On Fri, 21 Dec 2018, Vlastimil Babka wrote: > > I agree that it isn't terribly urgent, we can simply blacklist the kernel > > versions where the problem occurs. If others complain I would agree it > > warrants backport to stable, but the stable rules specifically say it must > > affect more than one user? I don't think it would be best to guess. But > > when userspace was implemented to determine thp eligibility using the only > > possible means to do so, it seems important to not break that. We caught > > that it broke us, hence the patch. > > There's also Michal's series to expand THP eligibility reporting. If it > would be feasible for you to switch to that implementation in your > userspace, and perhaps only locally and temporarily add your patch to > your currently used older kernel versions, then the benefit would be one > less obscure exception in the kernel API. > > If it's infeasible then I don't think further discussion is needed and > we should abide by the do-not-break-userspace rule. I'd say stable is up > to you as well in this case, whether you're going to consume the fixed > stable versions. > Re-reading Documentation/process/stable-kernel-rules.rst it appears there is no "one user" rule, I had thought that existed in the past. So I think this patch does meet stable criteria since it is known to break userspace. Andrew, would you mind adding the following? Cc: stable@xxxxxxxxxxxxxxx [4.13+]
