Skip to contentSkip to conversation

gpio-rcar: Use OUTDT when reading GPIOs configured as output

PATCH In review Magnus Damm 3 participants 5 messages Jun 16, 2013 – Jun 17, 2013

Subsystems (3)
GPIO SuperH Arm/RISC-V/Renesas architecture
lore
Share & tools
lore

Magnus Damm

author maintainer
#0 lore raw

PATCH 1/1 gpio-rcar: Use OUTDT when reading GPIOs configured as output

From: Magnus Damm <[email protected]>

Testing on r8a7790 shows that INDT does not indicate the correct pin state when reading a GPIO configured as output, so update the gpio_rcar_get() function to handle this case.

Signed-off-by Magnus Damm <[email protected]>

---

 drivers/gpio/gpio-rcar.c |    9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
drivers/gpio/gpio-rcar.c +8 -1
@@ -230,7 +230,14 @@ static int gpio_rcar_direction_input(str
 
 static int gpio_rcar_get(struct gpio_chip *chip, unsigned offset)
 {
-	return (int)(gpio_rcar_read(gpio_to_priv(chip), INDT) & BIT(offset));
+	u32 bit = BIT(offset);
+
+	/* testing on r8a7790 shows that INDT does not show correct pin state
+	 * when configured as output, so use OUTDT in case of output pins */
+	if (gpio_rcar_read(gpio_to_priv(chip), INOUTSEL) & bit)
+		return (int)(gpio_rcar_read(gpio_to_priv(chip), OUTDT) & bit);
+	else
+		return (int)(gpio_rcar_read(gpio_to_priv(chip), INDT) & bit);
 }
 
 static void gpio_rcar_set(struct gpio_chip *chip, unsigned offset, int value)
#1 lore raw
Quoted 7 lines · Magnus Damm

From: Magnus Damm <[email protected]>

Testing on r8a7790 shows that INDT does not indicate the correct pin state when reading a GPIO configured as output, so update the gpio_rcar_get() function to handle this case.

Signed-off-by Magnus Damm <[email protected]>

Patch applied. Unless you need this to go through some SH tree, I'm carrying it in the GPIO tree.

Yours,
Linus Walleij

author maintainer
#2 lore raw
In reply to Magnus Damm

Hi Linus,

Quoted 4 lines · Linus Walleij
Quoted 2 lines · Magnus Damm

From: Magnus Damm <[email protected]>

Quoted 8 lines
Quoted 5 lines

Testing on r8a7790 shows that INDT does not indicate the correct pin state when reading a GPIO configured as output, so update the gpio_rcar_get() function to handle this case.

Signed-off-by Magnus Damm <[email protected]>

Patch applied. Unless you need this to go through some SH tree, I'm carrying it in the GPIO tree.

Thanks a lot! You carrying it in the GPIO tree sounds perfect to me.

Cheers,

/ magnus

Acked-by on v1 1/1
#3 lore raw
In reply to Magnus Damm

Hi Magnus,

Quoted 5 lines · On Monday 17 June 2013 08:41:52 Magnus Damm wrote:

From: Magnus Damm <[email protected]>

Testing on r8a7790 shows that INDT does not indicate the correct pin state when reading a GPIO configured as output, so update the gpio_rcar_get() function to handle this case.

Have you checked whether this is true on the other SoCs as well ? Reading back the output register instead of the input register will not allow detection of conflicts. If the hardware can't provide that, fine, but otherwise it would be a nice feature to keep (although it might not get used in practice).

Quoted 1 line
Signed-off-by Magnus Damm <[email protected]>
Acked-by Laurent Pinchart <[email protected]>
Quoted 24 lines

---

 drivers/gpio/gpio-rcar.c |    9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
drivers/gpio/gpio-rcar.c +8 -1
@@ -230,7 +230,14 @@ static int gpio_rcar_direction_input(str
 
 static int gpio_rcar_get(struct gpio_chip *chip, unsigned offset)
 {
-	return (int)(gpio_rcar_read(gpio_to_priv(chip), INDT) & BIT(offset));
+	u32 bit = BIT(offset);
+
+	/* testing on r8a7790 shows that INDT does not show correct pin state
+	 * when configured as output, so use OUTDT in case of output pins */
+	if (gpio_rcar_read(gpio_to_priv(chip), INOUTSEL) & bit)
+		return (int)(gpio_rcar_read(gpio_to_priv(chip), OUTDT) & bit);
+	else
+		return (int)(gpio_rcar_read(gpio_to_priv(chip), INDT) & bit);
 }
 
 static void gpio_rcar_set(struct gpio_chip *chip, unsigned offset, int

value)

Signature · 3 lines
Regards,

Laurent Pinchart
author maintainer
#4 lore raw
In reply to Magnus Damm

Hi Laurent,

Quoted 5 lines · Laurent Pinchart

Hi Magnus,

Quoted 2 lines · On Monday 17 June 2013 08:41:52 Magnus Damm wrote:

From: Magnus Damm <[email protected]>

Quoted 8 lines
Quoted 3 lines

Testing on r8a7790 shows that INDT does not indicate the correct pin state when reading a GPIO configured as output, so update the gpio_rcar_get() function to handle this case.

Have you checked whether this is true on the other SoCs as well ? Reading back the output register instead of the input register will not allow detection of conflicts. If the hardware can't provide that, fine, but otherwise it would be a nice feature to keep (although it might not get used in practice).

I have not tested this patch myself, but I've received feedback from an internal team at Renesas. Regarding other SoCs, I'm quite confident in this change, the data sheet for both R-Car H2 and R-Car H1 have the following sentence in them:

"INDT is valid only for the ports for which general input/output mode is selected by the general IO/interrupt switching
register and then general input mode is selected by the general input/output switching register"

I agree it would be nice with hardware that allowed reading the pin state regardless independently of any mode setting. =)

Quoted 3 lines
Quoted 1 line
Signed-off-by Magnus Damm <[email protected]>
Acked-by Laurent Pinchart <[email protected]>

Thanks!

/ magnus