Re: [RFC] gpio: expose pull-up/pull-down line flags to userspace

[Bartosz Golaszewski <brgl@bgdev.pl>]

śr., 9 paź 2019 o 01:56 Kent Gibson <warthog618@gmail.com> napisał(a):
>
> On Wed, Oct 09, 2019 at 01:30:18AM +0200, Bartosz Golaszewski wrote:
> > śr., 9 paź 2019 o 01:21 Kent Gibson <warthog618@gmail.com> napisał(a):
> > >
> > > On Tue, Oct 08, 2019 at 10:56:18PM +0200, Bartosz Golaszewski wrote:
> > > > wt., 8 paź 2019 o 08:15 Kent Gibson <warthog618@gmail.com> napisał(a):
> > > > >
> > > > > On Sat, Sep 21, 2019 at 12:25:23PM +0200, Drew Fustini wrote:
> > > > > > Add pull-up/pull-down flags to the gpio line get and
> > > > > > set ioctl() calls.  Use cases include a push button
> > > > > > that does not have an external resistor.
> > > > > >
> > > > > > Addition use cases described by Limor Fried (ladyada) of
> > > > > > Adafruit in this PR for Adafruit_Blinka Python lib:
> > > > > > https://github.com/adafruit/Adafruit_Blinka/pull/59
> > > > > >
> > > > > > Signed-off-by: Drew Fustini <drew@pdp7.com>
> > > > > > ---
> > > > > >  drivers/gpio/gpiolib.c    | 12 ++++++++++++
> > > > > >  include/uapi/linux/gpio.h |  4 ++++
> > > > > >  2 files changed, 16 insertions(+)
> > > > > >
> > > > > > diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> > > > > > index d9074191edef..9da1093cc7f5 100644
> > > > > > --- a/drivers/gpio/gpiolib.c
> > > > > > +++ b/drivers/gpio/gpiolib.c
> > > > > > @@ -427,6 +427,8 @@ struct linehandle_state {
> > > > > >       (GPIOHANDLE_REQUEST_INPUT | \
> > > > > >       GPIOHANDLE_REQUEST_OUTPUT | \
> > > > > >       GPIOHANDLE_REQUEST_ACTIVE_LOW | \
> > > > > > +     GPIOHANDLE_REQUEST_PULL_UP | \
> > > > > > +     GPIOHANDLE_REQUEST_PULL_DOWN | \
> > > > > >       GPIOHANDLE_REQUEST_OPEN_DRAIN | \
> > > > > >       GPIOHANDLE_REQUEST_OPEN_SOURCE)
> > > > > >
> > > > > > @@ -598,6 +600,10 @@ static int linehandle_create(struct gpio_device *gdev, void __user *ip)
> > > > > >                       set_bit(FLAG_OPEN_DRAIN, &desc->flags);
> > > > > >               if (lflags & GPIOHANDLE_REQUEST_OPEN_SOURCE)
> > > > > >                       set_bit(FLAG_OPEN_SOURCE, &desc->flags);
> > > > > > +             if (lflags & GPIOHANDLE_REQUEST_PULL_DOWN)
> > > > > > +                     set_bit(FLAG_PULL_DOWN, &desc->flags);
> > > > > > +             if (lflags & GPIOHANDLE_REQUEST_PULL_UP)
> > > > > > +                     set_bit(FLAG_PULL_UP, &desc->flags);
> > > > > >
> > > > > >               ret = gpiod_set_transitory(desc, false);
> > > > > >               if (ret < 0)
> > > > > > @@ -1102,6 +1108,10 @@ static long gpio_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
> > > > > >               if (test_bit(FLAG_OPEN_SOURCE, &desc->flags))
> > > > > >                       lineinfo.flags |= (GPIOLINE_FLAG_OPEN_SOURCE |
> > > > > >                                          GPIOLINE_FLAG_IS_OUT);
> > > > > > +             if (test_bit(FLAG_PULL_DOWN, &desc->flags))
> > > > > > +                     lineinfo.flags |= GPIOLINE_FLAG_PULL_DOWN;
> > > > > > +             if (test_bit(FLAG_PULL_UP, &desc->flags))
> > > > > > +                     lineinfo.flags |= GPIOLINE_FLAG_PULL_UP;
> > > > > >
> > > > > >               if (copy_to_user(ip, &lineinfo, sizeof(lineinfo)))
> > > > > >                       return -EFAULT;
> > > > > > @@ -2475,6 +2485,8 @@ static bool gpiod_free_commit(struct gpio_desc *desc)
> > > > > >               clear_bit(FLAG_REQUESTED, &desc->flags);
> > > > > >               clear_bit(FLAG_OPEN_DRAIN, &desc->flags);
> > > > > >               clear_bit(FLAG_OPEN_SOURCE, &desc->flags);
> > > > > > +             clear_bit(FLAG_PULL_UP, &desc->flags);
> > > > > > +             clear_bit(FLAG_PULL_DOWN, &desc->flags);
> > > > > >               clear_bit(FLAG_IS_HOGGED, &desc->flags);
> > > > > >               ret = true;
> > > > > >       }
> > > > > > diff --git a/include/uapi/linux/gpio.h b/include/uapi/linux/gpio.h
> > > > > > index 4ebfe0ac6c5b..c2d1f7d908d6 100644
> > > > > > --- a/include/uapi/linux/gpio.h
> > > > > > +++ b/include/uapi/linux/gpio.h
> > > > > > @@ -33,6 +33,8 @@ struct gpiochip_info {
> > > > > >  #define GPIOLINE_FLAG_ACTIVE_LOW     (1UL << 2)
> > > > > >  #define GPIOLINE_FLAG_OPEN_DRAIN     (1UL << 3)
> > > > > >  #define GPIOLINE_FLAG_OPEN_SOURCE    (1UL << 4)
> > > > > > +#define GPIOLINE_FLAG_PULL_UP        (1UL << 5)
> > > > > > +#define GPIOLINE_FLAG_PULL_DOWN      (1UL << 6)
> > > > > >
> > > > > >  /**
> > > > > >   * struct gpioline_info - Information about a certain GPIO line
> > > > > > @@ -62,6 +64,8 @@ struct gpioline_info {
> > > > > >  #define GPIOHANDLE_REQUEST_ACTIVE_LOW        (1UL << 2)
> > > > > >  #define GPIOHANDLE_REQUEST_OPEN_DRAIN        (1UL << 3)
> > > > > >  #define GPIOHANDLE_REQUEST_OPEN_SOURCE       (1UL << 4)
> > > > > > +#define GPIOHANDLE_REQUEST_PULL_UP   (1UL << 5)
> > > > > > +#define GPIOHANDLE_REQUEST_PULL_DOWN (1UL << 6)
> > > > > >
> > > > > >  /**
> > > > > >   * struct gpiohandle_request - Information about a GPIO handle request
> > > > > > --
> > > > > > 2.20.1
> > > > > >
> > > > > Sorry for the late feedback, but I'm still not sure whether this patch
> > > > > is on track to be applied or not.  I had thought not, at least not yet,
> > > > > but just in case...
> > > > >
> > > >
> > > > It still needs some fixes, but Linus seems to be fine with the general idea.
> > > >
> > > > > You have added the flags to linehandle_create, but not lineevent_create.
> > > > > Given that your primary use case is push buttons, isn't the event request
> > > > > at least as important as the handle request?  Even ignoring your use
> > > > > case, I'd add them to lineevent_create just to maintain symmetry.
> > > > >
> > > > > Also, is it valid to set PULL_UP and PULL_DOWN simulaneously?
> > > > > I would think not, but I could be wrong.
> > > > > If not then that combination should be rejected with EINVAL.
> > > > >
> > > >
> > > > Yes, some validity checks of the flags must be added. I even did it in
> > > > my local branch[1].
> > > >
> > > Your changes for linehandle_create look ok, but for lineevent_create
> > > you explicitly disabled PULL_UP and PULL_DOWN, and in a block labelled
> > > "This is just wrong: we don't look for events on output lines" at that.
> > >
> >
> > Yeah, feel free to change it.
> >
> Yeah, you see I didn't feel free to change it, as I thought the
> appropriate etiquette was to send comments to the original author.
> At least until the patch has been applied.
> Or are RFCs a special case?
>

So either you review it and the author changes it and resends, or - if
for example you want to build on top of someone else's changes - you
take the patch, keep the original authorship, modify it and add your
signed-off-by tag with a small note on your changes in square
brackets.

> > > > Re: direction and configuration changes on requested lines: I think
> > > > it's quite useful to add in the form of a new ioctl():
> > > > GPIOHANDLE_SET_CONFIG_IOCTL or something. I started hacking on this
> > > > but eventually got more busy this week than I anticipated. I'm still
> > > > planning on sending an RFC by the end of the week though.
> > > >
> > > I have the reverse problem - bored and looking for something to do, so
> > > more than willing to help out - if you want it.
> > >
> >
> > Sure, in that case let's just wait for your patches. You can just
> > extend what I started, if you wish.
> >
> Yep, you sound tired.
> Are we talking about the SET_CONFIG here?  If you have a link to what
> you've started I'll take a look, but as I've already said - I'm not
> about to go writing new ioctls without approval and guidance from you
> and Linus.  If you have the ioctl defined I'm happy to take it from
> there. Well happy-ish.
>

But there's absolutely nothing wrong with writing code even if you're
not sure about the solution - it's even encouraged[1]. It's much
easier to discuss existing code than potential solutions.

[1] https://lkml.org/lkml/2000/8/25/132

> > > And while we're talking, does the gpio-mockup support pull up/down
> > > being set from the kernel side?
> > >
> >
> > No, but feel free to add it.
> >
> Ok, will have a look - cos I would like to add tests for the above and
> that will require the mockup supporting it as wel

Do you also plan on extending libgpiod by any chance?

Bart

>
> Cheers,
> Kent.
>