linux-kernel - Re: [PATCH 2/2] input: Imagis: add support for the IST3032C touchscreen

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: <ZTVZQ+Po20pcDWm8@nixie71>
Date:   Sun, 22 Oct 2023 12:17:55 -0500
From:   Jeff LaBundy <jeff@...undy.com>
To:     Karel Balej <karelb@...li.ms.mff.cuni.cz>
Cc:     Karel Balej <balejk@...fyz.cz>,
        Dmitry Torokhov <dmitry.torokhov@...il.com>,
        Rob Herring <robh+dt@...nel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Conor Dooley <conor+dt@...nel.org>,
        Markuss Broks <markuss.broks@...il.com>,
        linux-input@...r.kernel.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org,
        Duje Mihanović <duje.mihanovic@...le.hr>,
        ~postmarketos/upstreaming@...ts.sr.ht
Subject: Re: [PATCH 2/2] input: Imagis: add support for the IST3032C
 touchscreen

Hi Karel,

On Thu, Sep 28, 2023 at 07:56:45PM +0200, Karel Balej wrote:
> Hello, Jeff,
> 
> thank you very much for your feedback.
> 
> > > +	if (chip_id == IST3032C_WHOAMI) {
> > > +		/*
> > > +		 * if the regulator voltage is not set like this, the touchscreen
> > > +		 * generates random touches without user interaction
> > > +		 */
> > > +		error = regulator_set_voltage(ts->supplies[0].consumer, 3100000, 3100000);
> > > +		if (error)
> > > +			dev_warn(dev, "failed to set regulator voltage\n");
> > > +	}
> > > +
> >
> > Opinions may vary, but mine is that this kind of hard-coded board-level policy
> > does not belong in the driver. Surely the supply need not be equal to exactly
> > 3.1 V and this is merely an upper or lower limit? Assuming so, what if the board
> > designer opts to share this supply with another consumer that requires a specific
> > voltage not equal to 3.1 V, but still within the safe range of IST3032C?
> >
> > To me, this restriction belongs in dts, specifically within the regulator child
> > node referenced by the client which bears the new 'ist3032c' compatible string.
> > Maybe a better solution is to simply note this presumed silicon erratum in the
> > description of the vdd supply in the binding which, as Conor states, must not be
> > clubbed with driver patches.
> 
> I agree that the voltage should not be hardcoded. I do not know what the
> safe range for the touchscreen is though, because the downstream driver
> does exactly this. I will try to test it with several values within the
> range allowed by the regulator and see if I can determine some limits on
> when the "ghost" touches do not appear.
> 
> However I am not sure whether this setting should be moved to the
> regulator DT - it is my understanding that the DT for the regulator
> should list the min/max range *supported* by the regulator, not conform
> to requirements of its consumers, which should instead ask for the
> regulator to be set to a range they require themselves, via their driver
> - is it not so?
> 
> The regulator driver is not mainlined yet (although I managed to get the
> downstream code working with mainline), however the downstream DT
> contains much wider range of supported voltage (compared to those 3.1 V
> used by the touchscreen) - an information which would get lost if I set
> the DT for the regulator by the requirements of the touchscreen, which I
> believe would have similiar implications as what you said regarding
> using this regulator with other consumers.
> 
> What would seem a reasonable solution to me would be to move the voltage
> range values to the touchscreen DT (which incidentally is what the
> downstream driver does also, except it uses one value for both min and
> max), so that they would be set by the driver but not hardcoded in the
> code - what do you think about this?

I believe this has been clarified in the other thread with Markuss, but
just to close this out: in general, individual drivers should not be
setting the output value of a regulator. Instead, they should merely mark
themselves as a consumer of a regulator, and increment or decrement its
usage counter by enabling and disabling the regulator, respectively.

You are correct that the regulator node typically specifies the minimum
and maximum voltages supported by the regulator, but if your board has
a stricter range because said regulator is tied to something such as this
touchscreen controller, then you can override the maximum voltage with
something smaller (e.g. 3.1 V). The regulator framework will then set the
output voltage according to the ultimate range defined in the device tree.

Often times the baseline regulator nodes are defined in a PMIC-specific
.dtsi file, or part of a more generic board definition. Then, some nodes
are overridden at a higher level in the heirarchy tailored to the board,
or an overlay applied by the bootloader.

> 
> Best regards,
> K. B.

Kind regards,
Jeff LaBundy