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, 1 Jun 2022 10:59:52 +0530
From:   Pratyush Yadav <p.yadav@ti.com>
To:     Rob Herring <robh@kernel.org>
CC:     Serge Semin <fancer.lancer@gmail.com>,
        Mark Brown <broonie@kernel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        <linux-spi@vger.kernel.org>, <devicetree@vger.kernel.org>,
        <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] spi: dt-bindings: Move 'rx-sample-delay-ns' to
 spi-peripheral-props.yaml
Message-ID: <20220601052952.4htn4ycta2dkvd3l@ti.com>
References: <20220525210053.2488756-1-robh@kernel.org>
 <20220526054642.zw44mgw2bd2u5v76@ti.com>
 <20220526135404.GA3831942-robh@kernel.org>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Disposition: inline
In-Reply-To: <20220526135404.GA3831942-robh@kernel.org>
Precedence: bulk

On 26/05/22 08:54AM, Rob Herring wrote:
> On Thu, May 26, 2022 at 11:16:42AM +0530, Pratyush Yadav wrote:
> > Hi Rob,
> > 
> > On 25/05/22 04:00PM, Rob Herring wrote:
> > > SPI bus per device properties must be defined in spi-peripheral-props.yaml
> > > for unevaluatedProperties checks to work correctly on device nodes.
> > > 
> > > This has the side effect of promoting 'rx-sample-delay-ns' to be a
> > > common property, but functionally it's no different if it was defined in
> > > a Synopsys specific schema file.
> > 
> > Functionally it is no different, but does this property make sense for 
> > other controllers? If not then I don't see why we should pollute the 
> > common list with controller-specific ones. For one, this now no longer 
> > makes it obvious that this property should only be used with the 
> > Synopsys controller. And if you keep making small exceptions for other 
> > controllers too, soon the common list will be full of controller 
> > properties and it will be a mess finding out what belongs to who.
> 
> There's at least one other case already:
> 
>   cdns,read-delay:
>     $ref: /schemas/types.yaml#/definitions/uint32
>     description:
>       Delay for read capture logic, in clock cycles.
> 
> 
> Too many common properties is not a problem we have. Too many custom 
> properties doing the same thing is the problem.

I agree. But in this case these two properties have different units. 
rx-sample-delay-ns is obviously in nanoseconds. cdns,read-delay is in 
number of ref clock cycles. If other controllers also use this property, 
it could make sense to make rx-sample-delay-ns the default/common 
property and drivers can then make conversions between the units that 
should actually be programmed.

-- 
Regards,
Pratyush Yadav
Texas Instruments Inc.