bpo-40609: Remove _Py_hashtable_t.key_size (GH-20060)

Rewrite _Py_hashtable_t type to always store the key as
a "const void *" pointer. Add an explicit "key" member to
_Py_hashtable_entry_t.

Remove _Py_hashtable_t.key_size member.

hash and compare functions drop their hash table parameter, and their
'key' parameter type becomes "const void *".

Include/internal/pycore_hashtable.h

@@ -30,32 +30,13 @@ typedef struct {
     _Py_slist_item_t _Py_slist_item;
 
     Py_uhash_t key_hash;
-
-    /* key (key_size bytes) and then data (data_size bytes) follows */
+    void *key;
+    /* data (data_size bytes) follows */
 } _Py_hashtable_entry_t;
 
-#define _Py_HASHTABLE_ENTRY_PKEY(ENTRY) \
-        ((const void *)((char *)(ENTRY) \
-                        + sizeof(_Py_hashtable_entry_t)))
-
 #define _Py_HASHTABLE_ENTRY_PDATA(TABLE, ENTRY) \
         ((const void *)((char *)(ENTRY) \
-                        + sizeof(_Py_hashtable_entry_t) \
-                        + (TABLE)->key_size))
-
-/* Get a key value from pkey: use memcpy() rather than a pointer dereference
-   to avoid memory alignment issues. */
-#define _Py_HASHTABLE_READ_KEY(TABLE, PKEY, DST_KEY) \
-    do { \
-        assert(sizeof(DST_KEY) == (TABLE)->key_size); \
-        memcpy(&(DST_KEY), (PKEY), sizeof(DST_KEY)); \
-    } while (0)
-
-#define _Py_HASHTABLE_ENTRY_READ_KEY(TABLE, ENTRY, KEY) \
-    do { \
-        assert(sizeof(KEY) == (TABLE)->key_size); \
-        memcpy(&(KEY), _Py_HASHTABLE_ENTRY_PKEY(ENTRY), sizeof(KEY)); \
-    } while (0)
+                        + sizeof(_Py_hashtable_entry_t)))
 
 #define _Py_HASHTABLE_ENTRY_READ_DATA(TABLE, ENTRY, DATA) \
     do { \
@@ -78,15 +59,12 @@ typedef struct {
 struct _Py_hashtable_t;
 typedef struct _Py_hashtable_t _Py_hashtable_t;
 
-typedef Py_uhash_t (*_Py_hashtable_hash_func) (_Py_hashtable_t *ht,
-                                               const void *pkey);
-typedef int (*_Py_hashtable_compare_func) (_Py_hashtable_t *ht,
-                                           const void *pkey,
-                                           const _Py_hashtable_entry_t *he);
+typedef Py_uhash_t (*_Py_hashtable_hash_func) (const void *key);
+typedef int (*_Py_hashtable_compare_func) (const void *key1, const void *key2);
 typedef _Py_hashtable_entry_t* (*_Py_hashtable_get_entry_func)(_Py_hashtable_t *ht,
-                                                               const void *pkey);
+                                                               const void *key);
 typedef int (*_Py_hashtable_get_func) (_Py_hashtable_t *ht,
-                                       const void *pkey, void *data);
+                                       const void *key, void *data);
 
 typedef struct {
     /* allocate a memory block */
@@ -102,7 +80,6 @@ struct _Py_hashtable_t {
     size_t num_buckets;
     size_t entries; /* Total number of entries in the table. */
     _Py_slist_t *buckets;
-    size_t key_size;
     size_t data_size;
 
     _Py_hashtable_get_func get_func;
@@ -113,24 +90,19 @@ struct _Py_hashtable_t {
 };
 
 /* hash a pointer (void*) */
-PyAPI_FUNC(Py_uhash_t) _Py_hashtable_hash_ptr(
-    struct _Py_hashtable_t *ht,
-    const void *pkey);
+PyAPI_FUNC(Py_uhash_t) _Py_hashtable_hash_ptr(const void *key);
 
 /* comparison using memcmp() */
 PyAPI_FUNC(int) _Py_hashtable_compare_direct(
-    _Py_hashtable_t *ht,
-    const void *pkey,
-    const _Py_hashtable_entry_t *entry);
+    const void *key1,
+    const void *key2);
 
 PyAPI_FUNC(_Py_hashtable_t *) _Py_hashtable_new(
-    size_t key_size,
     size_t data_size,
     _Py_hashtable_hash_func hash_func,
     _Py_hashtable_compare_func compare_func);
 
 PyAPI_FUNC(_Py_hashtable_t *) _Py_hashtable_new_full(
-    size_t key_size,
     size_t data_size,
     size_t init_size,
     _Py_hashtable_hash_func hash_func,
@@ -165,16 +137,15 @@ PyAPI_FUNC(size_t) _Py_hashtable_size(_Py_hashtable_t *ht);
    but use _Py_HASHTABLE_SET() and _Py_HASHTABLE_SET_NODATA() macros */
 PyAPI_FUNC(int) _Py_hashtable_set(
     _Py_hashtable_t *ht,
-    size_t key_size,
-    const void *pkey,
+    const void *key,
     size_t data_size,
     const void *data);
 
 #define _Py_HASHTABLE_SET(TABLE, KEY, DATA) \
-    _Py_hashtable_set(TABLE, sizeof(KEY), &(KEY), sizeof(DATA), &(DATA))
+    _Py_hashtable_set(TABLE, (KEY), sizeof(DATA), &(DATA))
 
 #define _Py_HASHTABLE_SET_NODATA(TABLE, KEY) \
-    _Py_hashtable_set(TABLE, sizeof(KEY), &(KEY), 0, NULL)
+    _Py_hashtable_set(TABLE, (KEY), 0, NULL)
 
 
 /* Get an entry.
@@ -183,14 +154,13 @@ PyAPI_FUNC(int) _Py_hashtable_set(
    Don't call directly this function, but use _Py_HASHTABLE_GET_ENTRY()
    macro */
 static inline _Py_hashtable_entry_t *
-_Py_hashtable_get_entry(_Py_hashtable_t *ht, size_t key_size, const void *pkey)
+_Py_hashtable_get_entry(_Py_hashtable_t *ht, const void *key)
 {
-    assert(key_size == ht->key_size);
-    return ht->get_entry_func(ht, pkey);
+    return ht->get_entry_func(ht, key);
 }
 
 #define _Py_HASHTABLE_GET_ENTRY(TABLE, KEY) \
-    _Py_hashtable_get_entry(TABLE, sizeof(KEY), &(KEY))
+    _Py_hashtable_get_entry(TABLE, (const void *)(KEY))
 
 
 /* Get data from an entry. Copy entry data into data and return 1 if the entry
@@ -198,28 +168,26 @@ _Py_hashtable_get_entry(_Py_hashtable_t *ht, size_t key_size, const void *pkey)
 
    Don't call directly this function, but use _Py_HASHTABLE_GET() macro */
 static inline int
-_Py_hashtable_get(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
+_Py_hashtable_get(_Py_hashtable_t *ht, const void *key,
                   size_t data_size, void *data)
 {
-    assert(key_size == ht->key_size);
     assert(data_size == ht->data_size);
-    return ht->get_func(ht, pkey, data);
+    return ht->get_func(ht, key, data);
 }
 
 #define _Py_HASHTABLE_GET(TABLE, KEY, DATA) \
-    _Py_hashtable_get(TABLE, sizeof(KEY), &(KEY), sizeof(DATA), &(DATA))
+    _Py_hashtable_get(TABLE, (KEY), sizeof(DATA), &(DATA))
 
 
 /* Don't call directly this function, but use _Py_HASHTABLE_POP() macro */
 PyAPI_FUNC(int) _Py_hashtable_pop(
     _Py_hashtable_t *ht,
-    size_t key_size,
-    const void *pkey,
+    const void *key,
     size_t data_size,
     void *data);
 
 #define _Py_HASHTABLE_POP(TABLE, KEY, DATA) \
-    _Py_hashtable_pop(TABLE, sizeof(KEY), &(KEY), sizeof(DATA), &(DATA))
+    _Py_hashtable_pop(TABLE, (KEY), sizeof(DATA), &(DATA))
 
 
 #ifdef __cplusplus

Modules/_tracemalloc.c

@@ -23,6 +23,9 @@ static void raw_free(void *ptr);
 #  define TRACE_DEBUG
 #endif
 
+#define TO_PTR(key) ((const void *)(uintptr_t)key)
+#define FROM_PTR(key) ((uintptr_t)key)
+
 /* Protected by the GIL */
 static struct {
     PyMemAllocatorEx mem;
@@ -203,47 +206,42 @@ set_reentrant(int reentrant)
 
 
 static Py_uhash_t
-hashtable_hash_pyobject(_Py_hashtable_t *ht, const void *pkey)
+hashtable_hash_pyobject(const void *key)
 {
-    PyObject *obj;
-
-    _Py_HASHTABLE_READ_KEY(ht, pkey, obj);
+    PyObject *obj = (PyObject *)key;
     return PyObject_Hash(obj);
 }
 
 
 static int
-hashtable_compare_unicode(_Py_hashtable_t *ht, const void *pkey,
-                          const _Py_hashtable_entry_t *entry)
+hashtable_compare_unicode(const void *key1, const void *key2)
 {
-    PyObject *key1, *key2;
-
-    _Py_HASHTABLE_READ_KEY(ht, pkey, key1);
-    _Py_HASHTABLE_ENTRY_READ_KEY(ht, entry, key2);
-
-    if (key1 != NULL && key2 != NULL)
-        return (PyUnicode_Compare(key1, key2) == 0);
-    else
-        return key1 == key2;
+    PyObject *obj1 = (PyObject *)key1;
+    PyObject *obj2 = (PyObject *)key2;
+    if (obj1 != NULL && obj2 != NULL) {
+        return (PyUnicode_Compare(obj1, obj2) == 0);
+    }
+    else {
+        return obj1 == obj2;
+    }
 }
 
 
 static Py_uhash_t
-hashtable_hash_uint(_Py_hashtable_t *ht, const void *pkey)
+hashtable_hash_uint(const void *key_raw)
 {
-    unsigned int key;
-    _Py_HASHTABLE_READ_KEY(ht, pkey, key);
+    unsigned int key = (unsigned int)FROM_PTR(key_raw);
     return (Py_uhash_t)key;
 }
 
 
 static _Py_hashtable_t *
-hashtable_new(size_t key_size, size_t data_size,
+hashtable_new(size_t data_size,
               _Py_hashtable_hash_func hash_func,
               _Py_hashtable_compare_func compare_func)
 {
     _Py_hashtable_allocator_t hashtable_alloc = {malloc, free};
-    return _Py_hashtable_new_full(key_size, data_size, 0,
+    return _Py_hashtable_new_full(data_size, 0,
                                   hash_func, compare_func,
                                   &hashtable_alloc);
 }
@@ -263,39 +261,33 @@ raw_free(void *ptr)
 
 
 static Py_uhash_t
-hashtable_hash_traceback(_Py_hashtable_t *ht, const void *pkey)
+hashtable_hash_traceback(const void *key)
 {
-    traceback_t *traceback;
-
-    _Py_HASHTABLE_READ_KEY(ht, pkey, traceback);
+    const traceback_t *traceback = (const traceback_t *)key;
     return traceback->hash;
 }
 
 
 static int
-hashtable_compare_traceback(_Py_hashtable_t *ht, const void *pkey,
-                            const _Py_hashtable_entry_t *entry)
+hashtable_compare_traceback(const void *key1, const void *key2)
 {
-    traceback_t *traceback1, *traceback2;
-    const frame_t *frame1, *frame2;
-    int i;
+    const traceback_t *traceback1 = (const traceback_t *)key1;
+    const traceback_t *traceback2 = (const traceback_t *)key2;
 
-    _Py_HASHTABLE_READ_KEY(ht, pkey, traceback1);
-    _Py_HASHTABLE_ENTRY_READ_KEY(ht, entry, traceback2);
-
-    if (traceback1->nframe != traceback2->nframe)
+    if (traceback1->nframe != traceback2->nframe) {
         return 0;
-
-    if (traceback1->total_nframe != traceback2->total_nframe)
+    }
+    if (traceback1->total_nframe != traceback2->total_nframe) {
         return 0;
+    }
 
-    for (i=0; i < traceback1->nframe; i++) {
-        frame1 = &traceback1->frames[i];
-        frame2 = &traceback2->frames[i];
+    for (int i=0; i < traceback1->nframe; i++) {
+        const frame_t *frame1 = &traceback1->frames[i];
+        const frame_t *frame2 = &traceback2->frames[i];
 
-        if (frame1->lineno != frame2->lineno)
+        if (frame1->lineno != frame2->lineno) {
             return 0;
-
+        }
         if (frame1->filename != frame2->filename) {
             assert(PyUnicode_Compare(frame1->filename, frame2->filename) != 0);
             return 0;
@@ -349,7 +341,7 @@ tracemalloc_get_frame(PyFrameObject *pyframe, frame_t *frame)
     _Py_hashtable_entry_t *entry;
     entry = _Py_HASHTABLE_GET_ENTRY(tracemalloc_filenames, filename);
     if (entry != NULL) {
-        _Py_HASHTABLE_ENTRY_READ_KEY(tracemalloc_filenames, entry, filename);
+        filename = (PyObject *)entry->key;
     }
     else {
         /* tracemalloc_filenames is responsible to keep a reference
@@ -444,7 +436,7 @@ traceback_new(void)
     /* intern the traceback */
     entry = _Py_HASHTABLE_GET_ENTRY(tracemalloc_tracebacks, traceback);
     if (entry != NULL) {
-        _Py_HASHTABLE_ENTRY_READ_KEY(tracemalloc_tracebacks, entry, traceback);
+        traceback = (traceback_t *)entry->key;
     }
     else {
         traceback_t *copy;
@@ -477,8 +469,7 @@ traceback_new(void)
 static _Py_hashtable_t*
 tracemalloc_create_traces_table(void)
 {
-    return hashtable_new(sizeof(uintptr_t),
-                         sizeof(trace_t),
+    return hashtable_new(sizeof(trace_t),
                          _Py_hashtable_hash_ptr,
                          _Py_hashtable_compare_direct);
 }
@@ -487,8 +478,7 @@ tracemalloc_create_traces_table(void)
 static _Py_hashtable_t*
 tracemalloc_create_domains_table(void)
 {
-    return hashtable_new(sizeof(unsigned int),
-                         sizeof(_Py_hashtable_t *),
+    return hashtable_new(sizeof(_Py_hashtable_t *),
                          hashtable_hash_uint,
                          _Py_hashtable_compare_direct);
 }
@@ -522,7 +512,7 @@ tracemalloc_get_traces_table(unsigned int domain)
     }
     else {
         _Py_hashtable_t *traces = NULL;
-        (void)_Py_HASHTABLE_GET(tracemalloc_domains, domain, traces);
+        (void)_Py_HASHTABLE_GET(tracemalloc_domains, TO_PTR(domain), traces);
         return traces;
     }
 }
@@ -539,7 +529,7 @@ tracemalloc_remove_trace(unsigned int domain, uintptr_t ptr)
     }
 
     trace_t trace;
-    if (!_Py_HASHTABLE_POP(traces, ptr, trace)) {
+    if (!_Py_HASHTABLE_POP(traces, TO_PTR(ptr), trace)) {
         return;
     }
     assert(tracemalloc_traced_memory >= trace.size);
@@ -568,7 +558,7 @@ tracemalloc_add_trace(unsigned int domain, uintptr_t ptr,
             return -1;
         }
 
-        if (_Py_HASHTABLE_SET(tracemalloc_domains, domain, traces) < 0) {
+        if (_Py_HASHTABLE_SET(tracemalloc_domains, TO_PTR(domain), traces) < 0) {
             _Py_hashtable_destroy(traces);
             return -1;
         }
@@ -590,7 +580,7 @@ tracemalloc_add_trace(unsigned int domain, uintptr_t ptr,
         trace.size = size;
         trace.traceback = traceback;
 
-        int res = _Py_HASHTABLE_SET(traces, ptr, trace);
+        int res = _Py_HASHTABLE_SET(traces, TO_PTR(ptr), trace);
         if (res != 0) {
             return res;
         }
@@ -859,9 +849,7 @@ static int
 tracemalloc_clear_filename(_Py_hashtable_t *ht, _Py_hashtable_entry_t *entry,
                            void *user_data)
 {
-    PyObject *filename;
-
-    _Py_HASHTABLE_ENTRY_READ_KEY(ht, entry, filename);
+    PyObject *filename = (PyObject *)entry->key;
     Py_DECREF(filename);
     return 0;
 }
@@ -871,9 +859,7 @@ static int
 traceback_free_traceback(_Py_hashtable_t *ht, _Py_hashtable_entry_t *entry,
                          void *user_data)
 {
-    traceback_t *traceback;
-
-    _Py_HASHTABLE_ENTRY_READ_KEY(ht, entry, traceback);
+    traceback_t *traceback = (traceback_t *)entry->key;
     raw_free(traceback);
     return 0;
 }
@@ -936,11 +922,11 @@ tracemalloc_init(void)
     }
 #endif
 
-    tracemalloc_filenames = hashtable_new(sizeof(PyObject *), 0,
+    tracemalloc_filenames = hashtable_new(0,
                                           hashtable_hash_pyobject,
                                           hashtable_compare_unicode);
 
-    tracemalloc_tracebacks = hashtable_new(sizeof(traceback_t *), 0,
+    tracemalloc_tracebacks = hashtable_new(0,
                                            hashtable_hash_traceback,
                                            hashtable_compare_traceback);
 
@@ -1154,7 +1140,7 @@ traceback_to_pyobject(traceback_t *traceback, _Py_hashtable_t *intern_table)
     PyObject *frames, *frame;
 
     if (intern_table != NULL) {
-        if (_Py_HASHTABLE_GET(intern_table, traceback, frames)) {
+        if (_Py_HASHTABLE_GET(intern_table, (const void *)traceback, frames)) {
             Py_INCREF(frames);
             return frames;
         }
@@ -1244,13 +1230,12 @@ tracemalloc_get_traces_copy_domain(_Py_hashtable_t *domains,
 {
     get_traces_t *get_traces = user_data;
 
-    unsigned int domain;
-    _Py_HASHTABLE_ENTRY_READ_KEY(domains, entry, domain);
+    unsigned int domain = (unsigned int)FROM_PTR(entry->key);
     _Py_hashtable_t *traces;
     _Py_HASHTABLE_ENTRY_READ_DATA(domains, entry, traces);
 
     _Py_hashtable_t *traces2 = _Py_hashtable_copy(traces);
-    if (_Py_HASHTABLE_SET(get_traces->domains, domain, traces2) < 0) {
+    if (_Py_HASHTABLE_SET(get_traces->domains, TO_PTR(domain), traces2) < 0) {
         _Py_hashtable_destroy(traces2);
         return -1;
     }
@@ -1289,8 +1274,7 @@ tracemalloc_get_traces_domain(_Py_hashtable_t *domains,
 {
     get_traces_t *get_traces = user_data;
 
-    unsigned int domain;
-    _Py_HASHTABLE_ENTRY_READ_KEY(domains, entry, domain);
+    unsigned int domain = (unsigned int)FROM_PTR(entry->key);
     _Py_hashtable_t *traces;
     _Py_HASHTABLE_ENTRY_READ_DATA(domains, entry, traces);
 
@@ -1343,8 +1327,7 @@ _tracemalloc__get_traces_impl(PyObject *module)
 
     /* the traceback hash table is used temporarily to intern traceback tuple
        of (filename, lineno) tuples */
-    get_traces.tracebacks = hashtable_new(sizeof(traceback_t *),
-                                          sizeof(PyObject *),
+    get_traces.tracebacks = hashtable_new(sizeof(PyObject *),
                                           _Py_hashtable_hash_ptr,
                                           _Py_hashtable_compare_direct);
     if (get_traces.tracebacks == NULL) {
@@ -1425,7 +1408,7 @@ tracemalloc_get_traceback(unsigned int domain, uintptr_t ptr)
     TABLES_LOCK();
     _Py_hashtable_t *traces = tracemalloc_get_traces_table(domain);
     if (traces) {
-        found = _Py_HASHTABLE_GET(traces, ptr, trace);
+        found = _Py_HASHTABLE_GET(traces, TO_PTR(ptr), trace);
     }
     else {
         found = 0;

Python/hashtable.c

@@ -59,7 +59,7 @@
 #define ENTRY_NEXT(ENTRY) \
         ((_Py_hashtable_entry_t *)_Py_SLIST_ITEM_NEXT(ENTRY))
 #define HASHTABLE_ITEM_SIZE(HT) \
-        (sizeof(_Py_hashtable_entry_t) + (HT)->key_size + (HT)->data_size)
+        (sizeof(_Py_hashtable_entry_t) + (HT)->data_size)
 
 #define ENTRY_READ_PDATA(TABLE, ENTRY, DATA_SIZE, PDATA) \
     do { \
@@ -105,20 +105,16 @@ _Py_slist_remove(_Py_slist_t *list, _Py_slist_item_t *previous,
 
 
 Py_uhash_t
-_Py_hashtable_hash_ptr(struct _Py_hashtable_t *ht, const void *pkey)
+_Py_hashtable_hash_ptr(const void *key)
 {
-    void *key;
-    _Py_HASHTABLE_READ_KEY(ht, pkey, key);
     return (Py_uhash_t)_Py_HashPointerRaw(key);
 }
 
 
 int
-_Py_hashtable_compare_direct(_Py_hashtable_t *ht, const void *pkey,
-                             const _Py_hashtable_entry_t *entry)
+_Py_hashtable_compare_direct(const void *key1, const void *key2)
 {
-    const void *pkey2 = _Py_HASHTABLE_ENTRY_PKEY(entry);
-    return (memcmp(pkey, pkey2, ht->key_size) == 0);
+    return (key1 == key2);
 }
 
 
@@ -195,16 +191,16 @@ _Py_hashtable_print_stats(_Py_hashtable_t *ht)
 
 
 _Py_hashtable_entry_t *
-_Py_hashtable_get_entry_generic(_Py_hashtable_t *ht, const void *pkey)
+_Py_hashtable_get_entry_generic(_Py_hashtable_t *ht, const void *key)
 {
-    Py_uhash_t key_hash = ht->hash_func(ht, pkey);
+    Py_uhash_t key_hash = ht->hash_func(key);
     size_t index = key_hash & (ht->num_buckets - 1);
     _Py_hashtable_entry_t *entry = entry = TABLE_HEAD(ht, index);
     while (1) {
         if (entry == NULL) {
             return NULL;
         }
-        if (entry->key_hash == key_hash && ht->compare_func(ht, pkey, entry)) {
+        if (entry->key_hash == key_hash && ht->compare_func(key, entry->key)) {
             break;
         }
         entry = ENTRY_NEXT(entry);
@@ -214,28 +210,27 @@ _Py_hashtable_get_entry_generic(_Py_hashtable_t *ht, const void *pkey)
 
 
 static int
-_Py_hashtable_pop_entry(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
+_Py_hashtable_pop_entry(_Py_hashtable_t *ht, const void *key,
                         void *data, size_t data_size)
 {
-    Py_uhash_t key_hash;
-    size_t index;
-    _Py_hashtable_entry_t *entry, *previous;
 
-    assert(key_size == ht->key_size);
-
-    key_hash = ht->hash_func(ht, pkey);
-    index = key_hash & (ht->num_buckets - 1);
+    Py_uhash_t key_hash = ht->hash_func(key);
+    size_t index = key_hash & (ht->num_buckets - 1);
 
-    previous = NULL;
-    for (entry = TABLE_HEAD(ht, index); entry != NULL; entry = ENTRY_NEXT(entry)) {
-        if (entry->key_hash == key_hash && ht->compare_func(ht, pkey, entry))
+    _Py_hashtable_entry_t *entry = TABLE_HEAD(ht, index);
+    _Py_hashtable_entry_t *previous = NULL;
+    while (1) {
+        if (entry == NULL) {
+            // not found
+            return 0;
+        }
+        if (entry->key_hash == key_hash && ht->compare_func(key, entry->key)) {
             break;
+        }
         previous = entry;
+        entry = ENTRY_NEXT(entry);
     }
 
-    if (entry == NULL)
-        return 0;
-
     _Py_slist_remove(&ht->buckets[index], (_Py_slist_item_t *)previous,
                      (_Py_slist_item_t *)entry);
     ht->entries--;
@@ -251,26 +246,22 @@ _Py_hashtable_pop_entry(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
 
 
 int
-_Py_hashtable_set(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
+_Py_hashtable_set(_Py_hashtable_t *ht, const void *key,
                   size_t data_size, const void *data)
 {
-    Py_uhash_t key_hash;
-    size_t index;
     _Py_hashtable_entry_t *entry;
 
-    assert(key_size == ht->key_size);
-
     assert(data != NULL || data_size == 0);
 #ifndef NDEBUG
     /* Don't write the assertion on a single line because it is interesting
        to know the duplicated entry if the assertion failed. The entry can
        be read using a debugger. */
-    entry = ht->get_entry_func(ht, pkey);
+    entry = ht->get_entry_func(ht, key);
     assert(entry == NULL);
 #endif
 
-    key_hash = ht->hash_func(ht, pkey);
-    index = key_hash & (ht->num_buckets - 1);
+    Py_uhash_t key_hash = ht->hash_func(key);
+    size_t index = key_hash & (ht->num_buckets - 1);
 
     entry = ht->alloc.malloc(HASHTABLE_ITEM_SIZE(ht));
     if (entry == NULL) {
@@ -279,9 +270,10 @@ _Py_hashtable_set(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
     }
 
     entry->key_hash = key_hash;
-    memcpy((void *)_Py_HASHTABLE_ENTRY_PKEY(entry), pkey, ht->key_size);
-    if (data)
+    entry->key = (void *)key;
+    if (data) {
         ENTRY_WRITE_PDATA(ht, entry, data_size, data);
+    }
 
     _Py_slist_prepend(&ht->buckets[index], (_Py_slist_item_t*)entry);
     ht->entries++;
@@ -293,10 +285,10 @@ _Py_hashtable_set(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
 
 
 int
-_Py_hashtable_get_generic(_Py_hashtable_t *ht, const void *pkey, void *data)
+_Py_hashtable_get_generic(_Py_hashtable_t *ht, const void *key, void *data)
 {
     assert(data != NULL);
-    _Py_hashtable_entry_t *entry = ht->get_entry_func(ht, pkey);
+    _Py_hashtable_entry_t *entry = ht->get_entry_func(ht, key);
     if (entry != NULL) {
         ENTRY_READ_PDATA(ht, entry, ht->data_size, data);
         return 1;
@@ -308,13 +300,12 @@ _Py_hashtable_get_generic(_Py_hashtable_t *ht, const void *pkey, void *data)
 
 
 // Specialized for:
-// key_size == sizeof(void*)
 // hash_func == _Py_hashtable_hash_ptr
 // compare_func == _Py_hashtable_compare_direct
 _Py_hashtable_entry_t *
-_Py_hashtable_get_entry_ptr(_Py_hashtable_t *ht, const void *pkey)
+_Py_hashtable_get_entry_ptr(_Py_hashtable_t *ht, const void *key)
 {
-    Py_uhash_t key_hash = _Py_hashtable_hash_ptr(ht, pkey);
+    Py_uhash_t key_hash = _Py_hashtable_hash_ptr(key);
     size_t index = key_hash & (ht->num_buckets - 1);
     _Py_hashtable_entry_t *entry = entry = TABLE_HEAD(ht, index);
     while (1) {
@@ -322,8 +313,7 @@ _Py_hashtable_get_entry_ptr(_Py_hashtable_t *ht, const void *pkey)
             return NULL;
         }
         if (entry->key_hash == key_hash) {
-            const void *pkey2 = _Py_HASHTABLE_ENTRY_PKEY(entry);
-            if (memcmp(pkey, pkey2, sizeof(void*)) == 0) {
+            if (entry->key == key) {
                 break;
             }
         }
@@ -334,14 +324,13 @@ _Py_hashtable_get_entry_ptr(_Py_hashtable_t *ht, const void *pkey)
 
 
 // Specialized for:
-// key_size == sizeof(void*)
 // hash_func == _Py_hashtable_hash_ptr
 // compare_func == _Py_hashtable_compare_direct
 int
-_Py_hashtable_get_ptr(_Py_hashtable_t *ht, const void *pkey, void *data)
+_Py_hashtable_get_ptr(_Py_hashtable_t *ht, const void *key, void *data)
 {
     assert(data != NULL);
-    _Py_hashtable_entry_t *entry = _Py_hashtable_get_entry_ptr(ht, pkey);
+    _Py_hashtable_entry_t *entry = _Py_hashtable_get_entry_ptr(ht, key);
     if (entry != NULL) {
         ENTRY_READ_PDATA(ht, entry, ht->data_size, data);
         return 1;
@@ -353,24 +342,24 @@ _Py_hashtable_get_ptr(_Py_hashtable_t *ht, const void *pkey, void *data)
 
 
 int
-_Py_hashtable_pop(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
+_Py_hashtable_pop(_Py_hashtable_t *ht, const void *key,
                   size_t data_size, void *data)
 {
     assert(data != NULL);
-    return _Py_hashtable_pop_entry(ht, key_size, pkey, data, data_size);
+    return _Py_hashtable_pop_entry(ht, key, data, data_size);
 }
 
 
 /* Code commented since the function is not needed in Python */
 #if 0
 void
-_Py_hashtable_delete(_Py_hashtable_t *ht, size_t key_size, const void *pkey)
+_Py_hashtable_delete(_Py_hashtable_t *ht, size_t const void *key)
 {
 #ifndef NDEBUG
-    int found = _Py_hashtable_pop_entry(ht, key_size, pkey, NULL, 0);
+    int found = _Py_hashtable_pop_entry(ht, key, NULL, 0);
     assert(found);
 #else
-    (void)_Py_hashtable_pop_entry(ht, key_size, pkey, NULL, 0);
+    (void)_Py_hashtable_pop_entry(ht, key, NULL, 0);
 #endif
 }
 #endif
@@ -427,7 +416,7 @@ hashtable_rehash(_Py_hashtable_t *ht)
             size_t entry_index;
 
 
-            assert(ht->hash_func(ht, _Py_HASHTABLE_ENTRY_PKEY(entry)) == entry->key_hash);
+            assert(ht->hash_func(entry->key) == entry->key_hash);
             next = ENTRY_NEXT(entry);
             entry_index = entry->key_hash & (new_size - 1);
 
@@ -440,8 +429,7 @@ hashtable_rehash(_Py_hashtable_t *ht)
 
 
 _Py_hashtable_t *
-_Py_hashtable_new_full(size_t key_size, size_t data_size,
-                       size_t init_size,
+_Py_hashtable_new_full(size_t data_size, size_t init_size,
                        _Py_hashtable_hash_func hash_func,
                        _Py_hashtable_compare_func compare_func,
                        _Py_hashtable_allocator_t *allocator)
@@ -464,7 +452,6 @@ _Py_hashtable_new_full(size_t key_size, size_t data_size,
 
     ht->num_buckets = round_size(init_size);
     ht->entries = 0;
-    ht->key_size = key_size;
     ht->data_size = data_size;
 
     buckets_size = ht->num_buckets * sizeof(ht->buckets[0]);
@@ -480,8 +467,7 @@ _Py_hashtable_new_full(size_t key_size, size_t data_size,
     ht->hash_func = hash_func;
     ht->compare_func = compare_func;
     ht->alloc = alloc;
-    if (ht->key_size == sizeof(void*)
-        && ht->hash_func == _Py_hashtable_hash_ptr
+    if (ht->hash_func == _Py_hashtable_hash_ptr
         && ht->compare_func == _Py_hashtable_compare_direct)
     {
         ht->get_func = _Py_hashtable_get_ptr;
@@ -492,12 +478,11 @@ _Py_hashtable_new_full(size_t key_size, size_t data_size,
 
 
 _Py_hashtable_t *
-_Py_hashtable_new(size_t key_size, size_t data_size,
+_Py_hashtable_new(size_t data_size,
                   _Py_hashtable_hash_func hash_func,
                   _Py_hashtable_compare_func compare_func)
 {
-    return _Py_hashtable_new_full(key_size, data_size,
-                                  HASHTABLE_MIN_SIZE,
+    return _Py_hashtable_new_full(data_size, HASHTABLE_MIN_SIZE,
                                   hash_func, compare_func,
                                   NULL);
 }
@@ -543,15 +528,13 @@ _Py_hashtable_destroy(_Py_hashtable_t *ht)
 _Py_hashtable_t *
 _Py_hashtable_copy(_Py_hashtable_t *src)
 {
-    const size_t key_size = src->key_size;
     const size_t data_size = src->data_size;
     _Py_hashtable_t *dst;
     _Py_hashtable_entry_t *entry;
     size_t bucket;
     int err;
 
-    dst = _Py_hashtable_new_full(key_size, data_size,
-                                 src->num_buckets,
+    dst = _Py_hashtable_new_full(data_size, src->num_buckets,
                                  src->hash_func,
                                  src->compare_func,
                                  &src->alloc);
@@ -561,9 +544,9 @@ _Py_hashtable_copy(_Py_hashtable_t *src)
     for (bucket=0; bucket < src->num_buckets; bucket++) {
         entry = TABLE_HEAD(src, bucket);
         for (; entry; entry = ENTRY_NEXT(entry)) {
-            const void *pkey = _Py_HASHTABLE_ENTRY_PKEY(entry);
+            const void *key = entry->key;
             const void *pdata = _Py_HASHTABLE_ENTRY_PDATA(src, entry);
-            err = _Py_hashtable_set(dst, key_size, pkey, data_size, pdata);
+            err = _Py_hashtable_set(dst, key, data_size, pdata);
             if (err) {
                 _Py_hashtable_destroy(dst);
                 return NULL;

Python/marshal.c

@@ -549,7 +549,7 @@ static int
 w_init_refs(WFILE *wf, int version)
 {
     if (version >= 3) {
-        wf->hashtable = _Py_hashtable_new(sizeof(PyObject *), sizeof(int),
+        wf->hashtable = _Py_hashtable_new(sizeof(int),
                                           _Py_hashtable_hash_ptr,
                                           _Py_hashtable_compare_direct);
         if (wf->hashtable == NULL) {
@@ -564,9 +564,7 @@ static int
 w_decref_entry(_Py_hashtable_t *ht, _Py_hashtable_entry_t *entry,
                void *Py_UNUSED(data))
 {
-    PyObject *entry_key;
-
-    _Py_HASHTABLE_ENTRY_READ_KEY(ht, entry, entry_key);
+    PyObject *entry_key = (PyObject *)entry->key;
     Py_XDECREF(entry_key);
     return 0;
 }