On Monday, December 18, 2017 3:41:41 PM PST Huang, Ying wrote: > > A version implemented via stop_machine() could be gotten via a small > patch as below. If you still prefer stop_machine(), I can resend a > version implemented with stop_machine(). >
For the stop_machine() version, would it work to just put preempt_disable/enable at the start and end of lock_cluster() rather than introducing get/put_swap_device? Something like that might be simpler and would also disable preemption for less duration. Thanks, Junaid
