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
Mail-Followup-To: cygwin AT cygwin DOT com
Delivered-To: mailing list cygwin AT cygwin DOT com
Date: Mon, 17 Jun 2002 16:13:43 +0200
From: Pavel Tsekov <ptsekov AT syntrex DOT com>
Reply-To: Pavel Tsekov <cygwin AT cygwin DOT com>
Organization: Syntrex, Inc.
X-Priority: 3 (Normal)
Message-ID: <17824658567.20020617161343@syntrex.com>
To: cygwin AT cygwin DOT com
Subject: Re[2]: ioctl(), TIOCSWINSZ and pseudo terminal slaves
In-Reply-To: <20020617133403.B30892@cygbert.vinschen.de>
References: <8114354270 DOT 20020617132158 AT syntrex DOT com>
 <20020617133403 DOT B30892 AT cygbert DOT vinschen DOT de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

CV> On Mon, Jun 17, 2002 at 01:21:58PM +0200, Pavel Tsekov wrote:
>> This is just a heads up message :)
>> 
>> The above combination doesn't succeed (-1 is returned), but the errno
>> is not set approapriately.

CV> What's the expected errno?

My report wasn't quite right for which I apologize. errno is not set,
so its value is 0 (No error). I don't know which the correct value for
errno is in this case though :(

However I investiagted a bit more... From the source I can see that
this ioctl () function seems to be not implemented for pty slaves
when the master is opened with fhandler_pty_master::open() method -
I refer to this comment from fhander_tty.cc, line 461:

  /* output_done_event may or may not exist.  It will exist if the tty
     was opened by fhandler_tty_master::init, normally called at
     startup if use_tty is non-zero.  It will not exist if this is a
     pty opened by fhandler_pty_master::open.  In the former case, tty
     output is handled by a separate thread which controls output.  */

and some lines below this one (line 489):

  /* The ioctl events may or may not exist.  See output_done_event,
     above.  */

Now in this case (pty's) the following code from line 976

      get_ttyp ()->ioctl_retval = -1;
      get_ttyp ()->arg.winsize = * (struct winsize *) arg;
      if (ioctl_request_event) // This will be NULL
    SetEvent (ioctl_request_event);
      if (ioctl_done_event) // and this also
    WaitForSingleObject (ioctl_done_event, INFINITE);
      break;

will cause the return value of the ioctl() to be -1 but doesn't
set errno.

I see two possible solutions for this problem, though I don't have the insights to
decide which one is correct (of course maybe there are others):

1. The ioctl* events should not be handled like the read/write events
i.e. maybe they need to be created both for TTY masters and PTY
masters in tty::common_init().

2. Check in fhandler_tty_slave::ioctl() if the master is a PTY and set
ENOSYS for TIOCSWINSZ function.

I have to note that my knowledge on PTYs is not good at all and I
can't forsee other solutions without further information. I noted this
problem in my effort to port Midnight Commander to cygwin.


--
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/