gh-155835: fix digest_size and block_size data races on SHA-3 objects (GH-155838)

(cherry picked from commit 90ac539)

Co-authored-by: Bénédikt Tran <10796600+picnixz@users.noreply.github.com>

Author: picnixz

Lib/test/test_hashlib.py

@@ -18,6 +18,8 @@
 import tempfile
 import threading
 import unittest
+from functools import partial
+from operator import attrgetter
 from test import support
 from test.support import _4G, bigmemtest
 from test.support import hashlib_helper
@@ -52,18 +54,39 @@
     def get_fips_mode():
         return 0
 
+
+try:
+    import _md5
+except ImportError:
+    _md5 = None
+requires_md5 = unittest.skipUnless(_md5, 'requires _md5')
+
+
 try:
     import _blake2
 except ImportError:
     _blake2 = None
-
 requires_blake2 = unittest.skipUnless(_blake2, 'requires _blake2')
 
+
+try:
+    import _sha1
+except ImportError:
+    _sha1 = None
+requires_sha1 = unittest.skipUnless(_sha1, 'requires _sha1')
+
+
+try:
+    import _sha2
+except ImportError:
+    _sha2 = None
+requires_sha2 = unittest.skipUnless(_sha2, 'requires _sha2')
+
+
 try:
     import _sha3
 except ImportError:
     _sha3 = None
-
 requires_sha3 = unittest.skipUnless(_sha3, 'requires _sha3')
 
 
@@ -1418,5 +1441,70 @@ def scrypt(password=b"password", /, **kwargs):
         self.assertRaises(numeric_exc_types, scrypt, dklen=MAX_DKLEN + 1)
 
 
+@threading_helper.requires_working_threading()
+class TestTSAN(unittest.TestCase):
+
+    @threading_helper.reap_threads
+    def check_attribute(self, write, read, expected, nthreads=8):
+        ready = threading.Event()
+        barrier = threading.Barrier(nthreads)
+
+        def writer():
+            barrier.wait()
+            while not ready.is_set():
+                write()
+
+        def reader():
+            barrier.wait()
+            while not ready.is_set():
+                self.assertEqual(read(), expected)
+
+        targets = [writer if i % 2 else reader for i in range(nthreads)]
+        workers = [threading.Thread(target=target) for target in targets]
+        with threading_helper.start_threads(workers, unlock=ready.set):
+            pass
+
+    def check_HACL_attribute(self, module, version, attrname):
+        blob = b"A" * 65536
+        obj = getattr(module, version)()
+        update = partial(obj.update, blob)
+        read = attrgetter(attrname)
+        self.check_attribute(update, partial(read, obj), read(obj))
+
+    @requires_md5
+    @support.subTests("attrname", ["block_size", "digest_size"])
+    def test_HACL_md5_attributes(self, attrname):
+        self.check_HACL_attribute(_md5, "md5", attrname)
+
+    @requires_sha1
+    @support.subTests("attrname", ["block_size", "digest_size"])
+    def test_HACL_sha1_attributes(self, attrname):
+        self.check_HACL_attribute(_sha1, "sha1", attrname)
+
+    @requires_sha2
+    @support.subTests("size", [224, 256, 384, 512])
+    @support.subTests("attrname", ["block_size", "digest_size"])
+    def test_HACL_sha2_attributes(self, size, attrname):
+        self.check_HACL_attribute(_sha2, f"sha{size}", attrname)
+
+    @requires_sha3
+    @support.subTests("size", [224, 256, 384, 512])
+    @support.subTests(
+        "attrname",
+        ["block_size", "digest_size", "_capacity_bits", "_rate_bits"],
+    )
+    def test_HACL_sha3_attributes(self, size, attrname):
+        self.check_HACL_attribute(_sha3, f"sha3_{size}", attrname)
+
+    @requires_sha3
+    @support.subTests("size", [128, 256])
+    @support.subTests(
+        "attrname",
+        ["block_size", "digest_size", "_capacity_bits", "_rate_bits"],
+    )
+    def test_HACL_shake_attributes(self, size, attrname):
+        self.check_HACL_attribute(_sha3, f"shake_{size}", attrname)
+
+
 if __name__ == "__main__":
     unittest.main()

Misc/NEWS.d/next/Library/2026-08-15-13-12-09.gh-issue-155835.DsbSmy.rst

@@ -0,0 +1,3 @@
+:mod:`hashlib`: Fix data races when accessing
+:attr:`~hashlib.hash.digest_size` and :attr:`~hashlib.hash.block_size` on
+SHA-3 objects. Patch by Bénédikt Tran.

Modules/sha3module.c

@@ -67,6 +67,12 @@ sha3_get_state(PyObject *module)
 typedef struct {
     HASHLIB_OBJECT_HEAD
     Hacl_Hash_SHA3_state_t *hash_state;
+    // HACL* update functions entirely replace the state, which can lead
+    // to races on the free-threaded build. Since the kind of hash is static,
+    // we can store its corresponding metadata once.
+    uint32_t digest_size;
+    uint32_t block_size;
+    int is_shake;
 } SHA3object;
 
 #define _SHA3object_CAST(op)    ((SHA3object *)(op))
@@ -96,7 +102,7 @@ newSHA3object(PyTypeObject *type)
         return NULL;
     }
     HASHLIB_INIT_MUTEX(newobj);
-
+    newobj->digest_size = newobj->block_size = 0;
     PyObject_GC_Track(newobj);
     return newobj;
 }
@@ -179,6 +185,11 @@ py_sha3_new_impl(PyTypeObject *type, PyObject *data_obj, int usedforsecurity,
         goto error;
     }
 
+    // set the metadata once we know that the state is valid
+    int is_shake = Hacl_Hash_SHA3_is_shake(self->hash_state);
+    self->digest_size = is_shake ? 0 : Hacl_Hash_SHA3_hash_len(self->hash_state);
+    self->block_size = Hacl_Hash_SHA3_block_len(self->hash_state);
+
     if (data) {
         GET_BUFFER_VIEW_OR_ERROR(data, &buf, goto error);
         /* Do not use self->mutex here as this is the constructor
@@ -253,6 +264,8 @@ _sha3_sha3_224_copy_impl(SHA3object *self, PyTypeObject *cls)
         Py_DECREF(newobj);
         return PyErr_NoMemory();
     }
+    newobj->digest_size = self->digest_size;
+    newobj->block_size = self->block_size;
     return (PyObject *)newobj;
 }
 
@@ -273,8 +286,7 @@ _sha3_sha3_224_digest_impl(SHA3object *self)
     HASHLIB_ACQUIRE_LOCK(self);
     (void)Hacl_Hash_SHA3_digest(self->hash_state, digest);
     HASHLIB_RELEASE_LOCK(self);
-    return PyBytes_FromStringAndSize((const char *)digest,
-        Hacl_Hash_SHA3_hash_len(self->hash_state));
+    return PyBytes_FromStringAndSize((const char *)digest, self->digest_size);
 }
 
 
@@ -292,8 +304,7 @@ _sha3_sha3_224_hexdigest_impl(SHA3object *self)
     HASHLIB_ACQUIRE_LOCK(self);
     (void)Hacl_Hash_SHA3_digest(self->hash_state, digest);
     HASHLIB_RELEASE_LOCK(self);
-    return _Py_strhex((const char *)digest,
-        Hacl_Hash_SHA3_hash_len(self->hash_state));
+    return _Py_strhex((const char *)digest, self->digest_size);
 }
 
 
@@ -334,8 +345,7 @@ static PyObject *
 SHA3_get_block_size(PyObject *op, void *Py_UNUSED(closure))
 {
     SHA3object *self = _SHA3object_CAST(op);
-    uint32_t rate = Hacl_Hash_SHA3_block_len(self->hash_state);
-    return PyLong_FromLong(rate);
+    return PyLong_FromLong(self->block_size);
 }
 
 
@@ -371,18 +381,15 @@ SHA3_get_digest_size(PyObject *op, void *Py_UNUSED(closure))
 {
     // Preserving previous behavior: variable-length algorithms return 0
     SHA3object *self = _SHA3object_CAST(op);
-    if (Hacl_Hash_SHA3_is_shake(self->hash_state))
-      return PyLong_FromLong(0);
-    else
-      return PyLong_FromLong(Hacl_Hash_SHA3_hash_len(self->hash_state));
+    return PyLong_FromLong(self->digest_size);
 }
 
 
 static PyObject *
 SHA3_get_capacity_bits(PyObject *op, void *Py_UNUSED(closure))
 {
     SHA3object *self = _SHA3object_CAST(op);
-    uint32_t rate = Hacl_Hash_SHA3_block_len(self->hash_state) * 8;
+    uint32_t rate = self->block_size * 8;
     assert(rate <= 1600);
     int capacity = 1600 - rate;
     return PyLong_FromLong(capacity);
@@ -393,8 +400,7 @@ static PyObject *
 SHA3_get_rate_bits(PyObject *op, void *Py_UNUSED(closure))
 {
     SHA3object *self = _SHA3object_CAST(op);
-    uint32_t rate = Hacl_Hash_SHA3_block_len(self->hash_state) * 8;
-    return PyLong_FromLong(rate);
+    return PyLong_FromLong(self->block_size * 8);
 }
 
 static PyObject *