From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S264093AbTLUUnh (ORCPT ); Sun, 21 Dec 2003 15:43:37 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S264106AbTLUUng (ORCPT ); Sun, 21 Dec 2003 15:43:36 -0500 Received: from zxa8020.lanisdn-gte.net ([206.46.31.146]:38533 "EHLO links.magenta.com") by vger.kernel.org with ESMTP id S264093AbTLUUne (ORCPT ); Sun, 21 Dec 2003 15:43:34 -0500 Date: Sun, 21 Dec 2003 15:43:31 -0500 From: Raul Miller To: linux-kernel@vger.kernel.org Subject: problem with usb duo mouse and keyboard Message-ID: <20031221154331.Z28449@links.magenta.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5.1i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Linux 2.6.0 (and 2.6.0-test11) appears to have a significant problem with the mouse of the logitech 700mx duo when it's plugged in using usb. The mouse works fine* when plugged in using the ps/2 mouse connector, and the keyboard works ok even when plugged in using usb. However, when using usb: o horizontal mouse movement is interpreted as vertical movement o vertical mouse movement is ignored o strange things happen with mouse button presses (I didn't bother working with this configuration long enough to figure out the details). This occurs under both x and gpm, which is mild evidence that it's not a user space problem. It might also be worth noting that some of the extra keys on the keyboard do odd mouse-like things (like move the mouse's pointer horizontally a short distance) when it's a usb keyboard. However, these keys work reasonably normally (or, in two cases**, generate entries in syslog) when it's plugged in as a ps/2 keyboard. My guess is that the respective usb drivers haven't taken into account that both a keyboard and a mouse might appear on the same physical connection. [But I'm still coming up to speed, and haven't yet figured out how to confirm/disprove that guess.] -- Raul Miller moth@magenta.com * one mouse button is ignored, however it is so awkwardly placed that I doubt anyone would want to use it. ** the syslog entries from the unsupported keys: Dec 21 15:28:07 localhost kernel: atkbd.c: Unknown key pressed (translated set 2, code 0x11d, data 0x11, on isa0060/serio0). Dec 21 15:28:07 localhost kernel: atkbd.c: Unknown key released (translated set 2, code 0x11d, data 0x91, on isa0060/serio0). Dec 21 15:28:10 localhost kernel: atkbd.c: Unknown key pressed (translated set 2, code 0x12c, data 0x14, on isa0060/serio0). Dec 21 15:28:10 localhost kernel: atkbd.c: Unknown key released (translated set 2, code 0x12c, data 0x94, on isa0060/serio0). Perhaps it's also worth noting that most of the other extra keys send two byte sequences when pressed (as reported by showkey -s), and showkey -k seems to be reporting the extra information as extra key releases, I think I've seen something similar on other keyboards, but I don't have one of those handy to test on). If anyone cares, I can document this for them.