On Fri, Dec 29, 2023 at 01:47:35AM +0000, Shuming [范書銘] wrote:

...

...

This patch makes sure that the workqueue is completed before the system

...

...

• cancel_delayed_work_sync(&rt5663->jack_detect_work);
• cancel_delayed_work_sync(&rt5663->jd_unplug_work);

...

Does something ensure that no more interrupts can fire after the work is cancelled?

As I knew, the JD source is not the wake-up source when the system suspends. This patch just makes sure the work is done to avoid the unexpected problem.

The issue isn't unexpected wakes - it's if the work manages to start running again before the system suspends. For example if a jack detect interrupt is triggered while processing the second cancel there then the jack_detect_work will be rescheduled during suspend. Interrupts will still be active until a reasonable time later in the suspend process.