Change the printing/parsing behavior for Attributes used in declarative assembly format

[mehdi_amini]

Summary

The new form of printing attribute in the declarative assembly is eliding the #dialect.mnemonic prefix to only keep the <....> part.

This is going along with https://llvm.discourse.group/t/rfc-elide-type-attribute-prefixes-when-using-declarative-assembly/4759

[rriddle]

mlir/include/mlir/IR/BuiltinAttributeInterfaces.h

22 ↗

(On Diff #387164)

Why is this necessary?

mlir/include/mlir/IR/BuiltinAttributeInterfaces.td

420 ↗

(On Diff #387164)

Unrelated?

mlir/include/mlir/IR/OpImplementation.h

85–91

Do we need to publicly expose this? Or can we move this to a protected/private section?

653

Can we drop the llvm:: here?

730–735

Can we drop all of the ::mlir:: here?

767–774

Same comment about dropping ::mlir::.

868

Can we drop the llvm::?

909–913

Same comment about ::mlir::.

mlir/lib/IR/BuiltinAttributes.cpp

66–72

While I think it would be nice to write these in terms of AsmParser/AsmPrinter at some point when we can, do we need to provide any of these given that we can use SFINAE to avoid accidentally calling these?

mlir/test/lib/Dialect/Test/TestTypes.cpp

305–306 ↗

(On Diff #387164)

Why do we have a static_assert for an attribute in TestTypes? Also, maybe add a comment here.

mlir/tools/mlir-tblgen/OpFormatGen.cpp

1371–1372

else if? Could also use TypeSwitch with a generic lambda here, but no preference.

2079–2082

Same here.

[mehdi_amini]

mlir/lib/IR/BuiltinAttributes.cpp

66–72

If you noticed, I am only providing the implementation: these classes already declare these methods!

So we already use SFINAE, but that leads to a linker error right now.

[rriddle]

mlir/lib/IR/BuiltinAttributes.cpp

66–72

Why do they declare these methods? Can we avoid that?

[mehdi_amini]

mlir/lib/IR/BuiltinAttributes.cpp

66–72

Right now I don't think we can, this is part of the contract for AttrDef. We could special case the builtin dialect, but frankly I'd rather move toward having less special case (like moving the handling for the builtin stuff outside of the parser instead.

mlir/test/lib/Dialect/Test/TestTypes.cpp

305–306 ↗

(On Diff #387164)

That's debugging leftover

[mehdi_amini]

Address comments

[mehdi_amini]

Remove more ::mlir:: qualifiers

[rriddle]

mlir/include/mlir/IR/OpImplementation.h

67–68

Can you not use the enable_if in the template? or as the return type? Using it as a last param always kind of makes me feel weird, because it could be provided accidentally.

767–769

910–912

mlir/lib/IR/BuiltinAttributeInterfaces.cpp

12 ↗

(On Diff #391882)

Why is this one necessary?

mlir/lib/Parser/AsmParserImpl.h

352–355

Is the top-check necessary? Looks like it's doing the same thing as the else branch here.

363–366

Same comment here.

mlir/tools/mlir-tblgen/OpFormatGen.cpp

2087–2094

Is there a downside to always calling the stripped version? (if it isn't a derived type we will revert to the fallback anyways.

[mehdi_amini]

mlir/lib/IR/BuiltinAttributes.cpp

66–72

Jeff's recent patch fixed this: they are all gone now!

mlir/lib/Parser/AsmParserImpl.h

352–355

Not really, it is calling the callback while the else calls the generic parser

mlir/tools/mlir-tblgen/OpFormatGen.cpp

2087–2094

We could call the _odsPrinter.printStrippedAttrOrType in the else branch, but we have an mlir::Type so the SFINAE for printStrippedAttrOrType will resolve to the implementation that is:

template <typename AttrOrType,
    std::enable_if_t<!detect_has_print_method<AttrOrType>::value> *sfinae =
        nullptr>
void printStrippedAttrOrType(
    AttrOrType attrOrType) {
  *this << attrOrType;
}

so inline it here seems more clear to me.

[mehdi_amini]

Address comments

[mehdi_amini]

Remove unused include

mlir/include/mlir/Dialect/ArmSVE/ArmSVE.td

def ArmSVE_Dialect : Dialect {
   let name = "arm_sve";
   let cppNamespace = "::mlir::arm_sve";
   let summary = "Basic dialect to target Arm SVE architectures";
   let description = [{
     This dialect contains the definitions necessary to target Arm SVE scalable
     vector operations, including a scalable vector type and intrinsics for
     some Arm SVE instructions.
   }];
+  let useDefaultTypePrinterParser = 1;
 }
 
 //===----------------------------------------------------------------------===//
 // ArmSVE type definitions
 //===----------------------------------------------------------------------===//
 
 def ArmSVE_ScalableVectorType : DialectType<ArmSVE_Dialect,
     CPred<"$_self.isa<ScalableVectorType>()">,
let description = [{
     run time.
   }];
 
   let parameters = (ins
     ArrayRefParameter<"int64_t", "Vector shape">:$shape,
     "Type":$elementType
   );
 
-  let printer = [{
-    $_printer << "<";
-    for (int64_t dim : getShape())
-      $_printer << dim << 'x';
-    $_printer << getElementType() << '>';
-  }];
-
-  let parser = [{
-    VectorType vector;
-    if ($_parser.parseType(vector))
-      return Type();
-    return get($_ctxt, vector.getShape(), vector.getElementType());
-  }];
-
   let extraClassDeclaration = [{
     bool hasStaticShape() const {
       return llvm::none_of(getShape(), ShapedType::isDynamic);
     }
     int64_t getNumElements() const {
       assert(hasStaticShape() &&
              "cannot get element count of dynamic shaped type");
       ArrayRef<int64_t> shape = getShape();

mlir/include/mlir/IR/DialectImplementation.h

 
 /// Parse an attribute.
 template <typename AttributeT>
 struct FieldParser<
     AttributeT, std::enable_if_t<std::is_base_of<Attribute, AttributeT>::value,
                                  AttributeT>> {
   static FailureOr<AttributeT> parse(AsmParser &parser) {
     AttributeT value;
-    if (parser.parseAttribute(value))
+    if (parser.parseCustomAttributeWithFallback(value))
       return failure();
     return value;
   }
 };
 
-/// Parse a type.
+/// Parse an attribute.
 template <typename TypeT>
 struct FieldParser<
     TypeT, std::enable_if_t<std::is_base_of<Type, TypeT>::value, TypeT>> {
   static FailureOr<TypeT> parse(AsmParser &parser) {
     TypeT value;
-    if (parser.parseType(value))
+    if (parser.parseCustomTypeWithFallback(value))
       return failure();
     return value;
   }
 };
 
 /// Parse any integer.
 template <typename IntT>
 struct FieldParser<IntT,

mlir/include/mlir/IR/OpBase.td

 }
 
 // Define a new type, named `name`, belonging to `dialect` that inherits from
 // the given C++ base class.
 class TypeDef<Dialect dialect, string name, list<Trait> traits = [],
               string baseCppClass = "::mlir::Type">
     : DialectType<dialect, CPred<"">, /*descr*/"", name # "Type">,
       AttrOrTypeDef<"Type", name, traits, baseCppClass> {
+  // Make it possible to use such type as parameters for other types.
+  string cppType = dialect.cppNamespace # "::" # cppClassName;
+
   // A constant builder provided when the type has no parameters.
   let builderCall = !if(!empty(parameters),
                            "$_builder.getType<" # dialect.cppNamespace #
                                "::" # cppClassName # ">()",
                            "");
   // The predicate for when this def is used as a constraint.
   let predicate = CPred<"$_self.isa<" # dialect.cppNamespace #
                                  "::" # cppClassName # ">()">;

mlir/include/mlir/IR/OpImplementation.h

public:
   /// Print the given floating point value in a stabilized form that can be
   /// roundtripped through the IR. This is the companion to the 'parseFloat'
   /// hook on the AsmParser.
   virtual void printFloat(const APFloat &value);
 
   virtual void printType(Type type);
   virtual void printAttribute(Attribute attr);
 
+  /// Trait to check if `AttrType` provides a `print` method.
+  template <typename AttrOrType>
+  using has_print_method =
+      decltype(std::declval<AttrOrType>().print(std::declval<AsmPrinter &>()));
+  template <typename AttrOrType>
+  using detect_has_print_method =
+      llvm::is_detected<has_print_method, AttrOrType>;
+
+  /// Print the provided attribute in the context of an operation custom
+  /// printer/parser: this will invoke directly the print method on the
+  /// attribute class and skip the `#dialect.mnemonic` prefix in most cases.
+  template <typename AttrOrType,
+            std::enable_if_t<detect_has_print_method<AttrOrType>::value>
+                *sfinae = nullptr>
+  void printStrippedAttrOrType(AttrOrType attrOrType) {
+    if (succeeded(printAlias(attrOrType)))
+      return;
+    attrOrType.print(*this);
+  }
+
+  /// SFINAE for printing the provided attribute in the context of an operation
+  /// custom printer in the case where the attribute does not define a print
+  /// method.
+  template <typename AttrOrType,
+            std::enable_if_t<!detect_has_print_method<AttrOrType>::value>
+                *sfinae = nullptr>
+  void printStrippedAttrOrType(AttrOrType attrOrType) {
+    *this << attrOrType;
+  }
+
   /// Print the given attribute without its type. The corresponding parser must
   /// provide a valid type for the attribute.
   virtual void printAttributeWithoutType(Attribute attr);
 
   /// Print the given string as a keyword, or a quoted and escaped string if it
   /// has any special or non-printable characters in it.
   virtual void printKeywordOrString(StringRef keyword);
 
   /// Print the given string as a symbol reference, i.e. a form representable by
   /// a SymbolRefAttr. A symbol reference is represented as a string prefixed
   /// with '@'. The reference is surrounded with ""'s and escaped if it has any
   /// special or non-printable characters in it.
   virtual void printSymbolName(StringRef symbolRef);
 
   /// Print an optional arrow followed by a type list.
   template <typename TypeRange>
   void printOptionalArrowTypeList(TypeRange &&types) {
protected:
   /// Initialize the printer with no internal implementation. In this case, all
   /// virtual methods of this class must be overriden.
   AsmPrinter() : impl(nullptr) {}
 
 private:
   AsmPrinter(const AsmPrinter &) = delete;
   void operator=(const AsmPrinter &) = delete;
 
+  /// Print the alias for the given attribute, return failure if no alias could
+  /// be printed.
+  virtual LogicalResult printAlias(Attribute attr);
+
+  /// Print the alias for the given type, return failure if no alias could
+  /// be printed.
+  virtual LogicalResult printAlias(Type type);
+
   /// The internal implementation of the printer.
   Impl *impl;
 };
 
 template <typename AsmPrinterT>
 inline std::enable_if_t<std::is_base_of<AsmPrinter, AsmPrinterT>::value,
                         AsmPrinterT &>
 operator<<(AsmPrinterT &p, Type type) {
public:
 
   //===--------------------------------------------------------------------===//
   // Attribute Parsing
   //===--------------------------------------------------------------------===//
 
   /// Parse an arbitrary attribute of a given type and return it in result.
   virtual ParseResult parseAttribute(Attribute &result, Type type = {}) = 0;
 
+  /// Parse a custom attribute with the provided callback, unless the next
+  /// token is `#`, in which case the generic parser is invoked.
+  virtual ParseResult parseCustomAttributeWithFallback(
+      Attribute &result, Type type,
+      function_ref<ParseResult(Attribute &result, Type type)>
+          parseAttribute) = 0;
+
   /// Parse an attribute of a specific kind and type.
   template <typename AttrType>
   ParseResult parseAttribute(AttrType &result, Type type = {}) {
     llvm::SMLoc loc = getCurrentLocation();
 
     // Parse any kind of attribute.
     Attribute attr;
     if (parseAttribute(attr, type))
public:
 
   /// Parse an attribute of a specific kind and type.
   template <typename AttrType>
   ParseResult parseAttribute(AttrType &result, StringRef attrName,
                              NamedAttrList &attrs) {
     return parseAttribute(result, Type(), attrName, attrs);
   }
 
-  /// Parse an arbitrary attribute of a given type and return it in result. This
-  /// also adds the attribute to the specified attribute list with the specified
-  /// name.
+  /// Parse an arbitrary attribute of a given type and populate it in `result`.
+  /// This also adds the attribute to the specified attribute list with the
+  /// specified name.
   template <typename AttrType>
   ParseResult parseAttribute(AttrType &result, Type type, StringRef attrName,
                              NamedAttrList &attrs) {
     llvm::SMLoc loc = getCurrentLocation();
 
     // Parse any kind of attribute.
     Attribute attr;
     if (parseAttribute(attr, type))
       return failure();
 
     // Check for the right kind of attribute.
     result = attr.dyn_cast<AttrType>();
     if (!result)
       return emitError(loc, "invalid kind of attribute specified");
 
     attrs.append(attrName, result);
     return success();
   }
 
+  /// Trait to check if `AttrType` provides a `parse` method.
+  template <typename AttrType>
+  using has_parse_method = decltype(AttrType::parse(std::declval<AsmParser &>(),
+                                                    std::declval<Type>()));
+  template <typename AttrType>
+  using detect_has_parse_method = llvm::is_detected<has_parse_method, AttrType>;
+
+  /// Parse a custom attribute of a given type unless the next token is `#`, in
+  /// which case the generic parser is invoked. The parsed attribute is
+  /// populated in `result` and also added to the specified attribute list with
+  /// the specified name.
+  template <typename AttrType>
+  std::enable_if_t<detect_has_parse_method<AttrType>::value, ParseResult>
+  parseCustomAttributeWithFallback(AttrType &result, Type type,
+                                   StringRef attrName, NamedAttrList &attrs) {
+    llvm::SMLoc loc = getCurrentLocation();
+
+    // Parse any kind of attribute.
+    Attribute attr;
+    if (parseCustomAttributeWithFallback(
+            attr, type, [&](Attribute &result, Type type) -> ParseResult {
+              result = AttrType::parse(*this, type);
+              if (!result)
+                return failure();
+              return success();
+            }))
+      return failure();
+
+    // Check for the right kind of attribute.
+    result = attr.dyn_cast<AttrType>();
+    if (!result)
+      return emitError(loc, "invalid kind of attribute specified");
+
+    attrs.append(attrName, result);
+    return success();
+  }
+
+  /// SFINAE parsing method for Attribute that don't implement a parse method.
+  template <typename AttrType>
+  std::enable_if_t<!detect_has_parse_method<AttrType>::value, ParseResult>
+  parseCustomAttributeWithFallback(AttrType &result, Type type,
+                                   StringRef attrName, NamedAttrList &attrs) {
+    return parseAttribute(result, type, attrName, attrs);
+  }
+
+  /// Parse a custom attribute of a given type unless the next token is `#`, in
+  /// which case the generic parser is invoked. The parsed attribute is
+  /// populated in `result`.
+  template <typename AttrType>
+  std::enable_if_t<detect_has_parse_method<AttrType>::value, ParseResult>
+  parseCustomAttributeWithFallback(AttrType &result) {
+    llvm::SMLoc loc = getCurrentLocation();
+
+    // Parse any kind of attribute.
+    Attribute attr;
+    if (parseCustomAttributeWithFallback(
+            attr, {}, [&](Attribute &result, Type type) -> ParseResult {
+              result = AttrType::parse(*this, type);
+              return success(!!result);
+            }))
+      return failure();
+
+    // Check for the right kind of attribute.
+    result = attr.dyn_cast<AttrType>();
+    if (!result)
+      return emitError(loc, "invalid kind of attribute specified");
+    return success();
+  }
+
+  /// SFINAE parsing method for Attribute that don't implement a parse method.
+  template <typename AttrType>
+  std::enable_if_t<!detect_has_parse_method<AttrType>::value, ParseResult>
+  parseCustomAttributeWithFallback(AttrType &result) {
+    return parseAttribute(result);
+  }
+
   /// Parse an arbitrary optional attribute of a given type and return it in
   /// result.
   virtual OptionalParseResult parseOptionalAttribute(Attribute &result,
                                                      Type type = {}) = 0;
 
   /// Parse an optional array attribute and return it in result.
   virtual OptionalParseResult parseOptionalAttribute(ArrayAttr &result,
                                                      Type type = {}) = 0;
public:
 
   //===--------------------------------------------------------------------===//
   // Type Parsing
   //===--------------------------------------------------------------------===//
 
   /// Parse a type.
   virtual ParseResult parseType(Type &result) = 0;
 
+  /// Parse a custom type with the provided callback, unless the next
+  /// token is `#`, in which case the generic parser is invoked.
+  virtual ParseResult parseCustomTypeWithFallback(
+      Type &result, function_ref<ParseResult(Type &result)> parseType) = 0;
+
   /// Parse an optional type.
   virtual OptionalParseResult parseOptionalType(Type &result) = 0;
 
   /// Parse a type of a specific type.
   template <typename TypeT>
   ParseResult parseType(TypeT &result) {
     llvm::SMLoc loc = getCurrentLocation();
 
     // Parse any kind of type.
     Type type;
     if (parseType(type))
       return failure();
 
-    // Check for the right kind of attribute.
+    // Check for the right kind of type.
     result = type.dyn_cast<TypeT>();
     if (!result)
       return emitError(loc, "invalid kind of type specified");
 
     return success();
   }
 
+  /// Trait to check if `TypeT` provides a `parse` method.
+  template <typename TypeT>
+  using type_has_parse_method =
+      decltype(TypeT::parse(std::declval<AsmParser &>()));
+  template <typename TypeT>
+  using detect_type_has_parse_method =
+      llvm::is_detected<type_has_parse_method, TypeT>;
+
+  /// Parse a custom Type of a given type unless the next token is `#`, in
+  /// which case the generic parser is invoked. The parsed Type is
+  /// populated in `result`.
+  template <typename TypeT>
+  std::enable_if_t<detect_type_has_parse_method<TypeT>::value, ParseResult>
+  parseCustomTypeWithFallback(TypeT &result) {
+    llvm::SMLoc loc = getCurrentLocation();
+
+    // Parse any kind of Type.
+    Type type;
+    if (parseCustomTypeWithFallback(type, [&](Type &result) -> ParseResult {
+          result = TypeT::parse(*this);
+          return success(!!result);
+        }))
+      return failure();
+
+    // Check for the right kind of Type.
+    result = type.dyn_cast<TypeT>();
+    if (!result)
+      return emitError(loc, "invalid kind of Type specified");
+    return success();
+  }
+
+  /// SFINAE parsing method for Type that don't implement a parse method.
+  template <typename TypeT>
+  std::enable_if_t<!detect_type_has_parse_method<TypeT>::value, ParseResult>
+  parseCustomTypeWithFallback(TypeT &result) {
+    return parseType(result);
+  }
+
   /// Parse a type list.
   ParseResult parseTypeList(SmallVectorImpl<Type> &result) {
     do {
       Type type;
       if (parseType(type))
         return failure();
       result.push_back(type);
     } while (succeeded(parseOptionalComma()));
public:
   ParseResult parseColonType(TypeType &result) {
     llvm::SMLoc loc = getCurrentLocation();
 
     // Parse any kind of type.
     Type type;
     if (parseColonType(type))
       return failure();
 
-    // Check for the right kind of attribute.
+    // Check for the right kind of type.
     result = type.dyn_cast<TypeType>();
     if (!result)
       return emitError(loc, "invalid kind of type specified");
 
     return success();
   }
 
   /// Parse a colon followed by a type list, which must have at least one type.

mlir/lib/Dialect/ArmSVE/IR/ArmSVEDialect.cpp

 #include "mlir/Dialect/ArmSVE/ArmSVETypes.cpp.inc"
       >();
 }
 
 //===----------------------------------------------------------------------===//
 // ScalableVectorType
 //===----------------------------------------------------------------------===//
 
-Type ArmSVEDialect::parseType(DialectAsmParser &parser) const {
-  llvm::SMLoc typeLoc = parser.getCurrentLocation();
-  {
-    Type genType;
-    auto parseResult = generatedTypeParser(parser, "vector", genType);
-    if (parseResult.hasValue())
-      return genType;
-  }
-  parser.emitError(typeLoc, "unknown type in ArmSVE dialect");
-  return Type();
+void ScalableVectorType::print(AsmPrinter &printer) const {
+  printer << "<";
+  for (int64_t dim : getShape())
+    printer << dim << 'x';
+  printer << getElementType() << '>';
 }
 
-void ArmSVEDialect::printType(Type type, DialectAsmPrinter &os) const {
-  if (failed(generatedTypePrinter(type, os)))
-    llvm_unreachable("unexpected 'arm_sve' type kind");
+Type ScalableVectorType::parse(AsmParser &parser) {
+  SmallVector<int64_t> dims;
+  Type eltType;
+  if (parser.parseLess() ||
+      parser.parseDimensionList(dims, /*allowDynamic=*/false) ||
+      parser.parseType(eltType) || parser.parseGreater())
+    return {};
+  return ScalableVectorType::get(eltType.getContext(), dims, eltType);
 }
 
 //===----------------------------------------------------------------------===//
 // ScalableVector versions of general helpers for comparison ops
 //===----------------------------------------------------------------------===//
 
 // Return the scalable vector of the same shape and containing i1.
 static Type getI1SameShape(Type type) {

mlir/lib/Dialect/Vector/VectorOps.cpp

static constexpr const CombiningKind combiningKindsList[] = {
     CombiningKind::MAXF,
     CombiningKind::AND,
     CombiningKind::OR,
     CombiningKind::XOR,
     // clang-format on
 };
 
 void CombiningKindAttr::print(AsmPrinter &printer) const {
-  printer << "kind<";
+  printer << "<";
   auto kinds = llvm::make_filter_range(combiningKindsList, [&](auto kind) {
     return bitEnumContains(this->getKind(), kind);
   });
   llvm::interleaveComma(kinds, printer,
                         [&](auto kind) { printer << stringifyEnum(kind); });
   printer << ">";
 }
 
if (attrKind == "kind")
     return CombiningKindAttr::parse(parser, {});
 
   parser.emitError(parser.getNameLoc(), "Unknown attribute type: ") << attrKind;
   return {};
 }
 
 void VectorDialect::printAttribute(Attribute attr,
                                    DialectAsmPrinter &os) const {
-  if (auto ck = attr.dyn_cast<CombiningKindAttr>())
+  if (auto ck = attr.dyn_cast<CombiningKindAttr>()) {
+    os << "kind";
     ck.print(os);
-  else
-    llvm_unreachable("Unknown attribute type");
+    return;
+  }
 }
 
 //===----------------------------------------------------------------------===//
 // VectorDialect
 //===----------------------------------------------------------------------===//
 
 void VectorDialect::initialize() {
   addAttributes<CombiningKindAttr>();

mlir/lib/Dialect/Vector/VectorTransforms.cpp

static void incIdx(SmallVector<int64_t, 4> &idx, VectorType tp, int64_t r) {
     }
   }
 };
 
 /// Convert MulIOp/MulFOp + MultiDimReductionOp<add> into ContractionOp.
 /// Ex:
 /// ```
 ///   %0 = arith.mulf %arg0, %arg1 : vector<8x32x16xf32>
-///   %1 = vector.multi_reduction #vector.kind<add>, %0 [1]
+///   %1 = vector.multi_reduction add, %0 [1]
 ///     : vector<8x32x16xf32> to vector<8x16xf32>
 /// ```
 /// Gets converted to:
 /// ```
 ///   %1 = vector.contract {indexing_maps = [
 ///         affine_map<(d0, d1, d2) -> (d0, d1, d2)>,
 ///         affine_map<(d0, d1, d2) -> (d0, d1, d2)>,
 ///         affine_map<(d0, d1, d2) -> (d0, d1)>],
 ///    iterator_types = ["parallel", "parallel", "reduction"],
-///    kind = #vector.kind<add>} %0, %arg1, %cst_f0
+///    kind = add} %0, %arg1, %cst_f0
 ///    : vector<8x32x16xf32>, vector<8x32x16xf32> into vector<8x32xf32>
 ///  ```
 struct MultiReduceToContract
     : public OpRewritePattern<vector::MultiDimReductionOp> {
   using OpRewritePattern<vector::MultiDimReductionOp>::OpRewritePattern;
 
   LogicalResult matchAndRewrite(vector::MultiDimReductionOp reduceOp,
                                 PatternRewriter &rewriter) const override {

 /// ```
 ///   %0 = vector.transpose %arg0, [2, 0, 1]
 ///     : vector<32x16x8xf32> to vector<8x32x16xf32>
 ///   %1 = vector.contract {indexing_maps = [
 ///         affine_map<(d0, d1, d2) -> (d0, d1, d2)>,
 ///         affine_map<(d0, d1, d2) -> (d0, d1, d2)>,
 ///         affine_map<(d0, d1, d2) -> (d0, d1)>],
 ///    iterator_types = ["parallel", "parallel", "reduction"],
-///    kind = #vector.kind<add>} %0, %arg1, %cst_f0
+///    kind = add} %0, %arg1, %cst_f0
 ///    : vector<8x32x16xf32>, vector<8x32x16xf32> into vector<8x32xf32>
 /// ```
 /// Gets converted to:
 /// ```
 ///   %1 = vector.contract {indexing_maps = [
 ///         affine_map<(d0, d1, d2) -> (d1, d2, d0)>,
 ///         affine_map<(d0, d1, d2) -> (d0, d1, d2)>,
 ///         affine_map<(d0, d1, d2) -> (d0, d1)>],
 ///    iterator_types = ["parallel", "parallel", "reduction"],
-///    kind = #vector.kind<add>} %arg0, %arg1, %cst_f0
+///    kind = add} %arg0, %arg1, %cst_f0
 ///    : vector<8x32x16xf32>, vector<8x32x16xf32> into vector<8x32xf32>
 ///  ```
 struct CombineContractTranspose
     : public OpRewritePattern<vector::ContractionOp> {
   using OpRewritePattern<vector::ContractionOp>::OpRewritePattern;
 
   LogicalResult matchAndRewrite(vector::ContractionOp contractOp,
                                 PatternRewriter &rewriter) const override {

 /// Ex:
 /// ```
 ///   %0 = vector.broadcast %arg0 : vector<32x16xf32> to vector<8x32x16xf32>
 ///   %1 = vector.contract {indexing_maps = [
 ///         affine_map<(d0, d1, d2) -> (d0, d1, d2)>,
 ///         affine_map<(d0, d1, d2) -> (d0, d1, d2)>,
 ///         affine_map<(d0, d1, d2) -> (d0, d1)>],
 ///    iterator_types = ["parallel", "parallel", "reduction"],
-///    kind = #vector.kind<add>} %0, %arg1, %cst_f0
+///    kind = add} %0, %arg1, %cst_f0
 ///    : vector<8x32x16xf32>, vector<8x32x16xf32> into vector<8x32xf32>
 /// ```
 /// Gets converted to:
 /// ```
 ///   %1 = vector.contract {indexing_maps = [
 ///         affine_map<(d0, d1, d2) -> (d1, d2)>,
 ///         affine_map<(d0, d1, d2) -> (d0, d1, d2)>,
 ///         affine_map<(d0, d1, d2) -> (d0, d1)>],
 ///    iterator_types = ["parallel", "parallel", "reduction"],
-///    kind = #vector.kind<add>} %arg0, %arg1, %cst_f0
+///    kind = add} %arg0, %arg1, %cst_f0
 ///    : vector<32x16xf32>, vector<8x32x16xf32> into vector<8x32xf32>
 ///  ```
 struct CombineContractBroadcast
     : public OpRewritePattern<vector::ContractionOp> {
   using OpRewritePattern<vector::ContractionOp>::OpRewritePattern;
 
   LogicalResult matchAndRewrite(vector::ContractionOp contractOp,
                                 PatternRewriter &rewriter) const override {

mlir/lib/IR/AsmPrinter.cpp

private:
   /// Consider the given type to be printed for an alias.
   void printType(Type type) override { initializer.visit(type); }
 
   /// Consider the given attribute to be printed for an alias.
   void printAttribute(Attribute attr) override { initializer.visit(attr); }
   void printAttributeWithoutType(Attribute attr) override {
     printAttribute(attr);
   }
+  LogicalResult printAlias(Attribute attr) override {
+    initializer.visit(attr);
+    return success();
+  }
+  LogicalResult printAlias(Type type) override {
+    initializer.visit(type);
+    return success();
+  }
 
   /// Print the given set of attributes with names not included within
   /// 'elidedAttrs'.
   void printOptionalAttrDict(ArrayRef<NamedAttribute> attrs,
                              ArrayRef<StringRef> elidedAttrs = {}) override {
     if (attrs.empty())
       return;
     if (elidedAttrs.empty()) {
enum class AttrTypeElision {
     /// The type must be elided.
     Must
   };
 
   /// Print the given attribute.
   void printAttribute(Attribute attr,
                       AttrTypeElision typeElision = AttrTypeElision::Never);
 
+  /// Print the alias for the given attribute, return failure if no alias could
+  /// be printed.
+  LogicalResult printAlias(Attribute attr);
+
   void printType(Type type);
 
+  /// Print the alias for the given type, return failure if no alias could
+  /// be printed.
+  LogicalResult printAlias(Type type);
+
   /// Print the given location to the stream. If `allowAlias` is true, this
   /// allows for the internal location to use an attribute alias.
   void printLocation(LocationAttr loc, bool allowAlias = false);
 
   void printAffineMap(AffineMap map);
   void
   printAffineExpr(AffineExpr expr,
                   function_ref<void(unsigned, bool)> printValueName = nullptr);

 //
 // Unfortunately, neither of the strings of an opaque ElementsAttr literal will
 // accept the string "elided". The first string must be a registered dialect
 // name and the latter must be a hex constant.
 static void printElidedElementsAttr(raw_ostream &os) {
   os << R"(opaque<"_", "0xDEADBEEF">)";
 }
 
+LogicalResult AsmPrinter::Impl::printAlias(Attribute attr) {
+  return success(state && succeeded(state->getAliasState().getAlias(attr, os)));
+}
+
+LogicalResult AsmPrinter::Impl::printAlias(Type type) {
+  return success(state && succeeded(state->getAliasState().getAlias(type, os)));
+}
+
 void AsmPrinter::Impl::printAttribute(Attribute attr,
                                       AttrTypeElision typeElision) {
   if (!attr) {
     os << "<<NULL ATTRIBUTE>>";
     return;
   }
 
   // Try to print an alias for this attribute.
-  if (state && succeeded(state->getAliasState().getAlias(attr, os)))
+  if (succeeded(printAlias(attr)))
     return;
 
   if (!isa<BuiltinDialect>(attr.getDialect()))
     return printDialectAttribute(attr);
 
   auto attrType = attr.getType();
   if (auto opaqueAttr = attr.dyn_cast<OpaqueAttr>()) {
     printDialectSymbol(os, "#", opaqueAttr.getDialectNamespace(),
void AsmPrinter::printType(Type type) {
   impl->printType(type);
 }
 
 void AsmPrinter::printAttribute(Attribute attr) {
   assert(impl && "expected AsmPrinter::printAttribute to be overriden");
   impl->printAttribute(attr);
 }
 
+LogicalResult AsmPrinter::printAlias(Attribute attr) {
+  assert(impl && "expected AsmPrinter::printAlias to be overriden");
+  return impl->printAlias(attr);
+}
+
+LogicalResult AsmPrinter::printAlias(Type type) {
+  assert(impl && "expected AsmPrinter::printAlias to be overriden");
+  return impl->printAlias(type);
+}
+
 void AsmPrinter::printAttributeWithoutType(Attribute attr) {
   assert(impl &&
          "expected AsmPrinter::printAttributeWithoutType to be overriden");
   impl->printAttribute(attr, Impl::AttrTypeElision::Must);
 }
 
 void AsmPrinter::printKeywordOrString(StringRef keyword) {
   assert(impl && "expected AsmPrinter::printKeywordOrString to be overriden");

mlir/lib/IR/BuiltinAttributes.cpp

 //===- BuiltinAttributes.cpp - MLIR Builtin Attribute Classes -------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #include "mlir/IR/BuiltinAttributes.h"
 #include "AttributeDetail.h"
 #include "mlir/IR/AffineMap.h"
 #include "mlir/IR/BuiltinDialect.h"
 #include "mlir/IR/Dialect.h"
 #include "mlir/IR/IntegerSet.h"
+#include "mlir/IR/OpImplementation.h"
 #include "mlir/IR/Operation.h"
 #include "mlir/IR/SymbolTable.h"
 #include "mlir/IR/Types.h"
 #include "mlir/Interfaces/DecodeAttributesInterfaces.h"
 #include "llvm/ADT/APSInt.h"
 #include "llvm/ADT/Sequence.h"
 #include "llvm/Support/Endian.h"
 
SubElementAttrInterface ArrayAttr::replaceImmediateSubAttribute(
     ArrayRef<std::pair<size_t, Attribute>> replacements) const {
   std::vector<Attribute> vector = getValue().vec();
   for (auto &it : replacements) {
     vector[it.first] = it.second;
   }
   return get(getContext(), vector);
 }
 
 //===----------------------------------------------------------------------===//
 // DictionaryAttr
 //===----------------------------------------------------------------------===//
 
 /// Helper function that does either an in place sort or sorts from source array
 /// into destination. If inPlace then storage is both the source and the
 /// destination, else value is the source and storage destination. Returns
 /// whether source was sorted.
 template <bool inPlace>
 static bool dictionaryAttrSort(ArrayRef<NamedAttribute> value,
                                SmallVectorImpl<NamedAttribute> &storage) {
   // Specialize for the common case.
   switch (value.size()) {
   case 0:
     // Zero already sorted.

 
 BoolAttr IntegerAttr::getBoolAttrUnchecked(IntegerType type, bool value) {
   auto attr = Base::get(type.getContext(), type, APInt(/*numBits=*/1, value));
   return attr.cast<BoolAttr>();
 }
 
 //===----------------------------------------------------------------------===//
 // BoolAttr
+//===----------------------------------------------------------------------===//
 
 bool BoolAttr::getValue() const {
   auto *storage = reinterpret_cast<IntegerAttrStorage *>(impl);
   return storage->value.getBoolValue();
 }
 
 bool BoolAttr::classof(Attribute attr) {
   IntegerAttr intAttr = attr.dyn_cast<IntegerAttr>();

mlir/lib/IR/BuiltinTypes.cpp

 //===- BuiltinTypes.cpp - MLIR Builtin Type Classes -----------------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #include "mlir/IR/BuiltinTypes.h"
 #include "TypeDetail.h"
 #include "mlir/IR/AffineExpr.h"
 #include "mlir/IR/AffineMap.h"
 #include "mlir/IR/BuiltinAttributes.h"
 #include "mlir/IR/BuiltinDialect.h"
 #include "mlir/IR/Diagnostics.h"
 #include "mlir/IR/Dialect.h"
+#include "mlir/IR/OpImplementation.h"
 #include "mlir/IR/TensorEncoding.h"
 #include "llvm/ADT/APFloat.h"
 #include "llvm/ADT/BitVector.h"
 #include "llvm/ADT/Sequence.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/ADT/TypeSwitch.h"
 
 using namespace mlir;
bool mlir::detail::isSupportedMemorySpace(Attribute memorySpace) {
   if (!memorySpace)
     return true;
 
   // Supported built-in attributes.
   if (memorySpace.isa<IntegerAttr, StringAttr, DictionaryAttr>())
     return true;
 
   // Allow custom dialect attributes.
-  if (!::mlir::isa<BuiltinDialect>(memorySpace.getDialect()))
+  if (!isa<BuiltinDialect>(memorySpace.getDialect()))
     return true;
 
   return false;
 }
 
 Attribute mlir::detail::wrapIntegerMemorySpace(unsigned memorySpace,
                                                MLIRContext *ctx) {
   if (memorySpace == 0)

mlir/lib/Parser/AsmParserImpl.h

public:
   //===--------------------------------------------------------------------===//
 
   /// Parse an arbitrary attribute and return it in result.
   ParseResult parseAttribute(Attribute &result, Type type) override {
     result = parser.parseAttribute(type);
     return success(static_cast<bool>(result));
   }
 
+  /// Parse a custom attribute with the provided callback, unless the next
+  /// token is `#`, in which case the generic parser is invoked.
+  ParseResult parseCustomAttributeWithFallback(
+      Attribute &result, Type type,
+      function_ref<ParseResult(Attribute &result, Type type)> parseAttribute)
+      override {
+    if (parser.getToken().isNot(Token::hash_identifier))
+      return parseAttribute(result, type);
+    result = parser.parseAttribute(type);
+    return success(static_cast<bool>(result));
+  }
+
+  /// Parse a custom attribute with the provided callback, unless the next
+  /// token is `#`, in which case the generic parser is invoked.
+  ParseResult parseCustomTypeWithFallback(
+      Type &result,
+      function_ref<ParseResult(Type &result)> parseType) override {
+    if (parser.getToken().isNot(Token::exclamation_identifier))
+      return parseType(result);
+    result = parser.parseType();
+    return success(static_cast<bool>(result));
+  }
+
   OptionalParseResult parseOptionalAttribute(Attribute &result,
                                              Type type) override {
     return parser.parseOptionalAttribute(result, type);
   }
   OptionalParseResult parseOptionalAttribute(ArrayAttr &result,
                                              Type type) override {
     return parser.parseOptionalAttribute(result, type);
   }

mlir/test/Dialect/ArmSVE/roundtrip.mlir

 // RUN: mlir-opt -verify-diagnostics %s | mlir-opt | FileCheck %s
 
 func @arm_sve_sdot(%a: !arm_sve.vector<16xi8>,
                    %b: !arm_sve.vector<16xi8>,
                    %c: !arm_sve.vector<4xi32>) -> !arm_sve.vector<4xi32> {
-  // CHECK: arm_sve.sdot {{.*}}: !arm_sve.vector<16xi8> to !arm_sve.vector<4xi32
+  // CHECK: arm_sve.sdot {{.*}}: <16xi8> to <4xi32
   %0 = arm_sve.sdot %c, %a, %b :
              !arm_sve.vector<16xi8> to !arm_sve.vector<4xi32>
   return %0 : !arm_sve.vector<4xi32>
 }
 
 func @arm_sve_smmla(%a: !arm_sve.vector<16xi8>,
                     %b: !arm_sve.vector<16xi8>,
                     %c: !arm_sve.vector<4xi32>) -> !arm_sve.vector<4xi32> {
-  // CHECK: arm_sve.smmla {{.*}}: !arm_sve.vector<16xi8> to !arm_sve.vector<4xi3
+  // CHECK: arm_sve.smmla {{.*}}: <16xi8> to <4xi3
   %0 = arm_sve.smmla %c, %a, %b :
              !arm_sve.vector<16xi8> to !arm_sve.vector<4xi32>
   return %0 : !arm_sve.vector<4xi32>
 }
 
 func @arm_sve_udot(%a: !arm_sve.vector<16xi8>,
                    %b: !arm_sve.vector<16xi8>,
                    %c: !arm_sve.vector<4xi32>) -> !arm_sve.vector<4xi32> {
-  // CHECK: arm_sve.udot {{.*}}: !arm_sve.vector<16xi8> to !arm_sve.vector<4xi32
+  // CHECK: arm_sve.udot {{.*}}: <16xi8> to <4xi32
   %0 = arm_sve.udot %c, %a, %b :
              !arm_sve.vector<16xi8> to !arm_sve.vector<4xi32>
   return %0 : !arm_sve.vector<4xi32>
 }
 
 func @arm_sve_ummla(%a: !arm_sve.vector<16xi8>,
                     %b: !arm_sve.vector<16xi8>,
                     %c: !arm_sve.vector<4xi32>) -> !arm_sve.vector<4xi32> {
-  // CHECK: arm_sve.ummla {{.*}}: !arm_sve.vector<16xi8> to !arm_sve.vector<4xi3
+  // CHECK: arm_sve.ummla {{.*}}: <16xi8> to <4xi3
   %0 = arm_sve.ummla %c, %a, %b :
              !arm_sve.vector<16xi8> to !arm_sve.vector<4xi32>
   return %0 : !arm_sve.vector<4xi32>
 }
 
 func @arm_sve_arithi(%a: !arm_sve.vector<4xi32>,
                      %b: !arm_sve.vector<4xi32>,
                      %c: !arm_sve.vector<4xi32>) -> !arm_sve.vector<4xi32> {

mlir/test/Dialect/Async/async-to-async-runtime-eliminate-blocking.mlir

 // CHECK: %[[ID:.*]] = async.coro.id
 // CHECK: %[[HDL:.*]] = async.coro.begin %[[ID]]
 // CHECK: br ^[[ORIGINAL_ENTRY:.*]]
 // CHECK  ^[[ORIGINAL_ENTRY]]:
 // CHECK:   %[[VAL:.*]] = arith.addf %[[ARG]], %[[ARG]] : f32
   %0 = arith.addf %arg0, %arg0 : f32
 // CHECK:   %[[VAL_STORAGE:.*]] = async.runtime.create : !async.value<f32>
   %1 = async.runtime.create: !async.value<f32>
-// CHECK:   async.runtime.store %[[VAL]], %[[VAL_STORAGE]] : !async.value<f32>
+// CHECK:   async.runtime.store %[[VAL]], %[[VAL_STORAGE]] : <f32>
   async.runtime.store %0, %1: !async.value<f32>
 // CHECK:   async.runtime.set_available %[[VAL_STORAGE]] : !async.value<f32>
   async.runtime.set_available %1: !async.value<f32>
 
 // CHECK:   %[[SAVED:.*]] = async.coro.save %[[HDL]]
 // CHECK:   async.runtime.await_and_resume %[[VAL_STORAGE]], %[[HDL]]
 // CHECK:   async.coro.suspend %[[SAVED]]
 // CHECK-SAME: ^[[SUSPEND:.*]], ^[[RESUME:.*]], ^[[CLEANUP:.*]]
   %2 = async.await %1 : !async.value<f32>
 
 // CHECK: ^[[RESUME]]:
 // CHECK:   %[[IS_ERROR:.*]] = async.runtime.is_error %[[VAL_STORAGE]] : !async.value<f32>
 // CHECK:   cond_br %[[IS_ERROR]], ^[[BRANCH_ERROR:.*]], ^[[BRANCH_OK:.*]]
 
 // CHECK: ^[[BRANCH_OK]]:
-// CHECK:   %[[LOADED:.*]] = async.runtime.load %[[VAL_STORAGE]] : !async.value<f32>
+// CHECK:   %[[LOADED:.*]] = async.runtime.load %[[VAL_STORAGE]] : <f32>
 // CHECK:   %[[RETURNED:.*]] = arith.mulf %[[ARG]], %[[LOADED]] : f32
-// CHECK:   async.runtime.store %[[RETURNED]], %[[RETURNED_STORAGE]] : !async.value<f32>
+// CHECK:   async.runtime.store %[[RETURNED]], %[[RETURNED_STORAGE]] : <f32>
 // CHECK:   async.runtime.set_available %[[RETURNED_STORAGE]]
 // CHECK:   async.runtime.set_available %[[TOKEN]]
 // CHECK:   br ^[[CLEANUP]]
   %3 = arith.mulf %arg0, %2 : f32
   return %3: f32
 
 // CHECK: ^[[BRANCH_ERROR]]:
 // CHECK:   async.runtime.set_error %[[TOKEN]]

 // CHECK:   %[[IS_TOKEN_ERROR:.*]] = async.runtime.is_error %[[RETURNED_TO_CALLER]]#0 : !async.token
 // CHECK:   cond_br %[[IS_TOKEN_ERROR]], ^[[BRANCH_ERROR:.*]], ^[[BRANCH_TOKEN_OK:.*]]
 
 // CHECK: ^[[BRANCH_TOKEN_OK]]:
 // CHECK:   %[[IS_VALUE_ERROR:.*]] = async.runtime.is_error %[[RETURNED_TO_CALLER]]#1 : !async.value<f32>
 // CHECK:   cond_br %[[IS_VALUE_ERROR]], ^[[BRANCH_ERROR:.*]], ^[[BRANCH_VALUE_OK:.*]]
 
 // CHECK: ^[[BRANCH_VALUE_OK]]:
-// CHECK:   %[[LOADED:.*]] = async.runtime.load %[[RETURNED_TO_CALLER]]#1 : !async.value<f32>
-// CHECK:   async.runtime.store %[[LOADED]], %[[RETURNED_STORAGE]] : !async.value<f32>
+// CHECK:   %[[LOADED:.*]] = async.runtime.load %[[RETURNED_TO_CALLER]]#1 : <f32>
+// CHECK:   async.runtime.store %[[LOADED]], %[[RETURNED_STORAGE]] : <f32>
 // CHECK:   async.runtime.set_available %[[RETURNED_STORAGE]]
 // CHECK:   async.runtime.set_available %[[TOKEN]]
 // CHECK:   br ^[[CLEANUP]]
   return %r: f32
 // CHECK: ^[[BRANCH_ERROR]]:
 // CHECK:   async.runtime.set_error %[[TOKEN]]
 // CHECK:   async.runtime.set_error %[[RETURNED_STORAGE]]
 // CHECK:   br ^[[CLEANUP]]

 // CHECK:   %[[IS_TOKEN_ERROR_1:.*]] = async.runtime.is_error %[[RETURNED_TO_CALLER_1]]#0 : !async.token
 // CHECK:   cond_br %[[IS_TOKEN_ERROR_1]], ^[[BRANCH_ERROR:.*]], ^[[BRANCH_TOKEN_OK_1:.*]]
 
 // CHECK: ^[[BRANCH_TOKEN_OK_1]]:
 // CHECK:   %[[IS_VALUE_ERROR_1:.*]] = async.runtime.is_error %[[RETURNED_TO_CALLER_1]]#1 : !async.value<f32>
 // CHECK:   cond_br %[[IS_VALUE_ERROR_1]], ^[[BRANCH_ERROR:.*]], ^[[BRANCH_VALUE_OK_1:.*]]
 
 // CHECK: ^[[BRANCH_VALUE_OK_1]]:
-// CHECK:   %[[LOADED_1:.*]] = async.runtime.load %[[RETURNED_TO_CALLER_1]]#1 : !async.value<f32>
+// CHECK:   %[[LOADED_1:.*]] = async.runtime.load %[[RETURNED_TO_CALLER_1]]#1 : <f32>
 // CHECK:   %[[RETURNED_TO_CALLER_2:.*]]:2 = call @simple_callee(%[[LOADED_1]]) : (f32) -> (!async.token, !async.value<f32>)
 // CHECK:   %[[SAVED_2:.*]] = async.coro.save %[[HDL]]
 // CHECK:   async.runtime.await_and_resume %[[RETURNED_TO_CALLER_2]]#0, %[[HDL]]
 // CHECK:   async.coro.suspend %[[SAVED_2]]
 // CHECK-SAME: ^[[SUSPEND:.*]], ^[[RESUME_2:.*]], ^[[CLEANUP:.*]]
   %s = call @simple_callee(%r): (f32) -> f32
 
 // CHECK: ^[[RESUME_2]]:
 // CHECK:   %[[IS_TOKEN_ERROR_2:.*]] = async.runtime.is_error %[[RETURNED_TO_CALLER_2]]#0 : !async.token
 // CHECK:   cond_br %[[IS_TOKEN_ERROR_2]], ^[[BRANCH_ERROR:.*]], ^[[BRANCH_TOKEN_OK_2:.*]]
 
 // CHECK: ^[[BRANCH_TOKEN_OK_2]]:
 // CHECK:   %[[IS_VALUE_ERROR_2:.*]] = async.runtime.is_error %[[RETURNED_TO_CALLER_2]]#1 : !async.value<f32>
 // CHECK:   cond_br %[[IS_VALUE_ERROR_2]], ^[[BRANCH_ERROR:.*]], ^[[BRANCH_VALUE_OK_2:.*]]
 
 // CHECK: ^[[BRANCH_VALUE_OK_2]]:
-// CHECK:   %[[LOADED_2:.*]] = async.runtime.load %[[RETURNED_TO_CALLER_2]]#1 : !async.value<f32>
-// CHECK:   async.runtime.store %[[LOADED_2]], %[[RETURNED_STORAGE]] : !async.value<f32>
+// CHECK:   %[[LOADED_2:.*]] = async.runtime.load %[[RETURNED_TO_CALLER_2]]#1 : <f32>
+// CHECK:   async.runtime.store %[[LOADED_2]], %[[RETURNED_STORAGE]] : <f32>
 // CHECK:   async.runtime.set_available %[[RETURNED_STORAGE]]
 // CHECK:   async.runtime.set_available %[[TOKEN]]
 // CHECK:   br ^[[CLEANUP]]
   return %s: f32
 // CHECK: ^[[BRANCH_ERROR]]:
 // CHECK:   async.runtime.set_error %[[TOKEN]]
 // CHECK:   async.runtime.set_error %[[RETURNED_STORAGE]]
 // CHECK:   br ^[[CLEANUP]]

mlir/test/Dialect/Async/async-to-async-runtime.mlir

 func @execute_and_return_f32() -> f32 {
  // CHECK: %[[RET:.*]]:2 = call @async_execute_fn
   %token, %result = async.execute -> !async.value<f32> {
     %c0 = arith.constant 123.0 : f32
     async.yield %c0 : f32
   }
 
   // CHECK: async.runtime.await %[[RET]]#1 : !async.value<f32>
-  // CHECK: %[[VALUE:.*]] = async.runtime.load %[[RET]]#1 : !async.value<f32>
+  // CHECK: %[[VALUE:.*]] = async.runtime.load %[[RET]]#1 : <f32>
   %0 = async.await %result : !async.value<f32>
 
   // CHECK: return %[[VALUE]]
   return %0 : f32
 }
 
 // Function outlined from the async.execute operation.
 // CHECK-LABEL: func private @async_execute_fn()

 
 // Check the error of the awaited token after resumption.
 // CHECK: ^[[RESUME_1]]:
 // CHECK:   %[[ERR:.*]] = async.runtime.is_error %[[ARG]]
 // CHECK:   cond_br %[[ERR]], ^[[SET_ERROR:.*]], ^[[CONTINUATION:.*]]
 
 // // Load from the async.value argument after error checking.
 // CHECK: ^[[CONTINUATION:.*]]:
-// CHECK:   %[[LOADED:.*]] = async.runtime.load %[[ARG]] : !async.value<f32
+// CHECK:   %[[LOADED:.*]] = async.runtime.load %[[ARG]] : <f32
 // CHECK:   arith.addf %[[LOADED]], %[[LOADED]] : f32
 // CHECK:   async.runtime.set_available %[[TOKEN]]
 
 // CHECK: ^[[CLEANUP]]:
 // CHECK: ^[[SUSPEND]]:
 
 // -----
 

mlir/test/Dialect/Async/runtime.mlir

 func @resume(%arg0: !async.coro.handle) {
   // CHECK: async.runtime.resume %arg0
   async.runtime.resume %arg0
   return
 }
 
 // CHECK-LABEL: @store
 func @store(%arg0: f32, %arg1: !async.value<f32>) {
-  // CHECK: async.runtime.store %arg0, %arg1 : !async.value<f32>
-  async.runtime.store %arg0, %arg1 : !async.value<f32>
+  // CHECK: async.runtime.store %arg0, %arg1 : <f32>
+  async.runtime.store %arg0, %arg1 : <f32>
   return
 }
 
 // CHECK-LABEL: @load
 func @load(%arg0: !async.value<f32>) -> f32 {
-  // CHECK: %0 = async.runtime.load %arg0 : !async.value<f32>
+  // CHECK: %0 = async.runtime.load %arg0 : <f32>
   // CHECK: return %0 : f32
-  %0 = async.runtime.load %arg0 : !async.value<f32>
+  %0 = async.runtime.load %arg0 : <f32>
   return %0 : f32
 }
 
 // CHECK-LABEL: @add_to_group
 func @add_to_group(%arg0: !async.token, %arg1: !async.value<f32>,
                    %arg2: !async.group) {
   // CHECK: async.runtime.add_to_group %arg0, %arg2 : !async.token
   async.runtime.add_to_group %arg0, %arg2 : !async.token

mlir/test/Dialect/Linalg/vectorization.mlir

 // RUN: mlir-opt %s -test-linalg-transform-patterns=test-linalg-to-vector-patterns -split-input-file | FileCheck %s
 
 // -----
 
 // CHECK-LABEL: contraction_dot
 func @contraction_dot(%A: memref<1584xf32>, %B: memref<1584xf32>, %C: memref<f32>) {
 
 // CHECK: arith.mulf %{{.*}}, %{{.*}} : vector<1584xf32>
-// CHECK: vector.multi_reduction #vector.kind<add>, %{{.*}} [0] : vector<1584xf32> to f32
+// CHECK: vector.multi_reduction <add>, %{{.*}} [0] : vector<1584xf32> to f32
 // CHECK: arith.addf %{{.*}}, %{{.*}} : f32
   linalg.dot ins(%A, %B: memref<1584xf32>, memref<1584xf32>)
             outs(%C: memref<f32>)
   return
 }
 
 // -----
 
 // CHECK-LABEL: contraction_matvec
 func @contraction_matvec(%A: memref<1584x1584xf32>, %B: memref<1584xf32>, %C: memref<1584xf32>) {
 
 // CHECK: arith.mulf %{{.*}}, %{{.*}} : vector<1584x1584xf32>
-// CHECK: vector.multi_reduction #vector.kind<add>, %{{.*}} [1] : vector<1584x1584xf32> to vector<1584xf32>
+// CHECK: vector.multi_reduction <add>, %{{.*}} [1] : vector<1584x1584xf32> to vector<1584xf32>
 // CHECK: arith.addf %{{.*}}, %{{.*}} : vector<1584xf32>
   linalg.matvec ins(%A, %B: memref<1584x1584xf32>, memref<1584xf32>)
             outs(%C: memref<1584xf32>)
   return
 }
 
 // -----
 
 // CHECK-LABEL: contraction_matmul
 func @contraction_matmul(%A: memref<1584x1584xf32>, %B: memref<1584x1584xf32>, %C: memref<1584x1584xf32>) {
 // CHECK: arith.mulf %{{.*}}, %{{.*}} : vector<1584x1584x1584xf32>
-// CHECK: vector.multi_reduction #vector.kind<add>, %{{.*}} [2] : vector<1584x1584x1584xf32> to vector<1584x1584xf32>
+// CHECK: vector.multi_reduction <add>, %{{.*}} [2] : vector<1584x1584x1584xf32> to vector<1584x1584xf32>
 // CHECK: arith.addf %{{.*}}, %{{.*}} : vector<1584x1584xf32>
   linalg.matmul ins(%A, %B: memref<1584x1584xf32>, memref<1584x1584xf32>)
             outs(%C: memref<1584x1584xf32>)
   return
 }
 
 // -----
 
 // CHECK-LABEL: contraction_batch_matmul
 func @contraction_batch_matmul(%A: memref<1584x1584x1584xf32>, %B: memref<1584x1584x1584xf32>, %C: memref<1584x1584x1584xf32>) {
 // CHECK: arith.mulf %{{.*}}, %{{.*}} : vector<1584x1584x1584x1584xf32>
-// CHECK: vector.multi_reduction #vector.kind<add>, %{{.*}} [3] : vector<1584x1584x1584x1584xf32> to vector<1584x1584x1584xf32>
+// CHECK: vector.multi_reduction <add>, %{{.*}} [3] : vector<1584x1584x1584x1584xf32> to vector<1584x1584x1584xf32>
 // CHECK: arith.addf %{{.*}}, %{{.*}} : vector<1584x1584x1584xf32>
   linalg.batch_matmul
     ins(%A, %B: memref<1584x1584x1584xf32>, memref<1584x1584x1584xf32>)
    outs(%C: memref<1584x1584x1584xf32>)
   return
 }
 
 // -----

 
 // CHECK-LABEL: func @vectorization_test
 func @vectorization_test(%A: memref<8x16xf32>, %B: memref<16x32xf32>,
                          %C: memref<8x32xf32>) {
   //       CHECK: vector.transfer_read %{{.*}} : memref<8x16xf32>, vector<8x32x16xf32>
   //       CHECK: vector.transfer_read %{{.*}} : memref<16x32xf32>, vector<8x32x16xf32>
   //       CHECK: vector.transfer_read %{{.*}} : memref<8x32xf32>, vector<8x32xf32>
   //       CHECK: %[[MUL:.*]] = arith.mulf %{{.*}}, %{{.*}} : vector<8x32x16xf32>
-  //       CHECK: %[[R:.*]] = vector.multi_reduction #vector.kind<add>, %[[MUL]] [2] : vector<8x32x16xf32> to vector<8x32xf32>
+  //       CHECK: %[[R:.*]] = vector.multi_reduction <add>, %[[MUL]] [2] : vector<8x32x16xf32> to vector<8x32xf32>
   //       CHECK: arith.addf %[[R]], %{{.*}} : vector<8x32xf32>
   //       CHECK: vector.transfer_write %{{.*}}, %{{.*}} : vector<8x32xf32>, memref<8x32xf32>
   linalg.generic #matmul_trait
     ins(%A, %B : memref<8x16xf32>, memref<16x32xf32>)
    outs(%C : memref<8x32xf32>) {
     ^bb(%a: f32, %b: f32, %c: f32) :
       %d = arith.mulf %a, %b: f32
       %e = arith.addf %c, %d: f32

 
 // CHECK-LABEL: func @generic_output_transpose
 func @generic_output_transpose(%A: memref<8x16xf32>, %B: memref<16x32xf32>,
                          %C: memref<32x8xf32>) {
   //       CHECK: vector.transfer_read %{{.*}} : memref<8x16xf32>, vector<8x32x16xf32>
   //       CHECK: vector.transfer_read %{{.*}} : memref<16x32xf32>, vector<8x32x16xf32>
   //       CHECK: vector.transfer_read %{{.*}} : memref<32x8xf32>, vector<8x32xf32>
   //       CHECK: %[[MUL:.*]] = arith.mulf %{{.*}}, %{{.*}} : vector<8x32x16xf32>
-  //       CHECK: %[[R:.*]] = vector.multi_reduction #vector.kind<add>, %[[MUL]] [2] : vector<8x32x16xf32> to vector<8x32xf32>
+  //       CHECK: %[[R:.*]] = vector.multi_reduction <add>, %[[MUL]] [2] : vector<8x32x16xf32> to vector<8x32xf32>
   //       CHECK: arith.addf %[[R]], %{{.*}} : vector<8x32xf32>
   //       CHECK: vector.transfer_write %{{.*}}, %{{.*}} : vector<8x32xf32>, memref<32x8xf32>
   linalg.generic #matmul_transpose_out_trait
     ins(%A, %B : memref<8x16xf32>, memref<16x32xf32>)
    outs(%C : memref<32x8xf32>) {
     ^bb(%a: f32, %b: f32, %c: f32) :
       %d = arith.mulf %a, %b: f32
       %e = arith.addf %c, %d: f32

 
 // CHECK-LABEL: func @vectorization_test_integer
 func @vectorization_test_integer(%A: memref<8x16xi32>, %B: memref<16x32xi32>,
                                  %C: memref<8x32xi32>) {
   //       CHECK: vector.transfer_read %{{.*}} : memref<8x16xi32>, vector<8x32x16xi32>
   //       CHECK: vector.transfer_read %{{.*}} : memref<16x32xi32>, vector<8x32x16xi32>
   //       CHECK: vector.transfer_read %{{.*}} : memref<8x32xi32>, vector<8x32xi32>
   //       CHECK: %[[MUL:.*]] = arith.muli %{{.*}}, %{{.*}} : vector<8x32x16xi32>
-  //       CHECK: %[[R:.*]] = vector.multi_reduction #vector.kind<add>, %[[MUL]] [2] : vector<8x32x16xi32> to vector<8x32xi32>
+  //       CHECK: %[[R:.*]] = vector.multi_reduction <add>, %[[MUL]] [2] : vector<8x32x16xi32> to vector<8x32xi32>
   //       CHECK: arith.addi %[[R]], %{{.*}} : vector<8x32xi32>
 
   //       CHECK: vector.transfer_write %{{.*}}, %{{.*}} : vector<8x32xi32>, memref<8x32xi32>
   linalg.generic #matmul_trait
     ins(%A, %B : memref<8x16xi32>, memref<16x32xi32>)
    outs(%C : memref<8x32xi32>) {
     ^bb(%a: i32, %b: i32, %c: i32) :
       %d = arith.muli %a, %b: i32
       %e = arith.addi %c, %d: i32
       linalg.yield %e : i32
   }
   return
 }
 
 // -----
 
 // CHECK-LABEL: func @vectorization_test_2
 func @vectorization_test_2(%A: memref<8x16xf32>, %B: memref<16x32xf32>,
                          %C: memref<8x32xf32>) {
   //       CHECK: arith.mulf %{{.*}}, %{{.*}} : vector<8x32x16xf32>
-  //       CHECK: vector.multi_reduction #vector.kind<add>, %{{.*}} [2] : vector<8x32x16xf32> to vector<8x32xf32>
+  //       CHECK: vector.multi_reduction <add>, %{{.*}} [2] : vector<8x32x16xf32> to vector<8x32xf32>
   //       CHECK: arith.addf %{{.*}}, %{{.*}} : vector<8x32xf32>
   linalg.matmul
     ins(%A, %B: memref<8x16xf32>, memref<16x32xf32>)
    outs(%C: memref<8x32xf32>)
   return
 }
 
 // -----
func @matmul_tensors(
   //   CHECK-DAG:   %[[C0:.*]] = arith.constant 0 : index
   //   CHECK-DAG:   %[[V0:.*]] = vector.transfer_read %[[ARG0]][%[[C0]], %[[C0]]], {{.*}} : tensor<8x4xf32>, vector<8x12x4xf32>
   //   CHECK-DAG:   %[[V1:.*]] = vector.transfer_read %[[ARG1]][%[[C0]], %[[C0]]], {{.*}} : tensor<4x12xf32>, vector<8x12x4xf32>
   //   CHECK-DAG:   %[[V2:.*]] = vector.transfer_read %[[ARG2]][%[[C0]], %[[C0]]], {{.*}} : tensor<8x12xf32>, vector<8x12xf32>
   //
   // linalg matmul lowers gets expanded to a 3D reduction, canonicalization later
   // convert it to a 2D contract.
   //       CHECK:   %[[MUL:.*]] = arith.mulf %[[V0]], %[[V1]] : vector<8x12x4xf32>
-  //       CHECK:   %[[R:.*]] = vector.multi_reduction #vector.kind<add>, %[[MUL]] [2] : vector<8x12x4xf32> to vector<8x12xf32>
+  //       CHECK:   %[[R:.*]] = vector.multi_reduction <add>, %[[MUL]] [2] : vector<8x12x4xf32> to vector<8x12xf32>
   //       CHECK:   %[[ADD:.*]] = arith.addf %[[R]], %[[V2]] : vector<8x12xf32>
   //       CHECK:   %[[W:.*]] = vector.transfer_write %[[ADD]], %[[ARG2]][%[[C0]], %[[C0]]] {in_bounds = [true, true]} : vector<8x12xf32>, tensor<8x12xf32>
   %0 = linalg.matmul  ins(%arg0, %arg1: tensor<8x4xf32>, tensor<4x12xf32>)
                      outs(%arg2: tensor<8x12xf32>)
     -> tensor<8x12xf32>
   //       CHECK:   return %[[W]] : tensor<8x12xf32>
   return %0 : tensor<8x12xf32>
 }

 
 // CHECK-LABEL: func @sum_exp
 func @sum_exp(%input: tensor<4x16x8xf32>, %output: tensor<4x16xf32>)
   -> tensor<4x16xf32>
 {
   // CHECK: vector.transfer_read {{.*}} : tensor<4x16x8xf32>, vector<4x16x8xf32>
   // CHECK: vector.transfer_read {{.*}} {in_bounds = [true, true]} : tensor<4x16xf32>, vector<4x16xf32>
   // CHECK: math.exp {{.*}} : vector<4x16x8xf32>
-  // CHECK: vector.multi_reduction #vector.kind<add>, %{{.*}} [2] : vector<4x16x8xf32> to vector<4x16xf32>
+  // CHECK: vector.multi_reduction <add>, %{{.*}} [2] : vector<4x16x8xf32> to vector<4x16xf32>
   // CHECK: addf {{.*}} : vector<4x16xf32>
   // CHECK: vector.transfer_write {{.*}} : vector<4x16xf32>, tensor<4x16xf32>
   // CHECK: return {{.*}} : tensor<4x16xf32>
   %0 = linalg.generic {
       indexing_maps = [
         affine_map<(d0, d1, d2) -> (d0, d1, d2)>,
         affine_map<(d0, d1, d2) -> (d0, d1)>
       ],
func @sum_exp_2(%input: tensor<3x2xf32>, %input_2: tensor<5x4xf32>, %output: tensor<5x2xf32>)
   -> tensor<5x2xf32>
 {
   // CHECK: vector.transfer_read {{.*}} {in_bounds = [true, true, true, true], permutation_map = #[[$M1]]} : tensor<3x2xf32>, vector<2x3x4x5xf32>
   // CHECK: vector.transfer_read {{.*}} {in_bounds = [true, true, true, true], permutation_map = #[[$M2]]} : tensor<5x4xf32>, vector<2x3x4x5xf32>
   // CHECK: vector.transfer_read {{.*}} {in_bounds = [true, true], permutation_map = #[[$M3]]} : tensor<5x2xf32>, vector<2x5xf32>
   // CHECK: math.exp {{.*}} : vector<2x3x4x5xf32>
   // CHECK: math.exp {{.*}} : vector<2x3x4x5xf32>
   // CHECK: addf {{.*}} : vector<2x3x4x5xf32>
-  // CHECK: vector.multi_reduction #vector.kind<add>, {{.*}}  [1, 2] : vector<2x3x4x5xf32> to vector<2x5xf32>
+  // CHECK: vector.multi_reduction <add>, {{.*}}  [1, 2] : vector<2x3x4x5xf32> to vector<2x5xf32>
   // CHECK: addf {{.*}} : vector<2x5xf32>
   // CHECK: vector.transfer_write {{.*}} {in_bounds = [true, true], permutation_map = #[[$M3]]} : vector<2x5xf32>, tensor<5x2xf32>
   // CHECK: return {{.*}} : tensor<5x2xf32>
   %0 = linalg.generic {
       indexing_maps = [
         affine_map<(d0, d1, d2, d3) -> (d1, d0)>,
         affine_map<(d0, d1, d2, d3) -> (d3, d2)>,
         affine_map<(d0, d1, d2, d3) -> (d3, d0)>

 
 // -----
 
 // CHECK-LABEL:   func @red_max_2d(
 func @red_max_2d(%arg0: tensor<4x4xf32>) -> tensor<4xf32> {
   // CHECK: %[[CMINF:.+]] = arith.constant dense<-3.402820e+38> : vector<4xf32>
   // CHECK: linalg.init_tensor [4] : tensor<4xf32>
   // CHECK: vector.transfer_write {{.*}} : vector<4xf32>, tensor<4xf32>
-  // CHECK: %[[R:.+]] = vector.multi_reduction #vector.kind<maxf>, {{.*}} [1] : vector<4x4xf32> to vector<4xf32>
+  // CHECK: %[[R:.+]] = vector.multi_reduction <maxf>, {{.*}} [1] : vector<4x4xf32> to vector<4xf32>
   // CHECK: maxf %[[R]], %[[CMINF]] : vector<4xf32>
   // CHECK: vector.transfer_write {{.*}} : vector<4xf32>, tensor<4xf32>
   %ident = arith.constant -3.40282e+38 : f32
   %init = linalg.init_tensor [4] : tensor<4xf32>
   %fill = linalg.fill(%ident, %init) : f32, tensor<4xf32> -> tensor<4xf32>
   %red = linalg.generic {indexing_maps = [affine_map<(d0, d1) -> (d0, d1)>,
                                           affine_map<(d0, d1) -> (d0)>],
                          iterator_types = ["parallel", "reduction"]}
                          ins(%arg0 : tensor<4x4xf32>) outs(%fill : tensor<4xf32>) {
   ^bb0(%in0: f32, %out0: f32):  // no predecessors
     %max = arith.maxf %in0, %out0 : f32
     linalg.yield %max : f32
   } -> tensor<4xf32>
   return %red : tensor<4xf32>
 }
 
 // -----
 
 // CHECK-LABEL:   func @red_min_2d(
 func @red_min_2d(%arg0: tensor<4x4xf32>) -> tensor<4xf32> {
   // CHECK: %[[CMAXF:.+]] = arith.constant dense<3.402820e+38> : vector<4xf32>
   // CHECK: linalg.init_tensor [4] : tensor<4xf32>
   // CHECK: vector.transfer_write {{.*}} : vector<4xf32>, tensor<4xf32>
   // CHECK: vector.transfer_read {{.*}} : tensor<4x4xf32>, vector<4x4xf32>
-  // CHECK: %[[R:.+]] = vector.multi_reduction #vector.kind<minf>, {{.*}} [1] : vector<4x4xf32> to vector<4xf32>
+  // CHECK: %[[R:.+]] = vector.multi_reduction <minf>, {{.*}} [1] : vector<4x4xf32> to vector<4xf32>
   // CHECK: arith.minf %[[R]], %[[CMAXF]] : vector<4xf32>
   // CHECK: vector.transfer_write {{.*}} : vector<4xf32>, tensor<4xf32>
   %maxf32 = arith.constant 3.40282e+38 : f32
   %init = linalg.init_tensor [4] : tensor<4xf32>
   %fill = linalg.fill(%maxf32, %init) : f32, tensor<4xf32> -> tensor<4xf32>
   %red = linalg.generic {indexing_maps = [affine_map<(d0, d1) -> (d0, d1)>,
                                           affine_map<(d0, d1) -> (d0)>],
                          iterator_types = ["parallel", "reduction"]}
                          ins(%arg0 : tensor<4x4xf32>) outs(%fill : tensor<4xf32>) {
   ^bb0(%in0: f32, %out0: f32):  // no predecessors
     %min = arith.minf %out0, %in0 : f32
     linalg.yield %min : f32
   } -> tensor<4xf32>
   return %red : tensor<4xf32>
 }
 
 // -----
 
 // CHECK-LABEL:   func @red_mul_2d(
 func @red_mul_2d(%arg0: tensor<4x4xf32>) -> tensor<4xf32> {
   // CHECK: linalg.init_tensor [4] : tensor<4xf32>
   // CHECK: vector.transfer_write {{.*}} : vector<4xf32>, tensor<4xf32>
   // CHECK: vector.transfer_read {{.*}} : tensor<4x4xf32>, vector<4x4xf32>
-  // CHECK: vector.multi_reduction #vector.kind<mul>, {{.*}} [1] : vector<4x4xf32> to vector<4xf32>
+  // CHECK: vector.multi_reduction <mul>, {{.*}} [1] : vector<4x4xf32> to vector<4xf32>
   // CHECK: mulf {{.*}} : vector<4xf32>
   // CHECK: vector.transfer_write {{.*}} : vector<4xf32>, tensor<4xf32>
   %ident = arith.constant 1.0 : f32
   %init = linalg.init_tensor [4] : tensor<4xf32>
   %fill = linalg.fill(%ident, %init) : f32, tensor<4xf32> -> tensor<4xf32>
   %red = linalg.generic {indexing_maps = [affine_map<(d0, d1) -> (d0, d1)>,
                                           affine_map<(d0, d1) -> (d0)>],
                          iterator_types = ["parallel", "reduction"]}
                          ins(%arg0 : tensor<4x4xf32>) outs(%fill : tensor<4xf32>) {
   ^bb0(%in0: f32, %out0: f32):  // no predecessors
     %mul = arith.mulf %in0, %out0 : f32
     linalg.yield %mul : f32
   } -> tensor<4xf32>
   return %red : tensor<4xf32>
 }
 
 // -----
 
 // CHECK-LABEL:   func @red_or_2d(
 func @red_or_2d(%arg0: tensor<4x4xi1>) -> tensor<4xi1> {
   // CHECK: linalg.init_tensor [4] : tensor<4xi1>
   // CHECK: vector.transfer_write {{.*}} : vector<4xi1>, tensor<4xi1>
   // CHECK: vector.transfer_read {{.*}} : tensor<4x4xi1>, vector<4x4xi1>
-  // CHECK: vector.multi_reduction #vector.kind<or>, {{.*}} [1] : vector<4x4xi1> to vector<4xi1>
+  // CHECK: vector.multi_reduction <or>, {{.*}} [1] : vector<4x4xi1> to vector<4xi1>
   // CHECK: vector.transfer_write {{.*}} : vector<4xi1>, tensor<4xi1>
   %ident = arith.constant false
   %init = linalg.init_tensor [4] : tensor<4xi1>
   %fill = linalg.fill(%ident, %init) : i1, tensor<4xi1> -> tensor<4xi1>
   %red = linalg.generic {indexing_maps = [affine_map<(d0, d1) -> (d0, d1)>,
                                           affine_map<(d0, d1) -> (d0)>],
                          iterator_types = ["parallel", "reduction"]}
                          ins(%arg0 : tensor<4x4xi1>) outs(%fill : tensor<4xi1>) {
   ^bb0(%in0: i1, %out0: i1):  // no predecessors
     %or = arith.ori %in0, %out0 : i1
     linalg.yield %or : i1
   } -> tensor<4xi1>
   return %red : tensor<4xi1>
 }
 
 // -----
 
 // CHECK-LABEL:   func @red_and_2d(
 func @red_and_2d(%arg0: tensor<4x4xi1>) -> tensor<4xi1> {
   // CHECK: linalg.init_tensor [4] : tensor<4xi1>
   // CHECK: vector.transfer_write {{.*}} : vector<4xi1>, tensor<4xi1>
   // CHECK: vector.transfer_read {{.*}} : tensor<4x4xi1>, vector<4x4xi1>
-  // CHECK: vector.multi_reduction #vector.kind<and>, {{.*}} [1] : vector<4x4xi1> to vector<4xi1>
+  // CHECK: vector.multi_reduction <and>, {{.*}} [1] : vector<4x4xi1> to vector<4xi1>
   // CHECK: vector.transfer_write {{.*}} : vector<4xi1>, tensor<4xi1>
   %ident = arith.constant true
   %init = linalg.init_tensor [4] : tensor<4xi1>
   %fill = linalg.fill(%ident, %init) : i1, tensor<4xi1> -> tensor<4xi1>
   %red = linalg.generic {indexing_maps = [affine_map<(d0, d1) -> (d0, d1)>,
                                           affine_map<(d0, d1) -> (d0)>],
                          iterator_types = ["parallel", "reduction"]}
                          ins(%arg0 : tensor<4x4xi1>) outs(%fill : tensor<4xi1>) {
   ^bb0(%in0: i1, %out0: i1):  // no predecessors
     %and = arith.andi %in0, %out0 : i1
     linalg.yield %and : i1
   } -> tensor<4xi1>
   return %red : tensor<4xi1>
 }
 
 // -----
 
 // CHECK-LABEL:   func @red_xor_2d(
 func @red_xor_2d(%arg0: tensor<4x4xi1>) -> tensor<4xi1> {
   // CHECK: linalg.init_tensor [4] : tensor<4xi1>
   // CHECK: vector.transfer_write {{.*}} : vector<4xi1>, tensor<4xi1>
   // CHECK: vector.transfer_read {{.*}} : tensor<4x4xi1>, vector<4x4xi1>
-  // CHECK: vector.multi_reduction #vector.kind<xor>, {{.*}} [1] : vector<4x4xi1> to vector<4xi1>
+  // CHECK: vector.multi_reduction <xor>, {{.*}} [1] : vector<4x4xi1> to vector<4xi1>
   // CHECK: vector.transfer_write {{.*}} : vector<4xi1>, tensor<4xi1>
   %ident = arith.constant false
   %init = linalg.init_tensor [4] : tensor<4xi1>
   %fill = linalg.fill(%ident, %init) : i1, tensor<4xi1> -> tensor<4xi1>
   %red = linalg.generic {indexing_maps = [affine_map<(d0, d1) -> (d0, d1)>,
                                           affine_map<(d0, d1) -> (d0)>],
                          iterator_types = ["parallel", "reduction"]}
                          ins(%arg0 : tensor<4x4xi1>) outs(%fill : tensor<4xi1>) {

 // CHECK-DAG: #[[$M6:.*]] = affine_map<(d0, d1) -> (d0, 0)>
 
 // CHECK-LABEL:   func @fused_broadcast_red_2d
 func @fused_broadcast_red_2d(%arg0: tensor<4x4xf32>, %arg1: tensor<4x1xf32>) -> tensor<4xf32> {
   // CHECK: vector.transfer_read {{.*}} {in_bounds = [true, true]} : tensor<4x4xf32>, vector<4x4xf32>
   // CHECK: vector.transfer_read {{.*}} {in_bounds = [true, true], permutation_map = #[[$M6]]} : tensor<4x1xf32>, vector<4x4xf32>
   // CHECK: subf {{.*}} : vector<4x4xf32>
   // CHECK: math.exp {{.*}} : vector<4x4xf32>
-  // CHECK: vector.multi_reduction #vector.kind<add>, {{.*}} : vector<4x4xf32> to vector<4xf32>
+  // CHECK: vector.multi_reduction <add>, {{.*}} : vector<4x4xf32> to vector<4xf32>
   // CHECK: addf {{.*}} : vector<4xf32>
   // CHECK: vector.transfer_write {{.*}} {in_bounds = [true]} : vector<4xf32>, tensor<4xf32>
   %c0 = arith.constant 0.0 : f32
   %init = linalg.init_tensor [4] : tensor<4xf32>
   %fill = linalg.fill(%c0, %init) : f32, tensor<4xf32> -> tensor<4xf32>
   %red = linalg.generic {indexing_maps = [affine_map<(d0, d1) -> (d0, d1)>,
                                           affine_map<(d0, d1) -> (d0, 0)>,
                                           affine_map<(d0, d1) -> (d0)>],
func @reduce_1d(%arg0: tensor<32xf32>) -> tensor<f32> {
   %0 = linalg.init_tensor [] : tensor<f32>
 
   //      CHECK: %[[f:.*]] = vector.transfer_write %[[vF0]], %[[init]][]
   // CHECK-SAME:   : vector<f32>, tensor<f32>
   %1 = linalg.fill(%f0, %0) : f32, tensor<f32> -> tensor<f32>
   //      CHECK: %[[r:.*]] = vector.transfer_read %[[A]][%[[C0]]]
   // CHECK-SAME:   : tensor<32xf32>, vector<32xf32>
   //      CHECK: %[[f0:.*]] = vector.extractelement %[[vF0]][] : vector<f32>
-  //      CHECK: %[[red:.*]] = vector.multi_reduction #vector.kind<add>, %[[r]] [0]
+  //      CHECK: %[[red:.*]] = vector.multi_reduction <add>, %[[r]] [0]
   // CHECK-SAME:   : vector<32xf32> to f32
   //      CHECK: %[[a:.*]] = arith.addf %[[red]], %[[f0]] : f32
   //      CHECK: %[[red_v1:.*]] = vector.broadcast %[[a]] : f32 to vector<f32>
   //      CHECK: %[[res:.*]] = vector.transfer_write %[[red_v1]], %[[f]][]
   // CHECK-SAME:   : vector<f32>, tensor<f32>
   %2 = linalg.generic {
          indexing_maps = [affine_map<(d0) -> (d0)>,
                           affine_map<(d0) -> ()>],

mlir/test/Dialect/Vector/canonicalize.mlir

func @insert_slice_of_transfer_write_rank_extending(%t1 : tensor<?x?x12xf32>, %v : vector<5x6xf32>, %s : index, %t2 : tensor<5x6xf32>) -> tensor<?x?x12xf32> {
   return %1 : tensor<?x?x12xf32>
 }
 
 // -----
 
 // CHECK-LABEL: func @vector_multi_reduction_single_parallel(
 //  CHECK-SAME:     %[[v:.*]]: vector<2xf32>
 func @vector_multi_reduction_single_parallel(%arg0: vector<2xf32>) -> vector<2xf32> {
-    %0 = vector.multi_reduction #vector.kind<mul>, %arg0 [] : vector<2xf32> to vector<2xf32>
+    %0 = vector.multi_reduction <mul>, %arg0 [] : vector<2xf32> to vector<2xf32>
 
 //       CHECK:     return %[[v]] : vector<2xf32>
     return %0 : vector<2xf32>
 }
 
 // -----
 
 // CHECK-LABEL: func @insert_strided_slice_full_range

mlir/test/Dialect/Vector/invalid.mlir

 // RUN: mlir-opt %s -split-input-file -verify-diagnostics
 
 // -----
 
 func @broadcast_to_scalar(%arg0: f32) -> f32 {
-  // expected-error@+1 {{'vector.broadcast' op result #0 must be vector of any type values, but got 'f32'}}
+  // expected-error@+1 {{custom op 'vector.broadcast' invalid kind of type specified}}
   %0 = vector.broadcast %arg0 : f32 to f32
 }
 
 // -----
 
 func @broadcast_rank_too_high(%arg0: vector<4x4xf32>) {
   // expected-error@+1 {{'vector.broadcast' op source rank higher than destination rank}}
   %1 = vector.broadcast %arg0 : vector<4x4xf32> to vector<4xf32>

 func @shape_cast_invalid_rank_expansion(%arg0 : vector<15x2xf32>) {
   // expected-error@+1 {{invalid shape cast}}
   %0 = vector.shape_cast %arg0 : vector<15x2xf32> to vector<5x2x3x1xf32>
 }
 
 // -----
 
 func @bitcast_not_vector(%arg0 : vector<5x1x3x2xf32>) {
-  // expected-error@+1 {{must be vector of any type values}}
+  // expected-error@+1 {{'vector.bitcast' invalid kind of type specified}}
   %0 = vector.bitcast %arg0 : vector<5x1x3x2xf32> to f32
 }
 
 // -----
 
 func @bitcast_rank_mismatch_to_0d(%arg0 : vector<1xf32>) {
   // expected-error@+1 {{op failed to verify that all of {source, result} have same rank}}
   %0 = vector.bitcast %arg0 : vector<1xf32> to vector<f32>

mlir/test/Dialect/Vector/ops.mlir

func @extract_insert_map(%v: vector<32xf32>, %v2: vector<16x32xf32>,
   // CHECK: %[[R1:.*]] = vector.insert_map %[[V1]], %{{.*}}[%{{.*}}, %{{.*}}] : vector<4x2xf32> into vector<16x32xf32>
   %r2 = vector.insert_map %vd2, %v2[%id0, %id1] : vector<4x2xf32> into vector<16x32xf32>
   // CHECK: return %[[R]], %[[R1]] : vector<32xf32>, vector<16x32xf32>
   return %r, %r2 : vector<32xf32>, vector<16x32xf32>
 }
 
 // CHECK-LABEL: @multi_reduction
 func @multi_reduction(%0: vector<4x8x16x32xf32>) -> f32 {
-  %1 = vector.multi_reduction #vector.kind<add>, %0 [1, 3] :
+  %1 = vector.multi_reduction <add>, %0 [1, 3] :
     vector<4x8x16x32xf32> to vector<4x16xf32>
-  %2 = vector.multi_reduction #vector.kind<add>, %1 [0, 1] :
+  %2 = vector.multi_reduction <add>, %1 [0, 1] :
     vector<4x16xf32> to f32
   return %2 : f32
 }

mlir/test/Dialect/Vector/vector-multi-reduction-lowering.mlir

 // RUN: mlir-opt %s -test-vector-multi-reduction-lowering-patterns | FileCheck %s
 
 func @vector_multi_reduction(%arg0: vector<2x4xf32>) -> vector<2xf32> {
-    %0 = vector.multi_reduction #vector.kind<mul>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
+    %0 = vector.multi_reduction <mul>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
     return %0 : vector<2xf32>
 }
 // CHECK-LABEL: func @vector_multi_reduction
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<2x4xf32>
 //       CHECK:       %[[RESULT_VEC_0:.+]] = arith.constant dense<{{.*}}> : vector<2xf32>
 //       CHECK:       %[[C0:.+]] = arith.constant 0 : index
 //       CHECK:       %[[C1:.+]] = arith.constant 1 : index
 //       CHECK:       %[[V0:.+]] = vector.extract %[[INPUT]][0]
 //       CHECK:       %[[RV0:.+]] = vector.reduction "mul", %[[V0]] : vector<4xf32> into f32
 //       CHECK:       %[[RESULT_VEC_1:.+]] = vector.insertelement %[[RV0:.+]], %[[RESULT_VEC_0]][%[[C0]] : index] : vector<2xf32>
 //       CHECK:       %[[V1:.+]] = vector.extract %[[INPUT]][1]
 //       CHECK:       %[[RV1:.+]] = vector.reduction "mul", %[[V1]] : vector<4xf32> into f32
 //       CHECK:       %[[RESULT_VEC:.+]] = vector.insertelement %[[RV1:.+]], %[[RESULT_VEC_1]][%[[C1]] : index] : vector<2xf32>
 //       CHECK:       return %[[RESULT_VEC]]
 
 func @vector_multi_reduction_to_scalar(%arg0: vector<2x4xf32>) -> f32 {
-    %0 = vector.multi_reduction #vector.kind<mul>, %arg0 [0, 1] : vector<2x4xf32> to f32
+    %0 = vector.multi_reduction <mul>, %arg0 [0, 1] : vector<2x4xf32> to f32
     return %0 : f32
 }
 // CHECK-LABEL: func @vector_multi_reduction_to_scalar
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<2x4xf32>
 //       CHECK:   %[[CASTED:.*]] = vector.shape_cast %[[INPUT]] : vector<2x4xf32> to vector<8xf32>
 //       CHECK:   %[[REDUCED:.*]] = vector.reduction "mul", %[[CASTED]] : vector<8xf32> into f32
 //       CHECK:   %[[INSERTED:.*]] = vector.insertelement %[[REDUCED]], {{.*}} : vector<1xf32>
 //       CHECK:   %[[RES:.*]] = vector.extract %[[INSERTED]][0] : vector<1xf32>
 //       CHECK:   return %[[RES]]
 
 func @vector_reduction_inner(%arg0: vector<2x3x4x5xi32>) -> vector<2x3xi32> {
-    %0 = vector.multi_reduction #vector.kind<add>, %arg0 [2, 3] : vector<2x3x4x5xi32> to vector<2x3xi32>
+    %0 = vector.multi_reduction <add>, %arg0 [2, 3] : vector<2x3x4x5xi32> to vector<2x3xi32>
     return %0 : vector<2x3xi32>
 }
 // CHECK-LABEL: func @vector_reduction_inner
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<2x3x4x5xi32>
 //       CHECK:       %[[FLAT_RESULT_VEC_0:.+]] = arith.constant dense<0> : vector<6xi32>
 //   CHECK-DAG:       %[[C0:.+]] = arith.constant 0 : index
 //   CHECK-DAG:       %[[C1:.+]] = arith.constant 1 : index
 //   CHECK-DAG:       %[[C2:.+]] = arith.constant 2 : index

 ///       CHECK:      %[[V5:.+]] = vector.extract %[[RESHAPED_INPUT]][5] : vector<6x20xi32>
 //       CHECK:       %[[V5R:.+]] = vector.reduction "add", %[[V5]] : vector<20xi32> into i32
 //       CHECK:       %[[FLAT_RESULT_VEC:.+]] = vector.insertelement %[[V5R]], %[[FLAT_RESULT_VEC_5]][%[[C5]] : index] : vector<6xi32>
 //       CHECK:       %[[RESULT:.+]] = vector.shape_cast %[[FLAT_RESULT_VEC]] : vector<6xi32> to vector<2x3xi32>
 //       CHECK:       return %[[RESULT]]
 
 
 func @vector_multi_reduction_transposed(%arg0: vector<2x3x4x5xf32>) -> vector<2x5xf32> {
-    %0 = vector.multi_reduction #vector.kind<add>, %arg0 [1, 2] : vector<2x3x4x5xf32> to vector<2x5xf32>
+    %0 = vector.multi_reduction <add>, %arg0 [1, 2] : vector<2x3x4x5xf32> to vector<2x5xf32>
     return %0 : vector<2x5xf32>
 }
 
 // CHECK-LABEL: func @vector_multi_reduction_transposed
 //  CHECK-SAME:    %[[INPUT:.+]]: vector<2x3x4x5xf32>
 //       CHECK:     %[[TRANSPOSED_INPUT:.+]] = vector.transpose %[[INPUT]], [0, 3, 1, 2] : vector<2x3x4x5xf32> to vector<2x5x3x4xf32>
 //       CHECK:     vector.shape_cast %[[TRANSPOSED_INPUT]] : vector<2x5x3x4xf32> to vector<10x12xf32>
 //       CHECK:     %[[RESULT:.+]] = vector.shape_cast %{{.*}} : vector<10xf32> to vector<2x5xf32>
 //       CHECK:       return %[[RESULT]]
 
 func @vector_multi_reduction_ordering(%arg0: vector<3x2x4xf32>) -> vector<2x4xf32> {
-    %0 = vector.multi_reduction #vector.kind<mul>, %arg0 [0] : vector<3x2x4xf32> to vector<2x4xf32>
+    %0 = vector.multi_reduction <mul>, %arg0 [0] : vector<3x2x4xf32> to vector<2x4xf32>
     return %0 : vector<2x4xf32>
 }
 // CHECK-LABEL: func @vector_multi_reduction_ordering
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<3x2x4xf32>
 //       CHECK:       %[[RESULT_VEC_0:.+]] = arith.constant dense<{{.*}}> : vector<8xf32>
 //       CHECK:       %[[C0:.+]] = arith.constant 0 : index
 //       CHECK:       %[[C1:.+]] = arith.constant 1 : index
 //       CHECK:       %[[C2:.+]] = arith.constant 2 : index

mlir/test/Dialect/Vector/vector-multi-reduction-outer-lowering.mlir

 // RUN: mlir-opt %s -test-vector-multi-reduction-lowering-patterns="use-outer-reductions" | FileCheck %s
 
 func @vector_multi_reduction(%arg0: vector<2x4xf32>) -> vector<2xf32> {
-    %0 = vector.multi_reduction #vector.kind<mul>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
+    %0 = vector.multi_reduction <mul>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
     return %0 : vector<2xf32>
 }
 
 // CHECK-LABEL: func @vector_multi_reduction
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<2x4xf32>
 //       CHECK:   %[[TRANSPOSED:.+]] = vector.transpose %[[INPUT]], [1, 0] : vector<2x4xf32> to vector<4x2xf32>
 //       CHECK:   %[[V0:.+]] = vector.extract %[[TRANSPOSED]][0] : vector<4x2xf32>
 //       CHECK:   %[[V1:.+]] = vector.extract %[[TRANSPOSED]][1] : vector<4x2xf32>
 //       CHECK:   %[[RV01:.+]] = arith.mulf %[[V1]], %[[V0]] : vector<2xf32>
 //       CHECK:   %[[V2:.+]] = vector.extract %[[TRANSPOSED]][2] : vector<4x2xf32>
 //       CHECK:   %[[RV012:.+]] = arith.mulf %[[V2]], %[[RV01]] : vector<2xf32>
 //       CHECK:   %[[V3:.+]] = vector.extract %[[TRANSPOSED]][3] : vector<4x2xf32>
 //       CHECK:   %[[RESULT_VEC:.+]] = arith.mulf %[[V3]], %[[RV012]] : vector<2xf32>
 //       CHECK:   return %[[RESULT_VEC]] : vector<2xf32>
 
 func @vector_multi_reduction_min(%arg0: vector<2x4xf32>) -> vector<2xf32> {
-    %0 = vector.multi_reduction #vector.kind<minf>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
+    %0 = vector.multi_reduction <minf>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
     return %0 : vector<2xf32>
 }
 
 // CHECK-LABEL: func @vector_multi_reduction_min
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<2x4xf32>
 //       CHECK:   %[[TRANSPOSED:.+]] = vector.transpose %[[INPUT]], [1, 0] : vector<2x4xf32> to vector<4x2xf32>
 //       CHECK:   %[[V0:.+]] = vector.extract %[[TRANSPOSED]][0] : vector<4x2xf32>
 //       CHECK:   %[[V1:.+]] = vector.extract %[[TRANSPOSED]][1] : vector<4x2xf32>
 //       CHECK:   %[[RV01:.+]] = arith.minf %[[V1]], %[[V0]] : vector<2xf32>
 //       CHECK:   %[[V2:.+]] = vector.extract %[[TRANSPOSED]][2] : vector<4x2xf32>
 //       CHECK:   %[[RV012:.+]] = arith.minf %[[V2]], %[[RV01]] : vector<2xf32>
 //       CHECK:   %[[V3:.+]] = vector.extract %[[TRANSPOSED]][3] : vector<4x2xf32>
 //       CHECK:   %[[RESULT_VEC:.+]] = arith.minf %[[V3]], %[[RV012]] : vector<2xf32>
 //       CHECK:   return %[[RESULT_VEC]] : vector<2xf32>
 
 func @vector_multi_reduction_max(%arg0: vector<2x4xf32>) -> vector<2xf32> {
-    %0 = vector.multi_reduction #vector.kind<maxf>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
+    %0 = vector.multi_reduction <maxf>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
     return %0 : vector<2xf32>
 }
 
 // CHECK-LABEL: func @vector_multi_reduction_max
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<2x4xf32>
 //       CHECK:   %[[TRANSPOSED:.+]] = vector.transpose %[[INPUT]], [1, 0] : vector<2x4xf32> to vector<4x2xf32>
 //       CHECK:   %[[V0:.+]] = vector.extract %[[TRANSPOSED]][0] : vector<4x2xf32>
 //       CHECK:   %[[V1:.+]] = vector.extract %[[TRANSPOSED]][1] : vector<4x2xf32>
 //       CHECK:   %[[RV01:.+]] = arith.maxf %[[V1]], %[[V0]] : vector<2xf32>
 //       CHECK:   %[[V2:.+]] = vector.extract %[[TRANSPOSED]][2] : vector<4x2xf32>
 //       CHECK:   %[[RV012:.+]] = arith.maxf %[[V2]], %[[RV01]] : vector<2xf32>
 //       CHECK:   %[[V3:.+]] = vector.extract %[[TRANSPOSED]][3] : vector<4x2xf32>
 //       CHECK:   %[[RESULT_VEC:.+]] = arith.maxf %[[V3]], %[[RV012]] : vector<2xf32>
 //       CHECK:   return %[[RESULT_VEC]] : vector<2xf32>
 
 func @vector_multi_reduction_and(%arg0: vector<2x4xi32>) -> vector<2xi32> {
-    %0 = vector.multi_reduction #vector.kind<and>, %arg0 [1] : vector<2x4xi32> to vector<2xi32>
+    %0 = vector.multi_reduction <and>, %arg0 [1] : vector<2x4xi32> to vector<2xi32>
     return %0 : vector<2xi32>
 }
 
 // CHECK-LABEL: func @vector_multi_reduction_and
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<2x4xi32>
 //       CHECK:   %[[TRANSPOSED:.+]] = vector.transpose %[[INPUT]], [1, 0] : vector<2x4xi32> to vector<4x2xi32>
 //       CHECK:   %[[V0:.+]] = vector.extract %[[TRANSPOSED]][0] : vector<4x2xi32>
 //       CHECK:   %[[V1:.+]] = vector.extract %[[TRANSPOSED]][1] : vector<4x2xi32>
 //       CHECK:   %[[RV01:.+]] = arith.andi %[[V1]], %[[V0]] : vector<2xi32>
 //       CHECK:   %[[V2:.+]] = vector.extract %[[TRANSPOSED]][2] : vector<4x2xi32>
 //       CHECK:   %[[RV012:.+]] = arith.andi %[[V2]], %[[RV01]] : vector<2xi32>
 //       CHECK:   %[[V3:.+]] = vector.extract %[[TRANSPOSED]][3] : vector<4x2xi32>
 //       CHECK:   %[[RESULT_VEC:.+]] = arith.andi %[[V3]], %[[RV012]] : vector<2xi32>
 //       CHECK:   return %[[RESULT_VEC]] : vector<2xi32>
 
 func @vector_multi_reduction_or(%arg0: vector<2x4xi32>) -> vector<2xi32> {
-    %0 = vector.multi_reduction #vector.kind<or>, %arg0 [1] : vector<2x4xi32> to vector<2xi32>
+    %0 = vector.multi_reduction <or>, %arg0 [1] : vector<2x4xi32> to vector<2xi32>
     return %0 : vector<2xi32>
 }
 
 // CHECK-LABEL: func @vector_multi_reduction_or
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<2x4xi32>
 //       CHECK:   %[[TRANSPOSED:.+]] = vector.transpose %[[INPUT]], [1, 0] : vector<2x4xi32> to vector<4x2xi32>
 //       CHECK:   %[[V0:.+]] = vector.extract %[[TRANSPOSED]][0] : vector<4x2xi32>
 //       CHECK:   %[[V1:.+]] = vector.extract %[[TRANSPOSED]][1] : vector<4x2xi32>
 //       CHECK:   %[[RV01:.+]] = arith.ori %[[V1]], %[[V0]] : vector<2xi32>
 //       CHECK:   %[[V2:.+]] = vector.extract %[[TRANSPOSED]][2] : vector<4x2xi32>
 //       CHECK:   %[[RV012:.+]] = arith.ori %[[V2]], %[[RV01]] : vector<2xi32>
 //       CHECK:   %[[V3:.+]] = vector.extract %[[TRANSPOSED]][3] : vector<4x2xi32>
 //       CHECK:   %[[RESULT_VEC:.+]] = arith.ori %[[V3]], %[[RV012]] : vector<2xi32>
 //       CHECK:   return %[[RESULT_VEC]] : vector<2xi32>
 
 func @vector_multi_reduction_xor(%arg0: vector<2x4xi32>) -> vector<2xi32> {
-    %0 = vector.multi_reduction #vector.kind<xor>, %arg0 [1] : vector<2x4xi32> to vector<2xi32>
+    %0 = vector.multi_reduction <xor>, %arg0 [1] : vector<2x4xi32> to vector<2xi32>
     return %0 : vector<2xi32>
 }
 
 // CHECK-LABEL: func @vector_multi_reduction_xor
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<2x4xi32>
 //       CHECK:   %[[TRANSPOSED:.+]] = vector.transpose %[[INPUT]], [1, 0] : vector<2x4xi32> to vector<4x2xi32>
 //       CHECK:   %[[V0:.+]] = vector.extract %[[TRANSPOSED]][0] : vector<4x2xi32>
 //       CHECK:   %[[V1:.+]] = vector.extract %[[TRANSPOSED]][1] : vector<4x2xi32>
 //       CHECK:   %[[RV01:.+]] = arith.xori %[[V1]], %[[V0]] : vector<2xi32>
 //       CHECK:   %[[V2:.+]] = vector.extract %[[TRANSPOSED]][2] : vector<4x2xi32>
 //       CHECK:   %[[RV012:.+]] = arith.xori %[[V2]], %[[RV01]] : vector<2xi32>
 //       CHECK:   %[[V3:.+]] = vector.extract %[[TRANSPOSED]][3] : vector<4x2xi32>
 //       CHECK:   %[[RESULT_VEC:.+]] = arith.xori %[[V3]], %[[RV012]] : vector<2xi32>
 //       CHECK:   return %[[RESULT_VEC]] : vector<2xi32>
 
 
 func @vector_reduction_outer(%arg0: vector<2x3x4x5xi32>) -> vector<2x3xi32> {
-    %0 = vector.multi_reduction #vector.kind<add>, %arg0 [2, 3] : vector<2x3x4x5xi32> to vector<2x3xi32>
+    %0 = vector.multi_reduction <add>, %arg0 [2, 3] : vector<2x3x4x5xi32> to vector<2x3xi32>
     return %0 : vector<2x3xi32>
 }
 
 // CHECK-LABEL: func @vector_reduction_outer
 //  CHECK-SAME:   %[[INPUT:.+]]: vector<2x3x4x5xi32>
 //       CHECK:   %[[TRANSPOSED:.+]] = vector.transpose %[[INPUT]], [2, 3, 0, 1] : vector<2x3x4x5xi32> to vector<4x5x2x3xi32>
 //       CHECK:   %[[RESHAPED:.+]] = vector.shape_cast %[[TRANSPOSED]] : vector<4x5x2x3xi32> to vector<20x6xi32>
 //       CHECK:   %[[V0:.+]] = vector.extract %[[RESHAPED]][0] : vector<20x6xi32>

mlir/test/Dialect/Vector/vector-reduce-to-contract.mlir

 // RUN: mlir-opt %s -test-vector-reduction-to-contract-patterns -split-input-file | FileCheck %s
 
 // CHECK-DAG: #[[$map0:.*]] = affine_map<(d0, d1, d2) -> (d0, d1, d2)>
 // CHECK-DAG: #[[$map1:.*]] = affine_map<(d0, d1, d2) -> (d0, d2)>
 
 // CHECK-LABEL: multidimreduction_contract
 //  CHECK-NEXT:   %[[C0:.+]] = arith.constant dense<0.000000e+00> : vector<8x16xf32>
 //  CHECK-NEXT:   %[[R:.+]] = vector.contract {indexing_maps = [#[[$map0]], #[[$map0]], #[[$map1]]],
 //  CHECK-SAME:   iterator_types = ["parallel", "reduction", "parallel"], kind = #vector.kind<add>}
 //  CHECK-SAME:   %{{.*}}, %{{.*}}, %[[C0]] : vector<8x32x16xf32>, vector<8x32x16xf32> into vector<8x16xf32>
 //  CHECK-NEXT:   return %[[R]] : vector<8x16xf32>
 func @multidimreduction_contract(
   %arg0: vector<8x32x16xf32>,%arg1: vector<8x32x16xf32>) -> vector<8x16xf32> {
   %0 = arith.mulf %arg0, %arg1 : vector<8x32x16xf32>
-  %1 = vector.multi_reduction #vector.kind<add>, %0 [1] : vector<8x32x16xf32> to vector<8x16xf32>
+  %1 = vector.multi_reduction <add>, %0 [1] : vector<8x32x16xf32> to vector<8x16xf32>
   return %1 : vector<8x16xf32>
 }
 
 // -----
 
 // CHECK-DAG: #[[$map0:.*]] = affine_map<(d0, d1, d2) -> (d0, d1, d2)>
 // CHECK-DAG: #[[$map1:.*]] = affine_map<(d0, d1, d2) -> (d0, d2)>
 
 // CHECK-LABEL: multidimreduction_contract_int
 //  CHECK-NEXT:   %[[C0:.+]] = arith.constant dense<0> : vector<8x16xi32>
 //  CHECK-NEXT:   %[[R:.+]] = vector.contract {indexing_maps = [#[[$map0]], #[[$map0]], #[[$map1]]],
 //  CHECK-SAME:   iterator_types = ["parallel", "reduction", "parallel"], kind = #vector.kind<add>}
 //  CHECK-SAME:   %{{.*}}, %{{.*}}, %[[C0]] : vector<8x32x16xi32>, vector<8x32x16xi32> into vector<8x16xi32>
 //  CHECK-NEXT:   return %[[R]] : vector<8x16xi32>
 func @multidimreduction_contract_int(
   %arg0: vector<8x32x16xi32>,%arg1: vector<8x32x16xi32>) -> vector<8x16xi32> {
   %0 = arith.muli %arg0, %arg1 : vector<8x32x16xi32>
-  %1 = vector.multi_reduction #vector.kind<add>, %0 [1] : vector<8x32x16xi32> to vector<8x16xi32>
+  %1 = vector.multi_reduction <add>, %0 [1] : vector<8x32x16xi32> to vector<8x16xi32>
   return %1 : vector<8x16xi32>
 }
 
 // -----
 
 #map0 = affine_map<(d0, d1, d2) -> (d0, d1, d2)>
 #map1 = affine_map<(d0, d1, d2) -> (d0, d1)>
 

mlir/test/lib/Dialect/Test/TestAttrDefs.td

 //===-- TestAttrDefs.td - Test dialect attr definitions ----*- tablegen -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // TableGen data attribute definitions for Test dialect.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef TEST_ATTRDEFS
 #define TEST_ATTRDEFS
 
 // To get the test dialect definition.
-include "TestOps.td"
+include "TestDialect.td"
 include "mlir/IR/BuiltinAttributeInterfaces.td"
 include "mlir/IR/SubElementInterfaces.td"
 
 // All of the attributes will extend this class.
 class Test_Attr<string name, list<Trait> traits = []>
     : AttrDef<Test_Dialect, name, traits>;
 
 def SimpleAttrA : Test_Attr<"SimpleA"> {
def TestSubElementsAccessAttr : Test_Attr<"TestSubElementsAccess", [
 
   let parameters = (ins
     "::mlir::Attribute":$first,
     "::mlir::Attribute":$second,
     "::mlir::Attribute":$third
   );
 }
 
+// A more complex parameterized attribute with multiple level of nesting.
+def CompoundNestedInner : Test_Attr<"CompoundNestedInner"> {
+  let mnemonic = "cmpnd_nested_inner";
+  // List of type parameters.
+  let parameters = (
+    ins
+    "int":$some_int,
+    CompoundAttrA:$cmpdA
+  );
+  let assemblyFormat = "`<` $some_int $cmpdA `>`";
+}
+
+def CompoundNestedOuter : Test_Attr<"CompoundNestedOuter"> {
+  let mnemonic = "cmpnd_nested_outer";
+
+  // List of type parameters.
+  let parameters = (
+    ins
+    CompoundNestedInner:$inner
+  );
+  let assemblyFormat = "`<` `i`  $inner `>`";
+}
+
 def TestParamOne : AttrParameter<"int64_t", ""> {}
 
 def TestParamTwo : AttrParameter<"std::string", "", "llvm::StringRef"> {
   let printer = "$_printer << '\"' << $_self << '\"'";
 }
 
 def TestParamFour : ArrayRefParameter<"int", ""> {
   let cppStorageType = "llvm::SmallVector<int>";

mlir/test/lib/Dialect/Test/TestDialect.td

+//===-- TestDialect.td - Test dialect definition -----------*- tablegen -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef TEST_DIALECT
+#define TEST_DIALECT
+
+include "mlir/IR/OpBase.td"
+
+def Test_Dialect : Dialect {
+  let name = "test";
+  let cppNamespace = "::test";
+  let emitAccessorPrefix = kEmitAccessorPrefix_Prefixed;
+  let hasCanonicalizer = 1;
+  let hasConstantMaterializer = 1;
+  let hasOperationAttrVerify = 1;
+  let hasRegionArgAttrVerify = 1;
+  let hasRegionResultAttrVerify = 1;
+  let hasOperationInterfaceFallback = 1;
+  let hasNonDefaultDestructor = 1;
+  let useDefaultAttributePrinterParser = 1;
+  let dependentDialects = ["::mlir::DLTIDialect"];
+
+  let extraClassDeclaration = [{
+    void registerAttributes();
+    void registerTypes();
+
+    // Provides a custom printing/parsing for some operations.
+    ::llvm::Optional<ParseOpHook>
+      getParseOperationHook(::llvm::StringRef opName) const override;
+    ::llvm::unique_function<void(::mlir::Operation *,
+                                 ::mlir::OpAsmPrinter &printer)>
+     getOperationPrinter(::mlir::Operation *op) const override;
+
+  private:
+    // Storage for a custom fallback interface.
+    void *fallbackEffectOpInterfaces;
+
+  }];
+}
+
+#endif // TEST_DIALECT

mlir/test/lib/Dialect/Test/TestOps.td

 //===-- TestOps.td - Test dialect operation definitions ----*- tablegen -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef TEST_OPS
 #define TEST_OPS
 
+include "TestDialect.td"
 include "mlir/Dialect/DLTI/DLTIBase.td"
 include "mlir/IR/OpBase.td"
 include "mlir/IR/OpAsmInterface.td"
 include "mlir/IR/RegionKindInterface.td"
 include "mlir/IR/SymbolInterfaces.td"
 include "mlir/Interfaces/CallInterfaces.td"
 include "mlir/Interfaces/ControlFlowInterfaces.td"
 include "mlir/Interfaces/CopyOpInterface.td"
 include "mlir/Interfaces/DataLayoutInterfaces.td"
 include "mlir/Interfaces/InferTypeOpInterface.td"
 include "mlir/Interfaces/SideEffectInterfaces.td"
 include "mlir/Dialect/Linalg/IR/LinalgInterfaces.td"
 include "TestInterfaces.td"
 
-def Test_Dialect : Dialect {
-  let name = "test";
-  let cppNamespace = "::test";
-  let emitAccessorPrefix = kEmitAccessorPrefix_Prefixed;
-  let hasCanonicalizer = 1;
-  let hasConstantMaterializer = 1;
-  let hasOperationAttrVerify = 1;
-  let hasRegionArgAttrVerify = 1;
-  let hasRegionResultAttrVerify = 1;
-  let hasOperationInterfaceFallback = 1;
-  let hasNonDefaultDestructor = 1;
-  let useDefaultAttributePrinterParser = 1;
-  let dependentDialects = ["::mlir::DLTIDialect"];
-
-  let extraClassDeclaration = [{
-    void registerAttributes();
-    void registerTypes();
-
-    // Provides a custom printing/parsing for some operations.
-    ::llvm::Optional<ParseOpHook>
-      getParseOperationHook(::llvm::StringRef opName) const override;
-    ::llvm::unique_function<void(::mlir::Operation *,
-                                 ::mlir::OpAsmPrinter &printer)>
-     getOperationPrinter(::mlir::Operation *op) const override;
-
-  private:
-    // Storage for a custom fallback interface.
-    void *fallbackEffectOpInterfaces;
-
-  }];
-}
 
 // Include the attribute definitions.
 include "TestAttrDefs.td"
+// Include the type definitions.
+include "TestTypeDefs.td"
 
 
 class TEST_Op<string mnemonic, list<OpTrait> traits = []> :
     Op<Test_Dialect, mnemonic, traits>;
 
 //===----------------------------------------------------------------------===//
 // Test Types
 //===----------------------------------------------------------------------===//
def FormatCompoundAttr : TEST_Op<"format_compound_attr"> {
   let assemblyFormat = "$compound attr-dict-with-keyword";
 }
 
 def FormatNestedAttr : TEST_Op<"format_nested_attr"> {
   let arguments = (ins CompoundAttrNested:$nested);
   let assemblyFormat = "$nested attr-dict-with-keyword";
 }
 
+def FormatNestedCompoundAttr : TEST_Op<"format_cpmd_nested_attr"> {
+  let arguments = (ins CompoundNestedOuter:$nested);
+  let assemblyFormat = "`nested` $nested attr-dict-with-keyword";
+}
+
+def FormatNestedType : TEST_Op<"format_cpmd_nested_type"> {
+  let arguments = (ins CompoundNestedOuterType:$nested);
+  let assemblyFormat = "$nested `nested` type($nested) attr-dict-with-keyword";
+}
+
 //===----------------------------------------------------------------------===//
 // Custom Directives
 
 def FormatCustomDirectiveOperands
     : TEST_Op<"format_custom_directive_operands", [AttrSizedOperandSegments]> {
   let arguments = (ins I64:$operand, Optional<I64>:$optOperand,
                        Variadic<I64>:$varOperands);
   let assemblyFormat = [{

mlir/test/lib/Dialect/Test/TestTypeDefs.td

 //===-- TestTypeDefs.td - Test dialect type definitions ----*- tablegen -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // TableGen data type definitions for Test dialect.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef TEST_TYPEDEFS
 #define TEST_TYPEDEFS
 
 // To get the test dialect def.
-include "TestOps.td"
+include "TestDialect.td"
 include "TestAttrDefs.td"
+include "TestInterfaces.td"
 include "mlir/IR/BuiltinTypes.td"
 include "mlir/Interfaces/DataLayoutInterfaces.td"
 
 // All of the types will extend this class.
 class Test_Type<string name, list<Trait> traits = []>
   : TypeDef<Test_Dialect, name, traits>;
 
 def SimpleTypeA : Test_Type<"SimpleA"> {
ArrayRefParameter<
       >: $arrayOfInts
   );
 
   let extraClassDeclaration = [{
     struct SomeCppStruct {};
   }];
 }
 
+// A more complex and nested parameterized type.
+def CompoundNestedInnerType : Test_Type<"CompoundNestedInner"> {
+  let mnemonic = "cmpnd_inner";
+  // List of type parameters.
+  let parameters = (
+    ins
+    "int":$some_int,
+    CompoundTypeA:$cmpdA
+  );
+  let assemblyFormat = "`<` $some_int $cmpdA `>`";
+}
+
+def CompoundNestedOuterType : Test_Type<"CompoundNestedOuter"> {
+  let mnemonic = "cmpnd_nested_outer";
+
+  // List of type parameters.
+  let parameters = (
+    ins
+    CompoundNestedInnerType:$inner
+  );
+  let assemblyFormat = "`<` `i`  $inner `>`";
+}
+
 // An example of how one could implement a standard integer.
 def IntegerType : Test_Type<"TestInteger"> {
   let mnemonic = "int";
   let genVerifyDecl = 1;
   let parameters = (
     ins
     "unsigned":$width,
     // SignednessSemantics is defined below.

mlir/test/lib/Dialect/Test/TestTypes.h

 #include "mlir/IR/Diagnostics.h"
 #include "mlir/IR/Dialect.h"
 #include "mlir/IR/DialectImplementation.h"
 #include "mlir/IR/Operation.h"
 #include "mlir/IR/Types.h"
 #include "mlir/Interfaces/DataLayoutInterfaces.h"
 
 namespace test {
+class TestAttrWithFormatAttr;
 
 /// FieldInfo represents a field in the StructType data type. It is used as a
 /// parameter in TestTypeDefs.td.
 struct FieldInfo {
   ::llvm::StringRef name;
   ::mlir::Type type;
 
   // Custom allocation called from generated constructor code

 struct FieldParser<test::CustomParam> {
   static FailureOr<test::CustomParam> parse(AsmParser &parser) {
     auto value = FieldParser<int>::parse(parser);
     if (failed(value))
       return failure();
     return test::CustomParam{value.getValue()};
   }
 };
-} // end namespace mlir
-
 inline mlir::AsmPrinter &operator<<(mlir::AsmPrinter &printer,
-                                    const test::CustomParam &param) {
+                                    test::CustomParam param) {
   return printer << param.value;
 }
 
+} // end namespace mlir
+
 #include "TestTypeInterfaces.h.inc"
 
 #define GET_TYPEDEF_CLASSES
 #include "TestTypeDefs.h.inc"
 
 namespace test {
 
 /// Storage for simple named recursive types, where the type is identified by

mlir/test/mlir-tblgen/attr-or-type-format.td

 // ATTR:                         _result_value.getValue(),
 // ATTR:                         _result_complex.getValue());
 // ATTR: }
 
 // ATTR: void TestAAttr::print(::mlir::AsmPrinter &printer) const {
 // ATTR:   printer << ' ' << "hello";
 // ATTR:   printer << ' ' << "=";
 // ATTR:   printer << ' ';
-// ATTR:   printer << getValue();
+// ATTR:   printer.printStrippedAttrOrType(getValue());
 // ATTR:   printer << ",";
 // ATTR:   printer << ' ';
 // ATTR:   ::printAttrParamA(printer, getComplex());
 // ATTR:   printer << ")";
 // ATTR: }
 
 def AttrA : TestAttr<"TestA"> {
   let parameters = (ins

 // ATTR:     return {};
 // ATTR:   return TestFAttr::get(parser.getContext(),
 // ATTR:     _result_v0.getValue(),
 // ATTR:     _result_v1.getValue());
 // ATTR: }
 
 // ATTR: void TestFAttr::print(::mlir::AsmPrinter &printer) const {
 // ATTR:   printer << ' ';
-// ATTR:   printer << getV0();
+// ATTR:   printer.printStrippedAttrOrType(getV0());
 // ATTR:   printer << ",";
 // ATTR:   printer << ' ';
-// ATTR:   printer << getV1();
+// ATTR:   printer.printStrippedAttrOrType(getV1());
 // ATTR: }
 
 def AttrC : TestAttr<"TestF"> {
   let parameters = (ins "int":$v0, "int":$v1);
 
   let mnemonic = "attr_c";
   let assemblyFormat = "params";
 }

 
 // TYPE: void TestCType::print(::mlir::AsmPrinter &printer) const {
 // TYPE:   printer << ' ' << "foo";
 // TYPE:   printer << ",";
 // TYPE:   printer << ' ' << ":";
 // TYPE:   printer << ' ' << "bob";
 // TYPE:   printer << ' ' << "bar";
 // TYPE:   printer << ' ';
-// TYPE:   printer << getValue();
+// TYPE:   printer.printStrippedAttrOrType(getValue());
 // TYPE:   printer << ' ' << "complex";
 // TYPE:   printer << ' ' << "=";
 // TYPE:   printer << ' ';
 // TYPE:   printer << getComplex();
 // TYPE:   printer << ")";
 // TYPE: }
 
 def TypeA : TestType<"TestC"> {

 // TYPE:     _result_v2.getValue(),
 // TYPE:     _result_v3.getValue());
 // TYPE: }
 
 // TYPE: void TestEType::print(::mlir::AsmPrinter &printer) const {
 // TYPE:   printer << "v0";
 // TYPE:   printer << ' ' << "=";
 // TYPE:   printer << ' ';
-// TYPE:   printer << getV0();
+// TYPE:   printer.printStrippedAttrOrType(getV0());
 // TYPE:   printer << ",";
 // TYPE:   printer << ' ' << "v2";
 // TYPE:   printer << ' ' << "=";
 // TYPE:   printer << ' ';
-// TYPE:   printer << getV2();
+// TYPE:   printer.printStrippedAttrOrType(getV2());
 // TYPE:   printer << "v1";
 // TYPE:   printer << ' ' << "=";
 // TYPE:   printer << ' ';
-// TYPE:   printer << getV1();
+// TYPE:   printer.printStrippedAttrOrType(getV1());
 // TYPE:   printer << ",";
 // TYPE:   printer << ' ' << "v3";
 // TYPE:   printer << ' ' << "=";
 // TYPE:   printer << ' ';
-// TYPE:   printer << getV3();
+// TYPE:   printer.printStrippedAttrOrType(getV3());
 // TYPE: }
 
 def TypeC : TestType<"TestE"> {
   let parameters = (ins
       "IntegerAttr":$v0,
       "IntegerAttr":$v1,
       "IntegerAttr":$v2,
       "IntegerAttr":$v3
   );
 
   let mnemonic = "type_e";
   let assemblyFormat = "`{` struct($v0, $v2) `}` `{` struct($v1, $v3) `}`";
 }

mlir/test/mlir-tblgen/op-format.mlir

 
 // CHECK: test.format_optional_else else
 test.format_optional_else else
 
 //===----------------------------------------------------------------------===//
 // Format a custom attribute
 //===----------------------------------------------------------------------===//
 
-// CHECK: test.format_compound_attr #test.cmpnd_a<1, !test.smpla, [5, 6]>
-test.format_compound_attr #test.cmpnd_a<1, !test.smpla, [5, 6]>
+// CHECK: test.format_compound_attr <1, !test.smpla, [5, 6]>
+test.format_compound_attr <1, !test.smpla, [5, 6]>
 
-// CHECK:  module attributes {test.nested = #test.cmpnd_nested<nested = #test.cmpnd_a<1, !test.smpla, [5, 6]>>} {
+//-----
+
+
+// CHECK:   module attributes {test.nested = #test.cmpnd_nested<nested = <1, !test.smpla, [5, 6]>>} {
+module attributes {test.nested = #test.cmpnd_nested<nested = <1, !test.smpla, [5, 6]>>} {
+}
+
+//-----
+
+// Same as above, but fully spelling the inner attribute prefix `#test.cmpnd_a`.
+// CHECK:   module attributes {test.nested = #test.cmpnd_nested<nested = <1, !test.smpla, [5, 6]>>} {
 module attributes {test.nested = #test.cmpnd_nested<nested = #test.cmpnd_a<1, !test.smpla, [5, 6]>>} {
 }
 
-// CHECK: test.format_nested_attr #test.cmpnd_nested<nested = #test.cmpnd_a<1, !test.smpla, [5, 6]>>
+// CHECK: test.format_nested_attr <nested = <1, !test.smpla, [5, 6]>>
+test.format_nested_attr #test.cmpnd_nested<nested = <1, !test.smpla, [5, 6]>>
+
+//-----
+
+// Same as above, but fully spelling the inner attribute prefix `#test.cmpnd_a`.
+// CHECK: test.format_nested_attr <nested = <1, !test.smpla, [5, 6]>>
 test.format_nested_attr #test.cmpnd_nested<nested = #test.cmpnd_a<1, !test.smpla, [5, 6]>>
 
+//-----
+
+// CHECK: module attributes {test.someAttr = #test.cmpnd_nested_inner<42 <1, !test.smpla, [5, 6]>>}
+module attributes {test.someAttr = #test.cmpnd_nested_inner<42 <1, !test.smpla, [5, 6]>>}
+{
+}
+
+//-----
+
+// CHECK: module attributes {test.someAttr = #test.cmpnd_nested_outer<i <42 <1, !test.smpla, [5, 6]>>>}
+module attributes {test.someAttr = #test.cmpnd_nested_outer<i <42 <1, !test.smpla, [5, 6]>>>}
+{
+}
+
+//-----
+
+// CHECK: test.format_cpmd_nested_attr nested <i <42 <1, !test.smpla, [5, 6]>>>
+test.format_cpmd_nested_attr nested <i <42 <1, !test.smpla, [5, 6]>>>
 
 //===----------------------------------------------------------------------===//
 // Format custom directives
 //===----------------------------------------------------------------------===//
 
 // CHECK: test.format_custom_directive_operands %[[I64]], %[[I64]] -> (%[[I64]])
 test.format_custom_directive_operands %i64, %i64 -> (%i64)
 

mlir/test/mlir-tblgen/testdialect-typedefs.mlir

 // RUN: mlir-opt %s | mlir-opt -verify-diagnostics | FileCheck %s
 
 //////////////
 // Tests the types in the 'Test' dialect, not the ones in 'typedefs.mlir'
 
 // CHECK: @simpleA(%arg0: !test.smpla)
 func @simpleA(%A : !test.smpla) -> () {
   return
 }
 
 // CHECK: @compoundA(%arg0: !test.cmpnd_a<1, !test.smpla, [5, 6]>)
 func @compoundA(%A : !test.cmpnd_a<1, !test.smpla, [5, 6]>)-> () {
   return
 }
 
+// CHECK: @compoundNested(%arg0: !test.cmpnd_nested_outer<i <42 <1, !test.smpla, [5, 6]>>>)
+func @compoundNested(%arg0: !test.cmpnd_nested_outer<i <42 <1, !test.smpla, [5, 6]>>>) -> () {
+  return
+}
+
+// Same as above, but we're parsing the complete spec for the inner type
+// CHECK: @compoundNestedExplicit(%arg0: !test.cmpnd_nested_outer<i <42 <1, !test.smpla, [5, 6]>>>)
+func @compoundNestedExplicit(%arg0: !test.cmpnd_nested_outer<i !test.cmpnd_inner<42 <1, !test.smpla, [5, 6]>>>) -> () {
+// Verify that the type prefix is elided and optional
+// CHECK: format_cpmd_nested_type %arg0 nested <i <42 <1, !test.smpla, [5, 6]>>>
+// CHECK: format_cpmd_nested_type %arg0 nested <i <42 <1, !test.smpla, [5, 6]>>>
+  test.format_cpmd_nested_type %arg0 nested !test.cmpnd_nested_outer<i !test.cmpnd_inner<42 <1, !test.smpla, [5, 6]>>>
+  test.format_cpmd_nested_type %arg0 nested <i <42 <1, !test.smpla, [5, 6]>>>
+  return
+}
+
 // CHECK: @testInt(%arg0: !test.int<signed, 8>, %arg1: !test.int<unsigned, 2>, %arg2: !test.int<none, 1>)
 func @testInt(%A : !test.int<s, 8>, %B : !test.int<unsigned, 2>, %C : !test.int<n, 1>) {
   return
 }
 
 // CHECK: @structTest(%arg0: !test.struct<{field1,!test.smpla},{field2,!test.int<none, 3>}>)
 func @structTest (%A : !test.struct< {field1, !test.smpla}, {field2, !test.int<none, 3>} > ) {
   return
 }

mlir/tools/mlir-tblgen/AttrOrTypeFormatGen.cpp

 // Format Strings
 //===----------------------------------------------------------------------===//
 
 /// Default parser for attribute or type parameters.
 static const char *const defaultParameterParser =
     "::mlir::FieldParser<$0>::parse($_parser)";
 
 /// Default printer for attribute or type parameters.
-static const char *const defaultParameterPrinter = "$_printer << $_self";
+static const char *const defaultParameterPrinter =
+    "$_printer.printStrippedAttrOrType($_self)";
 
 /// Print an error when failing to parse an element.
 ///
 /// $0: The parameter C++ class name.
 static const char *const parseErrorStr =
     "$_parser.emitError($_parser.getCurrentLocation(), ";
 
 /// Loop declaration for struct parser.

mlir/tools/mlir-tblgen/OpFormatGen.cpp

static bool shouldFormatSymbolNameAttr(const NamedAttribute *attr) {
   return attr->attr.getBaseAttr().getAttrDefName() == "SymbolNameAttr";
 }
 
 /// The code snippet used to generate a parser call for an attribute.
 ///
 /// {0}: The name of the attribute.
 /// {1}: The type for the attribute.
 const char *const attrParserCode = R"(
-  if (parser.parseAttribute({0}Attr{1}, "{0}", result.attributes))
+  if (parser.parseCustomAttributeWithFallback({0}Attr, {1}, "{0}",
+          result.attributes)) {{
     return ::mlir::failure();
+  }
 )";
+
+/// The code snippet used to generate a parser call for an attribute.
+///
+/// {0}: The name of the attribute.
+/// {1}: The type for the attribute.
+const char *const genericAttrParserCode = R"(
+  if (parser.parseAttribute({0}Attr, {1}, "{0}", result.attributes))
+    return ::mlir::failure();
+)";
+
 const char *const optionalAttrParserCode = R"(
   {
     ::mlir::OptionalParseResult parseResult =
-      parser.parseOptionalAttribute({0}Attr{1}, "{0}", result.attributes);
+      parser.parseOptionalAttribute({0}Attr, {1}, "{0}", result.attributes);
     if (parseResult.hasValue() && failed(*parseResult))
       return ::mlir::failure();
   }
 )";
 
 /// The code snippet used to generate a parser call for a symbol name attribute.
 ///
 /// {0}: The name of the attribute.
const char *const optionalTypeParserCode = R"(
     if (parseResult.hasValue()) {
       if (failed(*parseResult))
         return ::mlir::failure();
       {0}Types.push_back(optionalType);
     }
   }
 )";
 const char *const typeParserCode = R"(
-  if (parser.parseType({0}RawTypes[0]))
+  {
+    {0} type;
+    if (parser.parseCustomTypeWithFallback(type))
       return ::mlir::failure();
+    {1}RawTypes[0] = type;
+  }
 )";
 
 /// The code snippet used to generate a parser call for a functional type.
 ///
 /// {0}: The name for the input type list.
 /// {1}: The name for the result type list.
 const char *const functionalTypeParserCode = R"(
   ::mlir::FunctionType {0}__{1}_functionType;
if (shouldFormatSymbolNameAttr(var)) {
       return;
     }
 
     // If this attribute has a buildable type, use that when parsing the
     // attribute.
     std::string attrTypeStr;
     if (Optional<StringRef> typeBuilder = attr->getTypeBuilder()) {
       llvm::raw_string_ostream os(attrTypeStr);
-      os << ", " << tgfmt(*typeBuilder, &attrTypeCtx);
+      os << tgfmt(*typeBuilder, &attrTypeCtx);
+    } else {
+      attrTypeStr = "Type{}";
+    }
+    if (var->attr.isOptional()) {
+      body << formatv(optionalAttrParserCode, var->name, attrTypeStr);
+    } else {
+      if (var->attr.getStorageType() == "::mlir::Attribute")
+        body << formatv(genericAttrParserCode, var->name, attrTypeStr);
+      else
+        body << formatv(attrParserCode, var->name, attrTypeStr);
     }
 
-    body << formatv(var->attr.isOptional() ? optionalAttrParserCode
-                                           : attrParserCode,
-                    var->name, attrTypeStr);
   } else if (auto *operand = dyn_cast<OperandVariable>(element)) {
     ArgumentLengthKind lengthKind = getArgumentLengthKind(operand->getVar());
     StringRef name = operand->getVar()->name;
     if (lengthKind == ArgumentLengthKind::VariadicOfVariadic)
       body << llvm::formatv(
           variadicOfVariadicOperandParserCode, name,
           operand->getVar()->constraint.getVariadicOfVariadicSegmentSizeAttr());
     else if (lengthKind == ArgumentLengthKind::Variadic)
else if (hasSingleBlockTrait)
       body << llvm::formatv(regionListEnsureSingleBlockParserCode, "full");
 
   } else if (isa<SuccessorsDirective>(element)) {
     body << llvm::formatv(successorListParserCode, "full");
 
   } else if (auto *dir = dyn_cast<TypeDirective>(element)) {
     ArgumentLengthKind lengthKind;
     StringRef listName = getTypeListName(dir->getOperand(), lengthKind);
-    if (lengthKind == ArgumentLengthKind::VariadicOfVariadic)
+    if (lengthKind == ArgumentLengthKind::VariadicOfVariadic) {
       body << llvm::formatv(variadicOfVariadicTypeParserCode, listName);
-    else if (lengthKind == ArgumentLengthKind::Variadic)
+    } else if (lengthKind == ArgumentLengthKind::Variadic) {
       body << llvm::formatv(variadicTypeParserCode, listName);
-    else if (lengthKind == ArgumentLengthKind::Optional)
+    } else if (lengthKind == ArgumentLengthKind::Optional) {
       body << llvm::formatv(optionalTypeParserCode, listName);
-    else
-      body << formatv(typeParserCode, listName);
+    } else {
+      TypeSwitch<Element *>(dir->getOperand())
+          .Case<OperandVariable, ResultVariable>([&](auto operand) {
+            body << formatv(typeParserCode,
+                            operand->getVar()->constraint.getCPPClassName(),
+                            listName);
+          })
+          .Default([&](auto operand) {
+            body << formatv(typeParserCode, "Type", listName);
+          });
+    }
   } else if (auto *dir = dyn_cast<FunctionalTypeDirective>(element)) {
     ArgumentLengthKind ignored;
     body << formatv(functionalTypeParserCode,
                     getTypeListName(dir->getInputs(), ignored),
                     getTypeListName(dir->getResults(), ignored));
   } else {
     llvm_unreachable("unknown format element");
   }
static void genVariadicRegionPrinter(const Twine &regionListName,
   body << "    llvm::interleaveComma(" << regionListName
        << ", _odsPrinter, [&](::mlir::Region &region) {\n      ";
   genRegionPrinter("region", body, hasImplicitTermTrait);
   body << "    });\n";
 }
 
 /// Generate the C++ for an operand to a (*-)type directive.
 static MethodBody &genTypeOperandPrinter(Element *arg, const Operator &op,
-                                         MethodBody &body) {
+                                         MethodBody &body,
+                                         bool useArrayRef = true) {
   if (isa<OperandsDirective>(arg))
     return body << "getOperation()->getOperandTypes()";
   if (isa<ResultsDirective>(arg))
     return body << "getOperation()->getResultTypes()";
   auto *operand = dyn_cast<OperandVariable>(arg);
   auto *var = operand ? operand->getVar() : cast<ResultVariable>(arg)->getVar();
   if (var->isVariadicOfVariadic())
     return body << llvm::formatv("{0}().join().getTypes()",
                                  op.getGetterName(var->name));
   if (var->isVariadic())
     return body << op.getGetterName(var->name) << "().getTypes()";
   if (var->isOptional())
     return body << llvm::formatv(
                "({0}() ? ::llvm::ArrayRef<::mlir::Type>({0}().getType()) : "
                "::llvm::ArrayRef<::mlir::Type>())",
                op.getGetterName(var->name));
+  if (useArrayRef)
     return body << "::llvm::ArrayRef<::mlir::Type>("
                 << op.getGetterName(var->name) << "().getType())";
+  return body << op.getGetterName(var->name) << "().getType()";
 }
 
 /// Generate the printer for an enum attribute.
 static void genEnumAttrPrinter(const NamedAttribute *var, const Operator &op,
                                MethodBody &body) {
   Attribute baseAttr = var->attr.getBaseAttr();
   const EnumAttr &enumAttr = cast<EnumAttr>(baseAttr);
   std::vector<EnumAttrCase> cases = enumAttr.getAllCases();
if (shouldFormatSymbolNameAttr(var)) {
            << "Attr().getValue());\n";
       return;
     }
 
     // Elide the attribute type if it is buildable.
     if (attr->getTypeBuilder())
       body << "  _odsPrinter.printAttributeWithoutType("
            << op.getGetterName(var->name) << "Attr());\n";
-    else
-      body << "  _odsPrinter.printAttribute(" << op.getGetterName(var->name)
+    else if (var->attr.isOptional())
+      body << "_odsPrinter.printAttribute(" << op.getGetterName(var->name)
            << "Attr());\n";
+    else if (var->attr.getStorageType() == "::mlir::Attribute")
+      body << "  _odsPrinter.printAttribute(" << op.getGetterName(var->name)
+           << "Attr());\n";
+    else
+      body << "_odsPrinter.printStrippedAttrOrType("
+           << op.getGetterName(var->name) << "Attr());\n";
   } else if (auto *operand = dyn_cast<OperandVariable>(element)) {
     if (operand->getVar()->isVariadicOfVariadic()) {
       body << "  ::llvm::interleaveComma("
            << op.getGetterName(operand->getVar()->name)
            << "(), _odsPrinter, [&](const auto &operands) { _odsPrinter << "
               "\"(\" << operands << "
               "\")\"; });\n";
 
if (auto *operand = dyn_cast<OperandVariable>(dir->getOperand())) {
         body << llvm::formatv(
             "  ::llvm::interleaveComma({0}().getTypes(), _odsPrinter, "
             "[&](::mlir::TypeRange types) {{ _odsPrinter << \"(\" << "
             "types << \")\"; });\n",
             op.getGetterName(operand->getVar()->name));
         return;
       }
     }
+    const NamedTypeConstraint *var = nullptr;
+    {
+      if (auto *operand = dyn_cast<OperandVariable>(dir->getOperand()))
+        var = operand->getVar();
+      else if (auto *operand = dyn_cast<ResultVariable>(dir->getOperand()))
+        var = operand->getVar();
+    }
+    if (var && !var->isVariadicOfVariadic() && !var->isVariadic() &&
+        !var->isOptional()) {
+      std::string cppClass = var->constraint.getCPPClassName();
+      body << "  {\n"
+           << "    auto type = " << op.getGetterName(var->name)
+           << "().getType();\n"
+           << "    if (auto validType = type.dyn_cast<" << cppClass << ">())\n"
+           << "      _odsPrinter.printStrippedAttrOrType(validType);\n"
+           << "   else\n"
+           << "     _odsPrinter << type;\n"
+           << "  }\n";
+      return;
+    }
     body << "  _odsPrinter << ";
-    genTypeOperandPrinter(dir->getOperand(), op, body) << ";\n";
+    genTypeOperandPrinter(dir->getOperand(), op, body, /*useArrayRef=*/false)
+        << ";\n";
   } else if (auto *dir = dyn_cast<FunctionalTypeDirective>(element)) {
     body << "  _odsPrinter.printFunctionalType(";
     genTypeOperandPrinter(dir->getInputs(), op, body) << ", ";
     genTypeOperandPrinter(dir->getResults(), op, body) << ");\n";
   } else {
     llvm_unreachable("unknown format element");
   }
 }