public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Greg KH <gregkh@suse.de>
To: Kay Sievers <kay.sievers@vrfy.org>
Cc: Jani Nikula <ext-jani.1.nikula@nokia.com>,
	"dbrownell@users.sourceforge.net"
	<dbrownell@users.sourceforge.net>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"dsilvers@simtec.co.uk" <dsilvers@simtec.co.uk>,
	"ben@simtec.co.uk" <ben@simtec.co.uk>,
	"Bityutskiy Artem (Nokia-D/Helsinki)"
	<Artem.Bityutskiy@nokia.com>,
	"akpm@linux-foundation.org" <akpm@linux-foundation.org>
Subject: Re: [PATCH 2/3] gpiolib: add support for having symlinks under gpio class directory
Date: Thu, 10 Dec 2009 07:24:52 -0800	[thread overview]
Message-ID: <20091210152452.GA13519@suse.de> (raw)
In-Reply-To: <ac3eb2510912100717p11e3c515l2d9155a764befc71@mail.gmail.com>

On Thu, Dec 10, 2009 at 04:17:00PM +0100, Kay Sievers wrote:
> On Thu, Dec 10, 2009 at 15:49, Greg KH <gregkh@suse.de> wrote:
> > On Thu, Dec 10, 2009 at 04:32:17PM +0200, Jani Nikula wrote:
> >> We simply decided to extend gpio_export_link(). When it is called
> >> with 'dev == NULL' (no driver), it creates an additional symlink
> >> in /sys/class/gpio/
> >>
> >> # ls -l /sys/class/gpio/
> >> --w-------    1 root     0            4096 Jan  1 00:00 export
> >> lrwxrwxrwx    1 root     0               0 Jan  1 00:00 power_button -> ../../devices/virtual/gpio/gpio25
> >> lrwxrwxrwx    1 root     0               0 Jan  1 00:00 gpio25 -> ../../devices/virtual/gpio/gpio25
> >> lrwxrwxrwx    1 root     0               0 Jan  1 00:00 gpio38 -> ../../devices/virtual/gpio/gpio38
> >> ...
> >>
> >> This is exactly what our patchset allows to do.
> >>
> >> An alternative would be a dummy driver just to create a home in sysfs
> >> for the standalone GPIOs and use the gpio_export_link() to make the
> >> links there.
> >>
> >> Any other suggestions?
> 
> We can not do anything like that. The symlink name must always match
> the name in /sys/devices/, and it must be a 1:1 relation. We can not
> allow anything else, it confuses tools which rightfully expect this.
> 
> You could convert the class to a "bus", the /sys/class layout is
> pretty much broken by design, and should better be avoided. It's not
> extensible, flat in its layout, and today just a too dumb version of
> the "bus" subsystem which does not have all these limitations.
> 
> All the device links would show up in /sys/bus/gpio/devices/, here you
> have the same rules as in /sys/class/: no additional symlinks, the
> same names as in /sys/devices -- but you could add a custom folder to
> the  /sys/bus/gpio/ directory where you could do whatever fits your
> needs. :)

Ah, yeah, good point, the pci "slots" directory does something like this
today, with symlinks pointing out of that directory to other devices.

It's messy, but it somehow works...

thanks,

greg k-h

  reply	other threads:[~2009-12-10 15:25 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-12-09 13:49 [PATCH 0/3] gpiolib: gpio naming in sysfs Jani Nikula
2009-12-09 13:49 ` [PATCH 1/3] device class: add symlink creation helpers Jani Nikula
2009-12-10  2:49   ` Greg KH
2009-12-09 13:49 ` [PATCH 2/3] gpiolib: add support for having symlinks under gpio class directory Jani Nikula
2009-12-10  2:48   ` Greg KH
2009-12-10 14:32     ` Jani Nikula
2009-12-10 14:49       ` Greg KH
2009-12-10 15:17         ` Kay Sievers
2009-12-10 15:24           ` Greg KH [this message]
2009-12-11  8:41         ` Jani Nikula
2009-12-11 15:38           ` Greg KH
2009-12-11  3:35       ` David Brownell
2009-12-09 13:49 ` [PATCH 3/3] gpiolib: use chip->names for symlinks, always use gpioN for device names Jani Nikula
2009-12-11  3:39   ` David Brownell
2009-12-11  3:47     ` Greg KH
2009-12-11  4:13       ` David Brownell
2009-12-11  4:38         ` Greg KH
2009-12-11  5:13           ` David Brownell
2009-12-11  5:18             ` Greg KH
2009-12-11  5:36           ` Artem Bityutskiy
2009-12-11  5:46             ` Greg KH
2009-12-11  7:51               ` Artem Bityutskiy
2009-12-11 15:36                 ` Greg KH
2009-12-11 13:23             ` [PATCH]crypto: Fix complain about lack test for internal used algorithm Youquan,Song
2009-12-11  6:04               ` Herbert Xu
2009-12-19  9:40                 ` Youquan,Song
2009-12-19  2:29                   ` Herbert Xu
2009-12-19 15:07                     ` Youquan,Song
2009-12-19  9:42                       ` Herbert Xu
2009-12-21 10:38                         ` [Resend PATCH]crypto: " Youquan,Song
2009-12-23 11:59                           ` Herbert Xu
2009-12-11  5:22         ` [PATCH 3/3] gpiolib: use chip->names for symlinks, always use gpioN for device names Ben Nizette
2009-12-11  5:12       ` Ben Nizette
2009-12-14 11:16         ` Jani Nikula
2009-12-14 22:27           ` Ben Nizette
2009-12-10  0:02 ` [PATCH 0/3] gpiolib: gpio naming in sysfs Andrew Morton

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=20091210152452.GA13519@suse.de \
    --to=gregkh@suse.de \
    --cc=Artem.Bityutskiy@nokia.com \
    --cc=akpm@linux-foundation.org \
    --cc=ben@simtec.co.uk \
    --cc=dbrownell@users.sourceforge.net \
    --cc=dsilvers@simtec.co.uk \
    --cc=ext-jani.1.nikula@nokia.com \
    --cc=kay.sievers@vrfy.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