diff --git a/Modules/_xxinterpchannelsmodule.c b/Modules/_xxinterpchannelsmodule.c
index fead12c963da26..ef1cdcab952271 100644
--- a/Modules/_xxinterpchannelsmodule.c
+++ b/Modules/_xxinterpchannelsmodule.c
@@ -10,9 +10,77 @@
#include "pycore_interpreteridobject.h"
+/*
+This module has the following process-global state:
+
+_globals (static struct globals):
+ module_count (int)
+ channels (struct _channels):
+ numopen (int64_t)
+ next_id; (int64_t)
+ mutex (PyThread_type_lock)
+ head (linked list of struct _channelref *):
+ id (int64_t)
+ objcount (Py_ssize_t)
+ next (struct _channelref *):
+ ...
+ chan (struct _channel *):
+ open (int)
+ mutex (PyThread_type_lock)
+ closing (struct _channel_closing *):
+ ref (struct _channelref *):
+ ...
+ ends (struct _channelends *):
+ numsendopen (int64_t)
+ numrecvopen (int64_t)
+ send (struct _channelend *):
+ interp (int64_t)
+ open (int)
+ next (struct _channelend *)
+ recv (struct _channelend *):
+ ...
+ queue (struct _channelqueue *):
+ count (int64_t)
+ first (struct _channelitem *):
+ next (struct _channelitem *):
+ ...
+ data (_PyCrossInterpreterData *):
+ data (void *)
+ obj (PyObject *)
+ interp (int64_t)
+ new_object (xid_newobjectfunc)
+ free (xid_freefunc)
+ last (struct _channelitem *):
+ ...
+
+The above state includes the following allocations by the module:
+
+* 1 top-level mutex (to protect the rest of the state)
+* for each channel:
+ * 1 struct _channelref
+ * 1 struct _channel
+ * 0-1 struct _channel_closing
+ * 1 struct _channelends
+ * 2 struct _channelend
+ * 1 struct _channelqueue
+* for each item in each channel:
+ * 1 struct _channelitem
+ * 1 _PyCrossInterpreterData
+
+The only objects in that global state are the references held by each
+channel's queue, which are safely managed via the _PyCrossInterpreterData_*()
+API.. The module does not create any objects that are shared globally.
+*/
+
#define MODULE_NAME "_xxinterpchannels"
+#define GLOBAL_MALLOC(TYPE) \
+ PyMem_RawMalloc(sizeof(TYPE))
+#define GLOBAL_FREE(VAR) \
+ PyMem_RawFree(VAR)
+
+
static PyInterpreterState *
_get_current_interp(void)
{
@@ -301,7 +369,7 @@ typedef struct _channelitem {
static _channelitem *
_channelitem_new(void)
{
- _channelitem *item = PyMem_NEW(_channelitem, 1);
+ _channelitem *item = GLOBAL_MALLOC(_channelitem);
if (item == NULL) {
PyErr_NoMemory();
return NULL;
@@ -316,7 +384,8 @@ _channelitem_clear(_channelitem *item)
{
if (item->data != NULL) {
(void)_release_xid_data(item->data, 1);
- PyMem_Free(item->data);
+ // It was allocated in _channel_send().
+ GLOBAL_FREE(item->data);
item->data = NULL;
}
item->next = NULL;
@@ -326,7 +395,7 @@ static void
_channelitem_free(_channelitem *item)
{
_channelitem_clear(item);
- PyMem_Free(item);
+ GLOBAL_FREE(item);
}
static void
@@ -357,7 +426,7 @@ typedef struct _channelqueue {
static _channelqueue *
_channelqueue_new(void)
{
- _channelqueue *queue = PyMem_NEW(_channelqueue, 1);
+ _channelqueue *queue = GLOBAL_MALLOC(_channelqueue);
if (queue == NULL) {
PyErr_NoMemory();
return NULL;
@@ -381,7 +450,7 @@ static void
_channelqueue_free(_channelqueue *queue)
{
_channelqueue_clear(queue);
- PyMem_Free(queue);
+ GLOBAL_FREE(queue);
}
static int
@@ -433,7 +502,7 @@ typedef struct _channelend {
static _channelend *
_channelend_new(int64_t interp)
{
- _channelend *end = PyMem_NEW(_channelend, 1);
+ _channelend *end = GLOBAL_MALLOC(_channelend);
if (end == NULL) {
PyErr_NoMemory();
return NULL;
@@ -447,7 +516,7 @@ _channelend_new(int64_t interp)
static void
_channelend_free(_channelend *end)
{
- PyMem_Free(end);
+ GLOBAL_FREE(end);
}
static void
@@ -492,7 +561,7 @@ typedef struct _channelassociations {
static _channelends *
_channelends_new(void)
{
- _channelends *ends = PyMem_NEW(_channelends, 1);
+ _channelends *ends = GLOBAL_MALLOC(_channelends);
if (ends== NULL) {
return NULL;
}
@@ -519,7 +588,7 @@ static void
_channelends_free(_channelends *ends)
{
_channelends_clear(ends);
- PyMem_Free(ends);
+ GLOBAL_FREE(ends);
}
static _channelend *
@@ -660,20 +729,20 @@ typedef struct _channel {
static _PyChannelState *
_channel_new(PyThread_type_lock mutex)
{
- _PyChannelState *chan = PyMem_NEW(_PyChannelState, 1);
+ _PyChannelState *chan = GLOBAL_MALLOC(_PyChannelState);
if (chan == NULL) {
return NULL;
}
chan->mutex = mutex;
chan->queue = _channelqueue_new();
if (chan->queue == NULL) {
- PyMem_Free(chan);
+ GLOBAL_FREE(chan);
return NULL;
}
chan->ends = _channelends_new();
if (chan->ends == NULL) {
_channelqueue_free(chan->queue);
- PyMem_Free(chan);
+ GLOBAL_FREE(chan);
return NULL;
}
chan->open = 1;
@@ -691,7 +760,7 @@ _channel_free(_PyChannelState *chan)
PyThread_release_lock(chan->mutex);
PyThread_free_lock(chan->mutex);
- PyMem_Free(chan);
+ GLOBAL_FREE(chan);
}
static int
@@ -814,7 +883,7 @@ typedef struct _channelref {
static _channelref *
_channelref_new(int64_t id, _PyChannelState *chan)
{
- _channelref *ref = PyMem_NEW(_channelref, 1);
+ _channelref *ref = GLOBAL_MALLOC(_channelref);
if (ref == NULL) {
return NULL;
}
@@ -841,7 +910,7 @@ _channelref_free(_channelref *ref)
_channel_clear_closing(ref->chan);
}
//_channelref_clear(ref);
- PyMem_Free(ref);
+ GLOBAL_FREE(ref);
}
static _channelref *
@@ -1163,7 +1232,7 @@ _channel_set_closing(struct _channelref *ref, PyThread_type_lock mutex) {
res = ERR_CHANNEL_CLOSED;
goto done;
}
- chan->closing = PyMem_NEW(struct _channel_closing, 1);
+ chan->closing = GLOBAL_MALLOC(struct _channel_closing);
if (chan->closing == NULL) {
goto done;
}
@@ -1179,7 +1248,7 @@ static void
_channel_clear_closing(struct _channel *chan) {
PyThread_acquire_lock(chan->mutex, WAIT_LOCK);
if (chan->closing != NULL) {
- PyMem_Free(chan->closing);
+ GLOBAL_FREE(chan->closing);
chan->closing = NULL;
}
PyThread_release_lock(chan->mutex);
@@ -1257,14 +1326,14 @@ _channel_send(_channels *channels, int64_t id, PyObject *obj)
}
// Convert the object to cross-interpreter data.
- _PyCrossInterpreterData *data = PyMem_NEW(_PyCrossInterpreterData, 1);
+ _PyCrossInterpreterData *data = GLOBAL_MALLOC(_PyCrossInterpreterData);
if (data == NULL) {
PyThread_release_lock(mutex);
return -1;
}
if (_PyObject_GetCrossInterpreterData(obj, data) != 0) {
PyThread_release_lock(mutex);
- PyMem_Free(data);
+ GLOBAL_FREE(data);
return -1;
}
@@ -1274,7 +1343,7 @@ _channel_send(_channels *channels, int64_t id, PyObject *obj)
if (res != 0) {
// We may chain an exception here:
(void)_release_xid_data(data, 0);
- PyMem_Free(data);
+ GLOBAL_FREE(data);
return res;
}
@@ -1323,11 +1392,13 @@ _channel_recv(_channels *channels, int64_t id, PyObject **res)
if (obj == NULL) {
assert(PyErr_Occurred());
(void)_release_xid_data(data, 1);
- PyMem_Free(data);
+ // It was allocated in _channel_send().
+ GLOBAL_FREE(data);
return -1;
}
int release_res = _release_xid_data(data, 0);
- PyMem_Free(data);
+ // It was allocated in _channel_send().
+ GLOBAL_FREE(data);
if (release_res < 0) {
// The source interpreter has been destroyed already.
assert(PyErr_Occurred());
pFad - Phonifier reborn
Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies:
Alternative Proxy
pFad Proxy
pFad v3 Proxy
pFad v4 Proxy