[pull] main from python:main

Doc/library/asyncio-queue.rst

@@ -107,7 +107,7 @@ Queue
       The queue can no longer grow.
       Future calls to :meth:`~Queue.put` raise :exc:`QueueShutDown`.
       Currently blocked callers of :meth:`~Queue.put` will be unblocked
-      and will raise :exc:`QueueShutDown` in the formerly blocked thread.
+      and will raise :exc:`QueueShutDown` in the formerly awaiting task.
 
       If *immediate* is false (the default), the queue can be wound
       down normally with :meth:`~Queue.get` calls to extract tasks

Doc/whatsnew/3.15.rst

@@ -73,6 +73,7 @@ Summary -- Release highlights
   <whatsnew315-utf8-default>`
 * :pep:`782`: :ref:`A new PyBytesWriter C API to create a Python bytes object
   <whatsnew315-pep782>`
+* :ref:`The JIT compiler has been significantly upgraded <whatsnew315-jit>`
 * :ref:`Improved error messages <whatsnew315-improved-error-messages>`
 
 
@@ -850,6 +851,91 @@ csv
   (Contributed by Maurycy Pawłowski-Wieroński in :gh:`137628`.)
 
 
+.. _whatsnew315-jit:
+
+Upgraded JIT compiler
+=====================
+
+Results from the `pyperformance <https://github.com/python/pyperformance>`__
+benchmark suite report
+`3-4% <https://github.com/facebookexperimental/free-threading-benchmarking/blob/main/results/bm-20251214-3.15.0a2%2B-6cddf04-JIT/bm-20251214-vultr-x86_64-python-6cddf04344a1e8ca9df5-3.15.0a2%2B-6cddf04-vs-base.svg>`__
+geometric mean performance improvement for the JIT over the standard CPython
+interpreter built with all optimizations enabled. The speedups for JIT
+builds versus no JIT builds range from roughly 20% slowdown to over
+100% speedup (ignoring the ``unpack_sequence`` microbenchmark) on
+x86-64 Linux and AArch64 macOS systems.
+
+.. attention::
+    These results are not yet final.
+
+The major upgrades to the JIT are:
+
+* LLVM 21 build-time dependency
+* New tracing frontend
+* Basic register allocation in the JIT
+* More JIT optimizations
+* Better machine code generation
+
+.. rubric:: LLVM 21 build-time dependency
+
+The JIT compiler now uses LLVM 21 for build-time stencil generation. As
+always, LLVM is only needed when building CPython with the JIT enabled;
+end users running Python do not need LLVM installed. Instructions for
+installing LLVM can be found in the `JIT compiler documentation
+<https://github.com/python/cpython/blob/main/Tools/jit/README.md>`__
+for all supported platforms.
+
+(Contributed by Savannah Ostrowski in :gh:`140973`.)
+
+.. rubric:: A new tracing frontend
+
+The JIT compiler now supports significantly more bytecode operations and
+control flow than in Python 3.14, enabling speedups on a wider variety of
+code. For example, simple Python object creation is now understood by the
+3.15 JIT compiler. Overloaded operations and generators are also partially
+supported. This was made possible by an overhauled JIT tracing frontend
+that records actual execution paths through code, rather than estimating
+them as the previous implementation did.
+
+(Contributed by Ken Jin in :gh:`139109`. Support for Windows added by
+Mark Shannon in :gh:`141703`.)
+
+.. rubric:: Basic register allocation in the JIT
+
+A basic form of register allocation has been added to the JIT compiler's
+optimizer. This allows the JIT compiler to avoid certain stack operations
+altogether and instead operate on registers. This allows the JIT to produce
+more efficient traces by avoiding reads and writes to memory.
+
+(Contributed by Mark Shannon in :gh:`135379`.)
+
+.. rubric:: More JIT optimizations
+
+More `constant-propagation <https://en.wikipedia.org/wiki/Constant_folding>`__
+is now performed. This means when the JIT compiler detects that certain user
+code results in constants, the code can be simplified by the JIT.
+
+(Contributed by Ken Jin and Savannah Ostrowski in :gh:`132732`.)
+
+The JIT avoids :term:`reference count`\ s where possible. This generally
+reduces the cost of most operations in Python.
+
+(Contributed by Ken Jin, Donghee Na, Zheao Li, Savannah Ostrowski,
+Noam Cohen, Tomas Roun, PuQing in :gh:`134584`.)
+
+.. rubric:: Better machine code generation
+
+The JIT compiler's machine code generator now produces better machine code
+for x86-64 and AArch64 macOS and Linux targets. In general, users should
+experience lower memory usage for generated machine code and more efficient
+machine code versus the old JIT.
+
+(Contributed by Brandt Bucher in :gh:`136528` and :gh:`136528`.
+Implementation for AArch64 contributed by Mark Shannon in :gh:`139855`.
+Additional optimizations for AArch64 contributed by Mark Shannon and
+Diego Russo in :gh:`140683` and :gh:`142305`.)
+
+
 Removed
 =======
 

Lib/profiling/sampling/__main__.py

@@ -46,6 +46,7 @@
 """
 
 from .cli import main
+from .errors import SamplingUnknownProcessError, SamplingModuleNotFoundError, SamplingScriptNotFoundError
 
 def handle_permission_error():
     """Handle PermissionError by displaying appropriate error message."""
@@ -64,3 +65,9 @@ def handle_permission_error():
         main()
     except PermissionError:
         handle_permission_error()
+    except SamplingUnknownProcessError as err:
+        print(f"Tachyon cannot find the process: {err}", file=sys.stderr)
+        sys.exit(1)
+    except (SamplingModuleNotFoundError, SamplingScriptNotFoundError) as err:
+        print(f"Tachyon cannot find the target: {err}", file=sys.stderr)
+        sys.exit(1)

Lib/profiling/sampling/cli.py

@@ -10,7 +10,8 @@
 import time
 from contextlib import nullcontext
 
-from .sample import sample, sample_live
+from .errors import SamplingUnknownProcessError, SamplingModuleNotFoundError, SamplingScriptNotFoundError
+from .sample import sample, sample_live, _is_process_running
 from .pstats_collector import PstatsCollector
 from .stack_collector import CollapsedStackCollector, FlamegraphCollector
 from .heatmap_collector import HeatmapCollector
@@ -743,6 +744,8 @@ def main():
 
 def _handle_attach(args):
     """Handle the 'attach' command."""
+    if not _is_process_running(args.pid):
+        raise SamplingUnknownProcessError(args.pid)
     # Check if live mode is requested
     if args.live:
         _handle_live_attach(args, args.pid)
@@ -792,13 +795,13 @@ def _handle_run(args):
             added_cwd = True
         try:
             if importlib.util.find_spec(args.target) is None:
-                sys.exit(f"Error: Module not found: {args.target}")
+                raise SamplingModuleNotFoundError(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}")
+            raise SamplingScriptNotFoundError(args.target)
 
     # Check if live mode is requested
     if args.live:

Lib/profiling/sampling/errors.py

@@ -0,0 +1,19 @@
+"""Custom exceptions for the sampling profiler."""
+
+class SamplingProfilerError(Exception):
+    """Base exception for sampling profiler errors."""
+
+class SamplingUnknownProcessError(SamplingProfilerError):
+    def __init__(self, pid):
+        self.pid = pid
+        super().__init__(f"Process with PID '{pid}' does not exist.")
+
+class SamplingScriptNotFoundError(SamplingProfilerError):
+    def __init__(self, script_path):
+        self.script_path = script_path
+        super().__init__(f"Script '{script_path}' not found.")
+
+class SamplingModuleNotFoundError(SamplingProfilerError):
+    def __init__(self, module_name):
+        self.module_name = module_name
+        super().__init__(f"Module '{module_name}' not found.")

Lib/profiling/sampling/sample.py

@@ -34,23 +34,29 @@ def __init__(self, pid, sample_interval_usec, all_threads, *, mode=PROFILING_MOD
         self.all_threads = all_threads
         self.mode = mode  # Store mode for later use
         self.collect_stats = collect_stats
+        try:
+            self.unwinder = self._new_unwinder(native, gc, opcodes, skip_non_matching_threads)
+        except RuntimeError as err:
+            raise SystemExit(err) from err
+        # Track sample intervals and total sample count
+        self.sample_intervals = deque(maxlen=100)
+        self.total_samples = 0
+        self.realtime_stats = False
+
+    def _new_unwinder(self, native, gc, opcodes, skip_non_matching_threads):
         if _FREE_THREADED_BUILD:
-            self.unwinder = _remote_debugging.RemoteUnwinder(
-                self.pid, all_threads=self.all_threads, mode=mode, native=native, gc=gc,
+            unwinder = _remote_debugging.RemoteUnwinder(
+                self.pid, all_threads=self.all_threads, mode=self.mode, native=native, gc=gc,
                 opcodes=opcodes, skip_non_matching_threads=skip_non_matching_threads,
-                cache_frames=True, stats=collect_stats
+                cache_frames=True, stats=self.collect_stats
             )
         else:
-            only_active_threads = bool(self.all_threads)
-            self.unwinder = _remote_debugging.RemoteUnwinder(
-                self.pid, only_active_thread=only_active_threads, mode=mode, native=native, gc=gc,
+            unwinder = _remote_debugging.RemoteUnwinder(
+                self.pid, only_active_thread=bool(self.all_threads), mode=self.mode, native=native, gc=gc,
                 opcodes=opcodes, skip_non_matching_threads=skip_non_matching_threads,
-                cache_frames=True, stats=collect_stats
+                cache_frames=True, stats=self.collect_stats
             )
-        # Track sample intervals and total sample count
-        self.sample_intervals = deque(maxlen=100)
-        self.total_samples = 0
-        self.realtime_stats = False
+        return unwinder
 
     def sample(self, collector, duration_sec=10, *, async_aware=False):
         sample_interval_sec = self.sample_interval_usec / 1_000_000
@@ -86,7 +92,7 @@ def sample(self, collector, duration_sec=10, *, async_aware=False):
                         collector.collect_failed_sample()
                         errors += 1
                     except Exception as e:
-                        if not self._is_process_running():
+                        if not _is_process_running(self.pid):
                             break
                         raise e from None
 
@@ -148,22 +154,6 @@ def sample(self, collector, duration_sec=10, *, async_aware=False):
                 f"({(expected_samples - num_samples) / expected_samples * 100:.2f}%)"
             )
 
-    def _is_process_running(self):
-        if sys.platform == "linux" or sys.platform == "darwin":
-            try:
-                os.kill(self.pid, 0)
-                return True
-            except ProcessLookupError:
-                return False
-        elif sys.platform == "win32":
-            try:
-                _remote_debugging.RemoteUnwinder(self.pid)
-            except Exception:
-                return False
-            return True
-        else:
-            raise ValueError(f"Unsupported platform: {sys.platform}")
-
     def _print_realtime_stats(self):
         """Print real-time sampling statistics."""
         if len(self.sample_intervals) < 2:
@@ -279,6 +269,28 @@ def _print_unwinder_stats(self):
             print(f"  {ANSIColors.YELLOW}Stale cache invalidations: {stale_invalidations}{ANSIColors.RESET}")
 
 
+def _is_process_running(pid):
+    if pid <= 0:
+        return False
+    if os.name == "posix":
+        try:
+            os.kill(pid, 0)
+            return True
+        except ProcessLookupError:
+            return False
+        except PermissionError:
+            # EPERM means process exists but we can't signal it
+            return True
+    elif sys.platform == "win32":
+        try:
+            _remote_debugging.RemoteUnwinder(pid)
+        except Exception:
+            return False
+        return True
+    else:
+        raise ValueError(f"Unsupported platform: {sys.platform}")
+
+
 def sample(
     pid,
     collector,

Lib/test/_test_atexit.py

@@ -135,6 +135,19 @@ def func():
         finally:
             atexit.unregister(func)
 
+    def test_eq_unregister_clear(self):
+        # Issue #112127: callback's __eq__ may call unregister or _clear
+        class Evil:
+            def __eq__(self, other):
+                action(other)
+                return NotImplemented
+
+        for action in atexit.unregister, lambda o: atexit._clear():
+            with self.subTest(action=action):
+                atexit.register(lambda: None)
+                atexit.unregister(Evil())
+                atexit._clear()
+
 
 if __name__ == "__main__":
     unittest.main()