Re: [PATCH v4 1/2] Input: enable i8042-level wakeup control

On Wed, Aug 3, 2011 at 8:38 PM, Dmitry Torokhov
[off-list ref] wrote:

I believe we can and do mark devices such as USB as wakeup capable on
other arches, you do not have control here. That is why I am uneasy with
doing this in input core.

That's possible, but do note that the latest iteration of the patch
only looks at the wakeup capability of the struct input_device. It is
unlikely that another part of the kernel marks it as wakeup-capable
without taking these considerations into account.

It is more likely that, in the USB case, the wakeup field of the
struct usb_device is marked wakeup capable. With Alan's ACPI case
earlier in the discussion, this was true: we found that only the ACPI
device gets marked as wakeup-capable/wakeup-enabled, and not the input
device or even the i8042/serio devices, even though it is implementing
(some form of) keyboard wakeup.

Anyway, I'll look at implementing it according to how you suggested,
which wouldn't involve such wide-reaching changes.

Is there any keys that are not autorepeating. For example regular
(non-OLPC) laptops usually do not repeat suspend and other special keys.
In fact, they quite often forget to send release events for them ;)

I just tested, and all of our keys autorepeat in that fashion. Even
the odd keys like "rotate" and "change language". I guess that makes
us irregular ;)

Daniel