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;
Date:   Wed, 31 May 2023 14:06:00 +0300
From:   Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To:     Tommaso Merciai <tomm.merciai@gmail.com>
Cc:     Sakari Ailus <sakari.ailus@linux.intel.com>,
        jacopo.mondi@ideasonboard.com, martin.hecht@avnet.eu,
        linuxfancy@googlegroups.com,
        Mauro Carvalho Chehab <mchehab@kernel.org>,
        Rob Herring <robh+dt@kernel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        Conor Dooley <conor+dt@kernel.org>,
        Hans Verkuil <hverkuil@xs4all.nl>,
        Marco Felsch <m.felsch@pengutronix.de>,
        Gerald Loacker <gerald.loacker@wolfvision.net>,
        Krzysztof =?utf-8?Q?Ha=C5=82asa?= <khalasa@piap.pl>,
        Shawn Tu <shawnx.tu@intel.com>,
        Linus Walleij <linus.walleij@linaro.org>,
        Benjamin Mugnier <benjamin.mugnier@foss.st.com>,
        Mikhail Rudenko <mike.rudenko@gmail.com>,
        Nicholas Roth <nicholas@rothemail.net>,
        linux-media@vger.kernel.org, devicetree@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 1/2] media: dt-bindings: alvium: add document YAML
 binding
Message-ID: <20230531110600.GB27043@pendragon.ideasonboard.com>
References: <20230526173955.797226-1-tomm.merciai@gmail.com>
 <20230526173955.797226-2-tomm.merciai@gmail.com>
 <ZHPElYOeD2C1qo4R@kekkonen.localdomain>
 <20230529063907.GB25984@pendragon.ideasonboard.com>
 <20230529064326.GC25984@pendragon.ideasonboard.com>
 <ZHcff4Gpr9QSgpOs@tom-HP-ZBook-Fury-15-G7-Mobile-Workstation>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <ZHcff4Gpr9QSgpOs@tom-HP-ZBook-Fury-15-G7-Mobile-Workstation>
Precedence: bulk

Hi Tommaso,

On Wed, May 31, 2023 at 12:20:47PM +0200, Tommaso Merciai wrote:
> On Mon, May 29, 2023 at 09:43:26AM +0300, Laurent Pinchart wrote:
> > On Mon, May 29, 2023 at 09:39:13AM +0300, Laurent Pinchart wrote:
> > > On Sun, May 28, 2023 at 09:16:05PM +0000, Sakari Ailus wrote:
> > > > On Fri, May 26, 2023 at 07:39:43PM +0200, Tommaso Merciai wrote:
> > > > > Add documentation of device tree in YAML schema for the ALVIUM
> > > > > Camera from Allied Vision Inc.
> > > > > 
> > > > > References:
> > > > >  - https://www.alliedvision.com/en/products/embedded-vision-solutions
> > > > > 
> > > > > Signed-off-by: Tommaso Merciai <tomm.merciai@gmail.com>
> > > > > ---
> > > > > Changes since v1:
> > > > >  - Fixed build error as suggested by RHerring bot
> > > > > 
> > > > >  .../media/i2c/alliedvision,alvium.yaml        | 115 ++++++++++++++++++
> > > > >  1 file changed, 115 insertions(+)
> > > > >  create mode 100644 Documentation/devicetree/bindings/media/i2c/alliedvision,alvium.yaml
> > > > > 
> > > > > diff --git a/Documentation/devicetree/bindings/media/i2c/alliedvision,alvium.yaml b/Documentation/devicetree/bindings/media/i2c/alliedvision,alvium.yaml
> > > > > new file mode 100644
> > > > > index 000000000000..81e9e560c99d
> > > > > --- /dev/null
> > > > > +++ b/Documentation/devicetree/bindings/media/i2c/alliedvision,alvium.yaml
> > > > > @@ -0,0 +1,115 @@
> > > > > +# SPDX-License-Identifier: GPL-2.0
> > > > > +%YAML 1.2
> > > > > +---
> > > > > +$id: http://devicetree.org/schemas/media/i2c/alliedvision,alvium.yaml#
> > > > > +$schema: http://devicetree.org/meta-schemas/core.yaml#
> > > > > +
> > > > > +title: Alliedvision Alvium Camera
> > > 
> > > s/Alliedvision/Allied Vision/
> > > 
> > > > > +
> > > > > +maintainers:
> > > > > +  - Tommaso Merciai <tomm.merciai@gmail.com>
> > > > > +  - Martin Hecht <martin.hecht@avnet.eu>
> > > > > +
> > > > > +allOf:
> > > > > +  - $ref: /schemas/media/video-interface-devices.yaml#
> > > > > +
> > > > > +properties:
> > > > > +  compatible:
> > > > > +    const: alliedvision,alvium
> > > 
> > > The name is very generic. There are Alvium camera modules that have a
> > > GMSL or FPD-Link interface, and I'm pretty sure those will require a
> > > different driver. I would add module-specific compatible strings (e.g.
> > > "alliedvision,alvium-1500c", ...) here, with a generic fallback.
> > > "alliedvision,alvium" isn't good as it won't cover GMSL or FPD-Link,
> > > maybe "alliedvision,alvium-csi2" would be an option.
> > 
> > Actually, "alvium-1500c" as a specific compatible string won't do. You
> > need the exact model in the compatible string, otherwise it won't be
> > possible for the driver to handle device-specific configuration (for
> > instance accessing registers of the camera sensor for fine-grained
> > configuration). I would thus recommend using "alliedvision,alvium-1500c"
> > and "alliedvision,alvium-1800c" as generic fallbacks, along compatible
> > strings that include the exact device model.
> 
> Agree with alliedvision,alvium-csi2 and thanks for your suggestion.
> In my opinion we don’t need names for 1500c and 
> others because the same driver can drive all the alvium models.
> Alvium is taking care of different sensor abstractions.
> 
> I test with this driver with the following models:
> 
>  - 1800 C-1240c
>  - 1800 C-040c
>  - 1500 C-500
> 
> What do you think about?

As far as I understand, the camera modules allow accessing sensors
registers from the host (through the ISP) for fine-grained
configuration. Even if that's not implemented in the driver at the
moment, I think it's an important feature to eventually support, and it
will require a way for the system to identify the camera module
precisely, to know which sensor the module uses. That's why I would like
that information to be available in DT, in the form of a compatible
string. For instance,

	compatible = "alliedvision,alvium-1500c-1240c",
		     "alliedvision,alvium-1500c";

The driver will only need DT match entries for
"alliedvision,alvium-1500c" and "alliedvision,alvium-1800c".

> > > > > +
> > > > > +  reg:
> > > > > +    maxItems: 1
> > > > > +
> > > > > +  clocks:
> > > > > +    description: XCLK Input Clock
> > > > > +
> > > > > +  clock-names:
> > > > > +    const: xclk
> > > > 
> > > > I'd also drop this as you have a single clock only: it's redundant.
> > > > 
> > > > > +
> > > > > +  powerdown-gpios:
> > > > > +    maxItems: 1
> > > > > +    description: >
> > > > > +      Reference to the GPIO connected to the powerdown pin, if any.
> > > > > +
> > > > > +  reset-gpios:
> > > > > +    maxItems: 1
> > > > > +    description: >
> > > > > +      Reference to the GPIO connected to the reset pin, if any.
> > > 
> > > Reading the Alvium CSI-2 Cameras User Guide, I don't see any powerdown
> > > or reset pin on the 22-pin connector. Am I missing something ? There are
> > > however two GPIOs (in addition to the I2C signals that are also
> > > documented as GPIOs), do you plan to support those ?
> > > 
> > > > > +
> > > > > +  streamon-delay:
> > > > > +    maxItems: 1
> > > > > +    description: >
> > > > > +      Delay before camera start capturing frames in us.
> > > 
> > > Add "-us" to the property name to indicate the unit.
> > > 
> > > This is a vendor-specific property, and should thus have a vendor
> > > prefix.
> > > 
> > > A longer description is needed, from that single line I have no idea
> > > what the property does exactly.
> > > 
> > > > > +
> > > > > +  rotation:
> > > > > +    enum:
> > > > > +      - 0
> > > > > +      - 180
> > > 
> > > Why is the rotation restricted to 0 or 180 ? Someone could mount the
> > > module with  90 degrees rotation, shouldn't the DT bindings allow
> > > describing that ?
> > > 
> > > You need a property for the vcc-ext-in supply.
> > > 
> > > > > +
> > > > > +  port:
> > > > > +    description: Digital Output Port
> > > > > +    $ref: /schemas/graph.yaml#/$defs/port-base
> > > > > +    additionalProperties: false
> > > > > +
> > > > > +    properties:
> > > > > +      endpoint:
> > > > > +        $ref: /schemas/media/video-interfaces.yaml#
> > > > > +        unevaluatedProperties: false
> > > > > +
> > > > > +        properties:
> > > > > +          clock-lanes:
> > > > > +            const: 0
> > > > 
> > > > The driver can know this, no need to have it in DT, i.e. please drop it.
> > > > 
> > > > > +          data-lanes:
> > > > > +            minItems: 1
> > > > > +            maxItems: 4
> > > > > +          link-frequencies: true
> > > > > +
> > > > > +        required:
> > > > > +          - data-lanes
> > > > > +          - link-frequencies
> > > > > +
> > > > > +required:
> > > > > +  - compatible
> > > > > +  - reg
> > > > > +  - clocks
> > > > > +  - clock-names
> > > > > +  - port
> > > > > +
> > > > > +additionalProperties: false
> > > > > +
> > > > > +examples:
> > > > > +  - |
> > > > > +      #include <dt-bindings/gpio/gpio.h>
> > > > > +      #include <dt-bindings/clock/imx8mp-clock.h>
> > > > > +
> > > > > +      i2c {
> > > > > +          #address-cells = <1>;
> > > > > +          #size-cells = <0>;
> > > > > +
> > > > > +          camera: alvium@3c {
> > > > > +              compatible = "alliedvision,alvium";
> > > 
> > > The "alliedvision" prefix is missing from
> > > Documentation/devicetree/bindings/vendor-prefixes.yaml.
> > > 
> > > > > +              pinctrl-names = "default";
> > > > > +              pinctrl-0 = <&pinctrl_csi0_pwn>, <&pinctrl_csi0_rst>, <&pinctrl_csi_mclk>;
> > > 
> > > I'd drop pinctrl, it makes the example longer without adding much value.
> > > 
> > > > > +              reg = <0x3c>;
> > > > > +              clocks = <&clk IMX8MP_CLK_IPP_DO_CLKO2>;
> > > > > +              clock-names = "xclk";
> > > > > +              assigned-clocks = <&clk IMX8MP_CLK_IPP_DO_CLKO2>;
> > > > > +              assigned-clock-parents = <&clk IMX8MP_CLK_24M>;
> > > > > +              assigned-clock-rates = <24000000>;
> > > > > +              streamon-delay = <20>;
> > > > > +              powerdown-gpios = <&gpio2 11 GPIO_ACTIVE_HIGH>;
> > > > > +              reset-gpios = <&gpio1 6 GPIO_ACTIVE_LOW>;
> > > > > +              status = "okay";
> > > > > +
> > > > > +              port {
> > > > > +                  alvium_out: endpoint {
> > > > > +                      remote-endpoint = <&mipi_csi_0_in>;
> > > > > +                      data-lanes = <1 2 3 4>;
> > > > > +                      link-frequencies = /bits/ 64 <681250000>;
> > > > > +                      clock-lanes = <0>;
> > > > > +                  };
> > > > > +              };
> > > > > +          };
> > > > > +      };
> > > > > +
> > > > > +...

-- 
Regards,

Laurent Pinchart