Sender: Tejun Heo <htejun@gmail.com>
Date:   Thu, 23 Feb 2023 08:22:09 -1000
From:   Tejun Heo <tj@kernel.org>
To:     Jason Gunthorpe <jgg@nvidia.com>
Cc:     Yosry Ahmed <yosryahmed@google.com>,
        "T.J. Mercier" <tjmercier@google.com>,
        Alistair Popple <apopple@nvidia.com>,
        Michal Hocko <mhocko@suse.com>, linux-mm@kvack.org,
        cgroups@vger.kernel.org, linux-kernel@vger.kernel.org,
        jhubbard@nvidia.com, hannes@cmpxchg.org, surenb@google.com,
        mkoutny@suse.com, daniel@ffwll.ch,
        "Daniel P . Berrange" <berrange@redhat.com>,
        Alex Williamson <alex.williamson@redhat.com>,
        Zefan Li <lizefan.x@bytedance.com>,
        Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH 14/19] mm: Introduce a cgroup for pinned memory
Message-ID: <Y/eu0cUDGe5JNOlq@slm.duckdns.org>
References: <87o7pmnd0p.fsf@nvidia.com>
 <Y/YRJNwwvqp7nKKt@nvidia.com>
 <87k009nvnr.fsf@nvidia.com>
 <Y/bHNO7A8T3QQ5T+@nvidia.com>
 <CABdmKX18MY19bnsxN5W38Z9zmoaZx+S4+zzN_5XCYDBruwPrLg@mail.gmail.com>
 <Y/eiLBo88pgr2IUm@nvidia.com>
 <CAJD7tkadBRP22qP63-SjKSch1im9sHLoMzc6c2h10+ggbuxqMg@mail.gmail.com>
 <Y/esMBOyTaJnv5CW@nvidia.com>
 <Y/etNfwxHv8XO3iC@slm.duckdns.org>
 <Y/etrtvyaiwRIo6f@nvidia.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <Y/etrtvyaiwRIo6f@nvidia.com>
Precedence: bulk

On Thu, Feb 23, 2023 at 02:17:18PM -0400, Jason Gunthorpe wrote:
> On Thu, Feb 23, 2023 at 08:15:17AM -1000, Tejun Heo wrote:
> > On Thu, Feb 23, 2023 at 02:10:56PM -0400, Jason Gunthorpe wrote:
> > > > I am not familiar with memfd, but based on
> > > > mem_cgroup_swapin_charge_folio() it seems like if cgroup B swapped in
> > > > the pages they will remain charged to cgroup A, unless cgroup A is
> > > > removed/offlined. Am I missing something?
> > > 
> > > Ah, I don't know, Tejun said:
> > > 
> > > "but it can converge when page usage transfers across cgroups
> > > if needed."
> > > 
> > > Which I assumed was swap related but I don't know how convergence
> > > works.
> > 
> > That'd work for pagecache. For swap-backed, I think Yosry is right. Is
> > MAP_SHARED | MAP_ANONYMOUS a concern? Such mappings can only be shared
> > through forking, so it's not a common thing to be shared across different
> > resource domains.
> 
> Isn't memfd also in the same boat?

I see. Yeah, that's looks like named shared anon. The first one
instantiating a page would always be the owner.

Thanks.

-- 
tejun