[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAMRc=Md2iFVMw14d4v-AjmcBniWF6r65VhcE5+HFsvH=VTdSxA@mail.gmail.com>
Date: Thu, 25 May 2023 16:35:23 +0200
From: Bartosz Golaszewski <brgl@...ev.pl>
To: Andy Shevchenko <andy.shevchenko@...il.com>
Cc: Chris Packham <Chris.Packham@...iedtelesis.co.nz>,
Kent Gibson <warthog618@...il.com>,
"linux-gpio@...r.kernel.org" <linux-gpio@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: using libgpiod to replace sysfs ABI (was Re: [PATCH] gpiolib:
Avoid side effects in gpio_is_visible())
On Thu, May 25, 2023 at 11:13 AM Andy Shevchenko
<andy.shevchenko@...il.com> wrote:
>
> On Thu, May 25, 2023 at 2:53 AM Chris Packham
> <Chris.Packham@...iedtelesis.co.nz> wrote:
> > On 24/05/23 17:41, Kent Gibson wrote:
>
> ...
>
> > It'd also be great if there was some way of ensuring that a line's state
> > is kept after the application has released the request (i.e. the txdis
> > case I mentioned). But that probably needs work on the kernel side to
> > make such guarantees.
>
> Won't happen. It will require too much of strictness to be added into
> the kernel with likely breakage of the existing code and
> documentation. What is being discussed is a D-Bus (like?) daemon +
> Policy in user space that will allow user / process / cgroup / etc to
> "own" the line and track its state.
>
It's already WiP[1]. I'm trying to keep the footprint minimal with
only GLib and dbus required at run-time.
Bart
[1] https://github.com/brgl/libgpiod-private/tree/topic/dbus
Powered by blists - more mailing lists