Re: [PATCH] tty: Add sysfs symlink for console name->tty device

[Peter Hurley <peter@hurleysoftware.com>]

On 02/27/2014 08:36 AM, Kay Sievers wrote:
> On Thu, Feb 27, 2014 at 2:31 PM, Peter Hurley <peter@hurleysoftware.com> wrote:
>> On 02/27/2014 06:13 AM, Kay Sievers wrote:
>>>
>>> On Wed, Feb 26, 2014 at 3:40 PM, Peter Hurley <peter@hurleysoftware.com>
>>> wrote:
>>>>
>>>> Enable a user-space process to discover the underlying tty device
>>>> for a console, if one exists, and when the tty device is later
>>>> created or destroyed.
>>>>
>>>> Add sysfs symlinks for registered consoles to their respective
>>>> devices in [sys/class,sys/devices/virtual]/tty/console.
>>>> Scan consoles at tty device (un)registration to handle deferred
>>>> console<->device (un)binding.
>>>
>>> What tool is supposed to read that? I can't think of anything
>>> interested in this, as soon as we have fixed the "active" console
>>> output.
>
>> With all due respect, that "fix" is a ridiculous hack,
>
> No, it is not. It's fine to handle tty0 special, as it is special.

I wasn't just referring to the plymouth workaround.

>> being done
>> for self-serving expedience.
>
> I don't see the problem.
>
>> It already caused one user-space breakage
>> which you did not expect.
>
> That is normal way to do things, only people who don't do things don't
> break things. And broken things get fixed, and the "active" file is still
> fixable, and that is what we should do.

But what's the plan when more user-space breakage is uncovered after that
change has been in 10 kernel releases?

Then everyone will point to how much user-space breakage reverting it
will cause.

> We don't need to invent new things because we did not get things right
> with the first try.
>
>> This sysfs interface is superior in every way.
>
> But nothing uses it now, and probably never will, so I don't see the
> need for it at this moment.

Let's just avoid the mess right up-front.

Regards,
Peter Hurley