Re: [PATCH v5 07/12] leds: leds-gpio: Add support for GPIO descriptors

[Mika Westerberg <mika.westerberg@linux.intel.com>]

On Tue, Oct 28, 2014 at 10:56:09PM +0100, Rafael J. Wysocki wrote:
> On Tuesday, October 28, 2014 04:26:25 PM Linus Walleij wrote:
> > On Fri, Oct 17, 2014 at 2:11 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> > 
> > > From: Mika Westerberg <mika.westerberg@linux.intel.com>
> > >
> > > GPIO descriptors are the preferred way over legacy GPIO numbers
> > > nowadays. Convert the driver to use GPIO descriptors internally but
> > > still allow passing legacy GPIO numbers from platform data to support
> > > existing platforms.
> > >
> > > Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
> > > Acked-by: Alexandre Courbot <acourbot@nvidia.com>
> > > Acked-by: Bryan Wu <cooloney@gmail.com>
> > > Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > (...)
> > 
> > >         if (led_dat->blinking) {
> > > -               led_dat->platform_gpio_blink_set(led_dat->gpio,
> > > -                                                led_dat->new_level,
> > > -                                                NULL, NULL);
> > > +               int gpio = desc_to_gpio(led_dat->gpiod);
> > > +               int level = led_dat->new_level;
> > 
> > So this desc_to_gpio() is done only to call the legacy callback below?
> > 
> > > +               if (gpiod_is_active_low(led_dat->gpiod))
> > > +                       level = !level;
> > 
> > And that leads to making it necessary to have this helper variable
> > to invert the level since that callback does not pass a descriptor
> > (which would inherently know if it's active low)....
> > 
> > > +
> > > +               led_dat->platform_gpio_blink_set(gpio, level, NULL, NULL);
> > 
> > Is it *really* impossible to change all the users of this callback?
> 
> You said it could be done in a followup patch.  Here:
> http://marc.info/?l=linux-acpi&m=141154536921643&w=4
> 
> And Mika said he would add that to his TODO list:
> http://marc.info/?l=linux-acpi&m=141155173924101&w=4
> 
> I suppose that is still valid.

Yes, I'll just let dust to settle before sending out a patch that
converts the existing users of platform_gpio_blink_set() callback to
gpio descriptors.

> 
> > 
> > >                 led_dat->blinking = 0;
> > >         } else
> > > -               gpio_set_value_cansleep(led_dat->gpio, led_dat->new_level);
> > > +               gpiod_set_value_cansleep(led_dat->gpiod, led_dat->new_level);
> > 
> > (...)
> > >         /* Setting GPIOs with I2C/etc requires a task context, and we don't
> > >          * seem to have a reliable way to know if we're already in one; so
> > >          * let's just assume the worst.
> > > @@ -72,11 +73,16 @@ static void gpio_led_set(struct led_clas
> > >                 schedule_work(&led_dat->work);
> > >         } else {
> > >                 if (led_dat->blinking) {
> > > -                       led_dat->platform_gpio_blink_set(led_dat->gpio, level,
> > > -                                                        NULL, NULL);
> > > +                       int gpio = desc_to_gpio(led_dat->gpiod);
> > > +
> > > +                       if (gpiod_is_active_low(led_dat->gpiod))
> > > +                               level = !level;
> > > +
> > > +                       led_dat->platform_gpio_blink_set(gpio, level, NULL,
> > > +                                                        NULL);
> > 
> > Same comment.
> > 
> > > @@ -85,9 +91,10 @@ static int gpio_blink_set(struct led_cla
> > >  {
> > >         struct gpio_led_data *led_dat =
> > >                 container_of(led_cdev, struct gpio_led_data, cdev);
> > > +       int gpio = desc_to_gpio(led_dat->gpiod);
> > >
> > >         led_dat->blinking = 1;
> > > -       return led_dat->platform_gpio_blink_set(led_dat->gpio, GPIO_LED_BLINK,
> > > +       return led_dat->platform_gpio_blink_set(gpio, GPIO_LED_BLINK,
> > >                                                 delay_on, delay_off);
> > 
> > Same comment.
> > 
> > > @@ -97,24 +104,33 @@ static int create_gpio_led(const struct
> > >  {
> > >         int ret, state;
> > >
> > > -       led_dat->gpio = -1;
> > > +       if (!template->gpiod) {
> > > +               unsigned long flags = 0;
> > >
> > > -       /* skip leds that aren't available */
> > > -       if (!gpio_is_valid(template->gpio)) {
> > > -               dev_info(parent, "Skipping unavailable LED gpio %d (%s)\n",
> > > -                               template->gpio, template->name);
> > > -               return 0;
> > > +               /* skip leds that aren't available */
> > > +               if (!gpio_is_valid(template->gpio)) {
> > > +                       dev_info(parent, "Skipping unavailable LED gpio %d (%s)\n",
> > > +                                       template->gpio, template->name);
> > > +                       return 0;
> > > +               }
> > > +
> > > +               if (template->active_low)
> > > +                       flags |= GPIOF_ACTIVE_LOW;
> > > +
> > > +               ret = devm_gpio_request_one(parent, template->gpio, flags,
> > > +                                           template->name);
> > > +               if (ret < 0)
> > > +                       return ret;
> > > +
> > > +               led_dat->gpiod = gpio_to_desc(template->gpio);
> > > +               if (IS_ERR(led_dat->gpiod))
> > > +                       return PTR_ERR(led_dat->gpiod);
> > >         }
> > 
> > OK so this is the legacy codepath: point it out in a big fat
> > comment that this is the legacy codepath.
> 
> That looks like it could be done in a followup patch too.
> 
> Since the series is in my linux-next branch at this point, I really wouldn't
> like to reshuffle commits in it if that can be avoided.
> 
> > > Index: linux-pm/include/linux/leds.h
> > > ===================================================================
> > > --- linux-pm.orig/include/linux/leds.h
> > > +++ linux-pm/include/linux/leds.h
> > > @@ -251,6 +251,7 @@ struct gpio_led {
> > >         unsigned        retain_state_suspended : 1;
> > >         unsigned        default_state : 2;
> > >         /* default_state should be one of LEDS_GPIO_DEFSTATE_(ON|OFF|KEEP) */
> > > +       struct gpio_desc *gpiod;
> > 
> > Put the new struct member right below the current "gpio"
> > member,
> 
> It was done like that in previous versions, but turned out to cause problems
> to happen in testing.  Unfortunately, I don't seem to be able to find a pointer
> to the original report ATM, but perhaps Mika can.  Mika?

It is burried inside this thread:

http://www.spinics.net/lists/arm-kernel/msg369522.html