> Date: Thu, 10 Mar 1994 07:55:50 -0800
> From: Adam Glass <glass@sun-lamp.cs.berkeley.edu>
> 
> [ I asked "Where is cdevsw?" ]
> sys/arch/sun3/sun3/conf.c

I guess I was tired.  It was right under my nose... (sigh)

The reason my driver open routine never got called was because the
conf.c never got rebuilt, because it (and many others) do not have
complete dependency lists in the makefile.  Oh well...

Anyway, I got the zs driver working.  I'll send the changes in
separate messages to categorize them somewhat.  Note that some of the
diffs are against files in the sparc port.  The path names shown in
the diffs reflect this, so use patch -p0 (or copy the file first).

So far, the zs driver works as a normal (non-console) device, and
it has console getchar/putchar routines that work, but I have not yet
tried to use it as a console.  It's not clear how to proceed here...

The sparc port uses a somewhat complicated scheme to implement the
switchable console stuff ala SunOS (i.e. TIOCCONS), and the current
sun3 port is trying to use a scheme similar to the i386 port.  The
later scheme is somewhat cleaner, but using it will require some
changes to the kbd, event, and ms drivers to make one of them own
the tty structure (i.e. make kbd a real tty) so the switchable console
can point at it.  Otherwise, we could just go with the sparc way...

I'd appreciate some advice on how to make a "keyboard/screen" tty
that the switchable console can point at.

Then there are these cngetc/cnputc functions.  If we use the ones in
the zs driver, then obio_internal_configure will need to map in the
zs chip so it is available when consinit is called.  Also, the one
in the keyboard driver will be really complicated.  I'm beginning to
think we should just always use the prom versions of cngetc/cnputc.
What do you think?

Gordon

------------------------------------------------------------------------------