Re: socket/fdopen/exec problem

2002-01-28 Thread Christopher Faylor
On Sat, Jan 26, 2002 at 11:51:38PM -0500, Pierre A. Humblet wrote: >I have no idea where to start debugging that. I am attaching the demo program. >It is used as shown in the previous message, and also using ps -W. Attach gdb to the process. cgf -- Unsubscribe info: http://cygwin.com/ml/#

Re: socket/fdopen/exec problem

2002-01-28 Thread Pierre A. Humblet
Corinna Vinschen wrote: > > > > > Any thoughts? > > Not immediately. Two questions: > > - Did you encounter the same on NT? Corinna, I have now duplicated on NT the tests reported in http://cygwin.com/ml/cygwin/2002-01/msg01642.html (saturday evening) On NT as on Win98/Me, ps -W shows one

Re: socket/fdopen/exec problem

2002-01-26 Thread Pierre A. Humblet
At 05:10 PM 1/25/02 +0100, Corinna Vinschen wrote: >On Fri, Jan 25, 2002 at 10:39:30AM -0500, Pierre A. Humblet wrote: >> Any thoughts? > >Not immediately. Two questions: > >- Could you patch Cygwin fhandler_socket.cc, fhandler_socket::close() > and rip out the whole linger stuff so that onl

Re: socket/fdopen/exec problem

2002-01-25 Thread Pierre A. Humblet
Corinna Vinschen wrote: > > > Any thoughts? > > Not immediately. Two questions: > > - Did you encounter the same on NT? The demo output I gave you, with the temporary non-acceptance, was on NT. The "never accepting" condition of exim is on Win98 (where most of my testing takes place) and was n

Re: socket/fdopen/exec problem

2002-01-25 Thread Corinna Vinschen
On Fri, Jan 25, 2002 at 10:39:30AM -0500, Pierre A. Humblet wrote: > Consider the following > > /> netstat -a | fgrep 999<=== daemon is running > TCPPHumblet:999 0.0.0.0:0 LISTENING > /> kill -HUP 293 < re-exec > /> netstat -a | fgrep 999<

Re: socket/fdopen/exec problem

2002-01-25 Thread Pierre A. Humblet
Corinna Vinschen wrote: > > Did you set the SO_REUSEADDR flag? > Yes. The background of this is that I am trying to port an application (exim, a mail transfer agent). I see it does not answer incoming calls (Win98) after a re-exec following kill -HUP. I am trying to reproduce the behavior in a

Re: socket/fdopen/exec problem

2002-01-25 Thread Corinna Vinschen
On Thu, Jan 24, 2002 at 04:47:36PM -0500, Pierre A. Humblet wrote: > Corinna Vinschen wrote: > > > Isn't that just the usual effect that sockets remain in TIME_WAIT > > state for 2MSL seconds? > > I realize I am mixing two things. Seeing reused ports in netstat > isn't much of a problem in itsel

Re: socket/fdopen/exec problem

2002-01-24 Thread Pierre A. Humblet
Corinna Vinschen wrote: > Isn't that just the usual effect that sockets remain in TIME_WAIT > state for 2MSL seconds? I realize I am mixing two things. Seeing reused ports in netstat isn't much of a problem in itself, as long as new calls can come in. That's the OK case I see now all the time on

Re: socket/fdopen/exec problem

2002-01-24 Thread Corinna Vinschen
On Thu, Jan 24, 2002 at 01:27:24PM -0500, Pierre A. Humblet wrote: > The attached file demonstrates a problem with a > combination of socket, fdopen and exec. It was observed > on Win98/ME/NT4/2000. Latest everything. > Please cc: me directly. > > A server is listening on port 999 and forking wo

socket/fdopen/exec problem

2002-01-24 Thread Pierre A. Humblet
The attached file demonstrates a problem with a combination of socket, fdopen and exec. It was observed on Win98/ME/NT4/2000. Latest everything. Please cc: me directly. A server is listening on port 999 and forking worker processes. It can be restarted by kill -HUP. Under some conditions, new