Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220825180417.1259360-1-mail@conchuod.ie> <20220825180417.1259360-2-mail@conchuod.ie>
 <5f00ab85-d5ac-728d-2157-e70f2a46cc90@canonical.com>
In-Reply-To: <5f00ab85-d5ac-728d-2157-e70f2a46cc90@canonical.com>
From:   Rob Herring <robh+dt@kernel.org>
Date:   Tue, 30 Aug 2022 15:57:10 -0500
Message-ID: <CAL_JsqLdYT5Bqn1aCd+vDxgOQAH_B2nMwsP9siATdK6DRte0QA@mail.gmail.com>
Subject: Re: [PATCH 1/2] dt-bindings: riscv: sifive-l2: add a PolarFire SoC compatible
To:     Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Cc:     Conor Dooley <conor.dooley@microchip.com>,
        Sagar Kadam <sagar.kadam@sifive.com>,
        Atish Patra <atishp@atishpatra.org>,
        Paul Walmsley <paul.walmsley@sifive.com>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        devicetree@vger.kernel.org,
        linux-riscv <linux-riscv@lists.infradead.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        Albert Ou <aou@eecs.berkeley.edu>,
        Daire McNamara <daire.mcnamara@microchip.com>,
        Palmer Dabbelt <palmer@dabbelt.com>,
        Conor Dooley <mail@conchuod.ie>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Aug 25, 2022 at 1:36 PM Heinrich Schuchardt
<heinrich.schuchardt@canonical.com> wrote:
>
> On 8/25/22 20:04, Conor Dooley wrote:
> > From: Conor Dooley <conor.dooley@microchip.com>
> >
> > The l2 cache on PolarFire SoC is cross between that of the fu540 and
> > the fu740. It has the extra interrupt from the fu740 but the lower
> > number of cache-sets. Add a specific compatible to avoid the likes
> > of:
> >
> > mpfs-polarberry.dtb: cache-controller@2010000: interrupts: [[1], [3], [4], [2]] is too long
>
> Where is such a message written? I couldn't find the string in
> next-20220825 (git grep -n 'is too long"').
>
> Why should a different number of cache sets require an extra compatible
> string. cache-size is simply a parameter going with the existing
> compatible strings.
>
> I would assume that you only need an extra compatible string if there is
> a functional difference that can not be expressed with the existing
> parameters.

Correct, but you have to account for unknown functional differences
aka errata as well. Otherwise, we need firmware updates to enable the
OS to handle errata.

> > Fixes: 34fc9cc3aebe ("riscv: dts: microchip: correct L2 cache interrupts")
> > Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
> > ---
> >   .../bindings/riscv/sifive-l2-cache.yaml       | 79 ++++++++++++-------
> >   1 file changed, 49 insertions(+), 30 deletions(-)
> >
> > diff --git a/Documentation/devicetree/bindings/riscv/sifive-l2-cache.yaml b/Documentation/devicetree/bindings/riscv/sifive-l2-cache.yaml
> > index 69cdab18d629..ca3b9be58058 100644
> > --- a/Documentation/devicetree/bindings/riscv/sifive-l2-cache.yaml
> > +++ b/Documentation/devicetree/bindings/riscv/sifive-l2-cache.yaml
> > @@ -17,9 +17,6 @@ description:
> >     acts as directory-based coherency manager.
> >     All the properties in ePAPR/DeviceTree specification applies for this platform.
> >
> > -allOf:
> > -  - $ref: /schemas/cache-controller.yaml#
> > -
> >   select:
> >     properties:
> >       compatible:
> > @@ -33,11 +30,16 @@ select:
> >
> >   properties:
> >     compatible:
> > -    items:
> > -      - enum:
> > -          - sifive,fu540-c000-ccache
> > -          - sifive,fu740-c000-ccache
>
> Why can't you simply add microchip,mpfs-ccache here?
>
> > -      - const: cache
> > +    oneOf:
> > +      - items:
> > +          - enum:
> > +              - sifive,fu540-c000-ccache
> > +              - sifive,fu740-c000-ccache
> > +          - const: cache
> > +      - items:
> > +          - const: microchip,mpfs-ccache
> > +          - const: sifive,fu540-c000-ccache
>
> Why do we need 'sifive,fu540-c000-ccache' twice?

Because it is in 2 different positions. While we can express that the
last N entries in a list are optional, there is no way in json-schema
to express entries at the beginning or in the middle are optional.

Rob