On 2022-12-16 15:58:04 [+0000], Will Deacon wrote: > I guess bigeasy can give the weaker barrier a try if he has the time, but > otherwise we can leave the change as-is. I can't give a try because I have no HW. All I contributed here so far was based on what you wrote in the previous email and then I spotted the lack of the barrier of any sorts and asked about it. I _would_ assume that the cmpxchg_barrier() here would work but I'm far from knowing. If the explicit barrier after the cmpxchg_relaxed() is what you two agree on and it is better/ cheaper/ more obvious then fine. Do it ;) > Cheers, > > Will Sebastian
