Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20181128104350.31902-1-linus.walleij@linaro.org>
 <CAMpxmJUsAccSKELNxn6DEWO6_3aCbzM=hZyaXUCr30S0-Eh0mA@mail.gmail.com> <20181129190132.GM2089@sirena.org.uk>
In-Reply-To: <20181129190132.GM2089@sirena.org.uk>
From:   Bartosz Golaszewski <brgl@bgdev.pl>
Date:   Fri, 30 Nov 2018 09:35:09 +0100
Message-ID: <CAMRc=McApz-8qUo9SiqZgYXor4_DkBRY61nMhiFQ2QSSBj6gWg@mail.gmail.com>
Subject: Re: [PATCH 00/10] Regulator ena_gpiod fixups
To:     Mark Brown <broonie@kernel.org>
Cc:     Bartosz Golaszewski <bgolaszewski@baylibre.com>,
        Linus Walleij <linus.walleij@linaro.org>, lgirdwood@gmail.com,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        ckeepax@opensource.cirrus.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

czw., 29 lis 2018 o 20:01 Mark Brown <broonie@kernel.org> napisa=C5=82(a):
>
> On Thu, Nov 29, 2018 at 07:38:20PM +0100, Bartosz Golaszewski wrote:
>
> > I'm wondering if instead of using the non-devm variants of
> > gpiod_get_*() routines, we shouldn't provide helpers in the regulator
> > framework that would be named accordingly, for example:
> > regulator_gpiod_get_optional() etc. even if all they do is call the
> > relevant gpiolib function. Those helpers could then be documented as
> > passing the control over GPIO lines over to the regulator subsystem.
>
> > The reason for that is that most driver developers will automatically
> > use devm functions whenever available and having a single non-devm
> > function without any comment used in a driver normally using devres
> > looks like a bug. Expect people sending "fixes" in a couple months.
>
> I predict that people would then immediately start demanding devm_
> variants of that function...

At least we could document it in the code.

If I wouldn't know about the reason for not using devm and saw a stray
gpiod_get() without a corresponding put() I'd probably send a patch to
fix it, but if I saw something like regulator_gpiod_get(), I'd look at
what this routine does.

Matter of taste I guess, but I'd prefer the latter. At the very least
we could add a comment to each call saying that the regulator
framework will take care of that resource.

Bart