Skip to contentSkip to conversation

gpio: gpio-omap: Remove conditional pm_runtime handling for GPIO interrupts

PATCH In review Tony Lindgren 3 participants 3 messages Apr 10, 2019 – Apr 11, 2019

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 gpio: gpio-omap: Remove conditional pm_runtime handling for GPIO interrupts

From: Russell King <[email protected]>

Commit b764a5863fd8 ("gpio: omap: Remove custom PM calls and use cpu_pm instead") moved interrupt using GPIO banks to idle with cpu_pm in order to drop the use of pm_runtime_irq_safe() in a later patch. The GPIO banks with no interrupts claimed are still being idled based on PM runtime calls. However this caused a regression for am437x suspend for rtc+ddr idle mode where the device cannot enter idle state as reported by Keerthy <[email protected]>.

To fix the issue, we must not fail the pm_runtime callbacks. For GPIO interrupts, we already have irq_chip_pm_get increment the PM runtime use count as pointed out by Grygorii Strashko <[email protected]>.

So all we need to do is remove the conditional handling in the runtime_suspend and resume functions and let the CPU PM notifier idle the GPIO banks.

Cc Aaro Koskinen <[email protected]>
Cc Grygorii Strashko <[email protected]>
Cc Keerthy <[email protected]>
Cc Peter Ujfalusi <[email protected]>
Cc Tero Kristo <[email protected]>
Fixes b764a5863fd8 ("gpio: omap: Remove custom PM calls and use cpu_pm instead")
[[email protected]: updated patch description, dropped runtime count changes]

---

Changes since v1:
- Dropped PM runtime changes to increment count in omap_gpio_irq_startup as that's not needed based on comments from Grygorii

- Updated subject and description accordingly

---
 drivers/gpio/gpio-omap.c | 18 ++----------------
 1 file changed, 2 insertions(+), 16 deletions(-)
drivers/gpio/gpio-omap.c +2 -16
@@ -1719,40 +1719,26 @@ static int __maybe_unused omap_gpio_runtime_suspend(struct device *dev)
 {
 	struct gpio_bank *bank = dev_get_drvdata(dev);
 	unsigned long flags;
-	int error = 0;
 
 	raw_spin_lock_irqsave(&bank->lock, flags);
-	/* Must be idled only by CPU_CLUSTER_PM_ENTER? */
-	if (bank->irq_usage) {
-		error = -EBUSY;
-		goto unlock;
-	}
 	omap_gpio_idle(bank, true);
 	bank->is_suspended = true;
-unlock:
 	raw_spin_unlock_irqrestore(&bank->lock, flags);
 
-	return error;
+	return 0;
 }
 
 static int __maybe_unused omap_gpio_runtime_resume(struct device *dev)
 {
 	struct gpio_bank *bank = dev_get_drvdata(dev);
 	unsigned long flags;
-	int error = 0;
 
 	raw_spin_lock_irqsave(&bank->lock, flags);
-	/* Must be unidled only by CPU_CLUSTER_PM_ENTER? */
-	if (bank->irq_usage) {
-		error = -EBUSY;
-		goto unlock;
-	}
 	omap_gpio_unidle(bank);
 	bank->is_suspended = false;
-unlock:
 	raw_spin_unlock_irqrestore(&bank->lock, flags);
 
-	return error;
+	return 0;
 }
 
 static const struct dev_pm_ops gpio_pm_ops = {
Signature · 1 line
2.21.0
Tested-by Reviewed-by on v2 1/1
#1 lore raw
Quoted 10 lines · Tony Lindgren

From: Russell King <[email protected]>

Commit b764a5863fd8 ("gpio: omap: Remove custom PM calls and use cpu_pm instead") moved interrupt using GPIO banks to idle with cpu_pm in order to drop the use of pm_runtime_irq_safe() in a later patch. The GPIO banks with no interrupts claimed are still being idled based on PM runtime calls. However this caused a regression for am437x suspend for rtc+ddr idle mode where the device cannot enter idle state as reported by Keerthy <[email protected]>.

Quoted 7 lines

To fix the issue, we must not fail the pm_runtime callbacks. For GPIO interrupts, we already have irq_chip_pm_get increment the PM runtime use count as pointed out by Grygorii Strashko <[email protected]>.

So all we need to do is remove the conditional handling in the runtime_suspend and resume functions and let the CPU PM notifier idle the GPIO banks.

On am437x-gp-evm Tested for rtc+ddr mode and DS0 back & forth On am335x-beaglebone-black tested for DS0

Quoted 70 lines
Cc Aaro Koskinen <[email protected]>
Cc Grygorii Strashko <[email protected]>
Cc Keerthy <[email protected]>
Cc Peter Ujfalusi <[email protected]>
Cc Tero Kristo <[email protected]>
Fixes b764a5863fd8 ("gpio: omap: Remove custom PM calls and use cpu_pm instead")
[[email protected]: updated patch description, dropped runtime count changes]

---

Changes since v1:
- Dropped PM runtime changes to increment count in omap_gpio_irq_startup as that's not needed based on comments from Grygorii

- Updated subject and description accordingly

---
  drivers/gpio/gpio-omap.c | 18 ++----------------
  1 file changed, 2 insertions(+), 16 deletions(-)
drivers/gpio/gpio-omap.c +2 -16
@@ -1719,40 +1719,26 @@ static int __maybe_unused omap_gpio_runtime_suspend(struct device *dev)
  {
  	struct gpio_bank *bank = dev_get_drvdata(dev);
  	unsigned long flags;
-	int error = 0;
 
  	raw_spin_lock_irqsave(&bank->lock, flags);
-	/* Must be idled only by CPU_CLUSTER_PM_ENTER? */
-	if (bank->irq_usage) {
-		error = -EBUSY;
-		goto unlock;
-	}
  	omap_gpio_idle(bank, true);
  	bank->is_suspended = true;
-unlock:
  	raw_spin_unlock_irqrestore(&bank->lock, flags);
 
-	return error;
+	return 0;
  }
 
  static int __maybe_unused omap_gpio_runtime_resume(struct device *dev)
  {
  	struct gpio_bank *bank = dev_get_drvdata(dev);
  	unsigned long flags;
-	int error = 0;
 
  	raw_spin_lock_irqsave(&bank->lock, flags);
-	/* Must be unidled only by CPU_CLUSTER_PM_ENTER? */
-	if (bank->irq_usage) {
-		error = -EBUSY;
-		goto unlock;
-	}
  	omap_gpio_unidle(bank);
  	bank->is_suspended = false;
-unlock:
  	raw_spin_unlock_irqrestore(&bank->lock, flags);
 
-	return error;
+	return 0;
  }
 
  static const struct dev_pm_ops gpio_pm_ops = {
#2 lore raw
In reply to Tony Lindgren
Quoted 28 lines · Tony Lindgren

From: Russell King <[email protected]>

Commit b764a5863fd8 ("gpio: omap: Remove custom PM calls and use cpu_pm instead") moved interrupt using GPIO banks to idle with cpu_pm in order to drop the use of pm_runtime_irq_safe() in a later patch. The GPIO banks with no interrupts claimed are still being idled based on PM runtime calls. However this caused a regression for am437x suspend for rtc+ddr idle mode where the device cannot enter idle state as reported by Keerthy <[email protected]>.

To fix the issue, we must not fail the pm_runtime callbacks. For GPIO interrupts, we already have irq_chip_pm_get increment the PM runtime use count as pointed out by Grygorii Strashko <[email protected]>.

So all we need to do is remove the conditional handling in the runtime_suspend and resume functions and let the CPU PM notifier idle the GPIO banks.

Cc Aaro Koskinen <[email protected]>
Cc Grygorii Strashko <[email protected]>
Cc Keerthy <[email protected]>
Cc Peter Ujfalusi <[email protected]>
Cc Tero Kristo <[email protected]>
Fixes b764a5863fd8 ("gpio: omap: Remove custom PM calls and use cpu_pm instead")
[[email protected]: updated patch description, dropped runtime count changes]
Quoted 7 lines

---

Changes since v1:
- Dropped PM runtime changes to increment count in omap_gpio_irq_startup as that's not needed based on comments from Grygorii

- Updated subject and description accordingly

Applied this v2 version.

Yours,
Linus Walleij