coredump with nchan_store_init_worker: Assertion `procslot_found == 1'

[DomrachevIvan]

I have got errors:

src/nginx-1.20.2/objs/nginx -p /home/user/nginx -e /home/user/nginx/error -c /home/user/nginx/nginx.conf: /home/user/src/nchan-master/src/store/memory/memstore.c:723: nchan_store_init_worker: Assertion `procslot_found == 1' failed.
2022/02/22 13:41:50 [alert] 1192#0: worker process 1878 exited on signal 6 (core dumped)
2022/02/22 13:41:50 [alert] 1192#0: shared memory zone "memstore" was locked by 1878

versions:
nginx-1.20.2
nchan from current master 7994951

os:
ubuntu-21.10

steps:
setup virtual machine with new ubuntu. one processor is enough. user=user.
take clean sources of nginx and nchan.
configure: ./configure --add-module=/home/user/src/nchan-master
nginx config:

 worker_processes 16;
 error_log /home/user/nginx/error warn;
 pid /home/user/nginx/pid;
 events {}
 http {
  server {
   listen localhost:8080 default;
   access_log /home/user/nginx/access combined;
   location /laggy_backend {
    proxy_ignore_client_abort on;
    proxy_pass http://10.100.0.1;
   }
  }
 }

change 10.100.0.1 to any ip that will do very long answer (more than 60sec).
start nginx and run this script:

 #!/usr/bin/env perl
 use Socket;
 for(;;){
  print $n++."\n";
  system('kill -s hup `cat /home/user/nginx/pid`');
  socket(S,PF_INET,SOCK_STREAM,getprotobyname("tcp")) || die "socket: $!";
  connect(S,pack_sockaddr_in(8080,inet_aton('127.0.0.1'))) || die "connect: $!";
  send(S,"GET /laggy_backend/$n HTTP/1.1\r\nHost: nchan\r\nConnection: Close\r\n\r\n",0);
  sleep(1);
 }

after about 30-60 seconds i usually get coredump.

I think problem is in some factors:
1 nginx reload (sig hup)
2 many worker_processes (more than 4)
3 long backend replys

and some more notes:
no need any of nchan configs and when script is runing, process begin appear "worker process is shutting down".

[DomrachevIvan]

how can i help to resolve this problem?