diff --git a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
--- a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
@@ -245,8 +245,8 @@
                              const SIRegisterInfo *TRI, unsigned OpNo) const;
 
   bool counterOutOfOrder(InstCounterType T) const;
-  bool simplifyWaitcnt(AMDGPU::Waitcnt &Wait) const;
-  bool simplifyWaitcnt(InstCounterType T, unsigned &Count) const;
+  void simplifyWaitcnt(AMDGPU::Waitcnt &Wait) const;
+  void simplifyWaitcnt(InstCounterType T, unsigned &Count) const;
   void determineWait(InstCounterType T, unsigned ScoreToWait,
                      AMDGPU::Waitcnt &Wait) const;
   void applyWaitcnt(const AMDGPU::Waitcnt &Wait);
@@ -418,7 +418,7 @@
     }
 
     if (DebugCounter::isCounterSet(ForceLgkmCounter) &&
-         DebugCounter::shouldExecute(ForceLgkmCounter)) {
+        DebugCounter::shouldExecute(ForceLgkmCounter)) {
       ForceEmitWaitcnt[LGKM_CNT] = true;
     } else {
       ForceEmitWaitcnt[LGKM_CNT] = false;
@@ -442,6 +442,9 @@
                                WaitcntBrackets *ScoreBrackets);
   bool insertWaitcntInBlock(MachineFunction &MF, MachineBasicBlock &Block,
                             WaitcntBrackets &ScoreBrackets);
+  bool applyPreexistingWaitcnt(WaitcntBrackets &ScoreBrackets,
+                               MachineInstr &OldWaitcntInstr,
+                               AMDGPU::Waitcnt &Wait, const MachineInstr *MI);
 };
 
 } // end anonymous namespace
@@ -708,22 +711,23 @@
 
 /// Simplify the waitcnt, in the sense of removing redundant counts, and return
 /// whether a waitcnt instruction is needed at all.
-bool WaitcntBrackets::simplifyWaitcnt(AMDGPU::Waitcnt &Wait) const {
-  return simplifyWaitcnt(VM_CNT, Wait.VmCnt) |
-         simplifyWaitcnt(EXP_CNT, Wait.ExpCnt) |
-         simplifyWaitcnt(LGKM_CNT, Wait.LgkmCnt) |
-         simplifyWaitcnt(VS_CNT, Wait.VsCnt);
+void WaitcntBrackets::simplifyWaitcnt(AMDGPU::Waitcnt &Wait) const {
+  simplifyWaitcnt(VM_CNT, Wait.VmCnt);
+  simplifyWaitcnt(EXP_CNT, Wait.ExpCnt);
+  simplifyWaitcnt(LGKM_CNT, Wait.LgkmCnt);
+  simplifyWaitcnt(VS_CNT, Wait.VsCnt);
 }
 
-bool WaitcntBrackets::simplifyWaitcnt(InstCounterType T,
+void WaitcntBrackets::simplifyWaitcnt(InstCounterType T,
                                       unsigned &Count) const {
   const unsigned LB = getScoreLB(T);
   const unsigned UB = getScoreUB(T);
-  if (Count < UB && UB - Count > LB)
-    return true;
 
-  Count = ~0u;
-  return false;
+  // The number of outstanding events for this type, T, can be calculated
+  // as (UB - LB). If the current Count is greater than or equal to the number
+  // of outstanding events, then the wait for this counter is redundant.
+  if (Count >= UB - LB)
+    Count = ~0u;
 }
 
 void WaitcntBrackets::determineWait(InstCounterType T, unsigned ScoreToWait,
@@ -798,6 +802,107 @@
   return new SIInsertWaitcnts();
 }
 
+/// Combine consecutive waitcnt instructions that precede \p MI and follow
+/// \p OldWaitcntInstr and apply any extra wait from waitcnt that were added
+/// by previous passes. Currently this pass conservatively assumes that these
+/// preexisting waitcnt are required for correctness.
+bool SIInsertWaitcnts::applyPreexistingWaitcnt(WaitcntBrackets &ScoreBrackets,
+                                               MachineInstr &OldWaitcntInstr,
+                                               AMDGPU::Waitcnt &Wait,
+                                               const MachineInstr *MI) {
+  bool Modified = false;
+  MachineInstr *WaitcntInstr = nullptr;
+  MachineInstr *WaitcntVsCntInstr = nullptr;
+  for (auto II = OldWaitcntInstr.getIterator(), NextI = std::next(II);
+       &*II != MI; II = NextI, ++NextI) {
+    if (II->isMetaInstruction())
+      continue;
+
+    if (II->getOpcode() == AMDGPU::S_WAITCNT) {
+      // Conservatively update required wait if this waitcnt was added in an
+      // earlier pass. In this case it will not exist in the tracked waitcnt
+      // set.
+      if (!TrackedWaitcntSet.count(&*II)) {
+        unsigned IEnc = II->getOperand(0).getImm();
+        AMDGPU::Waitcnt OldWait = AMDGPU::decodeWaitcnt(IV, IEnc);
+        Wait = Wait.combined(OldWait);
+      }
+
+      // Merge consecutive waitcnt of the same type by erasing multiples.
+      if (!WaitcntInstr) {
+        WaitcntInstr = &*II;
+      } else {
+        II->eraseFromParent();
+        Modified = true;
+      }
+
+    } else {
+      assert(II->getOpcode() == AMDGPU::S_WAITCNT_VSCNT);
+      assert(II->getOperand(0).getReg() == AMDGPU::SGPR_NULL);
+      if (!TrackedWaitcntSet.count(&*II)) {
+        unsigned OldVSCnt =
+            TII->getNamedOperand(*II, AMDGPU::OpName::simm16)->getImm();
+        Wait.VsCnt = std::min(Wait.VsCnt, OldVSCnt);
+      }
+
+      if (!WaitcntVsCntInstr) {
+        WaitcntVsCntInstr = &*II;
+      } else {
+        II->eraseFromParent();
+        Modified = true;
+      }
+    }
+  }
+
+  // Updated encoding of merged waitcnt with the required wait.
+  if (WaitcntInstr) {
+    if (Wait.hasWaitExceptVsCnt()) {
+      unsigned NewEnc = AMDGPU::encodeWaitcnt(IV, Wait);
+      unsigned OldEnc = WaitcntInstr->getOperand(0).getImm();
+      if (OldEnc != NewEnc) {
+        WaitcntInstr->getOperand(0).setImm(NewEnc);
+        Modified = true;
+      }
+      ScoreBrackets.applyWaitcnt(Wait);
+      Wait.VmCnt = ~0u;
+      Wait.LgkmCnt = ~0u;
+      Wait.ExpCnt = ~0u;
+
+      LLVM_DEBUG(dbgs() << "generateWaitcntInstBefore\n"
+                        << "Old Instr: " << MI << "New Instr: " << *WaitcntInstr
+                        << '\n');
+    } else {
+      WaitcntInstr->eraseFromParent();
+      Modified = true;
+    }
+  }
+
+  if (WaitcntVsCntInstr) {
+    if (Wait.hasWaitVsCnt()) {
+      assert(ST->hasVscnt());
+      unsigned OldVSCnt =
+          TII->getNamedOperand(*WaitcntVsCntInstr, AMDGPU::OpName::simm16)
+              ->getImm();
+      if (Wait.VsCnt != OldVSCnt) {
+        TII->getNamedOperand(*WaitcntVsCntInstr, AMDGPU::OpName::simm16)
+            ->setImm(Wait.VsCnt);
+        Modified = true;
+      }
+      ScoreBrackets.applyWaitcnt(Wait);
+      Wait.VsCnt = ~0u;
+
+      LLVM_DEBUG(dbgs() << "generateWaitcntInstBefore\n"
+                        << "Old Instr: " << MI
+                        << "New Instr: " << *WaitcntVsCntInstr << '\n');
+    } else {
+      WaitcntVsCntInstr->eraseFromParent();
+      Modified = true;
+    }
+  }
+
+  return Modified;
+}
+
 static bool readsVCCZ(const MachineInstr &MI) {
   unsigned Opc = MI.getOpcode();
   return (Opc == AMDGPU::S_CBRANCH_VCCNZ || Opc == AMDGPU::S_CBRANCH_VCCZ) &&
@@ -833,12 +938,12 @@
     MachineInstr &MI, WaitcntBrackets &ScoreBrackets,
     MachineInstr *OldWaitcntInstr) {
   setForceEmitWaitcnt();
-  bool IsForceEmitWaitcnt = isForceEmitWaitcnt();
 
   if (MI.isMetaInstruction())
     return false;
 
   AMDGPU::Waitcnt Wait;
+  bool Modified = false;
 
   // See if this instruction has a forced S_WAITCNT VM.
   // TODO: Handle other cases of NeedsWaitcntVmBefore()
@@ -1053,32 +1158,8 @@
     }
   }
 
-  // Early-out if no wait is indicated.
-  if (!ScoreBrackets.simplifyWaitcnt(Wait) && !IsForceEmitWaitcnt) {
-    bool Modified = false;
-    if (OldWaitcntInstr) {
-      for (auto II = OldWaitcntInstr->getIterator(), NextI = std::next(II);
-           &*II != &MI; II = NextI, ++NextI) {
-        if (II->isDebugInstr())
-          continue;
-
-        if (TrackedWaitcntSet.count(&*II)) {
-          TrackedWaitcntSet.erase(&*II);
-          II->eraseFromParent();
-          Modified = true;
-        } else if (II->getOpcode() == AMDGPU::S_WAITCNT) {
-          int64_t Imm = II->getOperand(0).getImm();
-          ScoreBrackets.applyWaitcnt(AMDGPU::decodeWaitcnt(IV, Imm));
-        } else {
-          assert(II->getOpcode() == AMDGPU::S_WAITCNT_VSCNT);
-          assert(II->getOperand(0).getReg() == AMDGPU::SGPR_NULL);
-          auto W = TII->getNamedOperand(*II, AMDGPU::OpName::simm16)->getImm();
-          ScoreBrackets.applyWaitcnt(AMDGPU::Waitcnt(~0u, ~0u, ~0u, W));
-        }
-      }
-    }
-    return Modified;
-  }
+  // Verify that the wait is actually needed.
+  ScoreBrackets.simplifyWaitcnt(Wait);
 
   if (ForceEmitZeroWaitcnts)
     Wait = AMDGPU::Waitcnt::allZero(ST->hasVscnt());
@@ -1092,57 +1173,19 @@
   if (ForceEmitWaitcnt[VS_CNT])
     Wait.VsCnt = 0;
 
-  ScoreBrackets.applyWaitcnt(Wait);
-
-  AMDGPU::Waitcnt OldWait;
-  bool Modified = false;
-
   if (OldWaitcntInstr) {
-    for (auto II = OldWaitcntInstr->getIterator(), NextI = std::next(II);
-         &*II != &MI; II = NextI, NextI++) {
-      if (II->isDebugInstr())
-        continue;
-
-      if (II->getOpcode() == AMDGPU::S_WAITCNT) {
-        unsigned IEnc = II->getOperand(0).getImm();
-        AMDGPU::Waitcnt IWait = AMDGPU::decodeWaitcnt(IV, IEnc);
-        OldWait = OldWait.combined(IWait);
-        if (!TrackedWaitcntSet.count(&*II))
-          Wait = Wait.combined(IWait);
-        unsigned NewEnc = AMDGPU::encodeWaitcnt(IV, Wait);
-        if (IEnc != NewEnc) {
-          II->getOperand(0).setImm(NewEnc);
-          Modified = true;
-        }
-        Wait.VmCnt = ~0u;
-        Wait.LgkmCnt = ~0u;
-        Wait.ExpCnt = ~0u;
-      } else {
-        assert(II->getOpcode() == AMDGPU::S_WAITCNT_VSCNT);
-        assert(II->getOperand(0).getReg() == AMDGPU::SGPR_NULL);
-
-        unsigned ICnt = TII->getNamedOperand(*II, AMDGPU::OpName::simm16)
-                        ->getImm();
-        OldWait.VsCnt = std::min(OldWait.VsCnt, ICnt);
-        if (!TrackedWaitcntSet.count(&*II))
-          Wait.VsCnt = std::min(Wait.VsCnt, ICnt);
-        if (Wait.VsCnt != ICnt) {
-          TII->getNamedOperand(*II, AMDGPU::OpName::simm16)->setImm(Wait.VsCnt);
-          Modified = true;
-        }
-        Wait.VsCnt = ~0u;
-      }
-
-      LLVM_DEBUG(dbgs() << "generateWaitcntInstBefore\n"
-                        << "Old Instr: " << MI
-                        << "New Instr: " << *II << '\n');
-
-      if (!Wait.hasWait())
-        return Modified;
-    }
+    // Try to merge the required wait with preexisting waitcnt instructions.
+    // Also erase redundant waitcnt.
+    Modified =
+        applyPreexistingWaitcnt(ScoreBrackets, *OldWaitcntInstr, Wait, &MI);
+  } else {
+    // Update waitcnt brackets after determining the required wait.
+    ScoreBrackets.applyWaitcnt(Wait);
   }
 
-  if (Wait.VmCnt != ~0u || Wait.LgkmCnt != ~0u || Wait.ExpCnt != ~0u) {
+  // Build new waitcnt instructions unless no wait is needed or the old waitcnt
+  // instruction was modified to handle the required wait.
+  if (Wait.hasWaitExceptVsCnt()) {
     unsigned Enc = AMDGPU::encodeWaitcnt(IV, Wait);
     auto SWaitInst = BuildMI(*MI.getParent(), MI.getIterator(),
                              MI.getDebugLoc(), TII->get(AMDGPU::S_WAITCNT))
@@ -1155,7 +1198,7 @@
                       << "New Instr: " << *SWaitInst << '\n');
   }
 
-  if (Wait.VsCnt != ~0u) {
+  if (Wait.hasWaitVsCnt()) {
     assert(ST->hasVscnt());
 
     auto SWaitInst =
@@ -1430,7 +1473,8 @@
        Iter != E;) {
     MachineInstr &Inst = *Iter;
 
-    // Track pre-existing waitcnts from earlier iterations.
+    // Track pre-existing waitcnts that were added in earlier iterations or by
+    // the memory legalizer.
     if (Inst.getOpcode() == AMDGPU::S_WAITCNT ||
         (Inst.getOpcode() == AMDGPU::S_WAITCNT_VSCNT &&
          Inst.getOperand(0).isReg() &&
diff --git a/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.h b/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.h
--- a/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.h
+++ b/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.h
@@ -477,6 +477,14 @@
     return VmCnt != ~0u || ExpCnt != ~0u || LgkmCnt != ~0u || VsCnt != ~0u;
   }
 
+  bool hasWaitExceptVsCnt() const {
+    return VmCnt != ~0u || ExpCnt != ~0u || LgkmCnt != ~0u;
+  }
+
+  bool hasWaitVsCnt() const {
+    return VsCnt != ~0u;
+  }
+
   bool dominates(const Waitcnt &Other) const {
     return VmCnt <= Other.VmCnt && ExpCnt <= Other.ExpCnt &&
            LgkmCnt <= Other.LgkmCnt && VsCnt <= Other.VsCnt;
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/flat-scratch.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/flat-scratch.ll
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/flat-scratch.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/flat-scratch.ll
@@ -184,7 +184,6 @@
 ; GFX9-NEXT:    s_lshl_b32 s1, s0, 2
 ; GFX9-NEXT:    s_and_b32 s0, s0, 15
 ; GFX9-NEXT:    s_lshl_b32 s0, s0, 2
-; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX9-NEXT:    s_add_u32 s1, 0x104, s1
 ; GFX9-NEXT:    scratch_store_dword off, v0, s1
@@ -357,7 +356,6 @@
 ; GFX9-NEXT:    s_lshl_b32 s1, s0, 2
 ; GFX9-NEXT:    s_and_b32 s0, s0, 15
 ; GFX9-NEXT:    s_lshl_b32 s0, s0, 2
-; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX9-NEXT:    s_add_u32 s1, 0x4004, s1
 ; GFX9-NEXT:    scratch_store_dword off, v0, s1
diff --git a/llvm/test/CodeGen/AMDGPU/atomicrmw-nand.ll b/llvm/test/CodeGen/AMDGPU/atomicrmw-nand.ll
--- a/llvm/test/CodeGen/AMDGPU/atomicrmw-nand.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomicrmw-nand.ll
@@ -72,7 +72,6 @@
 ; GCN-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] glc
 ; GCN-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GCN-NEXT:    buffer_wbinvl1_vol
-; GCN-NEXT:    s_waitcnt lgkmcnt(0)
 ; GCN-NEXT:    v_cmp_eq_u32_e32 vcc, v2, v3
 ; GCN-NEXT:    s_or_b64 s[4:5], vcc, s[4:5]
 ; GCN-NEXT:    s_andn2_b64 exec, exec, s[4:5]
diff --git a/llvm/test/CodeGen/AMDGPU/flat-scratch.ll b/llvm/test/CodeGen/AMDGPU/flat-scratch.ll
--- a/llvm/test/CodeGen/AMDGPU/flat-scratch.ll
+++ b/llvm/test/CodeGen/AMDGPU/flat-scratch.ll
@@ -788,7 +788,6 @@
 ; GFX9-NEXT:    s_lshl_b32 s1, s0, 2
 ; GFX9-NEXT:    s_and_b32 s0, s0, 15
 ; GFX9-NEXT:    s_lshl_b32 s0, s0, 2
-; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX9-NEXT:    s_add_u32 s1, 0x104, s1
 ; GFX9-NEXT:    scratch_store_dword off, v0, s1
@@ -1419,7 +1418,6 @@
 ; GFX9-NEXT:    s_lshl_b32 s1, s0, 2
 ; GFX9-NEXT:    s_and_b32 s0, s0, 15
 ; GFX9-NEXT:    s_lshl_b32 s0, s0, 2
-; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 15
 ; GFX9-NEXT:    s_add_u32 s1, 0x4004, s1
 ; GFX9-NEXT:    scratch_store_dword off, v0, s1
diff --git a/llvm/test/CodeGen/AMDGPU/fp64-atomics-gfx90a.ll b/llvm/test/CodeGen/AMDGPU/fp64-atomics-gfx90a.ll
--- a/llvm/test/CodeGen/AMDGPU/fp64-atomics-gfx90a.ll
+++ b/llvm/test/CodeGen/AMDGPU/fp64-atomics-gfx90a.ll
@@ -630,7 +630,6 @@
 ; GFX90A-NEXT:    flat_atomic_cmpswap_x2 v[0:1], v[4:5], v[0:3] glc
 ; GFX90A-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NEXT:    v_cmp_eq_u64_e32 vcc, v[0:1], v[2:3]
 ; GFX90A-NEXT:    s_or_b64 s[2:3], vcc, s[2:3]
 ; GFX90A-NEXT:    v_pk_mov_b32 v[2:3], v[0:1], v[0:1] op_sel:[0,1]
@@ -706,7 +705,6 @@
 ; GFX90A-NEXT:    flat_atomic_cmpswap_x2 v[2:3], v[0:1], v[2:5] glc
 ; GFX90A-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NEXT:    v_cmp_eq_u64_e32 vcc, v[2:3], v[4:5]
 ; GFX90A-NEXT:    s_or_b64 s[4:5], vcc, s[4:5]
 ; GFX90A-NEXT:    s_andn2_b64 exec, exec, s[4:5]
@@ -731,7 +729,6 @@
 ; GFX90A-NEXT:    flat_atomic_add_f64 v[0:1], v[0:1], v[2:3] glc
 ; GFX90A-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NEXT:    s_setpc_b64 s[30:31]
 main_body:
   %ret = atomicrmw fadd double* %ptr, double 4.0 syncscope("agent") seq_cst
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ds.gws.barrier.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ds.gws.barrier.ll
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ds.gws.barrier.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ds.gws.barrier.ll
@@ -192,7 +192,6 @@
 ; NOLOOP: s_mov_b32 m0, 0{{$}}
 ; NOLOOP: ds_gws_barrier v{{[0-9]+}} offset:7 gds
 ; NOLOOP-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; NOLOOP-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-NEXT: s_waitcnt_vscnt null, 0x0
 ; NOLOOP-NEXT: load_dword
 define amdgpu_kernel void @gws_barrier_fence_after(i32 %val, i32 addrspace(1)* %ptr) #0 {
@@ -220,7 +219,6 @@
 ; NOLOOP: s_mov_b32 m0, 0
 ; NOLOOP: ds_gws_init v0 offset:7 gds
 ; NOLOOP-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; NOLOOP-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-NEXT: s_waitcnt_vscnt null, 0x0
 ; NOLOOP-NEXT: ds_gws_barrier v0 offset:7 gds
 ; NOLOOP-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
diff --git a/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-agent.ll b/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-agent.ll
--- a/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-agent.ll
+++ b/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-agent.ll
@@ -188,7 +188,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s3
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[2:3], v0
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -204,7 +203,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -220,7 +218,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -248,7 +245,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
 ; GFX90A-NOTTGSPLIT-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX90A-NOTTGSPLIT-NEXT:    v_mov_b32_e32 v3, s3
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[2:3], v0
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -285,7 +281,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s3
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[2:3], v0
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -303,7 +298,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -321,7 +315,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -351,7 +344,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
 ; GFX90A-NOTTGSPLIT-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX90A-NOTTGSPLIT-NEXT:    v_mov_b32_e32 v3, s3
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[2:3], v0
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -1121,7 +1113,6 @@
 ; GFX7-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX7-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7-NEXT:    buffer_wbinvl1_vol
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -1138,7 +1129,6 @@
 ; GFX10-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -1155,7 +1145,6 @@
 ; GFX10-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-CU-NEXT:    buffer_gl0_inv
 ; GFX10-CU-NEXT:    buffer_gl1_inv
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -1182,7 +1171,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -1218,7 +1206,6 @@
 ; GFX7-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX7-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7-NEXT:    buffer_wbinvl1_vol
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -1237,7 +1224,6 @@
 ; GFX10-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -1256,7 +1242,6 @@
 ; GFX10-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-CU-NEXT:    buffer_gl0_inv
 ; GFX10-CU-NEXT:    buffer_gl1_inv
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -1285,7 +1270,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -1322,7 +1306,6 @@
 ; GFX7-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX7-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7-NEXT:    buffer_wbinvl1_vol
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -1341,7 +1324,6 @@
 ; GFX10-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -1360,7 +1342,6 @@
 ; GFX10-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-CU-NEXT:    buffer_gl0_inv
 ; GFX10-CU-NEXT:    buffer_gl1_inv
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -1389,7 +1370,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -2456,7 +2436,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -2478,7 +2457,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2500,7 +2478,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -2532,7 +2509,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -2575,7 +2551,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -2599,7 +2574,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2623,7 +2597,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -2657,7 +2630,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -2701,7 +2673,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -2725,7 +2696,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2749,7 +2719,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -2783,7 +2752,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -2826,7 +2794,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -2848,7 +2815,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2870,7 +2836,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -2902,7 +2867,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -2945,7 +2909,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -2969,7 +2932,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2993,7 +2955,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -3027,7 +2988,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -3071,7 +3031,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -3095,7 +3054,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -3119,7 +3077,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -3153,7 +3110,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -3197,7 +3153,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -3221,7 +3176,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -3245,7 +3199,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -3279,7 +3232,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -3323,7 +3275,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -3347,7 +3298,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -3371,7 +3321,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -3405,7 +3354,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
diff --git a/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-system.ll b/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-system.ll
--- a/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-system.ll
+++ b/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-system.ll
@@ -188,7 +188,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s3
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[2:3], v0
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -204,7 +203,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -220,7 +218,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -248,7 +245,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
 ; GFX90A-NOTTGSPLIT-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX90A-NOTTGSPLIT-NEXT:    v_mov_b32_e32 v3, s3
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[2:3], v0
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -285,7 +281,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s3
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[2:3], v0
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -303,7 +298,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -321,7 +315,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -351,7 +344,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
 ; GFX90A-NOTTGSPLIT-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX90A-NOTTGSPLIT-NEXT:    v_mov_b32_e32 v3, s3
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[2:3], v0
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -1121,7 +1113,6 @@
 ; GFX7-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX7-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7-NEXT:    buffer_wbinvl1_vol
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -1138,7 +1129,6 @@
 ; GFX10-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -1155,7 +1145,6 @@
 ; GFX10-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-CU-NEXT:    buffer_gl0_inv
 ; GFX10-CU-NEXT:    buffer_gl1_inv
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -1182,7 +1171,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -1218,7 +1206,6 @@
 ; GFX7-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX7-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7-NEXT:    buffer_wbinvl1_vol
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -1237,7 +1224,6 @@
 ; GFX10-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -1256,7 +1242,6 @@
 ; GFX10-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-CU-NEXT:    buffer_gl0_inv
 ; GFX10-CU-NEXT:    buffer_gl1_inv
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -1285,7 +1270,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -1322,7 +1306,6 @@
 ; GFX7-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX7-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7-NEXT:    buffer_wbinvl1_vol
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -1341,7 +1324,6 @@
 ; GFX10-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -1360,7 +1342,6 @@
 ; GFX10-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-CU-NEXT:    buffer_gl0_inv
 ; GFX10-CU-NEXT:    buffer_gl1_inv
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -1389,7 +1370,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -2456,7 +2436,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -2478,7 +2457,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2500,7 +2478,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -2532,7 +2509,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -2575,7 +2551,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -2599,7 +2574,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2623,7 +2597,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -2657,7 +2630,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -2701,7 +2673,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -2725,7 +2696,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2749,7 +2719,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -2783,7 +2752,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -2826,7 +2794,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -2848,7 +2815,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2870,7 +2836,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -2902,7 +2867,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -2945,7 +2909,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -2969,7 +2932,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2993,7 +2955,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -3027,7 +2988,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -3071,7 +3031,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -3095,7 +3054,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -3119,7 +3077,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -3153,7 +3110,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -3197,7 +3153,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -3221,7 +3176,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -3245,7 +3199,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -3279,7 +3232,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
@@ -3323,7 +3275,6 @@
 ; GFX7-NEXT:    buffer_wbinvl1_vol
 ; GFX7-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    flat_store_dword v[0:1], v2
 ; GFX7-NEXT:    s_endpgm
 ;
@@ -3347,7 +3298,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl1_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -3371,7 +3321,6 @@
 ; GFX10-CU-NEXT:    buffer_gl1_inv
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-CU-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-CU-NEXT:    s_endpgm
 ;
@@ -3405,7 +3354,6 @@
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_atomic_cmpswap v2, v[0:1], v[2:3] offset:16 glc
 ; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    buffer_wbinvl1_vol
-; GFX90A-NOTTGSPLIT-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX90A-NOTTGSPLIT-NEXT:    flat_store_dword v[0:1], v2
 ; GFX90A-NOTTGSPLIT-NEXT:    s_endpgm
 ;
diff --git a/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-volatile.ll b/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-volatile.ll
--- a/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-volatile.ll
+++ b/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-volatile.ll
@@ -298,7 +298,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
diff --git a/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-workgroup.ll b/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-workgroup.ll
--- a/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-workgroup.ll
+++ b/llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-workgroup.ll
@@ -202,7 +202,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -297,7 +296,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s3
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -1103,7 +1101,6 @@
 ; GFX10-WGP-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX10-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -1194,7 +1191,6 @@
 ; GFX10-WGP-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX10-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -1289,7 +1285,6 @@
 ; GFX10-WGP-NEXT:    flat_atomic_swap v2, v[0:1], v2 glc
 ; GFX10-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2363,7 +2358,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2479,7 +2473,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2599,7 +2592,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2716,7 +2708,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2832,7 +2823,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -2952,7 +2942,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -3072,7 +3061,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
@@ -3192,7 +3180,6 @@
 ; GFX10-WGP-NEXT:    buffer_gl0_inv
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX10-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-WGP-NEXT:    flat_store_dword v[0:1], v2
 ; GFX10-WGP-NEXT:    s_endpgm
 ;
diff --git a/llvm/test/CodeGen/AMDGPU/waitcnt-debug.mir b/llvm/test/CodeGen/AMDGPU/waitcnt-debug.mir
--- a/llvm/test/CodeGen/AMDGPU/waitcnt-debug.mir
+++ b/llvm/test/CodeGen/AMDGPU/waitcnt-debug.mir
@@ -25,6 +25,7 @@
 # VM-NEXT: S_NOP 0
 
 # ZERO: S_WAITCNT 0
+# ZERO-NEXT: S_NOP 0
 # ZERO-NEXT: S_WAITCNT 0
 # ZERO-NEXT: S_NOP 0
 # ZERO-NEXT: S_WAITCNT 0
@@ -32,6 +33,8 @@
 
 name:            waitcnt-debug
 liveins:
+machineFunctionInfo:
+  isEntryFunction: true
 body:             |
   bb.0:
     S_NOP 0
diff --git a/llvm/test/CodeGen/AMDGPU/waitcnt-preexisting-vscnt.mir b/llvm/test/CodeGen/AMDGPU/waitcnt-preexisting-vscnt.mir
new file mode 100644
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/waitcnt-preexisting-vscnt.mir
@@ -0,0 +1,131 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs -run-pass si-insert-waitcnts -o - %s | FileCheck -check-prefix=GFX10 %s
+
+---
+name:            test_waitcnt_preexisting_vscnt_unmodified
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1, $vgpr2
+
+    ; GFX10-LABEL: name: test_waitcnt_preexisting_vscnt_unmodified
+    ; GFX10: S_WAITCNT 0
+    ; GFX10: S_WAITCNT_VSCNT undef $sgpr_null, 0
+    ; GFX10: GLOBAL_STORE_DWORD $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec
+    ; GFX10: S_WAITCNT_VSCNT undef $sgpr_null, 0
+    ; GFX10: S_BARRIER
+    ; GFX10: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX10: S_WAITCNT 112
+    ; GFX10: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX10: S_ENDPGM 0
+    GLOBAL_STORE_DWORD $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec
+    S_WAITCNT_VSCNT undef $sgpr_null, 0
+    S_BARRIER
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    S_ENDPGM 0
+...
+
+---
+name:            test_waitcnt_preexisting_vscnt_needs_vscnt
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1, $vgpr2
+
+    ; GFX10-LABEL: name: test_waitcnt_preexisting_vscnt_needs_vscnt
+    ; GFX10: S_WAITCNT 0
+    ; GFX10: S_WAITCNT_VSCNT undef $sgpr_null, 0
+    ; GFX10: GLOBAL_STORE_DWORD $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec
+    ; GFX10: S_WAITCNT_VSCNT undef $sgpr_null, 0
+    ; GFX10: S_BARRIER
+    ; GFX10: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX10: S_WAITCNT 112
+    ; GFX10: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX10: S_ENDPGM 0
+    GLOBAL_STORE_DWORD $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec
+    S_WAITCNT_VSCNT undef $sgpr_null, 1
+    S_BARRIER
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    S_ENDPGM 0
+...
+
+---
+name:            test_waitcnt_preexisting_vscnt_with_other_waitcnt
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1, $vgpr2
+
+    ; GFX10-LABEL: name: test_waitcnt_preexisting_vscnt_with_other_waitcnt
+    ; GFX10: S_WAITCNT 0
+    ; GFX10: S_WAITCNT_VSCNT undef $sgpr_null, 0
+    ; GFX10: GLOBAL_STORE_DWORD $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec
+    ; GFX10: S_WAITCNT 112
+    ; GFX10: S_WAITCNT_VSCNT undef $sgpr_null, 0
+    ; GFX10: S_BARRIER
+    ; GFX10: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX10: S_WAITCNT 112
+    ; GFX10: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX10: S_ENDPGM 0
+    GLOBAL_STORE_DWORD $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec
+    S_WAITCNT 112
+    S_WAITCNT_VSCNT undef $sgpr_null, 0
+    S_BARRIER
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    S_ENDPGM 0
+...
+
+---
+name:            test_waitcnt_preexisting_vscnt_combined
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1, $vgpr2
+
+    ; GFX10-LABEL: name: test_waitcnt_preexisting_vscnt_combined
+    ; GFX10: S_WAITCNT 0
+    ; GFX10: S_WAITCNT_VSCNT undef $sgpr_null, 0
+    ; GFX10: GLOBAL_STORE_DWORD $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec
+    ; GFX10: S_WAITCNT_VSCNT undef $sgpr_null, 0
+    ; GFX10: S_BARRIER
+    ; GFX10: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX10: S_WAITCNT 112
+    ; GFX10: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX10: S_ENDPGM 0
+    GLOBAL_STORE_DWORD $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec
+    S_WAITCNT_VSCNT undef $sgpr_null, 0
+    S_WAITCNT_VSCNT undef $sgpr_null, 1
+    S_WAITCNT_VSCNT undef $sgpr_null, 2
+    S_BARRIER
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    S_ENDPGM 0
+...
+
+---
+name:            test_waitcnt_preexisting_vscnt_combined_both_types
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1, $vgpr2
+
+    ; GFX10-LABEL: name: test_waitcnt_preexisting_vscnt_combined_both_types
+    ; GFX10: S_WAITCNT 0
+    ; GFX10: S_WAITCNT_VSCNT undef $sgpr_null, 0
+    ; GFX10: GLOBAL_STORE_DWORD $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec
+    ; GFX10: S_WAITCNT 0
+    ; GFX10: S_WAITCNT_VSCNT undef $sgpr_null, 0
+    ; GFX10: S_BARRIER
+    ; GFX10: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX10: S_WAITCNT 112
+    ; GFX10: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX10: S_ENDPGM 0
+    GLOBAL_STORE_DWORD $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec
+    S_WAITCNT 0
+    S_WAITCNT_VSCNT undef $sgpr_null, 1
+    S_WAITCNT 0
+    S_WAITCNT_VSCNT undef $sgpr_null, 2
+    S_WAITCNT 0
+    S_BARRIER
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    S_ENDPGM 0
+...
diff --git a/llvm/test/CodeGen/AMDGPU/waitcnt-preexisting.mir b/llvm/test/CodeGen/AMDGPU/waitcnt-preexisting.mir
--- a/llvm/test/CodeGen/AMDGPU/waitcnt-preexisting.mir
+++ b/llvm/test/CodeGen/AMDGPU/waitcnt-preexisting.mir
@@ -1,37 +1,195 @@
-# RUN: llc -march=amdgcn -mcpu=hawaii -verify-machineinstrs -run-pass si-insert-waitcnts -o - %s | FileCheck -check-prefixes=GCN %s
-
-# GCN-LABEL: name: test{{$}}
-# GCN: S_WAITCNT -16257
-# GCN: DS_READ2_B32
-# GCN: DS_READ2_B32
-# GCN: S_WAITCNT 383{{$}}
-# GCN-NEXT: $vgpr1 = V_OR_B32_e32 1, killed $vgpr1, implicit $exec
-# GCN-NEXT: $vgpr1 = V_MAX_U32_e32 killed $vgpr0, killed $vgpr1, implicit $exec
-# GCN-NEXT: S_WAITCNT 127{{$}}
-# GCN-NEXT: $vgpr1 = V_MAX_U32_e32 killed $vgpr2, killed $vgpr1, implicit $exec
---- |
-  define amdgpu_cs void @test() {
-    ret void
-  }
-...
----
-name:            test
-body:             |
-  bb.0:
-    liveins: $sgpr0, $sgpr1, $vgpr0
-
-    renamable $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 = S_LOAD_DWORDX8_IMM renamable $sgpr0_sgpr1, 480, 0
-    renamable $vgpr13 = V_LSHLREV_B32_e32 2, killed $vgpr0, implicit $exec
-    S_WAITCNT -16257
-    renamable $vgpr0_vgpr1 = DS_READ2_B32 renamable $vgpr13, 0, 1, 0, implicit $m0, implicit $exec
-    renamable $vgpr2_vgpr3 = DS_READ2_B32 renamable $vgpr13, 2, 3, 0, implicit $m0, implicit $exec
-    renamable $vgpr1 = V_OR_B32_e32 1, killed $vgpr1, implicit $exec
-    renamable $vgpr1 = V_MAX_U32_e32 killed $vgpr0, killed $vgpr1, implicit $exec
-    renamable $vgpr1 = V_MAX_U32_e32 killed $vgpr2, killed $vgpr1, implicit $exec
-    renamable $vgpr1 = V_MAX_U32_e32 killed $vgpr3, killed $vgpr1, implicit $exec
-    $vgpr0 = V_MOV_B32_e32 $vgpr1, implicit $exec, implicit $exec
-    $vgpr2 = V_MOV_B32_e32 $vgpr1, implicit $exec, implicit $exec
-    $vgpr3 = V_MOV_B32_e32 $vgpr1, implicit $exec, implicit $exec
-    IMAGE_STORE_V4_V2 killed renamable $vgpr0_vgpr1_vgpr2_vgpr3, killed renamable $vgpr0_vgpr1, killed renamable $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, 15, -1, 1, 0, 0, 0, 0, 0, implicit $exec :: (store 16)
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=gfx908 -verify-machineinstrs -run-pass si-insert-waitcnts -o - %s | FileCheck -check-prefixes=GFX9 %s
+
+---
+name:            test_waitcnt_preexisting_lgkmcnt_unmodified
+body:             |
+  bb.0:
+    liveins: $vgpr0
+
+    ; GFX9-LABEL: name: test_waitcnt_preexisting_lgkmcnt_unmodified
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: $vgpr0_vgpr1 = DS_READ2_B32 $vgpr0, 0, 1, 0, implicit $m0, implicit $exec
+    ; GFX9: S_WAITCNT 49279
+    ; GFX9: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_WAITCNT 112
+    ; GFX9: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_ENDPGM 0
+    $vgpr0_vgpr1 = DS_READ2_B32 $vgpr0, 0, 1, 0, implicit $m0, implicit $exec
+    S_WAITCNT 49279
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    S_ENDPGM 0
+...
+
+---
+name:            test_waitcnt_preexisting_vmcnt_unmodified
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+
+    ; GFX9-LABEL: name: test_waitcnt_preexisting_vmcnt_unmodified
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: $vgpr0_vgpr1 = GLOBAL_LOAD_DWORDX2 $vgpr0_vgpr1, 0, 0, implicit $exec
+    ; GFX9: S_WAITCNT 3952
+    ; GFX9: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_WAITCNT 112
+    ; GFX9: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_ENDPGM 0
+    $vgpr0_vgpr1 = GLOBAL_LOAD_DWORDX2 $vgpr0_vgpr1, 0, 0, implicit $exec
+    S_WAITCNT 3952
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    S_ENDPGM 0
+...
+
+# Respect preexisting waitcnt and add required wait.
+
+---
+name:            test_waitcnt_preexisting_vmcnt_needs_lgkmcnt
+body:             |
+  bb.0:
+    liveins: $vgpr0
+
+    ; GFX9-LABEL: name: test_waitcnt_preexisting_vmcnt_needs_lgkmcnt
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: $vgpr0_vgpr1 = DS_READ2_B32 $vgpr0, 0, 1, 0, implicit $m0, implicit $exec
+    ; GFX9: S_WAITCNT 112
+    ; GFX9: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_WAITCNT 112
+    ; GFX9: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_ENDPGM 0
+    $vgpr0_vgpr1 = DS_READ2_B32 $vgpr0, 0, 1, 0, implicit $m0, implicit $exec
+    S_WAITCNT 3952
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    S_ENDPGM 0
+...
+
+---
+name:            test_waitcnt_preexisting_lgkmcnt_needs_vmcnt
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+
+    ; GFX9-LABEL: name: test_waitcnt_preexisting_lgkmcnt_needs_vmcnt
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: $vgpr0_vgpr1 = GLOBAL_LOAD_DWORDX2 $vgpr0_vgpr1, 0, 0, implicit $exec
+    ; GFX9: S_WAITCNT 112
+    ; GFX9: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_WAITCNT 112
+    ; GFX9: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_ENDPGM 0
+    $vgpr0_vgpr1 = GLOBAL_LOAD_DWORDX2 $vgpr0_vgpr1, 0, 0, implicit $exec
+    S_WAITCNT 49279
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    S_ENDPGM 0
+...
+
+# Apply wait for all counters from preexisting waitcnt regardless of the wait
+# required by the next instruction.
+
+---
+name:            test_waitcnt_preexisting_apply_all_counters
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1, $vgpr2
+
+    ; GFX9-LABEL: name: test_waitcnt_preexisting_apply_all_counters
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: $vgpr4_vgpr5 = GLOBAL_LOAD_DWORDX2 $vgpr0_vgpr1, 0, 0, implicit $exec
+    ; GFX9: $vgpr6_vgpr7 = DS_READ2_B32 $vgpr2, 0, 1, 0, implicit $m0, implicit $exec
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: $vgpr6 = V_OR_B32_e32 1, killed $vgpr6, implicit $exec
+    ; GFX9: $vgpr0 = FLAT_LOAD_DWORD $vgpr4_vgpr5, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_WAITCNT 112
+    ; GFX9: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    $vgpr4_vgpr5 = GLOBAL_LOAD_DWORDX2 $vgpr0_vgpr1, 0, 0, implicit $exec
+    $vgpr6_vgpr7 = DS_READ2_B32 $vgpr2, 0, 1, 0, implicit $m0, implicit $exec
+    S_WAITCNT 0
+    $vgpr6 = V_OR_B32_e32 1, killed $vgpr6, implicit $exec
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr4_vgpr5, 0, 0, implicit $exec, implicit $flat_scr
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+...
+
+---
+name:            test_waitcnt_preexisting_combine_waitcnt
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+
+    ; GFX9-LABEL: name: test_waitcnt_preexisting_combine_waitcnt
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    S_WAITCNT 0
+    S_WAITCNT 0
+    S_WAITCNT 0
+    S_WAITCNT 0
+    S_WAITCNT 0
+    S_WAITCNT 0
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+...
+
+---
+name:            test_waitcnt_preexisting_combine_waitcnt_diff_counters
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+
+    ; GFX9-LABEL: name: test_waitcnt_preexisting_combine_waitcnt_diff_counters
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_WAITCNT 112
+    ; GFX9: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    S_WAITCNT 49279
+    S_WAITCNT 3952
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+...
+
+# Apply preexisting waitcnt when no wait is immediately needed.
+# FIXME: Move waitcnt as late as possible.
+
+---
+name:            test_waitcnt_preexisting_early_wait
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+
+    ; GFX9-LABEL: name: test_waitcnt_preexisting_early_wait
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: S_NOP 0
+    ; GFX9: S_NOP 0
+    ; GFX9: S_NOP 0
+    ; GFX9: FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_ENDPGM 0
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    S_WAITCNT 0
+    S_NOP 0
+    S_NOP 0
+    S_NOP 0
+    FLAT_STORE_DWORD $vgpr0_vgpr1, $vgpr0, 0, 0, implicit $exec, implicit $flat_scr
     S_ENDPGM 0
 ...
+
+---
+name:            test_waitcnt_preexisting_ignore_kill
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+
+    ; GFX9-LABEL: name: test_waitcnt_preexisting_ignore_kill
+    ; GFX9: S_WAITCNT 0
+    ; GFX9: $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    ; GFX9: S_WAITCNT 3952
+    ; GFX9: KILL $vgpr0
+    $vgpr0 = FLAT_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec, implicit $flat_scr
+    S_WAITCNT 3952
+    KILL $vgpr0
+...