Re: [REGRESSION/PATCH] acpi: blacklist win8 OSI for ASUS Zenbok Prime UX31A

[Matthew Garrett <mjg59@srcf.ucam.org>]

On Wed, Jul 31, 2013 at 12:46:04PM -0500, Felipe Contreras wrote:

> That doesn't change the fact that you were wrong, and there *is*
> actually a way. The fact that you don't want to go there doesn't mean
> it's not there.

A quirk list will be incomplete, and as such there's no way to guarantee 
whether or not a value of 0 will turn off the backlight. This is why the 
interface doesn't make that guarantee, and why any userspace that 
depends upon that behaviour is behaving incorrectly.

> Here's another: device tree.

There's no functional distinction between device tree and a quirk list 
on x86 - they're both static data sources provided by something other 
than the system firmware. As a result, they will both be incomplete.

> There are ways to provide a consistent backlight interface to user-space.

No, there aren't. What you *can* do is propose to change the ABI 
description for the sysfs backlight interface such that any system where 
0 turns off the backlight is considered buggy, but that won't make those 
systems vanish. A good interface doesn't promise things it can't 
guarantee, so I'd expect that any such proposal would be rejected.

-- 
Matthew Garrett | mjg59@srcf.ucam.org