starting to work on decomposition functions

CMakeLists.txt

@@ -55,11 +55,17 @@ set(Theodosius_SOURCES "")
 
 list(APPEND Theodosius_SOURCES
 	"include/comp/comp.hpp"
+	"include/comp/reloc.hpp"
 	"include/comp/symbol_table.hpp"
 	"include/decomp/decomp.hpp"
 	"include/decomp/routine.hpp"
 	"include/decomp/symbol.hpp"
 	"include/theo.hpp"
+	"src/theo/comp/comp.cpp"
+	"src/theo/comp/symbol_table.cpp"
+	"src/theo/decomp/decomp.cpp"
+	"src/theo/decomp/routine.cpp"
+	"src/theo/decomp/symbol.cpp"
 	"src/theo/theo.cpp"
 )
 

include/comp/comp.hpp

@@ -17,6 +17,7 @@ class comp_t {
                   copier_t copy,
                   resolver_t resolve);
 
+  std::optional<std::uint32_t> compose();
   void allocator(allocator_t alloc);
   void copier(copier_t copy);
   void resolver(resolver_t resolve);

include/comp/reloc.hpp

@@ -0,0 +1,18 @@
+#pragma once
+#include <cstddef>
+#include <cstdint>
+
+namespace theo::comp {
+class reloc_t {
+ public:
+  explicit reloc_t(std::uint16_t offset, std::size_t hash)
+      : m_offset(offset), m_hash(hash) {}
+
+  std::size_t hash() { return m_hash; }
+  std::uint16_t offset() { return m_offset; }
+
+ private:
+  std::size_t m_hash;
+  std::uint16_t m_offset;
+};
+}  // namespace theo::comp

include/comp/symbol_table.hpp

@@ -1,4 +1,5 @@
 #pragma once
+#include <algorithm>
 #include <functional>
 #include <map>
 #include <vector>
@@ -8,19 +9,22 @@
 namespace theo::comp {
 class symbol_table_t {
  public:
-  symbol_table_t();
+  symbol_table_t() {}
   symbol_table_t(const std::vector<decomp::symbol_t>&& syms);
 
   void add_symbol(decomp::symbol_t& sym);
   void add_symbols(std::vector<decomp::symbol_t>& syms);
 
-  void update(std::string& name, decomp::symbol_t& sym);
-  void update(std::string& name, std::uintptr_t location);
+  decomp::symbol_t sym_from_hash(std::size_t hash);
+  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);
 
   void for_each(std::function<bool(decomp::symbol_t& sym)> fn);
   std::uint32_t size();
 
  private:
-  std::map<std::string, decomp::symbol_t> m_table;
+  std::map<std::size_t, decomp::symbol_t> m_table;
 };
 }  // namespace theo::comp

include/decomp/decomp.hpp

@@ -1,28 +1,30 @@
 #pragma once
-#include <coff/image.hpp>
 #include <cstdint>
 #include <linuxpe>
-#include <vector>
 #include <optional>
+#include <vector>
+#include <spdlog/spdlog.h>
 
 #include <comp/symbol_table.hpp>
 #include <decomp/routine.hpp>
 
+#include <coff/archive.hpp>
+#include <coff/image.hpp>
+
 namespace theo::decomp {
 class decomp_t {
  public:
-  explicit decomp_t(std::vector<std::uint8_t>& lib_data,
-                    comp::symbol_table_t* syms);
+  explicit decomp_t(std::vector<std::uint8_t>& lib, comp::symbol_table_t* syms);
 
   std::vector<routine_t> rtns();
   std::vector<std::uint8_t> lib();
-  std::vector<coff::image_t> objs();
+  std::vector<coff::image_t*> objs();
   comp::symbol_table_t* syms();
   std::optional<comp::symbol_table_t*> decompose();
 
  private:
-  std::vector<std::uint8_t> m_lib_data;
-  std::vector<coff::image_t> m_obj_imgs;
+  const std::vector<std::uint8_t> m_lib;
+  std::vector<coff::image_t*> m_objs;
   std::vector<routine_t> m_rtns;
   comp::symbol_table_t* m_syms;
 };

include/decomp/symbol.hpp

@@ -1,29 +1,35 @@
 #pragma once
 #include <coff/image.hpp>
+#include <comp/reloc.hpp>
 #include <cstdint>
 #include <string>
+#include <vector>
 
 namespace theo::decomp {
 class symbol_t {
  public:
-  explicit symbol_t(const std::string&& name,
-                    std::uintptr_t location,
-                    const std::vector<std::uint8_t>&& data,
-                    coff::section_header_t scn_hdr);
+  explicit symbol_t(std::string name,
+                    std::uintptr_t offset,
+                    std::vector<std::uint8_t> data,
+                    coff::section_header_t scn_hdr,
+                    std::vector<comp::reloc_t> relocs);
 
   std::string name() const;
-  std::uintptr_t location() const;
+  std::uintptr_t offset() const;
+  std::uintptr_t allocated_at() const;
   std::uint32_t size() const;
   std::vector<std::uint8_t> data() const;
 
-  void name(const std::string&& name);
-  void location(std::uintptr_t location);
-  void data(const std::vector<std::uint8_t>&& data);
+  void allocated_at(std::uintptr_t allocated_at);
+
+  std::size_t hash();
+  static std::size_t hash(const std::string& sym);
 
  private:
   std::string m_name;
-  std::uintptr_t m_location;
+  std::uintptr_t m_offset, m_allocated_at;
   std::vector<std::uint8_t> m_data;
-  const coff::section_header_t m_scn_hdr;
+  coff::section_header_t m_scn_hdr;
+  std::vector<comp::reloc_t> m_relocs;
 };
 }  // namespace theo::decomp

src/tests/demo/main.cpp

@@ -8,9 +8,6 @@
 namespace fs = std::filesystem;
 
 int main(int argc, char* argv[]) {
-  spdlog::info("this is info...");
-  spdlog::error("this is an error...");
-
   if (argc < 2)
     return -1;
 

src/theo/comp/comp.cpp

@@ -0,0 +1,28 @@
+#include <comp/comp.hpp>
+
+namespace theo::comp {
+comp_t::comp_t(decomp::decomp_t* dcmp) : m_dcmp(dcmp) {}
+comp_t::comp_t(decomp::decomp_t* dcmp,
+               allocator_t alloc,
+               copier_t copy,
+               resolver_t resolve)
+    : m_dcmp(dcmp), m_allocator(alloc), m_copier(copy), m_resolver(resolve) {}
+
+void comp_t::allocator(allocator_t alloc) {
+  m_allocator = alloc;
+}
+
+void comp_t::copier(copier_t copy) {
+  m_copier = copy;
+}
+
+void comp_t::resolver(resolver_t resolve) {
+  m_resolver = resolve;
+}
+
+std::uintptr_t comp_t::resolve(std::string&& sym) {
+  return m_dcmp->syms()
+      ->sym_from_hash(decomp::symbol_t::hash(sym))
+      .allocated_at();
+}
+}  // namespace theo::comp

src/theo/comp/symbol_table.cpp

@@ -0,0 +1,48 @@
+#include <comp/symbol_table.hpp>
+
+namespace theo::comp {
+symbol_table_t::symbol_table_t(const std::vector<decomp::symbol_t>&& syms) {
+  std::for_each(syms.begin(), syms.end(), [&](decomp::symbol_t sym) {
+    m_table.insert({sym.hash(), sym});
+  });
+}
+
+void symbol_table_t::add_symbol(decomp::symbol_t& sym) {
+  m_table.insert({sym.hash(), sym});
+}
+
+void symbol_table_t::add_symbols(std::vector<decomp::symbol_t>& syms) {
+  std::for_each(syms.begin(), syms.end(),
+                [&](decomp::symbol_t sym) { add_symbol(sym); });
+}
+
+void symbol_table_t::update(std::size_t hash, decomp::symbol_t& sym) {
+  m_table.insert({hash, sym});
+}
+
+void symbol_table_t::update(std::size_t hash, std::uintptr_t allocated_at) {
+  auto v = m_table.at(hash);
+  v.allocated_at(allocated_at);
+  m_table.insert({hash, v});
+}
+
+void symbol_table_t::for_each(std::function<bool(decomp::symbol_t& sym)> fn) {
+  std::for_each(m_table.begin(), m_table.end(), [&](auto v) { fn(v.second); });
+}
+
+decomp::symbol_t symbol_table_t::sym_from_hash(std::size_t hash) {
+  return m_table.at(hash);
+}
+
+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::pair<const std::size_t, decomp::symbol_t> itr) {
+                        return itr.second.allocated_at() == allocated_at;
+                      })
+      ->second;
+}
+
+std::uint32_t symbol_table_t::size() {
+  return m_table.size();
+}
+}  // namespace theo::comp

src/theo/decomp/decomp.cpp

@@ -0,0 +1,37 @@
+#include <decomp/decomp.hpp>
+
+namespace theo::decomp {
+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() {
+  ar::view<false> lib(m_lib.data(), m_lib.size());
+  std::for_each(
+      lib.begin(), lib.end(),
+      [&](std::pair<std::string_view, ar::entry_t&> itr) {
+        // if the entry isnt the symbol table or the string table
+        // then we know its an obj file...
+        //
+        if (!itr.second.is_symbol_table() && !itr.second.is_string_table()) {
+          spdlog::info("extracted obj from archive: {}", itr.first);
+          m_objs.push_back(reinterpret_cast<coff::image_t*>(itr.second.data()));
+        }
+      });
+}
+
+std::vector<routine_t> decomp_t::rtns() {
+  return m_rtns;
+}
+
+std::vector<std::uint8_t> decomp_t::lib() {
+  return m_lib;
+}
+
+std::vector<coff::image_t*> decomp_t::objs() {
+  return m_objs;
+}
+
+comp::symbol_table_t* decomp_t::syms() {
+  return m_syms;
+}
+}  // namespace theo::decomp

src/theo/decomp/symbol.cpp

@@ -0,0 +1,48 @@
+#include <decomp/symbol.hpp>
+
+namespace theo::decomp {
+symbol_t::symbol_t(std::string name,
+                   std::uintptr_t offset,
+                   std::vector<std::uint8_t> data,
+                   coff::section_header_t scn_hdr,
+                   std::vector<comp::reloc_t> relocs)
+    : m_name(name),
+      m_offset(offset),
+      m_data(data),
+      m_scn_hdr(scn_hdr),
+      m_relocs(relocs),
+      m_allocated_at(0) {}
+
+std::string symbol_t::name() const {
+  return m_name;
+}
+
+std::uintptr_t symbol_t::offset() const {
+  return m_offset;
+}
+
+std::uintptr_t symbol_t::allocated_at() const {
+  return m_allocated_at;
+}
+
+std::uint32_t symbol_t::size() const {
+  return m_data.size();
+}
+
+std::vector<std::uint8_t> symbol_t::data() const {
+  return m_data;
+}
+
+void symbol_t::allocated_at(std::uintptr_t allocated_at) {
+  m_allocated_at = allocated_at;
+}
+
+std::size_t symbol_t::hash() {
+  return hash(m_name);
+}
+
+std::size_t symbol_t::hash(const std::string& sym) 
+{
+  return std::hash<std::string>{}(sym);
+}
+}  // namespace theo::decomp