On Apr 13 21:20, Corinna Vinschen wrote: > On Apr 13 10:43, David Rothenberger wrote: > > While configuring apr-1.4.2 with Cygwin 1.7.5, the test for accept4 > > hung. I don't know much about this API, so I've just disabled it in > > libapr1 for now, but I wanted to report the hang in case it was a > > Cygwin bug. > > Boy, this testcase is really stressing the limits. > > The problem is that connect() is called in that loop so fast, that at > one point the AF_UNIX socket file already exists, but is not yet filled > with the information necessary to be recognized as AF_UNIX socket. > Consequentially, at that point connect() fails with EBADF, since it > assumes you're trying to connect to a file which is not a socket. > > The hang is just a followup effect due to the parent waiting in accept() > for a connect() which will never happen, because the child process > already left the connect() loop. > > I try to come up with a more robust solution which also catches this > scenario.
In CVS I just changed the way the socket file gets created in bind() and the way it's tested for being a socket file in connect() so this timing problem shouldn't occur anymore. Would you mind to test either CVS or the next developers snapshot from http://cygwin.com/snapshots/ ? Thanks, Corinna -- Corinna Vinschen Please, send mails regarding Cygwin to Cygwin Project Co-Leader cygwin AT cygwin DOT com Red Hat -- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
