Re: [PATCH v2] pinctrl: qcom: support gpio_chip .set_config call

From: Linus Walleij <linus.walleij@linaro.org>
To: Bjorn Andersson <bjorn.andersson@linaro.org>
Cc: Shawn Guo <shawn.guo@linaro.org>,
	Andy Shevchenko <andriy.shevchenko@intel.com>,
	"open list:GPIO SUBSYSTEM" <linux-gpio@vger.kernel.org>,
	MSM <linux-arm-msm@vger.kernel.org>
Subject: Re: [PATCH v2] pinctrl: qcom: support gpio_chip .set_config call
Date: Thu, 4 Mar 2021 09:41:05 +0100	[thread overview]
Message-ID: <CACRpkdbZNJR5XaNaEWxwKdxED2mXnN_bN+mLjfPRMxyxVP3TVw@mail.gmail.com> (raw)
In-Reply-To: <YD+iWuLS/9knWLFb@builder.lan>

On Wed, Mar 3, 2021 at 3:51 PM Bjorn Andersson
<bjorn.andersson@linaro.org> wrote:

> I like the fact that this solves your gpio configuration issue, but I'm
> uncertain if just adding support for configuring pins (in addition to
> groups) in the driver is the right solution.
>
> @Linus, to summarize, the Qualcomm TLMM configures pingroups, but all
> gpios are defined as a single pin. pinctrl_gpio_set_config() is invoked
> based on the configuration provided in the ACPI tables, so Shawn's
> proposal is to just implement "config by pin" as well.
> Would this not be a problem shared with all pinctrl drivers that
> configure gpios in groups?

It is done as Shawn does it in e.g. the Intel drivers.

This is a side effect of ACPI: ACPI thinks about the world mostly
in term of GPIO pins, there was a pin ctrl draft at one point but I don't
think it ever got off the ground. The standards committe just has not
been able to think about the world in terms of pin control. Or they
think the pin control abstraction is just wrong. Could be either.

This means that on ACPI systems pin config will be done with
this mechanism but on DT systems it will be done another way.
The mechanisms are essentially orthogonal usecase-wise, it should
work as long as there is some proper testing and concern
for both cases.

Yours,
Linus Walleij