Hi Leo,

> But wouldn't EINVAL be the most logical return value?

I' m not sure. I'd probably argue that TIOCFLAG_MDMBUF isn't
really an invalid argument or invalid combination of arguments
(like the softcar+clocal case). But then, I just looked in
src/lib/libc/gen/errlist.c and figured that "Operation not
supported by device" might be most appropriate from a users
point of view. I didn't consider the tech-kern aspects.

> (Trying to defend guru status ;-) )

I take this as a hint. ;-)

> > --- zs.c.orig	Thu Nov 16 13:31:39 1995
> > +++ zs.c	Tue Nov 28 10:41:42 1995
> > @@ -822,11 +822,11 @@
> >  		 # defaulting to software flow control.
> >  		 */
> >  		if(userbits & TIOCFLAG_SOFTCAR && userbits & TIOCFLAG_CLOCAL)
> >  			return(EINVAL);
> >  		if(userbits & TIOCFLAG_MDMBUF)	/* don't support this (yet?) */
> > -			return(ENXIO);
> > +			return(ENODEV);
> >  
> >  		s = splzs();
> >  		if((userbits & TIOCFLAG_SOFTCAR)) {
> >  			cs->cs_softcar = 1;	/* turn on softcar */
> >  			cs->cs_preg[15] &= ~ZSWR15_DCD_IE; /* turn off dcd */

Regards,
Waldi


PS Michael, is the `disklabel -e' => "partition extends ..." stuff solved?