linux-kernel - Re: [PATCH v5 3/6] dt-bindings: touchscreen: add touchscreen-glitch-threshold-ns property

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]

Message-ID: <CABGWkvr8X5a0ezeu6HDCMfjh+xbg-bQq4cLwzRD2BvoJsvH_BA@mail.gmail.com>
Date: Sat, 20 Sep 2025 11:39:59 +0200
From: Dario Binacchi <dario.binacchi@...rulasolutions.com>
To: Rob Herring <robh@...nel.org>
Cc: linux-kernel@...r.kernel.org, Frank Li <Frank.Li@....com>, 
	linux-amarula@...rulasolutions.com, Conor Dooley <conor.dooley@...rochip.com>, 
	Conor Dooley <conor+dt@...nel.org>, Dmitry Torokhov <dmitry.torokhov@...il.com>, 
	Javier Carrasco <javier.carrasco@...fvision.net>, Jeff LaBundy <jeff@...undy.com>, 
	Krzysztof Kozlowski <krzk+dt@...nel.org>, devicetree@...r.kernel.org, 
	linux-input@...r.kernel.org
Subject: Re: [PATCH v5 3/6] dt-bindings: touchscreen: add touchscreen-glitch-threshold-ns
 property

On Fri, Sep 19, 2025 at 10:44 PM Rob Herring <robh@...nel.org> wrote:
>
> On Fri, Sep 19, 2025 at 05:12:42PM +0200, Dario Binacchi wrote:
> > On Fri, Sep 19, 2025 at 4:38 PM Rob Herring <robh@...nel.org> wrote:
> > >
> > > On Thu, Sep 18, 2025 at 10:37:37PM +0200, Dario Binacchi wrote:
> > > > On Thu, Sep 18, 2025 at 10:04 PM Rob Herring <robh@...nel.org> wrote:
> > > > >
> > > > > On Thu, Sep 18, 2025 at 05:52:31PM +0200, Dario Binacchi wrote:
> > > > > > Add support for glitch threshold configuration. A detected signal is valid
> > > > > > only if it lasts longer than the set threshold; otherwise, it is regarded
> > > > > > as a glitch.
> > > > > >
> > > > > > Signed-off-by: Dario Binacchi <dario.binacchi@...rulasolutions.com>
> > > > > > Acked-by: Conor Dooley <conor.dooley@...rochip.com>
> > > > > >
> > > > > > ---
> > > > > >
> > > > > > Changes in v5:
> > > > > > - Add Acked-by tag of Conor Dooley
> > > > > >
> > > > > > Changes in v2:
> > > > > > - Added in v2.
> > > > > >
> > > > > >  .../devicetree/bindings/input/touchscreen/touchscreen.yaml    | 4 ++++
> > > > > >  1 file changed, 4 insertions(+)
> > > > > >
> > > > > > diff --git a/Documentation/devicetree/bindings/input/touchscreen/touchscreen.yaml b/Documentation/devicetree/bindings/input/touchscreen/touchscreen.yaml
> > > > > > index 3e3572aa483a..a60b4d08620d 100644
> > > > > > --- a/Documentation/devicetree/bindings/input/touchscreen/touchscreen.yaml
> > > > > > +++ b/Documentation/devicetree/bindings/input/touchscreen/touchscreen.yaml
> > > > > > @@ -206,6 +206,10 @@ properties:
> > > > > >
> > > > > >          unevaluatedProperties: false
> > > > > >
> > > > > > +  touchscreen-glitch-threshold-ns:
> > > > > > +    description: Minimum duration in nanoseconds a signal must remain stable
> > > > > > +      to be considered valid.
> > > > >
> > > > > What's wrong with debounce-delay-ms?
> > > >
> > > > Do you mean that I should rename touchscreen-glitch-threshold-ns to
> > > > debounce-delay-ms?
> > >
> > > I mean that's the common property we already have, so use it or explain
> > > why you aren't using it. I suppose the definition is technically a bit
> > > different if it's purely a s/w delay vs. h/w monitoring of the signal
> > > state. I don't think it matters if the interpretation by each driver is
> > > a bit different.
> > >
> > > Maybe msec is not enough resolution for you could be another reason?
> >
> > Yes, this is the main reason. As specified in the following patch:
> >   v5 4/6 dt-bindings: touchscreen: fsl,imx6ul-tsc: support glitch threshold
> >
> > Drivers must convert this value to IPG clock cycles and map
> > it to one of the four discrete thresholds exposed by the
> > TSC_DEBUG_MODE2 register:
> >
> >   0: 8191 IPG cycles
> >   1: 4095 IPG cycles
> >   2: 2047 IPG cycles
> >   3: 1023 IPG cycles
> >
> > In my case, the IPG clock runs at 66 MHz, which corresponds to:
> >
> > 124 µs for 0
> > 62 µs for 1
> > 31 us for 2
> > 15 us for 3
> >
> > So using milliseconds would not fit my use case. A possible trade-off
> > could be to use debounce-delay-us. Would that be acceptable?
>
> I agree it wouldn't map to what the h/w provides, but is what the h/w
> provides actually useful? There's plenty of h/w designed that's not
> useful. 15us is quite short for a glitch. Do you have an actual cases
> where the different values above are needed?

Considering an IPG clock at 66 MHz, currently at reset the deglitch
filter is set to 124 µs,
the driver sets it to 31 µs with a hardcoded value, and in my use case
I need to set it to 62 µs,
as you can see in the patch:
https://lore.kernel.org/all/20250918155240.2536852-6-dario.binacchi@amarulasolutions.com/
and its handling in
https://lore.kernel.org/all/20250918155240.2536852-7-dario.binacchi@amarulasolutions.com/

Another option could be to use a specific binding for the
fsl,imx6ul-tsc controller, as I did in the
earlier versions of the series.

Thanks and regards,
Dario

>
> Rob



-- 

Dario Binacchi

Senior Embedded Linux Developer

dario.binacchi@...rulasolutions.com

__________________________________


Amarula Solutions SRL

Via Le Canevare 30, 31100 Treviso, Veneto, IT

T. +39 042 243 5310
info@...rulasolutions.com

www.amarulasolutions.com