On Wed, 20 Jul 2022 19:37:58 +0200 Daniel Bristot de Oliveira <bristot@xxxxxxxxxx> wrote: > > That has logic that looks to require a lock to protect things from changing > > from underneath. > > ack, so the only variable I see we can use READ_ONCE/WRITE_ONCE is the reacting_on... WRITE_ONCE() does not supply memory barriers, which you may need. I'm only at patch 3, I'm not sure what the full requirements of that is. -- Steve
