Re: [PATCH v3 2/4] gpiolib: Add ability to get GPIO pin direction

[Uwe Kleine-König]

Hello Peter,

On Fri, Feb 18, 2011 at 11:36:26AM -0600, Peter Tyser wrote:
> On Fri, 2011-02-18 at 09:57 +0100, Uwe Kleine-König wrote:
> > On Thu, Feb 17, 2011 at 05:03:17PM -0600, Peter Tyser wrote:
> > > Add a new get_direction() function to the gpio_chip structure.  This is
> > > useful so that the direction of a pin can be determined when its
> > > initially exported.  Previously, the direction defaulted to "unknown"
> > > regardless of the actual configuration of the GPIO pin.
> > > 
> > > If a GPIO driver implements get_direction(), it is called in
> > > gpio_request() to set the initial direction of the pin accurately.
> > IMHO the commit log is conceptually wrong, because it talks about a
> > "pin".  Better use gpio here.
> 
> I don't follow.  I used "pin" to make it clear that the get_direction()
> function operates on a pin-by-pin basis, and to help reduce any
> ambiguity about if a gpio chip or gpio pin is being referred to.  Would
> you prefer that the "pin" references are clarified to be "GPIO pin"?
Maybe it's just that we use different terms.  For me a "pin" is an
entry/exit point into/from a cpu or other chip.  A gpio is (hmm, how
should I say) a concept how to drive a pin.  A gpio might or might not
be "connected" to a pin at a given time.

> > > Cc: Alek Du <alek.du@xxxxxxxxx>
> > > Cc: Samuel Ortiz <sameo@xxxxxxxxxxxxxxx>
> > > Cc: David Brownell <dbrownell@xxxxxxxxxxxxxxxxxxxxx>
> > > Cc: Eric Miao <eric.y.miao@xxxxxxxxx>
> > > Cc: Uwe Kleine-K?nig <u.kleine-koenig@xxxxxxxxxxxxxx>
> > > Cc: Mark Brown <broonie@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
> > > Cc: Joe Perches <joe@xxxxxxxxxxx>
> > > Cc: Alan Cox <alan@xxxxxxxxxxxxxxxxxxx>
> > > Cc: Grant Likely <grant.likely@xxxxxxxxxxxx>
> > > Signed-off-by: Peter Tyser <ptyser@xxxxxxxxxxx>
> > > ---
> > > Changes since v1:
> > > - Add support for "unknown" direction
> > > 
> > > Changes since v2:
> > >   Based on Wolfram's feedback:
> > > - Use GPIOF_DIR_* flags as returns from get_direction()
> > > - Call spin_lock_irqsave() to before setting flags
> > > 
> > >  drivers/gpio/gpiolib.c     |   23 +++++++++++++++++++++++
> > >  include/asm-generic/gpio.h |    4 ++++
> > >  2 files changed, 27 insertions(+), 0 deletions(-)
> > > 
> > > diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> > > index eb74311..a656a2c 100644
> > > --- a/drivers/gpio/gpiolib.c
> > > +++ b/drivers/gpio/gpiolib.c
> > > @@ -1174,6 +1174,7 @@ int gpio_request(unsigned gpio, const char *label)
> > >  	struct gpio_desc	*desc;
> > >  	struct gpio_chip	*chip;
> > >  	int			status = -EINVAL;
> > > +	int			dir;
> > >  	unsigned long		flags;
> > >  
> > >  	spin_lock_irqsave(&gpio_lock, flags);
> > > @@ -1214,6 +1215,28 @@ int gpio_request(unsigned gpio, const char *label)
> > >  		}
> > >  	}
> > >  
> > > +	if (chip->get_direction) {
> > > +		/* chip->get_direction may sleep */
> > > +		spin_unlock_irqrestore(&gpio_lock, flags);
> > might_sleep_if(chip->can_sleep) ?
> 
> Makes sense.  I was following the lead of chip->request() in the same
> function, which doesn't use might_sleep_if().  I assume might_sleep_if()
> should be added to it as well in a separate patch?
I was there first :-)
http://mid.gmane.org/1297977533-17794-1-git-send-email-u.kleine-koenig@xxxxxxxxxxxxxx

> > > +		dir = chip->get_direction(chip, gpio - chip->base);
> > > +		spin_lock_irqsave(&gpio_lock, flags);
> > > +		switch (dir) {
> > > +		case GPIOF_DIR_OUT:
> > > +			set_bit(FLAG_DIR_OUT, &desc->flags);
> > > +			clear_bit(FLAG_DIR_IN, &desc->flags);
> > > +			break;
> > > +		case GPIOF_DIR_IN:
> > > +			set_bit(FLAG_DIR_IN, &desc->flags);
> > > +			clear_bit(FLAG_DIR_OUT, &desc->flags);
> > > +			break;
> > > +		default:
> > > +			/* Direction isn't known */
> > > +			clear_bit(FLAG_DIR_OUT, &desc->flags);
> > > +			clear_bit(FLAG_DIR_IN, &desc->flags);
> > > +			break;
> > > +		}
> > Alternatively to my suggestion for patch1:
> > 	} else {
> > 		clear_bit(FLAG_DIR_OUT, &desc->flags);
> > 		clear_bit(FLAG_DIR_IN, &desc->flags);
> 
> I like this way better too.  I'll initialize dir = -1 and pull the
> switch statement out of the conditional, like:
> 
> 	if (chip->get_direction) {
> 		/* chip->get_direction may sleep */
> 		spin_unlock_irqrestore(&gpio_lock, flags);
> 		might_sleep_if(chip->can_sleep);
> 		dir = chip->get_direction(chip, gpio - chip->base);
> 		spin_lock_irqsave(&gpio_lock, flags);
> 	}
> 
> 	switch (dir) {
> 	case GPIOF_DIR_OUT:
> 		set_bit(FLAG_DIR_OUT, &desc->flags);
> 		clear_bit(FLAG_DIR_IN, &desc->flags);
> 		break;
> 	case GPIOF_DIR_IN:
> 		set_bit(FLAG_DIR_IN, &desc->flags);
> 		clear_bit(FLAG_DIR_OUT, &desc->flags);
> 		break;
> 	default:
> 		/* Direction isn't known */
> 		clear_bit(FLAG_DIR_OUT, &desc->flags);
> 		clear_bit(FLAG_DIR_IN, &desc->flags);
> 		break;
> 	}
fine

Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/