>Number:         6860
>Category:       lib
>Synopsis:       Kerberos, like X11, hosed without sgttyb (COMPAT_43)
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    lib-bug-people (Library Bug People)
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Jan 21 13:50:01 1999
>Last-Modified:
>Originator:     Charlie Root
>Organization:
	
>Release:        19981026
>Environment:
System: NetBSD edgerton.ops.cp.net 1.3H NetBSD 1.3H (PUSTULE) #4: Sun Jan 17 03:44:25 PST 1999 eric@edgerton.ops.cp.net:/usr/src/sys/arch/i386/compile/PUSTULE i386


>Description:
	kinit doesn't work, because krb_get_pw_in_tkt doesn't work, because
	des_read_pw doesn't work, because read_pw doesn't work, because it
	tries to do a TIOCGETP, which doesn't work without COMPAT_43 in the
	kernel.
>How-To-Repeat:
	try to use our /usr/bin/kinit on a kernel without COMPAT_43 in it
>Fix:
	workaround: use COMPAT_43.  fix: eliminate dependency on sgtty;
	there's already code to not use it.

	This is a more general problem -- we ship source which is configured
	to use sgtty when it already knows how to use termios.  A totally
	absurd example of this is that our own X distribution fails in
	various ways (e.g. in xterm) if COMPAT_43 and the sgtty interface
	aren't in the kernel -- a total lose.

	Maybe should find a way to remove public evidence of the sgtty
	interface so configurator scripts don't try to use it, while
	leaving the compat code in the kernel so old binaries work if
	COMPAT_43 is on.  With some header-file magic, we could probably
	arrange for sgtty ioctls to run through a wrapper function that
	spat out a linker warning like gets() does, which we _really_
	should do.

>Audit-Trail:
>Unformatted: