linux-kernel - Re: [PATCH 1/2] dt-bindings: hwmon: max31827: use supply pin name

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, 24 Sep 2023 05:02:14 -0700
From:   Guenter Roeck <linux@...ck-us.net>
To:     "Miclaus, Antoniu" <Antoniu.Miclaus@...log.com>
Cc:     Rob Herring <robh@...nel.org>,
        "Matyas, Daniel" <Daniel.Matyas@...log.com>,
        Jean Delvare <jdelvare@...e.com>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Conor Dooley <conor+dt@...nel.org>,
        "linux-hwmon@...r.kernel.org" <linux-hwmon@...r.kernel.org>,
        "devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/2] dt-bindings: hwmon: max31827: use supply pin name

On Sat, Sep 23, 2023 at 02:19:45PM +0000, Miclaus, Antoniu wrote:
> 
> 
> > On Thu, Sep 21, 2023 at 05:20:03PM +0300, Antoniu Miclaus wrote:
> > > The actual hardware pin name for the supply of max31827 is vdd.
> > > Update the dt-binding to reflect the hardware properties accordingly.
> > 
> > Changing this breaks the ABI. I see the old one wasn't used by the
> > driver, but that's just one driver potentially. You need some
> > justification here why it's okay to break the ABI.
> > 
> As I mentioned also in the commit description, the supply should match the
> actual hardware pin name. Otherwise it might create confusion. Usually vref
> refers to an external voltage reference pin used for ADC/DACs which is not
> exactly the case for this part, taking into account that there is no "reference"
> word mentioned in the datasheet at all. VREF and VDD are usually separate
> hardware pins. There is a hint indeed in the dts example that the vref-supply 
> might be referenced to a vdd regulator node, but from my point of view
> that is not enough. Moreover the current vref-supply is not handled at all in
> the driver, it is only mentioned in the dt-binding (That's why I added a second
> patch in the series handling the supply).
> 
> If the justification is not enough to apply this change, then I can keep only the
> second patch, which handles the regulator in the driver and use the  old `vref`
> naming which currently appears only in the dt-binding.
> 

That would have been a good argument when the property was introduced, but if
there are any systems with existing bindings out there they will use the old
name and fail after this change is applied.

I don't thnk it is mandated that every system in the world would publish their
devicetree bindings in the kernel. That would not scale. So any argument along
the line of "this binding is not used" is not really a valid argument.

Guenter