[REFACTOR] TVM_REGISTER_API -> TVM_REGISTER_GLOBAL (#4621)

TVM_REGSISTER_API is an alias of TVM_REGISTER_GLOBAL.
In the spirit of simplify redirections, this PR removes
the original TVM_REGISTER_API macro and directly use TVM_REGISTER_GLOBAL.

This type of refactor will also simplify the IDE navigation tools
such as FFI navigator to provide better code reading experiences.

Move EnvFunc's definition to node.

docs/dev/codebase_walkthrough.rst

@@ -72,7 +72,7 @@ The Node system is the basis of exposing C++ types to frontend languages, includ
 
 ::
 
-   TVM_REGISTER_API("_ComputeOp")
+   TVM_REGISTER_GLOBAL("_ComputeOp")
    .set_body([](TVMArgs args,  TVMRetValue* ret) {
        *ret = ComputeOpNode::make(args[0],
                                   args[1],
@@ -174,7 +174,7 @@ The ``Build()`` function looks up the code generator for the given target in the
 
 ::
 
-   TVM_REGISTER_API("codegen.build_cuda")
+   TVM_REGISTER_GLOBAL("codegen.build_cuda")
    .set_body([](TVMArgs args, TVMRetValue* rv) {
        *rv = BuildCUDA(args[0]);
      });

docs/dev/relay_add_op.rst

@@ -96,7 +96,7 @@ the arguments to the call node, as below.
 
 .. code:: c
 
-    TVM_REGISTER_API("relay.op._make.add")
+    TVM_REGISTER_GLOBAL("relay.op._make.add")
         .set_body_typed<Expr(Expr, Expr)>([](Expr lhs, Expr rhs) {
             static const Op& op = Op::Get("add");
           return CallNode::make(op, {lhs, rhs}, Attrs(), {});
@@ -106,7 +106,7 @@ Including a Python API Hook
 ---------------------------
 
 It is generally the convention in Relay, that functions exported
-through ``TVM_REGISTER_API`` should be wrapped in a separate
+through ``TVM_REGISTER_GLOBAL`` should be wrapped in a separate
 Python function rather than called directly in Python. In the case
 of the functions that produce calls to operators, it may be convenient
 to bundle them, as in ``python/tvm/relay/op/tensor.py``, where

docs/dev/relay_pass_infra.rst

@@ -131,13 +131,13 @@ Python APIs to create a compilation pipeline using pass context.
       TVM_DLL static PassContext Create();
       TVM_DLL static PassContext Current();
       /* Other fields are omitted. */
-    
+
      private:
       // The entry of a pass context scope.
       TVM_DLL void EnterWithScope();
       // The exit of a pass context scope.
       TVM_DLL void ExitWithScope();
-    
+
       // Classes to get the Python `with` like syntax.
       friend class tvm::With<PassContext>;
     };
@@ -225,7 +225,7 @@ cannot add or delete a function through these passes as they are not aware of
 the global information.
 
 .. code:: c++
-   
+
     class FunctionPassNode : PassNode {
       PassInfo pass_info;
       runtime::TypedPackedFunc<Function(Function, Module, PassContext)> pass_func;
@@ -319,7 +319,7 @@ favorably use Python APIs to create a specific pass object.
     ModulePass CreateModulePass(std::string name,
                                 int opt_level,
                                 PassFunc pass_func);
-    
+
     SequentialPass CreateSequentialPass(std::string name,
                                         int opt_level,
                                         Array<Pass> passes,
@@ -347,14 +347,14 @@ registration.
     auto tensor_type = relay::TensorTypeNode::make({}, tvm::Bool());
     auto x = relay::VarNode::make("x", relay::Type());
     auto f = relay::FunctionNode::make(tvm::Array<relay::Var>{ x }, x, relay::Type(), {});
-    
+
     auto y = relay::VarNode::make("y", tensor_type);
     auto call = relay::CallNode::make(f, tvm::Array<relay::Expr>{ y });
     auto fx = relay::FunctionNode::make(tvm::Array<relay::Var>{ y }, call, relay::Type(), {});
-    
+
     // Create a module for optimization.
     auto mod = relay::ModuleNode::FromExpr(fx);
-    
+
     // Create a sequential pass.
     tvm::Array<relay::transform::Pass> pass_seqs{
        relay::transform::InferType(),
@@ -363,7 +363,7 @@ registration.
        relay::transform::AlterOpLayout()
     };
     relay::transform::Pass seq = relay::transform::Sequential(pass_seqs);
-    
+
     // Create a pass context for the optimization.
     auto ctx = relay::transform::PassContext::Create();
     ctx->opt_level = 2;
@@ -421,7 +421,7 @@ Python when needed.
       return CreateFunctionPass(pass_func, 2, "FoldConstant", {});
     }
 
-    TVM_REGISTER_API("relay._transform.FoldConstant")
+    TVM_REGISTER_GLOBAL("relay._transform.FoldConstant")
     .set_body_typed(FoldConstant);
 
     }  // namespace transform
@@ -457,10 +457,10 @@ a certain scope.
         def __enter__(self):
             _transform.EnterPassContext(self)
             return self
-    
+
         def __exit__(self, ptype, value, trace):
             _transform.ExitPassContext(self)
-    
+
         @staticmethod
         def current():
             """Return the current pass context."""
@@ -580,18 +580,18 @@ using ``Sequential`` associated with other types of passes.
     z1 = relay.add(y, c)
     z2 = relay.add(z, z1)
     func = relay.Function([x], z2)
-  
-    # Customize the optimization pipeline. 
+
+    # Customize the optimization pipeline.
     seq = _transform.Sequential([
         relay.transform.InferType(),
         relay.transform.FoldConstant(),
         relay.transform.EliminateCommonSubexpr(),
         relay.transform.AlterOpLayout()
     ])
-  
+
     # Create a module to perform optimizations.
     mod = relay.Module({"main": func})
-    
+
     # Users can disable any passes that they don't want to execute by providing
     # a list, e.g. disabled_pass=["EliminateCommonSubexpr"].
     with relay.build_config(opt_level=3):
@@ -629,7 +629,7 @@ For more pass infra related examples in Python and C++, please refer to
 
 .. _Block: https://mxnet.incubator.apache.org/api/python/docs/api/gluon/block.html#gluon-block
 
-.. _Relay module: https://docs.tvm.ai/langref/relay_expr.html#module-and-global-functions 
+.. _Relay module: https://docs.tvm.ai/langref/relay_expr.html#module-and-global-functions
 
 .. _include/tvm/relay/transform.h: https://github.com/apache/incubator-tvm/blob/master/include/tvm/relay/transform.h
 

include/tvm/codegen.h

@@ -28,7 +28,6 @@
 #include "base.h"
 #include "expr.h"
 #include "lowered_func.h"
-#include "api_registry.h"
 #include "runtime/packed_func.h"
 
 namespace tvm {

include/tvm/api_registry.h → include/tvm/node/env_func.h

@@ -18,33 +18,19 @@
  */
 
 /*!
- * \file tvm/api_registry.h
- * \brief This file contains utilities related to
- *  the TVM's global function registry.
+ * \file tvm/node/env_func.h
+ * \brief Serializable global function.
  */
-#ifndef TVM_API_REGISTRY_H_
-#define TVM_API_REGISTRY_H_
+#ifndef TVM_NODE_ENV_FUNC_H_
+#define TVM_NODE_ENV_FUNC_H_
+
+#include <tvm/node/reflection.h>
 
 #include <string>
 #include <utility>
-#include "base.h"
-#include "packed_func_ext.h"
-#include "runtime/registry.h"
 
-namespace tvm {
-/*!
- * \brief Register an API function globally.
- * It simply redirects to TVM_REGISTER_GLOBAL
- *
- * \code
- *   TVM_REGISTER_API(MyPrint)
- *   .set_body([](TVMArgs args, TVMRetValue* rv) {
- *     // my code.
- *   });
- * \endcode
- */
-#define TVM_REGISTER_API(OpName) TVM_REGISTER_GLOBAL(OpName)
 
+namespace tvm {
 /*!
  * \brief Node container of EnvFunc
  * \sa EnvFunc
@@ -54,7 +40,7 @@ class EnvFuncNode : public Object {
   /*! \brief Unique name of the global function */
   std::string name;
   /*! \brief The internal packed function */
-  PackedFunc func;
+  runtime::PackedFunc func;
   /*! \brief constructor */
   EnvFuncNode() {}
 
@@ -154,4 +140,4 @@ class TypedEnvFunc<R(Args...)> : public ObjectRef {
 };
 
 }  // namespace tvm
-#endif  // TVM_API_REGISTRY_H_
+#endif  // TVM_NODE_ENV_FUNC_H_

include/tvm/relay/base.h

@@ -24,7 +24,7 @@
 #ifndef TVM_RELAY_BASE_H_
 #define TVM_RELAY_BASE_H_
 
-#include <tvm/api_registry.h>
+
 #include <tvm/ir/span.h>
 #include <tvm/ir.h>
 #include <tvm/node/node.h>

include/tvm/relay/type.h

@@ -24,8 +24,12 @@
 #ifndef TVM_RELAY_TYPE_H_
 #define TVM_RELAY_TYPE_H_
 
-#include <tvm/api_registry.h>
+
 #include <tvm/ir/type.h>
+#include <tvm/runtime/registry.h>
+#include <tvm/packed_func_ext.h>
+#include <tvm/node/env_func.h>
+
 #include <tvm/ir.h>
 #include <string>
 

include/tvm/runtime/registry.h

@@ -70,7 +70,7 @@ class Registry {
    *
    * \code
    *
-   * TVM_REGISTER_API("addone")
+   * TVM_REGISTER_GLOBAL("addone")
    * .set_body_typed<int(int)>([](int x) { return x + 1; });
    *
    * \endcode
@@ -96,7 +96,7 @@ class Registry {
    *   return x * y;
    * }
    *
-   * TVM_REGISTER_API("multiply")
+   * TVM_REGISTER_GLOBAL("multiply")
    * .set_body_typed(multiply); // will have type int(int, int)
    *
    * \endcode
@@ -120,7 +120,7 @@ class Registry {
    * struct Example {
    *    int doThing(int x);
    * }
-   * TVM_REGISTER_API("Example_doThing")
+   * TVM_REGISTER_GLOBAL("Example_doThing")
    * .set_body_method(&Example::doThing); // will have type int(Example, int)
    *
    * \endcode
@@ -148,7 +148,7 @@ class Registry {
    * struct Example {
    *    int doThing(int x);
    * }
-   * TVM_REGISTER_API("Example_doThing")
+   * TVM_REGISTER_GLOBAL("Example_doThing")
    * .set_body_method(&Example::doThing); // will have type int(Example, int)
    *
    * \endcode
@@ -181,7 +181,7 @@ class Registry {
    * // noderef subclass
    * struct Example;
    *
-   * TVM_REGISTER_API("Example_doThing")
+   * TVM_REGISTER_GLOBAL("Example_doThing")
    * .set_body_method<Example>(&ExampleNode::doThing); // will have type int(Example, int)
    *
    * // note that just doing:
@@ -221,7 +221,7 @@ class Registry {
    * // noderef subclass
    * struct Example;
    *
-   * TVM_REGISTER_API("Example_doThing")
+   * TVM_REGISTER_GLOBAL("Example_doThing")
    * .set_body_method<Example>(&ExampleNode::doThing); // will have type int(Example, int)
    *
    * // note that just doing:

src/api/api_arith.cc

@@ -23,29 +23,31 @@
  */
 #include <tvm/expr.h>
 #include <tvm/ir.h>
-#include <tvm/api_registry.h>
+#include <tvm/runtime/registry.h>
+#include <tvm/packed_func_ext.h>
+
 #include <tvm/tensor.h>
 
 namespace tvm {
 namespace arith {
 
-TVM_REGISTER_API("arith.intset_single_point")
+TVM_REGISTER_GLOBAL("arith.intset_single_point")
 .set_body_typed(IntSet::single_point);
 
-TVM_REGISTER_API("arith.intset_vector")
+TVM_REGISTER_GLOBAL("arith.intset_vector")
 .set_body_typed(IntSet::vector);
 
-TVM_REGISTER_API("arith.intset_interval")
+TVM_REGISTER_GLOBAL("arith.intset_interval")
 .set_body_typed(IntSet::interval);
 
 
-TVM_REGISTER_API("arith.DetectLinearEquation")
+TVM_REGISTER_GLOBAL("arith.DetectLinearEquation")
 .set_body_typed(DetectLinearEquation);
 
-TVM_REGISTER_API("arith.DetectClipBound")
+TVM_REGISTER_GLOBAL("arith.DetectClipBound")
 .set_body_typed(DetectClipBound);
 
-TVM_REGISTER_API("arith.DeduceBound")
+TVM_REGISTER_GLOBAL("arith.DeduceBound")
 .set_body_typed<IntSet(Expr, Expr, Map<Var, IntSet>, Map<Var, IntSet>)>([](
   Expr v, Expr cond,
   const Map<Var, IntSet> hint_map,
@@ -55,36 +57,36 @@ TVM_REGISTER_API("arith.DeduceBound")
 });
 
 
-TVM_REGISTER_API("arith.DomainTouched")
+TVM_REGISTER_GLOBAL("arith.DomainTouched")
 .set_body_typed(DomainTouched);
 
-TVM_REGISTER_API("_IntervalSetGetMin")
+TVM_REGISTER_GLOBAL("_IntervalSetGetMin")
 .set_body_method(&IntSet::min);
 
-TVM_REGISTER_API("_IntervalSetGetMax")
+TVM_REGISTER_GLOBAL("_IntervalSetGetMax")
 .set_body_method(&IntSet::max);
 
-TVM_REGISTER_API("_IntSetIsNothing")
+TVM_REGISTER_GLOBAL("_IntSetIsNothing")
 .set_body_method(&IntSet::is_nothing);
 
-TVM_REGISTER_API("_IntSetIsEverything")
+TVM_REGISTER_GLOBAL("_IntSetIsEverything")
 .set_body_method(&IntSet::is_everything);
 
 ConstIntBound MakeConstIntBound(int64_t min_value, int64_t max_value) {
   return ConstIntBound(min_value, max_value);
 }
 
-TVM_REGISTER_API("arith._make_ConstIntBound")
+TVM_REGISTER_GLOBAL("arith._make_ConstIntBound")
 .set_body_typed(MakeConstIntBound);
 
 ModularSet MakeModularSet(int64_t coeff, int64_t base) {
   return ModularSet(coeff, base);
 }
 
-TVM_REGISTER_API("arith._make_ModularSet")
+TVM_REGISTER_GLOBAL("arith._make_ModularSet")
 .set_body_typed(MakeModularSet);
 
-TVM_REGISTER_API("arith._CreateAnalyzer")
+TVM_REGISTER_GLOBAL("arith._CreateAnalyzer")
 .set_body([](TVMArgs args, TVMRetValue* ret) {
     using runtime::PackedFunc;
     using runtime::TypedPackedFunc;