Return-Path: MIME-Version: 1.0 From: Adam Bliss Date: Sun, 14 Feb 2010 22:29:31 -0800 Message-ID: Subject: SolidTek ASK-3962 keyboard works but mouse doesn't [patch] To: linux-bluetooth@vger.kernel.org Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-bluetooth-owner@vger.kernel.org List-ID: Hi there, I bought a SolidTek ASK-3962 Keyboard/Mouse for use with my Home Theatre PC. The keyboard connects and works great with bluez 4.60 -- hooray! However, the mouse does not work. I traced this down to the subclass the device reports when it connects. It reports 0x40, while bluez expects it to include 0x80 to be treated as a mouse. I found this patch to work great: --- bluez-4.60/input/device.c 2010-01-09 12:52:17.000000000 -0800 +++ bluez-4.60-new/input/device.c 2010-02-14 22:16:22.000000000 -0800 @@ -647,6 +647,12 @@ if (idev->name) strncpy(req->name, idev->name, sizeof(req->name) - 1); + if (req->vendor == 0x0dc6 && req->product == 0x3752) { + // The SolidTek ask-3962 has a keyboard and a mouse, but + // incorrectly reports its subclass. + req->subclass |= 0x80; + } + /* Encryption is mandatory for keyboards */ if (req->subclass & 0x40) { err = bt_acl_encrypt(&idev->src, &idev->dst, encrypt_completed, req); (You can also do this with the old hidd program by passing "-b 192" to force the subclass.) I don't know if one-off hacks in device.c are the right way to solve this problem -- I was "inspired" by the one-off patch in there already for 054c:0268. Let me know if there's a more appropriate way to fix this, and I'll be happy to send in another patch. Cheers, --Adam Bliss