linux-kernel - Re: [PATCH 1/2] dt-bindings: usb: parade,ps5511: Disallow unevaluated properties

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: <CAL_JsqJ31NN7vNxPxRFm6DfVtE=GvrPwPEOcVY4_TZJRG70R6g@mail.gmail.com>
Date: Mon, 12 Jan 2026 17:17:31 -0600
From: Rob Herring <robh@...nel.org>
To: Krzysztof Kozlowski <krzysztof.kozlowski@....qualcomm.com>
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>, Krzysztof Kozlowski <krzk+dt@...nel.org>, 
	Conor Dooley <conor+dt@...nel.org>, Pin-yen Lin <treapking@...omium.org>, 
	Matthias Kaehlcke <mka@...omium.org>, linux-usb@...r.kernel.org, devicetree@...r.kernel.org, 
	linux-kernel@...r.kernel.org, stable@...r.kernel.org
Subject: Re: [PATCH 1/2] dt-bindings: usb: parade,ps5511: Disallow unevaluated properties

On Mon, Jan 12, 2026 at 2:20 PM Rob Herring <robh@...nel.org> wrote:
>
> On Mon, Jan 12, 2026 at 10:01:50AM +0100, Krzysztof Kozlowski wrote:
> > Review given to v2 [1] of commit fc259b024cb3 ("dt-bindings: usb: Add
> > binding for PS5511 hub controller") asked to use unevaluatedProperties,
> > but this was ignored by the author probably because current dtschema
> > does not allow to use both additionalProperties and
> > unevaluatedProperties.  As an effect, this binding does not end with
> > unevaluatedProperties and allows any properties to be added.
> >
> > Fix this by reverting the approach suggested at v2 review and using
> > simpler definition of "reg" constraints.
> >
> > Link: https://lore.kernel.org/r/20250416180023.GB3327258-robh@kernel.org/ [1]
> > Fixes: fc259b024cb3 ("dt-bindings: usb: Add binding for PS5511 hub controller")
> > Cc: <stable@...r.kernel.org>
> > Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@....qualcomm.com>
> > ---
> >  .../devicetree/bindings/usb/parade,ps5511.yaml       | 12 ++++++------
> >  1 file changed, 6 insertions(+), 6 deletions(-)
> >
> > diff --git a/Documentation/devicetree/bindings/usb/parade,ps5511.yaml b/Documentation/devicetree/bindings/usb/parade,ps5511.yaml
> > index 10d002f09db8..154d779e507a 100644
> > --- a/Documentation/devicetree/bindings/usb/parade,ps5511.yaml
> > +++ b/Documentation/devicetree/bindings/usb/parade,ps5511.yaml
> > @@ -15,6 +15,10 @@ properties:
> >        - usb1da0,5511
> >        - usb1da0,55a1
> >
> > +  reg:
> > +    minimum: 1
> > +    maximum: 5
> > +
>
> This 'reg' would be the upstream USB port. We have no idea what its
> constraints are for the value.
>
> >    reset-gpios:
> >      items:
> >        - description: GPIO specifier for RESETB pin.
> > @@ -41,12 +45,6 @@ properties:
> >              minimum: 1
> >              maximum: 5
> >
> > -additionalProperties:
> > -  properties:
> > -    reg:
> > -      minimum: 1
> > -      maximum: 5
>
> Removing this is wrong. This is defining the number of downstream USB
> ports for this hub.
>
> What's wrong here is 'type: object' is missing, so any property that's
> not a object passes (no, 'properties' doesn't imply it's an object).
>
> We should fix dtschema to allow additionalProperties when not a
> boolean property to coexist with unevaluatedProperties. I'll look into
> it.

Actually, allowing both wouldn't make sense here.

If you rely on usb-hub.yaml (via usb-device.yaml) to define 'reg', the
additionalProperties schema is still going to be applied to 'reg' as
unevaluatedProperties has no impact on it. That would happen to work
since there's no 'type: object'.

I think the additionalProperties should be a patternProperties instead
because we need to define the unit-address format:

patternProperties:
  '@[1-5$':
    type: object
    ...

Rob