Merge pull request #2 from Nonannet/dev

Dev

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pelfy"
-version = "1.0.4"
+version = "1.0.7"
 authors = [
   { name="Nicolas Kruse", email="nicolas.kruse@nonan.net" },
 ]

src/pelfy/_fields_data.py

@@ -406,6 +406,8 @@ relocation_table_types = {
         14: ("R_ARM_THM_SWI8", 8, "S + A"),
         15: ("R_ARM_XPC25", 25, ""),
         16: ("R_ARM_THM_XPC22", 22, ""),
+        28: ("R_ARM_CALL", 24, "((S + A) - P) >> 2"),
+        29: ("R_ARM_JUMP24", 24, "((S + A) - P) >> 2"),
         30: ("R_ARM_TLS_DESC", 0, ""),
         32: ("R_ARM_ALU_PCREL_7_0", 7, "(S - P + A) & 0x000000FF"),
         33: ("R_ARM_ALU_PCREL_15_8", 15, "(S - P + A) & 0x0000FF00"),
@@ -418,6 +420,8 @@ relocation_table_types = {
         40: ("R_ARM_V4BX", 0, ""),
         41: ("R_ARM_STKCHK", 0, ""),
         42: ("R_ARM_THM_STKCHK", 0, ""),
+        43: ("R_ARM_MOVW_ABS_NC", 16, "S + A"),
+        44: ("R_ARM_MOVT_ABS", 16, "S + A")
     },
     "EM_AARCH64": {
         0: ("R_AARCH64_NONE", 0, ""),

src/pelfy/_main.py

@@ -265,6 +265,14 @@ class elf_list(Generic[_T]):
     def __iter__(self) -> Iterator[_T]:
         return iter(self._data)
 
+    def __contains__(self, item: Union[str, int, _T]) -> bool:
+        if isinstance(item, str):
+            return any(getattr(el, 'name', '') == item for el in self._data)
+        elif isinstance(item, int):
+            return 0 <= item < len(self._data)
+        else:
+            return item in self._data
+
     def _compact_table(self) -> tuple[list[str], list[list[Union[str, int]]], list[str]]:
         return [], [[]], []
 
@@ -483,22 +491,61 @@ class elf_file:
         Returns:
             Relocations from specified elf_section
         """
-        offs = sh['sh_offset']
+        sh_offset = sh['sh_offset']
-        for i, el_off in enumerate(range(offs, sh['sh_size'] + offs, sh['sh_entsize'])):
+        for i, el_off in enumerate(range(sh_offset, sh['sh_size'] + sh_offset, sh['sh_entsize'])):
             ret: dict[str, int] = dict()
 
             if self.bit_width == 32:
-                ret['r_offset'] = self.read_int(el_off, 4)
+                r_offset = self.read_int(el_off, 4)
                 r_info = self.read_int(el_off + 4, 4)
-                ret['r_info'] = r_info
+                symbol_index = r_info >> 8
-                ret['r_addend'] = self.read_int(el_off + 8, 4, True) if sh.type == 'SHT_RELA' else 0
+                relocation_type = r_info & 0xFF
-                yield elf_relocation(self, ret, r_info >> 8, r_info & 0xFF, sh['sh_info'], i)
+                ret['r_addend'] = self.read_int(el_off + 8, 4, True) \
+                    if sh.type == 'SHT_RELA' else self._get_rel_addend(relocation_type, r_offset, sh)
             elif self.bit_width == 64:
-                ret['r_offset'] = self.read_int(el_off, 8)
+                r_offset = self.read_int(el_off, 8)
                 r_info = self.read_int(el_off + 8, 8)
-                ret['r_info'] = r_info
+                symbol_index = r_info >> 32
-                ret['r_addend'] = self.read_int(el_off + 16, 8, True) if sh.type == 'SHT_RELA' else 0
+                relocation_type = r_info & 0xFFFFFFFF
-                yield elf_relocation(self, ret, r_info >> 32, r_info & 0xFFFFFFFF, sh['sh_info'], i)
+                ret['r_addend'] = self.read_int(el_off + 16, 8, True) \
+                    if sh.type == 'SHT_RELA' else self._get_rel_addend(relocation_type, r_offset, sh)
+            else:
+                raise NotImplementedError(f"{self.bit_width} bit is not supported")
+
+            ret['r_offset'] = r_offset
+            ret['r_info'] = r_info
+            yield elf_relocation(self, ret, symbol_index, relocation_type, sh['sh_info'], i)
+
+    def _get_rel_addend(self, relocation_type: int, r_offset: int, reloc_section: elf_section) -> int:
+        reloc_types = fdat.relocation_table_types.get(self.architecture)
+        if reloc_types and 'A' in reloc_types[relocation_type][2]:
+            name = reloc_types[relocation_type][0]
+            sh = self.sections[reloc_section['sh_info']]
+            field = self.read_int(r_offset + sh['sh_offset'], 4, True)
+            if name in ('R_386_PC32', 'R_386_32', 'R_X86_64_PC32', 'R_X86_64_PLT32', 'R_ARM_REL32', 'R_ARM_ABS32'):
+                return field
+            if name == 'R_ARM_MOVW_ABS_NC':
+                imm4 = (field >> 16) & 0xF
+                imm12 = field & 0xFFF
+                return imm12 | (imm4 << 12)
+            if name == 'R_ARM_MOVT_ABS':
+                imm4 = (field >> 16) & 0xF
+                imm12 = field & 0xFFF
+                return (imm12 << 16) | (imm4 << 28)  # Upper 16 bit
+            if name in ('R_ARM_JUMP24', 'R_ARM_CALL'):
+                imm24 = field & 0xFFFFFF
+                if imm24 & 0x800000:
+                    imm24 |= ~0xFFFFFF
+                return imm24 << 2
+            if '_THM_' in name:
+                print('Warning: Thumb relocation addend extraction is not implemented')
+                return 0
+            if '_MIPS_' in name:
+                print('Warning: MIPS relocations addend extraction is not implemented')
+                return 0
+            raise NotImplementedError(f"Relocation addend extraction for {name} is not implemented")
+
+        return 0
 
     def read_bytes(self, offset: int, num_bytes: int) -> bytes:
         """Read bytes from ELF file.

tests/test_example_elfs.py

@@ -38,6 +38,13 @@ def test_simple_c() -> None:
         for reloc in elf.get_relocations():
             assert known_name(reloc.type), f"Relocation type {reloc.type} for {elf.architecture} in {path} is unknown."
 
+        assert 'imageWidth' in elf.objects or 'read_float_ret' in elf.objects, path
+        assert 'leet456456456n4ghn4hf56n4f' not in elf.objects
+        assert 0 in elf.objects
+        assert 1000 not in elf.objects
+        assert elf.objects[0] in elf.symbols
+        assert elf.objects[0] not in elf.functions
+
 
 if __name__ == '__main__':
     test_simple_c()