On 2022-01-26 11:06:40 [+0100], Vlastimil Babka wrote:
> So it's like c68ed7945701 ("mm/vmstat: protect per cpu variables with
> preempt disable on RT") but we still don't want a wrapper of those
> constructs so they don't spread further, right :)
Right. We only have them because of assumption based on spin_lock_irq()
which are not true on PREEMPT_RT. Having a generic one might let people
to use them for the wrong reasons.
The commit you mentioned changed all users while here I only changed
those which missed it. Also I wasn't sure if preemption should be
disabled or interrupts (to align with the other local_irq_save()).
> Acked-by: Vlastimil Babka <vbabka@xxxxxxx>
Thank you.
Sebastian
