Remove SequentialType from the type heirarchy.

[efriedma]

Summary

Now that we have scalable vectors, there's a distinction that isn't getting captured in the original SequentialType: some vectors don't have a known element count, so counting the number of elements doesn't make sense.

In some cases, there's a better way to express the commonality using other methods. If we're dealing with GEPs, there's GEP methods; if we're dealing with a ConstantDataSequential, we can query its element type directly.

In the relatively few remaining cases, I just decided to write out the type checks. We're talking about relatively few places, and I think the abstraction doesn't really carry its weight. (See thread "[RFC] Refactor class hierarchy of VectorType in the IR" on llvmdev.)

[efriedma]

Rebase. Fix a test failure. A few other minor tweaks.

[efriedma]

I looked at getting rid of getSequentialNumElements/getSequentialElementType, but I'm not sure that would actually make the code more clear. I avoided them in cases where they clearly weren't helpful.

clang/lib/CodeGen/CGDecl.cpp

1053 ↗

(On Diff #250910)

Despite the appearance, this doesn't actually affect the result for vectors: we try to recursively rewrite the elements, but we don't actually support any rewrites for types other than StructType and ArrayType, so the rewrite just produces the input constant.

[efriedma]

Minor fix to AMDGPU changes.

[nhaehnle]

The AMDGPU changes LGTM, modulo a stylistic nitpick.

As for the overall change, I'm perfectly fine with the general direction of SequentialType, though I don't know where we stand in terms of enough people having had a chance to look at it.

I'm concerned though about having getSequentialElementType and getSequentialNumElements apply to different sets of types. That's bound to lead to confusion which causes bugs. Can we have one name for all types that are homogenous sequences of the same element type, and a different name for all types which are *fixed-length* homogenous sequences?

llvm/lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp

370–372

Please put braces for the outer if (multi-line body).

[efriedma]

I guess the API names could be a bit confusing, yes. Any suggestions for better names?

[ctetreau]

I'm concerned though about having getSequentialElementType and getSequentialNumElements apply to different sets of types. That's bound to lead to confusion which causes bugs. Can we have one name for all types that are homogenous sequences of the same element type, and a different name for all types which are *fixed-length* homogenous sequences?

isSequentialType() returns true for VectorType and ArrayType. It should be changed to only return true if the vector type is not scalable, and getSequentialNumElements() should assert isSequentialType() before attempting the casts.

In the future, when my proposed FixedVectorType is added, isSequentialType will return true for ArrayType and FixedVectorType only.

[efriedma]

I went through the exercise of completely removing the notion of a "sequential" type; the result comes out something like this.

Broadly, the places that might be able to use some notion of "sequential" mostly fall into two categories:

1. Places where we're examining indexes of a GEP.
2. Constants of array/vector type.

For GEPs, I think using GEP-specific API makes sense, to make it clear what notion of "element" we're considering. For the other cases, not really sure if an abstraction would be helpful.

[sdesmalen]

Thank you for this patch! Personally I find the code more readable without the SequentialType abstraction and the use of the GEP interface (getTypeAtIndex(Type, Idx)) you added in D75660.
Like @nhaehnle I don't know whether this patch needs some more eyes before you commit this, but it looks good to me!

llvm/lib/Transforms/IPO/GlobalOpt.cpp

442

nit: With only two uses, I'm not sure these functions make the code much more readable than just expanding it in their uses below.

[ctetreau]

clang/lib/CodeGen/CGDecl.cpp

1059 ↗

(On Diff #251218)

I assume the reasoning here is that [-0.0f, ...] isn't the zero initialized value, but why make this change now? Seems unrelated.

1078 ↗

(On Diff #251218)

The fact that you have to ask this question tells me that you should probably just make this handle vectors.

You could add a templated helper function above this function that is basically just the original body of the SequentialType branch.

[efriedma]

clang/lib/CodeGen/CGDecl.cpp

1059 ↗

(On Diff #251218)

I'll post this separately for review, since it's a little complicated.

1078 ↗

(On Diff #251218)

Well, no, the original code doesn't handle vectors either. The issue here would be that we need to pad out the vector with additional elements, or something like that.

[ctetreau]

clang/lib/CodeGen/CGDecl.cpp

1078 ↗

(On Diff #251218)

At a cursory glance, it seemed to me that the addition of this comment implied that it used to handle the tail padding. However, after having dug into what this function does, assuming that vectors only ever actually contain scalars, then it seems that this does currently just return the original vector and that this patch does not change the behavior.

Given this, your comment really kind of just says "FIXME: is this function correct?", a question that applies to basically every function. If you have reason to believe that the answer to your question is yes, you should probably add why you think so to this comment. Otherwise, I think you should remove it.

[efriedma]

Rebased. (CGDecl.cpp changes landed separately.)

[ctetreau]

I'd like the see the clang-format complaints actually addressed, but don't bother unless you need to upload another patch for some other reason anyways.

That aside, LGTM. This has been looked at by 3 people (plus yourself). So far, nobody has had any serious objections. Perhaps enough has been enough?

[efriedma]

Address lint comments

[efriedma]

Fix bug in GlobalOpt. Add fixes for MLIR.

[ctetreau]

mlir/lib/Target/LLVMIR/ModuleTranslation.cpp

81

Readability: This works, but is a little strange. I'd rewrite:

static llvm::Type *getInnermostElementType(llvm::Type *type) {
  while (isa<llvm::ArrayType>(type) || isa<llvm::VectorType>(type)) {
    if (auto *arrayTy = dyn_cast<llvm::ArrayType>(type))
      type = arrayTy->getElementType();
    else
      type = cast<llvm::VectorType>(type)->getElementType();
  }
  return type;
}

[ctetreau]

What's the status on this?

[ftynse]

LGTM for MLIR part.

mlir/lib/Target/LLVMIR/ModuleTranslation.cpp

60

Nit: can we update the error message not to mention "sequential LLVM type" anymore if sequential type is no longer a thing?

[efriedma]

mlir/lib/Target/LLVMIR/ModuleTranslation.cpp

60

Other MLIR code in this file still refers to them as sequential types, but I guess it might make sense to change in this context.

Is it realistically possible to hit this path? I'm not sure how to write a testcase.

clang/lib/CodeGen/CGExprConstant.cpp

 /// Hint indicates the location at which we'd like to split, but may be
 /// ignored.
 bool ConstantAggregateBuilder::split(size_t Index, CharUnits Hint) {
   NaturalLayout = false;
   llvm::Constant *C = Elems[Index];
   CharUnits Offset = Offsets[Index];
 
   if (auto *CA = dyn_cast<llvm::ConstantAggregate>(C)) {
+    // Expand the sequence into its contained elements.
+    // FIXME: This assumes vector elements are byte-sized.
     replace(Elems, Index, Index + 1,
             llvm::map_range(llvm::seq(0u, CA->getNumOperands()),
                             [&](unsigned Op) { return CA->getOperand(Op); }));
-    if (auto *Seq = dyn_cast<llvm::SequentialType>(CA->getType())) {
+    if (isa<llvm::ArrayType>(CA->getType()) ||
+        isa<llvm::VectorType>(CA->getType())) {
       // Array or vector.
-      CharUnits ElemSize = getSize(Seq->getElementType());
+      llvm::Type *ElemTy =
+          llvm::GetElementPtrInst::getTypeAtIndex(CA->getType(), (uint64_t)0);
+      CharUnits ElemSize = getSize(ElemTy);
       replace(
           Offsets, Index, Index + 1,
           llvm::map_range(llvm::seq(0u, CA->getNumOperands()),
                           [&](unsigned Op) { return Offset + Op * ElemSize; }));
     } else {
       // Must be a struct.
       auto *ST = cast<llvm::StructType>(CA->getType());
       const llvm::StructLayout *Layout =
           CGM.getDataLayout().getStructLayout(ST);
       replace(Offsets, Index, Index + 1,
               llvm::map_range(
                   llvm::seq(0u, CA->getNumOperands()), [&](unsigned Op) {
                     return Offset + CharUnits::fromQuantity(
                                         Layout->getElementOffset(Op));
                   }));
     }
     return true;
   }
 
   if (auto *CDS = dyn_cast<llvm::ConstantDataSequential>(C)) {
+    // Expand the sequence into its contained elements.
+    // FIXME: This assumes vector elements are byte-sized.
     // FIXME: If possible, split into two ConstantDataSequentials at Hint.
     CharUnits ElemSize = getSize(CDS->getElementType());
     replace(Elems, Index, Index + 1,
             llvm::map_range(llvm::seq(0u, CDS->getNumElements()),
                             [&](unsigned Elem) {
                               return CDS->getElementAsConstant(Elem);
                             }));
     replace(Offsets, Index, Index + 1,
             llvm::map_range(
                 llvm::seq(0u, CDS->getNumElements()),
                 [&](unsigned Elem) { return Offset + Elem * ElemSize; }));
     return true;
   }
 
   if (isa<llvm::ConstantAggregateZero>(C)) {
+    // Split into two zeros at the hinted offset.
     CharUnits ElemSize = getSize(C);
     assert(Hint > Offset && Hint < Offset + ElemSize && "nothing to split");
     replace(Elems, Index, Index + 1,
             {getZeroes(Hint - Offset), getZeroes(Offset + ElemSize - Hint)});
     replace(Offsets, Index, Index + 1, {Offset, Hint});
     return true;
   }
 
   if (isa<llvm::UndefValue>(C)) {
+    // Drop undef; it doesn't contribute to the final layout.
     replace(Elems, Index, Index + 1, {});
     replace(Offsets, Index, Index + 1, {});
     return true;
   }
 
   // FIXME: We could split a ConstantInt if the need ever arose.
   // We don't need to do this to handle bit-fields because we always eagerly
   // split them into 1-byte chunks.

llvm/include/llvm/IR/Constants.h

 #include <cstddef>
 #include <cstdint>
 
 namespace llvm {
 
 class ArrayType;
 class IntegerType;
 class PointerType;
-class SequentialType;
 class StructType;
 class VectorType;
 template <class ConstantClass> struct ConstantAggrKeyType;
 
 /// Base class for constants with no operands.
 ///
 /// These constants have no operands; they represent their data directly.
 /// Since they can be in use by unrelated modules (and are never based on
public:
   /// element as a double.
   double getElementAsDouble(unsigned i) const;
 
   /// Return a Constant for a specified index's element.
   /// Note that this has to compute a new constant to return, so it isn't as
   /// efficient as getElementAsInteger/Float/Double.
   Constant *getElementAsConstant(unsigned i) const;
 
-  /// Specialize the getType() method to always return a SequentialType, which
-  /// reduces the amount of casting needed in parts of the compiler.
-  inline SequentialType *getType() const {
-    return cast<SequentialType>(Value::getType());
-  }
-
   /// Return the element type of the array/vector.
   Type *getElementType() const;
 
   /// Return the number of elements in the array or vector.
   unsigned getNumElements() const;
 
   /// Return the size (in bytes) of each element in the array/vector.
   /// The size of the elements is known to be a multiple of one byte.

llvm/include/llvm/IR/DerivedTypes.h

 unsigned Type::getStructNumElements() const {
   return cast<StructType>(this)->getNumElements();
 }
 
 Type *Type::getStructElementType(unsigned N) const {
   return cast<StructType>(this)->getElementType(N);
 }
 
-/// This is the superclass of the array and vector type classes. Both of these
-/// represent "arrays" in memory. The array type represents a specifically sized
-/// array, and the vector type represents a specifically sized array that allows
-/// for use of SIMD instructions. SequentialType holds the common features of
-/// both, which stem from the fact that both lay their components out in memory
-/// identically.
-class SequentialType : public Type {
-  Type *ContainedType;               ///< Storage for the single contained type.
+/// Class to represent array types.
+class ArrayType : public Type {
+  /// The element type of the array.
+  Type *ContainedType;
+  /// Number of elements in the array.
   uint64_t NumElements;
 
-protected:
-  SequentialType(TypeID TID, Type *ElType, uint64_t NumElements)
-      : Type(ElType->getContext(), TID), ContainedType(ElType),
-        NumElements(NumElements) {
-    ContainedTys = &ContainedType;
-    NumContainedTys = 1;
-  }
-
-public:
-  SequentialType(const SequentialType &) = delete;
-  SequentialType &operator=(const SequentialType &) = delete;
-
-  /// For scalable vectors, this will return the minimum number of elements
-  /// in the vector.
-  uint64_t getNumElements() const { return NumElements; }
-  Type *getElementType() const { return ContainedType; }
-
-  /// Methods for support type inquiry through isa, cast, and dyn_cast.
-  static bool classof(const Type *T) {
-    return T->getTypeID() == ArrayTyID || T->getTypeID() == VectorTyID;
-  }
-};
-
-/// Class to represent array types.
-class ArrayType : public SequentialType {
   ArrayType(Type *ElType, uint64_t NumEl);
 
 public:
   ArrayType(const ArrayType &) = delete;
   ArrayType &operator=(const ArrayType &) = delete;
 
+  uint64_t getNumElements() const { return NumElements; }
+  Type *getElementType() const { return ContainedType; }
+
   /// This static method is the primary way to construct an ArrayType
   static ArrayType *get(Type *ElementType, uint64_t NumElements);
 
   /// Return true if the specified type is valid as a element type.
   static bool isValidElementType(Type *ElemTy);
 
   /// Methods for support type inquiry through isa, cast, and dyn_cast.
   static bool classof(const Type *T) {
     return T->getTypeID() == ArrayTyID;
   }
 };
 
 uint64_t Type::getArrayNumElements() const {
   return cast<ArrayType>(this)->getNumElements();
 }
 
 /// Class to represent vector types.
-class VectorType : public SequentialType {
+class VectorType : public Type {
   /// A fully specified VectorType is of the form <vscale x n x Ty>. 'n' is the
   /// minimum number of elements of type Ty contained within the vector, and
   /// 'vscale x' indicates that the total element count is an integer multiple
   /// of 'n', where the multiple is either guaranteed to be one, or is
   /// statically unknown at compile time.
   ///
   /// If the multiple is known to be 1, then the extra term is discarded in
   /// textual IR:
   ///
   /// <4 x i32>          - a vector containing 4 i32s
   /// <vscale x 4 x i32> - a vector containing an unknown integer multiple
   ///                      of 4 i32s
 
+  /// The element type of the vector.
+  Type *ContainedType;
+  /// Minumum number of elements in the vector.
+  uint64_t NumElements;
+
   VectorType(Type *ElType, unsigned NumEl, bool Scalable = false);
   VectorType(Type *ElType, ElementCount EC);
 
   // If true, the total number of elements is an unknown multiple of the
-  // minimum 'NumElements' from SequentialType. Otherwise the total number
-  // of elements is exactly equal to 'NumElements'.
+  // minimum 'NumElements'. Otherwise the total number of elements is exactly
+  // equal to 'NumElements'.
   bool Scalable;
 
 public:
   VectorType(const VectorType &) = delete;
   VectorType &operator=(const VectorType &) = delete;
 
+  /// For scalable vectors, this will return the minimum number of elements
+  /// in the vector.
+  uint64_t getNumElements() const { return NumElements; }
+  Type *getElementType() const { return ContainedType; }
+
   /// This static method is the primary way to construct an VectorType.
   static VectorType *get(Type *ElementType, ElementCount EC);
   static VectorType *get(Type *ElementType, unsigned NumElements,
                          bool Scalable = false) {
     return VectorType::get(ElementType, {NumElements, Scalable});
   }
 
   /// This static method gets a VectorType with the same number of elements as

llvm/include/llvm/IR/GetElementPtrTypeIterator.h

Type *getIndexedType() const {
         return T;
       return CurTy.get<StructType *>()->getTypeAtIndex(getOperand());
     }
 
     Value *getOperand() const { return const_cast<Value *>(&**OpIt); }
 
     generic_gep_type_iterator& operator++() {   // Preincrement
       Type *Ty = getIndexedType();
-      if (auto *STy = dyn_cast<SequentialType>(Ty)) {
-        CurTy = STy->getElementType();
-        NumElements = STy->getNumElements();
+      if (auto *ATy = dyn_cast<ArrayType>(Ty)) {
+        CurTy = ATy->getElementType();
+        NumElements = ATy->getNumElements();
+      } else if (auto *VTy = dyn_cast<VectorType>(Ty)) {
+        CurTy = VTy->getElementType();
+        if (VTy->isScalable())
+          NumElements = Unbounded;
+        else
+          NumElements = VTy->getNumElements();
       } else
         CurTy = dyn_cast<StructType>(Ty);
       ++OpIt;
       return *this;
     }
 
     generic_gep_type_iterator operator++(int) { // Postincrement
       generic_gep_type_iterator tmp = *this; ++*this; return tmp;

llvm/include/llvm/IR/Type.h

protected:
 
   /// A pointer to the array of Types contained by this Type. For example, this
   /// includes the arguments of a function type, the elements of a structure,
   /// the pointee of a pointer, the element type of an array, etc. This pointer
   /// may be 0 for types that don't contain other types (Integer, Double,
   /// Float).
   Type * const *ContainedTys = nullptr;
 
-  static bool isSequentialType(TypeID TyID) {
-    return TyID == ArrayTyID || TyID == VectorTyID;
-  }
-
 public:
   /// Print the current type.
   /// Omit the type details if \p NoDetails == true.
   /// E.g., let %st = type { i32, i16 }
   /// When \p NoDetails is true, we only print %st.
   /// Put differently, \p NoDetails prints the type as if
   /// inlined with the operands when printing an instruction.
   void print(raw_ostream &O, bool IsForDebug = false,
public:
   inline Type *getFunctionParamType(unsigned i) const;
   inline unsigned getFunctionNumParams() const;
   inline bool isFunctionVarArg() const;
 
   inline StringRef getStructName() const;
   inline unsigned getStructNumElements() const;
   inline Type *getStructElementType(unsigned N) const;
 
-  inline Type *getSequentialElementType() const {
-    assert(isSequentialType(getTypeID()) && "Not a sequential type!");
-    return ContainedTys[0];
-  }
-
   inline uint64_t getArrayNumElements() const;
 
   Type *getArrayElementType() const {
     assert(getTypeID() == ArrayTyID);
     return ContainedTys[0];
   }
 
   inline bool getVectorIsScalable() const;

llvm/lib/Analysis/BasicAliasAnalysis.cpp

if (!isa<ArrayType>(GetElementPtrInst::getIndexedType(
       return MayAlias;
     IntermediateIndices.push_back(GEP1->getOperand(i + 1));
   }
 
   auto *Ty = GetElementPtrInst::getIndexedType(
     GEP1->getSourceElementType(), IntermediateIndices);
   StructType *LastIndexedStruct = dyn_cast<StructType>(Ty);
 
-  if (isa<SequentialType>(Ty)) {
+  if (isa<ArrayType>(Ty) || isa<VectorType>(Ty)) {
     // We know that:
     // - both GEPs begin indexing from the exact same pointer;
     // - the last indices in both GEPs are constants, indexing into a sequential
-    //   type (array or pointer);
+    //   type (array or vector);
     // - both GEPs only index through arrays prior to that.
     //
     // Because array indices greater than the number of elements are valid in
     // GEPs, unless we know the intermediate indices are identical between
     // GEP1 and GEP2 we cannot guarantee that the last indexed arrays don't
     // partially overlap. We also need to check that the loaded size matches
     // the element size, otherwise we could still have overlap.
-    const uint64_t ElementSize =
-        DL.getTypeStoreSize(cast<SequentialType>(Ty)->getElementType());
+    Type *LastElementTy = GetElementPtrInst::getTypeAtIndex(Ty, (uint64_t)0);
+    const uint64_t ElementSize = DL.getTypeStoreSize(LastElementTy);
     if (V1Size != ElementSize || V2Size != ElementSize)
       return MayAlias;
 
     for (unsigned i = 0, e = GEP1->getNumIndices() - 1; i != e; ++i)
       if (GEP1->getOperand(i + 1) != GEP2->getOperand(i + 1))
         return MayAlias;
 
     // Now we know that the array/pointer that GEP1 indexes into and that

llvm/lib/Analysis/ConstantFolding.cpp

while (true) {
       ByteOffset = 0;
       CurEltOffset = NextEltOffset;
     }
     // not reached.
   }
 
   if (isa<ConstantArray>(C) || isa<ConstantVector>(C) ||
       isa<ConstantDataSequential>(C)) {
-    Type *EltTy = C->getType()->getSequentialElementType();
-    uint64_t EltSize = DL.getTypeAllocSize(EltTy);
-    uint64_t Index = ByteOffset / EltSize;
-    uint64_t Offset = ByteOffset - Index * EltSize;
     uint64_t NumElts;
-    if (auto *AT = dyn_cast<ArrayType>(C->getType()))
+    Type *EltTy;
+    if (auto *AT = dyn_cast<ArrayType>(C->getType())) {
       NumElts = AT->getNumElements();
-    else
+      EltTy = AT->getElementType();
+    } else {
       NumElts = C->getType()->getVectorNumElements();
+      EltTy = C->getType()->getVectorElementType();
+    }
+    uint64_t EltSize = DL.getTypeAllocSize(EltTy);
+    uint64_t Index = ByteOffset / EltSize;
+    uint64_t Offset = ByteOffset - Index * EltSize;
 
     for (; Index != NumElts; ++Index) {
       if (!ReadDataFromGlobal(C->getAggregateElement(Index), Offset, CurPtr,
                               BytesLeft, DL))
         return false;
 
       uint64_t BytesWritten = EltSize - Offset;
       assert(BytesWritten <= EltSize && "Not indexing into this element?");
if (!Ty->isStructTy()) {
         if (!NewIdxs.empty())
           break;
 
         Ty = SrcElemTy;
 
         // Only handle pointers to sized types, not pointers to functions.
         if (!Ty->isSized())
           return nullptr;
-      } else if (auto *ATy = dyn_cast<SequentialType>(Ty)) {
-        Ty = ATy->getElementType();
       } else {
-        // We've reached some non-indexable type.
+        Type *NextTy = GetElementPtrInst::getTypeAtIndex(Ty, (uint64_t)0);
+        if (!NextTy)
           break;
+        Ty = NextTy;
       }
 
       // Determine which element of the array the offset points into.
       APInt ElemSize(BitWidth, DL.getTypeAllocSize(Ty));
       if (ElemSize == 0) {
         // The element size is 0. This may be [0 x Ty]*, so just use a zero
         // index for this level and proceed to the next level to see if it can
         // accommodate the offset.

llvm/lib/Analysis/ScalarEvolution.cpp

for (const SCEV *IndexExpr : IndexExprs) {
     } else {
       // Update CurTy to its element type.
       if (FirstIter) {
         assert(isa<PointerType>(CurTy) &&
                "The first index of a GEP indexes a pointer");
         CurTy = GEP->getSourceElementType();
         FirstIter = false;
       } else {
-        CurTy = cast<SequentialType>(CurTy)->getElementType();
+        CurTy = GetElementPtrInst::getTypeAtIndex(CurTy, (uint64_t)0);
       }
       // For an array, add the element offset, explicitly scaled.
       const SCEV *ElementSize = getSizeOfExpr(IntIdxTy, CurTy);
       // Getelementptr indices are signed.
       IndexExpr = getTruncateOrSignExtend(IndexExpr, IntIdxTy);
 
       // Multiply the index by the element size to compute the element offset.
       const SCEV *LocalOffset = getMulExpr(IndexExpr, ElementSize, Wrap);

llvm/lib/Bitcode/Reader/BitcodeReader.cpp

case bitc::CST_CODE_CSTRING: { // CSTRING: [values]
       V = ConstantDataArray::getString(Context, Elts,
                                        BitCode == bitc::CST_CODE_CSTRING);
       break;
     }
     case bitc::CST_CODE_DATA: {// DATA: [n x value]
       if (Record.empty())
         return error("Invalid record");
 
-      Type *EltTy = cast<SequentialType>(CurTy)->getElementType();
+      Type *EltTy;
+      if (auto *Array = dyn_cast<ArrayType>(CurTy))
+        EltTy = Array->getElementType();
+      else
+        EltTy = cast<VectorType>(CurTy)->getElementType();
       if (EltTy->isIntegerTy(8)) {
         SmallVector<uint8_t, 16> Elts(Record.begin(), Record.end());
         if (isa<VectorType>(CurTy))
           V = ConstantDataVector::get(Context, Elts);
         else
           V = ConstantDataArray::get(Context, Elts);
       } else if (EltTy->isIntegerTy(16)) {
         SmallVector<uint16_t, 16> Elts(Record.begin(), Record.end());

llvm/lib/Bitcode/Writer/BitcodeWriter.cpp

if (C->isNullValue()) {
 
       if (isCStrChar6)
         AbbrevToUse = CString6Abbrev;
       else if (isCStr7)
         AbbrevToUse = CString7Abbrev;
     } else if (const ConstantDataSequential *CDS =
                   dyn_cast<ConstantDataSequential>(C)) {
       Code = bitc::CST_CODE_DATA;
-      Type *EltTy = CDS->getType()->getElementType();
+      Type *EltTy = CDS->getElementType();
       if (isa<IntegerType>(EltTy)) {
         for (unsigned i = 0, e = CDS->getNumElements(); i != e; ++i)
           Record.push_back(CDS->getElementAsInteger(i));
       } else {
         for (unsigned i = 0, e = CDS->getNumElements(); i != e; ++i)
           Record.push_back(
               CDS->getElementAsAPFloat(i).bitcastToAPInt().getLimitedValue());
       }

llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp

if (isa<IntegerType>(CDS->getElementType())) {
     }
   } else {
     Type *ET = CDS->getElementType();
     for (unsigned I = 0, E = CDS->getNumElements(); I != E; ++I)
       emitGlobalConstantFP(CDS->getElementAsAPFloat(I), ET, AP);
   }
 
   unsigned Size = DL.getTypeAllocSize(CDS->getType());
-  unsigned EmittedSize = DL.getTypeAllocSize(CDS->getType()->getElementType()) *
-                        CDS->getNumElements();
+  unsigned EmittedSize =
+      DL.getTypeAllocSize(CDS->getElementType()) * CDS->getNumElements();
   assert(EmittedSize <= Size && "Size cannot be less than EmittedSize!");
   if (unsigned Padding = Size - EmittedSize)
     AP.OutStreamer->emitZeros(Padding);
 }
 
 static void emitGlobalConstantArray(const DataLayout &DL,
                                     const ConstantArray *CA, AsmPrinter &AP,
                                     const Constant *BaseCV, uint64_t Offset) {

llvm/lib/IR/ConstantFold.cpp

if (PointerType *PTy = dyn_cast<PointerType>(V->getType()))
     if (PointerType *DPTy = dyn_cast<PointerType>(DestTy))
       if (PTy->getAddressSpace() == DPTy->getAddressSpace()
           && PTy->getElementType()->isSized()) {
         SmallVector<Value*, 8> IdxList;
         Value *Zero =
           Constant::getNullValue(Type::getInt32Ty(DPTy->getContext()));
         IdxList.push_back(Zero);
         Type *ElTy = PTy->getElementType();
-        while (ElTy != DPTy->getElementType()) {
-          if (StructType *STy = dyn_cast<StructType>(ElTy)) {
-            if (STy->getNumElements() == 0) break;
-            ElTy = STy->getElementType(0);
-            IdxList.push_back(Zero);
-          } else if (SequentialType *STy =
-                     dyn_cast<SequentialType>(ElTy)) {
-            ElTy = STy->getElementType();
-            IdxList.push_back(Zero);
-          } else {
-            break;
-          }
+        while (ElTy && ElTy != DPTy->getElementType()) {
+          ElTy = GetElementPtrInst::getTypeAtIndex(ElTy, (uint64_t)0);
         }
 
         if (ElTy == DPTy->getElementType())
           // This GEP is inbounds because all indices are zero.
           return ConstantExpr::getInBoundsGetElementPtr(PTy->getElementType(),
                                                         V, IdxList);
       }
 
Constant *llvm::ConstantFoldInsertValueInstruction(Constant *Agg,
   // Base case: no indices, so replace the entire value.
   if (Idxs.empty())
     return Val;
 
   unsigned NumElts;
   if (StructType *ST = dyn_cast<StructType>(Agg->getType()))
     NumElts = ST->getNumElements();
   else
-    NumElts = cast<SequentialType>(Agg->getType())->getNumElements();
+    NumElts = cast<ArrayType>(Agg->getType())->getNumElements();
 
   SmallVector<Constant*, 32> Result;
   for (unsigned i = 0; i != NumElts; ++i) {
     Constant *C = Agg->getAggregateElement(i);
     if (!C) return nullptr;
 
     if (Idxs[0] == i)
       C = ConstantFoldInsertValueInstruction(C, Val, Idxs.slice(1));
 
     Result.push_back(C);
   }
 
   if (StructType *ST = dyn_cast<StructType>(Agg->getType()))
     return ConstantStruct::get(ST, Result);
-  if (ArrayType *AT = dyn_cast<ArrayType>(Agg->getType()))
-    return ConstantArray::get(AT, Result);
-  return ConstantVector::get(Result);
+  return ConstantArray::get(cast<ArrayType>(Agg->getType()), Result);
 }
 
 Constant *llvm::ConstantFoldUnaryInstruction(unsigned Opcode, Constant *C) {
   assert(Instruction::isUnaryOp(Opcode) && "Non-unary instruction detected");
 
   // Handle scalar UndefValue and scalable vector UndefValue. Fixed-length
   // vectors are always evaluated per element.
   bool IsScalableVector =
if (InRangeIndex && i == *InRangeIndex + 1) {
       // the following index, as that will cause the inrange index to point to
       // the wrong element.
       continue;
     }
     if (isa<StructType>(Ty)) {
       // The verify makes sure that GEPs into a struct are in range.
       continue;
     }
-    auto *STy = cast<SequentialType>(Ty);
-    if (isa<VectorType>(STy)) {
+    if (isa<VectorType>(Ty)) {
       // There can be awkward padding in after a non-power of two vector.
       Unknown = true;
       continue;
     }
+    auto *STy = cast<ArrayType>(Ty);
     if (ConstantInt *CI = dyn_cast<ConstantInt>(Idxs[i])) {
       if (isIndexInRangeOfArrayType(STy->getNumElements(), CI))
         // It's in range, skip to the next index.
         continue;
       if (CI->getSExtValue() < 0) {
         // It's out of range and negative, don't try to factor it.
         Unknown = true;
         continue;

llvm/lib/IR/Constants.cpp

void ConstantFP::destroyConstantImpl() {
   llvm_unreachable("You can't ConstantFP->destroyConstantImpl()!");
 }
 
 //===----------------------------------------------------------------------===//
 //                   ConstantAggregateZero Implementation
 //===----------------------------------------------------------------------===//
 
 Constant *ConstantAggregateZero::getSequentialElement() const {
-  return Constant::getNullValue(getType()->getSequentialElementType());
+  if (auto *AT = dyn_cast<ArrayType>(getType()))
+    return Constant::getNullValue(AT->getElementType());
+  return Constant::getNullValue(cast<VectorType>(getType())->getElementType());
 }
 
 Constant *ConstantAggregateZero::getStructElement(unsigned Elt) const {
   return Constant::getNullValue(getType()->getStructElementType(Elt));
 }
 
 Constant *ConstantAggregateZero::getElementValue(Constant *C) const {
-  if (isa<SequentialType>(getType()))
+  if (isa<ArrayType>(getType()) || isa<VectorType>(getType()))
     return getSequentialElement();
   return getStructElement(cast<ConstantInt>(C)->getZExtValue());
 }
 
 Constant *ConstantAggregateZero::getElementValue(unsigned Idx) const {
-  if (isa<SequentialType>(getType()))
+  if (isa<ArrayType>(getType()) || isa<VectorType>(getType()))
     return getSequentialElement();
   return getStructElement(Idx);
 }
 
 unsigned ConstantAggregateZero::getNumElements() const {
   Type *Ty = getType();
   if (auto *AT = dyn_cast<ArrayType>(Ty))
     return AT->getNumElements();
   if (auto *VT = dyn_cast<VectorType>(Ty))
     return VT->getNumElements();
   return Ty->getStructNumElements();
 }
 
 //===----------------------------------------------------------------------===//
 //                         UndefValue Implementation
 //===----------------------------------------------------------------------===//
 
 UndefValue *UndefValue::getSequentialElement() const {
-  return UndefValue::get(getType()->getSequentialElementType());
+  if (ArrayType *ATy = dyn_cast<ArrayType>(getType()))
+    return UndefValue::get(ATy->getElementType());
+  return UndefValue::get(cast<VectorType>(getType())->getElementType());
 }
 
 UndefValue *UndefValue::getStructElement(unsigned Elt) const {
   return UndefValue::get(getType()->getStructElementType(Elt));
 }
 
 UndefValue *UndefValue::getElementValue(Constant *C) const {
-  if (isa<SequentialType>(getType()))
+  if (isa<ArrayType>(getType()) || isa<VectorType>(getType()))
     return getSequentialElement();
   return getStructElement(cast<ConstantInt>(C)->getZExtValue());
 }
 
 UndefValue *UndefValue::getElementValue(unsigned Idx) const {
-  if (isa<SequentialType>(getType()))
+  if (isa<ArrayType>(getType()) || isa<VectorType>(getType()))
     return getSequentialElement();
   return getStructElement(Idx);
 }
 
 unsigned UndefValue::getNumElements() const {
   Type *Ty = getType();
-  if (auto *ST = dyn_cast<SequentialType>(Ty))
-    return ST->getNumElements();
+  if (auto *AT = dyn_cast<ArrayType>(Ty))
+    return AT->getNumElements();
+  if (auto *VT = dyn_cast<VectorType>(Ty))
+    return VT->getNumElements();
   return Ty->getStructNumElements();
 }
 
 //===----------------------------------------------------------------------===//
 //                            ConstantXXX Classes
 //===----------------------------------------------------------------------===//
 
 template <typename ItTy, typename EltTy>

 Type *GetElementPtrConstantExpr::getResultElementType() const {
   return ResElementTy;
 }
 
 //===----------------------------------------------------------------------===//
 //                       ConstantData* implementations
 
 Type *ConstantDataSequential::getElementType() const {
-  return getType()->getElementType();
+  if (ArrayType *ATy = dyn_cast<ArrayType>(getType()))
+    return ATy->getElementType();
+  return cast<VectorType>(getType())->getElementType();
 }
 
 StringRef ConstantDataSequential::getRawDataValues() const {
   return StringRef(DataElements, getNumElements()*getElementByteSize());
 }
 
 bool ConstantDataSequential::isElementTypeCompatible(Type *Ty) {
   if (Ty->isHalfTy() || Ty->isFloatTy() || Ty->isDoubleTy()) return true;
static bool isAllZeros(StringRef Arr) {
   return true;
 }
 
 /// This is the underlying implementation of all of the
 /// ConstantDataSequential::get methods.  They all thunk down to here, providing
 /// the correct element type.  We take the bytes in as a StringRef because
 /// we *want* an underlying "char*" to avoid TBAA type punning violations.
 Constant *ConstantDataSequential::getImpl(StringRef Elements, Type *Ty) {
-  assert(isElementTypeCompatible(Ty->getSequentialElementType()));
+#ifndef NDEBUG
+  if (ArrayType *ATy = dyn_cast<ArrayType>(Ty))
+    assert(isElementTypeCompatible(ATy->getElementType()));
+  else
+    assert(isElementTypeCompatible(cast<VectorType>(Ty)->getElementType()));
+#endif
   // If the elements are all zero or there are no elements, return a CAZ, which
   // is more dense and canonical.
   if (isAllZeros(Elements))
     return ConstantAggregateZero::get(Ty);
 
   // Do a lookup to see if we have already formed one of these.
   auto &Slot =
       *Ty->getContext()

llvm/lib/IR/Core.cpp

 LLVMTypeRef LLVMVectorType(LLVMTypeRef ElementType, unsigned ElementCount) {
   return wrap(VectorType::get(unwrap(ElementType), ElementCount));
 }
 
 LLVMTypeRef LLVMGetElementType(LLVMTypeRef WrappedTy) {
   auto *Ty = unwrap<Type>(WrappedTy);
   if (auto *PTy = dyn_cast<PointerType>(Ty))
     return wrap(PTy->getElementType());
-  return wrap(cast<SequentialType>(Ty)->getElementType());
+  if (auto *ATy = dyn_cast<ArrayType>(Ty))
+    return wrap(ATy->getElementType());
+  return wrap(cast<VectorType>(Ty)->getElementType());
 }
 
 unsigned LLVMGetNumContainedTypes(LLVMTypeRef Tp) {
     return unwrap(Tp)->getNumContainedTypes();
 }
 
 unsigned LLVMGetArrayLength(LLVMTypeRef ArrayTy) {
   return unwrap<ArrayType>(ArrayTy)->getNumElements();

llvm/lib/IR/Type.cpp

bool StructType::indexValid(const Value *V) const {
   return CU && CU->getZExtValue() < getNumElements();
 }
 
 //===----------------------------------------------------------------------===//
 //                           ArrayType Implementation
 //===----------------------------------------------------------------------===//
 
 ArrayType::ArrayType(Type *ElType, uint64_t NumEl)
-  : SequentialType(ArrayTyID, ElType, NumEl) {}
+    : Type(ElType->getContext(), ArrayTyID), ContainedType(ElType),
+      NumElements(NumEl) {
+  ContainedTys = &ContainedType;
+  NumContainedTys = 1;
+}
 
 ArrayType *ArrayType::get(Type *ElementType, uint64_t NumElements) {
   assert(isValidElementType(ElementType) && "Invalid type for array element!");
 
   LLVMContextImpl *pImpl = ElementType->getContext().pImpl;
   ArrayType *&Entry =
     pImpl->ArrayTypes[std::make_pair(ElementType, NumElements)];
 
return !ElemTy->isVoidTy() && !ElemTy->isLabelTy() &&
          !ElemTy->isTokenTy();
 }
 
 //===----------------------------------------------------------------------===//
 //                          VectorType Implementation
 //===----------------------------------------------------------------------===//
 
 VectorType::VectorType(Type *ElType, ElementCount EC)
-  : SequentialType(VectorTyID, ElType, EC.Min), Scalable(EC.Scalable) {}
+    : Type(ElType->getContext(), VectorTyID), ContainedType(ElType),
+      NumElements(EC.Min), Scalable(EC.Scalable) {
+  ContainedTys = &ContainedType;
+  NumContainedTys = 1;
+}
 
 VectorType *VectorType::get(Type *ElementType, ElementCount EC) {
   assert(EC.Min > 0 && "#Elements of a VectorType must be greater than 0");
   assert(isValidElementType(ElementType) && "Element type of a VectorType must "
                                             "be an integer, floating point, or "
                                             "pointer type.");
 
   LLVMContextImpl *pImpl = ElementType->getContext().pImpl;

llvm/lib/Linker/IRMover.cpp

bool TypeMapTy::areTypesIsomorphic(Type *DstTy, Type *SrcTy) {
   } else if (FunctionType *FT = dyn_cast<FunctionType>(DstTy)) {
     if (FT->isVarArg() != cast<FunctionType>(SrcTy)->isVarArg())
       return false;
   } else if (StructType *DSTy = dyn_cast<StructType>(DstTy)) {
     StructType *SSTy = cast<StructType>(SrcTy);
     if (DSTy->isLiteral() != SSTy->isLiteral() ||
         DSTy->isPacked() != SSTy->isPacked())
       return false;
-  } else if (auto *DSeqTy = dyn_cast<SequentialType>(DstTy)) {
-    if (DSeqTy->getNumElements() !=
-        cast<SequentialType>(SrcTy)->getNumElements())
+  } else if (auto *DArrTy = dyn_cast<ArrayType>(DstTy)) {
+    if (DArrTy->getNumElements() != cast<ArrayType>(SrcTy)->getNumElements())
+      return false;
+  } else if (auto *DVecTy = dyn_cast<VectorType>(DstTy)) {
+    if (DVecTy->getElementCount() != cast<VectorType>(SrcTy)->getElementCount())
       return false;
   }
 
   // Otherwise, we speculate that these two types will line up and recursively
   // check the subelements.
   Entry = DstTy;
   SpeculativeTypes.push_back(SrcTy);
 

llvm/lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp

 
 static bool tryPromoteAllocaToVector(AllocaInst *Alloca) {
 
   if (DisablePromoteAllocaToVector) {
     LLVM_DEBUG(dbgs() << "  Promotion alloca to vector is disabled\n");
     return false;
   }
 
-  Type *AT = Alloca->getAllocatedType();
-  SequentialType *AllocaTy = dyn_cast<SequentialType>(AT);
+  Type *AllocaTy = Alloca->getAllocatedType();
+  VectorType *VectorTy = dyn_cast<VectorType>(AllocaTy);
+  if (auto *ArrayTy = dyn_cast<ArrayType>(AllocaTy)) {
+    if (VectorType::isValidElementType(ArrayTy->getElementType()) &&
+        ArrayTy->getNumElements() > 0)
+      VectorTy = arrayTypeToVecType(ArrayTy);
+  }
 
   LLVM_DEBUG(dbgs() << "Alloca candidate for vectorization\n");
 
   // FIXME: There is no reason why we can't support larger arrays, we
   // are just being conservative for now.
   // FIXME: We also reject alloca's of the form [ 2 x [ 2 x i32 ]] or equivalent. Potentially these
   // could also be promoted but we don't currently handle this case
-  if (!AllocaTy ||
-      AllocaTy->getNumElements() > 16 ||
-      AllocaTy->getNumElements() < 2 ||
-      !VectorType::isValidElementType(AllocaTy->getElementType())) {
+  if (!VectorTy || VectorTy->getNumElements() > 16 ||
+      VectorTy->getNumElements() < 2) {
     LLVM_DEBUG(dbgs() << "  Cannot convert type to vector\n");
     return false;
   }
 
   std::map<GetElementPtrInst*, Value*> GEPVectorIdx;
   std::vector<Value*> WorkList;
   for (User *AllocaUser : Alloca->users()) {
     GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(AllocaUser);
for (User *AllocaUser : Alloca->users()) {
     for (User *GEPUser : AllocaUser->users()) {
       if (!canVectorizeInst(cast<Instruction>(GEPUser), AllocaUser))
         return false;
 
       WorkList.push_back(GEPUser);
     }
   }
 
-  VectorType *VectorTy = dyn_cast<VectorType>(AllocaTy);
-  if (!VectorTy)
-    VectorTy = arrayTypeToVecType(cast<ArrayType>(AllocaTy));
-
   LLVM_DEBUG(dbgs() << "  Converting alloca to vector " << *AllocaTy << " -> "
                     << *VectorTy << '\n');
 
   for (Value *V : WorkList) {
     Instruction *Inst = cast<Instruction>(V);
     IRBuilder<> Builder(Inst);
     switch (Inst->getOpcode()) {
     case Instruction::Load: {
-      if (Inst->getType() == AT)
+      if (Inst->getType() == AllocaTy)
         break;
 
       Type *VecPtrTy = VectorTy->getPointerTo(AMDGPUAS::PRIVATE_ADDRESS);
       Value *Ptr = cast<LoadInst>(Inst)->getPointerOperand();
       Value *Index = calculateVectorIndex(Ptr, GEPVectorIdx);
 
       Value *BitCast = Builder.CreateBitCast(Alloca, VecPtrTy);
       Value *VecValue = Builder.CreateLoad(VectorTy, BitCast);
       Value *ExtractElement = Builder.CreateExtractElement(VecValue, Index);
       Inst->replaceAllUsesWith(ExtractElement);
       Inst->eraseFromParent();
       break;
     }
     case Instruction::Store: {
       StoreInst *SI = cast<StoreInst>(Inst);
-      if (SI->getValueOperand()->getType() == AT)
+      if (SI->getValueOperand()->getType() == AllocaTy)
         break;
 
       Type *VecPtrTy = VectorTy->getPointerTo(AMDGPUAS::PRIVATE_ADDRESS);
       Value *Ptr = SI->getPointerOperand();
       Value *Index = calculateVectorIndex(Ptr, GEPVectorIdx);
       Value *BitCast = Builder.CreateBitCast(Alloca, VecPtrTy);
       Value *VecValue = Builder.CreateLoad(VectorTy, BitCast);
       Value *NewVecValue = Builder.CreateInsertElement(VecValue,

llvm/lib/Target/Hexagon/HexagonCommonGEP.cpp

struct GepNode {
     }
 
     friend raw_ostream &operator<< (raw_ostream &OS, const GepNode &GN);
   };
 
   Type *next_type(Type *Ty, Value *Idx) {
     if (auto *PTy = dyn_cast<PointerType>(Ty))
       return PTy->getElementType();
-    // Advance the type.
-    if (!Ty->isStructTy()) {
-      Type *NexTy = cast<SequentialType>(Ty)->getElementType();
-      return NexTy;
-    }
-    // Otherwise it is a struct type.
-    ConstantInt *CI = dyn_cast<ConstantInt>(Idx);
-    assert(CI && "Struct type with non-constant index");
-    int64_t i = CI->getValue().getSExtValue();
-    Type *NextTy = cast<StructType>(Ty)->getElementType(i);
-    return NextTy;
+    return GetElementPtrInst::getTypeAtIndex(Ty, Idx);
   }
 
   raw_ostream &operator<< (raw_ostream &OS, const GepNode &GN) {
     OS << "{ {";
     bool Comma = false;
     if (GN.Flags & GepNode::Root) {
       OS << "root";
       Comma = true;

llvm/lib/Transforms/IPO/ArgumentPromotion.cpp

bool ArgumentPromotionPass::isDenselyPacked(Type *type, const DataLayout &DL) {
   if (!type->isSized())
     return false;
 
   // If the alloc size is not equal to the storage size, then there are padding
   // bytes. For x86_fp80 on x86-64, size: 80 alloc size: 128.
   if (DL.getTypeSizeInBits(type) != DL.getTypeAllocSizeInBits(type))
     return false;
 
-  if (!isa<StructType>(type) && !isa<SequentialType>(type))
-    return true;
+  // FIXME: This isn't the right way to check for padding in vectors with
+  // non-byte-size elements.
+  if (VectorType *seqTy = dyn_cast<VectorType>(type))
+    return isDenselyPacked(seqTy->getElementType(), DL);
 
-  // For homogenous sequential types, check for padding within members.
-  if (SequentialType *seqTy = dyn_cast<SequentialType>(type))
+  // For array types, check for padding within members.
+  if (ArrayType *seqTy = dyn_cast<ArrayType>(type))
     return isDenselyPacked(seqTy->getElementType(), DL);
 
+  if (!isa<StructType>(type))
+    return true;
+
   // Check for padding within and between elements of a struct.
   StructType *StructTy = cast<StructType>(type);
   const StructLayout *Layout = DL.getStructLayout(StructTy);
   uint64_t StartPos = 0;
   for (unsigned i = 0, E = StructTy->getNumElements(); i < E; ++i) {
     Type *ElTy = StructTy->getElementType(i);
     if (!isDenselyPacked(ElTy, DL))
       return false;

llvm/lib/Transforms/IPO/GlobalOpt.cpp

static bool isLeakCheckerRoot(GlobalVariable *GV) {
   SmallVector<Type *, 4> Types;
   Types.push_back(GV->getValueType());
 
   unsigned Limit = 20;
   do {
     Type *Ty = Types.pop_back_val();
     switch (Ty->getTypeID()) {
       default: break;
-      case Type::PointerTyID: return true;
+      case Type::PointerTyID:
+        return true;
+      case Type::VectorTyID:
+        if (cast<VectorType>(Ty)->getElementType()->isPointerTy())
+          return true;
+        break;
       case Type::ArrayTyID:
-      case Type::VectorTyID: {
-        SequentialType *STy = cast<SequentialType>(Ty);
-        Types.push_back(STy->getElementType());
+        Types.push_back(cast<ArrayType>(Ty)->getElementType());
         break;
-      }
       case Type::StructTyID: {
         StructType *STy = cast<StructType>(Ty);
         if (STy->isOpaque()) return true;
         for (StructType::element_iterator I = STy->element_begin(),
                  E = STy->element_end(); I != E; ++I) {
           Type *InnerTy = *I;
           if (isa<PointerType>(InnerTy)) return true;
-          if (isa<StructType>(InnerTy) || isa<SequentialType>(InnerTy))
+          if (isa<StructType>(InnerTy) || isa<ArrayType>(InnerTy) ||
+              isa<VectorType>(InnerTy))
             Types.push_back(InnerTy);
         }
         break;
       }
     }
     if (--Limit == 0) return true;
   } while (!Types.empty());
   return false;
for (User *U : GV->users()) {
     // Check the gep and it's users are safe to SRA
     if (!isSafeSROAGEP(U))
       return false;
   }
 
   return true;
 }
 
+static bool IsSRASequential(Type *T) {
+  return isa<ArrayType>(T) || isa<VectorType>(T);
+}
+static uint64_t GetSRASequentialNumElements(Type *T) {
+  if (ArrayType *AT = dyn_cast<ArrayType>(T))
+    return AT->getNumElements();
+  return cast<VectorType>(T)->getNumElements();
+}
+static Type *GetSRASequentialElementType(Type *T) {
+  if (ArrayType *AT = dyn_cast<ArrayType>(T))
+    return AT->getElementType();
+  return cast<VectorType>(T)->getElementType();
+}
 static bool CanDoGlobalSRA(GlobalVariable *GV) {
   Constant *Init = GV->getInitializer();
 
   if (isa<StructType>(Init->getType())) {
     // nothing to check
-  } else if (SequentialType *STy = dyn_cast<SequentialType>(Init->getType())) {
-    if (STy->getNumElements() > 16 && GV->hasNUsesOrMore(16))
+  } else if (IsSRASequential(Init->getType())) {
+    if (GetSRASequentialNumElements(Init->getType()) > 16 &&
+        GV->hasNUsesOrMore(16))
       return false; // It's not worth it.
   } else
     return false;
 
   return GlobalUsersSafeToSRA(GV);
 }
 
 /// Copy over the debug info for a variable to its SRA replacements.
for (User *GEP : GV->users()) {
     unsigned ElementIdx = cast<ConstantInt>(GEP->getOperand(2))->getZExtValue();
     if (NewGlobals.count(ElementIdx) == 1)
       continue; // we`ve already created replacement variable
     assert(NewGlobals.count(ElementIdx) == 0);
 
     Type *ElTy = nullptr;
     if (StructType *STy = dyn_cast<StructType>(Ty))
       ElTy = STy->getElementType(ElementIdx);
-    else if (SequentialType *STy = dyn_cast<SequentialType>(Ty))
-      ElTy = STy->getElementType();
+    else
+      ElTy = GetSRASequentialElementType(Ty);
     assert(ElTy);
 
     Constant *In = Init->getAggregateElement(ElementIdx);
     assert(In && "Couldn't get element of initializer?");
 
     GlobalVariable *NGV = new GlobalVariable(
         ElTy, false, GlobalVariable::InternalLinkage, In,
         GV->getName() + "." + Twine(ElementIdx), GV->getThreadLocalMode(),
if (StructType *STy = dyn_cast<StructType>(Ty)) {
           Align(DL.getABITypeAlignment(STy->getElementType(ElementIdx))))
         NGV->setAlignment(NewAlign);
 
       // Copy over the debug info for the variable.
       uint64_t Size = DL.getTypeAllocSizeInBits(NGV->getValueType());
       uint64_t FragmentOffsetInBits = Layout.getElementOffsetInBits(ElementIdx);
       transferSRADebugInfo(GV, NGV, FragmentOffsetInBits, Size,
                            STy->getNumElements());
-    } else if (SequentialType *STy = dyn_cast<SequentialType>(Ty)) {
+    } else {
       uint64_t EltSize = DL.getTypeAllocSize(ElTy);
       Align EltAlign(DL.getABITypeAlignment(ElTy));
       uint64_t FragmentSizeInBits = DL.getTypeAllocSizeInBits(ElTy);
 
       // Calculate the known alignment of the field.  If the original aggregate
       // had 256 byte alignment for example, something might depend on that:
       // propagate info to each field.
       Align NewAlign(MinAlign(StartAlignment, EltSize * ElementIdx));
       if (NewAlign > EltAlign)
         NGV->setAlignment(NewAlign);
       transferSRADebugInfo(GV, NGV, FragmentSizeInBits * ElementIdx,
-                           FragmentSizeInBits, STy->getNumElements());
+                           FragmentSizeInBits, GetSRASequentialNumElements(Ty));
     }
   }
 
   if (NewGlobals.empty())
     return nullptr;
 
   Module::GlobalListType &Globals = GV->getParent()->getGlobalList();
   for (auto NewGlobalVar : NewGlobals)
if (StructType *STy = dyn_cast<StructType>(Init->getType())) {
     assert(Idx < STy->getNumElements() && "Struct index out of range!");
     Elts[Idx] = EvaluateStoreInto(Elts[Idx], Val, Addr, OpNo+1);
 
     // Return the modified struct.
     return ConstantStruct::get(STy, Elts);
   }
 
   ConstantInt *CI = cast<ConstantInt>(Addr->getOperand(OpNo));
-  SequentialType *InitTy = cast<SequentialType>(Init->getType());
-  uint64_t NumElts = InitTy->getNumElements();
+  uint64_t NumElts;
+  if (ArrayType *ATy = dyn_cast<ArrayType>(Init->getType()))
+    NumElts = ATy->getNumElements();
+  else
+    NumElts = cast<VectorType>(Init->getType())->getNumElements();
 
   // Break up the array into elements.
   for (uint64_t i = 0, e = NumElts; i != e; ++i)
     Elts.push_back(Init->getAggregateElement(i));
 
   assert(CI->getZExtValue() < NumElts);
   Elts[CI->getZExtValue()] =
     EvaluateStoreInto(Elts[CI->getZExtValue()], Val, Addr, OpNo+1);
 
   if (Init->getType()->isArrayTy())
-    return ConstantArray::get(cast<ArrayType>(InitTy), Elts);
+    return ConstantArray::get(cast<ArrayType>(Init->getType()), Elts);
   return ConstantVector::get(Elts);
 }
 
 /// We have decided that Addr (which satisfies the predicate
 /// isSimpleEnoughPointerToCommit) should get Val as its value.  Make it happen.
 static void CommitValueTo(Constant *Val, Constant *Addr) {
   if (GlobalVariable *GV = dyn_cast<GlobalVariable>(Addr)) {
     assert(GV->hasInitializer());
if (Update) {
       if (CurrentGV == GV)
         return;
       // Need to clear and set up cache for new initializer.
       CurrentGV = GV;
       Elts.clear();
       unsigned NumElts;
       if (auto *STy = dyn_cast<StructType>(Ty))
         NumElts = STy->getNumElements();
+      else if (auto *ATy = dyn_cast<ArrayType>(Ty))
+        NumElts = ATy->getNumElements();
       else
-        NumElts = cast<SequentialType>(Ty)->getNumElements();
+        NumElts = cast<VectorType>(Ty)->getNumElements();
       for (unsigned i = 0, e = NumElts; i != e; ++i)
         Elts.push_back(Init->getAggregateElement(i));
     }
   };
 
   for (auto CEPair : SimpleCEs) {
     ConstantExpr *GEP = CEPair.first;
     Constant *Val = CEPair.second;

llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp

#endif
   // We use multiplication by 2**N instead of shift to cover the case of
   // multiplication by 0, which may occur in some elements of a vector operand.
   void handleMulByConstant(BinaryOperator &I, Constant *ConstArg,
                            Value *OtherArg) {
     Constant *ShadowMul;
     Type *Ty = ConstArg->getType();
     if (Ty->isVectorTy()) {
       unsigned NumElements = Ty->getVectorNumElements();
-      Type *EltTy = Ty->getSequentialElementType();
+      Type *EltTy = Ty->getVectorElementType();
       SmallVector<Constant *, 16> Elements;
       for (unsigned Idx = 0; Idx < NumElements; ++Idx) {
         if (ConstantInt *Elt =
                 dyn_cast<ConstantInt>(ConstArg->getAggregateElement(Idx))) {
           const APInt &V = Elt->getValue();
           APInt V2 = APInt(V.getBitWidth(), 1) << V.countTrailingZeros();
           Elements.push_back(ConstantInt::get(EltTy, V2));
         } else {

llvm/lib/Transforms/Scalar/SROA.cpp

 ///
 /// Note that this routine is very strict and tries to find a partition of the
 /// type which produces the *exact* right offset and size. It is not forgiving
 /// when the size or offset cause either end of type-based partition to be off.
 /// Also, this is a best-effort routine. It is reasonable to give up and not
 /// return a type if necessary.
 static Type *getTypePartition(const DataLayout &DL, Type *Ty, uint64_t Offset,
                               uint64_t Size) {
+  if (Ty->isVectorTy() && Ty->getVectorIsScalable())
+    return nullptr;
+
   if (Offset == 0 && DL.getTypeAllocSize(Ty) == Size)
     return stripAggregateTypeWrapping(DL, Ty);
   if (Offset > DL.getTypeAllocSize(Ty) ||
       (DL.getTypeAllocSize(Ty) - Offset) < Size)
     return nullptr;
 
-  if (SequentialType *SeqTy = dyn_cast<SequentialType>(Ty)) {
-    Type *ElementTy = SeqTy->getElementType();
+  if (isa<ArrayType>(Ty) || isa<VectorType>(Ty)) {
+    Type *ElementTy;
+    uint64_t TyNumElements;
+    if (auto *AT = dyn_cast<ArrayType>(Ty)) {
+      ElementTy = AT->getElementType();
+      TyNumElements = AT->getNumElements();
+    } else {
+      // FIXME: This isn't right for vectors with non-byte-sized or
+      // non-power-of-two sized elements.
+      auto *VT = cast<VectorType>(Ty);
+      ElementTy = VT->getElementType();
+      TyNumElements = VT->getNumElements();
+    }
     uint64_t ElementSize = DL.getTypeAllocSize(ElementTy);
     uint64_t NumSkippedElements = Offset / ElementSize;
-    if (NumSkippedElements >= SeqTy->getNumElements())
+    if (NumSkippedElements >= TyNumElements)
       return nullptr;
     Offset -= NumSkippedElements * ElementSize;
 
     // First check if we need to recurse.
     if (Offset > 0 || Size < ElementSize) {
       // Bail if the partition ends in a different array element.
       if ((Offset + Size) > ElementSize)
         return nullptr;

llvm/lib/Transforms/Utils/FunctionComparator.cpp

case Type::FunctionTyID: {
 
     for (unsigned i = 0, e = FTyL->getNumParams(); i != e; ++i) {
       if (int Res = cmpTypes(FTyL->getParamType(i), FTyR->getParamType(i)))
         return Res;
     }
     return 0;
   }
 
-  case Type::ArrayTyID:
-  case Type::VectorTyID: {
-    auto *STyL = cast<SequentialType>(TyL);
-    auto *STyR = cast<SequentialType>(TyR);
+  case Type::ArrayTyID: {
+    auto *STyL = cast<ArrayType>(TyL);
+    auto *STyR = cast<ArrayType>(TyR);
     if (STyL->getNumElements() != STyR->getNumElements())
       return cmpNumbers(STyL->getNumElements(), STyR->getNumElements());
     return cmpTypes(STyL->getElementType(), STyR->getElementType());
   }
+  case Type::VectorTyID: {
+    auto *STyL = cast<VectorType>(TyL);
+    auto *STyR = cast<VectorType>(TyR);
+    if (STyL->getElementCount().Scalable != STyR->getElementCount().Scalable)
+      return cmpNumbers(STyL->getElementCount().Scalable,
+                        STyR->getElementCount().Scalable);
+    if (STyL->getElementCount().Min != STyR->getElementCount().Min)
+      return cmpNumbers(STyL->getElementCount().Min,
+                        STyR->getElementCount().Min);
+    return cmpTypes(STyL->getElementType(), STyR->getElementType());
+  }
   }
 }
 
 // Determine whether the two operations are the same except that pointer-to-A
 // and pointer-to-B are equivalent. This should be kept in sync with
 // Instruction::isSameOperationAs.
 // Read method declaration comments for more details.
 int FunctionComparator::cmpOperations(const Instruction *L,

llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

default:
       return;
   }
 }
 
 unsigned BoUpSLP::canMapToVector(Type *T, const DataLayout &DL) const {
   unsigned N = 1;
   Type *EltTy = T;
 
-  while (isa<StructType>(EltTy) || isa<SequentialType>(EltTy)) {
+  while (isa<StructType>(EltTy) || isa<ArrayType>(EltTy) ||
+         isa<VectorType>(EltTy)) {
     if (auto *ST = dyn_cast<StructType>(EltTy)) {
       // Check that struct is homogeneous.
       for (const auto *Ty : ST->elements())
         if (Ty != *ST->element_begin())
           return 0;
       N *= ST->getNumElements();
       EltTy = *ST->element_begin();
+    } else if (auto *AT = dyn_cast<ArrayType>(EltTy)) {
+      N *= AT->getNumElements();
+      EltTy = AT->getElementType();
     } else {
-      auto *SeqT = cast<SequentialType>(EltTy);
-      N *= SeqT->getNumElements();
-      EltTy = SeqT->getElementType();
+      auto *VT = cast<VectorType>(EltTy);
+      N *= VT->getNumElements();
+      EltTy = VT->getElementType();
     }
   }
 
   if (!isValidElementType(EltTy))
     return 0;
   uint64_t VTSize = DL.getTypeStoreSizeInBits(VectorType::get(EltTy, N));
   if (VTSize < MinVecRegSize || VTSize > MaxVecRegSize || VTSize != DL.getTypeStoreSizeInBits(T))
     return 0;

mlir/lib/Target/LLVMIR/ModuleTranslation.cpp

buildSequentialConstant(ArrayRef<llvm::Constant *> &constants,
                         ArrayRef<int64_t> shape, llvm::Type *type,
                         Location loc) {
   if (shape.empty()) {
     llvm::Constant *result = constants.front();
     constants = constants.drop_front();
     return result;
   }
 
-  if (!isa<llvm::SequentialType>(type)) {
+  llvm::Type *elementType;
+  if (auto *arrayTy = dyn_cast<llvm::ArrayType>(type)) {
+    elementType = arrayTy->getElementType();
+  } else if (auto *vectorTy = dyn_cast<llvm::VectorType>(type)) {
+    elementType = vectorTy->getElementType();
+  } else {
     emitError(loc) << "expected sequential LLVM types wrapping a scalar";
     return nullptr;
   }
 
-  llvm::Type *elementType = type->getSequentialElementType();
   SmallVector<llvm::Constant *, 8> nested;
   nested.reserve(shape.front());
   for (int64_t i = 0; i < shape.front(); ++i) {
     nested.push_back(buildSequentialConstant(constants, shape.drop_front(),
                                              elementType, loc));
     if (!nested.back())
       return nullptr;
   }
 
   if (shape.size() == 1 && type->isVectorTy())
     return llvm::ConstantVector::get(nested);
   return llvm::ConstantArray::get(
       llvm::ArrayType::get(elementType, shape.front()), nested);
 }
 
 /// Returns the first non-sequential type nested in sequential types.
 static llvm::Type *getInnermostElementType(llvm::Type *type) {
-  while (isa<llvm::SequentialType>(type))
-    type = type->getSequentialElementType();
+  do {
+    if (auto *arrayTy = dyn_cast<llvm::ArrayType>(type)) {
+      type = arrayTy->getElementType();
+    } else if (auto *vectorTy = dyn_cast<llvm::VectorType>(type)) {
+      type = vectorTy->getElementType();
+    } else {
       return type;
     }
+  } while (1);
+}
 
 /// Create an LLVM IR constant of `llvmType` from the MLIR attribute `attr`.
 /// This currently supports integer, floating point, splat and dense element
 /// attributes and combinations thereof.  In case of error, report it to `loc`
 /// and return nullptr.
 llvm::Constant *ModuleTranslation::getLLVMConstant(llvm::Type *llvmType,
                                                    Attribute attr,
                                                    Location loc) {
   if (!attr)
     return llvm::UndefValue::get(llvmType);
   if (llvmType->isStructTy()) {
     emitError(loc, "struct types are not supported in constants");
     return nullptr;
   }
   if (auto intAttr = attr.dyn_cast<IntegerAttr>())
     return llvm::ConstantInt::get(llvmType, intAttr.getValue());
   if (auto boolAttr = attr.dyn_cast<BoolAttr>())
     return llvm::ConstantInt::get(llvmType, boolAttr.getValue());
   if (auto floatAttr = attr.dyn_cast<FloatAttr>())
     return llvm::ConstantFP::get(llvmType, floatAttr.getValue());
   if (auto funcAttr = attr.dyn_cast<FlatSymbolRefAttr>())
     return llvm::ConstantExpr::getBitCast(
         functionMapping.lookup(funcAttr.getValue()), llvmType);
   if (auto splatAttr = attr.dyn_cast<SplatElementsAttr>()) {
-    auto *sequentialType = cast<llvm::SequentialType>(llvmType);
-    auto elementType = sequentialType->getElementType();
-    uint64_t numElements = sequentialType->getNumElements();
+    llvm::Type *elementType;
+    uint64_t numElements;
+    if (auto *arrayTy = dyn_cast<llvm::ArrayType>(llvmType)) {
+      elementType = arrayTy->getElementType();
+      numElements = arrayTy->getNumElements();
+    } else {
+      elementType = llvmType->getVectorElementType();
+      numElements = llvmType->getVectorNumElements();
+    }
     // Splat value is a scalar. Extract it only if the element type is not
     // another sequence type. The recursion terminates because each step removes
     // one outer sequential type.
+    bool elementTypeSequential =
+        isa<llvm::ArrayType>(elementType) || isa<llvm::VectorType>(elementType);
     llvm::Constant *child = getLLVMConstant(
         elementType,
-        isa<llvm::SequentialType>(elementType) ? splatAttr
-                                               : splatAttr.getSplatValue(),
-        loc);
+        elementTypeSequential ? splatAttr : splatAttr.getSplatValue(), loc);
     if (!child)
       return nullptr;
     if (llvmType->isVectorTy())
       return llvm::ConstantVector::getSplat(
           llvm::ElementCount(numElements, /*Scalable=*/false), child);
     if (llvmType->isArrayTy()) {
       auto arrayType = llvm::ArrayType::get(elementType, numElements);
       SmallVector<llvm::Constant *, 8> constants(numElements, child);