aboutsummaryrefslogtreecommitdiff
path: root/llvm/lib/CodeGen
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/lib/CodeGen')
-rw-r--r--llvm/lib/CodeGen/AsmPrinter/DwarfCompileUnit.cpp15
-rw-r--r--llvm/lib/CodeGen/AsmPrinter/DwarfCompileUnit.h3
-rw-r--r--llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp22
-rw-r--r--llvm/lib/CodeGen/GlobalISel/CallLowering.cpp9
-rw-r--r--llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp7
-rw-r--r--llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp53
-rw-r--r--llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp18
-rw-r--r--llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp25
-rw-r--r--llvm/lib/CodeGen/TargetLoweringBase.cpp4
9 files changed, 120 insertions, 36 deletions
diff --git a/llvm/lib/CodeGen/AsmPrinter/DwarfCompileUnit.cpp b/llvm/lib/CodeGen/AsmPrinter/DwarfCompileUnit.cpp
index f9d7e76..67f526f 100644
--- a/llvm/lib/CodeGen/AsmPrinter/DwarfCompileUnit.cpp
+++ b/llvm/lib/CodeGen/AsmPrinter/DwarfCompileUnit.cpp
@@ -1292,12 +1292,10 @@ DwarfCompileUnit::getDwarf5OrGNULocationAtom(dwarf::LocationAtom Loc) const {
}
}
-DIE &DwarfCompileUnit::constructCallSiteEntryDIE(DIE &ScopeDIE,
- const DISubprogram *CalleeSP,
- bool IsTail,
- const MCSymbol *PCAddr,
- const MCSymbol *CallAddr,
- unsigned CallReg) {
+DIE &DwarfCompileUnit::constructCallSiteEntryDIE(
+ DIE &ScopeDIE, const DISubprogram *CalleeSP, bool IsTail,
+ const MCSymbol *PCAddr, const MCSymbol *CallAddr, unsigned CallReg,
+ DIType *AllocSiteTy) {
// Insert a call site entry DIE within ScopeDIE.
DIE &CallSiteDIE = createAndAddDIE(getDwarf5OrGNUTag(dwarf::DW_TAG_call_site),
ScopeDIE, nullptr);
@@ -1306,7 +1304,7 @@ DIE &DwarfCompileUnit::constructCallSiteEntryDIE(DIE &ScopeDIE,
// Indirect call.
addAddress(CallSiteDIE, getDwarf5OrGNUAttr(dwarf::DW_AT_call_target),
MachineLocation(CallReg));
- } else {
+ } else if (CalleeSP) {
DIE *CalleeDIE = getOrCreateSubprogramDIE(CalleeSP);
assert(CalleeDIE && "Could not create DIE for call site entry origin");
if (AddLinkageNamesToDeclCallOriginsForTuning(DD) &&
@@ -1351,6 +1349,9 @@ DIE &DwarfCompileUnit::constructCallSiteEntryDIE(DIE &ScopeDIE,
getDwarf5OrGNUAttr(dwarf::DW_AT_call_return_pc), PCAddr);
}
+ if (AllocSiteTy)
+ addType(CallSiteDIE, AllocSiteTy, dwarf::DW_AT_LLVM_alloc_type);
+
return CallSiteDIE;
}
diff --git a/llvm/lib/CodeGen/AsmPrinter/DwarfCompileUnit.h b/llvm/lib/CodeGen/AsmPrinter/DwarfCompileUnit.h
index 09be22c..c2f6ca0 100644
--- a/llvm/lib/CodeGen/AsmPrinter/DwarfCompileUnit.h
+++ b/llvm/lib/CodeGen/AsmPrinter/DwarfCompileUnit.h
@@ -289,7 +289,8 @@ public:
/// the \p CallReg is set to 0.
DIE &constructCallSiteEntryDIE(DIE &ScopeDIE, const DISubprogram *CalleeSP,
bool IsTail, const MCSymbol *PCAddr,
- const MCSymbol *CallAddr, unsigned CallReg);
+ const MCSymbol *CallAddr, unsigned CallReg,
+ DIType *AllocSiteTy);
/// Construct call site parameter DIEs for the \p CallSiteDIE. The \p Params
/// were collected by the \ref collectCallSiteParameters.
/// Note: The order of parameters does not matter, since debuggers recognize
diff --git a/llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp b/llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
index 5ae2d2a..c27f100 100644
--- a/llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
+++ b/llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
@@ -936,6 +936,8 @@ void DwarfDebug::constructCallSiteEntryDIEs(const DISubprogram &SP,
if (MI.hasDelaySlot() && !delaySlotSupported(*&MI))
return;
+ DIType *AllocSiteTy = dyn_cast_or_null<DIType>(MI.getHeapAllocMarker());
+
// If this is a direct call, find the callee's subprogram.
// In the case of an indirect call find the register that holds
// the callee.
@@ -950,23 +952,23 @@ void DwarfDebug::constructCallSiteEntryDIEs(const DISubprogram &SP,
PhysRegCalleeOperand =
PhysRegCalleeOperand && MCOI.OperandType == MCOI::OPERAND_REGISTER;
}
- if (!CalleeOp.isGlobal() && !PhysRegCalleeOperand)
- continue;
unsigned CallReg = 0;
const DISubprogram *CalleeSP = nullptr;
const Function *CalleeDecl = nullptr;
if (PhysRegCalleeOperand) {
- CallReg = CalleeOp.getReg();
- if (!CallReg)
- continue;
- } else {
+ CallReg = CalleeOp.getReg(); // might be zero
+ } else if (CalleeOp.isGlobal()) {
CalleeDecl = dyn_cast<Function>(CalleeOp.getGlobal());
- if (!CalleeDecl || !CalleeDecl->getSubprogram())
- continue;
- CalleeSP = CalleeDecl->getSubprogram();
+ if (CalleeDecl)
+ CalleeSP = CalleeDecl->getSubprogram(); // might be nullptr
}
+ // Omit DIE if we can't tell where the call goes *and* we don't want to
+ // add metadata to it.
+ if (CalleeSP == nullptr && CallReg == 0 && AllocSiteTy == nullptr)
+ continue;
+
// TODO: Omit call site entries for runtime calls (objc_msgSend, etc).
bool IsTail = TII->isTailCall(MI);
@@ -1000,7 +1002,7 @@ void DwarfDebug::constructCallSiteEntryDIEs(const DISubprogram &SP,
<< (IsTail ? " [IsTail]" : "") << "\n");
DIE &CallSiteDIE = CU.constructCallSiteEntryDIE(
- ScopeDIE, CalleeSP, IsTail, PCAddr, CallAddr, CallReg);
+ ScopeDIE, CalleeSP, IsTail, PCAddr, CallAddr, CallReg, AllocSiteTy);
// Optionally emit call-site-param debug info.
if (emitDebugEntryValues()) {
diff --git a/llvm/lib/CodeGen/GlobalISel/CallLowering.cpp b/llvm/lib/CodeGen/GlobalISel/CallLowering.cpp
index 9ba1782..0f3ec8b 100644
--- a/llvm/lib/CodeGen/GlobalISel/CallLowering.cpp
+++ b/llvm/lib/CodeGen/GlobalISel/CallLowering.cpp
@@ -132,9 +132,10 @@ bool CallLowering::lowerCall(MachineIRBuilder &MIRBuilder, const CallBase &CB,
unsigned i = 0;
unsigned NumFixedArgs = CB.getFunctionType()->getNumParams();
for (const auto &Arg : CB.args()) {
- ArgInfo OrigArg{ArgRegs[i], *Arg.get(), i, getAttributesForArgIdx(CB, i),
- i < NumFixedArgs};
+ ArgInfo OrigArg{ArgRegs[i], *Arg.get(), i, getAttributesForArgIdx(CB, i)};
setArgFlags(OrigArg, i + AttributeList::FirstArgIndex, DL, CB);
+ if (i >= NumFixedArgs)
+ OrigArg.Flags[0].setVarArg();
// If we have an explicit sret argument that is an Instruction, (i.e., it
// might point to function-local memory), we can't meaningfully tail-call.
@@ -301,7 +302,7 @@ void CallLowering::splitToValueTypes(const ArgInfo &OrigArg,
// double] -> double).
SplitArgs.emplace_back(OrigArg.Regs[0], SplitVTs[0].getTypeForEVT(Ctx),
OrigArg.OrigArgIndex, OrigArg.Flags[0],
- OrigArg.IsFixed, OrigArg.OrigValue);
+ OrigArg.OrigValue);
return;
}
@@ -313,7 +314,7 @@ void CallLowering::splitToValueTypes(const ArgInfo &OrigArg,
for (unsigned i = 0, e = SplitVTs.size(); i < e; ++i) {
Type *SplitTy = SplitVTs[i].getTypeForEVT(Ctx);
SplitArgs.emplace_back(OrigArg.Regs[i], SplitTy, OrigArg.OrigArgIndex,
- OrigArg.Flags[0], OrigArg.IsFixed);
+ OrigArg.Flags[0]);
if (NeedsRegBlock)
SplitArgs.back().Flags[0].setInConsecutiveRegs();
}
diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index d70e96938..7341914 100644
--- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -9390,8 +9390,7 @@ SDValue DAGCombiner::mergeTruncStores(StoreSDNode *N) {
LLVMContext &Context = *DAG.getContext();
unsigned NumStores = Stores.size();
unsigned WideNumBits = NumStores * NarrowNumBits;
- EVT WideVT = EVT::getIntegerVT(Context, WideNumBits);
- if (WideVT != MVT::i16 && WideVT != MVT::i32 && WideVT != MVT::i64)
+ if (WideNumBits != 16 && WideNumBits != 32 && WideNumBits != 64)
return SDValue();
// Check if all bytes of the source value that we are looking at are stored
@@ -9445,7 +9444,7 @@ SDValue DAGCombiner::mergeTruncStores(StoreSDNode *N) {
SourceValue = WideVal;
// Give up if the source value type is smaller than the store size.
- if (SourceValue.getScalarValueSizeInBits() < WideVT.getScalarSizeInBits())
+ if (SourceValue.getScalarValueSizeInBits() < WideNumBits)
return SDValue();
}
@@ -9469,6 +9468,8 @@ SDValue DAGCombiner::mergeTruncStores(StoreSDNode *N) {
OffsetMap[Offset] = ByteOffsetFromBase;
}
+ EVT WideVT = EVT::getIntegerVT(Context, WideNumBits);
+
assert(FirstOffset != INT64_MAX && "First byte offset must be set");
assert(FirstStore && "First store must be set");
diff --git a/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp b/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
index 71a175d..b9e72c9 100644
--- a/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
@@ -5630,6 +5630,7 @@ bool SelectionDAG::canCreateUndefOrPoison(SDValue Op, const APInt &DemandedElts,
case ISD::FDIV:
case ISD::FREM:
case ISD::FCOPYSIGN:
+ case ISD::FP_EXTEND:
// No poison except from flags (which is handled above)
return false;
@@ -6422,6 +6423,20 @@ SDValue SelectionDAG::getNode(unsigned Opcode, const SDLoc &DL, EVT VT,
if (N1.isUndef())
// sext(undef) = 0, because the top bits will all be the same.
return getConstant(0, DL, VT);
+
+ // Skip unnecessary sext_inreg pattern:
+ // (sext (trunc x)) -> x iff the upper bits are all signbits.
+ if (OpOpcode == ISD::TRUNCATE) {
+ SDValue OpOp = N1.getOperand(0);
+ if (OpOp.getValueType() == VT) {
+ unsigned NumSignExtBits =
+ VT.getScalarSizeInBits() - N1.getScalarValueSizeInBits();
+ if (ComputeNumSignBits(OpOp) > NumSignExtBits) {
+ transferDbgValues(N1, OpOp);
+ return OpOp;
+ }
+ }
+ }
break;
case ISD::ZERO_EXTEND:
assert(VT.isInteger() && N1.getValueType().isInteger() &&
@@ -8874,6 +8889,44 @@ static void checkAddrSpaceIsValidForLibcall(const TargetLowering *TLI,
}
}
+std::pair<SDValue, SDValue>
+SelectionDAG::getMemcmp(SDValue Chain, const SDLoc &dl, SDValue Mem0,
+ SDValue Mem1, SDValue Size, const CallInst *CI) {
+ const char *LibCallName = TLI->getLibcallName(RTLIB::MEMCMP);
+ if (!LibCallName)
+ return {};
+
+ // Emit a library call.
+ auto GetEntry = [](Type *Ty, SDValue &SDV) {
+ TargetLowering::ArgListEntry E;
+ E.Ty = Ty;
+ E.Node = SDV;
+ return E;
+ };
+
+ PointerType *PT = PointerType::getUnqual(*getContext());
+ TargetLowering::ArgListTy Args = {
+ GetEntry(PT, Mem0), GetEntry(PT, Mem1),
+ GetEntry(getDataLayout().getIntPtrType(*getContext()), Size)};
+
+ TargetLowering::CallLoweringInfo CLI(*this);
+ bool IsTailCall = false;
+ bool ReturnsFirstArg = CI && funcReturnsFirstArgOfCall(*CI);
+ IsTailCall = CI && CI->isTailCall() &&
+ isInTailCallPosition(*CI, getTarget(), ReturnsFirstArg);
+
+ CLI.setDebugLoc(dl)
+ .setChain(Chain)
+ .setLibCallee(
+ TLI->getLibcallCallingConv(RTLIB::MEMCMP),
+ Type::getInt32Ty(*getContext()),
+ getExternalSymbol(LibCallName, TLI->getPointerTy(getDataLayout())),
+ std::move(Args))
+ .setTailCall(IsTailCall);
+
+ return TLI->LowerCallTo(CLI);
+}
+
SDValue SelectionDAG::getMemcpy(
SDValue Chain, const SDLoc &dl, SDValue Dst, SDValue Src, SDValue Size,
Align Alignment, bool isVol, bool AlwaysInline, const CallInst *CI,
diff --git a/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp b/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
index d0815e9..f5f5c14 100644
--- a/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
@@ -2273,9 +2273,8 @@ void SelectionDAGBuilder::visitRet(const ReturnInst &I) {
Flags.setNoExt();
for (unsigned i = 0; i < NumParts; ++i) {
- Outs.push_back(ISD::OutputArg(Flags,
- Parts[i].getValueType().getSimpleVT(),
- VT, /*isfixed=*/true, 0, 0));
+ Outs.push_back(ISD::OutputArg(
+ Flags, Parts[i].getValueType().getSimpleVT(), VT, 0, 0));
OutVals.push_back(Parts[i]);
}
}
@@ -2291,9 +2290,9 @@ void SelectionDAGBuilder::visitRet(const ReturnInst &I) {
assert(SwiftError.getFunctionArg() && "Need a swift error argument");
ISD::ArgFlagsTy Flags = ISD::ArgFlagsTy();
Flags.setSwiftError();
- Outs.push_back(ISD::OutputArg(
- Flags, /*vt=*/TLI.getPointerTy(DL), /*argvt=*/EVT(TLI.getPointerTy(DL)),
- /*isfixed=*/true, /*origidx=*/1, /*partOffs=*/0));
+ Outs.push_back(ISD::OutputArg(Flags, /*vt=*/TLI.getPointerTy(DL),
+ /*argvt=*/EVT(TLI.getPointerTy(DL)),
+ /*origidx=*/1, /*partOffs=*/0));
// Create SDNode for the swifterror virtual register.
OutVals.push_back(
DAG.getRegister(SwiftError.getOrCreateVRegUseAt(
@@ -9091,7 +9090,7 @@ bool SelectionDAGBuilder::visitMemCmpBCmpCall(const CallInst &I) {
const SelectionDAGTargetInfo &TSI = DAG.getSelectionDAGInfo();
std::pair<SDValue, SDValue> Res = TSI.EmitTargetCodeForMemcmp(
DAG, getCurSDLoc(), DAG.getRoot(), getValue(LHS), getValue(RHS),
- getValue(Size), MachinePointerInfo(LHS), MachinePointerInfo(RHS));
+ getValue(Size), &I);
if (Res.first.getNode()) {
processIntegerCallValue(I, Res.first, true);
PendingLoads.push_back(Res.second);
@@ -11124,6 +11123,8 @@ TargetLowering::LowerCallTo(TargetLowering::CallLoweringInfo &CLI) const {
const Align OriginalAlignment(getABIAlignmentForCallingConv(ArgTy, DL));
Flags.setOrigAlign(OriginalAlignment);
+ if (i >= CLI.NumFixedArgs)
+ Flags.setVarArg();
if (Args[i].Ty->isPointerTy()) {
Flags.setPointer();
Flags.setPointerAddrSpace(
@@ -11246,8 +11247,7 @@ TargetLowering::LowerCallTo(TargetLowering::CallLoweringInfo &CLI) const {
// For scalable vectors the scalable part is currently handled
// by individual targets, so we just use the known minimum size here.
ISD::OutputArg MyFlags(
- Flags, Parts[j].getValueType().getSimpleVT(), VT,
- i < CLI.NumFixedArgs, i,
+ Flags, Parts[j].getValueType().getSimpleVT(), VT, i,
j * Parts[j].getValueType().getStoreSize().getKnownMinValue());
if (NumParts > 1 && j == 0)
MyFlags.Flags.setSplit();
diff --git a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
index a68f521..e235d14 100644
--- a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
@@ -5118,6 +5118,20 @@ SDValue TargetLowering::SimplifySetCC(EVT VT, SDValue N0, SDValue N1,
Cond == ISD::SETEQ ? ISD::SETLT : ISD::SETGE);
}
+ // fold (setcc (trunc x) c) -> (setcc x c)
+ if (N0.getOpcode() == ISD::TRUNCATE &&
+ ((N0->getFlags().hasNoUnsignedWrap() && !ISD::isSignedIntSetCC(Cond)) ||
+ (N0->getFlags().hasNoSignedWrap() &&
+ !ISD::isUnsignedIntSetCC(Cond))) &&
+ isTypeDesirableForOp(ISD::SETCC, N0.getOperand(0).getValueType())) {
+ EVT NewVT = N0.getOperand(0).getValueType();
+ SDValue NewConst = DAG.getConstant(ISD::isSignedIntSetCC(Cond)
+ ? C1.sext(NewVT.getSizeInBits())
+ : C1.zext(NewVT.getSizeInBits()),
+ dl, NewVT);
+ return DAG.getSetCC(dl, VT, N0.getOperand(0), NewConst, Cond);
+ }
+
if (SDValue V =
optimizeSetCCOfSignedTruncationCheck(VT, N0, N1, Cond, DCI, dl))
return V;
@@ -5654,6 +5668,17 @@ SDValue TargetLowering::SimplifySetCC(EVT VT, SDValue N0, SDValue N1,
return N0;
}
+ // Fold (setcc (trunc x) (trunc y)) -> (setcc x y)
+ if (N0.getOpcode() == ISD::TRUNCATE && N1.getOpcode() == ISD::TRUNCATE &&
+ N0.getOperand(0).getValueType() == N1.getOperand(0).getValueType() &&
+ ((!ISD::isSignedIntSetCC(Cond) && N0->getFlags().hasNoUnsignedWrap() &&
+ N1->getFlags().hasNoUnsignedWrap()) ||
+ (!ISD::isUnsignedIntSetCC(Cond) && N0->getFlags().hasNoSignedWrap() &&
+ N1->getFlags().hasNoSignedWrap())) &&
+ isTypeDesirableForOp(ISD::SETCC, N0.getOperand(0).getValueType())) {
+ return DAG.getSetCC(dl, VT, N0.getOperand(0), N1.getOperand(0), Cond);
+ }
+
// Could not fold it.
return SDValue();
}
diff --git a/llvm/lib/CodeGen/TargetLoweringBase.cpp b/llvm/lib/CodeGen/TargetLoweringBase.cpp
index 9f525ea..bf4c9f9 100644
--- a/llvm/lib/CodeGen/TargetLoweringBase.cpp
+++ b/llvm/lib/CodeGen/TargetLoweringBase.cpp
@@ -1008,7 +1008,7 @@ unsigned TargetLoweringBase::getBitWidthForCttzElements(
CR = CR.subtract(APInt(64, 1));
unsigned EltWidth = RetTy->getScalarSizeInBits();
- EltWidth = std::min(EltWidth, (unsigned)CR.getActiveBits());
+ EltWidth = std::min(EltWidth, CR.getActiveBits());
EltWidth = std::max(llvm::bit_ceil(EltWidth), (unsigned)8);
return EltWidth;
@@ -1772,7 +1772,7 @@ void llvm::GetReturnInfo(CallingConv::ID CC, Type *ReturnType,
Flags.setZExt();
for (unsigned i = 0; i < NumParts; ++i)
- Outs.push_back(ISD::OutputArg(Flags, PartVT, VT, /*isfixed=*/true, 0, 0));
+ Outs.push_back(ISD::OutputArg(Flags, PartVT, VT, 0, 0));
}
}
generated by cgit v1.1 at 2025-08-07 21:40:52 +0000