On 2020-04-16 14:40:02 [-0400], Steven Rostedt wrote:
>
> I wonder if we should have a this_cpu_spin_lock_irqsave(), that would
> encapsulate the migrate_disable()?
>
> static inline this_cpu_spin_lock_irqsave(struct spinlock __per_cpu *lock)
> {
> migrate_disable();
> spin_lock_irqsave(this_cpu_ptr(lock));
> migrate_enable();
> }
>
I would postpone it until we have more potential users. If Uladzislau
wouldn't insist on avoidin a potential migration spot we wouldn't have
this conversation. Most code, as I remember it, is using locks so they
can access the data structure cross-CPU so the possible migration
doesn't matter much.
I would like to avoid the local-lock like zoo where we have/had a lot of
locking based function with two users or so.
>
> ?
>
> -- Steve
Sebastian
