Fix for 1516: no sibbos on ELF/PPC

Author: Seva Alekseyev

dwex/__main__.py

@@ -19,7 +19,7 @@
 from .treedlg import TreeDlg
 
 # Sync with version in setup.py
-version = (4, 54)
+version = (4, 55)
 the_app = None
 
 # TODO:

dwex/formats.py

@@ -13,6 +13,13 @@ class FormatError(Exception):
     def __init__(self, s):
         Exception.__init__(self, s)
 
+def decorate_di(di, f, a, sa):
+    di._format = f
+    di._arch_code = a
+    di._start_address = sa
+    di._frames = None
+    di._use_siblings = not(f in (0, 4) and a in ("EM_PPC", 'EM_PPC64'))
+
 def read_pe(filename):
     from filebytes.pe import PE, IMAGE_FILE_MACHINE, BinaryError
     import struct, zlib
@@ -76,9 +83,7 @@ def read_section(name, is_compressed, section, virtual_size, raw_size):
             gnu_debugaltlink_sec = data.get('.gnu_debugaltlink'),
             debug_types_sec = data.get('.debug_types')
         )
-        di._format = 2
-        di._arch_code = machine
-        di._start_address = pefile.imageNtHeaders.header.OptionalHeader.ImageBase
+        decorate_di(di, 2, machine, pefile.imageNtHeaders.header.OptionalHeader.ImageBase)
         di._frames = None
         return di
     except BinaryError as err:
@@ -328,14 +333,11 @@ def string_at(off):
         gnu_debugaltlink_sec = data.get('__gnu_debugaltlink'),
         debug_types_sec = data.get('__debug_types'),
     )
+    text_cmd = next((cmd for cmd in macho.loadCommands if cmd.header.cmd in (LC.SEGMENT, LC.SEGMENT_64) and cmd.name == "__TEXT"), False)
+    decorate_di(di, 1, macho_arch_code(macho), text_cmd.header.vmaddr if text_cmd else 0)
     di._unwind_sec = sections.get('__unwind_info') # VERY unlikely to be None
-    di._format = 1 # Will be adjusted later if Mach-O within A within fat Mach-O
-    di._arch_code = macho_arch_code(macho)
     di._slice_code = slice_code
     di._uuid = uuid
-    text_cmd = next((cmd for cmd in macho.loadCommands if cmd.header.cmd in (LC.SEGMENT, LC.SEGMENT_64) and cmd.name == "__TEXT"), False)
-    di._start_address = text_cmd.header.vmaddr if text_cmd else 0
-    di._frames = None
     di._has_exec = False
     return di
 
@@ -495,10 +497,7 @@ def read_wasm(file):
         gnu_debugaltlink_sec = None,
         debug_types_sec = data.get('.debug_types')
     )
-    di._format = 3
-    di._arch_code = None #N/A
-    di._start_address = 0
-    di._frames = None
+    decorate_di(di, 3, None, 0)
     return di
 
 # Filename is only needed for supplemental DWARF resolution
@@ -519,10 +518,7 @@ def read_elf(file, filename):
         di = parse_dwarf1(elffile)
 
     if di:
-        di._format = 0
-        di._start_address = start_address
-        di._arch_code = elffile.header.e_machine
-        di._frames = None
+        decorate_di(di, 0, elffile.header.e_machine, start_address)
     return di
 
 ###########################################################################

dwex/patch.py

@@ -127,6 +127,60 @@ def _init_dispatch_table(structs):
         return dt
     elftools.dwarf.dwarf_expr._init_dispatch_table = _init_dispatch_table
 
+    # Fix for 1516: ignore DW_AT_sibling for child enumeration on ELF/PPC
+    def iter_DIE_children(self, die):
+        """ Given a DIE, yields either its children, without null DIE list
+            terminator, or nothing, if that DIE has no children.
+
+            The null DIE terminator is saved in that DIE when iteration ended.
+        """
+        if not die.has_children:
+            return
+
+        # `cur_offset` tracks the stream offset of the next DIE to yield
+        # as we iterate over our children,
+        cur_offset = die.offset + die.size
+
+        while True:
+            child = self._get_cached_DIE(cur_offset)
+
+            child.set_parent(die)
+
+            if child.is_null():
+                die._terminator = child
+                return
+
+            yield child
+
+            if not child.has_children:
+                cur_offset += child.size
+            elif "DW_AT_sibling" in child.attributes and self.dwarfinfo._use_siblings:
+                sibling = child.attributes["DW_AT_sibling"]
+                if sibling.form in ('DW_FORM_ref1', 'DW_FORM_ref2',
+                                    'DW_FORM_ref4', 'DW_FORM_ref8',
+                                    'DW_FORM_ref', 'DW_FORM_ref_udata'):
+                    cur_offset = sibling.value + self.cu_offset
+                elif sibling.form == 'DW_FORM_ref_addr':
+                    cur_offset = sibling.value
+                else:
+                    raise NotImplementedError('sibling in form %s' % sibling.form)
+            else:
+                # If no DW_AT_sibling attribute is provided by the producer
+                # then the whole child subtree must be parsed to find its next
+                # sibling. There is one zero byte representing null DIE
+                # terminating children list. It is used to locate child subtree
+                # bounds.
+
+                # If children are not parsed yet, this instruction will manage
+                # to recursive call of this function which will result in
+                # setting of `_terminator` attribute of the `child`.
+                if child._terminator is None:
+                    for _ in self.iter_DIE_children(child):
+                        pass
+
+                cur_offset = child._terminator.offset + child._terminator.size
+    elftools.dwarf.compileunit.CompileUnit.iter_DIE_children = iter_DIE_children
+
     # Short out import directory parsing for now
     filebytes.pe.PE._parseDataDirectory = lambda self,a,b,c: None
 

setup.py

@@ -123,7 +123,7 @@ def run(self):
 
 setup(
     name='dwex',
-    version='4.54',  # Sync with version in __main__
+    version='4.55',  # Sync with version in __main__
     packages=['dwex'],
     url="https://github.com/sevaa/dwex/",
     entry_points={"gui_scripts": ["dwex = dwex.__main__:main"]},