Re: [PATCH 0/6] GPIO character device skeleton

[Johan Hovold <johan@kernel.org>]

On Thu, Nov 05, 2015 at 03:11:48PM +0100, Linus Walleij wrote:
> On Thu, Nov 5, 2015 at 10:40 AM, Johan Hovold <johan@kernel.org> wrote:
> > On Tue, Nov 03, 2015 at 06:18:42PM +0100, Linus Walleij wrote:
> 
> >> It'd be nice if devices could be uniquely identified in the chardev alone
> >> I think, then we don't need to much reliance on external assumptions
> >> and traversing sysfs somehow for more info.
> >
> > Point is that this is not a gpio-specific problem. Userspace needs to
> > deal with this for any resource it wants to access (e.g. i2c or spi).
> 
> Yeah let's see what others have to say.
> 
> > And you should be able to use libudev for tree-traversal without actually
> > using udevd, right?
> 
> Well libudev depends on libsystemd and is using it's subcomponent
> sd-device to traverse sysfs. It's a pretty fat dependency for a
> space-constrained embedded system, and the source code is not
> maintained outside systemd anymore, leading to an
> all-or-nothing approach it seems to me.

Seems to me like you're trying to solve a user-space problem in the
kernel. Even the network-device links you mentioned is something that
udev provides.

> I don't think it used to be like that but it is like that now...

Yeah, I'm not using the latest and greatest, but the libudev0 182 I took
a look at has no such dependencies and the lib seemed reasonably small
(40k).

IIRC there has been discussions about a standalone (mdev) version of it
as well.

Johan