On Sat, 5 Nov 2022 at 23:46, Alexei Starovoitov <alexei.starovoitov@xxxxxxxxx> wrote: > > On Fri, Nov 4, 2022 at 7:15 PM Dave Marchevsky <davemarchevsky@xxxxxxxx> wrote: > > > > > I was contemplating whether to simply drop this whole set_release_on_unlock > > > logic entirely. Not sure it's worth the added complexity, atleast for now. Once > > > you push you simply lose ownership of the object and any registers are > > > immediately killed. > > > > I think that being able to read / modify the datastructure node after it's been > > added is pretty critical, at least from a UX perspective. > > > > Totally fine with it being dropped from the series and experimented with > > later, though. > > Kumar, > please split release_on_unlock logic into separate patch. > afaics it doesn't have to be introduced together with these kfuncs. Ack.
