Re: [PATCH 1/3] TTY: add support for "tty slave" devices.

[Pavel Machek <pavel@ucw.cz>]

On Mon 2015-01-05 15:41:41, One Thousand Gnomes wrote:
> > It would be really nice if the uart would register the line disciple as a
> > child device, then the line discipline would register whatever it wants.
> 
> For almost every case this doesn't work. You need a tty interface as well
> because thats how you manage it.
> 
> > But that isn't how it works.  The line discipline doesn't talk to the uart.
> > Rather the tty layer talks to the uart (through tty_operations) and to the
> > line discipline (through tty_ldisc_ops) and also registers the char_dev.
> 
> This is intentional. An ldisc has no business knowing what it's connected
> to. Try running bluetooth over a tty/pty pair remotely to a dongle - and
> you can do it or faking a GSM mux over a tty/pty pair for testing.

Not all cases are like that. IIRC Neil's hardware uses CTS/RTS in an
interesting way. bc2048 is connected to serial+clocks+gpios...

> There are lots of cases where we know the correct ldisc from information
> in the ACPI, DT or even USB identifiers in order to plug in the right
> ldisc and daemon but I think that pretty much has to be in user space.
> The kernel might be able to set an ldisc but it can't go around starting
> bluetooth daemons, doing modem chats to go into mux mode or firing up
> JMRI and java when it sees a Sprog. That's a job for systemd/udev.

No need to start bluetooth daemons: userspace can start them
itself. Just like in btusb case.

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html