[CGData] Global Merge Functions

Author: kyulee-com

llvm/include/llvm/CGData/CodeGenData.h

@@ -145,6 +145,9 @@ class CodeGenData {
   const OutlinedHashTree *getOutlinedHashTree() {
     return PublishedHashTree.get();
   }
+  const StableFunctionMap *getStableFunctionMap() {
+    return PublishedStableFunctionMap.get();
+  }
 
   /// Returns true if we should write codegen data.
   bool emitCGData() { return EmitCGData; }
@@ -169,10 +172,18 @@ inline bool hasOutlinedHashTree() {
   return CodeGenData::getInstance().hasOutlinedHashTree();
 }
 
+inline bool hasStableFunctionMap() {
+  return CodeGenData::getInstance().hasStableFunctionMap();
+}
+
 inline const OutlinedHashTree *getOutlinedHashTree() {
   return CodeGenData::getInstance().getOutlinedHashTree();
 }
 
+inline const StableFunctionMap *getStableFunctionMap() {
+  return CodeGenData::getInstance().getStableFunctionMap();
+}
+
 inline bool emitCGData() { return CodeGenData::getInstance().emitCGData(); }
 
 inline void

llvm/include/llvm/InitializePasses.h

@@ -123,6 +123,7 @@ void initializeGCEmptyBasicBlocksPass(PassRegistry &);
 void initializeGCMachineCodeAnalysisPass(PassRegistry &);
 void initializeGCModuleInfoPass(PassRegistry &);
 void initializeGVNLegacyPassPass(PassRegistry &);
+void initializeGlobalMergeFuncPass(PassRegistry &);
 void initializeGlobalMergePass(PassRegistry &);
 void initializeGlobalsAAWrapperPassPass(PassRegistry &);
 void initializeHardwareLoopsLegacyPass(PassRegistry &);

llvm/include/llvm/LinkAllPasses.h

@@ -79,6 +79,7 @@ struct ForcePassLinking {
     (void)llvm::createDomOnlyViewerWrapperPassPass();
     (void)llvm::createDomViewerWrapperPassPass();
     (void)llvm::createAlwaysInlinerLegacyPass();
+    (void)llvm::createGlobalMergeFuncPass();
     (void)llvm::createGlobalsAAWrapperPass();
     (void)llvm::createInstSimplifyLegacyPass();
     (void)llvm::createInstructionCombiningPass();

llvm/include/llvm/Passes/CodeGenPassBuilder.h

@@ -74,6 +74,7 @@
 #include "llvm/Target/CGPassBuilderOption.h"
 #include "llvm/Target/TargetMachine.h"
 #include "llvm/Transforms/CFGuard.h"
+#include "llvm/Transforms/IPO/GlobalMergeFunctions.h"
 #include "llvm/Transforms/Scalar/ConstantHoisting.h"
 #include "llvm/Transforms/Scalar/LoopPassManager.h"
 #include "llvm/Transforms/Scalar/LoopStrengthReduce.h"

llvm/include/llvm/Transforms/IPO.h

@@ -55,6 +55,8 @@ enum class PassSummaryAction {
   Export, ///< Export information to summary.
 };
 
+Pass *createGlobalMergeFuncPass();
+
 } // End llvm namespace
 
 #endif

llvm/include/llvm/Transforms/IPO/GlobalMergeFunctions.h

@@ -0,0 +1,73 @@
+//===------ GlobalMergeFunctions.h - Global merge functions -----*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// This file defines global merge functions pass and related data structure.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef PIKA_TRANSFORMS_UTILS_GLOBALMERGEFUNCTIONS_H
+#define PIKA_TRANSFORMS_UTILS_GLOBALMERGEFUNCTIONS_H
+
+#include "llvm/ADT/DenseMap.h"
+#include "llvm/ADT/StableHashing.h"
+#include "llvm/ADT/StringRef.h"
+#include "llvm/CGData/StableFunctionMap.h"
+#include "llvm/IR/Instructions.h"
+#include "llvm/IR/Module.h"
+#include "llvm/Pass.h"
+#include <map>
+#include <mutex>
+
+enum class HashFunctionMode {
+  Local,
+  BuildingHashFuncion,
+  UsingHashFunction,
+};
+
+namespace llvm {
+
+// A vector of locations (the pair of (instruction, operand) indices) reachable
+// from a parameter.
+using ParamLocs = SmallVector<IndexPair, 4>;
+// A vector of parameters
+using ParamLocsVecTy = SmallVector<ParamLocs, 8>;
+// A map of stable hash to a vector of stable functions
+
+/// GlobalMergeFunc finds functions which only differ by constants in
+/// certain instructions, e.g. resulting from specialized functions of layout
+/// compatible types.
+/// Unlike PikaMergeFunc that directly compares IRs, this uses stable function
+/// hash to find the merge candidate. Similar to the global outliner, we can run
+/// codegen twice to collect function merge candidate in the first round, and
+/// merge functions globally in the second round.
+class GlobalMergeFunc : public ModulePass {
+  HashFunctionMode MergerMode = HashFunctionMode::Local;
+
+  std::unique_ptr<StableFunctionMap> LocalFunctionMap;
+
+public:
+  static char ID;
+
+  GlobalMergeFunc();
+
+  void initializeMergerMode(const Module &M);
+
+  bool runOnModule(Module &M) override;
+
+  /// Analyze module to create stable function into LocalFunctionMap.
+  void analyze(Module &M);
+
+  /// Emit LocalFunctionMap into __llvm_merge section.
+  void emitFunctionMap(Module &M);
+
+  /// Merge functions in the module using the global function map.
+  bool merge(Module &M, const StableFunctionMap *FunctionMap);
+};
+
+} // end namespace llvm
+#endif // PIKA_TRANSFORMS_UTILS_GLOBALMERGEFUNCTIONS_H

llvm/lib/CodeGen/TargetPassConfig.cpp

@@ -141,6 +141,9 @@ static cl::opt<RunOutliner> EnableMachineOutliner(
                           "Disable all outlining"),
                // Sentinel value for unspecified option.
                clEnumValN(RunOutliner::AlwaysOutline, "", "")));
+cl::opt<bool> EnableGlobalMergeFunc(
+    "enable-global-merge-func", cl::Hidden,
+    cl::desc("Enable global merge functions that are based on hash function"));
 // Disable the pass to fix unwind information. Whether the pass is included in
 // the pipeline is controlled via the target options, this option serves as
 // manual override.

llvm/lib/LTO/LTO.cpp

@@ -53,6 +53,7 @@
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetOptions.h"
 #include "llvm/Transforms/IPO.h"
+#include "llvm/Transforms/IPO/GlobalMergeFunctions.h"
 #include "llvm/Transforms/IPO/MemProfContextDisambiguation.h"
 #include "llvm/Transforms/IPO/WholeProgramDevirt.h"
 #include "llvm/Transforms/Utils/FunctionImportUtils.h"

llvm/lib/Transforms/IPO/CMakeLists.txt

@@ -21,6 +21,7 @@ add_llvm_component_library(LLVMipo
   GlobalDCE.cpp
   GlobalOpt.cpp
   GlobalSplit.cpp
+  GlobalMergeFunctions.cpp
   HotColdSplitting.cpp
   IPO.cpp
   IROutliner.cpp
@@ -60,6 +61,7 @@ add_llvm_component_library(LLVMipo
   Analysis
   BitReader
   BitWriter
+  CGData
   Core
   FrontendOpenMP
   InstCombine