目录
- 前言
- 一、反调试技术概况
- 二、静态反调试技术
- 1、PEB
- 2、NtQueryInformationProcess()
- (1)ProcessDebugPort(0x7)
- (2)ProcessDebugObjectHandle(0x1E)
- (3)ProcessDebugFlags(0x1F)
- (4)例子
- (5)破解之法
- 3、NtQuerySystemInformation()
- 4、NtQueryObject()
- 5、ZwSerInformationThread()
- 6、ETC
- 三、动态反调试技术
- 1、异常
- (1)SEH
- (2)SetUnhandledExceptionFilter()
- 2、Timing Check
- 3、陷阱标志
- 4、0xCC探测
- (1)API断点
- (2)比较校验和
- 四、高级反调试技术
- 1、垃圾代码
- 2、扰乱代码对齐
- 3、加密/解密
- 4、Stolen Bytes
- 5、API重定向
- 6、Debug Blocker
- 结语
前言
继续学习《逆向工程核心原理》,本篇笔记是第七部分:反调试技术,包括一些静态反调试技术和动态反调试技术
值得注意的是,这本书距今有将近10年了,故这里这些都是比较老的东西了
一、反调试技术概况
反调试技术对调试器和OS有很强的的依赖性,分类如下所示:
二、静态反调试技术
静态反调试技术主要是通过一些API探测调试器,并使程序无法运行
1、PEB
PEB结构体信息可以判断进程是否处于调试状态
几个用到的成员如下:
结构体成员 | 调试信息 | 破解之法 |
| 调试时值为 | 用OD将其修改为 |
| 未使用的堆内存区域全部填充 | 将 |
| 指向 | 将这两个值改回 |
| 调试时值为 | 用OD将其修改为 |
一个例子如下:
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
void PEB()
{
HMODULE hMod = NULL;
FARPROC pProc = NULL;
LPBYTE pTEB = NULL;
LPBYTE pPEB = NULL;
BOOL bIsDebugging = FALSE;
// IsDebuggerPresent()
bIsDebugging = IsDebuggerPresent();
printf("IsDebuggerPresent() = %d\n", bIsDebugging);
if( bIsDebugging ) printf(" => Debugging!!!\n\n");
else printf(" => Not debugging...\n\n");
// Ldr
pProc = GetProcAddress(GetModuleHandle(L"ntdll.dll"), "NtCurrentTeb");
pTEB = (LPBYTE)(*pProc)(); // address of TEB
pPEB = (LPBYTE)*(LPDWORD)(pTEB+0x30); // address of PEB
printf("PEB.Ldr\n");
DWORD pLdrSig[4] = { 0xEEFEEEFE, 0xEEFEEEFE, 0xEEFEEEFE, 0xEEFEEEFE };
LPBYTE pLdr = (LPBYTE)*(LPDWORD)(pPEB+0xC);
__try
{
while( TRUE )
{
if( !memcmp(pLdr, pLdrSig, sizeof(pLdrSig)) )
{
printf(" => Debugging!!!\n\n");
break;
}
pLdr++;
}
}
__except (EXCEPTION_EXECUTE_HANDLER)
{
printf(" => Not debugging...\n\n");
}
// Process Heap - Flags
bIsDebugging = FALSE;
LPBYTE pHeap = (LPBYTE)*(LPDWORD)(pPEB+0x18);
DWORD dwFlags = *(LPDWORD)(pHeap+0xC);
printf("PEB.ProcessHeap.Flags = 0x%X\n", dwFlags);
if( dwFlags != 0x2 ) printf(" => Debugging!!!\n\n");
else printf(" => Not debugging...\n\n");
// Process Heap - ForceFlags
bIsDebugging = FALSE;
DWORD dwForceFlags = *(LPDWORD)(pHeap+0x10);
printf("PEB.ProcessHeap.ForceFlags = 0x%X\n", dwForceFlags);
if( dwForceFlags != 0x0 ) printf(" => Debugging!!!\n\n");
else printf(" => Not debugging...\n\n");
// NtGlobalFlag
bIsDebugging = FALSE;
DWORD dwNtGlobalFlag = *(LPDWORD)(pPEB+0x68);
printf("PEB.NtGlobalFlag = 0x%X\n", dwNtGlobalFlag);
if( (dwNtGlobalFlag & 0x70) == 0x70 ) printf(" => Debugging!!!\n\n");
else printf(" => Not debugging...\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
PEB();
printf("\npress any key to quit...\n");
_gettch();
return 0;
}
2、NtQueryInformationProcess()
NtQueryInformationProcess()
定义如下:
NTSYSAPI NTSTATUS NTAPI NtQueryInformationProcess (
IN HANDLE ProcessHandle, // 进程句柄
IN PROCESSINFOCLASS InformationClass, // 信息类型
OUT PVOID ProcessInformation, // 缓冲指针
IN ULONG ProcessInformationLength, // 以字节为单位的缓冲大小
OUT PULONG ReturnLength OPTIONAL // 写入缓冲的字节数
);
第二个参数PROCESSINFOCLASS
是枚举类型,其中与反调试有关的成员有三个:
-
ProcessDebugPort
(0x7
) -
ProcessDebugObjectHandle
(0x1E
) -
ProcessDebugFlags
(0x1F
)
(1)ProcessDebugPort(0x7)
进程处于调试状态时,操作系统会为他分配1个调试端口(Debug Port)
PROCESSINFOCLASS
设为ProcessDebugPort(0x07)
时,调用NtQueryInformationProcess()
函数就可以获取调试端口
- 若处于调试状态,
dwDebugPort
会被置为0xFFFFFFFF
- 若处于非调试状态,
dwDebugPort
值会被设置为0
(2)ProcessDebugObjectHandle(0x1E)
调试进程时,会生成一个调试对象(Debug Obiect)
NtQueryInformationProcess()
函数的第二个参数值PROCESSINFOCLASS
为ProcessDebugObjectHandle(0x1E)
时,函数的第三个参数就能获取到调试对象句柄
- 进程处于调试状态->调试句柄存在->返回值不为 NULL
- 处于非调试状态 , 返回值为 NULL
(3)ProcessDebugFlags(0x1F)
调试标志(Debug Flags) 的值也可以判断进程是否处于被调试状态
当 NtQueryInformationProcess()
第二个参数PROCESSINFOCLASS
为ProcessDebugFlags(0x1F)
时,第三个参数:
- 调试状态:
0
- 非调试状态:
1
(4)例子
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
enum PROCESSINFOCLASS
{
ProcessBasicInformation = 0,
ProcessQuotaLimits,
ProcessIoCounters,
ProcessVmCounters,
ProcessTimes,
ProcessBasePriority,
ProcessRaisePriority,
ProcessDebugPort = 7,
ProcessExceptionPort,
ProcessAccessToken,
ProcessLdtInformation,
ProcessLdtSize,
ProcessDefaultHardErrorMode,
ProcessIoPortHandlers,
ProcessPooledUsageAndLimits,
ProcessWorkingSetWatch,
ProcessUserModeIOPL,
ProcessEnableAlignmentFaultFixup,
ProcessPriorityClass,
ProcessWx86Information,
ProcessHandleCount,
ProcessAffinityMask,
ProcessPriorityBoost,
MaxProcessInfoClass,
ProcessWow64Information = 26,
ProcessImageFileName = 27,
ProcessDebugObjectHandle = 30,
ProcessDebugFlags = 31,
SystemKernelDebuggerInformation = 35
};
void MyNtQueryInformationProcess()
{
typedef NTSTATUS (WINAPI *NTQUERYINFORMATIONPROCESS)(
HANDLE ProcessHandle,
PROCESSINFOCLASS ProcessInformationClass,
PVOID ProcessInformation,
ULONG ProcessInformationLength,
PULONG ReturnLength
);
NTQUERYINFORMATIONPROCESS pNtQueryInformationProcess = NULL;
pNtQueryInformationProcess = (NTQUERYINFORMATIONPROCESS)
GetProcAddress(GetModuleHandle(L"ntdll.dll"),
"NtQueryInformationProcess");
// ProcessDebugPort (0x7)
DWORD dwDebugPort = 0;
pNtQueryInformationProcess(GetCurrentProcess(),
ProcessDebugPort,
&dwDebugPort,
sizeof(dwDebugPort),
NULL);
printf("NtQueryInformationProcess(ProcessDebugPort) = 0x%X\n", dwDebugPort);
if( dwDebugPort != 0x0 ) printf(" => Debugging!!!\n\n");
else printf(" => Not debugging...\n\n");
// ProcessDebugObjectHandle (0x1E)
HANDLE hDebugObject = NULL;
pNtQueryInformationProcess(GetCurrentProcess(),
ProcessDebugObjectHandle,
&hDebugObject,
sizeof(hDebugObject),
NULL);
printf("NtQueryInformationProcess(ProcessDebugObjectHandle) = 0x%X\n", hDebugObject);
if( hDebugObject != 0x0 ) printf(" => Debugging!!!\n\n");
else printf(" => Not debugging...\n\n");
// ProcessDebugFlags (0x1F)
BOOL bDebugFlag = TRUE;
pNtQueryInformationProcess(GetCurrentProcess(),
ProcessDebugFlags,
&bDebugFlag,
sizeof(bDebugFlag),
NULL);
printf("NtQueryInformationProcess(ProcessDebugFlags) = 0x%X\n", bDebugFlag);
if( bDebugFlag == 0x0 ) printf(" => Debugging!!!\n\n");
else printf(" => Not debugging...\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
MyNtQueryInformationProcess();
printf("\npress any key to quit...\n");
_gettch();
return 0;
}
(5)破解之法
对PROCESSINFOCLASS
的值进行操作
- 直接 HOOK API
- 修改返回值
- 利用 OD 插件 strong OD 中 KernelMode 可以绕过
3、NtQuerySystemInformation()
ntdll!NtQuerySystemInformation()
API是系统函数,用来获取当前运行的多种OS信息
NTSTATUS WINAPI NtQuerySystemInformation(
_In_ SYSTEM_INFORMATION_CLASS SystemInformationClass, //指定需要的系统信息类型
_Inout_ PVOID SystemInformation, //结构体地址
_In_ ULONG SystemInformationLength,
_Out_opt_ PULONG ReturnLength
);
SYSTEM_INFORMATION_CLASS
是枚举类型,如下:
调试状态下 SYSTEM_KERNEL_DEBUGGER_INFORMATION.DebuggerEnabled
值为1
破解之法:
- Win XP 编辑boot.ini,删除
/debugport=com1 /baudrate=115200 /Debug
- Win 7 执行
bcdedit /debug off
可以参考下:NtQuerySystemInformation用法详解
一个例子如下:
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
void MyNtQuerySystemInformation()
{
typedef NTSTATUS (WINAPI *NTQUERYSYSTEMINFORMATION)(
ULONG SystemInformationClass,
PVOID SystemInformation,
ULONG SystemInformationLength,
PULONG ReturnLength
);
typedef struct _SYSTEM_KERNEL_DEBUGGER_INFORMATION
{
BOOLEAN DebuggerEnabled;
BOOLEAN DebuggerNotPresent;
} SYSTEM_KERNEL_DEBUGGER_INFORMATION, *PSYSTEM_KERNEL_DEBUGGER_INFORMATION;
NTQUERYSYSTEMINFORMATION NtQuerySystemInformation;
NtQuerySystemInformation = (NTQUERYSYSTEMINFORMATION)
GetProcAddress(GetModuleHandle(L"ntdll"),
"NtQuerySystemInformation");
ULONG SystemKernelDebuggerInformation = 0x23;
ULONG ulReturnedLength = 0;
SYSTEM_KERNEL_DEBUGGER_INFORMATION DebuggerInfo = {0,};
NtQuerySystemInformation(SystemKernelDebuggerInformation,
(PVOID) &DebuggerInfo,
sizeof(DebuggerInfo), // 2 bytes
&ulReturnedLength);
printf("NtQuerySystemInformation(SystemKernelDebuggerInformation) = 0x%X 0x%X\n",
DebuggerInfo.DebuggerEnabled, DebuggerInfo.DebuggerNotPresent);
if( DebuggerInfo.DebuggerEnabled ) printf(" => Debugging!!!\n\n");
else printf(" => Not debugging...\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
MyNtQuerySystemInformation();
printf("\npress any key to quit...\n");
_gettch();
return 0;
}
4、NtQueryObject()
ntdll!NtQueryObject()
API用来获取各种内核对象的信息
NTSTATUS NtQueryObject(
_In_opt_ HANDLE Handle,
_In_ OBJECT_INFORMATION_CLASS objectInformationClass,
_Out_opt_ PVOID ObjectInformation,
_In_ ULONG ObjectInformationLength,
_Out_opt_ PULONG ReturnLength
);
类似的,是第二个参数是枚举类型,相关信息的结构体指针返回到第三个参数
使用ObjectAllTypesInformation
获取系统所有对象信息,从中检测是否存在调试对象
破解之法:在调用 ntdll.ZwQueryObject()
API的CALL ESI
指令下断点,然后将栈中ObjectAllTypesInformation
的值改为0
一个例子:
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
typedef enum _OBJECT_INFORMATION_CLASS {
ObjectBasicInformation,
ObjectNameInformation,
ObjectTypeInformation,
ObjectAllTypesInformation,
ObjectHandleInformation
} OBJECT_INFORMATION_CLASS, *POBJECT_INFORMATION_CLASS;
void MyNtQueryObject()
{
typedef struct _LSA_UNICODE_STRING {
USHORT Length;
USHORT MaximumLength;
PWSTR Buffer;
} LSA_UNICODE_STRING, *PLSA_UNICODE_STRING, UNICODE_STRING, *PUNICODE_STRING;
typedef NTSTATUS (WINAPI *NTQUERYOBJECT)(
HANDLE Handle,
OBJECT_INFORMATION_CLASS ObjectInformationClass,
PVOID ObjectInformation,
ULONG ObjectInformationLength,
PULONG ReturnLength
);
#pragma pack(1)
typedef struct _OBJECT_TYPE_INFORMATION {
UNICODE_STRING TypeName;
ULONG TotalNumberOfHandles;
ULONG TotalNumberOfObjects;
}OBJECT_TYPE_INFORMATION, *POBJECT_TYPE_INFORMATION;
typedef struct _OBJECT_ALL_INFORMATION {
ULONG NumberOfObjectsTypes;
OBJECT_TYPE_INFORMATION ObjectTypeInformation[1];
} OBJECT_ALL_INFORMATION, *POBJECT_ALL_INFORMATION;
#pragma pack()
POBJECT_ALL_INFORMATION pObjectAllInfo = NULL;
void *pBuf = NULL;
ULONG lSize = 0;
BOOL bDebugging = FALSE;
NTQUERYOBJECT pNtQueryObject = (NTQUERYOBJECT)
GetProcAddress(GetModuleHandle(L"ntdll.dll"),
"NtQueryObject");
// Get the size of the list
pNtQueryObject(NULL, ObjectAllTypesInformation, &lSize, sizeof(lSize), &lSize);
// Allocate list buffer
pBuf = VirtualAlloc(NULL, lSize, MEM_RESERVE | MEM_COMMIT, PAGE_READWRITE);
// Get the actual list
pNtQueryObject((HANDLE)0xFFFFFFFF, ObjectAllTypesInformation, pBuf, lSize, NULL);
pObjectAllInfo = (POBJECT_ALL_INFORMATION)pBuf;
UCHAR *pObjInfoLocation = (UCHAR *)pObjectAllInfo->ObjectTypeInformation;
POBJECT_TYPE_INFORMATION pObjectTypeInfo = NULL;
for( UINT i = 0; i < pObjectAllInfo->NumberOfObjectsTypes; i++ )
{
pObjectTypeInfo = (POBJECT_TYPE_INFORMATION)pObjInfoLocation;
if( wcscmp(L"DebugObject", pObjectTypeInfo->TypeName.Buffer) == 0 )
{
bDebugging = (pObjectTypeInfo->TotalNumberOfObjects > 0) ? TRUE : FALSE;
break;
}
// calculate next struct
pObjInfoLocation = (UCHAR*)pObjectTypeInfo->TypeName.Buffer;
pObjInfoLocation += pObjectTypeInfo->TypeName.Length;
pObjInfoLocation = (UCHAR*)(((ULONG)pObjInfoLocation & 0xFFFFFFFC) + sizeof(ULONG));
}
if( pBuf )
VirtualFree(pBuf, 0, MEM_RELEASE);
printf("NtQueryObject(ObjectAllTypesInformation)\n");
if( bDebugging ) printf(" => Debugging!!!\n\n");
else printf(" => Not debugging...\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
MyNtQueryObject();
printf("\npress any key to quit...\n");
_gettch();
return 0;
}
5、ZwSerInformationThread()
ZwSerInformationThread()
等同于 NtSetInformationThread
,通过将ThreadInformationClass
设置 ThreadHideFromDebugger(0x11)
,可以禁止线程产生调试事件。函数原型如下:
typedef enum _THREAD_INFORMATION_CLASS {
ThreadBasicInformation,
ThreadTimes,
ThreadPriority,
ThreadBasePriority,
ThreadAffinityMask,
ThreadImpersonationToken,
ThreadDescriptorTableEntry,
ThreadEnableAlignmentFaultFixup,
ThreadEventPair,
ThreadQuerySetWin32StartAddress,
ThreadZeroTlsCell,
ThreadPerformanceCount,
ThreadAmILastThread,
ThreadIdealProcessor,
ThreadPriorityBoost,
ThreadSetTlsArrayAddress,
ThreadIsIoPending,
ThreadHideFromDebugger // 17 (0x11)
} THREAD_INFORMATION_CLASS, *PTHREAD_INFORMATION_CLASS;
typedef NTSTATUS (WINAPI* ZWSETINFORMATIONTHREAD)(
HANDLE ThreadHandle, //接收当前线程的句柄
THREAD_INFORMATION_CLASS ThreadInformationClass, //表示线程信息类型
PVOID ThreadInformation,
ULONG ThreadInformationLength
);
破解方法:调试执行到该函数时,若发现第ThreadInformationClass
参数值为 0x11
,跳过或者将修改为0
一个例子:
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
void DetachDebugger()
{
typedef enum _THREAD_INFORMATION_CLASS {
ThreadBasicInformation,
ThreadTimes,
ThreadPriority,
ThreadBasePriority,
ThreadAffinityMask,
ThreadImpersonationToken,
ThreadDescriptorTableEntry,
ThreadEnableAlignmentFaultFixup,
ThreadEventPair,
ThreadQuerySetWin32StartAddress,
ThreadZeroTlsCell,
ThreadPerformanceCount,
ThreadAmILastThread,
ThreadIdealProcessor,
ThreadPriorityBoost,
ThreadSetTlsArrayAddress,
ThreadIsIoPending,
ThreadHideFromDebugger // 17 (0x11)
} THREAD_INFORMATION_CLASS, *PTHREAD_INFORMATION_CLASS;
typedef NTSTATUS (WINAPI* ZWSETINFORMATIONTHREAD)(
HANDLE ThreadHandle,
THREAD_INFORMATION_CLASS ThreadInformationClass,
PVOID ThreadInformation,
ULONG ThreadInformationLength
);
ZWSETINFORMATIONTHREAD pZwSetInformationThread = NULL;
pZwSetInformationThread = (ZWSETINFORMATIONTHREAD)
GetProcAddress(GetModuleHandle(L"ntdll.dll"),
"ZwSetInformationThread");
pZwSetInformationThread(GetCurrentThread(), ThreadHideFromDebugger, NULL, 0);
printf("ZwSetInformationThread() -> Debugger detached!!!\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
DetachDebugger();
printf("\npress any key to quit...\n");
_gettch();
return 0;
}
6、ETC
更简单的思路:判断当前系统是否为逆向分析专用系统,一些例子如下
一个例子:
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
void FindDebuggerWindow()
{
BOOL bDebugging = FALSE;
// using ClassName
if( FindWindow(L"OllyDbg", NULL) || // OllyDbg
FindWindow(L"TIdaWindow", NULL) || // IDA Pro
FindWindow(L"WinDbgFrameClass", NULL) ) // Windbg
bDebugging = TRUE;
printf("FindWindow()\n");
if( bDebugging ) printf(" => Found a debugger window!!!\n\n");
else printf(" => Not found a debugger window...\n\n");
// using WindowName
bDebugging = FALSE;
TCHAR szWindow[MAX_PATH] = {0,};
HWND hWnd = GetDesktopWindow();
hWnd = GetWindow(hWnd, GW_CHILD);
hWnd = GetWindow(hWnd, GW_HWNDFIRST);
while( hWnd )
{
if( GetWindowText(hWnd, szWindow, MAX_PATH) )
{
if( _tcsstr(szWindow, L"IDA") ||
_tcsstr(szWindow, L"OllyDbg") ||
_tcsstr(szWindow, L"WinDbg") )
{
bDebugging = TRUE;
break;
}
}
hWnd = GetWindow(hWnd, GW_HWNDNEXT);
}
printf("GetWindowText()\n");
if( bDebugging ) printf(" => Found a debugger window!!!\n\n");
else printf(" => Not found a debugger window...\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
FindDebuggerWindow();
printf("\npress any key to quit...\n");
_gettch();
return 0;
}
三、动态反调试技术
动态反调试技术扰乱调试器跟踪功能,使程序中的代码和数据无法查看
1、异常
(1)SEH
以基于INT3异常为例
代码执行流如下:
源码如下:
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
void AD_BreakPoint()
{
printf("SEH : BreakPoint\n");
__asm {
// install SEH
push handler
push DWORD ptr fs:[0]
mov DWORD ptr fs:[0], esp
// generating exception
int 3
// 1) debugging
// go to terminating code
mov eax, 0xFFFFFFFF
jmp eax // process terminating!!!
// 2) not debugging
// go to normal code
handler:
mov eax, dword ptr ss:[esp+0xc]
mov ebx, normal_code
mov dword ptr ds:[eax+0xb8], ebx
xor eax, eax
retn
normal_code:
// remove SEH
pop dword ptr fs:[0]
add esp, 4
}
printf(" => Not debugging...\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
AD_BreakPoint();
return 0;
}
破解之法:
(2)SetUnhandledExceptionFilter()
进程中发生异常时,若SEH未处理,则会调用kernel32!SetUnhandledExceptionFilter()
然后弹出进程停止工作的弹窗
一个例子:
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
LPVOID g_pOrgFilter = 0;
LONG WINAPI ExceptionFilter(PEXCEPTION_POINTERS pExcept)
{
SetUnhandledExceptionFilter((LPTOP_LEVEL_EXCEPTION_FILTER)g_pOrgFilter);
// 8900 MOV DWORD PTR DS:[EAX], EAX
// FFE0 JMP EAX
pExcept->ContextRecord->Eip += 4;
return EXCEPTION_CONTINUE_EXECUTION;
}
void AD_SetUnhandledExceptionFilter()
{
printf("SEH : SetUnhandledExceptionFilter()\n");
g_pOrgFilter = (LPVOID)SetUnhandledExceptionFilter(
(LPTOP_LEVEL_EXCEPTION_FILTER)ExceptionFilter);
__asm {
xor eax, eax;
mov dword ptr [eax], eax
jmp eax
}
printf(" => Not debugging...\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
AD_SetUnhandledExceptionFilter();
return 0;
}
2、Timing Check
调试必然要耗费更多时间,故可以通过计算运行时间来判断是否处于调试状态,有两种:
- 用CPU的计数器(Counter)
- 用系统的实际时间(Time)
一个例子:基于RDTSC(Read Time Stamp Counter,读取时间戳计数器)
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
void DynAD_RDTSC()
{
DWORD dwDelta = 0;
printf("Timing Check (RDTSC method)");
__asm {
pushad
// 第一次执行RDTSC
rdtsc
// 将结果TSC放入栈
push edx
push eax
// 用于消耗时间的循环
xor eax, eax
mov ecx, 0x3e8
_LOOP_START:
inc eax
loop _LOOP_START
// 第二次执行RDTSC
rdtsc
// 在栈中输入第一次求得的TSC
pop esi // eax
pop edi // edx
// check high order bits
cmp edx, edi
ja _DEBUGGER_FOUND
// check low order bits
sub eax, esi
mov dwDelta, eax
cmp eax, 0xffffff //若比这个值大,则判定为调试状态
jb _DEBUGGER_NOT_FOUND
// debugger found -> crash!!!
_DEBUGGER_FOUND:
xor eax, eax
mov [eax], eax
// debugger not found
_DEBUGGER_NOT_FOUND:
popad
}
printf(" : delta = %X (ticks)\n", dwDelta);
printf(" => Not debugging...\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
DynAD_RDTSC();
return 0;
}
破解之法:
- 直接用RUN命令越过相关代码
- 将第二个RDTSC值修改为与第一个相同
- 操纵条件分支指令
3、陷阱标志
陷阱标志是EFLAGS寄存器的第9位TF,TF为1时,CPU进入单步执行模式:CPU执行1条指令后,触发EXCEPTION_SINGLE_STEP异常,然后TF归0
这可以和SEH结合起来,例子如下:
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
void DynAD_SingleStep()
{
printf("Trap Flag (Single Step)\n");
__asm {
// install SEH
push handler
push DWORD ptr fs:[0]
mov DWORD ptr fs:[0], esp
// 因无法直接修改EFLAGS,故通过栈修改
pushfd // 将EFLAGS的值压入栈
or dword ptr ss:[esp], 0x100 //将TF值改为1
popfd //将修改后的TF值存入EFLAGS
nop
// 1) debugging
// go to terminating code
mov eax, 0xFFFFFFFF
jmp eax // process terminating!!!
// 2) not debugging
// go to normal code
handler:
mov eax, dword ptr ss:[esp+0xc]
mov ebx, normal_code
mov dword ptr ds:[eax+0xb8], ebx
xor eax, eax
retn
normal_code:
// remove SEH
pop dword ptr fs:[0]
add esp, 4
}
printf(" => Not debugging...\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
DynAD_SingleStep();
return 0;
}
破解之法:
同时在注册SEH的地址和新的EIP地址分别设置断点
4、0xCC探测
0xCC
是断点对应的x86指令,通过探测该指令同样可以判断是否处于调试状态
(1)API断点
调试时,常常是把断点设置在API代码的开始部分,故检测API代码的第一个字节
常用API如下:
(2)比较校验和
调试时做了些操作会导致校验和不同
一个例子
#include "stdio.h"
#include "windows.h"
#include "tchar.h"
DWORD g_dwOrgChecksum = 0xF5934986;
int _tmain(int argc, TCHAR* argv[]);
void DynAD_Checksum()
{
BOOL bDebugging = FALSE;
DWORD dwSize = 0;
printf("Checksum\n");
__asm {
mov ecx, offset _tmain
mov esi, offset DynAD_Checksum
sub ecx, esi // ecx : loop count (buf size)
xor eax, eax // eax : checksum
xor ebx, ebx
_CALC_CHECKSUM:
movzx ebx, byte ptr ds:[esi]
add eax, ebx
rol eax, 1
inc esi
loop _CALC_CHECKSUM
cmp eax, g_dwOrgChecksum
je _NOT_DEBUGGING
mov bDebugging, 1
_NOT_DEBUGGING:
}
if( bDebugging ) printf(" => Debugging!!!\n\n");
else printf(" => Not debugging...\n\n");
}
int _tmain(int argc, TCHAR* argv[])
{
DynAD_Checksum();
return 0;
}
破解之法:修改CRC的比较语句,当然这有难度
四、高级反调试技术
1、垃圾代码
简单讲就是添加大量无意义的代码,更甚的是其中可能还含有有用的代码或反调试技术
一个例子:用一堆SUB和ADD来给EBX设值,然后JMP EBX
2、扰乱代码对齐
向代码插入设计过的不必要的代码来降低反汇编代码的可读性
一个例子:
就A3
对应MOV
,用来处理4个字节的立即数
但是这里的A368
是故意添加的,实际上是7201
,F7进入后可得真实代码
OD和IDA有时候都对这个扰乱束手无策,可能还是停掉智能分析好点
3、加密/解密
一段包含有垃圾代码的加解密如下:
去掉垃圾代码后如下:
其中最核心的指令如下:
EAX为解码循环计数
EBX存放了要解密区域的地址
4、Stolen Bytes
将部分源码(主要是OEP)转移到压缩器创建的内存区域中运行
一个例子:
正常程序如下:
保护后的程序如下:OEP区域已删除
5、API重定向
将主要的API代码复制到其他内存区域,然后修改调用API的代码,从而使复制的API得以执行,这样的话,即使在原API处设置断点也没用
书中给了个结合垃圾代码和扰乱代码对齐的例子,有点复杂,一时没看太明白Q.Q,先记一笔,回头再看看
6、Debug Blocker
用调试模式运行自身程序,这样别的调试器就无法调试正在被调试的进程
- 自我创建技术:父进程负责创建子进程,但是由子进程负责运行实际代码,所以调试器如果调试父进程则得不到OEP,但是如果调试子进程就能找到OEP代码
- Debug Blocker技术弥补了这一缺点,即使是子进程也无法调试
SEH反调试是在同一内存空间处理异常,但是Debug Blocker是在调试进程的内存空间处理被调试进程的异常,所以其他调试器不能附加目标进程,要附加目标进程就先得断开与父进程的联系,但是断开之后子进程也不能继续运行,这就是Debug Blocker的精(bian)妙(tai)之处
另外有进阶版本的Debug Blocker——Nanomite技术:将被调试进程的所有条件跳转指令修改为INT3(软件断点0xCC),或者触发异常的指令,然后会转移到调试者(父进程)执行,调试进程里有条件跳转指令的表格,记录所有跳转地址,取出需要的地址再传给被调试进程,然后继续指令执行。所以要调试Nanomite保护的程序,要恢复出源程序,就得编程自动化实现