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;
MIME-Version: 1.0
References: <20230606014234.29491-1-nick.hawkins@hpe.com> <20230606014234.29491-3-nick.hawkins@hpe.com>
 <ZH76x6E9cbfrJHeu@surfacebook> <DM4PR84MB19278E9DD0BF34C570F714708852A@DM4PR84MB1927.NAMPRD84.PROD.OUTLOOK.COM>
 <CAHp75Vcc9hh6MYPSCJa1vniU=rC4Y+j9pCS4Np-b-Pyh5eUwNg@mail.gmail.com> <81CFCEE8-5276-475A-A2AB-A814DEDA04B2@hpe.com>
In-Reply-To: <81CFCEE8-5276-475A-A2AB-A814DEDA04B2@hpe.com>
From:   Andy Shevchenko <andy.shevchenko@gmail.com>
Date:   Wed, 7 Jun 2023 19:30:58 +0300
Message-ID: <CAHp75VeZkLSTw8OFmDMYb+im0qK0NQRHpndzBM7fMHm=HZCzeA@mail.gmail.com>
Subject: Re: [PATCH v3 2/5] gpio: gxp: Add HPE GXP GPIO
To:     "Hawkins, Nick" <nick.hawkins@hpe.com>
Cc:     "Verdun, Jean-Marie" <verdun@hpe.com>,
        "linus.walleij@linaro.org" <linus.walleij@linaro.org>,
        "brgl@bgdev.pl" <brgl@bgdev.pl>,
        "robh+dt@kernel.org" <robh+dt@kernel.org>,
        "krzysztof.kozlowski+dt@linaro.org" 
        <krzysztof.kozlowski+dt@linaro.org>,
        "jdelvare@suse.com" <jdelvare@suse.com>,
        "linux@roeck-us.net" <linux@roeck-us.net>,
        "linux-gpio@vger.kernel.org" <linux-gpio@vger.kernel.org>,
        "devicetree@vger.kernel.org" <devicetree@vger.kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "linux-hwmon@vger.kernel.org" <linux-hwmon@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Wed, Jun 7, 2023 at 7:07=E2=80=AFPM Hawkins, Nick <nick.hawkins@hpe.com>=
 wrote:

...

> > > > Why are you not using gpio-regmap?
> > >
> > > Is there are good example or previous commit you would recommend
> > > looking at that shows how to convert from regmap to gpio-regmap?
> > > Later in the code I am using regmap_read and regmap_update_bits
> > > with large differences in offset registers, and not so much a
> > > contiguous block.
>
>
> > I don't know how good these are, but that's what we have currently as
> > most prominent use of gpio-regmap
>
>
> > 1) (ongoing) https://lore.kernel.org/linux-gpio/20230606092107.764621-6=
-jiawenwu@trustnetic.com <mailto:20230606092107.764621-6-jiawenwu@trustneti=
c.com>/
> > 2) (in the repo)
> > https://elixir.bootlin.com/linux/v6.4-rc4/source/drivers/gpio/gpio-i825=
5.c <https://elixir.bootlin.com/linux/v6.4-rc4/source/drivers/gpio/gpio-i82=
55.c>
> > 3) (in the repo)
> > https://elixir.bootlin.com/linux/v6.4-rc4/source/drivers/gpio/gpio-104-=
idi-48.c <https://elixir.bootlin.com/linux/v6.4-rc4/source/drivers/gpio/gpi=
o-104-idi-48.c>
>
>
> > 2) & 3) were converted, so you may see by executing respective `git
> > log -p -- drivers/gpio/...`.
>
> Greetings Andy,
>
> Thank you for those links, I have observed the gpio_regmap code
> they have implemented in that case. It appears that the regmap
> code is opening the entire range of memory to be read. For my
> particular purpose I am not wanting to expose all the 0-0xff byte
> range of the GPIOs.

This is also supported by regmap (and regmap has caches for the sparse
registers as well).

> In my case is it still necessary to use the
> gpio_regmap code?

It does care about things the average GPIO controller driver needs to
repeat. So at least you may try and see how it will look.

> If gpio_regmap is required, how do I create a direct correlation
> between a specific gpio-line and a register offset? For example, in
> gpio-gxp-pl.c. Gpio-line at offset 0 (IOPLED) is at register 0x04. The
> gpio-line at offset 8 (FAN_INST) is at register 0x27.

You may remap registers. See, for example, gpio-pca953x, where some of
the registers (with high bit set) are actually virtual rather than
real offsets. Similar idea can be used in your case.

> Additionally, is it required to remove gpio_chip if gpio_regmap is
> used?

I don't think gpio_chip will be needed.

--=20
With Best Regards,
Andy Shevchenko