demo is working with mapping entire functions... still need to make

relocation transformations and such...

include/comp/comp.hpp

@@ -17,7 +17,10 @@ class comp_t {
                   copier_t copy,
                   resolver_t resolve);
 
-  void compose();
+  void allocate();
+  void resolve();
+  void copy_syms();
+
   void allocator(allocator_t alloc);
   void copier(copier_t copy);
   void resolver(resolver_t resolve);

include/comp/reloc.hpp

@@ -5,13 +5,17 @@
 namespace theo::comp {
 class reloc_t {
  public:
-  explicit reloc_t(std::uint16_t offset, std::size_t hash)
-      : m_offset(offset), m_hash(hash) {}
+  explicit reloc_t(std::uint16_t offset,
+                   std::size_t hash,
+                   const std::string&& sym_name)
+      : m_offset(offset), m_hash(hash), m_sym_name(sym_name) {}
 
   std::size_t hash() { return m_hash; }
+  std::string name() { return m_sym_name; }
   std::uint16_t offset() { return m_offset; }
 
  private:
+  std::string m_sym_name;
   std::size_t m_hash;
   std::uint16_t m_offset;
 };

include/comp/symbol_table.hpp

@@ -3,6 +3,7 @@
 #include <functional>
 #include <map>
 #include <vector>
+#include <optional>
 
 #include <decomp/symbol.hpp>
 
@@ -15,8 +16,8 @@ class symbol_table_t {
   void add_symbol(decomp::symbol_t& sym);
   void add_symbols(std::vector<decomp::symbol_t>& syms);
 
-  decomp::symbol_t sym_from_hash(std::size_t hash);
-  decomp::symbol_t sym_from_alloc(std::uintptr_t allocated_at);
+  std::optional<decomp::symbol_t> sym_from_hash(std::size_t hash);
+  std::optional<decomp::symbol_t> sym_from_alloc(std::uintptr_t allocated_at);
 
   void update(std::size_t hash, decomp::symbol_t& sym);
   void update(std::size_t hash, std::uintptr_t allocated_at);

include/decomp/decomp.hpp

@@ -1,9 +1,9 @@
 #pragma once
+#include <spdlog/spdlog.h>
 #include <cstdint>
 #include <linuxpe>
 #include <optional>
 #include <vector>
-#include <spdlog/spdlog.h>
 
 #include <comp/symbol_table.hpp>
 #include <decomp/routine.hpp>
@@ -20,12 +20,14 @@ class decomp_t {
   std::vector<std::uint8_t> lib();
   std::vector<coff::image_t*> objs();
   comp::symbol_table_t* syms();
+  std::map<coff::section_header_t*, std::size_t>& scn_hash_tbl();
   std::optional<comp::symbol_table_t*> decompose();
 
  private:
   const std::vector<std::uint8_t> m_lib;
   std::vector<coff::image_t*> m_objs;
   std::vector<routine_t> m_rtns;
+  std::map<coff::section_header_t*, std::size_t> m_scn_hash_tbl;
   comp::symbol_table_t* m_syms;
 };
 }  // namespace theo::decomp

include/decomp/routine.hpp

@@ -21,7 +21,7 @@ class routine_t {
                      coff::image_t* img,
                      coff::section_header_t* scn,
                      std::vector<std::uint8_t>& fn,
-                     decomp_type_t dcmp_type);
+                     sym_type_t dcmp_type);
 
   std::vector<decomp::symbol_t> decompose();
   coff::section_header_t* scn();
@@ -32,6 +32,6 @@ class routine_t {
   std::vector<std::uint8_t> m_data;
   coff::image_t* m_img;
   coff::section_header_t* m_scn;
-  decomp_type_t m_dcmp_type;
+  sym_type_t m_dcmp_type;
 };
 }  // namespace theo::decomp

include/decomp/symbol.hpp

@@ -6,33 +6,31 @@
 #include <vector>
 
 namespace theo::decomp {
-enum decomp_type_t { none, instr_split };
+enum sym_type_t { function, inst_split, data, section };
 
 class symbol_t {
  public:
   explicit symbol_t(std::string name,
                     std::uintptr_t offset,
                     std::vector<std::uint8_t> data,
-                    coff::section_header_t* scn,
-                    coff::symbol_t* sym,
-                    std::vector<comp::reloc_t> relocs,
-                    decomp_type_t dcmp_type);
+                    coff::section_header_t* scn = {},
+                    coff::symbol_t* sym = {},
+                    std::vector<comp::reloc_t> relocs = {},
+                    sym_type_t dcmp_type = {});
 
   std::string name() const;
   std::uintptr_t offset() const;
   std::uintptr_t allocated_at() const;
   std::uint32_t size() const;
   coff::section_header_t* scn() const;
-  std::vector<std::uint8_t> data() const;
+  std::vector<std::uint8_t>& data();
   coff::symbol_t* sym() const;
-  decomp_type_t dcmp_type() const;
+  sym_type_t sym_type() const;
   std::vector<comp::reloc_t>& relocs();
   void allocated_at(std::uintptr_t allocated_at);
 
   std::size_t hash();
   static std::size_t hash(const std::string& sym);
-  static std::size_t scn_hash(coff::symbol_t* sym, coff::section_header_t* scn,
-                              coff::image_t* img);
 
  private:
   std::string m_name;
@@ -40,7 +38,7 @@ class symbol_t {
   std::vector<std::uint8_t> m_data;
   coff::section_header_t* m_scn;
   std::vector<comp::reloc_t> m_relocs;
-  decomp_type_t m_dcmp_type;
+  sym_type_t m_sym_type;
   coff::symbol_t* m_sym;
 };
 }  // namespace theo::decomp

src/tests/demo/main.cpp

@@ -47,4 +47,5 @@ int main(int argc, char* argv[]) {
   spdlog::info("decomposed {} symbols...", res.value());
   auto entry_pnt = t.compose("EntryPoint");
   spdlog::info("entry point address: {:X}", entry_pnt);
+  reinterpret_cast<void (*)()>(entry_pnt)();
 }

src/tests/demolib/main.cpp

@@ -1,9 +1,9 @@
-extern "C" int MessageBox(void* hWnd,
+extern "C" int MessageBoxA(void* hWnd,
                           char* lpText,
                           char* lpCaption,
                           void* uType);
 
-__declspec(code_seg(".obf")) extern "C" void EntryPoint() {
-  MessageBox(nullptr, "Hello World", "Hello World", nullptr);
-  MessageBox(nullptr, "Hello World 1", "Hello World 1", nullptr);
+extern "C" void EntryPoint() {
+  MessageBoxA(nullptr, "Hello World", "Hello World", nullptr);
+  MessageBoxA(nullptr, "Hello World 1", "Hello World 1", nullptr);
 }

src/theo/comp/comp.cpp

@@ -8,35 +8,103 @@ comp_t::comp_t(decomp::decomp_t* dcmp,
                resolver_t resolve)
     : m_dcmp(dcmp), m_allocator(alloc), m_copier(copy), m_resolver(resolve) {}
 
-void comp_t::compose() {
+void comp_t::allocate() {
+  // map code & data/rdata/bss sections first...
+  //
   m_dcmp->syms()->for_each([&](theo::decomp::symbol_t& sym) {
-    // if this symbol is a function..
+    switch (sym.sym_type()) {
+      case decomp::sym_type_t::section:
+      case decomp::sym_type_t::function: {
+        sym.allocated_at(m_allocator(sym.size(), sym.scn()->characteristics));
+        break;
+      }
+      case decomp::sym_type_t::inst_split: {
+        // TODO: call into reloc_t static methods to generate random code to
+        // obfuscate relocations...
         //
-    if (sym.scn()->characteristics.mem_execute) {
-      // if comp type is none (meaning entire function) then we just allocate
-      // it...
+        break;
+      }
+      default:
+        break;
+    }
+  });
+
+  // then map data/rdata/bss symbols to the allocated sections...
   //
-      if (sym.dcmp_type() == decomp::decomp_type_t::none) {
-        sym.allocated_at(m_allocator(sym.size(), sym.scn()->characteristics));
-        spdlog::info("allocated entire function: {} at address: {:X} size: {}",
-                     sym.name(), sym.allocated_at(), sym.size());
-      } else {  // else the dcmp_type is split instrs we are going to apply
-                // transformations...
+  m_dcmp->syms()->for_each([&](theo::decomp::symbol_t& sym) {
+    if (sym.sym_type() == decomp::sym_type_t::data) {
+      auto scn_sym =
+          m_dcmp->syms()->sym_from_hash(m_dcmp->scn_hash_tbl()[sym.scn()]);
 
-        sym.allocated_at(m_allocator(sym.size(), sym.scn()->characteristics));
+      if (!scn_sym.has_value()) {
+        spdlog::error("failed to locate section: {} for symbol: {}",
+                      sym.scn()->name.to_string(), sym.name());
 
-        spdlog::info(
-            "transformed relocations of symbol: {} size is now: {} allocated "
-            "at: {:X}",
-            sym.name(), sym.size(), sym.allocated_at());
+        assert(scn_sym.has_value());
       }
-    } else {  // else the allocation is data which means we dont have any
-              // relocs...
-      sym.allocated_at(m_allocator(sym.size(), sym.scn()->characteristics));
+
+      sym.allocated_at(scn_sym.value().allocated_at() + sym.offset());
     }
   });
 }
 
+void comp_t::resolve() {
+  // resolve relocations in all symbols...
+  //
+  m_dcmp->syms()->for_each([&](theo::decomp::symbol_t& sym) {
+    std::for_each(sym.relocs().begin(), sym.relocs().end(), [&](reloc_t reloc) {
+      if (reloc.offset() > sym.data().size()) {
+        spdlog::error(
+            "invalid relocation... writing outside of symbol length... offset: "
+            "{} sym size: {}",
+            sym.offset(), sym.data().size());
+
+        assert(reloc.offset() > sym.data().size());
+      }
+
+      // try and resolve the symbol by refering to the internal symbol table
+      // first... if there is no symbol then refer to the resolver...
+      //
+      auto reloc_sym = m_dcmp->syms()->sym_from_hash(reloc.hash());
+      auto allocated_at = reloc_sym.has_value()
+                              ? reloc_sym.value().allocated_at()
+                              : m_resolver(reloc.name());
+
+      if (!allocated_at) {
+        spdlog::error("failed to resolve reloc from symbol: {} to symbol: {}",
+                      sym.name(), reloc.name());
+
+        assert(allocated_at);
+      }
+
+      switch (sym.sym_type()) {
+        case decomp::sym_type_t::function: {
+          *reinterpret_cast<std::uintptr_t*>(sym.data().data() +
+                                             reloc.offset()) = allocated_at;
+          break;
+        }
+        case decomp::sym_type_t::inst_split: {
+          // TODO: run the vector of transformation operations here if the
+          // symbol is of type inst_split... the transformations will be applied
+          // to allocate_at() result on the symbol...
+          //
+          break;
+        }
+        default:
+          break;
+      }
+    });
+  });
+}
+
+void comp_t::copy_syms() {
+  // copy symbols into memory using the copier supplied...
+  //
+  m_dcmp->syms()->for_each([&](theo::decomp::symbol_t& sym) {
+    m_copier(sym.allocated_at(), sym.data().data(), sym.data().size());
+  });
+}
+
 void comp_t::allocator(allocator_t alloc) {
   m_allocator = alloc;
 }
@@ -50,8 +118,7 @@ void comp_t::resolver(resolver_t resolve) {
 }
 
 std::uintptr_t comp_t::resolve(const std::string&& sym) {
-  return m_dcmp->syms()
-      ->sym_from_hash(decomp::symbol_t::hash(sym))
-      .allocated_at();
+  auto res = m_dcmp->syms()->sym_from_hash(decomp::symbol_t::hash(sym));
+  return res.has_value() ? res->allocated_at() : 0;
 }
 }  // namespace theo::comp

src/theo/comp/symbol_table.cpp

@@ -31,16 +31,21 @@ void symbol_table_t::for_each(std::function<void(decomp::symbol_t& sym)> fn) {
     fn(itr->second);
 }
 
-decomp::symbol_t symbol_table_t::sym_from_hash(std::size_t hash) {
-  return m_table.at(hash);
+std::optional<decomp::symbol_t> symbol_table_t::sym_from_hash(
+    std::size_t hash) {
+  return m_table.count(hash) ? m_table.at(hash)
+                             : std::optional<decomp::symbol_t>{};
 }
 
-decomp::symbol_t symbol_table_t::sym_from_alloc(std::uintptr_t allocated_at) {
-  return std::find_if(m_table.begin(), m_table.end(),
+std::optional<decomp::symbol_t> symbol_table_t::sym_from_alloc(
+    std::uintptr_t allocated_at) {
+  auto res =
+      std::find_if(m_table.begin(), m_table.end(),
                    [&](std::pair<const std::size_t, decomp::symbol_t> itr) {
                      return itr.second.allocated_at() == allocated_at;
-                      })
-      ->second;
+                   });
+
+  return res != m_table.end() ? res->second : std::optional<decomp::symbol_t>{};
 }
 
 std::uint32_t symbol_table_t::size() {

src/theo/decomp/decomp.cpp

@@ -5,6 +5,8 @@ decomp_t::decomp_t(std::vector<std::uint8_t>& lib, comp::symbol_table_t* syms)
     : m_lib(lib), m_syms(syms) {}
 
 std::optional<comp::symbol_table_t*> decomp_t::decompose() {
+  // extract obj files from the archive file...
+  //
   ar::view<false> lib(m_lib.data(), m_lib.size());
   std::for_each(
       lib.begin(), lib.end(),
@@ -18,7 +20,27 @@ std::optional<comp::symbol_table_t*> decomp_t::decompose() {
         }
       });
 
+  // generate symbols, populate section hash table, for each object file
+  // extracted from the archive file...
+  //
   std::for_each(m_objs.begin(), m_objs.end(), [&](coff::image_t* img) {
+    // populate section hash table...
+    //
+    for (auto idx = 0u; idx < img->file_header.num_sections; ++idx) {
+      auto scn = img->get_section(idx);
+      auto scn_sym_name =
+          std::string(scn->name.to_string(img->get_strings()))
+              .append("#")
+              .append(std::to_string(idx))
+              .append("!")
+              .append(std::to_string(img->file_header.timedate_stamp));
+
+      // hash the name of the section + the index + thhe timestamp of the obj
+      // file it is in...
+      //
+      m_scn_hash_tbl.insert({scn, decomp::symbol_t::hash(scn_sym_name)});
+    }
+
     auto syms_cnt = img->file_header.num_symbols;
     for (auto idx = 0u; idx < syms_cnt; ++idx) {
       auto sym = img->get_symbol(idx);
@@ -31,8 +53,8 @@ std::optional<comp::symbol_table_t*> decomp_t::decompose() {
         auto scn = img->get_section(sym->section_index - 1);
         auto dcmp_type =
             scn->name.to_string(img->get_strings()) == INSTR_SPLIT_SECTION_NAME
-                ? decomp::decomp_type_t::instr_split
-                : decomp::decomp_type_t::none;
+                ? decomp::sym_type_t::inst_split
+                : decomp::sym_type_t::function;
         auto fn_size = scn->size_raw_data;
         auto fn_bgn = scn->ptr_raw_data + reinterpret_cast<std::uint8_t*>(img);
 
@@ -48,14 +70,54 @@ std::optional<comp::symbol_table_t*> decomp_t::decompose() {
       } else if (sym->has_section() &&
                  sym->storage_class == coff::storage_class_id::public_symbol) {
         auto scn = img->get_section(sym->section_index - 1);
-        spdlog::info("{} allocated in section: {} with size: {}",
-                     sym->name.to_string(img->get_strings()),
-                     scn->name.to_string(img->get_strings()),
+        auto scn_sym = m_syms->sym_from_hash(m_scn_hash_tbl[scn]);
+
+        // if the section doesnt have a symbol then make one and put it into the
+        // symbol table...
+        //
+        if (!scn_sym.has_value()) {
+          auto scn_sym_name =
+              std::string(scn->name.to_string(img->get_strings()))
+                  .append("#")
+                  .append(std::to_string(sym->section_index - 1))
+                  .append("!")
+                  .append(std::to_string(img->file_header.timedate_stamp));
+
+          std::vector<std::uint8_t> scn_data(
+              reinterpret_cast<std::uint8_t*>(img) + scn->ptr_raw_data,
+              reinterpret_cast<std::uint8_t*>(img) + scn->ptr_raw_data +
                   scn->size_raw_data);
+
+          decomp::symbol_t new_scn_sym(scn_sym_name, 0, scn_data, scn, {}, {},
+                                       sym_type_t::section);
+
+          spdlog::info(
+              "generating symbol for section: {} sym name: {} hash: {:X}",
+              scn->name.to_string(img->get_strings()), new_scn_sym.name(),
+              new_scn_sym.hash());
+
+          m_syms->add_symbol(new_scn_sym);
+        }
+
+        // create a symbol for the data...
+        //
+        decomp::symbol_t new_sym(sym->name.to_string(img->get_strings()).data(),
+                                 sym->value, {}, scn, sym, {},
+                                 sym_type_t::data);
+
+        spdlog::info("adding data symbol: {} located inside of section: {}",
+                     new_sym.name(),
+                     m_syms->sym_from_hash(m_scn_hash_tbl[new_sym.scn()])
+                         .value()
+                         .name());
+
+        m_syms->add_symbol(new_sym);
       }
     }
   });
 
+  // return the extract symbols to the caller...
+  //
   return m_syms;
 }
 
@@ -74,4 +136,8 @@ std::vector<coff::image_t*> decomp_t::objs() {
 comp::symbol_table_t* decomp_t::syms() {
   return m_syms;
 }
+
+std::map<coff::section_header_t*, std::size_t>& decomp_t::scn_hash_tbl() {
+  return m_scn_hash_tbl;
+}
 }  // namespace theo::decomp

src/theo/decomp/routine.cpp

@@ -5,14 +5,14 @@ routine_t::routine_t(coff::symbol_t* sym,
                      coff::image_t* img,
                      coff::section_header_t* scn,
                      std::vector<std::uint8_t>& fn,
-                     decomp_type_t dcmp_type)
+                     sym_type_t dcmp_type)
     : m_img(img), m_scn(scn), m_data(fn), m_dcmp_type(dcmp_type), m_sym(sym) {}
 
 std::vector<decomp::symbol_t> routine_t::decompose() {
   std::vector<decomp::symbol_t> result;
 
   switch (m_dcmp_type) {
-    case none: {
+    case function: {
       std::vector<comp::reloc_t> relocs;
       auto scn_relocs = reinterpret_cast<coff::reloc_t*>(
           m_scn->ptr_relocs + reinterpret_cast<std::uint8_t*>(m_img));
@@ -23,19 +23,20 @@ std::vector<decomp::symbol_t> routine_t::decompose() {
         auto sym_name = sym_reloc->name.to_string(m_img->get_strings());
         auto sym_hash = decomp::symbol_t::hash(sym_name.data());
 
-        spdlog::info("{} reloc to: {} at offset: {}",
+        spdlog::info("{} reloc to: {} hash: {:X} at offset: {}",
                      m_sym->name.to_string(m_img->get_strings()), sym_name,
-                     scn_reloc->virtual_address);
+                     sym_hash, scn_reloc->virtual_address);
 
-        relocs.push_back(comp::reloc_t(scn_reloc->virtual_address, sym_hash));
+        relocs.push_back(comp::reloc_t(scn_reloc->virtual_address, sym_hash,
+                                       sym_name.data()));
       }
 
       result.push_back(decomp::symbol_t(
           m_sym->name.to_string(m_img->get_strings()).data(), m_sym->value,
-          m_data, m_scn, m_sym, relocs, decomp_type_t::none));
+          m_data, m_scn, m_sym, relocs, sym_type_t::function));
       break;
     }
-    case instr_split: {
+    case inst_split: {
       std::uint32_t offset = 0u;
       xed_error_enum_t err;
 
@@ -83,7 +84,8 @@ std::vector<decomp::symbol_t> routine_t::decompose() {
           spdlog::info("{} reloc to: {} at offset: {}", new_sym_name, sym_name,
                        reloc_offset);
 
-          relocs.push_back(comp::reloc_t(reloc_offset, sym_hash));
+          relocs.push_back(
+              comp::reloc_t(reloc_offset, sym_hash, sym_name.data()));
         }
 
         // add a reloc to the next instruction...
@@ -96,8 +98,8 @@ std::vector<decomp::symbol_t> routine_t::decompose() {
                 .append(std::to_string(offset +
                                        xed_decoded_inst_get_length(&instr)));
 
-        relocs.push_back(
-            comp::reloc_t(0, decomp::symbol_t::hash(next_inst_sym)));
+        relocs.push_back(comp::reloc_t(0, decomp::symbol_t::hash(next_inst_sym),
+                                       next_inst_sym.data()));
 
         // get the instructions bytes
         //
@@ -107,7 +109,7 @@ std::vector<decomp::symbol_t> routine_t::decompose() {
 
         result.push_back(decomp::symbol_t(new_sym_name, offset, inst_bytes,
                                           m_scn, m_sym, relocs,
-                                          decomp_type_t::instr_split));
+                                          sym_type_t::inst_split));
 
         // after creating the symbol and dealing with relocs then print the
         // information we have concluded...

src/theo/decomp/symbol.cpp

@@ -7,13 +7,13 @@ symbol_t::symbol_t(std::string name,
                    coff::section_header_t* scn,
                    coff::symbol_t* sym,
                    std::vector<comp::reloc_t> relocs,
-                   decomp_type_t dcmp_type)
+                   sym_type_t dcmp_type)
     : m_name(name),
       m_offset(offset),
       m_data(data),
       m_scn(scn),
       m_relocs(relocs),
-      m_dcmp_type(dcmp_type),
+      m_sym_type(dcmp_type),
       m_sym(sym),
       m_allocated_at(0) {}
 
@@ -37,12 +37,12 @@ std::uint32_t symbol_t::size() const {
   return m_data.size();
 }
 
-std::vector<std::uint8_t> symbol_t::data() const {
+std::vector<std::uint8_t>& symbol_t::data() {
   return m_data;
 }
 
-decomp_type_t symbol_t::dcmp_type() const {
-  return m_dcmp_type;
+sym_type_t symbol_t::sym_type() const {
+  return m_sym_type;
 }
 
 void symbol_t::allocated_at(std::uintptr_t allocated_at) {
@@ -64,14 +64,4 @@ std::vector<comp::reloc_t>& symbol_t::relocs() {
 std::size_t symbol_t::hash(const std::string& sym) {
   return std::hash<std::string>{}(sym);
 }
-
-std::size_t symbol_t::scn_hash(coff::symbol_t* sym,
-                               coff::section_header_t* scn,
-                               coff::image_t* img) {
-  return hash(std::string(scn->name.to_string(img->get_strings()))
-                  .append("#")
-                  .append(std::to_string(sym->section_index - 1))
-                  .append("!")
-                  .append(std::to_string(img->file_header.timedate_stamp)));
-}
 }  // namespace theo::decomp

src/theo/theo.cpp

@@ -28,7 +28,9 @@ std::optional<std::uint32_t> theo_t::decompose() {
 }
 
 std::uintptr_t theo_t::compose(const std::string&& entry_sym) {
-  m_cmp.compose();
+  m_cmp.allocate();
+  m_cmp.resolve();
+  m_cmp.copy_syms();
   return m_cmp.resolve(entry_sym.data());
 }
 }  // namespace theo