Re: Using EV_MSC or extending KEY_*

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

On Tue, Sep 15, 2009 at 01:05:57PM -0600, Rick L. Vinyard, Jr. wrote:
> Dmitry Torokhov wrote:
> > On Tue, Sep 15, 2009 at 10:41:30PM +0530, Trilok Soni wrote:
> >> Adding linux-input.
> >>
> >> On Tue, Sep 15, 2009 at 9:49 PM, Rick L. Vinyard, Jr.
> >> <rvinyard@cs.nmsu.edu> wrote:
> >> > On the Logitech G13 there are a set of 10 special keys at the top of
> >> the
> >> > device.
> >> >
> >> > I've written the driver so that it maintains a map from the general
> >> keys
> >> > (G1-G22) to Linux' input subsystem keys.
> >> >
> >> > The map can be changed on the fly, allowing multiple keymaps to be
> >> loaded.
> >> >
> >> > As a side note I haven't provided for this functionality (multiple
> >> > keymaps) in the driver since this can be maintained from userspace,
> >> but it
> >> > does need to know about the currently loaded keymap so that when the
> >> raw
> >> > HID events are parsed the appropriate EV_KEY events can be input.
> >> >
> >> > I could set it up so that the special keys also map to a KEY_* code.
> >> > However, I'd prefer to set things up so that any G1-G22 key can be
> >> mapped
> >> > to any input subsystem KEY_* code and not reserve a set of 10 from the
> >> > current KEY_* codes for the 10 special keys.
> >> >
> >> > One way I was thinking of approaching this was to generate EV_MSC type
> >> > events (with a MSC_RAW code) using the value to indicate the state
> >> > (pressed/released) and special key.
> >
> > Doesn't HID already generate EV_MSC/MSC_SCAN for exactly this purpose
> > so userspace knows what keys can be remapped? You may initially map all
> > keys you don't know the usage to KEY_UNKNOWN.
> 
> As I understand it the EV_MSC/MSC_SCAN is used to trigger a prompting of
> the user to map a key, which would be useful for the G1-G22 keys and
> perhaps 5 of the 10 special keys. But, that's not the behavior I want for
> the M1, M2, M3, MR and backlight keys.
> 
> Somehow they need to be distinct from any existing KEY_* codes or not have
> a code at all... i.e. only emitting the raw misc events.
> 
> Perhaps the BTN_MISC codes are better suited for these keys?
> 

I am sorry, I don't follow... why would yo not want to assign a standard
code to the M* keys?

-- 
Dmitry