Message-ID: <20250212-estate-tapeless-08fcdf5b5ca5@spud>
Date: Wed, 12 Feb 2025 18:56:16 +0000
From: Conor Dooley <conor@...nel.org>
To: Andrew Jeffery <andrew@...econstruct.com.au>
Cc: Naresh Solanki <naresh.solanki@...ements.com>,
	Guenter Roeck <linux@...ck-us.net>, broonie@...nel.org,
	Jean Delvare <jdelvare@...e.com>, Rob Herring <robh@...nel.org>,
	Krzysztof Kozlowski <krzk+dt@...nel.org>,
	Conor Dooley <conor+dt@...nel.org>,
	Liam Girdwood <lgirdwood@...il.com>, linux-hwmon@...r.kernel.org,
	devicetree@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/2] dt-bindings: hwmon: ir38060: Move & update dt binding

On Wed, Feb 12, 2025 at 09:13:11PM +1030, Andrew Jeffery wrote:
> On Thu, 2025-02-06 at 18:09 +0000, Conor Dooley wrote:
> > On Thu, Feb 06, 2025 at 09:23:03PM +0530, Naresh Solanki wrote:
> > > On Thu, 6 Feb 2025 at 01:43, Conor Dooley <conor@...nel.org> wrote:
> > > > On Wed, Feb 05, 2025 at 03:51:25PM +0530, Naresh Solanki wrote:
> > > > > On Wed, 5 Feb 2025 at 00:52, Conor Dooley <conor@...nel.org>
> > > > > wrote:
> > > > > > On Tue, Feb 04, 2025 at 11:33:03PM +0530, Naresh Solanki
> > > > > > wrote:
> > > > > > > +  regulators:
> > > > > > > +    type: object
> > > > > > > +    description:
> > > > > > > +      list of regulators provided by this controller.
> > > > > > 
> > > > > > Can you explain why this change is justified? Your commit
> > > > > > message is
> > > > > > explaining what you're doing but not why it's okay to do.
> > > > 
> > > > > This is based on other similar dt-bindings under hwmon/pmbus.
> > > > 
> > > > Okay, but what I am looking for is an explanation of why it is
> > > > okay to
> > > > change the node from
> > > > 
> > > > > regulator@34 {
> > > > >   compatible = "infineon,ir38060";
> > > > >   reg = <0x34>;
> > > > > 
> > > > >   regulator-min-microvolt = <437500>;
> > > > >   regulator-max-microvolt = <1387500>;
> > > > > };
> > > As I have understood the driver, this isn't supported.
> > > > 
> > > > to
> > > > 
> > > > > regulator@34 {
> > > > >     compatible = "infineon,ir38060";
> > > > >     reg = <0x34>;
> > > > > 
> > > > >     regulators {
> > > > >         vout {
> > > > >             regulator-name = "p5v_aux";
> > > > >             regulator-min-microvolt = <437500>;
> > > > >             regulator-max-microvolt = <1387500>;
> > > > >         };
> > > > >     };
> > > Above is the typical approach in other pmbus dt bindings.
> > > Even pmbus driver expects this approach.
> > > > 
> > > > ?
> > > > 
> > > > Will the driver handle both of these identically? Is backwards
> > > > compatibility with the old format maintained? Was the original
> > > > format
> > > > wrong and does not work? Why is a list of regulators needed when
> > > > the
> > > > device only provides one?
> > > Driver doesn't support both.
> > > Based on the pmbus driver original format was wrong.
> > > pmbus driver looks for a regulator node to start with.
> > > 
> > > Reference:
> > > https://github.com/torvalds/linux/blob/master/drivers/hwmon/pmbus/pmbus.h#L515
> > 
> > Then all of the in-tree users are all just broken? They're in aspeed
> > bmcs, so I would not be surprised at all if that were the case.
> 
> Can you unpack the intent of this remark for me a little?
> 
> The history of the problem from what I can see looks like:
> 
>    1. pmbus regulator support exploiting "regulators" as an OF child
>       node was merged for 3.19[1]
>    2. The infineon driver support was merged with trivial bindings[2]
>       and released in v5.17
>    3. A patch was proposed that extracted the Infineon regulator
>       compatibles and provided a dedicated binding[3], however it
>       lacked the "regulators" object property 
>    4. The patch in 3 was merged as [4] with relevant tags, and was
>       released in v6.9
>    5. The system1 devicetree was merged and released in v6.10, and sbp1
>       is merged in v6.14-rc1 for release in v6.14. Both devicetrees, as
>       far as I'm aware, conform to the binding as written.
> 
> In addition to keeping an eye out for Rob's bot, I check all Aspeed-
> related devicetree patches against the bindings using the usual tooling
> while applying them. I would like to avoid diving into driver
> implementations as a blocker to applying devicetree patches where
> possible - the formalised bindings and tooling should exist to separate
> us from having to do that.
> 
> If the complaint is that people submitting Aspeed devicetree patches
> are regularly not testing them to make sure they behave correctly on
> hardware, then sure, that's something to complain about. Otherwise, I'm
> well aware of the (Aspeed) bindings and warnings situation; we've
> spoken about it previously. If there's something I should change in my
> process (beyond eventually addressing all the warnings) please let me
> know, but I don't see that there is in this specific instance.

Ye, it's not a jab at aspeed maintainership, it's about the bmc stuff in
particular. I saw far too many warnings from Rob's bot on series with a
version number where the submitter should know better, so the idea that
it had not been tested in other ways wasn't exactly a stretch.

I made a mistake how I pulled these devices out of trivial-devices.yaml,
given the existing driver didn't work with that binding, but I don't
really see why there's a requirement for a regulator child here in the
first place. I get it for something like the lm25066 that is a monitor
IC that you connect a regulator to, as the regulator is a distinct
device - but the ir38060 is a regulator that has a pmbus interface so
both describe the same device.

Download attachment "signature.asc" of type "application/pgp-signature" (229 bytes)

Powered by blists - more mailing lists