linux-acpi.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Andy Shevchenko <andy.shevchenko@gmail.com>
To: Flavio Suligoi <f.suligoi@asem.it>
Cc: Bartosz Golaszewski <brgl@bgdev.pl>,
	Andy Shevchenko <andriy.shevchenko@linux.intel.com>,
	"linux-gpio@vger.kernel.org" <linux-gpio@vger.kernel.org>,
	"linux-acpi@vger.kernel.org" <linux-acpi@vger.kernel.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: How to use an ACPI declared GPIO in a userspace ...
Date: Tue, 29 Sep 2020 19:27:26 +0300	[thread overview]
Message-ID: <CAHp75Vdd2QjvJvLGHa1x=RaSknEG+O+YB4eJA6+2htnZ=Gf52g@mail.gmail.com> (raw)
In-Reply-To: <feb8567c830748c483c8c66dd4717003@asem.it>

On Tue, Sep 29, 2020 at 7:21 PM Flavio Suligoi <f.suligoi@asem.it> wrote:
> > > > I need to expose to the userspace a GPIO, physically connected to a
> > board
> > > > push-button. This GPIO must expose a pre-defined name, such as
> > > > "user-push-button", so that the userspace applications can use it
> > without
> > > > know any physical GPIO details.
> > > >
> > > > I can customize the board BIOS and so my goal is to add an ACPI table
> > with
> > > > a content like this:
> > > >
> > > > ...
> > > > Scope (\_SB.GPO1)
> > > >         {
> > > >                 Device (BTNS)
> > > >                 {
> > > >                         Name (_HID, "PRP0001")
> > > >                         Name (_DDN, "GPIO buttons device")
> > > >
> > > >                         Name (_CRS, ResourceTemplate ()
> > > >                         {
> > > >                                 GpioIo (
> > > >                                 Exclusive,               // Not shared
> > > >                                 PullNone,                // No need
> > for pulls
> > > >                                 0,                       // Debounce
> > timeout
> > > >                                 0,                       // Drive
> > strength
> > > >                                 IoRestrictionInputOnly,  // Only used
> > as input
> > > >                                 "\\_SB.GPO1",            // GPIO
> > controller
> > > >                                 0, ResourceConsumer, , ) // Must be 0
> > > >                                 {
> > > >                                         25,              // GPIO
> > number
> > > >                                 }
> > > > ...
> > > >
> > > > I know that this GPIO can be used from other drivers.
> > > > For example I successfully tested it using the "gpio-keys" device
> > driver,
> > > > giving to my GPIO a key-code and emulating in this way a keyboard key.
> > > > This could be a possible solution.
> > > >
> > > > But I prefer to expose my GPIO as a classic GPIO, not as a keyboard
> > key.
> > > >
> > > > I was wondering if there is a generic GPIO driver that I can use to
> > expose
> > > > this GPIO with its pre-defined name (caming from the ACPI table
> > declaration),
> > > > to the userspace...
> >
> > Unfortunately what you are describing in the second part is rather
> > property of the controller which can hog the line, but this is not
> > what you want in the first part.
> > The Linux kernel, in many ways, is designed that you need a driver
> > (I²C user space device node is rather a mistake, but compromise for
> > that time when most of the devices have access from user space
> > drivers). So, the proper way is to define this as gpio-keys (either
> > interrupt version or polling one) and connect a listener to the event.
> >
> > Summarize: you need to describe pin(s) via "gpio-line-names" property
> > of the controller (it's not so easy task if ACPI tables already have
> > parts of it, but I think your case should be feasible). And either
> > provide a gpio-keys device, or use line directly by name as (libgpiod
> > example):
> >  gpiodetect
> >  gpioinfo gpiochipX
> >  gpiofind $GPIO_LINE_NAME
> >  gpiomon gpiochipX $(gpiofind $GPIO_LINE_NAME) &
> >
> > Examples of ACPI are here [1] for controller part (look at the name
> > list) and for device part [2]. You may look into other folders as
> > well, though they are not so reach of examples.
> >
> > [1]: https://github.com/westeri/meta-acpi/blob/master/recipes-bsp/acpi-
> > tables/samples/edison/arduino.asli
> > [2]: https://github.com/westeri/meta-acpi/blob/master/recipes-bsp/acpi-
> > tables/samples/edison/buttons.asli
>
> I have already written and ACPI table, not in the BIOS but as separate
> SSDT, loaded manually at runtime, using the gpio-keys (with interrupt)
> and in this way all works good. So I have already tested this solution.
>
> But I prefer obtain this result in the classic way, with GPIO...
>
> So I think I'll write a device driver for it. A device driver which
> reads the ACPI table and publishes the GPIO, with its name, in sysfs...

Maybe I was not so clear, but as Bart mentioned the least you can do
is simply define line name via "gpio-line-names" property. The problem
here is when and how you would like to have them incorporated.
When: if ACPI tables are being provided by firmware which you may not
alter, then you must use initramfs type of solution (no configfs,
don't know about EFI var though). How: In that case you might have a
chance to incorporate _DSD() method into *existing* _CRS() one.
Possible impediments: if ACPI table from firmware already has a _DSD()
defined or above is not working for some reason. In such a case you
must upgrade entire DSDT via initramfs.

> > > Adding Andy who knows ACPI GPIO well.
> >
> > Thanks.
> >
> > > In general, the "gpio-line-names" property is used for that and it's
> > > supported both for device tree as well as ACPI, although I have only
> > > ever used the former.
> >
> > Right. ACPI supports properties via _DSD() method.

-- 
With Best Regards,
Andy Shevchenko

  reply	other threads:[~2020-09-29 16:27 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-09-29 15:37 How to use an ACPI declared GPIO in a userspace Flavio Suligoi
2020-09-29 15:47 ` Bartosz Golaszewski
2020-09-29 16:10   ` Andy Shevchenko
2020-09-29 16:21     ` Flavio Suligoi
2020-09-29 16:27       ` Andy Shevchenko [this message]
2020-09-30 12:04         ` Flavio Suligoi
2020-09-30 13:01           ` Andy Shevchenko
2020-09-30 15:39             ` Flavio Suligoi
2020-09-30 15:54               ` Andy Shevchenko
2020-09-30 16:10                 ` Andy Shevchenko
2020-10-02 10:02                   ` Flavio Suligoi
2020-10-02 12:48                     ` Andy Shevchenko
2020-10-02 13:23                       ` Andy Shevchenko
2020-10-02 13:29                         ` Flavio Suligoi
2020-10-02 10:26                   ` Flavio Suligoi
2020-10-02 12:40                     ` Andy Shevchenko
2020-09-29 16:10   ` Flavio Suligoi

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAHp75Vdd2QjvJvLGHa1x=RaSknEG+O+YB4eJA6+2htnZ=Gf52g@mail.gmail.com' \
    --to=andy.shevchenko@gmail.com \
    --cc=andriy.shevchenko@linux.intel.com \
    --cc=brgl@bgdev.pl \
    --cc=f.suligoi@asem.it \
    --cc=linux-acpi@vger.kernel.org \
    --cc=linux-gpio@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).