Hi Joao

On Thu, May 3, 2012 at 5:33 PM, David Herrmann
[off-list ref] wrote:

Hi Joao

On Wed, May 2, 2012 at 9:09 PM, Joao Paulo Rechi Vita
[off-list ref] wrote:

quoted

On Mon, Apr 30, 2012 at 12:27 PM, David Herrmann
[off-list ref] wrote:

quoted

Hi

This implements a hid_ll_driver user-space interface similar to uinput =

for the

quoted

quoted

input devices. It allows to implement the HID transport-layer in user-s=

pace as

quoted

quoted

required by Bluetooth Low-Energy (HoG).

This is my "first final revision" for the module. Please review and tes=

t.

quoted

quoted

Changes from the previous RFCs include:
=A0- The feature-report mechanism is now implemented
=A0- Several min_t() fixes
=A0- Minor coding-style issues

I did not change __attribute__((__packed__)) to __packed as the latter =

one is

quoted

quoted

not provided to user-space.

If there are any more issues, please tell me. And for the record, the H=

oG devs

quoted

quoted

told me that they are working since several months with it now so we ha=

ve a real

quoted

quoted

user-space application that needs it ;)

If there are no more issues (which i doubt) I can also resend this spli=

tted into

quoted

quoted

multiple patches.

Thanks for your work, David. I'll test and review this new version.
Did you find out the what is the problem when passing BUS_BLUETOOTH on
the 'bus' field of the create request?

I've spent some time looking into that but I haven't found the
problem, yet. Maybe Jiri can comment here.

hid-next should fix this:
http://git.kernel.org/?p=3Dlinux/kernel/git/jikos/hid.git;a=3Dcommitdiff;h=
=3D8215d557e5f3a70e50e07c857d35c250fee62a73

I actually haven't tested UHID with hid-next. Some core interfaces are
changed and I will update UHID as soon as Linus pulls the HID updates.

The BUS_USB problems should be gone, though.

quoted

Also, how does the HID subsystem differentiate from HID protocol
versions? We've just found out that HoG exposes what HID protocol
version the device implements, is it the case of having an extra field
on the uhid_create_req structure to pass this information, or is it
already present on the HID descriptor (sorry for my limited knowledge
on the HID protocol itself)?

There is already a "version" field in UHID_CREATE. It isn't mandatory,
though, so setting it to 0 works. I even think the HID core doesn't
even look into that value except for printk messages.

quoted

--
Jo=E3o Paulo Rechi Vita
Openbossa Labs - INdT

Thanks
David

Regards
David