On Thu, Jun 11, 2020 at 10:43:20AM +0900, Jaewon Kim wrote: > > That's fine, I simply wanted to illustrate where I thought the check > > should go to minimise the impact to the majority of allocations. > Hello Mel. > Can I understand that you also agrees on checking highatomic reserved? > Yes, I think it should be ok. > Additionally I've wondered why the number of highatomic free pages is not > accurately counted like cma free. Is there any concern on counting it? At the time, the cost of tracking it with 100% accuracy was not worth it due to a reliance on the pageblock types to protect the regions from other allocation types. -- Mel Gorman SUSE Labs
