[executorch][kernel reg] Allow kernel manual registration (#110086)

Summary:
Exposing a codegen mode for generating a hook for user to register their kernels.

If we pass `--manual-registration` flag to `gen_executorch.py`, we will generate the following files:
1. RegisterKernels.h which declares a `register_all_kernels()` API inside `torch::executor` namespace.
2. RegisterKernelsEverything.cpp which implements `register_all_kernels()` by defining an array of generated kernels.

This way user can depend on the library declared by `executorch_generated_lib` macro (with `manual_registration=True`) and be able to include `RegisterKernels.h`. Then they can manually call `register_all_kernels()` instead of relying on C++ static initialization mechanism which is not available in some embedded systems.

Test Plan:
Rely on the unit test:

```
buck2 test fbcode//executorch/runtime/kernel/test:test_kernel_manual_registration
```

Reviewed By: cccclai

Differential Revision: D49439673

Pull Request resolved: https://github.com/pytorch/pytorch/pull/110086
Approved by: https://github.com/cccclai

test/edge/templates/RegisterKernels.h

@@ -0,0 +1,22 @@
+/*
+ * 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.
+ */
+
+// ${generated_comment}
+// Exposing an API for registering all kernels at once.
+#include <executorch/runtime/core/evalue.h>
+#include <executorch/runtime/core/exec_aten/exec_aten.h>
+#include <executorch/runtime/kernel/operator_registry.h>
+#include <executorch/runtime/platform/profiler.h>
+
+namespace torch {
+namespace executor {
+
+Error register_all_kernels();
+
+} // namespace executor
+} // namespace torch

torchgen/gen_executorch.py

@@ -254,6 +254,7 @@ def gen_unboxing(
     selector: SelectiveBuilder,
     use_aten_lib: bool,
     kernel_index: ETKernelIndex,
+    manual_registration: bool,
 ) -> None:
     # Iterable type for write_sharded is a Tuple of (native_function, (kernel_key, metadata))
     def key_func(
@@ -268,9 +269,13 @@ def gen_unboxing(
     ]
 
     header = ["Functions.h" if use_aten_lib else "NativeFunctions.h"]
-
+    filename = (
+        "RegisterKernels.cpp"
+        if manual_registration
+        else "RegisterCodegenUnboxedKernels.cpp"
+    )
     cpu_fm.write_sharded(
-        "RegisterCodegenUnboxedKernels.cpp",
+        filename,
         items,
         key_fn=key_func,
         env_callable=lambda unbox_kernel_entry: {
@@ -449,6 +454,12 @@ def gen_headers(
             ),
         },
     )
+    cpu_fm.write(
+        "RegisterKernels.h",
+        lambda: {
+            "generated_comment": "@" + "generated by torchgen/gen_executorch.py",
+        },
+    )
     headers = {
         "headers": [
             "#include <executorch/runtime/core/exec_aten/exec_aten.h> // at::Tensor etc.",
@@ -864,6 +875,13 @@ def main() -> None:
         help="a boolean flag to indicate whether we use ATen kernels or not, in the future this flag will be per "
         "operator",
     )
+    parser.add_argument(
+        "--manual_registration",
+        "--manual-registration",
+        action="store_true",
+        help="a boolean flag to indicate whether we want to maually call"
+        "register_kernels() or rely on static init. ",
+    )
     parser.add_argument(
         "--generate",
         type=str,
@@ -917,6 +935,7 @@ def main() -> None:
             selector=selector,
             use_aten_lib=options.use_aten_lib,
             kernel_index=kernel_index,
+            manual_registration=options.manual_registration,
         )
         if custom_ops_native_functions:
             gen_custom_ops(