Re: ukbd_cnattach() called unconditionally

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Sat, Sep 22, 2018 at 09:04:41AM -0300, Jared McNeill wrote:
> 
> 
> On Thu, 20 Sep 2018, Manuel Bouyer wrote:
> 
> > Hello
> > I just noticed a console issue on my SUNXI board.
> > 
> > First I had to disable WSDISPLAY_MULTICONS to have a usable serial console.
> > Otherwise I couldn't enter ddb.
> 
> I think ddb should still work, but cnmagic is different after wsdisplay
> takes over. Can you try playing with that?

The problem was not entering ddb, I think it somewhat tried to enter
(the screen did become black, so it tried to switch back to text) but
it looks like it did hang (nothing in serial console or no text on screen).
After disabling WSDISPLAY_MULTICONS, I got some ddb output on the serial port
before the hang (ultimately ddb was still not usable, but disabling
WSDISPLAY_MULTICONS got me some more diag output).

> 
> > Second, when tty0 is the (only) console, ukbd_cnattach() is still called, which
> > cause ukbd to think it's the console keyboard. The result is that the
> > keyboard is not attached to the framebuffer and it's unusable.
> 
> USB keyboard console works for me on Pinebook..

the problem is when the USB keyboard should *not* be the console
(console=serial and WSDISPLAY_MULTICONS disabled).

> 
> > shouldn't ukbd_cnattach() be called from simplefb_console_consinit()
> > instead ?
> 
> That will only ever be called if console=fb is present in bootargs, and none
> of our boot scripts do this.

it's not called in the WSDISPLAY_MULTICONS case ?

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--