Return-Path: MIME-Version: 1.0 In-Reply-To: References: <1332793209-2950-1-git-send-email-dh.herrmann@googlemail.com> Date: Thu, 26 Apr 2012 15:19:20 -0300 Message-ID: Subject: Re: [RFC v2 0/1] User-space HID I/O Driver From: Joao Paulo Rechi Vita To: David Herrmann Cc: Claudio Takahasi , Jiri Kosina , linux-input@vger.kernel.org, chen.ganir@ti.com, linux-bluetooth@vger.kernel.org, anderson.lizardo@openbossa.org, marcel@holtmann.org Content-Type: text/plain; charset=UTF-8 List-ID: On Thu, Apr 26, 2012 at 2:54 PM, David Herrmann wrote: > Hi Claudio > > On Thu, Apr 26, 2012 at 7:22 PM, Claudio Takahasi > wrote: >> Hi David/Jiri, > [snip] >> >> The basic features are now working, we have tested HoG keyboard and >> mouse. The major features missing related to HID are output/feature >> reports, and suspend. > > Output reports should be working. Only thing missing is feature input > reports. They are not implemented yet. However, what driver (except > hidraw and picolcd) uses them? > >> My HID knowledge is quite limited but it seems that output reports are >> not working on uhid, at least I didn't see events from the host. Even >> for USB keyboard it doesn't work, do you have any pointers? > > Do you actually handle the UHID_OUTPUT and UHID_OUTPUT_EV events from > UHID? I have looked at the HoG code recently and haven't seen any code > that correctly parses UHID_OUTPUT_EV packets? These packets contain > LED (capslock, numlock etc.) updates so the keyboard can actually set > them. They also contain BELL/SOUND codes for such devices. Only other > non-standard stuff is sent through UHID_OUTPUT reports. If there are > problems with it, I can help implementing this. > We're handling these events, but this code has not been sent to the mailing list yet. We're going to re-send the same patchset as before, with comments fixed, and send a RFC for the output reports handling afterwards. > Regarding UHID, I will send an updated version this weekend which > mainly improves the internal buffer handling. The API will stay the > same, however, I might implement the feature input reports, then > everything should be ready for merging upstream (if the review cycles > go well ;)). > Good to hear that! We've been using it on a daily basis to work on the HoG implementation and haven't noticed any major issue. --=20 Jo=C3=A3o Paulo Rechi Vita Openbossa Labs - INdT