linux-kernel - Re: [PATCH 2/6] mm/gup: don't pin migrated cma pages in movable zone

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 3 Dec 2020 09:58:56 -0500
From:   Pavel Tatashin <pasha.tatashin@...een.com>
To:     Michal Hocko <mhocko@...e.com>
Cc:     LKML <linux-kernel@...r.kernel.org>, linux-mm <linux-mm@...ck.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Vlastimil Babka <vbabka@...e.cz>,
        David Hildenbrand <david@...hat.com>,
        Oscar Salvador <osalvador@...e.de>,
        Dan Williams <dan.j.williams@...el.com>,
        Sasha Levin <sashal@...nel.org>,
        Tyler Hicks <tyhicks@...ux.microsoft.com>,
        Joonsoo Kim <iamjoonsoo.kim@....com>, mike.kravetz@...cle.com,
        Steven Rostedt <rostedt@...dmis.org>,
        Ingo Molnar <mingo@...hat.com>, Jason Gunthorpe <jgg@...pe.ca>,
        Peter Zijlstra <peterz@...radead.org>,
        Mel Gorman <mgorman@...e.de>,
        Matthew Wilcox <willy@...radead.org>,
        David Rientjes <rientjes@...gle.com>,
        John Hubbard <jhubbard@...dia.com>
Subject: Re: [PATCH 2/6] mm/gup: don't pin migrated cma pages in movable zone

On Thu, Dec 3, 2020 at 3:46 AM Michal Hocko <mhocko@...e.com> wrote:
>
> On Wed 02-12-20 00:23:26, Pavel Tatashin wrote:
> > In order not to fragment CMA the pinned pages are migrated. However,
> > they are migrated to ZONE_MOVABLE, which also should not have pinned pages.
> >
> > Remove __GFP_MOVABLE, so pages can be migrated to zones where pinning
> > is allowed.
>
> I was wondering why we do have __GFP_MOVABLE at all. Took a shovel
> and... 41b4dc14ee807 says:
> : We have well defined scope API to exclude CMA region.  Use it rather than
> : manipulating gfp_mask manually.  With this change, we can now restore
> : __GFP_MOVABLE for gfp_mask like as usual migration target allocation.  It
> : would result in that the ZONE_MOVABLE is also searched by page allocator.
> : For hugetlb, gfp_mask is redefined since it has a regular allocation mask
> : filter for migration target.  __GPF_NOWARN is added to hugetlb gfp_mask
> : filter since a new user for gfp_mask filter, gup, want to be silent when
> : allocation fails.
>
> This clearly states that the priority was to increase the migration
> target success rate rather than bother with the pinning aspect of the
> target page. So I believe we have simply ignored/missed the point of the
> movable zone guarantees back then and that was a mistake.
>
> > Signed-off-by: Pavel Tatashin <pasha.tatashin@...een.com>
>
> I have to admit I am not really sure about the failure path. The code is
> just too convoluted to follow. I presume the pin will fail in that case.
> Anyway this wouldn't be anything new in this path. Movable zone
> exclusion can make the failure slightly more possible in some setups but
> fundamentally nothing new there.

I've been trying to keep this series simple for easier backport, and
not to introduce new changes beside increasing the scope of pages
which are not allowed to be pinned. This area, however, requires some
inspection and fixes, something that Jason also mentioned in another
patch.

> Acked-by: Michal Hocko <mhocko@...e.com>

Thank you,
Pasha