gh-135871: Fix needless spinning in _PyMutex_LockTimed (timeout==0)

jtibbertsma · jtibbertsma · commit 45811fd6631a · 2025-06-23T18:01:35.000-07:00
* Move the timeout == 0 guard outside the else so a non-blocking call
  returns immediately after a failed CAS instead of entering the spin loop.
* Reload v on every spin iteration, allowing timed/blocking callers to
  notice an unlock promptly.

No-GIL builds now honor the semantics of non-blocking attempts and avoid
wasted CPU; GIL builds are unaffected (MAX_SPIN_COUNT == 0).
diff --git a/Misc/NEWS.d/next/Core and Builtins/135871.mutex-spin.rst b/Misc/NEWS.d/next/Core and Builtins/135871.mutex-spin.rst
@@ -0,0 +1 @@
+bpo-135871: _PyMutex_LockTimed() no longer spins or fails for timeout==0 in no-GIL builds; the spin loop now reloads the lock state.
diff --git a/Python/lock.c b/Python/lock.c
@@ -58,7 +58,7 @@ _PyMutex_LockTimed(PyMutex *m, PyTime_t timeout, _PyLockFlags flags)
             return PY_LOCK_ACQUIRED;
         }
     }
-    else if (timeout == 0) {
+    if (timeout == 0) {
         return PY_LOCK_FAILURE;
     }
 
@@ -89,6 +89,7 @@ _PyMutex_LockTimed(PyMutex *m, PyTime_t timeout, _PyLockFlags flags)
             // Spin for a bit.
             _Py_yield();
             spin_count++;
+            v = _Py_atomic_load_uint8_relaxed(&m->_bits);
             continue;
         }
 

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1 @@`
	`1`	`+bpo-135871: _PyMutex_LockTimed() no longer spins or fails for timeout==0 in no-GIL builds; the spin loop now reloads the lock state.`
Original file line number	Diff line number	Diff line change
`@@ -58,7 +58,7 @@ _PyMutex_LockTimed(PyMutex *m, PyTime_t timeout, _PyLockFlags flags)`
`58`	`58`	`return PY_LOCK_ACQUIRED;`
`59`	`59`	`}`
`60`	`60`	`}`
`61`		`- else if (timeout == 0) {`
	`61`	`+ if (timeout == 0) {`
`62`	`62`	`return PY_LOCK_FAILURE;`
`63`	`63`	`}`
`64`	`64`
`@@ -89,6 +89,7 @@ _PyMutex_LockTimed(PyMutex *m, PyTime_t timeout, _PyLockFlags flags)`
`89`	`89`	`// Spin for a bit.`
`90`	`90`	`_Py_yield();`
`91`	`91`	`spin_count++;`
	`92`	`+ v = _Py_atomic_load_uint8_relaxed(&m->_bits);`
`92`	`93`	`continue;`
`93`	`94`	`}`
`94`	`95`