profilers will now have access to unicorn-engine traces so generators

can grab decrypted values using uc_reg_read... powerful!

.gitmodules

@@ -7,3 +7,6 @@
 [submodule "tests/deps/cli-parser"]
 	path = tests/deps/cli-parser
 	url = https://githacks.org/_xeroxz/cli-parser
+[submodule "deps/unicorn"]
+	path = deps/unicorn
+	url = https://github.com/unicorn-engine/unicorn.git

CMakeLists.txt

@@ -90,6 +90,7 @@ target_include_directories(vmprofiler PUBLIC
 target_link_libraries(vmprofiler PUBLIC
 	Zydis
 	linux-pe
+	unicorn
 )
 
 unset(CMKR_TARGET)

cmake.toml

@@ -17,6 +17,7 @@ include-directories = [
 link-libraries = [
     "Zydis",
     "linux-pe",
+    "unicorn"
 ]
 
 compile-definitions = [

deps/CMakeLists.txt

@@ -1,6 +1,23 @@
 # This file is automatically generated from cmake.toml - DO NOT EDIT
 # See https://github.com/build-cpp/cmkr for more information
 
+cmake_minimum_required(VERSION 3.15)
+
+# Regenerate CMakeLists.txt automatically in the root project
+set(CMKR_ROOT_PROJECT OFF)
+if(CMAKE_CURRENT_SOURCE_DIR STREQUAL CMAKE_SOURCE_DIR)
+	set(CMKR_ROOT_PROJECT ON)
+
+	# Bootstrap cmkr
+	include(cmkr.cmake OPTIONAL RESULT_VARIABLE CMKR_INCLUDE_RESULT)
+	if(CMKR_INCLUDE_RESULT)
+		cmkr()
+	endif()
+
+	# Enable folder support
+	set_property(GLOBAL PROPERTY USE_FOLDERS ON)
+endif()
+
 # Create a configure-time dependency on cmake.toml to improve IDE support
 if(CMKR_ROOT_PROJECT)
 	configure_file(cmake.toml cmake.toml COPYONLY)
@@ -23,6 +40,16 @@ endif()
 add_subdirectory(zydis)
 set(CMAKE_FOLDER ${CMKR_CMAKE_FOLDER})
 
+# unicorn
+set(CMKR_CMAKE_FOLDER ${CMAKE_FOLDER})
+if(CMAKE_FOLDER)
+	set(CMAKE_FOLDER "${CMAKE_FOLDER}/unicorn")
+else()
+	set(CMAKE_FOLDER unicorn)
+endif()
+add_subdirectory(unicorn)
+set(CMAKE_FOLDER ${CMKR_CMAKE_FOLDER})
+
 # Target linux-pe
 set(CMKR_TARGET linux-pe)
 set(linux-pe_SOURCES "")

deps/cmake.toml

@@ -6,6 +6,7 @@ ZYDIS_FUZZ_AFL_FAST = false
 ZYDIS_LIBFUZZER = false
 
 [subdir.zydis]
+[subdir.unicorn]
 
 [target.linux-pe]
 type = "interface"

deps/unicorn

@@ -0,0 +1 @@
+Subproject commit 63a445cbba18bf1313ac3699b5d25462b5d529f4

include/vmctx.hpp

@@ -1,4 +1,5 @@
 #pragma once
+#include <vminstrs.hpp>
 #include <vmutils.hpp>
 
 namespace vm {

include/vminstrs.hpp

@@ -1,4 +1,5 @@
 #pragma once
+#include <unicorn/unicorn.h>
 #include <vmutils.hpp>
 
 namespace vm::instrs {
@@ -68,6 +69,24 @@ struct vinstr_t {
   } imm;
 };
 
+/// <summary>
+/// emu instruction containing current cpu register values and such...
+/// </summary>
+struct emu_instr_t {
+  zydis_decoded_instr_t m_instr;
+  uc_context* m_cpu;
+};
+
+/// <summary>
+/// handler trace containing information about a stream of instructions... also
+/// contains some information about the virtual machine such as vip and vsp...
+/// </summary>
+struct hndlr_trace_t {
+  std::uintptr_t m_hndlr_addr;
+  zydis_reg_t m_vip, m_vsp;
+  std::vector<emu_instr_t> m_instrs;
+};
+
 /// <summary>
 /// matcher function which returns true if an instruction matches a desired
 /// one...
@@ -80,9 +99,10 @@ using matcher_t = std::function<bool(const zydis_reg_t vip,
 /// virtual instruction structure generator... this can update the vip and vsp
 /// argument... it cannot update the instruction stream (hndlr)...
 /// </summary>
-using vinstr_gen_t = std::function<std::optional<vinstr_t>(zydis_reg_t& vip,
+using vinstr_gen_t =
-                                                           zydis_reg_t& vsp,
+    std::function<std::optional<vinstr_t>(zydis_reg_t& vip,
-                                                           zydis_rtn_t& hndlr)>;
+                                          zydis_reg_t& vsp,
+                                          hndlr_trace_t& hndlr)>;
 
 /// <summary>
 /// each virtual instruction has its own profiler_t structure which can generate
@@ -111,21 +131,52 @@ struct profiler_t {
   vinstr_gen_t generate;
 };
 
+/// <summary>
+/// list of all profiles here...
+/// </summary>
 extern profiler_t jmp;
+extern profiler_t sreg;
+
+/// <summary>
+/// unsorted vector of profiles... they get sorted once at runtime...
+/// </summary>
 inline std::vector<profiler_t*> profiles = {&jmp};
 
+/// <summary>
+/// sorts the profiles by descending order of matchers... this will prevent a
+/// smaller profiler with less matchers from being used when it should not be...
+///
+/// this function can be called multiple times...
+/// </summary>
+inline void init() {
+  if (static std::atomic_bool once = true; once.exchange(false))
+    std::sort(profiles.begin(), profiles.end(),
+              [&](profiler_t* a, profiler_t* b) -> bool {
+                return a->matchers.size() > b->matchers.size();
+              });
+}
+
+/// <summary>
+/// determines the virtual instruction for the vm handler given vsp and vip...
+/// </summary>
+/// <param name="vip">vip native register...</param>
+/// <param name="vsp">vsp native register...</param>
+/// <param name="hndlr"></param>
+/// <returns>returns vinstr_t structure...</returns>
 inline vinstr_t determine(zydis_reg_t& vip,
                           zydis_reg_t& vsp,
-                          zydis_rtn_t& hndlr) {
+                          hndlr_trace_t& hndlr) {
+  const auto& instrs = hndlr.m_instrs;
   const auto profile = std::find_if(
       profiles.begin(), profiles.end(), [&](profiler_t* profile) -> bool {
         for (auto& matcher : profile->matchers) {
-          const auto matched = std::find_if(hndlr.begin(), hndlr.end(),
+          const auto matched =
-                                            [&](zydis_instr_t& instr) -> bool {
+              std::find_if(instrs.begin(), instrs.end(),
-                                              const auto& i = instr.instr;
+                           [&](const emu_instr_t& instr) -> bool {
-                                              return matcher(vip, vsp, i);
+                             const auto& i = instr.m_instr;
-                                            });
+                             return matcher(vip, vsp, i);
-          if (matched == hndlr.end())
+                           });
+          if (matched == instrs.end())
             return false;
         }
         return true;
@@ -137,4 +188,21 @@ inline vinstr_t determine(zydis_reg_t& vip,
   auto result = (*profile)->generate(vip, vsp, hndlr);
   return result.has_value() ? result.value() : vinstr_t{mnemonic_t::unknown};
 }
+
+/// <summary>
+/// get profile from mnemonic...
+/// </summary>
+/// <param name="mnemonic">mnemonic of the profile to get...</param>
+/// <returns>pointer to the profile...</returns>
+inline profiler_t* get_profile(mnemonic_t mnemonic) {
+  if (mnemonic == mnemonic_t::unknown)
+    return nullptr;
+
+  const auto res = std::find_if(profiles.begin(), profiles.end(),
+                                [&](profiler_t* profile) -> bool {
+                                  return profile->mnemonic == mnemonic;
+                                });
+
+  return res == profiles.end() ? nullptr : *res;
+}
 }  // namespace vm::instrs

src/vmctx.cpp

@@ -11,6 +11,9 @@ vmctx_t::vmctx_t(std::uintptr_t module_base,
       m_image_size(image_size) {}
 
 bool vmctx_t::init() {
+  vm::utils::init();
+  vm::instrs::init();
+
   // flatten and deobfuscate the vm entry...
   if (!vm::utils::flatten(m_vm_entry, m_module_base + m_vm_entry_rva))
     return false;

src/vmprofiles/jmp.cpp

@@ -4,8 +4,8 @@ namespace vm::instrs {
 profiler_t jmp = {
     "JMP",
     mnemonic_t::jmp,
-    // MOV REG, [VSP]
+    {{// MOV REG, [VSP]
-    {{[&](const zydis_reg_t vip,
+      [&](const zydis_reg_t vip,
           const zydis_reg_t vsp,
           const zydis_decoded_instr_t& instr) -> bool {
         return instr.mnemonic == ZYDIS_MNEMONIC_MOV &&
@@ -34,10 +34,11 @@ profiler_t jmp = {
       }}},
     [&](zydis_reg_t& vip,
         zydis_reg_t& vsp,
-        zydis_rtn_t& hndlr) -> std::optional<vinstr_t> {
+        hndlr_trace_t& hndlr) -> std::optional<vinstr_t> {
+      const auto& instrs = hndlr.m_instrs;
       const auto xchg = std::find_if(
-          hndlr.begin(), hndlr.end(), [&](const zydis_instr_t& instr) -> bool {
+          instrs.begin(), instrs.end(), [&](const emu_instr_t& instr) -> bool {
-            const auto& i = instr.instr;
+            const auto& i = instr.m_instr;
             return i.mnemonic == ZYDIS_MNEMONIC_XCHG &&
                    i.operands[0].type == ZYDIS_OPERAND_TYPE_REGISTER &&
                    i.operands[1].type == ZYDIS_OPERAND_TYPE_REGISTER &&
@@ -50,77 +51,77 @@ profiler_t jmp = {
           });
 
       // this JMP virtual instruction changes VSP as well as VIP...
-      if (xchg != hndlr.end()) {
+      if (xchg != instrs.end()) {
         // grab the register that isnt VSP in the XCHG...
         // xchg reg, vsp or xchg vsp, reg...
-        zydis_reg_t write_dep = xchg->instr.operands[0].reg.value != vsp
+        zydis_reg_t write_dep = xchg->m_instr.operands[0].reg.value != vsp
-                                    ? xchg->instr.operands[0].reg.value
+                                    ? xchg->m_instr.operands[0].reg.value
-                                    : xchg->instr.operands[1].reg.value;
+                                    : xchg->m_instr.operands[1].reg.value;
 
         // update VIP... VSP becomes VIP... with the XCHG...
-        vip = xchg->instr.operands[0].reg.value != vsp
+        vip = xchg->m_instr.operands[0].reg.value != vsp
-                  ? xchg->instr.operands[1].reg.value
+                  ? xchg->m_instr.operands[1].reg.value
-                  : xchg->instr.operands[0].reg.value;
+                  : xchg->m_instr.operands[0].reg.value;
 
         // find the next MOV REG, write_dep... this REG will be VSP...
         const auto mov_reg_write_dep = std::find_if(
-            xchg, hndlr.end(), [&](const zydis_instr_t& instr) -> bool {
+            xchg, instrs.end(), [&](const emu_instr_t& instr) -> bool {
-              const auto& i = instr.instr;
+              const auto& i = instr.m_instr;
               return i.mnemonic == ZYDIS_MNEMONIC_MOV &&
                      i.operands[0].type == ZYDIS_OPERAND_TYPE_REGISTER &&
                      i.operands[1].type == ZYDIS_OPERAND_TYPE_REGISTER &&
                      i.operands[1].reg.value == write_dep;
             });
 
-        if (mov_reg_write_dep == hndlr.end())
+        if (mov_reg_write_dep == instrs.end())
           return {};
 
-        vsp = mov_reg_write_dep->instr.operands[0].reg.value;
+        vsp = mov_reg_write_dep->m_instr.operands[0].reg.value;
       } else {
         // find the MOV REG, [VSP] instruction...
         const auto mov_reg_deref_vsp = std::find_if(
-            hndlr.begin(), hndlr.end(),
+            instrs.begin(), instrs.end(),
-            [&](const zydis_instr_t& instr) -> bool {
+            [&](const emu_instr_t& instr) -> bool {
-              const auto& i = instr.instr;
+              const auto& i = instr.m_instr;
               return i.mnemonic == ZYDIS_MNEMONIC_MOV &&
                      i.operands[0].type == ZYDIS_OPERAND_TYPE_REGISTER &&
                      i.operands[1].type == ZYDIS_OPERAND_TYPE_MEMORY &&
                      i.operands[1].mem.base == vsp;
             });
 
-        if (mov_reg_deref_vsp == hndlr.end())
+        if (mov_reg_deref_vsp == instrs.end())
           return {};
 
         // find the MOV REG, mov_reg_deref_vsp->operands[0].reg.value
         const auto mov_vip_reg = std::find_if(
-            mov_reg_deref_vsp, hndlr.end(),
+            mov_reg_deref_vsp, instrs.end(),
-            [&](const zydis_instr_t& instr) -> bool {
+            [&](const emu_instr_t& instr) -> bool {
-              const auto& i = instr.instr;
+              const auto& i = instr.m_instr;
               return i.mnemonic == ZYDIS_MNEMONIC_MOV &&
                      i.operands[0].type == ZYDIS_OPERAND_TYPE_REGISTER &&
                      i.operands[1].type == ZYDIS_OPERAND_TYPE_REGISTER &&
                      i.operands[1].reg.value ==
-                         mov_reg_deref_vsp->instr.operands[0].reg.value;
+                         mov_reg_deref_vsp->m_instr.operands[0].reg.value;
             });
 
-        if (mov_vip_reg == hndlr.end())
+        if (mov_vip_reg == instrs.end())
           return {};
 
-        vip = mov_vip_reg->instr.operands[0].reg.value;
+        vip = mov_vip_reg->m_instr.operands[0].reg.value;
 
         // see if VSP gets updated as well...
         const auto mov_reg_vsp = std::find_if(
-            mov_reg_deref_vsp, hndlr.end(),
+            mov_reg_deref_vsp, instrs.end(),
-            [&](const zydis_instr_t& instr) -> bool {
+            [&](const emu_instr_t& instr) -> bool {
-              const auto& i = instr.instr;
+              const auto& i = instr.m_instr;
               return i.mnemonic == ZYDIS_MNEMONIC_MOV &&
                      i.operands[0].type == ZYDIS_OPERAND_TYPE_REGISTER &&
                      i.operands[1].type == ZYDIS_OPERAND_TYPE_REGISTER &&
                      i.operands[1].reg.value == vsp;
             });
 
-        if (mov_reg_vsp != hndlr.end())
+        if (mov_reg_vsp != instrs.end())
-          vsp = mov_reg_vsp->instr.operands[0].reg.value;
+          vsp = mov_reg_vsp->m_instr.operands[0].reg.value;
       }
       return vinstr_t{mnemonic_t::jmp};
     }};

src/vmprofiles/sreg.cpp

@@ -0,0 +1,48 @@
+#include <vminstrs.hpp>
+
+namespace vm::instrs {
+profiler_t sreg = {
+    "SREG",
+    mnemonic_t::sreg,
+    {{// MOV REG, [VSP]
+      [&](const zydis_reg_t vip,
+          const zydis_reg_t vsp,
+          const zydis_decoded_instr_t& instr) -> bool {
+        return instr.mnemonic == ZYDIS_MNEMONIC_MOV &&
+               instr.operands[0].type == ZYDIS_OPERAND_TYPE_REGISTER &&
+               instr.operands[1].type == ZYDIS_OPERAND_TYPE_MEMORY &&
+               instr.operands[1].mem.base == vsp;
+      },
+      // ADD VSP, 8
+      [&](const zydis_reg_t vip,
+          const zydis_reg_t vsp,
+          const zydis_decoded_instr_t& instr) -> bool {
+        return instr.mnemonic == ZYDIS_MNEMONIC_ADD &&
+               instr.operands[0].type == ZYDIS_OPERAND_TYPE_REGISTER &&
+               instr.operands[0].reg.value == vsp &&
+               instr.operands[1].type == ZYDIS_OPERAND_TYPE_IMMEDIATE &&
+               instr.operands[1].imm.value.u == 8;
+      },
+      // MOV REG, [VIP]
+      [&](const zydis_reg_t vip,
+          const zydis_reg_t vsp,
+          const zydis_decoded_instr_t& instr) -> bool {
+        return instr.mnemonic == ZYDIS_MNEMONIC_MOV &&
+               instr.operands[0].type == ZYDIS_OPERAND_TYPE_REGISTER &&
+               instr.operands[1].type == ZYDIS_OPERAND_TYPE_MEMORY &&
+               instr.operands[1].mem.base == vip;
+      },
+      // MOV [RSP+REG], REG
+      [&](const zydis_reg_t vip,
+          const zydis_reg_t vsp,
+          const zydis_decoded_instr_t& instr) -> bool {
+        return instr.mnemonic == ZYDIS_MNEMONIC_MOV &&
+               instr.operands[0].type == ZYDIS_OPERAND_TYPE_MEMORY &&
+               instr.operands[0].mem.base == ZYDIS_REGISTER_RSP &&
+               instr.operands[0].mem.index != ZYDIS_REGISTER_NONE &&
+               instr.operands[1].type == ZYDIS_OPERAND_TYPE_REGISTER;
+      }}},
+    [&](zydis_reg_t& vip,
+        zydis_reg_t& vsp,
+        hndlr_trace_t& hndlr) -> std::optional<vinstr_t> { return {}; }};
+}

src/vmutils.cpp

@@ -119,8 +119,7 @@ void deobfuscate(zydis_rtn_t& routine) {
   std::uint32_t last_size = 0u;
   static const std::vector<ZydisMnemonic> blacklist = {
       ZYDIS_MNEMONIC_CLC, ZYDIS_MNEMONIC_BT,  ZYDIS_MNEMONIC_TEST,
-      ZYDIS_MNEMONIC_CMP, ZYDIS_MNEMONIC_CMC, ZYDIS_MNEMONIC_STC,
+      ZYDIS_MNEMONIC_CMP, ZYDIS_MNEMONIC_CMC, ZYDIS_MNEMONIC_STC};
-      ZYDIS_MNEMONIC_JMP};
 
   static const std::vector<ZydisMnemonic> whitelist = {
       ZYDIS_MNEMONIC_PUSH, ZYDIS_MNEMONIC_POP, ZYDIS_MNEMONIC_CALL,
@@ -139,6 +138,11 @@ void deobfuscate(zydis_rtn_t& routine) {
         break;
       }
 
+      if (is_jmp(itr->instr)) {
+        routine.erase(itr);
+        break;
+      }
+
       zydis_reg_t reg = ZYDIS_REGISTER_NONE;
       // look for operands with writes to a register...
       for (auto op_idx = 0u; op_idx < itr->instr.operand_count; ++op_idx)