Cleanup hashtable.h

Issue #26588:

* Pass the hash table rather than the key size to hash and compare functions
* _Py_HASHTABLE_READ_KEY() and _Py_HASHTABLE_ENTRY_READ_KEY() macros now expect
  the hash table as the first parameter, rather than the key size
* tracemalloc_get_traces_fill(): use _Py_HASHTABLE_ENTRY_READ_DATA() rather
  than pointer dereference
* Remove the _Py_HASHTABLE_ENTRY_WRITE_PKEY() macro
* Move "PKEY" and "PDATA" macros inside hashtable.c

Modules/_tracemalloc.c

@@ -223,23 +223,23 @@ set_reentrant(int reentrant)
 
 
 static Py_uhash_t
-hashtable_hash_pyobject(size_t key_size, const void *pkey)
+hashtable_hash_pyobject(_Py_hashtable_t *ht, const void *pkey)
 {
     PyObject *obj;
 
-    _Py_HASHTABLE_READ_KEY(key_size, pkey, obj);
+    _Py_HASHTABLE_READ_KEY(ht, pkey, obj);
     return PyObject_Hash(obj);
 }
 
 
 static int
-hashtable_compare_unicode(size_t key_size, const void *pkey,
+hashtable_compare_unicode(_Py_hashtable_t *ht, const void *pkey,
                           const _Py_hashtable_entry_t *entry)
 {
     PyObject *key1, *key2;
 
-    _Py_HASHTABLE_READ_KEY(key_size, pkey, key1);
-    _Py_HASHTABLE_ENTRY_READ_KEY(key_size, entry, 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);
@@ -249,12 +249,12 @@ hashtable_compare_unicode(size_t key_size, const void *pkey,
 
 
 static Py_uhash_t
-hashtable_hash_pointer_t(size_t key_size, const void *pkey)
+hashtable_hash_pointer_t(_Py_hashtable_t *ht, const void *pkey)
 {
     pointer_t ptr;
     Py_uhash_t hash;
 
-    _Py_HASHTABLE_READ_KEY(key_size, pkey, ptr);
+    _Py_HASHTABLE_READ_KEY(ht, pkey, ptr);
 
     hash = (Py_uhash_t)_Py_HashPointer((void*)ptr.ptr);
     hash ^= ptr.domain;
@@ -263,13 +263,13 @@ hashtable_hash_pointer_t(size_t key_size, const void *pkey)
 
 
 int
-hashtable_compare_pointer_t(size_t key_size, const void *pkey,
+hashtable_compare_pointer_t(_Py_hashtable_t *ht, const void *pkey,
                             const _Py_hashtable_entry_t *entry)
 {
     pointer_t ptr1, ptr2;
 
-    _Py_HASHTABLE_READ_KEY(key_size, pkey, ptr1);
-    _Py_HASHTABLE_ENTRY_READ_KEY(key_size, entry, ptr2);
+    _Py_HASHTABLE_READ_KEY(ht, pkey, ptr1);
+    _Py_HASHTABLE_ENTRY_READ_KEY(ht, entry, ptr2);
 
     /* compare pointer before domain, because pointer is more likely to be
        different */
@@ -304,25 +304,25 @@ raw_free(void *ptr)
 
 
 static Py_uhash_t
-hashtable_hash_traceback(size_t key_size, const void *pkey)
+hashtable_hash_traceback(_Py_hashtable_t *ht, const void *pkey)
 {
     traceback_t *traceback;
 
-    _Py_HASHTABLE_READ_KEY(key_size, pkey, traceback);
+    _Py_HASHTABLE_READ_KEY(ht, pkey, traceback);
     return traceback->hash;
 }
 
 
 static int
-hashtable_compare_traceback(size_t key_size, const void *pkey,
-                            const _Py_hashtable_entry_t *he)
+hashtable_compare_traceback(_Py_hashtable_t *ht, const void *pkey,
+                            const _Py_hashtable_entry_t *entry)
 {
     traceback_t *traceback1, *traceback2;
     const frame_t *frame1, *frame2;
     int i;
 
-    _Py_HASHTABLE_READ_KEY(key_size, pkey, traceback1);
-    _Py_HASHTABLE_ENTRY_READ_KEY(key_size, he, traceback2);
+    _Py_HASHTABLE_READ_KEY(ht, pkey, traceback1);
+    _Py_HASHTABLE_ENTRY_READ_KEY(ht, entry, traceback2);
 
     if (traceback1->nframe != traceback2->nframe)
         return 0;
@@ -395,8 +395,7 @@ tracemalloc_get_frame(PyFrameObject *pyframe, frame_t *frame)
     /* intern the filename */
     entry = _Py_HASHTABLE_GET_ENTRY(tracemalloc_filenames, filename);
     if (entry != NULL) {
-        _Py_HASHTABLE_ENTRY_READ_KEY(tracemalloc_filenames->key_size, entry,
-                                     filename);
+        _Py_HASHTABLE_ENTRY_READ_KEY(tracemalloc_filenames, entry, filename);
     }
     else {
         /* tracemalloc_filenames is responsible to keep a reference
@@ -490,8 +489,7 @@ traceback_new(void)
     /* intern the traceback */
     entry = _Py_HASHTABLE_GET_ENTRY(tracemalloc_tracebacks, traceback);
     if (entry != NULL) {
-        _Py_HASHTABLE_ENTRY_READ_KEY(tracemalloc_tracebacks->key_size, entry,
-                                     traceback);
+        _Py_HASHTABLE_ENTRY_READ_KEY(tracemalloc_tracebacks, entry, traceback);
     }
     else {
         traceback_t *copy;
@@ -873,7 +871,7 @@ tracemalloc_clear_filename(_Py_hashtable_t *ht, _Py_hashtable_entry_t *entry,
 {
     PyObject *filename;
 
-    _Py_HASHTABLE_ENTRY_READ_KEY(ht->key_size, entry, filename);
+    _Py_HASHTABLE_ENTRY_READ_KEY(ht, entry, filename);
     Py_DECREF(filename);
     return 0;
 }
@@ -885,7 +883,7 @@ traceback_free_traceback(_Py_hashtable_t *ht, _Py_hashtable_entry_t *entry,
 {
     traceback_t *traceback;
 
-    _Py_HASHTABLE_ENTRY_READ_KEY(ht->key_size, entry, traceback);
+    _Py_HASHTABLE_ENTRY_READ_KEY(ht, entry, traceback);
     raw_free(traceback);
     return 0;
 }
@@ -1246,21 +1244,21 @@ tracemalloc_get_traces_fill(_Py_hashtable_t *traces, _Py_hashtable_entry_t *entr
 {
     get_traces_t *get_traces = user_data;
     _PyTraceMalloc_domain_t domain;
-    trace_t *trace;
+    trace_t trace;
     PyObject *tracemalloc_obj;
     int res;
 
     if (tracemalloc_config.use_domain) {
         pointer_t key;
-        _Py_HASHTABLE_ENTRY_READ_KEY(traces->key_size, entry, key);
+        _Py_HASHTABLE_ENTRY_READ_KEY(traces, entry, key);
         domain = key.domain;
     }
     else {
         domain = DEFAULT_DOMAIN;
     }
-    trace = (trace_t *)_Py_HASHTABLE_ENTRY_DATA(traces, entry);
+    _Py_HASHTABLE_ENTRY_READ_DATA(traces, entry, trace);
 
-    tracemalloc_obj = trace_to_pyobject(domain, trace, get_traces->tracebacks);
+    tracemalloc_obj = trace_to_pyobject(domain, &trace, get_traces->tracebacks);
     if (tracemalloc_obj == NULL)
         return 1;
 

Modules/hashtable.c

@@ -61,6 +61,20 @@
 #define HASHTABLE_ITEM_SIZE(HT) \
         (sizeof(_Py_hashtable_entry_t) + (HT)->key_size + (HT)->data_size)
 
+#define ENTRY_READ_PDATA(TABLE, ENTRY, DATA_SIZE, PDATA) \
+    do { \
+        assert((DATA_SIZE) == (TABLE)->data_size); \
+        Py_MEMCPY((PDATA), _Py_HASHTABLE_ENTRY_PDATA(TABLE, (ENTRY)), \
+                  (DATA_SIZE)); \
+    } while (0)
+
+#define ENTRY_WRITE_PDATA(TABLE, ENTRY, DATA_SIZE, PDATA) \
+    do { \
+        assert((DATA_SIZE) == (TABLE)->data_size); \
+        Py_MEMCPY((void *)_Py_HASHTABLE_ENTRY_PDATA((TABLE), (ENTRY)), \
+                  (PDATA), (DATA_SIZE)); \
+    } while (0)
+
 /* Forward declaration */
 static void hashtable_rehash(_Py_hashtable_t *ht);
 
@@ -91,21 +105,21 @@ _Py_slist_remove(_Py_slist_t *list, _Py_slist_item_t *previous,
 
 
 Py_uhash_t
-_Py_hashtable_hash_ptr(size_t key_size, const void *pkey)
+_Py_hashtable_hash_ptr(struct _Py_hashtable_t *ht, const void *pkey)
 {
     void *key;
 
-    _Py_HASHTABLE_READ_KEY(key_size, pkey, key);
-    return (Py_uhash_t)_Py_HashPointer((void *)key);
+    _Py_HASHTABLE_READ_KEY(ht, pkey, key);
+    return (Py_uhash_t)_Py_HashPointer(key);
 }
 
 
 int
-_Py_hashtable_compare_direct(size_t key_size, const void *pkey,
+_Py_hashtable_compare_direct(_Py_hashtable_t *ht, const void *pkey,
                              const _Py_hashtable_entry_t *entry)
 {
-    const void *pkey2 = _Py_HASHTABLE_ENTRY_KEY(entry);
-    return (memcmp(pkey, pkey2, key_size) == 0);
+    const void *pkey2 = _Py_HASHTABLE_ENTRY_PKEY(entry);
+    return (memcmp(pkey, pkey2, ht->key_size) == 0);
 }
 
 
@@ -245,12 +259,11 @@ _Py_hashtable_get_entry(_Py_hashtable_t *ht,
 
     assert(key_size == ht->key_size);
 
-    key_hash = ht->hash_func(key_size, pkey);
+    key_hash = ht->hash_func(ht, pkey);
     index = key_hash & (ht->num_buckets - 1);
 
     for (entry = TABLE_HEAD(ht, index); entry != NULL; entry = ENTRY_NEXT(entry)) {
-        if (entry->key_hash == key_hash
-           && ht->compare_func(key_size, pkey, entry))
+        if (entry->key_hash == key_hash && ht->compare_func(ht, pkey, entry))
             break;
     }
 
@@ -268,13 +281,12 @@ _Py_hashtable_pop_entry(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
 
     assert(key_size == ht->key_size);
 
-    key_hash = ht->hash_func(key_size, pkey);
+    key_hash = ht->hash_func(ht, pkey);
     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(key_size, pkey, entry))
+        if (entry->key_hash == key_hash && ht->compare_func(ht, pkey, entry))
             break;
         previous = entry;
     }
@@ -287,7 +299,7 @@ _Py_hashtable_pop_entry(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
     ht->entries--;
 
     if (data != NULL)
-        _Py_HASHTABLE_ENTRY_READ_PDATA(ht, entry, data_size, data);
+        ENTRY_READ_PDATA(ht, entry, data_size, data);
     ht->alloc.free(entry);
 
     if ((float)ht->entries / (float)ht->num_buckets < HASHTABLE_LOW)
@@ -315,7 +327,7 @@ _Py_hashtable_set(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
     assert(entry == NULL);
 #endif
 
-    key_hash = ht->hash_func(key_size, pkey);
+    key_hash = ht->hash_func(ht, pkey);
     index = key_hash & (ht->num_buckets - 1);
 
     entry = ht->alloc.malloc(HASHTABLE_ITEM_SIZE(ht));
@@ -325,8 +337,8 @@ _Py_hashtable_set(_Py_hashtable_t *ht, size_t key_size, const void *pkey,
     }
 
     entry->key_hash = key_hash;
-    _Py_HASHTABLE_ENTRY_WRITE_PKEY(key_size, entry, pkey);
-    _Py_HASHTABLE_ENTRY_WRITE_PDATA(ht, entry, data_size, data);
+    Py_MEMCPY((void *)_Py_HASHTABLE_ENTRY_PKEY(entry), pkey, ht->key_size);
+    ENTRY_WRITE_PDATA(ht, entry, data_size, data);
 
     _Py_slist_prepend(&ht->buckets[index], (_Py_slist_item_t*)entry);
     ht->entries++;
@@ -348,7 +360,7 @@ _Py_hashtable_get(_Py_hashtable_t *ht, size_t key_size,const void *pkey,
     entry = _Py_hashtable_get_entry(ht, key_size, pkey);
     if (entry == NULL)
         return 0;
-    _Py_HASHTABLE_ENTRY_READ_PDATA(ht, entry, data_size, data);
+    ENTRY_READ_PDATA(ht, entry, data_size, data);
     return 1;
 }
 
@@ -399,7 +411,6 @@ _Py_hashtable_foreach(_Py_hashtable_t *ht,
 static void
 hashtable_rehash(_Py_hashtable_t *ht)
 {
-    const size_t key_size = ht->key_size;
     size_t buckets_size, new_size, bucket;
     _Py_slist_t *old_buckets = NULL;
     size_t old_num_buckets;
@@ -429,7 +440,7 @@ hashtable_rehash(_Py_hashtable_t *ht)
             size_t entry_index;
 
 
-            assert(ht->hash_func(key_size, _Py_HASHTABLE_ENTRY_KEY(entry)) == entry->key_hash);
+            assert(ht->hash_func(ht, _Py_HASHTABLE_ENTRY_PKEY(entry)) == entry->key_hash);
             next = ENTRY_NEXT(entry);
             entry_index = entry->key_hash & (new_size - 1);
 
@@ -499,9 +510,9 @@ _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_KEY(entry);
-            const void *data = _Py_HASHTABLE_ENTRY_DATA(src, entry);
-            err = _Py_hashtable_set(dst, key_size, pkey, data_size, data);
+            const void *pkey = _Py_HASHTABLE_ENTRY_PKEY(entry);
+            const void *pdata = _Py_HASHTABLE_ENTRY_PDATA(src, entry);
+            err = _Py_hashtable_set(dst, key_size, pkey, data_size, pdata);
             if (err) {
                 _Py_hashtable_destroy(dst);
                 return NULL;

Modules/hashtable.h

@@ -29,60 +29,52 @@ typedef struct {
     /* key (key_size bytes) and then data (data_size bytes) follows */
 } _Py_hashtable_entry_t;
 
-#define _Py_HASHTABLE_ENTRY_KEY(ENTRY) \
+#define _Py_HASHTABLE_ENTRY_PKEY(ENTRY) \
         ((const void *)((char *)(ENTRY) \
                         + sizeof(_Py_hashtable_entry_t)))
 
-#define _Py_HASHTABLE_ENTRY_DATA(TABLE, ENTRY) \
+#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(KEY_SIZE, PKEY, DST_KEY) \
+#define _Py_HASHTABLE_READ_KEY(TABLE, PKEY, DST_KEY) \
     do { \
-        assert(sizeof(DST_KEY) == (KEY_SIZE)); \
-        memcpy(&(DST_KEY), (PKEY), sizeof(DST_KEY)); \
+        assert(sizeof(DST_KEY) == (TABLE)->key_size); \
+        Py_MEMCPY(&(DST_KEY), (PKEY), sizeof(DST_KEY)); \
     } while (0)
 
-#define _Py_HASHTABLE_ENTRY_READ_KEY(KEY_SIZE, ENTRY, KEY) \
+#define _Py_HASHTABLE_ENTRY_READ_KEY(TABLE, ENTRY, KEY) \
     do { \
-        assert(sizeof(KEY) == (KEY_SIZE)); \
-        memcpy(&(KEY), _Py_HASHTABLE_ENTRY_KEY(ENTRY), sizeof(KEY)); \
-    } while (0)
-
-#define _Py_HASHTABLE_ENTRY_WRITE_PKEY(KEY_SIZE, ENTRY, PKEY) \
-    do { \
-        memcpy((void *)_Py_HASHTABLE_ENTRY_KEY(ENTRY), (PKEY), (KEY_SIZE)); \
-    } while (0)
-
-#define _Py_HASHTABLE_ENTRY_READ_PDATA(TABLE, ENTRY, DATA_SIZE, PDATA) \
-    do { \
-        assert((DATA_SIZE) == (TABLE)->data_size); \
-        memcpy((PDATA), _Py_HASHTABLE_ENTRY_DATA(TABLE, (ENTRY)), \
-               (DATA_SIZE)); \
+        assert(sizeof(KEY) == (TABLE)->key_size); \
+        Py_MEMCPY(&(KEY), _Py_HASHTABLE_ENTRY_PKEY(ENTRY), sizeof(KEY)); \
     } while (0)
 
 #define _Py_HASHTABLE_ENTRY_READ_DATA(TABLE, ENTRY, DATA) \
-    _Py_HASHTABLE_ENTRY_READ_PDATA((TABLE), (ENTRY), sizeof(DATA), &(DATA))
-
-#define _Py_HASHTABLE_ENTRY_WRITE_PDATA(TABLE, ENTRY, DATA_SIZE, PDATA) \
     do { \
-        assert((DATA_SIZE) == (TABLE)->data_size); \
-        memcpy((void *)_Py_HASHTABLE_ENTRY_DATA((TABLE), (ENTRY)), \
-               (PDATA), (DATA_SIZE)); \
+        assert(sizeof(DATA) == (TABLE)->data_size); \
+        Py_MEMCPY(&(DATA), _Py_HASHTABLE_ENTRY_PDATA(TABLE, (ENTRY)), \
+                  sizeof(DATA)); \
     } while (0)
 
 #define _Py_HASHTABLE_ENTRY_WRITE_DATA(TABLE, ENTRY, DATA) \
-    _Py_HASHTABLE_ENTRY_WRITE_PDATA(TABLE, ENTRY, sizeof(DATA), &(DATA))
+    do { \
+        assert(sizeof(DATA) == (TABLE)->data_size); \
+        Py_MEMCPY((void *)_Py_HASHTABLE_ENTRY_PDATA((TABLE), (ENTRY)), \
+                  &(DATA), sizeof(DATA)); \
+    } while (0)
 
 
 /* _Py_hashtable: prototypes */
 
-typedef Py_uhash_t (*_Py_hashtable_hash_func) (size_t key_size,
+/* Forward declaration */
+struct _Py_hashtable_t;
+
+typedef Py_uhash_t (*_Py_hashtable_hash_func) (struct _Py_hashtable_t *ht,
                                                const void *pkey);
-typedef int (*_Py_hashtable_compare_func) (size_t key_size,
+typedef int (*_Py_hashtable_compare_func) (struct _Py_hashtable_t *ht,
                                            const void *pkey,
                                            const _Py_hashtable_entry_t *he);
 
@@ -97,7 +89,7 @@ typedef struct {
 
 /* _Py_hashtable: table */
 
-typedef struct {
+typedef struct _Py_hashtable_t {
     size_t num_buckets;
     size_t entries; /* Total number of entries in the table. */
     _Py_slist_t *buckets;
@@ -111,12 +103,12 @@ typedef struct {
 
 /* hash a pointer (void*) */
 PyAPI_FUNC(Py_uhash_t) _Py_hashtable_hash_ptr(
-    size_t key_size,
+    struct _Py_hashtable_t *ht,
     const void *pkey);
 
 /* comparison using memcmp() */
 PyAPI_FUNC(int) _Py_hashtable_compare_direct(
-    size_t key_size,
+    _Py_hashtable_t *ht,
     const void *pkey,
     const _Py_hashtable_entry_t *entry);
 

Python/marshal.c

@@ -588,7 +588,7 @@ w_decref_entry(_Py_hashtable_t *ht, _Py_hashtable_entry_t *entry,
 {
     PyObject *entry_key;
 
-    _Py_HASHTABLE_ENTRY_READ_KEY(ht->key_size, entry, entry_key);
+    _Py_HASHTABLE_ENTRY_READ_KEY(ht, entry, entry_key);
     Py_XDECREF(entry_key);
     return 0;
 }