//===----------------------------------------------------------------------===//
#include "AArch64ISelLowering.h"
+#include "AArch64MachineFunctionInfo.h"
#include "AArch64PerfectShuffle.h"
#include "AArch64Subtarget.h"
-#include "AArch64MachineFunctionInfo.h"
#include "AArch64TargetMachine.h"
#include "AArch64TargetObjectFile.h"
#include "MCTargetDesc/AArch64AddressingModes.h"
setOperationAction(ISD::FROUND, Ty, Legal);
}
}
+
+ // Prefer likely predicted branches to selects on out-of-order cores.
+ if (Subtarget->isCortexA57())
+ PredictableSelectIsExpensive = true;
}
void AArch64TargetLowering::addTypeForNEON(EVT VT, EVT PromotedBitwiseVT) {
// EndBB:
// Dest = PHI [IfTrue, TrueBB], [IfFalse, OrigBB]
- const TargetInstrInfo *TII = getTargetMachine().getInstrInfo();
+ const TargetInstrInfo *TII =
+ getTargetMachine().getSubtargetImpl()->getInstrInfo();
MachineFunction *MF = MBB->getParent();
const BasicBlock *LLVM_BB = MBB->getBasicBlock();
DebugLoc DL = MI->getDebugLoc();
SDLoc DL(Op);
unsigned IsWrite = cast<ConstantSDNode>(Op.getOperand(2))->getZExtValue();
unsigned Locality = cast<ConstantSDNode>(Op.getOperand(3))->getZExtValue();
- // The data thing is not used.
- // unsigned isData = cast<ConstantSDNode>(Op.getOperand(4))->getZExtValue();
+ unsigned IsData = cast<ConstantSDNode>(Op.getOperand(4))->getZExtValue();
bool IsStream = !Locality;
// When the locality number is set
// built the mask value encoding the expected behavior.
unsigned PrfOp = (IsWrite << 4) | // Load/Store bit
+ (!IsData << 3) | // IsDataCache bit
(Locality << 1) | // Cache level bits
(unsigned)IsStream; // Stream bit
return DAG.getNode(AArch64ISD::PREFETCH, DL, MVT::Other, Op.getOperand(0),
// Assign locations to all of the incoming arguments.
SmallVector<CCValAssign, 16> ArgLocs;
- CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(),
- getTargetMachine(), ArgLocs, *DAG.getContext());
+ CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(), ArgLocs,
+ *DAG.getContext());
// At this point, Ins[].VT may already be promoted to i32. To correctly
// handle passing i8 as i8 instead of i32 on stack, we pass in both i32 and
ArgValue = DAG.getExtLoad(ExtType, DL, VA.getLocVT(), Chain, FIN,
MachinePointerInfo::getFixedStack(FI),
- MemVT, false, false, false, nullptr);
+ MemVT, false, false, false, 0, nullptr);
InVals.push_back(ArgValue);
}
: RetCC_AArch64_AAPCS;
// Assign locations to each value returned by this call.
SmallVector<CCValAssign, 16> RVLocs;
- CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(),
- getTargetMachine(), RVLocs, *DAG.getContext());
+ CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(), RVLocs,
+ *DAG.getContext());
CCInfo.AnalyzeCallResult(Ins, RetCC);
// Copy all of the result registers out of their specified physreg.
// FIXME: for now we take the most conservative of these in both cases:
// disallow all variadic memory operands.
SmallVector<CCValAssign, 16> ArgLocs;
- CCState CCInfo(CalleeCC, isVarArg, DAG.getMachineFunction(),
- getTargetMachine(), ArgLocs, *DAG.getContext());
+ CCState CCInfo(CalleeCC, isVarArg, DAG.getMachineFunction(), ArgLocs,
+ *DAG.getContext());
CCInfo.AnalyzeCallOperands(Outs, CCAssignFnForCall(CalleeCC, true));
for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i)
// results are returned in the same way as what the caller expects.
if (!CCMatch) {
SmallVector<CCValAssign, 16> RVLocs1;
- CCState CCInfo1(CalleeCC, false, DAG.getMachineFunction(),
- getTargetMachine(), RVLocs1, *DAG.getContext());
+ CCState CCInfo1(CalleeCC, false, DAG.getMachineFunction(), RVLocs1,
+ *DAG.getContext());
CCInfo1.AnalyzeCallResult(Ins, CCAssignFnForCall(CalleeCC, isVarArg));
SmallVector<CCValAssign, 16> RVLocs2;
- CCState CCInfo2(CallerCC, false, DAG.getMachineFunction(),
- getTargetMachine(), RVLocs2, *DAG.getContext());
+ CCState CCInfo2(CallerCC, false, DAG.getMachineFunction(), RVLocs2,
+ *DAG.getContext());
CCInfo2.AnalyzeCallResult(Ins, CCAssignFnForCall(CallerCC, isVarArg));
if (RVLocs1.size() != RVLocs2.size())
return true;
SmallVector<CCValAssign, 16> ArgLocs;
- CCState CCInfo(CalleeCC, isVarArg, DAG.getMachineFunction(),
- getTargetMachine(), ArgLocs, *DAG.getContext());
+ CCState CCInfo(CalleeCC, isVarArg, DAG.getMachineFunction(), ArgLocs,
+ *DAG.getContext());
CCInfo.AnalyzeCallOperands(Outs, CCAssignFnForCall(CalleeCC, isVarArg));
// Analyze operands of the call, assigning locations to each operand.
SmallVector<CCValAssign, 16> ArgLocs;
- CCState CCInfo(CallConv, IsVarArg, DAG.getMachineFunction(),
- getTargetMachine(), ArgLocs, *DAG.getContext());
+ CCState CCInfo(CallConv, IsVarArg, DAG.getMachineFunction(), ArgLocs,
+ *DAG.getContext());
if (IsVarArg) {
// Handle fixed and variable vector arguments differently.
DAG.getConstant(Outs[i].Flags.getByValSize(), MVT::i64);
SDValue Cpy = DAG.getMemcpy(
Chain, DL, DstAddr, Arg, SizeNode, Outs[i].Flags.getByValAlign(),
- /*isVolatile = */ false,
- /*alwaysInline = */ false, DstInfo, MachinePointerInfo());
+ /*isVol = */ false,
+ /*AlwaysInline = */ false, DstInfo, MachinePointerInfo());
MemOpChains.push_back(Cpy);
} else {
// Add a register mask operand representing the call-preserved registers.
const uint32_t *Mask;
- const TargetRegisterInfo *TRI = getTargetMachine().getRegisterInfo();
+ const TargetRegisterInfo *TRI =
+ getTargetMachine().getSubtargetImpl()->getRegisterInfo();
const AArch64RegisterInfo *ARI =
static_cast<const AArch64RegisterInfo *>(TRI);
if (IsThisReturn) {
? RetCC_AArch64_WebKit_JS
: RetCC_AArch64_AAPCS;
SmallVector<CCValAssign, 16> RVLocs;
- CCState CCInfo(CallConv, isVarArg, MF, getTargetMachine(), RVLocs, Context);
+ CCState CCInfo(CallConv, isVarArg, MF, RVLocs, Context);
return CCInfo.CheckReturn(Outs, RetCC);
}
? RetCC_AArch64_WebKit_JS
: RetCC_AArch64_AAPCS;
SmallVector<CCValAssign, 16> RVLocs;
- CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(),
- getTargetMachine(), RVLocs, *DAG.getContext());
+ CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(), RVLocs,
+ *DAG.getContext());
CCInfo.AnalyzeReturn(Outs, RetCC);
// Copy the result values into the output registers.
// TLS calls preserve all registers except those that absolutely must be
// trashed: X0 (it takes an argument), LR (it's a call) and NZCV (let's not be
// silly).
- const TargetRegisterInfo *TRI = getTargetMachine().getRegisterInfo();
+ const TargetRegisterInfo *TRI =
+ getTargetMachine().getSubtargetImpl()->getRegisterInfo();
const AArch64RegisterInfo *ARI =
static_cast<const AArch64RegisterInfo *>(TRI);
const uint32_t *Mask = ARI->getTLSCallPreservedMask();
// TLS calls preserve all registers except those that absolutely must be
// trashed: X0 (it takes an argument), LR (it's a call) and NZCV (let's not be
// silly).
- const TargetRegisterInfo *TRI = getTargetMachine().getRegisterInfo();
+ const TargetRegisterInfo *TRI =
+ getTargetMachine().getSubtargetImpl()->getRegisterInfo();
const AArch64RegisterInfo *ARI =
static_cast<const AArch64RegisterInfo *>(TRI);
const uint32_t *Mask = ARI->getTLSCallPreservedMask();
isPowerOf2_64(LHS.getConstantOperandVal(1))) {
SDValue Test = LHS.getOperand(0);
uint64_t Mask = LHS.getConstantOperandVal(1);
-
- // TBZ only operates on i64's, but the ext should be free.
- if (Test.getValueType() == MVT::i32)
- Test = DAG.getAnyExtOrTrunc(Test, dl, MVT::i64);
-
return DAG.getNode(AArch64ISD::TBZ, dl, MVT::Other, Chain, Test,
DAG.getConstant(Log2_64(Mask), MVT::i64), Dest);
}
isPowerOf2_64(LHS.getConstantOperandVal(1))) {
SDValue Test = LHS.getOperand(0);
uint64_t Mask = LHS.getConstantOperandVal(1);
-
- // TBNZ only operates on i64's, but the ext should be free.
- if (Test.getValueType() == MVT::i32)
- Test = DAG.getAnyExtOrTrunc(Test, dl, MVT::i64);
-
return DAG.getNode(AArch64ISD::TBNZ, dl, MVT::Other, Chain, Test,
DAG.getConstant(Log2_64(Mask), MVT::i64), Dest);
}
return DAG.getNode(AArch64ISD::CBNZ, dl, MVT::Other, Chain, LHS, Dest);
+ } else if (CC == ISD::SETLT && LHS.getOpcode() != ISD::AND) {
+ // Don't combine AND since emitComparison converts the AND to an ANDS
+ // (a.k.a. TST) and the test in the test bit and branch instruction
+ // becomes redundant. This would also increase register pressure.
+ uint64_t Mask = LHS.getValueType().getSizeInBits() - 1;
+ return DAG.getNode(AArch64ISD::TBNZ, dl, MVT::Other, Chain, LHS,
+ DAG.getConstant(Mask, MVT::i64), Dest);
}
}
+ if (RHSC && RHSC->getSExtValue() == -1 && CC == ISD::SETGT &&
+ LHS.getOpcode() != ISD::AND) {
+ // Don't combine AND since emitComparison converts the AND to an ANDS
+ // (a.k.a. TST) and the test in the test bit and branch instruction
+ // becomes redundant. This would also increase register pressure.
+ uint64_t Mask = LHS.getValueType().getSizeInBits() - 1;
+ return DAG.getNode(AArch64ISD::TBZ, dl, MVT::Other, Chain, LHS,
+ DAG.getConstant(Mask, MVT::i64), Dest);
+ }
SDValue CCVal;
SDValue Cmp = getAArch64Cmp(LHS, RHS, CC, CCVal, DAG, dl);
return;
case 'J': {
uint64_t NVal = -C->getSExtValue();
- if (isUInt<12>(NVal) || isShiftedUInt<12, 12>(NVal))
+ if (isUInt<12>(NVal) || isShiftedUInt<12, 12>(NVal)) {
+ CVal = C->getSExtValue();
break;
+ }
return;
}
// The K and L constraints apply *only* to logical immediates, including
EVT VT = Op.getValueType();
unsigned NumElts = VT.getVectorNumElements();
- SmallVector<SDValue, 2> SourceVecs;
- SmallVector<unsigned, 2> MinElts;
- SmallVector<unsigned, 2> MaxElts;
+ struct ShuffleSourceInfo {
+ SDValue Vec;
+ unsigned MinElt;
+ unsigned MaxElt;
+
+ // We may insert some combination of BITCASTs and VEXT nodes to force Vec to
+ // be compatible with the shuffle we intend to construct. As a result
+ // ShuffleVec will be some sliding window into the original Vec.
+ SDValue ShuffleVec;
+
+ // Code should guarantee that element i in Vec starts at element "WindowBase
+ // + i * WindowScale in ShuffleVec".
+ int WindowBase;
+ int WindowScale;
+
+ bool operator ==(SDValue OtherVec) { return Vec == OtherVec; }
+ ShuffleSourceInfo(SDValue Vec)
+ : Vec(Vec), MinElt(UINT_MAX), MaxElt(0), ShuffleVec(Vec), WindowBase(0),
+ WindowScale(1) {}
+ };
+ // First gather all vectors used as an immediate source for this BUILD_VECTOR
+ // node.
+ SmallVector<ShuffleSourceInfo, 2> Sources;
for (unsigned i = 0; i < NumElts; ++i) {
SDValue V = Op.getOperand(i);
if (V.getOpcode() == ISD::UNDEF)
return SDValue();
}
- // Record this extraction against the appropriate vector if possible...
+ // Add this element source to the list if it's not already there.
SDValue SourceVec = V.getOperand(0);
- unsigned EltNo = cast<ConstantSDNode>(V.getOperand(1))->getZExtValue();
- bool FoundSource = false;
- for (unsigned j = 0; j < SourceVecs.size(); ++j) {
- if (SourceVecs[j] == SourceVec) {
- if (MinElts[j] > EltNo)
- MinElts[j] = EltNo;
- if (MaxElts[j] < EltNo)
- MaxElts[j] = EltNo;
- FoundSource = true;
- break;
- }
- }
+ auto Source = std::find(Sources.begin(), Sources.end(), SourceVec);
+ if (Source == Sources.end())
+ Sources.push_back(ShuffleSourceInfo(SourceVec));
- // Or record a new source if not...
- if (!FoundSource) {
- SourceVecs.push_back(SourceVec);
- MinElts.push_back(EltNo);
- MaxElts.push_back(EltNo);
- }
+ // Update the minimum and maximum lane number seen.
+ unsigned EltNo = cast<ConstantSDNode>(V.getOperand(1))->getZExtValue();
+ Source->MinElt = std::min(Source->MinElt, EltNo);
+ Source->MaxElt = std::max(Source->MaxElt, EltNo);
}
// Currently only do something sane when at most two source vectors
- // involved.
- if (SourceVecs.size() > 2)
+ // are involved.
+ if (Sources.size() > 2)
return SDValue();
- SDValue ShuffleSrcs[2] = { DAG.getUNDEF(VT), DAG.getUNDEF(VT) };
- int VEXTOffsets[2] = { 0, 0 };
- int OffsetMultipliers[2] = { 1, 1 };
-
- // This loop extracts the usage patterns of the source vectors
- // and prepares appropriate SDValues for a shuffle if possible.
- for (unsigned i = 0; i < SourceVecs.size(); ++i) {
- unsigned NumSrcElts = SourceVecs[i].getValueType().getVectorNumElements();
- SDValue CurSource = SourceVecs[i];
- if (SourceVecs[i].getValueType().getVectorElementType() !=
- VT.getVectorElementType()) {
- // It may hit this case if SourceVecs[i] is AssertSext/AssertZext.
- // Then bitcast it to the vector which holds asserted element type,
- // and record the multiplier of element width between SourceVecs and
- // Build_vector which is needed to extract the correct lanes later.
- EVT CastVT =
- EVT::getVectorVT(*DAG.getContext(), VT.getVectorElementType(),
- SourceVecs[i].getValueSizeInBits() /
- VT.getVectorElementType().getSizeInBits());
-
- CurSource = DAG.getNode(ISD::BITCAST, dl, CastVT, SourceVecs[i]);
- OffsetMultipliers[i] = CastVT.getVectorNumElements() / NumSrcElts;
- NumSrcElts *= OffsetMultipliers[i];
- MaxElts[i] *= OffsetMultipliers[i];
- MinElts[i] *= OffsetMultipliers[i];
+ // Find out the smallest element size among result and two sources, and use
+ // it as element size to build the shuffle_vector.
+ EVT SmallestEltTy = VT.getVectorElementType();
+ for (auto &Source : Sources) {
+ EVT SrcEltTy = Source.Vec.getValueType().getVectorElementType();
+ if (SrcEltTy.bitsLT(SmallestEltTy)) {
+ SmallestEltTy = SrcEltTy;
}
+ }
+ unsigned ResMultiplier =
+ VT.getVectorElementType().getSizeInBits() / SmallestEltTy.getSizeInBits();
+ NumElts = VT.getSizeInBits() / SmallestEltTy.getSizeInBits();
+ EVT ShuffleVT = EVT::getVectorVT(*DAG.getContext(), SmallestEltTy, NumElts);
+
+ // If the source vector is too wide or too narrow, we may nevertheless be able
+ // to construct a compatible shuffle either by concatenating it with UNDEF or
+ // extracting a suitable range of elements.
+ for (auto &Src : Sources) {
+ EVT SrcVT = Src.ShuffleVec.getValueType();
- if (CurSource.getValueType() == VT) {
- // No VEXT necessary
- ShuffleSrcs[i] = CurSource;
- VEXTOffsets[i] = 0;
+ if (SrcVT.getSizeInBits() == VT.getSizeInBits())
continue;
- } else if (NumSrcElts < NumElts) {
+
+ // This stage of the search produces a source with the same element type as
+ // the original, but with a total width matching the BUILD_VECTOR output.
+ EVT EltVT = SrcVT.getVectorElementType();
+ EVT DestVT = EVT::getVectorVT(*DAG.getContext(), EltVT,
+ VT.getSizeInBits() / EltVT.getSizeInBits());
+
+ if (SrcVT.getSizeInBits() < VT.getSizeInBits()) {
+ assert(2 * SrcVT.getSizeInBits() == VT.getSizeInBits());
// We can pad out the smaller vector for free, so if it's part of a
// shuffle...
- ShuffleSrcs[i] = DAG.getNode(ISD::CONCAT_VECTORS, dl, VT, CurSource,
- DAG.getUNDEF(CurSource.getValueType()));
+ Src.ShuffleVec =
+ DAG.getNode(ISD::CONCAT_VECTORS, dl, DestVT, Src.ShuffleVec,
+ DAG.getUNDEF(Src.ShuffleVec.getValueType()));
continue;
}
- // Since only 64-bit and 128-bit vectors are legal on ARM and
- // we've eliminated the other cases...
- assert(NumSrcElts == 2 * NumElts &&
- "unexpected vector sizes in ReconstructShuffle");
+ assert(SrcVT.getSizeInBits() == 2 * VT.getSizeInBits());
- if (MaxElts[i] - MinElts[i] >= NumElts) {
+ if (Src.MaxElt - Src.MinElt >= NumElts) {
// Span too large for a VEXT to cope
return SDValue();
}
- if (MinElts[i] >= NumElts) {
+ if (Src.MinElt >= NumElts) {
// The extraction can just take the second half
- VEXTOffsets[i] = NumElts;
- ShuffleSrcs[i] = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, VT, CurSource,
- DAG.getIntPtrConstant(NumElts));
- } else if (MaxElts[i] < NumElts) {
+ Src.ShuffleVec =
+ DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, DestVT, Src.ShuffleVec,
+ DAG.getIntPtrConstant(NumElts));
+ Src.WindowBase = -NumElts;
+ } else if (Src.MaxElt < NumElts) {
// The extraction can just take the first half
- VEXTOffsets[i] = 0;
- ShuffleSrcs[i] = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, VT, CurSource,
- DAG.getIntPtrConstant(0));
+ Src.ShuffleVec = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, DestVT,
+ Src.ShuffleVec, DAG.getIntPtrConstant(0));
} else {
// An actual VEXT is needed
- VEXTOffsets[i] = MinElts[i];
- SDValue VEXTSrc1 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, VT, CurSource,
- DAG.getIntPtrConstant(0));
- SDValue VEXTSrc2 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, VT, CurSource,
- DAG.getIntPtrConstant(NumElts));
- unsigned Imm = VEXTOffsets[i] * getExtFactor(VEXTSrc1);
- ShuffleSrcs[i] = DAG.getNode(AArch64ISD::EXT, dl, VT, VEXTSrc1, VEXTSrc2,
- DAG.getConstant(Imm, MVT::i32));
+ SDValue VEXTSrc1 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, DestVT,
+ Src.ShuffleVec, DAG.getIntPtrConstant(0));
+ SDValue VEXTSrc2 =
+ DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, DestVT, Src.ShuffleVec,
+ DAG.getIntPtrConstant(NumElts));
+ unsigned Imm = Src.MinElt * getExtFactor(VEXTSrc1);
+
+ Src.ShuffleVec = DAG.getNode(AArch64ISD::EXT, dl, DestVT, VEXTSrc1,
+ VEXTSrc2, DAG.getConstant(Imm, MVT::i32));
+ Src.WindowBase = -Src.MinElt;
}
}
- SmallVector<int, 8> Mask;
-
- for (unsigned i = 0; i < NumElts; ++i) {
+ // Another possible incompatibility occurs from the vector element types. We
+ // can fix this by bitcasting the source vectors to the same type we intend
+ // for the shuffle.
+ for (auto &Src : Sources) {
+ EVT SrcEltTy = Src.ShuffleVec.getValueType().getVectorElementType();
+ if (SrcEltTy == SmallestEltTy)
+ continue;
+ assert(ShuffleVT.getVectorElementType() == SmallestEltTy);
+ Src.ShuffleVec = DAG.getNode(ISD::BITCAST, dl, ShuffleVT, Src.ShuffleVec);
+ Src.WindowScale = SrcEltTy.getSizeInBits() / SmallestEltTy.getSizeInBits();
+ Src.WindowBase *= Src.WindowScale;
+ }
+
+ // Final sanity check before we try to actually produce a shuffle.
+ DEBUG(
+ for (auto Src : Sources)
+ assert(Src.ShuffleVec.getValueType() == ShuffleVT);
+ );
+
+ // The stars all align, our next step is to produce the mask for the shuffle.
+ SmallVector<int, 8> Mask(ShuffleVT.getVectorNumElements(), -1);
+ int BitsPerShuffleLane = ShuffleVT.getVectorElementType().getSizeInBits();
+ for (unsigned i = 0; i < VT.getVectorNumElements(); ++i) {
SDValue Entry = Op.getOperand(i);
- if (Entry.getOpcode() == ISD::UNDEF) {
- Mask.push_back(-1);
+ if (Entry.getOpcode() == ISD::UNDEF)
continue;
- }
- SDValue ExtractVec = Entry.getOperand(0);
- int ExtractElt =
- cast<ConstantSDNode>(Op.getOperand(i).getOperand(1))->getSExtValue();
- if (ExtractVec == SourceVecs[0]) {
- Mask.push_back(ExtractElt * OffsetMultipliers[0] - VEXTOffsets[0]);
- } else {
- Mask.push_back(ExtractElt * OffsetMultipliers[1] + NumElts -
- VEXTOffsets[1]);
- }
+ auto Src = std::find(Sources.begin(), Sources.end(), Entry.getOperand(0));
+ int EltNo = cast<ConstantSDNode>(Entry.getOperand(1))->getSExtValue();
+
+ // EXTRACT_VECTOR_ELT performs an implicit any_ext; BUILD_VECTOR an implicit
+ // trunc. So only std::min(SrcBits, DestBits) actually get defined in this
+ // segment.
+ EVT OrigEltTy = Entry.getOperand(0).getValueType().getVectorElementType();
+ int BitsDefined = std::min(OrigEltTy.getSizeInBits(),
+ VT.getVectorElementType().getSizeInBits());
+ int LanesDefined = BitsDefined / BitsPerShuffleLane;
+
+ // This source is expected to fill ResMultiplier lanes of the final shuffle,
+ // starting at the appropriate offset.
+ int *LaneMask = &Mask[i * ResMultiplier];
+
+ int ExtractBase = EltNo * Src->WindowScale + Src->WindowBase;
+ ExtractBase += NumElts * (Src - Sources.begin());
+ for (int j = 0; j < LanesDefined; ++j)
+ LaneMask[j] = ExtractBase + j;
}
// Final check before we try to produce nonsense...
- if (isShuffleMaskLegal(Mask, VT))
- return DAG.getVectorShuffle(VT, dl, ShuffleSrcs[0], ShuffleSrcs[1],
- &Mask[0]);
+ if (!isShuffleMaskLegal(Mask, ShuffleVT))
+ return SDValue();
- return SDValue();
+ SDValue ShuffleOps[] = { DAG.getUNDEF(ShuffleVT), DAG.getUNDEF(ShuffleVT) };
+ for (unsigned i = 0; i < Sources.size(); ++i)
+ ShuffleOps[i] = Sources[i].ShuffleVec;
+
+ SDValue Shuffle = DAG.getVectorShuffle(ShuffleVT, dl, ShuffleOps[0],
+ ShuffleOps[1], &Mask[0]);
+ return DAG.getNode(ISD::BITCAST, dl, VT, Shuffle);
}
// check if an EXT instruction can handle the shuffle mask when the
!F->getAttributes().hasAttribute(AttributeSet::FunctionIndex,
Attribute::NoImplicitFloat) &&
(memOpAlign(SrcAlign, DstAlign, 16) ||
- (allowsUnalignedMemoryAccesses(MVT::f128, 0, &Fast) && Fast)))
+ (allowsMisalignedMemoryAccesses(MVT::f128, 0, 1, &Fast) && Fast)))
return MVT::f128;
return Size >= 8 ? MVT::i64 : MVT::i32;
return Inst->getType()->getPrimitiveSizeInBits() <= 128;
}
+bool AArch64TargetLowering::useLoadStackGuardNode() const {
+ return true;
+}
+
TargetLoweringBase::LegalizeTypeAction
AArch64TargetLowering::getPreferredVectorAction(EVT VT) const {
MVT SVT = VT.getSimpleVT();