Fwiw, I needed to get back into Relacy. Wrt Bonita, here is a Relacy
test unit for an old acquaintance of mine, Alex from over on
comp.programming.threads many years ago. A test of one of his fast
mutexes for windows, the code is used in:

Take notice that Relacy can model windows sync primitives, and C++...

rl::rl_HANDLE, rl_CreateEvent, ect, and such... ;^)

https://github.com/dvyukov/relacy

Just a quick test of Alex's original code:
______________________________________
#include <iostream>
#include <relacy/relacy.hpp>

#define CT_THREAD_N 3

// Testing Alex's mutex work from over on
// comp.programming.threads wrt windows...
// its on https://sourceware.org/pthreads-win32/
// anyway....

struct ct_alex_mutex_test_windows
{ rl::rl_HANDLE m_waitset;
rl::atomic<unsigned long> m_state;

ct_alex_mutex_test_windows()
: m_waitset(rl_CreateEvent(nullptr, false, false, nullptr)),
m_state(0)
{
if (! m_waitset) throw;
}

~ct_alex_mutex_test_windows()
{
CloseHandle(m_waitset);
}

void lock()
{
if (m_state.exchange(1, rl::mo_relaxed, $))
{
while (m_state.exchange(2, rl::mo_relaxed, $))
{
rl_WaitForSingleObject(m_waitset, INFINITE, $);
}
}

rl::atomic_thread_fence(rl::mo_acquire, $);
}

void unlock()
{
rl::atomic_thread_fence(rl::mo_release, $);

if (m_state.exchange(0, rl::mo_relaxed, $) == 2)
{
rl_SetEvent(m_waitset, $);
}
}
};

struct ct_relacy_test_fun : rl::test_suite<ct_relacy_test_fun, CT_THREAD_N>
{ VAR_T(unsigned long) m_test_0;
ct_alex_mutex_test_windows m_mutex;

void before()
{
VAR(m_test_0) = 0;

//std::cout << "ct_relacy_test_fun::before()" << std::endl;
}

void after()
{
// std::cout << "ct_relacy_test_fun::after()" << std::endl;

RL_ASSERT(VAR(m_test_0) == CT_THREAD_N);
}

void thread(unsigned int tidx)
{
m_mutex.lock();
VAR(m_test_0) += 1;
m_mutex.unlock();
}
};

int
main()
{ std::cout << "Relacy Testing 123... ;^)" << std::endl;

{
rl::test_params p;

//p.iteration_count = 10000000;
//p.execution_depth_limit = 33333;
//p.search_type = rl::sched_bound;
p.search_type = rl::fair_full_search_scheduler_type;
//p.search_type = rl::fair_context_bound_scheduler_type;

rl::simulate<ct_relacy_test_fun>(p);
}

return 0;
} ______________________________________

My output:
______________________________________
Relacy Testing 123... ;^)
struct ct_relacy_test_fun
5% (65536/1129586)
iterations: 90732
total time: 1609
throughput: 56390
______________________________________

Nice! Working like a charm. Now I can get into playing around with my
new'ish queue algorithm. I basically need to distribute it.

On 11/9/2023 12:12 AM, Chris M. Thomasson wrote:
> #include <iostream>
> #include <relacy/relacy.hpp>
>
>
> #define CT_THREAD_N 3
>
>
> // Testing Alex's mutex work from over on
> // comp.programming.threads wrt windows...
> // its on https://sourceware.org/pthreads-win32/
> // anyway....
>
> struct ct_alex_mutex_test_windows
> {
> [...]
>     void lock()
>     {
>         if (m_state.exchange(1, rl::mo_relaxed, $))
>         {
>             while (m_state.exchange(2, rl::mo_relaxed, $))
>             {
>                 rl_WaitForSingleObject(m_waitset, INFINITE, $);
>             }
>         }
>

>         rl::atomic_thread_fence(rl::mo_acquire, $);
//^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>     }
[...]

Relacy kicks some ass. If I comment out that acquire thread fence:

rl::atomic_thread_fence(rl::mo_acquire, $);

Relacy knows things are really wrong here, and gives me the following
output:

Nice!
_______________________________
Relacy Testing 123... ;^)
struct ct_relacy_test_fun
DATA RACE (data race detected)
iteration: 1

execution history (16):
[0] 0: [CTOR BEGIN], in rl::context_impl<struct ct_relacy_test_fun,class
rl::full_search_scheduler<3> >::fiber_proc_impl, context.hpp(457)
[1] 0: memory allocation: addr=0000022E2E825B60, size=24, in
ct_alex_mutex_test_windows::ct_alex_mutex_test_windows, ct_main.cpp(19)
[2] 0: <0000022E2E81C890> atomic store, value=0, (prev value=0),
order=relaxed, in rl::atomic<unsigned long>::atomic, atomic.hpp(594)
[3] 0: [CTOR END], in rl::context_impl<struct ct_relacy_test_fun,class
rl::full_search_scheduler<3> >::fiber_proc_impl, context.hpp(457)
[4] 0: [BEFORE BEGIN], in rl::context_impl<struct
ct_relacy_test_fun,class rl::full_search_scheduler<3>
>::fiber_proc_impl, context.hpp(457)
[5] 0: <0000022E2E81C870> store, value=0, in ct_relacy_test_fun::before,
ct_main.cpp(63)
[6] 0: [BEFORE END], in rl::context_impl<struct ct_relacy_test_fun,class
rl::full_search_scheduler<3> >::fiber_proc_impl, context.hpp(457)
[7] 0: <0000022E2E81C890> exchange , prev=0, arg=1, new=1,
order=relaxed, in ct_alex_mutex_test_windows::lock, ct_main.cpp(32)
[8] 0: <0000022E2E81C870> load, value=0, in ct_relacy_test_fun::thread,
ct_main.cpp(78)
[9] 0: <0000022E2E81C870> store, value=1, in ct_relacy_test_fun::thread,
ct_main.cpp(78)
[10] 0: release fence, in ct_alex_mutex_test_windows::unlock,
ct_main.cpp(45)
[11] 0: <0000022E2E81C890> exchange , prev=1, arg=0, new=0,
order=relaxed, in ct_alex_mutex_test_windows::unlock, ct_main.cpp(47)
[12] 0: [THREAD FINISHED], in rl::context_impl<struct
ct_relacy_test_fun,class rl::full_search_scheduler<3>
>::fiber_proc_impl, context.hpp(457)
[13] 1: <0000022E2E81C890> exchange , prev=0, arg=1, new=1,
order=relaxed, in ct_alex_mutex_test_windows::lock, ct_main.cpp(32)
[14] 1: <0000022E2E81C870> load, value=0, in ct_relacy_test_fun::thread,
ct_main.cpp(78)
[15] 1: DATA RACE (data race detected), in ct_relacy_test_fun::thread,
ct_main.cpp(78)

thread 0:
[0] 0: [CTOR BEGIN], in rl::context_impl<struct ct_relacy_test_fun,class
rl::full_search_scheduler<3> >::fiber_proc_impl, context.hpp(457)
[1] 0: memory allocation: addr=0000022E2E825B60, size=24, in
ct_alex_mutex_test_windows::ct_alex_mutex_test_windows, ct_main.cpp(19)
[2] 0: <0000022E2E81C890> atomic store, value=0, (prev value=0),
order=relaxed, in rl::atomic<unsigned long>::atomic, atomic.hpp(594)
[3] 0: [CTOR END], in rl::context_impl<struct ct_relacy_test_fun,class
rl::full_search_scheduler<3> >::fiber_proc_impl, context.hpp(457)
[4] 0: [BEFORE BEGIN], in rl::context_impl<struct
ct_relacy_test_fun,class rl::full_search_scheduler<3>
>::fiber_proc_impl, context.hpp(457)
[5] 0: <0000022E2E81C870> store, value=0, in ct_relacy_test_fun::before,
ct_main.cpp(63)
[6] 0: [BEFORE END], in rl::context_impl<struct ct_relacy_test_fun,class
rl::full_search_scheduler<3> >::fiber_proc_impl, context.hpp(457)
[7] 0: <0000022E2E81C890> exchange , prev=0, arg=1, new=1,
order=relaxed, in ct_alex_mutex_test_windows::lock, ct_main.cpp(32)
[8] 0: <0000022E2E81C870> load, value=0, in ct_relacy_test_fun::thread,
ct_main.cpp(78)
[9] 0: <0000022E2E81C870> store, value=1, in ct_relacy_test_fun::thread,
ct_main.cpp(78)
[10] 0: release fence, in ct_alex_mutex_test_windows::unlock,
ct_main.cpp(45)
[11] 0: <0000022E2E81C890> exchange , prev=1, arg=0, new=0,
order=relaxed, in ct_alex_mutex_test_windows::unlock, ct_main.cpp(47)
[12] 0: [THREAD FINISHED], in rl::context_impl<struct
ct_relacy_test_fun,class rl::full_search_scheduler<3>
>::fiber_proc_impl, context.hpp(457)

thread 1:
[13] 1: <0000022E2E81C890> exchange , prev=0, arg=1, new=1,
order=relaxed, in ct_alex_mutex_test_windows::lock, ct_main.cpp(32)
[14] 1: <0000022E2E81C870> load, value=0, in ct_relacy_test_fun::thread,
ct_main.cpp(78)
[15] 1: DATA RACE (data race detected), in ct_relacy_test_fun::thread,
ct_main.cpp(78)

thread 2:
_______________________________

Pretty good!

:^D

On 11/9/2023 12:12 AM, Chris M. Thomasson wrote:
> Fwiw, I needed to get back into Relacy. Wrt Bonita, here is a Relacy
> test unit for an old acquaintance of mine, Alex from over on
> comp.programming.threads many years ago. A test of one of his fast
> mutexes for windows, the code is used in:
[...]

https://sourceware.org/pthreads-win32/

On 11/9/2023 12:12 AM, Chris M. Thomasson wrote:
> Fwiw, I needed to get back into Relacy. Wrt Bonita, here is a Relacy
> test unit for an old acquaintance of mine, Alex from over on
> comp.programming.threads many years ago. A test of one of his fast
> mutexes for windows, the code is used in:
>
> Take notice that Relacy can model windows sync primitives, and C++...
>
> rl::rl_HANDLE, rl_CreateEvent, ect, and such... ;^)
>
>
> https://github.com/dvyukov/relacy
>
> Just a quick test of Alex's original code:
[...]

Some music that reminds of Relacy processing a long test:

https://youtu.be/QQ77ejz6i4M

On 11/9/2023 12:12 AM, Chris M. Thomasson wrote:
> Fwiw, I needed to get back into Relacy. Wrt Bonita, here is a Relacy
> test unit for an old acquaintance of mine, Alex from over on
> comp.programming.threads many years ago. A test of one of his fast
> mutexes for windows, the code is used in:
>
> Take notice that Relacy can model windows sync primitives, and C++...
>
> rl::rl_HANDLE, rl_CreateEvent, ect, and such... ;^)
>
>
> https://github.com/dvyukov/relacy
[...]

Fwiw, you do not need to have windows installed to play around with
Relacy. It simulates windows primitives, and even PThread primitives.

On 11/9/2023 12:12 AM, Chris M. Thomasson wrote:
[...]

Almost ready to start work on my new queue. I have my plotter ready, and
it does create volumetric images of some of my experimental fractals. A
low res example at 256^3 volume:

https://i.ibb.co/8XfnmrR/image.png