Merge pull request #4 from Nonannet/dev

Dev

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pelfy"
-version = "1.0.7"
+version = "1.0.8"
 authors = [
   { name="Nicolas Kruse", email="nicolas.kruse@nonan.net" },
 ]

src/pelfy/_fields_data.py

@@ -390,38 +390,42 @@ relocation_table_types = {
     },
     "EM_ARM": {
         0: ("R_ARM_NONE", 0, ""),
-        1: ("R_ARM_PC24", 24, "S - P + A"),
-        2: ("R_ARM_ABS32", 32, "S + A"),
-        3: ("R_ARM_REL32", 32, "S - P + A"),
-        4: ("R_ARM_PC13", 13, "S - P + A"),
+        1: ("R_ARM_PC24", 24, "((S + A) | T) - P"),
+        2: ("R_ARM_ABS32", 32, "(S + A) | T"),
+        3: ("R_ARM_REL32", 32, "((S + A) | T) - P"),
+        4: ("R_ARM_LDR_PC_G0", 12, "S + A - P"),
         5: ("R_ARM_ABS16", 16, "S + A"),
         6: ("R_ARM_ABS12", 12, "S + A"),
         7: ("R_ARM_THM_ABS5", 5, "S + A"),
         8: ("R_ARM_ABS8", 8, "S + A"),
-        9: ("R_ARM_SBREL32", 32, "S - B + A"),
-        10: ("R_ARM_THM_PC22", 22, "S - P + A"),
-        11: ("R_ARM_THM_PC8", 8, "S - P + A"),
-        12: ("Reserved", 0, ""),
-        13: ("R_ARM_SWI24", 24, "S + A"),
+        9: ("R_ARM_SBREL32", 32, "((S + A) | T) - B(S)"),
+        10: ("R_ARM_THM_CALL", 22, "((S + A) | T) - P"),
+        11: ("R_ARM_THM_PC8", 8, "S + A - Pa"),
+        12: ("R_ARM_BREL_ADJ", 0, "ChangeIn[B(S)] + A"),
+        13: ("R_ARM_TLS_DESC", 0, ""),
         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"),
-        34: ("R_ARM_ALU_PCREL_23_15", 23, "(S - P + A) & 0x00FF0000"),
-        35: ("R_ARM_LDR_SBREL_11_0", 11, "(S - B + A) & 0x00000FFF"),
-        36: ("R_ARM_ALU_SBREL_19_12", 19, "(S - B + A) & 0x000FF000"),
-        37: ("R_ARM_ALU_SBREL_27_20", 27, "(S - B + A) & 0x0FF00000"),
-        38: ("R_ARM_RELABS32", 32, "S + A or S - P + A"),
-        39: ("R_ARM_ROSEGREL32", 32, "S - E + A"),
+        17: ("R_ARM_TLS_DTPMOD32", 0, "Module[S]"),
+        18: ("R_ARM_TLS_DTPOFF32", 0, "S + A - TLS"),
+        19: ("R_ARM_TLS_TPOFF32", 0, "S + A - tp"),
+        20: ("R_ARM_COPY", 0, ""),
+        21: ("R_ARM_GLOB_DAT", 0, "(S + A) | T"),
+        22: ("R_ARM_JUMP_SLOT", 0, "(S + A) | T"),
+        23: ("R_ARM_RELATIVE", 0, "B(S) + A"),
+        24: ("R_ARM_GOTOFF32", 0, "((S + A) | T) - GOT_ORG"),
+        25: ("R_ARM_BASE_PREL", 0, "B(S) + A - P"),
+        26: ("R_ARM_GOT_BREL", 0, "GOT(S) + A - GOT_ORG"),
+        27: ("R_ARM_PLT32", 0, "((S + A) | T) - P"),
+        28: ("R_ARM_CALL", 24, "((S + A) | T) - P"),
+        29: ("R_ARM_JUMP24", 24, "((S + A) | T) - P"),
+        30: ("R_ARM_THM_JUMP24", 24, "((S + A) | T) - P"),
         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")
+        42: ("R_ARM_PREL31", 31, "((S + A) | T) - P"),
+        43: ("R_ARM_MOVW_ABS_NC", 16, "(S + A) | T"),
+        44: ("R_ARM_MOVT_ABS", 16, "S + A"),
+        47: ("R_ARM_THM_MOVW_ABS_NC", 16, "(S + A) | T"),
+        48: ("R_ARM_THM_MOVT_ABS", 16, "S + A"),
     },
     "EM_AARCH64": {
         0: ("R_AARCH64_NONE", 0, ""),

src/pelfy/_main.py

@@ -9,6 +9,7 @@ Typical usage example:
 from . import _fields_data as fdat
 from . import _output_formatter
 from typing import TypeVar, Literal, Iterable, Generic, Iterator, Generator, Optional, Union
+import warnings
 
 _T = TypeVar('_T')
 
@@ -26,6 +27,14 @@ def open_elf_file(file_path: str) -> 'elf_file':
         return elf_file(f.read())
 
 
+def _sign_extend(value: int, bits: int) -> int:
+    """Sign-extend a value with given bit width to 32 bits."""
+    sign_bit = 1 << (bits - 1)
+    if value & sign_bit:
+        return value - (1 << bits)
+    return value
+
+
 class elf_symbol():
     """A class for representing data of an ELF symbol
 
@@ -42,6 +51,7 @@ class elf_symbol():
         offset_in_section: Position of first symbol byte
             relative to section start
         offset_in_file: Position of first symbol byte in object file
+        size: size of symbol in bytes
         fields: All symbol header fields as dict
     """
 
@@ -71,19 +81,17 @@ class elf_symbol():
         self.thumb_mode = bool((file.architecture == 'EM_ARM') & fields['st_value'] & 1)
         self.offset_in_section = fields['st_value'] & ~int(self.thumb_mode)
         self.offset_in_file = self.section['sh_offset'] + self.offset_in_section if self.section else 0
+        self.size = self.fields['st_size']
 
     @property
     def data(self) -> bytes:
         """Returns the binary data the symbol is pointing to.
-        The offset in the ELF file is calculated by:
-        sections[symbol.st_shndx].sh_offset + symbol.st_value
         """
         assert self.section, 'This symbol is not associated to a data section'
         if self.section.type == 'SHT_NOBITS':
             return b'\x00' * self['st_size']
         else:
-            offset = self.section['sh_offset'] + self['st_value']
-            return self.file.read_bytes(offset, self['st_size'])
+            return self.file.read_bytes(self.offset_in_file, self['st_size'])
 
     @property
     def data_hex(self) -> str:
@@ -102,7 +110,7 @@ class elf_symbol():
         assert self.section and self.section.type == 'SHT_PROGBITS'
         for reloc in self.file.get_relocations():
             if reloc.target_section == self.section:
-                offset = reloc['r_offset'] - self['st_value']
+                offset = reloc['r_offset'] - self.offset_in_section
                 if 0 <= offset < self['st_size']:
                     ret.append(reloc)
         return relocation_list(ret)
@@ -528,7 +536,7 @@ class elf_file:
         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)
+            field = self.read_int(r_offset + sh['sh_offset'], 4, False)
             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':
@@ -544,11 +552,45 @@ class elf_file:
                 if imm24 & 0x800000:
                     imm24 |= ~0xFFFFFF
                 return imm24 << 2
-            if '_THM_' in name:
-                print('Warning: Thumb relocation addend extraction is not implemented')
-                return 0
+            if name == 'R_ARM_THM_PC22':
+                # BL Encoding T2
+                h1 = field & 0xFFFF
+                h2 = (field >> 16) & 0xFFFF
+                s = (h1 >> 10) & 0x1  # Bit 10 of h1
+                imm10H = h1 & 0x3FF  # Bits 9..0 of h1
+                j1 = (h2 >> 13) & 0x1  # Bit 13 of h2
+                j2 = (h2 >> 11) & 0x1  # Bit 11 of h2
+                imm10L = (h2 >> 1) & 0x3FF  # Bits 10..1 of h2
+                i1 = ~(j1 ^ s) & 0x1
+                i2 = ~(j2 ^ s) & 0x1
+                imm22 = (s << 21) | (i1 << 20) | (i2 << 19) | (imm10H << 9) | imm10L
+                imm32 = _sign_extend(imm22 << 1, 23)  # Shift left by 1, then sign-extend 23 bits
+                return imm32
+            if name in ('R_ARM_THM_JUMP24', 'R_ARM_THM_CALL'):
+                # B.W Encoding T4
+                h1 = field & 0xFFFF
+                h2 = (field >> 16) & 0xFFFF
+                s = (h1 >> 10) & 0x1  # Bit 10 of h1
+                imm10 = h1 & 0x3FF  # Bits 9..0 of h1
+                j1 = (h2 >> 13) & 0x1  # Bit 13 of h2
+                j2 = (h2 >> 11) & 0x1  # Bit 11 of h2
+                imm11 = h2 & 0x7FF  # Bits 10..0 of h2
+                i1 = ~(j1 ^ s) & 0x1
+                i2 = ~(j2 ^ s) & 0x1
+                imm24 = (s << 23) | (i1 << 22) | (i2 << 21) | (imm10 << 11) | imm11
+                imm32 = _sign_extend(imm24 << 1, 25)  # Shift left by 1, then sign-extend 25 bits
+                return imm32
+            if name == 'R_ARM_THM_MOVW_ABS_NC' or name == 'R_ARM_THM_MOVT_ABS':
+                i = (field >> 10) & 1
+                imm4 = field & 0xF
+                imm3 = (field >> 28) & 0x7
+                imm8 = (field >> 16) & 0xFF
+                imm16 = imm8 | (imm3 << 8) | (i << 11) | (imm4 << 12)
+                if name == 'R_ARM_THM_MOVT_ABS':
+                    return imm16 << 16
+                return imm16
             if '_MIPS_' in name:
-                print('Warning: MIPS relocations addend extraction is not implemented')
+                warnings.warn('Warning: MIPS relocations addend extraction is not implemented', stacklevel=2)
                 return 0
             raise NotImplementedError(f"Relocation addend extraction for {name} is not implemented")
 

tests/test_arm_addend.py

@@ -0,0 +1,62 @@
+import pelfy._main as _main
+import os
+
+
+def test_arm_addend_extraction() -> None:
+    # Path to the test object file
+    obj_path = os.path.join('tests', 'obj', 'stencils_armv7_O3.o')
+    elf = _main.open_elf_file(obj_path)
+
+    # Collect all ARM relocations
+    reloc_addends: list[tuple[str, int, int, str]] = []
+    for reloc in elf.get_relocations():
+        if reloc.type.startswith('R_ARM'):
+            reloc_addends.append((reloc.type, reloc['r_offset'], reloc['r_addend'], reloc.symbol.name))
+
+    # Reference values from stencils_armv7_O3.asm (addend = 0 for V4BX, -8 for JUMP24/CALL)
+    reference = [
+        ('R_ARM_V4BX', 0xB4, 0, ''),
+        ('R_ARM_V4BX', 0xC0, 0, ''),
+        ('R_ARM_V4BX', 0xD0, 0, ''),
+        ('R_ARM_V4BX', 0x114, 0, ''),
+        ('R_ARM_V4BX', 0x144, 0, ''),
+        ('R_ARM_V4BX', 0x148, 0, ''),
+        ('R_ARM_JUMP24', 0x124, -8, '__aeabi_idiv0'),
+        ('R_ARM_JUMP24', 0x14, -8, 'auxsub_get_42'),
+        ('R_ARM_CALL', 0x10, -8, 'result_int'),
+        ('R_ARM_JUMP24', 0xC, -8, 'result_float_int'),
+        ('R_ARM_JUMP24', 0xC, -8, 'result_float_float'),
+        ('R_ARM_JUMP24', 0xC, -8, 'result_int_int'),
+        ('R_ARM_JUMP24', 0xC, -8, 'result_int_float'),
+        ('R_ARM_CALL', 0xC, -8, 'aux_get_42'),
+        ('R_ARM_JUMP24', 0x14, -8, 'result_float'),
+        ('R_ARM_CALL', 0x4, -8, 'aux_get_42'),
+        ('R_ARM_JUMP24', 0xC, -8, 'result_float'),
+        ('R_ARM_JUMP24', 0x4, -8, 'result_int'),
+        ('R_ARM_JUMP24', 0x4, -8, 'result_float'),
+        ('R_ARM_JUMP24', 0x2C, -8, 'result_float'),
+        ('R_ARM_CALL', 0x30, -8, 'sqrtf'),
+        ('R_ARM_JUMP24', 0x24, -8, 'result_float'),
+        ('R_ARM_CALL', 0x28, -8, 'sqrtf'),
+        ('R_ARM_CALL', 0xC, -8, 'expf'),
+        ('R_ARM_JUMP24', 0x14, -8, 'result_float'),
+        ('R_ARM_CALL', 0x4, -8, 'expf'),
+        ('R_ARM_JUMP24', 0xC, -8, 'result_float'),
+        ('R_ARM_CALL', 0xC, -8, 'logf'),
+        ('R_ARM_JUMP24', 0x14, -8, 'result_float'),
+        ('R_ARM_CALL', 0x4, -8, 'logf'),
+        ('R_ARM_JUMP24', 0xC, -8, 'result_float'),
+        ('R_ARM_CALL', 0xC, -8, 'sinf'),
+        ('R_ARM_JUMP24', 0x14, -8, 'result_float'),
+        ('R_ARM_CALL', 0x4, -8, 'sinf'),
+        ('R_ARM_JUMP24', 0xC, -8, 'result_float'),
+    ]
+    # For each reference, check that at least one matching relocation has the expected addend
+    for ref_type, ref_offset, ref_addend, ref_symbol in reference:
+        found = False
+        addend = None
+        for typ, offset, addend, symbol in reloc_addends:
+            if typ == ref_type and offset == ref_offset and symbol == ref_symbol and addend == ref_addend:
+                found = True
+                break
+        assert found, f"Missing or incorrect addend for {ref_type} offset=0x{ref_offset:X} symbol={ref_symbol} (value={addend}, expected {ref_addend})"

tests/test_thumb_addend.py

@@ -0,0 +1,67 @@
+import pelfy._main as _main
+import os
+from typing import Union
+
+
+def test_thumb_addend_extraction() -> None:
+    # Path to the test object file
+    obj_path = os.path.join('tests', 'obj', 'stencils_armv7thumb_O3_THM_MOVW.o')
+    elf = _main.open_elf_file(obj_path)
+
+    # Collect all relocations of interest
+    reloc_addends: list[tuple[str, int, int, str]] = []
+    for reloc in elf.get_relocations():
+        if reloc.type.startswith('R_ARM_THM'):
+            reloc_addends.append((reloc.type, reloc['r_offset'], reloc['r_addend'], reloc.symbol.name))
+
+    # Reference values from the .asm file (addend = 0 for all Thumb relocations)
+    reference: list[tuple[str, Union[int, None], int, str]] = [
+        ('R_ARM_THM_MOVW_ABS_NC', None, 0, 'dummy_int'),
+        ('R_ARM_THM_MOVT_ABS', None, 0, 'dummy_int'),
+        ('R_ARM_THM_MOVW_ABS_NC', None, 0, 'dummy_float'),
+        ('R_ARM_THM_MOVT_ABS', None, 0, 'dummy_float'),
+        ('R_ARM_THM_JUMP24', 0x14, -4, 'auxsub_get_42'),
+        ('R_ARM_THM_CALL', 0xA, -4, 'result_int'),
+        ('R_ARM_THM_JUMP24', 0xA, -4, 'result_float_int'),
+        ('R_ARM_THM_JUMP24', 0xC, -4, 'result_float_float'),
+        ('R_ARM_THM_JUMP24', 0xA, -4, 'result_int_int'),
+        ('R_ARM_THM_JUMP24', 0xC, -4, 'result_int_float'),
+        ('R_ARM_THM_CALL', 0xA, -4, 'aux_get_42'),
+        ('R_ARM_THM_JUMP24', 0x12, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, -4, 'aux_get_42'),
+        ('R_ARM_THM_JUMP24', 0xA, -4, 'result_float'),
+        ('R_ARM_THM_JUMP24', 0x2, -4, 'result_int'),
+        ('R_ARM_THM_JUMP24', 0x4, -4, 'result_float'),
+        ('R_ARM_THM_JUMP24', 0x28, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2C, -4, 'sqrtf'),
+        ('R_ARM_THM_JUMP24', 0x20, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0x24, -4, 'sqrtf'),
+        ('R_ARM_THM_CALL', 0xA, -4, 'expf'),
+        ('R_ARM_THM_JUMP24', 0x12, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, -4, 'expf'),
+        ('R_ARM_THM_JUMP24', 0xA, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0xA, -4, 'logf'),
+        ('R_ARM_THM_JUMP24', 0x12, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, -4, 'logf'),
+        ('R_ARM_THM_JUMP24', 0xA, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0xA, -4, 'sinf'),
+        ('R_ARM_THM_JUMP24', 0x12, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, -4, 'sinf'),
+        ('R_ARM_THM_JUMP24', 0xA, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0xA, -4, 'cosf'),
+        ('R_ARM_THM_JUMP24', 0x12, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, -4, 'cosf'),
+        ('R_ARM_THM_JUMP24', 0xA, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0xA, -4, 'tanf'),
+        ('R_ARM_THM_JUMP24', 0x12, -4, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, -4, 'tanf'),
+    ]
+    # For each reference, check that at least one matching relocation has the expected addend
+    for ref_type, _, ref_addend, ref_symbol in reference:
+        found = False
+        addend = 0
+        for typ, offset, addend, symbol in reloc_addends:
+            if typ == ref_type and symbol == ref_symbol and addend == ref_addend:
+                found = True
+                break
+        assert found, f"Missing or incorrect addend for {ref_type} {ref_symbol} (value={addend:X}, expected {ref_addend})"