[PATCH] gpio: samsung: Fix off-by-one bug in gpio addresses

[PATCH] gpio: samsung: Fix off-by-one bug in gpio addresses

[Sean Paul]

Move gpc4 to the end of the automatically processed gpio controllers so
we don't taint the automatic offset calculation.

This bug caused all controllers coming after gpc4 to map to the
incorrect address. The result is <&gpd1 0 0 0 0> would actually map to
GPIO 0 in gpd0.

Signed-off-by: Sean Paul <seanpaul@chromium.org>
---
 drivers/gpio/gpio-samsung.c |   14 +++++++-------
 1 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/drivers/gpio/gpio-samsung.c b/drivers/gpio/gpio-samsung.c
index b6453d0..e4599b4 100644
--- a/drivers/gpio/gpio-samsung.c
+++ b/drivers/gpio/gpio-samsung.c
@@ -2454,12 +2454,6 @@ static struct samsung_gpio_chip exynos5_gpios_1[] = {
 		},
 	}, {
 		.chip	= {
-			.base	= EXYNOS5_GPC4(0),
-			.ngpio	= EXYNOS5_GPIO_C4_NR,
-			.label	= "GPC4",
-		},
-	}, {
-		.chip	= {
 			.base	= EXYNOS5_GPD0(0),
 			.ngpio	= EXYNOS5_GPIO_D0_NR,
 			.label	= "GPD0",
@@ -2513,6 +2507,12 @@ static struct samsung_gpio_chip exynos5_gpios_1[] = {
 			.label	= "GPY6",
 		},
 	}, {
+		.chip	= {
+			.base	= EXYNOS5_GPC4(0),
+			.ngpio	= EXYNOS5_GPIO_C4_NR,
+			.label	= "GPC4",
+		},
+	}, {
 		.config	= &samsung_gpio_cfgs[9],
 		.irq_base = IRQ_EINT(0),
 		.chip	= {
@@ -2833,7 +2833,7 @@ static __init void exynos5_gpiolib_init(void)
 	}
 
 	/* need to set base address for gpc4 */
-	exynos5_gpios_1[11].base = gpio_base1 + 0x2E0;
+	exynos5_gpios_1[20].base = gpio_base1 + 0x2E0;
 
 	/* need to set base address for gpx */
 	chip = &exynos5_gpios_1[21];
-- 
1.7.7.3

Re: [PATCH] gpio: samsung: Fix off-by-one bug in gpio addresses

[Doug Anderson]

Acked-by: Doug Anderson <dianders@chromium.org>

---

On Fri, Jul 20, 2012 at 1:58 PM, Sean Paul <seanpaul@chromium.org> wrote:
>
> Move gpc4 to the end of the automatically processed gpio controllers so
> we don't taint the automatic offset calculation.
>
> This bug caused all controllers coming after gpc4 to map to the
> incorrect address. The result is <&gpd1 0 0 0 0> would actually map to
> GPIO 0 in gpd0.
>
> Signed-off-by: Sean Paul <seanpaul@chromium.org>
> ---
>  drivers/gpio/gpio-samsung.c |   14 +++++++-------
>  1 files changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/gpio/gpio-samsung.c b/drivers/gpio/gpio-samsung.c
> index b6453d0..e4599b4 100644
> --- a/drivers/gpio/gpio-samsung.c
> +++ b/drivers/gpio/gpio-samsung.c
> @@ -2454,12 +2454,6 @@ static struct samsung_gpio_chip exynos5_gpios_1[] = {
>                 },
>         }, {
>                 .chip   = {
> -                       .base   = EXYNOS5_GPC4(0),
> -                       .ngpio  = EXYNOS5_GPIO_C4_NR,
> -                       .label  = "GPC4",
> -               },
> -       }, {
> -               .chip   = {
>                         .base   = EXYNOS5_GPD0(0),
>                         .ngpio  = EXYNOS5_GPIO_D0_NR,
>                         .label  = "GPD0",
> @@ -2513,6 +2507,12 @@ static struct samsung_gpio_chip exynos5_gpios_1[] = {
>                         .label  = "GPY6",
>                 },
>         }, {
> +               .chip   = {
> +                       .base   = EXYNOS5_GPC4(0),
> +                       .ngpio  = EXYNOS5_GPIO_C4_NR,
> +                       .label  = "GPC4",
> +               },
> +       }, {
>                 .config = &samsung_gpio_cfgs[9],
>                 .irq_base = IRQ_EINT(0),
>                 .chip   = {
> @@ -2833,7 +2833,7 @@ static __init void exynos5_gpiolib_init(void)
>         }
>
>         /* need to set base address for gpc4 */
> -       exynos5_gpios_1[11].base = gpio_base1 + 0x2E0;
> +       exynos5_gpios_1[20].base = gpio_base1 + 0x2E0;
>
>         /* need to set base address for gpx */
>         chip = &exynos5_gpios_1[21];
> --
> 1.7.7.3
>

Re: [PATCH] gpio: samsung: Fix off-by-one bug in gpio addresses

[Linus Walleij]

On Fri, Jul 20, 2012 at 10:58 PM, Sean Paul <seanpaul@chromium.org> wrote:

> Move gpc4 to the end of the automatically processed gpio controllers so
> we don't taint the automatic offset calculation.
>
> This bug caused all controllers coming after gpc4 to map to the
> incorrect address. The result is <&gpd1 0 0 0 0> would actually map to
> GPIO 0 in gpd0.
>
> Signed-off-by: Sean Paul <seanpaul@chromium.org>

Samsung people, please comment on this patch.

Yours,
Linus Walleij

RE: [PATCH] gpio: samsung: Fix off-by-one bug in gpio addresses

[박상수]

> -----Original Message-----
> From: dianders@google.com [mailto:dianders@google.com] On Behalf Of Doug Anderson
> Sent: Saturday, July 21, 2012 6:07 AM
> To: Sean Paul
> Cc: grant.likely@secretlab.ca; linus.walleij@stericsson.com; linux-kernel@vger.kernel.org;
> olofj@chromium.org; Sangsu Park; Thomas Abraham; Kukjin Kim
> Subject: Re: [PATCH] gpio: samsung: Fix off-by-one bug in gpio addresses
> 
> Acked-by: Doug Anderson <dianders@chromium.org>
> 
> ---
> 
> On Fri, Jul 20, 2012 at 1:58 PM, Sean Paul <seanpaul@chromium.org> wrote:
> >
> > Move gpc4 to the end of the automatically processed gpio controllers
> > so we don't taint the automatic offset calculation.
> >
> > This bug caused all controllers coming after gpc4 to map to the
> > incorrect address. The result is <&gpd1 0 0 0 0> would actually map to
> > GPIO 0 in gpd0.
> >
> > Signed-off-by: Sean Paul <seanpaul@chromium.org>
> > ---
> >  drivers/gpio/gpio-samsung.c |   14 +++++++-------
> >  1 files changed, 7 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/gpio/gpio-samsung.c b/drivers/gpio/gpio-samsung.c
> > index b6453d0..e4599b4 100644
> > --- a/drivers/gpio/gpio-samsung.c
> > +++ b/drivers/gpio/gpio-samsung.c
> > @@ -2454,12 +2454,6 @@ static struct samsung_gpio_chip exynos5_gpios_1[] = {
> >                 },
> >         }, {
> >                 .chip   = {
> > -                       .base   = EXYNOS5_GPC4(0),
> > -                       .ngpio  = EXYNOS5_GPIO_C4_NR,
> > -                       .label  = "GPC4",
> > -               },
> > -       }, {
> > -               .chip   = {
> >                         .base   = EXYNOS5_GPD0(0),
> >                         .ngpio  = EXYNOS5_GPIO_D0_NR,
> >                         .label  = "GPD0", @@ -2513,6 +2507,12 @@
> > static struct samsung_gpio_chip exynos5_gpios_1[] = {
> >                         .label  = "GPY6",
> >                 },
> >         }, {
> > +               .chip   = {
> > +                       .base   = EXYNOS5_GPC4(0),
> > +                       .ngpio  = EXYNOS5_GPIO_C4_NR,
> > +                       .label  = "GPC4",
> > +               },
> > +       }, {
> >                 .config = &samsung_gpio_cfgs[9],
> >                 .irq_base = IRQ_EINT(0),
> >                 .chip   = {
> > @@ -2833,7 +2833,7 @@ static __init void exynos5_gpiolib_init(void)
> >         }
> >
> >         /* need to set base address for gpc4 */
> > -       exynos5_gpios_1[11].base = gpio_base1 + 0x2E0;
> > +       exynos5_gpios_1[20].base = gpio_base1 + 0x2E0;
> >
> >         /* need to set base address for gpx */
> >         chip = &exynos5_gpios_1[21];
> > --
> > 1.7.7.3
> >

I think that this modification is right.
Thanks.

Re: [PATCH] gpio: samsung: Fix off-by-one bug in gpio addresses

[Thomas Abraham]

On 30 July 2012 13:34, 박상수 <sangsu4u.park@samsung.com> wrote:
>
>
>> -----Original Message-----
>> From: dianders@google.com [mailto:dianders@google.com] On Behalf Of Doug Anderson
>> Sent: Saturday, July 21, 2012 6:07 AM
>> To: Sean Paul
>> Cc: grant.likely@secretlab.ca; linus.walleij@stericsson.com; linux-kernel@vger.kernel.org;
>> olofj@chromium.org; Sangsu Park; Thomas Abraham; Kukjin Kim
>> Subject: Re: [PATCH] gpio: samsung: Fix off-by-one bug in gpio addresses
>>
>> Acked-by: Doug Anderson <dianders@chromium.org>
>>
>> ---
>>
>> On Fri, Jul 20, 2012 at 1:58 PM, Sean Paul <seanpaul@chromium.org> wrote:
>> >
>> > Move gpc4 to the end of the automatically processed gpio controllers
>> > so we don't taint the automatic offset calculation.
>> >
>> > This bug caused all controllers coming after gpc4 to map to the
>> > incorrect address. The result is <&gpd1 0 0 0 0> would actually map to
>> > GPIO 0 in gpd0.
>> >
>> > Signed-off-by: Sean Paul <seanpaul@chromium.org>
>> > ---
>> >  drivers/gpio/gpio-samsung.c |   14 +++++++-------
>> >  1 files changed, 7 insertions(+), 7 deletions(-)
>> >
>> > diff --git a/drivers/gpio/gpio-samsung.c b/drivers/gpio/gpio-samsung.c
>> > index b6453d0..e4599b4 100644
>> > --- a/drivers/gpio/gpio-samsung.c
>> > +++ b/drivers/gpio/gpio-samsung.c
>> > @@ -2454,12 +2454,6 @@ static struct samsung_gpio_chip exynos5_gpios_1[] = {
>> >                 },
>> >         }, {
>> >                 .chip   = {
>> > -                       .base   = EXYNOS5_GPC4(0),
>> > -                       .ngpio  = EXYNOS5_GPIO_C4_NR,
>> > -                       .label  = "GPC4",
>> > -               },
>> > -       }, {
>> > -               .chip   = {
>> >                         .base   = EXYNOS5_GPD0(0),
>> >                         .ngpio  = EXYNOS5_GPIO_D0_NR,
>> >                         .label  = "GPD0", @@ -2513,6 +2507,12 @@
>> > static struct samsung_gpio_chip exynos5_gpios_1[] = {
>> >                         .label  = "GPY6",
>> >                 },
>> >         }, {
>> > +               .chip   = {
>> > +                       .base   = EXYNOS5_GPC4(0),
>> > +                       .ngpio  = EXYNOS5_GPIO_C4_NR,
>> > +                       .label  = "GPC4",
>> > +               },
>> > +       }, {
>> >                 .config = &samsung_gpio_cfgs[9],
>> >                 .irq_base = IRQ_EINT(0),
>> >                 .chip   = {
>> > @@ -2833,7 +2833,7 @@ static __init void exynos5_gpiolib_init(void)
>> >         }
>> >
>> >         /* need to set base address for gpc4 */
>> > -       exynos5_gpios_1[11].base = gpio_base1 + 0x2E0;
>> > +       exynos5_gpios_1[20].base = gpio_base1 + 0x2E0;
>> >
>> >         /* need to set base address for gpx */
>> >         chip = &exynos5_gpios_1[21];
>> > --
>> > 1.7.7.3
>> >
>
> I think that this modification is right.
> Thanks.

Reviewed-by: Thomas Abraham <thomas.abraham@linaro.org>

RE: [PATCH] gpio: samsung: Fix off-by-one bug in gpio addresses

[Kukjin Kim]

Linus Walleij wrote:
> 
> On Fri, Jul 20, 2012 at 10:58 PM, Sean Paul <seanpaul@chromium.org> wrote:
> 
> > Move gpc4 to the end of the automatically processed gpio controllers so
> > we don't taint the automatic offset calculation.
> >
> > This bug caused all controllers coming after gpc4 to map to the
> > incorrect address. The result is <&gpd1 0 0 0 0> would actually map to
> > GPIO 0 in gpd0.
> >
> > Signed-off-by: Sean Paul <seanpaul@chromium.org>
> 
> Samsung people, please comment on this patch.
> 
Looks good to me, please keep going on.
And feel free to add my ack on this.

Acked-by: Kukjin Kim <kgene.kim@samsung.com>

Thanks.

Best regards,
Kgene.
--
Kukjin Kim <kgene.kim@samsung.com>, Senior Engineer,
SW Solution Development Team, Samsung Electronics Co., Ltd.

Re: [PATCH] gpio: samsung: Fix off-by-one bug in gpio addresses

[Linus Walleij]

On Fri, Jul 20, 2012 at 10:58 PM, Sean Paul <seanpaul@chromium.org> wrote:

> Move gpc4 to the end of the automatically processed gpio controllers so
> we don't taint the automatic offset calculation.
>
> This bug caused all controllers coming after gpc4 to map to the
> incorrect address. The result is <&gpd1 0 0 0 0> would actually map to
> GPIO 0 in gpd0.
>
> Signed-off-by: Sean Paul <seanpaul@chromium.org>

Patch applied with some Acks and Reviewed-by's.

Thanks,
Linus Walleij