Fix the internals of our hash functions to used unsigned values during hash

computation as the overflow behavior of signed integers is undefined.

In practice we require compiling everything with -fwrapv which forces overflow
to be defined as twos compliment but this keeps the code cleaner for checkers
or in the case where someone has compiled it without -fwrapv or their
compiler's equivalent.

Found by Clang trunk's Undefined Behavior Sanitizer (UBSan).

Cleanup only - no functionality or hash values change.

diff --git a/Include/pyport.h b/Include/pyport.h
index 9666cf28f972374f39f596bd7905fb41e850cdf9..c74ff9c7a4b4628746d4c9ced50a95c3580351bd 100644 (file)
--- a/Include/pyport.h
+++ b/Include/pyport.h
@@ -145,7 +145,7 @@ Used in:  PY_LONG_LONG
 #endif
 
 /* Prime multiplier used in string and various other hashes. */
-#define _PyHASH_MULTIPLIER 1000003  /* 0xf4243 */
+#define _PyHASH_MULTIPLIER 1000003UL  /* 0xf4243 */
 
 /* Parameters used for the numeric hash implementation.  See notes for
    _PyHash_Double in Objects/object.c.  Numeric hashes are based on

diff --git a/Objects/bytesobject.c b/Objects/bytesobject.c
index 796e400a822051afa6cccecc4da3ebba6b473785..ef3a5a12a2361e241b3efc1d7601eb7d1d7d6ae9 100644 (file)
--- a/Objects/bytesobject.c
+++ b/Objects/bytesobject.c
@@ -873,7 +873,7 @@ bytes_hash(PyBytesObject *a)
 {
     register Py_ssize_t len;
     register unsigned char *p;
-    register Py_hash_t x;
+    register Py_uhash_t x;  /* Unsigned for defined overflow behavior. */
 
 #ifdef Py_DEBUG
     assert(_Py_HashSecret_Initialized);

diff --git a/Objects/setobject.c b/Objects/setobject.c
index 3abeefb97d9400a5945e5b9c9ad67e4325b89c89..d8401f4ae746cc722249bbcacb26cc55a93c7644 100644 (file)
--- a/Objects/setobject.c
+++ b/Objects/setobject.c
@@ -77,7 +77,7 @@ NULL if the rich comparison returns an error.
 static setentry *
 set_lookkey(PySetObject *so, PyObject *key, register Py_hash_t hash)
 {
-    register Py_ssize_t i;
+    register size_t i;  /* Unsigned for defined overflow behavior. */
     register size_t perturb;
     register setentry *freeslot;
     register size_t mask = so->mask;
@@ -159,7 +159,7 @@ set_lookkey(PySetObject *so, PyObject *key, register Py_hash_t hash)
 static setentry *
 set_lookkey_unicode(PySetObject *so, PyObject *key, register Py_hash_t hash)
 {
-    register Py_ssize_t i;
+    register size_t i;  /* Unsigned for defined overflow behavior. */
     register size_t perturb;
     register setentry *freeslot;
     register size_t mask = so->mask;
@@ -768,7 +768,7 @@ static Py_hash_t
 frozenset_hash(PyObject *self)
 {
     PySetObject *so = (PySetObject *)self;
-    Py_hash_t h, hash = 1927868237L;
+    Py_uhash_t h, hash = 1927868237UL;
     setentry *entry;
     Py_ssize_t pos = 0;
 
@@ -783,11 +783,11 @@ frozenset_hash(PyObject *self)
            hashes so that many distinct combinations collapse to only
            a handful of distinct hash values. */
         h = entry->hash;
-        hash ^= (h ^ (h << 16) ^ 89869747L)  * 3644798167u;
+        hash ^= (h ^ (h << 16) ^ 89869747UL)  * 3644798167UL;
     }
-    hash = hash * 69069L + 907133923L;
+    hash = hash * 69069UL + 907133923UL;
     if (hash == -1)
-        hash = 590923713L;
+        hash = 590923713UL;
     so->hash = hash;
     return hash;
 }

diff --git a/Objects/tupleobject.c b/Objects/tupleobject.c
index b3454600c9ba087f34fbf51113a2d567a222ac7a..c725227979bc09d12107459f79e0edacad8981f3 100644 (file)
--- a/Objects/tupleobject.c
+++ b/Objects/tupleobject.c
@@ -315,11 +315,11 @@ error:
 static Py_hash_t
 tuplehash(PyTupleObject *v)
 {
-    register Py_hash_t x, y;
+    register Py_uhash_t x, y;  /* Unsigned for defined overflow behavior. */
     register Py_ssize_t len = Py_SIZE(v);
     register PyObject **p;
-    Py_hash_t mult = _PyHASH_MULTIPLIER;
-    x = 0x345678L;
+    Py_uhash_t mult = _PyHASH_MULTIPLIER;
+    x = 0x345678UL;
     p = v->ob_item;
     while (--len >= 0) {
         y = PyObject_Hash(*p++);
@@ -327,9 +327,9 @@ tuplehash(PyTupleObject *v)
             return -1;
         x = (x ^ y) * mult;
         /* the cast might truncate len; that doesn't change hash stability */
-        mult += (Py_hash_t)(82520L + len + len);
+        mult += (Py_uhash_t)(82520UL + len + len);
     }
-    x += 97531L;
+    x += 97531UL;
     if (x == -1)
         x = -2;
     return x;

diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index 565d2982708154c008d731620dd62fea9abba394..bb45b20b90cd46a1a10f25d78dc25c9304d48840 100644 (file)
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -7686,7 +7686,7 @@ unicode_hash(PyUnicodeObject *self)
 {
     Py_ssize_t len;
     Py_UNICODE *p;
-    Py_hash_t x;
+    Py_uhash_t x;  /* Unsigned for defined overflow behavior. */
 
 #ifdef Py_DEBUG
     assert(_Py_HashSecret_Initialized);