[llvm] [refactor] (Decomp of #5251 2/n) Make modulegen a virtual function and let LLVMCompiledData replace ModuleGenValue

taichi/codegen/codegen.h

@@ -1,9 +1,14 @@
 // Driver class for kernel code generators.
 
 #pragma once
-
+#include "taichi/ir/ir.h"
 #include "taichi/program/program.h"
-
+#ifdef TI_WITH_LLVM
+#include "llvm/IR/Module.h"
+#include "taichi/codegen/llvm/codegen_llvm.h"
+#include "taichi/runtime/llvm/launch_arg_info.h"
+#include "taichi/codegen/llvm/llvm_codegen_utils.h"
+#endif
 TLANG_NAMESPACE_BEGIN
 
 class KernelCodeGen {
@@ -22,6 +27,13 @@ class KernelCodeGen {
                                                Stmt *stmt = nullptr);
 
   virtual FunctionType codegen() = 0;
+#ifdef TI_WITH_LLVM
+  virtual LLVMCompiledData modulegen(
+      std::unique_ptr<llvm::Module> &&module = nullptr,
+      OffloadedStmt *stmt = nullptr) {
+    TI_NOT_IMPLEMENTED
+  }
+#endif
 };
 
 TLANG_NAMESPACE_END

taichi/codegen/cuda/codegen_cuda.cpp

@@ -42,8 +42,8 @@ class CodeGenLLVMCUDA : public CodeGenLLVM {
     CUDAModuleToFunctionConverter converter{tlctx,
                                             llvm_prog->get_runtime_executor()};
 
-    return converter.convert(this->kernel, std::move(compiled_res.llvm_module),
-                             std::move(compiled_res.offloaded_tasks));
+    return converter.convert(this->kernel, std::move(compiled_res.module),
+                             std::move(compiled_res.tasks));
   }
 
   llvm::Value *create_print(std::string tag,

taichi/codegen/llvm/codegen_llvm.cpp

@@ -2327,14 +2327,14 @@ void CodeGenLLVM::emit_to_module() {
   ir->accept(this);
 }
 
-CodeGenLLVM::CompiledData CodeGenLLVM::run_compilation() {
+LLVMCompiledData CodeGenLLVM::run_compilation() {
   const auto &config = prog->config;
   std::string kernel_key =
       get_hashed_offline_cache_key(&kernel->program->config, kernel);
   kernel->set_kernel_key_for_cache(kernel_key);
   if (config.offline_cache && !config.async_mode &&
       this->supports_offline_cache() && !kernel->is_evaluator) {
-    CompiledData res;
+    LLVMCompiledData res;
     const bool ok = maybe_read_compilation_from_cache(kernel_key, &res);
     if (ok) {
       return res;
@@ -2353,15 +2353,15 @@ CodeGenLLVM::CompiledData CodeGenLLVM::run_compilation() {
     cache_module(kernel_key);
   }
 
-  CompiledData res;
-  res.offloaded_tasks = std::move(this->offloaded_tasks);
-  res.llvm_module = std::move(this->module);
+  LLVMCompiledData res;
+  res.tasks = std::move(this->offloaded_tasks);
+  res.module = std::move(this->module);
   return res;
 }
 
 bool CodeGenLLVM::maybe_read_compilation_from_cache(
     const std::string &kernel_key,
-    CompiledData *data) {
+    LLVMCompiledData *data) {
   const auto &config = prog->config;
   auto reader =
       LlvmOfflineCacheFileReader::make(config.offline_cache_file_path);
@@ -2384,8 +2384,8 @@ bool CodeGenLLVM::maybe_read_compilation_from_cache(
     t.grid_dim = task.grid_dim;
   }
   kernel->set_from_offline_cache();
-  data->offloaded_tasks = std::move(this->offloaded_tasks);
-  data->llvm_module = std::move(this->module);
+  data->tasks = std::move(this->offloaded_tasks);
+  data->module = std::move(this->module);
   return true;
 }
 
@@ -2394,8 +2394,8 @@ FunctionType CodeGenLLVM::gen() {
 
   ModuleToFunctionConverter converter{
       tlctx, get_llvm_program(prog)->get_runtime_executor()};
-  return converter.convert(kernel, std::move(compiled_res.llvm_module),
-                           std::move(compiled_res.offloaded_tasks));
+  return converter.convert(kernel, std::move(compiled_res.module),
+                           std::move(compiled_res.tasks));
 }
 
 llvm::Value *CodeGenLLVM::create_xlogue(std::unique_ptr<Block> &block) {

taichi/codegen/llvm/codegen_llvm.h

@@ -44,6 +44,11 @@ class FunctionCreationGuard {
   ~FunctionCreationGuard();
 };
 
+struct LLVMCompiledData {
+  std::vector<OffloadedTask> tasks;
+  std::unique_ptr<llvm::Module> module{nullptr};
+};
+
 class CodeGenLLVM : public IRVisitor, public LLVMModuleBuilder {
  public:
   Kernel *kernel;
@@ -121,18 +126,14 @@ class CodeGenLLVM : public IRVisitor, public LLVMModuleBuilder {
 
   void eliminate_unused_functions();
 
-  struct CompiledData {
-    std::vector<OffloadedTask> offloaded_tasks;
-    std::unique_ptr<llvm::Module> llvm_module{nullptr};
-  };
   /**
    * @brief Runs the codegen and produces the compiled result.
    *
-   * After this call, `module` and `offloaded_tasks` will be moved.
+   * After this call, `module` and `tasks` will be moved.
    *
-   * @return CompiledData
+   * @return LLVMCompiledData
    */
-  CompiledData run_compilation();
+  LLVMCompiledData run_compilation();
 
   // TODO: This function relies largely on `run_compilation()`. Name it better.
   virtual FunctionType gen();
@@ -406,7 +407,7 @@ class CodeGenLLVM : public IRVisitor, public LLVMModuleBuilder {
 
  private:
   bool maybe_read_compilation_from_cache(const std::string &kernel_key,
-                                         CompiledData *data);
+                                         LLVMCompiledData *data);
 
   void cache_module(const std::string &kernel_key);
 };

taichi/codegen/wasm/codegen_wasm.cpp

@@ -246,28 +246,28 @@ FunctionType CodeGenWASM::codegen() {
   return CodeGenLLVMWASM(kernel, ir).gen();
 }
 
-std::unique_ptr<ModuleGenValue> CodeGenWASM::modulegen(
-    std::unique_ptr<llvm::Module> &&module) {
+LLVMCompiledData CodeGenWASM::modulegen(std::unique_ptr<llvm::Module> &&module,
+                                        OffloadedStmt *stmt) {
   bool init_flag = module == nullptr;
-  std::vector<std::string> name_list;
-
+  std::vector<OffloadedTask> name_list;
   auto gen = std::make_unique<CodeGenLLVMWASM>(kernel, ir, std::move(module));
 
-  name_list.push_back(gen->init_taichi_kernel_function());
+  name_list.emplace_back(nullptr);
+  name_list[0].name = gen->init_taichi_kernel_function();
   gen->emit_to_module();
   gen->finalize_taichi_kernel_function();
 
   // TODO: move the following functions to dump process in AOT.
   if (init_flag) {
     for (auto &name : kPreloadedFuncNames) {
-      name_list.emplace_back(name);
+      name_list.emplace_back(nullptr);
+      name_list.back().name = name;
     }
   }
 
   gen->tlctx->jit->global_optimize_module(gen->module.get());
 
-  return std::make_unique<ModuleGenValue>(std::move(gen->module), name_list);
+  return {name_list, std::move(gen->module)};
 }
-
 }  // namespace lang
 }  // namespace taichi

taichi/codegen/wasm/codegen_wasm.h

@@ -11,18 +11,6 @@
 namespace taichi {
 namespace lang {
 
-#ifdef TI_WITH_LLVM
-class ModuleGenValue {
- public:
-  ModuleGenValue(std::unique_ptr<llvm::Module> module,
-                 const std::vector<std::string> &name_list)
-      : module(std::move(module)), name_list(name_list) {
-  }
-  std::unique_ptr<llvm::Module> module;
-  std::vector<std::string> name_list;
-};
-#endif
-
 class CodeGenWASM : public KernelCodeGen {
  public:
   CodeGenWASM(Kernel *kernel, IRNode *ir = nullptr)
@@ -32,8 +20,9 @@ class CodeGenWASM : public KernelCodeGen {
   FunctionType codegen() override;
 
 #ifdef TI_WITH_LLVM
-  std::unique_ptr<ModuleGenValue> modulegen(
-      std::unique_ptr<llvm::Module> &&module);  // AOT Module Gen
+  LLVMCompiledData modulegen(
+      std::unique_ptr<llvm::Module> &&module = nullptr,
+      OffloadedStmt *stmt = nullptr) override;  // AOT Module Gen
 #endif
 };
 

taichi/runtime/cpu/aot_module_builder_impl.cpp

@@ -9,7 +9,7 @@ namespace taichi {
 namespace lang {
 namespace cpu {
 
-CodeGenLLVM::CompiledData AotModuleBuilderImpl::compile_kernel(Kernel *kernel) {
+LLVMCompiledData AotModuleBuilderImpl::compile_kernel(Kernel *kernel) {
   auto cgen = CodeGenCPU::make_codegen_llvm(kernel, /*ir=*/nullptr);
   return cgen->run_compilation();
 }

taichi/runtime/cpu/aot_module_builder_impl.h

@@ -15,7 +15,7 @@ class AotModuleBuilderImpl : public LlvmAotModuleBuilder {
   }
 
  private:
-  CodeGenLLVM::CompiledData compile_kernel(Kernel *kernel) override;
+  LLVMCompiledData compile_kernel(Kernel *kernel) override;
 };
 
 }  // namespace cpu

taichi/runtime/cuda/aot_module_builder_impl.cpp

@@ -9,7 +9,7 @@ namespace taichi {
 namespace lang {
 namespace cuda {
 
-CodeGenLLVM::CompiledData AotModuleBuilderImpl::compile_kernel(Kernel *kernel) {
+LLVMCompiledData AotModuleBuilderImpl::compile_kernel(Kernel *kernel) {
   auto cgen = CodeGenCUDA::make_codegen_llvm(kernel, /*ir=*/nullptr);
   return cgen->run_compilation();
 }

taichi/runtime/cuda/aot_module_builder_impl.h

@@ -15,7 +15,7 @@ class AotModuleBuilderImpl : public LlvmAotModuleBuilder {
   }
 
  private:
-  CodeGenLLVM::CompiledData compile_kernel(Kernel *kernel) override;
+  LLVMCompiledData compile_kernel(Kernel *kernel) override;
 };
 
 }  // namespace cuda

taichi/runtime/llvm/llvm_aot_module_builder.cpp

@@ -22,9 +22,9 @@ void LlvmAotModuleBuilder::add_per_backend(const std::string &identifier,
   auto compiled = compile_kernel(kernel);
   LlvmOfflineCache::KernelCacheData kcache;
   kcache.kernel_key = identifier;
-  kcache.module = compiled.llvm_module.get();
-  kcache.owned_module = std::move(compiled.llvm_module);
-  const auto &tasks = compiled.offloaded_tasks;
+  kcache.module = compiled.module.get();
+  kcache.owned_module = std::move(compiled.module);
+  const auto &tasks = compiled.tasks;
   kcache.args = infer_launch_args(kernel);
   kcache.offloaded_task_list.resize(tasks.size());
   std::transform(tasks.begin(), tasks.end(), kcache.offloaded_task_list.begin(),

taichi/runtime/llvm/llvm_aot_module_builder.h

@@ -17,7 +17,7 @@ class LlvmAotModuleBuilder : public AotModuleBuilder {
 
  protected:
   void add_per_backend(const std::string &identifier, Kernel *kernel) override;
-  virtual CodeGenLLVM::CompiledData compile_kernel(Kernel *kernel) = 0;
+  virtual LLVMCompiledData compile_kernel(Kernel *kernel) = 0;
 
   void add_field_per_backend(const std::string &identifier,
                              const SNode *rep_snode,

taichi/runtime/wasm/aot_module_builder_impl.cpp

@@ -36,10 +36,10 @@ void AotModuleBuilderImpl::dump(const std::string &output_dir,
 void AotModuleBuilderImpl::add_per_backend(const std::string &identifier,
                                            Kernel *kernel) {
   auto module_info = CodeGenWASM(kernel, nullptr).modulegen(std::move(module_));
-  module_ = std::move(module_info->module);
+  module_ = std::move(module_info.module);
 
-  for (auto &name : module_info->name_list)
-    name_list_.push_back(name);
+  for (auto &task : module_info.tasks)
+    name_list_.push_back(task.name);
 }
 
 void AotModuleBuilderImpl::add_field_per_backend(const std::string &identifier,