Re: HID Output and Feature Reports

Hello David,

On Sat, May 26, 2012 at 7:44 AM, David Herrmann
[off-list ref] wrote:

Hi Joao

I CC'ed Jiri as he is the HID maintainer.

On Fri, May 25, 2012 at 10:51 PM, Joao Paulo Rechi Vita
[off-list ref] wrote:

quoted

Hello all,

I'm working to support HID over GATT (HoG), which is HID over
Bluetooth Low Energy links, both on kernel and userspace. Traffic
coming from the remote peripheral through a L2CAP socket will be
handled in usespace, where the HID input reports are decapsulated from
GATT and injected back into the kernel through the new uHID module
written by David Herrmann. This module works similar to uinput, but
dealing with HID traffic.

The input part is working without problems and now I looking how to
support Output and Feature reports. Doing some basic testing it seems
we don't support Output reports very well. A basic use-case for that
is synchronizing keyboard leds (CAPS, NUM, etc) when there is more
than one keyboard connected, and I couldn't get this to work on my
tests. I didn't try to test feature reports since it's still not very
clear to me when they're used. The HID spec says that they exist for
setting values/controls on the peripheral that are not visible to the
user, but I haven't discovered any real-world use case for them. What
is the current state for these reports, do we support them somehow?

I think my last revision of uhid added support for feature-reports.
However, they are not used by the generic hid-input driver. If there
is currently no specialized bluetooth-LE HID driver, you also do not
need to worry about it.

Yes, I've seen the new event types for them, what I'm looking for is
how to trigger them.

I think the new generic HID battery-detection code uses the
feature-reports. However, its disabled by default if I remember
correctly.

I will check this out, thanks for the hint.

About output-reports: Do you correctly receive the UHID_OUTPUT_EV
request in user-space? Do you send the packet to the device? How do
you format it?

I was correctly receiving it with the previous version of your patch.
With this new version I still receive an event coming from the host,
but with ev.type value out of the enum range (different values like
this: 2062073856).

What is the difference between UHID_OUTPUT_EV and UHID_OUTPUT events?
Should both of them be written to the device?

quoted

Also, I've found usbhid-dump by Greg K-H, but it doesn't seem to dump
data coming from the host to the device. Is this on the plans for this
tool? I can try to send a patch for it, if someone can guide me on how
this whole output/feature reports thing should work.

Regards
David

-- 
João Paulo Rechi Vita
Openbossa Labs - INdT
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

On Tue, 29 May 2012, Joao Paulo Rechi Vita wrote:

quoted

I think the new generic HID battery-detection code uses the
feature-reports. However, its disabled by default if I remember
correctly.

I will check this out, thanks for the hint.

Yes, battery power level is one place where feature reports are currently 
being used.

There are also several devices that need specific feature report to be 
sent to them in order to be able to use them at all (initialization). 
Magicmouse, to name another example, uses feature report to switch it 
between singletouch and multitouch modes. Etc.

-- 
Jiri Kosina
SUSE Labs