Skip to contentSkip to conversation

gpiolib: fix up emulated open drain outputs

PATCH In review Russell King 4 participants 7 messages Dec 7, 2019 – Dec 11, 2019 GPIO

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 gpiolib: fix up emulated open drain outputs

gpiolib has a corner case with open drain outputs that are emulated. When such outputs are outputting a logic 1, emulation will set the hardware to input mode, which will cause gpiod_get_direction() to report that it is in input mode. This is different from the behaviour with a true open-drain output.

Unify the semantics here.

---
 drivers/gpio/gpiolib.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/gpio/gpiolib.c +8 -0
index 32c2048deb8c..70c60aac41cc 100644
@@ -220,6 +220,14 @@ int gpiod_get_direction(struct gpio_desc *desc)
 	chip = gpiod_to_chip(desc);
 	offset = gpio_chip_hwgpio(desc);
 
+	/*
+	 * Open drain emulation using input mode may incorrectly report
+	 * input here, fix that up.
+	 */
+	if (test_bit(FLAG_OPEN_DRAIN, &desc->flags) &&
+	    test_bit(FLAG_IS_OUT, &desc->flags))
+		return 0;
+
 	if (!chip->get_direction)
 		return -ENOTSUPP;
 
Signature · 1 line
2.20.1
#1 lore raw
Quoted 20 lines · Russell King

gpiolib has a corner case with open drain outputs that are emulated. When such outputs are outputting a logic 1, emulation will set the hardware to input mode, which will cause gpiod_get_direction() to report that it is in input mode. This is different from the behaviour with a true open-drain output.

Unify the semantics here.

---
 drivers/gpio/gpiolib.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/gpio/gpiolib.c +0 -0
index 32c2048deb8c..70c60aac41cc 100644
@@ -220,6 +220,14 @@ int gpiod_get_direction(struct gpio_desc *desc)
 	chip = gpiod_to_chip(desc);

drivers/gpio/gpiolib.c int gpiod_get_direction(struct gpio_desc *desc) · line 230

Quoted 10 lines
drivers/gpio/gpiolib.c +8 -0
 	offset = gpio_chip_hwgpio(desc);
 
+	/*
+	 * Open drain emulation using input mode may incorrectly report
+	 * input here, fix that up.
+	 */
+	if (test_bit(FLAG_OPEN_DRAIN, &desc->flags) &&
+	    test_bit(FLAG_IS_OUT, &desc->flags))
+		return 0;
+

What about the FLAG_OPEN_SOURCE case?

Kent.

Quoted 6 lines
drivers/gpio/gpiolib.c +0 -0
 	if (!chip->get_direction)
 		return -ENOTSUPP;
 
Signature · 1 line
2.20.1
#2 lore raw
Quoted 23 lines · Kent Gibson
Quoted 22 lines · Russell King

gpiolib has a corner case with open drain outputs that are emulated. When such outputs are outputting a logic 1, emulation will set the hardware to input mode, which will cause gpiod_get_direction() to report that it is in input mode. This is different from the behaviour with a true open-drain output.

Unify the semantics here.

---
 drivers/gpio/gpiolib.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/gpio/gpiolib.c +0 -0
index 32c2048deb8c..70c60aac41cc 100644
@@ -220,6 +220,14 @@ int gpiod_get_direction(struct gpio_desc *desc)
 	chip = gpiod_to_chip(desc);
 	offset = gpio_chip_hwgpio(desc);
 
Quoted 10 lines

drivers/gpio/gpiolib.c int gpiod_get_direction(struct gpio_desc *desc) · line 230

Quoted 8 lines
drivers/gpio/gpiolib.c +8 -0
+	/*
+	 * Open drain emulation using input mode may incorrectly report
+	 * input here, fix that up.
+	 */
+	if (test_bit(FLAG_OPEN_DRAIN, &desc->flags) &&
+	    test_bit(FLAG_IS_OUT, &desc->flags))
+		return 0;
+

What about the FLAG_OPEN_SOURCE case?

do you have a scenario you can test?

Signature · 3 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up
According to speedtest.net: 11.9Mbps down 500kbps up
#3 lore raw
Quoted 35 lines · Russell King - ARM Linux admin
Quoted 23 lines · Kent Gibson
Quoted 22 lines · Russell King

gpiolib has a corner case with open drain outputs that are emulated. When such outputs are outputting a logic 1, emulation will set the hardware to input mode, which will cause gpiod_get_direction() to report that it is in input mode. This is different from the behaviour with a true open-drain output.

Unify the semantics here.

---
 drivers/gpio/gpiolib.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/gpio/gpiolib.c +0 -0
index 32c2048deb8c..70c60aac41cc 100644
@@ -220,6 +220,14 @@ int gpiod_get_direction(struct gpio_desc *desc)
 	chip = gpiod_to_chip(desc);
 	offset = gpio_chip_hwgpio(desc);
 
Quoted 10 lines

drivers/gpio/gpiolib.c int gpiod_get_direction(struct gpio_desc *desc) · line 230

Quoted 8 lines
drivers/gpio/gpiolib.c +8 -0
+	/*
+	 * Open drain emulation using input mode may incorrectly report
+	 * input here, fix that up.
+	 */
+	if (test_bit(FLAG_OPEN_DRAIN, &desc->flags) &&
+	    test_bit(FLAG_IS_OUT, &desc->flags))
+		return 0;
+

What about the FLAG_OPEN_SOURCE case?

Quoted 2 lines

do you have a scenario you can test?

No I don't - if I had a scenario that had tripped over this problem then I would've submitted a patch already ;-).

I'm simply pointing out that the logic that applies to emulating OPEN_DRAIN also applies to emulating OPEN_SOURCE. IMHO if you are fixing this for one then it should be fixed for both.

I'd love to see some test cases for the gpiolib driver-api, including this function.
But I'm relatively new here, and am generally coming from the userspace side, so will have to admit ignorance to anything related to testing APIs within the kernel.

Kent.

Quoted 4 lines
Signature · 3 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up
According to speedtest.net: 11.9Mbps down 500kbps up
#4 lore raw
Quoted 39 lines · Kent Gibson
Quoted 35 lines · Russell King - ARM Linux admin
Quoted 23 lines · Kent Gibson
Quoted 22 lines · Russell King

gpiolib has a corner case with open drain outputs that are emulated. When such outputs are outputting a logic 1, emulation will set the hardware to input mode, which will cause gpiod_get_direction() to report that it is in input mode. This is different from the behaviour with a true open-drain output.

Unify the semantics here.

---
 drivers/gpio/gpiolib.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/gpio/gpiolib.c +0 -0
index 32c2048deb8c..70c60aac41cc 100644
@@ -220,6 +220,14 @@ int gpiod_get_direction(struct gpio_desc *desc)
 	chip = gpiod_to_chip(desc);
 	offset = gpio_chip_hwgpio(desc);
 
Quoted 10 lines

drivers/gpio/gpiolib.c int gpiod_get_direction(struct gpio_desc *desc) · line 230

Quoted 8 lines
drivers/gpio/gpiolib.c +8 -0
+	/*
+	 * Open drain emulation using input mode may incorrectly report
+	 * input here, fix that up.
+	 */
+	if (test_bit(FLAG_OPEN_DRAIN, &desc->flags) &&
+	    test_bit(FLAG_IS_OUT, &desc->flags))
+		return 0;
+

What about the FLAG_OPEN_SOURCE case?

Quoted 2 lines

do you have a scenario you can test?

Quoted 6 lines

No I don't - if I had a scenario that had tripped over this problem then I would've submitted a patch already ;-).

I'm simply pointing out that the logic that applies to emulating OPEN_DRAIN also applies to emulating OPEN_SOURCE. IMHO if you are fixing this for one then it should be fixed for both.

That would be nice, but it would also be nice to be sure that the fix works there _and_ it doesn't break anything by fixing it.

I regard this as a risky change: with open drain/open source "outputs" it is quite obvious when the pin is being driven, it is in output mode. When the driver is off though, it is debatable whether it should be regarded as in input or output mode.

Signature · 3 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up
According to speedtest.net: 11.9Mbps down 500kbps up
#5 lore raw
Quoted 47 lines · Russell King - ARM Linux admin
Quoted 39 lines · Kent Gibson
Quoted 35 lines · Russell King - ARM Linux admin
Quoted 23 lines · Kent Gibson
Quoted 22 lines · Russell King

gpiolib has a corner case with open drain outputs that are emulated. When such outputs are outputting a logic 1, emulation will set the hardware to input mode, which will cause gpiod_get_direction() to report that it is in input mode. This is different from the behaviour with a true open-drain output.

Unify the semantics here.

---
 drivers/gpio/gpiolib.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/gpio/gpiolib.c +0 -0
index 32c2048deb8c..70c60aac41cc 100644
@@ -220,6 +220,14 @@ int gpiod_get_direction(struct gpio_desc *desc)
 	chip = gpiod_to_chip(desc);
 	offset = gpio_chip_hwgpio(desc);
 
Quoted 10 lines

drivers/gpio/gpiolib.c int gpiod_get_direction(struct gpio_desc *desc) · line 230

Quoted 8 lines
drivers/gpio/gpiolib.c +8 -0
+	/*
+	 * Open drain emulation using input mode may incorrectly report
+	 * input here, fix that up.
+	 */
+	if (test_bit(FLAG_OPEN_DRAIN, &desc->flags) &&
+	    test_bit(FLAG_IS_OUT, &desc->flags))
+		return 0;
+

What about the FLAG_OPEN_SOURCE case?

Quoted 2 lines

do you have a scenario you can test?

Quoted 6 lines

No I don't - if I had a scenario that had tripped over this problem then I would've submitted a patch already ;-).

I'm simply pointing out that the logic that applies to emulating OPEN_DRAIN also applies to emulating OPEN_SOURCE. IMHO if you are fixing this for one then it should be fixed for both.

Quoted 8 lines

That would be nice, but it would also be nice to be sure that the fix works there _and_ it doesn't break anything by fixing it.

I regard this as a risky change: with open drain/open source "outputs" it is quite obvious when the pin is being driven, it is in output mode. When the driver is off though, it is debatable whether it should be regarded as in input or output mode.

Higher powers can make the call on that.

I just wanted to point out that the fix only deals with one of the two cases that need to be fixed - in case that slipped by.

Cheers,
Kent.

#6 lore raw

niedz., 8 gru 2019 o 14:34 Kent Gibson <[email protected]> napisał(a):

Quoted 23 lines
Quoted 14 lines
Quoted 7 lines
Quoted 6 lines
Quoted 3 lines

What about the FLAG_OPEN_SOURCE case?

do you have a scenario you can test?

Quoted 6 lines

No I don't - if I had a scenario that had tripped over this problem then I would've submitted a patch already ;-).

I'm simply pointing out that the logic that applies to emulating OPEN_DRAIN also applies to emulating OPEN_SOURCE. IMHO if you are fixing this for one then it should be fixed for both.

Quoted 8 lines

That would be nice, but it would also be nice to be sure that the fix works there _and_ it doesn't break anything by fixing it.

I regard this as a risky change: with open drain/open source "outputs" it is quite obvious when the pin is being driven, it is in output mode. When the driver is off though, it is debatable whether it should be regarded as in input or output mode.

Quoted 5 lines

Higher powers can make the call on that.

I just wanted to point out that the fix only deals with one of the two cases that need to be fixed - in case that slipped by.

I'll apply this patch as is and if someone reports problems with the opposite condition, then we'll fix that too.

Bartosz