fork dead lock

[jean-christophe81]

Hello,
I'm using boost::process::v2 on Windows and linux (alma and debian) in a multithread process.
When the child process calls notify_fork and tries to lock registered_descriptors_mutex_ (epoll_reactor.ipp:130), registered_descriptors_mutex_ is already locked by another thread of parent process.

In my mind, I think that when parent process forks, another thread has locked registered_descriptors_mutex_ and child process has started with registered_descriptors_mutex_ already locked.

Have you a workaround?

Thanks

Jean Christophe