merge main

Signed-off-by: Manjusaka <me@manjusaka.me>

Author: Zheaoli

Doc/library/os.rst

@@ -1556,6 +1556,15 @@ or `the MSDN <https://msdn.microsoft.com/en-us/library/z0kc8e3z.aspx>`_ on Windo
    .. versionadded:: 3.15
 
 
+.. data:: RWF_ATOMIC
+
+   Write data atomically. Requires alignment to the device's atomic write unit.
+
+   .. availability:: Linux >= 6.11
+
+   .. versionadded:: next
+
+
 .. function:: ptsname(fd, /)
 
    Return the name of the slave pseudo-terminal device associated with the
@@ -1598,6 +1607,7 @@ or `the MSDN <https://msdn.microsoft.com/en-us/library/z0kc8e3z.aspx>`_ on Windo
    - :data:`RWF_SYNC`
    - :data:`RWF_APPEND`
    - :data:`RWF_DONTCACHE`
+   - :data:`RWF_ATOMIC`
 
    Return the total number of bytes actually written.
 

Doc/whatsnew/3.15.rst

@@ -891,7 +891,11 @@ 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
+interpreter built with all optimizations enabled on x86-64 Linux. On AArch64
+macOS, the JIT has a
+`7-8% <https://raw.githubusercontent.com/facebookexperimental/free-threading-benchmarking/refs/heads/main/results/bm-20260103-3.15.0a3%2B-9609574-JIT/bm-20260103-macm4pro-arm64-python-9609574e7fd36edfaa8b-3.15.0a3%2B-9609574-vs-base.svg>`__
+speedup over the :ref:`tail calling interpreter <whatsnew314-tail-call-interpreter>`
+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.

Include/internal/pycore_opcode_metadata.h

@@ -2952,6 +2952,25 @@ def f(n):
         self.assertLessEqual(count_ops(ex, "_POP_TOP"), 3)
         self.assertIn("_POP_TOP_NOP", uops)
 
+    def test_to_bool_always_true(self):
+        def testfunc(n):
+            class A:
+                pass
+
+            a = A()
+            for _ in range(n):
+                if not a:
+                    return 0
+            return 1
+
+        res, ex = self._run_with_optimizer(testfunc, TIER2_THRESHOLD)
+        self.assertEqual(res, 1)
+        self.assertIsNotNone(ex)
+        uops = get_opnames(ex)
+        self.assertNotIn("_REPLACE_WITH_TRUE", uops)
+        self.assertIn("_INSERT_1_LOAD_CONST_INLINE_BORROW", uops)
+        self.assertEqual(count_ops(ex, "_POP_TOP_NOP"), 1)
+
     def test_attr_promotion_failure(self):
         # We're not testing for any specific uops here, just
         # testing it doesn't crash.