Introduce proper file stripping and start adding parser tests

test/templates/common.py

@@ -21,6 +21,9 @@ def error(msg):
 
 
 def _run(cmd, stdin=None):
+    """
+    Run process and abort on error.
+    """
     if verbose:
         print(f"{_me}: running command: {' '.join(cmd)}")
     with subprocess.Popen(cmd, stdin=stdin, stdout=subprocess.PIPE,
@@ -36,27 +39,38 @@ def _run(cmd, stdin=None):
 
 
 def gcc(args):
+    """
+    Run compiler with given arguments.
+    """
     return _run(['gcc'] + args)
 
 
-def disasm(file, objdump_or_gdb, strip=None, symbol=None):
+def disasm(file, objdump_or_gdb, symbol, start, finish):
+    """
+    Disassemble binary file.
+    """
     if objdump_or_gdb:
         out = _run(['objdump', '-d', file])
-    elif strip:
-        # This is tricky as we can not use symbol name
-        start, finish = _find_address(file, symbol)
-        out = _run(['gdb', '-batch', '-ex', f'disassemble {start},{finish}', file])
-    else:
+    elif symbol is not None:
         out = _run(['gdb', '-batch', '-ex', f'disassemble {symbol}', file])
+    else:
+        out = _run(['gdb', '-batch', '-ex', f'disassemble {start},{finish}', file])
     return out
 
 
+def strip_binary(file):
+    """
+    Strip symbol info from binary file.
+    """
+    _run(['strip', '-s', file])
+
+
 def grep(s, regex):
     lines = s.split('\n')
     return list(filter(lambda s: re.search(regex, s), lines))
 
 
-def _find_address(file, name):
+def find_address(file, name):
     out = _run(['readelf', '-sW', file])
     lines = grep(out, fr'{name}$')
     assert len(lines) >= 1, f"failed to locate symbol {name} in\n{out}"

test/templates/gen_calls.py

@@ -6,12 +6,11 @@
 
 # TODO: find a way to produce other snippets we see in disasm:
 #  addr32 call 0x5555555733e0
-#  call   0x555555576a00
 
 import os.path
 import itertools
 
-from common import set_basename, gcc, disasm, grep
+from common import set_basename, gcc, disasm, grep, strip_binary, find_address
 
 set_basename(os.path.basename(__file__))
 
@@ -42,23 +41,22 @@
     # Force non-PLT call for PIC code?
     if direct and pic:
         flags.append('-DHIDDEN')
-    # Include debuginfo?
-    if not strip:
-        # FIXME: for real stripping of symtab we need to run
-        # `strip a.out` and `strip -s a.out`.
-        # This should be done for other tests too.
-        flags.append('-g')
 
     gcc(flags)
 
+    caller = 'bar'
+    start, finish = find_address('a.out', caller)
+    if strip:
+        strip_binary('a.out')
+        caller = None
+
     # Generate disasm
 
-    caller = 'bar'
-    out = disasm('a.out', not gdb, strip, caller)
+    out = disasm('a.out', not gdb, caller, start, finish)
 
     # Print snippets
 
-    headers = grep(out, fr'<{caller}>:|Dump of')
+    headers = grep(out, r'<bar>:|Dump of')
     calls = grep(out, r'call')
     print('''\
   headers:

test/templates/gen_funtable.py

@@ -7,7 +7,7 @@
 import os.path
 import itertools
 
-from common import set_basename, gcc, disasm, grep
+from common import set_basename, gcc, disasm, grep, strip_binary, find_address
 
 set_basename(os.path.basename(__file__))
 
@@ -34,10 +34,17 @@
 
     gcc(flags)
 
-    # Generate disasm
+    # Strip
 
     caller = 'bar'
-    out = disasm('a.out', not gdb, strip, caller)
+    start, finish = find_address('a.out', caller)
+    if strip:
+        strip_binary('a.out')
+        caller = None
+
+    # Generate disasm
+
+    out = disasm('a.out', not gdb, caller, start, finish)
 
     # Print snippets
 

test/templates/gen_jumps.py

@@ -7,7 +7,7 @@
 import os.path
 import itertools
 
-from common import set_basename, gcc, disasm, grep
+from common import set_basename, gcc, disasm, grep, strip_binary, find_address
 
 set_basename(os.path.basename(__file__))
 
@@ -32,16 +32,20 @@
         flags += ['-fPIC', '-shared']
     if opt:
         flags.append('-O2')
-    # Include debuginfo?
-    if not strip:
-        flags.append('-g')
 
     gcc(flags)
 
-    # Generate disasm
+    # Strip
 
     caller = 'bar'
-    out = disasm('a.out', not gdb, strip, caller)
+    start, finish = find_address('a.out', caller)
+    if strip:
+        strip_binary('a.out')
+        caller = None
+
+    # Generate disasm
+
+    out = disasm('a.out', not gdb, caller, start, finish)
 
     # Print snippets
 

test/templates/gen_jumptable.py

@@ -7,7 +7,7 @@
 import os.path
 import itertools
 
-from common import set_basename, gcc, disasm, grep
+from common import set_basename, gcc, disasm, grep, strip_binary, find_address
 
 set_basename(os.path.basename(__file__))
 
@@ -28,16 +28,20 @@
     # DLL or executable?
     if pic:
         flags += ['-fPIC', '-shared']
-    # Include debuginfo?
-    if not strip:
-        flags.append('-g')
 
     gcc(flags)
 
-    # Generate disasm
+    # Strip
 
     caller = 'bar'
-    out = disasm('a.out', not gdb, strip, caller)
+    start, finish = find_address('a.out', caller)
+    if strip:
+        strip_binary('a.out')
+        caller = None
+
+    # Generate disasm
+
+    out = disasm('a.out', not gdb, caller, start, finish)
 
     # Print snippets
 

test/test_parser.py

@@ -0,0 +1,84 @@
+import unittest
+
+from src.asm2cfg import asm2cfg
+
+
+def _get_the_source_block(blocks):
+    sources = [block for _, block in blocks.items() if block.jump_edge is not None]
+    return sources[0] if len(sources) == 1 else None
+
+
+class ParseLineTestCase(unittest.TestCase):
+    @unittest.expectedFailure
+    def test_linear_sequence(self):
+        lines = '''\
+Dump of assembler code for function main:
+   0x000055555556f952 <+2>:	mov    $0x1,%ecx
+   0x000055555556f957 <+7>:	push   %r14
+   0x000055555556f959 <+9>:	push   %r13
+   0x000055555556f95b <+11>:	push   %r12
+   0x000055555556f95d <+13>:	push   %rbp
+   0x000055555556f95e <+14>:	push   %rbx\
+'''.split('\n')
+        _, blocks = asm2cfg.parse_lines(lines, False)
+
+        self.assertEqual(len(blocks), 1)
+        _, block = blocks.popitem()
+        self.assertEqual(len(block.instructions), 6)  # FIXME
+        self.assertIs(block.jump_edge, None)
+        self.assertIs(block.no_jump_edge, None)
+
+    def test_unconditional(self):
+        lines = '''\
+Dump of assembler code for function main:
+   0x000055555556fffb <+1707>:	jmp    0x555555570058 <main+1800>
+   0x0000555555570058 <+1800>:	mov    0xe0(%rsp),%rdi
+   0x0000555555570060 <+1808>:	test   %rdi,%rdi
+'''.split('\n')
+        _, blocks = asm2cfg.parse_lines(lines, False)
+
+        self.assertEqual(len(blocks), 2)
+
+        source_block = _get_the_source_block(blocks)
+        self.assertIsNot(source_block.jump_edge, None)
+        self.assertIs(source_block.no_jump_edge, None)
+        self.assertEqual(len(source_block.instructions), 1)
+
+        dst_block = blocks[source_block.jump_edge]
+        self.assertIs(dst_block.jump_edge, None)
+        self.assertIs(dst_block.no_jump_edge, None)
+        self.assertEqual(len(dst_block.instructions), 2)
+
+    def test_conditional(self):
+        lines = '''\
+Dump of assembler code for function main:
+   0x000055555556fffb <+1707>:	je    0x555555570058 <main+1800>
+   0x000055555556fffd <+1709>:	push   %rbx
+   0x000055555556fffe <+1710>:	mov    %r15,%r8
+   0x000055555556fffe <+1710>:	mov    %r15,%r8
+   0x0000555555570058 <+1800>:	mov    0xe0(%rsp),%rdi
+   0x0000555555570060 <+1808>:	test   %rdi,%rdi
+'''.split('\n')
+        _, blocks = asm2cfg.parse_lines(lines, False)
+
+        self.assertEqual(len(blocks), 3)
+
+        source_block = _get_the_source_block(blocks)
+        self.assertIsNot(source_block.jump_edge, None)
+        self.assertIsNot(source_block.no_jump_edge, None)
+        self.assertEqual(len(source_block.instructions), 1)
+
+        fall_block = blocks[source_block.no_jump_edge]
+        self.assertIs(fall_block.jump_edge, None)
+        self.assertIsNot(fall_block.no_jump_edge, None)
+        self.assertEqual(len(fall_block.instructions), 3)
+
+        dst_block = blocks[source_block.jump_edge]
+        self.assertIs(dst_block.jump_edge, None)
+        self.assertIs(dst_block.no_jump_edge, None)
+        self.assertEqual(len(dst_block.instructions), 2)
+
+    # TODO:
+    # - functions (with and w/o calls)
+    # - jumptables
+    # - skip calls

test/test_regex.py

@@ -211,3 +211,13 @@ def test_objdump_jumptable(self):
         self.assertIsNot(jump_match, None)
         self.assertEqual(jump_match[1], '101d')
         self.assertEqual(jump_match[2], '')
+
+    @unittest.expectedFailure
+    def test_objdump_funnyjump(self):
+        line = '1044:	f2 ff 25 d5 2f 00 00 	bnd jmpq *0x2fd5(%rip)        # 4020 <foo+0x2fd0>'
+        pattern = asm2cfg.get_jump_pattern(False, 'does_not_matter')
+        jump_match = pattern.search(line)
+
+        self.assertIsNot(jump_match, None)
+        self.assertEqual(jump_match[1], '4020')
+        self.assertEqual(jump_match[2], '0x2fd0')