[CodeGen] Port ExpandMemCmp to new pass manager

Author: paperchalice

llvm/include/llvm/CodeGen/CodeGenPassBuilder.h

@@ -25,6 +25,7 @@
 #include "llvm/Analysis/TypeBasedAliasAnalysis.h"
 #include "llvm/CodeGen/CallBrPrepare.h"
 #include "llvm/CodeGen/DwarfEHPrepare.h"
+#include "llvm/CodeGen/ExpandMemCmp.h"
 #include "llvm/CodeGen/ExpandReductions.h"
 #include "llvm/CodeGen/GCMetadata.h"
 #include "llvm/CodeGen/InterleavedAccess.h"
@@ -627,7 +628,7 @@ void CodeGenPassBuilder<Derived>::addIRPasses(AddIRPass &addPass) const {
     // target lowering hook.
     if (!Opt.DisableMergeICmps)
       addPass(MergeICmpsPass());
-    addPass(ExpandMemCmpPass());
+    addPass(ExpandMemCmpPass(&TM));
   }
 
   // Run GC lowering passes for builtin collectors

llvm/include/llvm/CodeGen/ExpandMemCmp.h

@@ -0,0 +1,29 @@
+//===--- ExpandMemCmp.h - Expand memcmp() to load/stores --------*- 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_EXPANDMEMCMP_H
+#define LLVM_CODEGEN_EXPANDMEMCMP_H
+
+#include "llvm/IR/PassManager.h"
+
+namespace llvm {
+
+class TargetMachine;
+
+class ExpandMemCmpPass : public PassInfoMixin<ExpandMemCmpPass> {
+  const TargetMachine *TM;
+
+public:
+  explicit ExpandMemCmpPass(const TargetMachine *TM_) : TM(TM_) {}
+
+  PreservedAnalyses run(Function &F, FunctionAnalysisManager &FAM);
+};
+
+} // namespace llvm
+
+#endif // LLVM_CODEGEN_EXPANDMEMCMP_H

llvm/include/llvm/CodeGen/MachinePassRegistry.def

@@ -46,6 +46,7 @@ FUNCTION_PASS("dwarf-eh-prepare", DwarfEHPreparePass, (TM))
 FUNCTION_PASS("ee-instrument", EntryExitInstrumenterPass, (false))
 FUNCTION_PASS("expand-large-div-rem", ExpandLargeDivRemPass, ())
 FUNCTION_PASS("expand-large-fp-convert", ExpandLargeFpConvertPass, ())
+FUNCTION_PASS("expand-memcmp", ExpandMemCmpPass, (TM))
 FUNCTION_PASS("expand-reductions", ExpandReductionsPass, ())
 FUNCTION_PASS("expandvp", ExpandVectorPredicationPass, ())
 FUNCTION_PASS("interleaved-access", InterleavedAccessPass, (TM))
@@ -129,7 +130,6 @@ MACHINE_FUNCTION_ANALYSIS("pass-instrumentation", PassInstrumentationAnalysis,
 #endif
 DUMMY_FUNCTION_PASS("atomic-expand", AtomicExpandPass, ())
 DUMMY_FUNCTION_PASS("codegenprepare", CodeGenPreparePass, ())
-DUMMY_FUNCTION_PASS("expandmemcmp", ExpandMemCmpPass, ())
 DUMMY_FUNCTION_PASS("gc-lowering", GCLoweringPass, ())
 DUMMY_FUNCTION_PASS("indirectbr-expand", IndirectBrExpandPass, ())
 DUMMY_FUNCTION_PASS("shadow-stack-gc-lowering", ShadowStackGCLoweringPass, ())

llvm/lib/CodeGen/ExpandMemCmp.cpp

@@ -11,6 +11,7 @@
 //
 //===----------------------------------------------------------------------===//
 
+#include "llvm/CodeGen/ExpandMemCmp.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Analysis/ConstantFolding.h"
 #include "llvm/Analysis/DomTreeUpdater.h"
@@ -38,7 +39,7 @@ namespace llvm {
 class TargetLowering;
 }
 
-#define DEBUG_TYPE "expandmemcmp"
+#define DEBUG_TYPE "expand-memcmp"
 
 STATISTIC(NumMemCmpCalls, "Number of memcmp calls");
 STATISTIC(NumMemCmpNotConstant, "Number of memcmp calls without constant size");
@@ -886,6 +887,18 @@ static bool expandMemCmp(CallInst *CI, const TargetTransformInfo *TTI,
   return true;
 }
 
+// Returns true if a change was made.
+static bool runOnBlock(BasicBlock &BB, const TargetLibraryInfo *TLI,
+                       const TargetTransformInfo *TTI, const TargetLowering *TL,
+                       const DataLayout &DL, ProfileSummaryInfo *PSI,
+                       BlockFrequencyInfo *BFI, DomTreeUpdater *DTU);
+
+static PreservedAnalyses runImpl(Function &F, const TargetLibraryInfo *TLI,
+                                 const TargetTransformInfo *TTI,
+                                 const TargetLowering *TL,
+                                 ProfileSummaryInfo *PSI,
+                                 BlockFrequencyInfo *BFI, DominatorTree *DT);
+
 class ExpandMemCmpLegacyPass : public FunctionPass {
 public:
   static char ID;
@@ -928,25 +941,13 @@ class ExpandMemCmpLegacyPass : public FunctionPass {
     LazyBlockFrequencyInfoPass::getLazyBFIAnalysisUsage(AU);
     FunctionPass::getAnalysisUsage(AU);
   }
-
-  PreservedAnalyses runImpl(Function &F, const TargetLibraryInfo *TLI,
-                            const TargetTransformInfo *TTI,
-                            const TargetLowering *TL, ProfileSummaryInfo *PSI,
-                            BlockFrequencyInfo *BFI, DominatorTree *DT);
-  // Returns true if a change was made.
-  bool runOnBlock(BasicBlock &BB, const TargetLibraryInfo *TLI,
-                  const TargetTransformInfo *TTI, const TargetLowering *TL,
-                  const DataLayout &DL, ProfileSummaryInfo *PSI,
-                  BlockFrequencyInfo *BFI, DomTreeUpdater *DTU);
 };
 
-bool ExpandMemCmpLegacyPass::runOnBlock(BasicBlock &BB, const TargetLibraryInfo *TLI,
-                                  const TargetTransformInfo *TTI,
-                                  const TargetLowering *TL,
-                                  const DataLayout &DL, ProfileSummaryInfo *PSI,
-                                  BlockFrequencyInfo *BFI,
-                                  DomTreeUpdater *DTU) {
-  for (Instruction& I : BB) {
+bool runOnBlock(BasicBlock &BB, const TargetLibraryInfo *TLI,
+                const TargetTransformInfo *TTI, const TargetLowering *TL,
+                const DataLayout &DL, ProfileSummaryInfo *PSI,
+                BlockFrequencyInfo *BFI, DomTreeUpdater *DTU) {
+  for (Instruction &I : BB) {
     CallInst *CI = dyn_cast<CallInst>(&I);
     if (!CI) {
       continue;
@@ -961,8 +962,7 @@ bool ExpandMemCmpLegacyPass::runOnBlock(BasicBlock &BB, const TargetLibraryInfo
   return false;
 }
 
-PreservedAnalyses
-ExpandMemCmpLegacyPass::runImpl(Function &F, const TargetLibraryInfo *TLI,
+PreservedAnalyses runImpl(Function &F, const TargetLibraryInfo *TLI,
                           const TargetTransformInfo *TTI,
                           const TargetLowering *TL, ProfileSummaryInfo *PSI,
                           BlockFrequencyInfo *BFI, DominatorTree *DT) {
@@ -994,16 +994,30 @@ ExpandMemCmpLegacyPass::runImpl(Function &F, const TargetLibraryInfo *TLI,
 
 } // namespace
 
+PreservedAnalyses ExpandMemCmpPass::run(Function &F,
+                                        FunctionAnalysisManager &FAM) {
+  const auto *TL = TM->getSubtargetImpl(F)->getTargetLowering();
+  const auto &TLI = FAM.getResult<TargetLibraryAnalysis>(F);
+  const auto &TTI = FAM.getResult<TargetIRAnalysis>(F);
+  auto *PSI = FAM.getResult<ModuleAnalysisManagerFunctionProxy>(F)
+                  .getCachedResult<ProfileSummaryAnalysis>(*F.getParent());
+  BlockFrequencyInfo *BFI = (PSI && PSI->hasProfileSummary())
+                                ? &FAM.getResult<BlockFrequencyAnalysis>(F)
+                                : nullptr;
+  auto *DT = FAM.getCachedResult<DominatorTreeAnalysis>(F);
+
+  return runImpl(F, &TLI, &TTI, TL, PSI, BFI, DT);
+}
 
 char ExpandMemCmpLegacyPass::ID = 0;
-INITIALIZE_PASS_BEGIN(ExpandMemCmpLegacyPass, "expandmemcmp",
+INITIALIZE_PASS_BEGIN(ExpandMemCmpLegacyPass, DEBUG_TYPE,
                       "Expand memcmp() to load/stores", false, false)
 INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(TargetTransformInfoWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(LazyBlockFrequencyInfoPass)
 INITIALIZE_PASS_DEPENDENCY(ProfileSummaryInfoWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
-INITIALIZE_PASS_END(ExpandMemCmpLegacyPass, "expandmemcmp",
+INITIALIZE_PASS_END(ExpandMemCmpLegacyPass, DEBUG_TYPE,
                     "Expand memcmp() to load/stores", false, false)
 
 FunctionPass *llvm::createExpandMemCmpLegacyPass() {

llvm/lib/Passes/PassBuilder.cpp

@@ -76,6 +76,7 @@
 #include "llvm/CodeGen/DwarfEHPrepare.h"
 #include "llvm/CodeGen/ExpandLargeDivRem.h"
 #include "llvm/CodeGen/ExpandLargeFpConvert.h"
+#include "llvm/CodeGen/ExpandMemCmp.h"
 #include "llvm/CodeGen/GCMetadata.h"
 #include "llvm/CodeGen/HardwareLoops.h"
 #include "llvm/CodeGen/InterleavedAccess.h"

llvm/lib/Passes/PassRegistry.def

@@ -308,6 +308,7 @@ FUNCTION_PASS("dse", DSEPass())
 FUNCTION_PASS("dwarf-eh-prepare", DwarfEHPreparePass(TM))
 FUNCTION_PASS("expand-large-div-rem", ExpandLargeDivRemPass(TM))
 FUNCTION_PASS("expand-large-fp-convert", ExpandLargeFpConvertPass(TM))
+FUNCTION_PASS("expand-memcmp", ExpandMemCmpPass(TM))
 FUNCTION_PASS("fix-irreducible", FixIrreduciblePass())
 FUNCTION_PASS("flattencfg", FlattenCFGPass())
 FUNCTION_PASS("float2int", Float2IntPass())

llvm/test/CodeGen/PowerPC/memcmp-mergeexpand.ll

@@ -1,7 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -verify-machineinstrs -mcpu=pwr8 -mtriple=powerpc64le-unknown-gnu-linux  < %s | FileCheck %s -check-prefix=PPC64LE
 
-; This tests interaction between MergeICmp and ExpandMemCmp.
+; This tests interaction between MergeICmp and expand-memcmp.
 
 %"struct.std::pair" = type { i32, i32 }
 

llvm/test/Transforms/ExpandMemCmp/AArch64/memcmp.ll

@@ -1,5 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3
-; RUN: opt -S -expandmemcmp -memcmp-num-loads-per-block=1 -mtriple=aarch64-unknown-unknown < %s | FileCheck %s
+; RUN: opt -S -expand-memcmp -memcmp-num-loads-per-block=1 -mtriple=aarch64-unknown-unknown < %s | FileCheck %s
+; RUN: opt -S -passes=expand-memcmp -memcmp-num-loads-per-block=1 -mtriple=aarch64-unknown-unknown < %s | FileCheck %s
 
 declare i32 @memcmp(ptr nocapture, ptr nocapture, i64)
 

llvm/test/Transforms/ExpandMemCmp/X86/bcmp.ll

@@ -1,5 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -S -expandmemcmp -memcmp-num-loads-per-block=1 -mtriple=x86_64-unknown-unknown -data-layout=e-m:o-i64:64-f80:128-n8:16:32:64-S128         < %s | FileCheck %s --check-prefix=X64
+; RUN: opt -S -expand-memcmp -memcmp-num-loads-per-block=1 -mtriple=x86_64-unknown-unknown -data-layout=e-m:o-i64:64-f80:128-n8:16:32:64-S128         < %s | FileCheck %s --check-prefix=X64
+; RUN: opt -S -passes=expand-memcmp -memcmp-num-loads-per-block=1 -mtriple=x86_64-unknown-unknown -data-layout=e-m:o-i64:64-f80:128-n8:16:32:64-S128         < %s | FileCheck %s --check-prefix=X64
 
 declare i32 @bcmp(ptr nocapture, ptr nocapture, i64)
 

llvm/test/Transforms/ExpandMemCmp/X86/memcmp-x32.ll

@@ -1,5 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -S -expandmemcmp -mtriple=i686-unknown-unknown   -data-layout=e-m:o-p:32:32-f64:32:64-f80:128-n8:16:32-S128 < %s | FileCheck %s --check-prefix=X32
+; RUN: opt -S -expand-memcmp -mtriple=i686-unknown-unknown   -data-layout=e-m:o-p:32:32-f64:32:64-f80:128-n8:16:32-S128 < %s | FileCheck %s --check-prefix=X32
+; RUN: opt -S -passes=expand-memcmp -mtriple=i686-unknown-unknown   -data-layout=e-m:o-p:32:32-f64:32:64-f80:128-n8:16:32-S128 < %s | FileCheck %s --check-prefix=X32
 
 declare i32 @memcmp(ptr nocapture, ptr nocapture, i32)
 

llvm/test/Transforms/ExpandMemCmp/X86/memcmp.ll

@@ -1,6 +1,8 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -S -expandmemcmp -memcmp-num-loads-per-block=1 -mtriple=x86_64-unknown-unknown -data-layout=e-m:o-i64:64-f80:128-n8:16:32:64-S128         < %s | FileCheck %s --check-prefix=X64 --check-prefix=X64_1LD
-; RUN: opt -S -expandmemcmp -memcmp-num-loads-per-block=2 -mtriple=x86_64-unknown-unknown -data-layout=e-m:o-i64:64-f80:128-n8:16:32:64-S128         < %s | FileCheck %s --check-prefix=X64 --check-prefix=X64_2LD
+; RUN: opt -S -expand-memcmp -memcmp-num-loads-per-block=1 -mtriple=x86_64-unknown-unknown -data-layout=e-m:o-i64:64-f80:128-n8:16:32:64-S128         < %s | FileCheck %s --check-prefix=X64 --check-prefix=X64_1LD
+; RUN: opt -S -expand-memcmp -memcmp-num-loads-per-block=2 -mtriple=x86_64-unknown-unknown -data-layout=e-m:o-i64:64-f80:128-n8:16:32:64-S128         < %s | FileCheck %s --check-prefix=X64 --check-prefix=X64_2LD
+; RUN: opt -S -passes=expand-memcmp -memcmp-num-loads-per-block=1 -mtriple=x86_64-unknown-unknown -data-layout=e-m:o-i64:64-f80:128-n8:16:32:64-S128         < %s | FileCheck %s --check-prefix=X64 --check-prefix=X64_1LD
+; RUN: opt -S -passes=expand-memcmp -memcmp-num-loads-per-block=2 -mtriple=x86_64-unknown-unknown -data-layout=e-m:o-i64:64-f80:128-n8:16:32:64-S128         < %s | FileCheck %s --check-prefix=X64 --check-prefix=X64_2LD
 
 declare i32 @memcmp(ptr nocapture, ptr nocapture, i64)
 

llvm/tools/opt/opt.cpp

@@ -356,7 +356,7 @@ static bool shouldPinPassToLegacyPM(StringRef Pass) {
       "expand-reductions",
       "indirectbr-expand",
       "generic-to-nvvm",
-      "expandmemcmp",
+      "expand-memcmp",
       "loop-reduce",
       "lower-amx-type",
       "lower-amx-intrinsics",