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;
Feedback-ID: i56a14606:Fastmail
User-Agent: Cyrus-JMAP/3.7.0-alpha0-934-g6274855a4c-fm-20220913.002-g6274855a
Mime-Version: 1.0
Message-Id: <0485fd4d-21f3-4972-8667-91959180e60f@www.fastmail.com>
In-Reply-To: <20220914142713.29351-2-lukas.bulwahn@gmail.com>
References: <20220914142713.29351-1-lukas.bulwahn@gmail.com>
 <20220914142713.29351-2-lukas.bulwahn@gmail.com>
Date:   Wed, 14 Sep 2022 16:33:13 +0200
From:   "Arnd Bergmann" <arnd@arndb.de>
To:     "Lukas Bulwahn" <lukas.bulwahn@gmail.com>,
        "Linus Walleij" <linus.walleij@linaro.org>,
        "Russell King" <linux@armlinux.org.uk>,
        "Catalin Marinas" <catalin.marinas@arm.com>,
        "Will Deacon" <will@kernel.org>,
        "Damien Le Moal" <damien.lemoal@opensource.wdc.com>,
        linux-arm-kernel@lists.infradead.org, linux-ide@vger.kernel.org
Cc:     kernel-janitors@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 2/2] ata: make PATA_PLATFORM selectable only for suitable
 architectures
Content-Type: text/plain
Precedence: bulk

On Wed, Sep 14, 2022, at 4:27 PM, Lukas Bulwahn wrote:
> It is currently possible to select "Generic platform device PATA support"
> in two situations:
>
>   - architecture allows the generic platform device PATA support and
>     indicates that with "select HAVE_PATA_PLATFORM".
>   - if the user claims to be an EXPERT by setting CONFIG_EXPERT to yes
>
> However, there is no use case to have Generic platform device PATA support
> in a kernel build if the architecture definition, i.e., the selection of
> configs by an architecture, does not support it.
>
> If the architecture definition is wrong, i.e., it just misses a 'select
> HAVE_PATA_PLATFORM', then even an expert that configures the kernel build
> should not just fix that by overruling the claimed support by an
> architecture. If the architecture definition is wrong, the expert should
> just provide a patch to correct the architecture definition instead---in
> the end, if the user is an expert, sending a quick one-line patch should
> not be an issue.
>
> In other words, I do not see the deeper why an expert can overrule the
> architecture definition in this case, as the expert may not overrule the
> config selections defined by the architecture in the large majority
> ---or probably all other (modulo some mistakes)---of similar cases.
>
> Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>

Sounds reasonable. My best guess about the intention of the EXPERT
dependency is that it would be used for users with third-party
board files or dts files referencing these. We can't really help
users with out-of-tree boardfiles, and the external dts case would
be covered by your patch 1/2.

Reviewed-by: Arnd Bergmann <arnd@arndb.de>