pytorch/torch/_dynamo/allowed_functions.py

import builtins
import collections
import copy
import functools
import inspect
import itertools
import math
import operator
import types
import warnings
from typing import cast, Dict, Optional, Set

import torch
from torch.fx._symbolic_trace import is_fx_tracing

from . import config
from .external_utils import is_compiling
from .utils import HAS_NUMPY, is_safe_constant, np, NP_SUPPORTED_MODULES

"""
A note on allowed functions:

Dynamo consults this file to determine if a particular function/module
is allowed to appear as a node in its fx output.

If a function is disallowed, it may either be traced-through, or skipped.

Trace-through means dynamo will continue to trace the interior code for
the function/module rather than stopping at its boundary and recording it
as a node in the fx graph. Whether tracing through or allowing, the functionality
of the function/module is part of the dynamo graph.  Caveat: if tracing through,
any interior operation could trigger its own graph-break.

Skips are determined by (torch/_dynamo/skipfiles.py) - see "a note on
skipfiles" there.
"""


def make_function_id_set(lazy_initializer):
    """
    Track a set of `id()`s of objects which are either allowed or not
    allowed to go into the generated FX graph.  Use to test for torch.*,
    numpy.*, builtins.*, etc.

    Support user modification to permit customization of what can be
    added to the graph and what will cause a graph break.
    """

    class FunctionIdSet:
        function_ids: Optional[Set[int]] = None
        function_names: Optional[Dict[int, str]] = None

        def __call__(self):
            if self.function_ids is None:
                value = lazy_initializer()
                if isinstance(value, dict):
                    self.function_ids = set(value.keys())
                    self.function_names = value
                else:
                    assert isinstance(value, set)
                    self.function_ids = value
            return self.function_ids

        def get_name(self, idx: int, default: str):
            self()  # lazy init
            return self.function_names.get(idx, default)

        def add(self, idx: int):
            self()  # lazy init
            self.function_ids.add(idx)

        def remove(self, idx: int):
            if idx in self():
                self.function_ids.remove(idx)

        def __contains__(self, idx: int):
            return idx in self()

    return FunctionIdSet()


@make_function_id_set
def _disallowed_function_ids():
    remove = [
        True,
        False,
        None,
        collections.OrderedDict,
        copy.copy,
        copy.deepcopy,
        inspect.signature,
        math.__package__,
        torch.__builtins__,
        torch.autocast_decrement_nesting,
        torch.autocast_increment_nesting,
        torch.autograd.grad,
        torch.clear_autocast_cache,
        torch.cuda.current_device,
        torch.cuda.set_device,
        torch.distributions.constraints.is_dependent,
        torch.distributions.normal.Normal,
        torch.inference_mode,
        torch.set_anomaly_enabled,
        torch.set_autocast_cache_enabled,
        torch.set_autocast_cpu_dtype,
        torch.set_autocast_cpu_enabled,
        torch.set_autocast_enabled,
        torch.set_autocast_gpu_dtype,
        warnings.warn,
        torch._C._dynamo.eval_frame.unsupported,
    ]
    if torch.distributed.is_available():
        from torch.distributed import _functional_collectives

        config.skipfiles_inline_module_allowlist.add(_functional_collectives)

    # extract all dtypes from torch
    dtypes = [
        obj for obj in torch.__dict__.values() if isinstance(obj, type(torch.float32))
    ]
    remove += dtypes
    storage = [
        obj
        for obj in torch.__dict__.values()
        if isinstance(obj, type(torch.FloatStorage))
    ]
    remove += storage

    # Distributed APIs don't work well with torch.compile.
    if torch.distributed.is_available():
        remove.extend(
            torch.distributed.distributed_c10d.dynamo_unsupported_distributed_c10d_ops
        )

    return {id(x) for x in remove}


@make_function_id_set
def _allowed_function_ids():
    """
    Walk torch.* and get the ids of all the stuff in it
    """
    warnings.filterwarnings("ignore", category=UserWarning, module="torch.distributed")
    torch_object_ids = dict()

    def _is_allowed_module_prefix(obj):
        allowed_modules = ("torch", "math")
        # torch.nn.modules.rnn is disallowed because these modules internally
        # flatten their parameters.  This flattening process will call
        # Tensor.set_ with a Storage, and Storages cannot be traced with
        # AOTAutograd; so we need to graph-break. To ensure this, we inline
        # these functions, rather than keep them opaque-ly in the graph.
        disallowed_modules = (
            "torch.optim.",
            "torch.utils._foreach_utils",  # omit the period so we match all the functions in this module
            "torch.nn.modules.rnn.",
            "torch._dynamo.",
            "torch._C._dynamo.",
            "torch._inductor.",
            "torch._C.inductor.",
            "torch.fx.",
            "torch.distributed.fsdp.",
            "torch.distributed._tensor.",
        )
        allowed_modules_dot = tuple([x + "." for x in allowed_modules])
        module = inspect.getmodule(obj)
        if module is None:
            return False

        mod_name = module.__name__

        if any(mod_name.startswith(m) for m in disallowed_modules):
            return False

        return mod_name in allowed_modules or mod_name.startswith(allowed_modules_dot)

    def _find_torch_objects(module):
        if any(
            module.__name__.startswith(mod_name)
            for mod_name in config.allowed_functions_module_string_ignorelist
        ):
            return
        torch_object_ids[id(module)] = module.__name__
        for name, obj in list(module.__dict__.items()):
            if id(obj) not in torch_object_ids:
                # Dynamo allows all builtins into the graph and does not attempt
                # to introspect into them. We don't want to allow instances of
                # HigherOrderOperator into the graph all the time (Dynamo needs
                # to introspect the body functions of these HigherOrderOperator
                # first, decide they are safe, and then allow them into the graph).
                # So we exclude HigherOrderOperator from being a builtin.
                import torch._ops

                if isinstance(obj, torch._ops.HigherOrderOperator):
                    continue
                # We want to trace through `grad`
                if obj is torch.func.grad:
                    continue

                if isinstance(obj, types.ModuleType):
                    if obj.__name__.startswith("torch.") and _is_allowed_module_prefix(
                        obj
                    ):
                        torch_object_ids[id(obj)] = f"{module.__name__}.{name}"
                        _find_torch_objects(obj)
                elif _is_allowed_module_prefix(obj):
                    torch_object_ids[id(obj)] = f"{module.__name__}.{name}"
                elif inspect.getmodule(obj) is None and not is_safe_constant(obj):
                    torch_object_ids[id(obj)] = f"{module.__name__}.{name}"

    _find_torch_objects(torch)
    _find_torch_objects(math)

    # torch.Tensor.{fn}
    for name in dir(torch.Tensor):
        method = getattr(torch.Tensor, name)
        if isinstance(method, types.MethodDescriptorType):
            torch_object_ids[id(method)] = f"torch.Tensor.{name}"

    for idx in _disallowed_function_ids():
        if idx in torch_object_ids:
            del torch_object_ids[idx]

    for extra in (is_fx_tracing, is_compiling):
        torch_object_ids[id(extra)] = f"{extra.__module__}.{extra.__name__}"

    return torch_object_ids


@make_function_id_set
def _builtin_function_ids():
    rv = {
        id(v): f"builtins.{k}"
        for k, v in builtins.__dict__.items()
        if not k.startswith("_") and callable(v)
    }
    rv.update(
        {
            id(v): f"operator.{k}"
            for k, v in operator.__dict__.items()
            if not k.startswith("_") and callable(v)
        }
    )
    rv.update(
        {id(v): f"functools.{v.__name__}" for v in (itertools.chain, itertools.islice)}
    )
    rv.update({id(cast): "typing.cast"})
    rv[id(functools.reduce)] = "functools.reduce"
    return rv


@make_function_id_set
def _numpy_function_ids():
    rv = dict()
    if HAS_NUMPY:
        for mod in NP_SUPPORTED_MODULES:
            rv.update(
                {
                    id(v): f"{mod.__name__}.{k}"
                    for k, v in mod.__dict__.items()
                    if callable(v)
                    and (getattr(v, "__module__", None) or mod.__name__) == mod.__name__
                }
            )
    return rv


@make_function_id_set
def _builtin_constant_ids():
    """
    Collects constant builtins by eliminating callable items.
    """
    rv = {
        id(v): f"builtins.{k}"
        for k, v in builtins.__dict__.items()
        if not k.startswith("_") and not callable(v)
    }
    return rv


def is_allowed(obj):
    """Is this safe to trace like torch.add ?"""
    # torch.ops is populated lazily so we don't necessarily have them in
    # _allowed_function_ids.  Figure it out by testing the type instead
    # in those cases
    if id(obj) in _disallowed_function_ids:
        return False

    return id(obj) in _allowed_function_ids or isinstance(
        obj,
        (torch._ops.OpOverloadPacket, torch._ops.OpOverload, torch._ops._OpNamespace),
    )


def torch_get_name(obj, default):
    """Convert a torch.* function to a string"""
    return _allowed_function_ids.get_name(id(obj), default)


def is_builtin_callable(obj):
    return id(obj) in _builtin_function_ids


def is_builtin_constant(obj):
    return id(obj) in _builtin_constant_ids


def is_numpy(obj):
    if HAS_NUMPY:
        return isinstance(obj, np.ndarray) or id(obj) in _numpy_function_ids
    else:
        return False