[RISCV] Add RISCVISD::BR_CC similar to RISCVISD::SELECT_CC.

[craig.topper]

Summary

This allows me to introduce similar combines for branches as
we have recently added for SELECT_CC. Some of them are less
useful for standalone setccs and only help branch instructions.
By having a BR_CC node its easier to only affect branches.

I'm using CondCodeSDNode to make isel patterns easier to
write so we can refer to the codes by name. SELECT_CC uses a
constant instead.

I've normalizing the condition code just like SELECT_CC so
we need less patterns for the swapped conditions. I've also
folded the setlt X, -1 handling into the normalize to get rid
of that special case isel pattern.

computeKnownBitsForTargetNode support for SELECT_CC is added
to allow MaskedValueIsZero to work for cases where the true
and false values of the SELECT_CC are setccs and the
result of the SELECT_CC is used by a BR_CC. This was needed
to avoid regressions in some of the overflow tests.

[craig.topper]

Rebase. The change to normaliseSetCC was already commited in a separate patch.

[luismarques]

LGTM.

(Perhaps it would have been better to have split it into the NFC patch and the rest?)

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

4297

Add empty line before?

[craig.topper]

In D98159#2625463, @luismarques wrote:

LGTM.

(Perhaps it would have been better to have split it into the NFC patch and the rest?)

I'm not sure there was much of a clean separation, but I'll see if I can split some of it when I commit.

[craig.topper]

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

1887

I'll split this out or drop it. It was a little less bad when we were also changing the signature of normaliseSetCC in this patch.

[craig.topper]

Remove non-functional change in LowerSELECT.
Add blank line.

Waiting for D98449 to commit before committing this since the isel pattern
for that will become obsolete, but the test from D98449 is still valid.

llvm/lib/Target/RISCV/RISCVISelLowering.h

enum NodeType : unsigned {
   CALL,
   /// Select with condition operator - This selects between a true value and
   /// a false value (ops #3 and #4) based on the boolean result of comparing
   /// the lhs and rhs (ops #0 and #1) of a conditional expression with the
   /// condition code in op #2, a XLenVT constant from the ISD::CondCode enum.
   /// The lhs and rhs are XLenVT integers. The true and false values can be
   /// integer or floating point.
   SELECT_CC,
+  BR_CC,
   BuildPairF64,
   SplitF64,
   TAIL,
   // RV64I shifts, directly matching the semantics of the named RISC-V
   // instructions.
   SLLW,
   SRAW,
   SRLW,
private:
   SDValue getDynamicTLSAddr(GlobalAddressSDNode *N, SelectionDAG &DAG) const;
 
   SDValue lowerGlobalAddress(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerBlockAddress(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerConstantPool(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerJumpTable(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerSELECT(SDValue Op, SelectionDAG &DAG) const;
+  SDValue lowerBRCOND(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerVASTART(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerRETURNADDR(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerShiftLeftParts(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerShiftRightParts(SDValue Op, SelectionDAG &DAG, bool IsSRA) const;
   SDValue lowerSPLAT_VECTOR_PARTS(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerVectorMaskExt(SDValue Op, SelectionDAG &DAG,
                              int64_t ExtTrueVal) const;

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
   for (auto N : {ISD::EXTLOAD, ISD::SEXTLOAD, ISD::ZEXTLOAD})
     setLoadExtAction(N, XLenVT, MVT::i1, Promote);
 
   // TODO: add all necessary setOperationAction calls.
   setOperationAction(ISD::DYNAMIC_STACKALLOC, XLenVT, Expand);
 
   setOperationAction(ISD::BR_JT, MVT::Other, Expand);
   setOperationAction(ISD::BR_CC, XLenVT, Expand);
+  setOperationAction(ISD::BRCOND, MVT::Other, Custom);
   setOperationAction(ISD::SELECT_CC, XLenVT, Expand);
 
   setOperationAction(ISD::STACKSAVE, MVT::Other, Expand);
   setOperationAction(ISD::STACKRESTORE, MVT::Other, Expand);
 
   setOperationAction(ISD::VASTART, MVT::Other, Custom);
   setOperationAction(ISD::VAARG, MVT::Other, Expand);
   setOperationAction(ISD::VACOPY, MVT::Other, Expand);
RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
   setMinimumJumpTableEntries(5);
 
   // Jumps are expensive, compared to logic
   setJumpIsExpensive();
 
   // We can use any register for comparisons
   setHasMultipleConditionRegisters();
 
-  setTargetDAGCombine(ISD::SETCC);
   if (Subtarget.hasStdExtZbp()) {
     setTargetDAGCombine(ISD::OR);
   }
   if (Subtarget.hasStdExtV())
     setTargetDAGCombine(ISD::FCOPYSIGN);
 }
 
 EVT RISCVTargetLowering::getSetCCResultType(const DataLayout &DL,
SDValue RISCVTargetLowering::LowerOperation(SDValue Op,
   case ISD::ConstantPool:
     return lowerConstantPool(Op, DAG);
   case ISD::JumpTable:
     return lowerJumpTable(Op, DAG);
   case ISD::GlobalTLSAddress:
     return lowerGlobalTLSAddress(Op, DAG);
   case ISD::SELECT:
     return lowerSELECT(Op, DAG);
+  case ISD::BRCOND:
+    return lowerBRCOND(Op, DAG);
   case ISD::VASTART:
     return lowerVASTART(Op, DAG);
   case ISD::FRAMEADDR:
     return lowerFRAMEADDR(Op, DAG);
   case ISD::RETURNADDR:
     return lowerRETURNADDR(Op, DAG);
   case ISD::SHL_PARTS:
     return lowerShiftLeftParts(Op, DAG);
SDValue RISCVTargetLowering::lowerSELECT(SDValue Op, SelectionDAG &DAG) const {
 
   // If the result type is XLenVT and CondV is the output of a SETCC node
   // which also operated on XLenVT inputs, then merge the SETCC node into the
   // lowered RISCVISD::SELECT_CC to take advantage of the integer
   // compare+branch instructions. i.e.:
   // (select (setcc lhs, rhs, cc), truev, falsev)
   // -> (riscvisd::select_cc lhs, rhs, cc, truev, falsev)
   if (Op.getSimpleValueType() == XLenVT && CondV.getOpcode() == ISD::SETCC &&
       CondV.getOperand(0).getSimpleValueType() == XLenVT) {
     SDValue LHS = CondV.getOperand(0);
     SDValue RHS = CondV.getOperand(1);
     auto CC = cast<CondCodeSDNode>(CondV.getOperand(2));
     ISD::CondCode CCVal = CC->get();
 
     translateSetCCForBranch(DL, LHS, RHS, CCVal, DAG);
 
     SDValue TargetCC = DAG.getConstant(CCVal, DL, XLenVT);
     SDValue Ops[] = {LHS, RHS, TargetCC, TrueV, FalseV};
     return DAG.getNode(RISCVISD::SELECT_CC, DL, Op.getValueType(), Ops);
   }
 
   // Otherwise:
   // (select condv, truev, falsev)
   // -> (riscvisd::select_cc condv, zero, setne, truev, falsev)
   SDValue Zero = DAG.getConstant(0, DL, XLenVT);
   SDValue SetNE = DAG.getConstant(ISD::SETNE, DL, XLenVT);
 
   SDValue Ops[] = {CondV, Zero, SetNE, TrueV, FalseV};
 
   return DAG.getNode(RISCVISD::SELECT_CC, DL, Op.getValueType(), Ops);
 }
 
+SDValue RISCVTargetLowering::lowerBRCOND(SDValue Op, SelectionDAG &DAG) const {
+  SDValue CondV = Op.getOperand(1);
+  SDLoc DL(Op);
+  MVT XLenVT = Subtarget.getXLenVT();
+
+  if (CondV.getOpcode() == ISD::SETCC &&
+      CondV.getOperand(0).getValueType() == XLenVT) {
+    SDValue LHS = CondV.getOperand(0);
+    SDValue RHS = CondV.getOperand(1);
+    ISD::CondCode CCVal = cast<CondCodeSDNode>(CondV.getOperand(2))->get();
+
+    translateSetCCForBranch(DL, LHS, RHS, CCVal, DAG);
+
+    SDValue TargetCC = DAG.getCondCode(CCVal);
+    return DAG.getNode(RISCVISD::BR_CC, DL, Op.getValueType(), Op.getOperand(0),
+                       LHS, RHS, TargetCC, Op.getOperand(2));
+  }
+
+  return DAG.getNode(RISCVISD::BR_CC, DL, Op.getValueType(), Op.getOperand(0),
+                     CondV, DAG.getConstant(0, DL, XLenVT),
+                     DAG.getCondCode(ISD::SETNE), Op.getOperand(2));
+}
+
 SDValue RISCVTargetLowering::lowerVASTART(SDValue Op, SelectionDAG &DAG) const {
   MachineFunction &MF = DAG.getMachineFunction();
   RISCVMachineFunctionInfo *FuncInfo = MF.getInfo<RISCVMachineFunctionInfo>();
 
   SDLoc DL(Op);
   SDValue FI = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(),
                                  getPointerTy(MF.getDataLayout()));
 
if (isOneConstant(RHS) && DAG.MaskedValueIsZero(LHS, Mask)) {
       RHS = DAG.getConstant(0, DL, LHS.getValueType());
       return DAG.getNode(
           RISCVISD::SELECT_CC, DL, N->getValueType(0),
           {LHS, RHS, TargetCC, N->getOperand(3), N->getOperand(4)});
     }
 
     break;
   }
-  case ISD::SETCC: {
-    // (setcc X, 1, setne) -> (setcc X, 0, seteq) if we can prove X is 0/1.
-    // Comparing with 0 may allow us to fold into bnez/beqz.
-    SDValue LHS = N->getOperand(0);
-    SDValue RHS = N->getOperand(1);
-    if (LHS.getValueType().isScalableVector())
+  case RISCVISD::BR_CC: {
+    SDValue LHS = N->getOperand(1);
+    SDValue RHS = N->getOperand(2);
+    ISD::CondCode CCVal = cast<CondCodeSDNode>(N->getOperand(3))->get();
+    if (!ISD::isIntEqualitySetCC(CCVal))
       break;
-    auto CC = cast<CondCodeSDNode>(N->getOperand(2))->get();
+
+    // Fold (br_cc (setlt X, Y), 0, ne, dest) ->
+    //      (br_cc X, Y, lt, dest)
+    // Sometimes the setcc is introduced after br_cc has been formed.
+    if (LHS.getOpcode() == ISD::SETCC && isNullConstant(RHS) &&
+        LHS.getOperand(0).getValueType() == Subtarget.getXLenVT()) {
+      // If we're looking for eq 0 instead of ne 0, we need to invert the
+      // condition.
+      bool Invert = CCVal == ISD::SETEQ;
+      CCVal = cast<CondCodeSDNode>(LHS.getOperand(2))->get();
+      if (Invert)
+        CCVal = ISD::getSetCCInverse(CCVal, LHS.getValueType());
+
+      SDLoc DL(N);
+      RHS = LHS.getOperand(1);
+      LHS = LHS.getOperand(0);
+      translateSetCCForBranch(DL, LHS, RHS, CCVal, DAG);
+
+      return DAG.getNode(RISCVISD::BR_CC, DL, N->getValueType(0),
+                         N->getOperand(0), LHS, RHS, DAG.getCondCode(CCVal),
+                         N->getOperand(4));
+    }
+
+    // Fold (br_cc (xor X, Y), 0, eq/ne, dest) ->
+    //      (br_cc X, Y, eq/ne, trueV, falseV)
+    if (LHS.getOpcode() == ISD::XOR && isNullConstant(RHS))
+      return DAG.getNode(RISCVISD::BR_CC, SDLoc(N), N->getValueType(0),
+                         N->getOperand(0), LHS.getOperand(0), LHS.getOperand(1),
+                         N->getOperand(3), N->getOperand(4));
+
+    // (br_cc X, 1, setne, br_cc) ->
+    // (br_cc X, 0, seteq, br_cc) if we can prove X is 0/1.
+    // This can occur when legalizing some floating point comparisons.
     APInt Mask = APInt::getBitsSetFrom(LHS.getValueSizeInBits(), 1);
-    if (isOneConstant(RHS) && ISD::isIntEqualitySetCC(CC) &&
-        DAG.MaskedValueIsZero(LHS, Mask)) {
+    if (isOneConstant(RHS) && DAG.MaskedValueIsZero(LHS, Mask)) {
       SDLoc DL(N);
-      SDValue Zero = DAG.getConstant(0, DL, LHS.getValueType());
-      CC = ISD::getSetCCInverse(CC, LHS.getValueType());
-      return DAG.getSetCC(DL, N->getValueType(0), LHS, Zero, CC);
+      CCVal = ISD::getSetCCInverse(CCVal, LHS.getValueType());
+      SDValue TargetCC = DAG.getCondCode(CCVal);
+      RHS = DAG.getConstant(0, DL, LHS.getValueType());
+      return DAG.getNode(RISCVISD::BR_CC, DL, N->getValueType(0),
+                         N->getOperand(0), LHS, RHS, TargetCC,
+                         N->getOperand(4));
     }
     break;
   }
   case ISD::FCOPYSIGN: {
     EVT VT = N->getValueType(0);
     if (!VT.isVector())
       break;
     // There is a form of VFSGNJ which injects the negated sign of its second
assert((Opc >= ISD::BUILTIN_OP_END ||
           Opc == ISD::INTRINSIC_W_CHAIN ||
           Opc == ISD::INTRINSIC_VOID) &&
          "Should use MaskedValueIsZero if you don't know whether Op"
          " is a target node!");
 
   Known.resetAll();
   switch (Opc) {
   default: break;
+  case RISCVISD::SELECT_CC: {
+    Known = DAG.computeKnownBits(Op.getOperand(4), Depth + 1);
+    // If we don't know any bits, early out.
+    if (Known.isUnknown())
+      break;
+    KnownBits Known2 = DAG.computeKnownBits(Op.getOperand(3), Depth + 1);
+
+    // Only known if known in both the LHS and RHS.
+    Known = KnownBits::commonBits(Known, Known2);
+    break;
+  }
   case RISCVISD::REMUW: {
     KnownBits Known2;
     Known = DAG.computeKnownBits(Op.getOperand(0), DemandedElts, Depth + 1);
     Known2 = DAG.computeKnownBits(Op.getOperand(1), DemandedElts, Depth + 1);
     // We only care about the lower 32 bits.
     Known = KnownBits::urem(Known.trunc(32), Known2.trunc(32));
     // Restore the original width by sign extending.
     Known = Known.sext(BitWidth);
#define NODE_NAME_CASE(NODE)                                                   \
   case RISCVISD::FIRST_NUMBER:
     break;
   NODE_NAME_CASE(RET_FLAG)
   NODE_NAME_CASE(URET_FLAG)
   NODE_NAME_CASE(SRET_FLAG)
   NODE_NAME_CASE(MRET_FLAG)
   NODE_NAME_CASE(CALL)
   NODE_NAME_CASE(SELECT_CC)
+  NODE_NAME_CASE(BR_CC)
   NODE_NAME_CASE(BuildPairF64)
   NODE_NAME_CASE(SplitF64)
   NODE_NAME_CASE(TAIL)
   NODE_NAME_CASE(SLLW)
   NODE_NAME_CASE(SRAW)
   NODE_NAME_CASE(SRLW)
   NODE_NAME_CASE(DIVW)
   NODE_NAME_CASE(DIVUW)

llvm/lib/Target/RISCV/RISCVInstrInfo.td

 def SDT_CallSeqEnd   : SDCallSeqEnd<[SDTCisVT<0, i32>,
                                      SDTCisVT<1, i32>]>;
 
 // Target-dependent type requirements.
 def SDT_RISCVCall     : SDTypeProfile<0, -1, [SDTCisVT<0, XLenVT>]>;
 def SDT_RISCVSelectCC : SDTypeProfile<1, 5, [SDTCisSameAs<1, 2>,
                                              SDTCisSameAs<0, 4>,
                                              SDTCisSameAs<4, 5>]>;
+def SDT_RISCVBrCC : SDTypeProfile<0, 4, [SDTCisSameAs<0, 1>,
+                                         SDTCisVT<2, OtherVT>,
+                                         SDTCisVT<3, OtherVT>]>;
 def SDT_RISCVReadCycleWide : SDTypeProfile<2, 0, [SDTCisVT<0, i32>,
                                                   SDTCisVT<1, i32>]>;
 def SDT_RISCVIntBinOpW : SDTypeProfile<1, 2, [
   SDTCisSameAs<0, 1>, SDTCisSameAs<0, 2>, SDTCisVT<0, i64>
 ]>;
 
 // Target-independent nodes, but with target-specific formats.
 def callseq_start : SDNode<"ISD::CALLSEQ_START", SDT_CallSeqStart,
def riscv_ret_flag  : SDNode<"RISCVISD::RET_FLAG", SDTNone,
                              [SDNPHasChain, SDNPOptInGlue, SDNPVariadic]>;
 def riscv_uret_flag : SDNode<"RISCVISD::URET_FLAG", SDTNone,
                              [SDNPHasChain, SDNPOptInGlue]>;
 def riscv_sret_flag : SDNode<"RISCVISD::SRET_FLAG", SDTNone,
                              [SDNPHasChain, SDNPOptInGlue]>;
 def riscv_mret_flag : SDNode<"RISCVISD::MRET_FLAG", SDTNone,
                              [SDNPHasChain, SDNPOptInGlue]>;
 def riscv_selectcc  : SDNode<"RISCVISD::SELECT_CC", SDT_RISCVSelectCC>;
+def riscv_brcc      : SDNode<"RISCVISD::BR_CC", SDT_RISCVBrCC,
+                             [SDNPHasChain]>;
 def riscv_tail      : SDNode<"RISCVISD::TAIL", SDT_RISCVCall,
                              [SDNPHasChain, SDNPOptInGlue, SDNPOutGlue,
                               SDNPVariadic]>;
 def riscv_sllw      : SDNode<"RISCVISD::SLLW", SDT_RISCVIntBinOpW>;
 def riscv_sraw      : SDNode<"RISCVISD::SRAW", SDT_RISCVIntBinOpW>;
 def riscv_srlw      : SDNode<"RISCVISD::SRLW", SDT_RISCVIntBinOpW>;
 
 def riscv_read_cycle_wide : SDNode<"RISCVISD::READ_CYCLE_WIDE",
: Pseudo<(outs valty:$dst),
               valty:$truev, valty:$falsev),
              [(set valty:$dst, (riscv_selectcc cmpty:$lhs, cmpty:$rhs,
               (XLenVT imm:$imm), valty:$truev, valty:$falsev))]>;
 
 def Select_GPR_Using_CC_GPR : SelectCC_rrirr<GPR, GPR>;
 
 /// Branches and jumps
 
-// Match `(brcond (CondOp ..), ..)` and lower to the appropriate RISC-V branch
-// instruction.
-class BccPat<PatFrag CondOp, RVInstB Inst>
-    : Pat<(brcond (XLenVT (CondOp GPR:$rs1, GPR:$rs2)), bb:$imm12),
+// Match `riscv_brcc` and lower to the appropriate RISC-V branch instruction.
+class BccPat<CondCode Cond, RVInstB Inst>
+    : Pat<(riscv_brcc GPR:$rs1, GPR:$rs2, Cond, bb:$imm12),
           (Inst GPR:$rs1, GPR:$rs2, simm13_lsb0:$imm12)>;
 
-def : BccPat<seteq, BEQ>;
-def : BccPat<setne, BNE>;
-def : BccPat<setlt, BLT>;
-def : BccPat<setge, BGE>;
-def : BccPat<setult, BLTU>;
-def : BccPat<setuge, BGEU>;
-
-class BccSwapPat<PatFrag CondOp, RVInst InstBcc>
-    : Pat<(brcond (XLenVT (CondOp GPR:$rs1, GPR:$rs2)), bb:$imm12),
-          (InstBcc GPR:$rs2, GPR:$rs1, bb:$imm12)>;
-
-// Condition codes that don't have matching RISC-V branch instructions, but
-// are trivially supported by swapping the two input operands
-def : BccSwapPat<setgt, BLT>;
-def : BccSwapPat<setle, BGE>;
-def : BccSwapPat<setugt, BLTU>;
-def : BccSwapPat<setule, BGEU>;
-
-// Extra patterns are needed for a brcond without a setcc (i.e. where the
-// condition was calculated elsewhere).
-def : Pat<(brcond GPR:$cond, bb:$imm12), (BNE GPR:$cond, X0, bb:$imm12)>;
-// In this pattern, the `(xor $cond, 1)` functions like (boolean) `not`, as the
-// `brcond` only uses the lowest bit.
-def : Pat<(brcond (XLenVT (xor GPR:$cond, 1)), bb:$imm12),
-          (BEQ GPR:$cond, X0, bb:$imm12)>;
-
-// Match X > -1, the canonical form of X >= 0, to the bgez pattern.
-def : Pat<(brcond (XLenVT (setgt GPR:$rs1, -1)), bb:$imm12),
-          (BGE GPR:$rs1, X0, bb:$imm12)>;
-// Lower (a < 1) as (0 >= a) into the blez pattern.
-def : Pat<(brcond (XLenVT (setlt GPR:$lhs, 1)), bb:$imm12),
-          (BGE X0, GPR:$lhs, bb:$imm12)>;
+def : BccPat<SETEQ, BEQ>;
+def : BccPat<SETNE, BNE>;
+def : BccPat<SETLT, BLT>;
+def : BccPat<SETGE, BGE>;
+def : BccPat<SETULT, BLTU>;
+def : BccPat<SETUGE, BGEU>;
 
 let isBarrier = 1, isBranch = 1, isTerminator = 1 in
 def PseudoBR : Pseudo<(outs), (ins simm21_lsb0_jal:$imm20), [(br bb:$imm20)]>,
                PseudoInstExpansion<(JAL X0, simm21_lsb0_jal:$imm20)>;
 
 let isBarrier = 1, isBranch = 1, isIndirectBranch = 1, isTerminator = 1 in
 def PseudoBRIND : Pseudo<(outs), (ins GPR:$rs1, simm12:$imm12), []>,
                   PseudoInstExpansion<(JALR X0, GPR:$rs1, simm12:$imm12)>;

llvm/test/CodeGen/RISCV/xaluo.ll

 ; Check the use of the overflow bit in combination with a branch instruction.
 ;
 define zeroext i1 @saddo.br.i32(i32 %v1, i32 %v2) {
 ; RV32-LABEL: saddo.br.i32:
 ; RV32:       # %bb.0: # %entry
 ; RV32-NEXT:    add a2, a0, a1
 ; RV32-NEXT:    slt a0, a2, a0
 ; RV32-NEXT:    slti a1, a1, 0
-; RV32-NEXT:    xor a0, a1, a0
-; RV32-NEXT:    beqz a0, .LBB46_2
+; RV32-NEXT:    beq a1, a0, .LBB46_2
 ; RV32-NEXT:  # %bb.1: # %overflow
 ; RV32-NEXT:    mv a0, zero
 ; RV32-NEXT:    ret
 ; RV32-NEXT:  .LBB46_2: # %continue
 ; RV32-NEXT:    addi a0, zero, 1
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: saddo.br.i32:

 ; RV32-NEXT:    addi a0, zero, 1
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: saddo.br.i64:
 ; RV64:       # %bb.0: # %entry
 ; RV64-NEXT:    add a2, a0, a1
 ; RV64-NEXT:    slt a0, a2, a0
 ; RV64-NEXT:    slti a1, a1, 0
-; RV64-NEXT:    xor a0, a1, a0
-; RV64-NEXT:    beqz a0, .LBB47_2
+; RV64-NEXT:    beq a1, a0, .LBB47_2
 ; RV64-NEXT:  # %bb.1: # %overflow
 ; RV64-NEXT:    mv a0, zero
 ; RV64-NEXT:    ret
 ; RV64-NEXT:  .LBB47_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 %v1, i64 %v2)

 }
 
 define zeroext i1 @ssubo.br.i32(i32 %v1, i32 %v2) {
 ; RV32-LABEL: ssubo.br.i32:
 ; RV32:       # %bb.0: # %entry
 ; RV32-NEXT:    sgtz a2, a1
 ; RV32-NEXT:    sub a1, a0, a1
 ; RV32-NEXT:    slt a0, a1, a0
-; RV32-NEXT:    xor a0, a2, a0
-; RV32-NEXT:    beqz a0, .LBB50_2
+; RV32-NEXT:    beq a2, a0, .LBB50_2
 ; RV32-NEXT:  # %bb.1: # %overflow
 ; RV32-NEXT:    mv a0, zero
 ; RV32-NEXT:    ret
 ; RV32-NEXT:  .LBB50_2: # %continue
 ; RV32-NEXT:    addi a0, zero, 1
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ssubo.br.i32:

 ; RV32-NEXT:    addi a0, zero, 1
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ssubo.br.i64:
 ; RV64:       # %bb.0: # %entry
 ; RV64-NEXT:    sgtz a2, a1
 ; RV64-NEXT:    sub a1, a0, a1
 ; RV64-NEXT:    slt a0, a1, a0
-; RV64-NEXT:    xor a0, a2, a0
-; RV64-NEXT:    beqz a0, .LBB51_2
+; RV64-NEXT:    beq a2, a0, .LBB51_2
 ; RV64-NEXT:  # %bb.1: # %overflow
 ; RV64-NEXT:    mv a0, zero
 ; RV64-NEXT:    ret
 ; RV64-NEXT:  .LBB51_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.ssub.with.overflow.i64(i64 %v1, i64 %v2)