Refactor module loader

Author: serpilliere

example/jitter/run_with_linuxenv.py

@@ -58,6 +58,7 @@
 ld_path = linux_env.filesystem.resolve_path(ld_path)
 cont_ld = Container.from_stream(
     open(ld_path, "rb"),
+    loc_db=loc_db,
     vm=jitter.vm,
     addr=0x80000000,
     apply_reloc=True

example/jitter/unpack_upx.py

@@ -25,7 +25,9 @@ def kernel32_GetProcAddress(jitter):
 
     # Get the generated address of the library, and store it in memory to
     # dst_ad
-    ad = sb.loader.lib_get_add_func(args.libbase, fname, dst_ad)
+    #ad = sb.loader.lib_get_add_func(args.libbase, fname, dst_ad)
+    name = sb.loader.module_base_address_to_name[args.libbase]
+    ad = sb.loader.resolve_function(name, fname, dst_ad=dst_ad)
     # Add a breakpoint in case of a call on the resolved function
     # NOTE: never happens in UPX, just for skeleton
     jitter.handle_function(ad)

miasm/analysis/dse.py

@@ -257,14 +257,14 @@ def add_lib_handler(self, loader, namespace):
 
         # lambda cannot contain statement
         def default_func(dse):
-            fname = loader.fad2cname[dse.jitter.pc]
+            fname = loader.module_base_address_to_name(dse.jitter.pc)
             if isinstance(fname, tuple):
                 fname = b"%s_%d_symb" % (force_bytes(fname[0]), fname[1])
             else:
                 fname = b"%s_symb" % force_bytes(fname)
             raise RuntimeError("Symbolic stub '%s' not found" % fname)
 
-        for addr, fname in viewitems(loader.fad2cname):
+        for addr, fname in viewitems(loader.function_address_to_canonical_name):
             if isinstance(fname, tuple):
                 fname = b"%s_%d_symb" % (force_bytes(fname[0]), fname[1])
             else:

miasm/analysis/sandbox.py

@@ -232,7 +232,7 @@ def init_loader(self, options):
             loader_start_address = int(options.loader_start_address, 0)
         else:
             loader_start_address = None
-        loader = LoaderWindows(loader_start_address=loader_start_address)
+        loader = LoaderWindows(self.jitter.vm, loader_start_address=loader_start_address)
         self.loader = loader
         winobjs.loader = loader
 
@@ -242,60 +242,19 @@ def use_windows_structs(self):
         win_api_x86_32_seh.main_pe_name = self.fname_basename
         win_api_x86_32_seh.main_pe = self.pe
         win_api_x86_32.winobjs.hcurmodule = self.pe.NThdr.ImageBase
-        win_api_x86_32_seh.name2module = self.name2module
+        win_api_x86_32_seh.name2module = self.loader.module_name_to_module
         win_api_x86_32_seh.set_win_fs_0(self.jitter)
         win_api_x86_32_seh.init_seh(self.jitter)
 
     def load_main_pe(self, options):
         from miasm.jitter.loader.pe import vm_load_pe
         from miasm.os_dep.win_api_x86_32 import winobjs
 
-        self.name2module = {}
-
-        # Load main pe
-        with open(options.filename, "rb") as fstream:
-            self.pe = vm_load_pe(
-                self.jitter.vm,
-                fstream.read(),
-                load_hdr=options.load_hdr,
-                name=options.filename,
-                winobjs=winobjs,
-            )
-            self.name2module[self.fname_basename] = self.pe
+        module_image_base = self.loader.load_module(options.filename)
+        name = self.loader.module_base_address_to_name[module_image_base]
+        self.pe = self.loader.module_name_to_module[name]
         winobjs.current_pe = self.pe
 
-    def load_base_dll(self):
-        from miasm.os_dep.win_api_x86_32 import winobjs
-        from miasm.jitter.loader.pe import vm_load_pe_libs, fix_pe_imports
-
-        # Load libs in memory
-        self.name2module.update(
-            vm_load_pe_libs(
-                self.jitter.vm,
-                self.LOADED_DLLS,
-                self.loader,
-                self.PATH_DLLS,
-                winobjs=winobjs,
-            )
-        )
-
-        # Patch libs imports
-        for name, pe in self.name2module.items():
-            fix_pe_imports(self.jitter.vm, pe, self.loader, pe_name=name)
-
-    def load_dependencies(self):
-        from miasm.os_dep.win_api_x86_32 import winobjs
-        from miasm.jitter.loader.pe import vm_load_pe_and_dependencies
-
-        vm_load_pe_and_dependencies(
-            self.jitter.vm,
-            self.fname_basename,
-            self.name2module,
-            self.loader,
-            self.PATH_DLLS,
-            winobjs=winobjs,
-        )
-
     def set_call_handler(self, custom_methods):
         # Library calls handler
         from miasm.os_dep import win_api_x86_32
@@ -306,27 +265,14 @@ def set_call_handler(self, custom_methods):
         methods.update(custom_methods)
         self.jitter.add_lib_handler(self.loader, methods)
 
-    def fix_pe_imports(self):
-        # Fix pe imports
-        from miasm.jitter.loader.pe import fix_pe_imports
-
-        fix_pe_imports(
-            self.jitter.vm, self.pe, self.loader, pe_name=self.fname_basename
-        )
-
     def __init__(self, jitter, options, custom_methods=None):
         self.fname_basename = os.path.basename(options.filename).lower()
         self.jitter = jitter
 
         self.init_stack()
         self.init_loader(options)
         self.load_main_pe(options)
-        if options.loadbasedll:
-            self.load_base_dll()
-        if options.dependencies:
-            self.load_dependencies()
 
-        self.fix_pe_imports()
         self.set_call_handler(custom_methods)
 
         # Manage SEH
@@ -419,7 +365,7 @@ def init_loader(self, options):
             loader_start_address = int(options.loader_start_address, 0)
         else:
             loader_start_address = None
-        loader = LoaderWindows(apiset=apiset, loader_start_address=loader_start_address)
+        loader = LoaderWindows(self.jitter.vm, apiset=apiset, loader_start_address=loader_start_address)
         self.loader = loader
         winobjs.loader = loader
 
@@ -503,7 +449,7 @@ def __init__(self, jitter, options, custom_methods=None):
         self.jitter.init_stack()
 
         # Import manager
-        self.loader = LoaderUnix()
+        self.loader = LoaderUnix(self.jitter.vm)
 
         with open(options.filename, "rb") as fstream:
             self.elf = vm_load_elf(
@@ -562,7 +508,7 @@ def __init__(self, jitter, options, custom_methods=None):
         self.jitter.init_stack()
 
         # Import manager
-        self.loader = LoaderUnix()
+        self.loader = LoaderUnix(self.jitter.vm)
 
         data = open(options.filename, "rb").read()
         options.load_base_addr = int(options.load_base_addr, 0)

miasm/jitter/jitload.py

@@ -491,7 +491,7 @@ def handle_lib(jitter):
         """Resolve the name of the function which cause the handler call. Then
         call the corresponding handler from users callback.
         """
-        fname = jitter.libs.fad2cname[jitter.pc]
+        fname = jitter.loader.function_address_to_canonical_name[jitter.pc]
         if fname in jitter.user_globals:
             func = jitter.user_globals[fname]
         else:
@@ -510,21 +510,21 @@ def handle_function(self, f_addr):
         """Add a breakpoint which will trigger the function handler"""
         self.add_breakpoint(f_addr, self.handle_lib)
 
-    def add_lib_handler(self, libs, user_globals=None):
-        """Add a function to handle libs call with breakpoints
-        @libs: libimp instance
+    def add_lib_handler(self, loader, user_globals=None):
+        """Add a function to handle loader call with breakpoints
+        @loader: Loader instance
         @user_globals: dictionary for defined user function
         """
         if user_globals is None:
             user_globals = {}
 
-        self.libs = libs
+        self.loader = loader
         out = {}
         for name, func in viewitems(user_globals):
             out[name] = func
         self.user_globals = out
 
-        for f_addr in libs.fad2cname:
+        for f_addr in loader.function_address_to_canonical_name:
             self.handle_function(f_addr)
 
     def eval_expr(self, expr):

miasm/jitter/loader/elf.py

@@ -43,7 +43,7 @@ def preload_elf(vm, e, loader, patch_vm_imp=True, loc_db=None):
             continue
         for ad in ads:
             ad_base_lib = loader.lib_get_add_base(libname)
-            ad_libfunc = loader.lib_get_add_func(ad_base_lib, libfunc, ad)
+            ad_libfunc = loader.resolve_function(vm, ad_base_lib, libfunc, ad)
 
             libname_s = canon_libname_libfunc(libname, libfunc)
             dyn_funcs[libname_s] = ad_libfunc
@@ -317,7 +317,17 @@ def vm_load_elf(vm, fdata, name="", base_addr=0, loc_db=None, apply_reloc=False,
 
 
 class LoaderUnix(Loader):
-    pass
+
+    def lib_get_add_base(self, name):
+        name = name.lower().strip(' ')
+        if name in self.module_name_to_base_address:
+            ad = self.module_name_to_base_address[name]
+        else:
+            ad = self.fake_library_entry(name)
+        return ad
+
+    def resolve_function(self, vm, libad, imp_ord_or_name, dst_ad=None):
+        return self.fake_resolve_function(libad, imp_ord_or_name, dst_ad=dst_ad)
 
 
 class libimp_elf(LoaderUnix):