bpo-33608: Factor out a private, per-interpreter _Py_AddPendingCall(). (gh-12360)

This is effectively an un-revert of #11617 and #12024 (reverted in #12159). Portions of those were merged in other PRs (with lower risk) and this represents the remainder. Note that I found 3 different bugs in the original PRs and have fixed them here.

3 files changed, 17 insertions, 6 deletions

diff --git a/Include/ceval.h b/Include/ceval.h
index 11283c0..9c6d420 100644
--- a/Include/ceval.h
+++ b/Include/ceval.h
@@ -221,7 +221,7 @@ PyAPI_FUNC(Py_ssize_t) _PyEval_RequestCodeExtraIndex(freefunc);
 #ifndef Py_LIMITED_API
 PyAPI_FUNC(int) _PyEval_SliceIndex(PyObject *, Py_ssize_t *);
 PyAPI_FUNC(int) _PyEval_SliceIndexNotNone(PyObject *, Py_ssize_t *);
-PyAPI_FUNC(void) _PyEval_SignalAsyncExc(void);
+PyAPI_FUNC(void) _PyEval_SignalAsyncExc(PyInterpreterState *);
 #endif
 
 /* Masks and values used by FORMAT_VALUE opcode. */
diff --git a/Include/internal/pycore_ceval.h b/Include/internal/pycore_ceval.h
index 2ead96c..1bdcdf5 100644
--- a/Include/internal/pycore_ceval.h
+++ b/Include/internal/pycore_ceval.h
@@ -11,7 +11,11 @@ extern "C" {
 #include "pycore_atomic.h"
 #include "pythread.h"
 
-PyAPI_FUNC(void) _Py_FinishPendingCalls(void);
+struct _is;  // See PyInterpreterState in cpython/pystate.h.
+
+PyAPI_FUNC(int) _Py_AddPendingCall(struct _is*, unsigned long, int (*)(void *), void *);
+PyAPI_FUNC(int) _Py_MakePendingCalls(struct _is*);
+PyAPI_FUNC(void) _Py_FinishPendingCalls(struct _is*);
 
 struct _pending_calls {
     int finishing;
@@ -24,6 +28,7 @@ struct _pending_calls {
     int async_exc;
 #define NPENDINGCALLS 32
     struct {
+        unsigned long thread_id;
         int (*func)(void *);
         void *arg;
     } calls[NPENDINGCALLS];
@@ -31,6 +36,13 @@ struct _pending_calls {
     int last;
 };
 
+struct _ceval_interpreter_state {
+    /* This single variable consolidates all requests to break out of
+       the fast path in the eval loop. */
+    _Py_atomic_int eval_breaker;
+    struct _pending_calls pending;
+};
+
 #include "pycore_gil.h"
 
 struct _ceval_runtime_state {
@@ -41,12 +53,8 @@ struct _ceval_runtime_state {
        c_tracefunc.  This speeds up the if statement in
        PyEval_EvalFrameEx() after fast_next_opcode. */
     int tracing_possible;
-    /* This single variable consolidates all requests to break out of
-       the fast path in the eval loop. */
-    _Py_atomic_int eval_breaker;
     /* Request for dropping the GIL */
     _Py_atomic_int gil_drop_request;
-    struct _pending_calls pending;
     /* Request for checking signals. */
     _Py_atomic_int signals_pending;
     struct _gil_runtime_state gil;
diff --git a/Include/internal/pycore_pystate.h b/Include/internal/pycore_pystate.h
index df3730f..3ae2e0c 100644
--- a/Include/internal/pycore_pystate.h
+++ b/Include/internal/pycore_pystate.h
@@ -12,6 +12,7 @@ extern "C" {
 #include "pystate.h"
 #include "pythread.h"
 
+#include "pycore_atomic.h"
 #include "pycore_ceval.h"
 #include "pycore_pathconfig.h"
 #include "pycore_pymem.h"
@@ -83,6 +84,8 @@ struct _is {
     PyObject *pyexitmodule;
 
     uint64_t tstate_next_unique_id;
+
+    struct _ceval_interpreter_state ceval;
 };
 
 PyAPI_FUNC(struct _is*) _PyInterpreterState_LookUpID(PY_INT64_T);