[llvm-branch-commits] [llvm] 686988a - [VE] Optimize prologue/epilogue instructions

Kazushi Marukawa via llvm-branch-commits llvm-branch-commits at lists.llvm.org
Mon Nov 30 05:26:50 PST 2020


Author: Kazushi (Jam) Marukawa
Date: 2020-11-30T22:22:33+09:00
New Revision: 686988a50f5009df5a7f184b7debfe012b29bbf8

URL: https://github.com/llvm/llvm-project/commit/686988a50f5009df5a7f184b7debfe012b29bbf8
DIFF: https://github.com/llvm/llvm-project/commit/686988a50f5009df5a7f184b7debfe012b29bbf8.diff

LOG: [VE] Optimize prologue/epilogue instructions

Optimize eliminate FP mechanism.  This time optimize a function which has
no call but fixed stack objects.  LLVM eliminates FP on such functions now.
Also, optimize GOT/PLT registers save/restore instructions if a given
function doesn't uses them.  In addition, remove generating mechanism of
`.cfi` instructions since those are taken from other architectures and not
inspected yet.  Update regression tests, also.

Reviewed By: simoll

Differential Revision: https://reviews.llvm.org/D92251

Added: 
    

Modified: 
    llvm/lib/Target/VE/VEFrameLowering.cpp
    llvm/lib/Target/VE/VEFrameLowering.h
    llvm/test/CodeGen/VE/Scalar/callee.ll
    llvm/test/CodeGen/VE/Scalar/fabs.ll
    llvm/test/CodeGen/VE/Scalar/fcopysign.ll
    llvm/test/CodeGen/VE/Scalar/fp_fneg.ll
    llvm/test/CodeGen/VE/Scalar/function_prologue_epilogue.ll
    llvm/test/CodeGen/VE/Scalar/load-align1.ll
    llvm/test/CodeGen/VE/Scalar/load-align2.ll
    llvm/test/CodeGen/VE/Scalar/load-align4.ll
    llvm/test/CodeGen/VE/Scalar/load-align8.ll
    llvm/test/CodeGen/VE/Scalar/load.ll
    llvm/test/CodeGen/VE/Scalar/loadrri.ll
    llvm/test/CodeGen/VE/Scalar/sext_zext_load.ll
    llvm/test/CodeGen/VE/Scalar/stackframe_align.ll
    llvm/test/CodeGen/VE/Scalar/stackframe_call.ll
    llvm/test/CodeGen/VE/Scalar/stackframe_nocall.ll
    llvm/test/CodeGen/VE/Scalar/stackframe_size.ll
    llvm/test/CodeGen/VE/Scalar/store-align1.ll
    llvm/test/CodeGen/VE/Scalar/store-align2.ll
    llvm/test/CodeGen/VE/Scalar/store-align4.ll
    llvm/test/CodeGen/VE/Scalar/store-align8.ll
    llvm/test/CodeGen/VE/Scalar/store.ll
    llvm/test/CodeGen/VE/Vector/fastcc_callee.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/VE/VEFrameLowering.cpp b/llvm/lib/Target/VE/VEFrameLowering.cpp
index 951430be8f61..7042068f1154 100644
--- a/llvm/lib/Target/VE/VEFrameLowering.cpp
+++ b/llvm/lib/Target/VE/VEFrameLowering.cpp
@@ -137,45 +137,47 @@ void VEFrameLowering::emitPrologueInsns(MachineFunction &MF,
                                         MachineBasicBlock::iterator MBBI,
                                         uint64_t NumBytes,
                                         bool RequireFPUpdate) const {
+  const VEMachineFunctionInfo *FuncInfo = MF.getInfo<VEMachineFunctionInfo>();
   DebugLoc DL;
-  const VEInstrInfo &TII =
-      *static_cast<const VEInstrInfo *>(MF.getSubtarget().getInstrInfo());
+  const VEInstrInfo &TII = *STI.getInstrInfo();
 
   // Insert following codes here as prologue
   //
-  //    st %fp, 0(,%sp)
-  //    st %lr, 8(,%sp)
-  //    st %got, 24(,%sp)
-  //    st %plt, 32(,%sp)
-  //    st %s17, 40(,%sp) iff this function is using s17 as BP
-  //    or %fp, 0, %sp
-  BuildMI(MBB, MBBI, DL, TII.get(VE::STrii))
-      .addReg(VE::SX11)
-      .addImm(0)
-      .addImm(0)
-      .addReg(VE::SX9);
-  BuildMI(MBB, MBBI, DL, TII.get(VE::STrii))
-      .addReg(VE::SX11)
-      .addImm(0)
-      .addImm(8)
-      .addReg(VE::SX10);
-  BuildMI(MBB, MBBI, DL, TII.get(VE::STrii))
-      .addReg(VE::SX11)
-      .addImm(0)
-      .addImm(24)
-      .addReg(VE::SX15);
-  BuildMI(MBB, MBBI, DL, TII.get(VE::STrii))
-      .addReg(VE::SX11)
-      .addImm(0)
-      .addImm(32)
-      .addReg(VE::SX16);
+  //    st %fp, 0(, %sp)   iff !isLeafProc
+  //    st %lr, 8(, %sp)   iff !isLeafProc
+  //    st %got, 24(, %sp) iff hasGOT
+  //    st %plt, 32(, %sp) iff hasGOT
+  //    st %s17, 40(, %sp) iff hasBP
+  if (!FuncInfo->isLeafProc()) {
+    BuildMI(MBB, MBBI, DL, TII.get(VE::STrii))
+        .addReg(VE::SX11)
+        .addImm(0)
+        .addImm(0)
+        .addReg(VE::SX9);
+    BuildMI(MBB, MBBI, DL, TII.get(VE::STrii))
+        .addReg(VE::SX11)
+        .addImm(0)
+        .addImm(8)
+        .addReg(VE::SX10);
+  }
+  if (hasGOT(MF)) {
+    BuildMI(MBB, MBBI, DL, TII.get(VE::STrii))
+        .addReg(VE::SX11)
+        .addImm(0)
+        .addImm(24)
+        .addReg(VE::SX15);
+    BuildMI(MBB, MBBI, DL, TII.get(VE::STrii))
+        .addReg(VE::SX11)
+        .addImm(0)
+        .addImm(32)
+        .addReg(VE::SX16);
+  }
   if (hasBP(MF))
     BuildMI(MBB, MBBI, DL, TII.get(VE::STrii))
         .addReg(VE::SX11)
         .addImm(0)
         .addImm(40)
         .addReg(VE::SX17);
-  BuildMI(MBB, MBBI, DL, TII.get(VE::ORri), VE::SX9).addReg(VE::SX11).addImm(0);
 }
 
 void VEFrameLowering::emitEpilogueInsns(MachineFunction &MF,
@@ -183,40 +185,42 @@ void VEFrameLowering::emitEpilogueInsns(MachineFunction &MF,
                                         MachineBasicBlock::iterator MBBI,
                                         uint64_t NumBytes,
                                         bool RequireFPUpdate) const {
+  const VEMachineFunctionInfo *FuncInfo = MF.getInfo<VEMachineFunctionInfo>();
   DebugLoc DL;
-  const VEInstrInfo &TII =
-      *static_cast<const VEInstrInfo *>(MF.getSubtarget().getInstrInfo());
+  const VEInstrInfo &TII = *STI.getInstrInfo();
 
   // Insert following codes here as epilogue
   //
-  //    or %sp, 0, %fp
-  //    ld %s17, 40(,%sp) iff this function is using s17 as BP
-  //    ld %plt, 32(,%sp)
-  //    ld %got, 24(,%sp)
-  //    ld %lr, 8(,%sp)
-  //    ld %fp, 0(,%sp)
-  BuildMI(MBB, MBBI, DL, TII.get(VE::ORri), VE::SX11).addReg(VE::SX9).addImm(0);
+  //    ld %s17, 40(, %sp) iff hasBP
+  //    ld %plt, 32(, %sp) iff hasGOT
+  //    ld %got, 24(, %sp) iff hasGOT
+  //    ld %lr, 8(, %sp)   iff !isLeafProc
+  //    ld %fp, 0(, %sp)   iff !isLeafProc
   if (hasBP(MF))
     BuildMI(MBB, MBBI, DL, TII.get(VE::LDrii), VE::SX17)
         .addReg(VE::SX11)
         .addImm(0)
         .addImm(40);
-  BuildMI(MBB, MBBI, DL, TII.get(VE::LDrii), VE::SX16)
-      .addReg(VE::SX11)
-      .addImm(0)
-      .addImm(32);
-  BuildMI(MBB, MBBI, DL, TII.get(VE::LDrii), VE::SX15)
-      .addReg(VE::SX11)
-      .addImm(0)
-      .addImm(24);
-  BuildMI(MBB, MBBI, DL, TII.get(VE::LDrii), VE::SX10)
-      .addReg(VE::SX11)
-      .addImm(0)
-      .addImm(8);
-  BuildMI(MBB, MBBI, DL, TII.get(VE::LDrii), VE::SX9)
-      .addReg(VE::SX11)
-      .addImm(0)
-      .addImm(0);
+  if (hasGOT(MF)) {
+    BuildMI(MBB, MBBI, DL, TII.get(VE::LDrii), VE::SX16)
+        .addReg(VE::SX11)
+        .addImm(0)
+        .addImm(32);
+    BuildMI(MBB, MBBI, DL, TII.get(VE::LDrii), VE::SX15)
+        .addReg(VE::SX11)
+        .addImm(0)
+        .addImm(24);
+  }
+  if (!FuncInfo->isLeafProc()) {
+    BuildMI(MBB, MBBI, DL, TII.get(VE::LDrii), VE::SX10)
+        .addReg(VE::SX11)
+        .addImm(0)
+        .addImm(8);
+    BuildMI(MBB, MBBI, DL, TII.get(VE::LDrii), VE::SX9)
+        .addReg(VE::SX11)
+        .addImm(0)
+        .addImm(0);
+  }
 }
 
 void VEFrameLowering::emitSPAdjustment(MachineFunction &MF,
@@ -270,8 +274,7 @@ void VEFrameLowering::emitSPAdjustment(MachineFunction &MF,
 void VEFrameLowering::emitSPExtend(MachineFunction &MF, MachineBasicBlock &MBB,
                                    MachineBasicBlock::iterator MBBI) const {
   DebugLoc DL;
-  const VEInstrInfo &TII =
-      *static_cast<const VEInstrInfo *>(MF.getSubtarget().getInstrInfo());
+  const VEInstrInfo &TII = *STI.getInstrInfo();
 
   // Emit following codes.  It is not possible to insert multiple
   // BasicBlocks in PEI pass, so we emit two pseudo instructions here.
@@ -327,12 +330,17 @@ void VEFrameLowering::emitPrologue(MachineFunction &MF,
                        "stack re-alignment, but LLVM couldn't handle it "
                        "(probably because it has a dynamic alloca).");
 
-  // Get the number of bytes to allocate from the FrameInfo
+  // Get the number of bytes to allocate from the FrameInfo.
+  // This number of bytes is already aligned to ABI stack alignment.
   uint64_t NumBytes = MFI.getStackSize();
 
-  // The VE ABI requires a reserved area at the top of stack as described
-  // in VESubtarget.cpp.  So, we adjust it here.
-  NumBytes = STI.getAdjustedFrameSize(NumBytes);
+  // Adjust stack size if this function is not a leaf function since the
+  // VE ABI requires a reserved area at the top of stack as described in
+  // VEFrameLowering.cpp.
+  if (!FuncInfo->isLeafProc()) {
+    // NOTE: The number is aligned to ABI stack alignment after adjustment.
+    NumBytes = STI.getAdjustedFrameSize(NumBytes);
+  }
 
   // Finally, ensure that the size is sufficiently aligned for the
   // data on the stack.
@@ -341,15 +349,22 @@ void VEFrameLowering::emitPrologue(MachineFunction &MF,
   // Update stack size with corrected value.
   MFI.setStackSize(NumBytes);
 
-  if (FuncInfo->isLeafProc())
-    return;
-
   // Emit Prologue instructions to save multiple registers.
   emitPrologueInsns(MF, MBB, MBBI, NumBytes, true);
 
+  // Emit instructions to save SP in FP as follows if this is not a leaf
+  // function:
+  //    or %fp, 0, %sp
+  if (!FuncInfo->isLeafProc())
+    BuildMI(MBB, MBBI, DL, TII.get(VE::ORri), VE::SX9)
+        .addReg(VE::SX11)
+        .addImm(0);
+
   // Emit stack adjust instructions
   MaybeAlign RuntimeAlign =
       NeedsStackRealignment ? MaybeAlign(MFI.getMaxAlign()) : None;
+  assert((RuntimeAlign == None || !FuncInfo->isLeafProc()) &&
+         "SP has to be saved in order to align variable sized stack object!");
   emitSPAdjustment(MF, MBB, MBBI, -(int64_t)NumBytes, RuntimeAlign);
 
   if (hasBP(MF)) {
@@ -360,20 +375,8 @@ void VEFrameLowering::emitPrologue(MachineFunction &MF,
   }
 
   // Emit stack extend instructions
-  emitSPExtend(MF, MBB, MBBI);
-
-  Register RegFP = RegInfo.getDwarfRegNum(VE::SX9, true);
-
-  // Emit ".cfi_def_cfa_register 30".
-  unsigned CFIIndex =
-      MF.addFrameInst(MCCFIInstruction::createDefCfaRegister(nullptr, RegFP));
-  BuildMI(MBB, MBBI, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
-      .addCFIIndex(CFIIndex);
-
-  // Emit ".cfi_window_save".
-  CFIIndex = MF.addFrameInst(MCCFIInstruction::createWindowSave(nullptr));
-  BuildMI(MBB, MBBI, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
-      .addCFIIndex(CFIIndex);
+  if (NumBytes != 0)
+    emitSPExtend(MF, MBB, MBBI);
 }
 
 MachineBasicBlock::iterator VEFrameLowering::eliminateCallFramePseudoInstr(
@@ -394,13 +397,24 @@ MachineBasicBlock::iterator VEFrameLowering::eliminateCallFramePseudoInstr(
 void VEFrameLowering::emitEpilogue(MachineFunction &MF,
                                    MachineBasicBlock &MBB) const {
   const VEMachineFunctionInfo *FuncInfo = MF.getInfo<VEMachineFunctionInfo>();
+  DebugLoc DL;
   MachineBasicBlock::iterator MBBI = MBB.getLastNonDebugInstr();
   MachineFrameInfo &MFI = MF.getFrameInfo();
+  const VEInstrInfo &TII = *STI.getInstrInfo();
 
   uint64_t NumBytes = MFI.getStackSize();
 
-  if (FuncInfo->isLeafProc())
-    return;
+  // Emit instructions to retrieve original SP.
+  if (!FuncInfo->isLeafProc()) {
+    // If SP is saved in FP, retrieve it as follows:
+    //    or %sp, 0, %fp     iff !isLeafProc
+    BuildMI(MBB, MBBI, DL, TII.get(VE::ORri), VE::SX11)
+        .addReg(VE::SX9)
+        .addImm(0);
+  } else {
+    // Emit stack adjust instructions.
+    emitSPAdjustment(MF, MBB, MBBI, NumBytes, None);
+  }
 
   // Emit Epilogue instructions to restore multiple registers.
   emitEpilogueInsns(MF, MBB, MBBI, NumBytes, true);
@@ -425,6 +439,13 @@ bool VEFrameLowering::hasBP(const MachineFunction &MF) const {
   return MFI.hasVarSizedObjects() && TRI->needsStackRealignment(MF);
 }
 
+bool VEFrameLowering::hasGOT(const MachineFunction &MF) const {
+  const VEMachineFunctionInfo *FuncInfo = MF.getInfo<VEMachineFunctionInfo>();
+
+  // If a global base register is assigned (!= 0), GOT is used.
+  return FuncInfo->getGlobalBaseReg() != 0;
+}
+
 StackOffset VEFrameLowering::getFrameIndexReference(const MachineFunction &MF,
                                                     int FI,
                                                     Register &FrameReg) const {
@@ -471,11 +492,10 @@ void VEFrameLowering::determineCalleeSaves(MachineFunction &MF,
                                            BitVector &SavedRegs,
                                            RegScavenger *RS) const {
   TargetFrameLowering::determineCalleeSaves(MF, SavedRegs, RS);
-  const MachineFrameInfo &MFI = MF.getFrameInfo();
 
   // Functions having BP or stack objects need to emit prologue and epilogue
   // to allocate local buffer on the stack.
-  if (isLeafProc(MF) && !hasBP(MF) && !MFI.hasStackObjects()) {
+  if (isLeafProc(MF) && !hasBP(MF)) {
     VEMachineFunctionInfo *FuncInfo = MF.getInfo<VEMachineFunctionInfo>();
     FuncInfo->setLeafProc(true);
   }

diff  --git a/llvm/lib/Target/VE/VEFrameLowering.h b/llvm/lib/Target/VE/VEFrameLowering.h
index d3497d8b8096..99eb41189b25 100644
--- a/llvm/lib/Target/VE/VEFrameLowering.h
+++ b/llvm/lib/Target/VE/VEFrameLowering.h
@@ -41,6 +41,7 @@ class VEFrameLowering : public TargetFrameLowering {
 
   bool hasFP(const MachineFunction &MF) const override;
   bool hasBP(const MachineFunction &MF) const;
+  bool hasGOT(const MachineFunction &MF) const;
 
   // VE reserves argument space always for call sites in the function
   // immediately on entry of the current function.

diff  --git a/llvm/test/CodeGen/VE/Scalar/callee.ll b/llvm/test/CodeGen/VE/Scalar/callee.ll
index b1dee4806ed0..2f5733c0e537 100644
--- a/llvm/test/CodeGen/VE/Scalar/callee.ll
+++ b/llvm/test/CodeGen/VE/Scalar/callee.ll
@@ -2,40 +2,40 @@
 
 define i32 @stack_stack_arg_i32_r9(i1 %0, i8 %1, i16 %2, i32 %3, i64 %4, i32 %5, i32 %6, i32 %7, i32 %8, i32 %9) {
 ; CHECK-LABEL: stack_stack_arg_i32_r9:
-; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 424(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ldl.sx %s0, 248(, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
   ret i32 %9
 }
 
 define i64 @stack_stack_arg_i64_r9(i1 %0, i8 %1, i16 %2, i32 %3, i64 %4, i64 %5, i64 %6, i64 %7, i64 %8, i64 %9) {
 ; CHECK-LABEL: stack_stack_arg_i64_r9:
-; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 424(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ld %s0, 248(, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
   ret i64 %9
 }
 
 define float @stack_stack_arg_f32_r9(float %p0, float %p1, float %p2, float %p3, float %p4, float %p5, float %p6, float %p7, float %s0, float %s1) {
 ; CHECK-LABEL: stack_stack_arg_f32_r9:
-; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldu %s0, 428(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ldu %s0, 252(, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
   ret float %s1
 }
 
 define i32 @stack_stack_arg_i32f32_r8(i32 %p0, float %p1, i32 %p2, float %p3, i32 %p4, float %p5, i32 %p6, float %p7, i32 %s0, float %s1) {
 ; CHECK-LABEL: stack_stack_arg_i32f32_r8:
-; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 416(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ldl.sx %s0, 240(, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
   ret i32 %s0
 }
 
 define float @stack_stack_arg_i32f32_r9(i32 %p0, float %p1, i32 %p2, float %p3, i32 %p4, float %p5, i32 %p6, float %p7, i32 %s0, float %s1) {
 ; CHECK-LABEL: stack_stack_arg_i32f32_r9:
-; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldu %s0, 428(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ldu %s0, 252(, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
   ret float %s1
 }

diff  --git a/llvm/test/CodeGen/VE/Scalar/fabs.ll b/llvm/test/CodeGen/VE/Scalar/fabs.ll
index 1943b0a277ce..a68e561d0098 100644
--- a/llvm/test/CodeGen/VE/Scalar/fabs.ll
+++ b/llvm/test/CodeGen/VE/Scalar/fabs.ll
@@ -59,14 +59,15 @@ declare double @llvm.fabs.f64(double)
 define fp128 @fabs_quad_var(fp128 %0) {
 ; CHECK-LABEL: fabs_quad_var:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s1, 176(, %s11)
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
+; CHECK-NEXT:    st %s1, (, %s11)
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
 ; CHECK-NEXT:    and %s0, %s0, (57)0
-; CHECK-NEXT:    st1b %s0, 191(, %s11)
-; CHECK-NEXT:    ld %s1, 176(, %s11)
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st1b %s0, 15(, %s11)
+; CHECK-NEXT:    ld %s1, (, %s11)
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %2 = tail call fast fp128 @llvm.fabs.f128(fp128 %0)
   ret fp128 %2
 }

diff  --git a/llvm/test/CodeGen/VE/Scalar/fcopysign.ll b/llvm/test/CodeGen/VE/Scalar/fcopysign.ll
index 6ea7f5fdfe00..552f6ed59f59 100644
--- a/llvm/test/CodeGen/VE/Scalar/fcopysign.ll
+++ b/llvm/test/CodeGen/VE/Scalar/fcopysign.ll
@@ -67,20 +67,21 @@ declare double @llvm.copysign.f64(double, double)
 define fp128 @copysign_quad_var(fp128 %0, fp128 %1) {
 ; CHECK-LABEL: copysign_quad_var:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s3, 192(, %s11)
-; CHECK-NEXT:    st %s2, 200(, %s11)
-; CHECK-NEXT:    st %s1, 176(, %s11)
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    ld1b.zx %s0, 207(, %s11)
-; CHECK-NEXT:    ld1b.zx %s1, 191(, %s11)
+; CHECK-NEXT:    st %s3, 16(, %s11)
+; CHECK-NEXT:    st %s2, 24(, %s11)
+; CHECK-NEXT:    st %s1, (, %s11)
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    ld1b.zx %s0, 31(, %s11)
+; CHECK-NEXT:    ld1b.zx %s1, 15(, %s11)
 ; CHECK-NEXT:    lea %s2, 128
 ; CHECK-NEXT:    and %s0, %s0, %s2
 ; CHECK-NEXT:    and %s1, %s1, (57)0
 ; CHECK-NEXT:    or %s0, %s1, %s0
-; CHECK-NEXT:    st1b %s0, 191(, %s11)
-; CHECK-NEXT:    ld %s1, 176(, %s11)
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st1b %s0, 15(, %s11)
+; CHECK-NEXT:    ld %s1, (, %s11)
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 32, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %3 = tail call fast fp128 @llvm.copysign.f128(fp128 %0, fp128 %1)
   ret fp128 %3
 }
@@ -121,20 +122,21 @@ define fp128 @copysign_quad_zero(fp128 %0) {
 ; CHECK-NEXT:    lea.sl %s2, .LCPI{{[0-9]+}}_0 at hi(, %s2)
 ; CHECK-NEXT:    ld %s4, 8(, %s2)
 ; CHECK-NEXT:    ld %s5, (, %s2)
-; CHECK-NEXT:    st %s1, 192(, %s11)
-; CHECK-NEXT:    st %s0, 200(, %s11)
-; CHECK-NEXT:    st %s5, 176(, %s11)
-; CHECK-NEXT:    st %s4, 184(, %s11)
-; CHECK-NEXT:    ld1b.zx %s0, 207(, %s11)
-; CHECK-NEXT:    ld1b.zx %s1, 191(, %s11)
+; CHECK-NEXT:    st %s1, 16(, %s11)
+; CHECK-NEXT:    st %s0, 24(, %s11)
+; CHECK-NEXT:    st %s5, (, %s11)
+; CHECK-NEXT:    st %s4, 8(, %s11)
+; CHECK-NEXT:    ld1b.zx %s0, 31(, %s11)
+; CHECK-NEXT:    ld1b.zx %s1, 15(, %s11)
 ; CHECK-NEXT:    lea %s2, 128
 ; CHECK-NEXT:    and %s0, %s0, %s2
 ; CHECK-NEXT:    and %s1, %s1, (57)0
 ; CHECK-NEXT:    or %s0, %s1, %s0
-; CHECK-NEXT:    st1b %s0, 191(, %s11)
-; CHECK-NEXT:    ld %s1, 176(, %s11)
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st1b %s0, 15(, %s11)
+; CHECK-NEXT:    ld %s1, (, %s11)
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 32, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %2 = tail call fast fp128 @llvm.copysign.f128(fp128 0xL00000000000000000000000000000000, fp128 %0)
   ret fp128 %2
 }
@@ -176,20 +178,21 @@ define fp128 @copysign_quad_const(fp128 %0) {
 ; CHECK-NEXT:    lea.sl %s2, .LCPI{{[0-9]+}}_0 at hi(, %s2)
 ; CHECK-NEXT:    ld %s4, 8(, %s2)
 ; CHECK-NEXT:    ld %s5, (, %s2)
-; CHECK-NEXT:    st %s1, 192(, %s11)
-; CHECK-NEXT:    st %s0, 200(, %s11)
-; CHECK-NEXT:    st %s5, 176(, %s11)
-; CHECK-NEXT:    st %s4, 184(, %s11)
-; CHECK-NEXT:    ld1b.zx %s0, 207(, %s11)
-; CHECK-NEXT:    ld1b.zx %s1, 191(, %s11)
+; CHECK-NEXT:    st %s1, 16(, %s11)
+; CHECK-NEXT:    st %s0, 24(, %s11)
+; CHECK-NEXT:    st %s5, (, %s11)
+; CHECK-NEXT:    st %s4, 8(, %s11)
+; CHECK-NEXT:    ld1b.zx %s0, 31(, %s11)
+; CHECK-NEXT:    ld1b.zx %s1, 15(, %s11)
 ; CHECK-NEXT:    lea %s2, 128
 ; CHECK-NEXT:    and %s0, %s0, %s2
 ; CHECK-NEXT:    and %s1, %s1, (57)0
 ; CHECK-NEXT:    or %s0, %s1, %s0
-; CHECK-NEXT:    st1b %s0, 191(, %s11)
-; CHECK-NEXT:    ld %s1, 176(, %s11)
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st1b %s0, 15(, %s11)
+; CHECK-NEXT:    ld %s1, (, %s11)
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 32, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %2 = tail call fast fp128 @llvm.copysign.f128(fp128 0xL0000000000000000C000000000000000, fp128 %0)
   ret fp128 %2
 }

diff  --git a/llvm/test/CodeGen/VE/Scalar/fp_fneg.ll b/llvm/test/CodeGen/VE/Scalar/fp_fneg.ll
index 30c307ad5039..2f4e6d914fa2 100644
--- a/llvm/test/CodeGen/VE/Scalar/fp_fneg.ll
+++ b/llvm/test/CodeGen/VE/Scalar/fp_fneg.ll
@@ -53,15 +53,16 @@ define double @fneg_double(double %0) {
 define fp128 @fneg_quad(fp128 %0) {
 ; CHECK-LABEL: fneg_quad:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s1, 176(, %s11)
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
+; CHECK-NEXT:    st %s1, (, %s11)
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
 ; CHECK-NEXT:    lea %s1, 128
 ; CHECK-NEXT:    xor %s0, %s0, %s1
-; CHECK-NEXT:    st1b %s0, 191(, %s11)
-; CHECK-NEXT:    ld %s1, 176(, %s11)
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st1b %s0, 15(, %s11)
+; CHECK-NEXT:    ld %s1, (, %s11)
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %2 = fneg fp128 %0
   ret fp128 %2
 }

diff  --git a/llvm/test/CodeGen/VE/Scalar/function_prologue_epilogue.ll b/llvm/test/CodeGen/VE/Scalar/function_prologue_epilogue.ll
index bc3f9d633e95..19c3dd2ce42c 100644
--- a/llvm/test/CodeGen/VE/Scalar/function_prologue_epilogue.ll
+++ b/llvm/test/CodeGen/VE/Scalar/function_prologue_epilogue.ll
@@ -68,12 +68,7 @@ define i8 @func_gv() {
 define i32 @func_alloca(i32 signext %0) {
 ; CHECK-LABEL: func_alloca:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -192(, %s11)
+; CHECK-NEXT:    adds.l %s11, -16, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB2_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -85,22 +80,13 @@ define i32 @func_alloca(i32 signext %0) {
 ; CHECK-NEXT:    monc
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB2_2:
-; CHECK-NEXT:    stl %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    stl %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
 ; PIC-LABEL: func_alloca:
 ; PIC:       # %bb.0:
-; PIC-NEXT:    st %s9, (, %s11)
-; PIC-NEXT:    st %s10, 8(, %s11)
-; PIC-NEXT:    st %s15, 24(, %s11)
-; PIC-NEXT:    st %s16, 32(, %s11)
-; PIC-NEXT:    or %s9, 0, %s11
-; PIC-NEXT:    lea %s11, -192(, %s11)
+; PIC-NEXT:    adds.l %s11, -16, %s11
 ; PIC-NEXT:    brge.l.t %s11, %s8, .LBB2_2
 ; PIC-NEXT:  # %bb.1:
 ; PIC-NEXT:    ld %s61, 24(, %s14)
@@ -112,12 +98,8 @@ define i32 @func_alloca(i32 signext %0) {
 ; PIC-NEXT:    monc
 ; PIC-NEXT:    or %s0, 0, %s62
 ; PIC-NEXT:  .LBB2_2:
-; PIC-NEXT:    stl %s0, 188(, %s11)
-; PIC-NEXT:    or %s11, 0, %s9
-; PIC-NEXT:    ld %s16, 32(, %s11)
-; PIC-NEXT:    ld %s15, 24(, %s11)
-; PIC-NEXT:    ld %s10, 8(, %s11)
-; PIC-NEXT:    ld %s9, (, %s11)
+; PIC-NEXT:    stl %s0, 12(, %s11)
+; PIC-NEXT:    adds.l %s11, 16, %s11
 ; PIC-NEXT:    b.l.t (, %s10)
   %2 = alloca i32, align 4
   store i32 %0, i32* %2, align 4

diff  --git a/llvm/test/CodeGen/VE/Scalar/load-align1.ll b/llvm/test/CodeGen/VE/Scalar/load-align1.ll
index 29c5687100ca..8c7a8d2cf497 100644
--- a/llvm/test/CodeGen/VE/Scalar/load-align1.ll
+++ b/llvm/test/CodeGen/VE/Scalar/load-align1.ll
@@ -11,8 +11,9 @@
 define double @loadf64stk() {
 ; CHECK-LABEL: loadf64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca double, align 1
   %1 = load double, double* %addr, align 1
   ret double %1
@@ -22,8 +23,9 @@ define double @loadf64stk() {
 define float @loadf32stk() {
 ; CHECK-LABEL: loadf32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldu %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldu %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca float, align 1
   %1 = load float, float* %addr, align 1
   ret float %1
@@ -33,8 +35,9 @@ define float @loadf32stk() {
 define i64 @loadi64stk() {
 ; CHECK-LABEL: loadi64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i64, align 1
   %1 = load i64, i64* %addr, align 1
   ret i64 %1
@@ -44,8 +47,9 @@ define i64 @loadi64stk() {
 define i32 @loadi32stk() {
 ; CHECK-LABEL: loadi32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldl.sx %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i32, align 1
   %1 = load i32, i32* %addr, align 1
   ret i32 %1
@@ -55,8 +59,9 @@ define i32 @loadi32stk() {
 define i16 @loadi16stk() {
 ; CHECK-LABEL: loadi16stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.zx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.zx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i16, align 1
   %1 = load i16, i16* %addr, align 1
   ret i16 %1
@@ -66,8 +71,9 @@ define i16 @loadi16stk() {
 define i8 @loadi8stk() {
 ; CHECK-LABEL: loadi8stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i8, align 1
   %1 = load i8, i8* %addr, align 1
   ret i8 %1

diff  --git a/llvm/test/CodeGen/VE/Scalar/load-align2.ll b/llvm/test/CodeGen/VE/Scalar/load-align2.ll
index 791485a23652..073664611651 100644
--- a/llvm/test/CodeGen/VE/Scalar/load-align2.ll
+++ b/llvm/test/CodeGen/VE/Scalar/load-align2.ll
@@ -11,8 +11,9 @@
 define double @loadf64stk() {
 ; CHECK-LABEL: loadf64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca double, align 2
   %1 = load double, double* %addr, align 2
   ret double %1
@@ -22,8 +23,9 @@ define double @loadf64stk() {
 define float @loadf32stk() {
 ; CHECK-LABEL: loadf32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldu %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldu %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca float, align 2
   %1 = load float, float* %addr, align 2
   ret float %1
@@ -33,8 +35,9 @@ define float @loadf32stk() {
 define i64 @loadi64stk() {
 ; CHECK-LABEL: loadi64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i64, align 2
   %1 = load i64, i64* %addr, align 2
   ret i64 %1
@@ -44,8 +47,9 @@ define i64 @loadi64stk() {
 define i32 @loadi32stk() {
 ; CHECK-LABEL: loadi32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldl.sx %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i32, align 2
   %1 = load i32, i32* %addr, align 2
   ret i32 %1
@@ -55,8 +59,9 @@ define i32 @loadi32stk() {
 define i16 @loadi16stk() {
 ; CHECK-LABEL: loadi16stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.zx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.zx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i16, align 2
   %1 = load i16, i16* %addr, align 2
   ret i16 %1
@@ -66,8 +71,9 @@ define i16 @loadi16stk() {
 define i8 @loadi8stk() {
 ; CHECK-LABEL: loadi8stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i8, align 2
   %1 = load i8, i8* %addr, align 2
   ret i8 %1

diff  --git a/llvm/test/CodeGen/VE/Scalar/load-align4.ll b/llvm/test/CodeGen/VE/Scalar/load-align4.ll
index 4c04ccdda433..7ec245ddff57 100644
--- a/llvm/test/CodeGen/VE/Scalar/load-align4.ll
+++ b/llvm/test/CodeGen/VE/Scalar/load-align4.ll
@@ -11,8 +11,9 @@
 define double @loadf64stk() {
 ; CHECK-LABEL: loadf64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca double, align 4
   %1 = load double, double* %addr, align 4
   ret double %1
@@ -22,8 +23,9 @@ define double @loadf64stk() {
 define float @loadf32stk() {
 ; CHECK-LABEL: loadf32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldu %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldu %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca float, align 4
   %1 = load float, float* %addr, align 4
   ret float %1
@@ -33,8 +35,9 @@ define float @loadf32stk() {
 define i64 @loadi64stk() {
 ; CHECK-LABEL: loadi64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i64, align 4
   %1 = load i64, i64* %addr, align 4
   ret i64 %1
@@ -44,8 +47,9 @@ define i64 @loadi64stk() {
 define i32 @loadi32stk() {
 ; CHECK-LABEL: loadi32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldl.sx %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i32, align 4
   %1 = load i32, i32* %addr, align 4
   ret i32 %1
@@ -55,8 +59,9 @@ define i32 @loadi32stk() {
 define i16 @loadi16stk() {
 ; CHECK-LABEL: loadi16stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.zx %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.zx %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i16, align 4
   %1 = load i16, i16* %addr, align 4
   ret i16 %1
@@ -66,8 +71,9 @@ define i16 @loadi16stk() {
 define i8 @loadi8stk() {
 ; CHECK-LABEL: loadi8stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i8, align 4
   %1 = load i8, i8* %addr, align 4
   ret i8 %1

diff  --git a/llvm/test/CodeGen/VE/Scalar/load-align8.ll b/llvm/test/CodeGen/VE/Scalar/load-align8.ll
index c7dafbcbc276..b1457a560ba5 100644
--- a/llvm/test/CodeGen/VE/Scalar/load-align8.ll
+++ b/llvm/test/CodeGen/VE/Scalar/load-align8.ll
@@ -11,8 +11,9 @@
 define double @loadf64stk() {
 ; CHECK-LABEL: loadf64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca double, align 8
   %1 = load double, double* %addr, align 8
   ret double %1
@@ -22,8 +23,9 @@ define double @loadf64stk() {
 define float @loadf32stk() {
 ; CHECK-LABEL: loadf32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldu %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldu %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca float, align 8
   %1 = load float, float* %addr, align 8
   ret float %1
@@ -33,8 +35,9 @@ define float @loadf32stk() {
 define i64 @loadi64stk() {
 ; CHECK-LABEL: loadi64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i64, align 8
   %1 = load i64, i64* %addr, align 8
   ret i64 %1
@@ -44,8 +47,9 @@ define i64 @loadi64stk() {
 define i32 @loadi32stk() {
 ; CHECK-LABEL: loadi32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldl.sx %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i32, align 8
   %1 = load i32, i32* %addr, align 8
   ret i32 %1
@@ -55,8 +59,9 @@ define i32 @loadi32stk() {
 define i16 @loadi16stk() {
 ; CHECK-LABEL: loadi16stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.zx %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.zx %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i16, align 8
   %1 = load i16, i16* %addr, align 8
   ret i16 %1
@@ -66,8 +71,9 @@ define i16 @loadi16stk() {
 define i8 @loadi8stk() {
 ; CHECK-LABEL: loadi8stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i8, align 8
   %1 = load i8, i8* %addr, align 8
   ret i8 %1

diff  --git a/llvm/test/CodeGen/VE/Scalar/load.ll b/llvm/test/CodeGen/VE/Scalar/load.ll
index 2efc5e8cc641..fb4ab95fa0be 100644
--- a/llvm/test/CodeGen/VE/Scalar/load.ll
+++ b/llvm/test/CodeGen/VE/Scalar/load.ll
@@ -155,9 +155,10 @@ define i64 @loadi8zext(i8* nocapture readonly %0) {
 define fp128 @loadf128stk() {
 ; CHECK-LABEL: loadf128stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s1, 176(, %s11)
-; CHECK-NEXT:    ld %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s1, (, %s11)
+; CHECK-NEXT:    ld %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca fp128, align 16
   %1 = load fp128, fp128* %addr, align 16
   ret fp128 %1
@@ -167,8 +168,9 @@ define fp128 @loadf128stk() {
 define double @loadf64stk() {
 ; CHECK-LABEL: loadf64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca double, align 16
   %1 = load double, double* %addr, align 16
   ret double %1
@@ -178,8 +180,9 @@ define double @loadf64stk() {
 define float @loadf32stk() {
 ; CHECK-LABEL: loadf32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldu %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldu %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca float, align 16
   %1 = load float, float* %addr, align 16
   ret float %1
@@ -189,9 +192,10 @@ define float @loadf32stk() {
 define i128 @loadi128stk() {
 ; CHECK-LABEL: loadi128stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 176(, %s11)
-; CHECK-NEXT:    ld %s1, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, (, %s11)
+; CHECK-NEXT:    ld %s1, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i128, align 16
   %1 = load i128, i128* %addr, align 16
   ret i128 %1
@@ -201,8 +205,9 @@ define i128 @loadi128stk() {
 define i64 @loadi64stk() {
 ; CHECK-LABEL: loadi64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i64, align 16
   %1 = load i64, i64* %addr, align 16
   ret i64 %1
@@ -212,8 +217,9 @@ define i64 @loadi64stk() {
 define i32 @loadi32stk() {
 ; CHECK-LABEL: loadi32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldl.sx %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i32, align 16
   %1 = load i32, i32* %addr, align 16
   ret i32 %1
@@ -223,8 +229,9 @@ define i32 @loadi32stk() {
 define i16 @loadi16stk() {
 ; CHECK-LABEL: loadi16stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.zx %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.zx %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i16, align 16
   %1 = load i16, i16* %addr, align 16
   ret i16 %1
@@ -234,8 +241,9 @@ define i16 @loadi16stk() {
 define i8 @loadi8stk() {
 ; CHECK-LABEL: loadi8stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i8, align 16
   %1 = load i8, i8* %addr, align 16
   ret i8 %1

diff  --git a/llvm/test/CodeGen/VE/Scalar/loadrri.ll b/llvm/test/CodeGen/VE/Scalar/loadrri.ll
index 76df63d484f3..18e4f837c6de 100644
--- a/llvm/test/CodeGen/VE/Scalar/loadrri.ll
+++ b/llvm/test/CodeGen/VE/Scalar/loadrri.ll
@@ -28,9 +28,10 @@ define signext i8 @func_fr(%struct.data* readonly %0, i32 signext %1) {
 ; CHECK:       .LBB{{[0-9]+}}_2:
 ; CHECK-NEXT:    sll %s1, %s1, 2
 ; CHECK-NEXT:    ldl.sx %s0, (%s1, %s0)
-; CHECK-NEXT:    stl %s0, 184(%s1, %s11)
-; CHECK-NEXT:    ld1b.sx %s0, 184(%s1, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stl %s0, 8(%s1, %s11)
+; CHECK-NEXT:    ld1b.sx %s0, 8(%s1, %s11)
+; CHECK-NEXT:    adds.l %s11, 48, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %3 = alloca [10 x %struct.data], align 1
   %4 = getelementptr inbounds [10 x %struct.data], [10 x %struct.data]* %3, i64 0, i64 0, i32 0, i64 0
   call void @llvm.lifetime.start.p0i8(i64 40, i8* nonnull %4)
@@ -53,8 +54,9 @@ declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture)
 define signext i8 @func_rf(i8* readonly %0, i64 %1, i32 signext %2) {
 ; CHECK-LABEL: func_rf:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.sx %s0, 184(%s1, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.sx %s0, 8(%s1, %s11)
+; CHECK-NEXT:    adds.l %s11, 32, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %buf = alloca %"basic_string", align 8
 
   %sub631 = add nsw i64 %1, -1

diff  --git a/llvm/test/CodeGen/VE/Scalar/sext_zext_load.ll b/llvm/test/CodeGen/VE/Scalar/sext_zext_load.ll
index 600a02a5b130..f9f8aa8cebbf 100644
--- a/llvm/test/CodeGen/VE/Scalar/sext_zext_load.ll
+++ b/llvm/test/CodeGen/VE/Scalar/sext_zext_load.ll
@@ -3,8 +3,9 @@
 define signext i16 @func1() {
 ; CHECK-LABEL: func1:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.sx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.sx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = sext i8 %a.val to i16
@@ -14,8 +15,9 @@ define signext i16 @func1() {
 define i32 @func2() {
 ; CHECK-LABEL: func2:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.sx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.sx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = sext i8 %a.val to i32
@@ -25,8 +27,9 @@ define i32 @func2() {
 define i64 @func3() {
 ; CHECK-LABEL: func3:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.sx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.sx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = sext i8 %a.val to i64
@@ -36,9 +39,10 @@ define i64 @func3() {
 define zeroext i16 @func5() {
 ; CHECK-LABEL: func5:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.sx %s0, 191(, %s11)
+; CHECK-NEXT:    ld1b.sx %s0, 15(, %s11)
 ; CHECK-NEXT:    and %s0, %s0, (48)0
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = sext i8 %a.val to i16
@@ -48,8 +52,9 @@ define zeroext i16 @func5() {
 define i32 @func6() {
 ; CHECK-LABEL: func6:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.sx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.sx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = sext i8 %a.val to i32
@@ -59,8 +64,9 @@ define i32 @func6() {
 define i64 @func7() {
 ; CHECK-LABEL: func7:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.sx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.sx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = sext i8 %a.val to i64
@@ -70,8 +76,9 @@ define i64 @func7() {
 define signext i16 @func9() {
 ; CHECK-LABEL: func9:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = zext i8 %a.val to i16
@@ -81,8 +88,9 @@ define signext i16 @func9() {
 define i32 @func10() {
 ; CHECK-LABEL: func10:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = zext i8 %a.val to i32
@@ -92,8 +100,9 @@ define i32 @func10() {
 define i64 @func11() {
 ; CHECK-LABEL: func11:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = zext i8 %a.val to i64
@@ -103,8 +112,9 @@ define i64 @func11() {
 define zeroext i16 @func13() {
 ; CHECK-LABEL: func13:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = zext i8 %a.val to i16
@@ -114,8 +124,9 @@ define zeroext i16 @func13() {
 define zeroext i16 @func14() {
 ; CHECK-LABEL: func14:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = zext i8 %a.val to i16
@@ -125,8 +136,9 @@ define zeroext i16 @func14() {
 define i64 @func15() {
 ; CHECK-LABEL: func15:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i8, align 1
   %a.val = load i8, i8* %a, align 1
   %a.conv = zext i8 %a.val to i64
@@ -136,8 +148,9 @@ define i64 @func15() {
 define i32 @func17() {
 ; CHECK-LABEL: func17:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.sx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.sx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i16, align 2
   %a.val = load i16, i16* %a, align 2
   %a.conv = sext i16 %a.val to i32
@@ -147,8 +160,9 @@ define i32 @func17() {
 define i64 @func18() {
 ; CHECK-LABEL: func18:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.sx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.sx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i16, align 2
   %a.val = load i16, i16* %a, align 2
   %a.conv = sext i16 %a.val to i64
@@ -158,8 +172,9 @@ define i64 @func18() {
 define zeroext i16 @func20() {
 ; CHECK-LABEL: func20:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.zx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.zx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i16, align 2
   %a.conv = load i16, i16* %a, align 2
   ret i16 %a.conv
@@ -168,8 +183,9 @@ define zeroext i16 @func20() {
 define i64 @func21() {
 ; CHECK-LABEL: func21:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.sx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.sx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i16, align 2
   %a.val = load i16, i16* %a, align 2
   %a.conv = sext i16 %a.val to i64
@@ -179,8 +195,9 @@ define i64 @func21() {
 define i32 @func23() {
 ; CHECK-LABEL: func23:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.zx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.zx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i16, align 2
   %a.val = load i16, i16* %a, align 2
   %a.conv = zext i16 %a.val to i32
@@ -190,8 +207,9 @@ define i32 @func23() {
 define i64 @func24() {
 ; CHECK-LABEL: func24:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.zx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.zx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i16, align 2
   %a.val = load i16, i16* %a, align 2
   %a.conv = zext i16 %a.val to i64
@@ -201,8 +219,9 @@ define i64 @func24() {
 define zeroext i16 @func26() {
 ; CHECK-LABEL: func26:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.zx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.zx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i16, align 2
   %a.conv = load i16, i16* %a, align 2
   ret i16 %a.conv
@@ -211,8 +230,9 @@ define zeroext i16 @func26() {
 define i64 @func27() {
 ; CHECK-LABEL: func27:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld2b.zx %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld2b.zx %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i16, align 2
   %a.val = load i16, i16* %a, align 2
   %a.conv = zext i16 %a.val to i64
@@ -222,8 +242,9 @@ define i64 @func27() {
 define i64 @func29() {
 ; CHECK-LABEL: func29:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldl.sx %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i32, align 4
   %a.val = load i32, i32* %a, align 4
   %a.conv = sext i32 %a.val to i64
@@ -233,8 +254,9 @@ define i64 @func29() {
 define i64 @func31() {
 ; CHECK-LABEL: func31:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldl.sx %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i32, align 4
   %a.val = load i32, i32* %a, align 4
   %a.conv = sext i32 %a.val to i64
@@ -244,8 +266,9 @@ define i64 @func31() {
 define i64 @func33() {
 ; CHECK-LABEL: func33:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.zx %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldl.zx %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i32, align 4
   %a.val = load i32, i32* %a, align 4
   %a.conv = zext i32 %a.val to i64
@@ -255,8 +278,9 @@ define i64 @func33() {
 define i64 @func35() {
 ; CHECK-LABEL: func35:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.zx %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ldl.zx %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i32, align 4
   %a.val = load i32, i32* %a, align 4
   %a.conv = zext i32 %a.val to i64
@@ -266,10 +290,11 @@ define i64 @func35() {
 define signext i8 @func37() {
 ; CHECK-LABEL: func37:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
 ; CHECK-NEXT:    sll %s0, %s0, 63
 ; CHECK-NEXT:    sra.l %s0, %s0, 63
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i1, align 1
   %a.val = load i1, i1* %a, align 1
   %a.conv = sext i1 %a.val to i8
@@ -279,10 +304,11 @@ define signext i8 @func37() {
 define signext i16 @func38() {
 ; CHECK-LABEL: func38:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
 ; CHECK-NEXT:    sll %s0, %s0, 63
 ; CHECK-NEXT:    sra.l %s0, %s0, 63
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i1, align 1
   %a.val = load i1, i1* %a, align 1
   %a.conv = sext i1 %a.val to i16
@@ -292,10 +318,11 @@ define signext i16 @func38() {
 define signext i32 @func39() {
 ; CHECK-LABEL: func39:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
 ; CHECK-NEXT:    sll %s0, %s0, 63
 ; CHECK-NEXT:    sra.l %s0, %s0, 63
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i1, align 1
   %a.val = load i1, i1* %a, align 1
   %a.conv = sext i1 %a.val to i32
@@ -305,10 +332,11 @@ define signext i32 @func39() {
 define signext i64 @func40() {
 ; CHECK-LABEL: func40:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
 ; CHECK-NEXT:    sll %s0, %s0, 63
 ; CHECK-NEXT:    sra.l %s0, %s0, 63
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i1, align 1
   %a.val = load i1, i1* %a, align 1
   %a.conv = sext i1 %a.val to i64
@@ -318,8 +346,9 @@ define signext i64 @func40() {
 define signext i8 @func42() {
 ; CHECK-LABEL: func42:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i1, align 1
   %a.val = load i1, i1* %a, align 1
   %a.conv = zext i1 %a.val to i8
@@ -329,8 +358,9 @@ define signext i8 @func42() {
 define signext i16 @func43() {
 ; CHECK-LABEL: func43:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i1, align 1
   %a.val = load i1, i1* %a, align 1
   %a.conv = zext i1 %a.val to i16
@@ -340,8 +370,9 @@ define signext i16 @func43() {
 define signext i32 @func44() {
 ; CHECK-LABEL: func44:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i1, align 1
   %a.val = load i1, i1* %a, align 1
   %a.conv = zext i1 %a.val to i32
@@ -351,8 +382,9 @@ define signext i32 @func44() {
 define signext i64 @func45() {
 ; CHECK-LABEL: func45:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld1b.zx %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld1b.zx %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %a = alloca i1, align 1
   %a.val = load i1, i1* %a, align 1
   %a.conv = zext i1 %a.val to i64

diff  --git a/llvm/test/CodeGen/VE/Scalar/stackframe_align.ll b/llvm/test/CodeGen/VE/Scalar/stackframe_align.ll
index 6717e38ff132..7270bf0b9e84 100644
--- a/llvm/test/CodeGen/VE/Scalar/stackframe_align.ll
+++ b/llvm/test/CodeGen/VE/Scalar/stackframe_align.ll
@@ -12,12 +12,7 @@
 define i8* @test_frame7(i8* %0) {
 ; CHECK-LABEL: test_frame7:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -192(, %s11)
+; CHECK-NEXT:    adds.l %s11, -16, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB0_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -30,21 +25,15 @@ define i8* @test_frame7(i8* %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB0_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 185(, %s11)
-; CHECK-NEXT:    st1b %s1, 185(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, 9(, %s11)
+; CHECK-NEXT:    st1b %s1, 9(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
 ; CHECKFP-LABEL: test_frame7:
 ; CHECKFP:       # %bb.0:
 ; CHECKFP-NEXT:    st %s9, (, %s11)
 ; CHECKFP-NEXT:    st %s10, 8(, %s11)
-; CHECKFP-NEXT:    st %s15, 24(, %s11)
-; CHECKFP-NEXT:    st %s16, 32(, %s11)
 ; CHECKFP-NEXT:    or %s9, 0, %s11
 ; CHECKFP-NEXT:    lea %s11, -192(, %s11)
 ; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB0_2
@@ -62,8 +51,6 @@ define i8* @test_frame7(i8* %0) {
 ; CHECKFP-NEXT:    lea %s0, -7(, %s9)
 ; CHECKFP-NEXT:    st1b %s1, -7(, %s9)
 ; CHECKFP-NEXT:    or %s11, 0, %s9
-; CHECKFP-NEXT:    ld %s16, 32(, %s11)
-; CHECKFP-NEXT:    ld %s15, 24(, %s11)
 ; CHECKFP-NEXT:    ld %s10, 8(, %s11)
 ; CHECKFP-NEXT:    ld %s9, (, %s11)
 ; CHECKFP-NEXT:    b.l.t (, %s10)
@@ -81,12 +68,7 @@ define i8* @test_frame7(i8* %0) {
 define i8* @test_frame7_align8(i8* %0) {
 ; CHECK-LABEL: test_frame7_align8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -192(, %s11)
+; CHECK-NEXT:    adds.l %s11, -16, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB1_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -99,21 +81,15 @@ define i8* @test_frame7_align8(i8* %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB1_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 184(, %s11)
-; CHECK-NEXT:    st1b %s1, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, 8(, %s11)
+; CHECK-NEXT:    st1b %s1, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
 ; CHECKFP-LABEL: test_frame7_align8:
 ; CHECKFP:       # %bb.0:
 ; CHECKFP-NEXT:    st %s9, (, %s11)
 ; CHECKFP-NEXT:    st %s10, 8(, %s11)
-; CHECKFP-NEXT:    st %s15, 24(, %s11)
-; CHECKFP-NEXT:    st %s16, 32(, %s11)
 ; CHECKFP-NEXT:    or %s9, 0, %s11
 ; CHECKFP-NEXT:    lea %s11, -192(, %s11)
 ; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB1_2
@@ -131,8 +107,6 @@ define i8* @test_frame7_align8(i8* %0) {
 ; CHECKFP-NEXT:    lea %s0, -8(, %s9)
 ; CHECKFP-NEXT:    st1b %s1, -8(, %s9)
 ; CHECKFP-NEXT:    or %s11, 0, %s9
-; CHECKFP-NEXT:    ld %s16, 32(, %s11)
-; CHECKFP-NEXT:    ld %s15, 24(, %s11)
 ; CHECKFP-NEXT:    ld %s10, 8(, %s11)
 ; CHECKFP-NEXT:    ld %s9, (, %s11)
 ; CHECKFP-NEXT:    b.l.t (, %s10)
@@ -150,12 +124,7 @@ define i8* @test_frame7_align8(i8* %0) {
 define i8* @test_frame16_align16(i8* %0) {
 ; CHECK-LABEL: test_frame16_align16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -192(, %s11)
+; CHECK-NEXT:    adds.l %s11, -16, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB2_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -168,21 +137,15 @@ define i8* @test_frame16_align16(i8* %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB2_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, (, %s11)
+; CHECK-NEXT:    st1b %s1, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
 ; CHECKFP-LABEL: test_frame16_align16:
 ; CHECKFP:       # %bb.0:
 ; CHECKFP-NEXT:    st %s9, (, %s11)
 ; CHECKFP-NEXT:    st %s10, 8(, %s11)
-; CHECKFP-NEXT:    st %s15, 24(, %s11)
-; CHECKFP-NEXT:    st %s16, 32(, %s11)
 ; CHECKFP-NEXT:    or %s9, 0, %s11
 ; CHECKFP-NEXT:    lea %s11, -192(, %s11)
 ; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB2_2
@@ -200,8 +163,6 @@ define i8* @test_frame16_align16(i8* %0) {
 ; CHECKFP-NEXT:    lea %s0, -16(, %s9)
 ; CHECKFP-NEXT:    st1b %s1, -16(, %s9)
 ; CHECKFP-NEXT:    or %s11, 0, %s9
-; CHECKFP-NEXT:    ld %s16, 32(, %s11)
-; CHECKFP-NEXT:    ld %s15, 24(, %s11)
 ; CHECKFP-NEXT:    ld %s10, 8(, %s11)
 ; CHECKFP-NEXT:    ld %s9, (, %s11)
 ; CHECKFP-NEXT:    b.l.t (, %s10)
@@ -222,8 +183,6 @@ define i8* @test_frame16_align32(i8* %0) {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -224(, %s11)
 ; CHECK-NEXT:    and %s11, %s11, (59)1
@@ -242,8 +201,6 @@ define i8* @test_frame16_align32(i8* %0) {
 ; CHECK-NEXT:    lea %s0, 192(, %s11)
 ; CHECK-NEXT:    st1b %s1, 192(, %s11)
 ; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
@@ -252,8 +209,6 @@ define i8* @test_frame16_align32(i8* %0) {
 ; CHECKFP:       # %bb.0:
 ; CHECKFP-NEXT:    st %s9, (, %s11)
 ; CHECKFP-NEXT:    st %s10, 8(, %s11)
-; CHECKFP-NEXT:    st %s15, 24(, %s11)
-; CHECKFP-NEXT:    st %s16, 32(, %s11)
 ; CHECKFP-NEXT:    or %s9, 0, %s11
 ; CHECKFP-NEXT:    lea %s11, -224(, %s11)
 ; CHECKFP-NEXT:    and %s11, %s11, (59)1
@@ -272,8 +227,6 @@ define i8* @test_frame16_align32(i8* %0) {
 ; CHECKFP-NEXT:    lea %s0, 192(, %s11)
 ; CHECKFP-NEXT:    st1b %s1, 192(, %s11)
 ; CHECKFP-NEXT:    or %s11, 0, %s9
-; CHECKFP-NEXT:    ld %s16, 32(, %s11)
-; CHECKFP-NEXT:    ld %s15, 24(, %s11)
 ; CHECKFP-NEXT:    ld %s10, 8(, %s11)
 ; CHECKFP-NEXT:    ld %s9, (, %s11)
 ; CHECKFP-NEXT:    b.l.t (, %s10)
@@ -294,8 +247,6 @@ define i8* @test_frame32_align32(i8* %0) {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -224(, %s11)
 ; CHECK-NEXT:    and %s11, %s11, (59)1
@@ -314,8 +265,6 @@ define i8* @test_frame32_align32(i8* %0) {
 ; CHECK-NEXT:    lea %s0, 192(, %s11)
 ; CHECK-NEXT:    st1b %s1, 192(, %s11)
 ; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
@@ -324,8 +273,6 @@ define i8* @test_frame32_align32(i8* %0) {
 ; CHECKFP:       # %bb.0:
 ; CHECKFP-NEXT:    st %s9, (, %s11)
 ; CHECKFP-NEXT:    st %s10, 8(, %s11)
-; CHECKFP-NEXT:    st %s15, 24(, %s11)
-; CHECKFP-NEXT:    st %s16, 32(, %s11)
 ; CHECKFP-NEXT:    or %s9, 0, %s11
 ; CHECKFP-NEXT:    lea %s11, -224(, %s11)
 ; CHECKFP-NEXT:    and %s11, %s11, (59)1
@@ -344,8 +291,6 @@ define i8* @test_frame32_align32(i8* %0) {
 ; CHECKFP-NEXT:    lea %s0, 192(, %s11)
 ; CHECKFP-NEXT:    st1b %s1, 192(, %s11)
 ; CHECKFP-NEXT:    or %s11, 0, %s9
-; CHECKFP-NEXT:    ld %s16, 32(, %s11)
-; CHECKFP-NEXT:    ld %s15, 24(, %s11)
 ; CHECKFP-NEXT:    ld %s10, 8(, %s11)
 ; CHECKFP-NEXT:    ld %s9, (, %s11)
 ; CHECKFP-NEXT:    b.l.t (, %s10)
@@ -370,8 +315,6 @@ define i8* @test_frame_dynalign16(i8* %0, i64 %1) {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -240(, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB5_2
@@ -396,8 +339,6 @@ define i8* @test_frame_dynalign16(i8* %0, i64 %1) {
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s2)
 ; CHECK-NEXT:    st1b %s1, (, %s0)
 ; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
@@ -406,8 +347,6 @@ define i8* @test_frame_dynalign16(i8* %0, i64 %1) {
 ; CHECKFP:       # %bb.0:
 ; CHECKFP-NEXT:    st %s9, (, %s11)
 ; CHECKFP-NEXT:    st %s10, 8(, %s11)
-; CHECKFP-NEXT:    st %s15, 24(, %s11)
-; CHECKFP-NEXT:    st %s16, 32(, %s11)
 ; CHECKFP-NEXT:    or %s9, 0, %s11
 ; CHECKFP-NEXT:    lea %s11, -240(, %s11)
 ; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB5_2
@@ -432,8 +371,6 @@ define i8* @test_frame_dynalign16(i8* %0, i64 %1) {
 ; CHECKFP-NEXT:    ld1b.zx %s1, (, %s2)
 ; CHECKFP-NEXT:    st1b %s1, (, %s0)
 ; CHECKFP-NEXT:    or %s11, 0, %s9
-; CHECKFP-NEXT:    ld %s16, 32(, %s11)
-; CHECKFP-NEXT:    ld %s15, 24(, %s11)
 ; CHECKFP-NEXT:    ld %s10, 8(, %s11)
 ; CHECKFP-NEXT:    ld %s9, (, %s11)
 ; CHECKFP-NEXT:    b.l.t (, %s10)
@@ -459,8 +396,6 @@ define i8* @test_frame16_align16_dynalign32(i8* %0, i64 %n) {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    st %s17, 40(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -288(, %s11)
@@ -493,8 +428,6 @@ define i8* @test_frame16_align16_dynalign32(i8* %0, i64 %n) {
 ; CHECK-NEXT:    st1b %s1, (, %s0)
 ; CHECK-NEXT:    or %s11, 0, %s9
 ; CHECK-NEXT:    ld %s17, 40(, %s11)
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
@@ -503,8 +436,6 @@ define i8* @test_frame16_align16_dynalign32(i8* %0, i64 %n) {
 ; CHECKFP:       # %bb.0:
 ; CHECKFP-NEXT:    st %s9, (, %s11)
 ; CHECKFP-NEXT:    st %s10, 8(, %s11)
-; CHECKFP-NEXT:    st %s15, 24(, %s11)
-; CHECKFP-NEXT:    st %s16, 32(, %s11)
 ; CHECKFP-NEXT:    st %s17, 40(, %s11)
 ; CHECKFP-NEXT:    or %s9, 0, %s11
 ; CHECKFP-NEXT:    lea %s11, -288(, %s11)
@@ -537,8 +468,6 @@ define i8* @test_frame16_align16_dynalign32(i8* %0, i64 %n) {
 ; CHECKFP-NEXT:    st1b %s1, (, %s0)
 ; CHECKFP-NEXT:    or %s11, 0, %s9
 ; CHECKFP-NEXT:    ld %s17, 40(, %s11)
-; CHECKFP-NEXT:    ld %s16, 32(, %s11)
-; CHECKFP-NEXT:    ld %s15, 24(, %s11)
 ; CHECKFP-NEXT:    ld %s10, 8(, %s11)
 ; CHECKFP-NEXT:    ld %s9, (, %s11)
 ; CHECKFP-NEXT:    b.l.t (, %s10)

diff  --git a/llvm/test/CodeGen/VE/Scalar/stackframe_call.ll b/llvm/test/CodeGen/VE/Scalar/stackframe_call.ll
index 3696af529d08..560502611add 100644
--- a/llvm/test/CodeGen/VE/Scalar/stackframe_call.ll
+++ b/llvm/test/CodeGen/VE/Scalar/stackframe_call.ll
@@ -17,8 +17,6 @@ define i8* @test_frame0(i8* %0, i8* %1) {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -240(, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB0_2
@@ -37,8 +35,6 @@ define i8* @test_frame0(i8* %0, i8* %1) {
 ; CHECK-NEXT:    lea.sl %s12, fun at hi(, %s2)
 ; CHECK-NEXT:    bsic %s10, (, %s12)
 ; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
@@ -89,8 +85,6 @@ define i8* @test_frame32(i8* %0) {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -272(, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB1_2
@@ -111,8 +105,6 @@ define i8* @test_frame32(i8* %0) {
 ; CHECK-NEXT:    lea %s0, 240(, %s11)
 ; CHECK-NEXT:    bsic %s10, (, %s12)
 ; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
@@ -173,8 +165,6 @@ define i8* @test_align32(i32 signext %0, i8* nocapture readnone %1) {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    st %s17, 40(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -288(, %s11)
@@ -208,8 +198,6 @@ define i8* @test_align32(i32 signext %0, i8* nocapture readnone %1) {
 ; CHECK-NEXT:    bsic %s10, (, %s12)
 ; CHECK-NEXT:    or %s11, 0, %s9
 ; CHECK-NEXT:    ld %s17, 40(, %s11)
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
@@ -280,8 +268,6 @@ define i8* @test_frame0_var(i8* %0, i8* %1) {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -240(, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB3_2
@@ -305,8 +291,6 @@ define i8* @test_frame0_var(i8* %0, i8* %1) {
 ; CHECK-NEXT:    lea.sl %s12, fun at hi(, %s2)
 ; CHECK-NEXT:    bsic %s10, (, %s12)
 ; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
@@ -363,8 +347,6 @@ define i8* @test_frame32_var(i8* %0) {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -272(, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB4_2
@@ -390,8 +372,6 @@ define i8* @test_frame32_var(i8* %0) {
 ; CHECK-NEXT:    lea %s0, 240(, %s11)
 ; CHECK-NEXT:    bsic %s10, (, %s12)
 ; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
@@ -454,8 +434,6 @@ define i8* @test_align32_var(i32 signext %0, i8* nocapture readnone %1) {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    st %s17, 40(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -288(, %s11)
@@ -494,8 +472,6 @@ define i8* @test_align32_var(i32 signext %0, i8* nocapture readnone %1) {
 ; CHECK-NEXT:    bsic %s10, (, %s12)
 ; CHECK-NEXT:    or %s11, 0, %s9
 ; CHECK-NEXT:    ld %s17, 40(, %s11)
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)

diff  --git a/llvm/test/CodeGen/VE/Scalar/stackframe_nocall.ll b/llvm/test/CodeGen/VE/Scalar/stackframe_nocall.ll
index e7f8b23c647a..f157c54c36fc 100644
--- a/llvm/test/CodeGen/VE/Scalar/stackframe_nocall.ll
+++ b/llvm/test/CodeGen/VE/Scalar/stackframe_nocall.ll
@@ -29,12 +29,7 @@ define i8* @test_frame0(i8* nocapture readnone %0, i8* readnone returned %1) {
 define nonnull i8* @test_frame32(i8* nocapture readonly %0) {
 ; CHECK-LABEL: test_frame32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -208(, %s11)
+; CHECK-NEXT:    adds.l %s11, -32, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB1_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -47,23 +42,14 @@ define nonnull i8* @test_frame32(i8* nocapture readonly %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB1_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, (, %s11)
+; CHECK-NEXT:    st1b %s1, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 32, %s11
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
 ; PIC-LABEL: test_frame32:
 ; PIC:       # %bb.0:
-; PIC-NEXT:    st %s9, (, %s11)
-; PIC-NEXT:    st %s10, 8(, %s11)
-; PIC-NEXT:    st %s15, 24(, %s11)
-; PIC-NEXT:    st %s16, 32(, %s11)
-; PIC-NEXT:    or %s9, 0, %s11
-; PIC-NEXT:    lea %s11, -208(, %s11)
+; PIC-NEXT:    adds.l %s11, -32, %s11
 ; PIC-NEXT:    brge.l.t %s11, %s8, .LBB1_2
 ; PIC-NEXT:  # %bb.1:
 ; PIC-NEXT:    ld %s61, 24(, %s14)
@@ -76,13 +62,9 @@ define nonnull i8* @test_frame32(i8* nocapture readonly %0) {
 ; PIC-NEXT:    or %s0, 0, %s62
 ; PIC-NEXT:  .LBB1_2:
 ; PIC-NEXT:    ld1b.zx %s1, (, %s0)
-; PIC-NEXT:    lea %s0, 176(, %s11)
-; PIC-NEXT:    st1b %s1, 176(, %s11)
-; PIC-NEXT:    or %s11, 0, %s9
-; PIC-NEXT:    ld %s16, 32(, %s11)
-; PIC-NEXT:    ld %s15, 24(, %s11)
-; PIC-NEXT:    ld %s10, 8(, %s11)
-; PIC-NEXT:    ld %s9, (, %s11)
+; PIC-NEXT:    lea %s0, (, %s11)
+; PIC-NEXT:    st1b %s1, (, %s11)
+; PIC-NEXT:    adds.l %s11, 32, %s11
 ; PIC-NEXT:    b.l.t (, %s10)
   %2 = alloca [32 x i8], align 1
   %3 = getelementptr inbounds [32 x i8], [32 x i8]* %2, i64 0, i64 0
@@ -105,8 +87,6 @@ define noalias nonnull i8* @test_align32(i32 signext %0, i8* nocapture readonly
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    st %s17, 40(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -288(, %s11)
@@ -139,8 +119,6 @@ define noalias nonnull i8* @test_align32(i32 signext %0, i8* nocapture readonly
 ; CHECK-NEXT:    st1b %s1, (, %s2)
 ; CHECK-NEXT:    or %s11, 0, %s9
 ; CHECK-NEXT:    ld %s17, 40(, %s11)
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
@@ -259,12 +237,7 @@ define i8* @test_frame0_var(i8* returned %0, i8* nocapture readnone %1) {
 define nonnull i8* @test_frame32_var(i8* nocapture readnone %0) {
 ; CHECK-LABEL: test_frame32_var:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -208(, %s11)
+; CHECK-NEXT:    adds.l %s11, -32, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB4_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -280,13 +253,9 @@ define nonnull i8* @test_frame32_var(i8* nocapture readnone %0) {
 ; CHECK-NEXT:    and %s0, %s0, (32)0
 ; CHECK-NEXT:    lea.sl %s0, data at hi(, %s0)
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, (, %s11)
+; CHECK-NEXT:    st1b %s1, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 32, %s11
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
 ; PIC-LABEL: test_frame32_var:
@@ -340,8 +309,6 @@ define noalias nonnull i8* @test_align32_var(i32 signext %0, i8* nocapture reado
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    st %s17, 40(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s11, -288(, %s11)
@@ -374,8 +341,6 @@ define noalias nonnull i8* @test_align32_var(i32 signext %0, i8* nocapture reado
 ; CHECK-NEXT:    st1b %s1, (, %s2)
 ; CHECK-NEXT:    or %s11, 0, %s9
 ; CHECK-NEXT:    ld %s17, 40(, %s11)
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)

diff  --git a/llvm/test/CodeGen/VE/Scalar/stackframe_size.ll b/llvm/test/CodeGen/VE/Scalar/stackframe_size.ll
index 5919b6f479fb..0d40b3bed71b 100644
--- a/llvm/test/CodeGen/VE/Scalar/stackframe_size.ll
+++ b/llvm/test/CodeGen/VE/Scalar/stackframe_size.ll
@@ -19,12 +19,7 @@ define signext i32 @test_frame0(i32 signext %0) {
 define i8* @test_frame8(i8* %0) {
 ; CHECK-LABEL: test_frame8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -192(, %s11)
+; CHECK-NEXT:    adds.l %s11, -16, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB1_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -37,13 +32,9 @@ define i8* @test_frame8(i8* %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB1_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 184(, %s11)
-; CHECK-NEXT:    st1b %s1, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, 8(, %s11)
+; CHECK-NEXT:    st1b %s1, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
 ; CHECK-NEXT:    b.l.t (, %s10)
   %2 = alloca [8 x i8], align 1
   %3 = getelementptr inbounds [8 x i8], [8 x i8]* %2, i64 0, i64 0
@@ -56,12 +47,7 @@ define i8* @test_frame8(i8* %0) {
 define i8* @test_frame16(i8* %0) {
 ; CHECK-LABEL: test_frame16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -192(, %s11)
+; CHECK-NEXT:    adds.l %s11, -16, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB2_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -74,13 +60,9 @@ define i8* @test_frame16(i8* %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB2_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, (, %s11)
+; CHECK-NEXT:    st1b %s1, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
 ; CHECK-NEXT:    b.l.t (, %s10)
   %2 = alloca [16 x i8], align 1
   %3 = getelementptr inbounds [16 x i8], [16 x i8]* %2, i64 0, i64 0
@@ -93,12 +75,7 @@ define i8* @test_frame16(i8* %0) {
 define i8* @test_frame32(i8* %0) {
 ; CHECK-LABEL: test_frame32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -208(, %s11)
+; CHECK-NEXT:    adds.l %s11, -32, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB3_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -111,13 +88,9 @@ define i8* @test_frame32(i8* %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB3_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, (, %s11)
+; CHECK-NEXT:    st1b %s1, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 32, %s11
 ; CHECK-NEXT:    b.l.t (, %s10)
   %2 = alloca [32 x i8], align 1
   %3 = getelementptr inbounds [32 x i8], [32 x i8]* %2, i64 0, i64 0
@@ -130,12 +103,7 @@ define i8* @test_frame32(i8* %0) {
 define i8* @test_frame64(i8* %0) {
 ; CHECK-LABEL: test_frame64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -240(, %s11)
+; CHECK-NEXT:    adds.l %s11, -64, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB4_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -148,13 +116,9 @@ define i8* @test_frame64(i8* %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB4_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, (, %s11)
+; CHECK-NEXT:    st1b %s1, (, %s11)
+; CHECK-NEXT:    lea %s11, 64(, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
   %2 = alloca [64 x i8], align 1
   %3 = getelementptr inbounds [64 x i8], [64 x i8]* %2, i64 0, i64 0
@@ -167,12 +131,7 @@ define i8* @test_frame64(i8* %0) {
 define i8* @test_frame128(i8* %0) {
 ; CHECK-LABEL: test_frame128:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -304(, %s11)
+; CHECK-NEXT:    lea %s11, -128(, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB5_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -185,13 +144,9 @@ define i8* @test_frame128(i8* %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB5_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, (, %s11)
+; CHECK-NEXT:    st1b %s1, (, %s11)
+; CHECK-NEXT:    lea %s11, 128(, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
   %2 = alloca [128 x i8], align 1
   %3 = getelementptr inbounds [128 x i8], [128 x i8]* %2, i64 0, i64 0
@@ -204,12 +159,7 @@ define i8* @test_frame128(i8* %0) {
 define i8* @test_frame65536(i8* %0) {
 ; CHECK-LABEL: test_frame65536:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s11, -65712(, %s11)
+; CHECK-NEXT:    lea %s11, -65536(, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB6_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -222,13 +172,9 @@ define i8* @test_frame65536(i8* %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB6_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, (, %s11)
+; CHECK-NEXT:    st1b %s1, (, %s11)
+; CHECK-NEXT:    lea %s11, 65536(, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
   %2 = alloca [65536 x i8], align 1
   %3 = getelementptr inbounds [65536 x i8], [65536 x i8]* %2, i64 0, i64 0
@@ -241,14 +187,9 @@ define i8* @test_frame65536(i8* %0) {
 define i8* @test_frame4294967296(i8* %0) {
 ; CHECK-LABEL: test_frame4294967296:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s13, -176
+; CHECK-NEXT:    lea %s13, 0
 ; CHECK-NEXT:    and %s13, %s13, (32)0
-; CHECK-NEXT:    lea.sl %s11, -2(%s13, %s11)
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB7_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
@@ -261,13 +202,11 @@ define i8* @test_frame4294967296(i8* %0) {
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB7_2:
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s0, (, %s11)
+; CHECK-NEXT:    st1b %s1, (, %s11)
+; CHECK-NEXT:    lea %s13, 0
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, 1(%s13, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
   %2 = alloca [4294967296 x i8], align 1
   %3 = getelementptr inbounds [4294967296 x i8], [4294967296 x i8]* %2, i64 0, i64 0

diff  --git a/llvm/test/CodeGen/VE/Scalar/store-align1.ll b/llvm/test/CodeGen/VE/Scalar/store-align1.ll
index dc1400bfca20..fa8076aa1256 100644
--- a/llvm/test/CodeGen/VE/Scalar/store-align1.ll
+++ b/llvm/test/CodeGen/VE/Scalar/store-align1.ll
@@ -11,8 +11,9 @@
 define void @storef64stk(double %0) {
 ; CHECK-LABEL: storef64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca double, align 1
   store double %0, double* %addr, align 1
   ret void
@@ -22,8 +23,9 @@ define void @storef64stk(double %0) {
 define void @storef32stk(float %0) {
 ; CHECK-LABEL: storef32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    stu %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stu %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca float, align 1
   store float %0, float* %addr, align 1
   ret void
@@ -33,8 +35,9 @@ define void @storef32stk(float %0) {
 define void @storei64stk(i64 %0) {
 ; CHECK-LABEL: storei64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i64, align 1
   store i64 %0, i64* %addr, align 1
   ret void
@@ -44,8 +47,9 @@ define void @storei64stk(i64 %0) {
 define void @storei32stk(i32 %0) {
 ; CHECK-LABEL: storei32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    stl %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stl %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i32, align 1
   store i32 %0, i32* %addr, align 1
   ret void
@@ -55,8 +59,9 @@ define void @storei32stk(i32 %0) {
 define void @storei16stk(i16 %0) {
 ; CHECK-LABEL: storei16stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st2b %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st2b %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i16, align 1
   store i16 %0, i16* %addr, align 1
   ret void
@@ -66,8 +71,9 @@ define void @storei16stk(i16 %0) {
 define void @storei8stk(i8 %0) {
 ; CHECK-LABEL: storei8stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st1b %s0, 191(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st1b %s0, 15(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i8, align 1
   store i8 %0, i8* %addr, align 1
   ret void

diff  --git a/llvm/test/CodeGen/VE/Scalar/store-align2.ll b/llvm/test/CodeGen/VE/Scalar/store-align2.ll
index f64af8f33874..7b4596ce4ae5 100644
--- a/llvm/test/CodeGen/VE/Scalar/store-align2.ll
+++ b/llvm/test/CodeGen/VE/Scalar/store-align2.ll
@@ -11,8 +11,9 @@
 define void @storef64stk(double %0) {
 ; CHECK-LABEL: storef64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca double, align 2
   store double %0, double* %addr, align 2
   ret void
@@ -22,8 +23,9 @@ define void @storef64stk(double %0) {
 define void @storef32stk(float %0) {
 ; CHECK-LABEL: storef32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    stu %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stu %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca float, align 2
   store float %0, float* %addr, align 2
   ret void
@@ -33,8 +35,9 @@ define void @storef32stk(float %0) {
 define void @storei64stk(i64 %0) {
 ; CHECK-LABEL: storei64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i64, align 2
   store i64 %0, i64* %addr, align 2
   ret void
@@ -44,8 +47,9 @@ define void @storei64stk(i64 %0) {
 define void @storei32stk(i32 %0) {
 ; CHECK-LABEL: storei32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    stl %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stl %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i32, align 2
   store i32 %0, i32* %addr, align 2
   ret void
@@ -55,8 +59,9 @@ define void @storei32stk(i32 %0) {
 define void @storei16stk(i16 %0) {
 ; CHECK-LABEL: storei16stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st2b %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st2b %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i16, align 2
   store i16 %0, i16* %addr, align 2
   ret void
@@ -66,8 +71,9 @@ define void @storei16stk(i16 %0) {
 define void @storei8stk(i8 %0) {
 ; CHECK-LABEL: storei8stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st1b %s0, 190(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st1b %s0, 14(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i8, align 2
   store i8 %0, i8* %addr, align 2
   ret void

diff  --git a/llvm/test/CodeGen/VE/Scalar/store-align4.ll b/llvm/test/CodeGen/VE/Scalar/store-align4.ll
index 49bd127a0ffb..153c219343cf 100644
--- a/llvm/test/CodeGen/VE/Scalar/store-align4.ll
+++ b/llvm/test/CodeGen/VE/Scalar/store-align4.ll
@@ -11,8 +11,9 @@
 define void @storef64stk(double %0) {
 ; CHECK-LABEL: storef64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca double, align 4
   store double %0, double* %addr, align 4
   ret void
@@ -22,8 +23,9 @@ define void @storef64stk(double %0) {
 define void @storef32stk(float %0) {
 ; CHECK-LABEL: storef32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    stu %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stu %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca float, align 4
   store float %0, float* %addr, align 4
   ret void
@@ -33,8 +35,9 @@ define void @storef32stk(float %0) {
 define void @storei64stk(i64 %0) {
 ; CHECK-LABEL: storei64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i64, align 4
   store i64 %0, i64* %addr, align 4
   ret void
@@ -44,8 +47,9 @@ define void @storei64stk(i64 %0) {
 define void @storei32stk(i32 %0) {
 ; CHECK-LABEL: storei32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    stl %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stl %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i32, align 4
   store i32 %0, i32* %addr, align 4
   ret void
@@ -55,8 +59,9 @@ define void @storei32stk(i32 %0) {
 define void @storei16stk(i16 %0) {
 ; CHECK-LABEL: storei16stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st2b %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st2b %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i16, align 4
   store i16 %0, i16* %addr, align 4
   ret void
@@ -66,8 +71,9 @@ define void @storei16stk(i16 %0) {
 define void @storei8stk(i8 %0) {
 ; CHECK-LABEL: storei8stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st1b %s0, 188(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st1b %s0, 12(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i8, align 4
   store i8 %0, i8* %addr, align 4
   ret void

diff  --git a/llvm/test/CodeGen/VE/Scalar/store-align8.ll b/llvm/test/CodeGen/VE/Scalar/store-align8.ll
index be61b69fae24..4d166cc5e5ed 100644
--- a/llvm/test/CodeGen/VE/Scalar/store-align8.ll
+++ b/llvm/test/CodeGen/VE/Scalar/store-align8.ll
@@ -11,8 +11,9 @@
 define void @storef64stk(double %0) {
 ; CHECK-LABEL: storef64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca double, align 8
   store double %0, double* %addr, align 8
   ret void
@@ -22,8 +23,9 @@ define void @storef64stk(double %0) {
 define void @storef32stk(float %0) {
 ; CHECK-LABEL: storef32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    stu %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stu %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca float, align 8
   store float %0, float* %addr, align 8
   ret void
@@ -33,8 +35,9 @@ define void @storef32stk(float %0) {
 define void @storei64stk(i64 %0) {
 ; CHECK-LABEL: storei64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i64, align 8
   store i64 %0, i64* %addr, align 8
   ret void
@@ -44,8 +47,9 @@ define void @storei64stk(i64 %0) {
 define void @storei32stk(i32 %0) {
 ; CHECK-LABEL: storei32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    stl %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stl %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i32, align 8
   store i32 %0, i32* %addr, align 8
   ret void
@@ -55,8 +59,9 @@ define void @storei32stk(i32 %0) {
 define void @storei16stk(i16 %0) {
 ; CHECK-LABEL: storei16stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st2b %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st2b %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i16, align 8
   store i16 %0, i16* %addr, align 8
   ret void
@@ -66,8 +71,9 @@ define void @storei16stk(i16 %0) {
 define void @storei8stk(i8 %0) {
 ; CHECK-LABEL: storei8stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st1b %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st1b %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i8, align 8
   store i8 %0, i8* %addr, align 8
   ret void

diff  --git a/llvm/test/CodeGen/VE/Scalar/store.ll b/llvm/test/CodeGen/VE/Scalar/store.ll
index 0581866f732b..748607d1b358 100644
--- a/llvm/test/CodeGen/VE/Scalar/store.ll
+++ b/llvm/test/CodeGen/VE/Scalar/store.ll
@@ -119,9 +119,10 @@ define void @storei8tr(i8* nocapture %0, i64 %1) {
 define void @storef128stk(fp128 %0) {
 ; CHECK-LABEL: storef128stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s1, 176(, %s11)
-; CHECK-NEXT:    st %s0, 184(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s1, (, %s11)
+; CHECK-NEXT:    st %s0, 8(, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca fp128, align 16
   store fp128 %0, fp128* %addr, align 16
   ret void
@@ -131,8 +132,9 @@ define void @storef128stk(fp128 %0) {
 define void @storef64stk(double %0) {
 ; CHECK-LABEL: storef64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca double, align 16
   store double %0, double* %addr, align 16
   ret void
@@ -142,8 +144,9 @@ define void @storef64stk(double %0) {
 define void @storef32stk(float %0) {
 ; CHECK-LABEL: storef32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    stu %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stu %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca float, align 16
   store float %0, float* %addr, align 16
   ret void
@@ -153,9 +156,10 @@ define void @storef32stk(float %0) {
 define void @storei128stk(i128 %0) {
 ; CHECK-LABEL: storei128stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s1, 184(, %s11)
-; CHECK-NEXT:    st %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s1, 8(, %s11)
+; CHECK-NEXT:    st %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i128, align 16
   store i128 %0, i128* %addr, align 16
   ret void
@@ -165,8 +169,9 @@ define void @storei128stk(i128 %0) {
 define void @storei64stk(i64 %0) {
 ; CHECK-LABEL: storei64stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i64, align 16
   store i64 %0, i64* %addr, align 16
   ret void
@@ -176,8 +181,9 @@ define void @storei64stk(i64 %0) {
 define void @storei32stk(i32 %0) {
 ; CHECK-LABEL: storei32stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    stl %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    stl %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i32, align 16
   store i32 %0, i32* %addr, align 16
   ret void
@@ -187,8 +193,9 @@ define void @storei32stk(i32 %0) {
 define void @storei16stk(i16 %0) {
 ; CHECK-LABEL: storei16stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st2b %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st2b %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i16, align 16
   store i16 %0, i16* %addr, align 16
   ret void
@@ -198,8 +205,9 @@ define void @storei16stk(i16 %0) {
 define void @storei8stk(i8 %0) {
 ; CHECK-LABEL: storei8stk:
 ; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    st1b %s0, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    st1b %s0, (, %s11)
+; CHECK-NEXT:    adds.l %s11, 16, %s11
+; CHECK-NEXT:    b.l.t (, %s10)
   %addr = alloca i8, align 16
   store i8 %0, i8* %addr, align 16
   ret void

diff  --git a/llvm/test/CodeGen/VE/Vector/fastcc_callee.ll b/llvm/test/CodeGen/VE/Vector/fastcc_callee.ll
index 433647b57ac9..fe601b082192 100644
--- a/llvm/test/CodeGen/VE/Vector/fastcc_callee.ll
+++ b/llvm/test/CodeGen/VE/Vector/fastcc_callee.ll
@@ -4,41 +4,41 @@
 
 define fastcc i32 @stack_stack_arg_i32_r9(i1 %0, i8 %1, i16 %2, i32 %3, i64 %4, i32 %5, i32 %6, i32 %7, i32 %8, i32 %9) {
 ; CHECK-LABEL: stack_stack_arg_i32_r9:
-; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 424(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ldl.sx %s0, 248(, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
   ret i32 %9
 }
 
 define fastcc i64 @stack_stack_arg_i64_r9(i1 %0, i8 %1, i16 %2, i32 %3, i64 %4, i64 %5, i64 %6, i64 %7, i64 %8, i64 %9) {
 ; CHECK-LABEL: stack_stack_arg_i64_r9:
-; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ld %s0, 424(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ld %s0, 248(, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
   ret i64 %9
 }
 
 define fastcc float @stack_stack_arg_f32_r9(float %p0, float %p1, float %p2, float %p3, float %p4, float %p5, float %p6, float %p7, float %s0, float %s1) {
 ; CHECK-LABEL: stack_stack_arg_f32_r9:
-; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldu %s0, 428(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ldu %s0, 252(, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
   ret float %s1
 }
 
 define fastcc i32 @stack_stack_arg_i32f32_r8(i32 %p0, float %p1, i32 %p2, float %p3, i32 %p4, float %p5, i32 %p6, float %p7, i32 %s0, float %s1) {
 ; CHECK-LABEL: stack_stack_arg_i32f32_r8:
-; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldl.sx %s0, 416(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ldl.sx %s0, 240(, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
   ret i32 %s0
 }
 
 define fastcc float @stack_stack_arg_i32f32_r9(i32 %p0, float %p1, i32 %p2, float %p3, i32 %p4, float %p5, i32 %p6, float %p7, i32 %s0, float %s1) {
 ; CHECK-LABEL: stack_stack_arg_i32f32_r9:
-; CHECK:       .LBB{{[0-9]+}}_2:
-; CHECK-NEXT:    ldu %s0, 428(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ldu %s0, 252(, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
   ret float %s1
 }
 


        


More information about the llvm-branch-commits mailing list