[3.13] gh-117657: Fix TSAN race involving import lock (GH-118523) (#120169)

This adds a `_PyRecursiveMutex` type based on `PyMutex` and uses that
for the import lock. This fixes some data races in the free-threaded
build and generally simplifies the import lock code.
(cherry picked from commit e21057b99967eb5323320e6d1121955e0cd2985e)

Co-authored-by: Sam Gross <colesbury@gmail.com>

2 files changed, 14 insertions, 16 deletions

diff --git a/Include/internal/pycore_import.h b/Include/internal/pycore_import.h
index bd40707..3806e0d 100644
--- a/Include/internal/pycore_import.h
+++ b/Include/internal/pycore_import.h
@@ -20,7 +20,7 @@ PyAPI_FUNC(int) _PyImport_SetModule(PyObject *name, PyObject *module);
 extern int _PyImport_SetModuleString(const char *name, PyObject* module);
 
 extern void _PyImport_AcquireLock(PyInterpreterState *interp);
-extern int _PyImport_ReleaseLock(PyInterpreterState *interp);
+extern void _PyImport_ReleaseLock(PyInterpreterState *interp);
 
 // This is used exclusively for the sys and builtins modules:
 extern int _PyImport_FixupBuiltin(
@@ -94,11 +94,7 @@ struct _import_state {
 #endif
     PyObject *import_func;
     /* The global import lock. */
-    struct {
-        PyThread_type_lock mutex;
-        unsigned long thread;
-        int level;
-    } lock;
+    _PyRecursiveMutex lock;
     /* diagnostic info in PyImport_ImportModuleLevelObject() */
     struct {
         int import_level;
@@ -123,11 +119,6 @@ struct _import_state {
 #define IMPORTS_INIT \
     { \
         DLOPENFLAGS_INIT \
-        .lock = { \
-            .mutex = NULL, \
-            .thread = PYTHREAD_INVALID_THREAD_ID, \
-            .level = 0, \
-        }, \
         .find_and_load = { \
             .header = 1, \
         }, \
@@ -180,11 +171,6 @@ extern void _PyImport_FiniCore(PyInterpreterState *interp);
 extern void _PyImport_FiniExternal(PyInterpreterState *interp);
 
 
-#ifdef HAVE_FORK
-extern PyStatus _PyImport_ReInitLock(PyInterpreterState *interp);
-#endif
-
-
 extern PyObject* _PyImport_GetBuiltinModuleNames(void);
 
 struct _module_alias {
diff --git a/Include/internal/pycore_lock.h b/Include/internal/pycore_lock.h
index a5b28e4..d5853b2 100644
--- a/Include/internal/pycore_lock.h
+++ b/Include/internal/pycore_lock.h
@@ -219,6 +219,18 @@ _PyOnceFlag_CallOnce(_PyOnceFlag *flag, _Py_once_fn_t *fn, void *arg)
     return _PyOnceFlag_CallOnceSlow(flag, fn, arg);
 }
 
+// A recursive mutex. The mutex should zero-initialized.
+typedef struct {
+    PyMutex mutex;
+    unsigned long long thread;  // i.e., PyThread_get_thread_ident_ex()
+    size_t level;
+} _PyRecursiveMutex;
+
+PyAPI_FUNC(int) _PyRecursiveMutex_IsLockedByCurrentThread(_PyRecursiveMutex *m);
+PyAPI_FUNC(void) _PyRecursiveMutex_Lock(_PyRecursiveMutex *m);
+PyAPI_FUNC(void) _PyRecursiveMutex_Unlock(_PyRecursiveMutex *m);
+
+
 // A readers-writer (RW) lock. The lock supports multiple concurrent readers or
 // a single writer. The lock is write-preferring: if a writer is waiting while
 // the lock is read-locked then, new readers will be blocked. This avoids