Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Mon, 16 Jul 2018 12:18:24 +0200
From:   Johan Hovold <johan@kernel.org>
To:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc:     Johan Hovold <johan@kernel.org>, Jiri Slaby <jslaby@suse.com>,
        Alan Cox <gnomes@lxorguk.ukuu.org.uk>,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH 3/3] tty: support CIBAUD without BOTHER
Message-ID: <20180716101824.GK10204@localhost>
References: <20180715133935.24964-1-johan@kernel.org>
 <20180715133935.24964-4-johan@kernel.org>
 <20180716100028.GA20101@kroah.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20180716100028.GA20101@kroah.com>
User-Agent: Mutt/1.10.0 (2018-05-17)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Jul 16, 2018 at 12:00:28PM +0200, Greg Kroah-Hartman wrote:
> On Sun, Jul 15, 2018 at 03:39:35PM +0200, Johan Hovold wrote:
> > Since commit edc6afc54968 ("[PATCH] tty: switch to ktermios and new
> > framework") arbitrary baud rates can be requested using BOTHER and input
> > rates can be requested using the termios CIBAUD bits (CBAUD shifted
> > IBSHIFT bits).
> > 
> > This functionality has been conditionally compiled depending on whether
> > an architecture defines BOTHER and IBSHIFT respectively, but would in
> > fact fail to compile unless both symbols were defined due to cross
> > dependencies.
> > 
> > Relax the IBSHIFT => BOTHER dependency so that an architecture could
> > theoretically support CIBAUD without the Linux-specific BOTHER, while
> > hopefully making the current conditional-compilation directives a bit
> > less confusing.
> > 
> > Note that the long-term goal is still to have all architectures support
> > both features, so an alternative could just be to have the lot depend on
> > BOTHER.
> 
> I thought we had all arches converted to use BOTHER already, what ones
> are not yet done?  It's hard to unwind the asm-generic use of termbits.h
> to obviously see which ones are not doing this yet, any ideas?

It looks like alpha does not yet define BOTHER at least.

> Oh, and thanks for fixing this all up, odd that no one has noticed it
> before.

Probably due to there being no in-tree drivers that support separate
input rates. And with no glibc support for BOTHER (still), it's somewhat
less likely that people will trigger the bug that could end up setting
CIBAUD for them.

Thanks,
Johan