[AArch64][GlobalISel] Improve codegen for some constant vectors by using constant pool loads

[aemerson]

Summary

There's more smarts in AArch64ISelLowering that we don't have yet, but this change incrementally improves some of the more common patterns. I think future iterations will want to use some combination of PostLegalizerCombiner and the selector to catch the other cases.

[aemerson]

llvm/test/CodeGen/AArch64/combine-loads.ll

7

I'm not entirely sure what this test was meant to demonstrate except poor codegen on -O0 (surprise surprise).

[paquette]

llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp

4503–4509

Can getConstantVRegValWithLookThrough handle this?

It has a HandleFConstants parameter.

[aemerson]

llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp

4503–4509

getConstantVRegWithLookThrough tries to return an int64_t, but here I need to find the actual ConstantInt/ConstantFP IR values in order to generate the constant pool entries.

[paquette]

LGTM

llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp

 #include "llvm/CodeGen/MachineBasicBlock.h"
 #include "llvm/CodeGen/MachineConstantPool.h"
 #include "llvm/CodeGen/MachineFunction.h"
 #include "llvm/CodeGen/MachineInstr.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
 #include "llvm/CodeGen/MachineOperand.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/CodeGen/TargetOpcodes.h"
+#include "llvm/IR/Constants.h"
 #include "llvm/IR/Type.h"
 #include "llvm/IR/IntrinsicsAArch64.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/raw_ostream.h"
 
 #define DEBUG_TYPE "aarch64-isel"
 
 using namespace llvm;
private:
   /// The lane inserted into is defined by \p LaneIdx. The vector source
   /// register is given by \p SrcReg. The register containing the element is
   /// given by \p EltReg.
   MachineInstr *emitLaneInsert(Optional<Register> DstReg, Register SrcReg,
                                Register EltReg, unsigned LaneIdx,
                                const RegisterBank &RB,
                                MachineIRBuilder &MIRBuilder) const;
   bool selectInsertElt(MachineInstr &I, MachineRegisterInfo &MRI) const;
+  bool tryOptConstantBuildVec(MachineInstr &MI, LLT DstTy,
+                              MachineRegisterInfo &MRI) const;
   bool selectBuildVector(MachineInstr &I, MachineRegisterInfo &MRI) const;
   bool selectMergeValues(MachineInstr &I, MachineRegisterInfo &MRI) const;
   bool selectUnmergeValues(MachineInstr &I, MachineRegisterInfo &MRI) const;
 
   bool selectShuffleVector(MachineInstr &I, MachineRegisterInfo &MRI) const;
   bool selectExtractElt(MachineInstr &I, MachineRegisterInfo &MRI) const;
   bool selectConcatVectors(MachineInstr &I, MachineRegisterInfo &MRI) const;
   bool selectSplitVectorUnmerge(MachineInstr &I,
if (VecSize < 128) {
     InsMI->getOperand(0).setReg(DstReg);
     constrainSelectedInstRegOperands(*InsMI, TII, TRI, RBI);
   }
 
   I.eraseFromParent();
   return true;
 }
 
+bool AArch64InstructionSelector::tryOptConstantBuildVec(
+    MachineInstr &I, LLT DstTy, MachineRegisterInfo &MRI) const {
+  assert(I.getOpcode() == TargetOpcode::G_BUILD_VECTOR);
+  assert(DstTy.getSizeInBits() <= 128 && "Unexpected build_vec type!");
+  if (DstTy.getSizeInBits() < 32)
+    return false;
+  // Check if we're building a constant vector, in which case we want to
+  // generate a constant pool load instead of a vector insert sequence.
+  SmallVector<Constant *, 16> Csts;
+  for (unsigned Idx = 1; Idx < I.getNumOperands(); ++Idx) {
+    // Try to find G_CONSTANT or G_FCONSTANT
+    auto *OpMI =
+        getOpcodeDef(TargetOpcode::G_CONSTANT, I.getOperand(Idx).getReg(), MRI);
+    if (OpMI)
+      Csts.emplace_back(
+          const_cast<ConstantInt *>(OpMI->getOperand(1).getCImm()));
+    else if ((OpMI = getOpcodeDef(TargetOpcode::G_FCONSTANT,
+                                  I.getOperand(Idx).getReg(), MRI)))
+      Csts.emplace_back(
+          const_cast<ConstantFP *>(OpMI->getOperand(1).getFPImm()));
+    else
+      return false;
+  }
+  Constant *CV = ConstantVector::get(Csts);
+  MachineIRBuilder MIB(I);
+  auto *CPLoad = emitLoadFromConstantPool(CV, MIB);
+  if (!CPLoad) {
+    LLVM_DEBUG(dbgs() << "Could not generate cp load for build_vector");
+    return false;
+  }
+  MIB.buildCopy(I.getOperand(0), CPLoad->getOperand(0));
+  RBI.constrainGenericRegister(I.getOperand(0).getReg(),
+                               *MRI.getRegClass(CPLoad->getOperand(0).getReg()),
+                               MRI);
+  I.eraseFromParent();
+  return true;
+}
+
 bool AArch64InstructionSelector::selectBuildVector(
     MachineInstr &I, MachineRegisterInfo &MRI) const {
   assert(I.getOpcode() == TargetOpcode::G_BUILD_VECTOR);
   // Until we port more of the optimized selections, for now just use a vector
   // insert sequence.
   const LLT DstTy = MRI.getType(I.getOperand(0).getReg());
   const LLT EltTy = MRI.getType(I.getOperand(1).getReg());
   unsigned EltSize = EltTy.getSizeInBits();
+
+  if (tryOptConstantBuildVec(I, DstTy, MRI))
+    return true;
   if (EltSize < 16 || EltSize > 64)
     return false; // Don't support all element types yet.
   const RegisterBank &RB = *RBI.getRegBank(I.getOperand(1).getReg(), MRI, TRI);
   MachineIRBuilder MIRBuilder(I);
 
   const TargetRegisterClass *DstRC = &AArch64::FPR128RegClass;
   MachineInstr *ScalarToVec =
       emitScalarToVector(DstTy.getElementType().getSizeInBits(), DstRC,

llvm/test/CodeGen/AArch64/GlobalISel/select-const-vector.mir

+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=aarch64-- -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s
+---
+name:            test_constant_vec_pool_v2f64
+alignment:       4
+legalized:       true
+regBankSelected: true
+tracksRegLiveness: true
+liveins:
+  - { reg: '$x0' }
+frameInfo:
+  maxAlignment:    1
+machineFunctionInfo: {}
+body:             |
+  bb.1:
+    liveins: $x0
+
+    ; CHECK-LABEL: name: test_constant_vec_pool_v2f64
+    ; CHECK: liveins: $x0
+    ; CHECK: [[COPY:%[0-9]+]]:gpr64sp = COPY $x0
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRQui:%[0-9]+]]:fpr128 = LDRQui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: STRQui [[LDRQui]], [[COPY]], 0 :: (store 16)
+    ; CHECK: RET_ReallyLR
+    %0:gpr(p0) = COPY $x0
+    %3:fpr(s64) = G_FCONSTANT double 5.000000e-01
+    %2:fpr(s64) = G_FCONSTANT double 1.600000e+01
+    %1:fpr(<2 x s64>) = G_BUILD_VECTOR %2(s64), %3(s64)
+    G_STORE %1(<2 x s64>), %0(p0) :: (store 16)
+    RET_ReallyLR
+
+...
+---
+name:            test_constant_vec_pool_v4f32
+alignment:       4
+legalized:       true
+regBankSelected: true
+tracksRegLiveness: true
+liveins:
+  - { reg: '$x0' }
+frameInfo:
+  maxAlignment:    1
+body:             |
+  bb.1:
+    liveins: $x0
+
+    ; CHECK-LABEL: name: test_constant_vec_pool_v4f32
+    ; CHECK: liveins: $x0
+    ; CHECK: [[COPY:%[0-9]+]]:gpr64sp = COPY $x0
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRQui:%[0-9]+]]:fpr128 = LDRQui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: STRQui [[LDRQui]], [[COPY]], 0 :: (store 16)
+    ; CHECK: RET_ReallyLR
+    %0:gpr(p0) = COPY $x0
+    %3:fpr(s32) = G_FCONSTANT float 5.000000e-01
+    %2:fpr(s32) = G_FCONSTANT float 1.600000e+01
+    %1:fpr(<4 x s32>) = G_BUILD_VECTOR %2(s32), %3(s32), %3(s32), %3(s32)
+    G_STORE %1(<4 x s32>), %0(p0) :: (store 16)
+    RET_ReallyLR
+
+...
+---
+name:            test_constant_vec_pool_v2i64
+alignment:       4
+legalized:       true
+regBankSelected: true
+tracksRegLiveness: true
+liveins:
+  - { reg: '$x0' }
+frameInfo:
+  maxAlignment:    1
+body:             |
+  bb.1:
+    liveins: $x0
+
+    ; CHECK-LABEL: name: test_constant_vec_pool_v2i64
+    ; CHECK: liveins: $x0
+    ; CHECK: [[COPY:%[0-9]+]]:gpr64sp = COPY $x0
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRQui:%[0-9]+]]:fpr128 = LDRQui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: STRQui [[LDRQui]], [[COPY]], 0 :: (store 16)
+    ; CHECK: RET_ReallyLR
+    %0:gpr(p0) = COPY $x0
+    %3:gpr(s64) = G_CONSTANT i64 67839
+    %2:gpr(s64) = G_CONSTANT i64 12375
+    %1:fpr(<2 x s64>) = G_BUILD_VECTOR %2(s64), %3(s64)
+    G_STORE %1(<2 x s64>), %0(p0) :: (store 16)
+    RET_ReallyLR
+
+...
+---
+name:            test_constant_vec_pool_v4i32
+alignment:       4
+legalized:       true
+regBankSelected: true
+tracksRegLiveness: true
+liveins:
+  - { reg: '$x0' }
+frameInfo:
+  maxAlignment:    1
+body:             |
+  bb.1:
+    liveins: $x0
+
+    ; CHECK-LABEL: name: test_constant_vec_pool_v4i32
+    ; CHECK: liveins: $x0
+    ; CHECK: [[COPY:%[0-9]+]]:gpr64sp = COPY $x0
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRQui:%[0-9]+]]:fpr128 = LDRQui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: STRQui [[LDRQui]], [[COPY]], 0 :: (store 16)
+    ; CHECK: RET_ReallyLR
+    %0:gpr(p0) = COPY $x0
+    %3:gpr(s32) = G_CONSTANT i32 67839
+    %2:gpr(s32) = G_CONSTANT i32 12375
+    %1:fpr(<4 x s32>) = G_BUILD_VECTOR %2(s32), %3(s32), %2(s32), %3(s32)
+    G_STORE %1(<4 x s32>), %0(p0) :: (store 16)
+    RET_ReallyLR
+
+...
+
+---
+name:            test_constant_vec_pool_v2i32
+alignment:       4
+legalized:       true
+regBankSelected: true
+tracksRegLiveness: true
+liveins:
+  - { reg: '$x0' }
+frameInfo:
+  maxAlignment:    1
+body:             |
+  bb.1:
+    liveins: $x0
+
+    ; CHECK-LABEL: name: test_constant_vec_pool_v2i32
+    ; CHECK: liveins: $x0
+    ; CHECK: [[COPY:%[0-9]+]]:gpr64sp = COPY $x0
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: STRDui [[LDRDui]], [[COPY]], 0 :: (store 8)
+    ; CHECK: RET_ReallyLR
+    %0:gpr(p0) = COPY $x0
+    %3:gpr(s32) = G_CONSTANT i32 67839
+    %2:gpr(s32) = G_CONSTANT i32 12375
+    %1:fpr(<2 x s32>) = G_BUILD_VECTOR %2(s32), %3(s32)
+    G_STORE %1(<2 x s32>), %0(p0) :: (store 8)
+    RET_ReallyLR
+
+...

llvm/test/CodeGen/AArch64/GlobalISel/select-vector-icmp.mir

 body:             |
   bb.1 (%ir-block.0):
     liveins: $d0, $d1
 
     ; CHECK-LABEL: name: test_v2i16_eq
     ; CHECK: liveins: $d0, $d1
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
     ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY $d1
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY2:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
-    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[COPY2]]
-    ; CHECK: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF1:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG1:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF1]], [[MOVi32imm1]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr1:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG1]], 1, [[MOVi32imm1]]
-    ; CHECK: [[COPY3:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr1]].dsub
-    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[COPY3]]
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[LDRDui]]
+    ; CHECK: [[ADRP1:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui1:%[0-9]+]]:fpr64 = LDRDui [[ADRP1]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[LDRDui1]]
     ; CHECK: [[CMEQv2i32_:%[0-9]+]]:fpr64 = CMEQv2i32 [[ANDv8i8_]], [[ANDv8i8_1]]
     ; CHECK: $d0 = COPY [[CMEQv2i32_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %2:fpr(<2 x s32>) = COPY $d0
     %3:fpr(<2 x s32>) = COPY $d1
     %13:gpr(s32) = G_CONSTANT i32 65535
     %14:fpr(<2 x s32>) = G_BUILD_VECTOR %13(s32), %13(s32)
     %15:fpr(<2 x s32>) = COPY %2(<2 x s32>)

 body:             |
   bb.1 (%ir-block.0):
     liveins: $d0, $d1
 
     ; CHECK-LABEL: name: test_v2i16_ne
     ; CHECK: liveins: $d0, $d1
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
     ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY $d1
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY2:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
-    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[COPY2]]
-    ; CHECK: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF1:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG1:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF1]], [[MOVi32imm1]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr1:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG1]], 1, [[MOVi32imm1]]
-    ; CHECK: [[COPY3:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr1]].dsub
-    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[COPY3]]
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[LDRDui]]
+    ; CHECK: [[ADRP1:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui1:%[0-9]+]]:fpr64 = LDRDui [[ADRP1]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[LDRDui1]]
     ; CHECK: [[CMEQv2i32_:%[0-9]+]]:fpr64 = CMEQv2i32 [[ANDv8i8_]], [[ANDv8i8_1]]
     ; CHECK: [[NOTv8i8_:%[0-9]+]]:fpr64 = NOTv8i8 [[CMEQv2i32_]]
     ; CHECK: $d0 = COPY [[NOTv8i8_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %2:fpr(<2 x s32>) = COPY $d0
     %3:fpr(<2 x s32>) = COPY $d1
     %13:gpr(s32) = G_CONSTANT i32 65535
     %14:fpr(<2 x s32>) = G_BUILD_VECTOR %13(s32), %13(s32)

 body:             |
   bb.1 (%ir-block.0):
     liveins: $d0, $d1
 
     ; CHECK-LABEL: name: test_v2i16_ugt
     ; CHECK: liveins: $d0, $d1
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
     ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY $d1
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY2:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
-    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[COPY2]]
-    ; CHECK: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF1:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG1:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF1]], [[MOVi32imm1]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr1:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG1]], 1, [[MOVi32imm1]]
-    ; CHECK: [[COPY3:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr1]].dsub
-    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[COPY3]]
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[LDRDui]]
+    ; CHECK: [[ADRP1:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui1:%[0-9]+]]:fpr64 = LDRDui [[ADRP1]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[LDRDui1]]
     ; CHECK: [[CMHIv2i32_:%[0-9]+]]:fpr64 = CMHIv2i32 [[ANDv8i8_]], [[ANDv8i8_1]]
     ; CHECK: $d0 = COPY [[CMHIv2i32_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %2:fpr(<2 x s32>) = COPY $d0
     %3:fpr(<2 x s32>) = COPY $d1
     %13:gpr(s32) = G_CONSTANT i32 65535
     %14:fpr(<2 x s32>) = G_BUILD_VECTOR %13(s32), %13(s32)
     %15:fpr(<2 x s32>) = COPY %2(<2 x s32>)

 body:             |
   bb.1 (%ir-block.0):
     liveins: $d0, $d1
 
     ; CHECK-LABEL: name: test_v2i16_uge
     ; CHECK: liveins: $d0, $d1
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
     ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY $d1
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY2:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
-    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[COPY2]]
-    ; CHECK: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF1:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG1:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF1]], [[MOVi32imm1]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr1:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG1]], 1, [[MOVi32imm1]]
-    ; CHECK: [[COPY3:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr1]].dsub
-    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[COPY3]]
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[LDRDui]]
+    ; CHECK: [[ADRP1:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui1:%[0-9]+]]:fpr64 = LDRDui [[ADRP1]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[LDRDui1]]
     ; CHECK: [[CMHSv2i32_:%[0-9]+]]:fpr64 = CMHSv2i32 [[ANDv8i8_]], [[ANDv8i8_1]]
     ; CHECK: $d0 = COPY [[CMHSv2i32_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %2:fpr(<2 x s32>) = COPY $d0
     %3:fpr(<2 x s32>) = COPY $d1
     %13:gpr(s32) = G_CONSTANT i32 65535
     %14:fpr(<2 x s32>) = G_BUILD_VECTOR %13(s32), %13(s32)
     %15:fpr(<2 x s32>) = COPY %2(<2 x s32>)

 body:             |
   bb.1 (%ir-block.0):
     liveins: $d0, $d1
 
     ; CHECK-LABEL: name: test_v2i16_ult
     ; CHECK: liveins: $d0, $d1
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
     ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY $d1
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY2:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
-    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[COPY2]]
-    ; CHECK: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF1:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG1:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF1]], [[MOVi32imm1]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr1:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG1]], 1, [[MOVi32imm1]]
-    ; CHECK: [[COPY3:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr1]].dsub
-    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[COPY3]]
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[LDRDui]]
+    ; CHECK: [[ADRP1:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui1:%[0-9]+]]:fpr64 = LDRDui [[ADRP1]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[LDRDui1]]
     ; CHECK: [[CMHIv2i32_:%[0-9]+]]:fpr64 = CMHIv2i32 [[ANDv8i8_1]], [[ANDv8i8_]]
     ; CHECK: $d0 = COPY [[CMHIv2i32_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %2:fpr(<2 x s32>) = COPY $d0
     %3:fpr(<2 x s32>) = COPY $d1
     %13:gpr(s32) = G_CONSTANT i32 65535
     %14:fpr(<2 x s32>) = G_BUILD_VECTOR %13(s32), %13(s32)
     %15:fpr(<2 x s32>) = COPY %2(<2 x s32>)

 body:             |
   bb.1 (%ir-block.0):
     liveins: $d0, $d1
 
     ; CHECK-LABEL: name: test_v2i16_ule
     ; CHECK: liveins: $d0, $d1
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
     ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY $d1
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY2:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
-    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[COPY2]]
-    ; CHECK: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 65535
-    ; CHECK: [[DEF1:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG1:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF1]], [[MOVi32imm1]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr1:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG1]], 1, [[MOVi32imm1]]
-    ; CHECK: [[COPY3:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr1]].dsub
-    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[COPY3]]
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY]], [[LDRDui]]
+    ; CHECK: [[ADRP1:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui1:%[0-9]+]]:fpr64 = LDRDui [[ADRP1]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[ANDv8i8_1:%[0-9]+]]:fpr64 = ANDv8i8 [[COPY1]], [[LDRDui1]]
     ; CHECK: [[CMHSv2i32_:%[0-9]+]]:fpr64 = CMHSv2i32 [[ANDv8i8_1]], [[ANDv8i8_]]
     ; CHECK: $d0 = COPY [[CMHSv2i32_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %2:fpr(<2 x s32>) = COPY $d0
     %3:fpr(<2 x s32>) = COPY $d1
     %13:gpr(s32) = G_CONSTANT i32 65535
     %14:fpr(<2 x s32>) = G_BUILD_VECTOR %13(s32), %13(s32)
     %15:fpr(<2 x s32>) = COPY %2(<2 x s32>)

 body:             |
   bb.1 (%ir-block.0):
     liveins: $d0, $d1
 
     ; CHECK-LABEL: name: test_v2i16_sgt
     ; CHECK: liveins: $d0, $d1
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
     ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY $d1
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 16
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY2:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
     ; CHECK: [[SHLv2i32_shift:%[0-9]+]]:fpr64 = SHLv2i32_shift [[COPY]], 16
-    ; CHECK: [[NEGv2i32_:%[0-9]+]]:fpr64 = NEGv2i32 [[COPY2]]
+    ; CHECK: [[NEGv2i32_:%[0-9]+]]:fpr64 = NEGv2i32 [[LDRDui]]
     ; CHECK: [[SSHLv2i32_:%[0-9]+]]:fpr64 = SSHLv2i32 [[SHLv2i32_shift]], [[NEGv2i32_]]
-    ; CHECK: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 16
-    ; CHECK: [[DEF1:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG1:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF1]], [[MOVi32imm1]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr1:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG1]], 1, [[MOVi32imm1]]
-    ; CHECK: [[COPY3:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr1]].dsub
+    ; CHECK: [[ADRP1:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui1:%[0-9]+]]:fpr64 = LDRDui [[ADRP1]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
     ; CHECK: [[SHLv2i32_shift1:%[0-9]+]]:fpr64 = SHLv2i32_shift [[COPY1]], 16
-    ; CHECK: [[NEGv2i32_1:%[0-9]+]]:fpr64 = NEGv2i32 [[COPY3]]
+    ; CHECK: [[NEGv2i32_1:%[0-9]+]]:fpr64 = NEGv2i32 [[LDRDui1]]
     ; CHECK: [[SSHLv2i32_1:%[0-9]+]]:fpr64 = SSHLv2i32 [[SHLv2i32_shift1]], [[NEGv2i32_1]]
     ; CHECK: [[CMGTv2i32_:%[0-9]+]]:fpr64 = CMGTv2i32 [[SSHLv2i32_]], [[SSHLv2i32_1]]
     ; CHECK: $d0 = COPY [[CMGTv2i32_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %2:fpr(<2 x s32>) = COPY $d0
     %3:fpr(<2 x s32>) = COPY $d1
     %14:gpr(s32) = G_CONSTANT i32 16
     %15:fpr(<2 x s32>) = G_BUILD_VECTOR %14(s32), %14(s32)

 body:             |
   bb.1 (%ir-block.0):
     liveins: $d0, $d1
 
     ; CHECK-LABEL: name: test_v2i16_sge
     ; CHECK: liveins: $d0, $d1
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
     ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY $d1
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 16
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY2:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
     ; CHECK: [[SHLv2i32_shift:%[0-9]+]]:fpr64 = SHLv2i32_shift [[COPY]], 16
-    ; CHECK: [[NEGv2i32_:%[0-9]+]]:fpr64 = NEGv2i32 [[COPY2]]
+    ; CHECK: [[NEGv2i32_:%[0-9]+]]:fpr64 = NEGv2i32 [[LDRDui]]
     ; CHECK: [[SSHLv2i32_:%[0-9]+]]:fpr64 = SSHLv2i32 [[SHLv2i32_shift]], [[NEGv2i32_]]
-    ; CHECK: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 16
-    ; CHECK: [[DEF1:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG1:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF1]], [[MOVi32imm1]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr1:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG1]], 1, [[MOVi32imm1]]
-    ; CHECK: [[COPY3:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr1]].dsub
+    ; CHECK: [[ADRP1:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui1:%[0-9]+]]:fpr64 = LDRDui [[ADRP1]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
     ; CHECK: [[SHLv2i32_shift1:%[0-9]+]]:fpr64 = SHLv2i32_shift [[COPY1]], 16
-    ; CHECK: [[NEGv2i32_1:%[0-9]+]]:fpr64 = NEGv2i32 [[COPY3]]
+    ; CHECK: [[NEGv2i32_1:%[0-9]+]]:fpr64 = NEGv2i32 [[LDRDui1]]
     ; CHECK: [[SSHLv2i32_1:%[0-9]+]]:fpr64 = SSHLv2i32 [[SHLv2i32_shift1]], [[NEGv2i32_1]]
     ; CHECK: [[CMGEv2i32_:%[0-9]+]]:fpr64 = CMGEv2i32 [[SSHLv2i32_]], [[SSHLv2i32_1]]
     ; CHECK: $d0 = COPY [[CMGEv2i32_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %2:fpr(<2 x s32>) = COPY $d0
     %3:fpr(<2 x s32>) = COPY $d1
     %14:gpr(s32) = G_CONSTANT i32 16
     %15:fpr(<2 x s32>) = G_BUILD_VECTOR %14(s32), %14(s32)

 body:             |
   bb.1 (%ir-block.0):
     liveins: $d0, $d1
 
     ; CHECK-LABEL: name: test_v2i16_slt
     ; CHECK: liveins: $d0, $d1
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
     ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY $d1
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 16
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY2:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
     ; CHECK: [[SHLv2i32_shift:%[0-9]+]]:fpr64 = SHLv2i32_shift [[COPY]], 16
-    ; CHECK: [[NEGv2i32_:%[0-9]+]]:fpr64 = NEGv2i32 [[COPY2]]
+    ; CHECK: [[NEGv2i32_:%[0-9]+]]:fpr64 = NEGv2i32 [[LDRDui]]
     ; CHECK: [[SSHLv2i32_:%[0-9]+]]:fpr64 = SSHLv2i32 [[SHLv2i32_shift]], [[NEGv2i32_]]
-    ; CHECK: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 16
-    ; CHECK: [[DEF1:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG1:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF1]], [[MOVi32imm1]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr1:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG1]], 1, [[MOVi32imm1]]
-    ; CHECK: [[COPY3:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr1]].dsub
+    ; CHECK: [[ADRP1:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui1:%[0-9]+]]:fpr64 = LDRDui [[ADRP1]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
     ; CHECK: [[SHLv2i32_shift1:%[0-9]+]]:fpr64 = SHLv2i32_shift [[COPY1]], 16
-    ; CHECK: [[NEGv2i32_1:%[0-9]+]]:fpr64 = NEGv2i32 [[COPY3]]
+    ; CHECK: [[NEGv2i32_1:%[0-9]+]]:fpr64 = NEGv2i32 [[LDRDui1]]
     ; CHECK: [[SSHLv2i32_1:%[0-9]+]]:fpr64 = SSHLv2i32 [[SHLv2i32_shift1]], [[NEGv2i32_1]]
     ; CHECK: [[CMGTv2i32_:%[0-9]+]]:fpr64 = CMGTv2i32 [[SSHLv2i32_1]], [[SSHLv2i32_]]
     ; CHECK: $d0 = COPY [[CMGTv2i32_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %2:fpr(<2 x s32>) = COPY $d0
     %3:fpr(<2 x s32>) = COPY $d1
     %14:gpr(s32) = G_CONSTANT i32 16
     %15:fpr(<2 x s32>) = G_BUILD_VECTOR %14(s32), %14(s32)

 body:             |
   bb.1 (%ir-block.0):
     liveins: $d0, $d1
 
     ; CHECK-LABEL: name: test_v2i16_sle
     ; CHECK: liveins: $d0, $d1
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
     ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY $d1
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 16
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY2:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
     ; CHECK: [[SHLv2i32_shift:%[0-9]+]]:fpr64 = SHLv2i32_shift [[COPY]], 16
-    ; CHECK: [[NEGv2i32_:%[0-9]+]]:fpr64 = NEGv2i32 [[COPY2]]
+    ; CHECK: [[NEGv2i32_:%[0-9]+]]:fpr64 = NEGv2i32 [[LDRDui]]
     ; CHECK: [[SSHLv2i32_:%[0-9]+]]:fpr64 = SSHLv2i32 [[SHLv2i32_shift]], [[NEGv2i32_]]
-    ; CHECK: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 16
-    ; CHECK: [[DEF1:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG1:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF1]], [[MOVi32imm1]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr1:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG1]], 1, [[MOVi32imm1]]
-    ; CHECK: [[COPY3:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr1]].dsub
+    ; CHECK: [[ADRP1:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui1:%[0-9]+]]:fpr64 = LDRDui [[ADRP1]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
     ; CHECK: [[SHLv2i32_shift1:%[0-9]+]]:fpr64 = SHLv2i32_shift [[COPY1]], 16
-    ; CHECK: [[NEGv2i32_1:%[0-9]+]]:fpr64 = NEGv2i32 [[COPY3]]
+    ; CHECK: [[NEGv2i32_1:%[0-9]+]]:fpr64 = NEGv2i32 [[LDRDui1]]
     ; CHECK: [[SSHLv2i32_1:%[0-9]+]]:fpr64 = SSHLv2i32 [[SHLv2i32_shift1]], [[NEGv2i32_1]]
     ; CHECK: [[CMGEv2i32_:%[0-9]+]]:fpr64 = CMGEv2i32 [[SSHLv2i32_1]], [[SSHLv2i32_]]
     ; CHECK: $d0 = COPY [[CMGEv2i32_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %2:fpr(<2 x s32>) = COPY $d0
     %3:fpr(<2 x s32>) = COPY $d1
     %14:gpr(s32) = G_CONSTANT i32 16
     %15:fpr(<2 x s32>) = G_BUILD_VECTOR %14(s32), %14(s32)

llvm/test/CodeGen/AArch64/GlobalISel/select-vector-shift.mir

 machineFunctionInfo: {}
 body:             |
   bb.1:
     liveins: $d0
 
     ; CHECK-LABEL: name: shl_v2i32_imm_out_of_range
     ; CHECK: liveins: $d0
     ; CHECK: [[COPY:%[0-9]+]]:fpr64 = COPY $d0
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 40
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[MOVi32imm]], %subreg.ssub
-    ; CHECK: [[INSvi32gpr:%[0-9]+]]:fpr128 = INSvi32gpr [[INSERT_SUBREG]], 1, [[MOVi32imm]]
-    ; CHECK: [[COPY1:%[0-9]+]]:fpr64 = COPY [[INSvi32gpr]].dsub
-    ; CHECK: [[USHLv2i32_:%[0-9]+]]:fpr64 = USHLv2i32 [[COPY]], [[COPY1]]
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRDui:%[0-9]+]]:fpr64 = LDRDui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[USHLv2i32_:%[0-9]+]]:fpr64 = USHLv2i32 [[COPY]], [[LDRDui]]
     ; CHECK: $d0 = COPY [[USHLv2i32_]]
     ; CHECK: RET_ReallyLR implicit $d0
     %0:fpr(<2 x s32>) = COPY $d0
     %2:gpr(s32) = G_CONSTANT i32 40
     %1:fpr(<2 x s32>) = G_BUILD_VECTOR %2(s32), %2(s32)
     %3:fpr(<2 x s32>) = G_SHL %0, %1(<2 x s32>)
     $d0 = COPY %3(<2 x s32>)
     RET_ReallyLR implicit $d0

 machineFunctionInfo: {}
 body:             |
   bb.1:
     liveins: $q0
 
     ; CHECK-LABEL: name: shl_v2i64_imm_out_of_range
     ; CHECK: liveins: $q0
     ; CHECK: [[COPY:%[0-9]+]]:fpr128 = COPY $q0
-    ; CHECK: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 70
-    ; CHECK: [[SUBREG_TO_REG:%[0-9]+]]:gpr64 = SUBREG_TO_REG 0, [[MOVi32imm]], %subreg.sub_32
-    ; CHECK: [[DEF:%[0-9]+]]:fpr128 = IMPLICIT_DEF
-    ; CHECK: [[INSERT_SUBREG:%[0-9]+]]:fpr128 = INSERT_SUBREG [[DEF]], [[SUBREG_TO_REG]], %subreg.dsub
-    ; CHECK: [[INSvi64gpr:%[0-9]+]]:fpr128 = INSvi64gpr [[INSERT_SUBREG]], 1, [[SUBREG_TO_REG]]
-    ; CHECK: [[USHLv2i64_:%[0-9]+]]:fpr128 = USHLv2i64 [[COPY]], [[INSvi64gpr]]
+    ; CHECK: [[ADRP:%[0-9]+]]:gpr64common = ADRP target-flags(aarch64-page) %const.0
+    ; CHECK: [[LDRQui:%[0-9]+]]:fpr128 = LDRQui [[ADRP]], target-flags(aarch64-pageoff, aarch64-nc) %const.0
+    ; CHECK: [[USHLv2i64_:%[0-9]+]]:fpr128 = USHLv2i64 [[COPY]], [[LDRQui]]
     ; CHECK: $q0 = COPY [[USHLv2i64_]]
     ; CHECK: RET_ReallyLR implicit $q0
     %0:fpr(<2 x s64>) = COPY $q0
     %2:gpr(s64) = G_CONSTANT i64 70
     %1:fpr(<2 x s64>) = G_BUILD_VECTOR %2(s64), %2(s64)
     %3:fpr(<2 x s64>) = G_SHL %0, %1(<2 x s64>)
     $q0 = COPY %3(<2 x s64>)
     RET_ReallyLR implicit $q0

llvm/test/CodeGen/AArch64/combine-loads.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -O0 -mtriple=aarch64-unknown-unknown | FileCheck %s
 
 define <2 x i64> @z(i64* nocapture nonnull readonly %p) {
 ; CHECK-LABEL: z:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    mov x8, xzr
-; CHECK-NEXT:    // implicit-def: $q0
-; CHECK-NEXT:    fmov d0, x8
-; CHECK-NEXT:    mov v0.d[1], x8
+; CHECK-NEXT:    adrp x8, .LCPI0_0
+; CHECK-NEXT:    ldr q0, [x8, :lo12:.LCPI0_0]
 ; CHECK-NEXT:    ldr x8, [x0]
 ; CHECK-NEXT:    ldr x9, [x0, #8]
 ; CHECK-NEXT:    mov v0.d[0], x8
 ; CHECK-NEXT:    mov v0.d[1], x9
 ; CHECK-NEXT:    ret
   %b = load i64, i64* %p
   %p2 = getelementptr i64, i64* %p, i64 1
   %bb = load i64, i64* %p2
   %r1 = insertelement <2 x i64> zeroinitializer, i64 %b, i32 0
   %r2 = insertelement <2 x i64> %r1, i64 %bb, i32 1
   ret <2 x i64> %r2
 }