Anti-unpacker tricks – part thirteen

2010-10-01

Peter Ferrie

Microsoft, USA
Editor: Helen Martin

Abstract

Last year, a series of articles described some tricks that might become common in the future, along with some countermeasures. Now, the series continues with a look at tricks that are specific to the IDA plug-in, IDA Stealth.


Table of contents

New anti-unpacking tricks continue to be developed as older ones are constantly being defeated. Last year, a series of articles described some tricks that might become common in the future, along with some countermeasures [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13].

In this article we look at some anti-unpacking tricks that are specific to the IDA plug-in IDA Stealth.

Unless stated otherwise, all of the techniques described here were discovered and developed by the author.

IDA plug-ins

A number of packers have been written to detect the IDA debugger, so one plug-in (so far) has been written to attempt to hide it from those packers. The following is a description of that plug-in, with a list of vulnerabilities that could be used to detect it.

IDA plug-ins

The IDA Stealth plug-in was described in a previous paper [8]. What follows is a description of the changes from the previous version, and behaviour that is specific to more recent versions of Windows.

IDA Stealth sets the PEB->Heap->ForceFlags flags to zero, and clears all but the HEAP_GROWABLE flag in the PEB->Heap->Flags flags. However, the location of these fields was moved in Windows Vista, so the plug-in fails to hide the changes. IDA Stealth also clears all but the FLG_STOP_ON_EXCEPTION, FLG_SHOW_LDR_SNAPS, FLG_DEBUG_INITIAL_COMMAND, FLG_STOP_ON_HUNG_GUI and FLG_HEAP_VALIDATE_ALL bits in the PEB->NtGlobalFlag field. Whilst not as wrong as setting bits arbitrarily, this behaviour is still incorrect.

IDA Stealth patches the debuggee’s ntdll RtlGetNtGlobalFlags() function code to always return zero.

IDA Stealth hooks the debuggee’s ntdll NtQuerySystemInformation() function by replacing its first five bytes with a relative jump to an injected DLL. The hook intercepts any attempt to call the ntdll NtQuerySystemInformation() function with the SystemKernelDebuggerInformation class. When that occurs, the hook calls the original ntdll NtQuerySystemInformation() function, and exits if an error occurs. If no error occurs, then IDA Stealth will store a value that corresponds to the cleared KdDebuggerEnabled flag and the KdDebuggerNotPresent flag that is set. However, it is unclear why this function is intercepted, since IDA is not a kernel-mode debugger.

The hook also checks if the ntdll NtQuerySystemInformation() function was called with the SystemProcessInformation class. If so, then the hook calls the original ntdll NtQuerySystemInformation() function. If the call is successful, and the ‘hide IDA’ option is enabled, then the hook searches within the returned buffer for ‘idag.exe’ or ‘idaw.exe’, then erases all copies of the name that are found.

If the ‘fake parent’ option is enabled, then the hook replaces the process ID of the IDA debugger with the process ID of EXPLORER.EXE in the InheritedFromUniqueProcessId field. This could be considered a bug, since the true parent might not be Explorer. The proper behaviour would be to use the process ID of IDA’s parent.

IDA Stealth also hooks the debuggee’s ntdll NtQueryInformationProcess() function by replacing its first five bytes with a relative jump to an injected DLL. The hook intercepts any attempt to call the ntdll NtQueryInformationProcess() function with the ProcessDebugPort class, and returns a zero for the debug port if the current process ID matches the requested process ID.

The hook also checks whether the ntdll NtQueryInformationProcess() function was called with the ProcessBasicInformation class, and that the current process ID matches the requested process ID. If both of those conditions are true, then the hook replaces the parent process ID with that of the shell window in the InheritedFromUniqueProcessId field. This could be considered a bug, since the true parent might not be the shell. The proper behaviour would be to use the process ID of IDA’s parent.

IDA Stealth hooks the debuggee’s ntdll NtQueryObject() function by replacing the first five bytes of the function with a relative jump to an injected DLL. The hook intercepts attempts to call the ntdll NtQueryObject() function with the ObjectAllTypesInformation class. When that occurs, the hook calls the original ntdll NtQueryObject() function, then searches within the returned buffer for the ‘DebugObject’ string. The hook sets the object count to zero if the DebugObject is found.

The plug-in no longer hooks the debuggee’s ntdll NtClose() function. Instead, it patches the ntdll KiUserExceptionDispatcher() function by replacing the function’s first byte with a ‘C3’ opcode (‘RET’ instruction). This has the effect of disabling the exception that is raised when an invalid handle is passed to the ntdll NtClose() function. However, this technique only works on the 32-bit versions of Windows. On 64-bit versions of Windows, the ntdll KiRaiseExceptionDispatcher() function is called as before, but it is the 64-bit version of the function that is being called. That function calls the 64-bit kernel32 RaiseException() function, which eventually delivers the exception to the environment. As a result, the exception remains visible.

Apparently, it would be possible to apply the same first-byte replacement on the 64-bit platform, but it would require the use of an interesting trick. Specifically, the code must use a gate to enter 64-bit mode. From there, it would be a simple matter to call the ntdll NtProtectVirtualMemory() function to unprotect the memory, write the opcode as usual, and then call the ntdll NtProtectVirtualMemory() function again to re-protect the memory. Finally, the code must return to 32-bit mode through the gate. The existence of such a gate has been disclosed publicly [14]. This gate has some interesting properties. For example, it is impossible to query its attributes from user mode. The problem is that the 32-bit ntdll NtQueryInformationThread() function supports only three selectors for a selector query: 0x20 (which corresponds to the CS selector), 0x28 (which corresponds to the DS/ES/FS/GS selectors), and 0x50 (which corresponds to the FS selector). The results are also hard coded, and the behaviour is contained entirely within the wow64.dll file. The call never reaches ntoskrnl.exe.

In any case, disabling the exception without reference to the ‘HKLM\System\CurrentControlSet\Control\Session Manager\GlobalFlag’ registry value means that the absence of the exception might reveal the presence of IDA Stealth.

IDA Stealth hooks the debuggee’s ntdll NtSetInformationThread() function by replacing its first five bytes with a relative jump to an injected DLL. The hook intercepts attempts to call the ntdll NtSetInformationThread() function with the HideThreadFromDebugger class. The hook detects an invalid handle by attempting to duplicate the handle. If the handle is valid, then the hook ignores the request and returns successfully.

The plug-in hooks the debuggee’s kernel32 SuspendThread() function by replacing its first five bytes with a relative jump to an injected DLL. The hook detects an invalid handle by attempting to duplicate it. If the handle is valid, then the hook ignores the request and returns successfully.

IDA Stealth hooks the debuggee’s kernel32 GetTickCount() function by replacing the first five bytes of the function with a relative jump to an injected DLL. When the hook is reached for the first time, it calls the kernel32 QueryPerformanceCounter() function to get an initial value for the tick count. Subsequent calls to the hook cause it to return a tick count that increases in a non-linear fashion.

The plug-in hooks the debuggee’s user32 BlockInput() function by replacing its first five bytes with a relative jump to an injected DLL. When the hook is reached, it simply returns successfully. However, this behaviour is incorrect. Windows will not allow the input to be blocked twice, nor will it allow the input to be enabled twice. Thus, if the same state is passed to the function twice, the result should be different.

Example code looks like this:

push 1
call BlockInput
xchg ebx, eax
push 1
call BlockInput
xor ebx, eax
je being_debugged

IDA Stealth hooks the debuggee’s kernel32 OpenProcess() function once again by replacing its first five bytes with a relative jump to an injected DLL. When the hook is reached, it enumerates the list of processes in order to find the CSRSS.EXE process. If that process is found, then its process ID is compared to the requested process ID. If there is a match, then the hook returns an error code. Otherwise, it calls the original function.

IDA Stealth hooks the debuggee’s user32 SwitchDesktop() function by replacing the first five bytes of the function with a relative jump to an injected DLL. The hook detects an invalid handle by attempting to duplicate the handle. If the handle is valid, then the hook ignores the request and returns successfully.

The plug-in hooks the debugger’s ntdll DbgUiConvertStateChangeStructure() function, if it is available (the API was introduced in Windows XP), by replacing its first five bytes with a relative jump to the plug-in. When the hook is reached, it checks for the DBG_PRINTEXCEPTION_C (0x40010006) exception, and then simply returns success if it is seen. Otherwise, it calls the original function. This allows the exception to be delivered to the debuggee.

IDA Stealth hooks the debuggee’s kernel32 GetThreadContext() function by replacing its first five bytes with a relative jump to an injected DLL. When the hook is reached, it calls the original kernel32 GetThreadContext() function, and then either zeroes the contents of the debug registers, or returns a cached copy of the debug registers. There is a bug in this code which is that the register cache is identified by the current thread ID, instead of the ID of the thread for which the context is being retrieved. Thus, if multiple threads request the same information for the same thread, they might receive different results.

IDA Stealth hooks the debuggee’s kernel32 SetThreadContext() function by replacing its first five bytes with a relative jump to an injected DLL. When the hook is reached, it caches a copy of the debug registers, and then removes the CONTEXT_DEBUG_REGISTERS flag from the ContextFlags field, before completing the call. There are two bugs in that code. The first is that the register cache is identified by the current thread ID, instead of the ID of the thread for which the context is being set. Thus, if one thread sets the context for a second thread, and a third thread retrieves the context for the second thread, then the context might be different. The second bug is that the hook does not check if the lpContext parameter points to a valid memory address, that the lpContext range is readable, or that the first four bytes of the lpContext range are writable. If the lpContext pointer is invalid, not fully readable, or the first four bytes are not writable, then IDA Stealth will cause an exception. The IDA debugger will trap the exception, but the debugging session will be interrupted.

IDA Stealth hooks the debuggee’s ntdll NtYieldExecution() function by replacing its first five bytes with a relative jump to an injected DLL. When the hook is reached, it calls the original ntdll NtYieldExecution() function, then returns successfully.

The plug-in hooks the debuggee’s ntdll NtTerminateThread() function by replacing its first five bytes with a relative jump to an injected DLL. The hook detects an invalid handle by attempting to duplicate the handle. If the handle is valid, then the hook ignores the request and returns successfully. This could be considered a bug, because it disallows the intentional termination of user-created threads.

Similarly, IDA Stealth hooks the debuggee’s ntdll NtTerminateProcess() function by replacing its first five bytes with a relative jump to an injected DLL. The hook detects an invalid handle by attempting to duplicate it. If the handle is valid, then the hook ignores the request and returns successfully. This could be considered a bug, because it disallows the intentional termination of user-created processes.

IDA Stealth hooks the debuggee’s ntdll RtlGetVersion() function, if it exists (the API was introduced in Windows 2000), by replacing the first five bytes of the function with a relative jump to an injected DLL. When the hook is reached, it checks if the RTL_OSVERSIONINFOEXW format was requested. If the RTL_OSVERSIONINFOEXW format was not requested, then the hook assumes that the requested format was the RTL_OSVERSIONINFOW format. This behaviour is identical to that of Windows XP. However, the different versions of Windows behave in different ways regarding this function. The exact problem is how to behave if the size field specifies a buffer that is not large enough to receive the full data. Specifically, Windows 2000 always writes 0x14 bytes before checking the value in this field; Windows XP writes 0x14 bytes and the service pack string before checking the value in this field; Windows Vista and later versions write 0x14 bytes and the service pack string before checking the value in this field, but limit the copy to a maximum of 0xfe bytes. As a result, the function can behave in one of three ways.

Example code looks like this:

   mov eax, fs:[30h]
   mov d [eax+1f4h], offset l2
   push offset l1
   call RtlGetVersion
   ...
l1: db     16h dup (1)
l2: db     100h dup (2), 0, 0

On Windows 2000, the byte at l1+0x14 has a value of 1, because the data is not copied at all. On Windows XP, the byte at l1+0x14 has a value of 2, because the data is copied irrespective of size. On Windows Vista, the byte at l1+0x14 has a value of 0, because the data is copied until the maximum length is reached, and then the value is zeroed because the string is too long. This behaviour allows IDA Stealth to be detected on specific platforms.

Example code looks like this:

   call GetVersion
   cmp  al, 5
   jb   not_supported
   xchg ebx, eax
   mov  eax, fs:[30h]
   mov  d [eax+1f4h], offset l8
   push offset l5
   call GetModuleHandleA
   push offset l6
   push eax
   call GetProcAddress
   push offset l7
   call eax
   movzx    ecx, b [offset l7+14h]
   jecxz    l3
   loop l2
   ;looks like Windows 2000
   ;fail if doesn’t behave like it
   cmp  bx, 5
l1: je l4
   ;fail if unrecognised value
l2: loop   being_debugged
   ;looks like XP
   ;fail if doesn’t behave like it
   cmp  bx, 105h
   jmp  l1
   ;looks like Windows Vista+
   ;fail if doesn’t behave like it
l3: cmp    bl, 6
   jb   being_debugged
l4: ...
l5: db “ntdll”, 0
l6: db “RtlGetVersion”, 0
l7: db 16h dup (1)
l8: db 100h dup (2), 0, 0

IDA Stealth hooks the debugger’s kernel32 DebugActiveProcess() function by replacing its first five bytes with a relative jump to the plug-in. When the hook is reached, it overwrites the entire contents of the debuggee’s ntdll.dll code section with that of the debugger’s ntdll.dll code section, whose size is specified by the SizeOfCode field in the PE header. This has the effect of removing any changes that the debuggee might have made, in an attempt to prevent a debugger from attaching to the process. However, this technique is detected very easily.

Example code looks like this:

   push  offset l3
   call  GetModuleHandleA
   push  offset l4
   push  eax
   call  GetProcAddress
   push  eax
   push  esp
   push  40h ;PAGE_EXECUTE_READWRITE
   push  1
   push  eax
   xchg  ebx, eax
   call  VirtualProtect
   mov   b [ebx], 0c3h
   push  eax
   push  esp
   xor   eax, eax
   push  eax
   push  ebx
   push  offset l1
   push  eax
   push  eax
   call  CreateThread
   ...
l1: pop  eax
   pop  eax
l2: cmp  b [eax], 0c3h
   je   l2
   jmp  being_debugged
l3: db “ntdll”, 0
   ;use a less common API
l4: db “DbgUserBreakPoint”, 0

The plug-in installs a driver that makes the RDTSC instruction illegal when called from ring 3. The driver’s name is ‘rdtscemu’ by default, but it can optionally be a random string value returned by either the kernel32 QueryPerformanceCounter() function or the kernel32 GetTickCount() function.

The driver intercepts the exception that occurs when the instruction is issued. When the exception occurs, the driver executes the RDTSC instruction in ring 0, and then returns a value that is controlled by the driver, as the elapsed time since the last time the RDTSC instruction was executed. The value has a delta applied to it, which is specified as part of a DeviceIoControl() call.

The author of IDA Stealth responded to the report. Some things were changed in version 1.1.1, such as adding support for the heap flags location for Windows Vista. However, the NtClose() problem remains on 64-bit Windows.

The author of IDA Stealth also released version 1.2 shortly afterwards, which introduced a new ‘stealth’ driver, but which also introduced some new bugs. The driver hooks the ntoskrnl NtQueryInformationProcess() function directly in the Service Descriptor Table. The hook calls the original ntoskrnl NtQueryInformationProcess() function, then checks if an error occurred. If no error occurred, then the hook checks if the ProcessInformationClass is the ProcessDebugPort class, and zeroes the port if so. There is a bug in that code, which is that the process handle is not checked. The correct behaviour would have been to zero the port only if the current process is specified.

The hook checks if the ProcessInformationClass is the ProcessDebugObjectHandle class. If it is, then the hook returns a handle value of one. There are three bugs in this code. The first is that the return value is incorrect. When the IDA debugger is active, the function will return successfully. That result alone is enough to reveal the presence of IDA Stealth. The second bug is that by changing the handle value, any legitimate use of that handle becomes impossible. The third bug is that the process handle is not checked. The correct behaviour would have been to return a failure with the correct error code, but only if the current process is specified.

The hook checks if the ProcessInformationClass is the ProcessDebugFlags class, and sets the flags to true if so, signifying that no debugger is present. There is a bug in this code, which is that the process handle is not checked. The correct behaviour would have been to zero the port only if the current process is specified.

Example code demonstrating these techniques was published in a previous paper [3].

The driver hooks the ntoskrnl NtSetInformationThread() function directly in the Service Descriptor Table. The hook intercepts attempts to call the ntdll NtSetInformationThread() function with the HideThreadFromDebugger class. The hook detects an invalid handle by attempting to duplicate the handle. If the handle is valid, then the hook ignores the request and returns successfully.

The author of IDA Stealth responded very quickly to the report. The bugs will be fixed in a future version.

The final part of this series will look at anti-unpacking by emulating.

The text of this paper was produced without reference to any Microsoft source code or personnel.

twitter.png
fb.png
linkedin.png
hackernews.png
reddit.png

 

Latest articles:

Nexus Android banking botnet – compromising C&C panels and dissecting mobile AppInjects

Aditya Sood & Rohit Bansal provide details of a security vulnerability in the Nexus Android botnet C&C panel that was exploited to compromise the C&C panel in order to gather threat intelligence, and present a model of mobile AppInjects.

Cryptojacking on the fly: TeamTNT using NVIDIA drivers to mine cryptocurrency

TeamTNT is known for attacking insecure and vulnerable Kubernetes deployments in order to infiltrate organizations’ dedicated environments and transform them into attack launchpads. In this article Aditya Sood presents a new module introduced by…

Collector-stealer: a Russian origin credential and information extractor

Collector-stealer, a piece of malware of Russian origin, is heavily used on the Internet to exfiltrate sensitive data from end-user systems and store it in its C&C panels. In this article, researchers Aditya K Sood and Rohit Chaturvedi present a 360…

Fighting Fire with Fire

In 1989, Joe Wells encountered his first virus: Jerusalem. He disassembled the virus, and from that moment onward, was intrigued by the properties of these small pieces of self-replicating code. Joe Wells was an expert on computer viruses, was partly…

Run your malicious VBA macros anywhere!

Kurt Natvig wanted to understand whether it’s possible to recompile VBA macros to another language, which could then easily be ‘run’ on any gateway, thus revealing a sample’s true nature in a safe manner. In this article he explains how he recompiled…


Bulletin Archive

We have placed cookies on your device in order to improve the functionality of this site, as outlined in our cookies policy. However, you may delete and block all cookies from this site and your use of the site will be unaffected. By continuing to browse this site, you are agreeing to Virus Bulletin's use of data as outlined in our privacy policy.