> I'm not sure this is a false-positive: if you traverse an RCU-list > without RCU watching the CPU, then a grace period may pass, the next > list can be garbage and you can go anywhere from the list. Or am I > missing something that a CPU hotplug can block a grace period? Thanks for the feedback! Yes, calling it a false positive is probably not correct, I will remove that. However, as Russell has explained in a previous email, the spinlock is fundamental for ASID handling and cannot be removed. @Russell, I will submit to the patch tracking system then.
