[LoongArch] Add support for the BranchRelaxation pass

[XiaodongLoong]

Summary

When the branch target is out of the range represented by the current
branch instruction's immediate, branch relaxation is required. There
are three types of immediate for branch instructions on LoongArch,
including simm16, simm21 and simm26. And the real branch target
address is PC + sext(simmXX << 2). In addition, the indirect branch
way is implemented to support larger branch target.

BranchRelaxation pass calls RenumberBlocks to renumber all of the
machine basic blocks in the function. So the machine basic blocks
number changed in some test cases.

[arsenm]

llvm/lib/Target/LoongArch/LoongArchInstrInfo.cpp

265

MC can handle relaxing these different immediate offset types; you probably only want to handle the > 28 bit case here (and even then there may be a better way to deal with it which I don't know about it, if you have dedicated registers for it)

[SixWeining]

About the title, you'd better to add a [LoongArch] prefix.

llvm/lib/Target/LoongArch/LoongArchInstrInfo.cpp

377

How about when assert = off? Will generate wrong code if Scav == LoongArch::NoRegister?

[XiaodongLoong]

rebase code

[XiaodongLoong]

llvm/lib/Target/LoongArch/LoongArchInstrInfo.cpp

265

MC can handle relaxing these different immediate offset types; you probably only want to handle the > 28 bit case here (and even then there may be a better way to deal with it which I don't know about it, if you have dedicated registers for it)

The BranchRelaxation pass could handle both condition branch and uncondition branch after I read the code. And I also read the patch from RISCV https://reviews.llvm.org/D40830. I think it might be OK to handle branch relaxing here on LoongArch. As far as I can see, there is no dedicated registers on LoongArch.
Thanks!

377

How about when assert = off? Will generate wrong code if Scav == LoongArch::NoRegister?

I will handle the condition in other patch. Thanks!

[XiaodongLoong]

rebase code

[StephenFan]

llvm/test/CodeGen/LoongArch/branch-relaxation.ll

4

Is it possible to use -filetype=null if the output is not needed?

[XiaodongLoong]

llvm/test/CodeGen/LoongArch/branch-relaxation.ll

4

Is it possible to use -filetype=null if the output is not needed?

Sorry, it is not possible to use -filetype=null.
When I use -filetype=null, I can not reproduce the errors I want to fix showed as following:

<unknown>:0: error: fixup value out of range [-131072, 131071]
<unknown>:0: error: fixup value out of range [-4194304, 4194303]
<unknown>:0: error: fixup value out of range [-4194304, 4194303]

[SixWeining]

LGTM except a small nit.

llvm/lib/Target/LoongArch/LoongArchInstrInfo.cpp

160

We can directly use MI.getOpcode() here (at least for now).

[XiaodongLoong]

rebase code and fix nits.

[XiaodongLoong]

llvm/lib/Target/LoongArch/LoongArchInstrInfo.cpp

160

Thanks! I changed it.

llvm/lib/Target/LoongArch/LoongArchInstrInfo.h

public:
 
   MachineBasicBlock *getBranchDestBlock(const MachineInstr &MI) const override;
 
   bool analyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
                      MachineBasicBlock *&FBB,
                      SmallVectorImpl<MachineOperand> &Cond,
                      bool AllowModify) const override;
 
+  bool isBranchOffsetInRange(unsigned BranchOpc,
+                             int64_t BrOffset) const override;
+
   unsigned removeBranch(MachineBasicBlock &MBB,
                         int *BytesRemoved = nullptr) const override;
 
   unsigned insertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
                         MachineBasicBlock *FBB, ArrayRef<MachineOperand> Cond,
                         const DebugLoc &dl,
                         int *BytesAdded = nullptr) const override;
 
+  void insertIndirectBranch(MachineBasicBlock &MBB,
+                            MachineBasicBlock &NewDestBB,
+                            MachineBasicBlock &RestoreBB, const DebugLoc &DL,
+                            int64_t BrOffset, RegScavenger *RS) const override;
+
   bool
   reverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const override;
 
 protected:
   const LoongArchSubtarget &STI;
 };
 
 } // end namespace llvm
 #endif // LLVM_LIB_TARGET_LOONGARCH_LOONGARCHINSTRINFO_H

llvm/lib/Target/LoongArch/LoongArchInstrInfo.cpp

 //=- LoongArchInstrInfo.cpp - LoongArch Instruction Information -*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file contains the LoongArch implementation of the TargetInstrInfo class.
 //
 //===----------------------------------------------------------------------===//
 
 #include "LoongArchInstrInfo.h"
 #include "LoongArch.h"
 #include "LoongArchMachineFunctionInfo.h"
 #include "MCTargetDesc/LoongArchMatInt.h"
+#include "llvm/CodeGen/RegisterScavenging.h"
 
 using namespace llvm;
 
 #define GET_INSTRINFO_CTOR_DTOR
 #include "LoongArchGenInstrInfo.inc"
 
 LoongArchInstrInfo::LoongArchInstrInfo(LoongArchSubtarget &STI)
     : LoongArchGenInstrInfo(LoongArch::ADJCALLSTACKDOWN,
for (auto &Inst : Seq) {
     }
 
     // Only the first instruction has $zero as its source.
     SrcReg = DstReg;
   }
 }
 
 unsigned LoongArchInstrInfo::getInstSizeInBytes(const MachineInstr &MI) const {
+  if (MI.getOpcode() == TargetOpcode::INLINEASM) {
+    const MachineFunction *MF = MI.getParent()->getParent();
+    const MCAsmInfo *MAI = MF->getTarget().getMCAsmInfo();
+    return getInlineAsmLength(MI.getOperand(0).getSymbolName(), *MAI);
+  }
   return MI.getDesc().getSize();
 }
 
 MachineBasicBlock *
 LoongArchInstrInfo::getBranchDestBlock(const MachineInstr &MI) const {
   assert(MI.getDesc().isBranch() && "Unexpected opcode!");
   // The branch target is always the last operand.
   return MI.getOperand(MI.getNumExplicitOperands() - 1).getMBB();
if (NumTerminators == 2 && std::prev(I)->getDesc().isConditionalBranch() &&
     FBB = getBranchDestBlock(*I);
     return false;
   }
 
   // Otherwise, we can't handle this.
   return true;
 }
 
+bool LoongArchInstrInfo::isBranchOffsetInRange(unsigned BranchOp,
+                                               int64_t BrOffset) const {
+  switch (BranchOp) {
+  default:
+    llvm_unreachable("Unknown branch instruction!");
+  case LoongArch::BEQ:
+  case LoongArch::BNE:
+  case LoongArch::BLT:
+  case LoongArch::BGE:
+  case LoongArch::BLTU:
+  case LoongArch::BGEU:
+    return isInt<18>(BrOffset);
+  case LoongArch::BEQZ:
+  case LoongArch::BNEZ:
+  case LoongArch::BCEQZ:
+  case LoongArch::BCNEZ:
+    return isInt<23>(BrOffset);
+  case LoongArch::B:
+  case LoongArch::PseudoBR:
+    return isInt<28>(BrOffset);
+  }
+}
+
 unsigned LoongArchInstrInfo::removeBranch(MachineBasicBlock &MBB,
                                           int *BytesRemoved) const {
   if (BytesRemoved)
     *BytesRemoved = 0;
   MachineBasicBlock::iterator I = MBB.getLastNonDebugInstr();
   if (I == MBB.end())
     return 0;
 
unsigned LoongArchInstrInfo::insertBranch(
 
   // Two-way conditional branch.
   MachineInstr &MI = *BuildMI(&MBB, DL, get(LoongArch::PseudoBR)).addMBB(FBB);
   if (BytesAdded)
     *BytesAdded += getInstSizeInBytes(MI);
   return 2;
 }
 
+void LoongArchInstrInfo::insertIndirectBranch(MachineBasicBlock &MBB,
+                                              MachineBasicBlock &DestBB,
+                                              MachineBasicBlock &RestoreBB,
+                                              const DebugLoc &DL,
+                                              int64_t BrOffset,
+                                              RegScavenger *RS) const {
+  assert(RS && "RegScavenger required for long branching");
+  assert(MBB.empty() &&
+         "new block should be inserted for expanding unconditional branch");
+  assert(MBB.pred_size() == 1);
+
+  MachineFunction *MF = MBB.getParent();
+  MachineRegisterInfo &MRI = MF->getRegInfo();
+
+  if (!isInt<32>(BrOffset))
+    report_fatal_error(
+        "Branch offsets outside of the signed 32-bit range not supported");
+
+  Register ScratchReg = MRI.createVirtualRegister(&LoongArch::GPRRegClass);
+  auto II = MBB.end();
+
+  MachineInstr &MI =
+      *BuildMI(MBB, II, DL, get(LoongArch::PCALAU12I), ScratchReg)
+           .addMBB(&DestBB, LoongArchII::MO_PCREL_HI);
+  BuildMI(MBB, II, DL,
+          get(STI.is64Bit() ? LoongArch::ADDI_D : LoongArch::ADDI_W),
+          ScratchReg)
+      .addReg(ScratchReg)
+      .addMBB(&DestBB, LoongArchII::MO_PCREL_LO);
+  BuildMI(MBB, II, DL, get(LoongArch::PseudoBRIND))
+      .addReg(ScratchReg, RegState::Kill)
+      .addImm(0);
+
+  RS->enterBasicBlockEnd(MBB);
+  Register Scav = RS->scavengeRegisterBackwards(LoongArch::GPRRegClass,
+                                                MI.getIterator(), false, 0);
+  // TODO: When there is no scavenged register, it needs to specify a register.
+  assert(Scav != LoongArch::NoRegister && "No register is scavenged!");
+  MRI.replaceRegWith(ScratchReg, Scav);
+  MRI.clearVirtRegs();
+  RS->setRegUsed(Scav);
+}
+
 static unsigned getOppositeBranchOpc(unsigned Opc) {
   switch (Opc) {
   default:
     llvm_unreachable("Unrecognized conditional branch");
   case LoongArch::BEQ:
     return LoongArch::BNE;
   case LoongArch::BNE:
     return LoongArch::BEQ;

llvm/lib/Target/LoongArch/LoongArchTargetMachine.cpp

LoongArchPassConfig(LoongArchTargetMachine &TM, PassManagerBase &PM)
       : TargetPassConfig(TM, PM) {}
 
   LoongArchTargetMachine &getLoongArchTargetMachine() const {
     return getTM<LoongArchTargetMachine>();
   }
 
   void addIRPasses() override;
   bool addInstSelector() override;
+  void addPreEmitPass() override;
   void addPreEmitPass2() override;
   void addPreRegAlloc() override;
 };
 } // end namespace
 
 TargetPassConfig *
 LoongArchTargetMachine::createPassConfig(PassManagerBase &PM) {
   return new LoongArchPassConfig(*this, PM);
 }
 
 void LoongArchPassConfig::addIRPasses() {
   addPass(createAtomicExpandPass());
 
   TargetPassConfig::addIRPasses();
 }
 
 bool LoongArchPassConfig::addInstSelector() {
   addPass(createLoongArchISelDag(getLoongArchTargetMachine()));
 
   return false;
 }
 
+void LoongArchPassConfig::addPreEmitPass() { addPass(&BranchRelaxationPassID); }
+
 void LoongArchPassConfig::addPreEmitPass2() {
   // Schedule the expansion of AtomicPseudos at the last possible moment,
   // avoiding the possibility for other passes to break the requirements for
   // forward progress in the LL/SC block.
   addPass(createLoongArchExpandAtomicPseudoPass());
 }
 
 void LoongArchPassConfig::addPreRegAlloc() {
   addPass(createLoongArchPreRAExpandPseudoPass());
 }

llvm/test/CodeGen/LoongArch/analyze-branch.ll

 }
 
 define void @test_bcc_fallthrough_nottaken(i64 %in) nounwind {
 ; CHECK-LABEL: test_bcc_fallthrough_nottaken:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi.d $sp, $sp, -16
 ; CHECK-NEXT:    st.d $ra, $sp, 8 # 8-byte Folded Spill
 ; CHECK-NEXT:    ori $a1, $zero, 42
-; CHECK-NEXT:    beq $a0, $a1, .LBB1_1
-; CHECK-NEXT:  # %bb.3: # %false
+; CHECK-NEXT:    beq $a0, $a1, .LBB1_3
+; CHECK-NEXT:  # %bb.1: # %false
 ; CHECK-NEXT:    bl %plt(test_false)
 ; CHECK-NEXT:  .LBB1_2: # %true
 ; CHECK-NEXT:    ld.d $ra, $sp, 8 # 8-byte Folded Reload
 ; CHECK-NEXT:    addi.d $sp, $sp, 16
 ; CHECK-NEXT:    ret
-; CHECK-NEXT:  .LBB1_1: # %true
+; CHECK-NEXT:  .LBB1_3: # %true
 ; CHECK-NEXT:    bl %plt(test_true)
 ; CHECK-NEXT:    b .LBB1_2
   %tst = icmp eq i64 %in, 42
   br i1 %tst, label %true, label %false, !prof !1
 
 ;; Expected layout order is: Entry, FalseBlock, TrueBlock
 ;; Entry->FalseBlock is the common path, which should be taken whenever the
 ;; conditional branch is false.

llvm/test/CodeGen/LoongArch/branch-relaxation.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc --mtriple=loongarch32 --filetype=obj --verify-machineinstrs < %s \
+; RUN:   -o /dev/null 2>&1
+; RUN: llc --mtriple=loongarch64 --filetype=obj --verify-machineinstrs < %s \
+; RUN:   -o /dev/null 2>&1
+; RUN: llc --mtriple=loongarch32 --verify-machineinstrs < %s | FileCheck %s --check-prefix=LA32
+; RUN: llc --mtriple=loongarch64 --verify-machineinstrs < %s | FileCheck %s --check-prefix=LA64
+
+define i32 @relax_b18(i32 signext %a, i32 signext %b) {
+; LA32-LABEL: relax_b18:
+; LA32:       # %bb.0:
+; LA32-NEXT:    beq $a0, $a1, .LBB0_1
+; LA32-NEXT:    b .LBB0_2
+; LA32-NEXT:  .LBB0_1: # %iftrue
+; LA32-NEXT:    #APP
+; LA32-NEXT:    .space 1048576
+; LA32-NEXT:    #NO_APP
+; LA32-NEXT:    ori $a0, $zero, 1
+; LA32-NEXT:    ret
+; LA32-NEXT:  .LBB0_2: # %iffalse
+; LA32-NEXT:    move $a0, $zero
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: relax_b18:
+; LA64:       # %bb.0:
+; LA64-NEXT:    beq $a0, $a1, .LBB0_1
+; LA64-NEXT:    b .LBB0_2
+; LA64-NEXT:  .LBB0_1: # %iftrue
+; LA64-NEXT:    #APP
+; LA64-NEXT:    .space 1048576
+; LA64-NEXT:    #NO_APP
+; LA64-NEXT:    ori $a0, $zero, 1
+; LA64-NEXT:    ret
+; LA64-NEXT:  .LBB0_2: # %iffalse
+; LA64-NEXT:    move $a0, $zero
+; LA64-NEXT:    ret
+  %cond = icmp eq i32 %a, %b
+  br i1 %cond, label %iftrue, label %iffalse
+
+iftrue:
+  call void asm sideeffect ".space 1048576", ""()
+  ret i32 1
+
+iffalse:
+  ret i32 0
+}
+
+define i32 @relax_b23(i1 %a) {
+; LA32-LABEL: relax_b23:
+; LA32:       # %bb.0:
+; LA32-NEXT:    andi $a0, $a0, 1
+; LA32-NEXT:    bnez $a0, .LBB1_1
+; LA32-NEXT:    b .LBB1_2
+; LA32-NEXT:  .LBB1_1: # %iftrue
+; LA32-NEXT:    #APP
+; LA32-NEXT:    .space 16777216
+; LA32-NEXT:    #NO_APP
+; LA32-NEXT:    ori $a0, $zero, 1
+; LA32-NEXT:    ret
+; LA32-NEXT:  .LBB1_2: # %iffalse
+; LA32-NEXT:    move $a0, $zero
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: relax_b23:
+; LA64:       # %bb.0:
+; LA64-NEXT:    andi $a0, $a0, 1
+; LA64-NEXT:    bnez $a0, .LBB1_1
+; LA64-NEXT:    b .LBB1_2
+; LA64-NEXT:  .LBB1_1: # %iftrue
+; LA64-NEXT:    #APP
+; LA64-NEXT:    .space 16777216
+; LA64-NEXT:    #NO_APP
+; LA64-NEXT:    ori $a0, $zero, 1
+; LA64-NEXT:    ret
+; LA64-NEXT:  .LBB1_2: # %iffalse
+; LA64-NEXT:    move $a0, $zero
+; LA64-NEXT:    ret
+  br i1 %a, label %iftrue, label %iffalse
+
+iftrue:
+  call void asm sideeffect ".space 16777216", ""()
+  ret i32 1
+
+iffalse:
+  ret i32 0
+}
+
+define i32 @relax_b28(i1 %a) {
+; LA32-LABEL: relax_b28:
+; LA32:       # %bb.0:
+; LA32-NEXT:    andi $a0, $a0, 1
+; LA32-NEXT:    bnez $a0, .LBB2_1
+; LA32-NEXT:  # %bb.3:
+; LA32-NEXT:    pcalau12i $a0, %pc_hi20(.LBB2_2)
+; LA32-NEXT:    addi.w $a0, $a0, %pc_lo12(.LBB2_2)
+; LA32-NEXT:    jr $a0
+; LA32-NEXT:  .LBB2_1: # %iftrue
+; LA32-NEXT:    #APP
+; LA32-NEXT:    .space 536870912
+; LA32-NEXT:    #NO_APP
+; LA32-NEXT:    ori $a0, $zero, 1
+; LA32-NEXT:    ret
+; LA32-NEXT:  .LBB2_2: # %iffalse
+; LA32-NEXT:    move $a0, $zero
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: relax_b28:
+; LA64:       # %bb.0:
+; LA64-NEXT:    andi $a0, $a0, 1
+; LA64-NEXT:    bnez $a0, .LBB2_1
+; LA64-NEXT:  # %bb.3:
+; LA64-NEXT:    pcalau12i $a0, %pc_hi20(.LBB2_2)
+; LA64-NEXT:    addi.d $a0, $a0, %pc_lo12(.LBB2_2)
+; LA64-NEXT:    jr $a0
+; LA64-NEXT:  .LBB2_1: # %iftrue
+; LA64-NEXT:    #APP
+; LA64-NEXT:    .space 536870912
+; LA64-NEXT:    #NO_APP
+; LA64-NEXT:    ori $a0, $zero, 1
+; LA64-NEXT:    ret
+; LA64-NEXT:  .LBB2_2: # %iffalse
+; LA64-NEXT:    move $a0, $zero
+; LA64-NEXT:    ret
+  br i1 %a, label %iftrue, label %iffalse
+
+iftrue:
+  call void asm sideeffect ".space 536870912", ""()
+  ret i32 1
+
+iffalse:
+  ret i32 0
+}

llvm/test/CodeGen/LoongArch/exception-pointer-register.ll

 ; LA32-NEXT:    .cfi_def_cfa_offset 16
 ; LA32-NEXT:    st.w $ra, $sp, 12 # 4-byte Folded Spill
 ; LA32-NEXT:    st.w $fp, $sp, 8 # 4-byte Folded Spill
 ; LA32-NEXT:    st.w $s0, $sp, 4 # 4-byte Folded Spill
 ; LA32-NEXT:    .cfi_offset 1, -4
 ; LA32-NEXT:    .cfi_offset 22, -8
 ; LA32-NEXT:    .cfi_offset 23, -12
 ; LA32-NEXT:    move $fp, $a0
-; LA32-NEXT:    beqz $a0, .LBB0_1
-; LA32-NEXT:  # %bb.2: # %bb2
+; LA32-NEXT:    beqz $a0, .LBB0_2
+; LA32-NEXT:  # %bb.1: # %bb2
 ; LA32-NEXT:  .Ltmp0:
 ; LA32-NEXT:    move $a0, $fp
 ; LA32-NEXT:    bl %plt(bar)
 ; LA32-NEXT:  .Ltmp1:
 ; LA32-NEXT:    b .LBB0_3
-; LA32-NEXT:  .LBB0_1: # %bb1
+; LA32-NEXT:  .LBB0_2: # %bb1
 ; LA32-NEXT:  .Ltmp2:
 ; LA32-NEXT:    move $a0, $fp
 ; LA32-NEXT:    bl %plt(foo)
 ; LA32-NEXT:  .Ltmp3:
 ; LA32-NEXT:  .LBB0_3: # %end2
 ; LA32-NEXT:    ld.w $s0, $sp, 4 # 4-byte Folded Reload
 ; LA32-NEXT:    ld.w $fp, $sp, 8 # 4-byte Folded Reload
 ; LA32-NEXT:    ld.w $ra, $sp, 12 # 4-byte Folded Reload

 ; LA64-NEXT:    .cfi_def_cfa_offset 32
 ; LA64-NEXT:    st.d $ra, $sp, 24 # 8-byte Folded Spill
 ; LA64-NEXT:    st.d $fp, $sp, 16 # 8-byte Folded Spill
 ; LA64-NEXT:    st.d $s0, $sp, 8 # 8-byte Folded Spill
 ; LA64-NEXT:    .cfi_offset 1, -8
 ; LA64-NEXT:    .cfi_offset 22, -16
 ; LA64-NEXT:    .cfi_offset 23, -24
 ; LA64-NEXT:    move $fp, $a0
-; LA64-NEXT:    beqz $a0, .LBB0_1
-; LA64-NEXT:  # %bb.2: # %bb2
+; LA64-NEXT:    beqz $a0, .LBB0_2
+; LA64-NEXT:  # %bb.1: # %bb2
 ; LA64-NEXT:  .Ltmp0:
 ; LA64-NEXT:    move $a0, $fp
 ; LA64-NEXT:    bl %plt(bar)
 ; LA64-NEXT:  .Ltmp1:
 ; LA64-NEXT:    b .LBB0_3
-; LA64-NEXT:  .LBB0_1: # %bb1
+; LA64-NEXT:  .LBB0_2: # %bb1
 ; LA64-NEXT:  .Ltmp2:
 ; LA64-NEXT:    move $a0, $fp
 ; LA64-NEXT:    bl %plt(foo)
 ; LA64-NEXT:  .Ltmp3:
 ; LA64-NEXT:  .LBB0_3: # %end2
 ; LA64-NEXT:    ld.d $s0, $sp, 8 # 8-byte Folded Reload
 ; LA64-NEXT:    ld.d $fp, $sp, 16 # 8-byte Folded Reload
 ; LA64-NEXT:    ld.d $ra, $sp, 24 # 8-byte Folded Reload

llvm/test/CodeGen/LoongArch/jump-table.ll

 
 define void @switch_4_arms(i32 %in, ptr %out) nounwind {
 ; LA32-LABEL: switch_4_arms:
 ; LA32:       # %bb.0: # %entry
 ; LA32-NEXT:    ori $a2, $zero, 2
 ; LA32-NEXT:    blt $a2, $a0, .LBB0_4
 ; LA32-NEXT:  # %bb.1: # %entry
 ; LA32-NEXT:    ori $a3, $zero, 1
-; LA32-NEXT:    beq $a0, $a3, .LBB0_8
+; LA32-NEXT:    beq $a0, $a3, .LBB0_7
 ; LA32-NEXT:  # %bb.2: # %entry
-; LA32-NEXT:    bne $a0, $a2, .LBB0_10
+; LA32-NEXT:    bne $a0, $a2, .LBB0_9
 ; LA32-NEXT:  # %bb.3: # %bb2
 ; LA32-NEXT:    ori $a0, $zero, 3
 ; LA32-NEXT:    st.w $a0, $a1, 0
 ; LA32-NEXT:    ret
 ; LA32-NEXT:  .LBB0_4: # %entry
 ; LA32-NEXT:    ori $a3, $zero, 3
-; LA32-NEXT:    beq $a0, $a3, .LBB0_9
+; LA32-NEXT:    beq $a0, $a3, .LBB0_8
 ; LA32-NEXT:  # %bb.5: # %entry
 ; LA32-NEXT:    ori $a2, $zero, 4
-; LA32-NEXT:    bne $a0, $a2, .LBB0_10
+; LA32-NEXT:    bne $a0, $a2, .LBB0_9
 ; LA32-NEXT:  # %bb.6: # %bb4
 ; LA32-NEXT:    ori $a0, $zero, 1
 ; LA32-NEXT:    st.w $a0, $a1, 0
 ; LA32-NEXT:    ret
-; LA32-NEXT:  .LBB0_8: # %bb1
+; LA32-NEXT:  .LBB0_7: # %bb1
 ; LA32-NEXT:    ori $a0, $zero, 4
 ; LA32-NEXT:    st.w $a0, $a1, 0
 ; LA32-NEXT:    ret
-; LA32-NEXT:  .LBB0_9: # %bb3
+; LA32-NEXT:  .LBB0_8: # %bb3
 ; LA32-NEXT:    st.w $a2, $a1, 0
-; LA32-NEXT:  .LBB0_10: # %exit
+; LA32-NEXT:  .LBB0_9: # %exit
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: switch_4_arms:
 ; LA64:       # %bb.0: # %entry
 ; LA64-NEXT:    bstrpick.d $a0, $a0, 31, 0
 ; LA64-NEXT:    ori $a2, $zero, 2
 ; LA64-NEXT:    blt $a2, $a0, .LBB0_4
 ; LA64-NEXT:  # %bb.1: # %entry
 ; LA64-NEXT:    ori $a3, $zero, 1
-; LA64-NEXT:    beq $a0, $a3, .LBB0_8
+; LA64-NEXT:    beq $a0, $a3, .LBB0_7
 ; LA64-NEXT:  # %bb.2: # %entry
-; LA64-NEXT:    bne $a0, $a2, .LBB0_10
+; LA64-NEXT:    bne $a0, $a2, .LBB0_9
 ; LA64-NEXT:  # %bb.3: # %bb2
 ; LA64-NEXT:    ori $a0, $zero, 3
 ; LA64-NEXT:    st.w $a0, $a1, 0
 ; LA64-NEXT:    ret
 ; LA64-NEXT:  .LBB0_4: # %entry
 ; LA64-NEXT:    ori $a3, $zero, 3
-; LA64-NEXT:    beq $a0, $a3, .LBB0_9
+; LA64-NEXT:    beq $a0, $a3, .LBB0_8
 ; LA64-NEXT:  # %bb.5: # %entry
 ; LA64-NEXT:    ori $a2, $zero, 4
-; LA64-NEXT:    bne $a0, $a2, .LBB0_10
+; LA64-NEXT:    bne $a0, $a2, .LBB0_9
 ; LA64-NEXT:  # %bb.6: # %bb4
 ; LA64-NEXT:    ori $a0, $zero, 1
 ; LA64-NEXT:    st.w $a0, $a1, 0
 ; LA64-NEXT:    ret
-; LA64-NEXT:  .LBB0_8: # %bb1
+; LA64-NEXT:  .LBB0_7: # %bb1
 ; LA64-NEXT:    ori $a0, $zero, 4
 ; LA64-NEXT:    st.w $a0, $a1, 0
 ; LA64-NEXT:    ret
-; LA64-NEXT:  .LBB0_9: # %bb3
+; LA64-NEXT:  .LBB0_8: # %bb3
 ; LA64-NEXT:    st.w $a2, $a1, 0
-; LA64-NEXT:  .LBB0_10: # %exit
+; LA64-NEXT:  .LBB0_9: # %exit
 ; LA64-NEXT:    ret
 ;
 ; LA32-JT-LABEL: switch_4_arms:
 ; LA32-JT:       # %bb.0: # %entry
 ; LA32-JT-NEXT:    addi.w $a2, $a0, -1
 ; LA32-JT-NEXT:    ori $a0, $zero, 3
 ; LA32-JT-NEXT:    bltu $a0, $a2, .LBB0_6
 ; LA32-JT-NEXT:  # %bb.1: # %entry

llvm/test/tools/UpdateTestChecks/update_llc_test_checks/Inputs/loongarch_generated_funcs.ll.generated.expected

 ; CHECK-NEXT:    st.w $ra, $sp, 28 # 4-byte Folded Spill
 ; CHECK-NEXT:    st.w $fp, $sp, 24 # 4-byte Folded Spill
 ; CHECK-NEXT:    .cfi_offset 1, -4
 ; CHECK-NEXT:    .cfi_offset 22, -8
 ; CHECK-NEXT:    addi.w $fp, $sp, 32
 ; CHECK-NEXT:    .cfi_def_cfa 22, 0
 ; CHECK-NEXT:    st.w $zero, $fp, -16
 ; CHECK-NEXT:    st.w $zero, $fp, -12
-; CHECK-NEXT:    beqz $zero, .LBB0_1
-; CHECK-NEXT:  # %bb.2:
+; CHECK-NEXT:    beqz $zero, .LBB0_3
+; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ori $a0, $zero, 1
 ; CHECK-NEXT:    st.w $a0, $fp, -24
 ; CHECK-NEXT:    ld.w $a0, $fp, -16
 ; CHECK-NEXT:    beqz $a0, .LBB0_4
-; CHECK-NEXT:  .LBB0_5:
+; CHECK-NEXT:  .LBB0_2:
 ; CHECK-NEXT:    ori $a0, $zero, 1
 ; CHECK-NEXT:    st.w $a0, $fp, -24
-; CHECK-NEXT:    b .LBB0_6
-; CHECK-NEXT:  .LBB0_1:
+; CHECK-NEXT:    b .LBB0_5
+; CHECK-NEXT:  .LBB0_3:
 ; CHECK-NEXT:    ori $a0, $zero, 2
 ; CHECK-NEXT:    st.w $a0, $fp, -20
 ; CHECK-NEXT:    ori $a0, $zero, 1
 ; CHECK-NEXT:    st.w $a0, $fp, -16
 ; CHECK-NEXT:    ori $a0, $zero, 3
 ; CHECK-NEXT:    st.w $a0, $fp, -24
 ; CHECK-NEXT:    ori $a0, $zero, 4
 ; CHECK-NEXT:    st.w $a0, $fp, -28
 ; CHECK-NEXT:    ld.w $a0, $fp, -16
-; CHECK-NEXT:    bnez $a0, .LBB0_5
+; CHECK-NEXT:    bnez $a0, .LBB0_2
 ; CHECK-NEXT:  .LBB0_4:
 ; CHECK-NEXT:    ori $a0, $zero, 2
 ; CHECK-NEXT:    st.w $a0, $fp, -20
 ; CHECK-NEXT:    ori $a0, $zero, 1
 ; CHECK-NEXT:    st.w $a0, $fp, -16
 ; CHECK-NEXT:    ori $a0, $zero, 3
 ; CHECK-NEXT:    st.w $a0, $fp, -24
 ; CHECK-NEXT:    ori $a0, $zero, 4
 ; CHECK-NEXT:    st.w $a0, $fp, -28
-; CHECK-NEXT:  .LBB0_6:
+; CHECK-NEXT:  .LBB0_5:
 ; CHECK-NEXT:    move $a0, $zero
 ; CHECK-NEXT:    ld.w $fp, $sp, 24 # 4-byte Folded Reload
 ; CHECK-NEXT:    ld.w $ra, $sp, 28 # 4-byte Folded Reload
 ; CHECK-NEXT:    addi.w $sp, $sp, 32
 ; CHECK-NEXT:    ret
 ;
 ; CHECK-LABEL: main:
 ; CHECK:       # %bb.0:

llvm/test/tools/UpdateTestChecks/update_llc_test_checks/Inputs/loongarch_generated_funcs.ll.nogenerated.expected

 ; CHECK-NEXT:    st.w $ra, $sp, 28 # 4-byte Folded Spill
 ; CHECK-NEXT:    st.w $fp, $sp, 24 # 4-byte Folded Spill
 ; CHECK-NEXT:    .cfi_offset 1, -4
 ; CHECK-NEXT:    .cfi_offset 22, -8
 ; CHECK-NEXT:    addi.w $fp, $sp, 32
 ; CHECK-NEXT:    .cfi_def_cfa 22, 0
 ; CHECK-NEXT:    st.w $zero, $fp, -16
 ; CHECK-NEXT:    st.w $zero, $fp, -12
-; CHECK-NEXT:    beqz $zero, .LBB0_1
-; CHECK-NEXT:  # %bb.2:
+; CHECK-NEXT:    beqz $zero, .LBB0_3
+; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ori $a0, $zero, 1
 ; CHECK-NEXT:    st.w $a0, $fp, -24
 ; CHECK-NEXT:    ld.w $a0, $fp, -16
 ; CHECK-NEXT:    beqz $a0, .LBB0_4
-; CHECK-NEXT:  .LBB0_5:
+; CHECK-NEXT:  .LBB0_2:
 ; CHECK-NEXT:    ori $a0, $zero, 1
 ; CHECK-NEXT:    st.w $a0, $fp, -24
-; CHECK-NEXT:    b .LBB0_6
-; CHECK-NEXT:  .LBB0_1:
+; CHECK-NEXT:    b .LBB0_5
+; CHECK-NEXT:  .LBB0_3:
 ; CHECK-NEXT:    ori $a0, $zero, 2
 ; CHECK-NEXT:    st.w $a0, $fp, -20
 ; CHECK-NEXT:    ori $a0, $zero, 1
 ; CHECK-NEXT:    st.w $a0, $fp, -16
 ; CHECK-NEXT:    ori $a0, $zero, 3
 ; CHECK-NEXT:    st.w $a0, $fp, -24
 ; CHECK-NEXT:    ori $a0, $zero, 4
 ; CHECK-NEXT:    st.w $a0, $fp, -28
 ; CHECK-NEXT:    ld.w $a0, $fp, -16
-; CHECK-NEXT:    bnez $a0, .LBB0_5
+; CHECK-NEXT:    bnez $a0, .LBB0_2
 ; CHECK-NEXT:  .LBB0_4:
 ; CHECK-NEXT:    ori $a0, $zero, 2
 ; CHECK-NEXT:    st.w $a0, $fp, -20
 ; CHECK-NEXT:    ori $a0, $zero, 1
 ; CHECK-NEXT:    st.w $a0, $fp, -16
 ; CHECK-NEXT:    ori $a0, $zero, 3
 ; CHECK-NEXT:    st.w $a0, $fp, -24
 ; CHECK-NEXT:    ori $a0, $zero, 4
 ; CHECK-NEXT:    st.w $a0, $fp, -28
-; CHECK-NEXT:  .LBB0_6:
+; CHECK-NEXT:  .LBB0_5:
 ; CHECK-NEXT:    move $a0, $zero
 ; CHECK-NEXT:    ld.w $fp, $sp, 24 # 4-byte Folded Reload
 ; CHECK-NEXT:    ld.w $ra, $sp, 28 # 4-byte Folded Reload
 ; CHECK-NEXT:    addi.w $sp, $sp, 32
 ; CHECK-NEXT:    ret
   %1 = alloca i32, align 4
   %2 = alloca i32, align 4
   %3 = alloca i32, align 4