[LTO] enable `ObjCARCContractPass` only on optimized build #101114

DataCorrupted · 2024-07-30T03:20:49Z

Summary:
#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:

reverts the revert.
createObjCARCContractPass only on optimized builds.

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

llvmbot · 2024-07-30T03:21:21Z

@llvm/pr-subscribers-llvm-transforms
@llvm/pr-subscribers-backend-loongarch
@llvm/pr-subscribers-lto
@llvm/pr-subscribers-backend-arm
@llvm/pr-subscribers-lld
@llvm/pr-subscribers-lld-macho
@llvm/pr-subscribers-backend-amdgpu

@llvm/pr-subscribers-clang-codegen

Author: Peter Rong (DataCorrupted)

Changes

Summary:
#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:

reverts the revert.
createObjCARCContractPass only on optimized builds.

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

Full diff: https://github.com/llvm/llvm-project/pull/101114.diff

15 Files Affected:

(modified) clang/lib/CodeGen/BackendUtil.cpp (-6)
(added) clang/test/CodeGen/thinlto-distributed-objc-contract-pass.ll (+19)
(modified) lld/MachO/LTO.cpp (-3)
(modified) llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h (+20-20)
(modified) llvm/lib/CodeGen/TargetPassConfig.cpp (+4)
(modified) llvm/lib/LTO/LTOCodeGenerator.cpp (-4)
(modified) llvm/lib/LTO/ThinLTOCodeGenerator.cpp (-4)
(modified) llvm/test/CodeGen/AArch64/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/AMDGPU/llc-pipeline.ll (+16)
(modified) llvm/test/CodeGen/ARM/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/LoongArch/opt-pipeline.ll (+4)
(modified) llvm/test/CodeGen/M68k/pipeline.ll (+4)
(modified) llvm/test/CodeGen/PowerPC/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/RISCV/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/X86/opt-pipeline.ll (+4)

diff --git a/clang/lib/CodeGen/BackendUtil.cpp b/clang/lib/CodeGen/BackendUtil.cpp
index e765bbf637a66..81e6702d5de66 100644
--- a/clang/lib/CodeGen/BackendUtil.cpp
+++ b/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);
diff --git a/clang/test/CodeGen/thinlto-distributed-objc-contract-pass.ll b/clang/test/CodeGen/thinlto-distributed-objc-contract-pass.ll
new file mode 100644
index 0000000000000..e0256da314306
--- /dev/null
+++ b/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
diff --git a/lld/MachO/LTO.cpp b/lld/MachO/LTO.cpp
index 7a9a9223a0322..6527cbb68f249 100644
--- a/lld/MachO/LTO.cpp
+++ b/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();
 
diff --git a/llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h b/llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h
index ccf859922e163..5ac6ad47f7799 100644
--- a/llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h
+++ b/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
diff --git a/llvm/lib/CodeGen/TargetPassConfig.cpp b/llvm/lib/CodeGen/TargetPassConfig.cpp
index 3658e8320a0cc..d833555d44b25 100644
--- a/llvm/lib/CodeGen/TargetPassConfig.cpp
+++ b/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>
@@ -949,6 +950,9 @@ 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);
diff --git a/llvm/lib/LTO/LTOCodeGenerator.cpp b/llvm/lib/LTO/LTOCodeGenerator.cpp
index 34aacb660144f..a192392e04585 100644
--- a/llvm/lib/LTO/LTOCodeGenerator.cpp
+++ b/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;
 }
diff --git a/llvm/lib/LTO/ThinLTOCodeGenerator.cpp b/llvm/lib/LTO/ThinLTOCodeGenerator.cpp
index 57c94e1988b79..b982df7c6e5d3 100644
--- a/llvm/lib/LTO/ThinLTOCodeGenerator.cpp
+++ b/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))
diff --git a/llvm/test/CodeGen/AArch64/O3-pipeline.ll b/llvm/test/CodeGen/AArch64/O3-pipeline.ll
index c5d604a5a2783..8fe2374103e44 100644
--- a/llvm/test/CodeGen/AArch64/O3-pipeline.ll
+++ b/llvm/test/CodeGen/AArch64/O3-pipeline.ll
@@ -104,6 +104,10 @@
 ; 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
diff --git a/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll b/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
index b61838c06a1f9..4f64a6bb898a4 100644
--- a/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
+++ b/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
@@ -283,6 +283,9 @@
 ; 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
@@ -577,6 +580,9 @@
 ; 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
@@ -884,6 +890,11 @@
 ; 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
@@ -1204,6 +1215,11 @@
 ; 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
diff --git a/llvm/test/CodeGen/ARM/O3-pipeline.ll b/llvm/test/CodeGen/ARM/O3-pipeline.ll
index 461920e1c5da1..18b8c7e3ced5d 100644
--- a/llvm/test/CodeGen/ARM/O3-pipeline.ll
+++ b/llvm/test/CodeGen/ARM/O3-pipeline.ll
@@ -66,6 +66,10 @@
 ; CHECK-NEXT:        Transform predicated vector loops to use MVE tail predication
 ; CHECK-NEXT:      A No-Op Barrier Pass
 ; CHECK-NEXT:      FunctionPass Manager
+; 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
diff --git a/llvm/test/CodeGen/LoongArch/opt-pipeline.ll b/llvm/test/CodeGen/LoongArch/opt-pipeline.ll
index 4e5a5433596db..fd3fb293ae7e6 100644
--- a/llvm/test/CodeGen/LoongArch/opt-pipeline.ll
+++ b/llvm/test/CodeGen/LoongArch/opt-pipeline.ll
@@ -72,10 +72,14 @@
 ; LAXX-NEXT:       CodeGen Prepare
 ; LAXX-NEXT:       Dominator Tree Construction
 ; LAXX-NEXT:       Exception handling preparation
+; LAXX-NEXT:       Basic Alias Analysis (stateless AA impl)
+; LAXX-NEXT:       Function Alias Analysis Results
+; LAXX-NEXT:       ObjC ARC contraction
 ; LAXX-NEXT:       Prepare callbr
 ; LAXX-NEXT:       Safe Stack instrumentation pass
 ; LAXX-NEXT:       Insert stack protectors
 ; LAXX-NEXT:       Module Verifier
+; LAXX-NEXT:       Dominator Tree Construction
 ; LAXX-NEXT:       Basic Alias Analysis (stateless AA impl)
 ; LAXX-NEXT:       Function Alias Analysis Results
 ; LAXX-NEXT:       Natural Loop Information
diff --git a/llvm/test/CodeGen/M68k/pipeline.ll b/llvm/test/CodeGen/M68k/pipeline.ll
index e7d0b20640f7a..1c9184a4d7e71 100644
--- a/llvm/test/CodeGen/M68k/pipeline.ll
+++ b/llvm/test/CodeGen/M68k/pipeline.ll
@@ -42,10 +42,14 @@
 ; CHECK-NEXT:      CodeGen Prepare
 ; CHECK-NEXT:      Dominator Tree Construction
 ; CHECK-NEXT:      Exception handling preparation
+; 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
diff --git a/llvm/test/CodeGen/PowerPC/O3-pipeline.ll b/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
index a564a5517f579..2954d051a2ea9 100644
--- a/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
+++ b/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
@@ -83,10 +83,14 @@
 ; CHECK-NEXT:       Lazy Block Frequency Analysis
 ; CHECK-NEXT:       Optimization Remark Emitter
 ; CHECK-NEXT:       Hardware Loop Insertion
+; 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
diff --git a/llvm/test/CodeGen/RISCV/O3-pipeline.ll b/llvm/test/CodeGen/RISCV/O3-pipeline.ll
index e305a74d7aef3..af5151b6f938c 100644
--- a/llvm/test/CodeGen/RISCV/O3-pipeline.ll
+++ b/llvm/test/CodeGen/RISCV/O3-pipeline.ll
@@ -75,6 +75,10 @@
 ; CHECK-NEXT:       Exception handling preparation
 ; CHECK-NEXT:     A No-Op Barrier Pass
 ; CHECK-NEXT:     FunctionPass Manager
+; 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
diff --git a/llvm/test/CodeGen/X86/opt-pipeline.ll b/llvm/test/CodeGen/X86/opt-pipeline.ll
index 19774b7205109..d9d6170585035 100644
--- a/llvm/test/CodeGen/X86/opt-pipeline.ll
+++ b/llvm/test/CodeGen/X86/opt-pipeline.ll
@@ -73,10 +73,14 @@
 ; CHECK-NEXT:       CodeGen Prepare
 ; CHECK-NEXT:       Dominator Tree Construction
 ; CHECK-NEXT:       Exception handling preparation
+; 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

llvmbot · 2024-07-30T03:21:21Z

@llvm/pr-subscribers-backend-powerpc

Author: Peter Rong (DataCorrupted)

Changes

Summary:
#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:

reverts the revert.
createObjCARCContractPass only on optimized builds.

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

Full diff: https://github.com/llvm/llvm-project/pull/101114.diff

15 Files Affected:

(modified) clang/lib/CodeGen/BackendUtil.cpp (-6)
(added) clang/test/CodeGen/thinlto-distributed-objc-contract-pass.ll (+19)
(modified) lld/MachO/LTO.cpp (-3)
(modified) llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h (+20-20)
(modified) llvm/lib/CodeGen/TargetPassConfig.cpp (+4)
(modified) llvm/lib/LTO/LTOCodeGenerator.cpp (-4)
(modified) llvm/lib/LTO/ThinLTOCodeGenerator.cpp (-4)
(modified) llvm/test/CodeGen/AArch64/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/AMDGPU/llc-pipeline.ll (+16)
(modified) llvm/test/CodeGen/ARM/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/LoongArch/opt-pipeline.ll (+4)
(modified) llvm/test/CodeGen/M68k/pipeline.ll (+4)
(modified) llvm/test/CodeGen/PowerPC/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/RISCV/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/X86/opt-pipeline.ll (+4)

diff --git a/clang/lib/CodeGen/BackendUtil.cpp b/clang/lib/CodeGen/BackendUtil.cpp
index e765bbf637a66..81e6702d5de66 100644
--- a/clang/lib/CodeGen/BackendUtil.cpp
+++ b/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);
diff --git a/clang/test/CodeGen/thinlto-distributed-objc-contract-pass.ll b/clang/test/CodeGen/thinlto-distributed-objc-contract-pass.ll
new file mode 100644
index 0000000000000..e0256da314306
--- /dev/null
+++ b/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
diff --git a/lld/MachO/LTO.cpp b/lld/MachO/LTO.cpp
index 7a9a9223a0322..6527cbb68f249 100644
--- a/lld/MachO/LTO.cpp
+++ b/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();
 
diff --git a/llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h b/llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h
index ccf859922e163..5ac6ad47f7799 100644
--- a/llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h
+++ b/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
diff --git a/llvm/lib/CodeGen/TargetPassConfig.cpp b/llvm/lib/CodeGen/TargetPassConfig.cpp
index 3658e8320a0cc..d833555d44b25 100644
--- a/llvm/lib/CodeGen/TargetPassConfig.cpp
+++ b/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>
@@ -949,6 +950,9 @@ 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);
diff --git a/llvm/lib/LTO/LTOCodeGenerator.cpp b/llvm/lib/LTO/LTOCodeGenerator.cpp
index 34aacb660144f..a192392e04585 100644
--- a/llvm/lib/LTO/LTOCodeGenerator.cpp
+++ b/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;
 }
diff --git a/llvm/lib/LTO/ThinLTOCodeGenerator.cpp b/llvm/lib/LTO/ThinLTOCodeGenerator.cpp
index 57c94e1988b79..b982df7c6e5d3 100644
--- a/llvm/lib/LTO/ThinLTOCodeGenerator.cpp
+++ b/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))
diff --git a/llvm/test/CodeGen/AArch64/O3-pipeline.ll b/llvm/test/CodeGen/AArch64/O3-pipeline.ll
index c5d604a5a2783..8fe2374103e44 100644
--- a/llvm/test/CodeGen/AArch64/O3-pipeline.ll
+++ b/llvm/test/CodeGen/AArch64/O3-pipeline.ll
@@ -104,6 +104,10 @@
 ; 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
diff --git a/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll b/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
index b61838c06a1f9..4f64a6bb898a4 100644
--- a/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
+++ b/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
@@ -283,6 +283,9 @@
 ; 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
@@ -577,6 +580,9 @@
 ; 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
@@ -884,6 +890,11 @@
 ; 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
@@ -1204,6 +1215,11 @@
 ; 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
diff --git a/llvm/test/CodeGen/ARM/O3-pipeline.ll b/llvm/test/CodeGen/ARM/O3-pipeline.ll
index 461920e1c5da1..18b8c7e3ced5d 100644
--- a/llvm/test/CodeGen/ARM/O3-pipeline.ll
+++ b/llvm/test/CodeGen/ARM/O3-pipeline.ll
@@ -66,6 +66,10 @@
 ; CHECK-NEXT:        Transform predicated vector loops to use MVE tail predication
 ; CHECK-NEXT:      A No-Op Barrier Pass
 ; CHECK-NEXT:      FunctionPass Manager
+; 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
diff --git a/llvm/test/CodeGen/LoongArch/opt-pipeline.ll b/llvm/test/CodeGen/LoongArch/opt-pipeline.ll
index 4e5a5433596db..fd3fb293ae7e6 100644
--- a/llvm/test/CodeGen/LoongArch/opt-pipeline.ll
+++ b/llvm/test/CodeGen/LoongArch/opt-pipeline.ll
@@ -72,10 +72,14 @@
 ; LAXX-NEXT:       CodeGen Prepare
 ; LAXX-NEXT:       Dominator Tree Construction
 ; LAXX-NEXT:       Exception handling preparation
+; LAXX-NEXT:       Basic Alias Analysis (stateless AA impl)
+; LAXX-NEXT:       Function Alias Analysis Results
+; LAXX-NEXT:       ObjC ARC contraction
 ; LAXX-NEXT:       Prepare callbr
 ; LAXX-NEXT:       Safe Stack instrumentation pass
 ; LAXX-NEXT:       Insert stack protectors
 ; LAXX-NEXT:       Module Verifier
+; LAXX-NEXT:       Dominator Tree Construction
 ; LAXX-NEXT:       Basic Alias Analysis (stateless AA impl)
 ; LAXX-NEXT:       Function Alias Analysis Results
 ; LAXX-NEXT:       Natural Loop Information
diff --git a/llvm/test/CodeGen/M68k/pipeline.ll b/llvm/test/CodeGen/M68k/pipeline.ll
index e7d0b20640f7a..1c9184a4d7e71 100644
--- a/llvm/test/CodeGen/M68k/pipeline.ll
+++ b/llvm/test/CodeGen/M68k/pipeline.ll
@@ -42,10 +42,14 @@
 ; CHECK-NEXT:      CodeGen Prepare
 ; CHECK-NEXT:      Dominator Tree Construction
 ; CHECK-NEXT:      Exception handling preparation
+; 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
diff --git a/llvm/test/CodeGen/PowerPC/O3-pipeline.ll b/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
index a564a5517f579..2954d051a2ea9 100644
--- a/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
+++ b/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
@@ -83,10 +83,14 @@
 ; CHECK-NEXT:       Lazy Block Frequency Analysis
 ; CHECK-NEXT:       Optimization Remark Emitter
 ; CHECK-NEXT:       Hardware Loop Insertion
+; 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
diff --git a/llvm/test/CodeGen/RISCV/O3-pipeline.ll b/llvm/test/CodeGen/RISCV/O3-pipeline.ll
index e305a74d7aef3..af5151b6f938c 100644
--- a/llvm/test/CodeGen/RISCV/O3-pipeline.ll
+++ b/llvm/test/CodeGen/RISCV/O3-pipeline.ll
@@ -75,6 +75,10 @@
 ; CHECK-NEXT:       Exception handling preparation
 ; CHECK-NEXT:     A No-Op Barrier Pass
 ; CHECK-NEXT:     FunctionPass Manager
+; 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
diff --git a/llvm/test/CodeGen/X86/opt-pipeline.ll b/llvm/test/CodeGen/X86/opt-pipeline.ll
index 19774b7205109..d9d6170585035 100644
--- a/llvm/test/CodeGen/X86/opt-pipeline.ll
+++ b/llvm/test/CodeGen/X86/opt-pipeline.ll
@@ -73,10 +73,14 @@
 ; CHECK-NEXT:       CodeGen Prepare
 ; CHECK-NEXT:       Dominator Tree Construction
 ; CHECK-NEXT:       Exception handling preparation
+; 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

llvmbot · 2024-07-30T03:21:22Z

@llvm/pr-subscribers-clang

Author: Peter Rong (DataCorrupted)

Changes

Summary:
#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:

reverts the revert.
createObjCARCContractPass only on optimized builds.

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

Full diff: https://github.com/llvm/llvm-project/pull/101114.diff

15 Files Affected:

(modified) clang/lib/CodeGen/BackendUtil.cpp (-6)
(added) clang/test/CodeGen/thinlto-distributed-objc-contract-pass.ll (+19)
(modified) lld/MachO/LTO.cpp (-3)
(modified) llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h (+20-20)
(modified) llvm/lib/CodeGen/TargetPassConfig.cpp (+4)
(modified) llvm/lib/LTO/LTOCodeGenerator.cpp (-4)
(modified) llvm/lib/LTO/ThinLTOCodeGenerator.cpp (-4)
(modified) llvm/test/CodeGen/AArch64/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/AMDGPU/llc-pipeline.ll (+16)
(modified) llvm/test/CodeGen/ARM/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/LoongArch/opt-pipeline.ll (+4)
(modified) llvm/test/CodeGen/M68k/pipeline.ll (+4)
(modified) llvm/test/CodeGen/PowerPC/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/RISCV/O3-pipeline.ll (+4)
(modified) llvm/test/CodeGen/X86/opt-pipeline.ll (+4)

diff --git a/clang/lib/CodeGen/BackendUtil.cpp b/clang/lib/CodeGen/BackendUtil.cpp
index e765bbf637a66..81e6702d5de66 100644
--- a/clang/lib/CodeGen/BackendUtil.cpp
+++ b/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);
diff --git a/clang/test/CodeGen/thinlto-distributed-objc-contract-pass.ll b/clang/test/CodeGen/thinlto-distributed-objc-contract-pass.ll
new file mode 100644
index 0000000000000..e0256da314306
--- /dev/null
+++ b/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
diff --git a/lld/MachO/LTO.cpp b/lld/MachO/LTO.cpp
index 7a9a9223a0322..6527cbb68f249 100644
--- a/lld/MachO/LTO.cpp
+++ b/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();
 
diff --git a/llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h b/llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h
index ccf859922e163..5ac6ad47f7799 100644
--- a/llvm/include/llvm/Analysis/ObjCARCAnalysisUtils.h
+++ b/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
diff --git a/llvm/lib/CodeGen/TargetPassConfig.cpp b/llvm/lib/CodeGen/TargetPassConfig.cpp
index 3658e8320a0cc..d833555d44b25 100644
--- a/llvm/lib/CodeGen/TargetPassConfig.cpp
+++ b/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>
@@ -949,6 +950,9 @@ 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);
diff --git a/llvm/lib/LTO/LTOCodeGenerator.cpp b/llvm/lib/LTO/LTOCodeGenerator.cpp
index 34aacb660144f..a192392e04585 100644
--- a/llvm/lib/LTO/LTOCodeGenerator.cpp
+++ b/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;
 }
diff --git a/llvm/lib/LTO/ThinLTOCodeGenerator.cpp b/llvm/lib/LTO/ThinLTOCodeGenerator.cpp
index 57c94e1988b79..b982df7c6e5d3 100644
--- a/llvm/lib/LTO/ThinLTOCodeGenerator.cpp
+++ b/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))
diff --git a/llvm/test/CodeGen/AArch64/O3-pipeline.ll b/llvm/test/CodeGen/AArch64/O3-pipeline.ll
index c5d604a5a2783..8fe2374103e44 100644
--- a/llvm/test/CodeGen/AArch64/O3-pipeline.ll
+++ b/llvm/test/CodeGen/AArch64/O3-pipeline.ll
@@ -104,6 +104,10 @@
 ; 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
diff --git a/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll b/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
index b61838c06a1f9..4f64a6bb898a4 100644
--- a/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
+++ b/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
@@ -283,6 +283,9 @@
 ; 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
@@ -577,6 +580,9 @@
 ; 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
@@ -884,6 +890,11 @@
 ; 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
@@ -1204,6 +1215,11 @@
 ; 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
diff --git a/llvm/test/CodeGen/ARM/O3-pipeline.ll b/llvm/test/CodeGen/ARM/O3-pipeline.ll
index 461920e1c5da1..18b8c7e3ced5d 100644
--- a/llvm/test/CodeGen/ARM/O3-pipeline.ll
+++ b/llvm/test/CodeGen/ARM/O3-pipeline.ll
@@ -66,6 +66,10 @@
 ; CHECK-NEXT:        Transform predicated vector loops to use MVE tail predication
 ; CHECK-NEXT:      A No-Op Barrier Pass
 ; CHECK-NEXT:      FunctionPass Manager
+; 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
diff --git a/llvm/test/CodeGen/LoongArch/opt-pipeline.ll b/llvm/test/CodeGen/LoongArch/opt-pipeline.ll
index 4e5a5433596db..fd3fb293ae7e6 100644
--- a/llvm/test/CodeGen/LoongArch/opt-pipeline.ll
+++ b/llvm/test/CodeGen/LoongArch/opt-pipeline.ll
@@ -72,10 +72,14 @@
 ; LAXX-NEXT:       CodeGen Prepare
 ; LAXX-NEXT:       Dominator Tree Construction
 ; LAXX-NEXT:       Exception handling preparation
+; LAXX-NEXT:       Basic Alias Analysis (stateless AA impl)
+; LAXX-NEXT:       Function Alias Analysis Results
+; LAXX-NEXT:       ObjC ARC contraction
 ; LAXX-NEXT:       Prepare callbr
 ; LAXX-NEXT:       Safe Stack instrumentation pass
 ; LAXX-NEXT:       Insert stack protectors
 ; LAXX-NEXT:       Module Verifier
+; LAXX-NEXT:       Dominator Tree Construction
 ; LAXX-NEXT:       Basic Alias Analysis (stateless AA impl)
 ; LAXX-NEXT:       Function Alias Analysis Results
 ; LAXX-NEXT:       Natural Loop Information
diff --git a/llvm/test/CodeGen/M68k/pipeline.ll b/llvm/test/CodeGen/M68k/pipeline.ll
index e7d0b20640f7a..1c9184a4d7e71 100644
--- a/llvm/test/CodeGen/M68k/pipeline.ll
+++ b/llvm/test/CodeGen/M68k/pipeline.ll
@@ -42,10 +42,14 @@
 ; CHECK-NEXT:      CodeGen Prepare
 ; CHECK-NEXT:      Dominator Tree Construction
 ; CHECK-NEXT:      Exception handling preparation
+; 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
diff --git a/llvm/test/CodeGen/PowerPC/O3-pipeline.ll b/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
index a564a5517f579..2954d051a2ea9 100644
--- a/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
+++ b/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
@@ -83,10 +83,14 @@
 ; CHECK-NEXT:       Lazy Block Frequency Analysis
 ; CHECK-NEXT:       Optimization Remark Emitter
 ; CHECK-NEXT:       Hardware Loop Insertion
+; 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
diff --git a/llvm/test/CodeGen/RISCV/O3-pipeline.ll b/llvm/test/CodeGen/RISCV/O3-pipeline.ll
index e305a74d7aef3..af5151b6f938c 100644
--- a/llvm/test/CodeGen/RISCV/O3-pipeline.ll
+++ b/llvm/test/CodeGen/RISCV/O3-pipeline.ll
@@ -75,6 +75,10 @@
 ; CHECK-NEXT:       Exception handling preparation
 ; CHECK-NEXT:     A No-Op Barrier Pass
 ; CHECK-NEXT:     FunctionPass Manager
+; 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
diff --git a/llvm/test/CodeGen/X86/opt-pipeline.ll b/llvm/test/CodeGen/X86/opt-pipeline.ll
index 19774b7205109..d9d6170585035 100644
--- a/llvm/test/CodeGen/X86/opt-pipeline.ll
+++ b/llvm/test/CodeGen/X86/opt-pipeline.ll
@@ -73,10 +73,14 @@
 ; CHECK-NEXT:       CodeGen Prepare
 ; CHECK-NEXT:       Dominator Tree Construction
 ; CHECK-NEXT:       Exception handling preparation
+; 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

NuriAmari · 2024-07-30T03:37:27Z

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


Since we're not calling this anymore, I don't think we need this change.

NuriAmari · 2024-07-30T03:47:08Z

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


@cachemeifyoucan As far as I can tell, it's already the case that for non-LTO builds, this pass is only added for optimization level > 0. Is that ok to do for all builds?

I checked and directly providing clang with IR that contains llvm.objc.clang.arc.use and compiling it with -O0 fails instruction selection as a result of not running this pass, so I'm assuming these intrinsics are only introduced at -O1 or higher.

@ahatanak probably know better. I think when the arc intrinsics are present, the pass is mandatory but I am not sure if the intrinsics are only produced for -O1 and higher.

Don't run this optimization on O0 has been there since LLVM 3.1 (1170b08). The committer @sunfishcode probably did it for a reason

The intrinsic calls are needed to prevent ARC optimizer from moving releases to an earlier point that can cause objects to be deallocated prematurely. clang doesn't emit the intrinsic calls if OptimizationLevel is zero.

efriedma-quic · 2024-07-30T18:14:57Z

Is there some way we can avoid the additional domtree construction? Particularly in cases where the pass doesn't do anything.

DataCorrupted · 2024-07-30T20:47:05Z

Is there some way we can avoid the additional domtree construction? Particularly in cases where the pass doesn't do anything.

This patch only changes how the pass is called. The behavior of the pass itself is probably another patch. Besides, I don't think it would be easy for PassConfig to get a hold of the module that we are compiling.

NuriAmari · 2024-07-30T21:39:27Z

This pass is added in numerous places: https://github.com/search?q=repo%3Allvm%2Fllvm-project%20createObjCARCContractPass&type=code, and we were hoping to consolidate all the places and instead schedule it by default, or behind some kind of predicate. We haven’t been able to find a predicate that satisfies all the use cases, there always seems to be a case where the required analysis passes are scheduled when not necessarily always required.

In practice, I think this pass is already scheduled by default, or for optimized builds in almost all configurations. Auditing the places it’s added, the legacy LTO backend adds it unconditionally:

llvm-project/llvm/lib/LTO/ThinLTOCodeGenerator.cpp

Line 340 in 8364a6e

PM.add(createObjCARCContractPass());

llvm-project/llvm/lib/LTO/LTOCodeGenerator.cpp

Line 140 in f70f122

PM.add(createObjCARCContractPass());

LLD for MachO injects it into the new LTO backend via a hook:

llvm-project/lld/MachO/LTO.cpp

Line 52 in f70f122

pm.add(createObjCARCContractPass());

Clang adds it for non-LTO compiles here:

llvm-project/clang/lib/CodeGen/BackendUtil.cpp

Line 595 in f70f122

CodeGenPasses.add(createObjCARCContractPass());

Granted that leaves LTO builds that are using the new LTO backend (except LLD for MachO), and distributed thin-LTO builds which currently don’t schedule it. Ultimately, our goal here is to schedule the pass for distributed thin-LTO compiles as the new test demonstrates.

We could go back to a change like: 66ddf60

That would schedule the pass unconditionally for distributed thin-LTO compiles, on top of those configurations it already runs for. The analysis passes would still be run in some cases that are not required, but I don’t see how to avoid that. If we prefer that change, over the current, we could go that direction instead.

nikic · 2024-07-31T13:49:08Z

The pipeline test changes here still look problematic. Can you make the ObjCARCContract pass preserve the DT?

DataCorrupted · 2024-08-05T18:52:42Z

The pipeline test changes here still look problematic. Can you make the ObjCARCContract pass preserve the DT?

ObjCARCContractPass can't preserve the DT at static time, it may alter CFG (See more in this patch). Unfortunately, legacy PM won't let us preserve the DT at runtime.

Current situation:

ObjCARCContractPass have no overhead on O3 builds.
On distributed thin-lto builds, the overhead is very small (on average <0.6%)
What's more, we can’t build Darwin applications using distributed thin-lto without this change. We can gain a huge build speed improvement with distributed thin-lto.

Do you think the compile time regressions are acceptable if this change is restricted to optimized builds, or only enabled for distributed thin-lto builds?

nikic · 2024-08-05T19:07:04Z

The pipeline test changes here still look problematic. Can you make the ObjCARCContract pass preserve the DT?

ObjCARCContractPass can't preserve the DT at static time, it may alter CFG (See more in this patch). Unfortunately, legacy PM won't let us preserve the DT at runtime.

You don't need to dynamically preserve the DT, you should mark it as unconditionally preserved in AnalysisUsage. From a quick glance at the pass, the only CFG change it does is critical edge splitting, and DT is already passed while doing that, so it will be updated:

llvm-project/llvm/lib/Transforms/ObjCARC/ObjCARC.cpp

Line 61 in f1fd9d7

DestBB = SplitCriticalEdge(I, 0, CriticalEdgeSplittingOptions(DT));

You're currently just not telling the pass managers that this is the case.

DataCorrupted · 2024-08-06T17:23:34Z

Interesting, I didn't know DT is preserved during critical edge splitting. Just add the line to preserve DT. Many DT Construction is removed from the tests.

nikic · 2024-08-06T18:57:12Z

llvm/lib/Transforms/ObjCARC/ObjCARCContract.cpp

@@ -730,6 +740,7 @@ INITIALIZE_PASS_END(ObjCARCContractLegacyPass, "objc-arc-contract",
 void ObjCARCContractLegacyPass::getAnalysisUsage(AnalysisUsage &AU) const {
  AU.addRequired<AAResultsWrapperPass>();
  AU.addRequired<DominatorTreeWrapperPass>();
+  AU.addPreserved<DominatorTreeWrapperPass>();


Can you also add AAResultsWrapper and BasicAAWrapperPass here? I think that should avoid some more pipeline diffs.

This reverts commit 1579e9c.

Summary: \llvm#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](llvm@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 <PeterRong@meta.com>

Signed-off-by: Peter Rong <PeterRong@meta.com>

Tests have been changed to show that we don't have as much DT construction as before.

Signed-off-by: Peter Rong <PeterRong@meta.com>

nikic

LGTM

* 'main' of https://github.com/llvm/llvm-project: (700 commits) [SandboxIR][NFC] SingleLLVMInstructionImpl class (llvm#102687) [ThinLTO]Clean up 'import-assume-unique-local' flag. (llvm#102424) [nsan] Make #include more conventional [SandboxIR][NFC] Use Tracker.emplaceIfTracking() [libc] Moved range_reduction_double ifdef statement (llvm#102659) [libc] Fix CFP long double and add tests (llvm#102660) [TargetLowering] Handle vector types in expandFixedPointMul (llvm#102635) [compiler-rt][NFC] Replace environment variable with %t (llvm#102197) [UnitTests] Convert a test to use opaque pointers (llvm#102668) [CodeGen][NFCI] Don't re-implement parts of ASTContext::getIntWidth (llvm#101765) [SandboxIR] Clean up tracking code with the help of emplaceIfTracking() (llvm#102406) [mlir][bazel] remove extra blanks in mlir-tblgen test [NVPTX][NFC] Update tests to use bfloat type (llvm#101493) [mlir] Add support for parsing nested PassPipelineOptions (llvm#101118) [mlir][bazel] add missing td dependency in mlir-tblgen test [flang][cuda] Fix lib dependency [libc] Clean up remaining use of *_WIDTH macros in printf (llvm#102679) [flang][cuda] Convert cuf.alloc for box to fir.alloca in device context (llvm#102662) [SandboxIR] Implement the InsertElementInst class (llvm#102404) [libc] Fix use of cpp::numeric_limits<...>::digits (llvm#102674) [mlir][ODS] Verify type constraints in Types and Attributes (llvm#102326) [LTO] enable `ObjCARCContractPass` only on optimized build (llvm#101114) [mlir][ODS] Consistent `cppType` / `cppClassName` usage (llvm#102657) [lldb] Move definition of SBSaveCoreOptions dtor out of header (llvm#102539) [libc] Use cpp::numeric_limits in preference to C23 <limits.h> macros (llvm#102665) [clang] Implement -fptrauth-auth-traps. (llvm#102417) [LLVM][rtsan] rtsan transform to preserve CFGAnalyses (llvm#102651) Revert "[AMDGPU] Move `AMDGPUAttributorPass` to full LTO post link stage (llvm#102086)" [RISCV][GISel] Add missing tests for G_CTLZ/CTTZ instruction selection. NFC Return available function types for BindingDecls. (llvm#102196) [clang] Wire -fptrauth-returns to "ptrauth-returns" fn attribute. (llvm#102416) [RISCV] Remove riscv-experimental-rv64-legal-i32. (llvm#102509) [RISCV] Move PseudoVSET(I)VLI expansion to use PseudoInstExpansion. (llvm#102496) [NVPTX] support switch statement with brx.idx (reland) (llvm#102550) [libc][newhdrgen]sorted function names in yaml (llvm#102544) [GlobalIsel] Combine G_ADD and G_SUB with constants (llvm#97771) Suppress spurious warnings due to R_RISCV_SET_ULEB128 [scudo] Separated committed and decommitted entries. (llvm#101409) [MIPS] Fix missing ANDI optimization (llvm#97689) [Clang] Add env var for nvptx-arch/amdgpu-arch timeout (llvm#102521) [asan] Switch allocator to dynamic base address (llvm#98511) [AMDGPU] Move `AMDGPUAttributorPass` to full LTO post link stage (llvm#102086) [libc][math][c23] Add fadd{l,f128} C23 math functions (llvm#102531) [mlir][bazel] revert bazel rule change for DLTITransformOps [msan] Support vst{2,3,4}_lane instructions (llvm#101215) Revert "[MLIR][DLTI][Transform] Introduce transform.dlti.query (llvm#101561)" [X86] pr57673.ll - generate MIR test checks [mlir][vector][test] Split tests from vector-transfer-flatten.mlir (llvm#102584) [mlir][bazel] add bazel rule for DLTITransformOps OpenMPOpt: Remove dead include [IR] Add method to GlobalVariable to change type of initializer. (llvm#102553) [flang][cuda] Force default allocator in device code (llvm#102238) [llvm] Construct SmallVector<SDValue> with ArrayRef (NFC) (llvm#102578) [MLIR][DLTI][Transform] Introduce transform.dlti.query (llvm#101561) [AMDGPU][AsmParser][NFC] Remove a misleading comment. (llvm#102604) [Arm][AArch64][Clang] Respect function's branch protection attributes. (llvm#101978) [mlir] Verifier: steal bit to track seen instead of set. (llvm#102626) [Clang] Fix Handling of Init Capture with Parameter Packs in LambdaScopeForCallOperatorInstantiationRAII (llvm#100766) [X86] Convert truncsat clamping patterns to use SDPatternMatch. NFC. [gn] Give two scripts argparse.RawDescriptionHelpFormatter [bazel] Add missing dep for the SPIRVToLLVM target [Clang] Simplify specifying passes via -Xoffload-linker (llvm#102483) [bazel] Port for d45de80 [SelectionDAG] Use unaligned store/load to move AVX registers onto stack for `insertelement` (llvm#82130) [Clang][OMPX] Add the code generation for multi-dim `num_teams` (llvm#101407) [ARM] Regenerate big-endian-vmov.ll. NFC [AMDGPU][AsmParser][NFCI] All NamedIntOperands to be of the i32 type. (llvm#102616) [libc][math][c23] Add totalorderl function. (llvm#102564) [mlir][spirv] Support `memref` in `convert-to-spirv` pass (llvm#102534) [MLIR][GPU-LLVM] Convert `gpu.func` to `llvm.func` (llvm#101664) Fix a unit test input file (llvm#102567) [llvm-readobj][COFF] Dump hybrid objects for ARM64X files. (llvm#102245) AMDGPU/NewPM: Port SIFixSGPRCopies to new pass manager (llvm#102614) [MemoryBuiltins] Simplify getCalledFunction() helper (NFC) [AArch64] Add invalid 1 x vscale costs for reductions and reduction-operations. (llvm#102105) [MemoryBuiltins] Handle allocator attributes on call-site LSV/test/AArch64: add missing lit.local.cfg; fix build (llvm#102607) Revert "Enable logf128 constant folding for hosts with 128bit floats (llvm#96287)" [RISCV] Add Syntacore SCR5 RV32/64 processors definition (llvm#102285) [InstCombine] Remove unnecessary RUN line from test (NFC) [flang][OpenMP] Handle multiple ranges in `num_teams` clause (llvm#102535) [mlir][vector] Add tests for scalable vectors in one-shot-bufferize.mlir (llvm#102361) [mlir][vector] Disable `vector.matrix_multiply` for scalable vectors (llvm#102573) [clang] Implement CWG2627 Bit-fields and narrowing conversions (llvm#78112) [NFC] Use references to avoid copying (llvm#99863) Revert "[mlir][ArmSME] Pattern to swap shape_cast(tranpose) with transpose(shape_cast) (llvm#100731)" (llvm#102457) [IRBuilder] Generate nuw GEPs for struct member accesses (llvm#99538) [bazel] Port for 9b06e25 [CodeGen][NewPM] Improve start/stop pass error message CodeGenPassBuilder (llvm#102591) [AArch64] Implement TRBMPAM_EL1 system register (llvm#102485) [InstCombine] Fixing wrong select folding in vectors with undef elements (llvm#102244) [AArch64] Sink operands to fmuladd. (llvm#102297) LSV: document hang reported in llvm#37865 (llvm#102479) Enable logf128 constant folding for hosts with 128bit floats (llvm#96287) [RISCV][clang] Remove bfloat base type in non-zvfbfmin vcreate (llvm#102146) [RISCV][clang] Add missing `zvfbfmin` to `vget_v` intrinsic (llvm#102149) [mlir][vector] Add mask elimination transform (llvm#99314) [Clang][Interp] Fix display of syntactically-invalid note for member function calls (llvm#102170) [bazel] Port for 3fffa6d [DebugInfo][RemoveDIs] Use iterator-inserters in clang (llvm#102006) ... Signed-off-by: Edwiin Kusuma Jaya <kutemeikito0905@gmail.com>

mikaelholmen · 2024-08-13T12:04:23Z

I noticed that if I compile opt with UBSan enabled on this patch and then run

opt -passes=objc-arc-contract bbi-98007.ll -o /dev/null

we get an UBSan complaint like

../lib/Transforms/ObjCARC/ObjCARCContract.cpp:105:39: runtime error: load of value 255, which is not a valid value for type 'bool'
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior ../lib/Transforms/ObjCARC/ObjCARCContract.cpp:105:39 in

And the bbi-98007.ll input is just

define void @a() {
entry:
  ret void
}

DataCorrupted · 2024-08-14T23:46:41Z

@mikaelholmen I can't reproduce this on commit e40915b with -DLLVM_USE_SANITIZER=Undefined. Kindly let me know how to reproduce this.

Looking from the code, I suspect that it's caused by uninitialized bool (

llvm-project/llvm/lib/Transforms/ObjCARC/ObjCARCContract.cpp

Line 553 in 539bf49

Changed = CFGChanged = false;

)

mikaelholmen · 2024-08-15T11:11:39Z

@mikaelholmen I can't reproduce this on commit e40915b with -DLLVM_USE_SANITIZER=Undefined. Kindly let me know how to reproduce this.

Looking from the code, I suspect that it's caused by uninitialized bool (

llvm-project/llvm/lib/Transforms/ObjCARC/ObjCARCContract.cpp

Line 553 in 539bf49

Changed = CFGChanged = false;

)

I think the problem is that the

Changed = CFGChanged = false;

initialization done in

bool ObjCARCContract::run(Function &F, AAResults *A, DominatorTree *D) {
  if (!Run)
    return false;

  if (!EnableARCOpts)
    return false;

  Changed = CFGChanged = false;

is skipped if we bail out early due to "!Run" or "!EnableARCOpts".
But the pass always asks

  bool CFGChanged = OCAC.hasCFGChanged();

afterwards anyway.

DataCorrupted requested review from nikic, heiher and NuriAmari July 30, 2024 03:20

DataCorrupted self-assigned this Jul 30, 2024

llvmbot added clang Clang issues not falling into any other category lld backend:ARM backend:AArch64 backend:AMDGPU backend:PowerPC backend:X86 clang:codegen lld:MachO backend:m68k LTO Link time optimization (regular/full LTO or ThinLTO) backend:loongarch llvm:analysis labels Jul 30, 2024

NuriAmari reviewed Jul 30, 2024

View reviewed changes

NuriAmari requested review from fhahn and cachemeifyoucan July 30, 2024 03:39

NuriAmari reviewed Jul 30, 2024

View reviewed changes

cachemeifyoucan requested review from gottesmm and ahatanak July 30, 2024 16:31

llvmbot added the llvm:transforms label Aug 5, 2024

DataCorrupted force-pushed the DefaultARC branch from 2ed3f66 to 88ef973 Compare August 6, 2024 17:36

nikic reviewed Aug 6, 2024

View reviewed changes

DataCorrupted added 6 commits August 6, 2024 12:31

Reapply "Run ObjCContractPass in Default Codegen Pipeline (llvm#92331)"

ebbba2b

This reverts commit 1579e9c.

Revert unnecessary format changes in ObjCARCAnalysisUtils.h

4fc7867

Signed-off-by: Peter Rong <PeterRong@meta.com>

Add ModuleHasARC check back to skip the pass on non objc modules.

a45b3b9

Signed-off-by: Peter Rong <PeterRong@meta.com>

Indicate that DT is preserved in ObjCARCContractPass

34e6238

Tests have been changed to show that we don't have as much DT construction as before.

preserving AA analysis as well

c9c15bb

Signed-off-by: Peter Rong <PeterRong@meta.com>

DataCorrupted force-pushed the DefaultARC branch from 88ef973 to c9c15bb Compare August 6, 2024 20:06

nikic approved these changes Aug 7, 2024

View reviewed changes

DataCorrupted merged commit 74e4694 into llvm:main Aug 9, 2024
7 checks passed

jayfoad mentioned this pull request Aug 13, 2024

[LTO] Run ObjCARCContractPass according to the callgraph #103034

Merged

DataCorrupted deleted the DefaultARC branch February 6, 2025 17:44

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[LTO] enable `ObjCARCContractPass` only on optimized build #101114

[LTO] enable `ObjCARCContractPass` only on optimized build #101114

DataCorrupted commented Jul 30, 2024

llvmbot commented Jul 30, 2024 •

edited

Loading

llvmbot commented Jul 30, 2024

llvmbot commented Jul 30, 2024

NuriAmari Jul 30, 2024

NuriAmari Jul 30, 2024

cachemeifyoucan Jul 30, 2024

DataCorrupted Jul 30, 2024 •

edited

Loading

ahatanak Jul 31, 2024

efriedma-quic commented Jul 30, 2024

DataCorrupted commented Jul 30, 2024

NuriAmari commented Jul 30, 2024

nikic commented Jul 31, 2024

DataCorrupted commented Aug 5, 2024 •

edited

Loading

nikic commented Aug 5, 2024

DataCorrupted commented Aug 6, 2024

nikic Aug 6, 2024

nikic left a comment

mikaelholmen commented Aug 13, 2024

DataCorrupted commented Aug 14, 2024

mikaelholmen commented Aug 15, 2024

[LTO] enable ObjCARCContractPass only on optimized build #101114

[LTO] enable ObjCARCContractPass only on optimized build #101114

Conversation

DataCorrupted commented Jul 30, 2024

llvmbot commented Jul 30, 2024 • edited Loading

llvmbot commented Jul 30, 2024

llvmbot commented Jul 30, 2024

NuriAmari Jul 30, 2024

Choose a reason for hiding this comment

NuriAmari Jul 30, 2024

Choose a reason for hiding this comment

cachemeifyoucan Jul 30, 2024

Choose a reason for hiding this comment

DataCorrupted Jul 30, 2024 • edited Loading

Choose a reason for hiding this comment

ahatanak Jul 31, 2024

Choose a reason for hiding this comment

efriedma-quic commented Jul 30, 2024

DataCorrupted commented Jul 30, 2024

NuriAmari commented Jul 30, 2024

nikic commented Jul 31, 2024

DataCorrupted commented Aug 5, 2024 • edited Loading

nikic commented Aug 5, 2024

DataCorrupted commented Aug 6, 2024

nikic Aug 6, 2024

Choose a reason for hiding this comment

nikic left a comment

Choose a reason for hiding this comment

mikaelholmen commented Aug 13, 2024

DataCorrupted commented Aug 14, 2024

mikaelholmen commented Aug 15, 2024

[LTO] enable `ObjCARCContractPass` only on optimized build #101114

[LTO] enable `ObjCARCContractPass` only on optimized build #101114

llvmbot commented Jul 30, 2024 •

edited

Loading

DataCorrupted Jul 30, 2024 •

edited

Loading

DataCorrupted commented Aug 5, 2024 •

edited

Loading