[DAG] Fold (X & Y) != 0 --> zextOrTrunc(X & Y) iff everything but LSB is known zero (PR51312)

Fixes parity codegen issue where we know all but the lowest bit is zero, we can replace the ICMPNE with 0 comparison with a ext/trunc

Differential Revision: https://reviews.llvm.org/D117983
This commit is contained in:
Simon Pilgrim 2022-01-23 16:20:34 +00:00
parent 116ab78694
commit 7c66aaddb1
2 changed files with 17 additions and 6 deletions

View File

@ -3254,17 +3254,29 @@ bool TargetLowering::isExtendedTrueVal(const ConstantSDNode *N, EVT VT,
SDValue TargetLowering::foldSetCCWithAnd(EVT VT, SDValue N0, SDValue N1,
ISD::CondCode Cond, const SDLoc &DL,
DAGCombinerInfo &DCI) const {
// Match these patterns in any of their permutations:
// (X & Y) == Y
// (X & Y) != Y
if (N1.getOpcode() == ISD::AND && N0.getOpcode() != ISD::AND)
std::swap(N0, N1);
SelectionDAG &DAG = DCI.DAG;
EVT OpVT = N0.getValueType();
if (N0.getOpcode() != ISD::AND || !OpVT.isInteger() ||
(Cond != ISD::SETEQ && Cond != ISD::SETNE))
return SDValue();
// (X & Y) != 0 --> zextOrTrunc(X & Y)
// iff everything but LSB is known zero:
if (Cond == ISD::SETNE && isNullConstant(N1) &&
(getBooleanContents(VT) == TargetLowering::UndefinedBooleanContent ||
getBooleanContents(VT) == TargetLowering::ZeroOrOneBooleanContent)) {
unsigned NumEltBits = OpVT.getScalarSizeInBits();
APInt UpperBits = APInt::getHighBitsSet(NumEltBits, NumEltBits - 1);
if (DAG.MaskedValueIsZero(N0, UpperBits))
return DAG.getBoolExtOrTrunc(N0, DL, VT, OpVT);
}
// Match these patterns in any of their permutations:
// (X & Y) == Y
// (X & Y) != Y
SDValue X, Y;
if (N0.getOperand(0) == N1) {
X = N0.getOperand(1);
@ -3276,7 +3288,6 @@ SDValue TargetLowering::foldSetCCWithAnd(EVT VT, SDValue N0, SDValue N1,
return SDValue();
}
SelectionDAG &DAG = DCI.DAG;
SDValue Zero = DAG.getConstant(0, DL, OpVT);
if (DAG.isKnownToBeAPowerOfTwo(Y)) {
// Simplify X & Y == Y to X & Y != 0 if Y has exactly one bit set.

View File

@ -36,8 +36,8 @@ define i1 @canonical_parity(<16 x i1> %x) {
; POPCNT-NEXT: psllw $7, %xmm0
; POPCNT-NEXT: pmovmskb %xmm0, %eax
; POPCNT-NEXT: popcntl %eax, %eax
; POPCNT-NEXT: testb $1, %al
; POPCNT-NEXT: setne %al
; POPCNT-NEXT: andl $1, %eax
; POPCNT-NEXT: # kill: def $al killed $al killed $eax
; POPCNT-NEXT: retq
%i1 = bitcast <16 x i1> %x to i16
%i2 = call i16 @llvm.ctpop.i16(i16 %i1)