GH-142591: Tachyon does not handle non-existent file/module (#142592)

savannahostrowski · pablogsal · web-flow · commit f893e8f2568d · 2025-12-14T04:58:40.000Z
Co-authored-by: Pablo Galindo Salgado &lt;pablogsal@gmail.com&gt;
diff --git a/Lib/profiling/sampling/_sync_coordinator.py b/Lib/profiling/sampling/_sync_coordinator.py
@@ -5,6 +5,7 @@
 the startup of target processes. It should not be called directly by users.
 """
 
+import importlib.util
 import os
 import sys
 import socket
@@ -138,7 +139,7 @@ def _execute_module(module_name: str, module_args: List[str]) -> None:
     """
     # Replace sys.argv to match how Python normally runs modules
     # When running 'python -m module args', sys.argv is ["__main__.py", "args"]
-    sys.argv = [f"__main__.py"] + module_args
+    sys.argv = ["__main__.py"] + module_args
 
     try:
         runpy.run_module(module_name, run_name="__main__", alter_sys=True)
@@ -215,22 +216,31 @@ def main() -> NoReturn:
         # Set up execution environment
         _setup_environment(cwd)
 
-        # Signal readiness to profiler
-        _signal_readiness(sync_port)
-
-        # Execute the target
-        if target_args[0] == "-m":
-            # Module execution
+        # Determine execution type and validate target exists
+        is_module = target_args[0] == "-m"
+        if is_module:
             if len(target_args) < 2:
                 raise ArgumentError("Module name required after -m")
-
             module_name = target_args[1]
             module_args = target_args[2:]
-            _execute_module(module_name, module_args)
+
+            if importlib.util.find_spec(module_name) is None:
+                raise TargetError(f"Module not found: {module_name}")
         else:
-            # Script execution
             script_path = target_args[0]
             script_args = target_args[1:]
+            # Match the path resolution logic in _execute_script
+            check_path = script_path if os.path.isabs(script_path) else os.path.join(cwd, script_path)
+            if not os.path.isfile(check_path):
+                raise TargetError(f"Script not found: {script_path}")
+
+        # Signal readiness to profiler
+        _signal_readiness(sync_port)
+
+        # Execute the target
+        if is_module:
+            _execute_module(module_name, module_args)
+        else:
             _execute_script(script_path, script_args, cwd)
 
     except CoordinatorError as e:
diff --git a/Lib/profiling/sampling/cli.py b/Lib/profiling/sampling/cli.py
@@ -1,10 +1,13 @@
 """Command-line interface for the sampling profiler."""
 
 import argparse
+import importlib.util
 import os
+import selectors
 import socket
 import subprocess
 import sys
+import time
 
 from .sample import sample, sample_live
 from .pstats_collector import PstatsCollector
@@ -92,6 +95,54 @@ def _parse_mode(mode_string):
     return mode_map[mode_string]
 
 
+def _check_process_died(process):
+    """Check if process died and raise an error with stderr if available."""
+    if process.poll() is None:
+        return  # Process still running
+
+    # Process died - try to get stderr for error message
+    stderr_msg = ""
+    if process.stderr:
+        try:
+            stderr_msg = process.stderr.read().decode().strip()
+        except (OSError, UnicodeDecodeError):
+            pass
+
+    if stderr_msg:
+        raise RuntimeError(stderr_msg)
+    raise RuntimeError(f"Process exited with code {process.returncode}")
+
+
+def _wait_for_ready_signal(sync_sock, process, timeout):
+    """Wait for the ready signal from the subprocess, checking for early death."""
+    deadline = time.monotonic() + timeout
+    sel = selectors.DefaultSelector()
+    sel.register(sync_sock, selectors.EVENT_READ)
+
+    try:
+        while True:
+            _check_process_died(process)
+
+            remaining = deadline - time.monotonic()
+            if remaining <= 0:
+                raise socket.timeout("timed out")
+
+            if not sel.select(timeout=min(0.1, remaining)):
+                continue
+
+            conn, _ = sync_sock.accept()
+            try:
+                ready_signal = conn.recv(_RECV_BUFFER_SIZE)
+            finally:
+                conn.close()
+
+            if ready_signal != _READY_MESSAGE:
+                raise RuntimeError(f"Invalid ready signal received: {ready_signal!r}")
+            return
+    finally:
+        sel.close()
+
+
 def _run_with_sync(original_cmd, suppress_output=False):
     """Run a command with socket-based synchronization and return the process."""
     # Create a TCP socket for synchronization with better socket options
@@ -117,24 +168,24 @@ def _run_with_sync(original_cmd, suppress_output=False):
         ) + tuple(target_args)
 
         # Start the process with coordinator
-        # Suppress stdout/stderr if requested (for live mode)
+        # When suppress_output=True (live mode), capture stderr so we can
+        # report errors if the process dies before signaling ready.
+        # When suppress_output=False (normal mode), let stderr inherit so
+        # script errors print to the terminal.
         popen_kwargs = {}
         if suppress_output:
             popen_kwargs["stdin"] = subprocess.DEVNULL
             popen_kwargs["stdout"] = subprocess.DEVNULL
-            popen_kwargs["stderr"] = subprocess.DEVNULL
+            popen_kwargs["stderr"] = subprocess.PIPE
 
         process = subprocess.Popen(cmd, **popen_kwargs)
 
         try:
-            # Wait for ready signal with timeout
-            with sync_sock.accept()[0] as conn:
-                ready_signal = conn.recv(_RECV_BUFFER_SIZE)
+            _wait_for_ready_signal(sync_sock, process, _SYNC_TIMEOUT)
 
-                if ready_signal != _READY_MESSAGE:
-                    raise RuntimeError(
-                        f"Invalid ready signal received: {ready_signal!r}"
-                    )
+            # Close stderr pipe if we were capturing it
+            if process.stderr:
+                process.stderr.close()
 
         except socket.timeout:
             # If we timeout, kill the process and raise an error
@@ -632,6 +683,25 @@ def _handle_attach(args):
 
 def _handle_run(args):
     """Handle the 'run' command."""
+    # Validate target exists before launching subprocess
+    if args.module:
+        # Temporarily add cwd to sys.path so we can find modules in the
+        # current directory, matching the coordinator's behavior
+        cwd = os.getcwd()
+        added_cwd = False
+        if cwd not in sys.path:
+            sys.path.insert(0, cwd)
+            added_cwd = True
+        try:
+            if importlib.util.find_spec(args.target) is None:
+                sys.exit(f"Error: Module not found: {args.target}")
+        finally:
+            if added_cwd:
+                sys.path.remove(cwd)
+    else:
+        if not os.path.exists(args.target):
+            sys.exit(f"Error: Script not found: {args.target}")
+
     # Check if live mode is requested
     if args.live:
         _handle_live_run(args)
@@ -644,7 +714,10 @@ def _handle_run(args):
         cmd = (sys.executable, args.target, *args.args)
 
     # Run with synchronization
-    process = _run_with_sync(cmd, suppress_output=False)
+    try:
+        process = _run_with_sync(cmd, suppress_output=False)
+    except RuntimeError as e:
+        sys.exit(f"Error: {e}")
 
     # Use PROFILING_MODE_ALL for gecko format
     mode = (
@@ -732,7 +805,10 @@ def _handle_live_run(args):
         cmd = (sys.executable, args.target, *args.args)
 
     # Run with synchronization, suppressing output for live mode
-    process = _run_with_sync(cmd, suppress_output=True)
+    try:
+        process = _run_with_sync(cmd, suppress_output=True)
+    except RuntimeError as e:
+        sys.exit(f"Error: {e}")
 
     mode = _parse_mode(args.mode)
 
diff --git a/Lib/test/test_profiling/test_sampling_profiler/test_cli.py b/Lib/test/test_profiling/test_sampling_profiler/test_cli.py
