[PATCH] maple_tree: Fix mas_next() when already on the last node entry

[Andrew Morton]

On Tue, 17 May 2022 20:07:25 +0000 Liam Howlett <liam.howlett at oracle.com> wrote:

> > > However, as Heiko already wrote in another mail i would also like to
> > > request that the maple tree code isn't merged with the next merge
> > > window. These patches touch a lot of critical infrastructure, and i would like
> > > to have it in next for at least one development cycle, so we can be sure
> > > that we've seen and fixed most of the issues.
> > 
> > Agree, it isn't looking good.  If Linus does an -rc8 (unlikely) then
> > perhaps we'll be in better shape.
> > 
> 
> Yeah, it needs more soak time.  I've been testing on a lot of platforms
> and have a number of testcases, but it's difficult to cover this
> critical infrastructure with the variations of hardware and
> randomization of addresses.

I think we're pretty close.  Asymptotically speaking ;) I expect an
additional week would get us there but alas, we won't have it.

I've disabled the mapletree patches for now, so that Yang Shi's "Make
khugepaged collapse readonly FS THP more consistent" series can get a
run without mapletree possibly affecting its behaviour.  That's all
pushed out now.

This involved switching the order of the two series.  I think I screwed
that up last time so now I diffed all the dang files to ensure that the
end result was the same.  No more stray khugepaged_enter_vma_merge()s
hanging around.

I plan to bring back the current mapletree series once we're merged up
with Linus.  Or maybe you'd prefer to start afresh with a new series
(based on current mm-unstable, please).

New versions of patchsets worry me.  Sometimes I retain all the -fix
patches I've accumulated, check that they're all incorporated in the
new version.  Other times I'll diff the old and new trees, check that
any differences were intentional as per the changelogs.  Neither of
these approaches catches missed changelog updates.

I see a new version of MGLRU is on the wires so I expect I'll merge
that in shortly after mapletree, which will muddy the waters somewhat. 
Maybe a week after, to give mapletree some standalone time.  But MGLRU
can be disabled at compile time so that's the first question we'll
ask reporters.