Got it, I like the solution, thanks! On Sat, May 6, 2023 at 10:59 PM Robert Hoo <robert.hoo.linux@xxxxxxxxx> wrote: > > On 5/6/2023 3:12 PM, zhuangel570 wrote: > > The "never" parameter works for environments without ITLB MULTIHIT issue. But > > for vulnerable environments, should we prohibit users from turning off > > software mitigations? > > > > As for the nx_huge_page_recovery_thread worker thread, this is a solution to > > optimize software mitigation, maybe not needed in all cases. > > For example, on a vulnerable machine, software mitigations need to be enabled, > > but worker threads may not be needed when the VM determines that huge pages > > are not in use (not sure). > > Then nx_hugepage is totally not needed:) > > > > Do you think it is possible to introduce a new parameter to disable worker > > threads? > > I suggest no. I would perceive this kthread as ingredient of nx_hugepage > solution. > > -- —————————— zhuangel570 ——————————
