[Alignment] Migrate Attribute::getWith(Stack)Alignment

[gchatelet]

Summary

This is patch is part of a series to introduce an Alignment type.
See this thread for context: http://lists.llvm.org/pipermail/llvm-dev/2019-July/133851.html
See this patch for the introduction of the type: https://reviews.llvm.org/D64790

[gchatelet]

@courbet there's a slight difference in the way the field is serialized in the mir / ir format (hence not [NFC]) but there's no change in semantics.

[courbet]

llvm/lib/IR/Core.cpp

2792

Align

llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp

4168

const

llvm/lib/Target/AMDGPU/AMDGPULowerKernelArguments.cpp

78

Align

131–132

ditto

llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp

506

const

524

const

[gchatelet]

• Address comments

[gchatelet]

• Fix naming collision

clang/lib/CodeGen/CGCall.cpp

case ParameterABI::SwiftIndirectResult: {
       // Add 'noalias' in either case.
       Attrs.addAttribute(llvm::Attribute::NoAlias);
 
       // Add 'dereferenceable' and 'alignment'.
       auto PTy = ParamType->getPointeeType();
       if (!PTy->isIncompleteType() && PTy->isConstantSizeType()) {
         auto info = getContext().getTypeInfoInChars(PTy);
         Attrs.addDereferenceableAttr(info.first.getQuantity());
-        Attrs.addAttribute(llvm::Attribute::getWithAlignment(getLLVMContext(),
-                                                 info.second.getQuantity()));
+        Attrs.addAttribute(llvm::Attribute::getWithAlignment(
+            getLLVMContext(), info.second.getAsAlign()));
       }
       break;
     }
 
     case ParameterABI::SwiftErrorResult:
       Attrs.addAttribute(llvm::Attribute::SwiftError);
       break;
 

llvm/include/llvm/IR/Attributes.h

public:
   /// Return a uniquified Attribute object.
   static Attribute get(LLVMContext &Context, AttrKind Kind, uint64_t Val = 0);
   static Attribute get(LLVMContext &Context, StringRef Kind,
                        StringRef Val = StringRef());
   static Attribute get(LLVMContext &Context, AttrKind Kind, Type *Ty);
 
   /// Return a uniquified Attribute object that has the specific
   /// alignment set.
-  static Attribute getWithAlignment(LLVMContext &Context, uint64_t Align);
-  static Attribute getWithStackAlignment(LLVMContext &Context, uint64_t Align);
+  static Attribute getWithAlignment(LLVMContext &Context, Align Alignment);
+  static Attribute getWithStackAlignment(LLVMContext &Context, Align Alignment);
   static Attribute getWithDereferenceableBytes(LLVMContext &Context,
                                               uint64_t Bytes);
   static Attribute getWithDereferenceableOrNullBytes(LLVMContext &Context,
                                                      uint64_t Bytes);
   static Attribute getWithAllocSizeArgs(LLVMContext &Context,
                                         unsigned ElemSizeArg,
                                         const Optional<unsigned> &NumElemsArg);
   static Attribute getWithByValType(LLVMContext &Context, Type *Ty);

llvm/include/llvm/IR/IntrinsicInst.h

public:
 
     /// Set the specified arguments of the instruction.
     void setDest(Value *Ptr) {
       assert(getRawDest()->getType() == Ptr->getType() &&
              "setDest called with pointer of wrong type!");
       setArgOperand(ARG_DEST, Ptr);
     }
 
-    void setDestAlignment(unsigned Align) {
+    void setDestAlignment(unsigned Alignment) {
       removeParamAttr(ARG_DEST, Attribute::Alignment);
-      if (Align > 0)
-        addParamAttr(ARG_DEST,
-                     Attribute::getWithAlignment(getContext(), Align));
+      if (Alignment > 0)
+        addParamAttr(ARG_DEST, Attribute::getWithAlignment(getContext(),
+                                                           Align(Alignment)));
     }
 
     void setLength(Value *L) {
       assert(getLength()->getType() == L->getType() &&
              "setLength called with value of wrong type!");
       setArgOperand(ARG_LENGTH, L);
     }
   };
public:
     }
 
     void setSource(Value *Ptr) {
       assert(getRawSource()->getType() == Ptr->getType() &&
              "setSource called with pointer of wrong type!");
       BaseCL::setArgOperand(ARG_SOURCE, Ptr);
     }
 
-    void setSourceAlignment(unsigned Align) {
+    void setSourceAlignment(unsigned Alignment) {
       BaseCL::removeParamAttr(ARG_SOURCE, Attribute::Alignment);
-      if (Align > 0)
-        BaseCL::addParamAttr(ARG_SOURCE, Attribute::getWithAlignment(
-                                             BaseCL::getContext(), Align));
+      if (Alignment > 0)
+        BaseCL::addParamAttr(ARG_SOURCE,
+                             Attribute::getWithAlignment(BaseCL::getContext(),
+                                                         Align(Alignment)));
     }
   };
 
   /// Common base class for all memset intrinsics. Simply provides
   /// common methods.
   template <class BaseCL> class MemSetBase : public BaseCL {
   private:
     enum { ARG_VALUE = 1 };

llvm/lib/IR/Attributes.cpp

if (!PA) {
     PA = new TypeAttributeImpl(Kind, Ty);
     pImpl->AttrsSet.InsertNode(PA, InsertPoint);
   }
 
   // Return the Attribute that we found or created.
   return Attribute(PA);
 }
 
-Attribute Attribute::getWithAlignment(LLVMContext &Context, uint64_t Align) {
-  assert(isPowerOf2_32(Align) && "Alignment must be a power of two.");
-  assert(Align <= 0x40000000 && "Alignment too large.");
-  return get(Context, Alignment, Align);
+Attribute Attribute::getWithAlignment(LLVMContext &Context, Align A) {
+  assert(A <= 0x40000000 && "Alignment too large.");
+  return get(Context, Alignment, A.value());
 }
 
-Attribute Attribute::getWithStackAlignment(LLVMContext &Context,
-                                           uint64_t Align) {
-  assert(isPowerOf2_32(Align) && "Alignment must be a power of two.");
-  assert(Align <= 0x100 && "Alignment too large.");
-  return get(Context, StackAlignment, Align);
+Attribute Attribute::getWithStackAlignment(LLVMContext &Context, Align A) {
+  assert(A <= 0x100 && "Alignment too large.");
+  return get(Context, StackAlignment, A.value());
 }
 
 Attribute Attribute::getWithDereferenceableBytes(LLVMContext &Context,
                                                 uint64_t Bytes) {
   assert(Bytes && "Bytes must be non-zero.");
   return get(Context, Dereferenceable, Bytes);
 }
 
if (!B.contains(Kind))
       continue;
 
     Attribute Attr;
     switch (Kind) {
     case Attribute::ByVal:
       Attr = Attribute::getWithByValType(C, B.getByValType());
       break;
     case Attribute::Alignment:
-      Attr = Attribute::getWithAlignment(C, B.getAlignment());
+      Attr = Attribute::getWithAlignment(C, Align(B.getAlignment()));
       break;
     case Attribute::StackAlignment:
-      Attr = Attribute::getWithStackAlignment(C, B.getStackAlignment());
+      Attr = Attribute::getWithStackAlignment(C, Align(B.getStackAlignment()));
       break;
     case Attribute::Dereferenceable:
       Attr = Attribute::getWithDereferenceableBytes(
           C, B.getDereferenceableBytes());
       break;
     case Attribute::DereferenceableOrNull:
       Attr = Attribute::getWithDereferenceableOrNullBytes(
           C, B.getDereferenceableOrNullBytes());

llvm/lib/IR/Core.cpp

LLVMValueRef LLVMGetPreviousParam(LLVMValueRef Arg) {
   Argument *A = unwrap<Argument>(Arg);
   if (A->getArgNo() == 0)
     return nullptr;
   return wrap(&A->getParent()->arg_begin()[A->getArgNo() - 1]);
 }
 
 void LLVMSetParamAlignment(LLVMValueRef Arg, unsigned align) {
   Argument *A = unwrap<Argument>(Arg);
-  A->addAttr(Attribute::getWithAlignment(A->getContext(), align));
+  A->addAttr(Attribute::getWithAlignment(A->getContext(), Align(align)));
 }
 
 /*--.. Operations on ifuncs ................................................--*/
 
 LLVMValueRef LLVMAddGlobalIFunc(LLVMModuleRef M,
                                 const char *Name, size_t NameLen,
                                 LLVMTypeRef Ty, unsigned AddrSpace,
                                 LLVMValueRef Resolver) {

 void LLVMSetInstructionCallConv(LLVMValueRef Instr, unsigned CC) {
   return unwrap<CallBase>(Instr)->setCallingConv(
       static_cast<CallingConv::ID>(CC));
 }
 
 void LLVMSetInstrParamAlignment(LLVMValueRef Instr, unsigned index,
                                 unsigned align) {
   auto *Call = unwrap<CallBase>(Instr);
-  Attribute AlignAttr = Attribute::getWithAlignment(Call->getContext(), align);
+  Attribute AlignAttr =
+      Attribute::getWithAlignment(Call->getContext(), Align(align));
   Call->addAttribute(index, AlignAttr);
 }
 
 void LLVMAddCallSiteAttribute(LLVMValueRef C, LLVMAttributeIndex Idx,
                               LLVMAttributeRef A) {
   unwrap<CallBase>(C)->addAttribute(Idx, unwrap(A));
 }
 

llvm/lib/IR/IRBuilder.cpp

CallInst *IRBuilderBase::CreateElementUnorderedAtomicMemMove(
   Type *Tys[] = {Dst->getType(), Src->getType(), Size->getType()};
   Module *M = BB->getParent()->getParent();
   Function *TheFn = Intrinsic::getDeclaration(
       M, Intrinsic::memmove_element_unordered_atomic, Tys);
 
   CallInst *CI = createCallHelper(TheFn, Ops, this);
 
   // Set the alignment of the pointer args.
-  CI->addParamAttr(0, Attribute::getWithAlignment(CI->getContext(), DstAlign));
-  CI->addParamAttr(1, Attribute::getWithAlignment(CI->getContext(), SrcAlign));
+  CI->addParamAttr(
+      0, Attribute::getWithAlignment(CI->getContext(), Align(DstAlign)));
+  CI->addParamAttr(
+      1, Attribute::getWithAlignment(CI->getContext(), Align(SrcAlign)));
 
   // Set the TBAA info if present.
   if (TBAATag)
     CI->setMetadata(LLVMContext::MD_tbaa, TBAATag);
 
   // Set the TBAA Struct info if present.
   if (TBAAStructTag)
     CI->setMetadata(LLVMContext::MD_tbaa_struct, TBAAStructTag);

llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp

if (MFI->hasFlatScratchInit())
     Out.code_properties |= AMD_CODE_PROPERTY_ENABLE_SGPR_FLAT_SCRATCH_INIT;
 
   if (MFI->hasDispatchPtr())
     Out.code_properties |= AMD_CODE_PROPERTY_ENABLE_SGPR_DISPATCH_PTR;
 
   if (STM.isXNACKEnabled())
     Out.code_properties |= AMD_CODE_PROPERTY_IS_XNACK_SUPPORTED;
 
-  unsigned MaxKernArgAlign;
+  Align MaxKernArgAlign;
   Out.kernarg_segment_byte_size = STM.getKernArgSegmentSize(F, MaxKernArgAlign);
   Out.wavefront_sgpr_count = CurrentProgramInfo.NumSGPR;
   Out.workitem_vgpr_count = CurrentProgramInfo.NumVGPR;
   Out.workitem_private_segment_byte_size = CurrentProgramInfo.ScratchSize;
   Out.workgroup_group_segment_byte_size = CurrentProgramInfo.LDSSize;
 
-  // These alignment values are specified in powers of two, so alignment =
-  // 2^n.  The minimum alignment is 2^4 = 16.
-  Out.kernarg_segment_alignment = std::max<size_t>(4,
-      countTrailingZeros(MaxKernArgAlign));
+  // kernarg_segment_alignment is specified as log of the alignment.
+  // The minimum alignment is 16.
+  Out.kernarg_segment_alignment = Log2(std::max(Align(16), MaxKernArgAlign));
 }
 
 bool AMDGPUAsmPrinter::PrintAsmOperand(const MachineInstr *MI, unsigned OpNo,
                                        const char *ExtraCode, raw_ostream &O) {
   // First try the generic code, which knows about modifiers like 'c' and 'n'.
   if (!AsmPrinter::PrintAsmOperand(MI, OpNo, ExtraCode, O))
     return false;
 

llvm/lib/Target/AMDGPU/AMDGPUHSAMetadataStreamer.cpp

MetadataStreamerV2::getHSACodeProps(const MachineFunction &MF,
   const GCNSubtarget &STM = MF.getSubtarget<GCNSubtarget>();
   const SIMachineFunctionInfo &MFI = *MF.getInfo<SIMachineFunctionInfo>();
   HSAMD::Kernel::CodeProps::Metadata HSACodeProps;
   const Function &F = MF.getFunction();
 
   assert(F.getCallingConv() == CallingConv::AMDGPU_KERNEL ||
          F.getCallingConv() == CallingConv::SPIR_KERNEL);
 
-  unsigned MaxKernArgAlign;
+  Align MaxKernArgAlign;
   HSACodeProps.mKernargSegmentSize = STM.getKernArgSegmentSize(F,
                                                                MaxKernArgAlign);
   HSACodeProps.mGroupSegmentFixedSize = ProgramInfo.LDSSize;
   HSACodeProps.mPrivateSegmentFixedSize = ProgramInfo.ScratchSize;
-  HSACodeProps.mKernargSegmentAlign = std::max(MaxKernArgAlign, 4u);
+  HSACodeProps.mKernargSegmentAlign =
+      std::max(MaxKernArgAlign, Align(4)).value();
   HSACodeProps.mWavefrontSize = STM.getWavefrontSize();
   HSACodeProps.mNumSGPRs = ProgramInfo.NumSGPR;
   HSACodeProps.mNumVGPRs = ProgramInfo.NumVGPR;
   HSACodeProps.mMaxFlatWorkGroupSize = MFI.getMaxFlatWorkGroupSize();
   HSACodeProps.mIsDynamicCallStack = ProgramInfo.DynamicCallStack;
   HSACodeProps.mIsXNACKEnabled = STM.isXNACKEnabled();
   HSACodeProps.mNumSpilledSGPRs = MFI.getNumSpilledSGPRs();
   HSACodeProps.mNumSpilledVGPRs = MFI.getNumSpilledVGPRs();

 MetadataStreamerV3::getHSAKernelProps(const MachineFunction &MF,
                                       const SIProgramInfo &ProgramInfo) const {
   const GCNSubtarget &STM = MF.getSubtarget<GCNSubtarget>();
   const SIMachineFunctionInfo &MFI = *MF.getInfo<SIMachineFunctionInfo>();
   const Function &F = MF.getFunction();
 
   auto Kern = HSAMetadataDoc->getMapNode();
 
-  unsigned MaxKernArgAlign;
+  Align MaxKernArgAlign;
   Kern[".kernarg_segment_size"] = Kern.getDocument()->getNode(
       STM.getKernArgSegmentSize(F, MaxKernArgAlign));
   Kern[".group_segment_fixed_size"] =
       Kern.getDocument()->getNode(ProgramInfo.LDSSize);
   Kern[".private_segment_fixed_size"] =
       Kern.getDocument()->getNode(ProgramInfo.ScratchSize);
   Kern[".kernarg_segment_align"] =
-      Kern.getDocument()->getNode(std::max(uint32_t(4), MaxKernArgAlign));
+      Kern.getDocument()->getNode(std::max(Align(4), MaxKernArgAlign).value());
   Kern[".wavefront_size"] =
       Kern.getDocument()->getNode(STM.getWavefrontSize());
   Kern[".sgpr_count"] = Kern.getDocument()->getNode(ProgramInfo.NumSGPR);
   Kern[".vgpr_count"] = Kern.getDocument()->getNode(ProgramInfo.NumVGPR);
   Kern[".max_flat_workgroup_size"] =
       Kern.getDocument()->getNode(MFI.getMaxFlatWorkGroupSize());
   Kern[".sgpr_spill_count"] =
       Kern.getDocument()->getNode(MFI.getNumSpilledSGPRs());

llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp

 }
 
 uint32_t AMDGPUTargetLowering::getImplicitParameterOffset(
     const MachineFunction &MF, const ImplicitParameter Param) const {
   const AMDGPUMachineFunction *MFI = MF.getInfo<AMDGPUMachineFunction>();
   const AMDGPUSubtarget &ST =
       AMDGPUSubtarget::get(getTargetMachine(), MF.getFunction());
   unsigned ExplicitArgOffset = ST.getExplicitKernelArgOffset(MF.getFunction());
-  unsigned Alignment = ST.getAlignmentForImplicitArgPtr();
+  const Align Alignment = ST.getAlignmentForImplicitArgPtr();
   uint64_t ArgOffset = alignTo(MFI->getExplicitKernArgSize(), Alignment) +
                        ExplicitArgOffset;
   switch (Param) {
   case GRID_DIM:
     return ArgOffset;
   case GRID_OFFSET:
     return ArgOffset + 4;
   }

llvm/lib/Target/AMDGPU/AMDGPULowerKernelArguments.cpp

bool AMDGPULowerKernelArguments::runOnFunction(Function &F) {
 
   const TargetMachine &TM = TPC.getTM<TargetMachine>();
   const GCNSubtarget &ST = TM.getSubtarget<GCNSubtarget>(F);
   LLVMContext &Ctx = F.getParent()->getContext();
   const DataLayout &DL = F.getParent()->getDataLayout();
   BasicBlock &EntryBlock = *F.begin();
   IRBuilder<> Builder(&*EntryBlock.begin());
 
-  const unsigned KernArgBaseAlign = 16; // FIXME: Increase if necessary
+  const Align KernArgBaseAlign(16); // FIXME: Increase if necessary
   const uint64_t BaseOffset = ST.getExplicitKernelArgOffset(F);
 
-  unsigned MaxAlign;
+  Align MaxAlign;
   // FIXME: Alignment is broken broken with explicit arg offset.;
   const uint64_t TotalKernArgSize = ST.getKernArgSegmentSize(F, MaxAlign);
   if (TotalKernArgSize == 0)
     return false;
 
   CallInst *KernArgSegment =
       Builder.CreateIntrinsic(Intrinsic::amdgcn_kernarg_segment_ptr, {}, {},
                               nullptr, F.getName() + ".kernarg.segment");
for (Argument &Arg : F.args()) {
     VectorType *VT = dyn_cast<VectorType>(ArgTy);
     bool IsV3 = VT && VT->getNumElements() == 3;
     bool DoShiftOpt = Size < 32 && !ArgTy->isAggregateType();
 
     VectorType *V4Ty = nullptr;
 
     int64_t AlignDownOffset = alignDown(EltOffset, 4);
     int64_t OffsetDiff = EltOffset - AlignDownOffset;
-    unsigned AdjustedAlign = MinAlign(DoShiftOpt ? AlignDownOffset : EltOffset,
-                                      KernArgBaseAlign);
+    Align AdjustedAlign = commonAlignment(
+        KernArgBaseAlign, DoShiftOpt ? AlignDownOffset : EltOffset);
 
     Value *ArgPtr;
     Type *AdjustedArgTy;
     if (DoShiftOpt) { // FIXME: Handle aggregate types
       // Since we don't have sub-dword scalar loads, avoid doing an extload by
       // loading earlier than the argument address, and extracting the relevant
       // bits.
       //
if (IsV3 && Size >= 32) {
       V4Ty = VectorType::get(VT->getVectorElementType(), 4);
       // Use the hack that clang uses to avoid SelectionDAG ruining v3 loads
       AdjustedArgTy = V4Ty;
     }
 
     ArgPtr = Builder.CreateBitCast(ArgPtr, AdjustedArgTy->getPointerTo(AS),
                                    ArgPtr->getName() + ".cast");
     LoadInst *Load =
-        Builder.CreateAlignedLoad(AdjustedArgTy, ArgPtr, AdjustedAlign);
+        Builder.CreateAlignedLoad(AdjustedArgTy, ArgPtr, AdjustedAlign.value());
     Load->setMetadata(LLVMContext::MD_invariant_load, MDNode::get(Ctx, {}));
 
     MDBuilder MDB(Ctx);
 
     if (isa<PointerType>(ArgTy)) {
       if (Arg.hasNonNullAttr())
         Load->setMetadata(LLVMContext::MD_nonnull, MDNode::get(Ctx, {}));
 
if (DoShiftOpt) {
       Arg.replaceAllUsesWith(Shuf);
     } else {
       Load->setName(Arg.getName() + ".load");
       Arg.replaceAllUsesWith(Load);
     }
   }
 
   KernArgSegment->addAttribute(
       AttributeList::ReturnIndex,
       Attribute::getWithAlignment(Ctx, std::max(KernArgBaseAlign, MaxAlign)));
 
   return true;
 }
 
 INITIALIZE_PASS_BEGIN(AMDGPULowerKernelArguments, DEBUG_TYPE,
                       "AMDGPU Lower Kernel Arguments", false, false)
 INITIALIZE_PASS_END(AMDGPULowerKernelArguments, DEBUG_TYPE, "AMDGPU Lower Kernel Arguments",
                     false, false)
 
 char AMDGPULowerKernelArguments::ID = 0;
 
 FunctionPass *llvm::createAMDGPULowerKernelArgumentsPass() {
   return new AMDGPULowerKernelArguments();
 }

llvm/lib/Target/AMDGPU/AMDGPUMachineFunction.h

 
 class AMDGPUMachineFunction : public MachineFunctionInfo {
   /// A map to keep track of local memory objects and their offsets within the
   /// local memory space.
   SmallDenseMap<const GlobalValue *, unsigned, 4> LocalMemoryObjects;
 
 protected:
   uint64_t ExplicitKernArgSize; // Cache for this.
-  unsigned MaxKernArgAlign; // Cache for this.
+  Align MaxKernArgAlign;        // Cache for this.
 
   /// Number of bytes in the LDS that are being used.
   unsigned LDSSize;
 
   // Kernels + shaders. i.e. functions called by the driver and not called
   // by other functions.
   bool IsEntryFunction;
 
   bool NoSignedZerosFPMath;
 
   // Function may be memory bound.
   bool MemoryBound;
 
   // Kernel may need limited waves per EU for better performance.
   bool WaveLimiter;
 
 public:
   AMDGPUMachineFunction(const MachineFunction &MF);
 
   uint64_t getExplicitKernArgSize() const {
     return ExplicitKernArgSize;
   }
 
-  unsigned getMaxKernArgAlign() const {
-    return MaxKernArgAlign;
-  }
+  unsigned getMaxKernArgAlign() const { return MaxKernArgAlign.value(); }
 
   unsigned getLDSSize() const {
     return LDSSize;
   }
 
   bool isEntryFunction() const {
     return IsEntryFunction;
   }

llvm/lib/Target/AMDGPU/AMDGPUMachineFunction.cpp

 #include "llvm/CodeGen/MachineModuleInfo.h"
 
 using namespace llvm;
 
 AMDGPUMachineFunction::AMDGPUMachineFunction(const MachineFunction &MF) :
   MachineFunctionInfo(),
   LocalMemoryObjects(),
   ExplicitKernArgSize(0),
-  MaxKernArgAlign(0),
   LDSSize(0),
   IsEntryFunction(AMDGPU::isEntryFunctionCC(MF.getFunction().getCallingConv())),
   NoSignedZerosFPMath(MF.getTarget().Options.NoSignedZerosFPMath),
   MemoryBound(false),
   WaveLimiter(false) {
   const AMDGPUSubtarget &ST = AMDGPUSubtarget::get(MF);
 
   // FIXME: Should initialize KernArgSize based on ExplicitKernelArgOffset,

llvm/lib/Target/AMDGPU/AMDGPUSubtarget.h

public:
   unsigned getWavefrontSize() const {
     return WavefrontSize;
   }
 
   int getLocalMemorySize() const {
     return LocalMemorySize;
   }
 
-  unsigned getAlignmentForImplicitArgPtr() const {
-    return isAmdHsaOS() ? 8 : 4;
+  Align getAlignmentForImplicitArgPtr() const {
+    return isAmdHsaOS() ? Align(8) : Align(4);
   }
 
   /// Returns the offset in bytes from the start of the input buffer
   ///        of the first explicit kernel argument.
   unsigned getExplicitKernelArgOffset(const Function &F) const {
     return isAmdHsaOrMesa(F) ? 0 : 36;
   }
 
public:
 
   /// \returns Number of bytes of arguments that are passed to a shader or
   /// kernel in addition to the explicit ones declared for the function.
   unsigned getImplicitArgNumBytes(const Function &F) const {
     if (isMesaKernel(F))
       return 16;
     return AMDGPU::getIntegerAttribute(F, "amdgpu-implicitarg-num-bytes", 0);
   }
-  uint64_t getExplicitKernArgSize(const Function &F,
-                                  unsigned &MaxAlign) const;
-  unsigned getKernArgSegmentSize(const Function &F,
-                                 unsigned &MaxAlign) const;
+  uint64_t getExplicitKernArgSize(const Function &F, Align &MaxAlign) const;
+  unsigned getKernArgSegmentSize(const Function &F, Align &MaxAlign) const;
 
   virtual ~AMDGPUSubtarget() {}
 };
 
 class GCNSubtarget : public AMDGPUGenSubtargetInfo,
                      public AMDGPUSubtarget {
 
   using AMDGPUSubtarget::getMaxWavesPerEU;

llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp

bool AMDGPUSubtarget::makeLIDRangeMetadata(Instruction *I) const {
   MDBuilder MDB(I->getContext());
   MDNode *MaxWorkGroupSizeRange = MDB.createRange(APInt(32, MinSize),
                                                   APInt(32, MaxSize));
   I->setMetadata(LLVMContext::MD_range, MaxWorkGroupSizeRange);
   return true;
 }
 
 uint64_t AMDGPUSubtarget::getExplicitKernArgSize(const Function &F,
-                                                 unsigned &MaxAlign) const {
+                                                 Align &MaxAlign) const {
   assert(F.getCallingConv() == CallingConv::AMDGPU_KERNEL ||
          F.getCallingConv() == CallingConv::SPIR_KERNEL);
 
   const DataLayout &DL = F.getParent()->getDataLayout();
   uint64_t ExplicitArgBytes = 0;
-  MaxAlign = 1;
+  MaxAlign = Align::None();
 
   for (const Argument &Arg : F.args()) {
     Type *ArgTy = Arg.getType();
 
-    unsigned Align = DL.getABITypeAlignment(ArgTy);
+    const Align Alignment(DL.getABITypeAlignment(ArgTy));
     uint64_t AllocSize = DL.getTypeAllocSize(ArgTy);
-    ExplicitArgBytes = alignTo(ExplicitArgBytes, Align) + AllocSize;
-    MaxAlign = std::max(MaxAlign, Align);
+    ExplicitArgBytes = alignTo(ExplicitArgBytes, Alignment) + AllocSize;
+    MaxAlign = std::max(MaxAlign, Alignment);
   }
 
   return ExplicitArgBytes;
 }
 
 unsigned AMDGPUSubtarget::getKernArgSegmentSize(const Function &F,
-                                                unsigned &MaxAlign) const {
+                                                Align &MaxAlign) const {
   uint64_t ExplicitArgBytes = getExplicitKernArgSize(F, MaxAlign);
 
   unsigned ExplicitOffset = getExplicitKernelArgOffset(F);
 
   uint64_t TotalSize = ExplicitOffset + ExplicitArgBytes;
   unsigned ImplicitBytes = getImplicitArgNumBytes(F);
   if (ImplicitBytes != 0) {
-    unsigned Alignment = getAlignmentForImplicitArgPtr();
+    const Align Alignment = getAlignmentForImplicitArgPtr();
     TotalSize = alignTo(ExplicitArgBytes, Alignment) + ImplicitBytes;
   }
 
   // Being able to dereference past the end is useful for emitting scalar loads.
   return alignTo(TotalSize, 4);
 }
 
 R600Subtarget::R600Subtarget(const Triple &TT, StringRef GPU, StringRef FS,

llvm/lib/Target/AMDGPU/SIMachineFunctionInfo.cpp

 
 void yaml::SIMachineFunctionInfo::mappingImpl(yaml::IO &YamlIO) {
   MappingTraits<SIMachineFunctionInfo>::mapping(YamlIO, *this);
 }
 
 bool SIMachineFunctionInfo::initializeBaseYamlFields(
   const yaml::SIMachineFunctionInfo &YamlMFI) {
   ExplicitKernArgSize = YamlMFI.ExplicitKernArgSize;
-  MaxKernArgAlign = YamlMFI.MaxKernArgAlign;
+  MaxKernArgAlign = assumeAligned(YamlMFI.MaxKernArgAlign);
   LDSSize = YamlMFI.LDSSize;
   HighBitsOf32BitAddress = YamlMFI.HighBitsOf32BitAddress;
   IsEntryFunction = YamlMFI.IsEntryFunction;
   NoSignedZerosFPMath = YamlMFI.NoSignedZerosFPMath;
   MemoryBound = YamlMFI.MemoryBound;
   WaveLimiter = YamlMFI.WaveLimiter;
   return false;
 }

llvm/lib/Transforms/IPO/Attributor.cpp

struct AAAlignImpl : AAAlign {
   //       the existing manifest method and a new one for AAAlignImpl that value
   //       to avoid making the alignment explicit if it did not improve.
 
   /// See AbstractAttribute::getDeducedAttributes
   virtual void
   getDeducedAttributes(LLVMContext &Ctx,
                        SmallVectorImpl<Attribute> &Attrs) const override {
     if (getAssumedAlign() > 1)
-      Attrs.emplace_back(Attribute::getWithAlignment(Ctx, getAssumedAlign()));
+      Attrs.emplace_back(
+          Attribute::getWithAlignment(Ctx, Align(getAssumedAlign())));
   }
 
   /// See AbstractAttribute::getAsStr().
   const std::string getAsStr() const override {
     return getAssumedAlign() ? ("align<" + std::to_string(getKnownAlign()) +
                                 "-" + std::to_string(getAssumedAlign()) + ">")
                              : "unknown-align";
   }

llvm/test/CodeGen/MIR/AMDGPU/machine-function-info-no-ir.mir

 
 ...
 
 # FIXME: Should be able to not print section for simple
 ---
 # ALL-LABEL: name: no_mfi
 # FULL: machineFunctionInfo:
 # FULL-NEXT: explicitKernArgSize: 0
-# FULL-NEXT: maxKernArgAlign: 0
+# FULL-NEXT: maxKernArgAlign: 1
 # FULL-NEXT: ldsSize: 0
 # FULL-NEXT: isEntryFunction: false
 # FULL-NEXT: noSignedZerosFPMath: false
 # FULL-NEXT: memoryBound:     false
 # FULL-NEXT: waveLimiter:     false
 # FULL-NEXT: scratchRSrcReg:  '$private_rsrc_reg'
 # FULL-NEXT: scratchWaveOffsetReg:  '$scratch_wave_offset_reg'
 # FULL-NEXT: frameOffsetReg:  '$fp_reg'
 # FULL-NEXT: stackPtrOffsetReg:  '$sp_reg'
 # FULL-NEXT: argumentInfo:
 # FULL-NEXT: privateSegmentBuffer: { reg: '$sgpr0_sgpr1_sgpr2_sgpr3' }
 # FULL-NEXT: privateSegmentWaveByteOffset: { reg: '$sgpr33' }
 # FULL-NEXT: mode:
 # FULL-NEXT: ieee: true
 # FULL-NEXT: dx10-clamp: true
 # FULL-NEXT:  highBitsOf32BitAddress: 0
 # FULL-NEXT: body:
 
 # SIMPLE: machineFunctionInfo:
+# SIMPLE-NEXT: maxKernArgAlign: 1
 # SIMPLE-NEXT: argumentInfo:
 # SIMPLE-NEXT: privateSegmentBuffer: { reg: '$sgpr0_sgpr1_sgpr2_sgpr3' }
 # SIMPLE-NEXT: privateSegmentWaveByteOffset: { reg: '$sgpr33' }
 # SIMPLE-NEXT: body:
 
 name: no_mfi
 body:             |
   bb.0:
     S_ENDPGM 0
 
 ...
 
 ---
 # ALL-LABEL: name: empty_mfi
 # FULL: machineFunctionInfo:
 # FULL-NEXT: explicitKernArgSize: 0
-# FULL-NEXT: maxKernArgAlign: 0
+# FULL-NEXT: maxKernArgAlign: 1
 # FULL-NEXT: ldsSize: 0
 # FULL-NEXT: isEntryFunction: false
 # FULL-NEXT: noSignedZerosFPMath: false
 # FULL-NEXT: memoryBound:     false
 # FULL-NEXT: waveLimiter:     false
 # FULL-NEXT: scratchRSrcReg:  '$private_rsrc_reg'
 # FULL-NEXT: scratchWaveOffsetReg:  '$scratch_wave_offset_reg'
 # FULL-NEXT: frameOffsetReg:  '$fp_reg'
 # FULL-NEXT: stackPtrOffsetReg:  '$sp_reg'
 # FULL-NEXT: argumentInfo:
 # FULL-NEXT: privateSegmentBuffer: { reg: '$sgpr0_sgpr1_sgpr2_sgpr3' }
 # FULL-NEXT: privateSegmentWaveByteOffset: { reg: '$sgpr33' }
 # FULL-NEXT: mode:
 # FULL-NEXT: ieee: true
 # FULL-NEXT: dx10-clamp: true
 # FULL-NEXT:  highBitsOf32BitAddress: 0
 # FULL-NEXT: body:
 
 # SIMPLE: machineFunctionInfo:
+# SIMPLE-NEXT: maxKernArgAlign: 1
 # SIMPLE-NEXT: argumentInfo:
 # SIMPLE-NEXT: privateSegmentBuffer: { reg: '$sgpr0_sgpr1_sgpr2_sgpr3' }
 # SIMPLE-NEXT: privateSegmentWaveByteOffset: { reg: '$sgpr33' }
 # SIMPLE-NEXT: body:
 
 name: empty_mfi
 machineFunctionInfo:
 body:             |
   bb.0:
     S_ENDPGM 0
 
 ...
 
 ---
 # ALL-LABEL: name: empty_mfi_entry_func
 # FULL: machineFunctionInfo:
 # FULL-NEXT: explicitKernArgSize: 0
-# FULL-NEXT: maxKernArgAlign: 0
+# FULL-NEXT: maxKernArgAlign: 1
 # FULL-NEXT: ldsSize: 0
 # FULL-NEXT: isEntryFunction: true
 # FULL-NEXT: noSignedZerosFPMath: false
 # FULL-NEXT: memoryBound:     false
 # FULL-NEXT: waveLimiter:     false
 # FULL-NEXT: scratchRSrcReg:  '$private_rsrc_reg'
 # FULL-NEXT: scratchWaveOffsetReg:  '$scratch_wave_offset_reg'
 # FULL-NEXT: frameOffsetReg:  '$fp_reg'
 # FULL-NEXT: stackPtrOffsetReg:  '$sp_reg'
 # FULL-NEXT: argumentInfo:
 # FULL-NEXT: privateSegmentBuffer: { reg: '$sgpr0_sgpr1_sgpr2_sgpr3' }
 # FULL-NEXT: privateSegmentWaveByteOffset: { reg: '$sgpr33' }
 # FULL-NEXT: mode:
 # FULL-NEXT: ieee: true
 # FULL-NEXT: dx10-clamp: true
 # FULL-NEXT:  highBitsOf32BitAddress: 0
 # FULL-NEXT: body:
 
 # SIMPLE: machineFunctionInfo:
+# SIMPLE-NEXT: maxKernArgAlign: 1
 # SIMPLE-NEXT: isEntryFunction: true
 # SIMPLE-NEXT: argumentInfo:
 # SIMPLE-NEXT: privateSegmentBuffer: { reg: '$sgpr0_sgpr1_sgpr2_sgpr3' }
 # SIMPLE-NEXT: privateSegmentWaveByteOffset: { reg: '$sgpr33' }
 # SIMPLE-NEXT: body:
 
 name: empty_mfi_entry_func
 machineFunctionInfo:

llvm/test/CodeGen/MIR/AMDGPU/machine-function-info.ll

define amdgpu_kernel void @kernel(i32 %arg0, i64 %arg1, <16 x i32> %arg2) {
   %gep = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %arg0
   store float 0.0, float addrspace(3)* %gep, align 4
   ret void
 }
 
 ; CHECK-LABEL: {{^}}name: ps_shader
 ; CHECK: machineFunctionInfo:
 ; CHECK-NEXT: explicitKernArgSize: 0
-; CHECK-NEXT: maxKernArgAlign: 0
+; CHECK-NEXT: maxKernArgAlign: 1
 ; CHECK-NEXT: ldsSize: 0
 ; CHECK-NEXT: isEntryFunction: true
 ; CHECK-NEXT: noSignedZerosFPMath: false
 ; CHECK-NEXT: memoryBound: false
 ; CHECK-NEXT: waveLimiter: false
 ; CHECK-NEXT: scratchRSrcReg:  '$sgpr96_sgpr97_sgpr98_sgpr99'
 ; CHECK-NEXT: scratchWaveOffsetReg: '$sgpr101'
 ; CHECK-NEXT: frameOffsetReg:  '$sgpr101'
 ; CHECK-NEXT: stackPtrOffsetReg: '$sgpr101'
 ; CHECK-NEXT: argumentInfo:
 ; CHECK-NEXT: privateSegmentWaveByteOffset: { reg: '$sgpr3' }
 ; CHECK-NEXT: implicitBufferPtr: { reg: '$sgpr0_sgpr1' }
 ; CHECK-NEXT: mode:
 ; CHECK-NEXT: ieee: false
 ; CHECK-NEXT: dx10-clamp: true
 ; CHECK-NEXT: highBitsOf32BitAddress: 0
 ; CHECK-NEXT: body:
 define amdgpu_ps void @ps_shader(i32 %arg0, i32 inreg %arg1) {
   ret void
 }
 
 ; CHECK-LABEL: {{^}}name: function
 ; CHECK: machineFunctionInfo:
 ; CHECK-NEXT: explicitKernArgSize: 0
-; CHECK-NEXT: maxKernArgAlign: 0
+; CHECK-NEXT: maxKernArgAlign: 1
 ; CHECK-NEXT: ldsSize: 0
 ; CHECK-NEXT: isEntryFunction: false
 ; CHECK-NEXT: noSignedZerosFPMath: false
 ; CHECK-NEXT: memoryBound: false
 ; CHECK-NEXT: waveLimiter: false
 ; CHECK-NEXT: scratchRSrcReg: '$sgpr0_sgpr1_sgpr2_sgpr3'
 ; CHECK-NEXT: scratchWaveOffsetReg: '$sgpr33'
 ; CHECK-NEXT: frameOffsetReg: '$sgpr34'
 ; CHECK-NEXT: stackPtrOffsetReg: '$sgpr32'
 ; CHECK-NEXT: argumentInfo:
 ; CHECK-NEXT: privateSegmentBuffer: { reg: '$sgpr0_sgpr1_sgpr2_sgpr3' }
 ; CHECK-NEXT: privateSegmentWaveByteOffset: { reg: '$sgpr33' }
 ; CHECK-NEXT: mode:
 ; CHECK-NEXT: ieee: true
 ; CHECK-NEXT: dx10-clamp: true
 ; CHECK-NEXT: highBitsOf32BitAddress: 0
 ; CHECK-NEXT: body:
 define void @function() {
   ret void
 }
 
 ; CHECK-LABEL: {{^}}name: function_nsz
 ; CHECK: machineFunctionInfo:
 ; CHECK-NEXT: explicitKernArgSize: 0
-; CHECK-NEXT: maxKernArgAlign: 0
+; CHECK-NEXT: maxKernArgAlign: 1
 ; CHECK-NEXT: ldsSize: 0
 ; CHECK-NEXT: isEntryFunction: false
 ; CHECK-NEXT: noSignedZerosFPMath: true
 ; CHECK-NEXT: memoryBound: false
 ; CHECK-NEXT: waveLimiter: false
 ; CHECK-NEXT: scratchRSrcReg: '$sgpr0_sgpr1_sgpr2_sgpr3'
 ; CHECK-NEXT: scratchWaveOffsetReg: '$sgpr33'
 ; CHECK-NEXT: frameOffsetReg: '$sgpr34'

llvm/unittests/IR/AttributesTest.cpp

TEST(Attributes, AddAttributes) {
   AL = AL.addAttributes(C, AttributeList::ReturnIndex, B);
   EXPECT_TRUE(AL.hasAttribute(AttributeList::ReturnIndex, Attribute::SExt));
   EXPECT_TRUE(AL.hasFnAttribute(Attribute::NoReturn));
 }
 
 TEST(Attributes, RemoveAlign) {
   LLVMContext C;
 
-  Attribute AlignAttr = Attribute::getWithAlignment(C, 8);
-  Attribute StackAlignAttr = Attribute::getWithStackAlignment(C, 32);
+  Attribute AlignAttr = Attribute::getWithAlignment(C, Align(8));
+  Attribute StackAlignAttr = Attribute::getWithStackAlignment(C, Align(32));
   AttrBuilder B_align_readonly;
   B_align_readonly.addAttribute(AlignAttr);
   B_align_readonly.addAttribute(Attribute::ReadOnly);
   AttrBuilder B_align;
   B_align.addAttribute(AlignAttr);
   AttrBuilder B_stackalign_optnone;
   B_stackalign_optnone.addAttribute(StackAlignAttr);
   B_stackalign_optnone.addAttribute(Attribute::OptimizeNone);
TEST(Attributes, RemoveAlign) {
   EXPECT_FALSE(AL2.hasAttribute(0, Attribute::StackAlignment));
   EXPECT_TRUE(AL2.hasAttribute(0, Attribute::OptimizeNone));
 }
 
 TEST(Attributes, AddMatchingAlignAttr) {
   LLVMContext C;
   AttributeList AL;
   AL = AL.addAttribute(C, AttributeList::FirstArgIndex,
-                       Attribute::getWithAlignment(C, 8));
+                       Attribute::getWithAlignment(C, Align(8)));
   AL = AL.addAttribute(C, AttributeList::FirstArgIndex + 1,
-                       Attribute::getWithAlignment(C, 32));
+                       Attribute::getWithAlignment(C, Align(32)));
   EXPECT_EQ(8U, AL.getParamAlignment(0));
   EXPECT_EQ(32U, AL.getParamAlignment(1));
 
   AttrBuilder B;
   B.addAttribute(Attribute::NonNull);
   B.addAlignmentAttr(8);
   AL = AL.addAttributes(C, AttributeList::FirstArgIndex, B);
   EXPECT_EQ(8U, AL.getParamAlignment(0));