pytorch · atalman · Jun 26, 2024 · Jun 25, 2024 · Jun 25, 2024 · Jun 25, 2024
diff --git a/test/distributed/_tensor/test_dtensor.py b/test/distributed/_tensor/test_dtensor.py
@@ -536,6 +536,16 @@ def test_dtensor_save_load(self):
         buffer.seek(0)
         reloaded_st = torch.load(buffer)
         self.assertEqual(sharded_tensor, reloaded_st)
+        # Test weights_only load
+        try:
+            torch.serialization.add_safe_globals(
+                [DTensor, DeviceMesh, Shard, DTensorSpec, TensorMeta]
+            )
+            buffer.seek(0)
+            reloaded_st = torch.load(buffer, weights_only=True)
+            self.assertEqual(sharded_tensor, reloaded_st)
+        finally:
+            torch.serialization.clear_safe_globals()
 
 
 class DTensorMeshTest(DTensorTestBase):

diff --git a/test/test_nn.py b/test/test_nn.py
@@ -1801,26 +1801,35 @@ def test_parameterlistdict_pickle(self):
         m = nn.ParameterList(map(nn.Parameter, [torch.rand(2), torch.rand(2)]))
         with warnings.catch_warnings(record=True) as w:
             m = pickle.loads(pickle.dumps(m))
-        self.assertTrue(len(w) == 0)
+        # warning from torch.load call in _load_from_bytes
+        num_warnings = 2 if torch._dynamo.is_compiling() else 1
+        self.assertTrue(len(w) == num_warnings)
+        self.assertEqual(w[0].category, FutureWarning)
 
         # Test whether loading from older checkpoints works without triggering warnings
         m = nn.ParameterList(map(nn.Parameter, [torch.rand(2), torch.rand(2)]))
         del m._forward_pre_hooks, m._state_dict_hooks, m._load_state_dict_pre_hooks, m._non_persistent_buffers_set
         with warnings.catch_warnings(record=True) as w:
             m = pickle.loads(pickle.dumps(m))
-        self.assertTrue(len(w) == 0)
+        # warning from torch.load call in _load_from_bytes
+        self.assertTrue(len(w) == 1)
+        self.assertEqual(w[0].category, FutureWarning)
 
         m = nn.ParameterDict({"a": nn.Parameter(torch.rand(2)), "b": nn.Parameter(torch.rand(2))})
         with warnings.catch_warnings(record=True) as w:
             m = pickle.loads(pickle.dumps(m))
-        self.assertTrue(len(w) == 0)
+        # warning from torch.load call in _load_from_bytes
+        self.assertTrue(len(w) == 1)
+        self.assertEqual(w[0].category, FutureWarning)
 
         # Test whether loading from older checkpoints works without triggering warnings
         m = nn.ParameterDict({"a": nn.Parameter(torch.rand(2)), "b": nn.Parameter(torch.rand(2))})
         del m._forward_pre_hooks, m._state_dict_hooks, m._load_state_dict_pre_hooks, m._non_persistent_buffers_set
         with warnings.catch_warnings(record=True) as w:
             m = pickle.loads(pickle.dumps(m))
-        self.assertTrue(len(w) == 0)
+        # warning from torch.load call in _load_from_bytes
+        self.assertTrue(len(w) == 1)
+        self.assertEqual(w[0].category, FutureWarning)
 
     def test_weight_norm_pickle(self):
         m = torch.nn.utils.weight_norm(nn.Linear(5, 7))

diff --git a/test/test_serialization.py b/test/test_serialization.py
@@ -15,7 +15,7 @@
 import shutil
 import pathlib
 import platform
-from collections import OrderedDict
+from collections import namedtuple, OrderedDict
 from copy import deepcopy
 from itertools import product
 
@@ -804,6 +804,17 @@ def wrapper(*args, **kwargs):
     def __exit__(self, *args, **kwargs):
         torch.save = self.torch_save
 
+Point = namedtuple('Point', ['x', 'y'])
+
+class ClassThatUsesBuildInstruction:
+    def __init__(self, num):
+        self.num = num
+
+    def __reduce_ex__(self, proto):
+        # Third item, state here will cause pickle to push a BUILD instruction
+        return ClassThatUsesBuildInstruction, (self.num,), {'foo': 'bar'}
+
+
 @unittest.skipIf(IS_WINDOWS, "NamedTemporaryFile on windows")
 class TestBothSerialization(TestCase):
     @parametrize("weights_only", (True, False))
@@ -826,7 +837,6 @@ def test(f_new, f_old):
                 test(f_new, f_old)
             self.assertTrue(len(w) == 0, msg=f"Expected no warnings but got {[str(x) for x in w]}")
 
-
 class TestOldSerialization(TestCase, SerializationMixin):
     # unique_key is necessary because on Python 2.7, if a warning passed to
     # the warning module is the same, it is not raised again.
@@ -854,7 +864,8 @@ def import_module(name, filename):
                 loaded = torch.load(checkpoint)
                 self.assertTrue(isinstance(loaded, module.Net))
                 if can_retrieve_source:
-                    self.assertEqual(len(w), 0)
+                    self.assertEqual(len(w), 1)
+                    self.assertEqual(w[0].category, FutureWarning)
 
             # Replace the module with different source
             fname = get_file_path_2(os.path.dirname(os.path.dirname(torch.__file__)), 'torch', 'testing',
@@ -865,8 +876,8 @@ def import_module(name, filename):
                 loaded = torch.load(checkpoint)
                 self.assertTrue(isinstance(loaded, module.Net))
                 if can_retrieve_source:
-                    self.assertEqual(len(w), 1)
-                    self.assertTrue(w[0].category, 'SourceChangeWarning')
+                    self.assertEqual(len(w), 2)
+                    self.assertTrue(w[1].category, 'SourceChangeWarning')
 
     def test_serialization_container(self):
         self._test_serialization_container('file', tempfile.NamedTemporaryFile)
@@ -1040,8 +1051,63 @@ def __reduce__(self):
             self.assertIsNone(torch.load(f, weights_only=False))
             f.seek(0)
             # Safe load should assert
-            with self.assertRaisesRegex(pickle.UnpicklingError, "Unsupported global: GLOBAL __builtin__.print"):
+            with self.assertRaisesRegex(pickle.UnpicklingError, "Unsupported global: GLOBAL builtins.print"):
+                torch.load(f, weights_only=True)
+            try:
+                torch.serialization.add_safe_globals([print])
+                f.seek(0)
+                torch.load(f, weights_only=True)
+            finally:
+                torch.serialization.clear_safe_globals()
+
+    def test_weights_only_safe_globals_newobj(self):
+        # This will use NEWOBJ
+        p = Point(x=1, y=2)
+        with BytesIOContext() as f:
+            torch.save(p, f)
+            f.seek(0)
+            with self.assertRaisesRegex(pickle.UnpicklingError,
+                                        "GLOBAL __main__.Point was not an allowed global by default"):
                 torch.load(f, weights_only=True)
+            f.seek(0)
+            try:
+                torch.serialization.add_safe_globals([Point])
+                loaded_p = torch.load(f, weights_only=True)
+                self.assertEqual(loaded_p, p)
+            finally:
+                torch.serialization.clear_safe_globals()
+
+    def test_weights_only_safe_globals_build(self):
+        counter = 0
+
+        def fake_set_state(obj, *args):
+            nonlocal counter
+            counter += 1
+
+        c = ClassThatUsesBuildInstruction(2)
+        with BytesIOContext() as f:
+            torch.save(c, f)
+            f.seek(0)
+            with self.assertRaisesRegex(pickle.UnpicklingError,
+                                        "GLOBAL __main__.ClassThatUsesBuildInstruction was not an allowed global by default"):
+                torch.load(f, weights_only=True)
+            try:
+                torch.serialization.add_safe_globals([ClassThatUsesBuildInstruction])
+                # Test dict update path
+                f.seek(0)
+                loaded_c = torch.load(f, weights_only=True)
+                self.assertEqual(loaded_c.num, 2)
+                self.assertEqual(loaded_c.foo, 'bar')
+                # Test setstate path
+                ClassThatUsesBuildInstruction.__setstate__ = fake_set_state
+                f.seek(0)
+                loaded_c = torch.load(f, weights_only=True)
+                self.assertEqual(loaded_c.num, 2)
+                self.assertEqual(counter, 1)
+                self.assertFalse(hasattr(loaded_c, 'foo'))
+            finally:
+                torch.serialization.clear_safe_globals()
+                ClassThatUsesBuildInstruction.__setstate__ = None
 
     @parametrize('weights_only', (False, True))
     def test_serialization_math_bits(self, weights_only):

diff --git a/torch/_weights_only_unpickler.py b/torch/_weights_only_unpickler.py
@@ -23,6 +23,7 @@
 # weights = torch.load(buf, weights_only = True)
 
 import functools as _functools
+import warnings
 from collections import Counter, OrderedDict
 from pickle import (
     APPEND,
@@ -67,6 +68,16 @@
 from sys import maxsize
 from typing import Any, Dict, List
 
+try:
+    # We rely on this module in private cPython which provides dicts of
+    # modules/functions that had their names changed from Python 2 to 3
+    has_compat_pickle = True
+    from _compat_pickle import IMPORT_MAPPING, NAME_MAPPING
+except ImportError:
+    # To prevent warning on import torch, we warn in the Unpickler.load below
+    has_compat_pickle = False
+    IMPORT_MAPPING, NAME_MAPPING = dict(), dict()
+
 import torch
 
 _marked_safe_globals_list: List[Any] = []
@@ -97,7 +108,8 @@ def _clear_safe_globals():
 def _get_user_allowed_globals():
     rc: Dict[str, Any] = {}
     for f in _marked_safe_globals_list:
-        rc[f"{f.__module__}.{f.__name__}"] = f
+        module, name = f.__module__, f.__name__
+        rc[f"{module}.{name}"] = f
     return rc
 
 
@@ -170,12 +182,20 @@ def __init__(self, file, *, encoding: str = "bytes"):
         self.readline = file.readline
         self.read = file.read
         self.memo: Dict[int, Any] = {}
+        self.proto: int = -1
 
     def load(self):
         """Read a pickled object representation from the open file.
 
         Return the reconstituted object hierarchy specified in the file.
         """
+        if not has_compat_pickle:
+            warnings.warn(
+                "Could not import IMPORT_MAPPING and NAME_MAPPING from _compat_pickle. "
+                "If the default `pickle_protocol` was used at `torch.save` time, any functions or "
+                "classes that are in these maps might not behave correctly if allowlisted via "
+                "`torch.serialization.add_safe_globals()`."
+            )
         self.metastack = []
         self.stack: List[Any] = []
         self.append = self.stack.append
@@ -190,6 +210,13 @@ def load(self):
             if key[0] == GLOBAL[0]:
                 module = readline()[:-1].decode("utf-8")
                 name = readline()[:-1].decode("utf-8")
+                # Patch since torch.save default protocol is 2
+                # users will be running this code in python > 3
+                if self.proto == 2 and has_compat_pickle:
+                    if (module, name) in NAME_MAPPING:
+                        module, name = NAME_MAPPING[(module, name)]
+                    elif module in IMPORT_MAPPING:
+                        module = IMPORT_MAPPING[module]
                 full_path = f"{module}.{name}"
                 if full_path in _get_allowed_globals():
                     self.append(_get_allowed_globals()[full_path])
@@ -204,9 +231,12 @@ def load(self):
             elif key[0] == NEWOBJ[0]:
                 args = self.stack.pop()
                 cls = self.stack.pop()
-                if cls is not torch.nn.Parameter:
+                if cls is torch.nn.Parameter:
+                    self.append(torch.nn.Parameter(*args))
+                elif cls in _get_user_allowed_globals().values():
+                    self.append(cls.__new__(cls, *args))
+                else:
                     raise RuntimeError(f"Trying to instantiate unsupported class {cls}")
-                self.append(torch.nn.Parameter(*args))
             elif key[0] == REDUCE[0]:
                 args = self.stack.pop()
                 func = self.stack[-1]
@@ -228,9 +258,14 @@ def load(self):
                     inst.__setstate__(state)
                 elif type(inst) is OrderedDict:
                     inst.__dict__.update(state)
+                elif type(inst) in _get_user_allowed_globals().values():
+                    if hasattr(inst, "__setstate__"):
+                        inst.__setstate__(state)
+                    else:
+                        inst.__dict__.update(state)
                    else:
                     raise RuntimeError(
-                        f"Can only build Tensor, parameter or dict objects, but got {type(inst)}"
+                        f"Can only build Tensor, parameter or OrderedDict objects, but got {type(inst)}"
                     )
             # Stack manipulation
             elif key[0] == APPEND[0]:
@@ -334,8 +369,14 @@ def load(self):
                 self.append(decode_long(data))
             # First and last deserializer ops
             elif key[0] == PROTO[0]:
-                # Read and ignore proto version
-                read(1)[0]
+                self.proto = read(1)[0]
+                if self.proto != 2:
+                    warnings.warn(
+                        f"Detected pickle protocol {self.proto} in the checkpoint, which was "
+                        "not the default pickle protocol used by `torch.load` (2). The weights_only "
+                        "Unpickler might not support all instructions implemented by this protocol, "
+                        "please file an issue for adding support if you encounter this."
+                    )
             elif key[0] == STOP[0]:
                 rc = self.stack.pop()
                 return rc

@@ -165,12 +165,30 @@
     return _weights_only_unpickler._get_safe_globals()
 
 def add_safe_globals(safe_globals: List[Any]) -> None:
-    '''
-    Marks the given globals as safe for ``weights_only`` load.
+    """
+    Marks the given globals as safe for ``weights_only`` load. For example, functions
+    added to this list can be called during unpickling, classes could be instantiated
+    and have state set.
 
     Args:
         safe_globals (List[Any]): list of globals to mark as safe
-    '''
+
+    Example:
+        >>> # xdoctest: +SKIP("Can't torch.save(t, ...) as doctest thinks MyTensor is defined on torch.serialization")
+        >>> import tempfile
+        >>> class MyTensor(torch.Tensor):
+        ...     pass
+        >>> t = MyTensor(torch.randn(2, 3))
+        >>> with tempfile.NamedTemporaryFile() as f:
+        ...     torch.save(t, f.name)
+        # Running `torch.load(f.name, weights_only=True)` will fail with
+        # Unsupported global: GLOBAL __main__.MyTensor was not an allowed global by default.
+        # Check the code and make sure MyTensor is safe to be used when loaded from an arbitrary checkpoint.
+        ...     torch.serialization.add_safe_globals([MyTensor])
+        ...     torch.load(f.name, weights_only=True)
+        # MyTensor([[-0.5024, -1.8152, -0.5455],
+        #          [-0.8234,  2.0500, -0.3657]])
+    """
     _weights_only_unpickler._add_safe_globals(safe_globals)
 
 def _is_zipfile(f) -> bool:
@@ -867,12 +885,12 @@
        zip_file.write_record(name, storage, num_bytes)


 def load(
    f: FILE_LIKE,
     map_location: MAP_LOCATION = None,
     pickle_module: Any = None,
     *,
-    weights_only: bool = False,
+    weights_only: Optional[bool] = None,
     mmap: Optional[bool] = None,
     **pickle_load_args: Any
 ) -> Any:
@@ -982,6 +1000,11 @@
         " with `weights_only` please check the recommended steps in the following error message."
         " WeightsUnpickler error: "
     )
+    if weights_only is None:
+        weights_only, warn_weights_only = False, True
+    else:
+        warn_weights_only = False
+
     # Add ability to force safe only weight loads via environment variable
     if os.getenv("TORCH_FORCE_WEIGHTS_ONLY_LOAD", "0").lower() in ['1', 'y', 'yes', 'true']:
         weights_only = True
@@ -991,6 +1014,20 @@
             raise RuntimeError("Can not safely load weights when explicit pickle_module is specified")
     else:
         if pickle_module is None:
+            if warn_weights_only:
+                warnings.warn(
+                    "You are using `torch.load` with `weights_only=False` (the current default value), which uses "
+                    "the default pickle module implicitly. It is possible to construct malicious pickle data "
+                    "which will execute arbitrary code during unpickling (See "
+                    "https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). "
+                    "In a future release, the default value for `weights_only` will be flipped to `True`. This "
+                    "limits the functions that could be executed during unpickling. Arbitrary objects will no "
+                    "longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the "
+                    "user via `torch.serialization.add_safe_globals`. We recommend you start setting "
+                    "`weights_only=True` for any use case where you don't have full control of the loaded file. "
+                    "Please open an issue on GitHub for any issues related to this experimental feature.",
+                    FutureWarning,
+                )
             pickle_module = pickle
 
     # make flipping default BC-compatible