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;
Message-ID: <1ab7b38d-e356-6cb8-f101-9499eb34e026@linaro.org>
Date:   Mon, 8 Aug 2022 07:48:31 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.12.0
Subject: Re: [PATCH v2] dt-bindings: iio: gyroscope: bosch,bmg160: correct
 number of pins
Content-Language: en-US
To:     Jonathan Cameron <jic23@kernel.org>
Cc:     Lars-Peter Clausen <lars@metafoo.de>,
        Rob Herring <robh+dt@kernel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        "H. Nikolaus Schaller" <hns@goldelico.com>,
        linux-iio@vger.kernel.org, devicetree@vger.kernel.org,
        linux-kernel@vger.kernel.org
References: <20220805075503.16983-1-krzysztof.kozlowski@linaro.org>
 <20220806153221.1a90bd28@jic23-huawei>
From:   Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
In-Reply-To: <20220806153221.1a90bd28@jic23-huawei>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 06/08/2022 16:32, Jonathan Cameron wrote:
> On Fri,  5 Aug 2022 09:55:03 +0200
> Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> wrote:
> 
>> BMG160 has two interrupt pins to which interrupts can be freely mapped.
>> Correct the schema to express such case and fix warnings like:
>>
>>   qcom/msm8916-alcatel-idol347.dtb: gyroscope@68: interrupts: [[97, 1], [98, 1]] is too long
>>
>> However the basic issue still persists - the interrupts should come in a
>> defined order.
>>
>> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
>>
>> ---
>>
>> Changes since v1:
>> 1. Accept also INT2 as one interrupt (Jonathan).
> 
> This doesn't work. If we are going to support either interrupt, at the very least
> we need to require interrupt-names if the first one isn't INT1.  So your fix
> is right but not enough.
> 
> Driver may ignore interrupt-names for now (would be good to have a sanity check in driver
> though so the driver explicitly checks for INT2 and doesn't use the interrupt if
> it is INT2 - support for that being for a future 'feature' addition).
> 
> A hacky solution would be to require the first one to always be INT1 but that
> gives us no (backwards compatible) path forwards if someone does have a board
> where only INT2 is wired.
> 
> So minimum change I think will be to provide interrupt-names allowing any of
> INT1 (default if not specified)
> INT1, INT2
> INT2

This is exactly what my fix is doing. What else do you need?
interrupt-names is just a helper which anyway driver does not use, so
enforcing it now does not change much.


Best regards,
Krzysztof