mirror of
https://github.com/llvm/llvm-project.git
synced 2025-04-25 08:46:06 +00:00

Currently, `DistinctAttr` uses an allocator wrapped in a `ThreadLocalCache` to manage attribute storage allocations. This ensures all allocations are freed when the allocator is destroyed. However, this setup can cause use-after-free errors when `mlir::PassManager` runs its passes on a separate thread as a result of crash reproduction being enabled. Distinct attribute storages are created in the child thread's local storage and freed once the thread joins. Attempting to access these attributes after this can result in segmentation faults, such as during printing or alias analysis. Example: This invocation of `mlir-opt` demonstrates the segfault issue due to distinct attributes being created in a child thread and their storage being freed once the thread joins: ``` mlir-opt --mlir-pass-pipeline-crash-reproducer=. --test-distinct-attrs mlir/test/IR/test-builtin-distinct-attrs.mlir ``` This pull request changes the distinct attribute allocator to use different allocators depending on whether or not threading is enabled and whether or not the pass manager is running its passes in a separate thread. If multithreading is disabled, a non thread-local allocator is used. If threading remains enabled and the pass manager invokes its pass pipelines in a child thread, then a non-thread local but synchronised allocator is used. This ensures that the lifetime of allocated storage persists beyond the lifetime of the child thread. I have added two tests for the `-test-distinct-attrs` pass and the `-enable-debug-info-on-llvm-scope` passes that run them with crash reproduction enabled.
464 lines
18 KiB
C++
464 lines
18 KiB
C++
//===- AttributeDetail.h - MLIR Affine Map details Class --------*- C++ -*-===//
|
|
//
|
|
// 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
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
//
|
|
// This holds implementation details of Attribute.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#ifndef ATTRIBUTEDETAIL_H_
|
|
#define ATTRIBUTEDETAIL_H_
|
|
|
|
#include "mlir/IR/AffineMap.h"
|
|
#include "mlir/IR/AttributeSupport.h"
|
|
#include "mlir/IR/BuiltinAttributes.h"
|
|
#include "mlir/IR/BuiltinTypes.h"
|
|
#include "mlir/IR/IntegerSet.h"
|
|
#include "mlir/IR/MLIRContext.h"
|
|
#include "mlir/Support/StorageUniquer.h"
|
|
#include "mlir/Support/ThreadLocalCache.h"
|
|
#include "llvm/ADT/APFloat.h"
|
|
#include "llvm/ADT/PointerIntPair.h"
|
|
#include "llvm/Support/TrailingObjects.h"
|
|
#include <mutex>
|
|
|
|
namespace mlir {
|
|
namespace detail {
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Elements Attributes
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
/// Return the bit width which DenseElementsAttr should use for this type.
|
|
inline size_t getDenseElementBitWidth(Type eltType) {
|
|
// Align the width for complex to 8 to make storage and interpretation easier.
|
|
if (ComplexType comp = llvm::dyn_cast<ComplexType>(eltType))
|
|
return llvm::alignTo<8>(getDenseElementBitWidth(comp.getElementType())) * 2;
|
|
if (eltType.isIndex())
|
|
return IndexType::kInternalStorageBitWidth;
|
|
return eltType.getIntOrFloatBitWidth();
|
|
}
|
|
|
|
/// An attribute representing a reference to a dense vector or tensor object.
|
|
struct DenseElementsAttributeStorage : public AttributeStorage {
|
|
public:
|
|
DenseElementsAttributeStorage(ShapedType type, bool isSplat)
|
|
: type(type), isSplat(isSplat) {}
|
|
|
|
ShapedType type;
|
|
bool isSplat;
|
|
};
|
|
|
|
/// An attribute representing a reference to a dense vector or tensor object.
|
|
struct DenseIntOrFPElementsAttrStorage : public DenseElementsAttributeStorage {
|
|
DenseIntOrFPElementsAttrStorage(ShapedType ty, ArrayRef<char> data,
|
|
bool isSplat = false)
|
|
: DenseElementsAttributeStorage(ty, isSplat), data(data) {}
|
|
|
|
struct KeyTy {
|
|
KeyTy(ShapedType type, ArrayRef<char> data, llvm::hash_code hashCode,
|
|
bool isSplat = false)
|
|
: type(type), data(data), hashCode(hashCode), isSplat(isSplat) {}
|
|
|
|
/// The type of the dense elements.
|
|
ShapedType type;
|
|
|
|
/// The raw buffer for the data storage.
|
|
ArrayRef<char> data;
|
|
|
|
/// The computed hash code for the storage data.
|
|
llvm::hash_code hashCode;
|
|
|
|
/// A boolean that indicates if this data is a splat or not.
|
|
bool isSplat;
|
|
};
|
|
|
|
/// Compare this storage instance with the provided key.
|
|
bool operator==(const KeyTy &key) const {
|
|
return key.type == type && key.data == data;
|
|
}
|
|
|
|
/// Construct a key from a shaped type, raw data buffer, and a flag that
|
|
/// signals if the data is already known to be a splat. Callers to this
|
|
/// function are expected to tag preknown splat values when possible, e.g. one
|
|
/// element shapes.
|
|
static KeyTy getKey(ShapedType ty, ArrayRef<char> data, bool isKnownSplat) {
|
|
// Handle an empty storage instance.
|
|
if (data.empty())
|
|
return KeyTy(ty, data, 0);
|
|
|
|
// If the data is already known to be a splat, the key hash value is
|
|
// directly the data buffer.
|
|
bool isBoolData = ty.getElementType().isInteger(1);
|
|
if (isKnownSplat) {
|
|
if (isBoolData)
|
|
return getKeyForSplatBoolData(ty, data[0] != 0);
|
|
return KeyTy(ty, data, llvm::hash_value(data), isKnownSplat);
|
|
}
|
|
|
|
// Otherwise, we need to check if the data corresponds to a splat or not.
|
|
|
|
// Handle the simple case of only one element.
|
|
size_t numElements = ty.getNumElements();
|
|
assert(numElements != 1 && "splat of 1 element should already be detected");
|
|
|
|
// Handle boolean values directly as they are packed to 1-bit.
|
|
if (isBoolData)
|
|
return getKeyForBoolData(ty, data, numElements);
|
|
|
|
size_t elementWidth = getDenseElementBitWidth(ty.getElementType());
|
|
// Non 1-bit dense elements are padded to 8-bits.
|
|
size_t storageSize = llvm::divideCeil(elementWidth, CHAR_BIT);
|
|
assert(((data.size() / storageSize) == numElements) &&
|
|
"data does not hold expected number of elements");
|
|
|
|
// Create the initial hash value with just the first element.
|
|
auto firstElt = data.take_front(storageSize);
|
|
auto hashVal = llvm::hash_value(firstElt);
|
|
|
|
// Check to see if this storage represents a splat. If it doesn't then
|
|
// combine the hash for the data starting with the first non splat element.
|
|
for (size_t i = storageSize, e = data.size(); i != e; i += storageSize)
|
|
if (memcmp(data.data(), &data[i], storageSize))
|
|
return KeyTy(ty, data, llvm::hash_combine(hashVal, data.drop_front(i)));
|
|
|
|
// Otherwise, this is a splat so just return the hash of the first element.
|
|
return KeyTy(ty, firstElt, hashVal, /*isSplat=*/true);
|
|
}
|
|
|
|
/// Construct a key with a set of boolean data.
|
|
static KeyTy getKeyForBoolData(ShapedType ty, ArrayRef<char> data,
|
|
size_t numElements) {
|
|
ArrayRef<char> splatData = data;
|
|
bool splatValue = splatData.front() & 1;
|
|
|
|
// Check the simple case where the data matches the known splat value.
|
|
if (splatData == ArrayRef<char>(splatValue ? kSplatTrue : kSplatFalse))
|
|
return getKeyForSplatBoolData(ty, splatValue);
|
|
|
|
// Handle the case where the potential splat value is 1 and the number of
|
|
// elements is non 8-bit aligned.
|
|
size_t numOddElements = numElements % CHAR_BIT;
|
|
if (splatValue && numOddElements != 0) {
|
|
// Check that all bits are set in the last value.
|
|
char lastElt = splatData.back();
|
|
if (lastElt != llvm::maskTrailingOnes<unsigned char>(numOddElements))
|
|
return KeyTy(ty, data, llvm::hash_value(data));
|
|
|
|
// If this is the only element, the data is known to be a splat.
|
|
if (splatData.size() == 1)
|
|
return getKeyForSplatBoolData(ty, splatValue);
|
|
splatData = splatData.drop_back();
|
|
}
|
|
|
|
// Check that the data buffer corresponds to a splat of the proper mask.
|
|
char mask = splatValue ? ~0 : 0;
|
|
return llvm::all_of(splatData, [mask](char c) { return c == mask; })
|
|
? getKeyForSplatBoolData(ty, splatValue)
|
|
: KeyTy(ty, data, llvm::hash_value(data));
|
|
}
|
|
|
|
/// Return a key to use for a boolean splat of the given value.
|
|
static KeyTy getKeyForSplatBoolData(ShapedType type, bool splatValue) {
|
|
const char &splatData = splatValue ? kSplatTrue : kSplatFalse;
|
|
return KeyTy(type, splatData, llvm::hash_value(splatData),
|
|
/*isSplat=*/true);
|
|
}
|
|
|
|
/// Hash the key for the storage.
|
|
static llvm::hash_code hashKey(const KeyTy &key) {
|
|
return llvm::hash_combine(key.type, key.hashCode);
|
|
}
|
|
|
|
/// Construct a new storage instance.
|
|
static DenseIntOrFPElementsAttrStorage *
|
|
construct(AttributeStorageAllocator &allocator, KeyTy key) {
|
|
// If the data buffer is non-empty, we copy it into the allocator with a
|
|
// 64-bit alignment.
|
|
ArrayRef<char> copy, data = key.data;
|
|
if (!data.empty()) {
|
|
char *rawData = reinterpret_cast<char *>(
|
|
allocator.allocate(data.size(), alignof(uint64_t)));
|
|
std::memcpy(rawData, data.data(), data.size());
|
|
copy = ArrayRef<char>(rawData, data.size());
|
|
}
|
|
|
|
return new (allocator.allocate<DenseIntOrFPElementsAttrStorage>())
|
|
DenseIntOrFPElementsAttrStorage(key.type, copy, key.isSplat);
|
|
}
|
|
|
|
ArrayRef<char> data;
|
|
|
|
/// The values used to denote a boolean splat value.
|
|
// This is not using constexpr declaration due to compilation failure
|
|
// encountered with MSVC where it would inline these values, which makes it
|
|
// unsafe to refer by reference in KeyTy.
|
|
static const char kSplatTrue;
|
|
static const char kSplatFalse;
|
|
};
|
|
|
|
/// An attribute representing a reference to a dense vector or tensor object
|
|
/// containing strings.
|
|
struct DenseStringElementsAttrStorage : public DenseElementsAttributeStorage {
|
|
DenseStringElementsAttrStorage(ShapedType ty, ArrayRef<StringRef> data,
|
|
bool isSplat = false)
|
|
: DenseElementsAttributeStorage(ty, isSplat), data(data) {}
|
|
|
|
struct KeyTy {
|
|
KeyTy(ShapedType type, ArrayRef<StringRef> data, llvm::hash_code hashCode,
|
|
bool isSplat = false)
|
|
: type(type), data(data), hashCode(hashCode), isSplat(isSplat) {}
|
|
|
|
/// The type of the dense elements.
|
|
ShapedType type;
|
|
|
|
/// The raw buffer for the data storage.
|
|
ArrayRef<StringRef> data;
|
|
|
|
/// The computed hash code for the storage data.
|
|
llvm::hash_code hashCode;
|
|
|
|
/// A boolean that indicates if this data is a splat or not.
|
|
bool isSplat;
|
|
};
|
|
|
|
/// Compare this storage instance with the provided key.
|
|
bool operator==(const KeyTy &key) const {
|
|
if (key.type != type)
|
|
return false;
|
|
|
|
// Otherwise, we can default to just checking the data. StringRefs compare
|
|
// by contents.
|
|
return key.data == data;
|
|
}
|
|
|
|
/// Construct a key from a shaped type, StringRef data buffer, and a flag that
|
|
/// signals if the data is already known to be a splat. Callers to this
|
|
/// function are expected to tag preknown splat values when possible, e.g. one
|
|
/// element shapes.
|
|
static KeyTy getKey(ShapedType ty, ArrayRef<StringRef> data,
|
|
bool isKnownSplat) {
|
|
// Handle an empty storage instance.
|
|
if (data.empty())
|
|
return KeyTy(ty, data, 0);
|
|
|
|
// If the data is already known to be a splat, the key hash value is
|
|
// directly the data buffer.
|
|
if (isKnownSplat)
|
|
return KeyTy(ty, data, llvm::hash_value(data.front()), isKnownSplat);
|
|
|
|
// Handle the simple case of only one element.
|
|
assert(ty.getNumElements() != 1 &&
|
|
"splat of 1 element should already be detected");
|
|
|
|
// Create the initial hash value with just the first element.
|
|
const auto &firstElt = data.front();
|
|
auto hashVal = llvm::hash_value(firstElt);
|
|
|
|
// Check to see if this storage represents a splat. If it doesn't then
|
|
// combine the hash for the data starting with the first non splat element.
|
|
for (size_t i = 1, e = data.size(); i != e; i++)
|
|
if (firstElt != data[i])
|
|
return KeyTy(ty, data, llvm::hash_combine(hashVal, data.drop_front(i)));
|
|
|
|
// Otherwise, this is a splat so just return the hash of the first element.
|
|
return KeyTy(ty, data.take_front(), hashVal, /*isSplat=*/true);
|
|
}
|
|
|
|
/// Hash the key for the storage.
|
|
static llvm::hash_code hashKey(const KeyTy &key) {
|
|
return llvm::hash_combine(key.type, key.hashCode);
|
|
}
|
|
|
|
/// Construct a new storage instance.
|
|
static DenseStringElementsAttrStorage *
|
|
construct(AttributeStorageAllocator &allocator, KeyTy key) {
|
|
// If the data buffer is non-empty, we copy it into the allocator with a
|
|
// 64-bit alignment.
|
|
ArrayRef<StringRef> copy, data = key.data;
|
|
if (data.empty()) {
|
|
return new (allocator.allocate<DenseStringElementsAttrStorage>())
|
|
DenseStringElementsAttrStorage(key.type, copy, key.isSplat);
|
|
}
|
|
|
|
int numEntries = key.isSplat ? 1 : data.size();
|
|
|
|
// Compute the amount data needed to store the ArrayRef and StringRef
|
|
// contents.
|
|
size_t dataSize = sizeof(StringRef) * numEntries;
|
|
for (int i = 0; i < numEntries; i++)
|
|
dataSize += data[i].size();
|
|
|
|
char *rawData = reinterpret_cast<char *>(
|
|
allocator.allocate(dataSize, alignof(uint64_t)));
|
|
|
|
// Setup a mutable array ref of our string refs so that we can update their
|
|
// contents.
|
|
auto mutableCopy = MutableArrayRef<StringRef>(
|
|
reinterpret_cast<StringRef *>(rawData), numEntries);
|
|
auto *stringData = rawData + numEntries * sizeof(StringRef);
|
|
|
|
for (int i = 0; i < numEntries; i++) {
|
|
memcpy(stringData, data[i].data(), data[i].size());
|
|
mutableCopy[i] = StringRef(stringData, data[i].size());
|
|
stringData += data[i].size();
|
|
}
|
|
|
|
copy =
|
|
ArrayRef<StringRef>(reinterpret_cast<StringRef *>(rawData), numEntries);
|
|
|
|
return new (allocator.allocate<DenseStringElementsAttrStorage>())
|
|
DenseStringElementsAttrStorage(key.type, copy, key.isSplat);
|
|
}
|
|
|
|
ArrayRef<StringRef> data;
|
|
};
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// StringAttr
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
struct StringAttrStorage : public AttributeStorage {
|
|
StringAttrStorage(StringRef value, Type type)
|
|
: type(type), value(value), referencedDialect(nullptr) {}
|
|
|
|
/// The hash key is a tuple of the parameter types.
|
|
using KeyTy = std::pair<StringRef, Type>;
|
|
bool operator==(const KeyTy &key) const {
|
|
return value == key.first && type == key.second;
|
|
}
|
|
static ::llvm::hash_code hashKey(const KeyTy &key) {
|
|
return DenseMapInfo<KeyTy>::getHashValue(key);
|
|
}
|
|
|
|
/// Define a construction method for creating a new instance of this
|
|
/// storage.
|
|
static StringAttrStorage *construct(AttributeStorageAllocator &allocator,
|
|
const KeyTy &key) {
|
|
return new (allocator.allocate<StringAttrStorage>())
|
|
StringAttrStorage(allocator.copyInto(key.first), key.second);
|
|
}
|
|
|
|
/// Initialize the storage given an MLIRContext.
|
|
void initialize(MLIRContext *context);
|
|
|
|
/// The type of the string.
|
|
Type type;
|
|
/// The raw string value.
|
|
StringRef value;
|
|
/// If the string value contains a dialect namespace prefix (e.g.
|
|
/// dialect.blah), this is the dialect referenced.
|
|
Dialect *referencedDialect;
|
|
};
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// DistinctAttr
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
/// An attribute to store a distinct reference to another attribute.
|
|
struct DistinctAttrStorage : public AttributeStorage {
|
|
using KeyTy = Attribute;
|
|
|
|
DistinctAttrStorage(Attribute referencedAttr)
|
|
: referencedAttr(referencedAttr) {}
|
|
|
|
/// Returns the referenced attribute as key.
|
|
KeyTy getAsKey() const { return KeyTy(referencedAttr); }
|
|
|
|
/// The referenced attribute.
|
|
Attribute referencedAttr;
|
|
};
|
|
|
|
/// A specialized attribute uniquer for distinct attributes that always
|
|
/// allocates since the distinct attribute instances use the address of their
|
|
/// storage as unique identifier.
|
|
class DistinctAttributeUniquer {
|
|
public:
|
|
/// Creates a distinct attribute storage. Allocates every time since the
|
|
/// address of the storage serves as unique identifier.
|
|
template <typename T, typename... Args>
|
|
static T get(MLIRContext *context, Args &&...args) {
|
|
static_assert(std::is_same_v<typename T::ImplType, DistinctAttrStorage>,
|
|
"expects a distinct attribute storage");
|
|
DistinctAttrStorage *storage = DistinctAttributeUniquer::allocateStorage(
|
|
context, std::forward<Args>(args)...);
|
|
storage->initializeAbstractAttribute(
|
|
AbstractAttribute::lookup(DistinctAttr::getTypeID(), context));
|
|
return storage;
|
|
}
|
|
|
|
private:
|
|
/// Allocates a distinct attribute storage.
|
|
static DistinctAttrStorage *allocateStorage(MLIRContext *context,
|
|
Attribute referencedAttr);
|
|
};
|
|
|
|
/// An allocator for distinct attribute storage instances. It uses thread local
|
|
/// bump pointer allocators stored in a thread local cache to ensure the storage
|
|
/// is freed after the destruction of the distinct attribute allocator.
|
|
class DistinctAttributeAllocator {
|
|
public:
|
|
DistinctAttributeAllocator(bool threadingIsEnabled)
|
|
: threadingIsEnabled(threadingIsEnabled), useThreadLocalAllocator(true) {};
|
|
|
|
DistinctAttributeAllocator(DistinctAttributeAllocator &&) = delete;
|
|
DistinctAttributeAllocator(const DistinctAttributeAllocator &) = delete;
|
|
DistinctAttributeAllocator &
|
|
operator=(const DistinctAttributeAllocator &) = delete;
|
|
|
|
/// Allocates a distinct attribute storage using a thread local bump pointer
|
|
/// allocator to enable synchronization free parallel allocations.
|
|
DistinctAttrStorage *allocate(Attribute referencedAttr) {
|
|
if (!useThreadLocalAllocator && threadingIsEnabled) {
|
|
std::scoped_lock<std::mutex> lock(allocatorMutex);
|
|
return allocateImpl(referencedAttr);
|
|
}
|
|
return allocateImpl(referencedAttr);
|
|
}
|
|
|
|
/// Sets a flag that stores if multithreading is enabled. The flag is used to
|
|
/// decide if locking is needed when using a non thread-safe allocator.
|
|
void disableMultiThreading(bool disable = true) {
|
|
threadingIsEnabled = !disable;
|
|
}
|
|
|
|
/// Sets a flag to disable using thread local bump pointer allocators and use
|
|
/// a single thread-safe allocator. Use this to persist allocated storage
|
|
/// beyond the lifetime of a child thread calling this function while ensuring
|
|
/// thread-safe allocation.
|
|
void disableThreadLocalStorage(bool disable = true) {
|
|
useThreadLocalAllocator = !disable;
|
|
}
|
|
|
|
private:
|
|
DistinctAttrStorage *allocateImpl(Attribute referencedAttr) {
|
|
return new (getAllocatorInUse().Allocate<DistinctAttrStorage>())
|
|
DistinctAttrStorage(referencedAttr);
|
|
}
|
|
|
|
/// If threading is disabled on the owning MLIR context, a normal non
|
|
/// thread-local, non-thread safe bump pointer allocator is used instead to
|
|
/// prevent use-after-free errors whenever attribute storage created on a
|
|
/// crash recover thread is accessed after the thread joins.
|
|
llvm::BumpPtrAllocator &getAllocatorInUse() {
|
|
if (useThreadLocalAllocator)
|
|
return allocatorCache.get();
|
|
return allocator;
|
|
}
|
|
|
|
ThreadLocalCache<llvm::BumpPtrAllocator> allocatorCache;
|
|
llvm::BumpPtrAllocator allocator;
|
|
std::mutex allocatorMutex;
|
|
|
|
bool threadingIsEnabled : 1;
|
|
bool useThreadLocalAllocator : 1;
|
|
};
|
|
} // namespace detail
|
|
} // namespace mlir
|
|
|
|
#endif // ATTRIBUTEDETAIL_H_
|