llvm-project/mlir/lib/IR/BuiltinAttributes.cpp

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1818 lines
70 KiB
C++
Raw Normal View History

//===- 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/DialectResourceBlobManager.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 "llvm/ADT/APSInt.h"
#include "llvm/ADT/Sequence.h"
#include "llvm/Support/Endian.h"
using namespace mlir;
using namespace mlir::detail;
//===----------------------------------------------------------------------===//
/// Tablegen Attribute Definitions
//===----------------------------------------------------------------------===//
#define GET_ATTRDEF_CLASSES
#include "mlir/IR/BuiltinAttributes.cpp.inc"
//===----------------------------------------------------------------------===//
// BuiltinDialect
//===----------------------------------------------------------------------===//
void BuiltinDialect::registerAttributes() {
addAttributes<
#define GET_ATTRDEF_LIST
#include "mlir/IR/BuiltinAttributes.cpp.inc"
>();
}
//===----------------------------------------------------------------------===//
// ArrayAttr
//===----------------------------------------------------------------------===//
void ArrayAttr::walkImmediateSubElements(
function_ref<void(Attribute)> walkAttrsFn,
function_ref<void(Type)> walkTypesFn) const {
for (Attribute attr : getValue())
walkAttrsFn(attr);
}
Attribute
ArrayAttr::replaceImmediateSubElements(ArrayRef<Attribute> replAttrs,
ArrayRef<Type> replTypes) const {
return get(getContext(), replAttrs);
}
//===----------------------------------------------------------------------===//
// 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.
if (!inPlace)
storage.clear();
break;
case 1:
// One already sorted but may need to be copied.
if (!inPlace)
storage.assign({value[0]});
break;
case 2: {
bool isSorted = value[0] < value[1];
if (inPlace) {
if (!isSorted)
std::swap(storage[0], storage[1]);
} else if (isSorted) {
storage.assign({value[0], value[1]});
} else {
storage.assign({value[1], value[0]});
}
return !isSorted;
}
default:
if (!inPlace)
storage.assign(value.begin(), value.end());
// Check to see they are sorted already.
bool isSorted = llvm::is_sorted(value);
// If not, do a general sort.
if (!isSorted)
llvm::array_pod_sort(storage.begin(), storage.end());
return !isSorted;
}
return false;
}
/// Returns an entry with a duplicate name from the given sorted array of named
/// attributes. Returns llvm::None if all elements have unique names.
static Optional<NamedAttribute>
findDuplicateElement(ArrayRef<NamedAttribute> value) {
const Optional<NamedAttribute> none{llvm::None};
if (value.size() < 2)
return none;
if (value.size() == 2)
return value[0].getName() == value[1].getName() ? value[0] : none;
const auto *it = std::adjacent_find(value.begin(), value.end(),
[](NamedAttribute l, NamedAttribute r) {
return l.getName() == r.getName();
});
return it != value.end() ? *it : none;
}
bool DictionaryAttr::sort(ArrayRef<NamedAttribute> value,
SmallVectorImpl<NamedAttribute> &storage) {
bool isSorted = dictionaryAttrSort</*inPlace=*/false>(value, storage);
assert(!findDuplicateElement(storage) &&
"DictionaryAttr element names must be unique");
return isSorted;
}
bool DictionaryAttr::sortInPlace(SmallVectorImpl<NamedAttribute> &array) {
bool isSorted = dictionaryAttrSort</*inPlace=*/true>(array, array);
assert(!findDuplicateElement(array) &&
"DictionaryAttr element names must be unique");
return isSorted;
}
Optional<NamedAttribute>
DictionaryAttr::findDuplicate(SmallVectorImpl<NamedAttribute> &array,
bool isSorted) {
if (!isSorted)
dictionaryAttrSort</*inPlace=*/true>(array, array);
return findDuplicateElement(array);
}
DictionaryAttr DictionaryAttr::get(MLIRContext *context,
ArrayRef<NamedAttribute> value) {
if (value.empty())
return DictionaryAttr::getEmpty(context);
// We need to sort the element list to canonicalize it.
SmallVector<NamedAttribute, 8> storage;
if (dictionaryAttrSort</*inPlace=*/false>(value, storage))
value = storage;
assert(!findDuplicateElement(value) &&
"DictionaryAttr element names must be unique");
return Base::get(context, value);
}
/// Construct a dictionary with an array of values that is known to already be
/// sorted by name and uniqued.
DictionaryAttr DictionaryAttr::getWithSorted(MLIRContext *context,
ArrayRef<NamedAttribute> value) {
if (value.empty())
return DictionaryAttr::getEmpty(context);
// Ensure that the attribute elements are unique and sorted.
assert(llvm::is_sorted(
value, [](NamedAttribute l, NamedAttribute r) { return l < r; }) &&
"expected attribute values to be sorted");
assert(!findDuplicateElement(value) &&
"DictionaryAttr element names must be unique");
return Base::get(context, value);
}
/// Return the specified attribute if present, null otherwise.
Attribute DictionaryAttr::get(StringRef name) const {
auto it = impl::findAttrSorted(begin(), end(), name);
return it.second ? it.first->getValue() : Attribute();
}
Attribute DictionaryAttr::get(StringAttr name) const {
auto it = impl::findAttrSorted(begin(), end(), name);
return it.second ? it.first->getValue() : Attribute();
}
/// Return the specified named attribute if present, None otherwise.
Optional<NamedAttribute> DictionaryAttr::getNamed(StringRef name) const {
auto it = impl::findAttrSorted(begin(), end(), name);
return it.second ? *it.first : Optional<NamedAttribute>();
}
Optional<NamedAttribute> DictionaryAttr::getNamed(StringAttr name) const {
auto it = impl::findAttrSorted(begin(), end(), name);
return it.second ? *it.first : Optional<NamedAttribute>();
}
/// Return whether the specified attribute is present.
bool DictionaryAttr::contains(StringRef name) const {
return impl::findAttrSorted(begin(), end(), name).second;
}
bool DictionaryAttr::contains(StringAttr name) const {
return impl::findAttrSorted(begin(), end(), name).second;
}
DictionaryAttr::iterator DictionaryAttr::begin() const {
return getValue().begin();
}
DictionaryAttr::iterator DictionaryAttr::end() const {
return getValue().end();
}
size_t DictionaryAttr::size() const { return getValue().size(); }
DictionaryAttr DictionaryAttr::getEmptyUnchecked(MLIRContext *context) {
return Base::get(context, ArrayRef<NamedAttribute>());
}
void DictionaryAttr::walkImmediateSubElements(
function_ref<void(Attribute)> walkAttrsFn,
function_ref<void(Type)> walkTypesFn) const {
for (const NamedAttribute &attr : getValue())
walkAttrsFn(attr.getValue());
}
Attribute
DictionaryAttr::replaceImmediateSubElements(ArrayRef<Attribute> replAttrs,
ArrayRef<Type> replTypes) const {
std::vector<NamedAttribute> vec = getValue().vec();
for (auto &it : llvm::enumerate(replAttrs))
vec[it.index()].setValue(it.value());
// The above only modifies the mapped value, but not the key, and therefore
// not the order of the elements. It remains sorted
return getWithSorted(getContext(), vec);
}
//===----------------------------------------------------------------------===//
// StringAttr
//===----------------------------------------------------------------------===//
StringAttr StringAttr::getEmptyStringAttrUnchecked(MLIRContext *context) {
return Base::get(context, "", NoneType::get(context));
}
/// Twine support for StringAttr.
StringAttr StringAttr::get(MLIRContext *context, const Twine &twine) {
// Fast-path empty twine.
if (twine.isTriviallyEmpty())
return get(context);
SmallVector<char, 32> tempStr;
return Base::get(context, twine.toStringRef(tempStr), NoneType::get(context));
}
/// Twine support for StringAttr.
StringAttr StringAttr::get(const Twine &twine, Type type) {
SmallVector<char, 32> tempStr;
return Base::get(type.getContext(), twine.toStringRef(tempStr), type);
}
StringRef StringAttr::getValue() const { return getImpl()->value; }
[mlir] Remove types from attributes This patch removes the `type` field from `Attribute` along with the `Attribute::getType` accessor. Going forward, this means that attributes in MLIR will no longer have types as a first-class concept. This patch lays the groundwork to incrementally remove or refactor code that relies on generic attributes being typed. The immediate impact will be on attributes that rely on `Attribute` containing a type, such as `IntegerAttr`, `DenseElementsAttr`, and `ml_program::ExternAttr`, which will now need to define a type parameter on their storage classes. This will save memory as all other attribute kinds will no longer contain a type. Moreover, it will not be possible to generically query the type of an attribute directly. This patch provides an attribute interface `TypedAttr` that implements only one method, `getType`, which can be used to generically query the types of attributes that implement the interface. This interface can be used to retain the concept of a "typed attribute". The ODS-generated accessor for a `type` parameter automatically implements this method. Next steps will be to refactor the assembly formats of certain operations that rely on `parseAttribute(type)` and `printAttributeWithoutType` to remove special handling of type elision until `type` can be removed from the dialect parsing hook entirely; and incrementally remove uses of `TypedAttr`. Reviewed By: lattner, rriddle, jpienaar Differential Revision: https://reviews.llvm.org/D130092
2022-07-18 21:32:38 -07:00
Type StringAttr::getType() const { return getImpl()->type; }
Dialect *StringAttr::getReferencedDialect() const {
return getImpl()->referencedDialect;
}
//===----------------------------------------------------------------------===//
// FloatAttr
//===----------------------------------------------------------------------===//
double FloatAttr::getValueAsDouble() const {
return getValueAsDouble(getValue());
}
double FloatAttr::getValueAsDouble(APFloat value) {
if (&value.getSemantics() != &APFloat::IEEEdouble()) {
bool losesInfo = false;
value.convert(APFloat::IEEEdouble(), APFloat::rmNearestTiesToEven,
&losesInfo);
}
return value.convertToDouble();
}
[mlir][IR] Refactor the `getChecked` and `verifyConstructionInvariants` methods on Attributes/Types `verifyConstructionInvariants` is intended to allow for verifying the invariants of an attribute/type on construction, and `getChecked` is intended to enable more graceful error handling aside from an assert. There are a few problems with the current implementation of these methods: * `verifyConstructionInvariants` requires an mlir::Location for emitting errors, which is prohibitively costly in the situations that would most likely use them, e.g. the parser. This creates an unfortunate code duplication between the verifier code and the parser code, given that the parser operates on llvm::SMLoc and it is an undesirable overhead to pre-emptively convert from that to an mlir::Location. * `getChecked` effectively requires duplicating the definition of the `get` method, creating a quite clunky workflow due to the subtle different in its signature. This revision aims to talk the above problems by refactoring the implementation to use a callback for error emission. Using a callback allows for deferring the costly part of error emission until it is actually necessary. Due to the necessary signature change in each instance of these methods, this revision also takes this opportunity to cleanup the definition of these methods by: * restructuring the signature of `getChecked` such that it can be generated from the same code block as the `get` method. * renaming `verifyConstructionInvariants` to `verify` to match the naming scheme of the rest of the compiler. Differential Revision: https://reviews.llvm.org/D97100
2021-02-22 17:30:19 -08:00
LogicalResult FloatAttr::verify(function_ref<InFlightDiagnostic()> emitError,
Type type, APFloat value) {
// Verify that the type is correct.
if (!type.isa<FloatType>())
return emitError() << "expected floating point type";
// Verify that the type semantics match that of the value.
if (&type.cast<FloatType>().getFloatSemantics() != &value.getSemantics()) {
[mlir][IR] Refactor the `getChecked` and `verifyConstructionInvariants` methods on Attributes/Types `verifyConstructionInvariants` is intended to allow for verifying the invariants of an attribute/type on construction, and `getChecked` is intended to enable more graceful error handling aside from an assert. There are a few problems with the current implementation of these methods: * `verifyConstructionInvariants` requires an mlir::Location for emitting errors, which is prohibitively costly in the situations that would most likely use them, e.g. the parser. This creates an unfortunate code duplication between the verifier code and the parser code, given that the parser operates on llvm::SMLoc and it is an undesirable overhead to pre-emptively convert from that to an mlir::Location. * `getChecked` effectively requires duplicating the definition of the `get` method, creating a quite clunky workflow due to the subtle different in its signature. This revision aims to talk the above problems by refactoring the implementation to use a callback for error emission. Using a callback allows for deferring the costly part of error emission until it is actually necessary. Due to the necessary signature change in each instance of these methods, this revision also takes this opportunity to cleanup the definition of these methods by: * restructuring the signature of `getChecked` such that it can be generated from the same code block as the `get` method. * renaming `verifyConstructionInvariants` to `verify` to match the naming scheme of the rest of the compiler. Differential Revision: https://reviews.llvm.org/D97100
2021-02-22 17:30:19 -08:00
return emitError()
<< "FloatAttr type doesn't match the type implied by its value";
}
return success();
}
//===----------------------------------------------------------------------===//
// SymbolRefAttr
//===----------------------------------------------------------------------===//
SymbolRefAttr SymbolRefAttr::get(MLIRContext *ctx, StringRef value,
ArrayRef<FlatSymbolRefAttr> nestedRefs) {
return get(StringAttr::get(ctx, value), nestedRefs);
}
FlatSymbolRefAttr SymbolRefAttr::get(MLIRContext *ctx, StringRef value) {
return get(ctx, value, {}).cast<FlatSymbolRefAttr>();
}
FlatSymbolRefAttr SymbolRefAttr::get(StringAttr value) {
return get(value, {}).cast<FlatSymbolRefAttr>();
}
FlatSymbolRefAttr SymbolRefAttr::get(Operation *symbol) {
auto symName =
symbol->getAttrOfType<StringAttr>(SymbolTable::getSymbolAttrName());
assert(symName && "value does not have a valid symbol name");
return SymbolRefAttr::get(symName);
}
StringAttr SymbolRefAttr::getLeafReference() const {
ArrayRef<FlatSymbolRefAttr> nestedRefs = getNestedReferences();
return nestedRefs.empty() ? getRootReference() : nestedRefs.back().getAttr();
}
void SymbolRefAttr::walkImmediateSubElements(
function_ref<void(Attribute)> walkAttrsFn,
function_ref<void(Type)> walkTypesFn) const {
walkAttrsFn(getRootReference());
for (FlatSymbolRefAttr ref : getNestedReferences())
walkAttrsFn(ref);
}
Attribute
SymbolRefAttr::replaceImmediateSubElements(ArrayRef<Attribute> replAttrs,
ArrayRef<Type> replTypes) const {
ArrayRef<Attribute> rawNestedRefs = replAttrs.drop_front();
ArrayRef<FlatSymbolRefAttr> nestedRefs(
static_cast<const FlatSymbolRefAttr *>(rawNestedRefs.data()),
rawNestedRefs.size());
return get(replAttrs[0].cast<StringAttr>(), nestedRefs);
}
//===----------------------------------------------------------------------===//
// IntegerAttr
//===----------------------------------------------------------------------===//
int64_t IntegerAttr::getInt() const {
assert((getType().isIndex() || getType().isSignlessInteger()) &&
"must be signless integer");
return getValue().getSExtValue();
}
int64_t IntegerAttr::getSInt() const {
assert(getType().isSignedInteger() && "must be signed integer");
return getValue().getSExtValue();
}
uint64_t IntegerAttr::getUInt() const {
assert(getType().isUnsignedInteger() && "must be unsigned integer");
return getValue().getZExtValue();
}
/// Return the value as an APSInt which carries the signed from the type of
/// the attribute. This traps on signless integers types!
APSInt IntegerAttr::getAPSInt() const {
assert(!getType().isSignlessInteger() &&
"Signless integers don't carry a sign for APSInt");
return APSInt(getValue(), getType().isUnsignedInteger());
}
[mlir][IR] Refactor the `getChecked` and `verifyConstructionInvariants` methods on Attributes/Types `verifyConstructionInvariants` is intended to allow for verifying the invariants of an attribute/type on construction, and `getChecked` is intended to enable more graceful error handling aside from an assert. There are a few problems with the current implementation of these methods: * `verifyConstructionInvariants` requires an mlir::Location for emitting errors, which is prohibitively costly in the situations that would most likely use them, e.g. the parser. This creates an unfortunate code duplication between the verifier code and the parser code, given that the parser operates on llvm::SMLoc and it is an undesirable overhead to pre-emptively convert from that to an mlir::Location. * `getChecked` effectively requires duplicating the definition of the `get` method, creating a quite clunky workflow due to the subtle different in its signature. This revision aims to talk the above problems by refactoring the implementation to use a callback for error emission. Using a callback allows for deferring the costly part of error emission until it is actually necessary. Due to the necessary signature change in each instance of these methods, this revision also takes this opportunity to cleanup the definition of these methods by: * restructuring the signature of `getChecked` such that it can be generated from the same code block as the `get` method. * renaming `verifyConstructionInvariants` to `verify` to match the naming scheme of the rest of the compiler. Differential Revision: https://reviews.llvm.org/D97100
2021-02-22 17:30:19 -08:00
LogicalResult IntegerAttr::verify(function_ref<InFlightDiagnostic()> emitError,
Type type, APInt value) {
if (IntegerType integerType = type.dyn_cast<IntegerType>()) {
if (integerType.getWidth() != value.getBitWidth())
[mlir][IR] Refactor the `getChecked` and `verifyConstructionInvariants` methods on Attributes/Types `verifyConstructionInvariants` is intended to allow for verifying the invariants of an attribute/type on construction, and `getChecked` is intended to enable more graceful error handling aside from an assert. There are a few problems with the current implementation of these methods: * `verifyConstructionInvariants` requires an mlir::Location for emitting errors, which is prohibitively costly in the situations that would most likely use them, e.g. the parser. This creates an unfortunate code duplication between the verifier code and the parser code, given that the parser operates on llvm::SMLoc and it is an undesirable overhead to pre-emptively convert from that to an mlir::Location. * `getChecked` effectively requires duplicating the definition of the `get` method, creating a quite clunky workflow due to the subtle different in its signature. This revision aims to talk the above problems by refactoring the implementation to use a callback for error emission. Using a callback allows for deferring the costly part of error emission until it is actually necessary. Due to the necessary signature change in each instance of these methods, this revision also takes this opportunity to cleanup the definition of these methods by: * restructuring the signature of `getChecked` such that it can be generated from the same code block as the `get` method. * renaming `verifyConstructionInvariants` to `verify` to match the naming scheme of the rest of the compiler. Differential Revision: https://reviews.llvm.org/D97100
2021-02-22 17:30:19 -08:00
return emitError() << "integer type bit width (" << integerType.getWidth()
<< ") doesn't match value bit width ("
<< value.getBitWidth() << ")";
return success();
}
if (type.isa<IndexType>())
return success();
return emitError() << "expected integer or index type";
}
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>();
return intAttr && intAttr.getType().isSignlessInteger(1);
}
//===----------------------------------------------------------------------===//
// OpaqueAttr
//===----------------------------------------------------------------------===//
[mlir][IR] Refactor the `getChecked` and `verifyConstructionInvariants` methods on Attributes/Types `verifyConstructionInvariants` is intended to allow for verifying the invariants of an attribute/type on construction, and `getChecked` is intended to enable more graceful error handling aside from an assert. There are a few problems with the current implementation of these methods: * `verifyConstructionInvariants` requires an mlir::Location for emitting errors, which is prohibitively costly in the situations that would most likely use them, e.g. the parser. This creates an unfortunate code duplication between the verifier code and the parser code, given that the parser operates on llvm::SMLoc and it is an undesirable overhead to pre-emptively convert from that to an mlir::Location. * `getChecked` effectively requires duplicating the definition of the `get` method, creating a quite clunky workflow due to the subtle different in its signature. This revision aims to talk the above problems by refactoring the implementation to use a callback for error emission. Using a callback allows for deferring the costly part of error emission until it is actually necessary. Due to the necessary signature change in each instance of these methods, this revision also takes this opportunity to cleanup the definition of these methods by: * restructuring the signature of `getChecked` such that it can be generated from the same code block as the `get` method. * renaming `verifyConstructionInvariants` to `verify` to match the naming scheme of the rest of the compiler. Differential Revision: https://reviews.llvm.org/D97100
2021-02-22 17:30:19 -08:00
LogicalResult OpaqueAttr::verify(function_ref<InFlightDiagnostic()> emitError,
StringAttr dialect, StringRef attrData,
[mlir][IR] Refactor the `getChecked` and `verifyConstructionInvariants` methods on Attributes/Types `verifyConstructionInvariants` is intended to allow for verifying the invariants of an attribute/type on construction, and `getChecked` is intended to enable more graceful error handling aside from an assert. There are a few problems with the current implementation of these methods: * `verifyConstructionInvariants` requires an mlir::Location for emitting errors, which is prohibitively costly in the situations that would most likely use them, e.g. the parser. This creates an unfortunate code duplication between the verifier code and the parser code, given that the parser operates on llvm::SMLoc and it is an undesirable overhead to pre-emptively convert from that to an mlir::Location. * `getChecked` effectively requires duplicating the definition of the `get` method, creating a quite clunky workflow due to the subtle different in its signature. This revision aims to talk the above problems by refactoring the implementation to use a callback for error emission. Using a callback allows for deferring the costly part of error emission until it is actually necessary. Due to the necessary signature change in each instance of these methods, this revision also takes this opportunity to cleanup the definition of these methods by: * restructuring the signature of `getChecked` such that it can be generated from the same code block as the `get` method. * renaming `verifyConstructionInvariants` to `verify` to match the naming scheme of the rest of the compiler. Differential Revision: https://reviews.llvm.org/D97100
2021-02-22 17:30:19 -08:00
Type type) {
if (!Dialect::isValidNamespace(dialect.strref()))
[mlir][IR] Refactor the `getChecked` and `verifyConstructionInvariants` methods on Attributes/Types `verifyConstructionInvariants` is intended to allow for verifying the invariants of an attribute/type on construction, and `getChecked` is intended to enable more graceful error handling aside from an assert. There are a few problems with the current implementation of these methods: * `verifyConstructionInvariants` requires an mlir::Location for emitting errors, which is prohibitively costly in the situations that would most likely use them, e.g. the parser. This creates an unfortunate code duplication between the verifier code and the parser code, given that the parser operates on llvm::SMLoc and it is an undesirable overhead to pre-emptively convert from that to an mlir::Location. * `getChecked` effectively requires duplicating the definition of the `get` method, creating a quite clunky workflow due to the subtle different in its signature. This revision aims to talk the above problems by refactoring the implementation to use a callback for error emission. Using a callback allows for deferring the costly part of error emission until it is actually necessary. Due to the necessary signature change in each instance of these methods, this revision also takes this opportunity to cleanup the definition of these methods by: * restructuring the signature of `getChecked` such that it can be generated from the same code block as the `get` method. * renaming `verifyConstructionInvariants` to `verify` to match the naming scheme of the rest of the compiler. Differential Revision: https://reviews.llvm.org/D97100
2021-02-22 17:30:19 -08:00
return emitError() << "invalid dialect namespace '" << dialect << "'";
// Check that the dialect is actually registered.
MLIRContext *context = dialect.getContext();
if (!context->allowsUnregisteredDialects() &&
!context->getLoadedDialect(dialect.strref())) {
return emitError()
<< "#" << dialect << "<\"" << attrData << "\"> : " << type
<< " attribute created with unregistered dialect. If this is "
"intended, please call allowUnregisteredDialects() on the "
"MLIRContext, or use -allow-unregistered-dialect with "
"the MLIR opt tool used";
}
return success();
}
//===----------------------------------------------------------------------===//
// DenseElementsAttr Utilities
//===----------------------------------------------------------------------===//
/// Get the bitwidth of a dense element type within the buffer.
/// DenseElementsAttr requires bitwidths greater than 1 to be aligned by 8.
static size_t getDenseElementStorageWidth(size_t origWidth) {
return origWidth == 1 ? origWidth : llvm::alignTo<8>(origWidth);
}
static size_t getDenseElementStorageWidth(Type elementType) {
return getDenseElementStorageWidth(getDenseElementBitWidth(elementType));
}
/// Set a bit to a specific value.
static void setBit(char *rawData, size_t bitPos, bool value) {
if (value)
rawData[bitPos / CHAR_BIT] |= (1 << (bitPos % CHAR_BIT));
else
rawData[bitPos / CHAR_BIT] &= ~(1 << (bitPos % CHAR_BIT));
}
/// Return the value of the specified bit.
static bool getBit(const char *rawData, size_t bitPos) {
return (rawData[bitPos / CHAR_BIT] & (1 << (bitPos % CHAR_BIT))) != 0;
}
/// Copy actual `numBytes` data from `value` (APInt) to char array(`result`) for
/// BE format.
static void copyAPIntToArrayForBEmachine(APInt value, size_t numBytes,
char *result) {
assert(llvm::support::endian::system_endianness() == // NOLINT
llvm::support::endianness::big); // NOLINT
assert(value.getNumWords() * APInt::APINT_WORD_SIZE >= numBytes);
// Copy the words filled with data.
// For example, when `value` has 2 words, the first word is filled with data.
// `value` (10 bytes, BE):|abcdefgh|------ij| ==> `result` (BE):|abcdefgh|--|
size_t numFilledWords = (value.getNumWords() - 1) * APInt::APINT_WORD_SIZE;
std::copy_n(reinterpret_cast<const char *>(value.getRawData()),
numFilledWords, result);
// Convert last word of APInt to LE format and store it in char
// array(`valueLE`).
// ex. last word of `value` (BE): |------ij| ==> `valueLE` (LE): |ji------|
size_t lastWordPos = numFilledWords;
SmallVector<char, 8> valueLE(APInt::APINT_WORD_SIZE);
DenseIntOrFPElementsAttr::convertEndianOfCharForBEmachine(
reinterpret_cast<const char *>(value.getRawData()) + lastWordPos,
valueLE.begin(), APInt::APINT_BITS_PER_WORD, 1);
// Extract actual APInt data from `valueLE`, convert endianness to BE format,
// and store it in `result`.
// ex. `valueLE` (LE): |ji------| ==> `result` (BE): |abcdefgh|ij|
DenseIntOrFPElementsAttr::convertEndianOfCharForBEmachine(
valueLE.begin(), result + lastWordPos,
(numBytes - lastWordPos) * CHAR_BIT, 1);
}
/// Copy `numBytes` data from `inArray`(char array) to `result`(APINT) for BE
/// format.
static void copyArrayToAPIntForBEmachine(const char *inArray, size_t numBytes,
APInt &result) {
assert(llvm::support::endian::system_endianness() == // NOLINT
llvm::support::endianness::big); // NOLINT
assert(result.getNumWords() * APInt::APINT_WORD_SIZE >= numBytes);
// Copy the data that fills the word of `result` from `inArray`.
// For example, when `result` has 2 words, the first word will be filled with
// data. So, the first 8 bytes are copied from `inArray` here.
// `inArray` (10 bytes, BE): |abcdefgh|ij|
// ==> `result` (2 words, BE): |abcdefgh|--------|
size_t numFilledWords = (result.getNumWords() - 1) * APInt::APINT_WORD_SIZE;
std::copy_n(
inArray, numFilledWords,
const_cast<char *>(reinterpret_cast<const char *>(result.getRawData())));
// Convert array data which will be last word of `result` to LE format, and
// store it in char array(`inArrayLE`).
// ex. `inArray` (last two bytes, BE): |ij| ==> `inArrayLE` (LE): |ji------|
size_t lastWordPos = numFilledWords;
SmallVector<char, 8> inArrayLE(APInt::APINT_WORD_SIZE);
DenseIntOrFPElementsAttr::convertEndianOfCharForBEmachine(
inArray + lastWordPos, inArrayLE.begin(),
(numBytes - lastWordPos) * CHAR_BIT, 1);
// Convert `inArrayLE` to BE format, and store it in last word of `result`.
// ex. `inArrayLE` (LE): |ji------| ==> `result` (BE): |abcdefgh|------ij|
DenseIntOrFPElementsAttr::convertEndianOfCharForBEmachine(
inArrayLE.begin(),
const_cast<char *>(reinterpret_cast<const char *>(result.getRawData())) +
lastWordPos,
APInt::APINT_BITS_PER_WORD, 1);
}
/// Writes value to the bit position `bitPos` in array `rawData`.
static void writeBits(char *rawData, size_t bitPos, APInt value) {
size_t bitWidth = value.getBitWidth();
// If the bitwidth is 1 we just toggle the specific bit.
if (bitWidth == 1)
return setBit(rawData, bitPos, value.isOneValue());
// Otherwise, the bit position is guaranteed to be byte aligned.
assert((bitPos % CHAR_BIT) == 0 && "expected bitPos to be 8-bit aligned");
if (llvm::support::endian::system_endianness() ==
llvm::support::endianness::big) {
// Copy from `value` to `rawData + (bitPos / CHAR_BIT)`.
// Copying the first `llvm::divideCeil(bitWidth, CHAR_BIT)` bytes doesn't
// work correctly in BE format.
// ex. `value` (2 words including 10 bytes)
// ==> BE: |abcdefgh|------ij|, LE: |hgfedcba|ji------|
copyAPIntToArrayForBEmachine(value, llvm::divideCeil(bitWidth, CHAR_BIT),
rawData + (bitPos / CHAR_BIT));
} else {
std::copy_n(reinterpret_cast<const char *>(value.getRawData()),
llvm::divideCeil(bitWidth, CHAR_BIT),
rawData + (bitPos / CHAR_BIT));
}
}
/// Reads the next `bitWidth` bits from the bit position `bitPos` in array
/// `rawData`.
static APInt readBits(const char *rawData, size_t bitPos, size_t bitWidth) {
// Handle a boolean bit position.
if (bitWidth == 1)
return APInt(1, getBit(rawData, bitPos) ? 1 : 0);
// Otherwise, the bit position must be 8-bit aligned.
assert((bitPos % CHAR_BIT) == 0 && "expected bitPos to be 8-bit aligned");
APInt result(bitWidth, 0);
if (llvm::support::endian::system_endianness() ==
llvm::support::endianness::big) {
// Copy from `rawData + (bitPos / CHAR_BIT)` to `result`.
// Copying the first `llvm::divideCeil(bitWidth, CHAR_BIT)` bytes doesn't
// work correctly in BE format.
// ex. `result` (2 words including 10 bytes)
// ==> BE: |abcdefgh|------ij|, LE: |hgfedcba|ji------| This function
copyArrayToAPIntForBEmachine(rawData + (bitPos / CHAR_BIT),
llvm::divideCeil(bitWidth, CHAR_BIT), result);
} else {
std::copy_n(rawData + (bitPos / CHAR_BIT),
llvm::divideCeil(bitWidth, CHAR_BIT),
const_cast<char *>(
reinterpret_cast<const char *>(result.getRawData())));
}
return result;
}
/// Returns true if 'values' corresponds to a splat, i.e. one element, or has
/// the same element count as 'type'.
template <typename Values>
static bool hasSameElementsOrSplat(ShapedType type, const Values &values) {
return (values.size() == 1) ||
(type.getNumElements() == static_cast<int64_t>(values.size()));
}
//===----------------------------------------------------------------------===//
// DenseElementsAttr Iterators
//===----------------------------------------------------------------------===//
//===----------------------------------------------------------------------===//
// AttributeElementIterator
DenseElementsAttr::AttributeElementIterator::AttributeElementIterator(
DenseElementsAttr attr, size_t index)
: llvm::indexed_accessor_iterator<AttributeElementIterator, const void *,
Attribute, Attribute, Attribute>(
attr.getAsOpaquePointer(), index) {}
Attribute DenseElementsAttr::AttributeElementIterator::operator*() const {
auto owner = getFromOpaquePointer(base).cast<DenseElementsAttr>();
Type eltTy = owner.getElementType();
if (auto intEltTy = eltTy.dyn_cast<IntegerType>())
return IntegerAttr::get(eltTy, *IntElementIterator(owner, index));
if (eltTy.isa<IndexType>())
return IntegerAttr::get(eltTy, *IntElementIterator(owner, index));
if (auto floatEltTy = eltTy.dyn_cast<FloatType>()) {
IntElementIterator intIt(owner, index);
FloatElementIterator floatIt(floatEltTy.getFloatSemantics(), intIt);
return FloatAttr::get(eltTy, *floatIt);
}
if (auto complexTy = eltTy.dyn_cast<ComplexType>()) {
auto complexEltTy = complexTy.getElementType();
ComplexIntElementIterator complexIntIt(owner, index);
if (complexEltTy.isa<IntegerType>()) {
auto value = *complexIntIt;
auto real = IntegerAttr::get(complexEltTy, value.real());
auto imag = IntegerAttr::get(complexEltTy, value.imag());
return ArrayAttr::get(complexTy.getContext(),
ArrayRef<Attribute>{real, imag});
}
ComplexFloatElementIterator complexFloatIt(
complexEltTy.cast<FloatType>().getFloatSemantics(), complexIntIt);
auto value = *complexFloatIt;
auto real = FloatAttr::get(complexEltTy, value.real());
auto imag = FloatAttr::get(complexEltTy, value.imag());
return ArrayAttr::get(complexTy.getContext(),
ArrayRef<Attribute>{real, imag});
}
if (owner.isa<DenseStringElementsAttr>()) {
ArrayRef<StringRef> vals = owner.getRawStringData();
return StringAttr::get(owner.isSplat() ? vals.front() : vals[index], eltTy);
}
llvm_unreachable("unexpected element type");
}
//===----------------------------------------------------------------------===//
// BoolElementIterator
DenseElementsAttr::BoolElementIterator::BoolElementIterator(
DenseElementsAttr attr, size_t dataIndex)
: DenseElementIndexedIteratorImpl<BoolElementIterator, bool, bool, bool>(
attr.getRawData().data(), attr.isSplat(), dataIndex) {}
bool DenseElementsAttr::BoolElementIterator::operator*() const {
return getBit(getData(), getDataIndex());
}
//===----------------------------------------------------------------------===//
// IntElementIterator
DenseElementsAttr::IntElementIterator::IntElementIterator(
DenseElementsAttr attr, size_t dataIndex)
: DenseElementIndexedIteratorImpl<IntElementIterator, APInt, APInt, APInt>(
attr.getRawData().data(), attr.isSplat(), dataIndex),
bitWidth(getDenseElementBitWidth(attr.getElementType())) {}
APInt DenseElementsAttr::IntElementIterator::operator*() const {
return readBits(getData(),
getDataIndex() * getDenseElementStorageWidth(bitWidth),
bitWidth);
}
//===----------------------------------------------------------------------===//
// ComplexIntElementIterator
DenseElementsAttr::ComplexIntElementIterator::ComplexIntElementIterator(
DenseElementsAttr attr, size_t dataIndex)
: DenseElementIndexedIteratorImpl<ComplexIntElementIterator,
std::complex<APInt>, std::complex<APInt>,
std::complex<APInt>>(
attr.getRawData().data(), attr.isSplat(), dataIndex) {
auto complexType = attr.getElementType().cast<ComplexType>();
bitWidth = getDenseElementBitWidth(complexType.getElementType());
}
std::complex<APInt>
DenseElementsAttr::ComplexIntElementIterator::operator*() const {
size_t storageWidth = getDenseElementStorageWidth(bitWidth);
size_t offset = getDataIndex() * storageWidth * 2;
return {readBits(getData(), offset, bitWidth),
readBits(getData(), offset + storageWidth, bitWidth)};
}
//===----------------------------------------------------------------------===//
// DenseArrayAttr
//===----------------------------------------------------------------------===//
const bool *DenseArrayBaseAttr::value_begin_impl(OverloadToken<bool>) const {
return cast<DenseBoolArrayAttr>().asArrayRef().begin();
}
const int8_t *
DenseArrayBaseAttr::value_begin_impl(OverloadToken<int8_t>) const {
return cast<DenseI8ArrayAttr>().asArrayRef().begin();
}
const int16_t *
DenseArrayBaseAttr::value_begin_impl(OverloadToken<int16_t>) const {
return cast<DenseI16ArrayAttr>().asArrayRef().begin();
}
const int32_t *
DenseArrayBaseAttr::value_begin_impl(OverloadToken<int32_t>) const {
return cast<DenseI32ArrayAttr>().asArrayRef().begin();
}
const int64_t *
DenseArrayBaseAttr::value_begin_impl(OverloadToken<int64_t>) const {
return cast<DenseI64ArrayAttr>().asArrayRef().begin();
}
const float *DenseArrayBaseAttr::value_begin_impl(OverloadToken<float>) const {
return cast<DenseF32ArrayAttr>().asArrayRef().begin();
}
const double *
DenseArrayBaseAttr::value_begin_impl(OverloadToken<double>) const {
return cast<DenseF64ArrayAttr>().asArrayRef().begin();
}
void DenseArrayBaseAttr::print(AsmPrinter &printer) const {
print(printer.getStream());
}
void DenseArrayBaseAttr::printWithoutBraces(raw_ostream &os) const {
switch (getElementType()) {
case DenseArrayBaseAttr::EltType::I1:
this->cast<DenseBoolArrayAttr>().printWithoutBraces(os);
return;
case DenseArrayBaseAttr::EltType::I8:
this->cast<DenseI8ArrayAttr>().printWithoutBraces(os);
return;
case DenseArrayBaseAttr::EltType::I16:
this->cast<DenseI16ArrayAttr>().printWithoutBraces(os);
return;
case DenseArrayBaseAttr::EltType::I32:
this->cast<DenseI32ArrayAttr>().printWithoutBraces(os);
return;
case DenseArrayBaseAttr::EltType::I64:
this->cast<DenseI64ArrayAttr>().printWithoutBraces(os);
return;
case DenseArrayBaseAttr::EltType::F32:
this->cast<DenseF32ArrayAttr>().printWithoutBraces(os);
return;
case DenseArrayBaseAttr::EltType::F64:
this->cast<DenseF64ArrayAttr>().printWithoutBraces(os);
return;
}
llvm_unreachable("<unknown DenseArrayBaseAttr>");
}
void DenseArrayBaseAttr::print(raw_ostream &os) const {
os << "[";
printWithoutBraces(os);
os << "]";
}
template <typename T>
void DenseArrayAttr<T>::print(AsmPrinter &printer) const {
print(printer.getStream());
}
template <typename T>
void DenseArrayAttr<T>::printWithoutBraces(raw_ostream &os) const {
llvm::interleaveComma(asArrayRef(), os);
}
/// Specialization for bool to print `true` or `false`.
template <>
void DenseArrayAttr<bool>::printWithoutBraces(raw_ostream &os) const {
llvm::interleaveComma(asArrayRef(), os,
[&](bool v) { os << (v ? "true" : "false"); });
}
/// Specialization for int8_t for forcing printing as number instead of chars.
template <>
void DenseArrayAttr<int8_t>::printWithoutBraces(raw_ostream &os) const {
llvm::interleaveComma(asArrayRef(), os, [&](int64_t v) { os << v; });
}
template <typename T>
void DenseArrayAttr<T>::print(raw_ostream &os) const {
os << "[";
printWithoutBraces(os);
os << "]";
}
/// Parse a single element: generic template for int types, specialized for
/// floating point and boolean values below.
template <typename T>
static ParseResult parseDenseArrayAttrElt(AsmParser &parser, T &value) {
return parser.parseInteger(value);
}
template <>
ParseResult parseDenseArrayAttrElt<float>(AsmParser &parser, float &value) {
double doubleVal;
if (parser.parseFloat(doubleVal))
return failure();
value = doubleVal;
return success();
}
template <>
ParseResult parseDenseArrayAttrElt<double>(AsmParser &parser, double &value) {
return parser.parseFloat(value);
}
/// Parse a DenseArrayAttr without the braces: `1, 2, 3`
template <typename T>
Attribute DenseArrayAttr<T>::parseWithoutBraces(AsmParser &parser,
Type odsType) {
SmallVector<T> data;
if (failed(parser.parseCommaSeparatedList([&]() {
T value;
if (parseDenseArrayAttrElt(parser, value))
return failure();
data.push_back(value);
return success();
})))
return {};
return get(parser.getContext(), data);
}
/// Parse a DenseArrayAttr: `[ 1, 2, 3 ]`
template <typename T>
Attribute DenseArrayAttr<T>::parse(AsmParser &parser, Type odsType) {
if (parser.parseLSquare())
return {};
// Handle empty list case.
if (succeeded(parser.parseOptionalRSquare()))
return get(parser.getContext(), {});
Attribute result = parseWithoutBraces(parser, odsType);
if (parser.parseRSquare())
return {};
return result;
}
/// Conversion from DenseArrayAttr<T> to ArrayRef<T>.
template <typename T>
DenseArrayAttr<T>::operator ArrayRef<T>() const {
ArrayRef<char> raw = getRawData();
assert((raw.size() % sizeof(T)) == 0);
return ArrayRef<T>(reinterpret_cast<const T *>(raw.data()),
raw.size() / sizeof(T));
}
namespace {
/// Mapping from C++ element type to MLIR DenseArrayAttr internals.
template <typename T>
struct denseArrayAttrEltTypeBuilder;
template <>
struct denseArrayAttrEltTypeBuilder<bool> {
constexpr static auto eltType = DenseArrayBaseAttr::EltType::I1;
static ShapedType getShapedType(MLIRContext *context,
ArrayRef<int64_t> shape) {
return RankedTensorType::get(shape, IntegerType::get(context, 1));
}
};
template <>
struct denseArrayAttrEltTypeBuilder<int8_t> {
constexpr static auto eltType = DenseArrayBaseAttr::EltType::I8;
static ShapedType getShapedType(MLIRContext *context,
ArrayRef<int64_t> shape) {
return RankedTensorType::get(shape, IntegerType::get(context, 8));
}
};
template <>
struct denseArrayAttrEltTypeBuilder<int16_t> {
constexpr static auto eltType = DenseArrayBaseAttr::EltType::I16;
static ShapedType getShapedType(MLIRContext *context,
ArrayRef<int64_t> shape) {
return RankedTensorType::get(shape, IntegerType::get(context, 16));
}
};
template <>
struct denseArrayAttrEltTypeBuilder<int32_t> {
constexpr static auto eltType = DenseArrayBaseAttr::EltType::I32;
static ShapedType getShapedType(MLIRContext *context,
ArrayRef<int64_t> shape) {
return RankedTensorType::get(shape, IntegerType::get(context, 32));
}
};
template <>
struct denseArrayAttrEltTypeBuilder<int64_t> {
constexpr static auto eltType = DenseArrayBaseAttr::EltType::I64;
static ShapedType getShapedType(MLIRContext *context,
ArrayRef<int64_t> shape) {
return RankedTensorType::get(shape, IntegerType::get(context, 64));
}
};
template <>
struct denseArrayAttrEltTypeBuilder<float> {
constexpr static auto eltType = DenseArrayBaseAttr::EltType::F32;
static ShapedType getShapedType(MLIRContext *context,
ArrayRef<int64_t> shape) {
return RankedTensorType::get(shape, Float32Type::get(context));
}
};
template <>
struct denseArrayAttrEltTypeBuilder<double> {
constexpr static auto eltType = DenseArrayBaseAttr::EltType::F64;
static ShapedType getShapedType(MLIRContext *context,
ArrayRef<int64_t> shape) {
return RankedTensorType::get(shape, Float64Type::get(context));
}
};
} // namespace
/// Builds a DenseArrayAttr<T> from an ArrayRef<T>.
template <typename T>
DenseArrayAttr<T> DenseArrayAttr<T>::get(MLIRContext *context,
ArrayRef<T> content) {
auto size = static_cast<int64_t>(content.size());
auto shapedType =
denseArrayAttrEltTypeBuilder<T>::getShapedType(context, size);
auto eltType = denseArrayAttrEltTypeBuilder<T>::eltType;
auto rawArray = ArrayRef<char>(reinterpret_cast<const char *>(content.data()),
content.size() * sizeof(T));
return Base::get(context, shapedType, eltType, rawArray)
.template cast<DenseArrayAttr<T>>();
}
template <typename T>
bool DenseArrayAttr<T>::classof(Attribute attr) {
return attr.isa<DenseArrayBaseAttr>() &&
attr.cast<DenseArrayBaseAttr>().getElementType() ==
denseArrayAttrEltTypeBuilder<T>::eltType;
}
namespace mlir {
namespace detail {
// Explicit instantiation for all the supported DenseArrayAttr.
template class DenseArrayAttr<bool>;
template class DenseArrayAttr<int8_t>;
template class DenseArrayAttr<int16_t>;
template class DenseArrayAttr<int32_t>;
template class DenseArrayAttr<int64_t>;
template class DenseArrayAttr<float>;
template class DenseArrayAttr<double>;
} // namespace detail
} // namespace mlir
//===----------------------------------------------------------------------===//
// DenseElementsAttr
//===----------------------------------------------------------------------===//
/// Method for support type inquiry through isa, cast and dyn_cast.
bool DenseElementsAttr::classof(Attribute attr) {
return attr.isa<DenseIntOrFPElementsAttr, DenseStringElementsAttr>();
}
DenseElementsAttr DenseElementsAttr::get(ShapedType type,
ArrayRef<Attribute> values) {
assert(hasSameElementsOrSplat(type, values));
// If the element type is not based on int/float/index, assume it is a string
// type.
[mlir] Remove types from attributes This patch removes the `type` field from `Attribute` along with the `Attribute::getType` accessor. Going forward, this means that attributes in MLIR will no longer have types as a first-class concept. This patch lays the groundwork to incrementally remove or refactor code that relies on generic attributes being typed. The immediate impact will be on attributes that rely on `Attribute` containing a type, such as `IntegerAttr`, `DenseElementsAttr`, and `ml_program::ExternAttr`, which will now need to define a type parameter on their storage classes. This will save memory as all other attribute kinds will no longer contain a type. Moreover, it will not be possible to generically query the type of an attribute directly. This patch provides an attribute interface `TypedAttr` that implements only one method, `getType`, which can be used to generically query the types of attributes that implement the interface. This interface can be used to retain the concept of a "typed attribute". The ODS-generated accessor for a `type` parameter automatically implements this method. Next steps will be to refactor the assembly formats of certain operations that rely on `parseAttribute(type)` and `printAttributeWithoutType` to remove special handling of type elision until `type` can be removed from the dialect parsing hook entirely; and incrementally remove uses of `TypedAttr`. Reviewed By: lattner, rriddle, jpienaar Differential Revision: https://reviews.llvm.org/D130092
2022-07-18 21:32:38 -07:00
Type eltType = type.getElementType();
if (!eltType.isIntOrIndexOrFloat()) {
SmallVector<StringRef, 8> stringValues;
stringValues.reserve(values.size());
for (Attribute attr : values) {
assert(attr.isa<StringAttr>() &&
"expected string value for non integer/index/float element");
stringValues.push_back(attr.cast<StringAttr>().getValue());
}
return get(type, stringValues);
}
// Otherwise, get the raw storage width to use for the allocation.
size_t bitWidth = getDenseElementBitWidth(eltType);
size_t storageBitWidth = getDenseElementStorageWidth(bitWidth);
// Compress the attribute values into a character buffer.
SmallVector<char, 8> data(
llvm::divideCeil(storageBitWidth * values.size(), CHAR_BIT));
APInt intVal;
for (unsigned i = 0, e = values.size(); i < e; ++i) {
[mlir] Remove types from attributes This patch removes the `type` field from `Attribute` along with the `Attribute::getType` accessor. Going forward, this means that attributes in MLIR will no longer have types as a first-class concept. This patch lays the groundwork to incrementally remove or refactor code that relies on generic attributes being typed. The immediate impact will be on attributes that rely on `Attribute` containing a type, such as `IntegerAttr`, `DenseElementsAttr`, and `ml_program::ExternAttr`, which will now need to define a type parameter on their storage classes. This will save memory as all other attribute kinds will no longer contain a type. Moreover, it will not be possible to generically query the type of an attribute directly. This patch provides an attribute interface `TypedAttr` that implements only one method, `getType`, which can be used to generically query the types of attributes that implement the interface. This interface can be used to retain the concept of a "typed attribute". The ODS-generated accessor for a `type` parameter automatically implements this method. Next steps will be to refactor the assembly formats of certain operations that rely on `parseAttribute(type)` and `printAttributeWithoutType` to remove special handling of type elision until `type` can be removed from the dialect parsing hook entirely; and incrementally remove uses of `TypedAttr`. Reviewed By: lattner, rriddle, jpienaar Differential Revision: https://reviews.llvm.org/D130092
2022-07-18 21:32:38 -07:00
if (auto floatAttr = values[i].dyn_cast<FloatAttr>()) {
assert(floatAttr.getType() == eltType &&
"expected float attribute type to equal element type");
intVal = floatAttr.getValue().bitcastToAPInt();
} else {
auto intAttr = values[i].cast<IntegerAttr>();
assert(intAttr.getType() == eltType &&
"expected integer attribute type to equal element type");
intVal = intAttr.getValue();
}
assert(intVal.getBitWidth() == bitWidth &&
"expected value to have same bitwidth as element type");
writeBits(data.data(), i * storageBitWidth, intVal);
}
// Handle the special encoding of splat of bool.
[mlir] Remove types from attributes This patch removes the `type` field from `Attribute` along with the `Attribute::getType` accessor. Going forward, this means that attributes in MLIR will no longer have types as a first-class concept. This patch lays the groundwork to incrementally remove or refactor code that relies on generic attributes being typed. The immediate impact will be on attributes that rely on `Attribute` containing a type, such as `IntegerAttr`, `DenseElementsAttr`, and `ml_program::ExternAttr`, which will now need to define a type parameter on their storage classes. This will save memory as all other attribute kinds will no longer contain a type. Moreover, it will not be possible to generically query the type of an attribute directly. This patch provides an attribute interface `TypedAttr` that implements only one method, `getType`, which can be used to generically query the types of attributes that implement the interface. This interface can be used to retain the concept of a "typed attribute". The ODS-generated accessor for a `type` parameter automatically implements this method. Next steps will be to refactor the assembly formats of certain operations that rely on `parseAttribute(type)` and `printAttributeWithoutType` to remove special handling of type elision until `type` can be removed from the dialect parsing hook entirely; and incrementally remove uses of `TypedAttr`. Reviewed By: lattner, rriddle, jpienaar Differential Revision: https://reviews.llvm.org/D130092
2022-07-18 21:32:38 -07:00
if (values.size() == 1 && eltType.isInteger(1))
data[0] = data[0] ? -1 : 0;
return DenseIntOrFPElementsAttr::getRaw(type, data);
}
DenseElementsAttr DenseElementsAttr::get(ShapedType type,
ArrayRef<bool> values) {
assert(hasSameElementsOrSplat(type, values));
assert(type.getElementType().isInteger(1));
std::vector<char> buff(llvm::divideCeil(values.size(), CHAR_BIT));
if (!values.empty()) {
bool isSplat = true;
bool firstValue = values[0];
for (int i = 0, e = values.size(); i != e; ++i) {
isSplat &= values[i] == firstValue;
setBit(buff.data(), i, values[i]);
}
// Splat of bool is encoded as a byte with all-ones in it.
if (isSplat) {
buff.resize(1);
buff[0] = values[0] ? -1 : 0;
}
}
return DenseIntOrFPElementsAttr::getRaw(type, buff);
}
DenseElementsAttr DenseElementsAttr::get(ShapedType type,
ArrayRef<StringRef> values) {
assert(!type.getElementType().isIntOrFloat());
return DenseStringElementsAttr::get(type, values);
}
/// Constructs a dense integer elements attribute from an array of APInt
/// values. Each APInt value is expected to have the same bitwidth as the
/// element type of 'type'.
DenseElementsAttr DenseElementsAttr::get(ShapedType type,
ArrayRef<APInt> values) {
assert(type.getElementType().isIntOrIndex());
assert(hasSameElementsOrSplat(type, values));
size_t storageBitWidth = getDenseElementStorageWidth(type.getElementType());
return DenseIntOrFPElementsAttr::getRaw(type, storageBitWidth, values);
}
DenseElementsAttr DenseElementsAttr::get(ShapedType type,
ArrayRef<std::complex<APInt>> values) {
ComplexType complex = type.getElementType().cast<ComplexType>();
assert(complex.getElementType().isa<IntegerType>());
assert(hasSameElementsOrSplat(type, values));
size_t storageBitWidth = getDenseElementStorageWidth(complex) / 2;
ArrayRef<APInt> intVals(reinterpret_cast<const APInt *>(values.data()),
values.size() * 2);
return DenseIntOrFPElementsAttr::getRaw(type, storageBitWidth, intVals);
}
// Constructs a dense float elements attribute from an array of APFloat
// values. Each APFloat value is expected to have the same bitwidth as the
// element type of 'type'.
DenseElementsAttr DenseElementsAttr::get(ShapedType type,
ArrayRef<APFloat> values) {
assert(type.getElementType().isa<FloatType>());
assert(hasSameElementsOrSplat(type, values));
size_t storageBitWidth = getDenseElementStorageWidth(type.getElementType());
return DenseIntOrFPElementsAttr::getRaw(type, storageBitWidth, values);
}
DenseElementsAttr
DenseElementsAttr::get(ShapedType type,
ArrayRef<std::complex<APFloat>> values) {
ComplexType complex = type.getElementType().cast<ComplexType>();
assert(complex.getElementType().isa<FloatType>());
assert(hasSameElementsOrSplat(type, values));
ArrayRef<APFloat> apVals(reinterpret_cast<const APFloat *>(values.data()),
values.size() * 2);
size_t storageBitWidth = getDenseElementStorageWidth(complex) / 2;
return DenseIntOrFPElementsAttr::getRaw(type, storageBitWidth, apVals);
}
/// Construct a dense elements attribute from a raw buffer representing the
/// data for this attribute. Users should generally not use this methods as
/// the expected buffer format may not be a form the user expects.
DenseElementsAttr
DenseElementsAttr::getFromRawBuffer(ShapedType type, ArrayRef<char> rawBuffer) {
return DenseIntOrFPElementsAttr::getRaw(type, rawBuffer);
}
/// Returns true if the given buffer is a valid raw buffer for the given type.
bool DenseElementsAttr::isValidRawBuffer(ShapedType type,
ArrayRef<char> rawBuffer,
bool &detectedSplat) {
size_t storageWidth = getDenseElementStorageWidth(type.getElementType());
size_t rawBufferWidth = rawBuffer.size() * CHAR_BIT;
int64_t numElements = type.getNumElements();
// The initializer is always a splat if the result type has a single element.
detectedSplat = numElements == 1;
// Storage width of 1 is special as it is packed by the bit.
if (storageWidth == 1) {
// Check for a splat, or a buffer equal to the number of elements which
// consists of either all 0's or all 1's.
if (rawBuffer.size() == 1) {
auto rawByte = static_cast<uint8_t>(rawBuffer[0]);
if (rawByte == 0 || rawByte == 0xff) {
detectedSplat = true;
return true;
}
}
// This is a valid non-splat buffer if it has the right size.
return rawBufferWidth == llvm::alignTo<8>(numElements);
}
// All other types are 8-bit aligned, so we can just check the buffer width
// to know if only a single initializer element was passed in.
if (rawBufferWidth == storageWidth) {
detectedSplat = true;
return true;
}
// The raw buffer is valid if it has the right size.
return rawBufferWidth == storageWidth * numElements;
}
/// Check the information for a C++ data type, check if this type is valid for
/// the current attribute. This method is used to verify specific type
/// invariants that the templatized 'getValues' method cannot.
static bool isValidIntOrFloat(Type type, int64_t dataEltSize, bool isInt,
bool isSigned) {
// Make sure that the data element size is the same as the type element width.
if (getDenseElementBitWidth(type) !=
static_cast<size_t>(dataEltSize * CHAR_BIT))
return false;
// Check that the element type is either float or integer or index.
if (!isInt)
return type.isa<FloatType>();
if (type.isIndex())
return true;
auto intType = type.dyn_cast<IntegerType>();
if (!intType)
return false;
// Make sure signedness semantics is consistent.
if (intType.isSignless())
return true;
return intType.isSigned() ? isSigned : !isSigned;
}
/// Defaults down the subclass implementation.
DenseElementsAttr DenseElementsAttr::getRawComplex(ShapedType type,
ArrayRef<char> data,
int64_t dataEltSize,
bool isInt, bool isSigned) {
return DenseIntOrFPElementsAttr::getRawComplex(type, data, dataEltSize, isInt,
isSigned);
}
DenseElementsAttr DenseElementsAttr::getRawIntOrFloat(ShapedType type,
ArrayRef<char> data,
int64_t dataEltSize,
bool isInt,
bool isSigned) {
return DenseIntOrFPElementsAttr::getRawIntOrFloat(type, data, dataEltSize,
isInt, isSigned);
}
bool DenseElementsAttr::isValidIntOrFloat(int64_t dataEltSize, bool isInt,
bool isSigned) const {
return ::isValidIntOrFloat(getElementType(), dataEltSize, isInt, isSigned);
}
bool DenseElementsAttr::isValidComplex(int64_t dataEltSize, bool isInt,
bool isSigned) const {
return ::isValidIntOrFloat(
getElementType().cast<ComplexType>().getElementType(), dataEltSize / 2,
isInt, isSigned);
}
/// Returns true if this attribute corresponds to a splat, i.e. if all element
/// values are the same.
bool DenseElementsAttr::isSplat() const {
return static_cast<DenseElementsAttributeStorage *>(impl)->isSplat;
}
/// Return if the given complex type has an integer element type.
LLVM_ATTRIBUTE_UNUSED static bool isComplexOfIntType(Type type) {
return type.cast<ComplexType>().getElementType().isa<IntegerType>();
}
auto DenseElementsAttr::getComplexIntValues() const
-> iterator_range_impl<ComplexIntElementIterator> {
assert(isComplexOfIntType(getElementType()) &&
"expected complex integral type");
return {getType(), ComplexIntElementIterator(*this, 0),
ComplexIntElementIterator(*this, getNumElements())};
}
auto DenseElementsAttr::complex_value_begin() const
-> ComplexIntElementIterator {
assert(isComplexOfIntType(getElementType()) &&
"expected complex integral type");
return ComplexIntElementIterator(*this, 0);
}
auto DenseElementsAttr::complex_value_end() const -> ComplexIntElementIterator {
assert(isComplexOfIntType(getElementType()) &&
"expected complex integral type");
return ComplexIntElementIterator(*this, getNumElements());
}
/// Return the held element values as a range of APFloat. The element type of
/// this attribute must be of float type.
auto DenseElementsAttr::getFloatValues() const
-> iterator_range_impl<FloatElementIterator> {
auto elementType = getElementType().cast<FloatType>();
const auto &elementSemantics = elementType.getFloatSemantics();
return {getType(), FloatElementIterator(elementSemantics, raw_int_begin()),
FloatElementIterator(elementSemantics, raw_int_end())};
}
auto DenseElementsAttr::float_value_begin() const -> FloatElementIterator {
auto elementType = getElementType().cast<FloatType>();
return FloatElementIterator(elementType.getFloatSemantics(), raw_int_begin());
}
auto DenseElementsAttr::float_value_end() const -> FloatElementIterator {
auto elementType = getElementType().cast<FloatType>();
return FloatElementIterator(elementType.getFloatSemantics(), raw_int_end());
}
auto DenseElementsAttr::getComplexFloatValues() const
-> iterator_range_impl<ComplexFloatElementIterator> {
Type eltTy = getElementType().cast<ComplexType>().getElementType();
assert(eltTy.isa<FloatType>() && "expected complex float type");
const auto &semantics = eltTy.cast<FloatType>().getFloatSemantics();
return {getType(),
{semantics, {*this, 0}},
{semantics, {*this, static_cast<size_t>(getNumElements())}}};
}
auto DenseElementsAttr::complex_float_value_begin() const
-> ComplexFloatElementIterator {
Type eltTy = getElementType().cast<ComplexType>().getElementType();
assert(eltTy.isa<FloatType>() && "expected complex float type");
return {eltTy.cast<FloatType>().getFloatSemantics(), {*this, 0}};
}
auto DenseElementsAttr::complex_float_value_end() const
-> ComplexFloatElementIterator {
Type eltTy = getElementType().cast<ComplexType>().getElementType();
assert(eltTy.isa<FloatType>() && "expected complex float type");
return {eltTy.cast<FloatType>().getFloatSemantics(),
{*this, static_cast<size_t>(getNumElements())}};
}
/// Return the raw storage data held by this attribute.
ArrayRef<char> DenseElementsAttr::getRawData() const {
return static_cast<DenseIntOrFPElementsAttrStorage *>(impl)->data;
}
ArrayRef<StringRef> DenseElementsAttr::getRawStringData() const {
return static_cast<DenseStringElementsAttrStorage *>(impl)->data;
}
/// Return a new DenseElementsAttr that has the same data as the current
/// attribute, but has been reshaped to 'newType'. The new type must have the
/// same total number of elements as well as element type.
DenseElementsAttr DenseElementsAttr::reshape(ShapedType newType) {
ShapedType curType = getType();
if (curType == newType)
return *this;
assert(newType.getElementType() == curType.getElementType() &&
"expected the same element type");
assert(newType.getNumElements() == curType.getNumElements() &&
"expected the same number of elements");
return DenseIntOrFPElementsAttr::getRaw(newType, getRawData());
}
DenseElementsAttr DenseElementsAttr::resizeSplat(ShapedType newType) {
assert(isSplat() && "expected a splat type");
ShapedType curType = getType();
if (curType == newType)
return *this;
assert(newType.getElementType() == curType.getElementType() &&
"expected the same element type");
return DenseIntOrFPElementsAttr::getRaw(newType, getRawData());
}
/// Return a new DenseElementsAttr that has the same data as the current
/// attribute, but has bitcast elements such that it is now 'newType'. The new
/// type must have the same shape and element types of the same bitwidth as the
/// current type.
DenseElementsAttr DenseElementsAttr::bitcast(Type newElType) {
ShapedType curType = getType();
Type curElType = curType.getElementType();
if (curElType == newElType)
return *this;
assert(getDenseElementBitWidth(newElType) ==
getDenseElementBitWidth(curElType) &&
"expected element types with the same bitwidth");
return DenseIntOrFPElementsAttr::getRaw(curType.clone(newElType),
getRawData());
}
DenseElementsAttr
DenseElementsAttr::mapValues(Type newElementType,
function_ref<APInt(const APInt &)> mapping) const {
return cast<DenseIntElementsAttr>().mapValues(newElementType, mapping);
}
DenseElementsAttr DenseElementsAttr::mapValues(
Type newElementType, function_ref<APInt(const APFloat &)> mapping) const {
return cast<DenseFPElementsAttr>().mapValues(newElementType, mapping);
}
[mlir] Refactor ElementsAttr into an AttrInterface This revision refactors ElementsAttr into an Attribute Interface. This enables a common interface with which to interact with element attributes, without needing to modify the builtin dialect. It also removes a majority (if not all?) of the need for the current OpaqueElementsAttr, which was originally intended as a way to opaquely represent data that was not representable by the other builtin constructs. The new ElementsAttr interface not only allows for users to natively represent their data in the way that best suits them, it also allows for efficient opaque access and iteration of the underlying data. Attributes using the ElementsAttr interface can directly expose support for interacting with the held elements using any C++ data type they claim to support. For example, DenseIntOrFpElementsAttr supports iteration using various native C++ integer/float data types, as well as APInt/APFloat, and more. ElementsAttr instances that refer to DenseIntOrFpElementsAttr can use all of these data types for iteration: ```c++ DenseIntOrFpElementsAttr intElementsAttr = ...; ElementsAttr attr = intElementsAttr; for (uint64_t value : attr.getValues<uint64_t>()) ...; for (APInt value : attr.getValues<APInt>()) ...; for (IntegerAttr value : attr.getValues<IntegerAttr>()) ...; ``` ElementsAttr also supports failable range/iterator access, allowing for selective code paths depending on data type support: ```c++ ElementsAttr attr = ...; if (auto range = attr.tryGetValues<uint64_t>()) { for (uint64_t value : *range) ...; } ``` Differential Revision: https://reviews.llvm.org/D109190
2021-09-21 01:40:45 +00:00
ShapedType DenseElementsAttr::getType() const {
[mlir] Remove types from attributes This patch removes the `type` field from `Attribute` along with the `Attribute::getType` accessor. Going forward, this means that attributes in MLIR will no longer have types as a first-class concept. This patch lays the groundwork to incrementally remove or refactor code that relies on generic attributes being typed. The immediate impact will be on attributes that rely on `Attribute` containing a type, such as `IntegerAttr`, `DenseElementsAttr`, and `ml_program::ExternAttr`, which will now need to define a type parameter on their storage classes. This will save memory as all other attribute kinds will no longer contain a type. Moreover, it will not be possible to generically query the type of an attribute directly. This patch provides an attribute interface `TypedAttr` that implements only one method, `getType`, which can be used to generically query the types of attributes that implement the interface. This interface can be used to retain the concept of a "typed attribute". The ODS-generated accessor for a `type` parameter automatically implements this method. Next steps will be to refactor the assembly formats of certain operations that rely on `parseAttribute(type)` and `printAttributeWithoutType` to remove special handling of type elision until `type` can be removed from the dialect parsing hook entirely; and incrementally remove uses of `TypedAttr`. Reviewed By: lattner, rriddle, jpienaar Differential Revision: https://reviews.llvm.org/D130092
2022-07-18 21:32:38 -07:00
return static_cast<const DenseElementsAttributeStorage *>(impl)->type;
[mlir] Refactor ElementsAttr into an AttrInterface This revision refactors ElementsAttr into an Attribute Interface. This enables a common interface with which to interact with element attributes, without needing to modify the builtin dialect. It also removes a majority (if not all?) of the need for the current OpaqueElementsAttr, which was originally intended as a way to opaquely represent data that was not representable by the other builtin constructs. The new ElementsAttr interface not only allows for users to natively represent their data in the way that best suits them, it also allows for efficient opaque access and iteration of the underlying data. Attributes using the ElementsAttr interface can directly expose support for interacting with the held elements using any C++ data type they claim to support. For example, DenseIntOrFpElementsAttr supports iteration using various native C++ integer/float data types, as well as APInt/APFloat, and more. ElementsAttr instances that refer to DenseIntOrFpElementsAttr can use all of these data types for iteration: ```c++ DenseIntOrFpElementsAttr intElementsAttr = ...; ElementsAttr attr = intElementsAttr; for (uint64_t value : attr.getValues<uint64_t>()) ...; for (APInt value : attr.getValues<APInt>()) ...; for (IntegerAttr value : attr.getValues<IntegerAttr>()) ...; ``` ElementsAttr also supports failable range/iterator access, allowing for selective code paths depending on data type support: ```c++ ElementsAttr attr = ...; if (auto range = attr.tryGetValues<uint64_t>()) { for (uint64_t value : *range) ...; } ``` Differential Revision: https://reviews.llvm.org/D109190
2021-09-21 01:40:45 +00:00
}
Type DenseElementsAttr::getElementType() const {
return getType().getElementType();
}
int64_t DenseElementsAttr::getNumElements() const {
return getType().getNumElements();
}
//===----------------------------------------------------------------------===//
// DenseIntOrFPElementsAttr
//===----------------------------------------------------------------------===//
/// Utility method to write a range of APInt values to a buffer.
template <typename APRangeT>
static void writeAPIntsToBuffer(size_t storageWidth, std::vector<char> &data,
APRangeT &&values) {
size_t numValues = llvm::size(values);
data.resize(llvm::divideCeil(storageWidth * numValues, CHAR_BIT));
size_t offset = 0;
for (auto it = values.begin(), e = values.end(); it != e;
++it, offset += storageWidth) {
assert((*it).getBitWidth() <= storageWidth);
writeBits(data.data(), offset, *it);
}
// Handle the special encoding of splat of a boolean.
if (numValues == 1 && (*values.begin()).getBitWidth() == 1)
data[0] = data[0] ? -1 : 0;
}
/// Constructs a dense elements attribute from an array of raw APFloat values.
/// Each APFloat value is expected to have the same bitwidth as the element
/// type of 'type'. 'type' must be a vector or tensor with static shape.
DenseElementsAttr DenseIntOrFPElementsAttr::getRaw(ShapedType type,
size_t storageWidth,
ArrayRef<APFloat> values) {
std::vector<char> data;
auto unwrapFloat = [](const APFloat &val) { return val.bitcastToAPInt(); };
writeAPIntsToBuffer(storageWidth, data, llvm::map_range(values, unwrapFloat));
return DenseIntOrFPElementsAttr::getRaw(type, data);
}
/// Constructs a dense elements attribute from an array of raw APInt values.
/// Each APInt value is expected to have the same bitwidth as the element type
/// of 'type'.
DenseElementsAttr DenseIntOrFPElementsAttr::getRaw(ShapedType type,
size_t storageWidth,
ArrayRef<APInt> values) {
std::vector<char> data;
writeAPIntsToBuffer(storageWidth, data, values);
return DenseIntOrFPElementsAttr::getRaw(type, data);
}
DenseElementsAttr DenseIntOrFPElementsAttr::getRaw(ShapedType type,
ArrayRef<char> data) {
assert((type.isa<RankedTensorType, VectorType>()) &&
"type must be ranked tensor or vector");
assert(type.hasStaticShape() && "type must have static shape");
bool isSplat = false;
bool isValid = isValidRawBuffer(type, data, isSplat);
assert(isValid);
(void)isValid;
return Base::get(type.getContext(), type, data, isSplat);
}
/// Overload of the raw 'get' method that asserts that the given type is of
/// complex type. This method is used to verify type invariants that the
/// templatized 'get' method cannot.
DenseElementsAttr DenseIntOrFPElementsAttr::getRawComplex(ShapedType type,
ArrayRef<char> data,
int64_t dataEltSize,
bool isInt,
bool isSigned) {
assert(::isValidIntOrFloat(
type.getElementType().cast<ComplexType>().getElementType(),
dataEltSize / 2, isInt, isSigned));
int64_t numElements = data.size() / dataEltSize;
(void)numElements;
assert(numElements == 1 || numElements == type.getNumElements());
return getRaw(type, data);
}
/// Overload of the 'getRaw' method that asserts that the given type is of
/// integer type. This method is used to verify type invariants that the
/// templatized 'get' method cannot.
DenseElementsAttr
DenseIntOrFPElementsAttr::getRawIntOrFloat(ShapedType type, ArrayRef<char> data,
int64_t dataEltSize, bool isInt,
bool isSigned) {
assert(
::isValidIntOrFloat(type.getElementType(), dataEltSize, isInt, isSigned));
int64_t numElements = data.size() / dataEltSize;
assert(numElements == 1 || numElements == type.getNumElements());
(void)numElements;
return getRaw(type, data);
}
void DenseIntOrFPElementsAttr::convertEndianOfCharForBEmachine(
const char *inRawData, char *outRawData, size_t elementBitWidth,
size_t numElements) {
using llvm::support::ulittle16_t;
using llvm::support::ulittle32_t;
using llvm::support::ulittle64_t;
assert(llvm::support::endian::system_endianness() == // NOLINT
llvm::support::endianness::big); // NOLINT
// NOLINT to avoid warning message about replacing by static_assert()
// Following std::copy_n always converts endianness on BE machine.
switch (elementBitWidth) {
case 16: {
const ulittle16_t *inRawDataPos =
reinterpret_cast<const ulittle16_t *>(inRawData);
uint16_t *outDataPos = reinterpret_cast<uint16_t *>(outRawData);
std::copy_n(inRawDataPos, numElements, outDataPos);
break;
}
case 32: {
const ulittle32_t *inRawDataPos =
reinterpret_cast<const ulittle32_t *>(inRawData);
uint32_t *outDataPos = reinterpret_cast<uint32_t *>(outRawData);
std::copy_n(inRawDataPos, numElements, outDataPos);
break;
}
case 64: {
const ulittle64_t *inRawDataPos =
reinterpret_cast<const ulittle64_t *>(inRawData);
uint64_t *outDataPos = reinterpret_cast<uint64_t *>(outRawData);
std::copy_n(inRawDataPos, numElements, outDataPos);
break;
}
default: {
size_t nBytes = elementBitWidth / CHAR_BIT;
for (size_t i = 0; i < nBytes; i++)
std::copy_n(inRawData + (nBytes - 1 - i), 1, outRawData + i);
break;
}
}
}
void DenseIntOrFPElementsAttr::convertEndianOfArrayRefForBEmachine(
ArrayRef<char> inRawData, MutableArrayRef<char> outRawData,
ShapedType type) {
size_t numElements = type.getNumElements();
Type elementType = type.getElementType();
if (ComplexType complexTy = elementType.dyn_cast<ComplexType>()) {
elementType = complexTy.getElementType();
numElements = numElements * 2;
}
size_t elementBitWidth = getDenseElementStorageWidth(elementType);
assert(numElements * elementBitWidth == inRawData.size() * CHAR_BIT &&
inRawData.size() <= outRawData.size());
if (elementBitWidth <= CHAR_BIT)
std::memcpy(outRawData.begin(), inRawData.begin(), inRawData.size());
else
convertEndianOfCharForBEmachine(inRawData.begin(), outRawData.begin(),
elementBitWidth, numElements);
}
//===----------------------------------------------------------------------===//
// DenseFPElementsAttr
//===----------------------------------------------------------------------===//
template <typename Fn, typename Attr>
static ShapedType mappingHelper(Fn mapping, Attr &attr, ShapedType inType,
Type newElementType,
llvm::SmallVectorImpl<char> &data) {
size_t bitWidth = getDenseElementBitWidth(newElementType);
size_t storageBitWidth = getDenseElementStorageWidth(bitWidth);
ShapedType newArrayType;
if (inType.isa<RankedTensorType>())
newArrayType = RankedTensorType::get(inType.getShape(), newElementType);
else if (inType.isa<UnrankedTensorType>())
newArrayType = RankedTensorType::get(inType.getShape(), newElementType);
else if (auto vType = inType.dyn_cast<VectorType>())
newArrayType = VectorType::get(vType.getShape(), newElementType,
vType.getNumScalableDims());
else
assert(newArrayType && "Unhandled tensor type");
size_t numRawElements = attr.isSplat() ? 1 : newArrayType.getNumElements();
data.resize(llvm::divideCeil(storageBitWidth * numRawElements, CHAR_BIT));
// Functor used to process a single element value of the attribute.
auto processElt = [&](decltype(*attr.begin()) value, size_t index) {
auto newInt = mapping(value);
assert(newInt.getBitWidth() == bitWidth);
writeBits(data.data(), index * storageBitWidth, newInt);
};
// Check for the splat case.
if (attr.isSplat()) {
processElt(*attr.begin(), /*index=*/0);
return newArrayType;
}
// Otherwise, process all of the element values.
uint64_t elementIdx = 0;
for (auto value : attr)
processElt(value, elementIdx++);
return newArrayType;
}
DenseElementsAttr DenseFPElementsAttr::mapValues(
Type newElementType, function_ref<APInt(const APFloat &)> mapping) const {
llvm::SmallVector<char, 8> elementData;
auto newArrayType =
mappingHelper(mapping, *this, getType(), newElementType, elementData);
return getRaw(newArrayType, elementData);
}
/// Method for supporting type inquiry through isa, cast and dyn_cast.
bool DenseFPElementsAttr::classof(Attribute attr) {
[mlir] Remove types from attributes This patch removes the `type` field from `Attribute` along with the `Attribute::getType` accessor. Going forward, this means that attributes in MLIR will no longer have types as a first-class concept. This patch lays the groundwork to incrementally remove or refactor code that relies on generic attributes being typed. The immediate impact will be on attributes that rely on `Attribute` containing a type, such as `IntegerAttr`, `DenseElementsAttr`, and `ml_program::ExternAttr`, which will now need to define a type parameter on their storage classes. This will save memory as all other attribute kinds will no longer contain a type. Moreover, it will not be possible to generically query the type of an attribute directly. This patch provides an attribute interface `TypedAttr` that implements only one method, `getType`, which can be used to generically query the types of attributes that implement the interface. This interface can be used to retain the concept of a "typed attribute". The ODS-generated accessor for a `type` parameter automatically implements this method. Next steps will be to refactor the assembly formats of certain operations that rely on `parseAttribute(type)` and `printAttributeWithoutType` to remove special handling of type elision until `type` can be removed from the dialect parsing hook entirely; and incrementally remove uses of `TypedAttr`. Reviewed By: lattner, rriddle, jpienaar Differential Revision: https://reviews.llvm.org/D130092
2022-07-18 21:32:38 -07:00
if (auto denseAttr = attr.dyn_cast<DenseElementsAttr>())
return denseAttr.getType().getElementType().isa<FloatType>();
return false;
}
//===----------------------------------------------------------------------===//
// DenseIntElementsAttr
//===----------------------------------------------------------------------===//
DenseElementsAttr DenseIntElementsAttr::mapValues(
Type newElementType, function_ref<APInt(const APInt &)> mapping) const {
llvm::SmallVector<char, 8> elementData;
auto newArrayType =
mappingHelper(mapping, *this, getType(), newElementType, elementData);
return getRaw(newArrayType, elementData);
}
/// Method for supporting type inquiry through isa, cast and dyn_cast.
bool DenseIntElementsAttr::classof(Attribute attr) {
[mlir] Remove types from attributes This patch removes the `type` field from `Attribute` along with the `Attribute::getType` accessor. Going forward, this means that attributes in MLIR will no longer have types as a first-class concept. This patch lays the groundwork to incrementally remove or refactor code that relies on generic attributes being typed. The immediate impact will be on attributes that rely on `Attribute` containing a type, such as `IntegerAttr`, `DenseElementsAttr`, and `ml_program::ExternAttr`, which will now need to define a type parameter on their storage classes. This will save memory as all other attribute kinds will no longer contain a type. Moreover, it will not be possible to generically query the type of an attribute directly. This patch provides an attribute interface `TypedAttr` that implements only one method, `getType`, which can be used to generically query the types of attributes that implement the interface. This interface can be used to retain the concept of a "typed attribute". The ODS-generated accessor for a `type` parameter automatically implements this method. Next steps will be to refactor the assembly formats of certain operations that rely on `parseAttribute(type)` and `printAttributeWithoutType` to remove special handling of type elision until `type` can be removed from the dialect parsing hook entirely; and incrementally remove uses of `TypedAttr`. Reviewed By: lattner, rriddle, jpienaar Differential Revision: https://reviews.llvm.org/D130092
2022-07-18 21:32:38 -07:00
if (auto denseAttr = attr.dyn_cast<DenseElementsAttr>())
return denseAttr.getType().getElementType().isIntOrIndex();
return false;
}
//===----------------------------------------------------------------------===//
// DenseResourceElementsAttr
//===----------------------------------------------------------------------===//
DenseResourceElementsAttr
DenseResourceElementsAttr::get(ShapedType type,
DenseResourceElementsHandle handle) {
return Base::get(type.getContext(), type, handle);
}
DenseResourceElementsAttr DenseResourceElementsAttr::get(ShapedType type,
StringRef blobName,
AsmResourceBlob blob) {
// Extract the builtin dialect resource manager from context and construct a
// handle by inserting a new resource using the provided blob.
auto &manager =
DenseResourceElementsHandle::getManagerInterface(type.getContext());
return get(type, manager.insert(blobName, std::move(blob)));
}
//===----------------------------------------------------------------------===//
// DenseResourceElementsAttrBase
namespace {
/// Instantiations of this class provide utilities for interacting with native
/// data types in the context of DenseResourceElementsAttr.
template <typename T>
struct DenseResourceAttrUtil;
template <size_t width, bool isSigned>
struct DenseResourceElementsAttrIntUtil {
static bool checkElementType(Type eltType) {
IntegerType type = eltType.dyn_cast<IntegerType>();
if (!type || type.getWidth() != width)
return false;
return isSigned ? !type.isUnsigned() : !type.isSigned();
}
};
template <>
struct DenseResourceAttrUtil<bool> {
static bool checkElementType(Type eltType) {
return eltType.isSignlessInteger(1);
}
};
template <>
struct DenseResourceAttrUtil<int8_t>
: public DenseResourceElementsAttrIntUtil<8, true> {};
template <>
struct DenseResourceAttrUtil<uint8_t>
: public DenseResourceElementsAttrIntUtil<8, false> {};
template <>
struct DenseResourceAttrUtil<int16_t>
: public DenseResourceElementsAttrIntUtil<16, true> {};
template <>
struct DenseResourceAttrUtil<uint16_t>
: public DenseResourceElementsAttrIntUtil<16, false> {};
template <>
struct DenseResourceAttrUtil<int32_t>
: public DenseResourceElementsAttrIntUtil<32, true> {};
template <>
struct DenseResourceAttrUtil<uint32_t>
: public DenseResourceElementsAttrIntUtil<32, false> {};
template <>
struct DenseResourceAttrUtil<int64_t>
: public DenseResourceElementsAttrIntUtil<64, true> {};
template <>
struct DenseResourceAttrUtil<uint64_t>
: public DenseResourceElementsAttrIntUtil<64, false> {};
template <>
struct DenseResourceAttrUtil<float> {
static bool checkElementType(Type eltType) { return eltType.isF32(); }
};
template <>
struct DenseResourceAttrUtil<double> {
static bool checkElementType(Type eltType) { return eltType.isF64(); }
};
} // namespace
template <typename T>
DenseResourceElementsAttrBase<T>
DenseResourceElementsAttrBase<T>::get(ShapedType type, StringRef blobName,
AsmResourceBlob blob) {
// Check that the blob is in the form we were expecting.
assert(blob.getDataAlignment() == alignof(T) &&
"alignment mismatch between expected alignment and blob alignment");
assert(((blob.getData().size() % sizeof(T)) == 0) &&
"size mismatch between expected element width and blob size");
assert(DenseResourceAttrUtil<T>::checkElementType(type.getElementType()) &&
"invalid shape element type for provided type `T`");
return DenseResourceElementsAttr::get(type, blobName, std::move(blob))
.template cast<DenseResourceElementsAttrBase<T>>();
}
template <typename T>
Optional<ArrayRef<T>>
DenseResourceElementsAttrBase<T>::tryGetAsArrayRef() const {
if (AsmResourceBlob *blob = this->getRawHandle().getBlob())
return blob->template getDataAs<T>();
return llvm::None;
}
template <typename T>
bool DenseResourceElementsAttrBase<T>::classof(Attribute attr) {
auto resourceAttr = attr.dyn_cast<DenseResourceElementsAttr>();
return resourceAttr && DenseResourceAttrUtil<T>::checkElementType(
resourceAttr.getElementType());
}
namespace mlir {
namespace detail {
// Explicit instantiation for all the supported DenseResourceElementsAttr.
template class DenseResourceElementsAttrBase<bool>;
template class DenseResourceElementsAttrBase<int8_t>;
template class DenseResourceElementsAttrBase<int16_t>;
template class DenseResourceElementsAttrBase<int32_t>;
template class DenseResourceElementsAttrBase<int64_t>;
template class DenseResourceElementsAttrBase<uint8_t>;
template class DenseResourceElementsAttrBase<uint16_t>;
template class DenseResourceElementsAttrBase<uint32_t>;
template class DenseResourceElementsAttrBase<uint64_t>;
template class DenseResourceElementsAttrBase<float>;
template class DenseResourceElementsAttrBase<double>;
} // namespace detail
} // namespace mlir
//===----------------------------------------------------------------------===//
// SparseElementsAttr
//===----------------------------------------------------------------------===//
/// Get a zero APFloat for the given sparse attribute.
APFloat SparseElementsAttr::getZeroAPFloat() const {
auto eltType = getElementType().cast<FloatType>();
return APFloat(eltType.getFloatSemantics());
}
/// Get a zero APInt for the given sparse attribute.
APInt SparseElementsAttr::getZeroAPInt() const {
auto eltType = getElementType().cast<IntegerType>();
return APInt::getZero(eltType.getWidth());
}
/// Get a zero attribute for the given attribute type.
Attribute SparseElementsAttr::getZeroAttr() const {
auto eltType = getElementType();
// Handle floating point elements.
if (eltType.isa<FloatType>())
return FloatAttr::get(eltType, 0);
// Handle complex elements.
if (auto complexTy = eltType.dyn_cast<ComplexType>()) {
auto eltType = complexTy.getElementType();
Attribute zero;
if (eltType.isa<FloatType>())
zero = FloatAttr::get(eltType, 0);
else // must be integer
zero = IntegerAttr::get(eltType, 0);
return ArrayAttr::get(complexTy.getContext(),
ArrayRef<Attribute>{zero, zero});
}
// Handle string type.
if (getValues().isa<DenseStringElementsAttr>())
return StringAttr::get("", eltType);
// Otherwise, this is an integer.
return IntegerAttr::get(eltType, 0);
}
/// Flatten, and return, all of the sparse indices in this attribute in
/// row-major order.
std::vector<ptrdiff_t> SparseElementsAttr::getFlattenedSparseIndices() const {
std::vector<ptrdiff_t> flatSparseIndices;
// The sparse indices are 64-bit integers, so we can reinterpret the raw data
// as a 1-D index array.
auto sparseIndices = getIndices();
auto sparseIndexValues = sparseIndices.getValues<uint64_t>();
if (sparseIndices.isSplat()) {
SmallVector<uint64_t, 8> indices(getType().getRank(),
*sparseIndexValues.begin());
flatSparseIndices.push_back(getFlattenedIndex(indices));
return flatSparseIndices;
}
// Otherwise, reinterpret each index as an ArrayRef when flattening.
auto numSparseIndices = sparseIndices.getType().getDimSize(0);
size_t rank = getType().getRank();
for (size_t i = 0, e = numSparseIndices; i != e; ++i)
flatSparseIndices.push_back(getFlattenedIndex(
{&*std::next(sparseIndexValues.begin(), i * rank), rank}));
return flatSparseIndices;
}
LogicalResult
SparseElementsAttr::verify(function_ref<InFlightDiagnostic()> emitError,
ShapedType type, DenseIntElementsAttr sparseIndices,
DenseElementsAttr values) {
ShapedType valuesType = values.getType();
if (valuesType.getRank() != 1)
return emitError() << "expected 1-d tensor for sparse element values";
// Verify the indices and values shape.
ShapedType indicesType = sparseIndices.getType();
auto emitShapeError = [&]() {
return emitError() << "expected shape ([" << type.getShape()
<< "]); inferred shape of indices literal (["
<< indicesType.getShape()
<< "]); inferred shape of values literal (["
<< valuesType.getShape() << "])";
};
// Verify indices shape.
size_t rank = type.getRank(), indicesRank = indicesType.getRank();
if (indicesRank == 2) {
[mlir] Refactor ElementsAttr into an AttrInterface This revision refactors ElementsAttr into an Attribute Interface. This enables a common interface with which to interact with element attributes, without needing to modify the builtin dialect. It also removes a majority (if not all?) of the need for the current OpaqueElementsAttr, which was originally intended as a way to opaquely represent data that was not representable by the other builtin constructs. The new ElementsAttr interface not only allows for users to natively represent their data in the way that best suits them, it also allows for efficient opaque access and iteration of the underlying data. Attributes using the ElementsAttr interface can directly expose support for interacting with the held elements using any C++ data type they claim to support. For example, DenseIntOrFpElementsAttr supports iteration using various native C++ integer/float data types, as well as APInt/APFloat, and more. ElementsAttr instances that refer to DenseIntOrFpElementsAttr can use all of these data types for iteration: ```c++ DenseIntOrFpElementsAttr intElementsAttr = ...; ElementsAttr attr = intElementsAttr; for (uint64_t value : attr.getValues<uint64_t>()) ...; for (APInt value : attr.getValues<APInt>()) ...; for (IntegerAttr value : attr.getValues<IntegerAttr>()) ...; ``` ElementsAttr also supports failable range/iterator access, allowing for selective code paths depending on data type support: ```c++ ElementsAttr attr = ...; if (auto range = attr.tryGetValues<uint64_t>()) { for (uint64_t value : *range) ...; } ``` Differential Revision: https://reviews.llvm.org/D109190
2021-09-21 01:40:45 +00:00
if (indicesType.getDimSize(1) != static_cast<int64_t>(rank))
return emitShapeError();
} else if (indicesRank != 1 || rank != 1) {
return emitShapeError();
}
// Verify the values shape.
int64_t numSparseIndices = indicesType.getDimSize(0);
if (numSparseIndices != valuesType.getDimSize(0))
return emitShapeError();
// Verify that the sparse indices are within the value shape.
auto emitIndexError = [&](unsigned indexNum, ArrayRef<uint64_t> index) {
return emitError()
<< "sparse index #" << indexNum
<< " is not contained within the value shape, with index=[" << index
<< "], and type=" << type;
};
// Handle the case where the index values are a splat.
auto sparseIndexValues = sparseIndices.getValues<uint64_t>();
if (sparseIndices.isSplat()) {
SmallVector<uint64_t> indices(rank, *sparseIndexValues.begin());
if (!ElementsAttr::isValidIndex(type, indices))
return emitIndexError(0, indices);
return success();
}
// Otherwise, reinterpret each index as an ArrayRef.
for (size_t i = 0, e = numSparseIndices; i != e; ++i) {
ArrayRef<uint64_t> index(&*std::next(sparseIndexValues.begin(), i * rank),
rank);
if (!ElementsAttr::isValidIndex(type, index))
return emitIndexError(i, index);
}
return success();
}
//===----------------------------------------------------------------------===//
// TypeAttr
//===----------------------------------------------------------------------===//
void TypeAttr::walkImmediateSubElements(
function_ref<void(Attribute)> walkAttrsFn,
function_ref<void(Type)> walkTypesFn) const {
walkTypesFn(getValue());
}
Attribute
TypeAttr::replaceImmediateSubElements(ArrayRef<Attribute> replAttrs,
ArrayRef<Type> replTypes) const {
return get(replTypes[0]);
}