Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20221012081526.73067-1-huangjie.albert@bytedance.com>
 <2aaf2c3a-6e49-abb9-b9c8-19ce87404982@redhat.com> <CABKxMyOnxiS6RsJtQEQPrnzNn-cO0Z+OGJRCUpw1M4=WANx0Dw@mail.gmail.com>
 <2f41fc4c-68eb-ab7d-970b-fcb10f474fd4@redhat.com> <Y02YAyE1TnU6/dkA@monkey>
In-Reply-To: <Y02YAyE1TnU6/dkA@monkey>
From:   =?UTF-8?B?6buE5p2w?= <huangjie.albert@bytedance.com>
Date:   Tue, 18 Oct 2022 17:24:31 +0800
Message-ID: <CABKxMyPtm8qHRMZfetUaanqVpLCL6cVC7OJ-bdgEedz=8YYdGg@mail.gmail.com>
Subject: Re: [PATCH] mm: hugetlb: support get/set_policy for hugetlb_vm_ops
To:     Mike Kravetz <mike.kravetz@oracle.com>
Cc:     David Hildenbrand <david@redhat.com>,
        Muchun Song <songmuchun@bytedance.com>,
        Andrew Morton <akpm@linux-foundation.org>, linux-mm@kvack.org,
        linux-kernel <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Mike Kravetz <mike.kravetz@oracle.com> =E4=BA=8E2022=E5=B9=B410=E6=9C=8818=
=E6=97=A5=E5=91=A8=E4=BA=8C 01:59=E5=86=99=E9=81=93=EF=BC=9A
>
> On 10/17/22 13:33, David Hildenbrand wrote:
> > On 17.10.22 11:48, =E9=BB=84=E6=9D=B0 wrote:
> > > David Hildenbrand <david@redhat.com> =E4=BA=8E2022=E5=B9=B410=E6=9C=
=8817=E6=97=A5=E5=91=A8=E4=B8=80 16:44=E5=86=99=E9=81=93=EF=BC=9A
> > > >
> > > > On 12.10.22 10:15, Albert Huang wrote:
> > > > > From: "huangjie.albert" <huangjie.albert@bytedance.com>
> > > > >
> > > > > implement these two functions so that we can set the mempolicy to
> > > > > the inode of the hugetlb file. This ensures that the mempolicy of
> > > > > all processes sharing this huge page file is consistent.
> > > > >
> > > > > In some scenarios where huge pages are shared:
> > > > > if we need to limit the memory usage of vm within node0, so I set=
 qemu's
> > > > > mempilciy bind to node0, but if there is a process (such as virti=
ofsd)
> > > > > shared memory with the vm, in this case. If the page fault is tri=
ggered
> > > > > by virtiofsd, the allocated memory may go to node1 which  depends=
 on
> > > > > virtiofsd.
> > > > >
> > > >
> > > > Any VM that uses hugetlb should be preallocating memory. For exampl=
e,
> > > > this is the expected default under QEMU when using huge pages.
> > > >
> > > > Once preallocation does the right thing regarding NUMA policy, ther=
e is
> > > > no need to worry about it in other sub-processes.
> > > >
> > >
> > > Hi, David
> > > thanks for your reminder
> > >
> > > Yes, you are absolutely right, However, the pre-allocation mechanism
> > > does solve this problem.
> > > However, some scenarios do not like to use the pre-allocation mechani=
sm, such as
> > > scenarios that are sensitive to virtual machine startup time, or
> > > scenarios that require
> > > high memory utilization. The on-demand allocation mechanism may be be=
tter,
> > > so the key point is to find a way support for shared policy=E3=80=82
> >
> > Using hugetlb -- with a fixed pool size -- without preallocation is lik=
e
> > playing with fire. Hugetlb reservation makes one believe that on-demand
> > allocation is going to work, but there are various scenarios where that=
 can
> > go seriously wrong, and you can run out of huge pages.
>
> I absolutely agree with this cautionary note.
>
> hugetlb reservations guarantee that a sufficient number of huge pages exi=
st.
> However, there is no guarantee that those pages are on any specific node
> associated with a numa policy.  Therefore, an 'on demand' allocation coul=
d
> fail resulting in SIGBUS being set to the faulting process.
> -

Yes, supporting on-demand requires adding a lot of other code to
support, I have thought about this, but there is currently no code
that is suitable for submitting to the community.


> Mike Kravetz