pytorch/torchgen/gen_aoti_c_shim.py

import textwrap
from dataclasses import dataclass
from typing import Dict, List, Optional, Sequence, Tuple, Union

from torchgen.api.types import DispatcherSignature
from torchgen.api.types.signatures import CppSignature, CppSignatureGroup

from torchgen.context import method_with_native_function
from torchgen.model import (
    Argument,
    BackendIndex,
    BaseTy,
    BaseType,
    DispatchKey,
    FunctionSchema,
    ListType,
    NativeFunction,
    OptionalType,
    Type,
)
from torchgen.utils import mapMaybe


def returns_are_all_tensor(schema: FunctionSchema) -> bool:
    return len(schema.returns) != 0 and all(
        ret.type.is_tensor_like() for ret in schema.returns
    )


base_type_to_c_type = {
    BaseTy.Tensor: "AtenTensorHandle",
    BaseTy.bool: "int32_t",  # Use int to pass bool
    BaseTy.int: "int64_t",
    BaseTy.SymInt: "int64_t",  # Inductor-generated code won't see a SymInt
    BaseTy.Scalar: "double",  # Use double to pass both integer and floating point
    BaseTy.float: "double",  # TODO: how about other floating point types?
    BaseTy.str: "const char*",
    BaseTy.DeviceIndex: "int32_t",
    BaseTy.Layout: "int32_t",  # Represent enum as int
    BaseTy.MemoryFormat: "int32_t",  # Represent enum as int
    BaseTy.ScalarType: "int32_t",  # Represent enum as int
}

base_type_to_aten_type = {
    BaseTy.Tensor: "at::Tensor",
    BaseTy.bool: "bool",
    BaseTy.int: "int64_t",
    BaseTy.SymInt: "c10::SymInt",
    BaseTy.Scalar: "c10::Scalar",
    BaseTy.float: "double",
    BaseTy.str: "c10::string_view",
    BaseTy.DeviceIndex: "c10::DeviceIndex",
    BaseTy.Layout: "c10::Layout",
    BaseTy.MemoryFormat: "c10::MemoryFormat",
    BaseTy.ScalarType: "c10::ScalarType",
}

base_type_to_callsite_expr = {
    BaseTy.Tensor: "*tensor_handle_to_tensor_pointer",
    BaseTy.bool: "",
    BaseTy.int: "",
    BaseTy.SymInt: "",
    BaseTy.Scalar: "",
    BaseTy.float: "",
    BaseTy.str: "",
    BaseTy.DeviceIndex: "static_cast<c10::DeviceIndex>",
    BaseTy.Layout: "static_cast<c10::Layout>",
    BaseTy.MemoryFormat: "static_cast<c10::MemoryFormat>",
    BaseTy.ScalarType: "static_cast<c10::ScalarType>",
}


# convert args to C types, names in declarations, and expressions in function bodies
def convert_arg_type_and_name(typ: Type, name: str) -> Tuple[List[str], List[str], List[str], List[str]]:  # type: ignore[return]
    if isinstance(typ, BaseType):
        if typ.name in base_type_to_c_type:
            return (
                [base_type_to_c_type[typ.name]],
                [name],
                [base_type_to_aten_type[typ.name]],
                [
                    f"{base_type_to_callsite_expr[typ.name]}({name})"
                    if base_type_to_callsite_expr[typ.name]
                    else name
                ],
            )
        elif typ.name == BaseTy.Device:
            return (
                ["int32_t", "int32_t"],
                [name, name + "_index_"],
                ["c10::Device"],
                [
                    f"c10::Device(static_cast<c10::DeviceType>({name}), static_cast<c10::DeviceIndex>({name}_index_))"
                ],
            )
        else:
            # TODO: BaseTy.Dimname, BaseTy.Generator, etc.
            raise NotImplementedError(f"TODO: add support for arg type {repr(typ)}")
    elif isinstance(typ, OptionalType):
        c_types, names, aten_types, callsite_exprs = convert_arg_type_and_name(
            typ.elem, name
        )
        j = 0  # index for names
        new_aten_types = []
        new_callsite_exprs = []
        for i, aten_type in enumerate(aten_types):
            # Use pointer to denote optional type
            c_types[j] = c_types[j] + "*"
            if aten_type.startswith("c10::ArrayRef<"):
                # ArrayRef is passed as pointer + size, but no need to add "*" to the size argument
                new_aten_types.append(f"c10::optional<{aten_type}>")
                base_type = aten_type[len("c10::ArrayRef<") : -1]
                new_callsite_exprs.append(
                    f"pointer_to_optional_list<{base_type}>({names[j]}, {names[j+1]})"
                )
                j += 2
            elif aten_type == "c10::Device":
                # Device is passed as device_type + device_index
                new_aten_types.append("c10::optional<c10::Device>")
                new_callsite_exprs.append(
                    f"pointer_to_optional_device({names[j]}, {names[j+1]})"
                )
                j += 2
            else:
                new_aten_types.append(f"c10::optional<{aten_type}>")
                new_callsite_exprs.append(
                    f"pointer_to_optional<{aten_type}>({names[j]})"
                )
                j += 1

        return (
            c_types,
            names,
            new_aten_types,
            new_callsite_exprs,
        )
    elif isinstance(typ, ListType):
        # Need to explictly pass the list as pointer + length
        c_types, names, aten_types, _ = convert_arg_type_and_name(typ.elem, name)
        assert len(c_types) == 1, "ListType with unsupported element type " + repr(typ)

        # The list content should never be modified
        c_types[0] = f"const {c_types[0]}*"
        c_types.append("int64_t")
        name = names[0]
        names.append(name + "_len_")

        atype = aten_types[0]
        callsite_exprs = []
        if atype == "bool":
            # no converter from std::vector<bool> to c10::ArrayRef<bool>
            # construct std::array<bool, N> instead
            assert typ.size is not None
            callsite_exprs.append(f"pointer_to_list<{typ.size}>({name})")
        elif atype == "c10::optional<at::Tensor>":
            # convert from std::vector<c10::optional<at::Tensor>> to c10::List<c10::optional<at::Tensor>>
            callsite_exprs.append(
                f"c10::List<{atype}>(c10::ArrayRef<{atype}>(pointer_to_list<{atype}>({name}, {name}_len_)))"
            )
        else:
            callsite_exprs.append(f"pointer_to_list<{atype}>({name}, {name}_len_)")

        aten_types = [f"c10::ArrayRef<{t}>" for t in aten_types]
        return (
            c_types,
            names,
            aten_types,
            callsite_exprs,
        )


def zip_type_and_name(types: List[str], names: List[str]) -> List[str]:
    return [typ + " " + name for typ, name in zip(types, names)]


# Generate argument declarations and callsite expressions
def gen_arguments(flat_arguments: Sequence[Argument]) -> Tuple[List[str], List[str]]:
    types = []
    new_names = []
    callsite_exprs = []
    for arg in flat_arguments:
        new_types, names, _, new_callsite_exprs = convert_arg_type_and_name(
            arg.type, arg.name
        )
        types.extend(new_types)
        new_names.extend(names)
        callsite_exprs.extend(new_callsite_exprs)
    return zip_type_and_name(types, new_names), callsite_exprs


# Return values are passed out as pointer arguments because all the C shim functions
# are expected to return AOTITorchError.
# Generate returns as declarations and callsite expressions
def gen_returns(schema: FunctionSchema) -> Tuple[List[str], List[str]]:
    types = []
    names = []
    for idx, ret in enumerate(schema.returns):
        names.append(f"ret{idx}")
        if isinstance(ret.type, BaseType) and ret.type.name in base_type_to_c_type:
            types.append(base_type_to_c_type[ret.type.name] + "*")
        else:
            raise NotImplementedError(
                f"TODO: add support for return type {repr(ret.type)}"
            )

    def convert_return(typ: BaseType, val: str) -> str:
        if typ.name == BaseTy.Tensor:
            return f"new_tensor_handle(std::move({val}));"
        elif typ.name == BaseTy.SymInt:
            return f"{val}.expect_int()"
        elif typ.name == BaseTy.Scalar:
            return f"{val}.toDouble()"
        else:
            return val

    ret_pointer_can_be_null = False
    unambiguous_name = schema.name.unambiguous_name()
    for name in ["_scaled_dot_product_flash_attention"]:
        if name in unambiguous_name:
            ret_pointer_can_be_null = True
            break

    callsite_exprs: List[str] = []
    for idx, ret in enumerate(schema.returns):
        tmp = "tmp_result" if len(names) == 1 else f"std::get<{idx}>(tmp_result)"
        assert isinstance(ret.type, BaseType)
        rval = convert_return(ret.type, tmp)
        if ret_pointer_can_be_null:
            callsite_exprs.append(f"if ({names[idx]}) {{ *{names[idx]} = {rval}; }}")
        else:
            callsite_exprs.append(f"*{names[idx]} = {rval};")

    return zip_type_and_name(types, names), callsite_exprs


# gen.py generates header first and then src, so caching the result here to avoid duplicate work
declaration_definition_cache: Dict[Tuple[str, str, str], Tuple[str, str]] = {}


def gen_declaration_and_definition(
    schema: FunctionSchema, device: str, backend_call: str
) -> Tuple[str, str]:
    func_name = schema.name.unambiguous_name()

    global declaration_definition_cache
    if (func_name, device, backend_call) in declaration_definition_cache:
        return declaration_definition_cache[(func_name, device, backend_call)]

    if schema.is_out_fn():
        # out_variant has out arguments in the front, and it's ok to ignore return value
        # because C shim functions only return AOTITorchError
        # Somehow at::native out-variant functions have out arguments in the back
        args, callsite_exprs = gen_arguments(
            [*schema.arguments.flat_non_out, *schema.arguments.out]
            if "at::native" in backend_call
            else [*schema.arguments.out, *schema.arguments.flat_non_out],
        )
        ret_assignments: List[str] = []
    else:
        args, callsite_exprs = gen_arguments(schema.arguments.flat_all)
        ret_declarations, ret_assignments = gen_returns(schema)
        args.extend(ret_declarations)

    declaration = f"AOTITorchError aoti_torch_{device}_{func_name}({', '.join(args)})"

    tmp_result = "auto tmp_result = " if ret_assignments else ""
    ret_assignments_str = "\n" + "\n".join(ret_assignments) if ret_assignments else ""
    definition = f"""
{declaration} {{
    AOTI_TORCH_CONVERT_EXCEPTION_TO_ERROR_CODE({{
        {tmp_result}{backend_call}(
{textwrap.indent(', '.join(callsite_exprs), "            ")}
        );{textwrap.indent(ret_assignments_str, "        ")}
    }});
}}
"""
    declaration_definition_cache[(func_name, device, backend_call)] = (
        declaration,
        definition,
    )
    return declaration, definition


def gen_static_dispatch_backend_call_signature(
    sig: Union[CppSignature, DispatcherSignature],
    f: NativeFunction,
) -> CppSignature:
    sig = DispatcherSignature.from_schema(f.func)
    cpp_sigs = CppSignatureGroup.from_native_function(
        f, method=False, fallback_binding=False
    )
    if sig.symint and f.func.has_symint():
        cpp_sig = cpp_sigs.symint_signature
    else:
        cpp_sig = cpp_sigs.signature
    assert cpp_sig is not None
    return cpp_sig


def gen_static_dispatch_backend_call(
    f: NativeFunction,
    backend_index: BackendIndex,
) -> str:
    assert backend_index.has_kernel(f)
    sig = DispatcherSignature.from_schema(f.func)
    cpp_sig = gen_static_dispatch_backend_call_signature(sig, f)
    return f"at::{backend_index.dispatch_key.lower()}::{cpp_sig.name()}"


def get_backend_index_for_aoti(
    f: NativeFunction,
    dispatch_key: DispatchKey,
    backend_indices: Dict[DispatchKey, BackendIndex],
) -> Optional[BackendIndex]:
    if "pointwise" in f.tags:
        # TODO: No need to generate C shim for Inductor lowered ops.
        # Only skip pointwise kernels for now, and we can add more tags later.
        return None

    backend_index = None
    if backend_indices[dispatch_key].has_kernel(f):
        backend_index = backend_indices[dispatch_key]
    elif backend_indices[DispatchKey.CompositeExplicitAutograd].has_kernel(f):
        # We need to create C shim wrappers for CompositeExplicitAutograd kernels
        backend_index = backend_indices[DispatchKey.CompositeExplicitAutograd]
    elif backend_indices[DispatchKey.CompositeExplicitAutogradNonFunctional].has_kernel(
        f
    ):
        # We need to create C shim wrappers for CompositeExplicitAutogradNonFunctional kernels
        backend_index = backend_indices[
            DispatchKey.CompositeExplicitAutogradNonFunctional
        ]
    return backend_index


def gen_c_shim(
    f: NativeFunction,
    dispatch_key: DispatchKey,
    backend_indices: Dict[DispatchKey, BackendIndex],
    header: bool,
) -> Optional[str]:
    backend_index = get_backend_index_for_aoti(f, dispatch_key, backend_indices)
    if backend_index is None:
        return None

    schema = f.func
    device = dispatch_key.lower()
    backend_call = gen_static_dispatch_backend_call(
        f,
        backend_index,
    )

    try:
        if header:
            declaration, _ = gen_declaration_and_definition(
                schema, device, backend_call
            )
            return f"AOTI_TORCH_EXPORT {declaration};"
        else:
            _, definition = gen_declaration_and_definition(schema, device, backend_call)
            return definition

    except NotImplementedError:
        return None


@dataclass(frozen=True)
class ShimGenerator:
    dispatch_key: DispatchKey
    backend_indices: Dict[DispatchKey, BackendIndex]
    header: bool  # True to generate .h and False to generate .cpp

    @method_with_native_function
    def __call__(self, f: NativeFunction) -> Optional[str]:
        result = gen_c_shim(f, self.dispatch_key, self.backend_indices, self.header)
        return result


def gen_aoti_c_shim(
    native_functions: Sequence[NativeFunction],
    dispatch_key: DispatchKey,
    backend_indices: Dict[DispatchKey, BackendIndex],
    header: bool,
    includes: str = "",
) -> str:
    body = "\n".join(
        list(
            mapMaybe(
                ShimGenerator(dispatch_key, backend_indices, header),
                native_functions,
            )
        )
    )

    if header:
        return f"""
#pragma once

#include <torch/csrc/inductor/aoti_torch/c/shim.h>

#ifdef __cplusplus
extern "C" {{
#endif

{body}

#ifdef __cplusplus
}} // extern "C"
#endif

"""
    else:
        device = dispatch_key.lower()
        return f"""
#include <torch/csrc/inductor/aoti_torch/tensor_converter.h>
#include <torch/csrc/inductor/aoti_torch/utils.h>
#include <torch/csrc/inductor/aoti_torch/generated/c_shim_{device}.h>

#ifndef AT_PER_OPERATOR_HEADERS
#include <ATen/{str(dispatch_key)}Functions.h>
#include <ATen/CompositeExplicitAutogradFunctions.h>
#include <ATen/CompositeExplicitAutogradNonFunctionalFunctions.h>
#else
{includes}
#endif

using namespace torch::aot_inductor;

{body}

"""