Skip to contentSkip to conversation

gpio: Question about gpio_set_value() implementation in spear_gpio.c

U-Boot Axel Lin 5 participants 7 messages Jun 19, 2013 – Jun 20, 2013 GPIO

lore
Share & tools
lore

Axel Lin

author
#0 lore raw

Current code looks strange because no matter the value argument is 0 or 1 it always calls

	writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);

And then gpio_get_value() always return 1.

I'm wondering if it needs to be fixed, something like below change:

drivers/gpio/spear_gpio.c +4 -1
index d3c728e..8878608 100644
@@ -52,7 +52,10 @@ int gpio_set_value(unsigned gpio, int value)
 {
 	struct gpio_regs *regs = (struct gpio_regs *)CONFIG_GPIO_BASE;
 
-	writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+	if (value)
+		writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+	else
+		writel(0, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
 
 	return 0;
 }
#1 lore raw
Quoted 14 lines · Axel Lin

Current code looks strange because no matter the value argument is 0 or 1 it always calls

        writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);

And then gpio_get_value() always return 1.

I'm wondering if it needs to be fixed, something like below change:

drivers/gpio/spear_gpio.c +0 -0
index d3c728e..8878608 100644
@@ -52,7 +52,10 @@ int gpio_set_value(unsigned gpio, int value)
 {

drivers/gpio/spear_gpio.c int gpio_set_value(unsigned gpio, int value) · line 61

Quoted 10 lines
drivers/gpio/spear_gpio.c +4 -1
        struct gpio_regs *regs = (struct gpio_regs *)CONFIG_GPIO_BASE;
 
-       writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+       if (value)
+               writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+       else
+               writel(0, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
 
        return 0;
 }

writel(value << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);

Should do no? This would avoid the if block.

--
Otavio Salvador O.S. Systems
http://www.ossystems.com.br http://projetos.ossystems.com.br
Mobile: +55 (53) 9981-7854 Mobile: +1 (347) 903-9750

#2 lore raw

Dear Otavio Salvador,

Quoted 26 lines · Axel Lin
Quoted 25 lines

Current code looks strange because no matter the value argument is 0 or 1 it always calls

        writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);

And then gpio_get_value() always return 1.

I'm wondering if it needs to be fixed, something like below change:

drivers/gpio/spear_gpio.c +4 -1
index d3c728e..8878608 100644
@@ -52,7 +52,10 @@ int gpio_set_value(unsigned gpio, int value)
 
 {
 
        struct gpio_regs *regs = (struct gpio_regs *)CONFIG_GPIO_BASE;
 
-       writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+       if (value)
+               writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+       else
+               writel(0, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
 
Quoted 7 lines
Quoted 3 lines
        return 0;

}

writel(value << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);

Should do no? This would avoid the if block.

No, you need clrbits_le32() to unset the GPIO

Best regards,
Marek Vasut

#3 lore raw
In reply to Axel Lin

Dear Axel Lin,

Quoted 13 lines

Current code looks strange because no matter the value argument is 0 or 1 it always calls

	writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);

And then gpio_get_value() always return 1.

I'm wondering if it needs to be fixed, something like below change:

drivers/gpio/spear_gpio.c +0 -0
index d3c728e..8878608 100644

drivers/gpio/spear_gpio.c int gpio_set_value(unsigned gpio, int value) · line 58

Quoted 8 lines
drivers/gpio/spear_gpio.c +4 -1
@@ -52,7 +52,10 @@ int gpio_set_value(unsigned gpio, int value)
 {
 	struct gpio_regs *regs = (struct gpio_regs *)CONFIG_GPIO_BASE;
 
-	writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+	if (value)
+		writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+	else
+		writel(0, &regs->gpiodata[DATA_REG_ADDR(gpio)]);

You might want to use clrbits_le32() and setbits_le32() here, no?

drivers/gpio/spear_gpio.c int gpio_set_value(unsigned gpio, int value) · line 61

Quoted 2 lines
drivers/gpio/spear_gpio.c +0 -0
 	return 0;
 }

Best regards,
Marek Vasut

author
#4 lore raw

2013/6/19 Marek Vasut <[email protected]>:

Quoted 18 lines

Dear Axel Lin,

Quoted 16 lines

Current code looks strange because no matter the value argument is 0 or 1 it always calls

      writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);

And then gpio_get_value() always return 1.

I'm wondering if it needs to be fixed, something like below change:

drivers/gpio/spear_gpio.c +0 -0
index d3c728e..8878608 100644
@@ -52,7 +52,10 @@ int gpio_set_value(unsigned gpio, int value)
 {
      struct gpio_regs *regs = (struct gpio_regs *)CONFIG_GPIO_BASE;
 
Quoted 9 lines

drivers/gpio/spear_gpio.c int gpio_set_value(unsigned gpio, int value) · line 58

Quoted 5 lines
drivers/gpio/spear_gpio.c +4 -1
-     writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+     if (value)
+             writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+     else
+             writel(0, &regs->gpiodata[DATA_REG_ADDR(gpio)]);

You might want to use clrbits_le32() and setbits_le32() here, no?

Honestly, I ask this because I cannot find the detail datasheet for the gpio control of this device. ( Just found the code looks strange).

Usually we can use clrbits_le32() and setbits_le32() helpers to set/clear register bits. But if the only meaningful bit is "1 << gpio", simply use "write 1 << gpio" and "write 0" saves a register read operation.

So maybe Stefan or someone from ST can provide the information about gpio control on this hardware.

Regards,
Axel

#5 lore raw
Quoted 13 lines · Axel Lin
Quoted 9 lines

drivers/gpio/spear_gpio.c int gpio_set_value(unsigned gpio, int value) · line 58

Quoted 5 lines
drivers/gpio/spear_gpio.c +4 -1
-     writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+     if (value)
+             writel(1 << gpio, &regs->gpiodata[DATA_REG_ADDR(gpio)]);
+     else
+             writel(0, &regs->gpiodata[DATA_REG_ADDR(gpio)]);

You might want to use clrbits_le32() and setbits_le32() here, no?

Honestly, I ask this because I cannot find the detail datasheet for the gpio control of this device. ( Just found the code looks strange).

Quoted 6 lines

Usually we can use clrbits_le32() and setbits_le32() helpers to set/clear register bits. But if the only meaningful bit is "1 << gpio", simply use "write 1 << gpio" and "write 0" saves a register read operation.

So maybe Stefan or someone from ST can provide the information about gpio control on this hardware.

I honestly have no idea how I tested this GPIO driver. That was too long ago. And I don't have the time to dig out the hardware to do some re-testing. But the code definitely looks wrong, so thanks for catching this. Let's wait what the ST engineers have to comment here.

Thanks,
Stefan

#6 lore raw
In reply to Axel Lin
Quoted 13 lines · Axel Lin

Current code looks strange because no matter the value argument is 0 or 1 it always calls

	writel(1<<  gpio,&regs->gpiodata[DATA_REG_ADDR(gpio)]);

And then gpio_get_value() always return 1.

I'm wondering if it needs to be fixed, something like below change:

drivers/gpio/spear_gpio.c +0 -0
index d3c728e..8878608 100644

drivers/gpio/spear_gpio.c int gpio_set_value(unsigned gpio, int value) · line 59

Quoted 9 lines
drivers/gpio/spear_gpio.c +4 -1
@@ -52,7 +52,10 @@ int gpio_set_value(unsigned gpio, int value)
  {
  	struct gpio_regs *regs = (struct gpio_regs *)CONFIG_GPIO_BASE;
 
-	writel(1<<  gpio,&regs->gpiodata[DATA_REG_ADDR(gpio)]);
+	if (value)
+		writel(1<<  gpio,&regs->gpiodata[DATA_REG_ADDR(gpio)]);
+	else
+		writel(0,&regs->gpiodata[DATA_REG_ADDR(gpio)]);
 

Yes, this is the right way. It was a blunder. I am wondering no one ever tried to set a ZERO on any GPIO..

Thanks for pointing out

Regards
Vipin

Quoted 5 lines
drivers/gpio/spear_gpio.c +0 -0
  	return 0;
  }