[LTO] enable ObjCARCContractPass only on optimized build (#101114)

\#92331 tried to make `ObjCARCContractPass` by default, but it caused a
regression on O0 builds and was reverted.
This patch trys to bring that back by:

1. reverts the
[revert](1579e9c).
2. `createObjCARCContractPass` only on optimized builds.

Tests are updated to refelect the changes. Specifically, all `O0` tests
should not include `ObjCARCContractPass`

Signed-off-by: Peter Rong <[email protected]>

Author: DataCorrupted

clang/lib/CodeGen/BackendUtil.cpp

@@ -588,12 +588,6 @@ bool EmitAssemblyHelper::AddEmitPasses(legacy::PassManager &CodeGenPasses,
   // this also adds codegenerator level optimization passes.
   CodeGenFileType CGFT = getCodeGenFileType(Action);
 
-  // Add ObjC ARC final-cleanup optimizations. This is done as part of the
-  // "codegen" passes so that it isn't run multiple times when there is
-  // inlining happening.
-  if (CodeGenOpts.OptimizationLevel > 0)
-    CodeGenPasses.add(createObjCARCContractPass());
-
   if (TM->addPassesToEmitFile(CodeGenPasses, OS, DwoOS, CGFT,
                               /*DisableVerify=*/!CodeGenOpts.VerifyModule)) {
     Diags.Report(diag::err_fe_unable_to_interface_with_target);

clang/test/CodeGen/thinlto-distributed-objc-contract-pass.ll

@@ -0,0 +1,19 @@
+; RUN: opt -thinlto-bc -o %t.o %s
+
+; RUN: llvm-lto2 run -thinlto-distributed-indexes %t.o \
+; RUN:   -o %t2.index \
+; RUN:   -r=%t.o,_use_arc,px
+
+; RUN: %clang_cc1 -O2 -triple x86_64-apple-darwin \
+; RUN:   -emit-obj -fthinlto-index=%t.o.thinlto.bc \
+; RUN:   -o %t.native.o -x ir %t.o
+
+target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-apple-darwin"
+
+define void @use_arc(ptr %a, ptr %b) {
+  call void (...) @llvm.objc.clang.arc.use(ptr %a, ptr %b) nounwind
+  ret void
+}
+
+declare void @llvm.objc.clang.arc.use(...) nounwind

lld/MachO/LTO.cpp

@@ -48,9 +48,6 @@ static lto::Config createConfig() {
   c.CPU = getCPUStr();
   c.MAttrs = getMAttrs();
   c.DiagHandler = diagnosticHandler;
-  c.PreCodeGenPassesHook = [](legacy::PassManager &pm) {
-    pm.add(createObjCARCContractPass());
-  };
 
   c.AlwaysEmitRegularLTOObj = !config->ltoObjPath.empty();
 

llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h

@@ -41,26 +41,26 @@ extern bool EnableARCOpts;
 /// Test if the given module looks interesting to run ARC optimization
 /// on.
 inline bool ModuleHasARC(const Module &M) {
-  return
-    M.getNamedValue("llvm.objc.retain") ||
-    M.getNamedValue("llvm.objc.release") ||
-    M.getNamedValue("llvm.objc.autorelease") ||
-    M.getNamedValue("llvm.objc.retainAutoreleasedReturnValue") ||
-    M.getNamedValue("llvm.objc.unsafeClaimAutoreleasedReturnValue") ||
-    M.getNamedValue("llvm.objc.retainBlock") ||
-    M.getNamedValue("llvm.objc.autoreleaseReturnValue") ||
-    M.getNamedValue("llvm.objc.autoreleasePoolPush") ||
-    M.getNamedValue("llvm.objc.loadWeakRetained") ||
-    M.getNamedValue("llvm.objc.loadWeak") ||
-    M.getNamedValue("llvm.objc.destroyWeak") ||
-    M.getNamedValue("llvm.objc.storeWeak") ||
-    M.getNamedValue("llvm.objc.initWeak") ||
-    M.getNamedValue("llvm.objc.moveWeak") ||
-    M.getNamedValue("llvm.objc.copyWeak") ||
-    M.getNamedValue("llvm.objc.retainedObject") ||
-    M.getNamedValue("llvm.objc.unretainedObject") ||
-    M.getNamedValue("llvm.objc.unretainedPointer") ||
-    M.getNamedValue("llvm.objc.clang.arc.use");
+  return M.getNamedValue("llvm.objc.retain") ||
+         M.getNamedValue("llvm.objc.release") ||
+         M.getNamedValue("llvm.objc.autorelease") ||
+         M.getNamedValue("llvm.objc.retainAutoreleasedReturnValue") ||
+         M.getNamedValue("llvm.objc.unsafeClaimAutoreleasedReturnValue") ||
+         M.getNamedValue("llvm.objc.retainBlock") ||
+         M.getNamedValue("llvm.objc.autoreleaseReturnValue") ||
+         M.getNamedValue("llvm.objc.autoreleasePoolPush") ||
+         M.getNamedValue("llvm.objc.loadWeakRetained") ||
+         M.getNamedValue("llvm.objc.loadWeak") ||
+         M.getNamedValue("llvm.objc.destroyWeak") ||
+         M.getNamedValue("llvm.objc.storeWeak") ||
+         M.getNamedValue("llvm.objc.initWeak") ||
+         M.getNamedValue("llvm.objc.moveWeak") ||
+         M.getNamedValue("llvm.objc.copyWeak") ||
+         M.getNamedValue("llvm.objc.retainedObject") ||
+         M.getNamedValue("llvm.objc.unretainedObject") ||
+         M.getNamedValue("llvm.objc.unretainedPointer") ||
+         M.getNamedValue("llvm.objc.clang.arc.noop.use") ||
+         M.getNamedValue("llvm.objc.clang.arc.use");
 }
 
 /// This is a wrapper around getUnderlyingObject which also knows how to

llvm/lib/CodeGen/TargetPassConfig.cpp

@@ -46,6 +46,7 @@
 #include "llvm/Support/WithColor.h"
 #include "llvm/Target/CGPassBuilderOption.h"
 #include "llvm/Target/TargetMachine.h"
+#include "llvm/Transforms/ObjCARC.h"
 #include "llvm/Transforms/Scalar.h"
 #include "llvm/Transforms/Utils.h"
 #include <cassert>
@@ -943,6 +944,9 @@ void TargetPassConfig::addCodeGenPrepare() {
 void TargetPassConfig::addISelPrepare() {
   addPreISel();
 
+  if (getOptLevel() != CodeGenOptLevel::None)
+    addPass(createObjCARCContractPass());
+
   // Force codegen to run according to the callgraph.
   if (requiresCodeGenSCCOrder())
     addPass(new DummyCGSCCPass);

llvm/lib/LTO/LTOCodeGenerator.cpp

@@ -136,10 +136,6 @@ LTOCodeGenerator::LTOCodeGenerator(LLVMContext &Context)
 
   Config.CodeModel = std::nullopt;
   Config.StatsFile = LTOStatsFile;
-  Config.PreCodeGenPassesHook = [](legacy::PassManager &PM) {
-    PM.add(createObjCARCContractPass());
-  };
-
   Config.RunCSIRInstr = LTORunCSIRInstr;
   Config.CSIRProfile = LTOCSIRProfile;
 }

llvm/lib/LTO/ThinLTOCodeGenerator.cpp

@@ -335,10 +335,6 @@ std::unique_ptr<MemoryBuffer> codegenModule(Module &TheModule,
     raw_svector_ostream OS(OutputBuffer);
     legacy::PassManager PM;
 
-    // If the bitcode files contain ARC code and were compiled with optimization,
-    // the ObjCARCContractPass must be run, so do it unconditionally here.
-    PM.add(createObjCARCContractPass());
-
     // Setup the codegen now.
     if (TM.addPassesToEmitFile(PM, OS, nullptr, CodeGenFileType::ObjectFile,
                                /* DisableVerify */ true))

llvm/lib/Transforms/ObjCARC/ObjCARCContract.cpp

@@ -31,6 +31,7 @@
 #include "ProvenanceAnalysis.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Analysis/AliasAnalysis.h"
+#include "llvm/Analysis/BasicAliasAnalysis.h"
 #include "llvm/Analysis/ObjCARCUtil.h"
 #include "llvm/IR/Dominators.h"
 #include "llvm/IR/EHPersonalities.h"
@@ -71,6 +72,9 @@ class ObjCARCContract {
   ARCRuntimeEntryPoints EP;
   BundledRetainClaimRVs *BundledInsts = nullptr;
 
+  /// A flag indicating whether this optimization pass should run.
+  bool Run;
+
   /// The inline asm string to insert between calls and RetainRV calls to make
   /// the optimization work on targets which need it.
   const MDString *RVInstMarker;
@@ -527,6 +531,10 @@ bool ObjCARCContract::tryToPeepholeInstruction(
 //===----------------------------------------------------------------------===//
 
 bool ObjCARCContract::init(Module &M) {
+  Run = ModuleHasARC(M);
+  if (!Run)
+    return false;
+
   EP.init(&M);
 
   // Initialize RVInstMarker.
@@ -536,6 +544,9 @@ bool ObjCARCContract::init(Module &M) {
 }
 
 bool ObjCARCContract::run(Function &F, AAResults *A, DominatorTree *D) {
+  if (!Run)
+    return false;
+
   if (!EnableARCOpts)
     return false;
 
@@ -730,6 +741,9 @@ INITIALIZE_PASS_END(ObjCARCContractLegacyPass, "objc-arc-contract",
 void ObjCARCContractLegacyPass::getAnalysisUsage(AnalysisUsage &AU) const {
   AU.addRequired<AAResultsWrapperPass>();
   AU.addRequired<DominatorTreeWrapperPass>();
+  AU.addPreserved<AAResultsWrapperPass>();
+  AU.addPreserved<BasicAAWrapperPass>();
+  AU.addPreserved<DominatorTreeWrapperPass>();
 }
 
 Pass *llvm::createObjCARCContractPass() {

llvm/test/CodeGen/AArch64/O3-pipeline.ll

@@ -102,11 +102,14 @@
 ; CHECK-NEXT:         Dominator Tree Construction
 ; CHECK-NEXT:     FunctionPass Manager
 ; CHECK-NEXT:       Merge internal globals
+; CHECK-NEXT:       Dominator Tree Construction
+; CHECK-NEXT:       Basic Alias Analysis (stateless AA impl)
+; CHECK-NEXT:       Function Alias Analysis Results
+; CHECK-NEXT:       ObjC ARC contraction
 ; CHECK-NEXT:       Prepare callbr
 ; CHECK-NEXT:       Safe Stack instrumentation pass
 ; CHECK-NEXT:       Insert stack protectors
 ; CHECK-NEXT:       Module Verifier
-; CHECK-NEXT:       Dominator Tree Construction
 ; CHECK-NEXT:       Basic Alias Analysis (stateless AA impl)
 ; CHECK-NEXT:       Function Alias Analysis Results
 ; CHECK-NEXT:       Natural Loop Information

llvm/test/CodeGen/AMDGPU/llc-pipeline.ll

@@ -279,6 +279,9 @@
 ; GCN-O1-NEXT:        AMDGPU Rewrite Undef for PHI
 ; GCN-O1-NEXT:        LCSSA Verifier
 ; GCN-O1-NEXT:        Loop-Closed SSA Form Pass
+; GCN-O1-NEXT:        Basic Alias Analysis (stateless AA impl)
+; GCN-O1-NEXT:        Function Alias Analysis Results
+; GCN-O1-NEXT:        ObjC ARC contraction
 ; GCN-O1-NEXT:      DummyCGSCCPass
 ; GCN-O1-NEXT:      FunctionPass Manager
 ; GCN-O1-NEXT:        Prepare callbr
@@ -571,6 +574,9 @@
 ; GCN-O1-OPTS-NEXT:        AMDGPU Rewrite Undef for PHI
 ; GCN-O1-OPTS-NEXT:        LCSSA Verifier
 ; GCN-O1-OPTS-NEXT:        Loop-Closed SSA Form Pass
+; GCN-O1-OPTS-NEXT:        Basic Alias Analysis (stateless AA impl)
+; GCN-O1-OPTS-NEXT:        Function Alias Analysis Results
+; GCN-O1-OPTS-NEXT:        ObjC ARC contraction
 ; GCN-O1-OPTS-NEXT:      DummyCGSCCPass
 ; GCN-O1-OPTS-NEXT:      FunctionPass Manager
 ; GCN-O1-OPTS-NEXT:        Prepare callbr
@@ -876,6 +882,11 @@
 ; GCN-O2-NEXT:        LCSSA Verifier
 ; GCN-O2-NEXT:        Loop-Closed SSA Form Pass
 ; GCN-O2-NEXT:      Analysis if a function is memory bound
+; GCN-O2-NEXT:      FunctionPass Manager
+; GCN-O2-NEXT:        Dominator Tree Construction
+; GCN-O2-NEXT:        Basic Alias Analysis (stateless AA impl)
+; GCN-O2-NEXT:        Function Alias Analysis Results
+; GCN-O2-NEXT:        ObjC ARC contraction
 ; GCN-O2-NEXT:      DummyCGSCCPass
 ; GCN-O2-NEXT:      FunctionPass Manager
 ; GCN-O2-NEXT:        Prepare callbr
@@ -1194,6 +1205,11 @@
 ; GCN-O3-NEXT:        LCSSA Verifier
 ; GCN-O3-NEXT:        Loop-Closed SSA Form Pass
 ; GCN-O3-NEXT:      Analysis if a function is memory bound
+; GCN-O3-NEXT:      FunctionPass Manager
+; GCN-O3-NEXT:        Dominator Tree Construction
+; GCN-O3-NEXT:        Basic Alias Analysis (stateless AA impl)
+; GCN-O3-NEXT:        Function Alias Analysis Results
+; GCN-O3-NEXT:        ObjC ARC contraction
 ; GCN-O3-NEXT:      DummyCGSCCPass
 ; GCN-O3-NEXT:      FunctionPass Manager
 ; GCN-O3-NEXT:        Prepare callbr