[dy2static] support fallback for whole graph. (stage 1) (#50111)

* [dy2static] support fallback for whole graph. (stage 1)

* bug fix

* bug fix and add a new unittest

* fix code by code review

* fix coverage

python/paddle/fluid/tests/unittests/dygraph_to_static/test_fallback.py

+# Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import print_function
+
+import unittest
+
+import numpy as np
+
+import paddle
+
+
+def support_func(x):
+    return 2 * x
+
+
+def unsupport_func(x):
+    x = 2 * x
+    t = x.numpy()
+    t = np.ones(t)
+    return paddle.to_tensor(t)
+
+
+class SuppportNet(paddle.nn.Layer):
+    def __init__(self):
+        super().__init__()
+
+    def forward(self, x):
+        return support_func(x)
+
+
+class UnsuppportNet(paddle.nn.Layer):
+    def __init__(self):
+        super().__init__()
+
+    def forward(self, x):
+        if self.training:
+            return unsupport_func(x)
+        else:
+            return unsupport_func(x - 1)
+
+
+class TestFallback(unittest.TestCase):
+    def setUp(self):
+        self.x = paddle.to_tensor(2).astype('int')
+
+    def tearDown(self):
+        pass
+
+    def test_case_support(self):
+        output = paddle.jit.to_static(support_func)(self.x)
+        np.testing.assert_allclose(output.numpy(), 4)
+
+    def test_case_func_fallback(self):
+        build_strategy = paddle.static.BuildStrategy()
+        build_strategy.build_cinn_pass = True
+        output = paddle.jit.to_static(
+            unsupport_func, build_strategy=build_strategy
+        )(self.x)
+        np.testing.assert_allclose(output.numpy(), unsupport_func(self.x))
+
+    def test_case_net_fallback(self):
+        s_net = SuppportNet()
+        u_net = UnsuppportNet()
+        np.testing.assert_allclose(
+            paddle.jit.to_static(s_net)(self.x).numpy(), 4
+        )
+        build_strategy = paddle.static.BuildStrategy()
+        build_strategy.build_cinn_pass = True
+        np.testing.assert_allclose(
+            paddle.jit.to_static(u_net, build_strategy=build_strategy)(
+                self.x
+            ).numpy(),
+            u_net(self.x).numpy(),
+        )
+
+    def test_case_net_error(self):
+        s_net = SuppportNet()
+        u_net = UnsuppportNet()
+        np.testing.assert_allclose(
+            paddle.jit.to_static(s_net)(self.x).numpy(), 4
+        )
+        build_strategy = paddle.static.BuildStrategy()
+        build_strategy.build_cinn_pass = False
+        with self.assertRaises(TypeError):
+            np.testing.assert_allclose(
+                paddle.jit.to_static(u_net, build_strategy=build_strategy)(
+                    self.x
+                ).numpy(),
+                u_net(self.x).numpy(),
+            )
+
+    def test_case_training(self):
+        build_strategy = paddle.static.BuildStrategy()
+        build_strategy.build_cinn_pass = True
+        u_net = paddle.jit.to_static(
+            UnsuppportNet(), build_strategy=build_strategy
+        )
+        u_net.eval()
+        np.testing.assert_allclose(u_net(self.x).numpy(), [1, 1])
+        assert u_net.training is False, "Training must be false."
+
+    def test_case_save_error(self):
+        """
+        test the save will raise error.
+        """
+        u_net = UnsuppportNet()
+        u_net = paddle.jit.to_static(
+            u_net, input_spec=[paddle.static.InputSpec(name='x', shape=[1])]
+        )
+        with self.assertRaises(TypeError):
+            paddle.jit.save(u_net, path="model")
+
+    def test_case_save_error_2(self):
+        """
+        test the save will raise error.
+        """
+        u_net = UnsuppportNet()
+        build_strategy = paddle.static.BuildStrategy()
+        build_strategy.build_cinn_pass = True
+        u_net = paddle.jit.to_static(u_net, build_strategy=build_strategy)
+        u_net(self.x)
+        with self.assertRaises(RuntimeError):
+            print(u_net.forward.main_program)
+
+    def test_case_flag(self):
+        """
+        test the flags is working. TODO: add a global flags.
+        """
+        pass
+
+
+if __name__ == "__main__":
+    unittest.main()

python/paddle/jit/dy2static/program_translator.py

@@ -16,6 +16,7 @@ import collections
 import inspect
 import textwrap
 import threading
+import warnings
 import weakref
 
 from paddle.fluid import _non_static_mode, core, framework
@@ -1077,11 +1078,56 @@ class ParametersRecorder:
         return id(program)
 
 
+class FallbackProgramLayer(object):
+    __slots__ = [
+        '_instance',
+        '_dy_func',
+        'training',
+        '_cuda_graph_capture_mode',
+        '_cuda_graph_pool_id',
+    ]
+
+    def __init__(self, instance, dy_func):
+        self._instance = instance
+        self._dy_func = dy_func
+
+    def __call__(self, inputs):
+        return self._dy_func(*inputs)
+
+    def __getattr__(self, key):
+        if key not in self.__slots__:
+            raise RuntimeError(
+                "There raises a exception after applying `@paddle.jit.to_static()` and already switch into fallback mode. \n"
+                "You can't get attribute for a fallback program layer. Please check `to_static.error` file for detail."
+            )
+        elif key in ['training']:
+            if self._instance is not None:
+                return getattr(self._instance, key)
+            return
+
+        return super().__getattr__(key)
+
+    def __setattr__(self, key, value):
+        if key not in self.__slots__:
+            raise RuntimeError(
+                "There raises a exception after applying `@paddle.jit.to_static()` and already switch into fallback mode. \n"
+                "You can't get attribute for a fallback program layer. Please check `to_static.error` file for detail."
+            )
+        elif key in ['training']:
+            if self._instance is not None:
+                return setattr(self._instance, key, value)
+            return
+
+        return super().__setattr__(key, value)
+
+
 class ProgramCache:
     """
     Wrapper class for the program functions defined by dygraph function.
     """
 
+    dy2static_error_file = "to_static.error"
+
     def __init__(self):
         # {hash_id : (concrete_program, partial_layer)}
         self._caches = collections.OrderedDict()
@@ -1092,17 +1138,37 @@ class ProgramCache:
     def _build_once(self, cache_key):
         # TODO(Aurelius84): Need a gloabl FLAGS to enable/disable to_prim
         enable_prim = cache_key.kwargs['build_strategy'].build_cinn_pass
+        # NOTE(xiongkun): Need a global FLAGS to enable/disable fallback
+        enable_fallback = enable_prim
         if enable_prim:
             # TODO(Jiabin): Change this to True if we need this to be default option
             core.check_and_set_prim_all_enabled()
+        try:
+            concrete_program = ConcreteProgram.from_func_spec(
+                func_spec=cache_key.function_spec,
+                input_spec=cache_key.input_args_with_spec,
+                input_kwargs_spec=cache_key.input_kwargs_with_spec,
+                class_instance=cache_key.class_instance,
+                **cache_key.kwargs
+            )
+        except Exception as e:
+            if enable_fallback:
+                warnings.warn(
+                    "Exception is thrown while applying @paddle.jit.to_static. It will fallback into dygraph mode for training.\n"
+                    "1. You can check `to_static.error` file in current workspace directory for detail.\n"
+                    "2. In fallback mode, you can only do training, can't call paddle.jit.save(). Please modify model code according `to_static.error` firstly"
+                )
+                # TODO(xiongkun) change different file name to avoid overwrite.
+                with open(self.dy2static_error_file, "w") as fp:
+                    fp.write(str(e))
 
-        concrete_program = ConcreteProgram.from_func_spec(
-            func_spec=cache_key.function_spec,
-            input_spec=cache_key.input_args_with_spec,
-            input_kwargs_spec=cache_key.input_kwargs_with_spec,
-            class_instance=cache_key.class_instance,
-            **cache_key.kwargs
-        )
+                fallback_layer = FallbackProgramLayer(
+                    cache_key.class_instance,
+                    cache_key.function_spec.dygraph_function,
+                )
+                return fallback_layer, fallback_layer
+            else:
+                raise
 
         concrete_program._to_prim()
         return concrete_program, partial_program_from(concrete_program)