linux-kernel - Re: [RFC PATCH 0/2] Input: adc-joystick: add detachable devices support

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]

Date:   Sun, 6 Nov 2022 11:47:31 +0300
From:   Siarhei Volkau <lis8215@...il.com>
To:     Paul Cercueil <paul@...pouillou.net>
Cc:     Dmitry Torokhov <dmitry.torokhov@...il.com>,
        Rob Herring <robh+dt@...nel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Artur Rojek <contact@...ur-rojek.eu>,
        linux-input@...r.kernel.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH 0/2] Input: adc-joystick: add detachable devices support

вс, 6 нояб. 2022 г. в 02:39, Paul Cercueil <paul@...pouillou.net>:
>
> Hi Siarhei,
>
> Le lun. 31 oct. 2022 à 22:01:57 +0300, Siarhei Volkau
> <lis8215@...il.com> a écrit :
> > For detachable or lock-able joysticks the ADC lanes might
> > be biased to GND or AVDD when the joystick is detached/locked.
> >
> > One such kind of joystick is found in the Ritmix RZX-50 handheld.
> > The joystick is non-detachable, although ADC lane biased to power
> > supply when the "Hold" switch is activated.
>
> But the RZX-50 has no joystick...

Well, actually there's two versions in the wild (google "RZX-50 pictures"):
 - with analog joystick and speakers on the back side
 - without the joystick and speakers on the front side
I have only the first one at the moment, but I'm looking for another one.

> Or is the d-pad actually wired to the ADC instead of GPIOs?

The D-Pad is another kind of pain there - it's a part of the matrix-keypad
but the pad's row line is shared with the LCD HSYNC signal.

> > To avoid reporting old/broken measurements valid-range is introduced.
> > When measured value is outside valid-range the driver reports
> > safe center position for corresponding axis.
>
> First of all, you already have a "valid range", it is called
> "abs-range"; no need for a new one.
>
> Then, the driver has no business doing events filtering. Notice that
> when you activate the "hold" button and your joystick values go way
> off-range, you still get input events in userspace: that's because the
> kernel is not responsible for enforcing the deadzone, the userspace is.
>
> In your case, you need to update your userspace applications/libraries
> so that when the joystick values are way off-range, the assumed
> position is the center.

Many userspace apps use SDL library to handle joystick input, the SDL
unfortunately hides the fact that the joystick position is out of range -
it does normalization and clamping of the abs-range to an int16_t range.

I see two approaches to do that in userspace:
1. Make a quirk in the SDL library, which affects many types of joysticks
and lead to a fragmentation (special userspace apps for every hardware
isn't the right way in my opinion).
2. Make a special filter app which will filter out invalid joystick events and
route rest to the /dev/input/uinput. Not sure how to hide the real joystick
device from libraries like SDL then.
Both of them don't look like a "silver bullet", however feasible.

>
> Cheers,
> -Paul
>
>

Thank you.