Skip to content

FORMUS3IC_LAS3 / embb

Go to a project
Commit 812a9a26 by lucapegolotti
Options

Fix style of linearizability_tester.h

parent cc1e3c42
Inline Side-by-side
Showing with 1807 additions and 1807 deletions
linearizability_tester/src/linearizability_tester.h
...@@ -70,7 +70,7 @@ ...@@ -70,7 +70,7 @@
template<class T, class ...Args> template<class T, class ...Args>
std::unique_ptr<T> make_unique(Args&& ...args) std::unique_ptr<T> make_unique(Args&& ...args)
{ {
return std::unique_ptr<T>(new T(std::forward<Args>(args)...)); return std::unique_ptr<T>(new T(std::forward<Args>(args)...));
} }
#else #else
using std::make_unique; using std::make_unique;
...@@ -81,1917 +81,1917 @@ using std::make_unique; ...@@ -81,1917 +81,1917 @@ using std::make_unique;
/// Linearizability tester /// Linearizability tester
namespace lt namespace lt
{ {
/************* Core data structures && algorithms *************/ /************* Core data structures && algorithms *************/
template<class S> template<class S>
class Entry; class Entry;
/// Doubly-linked list of log entries /// Doubly-linked list of log entries
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
using EntryPtr = Entry<S>*; using EntryPtr = Entry<S>*;
/// Bounded stack of call entries that have been linearized /// Bounded stack of call entries that have been linearized
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
class Stack class Stack
{ {
private: private:
typedef std::tuple<EntryPtr<S>, S> Pair; typedef std::tuple<EntryPtr<S>, S> Pair;
typedef std::vector<Pair> Pairs; typedef std::vector<Pair> Pairs;
typedef typename Pairs::size_type SizeType; typedef typename Pairs::size_type SizeType;
// A constant-size vector // A constant-size vector
Pairs m_vector; Pairs m_vector;
SizeType m_top; SizeType m_top;
public: public:
/// Create a new stack of bounded height /// Create a new stack of bounded height
/// \post: if capacity is positive, then !is_full() /// \post: if capacity is positive, then !is_full()
Stack(SizeType capacity) Stack(SizeType capacity)
: m_vector(capacity), m_top{ 0U } : m_vector(capacity), m_top{ 0U }
{ {
assert(capacity == 0U || !is_full()); assert(capacity == 0U || !is_full());
} }
/// History length in the stack /// History length in the stack
SizeType size() const noexcept SizeType size() const noexcept
{ {
return m_top; return m_top;
} }
/// Is size() zero? /// Is size() zero?
bool is_empty() const noexcept bool is_empty() const noexcept
{ {
return 0U == size(); return 0U == size();
} }
/// Is size() equal to the stack's capacity? /// Is size() equal to the stack's capacity?
bool is_full() const noexcept bool is_full() const noexcept
{ {
return m_top == m_vector.size(); return m_top == m_vector.size();
} }
/// \pre: !is_empty() /// \pre: !is_empty()
const Pair& top() const noexcept const Pair& top() const noexcept
{ {
assert(!is_empty()); assert(!is_empty());
return m_vector[m_top - 1U]; return m_vector[m_top - 1U];
} }
/// Add an entry to the top() of the stack /// Add an entry to the top() of the stack
/// \pre: !is_full() /// \pre: !is_full()
/// \pre: ptr->is_call() /// \pre: ptr->is_call()
void push(EntryPtr<S>, S&&); void push(EntryPtr<S>, S&&);
/// Remove count entries from the stack /// Remove count entries from the stack
/// \pre: 0 < count <= size() /// \pre: 0 < count <= size()
void pop(unsigned count = 1U) void pop(unsigned count = 1U)
{ {
assert(0U < count); assert(0U < count);
assert(count <= size()); assert(count <= size());
m_top -= count; m_top -= count;
} }
/// \internal /// \internal
EntryPtr<S> entry_ptr(std::size_t pos) EntryPtr<S> entry_ptr(std::size_t pos)
{ {
assert(pos < m_top); assert(pos < m_top);
return std::get<0>(m_vector[pos]); return std::get<0>(m_vector[pos]);
} }
}; };
enum class Option : unsigned char enum class Option : unsigned char
{ {
NEVER_CACHE, NEVER_CACHE,
LRU_CACHE, LRU_CACHE,
ALWAYS_CACHE, ALWAYS_CACHE,
}; };
template<class S> class Entry; template<class S> class Entry;
template<class S> class Log; template<class S> class Log;
template<class S> class ConcurrentLog; template<class S> class ConcurrentLog;
template<class S> class Slicer; template<class S> class Slicer;
template<class S, Option> class LinearizabilityTester; template<class S, Option> class LinearizabilityTester;
/// A kind of "functor" in C++ terminology /// A kind of "functor" in C++ terminology
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
class Op class Op
{ {
private: private:
friend class Entry<S>; friend class Entry<S>;
// Is m_partition defined? // Is m_partition defined?
bool m_is_partitionable; bool m_is_partitionable;
unsigned m_partition; unsigned m_partition;
// modified by Entry // modified by Entry
unsigned ref_counter; unsigned ref_counter;
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
virtual std::ostream& print(std::ostream&) const = 0; virtual std::ostream& print(std::ostream&) const = 0;
#endif #endif
virtual std::pair<bool, S> internal_apply(const S&, const Op<S>&) virtual std::pair<bool, S> internal_apply(const S&, const Op<S>&)
{ {
return{}; return{};
} }
public: public:
Op() Op()
: m_is_partitionable{ false }, : m_is_partitionable{ false },
m_partition{ 0U }, m_partition{ 0U },
ref_counter{ 0U } {} ref_counter{ 0U } {}
Op(unsigned partition) Op(unsigned partition)
: m_is_partitionable{ true }, : m_is_partitionable{ true },
m_partition{ partition }, m_partition{ partition },
ref_counter{ 0U } {} ref_counter{ 0U } {}
Op(bool is_partitionable, unsigned partition) Op(bool is_partitionable, unsigned partition)
: m_is_partitionable{ is_partitionable }, : m_is_partitionable{ is_partitionable },
m_partition{ partition }, m_partition{ partition },
ref_counter{ 0U } {} ref_counter{ 0U } {}
virtual ~Op() virtual ~Op()
{ {
assert(ref_counter == 0); assert(ref_counter == 0);
} }
/// Is partition() defined? /// Is partition() defined?
bool is_partitionable() const noexcept bool is_partitionable() const noexcept
{ {
return m_is_partitionable; return m_is_partitionable;
} }
/// \pre: is_partitionable() /// \pre: is_partitionable()
unsigned partition() const unsigned partition() const
{ {
assert(m_is_partitionable); assert(m_is_partitionable);
return m_partition; return m_partition;
} }
/// Returns true exactly if the operation could be applied /// Returns true exactly if the operation could be applied
std::pair<bool, S> apply(const S& s, const Op<S>& op) std::pair<bool, S> apply(const S& s, const Op<S>& op)
{ {
return internal_apply(s, op); return internal_apply(s, op);
} }
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
friend std::ostream& operator<<(std::ostream& os, const Op& op) friend std::ostream& operator<<(std::ostream& os, const Op& op)
{ {
return op.print(os); return op.print(os);
} }
#endif #endif
}; };
/// Fixed-size set of bits with persistence features /// Fixed-size set of bits with persistence features
class Bitset class Bitset
{ {
public: public:
typedef std::size_t Pos; typedef std::size_t Pos;
private: private:
friend struct BitsetHash; friend struct BitsetHash;
friend class FlexibleBitset; friend class FlexibleBitset;
typedef unsigned long Block; typedef unsigned long Block;
typedef std::vector<Block> Blocks; typedef std::vector<Block> Blocks;
typedef Blocks::size_type BlockIndex; typedef Blocks::size_type BlockIndex;
/// Accessible bits in a Block /// Accessible bits in a Block
typedef unsigned BlockWidth; typedef unsigned BlockWidth;
static constexpr BlockWidth s_bits_per_block = static constexpr BlockWidth s_bits_per_block =
static_cast<BlockWidth>(sizeof(Block) * CHAR_BIT); static_cast<BlockWidth>(sizeof(Block) * CHAR_BIT);
static BlockIndex block_index(Pos pos) noexcept static BlockIndex block_index(Pos pos) noexcept
{ {
return pos / s_bits_per_block; return pos / s_bits_per_block;
} }
static BlockIndex blocks_size(Pos max_pos) noexcept static BlockIndex blocks_size(Pos max_pos) noexcept
{ {
return block_index(max_pos) + 1U; return block_index(max_pos) + 1U;
} }
static BlockWidth bit_index(Pos pos) noexcept static BlockWidth bit_index(Pos pos) noexcept
{ {
return static_cast<BlockWidth>(pos % s_bits_per_block); return static_cast<BlockWidth>(pos % s_bits_per_block);
} }
static Block bit_mask(Pos pos) noexcept static Block bit_mask(Pos pos) noexcept
{ {
return Block(1U) << bit_index(pos); return Block(1U) << bit_index(pos);
} }
/// only resized by FlexibleBitset /// only resized by FlexibleBitset
Blocks m_blocks; Blocks m_blocks;
std::size_t m_hash; std::size_t m_hash;
unsigned m_number_of_set_bits; unsigned m_number_of_set_bits;
Block& find_block(Pos pos) Block& find_block(Pos pos)
{ {
BlockIndex i{ block_index(pos) }; BlockIndex i{ block_index(pos) };
assert(i < m_blocks.size()); assert(i < m_blocks.size());
return m_blocks[i]; return m_blocks[i];
} }
// We exploit the fact that XOR forms an abelian group: // We exploit the fact that XOR forms an abelian group:
// first, clear hash of old block; then, hash new block. // first, clear hash of old block; then, hash new block.
void update_hash(Block old_block, Block new_block) void update_hash(Block old_block, Block new_block)
{ {
m_hash ^= old_block; m_hash ^= old_block;
m_hash ^= new_block; m_hash ^= new_block;
} }
public: public:
Bitset(Pos max_pos) Bitset(Pos max_pos)
: m_blocks(blocks_size(max_pos)), : m_blocks(blocks_size(max_pos)),
m_hash{ 0U }, m_hash{ 0U },
m_number_of_set_bits{ 0U } {} m_number_of_set_bits{ 0U } {}
bool is_empty() const noexcept bool is_empty() const noexcept
{ {
return m_number_of_set_bits == 0U; return m_number_of_set_bits == 0U;
} }
bool set(Pos pos) bool set(Pos pos)
{ {
Block& block = find_block(pos); Block& block = find_block(pos);
const Block copy_block{ block }; const Block copy_block{ block };
block |= bit_mask(pos); block |= bit_mask(pos);
update_hash(copy_block, block); update_hash(copy_block, block);
bool ok{ block != copy_block }; bool ok{ block != copy_block };
m_number_of_set_bits += ok; m_number_of_set_bits += ok;
return ok; return ok;
} }
Bitset immutable_set(Pos pos) const Bitset immutable_set(Pos pos) const
{ {
Bitset copy{ *this }; Bitset copy{ *this };
copy.set(pos); copy.set(pos);
return copy; return copy;
} }
bool is_set(Pos pos) const bool is_set(Pos pos) const
{ {
BlockIndex i{ block_index(pos) }; BlockIndex i{ block_index(pos) };
if (i < m_blocks.size()) if (i < m_blocks.size())
return (m_blocks[i] & bit_mask(pos)) != 0U; return (m_blocks[i] & bit_mask(pos)) != 0U;
return false; return false;
} }
bool reset(Pos pos) bool reset(Pos pos)
{ {
Block& block = find_block(pos); Block& block = find_block(pos);
const Block copy_block{ block }; const Block copy_block{ block };
block &= ~bit_mask(pos); block &= ~bit_mask(pos);
update_hash(copy_block, block); update_hash(copy_block, block);
bool ok{ block != copy_block }; bool ok{ block != copy_block };
m_number_of_set_bits -= ok; m_number_of_set_bits -= ok;
return ok; return ok;
} }
Bitset immutable_reset(Pos pos) const Bitset immutable_reset(Pos pos) const
{ {
Bitset copy{ *this }; Bitset copy{ *this };
copy.reset(pos); copy.reset(pos);
return copy; return copy;
} }
// Same number of blocks && identical bits in all those blocks? // Same number of blocks && identical bits in all those blocks?
bool operator==(const Bitset& other) const noexcept bool operator==(const Bitset& other) const noexcept
{ {
return m_number_of_set_bits == other.m_number_of_set_bits && return m_number_of_set_bits == other.m_number_of_set_bits &&
m_blocks == other.m_blocks; m_blocks == other.m_blocks;
} }
bool operator!=(const Bitset& other) const noexcept bool operator!=(const Bitset& other) const noexcept
{ {
return m_number_of_set_bits != other.m_number_of_set_bits || return m_number_of_set_bits != other.m_number_of_set_bits ||
m_blocks != other.m_blocks; m_blocks != other.m_blocks;
} }
}; };
/// Constant-time, O(1), hash function /// Constant-time, O(1), hash function
struct BitsetHash struct BitsetHash
{ {
std::size_t operator()(const Bitset& bitset) const noexcept std::size_t operator()(const Bitset& bitset) const noexcept
{ {
return bitset.m_hash; return bitset.m_hash;
} }
}; };
/// States of abstract data types /// States of abstract data types
namespace state namespace state
{ {
template<class T> template<class T>
struct Hash struct Hash
{ {
std::size_t operator()(const T&) const noexcept; std::size_t operator()(const T&) const noexcept;
}; };
} }
template<class S> template<class S>
using OpPtr = std::unique_ptr<Op<S>>; using OpPtr = std::unique_ptr<Op<S>>;
/// Call/ret log entry /// Call/ret log entry
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
class Entry class Entry
{ {
private: private:
friend class Log<S>; friend class Log<S>;
friend class Slicer<S>; friend class Slicer<S>;
friend class LinearizabilityTester<S, Option::NEVER_CACHE>; friend class LinearizabilityTester<S, Option::NEVER_CACHE>;
friend class LinearizabilityTester<S, Option::LRU_CACHE>; friend class LinearizabilityTester<S, Option::LRU_CACHE>;
friend class LinearizabilityTester<S, Option::ALWAYS_CACHE>; friend class LinearizabilityTester<S, Option::ALWAYS_CACHE>;
// Ref counted pointer because we need to copy logs so that we // Ref counted pointer because we need to copy logs so that we
// can experimentally compare different linearizability testers // can experimentally compare different linearizability testers
// //
// However, this is an implementation detail and the strict type // However, this is an implementation detail and the strict type
// of OpPtr<S> enforces at compile-time that we manage the // of OpPtr<S> enforces at compile-time that we manage the
// ownership of these kind of pointers on the user's behalf. // ownership of these kind of pointers on the user's behalf.
Op<S>* m_op_ptr; Op<S>* m_op_ptr;
unsigned m_entry_id; unsigned m_entry_id;
std::thread::id m_thread_id; std::thread::id m_thread_id;
EntryPtr<S> m_match; EntryPtr<S> m_match;
bool m_is_call; bool m_is_call;
void inc_ref_counter() const noexcept void inc_ref_counter() const noexcept
{ {
if (m_op_ptr != nullptr) if (m_op_ptr != nullptr)
++m_op_ptr->ref_counter; ++m_op_ptr->ref_counter;
} }
void dec_ref_counter() const void dec_ref_counter() const
{ {
assert(m_op_ptr == nullptr || 0 < m_op_ptr->ref_counter); assert(m_op_ptr == nullptr || 0 < m_op_ptr->ref_counter);
if (m_op_ptr != nullptr && --m_op_ptr->ref_counter == 0) if (m_op_ptr != nullptr && --m_op_ptr->ref_counter == 0)
delete m_op_ptr; delete m_op_ptr;
} }
/// Log head /// Log head
/// \post: if _next is !nullptr, then _next->prev == this /// \post: if _next is !nullptr, then _next->prev == this
Entry(EntryPtr<S> _next) Entry(EntryPtr<S> _next)
: m_op_ptr{ nullptr }, : m_op_ptr{ nullptr },
m_entry_id{}, m_entry_id{},
m_thread_id{}, m_thread_id{},
m_match{ nullptr }, m_match{ nullptr },
m_is_call{ false }, m_is_call{ false },
prev{ nullptr }, prev{ nullptr },
next{ _next } next{ _next }
{ {
if (_next != nullptr) if (_next != nullptr)
_next->prev = this; _next->prev = this;
} }
public: public:
~Entry() ~Entry()
{ {
dec_ref_counter(); dec_ref_counter();
} }
EntryPtr<S> prev; EntryPtr<S> prev;
EntryPtr<S> next; EntryPtr<S> next;
Entry() Entry()
: m_op_ptr{ nullptr }, : m_op_ptr{ nullptr },
m_entry_id{}, m_entry_id{},
m_thread_id{}, m_thread_id{},
m_match{ nullptr }, m_match{ nullptr },
m_is_call{ false }, m_is_call{ false },
prev{ nullptr }, prev{ nullptr },
next{ nullptr } {} next{ nullptr } {}
Entry(const Entry& entry) Entry(const Entry& entry)
: m_op_ptr{ entry.m_op_ptr }, : m_op_ptr{ entry.m_op_ptr },
m_entry_id{ entry.m_entry_id }, m_entry_id{ entry.m_entry_id },
m_thread_id{ entry.m_thread_id }, m_thread_id{ entry.m_thread_id },
m_match{ entry.m_match }, m_match{ entry.m_match },
m_is_call{ entry.m_is_call }, m_is_call{ entry.m_is_call },
prev{ entry.prev }, prev{ entry.prev },
next{ entry.next } next{ entry.next }
{ {
inc_ref_counter(); inc_ref_counter();
} }
Entry& operator=(const Entry& entry) Entry& operator=(const Entry& entry)
{ {
entry.inc_ref_counter(); entry.inc_ref_counter();
dec_ref_counter(); dec_ref_counter();
m_op_ptr = entry.m_op_ptr; m_op_ptr = entry.m_op_ptr;
m_entry_id = entry.m_entry_id; m_entry_id = entry.m_entry_id;
m_thread_id = entry.m_thread_id; m_thread_id = entry.m_thread_id;
m_match = entry.m_match; m_match = entry.m_match;
m_is_call = entry.m_is_call; m_is_call = entry.m_is_call;
prev = entry.prev; prev = entry.prev;
next = entry.next; next = entry.next;
return *this; return *this;
} }
Entry& operator=(Entry&& entry) Entry& operator=(Entry&& entry)
{ {
// only decrement required (due to move semantics) // only decrement required (due to move semantics)
dec_ref_counter(); dec_ref_counter();
m_op_ptr = entry.m_op_ptr; m_op_ptr = entry.m_op_ptr;
m_entry_id = entry.m_entry_id; m_entry_id = entry.m_entry_id;
m_thread_id = entry.m_thread_id; m_thread_id = entry.m_thread_id;
m_match = entry.m_match; m_match = entry.m_match;
m_is_call = entry.m_is_call; m_is_call = entry.m_is_call;
prev = entry.prev; prev = entry.prev;
next = entry.next; next = entry.next;
entry.m_op_ptr = nullptr; entry.m_op_ptr = nullptr;
entry.m_entry_id = 0; entry.m_entry_id = 0;
entry.m_thread_id = 0; entry.m_thread_id = 0;
entry.m_match = nullptr; entry.m_match = nullptr;
entry.m_is_call = false; entry.m_is_call = false;
entry.prev = nullptr; entry.prev = nullptr;
entry.next = nullptr; entry.next = nullptr;
return *this; return *this;
} }
/// \pre: set_match && set_op have been called with non-null arguments /// \pre: set_match && set_op have been called with non-null arguments
bool is_partitionable() const bool is_partitionable() const
{ {
assert(m_match != nullptr); assert(m_match != nullptr);
assert(m_match->m_op_ptr != nullptr); assert(m_match->m_op_ptr != nullptr);
assert(m_op_ptr->m_is_partitionable == m_match->m_op_ptr->m_is_partitionable); assert(m_op_ptr->m_is_partitionable == m_match->m_op_ptr->m_is_partitionable);
assert(m_op_ptr->m_partition == m_match->m_op_ptr->m_partition); assert(m_op_ptr->m_partition == m_match->m_op_ptr->m_partition);
return m_op_ptr->m_is_partitionable; return m_op_ptr->m_is_partitionable;
} }
void set_op(OpPtr<S>&& op_ptr) noexcept void set_op(OpPtr<S>&& op_ptr) noexcept
{ {
m_op_ptr = op_ptr.release(); m_op_ptr = op_ptr.release();
inc_ref_counter(); inc_ref_counter();
} }
Op<S>& op() const Op<S>& op() const
{ {
assert(m_op_ptr != nullptr); assert(m_op_ptr != nullptr);
return *m_op_ptr; return *m_op_ptr;
} }
const Op<S>* const op_ptr() const noexcept const Op<S>* const op_ptr() const noexcept
{ {
return m_op_ptr; return m_op_ptr;
} }
void set_entry_id(unsigned entry_id) noexcept void set_entry_id(unsigned entry_id) noexcept
{ {
m_entry_id = entry_id; m_entry_id = entry_id;
} }
unsigned entry_id() const noexcept unsigned entry_id() const noexcept
{ {
return m_entry_id; return m_entry_id;
} }
void set_thread_id(std::thread::id thread_id) noexcept void set_thread_id(std::thread::id thread_id) noexcept
{ {
m_thread_id = thread_id; m_thread_id = thread_id;
} }
std::thread::id thread_id() const noexcept std::thread::id thread_id() const noexcept
{ {
return m_thread_id; return m_thread_id;
} }
/// \pre: ret_entry_ptr->match() == nullptr /// \pre: ret_entry_ptr->match() == nullptr
/// \pre: !ret_entry_ptr->is_call() /// \pre: !ret_entry_ptr->is_call()
/// ///
/// \post: this->is_call() /// \post: this->is_call()
/// \post: this == ret_entry_ptr->match() /// \post: this == ret_entry_ptr->match()
/// \post: this->match() == ret_entry_ptr /// \post: this->match() == ret_entry_ptr
/// \post: this->entry_id() == ret_entry_ptr->entry_id() /// \post: this->entry_id() == ret_entry_ptr->entry_id()
/// \post: if this->is_partitionable() || ret_entry_ptr->is_partitionable(), /// \post: if this->is_partitionable() || ret_entry_ptr->is_partitionable(),
/// then this->op().partition() == ret_entry_ptr->op().partition() /// then this->op().partition() == ret_entry_ptr->op().partition()
void set_match(EntryPtr<S> ret_entry_ptr) noexcept void set_match(EntryPtr<S> ret_entry_ptr) noexcept
{ {
assert(ret_entry_ptr->m_match == nullptr); assert(ret_entry_ptr->m_match == nullptr);
assert(!ret_entry_ptr->is_call()); assert(!ret_entry_ptr->is_call());
ret_entry_ptr->m_match = this; ret_entry_ptr->m_match = this;
ret_entry_ptr->set_entry_id(m_entry_id); ret_entry_ptr->set_entry_id(m_entry_id);
if (ret_entry_ptr->op().m_is_partitionable) if (ret_entry_ptr->op().m_is_partitionable)
{ {
op().m_is_partitionable = ret_entry_ptr->op().m_is_partitionable; op().m_is_partitionable = ret_entry_ptr->op().m_is_partitionable;
op().m_partition = ret_entry_ptr->op().m_partition; op().m_partition = ret_entry_ptr->op().m_partition;
} }
else else
{ {
ret_entry_ptr->op().m_is_partitionable = op().m_is_partitionable; ret_entry_ptr->op().m_is_partitionable = op().m_is_partitionable;
ret_entry_ptr->op().m_partition = op().m_partition; ret_entry_ptr->op().m_partition = op().m_partition;
} }
m_match = ret_entry_ptr; m_match = ret_entry_ptr;
m_is_call = true; m_is_call = true;
assert(is_call()); assert(is_call());
assert(this == ret_entry_ptr->match()); assert(this == ret_entry_ptr->match());
assert(match() == ret_entry_ptr); assert(match() == ret_entry_ptr);
assert(entry_id() == ret_entry_ptr->entry_id()); assert(entry_id() == ret_entry_ptr->entry_id());
assert(op().m_is_partitionable == ret_entry_ptr->op().m_is_partitionable); assert(op().m_is_partitionable == ret_entry_ptr->op().m_is_partitionable);
assert(op().m_partition == ret_entry_ptr->op().m_partition); assert(op().m_partition == ret_entry_ptr->op().m_partition);
} }
EntryPtr<S> match() const noexcept EntryPtr<S> match() const noexcept
{ {
return m_match; return m_match;
} }
bool is_call() const noexcept bool is_call() const noexcept
{ {
return m_is_call; return m_is_call;
} }
}; };
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
std::ostream& operator<<(std::ostream& os, EntryPtr<S> entry_ptr) std::ostream& operator<<(std::ostream& os, EntryPtr<S> entry_ptr)
{ {
if (entry_ptr == nullptr) if (entry_ptr == nullptr)
return os << "entry id: none, thread id: none [nullptr]"; return os << "entry id: none, thread id: none [nullptr]";
const Entry<S>& entry = *entry_ptr; const Entry<S>& entry = *entry_ptr;
return os << return os <<
"entry id: " << entry.entry_id() << "entry id: " << entry.entry_id() <<
", thread id: " << entry.thread_id() << ", thread id: " << entry.thread_id() <<
", " << (entry.is_call() ? "call: " : "return: ") << ", " << (entry.is_call() ? "call: " : "return: ") <<
entry.op(); entry.op();
} }
#endif #endif
template<class S> template<class S>
void Stack<S>::push(EntryPtr<S> ptr, S&& s) void Stack<S>::push(EntryPtr<S> ptr, S&& s)
{ {
assert(!is_full()); assert(!is_full());
assert(ptr != nullptr); assert(ptr != nullptr);
assert(ptr->is_call()); assert(ptr->is_call());
// no overflow // no overflow
m_vector[m_top++] = std::make_pair(ptr, std::move(s)); m_vector[m_top++] = std::make_pair(ptr, std::move(s));
assert(0U != m_top); assert(0U != m_top);
} }
/// Input to linearizabilty testers /// Input to linearizabilty testers
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
class LogInfo class LogInfo
{ {
private: private:
friend class Slicer<S>; friend class Slicer<S>;
EntryPtr<S> m_log_head_ptr; EntryPtr<S> m_log_head_ptr;
std::size_t m_number_of_entries; std::size_t m_number_of_entries;
public: public:
/// \post: is_empty() /// \post: is_empty()
LogInfo() : m_log_head_ptr{ nullptr }, m_number_of_entries{ 0U } {} LogInfo() : m_log_head_ptr{ nullptr }, m_number_of_entries{ 0U } {}
/// \pre: number_of_entries is positive && even /// \pre: number_of_entries is positive && even
/// \pre: log_head_ptr is !nullptr /// \pre: log_head_ptr is !nullptr
/// \post: !is_empty() /// \post: !is_empty()
LogInfo(EntryPtr<S> log_head_ptr, std::size_t number_of_entries) LogInfo(EntryPtr<S> log_head_ptr, std::size_t number_of_entries)
: m_log_head_ptr{ log_head_ptr }, m_number_of_entries{ number_of_entries } : m_log_head_ptr{ log_head_ptr }, m_number_of_entries{ number_of_entries }
{ {
assert(log_head_ptr != nullptr); assert(log_head_ptr != nullptr);
assert(0U < m_number_of_entries); assert(0U < m_number_of_entries);
assert((m_number_of_entries & 1U) == 0U); assert((m_number_of_entries & 1U) == 0U);
} }
/// Ptr to the first entry in the log /// Ptr to the first entry in the log
EntryPtr<S> log_head_ptr() const noexcept EntryPtr<S> log_head_ptr() const noexcept
{ {
return m_log_head_ptr; return m_log_head_ptr;
} }
/// Total number of call entries plus return entries. /// Total number of call entries plus return entries.
/// Returns even number since every call is paired with a return /// Returns even number since every call is paired with a return
std::size_t number_of_entries() const noexcept std::size_t number_of_entries() const noexcept
{ {
return m_number_of_entries; return m_number_of_entries;
} }
bool is_empty() const noexcept bool is_empty() const noexcept
{ {
return m_log_head_ptr == nullptr && m_number_of_entries == 0U; return m_log_head_ptr == nullptr && m_number_of_entries == 0U;
} }
}; };
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
std::ostream& operator<<(std::ostream& os, const LogInfo<S>& log_info) std::ostream& operator<<(std::ostream& os, const LogInfo<S>& log_info)
{ {
EntryPtr<S> entry_ptr{ log_info.log_head_ptr() }; EntryPtr<S> entry_ptr{ log_info.log_head_ptr() };
os << "log info, number of entries: " << log_info.number_of_entries() << std::endl; os << "log info, number of entries: " << log_info.number_of_entries() << std::endl;
for (; entry_ptr != nullptr; entry_ptr = entry_ptr->next) for (; entry_ptr != nullptr; entry_ptr = entry_ptr->next)
os << entry_ptr << std::endl; os << entry_ptr << std::endl;
return os; return os;
} }
#endif #endif
/// Bounded history log /// Bounded history log
/// If you need thread-safety, use ConcurrentLog<S> instead. /// If you need thread-safety, use ConcurrentLog<S> instead.
/// ///
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
class Log class Log
{ {
private: private:
// fixed-size vector // fixed-size vector
typedef std::vector<Entry<S>> Entries; typedef std::vector<Entry<S>> Entries;
public: public:
typedef typename Entries::size_type Size; typedef typename Entries::size_type Size;
private: private:
// we never resize the vector and so pointers into it are stable // we never resize the vector and so pointers into it are stable
Size m_entry_id, m_index; Size m_entry_id, m_index;
Entries m_entries; Entries m_entries;
EntryPtr<S> m_last_entry_ptr; EntryPtr<S> m_last_entry_ptr;
void link(Entry<S>& entry) noexcept void link(Entry<S>& entry) noexcept
{ {
if (m_last_entry_ptr != nullptr) if (m_last_entry_ptr != nullptr)
m_last_entry_ptr->next = &entry; m_last_entry_ptr->next = &entry;
entry.prev = m_last_entry_ptr; entry.prev = m_last_entry_ptr;
} }
public: public:
Log(const Log&) = delete; Log(const Log&) = delete;
/// A history with at most capacity entries /// A history with at most capacity entries
Log(Size capacity) Log(Size capacity)
: m_entry_id{ 0U }, : m_entry_id{ 0U },
m_index{ 0U }, m_index{ 0U },
m_entries(capacity), m_entries(capacity),
m_last_entry_ptr{ nullptr } {} m_last_entry_ptr{ nullptr } {}
/// Copy entries /// Copy entries
Log(LogInfo<S> log_info) Log(LogInfo<S> log_info)
: m_entry_id{ 0U }, : m_entry_id{ 0U },
m_index{ 0U }, m_index{ 0U },
m_entries(log_info.number_of_entries()), m_entries(log_info.number_of_entries()),
m_last_entry_ptr{ nullptr } m_last_entry_ptr{ nullptr }
{ {
EntryPtr<S> entry_ptr{ log_info.log_head_ptr() }; EntryPtr<S> entry_ptr{ log_info.log_head_ptr() };
std::vector<unsigned> matches(log_info.number_of_entries() >> 1); std::vector<unsigned> matches(log_info.number_of_entries() >> 1);
while (entry_ptr != nullptr) while (entry_ptr != nullptr)
{ {
assert(m_index < m_entries.size()); assert(m_index < m_entries.size());
Entry<S>& new_entry = m_entries[m_index]; Entry<S>& new_entry = m_entries[m_index];
new_entry = *entry_ptr; new_entry = *entry_ptr;
new_entry.m_match = nullptr; new_entry.m_match = nullptr;
link(new_entry); link(new_entry);
if (new_entry.is_call()) if (new_entry.is_call())
{ {
matches[new_entry.entry_id()] = m_index; matches[new_entry.entry_id()] = m_index;
} }
else else
{ {
Entry<S>& call_entry = m_entries[matches[new_entry.entry_id()]]; Entry<S>& call_entry = m_entries[matches[new_entry.entry_id()]];
call_entry.set_match(&new_entry); call_entry.set_match(&new_entry);
} }
m_last_entry_ptr = &new_entry; m_last_entry_ptr = &new_entry;
entry_ptr = entry_ptr->next; entry_ptr = entry_ptr->next;
++m_index; ++m_index;
} }
assert(m_index == m_entries.size()); assert(m_index == m_entries.size());
assert(entry_ptr == nullptr); assert(entry_ptr == nullptr);
} }
EntryPtr<S> add_call(OpPtr<S>&& op_ptr) EntryPtr<S> add_call(OpPtr<S>&& op_ptr)
{ {
assert(m_index < m_entries.size()); assert(m_index < m_entries.size());
Entry<S>& entry = m_entries[m_index++]; Entry<S>& entry = m_entries[m_index++];
entry.set_op(std::move(op_ptr)); entry.set_op(std::move(op_ptr));
entry.set_entry_id(m_entry_id++); entry.set_entry_id(m_entry_id++);
link(entry); link(entry);
m_last_entry_ptr = &entry; m_last_entry_ptr = &entry;
return m_last_entry_ptr; return m_last_entry_ptr;
} }
/// \post: call_entry_ptr->is_call() /// \post: call_entry_ptr->is_call()
EntryPtr<S> add_ret(EntryPtr<S> call_entry_ptr, OpPtr<S>&& op_ptr) EntryPtr<S> add_ret(EntryPtr<S> call_entry_ptr, OpPtr<S>&& op_ptr)
{ {
assert(m_index < m_entries.size()); assert(m_index < m_entries.size());
Entry<S>& entry = m_entries[m_index++]; Entry<S>& entry = m_entries[m_index++];
entry.set_op(std::move(op_ptr)); entry.set_op(std::move(op_ptr));
link(entry); link(entry);
m_last_entry_ptr = &entry; m_last_entry_ptr = &entry;
call_entry_ptr->set_match(m_last_entry_ptr); call_entry_ptr->set_match(m_last_entry_ptr);
assert(call_entry_ptr->is_call()); assert(call_entry_ptr->is_call());
assert(m_entry_id <= m_index); assert(m_entry_id <= m_index);
return m_last_entry_ptr; return m_last_entry_ptr;
} }
EntryPtr<S> log_head_ptr() EntryPtr<S> log_head_ptr()
{ {
return &m_entries.front(); return &m_entries.front();
} }
/// Total number of call entries plus return entries. /// Total number of call entries plus return entries.
/// Returns even number since every call is paired with a return /// Returns even number since every call is paired with a return
std::size_t number_of_entries() const noexcept std::size_t number_of_entries() const noexcept
{ {
return m_index; return m_index;
} }
LogInfo<S> info() LogInfo<S> info()
{ {
return{ log_head_ptr(), number_of_entries() }; return{ log_head_ptr(), number_of_entries() };
} }
}; };
/// Output of linearizability tester /// Output of linearizability tester
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
class Result class Result
{ {
private: private:
friend class LinearizabilityTester<S, Option::NEVER_CACHE>; friend class LinearizabilityTester<S, Option::NEVER_CACHE>;
friend class LinearizabilityTester<S, Option::LRU_CACHE>; friend class LinearizabilityTester<S, Option::LRU_CACHE>;
friend class LinearizabilityTester<S, Option::ALWAYS_CACHE>; friend class LinearizabilityTester<S, Option::ALWAYS_CACHE>;
typedef std::vector<EntryPtr<S>> EntryPtrs; typedef std::vector<EntryPtr<S>> EntryPtrs;
bool m_is_linearizable; bool m_is_linearizable;
EntryPtrs m_entry_ptrs; EntryPtrs m_entry_ptrs;
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
unsigned m_cutoff_entry_id; unsigned m_cutoff_entry_id;
EntryPtr<S> m_log_head_ptr; EntryPtr<S> m_log_head_ptr;
#endif #endif
bool m_is_timeout; bool m_is_timeout;
double m_virtual_memory_usage; double m_virtual_memory_usage;
double m_resident_set_size; double m_resident_set_size;
void reset() void reset()
{ {
m_is_linearizable = true; m_is_linearizable = true;
m_entry_ptrs.clear(); m_entry_ptrs.clear();
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
m_cutoff_entry_id = 0U; m_cutoff_entry_id = 0U;
m_log_head_ptr = nullptr; m_log_head_ptr = nullptr;
#endif #endif
m_is_timeout = false; m_is_timeout = false;
m_virtual_memory_usage = 0.0; m_virtual_memory_usage = 0.0;
m_resident_set_size = 0.0; m_resident_set_size = 0.0;
} }
public: public:
/// Initially linearizable /// Initially linearizable
Result() Result()
: m_is_linearizable{ true }, : m_is_linearizable{ true },
m_entry_ptrs{}, m_entry_ptrs{},
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
m_cutoff_entry_id{ 0U }, m_cutoff_entry_id{ 0U },
m_log_head_ptr{ nullptr }, m_log_head_ptr{ nullptr },
#endif #endif
m_is_timeout{ false }, m_is_timeout{ false },
m_virtual_memory_usage{ 0.0 }, m_virtual_memory_usage{ 0.0 },
m_resident_set_size{ 0.0 } {} m_resident_set_size{ 0.0 } {}
/// \pre: !is_timeout() /// \pre: !is_timeout()
bool is_linearizable() const noexcept bool is_linearizable() const noexcept
{ {
assert(!is_timeout()); assert(!is_timeout());
return m_is_linearizable; return m_is_linearizable;
} }
bool is_timeout() const noexcept bool is_timeout() const noexcept
{ {
return m_is_timeout; return m_is_timeout;
} }
/// Zero if unknown, unit: MiB /// Zero if unknown, unit: MiB
double virtual_memory_usage() const noexcept double virtual_memory_usage() const noexcept
{ {
return m_virtual_memory_usage; return m_virtual_memory_usage;
} }
/// Zero if unknown, unit: MiB /// Zero if unknown, unit: MiB
double resident_set_size() const noexcept double resident_set_size() const noexcept
{ {
return m_resident_set_size; return m_resident_set_size;
} }
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
void debug(std::ostream& os, bool verbose = false) void debug(std::ostream& os, bool verbose = false)
{ {
os << "Linearizable: "; os << "Linearizable: ";
if (m_is_linearizable) if (m_is_linearizable)
{ {
os << "Yes" << std::endl; os << "Yes" << std::endl;
for (EntryPtr<S> entry_ptr : m_entry_ptrs) for (EntryPtr<S> entry_ptr : m_entry_ptrs)
os << entry_ptr << " : " << entry_ptr->match() << std::endl; os << entry_ptr << " : " << entry_ptr->match() << std::endl;
return; return;
} }
os << "No" << std::endl; os << "No" << std::endl;
EntryPtr<S> entry_ptr{ m_log_head_ptr }; EntryPtr<S> entry_ptr{ m_log_head_ptr };
for (; entry_ptr != nullptr; entry_ptr = entry_ptr->next) for (; entry_ptr != nullptr; entry_ptr = entry_ptr->next)
{ {
os << entry_ptr << std::endl; os << entry_ptr << std::endl;
if (entry_ptr->entry_id() == m_cutoff_entry_id) if (entry_ptr->entry_id() == m_cutoff_entry_id)
{ {
os << "^ previous entries cannot be linearized" << std::endl; os << "^ previous entries cannot be linearized" << std::endl;
if (!(verbose || entry_ptr->is_call())) if (!(verbose || entry_ptr->is_call()))
return; return;
} }
} }
} }
#endif #endif
}; };
#ifdef _LT_TIMEOUT_ #ifdef _LT_TIMEOUT_
template <typename Clock = std::chrono::steady_clock> template <typename Clock = std::chrono::steady_clock>
struct Timeout struct Timeout
{ {
const typename Clock::time_point start_time; const typename Clock::time_point start_time;
const typename Clock::duration max_duration; const typename Clock::duration max_duration;
Timeout() Timeout()
: start_time{ Clock::now() }, : start_time{ Clock::now() },
max_duration{ Clock::duration::max() } {} max_duration{ Clock::duration::max() } {}
Timeout(typename Clock::duration duration) Timeout(typename Clock::duration duration)
: start_time{ Clock::now() }, : start_time{ Clock::now() },
max_duration{ duration } {} max_duration{ duration } {}
bool is_expired() const bool is_expired() const
{ {
return max_duration < (Clock::now() - start_time); return max_duration < (Clock::now() - start_time);
} }
}; };
#endif #endif
/// Least-recently used cache eviction /// Least-recently used cache eviction
template<class Key, class Hash = std::hash<Key>> template<class Key, class Hash = std::hash<Key>>
class LruCache class LruCache
{ {
private: private:
typedef std::list<Key> List; typedef std::list<Key> List;
typedef typename List::iterator ListIterator; typedef typename List::iterator ListIterator;
typedef std::unordered_map<Key, ListIterator, Hash> UnorderedMap; typedef std::unordered_map<Key, ListIterator, Hash> UnorderedMap;
typedef typename UnorderedMap::size_type Capacity; typedef typename UnorderedMap::size_type Capacity;
const Capacity m_capacity; const Capacity m_capacity;
UnorderedMap m_unordered_map; UnorderedMap m_unordered_map;
List m_list; List m_list;
public: public:
typedef typename UnorderedMap::iterator Iterator; typedef typename UnorderedMap::iterator Iterator;
LruCache() LruCache()
: m_capacity{ 4096 }, : m_capacity{ 4096 },
m_unordered_map{ m_capacity } {} m_unordered_map{ m_capacity } {}
LruCache(Capacity capacity) LruCache(Capacity capacity)
: m_capacity{ capacity }, : m_capacity{ capacity },
m_unordered_map{ m_capacity } {} m_unordered_map{ m_capacity } {}
bool insert(Key&& key) bool insert(Key&& key)
{ {
std::pair<Iterator, bool> pair{ m_unordered_map.insert( std::pair<Iterator, bool> pair{ m_unordered_map.insert(
std::make_pair(std::move(key), m_list.end())) }; std::make_pair(std::move(key), m_list.end())) };
if (pair.second) if (pair.second)
pair.first->second = m_list.insert(m_list.end(), pair.first->first); pair.first->second = m_list.insert(m_list.end(), pair.first->first);
else else
m_list.splice(m_list.end(), m_list, pair.first->second); m_list.splice(m_list.end(), m_list, pair.first->second);
if (m_unordered_map.size() == m_capacity) if (m_unordered_map.size() == m_capacity)
{ {
auto iter = m_unordered_map.find(m_list.front()); auto iter = m_unordered_map.find(m_list.front());
assert(iter != m_unordered_map.cend()); assert(iter != m_unordered_map.cend());
m_unordered_map.erase(iter); m_unordered_map.erase(iter);
m_list.pop_front(); m_list.pop_front();
} }
return pair.second; return pair.second;
} }
}; };
namespace cache namespace cache
{ {
// regardless of caching, we need to keep track of the current state of type S // regardless of caching, we need to keep track of the current state of type S
template<class S> template<class S>
using State = std::pair<Bitset, S>; using State = std::pair<Bitset, S>;
// if caching is enabled, we use these hash functions // if caching is enabled, we use these hash functions
template<class S> template<class S>
class StateHash class StateHash
{ {
private: private:
// courtesy of Daniel Kroening, see CPROVER source file util/irep.cpp // courtesy of Daniel Kroening, see CPROVER source file util/irep.cpp
static inline size_t hash_rotl(std::size_t value, unsigned shift) static inline size_t hash_rotl(std::size_t value, unsigned shift)
{ {
return (value << shift) | (value >> ((sizeof(value) * 8U) - shift)); return (value << shift) | (value >> ((sizeof(value) * 8U) - shift));
} }
// courtesy of Daniel Kroening, see CPROVER source file util/irep.cpp // courtesy of Daniel Kroening, see CPROVER source file util/irep.cpp
static inline size_t hash_combine(std::size_t h1, std::size_t h2) static inline size_t hash_combine(std::size_t h1, std::size_t h2)
{ {
return hash_rotl(h1, 7U) ^ h2; return hash_rotl(h1, 7U) ^ h2;
} }
const BitsetHash m_bitset_hash; const BitsetHash m_bitset_hash;
const state::Hash<S> m_s_hash; const state::Hash<S> m_s_hash;
public: public:
StateHash() : m_bitset_hash{}, m_s_hash{} {} StateHash() : m_bitset_hash{}, m_s_hash{} {}
std::size_t operator()(const State<S>& state) const noexcept std::size_t operator()(const State<S>& state) const noexcept
{ {
return hash_combine(m_bitset_hash(state.first), m_s_hash(state.second)); return hash_combine(m_bitset_hash(state.first), m_s_hash(state.second));
} }
}; };
template<class S, Option option = Option::NEVER_CACHE> template<class S, Option option = Option::NEVER_CACHE>
struct Switch struct Switch
{ {
typedef std::nullptr_t Type; typedef std::nullptr_t Type;
static bool try_insert(const S& s, const EntryPtr<S> entry_ptr, static bool try_insert(const S& s, const EntryPtr<S> entry_ptr,
Type& cache, Bitset& bitset) Type& cache, Bitset& bitset)
{ {
return true; return true;
} }
}; };
template<class S> template<class S>
struct Switch<S, Option::LRU_CACHE> struct Switch<S, Option::LRU_CACHE>
{ {
typedef LruCache<State<S>, StateHash<S>> Type; typedef LruCache<State<S>, StateHash<S>> Type;
static bool try_insert(const S& s, const EntryPtr<S> entry_ptr, static bool try_insert(const S& s, const EntryPtr<S> entry_ptr,
Type& cache, Bitset& bitset) Type& cache, Bitset& bitset)
{ {
return cache.insert(std::make_pair(bitset.immutable_set(entry_ptr->entry_id()), s)); return cache.insert(std::make_pair(bitset.immutable_set(entry_ptr->entry_id()), s));
} }
}; };
template<class S> template<class S>
struct Switch<S, Option::ALWAYS_CACHE> struct Switch<S, Option::ALWAYS_CACHE>
{ {
typedef std::unordered_set<State<S>, StateHash<S>> Type; typedef std::unordered_set<State<S>, StateHash<S>> Type;
static bool try_insert(const S& s, const EntryPtr<S> entry_ptr, static bool try_insert(const S& s, const EntryPtr<S> entry_ptr,
Type& cache, Bitset& bitset) Type& cache, Bitset& bitset)
{ {
unsigned int pos = entry_ptr->entry_id(); unsigned int pos = entry_ptr->entry_id();
lt::Bitset bs = bitset.immutable_set(pos); lt::Bitset bs = bitset.immutable_set(pos);
return std::get<1>(cache.emplace(bs, s)); return std::get<1>(cache.emplace(bs, s));
} }
}; };
} }
/// S - sequential data type /// S - sequential data type
template<class S, Option option = Option::ALWAYS_CACHE> template<class S, Option option = Option::ALWAYS_CACHE>
class LinearizabilityTester class LinearizabilityTester
{ {
private: private:
typedef cache::Switch<S, option> Cache; typedef cache::Switch<S, option> Cache;
// Maximum number of call/ret entries, i.e. half of the // Maximum number of call/ret entries, i.e. half of the
// total number of entry pointers reachable in m_log_head // total number of entry pointers reachable in m_log_head
const std::size_t m_log_size; const std::size_t m_log_size;
// History to linearize, every call is matched by a return // History to linearize, every call is matched by a return
const Entry<S> m_log_head; const Entry<S> m_log_head;
// Invariants: // Invariants:
// //
// * for every EntryPtr<S> `e` in `m_calls`, `e->is_call()` holds // * for every EntryPtr<S> `e` in `m_calls`, `e->is_call()` holds
// * for every EntryPtr<S> `e`, if `e` in `m_calls`, then `e` is not reachable // * for every EntryPtr<S> `e`, if `e` in `m_calls`, then `e` is not reachable
// from `m_log_head` by following the next pointers. // from `m_log_head` by following the next pointers.
Stack<S> m_calls; Stack<S> m_calls;
#ifdef _LT_TIMEOUT_ #ifdef _LT_TIMEOUT_
Timeout<std::chrono::steady_clock> m_timeout; Timeout<std::chrono::steady_clock> m_timeout;
#endif #endif
// An approximation of the workload // An approximation of the workload
unsigned long m_number_of_iterations; unsigned long m_number_of_iterations;
// see http://stackoverflow.com/questions/669438/how-to-get-memory-usage-at-run-time-in-c // see http://stackoverflow.com/questions/669438/how-to-get-memory-usage-at-run-time-in-c
// //
// process_mem_usage(double &, double &) - takes two doubles by reference, // process_mem_usage(double &, double &) - takes two doubles by reference,
// attempts to read the system-dependent data for a process' virtual memory // attempts to read the system-dependent data for a process' virtual memory
// size && resident set size, && return the results in MiB. // size && resident set size, && return the results in MiB.
// //
// On failure, returns 0.0, 0.0 // On failure, returns 0.0, 0.0
static void process_mem_usage(double& vm_usage, double& resident_set) static void process_mem_usage(double& vm_usage, double& resident_set)
{ {
vm_usage = resident_set = 0.0; vm_usage = resident_set = 0.0;
} }
// Temporarily remove call_entry_ptr and call_entry_ptr->match() from the log // Temporarily remove call_entry_ptr and call_entry_ptr->match() from the log
// \pre: call_entry_ptr->is_call() // \pre: call_entry_ptr->is_call()
static void lift(const EntryPtr<S> call_entry_ptr) static void lift(const EntryPtr<S> call_entry_ptr)
{ {
const Entry<S>& call = *call_entry_ptr; const Entry<S>& call = *call_entry_ptr;
assert(call.is_call()); assert(call.is_call());
Entry<S>& match = *call.match(); Entry<S>& match = *call.match();
call.prev->next = call.next; call.prev->next = call.next;
call.next->prev = call.prev; call.next->prev = call.prev;
match.prev->next = match.next; match.prev->next = match.next;
if (match.next != nullptr) if (match.next != nullptr)
match.next->prev = match.prev; match.next->prev = match.prev;
} }
// Reinsert call_entry_ptr && call_entry_ptr->match() into the log // Reinsert call_entry_ptr && call_entry_ptr->match() into the log
// \pre: call_entry_ptr->is_call() // \pre: call_entry_ptr->is_call()
static void unlift(const EntryPtr<S> call_entry_ptr) static void unlift(const EntryPtr<S> call_entry_ptr)
{ {
const Entry<S>& call = *call_entry_ptr; const Entry<S>& call = *call_entry_ptr;
assert(call.is_call()); assert(call.is_call());
Entry<S>& match = *call.match(); Entry<S>& match = *call.match();
assert(match.prev->next == match.next); assert(match.prev->next == match.next);
match.prev->next = &match; match.prev->next = &match;
if (match.next != nullptr) if (match.next != nullptr)
match.next->prev = &match; match.next->prev = &match;
assert(call.prev->next == call.next); assert(call.prev->next == call.next);
call.prev->next = call_entry_ptr; call.prev->next = call_entry_ptr;
call.next->prev = call_entry_ptr; call.next->prev = call_entry_ptr;
} }
void internal_check(Result<S>& result, unsigned& global_linearized_entry_id) void internal_check(Result<S>& result, unsigned& global_linearized_entry_id)
{ {
S s, new_s; S s, new_s;
bool is_entry_linearizable; bool is_entry_linearizable;
typename Cache::Type cache; typename Cache::Type cache;
EntryPtr<S> pop_entry_ptr, entry_ptr{ m_log_head.next }; EntryPtr<S> pop_entry_ptr, entry_ptr{ m_log_head.next };
double virtual_memory_usage; double virtual_memory_usage;
double resident_set_size; double resident_set_size;
// fixing the size is !merely an optimization but // fixing the size is !merely an optimization but
// necessary for checking the equality of bitsets // necessary for checking the equality of bitsets
Bitset linearized_entries(m_log_size); Bitset linearized_entries(m_log_size);
while (m_log_head.next != nullptr) while (m_log_head.next != nullptr)
{ {
process_mem_usage(virtual_memory_usage, resident_set_size); process_mem_usage(virtual_memory_usage, resident_set_size);
result.m_virtual_memory_usage = std::max(result.m_virtual_memory_usage, virtual_memory_usage); result.m_virtual_memory_usage = std::max(result.m_virtual_memory_usage, virtual_memory_usage);
result.m_resident_set_size = std::max(result.m_resident_set_size, resident_set_size); result.m_resident_set_size = std::max(result.m_resident_set_size, resident_set_size);
#ifdef _LT_TIMEOUT_ #ifdef _LT_TIMEOUT_
if (m_timeout.is_expired()) if (m_timeout.is_expired())
{ {
result.m_is_timeout = true; result.m_is_timeout = true;
break; break;
} }
#endif #endif
++m_number_of_iterations; ++m_number_of_iterations;
assert(entry_ptr != nullptr); assert(entry_ptr != nullptr);
if (entry_ptr->is_call()) if (entry_ptr->is_call())
{ {
assert(!m_calls.is_full()); assert(!m_calls.is_full());
assert(entry_ptr->match() != nullptr); assert(entry_ptr->match() != nullptr);
assert(!linearized_entries.is_set(entry_ptr->entry_id())); assert(!linearized_entries.is_set(entry_ptr->entry_id()));
std::tie(is_entry_linearizable, new_s) = std::tie(is_entry_linearizable, new_s) =
entry_ptr->op().apply(s, entry_ptr->match()->op()); entry_ptr->op().apply(s, entry_ptr->match()->op());
if (is_entry_linearizable && Cache::try_insert(new_s, entry_ptr, cache, linearized_entries)) if (is_entry_linearizable && Cache::try_insert(new_s, entry_ptr, cache, linearized_entries))
{ {
// call entry is always matched up with a return entry // call entry is always matched up with a return entry
assert(entry_ptr->next != nullptr); assert(entry_ptr->next != nullptr);
// provisionally linearize the call entry together with // provisionally linearize the call entry together with
// the associated state produced by the new linearization // the associated state produced by the new linearization
m_calls.push(entry_ptr, std::move(s)); m_calls.push(entry_ptr, std::move(s));
s = std::move(new_s); s = std::move(new_s);
linearized_entries.set(entry_ptr->entry_id()); linearized_entries.set(entry_ptr->entry_id());
// provisionally remove the call && return entry from the history // provisionally remove the call && return entry from the history
lift(entry_ptr); lift(entry_ptr);
// restart from the beginning of the shortened history // restart from the beginning of the shortened history
entry_ptr = m_log_head.next; entry_ptr = m_log_head.next;
} }
else // cannot linearize call entry else // cannot linearize call entry
{ {
// get the next entry in the unmodified history // get the next entry in the unmodified history
entry_ptr = entry_ptr->next; entry_ptr = entry_ptr->next;
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
global_linearized_entry_id = std::max(global_linearized_entry_id, entry_ptr->entry_id()); global_linearized_entry_id = std::max(global_linearized_entry_id, entry_ptr->entry_id());
#endif #endif
} }
} }
else // handle "return" entry else // handle "return" entry
{ {
if (m_calls.is_empty()) if (m_calls.is_empty())
break; break;
assert(!m_calls.is_empty()); assert(!m_calls.is_empty());
// revert state change // revert state change
std::tie(pop_entry_ptr, s) = m_calls.top(); std::tie(pop_entry_ptr, s) = m_calls.top();
assert(pop_entry_ptr != nullptr); assert(pop_entry_ptr != nullptr);
linearized_entries.reset(pop_entry_ptr->entry_id()); linearized_entries.reset(pop_entry_ptr->entry_id());
m_calls.pop(); m_calls.pop();
// undo the provisional linearization // undo the provisional linearization
unlift(pop_entry_ptr); unlift(pop_entry_ptr);
// continue after the entry to which we have just backtracked // continue after the entry to which we have just backtracked
entry_ptr = pop_entry_ptr->next; entry_ptr = pop_entry_ptr->next;
} }
} }
// all call entries linearized? // all call entries linearized?
result.m_is_linearizable = m_calls.is_full(); result.m_is_linearizable = m_calls.is_full();
assert(result.m_is_linearizable == (m_log_head.next == nullptr)); assert(result.m_is_linearizable == (m_log_head.next == nullptr));
// witness linearization // witness linearization
std::size_t pos{ 0 }; std::size_t pos{ 0 };
result.m_entry_ptrs.resize(m_calls.size()); result.m_entry_ptrs.resize(m_calls.size());
for (EntryPtr<S>& entry_ptr : result.m_entry_ptrs) for (EntryPtr<S>& entry_ptr : result.m_entry_ptrs)
entry_ptr = m_calls.entry_ptr(pos++); entry_ptr = m_calls.entry_ptr(pos++);
} }
public: public:
LinearizabilityTester(LogInfo<S> log_info) LinearizabilityTester(LogInfo<S> log_info)
: m_log_size{ log_info.number_of_entries() >> 1 }, : m_log_size{ log_info.number_of_entries() >> 1 },
m_log_head{ log_info.log_head_ptr() }, m_log_head{ log_info.log_head_ptr() },
m_calls{ m_log_size }, m_calls{ m_log_size },
#ifdef _LT_TIMEOUT_ #ifdef _LT_TIMEOUT_
m_timeout{}, m_timeout{},
#endif #endif
m_number_of_iterations{} {} m_number_of_iterations{} {}
#ifdef _LT_TIMEOUT_ #ifdef _LT_TIMEOUT_
LinearizabilityTester(LogInfo<S> log_info, LinearizabilityTester(LogInfo<S> log_info,
std::chrono::steady_clock::duration max_duration) std::chrono::steady_clock::duration max_duration)
: m_log_size{ log_info.number_of_entries() >> 1 }, : m_log_size{ log_info.number_of_entries() >> 1 },
m_log_head{ log_info.log_head_ptr() }, m_log_head{ log_info.log_head_ptr() },
m_calls{ m_log_size }, m_calls{ m_log_size },
m_timeout{ max_duration }, m_timeout{ max_duration },
m_number_of_iterations{} {} m_number_of_iterations{} {}
#endif #endif
/// A rough approximation of the workload /// A rough approximation of the workload
unsigned long number_of_iterations() const noexcept unsigned long number_of_iterations() const noexcept
{ {
return m_number_of_iterations; return m_number_of_iterations;
} }
/// Is history linearizable? /// Is history linearizable?
/// Throws an exception on timeout /// Throws an exception on timeout
bool check() bool check()
{ {
Result<S> result; Result<S> result;
unsigned disregard_cutoff_entry_id; unsigned disregard_cutoff_entry_id;
internal_check(result, disregard_cutoff_entry_id); internal_check(result, disregard_cutoff_entry_id);
if (result.is_timeout()) if (result.is_timeout())
throw std::runtime_error("Timeout!"); throw std::runtime_error("Timeout!");
return result.is_linearizable(); return result.is_linearizable();
} }
void check(Result<S>& result) void check(Result<S>& result)
{ {
result.reset(); result.reset();
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
internal_check(result, result.m_cutoff_entry_id); internal_check(result, result.m_cutoff_entry_id);
result.m_log_head_ptr = m_log_head.next; result.m_log_head_ptr = m_log_head.next;
#else #else
unsigned disregard_cutoff_entry_id; unsigned disregard_cutoff_entry_id;
internal_check(result, disregard_cutoff_entry_id); internal_check(result, disregard_cutoff_entry_id);
#endif #endif
} }
}; };
template<class S, class Duration> template<class S, class Duration>
void compositional_check(Log<S>& log, Result<S> &result, void compositional_check(Log<S>& log, Result<S> &result,
unsigned number_of_partitions, Duration max_duration) unsigned number_of_partitions, Duration max_duration)
{ {
Slicer<S> slicer{ log.info(), number_of_partitions }; Slicer<S> slicer{ log.info(), number_of_partitions };
for (unsigned partition = 0; partition < slicer.number_of_partitions; ++partition) for (unsigned partition = 0; partition < slicer.number_of_partitions; ++partition)
{ {
LinearizabilityTester<S> tester{ slicer.sublog_info(partition), max_duration }; LinearizabilityTester<S> tester{ slicer.sublog_info(partition), max_duration };
tester.check(result); tester.check(result);
if (!(result.is_timeout() || result.is_linearizable())) if (!(result.is_timeout() || result.is_linearizable()))
break; break;
} }
} }
/// RAII class to ensure a thread becomes unjoinable on all paths /// RAII class to ensure a thread becomes unjoinable on all paths
class Thread class Thread
{ {
private: private:
std::thread m_thread; std::thread m_thread;
public: public:
Thread() Thread()
: m_thread{} {} : m_thread{} {}
Thread(std::thread&& thread) Thread(std::thread&& thread)
: m_thread(std::move(thread)) {} : m_thread(std::move(thread)) {}
template<typename F, typename... Args> template<typename F, typename... Args>
Thread(F&& f, Args&&... args) Thread(F&& f, Args&&... args)
: m_thread(std::forward<F>(f), std::forward<Args>(args)...) {} : m_thread(std::forward<F>(f), std::forward<Args>(args)...) {}
~Thread() ~Thread()
{ {
if (m_thread.joinable()) if (m_thread.joinable())
m_thread.join(); m_thread.join();
} }
/// \pre: joinable() /// \pre: joinable()
/// \post: not joinable() /// \post: not joinable()
/// Throws std::system_error if an error occurs. /// Throws std::system_error if an error occurs.
void join() void join()
{ {
m_thread.join(); m_thread.join();
} }
bool joinable() const noexcept bool joinable() const noexcept
{ {
return m_thread.joinable(); return m_thread.joinable();
} }
Thread& operator=(Thread&& thread) Thread& operator=(Thread&& thread)
{ {
m_thread = std::move(thread.m_thread); m_thread = std::move(thread.m_thread);
return *this; return *this;
} }
}; };
/// Partition history into sub-histories /// Partition history into sub-histories
/// A slicer partitions the history into independent sub-histories. /// A slicer partitions the history into independent sub-histories.
/// Our partitioning scheme hinges on Theorem 3.6.1 in "The Art of /// Our partitioning scheme hinges on Theorem 3.6.1 in "The Art of
/// Multiprocessor Programming" (Revised Ed.) by Herlihy && Shavit. /// Multiprocessor Programming" (Revised Ed.) by Herlihy && Shavit.
/// ///
/// Typically only associative concurrent abstract data types (ADTs) /// Typically only associative concurrent abstract data types (ADTs)
/// such as sets && hash tables are suitable for this partitioning /// such as sets && hash tables are suitable for this partitioning
/// scheme. && !all operations on such ADTs are always supported. /// scheme. && !all operations on such ADTs are always supported.
/// For example, the partitioning scheme is incompatible with 0-arg /// For example, the partitioning scheme is incompatible with 0-arg
/// operations such as "empty?" on sets. But it is very effective if /// operations such as "empty?" on sets. But it is very effective if
/// we want to only check linearizability of say "insert", "remove" /// we want to only check linearizability of say "insert", "remove"
/// && "contains". /// && "contains".
/// ///
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
class Slicer class Slicer
{ {
private: private:
typedef std::vector<LogInfo<S>> Sublogs; typedef std::vector<LogInfo<S>> Sublogs;
static void slice(const Entry<S>& log_head, Sublogs& sublogs) static void slice(const Entry<S>& log_head, Sublogs& sublogs)
{ {
const typename Sublogs::size_type n = sublogs.size(); const typename Sublogs::size_type n = sublogs.size();
EntryPtr<S> entry_ptr{ log_head.next }, next_entry_ptr; EntryPtr<S> entry_ptr{ log_head.next }, next_entry_ptr;
std::vector<EntryPtr<S>> last_entry_ptrs(sublogs.size()); std::vector<EntryPtr<S>> last_entry_ptrs(sublogs.size());
std::vector<unsigned> entry_ids(sublogs.size()); std::vector<unsigned> entry_ids(sublogs.size());
typename Sublogs::size_type i; typename Sublogs::size_type i;
unsigned new_entry_id; unsigned new_entry_id;
while (entry_ptr != nullptr) while (entry_ptr != nullptr)
{ {
i = entry_ptr->op().partition() % n; i = entry_ptr->op().partition() % n;
LogInfo<S>& log_info = sublogs[i]; LogInfo<S>& log_info = sublogs[i];
EntryPtr<S>& last_entry_ptr = last_entry_ptrs[i]; EntryPtr<S>& last_entry_ptr = last_entry_ptrs[i];
if (log_info.log_head_ptr() == nullptr) if (log_info.log_head_ptr() == nullptr)
{ {
// initialize sub-log // initialize sub-log
assert(entry_ptr->is_call()); assert(entry_ptr->is_call());
assert(last_entry_ptr == nullptr); assert(last_entry_ptr == nullptr);
log_info.m_log_head_ptr = entry_ptr; log_info.m_log_head_ptr = entry_ptr;
log_info.m_number_of_entries = 1U; log_info.m_number_of_entries = 1U;
} }
else else
{ {
// size of the sub-log increases // size of the sub-log increases
++log_info.m_number_of_entries; ++log_info.m_number_of_entries;
assert(last_entry_ptr != nullptr); assert(last_entry_ptr != nullptr);
last_entry_ptr->next = entry_ptr; last_entry_ptr->next = entry_ptr;
} }
if (entry_ptr->is_call()) if (entry_ptr->is_call())
{ {
new_entry_id = entry_ids[i]++; new_entry_id = entry_ids[i]++;
entry_ptr->set_entry_id(new_entry_id); entry_ptr->set_entry_id(new_entry_id);
entry_ptr->match()->set_entry_id(new_entry_id); entry_ptr->match()->set_entry_id(new_entry_id);
} }
next_entry_ptr = entry_ptr->next; next_entry_ptr = entry_ptr->next;
entry_ptr->prev = last_entry_ptr; entry_ptr->prev = last_entry_ptr;
entry_ptr->next = nullptr; entry_ptr->next = nullptr;
last_entry_ptr = entry_ptr; last_entry_ptr = entry_ptr;
entry_ptr = next_entry_ptr; entry_ptr = next_entry_ptr;
} }
} }
Sublogs m_sublogs; Sublogs m_sublogs;
unsigned m_current_partition; unsigned m_current_partition;
public: public:
const Entry<S> log_head; const Entry<S> log_head;
const unsigned number_of_partitions; const unsigned number_of_partitions;
Slicer(LogInfo<S> log_info, unsigned _number_of_partitions) Slicer(LogInfo<S> log_info, unsigned _number_of_partitions)
: m_sublogs(_number_of_partitions), : m_sublogs(_number_of_partitions),
m_current_partition{ 0U }, m_current_partition{ 0U },
log_head{ log_info.log_head_ptr() }, log_head{ log_info.log_head_ptr() },
number_of_partitions{ _number_of_partitions } number_of_partitions{ _number_of_partitions }
{ {
slice(log_head, m_sublogs); slice(log_head, m_sublogs);
} }
const LogInfo<S>& sublog_info(unsigned partition) const const LogInfo<S>& sublog_info(unsigned partition) const
{ {
return m_sublogs[partition]; return m_sublogs[partition];
} }
const LogInfo<S>& next_sublog_info() const LogInfo<S>& next_sublog_info()
{ {
static LogInfo<S> s_empty_log; static LogInfo<S> s_empty_log;
unsigned partition = m_current_partition; unsigned partition = m_current_partition;
++m_current_partition; ++m_current_partition;
if (partition < number_of_partitions) if (partition < number_of_partitions)
return m_sublogs[partition]; return m_sublogs[partition];
return s_empty_log; return s_empty_log;
} }
}; };
/// S - sequential data type /// S - sequential data type
template<class S> template<class S>
class ConcurrentLog class ConcurrentLog
{ {
private: private:
typedef std::vector<Entry<S>> Entries; typedef std::vector<Entry<S>> Entries;
typedef typename Entries::size_type Size; typedef typename Entries::size_type Size;
Entries m_entries; Entries m_entries;
std::atomic<Size> m_index; std::atomic<Size> m_index;
static void link(EntryPtr<S> last_entry_ptr, Entry<S>& entry) static void link(EntryPtr<S> last_entry_ptr, Entry<S>& entry)
{ {
if (last_entry_ptr != nullptr) if (last_entry_ptr != nullptr)
last_entry_ptr->next = &entry; last_entry_ptr->next = &entry;
entry.prev = last_entry_ptr; entry.prev = last_entry_ptr;
} }
public: public:
ConcurrentLog(Size capacity) ConcurrentLog(Size capacity)
: m_entries(capacity), : m_entries(capacity),
m_index{ 0U } {} m_index{ 0U } {}
/// \remark thread-safe /// \remark thread-safe
/// \pre: enough capacity /// \pre: enough capacity
EntryPtr<S> push_back(OpPtr<S>&& op_ptr) EntryPtr<S> push_back(OpPtr<S>&& op_ptr)
{ {
// we use the relaxed memory || der tag because we // we use the relaxed memory || der tag because we
// do !need to read any other memory locations // do !need to read any other memory locations
Size index = m_index.fetch_add(1U, std::memory_order_relaxed); Size index = m_index.fetch_add(1U, std::memory_order_relaxed);
assert(index < m_entries.size()); assert(index < m_entries.size());
// There is no data race, see [container.requirements.dataraces] // There is no data race, see [container.requirements.dataraces]
// in Section 23.2.2, paragraph 2, p. 734 in the C++11 language // in Section 23.2.2, paragraph 2, p. 734 in the C++11 language
// specification. Since the index was incremented atomically, // specification. Since the index was incremented atomically,
// each thread accesses a different element in the vector. // each thread accesses a different element in the vector.
Entry<S>& entry = m_entries[index]; Entry<S>& entry = m_entries[index];
entry.set_op(std::move(op_ptr)); entry.set_op(std::move(op_ptr));
entry.set_thread_id(std::this_thread::get_id()); entry.set_thread_id(std::this_thread::get_id());
return &entry; return &entry;
} }
/// \remark thread-safe /// \remark thread-safe
/// \pre: enough capacity /// \pre: enough capacity
/// \post: call_entry_ptr->is_call() /// \post: call_entry_ptr->is_call()
EntryPtr<S> push_back(EntryPtr<S> call_entry_ptr, OpPtr<S>&& op_ptr) EntryPtr<S> push_back(EntryPtr<S> call_entry_ptr, OpPtr<S>&& op_ptr)
{ {
EntryPtr<S> entry_ptr = push_back(std::move(op_ptr)); EntryPtr<S> entry_ptr = push_back(std::move(op_ptr));
call_entry_ptr->set_match(entry_ptr); call_entry_ptr->set_match(entry_ptr);
assert(call_entry_ptr->is_call()); assert(call_entry_ptr->is_call());
return entry_ptr; return entry_ptr;
} }
/// \warning !thread-safe /// \warning !thread-safe
EntryPtr<S> log_head_ptr() EntryPtr<S> log_head_ptr()
{ {
if (m_entries.front().next == nullptr) if (m_entries.front().next == nullptr)
{ {
unsigned entry_id{ 0U }; unsigned entry_id{ 0U };
Size index{ 0U }; Size index{ 0U };
EntryPtr<S> last_entry_ptr{ nullptr }; EntryPtr<S> last_entry_ptr{ nullptr };
for (Entry<S>& entry : m_entries) for (Entry<S>& entry : m_entries)
{ {
if (index == m_index) if (index == m_index)
break; break;
++index; ++index;
if (entry.is_call()) if (entry.is_call())
{ {
entry.set_entry_id(entry_id); entry.set_entry_id(entry_id);
entry.match()->set_entry_id(entry_id); entry.match()->set_entry_id(entry_id);
++entry_id; ++entry_id;
} }
link(last_entry_ptr, entry); link(last_entry_ptr, entry);
last_entry_ptr = &entry; last_entry_ptr = &entry;
} }
} }
return &m_entries.front(); return &m_entries.front();
} }
/// Total number of call entries plus return entries. /// Total number of call entries plus return entries.
/// Returns even number since every call is paired with a return /// Returns even number since every call is paired with a return
/// ///
/// \warning !thread-safe /// \warning !thread-safe
std::size_t number_of_entries() const noexcept std::size_t number_of_entries() const noexcept
{ {
return m_index.load(); return m_index.load();
} }
/// \warning !thread-safe /// \warning !thread-safe
LogInfo<S> info() LogInfo<S> info()
{ {
return{ log_head_ptr(), number_of_entries() }; return{ log_head_ptr(), number_of_entries() };
} }
}; };
/************* Models for sequential abstract data types *************/ /************* Models for sequential abstract data types *************/
class FlexibleBitset class FlexibleBitset
{ {
public: public:
typedef Bitset::Pos Pos; typedef Bitset::Pos Pos;
private: private:
Bitset m_bitset; Bitset m_bitset;
void allocate_blocks_if_neccessary(Pos pos) noexcept void allocate_blocks_if_neccessary(Pos pos) noexcept
{ {
if (pos < Bitset::s_bits_per_block) if (pos < Bitset::s_bits_per_block)
return; return;
assert(0U < pos); assert(0U < pos);
Bitset::BlockIndex new_size{ Bitset::blocks_size(pos) }; Bitset::BlockIndex new_size{ Bitset::blocks_size(pos) };
if (m_bitset.m_blocks.size() < new_size) { if (m_bitset.m_blocks.size() < new_size) {
m_bitset.m_blocks.resize(new_size); m_bitset.m_blocks.resize(new_size);
} }
} }
public: public:
FlexibleBitset() FlexibleBitset()
: m_bitset{ 1U } {} : m_bitset{ 1U } {}
FlexibleBitset(Pos max_pos) FlexibleBitset(Pos max_pos)
: m_bitset{ max_pos } {} : m_bitset{ max_pos } {}
bool is_empty() const noexcept bool is_empty() const noexcept
{ {
return m_bitset.is_empty(); return m_bitset.is_empty();
} }
bool set(Pos pos) bool set(Pos pos)
{ {
allocate_blocks_if_neccessary(pos); allocate_blocks_if_neccessary(pos);
return m_bitset.set(pos); return m_bitset.set(pos);
} }
bool is_set(Pos pos) const bool is_set(Pos pos) const
{ {
return m_bitset.is_set(pos); return m_bitset.is_set(pos);
} }
bool reset(Pos pos) bool reset(Pos pos)
{ {
allocate_blocks_if_neccessary(pos); allocate_blocks_if_neccessary(pos);
return m_bitset.reset(pos); return m_bitset.reset(pos);
} }
/// Same size && bits? /// Same size && bits?
bool operator==(const FlexibleBitset& other) const noexcept bool operator==(const FlexibleBitset& other) const noexcept
{ {
return m_bitset == other.m_bitset; return m_bitset == other.m_bitset;
} }
bool operator!=(const FlexibleBitset& other) const noexcept bool operator!=(const FlexibleBitset& other) const noexcept
{ {
return m_bitset != other.m_bitset; return m_bitset != other.m_bitset;
} }
std::size_t hash_code() const noexcept std::size_t hash_code() const noexcept
{ {
return m_bitset.m_hash; return m_bitset.m_hash;
} }
}; };
namespace state namespace state
{ {
namespace internal namespace internal
{ {
template<class S, class Ret> template<class S, class Ret>
struct RetOp : public Op<S> struct RetOp : public Op<S>
{ {
typedef RetOp<S, Ret> Base; typedef RetOp<S, Ret> Base;
const Ret ret; const Ret ret;
RetOp(Ret r) RetOp(Ret r)
: Op<S>(), ret{ r } {} : Op<S>(), ret{ r } {}
RetOp(Ret r, unsigned partition) RetOp(Ret r, unsigned partition)
: Op<S>(partition), ret{ r } {} : Op<S>(partition), ret{ r } {}
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
std::ostream& print(std::ostream& os) const override std::ostream& print(std::ostream& os) const override
{ {
return os << "ret: " << ret; return os << "ret: " << ret;
} }
#endif #endif
}; };
template<class S, const char* const op_name> template<class S, const char* const op_name>
struct ZeroArgOp : public Op<S> struct ZeroArgOp : public Op<S>
{ {
typedef ZeroArgOp<S, op_name> Base; typedef ZeroArgOp<S, op_name> Base;
ZeroArgOp() ZeroArgOp()
: Op<S>() {} : Op<S>() {}
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
std::ostream& print(std::ostream& os) const override std::ostream& print(std::ostream& os) const override
{ {
return os << op_name << "()"; return os << op_name << "()";
} }
#endif #endif
}; };
template<class S, class Value, const char* const op_name> template<class S, class Value, const char* const op_name>
struct ArgOp : public Op<S> struct ArgOp : public Op<S>
{ {
typedef ArgOp<S, Value, op_name> Base; typedef ArgOp<S, Value, op_name> Base;
const Value value; const Value value;
ArgOp(Value v) ArgOp(Value v)
: Op<S>(v), value{ v } {} : Op<S>(v), value{ v } {}
ArgOp(bool is_partitionable, Value v) ArgOp(bool is_partitionable, Value v)
: Op<S>(is_partitionable, v), value{ v } {} : Op<S>(is_partitionable, v), value{ v } {}
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
std::ostream& print(std::ostream& os) const override std::ostream& print(std::ostream& os) const override
{ {
return os << op_name << "(" << std::to_string(value) << ")"; return os << op_name << "(" << std::to_string(value) << ")";
} }
#endif #endif
}; };
} }
/// Byte read-write register with CAS /// Byte read-write register with CAS
class Atomic class Atomic
{ {
public: public:
typedef signed char Value; typedef signed char Value;
private: private:
static constexpr char s_read_op_name[5] = "read"; static constexpr char s_read_op_name[5] = "read";
static constexpr char s_write_op_name[6] = "write"; static constexpr char s_write_op_name[6] = "write";
struct ReadRetOp : public Op<Atomic> struct ReadRetOp : public Op<Atomic>
{ {
private: private:
const bool m_is_pending; const bool m_is_pending;
const Value m_value; const Value m_value;
public: public:
ReadRetOp(bool is_pending, Value value) ReadRetOp(bool is_pending, Value value)
: Op<Atomic>(), : Op<Atomic>(),
m_is_pending(is_pending), m_is_pending(is_pending),
m_value{ value } {} m_value{ value } {}
bool is_pending() const noexcept bool is_pending() const noexcept
{ {
return m_is_pending; return m_is_pending;
} }
/// \pre: !is_pending() /// \pre: !is_pending()
Value value() const Value value() const
{ {
assert(!m_is_pending); assert(!m_is_pending);
return m_value; return m_value;
} }
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
std::ostream& print(std::ostream& os) const override std::ostream& print(std::ostream& os) const override
{ {
if (m_is_pending) if (m_is_pending)
return os << "read() : pending"; return os << "read() : pending";
return os << "read() : " << std::to_string(m_value); return os << "read() : " << std::to_string(m_value);
} }
#endif #endif
}; };
struct ReadCallOp : public internal::ZeroArgOp<Atomic, s_read_op_name> struct ReadCallOp : public internal::ZeroArgOp<Atomic, s_read_op_name>
{ {
ReadCallOp() : Base() {} ReadCallOp() : Base() {}
std::pair<bool, Atomic> internal_apply(const Atomic& atomic, const Op<Atomic>& op) override std::pair<bool, Atomic> internal_apply(const Atomic& atomic, const Op<Atomic>& op) override
{ {
const ReadRetOp& read_ret = dynamic_cast<const ReadRetOp&>(op); const ReadRetOp& read_ret = dynamic_cast<const ReadRetOp&>(op);
if (read_ret.is_pending()) if (read_ret.is_pending())
return{ true, atomic }; return{ true, atomic };
return{ atomic.get() == read_ret.value(), atomic }; return{ atomic.get() == read_ret.value(), atomic };
} }
}; };
struct CASRetOp : public Op<Atomic> struct CASRetOp : public Op<Atomic>
{ {
private: private:
// 0: pending, 1: failed, 2: ok // 0: pending, 1: failed, 2: ok
const unsigned m_status; const unsigned m_status;
public: public:
CASRetOp(unsigned status) CASRetOp(unsigned status)
: Op<Atomic>(), : Op<Atomic>(),
m_status(status) {} m_status(status) {}
bool is_pending() const noexcept bool is_pending() const noexcept
{ {
return m_status == 0U; return m_status == 0U;
} }
/// \pre: !is_pending() /// \pre: !is_pending()
bool is_ok() const bool is_ok() const
{ {
assert(0U < m_status); assert(0U < m_status);
return m_status == 2U; return m_status == 2U;
} }
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
std::ostream& print(std::ostream& os) const override std::ostream& print(std::ostream& os) const override
{ {
os << "cas() : "; os << "cas() : ";
if (is_pending()) if (is_pending())
return os << "pending"; return os << "pending";
if (is_ok()) if (is_ok())
return os << "succeeded"; return os << "succeeded";
return os << "failed"; return os << "failed";
} }
#endif #endif
}; };
struct CASCallOp : public Op<Atomic> struct CASCallOp : public Op<Atomic>
{ {
const Value current_value, new_value; const Value current_value, new_value;
CASCallOp(Value current_v, Value new_v) CASCallOp(Value current_v, Value new_v)
: Op<Atomic>(), : Op<Atomic>(),
current_value{ current_v }, current_value{ current_v },
new_value{ new_v } {} new_value{ new_v } {}
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
std::ostream& print(std::ostream& os) const override std::ostream& print(std::ostream& os) const override
{ {
return os << "cas(" << std::to_string(current_value) << ", " << std::to_string(new_value) << ")"; return os << "cas(" << std::to_string(current_value) << ", " << std::to_string(new_value) << ")";
} }
#endif #endif
std::pair<bool, Atomic> internal_apply(const Atomic& atomic, const Op<Atomic>& op) override std::pair<bool, Atomic> internal_apply(const Atomic& atomic, const Op<Atomic>& op) override
{ {
const CASRetOp& cas_ret = dynamic_cast<const CASRetOp&>(op); const CASRetOp& cas_ret = dynamic_cast<const CASRetOp&>(op);
if (cas_ret.is_pending()) if (cas_ret.is_pending())
{ {
if (atomic.get() == current_value) if (atomic.get() == current_value)
return{ true, atomic.set(new_value) }; return{ true, atomic.set(new_value) };
return{ true, atomic }; return{ true, atomic };
} }
if (atomic.get() == current_value) if (atomic.get() == current_value)
return{ cas_ret.is_ok(), atomic.set(new_value) }; return{ cas_ret.is_ok(), atomic.set(new_value) };
return{ !cas_ret.is_ok(), atomic }; return{ !cas_ret.is_ok(), atomic };
} }
}; };
struct WriteRetOp : public Op<Atomic> struct WriteRetOp : public Op<Atomic>
{ {
const bool is_pending; const bool is_pending;
WriteRetOp(bool pending) WriteRetOp(bool pending)
: Op<Atomic>(), : Op<Atomic>(),
is_pending(pending) {} is_pending(pending) {}
#ifdef _LT_DEBUG_ #ifdef _LT_DEBUG_
std::ostream& print(std::ostream& os) const override std::ostream& print(std::ostream& os) const override
{ {
if (is_pending) if (is_pending)
return os << "write() : pending"; return os << "write() : pending";
return os << "write() : succeeded"; return os << "write() : succeeded";
} }
#endif #endif
}; };
struct WriteCallOp : public internal::ArgOp<Atomic, Value, s_write_op_name> struct WriteCallOp : public internal::ArgOp<Atomic, Value, s_write_op_name>
{ {
typedef internal::ArgOp<Atomic, Value, s_write_op_name> Base; typedef internal::ArgOp<Atomic, Value, s_write_op_name> Base;
WriteCallOp(Value new_value) WriteCallOp(Value new_value)
: Base(false, new_value) {} : Base(false, new_value) {}
std::pair<bool, Atomic> internal_apply(const Atomic& atomic, const Op<Atomic>& op) override std::pair<bool, Atomic> internal_apply(const Atomic& atomic, const Op<Atomic>& op) override
{ {
const WriteRetOp& write_ret = dynamic_cast<const WriteRetOp&>(op); const WriteRetOp& write_ret = dynamic_cast<const WriteRetOp&>(op);
// we don't need to check write_ret.is_pending because if the // we don't need to check write_ret.is_pending because if the
// write is pending then it could be still linearized last // write is pending then it could be still linearized last
return{ true, atomic.set(Base::value) }; return{ true, atomic.set(Base::value) };
} }
}; };
Value m_value; Value m_value;
Atomic(Value value) : m_value{ value } {} Atomic(Value value) : m_value{ value } {}
public: public:
typedef std::unique_ptr<Op<Atomic>> AtomicOpPtr; typedef std::unique_ptr<Op<Atomic>> AtomicOpPtr;
static AtomicOpPtr make_read_call() static AtomicOpPtr make_read_call()
{ {
return make_unique<ReadCallOp>(); return make_unique<ReadCallOp>();
} }
static AtomicOpPtr make_read_ret(Value v) static AtomicOpPtr make_read_ret(Value v)
{ {
return make_unique<ReadRetOp>(false, v); return make_unique<ReadRetOp>(false, v);
} }
static AtomicOpPtr make_read_pending() static AtomicOpPtr make_read_pending()
{ {
return make_unique<ReadRetOp>(true, '\0'); return make_unique<ReadRetOp>(true, '\0');
} }
static AtomicOpPtr make_write_call(Value v) static AtomicOpPtr make_write_call(Value v)
{ {
return make_unique<WriteCallOp>(v); return make_unique<WriteCallOp>(v);
} }
static AtomicOpPtr make_write_ret() static AtomicOpPtr make_write_ret()
{ {
return make_unique<WriteRetOp>(false); return make_unique<WriteRetOp>(false);
} }
static AtomicOpPtr make_write_pending() static AtomicOpPtr make_write_pending()
{ {
return make_unique<WriteRetOp>(true); return make_unique<WriteRetOp>(true);
} }
static AtomicOpPtr make_cas_call(Value curr_value, Value new_value) static AtomicOpPtr make_cas_call(Value curr_value, Value new_value)
{ {
return make_unique<CASCallOp>(curr_value, new_value); return make_unique<CASCallOp>(curr_value, new_value);
} }
static AtomicOpPtr make_cas_ret(bool ok) static AtomicOpPtr make_cas_ret(bool ok)
{ {
return make_unique<CASRetOp>(1U + ok); return make_unique<CASRetOp>(1U + ok);
} }
static AtomicOpPtr make_cas_pending() static AtomicOpPtr make_cas_pending()
{ {
return make_unique<CASRetOp>(0U); return make_unique<CASRetOp>(0U);
} }
/// Initially, register is negative /// Initially, register is negative
Atomic() : m_value{ -1 } {} Atomic() : m_value{ -1 } {}
Value get() const noexcept Value get() const noexcept
{ {
return m_value; return m_value;
} }
Atomic set(Value v) const noexcept Atomic set(Value v) const noexcept
{ {
return{ v }; return{ v };
} }
bool operator==(const Atomic& atomic) const bool operator==(const Atomic& atomic) const
{ {
return m_value == atomic.m_value; return m_value == atomic.m_value;
} }
bool operator!=(const Atomic& atomic) const bool operator!=(const Atomic& atomic) const
{ {
return m_value != atomic.m_value; return m_value != atomic.m_value;
} }
}; };
constexpr char Atomic::s_read_op_name[]; constexpr char Atomic::s_read_op_name[];
constexpr char Atomic::s_write_op_name[]; constexpr char Atomic::s_write_op_name[];
template<> template<>
struct Hash<Atomic> struct Hash<Atomic>
{ {
std::size_t operator()(const Atomic& atomic) const noexcept std::size_t operator()(const Atomic& atomic) const noexcept
{ {
return atomic.get() * 193U; return atomic.get() * 193U;
} }
}; };
} }
} }
#endif #endif
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or sign in to comment