wpi::hashing::detail Namespace Reference

Classes
struct	hash_combine_recursive_helper
	Helper class to manage the recursive combining of hash_combine arguments. More...
struct	hash_state
	The intermediate state used during hashing. More...
struct	is_hashable_data
	Trait to indicate whether a type's bits can be hashed directly. More...
struct	is_hashable_data< std::pair< T, U > >

Functions
uint64_t	fetch64 (const char *p)
uint32_t	fetch32 (const char *p)
uint64_t	rotate (uint64_t val, size_t shift)
	Bitwise right rotate.
uint64_t	shift_mix (uint64_t val)
uint64_t	hash_16_bytes (uint64_t low, uint64_t high)
uint64_t	hash_1to3_bytes (const char *s, size_t len, uint64_t seed)
uint64_t	hash_4to8_bytes (const char *s, size_t len, uint64_t seed)
uint64_t	hash_9to16_bytes (const char *s, size_t len, uint64_t seed)
uint64_t	hash_17to32_bytes (const char *s, size_t len, uint64_t seed)
uint64_t	hash_33to64_bytes (const char *s, size_t len, uint64_t seed)
uint64_t	hash_short (const char *s, size_t length, uint64_t seed)
uint64_t	get_execution_seed ()
	In LLVM_ENABLE_ABI_BREAKING_CHECKS builds, the seed is non-deterministic per process (address of a function in LLVMSupport) to prevent having users depend on the particular hash values.
template<typename T >
std::enable_if_t< is_hashable_data< T >::value, T >	get_hashable_data (const T &value)
	Helper to get the hashable data representation for a type.
template<typename T >
std::enable_if_t<!is_hashable_data< T >::value, size_t >	get_hashable_data (const T &value)
	Helper to get the hashable data representation for a type.
template<typename T >
bool	store_and_advance (char *&buffer_ptr, char *buffer_end, const T &value, size_t offset=0)
	Helper to store data from a value into a buffer and advance the pointer into that buffer.
template<typename InputIteratorT >
hash_code	hash_combine_range_impl (InputIteratorT first, InputIteratorT last)
	Implement the combining of integral values into a hash_code.
template<typename ValueT >
std::enable_if_t< is_hashable_data< ValueT >::value, hash_code >	hash_combine_range_impl (ValueT *first, ValueT *last)
	Implement the combining of integral values into a hash_code.
hash_code	hash_integer_value (uint64_t value)
	Helper to hash the value of a single integer.

Variables
static constexpr uint64_t	k0 = 0xc3a5c85c97cb3127ULL
	Some primes between 2^63 and 2^64 for various uses.
static constexpr uint64_t	k1 = 0xb492b66fbe98f273ULL
static constexpr uint64_t	k2 = 0x9ae16a3b2f90404fULL
static constexpr uint64_t	k3 = 0xc949d7c7509e6557ULL

Function Documentation

fetch32()

uint32_t wpi::hashing::detail::fetch32 ( const char * p )

inline

fetch64()

uint64_t wpi::hashing::detail::fetch64 ( const char * p )

inline

get_execution_seed()

uint64_t wpi::hashing::detail::get_execution_seed ( )

inline

In LLVM_ENABLE_ABI_BREAKING_CHECKS builds, the seed is non-deterministic per process (address of a function in LLVMSupport) to prevent having users depend on the particular hash values.

On platforms without ASLR, this is still likely non-deterministic per build.

get_hashable_data() [1/2]

template<typename T >

std::enable_if_t< is_hashable_data< T >::value, T > wpi::hashing::detail::get_hashable_data

(

const T &

value

)

Helper to get the hashable data representation for a type.

This variant is enabled when the type itself can be used.

get_hashable_data() [2/2]

template<typename T >

std::enable_if_t<!is_hashable_data< T >::value, size_t > wpi::hashing::detail::get_hashable_data

(

const T &

value

)

Helper to get the hashable data representation for a type.

This variant is enabled when we must first call hash_value and use the result as our data.

hash_16_bytes()

uint64_t wpi::hashing::detail::hash_16_bytes ( uint64_t low, uint64_t high )

inline

hash_17to32_bytes()

uint64_t wpi::hashing::detail::hash_17to32_bytes ( const char * s, size_t len, uint64_t seed )

inline

hash_1to3_bytes()

uint64_t wpi::hashing::detail::hash_1to3_bytes ( const char * s, size_t len, uint64_t seed )

inline

hash_33to64_bytes()

uint64_t wpi::hashing::detail::hash_33to64_bytes ( const char * s, size_t len, uint64_t seed )

inline

hash_4to8_bytes()

uint64_t wpi::hashing::detail::hash_4to8_bytes ( const char * s, size_t len, uint64_t seed )

inline

hash_9to16_bytes()

uint64_t wpi::hashing::detail::hash_9to16_bytes ( const char * s, size_t len, uint64_t seed )

inline

hash_combine_range_impl() [1/2]

template<typename InputIteratorT >

hash_code wpi::hashing::detail::hash_combine_range_impl	(	InputIteratorT	first,
		InputIteratorT	last )

Implement the combining of integral values into a hash_code.

This overload is selected when the value type of the iterator is integral. Rather than computing a hash_code for each object and then combining them, this (as an optimization) directly combines the integers.

hash_combine_range_impl() [2/2]

template<typename ValueT >

std::enable_if_t< is_hashable_data< ValueT >::value, hash_code > wpi::hashing::detail::hash_combine_range_impl	(	ValueT *	first,
		ValueT *	last )

Implement the combining of integral values into a hash_code.

This overload is selected when the value type of the iterator is integral and when the input iterator is actually a pointer. Rather than computing a hash_code for each object and then combining them, this (as an optimization) directly combines the integers. Also, because the integers are stored in contiguous memory, this routine avoids copying each value and directly reads from the underlying memory.

hash_integer_value()

hash_code wpi::hashing::detail::hash_integer_value ( uint64_t value )

inline

Helper to hash the value of a single integer.

Overloads for smaller integer types are not provided to ensure consistent behavior in the presence of integral promotions. Essentially, "hash_value('4')" and "hash_value('0' + 4)" should be the same.

hash_short()

uint64_t wpi::hashing::detail::hash_short ( const char * s, size_t length, uint64_t seed )

inline

rotate()

uint64_t wpi::hashing::detail::rotate ( uint64_t val, size_t shift )

inline

Bitwise right rotate.

Normally this will compile to a single instruction, especially if the shift is a manifest constant.

shift_mix()

uint64_t wpi::hashing::detail::shift_mix ( uint64_t val )

inline

store_and_advance()

template<typename T >

bool wpi::hashing::detail::store_and_advance	(	char *&	buffer_ptr,
		char *	buffer_end,
		const T &	value,
		size_t	offset = 0 )

Helper to store data from a value into a buffer and advance the pointer into that buffer.

This routine first checks whether there is enough space in the provided buffer, and if not immediately returns false. If there is space, it copies the underlying bytes of value into the buffer, advances the buffer_ptr past the copied bytes, and returns true.

Variable Documentation

k0

uint64_t wpi::hashing::detail::k0 = 0xc3a5c85c97cb3127ULL

staticconstexpr

Some primes between 2^63 and 2^64 for various uses.

k1

uint64_t wpi::hashing::detail::k1 = 0xb492b66fbe98f273ULL

staticconstexpr

k2

uint64_t wpi::hashing::detail::k2 = 0x9ae16a3b2f90404fULL

staticconstexpr

k3

uint64_t wpi::hashing::detail::k3 = 0xc949d7c7509e6557ULL

staticconstexpr