simpleperf: switch to llvm-symbolizer.

Bug: 74395415
Test: run test.py.
Change-Id: I4cc0013735bedcfd18813428e4892049277c38ff

3 files changed, 44 insertions, 29 deletions

diff --git a/simpleperf/scripts/pprof_proto_generator.py b/simpleperf/scripts/pprof_proto_generator.py
index 4c59cc56..6487a66e 100755
--- a/simpleperf/scripts/pprof_proto_generator.py
+++ b/simpleperf/scripts/pprof_proto_generator.py
@@ -442,8 +442,8 @@ class PprofProfileGenerator(object):
         if not self.config.get('binary_cache_dir'):
             log_info("Can't generate line information because binary_cache is missing.")
             return
-        if not find_tool_path('addr2line', self.config['ndk_path']):
-            log_info("Can't generate line information because can't find addr2line.")
+        if not find_tool_path('llvm-symbolizer', self.config['ndk_path']):
+            log_info("Can't generate line information because can't find llvm-symbolizer.")
             return
         addr2line = Addr2Nearestline(self.config['ndk_path'], self.config['binary_cache_dir'], True)
 
diff --git a/simpleperf/scripts/test.py b/simpleperf/scripts/test.py
index e24f58f1..1066404b 100755
--- a/simpleperf/scripts/test.py
+++ b/simpleperf/scripts/test.py
@@ -1016,7 +1016,7 @@ class TestTools(unittest.TestCase):
                 {
                     'func_addr': 0x840,
                     'addr': 0x840,
-                    'source': 'system/extras/simpleperf/runtest/two_functions.cpp:7',
+                    'source': 'system/extras/simpleperf/runtest/two_functions.cpp:6',
                     'function': 'Function1()',
                 },
                 {
@@ -1065,7 +1065,7 @@ class TestTools(unittest.TestCase):
                     expected_lines.append(int(items[1]))
                 for line in test_addr['function'].split('\n'):
                     expected_functions.append(line.strip())
-                self.assertEquals(len(expected_files), len(expected_functions))
+                self.assertEqual(len(expected_files), len(expected_functions))
 
                 actual_source = addr2line.get_addr_source(dso, test_addr['addr'])
                 self.assertTrue(actual_source is not None)
diff --git a/simpleperf/scripts/utils.py b/simpleperf/scripts/utils.py
index 9891e606..2321b80f 100644
--- a/simpleperf/scripts/utils.py
+++ b/simpleperf/scripts/utils.py
@@ -148,15 +148,16 @@ EXPECTED_TOOLS = {
     'adb': {
         'is_binutils': False,
         'test_option': 'version',
-        'path_in_ndk': '../platform-tools/adb',
+        'path_in_ndk': lambda _: '../platform-tools/adb',
     },
     'readelf': {
         'is_binutils': True,
         'accept_tool_without_arch': True,
     },
-    'addr2line': {
-        'is_binutils': True,
-        'accept_tool_without_arch': True
+    'llvm-symbolizer': {
+        'is_binutils': False,
+        'path_in_ndk':
+            lambda platform: 'toolchains/llvm/prebuilt/%s-x86_64/bin/llvm-symbolizer' % platform,
     },
     'objdump': {
         'is_binutils': True,
@@ -196,7 +197,7 @@ def find_tool_path(toolname, ndk_path=None, arch=None):
         toolname_with_arch, path_in_ndk = _get_binutils_path_in_ndk(toolname, arch, platform)
     else:
         toolname_with_arch = toolname
-        path_in_ndk = tool_info['path_in_ndk']
+        path_in_ndk = tool_info['path_in_ndk'](platform)
     path_in_ndk = path_in_ndk.replace('/', os.sep)
 
     # 1. Find tool in the given ndk path.
@@ -391,7 +392,7 @@ def find_real_dso_path(dso_path_in_record_file, binary_cache_path):
 
 
 class Addr2Nearestline(object):
-    """ Use addr2line to convert (dso_path, func_addr, addr) to (source_file, line) pairs.
+    """ Use llvm-symbolizer to convert (dso_path, func_addr, addr) to (source_file, line).
         For instructions generated by C++ compilers without a matching statement in source code
         (like stack corruption check, switch optimization, etc.), addr2line can't generate
         line information. However, we want to assign the instruction to the nearest line before
@@ -435,9 +436,9 @@ class Addr2Nearestline(object):
             self.source_lines = None
 
     def __init__(self, ndk_path, binary_cache_path, with_function_name):
-        self.addr2line_path = find_tool_path('addr2line', ndk_path)
-        if not self.addr2line_path:
-            log_exit("Can't find addr2line. Please set ndk path with --ndk_path option.")
+        self.symbolizer_path = find_tool_path('llvm-symbolizer', ndk_path)
+        if not self.symbolizer_path:
+            log_exit("Can't find llvm-symbolizer. Please set ndk path with --ndk_path option.")
         self.readelf = ReadElf(ndk_path)
         self.dso_map = {}  # map from dso_path to Dso.
         self.binary_cache_path = binary_cache_path
@@ -504,12 +505,11 @@ class Addr2Nearestline(object):
                     break
         if not addr_set:
             return
-        addr_request = '\n'.join(['%x' % addr for addr in sorted(addr_set)])
+        addr_request = '\n'.join(['0x%x' % addr for addr in sorted(addr_set)])
 
         # 2. Use addr2line to collect line info.
         try:
-            option = '-ai' + ('fC' if self.with_function_name else '')
-            subproc = subprocess.Popen([self.addr2line_path, option, '-e', real_path],
+            subproc = subprocess.Popen(self._build_symbolizer_args(real_path),
                                        stdin=subprocess.PIPE, stdout=subprocess.PIPE)
             (stdoutdata, _) = subproc.communicate(str_to_bytes(addr_request))
             stdoutdata = bytes_to_str(stdoutdata)
@@ -520,6 +520,9 @@ class Addr2Nearestline(object):
         need_function_name = self.with_function_name
         cur_function_name = None
         for line in stdoutdata.strip().split('\n'):
+            line = line.strip()
+            if not line:
+                continue
             if line[:2] == '0x':
                 # a new address
                 cur_line_list = addr_map[int(line, 16)] = []
@@ -528,27 +531,16 @@ class Addr2Nearestline(object):
                 need_function_name = False
             else:
                 need_function_name = self.with_function_name
-                # a file:line.
                 if cur_line_list is None:
                     continue
-                # Handle lines like "C:\Users\...\file:32".
-                items = line.rsplit(':', 1)
-                if len(items) != 2:
-                    continue
-                if '?' in line:
-                    # if ? in line, it doesn't have a valid line info.
+                file_path, line_number = self._parse_source_location(line)
+                if not file_path or not line_number:
                     # An addr can have a list of (file, line), when the addr belongs to an inlined
                     # function. Sometimes only part of the list has ? mark. In this case, we think
                     # the line info is valid if the first line doesn't have ? mark.
                     if not cur_line_list:
                         cur_line_list = None
                     continue
-                (file_path, line_number) = items
-                line_number = line_number.split()[0]  # Remove comments after line number
-                try:
-                    line_number = int(line_number)
-                except ValueError:
-                    continue
                 file_id = self._get_file_id(file_path)
                 if self.with_function_name:
                     func_id = self._get_func_id(cur_function_name)
@@ -570,6 +562,29 @@ class Addr2Nearestline(object):
                 if shifted_addr == addr_obj.func_addr:
                     break
 
+    def _build_symbolizer_args(self, binary_path):
+        args = [self.symbolizer_path, '-print-address', '-inlining', '-obj=%s' % binary_path]
+        if self.with_function_name:
+            args += ['-functions=linkage', '-demangle']
+        else:
+            args.append('-functions=none')
+        return args
+
+    def _parse_source_location(self, line):
+        file_path, line_number = None, None
+        # Handle lines in format filename:line:column, like "runtest/two_functions.cpp:14:25".
+        # Filename may contain ':' like "C:\Users\...\file".
+        items = line.rsplit(':', 2)
+        if len(items) == 3:
+            file_path, line_number = items[:2]
+        if not file_path or ('?' in file_path) or not line_number or ('?' in line_number):
+            return None, None
+        try:
+            line_number = int(line_number)
+        except ValueError:
+            return None, None
+        return file_path, line_number
+
     def _get_file_id(self, file_path):
         file_id = self.file_name_to_id.get(file_path)
         if file_id is None: