[AMDGPU] Lower kernel LDS into a sorted structure

[JonChesterfield]

Couple of minor comments above, overall approach looks sound to me

llvm/lib/IR/Value.cpp

535

since this hazard affects all the targets, we should probably fix it outside of the amdgpu specific LDS patch

llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.cpp

37

Is this safe (enough) for space? Wonder if it's another place for a heap allocated worklist

[rampitec]

llvm/lib/IR/Value.cpp

535

I can make a separate patch, but the testcase to exploit it here. It needs to see a constant use to fail.

[hsmhsm]

llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.cpp

39

For me, this logic looks incomplete in general sense. Constants are constants throughout compilation phase, meaning, once a constant is created, it is reused in multiple contexts when required. So this is possible case.

void kern_0() {

Inst_1(Const_1(Const_3(LDS)));

}

void kern_1() {

Inst_1(Const_2(Const_3(LDS)));

}

Here Inst_1 and Inst_2 are two different instances, even if they are exactly same, but Const_3 is one-and-the same for both kern_0() and kern_1(). Will this case works here?

47

This is fine, but combining the testing of two different requirements for both module lowering and kernel lowering within same function looks maintenance headache to me. May be it is useful to separate them into two different functions, and findVariablesToLower() call one of them, based on F. But it is upto you. I am fine either way.

[hsmhsm]

As I understand this patch - changes to Value::replaceUsesWithIf() is a fundamental change, so it is better that some one from non AMDGPU team also review it.

Crux of the whole logic within the patch is from two functions - (1) shouldLowerLDSToStruct() and (2) isUsedOnlyFromFunction(). Any bug here will affect the whole functionality of the pass. So we need to be extra careful while reviweing these two functions.

[hsmhsm]

llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp

232

There is some problem here - because of this, llvm -lit tests including your own newly added tests are failing.

After changing

Twine VarName(F ? "llvm.amdgcn.kernel." + Twine(F->getName() + ".lds") : "llvm.amdgcn.module.lds");

To

Twine VarName(F ? Twine("llvm.amdgcn.kernel.") + Twine(F->getName()) + Twine(".lds") : "llvm.amdgcn.module.lds");

it works.

[hsmhsm]

llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp

255

We end-up erasing and re-constructing "llvm.used" as many times as number of kernels which have undergone LDS lowering within them, plus for a very first module level lowering. Not sure about the side-effect it. Also may not worth, erasing and re-constructing "llvm.used" so many times in my openion. But it is left to you.

llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.cpp

39

Further, I verified that below test-case misses lowering because of above incomplete logic, but lowering is required here.

@kern = addrspace(3) global float undef, align 4

define amdgpu_kernel void @k0() {
  %ld = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
  ret void
}

define amdgpu_kernel void @k1() {
  %ld = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
  ret void
}

[hsmhsm]

llvm/test/CodeGen/AMDGPU/lower-module-lds-constantexpr.ll

44

@kern is not lowered here, and same below at line 46.

llvm/test/CodeGen/AMDGPU/lower-module-lds-inactive.ll

10

@var1 is kernel lowered. so it should have 0 use, and got erased. Why it is still there? and we should have CHECK-NOT test for @var1.

[hsmhsm]

llvm/lib/IR/Value.cpp

549

I do not understand, why are we not considering "GlobalVlaue" here. It is upto caller who calls Value::replaceUsesWithIf() to decide on it, and this caller is the one who decide on the logic of the callback of ShouldReplace(). If this caller decide to replace "this" by "New" even in global scope (which is probably an initializer to global), Value::replaceUsesWithIf() should honor it.

llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp

256

Further I have my own doubt about the why we actually doing this - removeFromUsedLists().

I think, it is problamatic even for moduleLDS lower().

When we erase and reconstruct llvm.used/llvm.compiler.used, are we sure that we do not land-up with stale references to (LDS) globals which were added to llvm.used/llvm.compiler.used as used?

As I understand it, the story with "usedList" is as below in this pass:

"When an LDS is used only within llvm.used/llvm.compiler.used, but *nowhere else*, then, we don't want to lower such LDS, that we take care while collecting the globals (for lowering) at the begining. On the other hand, when LDS is used within llvm.used/llvm.compiler.used, but also in other scope (function scope for module lowering), and (kernel scope for kernel lowering), then we want to lower such LDS. That means, we endup replacing LDS within lvm.used/llvm.compiler.used by its struct offset."

If above is what we want to achieve, then why do we need to have call to removeFromUsedLists()?

As far as I understand, call to removeFromUsedLists() would cause problems for kernel lowering because of repeated erasing and re-constructing of llvm.used/llvm.compiler.used.

[foad]

llvm/include/llvm/IR/Value.h

313–314

I guess you need to update the "does not support ... constant users" here?

321–322

... and here.

[rampitec]

Fixed tidy and twine issues.

llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp

232

clang-tidy also warns about local twine variable. Switched to std::string.

llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.cpp

37

It goes up the users and bails on a first non ConstantExpr use. The call stack is as deep as much nested ConstantExprs we have. To run into a problem we would need to have a ConstExpr of a form like gep(gep(gep(gep(...)))) nested really deep. If you still feel I need to switch to a worklist I will do it, it just looks unnecessary here to me.

[rampitec]

Addressed some review comments. Added test. Looking into llvm.used processing.

llvm/lib/IR/Value.cpp

549

This is simply a Constant handling code borrowed from replaceAllUsesWith(). See line 514 above.

llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.cpp

39

Right. This is exactly the case checked in the lower-kernel-lds-constexpr.ll, k0 and k1. Two kernels use the same constant and that cannot be lowered because we cannot replace the constant in one kernel without affecting another one. Handling of this case would require a transformation replacing a constant with an instruction. That is possible but not trivial and not handled in this pass.

At the same time lowering is NOT required. A safe action for kernel LDS lowering is to bail and not touch anything. This is optimization and not required for correct codegen.

39

For me, this logic looks incomplete in general sense. Constants are constants throughout compilation phase, meaning, once a constant is created, it is reused in multiple contexts when required. So this is possible case.

void kern_0() {

Inst_1(Const_1(Const_3(LDS)));

}

void kern_1() {

Inst_1(Const_2(Const_3(LDS)));

}

Here Inst_1 and Inst_2 are two different instances, even if they are exactly same, but Const_3 is one-and-the same for both kern_0() and kern_1(). Will this case works here?

Yes, this is the situation handled by handleOperandChange(), that is why a normal setUse() cannot be used in the first place. Added test k3 to the lower-kernel-lds-constexpr.ll.

47

This is fine, but combining the testing of two different requirements for both module lowering and kernel lowering within same function looks maintenance headache to me. May be it is useful to separate them into two different functions, and findVariablesToLower() call one of them, based on F. But it is upto you. I am fine either way.

I did not want to clone practically identical code. I hope after a series of changes there will be no module LDS at all, it will all go into a per-kernel allocation.

llvm/test/CodeGen/AMDGPU/lower-module-lds-constantexpr.ll

44

What do you mean? @kern is lowered. This is input IR, the output does not have it.

[rampitec]

llvm/lib/IR/Value.cpp

535

since this hazard affects all the targets, we should probably fix it outside of the amdgpu specific LDS patch

D103051

[rampitec]

Handled a case when a variable would not be removed even if lowered.

llvm/test/CodeGen/AMDGPU/lower-module-lds-inactive.ll

10

There was a dangling use of constantexpr bitcast from llvm.used, so use list was not empty. Added a call to C->removeDeadConstantUsers() while destroying llvm.used. Negative check is also added.

[rampitec]

Remove erased variables from local UsedList set.

llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp

256

Actually we do not just keep removing and recreating llvm.used, it is a converging process. We are only removing variables from there along the route when we replace them. We have to do it before we actually replace a variable because after replaceAllUsesWith() the llvm.used would contain a useless reference to the new structure (and verifier rightfully complaints about that spume).

For the kernel lowering an additional reason to drop it before the replacement is to have isUsedOnlyFromFunction() not to be stopped by a GlobalVariable use. Note that shouldLowerLDSToStruct() has a check that UsedList contains V->stripPointerCasts(), so llvm.used does not prevent the collection. isUsedOnlyFromFunction() does not need this handling because reference is already dropped by the removeFromUsedLists() call here.

For the case where a variable only used by an llvm.used we would not do anything at all since the variable simply will not get into the collection from findVariablesToLower().

We shall also consider the reason why an LDS global may appear in an used list in the first place. The use by a function creates an indirect use by a kernel (at least by an indirectly called function), I do not see any other reason to add the variable to a used list. For a kernel that is simply not needed because a use is direct and immediately visible. In reality this path is very unlikely to be taken in a kernel case, although shall work correctly nonetheless.

Moreover, when we get rid of module LDS, and will have only kernel LDS, plus will manage all the cases all LDS global will be completely replaced and lowered and unconditionally erased from used list. At that point we could just replace this logic with a bulk drop at the top of the runOnModule().

However, while exploring all of that I have realized there is a bug in the implementation, I have to erase a variable from local set UsedList when erasing so we do not have dangling pointers there. Fixed.

[hsmhsm]

llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.cpp

39

Why do we need a transformation replacing a constant with an instruction? I am not sure.

My understanding is this -

In the above example, defintely @kern is suppose to be lowered both within kernels @k0 and @k1. But since the struct offset for @kern would be different within both the kernels (assuming they also use other LDS globals), the current lowering flow - that is handling one kernel at a time is the limitation here as I understand it.

Instead let's assume below flow, then it is possible to achieve it. But, only problem is, this flow is overkill since we need to do some heavy work here which is bit against LLVM infrastructure.

1.  ConsolidatedGlobals = nullptr

2.  FOR (each kernel K) DO
        LDSGlobals = Collect LDS globals to be lowered
        Create struct type and struct instance for K, and save it.
        Set union LDSGlobals with ConsolidatedGlobals
     ENDFOR

3.   Call removeFromUsedList for ConsolidatedGlobals

4.   Construct an indexing table which given an LDS and a Kernel, it returns the corresponding struct offset.
 
5.   FOR (each lds global GV from ConsolidatedGlobals) DO
         FOR (each kernel K) DO
              Offset = Get offset for the pair (GV, K) from indexing table.
              GEF = Create a GEF constant expression using Offset.
              KernelUses = Collect all uses of GV within kernel K.
              FOR (each use U from KernelUses) DO
                  Replace GV within U by GEF
              ENDFOR
         ENDFOR
      ENDFOR

6.   FOR (each global GV from ConsolidatedGlobals) DO
         Erase GV from module.
      ENDFOR

llvm/test/CodeGen/AMDGPU/lower-module-lds-constantexpr.ll

44

My bad, I misunderstood it, please ignore it.

[rampitec]

llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.cpp

39

A constant is uniqued. That means being used from two different kernels it is still the same constant. To lower it differently in different kernels we need to get rid of that modality and duplicate it. One way is to construct a new constant, another is to construct an instruction(s).

The only reason now it is a constant is because it was simpler to implement, one do not need to find a proper insertion point. With kernels and uniqued constants that doesn't work anymore.

[rampitec]

Rebased.

[hsmhsm]

llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.cpp

39

Oh! yes, get it.

[hsmhsm]

Assuming that we have known limitations with handling of same constant expression appearing in two different kernels, this patch looks okay to me.

llvm/include/llvm/IR/Value.h

#endif
   /// Change non-metadata uses of this to point to a new Value.
   ///
   /// Go through the uses list for this definition and make each use point to
   /// "V" instead of "this". This function skips metadata entries in the list.
   void replaceNonMetadataUsesWith(Value *V);
 
   /// Go through the uses list for this definition and make each use point
   /// to "V" if the callback ShouldReplace returns true for the given Use.
   /// Unlike replaceAllUsesWith() this function does not support basic block
-  /// values or constant users.
+  /// values.
   void replaceUsesWithIf(Value *New,
-                         llvm::function_ref<bool(Use &U)> ShouldReplace) {
-    assert(New && "Value::replaceUsesWithIf(<null>) is invalid!");
-    assert(New->getType() == getType() &&
-           "replaceUses of value with new value of different type!");
-
-    for (use_iterator UI = use_begin(), E = use_end(); UI != E;) {
-      Use &U = *UI;
-      ++UI;
-      if (!ShouldReplace(U))
-        continue;
-      U.set(New);
-    }
-  }
+                         llvm::function_ref<bool(Use &U)> ShouldReplace);
 
   /// replaceUsesOutsideBlock - Go through the uses list for this definition and
   /// make each use point to "V" instead of "this" when the use is outside the
   /// block. 'This's use list is expected to have at least one element.
   /// Unlike replaceAllUsesWith() this function does not support basic block
-  /// values or constant users.
+  /// values.
   void replaceUsesOutsideBlock(Value *V, BasicBlock *BB);
 
   //----------------------------------------------------------------------
   // Methods for handling the chain of uses of this Value.
   //
   // Materializing a function can introduce new uses, so these methods come in
   // two variants:
   // The methods that start with materialized_ check the uses that are

llvm/lib/IR/Value.cpp

 void Value::replaceAllUsesWith(Value *New) {
   doRAUW(New, ReplaceMetadataUses::Yes);
 }
 
 void Value::replaceNonMetadataUsesWith(Value *New) {
   doRAUW(New, ReplaceMetadataUses::No);
 }
 
+void Value::replaceUsesWithIf(Value *New,
+                              llvm::function_ref<bool(Use &U)> ShouldReplace) {
+  assert(New && "Value::replaceUsesWithIf(<null>) is invalid!");
+  assert(New->getType() == getType() &&
+         "replaceUses of value with new value of different type!");
+
+  for (use_iterator UI = use_begin(), E = use_end(); UI != E;) {
+    Use &U = *UI;
+    ++UI;
+    if (!ShouldReplace(U))
+      continue;
+    // Must handle Constants specially, we cannot call replaceUsesOfWith on a
+    // constant because they are uniqued.
+    if (auto *C = dyn_cast<Constant>(U.getUser())) {
+      if (!isa<GlobalValue>(C)) {
+        C->handleOperandChange(this, New);
+        continue;
+      }
+    }
+    U.set(New);
+  }
+}
+
 /// Replace llvm.dbg.* uses of MetadataAsValue(ValueAsMetadata(V)) outside BB
 /// with New.
 static void replaceDbgUsesOutsideBlock(Value *V, Value *New, BasicBlock *BB) {
   SmallVector<DbgVariableIntrinsic *> DbgUsers;
   findDbgUsers(DbgUsers, V);
   for (auto *DVI : DbgUsers) {
     if (DVI->getParent() != BB)
       DVI->replaceVariableLocationOp(V, New);

llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp

static void removeFromUsedList(Module &M, StringRef Name,
     }
 
     if (Init.size() == CA->getNumOperands()) {
       return; // none to remove
     }
 
     GV->eraseFromParent();
 
+    for (Constant *C : ToRemove) {
+      C->removeDeadConstantUsers();
+    }
+
     if (!Init.empty()) {
       ArrayType *ATy =
           ArrayType::get(Type::getInt8PtrTy(M.getContext()), Init.size());
       GV =
           new llvm::GlobalVariable(M, ATy, false, GlobalValue::AppendingLinkage,
                                    ConstantArray::get(ATy, Init), Name);
       GV->setSection("llvm.metadata");
     }
static void markUsedByKernel(IRBuilder<> &Builder, Function *Func,
     Value *UseInstance[1] = {Builder.CreateInBoundsGEP(
         SGV->getValueType(), SGV, ConstantInt::get(Type::getInt32Ty(Ctx), 0))};
 
     Builder.CreateCall(FTy, Decl, {},
                        {OperandBundleDefT<Value *>("ExplicitUse", UseInstance)},
                        "");
   }
 
+private:
+  SmallPtrSet<GlobalValue *, 32> UsedList;
+
 public:
   static char ID;
 
   AMDGPULowerModuleLDS() : ModulePass(ID) {
     initializeAMDGPULowerModuleLDSPass(*PassRegistry::getPassRegistry());
   }
 
   bool runOnModule(Module &M) override {
+    UsedList = AMDGPU::getUsedList(M);
+
+    bool Changed = processUsedLDS(M);
+
+    for (Function &F : M.functions()) {
+      if (!AMDGPU::isKernelCC(&F))
+        continue;
+      Changed |= processUsedLDS(M, &F);
+    }
+
+    UsedList.clear();
+    return Changed;
+  }
+
+private:
+  bool processUsedLDS(Module &M, Function *F = nullptr) {
     LLVMContext &Ctx = M.getContext();
     const DataLayout &DL = M.getDataLayout();
-    SmallPtrSet<GlobalValue *, 32> UsedList = AMDGPU::getUsedList(M);
 
     // Find variables to move into new struct instance
     std::vector<GlobalVariable *> FoundLocalVars =
-        AMDGPU::findVariablesToLower(M, UsedList);
+        AMDGPU::findVariablesToLower(M, UsedList, F);
 
     if (FoundLocalVars.empty()) {
       // No variables to rewrite, no changes made.
       return false;
     }
 
     // Sort by alignment, descending, to minimise padding.
     // On ties, sort by size, descending, then by name, lexicographical.
bool processUsedLDS(Module &M, Function *F = nullptr) {
     }
 
     std::vector<Type *> LocalVarTypes;
     LocalVarTypes.reserve(LocalVars.size());
     std::transform(
         LocalVars.cbegin(), LocalVars.cend(), std::back_inserter(LocalVarTypes),
         [](const GlobalVariable *V) -> Type * { return V->getValueType(); });
 
-    StructType *LDSTy = StructType::create(
-        Ctx, LocalVarTypes, llvm::StringRef("llvm.amdgcn.module.lds.t"));
+    std::string VarName(
+        F ? (Twine("llvm.amdgcn.kernel.") + F->getName() + ".lds").str()
+          : "llvm.amdgcn.module.lds");
+    StructType *LDSTy =
+        StructType::create(Ctx, LocalVarTypes, VarName + ".t");
 
     Align MaxAlign =
         AMDGPU::getAlign(DL, LocalVars[0]); // was sorted on alignment
 
     GlobalVariable *SGV = new GlobalVariable(
         M, LDSTy, false, GlobalValue::InternalLinkage, UndefValue::get(LDSTy),
-        "llvm.amdgcn.module.lds", nullptr, GlobalValue::NotThreadLocal,
-        AMDGPUAS::LOCAL_ADDRESS, false);
+        VarName, nullptr, GlobalValue::NotThreadLocal, AMDGPUAS::LOCAL_ADDRESS,
+        false);
     SGV->setAlignment(MaxAlign);
+    if (!F) {
       appendToCompilerUsed(
           M, {static_cast<GlobalValue *>(
                  ConstantExpr::getPointerBitCastOrAddrSpaceCast(
                      cast<Constant>(SGV), Type::getInt8PtrTy(Ctx)))});
+    }
 
     // The verifier rejects used lists containing an inttoptr of a constant
     // so remove the variables from these lists before replaceAllUsesWith
     removeFromUsedLists(M, LocalVars);
 
     // Replace uses of ith variable with a constantexpr to the ith field of the
     // instance that will be allocated by AMDGPUMachineFunction
     Type *I32 = Type::getInt32Ty(Ctx);
     for (size_t I = 0; I < LocalVars.size(); I++) {
       GlobalVariable *GV = LocalVars[I];
       Constant *GEPIdx[] = {ConstantInt::get(I32, 0), ConstantInt::get(I32, I)};
-      GV->replaceAllUsesWith(
-          ConstantExpr::getGetElementPtr(LDSTy, SGV, GEPIdx));
+      Constant *GEP = ConstantExpr::getGetElementPtr(LDSTy, SGV, GEPIdx);
+      if (F) {
+        GV->replaceUsesWithIf(GEP, [F](Use &U) {
+          return AMDGPU::isUsedOnlyFromFunction(U.getUser(), F);
+        });
+      } else {
+        GV->replaceAllUsesWith(GEP);
+      }
+      if (GV->use_empty())
         GV->eraseFromParent();
     }
 
     // Mark kernels with asm that reads the address of the allocated structure
     // This is not necessary for lowering. This lets other passes, specifically
     // PromoteAlloca, accurately calculate how much LDS will be used by the
     // kernel after lowering.
-    {
+    if (!F) {
       IRBuilder<> Builder(Ctx);
       SmallPtrSet<Function *, 32> Kernels;
       for (auto &I : M.functions()) {
         Function *Func = &I;
         if (AMDGPU::isKernelCC(Func) && !Kernels.contains(Func)) {
           markUsedByKernel(Builder, Func, SGV);
           Kernels.insert(Func);
         }

llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.h

 #define LLVM_LIB_TARGET_AMDGPU_UTILS_AMDGPULDSUTILS_H
 
 #include "AMDGPU.h"
 
 namespace llvm {
 
 namespace AMDGPU {
 
-bool isKernelCC(Function *Func);
+bool isKernelCC(const Function *Func);
 
 Align getAlign(DataLayout const &DL, const GlobalVariable *GV);
 
-bool userRequiresLowering(const SmallPtrSetImpl<GlobalValue *> &UsedList,
-                          User *InitialUser);
+/// \returns true if an LDS global requres lowering to a module LDS structure
+/// if \p F is not given. If \p F is given it must be a kernel and function
+/// \returns true if an LDS global is directly used from that kernel and it
+/// is safe to replace its uses with a kernel LDS structure member.
+/// \p UsedList contains a union of llvm.used and llvm.compiler.used variables
+/// which do not count as a use.
+bool shouldLowerLDSToStruct(const SmallPtrSetImpl<GlobalValue *> &UsedList,
+                            const GlobalVariable &GV,
+                            const Function *F = nullptr);
 
 std::vector<GlobalVariable *>
-findVariablesToLower(Module &M, const SmallPtrSetImpl<GlobalValue *> &UsedList);
+findVariablesToLower(Module &M, const SmallPtrSetImpl<GlobalValue *> &UsedList,
+                     const Function *F = nullptr);
 
 SmallPtrSet<GlobalValue *, 32> getUsedList(Module &M);
 
+/// \returns true if all uses of \p U end up in a function \p F.
+bool isUsedOnlyFromFunction(const User *U, const Function *F);
+
 } // end namespace AMDGPU
 
 } // end namespace llvm
 
 #endif // LLVM_LIB_TARGET_AMDGPU_UTILS_AMDGPULDSUTILS_H

llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.cpp

 #include "llvm/IR/Constants.h"
 
 using namespace llvm;
 
 namespace llvm {
 
 namespace AMDGPU {
 
-bool isKernelCC(Function *Func) {
+bool isKernelCC(const Function *Func) {
   return AMDGPU::isModuleEntryFunctionCC(Func->getCallingConv());
 }
 
 Align getAlign(DataLayout const &DL, const GlobalVariable *GV) {
   return DL.getValueOrABITypeAlignment(GV->getPointerAlignment(DL),
                                        GV->getValueType());
 }
 
-bool userRequiresLowering(const SmallPtrSetImpl<GlobalValue *> &UsedList,
-                          User *InitialUser) {
+bool isUsedOnlyFromFunction(const User *U, const Function *F) {
+  if (auto *I = dyn_cast<Instruction>(U)) {
+    return I->getFunction() == F;
+  }
+
+  if (auto *C = dyn_cast<ConstantExpr>(U)) {
+    return all_of(U->users(),
+                  [F](const User *U) { return isUsedOnlyFromFunction(U, F); });
+  }
+
+  return false;
+}
+
+bool shouldLowerLDSToStruct(const SmallPtrSetImpl<GlobalValue *> &UsedList,
+                            const GlobalVariable &GV, const Function *F) {
   // Any LDS variable can be lowered by moving into the created struct
   // Each variable so lowered is allocated in every kernel, so variables
   // whose users are all known to be safe to lower without the transform
   // are left unchanged.
-  SmallPtrSet<User *, 8> Visited;
-  SmallVector<User *, 16> Stack;
-  Stack.push_back(InitialUser);
+  bool Ret = false;
+  SmallPtrSet<const User *, 8> Visited;
+  SmallVector<const User *, 16> Stack(GV.users());
+
+  assert(!F || isKernelCC(F));
 
   while (!Stack.empty()) {
-    User *V = Stack.pop_back_val();
+    const User *V = Stack.pop_back_val();
     Visited.insert(V);
 
     if (auto *G = dyn_cast<GlobalValue>(V->stripPointerCasts())) {
       if (UsedList.contains(G)) {
         continue;
       }
     }
 
     if (auto *I = dyn_cast<Instruction>(V)) {
-      if (isKernelCC(I->getFunction())) {
-        continue;
+      const Function *UF = I->getFunction();
+      if (UF == F) {
+        // Used from this kernel, we want to put it into the structure.
+        Ret = true;
+      } else if (!F) {
+        Ret |= !isKernelCC(UF);
       }
+      continue;
     }
 
     if (auto *E = dyn_cast<ConstantExpr>(V)) {
-      for (Value::user_iterator EU = E->user_begin(); EU != E->user_end();
-           ++EU) {
-        if (Visited.insert(*EU).second) {
-          Stack.push_back(*EU);
+      if (F) {
+        // Any use which does not end up an instruction disqualifies a
+        // variable to be put into a kernel's LDS structure because later
+        // we will need to replace only this kernel's uses for which we
+        // need to identify a using function.
+        return isUsedOnlyFromFunction(E, F);
+      }
+      for (const User *U : E->users()) {
+        if (Visited.insert(U).second) {
+          Stack.push_back(U);
         }
       }
       continue;
     }
 
-    // Unknown user, conservatively lower the variable
-    return true;
+    // Unknown user, conservatively lower the variable.
+    // For module LDS conservatively means place it into the module LDS struct.
+    // For kernel LDS it means lower as a standalone variable.
+    return !F;
   }
 
-  return false;
+  return Ret;
 }
 
 std::vector<GlobalVariable *>
-findVariablesToLower(Module &M,
-                     const SmallPtrSetImpl<GlobalValue *> &UsedList) {
+findVariablesToLower(Module &M, const SmallPtrSetImpl<GlobalValue *> &UsedList,
+                     const Function *F) {
   std::vector<llvm::GlobalVariable *> LocalVars;
   for (auto &GV : M.globals()) {
     if (GV.getType()->getPointerAddressSpace() != AMDGPUAS::LOCAL_ADDRESS) {
       continue;
     }
     if (!GV.hasInitializer()) {
       // addrspace(3) without initializer implies cuda/hip extern __shared__
       // the semantics for such a variable appears to be that all extern
       // __shared__ variables alias one another, in which case this transform
       // is not required
       continue;
     }
     if (!isa<UndefValue>(GV.getInitializer())) {
       // Initializers are unimplemented for local address space.
       // Leave such variables in place for consistent error reporting.
       continue;
     }
     if (GV.isConstant()) {
       // A constant undef variable can't be written to, and any load is
       // undef, so it should be eliminated by the optimizer. It could be
       // dropped by the back end if not. This pass skips over it.
       continue;
     }
-    if (std::none_of(GV.user_begin(), GV.user_end(), [&](User *U) {
-          return userRequiresLowering(UsedList, U);
-        })) {
+    if (!shouldLowerLDSToStruct(UsedList, GV, F)) {
       continue;
     }
     LocalVars.push_back(&GV);
   }
   return LocalVars;
 }
 
 SmallPtrSet<GlobalValue *, 32> getUsedList(Module &M) {

llvm/test/CodeGen/AMDGPU/32-bit-local-address-space.ll

define amdgpu_kernel void @mul_32bit_ptr(float addrspace(1)* %out, [3 x float] addrspace(3)* %lds, i32 %tid) {
   %val = load float, float addrspace(3)* %ptr
   store float %val, float addrspace(1)* %out
   ret void
 }
 
 @g_lds = addrspace(3) global float undef, align 4
 
 ; FUNC-LABEL: {{^}}infer_ptr_alignment_global_offset:
-; SI: v_mov_b32_e32 [[PTR:v[0-9]+]], g_lds@abs32@lo
+; SI: v_mov_b32_e32 [[PTR:v[0-9]+]], 0{{$}}
 ; SI: ds_read_b32 v{{[0-9]+}}, [[PTR]]
 define amdgpu_kernel void @infer_ptr_alignment_global_offset(float addrspace(1)* %out, i32 %tid) {
   %val = load float, float addrspace(3)* @g_lds
   store float %val, float addrspace(1)* %out
   ret void
 }
 
 

llvm/test/CodeGen/AMDGPU/GlobalISel/lds-global-value.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=bonaire -verify-machineinstrs < %s | FileCheck %s
 ; TODO: Replace with existing DAG tests
 
 @lds_512_4 = internal unnamed_addr addrspace(3) global [128 x i32] undef, align 4
 @lds_4_8 = addrspace(3) global i32 undef, align 8
 
 define amdgpu_kernel void @use_lds_globals(i32 addrspace(1)* %out, i32 addrspace(3)* %in) #0 {
 ; CHECK-LABEL: use_lds_globals:
 ; CHECK:       ; %bb.0: ; %entry
 ; CHECK-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
-; CHECK-NEXT:    v_mov_b32_e32 v0, 4
+; CHECK-NEXT:    v_mov_b32_e32 v0, 8
 ; CHECK-NEXT:    s_mov_b32 m0, -1
-; CHECK-NEXT:    ds_read_b32 v3, v0 offset:4
+; CHECK-NEXT:    ds_read_b32 v3, v0
 ; CHECK-NEXT:    v_mov_b32_e32 v2, 9
 ; CHECK-NEXT:    s_waitcnt lgkmcnt(0)
 ; CHECK-NEXT:    s_add_u32 s0, s0, 4
 ; CHECK-NEXT:    s_addc_u32 s1, s1, 0
 ; CHECK-NEXT:    v_mov_b32_e32 v0, s0
 ; CHECK-NEXT:    v_mov_b32_e32 v1, s1
 ; CHECK-NEXT:    flat_store_dword v[0:1], v3
 ; CHECK-NEXT:    v_mov_b32_e32 v0, 0

llvm/test/CodeGen/AMDGPU/GlobalISel/lds-relocs.ll

-; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -verify-machineinstrs -show-mc-encoding < %s | FileCheck -check-prefixes=GCN %s
+; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -amdgpu-enable-lower-module-lds=0 -verify-machineinstrs -show-mc-encoding < %s | FileCheck -check-prefixes=GCN %s
 ; FIXME: Merge with DAG test
 
 @lds.external = external unnamed_addr addrspace(3) global [0 x i32]
 @lds.defined = unnamed_addr addrspace(3) global [8 x i32] undef, align 8
 
 ; GCN-LABEL: {{^}}test_basic:
 ; GCN: s_add_u32 s0, lds.defined@abs32@lo, s0 ; encoding: [0xff,0x00,0x00,0x80,A,A,A,A]
 ; GCN: v_mov_b32_e32 v2, s0 ; encoding: [0x00,0x02,0x04,0x7e]

llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll

-; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --force-update
 ; RUN: llc -march=amdgcn -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX7LESS %s
 ; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX8 %s
 ; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX9 %s
 ; RUN: llc -march=amdgcn -mcpu=gfx1010 -mattr=-wavefrontsize32,+wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX10,GFX1064 %s
 ; RUN: llc -march=amdgcn -mcpu=gfx1010 -mattr=+wavefrontsize32,-wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX10,GFX1032 %s
 
 declare i32 @llvm.amdgcn.workitem.id.x()
 

 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s3, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB0_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
 ; GFX7LESS-NEXT:    s_mul_i32 s2, s2, 5
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:  BB0_2:
 ; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB0_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
 ; GFX8-NEXT:    s_mul_i32 s2, s2, 5
-; GFX8-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB0_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB0_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
 ; GFX9-NEXT:    s_mul_i32 s2, s2, 5
-; GFX9-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB0_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v1

 ; GFX1064-NEXT:    ; implicit-def: $vgpr1
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB0_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
-; GFX1064-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1064-NEXT:    s_mul_i32 s2, s2, 5
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB0_2:

 ; GFX1032-NEXT:    s_mov_b32 s3, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, s3, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB0_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s3, s3
-; GFX1032-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1032-NEXT:    s_mul_i32 s3, s3, 5
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, s3
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB0_2:

 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[6:7], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB1_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s1, s[2:3]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mul_i32 s1, s0, s1
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:  BB1_2:
 ; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[6:7]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[6:7], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB1_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s1, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_mul_i32 s1, s0, s1
-; GFX8-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB1_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[6:7]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[0:1], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB1_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s3, s[6:7]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_mul_i32 s3, s2, s3
-; GFX9-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB1_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[0:1]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mul_lo_u32 v0, s2, v0

 ; GFX1064-NEXT:    ; implicit-def: $vgpr1
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[0:1], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB1_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    s_bcnt1_i32_b64 s3, s[6:7]
-; GFX1064-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    s_mul_i32 s3, s2, s3
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, s3
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv

 ; GFX1032-NEXT:    s_mov_b32 s3, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, s3, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s0, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB1_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s1, s3
-; GFX1032-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    s_mul_i32 s1, s2, s1
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv

 }
 
 define amdgpu_kernel void @add_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: add_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_add_rtn_u32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB2_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_add_rtn_u32 v0, v0, v3
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB2_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB2_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_add_rtn_u32 v0, v0, v3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB2_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_writelane_b32 v3, s6, 48
 ; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB2_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v4, s7
 ; GFX1064-NEXT:    s_mov_b32 s3, s7
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_add_rtn_u32 v0, v0, v4
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB2_2:

 ; GFX1032-NEXT:    v_writelane_b32 v3, s3, 16
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB2_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v4, s4
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_add_rtn_u32 v0, v0, v4
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB2_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3
entry:
   %old = atomicrmw add i32 addrspace(3)* @local_var32, i32 %lane acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @add_i32_varying_nouse() {
 ; GFX7LESS-LABEL: add_i32_varying_nouse:
 ; GFX7LESS:       ; %bb.0: ; %entry
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_add_u32 v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: add_i32_varying_nouse:
 ; GFX8:       ; %bb.0: ; %entry

 ; GFX8-NEXT:    v_add_u32_dpp v1, vcc, v1, v1 row_bcast:31 row_mask:0xc bank_mask:0xf
 ; GFX8-NEXT:    v_readlane_b32 s2, v1, 63
 ; GFX8-NEXT:    s_mov_b64 exec, s[0:1]
 ; GFX8-NEXT:    s_mov_b32 s0, s2
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB3_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v2, s0
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_add_u32 v0, v2
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB3_2:
 ; GFX8-NEXT:    s_endpgm
 ;

 ; GFX9-NEXT:    v_add_u32_dpp v1, v1, v1 row_bcast:31 row_mask:0xc bank_mask:0xf
 ; GFX9-NEXT:    v_readlane_b32 s2, v1, 63
 ; GFX9-NEXT:    s_mov_b64 exec, s[0:1]
 ; GFX9-NEXT:    s_mov_b32 s0, s2
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB3_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_add_u32 v0, v2
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB3_2:
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: add_i32_varying_nouse:

 ; GFX1064-NEXT:    v_readlane_b32 s3, v1, 32
 ; GFX1064-NEXT:    s_mov_b64 exec, s[0:1]
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX1064-NEXT:    s_add_i32 s0, s2, s3
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB3_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v3, s0
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_add_u32 v0, v3
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB3_2:
 ; GFX1064-NEXT:    s_endpgm

 ; GFX1032-NEXT:    v_add_nc_u32_e32 v1, v1, v2
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s0
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v3, exec_lo, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v0, v1
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v3
 ; GFX1032-NEXT:    s_and_saveexec_b32 s0, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB3_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v3, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_add_u32 v3, v0
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB3_2:
 ; GFX1032-NEXT:    s_endpgm
 entry:

 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s4, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s5, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB4_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
-; GFX7LESS-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX7LESS-NEXT:    s_mul_i32 s4, s4, 5
-; GFX7LESS-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, s4
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7LESS-NEXT:    ds_add_rtn_u64 v[1:2], v3, v[1:2]
+; GFX7LESS-NEXT:    ds_add_rtn_u64 v[1:2], v2, v[1:2]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:  BB4_2:
 ; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX7LESS-NEXT:    v_readfirstlane_b32 s4, v2
 ; GFX7LESS-NEXT:    v_mul_hi_u32_u24_e32 v1, 5, v0
 ; GFX7LESS-NEXT:    v_mul_u32_u24_e32 v0, 5, v0

 ; GFX8-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB4_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
 ; GFX8-NEXT:    s_mul_i32 s4, s4, 5
 ; GFX8-NEXT:    v_mov_b32_e32 v1, s4
 ; GFX8-NEXT:    v_mov_b32_e32 v2, 0
-; GFX8-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    ds_add_rtn_u64 v[1:2], v3, v[1:2]
+; GFX8-NEXT:    ds_add_rtn_u64 v[1:2], v2, v[1:2]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB4_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX8-NEXT:    v_readfirstlane_b32 s3, v2
 ; GFX8-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX8-NEXT:    v_mov_b32_e32 v2, s3

 ; GFX9-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB4_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
 ; GFX9-NEXT:    s_mul_i32 s4, s4, 5
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s4
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
-; GFX9-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    ds_add_rtn_u64 v[1:2], v3, v[1:2]
+; GFX9-NEXT:    ds_add_rtn_u64 v[1:2], v2, v[1:2]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB4_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX9-NEXT:    v_readfirstlane_b32 s3, v2
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s3

 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, s5, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB4_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    s_mul_i32 s4, s4, 5
-; GFX1064-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, s4
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX1064-NEXT:    ds_add_rtn_u64 v[1:2], v3, v[1:2]
+; GFX1064-NEXT:    ds_add_rtn_u64 v[1:2], v2, v[1:2]
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB4_2:
 ; GFX1064-NEXT:    s_waitcnt_depctr 0xffe3
 ; GFX1064-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX1064-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX1064-NEXT:    v_readfirstlane_b32 s3, v2
 ; GFX1064-NEXT:    v_mad_u64_u32 v[0:1], s[2:3], v0, 5, s[2:3]

 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, s3, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB4_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s3, s3
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    s_mul_i32 s3, s3, 5
-; GFX1032-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX1032-NEXT:    ds_add_rtn_u64 v[1:2], v3, v[1:2]
+; GFX1032-NEXT:    ds_add_rtn_u64 v[1:2], v2, v[1:2]
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB4_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3
 ; GFX1032-NEXT:    s_or_b32 exec_lo, exec_lo, s2
 ; GFX1032-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX1032-NEXT:    v_readfirstlane_b32 s3, v2
 ; GFX1032-NEXT:    v_mad_u64_u32 v[0:1], s2, v0, 5, s[2:3]

 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s6, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s7, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB5_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]
-; GFX7LESS-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mul_i32 s7, s3, s6
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX7LESS-NEXT:    v_mul_hi_u32 v1, s2, v1
 ; GFX7LESS-NEXT:    s_mul_i32 s6, s2, s6
 ; GFX7LESS-NEXT:    v_add_i32_e32 v2, vcc, s7, v1
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1

 ; GFX8-NEXT:    s_cbranch_execz BB5_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]
 ; GFX8-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    v_mul_hi_u32 v1, s2, v1
 ; GFX8-NEXT:    s_mul_i32 s7, s3, s6
 ; GFX8-NEXT:    s_mul_i32 s6, s2, s6
-; GFX8-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX8-NEXT:    v_add_u32_e32 v2, vcc, s7, v1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_add_rtn_u64 v[1:2], v3, v[1:2]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB5_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[4:5]

 ; GFX9-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_mul_i32 s7, s3, s6
 ; GFX9-NEXT:    s_mul_hi_u32 s8, s2, s6
 ; GFX9-NEXT:    s_add_i32 s8, s8, s7
 ; GFX9-NEXT:    s_mul_i32 s6, s2, s6
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s8
-; GFX9-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_add_rtn_u64 v[1:2], v3, v[1:2]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB5_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mul_lo_u32 v3, s3, v0
 ; GFX9-NEXT:    v_mul_hi_u32 v4, s2, v0

 ; GFX1064-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB5_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]
-; GFX1064-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    s_mul_i32 s7, s3, s6
 ; GFX1064-NEXT:    s_mul_hi_u32 s8, s2, s6
 ; GFX1064-NEXT:    s_mul_i32 s6, s2, s6
 ; GFX1064-NEXT:    s_add_i32 s8, s8, s7
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, s8
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)

 ; GFX1032-NEXT:    s_mov_b32 s5, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, s5, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s4, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB5_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s5, s5
-; GFX1032-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    s_mul_i32 s6, s3, s5
 ; GFX1032-NEXT:    s_mul_hi_u32 s7, s2, s5
 ; GFX1032-NEXT:    s_mul_i32 s5, s2, s5
 ; GFX1032-NEXT:    s_add_i32 s7, s7, s6
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, s7
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)

 
 define amdgpu_kernel void @add_i64_varying(i64 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: add_i64_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7LESS-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7LESS-NEXT:    ds_add_rtn_u64 v[0:1], v2, v[0:1]
+; GFX7LESS-NEXT:    ds_add_rtn_u64 v[0:1], v1, v[0:1]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: add_i64_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
-; GFX8-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    ds_add_rtn_u64 v[0:1], v2, v[0:1]
+; GFX8-NEXT:    ds_add_rtn_u64 v[0:1], v1, v[0:1]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX8-NEXT:    s_mov_b32 s2, -1
 ; GFX8-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: add_i64_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    ds_add_rtn_u64 v[0:1], v2, v[0:1]
+; GFX9-NEXT:    ds_add_rtn_u64 v[0:1], v1, v[0:1]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s2, -1
 ; GFX9-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: add_i64_varying:
 ; GFX10:       ; %bb.0: ; %entry
 ; GFX10-NEXT:    v_mov_b32_e32 v1, 0
-; GFX10-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX10-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX10-NEXT:    s_mov_b32 s3, 0x31016000
 ; GFX10-NEXT:    s_mov_b32 s2, -1
 ; GFX10-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX10-NEXT:    ds_add_rtn_u64 v[0:1], v2, v[0:1]
+; GFX10-NEXT:    ds_add_rtn_u64 v[0:1], v1, v[0:1]
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    buffer_gl0_inv
 ; GFX10-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX10-NEXT:    s_endpgm
 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %zext = zext i32 %lane to i64
   %old = atomicrmw add i64 addrspace(3)* @local_var64, i64 %zext acq_rel

 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s3, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB7_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
 ; GFX7LESS-NEXT:    s_mul_i32 s2, s2, 5
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:  BB7_2:
 ; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB7_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
 ; GFX8-NEXT:    s_mul_i32 s2, s2, 5
-; GFX8-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB7_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB7_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
 ; GFX9-NEXT:    s_mul_i32 s2, s2, 5
-; GFX9-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB7_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v1

 ; GFX1064-NEXT:    ; implicit-def: $vgpr1
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB7_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
-; GFX1064-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1064-NEXT:    s_mul_i32 s2, s2, 5
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB7_2:

 ; GFX1032-NEXT:    s_mov_b32 s3, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, s3, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB7_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s3, s3
-; GFX1032-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1032-NEXT:    s_mul_i32 s3, s3, 5
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, s3
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB7_2:

 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[6:7], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB8_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s1, s[2:3]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mul_i32 s1, s0, s1
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:  BB8_2:
 ; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[6:7]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[6:7], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB8_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s1, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_mul_i32 s1, s0, s1
-; GFX8-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB8_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[6:7]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[0:1], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB8_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s3, s[6:7]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_mul_i32 s3, s2, s3
-; GFX9-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB8_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[0:1]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mul_lo_u32 v0, s2, v0

 ; GFX1064-NEXT:    ; implicit-def: $vgpr1
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[0:1], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB8_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    s_bcnt1_i32_b64 s3, s[6:7]
-; GFX1064-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    s_mul_i32 s3, s2, s3
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, s3
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv

 ; GFX1032-NEXT:    s_mov_b32 s3, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, s3, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s0, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB8_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s1, s3
-; GFX1032-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    s_mul_i32 s1, s2, s1
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv

 }
 
 define amdgpu_kernel void @sub_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: sub_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_sub_rtn_u32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB9_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_sub_rtn_u32 v0, v0, v3
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB9_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB9_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_sub_rtn_u32 v0, v0, v3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB9_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_writelane_b32 v3, s6, 48
 ; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB9_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v4, s7
 ; GFX1064-NEXT:    s_mov_b32 s3, s7
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_sub_rtn_u32 v0, v0, v4
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB9_2:

 ; GFX1032-NEXT:    v_writelane_b32 v3, s3, 16
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB9_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v4, s4
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_sub_rtn_u32 v0, v0, v4
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB9_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3
entry:
   %old = atomicrmw sub i32 addrspace(3)* @local_var32, i32 %lane acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @sub_i32_varying_nouse() {
 ; GFX7LESS-LABEL: sub_i32_varying_nouse:
 ; GFX7LESS:       ; %bb.0: ; %entry
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_sub_u32 v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: sub_i32_varying_nouse:
 ; GFX8:       ; %bb.0: ; %entry

 ; GFX8-NEXT:    v_add_u32_dpp v1, vcc, v1, v1 row_bcast:31 row_mask:0xc bank_mask:0xf
 ; GFX8-NEXT:    v_readlane_b32 s2, v1, 63
 ; GFX8-NEXT:    s_mov_b64 exec, s[0:1]
 ; GFX8-NEXT:    s_mov_b32 s0, s2
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB10_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v2, s0
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_sub_u32 v0, v2
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB10_2:
 ; GFX8-NEXT:    s_endpgm
 ;

 ; GFX9-NEXT:    v_add_u32_dpp v1, v1, v1 row_bcast:31 row_mask:0xc bank_mask:0xf
 ; GFX9-NEXT:    v_readlane_b32 s2, v1, 63
 ; GFX9-NEXT:    s_mov_b64 exec, s[0:1]
 ; GFX9-NEXT:    s_mov_b32 s0, s2
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB10_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_sub_u32 v0, v2
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB10_2:
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: sub_i32_varying_nouse:

 ; GFX1064-NEXT:    v_readlane_b32 s3, v1, 32
 ; GFX1064-NEXT:    s_mov_b64 exec, s[0:1]
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX1064-NEXT:    s_add_i32 s0, s2, s3
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB10_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v3, s0
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_sub_u32 v0, v3
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB10_2:
 ; GFX1064-NEXT:    s_endpgm

 ; GFX1032-NEXT:    v_add_nc_u32_e32 v1, v1, v2
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s0
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v3, exec_lo, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v0, v1
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v3
 ; GFX1032-NEXT:    s_and_saveexec_b32 s0, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB10_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v3, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_sub_u32 v3, v0
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB10_2:
 ; GFX1032-NEXT:    s_endpgm
 entry:

 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s4, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s5, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB11_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
-; GFX7LESS-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX7LESS-NEXT:    s_mul_i32 s4, s4, 5
-; GFX7LESS-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, s4
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7LESS-NEXT:    ds_sub_rtn_u64 v[1:2], v3, v[1:2]
+; GFX7LESS-NEXT:    ds_sub_rtn_u64 v[1:2], v2, v[1:2]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:  BB11_2:
 ; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX7LESS-NEXT:    v_readfirstlane_b32 s4, v2
 ; GFX7LESS-NEXT:    v_mul_hi_u32_u24_e32 v1, 5, v0
 ; GFX7LESS-NEXT:    v_mul_u32_u24_e32 v0, 5, v0

 ; GFX8-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB11_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
 ; GFX8-NEXT:    s_mul_i32 s4, s4, 5
 ; GFX8-NEXT:    v_mov_b32_e32 v1, s4
 ; GFX8-NEXT:    v_mov_b32_e32 v2, 0
-; GFX8-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    ds_sub_rtn_u64 v[1:2], v3, v[1:2]
+; GFX8-NEXT:    ds_sub_rtn_u64 v[1:2], v2, v[1:2]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB11_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    v_readfirstlane_b32 s3, v2
 ; GFX8-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX8-NEXT:    v_mul_hi_u32_u24_e32 v1, 5, v0
 ; GFX8-NEXT:    v_mul_u32_u24_e32 v0, 5, v0

 ; GFX9-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB11_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
 ; GFX9-NEXT:    s_mul_i32 s4, s4, 5
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s4
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
-; GFX9-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    ds_sub_rtn_u64 v[1:2], v3, v[1:2]
+; GFX9-NEXT:    ds_sub_rtn_u64 v[1:2], v2, v[1:2]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB11_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s3, v2
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX9-NEXT:    v_mul_hi_u32_u24_e32 v1, 5, v0
 ; GFX9-NEXT:    v_mul_u32_u24_e32 v0, 5, v0

 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, s5, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB11_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    s_mul_i32 s4, s4, 5
-; GFX1064-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, s4
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX1064-NEXT:    ds_sub_rtn_u64 v[1:2], v3, v[1:2]
+; GFX1064-NEXT:    ds_sub_rtn_u64 v[1:2], v2, v[1:2]
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB11_2:
 ; GFX1064-NEXT:    s_waitcnt_depctr 0xffe3
 ; GFX1064-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX1064-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX1064-NEXT:    v_mul_u32_u24_e32 v1, 5, v0
 ; GFX1064-NEXT:    v_readfirstlane_b32 s3, v2

 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, s3, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB11_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s3, s3
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    s_mul_i32 s3, s3, 5
-; GFX1032-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX1032-NEXT:    ds_sub_rtn_u64 v[1:2], v3, v[1:2]
+; GFX1032-NEXT:    ds_sub_rtn_u64 v[1:2], v2, v[1:2]
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB11_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3
 ; GFX1032-NEXT:    s_or_b32 exec_lo, exec_lo, s2
 ; GFX1032-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX1032-NEXT:    v_mul_u32_u24_e32 v1, 5, v0
 ; GFX1032-NEXT:    v_readfirstlane_b32 s3, v2

 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s6, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s7, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB12_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]
-; GFX7LESS-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mul_i32 s7, s3, s6
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX7LESS-NEXT:    v_mul_hi_u32 v1, s2, v1
 ; GFX7LESS-NEXT:    s_mul_i32 s6, s2, s6
 ; GFX7LESS-NEXT:    v_add_i32_e32 v2, vcc, s7, v1
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1

 ; GFX8-NEXT:    s_cbranch_execz BB12_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]
 ; GFX8-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    v_mul_hi_u32 v1, s2, v1
 ; GFX8-NEXT:    s_mul_i32 s7, s3, s6
 ; GFX8-NEXT:    s_mul_i32 s6, s2, s6
-; GFX8-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX8-NEXT:    v_add_u32_e32 v2, vcc, s7, v1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_sub_rtn_u64 v[1:2], v3, v[1:2]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB12_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[4:5]

 ; GFX9-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_mul_i32 s7, s3, s6
 ; GFX9-NEXT:    s_mul_hi_u32 s8, s2, s6
 ; GFX9-NEXT:    s_add_i32 s8, s8, s7
 ; GFX9-NEXT:    s_mul_i32 s6, s2, s6
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s8
-; GFX9-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_sub_rtn_u64 v[1:2], v3, v[1:2]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB12_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mul_lo_u32 v3, s3, v0
 ; GFX9-NEXT:    v_mul_hi_u32 v4, s2, v0

 ; GFX1064-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB12_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]
-; GFX1064-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    s_mul_i32 s7, s3, s6
 ; GFX1064-NEXT:    s_mul_hi_u32 s8, s2, s6
 ; GFX1064-NEXT:    s_mul_i32 s6, s2, s6
 ; GFX1064-NEXT:    s_add_i32 s8, s8, s7
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, s6
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, s8
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)

 ; GFX1032-NEXT:    s_mov_b32 s5, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, s5, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s4, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB12_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s5, s5
-; GFX1032-NEXT:    v_mov_b32_e32 v3, local_var64@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    s_mul_i32 s6, s3, s5
 ; GFX1032-NEXT:    s_mul_hi_u32 s7, s2, s5
 ; GFX1032-NEXT:    s_mul_i32 s5, s2, s5
 ; GFX1032-NEXT:    s_add_i32 s7, s7, s6
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, s7
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)

 
 define amdgpu_kernel void @sub_i64_varying(i64 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: sub_i64_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7LESS-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7LESS-NEXT:    ds_sub_rtn_u64 v[0:1], v2, v[0:1]
+; GFX7LESS-NEXT:    ds_sub_rtn_u64 v[0:1], v1, v[0:1]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: sub_i64_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
-; GFX8-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    ds_sub_rtn_u64 v[0:1], v2, v[0:1]
+; GFX8-NEXT:    ds_sub_rtn_u64 v[0:1], v1, v[0:1]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX8-NEXT:    s_mov_b32 s2, -1
 ; GFX8-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: sub_i64_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    ds_sub_rtn_u64 v[0:1], v2, v[0:1]
+; GFX9-NEXT:    ds_sub_rtn_u64 v[0:1], v1, v[0:1]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s2, -1
 ; GFX9-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: sub_i64_varying:
 ; GFX10:       ; %bb.0: ; %entry
 ; GFX10-NEXT:    v_mov_b32_e32 v1, 0
-; GFX10-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX10-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX10-NEXT:    s_mov_b32 s3, 0x31016000
 ; GFX10-NEXT:    s_mov_b32 s2, -1
 ; GFX10-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX10-NEXT:    ds_sub_rtn_u64 v[0:1], v2, v[0:1]
+; GFX10-NEXT:    ds_sub_rtn_u64 v[0:1], v1, v[0:1]
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    buffer_gl0_inv
 ; GFX10-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX10-NEXT:    s_endpgm
 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %zext = zext i32 %lane to i64
   %old = atomicrmw sub i64 addrspace(3)* @local_var64, i64 %zext acq_rel
   store i64 %old, i64 addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @and_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: and_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_and_rtn_b32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB14_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_and_rtn_b32 v0, v0, v3
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB14_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB14_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_and_rtn_b32 v0, v0, v3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB14_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_writelane_b32 v3, s6, 48
 ; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB14_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v4, s7
 ; GFX1064-NEXT:    s_mov_b32 s3, s7
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_and_rtn_b32 v0, v0, v4
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB14_2:

 ; GFX1032-NEXT:    v_writelane_b32 v3, s3, 16
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB14_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v4, s4
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_and_rtn_b32 v0, v0, v4
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB14_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

 }
 
 define amdgpu_kernel void @or_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: or_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_or_rtn_b32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB15_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_or_rtn_b32 v0, v0, v3
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB15_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB15_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_or_rtn_b32 v0, v0, v3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB15_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_writelane_b32 v3, s6, 48
 ; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB15_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v4, s7
 ; GFX1064-NEXT:    s_mov_b32 s3, s7
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_or_rtn_b32 v0, v0, v4
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB15_2:

 ; GFX1032-NEXT:    v_writelane_b32 v3, s3, 16
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB15_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v4, s4
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_or_rtn_b32 v0, v0, v4
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB15_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

 }
 
 define amdgpu_kernel void @xor_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: xor_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_xor_rtn_b32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB16_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_xor_rtn_b32 v0, v0, v3
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB16_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB16_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_xor_rtn_b32 v0, v0, v3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB16_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_writelane_b32 v3, s6, 48
 ; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB16_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v4, s7
 ; GFX1064-NEXT:    s_mov_b32 s3, s7
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_xor_rtn_b32 v0, v0, v4
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB16_2:

 ; GFX1032-NEXT:    v_writelane_b32 v3, s3, 16
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB16_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v4, s4
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_xor_rtn_b32 v0, v0, v4
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB16_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

 }
 
 define amdgpu_kernel void @max_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: max_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_max_rtn_i32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB17_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_max_rtn_i32 v0, v0, v3
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB17_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB17_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_max_rtn_i32 v0, v0, v3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB17_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_writelane_b32 v1, s6, 48
 ; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB17_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v4, s7
 ; GFX1064-NEXT:    s_mov_b32 s3, s7
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_max_rtn_i32 v0, v0, v4
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB17_2:

 ; GFX1032-NEXT:    v_writelane_b32 v1, s3, 16
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB17_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v4, s4
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_max_rtn_i32 v0, v0, v4
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB17_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, exec_hi, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB18_2
 ; GFX7LESS-NEXT:  ; %bb.1:
-; GFX7LESS-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_max_rtn_i64 v[0:1], v2, v[0:1]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:  BB18_2:
 ; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[2:3]

 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB18_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    v_mov_b32_e32 v0, 5
-; GFX8-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_max_rtn_i64 v[0:1], v2, v[0:1]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB18_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB18_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 5
-; GFX9-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_max_rtn_i64 v[0:1], v2, v[0:1]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB18_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB18_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    v_mov_b32_e32 v0, 5
-; GFX1064-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_max_rtn_i64 v[0:1], v2, v[0:1]
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB18_2:
 ; GFX1064-NEXT:    s_waitcnt_depctr 0xffe3

 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB18_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    v_mov_b32_e32 v0, 5
-; GFX1032-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_max_rtn_i64 v[0:1], v2, v[0:1]
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB18_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

 }
 
 define amdgpu_kernel void @min_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: min_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_min_rtn_i32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB19_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_min_rtn_i32 v0, v0, v3
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB19_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB19_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_min_rtn_i32 v0, v0, v3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB19_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_writelane_b32 v1, s6, 48
 ; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB19_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v4, s7
 ; GFX1064-NEXT:    s_mov_b32 s3, s7
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_min_rtn_i32 v0, v0, v4
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB19_2:

 ; GFX1032-NEXT:    v_writelane_b32 v1, s3, 16
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB19_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v4, s4
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_min_rtn_i32 v0, v0, v4
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB19_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, exec_hi, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB20_2
 ; GFX7LESS-NEXT:  ; %bb.1:
-; GFX7LESS-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_min_rtn_i64 v[0:1], v2, v[0:1]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:  BB20_2:
 ; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[2:3]

 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB20_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    v_mov_b32_e32 v0, 5
-; GFX8-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_min_rtn_i64 v[0:1], v2, v[0:1]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB20_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB20_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 5
-; GFX9-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_min_rtn_i64 v[0:1], v2, v[0:1]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB20_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s4, v0

 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB20_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    v_mov_b32_e32 v0, 5
-; GFX1064-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_min_rtn_i64 v[0:1], v2, v[0:1]
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB20_2:
 ; GFX1064-NEXT:    s_waitcnt_depctr 0xffe3

 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB20_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    v_mov_b32_e32 v0, 5
-; GFX1032-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_min_rtn_i64 v[0:1], v2, v[0:1]
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB20_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

 }
 
 define amdgpu_kernel void @umax_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: umax_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_max_rtn_u32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB21_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_max_rtn_u32 v0, v0, v3
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB21_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB21_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_max_rtn_u32 v0, v0, v3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB21_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_writelane_b32 v3, s6, 48
 ; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB21_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v4, s7
 ; GFX1064-NEXT:    s_mov_b32 s3, s7
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_max_rtn_u32 v0, v0, v4
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB21_2:

 ; GFX1032-NEXT:    v_writelane_b32 v3, s3, 16
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB21_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v4, s4
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_max_rtn_u32 v0, v0, v4
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB21_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, exec_hi, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB22_2
 ; GFX7LESS-NEXT:  ; %bb.1:
-; GFX7LESS-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_max_rtn_u64 v[0:1], v2, v[0:1]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:  BB22_2:
 ; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[2:3]

 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB22_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    v_mov_b32_e32 v0, 5
-; GFX8-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_max_rtn_u64 v[0:1], v2, v[0:1]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB22_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB22_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 5
-; GFX9-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_max_rtn_u64 v[0:1], v2, v[0:1]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB22_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB22_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    v_mov_b32_e32 v0, 5
-; GFX1064-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_max_rtn_u64 v[0:1], v2, v[0:1]
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB22_2:
 ; GFX1064-NEXT:    s_waitcnt_depctr 0xffe3

 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB22_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    v_mov_b32_e32 v0, 5
-; GFX1032-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_max_rtn_u64 v[0:1], v2, v[0:1]
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB22_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

 }
 
 define amdgpu_kernel void @umin_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: umin_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_min_rtn_u32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB23_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_min_rtn_u32 v0, v0, v3
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB23_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    v_mov_b32_dpp v1, v2 wave_shr:1 row_mask:0xf bank_mask:0xf
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB23_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_min_rtn_u32 v0, v0, v3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB23_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v0

 ; GFX1064-NEXT:    v_writelane_b32 v3, s6, 48
 ; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB23_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v4, s7
 ; GFX1064-NEXT:    s_mov_b32 s3, s7
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_min_rtn_u32 v0, v0, v4
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB23_2:

 ; GFX1032-NEXT:    v_writelane_b32 v3, s3, 16
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB23_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_mov_b32_e32 v0, local_var32@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v4, s4
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_min_rtn_u32 v0, v0, v4
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB23_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, exec_hi, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB24_2
 ; GFX7LESS-NEXT:  ; %bb.1:
-; GFX7LESS-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX7LESS-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_min_rtn_u64 v[0:1], v2, v[0:1]
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:  BB24_2:
 ; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[2:3]

 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB24_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    v_mov_b32_e32 v0, 5
-; GFX8-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    ds_min_rtn_u64 v[0:1], v2, v[0:1]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:  BB24_2:
 ; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB24_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 5
-; GFX9-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    ds_min_rtn_u64 v[0:1], v2, v[0:1]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:  BB24_2:
 ; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_readfirstlane_b32 s5, v1

 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB24_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    v_mov_b32_e32 v0, 5
-; GFX1064-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_min_rtn_u64 v[0:1], v2, v[0:1]
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:  BB24_2:
 ; GFX1064-NEXT:    s_waitcnt_depctr 0xffe3

 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB24_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    v_mov_b32_e32 v0, 5
-; GFX1032-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
+; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_min_rtn_u64 v[0:1], v2, v[0:1]
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:  BB24_2:
 ; GFX1032-NEXT:    s_waitcnt_depctr 0xffe3

llvm/test/CodeGen/AMDGPU/ds_read2.ll

 @sgemm.lA = internal unnamed_addr addrspace(3) global [264 x float] undef, align 4
 @sgemm.lB = internal unnamed_addr addrspace(3) global [776 x float] undef, align 4
 
 define amdgpu_kernel void @sgemm_inner_loop_read2_sequence(float addrspace(1)* %C, i32 %lda, i32 %ldb) #0 {
 ; CI-LABEL: sgemm_inner_loop_read2_sequence:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; CI-NEXT:    s_lshl_b32 s0, s2, 2
-; CI-NEXT:    s_add_i32 s1, s0, 0xc20
-; CI-NEXT:    s_addk_i32 s0, 0xc60
+; CI-NEXT:    s_add_i32 s1, s0, 0x8c40
+; CI-NEXT:    s_add_i32 s0, s0, 0x8c80
+; CI-NEXT:    v_mov_b32_e32 v2, s0
+; CI-NEXT:    v_lshlrev_b32_e32 v8, 2, v1
+; CI-NEXT:    s_mov_b32 s0, 0x8020
+; CI-NEXT:    v_add_i32_e32 v4, vcc, s0, v8
+; CI-NEXT:    s_mov_b32 s0, 0x80a0
+; CI-NEXT:    v_add_i32_e32 v6, vcc, s0, v8
 ; CI-NEXT:    v_mov_b32_e32 v0, s1
-; CI-NEXT:    v_mov_b32_e32 v4, s0
 ; CI-NEXT:    s_mov_b32 m0, -1
-; CI-NEXT:    ds_read2_b32 v[2:3], v0 offset1:1
+; CI-NEXT:    ds_read2_b32 v[0:1], v0 offset1:1
+; CI-NEXT:    ds_read2_b32 v[2:3], v2 offset1:1
 ; CI-NEXT:    ds_read2_b32 v[4:5], v4 offset1:1
-; CI-NEXT:    v_lshlrev_b32_e32 v8, 2, v1
-; CI-NEXT:    ds_read2_b32 v[0:1], v8 offset1:1
-; CI-NEXT:    ds_read2_b32 v[6:7], v8 offset0:32 offset1:33
-; CI-NEXT:    ds_read2_b32 v[8:9], v8 offset0:64 offset1:65
-; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    ds_read2_b32 v[6:7], v6 offset1:1
+; CI-NEXT:    v_add_i32_e32 v8, vcc, 0x8120, v8
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    v_add_f32_e32 v2, v2, v3
-; CI-NEXT:    v_add_f32_e32 v2, v2, v4
-; CI-NEXT:    v_add_f32_e32 v2, v2, v5
-; CI-NEXT:    v_add_f32_e32 v0, v2, v0
 ; CI-NEXT:    v_add_f32_e32 v0, v0, v1
+; CI-NEXT:    v_add_f32_e32 v0, v0, v2
+; CI-NEXT:    v_add_f32_e32 v0, v0, v3
+; CI-NEXT:    ds_read2_b32 v[8:9], v8 offset1:1
+; CI-NEXT:    v_add_f32_e32 v0, v0, v4
+; CI-NEXT:    v_add_f32_e32 v0, v0, v5
 ; CI-NEXT:    v_add_f32_e32 v0, v0, v6
 ; CI-NEXT:    v_add_f32_e32 v0, v0, v7
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    v_add_f32_e32 v0, v0, v8
+; CI-NEXT:    s_mov_b32 s7, 0xf000
 ; CI-NEXT:    s_mov_b32 s6, -1
 ; CI-NEXT:    v_add_f32_e32 v0, v0, v9
 ; CI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; CI-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: sgemm_inner_loop_read2_sequence:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_lshl_b32 s2, s2, 2
-; GFX9-NEXT:    s_add_i32 s3, s2, 0xc20
-; GFX9-NEXT:    s_addk_i32 s2, 0xc60
+; GFX9-NEXT:    s_add_i32 s3, s2, 0x8c40
+; GFX9-NEXT:    v_lshlrev_b32_e32 v8, 2, v1
+; GFX9-NEXT:    s_add_i32 s2, s2, 0x8c80
 ; GFX9-NEXT:    v_mov_b32_e32 v0, s3
-; GFX9-NEXT:    v_mov_b32_e32 v4, s2
-; GFX9-NEXT:    ds_read2_b32 v[2:3], v0 offset1:1
+; GFX9-NEXT:    v_mov_b32_e32 v2, s2
+; GFX9-NEXT:    v_add_u32_e32 v4, 0x8020, v8
+; GFX9-NEXT:    v_add_u32_e32 v6, 0x80a0, v8
+; GFX9-NEXT:    ds_read2_b32 v[0:1], v0 offset1:1
+; GFX9-NEXT:    ds_read2_b32 v[2:3], v2 offset1:1
 ; GFX9-NEXT:    ds_read2_b32 v[4:5], v4 offset1:1
-; GFX9-NEXT:    v_lshlrev_b32_e32 v8, 2, v1
-; GFX9-NEXT:    ds_read2_b32 v[0:1], v8 offset1:1
-; GFX9-NEXT:    ds_read2_b32 v[6:7], v8 offset0:32 offset1:33
-; GFX9-NEXT:    ds_read2_b32 v[8:9], v8 offset0:64 offset1:65
+; GFX9-NEXT:    ds_read2_b32 v[6:7], v6 offset1:1
+; GFX9-NEXT:    v_add_u32_e32 v8, 0x8120, v8
+; GFX9-NEXT:    s_waitcnt lgkmcnt(3)
+; GFX9-NEXT:    v_add_f32_e32 v0, v0, v1
+; GFX9-NEXT:    s_waitcnt lgkmcnt(2)
+; GFX9-NEXT:    v_add_f32_e32 v0, v0, v2
+; GFX9-NEXT:    v_add_f32_e32 v0, v0, v3
+; GFX9-NEXT:    ds_read2_b32 v[8:9], v8 offset1:1
+; GFX9-NEXT:    s_waitcnt lgkmcnt(2)
+; GFX9-NEXT:    v_add_f32_e32 v0, v0, v4
+; GFX9-NEXT:    v_add_f32_e32 v0, v0, v5
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    v_add_f32_e32 v2, v2, v3
-; GFX9-NEXT:    v_add_f32_e32 v2, v2, v4
-; GFX9-NEXT:    v_add_f32_e32 v2, v2, v5
-; GFX9-NEXT:    v_add_f32_e32 v0, v2, v0
-; GFX9-NEXT:    v_add_f32_e32 v0, v0, v1
 ; GFX9-NEXT:    v_add_f32_e32 v0, v0, v6
 ; GFX9-NEXT:    v_add_f32_e32 v0, v0, v7
 ; GFX9-NEXT:    v_add_f32_e32 v0, v0, v8
 ; GFX9-NEXT:    v_mov_b32_e32 v10, 0
 ; GFX9-NEXT:    v_add_f32_e32 v0, v0, v9
 ; GFX9-NEXT:    global_store_dword v10, v0, s[0:1]
 ; GFX9-NEXT:    s_endpgm
   %x.i = tail call i32 @llvm.amdgcn.workgroup.id.x() #1

 
 @v2i32_align1 = internal addrspace(3) global [100 x <2 x i32>] undef, align 1
 
 define amdgpu_kernel void @read2_v2i32_align1_odd_offset(<2 x i32> addrspace(1)* %out) {
 ; CI-LABEL: read2_v2i32_align1_odd_offset:
 ; CI:       ; %bb.0: ; %entry
 ; CI-NEXT:    v_mov_b32_e32 v0, 0
 ; CI-NEXT:    s_mov_b32 m0, -1
-; CI-NEXT:    ds_read_u8 v1, v0 offset:72
-; CI-NEXT:    ds_read_u8 v2, v0 offset:71
-; CI-NEXT:    ds_read_u8 v3, v0 offset:70
-; CI-NEXT:    ds_read_u8 v4, v0 offset:69
-; CI-NEXT:    ds_read_u8 v5, v0 offset:68
+; CI-NEXT:    ds_read_u8 v1, v0 offset:37032
+; CI-NEXT:    ds_read_u8 v2, v0 offset:37031
+; CI-NEXT:    ds_read_u8 v3, v0 offset:37030
+; CI-NEXT:    ds_read_u8 v4, v0 offset:37029
+; CI-NEXT:    ds_read_u8 v5, v0 offset:37028
 ; CI-NEXT:    s_waitcnt lgkmcnt(4)
 ; CI-NEXT:    v_lshlrev_b32_e32 v1, 8, v1
 ; CI-NEXT:    s_waitcnt lgkmcnt(3)
 ; CI-NEXT:    v_or_b32_e32 v1, v1, v2
 ; CI-NEXT:    s_waitcnt lgkmcnt(2)
 ; CI-NEXT:    v_lshlrev_b32_e32 v3, 8, v3
 ; CI-NEXT:    s_waitcnt lgkmcnt(1)
 ; CI-NEXT:    v_or_b32_e32 v3, v3, v4
 ; CI-NEXT:    v_lshlrev_b32_e32 v1, 16, v1
 ; CI-NEXT:    v_or_b32_e32 v1, v1, v3
-; CI-NEXT:    ds_read_u8 v2, v0 offset:67
-; CI-NEXT:    ds_read_u8 v3, v0 offset:66
-; CI-NEXT:    ds_read_u8 v0, v0 offset:65
+; CI-NEXT:    ds_read_u8 v2, v0 offset:37027
+; CI-NEXT:    ds_read_u8 v3, v0 offset:37026
+; CI-NEXT:    ds_read_u8 v0, v0 offset:37025
 ; CI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; CI-NEXT:    s_mov_b32 s3, 0xf000
 ; CI-NEXT:    s_mov_b32 s2, -1
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    v_lshlrev_b32_e32 v3, 8, v3
 ; CI-NEXT:    v_or_b32_e32 v0, v3, v0
 ; CI-NEXT:    v_lshlrev_b32_e32 v3, 8, v5
 ; CI-NEXT:    v_or_b32_e32 v2, v3, v2
 ; CI-NEXT:    v_lshlrev_b32_e32 v2, 16, v2
 ; CI-NEXT:    v_or_b32_e32 v0, v2, v0
 ; CI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; CI-NEXT:    s_endpgm
 ;
 ; GFX9-ALIGNED-LABEL: read2_v2i32_align1_odd_offset:
 ; GFX9-ALIGNED:       ; %bb.0: ; %entry
 ; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v2, 0
-; GFX9-ALIGNED-NEXT:    ds_read_u8 v0, v2 offset:65
-; GFX9-ALIGNED-NEXT:    ds_read_u8 v3, v2 offset:66
-; GFX9-ALIGNED-NEXT:    ds_read_u8 v4, v2 offset:67
-; GFX9-ALIGNED-NEXT:    ds_read_u8 v5, v2 offset:68
-; GFX9-ALIGNED-NEXT:    ds_read_u8 v1, v2 offset:69
-; GFX9-ALIGNED-NEXT:    ds_read_u8 v6, v2 offset:70
-; GFX9-ALIGNED-NEXT:    ds_read_u8 v7, v2 offset:71
-; GFX9-ALIGNED-NEXT:    ds_read_u8 v8, v2 offset:72
+; GFX9-ALIGNED-NEXT:    ds_read_u8 v0, v2 offset:37025
+; GFX9-ALIGNED-NEXT:    ds_read_u8 v3, v2 offset:37026
+; GFX9-ALIGNED-NEXT:    ds_read_u8 v4, v2 offset:37027
+; GFX9-ALIGNED-NEXT:    ds_read_u8 v5, v2 offset:37028
+; GFX9-ALIGNED-NEXT:    ds_read_u8 v1, v2 offset:37029
+; GFX9-ALIGNED-NEXT:    ds_read_u8 v6, v2 offset:37030
+; GFX9-ALIGNED-NEXT:    ds_read_u8 v7, v2 offset:37031
+; GFX9-ALIGNED-NEXT:    ds_read_u8 v8, v2 offset:37032
 ; GFX9-ALIGNED-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX9-ALIGNED-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-ALIGNED-NEXT:    v_lshlrev_b32_e32 v3, 8, v3
 ; GFX9-ALIGNED-NEXT:    v_lshlrev_b32_e32 v6, 8, v6
 ; GFX9-ALIGNED-NEXT:    v_or_b32_e32 v1, v6, v1
 ; GFX9-ALIGNED-NEXT:    v_lshlrev_b32_e32 v6, 8, v8
 ; GFX9-ALIGNED-NEXT:    v_or_b32_e32 v0, v3, v0
 ; GFX9-ALIGNED-NEXT:    v_lshlrev_b32_e32 v3, 8, v5
 ; GFX9-ALIGNED-NEXT:    v_or_b32_sdwa v6, v6, v7 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; GFX9-ALIGNED-NEXT:    v_or_b32_sdwa v3, v3, v4 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; GFX9-ALIGNED-NEXT:    v_or_b32_e32 v1, v6, v1
 ; GFX9-ALIGNED-NEXT:    v_or_b32_e32 v0, v3, v0
 ; GFX9-ALIGNED-NEXT:    global_store_dwordx2 v2, v[0:1], s[0:1]
 ; GFX9-ALIGNED-NEXT:    s_endpgm
 ;
 ; GFX9-UNALIGNED-LABEL: read2_v2i32_align1_odd_offset:
 ; GFX9-UNALIGNED:       ; %bb.0: ; %entry
-; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v0, 0x41
+; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v0, 0x90a1
 ; GFX9-UNALIGNED-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX9-UNALIGNED-NEXT:    ds_read2_b32 v[0:1], v0 offset1:1
 ; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-UNALIGNED-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-UNALIGNED-NEXT:    global_store_dwordx2 v2, v[0:1], s[0:1]
 ; GFX9-UNALIGNED-NEXT:    s_endpgm
 entry:
   %load = load <2 x i32>, <2 x i32> addrspace(3)* bitcast (i8 addrspace(3)* getelementptr (i8, i8 addrspace(3)* bitcast ([100 x <2 x i32>] addrspace(3)* @v2i32_align1 to i8 addrspace(3)*), i32 65) to <2 x i32> addrspace(3)*), align 1

llvm/test/CodeGen/AMDGPU/ds_write2.ll

-; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --force-update
 ; RUN: llc -mtriple=amdgcn--amdpal -mcpu=bonaire -verify-machineinstrs -mattr=+load-store-opt < %s | FileCheck -enable-var-scope --check-prefix=CI %s
 ; RUN: llc -mtriple=amdgcn--amdpal -mcpu=gfx900 -verify-machineinstrs -mattr=+load-store-opt,-unaligned-access-mode < %s | FileCheck -enable-var-scope -check-prefixes=GFX9,GFX9-ALIGNED %s
 ; RUN: llc -mtriple=amdgcn--amdpal -mcpu=gfx900 -verify-machineinstrs -mattr=+load-store-opt,+unaligned-access-mode < %s | FileCheck -enable-var-scope -check-prefixes=GFX9,GFX9-UNALIGNED %s
 
 @lds = addrspace(3) global [512 x float] undef, align 4
 @lds.f64 = addrspace(3) global [512 x double] undef, align 8
 
 define amdgpu_kernel void @simple_write2_one_val_f32(float addrspace(1)* %C, float addrspace(1)* %in) #0 {

 @sgemm.lA = internal unnamed_addr addrspace(3) global [264 x float] undef, align 4
 @sgemm.lB = internal unnamed_addr addrspace(3) global [776 x float] undef, align 4
 
 define amdgpu_kernel void @write2_sgemm_sequence(float addrspace(1)* %C, i32 %lda, i32 %ldb, float addrspace(1)* %in) #0 {
 ; CI-LABEL: write2_sgemm_sequence:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
 ; CI-NEXT:    s_lshl_b32 s2, s2, 2
-; CI-NEXT:    s_add_i32 s3, s2, 0xc20
+; CI-NEXT:    s_add_i32 s3, s2, 0x8c40
 ; CI-NEXT:    v_mov_b32_e32 v0, s3
-; CI-NEXT:    s_addk_i32 s2, 0xc60
+; CI-NEXT:    s_add_i32 s2, s2, 0x8c80
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    s_load_dword s0, s[0:1], 0x0
 ; CI-NEXT:    s_mov_b32 m0, -1
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    v_mov_b32_e32 v2, s0
 ; CI-NEXT:    v_mov_b32_e32 v3, s0
 ; CI-NEXT:    ds_write2_b32 v0, v2, v3 offset1:1
 ; CI-NEXT:    v_mov_b32_e32 v0, s2
 ; CI-NEXT:    ds_write2_b32 v0, v2, v3 offset1:1
 ; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v1
+; CI-NEXT:    v_add_i32_e32 v1, vcc, 0x8020, v0
+; CI-NEXT:    ds_write2_b32 v1, v2, v3 offset1:1
+; CI-NEXT:    v_add_i32_e32 v1, vcc, 0x80a0, v0
+; CI-NEXT:    v_add_i32_e32 v0, vcc, 0x8120, v0
+; CI-NEXT:    ds_write2_b32 v1, v2, v3 offset1:1
 ; CI-NEXT:    ds_write2_b32 v0, v2, v3 offset1:1
-; CI-NEXT:    ds_write2_b32 v0, v2, v3 offset0:32 offset1:33
-; CI-NEXT:    ds_write2_b32 v0, v2, v3 offset0:64 offset1:65
 ; CI-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: write2_sgemm_sequence:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-NEXT:    s_lshl_b32 s2, s2, 2
-; GFX9-NEXT:    s_add_i32 s3, s2, 0xc20
-; GFX9-NEXT:    s_addk_i32 s2, 0xc60
+; GFX9-NEXT:    s_add_i32 s3, s2, 0x8c40
+; GFX9-NEXT:    s_add_i32 s2, s2, 0x8c80
 ; GFX9-NEXT:    v_mov_b32_e32 v0, s3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s0
 ; GFX9-NEXT:    v_mov_b32_e32 v4, s0
 ; GFX9-NEXT:    ds_write2_b32 v0, v3, v4 offset1:1
 ; GFX9-NEXT:    ds_write2_b32 v2, v3, v4 offset1:1
 ; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v1
+; GFX9-NEXT:    v_add_u32_e32 v1, 0x8020, v0
+; GFX9-NEXT:    ds_write2_b32 v1, v3, v4 offset1:1
+; GFX9-NEXT:    v_add_u32_e32 v1, 0x80a0, v0
+; GFX9-NEXT:    v_add_u32_e32 v0, 0x8120, v0
+; GFX9-NEXT:    ds_write2_b32 v1, v3, v4 offset1:1
 ; GFX9-NEXT:    ds_write2_b32 v0, v3, v4 offset1:1
-; GFX9-NEXT:    ds_write2_b32 v0, v3, v4 offset0:32 offset1:33
-; GFX9-NEXT:    ds_write2_b32 v0, v3, v4 offset0:64 offset1:65
 ; GFX9-NEXT:    s_endpgm
   %x.i = tail call i32 @llvm.amdgcn.workgroup.id.x() #1
   %y.i = tail call i32 @llvm.amdgcn.workitem.id.y() #1
   %val = load float, float addrspace(1)* %in
   %arrayidx44 = getelementptr inbounds [264 x float], [264 x float] addrspace(3)* @sgemm.lA, i32 0, i32 %x.i
   store float %val, float addrspace(3)* %arrayidx44, align 4
   %add47 = add nsw i32 %x.i, 1
   %arrayidx48 = getelementptr inbounds [264 x float], [264 x float] addrspace(3)* @sgemm.lA, i32 0, i32 %add47

 @v2i32_align1 = internal addrspace(3) global [100 x <2 x i32>] undef, align 1
 
 define amdgpu_kernel void @write2_v2i32_align1_odd_offset() {
 ; CI-LABEL: write2_v2i32_align1_odd_offset:
 ; CI:       ; %bb.0: ; %entry
 ; CI-NEXT:    v_mov_b32_e32 v0, 0x7b
 ; CI-NEXT:    v_mov_b32_e32 v1, 0
 ; CI-NEXT:    s_mov_b32 m0, -1
-; CI-NEXT:    ds_write_b8 v1, v0 offset:65
+; CI-NEXT:    ds_write_b8 v1, v0 offset:37025
 ; CI-NEXT:    v_mov_b32_e32 v0, 1
-; CI-NEXT:    ds_write_b8 v1, v0 offset:70
+; CI-NEXT:    ds_write_b8 v1, v0 offset:37030
 ; CI-NEXT:    v_mov_b32_e32 v0, 0xc8
-; CI-NEXT:    ds_write_b8 v1, v0 offset:69
-; CI-NEXT:    ds_write_b8 v1, v1 offset:68
-; CI-NEXT:    ds_write_b8 v1, v1 offset:67
-; CI-NEXT:    ds_write_b8 v1, v1 offset:66
-; CI-NEXT:    ds_write_b8 v1, v1 offset:72
-; CI-NEXT:    ds_write_b8 v1, v1 offset:71
+; CI-NEXT:    ds_write_b8 v1, v0 offset:37029
+; CI-NEXT:    ds_write_b8 v1, v1 offset:37028
+; CI-NEXT:    ds_write_b8 v1, v1 offset:37027
+; CI-NEXT:    ds_write_b8 v1, v1 offset:37026
+; CI-NEXT:    ds_write_b8 v1, v1 offset:37032
+; CI-NEXT:    ds_write_b8 v1, v1 offset:37031
 ; CI-NEXT:    s_endpgm
 ;
 ; GFX9-ALIGNED-LABEL: write2_v2i32_align1_odd_offset:
 ; GFX9-ALIGNED:       ; %bb.0: ; %entry
 ; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v0, 0x7b
 ; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v0 offset:65
+; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v0 offset:37025
 ; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v0, 1
-; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v0 offset:70
+; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v0 offset:37030
 ; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v0, 0xc8
-; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v0 offset:69
-; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:68
-; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:67
-; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:66
-; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:72
-; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:71
+; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v0 offset:37029
+; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:37028
+; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:37027
+; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:37026
+; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:37032
+; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:37031
 ; GFX9-ALIGNED-NEXT:    s_endpgm
 ;
 ; GFX9-UNALIGNED-LABEL: write2_v2i32_align1_odd_offset:
 ; GFX9-UNALIGNED:       ; %bb.0: ; %entry
-; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v0, 0x41
+; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v0, 0x90a1
 ; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v1, 0x7b
 ; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v2, 0x1c8
 ; GFX9-UNALIGNED-NEXT:    ds_write2_b32 v0, v1, v2 offset1:1
 ; GFX9-UNALIGNED-NEXT:    s_endpgm
 entry:
   store <2 x i32> <i32 123, i32 456>, <2 x i32> addrspace(3)* bitcast (i8 addrspace(3)* getelementptr (i8, i8 addrspace(3)* bitcast ([100 x <2 x i32>] addrspace(3)* @v2i32_align1 to i8 addrspace(3)*), i32 65) to <2 x i32> addrspace(3)*), align 1
   ret void
 }

llvm/test/CodeGen/AMDGPU/lds-alignment.ll

define amdgpu_kernel void @test_round_size_2(i8 addrspace(1)* %out, i8 addrspace(1)* %in) #1 {
 
   %lds.align16.1.bc = bitcast [38 x i8] addrspace(3)* @lds.align16.1 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 4 %lds.align16.1.bc, i8 addrspace(1)* align 4 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 4 %out, i8 addrspace(3)* align 4 %lds.align16.1.bc, i32 38, i1 false)
 
   ret void
 }
 
-; 38 + (10 pad) + 38
+; 38 + (2 pad) + 38
 ; HSA-LABEL: {{^}}test_round_size_2_align_8:
-; HSA: workgroup_group_segment_byte_size = 86
+; HSA: workgroup_group_segment_byte_size = 78
 ; HSA: group_segment_alignment = 4
 define amdgpu_kernel void @test_round_size_2_align_8(i8 addrspace(1)* %out, i8 addrspace(1)* %in) #1 {
   %lds.align16.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align16.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align16.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align16.0.bc, i32 38, i1 false)
 
   %lds.align8.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align8.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align8.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)

 ; HSA: workgroup_group_segment_byte_size = 0
 ; HSA: group_segment_alignment = 4
 define amdgpu_kernel void @test_high_align_lds_arg(i8 addrspace(1)* %out, i8 addrspace(1)* %in, i8 addrspace(3)* align 64 %lds.arg) #1 {
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 64 %lds.arg, i8 addrspace(1)* align 64 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 64 %out, i8 addrspace(3)* align 64 %lds.arg, i32 38, i1 false)
   ret void
 }
 
-; (7 * 8) + (39 * 4) = 212
+; FIXME: missign alignment can be improved.
+; (39 * 4) + (4 pad) + (7 * 8) = 216
 ; HSA-LABEL: {{^}}test_missing_alignment_size_2_order0:
-; HSA: workgroup_group_segment_byte_size = 212
+; HSA: workgroup_group_segment_byte_size = 216
 ; HSA: group_segment_alignment = 4
 define amdgpu_kernel void @test_missing_alignment_size_2_order0(i8 addrspace(1)* %out, i8 addrspace(1)* %in) #1 {
   %lds.missing.align.0.bc = bitcast [39 x i32] addrspace(3)* @lds.missing.align.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 4 %lds.missing.align.0.bc, i8 addrspace(1)* align 4 %in, i32 160, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 4 %out, i8 addrspace(3)* align 4 %lds.missing.align.0.bc, i32 160, i1 false)
 
   %lds.missing.align.1.bc = bitcast [7 x i64] addrspace(3)* @lds.missing.align.1 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.missing.align.1.bc, i8 addrspace(1)* align 8 %in, i32 56, i1 false)
define amdgpu_kernel void @test_missing_alignment_size_2_order1(i8 addrspace(1)* %out, i8 addrspace(1)* %in) #1 {
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.missing.align.1.bc, i32 56, i1 false)
 
   %lds.missing.align.0.bc = bitcast [39 x i32] addrspace(3)* @lds.missing.align.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 4 %lds.missing.align.0.bc, i8 addrspace(1)* align 4 %in, i32 160, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 4 %out, i8 addrspace(3)* align 4 %lds.missing.align.0.bc, i32 160, i1 false)
 
   ret void
 }
-; Test how the size needed for padding changes based on when the
-; global is encountered during lowering. There should be a consistent
-; order to minimize padding waste.
-;
-; The way global addresses are lowered now, this is in inverse of
-; first use order which isn't great.
-;
-; This should be the optimal order for these globals. If sorted to
-; minimize padding, the minimum possible size is: align 32, align 8,
-; align 16
-
 
 ; align 32, 16, 8
-; 38 + (10 pad) + 38 + (10 pad) + 38 = 134
+; 38 + (10 pad) + 38 + (2 pad) + 38 = 126
 ; HSA-LABEL: {{^}}test_round_size_3_order0:
-; HSA: workgroup_group_segment_byte_size = 134
+; HSA: workgroup_group_segment_byte_size = 126
 ; HSA: group_segment_alignment = 4
 define amdgpu_kernel void @test_round_size_3_order0(i8 addrspace(1)* %out, i8 addrspace(1)* %in) #1 {
   %lds.align32.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align32.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align32.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align32.0.bc, i32 38, i1 false)
 
   %lds.align16.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align16.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align16.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align16.0.bc, i32 38, i1 false)
 
   %lds.align8.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align8.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align8.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align8.0.bc, i32 38, i1 false)
 
   ret void
 }
 
 ; align 32, 8, 16
-; 38 (+ 2 pad) + 38 + (18 pad) + 38 = 134
+; 38 (+ 10 pad) + 38 + (2 pad) + 38 = 126
 ; HSA-LABEL: {{^}}test_round_size_3_order1:
-; HSA: workgroup_group_segment_byte_size = 134
+; HSA: workgroup_group_segment_byte_size = 126
 ; HSA: group_segment_alignment = 4
 define amdgpu_kernel void @test_round_size_3_order1(i8 addrspace(1)* %out, i8 addrspace(1)* %in) #1 {
   %lds.align32.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align32.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align32.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align32.0.bc, i32 38, i1 false)
 
   %lds.align8.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align8.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align8.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align8.0.bc, i32 38, i1 false)
 
   %lds.align16.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align16.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align16.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align16.0.bc, i32 38, i1 false)
 
   ret void
 }
 
 ; align 16, 32, 8
-; 38 + (26 pad) + 38 + (10 pad) + 38 = 150
+; 38 + (10 pad) + 38 + (2 pad) + 38 = 126
 ; HSA-LABEL: {{^}}test_round_size_3_order2:
-; HSA: workgroup_group_segment_byte_size = 150
+; HSA: workgroup_group_segment_byte_size = 126
 ; HSA: group_segment_alignment = 4
 define amdgpu_kernel void @test_round_size_3_order2(i8 addrspace(1)* %out, i8 addrspace(1)* %in) #1 {
   %lds.align16.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align16.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align16.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align16.0.bc, i32 38, i1 false)
 
   %lds.align32.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align32.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align32.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align32.0.bc, i32 38, i1 false)
 
   %lds.align8.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align8.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align8.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align8.0.bc, i32 38, i1 false)
 
   ret void
 }
 
+; FIXME: Improve alignment
 ; align 16, 8, 32
-; 38 + (2 pad) + 38 + (2 pad) + 38
+; 38 + (10 pad) + 38 + (2 pad) + 38
 ; HSA-LABEL: {{^}}test_round_size_3_order3:
-; HSA: workgroup_group_segment_byte_size = 118
+; HSA: workgroup_group_segment_byte_size = 126
 ; HSA: group_segment_alignment = 4
 define amdgpu_kernel void @test_round_size_3_order3(i8 addrspace(1)* %out, i8 addrspace(1)* %in) #1 {
   %lds.align16.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align16.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align16.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align16.0.bc, i32 38, i1 false)
 
   %lds.align8.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align8.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align8.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align8.0.bc, i32 38, i1 false)
 
   %lds.align32.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align32.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align32.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align32.0.bc, i32 38, i1 false)
 
   ret void
 }
 
 ; align 8, 32, 16
-; 38 + (26 pad) + 38 + (2 pad) + 38 = 142
+; 38 + (10 pad) + 38 + (2 pad) + 38 = 126
 ; HSA-LABEL: {{^}}test_round_size_3_order4:
-; HSA: workgroup_group_segment_byte_size = 142
+; HSA: workgroup_group_segment_byte_size = 126
 ; HSA: group_segment_alignment = 4
 define amdgpu_kernel void @test_round_size_3_order4(i8 addrspace(1)* %out, i8 addrspace(1)* %in) #1 {
   %lds.align8.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align8.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align8.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)
   call void @llvm.memcpy.p1i8.p3i8.i32(i8 addrspace(1)* align 8 %out, i8 addrspace(3)* align 8 %lds.align8.0.bc, i32 38, i1 false)
 
   %lds.align32.0.bc = bitcast [38 x i8] addrspace(3)* @lds.align32.0 to i8 addrspace(3)*
   call void @llvm.memcpy.p3i8.p1i8.i32(i8 addrspace(3)* align 8 %lds.align32.0.bc, i8 addrspace(1)* align 8 %in, i32 38, i1 false)

llvm/test/CodeGen/AMDGPU/lds-relocs.ll

-; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -verify-machineinstrs -show-mc-encoding < %s | FileCheck -check-prefixes=GCN %s
-; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -filetype=obj < %s | llvm-readobj -r -t - | FileCheck -check-prefixes=ELF %s
+; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -amdgpu-enable-lower-module-lds=0 -verify-machineinstrs -show-mc-encoding < %s | FileCheck -check-prefixes=GCN %s
+; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -amdgpu-enable-lower-module-lds=0 -filetype=obj < %s | llvm-readobj -r -t - | FileCheck -check-prefixes=ELF %s
 
 @lds.external = external unnamed_addr addrspace(3) global [0 x i32]
 @lds.defined = unnamed_addr addrspace(3) global [8 x i32] undef, align 8
 
 ; ELF:      Relocations [
 ; ELF-NEXT:   Section (3) .rel.text {
 ; ELF-NEXT:     0x{{[0-9a-f]*}} R_AMDGPU_ABS32 lds.external
 ; ELF-NEXT:     0x{{[0-9a-f]*}} R_AMDGPU_ABS32 lds.defined

llvm/test/CodeGen/AMDGPU/llvm.amdgcn.atomic.dec.ll

 }
 
 @lds0 = addrspace(3) global [512 x i32] undef
 
 ; GCN-LABEL: {{^}}atomic_dec_shl_base_lds_0:
 ; CIVI-DAG: s_mov_b32 m0
 ; GFX9-NOT: m0
 
-; CIVI-DAG: v_lshlrev_b32_e32 [[OFS:v[0-9]+]], 2, {{v[0-9]+}}
-; CIVI-DAG: v_add_{{[ui]}}32_e32 [[PTR:v[0-9]+]], vcc, lds0@abs32@lo, [[OFS]]
-; GFX9-DAG: s_mov_b32 [[BASE:s[0-9]+]], lds0@abs32@lo
-; GFX9-DAG: v_lshl_add_u32 [[PTR:v[0-9]+]], {{v[0-9]+}}, 2, [[BASE]]
-
+; GCN-DAG:  v_lshlrev_b32_e32 [[PTR:v[0-9]+]], 2, {{v[0-9]+}}
 ; GCN: ds_dec_rtn_u32 {{v[0-9]+}}, [[PTR]], {{v[0-9]+}} offset:8
 define amdgpu_kernel void @atomic_dec_shl_base_lds_0(i32 addrspace(1)* %out, i32 addrspace(1)* %add_use) #0 {
   %tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
   %idx.0 = add nsw i32 %tid.x, 2
   %arrayidx0 = getelementptr inbounds [512 x i32], [512 x i32] addrspace(3)* @lds0, i32 0, i32 %idx.0
   %val0 = call i32 @llvm.amdgcn.atomic.dec.i32.p3i32(i32 addrspace(3)* %arrayidx0, i32 9, i32 0, i32 0, i1 false)
   store i32 %idx.0, i32 addrspace(1)* %add_use
   store i32 %val0, i32 addrspace(1)* %out

 }
 
 @lds1 = addrspace(3) global [512 x i64] undef, align 8
 
 ; GCN-LABEL: {{^}}atomic_dec_shl_base_lds_0_i64:
 ; CIVI-DAG: s_mov_b32 m0
 ; GFX9-NOT: m0
 
-; CIVI-DAG: v_lshlrev_b32_e32 [[OFS:v[0-9]+]], 3, {{v[0-9]+}}
-; CIVI-DAG: v_add_{{[ui]}}32_e32 [[PTR:v[0-9]+]], vcc, lds1@abs32@lo, [[OFS]]
-; GFX9-DAG: v_mov_b32_e32 [[BASE:v[0-9]+]], lds1@abs32@lo
-; GFX9-DAG: v_lshl_add_u32 [[PTR:v[0-9]+]], {{v[0-9]+}}, 3, [[BASE]]
-
+; GCN-DAG: v_lshlrev_b32_e32 [[PTR:v[0-9]+]], 3, {{v[0-9]+}}
 ; GCN: ds_dec_rtn_u64 v{{\[[0-9]+:[0-9]+\]}}, [[PTR]], v{{\[[0-9]+:[0-9]+\]}} offset:16
 define amdgpu_kernel void @atomic_dec_shl_base_lds_0_i64(i64 addrspace(1)* %out, i32 addrspace(1)* %add_use) #0 {
   %tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
   %idx.0 = add nsw i32 %tid.x, 2
   %arrayidx0 = getelementptr inbounds [512 x i64], [512 x i64] addrspace(3)* @lds1, i32 0, i32 %idx.0
   %val0 = call i64 @llvm.amdgcn.atomic.dec.i64.p3i64(i64 addrspace(3)* %arrayidx0, i64 9, i32 0, i32 0, i1 false)
   store i32 %idx.0, i32 addrspace(1)* %add_use
   store i64 %val0, i64 addrspace(1)* %out
   ret void
 }
 
 attributes #0 = { nounwind }
 attributes #1 = { nounwind readnone }
 attributes #2 = { nounwind argmemonly }

llvm/test/CodeGen/AMDGPU/llvm.amdgcn.atomic.inc.ll

define amdgpu_kernel void @global_atomic_inc_noret_i32_offset_addr64(i32 addrspace(1)* %ptr) #0 {
   %gep = getelementptr i32, i32 addrspace(1)* %gep.tid, i32 5
   %result = call i32 @llvm.amdgcn.atomic.inc.i32.p1i32(i32 addrspace(1)* %gep, i32 42, i32 0, i32 0, i1 false)
   ret void
 }
 
 @lds0 = addrspace(3) global [512 x i32] undef, align 4
 
 ; GCN-LABEL: {{^}}atomic_inc_shl_base_lds_0_i32:
-; CIVI-DAG: v_lshlrev_b32_e32 [[OFS:v[0-9]+]], 2, {{v[0-9]+}}
-; CIVI-DAG: v_add_{{[ui]}}32_e32 [[PTR:v[0-9]+]], vcc, lds0@abs32@lo, [[OFS]]
-; GFX9-DAG: s_mov_b32 [[BASE:s[0-9]+]], lds0@abs32@lo
-; GFX9-DAG: v_lshl_add_u32 [[PTR:v[0-9]+]], {{v[0-9]+}}, 2, [[BASE]]
+; GCN: v_lshlrev_b32_e32 [[PTR:v[0-9]+]], 2, {{v[0-9]+}}
 ; GCN: ds_inc_rtn_u32 {{v[0-9]+}}, [[PTR]], {{v[0-9]+}} offset:8
 define amdgpu_kernel void @atomic_inc_shl_base_lds_0_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %add_use) #0 {
   %tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
   %idx.0 = add nsw i32 %tid.x, 2
   %arrayidx0 = getelementptr inbounds [512 x i32], [512 x i32] addrspace(3)* @lds0, i32 0, i32 %idx.0
   %val0 = call i32 @llvm.amdgcn.atomic.inc.i32.p3i32(i32 addrspace(3)* %arrayidx0, i32 9, i32 0, i32 0, i1 false)
   store i32 %idx.0, i32 addrspace(1)* %add_use
   store i32 %val0, i32 addrspace(1)* %out
define amdgpu_kernel void @flat_atomic_inc_noret_i32_offset_addr64(i32* %ptr) #0 {
   %gep = getelementptr i32, i32* %gep.tid, i32 5
   %result = call i32 @llvm.amdgcn.atomic.inc.i32.p0i32(i32* %gep, i32 42, i32 0, i32 0, i1 false)
   ret void
 }
 
 @lds1 = addrspace(3) global [512 x i64] undef, align 8
 
 ; GCN-LABEL: {{^}}atomic_inc_shl_base_lds_0_i64:
-; CIVI-DAG: v_lshlrev_b32_e32 [[OFS:v[0-9]+]], 3, {{v[0-9]+}}
-; CIVI-DAG: v_add_{{[ui]}}32_e32 [[PTR:v[0-9]+]], vcc, lds1@abs32@lo, [[OFS]]
-; GFX9-DAG: v_mov_b32_e32 [[BASE:v[0-9]+]], lds1@abs32@lo
-; GFX9-DAG: v_lshl_add_u32 [[PTR:v[0-9]+]], {{v[0-9]+}}, 3, [[BASE]]
+; GCN: v_lshlrev_b32_e32 [[PTR:v[0-9]+]], 3, {{v[0-9]+}}
 ; GCN: ds_inc_rtn_u64 v{{\[[0-9]+:[0-9]+\]}}, [[PTR]], v{{\[[0-9]+:[0-9]+\]}} offset:16
 define amdgpu_kernel void @atomic_inc_shl_base_lds_0_i64(i64 addrspace(1)* %out, i32 addrspace(1)* %add_use) #0 {
   %tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
   %idx.0 = add nsw i32 %tid.x, 2
   %arrayidx0 = getelementptr inbounds [512 x i64], [512 x i64] addrspace(3)* @lds1, i32 0, i32 %idx.0
   %val0 = call i64 @llvm.amdgcn.atomic.inc.i64.p3i64(i64 addrspace(3)* %arrayidx0, i64 9, i32 0, i32 0, i1 false)
   store i32 %idx.0, i32 addrspace(1)* %add_use
   store i64 %val0, i64 addrspace(1)* %out

llvm/test/CodeGen/AMDGPU/local-memory.amdgcn.ll

 ; Check that the LDS size emitted correctly
 ; EG: .long 166120
 ; EG-NEXT: .long 8
 ; GCN: .long 47180
 ; GCN-NEXT: .long 32900
 
 ; GCN-LABEL: {{^}}local_memory_two_objects:
 ; GCN: v_lshlrev_b32_e32 [[ADDRW:v[0-9]+]], 2, v0
-; CI-DAG: v_sub_i32_e32 [[SUB:v[0-9]+]], vcc, 0, [[ADDRW]]
-; CI-DAG: ds_write2_b32 [[ADDRW]], {{v[0-9]+}}, {{v[0-9]+}} offset1:4
-; SI-DAG: ds_write2_b32 [[ADDRW]], {{v[0-9]+}}, {{v[0-9]+}} offset1:4
-; SI-DAG: v_sub_i32_e32 [[SUB0:v[0-9]+]], vcc, 28, [[ADDRW]]
+; CI-DAG:  v_sub_i32_e32 [[SUB0:v[0-9]+]], vcc, 0, [[ADDRW]]
+; SI-DAG:  v_sub_i32_e32 [[SUB0:v[0-9]+]], vcc, 12, [[ADDRW]]
+; GCN-DAG: v_sub_i32_e32 [[SUB1:v[0-9]+]], vcc, 16, [[ADDRW]]
+; GCN-DAG: ds_write2_b32 [[ADDRW]], {{v[0-9]+}}, {{v[0-9]+}} offset1:4
 
 ; GCN: s_barrier
 
-; SI-DAG: v_sub_i32_e32 [[SUB1:v[0-9]+]], vcc, 12, [[ADDRW]]
-; SI-DAG: ds_read_b32 v{{[0-9]+}}, [[SUB0]]
-; SI-DAG: ds_read_b32 v{{[0-9]+}}, [[SUB1]]
-; CI: ds_read2_b32 {{v\[[0-9]+:[0-9]+\]}}, [[SUB]] offset0:3 offset1:7
+; GCN-DAG: ds_read_b32 v{{[0-9]+}}, [[SUB0]]
+; GCN-DAG: ds_read_b32 v{{[0-9]+}}, [[SUB1]]
 
 define amdgpu_kernel void @local_memory_two_objects(i32 addrspace(1)* %out) #0 {
 entry:
   %x.i = call i32 @llvm.amdgcn.workitem.id.x()
   %arrayidx = getelementptr inbounds [4 x i32], [4 x i32] addrspace(3)* @local_memory_two_objects.local_mem0, i32 0, i32 %x.i
   store i32 %x.i, i32 addrspace(3)* %arrayidx, align 4
   %mul = shl nsw i32 %x.i, 1
   %arrayidx1 = getelementptr inbounds [4 x i32], [4 x i32] addrspace(3)* @local_memory_two_objects.local_mem1, i32 0, i32 %x.i

llvm/test/CodeGen/AMDGPU/local-memory.ll

 ; RUN: llc -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,FUNC %s
 ; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,FUNC %s
 ; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck -check-prefix=FUNC %s
 
 @local_memory.local_mem = internal unnamed_addr addrspace(3) global [128 x i32] undef, align 4
 
 @lds = addrspace(3) global [512 x i32] undef, align 4
 
 ; On SI we need to make sure that the base offset is a register and
 ; not an immediate.
 
 ; FUNC-LABEL: {{^}}load_i32_local_const_ptr:
-; GCN: v_mov_b32_e32 v[[PTR:[0-9]+]], lds@abs32@lo
+; GCN: v_mov_b32_e32 v[[PTR:[0-9]+]], 0{{$}}
 ; GCN: ds_read_b32 v{{[0-9]+}}, v[[PTR]] offset:4
 
 ; R600: LDS_READ_RET
 define amdgpu_kernel void @load_i32_local_const_ptr(i32 addrspace(1)* %out, i32 addrspace(3)* %in) #0 {
 entry:
   %tmp0 = getelementptr [512 x i32], [512 x i32] addrspace(3)* @lds, i32 0, i32 1
   %tmp1 = load i32, i32 addrspace(3)* %tmp0
   %tmp2 = getelementptr i32, i32 addrspace(1)* %out, i32 1

llvm/test/CodeGen/AMDGPU/loop_break.ll

-; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: opt -mtriple=amdgcn-- -S -structurizecfg -si-annotate-control-flow %s | FileCheck -check-prefix=OPT %s
 ; RUN: llc -march=amdgcn -verify-machineinstrs -disable-block-placement < %s | FileCheck -check-prefix=GCN %s
 
 ; Uses llvm.amdgcn.break
 
 define amdgpu_kernel void @break_loop(i32 %arg) #0 {
 ; OPT-LABEL: @break_loop(
 ; OPT-NEXT:  bb:

 ; GCN-NEXT:  BB0_1: ; %bb1
 ; GCN-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GCN-NEXT:    s_add_i32 s6, s6, 1
 ; GCN-NEXT:    s_or_b64 s[4:5], s[4:5], exec
 ; GCN-NEXT:    s_cmp_gt_i32 s6, -1
 ; GCN-NEXT:    s_cbranch_scc1 BB0_3
 ; GCN-NEXT:  ; %bb.2: ; %bb4
 ; GCN-NEXT:    ; in Loop: Header=BB0_1 Depth=1
-; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0
+; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0 glc
 ; GCN-NEXT:    s_waitcnt vmcnt(0)
 ; GCN-NEXT:    v_cmp_ge_i32_e32 vcc, v0, v1
 ; GCN-NEXT:    s_andn2_b64 s[4:5], s[4:5], exec
 ; GCN-NEXT:    s_and_b64 s[8:9], vcc, exec
 ; GCN-NEXT:    s_or_b64 s[4:5], s[4:5], s[8:9]
 ; GCN-NEXT:  BB0_3: ; %Flow
 ; GCN-NEXT:    ; in Loop: Header=BB0_1 Depth=1
 ; GCN-NEXT:    s_and_b64 s[8:9], exec, s[4:5]

 ; GCN-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GCN-NEXT:    s_andn2_b64 s[6:7], s[6:7], exec
 ; GCN-NEXT:    s_and_b64 s[8:9], s[0:1], exec
 ; GCN-NEXT:    s_or_b64 s[6:7], s[6:7], s[8:9]
 ; GCN-NEXT:    s_cmp_gt_i32 s4, -1
 ; GCN-NEXT:    s_cbranch_scc1 BB1_3
 ; GCN-NEXT:  ; %bb.2: ; %bb4
 ; GCN-NEXT:    ; in Loop: Header=BB1_1 Depth=1
-; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0
+; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0 glc
 ; GCN-NEXT:    s_waitcnt vmcnt(0)
 ; GCN-NEXT:    v_cmp_ge_i32_e32 vcc, v0, v1
 ; GCN-NEXT:    s_andn2_b64 s[6:7], s[6:7], exec
 ; GCN-NEXT:    s_and_b64 s[8:9], vcc, exec
 ; GCN-NEXT:    s_or_b64 s[6:7], s[6:7], s[8:9]
 ; GCN-NEXT:  BB1_3: ; %Flow
 ; GCN-NEXT:    ; in Loop: Header=BB1_1 Depth=1
 ; GCN-NEXT:    s_add_i32 s4, s4, 1

 ; OPT-NEXT:    call void @llvm.amdgcn.end.cf.i64(i64 [[TMP0]])
 ; OPT-NEXT:    store volatile i32 7, i32 addrspace(3)* undef
 ; OPT-NEXT:    ret void
 ;
 ; GCN-LABEL: constexpr_phi_cond_break_loop:
 ; GCN:       ; %bb.0: ; %bb
 ; GCN-NEXT:    s_load_dword s3, s[0:1], 0x9
 ; GCN-NEXT:    s_mov_b64 s[0:1], 0
-; GCN-NEXT:    s_mov_b32 s2, lds@abs32@lo
-; GCN-NEXT:    s_mov_b32 s6, -1
+; GCN-NEXT:    s_mov_b32 s2, -1
 ; GCN-NEXT:    s_waitcnt lgkmcnt(0)
 ; GCN-NEXT:    v_subrev_i32_e32 v0, vcc, s3, v0
-; GCN-NEXT:    s_mov_b32 s7, 0xf000
+; GCN-NEXT:    s_mov_b32 s3, 0xf000
 ; GCN-NEXT:    ; implicit-def: $sgpr4_sgpr5
-; GCN-NEXT:    ; implicit-def: $sgpr3
+; GCN-NEXT:    ; implicit-def: $sgpr6
 ; GCN-NEXT:  BB2_1: ; %bb1
 ; GCN-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GCN-NEXT:    v_cmp_ne_u32_e64 s[8:9], s2, 4
-; GCN-NEXT:    s_andn2_b64 s[4:5], s[4:5], exec
-; GCN-NEXT:    s_and_b64 s[8:9], s[8:9], exec
-; GCN-NEXT:    s_or_b64 s[4:5], s[4:5], s[8:9]
-; GCN-NEXT:    s_cmp_gt_i32 s3, -1
+; GCN-NEXT:    s_or_b64 s[4:5], s[4:5], exec
+; GCN-NEXT:    s_cmp_gt_i32 s6, -1
 ; GCN-NEXT:    s_cbranch_scc1 BB2_3
 ; GCN-NEXT:  ; %bb.2: ; %bb4
 ; GCN-NEXT:    ; in Loop: Header=BB2_1 Depth=1
-; GCN-NEXT:    buffer_load_dword v1, off, s[4:7], 0
+; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0 glc
 ; GCN-NEXT:    s_waitcnt vmcnt(0)
 ; GCN-NEXT:    v_cmp_ge_i32_e32 vcc, v0, v1
 ; GCN-NEXT:    s_andn2_b64 s[4:5], s[4:5], exec
 ; GCN-NEXT:    s_and_b64 s[8:9], vcc, exec
 ; GCN-NEXT:    s_or_b64 s[4:5], s[4:5], s[8:9]
 ; GCN-NEXT:  BB2_3: ; %Flow
 ; GCN-NEXT:    ; in Loop: Header=BB2_1 Depth=1
-; GCN-NEXT:    s_add_i32 s3, s3, 1
+; GCN-NEXT:    s_add_i32 s6, s6, 1
 ; GCN-NEXT:    s_and_b64 s[8:9], exec, s[4:5]
 ; GCN-NEXT:    s_or_b64 s[0:1], s[8:9], s[0:1]
 ; GCN-NEXT:    s_andn2_b64 exec, exec, s[0:1]
 ; GCN-NEXT:    s_cbranch_execnz BB2_1
 ; GCN-NEXT:  ; %bb.4: ; %bb9
 ; GCN-NEXT:    s_or_b64 exec, exec, s[0:1]
 ; GCN-NEXT:    v_mov_b32_e32 v0, 7
 ; GCN-NEXT:    s_mov_b32 m0, -1

 ; GCN-NEXT:    ; implicit-def: $sgpr6
 ; GCN-NEXT:  BB3_1: ; %bb1
 ; GCN-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GCN-NEXT:    s_or_b64 s[4:5], s[4:5], exec
 ; GCN-NEXT:    s_cmp_gt_i32 s6, -1
 ; GCN-NEXT:    s_cbranch_scc1 BB3_3
 ; GCN-NEXT:  ; %bb.2: ; %bb4
 ; GCN-NEXT:    ; in Loop: Header=BB3_1 Depth=1
-; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0
+; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0 glc
 ; GCN-NEXT:    s_waitcnt vmcnt(0)
 ; GCN-NEXT:    v_cmp_ge_i32_e32 vcc, v0, v1
 ; GCN-NEXT:    s_andn2_b64 s[4:5], s[4:5], exec
 ; GCN-NEXT:    s_and_b64 s[8:9], vcc, exec
 ; GCN-NEXT:    s_or_b64 s[4:5], s[4:5], s[8:9]
 ; GCN-NEXT:  BB3_3: ; %Flow
 ; GCN-NEXT:    ; in Loop: Header=BB3_1 Depth=1
 ; GCN-NEXT:    s_add_i32 s6, s6, 1

 ; GCN-NEXT:    ; implicit-def: $sgpr6
 ; GCN-NEXT:  BB4_1: ; %bb1
 ; GCN-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GCN-NEXT:    s_andn2_b64 s[4:5], s[4:5], exec
 ; GCN-NEXT:    s_cmp_gt_i32 s6, -1
 ; GCN-NEXT:    s_cbranch_scc1 BB4_3
 ; GCN-NEXT:  ; %bb.2: ; %bb4
 ; GCN-NEXT:    ; in Loop: Header=BB4_1 Depth=1
-; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0
+; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0 glc
 ; GCN-NEXT:    s_waitcnt vmcnt(0)
 ; GCN-NEXT:    v_cmp_ge_i32_e32 vcc, v0, v1
 ; GCN-NEXT:    s_andn2_b64 s[4:5], s[4:5], exec
 ; GCN-NEXT:    s_and_b64 s[8:9], vcc, exec
 ; GCN-NEXT:    s_or_b64 s[4:5], s[4:5], s[8:9]
 ; GCN-NEXT:  BB4_3: ; %Flow
 ; GCN-NEXT:    ; in Loop: Header=BB4_1 Depth=1
 ; GCN-NEXT:    s_add_i32 s6, s6, 1

 ; GCN-NEXT:    ; implicit-def: $sgpr6
 ; GCN-NEXT:  BB5_1: ; %bb1
 ; GCN-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GCN-NEXT:    s_or_b64 s[4:5], s[4:5], exec
 ; GCN-NEXT:    s_cmp_gt_i32 s6, -1
 ; GCN-NEXT:    s_cbranch_scc1 BB5_3
 ; GCN-NEXT:  ; %bb.2: ; %bb4
 ; GCN-NEXT:    ; in Loop: Header=BB5_1 Depth=1
-; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0
+; GCN-NEXT:    buffer_load_dword v1, off, s[0:3], 0 glc
 ; GCN-NEXT:    s_waitcnt vmcnt(0)
 ; GCN-NEXT:    v_cmp_ge_i32_e32 vcc, v0, v1
 ; GCN-NEXT:    s_andn2_b64 s[4:5], s[4:5], exec
 ; GCN-NEXT:    s_and_b64 s[8:9], vcc, exec
 ; GCN-NEXT:    s_or_b64 s[4:5], s[4:5], s[8:9]
 ; GCN-NEXT:  BB5_3: ; %Flow
 ; GCN-NEXT:    ; in Loop: Header=BB5_1 Depth=1
 ; GCN-NEXT:    s_xor_b64 s[8:9], s[4:5], -1

llvm/test/CodeGen/AMDGPU/lower-kernel-and-module-lds.ll

+; RUN: opt -S -mtriple=amdgcn-- -amdgpu-lower-module-lds < %s | FileCheck %s
+; RUN: opt -S -mtriple=amdgcn-- -passes=amdgpu-lower-module-lds < %s | FileCheck %s
+
[e-mail].1 = internal unnamed_addr addrspace(3) global [1 x i8] undef, align 1
[e-mail].2 = internal unnamed_addr addrspace(3) global [2 x i8] undef, align 2
[e-mail].4 = internal unnamed_addr addrspace(3) global [4 x i8] undef, align 4
[e-mail].8 = internal unnamed_addr addrspace(3) global [8 x i8] undef, align 8
[e-mail].16 = internal unnamed_addr addrspace(3) global [16 x i8] undef, align 16
+
+; CHECK: %llvm.amdgcn.module.lds.t = type { [8 x i8], [1 x i8] }
+; CHECK: %llvm.amdgcn.kernel.k0.lds.t = type { [16 x i8], [4 x i8], [2 x i8] }
+; CHECK: %llvm.amdgcn.kernel.k1.lds.t = type { [16 x i8], [4 x i8], [2 x i8] }
+; CHECK: %llvm.amdgcn.kernel..lds.t = type { [2 x i8] }
+; CHECK: %llvm.amdgcn.kernel..lds.t.0 = type { [4 x i8] }
+
+;.
+; CHECK: @llvm.amdgcn.module.lds = internal addrspace(3) global %llvm.amdgcn.module.lds.t undef, align 8
+; CHECK: @llvm.compiler.used = appending global [1 x i8*] [i8* addrspacecast (i8 addrspace(3)* getelementptr inbounds (%llvm.amdgcn.module.lds.t, %llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds, i32 0, i32 0, i32 0) to i8*)], section "llvm.metadata"
+; CHECK: @llvm.amdgcn.kernel.k0.lds = internal addrspace(3) global %llvm.amdgcn.kernel.k0.lds.t undef, align 16
+; CHECK: @llvm.amdgcn.kernel.k1.lds = internal addrspace(3) global %llvm.amdgcn.kernel.k1.lds.t undef, align 16
+; CHECK: @llvm.amdgcn.kernel..lds = internal addrspace(3) global %llvm.amdgcn.kernel..lds.t undef, align 2
+; CHECK: @llvm.amdgcn.kernel..lds.1 = internal addrspace(3) global %llvm.amdgcn.kernel..lds.t.0 undef, align 4
+;.
+define amdgpu_kernel void @k0() {
+; CHECK-LABEL: @k0(
+; CHECK-NEXT:    call void @llvm.donothing() [ "ExplicitUse"(%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds) ]
+; CHECK-NEXT:    %lds.size.1.align.1.bc = bitcast [1 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.module.lds.t, %llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds, i32 0, i32 1) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 1, i8 addrspace(3)* %lds.size.1.align.1.bc, align 1
+; CHECK-NEXT:    %lds.size.2.align.2.bc = bitcast [2 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k0.lds.t, %llvm.amdgcn.kernel.k0.lds.t addrspace(3)* @llvm.amdgcn.kernel.k0.lds, i32 0, i32 2) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 2, i8 addrspace(3)* %lds.size.2.align.2.bc, align 2
+; CHECK-NEXT:    %lds.size.4.align.4.bc = bitcast [4 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k0.lds.t, %llvm.amdgcn.kernel.k0.lds.t addrspace(3)* @llvm.amdgcn.kernel.k0.lds, i32 0, i32 1) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 4, i8 addrspace(3)* %lds.size.4.align.4.bc, align 4
+; CHECK-NEXT:    %lds.size.16.align.16.bc = bitcast [16 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k0.lds.t, %llvm.amdgcn.kernel.k0.lds.t addrspace(3)* @llvm.amdgcn.kernel.k0.lds, i32 0, i32 0) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 16, i8 addrspace(3)* %lds.size.16.align.16.bc, align 16
+; CHECK-NEXT:    ret void
+;
+  %lds.size.1.align.1.bc = bitcast [1 x i8] addrspace(3)* @lds.size.1.align.1 to i8 addrspace(3)*
+  store i8 1, i8 addrspace(3)* %lds.size.1.align.1.bc, align 1
+
+  %lds.size.2.align.2.bc = bitcast [2 x i8] addrspace(3)* @lds.size.2.align.2 to i8 addrspace(3)*
+  store i8 2, i8 addrspace(3)* %lds.size.2.align.2.bc, align 2
+
+  %lds.size.4.align.4.bc = bitcast [4 x i8] addrspace(3)* @lds.size.4.align.4 to i8 addrspace(3)*
+  store i8 4, i8 addrspace(3)* %lds.size.4.align.4.bc, align 4
+
+  %lds.size.16.align.16.bc = bitcast [16 x i8] addrspace(3)* @lds.size.16.align.16 to i8 addrspace(3)*
+  store i8 16, i8 addrspace(3)* %lds.size.16.align.16.bc, align 16
+
+  ret void
+}
+
+define amdgpu_kernel void @k1() {
+; CHECK-LABEL: @k1(
+; CHECK-NEXT:    call void @llvm.donothing() [ "ExplicitUse"(%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds) ]
+; CHECK-NEXT:    %lds.size.2.align.2.bc = bitcast [2 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k1.lds.t, %llvm.amdgcn.kernel.k1.lds.t addrspace(3)* @llvm.amdgcn.kernel.k1.lds, i32 0, i32 2) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 2, i8 addrspace(3)* %lds.size.2.align.2.bc, align 2
+; CHECK-NEXT:    %lds.size.4.align.4.bc = bitcast [4 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k1.lds.t, %llvm.amdgcn.kernel.k1.lds.t addrspace(3)* @llvm.amdgcn.kernel.k1.lds, i32 0, i32 1) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 4, i8 addrspace(3)* %lds.size.4.align.4.bc, align 4
+; CHECK-NEXT:    %lds.size.16.align.16.bc = bitcast [16 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k1.lds.t, %llvm.amdgcn.kernel.k1.lds.t addrspace(3)* @llvm.amdgcn.kernel.k1.lds, i32 0, i32 0) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 16, i8 addrspace(3)* %lds.size.16.align.16.bc, align 16
+; CHECK-NEXT:    ret void
+;
+  %lds.size.2.align.2.bc = bitcast [2 x i8] addrspace(3)* @lds.size.2.align.2 to i8 addrspace(3)*
+  store i8 2, i8 addrspace(3)* %lds.size.2.align.2.bc, align 2
+
+  %lds.size.4.align.4.bc = bitcast [4 x i8] addrspace(3)* @lds.size.4.align.4 to i8 addrspace(3)*
+  store i8 4, i8 addrspace(3)* %lds.size.4.align.4.bc, align 4
+
+  %lds.size.16.align.16.bc = bitcast [16 x i8] addrspace(3)* @lds.size.16.align.16 to i8 addrspace(3)*
+  store i8 16, i8 addrspace(3)* %lds.size.16.align.16.bc, align 16
+
+  ret void
+}
+
+define amdgpu_kernel void @0() {
+; CHECK-LABEL: @0(
+; CHECK-NEXT:    call void @llvm.donothing() [ "ExplicitUse"(%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds) ]
+; CHECK-NEXT:    %lds.size.2.align.2.bc = bitcast [2 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel..lds.t, %llvm.amdgcn.kernel..lds.t addrspace(3)* @llvm.amdgcn.kernel..lds, i32 0, i32 0) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 2, i8 addrspace(3)* %lds.size.2.align.2.bc, align 2
+; CHECK-NEXT:    ret void
+;
+  %lds.size.2.align.2.bc = bitcast [2 x i8] addrspace(3)* @lds.size.2.align.2 to i8 addrspace(3)*
+  store i8 2, i8 addrspace(3)* %lds.size.2.align.2.bc, align 2
+
+  ret void
+}
+
+define amdgpu_kernel void @1() {
+; CHECK-LABEL: @1(
+; CHECK-NEXT:    call void @llvm.donothing() [ "ExplicitUse"(%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds) ]
+; CHECK-NEXT:    %lds.size.4.align.4.bc = bitcast [4 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel..lds.t.0, %llvm.amdgcn.kernel..lds.t.0 addrspace(3)* @llvm.amdgcn.kernel..lds.1, i32 0, i32 0) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 4, i8 addrspace(3)* %lds.size.4.align.4.bc, align 4
+; CHECK-NEXT:    ret void
+;
+  %lds.size.4.align.4.bc = bitcast [4 x i8] addrspace(3)* @lds.size.4.align.4 to i8 addrspace(3)*
+  store i8 4, i8 addrspace(3)* %lds.size.4.align.4.bc, align 4
+
+  ret void
+}
+
+define void @f0() {
+; CHECK-LABEL: @f0(
+; CHECK-NEXT:    %lds.size.1.align.1.bc = bitcast [1 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.module.lds.t, %llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds, i32 0, i32 1) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 1, i8 addrspace(3)* %lds.size.1.align.1.bc, align 1
+; CHECK-NEXT:    %lds.size.8.align.8.bc = bitcast [8 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.module.lds.t, %llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds, i32 0, i32 0) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 8, i8 addrspace(3)* %lds.size.8.align.8.bc, align 4
+; CHECK-NEXT:    ret void
+;
+  %lds.size.1.align.1.bc = bitcast [1 x i8] addrspace(3)* @lds.size.1.align.1 to i8 addrspace(3)*
+  store i8 1, i8 addrspace(3)* %lds.size.1.align.1.bc, align 1
+
+  %lds.size.8.align.8.bc = bitcast [8 x i8] addrspace(3)* @lds.size.8.align.8 to i8 addrspace(3)*
+  store i8 8, i8 addrspace(3)* %lds.size.8.align.8.bc, align 4
+
+  ret void
+}
+;.
+; CHECK: attributes #0 = { nofree nosync nounwind readnone willreturn }
+;.

llvm/test/CodeGen/AMDGPU/lower-kernel-lds-constexpr.ll

+; RUN: opt -S -mtriple=amdgcn-- -amdgpu-lower-module-lds < %s | FileCheck %s
+; RUN: opt -S -mtriple=amdgcn-- -passes=amdgpu-lower-module-lds < %s | FileCheck %s
+
+; CHECK: %llvm.amdgcn.kernel.k2.lds.t = type { i32 }
+
+@lds.1 = internal unnamed_addr addrspace(3) global [2 x i8] undef, align 1
+
+; Use constant from different kernels
+;.
+; CHECK: @lds.1 = internal unnamed_addr addrspace(3) global [2 x i8] undef, align 1
+; CHECK: @llvm.amdgcn.kernel.k2.lds = internal addrspace(3) global %llvm.amdgcn.kernel.k2.lds.t undef, align 4
+;.
+define amdgpu_kernel void @k0(i64 %x) {
+; CHECK-LABEL: @k0(
+; CHECK-NEXT:    %ptr = getelementptr inbounds i8, i8* addrspacecast (i8 addrspace(3)* getelementptr inbounds ([2 x i8], [2 x i8] addrspace(3)* @lds.1, i32 0, i32 0) to i8*), i64 %x
+; CHECK-NEXT:    store i8 1, i8* %ptr, align 1
+; CHECK-NEXT:    ret void
+;
+  %ptr = getelementptr inbounds i8, i8* addrspacecast ([2 x i8] addrspace(3)* @lds.1 to i8*), i64 %x
+  store i8 1, i8 addrspace(0)* %ptr, align 1
+  ret void
+}
+
+define amdgpu_kernel void @k1(i64 %x) {
+; CHECK-LABEL: @k1(
+; CHECK-NEXT:    %ptr = getelementptr inbounds i8, i8* addrspacecast (i8 addrspace(3)* getelementptr inbounds ([2 x i8], [2 x i8] addrspace(3)* @lds.1, i32 0, i32 0) to i8*), i64 %x
+; CHECK-NEXT:    store i8 1, i8* %ptr, align 1
+; CHECK-NEXT:    ret void
+;
+  %ptr = getelementptr inbounds i8, i8* addrspacecast ([2 x i8] addrspace(3)* @lds.1 to i8*), i64 %x
+  store i8 1, i8 addrspace(0)* %ptr, align 1
+  ret void
+}
+
+@lds.2 = internal unnamed_addr addrspace(3) global i32 undef, align 4
+
+; Use constant twice from the same kernel
+define amdgpu_kernel void @k2(i64 %x) {
+; CHECK-LABEL: @k2(
+; CHECK-NEXT:    %ptr1 = bitcast i32 addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k2.lds.t, %llvm.amdgcn.kernel.k2.lds.t addrspace(3)* @llvm.amdgcn.kernel.k2.lds, i32 0, i32 0) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 1, i8 addrspace(3)* %ptr1, align 4
+; CHECK-NEXT:    %ptr2 = bitcast i32 addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k2.lds.t, %llvm.amdgcn.kernel.k2.lds.t addrspace(3)* @llvm.amdgcn.kernel.k2.lds, i32 0, i32 0) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 2, i8 addrspace(3)* %ptr2, align 4
+; CHECK-NEXT:    ret void
+;
+  %ptr1 = bitcast i32 addrspace(3)* @lds.2 to i8 addrspace(3)*
+  store i8 1, i8 addrspace(3)* %ptr1, align 4
+  %ptr2 = bitcast i32 addrspace(3)* @lds.2 to i8 addrspace(3)*
+  store i8 2, i8 addrspace(3)* %ptr2, align 4
+  ret void
+}
+
+@lds.3 = internal unnamed_addr addrspace(3) global [32 x i8] undef, align 1
+
+; Use constant twice from the same kernel but a different other constant.
+define amdgpu_kernel void @k3(i64 %x) {
+; CHECK-LABEL: @k3(
+; CHECK-NEXT:    %ptr1 = addrspacecast i64 addrspace(3)* bitcast (i8 addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k3.lds.t, %llvm.amdgcn.kernel.k3.lds.t addrspace(3)* @llvm.amdgcn.kernel.k3.lds, i32 0, i32 0, i32 16) to i64 addrspace(3)*) to i64*
+; CHECK-NEXT:    store i64 1, i64* %ptr1, align 1
+; CHECK-NEXT:    %ptr2 = addrspacecast i64 addrspace(3)* bitcast (i8 addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k3.lds.t, %llvm.amdgcn.kernel.k3.lds.t addrspace(3)* @llvm.amdgcn.kernel.k3.lds, i32 0, i32 0, i32 24) to i64 addrspace(3)*) to i64*
+; CHECK-NEXT:    store i64 2, i64* %ptr2, align 1
+;
+  %ptr1 = addrspacecast i64 addrspace(3)* bitcast (i8 addrspace(3)* getelementptr inbounds ([32 x i8], [32 x i8] addrspace(3)* @lds.3, i32 0, i32 16) to i64 addrspace(3)*) to i64*
+  store i64 1, i64* %ptr1, align 1
+  %ptr2 = addrspacecast i64 addrspace(3)* bitcast (i8 addrspace(3)* getelementptr inbounds ([32 x i8], [32 x i8] addrspace(3)* @lds.3, i32 0, i32 24) to i64 addrspace(3)*) to i64*
+  store i64 2, i64* %ptr2, align 1
+  ret void
+}

llvm/test/CodeGen/AMDGPU/lower-kernel-lds.ll

+; RUN: opt -S -mtriple=amdgcn-- -amdgpu-lower-module-lds < %s | FileCheck %s
+; RUN: opt -S -mtriple=amdgcn-- -passes=amdgpu-lower-module-lds < %s | FileCheck %s
+
[e-mail].1 = internal unnamed_addr addrspace(3) global [1 x i8] undef, align 1
[e-mail].2 = internal unnamed_addr addrspace(3) global [2 x i8] undef, align 2
[e-mail].4 = internal unnamed_addr addrspace(3) global [4 x i8] undef, align 4
[e-mail].8 = internal unnamed_addr addrspace(3) global [8 x i8] undef, align 8
[e-mail].16 = internal unnamed_addr addrspace(3) global [16 x i8] undef, align 16
+
+; CHECK: %llvm.amdgcn.kernel.k0.lds.t = type { [16 x i8], [4 x i8], [2 x i8], [1 x i8] }
+; CHECK: %llvm.amdgcn.kernel.k1.lds.t = type { [16 x i8], [4 x i8], [2 x i8] }
+
+;.
+; CHECK: @lds.size.8.align.8 = internal unnamed_addr addrspace(3) global [8 x i8] undef, align 8
+; CHECK: @llvm.amdgcn.kernel.k0.lds = internal addrspace(3) global %llvm.amdgcn.kernel.k0.lds.t undef, align 16
+; CHECK: @llvm.amdgcn.kernel.k1.lds = internal addrspace(3) global %llvm.amdgcn.kernel.k1.lds.t undef, align 16
+;.
+define amdgpu_kernel void @k0() {
+; CHECK-LABEL: @k0(
+; CHECK-NEXT:    %lds.size.1.align.1.bc = bitcast [1 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k0.lds.t, %llvm.amdgcn.kernel.k0.lds.t addrspace(3)* @llvm.amdgcn.kernel.k0.lds, i32 0, i32 3) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 1, i8 addrspace(3)* %lds.size.1.align.1.bc, align 1
+; CHECK-NEXT:    %lds.size.2.align.2.bc = bitcast [2 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k0.lds.t, %llvm.amdgcn.kernel.k0.lds.t addrspace(3)* @llvm.amdgcn.kernel.k0.lds, i32 0, i32 2) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 2, i8 addrspace(3)* %lds.size.2.align.2.bc, align 2
+; CHECK-NEXT:    %lds.size.4.align.4.bc = bitcast [4 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k0.lds.t, %llvm.amdgcn.kernel.k0.lds.t addrspace(3)* @llvm.amdgcn.kernel.k0.lds, i32 0, i32 1) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 4, i8 addrspace(3)* %lds.size.4.align.4.bc, align 4
+; CHECK-NEXT:    %lds.size.16.align.16.bc = bitcast [16 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k0.lds.t, %llvm.amdgcn.kernel.k0.lds.t addrspace(3)* @llvm.amdgcn.kernel.k0.lds, i32 0, i32 0) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 16, i8 addrspace(3)* %lds.size.16.align.16.bc, align 16
+; CHECK-NEXT:    ret void
+;
+  %lds.size.1.align.1.bc = bitcast [1 x i8] addrspace(3)* @lds.size.1.align.1 to i8 addrspace(3)*
+  store i8 1, i8 addrspace(3)* %lds.size.1.align.1.bc, align 1
+
+  %lds.size.2.align.2.bc = bitcast [2 x i8] addrspace(3)* @lds.size.2.align.2 to i8 addrspace(3)*
+  store i8 2, i8 addrspace(3)* %lds.size.2.align.2.bc, align 2
+
+  %lds.size.4.align.4.bc = bitcast [4 x i8] addrspace(3)* @lds.size.4.align.4 to i8 addrspace(3)*
+  store i8 4, i8 addrspace(3)* %lds.size.4.align.4.bc, align 4
+
+  %lds.size.16.align.16.bc = bitcast [16 x i8] addrspace(3)* @lds.size.16.align.16 to i8 addrspace(3)*
+  store i8 16, i8 addrspace(3)* %lds.size.16.align.16.bc, align 16
+
+  ret void
+}
+
+define amdgpu_kernel void @k1() {
+; CHECK-LABEL: @k1(
+; CHECK-NEXT:    %lds.size.2.align.2.bc = bitcast [2 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k1.lds.t, %llvm.amdgcn.kernel.k1.lds.t addrspace(3)* @llvm.amdgcn.kernel.k1.lds, i32 0, i32 2) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 2, i8 addrspace(3)* %lds.size.2.align.2.bc, align 2
+; CHECK-NEXT:    %lds.size.4.align.4.bc = bitcast [4 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k1.lds.t, %llvm.amdgcn.kernel.k1.lds.t addrspace(3)* @llvm.amdgcn.kernel.k1.lds, i32 0, i32 1) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 4, i8 addrspace(3)* %lds.size.4.align.4.bc, align 4
+; CHECK-NEXT:    %lds.size.16.align.16.bc = bitcast [16 x i8] addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.k1.lds.t, %llvm.amdgcn.kernel.k1.lds.t addrspace(3)* @llvm.amdgcn.kernel.k1.lds, i32 0, i32 0) to i8 addrspace(3)*
+; CHECK-NEXT:    store i8 16, i8 addrspace(3)* %lds.size.16.align.16.bc, align 16
+; CHECK-NEXT:    ret void
+;
+  %lds.size.2.align.2.bc = bitcast [2 x i8] addrspace(3)* @lds.size.2.align.2 to i8 addrspace(3)*
+  store i8 2, i8 addrspace(3)* %lds.size.2.align.2.bc, align 2
+
+  %lds.size.4.align.4.bc = bitcast [4 x i8] addrspace(3)* @lds.size.4.align.4 to i8 addrspace(3)*
+  store i8 4, i8 addrspace(3)* %lds.size.4.align.4.bc, align 4
+
+  %lds.size.16.align.16.bc = bitcast [16 x i8] addrspace(3)* @lds.size.16.align.16 to i8 addrspace(3)*
+  store i8 16, i8 addrspace(3)* %lds.size.16.align.16.bc, align 16
+
+  ret void
+}

llvm/test/CodeGen/AMDGPU/lower-module-lds-constantexpr.ll

 ; RUN: opt -S -mtriple=amdgcn-- -amdgpu-lower-module-lds < %s | FileCheck %s
 ; RUN: opt -S -mtriple=amdgcn-- -passes=amdgpu-lower-module-lds < %s | FileCheck %s
 
 ; CHECK: %llvm.amdgcn.module.lds.t = type { float, float }
 
 @func = addrspace(3) global float undef, align 4
 
-; @kern is only used from a kernel so it is left unchanged
-; CHECK: @kern = addrspace(3) global float undef, align 4
+; CHECK: %llvm.amdgcn.kernel.timestwo.lds.t = type { float }
+
 @kern = addrspace(3) global float undef, align 4
 
 ; @func is only used from a non-kernel function so is rewritten
 ; CHECK-NOT: @func
 ; @both is used from a non-kernel function so is rewritten
 ; CHECK-NOT: @both
 ; sorted both < func, so @both at null and @func at 4
 @both = addrspace(3) global float undef, align 4
 
 ; CHECK: @llvm.amdgcn.module.lds = internal addrspace(3) global %llvm.amdgcn.module.lds.t undef, align 4
+; CHECK: @llvm.amdgcn.kernel.timestwo.lds = internal addrspace(3) global %llvm.amdgcn.kernel.timestwo.lds.t undef, align 4
 
 ; CHECK-LABEL: @get_func()
 ; CHECK: %0 = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* getelementptr inbounds (%llvm.amdgcn.module.lds.t, %llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds, i32 0, i32 1) to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* getelementptr inbounds (%llvm.amdgcn.module.lds.t, %llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds, i32 0, i32 1) to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
 define i32 @get_func() local_unnamed_addr #0 {
 entry:
   %0 = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @func to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @func to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
   ret i32 %0
 }
 
 ; CHECK-LABEL: @set_func(i32 %x)
 ; CHECK:  store i32 %x, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
 define void @set_func(i32 %x) local_unnamed_addr #1 {
 entry:
   store i32 %x, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @both to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @both to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
   ret void
 }
 
 ; CHECK-LABEL: @timestwo()
 ; CHECK: call void @llvm.donothing() [ "ExplicitUse"(%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds) ]
-; CHECK: %ld = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
+; CHECK: %ld = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (%llvm.amdgcn.kernel.timestwo.lds.t addrspace(3)* @llvm.amdgcn.kernel.timestwo.lds to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
 ; CHECK: %mul = mul i32 %ld, 2
-; CHECK: store i32 %mul, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
+; CHECK: store i32 %mul, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (%llvm.amdgcn.kernel.timestwo.lds.t addrspace(3)* @llvm.amdgcn.kernel.timestwo.lds to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
 define amdgpu_kernel void @timestwo() {
   %ld = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @both to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
   %mul = mul i32 %ld, 2
   store i32 %mul, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @both to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
   ret void
 }

llvm/test/CodeGen/AMDGPU/lower-module-lds-inactive.ll

 ; RUN: opt -S -mtriple=amdgcn-- -amdgpu-lower-module-lds < %s | FileCheck %s
 ; RUN: opt -S -mtriple=amdgcn-- -passes=amdgpu-lower-module-lds < %s | FileCheck %s
 
 ; Variables that are not lowered by this pass are left unchanged
 ; CHECK-NOT: asm
 ; CHECK-NOT: llvm.amdgcn.module.lds
 ; CHECK-NOT: llvm.amdgcn.module.lds.t
 
 ; var1, var2 would be transformed were they used from a non-kernel function
-; CHECK: @var1 = addrspace(3) global i32 undef
+; CHECK-NOT: @var1 =
 ; CHECK: @var2 = addrspace(3) global float undef
 @var1 = addrspace(3) global i32 undef
 @var2 = addrspace(3) global float undef
 
 ; constant variables are left to the optimizer / error diagnostics
 ; CHECK: @const_undef = addrspace(3) constant i32 undef
 ; CHECK: @const_with_init = addrspace(3) constant i64 8
 @const_undef = addrspace(3) constant i32 undef

 ; CHECK: @addr4 = addrspace(4) global i64 undef
 @addr4 = addrspace(4) global i64 undef
 
 ; Assign to self is treated as any other initializer, i.e. ignored by this pass
 ; CHECK: @toself = addrspace(3) global float addrspace(3)* bitcast (float addrspace(3)* addrspace(3)* @toself to float addrspace(3)*), align 8
 @toself = addrspace(3) global float addrspace(3)* bitcast (float addrspace(3)* addrspace(3)* @toself to float addrspace(3)*), align 8
 
 ; Use by .used lists doesn't trigger lowering
-; CHECK: @llvm.used = appending global [1 x i8*] [i8* addrspacecast (i8 addrspace(3)* bitcast (i32 addrspace(3)* @var1 to i8 addrspace(3)*) to i8*)], section "llvm.metadata"
+; CHECK-NOT: @llvm.used =
 @llvm.used = appending global [1 x i8*] [i8* addrspacecast (i8 addrspace(3)* bitcast (i32 addrspace(3)* @var1 to i8 addrspace(3)*) to i8*)], section "llvm.metadata"
 
 ; CHECK: @llvm.compiler.used = appending global [1 x i8*] [i8* addrspacecast (i8 addrspace(3)* bitcast (float addrspace(3)* @var2 to i8 addrspace(3)*) to i8*)], section "llvm.metadata"
 @llvm.compiler.used = appending global [1 x i8*] [i8* addrspacecast (i8 addrspace(3)* bitcast (float addrspace(3)* @var2 to i8 addrspace(3)*) to i8*)], section "llvm.metadata"
 
 ; Access from a function would cause lowering for non-excluded cases
 ; CHECK-LABEL: @use_variables()
 ; CHECK: %c0 = load i32, i32 addrspace(3)* @const_undef, align 4
 ; CHECK: %c1 = load i64, i64 addrspace(3)* @const_with_init, align 4
 ; CHECK: %v0 = atomicrmw add i64 addrspace(3)* @with_init, i64 1 seq_cst
 ; CHECK: %v1 = cmpxchg i32 addrspace(3)* @extern, i32 4, i32 %c0 acq_rel monotonic
 ; CHECK: %v2 = atomicrmw add i64 addrspace(4)* @addr4, i64 %c1 monotonic
 define void @use_variables() {
   %c0 = load i32, i32 addrspace(3)* @const_undef, align 4
   %c1 = load i64, i64 addrspace(3)* @const_with_init, align 4
   %v0 = atomicrmw add i64 addrspace(3)* @with_init, i64 1 seq_cst
   %v1 = cmpxchg i32 addrspace(3)* @extern, i32 4, i32 %c0 acq_rel monotonic
   %v2 = atomicrmw add i64 addrspace(4)* @addr4, i64 %c1 monotonic
   ret void
 }
 
-; Use by kernel doesn't trigger lowering
 ; CHECK-LABEL: @kern_use()
-; CHECK: %inc = atomicrmw add i32 addrspace(3)* @var1, i32 1 monotonic
+; CHECK: %inc = atomicrmw add i32 addrspace(3)* getelementptr inbounds (%llvm.amdgcn.kernel.kern_use.lds.t, %llvm.amdgcn.kernel.kern_use.lds.t addrspace(3)* @llvm.amdgcn.kernel.kern_use.lds, i32 0, i32 0), i32 1 monotonic, align 4
 define amdgpu_kernel void @kern_use() {
   %inc = atomicrmw add i32 addrspace(3)* @var1, i32 1 monotonic
   call void @use_variables()
   ret void
 }

llvm/test/CodeGen/AMDGPU/lower-module-lds-offsets.ll

 ; GCN:     ds_write_b8 [[NULL]], [[ONE]]
 ; GCN:     v_mov_b32_e32 [[TWO:v[0-9]+]], 2
 ; GCN:     ds_write_b8 [[NULL]], [[TWO]] offset:16
 define amdgpu_kernel void @k0() {
 ; OPT-LABEL: @k0(
 ; OPT-NEXT:    call void @llvm.donothing() [ "ExplicitUse"([[LLVM_AMDGCN_MODULE_LDS_T:%.*]] addrspace(3)* @llvm.amdgcn.module.lds) ]
 ; OPT-NEXT:    [[LDS_SIZE_1_ALIGN_1_BC:%.*]] = bitcast [1 x i8] addrspace(3)* getelementptr inbounds ([[LLVM_AMDGCN_MODULE_LDS_T]], [[LLVM_AMDGCN_MODULE_LDS_T]] addrspace(3)* @llvm.amdgcn.module.lds, i32 0, i32 0) to i8 addrspace(3)*
 ; OPT-NEXT:    store i8 1, i8 addrspace(3)* [[LDS_SIZE_1_ALIGN_1_BC]], align 1
-; OPT-NEXT:    [[LDS_SIZE_16_ALIGN_16_BC:%.*]] = bitcast [16 x i8] addrspace(3)* @lds.size.16.align.16 to i8 addrspace(3)*
+; OPT-NEXT:    [[LDS_SIZE_16_ALIGN_16_BC:%.*]] = bitcast [16 x i8] addrspace(3)* getelementptr inbounds ([[LLVM_AMDGCN_KERNEL_K0_LDS_T:%.*]], [[LLVM_AMDGCN_KERNEL_K0_LDS_T]] addrspace(3)* @llvm.amdgcn.kernel.k0.lds, i32 0, i32 0) to i8 addrspace(3)*
 ; OPT-NEXT:    store i8 2, i8 addrspace(3)* [[LDS_SIZE_16_ALIGN_16_BC]], align 16
 ; OPT-NEXT:    ret void
 ;
   %lds.size.1.align.1.bc = bitcast [1 x i8] addrspace(3)* @lds.size.1.align.1 to i8 addrspace(3)*
   store i8 1, i8 addrspace(3)* %lds.size.1.align.1.bc, align 1
   %lds.size.16.align.16.bc = bitcast [16 x i8] addrspace(3)* @lds.size.16.align.16 to i8 addrspace(3)*
   store i8 2, i8 addrspace(3)* %lds.size.16.align.16.bc, align 16
   ret void

llvm/test/CodeGen/AMDGPU/promote-alloca-globals.ll

 ; RUN: opt -data-layout=A5 -S -mtriple=amdgcn-unknown-unknown -amdgpu-promote-alloca < %s | FileCheck -check-prefix=IR %s
 ; RUN: llc -march=amdgcn -mcpu=tonga < %s | FileCheck -check-prefix=ASM %s
 
 
 @global_array0 = internal unnamed_addr addrspace(3) global [750 x [10 x i32]] undef, align 4
 @global_array1 = internal unnamed_addr addrspace(3) global [750 x [10 x i32]] undef, align 4
 
 ; IR-LABEL: define amdgpu_kernel void @promote_alloca_size_256(i32 addrspace(1)* nocapture %out, i32 addrspace(1)* nocapture %in) {
 ; IR: alloca [10 x i32]
 ; ASM-LABEL: {{^}}promote_alloca_size_256:
-; ASM: .amdgpu_lds global_array0, 30000, 4
-; ASM: .amdgpu_lds global_array1, 30000, 4
+; ASM: .amdgpu_lds llvm.amdgcn.kernel.promote_alloca_size_256.lds, 60000, 4
+; ASM-NOT: .amdgpu_lds
 
 define amdgpu_kernel void @promote_alloca_size_256(i32 addrspace(1)* nocapture %out, i32 addrspace(1)* nocapture %in) {
 entry:
   %stack = alloca [10 x i32], align 4, addrspace(5)
   %tmp = load i32, i32 addrspace(1)* %in, align 4
   %arrayidx1 = getelementptr inbounds [10 x i32], [10 x i32] addrspace(5)* %stack, i32 0, i32 %tmp
   store i32 4, i32 addrspace(5)* %arrayidx1, align 4
   %arrayidx2 = getelementptr inbounds i32, i32 addrspace(1)* %in, i32 1

llvm/test/CodeGen/AMDGPU/promote-alloca-padding-size-estimate.ll

-; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=kaveri --amdhsa-code-object-version=2 -disable-promote-alloca-to-vector < %s | FileCheck -check-prefix=GCN %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=kaveri --amdhsa-code-object-version=2 -disable-promote-alloca-to-vector -amdgpu-enable-lower-module-lds=0 < %s | FileCheck -check-prefix=GCN %s
 
 ; This shows that the amount LDS size estimate should try to not be
 ; sensitive to the order of the LDS globals. This should try to
 ; estimate the worst case padding behavior to avoid overallocating
 ; LDS.
 
 ; These functions use the same amount of LDS, but the total, final
 ; size changes depending on the visit order of first use.

llvm/test/CodeGen/AMDGPU/shl_add_ptr.ll

define amdgpu_kernel void @load_shl_base_lds_0(float addrspace(1)* %out, i32 addrspace(1)* %add_use) #0 {
   ret void
 }
 
 ; Make sure once the first use is folded into the addressing mode, the
 ; remaining add use goes through the normal shl + add constant fold.
 
 ; GCN-LABEL: {{^}}load_shl_base_lds_1:
 ; GCN: v_lshlrev_b32_e32 [[OFS:v[0-9]+]], 2, {{v[0-9]+}}
-
-; TODO: integrate into the ds_read_b32 offset using a 16-bit relocation
-; GCN: v_add_{{[iu]}}32_e32 [[PTR:v[0-9]+]], vcc, lds0@abs32@lo, [[OFS]]
-
-; GCN: ds_read_b32 [[RESULT:v[0-9]+]], [[PTR]] offset:8
+; GCN: ds_read_b32 [[RESULT:v[0-9]+]], [[OFS]] offset:8
 ; GCN: v_add_{{[iu]}}32_e32 [[ADDUSE:v[0-9]+]], vcc, 8, v{{[0-9]+}}
 ; GCN-DAG: buffer_store_dword [[RESULT]]
 ; GCN-DAG: buffer_store_dword [[ADDUSE]]
 ; GCN: s_endpgm
 define amdgpu_kernel void @load_shl_base_lds_1(float addrspace(1)* %out, i32 addrspace(1)* %add_use) #0 {
   %tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
   %idx.0 = add nsw i32 %tid.x, 2
   %arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds0, i32 0, i32 %idx.0
define amdgpu_kernel void @load_shl_base_lds_max_offset(i8 addrspace(1)* %out, i8 addrspace(3)* %lds, i32 addrspace(1)* %add_use) #0 {
   store i32 %idx.0, i32 addrspace(1)* %add_use
   store i8 %val0, i8 addrspace(1)* %out
   ret void
 }
 
 ; The two globals are placed adjacent in memory, so the same base
 ; pointer can be used with an offset into the second one.
 
-; TODO: Recover the optimization of using ds_read2st64_b32 using alignment hints
-
 ; GCN-LABEL: {{^}}load_shl_base_lds_2:
-; GCN: v_lshlrev_b32_e32 [[OFS:v[0-9]+]], 2, {{v[0-9]+}}
-; GCN-DAG: v_add_{{[iu]}}32_e32 [[PTR0:v[0-9]+]], vcc, lds0@abs32@lo, [[OFS]]
-; GCN-DAG: v_add_{{[iu]}}32_e32 [[PTR1:v[0-9]+]], vcc, lds1@abs32@lo, [[OFS]]
-; GCN-DAG: s_mov_b32 m0, -1
-
-; GCN-DAG: ds_read_b32 {{v[0-9]+}}, [[PTR0]] offset:256
-; GCN-DAG: ds_read_b32 {{v[0-9]+}}, [[PTR1]] offset:256
-; TODO: ds_read2st64_b32 {{v\[[0-9]+:[0-9]+\]}}, [[PTR]] offset0:1 offset1:9
-
+; GCN: v_lshlrev_b32_e32 [[PTR:v[0-9]+]], 2, {{v[0-9]+}}
+; GCN: s_mov_b32 m0, -1
+; GCN: ds_read2st64_b32 {{v\[[0-9]+:[0-9]+\]}}, [[PTR]] offset0:1 offset1:9
 ; GCN: s_endpgm
 define amdgpu_kernel void @load_shl_base_lds_2(float addrspace(1)* %out) #0 {
   %tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
   %idx.0 = add nsw i32 %tid.x, 64
   %arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds0, i32 0, i32 %idx.0
   %val0 = load float, float addrspace(3)* %arrayidx0, align 4
   %arrayidx1 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds1, i32 0, i32 %idx.0
   %val1 = load float, float addrspace(3)* %arrayidx1, align 4

llvm/test/CodeGen/MIR/AMDGPU/machine-function-info.ll

 ; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=tahiti -stop-after finalize-isel -o %t.mir %s
 ; RUN: llc -run-pass=none -verify-machineinstrs %t.mir -o - | FileCheck %s
 
 ; Test that SIMachineFunctionInfo can be round trip serialized through
 ; MIR.
 
 @lds = addrspace(3) global [512 x float] undef, align 4
 
 ; CHECK-LABEL: {{^}}name: kernel
 ; CHECK: machineFunctionInfo:
 ; CHECK-NEXT: explicitKernArgSize: 128
 ; CHECK-NEXT: maxKernArgAlign: 64
-; CHECK-NEXT: ldsSize: 0
+; CHECK-NEXT: ldsSize: 2048
 ; CHECK-NEXT: dynLDSAlign: 1
 ; CHECK-NEXT: isEntryFunction: true
 ; CHECK-NEXT: noSignedZerosFPMath: false
 ; CHECK-NEXT: memoryBound: false
 ; CHECK-NEXT: waveLimiter: false
 ; CHECK-NEXT: hasSpilledSGPRs: false
 ; CHECK-NEXT: hasSpilledVGPRs: false
 ; CHECK-NEXT: scratchRSrcReg:  '$sgpr96_sgpr97_sgpr98_sgpr99'