Reapply "[CodeGen] Add new pass for late cleanup of redundant definitions."

Init captures added in processBlock() to avoid capturing structured bindings,
which caused the build problems (with clang).

RISCV has this disabled for now until problems relating to post RA pseudo
expansions are resolved.

diff --git a/llvm/include/llvm/CodeGen/CodeGenPassBuilder.h b/llvm/include/llvm/CodeGen/CodeGenPassBuilder.h
index 48aa4b0..c3fd2b2 100644 (file)
--- a/llvm/include/llvm/CodeGen/CodeGenPassBuilder.h
+++ b/llvm/include/llvm/CodeGen/CodeGenPassBuilder.h
@@ -1130,6 +1130,9 @@ void CodeGenPassBuilder<Derived>::addMachineLateOptimization(
   if (!TM.requiresStructuredCFG())
     addPass(TailDuplicatePass());
 
+  // Cleanup of redundant (identical) address/immediate loads.
+  addPass(MachineLateInstrsCleanupPass());
+
   // Copy propagation.
   addPass(MachineCopyPropagationPass());
 }

diff --git a/llvm/include/llvm/CodeGen/MachinePassRegistry.def b/llvm/include/llvm/CodeGen/MachinePassRegistry.def
index c1ceff9..0bb46e4 100644 (file)
--- a/llvm/include/llvm/CodeGen/MachinePassRegistry.def
+++ b/llvm/include/llvm/CodeGen/MachinePassRegistry.def
@@ -151,6 +151,7 @@ DUMMY_MACHINE_FUNCTION_PASS("implicit-null-checks", ImplicitNullChecksPass, ())
 DUMMY_MACHINE_FUNCTION_PASS("postmisched", PostMachineSchedulerPass, ())
 DUMMY_MACHINE_FUNCTION_PASS("machine-scheduler", MachineSchedulerPass, ())
 DUMMY_MACHINE_FUNCTION_PASS("machine-cp", MachineCopyPropagationPass, ())
+DUMMY_MACHINE_FUNCTION_PASS("machine-latecleanup", MachineLateInstrsCleanupPass, ())
 DUMMY_MACHINE_FUNCTION_PASS("post-RA-sched", PostRASchedulerPass, ())
 DUMMY_MACHINE_FUNCTION_PASS("fentry-insert", FEntryInserterPass, ())
 DUMMY_MACHINE_FUNCTION_PASS("xray-instrumentation", XRayInstrumentationPass, ())

diff --git a/llvm/include/llvm/CodeGen/Passes.h b/llvm/include/llvm/CodeGen/Passes.h
index 5701dd1..b121ecb 100644 (file)
--- a/llvm/include/llvm/CodeGen/Passes.h
+++ b/llvm/include/llvm/CodeGen/Passes.h
@@ -334,6 +334,10 @@ namespace llvm {
 
   MachineFunctionPass *createMachineCopyPropagationPass(bool UseCopyInstr);
 
+  /// MachineLateInstrsCleanup - This pass removes redundant identical
+  /// instructions after register allocation and rematerialization.
+  extern char &MachineLateInstrsCleanupID;
+
   /// PeepholeOptimizer - This pass performs peephole optimizations -
   /// like extension and comparison eliminations.
   extern char &PeepholeOptimizerID;

diff --git a/llvm/include/llvm/InitializePasses.h b/llvm/include/llvm/InitializePasses.h
index 26e2d7e..680cb37 100644 (file)
--- a/llvm/include/llvm/InitializePasses.h
+++ b/llvm/include/llvm/InitializePasses.h
@@ -277,6 +277,7 @@ void initializeMachineDominanceFrontierPass(PassRegistry&);
 void initializeMachineDominatorTreePass(PassRegistry&);
 void initializeMachineFunctionPrinterPassPass(PassRegistry&);
 void initializeMachineFunctionSplitterPass(PassRegistry &);
+void initializeMachineLateInstrsCleanupPass(PassRegistry&);
 void initializeMachineLICMPass(PassRegistry&);
 void initializeMachineLoopInfoPass(PassRegistry&);
 void initializeMachineModuleInfoWrapperPassPass(PassRegistry &);

diff --git a/llvm/lib/CodeGen/CMakeLists.txt b/llvm/lib/CodeGen/CMakeLists.txt
index db3b618..35d1012 100644 (file)
--- a/llvm/lib/CodeGen/CMakeLists.txt
+++ b/llvm/lib/CodeGen/CMakeLists.txt
@@ -119,6 +119,7 @@ add_llvm_component_library(LLVMCodeGen
   MachineFunctionSplitter.cpp
   MachineInstrBundle.cpp
   MachineInstr.cpp
+  MachineLateInstrsCleanup.cpp
   MachineLICM.cpp
   MachineLoopInfo.cpp
   MachineLoopUtils.cpp

diff --git a/llvm/lib/CodeGen/CodeGen.cpp b/llvm/lib/CodeGen/CodeGen.cpp
index a8bde3b..beedd94 100644 (file)
--- a/llvm/lib/CodeGen/CodeGen.cpp
+++ b/llvm/lib/CodeGen/CodeGen.cpp
@@ -78,6 +78,7 @@ void llvm::initializeCodeGen(PassRegistry &Registry) {
   initializeMachineCycleInfoWrapperPassPass(Registry);
   initializeMachineDominatorTreePass(Registry);
   initializeMachineFunctionPrinterPassPass(Registry);
+  initializeMachineLateInstrsCleanupPass(Registry);
   initializeMachineLICMPass(Registry);
   initializeMachineLoopInfoPass(Registry);
   initializeMachineModuleInfoWrapperPassPass(Registry);

diff --git a/llvm/lib/CodeGen/MachineLateInstrsCleanup.cpp b/llvm/lib/CodeGen/MachineLateInstrsCleanup.cpp
new file mode 100644 (file)
index 0000000..41a0223
--- /dev/null
+++ b/llvm/lib/CodeGen/MachineLateInstrsCleanup.cpp
@@ -0,0 +1,240 @@
+//==--- MachineLateInstrsCleanup.cpp - Late Instructions Cleanup Pass -----===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This simple pass removes any identical and redundant immediate or address
+// loads to the same register. The immediate loads removed can originally be
+// the result of rematerialization, while the addresses are redundant frame
+// addressing anchor points created during Frame Indices elimination.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/ADT/BitVector.h"
+#include "llvm/ADT/PostOrderIterator.h"
+#include "llvm/ADT/SmallPtrSet.h"
+#include "llvm/ADT/Statistic.h"
+#include "llvm/CodeGen/MachineBasicBlock.h"
+#include "llvm/CodeGen/MachineFunction.h"
+#include "llvm/CodeGen/MachineFunctionPass.h"
+#include "llvm/CodeGen/MachineInstr.h"
+#include "llvm/CodeGen/MachineOperand.h"
+#include "llvm/CodeGen/MachineRegisterInfo.h"
+#include "llvm/CodeGen/TargetInstrInfo.h"
+#include "llvm/CodeGen/TargetRegisterInfo.h"
+#include "llvm/CodeGen/TargetSubtargetInfo.h"
+#include "llvm/InitializePasses.h"
+#include "llvm/Pass.h"
+#include "llvm/Support/Debug.h"
+
+using namespace llvm;
+
+#define DEBUG_TYPE "machine-latecleanup"
+
+STATISTIC(NumRemoved, "Number of redundant instructions removed.");
+
+namespace {
+
+class MachineLateInstrsCleanup : public MachineFunctionPass {
+  const TargetRegisterInfo *TRI;
+  const TargetInstrInfo *TII;
+
+  // Data structures to map regs to their definitions per MBB.
+  using Reg2DefMap = std::map<Register, MachineInstr*>;
+  std::vector<Reg2DefMap> RegDefs;
+
+  // Walk through the instructions in MBB and remove any redundant
+  // instructions.
+  bool processBlock(MachineBasicBlock *MBB);
+
+public:
+  static char ID; // Pass identification, replacement for typeid
+
+  MachineLateInstrsCleanup() : MachineFunctionPass(ID) {
+    initializeMachineLateInstrsCleanupPass(*PassRegistry::getPassRegistry());
+  }
+
+  void getAnalysisUsage(AnalysisUsage &AU) const override {
+    AU.setPreservesCFG();
+    MachineFunctionPass::getAnalysisUsage(AU);
+  }
+
+  bool runOnMachineFunction(MachineFunction &MF) override;
+
+  MachineFunctionProperties getRequiredProperties() const override {
+    return MachineFunctionProperties().set(
+        MachineFunctionProperties::Property::NoVRegs);
+  }
+};
+
+} // end anonymous namespace
+
+char MachineLateInstrsCleanup::ID = 0;
+
+char &llvm::MachineLateInstrsCleanupID = MachineLateInstrsCleanup::ID;
+
+INITIALIZE_PASS(MachineLateInstrsCleanup, DEBUG_TYPE,
+                "Machine Late Instructions Cleanup Pass", false, false)
+
+bool MachineLateInstrsCleanup::runOnMachineFunction(MachineFunction &MF) {
+  if (skipFunction(MF.getFunction()))
+    return false;
+
+  bool Changed = false;
+
+  TRI = MF.getSubtarget().getRegisterInfo();
+  TII = MF.getSubtarget().getInstrInfo();
+
+  RegDefs.clear();
+  RegDefs.resize(MF.getNumBlockIDs());
+
+  // Visit all MBBs in an order that maximises the reuse from predecessors.
+  ReversePostOrderTraversal<MachineFunction *> RPOT(&MF);
+  for (MachineBasicBlock *MBB : RPOT)
+    Changed |= processBlock(MBB);
+
+  return Changed;
+}
+
+// Clear any previous kill flag on Reg found before I in MBB. Walk backwards
+// in MBB and if needed continue in predecessors until a use/def of Reg is
+// encountered. This seems to be faster in practice than tracking kill flags
+// in a map.
+static void clearKillsForDef(Register Reg, MachineBasicBlock *MBB,
+                             MachineBasicBlock::iterator I,
+                             BitVector &VisitedPreds,
+                             const TargetRegisterInfo *TRI) {
+  VisitedPreds.set(MBB->getNumber());
+  while (I != MBB->begin()) {
+    I--;
+    bool Found = false;
+    for (auto &MO : I->operands())
+      if (MO.isReg() && TRI->regsOverlap(MO.getReg(), Reg)) {
+        if (MO.isDef())
+          return;
+        if (MO.readsReg()) {
+          MO.setIsKill(false);
+          Found = true; // Keep going for an implicit kill of the super-reg.
+        }
+      }
+    if (Found)
+      return;
+  }
+
+  // If an earlier def is not in MBB, continue in predecessors.
+  if (!MBB->isLiveIn(Reg))
+    MBB->addLiveIn(Reg);
+  assert(!MBB->pred_empty() && "Predecessor def not found!");
+  for (MachineBasicBlock *Pred : MBB->predecessors())
+    if (!VisitedPreds.test(Pred->getNumber()))
+      clearKillsForDef(Reg, Pred, Pred->end(), VisitedPreds, TRI);
+}
+
+static void removeRedundantDef(MachineInstr *MI,
+                               const TargetRegisterInfo *TRI) {
+  Register Reg = MI->getOperand(0).getReg();
+  BitVector VisitedPreds(MI->getMF()->getNumBlockIDs());
+  clearKillsForDef(Reg, MI->getParent(), MI->getIterator(), VisitedPreds, TRI);
+  MI->eraseFromParent();
+  ++NumRemoved;
+}
+
+// Return true if MI is a potential candidate for reuse/removal and if so
+// also the register it defines in DefedReg.  A candidate is a simple
+// instruction that does not touch memory, has only one register definition
+// and the only reg it may use is FrameReg. Typically this is an immediate
+// load or a load-address instruction.
+static bool isCandidate(const MachineInstr *MI, Register &DefedReg,
+                        Register FrameReg) {
+  DefedReg = MCRegister::NoRegister;
+  bool SawStore = true;
+  if (!MI->isSafeToMove(nullptr, SawStore) || MI->isImplicitDef() ||
+      MI->isInlineAsm())
+    return false;
+  for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
+    const MachineOperand &MO = MI->getOperand(i);
+    if (MO.isReg()) {
+      if (MO.isDef()) {
+        if (i == 0 && !MO.isImplicit() && !MO.isDead())
+          DefedReg = MO.getReg();
+        else
+          return false;
+      } else if (MO.getReg() && MO.getReg() != FrameReg)
+        return false;
+    } else if (!(MO.isImm() || MO.isCImm() || MO.isFPImm() || MO.isCPI() ||
+                 MO.isGlobal() || MO.isSymbol()))
+      return false;
+  }
+  return DefedReg.isValid();
+}
+
+bool MachineLateInstrsCleanup::processBlock(MachineBasicBlock *MBB) {
+  bool Changed = false;
+
+  Reg2DefMap &MBBDefs = RegDefs[MBB->getNumber()];
+
+  // Find reusable definitions in the predecessor(s).
+  if (!MBB->pred_empty()) {
+    MachineBasicBlock *FirstPred = *MBB->pred_begin();
+    for (auto [Reg, DefMI] : RegDefs[FirstPred->getNumber()])
+      if (llvm::all_of(
+              drop_begin(MBB->predecessors()),
+              [&, &Reg = Reg, &DefMI = DefMI](const MachineBasicBlock *Pred) {
+                auto PredDefI = RegDefs[Pred->getNumber()].find(Reg);
+                return PredDefI != RegDefs[Pred->getNumber()].end() &&
+                       DefMI->isIdenticalTo(*PredDefI->second);
+              })) {
+        MBBDefs[Reg] = DefMI;
+        LLVM_DEBUG(dbgs() << "Reusable instruction from pred(s): in "
+                          << printMBBReference(*MBB) << ":  " << *DefMI;);
+      }
+  }
+
+  // Process MBB.
+  MachineFunction *MF = MBB->getParent();
+  const TargetRegisterInfo *TRI = MF->getSubtarget().getRegisterInfo();
+  Register FrameReg = TRI->getFrameRegister(*MF);
+  for (MachineInstr &MI : llvm::make_early_inc_range(*MBB)) {
+    // If FrameReg is modified, no previous load-address instructions are valid.
+    if (MI.modifiesRegister(FrameReg, TRI)) {
+      MBBDefs.clear();
+      continue;
+    }
+
+    Register DefedReg;
+    bool IsCandidate = isCandidate(&MI, DefedReg, FrameReg);
+
+    // Check for an earlier identical and reusable instruction.
+    if (IsCandidate) {
+      auto DefI = MBBDefs.find(DefedReg);
+      if (DefI != MBBDefs.end() && MI.isIdenticalTo(*DefI->second)) {
+        LLVM_DEBUG(dbgs() << "Removing redundant instruction in "
+                          << printMBBReference(*MBB) << ":  " << MI;);
+        removeRedundantDef(&MI, TRI);
+        Changed = true;
+        continue;
+      }
+    }
+
+    // Clear any entries in map that MI clobbers.
+    for (auto DefI = MBBDefs.begin(); DefI != MBBDefs.end();) {
+      Register Reg = DefI->first;
+      if (MI.modifiesRegister(Reg, TRI))
+        DefI = MBBDefs.erase(DefI);
+      else
+        ++DefI;
+    }
+
+    // Record this MI for potential later reuse.
+    if (IsCandidate) {
+      LLVM_DEBUG(dbgs() << "Found interesting instruction in "
+                        << printMBBReference(*MBB) << ":  " << MI;);
+      MBBDefs[DefedReg] = &MI;
+    }
+  }
+
+  return Changed;
+}

diff --git a/llvm/lib/CodeGen/TargetPassConfig.cpp b/llvm/lib/CodeGen/TargetPassConfig.cpp
index 26ec8ee..0f4a82d 100644 (file)
--- a/llvm/lib/CodeGen/TargetPassConfig.cpp
+++ b/llvm/lib/CodeGen/TargetPassConfig.cpp
@@ -1522,6 +1522,9 @@ void TargetPassConfig::addOptimizedRegAlloc() {
 
 /// Add passes that optimize machine instructions after register allocation.
 void TargetPassConfig::addMachineLateOptimization() {
+  // Cleanup of redundant immediate/address loads.
+  addPass(&MachineLateInstrsCleanupID);
+
   // Branch folding must be run after regalloc and prolog/epilog insertion.
   addPass(&BranchFolderPassID);
 

diff --git a/llvm/lib/Target/NVPTX/NVPTXTargetMachine.cpp b/llvm/lib/Target/NVPTX/NVPTXTargetMachine.cpp
index 56c605d..259ecbf 100644 (file)
--- a/llvm/lib/Target/NVPTX/NVPTXTargetMachine.cpp
+++ b/llvm/lib/Target/NVPTX/NVPTXTargetMachine.cpp
@@ -291,6 +291,7 @@ void NVPTXPassConfig::addIRPasses() {
   // of the PrologEpilogCodeInserter pass, so we emulate that behavior in the
   // NVPTXPrologEpilog pass (see NVPTXPrologEpilogPass.cpp).
   disablePass(&PrologEpilogCodeInserterID);
+  disablePass(&MachineLateInstrsCleanupID);
   disablePass(&MachineCopyPropagationID);
   disablePass(&TailDuplicateID);
   disablePass(&StackMapLivenessID);

diff --git a/llvm/lib/Target/RISCV/RISCVTargetMachine.cpp b/llvm/lib/Target/RISCV/RISCVTargetMachine.cpp
index dff8ccc..43e8c7f 100644 (file)
--- a/llvm/lib/Target/RISCV/RISCVTargetMachine.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetMachine.cpp
@@ -286,6 +286,10 @@ void RISCVPassConfig::addPreRegAlloc() {
 void RISCVPassConfig::addPostRegAlloc() {
   if (TM->getOptLevel() != CodeGenOpt::None && EnableRedundantCopyElimination)
     addPass(createRISCVRedundantCopyEliminationPass());
+
+  // Temporarily disabled until post-RA pseudo expansion problem is fixed,
+  // see D123394 and D139169.
+  disablePass(&MachineLateInstrsCleanupID);
 }
 
 yaml::MachineFunctionInfo *

diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyTargetMachine.cpp b/llvm/lib/Target/WebAssembly/WebAssemblyTargetMachine.cpp
index ffc3de2..497ba8d 100644 (file)
--- a/llvm/lib/Target/WebAssembly/WebAssemblyTargetMachine.cpp
+++ b/llvm/lib/Target/WebAssembly/WebAssemblyTargetMachine.cpp
@@ -501,6 +501,7 @@ void WebAssemblyPassConfig::addPostRegAlloc() {
   // them.
 
   // These functions all require the NoVRegs property.
+  disablePass(&MachineLateInstrsCleanupID);
   disablePass(&MachineCopyPropagationID);
   disablePass(&PostRAMachineSinkingID);
   disablePass(&PostRASchedulerID);

diff --git a/llvm/test/CodeGen/AArch64/O3-pipeline.ll b/llvm/test/CodeGen/AArch64/O3-pipeline.ll
index 90cf49e..15b906d 100644 (file)
--- a/llvm/test/CodeGen/AArch64/O3-pipeline.ll
+++ b/llvm/test/CodeGen/AArch64/O3-pipeline.ll
@@ -188,6 +188,7 @@
 ; CHECK-NEXT:       Machine Optimization Remark Emitter
 ; CHECK-NEXT:       Shrink Wrapping analysis
 ; CHECK-NEXT:       Prologue/Epilogue Insertion & Frame Finalization
+; CHECK-NEXT:       Machine Late Instructions Cleanup Pass
 ; CHECK-NEXT:       Control Flow Optimizer
 ; CHECK-NEXT:       Lazy Machine Block Frequency Analysis
 ; CHECK-NEXT:       Tail Duplication

diff --git a/llvm/test/CodeGen/AArch64/stack-guard-remat-bitcast.ll b/llvm/test/CodeGen/AArch64/stack-guard-remat-bitcast.ll
index bb3397e..b71e9e2 100644 (file)
--- a/llvm/test/CodeGen/AArch64/stack-guard-remat-bitcast.ll
+++ b/llvm/test/CodeGen/AArch64/stack-guard-remat-bitcast.ll
@@ -29,14 +29,8 @@ define i32 @test_stack_guard_remat2() ssp {
 ; CHECK-NEXT:  Lloh5:
 ; CHECK-NEXT:    ldr x9, [x9]
 ; CHECK-NEXT:    str x8, [sp]
-; CHECK-NEXT:  Lloh6:
-; CHECK-NEXT:    adrp x8, ___stack_chk_guard@GOTPAGE
 ; CHECK-NEXT:    stur x9, [x29, #-8]
-; CHECK-NEXT:  Lloh7:
-; CHECK-NEXT:    ldr x8, [x8, ___stack_chk_guard@GOTPAGEOFF]
 ; CHECK-NEXT:    ldur x9, [x29, #-8]
-; CHECK-NEXT:  Lloh8:
-; CHECK-NEXT:    ldr x8, [x8]
 ; CHECK-NEXT:    cmp x8, x9
 ; CHECK-NEXT:    b.ne LBB0_2
 ; CHECK-NEXT:  ; %bb.1: ; %entry
@@ -46,7 +40,6 @@ define i32 @test_stack_guard_remat2() ssp {
 ; CHECK-NEXT:    ret
 ; CHECK-NEXT:  LBB0_2: ; %entry
 ; CHECK-NEXT:    bl ___stack_chk_fail
-; CHECK-NEXT:    .loh AdrpLdrGotLdr Lloh6, Lloh7, Lloh8
 ; CHECK-NEXT:    .loh AdrpLdrGotLdr Lloh1, Lloh3, Lloh5
 ; CHECK-NEXT:    .loh AdrpLdrGotLdr Lloh0, Lloh2, Lloh4
 entry:

diff --git a/llvm/test/CodeGen/AArch64/sve-calling-convention-mixed.ll b/llvm/test/CodeGen/AArch64/sve-calling-convention-mixed.ll
index 554f9b9..34d4612 100644 (file)
--- a/llvm/test/CodeGen/AArch64/sve-calling-convention-mixed.ll
+++ b/llvm/test/CodeGen/AArch64/sve-calling-convention-mixed.ll
@@ -59,26 +59,23 @@ define float @foo2(double* %x0, double* %x1) nounwind {
 ; CHECK-NEXT:    addvl sp, sp, #-4
 ; CHECK-NEXT:    sub sp, sp, #16
 ; CHECK-NEXT:    ptrue p0.b
-; CHECK-NEXT:    add x9, sp, #16
+; CHECK-NEXT:    add x8, sp, #16
 ; CHECK-NEXT:    ld4d { z1.d - z4.d }, p0/z, [x0]
 ; CHECK-NEXT:    ld4d { z16.d - z19.d }, p0/z, [x1]
-; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    add x8, sp, #16
 ; CHECK-NEXT:    fmov s0, #1.00000000
 ; CHECK-NEXT:    mov w0, wzr
 ; CHECK-NEXT:    mov w1, #1
 ; CHECK-NEXT:    mov w2, #2
-; CHECK-NEXT:    st1d { z16.d }, p0, [x9]
-; CHECK-NEXT:    add x9, sp, #16
 ; CHECK-NEXT:    mov w3, #3
 ; CHECK-NEXT:    mov w4, #4
 ; CHECK-NEXT:    mov w5, #5
 ; CHECK-NEXT:    mov w6, #6
-; CHECK-NEXT:    st1d { z17.d }, p0, [x9, #1, mul vl]
-; CHECK-NEXT:    add x9, sp, #16
 ; CHECK-NEXT:    mov w7, #7
-; CHECK-NEXT:    st1d { z18.d }, p0, [x9, #2, mul vl]
 ; CHECK-NEXT:    add x9, sp, #16
+; CHECK-NEXT:    ptrue p0.d
+; CHECK-NEXT:    st1d { z16.d }, p0, [x9]
+; CHECK-NEXT:    st1d { z17.d }, p0, [x9, #1, mul vl]
+; CHECK-NEXT:    st1d { z18.d }, p0, [x9, #2, mul vl]
 ; CHECK-NEXT:    st1d { z19.d }, p0, [x9, #3, mul vl]
 ; CHECK-NEXT:    str x8, [sp]
 ; CHECK-NEXT:    bl callee2

diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/call-outgoing-stack-args.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/call-outgoing-stack-args.ll
index 37592a7..7ed11c7 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/call-outgoing-stack-args.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/call-outgoing-stack-args.ll
@@ -157,8 +157,6 @@ define amdgpu_kernel void @kernel_caller_byval() {
 ; FLATSCR-NEXT:    v_mov_b32_e32 v1, 0
 ; FLATSCR-NEXT:    s_mov_b32 vcc_lo, 0
 ; FLATSCR-NEXT:    s_mov_b32 vcc_hi, 0
-; FLATSCR-NEXT:    scratch_store_dwordx2 off, v[0:1], vcc_lo offset:8
-; FLATSCR-NEXT:    scratch_store_dwordx2 off, v[0:1], vcc_hi offset:16
 ; FLATSCR-NEXT:    s_mov_b32 s11, 0
 ; FLATSCR-NEXT:    s_mov_b32 s10, 0
 ; FLATSCR-NEXT:    s_mov_b32 s9, 0
@@ -171,9 +169,8 @@ define amdgpu_kernel void @kernel_caller_byval() {
 ; FLATSCR-NEXT:    s_mov_b32 s4, 0
 ; FLATSCR-NEXT:    s_mov_b32 s3, 0
 ; FLATSCR-NEXT:    s_mov_b32 s2, 0
-; FLATSCR-NEXT:    s_mov_b32 vcc_lo, 0
-; FLATSCR-NEXT:    s_mov_b32 vcc_hi, 0
-; FLATSCR-NEXT:    s_mov_b32 s40, 0
+; FLATSCR-NEXT:    scratch_store_dwordx2 off, v[0:1], vcc_lo offset:8
+; FLATSCR-NEXT:    scratch_store_dwordx2 off, v[0:1], vcc_hi offset:16
 ; FLATSCR-NEXT:    scratch_store_dwordx2 off, v[0:1], s11 offset:24
 ; FLATSCR-NEXT:    scratch_store_dwordx2 off, v[0:1], s10 offset:32
 ; FLATSCR-NEXT:    scratch_store_dwordx2 off, v[0:1], s9 offset:40
@@ -188,6 +185,7 @@ define amdgpu_kernel void @kernel_caller_byval() {
 ; FLATSCR-NEXT:    scratch_store_dwordx2 off, v[0:1], s2 offset:112
 ; FLATSCR-NEXT:    scratch_store_dwordx2 off, v[0:1], vcc_lo offset:120
 ; FLATSCR-NEXT:    scratch_store_dwordx2 off, v[0:1], vcc_hi offset:128
+; FLATSCR-NEXT:    s_mov_b32 s40, 0
 ; FLATSCR-NEXT:    scratch_load_dwordx2 v[0:1], off, s40 offset:8
 ; FLATSCR-NEXT:    s_mov_b32 s39, 0
 ; FLATSCR-NEXT:    scratch_load_dwordx2 v[2:3], off, s39 offset:16

diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.div.fmas.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.div.fmas.ll
index d8705bf..5f71713 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.div.fmas.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.div.fmas.ll
@@ -1354,7 +1354,6 @@ define amdgpu_kernel void @test_div_fmas_f32_i1_phi_vcc(ptr addrspace(1) %out, [
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    buffer_load_dwordx3 v[1:3], v[1:2], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b32 s6, 0
 ; GFX7-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7-NEXT:    s_cbranch_execz .LBB13_2

diff --git a/llvm/test/CodeGen/AMDGPU/cc-update.ll b/llvm/test/CodeGen/AMDGPU/cc-update.ll
index 83647a0..a3e22a3 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/cc-update.ll
+++ b/llvm/test/CodeGen/AMDGPU/cc-update.ll
@@ -537,7 +537,6 @@ define amdgpu_kernel void @test_sgpr_offset_kernel() #1 {
 ; GFX803-NEXT:    buffer_store_dword v0, off, s[0:3], s4 ; 4-byte Folded Spill
 ; GFX803-NEXT:    ;;#ASMSTART
 ; GFX803-NEXT:    ;;#ASMEND
-; GFX803-NEXT:    s_mov_b32 s4, 0x40000
 ; GFX803-NEXT:    buffer_load_dword v0, off, s[0:3], s4 ; 4-byte Folded Reload
 ; GFX803-NEXT:    s_waitcnt vmcnt(0)
 ; GFX803-NEXT:    buffer_store_dword v0, off, s[0:3], 0 offset:8
@@ -554,7 +553,6 @@ define amdgpu_kernel void @test_sgpr_offset_kernel() #1 {
 ; GFX900-NEXT:    buffer_store_dword v0, off, s[0:3], s4 ; 4-byte Folded Spill
 ; GFX900-NEXT:    ;;#ASMSTART
 ; GFX900-NEXT:    ;;#ASMEND
-; GFX900-NEXT:    s_mov_b32 s4, 0x40000
 ; GFX900-NEXT:    buffer_load_dword v0, off, s[0:3], s4 ; 4-byte Folded Reload
 ; GFX900-NEXT:    s_waitcnt vmcnt(0)
 ; GFX900-NEXT:    buffer_store_dword v0, off, s[0:3], 0 offset:8
@@ -569,8 +567,6 @@ define amdgpu_kernel void @test_sgpr_offset_kernel() #1 {
 ; GFX1010-NEXT:    buffer_load_dword v0, off, s[0:3], 0 offset:8 glc dlc
 ; GFX1010-NEXT:    s_waitcnt vmcnt(0)
 ; GFX1010-NEXT:    buffer_store_dword v0, off, s[0:3], s4 ; 4-byte Folded Spill
-; GFX1010-NEXT:    s_waitcnt_depctr 0xffe3
-; GFX1010-NEXT:    s_mov_b32 s4, 0x20000
 ; GFX1010-NEXT:    ;;#ASMSTART
 ; GFX1010-NEXT:    ;;#ASMEND
 ; GFX1010-NEXT:    buffer_load_dword v0, off, s[0:3], s4 ; 4-byte Folded Reload
@@ -585,7 +581,6 @@ define amdgpu_kernel void @test_sgpr_offset_kernel() #1 {
 ; GFX1100-NEXT:    s_waitcnt vmcnt(0)
 ; GFX1100-NEXT:    s_movk_i32 s0, 0x1000
 ; GFX1100-NEXT:    scratch_store_b32 off, v0, s0 ; 4-byte Folded Spill
-; GFX1100-NEXT:    s_movk_i32 s0, 0x1000
 ; GFX1100-NEXT:    ;;#ASMSTART
 ; GFX1100-NEXT:    ;;#ASMEND
 ; GFX1100-NEXT:    scratch_load_b32 v0, off, s0 ; 4-byte Folded Reload

diff --git a/llvm/test/CodeGen/AMDGPU/exec-mask-opt-cannot-create-empty-or-backward-segment.ll b/llvm/test/CodeGen/AMDGPU/exec-mask-opt-cannot-create-empty-or-backward-segment.ll
index b4c00f3..3f50130 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/exec-mask-opt-cannot-create-empty-or-backward-segment.ll
+++ b/llvm/test/CodeGen/AMDGPU/exec-mask-opt-cannot-create-empty-or-backward-segment.ll
@@ -76,12 +76,10 @@ define amdgpu_kernel void @cannot_create_empty_or_backwards_segment(i1 %arg, i1
 ; CHECK-NEXT:  ; %bb.10: ; %bb16
 ; CHECK-NEXT:    ; in Loop: Header=BB0_3 Depth=1
 ; CHECK-NEXT:    s_mov_b64 s[16:17], 0
-; CHECK-NEXT:    s_mov_b64 s[20:21], -1
 ; CHECK-NEXT:    s_mov_b64 s[22:23], s[10:11]
 ; CHECK-NEXT:    s_mov_b64 s[18:19], s[16:17]
 ; CHECK-NEXT:    s_branch .LBB0_2
 ; CHECK-NEXT:  .LBB0_11: ; in Loop: Header=BB0_3 Depth=1
-; CHECK-NEXT:    s_mov_b64 s[22:23], -1
 ; CHECK-NEXT:    s_mov_b64 s[20:21], 0
 ; CHECK-NEXT:    ; implicit-def: $sgpr16_sgpr17
 ; CHECK-NEXT:    s_mov_b64 s[18:19], s[16:17]

diff --git a/llvm/test/CodeGen/AMDGPU/flat-scratch.ll b/llvm/test/CodeGen/AMDGPU/flat-scratch.ll
index c0590ce..06ee6b4 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/flat-scratch.ll
+++ b/llvm/test/CodeGen/AMDGPU/flat-scratch.ll
@@ -22,18 +22,12 @@ define amdgpu_kernel void @zero_init_kernel() {
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s3
 ; GFX9-NEXT:    s_mov_b32 s1, 0
-; GFX9-NEXT:    s_mov_b32 s0, 0
 ; GFX9-NEXT:    s_mov_b32 vcc_lo, 0
 ; GFX9-NEXT:    s_mov_b32 vcc_hi, 0
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:52
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:36
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:20
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_hi offset:4
-; GFX9-NEXT:    s_nop 0
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: zero_init_kernel:
@@ -43,7 +37,6 @@ define amdgpu_kernel void @zero_init_kernel() {
 ; GFX10-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_LO), s0
 ; GFX10-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_HI), s1
 ; GFX10-NEXT:    s_mov_b32 s0, 0
-; GFX10-NEXT:    v_mov_b32_e32 v4, 4
 ; GFX10-NEXT:    s_mov_b32 s1, s0
 ; GFX10-NEXT:    s_mov_b32 s2, s0
 ; GFX10-NEXT:    s_mov_b32 s3, s0
@@ -55,15 +48,12 @@ define amdgpu_kernel void @zero_init_kernel() {
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:36
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:20
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:4
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v4
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: zero_init_kernel:
 ; GFX11:       ; %bb.0:
 ; GFX11-NEXT:    s_mov_b32 s0, 0
-; GFX11-NEXT:    v_mov_b32_e32 v4, 4
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-NEXT:    s_mov_b32 s1, s0
 ; GFX11-NEXT:    s_mov_b32 s2, s0
 ; GFX11-NEXT:    s_mov_b32 s3, s0
@@ -74,9 +64,6 @@ define amdgpu_kernel void @zero_init_kernel() {
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], off offset:36
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], off offset:20
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], off offset:4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v4
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
 ; GFX11-NEXT:    s_endpgm
 ;
@@ -100,16 +87,10 @@ define amdgpu_kernel void @zero_init_kernel() {
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v3, s3
 ; GFX9-PAL-NEXT:    s_mov_b32 s1, 0
-; GFX9-PAL-NEXT:    s_mov_b32 s0, 0
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:52
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:36
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:20
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_hi offset:4
-; GFX9-PAL-NEXT:    s_nop 0
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: zero_init_kernel:
@@ -124,11 +105,6 @@ define amdgpu_kernel void @zero_init_kernel() {
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:36
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:20
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:4
-; GFX940-NEXT:    s_nop 1
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX1010-PAL-LABEL: zero_init_kernel:
@@ -153,15 +129,10 @@ define amdgpu_kernel void @zero_init_kernel() {
 ; GFX1010-PAL-NEXT:    v_mov_b32_e32 v3, s3
 ; GFX1010-PAL-NEXT:    s_mov_b32 s2, 0
 ; GFX1010-PAL-NEXT:    s_mov_b32 s1, 0
-; GFX1010-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v4, 4
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s2 offset:52
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:36
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:20
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:4
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v4
-; GFX1010-PAL-NEXT:    ;;#ASMEND
 ; GFX1010-PAL-NEXT:    s_endpgm
 ;
 ; GFX1030-PAL-LABEL: zero_init_kernel:
@@ -176,7 +147,6 @@ define amdgpu_kernel void @zero_init_kernel() {
 ; GFX1030-PAL-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_LO), s2
 ; GFX1030-PAL-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_HI), s3
 ; GFX1030-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v4, 4
 ; GFX1030-PAL-NEXT:    s_mov_b32 s1, s0
 ; GFX1030-PAL-NEXT:    s_mov_b32 s2, s0
 ; GFX1030-PAL-NEXT:    s_mov_b32 s3, s0
@@ -188,15 +158,12 @@ define amdgpu_kernel void @zero_init_kernel() {
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:36
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:20
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:4
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v4
-; GFX1030-PAL-NEXT:    ;;#ASMEND
 ; GFX1030-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: zero_init_kernel:
 ; GFX11-PAL:       ; %bb.0:
 ; GFX11-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v4, 4
+; GFX11-PAL-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-PAL-NEXT:    s_mov_b32 s1, s0
 ; GFX11-PAL-NEXT:    s_mov_b32 s2, s0
 ; GFX11-PAL-NEXT:    s_mov_b32 s3, s0
@@ -207,15 +174,11 @@ define amdgpu_kernel void @zero_init_kernel() {
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], off offset:36
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], off offset:20
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], off offset:4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v4
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
 ; GFX11-PAL-NEXT:    s_endpgm
   %alloca = alloca [32 x i16], align 2, addrspace(5)
   %cast = bitcast [32 x i16] addrspace(5)* %alloca to i8 addrspace(5)*
   call void @llvm.memset.p5i8.i64(i8 addrspace(5)* align 2 dereferenceable(64) %cast, i8 0, i64 64, i1 false)
-  call void asm sideeffect "; use $0", "s"([32 x i16] addrspace(5)* %alloca) #0
   ret void
 }
 
@@ -235,11 +198,6 @@ define void @zero_init_foo() {
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:32
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:16
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s32
-; GFX9-NEXT:    s_nop 0
-; GFX9-NEXT:    v_mov_b32_e32 v0, s32
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -248,7 +206,6 @@ define void @zero_init_foo() {
 ; GFX10-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    s_mov_b32 s0, 0
-; GFX10-NEXT:    v_mov_b32_e32 v4, s32
 ; GFX10-NEXT:    s_mov_b32 s1, s0
 ; GFX10-NEXT:    s_mov_b32 s2, s0
 ; GFX10-NEXT:    s_mov_b32 s3, s0
@@ -260,9 +217,6 @@ define void @zero_init_foo() {
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:32
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:16
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s32
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v4
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -271,7 +225,7 @@ define void @zero_init_foo() {
 ; GFX11-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    s_mov_b32 s0, 0
-; GFX11-NEXT:    v_mov_b32_e32 v4, s32
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-NEXT:    s_mov_b32 s1, s0
 ; GFX11-NEXT:    s_mov_b32 s2, s0
 ; GFX11-NEXT:    s_mov_b32 s3, s0
@@ -282,9 +236,6 @@ define void @zero_init_foo() {
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:32
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:16
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s32
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v4
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -303,11 +254,6 @@ define void @zero_init_foo() {
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:32
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:16
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32
-; GFX9-PAL-NEXT:    s_nop 0
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, s32
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -324,11 +270,6 @@ define void @zero_init_foo() {
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:32
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:16
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s32
-; GFX940-NEXT:    s_nop 1
-; GFX940-NEXT:    v_mov_b32_e32 v0, s32
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
 ; GFX940-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -337,7 +278,6 @@ define void @zero_init_foo() {
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v4, s32
 ; GFX10-PAL-NEXT:    s_mov_b32 s1, s0
 ; GFX10-PAL-NEXT:    s_mov_b32 s2, s0
 ; GFX10-PAL-NEXT:    s_mov_b32 s3, s0
@@ -349,9 +289,6 @@ define void @zero_init_foo() {
 ; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:32
 ; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:16
 ; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v4
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -360,7 +297,7 @@ define void @zero_init_foo() {
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v4, s32
+; GFX11-PAL-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-PAL-NEXT:    s_mov_b32 s1, s0
 ; GFX11-PAL-NEXT:    s_mov_b32 s2, s0
 ; GFX11-PAL-NEXT:    s_mov_b32 s3, s0
@@ -371,15 +308,26 @@ define void @zero_init_foo() {
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:32
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:16
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s32
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v4
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: zero_init_foo:
+; GCN:       ; %bb.0:
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    s_mov_b32 s0, 0
+; GCN-NEXT:    s_mov_b32 s1, s0
+; GCN-NEXT:    s_mov_b32 s2, s0
+; GCN-NEXT:    s_mov_b32 s3, s0
+; GCN-NEXT:    v_mov_b64_e32 v[0:1], s[0:1]
+; GCN-NEXT:    v_mov_b64_e32 v[2:3], s[2:3]
+; GCN-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:48
+; GCN-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:32
+; GCN-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:16
+; GCN-NEXT:    scratch_store_dwordx4 off, v[0:3], s32
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
   %alloca = alloca [32 x i16], align 2, addrspace(5)
   %cast = bitcast [32 x i16] addrspace(5)* %alloca to i8 addrspace(5)*
   call void @llvm.memset.p5i8.i64(i8 addrspace(5)* align 2 dereferenceable(64) %cast, i8 0, i64 64, i1 false)
-  call void asm sideeffect "; use $0", "s"([32 x i16] addrspace(5)* %alloca) #0
   ret void
 }
 
@@ -400,10 +348,6 @@ define amdgpu_kernel void @store_load_sindex_kernel(i32 %idx) {
 ; GFX9-NEXT:    s_add_i32 s0, s0, 4
 ; GFX9-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_sindex_kernel:
@@ -424,10 +368,6 @@ define amdgpu_kernel void @store_load_sindex_kernel(i32 %idx) {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_sindex_kernel:
@@ -444,10 +384,6 @@ define amdgpu_kernel void @store_load_sindex_kernel(i32 %idx) {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_sindex_kernel:
@@ -470,10 +406,6 @@ define amdgpu_kernel void @store_load_sindex_kernel(i32 %idx) {
 ; GFX9-PAL-NEXT:    s_add_i32 s0, s0, 4
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_sindex_kernel:
@@ -490,10 +422,6 @@ define amdgpu_kernel void @store_load_sindex_kernel(i32 %idx) {
 ; GFX940-NEXT:    s_add_i32 s0, s0, 4
 ; GFX940-NEXT:    scratch_load_dword v0, off, s0 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX10-PAL-LABEL: store_load_sindex_kernel:
@@ -519,10 +447,6 @@ define amdgpu_kernel void @store_load_sindex_kernel(i32 %idx) {
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v0
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_sindex_kernel:
@@ -539,11 +463,22 @@ define amdgpu_kernel void @store_load_sindex_kernel(i32 %idx) {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
+; GCN-LABEL: store_load_sindex_kernel:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_load_dword s0, s[0:1], 0x24
+; GCN-NEXT:    v_mov_b32_e32 v0, 15
+; GCN-NEXT:    s_waitcnt lgkmcnt(0)
+; GCN-NEXT:    s_lshl_b32 s1, s0, 2
+; GCN-NEXT:    s_and_b32 s0, s0, 15
+; GCN-NEXT:    s_lshl_b32 s0, s0, 2
+; GCN-NEXT:    s_add_u32 s1, 4, s1
+; GCN-NEXT:    scratch_store_dword off, v0, s1 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_add_u32 s0, 4, s0
+; GCN-NEXT:    scratch_load_dword v0, off, s0 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_endpgm
 bb:
   %i = alloca [32 x float], align 4, addrspace(5)
   %i1 = bitcast [32 x float] addrspace(5)* %i to i8 addrspace(5)*
@@ -554,7 +489,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -573,10 +507,6 @@ define amdgpu_ps void @store_load_sindex_foo(i32 inreg %idx) {
 ; GFX9-NEXT:    s_add_i32 s0, s0, 4
 ; GFX9-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_sindex_foo:
@@ -595,10 +525,6 @@ define amdgpu_ps void @store_load_sindex_foo(i32 inreg %idx) {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, off, s0 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_sindex_foo:
@@ -613,10 +539,6 @@ define amdgpu_ps void @store_load_sindex_foo(i32 inreg %idx) {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_sindex_foo:
@@ -638,10 +560,6 @@ define amdgpu_ps void @store_load_sindex_foo(i32 inreg %idx) {
 ; GFX9-PAL-NEXT:    s_add_i32 s0, s0, 4
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_sindex_foo:
@@ -656,10 +574,6 @@ define amdgpu_ps void @store_load_sindex_foo(i32 inreg %idx) {
 ; GFX940-NEXT:    s_add_i32 s0, s0, 4
 ; GFX940-NEXT:    scratch_load_dword v0, off, s0 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX10-PAL-LABEL: store_load_sindex_foo:
@@ -683,10 +597,6 @@ define amdgpu_ps void @store_load_sindex_foo(i32 inreg %idx) {
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v0
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_sindex_foo:
@@ -701,11 +611,20 @@ define amdgpu_ps void @store_load_sindex_foo(i32 inreg %idx) {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
+; GCN-LABEL: store_load_sindex_foo:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_lshl_b32 s1, s0, 2
+; GCN-NEXT:    s_and_b32 s0, s0, 15
+; GCN-NEXT:    s_lshl_b32 s0, s0, 2
+; GCN-NEXT:    s_add_u32 s1, 4, s1
+; GCN-NEXT:    v_mov_b32_e32 v0, 15
+; GCN-NEXT:    scratch_store_dword off, v0, s1 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_add_u32 s0, 4, s0
+; GCN-NEXT:    scratch_load_dword v0, off, s0 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_endpgm
 bb:
   %i = alloca [32 x float], align 4, addrspace(5)
   %i1 = bitcast [32 x float] addrspace(5)* %i to i8 addrspace(5)*
@@ -716,7 +635,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -733,10 +651,6 @@ define amdgpu_kernel void @store_load_vindex_kernel() {
 ; GFX9-NEXT:    v_sub_u32_e32 v0, 4, v0
 ; GFX9-NEXT:    scratch_load_dword v0, v0, off offset:124 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_vindex_kernel:
@@ -753,10 +667,6 @@ define amdgpu_kernel void @store_load_vindex_kernel() {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, v0, off offset:124 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_vindex_kernel:
@@ -768,10 +678,6 @@ define amdgpu_kernel void @store_load_vindex_kernel() {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, v2, off offset:124 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_vindex_kernel:
@@ -791,10 +697,6 @@ define amdgpu_kernel void @store_load_vindex_kernel() {
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, v0, off offset:124 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_vindex_kernel:
@@ -806,10 +708,6 @@ define amdgpu_kernel void @store_load_vindex_kernel() {
 ; GFX940-NEXT:    v_sub_u32_e32 v0, 4, v0
 ; GFX940-NEXT:    scratch_load_dword v0, v0, off offset:124 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX10-PAL-LABEL: store_load_vindex_kernel:
@@ -831,10 +729,6 @@ define amdgpu_kernel void @store_load_vindex_kernel() {
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    scratch_load_dword v0, v0, off offset:124 glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v0
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_vindex_kernel:
@@ -846,11 +740,17 @@ define amdgpu_kernel void @store_load_vindex_kernel() {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, v2, off offset:124 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
+; GCN-LABEL: store_load_vindex_kernel:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; GCN-NEXT:    v_mov_b32_e32 v1, 15
+; GCN-NEXT:    scratch_store_dword v0, v1, off offset:4 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    v_sub_u32_e32 v0, 4, v0
+; GCN-NEXT:    scratch_load_dword v0, v0, off offset:124 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_endpgm
 bb:
   %i = alloca [32 x float], align 4, addrspace(5)
   %i1 = bitcast [32 x float] addrspace(5)* %i to i8 addrspace(5)*
@@ -863,7 +763,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -880,9 +779,6 @@ define void @store_load_vindex_foo(i32 %idx) {
 ; GFX9-NEXT:    v_lshl_add_u32 v0, v0, 2, v1
 ; GFX9-NEXT:    scratch_load_dword v0, v0, off glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v1
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX10-LABEL: store_load_vindex_foo:
@@ -897,10 +793,6 @@ define void @store_load_vindex_foo(i32 %idx) {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, v1, off glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, s32
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX11-LABEL: store_load_vindex_foo:
@@ -915,10 +807,6 @@ define void @store_load_vindex_foo(i32 %idx) {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, v1, s32 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, s32
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX9-PAL-LABEL: store_load_vindex_foo:
@@ -933,9 +821,6 @@ define void @store_load_vindex_foo(i32 %idx) {
 ; GFX9-PAL-NEXT:    v_lshl_add_u32 v0, v0, 2, v1
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, v0, off glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v1
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX940-LABEL: store_load_vindex_foo:
@@ -949,10 +834,6 @@ define void @store_load_vindex_foo(i32 %idx) {
 ; GFX940-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX940-NEXT:    scratch_load_dword v0, v0, s32 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, s32
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX10-PAL-LABEL: store_load_vindex_foo:
@@ -967,10 +848,6 @@ define void @store_load_vindex_foo(i32 %idx) {
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    scratch_load_dword v0, v1, off glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v0, s32
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v0
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX11-PAL-LABEL: store_load_vindex_foo:
@@ -985,11 +862,19 @@ define void @store_load_vindex_foo(i32 %idx) {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, v1, s32 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, s32
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: store_load_vindex_foo:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v2, 15
+; GCN-NEXT:    v_lshlrev_b32_e32 v1, 2, v0
+; GCN-NEXT:    v_and_b32_e32 v0, v0, v2
+; GCN-NEXT:    scratch_store_dword v1, v2, s32 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; GCN-NEXT:    scratch_load_dword v0, v0, s32 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
 bb:
   %i = alloca [32 x float], align 4, addrspace(5)
   %i1 = bitcast [32 x float] addrspace(5)* %i to i8 addrspace(5)*
@@ -1000,7 +885,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -1064,6 +948,13 @@ define void @private_ptr_foo(float addrspace(5)* nocapture %arg) {
 ; GFX11-PAL-NEXT:    scratch_store_b32 v0, v1, off offset:4
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: private_ptr_foo:
+; GCN:       ; %bb.0:
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v1, 0x41200000
+; GCN-NEXT:    scratch_store_dword v0, v1, off offset:4
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
   %gep = getelementptr inbounds float, float addrspace(5)* %arg, i32 1
   store float 1.000000e+01, float addrspace(5)* %gep, align 4
   ret void
@@ -1086,22 +977,12 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s3
 ; GFX9-NEXT:    s_mov_b32 s1, 0
-; GFX9-NEXT:    s_mov_b32 s0, 0
 ; GFX9-NEXT:    s_mov_b32 vcc_lo, 0
 ; GFX9-NEXT:    s_mov_b32 vcc_hi, 0
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:260
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:276
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:292
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_hi offset:308
-; GFX9-NEXT:    s_nop 0
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: zero_init_small_offset_kernel:
@@ -1113,7 +994,6 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX10-NEXT:    scratch_load_dword v0, off, off offset:4 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-NEXT:    s_mov_b32 s0, 0
-; GFX10-NEXT:    v_mov_b32_e32 v4, 4
 ; GFX10-NEXT:    s_mov_b32 s1, s0
 ; GFX10-NEXT:    s_mov_b32 s2, s0
 ; GFX10-NEXT:    s_mov_b32 s3, s0
@@ -1121,17 +1001,10 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX10-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX10-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX10-NEXT:    v_mov_b32_e32 v3, s3
-; GFX10-NEXT:    v_mov_b32_e32 v5, 0x104
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:260
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:276
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:292
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:308
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v4
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v5
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: zero_init_small_offset_kernel:
@@ -1139,7 +1012,7 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX11-NEXT:    scratch_load_b32 v0, off, off offset:4 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-NEXT:    s_mov_b32 s0, 0
-; GFX11-NEXT:    v_dual_mov_b32 v4, 4 :: v_dual_mov_b32 v5, 0x104
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-NEXT:    s_mov_b32 s1, s0
 ; GFX11-NEXT:    s_mov_b32 s2, s0
 ; GFX11-NEXT:    s_mov_b32 s3, s0
@@ -1150,12 +1023,6 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], off offset:276
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], off offset:292
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], off offset:308
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v4
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v5
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
 ; GFX11-NEXT:    s_endpgm
 ;
@@ -1182,20 +1049,10 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v3, s3
 ; GFX9-PAL-NEXT:    s_mov_b32 s1, 0
-; GFX9-PAL-NEXT:    s_mov_b32 s0, 0
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:260
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:276
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:292
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_hi offset:308
-; GFX9-PAL-NEXT:    s_nop 0
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: zero_init_small_offset_kernel:
@@ -1212,15 +1069,6 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:276
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:292
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:308
-; GFX940-NEXT:    s_nop 1
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX1010-PAL-LABEL: zero_init_small_offset_kernel:
@@ -1247,20 +1095,11 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX1010-PAL-NEXT:    v_mov_b32_e32 v3, s3
 ; GFX1010-PAL-NEXT:    s_mov_b32 s2, 0
 ; GFX1010-PAL-NEXT:    s_mov_b32 s1, 0
-; GFX1010-PAL-NEXT:    s_mov_b32 s0, 0
 ; GFX1010-PAL-NEXT:    s_mov_b32 vcc_lo, 0
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s2 offset:260
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v4, 4
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v5, 0x104
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:276
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:292
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:308
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v4
-; GFX1010-PAL-NEXT:    ;;#ASMEND
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v5
-; GFX1010-PAL-NEXT:    ;;#ASMEND
 ; GFX1010-PAL-NEXT:    s_endpgm
 ;
 ; GFX1030-PAL-LABEL: zero_init_small_offset_kernel:
@@ -1277,7 +1116,6 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX1030-PAL-NEXT:    scratch_load_dword v0, off, off offset:4 glc dlc
 ; GFX1030-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX1030-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v4, 4
 ; GFX1030-PAL-NEXT:    s_mov_b32 s1, s0
 ; GFX1030-PAL-NEXT:    s_mov_b32 s2, s0
 ; GFX1030-PAL-NEXT:    s_mov_b32 s3, s0
@@ -1285,17 +1123,10 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX1030-PAL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX1030-PAL-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX1030-PAL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v5, 0x104
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:260
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:276
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:292
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], off offset:308
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v4
-; GFX1030-PAL-NEXT:    ;;#ASMEND
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v5
-; GFX1030-PAL-NEXT:    ;;#ASMEND
 ; GFX1030-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: zero_init_small_offset_kernel:
@@ -1303,7 +1134,7 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, off offset:4 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX11-PAL-NEXT:    v_dual_mov_b32 v4, 4 :: v_dual_mov_b32 v5, 0x104
+; GFX11-PAL-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-PAL-NEXT:    s_mov_b32 s1, s0
 ; GFX11-PAL-NEXT:    s_mov_b32 s2, s0
 ; GFX11-PAL-NEXT:    s_mov_b32 s3, s0
@@ -1314,12 +1145,6 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], off offset:276
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], off offset:292
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], off offset:308
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v4
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v5
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
 ; GFX11-PAL-NEXT:    s_endpgm
   %padding = alloca [64 x i32], align 4, addrspace(5)
@@ -1328,8 +1153,6 @@ define amdgpu_kernel void @zero_init_small_offset_kernel() {
   %pad_load = load volatile i32, i32 addrspace(5)* %pad_gep, align 4
   %cast = bitcast [32 x i16] addrspace(5)* %alloca to i8 addrspace(5)*
   call void @llvm.memset.p5i8.i64(i8 addrspace(5)* align 2 dereferenceable(64) %cast, i8 0, i64 64, i1 false)
-  call void asm sideeffect "; use $0", "s"([64 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x i16] addrspace(5)* %alloca) #0
   ret void
 }
 
@@ -1351,15 +1174,6 @@ define void @zero_init_small_offset_foo() {
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:272
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:288
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:304
-; GFX9-NEXT:    s_add_i32 vcc_hi, s32, 0x100
-; GFX9-NEXT:    v_mov_b32_e32 v0, s32
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -1370,7 +1184,6 @@ define void @zero_init_small_offset_foo() {
 ; GFX10-NEXT:    scratch_load_dword v0, off, s32 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-NEXT:    s_mov_b32 s0, 0
-; GFX10-NEXT:    s_add_i32 vcc_lo, s32, 0x100
 ; GFX10-NEXT:    s_mov_b32 s1, s0
 ; GFX10-NEXT:    s_mov_b32 s2, s0
 ; GFX10-NEXT:    s_mov_b32 s3, s0
@@ -1378,18 +1191,10 @@ define void @zero_init_small_offset_foo() {
 ; GFX10-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX10-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX10-NEXT:    v_mov_b32_e32 v3, s3
-; GFX10-NEXT:    v_mov_b32_e32 v4, s32
-; GFX10-NEXT:    v_mov_b32_e32 v5, vcc_lo
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:256
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:272
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:288
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:304
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v4
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v5
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -1400,24 +1205,17 @@ define void @zero_init_small_offset_foo() {
 ; GFX11-NEXT:    scratch_load_b32 v0, off, s32 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-NEXT:    s_mov_b32 s0, 0
-; GFX11-NEXT:    s_add_i32 vcc_lo, s32, 0x100
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-NEXT:    s_mov_b32 s1, s0
 ; GFX11-NEXT:    s_mov_b32 s2, s0
 ; GFX11-NEXT:    s_mov_b32 s3, s0
 ; GFX11-NEXT:    v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1
 ; GFX11-NEXT:    v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3
-; GFX11-NEXT:    v_dual_mov_b32 v4, s32 :: v_dual_mov_b32 v5, vcc_lo
 ; GFX11-NEXT:    s_clause 0x3
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:256
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:272
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:288
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:304
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v4
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v5
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -1438,15 +1236,6 @@ define void @zero_init_small_offset_foo() {
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:272
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:288
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:304
-; GFX9-PAL-NEXT:    s_add_i32 vcc_hi, s32, 0x100
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, s32
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -1465,16 +1254,6 @@ define void @zero_init_small_offset_foo() {
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:272
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:288
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:304
-; GFX940-NEXT:    s_add_i32 vcc_hi, s32, 0x100
-; GFX940-NEXT:    s_nop 0
-; GFX940-NEXT:    v_mov_b32_e32 v0, s32
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
 ; GFX940-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -1485,7 +1264,6 @@ define void @zero_init_small_offset_foo() {
 ; GFX10-PAL-NEXT:    scratch_load_dword v0, off, s32 glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX10-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x100
 ; GFX10-PAL-NEXT:    s_mov_b32 s1, s0
 ; GFX10-PAL-NEXT:    s_mov_b32 s2, s0
 ; GFX10-PAL-NEXT:    s_mov_b32 s3, s0
@@ -1493,18 +1271,10 @@ define void @zero_init_small_offset_foo() {
 ; GFX10-PAL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX10-PAL-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX10-PAL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v4, s32
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v5, vcc_lo
 ; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:256
 ; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:272
 ; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:288
 ; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:304
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v4
-; GFX10-PAL-NEXT:    ;;#ASMEND
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v5
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -1515,34 +1285,42 @@ define void @zero_init_small_offset_foo() {
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, s32 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX11-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x100
+; GFX11-PAL-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-PAL-NEXT:    s_mov_b32 s1, s0
 ; GFX11-PAL-NEXT:    s_mov_b32 s2, s0
 ; GFX11-PAL-NEXT:    s_mov_b32 s3, s0
 ; GFX11-PAL-NEXT:    v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1
 ; GFX11-PAL-NEXT:    v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3
-; GFX11-PAL-NEXT:    v_dual_mov_b32 v4, s32 :: v_dual_mov_b32 v5, vcc_lo
 ; GFX11-PAL-NEXT:    s_clause 0x3
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:256
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:272
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:288
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s32 offset:304
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v4
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v5
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: zero_init_small_offset_foo:
+; GCN:       ; %bb.0:
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    scratch_load_dword v0, off, s32 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_mov_b32 s0, 0
+; GCN-NEXT:    s_mov_b32 s1, s0
+; GCN-NEXT:    s_mov_b32 s2, s0
+; GCN-NEXT:    s_mov_b32 s3, s0
+; GCN-NEXT:    v_mov_b64_e32 v[0:1], s[0:1]
+; GCN-NEXT:    v_mov_b64_e32 v[2:3], s[2:3]
+; GCN-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:256
+; GCN-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:272
+; GCN-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:288
+; GCN-NEXT:    scratch_store_dwordx4 off, v[0:3], s32 offset:304
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
   %padding = alloca [64 x i32], align 4, addrspace(5)
   %alloca = alloca [32 x i16], align 2, addrspace(5)
   %pad_gep = getelementptr inbounds [64 x i32], [64 x i32] addrspace(5)* %padding, i32 0, i32 undef
   %pad_load = load volatile i32, i32 addrspace(5)* %pad_gep, align 4
   %cast = bitcast [32 x i16] addrspace(5)* %alloca to i8 addrspace(5)*
   call void @llvm.memset.p5i8.i64(i8 addrspace(5)* align 2 dereferenceable(64) %cast, i8 0, i64 64, i1 false)
-  call void asm sideeffect "; use $0", "s"([64 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x i16] addrspace(5)* %alloca) #0
   ret void
 }
 
@@ -1565,14 +1343,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX9-NEXT:    s_addk_i32 s0, 0x104
 ; GFX9-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_sindex_small_offset_kernel:
@@ -1585,7 +1355,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX10-NEXT:    scratch_load_dword v0, off, off offset:4 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-NEXT:    v_mov_b32_e32 v0, 15
-; GFX10-NEXT:    v_mov_b32_e32 v1, 0x104
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    s_and_b32 s1, s0, 15
 ; GFX10-NEXT:    s_lshl_b32 s0, s0, 2
@@ -1596,13 +1365,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v1
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_sindex_small_offset_kernel:
@@ -1610,7 +1372,7 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX11-NEXT:    s_load_b32 s0, s[0:1], 0x24
 ; GFX11-NEXT:    scratch_load_b32 v0, off, off offset:4 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_dual_mov_b32 v0, 15 :: v_dual_mov_b32 v1, 0x104
+; GFX11-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-NEXT:    s_and_b32 s1, s0, 15
 ; GFX11-NEXT:    s_lshl_b32 s0, s0, 2
@@ -1621,13 +1383,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v1
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_sindex_small_offset_kernel:
@@ -1653,14 +1408,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX9-PAL-NEXT:    s_addk_i32 s0, 0x104
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_sindex_small_offset_kernel:
@@ -1679,14 +1426,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX940-NEXT:    s_addk_i32 s0, 0x104
 ; GFX940-NEXT:    scratch_load_dword v0, off, s0 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX1010-PAL-LABEL: store_load_sindex_small_offset_kernel:
@@ -1702,7 +1441,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX1010-PAL-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_HI), s5
 ; GFX1010-PAL-NEXT:    s_load_dword s0, s[0:1], 0x0
 ; GFX1010-PAL-NEXT:    s_mov_b32 vcc_lo, 0
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v1, 0x104
 ; GFX1010-PAL-NEXT:    scratch_load_dword v0, off, vcc_lo offset:4 glc dlc
 ; GFX1010-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX1010-PAL-NEXT:    v_mov_b32_e32 v0, 15
@@ -1716,13 +1454,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX1010-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1010-PAL-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX1010-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v0
-; GFX1010-PAL-NEXT:    ;;#ASMEND
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v1
-; GFX1010-PAL-NEXT:    ;;#ASMEND
 ; GFX1010-PAL-NEXT:    s_endpgm
 ;
 ; GFX1030-PAL-LABEL: store_load_sindex_small_offset_kernel:
@@ -1740,7 +1471,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX1030-PAL-NEXT:    scratch_load_dword v0, off, off offset:4 glc dlc
 ; GFX1030-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX1030-PAL-NEXT:    v_mov_b32_e32 v0, 15
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v1, 0x104
 ; GFX1030-PAL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1030-PAL-NEXT:    s_and_b32 s1, s0, 15
 ; GFX1030-PAL-NEXT:    s_lshl_b32 s0, s0, 2
@@ -1751,13 +1481,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX1030-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1030-PAL-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX1030-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v0
-; GFX1030-PAL-NEXT:    ;;#ASMEND
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v1
-; GFX1030-PAL-NEXT:    ;;#ASMEND
 ; GFX1030-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_sindex_small_offset_kernel:
@@ -1765,7 +1488,7 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX11-PAL-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, off offset:4 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_dual_mov_b32 v0, 15 :: v_dual_mov_b32 v1, 0x104
+; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX11-PAL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-PAL-NEXT:    s_and_b32 s1, s0, 15
 ; GFX11-PAL-NEXT:    s_lshl_b32 s0, s0, 2
@@ -1776,13 +1499,6 @@ define amdgpu_kernel void @store_load_sindex_small_offset_kernel(i32 %idx) {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v1
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
 bb:
   %padding = alloca [64 x i32], align 4, addrspace(5)
@@ -1797,8 +1513,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([64 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -1820,14 +1534,6 @@ define amdgpu_ps void @store_load_sindex_small_offset_foo(i32 inreg %idx) {
 ; GFX9-NEXT:    s_addk_i32 s0, 0x104
 ; GFX9-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_sindex_small_offset_foo:
@@ -1848,21 +1554,13 @@ define amdgpu_ps void @store_load_sindex_small_offset_foo(i32 inreg %idx) {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, off, s0 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    v_mov_b32_e32 v1, 0x104
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v1
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_sindex_small_offset_foo:
 ; GFX11:       ; %bb.0: ; %bb
 ; GFX11-NEXT:    scratch_load_b32 v0, off, off offset:4 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_dual_mov_b32 v0, 15 :: v_dual_mov_b32 v1, 0x104
+; GFX11-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX11-NEXT:    s_and_b32 s1, s0, 15
 ; GFX11-NEXT:    s_lshl_b32 s0, s0, 2
 ; GFX11-NEXT:    s_lshl_b32 s1, s1, 2
@@ -1872,13 +1570,6 @@ define amdgpu_ps void @store_load_sindex_small_offset_foo(i32 inreg %idx) {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v1
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_sindex_small_offset_foo:
@@ -1903,14 +1594,6 @@ define amdgpu_ps void @store_load_sindex_small_offset_foo(i32 inreg %idx) {
 ; GFX9-PAL-NEXT:    s_addk_i32 s0, 0x104
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_sindex_small_offset_foo:
@@ -1927,14 +1610,6 @@ define amdgpu_ps void @store_load_sindex_small_offset_foo(i32 inreg %idx) {
 ; GFX940-NEXT:    s_addk_i32 s0, 0x104
 ; GFX940-NEXT:    scratch_load_dword v0, off, s0 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX1010-PAL-LABEL: store_load_sindex_small_offset_foo:
@@ -1961,14 +1636,6 @@ define amdgpu_ps void @store_load_sindex_small_offset_foo(i32 inreg %idx) {
 ; GFX1010-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1010-PAL-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX1010-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v1, 0x104
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v0
-; GFX1010-PAL-NEXT:    ;;#ASMEND
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v1
-; GFX1010-PAL-NEXT:    ;;#ASMEND
 ; GFX1010-PAL-NEXT:    s_endpgm
 ;
 ; GFX1030-PAL-LABEL: store_load_sindex_small_offset_foo:
@@ -1994,21 +1661,13 @@ define amdgpu_ps void @store_load_sindex_small_offset_foo(i32 inreg %idx) {
 ; GFX1030-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1030-PAL-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX1030-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v1, 0x104
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v0
-; GFX1030-PAL-NEXT:    ;;#ASMEND
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v1
-; GFX1030-PAL-NEXT:    ;;#ASMEND
 ; GFX1030-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_sindex_small_offset_foo:
 ; GFX11-PAL:       ; %bb.0: ; %bb
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, off offset:4 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_dual_mov_b32 v0, 15 :: v_dual_mov_b32 v1, 0x104
+; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX11-PAL-NEXT:    s_and_b32 s1, s0, 15
 ; GFX11-PAL-NEXT:    s_lshl_b32 s0, s0, 2
 ; GFX11-PAL-NEXT:    s_lshl_b32 s1, s1, 2
@@ -2018,13 +1677,6 @@ define amdgpu_ps void @store_load_sindex_small_offset_foo(i32 inreg %idx) {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v1
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
 bb:
   %padding = alloca [64 x i32], align 4, addrspace(5)
@@ -2039,8 +1691,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([64 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -2060,14 +1710,6 @@ define amdgpu_kernel void @store_load_vindex_small_offset_kernel() {
 ; GFX9-NEXT:    v_sub_u32_e32 v0, 0x104, v0
 ; GFX9-NEXT:    scratch_load_dword v0, v0, off offset:124 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX9-NEXT:    v_mov_b32_e32 v1, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v1
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_vindex_small_offset_kernel:
@@ -2086,14 +1728,6 @@ define amdgpu_kernel void @store_load_vindex_small_offset_kernel() {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, v0, off offset:124 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    v_mov_b32_e32 v1, 0x104
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v1
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_vindex_small_offset_kernel:
@@ -2104,16 +1738,8 @@ define amdgpu_kernel void @store_load_vindex_small_offset_kernel() {
 ; GFX11-NEXT:    v_sub_nc_u32_e32 v2, 0x104, v0
 ; GFX11-NEXT:    scratch_store_b32 v0, v1, off offset:260 dlc
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-NEXT:    v_mov_b32_e32 v1, 0x104
 ; GFX11-NEXT:    scratch_load_b32 v0, v2, off offset:124 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v1
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_vindex_small_offset_kernel:
@@ -2136,14 +1762,6 @@ define amdgpu_kernel void @store_load_vindex_small_offset_kernel() {
 ; GFX9-PAL-NEXT:    v_sub_u32_e32 v0, 0x104, v0
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, v0, off offset:124 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v1, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v1
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_vindex_small_offset_kernel:
@@ -2157,14 +1775,6 @@ define amdgpu_kernel void @store_load_vindex_small_offset_kernel() {
 ; GFX940-NEXT:    v_sub_u32_e32 v0, 0x104, v0
 ; GFX940-NEXT:    scratch_load_dword v0, v0, off offset:124 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 0x104
-; GFX940-NEXT:    v_mov_b32_e32 v1, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v1
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX1010-PAL-LABEL: store_load_vindex_small_offset_kernel:
@@ -2189,14 +1799,6 @@ define amdgpu_kernel void @store_load_vindex_small_offset_kernel() {
 ; GFX1010-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1010-PAL-NEXT:    scratch_load_dword v0, v0, off offset:124 glc dlc
 ; GFX1010-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v1, 0x104
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v0
-; GFX1010-PAL-NEXT:    ;;#ASMEND
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v1
-; GFX1010-PAL-NEXT:    ;;#ASMEND
 ; GFX1010-PAL-NEXT:    s_endpgm
 ;
 ; GFX1030-PAL-LABEL: store_load_vindex_small_offset_kernel:
@@ -2220,14 +1822,6 @@ define amdgpu_kernel void @store_load_vindex_small_offset_kernel() {
 ; GFX1030-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1030-PAL-NEXT:    scratch_load_dword v0, v0, off offset:124 glc dlc
 ; GFX1030-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v1, 0x104
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v0
-; GFX1030-PAL-NEXT:    ;;#ASMEND
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v1
-; GFX1030-PAL-NEXT:    ;;#ASMEND
 ; GFX1030-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_vindex_small_offset_kernel:
@@ -2238,16 +1832,8 @@ define amdgpu_kernel void @store_load_vindex_small_offset_kernel() {
 ; GFX11-PAL-NEXT:    v_sub_nc_u32_e32 v2, 0x104, v0
 ; GFX11-PAL-NEXT:    scratch_store_b32 v0, v1, off offset:260 dlc
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v1, 0x104
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, v2, off offset:124 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v1
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
 bb:
   %padding = alloca [64 x i32], align 4, addrspace(5)
@@ -2264,8 +1850,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([64 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -2285,13 +1869,6 @@ define void @store_load_vindex_small_offset_foo(i32 %idx) {
 ; GFX9-NEXT:    v_lshl_add_u32 v0, v0, 2, v1
 ; GFX9-NEXT:    scratch_load_dword v0, v0, off glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, s32
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v1
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX10-LABEL: store_load_vindex_small_offset_foo:
@@ -2299,26 +1876,17 @@ define void @store_load_vindex_small_offset_foo(i32 %idx) {
 ; GFX10-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    v_and_b32_e32 v1, 15, v0
-; GFX10-NEXT:    s_add_i32 s1, s32, 0x100
 ; GFX10-NEXT:    s_add_i32 s0, s32, 0x100
-; GFX10-NEXT:    v_lshl_add_u32 v0, v0, 2, s1
+; GFX10-NEXT:    s_add_i32 vcc_lo, s32, 0x100
+; GFX10-NEXT:    v_lshl_add_u32 v0, v0, 2, s0
 ; GFX10-NEXT:    v_mov_b32_e32 v2, 15
-; GFX10-NEXT:    v_lshl_add_u32 v1, v1, 2, s0
+; GFX10-NEXT:    v_lshl_add_u32 v1, v1, 2, vcc_lo
 ; GFX10-NEXT:    scratch_load_dword v3, off, s32 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    s_add_i32 vcc_lo, s32, 0x100
 ; GFX10-NEXT:    scratch_store_dword v0, v2, off
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, v1, off glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, s32
-; GFX10-NEXT:    v_mov_b32_e32 v1, vcc_lo
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v1
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX11-LABEL: store_load_vindex_small_offset_foo:
@@ -2326,7 +1894,6 @@ define void @store_load_vindex_small_offset_foo(i32 %idx) {
 ; GFX11-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    v_dual_mov_b32 v2, 15 :: v_dual_and_b32 v1, 15, v0
-; GFX11-NEXT:    s_add_i32 vcc_lo, s32, 0x100
 ; GFX11-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-NEXT:    scratch_load_b32 v3, off, s32 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
@@ -2335,13 +1902,6 @@ define void @store_load_vindex_small_offset_foo(i32 %idx) {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, v1, s32 offset:256 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_dual_mov_b32 v0, s32 :: v_dual_mov_b32 v1, vcc_lo
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v1
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX9-PAL-LABEL: store_load_vindex_small_offset_foo:
@@ -2359,13 +1919,6 @@ define void @store_load_vindex_small_offset_foo(i32 %idx) {
 ; GFX9-PAL-NEXT:    v_lshl_add_u32 v0, v0, 2, v1
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, v0, off glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, s32
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v1
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX940-LABEL: store_load_vindex_small_offset_foo:
@@ -2381,15 +1934,6 @@ define void @store_load_vindex_small_offset_foo(i32 %idx) {
 ; GFX940-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX940-NEXT:    scratch_load_dword v0, v0, s32 offset:256 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, s32
-; GFX940-NEXT:    s_add_i32 vcc_hi, s32, 0x100
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX10-PAL-LABEL: store_load_vindex_small_offset_foo:
@@ -2397,26 +1941,17 @@ define void @store_load_vindex_small_offset_foo(i32 %idx) {
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    v_and_b32_e32 v1, 15, v0
-; GFX10-PAL-NEXT:    s_add_i32 s1, s32, 0x100
 ; GFX10-PAL-NEXT:    s_add_i32 s0, s32, 0x100
-; GFX10-PAL-NEXT:    v_lshl_add_u32 v0, v0, 2, s1
+; GFX10-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x100
+; GFX10-PAL-NEXT:    v_lshl_add_u32 v0, v0, 2, s0
 ; GFX10-PAL-NEXT:    v_mov_b32_e32 v2, 15
-; GFX10-PAL-NEXT:    v_lshl_add_u32 v1, v1, 2, s0
+; GFX10-PAL-NEXT:    v_lshl_add_u32 v1, v1, 2, vcc_lo
 ; GFX10-PAL-NEXT:    scratch_load_dword v3, off, s32 glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x100
 ; GFX10-PAL-NEXT:    scratch_store_dword v0, v2, off
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    scratch_load_dword v0, v1, off glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v0, s32
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v1, vcc_lo
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v0
-; GFX10-PAL-NEXT:    ;;#ASMEND
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v1
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX11-PAL-LABEL: store_load_vindex_small_offset_foo:
@@ -2424,7 +1959,6 @@ define void @store_load_vindex_small_offset_foo(i32 %idx) {
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    v_dual_mov_b32 v2, 15 :: v_dual_and_b32 v1, 15, v0
-; GFX11-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x100
 ; GFX11-PAL-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v3, off, s32 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
@@ -2433,14 +1967,21 @@ define void @store_load_vindex_small_offset_foo(i32 %idx) {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, v1, s32 offset:256 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_dual_mov_b32 v0, s32 :: v_dual_mov_b32 v1, vcc_lo
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v1
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: store_load_vindex_small_offset_foo:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    scratch_load_dword v1, off, s32 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v2, 15
+; GCN-NEXT:    v_lshlrev_b32_e32 v1, 2, v0
+; GCN-NEXT:    v_and_b32_e32 v0, v0, v2
+; GCN-NEXT:    scratch_store_dword v1, v2, s32 offset:256 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; GCN-NEXT:    scratch_load_dword v0, v0, s32 offset:256 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
 bb:
   %padding = alloca [64 x i32], align 4, addrspace(5)
   %i = alloca [32 x float], align 4, addrspace(5)
@@ -2454,8 +1995,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([64 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -2483,15 +2022,6 @@ define amdgpu_kernel void @zero_init_large_offset_kernel() {
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:16
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:32
 ; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_hi offset:48
-; GFX9-NEXT:    s_nop 0
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: zero_init_large_offset_kernel:
@@ -2514,18 +2044,10 @@ define amdgpu_kernel void @zero_init_large_offset_kernel() {
 ; GFX10-NEXT:    s_movk_i32 s2, 0x4004
 ; GFX10-NEXT:    s_movk_i32 s1, 0x4004
 ; GFX10-NEXT:    s_movk_i32 s0, 0x4004
-; GFX10-NEXT:    v_mov_b32_e32 v4, 4
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s2
-; GFX10-NEXT:    v_mov_b32_e32 v5, 0x4004
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:16
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:32
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:48
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v4
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v5
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: zero_init_large_offset_kernel:
@@ -2542,18 +2064,11 @@ define amdgpu_kernel void @zero_init_large_offset_kernel() {
 ; GFX11-NEXT:    s_movk_i32 s2, 0x4004
 ; GFX11-NEXT:    s_movk_i32 s1, 0x4004
 ; GFX11-NEXT:    s_movk_i32 s0, 0x4004
-; GFX11-NEXT:    v_dual_mov_b32 v4, 4 :: v_dual_mov_b32 v5, 0x4004
 ; GFX11-NEXT:    s_clause 0x3
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s2
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s1 offset:16
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s0 offset:32
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], vcc_lo offset:48
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v4
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v5
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
 ; GFX11-NEXT:    s_endpgm
 ;
@@ -2585,15 +2100,6 @@ define amdgpu_kernel void @zero_init_large_offset_kernel() {
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:16
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:32
 ; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_hi offset:48
-; GFX9-PAL-NEXT:    s_nop 0
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: zero_init_large_offset_kernel:
@@ -2614,15 +2120,6 @@ define amdgpu_kernel void @zero_init_large_offset_kernel() {
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:16
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:32
 ; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_hi offset:48
-; GFX940-NEXT:    s_nop 1
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX1010-PAL-LABEL: zero_init_large_offset_kernel:
@@ -2652,17 +2149,9 @@ define amdgpu_kernel void @zero_init_large_offset_kernel() {
 ; GFX1010-PAL-NEXT:    s_movk_i32 s0, 0x4004
 ; GFX1010-PAL-NEXT:    s_movk_i32 vcc_lo, 0x4004
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s2
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v4, 4
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v5, 0x4004
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:16
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:32
 ; GFX1010-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:48
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v4
-; GFX1010-PAL-NEXT:    ;;#ASMEND
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v5
-; GFX1010-PAL-NEXT:    ;;#ASMEND
 ; GFX1010-PAL-NEXT:    s_endpgm
 ;
 ; GFX1030-PAL-LABEL: zero_init_large_offset_kernel:
@@ -2690,18 +2179,10 @@ define amdgpu_kernel void @zero_init_large_offset_kernel() {
 ; GFX1030-PAL-NEXT:    s_movk_i32 s2, 0x4004
 ; GFX1030-PAL-NEXT:    s_movk_i32 s1, 0x4004
 ; GFX1030-PAL-NEXT:    s_movk_i32 s0, 0x4004
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v4, 4
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s2
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v5, 0x4004
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:16
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:32
 ; GFX1030-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:48
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v4
-; GFX1030-PAL-NEXT:    ;;#ASMEND
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v5
-; GFX1030-PAL-NEXT:    ;;#ASMEND
 ; GFX1030-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: zero_init_large_offset_kernel:
@@ -2718,18 +2199,11 @@ define amdgpu_kernel void @zero_init_large_offset_kernel() {
 ; GFX11-PAL-NEXT:    s_movk_i32 s2, 0x4004
 ; GFX11-PAL-NEXT:    s_movk_i32 s1, 0x4004
 ; GFX11-PAL-NEXT:    s_movk_i32 s0, 0x4004
-; GFX11-PAL-NEXT:    v_dual_mov_b32 v4, 4 :: v_dual_mov_b32 v5, 0x4004
 ; GFX11-PAL-NEXT:    s_clause 0x3
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s2
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s1 offset:16
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s0 offset:32
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], vcc_lo offset:48
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v4
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v5
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
 ; GFX11-PAL-NEXT:    s_endpgm
   %padding = alloca [4096 x i32], align 4, addrspace(5)
@@ -2738,8 +2212,6 @@ define amdgpu_kernel void @zero_init_large_offset_kernel() {
   %pad_load = load volatile i32, i32 addrspace(5)* %pad_gep, align 4
   %cast = bitcast [32 x i16] addrspace(5)* %alloca to i8 addrspace(5)*
   call void @llvm.memset.p5i8.i64(i8 addrspace(5)* align 2 dereferenceable(64) %cast, i8 0, i64 64, i1 false)
-  call void asm sideeffect "; use $0", "s"([4096 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x i16] addrspace(5)* %alloca) #0
   ret void
 }
 
@@ -2757,24 +2229,14 @@ define void @zero_init_large_offset_foo() {
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s3
-; GFX9-NEXT:    s_add_i32 s3, s32, 0x4004
-; GFX9-NEXT:    s_add_i32 s2, s32, 0x4004
 ; GFX9-NEXT:    s_add_i32 s1, s32, 0x4004
 ; GFX9-NEXT:    s_add_i32 s0, s32, 0x4004
-; GFX9-NEXT:    s_add_i32 vcc_lo, s32, 4
-; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s3
-; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s2 offset:16
-; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:32
-; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:48
+; GFX9-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
 ; GFX9-NEXT:    s_add_i32 vcc_hi, s32, 0x4004
-; GFX9-NEXT:    v_mov_b32_e32 v0, vcc_lo
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
+; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s1
+; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:16
+; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:32
+; GFX9-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_hi offset:48
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -2785,7 +2247,6 @@ define void @zero_init_large_offset_foo() {
 ; GFX10-NEXT:    scratch_load_dword v0, off, s32 offset:4 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-NEXT:    s_mov_b32 s0, 0
-; GFX10-NEXT:    s_add_i32 s4, s32, 0x4004
 ; GFX10-NEXT:    s_mov_b32 s1, s0
 ; GFX10-NEXT:    s_mov_b32 s2, s0
 ; GFX10-NEXT:    s_mov_b32 s3, s0
@@ -2793,23 +2254,14 @@ define void @zero_init_large_offset_foo() {
 ; GFX10-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX10-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX10-NEXT:    v_mov_b32_e32 v3, s3
-; GFX10-NEXT:    s_add_i32 s3, s32, 4
 ; GFX10-NEXT:    s_add_i32 s2, s32, 0x4004
 ; GFX10-NEXT:    s_add_i32 s1, s32, 0x4004
 ; GFX10-NEXT:    s_add_i32 s0, s32, 0x4004
 ; GFX10-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
-; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s4
-; GFX10-NEXT:    v_mov_b32_e32 v4, s3
-; GFX10-NEXT:    v_mov_b32_e32 v5, s2
+; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s2
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:16
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:32
 ; GFX10-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:48
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v4
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v5
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -2820,29 +2272,21 @@ define void @zero_init_large_offset_foo() {
 ; GFX11-NEXT:    scratch_load_b32 v0, off, s32 offset:4 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-NEXT:    s_mov_b32 s0, 0
-; GFX11-NEXT:    s_add_i32 s4, s32, 4
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-NEXT:    s_mov_b32 s1, s0
 ; GFX11-NEXT:    s_mov_b32 s2, s0
 ; GFX11-NEXT:    s_mov_b32 s3, s0
 ; GFX11-NEXT:    v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1
 ; GFX11-NEXT:    v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3
-; GFX11-NEXT:    s_add_i32 s3, s32, 0x4004
 ; GFX11-NEXT:    s_add_i32 s2, s32, 0x4004
 ; GFX11-NEXT:    s_add_i32 s1, s32, 0x4004
 ; GFX11-NEXT:    s_add_i32 s0, s32, 0x4004
 ; GFX11-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
-; GFX11-NEXT:    v_dual_mov_b32 v4, s4 :: v_dual_mov_b32 v5, s3
 ; GFX11-NEXT:    s_clause 0x3
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s2
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s1 offset:16
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], s0 offset:32
 ; GFX11-NEXT:    scratch_store_b128 off, v[0:3], vcc_lo offset:48
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v4
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v5
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -2859,24 +2303,14 @@ define void @zero_init_large_offset_foo() {
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX9-PAL-NEXT:    s_add_i32 s3, s32, 0x4004
-; GFX9-PAL-NEXT:    s_add_i32 s2, s32, 0x4004
 ; GFX9-PAL-NEXT:    s_add_i32 s1, s32, 0x4004
 ; GFX9-PAL-NEXT:    s_add_i32 s0, s32, 0x4004
-; GFX9-PAL-NEXT:    s_add_i32 vcc_lo, s32, 4
-; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s3
-; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s2 offset:16
-; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:32
-; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:48
+; GFX9-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
 ; GFX9-PAL-NEXT:    s_add_i32 vcc_hi, s32, 0x4004
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, vcc_lo
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
+; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s1
+; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:16
+; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:32
+; GFX9-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_hi offset:48
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -2891,25 +2325,14 @@ define void @zero_init_large_offset_foo() {
 ; GFX940-NEXT:    s_mov_b32 s3, s0
 ; GFX940-NEXT:    v_mov_b64_e32 v[0:1], s[0:1]
 ; GFX940-NEXT:    v_mov_b64_e32 v[2:3], s[2:3]
-; GFX940-NEXT:    s_add_i32 s3, s32, 0x4004
-; GFX940-NEXT:    s_add_i32 s2, s32, 0x4004
 ; GFX940-NEXT:    s_add_i32 s1, s32, 0x4004
 ; GFX940-NEXT:    s_add_i32 s0, s32, 0x4004
-; GFX940-NEXT:    s_add_i32 vcc_lo, s32, 4
-; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s3
-; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s2 offset:16
-; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:32
-; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:48
+; GFX940-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
 ; GFX940-NEXT:    s_add_i32 vcc_hi, s32, 0x4004
-; GFX940-NEXT:    s_nop 0
-; GFX940-NEXT:    v_mov_b32_e32 v0, vcc_lo
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
+; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s1
+; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:16
+; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:32
+; GFX940-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_hi offset:48
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
 ; GFX940-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -2920,7 +2343,6 @@ define void @zero_init_large_offset_foo() {
 ; GFX10-PAL-NEXT:    scratch_load_dword v0, off, s32 offset:4 glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX10-PAL-NEXT:    s_add_i32 s4, s32, 0x4004
 ; GFX10-PAL-NEXT:    s_mov_b32 s1, s0
 ; GFX10-PAL-NEXT:    s_mov_b32 s2, s0
 ; GFX10-PAL-NEXT:    s_mov_b32 s3, s0
@@ -2928,23 +2350,14 @@ define void @zero_init_large_offset_foo() {
 ; GFX10-PAL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX10-PAL-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX10-PAL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX10-PAL-NEXT:    s_add_i32 s3, s32, 4
 ; GFX10-PAL-NEXT:    s_add_i32 s2, s32, 0x4004
 ; GFX10-PAL-NEXT:    s_add_i32 s1, s32, 0x4004
 ; GFX10-PAL-NEXT:    s_add_i32 s0, s32, 0x4004
 ; GFX10-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
-; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s4
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v4, s3
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v5, s2
+; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s2
 ; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s1 offset:16
 ; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], s0 offset:32
 ; GFX10-PAL-NEXT:    scratch_store_dwordx4 off, v[0:3], vcc_lo offset:48
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v4
-; GFX10-PAL-NEXT:    ;;#ASMEND
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v5
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
@@ -2955,29 +2368,21 @@ define void @zero_init_large_offset_foo() {
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, s32 offset:4 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-PAL-NEXT:    s_mov_b32 s0, 0
-; GFX11-PAL-NEXT:    s_add_i32 s4, s32, 4
+; GFX11-PAL-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-PAL-NEXT:    s_mov_b32 s1, s0
 ; GFX11-PAL-NEXT:    s_mov_b32 s2, s0
 ; GFX11-PAL-NEXT:    s_mov_b32 s3, s0
 ; GFX11-PAL-NEXT:    v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1
 ; GFX11-PAL-NEXT:    v_dual_mov_b32 v2, s2 :: v_dual_mov_b32 v3, s3
-; GFX11-PAL-NEXT:    s_add_i32 s3, s32, 0x4004
 ; GFX11-PAL-NEXT:    s_add_i32 s2, s32, 0x4004
 ; GFX11-PAL-NEXT:    s_add_i32 s1, s32, 0x4004
 ; GFX11-PAL-NEXT:    s_add_i32 s0, s32, 0x4004
 ; GFX11-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
-; GFX11-PAL-NEXT:    v_dual_mov_b32 v4, s4 :: v_dual_mov_b32 v5, s3
 ; GFX11-PAL-NEXT:    s_clause 0x3
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s2
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s1 offset:16
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], s0 offset:32
 ; GFX11-PAL-NEXT:    scratch_store_b128 off, v[0:3], vcc_lo offset:48
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v4
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v5
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
   %padding = alloca [4096 x i32], align 4, addrspace(5)
@@ -2986,8 +2391,6 @@ define void @zero_init_large_offset_foo() {
   %pad_load = load volatile i32, i32 addrspace(5)* %pad_gep, align 4
   %cast = bitcast [32 x i16] addrspace(5)* %alloca to i8 addrspace(5)*
   call void @llvm.memset.p5i8.i64(i8 addrspace(5)* align 2 dereferenceable(64) %cast, i8 0, i64 64, i1 false)
-  call void asm sideeffect "; use $0", "s"([4096 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x i16] addrspace(5)* %alloca) #0
   ret void
 }
 
@@ -3010,14 +2413,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX9-NEXT:    s_addk_i32 s0, 0x4004
 ; GFX9-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_sindex_large_offset_kernel:
@@ -3030,7 +2425,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX10-NEXT:    scratch_load_dword v0, off, off offset:4 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-NEXT:    v_mov_b32_e32 v0, 15
-; GFX10-NEXT:    v_mov_b32_e32 v1, 0x4004
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    s_and_b32 s1, s0, 15
 ; GFX10-NEXT:    s_lshl_b32 s0, s0, 2
@@ -3041,13 +2435,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v1
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_sindex_large_offset_kernel:
@@ -3055,7 +2442,7 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX11-NEXT:    s_load_b32 s0, s[0:1], 0x24
 ; GFX11-NEXT:    scratch_load_b32 v0, off, off offset:4 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_dual_mov_b32 v0, 15 :: v_dual_mov_b32 v1, 0x4004
+; GFX11-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-NEXT:    s_and_b32 s1, s0, 15
 ; GFX11-NEXT:    s_lshl_b32 s0, s0, 2
@@ -3066,13 +2453,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v1
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_sindex_large_offset_kernel:
@@ -3098,14 +2478,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX9-PAL-NEXT:    s_addk_i32 s0, 0x4004
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_sindex_large_offset_kernel:
@@ -3124,14 +2496,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX940-NEXT:    s_addk_i32 s0, 0x4004
 ; GFX940-NEXT:    scratch_load_dword v0, off, s0 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX1010-PAL-LABEL: store_load_sindex_large_offset_kernel:
@@ -3147,7 +2511,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX1010-PAL-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_HI), s5
 ; GFX1010-PAL-NEXT:    s_load_dword s0, s[0:1], 0x0
 ; GFX1010-PAL-NEXT:    s_mov_b32 vcc_lo, 0
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v1, 0x4004
 ; GFX1010-PAL-NEXT:    scratch_load_dword v0, off, vcc_lo offset:4 glc dlc
 ; GFX1010-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX1010-PAL-NEXT:    v_mov_b32_e32 v0, 15
@@ -3161,13 +2524,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX1010-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1010-PAL-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX1010-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v0
-; GFX1010-PAL-NEXT:    ;;#ASMEND
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v1
-; GFX1010-PAL-NEXT:    ;;#ASMEND
 ; GFX1010-PAL-NEXT:    s_endpgm
 ;
 ; GFX1030-PAL-LABEL: store_load_sindex_large_offset_kernel:
@@ -3185,7 +2541,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX1030-PAL-NEXT:    scratch_load_dword v0, off, off offset:4 glc dlc
 ; GFX1030-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX1030-PAL-NEXT:    v_mov_b32_e32 v0, 15
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v1, 0x4004
 ; GFX1030-PAL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1030-PAL-NEXT:    s_and_b32 s1, s0, 15
 ; GFX1030-PAL-NEXT:    s_lshl_b32 s0, s0, 2
@@ -3196,13 +2551,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX1030-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1030-PAL-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX1030-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v0
-; GFX1030-PAL-NEXT:    ;;#ASMEND
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v1
-; GFX1030-PAL-NEXT:    ;;#ASMEND
 ; GFX1030-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_sindex_large_offset_kernel:
@@ -3210,7 +2558,7 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX11-PAL-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, off offset:4 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_dual_mov_b32 v0, 15 :: v_dual_mov_b32 v1, 0x4004
+; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX11-PAL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-PAL-NEXT:    s_and_b32 s1, s0, 15
 ; GFX11-PAL-NEXT:    s_lshl_b32 s0, s0, 2
@@ -3221,13 +2569,6 @@ define amdgpu_kernel void @store_load_sindex_large_offset_kernel(i32 %idx) {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v1
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
 bb:
   %padding = alloca [4096 x i32], align 4, addrspace(5)
@@ -3242,8 +2583,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([4096 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -3265,14 +2604,6 @@ define amdgpu_ps void @store_load_sindex_large_offset_foo(i32 inreg %idx) {
 ; GFX9-NEXT:    s_addk_i32 s0, 0x4004
 ; GFX9-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_sindex_large_offset_foo:
@@ -3293,21 +2624,13 @@ define amdgpu_ps void @store_load_sindex_large_offset_foo(i32 inreg %idx) {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, off, s0 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    v_mov_b32_e32 v1, 0x4004
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v1
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_sindex_large_offset_foo:
 ; GFX11:       ; %bb.0: ; %bb
 ; GFX11-NEXT:    scratch_load_b32 v0, off, off offset:4 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_dual_mov_b32 v0, 15 :: v_dual_mov_b32 v1, 0x4004
+; GFX11-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX11-NEXT:    s_and_b32 s1, s0, 15
 ; GFX11-NEXT:    s_lshl_b32 s0, s0, 2
 ; GFX11-NEXT:    s_lshl_b32 s1, s1, 2
@@ -3317,13 +2640,6 @@ define amdgpu_ps void @store_load_sindex_large_offset_foo(i32 inreg %idx) {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v1
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_sindex_large_offset_foo:
@@ -3348,14 +2664,6 @@ define amdgpu_ps void @store_load_sindex_large_offset_foo(i32 inreg %idx) {
 ; GFX9-PAL-NEXT:    s_addk_i32 s0, 0x4004
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, off, s0 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_sindex_large_offset_foo:
@@ -3372,14 +2680,6 @@ define amdgpu_ps void @store_load_sindex_large_offset_foo(i32 inreg %idx) {
 ; GFX940-NEXT:    s_addk_i32 s0, 0x4004
 ; GFX940-NEXT:    scratch_load_dword v0, off, s0 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX1010-PAL-LABEL: store_load_sindex_large_offset_foo:
@@ -3406,14 +2706,6 @@ define amdgpu_ps void @store_load_sindex_large_offset_foo(i32 inreg %idx) {
 ; GFX1010-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1010-PAL-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX1010-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v1, 0x4004
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v0
-; GFX1010-PAL-NEXT:    ;;#ASMEND
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v1
-; GFX1010-PAL-NEXT:    ;;#ASMEND
 ; GFX1010-PAL-NEXT:    s_endpgm
 ;
 ; GFX1030-PAL-LABEL: store_load_sindex_large_offset_foo:
@@ -3439,21 +2731,13 @@ define amdgpu_ps void @store_load_sindex_large_offset_foo(i32 inreg %idx) {
 ; GFX1030-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1030-PAL-NEXT:    scratch_load_dword v0, off, s1 glc dlc
 ; GFX1030-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v1, 0x4004
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v0
-; GFX1030-PAL-NEXT:    ;;#ASMEND
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v1
-; GFX1030-PAL-NEXT:    ;;#ASMEND
 ; GFX1030-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_sindex_large_offset_foo:
 ; GFX11-PAL:       ; %bb.0: ; %bb
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, off offset:4 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_dual_mov_b32 v0, 15 :: v_dual_mov_b32 v1, 0x4004
+; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX11-PAL-NEXT:    s_and_b32 s1, s0, 15
 ; GFX11-PAL-NEXT:    s_lshl_b32 s0, s0, 2
 ; GFX11-PAL-NEXT:    s_lshl_b32 s1, s1, 2
@@ -3463,13 +2747,6 @@ define amdgpu_ps void @store_load_sindex_large_offset_foo(i32 inreg %idx) {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, off, s1 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v1
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
 bb:
   %padding = alloca [4096 x i32], align 4, addrspace(5)
@@ -3484,8 +2761,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([4096 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -3505,14 +2780,6 @@ define amdgpu_kernel void @store_load_vindex_large_offset_kernel() {
 ; GFX9-NEXT:    v_sub_u32_e32 v0, 0x4004, v0
 ; GFX9-NEXT:    scratch_load_dword v0, v0, off offset:124 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX9-NEXT:    v_mov_b32_e32 v1, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v1
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_vindex_large_offset_kernel:
@@ -3531,14 +2798,6 @@ define amdgpu_kernel void @store_load_vindex_large_offset_kernel() {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, v0, off offset:124 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    v_mov_b32_e32 v1, 0x4004
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v1
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_vindex_large_offset_kernel:
@@ -3550,16 +2809,8 @@ define amdgpu_kernel void @store_load_vindex_large_offset_kernel() {
 ; GFX11-NEXT:    v_sub_nc_u32_e32 v2, 0x4004, v0
 ; GFX11-NEXT:    scratch_store_b32 v0, v1, vcc_lo dlc
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-NEXT:    v_mov_b32_e32 v1, 0x4004
 ; GFX11-NEXT:    scratch_load_b32 v0, v2, off offset:124 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v1
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_vindex_large_offset_kernel:
@@ -3582,14 +2833,6 @@ define amdgpu_kernel void @store_load_vindex_large_offset_kernel() {
 ; GFX9-PAL-NEXT:    v_sub_u32_e32 v0, 0x4004, v0
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, v0, off offset:124 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v1, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v1
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_vindex_large_offset_kernel:
@@ -3604,14 +2847,6 @@ define amdgpu_kernel void @store_load_vindex_large_offset_kernel() {
 ; GFX940-NEXT:    v_sub_u32_e32 v0, 0x4004, v0
 ; GFX940-NEXT:    scratch_load_dword v0, v0, off offset:124 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 0x4004
-; GFX940-NEXT:    v_mov_b32_e32 v1, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v1
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX1010-PAL-LABEL: store_load_vindex_large_offset_kernel:
@@ -3636,14 +2871,6 @@ define amdgpu_kernel void @store_load_vindex_large_offset_kernel() {
 ; GFX1010-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1010-PAL-NEXT:    scratch_load_dword v0, v0, off offset:124 glc dlc
 ; GFX1010-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v1, 0x4004
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v0
-; GFX1010-PAL-NEXT:    ;;#ASMEND
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v1
-; GFX1010-PAL-NEXT:    ;;#ASMEND
 ; GFX1010-PAL-NEXT:    s_endpgm
 ;
 ; GFX1030-PAL-LABEL: store_load_vindex_large_offset_kernel:
@@ -3667,14 +2894,6 @@ define amdgpu_kernel void @store_load_vindex_large_offset_kernel() {
 ; GFX1030-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1030-PAL-NEXT:    scratch_load_dword v0, v0, off offset:124 glc dlc
 ; GFX1030-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v1, 0x4004
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v0
-; GFX1030-PAL-NEXT:    ;;#ASMEND
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v1
-; GFX1030-PAL-NEXT:    ;;#ASMEND
 ; GFX1030-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_vindex_large_offset_kernel:
@@ -3686,16 +2905,8 @@ define amdgpu_kernel void @store_load_vindex_large_offset_kernel() {
 ; GFX11-PAL-NEXT:    v_sub_nc_u32_e32 v2, 0x4004, v0
 ; GFX11-PAL-NEXT:    scratch_store_b32 v0, v1, vcc_lo dlc
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v1, 0x4004
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, v2, off offset:124 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v1
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
 bb:
   %padding = alloca [4096 x i32], align 4, addrspace(5)
@@ -3712,8 +2923,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([4096 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -3723,8 +2932,8 @@ define void @store_load_vindex_large_offset_foo(i32 %idx) {
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    scratch_load_dword v1, off, s32 offset:4 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
-; GFX9-NEXT:    v_mov_b32_e32 v1, vcc_lo
+; GFX9-NEXT:    s_add_i32 vcc_hi, s32, 0x4004
+; GFX9-NEXT:    v_mov_b32_e32 v1, vcc_hi
 ; GFX9-NEXT:    v_lshl_add_u32 v2, v0, 2, v1
 ; GFX9-NEXT:    v_mov_b32_e32 v3, 15
 ; GFX9-NEXT:    v_and_b32_e32 v0, 15, v0
@@ -3733,14 +2942,6 @@ define void @store_load_vindex_large_offset_foo(i32 %idx) {
 ; GFX9-NEXT:    v_lshl_add_u32 v0, v0, 2, v1
 ; GFX9-NEXT:    scratch_load_dword v0, v0, off glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    s_add_i32 vcc_hi, s32, 4
-; GFX9-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v1
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX10-LABEL: store_load_vindex_large_offset_foo:
@@ -3748,27 +2949,17 @@ define void @store_load_vindex_large_offset_foo(i32 %idx) {
 ; GFX10-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    v_and_b32_e32 v1, 15, v0
-; GFX10-NEXT:    s_add_i32 s2, s32, 0x4004
-; GFX10-NEXT:    s_add_i32 s1, s32, 0x4004
-; GFX10-NEXT:    v_lshl_add_u32 v0, v0, 2, s2
+; GFX10-NEXT:    s_add_i32 s0, s32, 0x4004
+; GFX10-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
+; GFX10-NEXT:    v_lshl_add_u32 v0, v0, 2, s0
 ; GFX10-NEXT:    v_mov_b32_e32 v2, 15
-; GFX10-NEXT:    v_lshl_add_u32 v1, v1, 2, s1
+; GFX10-NEXT:    v_lshl_add_u32 v1, v1, 2, vcc_lo
 ; GFX10-NEXT:    scratch_load_dword v3, off, s32 offset:4 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    s_add_i32 s0, s32, 4
-; GFX10-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
 ; GFX10-NEXT:    scratch_store_dword v0, v2, off
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, v1, off glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-NEXT:    v_mov_b32_e32 v1, vcc_lo
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v1
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX11-LABEL: store_load_vindex_large_offset_foo:
@@ -3776,25 +2967,17 @@ define void @store_load_vindex_large_offset_foo(i32 %idx) {
 ; GFX11-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    v_dual_mov_b32 v2, 15 :: v_dual_and_b32 v1, 15, v0
-; GFX11-NEXT:    s_add_i32 s2, s32, 0x4004
-; GFX11-NEXT:    s_add_i32 s1, s32, 0x4004
-; GFX11-NEXT:    s_add_i32 s0, s32, 4
-; GFX11-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
 ; GFX11-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; GFX11-NEXT:    s_add_i32 s0, s32, 0x4004
+; GFX11-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_2)
 ; GFX11-NEXT:    v_lshlrev_b32_e32 v1, 2, v1
 ; GFX11-NEXT:    scratch_load_b32 v3, off, s32 offset:4 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    scratch_store_b32 v0, v2, s2 dlc
+; GFX11-NEXT:    scratch_store_b32 v0, v2, s0 dlc
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-NEXT:    scratch_load_b32 v0, v1, s1 glc dlc
+; GFX11-NEXT:    scratch_load_b32 v0, v1, vcc_lo glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, vcc_lo
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v1
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX9-PAL-LABEL: store_load_vindex_large_offset_foo:
@@ -3802,8 +2985,8 @@ define void @store_load_vindex_large_offset_foo(i32 %idx) {
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX9-PAL-NEXT:    scratch_load_dword v1, off, s32 offset:4 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v1, vcc_lo
+; GFX9-PAL-NEXT:    s_add_i32 vcc_hi, s32, 0x4004
+; GFX9-PAL-NEXT:    v_mov_b32_e32 v1, vcc_hi
 ; GFX9-PAL-NEXT:    v_lshl_add_u32 v2, v0, 2, v1
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v3, 15
 ; GFX9-PAL-NEXT:    v_and_b32_e32 v0, 15, v0
@@ -3812,14 +2995,6 @@ define void @store_load_vindex_large_offset_foo(i32 %idx) {
 ; GFX9-PAL-NEXT:    v_lshl_add_u32 v0, v0, 2, v1
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, v0, off glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    s_add_i32 vcc_hi, s32, 4
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v1
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX940-LABEL: store_load_vindex_large_offset_foo:
@@ -3829,24 +3004,14 @@ define void @store_load_vindex_large_offset_foo(i32 %idx) {
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
 ; GFX940-NEXT:    v_lshlrev_b32_e32 v1, 2, v0
 ; GFX940-NEXT:    v_mov_b32_e32 v2, 15
-; GFX940-NEXT:    s_add_i32 s1, s32, 0x4004
+; GFX940-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
 ; GFX940-NEXT:    v_and_b32_e32 v0, 15, v0
-; GFX940-NEXT:    scratch_store_dword v1, v2, s1 sc0 sc1
+; GFX940-NEXT:    scratch_store_dword v1, v2, vcc_lo sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
 ; GFX940-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; GFX940-NEXT:    s_add_i32 s0, s32, 0x4004
-; GFX940-NEXT:    scratch_load_dword v0, v0, s0 sc0 sc1
-; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    s_add_i32 vcc_lo, s32, 4
-; GFX940-NEXT:    v_mov_b32_e32 v0, vcc_lo
 ; GFX940-NEXT:    s_add_i32 vcc_hi, s32, 0x4004
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
-; GFX940-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
+; GFX940-NEXT:    scratch_load_dword v0, v0, vcc_hi sc0 sc1
+; GFX940-NEXT:    s_waitcnt vmcnt(0)
 ; GFX940-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX10-PAL-LABEL: store_load_vindex_large_offset_foo:
@@ -3854,27 +3019,17 @@ define void @store_load_vindex_large_offset_foo(i32 %idx) {
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    v_and_b32_e32 v1, 15, v0
-; GFX10-PAL-NEXT:    s_add_i32 s2, s32, 0x4004
-; GFX10-PAL-NEXT:    s_add_i32 s1, s32, 0x4004
-; GFX10-PAL-NEXT:    v_lshl_add_u32 v0, v0, 2, s2
+; GFX10-PAL-NEXT:    s_add_i32 s0, s32, 0x4004
+; GFX10-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
+; GFX10-PAL-NEXT:    v_lshl_add_u32 v0, v0, 2, s0
 ; GFX10-PAL-NEXT:    v_mov_b32_e32 v2, 15
-; GFX10-PAL-NEXT:    v_lshl_add_u32 v1, v1, 2, s1
+; GFX10-PAL-NEXT:    v_lshl_add_u32 v1, v1, 2, vcc_lo
 ; GFX10-PAL-NEXT:    scratch_load_dword v3, off, s32 offset:4 glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-PAL-NEXT:    s_add_i32 s0, s32, 4
-; GFX10-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
 ; GFX10-PAL-NEXT:    scratch_store_dword v0, v2, off
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    scratch_load_dword v0, v1, off glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v1, vcc_lo
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v0
-; GFX10-PAL-NEXT:    ;;#ASMEND
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v1
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX11-PAL-LABEL: store_load_vindex_large_offset_foo:
@@ -3882,26 +3037,34 @@ define void @store_load_vindex_large_offset_foo(i32 %idx) {
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    v_dual_mov_b32 v2, 15 :: v_dual_and_b32 v1, 15, v0
-; GFX11-PAL-NEXT:    s_add_i32 s2, s32, 0x4004
-; GFX11-PAL-NEXT:    s_add_i32 s1, s32, 0x4004
-; GFX11-PAL-NEXT:    s_add_i32 s0, s32, 4
-; GFX11-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
 ; GFX11-PAL-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; GFX11-PAL-NEXT:    s_add_i32 s0, s32, 0x4004
+; GFX11-PAL-NEXT:    s_add_i32 vcc_lo, s32, 0x4004
+; GFX11-PAL-NEXT:    s_delay_alu instid0(VALU_DEP_2)
 ; GFX11-PAL-NEXT:    v_lshlrev_b32_e32 v1, 2, v1
 ; GFX11-PAL-NEXT:    scratch_load_b32 v3, off, s32 offset:4 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    scratch_store_b32 v0, v2, s2 dlc
+; GFX11-PAL-NEXT:    scratch_store_b32 v0, v2, s0 dlc
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-PAL-NEXT:    scratch_load_b32 v0, v1, s1 glc dlc
+; GFX11-PAL-NEXT:    scratch_load_b32 v0, v1, vcc_lo glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, vcc_lo
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v1
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: store_load_vindex_large_offset_foo:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    scratch_load_dword v1, off, s32 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v2, 15
+; GCN-NEXT:    v_lshlrev_b32_e32 v1, 2, v0
+; GCN-NEXT:    v_and_b32_e32 v0, v0, v2
+; GCN-NEXT:    s_add_u32 vcc_hi, s32, 0x4000
+; GCN-NEXT:    scratch_store_dword v1, v2, vcc_hi sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; GCN-NEXT:    s_add_u32 vcc_hi, s32, 0x4000
+; GCN-NEXT:    scratch_load_dword v0, v0, vcc_hi sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
 bb:
   %padding = alloca [4096 x i32], align 4, addrspace(5)
   %i = alloca [32 x float], align 4, addrspace(5)
@@ -3915,8 +3078,6 @@ bb:
   %i10 = getelementptr inbounds [32 x float], [32 x float] addrspace(5)* %i, i32 0, i32 %i9
   %i11 = bitcast float addrspace(5)* %i10 to i32 addrspace(5)*
   %i12 = load volatile i32, i32 addrspace(5)* %i11, align 4
-  call void asm sideeffect "; use $0", "s"([4096 x i32] addrspace(5)* %padding) #0
-  call void asm sideeffect "; use $0", "s"([32 x float] addrspace(5)* %i) #0
   ret void
 }
 
@@ -3936,10 +3097,6 @@ define amdgpu_kernel void @store_load_large_imm_offset_kernel() {
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    scratch_load_dword v0, off, s0 offset:3712 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_large_imm_offset_kernel:
@@ -3958,10 +3115,6 @@ define amdgpu_kernel void @store_load_large_imm_offset_kernel() {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, off, s0 offset:1664 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_large_imm_offset_kernel:
@@ -3974,10 +3127,6 @@ define amdgpu_kernel void @store_load_large_imm_offset_kernel() {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, v1, off offset:3716 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_large_imm_offset_kernel:
@@ -4000,10 +3149,6 @@ define amdgpu_kernel void @store_load_large_imm_offset_kernel() {
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, off, s0 offset:3712 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_large_imm_offset_kernel:
@@ -4017,10 +3162,6 @@ define amdgpu_kernel void @store_load_large_imm_offset_kernel() {
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
 ; GFX940-NEXT:    scratch_load_dword v0, v0, off offset:3716 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX1010-PAL-LABEL: store_load_large_imm_offset_kernel:
@@ -4045,10 +3186,6 @@ define amdgpu_kernel void @store_load_large_imm_offset_kernel() {
 ; GFX1010-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1010-PAL-NEXT:    scratch_load_dword v0, off, s0 offset:1664 glc dlc
 ; GFX1010-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1010-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1010-PAL-NEXT:    ;;#ASMSTART
-; GFX1010-PAL-NEXT:    ; use v0
-; GFX1010-PAL-NEXT:    ;;#ASMEND
 ; GFX1010-PAL-NEXT:    s_endpgm
 ;
 ; GFX1030-PAL-LABEL: store_load_large_imm_offset_kernel:
@@ -4072,10 +3209,6 @@ define amdgpu_kernel void @store_load_large_imm_offset_kernel() {
 ; GFX1030-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1030-PAL-NEXT:    scratch_load_dword v0, off, s0 offset:1664 glc dlc
 ; GFX1030-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX1030-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX1030-PAL-NEXT:    ;;#ASMSTART
-; GFX1030-PAL-NEXT:    ; use v0
-; GFX1030-PAL-NEXT:    ;;#ASMEND
 ; GFX1030-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_large_imm_offset_kernel:
@@ -4088,10 +3221,6 @@ define amdgpu_kernel void @store_load_large_imm_offset_kernel() {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, v1, off offset:3716 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
 bb:
   %i = alloca [4096 x i32], align 4, addrspace(5)
@@ -4101,7 +3230,6 @@ bb:
   store volatile i32 15, i32 addrspace(5)* %i7, align 4
   %i10 = getelementptr inbounds [4096 x i32], [4096 x i32] addrspace(5)* %i, i32 0, i32 4000
   %i12 = load volatile i32, i32 addrspace(5)* %i10, align 4
-  call void asm sideeffect "; use $0", "s"([4096 x i32] addrspace(5)* %i) #0
   ret void
 }
 
@@ -4111,20 +3239,15 @@ define void @store_load_large_imm_offset_foo() {
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 13
 ; GFX9-NEXT:    s_movk_i32 s0, 0x3000
-; GFX9-NEXT:    s_add_i32 vcc_lo, s32, 4
+; GFX9-NEXT:    s_add_i32 vcc_hi, s32, 4
 ; GFX9-NEXT:    scratch_store_dword off, v0, s32 offset:4
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    s_add_i32 s0, s0, vcc_lo
+; GFX9-NEXT:    s_add_i32 s0, s0, vcc_hi
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX9-NEXT:    scratch_store_dword off, v0, s0 offset:3712
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    scratch_load_dword v0, off, s0 offset:3712 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    s_add_i32 vcc_hi, s32, 4
-; GFX9-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v0
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX10-LABEL: store_load_large_imm_offset_foo:
@@ -4134,19 +3257,14 @@ define void @store_load_large_imm_offset_foo() {
 ; GFX10-NEXT:    v_mov_b32_e32 v0, 13
 ; GFX10-NEXT:    v_mov_b32_e32 v1, 15
 ; GFX10-NEXT:    s_movk_i32 s0, 0x3800
-; GFX10-NEXT:    s_add_i32 s1, s32, 4
-; GFX10-NEXT:    s_add_i32 s0, s0, s1
+; GFX10-NEXT:    s_add_i32 vcc_lo, s32, 4
+; GFX10-NEXT:    s_add_i32 s0, s0, vcc_lo
 ; GFX10-NEXT:    scratch_store_dword off, v0, s32 offset:4
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_store_dword off, v1, s0 offset:1664
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, off, s0 offset:1664 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    s_add_i32 vcc_lo, s32, 4
-; GFX10-NEXT:    v_mov_b32_e32 v0, vcc_lo
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX11-LABEL: store_load_large_imm_offset_foo:
@@ -4155,17 +3273,12 @@ define void @store_load_large_imm_offset_foo() {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    v_dual_mov_b32 v0, 13 :: v_dual_mov_b32 v1, 0x3000
 ; GFX11-NEXT:    v_mov_b32_e32 v2, 15
-; GFX11-NEXT:    s_add_i32 vcc_lo, s32, 4
 ; GFX11-NEXT:    scratch_store_b32 off, v0, s32 offset:4 dlc
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_store_b32 v1, v2, s32 offset:3716 dlc
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, v1, s32 offset:3716 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, vcc_lo
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX9-PAL-LABEL: store_load_large_imm_offset_foo:
@@ -4173,20 +3286,15 @@ define void @store_load_large_imm_offset_foo() {
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 13
 ; GFX9-PAL-NEXT:    s_movk_i32 s0, 0x3000
-; GFX9-PAL-NEXT:    s_add_i32 vcc_lo, s32, 4
+; GFX9-PAL-NEXT:    s_add_i32 vcc_hi, s32, 4
 ; GFX9-PAL-NEXT:    scratch_store_dword off, v0, s32 offset:4
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    s_add_i32 s0, s0, vcc_lo
+; GFX9-PAL-NEXT:    s_add_i32 s0, s0, vcc_hi
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX9-PAL-NEXT:    scratch_store_dword off, v0, s0 offset:3712
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, off, s0 offset:3712 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    s_add_i32 vcc_hi, s32, 4
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v0
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX940-LABEL: store_load_large_imm_offset_foo:
@@ -4201,11 +3309,6 @@ define void @store_load_large_imm_offset_foo() {
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
 ; GFX940-NEXT:    scratch_load_dword v0, v0, s32 offset:3716 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    s_add_i32 vcc_hi, s32, 4
-; GFX940-NEXT:    v_mov_b32_e32 v0, vcc_hi
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX10-PAL-LABEL: store_load_large_imm_offset_foo:
@@ -4215,19 +3318,14 @@ define void @store_load_large_imm_offset_foo() {
 ; GFX10-PAL-NEXT:    v_mov_b32_e32 v0, 13
 ; GFX10-PAL-NEXT:    v_mov_b32_e32 v1, 15
 ; GFX10-PAL-NEXT:    s_movk_i32 s0, 0x3800
-; GFX10-PAL-NEXT:    s_add_i32 s1, s32, 4
-; GFX10-PAL-NEXT:    s_add_i32 s0, s0, s1
+; GFX10-PAL-NEXT:    s_add_i32 vcc_lo, s32, 4
+; GFX10-PAL-NEXT:    s_add_i32 s0, s0, vcc_lo
 ; GFX10-PAL-NEXT:    scratch_store_dword off, v0, s32 offset:4
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    scratch_store_dword off, v1, s0 offset:1664
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    scratch_load_dword v0, off, s0 offset:1664 glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-PAL-NEXT:    s_add_i32 vcc_lo, s32, 4
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v0, vcc_lo
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v0
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; GFX11-PAL-LABEL: store_load_large_imm_offset_foo:
@@ -4236,18 +3334,26 @@ define void @store_load_large_imm_offset_foo() {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    v_dual_mov_b32 v0, 13 :: v_dual_mov_b32 v1, 0x3000
 ; GFX11-PAL-NEXT:    v_mov_b32_e32 v2, 15
-; GFX11-PAL-NEXT:    s_add_i32 vcc_lo, s32, 4
 ; GFX11-PAL-NEXT:    scratch_store_b32 off, v0, s32 offset:4 dlc
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_store_b32 v1, v2, s32 offset:3716 dlc
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, v1, s32 offset:3716 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, vcc_lo
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: store_load_large_imm_offset_foo:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v0, 13
+; GCN-NEXT:    scratch_store_dword off, v0, s32 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v0, 0x3000
+; GCN-NEXT:    v_mov_b32_e32 v1, 15
+; GCN-NEXT:    scratch_store_dword v0, v1, s32 offset:3712 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    scratch_load_dword v0, v0, s32 offset:3712 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
 bb:
   %i = alloca [4096 x i32], align 4, addrspace(5)
   %i1 = getelementptr inbounds [4096 x i32], [4096 x i32] addrspace(5)* %i, i32 0, i32 undef
@@ -4256,7 +3362,6 @@ bb:
   store volatile i32 15, i32 addrspace(5)* %i7, align 4
   %i10 = getelementptr inbounds [4096 x i32], [4096 x i32] addrspace(5)* %i, i32 0, i32 4000
   %i12 = load volatile i32, i32 addrspace(5)* %i10, align 4
-  call void asm sideeffect "; use $0", "s"([4096 x i32] addrspace(5)* %i) #0
   ret void
 }
 
@@ -4267,17 +3372,14 @@ define amdgpu_kernel void @store_load_vidx_sidx_offset(i32 %sidx) {
 ; GFX9-NEXT:    s_add_u32 flat_scratch_lo, s2, s5
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 4
 ; GFX9-NEXT:    s_addc_u32 flat_scratch_hi, s3, 0
-; GFX9-NEXT:    v_mov_b32_e32 v2, 15
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_add_u32_e32 v0, s0, v0
 ; GFX9-NEXT:    v_lshl_add_u32 v0, v0, 2, v1
-; GFX9-NEXT:    scratch_store_dword v0, v2, off offset:1024
+; GFX9-NEXT:    v_mov_b32_e32 v1, 15
+; GFX9-NEXT:    scratch_store_dword v0, v1, off offset:1024
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    scratch_load_dword v0, v0, off offset:1024 glc
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use v1
-; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX10-LABEL: store_load_vidx_sidx_offset:
@@ -4295,10 +3397,6 @@ define amdgpu_kernel void @store_load_vidx_sidx_offset(i32 %sidx) {
 ; GFX10-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-NEXT:    scratch_load_dword v0, v0, off offset:1024 glc dlc
 ; GFX10-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-NEXT:    ;;#ASMSTART
-; GFX10-NEXT:    ; use v0
-; GFX10-NEXT:    ;;#ASMEND
 ; GFX10-NEXT:    s_endpgm
 ;
 ; GFX11-LABEL: store_load_vidx_sidx_offset:
@@ -4311,10 +3409,6 @@ define amdgpu_kernel void @store_load_vidx_sidx_offset(i32 %sidx) {
 ; GFX11-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-NEXT:    scratch_load_b32 v0, v0, off offset:1028 glc dlc
 ; GFX11-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-NEXT:    ;;#ASMSTART
-; GFX11-NEXT:    ; use v0
-; GFX11-NEXT:    ;;#ASMEND
 ; GFX11-NEXT:    s_endpgm
 ;
 ; GFX9-PAL-LABEL: store_load_vidx_sidx_offset:
@@ -4324,20 +3418,17 @@ define amdgpu_kernel void @store_load_vidx_sidx_offset(i32 %sidx) {
 ; GFX9-PAL-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0x0
 ; GFX9-PAL-NEXT:    v_mov_b32_e32 v1, 4
 ; GFX9-PAL-NEXT:    s_load_dword s0, s[0:1], 0x0
-; GFX9-PAL-NEXT:    v_mov_b32_e32 v2, 15
 ; GFX9-PAL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-PAL-NEXT:    s_and_b32 s5, s5, 0xffff
 ; GFX9-PAL-NEXT:    s_add_u32 flat_scratch_lo, s4, s3
 ; GFX9-PAL-NEXT:    v_add_u32_e32 v0, s0, v0
 ; GFX9-PAL-NEXT:    s_addc_u32 flat_scratch_hi, s5, 0
 ; GFX9-PAL-NEXT:    v_lshl_add_u32 v0, v0, 2, v1
-; GFX9-PAL-NEXT:    scratch_store_dword v0, v2, off offset:1024
+; GFX9-PAL-NEXT:    v_mov_b32_e32 v1, 15
+; GFX9-PAL-NEXT:    scratch_store_dword v0, v1, off offset:1024
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-PAL-NEXT:    scratch_load_dword v0, v0, off offset:1024 glc
 ; GFX9-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-PAL-NEXT:    ;;#ASMSTART
-; GFX9-PAL-NEXT:    ; use v1
-; GFX9-PAL-NEXT:    ;;#ASMEND
 ; GFX9-PAL-NEXT:    s_endpgm
 ;
 ; GFX940-LABEL: store_load_vidx_sidx_offset:
@@ -4350,10 +3441,6 @@ define amdgpu_kernel void @store_load_vidx_sidx_offset(i32 %sidx) {
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
 ; GFX940-NEXT:    scratch_load_dword v0, v0, off offset:1028 sc0 sc1
 ; GFX940-NEXT:    s_waitcnt vmcnt(0)
-; GFX940-NEXT:    v_mov_b32_e32 v0, 4
-; GFX940-NEXT:    ;;#ASMSTART
-; GFX940-NEXT:    ; use v0
-; GFX940-NEXT:    ;;#ASMEND
 ; GFX940-NEXT:    s_endpgm
 ;
 ; GFX10-PAL-LABEL: store_load_vidx_sidx_offset:
@@ -4376,10 +3463,6 @@ define amdgpu_kernel void @store_load_vidx_sidx_offset(i32 %sidx) {
 ; GFX10-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX10-PAL-NEXT:    scratch_load_dword v0, v0, off offset:1024 glc dlc
 ; GFX10-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX10-PAL-NEXT:    ;;#ASMSTART
-; GFX10-PAL-NEXT:    ; use v0
-; GFX10-PAL-NEXT:    ;;#ASMEND
 ; GFX10-PAL-NEXT:    s_endpgm
 ;
 ; GFX11-PAL-LABEL: store_load_vidx_sidx_offset:
@@ -4392,11 +3475,18 @@ define amdgpu_kernel void @store_load_vidx_sidx_offset(i32 %sidx) {
 ; GFX11-PAL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-PAL-NEXT:    scratch_load_b32 v0, v0, off offset:1028 glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
-; GFX11-PAL-NEXT:    v_mov_b32_e32 v0, 4
-; GFX11-PAL-NEXT:    ;;#ASMSTART
-; GFX11-PAL-NEXT:    ; use v0
-; GFX11-PAL-NEXT:    ;;#ASMEND
 ; GFX11-PAL-NEXT:    s_endpgm
+; GCN-LABEL: store_load_vidx_sidx_offset:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_load_dword s0, s[0:1], 0x24
+; GCN-NEXT:    v_mov_b32_e32 v1, 15
+; GCN-NEXT:    s_waitcnt lgkmcnt(0)
+; GCN-NEXT:    v_add_lshl_u32 v0, s0, v0, 2
+; GCN-NEXT:    scratch_store_dword v0, v1, off offset:1028 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    scratch_load_dword v0, v0, off offset:1028 sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_endpgm
 bb:
   %alloca = alloca [32 x i32], align 4, addrspace(5)
   %vidx = tail call i32 @llvm.amdgcn.workitem.id.x()
@@ -4405,7 +3495,6 @@ bb:
   %gep = getelementptr inbounds [32 x i32], [32 x i32] addrspace(5)* %alloca, i32 0, i32 %add2
   store volatile i32 15, i32 addrspace(5)* %gep, align 4
   %load = load volatile i32, i32 addrspace(5)* %gep, align 4
-  call void asm sideeffect "; use $0", "s"([32 x i32] addrspace(5)* %alloca) #0
   ret void
 }
 
@@ -4488,6 +3577,16 @@ define void @store_load_i64_aligned(i64 addrspace(5)* nocapture %arg) {
 ; GFX11-PAL-NEXT:    scratch_load_b64 v[0:1], v0, off glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: store_load_i64_aligned:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v2, 15
+; GCN-NEXT:    v_mov_b32_e32 v3, 0
+; GCN-NEXT:    scratch_store_dwordx2 v0, v[2:3], off sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    scratch_load_dwordx2 v[0:1], v0, off sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
 bb:
   store volatile i64 15, i64 addrspace(5)* %arg, align 8
   %load = load volatile i64, i64 addrspace(5)* %arg, align 8
@@ -4573,6 +3672,16 @@ define void @store_load_i64_unaligned(i64 addrspace(5)* nocapture %arg) {
 ; GFX11-PAL-NEXT:    scratch_load_b64 v[0:1], v0, off glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: store_load_i64_unaligned:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v2, 15
+; GCN-NEXT:    v_mov_b32_e32 v3, 0
+; GCN-NEXT:    scratch_store_dwordx2 v0, v[2:3], off sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    scratch_load_dwordx2 v[0:1], v0, off sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
 bb:
   store volatile i64 15, i64 addrspace(5)* %arg, align 1
   %load = load volatile i64, i64 addrspace(5)* %arg, align 1
@@ -4665,6 +3774,17 @@ define void @store_load_v3i32_unaligned(<3 x i32> addrspace(5)* nocapture %arg)
 ; GFX11-PAL-NEXT:    scratch_load_b96 v[0:2], v0, off glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: store_load_v3i32_unaligned:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v2, 1
+; GCN-NEXT:    v_mov_b32_e32 v3, 2
+; GCN-NEXT:    v_mov_b32_e32 v4, 3
+; GCN-NEXT:    scratch_store_dwordx3 v0, v[2:4], off sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    scratch_load_dwordx3 v[0:2], v0, off sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
 bb:
   store volatile <3 x i32> <i32 1, i32 2, i32 3>, <3 x i32> addrspace(5)* %arg, align 1
   %load = load volatile <3 x i32>, <3 x i32> addrspace(5)* %arg, align 1
@@ -4762,6 +3882,18 @@ define void @store_load_v4i32_unaligned(<4 x i32> addrspace(5)* nocapture %arg)
 ; GFX11-PAL-NEXT:    scratch_load_b128 v[0:3], v0, off glc dlc
 ; GFX11-PAL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-PAL-NEXT:    s_setpc_b64 s[30:31]
+; GCN-LABEL: store_load_v4i32_unaligned:
+; GCN:       ; %bb.0: ; %bb
+; GCN-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v2, 1
+; GCN-NEXT:    v_mov_b32_e32 v3, 2
+; GCN-NEXT:    v_mov_b32_e32 v4, 3
+; GCN-NEXT:    v_mov_b32_e32 v5, 4
+; GCN-NEXT:    scratch_store_dwordx4 v0, v[2:5], off sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    scratch_load_dwordx4 v[0:3], v0, off sc0 sc1
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_setpc_b64 s[30:31]
 bb:
   store volatile <4 x i32> <i32 1, i32 2, i32 3, i32 4>, <4 x i32> addrspace(5)* %arg, align 1
   %load = load volatile <4 x i32>, <4 x i32> addrspace(5)* %arg, align 1

diff --git a/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll b/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
index 8ceb5ec..33d1bd6 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
+++ b/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
@@ -374,6 +374,7 @@
 ; GCN-O1-NEXT:        Machine Optimization Remark Emitter
 ; GCN-O1-NEXT:        Shrink Wrapping analysis
 ; GCN-O1-NEXT:        Prologue/Epilogue Insertion & Frame Finalization
+; GCN-O1-NEXT:        Machine Late Instructions Cleanup Pass
 ; GCN-O1-NEXT:        Control Flow Optimizer
 ; GCN-O1-NEXT:        Lazy Machine Block Frequency Analysis
 ; GCN-O1-NEXT:        Tail Duplication
@@ -670,6 +671,7 @@
 ; GCN-O1-OPTS-NEXT:        Machine Optimization Remark Emitter
 ; GCN-O1-OPTS-NEXT:        Shrink Wrapping analysis
 ; GCN-O1-OPTS-NEXT:        Prologue/Epilogue Insertion & Frame Finalization
+; GCN-O1-OPTS-NEXT:        Machine Late Instructions Cleanup Pass
 ; GCN-O1-OPTS-NEXT:        Control Flow Optimizer
 ; GCN-O1-OPTS-NEXT:        Lazy Machine Block Frequency Analysis
 ; GCN-O1-OPTS-NEXT:        Tail Duplication
@@ -968,6 +970,7 @@
 ; GCN-O2-NEXT:        Machine Optimization Remark Emitter
 ; GCN-O2-NEXT:        Shrink Wrapping analysis
 ; GCN-O2-NEXT:        Prologue/Epilogue Insertion & Frame Finalization
+; GCN-O2-NEXT:        Machine Late Instructions Cleanup Pass
 ; GCN-O2-NEXT:        Control Flow Optimizer
 ; GCN-O2-NEXT:        Lazy Machine Block Frequency Analysis
 ; GCN-O2-NEXT:        Tail Duplication
@@ -1279,6 +1282,7 @@
 ; GCN-O3-NEXT:        Machine Optimization Remark Emitter
 ; GCN-O3-NEXT:        Shrink Wrapping analysis
 ; GCN-O3-NEXT:        Prologue/Epilogue Insertion & Frame Finalization
+; GCN-O3-NEXT:        Machine Late Instructions Cleanup Pass
 ; GCN-O3-NEXT:        Control Flow Optimizer
 ; GCN-O3-NEXT:        Lazy Machine Block Frequency Analysis
 ; GCN-O3-NEXT:        Tail Duplication

diff --git a/llvm/test/CodeGen/AMDGPU/multilevel-break.ll b/llvm/test/CodeGen/AMDGPU/multilevel-break.ll
index d125f43..463aacd 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/multilevel-break.ll
+++ b/llvm/test/CodeGen/AMDGPU/multilevel-break.ll
@@ -188,7 +188,6 @@ define amdgpu_kernel void @multi_if_break_loop(i32 %arg) #0 {
 ; GCN-NEXT:  ; %bb.3: ; %LeafBlock1
 ; GCN-NEXT:    ; in Loop: Header=BB1_2 Depth=1
 ; GCN-NEXT:    s_cmp_eq_u32 s8, 1
-; GCN-NEXT:    s_mov_b64 s[4:5], -1
 ; GCN-NEXT:    s_cbranch_scc0 .LBB1_5
 ; GCN-NEXT:  ; %bb.4: ; %case1
 ; GCN-NEXT:    ; in Loop: Header=BB1_2 Depth=1

diff --git a/llvm/test/CodeGen/AMDGPU/si-annotate-cf.ll b/llvm/test/CodeGen/AMDGPU/si-annotate-cf.ll
index 50a8d78..fb070e8 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/si-annotate-cf.ll
+++ b/llvm/test/CodeGen/AMDGPU/si-annotate-cf.ll
@@ -187,8 +187,6 @@ define amdgpu_kernel void @loop_land_info_assert(i32 %c0, i32 %c1, i32 %c2, i32
 ; SI-NEXT:    s_branch .LBB3_3
 ; SI-NEXT:  .LBB3_1: ; in Loop: Header=BB3_3 Depth=1
 ; SI-NEXT:    s_mov_b64 s[8:9], 0
-; SI-NEXT:    s_mov_b64 s[12:13], -1
-; SI-NEXT:    s_mov_b64 s[14:15], -1
 ; SI-NEXT:  .LBB3_2: ; %Flow
 ; SI-NEXT:    ; in Loop: Header=BB3_3 Depth=1
 ; SI-NEXT:    s_and_b64 vcc, exec, s[14:15]
@@ -206,7 +204,6 @@ define amdgpu_kernel void @loop_land_info_assert(i32 %c0, i32 %c1, i32 %c2, i32
 ; SI-NEXT:    s_cbranch_vccz .LBB3_1
 ; SI-NEXT:  ; %bb.5: ; %if.end
 ; SI-NEXT:    ; in Loop: Header=BB3_3 Depth=1
-; SI-NEXT:    s_mov_b64 s[14:15], -1
 ; SI-NEXT:    s_mov_b64 vcc, s[6:7]
 ; SI-NEXT:    s_cbranch_vccz .LBB3_7
 ; SI-NEXT:  ; %bb.6: ; %if.else
@@ -263,8 +260,6 @@ define amdgpu_kernel void @loop_land_info_assert(i32 %c0, i32 %c1, i32 %c2, i32
 ; FLAT-NEXT:    s_branch .LBB3_3
 ; FLAT-NEXT:  .LBB3_1: ; in Loop: Header=BB3_3 Depth=1
 ; FLAT-NEXT:    s_mov_b64 s[8:9], 0
-; FLAT-NEXT:    s_mov_b64 s[12:13], -1
-; FLAT-NEXT:    s_mov_b64 s[14:15], -1
 ; FLAT-NEXT:  .LBB3_2: ; %Flow
 ; FLAT-NEXT:    ; in Loop: Header=BB3_3 Depth=1
 ; FLAT-NEXT:    s_and_b64 vcc, exec, s[14:15]
@@ -282,7 +277,6 @@ define amdgpu_kernel void @loop_land_info_assert(i32 %c0, i32 %c1, i32 %c2, i32
 ; FLAT-NEXT:    s_cbranch_vccz .LBB3_1
 ; FLAT-NEXT:  ; %bb.5: ; %if.end
 ; FLAT-NEXT:    ; in Loop: Header=BB3_3 Depth=1
-; FLAT-NEXT:    s_mov_b64 s[14:15], -1
 ; FLAT-NEXT:    s_mov_b64 vcc, s[6:7]
 ; FLAT-NEXT:    s_cbranch_vccz .LBB3_7
 ; FLAT-NEXT:  ; %bb.6: ; %if.else

diff --git a/llvm/test/CodeGen/AMDGPU/si-unify-exit-multiple-unreachables.ll b/llvm/test/CodeGen/AMDGPU/si-unify-exit-multiple-unreachables.ll
index 9af9894..d94fed4 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/si-unify-exit-multiple-unreachables.ll
+++ b/llvm/test/CodeGen/AMDGPU/si-unify-exit-multiple-unreachables.ll
@@ -60,7 +60,6 @@ define amdgpu_kernel void @kernel(i32 %a, i32 addrspace(1)* %x, i32 noundef %n)
 ; CHECK-NEXT:    s_cmp_lg_u32 s10, 0
 ; CHECK-NEXT:    s_cbranch_scc1 .LBB0_14
 ; CHECK-NEXT:  ; %bb.3:
-; CHECK-NEXT:    s_mov_b64 s[2:3], 0
 ; CHECK-NEXT:    s_mov_b64 s[0:1], -1
 ; CHECK-NEXT:  .LBB0_4: ; %Flow3
 ; CHECK-NEXT:    s_and_b64 s[0:1], s[0:1], exec
@@ -103,7 +102,6 @@ define amdgpu_kernel void @kernel(i32 %a, i32 addrspace(1)* %x, i32 noundef %n)
 ; CHECK-NEXT:    s_branch .LBB0_10
 ; CHECK-NEXT:  .LBB0_14: ; %cond.false.i8
 ; CHECK-NEXT:    s_mov_b64 s[2:3], -1
-; CHECK-NEXT:    s_mov_b64 s[0:1], 0
 ; CHECK-NEXT:    s_trap 2
 ; CHECK-NEXT:    s_branch .LBB0_4
 entry:

diff --git a/llvm/test/CodeGen/AMDGPU/si-unify-exit-return-unreachable.ll b/llvm/test/CodeGen/AMDGPU/si-unify-exit-return-unreachable.ll
index c3d3993..56d7fc3 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/si-unify-exit-return-unreachable.ll
+++ b/llvm/test/CodeGen/AMDGPU/si-unify-exit-return-unreachable.ll
@@ -140,7 +140,6 @@ define void @my_func(i32 %0) {
 ; GCN-NEXT:    s_cbranch_scc1 .LBB0_10
 ; GCN-NEXT:  ; %bb.9:
 ; GCN-NEXT:    s_mov_b64 s[6:7], -1
-; GCN-NEXT:    s_mov_b64 s[4:5], 0
 ; GCN-NEXT:    s_and_saveexec_b64 s[8:9], s[6:7]
 ; GCN-NEXT:    s_cbranch_execnz .LBB0_3
 ; GCN-NEXT:    s_branch .LBB0_4
@@ -173,7 +172,6 @@ define void @my_func(i32 %0) {
 ; GCN-NEXT:  ; %bb.15: ; %LeafBlock9
 ; GCN-NEXT:    v_cmp_lt_i32_e32 vcc, 1, v0
 ; GCN-NEXT:    s_mov_b64 s[8:9], -1
-; GCN-NEXT:    s_mov_b64 s[4:5], 0
 ; GCN-NEXT:    s_and_saveexec_b64 s[12:13], vcc
 ; GCN-NEXT:  ; %bb.16: ; %do.body.i.i.i.i
 ; GCN-NEXT:    s_mov_b64 s[4:5], exec

diff --git a/llvm/test/CodeGen/AMDGPU/spill-offset-calculation.ll b/llvm/test/CodeGen/AMDGPU/spill-offset-calculation.ll
index 34bc752..b9b7a5d 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/spill-offset-calculation.ll
+++ b/llvm/test/CodeGen/AMDGPU/spill-offset-calculation.ll
@@ -34,7 +34,6 @@ define amdgpu_kernel void @test_inst_offset_kernel() {
 ; FLATSCR-NEXT:    scratch_store_dword off, v0, s0 ; 4-byte Folded Spill
 ; FLATSCR-NEXT:    ;;#ASMSTART
 ; FLATSCR-NEXT:    ;;#ASMEND
-; FLATSCR-NEXT:    s_movk_i32 s0, 0xffc
 ; FLATSCR-NEXT:    scratch_load_dword v0, off, s0 ; 4-byte Folded Reload
 ; FLATSCR-NEXT:    s_waitcnt vmcnt(0)
 ; FLATSCR-NEXT:    scratch_store_dword off, v0, vcc_hi offset:8
@@ -71,7 +70,6 @@ define amdgpu_kernel void @test_sgpr_offset_kernel() {
 ; MUBUF-NEXT:    buffer_store_dword v0, off, s[0:3], s4 ; 4-byte Folded Spill
 ; MUBUF-NEXT:    ;;#ASMSTART
 ; MUBUF-NEXT:    ;;#ASMEND
-; MUBUF-NEXT:    s_mov_b32 s4, 0x40000
 ; MUBUF-NEXT:    buffer_load_dword v0, off, s[0:3], s4 ; 4-byte Folded Reload
 ; MUBUF-NEXT:    s_waitcnt vmcnt(0)
 ; MUBUF-NEXT:    buffer_store_dword v0, off, s[0:3], 0 offset:8
@@ -90,7 +88,6 @@ define amdgpu_kernel void @test_sgpr_offset_kernel() {
 ; FLATSCR-NEXT:    scratch_store_dword off, v0, s0 ; 4-byte Folded Spill
 ; FLATSCR-NEXT:    ;;#ASMSTART
 ; FLATSCR-NEXT:    ;;#ASMEND
-; FLATSCR-NEXT:    s_movk_i32 s0, 0x1000
 ; FLATSCR-NEXT:    scratch_load_dword v0, off, s0 ; 4-byte Folded Reload
 ; FLATSCR-NEXT:    s_waitcnt vmcnt(0)
 ; FLATSCR-NEXT:    scratch_store_dword off, v0, vcc_hi offset:8
@@ -237,7 +234,6 @@ define amdgpu_kernel void @test_sgpr_offset_function_scavenge_fail_kernel() #3 {
 ; FLATSCR-NEXT:    ;;#ASMEND
 ; FLATSCR-NEXT:    ;;#ASMSTART
 ; FLATSCR-NEXT:    ;;#ASMEND
-; FLATSCR-NEXT:    s_movk_i32 s8, 0x1004
 ; FLATSCR-NEXT:    scratch_load_dword v0, off, s8 ; 4-byte Folded Reload
 ; FLATSCR-NEXT:    s_waitcnt vmcnt(0)
 ; FLATSCR-NEXT:    ;;#ASMSTART
@@ -320,7 +316,6 @@ define amdgpu_kernel void @test_sgpr_offset_subregs_kernel() {
 ; FLATSCR-NEXT:    ;;#ASMEND
 ; FLATSCR-NEXT:    scratch_load_dword v0, off, vcc_hi offset:8 glc
 ; FLATSCR-NEXT:    s_waitcnt vmcnt(0)
-; FLATSCR-NEXT:    s_movk_i32 s0, 0xff8
 ; FLATSCR-NEXT:    scratch_load_dwordx2 v[0:1], off, s0 ; 8-byte Folded Reload
 ; FLATSCR-NEXT:    s_waitcnt vmcnt(0)
 ; FLATSCR-NEXT:    ;;#ASMSTART
@@ -367,7 +362,6 @@ define amdgpu_kernel void @test_inst_offset_subregs_kernel() {
 ; MUBUF-NEXT:    ;;#ASMEND
 ; MUBUF-NEXT:    buffer_load_dword v0, off, s[0:3], 0 offset:8 glc
 ; MUBUF-NEXT:    s_waitcnt vmcnt(0)
-; MUBUF-NEXT:    s_mov_b32 s4, 0x3ff00
 ; MUBUF-NEXT:    buffer_load_dword v0, off, s[0:3], s4 ; 4-byte Folded Reload
 ; MUBUF-NEXT:    s_nop 0
 ; MUBUF-NEXT:    buffer_load_dword v1, off, s[0:3], s4 offset:4 ; 4-byte Folded Reload
@@ -391,7 +385,6 @@ define amdgpu_kernel void @test_inst_offset_subregs_kernel() {
 ; FLATSCR-NEXT:    ;;#ASMEND
 ; FLATSCR-NEXT:    scratch_load_dword v0, off, vcc_hi offset:8 glc
 ; FLATSCR-NEXT:    s_waitcnt vmcnt(0)
-; FLATSCR-NEXT:    s_movk_i32 s0, 0xffc
 ; FLATSCR-NEXT:    scratch_load_dwordx2 v[0:1], off, s0 ; 8-byte Folded Reload
 ; FLATSCR-NEXT:    s_waitcnt vmcnt(0)
 ; FLATSCR-NEXT:    ;;#ASMSTART

diff --git a/llvm/test/CodeGen/AMDGPU/spill-scavenge-offset.ll b/llvm/test/CodeGen/AMDGPU/spill-scavenge-offset.ll
index bdeb97c..ab8efa9 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/spill-scavenge-offset.ll
+++ b/llvm/test/CodeGen/AMDGPU/spill-scavenge-offset.ll
@@ -10551,7 +10551,6 @@ define amdgpu_kernel void @test_limited_sgpr(<64 x i32> addrspace(1)* %out, <64
 ; GFX6-NEXT:    s_waitcnt expcnt(0)
 ; GFX6-NEXT:    ;;#ASMSTART
 ; GFX6-NEXT:    ;;#ASMEND
-; GFX6-NEXT:    s_mov_b32 s2, 0x84800
 ; GFX6-NEXT:    buffer_load_dword v17, off, s[40:43], s2 ; 4-byte Folded Reload
 ; GFX6-NEXT:    buffer_load_dword v18, off, s[40:43], s2 offset:4 ; 4-byte Folded Reload
 ; GFX6-NEXT:    buffer_load_dword v19, off, s[40:43], s2 offset:8 ; 4-byte Folded Reload
@@ -10796,7 +10795,7 @@ define amdgpu_kernel void @test_limited_sgpr(<64 x i32> addrspace(1)* %out, <64
 ; GFX9-FLATSCR-NEXT:    scratch_store_dwordx4 off, v[20:23], s0 ; 16-byte Folded Spill
 ; GFX9-FLATSCR-NEXT:    s_movk_i32 s0, 0x2100
 ; GFX9-FLATSCR-NEXT:    scratch_store_dwordx4 off, v[8:11], s0 ; 16-byte Folded Spill
-; GFX9-FLATSCR-NEXT:    s_movk_i32 s0, 0x2100
+; GFX9-FLATSCR-NEXT:    s_nop 0
 ; GFX9-FLATSCR-NEXT:    ;;#ASMSTART
 ; GFX9-FLATSCR-NEXT:    ;;#ASMEND
 ; GFX9-FLATSCR-NEXT:    scratch_load_dwordx4 v[8:11], off, s0 ; 16-byte Folded Reload
@@ -11032,7 +11031,6 @@ define amdgpu_kernel void @test_limited_sgpr(<64 x i32> addrspace(1)* %out, <64
 ; GFX10-FLATSCR-NEXT:    v_mov_b32_e32 v35, v60
 ; GFX10-FLATSCR-NEXT:    ;;#ASMSTART
 ; GFX10-FLATSCR-NEXT:    ;;#ASMEND
-; GFX10-FLATSCR-NEXT:    s_movk_i32 s0, 0x2010
 ; GFX10-FLATSCR-NEXT:    v_mov_b32_e32 v36, v65
 ; GFX10-FLATSCR-NEXT:    v_mov_b32_e32 v37, v66
 ; GFX10-FLATSCR-NEXT:    v_mov_b32_e32 v38, v67

diff --git a/llvm/test/CodeGen/ARM/O3-pipeline.ll b/llvm/test/CodeGen/ARM/O3-pipeline.ll
index 253e17c..411335a 100644 (file)
--- a/llvm/test/CodeGen/ARM/O3-pipeline.ll
+++ b/llvm/test/CodeGen/ARM/O3-pipeline.ll
@@ -149,6 +149,7 @@
 ; CHECK-NEXT:      Machine Optimization Remark Emitter
 ; CHECK-NEXT:      Shrink Wrapping analysis
 ; CHECK-NEXT:      Prologue/Epilogue Insertion & Frame Finalization
+; CHECK-NEXT:      Machine Late Instructions Cleanup Pass
 ; CHECK-NEXT:      Control Flow Optimizer
 ; CHECK-NEXT:      Lazy Machine Block Frequency Analysis
 ; CHECK-NEXT:      Tail Duplication

diff --git a/llvm/test/CodeGen/ARM/arm-shrink-wrapping.ll b/llvm/test/CodeGen/ARM/arm-shrink-wrapping.ll
index b5c63af..fa94812 100644 (file)
--- a/llvm/test/CodeGen/ARM/arm-shrink-wrapping.ll
+++ b/llvm/test/CodeGen/ARM/arm-shrink-wrapping.ll
@@ -1652,7 +1652,6 @@ define void @infiniteloop3() "frame-pointer"="all" {
 ; THUMB-ENABLE-NEXT:    movs r0, #0
 ; THUMB-ENABLE-NEXT:    cbnz r0, LBB11_5
 ; THUMB-ENABLE-NEXT:  @ %bb.1: @ %loop2a.preheader
-; THUMB-ENABLE-NEXT:    movs r0, #0
 ; THUMB-ENABLE-NEXT:    movs r1, #0
 ; THUMB-ENABLE-NEXT:    mov r2, r0
 ; THUMB-ENABLE-NEXT:    b LBB11_3
@@ -1679,7 +1678,6 @@ define void @infiniteloop3() "frame-pointer"="all" {
 ; THUMB-DISABLE-NEXT:    movs r0, #0
 ; THUMB-DISABLE-NEXT:    cbnz r0, LBB11_5
 ; THUMB-DISABLE-NEXT:  @ %bb.1: @ %loop2a.preheader
-; THUMB-DISABLE-NEXT:    movs r0, #0
 ; THUMB-DISABLE-NEXT:    movs r1, #0
 ; THUMB-DISABLE-NEXT:    mov r2, r0
 ; THUMB-DISABLE-NEXT:    b LBB11_3

diff --git a/llvm/test/CodeGen/ARM/fpclamptosat.ll b/llvm/test/CodeGen/ARM/fpclamptosat.ll
index 4824142..18fa1ad 100644 (file)
--- a/llvm/test/CodeGen/ARM/fpclamptosat.ll
+++ b/llvm/test/CodeGen/ARM/fpclamptosat.ll
@@ -3764,7 +3764,6 @@ define i64 @stest_f32i64_mm(float %x) {
 ; SOFT-NEXT:  @ %bb.18: @ %entry
 ; SOFT-NEXT:    mov r3, r6
 ; SOFT-NEXT:  .LBB48_19: @ %entry
-; SOFT-NEXT:    ldr r0, .LCPI48_0
 ; SOFT-NEXT:    cmp r4, r0
 ; SOFT-NEXT:    ldr r4, [sp, #16] @ 4-byte Reload
 ; SOFT-NEXT:    beq .LBB48_21
@@ -4347,7 +4346,6 @@ define i64 @stest_f16i64_mm(half %x) {
 ; SOFT-NEXT:  @ %bb.18: @ %entry
 ; SOFT-NEXT:    mov r3, r6
 ; SOFT-NEXT:  .LBB51_19: @ %entry
-; SOFT-NEXT:    ldr r0, .LCPI51_0
 ; SOFT-NEXT:    cmp r4, r0
 ; SOFT-NEXT:    ldr r4, [sp, #16] @ 4-byte Reload
 ; SOFT-NEXT:    beq .LBB51_21

diff --git a/llvm/test/CodeGen/ARM/ifcvt-branch-weight-bug.ll b/llvm/test/CodeGen/ARM/ifcvt-branch-weight-bug.ll
index af2009c..6e5db3f 100644 (file)
--- a/llvm/test/CodeGen/ARM/ifcvt-branch-weight-bug.ll
+++ b/llvm/test/CodeGen/ARM/ifcvt-branch-weight-bug.ll
@@ -22,7 +22,7 @@ entry:
 ; for.body -> for.cond.backedge (100%)
 ;          -> cond.false.i (0%)
 ; CHECK: bb.1.for.body:
-; CHECK: successors: %bb.2(0x80000000), %bb.4(0x00000000)
+; CHECK: successors: %bb.2(0x80000000), %bb.5(0x00000000)
 for.body:
   br i1 undef, label %for.cond.backedge, label %lor.lhs.false.i, !prof !1
 

diff --git a/llvm/test/CodeGen/ARM/jump-table-islands.ll b/llvm/test/CodeGen/ARM/jump-table-islands.ll
index 755ca30..c327aff 100644 (file)
--- a/llvm/test/CodeGen/ARM/jump-table-islands.ll
+++ b/llvm/test/CodeGen/ARM/jump-table-islands.ll
@@ -1,6 +1,6 @@
 ; RUN: llc -mtriple=armv7-apple-ios8.0 -o - %s | FileCheck %s
 
-%BigInt = type i5500
+%BigInt = type i8500
 
 define %BigInt @test_moved_jumptable(i1 %tst, i32 %sw, %BigInt %l) {
 ; CHECK-LABEL: test_moved_jumptable:

diff --git a/llvm/test/CodeGen/ARM/reg_sequence.ll b/llvm/test/CodeGen/ARM/reg_sequence.ll
index 976dddc..db620f6 100644 (file)
--- a/llvm/test/CodeGen/ARM/reg_sequence.ll
+++ b/llvm/test/CodeGen/ARM/reg_sequence.ll
@@ -283,7 +283,6 @@ define arm_aapcs_vfpcc i32 @t10(float %x) nounwind {
 ; CHECK-NEXT:  vst1.32 {d17[1]}, [r0:32]
 ; CHECK-NEXT:  mov     r0, #0
 ; CHECK-NEXT:  cmp     r0, #0
-; CHECK-NEXT:  movne   r0, #0
 ; CHECK-NEXT:  bxne    lr
 ; CHECK-NEXT:  LBB9_1:
 ; CHECK-NEXT:  trap

diff --git a/llvm/test/CodeGen/BPF/objdump_cond_op_2.ll b/llvm/test/CodeGen/BPF/objdump_cond_op_2.ll
index a87c517..a1f8fdb 100644 (file)
--- a/llvm/test/CodeGen/BPF/objdump_cond_op_2.ll
+++ b/llvm/test/CodeGen/BPF/objdump_cond_op_2.ll
@@ -14,9 +14,8 @@ define i32 @test(i32, i32) local_unnamed_addr #0 {
 
 ; <label>:4:                                      ; preds = %2
   br label %5
-; CHECK: if r4 s>= r3 goto +11 <LBB0_3>
-; CHECK: r0 = 0
-; CHECK-LABEL: <LBB0_2>:
+; CHECK: if r4 s>= r3 goto +10 <LBB0_2>
+; CHECK-LABEL: <LBB0_1>:
 
 ; <label>:5:                                      ; preds = %4, %5
   %6 = phi i32 [ %9, %5 ], [ 0, %4 ]
@@ -28,12 +27,12 @@ define i32 @test(i32, i32) local_unnamed_addr #0 {
   %12 = icmp slt i32 %10, %11
   br i1 %12, label %5, label %13
 ; CHECK: r1 = r3
-; CHECK: if r2 s> r3 goto -10 <LBB0_2>
+; CHECK: if r2 s> r3 goto -10 <LBB0_1>
 
 ; <label>:13:                                     ; preds = %5, %2
   %14 = phi i32 [ 0, %2 ], [ %9, %5 ]
   ret i32 %14
-; CHECK-LABEL: <LBB0_3>:
+; CHECK-LABEL: <LBB0_2>:
 ; CHECK: exit
 }
 attributes #0 = { norecurse nounwind readnone }

diff --git a/llvm/test/CodeGen/Mips/llvm-ir/lshr.ll b/llvm/test/CodeGen/Mips/llvm-ir/lshr.ll
index ed2bfc9..9acdf25 100644 (file)
--- a/llvm/test/CodeGen/Mips/llvm-ir/lshr.ll
+++ b/llvm/test/CodeGen/Mips/llvm-ir/lshr.ll
@@ -841,7 +841,6 @@ define signext i128 @lshr_i128(i128 signext %a, i128 signext %b) {
 ; MMR3-NEXT:    or16 $6, $4
 ; MMR3-NEXT:    lw $4, 8($sp) # 4-byte Folded Reload
 ; MMR3-NEXT:    movn $1, $7, $4
-; MMR3-NEXT:    li16 $7, 0
 ; MMR3-NEXT:    movn $1, $6, $10
 ; MMR3-NEXT:    lw $4, 24($sp) # 4-byte Folded Reload
 ; MMR3-NEXT:    movz $1, $4, $16

diff --git a/llvm/test/CodeGen/Mips/llvm-ir/shl.ll b/llvm/test/CodeGen/Mips/llvm-ir/shl.ll
index a8d829b..b47a1f8 100644 (file)
--- a/llvm/test/CodeGen/Mips/llvm-ir/shl.ll
+++ b/llvm/test/CodeGen/Mips/llvm-ir/shl.ll
@@ -915,7 +915,6 @@ define signext i128 @shl_i128(i128 signext %a, i128 signext %b) {
 ; MMR3-NEXT:    or16 $5, $3
 ; MMR3-NEXT:    lw $3, 12($sp) # 4-byte Folded Reload
 ; MMR3-NEXT:    movn $8, $7, $3
-; MMR3-NEXT:    li16 $7, 0
 ; MMR3-NEXT:    movn $8, $5, $10
 ; MMR3-NEXT:    lw $3, 28($sp) # 4-byte Folded Reload
 ; MMR3-NEXT:    movz $8, $3, $16

diff --git a/llvm/test/CodeGen/PowerPC/O3-pipeline.ll b/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
index 6b87f60..8c51f75 100644 (file)
--- a/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
+++ b/llvm/test/CodeGen/PowerPC/O3-pipeline.ll
@@ -182,6 +182,7 @@
 ; CHECK-NEXT:       Machine Optimization Remark Emitter
 ; CHECK-NEXT:       Shrink Wrapping analysis
 ; CHECK-NEXT:       Prologue/Epilogue Insertion & Frame Finalization
+; CHECK-NEXT:       Machine Late Instructions Cleanup Pass
 ; CHECK-NEXT:       Control Flow Optimizer
 ; CHECK-NEXT:       Lazy Machine Block Frequency Analysis
 ; CHECK-NEXT:       Tail Duplication

diff --git a/llvm/test/CodeGen/PowerPC/cgp-select.ll b/llvm/test/CodeGen/PowerPC/cgp-select.ll
index abc868c..2c36233 100644 (file)
--- a/llvm/test/CodeGen/PowerPC/cgp-select.ll
+++ b/llvm/test/CodeGen/PowerPC/cgp-select.ll
@@ -11,7 +11,6 @@ define dso_local void @wibble(ptr nocapture readonly %arg, i32 signext %arg1, pt
 ; CHECK-NEXT:    blt 0, .LBB0_5
 ; CHECK-NEXT:  # %bb.1: # %bb6
 ; CHECK-NEXT:    clrldi 4, 4, 32
-; CHECK-NEXT:    li 7, 7
 ; CHECK-NEXT:    addi 4, 4, -1
 ; CHECK-NEXT:    mtctr 4
 ; CHECK-NEXT:    li 4, 8

diff --git a/llvm/test/CodeGen/PowerPC/fast-isel-branch.ll b/llvm/test/CodeGen/PowerPC/fast-isel-branch.ll
index 826ceea..4a424b6 100644 (file)
--- a/llvm/test/CodeGen/PowerPC/fast-isel-branch.ll
+++ b/llvm/test/CodeGen/PowerPC/fast-isel-branch.ll
@@ -53,7 +53,7 @@ define signext i32 @bar() #0 {
 ; AIX64-NEXT:  L..BB0_1: # %for.cond
 ; AIX64-NEXT:    #
 ; AIX64-NEXT:    lwz 3, 120(1)
-; AIX64-NEXT:    ld 4, L..C0(2) # @x
+; AIX64-NEXT:    ld 4, L..C0(2)
 ; AIX64-NEXT:    lwz 4, 0(4)
 ; AIX64-NEXT:    cmpw 3, 4
 ; AIX64-NEXT:    bge 0, L..BB0_4

diff --git a/llvm/test/CodeGen/PowerPC/fp-strict-conv-f128.ll b/llvm/test/CodeGen/PowerPC/fp-strict-conv-f128.ll
index 97f9ae1..8390227 100644 (file)
--- a/llvm/test/CodeGen/PowerPC/fp-strict-conv-f128.ll
+++ b/llvm/test/CodeGen/PowerPC/fp-strict-conv-f128.ll
@@ -618,7 +618,6 @@ define zeroext i32 @ppcq_to_u32(ppc_fp128 %m) #0 {
 ; P8-NEXT:    lfs f0, .LCPI13_0@toc@l(r3)
 ; P8-NEXT:    lis r3, -32768
 ; P8-NEXT:    fcmpo cr0, f2, f3
-; P8-NEXT:    xxlxor f3, f3, f3
 ; P8-NEXT:    fcmpo cr1, f1, f0
 ; P8-NEXT:    crand 4*cr5+lt, 4*cr1+eq, lt
 ; P8-NEXT:    crandc 4*cr5+gt, 4*cr1+lt, 4*cr1+eq
@@ -660,7 +659,6 @@ define zeroext i32 @ppcq_to_u32(ppc_fp128 %m) #0 {
 ; P9-NEXT:    lfs f0, .LCPI13_0@toc@l(r3)
 ; P9-NEXT:    fcmpo cr1, f2, f3
 ; P9-NEXT:    lis r3, -32768
-; P9-NEXT:    xxlxor f3, f3, f3
 ; P9-NEXT:    fcmpo cr0, f1, f0
 ; P9-NEXT:    crand 4*cr5+lt, eq, 4*cr1+lt
 ; P9-NEXT:    crandc 4*cr5+gt, lt, eq

diff --git a/llvm/test/CodeGen/PowerPC/ppcf128-constrained-fp-intrinsics.ll b/llvm/test/CodeGen/PowerPC/ppcf128-constrained-fp-intrinsics.ll
index bfb4f05..9bc88c9 100644 (file)
--- a/llvm/test/CodeGen/PowerPC/ppcf128-constrained-fp-intrinsics.ll
+++ b/llvm/test/CodeGen/PowerPC/ppcf128-constrained-fp-intrinsics.ll
@@ -1295,7 +1295,6 @@ define i32 @test_fptoui_ppc_i32_ppc_fp128(ppc_fp128 %first) #0 {
 ; PC64LE-NEXT:    lfs 0, .LCPI31_0@toc@l(3)
 ; PC64LE-NEXT:    lis 3, -32768
 ; PC64LE-NEXT:    fcmpo 0, 2, 3
-; PC64LE-NEXT:    xxlxor 3, 3, 3
 ; PC64LE-NEXT:    fcmpo 1, 1, 0
 ; PC64LE-NEXT:    crand 20, 6, 0
 ; PC64LE-NEXT:    crandc 21, 4, 6
@@ -1333,7 +1332,6 @@ define i32 @test_fptoui_ppc_i32_ppc_fp128(ppc_fp128 %first) #0 {
 ; PC64LE9-NEXT:    lfs 0, .LCPI31_0@toc@l(3)
 ; PC64LE9-NEXT:    fcmpo 1, 2, 3
 ; PC64LE9-NEXT:    lis 3, -32768
-; PC64LE9-NEXT:    xxlxor 3, 3, 3
 ; PC64LE9-NEXT:    fcmpo 0, 1, 0
 ; PC64LE9-NEXT:    crand 20, 2, 4
 ; PC64LE9-NEXT:    crandc 21, 0, 2

diff --git a/llvm/test/CodeGen/SystemZ/frame-28.mir b/llvm/test/CodeGen/SystemZ/frame-28.mir
new file mode 100644 (file)
index 0000000..dd5933a
--- /dev/null
+++ b/llvm/test/CodeGen/SystemZ/frame-28.mir
@@ -0,0 +1,327 @@
+# RUN: llc -mtriple=s390x-linux-gnu -start-before=prologepilog %s -o - -mcpu=z14 \
+# RUN:   -verify-machineinstrs 2>&1 | FileCheck %s
+# REQUIRES: asserts
+#
+# Test that redundant frame addressing anchor points are removed by
+# MachineLateInstrsCleanup.
+
+--- |
+  define void @fun1() { ret void }
+  define void @fun2() { ret void }
+  define void @fun3() { ret void }
+  define void @fun4() { ret void }
+  define void @fun5() { ret void }
+  define void @fun6() { ret void }
+  define void @fun7() { ret void }
+  define void @fun8() { ret void }
+
+  declare i32 @foo()
+
+  @ptr = external dso_local local_unnamed_addr global ptr
+---
+
+# Test elimination of redundant LAYs in successor blocks.
+# CHECK-LABEL: fun1:
+# CHECK: lay %r1, 4096(%r15)
+# CHECK: # %bb.1:
+# CHECK-NOT: lay
+# CHECK: .LBB0_2:
+# CHECK-NOT: lay
+---
+name:            fun1
+tracksRegLiveness: true
+stack:
+  - { id: 0, size: 5000 }
+  - { id: 1, size: 2500 }
+  - { id: 2, size: 2500 }
+
+machineFunctionInfo: {}
+body:             |
+  bb.0 (%ir-block.0):
+    liveins: $f16d
+    successors: %bb.2(0x00000001), %bb.1(0x7fffffff)
+
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.1, 0, $noreg
+    CHIMux undef $r0l, 3, implicit-def $cc
+    BRC 14, 8, %bb.2, implicit killed $cc
+    J %bb.1
+
+  bb.1:
+    liveins: $f16d
+    VST64 renamable $f16d, %stack.2, 0, $noreg
+    J %bb.2
+
+  bb.2:
+    liveins: $f16d
+    VST64 renamable $f16d, %stack.1, 0, $noreg
+    Return
+...
+
+# In this function the LAY in bb.1 will have a different offset, so the first
+# LAY in bb.2 must remain.
+# CHECK-LABEL: fun2:
+# CHECK: lay %r1, 4096(%r15)
+# CHECK: # %bb.1:
+# CHECK: lay %r1, 8192(%r15)
+# CHECK: .LBB1_2:
+# CHECK: lay %r1, 4096(%r15)
+# CHECK-NOT: lay
+---
+name:            fun2
+tracksRegLiveness: true
+stack:
+  - { id: 0, size: 5000 }
+  - { id: 1, size: 5000 }
+  - { id: 2, size: 2500 }
+
+machineFunctionInfo: {}
+body:             |
+  bb.0 (%ir-block.0):
+    liveins: $f16d
+    successors: %bb.2(0x00000001), %bb.1(0x7fffffff)
+
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.1, 0, $noreg
+    CHIMux undef $r0l, 3, implicit-def $cc
+    BRC 14, 8, %bb.2, implicit killed $cc
+    J %bb.1
+
+  bb.1:
+    liveins: $f16d
+    VST64 renamable $f16d, %stack.2, 0, $noreg
+    J %bb.2
+
+  bb.2:
+    liveins: $f16d
+    VST64 renamable $f16d, %stack.1, 0, $noreg
+    VST64 renamable $f16d, %stack.1, 0, $noreg
+    Return
+...
+
+# Test case with a loop (with room for improvement: since %r1 is not clobbered
+# inside the loop only the first LAY is needed).
+# CHECK-LABEL: fun3:
+# CHECK: lay %r1, 4096(%r15)
+# CHECK: .LBB2_1:
+# CHECK: lay %r1, 4096(%r15)
+# CHECK: .LBB2_2:
+# CHECK-NOT: lay %r1, 4096(%r15)
+---
+name:            fun3
+tracksRegLiveness: true
+stack:
+  - { id: 0, size: 5000 }
+  - { id: 1, size: 2500 }
+  - { id: 2, size: 2500 }
+
+machineFunctionInfo: {}
+body:             |
+  bb.0 (%ir-block.0):
+    liveins: $f16d
+    successors: %bb.2(0x00000001), %bb.1(0x7fffffff)
+
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.1, 0, $noreg
+    CHIMux undef $r0l, 3, implicit-def $cc
+    BRC 14, 8, %bb.2, implicit killed $cc
+    J %bb.1
+
+  bb.1:
+    liveins: $f16d
+    successors: %bb.2(0x00000001), %bb.1(0x7fffffff)
+
+    VST64 renamable $f16d, %stack.2, 0, $noreg
+    CHIMux undef $r0l, 3, implicit-def $cc
+    BRC 14, 8, %bb.1, implicit killed $cc
+    J %bb.2
+
+  bb.2:
+    liveins: $f16d
+    VST64 renamable $f16d, %stack.1, 0, $noreg
+    Return
+...
+
+# Test case with a call which clobbers r1: the second LAY after the call is needed.
+# CHECK-LABEL: fun4:
+# CHECK: lay %r1, 4096(%r15)
+# CHECK: brasl
+# CHECK: lay %r1, 4096(%r15)
+---
+name:            fun4
+tracksRegLiveness: true
+stack:
+  - { id: 0, size: 5000 }
+  - { id: 1, size: 2500 }
+
+machineFunctionInfo: {}
+body:             |
+  bb.0 (%ir-block.0):
+    liveins: $f16d
+
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.1, 0, $noreg
+    ADJCALLSTACKDOWN 0, 0
+    CallBRASL @foo, csr_systemz_elf, implicit-def dead $r14d, implicit-def dead $cc, implicit $fpc, implicit-def $r2l
+    ADJCALLSTACKUP 0, 0
+    $f17d = IMPLICIT_DEF
+    VST64 renamable $f17d, %stack.1, 0, $noreg
+    Return
+...
+
+# Test case where index reg is loaded instead of using an LAY. Only one LGHI is needed.
+# CHECK-LABEL: fun5:
+# CHECK: lghi %r1, 4096
+# CHECK-NOT: lghi
+---
+name:            fun5
+tracksRegLiveness: true
+stack:
+  - { id: 0, size: 5000 }
+  - { id: 1, size: 2500 }
+
+machineFunctionInfo: {}
+body:             |
+  bb.0 (%ir-block.0):
+    liveins: $f16d
+
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    VST64 renamable $f16d, %stack.0, 0, $noreg
+    $f0q = nofpexcept LXEB %stack.1, 0, $noreg, implicit $fpc
+    $f1q = nofpexcept LXEB %stack.1, 0, $noreg, implicit $fpc
+    Return
+...
+
+# Test where the constant is a Global. Only one LARL is needed.
+# CHECK-LABEL: fun6:
+# CHECK: larl %r1, ptr
+# CHECK-NOT: larl
+---
+name:            fun6
+alignment:       16
+tracksRegLiveness: true
+tracksDebugUserValues: true
+frameInfo:
+  maxAlignment:    1
+  maxCallFrameSize: 0
+fixedStack:
+  - { id: 0, offset: -160, size: 8, alignment: 8 }
+machineFunctionInfo: {}
+body:             |
+  bb.0:
+    successors: %bb.2(0x30000000), %bb.1(0x50000000)
+
+    renamable $r1d = LARL @ptr
+    CGHSI killed renamable $r1d, 0, 0, implicit-def $cc :: (volatile dereferenceable load (s64) from @ptr)
+    BRC 14, 8, %bb.2, implicit killed $cc
+    J %bb.1
+
+  bb.1:
+    renamable $r1d = LARL @ptr
+    MVGHI killed renamable $r1d, 0, 0
+
+  bb.2:
+    Return
+
+...
+
+# Load of an invariant location (GOT). Only one LGRL is needed.
+# CHECK-LABEL: fun7:
+# CHECK: lgrl %r1, ptr
+# CHECK-NOT: lgrl
+---
+name:            fun7
+alignment:       16
+tracksRegLiveness: true
+tracksDebugUserValues: true
+frameInfo:
+  maxAlignment:    1
+  maxCallFrameSize: 0
+fixedStack:
+  - { id: 0, offset: -160, size: 8, alignment: 8 }
+machineFunctionInfo: {}
+body:             |
+  bb.0:
+    successors: %bb.2(0x30000000), %bb.1(0x50000000)
+  
+    renamable $r1d = LGRL @ptr :: (load (s64) from got)
+    CGHSI killed renamable $r1d, 0, 0, implicit-def $cc :: (volatile dereferenceable load (s64) from @ptr)
+    BRC 14, 8, %bb.2, implicit killed $cc
+    J %bb.1
+  
+  bb.1:
+    renamable $r1d = LGRL @ptr :: (load (s64) from got)
+    MVGHI killed renamable $r1d, 0, 0
+  
+  bb.2:
+    Return
+
+...
+
+# Load from constant pool. Only one LARL is needed.
+# CHECK-LABEL: fun8:
+# CHECK: larl %r1, .LCPI7_0
+# CHECK-NOT: larl
+---
+name:            fun8
+alignment:       16
+tracksRegLiveness: true
+tracksDebugUserValues: true
+liveins:
+  - { reg: '$f0s' }
+frameInfo:
+  maxAlignment:    1
+  maxCallFrameSize: 0
+fixedStack:
+  - { id: 0, offset: -160, size: 8, alignment: 8 }
+constants:
+  - id:              0
+    value:           float 0x43E0000000000000
+    alignment:       4
+machineFunctionInfo: {}
+body:             |
+  bb.0 (%ir-block.0):
+    successors: %bb.1, %bb.2
+    liveins: $f0s
+
+    renamable $r1d = LARL %const.0
+    renamable $f1s = LE killed renamable $r1d, 0, $noreg :: (load (s32) from constant-pool)
+    nofpexcept CEBR renamable $f0s, renamable $f1s, implicit-def $cc, implicit $fpc
+    BRC 15, 11, %bb.2, implicit killed $cc
+
+  bb.1:
+    liveins: $f0s
+
+    J %bb.3
+
+  bb.2 (%ir-block.0):
+    liveins: $f0s, $f1s
+
+    renamable $r1d = LARL %const.0
+    renamable $f1s = LE killed renamable $r1d, 0, $noreg :: (load (s32) from constant-pool)
+
+  bb.3 (%ir-block.0):
+    liveins: $r2d
+
+    Return
+
+...

diff --git a/llvm/test/CodeGen/Thumb/frame-access.ll b/llvm/test/CodeGen/Thumb/frame-access.ll
index 6a98d12..e7d4aca 100644 (file)
--- a/llvm/test/CodeGen/Thumb/frame-access.ll
+++ b/llvm/test/CodeGen/Thumb/frame-access.ll
@@ -77,12 +77,9 @@ entry:
 ; CHECK-FP-ATPCS: adds r0, #8
 ; CHECK-FP-ATPCS: stm r0!, {r1, r2, r3}
 ; CHECK-FP-AAPCS: mov r0, r11
-; CHECK-FP-AAPCS: str r1, [r0, #8]
-; CHECK-FP-AAPCS: mov r0, r11
-; CHECK-FP-AAPCS: str r2, [r0, #12]
-; CHECK-FP-AAPCS: mov r0, r11
-; CHECK-FP-AAPCS: str r3, [r0, #16]
-
+; CHECK-FP-AAPCS: mov r7, r0
+; CHECK-FP-AAPCS: adds r7, #8
+; CHECK-FP-AAPCS: stm r7!, {r1, r2, r3}
 ; Re-aligned stack, access via FP
 ; int test_args_realign(int a, int b, int c, int d, int e) {
 ;   __attribute__((aligned(16))) int v[4];
@@ -148,11 +145,9 @@ entry:
 ; CHECK-ATPCS-NEXT: adds r0, #8
 ; CHECK-ATPCS-NEXT: stm r0!, {r1, r2, r3}
 ; CHECK-AAPCS: mov r0, r11
-; CHECK-AAPCS: str r1, [r0, #8]
-; CHECK-AAPCS: mov r0, r11
-; CHECK-AAPCS: str r2, [r0, #12]
-; CHECK-AAPCS: mov r0, r11
-; CHECK-AAPCS: str r3, [r0, #16]
+; CHECK-AAPCS: mov r7, r0
+; CHECK-AAPCS: adds r7, #8
+; CHECK-AAPCS: stm r7!, {r1, r2, r3}
 ; VLAs present, access via FP
 ; int test_args_vla(int a, int b, int c, int d, int e) {
 ;   int v[a];
@@ -308,11 +303,9 @@ entry:
 ; CHECK-FP-ATPCS-NEXT: adds r0, #8
 ; CHECK-FP-ATPCS-NEXT: stm r0!, {r1, r2, r3}
 ; CHECK-FP-AAPCS:      mov r0, r11
-; CHECK-FP-AAPCS-NEXT: str r1, [r0, #8]
-; CHECK-FP-AAPCS-NEXT: mov r0, r11
-; CHECK-FP-AAPCS-NEXT: str r2, [r0, #12]
-; CHECK-FP-AAPCS-NEXT: mov r0, r11
-; CHECK-FP-AAPCS-NEXT: str r3, [r0, #16]
+; CHECK-FP-AAPCS-NEXT: mov r5, r0
+; CHECK-FP-AAPCS-NEXT: adds r5, #8
+; CHECK-FP-AAPCS-NEXT: stm r5!, {r1, r2, r3}
 
 ; struct S { int x[128]; } s;
 ; int test(S a, int b) {

diff --git a/llvm/test/CodeGen/Thumb2/mve-fpclamptosat_vec.ll b/llvm/test/CodeGen/Thumb2/mve-fpclamptosat_vec.ll
index 7929bba..d3297d2 100644 (file)
--- a/llvm/test/CodeGen/Thumb2/mve-fpclamptosat_vec.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-fpclamptosat_vec.ll
@@ -1890,7 +1890,6 @@ define arm_aapcs_vfpcc <2 x i64> @stest_f64i64_mm(<2 x double> %x) {
 ; CHECK-NEXT:    cmp.w r2, #-1
 ; CHECK-NEXT:    ldr r2, [sp, #8] @ 4-byte Reload
 ; CHECK-NEXT:    csel r5, r5, r8, gt
-; CHECK-NEXT:    mov.w r8, #0
 ; CHECK-NEXT:    cmp r2, #0
 ; CHECK-NEXT:    ldr r2, [sp, #4] @ 4-byte Reload
 ; CHECK-NEXT:    csel r5, r6, r5, eq
@@ -2152,7 +2151,6 @@ define arm_aapcs_vfpcc <2 x i64> @stest_f32i64_mm(<2 x float> %x) {
 ; CHECK-NEXT:    cmp.w r2, #-1
 ; CHECK-NEXT:    ldr r2, [sp, #12] @ 4-byte Reload
 ; CHECK-NEXT:    csel r5, r5, r8, gt
-; CHECK-NEXT:    mov.w r8, #0
 ; CHECK-NEXT:    cmp r2, #0
 ; CHECK-NEXT:    ldr r2, [sp, #8] @ 4-byte Reload
 ; CHECK-NEXT:    csel r5, r6, r5, eq
@@ -2410,7 +2408,6 @@ define arm_aapcs_vfpcc <2 x i64> @stest_f16i64_mm(<2 x half> %x) {
 ; CHECK-NEXT:    cmp.w r2, #-1
 ; CHECK-NEXT:    ldr r2, [sp, #8] @ 4-byte Reload
 ; CHECK-NEXT:    csel r5, r5, r8, gt
-; CHECK-NEXT:    mov.w r8, #0
 ; CHECK-NEXT:    cmp r2, #0
 ; CHECK-NEXT:    ldr r2, [sp, #4] @ 4-byte Reload
 ; CHECK-NEXT:    csel r5, r6, r5, eq

diff --git a/llvm/test/CodeGen/X86/2008-04-09-BranchFolding.ll b/llvm/test/CodeGen/X86/2008-04-09-BranchFolding.ll
index 878af85..45b5ce5 100644 (file)
--- a/llvm/test/CodeGen/X86/2008-04-09-BranchFolding.ll
+++ b/llvm/test/CodeGen/X86/2008-04-09-BranchFolding.ll
@@ -18,7 +18,6 @@ define fastcc ptr @pushdecl(ptr %x) nounwind  {
 ; CHECK-NEXT:    xorl %eax, %eax
 ; CHECK-NEXT:    retl
 ; CHECK-NEXT:  .LBB0_1: # %bb160
-; CHECK-NEXT:    movb $1, %al
 ; CHECK-NEXT:    testb %al, %al
 ; CHECK-NEXT:    xorl %eax, %eax
 ; CHECK-NEXT:    retl

diff --git a/llvm/test/CodeGen/X86/2008-04-16-ReMatBug.ll b/llvm/test/CodeGen/X86/2008-04-16-ReMatBug.ll
index c5e1ff6..b32afdc 100644 (file)
--- a/llvm/test/CodeGen/X86/2008-04-16-ReMatBug.ll
+++ b/llvm/test/CodeGen/X86/2008-04-16-ReMatBug.ll
@@ -27,7 +27,6 @@ define i16 @SQLDriversW(ptr %henv, i16 zeroext  %fDir, ptr %szDrvDesc, i16 signe
 ; CHECK-NEXT:    jne LBB0_6
 ; CHECK-NEXT:  ## %bb.4: ## %bb37
 ; CHECK-NEXT:    movw $0, 40(%edi)
-; CHECK-NEXT:    movb $1, %al
 ; CHECK-NEXT:    testb %al, %al
 ; CHECK-NEXT:    leal (,%ecx,4), %ecx
 ; CHECK-NEXT:    leal (,%ebx,4), %edx

diff --git a/llvm/test/CodeGen/X86/AMX/amx-across-func.ll b/llvm/test/CodeGen/X86/AMX/amx-across-func.ll
index c67208d..ae0be9b 100644 (file)
--- a/llvm/test/CodeGen/X86/AMX/amx-across-func.ll
+++ b/llvm/test/CodeGen/X86/AMX/amx-across-func.ll
@@ -58,7 +58,6 @@ define dso_local void @test_api(i16 signext %0, i16 signext %1) nounwind {
 ; CHECK-NEXT:    tileloadd (%rax,%r14), %tmm0
 ; CHECK-NEXT:    movabsq $64, %rcx
 ; CHECK-NEXT:    tileloadd 1088(%rsp,%rcx), %tmm1 # 1024-byte Folded Reload
-; CHECK-NEXT:    movabsq $64, %rcx
 ; CHECK-NEXT:    tileloadd 64(%rsp,%rcx), %tmm2 # 1024-byte Folded Reload
 ; CHECK-NEXT:    tdpbssd %tmm2, %tmm1, %tmm0
 ; CHECK-NEXT:    tilestored %tmm0, (%rax,%r14)

diff --git a/llvm/test/CodeGen/X86/AMX/amx-spill-merge.ll b/llvm/test/CodeGen/X86/AMX/amx-spill-merge.ll
index e4a2279..9f0d8ae 100644 (file)
--- a/llvm/test/CodeGen/X86/AMX/amx-spill-merge.ll
+++ b/llvm/test/CodeGen/X86/AMX/amx-spill-merge.ll
@@ -46,7 +46,6 @@ define dso_local void @test_api(i16 signext %0, i16 signext %1) nounwind {
 ; CHECK-NEXT:    movabsq $64, %rax
 ; CHECK-NEXT:    tilestored %tmm5, 1088(%rsp,%rax) # 1024-byte Folded Spill
 ; CHECK-NEXT:    tdpbssd %tmm1, %tmm0, %tmm5
-; CHECK-NEXT:    movabsq $64, %rax
 ; CHECK-NEXT:    tilestored %tmm5, 64(%rsp,%rax) # 1024-byte Folded Spill
 ; CHECK-NEXT:    xorl %eax, %eax
 ; CHECK-NEXT:    vzeroupper
@@ -64,7 +63,6 @@ define dso_local void @test_api(i16 signext %0, i16 signext %1) nounwind {
 ; CHECK-NEXT:    movabsq $64, %rax
 ; CHECK-NEXT:    tilestored %tmm5, 1088(%rsp,%rax) # 1024-byte Folded Spill
 ; CHECK-NEXT:    tdpbssd %tmm3, %tmm2, %tmm5
-; CHECK-NEXT:    movabsq $64, %rax
 ; CHECK-NEXT:    tilestored %tmm5, 64(%rsp,%rax) # 1024-byte Folded Spill
 ; CHECK-NEXT:    xorl %eax, %eax
 ; CHECK-NEXT:    vzeroupper

diff --git a/llvm/test/CodeGen/X86/fast-isel-stackcheck.ll b/llvm/test/CodeGen/X86/fast-isel-stackcheck.ll
index 1398b30..10e192e 100644 (file)
--- a/llvm/test/CodeGen/X86/fast-isel-stackcheck.ll
+++ b/llvm/test/CodeGen/X86/fast-isel-stackcheck.ll
@@ -8,7 +8,6 @@ target triple = "x86_64-apple-macosx"
 
 ; CHECK-LABEL: foo:
 ; CHECK: movq ___stack_chk_guard@GOTPCREL(%rip), %rax
-; CHECK: movq ___stack_chk_guard@GOTPCREL(%rip), %rax
 define void @foo() #0 {
 entry:
   %_tags = alloca [3 x i32], align 4

diff --git a/llvm/test/CodeGen/X86/fshl.ll b/llvm/test/CodeGen/X86/fshl.ll
index eaf794c..bf561a1 100644 (file)
--- a/llvm/test/CodeGen/X86/fshl.ll
+++ b/llvm/test/CodeGen/X86/fshl.ll
@@ -338,26 +338,24 @@ define i128 @var_shift_i128(i128 %x, i128 %y, i128 %z) nounwind {
 ; X86-SLOW-NEXT:    movl %edx, %ebx
 ; X86-SLOW-NEXT:    movl %esi, %edx
 ; X86-SLOW-NEXT:    movl {{[0-9]+}}(%esp), %esi
-; X86-SLOW-NEXT:    jmp .LBB6_3
+; X86-SLOW-NEXT:    testb $32, %al
+; X86-SLOW-NEXT:    je .LBB6_5
+; X86-SLOW-NEXT:  .LBB6_4:
+; X86-SLOW-NEXT:    movl %esi, (%esp) # 4-byte Spill
+; X86-SLOW-NEXT:    movl %ebp, %esi
+; X86-SLOW-NEXT:    movl %edx, %ebp
+; X86-SLOW-NEXT:    movl %ecx, %edx
+; X86-SLOW-NEXT:    jmp .LBB6_6
 ; X86-SLOW-NEXT:  .LBB6_1:
 ; X86-SLOW-NEXT:    movl {{[0-9]+}}(%esp), %ebx
 ; X86-SLOW-NEXT:    movl {{[0-9]+}}(%esp), %ecx
-; X86-SLOW-NEXT:  .LBB6_3:
-; X86-SLOW-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-SLOW-NEXT:    testb $32, %al
 ; X86-SLOW-NEXT:    jne .LBB6_4
-; X86-SLOW-NEXT:  # %bb.5:
+; X86-SLOW-NEXT:  .LBB6_5:
 ; X86-SLOW-NEXT:    movl %ecx, %ebx
 ; X86-SLOW-NEXT:    movl %edi, (%esp) # 4-byte Spill
-; X86-SLOW-NEXT:    jmp .LBB6_6
-; X86-SLOW-NEXT:  .LBB6_4:
-; X86-SLOW-NEXT:    movl %esi, (%esp) # 4-byte Spill
-; X86-SLOW-NEXT:    movl %ebp, %esi
-; X86-SLOW-NEXT:    movl %edx, %ebp
-; X86-SLOW-NEXT:    movl %ecx, %edx
 ; X86-SLOW-NEXT:  .LBB6_6:
 ; X86-SLOW-NEXT:    movl %edx, %edi
-; X86-SLOW-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-SLOW-NEXT:    movl %eax, %ecx
 ; X86-SLOW-NEXT:    shll %cl, %edi
 ; X86-SLOW-NEXT:    shrl %ebx

diff --git a/llvm/test/CodeGen/X86/masked_load.ll b/llvm/test/CodeGen/X86/masked_load.ll
index 7bdc34b..e226279 100644 (file)
--- a/llvm/test/CodeGen/X86/masked_load.ll
+++ b/llvm/test/CodeGen/X86/masked_load.ll
@@ -247,7 +247,6 @@ define <4 x double> @load_v4f64_v4i32_zero(<4 x i32> %trigger, ptr %addr) {
 ; SSE-NEXT:    retq
 ; SSE-NEXT:  LBB3_1: ## %cond.load
 ; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSE-NEXT:    xorps %xmm1, %xmm1
 ; SSE-NEXT:    testb $2, %al
 ; SSE-NEXT:    je LBB3_4
 ; SSE-NEXT:  LBB3_3: ## %cond.load1
@@ -1129,7 +1128,6 @@ define <8 x float> @load_v8f32_v8i1_zero(<8 x i1> %mask, ptr %addr) {
 ; SSE2-NEXT:    retq
 ; SSE2-NEXT:  LBB10_1: ## %cond.load
 ; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    xorps %xmm1, %xmm1
 ; SSE2-NEXT:    testb $2, %al
 ; SSE2-NEXT:    je LBB10_4
 ; SSE2-NEXT:  LBB10_3: ## %cond.load1
@@ -1209,7 +1207,6 @@ define <8 x float> @load_v8f32_v8i1_zero(<8 x i1> %mask, ptr %addr) {
 ; SSE42-NEXT:    retq
 ; SSE42-NEXT:  LBB10_1: ## %cond.load
 ; SSE42-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE42-NEXT:    xorps %xmm1, %xmm1
 ; SSE42-NEXT:    testb $2, %al
 ; SSE42-NEXT:    je LBB10_4
 ; SSE42-NEXT:  LBB10_3: ## %cond.load1
@@ -2650,7 +2647,6 @@ define <8 x i32> @load_v8i32_v8i1_zero(<8 x i1> %mask, ptr %addr) {
 ; SSE2-NEXT:    retq
 ; SSE2-NEXT:  LBB20_1: ## %cond.load
 ; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    xorps %xmm1, %xmm1
 ; SSE2-NEXT:    testb $2, %al
 ; SSE2-NEXT:    je LBB20_4
 ; SSE2-NEXT:  LBB20_3: ## %cond.load1
@@ -2730,7 +2726,6 @@ define <8 x i32> @load_v8i32_v8i1_zero(<8 x i1> %mask, ptr %addr) {
 ; SSE42-NEXT:    retq
 ; SSE42-NEXT:  LBB20_1: ## %cond.load
 ; SSE42-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE42-NEXT:    pxor %xmm1, %xmm1
 ; SSE42-NEXT:    testb $2, %al
 ; SSE42-NEXT:    je LBB20_4
 ; SSE42-NEXT:  LBB20_3: ## %cond.load1

diff --git a/llvm/test/CodeGen/X86/oddshuffles.ll b/llvm/test/CodeGen/X86/oddshuffles.ll
index c28f30f..f8e9bf3 100644 (file)
--- a/llvm/test/CodeGen/X86/oddshuffles.ll
+++ b/llvm/test/CodeGen/X86/oddshuffles.ll
@@ -2231,7 +2231,6 @@ define <16 x i32> @splat_v3i32(ptr %ptr) {
 ; SSE42-NEXT:    pblendw {{.*#+}} xmm2 = xmm0[0,1],xmm2[2,3,4,5,6,7]
 ; SSE42-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
 ; SSE42-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,0,1]
-; SSE42-NEXT:    pxor %xmm1, %xmm1
 ; SSE42-NEXT:    xorps %xmm3, %xmm3
 ; SSE42-NEXT:    retq
 ;

diff --git a/llvm/test/CodeGen/X86/opt-pipeline.ll b/llvm/test/CodeGen/X86/opt-pipeline.ll
index 9de77d2..24f5445 100644 (file)
--- a/llvm/test/CodeGen/X86/opt-pipeline.ll
+++ b/llvm/test/CodeGen/X86/opt-pipeline.ll
@@ -170,6 +170,7 @@
 ; CHECK-NEXT:       Machine Optimization Remark Emitter
 ; CHECK-NEXT:       Shrink Wrapping analysis
 ; CHECK-NEXT:       Prologue/Epilogue Insertion & Frame Finalization
+; CHECK-NEXT:       Machine Late Instructions Cleanup Pass
 ; CHECK-NEXT:       Control Flow Optimizer
 ; CHECK-NEXT:       Lazy Machine Block Frequency Analysis
 ; CHECK-NEXT:       Tail Duplication

diff --git a/llvm/test/CodeGen/X86/sdiv_fix_sat.ll b/llvm/test/CodeGen/X86/sdiv_fix_sat.ll
index 78e4705..6bec366 100644 (file)
--- a/llvm/test/CodeGen/X86/sdiv_fix_sat.ll
+++ b/llvm/test/CodeGen/X86/sdiv_fix_sat.ll
@@ -1240,7 +1240,6 @@ define <4 x i32> @vec(<4 x i32> %x, <4 x i32> %y) nounwind {
 ; X86-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload
 ; X86-NEXT:    testl %ecx, %ecx
 ; X86-NEXT:    cmovsl %esi, %eax
-; X86-NEXT:    movl $0, %esi
 ; X86-NEXT:    movl $-1, %ebx
 ; X86-NEXT:    cmovsl %ebx, %edi
 ; X86-NEXT:    andl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload

diff --git a/llvm/test/CodeGen/X86/shift-i128.ll b/llvm/test/CodeGen/X86/shift-i128.ll
index 8fb34ff..1006477 100644 (file)
--- a/llvm/test/CodeGen/X86/shift-i128.ll
+++ b/llvm/test/CodeGen/X86/shift-i128.ll
@@ -533,7 +533,6 @@ define void @test_lshr_v2i128(<2 x i128> %x, <2 x i128> %a, ptr nocapture %r) no
 ; i686-NEXT:  .LBB6_9: # %entry
 ; i686-NEXT:    movl %edi, %esi
 ; i686-NEXT:    movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
-; i686-NEXT:    movl {{[0-9]+}}(%esp), %ebp
 ; i686-NEXT:    movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload
 ; i686-NEXT:    shrl %cl, %ebp
 ; i686-NEXT:    testb $32, %cl
@@ -846,7 +845,6 @@ define void @test_ashr_v2i128(<2 x i128> %x, <2 x i128> %a, ptr nocapture %r) no
 ; i686-NEXT:    movl {{[0-9]+}}(%esp), %esi
 ; i686-NEXT:    movb $64, %cl
 ; i686-NEXT:    subb %dl, %cl
-; i686-NEXT:    movl {{[0-9]+}}(%esp), %ebp
 ; i686-NEXT:    movl {{[0-9]+}}(%esp), %ebx
 ; i686-NEXT:    shldl %cl, %ebx, %ebp
 ; i686-NEXT:    movl %ebp, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill

diff --git a/llvm/test/CodeGen/X86/ushl_sat_vec.ll b/llvm/test/CodeGen/X86/ushl_sat_vec.ll
index 5246c8b..a73fa32 100644 (file)
--- a/llvm/test/CodeGen/X86/ushl_sat_vec.ll
+++ b/llvm/test/CodeGen/X86/ushl_sat_vec.ll
@@ -354,7 +354,6 @@ define <8 x i16> @vec_v8i16(<8 x i16> %x, <8 x i16> %y) nounwind {
 ; X86-NEXT:    cmpw %si, %dx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %edx
 ; X86-NEXT:    cmovnel %eax, %ebx
-; X86-NEXT:    movl $65535, %eax # imm = 0xFFFF
 ; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl %edx, %esi
 ; X86-NEXT:    shll %cl, %esi

diff --git a/llvm/test/CodeGen/X86/vec_extract.ll b/llvm/test/CodeGen/X86/vec_extract.ll
index dd0a50e..e753019 100644 (file)
--- a/llvm/test/CodeGen/X86/vec_extract.ll
+++ b/llvm/test/CodeGen/X86/vec_extract.ll
@@ -111,14 +111,12 @@ define <4 x i32> @ossfuzz15662(ptr %in) {
 ; X32:       # %bb.0:
 ; X32-NEXT:    xorps %xmm0, %xmm0
 ; X32-NEXT:    movaps %xmm0, (%eax)
-; X32-NEXT:    xorps %xmm0, %xmm0
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: ossfuzz15662:
 ; X64:       # %bb.0:
 ; X64-NEXT:    xorps %xmm0, %xmm0
 ; X64-NEXT:    movaps %xmm0, (%rax)
-; X64-NEXT:    xorps %xmm0, %xmm0
 ; X64-NEXT:    retq
    %C10 = icmp ule i1 false, false
    %C3 = icmp ule i1 true, undef

diff --git a/llvm/test/CodeGen/X86/vec_shift5.ll b/llvm/test/CodeGen/X86/vec_shift5.ll
index ab16e1a..b7cfdeb 100644 (file)
--- a/llvm/test/CodeGen/X86/vec_shift5.ll
+++ b/llvm/test/CodeGen/X86/vec_shift5.ll
@@ -178,14 +178,12 @@ define <4 x i32> @test17(<4 x i32> %a0, ptr %dummy) {
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    xorps %xmm0, %xmm0
 ; X86-NEXT:    movaps %xmm0, (%eax)
-; X86-NEXT:    xorps %xmm0, %xmm0
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: test17:
 ; X64:       # %bb.0:
 ; X64-NEXT:    xorps %xmm0, %xmm0
 ; X64-NEXT:    movaps %xmm0, (%rdi)
-; X64-NEXT:    xorps %xmm0, %xmm0
 ; X64-NEXT:    retq
   %a = call <4 x i32> @llvm.x86.sse2.pslli.d(<4 x i32> undef, i32 6)
   store <4 x i32> %a, ptr %dummy
@@ -199,14 +197,12 @@ define <4 x i32> @test18(<4 x i32> %a0, ptr %dummy) {
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    xorps %xmm0, %xmm0
 ; X86-NEXT:    movaps %xmm0, (%eax)
-; X86-NEXT:    xorps %xmm0, %xmm0
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: test18:
 ; X64:       # %bb.0:
 ; X64-NEXT:    xorps %xmm0, %xmm0
 ; X64-NEXT:    movaps %xmm0, (%rdi)
-; X64-NEXT:    xorps %xmm0, %xmm0
 ; X64-NEXT:    retq
   %a = call <4 x i32> @llvm.x86.sse2.pslli.d(<4 x i32> undef, i32 3)
   store <4 x i32> %a, ptr %dummy

diff --git a/llvm/test/CodeGen/XCore/scavenging.ll b/llvm/test/CodeGen/XCore/scavenging.ll
index adf313e..dc12698 100644 (file)
--- a/llvm/test/CodeGen/XCore/scavenging.ll
+++ b/llvm/test/CodeGen/XCore/scavenging.ll
@@ -87,13 +87,10 @@ declare void @g(i32*, i32*)
 ; CHECK: ldaw r0, sp[0]
 ; CHECK: ldw r5, cp[[[INDEX1]]]
 ; CHECK: stw r1, r0[r5]
-; CHECK: ldaw r0, sp[0]
 ; CHECK: ldw r1, cp[[[INDEX2]]]
 ; CHECK: stw r2, r0[r1]
-; CHECK: ldaw r0, sp[0]
 ; CHECK: ldw r1, cp[[[INDEX3]]]
 ; CHECK: stw r3, r0[r1]
-; CHECK: ldaw r0, sp[0]
 ; CHECK: ldw r1, cp[[[INDEX4]]]
 ; CHECK: stw r11, r0[r1]
 ; CHECK: ldaw sp, sp[65535]