linux-kernel - Re: [PATCH] Input: MT - Add support for balanced slot assignment

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

Message-ID: <CAN+gG=FNJ8O-=Q4kpsVSYYY+o3KLHpfm43OT-AgTPkHYF-zVLg@mail.gmail.com>
Date:	Mon, 26 Jan 2015 11:56:02 -0500
From:	Benjamin Tissoires <benjamin.tissoires@...il.com>
To:	Henrik Rydberg <rydberg@...math.org>
Cc:	Dmitry Torokhov <dmitry.torokhov@...il.com>,
	Benjamin Tissoires <benjamin.tissoires@...hat.com>,
	Peter Hutterer <peter.hutterer@...-t.net>,
	linux-input <linux-input@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] Input: MT - Add support for balanced slot assignment

On Thu, Jan 22, 2015 at 2:52 PM, Henrik Rydberg <rydberg@...math.org> wrote:
> Some devices are not fast enough to differentiate between a
> fast-moving contact and a new contact. This problem cannot be fully
> resolved because information is truly missing, but it is possible
> to safe-guard against obvious mistakes by restricting movement with
> a maximum displacement.
>
> The new problem formulation for dmax > 0 cannot benefit from the
> speedup for positive definite matrices, but since the convergence is
> faster, the result is about the same. For a handful of contacts, the
> latency difference is truly negligible.
>
> Suggested-by: Benjamin Tissoires <benjamin.tissoires@...hat.com>
> Not-yet-signed-off-by: Henrik Rydberg <rydberg@...math.org>
> ---
> Hi Dmitry, Benjamin,
>
> This patch incorporates the notion of a speed limit in the slot
> assignment function, without changing the behavior for existing code.
>
> There are multi-touch movement cases where the difference between this
> algorithm and a simple deassign-overspeeding-contacts algorithm which
> warrants this approach. One example is when a whole array of contacts
> are moved one step to the right in one frame. With the simple
> algorithm, all contacts will be deassigned, whereas with this
> algorithm, the leftmost contact is deassigned and a new contact
> created on the right. This is the interpretation of smallest change,
> and is consistent with the behavior one would get from two contacts,
> for both algorithms.
>
> I have tested this on a number of testcases in my tree, and I am
> running the patch with a speed limit on my device as I write this. If
> this patch solves Benjamin's problem, I think we are ok.
>

Tested this morning, and yes, it solves the problem.
I assumed that the dmax was in mm, and used "10 * priv->x_res", which
seemed to do the trick:
- tapping with two fingers side by side triggered the jumps
- in the general case (switching from the index to the thumb to
click), the jumps disappeared.

Maybe we should also add a doc telling which units the dmax should be.

When you'll sign this, you can add my tested-by.

Cheers,
Benjamin
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/