[PATCH] pinctrl/at91: Fix pin_to_mask

[PATCH] pinctrl/at91: Fix pin_to_mask

[Alexander Stein]

We need first to reduce the pin number to only a GPIO bank before we can
create the mask.
Otherwise only GPIO bank 0 has correct masks as the bits in the other
banks are shifted out of range.

Signed-off-by: Alexander Stein <alexanders83@web.de>
---
 drivers/pinctrl/pinctrl-at91.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/pinctrl/pinctrl-at91.c b/drivers/pinctrl/pinctrl-at91.c
index 63176f2..6669e13 100644
--- a/drivers/pinctrl/pinctrl-at91.c
+++ b/drivers/pinctrl/pinctrl-at91.c
@@ -316,7 +316,7 @@ static inline int pin_to_bank(unsigned pin)
 
 static unsigned pin_to_mask(unsigned int pin)
 {
-	return 1 << pin;
+	return 1 << (pin % MAX_NB_GPIO_PER_BANK);
 }
 
 static void at91_mux_disable_interrupt(void __iomem *pio, unsigned mask)
-- 
1.9.2

[PATCH] pinctrl/at91: Fix pin_to_mask

[Jean-Christophe PLAGNIOL-VILLARD]

On Apr 11, 2014, at 11:35 PM, Alexander Stein <alexanders83@web.de> wrote:

> 
> We need first to reduce the pin number to only a GPIO bank before we can
> create the mask.
> Otherwise only GPIO bank 0 has correct masks as the bits in the other
> banks are shifted out of range.
> 
> Signed-off-by: Alexander Stein <alexanders83@web.de>
> ---
> drivers/pinctrl/pinctrl-at91.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/pinctrl/pinctrl-at91.c b/drivers/pinctrl/pinctrl-at91.c
> index 63176f2..6669e13 100644
> --- a/drivers/pinctrl/pinctrl-at91.c
> +++ b/drivers/pinctrl/pinctrl-at91.c
> @@ -316,7 +316,7 @@ static inline int pin_to_bank(unsigned pin)
> 
> static unsigned pin_to_mask(unsigned int pin)
> {
> -	return 1 << pin;
> +	return 1 << (pin % MAX_NB_GPIO_PER_BANK);
> }
no need pin_to_mask is already called with it
> 
> 
> static void at91_mux_disable_interrupt(void __iomem *pio, unsigned mask)
> -- 
> 1.9.2
> 

[PATCH] pinctrl/at91: Fix pin_to_mask

[Alexander Stein]

On Friday 11 April 2014, 23:42:19 wrote Jean-Christophe PLAGNIOL-VILLARD:
> 
> On Apr 11, 2014, at 11:35 PM, Alexander Stein <alexanders83@web.de> wrote:
> 
> > 
> > We need first to reduce the pin number to only a GPIO bank before we can
> > create the mask.
> > Otherwise only GPIO bank 0 has correct masks as the bits in the other
> > banks are shifted out of range.
> > 
> > Signed-off-by: Alexander Stein <alexanders83@web.de>
> > ---
> > drivers/pinctrl/pinctrl-at91.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/drivers/pinctrl/pinctrl-at91.c b/drivers/pinctrl/pinctrl-at91.c
> > index 63176f2..6669e13 100644
> > --- a/drivers/pinctrl/pinctrl-at91.c
> > +++ b/drivers/pinctrl/pinctrl-at91.c
> > @@ -316,7 +316,7 @@ static inline int pin_to_bank(unsigned pin)
> > 
> > static unsigned pin_to_mask(unsigned int pin)
> > {
> > -	return 1 << pin;
> > +	return 1 << (pin % MAX_NB_GPIO_PER_BANK);
> > }
> no need pin_to_mask is already called with it

But this only true for the call within at91_pinconf_set, but not for those in
at91_gpio_dbg_show
at91_pmx_enable
at91_pmx_disable

Regards,
Alexander

[PATCH] pinctrl/at91: Fix pin_to_mask

[Mark Roszko]

The member struct pin inside the at91_pmx_pin structs is already the
bank pin number, it was documented at the top of the driver.

/**
   * struct at91_pmx_pin - describes an At91 pin mux
   * @bank: the bank of the pin
   * @pin: the pin number in the @bank
   * @mux: the mux mode : gpio or periph_x of the pin i.e. alternate function.
   * @conf: the configuration of the pin: PULL_UP, MULTIDRIVE etc...
   */

if the pmx_pin->pin member wasn't already a bank number, then other
logic would break such as if (pin->pin >= MAX_NB_GPIO_PER_BANK) in
pin_check_config. In fact pmx_enable calls pin_check_config and should
catch an error.

[PATCH] pinctrl/at91: Fix mask creation in at91_gpio_dbg_show

[Alexander Stein]

pin_to_mask expects a bank pin number. So do not add the chip base.

Signed-off-by: Alexander Stein <alexanders83@web.de>
---
Mark, thanks for the clarification that pin_to_mask only expects tha bank
pin number. This wasn't obvious to me.
 drivers/pinctrl/pinctrl-at91.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/pinctrl/pinctrl-at91.c b/drivers/pinctrl/pinctrl-at91.c
index 63176f2..2da0e6e 100644
--- a/drivers/pinctrl/pinctrl-at91.c
+++ b/drivers/pinctrl/pinctrl-at91.c
@@ -1205,8 +1205,7 @@ static void at91_gpio_dbg_show(struct seq_file *s, struct gpio_chip *chip)
 	void __iomem *pio = at91_gpio->regbase;
 
 	for (i = 0; i < chip->ngpio; i++) {
-		unsigned pin = chip->base + i;
-		unsigned mask = pin_to_mask(pin);
+		unsigned mask = pin_to_mask(i);
 		const char *gpio_label;
 		u32 pdsr;
 
-- 
1.9.2

[PATCH] pinctrl/at91: Fix mask creation in at91_gpio_dbg_show

[Alexandre Belloni]

On 11/04/2014 at 20:33:13 +0200, Alexander Stein wrote :
> pin_to_mask expects a bank pin number. So do not add the chip base.
> 
I would prefer that you add a description of the effects of the bug as
seen by the user.

> Signed-off-by: Alexander Stein <alexanders83@web.de>
else
Acked-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>

> ---
> Mark, thanks for the clarification that pin_to_mask only expects tha bank
> pin number. This wasn't obvious to me.
>  drivers/pinctrl/pinctrl-at91.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/pinctrl/pinctrl-at91.c b/drivers/pinctrl/pinctrl-at91.c
> index 63176f2..2da0e6e 100644
> --- a/drivers/pinctrl/pinctrl-at91.c
> +++ b/drivers/pinctrl/pinctrl-at91.c
> @@ -1205,8 +1205,7 @@ static void at91_gpio_dbg_show(struct seq_file *s, struct gpio_chip *chip)
>  	void __iomem *pio = at91_gpio->regbase;
>  
>  	for (i = 0; i < chip->ngpio; i++) {
> -		unsigned pin = chip->base + i;
> -		unsigned mask = pin_to_mask(pin);
> +		unsigned mask = pin_to_mask(i);
>  		const char *gpio_label;
>  		u32 pdsr;
>  
> -- 
> 1.9.2
> 
> 
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

-- 
Alexandre Belloni, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[PATCH] pinctrl/at91: Fix mask creation in at91_gpio_dbg_show

[Nicolas Ferre]

On 11/04/2014 20:33, Alexander Stein :
> pin_to_mask expects a bank pin number. So do not add the chip base.
> 
> Signed-off-by: Alexander Stein <alexanders83@web.de>

Indeed, seems to be a bug in this function:

Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>

Thanks a lot. Bye,


> ---
> Mark, thanks for the clarification that pin_to_mask only expects tha bank
> pin number. This wasn't obvious to me.
>  drivers/pinctrl/pinctrl-at91.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/pinctrl/pinctrl-at91.c b/drivers/pinctrl/pinctrl-at91.c
> index 63176f2..2da0e6e 100644
> --- a/drivers/pinctrl/pinctrl-at91.c
> +++ b/drivers/pinctrl/pinctrl-at91.c
> @@ -1205,8 +1205,7 @@ static void at91_gpio_dbg_show(struct seq_file *s, struct gpio_chip *chip)
>  	void __iomem *pio = at91_gpio->regbase;
>  
>  	for (i = 0; i < chip->ngpio; i++) {
> -		unsigned pin = chip->base + i;
> -		unsigned mask = pin_to_mask(pin);
> +		unsigned mask = pin_to_mask(i);
>  		const char *gpio_label;
>  		u32 pdsr;
>  
> 


-- 
Nicolas Ferre

[PATCH v2] pinctrl/at91: Fix mask creation in at91_gpio_dbg_show

[Alexander Stein]

pin_to_mask expects a bank pin number. So do not add the chip base.

Without that patch cat /sys/kernel/debug/gpio looks like that:
GPIOs 0-31, platform/fffff200.gpio, fffff200.gpio:
[spi32766.0] GPIOfffff200.gpio5: [gpio] set
[ads7846_pendown] GPIOfffff200.gpio15: [gpio] set
[ohci_vbus] GPIOfffff200.gpio21: [gpio] set
[ohci_vbus] GPIOfffff200.gpio24: [gpio] set
[button1] GPIOfffff200.gpio28: [gpio] clear
[button2] GPIOfffff200.gpio29: [gpio] clear

GPIOs 32-63, platform/fffff400.gpio, fffff400.gpio:
[sda] GPIOfffff400.gpio4: [periph A]
[scl] GPIOfffff400.gpio5: [periph A]
[spi32766.3] GPIOfffff400.gpio11: [periph A]
[error] GPIOfffff400.gpio22: [periph A]
[run] GPIOfffff400.gpio23: [periph A]

GPIOs 64-95, platform/fffff600.gpio, fffff600.gpio:
[reset_pin] GPIOfffff600.gpio29: [periph A]

GPIOs 96-127, platform/fffff800.gpio, fffff800.gpio:
[led1] GPIOfffff800.gpio5: [periph A]
[led2] GPIOfffff800.gpio6: [periph A]
[led3] GPIOfffff800.gpio7: [periph A]
[led4] GPIOfffff800.gpio8: [periph A]

GPIOs 128-159, platform/fffffa00.gpio, fffffa00.gpio:
[button3] GPIOfffffa00.gpio10: [periph A]
[button4] GPIOfffffa00.gpio12: [periph A]

Note that every bank despite bank 0 only shows "periph A" which are
obviously used as GPIOs.

Signed-off-by: Alexander Stein <alexanders83@web.de>
Acked-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
Changes in v2:
* Added 2 Acked-by's
* Add an example of wrong output

 drivers/pinctrl/pinctrl-at91.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/pinctrl/pinctrl-at91.c b/drivers/pinctrl/pinctrl-at91.c
index 63176f2..2da0e6e 100644
--- a/drivers/pinctrl/pinctrl-at91.c
+++ b/drivers/pinctrl/pinctrl-at91.c
@@ -1205,8 +1205,7 @@ static void at91_gpio_dbg_show(struct seq_file *s, struct gpio_chip *chip)
 	void __iomem *pio = at91_gpio->regbase;
 
 	for (i = 0; i < chip->ngpio; i++) {
-		unsigned pin = chip->base + i;
-		unsigned mask = pin_to_mask(pin);
+		unsigned mask = pin_to_mask(i);
 		const char *gpio_label;
 		u32 pdsr;
 
-- 
1.9.2

[PATCH v2] pinctrl/at91: Fix mask creation in at91_gpio_dbg_show

[Nicolas Ferre]

On 14/04/2014 20:53, Alexander Stein :
> pin_to_mask expects a bank pin number. So do not add the chip base.
> 
> Without that patch cat /sys/kernel/debug/gpio looks like that:
> GPIOs 0-31, platform/fffff200.gpio, fffff200.gpio:
> [spi32766.0] GPIOfffff200.gpio5: [gpio] set
> [ads7846_pendown] GPIOfffff200.gpio15: [gpio] set
> [ohci_vbus] GPIOfffff200.gpio21: [gpio] set
> [ohci_vbus] GPIOfffff200.gpio24: [gpio] set
> [button1] GPIOfffff200.gpio28: [gpio] clear
> [button2] GPIOfffff200.gpio29: [gpio] clear
> 
> GPIOs 32-63, platform/fffff400.gpio, fffff400.gpio:
> [sda] GPIOfffff400.gpio4: [periph A]
> [scl] GPIOfffff400.gpio5: [periph A]
> [spi32766.3] GPIOfffff400.gpio11: [periph A]
> [error] GPIOfffff400.gpio22: [periph A]
> [run] GPIOfffff400.gpio23: [periph A]
> 
> GPIOs 64-95, platform/fffff600.gpio, fffff600.gpio:
> [reset_pin] GPIOfffff600.gpio29: [periph A]
> 
> GPIOs 96-127, platform/fffff800.gpio, fffff800.gpio:
> [led1] GPIOfffff800.gpio5: [periph A]
> [led2] GPIOfffff800.gpio6: [periph A]
> [led3] GPIOfffff800.gpio7: [periph A]
> [led4] GPIOfffff800.gpio8: [periph A]
> 
> GPIOs 128-159, platform/fffffa00.gpio, fffffa00.gpio:
> [button3] GPIOfffffa00.gpio10: [periph A]
> [button4] GPIOfffffa00.gpio12: [periph A]

Maybe a little bit less context to the commit message: Linus might
remove some lines to it...

> Note that every bank despite bank 0 only shows "periph A" which are
> obviously used as GPIOs.
> 
> Signed-off-by: Alexander Stein <alexanders83@web.de>
> Acked-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>

Okay, actually this patch have to be routed to Linus Walleij, I add him
in the destination field.

Thanks,

> ---
> Changes in v2:
> * Added 2 Acked-by's
> * Add an example of wrong output
> 
>  drivers/pinctrl/pinctrl-at91.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/pinctrl/pinctrl-at91.c b/drivers/pinctrl/pinctrl-at91.c
> index 63176f2..2da0e6e 100644
> --- a/drivers/pinctrl/pinctrl-at91.c
> +++ b/drivers/pinctrl/pinctrl-at91.c
> @@ -1205,8 +1205,7 @@ static void at91_gpio_dbg_show(struct seq_file *s, struct gpio_chip *chip)
>  	void __iomem *pio = at91_gpio->regbase;
>  
>  	for (i = 0; i < chip->ngpio; i++) {
> -		unsigned pin = chip->base + i;
> -		unsigned mask = pin_to_mask(pin);
> +		unsigned mask = pin_to_mask(i);
>  		const char *gpio_label;
>  		u32 pdsr;
>  
> 


-- 
Nicolas Ferre

[PATCH v2] pinctrl/at91: Fix mask creation in at91_gpio_dbg_show

[Linus Walleij]

On Mon, Apr 14, 2014 at 8:53 PM, Alexander Stein <alexanders83@web.de> wrote:

> pin_to_mask expects a bank pin number. So do not add the chip base.
>
> Without that patch cat /sys/kernel/debug/gpio looks like that:
> GPIOs 0-31, platform/fffff200.gpio, fffff200.gpio:
> [spi32766.0] GPIOfffff200.gpio5: [gpio] set
> [ads7846_pendown] GPIOfffff200.gpio15: [gpio] set
> [ohci_vbus] GPIOfffff200.gpio21: [gpio] set
> [ohci_vbus] GPIOfffff200.gpio24: [gpio] set
> [button1] GPIOfffff200.gpio28: [gpio] clear
> [button2] GPIOfffff200.gpio29: [gpio] clear
>
> GPIOs 32-63, platform/fffff400.gpio, fffff400.gpio:
> [sda] GPIOfffff400.gpio4: [periph A]
> [scl] GPIOfffff400.gpio5: [periph A]
> [spi32766.3] GPIOfffff400.gpio11: [periph A]
> [error] GPIOfffff400.gpio22: [periph A]
> [run] GPIOfffff400.gpio23: [periph A]
>
> GPIOs 64-95, platform/fffff600.gpio, fffff600.gpio:
> [reset_pin] GPIOfffff600.gpio29: [periph A]
>
> GPIOs 96-127, platform/fffff800.gpio, fffff800.gpio:
> [led1] GPIOfffff800.gpio5: [periph A]
> [led2] GPIOfffff800.gpio6: [periph A]
> [led3] GPIOfffff800.gpio7: [periph A]
> [led4] GPIOfffff800.gpio8: [periph A]
>
> GPIOs 128-159, platform/fffffa00.gpio, fffffa00.gpio:
> [button3] GPIOfffffa00.gpio10: [periph A]
> [button4] GPIOfffffa00.gpio12: [periph A]
>
> Note that every bank despite bank 0 only shows "periph A" which are
> obviously used as GPIOs.
>
> Signed-off-by: Alexander Stein <alexanders83@web.de>
> Acked-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
> ---
> Changes in v2:
> * Added 2 Acked-by's
> * Add an example of wrong output

Patch applied.

Yours,
Linus Walleij