linux-kernel - Re: [PATCH 1/1] HID: multitouch: enable palm rejection if device implements confidence usage

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:	Tue, 21 Jun 2016 14:36:48 +0200
From:	Benjamin Tissoires <benjamin.tissoires@...il.com>
To:	Allen Hung <allen_hung@...l.com>
Cc:	Jiri Kosina <jikos@...nel.org>,
	LKML <linux-kernel@...r.kernel.org>,
	linux-input <linux-input@...r.kernel.org>
Subject: Re: [PATCH 1/1] HID: multitouch: enable palm rejection if device
 implements confidence usage

Hi Allen,

On Tue, Jun 21, 2016 at 11:20 AM, Allen Hung <allen_hung@...l.com> wrote:
> Hi Benjamin,
>
> On 01/04/2016 04:35 PM, Benjamin Tissoires wrote:
>> Hi Allen,
>>
>> On Mon, Dec 21, 2015 at 10:45 AM, Allen Hung <allen_hung@...l.com> wrote:
>>> On 12/18/2015 05:52 AM, Benjamin Tissoires wrote:
>>>> On Thu, Nov 26, 2015 at 3:21 PM, Jiri Kosina <jikos@...nel.org> wrote:
>>>>> On Fri, 20 Nov 2015, Allen Hung wrote:
>>>>>
>>>>>> The usage Confidence is mandary to Windows Precision Touchpad devices. The
>>>>>> appearance of this usage is checked in hidinput_connect but the quirk
>>>>>> MT_QUIRK_VALID_IS_CONFIDENCE is not applied to device accordingly.
>>>>>> Apply this quirk and also remove quirk MT_QUIRK_ALWAYS_VALID to enable palm
>>>>>> rejection for the WIN 8 touchpad devices which have implemented usage
>>>>>> Confidence in its input reports.
>>>>>>
>>>>>> Tested on Dell XPS 13 laptop.
>>>>>>
>>>>>> Signed-off-by: Allen Hung <allen_hung@...l.com>
>>>>> Applied to for-4.5/multitouch. Thanks,
>>>>>
>>>> Allen, looks like the Dell XPS13 2015 (or was it the 2014?) has
>>>> problems with your patch:
>>>> https://bugzilla.redhat.com/show_bug.cgi?id=1292583 (we backported
>>>> this change in the Fedora kernel 4.2.7).
>>>>
>>>> Would you mind checking that this patch does not create regressions on
>>>> older series of laptops?
>>>>
>>>> Thanks in advance.
>>>>
>>>> Cheers,
>>>> Benjamin
>>> Hi Benjamin,
>>>
>>> It looks like a XPS13 2014 but I can confirm this with my colleagues who are in charge of this product line. I don't mind checking this patch with more older series of laptops. I will do it and let you know the status. Should I also test the machines with Fedora 23, or just go with Ubuntu OS? The XPS machines are shipped with Ubuntu.
>>> Btw, the issue reporter said she/he cannot reproduce the issue (posted on Dec. 18).
>>>
>>
>> [Sorry for the late answer, but you guessed the holiday season did not help :)]
>>
>> I think I'll just ask the reporter to confirm he can not reproduce and
>> then close the downstream bug.
>> Unless other people complains, I think we can just keep the current
>> state and wait :)
>>
>> Cheers,
>> Benjamin
>>
> My this patch 25a84db15b3f3a24d3ea7d2baf90693bcff34b0c (HID: multitouch: enable
> palm rejection if device implements confidence usage) brings side effect as
> someone reported it on https://bugzilla.kernel.org/show_bug.cgi?id=112791
> My fix is attached in the thread and please have a look at it and my reply in
> comment #5 as the root cause.
> Let me try to explain my fix here - the quirk MT_QUIRK_VALID_IS_CONFIDENCE
> is implemented in hid-multitouch and I see it was only applied to some
> particular touchpad devices (see my list in the end) before my earlier patch
> applies it to Windows 8 precision touchpad. I believe the quirk is working
> very well for the devices in list, and it must have some stories behind that
> made the quirk designed in this way - simply dropped the reports with
> [confidence = 0].
> However, this quirk isn't working very properly for the Win8 precise touchpad.
> I think when a touch is initially detected as finger then
> later detected as a palm, the driver has to send the latter detections
> to input/mt as calling input_mt_report_slot_state(input, MT_TOOL_FINGER,
> active=0). In order not to break the devices in list, I add a new quirk
> MT_QUIRK_CONFIDENCE to do the work. Please provide your comments :)
>
> Device list:
> MT_USB_DEVICE(USB_VENDOR_ID_3M, USB_DEVICE_ID_3M1968)
> MT_USB_DEVICE(USB_VENDOR_ID_3M, USB_DEVICE_ID_3M2256)
> MT_USB_DEVICE(USB_VENDOR_ID_3M, USB_DEVICE_ID_3M3266)
> MT_USB_DEVICE(USB_VENDOR_ID_ASUS, USB_DEVICE_ID_ASUS_T91MT)
> MT_USB_DEVICE(USB_VENDOR_ID_ASUS, USB_DEVICE_ID_ASUSTEK_MULTITOUCH_YFO)
> MT_USB_DEVICE(USB_VENDOR_ID_TURBOX, USB_DEVICE_ID_TURBOX_TOUCHSCREEN_MOSART)
> MT_USB_DEVICE(USB_VENDOR_ID_QUANTA,     USB_DEVICE_ID_QUANTA_OPTICAL_TOUCH_3001)
> MT_USB_DEVICE(USB_VENDOR_ID_STANTUM_STM, USB_DEVICE_ID_MTP_STM)
>

I was discussing with Synaptics about that last week, and I was
planning on working on it this week. Looks like you got the patch
ready before me :)
I *think* the patch is valid, and should be carried in upstream and in
stable. I'd like to have a further look at it once you submit it
however before giving my rev-by.

Regarding the device list you provided, yes those devices should
reject the touch when the confidence bit is not set. These were
Windows 7 touchscreen and the spec was not very clear and we had to
tweak the code on a per-device basis.

I think I also have a few pending patches for hid-multitouch, but your
patch might supersede those.

Cheers,
Benjamin