gh-160529: Accept synthetic profiler frames and resolve BusyBox executables

Author: pablogsal

Lib/test/test_profiling/test_sampling_profiler/test_children.py

@@ -3,6 +3,7 @@
 import argparse
 import io
 import os
+import shutil
 import signal
 import subprocess
 import sys
@@ -85,12 +86,11 @@ def reader():
     return result[0]
 
 
-def _wait_for_process_exec(proc, name, timeout):
+def _wait_for_process_exec(proc, executable, timeout):
     # Wait until a child process calls exec() to replace the forked process
     # executable. Raise AssertionError on timeout.
     for _ in sleeping_retry(timeout, max_delay=0.1):
-        exe = os.readlink(f"/proc/{proc.pid}/exe")
-        if os.path.basename(exe) == name:
+        if os.path.samefile(f"/proc/{proc.pid}/exe", executable):
             break
 
 
@@ -617,15 +617,18 @@ def test_is_python_process_python_subprocess(self):
     @unittest.skipUnless(sys.platform == "linux", "Linux only test")
     def test_is_python_process_non_python_subprocess(self):
         """Test that a non-Python subprocess is not detected as Python."""
+        executable = shutil.which("sleep")
+        if executable is None:
+            self.skipTest("sleep command not found")
         # Start a non-Python subprocess (sleep command)
         proc = subprocess.Popen(
-            ["sleep", "60"],
+            [executable, "60"],
             stdout=subprocess.DEVNULL,
             stderr=subprocess.DEVNULL,
         )
 
         try:
-            _wait_for_process_exec(proc, "sleep", SHORT_TIMEOUT)
+            _wait_for_process_exec(proc, executable, SHORT_TIMEOUT)
 
             self.assertFalse(
                 is_python_process(proc.pid),

Lib/test/test_profiling/test_sampling_profiler/test_collectors.py

@@ -467,6 +467,24 @@ def test_collapsed_stack_collector_export(self):
         self.assertIn(stack1_expected, lines)
         self.assertIn(stack2_expected, lines)
 
+    def test_collapsed_stack_collector_export_synthetic_frames(self):
+        collapsed_out = tempfile.NamedTemporaryFile(delete=False)
+        self.addCleanup(close_and_unlink, collapsed_out)
+
+        for name in ("<GC>", "<native>", "<unreadable frame>"):
+            with self.subTest(name=name):
+                collector = CollapsedStackCollector(1000)
+                collector.collect([
+                    MockInterpreterInfo(0, [MockThreadInfo(1, [
+                        _remote_debugging.FrameInfo(("~", None, name, None)),
+                        MockFrameInfo("app.py", 10, "main"),
+                    ])])
+                ])
+                with captured_stdout():
+                    collector.export(collapsed_out.name)
+                with open(collapsed_out.name, encoding="utf-8") as f:
+                    self.assertEqual(f.read(), f"tid:1;app.py:main:10;{name} 1\n")
+
     def test_collapsed_stack_collector_export_non_ascii_names(self):
         # gh-156810: frame names are written verbatim, so the output must be
         # opened with an encoding that can represent non-ASCII and

Lib/test/test_profiling/test_sampling_profiler/test_integration.py

@@ -508,6 +508,9 @@ def test_sampling_with_collapsed_export(self):
                 if ";" in stack_trace:
                     stack_parts = stack_trace.split(";")
                     for part in stack_parts:
+                        # Synthetic frames contain only their name.
+                        if part in {"<GC>", "<native>", "<unreadable frame>"}:
+                            continue
                         # Each part should be file:function:line
                         self.assertIn(":", part)