Example #1
0
/// <summary>
/// Create worker RPC thread
/// </summary>
/// <returns>Thread ID</returns>
DWORD RemoteExec::CreateWorkerThread()
{
    AsmJit::Assembler a;
    AsmJitHelper ah( a );
    AsmJit::Label l_loop = a.newLabel();

    //
    // Create execution thread
    //
    if(!_hWorkThd.valid())
    {
        eModType mt = mt_default;
        if (_memory.core().native()->GetWow64Barrier().type == wow_64_32)
        {
            mt = mt_mod64;

            ah.SwitchTo64();

            // Align stack on 16 byte boundary
            a.and_( AsmJit::nsp, -16 );

            // Allocate new x64 activation stack
            auto createActStack = _mods.GetExport( _mods.GetModule( L"ntdll.dll", LdrList, mt ),
                                                   "RtlAllocateActivationContextStack" ).procAddress;
            if(createActStack)
            {
                ah.GenCall( static_cast<size_t>(createActStack), { _userData.ptr<size_t>() + 0x3000 } );
                a.mov( AsmJit::nax, _userData.ptr<size_t>() + 0x3000 );
                a.mov( AsmJit::nax, AsmJit::sysint_ptr( AsmJit::nax ) );

                ah.SetTebPtr();
                a.mov( AsmJit::sysint_ptr( AsmJit::ndx, 0x2c8 ), AsmJit::nax );
            }
        }

        auto proc = _mods.GetExport( _mods.GetModule( L"ntdll.dll", LdrList, mt ), "NtDelayExecution" ).procAddress;
        auto pExitThread = _mods.GetExport( _mods.GetModule( L"ntdll.dll" ), "NtTerminateThread" ).procAddress;
        if (proc == 0 || pExitThread == 0)
            return 0;

        /*
            for(;;)
                SleepEx(5, TRUE);

            ExitThread(SetEvent(m_hWaitEvent));
        */
        a.bind( l_loop );
        ah.GenCall( static_cast<size_t>(proc), { TRUE, _workerCode.ptr<size_t>() } );
        a.jmp( l_loop );

        ah.ExitThreadWithStatus( pExitThread, _userData.ptr<size_t>() );

        // Write code into process
        LARGE_INTEGER liDelay = { 0 };
        liDelay.QuadPart = -10 * 1000 * 5;

        _workerCode.Write( 0, liDelay );
        _workerCode.Write( sizeof(LARGE_INTEGER), a.getCodeSize(), a.make() );

        _hWorkThd = _threads.CreateNew( _workerCode.ptr<size_t>() + sizeof(LARGE_INTEGER), _userData.ptr<size_t>() );
    }

    return _hWorkThd.id();
}
Example #2
0
/// <summary>
/// Execute code in context of any existing thread
/// </summary>
/// <param name="pCode">Cde to execute</param>
/// <param name="size">Code size.</param>
/// <param name="callResult">Execution result</param>
/// <param name="thd">Target thread</param>
/// <returns>Status</returns>
NTSTATUS RemoteExec::ExecInAnyThread( PVOID pCode, size_t size, uint64_t& callResult, Thread& thd )
{
    NTSTATUS dwResult = STATUS_SUCCESS;
    CONTEXT_T ctx;

    // Prepare for remote exec
    CreateRPCEnvironment( true );

    // Write code
    dwResult = CopyCode( pCode, size );
    if (dwResult != STATUS_SUCCESS)
        return dwResult;

    if (_hWaitEvent)
        ResetEvent( _hWaitEvent );

    if (!thd.Suspend())
        return LastNtStatus();

    if (thd.GetContext( ctx, CONTEXT_ALL, true ))
    {
        AsmJit::Assembler a;
        AsmJitHelper ah( a );

#ifdef _M_AMD64
        const int count = 15;
        AsmJit::GPReg regs[] = { AsmJit::rax, AsmJit::rbx, AsmJit::rcx, AsmJit::rdx, AsmJit::rsi,
                                 AsmJit::rdi, AsmJit::r8,  AsmJit::r9,  AsmJit::r10, AsmJit::r11,
                                 AsmJit::r12, AsmJit::r13, AsmJit::r14, AsmJit::r15, AsmJit::rbp
                               };
        //
        // Preserve thread context
        // I don't care about FPU, XMM and anything else
        //
        a.sub(AsmJit::rsp, count * WordSize);  // Stack must be aligned on 16 bytes
        a.pushfq();                            //

        // Save registers
        for (int i = 0; i < count; i++)
            a.mov( AsmJit::Mem( AsmJit::rsp, i * WordSize ), regs[i] );

        ah.GenCall( _userCode.ptr<size_t>(), { _userData.ptr<size_t>() } );
        AddReturnWithEvent( ah, rt_int32, INTRET_OFFSET );

        // Restore registers
        for (int i = 0; i < count; i++)
            a.mov( regs[i], AsmJit::Mem( AsmJit::rsp, i * WordSize ) );

        a.popfq();
        a.add( AsmJit::rsp, count * WordSize );

        a.jmp( ctx.Rip );
#else
        a.pushad();
        a.pushfd();

        ah.GenCall( _userCode.ptr<size_t>(), { _userData.ptr<size_t>() } );
        AddReturnWithEvent( ah, rt_int32, INTRET_OFFSET );

        a.popfd();
        a.popad();

        a.push( ctx.NIP );
        a.ret();
#endif

        if (_userCode.Write( size, a.getCodeSize(), a.make() ) == STATUS_SUCCESS)
        {
            ctx.NIP = _userCode.ptr<size_t>() + size;

            if (!thd.SetContext( ctx, true ))
                dwResult = LastNtStatus();
        }
        else
            dwResult = LastNtStatus();
    }
    else
        dwResult = LastNtStatus();

    thd.Resume();

    if (dwResult == STATUS_SUCCESS)
    {
        WaitForSingleObject( _hWaitEvent, INFINITE );
        callResult = _userData.Read<size_t>( INTRET_OFFSET, 0 );
    }

    return dwResult;
}