perf: Correctly parse frames with " (" in them (#753)

Closes: #673 Co-authored-by: Marcin Olszewski <[email protected]>
intel · Jun 25, 2023 · e3d865a · e3d865a
1 parent 9e3c08d
commit e3d865a
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 2 deletions.
diff --git a/gprofiler/profilers/perf.py b/gprofiler/profilers/perf.py
@@ -52,7 +52,16 @@
 # ffffffff81082227 mmput+0x57 ([kernel.kallsyms])
 # 0 [unknown] ([unknown])
 # 7fe48f00faff __poll+0x4f (/lib/x86_64-linux-gnu/libc-2.31.so)
-FRAME_REGEX = re.compile(r"^\s*[0-9a-f]+ (.*?) \(\[?(.*?)\]?\)$")
+FRAME_REGEX = re.compile(
+    r"""
+    ^\s*[0-9a-f]+[ ]                                 # first a hexadecimal offset
+    (?P<symbol>.*)[ ]                                # a symbol name followed by a space
+    \( (?:                                           # dso name is either:
+        \[ (?P<dso_brackets> [^]]+) \]               # - text enclosed in square brackets, e.g.: [vdso]
+        | (?P<dso_plain> [^)]+(?:[ ]\(deleted\))? )  # - OR library name, optionally followed by " (deleted)" tag
+    ) \)$""",
+    re.VERBOSE,
+)
 SAMPLE_REGEX = re.compile(
     r"\s*(?P<comm>.+?)\s+(?P<pid>[\d-]+)/(?P<tid>[\d-]+)(?:\s+\[(?P<cpu>\d+)])?\s+(?P<time>\d+\.\d+):\s+"
     r"(?:(?P<freq>\d+)\s+)?(?P<event_family>[\w\-_/]+):(?:(?P<event>[\w-]+):)?(?P<suffix>[^\n]*)(?:\n(?P<stack>.*))?",
@@ -120,7 +129,7 @@ def _collapse_stack(comm: str, stack: str, insert_dso_name: bool = False) -> str
     for line in reversed(stack.splitlines()):
         m = FRAME_REGEX.match(line)
         assert m is not None, f"bad line: {line}"
-        sym, dso = m.groups()
+        sym, dso = m.group("symbol"), m.group("dso_brackets") or m.group("dso_plain")
         sym = sym.split("+")[0]  # strip the offset part.
         if sym == "[unknown]" and dso != "unknown":
             sym = f"({dso})"

diff --git a/tests/test_perf.py b/tests/test_perf.py
@@ -382,6 +382,14 @@ def test_get_average_frame_count(samples: str, count: float) -> None:
             ),
             id="mixed_java_stack",
         ),
+        pytest.param(
+            "   1234 std::__1::__function::__func<Envoy::Event::DispatcherImpl::createFileEvent(int, std::__1::function<void (unsigned int)>, Envoy::Event::FileTriggerType, unsigned int)::$_5, std::__1::allocator<Envoy::Event::DispatcherImpl::createFileEvent(int, std::__1::function<void (unsigned int)>, Envoy::Event::FileTriggerType, unsigned int)::$_5>, void (unsigned int)>::operator() (/path/to/envoy)\n",  # noqa
+            dict(
+                dso_true="std::__1::__function::__func<Envoy::Event::DispatcherImpl::createFileEvent(int, std::__1::function<void (unsigned int)>, Envoy::Event::FileTriggerType, unsigned int)::$_5, std::__1::allocator<Envoy::Event::DispatcherImpl::createFileEvent(int, std::__1::function<void (unsigned int)>, Envoy::Event::FileTriggerType, unsigned int)::$_5>, void (unsigned int)>::operator() (/path/to/envoy)",  # noqa
+                dso_false="std::__1::__function::__func<Envoy::Event::DispatcherImpl::createFileEvent(int, std::__1::function<void (unsigned int)>, Envoy::Event::FileTriggerType, unsigned int)::$_5, std::__1::allocator<Envoy::Event::DispatcherImpl::createFileEvent(int, std::__1::function<void (unsigned int)>, Envoy::Event::FileTriggerType, unsigned int)::$_5>, void (unsigned int)>::operator()",  # noqa
+            ),
+            id="frame_with_space_parenthesis",
+        ),
     ],
 )
 def test_collapse_stack_consider_dso(stack: str, insert_dso_name: bool, outcome_dict: Dict[str, str]) -> None: