Mailing-List: contact cygwin-developers-help AT cygwin DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-developers-subscribe AT cygwin DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin-developers/>
List-Post: <mailto:cygwin-developers AT cygwin DOT com>
List-Help: <mailto:cygwin-developers-help AT cygwin DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-developers-owner AT cygwin DOT com
Delivered-To: mailing list cygwin-developers AT cygwin DOT com
Date: Mon, 15 Jul 2002 13:10:19 +0200
From: Corinna Vinschen <vinschen AT redhat DOT com>
To: cygwin-developers AT cygwin DOT com
Subject: Re: How about this for passing file descriptors?
Message-ID: <20020715131019.C17700@cygbert.vinschen.de>
Reply-To: cygwin-developers AT cygwin DOT com
Mail-Followup-To: cygwin-developers AT cygwin DOT com
References: <Pine DOT LNX DOT 4 DOT 33 DOT 0207131534200 DOT 3194-100000 AT this> <20020715103127 DOT A6932 AT cygbert DOT vinschen DOT de> <006b01c22be7$15deadf0$1800a8c0 AT LAPTOP>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <006b01c22be7$15deadf0$1800a8c0@LAPTOP>
User-Agent: Mutt/1.3.22.1i

On Mon, Jul 15, 2002 at 08:05:01PM +1000, Robert Collins wrote:
> 
> ----- Original Message -----
> From: "Corinna Vinschen" <vinschen AT redhat DOT com>
> To: <cygwin-developers AT cygwin DOT com>
> Sent: Monday, July 15, 2002 6:31 PM
> 
> ...  Each of them needs some different
> > strategy so I wonder if it makes sense to create a new virtual fhandler
> > method to do the job.  Anyway, what I'm implementing is the following:
> 
> I think it does. Just off the cuff, we need a struct that has transfer
> information -
> struct transfer { void *buffer; int buflen; HANDLE *handles; int
> handlecount;}; and a couple of virtual functions for the fd class -
> 
> transfer prep_transfer () const;
> void complete_transfer (transfer const &);

Perhaps that's not even needed.  We already have a couple of methods
which are involved when duplicating handles between processes, namely
dup(), fixup_before_fork(), fixup_after_fork(), fixup_after_exec().

Basically they are only restricted in their functionality since they
only get the *parent* process handle.  If we generalize this stuff
to get a *source* and a *destination* process handle instead, the
whole stuff would be available for situations besides fork/exec...

Ah yes, I didn't take a closer look but if some of these methods
expect to be called in a fhandler inside of the fdtab, that expectation
would have to be corrected, too.

Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Developer                                mailto:cygwin AT cygwin DOT com
Red Hat, Inc.