Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT cygwin DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-help AT cygwin DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-owner AT cygwin DOT com
Delivered-To: mailing list cygwin AT cygwin DOT com
Message-ID: <3C587F40.FB1EED18@lapo.it>
Date: Thu, 31 Jan 2002 00:18:24 +0100
From: Lapo Luchini <lapo AT lapo DOT it>
X-Mailer: Mozilla 4.76 [en] (Windows NT 5.0; U)
X-Accept-Language: it,en,fr,es,ja
MIME-Version: 1.0
To: chuckm AT rochester DOT rr DOT com
CC: cygwin AT cygwin DOT com
Subject: Re: Cygwin port of rsync -- patch for Win2k (rsync 2.5.1)
References: <3C581EC5 DOT 20100 AT rochester DOT rr DOT com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

I don't think I have enough knowledge on the "fork() in cygwin" subject to
evaluate if there are better solutions, I reply also in the mailing list.

Lapo

BTW: I'll try the fix soon, just wanted some more "advice from the gurus"
about the issue ;)



Chuck Messenger wrote:

> Hi Lapo,
>
> I've been trying out the Cygwin port of rsync -- version 2.5.1.  I've
> found that it doesn't work in server (i.e. --daemon) mode on my Win2k
> box, but it does work on my Win98 box.
>
> I did some debugging, and came up with a fix.
>
> In socket.c, we have the following code which the server executes after
> being connected to by a client:
>
>          if (fork()==0) {
>              close(s);
>              /* open log file in child before possibly giving
>                 up privileges  */
>              log_open();
>              _exit(fn(fd));
>          }
>
>   >>> delay(1);    /* CHM addition */
>
>          close(fd);
>
> I've marked my addition with >>>.  What seems to happen on Win2k is that
> the close(fd) gets called by the mother process before the child process
> gets going.  By adding a 1 second delay after the fork in the mother
> process, it seems to give enough time for the child to replicate the
> socket, so the mother can safely close(fd) it.
>
> The code for delay is:
>
>      static void  /* CHM addition */
>      delay(
>          int secs)
>      {
>          struct timeval timeout;
>
>          timeout.tv_sec = secs;
>          timeout.tv_usec = 0;
>
>          select(0, NULL, NULL, NULL, &timeout);
>      }
>
> I've been running rsync with this mod, and it seems to work fine.
>
> I know it's ugly -- the real solution would be to fix fork() -- but at
> least it works...
>
>      - Chuck

--
Lapo 'Raist' Luchini
lapo AT lapo DOT it (PGP & X.509 keys available)
http://www.lapo.it (ICQ UIN: 529796)



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/