Extended test for ARM thumb relocations and added reference disassembly

src/pelfy/_fields_data.py

@@ -434,7 +434,7 @@ relocation_table_types = {
         44: ("R_ARM_MOVT_ABS", 16, "S + A"),
 
         47: ("R_ARM_THM_MOVW_ABS_NC", 16, "S + A"),
-        48: ("R_ARM_THM_MOVT_ABS_NC", 16, "S + A"),
+        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')
 
@@ -574,7 +575,7 @@ 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':
@@ -594,11 +595,20 @@ class elf_file:
                 return _decode_thumb_branch_imm(field, 22)
             if name in ('R_ARM_THM_JUMP24', 'R_ARM_THM_CALL'):
                 return _decode_thumb_branch_imm(field, 24)
+            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 '_THM_' in name:
-                print(f'Warning: Thumb relocation addend extraction is for {name} not implemented')
+                warnings.warn(f'Thumb relocation addend extraction is for {name} not implemented', stacklevel=2)
                 return 0
             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 +1,65 @@
+import pelfy._main as _main
+import os
+
+def test_arm_addend_extraction():
+    # 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})"
+
+        print(found, f"Missing or incorrect addend for {ref_type} offset=0x{ref_offset:X} symbol={ref_symbol} (value={addend}, expected {ref_addend})")
+
+    assert False

tests/test_thumb_addend.py

@@ -0,0 +1,65 @@
+import pelfy._main as _main
+import os
+
+def test_thumb_addend_extraction():
+    # 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 = [
+        ('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, 0, 'auxsub_get_42'),
+        ('R_ARM_THM_CALL', 0xA, 0, 'result_int'),
+        ('R_ARM_THM_JUMP24', 0xA, 0, 'result_float_int'),
+        ('R_ARM_THM_JUMP24', 0xC, 0, 'result_float_float'),
+        ('R_ARM_THM_JUMP24', 0xA, 0, 'result_int_int'),
+        ('R_ARM_THM_JUMP24', 0xC, 0, 'result_int_float'),
+        ('R_ARM_THM_CALL', 0xA, 0, 'aux_get_42'),
+        ('R_ARM_THM_JUMP24', 0x12, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, 0, 'aux_get_42'),
+        ('R_ARM_THM_JUMP24', 0xA, 0, 'result_float'),
+        ('R_ARM_THM_JUMP24', 0x2, 0, 'result_int'),
+        ('R_ARM_THM_JUMP24', 0x4, 0, 'result_float'),
+        ('R_ARM_THM_JUMP24', 0x28, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2C, 0, 'sqrtf'),
+        ('R_ARM_THM_JUMP24', 0x20, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0x24, 0, 'sqrtf'),
+        ('R_ARM_THM_CALL', 0xA, 0, 'expf'),
+        ('R_ARM_THM_JUMP24', 0x12, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, 0, 'expf'),
+        ('R_ARM_THM_JUMP24', 0xA, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0xA, 0, 'logf'),
+        ('R_ARM_THM_JUMP24', 0x12, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, 0, 'logf'),
+        ('R_ARM_THM_JUMP24', 0xA, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0xA, 0, 'sinf'),
+        ('R_ARM_THM_JUMP24', 0x12, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, 0, 'sinf'),
+        ('R_ARM_THM_JUMP24', 0xA, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0xA, 0, 'cosf'),
+        ('R_ARM_THM_JUMP24', 0x12, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, 0, 'cosf'),
+        ('R_ARM_THM_JUMP24', 0xA, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0xA, 0, 'tanf'),
+        ('R_ARM_THM_JUMP24', 0x12, 0, 'result_float'),
+        ('R_ARM_THM_CALL', 0x2, 0, '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})"