src_reg, dst_reg)
        x64_asm.append(asm)
 
    elif opcode == 2:
        print(f"{opcode} not impl")
        break
 
    elif opcode == 3:
        if opsize == 1:
            print(f"store16")
        if opsize == 2:
            print(f"store16")
        if opsize == 4:
            print(f"store32")
        if opsize == 8:
            print(f"store64")
        pc += 2
        dst_reg = get_reg()
        reg_index -= 1
        src_reg = get_reg_size()
        reg_index -= 1
        asm = "mov [%s], %s" % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 4:
        print(f"{opcode} not impl")
        break
 
    elif opcode == 5:
        if opsize == 1:
            print(f"store8")
        if opsize == 2:
            print(f"store16")
        if opsize == 4:
            print(f"store32u")
        if opsize == 8:
            print(f"store64")
        pc += 2
        dst_reg = get_reg()
        reg_index -= 1
        src_reg = get_reg_size()
        reg_index -= 1
        asm = "mov [%s], %s" % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 6:
        if opsize == 1:
            print(f"store8")
        if opsize == 2:
            print(f"store16")
        pc += 2
        dst_reg = get_reg()
        reg_index -= 1
        src_reg = get_reg_size()
        reg_index -= 1
        asm = "mov [%s], %s" % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 7:
        if opsize == 1:
            print(f"add16")
        if opsize == 2:
            print(f"add16")
        if opsize == 4:
            print(f"add32")
        if opsize == 8:
            print(f"add64")
        pc += 2
        src_reg = get_reg_size()
        reg_index -= 1
        dst_reg = get_reg_size()
        asm = 'add %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
     
 
    elif opcode == 8:
        if opsize == 1:
            print(f"sub16")
        if opsize == 2:
            print(f"sub16")
        if opsize == 4:
            print(f"sub32")
        if opsize == 8:
            print(f"sub64")
        pc += 2
        src_reg = get_reg_size()
        reg_index -= 1
        dst_reg = get_reg_size()
        asm = 'sub %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)
 
    elif opcode == 9 or opcode == 0xa:
        print(f"{opcode} not impl")
        break
 
    elif opcode == 0x0b:
        if opsize == 1:
            print(f"div16")
        if opsize == 2:
            print(f"div16")
        if opsize == 4:
            print(f"div32")
        if opsize == 8:
            print(f"div64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        asm = 'xor rdx, rdx'
        asm += '\nmov rax, %s' % dst_reg
        asm += '\nmov rcx, %s' % src_reg
        asm += '\ndiv rcx'
        asm += '\nmov %s, rax'% dst_reg
        x64_asm.append(asm)
 
    elif opcode == 0x0c:
        print(f"{opcode} not impl")
        break
 
    elif opcode == 0x0d:
        if opsize == 1:
            print(f"imul16")
        if opsize == 2:
            print(f"imul16")
        if opsize == 4:
            print(f"imul32")
        if opsize == 8:
            print(f"imul64")
        pc += 2
        src_reg = get_reg()
        reg_index -= 1
        dst_reg = get_reg()
        asm = 'imul %s, %s' % (dst_reg, src_reg)
        x64_asm.append(asm)

[招生]科锐逆向工程师培训(2024年11月15日实地,远程教学同时开班, 第51期)

最后于 3天前 被SleepAlone编辑 ,原因:
上传的附件:
收藏
免费 7
支持
分享
最新回复 (3)
雪    币: 10
能力值: ( LV1,RANK:0 )
在线值:
发帖
回帖
粉丝
2
这种vmp比商业的vmp难度如何呢
2天前
0
雪    币: 854
活跃值: (211)
能力值: ( LV5,RANK:60 )
在线值:
发帖
回帖
粉丝
3
哇,眼花缭乱
2天前
0
雪    币: 2502
活跃值: (715)
能力值: ( LV7,RANK:100 )
在线值:
发帖
回帖
粉丝
4
村民九号 哇,眼花缭乱
哪里没看懂可以问我 我写的还是比较详细的
2天前
0
游客
登录 | 注册 方可回帖
返回
//