linux-kernel - Re: [PATCH] Input: gpio_keys_polled - always use gpiod_get_value

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] [day] [month] [year] [list]

Message-ID: <20161020215301.GA31116@dtor-ws>
Date:   Thu, 20 Oct 2016 14:53:01 -0700
From:   Dmitry Torokhov <dmitry.torokhov@...il.com>
To:     Mika Westerberg <mika.westerberg@...ux.intel.com>
Cc:     linux-input@...r.kernel.org, linux-kernel@...r.kernel.org,
        Hans de Goede <hdegoede@...hat.com>,
        Geert Uytterhoeven <geert+renesas@...der.be>,
        Aaron Lu <aaron.lu@...el.com>,
        Linus Walleij <linus.walleij@...aro.org>
Subject: Re: [PATCH] Input: gpio_keys_polled - always use
 gpiod_get_value_cansleep

On Thu, Oct 20, 2016 at 01:45:19PM +0300, Mika Westerberg wrote:
> On Wed, Oct 19, 2016 at 04:41:07PM -0700, Dmitry Torokhov wrote:
> > It does not matter if given GPIO may sleep or not when reading state,
> > polling is always done in a non-atomic context, so we should always
> > be able to simply use gpiod_get_value_cansleep().
> > 
> > Also let's note in the logs when we fail to read gpio state.
> > 
> > Signed-off-by: Dmitry Torokhov <dmitry.torokhov@...il.com>
> > ---
> >  drivers/input/keyboard/gpio_keys_polled.c | 21 ++++++++++-----------
> >  1 file changed, 10 insertions(+), 11 deletions(-)
> > 
> > diff --git a/drivers/input/keyboard/gpio_keys_polled.c b/drivers/input/keyboard/gpio_keys_polled.c
> > index daef8ea..3c79158 100644
> > --- a/drivers/input/keyboard/gpio_keys_polled.c
> > +++ b/drivers/input/keyboard/gpio_keys_polled.c
> > @@ -34,7 +34,6 @@ struct gpio_keys_button_data {
> >  	int last_state;
> >  	int count;
> >  	int threshold;
> > -	int can_sleep;
> >  };
> >  
> >  struct gpio_keys_polled_dev {
> > @@ -76,16 +75,17 @@ static void gpio_keys_polled_check_state(struct input_polled_dev *dev,
> >  {
> >  	int state;
> >  
> > -	if (bdata->can_sleep)
> > -		state = !!gpiod_get_value_cansleep(bdata->gpiod);
> > -	else
> > -		state = !!gpiod_get_value(bdata->gpiod);
> > -
> > -	gpio_keys_button_event(dev, button, state);
> > +	state = gpiod_get_value_cansleep(bdata->gpiod);
> > +	if (unlikely(state < 0)) {
> 
> Is this unlikely() really bringing any performance benefits here?

Not really, I dropped it.

> 
> Otherwise this patch looks good to me (sans the below line which you
> mentioned in your followup email).
> 
> > +		dev_err(input->dev.parent,
> > +			"failed to get gpio state: %d\n", state);
> > +	} else {
> > +		gpio_keys_button_event(dev, button, state);
> >  
> > -	if (state != bdata->last_state) {
> > -		bdata->count = 0;
> > -		bdata->last_state = state;
> > +		if (state != bdata->last_state) {
> > +			bdata->count = 0;
> > +			bdata->last_state = state;
> > +		}
> >  	}
> >  }
> >  
> > @@ -341,7 +341,6 @@ static int gpio_keys_polled_probe(struct platform_device *pdev)
> >  			}
> >  		}
> >  
> > -		bdata->can_sleep = gpiod_cansleep(bdata->gpiod);
> >  		bdata->last_state = -1;
> >  		bdata->threshold = DIV_ROUND_UP(button->debounce_interval,
> >  						pdata->poll_interval);
> > -- 
> > 2.8.0.rc3.226.g39d4020
> > 
> > 
> > -- 
> > Dmitry

-- 
Dmitry