Skip to contentSkip to conversation

misc: microchip: pci1xxxx: Resolve IRQ handler kernel panic and incorrect IRQ status update issues

PATCH New Rengarajan S 1 participant 3 messages Mar 13, 2025

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover misc: microchip: pci1xxxx: Resolve IRQ handler kernel panic and incorrect IRQ status update issues

This patch series resolves couple of interrupt handling issues namely fixing the kernel panic during IRQ handler initialization and resolving the incorrect IRQ status update inside irq_ack.

Rengarajan S (2):
misc: microchip: pci1xxxx: Fix Kernel panic during IRQ handler
registration
misc: microchip: pci1xxxx: Fix incorrect IRQ status handling during
ack

 drivers/misc/mchp_pci1xxxx/mchp_pci1xxxx_gpio.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
Signature · 1 line
2.25.1
#1 lore raw

PATCH 1/2 misc: microchip: pci1xxxx: Fix Kernel panic during IRQ handler registration

Resolve kernel panic while accessing IRQ handler associated with the generated IRQ. This is done by acquiring the spinlock and storing the current interrupt state before handling the interrupt request using generic_handle_irq.

A previous fix patch was submitted where 'generic_handle_irq' was replaced with 'handle_nested_irq'. However, this change also causes the kernel panic where after determining which GPIO triggered the interrupt and attempting to call handle_nested_irq with the mapped IRQ number, leads to a failure in locating the registered handler.

Fixes 194f9f94a516 ("misc: microchip: pci1xxxx: Resolve kernel panic during GPIO IRQ handling")
---
 drivers/misc/mchp_pci1xxxx/mchp_pci1xxxx_gpio.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
drivers/misc/mchp_pci1xxxx/mchp_pci1xxxx_gpio.c +5 -1
index 3c1359d8d4e6..88a405b72125 100644
@@ -37,6 +37,7 @@
 struct pci1xxxx_gpio {
 	struct auxiliary_device *aux_dev;
 	void __iomem *reg_base;
+	raw_spinlock_t wa_lock;
 	struct gpio_chip gpio;
 	spinlock_t lock;
 	int irq_base;
@@ -254,6 +255,7 @@ static irqreturn_t pci1xxxx_gpio_irq_handler(int irq, void *dev_id)
 	struct pci1xxxx_gpio *priv = dev_id;
 	struct gpio_chip *gc =  &priv->gpio;
 	unsigned long int_status = 0;
+	unsigned long wa_flags;
 	unsigned long flags;
 	u8 pincount;
 	int bit;
@@ -277,7 +279,9 @@ static irqreturn_t pci1xxxx_gpio_irq_handler(int irq, void *dev_id)
 			writel(BIT(bit), priv->reg_base + INTR_STATUS_OFFSET(gpiobank));
 			spin_unlock_irqrestore(&priv->lock, flags);
 			irq = irq_find_mapping(gc->irq.domain, (bit + (gpiobank * 32)));
-			handle_nested_irq(irq);
+			raw_spin_lock_irqsave(&priv->wa_lock, wa_flags);
+			generic_handle_irq(irq);
+			raw_spin_unlock_irqrestore(&priv->wa_lock, wa_flags);
 		}
 	}
 	spin_lock_irqsave(&priv->lock, flags);
Signature · 1 line
2.25.1
#2 lore raw

PATCH 2/2 misc: microchip: pci1xxxx: Fix incorrect IRQ status handling during ack

In reply to Rengarajan S

Under irq_ack, pci1xxxx_assign_bit reads the current interrupt status, modifies and writes the entire value back. Since, the IRQ status bit gets cleared on writing back, the better approach is to directly write the bitmask to the register in order to preserve the value.

Fixes 1f4d8ae231f4 ("misc: microchip: pci1xxxx: Add gpio irq handler and irq helper functions irq_ack, irq_mask, irq_unmask and irq_

set_type of irq_chip.")

---
 drivers/misc/mchp_pci1xxxx/mchp_pci1xxxx_gpio.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/misc/mchp_pci1xxxx/mchp_pci1xxxx_gpio.c +1 -1
index 88a405b72125..55b892f982e9 100644
@@ -165,7 +165,7 @@ static void pci1xxxx_gpio_irq_ack(struct irq_data *data)
 	unsigned long flags;
 
 	spin_lock_irqsave(&priv->lock, flags);
-	pci1xxx_assign_bit(priv->reg_base, INTR_STAT_OFFSET(gpio), (gpio % 32), true);
+	writel(BIT(gpio % 32), priv->reg_base + INTR_STAT_OFFSET(gpio));
 	spin_unlock_irqrestore(&priv->lock, flags);
 }
 
Signature · 1 line
2.25.1