Fix the internals of our hash functions to used unsigned values durin… · native-api/cpython@c2176e4 · GitHub

1. native-api
2. cpython

Repository navigation

• Code
• Pull requests
• Actions
• Projects
• Security and quality
• Insights

Commit c2176e4

Browse filesBrowse the repository at this point in the historyBrowse files

gpshead

committed

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

computation as the overflow behavior of signed integers is undefined. NOTE: This change is smaller compared to 3.2 as much of this cleanup had already been done. I added the comment that my change in 3.2 added so that the code would match up. Otherwise this just adds or synchronizes appropriate UL designations on some constants to be pedantic. 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.

2 parents c0dd80e + 27cbcd6 commit c2176e4
Copy full SHA for c2176e4

4 files changed

+12-12Lines changed: 12 additions & 12 deletions

File tree

Expand file treeCollapse file tree

Open diff view settings

Filter options

• Include
  • pyport.h
• Objects
  • setobject.c
  • tupleobject.c
  • unicodeobject.c

Expand file treeCollapse file tree

Open diff view settings

Collapse file

‎Include/pyport.h‎

Copy file name to clipboardExpand all lines: Include/pyport.h

+1-1Lines changed: 1 addition & 1 deletion

Original file line number	Diff line number	Diff line change
@@ -145,7 +145,7 @@ Used in: PY_LONG_LONG
145	145	#endif
146	146
147	147	/* Prime multiplier used in string and various other hashes. */
148		- #define _PyHASH_MULTIPLIER 1000003 /* 0xf4243 */
	148	+ #define _PyHASH_MULTIPLIER 1000003UL /* 0xf4243 */
149	149
150	150	/* Parameters used for the numeric hash implementation. See notes for
151	151	_Py_HashDouble in Objects/object.c. Numeric hashes are based on

Collapse file

‎Objects/setobject.c‎

Copy file name to clipboardExpand all lines: Objects/setobject.c

+6-6Lines changed: 6 additions & 6 deletions

Original file line number	Diff line number	Diff line change
@@ -77,7 +77,7 @@ NULL if the rich comparison returns an error.
77	77	static setentry *
78	78	set_lookkey(PySetObject *so, PyObject *key, register Py_hash_t hash)
79	79	{
80		- register size_t i;
	80	+ register size_t i; /* Unsigned for defined overflow behavior. */
81	81	register size_t perturb;
82	82	register setentry *freeslot;
83	83	register size_t mask = so->mask;
@@ -159,7 +159,7 @@ set_lookkey(PySetObject *so, PyObject *key, register Py_hash_t hash)
159	159	static setentry *
160	160	set_lookkey_unicode(PySetObject *so, PyObject *key, register Py_hash_t hash)
161	161	{
162		- register size_t i;
	162	+ register size_t i; /* Unsigned for defined overflow behavior. */
163	163	register size_t perturb;
164	164	register setentry *freeslot;
165	165	register size_t mask = so->mask;
@@ -760,7 +760,7 @@ static Py_hash_t
760	760	frozenset_hash(PyObject *self)
761	761	{
762	762	PySetObject *so = (PySetObject *)self;
763		- Py_uhash_t h, hash = 1927868237U;
	763	+ Py_uhash_t h, hash = 1927868237UL;
764	764	setentry *entry;
765	765	Py_ssize_t pos = 0;
766	766
@@ -775,11 +775,11 @@ frozenset_hash(PyObject *self)
775	775	hashes so that many distinct combinations collapse to only
776	776	a handful of distinct hash values. */
777	777	h = entry->hash;
778		- hash ^= (h ^ (h << 16) ^ 89869747U) * 3644798167U;
	778	+ hash ^= (h ^ (h << 16) ^ 89869747UL) * 3644798167UL;
779	779	}
780		- hash = hash * 69069U + 907133923U;
	780	+ hash = hash * 69069U + 907133923UL;
781	781	if (hash == -1)
782		- hash = 590923713U;
	782	+ hash = 590923713UL;
783	783	so->hash = hash;
784	784	return hash;
785	785	}

Collapse file

‎Objects/tupleobject.c‎

Copy file name to clipboardExpand all lines: Objects/tupleobject.c

+4-4Lines changed: 4 additions & 4 deletions

Original file line number	Diff line number	Diff line change
@@ -327,22 +327,22 @@ tuplerepr(PyTupleObject *v)
327	327	static Py_hash_t
328	328	tuplehash(PyTupleObject *v)
329	329	{
330		- register Py_uhash_t x;
	330	+ register Py_uhash_t x; /* Unsigned for defined overflow behavior. */
331	331	register Py_hash_t y;
332	332	register Py_ssize_t len = Py_SIZE(v);
333	333	register PyObject **p;
334	334	Py_uhash_t mult = _PyHASH_MULTIPLIER;
335		- x = 0x345678;
	335	+ x = 0x345678UL;
336	336	p = v->ob_item;
337	337	while (--len >= 0) {
338	338	y = PyObject_Hash(*p++);
339	339	if (y == -1)
340	340	return -1;
341	341	x = (x ^ y) * mult;
342	342	/* the cast might truncate len; that doesn't change hash stability */
343		- mult += (Py_hash_t)(82520L + len + len);
	343	+ mult += (Py_hash_t)(82520UL + len + len);
344	344	}
345		- x += 97531L;
	345	+ x += 97531UL;
346	346	if (x == (Py_uhash_t)-1)
347	347	x = -2;
348	348	return x;

Collapse file

‎Objects/unicodeobject.c‎

Copy file name to clipboardExpand all lines: Objects/unicodeobject.c

+1-1Lines changed: 1 addition & 1 deletion

Original file line number	Diff line number	Diff line change
@@ -11008,7 +11008,7 @@ static Py_hash_t
11008	11008	unicode_hash(PyObject *self)
11009	11009	{
11010	11010	Py_ssize_t len;
11011		- Py_uhash_t x;
	11011	+ Py_uhash_t x; /* Unsigned for defined overflow behavior. */
11012	11012
11013	11013	#ifdef Py_DEBUG
11014	11014	assert(_Py_HashSecret_Initialized);

0 commit comments

Comments

0 (0)