accept doesn't block while running as NT service

[linamat@moldtelecom.md]

SYSTEM can use network resources without doubts, because UNIX domain
socket works under NT service, inetd also works as NT service.

Turning to accept() problem errno was set to EBADF.
I parsed fhandler_socket.cc's method accept(); there are 2 places where
EBADF might be set: 1st after sock_event handling if blocking (but I use
FASYNC | O_NONBLOCK), 2nd after ancestor's accept() which checks for
invalid handle.
It seems that socket descriptor is suddenly closed or invalidated under NT
service. I've even tried to run some very simple TCP socket servers
through SCM, but it fail. Don't forget that all OK with it while it runs
from console as application.
Is such behaveour normal?



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/