[CodeGen][NPM] Port BranchRelaxation to NPM (#130067)

This completes the PreEmitPasses.

Author: optimisan

llvm/include/llvm/CodeGen/BranchRelaxation.h

@@ -0,0 +1,25 @@
+//===- llvm/CodeGen/BranchRelaxation.h --------------------------*- 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
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CODEGEN_BRANCHRELAXATION_H
+#define LLVM_CODEGEN_BRANCHRELAXATION_H
+
+#include "llvm/CodeGen/MachinePassManager.h"
+
+namespace llvm {
+
+class BranchRelaxationPass : public PassInfoMixin<BranchRelaxationPass> {
+public:
+  PreservedAnalyses run(MachineFunction &MF,
+                        MachineFunctionAnalysisManager &MFAM);
+  static bool isRequired() { return true; }
+};
+
+} // namespace llvm
+
+#endif // LLVM_CODEGEN_BRANCHRELAXATION_H

llvm/include/llvm/InitializePasses.h

@@ -61,7 +61,7 @@ void initializeBasicAAWrapperPassPass(PassRegistry &);
 void initializeBlockFrequencyInfoWrapperPassPass(PassRegistry &);
 void initializeBranchFolderLegacyPass(PassRegistry &);
 void initializeBranchProbabilityInfoWrapperPassPass(PassRegistry &);
-void initializeBranchRelaxationPass(PassRegistry &);
+void initializeBranchRelaxationLegacyPass(PassRegistry &);
 void initializeBreakCriticalEdgesPass(PassRegistry &);
 void initializeBreakFalseDepsPass(PassRegistry &);
 void initializeCanonicalizeFreezeInLoopsPass(PassRegistry &);

llvm/include/llvm/Passes/MachinePassRegistry.def

@@ -137,6 +137,7 @@ MACHINE_FUNCTION_ANALYSIS("virtregmap", VirtRegMapAnalysis())
 #ifndef MACHINE_FUNCTION_PASS
 #define MACHINE_FUNCTION_PASS(NAME, CREATE_PASS)
 #endif
+MACHINE_FUNCTION_PASS("branch-relaxation", BranchRelaxationPass())
 MACHINE_FUNCTION_PASS("dead-mi-elimination", DeadMachineInstructionElimPass())
 MACHINE_FUNCTION_PASS("detect-dead-lanes", DetectDeadLanesPass())
 MACHINE_FUNCTION_PASS("early-ifcvt", EarlyIfConverterPass())

llvm/lib/CodeGen/BranchRelaxation.cpp

@@ -6,6 +6,7 @@
 //
 //===----------------------------------------------------------------------===//
 
+#include "llvm/CodeGen/BranchRelaxation.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/CodeGen/LivePhysRegs.h"
@@ -44,7 +45,7 @@ STATISTIC(NumUnconditionalRelaxed, "Number of unconditional branches relaxed");
 
 namespace {
 
-class BranchRelaxation : public MachineFunctionPass {
+class BranchRelaxation {
   /// BasicBlockInfo - Information about the offset and size of a single
   /// basic block.
   struct BasicBlockInfo {
@@ -115,23 +116,31 @@ class BranchRelaxation : public MachineFunctionPass {
   void dumpBBs();
   void verify();
 
+public:
+  bool run(MachineFunction &MF);
+};
+
+class BranchRelaxationLegacy : public MachineFunctionPass {
 public:
   static char ID;
 
-  BranchRelaxation() : MachineFunctionPass(ID) {}
+  BranchRelaxationLegacy() : MachineFunctionPass(ID) {}
 
-  bool runOnMachineFunction(MachineFunction &MF) override;
+  bool runOnMachineFunction(MachineFunction &MF) override {
+    return BranchRelaxation().run(MF);
+  }
 
   StringRef getPassName() const override { return BRANCH_RELAX_NAME; }
 };
 
 } // end anonymous namespace
 
-char BranchRelaxation::ID = 0;
+char BranchRelaxationLegacy::ID = 0;
 
-char &llvm::BranchRelaxationPassID = BranchRelaxation::ID;
+char &llvm::BranchRelaxationPassID = BranchRelaxationLegacy::ID;
 
-INITIALIZE_PASS(BranchRelaxation, DEBUG_TYPE, BRANCH_RELAX_NAME, false, false)
+INITIALIZE_PASS(BranchRelaxationLegacy, DEBUG_TYPE, BRANCH_RELAX_NAME, false,
+                false)
 
 /// verify - check BBOffsets, BBSizes, alignment of islands
 void BranchRelaxation::verify() {
@@ -744,7 +753,16 @@ bool BranchRelaxation::relaxBranchInstructions() {
   return Changed;
 }
 
-bool BranchRelaxation::runOnMachineFunction(MachineFunction &mf) {
+PreservedAnalyses
+BranchRelaxationPass::run(MachineFunction &MF,
+                          MachineFunctionAnalysisManager &MFAM) {
+  if (!BranchRelaxation().run(MF))
+    return PreservedAnalyses::all();
+
+  return getMachineFunctionPassPreservedAnalyses();
+}
+
+bool BranchRelaxation::run(MachineFunction &mf) {
   MF = &mf;
 
   LLVM_DEBUG(dbgs() << "***** BranchRelaxation *****\n");

llvm/lib/CodeGen/CodeGen.cpp

@@ -23,7 +23,7 @@ void llvm::initializeCodeGen(PassRegistry &Registry) {
   initializeBasicBlockPathCloningPass(Registry);
   initializeBasicBlockSectionsPass(Registry);
   initializeBranchFolderLegacyPass(Registry);
-  initializeBranchRelaxationPass(Registry);
+  initializeBranchRelaxationLegacyPass(Registry);
   initializeBreakFalseDepsPass(Registry);
   initializeCallBrPreparePass(Registry);
   initializeCFGuardLongjmpPass(Registry);

llvm/lib/Passes/PassBuilder.cpp

@@ -82,6 +82,7 @@
 #include "llvm/CodeGen/AtomicExpand.h"
 #include "llvm/CodeGen/BasicBlockSectionsProfileReader.h"
 #include "llvm/CodeGen/BranchFoldingPass.h"
+#include "llvm/CodeGen/BranchRelaxation.h"
 #include "llvm/CodeGen/CallBrPrepare.h"
 #include "llvm/CodeGen/CodeGenPrepare.h"
 #include "llvm/CodeGen/ComplexDeinterleavingPass.h"

llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp

@@ -66,6 +66,7 @@
 #include "llvm/Analysis/KernelInfo.h"
 #include "llvm/Analysis/UniformityAnalysis.h"
 #include "llvm/CodeGen/AtomicExpand.h"
+#include "llvm/CodeGen/BranchRelaxation.h"
 #include "llvm/CodeGen/DeadMachineInstructionElim.h"
 #include "llvm/CodeGen/GlobalISel/CSEInfo.h"
 #include "llvm/CodeGen/GlobalISel/IRTranslator.h"
@@ -2205,7 +2206,7 @@ void AMDGPUCodeGenPassBuilder::addPreEmitPass(AddMachinePass &addPass) const {
     addPass(AMDGPUInsertDelayAluPass());
   }
 
-  // TODO: addPass(BranchRelaxationPass());
+  addPass(BranchRelaxationPass());
 }
 
 bool AMDGPUCodeGenPassBuilder::isPassEnabled(const cl::opt<bool> &Opt,

llvm/test/CodeGen/AArch64/branch-relax-block-size.mir

@@ -1,5 +1,6 @@
 # REQUIRES: asserts
 # RUN: llc -mtriple=aarch64--linux-gnu -run-pass=branch-relaxation -debug-only=branch-relaxation %s -o /dev/null 2>&1 | FileCheck %s
+# RUN: llc -mtriple=aarch64--linux-gnu -passes=branch-relaxation -debug-only=branch-relaxation %s -o /dev/null 2>&1 | FileCheck %s
 
 # Ensure meta instructions (e.g. CFI_INSTRUCTION) don't contribute to the code
 # size of a basic block.

llvm/test/CodeGen/AArch64/branch-relax-cross-section.mir

@@ -1,6 +1,8 @@
 # RUN: llc -mtriple=aarch64-none-linux-gnu -run-pass branch-relaxation -aarch64-b-offset-bits=64 -aarch64-tbz-offset-bits=9 -aarch64-cbz-offset-bits=9 %s -o - | FileCheck %s
 # RUN: llc -mtriple=aarch64-none-linux-gnu -run-pass branch-relaxation -aarch64-tbz-offset-bits=9 -aarch64-cbz-offset-bits=9 %s -o - | FileCheck --check-prefix=INDIRECT %s
 
+# RUN: llc -mtriple=aarch64-none-linux-gnu -passes branch-relaxation -aarch64-tbz-offset-bits=9 -aarch64-cbz-offset-bits=9 %s -o - | FileCheck --check-prefix=INDIRECT %s
+
 --- |
   declare i32 @bar()
   declare i32 @baz()

llvm/test/CodeGen/AMDGPU/branch-relax-no-terminators.mir

@@ -1,5 +1,6 @@
 # NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
 # RUN: llc -verify-machineinstrs -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a --amdgpu-s-branch-bits=5 -run-pass branch-relaxation %s -o - | FileCheck %s
+# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a --amdgpu-s-branch-bits=5 -passes=branch-relaxation %s -o - | FileCheck %s
 
 ---
 name:            branch_no_terminators