Add support for SchemaKind.mutable to to_out_variant (#2541)

Summary:
Pull Request resolved: #2541

Ideally custom mutator ops would not be in aten OpOverload but would instead be edge ops. The  pass that results in mutator ops being in the graph is shared with core for now though and its hard to mix and match EP passes and GM passes. So just add the logic to the otherwise unused aten to_out_variant code.

Reviewed By: larryliu0820

Differential Revision: D55156605

fbshipit-source-id: 3003a51b83bb211345c1d6c6e5bb7e97c19dfeb1

Author: JacobSzwejbka

exir/operator/TARGETS

@@ -36,11 +36,12 @@ python_library(
 )
 
 python_unittest(
-    name = "test_util",
+    name = "test_operator",
     srcs = [
-        "test/test_util.py",
+        "test/test_operator.py",
     ],
     deps = [
+        ":convert",
         ":util",
         "//caffe2:torch",
     ],

exir/operator/convert.py

@@ -41,6 +41,7 @@
 # existing case with cache miss.
 _func_to_out_variant_map: Dict[OpOverload, Optional[OpOverload]] = {}
 _out_variant_to_scratch_map: Dict[OpOverload, Optional[OpOverload]] = {}
+_mutable_to_out_variant_map: Dict[OpOverload, Optional[OpOverload]] = {}
 
 # We've found a functional and an out variant with the same name, but their
 # schemas mismatch. This map collects all of these cases and provides proper
@@ -135,9 +136,9 @@ def schema_to_opoverload(schema: FunctionSchema) -> OpOverload:
 
 def set_mapping_for_op(op: OpOverload) -> None:
     """
-    op can either be a functional op or out variant op.
+    op can either be a functional op, mutable op, or out variant op.
     This method is only called if
-    1. either op is a functiona op and it's missing in the _func_to_out_variant_map cache.
+    1. either op is a functional op and it's missing in the _func_to_out_variant_map cache.
     2. or op is a out variant op and it's missing in the _out_variant_to_scratch_map cache.
 
     Setup entries in _func_to_out_variant_map and _out_variant_to_scratch_map for all ops sharing the same
@@ -148,13 +149,17 @@ def set_mapping_for_op(op: OpOverload) -> None:
     assert native_schema.kind() in (
         SchemaKind.functional,
         SchemaKind.out,
+        SchemaKind.mutable,
     )
     assert not (
         native_schema.kind() == SchemaKind.functional and op in _func_to_out_variant_map
     )
     assert not (
         native_schema.kind() == SchemaKind.out and op in _out_variant_to_scratch_map
     )
+    assert not (
+        native_schema.kind() == SchemaKind.mutable and op in _mutable_to_out_variant_map
+    )
     qualified_opname = str(op._schema.name)
 
     all_schemas = [
@@ -192,6 +197,7 @@ def set_mapping_for_op(op: OpOverload) -> None:
     for group_by_kind in group_by_signature.values():
         func_op_schema = group_by_kind.get(SchemaKind.functional)
         out_var_schema = group_by_kind.get(SchemaKind.out)
+        mutable_op_schema = group_by_kind.get(SchemaKind.mutable)
         scratch_schema = group_by_kind.get(SchemaKind.scratch)
 
         # update the map even if out_var_schema is None to cache the negative
@@ -216,30 +222,41 @@ def set_mapping_for_op(op: OpOverload) -> None:
             _out_variant_to_scratch_map[schema_to_opoverload(out_var_schema)] = (
                 schema_to_opoverload(scratch_schema) if scratch_schema else None
             )
+        if mutable_op_schema:
+            _mutable_to_out_variant_map[schema_to_opoverload(mutable_op_schema)] = (
+                schema_to_opoverload(out_var_schema) if out_var_schema else None
+            )
 
 
 def to_out_variant(op_overload: OpOverload) -> Tuple[OpOverload, Tuple[str]]:
     r"""
     Convert the passed in OpOverload to its out variant. Raise an exception if
     on return the op_overload is not guaranteed to be an out variant.
 
-    If a conversion is found, return the out variant OpOverlaod alongwith the name of out
+    If a conversion is found, return the out variant OpOverload alongwith the name of out
     arguments.
     """
     schema = _get_overload_schema(op_overload)
     if schema.is_out_fn():  # pyre-ignore
-        return op_overload, get_out_args_from_schema(schema)  # pyre-ignore
+        return op_overload, get_out_args_from_schema(schema)  # pyre-ignore[6]
 
-    # should be a functional op here
+    # should be a functionalish op here
     assert (
-        schema.kind() == SchemaKind.functional  # pyre-ignore
-    ), f"Expect an functional op, but get {schema}"
-
-    if op_overload not in _func_to_out_variant_map:
+        schema.kind() == SchemaKind.functional  # pyre-ignore[16]
+        or schema.kind() == SchemaKind.mutable
+    ), f"Expect a functionalish op, but get {schema.kind()} {schema}"
+
+    if (
+        op_overload not in _func_to_out_variant_map
+        and op_overload not in _mutable_to_out_variant_map
+    ):
         # setup out_var
         set_mapping_for_op(op_overload)
 
-    out_var = _func_to_out_variant_map.get(op_overload)
+    if op_overload in _mutable_to_out_variant_map:
+        out_var = _mutable_to_out_variant_map[op_overload]
+    else:
+        out_var = _func_to_out_variant_map.get(op_overload)
 
     if not out_var:
         msg = f"Missing out variant for functional op: {schema} . Make sure you have loaded your custom operator library for compiler. E.g., custom_ops_generated_lib"

exir/operator/test/test_operator.py

@@ -0,0 +1,84 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+# pyre-strict
+
+import unittest
+
+import torch
+from executorch.exir.operator.convert import _get_overload_schema, to_out_variant
+from executorch.exir.operator.util import gen_out_variant_schema
+from torch.library import _scoped_library, impl, impl_abstract
+
+
+class TestOperator(unittest.TestCase):
+    def setUp(self) -> None:
+        super().setUp()
+
+    def test_gen_out_variant_schema_from_functional(self) -> None:
+        func_schema = str(torch.ops.aten.mul.Scalar._schema)
+
+        out_schema = gen_out_variant_schema(func_schema)
+        self.assertEqual(out_schema, str(torch.ops.aten.mul.Scalar_out._schema))
+
+    def test_gen_out_variant_schema_from_inplace(self) -> None:
+        func_schema = str(torch.ops.aten.add_.Scalar._schema)
+
+        out_schema = gen_out_variant_schema(func_schema)
+        self.assertEqual(out_schema, str(torch.ops.aten.add.Scalar_out._schema))
+
+    def test_gen_out_variant_schema_for_custom_ops(self) -> None:
+        func_schema = "custom::foo(Tensor a, Tensor b) -> (Tensor c, Tensor d)"
+
+        out_schema = gen_out_variant_schema(func_schema)
+        self.assertEqual(
+            out_schema,
+            "custom::foo.out(Tensor a, Tensor b, *, Tensor(a!) out0, Tensor(b!) out1) -> (Tensor(a!), Tensor(b!))",
+        )
+
+    def test_to_out_variant_mutable(self) -> None:
+
+        with _scoped_library("DO_NOT_USE_TEST_ONLY", "DEF") as lib:
+
+            lib.define("custom_mutator(Tensor x, Tensor(a!) y) -> Tensor")
+            lib.define(
+                "custom_mutator.out(Tensor x, Tensor(a!) y, *, Tensor(b!) out) -> Tensor(b!)"
+            )
+
+            @impl(lib, "custom_mutator", "Meta")
+            def custom_mutator_meta(
+                x: torch.Tensor,
+                y: torch.Tensor,
+            ) -> torch.Tensor:
+                return torch.empty_like(x)
+
+            @impl(lib, "custom_mutator", "CompositeExplicitAutograd")
+            def custom_mutator(
+                x: torch.Tensor,
+                y: torch.Tensor,
+            ) -> torch.Tensor:
+                return x + y.add_(1)
+
+            @impl_abstract("DO_NOT_USE_TEST_ONLY::custom_mutator.out")
+            def custom_mutator_out(
+                x: torch.Tensor,
+                y: torch.Tensor,
+                out: torch.Tensor,
+            ) -> torch.Tensor:
+                out = custom_mutator_meta(
+                    x,
+                    y,
+                )
+                return out
+
+            out, _ = to_out_variant(
+                torch.ops.DO_NOT_USE_TEST_ONLY.custom_mutator.default
+            )
+            schema = _get_overload_schema(out)
+            self.assertEqual(
+                schema.__str__(),
+                "DO_NOT_USE_TEST_ONLY::custom_mutator.out(Tensor x, Tensor(a!) y, *, Tensor(b!) out) -> Tensor(b!)",
+            )