linux-kernel - Re: [PATCH v4 8/8] gpiolib: notify user-space about in-kernel line state changes

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAMRc=MfrGynY0J6ozG0B9GiPJgqLJywaT-Fw0_O3zZNKrsALAA@mail.gmail.com>
Date: Thu, 17 Oct 2024 17:04:13 +0200
From: Bartosz Golaszewski <brgl@...ev.pl>
To: Kent Gibson <warthog618@...il.com>
Cc: Linus Walleij <linus.walleij@...aro.org>, linux-gpio@...r.kernel.org, 
	linux-kernel@...r.kernel.org, 
	Bartosz Golaszewski <bartosz.golaszewski@...aro.org>
Subject: Re: [PATCH v4 8/8] gpiolib: notify user-space about in-kernel line
 state changes

On Thu, Oct 17, 2024 at 5:02 PM Kent Gibson <warthog618@...il.com> wrote:
>
> On Thu, Oct 17, 2024 at 04:59:46PM +0200, Bartosz Golaszewski wrote:
> > On Thu, Oct 17, 2024 at 4:20 PM Kent Gibson <warthog618@...il.com> wrote:
> > >
> > > On Thu, Oct 17, 2024 at 04:14:24PM +0200, Bartosz Golaszewski wrote:
> > > > On Thu, Oct 17, 2024 at 2:53 PM Kent Gibson <warthog618@...il.com> wrote:
> > > > >
> > > > > On Thu, Oct 17, 2024 at 10:14:16AM +0200, Bartosz Golaszewski wrote:
> > > > > > From: Bartosz Golaszewski <bartosz.golaszewski@...aro.org>
> > > > > >
> > > > > > @@ -1447,8 +1450,6 @@ static long linereq_set_config(struct linereq *lr, void __user *ip)
> > > > > >               }
> > > > > >
> > > > > >               WRITE_ONCE(line->edflags, edflags);
> > > > > > -
> > > > > > -             gpiod_line_state_notify(desc, GPIO_V2_LINE_CHANGED_CONFIG);
> > > > > >       }
> > > > > >       return 0;
> > > > > >  }
> > > > >
> > > > > I still get errors from this when reconfiguring lines with debounce.
> > > > > You should leave this notify in place and use _nonotify when setting the
> > > > > direction.
> > > > > i.e.
> > > > >
> > > > > @@ -1436,11 +1432,11 @@ static long linereq_set_config(struct linereq *lr, void __user *ip)
> > > > >                         int val = gpio_v2_line_config_output_value(&lc, i);
> > > > >
> > > > >                         edge_detector_stop(line);
> > > > > -                       ret = gpiod_direction_output(desc, val);
> > > > > +                       ret = gpiod_direction_output_nonotify(desc, val);
> > > > >                         if (ret)
> > > > >                                 return ret;
> > > > >                 } else {
> > > > > -                       ret = gpiod_direction_input(desc);
> > > > > +                       ret = gpiod_direction_input_nonotify(desc);
> > > > >                         if (ret)
> > > > >                                 return ret;
> > > > >
> > > > > @@ -1450,6 +1446,8 @@ static long linereq_set_config(struct linereq *lr, void __user *ip)
> > > > >                 }
> > > > >
> > > > >                 WRITE_ONCE(line->edflags, edflags);
> > > > > +
> > > > > +               gpiod_line_state_notify(desc, GPIO_V2_LINE_CHANGED_CONFIG);
> > > > >         }
> > > > >         return 0;
> > > > >  }
> > > > >
> > > > > Given that, all my current tests are passing.
> > > > >
> > > >
> > > > That looks good - after all we no longer notify from any place in
> > > > gpiolib-cdev.c anymore - but I'd like to learn what's wrong exactly.
> > > > Are you getting more events with debounce? Are you not getting any?
> > > >
> > >
> > > In linereq_set_config(), the notify comes from the gpiod_direction_input() -
> > > before the edge_detector_setup() is called (not visible in the patch) and that
> > > sets the debounce value in the desc.
> > > So you get an event without the debounce set, or with a stale value.
> > > Keeping the gpiod_line_state_notify() and using the _nonotify()
> > > functions means the notify comes after the debounce has been set.
> > >
> >
> > I see. I guess I should do the same both for linehandle_set_config()
> > and linereq_set_config()?
> >
>
> linehandles don't support debounce, so it's good as is.
>

Right, but I'm wondering if it isn't better for consistency.
Otherwise, someone may ask themselves why there's no event being
emitted if they're not familiar with the gpiod_direction_*()
internals.

Bart