LLVM 22.0.0git
Public Member Functions | Static Public Member Functions | List of all members
llvm::AMDGPUCallLowering Class Referencefinal

#include "Target/AMDGPU/AMDGPUCallLowering.h"

Inheritance diagram for llvm::AMDGPUCallLowering:
[legend]

Public Member Functions

 AMDGPUCallLowering (const AMDGPUTargetLowering &TLI)
bool lowerReturn (MachineIRBuilder &B, const Value *Val, ArrayRef< Register > VRegs, FunctionLoweringInfo &FLI) const override
 This hook behaves as the extended lowerReturn function, but for targets that do not support swifterror value promotion.
bool lowerFormalArgumentsKernel (MachineIRBuilder &B, const Function &F, ArrayRef< ArrayRef< Register > > VRegs) const
bool lowerFormalArguments (MachineIRBuilder &B, const Function &F, ArrayRef< ArrayRef< Register > > VRegs, FunctionLoweringInfo &FLI) const override
 This hook must be implemented to lower the incoming (formal) arguments, described by VRegs, for GlobalISel.
bool passSpecialInputs (MachineIRBuilder &MIRBuilder, CCState &CCInfo, SmallVectorImpl< std::pair< MCRegister, Register > > &ArgRegs, CallLoweringInfo &Info) const
bool doCallerAndCalleePassArgsTheSameWay (CallLoweringInfo &Info, MachineFunction &MF, SmallVectorImpl< ArgInfo > &InArgs) const
bool areCalleeOutgoingArgsTailCallable (CallLoweringInfo &Info, MachineFunction &MF, SmallVectorImpl< ArgInfo > &OutArgs) const
bool isEligibleForTailCallOptimization (MachineIRBuilder &MIRBuilder, CallLoweringInfo &Info, SmallVectorImpl< ArgInfo > &InArgs, SmallVectorImpl< ArgInfo > &OutArgs) const
 Returns true if the call can be lowered as a tail call.
void handleImplicitCallArguments (MachineIRBuilder &MIRBuilder, MachineInstrBuilder &CallInst, const GCNSubtarget &ST, const SIMachineFunctionInfo &MFI, CallingConv::ID CalleeCC, ArrayRef< std::pair< MCRegister, Register > > ImplicitArgRegs) const
bool lowerTailCall (MachineIRBuilder &MIRBuilder, CallLoweringInfo &Info, SmallVectorImpl< ArgInfo > &OutArgs) const
bool lowerChainCall (MachineIRBuilder &MIRBuilder, CallLoweringInfo &Info) const
 Lower a call to the @llvm.amdgcn.cs.chain intrinsic.
bool lowerCall (MachineIRBuilder &MIRBuilder, CallLoweringInfo &Info) const override
 This hook must be implemented to lower the given call instruction, including argument and return value marshalling.
Public Member Functions inherited from llvm::CallLowering
 CallLowering (const TargetLowering *TLI)
virtual ~CallLowering ()=default
virtual bool supportSwiftError () const
void insertSRetLoads (MachineIRBuilder &MIRBuilder, Type *RetTy, ArrayRef< Register > VRegs, Register DemoteReg, int FI) const
 Load the returned value from the stack into virtual registers in VRegs.
void insertSRetStores (MachineIRBuilder &MIRBuilder, Type *RetTy, ArrayRef< Register > VRegs, Register DemoteReg) const
 Store the return value given by VRegs into stack starting at the offset specified in DemoteReg.
void insertSRetIncomingArgument (const Function &F, SmallVectorImpl< ArgInfo > &SplitArgs, Register &DemoteReg, MachineRegisterInfo &MRI, const DataLayout &DL) const
 Insert the hidden sret ArgInfo to the beginning of SplitArgs.
void insertSRetOutgoingArgument (MachineIRBuilder &MIRBuilder, const CallBase &CB, CallLoweringInfo &Info) const
 For the call-base described by CB, insert the hidden sret ArgInfo to the OrigArgs field of Info.
bool checkReturn (CCState &CCInfo, SmallVectorImpl< BaseArgInfo > &Outs, CCAssignFn *Fn) const
void getReturnInfo (CallingConv::ID CallConv, Type *RetTy, AttributeList Attrs, SmallVectorImpl< BaseArgInfo > &Outs, const DataLayout &DL) const
 Get the type and the ArgFlags for the split components of RetTy as returned by ComputeValueVTs.
bool checkReturnTypeForCallConv (MachineFunction &MF) const
 Toplevel function to check the return type based on the target calling convention.
virtual bool lowerReturn (MachineIRBuilder &MIRBuilder, const Value *Val, ArrayRef< Register > VRegs, FunctionLoweringInfo &FLI, Register SwiftErrorVReg) const
 This hook must be implemented to lower outgoing return values, described by Val, into the specified virtual registers VRegs.
virtual bool fallBackToDAGISel (const MachineFunction &MF) const
bool lowerCall (MachineIRBuilder &MIRBuilder, const CallBase &Call, ArrayRef< Register > ResRegs, ArrayRef< ArrayRef< Register > > ArgRegs, Register SwiftErrorVReg, std::optional< PtrAuthInfo > PAI, Register ConvergenceCtrlToken, std::function< Register()> GetCalleeReg) const
 Lower the given call instruction, including argument and return value marshalling.
virtual bool enableBigEndian () const
 For targets which want to use big-endian can enable it with enableBigEndian() hook.
virtual bool isTypeIsValidForThisReturn (EVT Ty) const
 For targets which support the "returned" parameter attribute, returns true if the given type is a valid one to use with "returned".

Static Public Member Functions

static CCAssignFn * CCAssignFnForCall (CallingConv::ID CC, bool IsVarArg)
static CCAssignFn * CCAssignFnForReturn (CallingConv::ID CC, bool IsVarArg)

Additional Inherited Members

Protected Member Functions inherited from llvm::CallLowering
const TargetLowering * getTLI () const
 Getter for generic TargetLowering class.
template<class XXXTargetLowering>
const XXXTargetLowering * getTLI () const
 Getter for target specific TargetLowering class.
ISD::ArgFlagsTy getAttributesForArgIdx (const CallBase &Call, unsigned ArgIdx) const
ISD::ArgFlagsTy getAttributesForReturn (const CallBase &Call) const
void addArgFlagsFromAttributes (ISD::ArgFlagsTy &Flags, const AttributeList &Attrs, unsigned OpIdx) const
 Adds flags to Flags based off of the attributes in Attrs.
template<typename FuncInfoTy>
void setArgFlags (ArgInfo &Arg, unsigned OpIdx, const DataLayout &DL, const FuncInfoTy &FuncInfo) const
void splitToValueTypes (const ArgInfo &OrigArgInfo, SmallVectorImpl< ArgInfo > &SplitArgs, const DataLayout &DL, CallingConv::ID CallConv, SmallVectorImpl< uint64_t > *Offsets=nullptr) const
 Break OrigArgInfo into one or more pieces the calling convention can process, returned in SplitArgs.
bool determineAssignments (ValueAssigner &Assigner, SmallVectorImpl< ArgInfo > &Args, CCState &CCInfo) const
 Analyze the argument list in Args, using Assigner to populate CCInfo.
bool determineAndHandleAssignments (ValueHandler &Handler, ValueAssigner &Assigner, SmallVectorImpl< ArgInfo > &Args, MachineIRBuilder &MIRBuilder, CallingConv::ID CallConv, bool IsVarArg, ArrayRef< Register > ThisReturnRegs={}) const
 Invoke ValueAssigner::assignArg on each of the given Args and then use Handler to move them to the assigned locations.
bool handleAssignments (ValueHandler &Handler, SmallVectorImpl< ArgInfo > &Args, CCState &CCState, SmallVectorImpl< CCValAssign > &ArgLocs, MachineIRBuilder &MIRBuilder, ArrayRef< Register > ThisReturnRegs={}) const
 Use Handler to insert code to handle the argument/return values represented by Args.
bool parametersInCSRMatch (const MachineRegisterInfo &MRI, const uint32_t *CallerPreservedMask, const SmallVectorImpl< CCValAssign > &ArgLocs, const SmallVectorImpl< ArgInfo > &OutVals) const
 Check whether parameters to a call that are passed in callee saved registers are the same as from the calling function.
bool resultsCompatible (CallLoweringInfo &Info, MachineFunction &MF, SmallVectorImpl< ArgInfo > &InArgs, ValueAssigner &CalleeAssigner, ValueAssigner &CallerAssigner) const

Detailed Description

Definition at line 26 of file AMDGPUCallLowering.h.

Constructor & Destructor Documentation

◆ AMDGPUCallLowering()

AMDGPUCallLowering::AMDGPUCallLowering ( const AMDGPUTargetLowering & TLI)

Definition at line 257 of file AMDGPUCallLowering.cpp.

References llvm::CallLowering::CallLowering().

Member Function Documentation

◆ areCalleeOutgoingArgsTailCallable()

bool AMDGPUCallLowering::areCalleeOutgoingArgsTailCallable ( CallLoweringInfo & Info,
MachineFunction & MF,
SmallVectorImpl< ArgInfo > & OutArgs ) const

Definition at line 1088 of file AMDGPUCallLowering.cpp.

References llvm::dbgs(), llvm::CallLowering::determineAssignments(), llvm::SmallVectorTemplateCommon< T, typename >::empty(), getAssignFnsForCC(), llvm::SIMachineFunctionInfo::getBytesInStackArgArea(), llvm::Function::getCallingConv(), llvm::Function::getContext(), llvm::MachineFunction::getFunction(), llvm::MachineFunction::getInfo(), llvm::MachineFunction::getRegInfo(), llvm::CCState::getStackSize(), llvm::MachineFunction::getSubtarget(), llvm::CallLowering::getTLI(), LLVM_DEBUG, MRI, llvm::CallLowering::parametersInCSRMatch(), and TRI.

Referenced by isEligibleForTailCallOptimization().

◆ CCAssignFnForCall()

CCAssignFn * AMDGPUCallLowering::CCAssignFnForCall ( CallingConv::ID CC,
bool IsVarArg )
static

Definition at line 1132 of file AMDGPUISelLowering.cpp.

References llvm::CallingConv::AMDGPU_CS, llvm::CallingConv::AMDGPU_CS_Chain, llvm::CallingConv::AMDGPU_CS_ChainPreserve, llvm::CallingConv::AMDGPU_ES, llvm::CallingConv::AMDGPU_Gfx, llvm::CallingConv::AMDGPU_Gfx_WholeWave, llvm::CallingConv::AMDGPU_GS, llvm::CallingConv::AMDGPU_HS, llvm::CallingConv::AMDGPU_KERNEL, llvm::CallingConv::AMDGPU_LS, llvm::CallingConv::AMDGPU_PS, llvm::CallingConv::AMDGPU_VS, llvm::CallingConv::C, llvm::CallingConv::Cold, llvm::CallingConv::Fast, llvm::reportFatalUsageError(), and llvm::CallingConv::SPIR_KERNEL.

Referenced by llvm::AMDGPUTargetLowering::CCAssignFnForCall().

◆ CCAssignFnForReturn()

CCAssignFn * AMDGPUCallLowering::CCAssignFnForReturn ( CallingConv::ID CC,
bool IsVarArg )
static

Definition at line 1160 of file AMDGPUISelLowering.cpp.

References llvm::CallingConv::AMDGPU_CS, llvm::CallingConv::AMDGPU_CS_Chain, llvm::CallingConv::AMDGPU_CS_ChainPreserve, llvm::CallingConv::AMDGPU_ES, llvm::CallingConv::AMDGPU_Gfx, llvm::CallingConv::AMDGPU_Gfx_WholeWave, llvm::CallingConv::AMDGPU_GS, llvm::CallingConv::AMDGPU_HS, llvm::CallingConv::AMDGPU_KERNEL, llvm::CallingConv::AMDGPU_LS, llvm::CallingConv::AMDGPU_PS, llvm::CallingConv::AMDGPU_VS, llvm::CallingConv::C, llvm::CallingConv::Cold, llvm::CallingConv::Fast, llvm_unreachable, llvm::reportFatalUsageError(), and llvm::CallingConv::SPIR_KERNEL.

Referenced by llvm::AMDGPUTargetLowering::CCAssignFnForReturn().

◆ doCallerAndCalleePassArgsTheSameWay()

bool AMDGPUCallLowering::doCallerAndCalleePassArgsTheSameWay ( CallLoweringInfo & Info,
MachineFunction & MF,
SmallVectorImpl< ArgInfo > & InArgs ) const

Definition at line 1046 of file AMDGPUCallLowering.cpp.

References getAssignFnsForCC(), llvm::Function::getCallingConv(), llvm::MachineFunction::getFunction(), llvm::MachineFunction::getSubtarget(), llvm::CallLowering::getTLI(), llvm::CallLowering::resultsCompatible(), and TRI.

Referenced by isEligibleForTailCallOptimization().

◆ handleImplicitCallArguments()

void AMDGPUCallLowering::handleImplicitCallArguments ( MachineIRBuilder & MIRBuilder,
MachineInstrBuilder & CallInst,
const GCNSubtarget & ST,
const SIMachineFunctionInfo & MFI,
CallingConv::ID CalleeCC,
ArrayRef< std::pair< MCRegister, Register > > ImplicitArgRegs ) const

Definition at line 1194 of file AMDGPUCallLowering.cpp.

References llvm::MachineIRBuilder::buildCopy(), llvm::LLT::fixed_vector(), llvm::SIMachineFunctionInfo::getScratchRSrcReg(), llvm::RegState::Implicit, and llvm::AMDGPU::isChainCC().

Referenced by lowerCall(), and lowerTailCall().

◆ isEligibleForTailCallOptimization()

bool AMDGPUCallLowering::isEligibleForTailCallOptimization ( MachineIRBuilder & MIRBuilder,
CallLoweringInfo & Info,
SmallVectorImpl< ArgInfo > & InArgs,
SmallVectorImpl< ArgInfo > & OutArgs ) const

Returns true if the call can be lowered as a tail call.

Definition at line 1129 of file AMDGPUCallLowering.cpp.

References A(), llvm::any_of(), areCalleeOutgoingArgsTailCallable(), llvm::Function::args(), B(), llvm::AMDGPU::canGuaranteeTCO(), llvm::dbgs(), doCallerAndCalleePassArgsTheSameWay(), llvm::Function::getCallingConv(), llvm::MachineFunction::getFunction(), llvm::MachineFunction::getSubtarget(), llvm::MachineFunction::getTarget(), llvm::TargetOptions::GuaranteedTailCallOpt, LLVM_DEBUG, llvm::AMDGPU::mayTailCallThisCC(), llvm::TargetMachine::Options, and TRI.

Referenced by lowerCall().

◆ lowerCall()

bool AMDGPUCallLowering::lowerCall ( MachineIRBuilder & MIRBuilder,
CallLoweringInfo & Info ) const
overridevirtual

This hook must be implemented to lower the given call instruction, including argument and return value marshalling.

Returns
true if the lowering succeeded, false otherwise.

Reimplemented from llvm::CallLowering.

Definition at line 1498 of file AMDGPUCallLowering.cpp.

References addCallTargetOperands(), llvm::MachineInstrBuilder::addDef(), llvm::MachineInstrBuilder::addImm(), llvm::CallingConv::AMDGPU_Gfx, llvm::CallingConv::AMDGPU_Gfx_WholeWave, llvm::MachineIRBuilder::buildInstr(), llvm::MachineIRBuilder::buildInstrNoInsert(), llvm::cast(), llvm::constrainOperandRegClass(), llvm::MachineOperand::CreateGA(), llvm::dbgs(), llvm::CallLowering::determineAndHandleAssignments(), llvm::CallLowering::determineAssignments(), DL, F, getAssignFnsForCC(), getCallOpcode(), llvm::MachineFunction::getFunction(), llvm::MachineFunction::getInfo(), llvm::MachineIRBuilder::getMF(), llvm::MachineFunction::getRegInfo(), llvm::CCState::getStackSize(), llvm::MachineFunction::getSubtarget(), llvm::CallLowering::getTLI(), llvm::CallLowering::handleAssignments(), handleImplicitCallArguments(), llvm::RegState::Implicit, llvm::MachineIRBuilder::insertInstr(), llvm::CallLowering::insertSRetLoads(), isEligibleForTailCallOptimization(), LLVM_DEBUG, llvm_unreachable, lowerChainCall(), lowerTailCall(), MRI, llvm::MachineInstr::NoConvergent, Opc, passSpecialInputs(), llvm::MachineInstrBuilder::setMIFlag(), llvm::CallLowering::splitToValueTypes(), and TRI.

◆ lowerChainCall()

bool AMDGPUCallLowering::lowerChainCall ( MachineIRBuilder & MIRBuilder,
CallLoweringInfo & Info ) const

Lower a call to the @llvm.amdgcn.cs.chain intrinsic.

Definition at line 1457 of file AMDGPUCallLowering.cpp.

References llvm::all_of(), llvm::CallingConv::AMDGPU_CS_Chain, assert(), llvm::MachineOperand::CreateGA(), llvm::MachineOperand::CreateReg(), DL, llvm::dyn_cast(), F, llvm::CallLowering::BaseArgInfo::Flags, llvm::MachineFunction::getFunction(), llvm::MachineIRBuilder::getMF(), lowerTailCall(), llvm::none_of(), and llvm::CallLowering::splitToValueTypes().

Referenced by lowerCall().

◆ lowerFormalArguments()

bool AMDGPUCallLowering::lowerFormalArguments ( MachineIRBuilder & MIRBuilder,
const Function & F,
ArrayRef< ArrayRef< Register > > VRegs,
FunctionLoweringInfo & FLI ) const
overridevirtual

This hook must be implemented to lower the incoming (formal) arguments, described by VRegs, for GlobalISel.

Each argument must end up in the related virtual registers described by VRegs. In other words, the first argument should end up in VRegs[0], the second in VRegs[1], and so on. For each argument, there will be one register for each non-aggregate type, as returned by computeValueLLTs. MIRBuilder is set to the proper insertion for the argument lowering. FLI is required for sret demotion.

Returns
True if the lowering succeeded, false otherwise.

Reimplemented from llvm::CallLowering.

Definition at line 591 of file AMDGPUCallLowering.cpp.

References llvm::MachineFunction::addLiveIn(), llvm::CCState::AllocateReg(), llvm::CallingConv::AMDGPU_KERNEL, llvm::CallingConv::AMDGPU_PS, assert(), B(), llvm::FunctionLoweringInfo::CanLowerReturn, llvm::countr_zero(), llvm::FunctionLoweringInfo::DemoteRegister, llvm::CallLowering::determineAssignments(), DL, F, llvm::CCState::getFirstUnallocated(), llvm::MachineFunction::getInfo(), llvm::MachineFunction::getRegInfo(), llvm::MachineFunction::getSubtarget(), llvm::CallLowering::getTLI(), llvm::CallLowering::handleAssignments(), llvm::GCNUserSGPRUsageInfo::hasFlatScratchInit(), llvm::GCNUserSGPRUsageInfo::hasImplicitBufferPtr(), llvm::CallLowering::insertSRetIncomingArgument(), llvm::AMDGPU::isEntryFunctionCC(), llvm::AMDGPU::isGraphics(), lowerFormalArgumentsKernel(), MBB, MRI, llvm::CallLowering::setArgFlags(), llvm::size(), llvm::CallLowering::splitToValueTypes(), llvm::CallLowering::ValueAssigner::StackSize, and TRI.

◆ lowerFormalArgumentsKernel()

bool AMDGPUCallLowering::lowerFormalArgumentsKernel ( MachineIRBuilder & B,
const Function & F,
ArrayRef< ArrayRef< Register > > VRegs ) const

Definition at line 511 of file AMDGPUCallLowering.cpp.

References llvm::alignTo(), allocateHSAUserSGPRs(), assert(), B(), llvm::cast(), llvm::commonAlignment(), llvm::AMDGPUAS::CONSTANT_ADDRESS, llvm::dbgs(), DL, F, llvm::AMDGPUSubtarget::getExplicitKernelArgOffset(), llvm::MachineFunction::getInfo(), llvm::MachineFunction::getRegInfo(), llvm::GCNSubtarget::getRegisterInfo(), llvm::MachineFunction::getSubtarget(), llvm::CallLowering::getTLI(), LLVM_DEBUG, MRI, llvm::LLT::pointer(), llvm::CallLowering::setArgFlags(), llvm::size(), and TRI.

Referenced by lowerFormalArguments().

◆ lowerReturn()

bool AMDGPUCallLowering::lowerReturn ( MachineIRBuilder & MIRBuilder,
const Value * Val,
ArrayRef< Register > VRegs,
FunctionLoweringInfo & FLI ) const
overridevirtual

This hook behaves as the extended lowerReturn function, but for targets that do not support swifterror value promotion.

Reimplemented from llvm::CallLowering.

Definition at line 357 of file AMDGPUCallLowering.cpp.

References assert(), B(), llvm::FunctionLoweringInfo::CanLowerReturn, llvm::FunctionLoweringInfo::DemoteRegister, llvm::ArrayRef< T >::empty(), llvm::MachineFunction::getInfo(), llvm::Value::getType(), llvm::CallLowering::insertSRetStores(), llvm::AMDGPU::isKernel(), llvm::AMDGPU::isShader(), llvm::SIMachineFunctionInfo::isWholeWaveFunction(), llvm::SIMachineFunctionInfo::returnsVoid(), and llvm::SIMachineFunctionInfo::setIfReturnsVoid().

◆ lowerTailCall()

bool AMDGPUCallLowering::lowerTailCall ( MachineIRBuilder & MIRBuilder,
CallLoweringInfo & Info,
SmallVectorImpl< ArgInfo > & OutArgs ) const

Definition at line 1231 of file AMDGPUCallLowering.cpp.

References addCallTargetOperands(), llvm::MachineInstrBuilder::addImm(), llvm::alignTo(), llvm::CallingConv::AMDGPU_Gfx, llvm::CallingConv::AMDGPU_Gfx_WholeWave, assert(), llvm::MachineIRBuilder::buildInstr(), llvm::MachineIRBuilder::buildInstrNoInsert(), llvm::cast(), llvm::constrainOperandRegClass(), llvm::dbgs(), llvm::CallLowering::determineAssignments(), llvm::dyn_cast(), F, getAssignFnsForCC(), getCallOpcode(), llvm::MachineFunction::getFrameInfo(), llvm::MachineFunction::getFunction(), llvm::MachineFunction::getInfo(), llvm::MachineIRBuilder::getMF(), llvm::MachineFunction::getRegInfo(), llvm::CCState::getStackSize(), llvm::MachineFunction::getSubtarget(), llvm::MachineFunction::getTarget(), llvm::CallLowering::getTLI(), llvm::TargetOptions::GuaranteedTailCallOpt, llvm::CallLowering::handleAssignments(), handleImplicitCallArguments(), llvm::RegState::Implicit, llvm::MachineIRBuilder::insertInstr(), llvm::isAligned(), llvm::AMDGPU::isChainCC(), llvm::Type::isIntegerTy(), llvm::APInt::isOneBitSet(), llvm::APInt::isZero(), LLVM_DEBUG, MRI, Opc, llvm::TargetMachine::Options, llvm::CallLowering::ArgInfo::OrigValue, passSpecialInputs(), llvm::CallLowering::ArgInfo::Regs, llvm::MachineFrameInfo::setHasTailCall(), TII, TRI, and llvm::CallLowering::BaseArgInfo::Ty.

Referenced by lowerCall(), and lowerChainCall().

◆ passSpecialInputs()

bool AMDGPUCallLowering::passSpecialInputs ( MachineIRBuilder & MIRBuilder,
CCState & CCInfo,
SmallVectorImpl< std::pair< MCRegister, Register > > & ArgRegs,
CallLoweringInfo & Info ) const

Definition at line 780 of file AMDGPUCallLowering.cpp.

References llvm::all_of(), llvm::CCState::AllocateReg(), assert(), llvm::MachineIRBuilder::buildConstant(), llvm::MachineIRBuilder::buildOr(), llvm::MachineIRBuilder::buildShl(), llvm::MachineIRBuilder::buildUndef(), llvm::ArgDescriptor::createArg(), llvm::dbgs(), llvm::AMDGPUFunctionArgInfo::DISPATCH_ID, llvm::AMDGPUFunctionArgInfo::DISPATCH_PTR, llvm::StringRef::empty(), llvm::AMDGPUArgumentUsageInfo::FixedABIFunctionInfo, llvm::SIMachineFunctionInfo::getArgInfo(), llvm::MachineFunction::getFunction(), llvm::MachineFunction::getInfo(), llvm::AMDGPUMachineFunction::getLDSKernelIdMetadata(), llvm::MachineIRBuilder::getMF(), llvm::AMDGPUFunctionArgInfo::getPreloadedValue(), llvm::MachineInstrBuilder::getReg(), llvm::MachineFunction::getRegInfo(), llvm::ArgDescriptor::getRegister(), llvm::MachineFunction::getSubtarget(), I, llvm::AMDGPUFunctionArgInfo::IMPLICIT_ARG_PTR, llvm::ArgDescriptor::isMasked(), llvm::ArgDescriptor::isRegister(), llvm::AMDGPUFunctionArgInfo::LDS_KERNEL_ID, LLVM_DEBUG, MRI, llvm::AMDGPUFunctionArgInfo::QUEUE_PTR, llvm::report_fatal_error(), S32, llvm::LLT::scalar(), llvm::AMDGPUFunctionArgInfo::WORKGROUP_ID_X, llvm::AMDGPUFunctionArgInfo::WORKGROUP_ID_Y, llvm::AMDGPUFunctionArgInfo::WORKGROUP_ID_Z, llvm::AMDGPUFunctionArgInfo::WORKITEM_ID_X, llvm::AMDGPUFunctionArgInfo::WORKITEM_ID_Y, llvm::AMDGPUFunctionArgInfo::WORKITEM_ID_Z, llvm::AMDGPUFunctionArgInfo::WorkItemIDX, llvm::AMDGPUFunctionArgInfo::WorkItemIDY, llvm::AMDGPUFunctionArgInfo::WorkItemIDZ, and Y.

Referenced by lowerCall(), and lowerTailCall().

The documentation for this class was generated from the following files:
Generated on for LLVM by doxygen 1.14.0