Implement strip.invariant.group

[Prazek]

Summary

This patch introduce new intrinsic -
strip.invariant.group that was described in the
RFC: Devirtualization v2

[Prazek]

• Add deleted tests

[Prazek]

• small fixes in tests

[Prazek]

Reformat

[Prazek]

introduced mayBeDynamicClass and added more tests

[xbolva00]

clang/include/clang/AST/DeclCXX.h

778 ↗

(On Diff #147683)

maybeDynamicClass?

https://github.com/llvm-mirror/llvm/search?utf8=%E2%9C%93&q=maybe&type=

[Prazek]

Changed comments

[Prazek]

After rebasing

[Prazek]

rebase

[amharc]

clang/include/clang/AST/DeclCXX.h

778 ↗

(On Diff #147683)

I think mayBe is more appropriate, because it grammatically consists of 'may' and 'be'. In this form it appears in multiple places in the codebase, too. A few examples:

https://github.com/llvm-mirror/llvm/blob/0b24b74655b976aaba01b2f726659b8c745a16de/include/llvm/IR/GlobalValue.h#L131
https://github.com/llvm-mirror/clang/blob/647be32c6048e24f90f470c557e850597e5526c3/lib/AST/DeclCXX.cpp#L1800
https://github.com/llvm-mirror/llvm/blob/77438c3c98c0901fa9ef56f78bdefccedc16fcc4/lib/Analysis/ValueTracking.cpp#L3659

[rjmccall]

The changes to Clang generally seem reasonable, but I think you should split them into a separate commit from the commit that adds the intrinsic itself.

clang/lib/CodeGen/CGExpr.cpp

3858 ↗

(On Diff #147691)

Please add a comment explaining why this is necessary. (I'm actually not sure why it is, because surely the invariant groups we generate don't contain assumptions about memory from fields, right?)

clang/lib/CodeGen/CGExprScalar.cpp

1626 ↗

(On Diff #147691)

Again, comment.

1767 ↗

(On Diff #147691)

I think assigning to IntToPtr would be clearer here. You're changing the value returned, not introducing a reason to early-exit.

And a comment, please, on all of these cases.

[Prazek]

Slitted commit into defining and using intrinsic

clang/lib/CodeGen/CGExpr.cpp

3858 ↗

(On Diff #147691)

Short answer: you can only make virtual calls on a dynamic pointer that carries invariant.group and you can't do anything other because it could leak the information about this pointer (which when used with comparison could break devirtualization).

[rjmccall]

clang/lib/CodeGen/CGExpr.cpp

3858 ↗

(On Diff #147691)

Alright, sure.

[Prazek]

friendly ping

[Prazek]

Friendly ping @hfinkel @rsmith

[rsmith]

llvm/docs/LangRef.rst

12928 ↗

(On Diff #148313)

does carries -> carries

llvm/trunk/docs/LangRef.rst

 ::
 
       declare i8* @llvm.launder.invariant.group.p0i8(i8* <ptr>)
 
 Overview:
 """""""""
 
 The '``llvm.launder.invariant.group``' intrinsic can be used when an invariant
-established by invariant.group metadata no longer holds, to obtain a new pointer
-value that does not carry the invariant information. It is an experimental
-intrinsic, which means that its semantics might change in the future.
+established by ``invariant.group`` metadata no longer holds, to obtain a new
+pointer value that carries fresh invariant group information. It is an
+experimental intrinsic, which means that its semantics might change in the
+future.
 
 
 Arguments:
 """"""""""
 
-The ``llvm.launder.invariant.group`` takes only one argument, which is
-the pointer to the memory for which the ``invariant.group`` no longer holds.
+The ``llvm.launder.invariant.group`` takes only one argument, which is a pointer
+to the memory.
 
 Semantics:
 """"""""""
 
 Returns another pointer that aliases its argument but which is considered different
 for the purposes of ``load``/``store`` ``invariant.group`` metadata.
 It does not read any accessible memory and the execution can be speculated.
 
+'``llvm.strip.invariant.group``' Intrinsic
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Syntax:
+"""""""
+This is an overloaded intrinsic. The memory object can belong to any address
+space. The returned pointer must belong to the same address space as the
+argument.
+
+::
+
+      declare i8* @llvm.strip.invariant.group.p0i8(i8* <ptr>)
+
+Overview:
+"""""""""
+
+The '``llvm.strip.invariant.group``' intrinsic can be used when an invariant
+established by ``invariant.group`` metadata no longer holds, to obtain a new pointer
+value that does not carry the invariant information. It is an experimental
+intrinsic, which means that its semantics might change in the future.
+
+
+Arguments:
+""""""""""
+
+The ``llvm.strip.invariant.group`` takes only one argument, which is a pointer
+to the memory.
+
+Semantics:
+""""""""""
+
+Returns another pointer that aliases its argument but which has no associated
+``invariant.group`` metadata.
+It does not read any memory and can be speculated.
+
+
+
 .. _constrainedfp:
 
 Constrained Floating-Point Intrinsics
 -------------------------------------
 
 These intrinsics are used to provide special handling of floating-point
 operations when specific rounding mode or floating-point exception behavior is
 required.  By default, LLVM optimization passes assume that the rounding mode is

llvm/trunk/include/llvm/IR/IRBuilder.h

   }
 
   /// Create a launder.invariant.group intrinsic call. If Ptr type is
   /// different from pointer to i8, it's casted to pointer to i8 in the same
   /// address space before call and casted back to Ptr type after call.
   Value *CreateLaunderInvariantGroup(Value *Ptr) {
     assert(isa<PointerType>(Ptr->getType()) &&
            "launder.invariant.group only applies to pointers.");
+    // FIXME: we could potentially avoid casts to/from i8*.
     auto *PtrType = Ptr->getType();
     auto *Int8PtrTy = getInt8PtrTy(PtrType->getPointerAddressSpace());
     if (PtrType != Int8PtrTy)
       Ptr = CreateBitCast(Ptr, Int8PtrTy);
     Module *M = BB->getParent()->getParent();
     Function *FnLaunderInvariantGroup = Intrinsic::getDeclaration(
         M, Intrinsic::launder_invariant_group, {Int8PtrTy});
 
     assert(FnLaunderInvariantGroup->getReturnType() == Int8PtrTy &&
            FnLaunderInvariantGroup->getFunctionType()->getParamType(0) ==
                Int8PtrTy &&
            "LaunderInvariantGroup should take and return the same type");
 
     CallInst *Fn = CreateCall(FnLaunderInvariantGroup, {Ptr});
 
     if (PtrType != Int8PtrTy)
       return CreateBitCast(Fn, PtrType);
     return Fn;
   }
 
+  /// \brief Create a strip.invariant.group intrinsic call. If Ptr type is
+  /// different from pointer to i8, it's casted to pointer to i8 in the same
+  /// address space before call and casted back to Ptr type after call.
+  Value *CreateStripInvariantGroup(Value *Ptr) {
+    assert(isa<PointerType>(Ptr->getType()) &&
+           "strip.invariant.group only applies to pointers.");
+
+    // FIXME: we could potentially avoid casts to/from i8*.
+    auto *PtrType = Ptr->getType();
+    auto *Int8PtrTy = getInt8PtrTy(PtrType->getPointerAddressSpace());
+    if (PtrType != Int8PtrTy)
+      Ptr = CreateBitCast(Ptr, Int8PtrTy);
+    Module *M = BB->getParent()->getParent();
+    Function *FnStripInvariantGroup = Intrinsic::getDeclaration(
+        M, Intrinsic::strip_invariant_group, {Int8PtrTy});
+
+    assert(FnStripInvariantGroup->getReturnType() == Int8PtrTy &&
+           FnStripInvariantGroup->getFunctionType()->getParamType(0) ==
+               Int8PtrTy &&
+           "StripInvariantGroup should take and return the same type");
+
+    CallInst *Fn = CreateCall(FnStripInvariantGroup, {Ptr});
+
+    if (PtrType != Int8PtrTy)
+      return CreateBitCast(Fn, PtrType);
+    return Fn;
+  }
+
   /// Return a vector value that contains \arg V broadcasted to \p
   /// NumElts elements.
   Value *CreateVectorSplat(unsigned NumElts, Value *V, const Twine &Name = "") {
     assert(NumElts > 0 && "Cannot splat to an empty vector!");
 
     // First insert it into an undef vector so we can shuffle it.
     Type *I32Ty = getInt32Ty();
     Value *Undef = UndefValue::get(VectorType::get(V->getType(), NumElts));

llvm/trunk/include/llvm/IR/Intrinsics.td

 // The argument also can't be marked with 'returned' attribute, because
 // it would remove barrier.
 // Note that it is still experimental, which means that its semantics
 // might change in the future.
 def int_launder_invariant_group : Intrinsic<[llvm_anyptr_ty],
                                             [LLVMMatchType<0>],
                                             [IntrInaccessibleMemOnly, IntrSpeculatable]>;
 
+
+def int_strip_invariant_group : Intrinsic<[llvm_anyptr_ty],
+                                          [LLVMMatchType<0>],
+                                          [IntrSpeculatable, IntrNoMem]>;
+
 //===------------------------ Stackmap Intrinsics -------------------------===//
 //
 def int_experimental_stackmap : Intrinsic<[],
                                   [llvm_i64_ty, llvm_i32_ty, llvm_vararg_ty],
                                   [Throws]>;
 def int_experimental_patchpoint_void : Intrinsic<[],
                                                  [llvm_i64_ty, llvm_i32_ty,
                                                   llvm_ptr_ty, llvm_i32_ty,

llvm/trunk/lib/Analysis/BasicAliasAnalysis.cpp

if (Op->getOpcode() == Instruction::BitCast ||
         Op->getOpcode() == Instruction::AddrSpaceCast) {
       V = Op->getOperand(0);
       continue;
     }
 
     const GEPOperator *GEPOp = dyn_cast<GEPOperator>(Op);
     if (!GEPOp) {
       if (auto CS = ImmutableCallSite(V)) {
-        // Note: getArgumentAliasingToReturnedPointer keeps it in sync with
-        // CaptureTracking, which is needed for correctness.  This is because
-        // some intrinsics like launder.invariant.group returns pointers that
-        // are aliasing it's argument, which is known to CaptureTracking.
-        // If AliasAnalysis does not use the same information, it could assume
-        // that pointer returned from launder does not alias it's argument
-        // because launder could not return it if the pointer was not captured.
+        // CaptureTracking can know about special capturing properties of some
+        // intrinsics like launder.invariant.group, that can't be expressed with
+        // the attributes, but have properties like returning aliasing pointer.
+        // Because some analysis may assume that nocaptured pointer is not
+        // returned from some special intrinsic (because function would have to
+        // be marked with returns attribute), it is crucial to use this function
+        // because it should be in sync with CaptureTracking. Not using it may
+        // cause weird miscompilations where 2 aliasing pointers are assumed to
+        // noalias.
         if (auto *RP = getArgumentAliasingToReturnedPointer(CS)) {
           V = RP;
           continue;
         }
       }
 
       // If it's not a GEP, hand it off to SimplifyInstruction to see if it
       // can come up with something. This matches what GetUnderlyingObject does.

llvm/trunk/lib/Analysis/ConstantFolding.cpp

bool llvm::canConstantFoldCallTo(ImmutableCallSite CS, const Function *F) {
   case Intrinsic::bswap:
   case Intrinsic::ctpop:
   case Intrinsic::ctlz:
   case Intrinsic::cttz:
   case Intrinsic::fma:
   case Intrinsic::fmuladd:
   case Intrinsic::copysign:
   case Intrinsic::launder_invariant_group:
+  case Intrinsic::strip_invariant_group:
   case Intrinsic::round:
   case Intrinsic::masked_load:
   case Intrinsic::sadd_with_overflow:
   case Intrinsic::uadd_with_overflow:
   case Intrinsic::ssub_with_overflow:
   case Intrinsic::usub_with_overflow:
   case Intrinsic::smul_with_overflow:
   case Intrinsic::umul_with_overflow:
Constant *ConstantFoldScalarCall(StringRef Name, unsigned IntrinsicID, Type *Ty,
                                  const TargetLibraryInfo *TLI) {
   if (Operands.size() == 1) {
     if (isa<UndefValue>(Operands[0])) {
       // cosine(arg) is between -1 and 1. cosine(invalid arg) is NaN
       if (IntrinsicID == Intrinsic::cos)
         return Constant::getNullValue(Ty);
       if (IntrinsicID == Intrinsic::bswap ||
           IntrinsicID == Intrinsic::bitreverse ||
-          IntrinsicID == Intrinsic::launder_invariant_group)
+          IntrinsicID == Intrinsic::launder_invariant_group ||
+          IntrinsicID == Intrinsic::strip_invariant_group)
         return Operands[0];
     }
 
     if (isa<ConstantPointerNull>(Operands[0]) &&
         Operands[0]->getType()->getPointerAddressSpace() == 0) {
-      // launder(null) == null iff in addrspace 0
-      if (IntrinsicID == Intrinsic::launder_invariant_group)
+      // launder(null) == null == strip(null) iff in addrspace 0
+      if (IntrinsicID == Intrinsic::launder_invariant_group ||
+          IntrinsicID == Intrinsic::strip_invariant_group)
         return Operands[0];
       return nullptr;
     }
 
     if (auto *Op = dyn_cast<ConstantFP>(Operands[0])) {
       if (IntrinsicID == Intrinsic::convert_to_fp16) {
         APFloat Val(Op->getValueAPF());
 

llvm/trunk/lib/Analysis/ValueTracking.cpp

if (const Value *RV = CS.getReturnedArgOperand())
     return RV;
   // This can be used only as a aliasing property.
   if (isIntrinsicReturningPointerAliasingArgumentWithoutCapturing(CS))
     return CS.getArgOperand(0);
   return nullptr;
 }
 
 bool llvm::isIntrinsicReturningPointerAliasingArgumentWithoutCapturing(
-      ImmutableCallSite CS) {
-  return CS.getIntrinsicID() == Intrinsic::launder_invariant_group;
+  return CS.getIntrinsicID() == Intrinsic::launder_invariant_group ||
+         CS.getIntrinsicID() == Intrinsic::strip_invariant_group;
 }
 
 /// \p PN defines a loop-variant pointer to an object.  Check if the
 /// previous iteration of the loop was referring to the same object as \p PN.
 static bool isSameUnderlyingObjectInLoop(const PHINode *PN,
                                          const LoopInfo *LI) {
   // Find the loop-defined value.
   Loop *L = LI->getLoopFor(PN->getParent());
if (GEPOperator *GEP = dyn_cast<GEPOperator>(V)) {
       if (GA->isInterposable())
         return V;
       V = GA->getAliasee();
     } else if (isa<AllocaInst>(V)) {
       // An alloca can't be further simplified.
       return V;
     } else {
       if (auto CS = CallSite(V)) {
-        // Note: getArgumentAliasingToReturnedPointer keeps it in sync with
-        // CaptureTracking, which is needed for correctness.  This is because
-        // some intrinsics like launder.invariant.group returns pointers that
-        // are aliasing it's argument, which is known to CaptureTracking.
-        // If AliasAnalysis does not use the same information, it could assume
-        // that pointer returned from launder does not alias it's argument
-        // because launder could not return it if the pointer was not captured.
+        // CaptureTracking can know about special capturing properties of some
+        // intrinsics like launder.invariant.group, that can't be expressed with
+        // the attributes, but have properties like returning aliasing pointer.
+        // Because some analysis may assume that nocaptured pointer is not
+        // returned from some special intrinsic (because function would have to
+        // be marked with returns attribute), it is crucial to use this function
+        // because it should be in sync with CaptureTracking. Not using it may
+        // cause weird miscompilations where 2 aliasing pointers are assumed to
+        // noalias.
         if (auto *RP = getArgumentAliasingToReturnedPointer(CS)) {
           V = RP;
           continue;
         }
       }
 
       // See if InstructionSimplify knows any relevant tricks.
       if (Instruction *I = dyn_cast<Instruction>(V))

llvm/trunk/lib/CodeGen/CodeGenPrepare.cpp

case Intrinsic::aarch64_stxr: {
       // Sink a zext feeding stlxr/stxr before it, so it can be folded into it.
       ExtVal->moveBefore(CI);
       // Mark this instruction as "inserted by CGP", so that other
       // optimizations don't touch it.
       InsertedInsts.insert(ExtVal);
       return true;
     }
     case Intrinsic::launder_invariant_group:
+    case Intrinsic::strip_invariant_group:
       II->replaceAllUsesWith(II->getArgOperand(0));
       II->eraseFromParent();
       return true;
 
     case Intrinsic::cttz:
     case Intrinsic::ctlz:
       // If counting zeros is expensive, try to avoid it.
       return despeculateCountZeros(II, TLI, DL, ModifiedDT);

llvm/trunk/lib/CodeGen/SelectionDAG/FastISel.cpp

case Intrinsic::objectsize: {
     Constant *ResCI = ConstantInt::get(II->getType(), Res);
     unsigned ResultReg = getRegForValue(ResCI);
     if (!ResultReg)
       return false;
     updateValueMap(II, ResultReg);
     return true;
   }
   case Intrinsic::launder_invariant_group:
+  case Intrinsic::strip_invariant_group:
   case Intrinsic::expect: {
     unsigned ResultReg = getRegForValue(II->getArgOperand(0));
     if (!ResultReg)
       return false;
     updateValueMap(II, ResultReg);
     return true;
   }
   case Intrinsic::experimental_stackmap:

llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp

else
       Res = DAG.getConstant(0, sdl, Ty);
 
     setValue(&I, Res);
     return nullptr;
   }
   case Intrinsic::annotation:
   case Intrinsic::ptr_annotation:
   case Intrinsic::launder_invariant_group:
+  case Intrinsic::strip_invariant_group:
     // Drop the intrinsic, but forward the value
     setValue(&I, getValue(I.getOperand(0)));
     return nullptr;
   case Intrinsic::assume:
   case Intrinsic::var_annotation:
   case Intrinsic::sideeffect:
     // Discard annotate attributes, assumptions, and artificial side-effects.
     return nullptr;

llvm/trunk/lib/IR/Value.cpp

if (auto *GEP = dyn_cast<GEPOperator>(V)) {
         if (const Value *RV = CS.getReturnedArgOperand()) {
           V = RV;
           continue;
         }
         // The result of launder.invariant.group must alias it's argument,
         // but it can't be marked with returned attribute, that's why it needs
         // special case.
         if (StripKind == PSK_ZeroIndicesAndAliasesAndInvariantGroups &&
-            CS.getIntrinsicID() == Intrinsic::launder_invariant_group) {
+            (CS.getIntrinsicID() == Intrinsic::launder_invariant_group ||
+             CS.getIntrinsicID() == Intrinsic::strip_invariant_group)) {
           V = CS.getArgOperand(0);
           continue;
         }
       }
       return V;
     }
     assert(V->getType()->isPointerTy() && "Unexpected operand type!");
   } while (Visited.insert(V).second);

llvm/trunk/lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp

static bool isCallPromotable(CallInst *CI) {
   case Intrinsic::memcpy:
   case Intrinsic::memmove:
   case Intrinsic::memset:
   case Intrinsic::lifetime_start:
   case Intrinsic::lifetime_end:
   case Intrinsic::invariant_start:
   case Intrinsic::invariant_end:
   case Intrinsic::launder_invariant_group:
+  case Intrinsic::strip_invariant_group:
   case Intrinsic::objectsize:
     return true;
   default:
     return false;
   }
 }
 
 bool AMDGPUPromoteAlloca::binaryOpIsDerivedFromSameAlloca(Value *BaseAlloca,
case Intrinsic::memset: {
                            MemSet->getLength(), MemSet->getDestAlignment(),
                            MemSet->isVolatile());
       Intr->eraseFromParent();
       continue;
     }
     case Intrinsic::invariant_start:
     case Intrinsic::invariant_end:
     case Intrinsic::launder_invariant_group:
+    case Intrinsic::strip_invariant_group:
       Intr->eraseFromParent();
       // FIXME: I think the invariant marker should still theoretically apply,
       // but the intrinsics need to be changed to accept pointers with any
       // address space.
       continue;
     case Intrinsic::objectsize: {
       Value *Src = Intr->getOperand(0);
       Type *SrcTy = Src->getType()->getPointerElementType();

llvm/trunk/test/Analysis/ValueTracking/invariant.group.ll

 ; RUN: opt -S -instsimplify -instcombine < %s | FileCheck %s
 
-; CHECK-LABEL: define void @checkNonnull()
-define void @checkNonnull() {
+; CHECK-LABEL: define void @checkNonnullLaunder()
+define void @checkNonnullLaunder() {
 ; CHECK:   %p = call i8* @llvm.launder.invariant.group.p0i8(i8* nonnull %0)
 ; CHECK:   %p2 = call i8* @llvm.launder.invariant.group.p0i8(i8* nonnull %p)
 ; CHECK:   call void @use(i8* nonnull %p2)
 entry:
   %0 = alloca i8, align 8
 
   %p = call i8* @llvm.launder.invariant.group.p0i8(i8* %0)
   %p2 = call i8* @llvm.launder.invariant.group.p0i8(i8* %p)
   call void @use(i8* %p2)
 
   ret void
 }
 
+; CHECK-LABEL: define void @checkNonnullStrip()
+define void @checkNonnullStrip() {
+; CHECK:   %p = call i8* @llvm.strip.invariant.group.p0i8(i8* nonnull %0)
+; CHECK:   %p2 = call i8* @llvm.strip.invariant.group.p0i8(i8* nonnull %p)
+; CHECK:   call void @use(i8* nonnull %p2)
+entry:
+  %0 = alloca i8, align 8
+
+  %p = call i8* @llvm.strip.invariant.group.p0i8(i8* %0)
+  %p2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %p)
+  call void @use(i8* %p2)
+
+  ret void
+}
+
 declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
+
 declare void @use(i8*)

llvm/trunk/test/CodeGen/Generic/intrinsics.ll

define double @test_cos(float %F) {
         %G = call float @cosf( float %F )               ; <float> [#uses=1]
         %H = fpext float %G to double           ; <double> [#uses=1]
         %I = call double @cos( double %H )              ; <double> [#uses=1]
         ret double %I
 }
 
 declare i8* @llvm.launder.invariant.group(i8*)
 
-define i8* @barrier(i8* %p) {
+define i8* @launder(i8* %p) {
         %q = call i8* @llvm.launder.invariant.group(i8* %p)
         ret i8* %q
 }
 
+declare i8* @llvm.strip.invariant.group(i8*)
+
+define i8* @strip(i8* %p) {
+        %q = call i8* @llvm.strip.invariant.group(i8* %p)
+        ret i8* %q
+}
+
+
 ; sideeffect
 
 declare void @llvm.sideeffect()
 
 define void @test_sideeffect() {
     call void @llvm.sideeffect()
     ret void
 }

llvm/trunk/test/Other/invariant.group.ll

+; RUN: opt -S -early-cse < %s | FileCheck %s
+; RUN: opt -S -gvn < %s | FileCheck %s
+; RUN: opt -S -newgvn < %s | FileCheck %s
+; RUN: opt -S -O3 < %s | FileCheck %s
+
+; These tests checks if passes with CSE functionality can do CSE on
+; launder.invariant.group, that is prohibited if there is a memory clobber
+; between barriers call.
+
+; CHECK-LABEL: define i8 @optimizable()
+define i8 @optimizable() {
+entry:
+    %ptr = alloca i8
+    store i8 42, i8* %ptr, !invariant.group !0
+; CHECK: call i8* @llvm.launder.invariant.group.p0i8
+    %ptr2 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
+; FIXME: This one could be CSE
+; CHECK: call i8* @llvm.launder.invariant.group
+    %ptr3 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
+; CHECK: call void @clobber(i8* {{.*}}%ptr)
+    call void @clobber(i8* %ptr)
+
+; CHECK: call void @use(i8* {{.*}}%ptr2)
+    call void @use(i8* %ptr2)
+; CHECK: call void @use(i8* {{.*}}%ptr3)
+    call void @use(i8* %ptr3)
+; CHECK: load i8, i8* %ptr3, {{.*}}!invariant.group
+    %v = load i8, i8* %ptr3, !invariant.group !0
+
+    ret i8 %v
+}
+
+; CHECK-LABEL: define i8 @unoptimizable()
+define i8 @unoptimizable() {
+entry:
+    %ptr = alloca i8
+    store i8 42, i8* %ptr, !invariant.group !0
+; CHECK: call i8* @llvm.launder.invariant.group.p0i8
+    %ptr2 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
+    call void @clobber(i8* %ptr)
+; CHECK: call i8* @llvm.launder.invariant.group.p0i8
+    %ptr3 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
+; CHECK: call void @clobber(i8* {{.*}}%ptr)
+    call void @clobber(i8* %ptr)
+; CHECK: call void @use(i8* {{.*}}%ptr2)
+    call void @use(i8* %ptr2)
+; CHECK: call void @use(i8* {{.*}}%ptr3)
+    call void @use(i8* %ptr3)
+; CHECK: load i8, i8* %ptr3, {{.*}}!invariant.group
+    %v = load i8, i8* %ptr3, !invariant.group !0
+
+    ret i8 %v
+}
+
+; CHECK-LABEL: define i8 @unoptimizable2()
+define i8 @unoptimizable2() {
+    %ptr = alloca i8
+    store i8 42, i8* %ptr, !invariant.group !0
+; CHECK: call i8* @llvm.launder.invariant.group
+    %ptr2 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
+    store i8 43, i8* %ptr
+; CHECK: call i8* @llvm.launder.invariant.group
+    %ptr3 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
+; CHECK: call void @clobber(i8* {{.*}}%ptr)
+    call void @clobber(i8* %ptr)
+; CHECK: call void @use(i8* {{.*}}%ptr2)
+    call void @use(i8* %ptr2)
+; CHECK: call void @use(i8* {{.*}}%ptr3)
+    call void @use(i8* %ptr3)
+; CHECK: load i8, i8* %ptr3, {{.*}}!invariant.group
+    %v = load i8, i8* %ptr3, !invariant.group !0
+    ret i8 %v
+}
+
+; This test check if optimizer is not proving equality based on mustalias
+; CHECK-LABEL: define void @dontProveEquality(i8* %a)
+define void @dontProveEquality(i8* %a) {
+  %b = call i8* @llvm.launder.invariant.group.p0i8(i8* %a)
+  %r = icmp eq i8* %b, %a
+; CHECK: call void @useBool(i1 %r)
+  call void @useBool(i1 %r)
+
+  %b2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+  %r2 = icmp eq i8* %b2, %a
+; CHECK: call void @useBool(i1 %r2)
+  call void @useBool(i1 %r2)
+
+  ret void
+}
+
+declare void @use(i8* readonly)
+declare void @useBool(i1)
+
+declare void @clobber(i8*)
+; CHECK: Function Attrs: inaccessiblememonly nounwind speculatable{{$}}
+; CHECK-NEXT: declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+
+; CHECK: Function Attrs: nounwind readnone speculatable{{$}}
+; CHECK-NEXT: declare i8* @llvm.strip.invariant.group.p0i8(i8*)
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
+
+
+!0 = !{}
 No newline at end of file

llvm/trunk/test/Other/launder.invariant.group.ll

-; RUN: opt -S -early-cse < %s | FileCheck %s
-; RUN: opt -S -gvn < %s | FileCheck %s
-; RUN: opt -S -newgvn < %s | FileCheck %s
-; RUN: opt -S -O3 < %s | FileCheck %s
-
-; These tests checks if passes with CSE functionality can do CSE on
-; launder.invariant.group, that is prohibited if there is a memory clobber
-; between barriers call.
-
-; CHECK-LABEL: define i8 @optimizable()
-define i8 @optimizable() {
-entry:
-    %ptr = alloca i8
-    store i8 42, i8* %ptr, !invariant.group !0
-; CHECK: call i8* @llvm.launder.invariant.group.p0i8
-    %ptr2 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
-; FIXME: This one could be CSE
-; CHECK: call i8* @llvm.launder.invariant.group
-    %ptr3 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
-; CHECK: call void @clobber(i8* {{.*}}%ptr)
-    call void @clobber(i8* %ptr)
-
-; CHECK: call void @use(i8* {{.*}}%ptr2)
-    call void @use(i8* %ptr2)
-; CHECK: call void @use(i8* {{.*}}%ptr3)
-    call void @use(i8* %ptr3)
-; CHECK: load i8, i8* %ptr3, {{.*}}!invariant.group
-    %v = load i8, i8* %ptr3, !invariant.group !0
-
-    ret i8 %v
-}
-
-; CHECK-LABEL: define i8 @unoptimizable()
-define i8 @unoptimizable() {
-entry:
-    %ptr = alloca i8
-    store i8 42, i8* %ptr, !invariant.group !0
-; CHECK: call i8* @llvm.launder.invariant.group.p0i8
-    %ptr2 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
-    call void @clobber(i8* %ptr)
-; CHECK: call i8* @llvm.launder.invariant.group.p0i8
-    %ptr3 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
-; CHECK: call void @clobber(i8* {{.*}}%ptr)
-    call void @clobber(i8* %ptr)
-; CHECK: call void @use(i8* {{.*}}%ptr2)
-    call void @use(i8* %ptr2)
-; CHECK: call void @use(i8* {{.*}}%ptr3)
-    call void @use(i8* %ptr3)
-; CHECK: load i8, i8* %ptr3, {{.*}}!invariant.group
-    %v = load i8, i8* %ptr3, !invariant.group !0
-
-    ret i8 %v
-}
-
-; CHECK-LABEL: define i8 @unoptimizable2()
-define i8 @unoptimizable2() {
-    %ptr = alloca i8
-    store i8 42, i8* %ptr, !invariant.group !0
-; CHECK: call i8* @llvm.launder.invariant.group
-    %ptr2 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
-    store i8 43, i8* %ptr
-; CHECK: call i8* @llvm.launder.invariant.group
-    %ptr3 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr)
-; CHECK: call void @clobber(i8* {{.*}}%ptr)
-    call void @clobber(i8* %ptr)
-; CHECK: call void @use(i8* {{.*}}%ptr2)
-    call void @use(i8* %ptr2)
-; CHECK: call void @use(i8* {{.*}}%ptr3)
-    call void @use(i8* %ptr3)
-; CHECK: load i8, i8* %ptr3, {{.*}}!invariant.group
-    %v = load i8, i8* %ptr3, !invariant.group !0
-    ret i8 %v
-}
-
-; This test check if optimizer is not proving equality based on mustalias
-; CHECK-LABEL: define void @dontProveEquality(i8* %a)
-define void @dontProveEquality(i8* %a) {
-  %b = call i8* @llvm.launder.invariant.group.p0i8(i8* %a)
-  %r = icmp eq i8* %b, %a
-;CHECK: call void @useBool(i1 %r)
-  call void @useBool(i1 %r)
-  ret void
-}
-
-declare void @use(i8* readonly)
-declare void @useBool(i1)
-
-declare void @clobber(i8*)
-; CHECK: Function Attrs: inaccessiblememonly nounwind speculatable{{$}}
-; CHECK-NEXT: declare i8* @llvm.launder.invariant.group.p0i8(i8*)
-declare i8* @llvm.launder.invariant.group.p0i8(i8*)
-
-!0 = !{}
-

llvm/trunk/test/Transforms/CodeGenPrepare/invariant.group.ll

 ; RUN: opt -codegenprepare -S < %s | FileCheck %s
 
 @tmp = global i8 0
 
 ; CHECK-LABEL: define void @foo() {
 define void @foo() {
 enter:
   ; CHECK-NOT: !invariant.group
   ; CHECK-NOT: @llvm.launder.invariant.group.p0i8(
-  ; CHECK: %val = load i8, i8* @tmp, !tbaa
-  %val = load i8, i8* @tmp, !invariant.group !0, !tbaa !{!1, !1, i64 0}
+  ; CHECK: %val = load i8, i8* @tmp{{$}}
+  %val = load i8, i8* @tmp, !invariant.group !0
   %ptr = call i8* @llvm.launder.invariant.group.p0i8(i8* @tmp)
   
   ; CHECK: store i8 42, i8* @tmp{{$}}
   store i8 42, i8* %ptr, !invariant.group !0
   
   ret void
 }
 ; CHECK-LABEL: }
 
-declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+; CHECK-LABEL: define void @foo2() {
+define void @foo2() {
+enter:
+  ; CHECK-NOT: !invariant.group
+  ; CHECK-NOT: @llvm.strip.invariant.group.p0i8(
+  ; CHECK: %val = load i8, i8* @tmp{{$}}
+  %val = load i8, i8* @tmp, !invariant.group !0
+  %ptr = call i8* @llvm.strip.invariant.group.p0i8(i8* @tmp)
+
+  ; CHECK: store i8 42, i8* @tmp{{$}}
+  store i8 42, i8* %ptr, !invariant.group !0
 
+  ret void
+}
+; CHECK-LABEL: }
+
+
+declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
 !0 = !{}
-!1 = !{!"x", !0}

llvm/trunk/test/Transforms/DeadStoreElimination/launder.invariant.group.ll

; CHECK-NOT: store i8 43
   %ptr3 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr2)
   %ptr4 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr3)
 
 ; CHECK: store i8 44
   store i8 44, i8* %ptr4
   ret void
 }
 
+; CHECK-LABEL: void @skip3Barriers(i8* %ptr)
+define void @skip3Barriers(i8* %ptr) {
+; CHECK-NOT: store i8 42
+  store i8 42, i8* %ptr
+; CHECK: %ptr2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr)
+  %ptr2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr)
+; CHECK-NOT: store i8 43
+  store i8 43, i8* %ptr2
+  %ptr3 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr2)
+  %ptr4 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr3)
+
+; CHECK: store i8 44
+  store i8 44, i8* %ptr4
+  ret void
+}
+
+; CHECK-LABEL: void @skip4Barriers(i8* %ptr)
+define void @skip4Barriers(i8* %ptr) {
+; CHECK-NOT: store i8 42
+  store i8 42, i8* %ptr
+; CHECK: %ptr2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr)
+  %ptr2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr)
+; CHECK-NOT: store i8 43
+  store i8 43, i8* %ptr2
+  %ptr3 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr2)
+  %ptr4 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr3)
+  %ptr5 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr3)
+
+; CHECK: store i8 44
+  store i8 44, i8* %ptr5
+  ret void
+}
+
+
 declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
 No newline at end of file

llvm/trunk/test/Transforms/FunctionAttrs/nocapture.ll

 @g2 = global i8* null
 ; CHECK: define void @captureLaunder(i8* %p)
 define void @captureLaunder(i8* %p) {
   %b = call i8* @llvm.launder.invariant.group.p0i8(i8* %p)
   store i8* %b, i8** @g2
   ret void
 }
 
+; CHECK: @nocaptureStrip(i8* nocapture %p)
+define void @nocaptureStrip(i8* %p) {
+entry:
+  %b = call i8* @llvm.strip.invariant.group.p0i8(i8* %p)
+  store i8 42, i8* %b
+  ret void
+}
+
+@g3 = global i8* null
+; CHECK: define void @captureStrip(i8* %p)
+define void @captureStrip(i8* %p) {
+  %b = call i8* @llvm.strip.invariant.group.p0i8(i8* %p)
+  store i8* %b, i8** @g3
+  ret void
+}
+
 declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)

llvm/trunk/test/Transforms/GVN/invariant.group.ll

entry:
     
     call void @foo(i8* %ptr)
     %b = load i8, i8* %ptr, !invariant.group !0
     
 ; CHECK: ret i8 42
     ret i8 %b
 }
 
+; CHECK-LABEL: define i1 @proveEqualityForStrip(
+define i1 @proveEqualityForStrip(i8* %a) {
+; FIXME: The first call could be also removed by GVN. Right now
+; DCE removes it. The second call is CSE'd with the first one.
+; CHECK: %b1 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+  %b1 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+; CHECK-NOT: llvm.strip.invariant.group
+  %b2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+  %r = icmp eq i8* %b1, %b2
+; CHECK: ret i1 true
+  ret i1 %r
+}
 ; CHECK-LABEL: define i8 @unoptimizable1() {
 define i8 @unoptimizable1() {
 entry:
     %ptr = alloca i8
     store i8 42, i8* %ptr
     call void @foo(i8* %ptr)
     %a = load i8, i8* %ptr, !invariant.group !0
 ; CHECK: ret i8 %a

 declare void @foo2(i8*, i8)
 declare void @bar(i8)
 declare i8* @getPointer(i8*)
 declare void @_ZN1A3fooEv(%struct.A*)
 declare void @_ZN1AC1Ev(%struct.A*)
 declare void @fooBit(i1*, i1)
 
 declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
+
 
-; Function Attrs: nounwind
-declare void @llvm.assume(i1 %cmp.vtables) #0
+declare void @llvm.assume(i1 %cmp.vtables)
 
 
-attributes #0 = { nounwind }
 !0 = !{}
 No newline at end of file

llvm/trunk/test/Transforms/GlobalOpt/invariant.group.barrier.ll

-; RUN: opt -S -globalopt < %s | FileCheck %s
-
-; This test is hint, what could globalOpt optimize and what it can't
-; FIXME: @tmp and @tmp2 can be safely set to 42
-; CHECK: @tmp = local_unnamed_addr global i32 0
-; CHECK: @tmp2 = local_unnamed_addr global i32 0
-; CHECK: @tmp3 = global i32 0
-
-@tmp = global i32 0
-@tmp2 = global i32 0
-@tmp3 = global i32 0
-@ptrToTmp3 = global i32* null
-
[e-mail]_ctors = appending global [1 x { i32, void ()* }] [{ i32, void ()* } { i32 65535, void ()* @_GLOBAL__I_a }]
-
-define i32 @TheAnswerToLifeTheUniverseAndEverything() {
-  ret i32 42
-}
-
-define void @_GLOBAL__I_a() {
-enter:
-  call void @_optimizable()
-  call void @_not_optimizable()
-  ret void
-}
-
-define void @_optimizable() {
-enter:
-  %valptr = alloca i32
-  
-  %val = call i32 @TheAnswerToLifeTheUniverseAndEverything()
-  store i32 %val, i32* @tmp
-  store i32 %val, i32* %valptr
-  
-  %0 = bitcast i32* %valptr to i8*
-  %barr = call i8* @llvm.launder.invariant.group(i8* %0)
-  %1 = bitcast i8* %barr to i32*
-  
-  %val2 = load i32, i32* %1
-  store i32 %val2, i32* @tmp2
-  ret void
-}
-
-; We can't step through launder.invariant.group here, because that would change
-; this load in @usage_of_globals()
-; val = load i32, i32* %ptrVal, !invariant.group !0 
-; into 
-; %val = load i32, i32* @tmp3, !invariant.group !0
-; and then we could assume that %val and %val2 to be the same, which coud be 
-; false, because @changeTmp3ValAndCallBarrierInside() may change the value
-; of @tmp3.
-define void @_not_optimizable() {
-enter:
-  store i32 13, i32* @tmp3, !invariant.group !0
-  
-  %0 = bitcast i32* @tmp3 to i8*
-  %barr = call i8* @llvm.launder.invariant.group(i8* %0)
-  %1 = bitcast i8* %barr to i32*
-  
-  store i32* %1, i32** @ptrToTmp3
-  store i32 42, i32* %1, !invariant.group !0
-  
-  ret void
-}
-define void @usage_of_globals() {
-entry:
-  %ptrVal = load i32*, i32** @ptrToTmp3
-  %val = load i32, i32* %ptrVal, !invariant.group !0
-  
-  call void @changeTmp3ValAndCallBarrierInside()
-  %val2 = load i32, i32* @tmp3, !invariant.group !0
-  ret void;
-}
-
-declare void @changeTmp3ValAndCallBarrierInside()
-
-declare i8* @llvm.launder.invariant.group(i8*)
-
-!0 = !{}

llvm/trunk/test/Transforms/GlobalOpt/invariant.group.ll

+; RUN: opt -S -globalopt < %s | FileCheck %s
+
+; This test is hint, what could globalOpt optimize and what it can't
+; FIXME: @tmp and @tmp2 can be safely set to 42
+; CHECK: @tmp = local_unnamed_addr global i32 0
+; CHECK: @tmp2 = local_unnamed_addr global i32 0
+; CHECK: @tmp3 = global i32 0
+
+@tmp = global i32 0
+@tmp2 = global i32 0
+@tmp3 = global i32 0
+@ptrToTmp3 = global i32* null
+
[e-mail]_ctors = appending global [1 x { i32, void ()* }] [{ i32, void ()* } { i32 65535, void ()* @_GLOBAL__I_a }]
+
+define i32 @TheAnswerToLifeTheUniverseAndEverything() {
+  ret i32 42
+}
+
+define void @_GLOBAL__I_a() {
+enter:
+  call void @_optimizable()
+  call void @_not_optimizable()
+  ret void
+}
+
+define void @_optimizable() {
+enter:
+  %valptr = alloca i32
+
+  %val = call i32 @TheAnswerToLifeTheUniverseAndEverything()
+  store i32 %val, i32* @tmp
+  store i32 %val, i32* %valptr
+
+  %0 = bitcast i32* %valptr to i8*
+  %barr = call i8* @llvm.launder.invariant.group(i8* %0)
+  %1 = bitcast i8* %barr to i32*
+
+  %val2 = load i32, i32* %1
+  store i32 %val2, i32* @tmp2
+  ret void
+}
+
+; We can't step through launder.invariant.group here, because that would change
+; this load in @usage_of_globals()
+; val = load i32, i32* %ptrVal, !invariant.group !0
+; into
+; %val = load i32, i32* @tmp3, !invariant.group !0
+; and then we could assume that %val and %val2 to be the same, which coud be
+; false, because @changeTmp3ValAndCallBarrierInside() may change the value
+; of @tmp3.
+define void @_not_optimizable() {
+enter:
+  store i32 13, i32* @tmp3, !invariant.group !0
+
+  %0 = bitcast i32* @tmp3 to i8*
+  %barr = call i8* @llvm.launder.invariant.group(i8* %0)
+  %1 = bitcast i8* %barr to i32*
+
+  store i32* %1, i32** @ptrToTmp3
+  store i32 42, i32* %1, !invariant.group !0
+
+  ret void
+}
+define void @usage_of_globals() {
+entry:
+  %ptrVal = load i32*, i32** @ptrToTmp3
+  %val = load i32, i32* %ptrVal, !invariant.group !0
+
+  call void @changeTmp3ValAndCallBarrierInside()
+  %val2 = load i32, i32* @tmp3, !invariant.group !0
+  ret void;
+}
+
+declare void @changeTmp3ValAndCallBarrierInside()
+
+declare i8* @llvm.launder.invariant.group(i8*)
+
+!0 = !{}

llvm/trunk/test/Transforms/InstCombine/invariant.group.ll

 ; RUN: opt -instcombine -S < %s | FileCheck %s
 
+
 ; CHECK-LABEL: define i8* @simplifyNullLaunder()
 define i8* @simplifyNullLaunder() {
 ; CHECK-NEXT: ret i8* null
   %b2 = call i8* @llvm.launder.invariant.group.p0i8(i8* null)
   ret i8* %b2
 }
 
 ; CHECK-LABEL: define i8 addrspace(42)* @dontsimplifyNullLaunderForDifferentAddrspace()

 
 ; CHECK-LABEL: define i8 addrspace(42)* @simplifyUndefLaunder2()
 define i8 addrspace(42)* @simplifyUndefLaunder2() {
 ; CHECK-NEXT: ret i8 addrspace(42)* undef
   %b2 = call i8 addrspace(42)* @llvm.launder.invariant.group.p42i8(i8 addrspace(42)* undef)
   ret i8 addrspace(42)* %b2
 }
 
-
 declare i8* @llvm.launder.invariant.group.p0i8(i8*)
 declare i8 addrspace(42)* @llvm.launder.invariant.group.p42i8(i8 addrspace(42)*)
+
+
+; CHECK-LABEL: define i8* @simplifyNullStrip()
+define i8* @simplifyNullStrip() {
+; CHECK-NEXT: ret i8* null
+  %b2 = call i8* @llvm.strip.invariant.group.p0i8(i8* null)
+  ret i8* %b2
+}
+
+; CHECK-LABEL: define i8 addrspace(42)* @dontsimplifyNullStripForDifferentAddrspace()
+define i8 addrspace(42)* @dontsimplifyNullStripForDifferentAddrspace() {
+; CHECK: %b2 = call i8 addrspace(42)* @llvm.strip.invariant.group.p42i8(i8 addrspace(42)* null)
+; CHECK: ret i8 addrspace(42)* %b2
+  %b2 = call i8 addrspace(42)* @llvm.strip.invariant.group.p42i8(i8 addrspace(42)* null)
+  ret i8 addrspace(42)* %b2
+}
+
+; CHECK-LABEL: define i8* @simplifyUndefStrip()
+define i8* @simplifyUndefStrip() {
+; CHECK-NEXT: ret i8* undef
+  %b2 = call i8* @llvm.strip.invariant.group.p0i8(i8* undef)
+  ret i8* %b2
+}
+
+; CHECK-LABEL: define i8 addrspace(42)* @simplifyUndefStrip2()
+define i8 addrspace(42)* @simplifyUndefStrip2() {
+; CHECK-NEXT: ret i8 addrspace(42)* undef
+  %b2 = call i8 addrspace(42)* @llvm.strip.invariant.group.p42i8(i8 addrspace(42)* undef)
+  ret i8 addrspace(42)* %b2
+}
+
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
+declare i8 addrspace(42)* @llvm.strip.invariant.group.p42i8(i8 addrspace(42)*)
+

llvm/trunk/test/Transforms/NewGVN/invariant.group.ll

entry:
     
     call void @foo(i8* %ptr)
     %b = load i8, i8* %ptr, !invariant.group !0
     
 ; CHECK: ret i8 42
     ret i8 %b
 }
 
+; CHECK-LABEL: define i1 @proveEqualityForStrip(
+define i1 @proveEqualityForStrip(i8* %a) {
+; FIXME: The first call could be also removed by GVN. Right now
+; DCE removes it. The second call is CSE'd with the first one.
+; CHECK: %b1 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+  %b1 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+; CHECK-NOT: llvm.strip.invariant.group
+  %b2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+  %r = icmp eq i8* %b1, %b2
+; CHECK: ret i1 true
+  ret i1 %r
+}
+
 ; CHECK-LABEL: define i8 @unoptimizable1() {
 define i8 @unoptimizable1() {
 entry:
     %ptr = alloca i8
     store i8 42, i8* %ptr
     call void @foo(i8* %ptr)
     %a = load i8, i8* %ptr, !invariant.group !0
 ; CHECK: ret i8 %a

 declare i8* @llvm.launder.invariant.group.p0i8(i8*)
 
 ; Function Attrs: nounwind
 declare void @llvm.assume(i1 %cmp.vtables) #0
 
 
 attributes #0 = { nounwind }
 !0 = !{}
 No newline at end of file