This, especially noreentry_this_object is too convoluted to be a core language feature.

Such a keyword would have to be generic, covering both single- and multithreaded cases, and would require some state, that you can easily implement yourself (just like you did, with the mutex). 

void no_reentry_fn() {
    static std::mutex prevent;
    if (!prevent.try_lock()) return;
    std::scoped_lock prevent_l(prevent);
... 

On Thu, Feb 2, 2023, 12:59 Frederick Virchanza Gotham via Std-Proposals <std-proposals@lists.isocpp.org> wrote:
This week I had to fix a bug in a desktop GUI program. After doing
some testing, I was leaning toward the possibility that one of the
functions in the program was being re-entered (when it shouldn't be).
For a quick solution to test my theory, I protected the function with
a mutex as follows:

      int Func(void)
      {
          static mutex mtx;
          lock_guard<mutex> mylock;

          // The rest of the function goes here
      }

After making this change, the program locked up completely. I was a
little puzzled so I changed the 'mutex' to a 'recursive_mutex', and
then it no longer locked up, but the original bug was still presenting
itself.

So I figured that this function is being re-entered by the same
thread. This was because the function was an event handler for the
GUI, and the GUI sometimes created nested event loops, which meant
that an event handler could be re-entered by the same thread (for
example if you displayed a modal dialog box inside a event handler).

So I needed to protect this function from re-entry, not just by
multiple threads but also by the same thread. It would be nice if
there were a core-language feature for this, for example:

      int Func(void) noreentry(-1)
      {
         // The rest of the function goes here
      }

The idea here is that the function would immediately return -1 if you
tried to re-enter it. As 'Func' is a normal function (i.e. not a
member function), the prevention of reentry would apply to the entire
program.

If it were a member function, for example:

      struct MyClass {
           int MemberFunc(void) noreentry_this_object(-1)
            {
               // The rest of the function goes here
            }
      };

then the prevention of reentry only applies to the current object (not
to all objects). Alternatively, if you want to ensure prevention of
reentry of a member function for all objects, the syntax is:

      struct MyClass {
           int MemberFunc(void) noreentry_all_objects(-1)
            {
               // The rest of the function goes here
            }
      };

So you can only apply "noreentry" to a normal function, and you can
only apply "noreentry_this_object" and "noreentry_all_objects" to a
member function. Note that you cannot apply 'noreentry' to a member
function because I don't want the confusion.

Behind the scenes, here is how each of them would work. Firstly I've
made a 'lock_guard' class to be used with 'atomic_flag':

      #include <atomic>

      class atomic_flag_guard {
         std::atomic_flag &flag;
         bool const previous;
      public:
         atomic_flag_guard(std::atomic_flag &arg) noexcept
           : flag(arg), previous(flag.test_and_set()) {}

         ~atomic_flag_guard(void) noexcept
         {
             if ( false == previous ) flag.clear();
         }

         bool failed(void) const noexcept { return previous; }
      };

Here's the implementation of the normal function:

      int Func(void)
      {
          static std::atomic_flag flag;
          atomic_flag_guard myguard(flag);
          if ( myguard.failed() ) return -1;
          // The rest of the function goes here
      }

Here's the implementation of the member function:

      struct MyClass {
            std::atomic_flag flag;
            int MemberFunc(void)
            {
                atomic_flag_guard myguard(flag);
                if ( myguard.failed() ) return -1;
               // The rest of the function goes here
            }
      };

And here's the implementation of the member function whereby reentry
is prevented for all objects:

      struct MyClass {
            int MemberFunc(void)
            {
                static std::atomic_flag flag;
                atomic_flag_guard myguard(flag);
                if ( myguard.failed() ) return -1;
               // The rest of the function goes here
            }
      };

I had considered a few other little things, for example let's say you
want the function to throw an exception rather than return -1. The
syntax would be something like:

      int Func(void) noreentry( throw runtime_error("cannot re-enter") )
      {
           // The rest of the function goes here
      }

This new feature could be kept simple, or tt could made be very
complex with extra parameters to do stuff like:
* Prevent reentry by other threads but not by the same thread
* Prevent reentry by the same thread but not by other threads
* Prevent reentry by all threads except the one with the specified thread::id

Mostly this feature would be used in desktop GUI programming, although
someone might find another use for it. It would even be useful purely
as a debugging tool.
--
Std-Proposals mailing list
Std-Proposals@lists.isocpp.org
https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals