Re: [PATCH RFC v2 2/2] ufs: core: delegate the interrupt service routine to a threaded irq handler

[Bart Van Assche <bvanassche@acm.org>]

On 3/26/25 1:36 AM, Neil Armstrong wrote:
> +static irqreturn_t ufshcd_intr(int irq, void *__hba)
> +{
> +	struct ufs_hba *hba = __hba;
> +
> +	/*
> +	 * Move interrupt handling to thread when MCQ is not supported
> +	 * or when Interrupt Aggregation is not supported, leading to
> +	 * potentially longer interrupt handling.
> +	 */
> +	if (!is_mcq_supported(hba) || !ufshcd_is_intr_aggr_allowed(hba))
> +		return IRQ_WAKE_THREAD;
> +
> +	/* Directly handle interrupts since MCQ handlers does the hard job */
> +	return ufshcd_sl_intr(hba, ufshcd_readl(hba, REG_INTERRUPT_STATUS) &
> +				   ufshcd_readl(hba, REG_INTERRUPT_ENABLE));
> +}

Calling ufshcd_is_intr_aggr_allowed() from the above interrupt handler
seems wrong to me. I think you want to check whether or not ESI has been
disabled since only if ESI is disabled all I/O completions are handled
by a single interrupt if MCQ is enabled.

Thanks,

Bart.