Skip to contentSkip to conversation

gpiolib: handle gpio-hogs only once

PATCH Applied Daniel Drake 3 participants 5 messages Jun 8 – Jun 10 GPIO

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 gpiolib: handle gpio-hogs only once

Commit d1d564ec49929 ("gpio: move hogs into GPIO core") introduced a behaviour change that breaks boot on Raspberry Pi 5 when using the firmware-supplied device tree:

gpiochip_add_data_with_key: GPIOs 544..575

    (/soc@107c000000/gpio@7d517c00) failed to register, -22

brcmstb-gpio 107d517c00.gpio: Could not add gpiochip for bank 1 brcmstb-gpio 107d517c00.gpio: probe with driver brcmstb-gpio failed
with error -22

gpio-brcmstb registers two gpio_chips against the device tree node gpio@7d517c00, one for each bank. The firmware-supplied DT includes a gpio-hog on RP1 RUN, and this gpio-hog is attempted to be applied to *both* gpio_chips. This succeeds against bank 0 (which hosts the GPIO) and fails for bank 1 (which does not).

In the previous implementation, failures to apply gpio-hogs were quietly ignored. In the new code, the error code propagates and causes probe to fail.

Closely approximate the previous behaviour by using the OF_POPULATED flag to ensure that each gpio-hog is processed only once. The flag was previously being set before the gpio-hogs were processed, so as part of this change, the flag now gets set only after the gpio-hog is actioned. The handling of gpio-hogs on a DT node with multiple gpio_chips remains a bit incomplete/unclear, but this at least retains the ability to apply hogs to the first gpio_chip per node.

---
 drivers/gpio/gpiolib-of.c | 5 -----
 drivers/gpio/gpiolib.c    | 8 ++++++++
 2 files changed, 8 insertions(+), 5 deletions(-)

v2: move OF_POPULATED flag setting to happen after the gpio-hog has been applied (otherwise all hogs were considered already-applied and never applied at all, oops!)

This bug is only exposed by the RPi5 firmware-provided DT that has the gpio-hog. The DT shipped in the mainline kernel does not have the hog here. I'm not sure to what extent Linux cares about supporting the downstream firmware DT.

I'm also happy to consider other approaches. This multi-gpiochip setup is a bit weird and gpio-brcmstb could perhaps be converted to register only a single gpio_chip covering all banks. I verified that the other drivers that obviously follow this same multiple-gpiochip pattern (pinctrl-amlogic-a4, pinctrl-st and pinctrl-stm32) do not seem to be used by any board DTs that include gpio-hogs.

drivers/gpio/gpiolib-of.c +0 -5
index 2c923d17541f..813dbcb91f6f 100644
@@ -1066,11 +1066,6 @@ int of_gpiochip_add(struct gpio_chip *chip)
 
 	of_node_get(np);
 
-	for_each_available_child_of_node_scoped(np, child) {
-		if (of_property_read_bool(child, "gpio-hog"))
-			of_node_set_flag(child, OF_POPULATED);
-	}
-
 	return ret;
 }
 
drivers/gpio/gpiolib.c +8 -0
index 1e6dce430dca..b02d711289d0 100644
@@ -1031,9 +1031,17 @@ static int gpiochip_hog_lines(struct gpio_chip *gc)
 		if (!fwnode_property_present(fwnode, "gpio-hog"))
 			continue;
 
+		/* The hog may have been handled by another gpio_chip on the same fwnode */
+		if (is_of_node(fwnode) &&
+		    of_node_check_flag(to_of_node(fwnode), OF_POPULATED))
+			continue;
+
 		ret = gpiochip_add_hog(gc, fwnode);
 		if (ret)
 			return ret;
+
+		if (is_of_node(fwnode))
+			of_node_set_flag(to_of_node(fwnode), OF_POPULATED);
 	}
 
 	return 0;
Signature · 1 line
2.54.0
#1 lore raw
Quoted 28 lines · Daniel Drake

Commit d1d564ec49929 ("gpio: move hogs into GPIO core") introduced a behaviour change that breaks boot on Raspberry Pi 5 when using the firmware-supplied device tree:

gpiochip_add_data_with_key: GPIOs 544..575

    (/soc@107c000000/gpio@7d517c00) failed to register, -22

brcmstb-gpio 107d517c00.gpio: Could not add gpiochip for bank 1 brcmstb-gpio 107d517c00.gpio: probe with driver brcmstb-gpio failed
with error -22

gpio-brcmstb registers two gpio_chips against the device tree node gpio@7d517c00, one for each bank. The firmware-supplied DT includes a gpio-hog on RP1 RUN, and this gpio-hog is attempted to be applied to *both* gpio_chips. This succeeds against bank 0 (which hosts the GPIO) and fails for bank 1 (which does not).

In the previous implementation, failures to apply gpio-hogs were quietly ignored. In the new code, the error code propagates and causes probe to fail.

Closely approximate the previous behaviour by using the OF_POPULATED flag to ensure that each gpio-hog is processed only once. The flag was previously being set before the gpio-hogs were processed, so as part of this change, the flag now gets set only after the gpio-hog is actioned. The handling of gpio-hogs on a DT node with multiple gpio_chips remains a bit incomplete/unclear, but this at least retains the ability to apply hogs to the first gpio_chip per node.

Quoted 1 line

This needs a Fixes tag.

Quoted 5 lines
---
 drivers/gpio/gpiolib-of.c | 5 -----
 drivers/gpio/gpiolib.c    | 8 ++++++++
 2 files changed, 8 insertions(+), 5 deletions(-)
Quoted 9 lines

v2: move OF_POPULATED flag setting to happen after the gpio-hog has been applied (otherwise all hogs were considered already-applied and never applied at all, oops!)

This bug is only exposed by the RPi5 firmware-provided DT that has the gpio-hog. The DT shipped in the mainline kernel does not have the hog here. I'm not sure to what extent Linux cares about supporting the downstream firmware DT.

We care about not breaking working setups, I think this should be fixed.

Quoted 32 lines

I'm also happy to consider other approaches. This multi-gpiochip setup is a bit weird and gpio-brcmstb could perhaps be converted to register only a single gpio_chip covering all banks. I verified that the other drivers that obviously follow this same multiple-gpiochip pattern (pinctrl-amlogic-a4, pinctrl-st and pinctrl-stm32) do not seem to be used by any board DTs that include gpio-hogs.

drivers/gpio/gpiolib-of.c +0 -5
index 2c923d17541f..813dbcb91f6f 100644
@@ -1066,11 +1066,6 @@ int of_gpiochip_add(struct gpio_chip *chip)
 
 	of_node_get(np);
 
-	for_each_available_child_of_node_scoped(np, child) {
-		if (of_property_read_bool(child, "gpio-hog"))
-			of_node_set_flag(child, OF_POPULATED);
-	}
-
 	return ret;
 }
 
drivers/gpio/gpiolib.c +1 -0
index 1e6dce430dca..b02d711289d0 100644
@@ -1031,9 +1031,17 @@ static int gpiochip_hog_lines(struct gpio_chip *gc)
 		if (!fwnode_property_present(fwnode, "gpio-hog"))
 			continue;
 
+		/* The hog may have been handled by another gpio_chip on the same fwnode */

drivers/gpio/gpiolib.c static int gpiochip_hog_lines(struct gpio_chip *gc) · line 1044

Quoted 10 lines
drivers/gpio/gpiolib.c +7 -0
+		if (is_of_node(fwnode) &&
+		    of_node_check_flag(to_of_node(fwnode), OF_POPULATED))
+			continue;
+
 		ret = gpiochip_add_hog(gc, fwnode);
 		if (ret)
 			return ret;
+
+		if (is_of_node(fwnode))
+			of_node_set_flag(to_of_node(fwnode), OF_POPULATED);

Sashiko correctly points out that on errors, the state will be corrupted. We could maybe move the clearing of the flag to gpiochip_free_hogs() and track its state when processing fwnodes in order not to clear it incorrectly?

Quoted 7 lines
drivers/gpio/gpiolib.c +0 -1
 	}
 
 	return 0;
--

2.54.0

Thanks,
Bartosz

#2 lore raw
Quoted 9 lines · Bartosz Golaszewski
Quoted 8 lines · Daniel Drake
drivers/gpio/gpiolib.c +0 -0
index 1e6dce430dca..b02d711289d0 100644
@@ -1031,9 +1031,17 @@ static int gpiochip_hog_lines(struct gpio_chip *gc)
  		if (!fwnode_property_present(fwnode, "gpio-hog"))
  			continue;
 
Quoted 15 lines

drivers/gpio/gpiolib.c static int gpiochip_hog_lines(struct gpio_chip *gc) · line 1044

Quoted 11 lines
drivers/gpio/gpiolib.c +8 -0
+		/* The hog may have been handled by another gpio_chip on the same fwnode */
+		if (is_of_node(fwnode) &&
+		    of_node_check_flag(to_of_node(fwnode), OF_POPULATED))
+			continue;
+
  		ret = gpiochip_add_hog(gc, fwnode);
  		if (ret)
  			return ret;
+
+		if (is_of_node(fwnode))
+			of_node_set_flag(to_of_node(fwnode), OF_POPULATED);

Sashiko correctly points out that on errors, the state will be corrupted. We could maybe move the clearing of the flag to gpiochip_free_hogs() and track its state when processing fwnodes in order not to clear it incorrectly?

I guess you are referring to:

Quoted 9 lines

Does setting OF_POPULATED here cause state corruption if a secondary chip on a shared node fails to probe?
When multiple gpio_chip instances share a device node, the first chip processes its hogs and sets OF_POPULATED. If a subsequent chip fails probe (for example, returning -EPROBE_DEFER), its cleanup path calls of_gpiochip_remove() which clears the flag for all hogs.
If the flag is unconditionally cleared, will the deferred chip attempt to process the first chip's hogs on retry, fail due to a mismatch, and permanently abort probe?

I don't think this is actually an issue. If we have two gpio_chips sharing a device node, a first one with a hog that probes fine and a subsequent one that fails during probe, both of the gpio_chips will brought down and the flag is cleared. If it was a EPROBE_DEFER case which is then retried later, the first chip's hogs will be set up a 2nd time when the probe is retried.

It is true that the teardown of the 2nd gpio_chip would erase the OF_POPULATED flag of a gpio-hog node that it does not "own", but the first gpio_chip would also be torn down at the same time (and OF_POPULATED unset a 2nd time). This is not ideal, but harmless as far as I can see.

I don't quite follow the suggestion for doing the clearing better in gpiochip_free_hogs(). It would be neat if we could go from a hogged gpio_desc back to the fwnode, so that we could only unset OF_POPULATED on the fwnode at the point when we are really removing the hog, but I don't see a way to derive the gpio-hog fwnode from gpio_desc. Also, this would be complicated because one gpio-hog node can hog multiple gpios.

Let me know if I'm missing anything or if you have any preference to handle this differently!

Thanks
Daniel

#3 lore raw
Quoted 59 lines · Daniel Drake
Quoted 9 lines · Bartosz Golaszewski
Quoted 8 lines · Daniel Drake
drivers/gpio/gpiolib.c +0 -0
index 1e6dce430dca..b02d711289d0 100644
@@ -1031,9 +1031,17 @@ static int gpiochip_hog_lines(struct gpio_chip *gc)
  		if (!fwnode_property_present(fwnode, "gpio-hog"))
  			continue;
 
Quoted 15 lines

drivers/gpio/gpiolib.c static int gpiochip_hog_lines(struct gpio_chip *gc) · line 1044

Quoted 11 lines
drivers/gpio/gpiolib.c +8 -0
+		/* The hog may have been handled by another gpio_chip on the same fwnode */
+		if (is_of_node(fwnode) &&
+		    of_node_check_flag(to_of_node(fwnode), OF_POPULATED))
+			continue;
+
  		ret = gpiochip_add_hog(gc, fwnode);
  		if (ret)
  			return ret;
+
+		if (is_of_node(fwnode))
+			of_node_set_flag(to_of_node(fwnode), OF_POPULATED);

Sashiko correctly points out that on errors, the state will be corrupted. We could maybe move the clearing of the flag to gpiochip_free_hogs() and track its state when processing fwnodes in order not to clear it incorrectly?

I guess you are referring to:

Quoted 9 lines

Does setting OF_POPULATED here cause state corruption if a secondary chip on a shared node fails to probe?
When multiple gpio_chip instances share a device node, the first chip processes its hogs and sets OF_POPULATED. If a subsequent chip fails probe (for example, returning -EPROBE_DEFER), its cleanup path calls of_gpiochip_remove() which clears the flag for all hogs.
If the flag is unconditionally cleared, will the deferred chip attempt to process the first chip's hogs on retry, fail due to a mismatch, and permanently abort probe?

I don't think this is actually an issue. If we have two gpio_chips sharing a device node, a first one with a hog that probes fine and a subsequent one that fails during probe, both of the gpio_chips will brought down and the flag is cleared. If it was a EPROBE_DEFER case which is then retried later, the first chip's hogs will be set up a 2nd time when the probe is retried.

It is true that the teardown of the 2nd gpio_chip would erase the OF_POPULATED flag of a gpio-hog node that it does not "own", but the first gpio_chip would also be torn down at the same time (and OF_POPULATED unset a 2nd time). This is not ideal, but harmless as far as I can see.

I don't quite follow the suggestion for doing the clearing better in gpiochip_free_hogs(). It would be neat if we could go from a hogged gpio_desc back to the fwnode, so that we could only unset OF_POPULATED on the fwnode at the point when we are really removing the hog, but I don't see a way to derive the gpio-hog fwnode from gpio_desc. Also, this would be complicated because one gpio-hog node can hog multiple gpios.

Quoted 7 lines

Let me know if I'm missing anything or if you have any preference to handle this differently!

Thanks
Daniel

Ok, thanks for the clarification. Let's give this a chance then.

Bart

#4 lore raw
In reply to Daniel Drake
Quoted 11 lines · Daniel Drake

Commit d1d564ec49929 ("gpio: move hogs into GPIO core") introduced a behaviour change that breaks boot on Raspberry Pi 5 when using the firmware-supplied device tree:

gpiochip_add_data_with_key: GPIOs 544..575

    (/soc@107c000000/gpio@7d517c00) failed to register, -22

brcmstb-gpio 107d517c00.gpio: Could not add gpiochip for bank 1 brcmstb-gpio 107d517c00.gpio: probe with driver brcmstb-gpio failed
with error -22

[...]

Applied, thanks!

[1/1] gpiolib: handle gpio-hogs only once

      https://git.kernel.org/brgl/c/a23226b7c1f69eafd9ced4e037fb51c9758c0501

Best regards,

Signature · 1 line
Bartosz Golaszewski <[email protected]>