Re: [PATCH 3/3] mm/mincore: provide mapped status when cached status is not allowed

[Michal Hocko]

On Tue 12-02-19 14:09:03, Jiri Kosina wrote:
> On Tue, 12 Feb 2019, Michal Hocko wrote:
> 
> > I would go with patch 1 for 5.1. Patches 2 still sounds controversial or
> > incomplete to me. 
> 
> Is it because of the disagreement what 'non-blocking' really means, or do 
> you see something else missing?

Not only. See the remark from Dave [1] that the patch in its current
form seems to be incomplete. Also FS people were not involved
properly to evaluate all the potential fallouts. Even if the only way
forward is to "cripple" IOCB_NOWAIT then the documentation should go
along with the change rather than suprise people much later when the
system behaves unexpectedly. So I _think_ this patch is not really ready
yet.

Also I haven't heard any discussion whether we can reduce the effect of
the change in a similar way we do for mincore.

> Merging patch just patch 1 withouth patch 2 is probably sort of useless 
> excercise, unfortunately.

Why would that be the case. We know that mincore is the simplest way
_right now_. Closing it makes sense on its own.

[1] http://lkml.kernel.org/r/20190201014446.GU6173@dastard
-- 
Michal Hocko
SUSE Labs