[AArch64] Correctly align HFA arguments when passed on the stack

[chill]

Summary

When we pass a Homogeneous Floating-Point Aggregate (HFA) argument
with increased alignment requirements, for example

struct S {
  __attribute__ ((__aligned__(16))) double v[4];
};

Clang uses [4 x double] for the parameter, which is passed on the
stack at alignment 8, whereas it should be at alignment 16, following
Rule C.4 in AAPCS (https://github.com/ARM-software/abi-aa/blob/master/aapcs64/aapcs64.rst#642parameter-passing-rules)

Currently we don't have a way to express in LLVM IR the alignment
requirements of the function arguments. The align attribute is
applicable to pointers only, and only for some special ways of passing
arguments (e..g byval). When implementing AAPCS32/AAPCS64, clang
resorts to dubious hacks of coercing to types, which naturally have
the needed alignment. We don't have enough types to cover all the
cases, though.

This patch introduces a new use of the stackalign attribute to
control stack slot alignment, when and if an argument is passed in
memory.

The attribute align is left as an optimizer hint - it still applies
to pointer types only and pertains to the content of the pointer,
whereas the alignment of the pointer itself is determined by the
stackalign attribute.

For byval arguments, the stackalign attribute assumes the role,
previously perfomed by align, falling back to align if
stackalign` is absent.

Patch by Momchil Velikov and Lucas Prates.

[chill]

llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp

9428

Maybe we shouldn't be using OrigAlign and can reuse ByValOrByRefAlign (renamed).

[t.p.northover]

I think this looks reasonable. Preserving Darwin is probably the safest choice, though I don't know if we actually rely on it. If the description is actually the commit message make sure it's updated before committing because this patch does in fact change the Clang side too.

[chill]

This needs to be changed, cf. https://lists.llvm.org/pipermail/llvm-dev/2021-March/149405.html

[chill]

In D98794#2673766, @chill wrote:

This needs to be changed, cf. https://lists.llvm.org/pipermail/llvm-dev/2021-March/149405.html

After doing the prototype implementation here https://reviews.llvm.org/D100397
I'm now reasonably convinced it minimally affects *this* patch, and so will go ahead and commit it.

[chill]

Updated with only whitespace changes.

[chill]

Fixed a merge conflict with https://reviews.llvm.org/D100467

clang/include/clang/CodeGen/CGFunctionInfo.h

 
 private:
   llvm::Type *TypeData; // canHaveCoerceToType()
   union {
     llvm::Type *PaddingType; // canHavePaddingType()
     llvm::Type *UnpaddedCoerceAndExpandType; // isCoerceAndExpand()
   };
   union {
-    unsigned DirectOffset;     // isDirect() || isExtend()
-    unsigned IndirectAlign;    // isIndirect()
+    struct {
+      unsigned Offset;     // isDirect() || isExtend()
+      unsigned Align;
+    } DirectAttr;
+    struct {
+      unsigned Align;    // isIndirect()
+      unsigned AddrSpace;
+    } IndirectAttr;
     unsigned AllocaFieldIndex; // isInAlloca()
   };
   Kind TheKind;
-  unsigned IndirectAddrSpace : 24; // isIndirect()
   bool PaddingInReg : 1;
   bool InAllocaSRet : 1;    // isInAlloca()
   bool InAllocaIndirect : 1;// isInAlloca()
   bool IndirectByVal : 1;   // isIndirect()
   bool IndirectRealign : 1; // isIndirect()
   bool SRetAfterThis : 1;   // isIndirect()
   bool InReg : 1;           // isDirect() || isExtend() || isIndirect()
   bool CanBeFlattened: 1;   // isDirect()
private:
 
   void setUnpaddedCoerceToType(llvm::Type *T) {
     assert(isCoerceAndExpand());
     UnpaddedCoerceAndExpandType = T;
   }
 
 public:
   ABIArgInfo(Kind K = Direct)
-      : TypeData(nullptr), PaddingType(nullptr), DirectOffset(0), TheKind(K),
-        IndirectAddrSpace(0), PaddingInReg(false), InAllocaSRet(false),
+      : TypeData(nullptr), PaddingType(nullptr), DirectAttr{0, 0}, TheKind(K),
+        PaddingInReg(false), InAllocaSRet(false),
         InAllocaIndirect(false), IndirectByVal(false), IndirectRealign(false),
         SRetAfterThis(false), InReg(false), CanBeFlattened(false),
         SignExt(false) {}
 
   static ABIArgInfo getDirect(llvm::Type *T = nullptr, unsigned Offset = 0,
                               llvm::Type *Padding = nullptr,
-                              bool CanBeFlattened = true) {
+                              bool CanBeFlattened = true,
+                              unsigned Align = 0) {
     auto AI = ABIArgInfo(Direct);
     AI.setCoerceToType(T);
     AI.setPaddingType(Padding);
     AI.setDirectOffset(Offset);
+    AI.setDirectAlign(Align);
     AI.setCanBeFlattened(CanBeFlattened);
     return AI;
   }
   static ABIArgInfo getDirectInReg(llvm::Type *T = nullptr) {
     auto AI = getDirect(T);
     AI.setInReg(true);
     return AI;
   }
 
   static ABIArgInfo getSignExtend(QualType Ty, llvm::Type *T = nullptr) {
     assert(Ty->isIntegralOrEnumerationType() && "Unexpected QualType");
     auto AI = ABIArgInfo(Extend);
     AI.setCoerceToType(T);
     AI.setPaddingType(nullptr);
     AI.setDirectOffset(0);
+    AI.setDirectAlign(0);
     AI.setSignExt(true);
     return AI;
   }
 
   static ABIArgInfo getZeroExtend(QualType Ty, llvm::Type *T = nullptr) {
     assert(Ty->isIntegralOrEnumerationType() && "Unexpected QualType");
     auto AI = ABIArgInfo(Extend);
     AI.setCoerceToType(T);
     AI.setPaddingType(nullptr);
     AI.setDirectOffset(0);
+    AI.setDirectAlign(0);
     AI.setSignExt(false);
     return AI;
   }
 
   // ABIArgInfo will record the argument as being extended based on the sign
   // of its type.
   static ABIArgInfo getExtend(QualType Ty, llvm::Type *T = nullptr) {
     assert(Ty->isIntegralOrEnumerationType() && "Unexpected QualType");
#endif
 
   bool canHaveCoerceToType() const {
     return isDirect() || isExtend() || isCoerceAndExpand();
   }
 
   // Direct/Extend accessors
   unsigned getDirectOffset() const {
     assert((isDirect() || isExtend()) && "Not a direct or extend kind");
-    return DirectOffset;
+    return DirectAttr.Offset;
   }
   void setDirectOffset(unsigned Offset) {
     assert((isDirect() || isExtend()) && "Not a direct or extend kind");
-    DirectOffset = Offset;
+    DirectAttr.Offset = Offset;
+  }
+
+  unsigned getDirectAlign() const {
+    assert((isDirect() || isExtend()) && "Not a direct or extend kind");
+    return DirectAttr.Align;
+  }
+  void setDirectAlign(unsigned Align) {
+    assert((isDirect() || isExtend()) && "Not a direct or extend kind");
+    DirectAttr.Align = Align;
   }
 
   bool isSignExt() const {
     assert(isExtend() && "Invalid kind!");
     return SignExt;
   }
   void setSignExt(bool SExt) {
     assert(isExtend() && "Invalid kind!");
#endif
   void setInReg(bool IR) {
     assert((isDirect() || isExtend() || isIndirect()) && "Invalid kind!");
     InReg = IR;
   }
 
   // Indirect accessors
   CharUnits getIndirectAlign() const {
     assert((isIndirect() || isIndirectAliased()) && "Invalid kind!");
-    return CharUnits::fromQuantity(IndirectAlign);
+    return CharUnits::fromQuantity(IndirectAttr.Align);
   }
   void setIndirectAlign(CharUnits IA) {
     assert((isIndirect() || isIndirectAliased()) && "Invalid kind!");
-    IndirectAlign = IA.getQuantity();
+    IndirectAttr.Align = IA.getQuantity();
   }
 
   bool getIndirectByVal() const {
     assert(isIndirect() && "Invalid kind!");
     return IndirectByVal;
   }
   void setIndirectByVal(bool IBV) {
     assert(isIndirect() && "Invalid kind!");
     IndirectByVal = IBV;
   }
 
   unsigned getIndirectAddrSpace() const {
     assert(isIndirectAliased() && "Invalid kind!");
-    return IndirectAddrSpace;
+    return IndirectAttr.AddrSpace;
   }
 
   void setIndirectAddrSpace(unsigned AddrSpace) {
     assert(isIndirectAliased() && "Invalid kind!");
-    IndirectAddrSpace = AddrSpace;
+    IndirectAttr.AddrSpace = AddrSpace;
   }
 
   bool getIndirectRealign() const {
     assert((isIndirect() || isIndirectAliased()) && "Invalid kind!");
     return IndirectRealign;
   }
   void setIndirectRealign(bool IR) {
     assert((isIndirect() || isIndirectAliased()) && "Invalid kind!");

clang/lib/CodeGen/CGCall.cpp

case ABIArgInfo::Extend:
       else
         Attrs.addAttribute(llvm::Attribute::ZExt);
       LLVM_FALLTHROUGH;
     case ABIArgInfo::Direct:
       if (ArgNo == 0 && FI.isChainCall())
         Attrs.addAttribute(llvm::Attribute::Nest);
       else if (AI.getInReg())
         Attrs.addAttribute(llvm::Attribute::InReg);
+      Attrs.addStackAlignmentAttr(llvm::MaybeAlign(AI.getDirectAlign()));
       break;
 
     case ABIArgInfo::Indirect: {
       if (AI.getInReg())
         Attrs.addAttribute(llvm::Attribute::InReg);
 
       if (AI.getIndirectByVal())
         Attrs.addByValAttr(getTypes().ConvertTypeForMem(ParamType));

clang/lib/CodeGen/TargetInfo.cpp

if (IsEmpty && Size == 0)
       return ABIArgInfo::getIgnore();
     return ABIArgInfo::getDirect(llvm::Type::getInt8Ty(getVMContext()));
   }
 
   // Homogeneous Floating-point Aggregates (HFAs) need to be expanded.
   const Type *Base = nullptr;
   uint64_t Members = 0;
   if (isHomogeneousAggregate(Ty, Base, Members)) {
+    if (Kind != AArch64ABIInfo::AAPCS)
       return ABIArgInfo::getDirect(
           llvm::ArrayType::get(CGT.ConvertType(QualType(Base, 0)), Members));
+
+    // For alignment adjusted HFAs, cap the argument alignment to 16, leave it
+    // default otherwise.
+    unsigned Align =
+        getContext().getTypeUnadjustedAlignInChars(Ty).getQuantity();
+    unsigned BaseAlign = getContext().getTypeAlignInChars(Base).getQuantity();
+    Align = (Align > BaseAlign && Align >= 16) ? 16 : 0;
+    return ABIArgInfo::getDirect(
+        llvm::ArrayType::get(CGT.ConvertType(QualType(Base, 0)), Members), 0,
+        nullptr, true, Align);
   }
 
   // Aggregates <= 16 bytes are passed directly in registers or on the stack.
   if (Size <= 128) {
     // On RenderScript, coerce Aggregates <= 16 bytes to an integer array of
     // same size and alignment.
     if (getTarget().isRenderScriptTarget()) {
       return coerceToIntArray(Ty, getContext(), getVMContext());

clang/test/CodeGen/aarch64-args-hfa.c

+// RUN: %clang_cc1 -triple aarch64-none-eabi -emit-llvm -o - %s | FileCheck %s --check-prefixes=CHECK,CHECK-AAPCS
+// RUN: %clang_cc1 -triple arm64-apple-ios7.0 -target-abi darwinpcs -emit-llvm -o - %s | FileCheck %s --check-prefixes=CHECK,CHECK-DARWIN
+// RUN: %clang_cc1 -triple aarch64-linux-gnu -emit-llvm -o - -x c %s | FileCheck %s --check-prefixes=CHECK,CHECK-AAPCS
+
+typedef struct {
+  float v[2];
+} S0;
+
+// CHECK: define{{.*}} float @f0([2 x float] %h.coerce)
+float f0(S0 h) {
+  return h.v[0];
+}
+
+// CHECK: define{{.*}} float @f0_call()
+// CHECK: %call = call float @f0([2 x float] %1)
+float f0_call() {
+  S0 h = {1.0f, 2.0f};
+  return f0(h);
+}
+typedef struct {
+  double v[2];
+} S1;
+
+// CHECK: define{{.*}} double @f1([2 x double] %h.coerce)
+double f1(S1 h) {
+  return h.v[0];
+}
+
+// CHECK: define{{.*}} double @f1_call()
+// CHECK: %call = call double @f1([2 x double] %1
+double f1_call() {
+  S1 h = {1.0, 2.0};
+  return f1(h);
+}
+typedef struct {
+  __attribute__((__aligned__(16))) double v[2];
+} S2;
+
+// CHECK-AAPCS:  define{{.*}} double @f2([2 x double] alignstack(16) %h.coerce)
+// CHECK-DARWIN: define{{.*}} double @f2([2 x double] %h.coerce)
+double f2(S2 h) {
+  return h.v[0];
+}
+
+// CHECK: define{{.*}} double @f2_call()
+// CHECK-AAPCS:  %call = call double @f2([2 x double] alignstack(16) %1)
+// CHECK-DARWIN: %call = call double @f2([2 x double] %1
+double f2_call() {
+  S2 h = {1.0, 2.0};
+  return f2(h);
+}
+
+typedef struct {
+  __attribute__((__aligned__(32))) double v[4];
+} S3;
+
+// CHECK-AAPCS:  define{{.*}} double @f3([4 x double] alignstack(16) %h.coerce)
+// CHECK-DARWIN: define{{.*}} double @f3([4 x double] %h.coerce)
+double f3(S3 h) {
+  return h.v[0];
+}
+
+// CHECK: define{{.*}} double @f3_call()
+// CHECK-AAPCS:  %call = call double @f3([4 x double] alignstack(16) %1)
+// CHECK-DARWIN: %call = call double @f3([4 x double] %1
+double f3_call() {
+  S3 h = {1.0, 2.0};
+  return f3(h);
+}

llvm/docs/LangRef.rst

``immarg``
     call site or arbitrary function.
 
 ``noundef``
     This attribute applies to parameters and return values. If the value
     representation contains any undefined or poison bits, the behavior is
     undefined. Note that this does not refer to padding introduced by the
     type's storage representation.
 
+``alignstack(<n>)``
+    This indicates the alignment that should be considered by the backend when
+    assigning this parameter to a stack slot during calling convention
+    lowering. The enforcement of the specified alignment is target-dependent,
+    as target-specific calling convention rules may override this value. This
+    attribute serves the purpose of carrying language specific alignment
+    information that is not mapped to base types in the backend (for example,
+    over-alignment specification through language attributes).
+
 .. _gc:
 
 Garbage Collector Strategy Names
 --------------------------------
 
 Each function may specify a garbage collector strategy name, which is simply a
 string:
 

llvm/include/llvm/CodeGen/TargetCallingConv.h

private:
     unsigned IsPreallocated : 1; ///< ByVal without the copy
     unsigned IsSplitEnd : 1;   ///< Last part of a split
     unsigned IsSwiftSelf : 1;  ///< Swift self parameter
     unsigned IsSwiftError : 1; ///< Swift error parameter
     unsigned IsCFGuardTarget : 1; ///< Control Flow Guard target
     unsigned IsHva : 1;        ///< HVA field for
     unsigned IsHvaStart : 1;   ///< HVA structure start
     unsigned IsSecArgPass : 1; ///< Second argument
-    unsigned ByValOrByRefAlign : 4; ///< Log 2 of byval/byref alignment
+    unsigned MemAlign : 4;     ///< Log 2 of alignment when arg is passed in memory (including byval/byref)
     unsigned OrigAlign : 5;    ///< Log 2 of original alignment
     unsigned IsInConsecutiveRegsLast : 1;
     unsigned IsInConsecutiveRegs : 1;
     unsigned IsCopyElisionCandidate : 1; ///< Argument copy elision candidate
     unsigned IsPointer : 1;
 
     unsigned ByValOrByRefSize; ///< Byval or byref struct size
 
     unsigned PointerAddrSpace; ///< Address space of pointer argument
 
-    /// Set the alignment used by byref or byval parameters.
-    void setAlignImpl(Align A) {
-      ByValOrByRefAlign = encode(A);
-      assert(getNonZeroByValAlign() == A && "bitfield overflow");
-    }
-
   public:
     ArgFlagsTy()
       : IsZExt(0), IsSExt(0), IsInReg(0), IsSRet(0), IsByVal(0), IsByRef(0),
           IsNest(0), IsReturned(0), IsSplit(0), IsInAlloca(0), IsPreallocated(0),
           IsSplitEnd(0), IsSwiftSelf(0), IsSwiftError(0), IsCFGuardTarget(0),
-          IsHva(0), IsHvaStart(0), IsSecArgPass(0), ByValOrByRefAlign(0),
+          IsHva(0), IsHvaStart(0), IsSecArgPass(0), MemAlign(0),
           OrigAlign(0), IsInConsecutiveRegsLast(0), IsInConsecutiveRegs(0),
           IsCopyElisionCandidate(0), IsPointer(0), ByValOrByRefSize(0),
           PointerAddrSpace(0) {
       static_assert(sizeof(*this) == 3 * sizeof(unsigned), "flags are too big");
     }
 
     bool isZExt() const { return IsZExt; }
     void setZExt() { IsZExt = 1; }
public:
     void setSplitEnd()  { IsSplitEnd = 1; }
 
     bool isCopyElisionCandidate()  const { return IsCopyElisionCandidate; }
     void setCopyElisionCandidate() { IsCopyElisionCandidate = 1; }
 
     bool isPointer()  const { return IsPointer; }
     void setPointer() { IsPointer = 1; }
 
-    Align getNonZeroByValAlign() const {
-      MaybeAlign A = decodeMaybeAlign(ByValOrByRefAlign);
-      assert(A && "ByValAlign must be defined");
-      return *A;
+    Align getNonZeroMemAlign() const {
+      return decodeMaybeAlign(MemAlign).valueOrOne();
     }
-    void setByValAlign(Align A) {
-      assert(isByVal() && !isByRef());
-      setAlignImpl(A);
+
+    void setMemAlign(Align A) {
+      MemAlign = encode(A);
+      assert(getNonZeroMemAlign() == A && "bitfield overflow");
     }
 
-    void setByRefAlign(Align A) {
-      assert(!isByVal() && isByRef());
-      setAlignImpl(A);
+    Align getNonZeroByValAlign() const {
+      assert(isByVal());
+      MaybeAlign A = decodeMaybeAlign(MemAlign);
+      assert(A && "ByValAlign must be defined");
+      return *A;
     }
 
     Align getNonZeroOrigAlign() const {
       return decodeMaybeAlign(OrigAlign).valueOrOne();
     }
+
     void setOrigAlign(Align A) {
       OrigAlign = encode(A);
       assert(getNonZeroOrigAlign() == A && "bitfield overflow");
     }
 
     unsigned getByValSize() const {
       assert(isByVal() && !isByRef());
       return ByValOrByRefSize;

llvm/include/llvm/IR/Argument.h

public:
   /// If this is a byval or inalloca argument, return its alignment.
   /// FIXME: Remove this function once transition to Align is over.
   /// Use getParamAlign() instead.
   unsigned getParamAlignment() const;
 
   /// If this is a byval or inalloca argument, return its alignment.
   MaybeAlign getParamAlign() const;
 
+  MaybeAlign getParamStackAlign() const;
+
   /// If this is a byval argument, return its type.
   Type *getParamByValType() const;
 
   /// If this is an sret argument, return its type.
   Type *getParamStructRetType() const;
 
   /// If this is a byref argument, return its type.
   Type *getParamByRefType() const;

llvm/include/llvm/IR/Attributes.h

public:
   }
 
   /// Return the alignment of the return value.
   MaybeAlign getRetAlignment() const;
 
   /// Return the alignment for the specified function parameter.
   MaybeAlign getParamAlignment(unsigned ArgNo) const;
 
+  /// Return the stack alignment for the specified function parameter.
+  MaybeAlign getParamStackAlignment(unsigned ArgNo) const;
+
   /// Return the byval type for the specified function parameter.
   Type *getParamByValType(unsigned ArgNo) const;
 
   /// Return the sret type for the specified function parameter.
   Type *getParamStructRetType(unsigned ArgNo) const;
 
   /// Return the byref type for the specified function parameter.
   Type *getParamByRefType(unsigned ArgNo) const;

llvm/include/llvm/IR/Function.h

if (const auto MA = getParamAlign(ArgNo))
       return MA->value();
     return 0;
   }
 
   MaybeAlign getParamAlign(unsigned ArgNo) const {
     return AttributeSets.getParamAlignment(ArgNo);
   }
 
+  MaybeAlign getParamStackAlign(unsigned ArgNo) const {
+    return AttributeSets.getParamStackAlignment(ArgNo);
+  }
+
   /// Extract the byval type for a parameter.
   Type *getParamByValType(unsigned ArgNo) const {
     return AttributeSets.getParamByValType(ArgNo);
   }
 
   /// Extract the sret type for a parameter.
   Type *getParamStructRetType(unsigned ArgNo) const {
     return AttributeSets.getParamStructRetType(ArgNo);

llvm/include/llvm/IR/InstrTypes.h

LLVM_ATTRIBUTE_DEPRECATED(unsigned getParamAlignment(unsigned ArgNo) const,
     return 0;
   }
 
   /// Extract the alignment for a call or parameter (0=unknown).
   MaybeAlign getParamAlign(unsigned ArgNo) const {
     return Attrs.getParamAlignment(ArgNo);
   }
 
+  MaybeAlign getParamStackAlign(unsigned ArgNo) const {
+    return Attrs.getParamStackAlignment(ArgNo);
+  }
+
   /// Extract the byval type for a call or parameter.
   Type *getParamByValType(unsigned ArgNo) const {
     Type *Ty = Attrs.getParamByValType(ArgNo);
     return Ty ? Ty : getArgOperand(ArgNo)->getType()->getPointerElementType();
   }
 
   /// Extract the preallocated type for a call or parameter.
   Type *getParamPreallocatedType(unsigned ArgNo) const {

llvm/lib/AsmParser/LLParser.cpp

while (true) {
     }
     case lltok::kw_align: {
       MaybeAlign Alignment;
       if (parseOptionalAlignment(Alignment, true))
         return true;
       B.addAlignmentAttr(Alignment);
       continue;
     }
+    case lltok::kw_alignstack: {
+      unsigned Alignment;
+      if (parseOptionalStackAlignment(Alignment))
+        return true;
+      B.addStackAlignmentAttr(Alignment);
+      continue;
+    }
     case lltok::kw_byval: {
       Type *Ty;
       if (parseRequiredTypeAttr(Ty, lltok::kw_byval))
         return true;
       B.addByValAttr(Ty);
       continue;
     }
     case lltok::kw_sret: {
while (true) {
     case lltok::kw_returned:        B.addAttribute(Attribute::Returned); break;
     case lltok::kw_signext:         B.addAttribute(Attribute::SExt); break;
     case lltok::kw_swifterror:      B.addAttribute(Attribute::SwiftError); break;
     case lltok::kw_swiftself:       B.addAttribute(Attribute::SwiftSelf); break;
     case lltok::kw_writeonly:       B.addAttribute(Attribute::WriteOnly); break;
     case lltok::kw_zeroext:         B.addAttribute(Attribute::ZExt); break;
     case lltok::kw_immarg:          B.addAttribute(Attribute::ImmArg); break;
 
-    case lltok::kw_alignstack:
     case lltok::kw_alwaysinline:
     case lltok::kw_argmemonly:
     case lltok::kw_builtin:
     case lltok::kw_inlinehint:
     case lltok::kw_jumptable:
     case lltok::kw_minsize:
     case lltok::kw_mustprogress:
     case lltok::kw_naked:

llvm/lib/CodeGen/GlobalISel/CallLowering.cpp

 template <typename FuncInfoTy>
 void CallLowering::setArgFlags(CallLowering::ArgInfo &Arg, unsigned OpIdx,
                                const DataLayout &DL,
                                const FuncInfoTy &FuncInfo) const {
   auto &Flags = Arg.Flags[0];
   const AttributeList &Attrs = FuncInfo.getAttributes();
   addArgFlagsFromAttributes(Flags, Attrs, OpIdx);
 
+  Align MemAlign;
   if (Flags.isByVal() || Flags.isInAlloca() || Flags.isPreallocated()) {
     Type *ElementTy = cast<PointerType>(Arg.Ty)->getElementType();
 
     auto Ty = Attrs.getAttribute(OpIdx, Attribute::ByVal).getValueAsType();
     Flags.setByValSize(DL.getTypeAllocSize(Ty ? Ty : ElementTy));
 
     // For ByVal, alignment should be passed from FE.  BE will guess if
     // this info is not there but there are cases it cannot get right.
-    Align FrameAlign;
-    if (auto ParamAlign = FuncInfo.getParamAlign(OpIdx - 1))
-      FrameAlign = *ParamAlign;
+    if (auto ParamAlign = FuncInfo.getParamStackAlign(OpIdx - 1))
+      MemAlign = *ParamAlign;
+    else if ((ParamAlign = FuncInfo.getParamAlign(OpIdx - 1)))
+      MemAlign = *ParamAlign;
     else
-      FrameAlign = Align(getTLI()->getByValTypeAlignment(ElementTy, DL));
-    Flags.setByValAlign(FrameAlign);
+      MemAlign = Align(getTLI()->getByValTypeAlignment(ElementTy, DL));
+  } else if (auto ParamAlign = FuncInfo.getParamStackAlign(OpIdx - 1)) {
+    MemAlign = *ParamAlign;
+  } else {
+    MemAlign = Align(DL.getABITypeAlign(Arg.Ty));
   }
+  Flags.setMemAlign(MemAlign);
   Flags.setOrigAlign(DL.getABITypeAlign(Arg.Ty));
 
   // Don't try to use the returned attribute if the argument is marked as
   // swiftself, since it won't be passed in x0.
   if (Flags.isSwiftSelf())
     Flags.setReturned(false);
 }
 

llvm/lib/CodeGen/SelectionDAG/FastISel.cpp

if (Arg.IsPreallocated) {
       Flags.setPreallocated();
       // Set the byval flag for CCAssignFn callbacks that don't know about
       // preallocated. This way we can know how many bytes we should've
       // allocated and how many bytes a callee cleanup function will pop.  If we
       // port preallocated to more targets, we'll have to add custom
       // preallocated handling in the various CC lowering callbacks.
       Flags.setByVal();
     }
+    MaybeAlign MemAlign = Arg.Alignment;
     if (Arg.IsByVal || Arg.IsInAlloca || Arg.IsPreallocated) {
       PointerType *Ty = cast<PointerType>(Arg.Ty);
       Type *ElementTy = Ty->getElementType();
       unsigned FrameSize =
           DL.getTypeAllocSize(Arg.ByValType ? Arg.ByValType : ElementTy);
 
       // For ByVal, alignment should come from FE. BE will guess if this info
       // is not there, but there are cases it cannot get right.
-      MaybeAlign FrameAlign = Arg.Alignment;
-      if (!FrameAlign)
-        FrameAlign = Align(TLI.getByValTypeAlignment(ElementTy, DL));
+      if (!MemAlign)
+        MemAlign = Align(TLI.getByValTypeAlignment(ElementTy, DL));
       Flags.setByValSize(FrameSize);
-      Flags.setByValAlign(*FrameAlign);
+    } else if (!MemAlign) {
+      MemAlign = DL.getABITypeAlign(Arg.Ty);
     }
+    Flags.setMemAlign(*MemAlign);
     if (Arg.IsNest)
       Flags.setNest();
     if (NeedsRegBlock)
       Flags.setInConsecutiveRegs();
     Flags.setOrigAlign(DL.getABITypeAlign(Arg.Ty));
-
     CLI.OutVals.push_back(Arg.Val);
     CLI.OutFlags.push_back(Flags);
   }
 
   if (!fastLowerCall(CLI))
     return false;
 
   // Set all unused physreg defs as dead.

llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp

for (unsigned Value = 0, NumValues = ValueVTs.size(); Value != NumValues;
       SDValue Op = SDValue(Args[i].Node.getNode(),
                            Args[i].Node.getResNo() + Value);
       ISD::ArgFlagsTy Flags;
 
       // Certain targets (such as MIPS), may have a different ABI alignment
       // for a type depending on the context. Give the target a chance to
       // specify the alignment it wants.
       const Align OriginalAlignment(getABIAlignmentForCallingConv(ArgTy, DL));
+      Flags.setOrigAlign(OriginalAlignment);
 
       if (Args[i].Ty->isPointerTy()) {
         Flags.setPointer();
         Flags.setPointerAddrSpace(
             cast<PointerType>(Args[i].Ty)->getAddressSpace());
       }
       if (Args[i].IsZExt)
         Flags.setZExt();
       if (Args[i].IsSExt)
         Flags.setSExt();
for (unsigned Value = 0, NumValues = ValueVTs.size(); Value != NumValues;
         Flags.setInAlloca();
         // Set the byval flag for CCAssignFn callbacks that don't know about
         // inalloca.  This way we can know how many bytes we should've allocated
         // and how many bytes a callee cleanup function will pop.  If we port
         // inalloca to more targets, we'll have to add custom inalloca handling
         // in the various CC lowering callbacks.
         Flags.setByVal();
       }
+      Align MemAlign;
       if (Args[i].IsByVal || Args[i].IsInAlloca || Args[i].IsPreallocated) {
         PointerType *Ty = cast<PointerType>(Args[i].Ty);
         Type *ElementTy = Ty->getElementType();
 
         unsigned FrameSize = DL.getTypeAllocSize(
             Args[i].ByValType ? Args[i].ByValType : ElementTy);
         Flags.setByValSize(FrameSize);
 
         // info is not there but there are cases it cannot get right.
-        Align FrameAlign;
         if (auto MA = Args[i].Alignment)
-          FrameAlign = *MA;
+          MemAlign = *MA;
         else
-          FrameAlign = Align(getByValTypeAlignment(ElementTy, DL));
-        Flags.setByValAlign(FrameAlign);
+          MemAlign = Align(getByValTypeAlignment(ElementTy, DL));
+      } else if (auto MA = Args[i].Alignment) {
+        MemAlign = *MA;
+      } else {
+        MemAlign = OriginalAlignment;
       }
+      Flags.setMemAlign(MemAlign);
       if (Args[i].IsNest)
         Flags.setNest();
       if (NeedsRegBlock)
         Flags.setInConsecutiveRegs();
-      Flags.setOrigAlign(OriginalAlignment);
 
       MVT PartVT = getRegisterTypeForCallingConv(CLI.RetTy->getContext(),
                                                  CLI.CallConv, VT);
       unsigned NumParts = getNumRegistersForCallingConv(CLI.RetTy->getContext(),
                                                         CLI.CallConv, VT);
       SmallVector<SDValue, 4> Parts(NumParts);
       ISD::NodeType ExtendKind = ISD::ANY_EXTEND;
 
for (const Argument &Arg : F.args()) {
     bool NeedsRegBlock = TLI->functionArgumentNeedsConsecutiveRegisters(
         FinalType, F.getCallingConv(), F.isVarArg());
     for (unsigned Value = 0, NumValues = ValueVTs.size();
          Value != NumValues; ++Value) {
       EVT VT = ValueVTs[Value];
       Type *ArgTy = VT.getTypeForEVT(*DAG.getContext());
       ISD::ArgFlagsTy Flags;
 
-      // Certain targets (such as MIPS), may have a different ABI alignment
-      // for a type depending on the context. Give the target a chance to
-      // specify the alignment it wants.
-      const Align OriginalAlignment(
-          TLI->getABIAlignmentForCallingConv(ArgTy, DL));
 
       if (Arg.getType()->isPointerTy()) {
         Flags.setPointer();
         Flags.setPointerAddrSpace(
             cast<PointerType>(Arg.getType())->getAddressSpace());
       }
       if (Arg.hasAttribute(Attribute::ZExt))
         Flags.setZExt();
for (unsigned Value = 0, NumValues = ValueVTs.size();
         // Set the byval flag for CCAssignFn callbacks that don't know about
         // preallocated.  This way we can know how many bytes we should've
         // allocated and how many bytes a callee cleanup function will pop.  If
         // we port preallocated to more targets, we'll have to add custom
         // preallocated handling in the various CC lowering callbacks.
         Flags.setByVal();
       }
 
+      // Certain targets (such as MIPS), may have a different ABI alignment
+      // for a type depending on the context. Give the target a chance to
+      // specify the alignment it wants.
+      const Align OriginalAlignment(
+          TLI->getABIAlignmentForCallingConv(ArgTy, DL));
+      Flags.setOrigAlign(OriginalAlignment);
+
+      Align MemAlign;
       Type *ArgMemTy = nullptr;
       if (Flags.isByVal() || Flags.isInAlloca() || Flags.isPreallocated() ||
           Flags.isByRef()) {
         if (!ArgMemTy)
           ArgMemTy = Arg.getPointeeInMemoryValueType();
 
         uint64_t MemSize = DL.getTypeAllocSize(ArgMemTy);
 
         // For in-memory arguments, size and alignment should be passed from FE.
         // BE will guess if this info is not there but there are cases it cannot
         // get right.
-        MaybeAlign MemAlign = Arg.getParamAlign();
-        if (!MemAlign)
+        if (auto ParamAlign = Arg.getParamStackAlign())
+          MemAlign = *ParamAlign;
+        else if ((ParamAlign = Arg.getParamAlign()))
+          MemAlign = *ParamAlign;
+        else
           MemAlign = Align(TLI->getByValTypeAlignment(ArgMemTy, DL));
-
-        if (Flags.isByRef()) {
+        if (Flags.isByRef())
           Flags.setByRefSize(MemSize);
-          Flags.setByRefAlign(*MemAlign);
-        } else {
+        else
           Flags.setByValSize(MemSize);
-          Flags.setByValAlign(*MemAlign);
-        }
+      } else if (auto ParamAlign = Arg.getParamStackAlign()) {
+        MemAlign = *ParamAlign;
+      } else {
+        MemAlign = OriginalAlignment;
       }
+      Flags.setMemAlign(MemAlign);
 
       if (Arg.hasAttribute(Attribute::Nest))
         Flags.setNest();
       if (NeedsRegBlock)
         Flags.setInConsecutiveRegs();
-      Flags.setOrigAlign(OriginalAlignment);
       if (ArgCopyElisionCandidates.count(&Arg))
         Flags.setCopyElisionCandidate();
       if (Arg.hasAttribute(Attribute::Returned))
         Flags.setReturned();
 
       MVT RegisterVT = TLI->getRegisterTypeForCallingConv(
           *CurDAG->getContext(), F.getCallingConv(), VT);
       unsigned NumRegs = TLI->getNumRegistersForCallingConv(

llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp

void TargetLoweringBase::ArgListEntry::setAttributes(const CallBase *Call,
   IsSRet = Call->paramHasAttr(ArgIdx, Attribute::StructRet);
   IsNest = Call->paramHasAttr(ArgIdx, Attribute::Nest);
   IsByVal = Call->paramHasAttr(ArgIdx, Attribute::ByVal);
   IsPreallocated = Call->paramHasAttr(ArgIdx, Attribute::Preallocated);
   IsInAlloca = Call->paramHasAttr(ArgIdx, Attribute::InAlloca);
   IsReturned = Call->paramHasAttr(ArgIdx, Attribute::Returned);
   IsSwiftSelf = Call->paramHasAttr(ArgIdx, Attribute::SwiftSelf);
   IsSwiftError = Call->paramHasAttr(ArgIdx, Attribute::SwiftError);
-  Alignment = Call->getParamAlign(ArgIdx);
+  Alignment = Call->getParamStackAlign(ArgIdx);
   ByValType = nullptr;
-  if (IsByVal)
+  if (IsByVal) {
     ByValType = Call->getParamByValType(ArgIdx);
+    if (!Alignment)
+      Alignment = Call->getParamAlign(ArgIdx);
+  }
   PreallocatedType = nullptr;
   if (IsPreallocated)
     PreallocatedType = Call->getParamPreallocatedType(ArgIdx);
 }
 
 /// Generate a libcall taking the given operands as arguments and returning a
 /// result of type RetVT.
 std::pair<SDValue, SDValue>

llvm/lib/IR/Attributes.cpp

 MaybeAlign AttributeList::getRetAlignment() const {
   return getAttributes(ReturnIndex).getAlignment();
 }
 
 MaybeAlign AttributeList::getParamAlignment(unsigned ArgNo) const {
   return getAttributes(ArgNo + FirstArgIndex).getAlignment();
 }
 
+MaybeAlign AttributeList::getParamStackAlignment(unsigned ArgNo) const {
+  return getAttributes(ArgNo + FirstArgIndex).getStackAlignment();
+}
+
 Type *AttributeList::getParamByValType(unsigned Index) const {
   return getAttributes(Index+FirstArgIndex).getByValType();
 }
 
 Type *AttributeList::getParamStructRetType(unsigned Index) const {
   return getAttributes(Index + FirstArgIndex).getStructRetType();
 }
 

llvm/lib/IR/Function.cpp

unsigned Argument::getParamAlignment() const {
   return getParent()->getParamAlignment(getArgNo());
 }
 
 MaybeAlign Argument::getParamAlign() const {
   assert(getType()->isPointerTy() && "Only pointers have alignments");
   return getParent()->getParamAlign(getArgNo());
 }
 
+MaybeAlign Argument::getParamStackAlign() const {
+  return getParent()->getParamStackAlign(getArgNo());
+}
+
 Type *Argument::getParamByValType() const {
   assert(getType()->isPointerTy() && "Only pointers have byval types");
   return getParent()->getParamByValType(getArgNo());
 }
 
 Type *Argument::getParamStructRetType() const {
   assert(getType()->isPointerTy() && "Only pointers have sret types");
   return getParent()->getParamStructRetType(getArgNo());

llvm/lib/IR/Verifier.cpp

static bool isFuncOnlyAttr(Attribute::AttrKind Kind) {
   case Attribute::StackProtectReq:
   case Attribute::StackProtectStrong:
   case Attribute::SafeStack:
   case Attribute::ShadowCallStack:
   case Attribute::NoRedZone:
   case Attribute::NoImplicitFloat:
   case Attribute::Naked:
   case Attribute::InlineHint:
-  case Attribute::StackAlignment:
   case Attribute::UWTable:
   case Attribute::VScaleRange:
   case Attribute::NonLazyBind:
   case Attribute::ReturnsTwice:
   case Attribute::SanitizeAddress:
   case Attribute::SanitizeHWAddress:
   case Attribute::SanitizeMemTag:
   case Attribute::SanitizeThread:
static bool isFuncOnlyAttr(Attribute::AttrKind Kind) {
   return false;
 }
 
 /// Return true if this is a function attribute that can also appear on
 /// arguments.
 static bool isFuncOrArgAttr(Attribute::AttrKind Kind) {
   return Kind == Attribute::ReadOnly || Kind == Attribute::WriteOnly ||
          Kind == Attribute::ReadNone || Kind == Attribute::NoFree ||
-         Kind == Attribute::Preallocated;
+         Kind == Attribute::Preallocated || Kind == Attribute::StackAlignment;
 }
 
 void Verifier::verifyAttributeTypes(AttributeSet Attrs, bool IsFunction,
                                     const Value *V) {
   for (Attribute A : Attrs) {
     if (A.isStringAttribute())
       continue;
 
if (!PL || !PR)
     return false;
   return PL->getAddressSpace() == PR->getAddressSpace();
 }
 
 static AttrBuilder getParameterABIAttributes(int I, AttributeList Attrs) {
   static const Attribute::AttrKind ABIAttrs[] = {
       Attribute::StructRet,    Attribute::ByVal,     Attribute::InAlloca,
       Attribute::InReg,        Attribute::SwiftSelf, Attribute::SwiftError,
-      Attribute::Preallocated, Attribute::ByRef};
+      Attribute::Preallocated, Attribute::ByRef, Attribute::StackAlignment};
   AttrBuilder Copy;
   for (auto AK : ABIAttrs) {
     if (Attrs.hasParamAttribute(I, AK))
       Copy.addAttribute(AK);
   }
 
   // `align` is ABI-affecting only in combination with `byval` or `byref`.
   if (Attrs.hasParamAttribute(I, Attribute::Alignment) &&

llvm/lib/Target/AArch64/AArch64CallingConvention.cpp

for (int I = 0; I < 8; I++)
         State.DeallocateReg(ZRegList[I]);
 
     // All pending members have now been allocated
     PendingMembers.clear();
     return true;
   }
 
   unsigned Size = LocVT.getSizeInBits() / 8;
-  const Align StackAlign =
-      State.getMachineFunction().getDataLayout().getStackAlignment();
-  const Align OrigAlign = ArgFlags.getNonZeroOrigAlign();
-  const Align Alignment = std::min(OrigAlign, StackAlign);
-
   for (auto &It : PendingMembers) {
-    It.convertToMem(State.AllocateStack(Size, std::max(Alignment, SlotAlign)));
+    It.convertToMem(State.AllocateStack(Size, SlotAlign));
     State.addLoc(It);
     SlotAlign = Align(1);
   }
 
   // All pending members have now been allocated
   PendingMembers.clear();
   return true;
 }
static bool CC_AArch64_Custom_Block(unsigned &ValNo, MVT &ValVT, MVT &LocVT,
   }
 
   if (!LocVT.isScalableVector()) {
     // Mark all regs in the class as unavailable
     for (auto Reg : RegList)
       State.AllocateReg(Reg);
   }
 
-  const Align SlotAlign = Subtarget.isTargetDarwin() ? Align(1) : Align(8);
+  const Align StackAlign =
+      State.getMachineFunction().getDataLayout().getStackAlignment();
+  const Align MemAlign = ArgFlags.getNonZeroMemAlign();
+  Align SlotAlign = std::min(MemAlign, StackAlign);
+  if (!Subtarget.isTargetDarwin())
+    SlotAlign = std::max(SlotAlign, Align(8));
 
   return finishStackBlock(PendingMembers, LocVT, ArgFlags, State, SlotAlign);
 }
 
 // TableGen provides definitions of the calling convention analysis entry
 // points.
 #include "AArch64GenCallingConv.inc"

llvm/test/Bitcode/compatibility.ll

 declare i8* @f.param.returned(i8* returned)
 ; CHECK: declare i8* @f.param.returned(i8* returned)
 declare void @f.param.nonnull(i8* nonnull)
 ; CHECK: declare void @f.param.nonnull(i8* nonnull)
 declare void @f.param.dereferenceable(i8* dereferenceable(4))
 ; CHECK: declare void @f.param.dereferenceable(i8* dereferenceable(4))
 declare void @f.param.dereferenceable_or_null(i8* dereferenceable_or_null(4))
 ; CHECK: declare void @f.param.dereferenceable_or_null(i8* dereferenceable_or_null(4))
+declare void @f.param.stack_align([2 x double] alignstack(16))
+; CHECK: declare void @f.param.stack_align([2 x double] alignstack(16))
 
 ; Functions -- unnamed_addr and local_unnamed_addr
 declare void @f.unnamed_addr() unnamed_addr
 ; CHECK: declare void @f.unnamed_addr() unnamed_addr
 declare void @f.local_unnamed_addr() local_unnamed_addr
 ; CHECK: declare void @f.local_unnamed_addr() local_unnamed_addr
 
 ; Functions -- fn Attrs (Function attributes)

llvm/test/CodeGen/AArch64/arm64-abi-hfa-args.ll

+; RUN: llc < %s -mtriple=arm64-none-eabi | FileCheck %s
+
+; Over-aligned HFA argument placed on register - one element per register
+define double @test_hfa_align_arg_reg([2 x double] alignstack(16) %h.coerce) local_unnamed_addr #0 {
+entry:
+; CHECK-LABEL: test_hfa_align_arg_reg:
+; CHECK-NOT: mov
+; CHECK-NOT: ld
+; CHECK: ret
+  %h.coerce.fca.0.extract = extractvalue [2 x double] %h.coerce, 0
+  ret double %h.coerce.fca.0.extract
+}
+
+; Call with over-aligned HFA argument placed on register - one element per register
+define double @test_hfa_align_call_reg() local_unnamed_addr #0 {
+entry:
+; CHECK-LABEL: test_hfa_align_call_reg:
+; CHECK-DAG: fmov  d0, #1.00000000
+; CHECK-DAG: fmov  d1, #2.00000000
+; CHECK:     bl    test_hfa_align_arg_reg
+  %call = call double @test_hfa_align_arg_reg([2 x double] alignstack(16) [double 1.000000e+00, double 2.000000e+00])
+  ret double %call
+}
+
+; Over-aligned HFA argument placed on stack - stack round up to alignment
+define double @test_hfa_align_arg_stack(double %d0, double %d1, double %d2, double %d3, double %d4, double %d5, double %d6, double %d7, float %f, [2 x double] alignstack(16) %h.coerce) local_unnamed_addr #0 {
+entry:
+; CHECK-LABEL: test_hfa_align_arg_stack:
+; CHECK:       ldr  d0, [sp, #16]
+; CHECK-NEXT:  ret
+  %h.coerce.fca.0.extract = extractvalue [2 x double] %h.coerce, 0
+  ret double %h.coerce.fca.0.extract
+}