On Tue, Nov 25, 2025 at 10:08 PM Jonathan Wakely <[email protected]> wrote:
> This will allow us to extend atomic waiting functions to support a
> possible future 64-bit version of futex, as well as supporting
> futex-like wait/wake primitives on other targets (e.g. macOS has
> os_sync_wait_on_address and FreeBSD has _umtx_op).
>
> Before this change, the decision of whether to do a proxy wait or to
> wait on the atomic variable itself was made in the header at
> compile-time, which makes it an ABI property that would not have been
> possible to change later. That would have meant that
> std::atomic<uint64_t> would always have to do a proxy wait even if Linux
> gains support for 64-bit futex2(2) calls at some point in the future.
> The disadvantage of proxy waits is that several distinct atomic objects
> can share the same proxy state, leading to contention between threads
> even when they are not waiting on the same atomic object, similar to
> false sharing. It also result in spurious wake-ups because doing a
> notify on an atomic object that uses a proxy wait will wake all waiters
> sharing the proxy.
>
> For types that are known to definitely not need a proxy wait (e.g. int
> on Linux) the header can still choose a more efficient path at
> compile-time. But for other types, the decision of whether to do a proxy
> wait is deferred to runtime, inside the library internals. This will
> make it possible for future versions of libstdc++.so to extend the set
> of types which don't need to use proxy waits, without ABI changes.
>
> The way the change works is to stop using the __proxy_wait flag that was
> set by the inline code in the headers. Instead the __wait_args struct
> has an extra pointer member which the library internals populate with
> either the address of the atomic object or the _M_ver counter in the
> proxy state. There is also a new _M_obj_size member which stores the
> size of the atomic object, so that the library can decide whether a
> proxy is needed. So for example if linux gains 64-bit futex support then
> the library can decide not to use a proxy when _M_obj_size == 8.
> Finally, the _M_old member of the __wait_args struct is changed to
> uint64_t so that it has room to store 64-bit values, not just whatever
> size the __platform_wait_t type is (which is a 32-bit int on Linux).
> Similarly, the _M_val member of __wait_result_type changes to uint64_t
> too.
>
> libstdc++-v3/ChangeLog:
>
> * config/abi/pre/gnu.ver: Adjust exports.
> * include/bits/atomic_timed_wait.h
> (_GLIBCXX_HAVE_PLATFORM_TIMED_WAIT):
> Do not define this macro.
> (__atomic_wait_address_until_v, __atomic_wait_address_for_v):
> Adjust assertions to check that __platform_wait_uses_type is
> true.
> * include/bits/atomic_wait.h (__waitable): New concept.
> (__platform_wait_uses_type): Different separately for platforms
> with and without platform wait.
> (_GLIBCXX_HAVE_PLATFORM_WAIT): Do not define this macro.
> (__wait_value_type): New typedef.
> (__wait_result_type): Change _M_val to __wait_value_type.
> (__wait_flags): Remove __proxy_wait enumerator. Reduce range
> reserved for ABI version by the commented-out value.
> (__wait_args_base::_M_old): Change type to __wait_args_base.
> (__wait_args_base::_M_obj, __wait_args_base::_M_obj_size): New
> data members.
> (__wait_args::__wait_args): Set _M_obj and _M_obj_size on
> construction.
> (__wait_args::_M_setup_wait): Change void* parameter to deduced
> type. Adjust bit_cast to work for types of different sizes.
> (__wait_args::_M_load_proxy_wait_val): Remove function, replace
> with ...
> (__wait_args::_M_setup_wait_impl): New function.
> (__wait_args::_S_flags_for): Do not set __proxy_wait flag.
> (__atomic_wait_address_v): Adjust assertion to check that
> __platform_wait_uses_type is true.
> * src/c++20/atomic.cc (_GLIBCXX_HAVE_PLATFORM_WAIT): Define here
> instead of in header. Check _GLIBCXX_HAVE_PLATFORM_WAIT instead
> of _GLIBCXX_HAVE_PLATFORM_TIMED_WAIT.
> (__platform_wait, __platform_notify, __platform_wait_until): Add
> unused parameter for _M_obj_size.
> (__spin_impl): Adjust for 64-bit __wait_args_base::_M_old.
> (use_proxy_wait): New function.
> (__wait_args::_M_load_proxy_wait_val): Replace with ...
> (__wait_args::_M_setup_wait_impl): New function. Call
> use_proxy_wait to decide at runtime whether to wait on the
> pointer directly instead of using a proxy. If a proxy is needed,
> set _M_obj and _M_obj_size to refer to its _M_ver member. Adjust
> for change to type of _M_old.
> (__wait_impl): Wait on _M_obj unconditionally. Pass _M_obj_size
> to __platform_wait.
> (__notify_impl): Call use_proxy_wait to decide whether to notify
> on the address parameter or a proxy
> (__spin_until_impl): Adjust for change to type of _M_val.
> (__wait_until_impl): Wait on _M_obj unconditionally. Pass
> _M_obj_size to __platform_wait_until.
> ---
>
> Tested x86_64-linux and x86_64-freebsd14.
>
> v3:
> - Moved definition of __platform_wait_until adjacent to __platform_wait
> and __platform_notify.
> - add obj_sz parameter to __platform_notify.
>
Still have questions regarding the bit_cast in _M_setup_wait, as we seem
to use it consistently
(we either need one or do not need one).
>
> libstdc++-v3/config/abi/pre/gnu.ver | 3 +-
> libstdc++-v3/include/bits/atomic_timed_wait.h | 20 +-
> libstdc++-v3/include/bits/atomic_wait.h | 119 ++++++---
> libstdc++-v3/src/c++20/atomic.cc | 228 ++++++++++--------
> 4 files changed, 216 insertions(+), 154 deletions(-)
>
> diff --git a/libstdc++-v3/config/abi/pre/gnu.ver
> b/libstdc++-v3/config/abi/pre/gnu.ver
> index 2e48241d51f9..3c2bd4921730 100644
> --- a/libstdc++-v3/config/abi/pre/gnu.ver
> +++ b/libstdc++-v3/config/abi/pre/gnu.ver
> @@ -2553,7 +2553,8 @@ GLIBCXX_3.4.35 {
> _ZNSt8__detail11__wait_implEPKvRNS_16__wait_args_baseE;
> _ZNSt8__detail13__notify_implEPKvbRKNS_16__wait_args_baseE;
>
>
> _ZNSt8__detail17__wait_until_implEPKvRNS_16__wait_args_baseERKNSt6chrono8durationI[lx]St5ratioIL[lx]1EL[lx]1000000000EEEE;
> - _ZNSt8__detail11__wait_args22_M_load_proxy_wait_valEPKv;
> + _ZNSt8__detail11__wait_args18_M_setup_wait_implEPKv;
> + _ZNSt8__detail11__wait_args20_M_setup_notify_implEPKv;
>
> # std::chrono::gps_clock::now, tai_clock::now
> _ZNSt6chrono9gps_clock3nowEv;
> diff --git a/libstdc++-v3/include/bits/atomic_timed_wait.h
> b/libstdc++-v3/include/bits/atomic_timed_wait.h
> index 30f7ff616840..5b3158050668 100644
> --- a/libstdc++-v3/include/bits/atomic_timed_wait.h
> +++ b/libstdc++-v3/include/bits/atomic_timed_wait.h
> @@ -75,14 +75,6 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
> return chrono::ceil<__w_dur>(__atime);
> }
>
> -#ifdef _GLIBCXX_HAVE_LINUX_FUTEX
> -#define _GLIBCXX_HAVE_PLATFORM_TIMED_WAIT
> -#else
> -// define _GLIBCXX_HAVE_PLATFORM_TIMED_WAIT and implement
> __platform_wait_until
> -// if there is a more efficient primitive supported by the platform
> -// (e.g. __ulock_wait) which is better than pthread_cond_clockwait.
> -#endif // ! HAVE_LINUX_FUTEX
> -
> __wait_result_type
> __wait_until_impl(const void* __addr, __wait_args_base& __args,
> const __wait_clock_t::duration& __atime);
> @@ -156,9 +148,9 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
> const chrono::time_point<_Clock, _Dur>&
> __atime,
> bool __bare_wait = false) noexcept
> {
> -#ifndef _GLIBCXX_HAVE_PLATFORM_TIMED_WAIT
> - __glibcxx_assert(false); // This function can't be used for proxy
> wait.
> -#endif
> + // This function must not be used if __wait_impl might use a proxy
> wait:
> +
> __glibcxx_assert(__platform_wait_uses_type<__detail::__platform_wait_t>);
> +
> __detail::__wait_args __args{ __addr, __old, __order, __bare_wait };
> auto __res = __detail::__wait_until(__addr, __args, __atime);
> return !__res._M_timeout; // C++26 will also return last observed
> __val
> @@ -208,9 +200,9 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
> const chrono::duration<_Rep, _Period>&
> __rtime,
> bool __bare_wait = false) noexcept
> {
> -#ifndef _GLIBCXX_HAVE_PLATFORM_TIMED_WAIT
> - __glibcxx_assert(false); // This function can't be used for proxy
> wait.
> -#endif
> + // This function must not be used if __wait_impl might use a proxy
> wait:
> +
> __glibcxx_assert(__platform_wait_uses_type<__detail::__platform_wait_t>);
> +
> __detail::__wait_args __args{ __addr, __old, __order, __bare_wait };
> auto __res = __detail::__wait_for(__addr, __args, __rtime);
> return !__res._M_timeout; // C++26 will also return last observed
> __val
> diff --git a/libstdc++-v3/include/bits/atomic_wait.h
> b/libstdc++-v3/include/bits/atomic_wait.h
> index 95151479c120..a280d3534f46 100644
> --- a/libstdc++-v3/include/bits/atomic_wait.h
> +++ b/libstdc++-v3/include/bits/atomic_wait.h
> @@ -45,35 +45,50 @@
> namespace std _GLIBCXX_VISIBILITY(default)
> {
> _GLIBCXX_BEGIN_NAMESPACE_VERSION
> +
> namespace __detail
> {
> -#ifdef _GLIBCXX_HAVE_LINUX_FUTEX
> -#define _GLIBCXX_HAVE_PLATFORM_WAIT 1
> + // TODO: this needs to be false for types with padding, e.g. __int20.
> + // TODO: should this be true only for integral, enum, and pointer
> types?
> + template<typename _Tp>
> + concept __waitable
> + = is_scalar_v<_Tp> && (sizeof(_Tp) <= sizeof(__UINT64_TYPE__));
> + }
> +
> +#if defined _GLIBCXX_HAVE_LINUX_FUTEX
> + namespace __detail
> + {
> + // Use futex syscall on int objects.
> using __platform_wait_t = int;
> inline constexpr size_t __platform_wait_alignment = 4;
> + }
> + // Defined to true for a subset of __waitable types which are statically
> + // known to definitely be able to use futex, not a proxy wait.
> + template<typename _Tp>
> + inline constexpr bool __platform_wait_uses_type
> + = __detail::__waitable<_Tp>
> + && sizeof(_Tp) == sizeof(int) && alignof(_Tp) >= 4;
> #else
> // define _GLIBCX_HAVE_PLATFORM_WAIT and implement __platform_wait()
> // and __platform_notify() if there is a more efficient primitive
> supported
> // by the platform (e.g. __ulock_wait()/__ulock_wake()) which is better
> than
> // a mutex/condvar based wait.
> + namespace __detail
> + {
> # if ATOMIC_LONG_LOCK_FREE == 2
> using __platform_wait_t = unsigned long;
> # else
> using __platform_wait_t = unsigned int;
> # endif
> inline constexpr size_t __platform_wait_alignment
> - = __alignof__(__platform_wait_t);
> -#endif
> + = sizeof(__platform_wait_t) < __alignof__(__platform_wait_t)
> + ? __alignof__(__platform_wait_t) : sizeof(__platform_wait_t);
> } // namespace __detail
>
> - template<typename _Tp>
> - inline constexpr bool __platform_wait_uses_type
> -#ifdef _GLIBCXX_HAVE_PLATFORM_WAIT
> - = is_scalar_v<_Tp>
> - && ((sizeof(_Tp) == sizeof(__detail::__platform_wait_t))
> - && (alignof(_Tp) >= __detail::__platform_wait_alignment));
> -#else
> - = false;
> + // This must be false for the general case where we don't know of any
> + // futex-like syscall.
> + template<typename>
> + inline constexpr bool __platform_wait_uses_type = false;
> #endif
>
> namespace __detail
> @@ -105,10 +120,13 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
> return __builtin_memcmp(&__a, &__b, sizeof(_Tp)) == 0;
> }
>
> - // lightweight std::optional<__platform_wait_t>
> + // Storage for up to 64 bits of value, should be considered opaque
> bits.
> + using __wait_value_type = __UINT64_TYPE__;
> +
> + // lightweight std::optional<__wait_value_type>
> struct __wait_result_type
> {
> - __platform_wait_t _M_val;
> + __wait_value_type _M_val;
> unsigned char _M_has_val : 1; // _M_val value was loaded before
> return.
> unsigned char _M_timeout : 1; // Waiting function ended with
> timeout.
> unsigned char _M_unused : 6; // padding
> @@ -116,12 +134,12 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
>
> enum class __wait_flags : __UINT_LEAST32_TYPE__
> {
> - __abi_version = 0,
> - __proxy_wait = 1,
> + __abi_version = 0x00000000,
> + // currently unused = 1,
> __track_contention = 2,
> __do_spin = 4,
> __spin_only = 8, // Ignored unless __do_spin is also set.
> - // __abi_version_mask = 0xffff0000,
> + // __abi_version_mask = 0xff000000,
> };
>
> [[__gnu__::__always_inline__]]
> @@ -143,8 +161,10 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
> {
> __wait_flags _M_flags;
> int _M_order = __ATOMIC_ACQUIRE;
> - __platform_wait_t _M_old = 0;
> + __wait_value_type _M_old = 0;
> void* _M_wait_state = nullptr;
> + const void* _M_obj = nullptr; // The address of the object to wait
> on.
> + unsigned char _M_obj_size = 0; // The size of that object.
>
> // Test whether _M_flags & __flags is non-zero.
> bool
> @@ -162,53 +182,74 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
> explicit
> __wait_args(const _Tp* __addr, bool __bare_wait = false) noexcept
> : __wait_args_base{ _S_flags_for(__addr, __bare_wait) }
> - { }
> + {
> + _M_obj = __addr; // Might be replaced by _M_setup_wait
> + if constexpr (__waitable<_Tp>)
> + // __wait_impl might be able to wait directly on __addr
> + // instead of using a proxy, depending on its size.
> + _M_obj_size = sizeof(_Tp);
> + }
>
> __wait_args(const __platform_wait_t* __addr, __platform_wait_t
> __old,
> int __order, bool __bare_wait = false) noexcept
> - : __wait_args_base{ _S_flags_for(__addr, __bare_wait), __order,
> __old }
> - { }
> + : __wait_args(__addr, __bare_wait)
> + {
> + _M_order = __order;
> + _M_old = __old;
> + }
>
> __wait_args(const __wait_args&) noexcept = default;
> __wait_args& operator=(const __wait_args&) noexcept = default;
>
> - template<typename _ValFn,
> - typename _Tp = decay_t<decltype(std::declval<_ValFn&>()())>>
> + template<typename _Tp, typename _ValFn>
> _Tp
> - _M_setup_wait(const void* __addr, _ValFn __vfn,
> + _M_setup_wait(const _Tp* __addr, _ValFn __vfn,
> __wait_result_type __res = {})
> {
> + static_assert(is_same_v<_Tp, decay_t<decltype(__vfn())>>);
> +
> if constexpr (__platform_wait_uses_type<_Tp>)
> {
> - // If the wait is not proxied, the value we check when
> waiting
> - // is the value of the atomic variable itself.
> + // If we know for certain that this type can be waited on
> + // efficiently using something like a futex syscall,
> + // then we can avoid the overhead of _M_setup_wait_impl
> + // and just load the value and store it into _M_old.
>
> - if (__res._M_has_val) // The previous wait loaded a recent
> value.
> + if (__res._M_has_val) // A previous wait loaded a recent
> value.
> {
> _M_old = __res._M_val;
> - return __builtin_bit_cast(_Tp, __res._M_val);
> + return (_Tp)_M_old;
>
These two branches are not self consistent, in deteriminig what is required
when
going between interger and _Tp. When storing we are doing a bit_cast (i.e.
this
may be not integer) but when loading we simply cast it, assuming it is
integer.
So either the integral cast should be fine both ways, or loading the value
needs
something like:
if constexpr (sizeof(_Tp) == sizeof(__UINT32_TYPE__))
return__builtin_bit_cast(_Tp, (__UINT32_TYPE__)_M_old);
else if constexpr (sizeof(_Tp) == sizeof(__UINT64_TYPE__))
return__builtin_bit_cast(_Tp, (__UINT64_TYPE__)_M_old);
else
static_assert(false); // Unsupported size
}
> else // Load the value from __vfn
> {
> - _Tp __val = __vfn();
> - _M_old = __builtin_bit_cast(__platform_wait_t, __val);
> + auto __val = __vfn();
> + if constexpr (sizeof(_Tp) == sizeof(__UINT32_TYPE__))
> + _M_old = __builtin_bit_cast(__UINT32_TYPE__, __val);
> + else if constexpr (sizeof(_Tp) ==
> sizeof(__UINT64_TYPE__))
> + _M_old = __builtin_bit_cast(__UINT64_TYPE__, __val);
> + else
> + static_assert(false); // Unsupported size
> return __val;
> }
> }
> - else // It's a proxy wait and the proxy's _M_ver is used.
> + else // The library will decide whether to use a proxy wait.
> {
> if (__res._M_has_val) // The previous wait loaded a recent
> value.
> _M_old = __res._M_val;
> - else // Load _M_ver from the proxy (must happen before
> __vfn()).
> - _M_load_proxy_wait_val(__addr);
> + else // Let the library decide how to setup the wait.
> + {
> + // Set _M_obj to the address to be waited on (either
> __addr
> + // or a proxy) and load its current value into _M_old.
> + _M_setup_wait_impl(__addr);
> + }
> return __vfn();
> }
> }
>
> private:
> - // Populates _M_wait_state and _M_old from the proxy for __addr.
> + // Populates _M_wait_state and _M_old appropriately for __addr.
> void
> - _M_load_proxy_wait_val(const void* __addr);
> + _M_setup_wait_impl(const void* __addr);
>
> template<typename _Tp>
> static constexpr __wait_flags
> @@ -218,8 +259,6 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
> __wait_flags __res = __abi_version | __do_spin;
> if (!__bare_wait)
> __res |= __track_contention;
> - if constexpr (!__platform_wait_uses_type<_Tp>)
> - __res |= __proxy_wait;
> return __res;
> }
> };
> @@ -255,9 +294,9 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
> __detail::__platform_wait_t __old,
> int __order, bool __bare_wait = false)
> {
> -#ifndef _GLIBCXX_HAVE_PLATFORM_WAIT
> - __glibcxx_assert(false); // This function can't be used for proxy
> wait.
> -#endif
> + // This function must not be used if __wait_impl might use a proxy
> wait:
> +
> __glibcxx_assert(__platform_wait_uses_type<__detail::__platform_wait_t>);
> +
> __detail::__wait_args __args{ __addr, __old, __order, __bare_wait };
> // C++26 will not ignore the return value here
> __detail::__wait_impl(__addr, __args);
> diff --git a/libstdc++-v3/src/c++20/atomic.cc
> b/libstdc++-v3/src/c++20/atomic.cc
> index e280045b619d..dfbf4a95cd62 100644
> --- a/libstdc++-v3/src/c++20/atomic.cc
> +++ b/libstdc++-v3/src/c++20/atomic.cc
> @@ -27,25 +27,18 @@
> #if __glibcxx_atomic_wait
> #include <atomic>
> #include <bits/atomic_timed_wait.h>
> -#include <bits/functional_hash.h>
> -#include <cstdint>
> +#include <cstdint> // uint32_t, uint64_t
> +#include <climits> // INT_MAX
> +#include <cerrno> // errno, ETIMEDOUT, etc.
> #include <bits/std_mutex.h> // std::mutex, std::__condvar
> +#include <bits/functexcept.h> // __throw_system_error
> +#include <bits/functional_hash.h>
>
> #ifdef _GLIBCXX_HAVE_LINUX_FUTEX
> -# include <cerrno>
> -# include <climits>
> # include <unistd.h>
> # include <syscall.h>
> -# include <bits/functexcept.h>
> -# include <sys/time.h>
> -#endif
> -
> -#ifdef _GLIBCXX_HAVE_PLATFORM_WAIT
> -# ifndef _GLIBCXX_HAVE_PLATFORM_TIMED_WAIT
> -// __waitable_state assumes that we consistently use the same
> implementation
> -// (i.e. futex vs mutex+condvar) for timed and untimed waiting.
> -# error "This configuration is not currently supported"
> -# endif
> +# include <sys/time.h> // timespec
> +# define _GLIBCXX_HAVE_PLATFORM_WAIT 1
> #endif
>
> #pragma GCC diagnostic ignored "-Wmissing-field-initializers"
> @@ -77,25 +70,45 @@ namespace
> };
>
> void
> - __platform_wait(const int* __addr, int __val) noexcept
> + __platform_wait(const int* addr, int val, int /* obj_size */) noexcept
> {
> - auto __e = syscall (SYS_futex, __addr,
> -
> static_cast<int>(__futex_wait_flags::__wait_private),
> - __val, nullptr);
> - if (!__e || errno == EAGAIN)
> - return;
> - if (errno != EINTR)
> - __throw_system_error(errno);
> + if (syscall(SYS_futex, addr,
> + static_cast<int>(__futex_wait_flags::__wait_private),
> + val, nullptr))
> + if (errno != EAGAIN && errno != EINTR)
> + __throw_system_error(errno);
> }
>
> void
> - __platform_notify(const int* __addr, bool __all) noexcept
> + __platform_notify(const int* addr, bool all, int) noexcept
> {
> - syscall (SYS_futex, __addr,
> - static_cast<int>(__futex_wait_flags::__wake_private),
> - __all ? INT_MAX : 1);
> + syscall(SYS_futex, addr,
> + static_cast<int>(__futex_wait_flags::__wake_private),
> + all ? INT_MAX : 1);
> }
> -#endif
> +
> + // returns true if wait ended before timeout
> + bool
> + __platform_wait_until(const __platform_wait_t* addr,
> + __platform_wait_t val,
> + const __wait_clock_t::time_point& __atime,
> + int /* obj_size */) noexcept
> + {
> + struct timespec timeout = chrono::__to_timeout_timespec(__atime);
> +
> + if (syscall(SYS_futex, addr,
> +
> static_cast<int>(__futex_wait_flags::__wait_bitset_private),
> + val, &timeout, nullptr,
> + static_cast<int>(__futex_wait_flags::__bitset_match_any)))
> + {
> + if (errno == ETIMEDOUT)
> + return false;
> + if (errno != EAGAIN && errno != EINTR)
> + __throw_system_error(errno);
> + }
> + return true;
> + }
> +#endif // HAVE_LINUX_FUTEX
>
> // The state used by atomic waiting and notifying functions.
> struct __waitable_state
> @@ -107,7 +120,7 @@ namespace
> // Count of threads blocked waiting on this state.
> alignas(_S_align) __platform_wait_t _M_waiters = 0;
>
> -#ifndef _GLIBCXX_HAVE_PLATFORM_TIMED_WAIT
> +#ifndef _GLIBCXX_HAVE_PLATFORM_WAIT
> mutex _M_mtx;
>
> // This type meets the Cpp17BasicLockable requirements.
> @@ -123,7 +136,7 @@ namespace
> // use this for waiting and notifying functions instead.
> alignas(_S_align) __platform_wait_t _M_ver = 0;
>
> -#ifndef _GLIBCXX_HAVE_PLATFORM_TIMED_WAIT
> +#ifndef _GLIBCXX_HAVE_PLATFORM_WAIT
> __condvar _M_cv;
> #endif
>
> @@ -215,18 +228,18 @@ namespace
> __wait_result_type
> __spin_impl(const __platform_wait_t* __addr, const __wait_args_base&
> __args)
> {
> - __platform_wait_t __val{};
> + __wait_value_type wval;
> for (auto __i = 0; __i < __atomic_spin_count; ++__i)
> {
> - __atomic_load(__addr, &__val, __args._M_order);
> - if (__val != __args._M_old)
> - return { ._M_val = __val, ._M_has_val = true, ._M_timeout =
> false };
> + wval = __atomic_load_n(__addr, __args._M_order);
> + if (wval != __args._M_old)
> + return { ._M_val = wval, ._M_has_val = true, ._M_timeout = false
> };
> if (__i < __atomic_spin_count_relax)
> __thread_relax();
> else
> __thread_yield();
> }
> - return { ._M_val = __val, ._M_has_val = true, ._M_timeout = true };
> + return { ._M_val = wval, ._M_has_val = true, ._M_timeout = true };
> }
>
> inline __waitable_state*
> @@ -237,32 +250,58 @@ namespace
> return static_cast<__waitable_state*>(args._M_wait_state);
> }
>
> + [[gnu::always_inline]]
> + inline bool
> + use_proxy_wait(const __wait_args_base& args, const void* /* addr */)
> + {
> + if constexpr (__platform_wait_uses_type<uint32_t>)
> + if (args._M_obj_size == sizeof(uint32_t))
> + return false;
> +
> + if constexpr (__platform_wait_uses_type<uint64_t>)
> + if (args._M_obj_size == sizeof(uint64_t))
> + return false;
> +
> + // Use proxy wait for everything else:
> + return true;
> + }
> +
> } // namespace
>
> -// Called for a proxy wait
> +// Set _M_wait_state if using proxy wait, or caller wants contention
> tracking.
> +// Set _M_obj to &_M_wait_state->_M_ver if using proxy wait.
> +// Load the current value from _M_obj and store in _M_val.
> void
> -__wait_args::_M_load_proxy_wait_val(const void* addr)
> +__wait_args::_M_setup_wait_impl(const void* addr)
> {
> - // __glibcxx_assert( *this & __wait_flags::__proxy_wait );
> + if (!use_proxy_wait(*this, addr))
> + {
> + // We can wait on this address directly.
> + __glibcxx_assert(_M_obj == addr);
>
> - // We always need a waitable state for proxy waits.
> + int val;
> + __atomic_load(static_cast<const int*>(addr), &val,
> __ATOMIC_ACQUIRE);
> + _M_old = val;
>
This seems like we should load either 4B or 8B depending on the _M_size.
And also if we are not using proxy wait, we are setting _M_old to the value
produced by __vfn parameter
(functor). I think this function should return bool, indicating if the
value of _M_old
should be set, here. And the _M_setup_wait would like something like this:
if (__res._M_has_val) // A previous wait loaded a recent value.
{
_M_old = __res._M_val;
if constexpr (!__platform_wait_uses_type<_Tp>)
return __vfn(); // possibly proxy wait _M_old is not usabel
// We know that proxy wait is not used, so we can return
// return last loaded value instead of calling it __vfn() to load it.
else if constexpr (sizeof(_Tp) == sizeof(__UINT32_TYPE__))
return __builtin_bit_cast(_Tp, (__UINT32_TYPE__)_M_old);
else if f constexpr (sizeof(_Tp) == sizeof(__UINT64_TYPE__))
return __builtin_bit_cast(_Tp, (__UINT64_TYPE__)_M_old);
else
static_assert(false); // Unsupported size
}
// If we know for certain that this type can be waited on
// efficiently using something like a futex syscall,
// then we can avoid the overhead of _M_setup_wait_impl.
if (!__platform_wait_uses_type<_Tp>)
if (!_M_setup_wait_impl(__addr))
// proxy wait was used, and _M_old was set accordingly
return __vfn();
// No proxy wait, set _M_old based on value.
auto __val = __vfn();
if constexpr (sizeof(_Tp) == sizeof(__UINT32_TYPE__))
_M_old = __builtin_bit_cast(__UINT32_TYPE__, __val);
else if constexpr (sizeof(_Tp) == sizeof(__UINT64_TYPE__))
_M_old = __builtin_bit_cast(__UINT64_TYPE__, __val);
else
static_assert(false); // Unsupported size
return __val;
+
> + return;
> + }
> +
> + // This will be a proxy wait, so get a waitable state.
> auto state = set_wait_state(addr, *this);
>
> + // The address we will wait on is the version count of the waitable
> state:
> + _M_obj = &state->_M_ver;
> + // __wait_impl and __wait_until_impl need to know this size:
> + _M_obj_size = sizeof(state->_M_ver);
> +
> // Read the value of the _M_ver counter.
> - __atomic_load(&state->_M_ver, &_M_old, __ATOMIC_ACQUIRE);
> + _M_old = __atomic_load_n(&state->_M_ver, __ATOMIC_ACQUIRE);
> }
>
> __wait_result_type
> -__wait_impl(const void* __addr, __wait_args_base& __args)
> +__wait_impl([[maybe_unused]] const void* __addr, __wait_args_base& __args)
> {
> - auto __state = static_cast<__waitable_state*>(__args._M_wait_state);
> -
> - const __platform_wait_t* __wait_addr;
> -
> - if (__args & __wait_flags::__proxy_wait)
> - __wait_addr = &__state->_M_ver;
> - else
> - __wait_addr = static_cast<const __platform_wait_t*>(__addr);
> + auto* __wait_addr = static_cast<const
> __platform_wait_t*>(__args._M_obj);
>
> if (__args & __wait_flags::__do_spin)
> {
> @@ -277,7 +316,7 @@ __wait_impl(const void* __addr, __wait_args_base&
> __args)
> if (__args & __wait_flags::__track_contention)
> set_wait_state(__addr, __args); // scoped_wait needs a
> __waitable_state
> scoped_wait s(__args);
> - __platform_wait(__wait_addr, __args._M_old);
> + __platform_wait(__wait_addr, __args._M_old, __args._M_obj_size);
> // We haven't loaded a new value so return _M_has_val=false
> return { ._M_val = __args._M_old, ._M_has_val = false, ._M_timeout =
> false };
> #else
> @@ -286,6 +325,7 @@ __wait_impl(const void* __addr, __wait_args_base&
> __args)
> __atomic_load(__wait_addr, &__val, __args._M_order);
> if (__val == __args._M_old)
> {
> + auto __state = static_cast<__waitable_state*>(__args._M_wait_state);
> __state->_M_cv.wait(__state->_M_mtx);
> return { ._M_val = __val, ._M_has_val = false, ._M_timeout = false
> };
> }
> @@ -294,24 +334,40 @@ __wait_impl(const void* __addr, __wait_args_base&
> __args)
> }
>
> void
> -__notify_impl(const void* __addr, [[maybe_unused]] bool __all,
> +__notify_impl([[maybe_unused]] const void* __addr, [[maybe_unused]] bool
> __all,
> const __wait_args_base& __args)
> {
> - auto __state = static_cast<__waitable_state*>(__args._M_wait_state);
> - if (!__state)
> - __state = &__waitable_state::_S_state_for(__addr);
> + const bool __track_contention = __args &
> __wait_flags::__track_contention;
> + const bool proxy_wait = use_proxy_wait(__args, __addr);
>
> - [[maybe_unused]] const __platform_wait_t* __wait_addr;
> + [[maybe_unused]] auto* __wait_addr
> + = static_cast<const __platform_wait_t*>(__addr);
> +
> +#ifdef _GLIBCXX_HAVE_PLATFORM_WAIT
> + // Check whether it would be a non-proxy wait for this object.
> + // This condition must match the one in _M_setup_wait_impl to ensure
> that
> + // the address used for the notify matches the one used for the wait.
> + if (!proxy_wait)
> + {
> + if (__track_contention)
> + if (!__waitable_state::_S_state_for(__addr)._M_waiting())
> + return;
> +
> + __platform_notify(__wait_addr, __all, __args._M_obj_size);
> + return;
> + }
> +#endif
> +
> + // Either a proxy wait or we don't have platform wait/wake primitives.
> +
> + auto __state = &__waitable_state::_S_state_for(__addr);
>
> // Lock mutex so that proxied waiters cannot race with incrementing
> _M_ver
> // and see the old value, then sleep after the increment and
> notify_all().
> lock_guard __l{ *__state };
>
> - if (__args & __wait_flags::__proxy_wait)
> + if (proxy_wait)
> {
> - // Waiting for *__addr is actually done on the proxy's _M_ver.
> - __wait_addr = &__state->_M_ver;
> -
> // Increment _M_ver so that waiting threads see something changed.
> // This has to be atomic because the load in _M_load_proxy_wait_val
> // is done without the mutex locked.
> @@ -322,18 +378,18 @@ __notify_impl(const void* __addr, [[maybe_unused]]
> bool __all,
> // they can re-evaluate their conditions to see if they should
> // stop waiting or should wait again.
> __all = true;
> - }
> - else // Use the atomic variable's own address.
> - __wait_addr = static_cast<const __platform_wait_t*>(__addr);
>
> - if (__args & __wait_flags::__track_contention)
> + __wait_addr = &__state->_M_ver;
> + }
> +
> + if (__track_contention)
> {
> if (!__state->_M_waiting())
> return;
> }
>
> #ifdef _GLIBCXX_HAVE_PLATFORM_WAIT
> - __platform_notify(__wait_addr, __all);
> + __platform_notify(__wait_addr, __all, sizeof(__state->_M_ver));
> #else
> __state->_M_cv.notify_all();
> #endif
> @@ -343,30 +399,7 @@ __notify_impl(const void* __addr, [[maybe_unused]]
> bool __all,
>
> namespace
> {
> -#ifdef _GLIBCXX_HAVE_LINUX_FUTEX
> -// returns true if wait ended before timeout
> -bool
> -__platform_wait_until(const __platform_wait_t* __addr,
> - __platform_wait_t __old,
> - const __wait_clock_t::time_point& __atime) noexcept
> -{
> - struct timespec __rt = chrono::__to_timeout_timespec(__atime);
> -
> - if (syscall (SYS_futex, __addr,
> - static_cast<int>(__futex_wait_flags::__wait_bitset_private),
> - __old, &__rt, nullptr,
> - static_cast<int>(__futex_wait_flags::__bitset_match_any)))
> - {
> - if (errno == ETIMEDOUT)
> - return false;
> - if (errno != EINTR && errno != EAGAIN)
> - __throw_system_error(errno);
> - }
> - return true;
> -}
> -#endif // HAVE_LINUX_FUTEX
> -
> -#ifndef _GLIBCXX_HAVE_PLATFORM_TIMED_WAIT
> +#ifndef _GLIBCXX_HAVE_PLATFORM_WAIT
> bool
> __cond_wait_until(__condvar& __cv, mutex& __mx,
> const __wait_clock_t::time_point& __atime)
> @@ -381,7 +414,7 @@ __cond_wait_until(__condvar& __cv, mutex& __mx,
> __cv.wait_until(__mx, __ts);
> return __wait_clock_t::now() < __atime;
> }
> -#endif // ! HAVE_PLATFORM_TIMED_WAIT
> +#endif // ! HAVE_PLATFORM_WAIT
>
> // Unlike __spin_impl, does not always return _M_has_val == true.
> // If the deadline has already passed then no fresh value is loaded.
> @@ -414,7 +447,7 @@ __spin_until_impl(const __platform_wait_t* __addr,
> return __res;
> }
>
> - __atomic_load(__addr, &__res._M_val, __args._M_order);
> + __res._M_val = __atomic_load_n(__addr, __args._M_order);
> __res._M_has_val = true;
> if (__res._M_val != __args._M_old)
> {
> @@ -428,16 +461,11 @@ __spin_until_impl(const __platform_wait_t* __addr,
> } // namespace
>
> __wait_result_type
> -__wait_until_impl(const void* __addr, __wait_args_base& __args,
> +__wait_until_impl([[maybe_unused]] const void* __addr, __wait_args_base&
> __args,
> const __wait_clock_t::duration& __time)
> {
> const __wait_clock_t::time_point __atime(__time);
> - auto __state = static_cast<__waitable_state*>(__args._M_wait_state);
> - const __platform_wait_t* __wait_addr;
> - if (__args & __wait_flags::__proxy_wait)
> - __wait_addr = &__state->_M_ver;
> - else
> - __wait_addr = static_cast<const __platform_wait_t*>(__addr);
> + auto* __wait_addr = static_cast<const
> __platform_wait_t*>(__args._M_obj);
>
> if (__args & __wait_flags::__do_spin)
> {
> @@ -448,11 +476,12 @@ __wait_until_impl(const void* __addr,
> __wait_args_base& __args,
> return __res;
> }
>
> -#ifdef _GLIBCXX_HAVE_PLATFORM_TIMED_WAIT
> +#ifdef _GLIBCXX_HAVE_PLATFORM_WAIT
> if (__args & __wait_flags::__track_contention)
> - set_wait_state(__addr, __args);
> + set_wait_state(__addr, __args); // scoped_wait needs a
> __waitable_state
> scoped_wait s(__args);
> - bool timeout = !__platform_wait_until(__wait_addr, __args._M_old,
> __atime);
> + bool timeout = !__platform_wait_until(__wait_addr, __args._M_old,
> __atime,
> + __args._M_obj_size);
> return { ._M_val = __args._M_old, ._M_has_val = false, ._M_timeout =
> timeout };
> #else
> waiter_lock l(__args);
> @@ -460,6 +489,7 @@ __wait_until_impl(const void* __addr,
> __wait_args_base& __args,
> __atomic_load(__wait_addr, &__val, __args._M_order);
> if (__val == __args._M_old)
> {
> + auto __state = static_cast<__waitable_state*>(__args._M_wait_state);
> bool timeout = !__cond_wait_until(__state->_M_cv, __state->_M_mtx,
> __atime);
> return { ._M_val = __val, ._M_has_val = false, ._M_timeout =
> timeout };
> }
> --
> 2.51.1
>
>
