Skip to contentSkip to conversation

USB: dwc2: hande irq on dead controller correctly

PATCH In review Oliver Neukum 2 participants 3 messages Aug 22, 2023 – Aug 28, 2023

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

PATCH 1/1 USB: dwc2: hande irq on dead controller correctly

If the controller is dead, the honest answer to the question whether it has caused an irq is: unknown As the purpose of the irq return is to trigger switching off an IRQ, the correct response if you cannot determine if your device has caused the interrupt is IRQ_HANDLED

---
 drivers/usb/dwc2/hcd_intr.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/usb/dwc2/hcd_intr.c +3 -1
index c9740caa5974..0144ca8350c3 100644
@@ -2203,11 +2203,13 @@ static void dwc2_hc_intr(struct dwc2_hsotg *hsotg)
 irqreturn_t dwc2_handle_hcd_intr(struct dwc2_hsotg *hsotg)
 {
 	u32 gintsts, dbg_gintsts;
-	irqreturn_t retval = IRQ_NONE;
+	irqreturn_t retval = IRQ_HANDLED;
 
 	if (!dwc2_is_controller_alive(hsotg)) {
 		dev_warn(hsotg->dev, "Controller is dead\n");
 		return retval;
+	} else {
+		retval = IRQ_NONE;
 	}
 
 	spin_lock(&hsotg->lock);
Signature · 1 line
2.41.0
#1 lore raw

Hi Oliver,

Quoted 7 lines · Oliver Neukum

If the controller is dead, the honest answer to the question whether it has caused an irq is: unknown As the purpose of the irq return is to trigger switching off an IRQ, the correct response if you cannot determine if your device has caused the interrupt is IRQ_HANDLED

If core is dead then it can not assert interrupt, it's more possible that it's spurious interrupt or interrupt from other device (IRQ is shared) and nothing should be handled. In this case reply by IRQ_HANDLED is incorrect because nothing is handled.

Thanks,
Minas

Quoted 24 lines
---
  drivers/usb/dwc2/hcd_intr.c | 4 +++-
  1 file changed, 3 insertions(+), 1 deletion(-)
drivers/usb/dwc2/hcd_intr.c +3 -1
index c9740caa5974..0144ca8350c3 100644
@@ -2203,11 +2203,13 @@ static void dwc2_hc_intr(struct dwc2_hsotg *hsotg)
  irqreturn_t dwc2_handle_hcd_intr(struct dwc2_hsotg *hsotg)
  {
  	u32 gintsts, dbg_gintsts;
-	irqreturn_t retval = IRQ_NONE;
+	irqreturn_t retval = IRQ_HANDLED;
 
  	if (!dwc2_is_controller_alive(hsotg)) {
  		dev_warn(hsotg->dev, "Controller is dead\n");
  		return retval;
+	} else {
+		retval = IRQ_NONE;
  	}
 
  	spin_lock(&hsotg->lock);
#2 lore raw

On 28.08.23 12:52, Minas Harutyunyan wrote:

Hi,

Quoted 1 line

If core is dead then it can not assert interrupt, it's more possible

No, but it may have asserted the interrupt and then died.

Quoted 3 lines

that it's spurious interrupt or interrupt from other device (IRQ is shared) and nothing should be handled. In this case reply by IRQ_HANDLED is incorrect because nothing is handled.

In this case we cannot be sure if it was our interrupt. IRQ_NONE however means that we are sure. Likelihood is not a relevant question here. We must know for sure if we claim IRQ_NONE, because IRQ_NONE has consequences.

Regards
Oliver