diff mbox series

[v5,2/2] kexec: Prevent redundant IRQ masking by checking state before shutdown

Message ID 20241130201143.48808-3-farbere@amazon.com (mailing list archive)
State New
Headers show
Series Improve interrupt handling during machine kexec | expand

Commit Message

Eliav Farber Nov. 30, 2024, 8:11 p.m. UTC
During machine kexec, the function machine_kexec_mask_interrupts() is
responsible for disabling or masking all interrupts. While the irq_disable
hook ensures that an already-disabled IRQ is not disabled again, the
current implementation unconditionally invokes the irq_mask() function for
every interrupt descriptor, even when the interrupt is already masked.

A specific issue was observed in the crash kernel flow after unbinding a
device (prior to kexec) that used a GPIO as an IRQ source. The warning was
triggered by the gpiochip_disable_irq() function, which attempted to clear
the FLAG_IRQ_IS_ENABLED flag when FLAG_USED_AS_IRQ was not set:

```
void gpiochip_disable_irq(struct gpio_chip *gc, unsigned int offset)
{
	struct gpio_desc *desc = gpiochip_get_desc(gc, offset);

	if (!IS_ERR(desc) &&
	    !WARN_ON(!test_bit(FLAG_USED_AS_IRQ, &desc->flags)))
		clear_bit(FLAG_IRQ_IS_ENABLED, &desc->flags);
}
```

This issue surfaced after commit a8173820f441 ("gpio: gpiolib: Allow GPIO
IRQs to lazy disable") introduced lazy disablement for GPIO IRQs. It
replaced disable/enable hooks with mask/unmask hooks. Unlike the disable
hook, the mask hook doesn't handle already-masked IRQs.

When a GPIO-IRQ driver is unbound, the IRQ is released, triggering
__irq_disable() and irq_state_set_masked(). A subsequent call to
machine_kexec_mask_interrupts() re-invokes chip->irq_mask(). This results
in a call chain, including gpiochip_irq_mask() and gpiochip_disable_irq().
Since FLAG_USED_AS_IRQ was cleared earlier, a warning occurs.

Replace calls to irq_mask() and irq_disable() hooks with a simplified call
to irq_shutdown(), and check if the interrupt is started (irqd_is_started)
before calling the shutdown.

Signed-off-by: Eliav Farber <farbere@amazon.com>
---
V4 -> V5:
 - The phrase 'This patch' has been removed from the commit message.

 kernel/irq/kexec.c | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)
diff mbox series

Patch

diff --git a/kernel/irq/kexec.c b/kernel/irq/kexec.c
index 0f9548c1708d..1a3deffe6b5b 100644
--- a/kernel/irq/kexec.c
+++ b/kernel/irq/kexec.c
@@ -17,7 +17,7 @@  void machine_kexec_mask_interrupts(void)
 		int check_eoi = 1;
 
 		chip = irq_desc_get_chip(desc);
-		if (!chip)
+		if (!chip || !irqd_is_started(&desc->irq_data))
 			continue;
 
 		if (IS_ENABLED(CONFIG_GENERIC_IRQ_KEXEC_CLEAR_VM_FORWARD)) {
@@ -31,10 +31,6 @@  void machine_kexec_mask_interrupts(void)
 		if (check_eoi && chip->irq_eoi && irqd_irq_inprogress(&desc->irq_data))
 			chip->irq_eoi(&desc->irq_data);
 
-		if (chip->irq_mask)
-			chip->irq_mask(&desc->irq_data);
-
-		if (chip->irq_disable && !irqd_irq_disabled(&desc->irq_data))
-			chip->irq_disable(&desc->irq_data);
+		irq_shutdown(desc);
 	}
 }