Excerpts from Linus Torvalds's message of April 21, 2022 3:48 pm: > On Wed, Apr 20, 2022 at 8:25 PM Nicholas Piggin <npiggin@xxxxxxxxx> wrote: >> >> Why not just revert fac54e2bfb5b ? > > That would be stupid, with no sane way forward. Oh I missed this comment. Now I'm completely confused. Reverting the patch which caused the breakage *is* the sane way forward. Your tree is not broken after that so you're done. And if x86 wanted to select HUGE_VMALLOC in future then it can do so after fixing the issues it has with it, so that's that the sane way forward for that. What you have now is what's insane. HAVE_ARCH_HUGE_VMALLOC now means "you can ask for huge pages but it might crash in undocumented arch-specific circumstances so good luck". Thanks, Nick
