On Mon, 27 Mar 2023 at 16:54, Wolfram Sang <wsa@xxxxxxxxxx> wrote: > For the !CONFIG_PREEMPT_COUNT case, preemptible() is defined 0. So, > don't we lose the irqs_disabled() check in that case? Thanks for the feedback! PREEMPT_COUNT is selected by PREEMPTION, so I guess in the case of !PREEMPT_COUNT, we should be atomic (anyways)?