On Fri, Aug 21, 2020 at 07:49:54PM -0400, Peter Xu wrote:
> This is a small series that I picked up from Linus's suggestion [0] to simplify
> cow handling (and also more strict) by checking against page refcounts rather
> than mapcounts.
>
> I'm CCing the author and reviewer of commit 52d1e606ee73 on ksm ("mm: reuse
> only-pte-mapped KSM page in do_wp_page()", 2019-03-05). Please shoot if
> there's any reason to keep the logic, or it'll be removed in this series. For
> more information, please refer to [3,4].
>
> The new mm counter in the last patch can be seen as RFC, depending on whether
> anyone dislikes it... I used it majorly for observing the page reuses, so it is
> kind of optional.
>
> Two tests I did:
>
> - Run a busy loop dirty program [1] that uses 6G of memory, restrict to 1G
> RAM + 5G swap (cgroup). A few hours later, all things still look good.
> Make sure to observe (still massive) correct page reuses using the new
> counter using the last patch, probably when swapping in.
>
> - Run umapsort [2] to make sure uffd-wp will work again after applying this
> series upon master 5.9-rc1 (5.9-rc1 is broken).
>
> In all cases, I must confess it's quite pleased to post a series with diffstat
> like this... Hopefully this won't break anyone but only to make everything
> better.
>
> Please review, thanks.
>
> [0] https://lore.kernel.org/lkml/CAHk-=wjn90-=s6MBerxTuP=-FVEZtR-LpoH9eenEQ3A-QfKTZw@xxxxxxxxxxxxxx
> [1] https://github.com/xzpeter/clibs/blob/master/bsd/mig_mon/mig_mon.c
> [2] https://github.com/LLNL/umap-apps/blob/develop/src/umapsort/umapsort.cpp
> [3] https://lore.kernel.org/lkml/CAHk-=wh0syDtNzt9jGyHRV0r1pVX5gkdJWdenwmvy=dq0AL5mA@xxxxxxxxxxxxxx
> [4] https://lore.kernel.org/lkml/CAHk-=wj5Oyg0LeAxSw_vizerm=sLd=sHfcVecZMKPZn6kNbbXA@xxxxxxxxxxxxxx
Is there more comments for this series on either reviews, or suggested tests to
carry out (before a broader trial)?
Thanks,
--
Peter Xu
