Revert "bpo-33608: Factor out a private, per-interpreter _Py_AddPendingCall(). (gh-13714)" (GH-13780)

This reverts commit 6a150bcaeb190d1731b38ab9c7a5d1a352847ddc.

4 files changed, 175 insertions, 285 deletions

diff --git a/Python/ceval.c b/Python/ceval.c
index cb0275c..0a4af91 100644
--- a/Python/ceval.c
+++ b/Python/ceval.c
@@ -115,65 +115,66 @@ static size_t opcache_global_hits = 0;
 static size_t opcache_global_misses = 0;
 #endif
 
+#define GIL_REQUEST _Py_atomic_load_relaxed(&ceval->gil_drop_request)
 
 /* This can set eval_breaker to 0 even though gil_drop_request became
    1.  We believe this is all right because the eval loop will release
    the GIL eventually anyway. */
-#define COMPUTE_EVAL_BREAKER(ceval_r, ceval_i) \
+#define COMPUTE_EVAL_BREAKER(ceval) \
     _Py_atomic_store_relaxed( \
-        &(ceval_r)->eval_breaker, \
-        _Py_atomic_load_relaxed(&(ceval_r)->gil_drop_request) | \
-        _Py_atomic_load_relaxed(&(ceval_r)->signals_pending) | \
-        _Py_atomic_load_relaxed(&(ceval_i)->pending.calls_to_do) | \
-        (ceval_i)->pending.async_exc)
+        &(ceval)->eval_breaker, \
+        GIL_REQUEST | \
+        _Py_atomic_load_relaxed(&(ceval)->signals_pending) | \
+        _Py_atomic_load_relaxed(&(ceval)->pending.calls_to_do) | \
+        (ceval)->pending.async_exc)
 
-#define SET_GIL_DROP_REQUEST(ceval_r) \
+#define SET_GIL_DROP_REQUEST(ceval) \
     do { \
-        _Py_atomic_store_relaxed(&(ceval_r)->gil_drop_request, 1); \
-        _Py_atomic_store_relaxed(&(ceval_r)->eval_breaker, 1); \
+        _Py_atomic_store_relaxed(&(ceval)->gil_drop_request, 1); \
+        _Py_atomic_store_relaxed(&(ceval)->eval_breaker, 1); \
     } while (0)
 
-#define RESET_GIL_DROP_REQUEST(ceval_r, ceval_i) \
+#define RESET_GIL_DROP_REQUEST(ceval) \
     do { \
-        _Py_atomic_store_relaxed(&(ceval_r)->gil_drop_request, 0); \
-        COMPUTE_EVAL_BREAKER(ceval_r, ceval_i); \
+        _Py_atomic_store_relaxed(&(ceval)->gil_drop_request, 0); \
+        COMPUTE_EVAL_BREAKER(ceval); \
     } while (0)
 
 /* Pending calls are only modified under pending_lock */
-#define SIGNAL_PENDING_CALLS(ceval_r, ceval_i) \
+#define SIGNAL_PENDING_CALLS(ceval) \
     do { \
-        _Py_atomic_store_relaxed(&(ceval_i)->pending.calls_to_do, 1); \
-        _Py_atomic_store_relaxed(&(ceval_r)->eval_breaker, 1); \
+        _Py_atomic_store_relaxed(&(ceval)->pending.calls_to_do, 1); \
+        _Py_atomic_store_relaxed(&(ceval)->eval_breaker, 1); \
     } while (0)
 
-#define UNSIGNAL_PENDING_CALLS(ceval_r, ceval_i) \
+#define UNSIGNAL_PENDING_CALLS(ceval) \
     do { \
-        _Py_atomic_store_relaxed(&(ceval_i)->pending.calls_to_do, 0); \
-        COMPUTE_EVAL_BREAKER(ceval_r, ceval_i); \
+        _Py_atomic_store_relaxed(&(ceval)->pending.calls_to_do, 0); \
+        COMPUTE_EVAL_BREAKER(ceval); \
     } while (0)
 
-#define SIGNAL_PENDING_SIGNALS(ceval_r) \
+#define SIGNAL_PENDING_SIGNALS(ceval) \
     do { \
-        _Py_atomic_store_relaxed(&(ceval_r)->signals_pending, 1); \
-        _Py_atomic_store_relaxed(&(ceval_r)->eval_breaker, 1); \
+        _Py_atomic_store_relaxed(&(ceval)->signals_pending, 1); \
+        _Py_atomic_store_relaxed(&(ceval)->eval_breaker, 1); \
     } while (0)
 
-#define UNSIGNAL_PENDING_SIGNALS(ceval_r, ceval_i) \
+#define UNSIGNAL_PENDING_SIGNALS(ceval) \
     do { \
-        _Py_atomic_store_relaxed(&(ceval_r)->signals_pending, 0); \
-        COMPUTE_EVAL_BREAKER(ceval_r, ceval_i); \
+        _Py_atomic_store_relaxed(&(ceval)->signals_pending, 0); \
+        COMPUTE_EVAL_BREAKER(ceval); \
     } while (0)
 
-#define SIGNAL_ASYNC_EXC(ceval_r, ceval_i) \
+#define SIGNAL_ASYNC_EXC(ceval) \
     do { \
-        (ceval_i)->pending.async_exc = 1; \
-        _Py_atomic_store_relaxed(&(ceval_r)->eval_breaker, 1); \
+        (ceval)->pending.async_exc = 1; \
+        _Py_atomic_store_relaxed(&(ceval)->eval_breaker, 1); \
     } while (0)
 
-#define UNSIGNAL_ASYNC_EXC(ceval_r, ceval_i) \
+#define UNSIGNAL_ASYNC_EXC(ceval) \
     do { \
-        (ceval_i)->pending.async_exc = 0; \
-        COMPUTE_EVAL_BREAKER(ceval_r, ceval_i); \
+        (ceval)->pending.async_exc = 0; \
+        COMPUTE_EVAL_BREAKER(ceval); \
     } while (0)
 
 
@@ -193,8 +194,8 @@ void
 PyEval_InitThreads(void)
 {
     _PyRuntimeState *runtime = &_PyRuntime;
-    struct _ceval_runtime_state *ceval_r = &runtime->ceval;
-    struct _gil_runtime_state *gil = &ceval_r->gil;
+    struct _ceval_runtime_state *ceval = &runtime->ceval;
+    struct _gil_runtime_state *gil = &ceval->gil;
     if (gil_created(gil)) {
         return;
     }
@@ -202,15 +203,19 @@ PyEval_InitThreads(void)
     PyThread_init_thread();
     create_gil(gil);
     PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
-    take_gil(ceval_r, tstate);
+    take_gil(ceval, tstate);
 
-    // The pending calls mutex is initialized in PyInterpreterState_New().
+    struct _pending_calls *pending = &ceval->pending;
+    pending->lock = PyThread_allocate_lock();
+    if (pending->lock == NULL) {
+        Py_FatalError("Can't initialize threads for pending calls");
+    }
 }
 
 void
-_PyEval_FiniThreads(struct _ceval_runtime_state *ceval_r)
+_PyEval_FiniThreads(struct _ceval_runtime_state *ceval)
 {
-    struct _gil_runtime_state *gil = &ceval_r->gil;
+    struct _gil_runtime_state *gil = &ceval->gil;
     if (!gil_created(gil)) {
         return;
     }
@@ -218,24 +223,20 @@ _PyEval_FiniThreads(struct _ceval_runtime_state *ceval_r)
     destroy_gil(gil);
     assert(!gil_created(gil));
 
-    // The pending calls mutex is freed in PyInterpreterState_Delete().
+    struct _pending_calls *pending = &ceval->pending;
+    if (pending->lock != NULL) {
+        PyThread_free_lock(pending->lock);
+        pending->lock = NULL;
+    }
 }
 
 static inline void
 exit_thread_if_finalizing(PyThreadState *tstate)
 {
-    PyInterpreterState *interp = tstate->interp;
-    // Stop if thread/interpreter inalization already stated.
-    if (interp == NULL) {
-        return;
-    }
-    _PyRuntimeState *runtime = interp->runtime;
-    if (runtime == NULL) {
-        return;
-    }
-    // Don't exit if the main thread (i.e. of the main interpreter).
+    _PyRuntimeState *runtime = tstate->interp->runtime;
+    /* _Py_Finalizing is protected by the GIL */
     if (runtime->finalizing != NULL && !_Py_CURRENTLY_FINALIZING(runtime, tstate)) {
-        drop_gil(&runtime->ceval, &interp->ceval, tstate);
+        drop_gil(&runtime->ceval, tstate);
         PyThread_exit_thread();
     }
 }
@@ -273,12 +274,12 @@ void
 PyEval_AcquireLock(void)
 {
     _PyRuntimeState *runtime = &_PyRuntime;
-    struct _ceval_runtime_state *ceval_r = &runtime->ceval;
+    struct _ceval_runtime_state *ceval = &runtime->ceval;
     PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
     if (tstate == NULL) {
         Py_FatalError("PyEval_AcquireLock: current thread state is NULL");
     }
-    take_gil(ceval_r, tstate);
+    take_gil(ceval, tstate);
     exit_thread_if_finalizing(tstate);
 }
 
@@ -286,21 +287,12 @@ void
 PyEval_ReleaseLock(void)
 {
     _PyRuntimeState *runtime = &_PyRuntime;
+    PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
     /* This function must succeed when the current thread state is NULL.
        We therefore avoid PyThreadState_Get() which dumps a fatal error
        in debug mode.
     */
-    PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
-    // Fall back to the main interpreter if there is not active Python
-    // thread.  This only affects the eval_breaker.
-    PyInterpreterState *interp = runtime->interpreters.main;
-    if (tstate != NULL) {
-        interp = tstate->interp;
-        if (interp == NULL) {
-            Py_FatalError("PyEval_ReleaseLock: NULL interpreter state");
-        }
-    }
-    drop_gil(&runtime->ceval, &interp->ceval, tstate);
+    drop_gil(&runtime->ceval, tstate);
 }
 
 void
@@ -309,19 +301,14 @@ PyEval_AcquireThread(PyThreadState *tstate)
     if (tstate == NULL) {
         Py_FatalError("PyEval_AcquireThread: NULL new thread state");
     }
-    PyInterpreterState *interp = tstate->interp;
-    if (interp == NULL) {
-        Py_FatalError("PyEval_AcquireThread: NULL interpreter state");
-    }
-    _PyRuntimeState *runtime = interp->runtime;
-    if (runtime == NULL) {
-        Py_FatalError("PyEval_AcquireThread: NULL runtime state");
-    }
-    struct _ceval_runtime_state *ceval_r = &runtime->ceval;
+    assert(tstate->interp != NULL);
+
+    _PyRuntimeState *runtime = tstate->interp->runtime;
+    struct _ceval_runtime_state *ceval = &runtime->ceval;
 
     /* Check someone has called PyEval_InitThreads() to create the lock */
-    assert(gil_created(&ceval_r->gil));
-    take_gil(ceval_r, tstate);
+    assert(gil_created(&ceval->gil));
+    take_gil(ceval, tstate);
     exit_thread_if_finalizing(tstate);
     if (_PyThreadState_Swap(&runtime->gilstate, tstate) != NULL) {
         Py_FatalError("PyEval_AcquireThread: non-NULL old thread state");
@@ -334,20 +321,14 @@ PyEval_ReleaseThread(PyThreadState *tstate)
     if (tstate == NULL) {
         Py_FatalError("PyEval_ReleaseThread: NULL thread state");
     }
-    PyInterpreterState *interp = tstate->interp;
-    if (interp == NULL) {
-        Py_FatalError("PyEval_ReleaseThread: NULL interpreter state");
-    }
-    _PyRuntimeState *runtime = interp->runtime;
-    if (runtime == NULL) {
-        Py_FatalError("PyEval_ReleaseThread: NULL runtime state");
-    }
+    assert(tstate->interp != NULL);
 
+    _PyRuntimeState *runtime = tstate->interp->runtime;
     PyThreadState *new_tstate = _PyThreadState_Swap(&runtime->gilstate, NULL);
     if (new_tstate != tstate) {
         Py_FatalError("PyEval_ReleaseThread: wrong thread state");
     }
-    drop_gil(&runtime->ceval, &interp->ceval, tstate);
+    drop_gil(&runtime->ceval, tstate);
 }
 
 /* This function is called from PyOS_AfterFork_Child to destroy all threads
@@ -358,17 +339,15 @@ PyEval_ReleaseThread(PyThreadState *tstate)
 void
 _PyEval_ReInitThreads(_PyRuntimeState *runtime)
 {
-    struct _ceval_runtime_state *ceval_r = &runtime->ceval;
-    if (!gil_created(&ceval_r->gil)) {
+    struct _ceval_runtime_state *ceval = &runtime->ceval;
+    if (!gil_created(&ceval->gil)) {
         return;
     }
-    recreate_gil(&ceval_r->gil);
+    recreate_gil(&ceval->gil);
     PyThreadState *current_tstate = _PyRuntimeState_GetThreadState(runtime);
-    take_gil(ceval_r, current_tstate);
+    take_gil(ceval, current_tstate);
 
-    // Only the main interpreter remains, so ignore the rest.
-    PyInterpreterState *interp = _PyRuntime.interpreters.main;
-    struct _ceval_pending_calls *pending = &interp->ceval.pending;
+    struct _pending_calls *pending = &ceval->pending;
     pending->lock = PyThread_allocate_lock();
     if (pending->lock == NULL) {
         Py_FatalError("Can't initialize threads for pending calls");
@@ -382,28 +361,22 @@ _PyEval_ReInitThreads(_PyRuntimeState *runtime)
    raised. */
 
 void
-_PyEval_SignalAsyncExc(struct _ceval_runtime_state *ceval_r,
-                       struct _ceval_interpreter_state *ceval_i)
+_PyEval_SignalAsyncExc(struct _ceval_runtime_state *ceval)
 {
-    SIGNAL_ASYNC_EXC(ceval_r, ceval_i);
+    SIGNAL_ASYNC_EXC(ceval);
 }
 
 PyThreadState *
 PyEval_SaveThread(void)
 {
     _PyRuntimeState *runtime = &_PyRuntime;
-    struct _ceval_runtime_state *ceval_r = &runtime->ceval;
+    struct _ceval_runtime_state *ceval = &runtime->ceval;
     PyThreadState *tstate = _PyThreadState_Swap(&runtime->gilstate, NULL);
     if (tstate == NULL) {
         Py_FatalError("PyEval_SaveThread: NULL tstate");
     }
-    PyInterpreterState *interp = tstate->interp;
-    if (interp == NULL) {
-        Py_FatalError("PyEval_SaveThread: NULL interpreter state");
-    }
-
-    assert(gil_created(&ceval_r->gil));
-    drop_gil(ceval_r, &interp->ceval, tstate);
+    assert(gil_created(&ceval->gil));
+    drop_gil(ceval, tstate);
     return tstate;
 }
 
@@ -413,20 +386,14 @@ PyEval_RestoreThread(PyThreadState *tstate)
     if (tstate == NULL) {
         Py_FatalError("PyEval_RestoreThread: NULL tstate");
     }
-    PyInterpreterState *interp = tstate->interp;
-    if (interp == NULL) {
-        Py_FatalError("PyEval_RestoreThread: NULL interpreter state");
-    }
-    _PyRuntimeState *runtime = interp->runtime;
-    if (runtime == NULL) {
-        Py_FatalError("PyEval_RestoreThread: NULL runtime state");
-    }
-    struct _ceval_runtime_state *ceval_r = &runtime->ceval;
+    assert(tstate->interp != NULL);
 
-    assert(gil_created(&ceval_r->gil));
+    _PyRuntimeState *runtime = tstate->interp->runtime;
+    struct _ceval_runtime_state *ceval = &runtime->ceval;
+    assert(gil_created(&ceval->gil));
 
     int err = errno;
-    take_gil(ceval_r, tstate);
+    take_gil(ceval, tstate);
     exit_thread_if_finalizing(tstate);
     errno = err;
 
@@ -457,17 +424,17 @@ PyEval_RestoreThread(PyThreadState *tstate)
 */
 
 void
-_PyEval_SignalReceived(struct _ceval_runtime_state *ceval_r)
+_PyEval_SignalReceived(struct _ceval_runtime_state *ceval)
 {
     /* bpo-30703: Function called when the C signal handler of Python gets a
        signal. We cannot queue a callback using Py_AddPendingCall() since
        that function is not async-signal-safe. */
-    SIGNAL_PENDING_SIGNALS(ceval_r);
+    SIGNAL_PENDING_SIGNALS(ceval);
 }
 
 /* Push one item onto the queue while holding the lock. */
 static int
-_push_pending_call(struct _ceval_pending_calls *pending, unsigned long thread_id,
+_push_pending_call(struct _pending_calls *pending,
                    int (*func)(void *), void *arg)
 {
     int i = pending->last;
@@ -475,7 +442,6 @@ _push_pending_call(struct _ceval_pending_calls *pending, unsigned long thread_id
     if (j == pending->first) {
         return -1; /* Queue full */
     }
-    pending->calls[i].thread_id = thread_id;
     pending->calls[i].func = func;
     pending->calls[i].arg = arg;
     pending->last = j;
@@ -484,7 +450,7 @@ _push_pending_call(struct _ceval_pending_calls *pending, unsigned long thread_id
 
 /* Pop one item off the queue while holding the lock. */
 static void
-_pop_pending_call(struct _ceval_pending_calls *pending, unsigned long *thread_id,
+_pop_pending_call(struct _pending_calls *pending,
                   int (**func)(void *), void **arg)
 {
     int i = pending->first;
@@ -494,7 +460,6 @@ _pop_pending_call(struct _ceval_pending_calls *pending, unsigned long *thread_id
 
     *func = pending->calls[i].func;
     *arg = pending->calls[i].arg;
-    *thread_id = pending->calls[i].thread_id;
     pending->first = (i + 1) % NPENDINGCALLS;
 }
 
@@ -505,12 +470,10 @@ _pop_pending_call(struct _ceval_pending_calls *pending, unsigned long *thread_id
 
 int
 _PyEval_AddPendingCall(PyThreadState *tstate,
-                       struct _ceval_runtime_state *ceval_r,
-                       struct _ceval_interpreter_state *ceval_i,
-                       unsigned long thread_id,
+                       struct _ceval_runtime_state *ceval,
                        int (*func)(void *), void *arg)
 {
-    struct _ceval_pending_calls *pending = &ceval_i->pending;
+    struct _pending_calls *pending = &ceval->pending;
 
     PyThread_acquire_lock(pending->lock, WAIT_LOCK);
     if (pending->finishing) {
@@ -525,27 +488,20 @@ _PyEval_AddPendingCall(PyThreadState *tstate,
         _PyErr_Restore(tstate, exc, val, tb);
         return -1;
     }
-    int result = _push_pending_call(pending, thread_id, func, arg);
-
-    /* signal loop */
-    SIGNAL_PENDING_CALLS(ceval_r, ceval_i);
+    int result = _push_pending_call(pending, func, arg);
     PyThread_release_lock(pending->lock);
 
+    /* signal main loop */
+    SIGNAL_PENDING_CALLS(ceval);
     return result;
 }
 
-/* Py_AddPendingCall() is a simple wrapper for the sake
-   of backward-compatibility. */
 int
 Py_AddPendingCall(int (*func)(void *), void *arg)
 {
     _PyRuntimeState *runtime = &_PyRuntime;
-    PyInterpreterState *interp = runtime->interpreters.main;
     PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
-    return _PyEval_AddPendingCall(tstate,
-                                  &runtime->ceval, &interp->ceval,
-                                  runtime->main_thread,
-                                  func, arg);
+    return _PyEval_AddPendingCall(tstate, &runtime->ceval, func, arg);
 }
 
 static int
@@ -566,69 +522,47 @@ handle_signals(_PyRuntimeState *runtime)
         return 0;
     }
 
-    struct _ceval_runtime_state *ceval_r = &runtime->ceval;
-    struct _ceval_interpreter_state *ceval_i = &interp->ceval;
-    UNSIGNAL_PENDING_SIGNALS(ceval_r, ceval_i);
+    struct _ceval_runtime_state *ceval = &runtime->ceval;
+    UNSIGNAL_PENDING_SIGNALS(ceval);
     if (_PyErr_CheckSignals() < 0) {
-        SIGNAL_PENDING_SIGNALS(ceval_r); /* We're not done yet */
+        SIGNAL_PENDING_SIGNALS(ceval); /* We're not done yet */
         return -1;
     }
     return 0;
 }
 
 static int
-make_pending_calls(PyInterpreterState *interp)
+make_pending_calls(_PyRuntimeState *runtime)
 {
-    if (interp == NULL) {
-        Py_FatalError("make_pending_calls: NULL interpreter state");
-    }
-    _PyRuntimeState *runtime = interp->runtime;
-    if (runtime == NULL) {
-        Py_FatalError("make_pending_calls: NULL runtime state");
-    }
-    PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
-    if (tstate == NULL) {
-        Py_FatalError("make_pending_calls: NULL thread state");
-    }
-    if (tstate->interp == NULL || tstate->interp != interp) {
-        Py_FatalError("make_pending_calls: thread state mismatch");
-    }
     static int busy = 0;
 
+    /* only service pending calls on main thread */
+    if (PyThread_get_thread_ident() != runtime->main_thread) {
+        return 0;
+    }
+
     /* don't perform recursive pending calls */
     if (busy) {
         return 0;
     }
     busy = 1;
-    struct _ceval_runtime_state *ceval_r = &runtime->ceval;
-    struct _ceval_interpreter_state *ceval_i = &interp->ceval;
+    struct _ceval_runtime_state *ceval = &runtime->ceval;
     /* unsignal before starting to call callbacks, so that any callback
        added in-between re-signals */
-    UNSIGNAL_PENDING_CALLS(ceval_r, ceval_i);
+    UNSIGNAL_PENDING_CALLS(ceval);
     int res = 0;
 
     /* perform a bounded number of calls, in case of recursion */
-    struct _ceval_pending_calls *pending = &ceval_i->pending;
-    unsigned long thread_id = 0;
+    struct _pending_calls *pending = &ceval->pending;
     for (int i=0; i<NPENDINGCALLS; i++) {
         int (*func)(void *) = NULL;
         void *arg = NULL;
 
         /* pop one item off the queue while holding the lock */
         PyThread_acquire_lock(pending->lock, WAIT_LOCK);
-        _pop_pending_call(pending, &thread_id, &func, &arg);
+        _pop_pending_call(pending, &func, &arg);
         PyThread_release_lock(pending->lock);
 
-        if (thread_id && PyThread_get_thread_ident() != thread_id) {
-            // Thread mismatch, so move it to the end of the list
-            // and start over.
-            _PyEval_AddPendingCall(tstate,
-                                   &runtime->ceval, &interp->ceval,
-                                   thread_id,
-                                   func, arg);
-            goto error;
-        }
-
         /* having released the lock, perform the callback */
         if (func == NULL) {
             break;
@@ -644,16 +578,17 @@ make_pending_calls(PyInterpreterState *interp)
 
 error:
     busy = 0;
-    SIGNAL_PENDING_CALLS(ceval_r, ceval_i);
+    SIGNAL_PENDING_CALLS(ceval);
     return res;
 }
 
 void
-_PyEval_FinishPendingCalls(PyInterpreterState *interp)
+_Py_FinishPendingCalls(_PyRuntimeState *runtime)
 {
     assert(PyGILState_Check());
 
-    struct _ceval_pending_calls *pending = &interp->ceval.pending;
+    PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
+    struct _pending_calls *pending = &runtime->ceval.pending;
 
     PyThread_acquire_lock(pending->lock, WAIT_LOCK);
     pending->finishing = 1;
@@ -663,19 +598,12 @@ _PyEval_FinishPendingCalls(PyInterpreterState *interp)
         return;
     }
 
-    if (make_pending_calls(interp) < 0) {
-        _PyRuntimeState *runtime = interp->runtime;
-        if (runtime == NULL) {
-            runtime = &_PyRuntime;
-        }
-        PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
-        if (tstate != NULL) {
-            PyObject *exc, *val, *tb;
-            _PyErr_Fetch(tstate, &exc, &val, &tb);
-            PyErr_BadInternalCall();
-            _PyErr_ChainExceptions(exc, val, tb);
-            _PyErr_Print(tstate);
-        }
+    if (make_pending_calls(runtime) < 0) {
+        PyObject *exc, *val, *tb;
+        _PyErr_Fetch(tstate, &exc, &val, &tb);
+        PyErr_BadInternalCall();
+        _PyErr_ChainExceptions(exc, val, tb);
+        _PyErr_Print(tstate);
     }
 }
 
@@ -694,8 +622,7 @@ Py_MakePendingCalls(void)
         return res;
     }
 
-    PyInterpreterState *interp = _PyRuntime.interpreters.main;
-    res = make_pending_calls(interp);
+    res = make_pending_calls(runtime);
     if (res != 0) {
         return res;
     }
@@ -712,11 +639,11 @@ Py_MakePendingCalls(void)
 int _Py_CheckRecursionLimit = Py_DEFAULT_RECURSION_LIMIT;
 
 void
-_PyEval_Initialize(struct _ceval_runtime_state *ceval_r)
+_PyEval_Initialize(struct _ceval_runtime_state *state)
 {
-    ceval_r->recursion_limit = Py_DEFAULT_RECURSION_LIMIT;
+    state->recursion_limit = Py_DEFAULT_RECURSION_LIMIT;
     _Py_CheckRecursionLimit = Py_DEFAULT_RECURSION_LIMIT;
-    _gil_initialize(&ceval_r->gil);
+    _gil_initialize(&state->gil);
 }
 
 int
@@ -728,9 +655,9 @@ Py_GetRecursionLimit(void)
 void
 Py_SetRecursionLimit(int new_limit)
 {
-    struct _ceval_runtime_state *ceval_r = &_PyRuntime.ceval;
-    ceval_r->recursion_limit = new_limit;
-    _Py_CheckRecursionLimit = ceval_r->recursion_limit;
+    struct _ceval_runtime_state *ceval = &_PyRuntime.ceval;
+    ceval->recursion_limit = new_limit;
+    _Py_CheckRecursionLimit = ceval->recursion_limit;
 }
 
 /* the macro Py_EnterRecursiveCall() only calls _Py_CheckRecursiveCall()
@@ -779,7 +706,7 @@ _Py_CheckRecursiveCall(const char *where)
 static int do_raise(PyThreadState *tstate, PyObject *exc, PyObject *cause);
 static int unpack_iterable(PyThreadState *, PyObject *, int, int, PyObject **);
 
-#define _Py_TracingPossible(ceval_r) ((ceval_r)->tracing_possible)
+#define _Py_TracingPossible(ceval) ((ceval)->tracing_possible)
 
 
 PyObject *
@@ -825,10 +752,8 @@ _PyEval_EvalFrameDefault(PyFrameObject *f, int throwflag)
     PyObject *retval = NULL;            /* Return value */
     _PyRuntimeState * const runtime = &_PyRuntime;
     PyThreadState * const tstate = _PyRuntimeState_GetThreadState(runtime);
-    PyInterpreterState * const interp = tstate->interp;
-    struct _ceval_runtime_state * const ceval_r = &runtime->ceval;
-    struct _ceval_interpreter_state * const ceval_i = &interp->ceval;
-    _Py_atomic_int * const eval_breaker = &ceval_r->eval_breaker;
+    struct _ceval_runtime_state * const ceval = &runtime->ceval;
+    _Py_atomic_int * const eval_breaker = &ceval->eval_breaker;
     PyCodeObject *co;
 
     /* when tracing we set things up so that
@@ -916,7 +841,7 @@ _PyEval_EvalFrameDefault(PyFrameObject *f, int throwflag)
 #ifdef LLTRACE
 #define FAST_DISPATCH() \
     { \
-        if (!lltrace && !_Py_TracingPossible(ceval_r) && !PyDTrace_LINE_ENABLED()) { \
+        if (!lltrace && !_Py_TracingPossible(ceval) && !PyDTrace_LINE_ENABLED()) { \
             f->f_lasti = INSTR_OFFSET(); \
             NEXTOPARG(); \
             goto *opcode_targets[opcode]; \
@@ -926,7 +851,7 @@ _PyEval_EvalFrameDefault(PyFrameObject *f, int throwflag)
 #else
 #define FAST_DISPATCH() \
     { \
-        if (!_Py_TracingPossible(ceval_r) && !PyDTrace_LINE_ENABLED()) { \
+        if (!_Py_TracingPossible(ceval) && !PyDTrace_LINE_ENABLED()) { \
             f->f_lasti = INSTR_OFFSET(); \
             NEXTOPARG(); \
             goto *opcode_targets[opcode]; \
@@ -1295,27 +1220,27 @@ main_loop:
                 goto fast_next_opcode;
             }
 
-            if (_Py_atomic_load_relaxed(&ceval_r->signals_pending)) {
+            if (_Py_atomic_load_relaxed(&ceval->signals_pending)) {
                 if (handle_signals(runtime) != 0) {
                     goto error;
                 }
             }
-            if (_Py_atomic_load_relaxed(&ceval_i->pending.calls_to_do)) {
-                if (make_pending_calls(interp) != 0) {
+            if (_Py_atomic_load_relaxed(&ceval->pending.calls_to_do)) {
+                if (make_pending_calls(runtime) != 0) {
                     goto error;
                 }
             }
 
-            if (_Py_atomic_load_relaxed(&ceval_r->gil_drop_request)) {
+            if (_Py_atomic_load_relaxed(&ceval->gil_drop_request)) {
                 /* Give another thread a chance */
                 if (_PyThreadState_Swap(&runtime->gilstate, NULL) != tstate) {
                     Py_FatalError("ceval: tstate mix-up");
                 }
-                drop_gil(ceval_r, ceval_i, tstate);
+                drop_gil(ceval, tstate);
 
                 /* Other threads may run now */
 
-                take_gil(ceval_r, tstate);
+                take_gil(ceval, tstate);
 
                 /* Check if we should make a quick exit. */
                 exit_thread_if_finalizing(tstate);
@@ -1328,7 +1253,7 @@ main_loop:
             if (tstate->async_exc != NULL) {
                 PyObject *exc = tstate->async_exc;
                 tstate->async_exc = NULL;
-                UNSIGNAL_ASYNC_EXC(ceval_r, ceval_i);
+                UNSIGNAL_ASYNC_EXC(ceval);
                 _PyErr_SetNone(tstate, exc);
                 Py_DECREF(exc);
                 goto error;
@@ -1343,7 +1268,7 @@ main_loop:
 
         /* line-by-line tracing support */
 
-        if (_Py_TracingPossible(ceval_r) &&
+        if (_Py_TracingPossible(ceval) &&
             tstate->c_tracefunc != NULL && !tstate->tracing) {
             int err;
             /* see maybe_call_line_trace
diff --git a/Python/ceval_gil.h b/Python/ceval_gil.h
index b44d0ab..34d48c9 100644
--- a/Python/ceval_gil.h
+++ b/Python/ceval_gil.h
@@ -141,11 +141,9 @@ static void recreate_gil(struct _gil_runtime_state *gil)
 }
 
 static void
-drop_gil(struct _ceval_runtime_state *ceval_r,
-         struct _ceval_interpreter_state *ceval_i,
-         PyThreadState *tstate)
+drop_gil(struct _ceval_runtime_state *ceval, PyThreadState *tstate)
 {
-    struct _gil_runtime_state *gil = &ceval_r->gil;
+    struct _gil_runtime_state *gil = &ceval->gil;
     if (!_Py_atomic_load_relaxed(&gil->locked)) {
         Py_FatalError("drop_gil: GIL is not locked");
     }
@@ -165,12 +163,12 @@ drop_gil(struct _ceval_runtime_state *ceval_r,
     MUTEX_UNLOCK(gil->mutex);
 
 #ifdef FORCE_SWITCHING
-    if (_Py_atomic_load_relaxed(&ceval_r->gil_drop_request) && tstate != NULL) {
+    if (_Py_atomic_load_relaxed(&ceval->gil_drop_request) && tstate != NULL) {
         MUTEX_LOCK(gil->switch_mutex);
         /* Not switched yet => wait */
         if (((PyThreadState*)_Py_atomic_load_relaxed(&gil->last_holder)) == tstate)
         {
-            RESET_GIL_DROP_REQUEST(ceval_r, ceval_i);
+            RESET_GIL_DROP_REQUEST(ceval);
             /* NOTE: if COND_WAIT does not atomically start waiting when
                releasing the mutex, another thread can run through, take
                the GIL and drop it again, and reset the condition
@@ -183,19 +181,13 @@ drop_gil(struct _ceval_runtime_state *ceval_r,
 }
 
 static void
-take_gil(struct _ceval_runtime_state *ceval_r,
-         PyThreadState *tstate)
+take_gil(struct _ceval_runtime_state *ceval, PyThreadState *tstate)
 {
     if (tstate == NULL) {
         Py_FatalError("take_gil: NULL tstate");
     }
-    PyInterpreterState *interp = tstate->interp;
-    if (interp == NULL) {
-        Py_FatalError("take_gil: NULL interp");
-    }
-    struct _ceval_interpreter_state *ceval_i = &interp->ceval;
 
-    struct _gil_runtime_state *gil = &ceval_r->gil;
+    struct _gil_runtime_state *gil = &ceval->gil;
     int err = errno;
     MUTEX_LOCK(gil->mutex);
 
@@ -218,7 +210,7 @@ take_gil(struct _ceval_runtime_state *ceval_r,
             _Py_atomic_load_relaxed(&gil->locked) &&
             gil->switch_number == saved_switchnum)
         {
-            SET_GIL_DROP_REQUEST(ceval_r);
+            SET_GIL_DROP_REQUEST(ceval);
         }
     }
 _ready:
@@ -240,11 +232,11 @@ _ready:
     COND_SIGNAL(gil->switch_cond);
     MUTEX_UNLOCK(gil->switch_mutex);
 #endif
-    if (_Py_atomic_load_relaxed(&ceval_r->gil_drop_request)) {
-        RESET_GIL_DROP_REQUEST(ceval_r, ceval_i);
+    if (_Py_atomic_load_relaxed(&ceval->gil_drop_request)) {
+        RESET_GIL_DROP_REQUEST(ceval);
     }
     if (tstate->async_exc != NULL) {
-        _PyEval_SignalAsyncExc(ceval_r, ceval_i);
+        _PyEval_SignalAsyncExc(ceval);
     }
 
     MUTEX_UNLOCK(gil->mutex);
diff --git a/Python/pylifecycle.c b/Python/pylifecycle.c
index fc7e551..fca2ee6 100644
--- a/Python/pylifecycle.c
+++ b/Python/pylifecycle.c
@@ -1147,31 +1147,15 @@ Py_FinalizeEx(void)
         return status;
     }
 
-    /* Get current thread state and interpreter pointer */
-    PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
-    PyInterpreterState *interp = tstate->interp;
-
     // Wrap up existing "threading"-module-created, non-daemon threads.
     wait_for_thread_shutdown();
 
     // Make any remaining pending calls.
-    /* XXX For the moment we are going to ignore lingering pending calls.
-     * We've seen sporadic on some of the buildbots during finalization
-     * with the changes for per-interpreter pending calls (see bpo-33608),
-     * meaning the previous _PyEval_FinishPendincCalls() call here is
-     * a trigger, if not responsible.
-     *
-     * Ignoring pending calls at this point in the runtime lifecycle
-     * is okay (for now) for the following reasons:
-     *
-     *  * pending calls are still not a widely-used feature
-     *  * this only affects runtime finalization, where the process is
-     *    likely to end soon anyway (except for some embdding cases)
-     *
-     * See bpo-37127 about resolving the problem.  Ultimately the call
-     * here should be re-enabled.
-     */
-    //_PyEval_FinishPendingCalls(interp);
+    _Py_FinishPendingCalls(runtime);
+
+    /* Get current thread state and interpreter pointer */
+    PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
+    PyInterpreterState *interp = tstate->interp;
 
     /* The interpreter is still entirely intact at this point, and the
      * exit funcs may be relying on that.  In particular, if some thread
@@ -1599,9 +1583,6 @@ Py_EndInterpreter(PyThreadState *tstate)
     // Wrap up existing "threading"-module-created, non-daemon threads.
     wait_for_thread_shutdown();
 
-    // Make any remaining pending calls.
-    _PyEval_FinishPendingCalls(interp);
-
     call_py_exitfuncs(interp);
 
     if (tstate != interp->tstate_head || tstate->next != NULL)
diff --git a/Python/pystate.c b/Python/pystate.c
index a9f3389..2b7db0e 100644
--- a/Python/pystate.c
+++ b/Python/pystate.c
@@ -218,13 +218,6 @@ PyInterpreterState_New(void)
         return NULL;
     }
 
-    interp->ceval.pending.lock = PyThread_allocate_lock();
-    if (interp->ceval.pending.lock == NULL) {
-        PyErr_SetString(PyExc_RuntimeError,
-                        "failed to create interpreter ceval pending mutex");
-        return NULL;
-    }
-
     interp->eval_frame = _PyEval_EvalFrameDefault;
 #ifdef HAVE_DLOPEN
 #if HAVE_DECL_RTLD_NOW
@@ -352,10 +345,6 @@ PyInterpreterState_Delete(PyInterpreterState *interp)
     if (interp->id_mutex != NULL) {
         PyThread_free_lock(interp->id_mutex);
     }
-    if (interp->ceval.pending.lock != NULL) {
-        PyThread_free_lock(interp->ceval.pending.lock);
-        interp->ceval.pending.lock = NULL;
-    }
     PyMem_RawFree(interp);
 }
 
@@ -1025,7 +1014,7 @@ PyThreadState_SetAsyncExc(unsigned long id, PyObject *exc)
             p->async_exc = exc;
             HEAD_UNLOCK(runtime);
             Py_XDECREF(old_exc);
-            _PyEval_SignalAsyncExc(&runtime->ceval, &interp->ceval);
+            _PyEval_SignalAsyncExc(&runtime->ceval);
             return 1;
         }
     }
@@ -1455,7 +1444,7 @@ _PyObject_GetCrossInterpreterData(PyObject *obj, _PyCrossInterpreterData *data)
     return 0;
 }
 
-static int
+static void
 _release_xidata(void *arg)
 {
     _PyCrossInterpreterData *data = (_PyCrossInterpreterData *)arg;
@@ -1463,21 +1452,42 @@ _release_xidata(void *arg)
         data->free(data->data);
     }
     Py_XDECREF(data->obj);
-    PyMem_Free(data);
-    return 0;
+}
+
+static void
+_call_in_interpreter(struct _gilstate_runtime_state *gilstate,
+                     PyInterpreterState *interp,
+                     void (*func)(void *), void *arg)
+{
+    /* We would use Py_AddPendingCall() if it weren't specific to the
+     * main interpreter (see bpo-33608).  In the meantime we take a
+     * naive approach.
+     */
+    PyThreadState *save_tstate = NULL;
+    if (interp != _PyRuntimeGILState_GetThreadState(gilstate)->interp) {
+        // XXX Using the "head" thread isn't strictly correct.
+        PyThreadState *tstate = PyInterpreterState_ThreadHead(interp);
+        // XXX Possible GILState issues?
+        save_tstate = _PyThreadState_Swap(gilstate, tstate);
+    }
+
+    func(arg);
+
+    // Switch back.
+    if (save_tstate != NULL) {
+        _PyThreadState_Swap(gilstate, save_tstate);
+    }
 }
 
 void
 _PyCrossInterpreterData_Release(_PyCrossInterpreterData *data)
 {
-    _PyRuntimeState *runtime = &_PyRuntime;
-
     if (data->data == NULL && data->obj == NULL) {
         // Nothing to release!
         return;
     }
 
-    // Get the original interpreter.
+    // Switch to the original interpreter.
     PyInterpreterState *interp = _PyInterpreterState_LookUpID(data->interp);
     if (interp == NULL) {
         // The intepreter was already destroyed.
@@ -1486,28 +1496,10 @@ _PyCrossInterpreterData_Release(_PyCrossInterpreterData *data)
         }
         return;
     }
-    // XXX There's an ever-so-slight race here...
-    if (interp->finalizing) {
-        // XXX Someone leaked some memory...
-        return;
-    }
 
     // "Release" the data and/or the object.
-    _PyCrossInterpreterData *copied = PyMem_Malloc(sizeof(_PyCrossInterpreterData));
-    if (copied == NULL) {
-        PyErr_SetString(PyExc_MemoryError,
-                        "Not enough memory to preserve cross-interpreter data");
-        PyErr_Print();
-        return;
-    }
-    memcpy(copied, data, sizeof(_PyCrossInterpreterData));
-    PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
-    int res = _PyEval_AddPendingCall(tstate,
-                                     &runtime->ceval, &interp->ceval,
-                                     0, _release_xidata, copied);
-    if (res != 0) {
-        // XXX Queue full or couldn't get lock.  Try again somehow?
-    }
+    struct _gilstate_runtime_state *gilstate = &_PyRuntime.gilstate;
+    _call_in_interpreter(gilstate, interp, _release_xidata, data);
 }
 
 PyObject *