Re: [PATCH v4 08/23] dmaengine: ste_dma40: Return IRQ_NONE without interrupt status

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] dmaengine: ste_dma40: IRQ disabled due to unhandled DSP interrupts
--

commit 93721f73dfcbd1d64fc59b63723c4f215e77a3b2
Author: Linus Walleij <linusw@kernel.org>

dmaengine: ste_dma40: Return IRQ_NONE without interrupt status

This commit modifies the dma40 interrupt handler to track whether an
interrupt is acknowledged. If no matching status bit is found for a
registered channel, it returns IRQ_NONE to allow the generic spurious
interrupt detector to disable a stuck interrupt line.

> diff --git a/drivers/dma/ste_dma40.c b/drivers/dma/ste_dma40.c
> index 633f6a04fdc14..9f725369721b8 100644
> --- a/drivers/dma/ste_dma40.c
> +++ b/drivers/dma/ste_dma40.c
> @@ -1790,6 +1790,7 @@ static void dma_tasklet(struct tasklet_struct *t)
>  
>  static irqreturn_t d40_handle_interrupt(int irq, void *data)
>  {
> +	irqreturn_t handled = IRQ_NONE;
>  	int i;
>  	u32 idx;
>  	u32 row;

[ ... ]

> @@ -1838,6 +1839,7 @@ static irqreturn_t d40_handle_interrupt(int irq, void *data)
>  
>  		/* ACK interrupt */
>  		writel(BIT(idx), base->virtbase + il[row].clr);
> +		handled = IRQ_HANDLED;

[Severity: High]
Does returning IRQ_NONE here risk disabling the DMA interrupt line entirely
when another processor (like a DSP) is using the channel?

If a logical channel is owned by another processor, Linux receives the
hardware interrupt because interrupts are enabled for all logical channels.
In this scenario, d40c will be NULL, and d40_handle_interrupt() skips
acknowledging the interrupt and continues processing, leaving handled as
IRQ_NONE:

	if (!d40c) {
		/*
		 * No error because this can happen if something else
		 * in the system is using the channel.
		 */
		continue;
	}

Since other processors can generate frequent DMA completions, Linux could
rapidly accumulate IRQ_NONE returns. Would the core IRQ subsystem's spurious
detector log "nobody cared" and permanently disable the shared DMA interrupt
line, breaking all Linux DMA transfers?

>  
>  		spin_lock(&d40c->lock);
>  

[ ... ]

> @@ -1855,7 +1857,7 @@ static irqreturn_t d40_handle_interrupt(int irq, void *data)
>  	if (ret > 0)
>  		pm_runtime_put_autosuspend(base->dev);
>  
> -	return IRQ_HANDLED;
> +	return handled;
>  }

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260920-dma40-fixes-v4-0-d751b2d9c23f@kernel.org?part=8