On Fri, 29 Jul 2022 12:34:51 +0200,
Venkata Prasad Potturu wrote:
>
>
> On 7/28/22 18:19, Mark Brown wrote:
> Thanks for your time.
>
> On Thu, Jul 28, 2022 at 06:10:50PM +0530, Venkata Prasad Potturu wrote:
>
> @@ -104,14 +105,13 @@ static irqreturn_t i2s_irq_handler(int irq, void *data)
>
> ext_intr_stat = readl(ACP_EXTERNAL_INTR_STAT(adata, rsrc->irqp_used));
>
> - for (i = 0; i < ACP_MAX_STREAM; i++) {
> - stream = adata->stream[i];
> + spin_lock_irqsave(&adata->acp_lock, flags);
> + list_for_each_entry(stream, &adata->stream_list, list) {
>
> If we're already in an interrupt handler here (presumably not a threaded
> one) why are we using irqsave?
>
> Yes, your statement make sense, I have followed below statement in kernel
> document. so used irqsave in interrupt context as well.
>
> We will change it to spin_lock() and send it in the next version.
>
> statement:- spin_lock_irqsave() will turn off interrupts if they are on,
> otherwise does nothing (if we are already in an interrupt handler), hence
> these functions are safe to call from any context.
Also the open and close callbacks are certainly non-irq context, hence
you can use spin_lock_irq() instead of irqsave(), too.
Takashi
