Hi, On Mon, 2026-07-06 at 09:25 -1000, Tejun Heo wrote: > Hello, > > On Fri, Jul 03, 2026 at 03:05:38PM +0200, Thomas Hellström wrote: > > Also honor O_NONBLOCK so that if that flag is set during the > > max value write, no reclaim is initiated. The idea is to avoid > > charging the reclaim cost to the writer of the max value. > > Is this really necessary? I'm not necessarily against it but this is > trivial > to work around from userspace and feels like a gratuitous addition. > What's > the use case?
This was something that was added at the request of Maarten. The idea is to mimic the memcg behaviour. For memcg, the use-case was to avoid having the reclaim cost (I assume in terms of cpu time) land on the next allocator instead of the process writing the new max value. I think there are more details in the memcg commit history, but the presence here is solely to mimic memcg WRT this. Thanks, Thomas > > Thanks.
