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: <ZEM5Zq8oo+xnApW9@google.com> <diqz8re2ftzb.fsf@ackerleytng-ctop.c.googlers.com>
 <ZFWli2/H5M8MZRiY@google.com>
In-Reply-To: <ZFWli2/H5M8MZRiY@google.com>
From:   Vishal Annapurve <vannapurve@google.com>
Date:   Fri, 5 May 2023 18:17:54 -0700
Message-ID: <CAGtprH9UTL5_3F4Zad38vasTWse43GEP3HHdCHtYyWkCTwu_gA@mail.gmail.com>
Subject: Re: Rename restrictedmem => guardedmem? (was: Re: [PATCH v10 0/9]
 KVM: mm: fd-based approach for supporting KVM)
To:     Sean Christopherson <seanjc@google.com>
Cc:     Ackerley Tng <ackerleytng@google.com>, david@redhat.com,
        chao.p.peng@linux.intel.com, pbonzini@redhat.com,
        vkuznets@redhat.com, jmattson@google.com, joro@8bytes.org,
        mail@maciej.szmigiero.name, vbabka@suse.cz,
        yu.c.zhang@linux.intel.com, kirill.shutemov@linux.intel.com,
        dhildenb@redhat.com, qperret@google.com, tabba@google.com,
        michael.roth@amd.com, wei.w.wang@intel.com, rppt@kernel.org,
        liam.merwick@oracle.com, isaku.yamahata@gmail.com,
        jarkko@kernel.org, kvm@vger.kernel.org,
        linux-kernel@vger.kernel.org, hughd@google.com, brauner@kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Fri, May 5, 2023 at 5:55=E2=80=AFPM Sean Christopherson <seanjc@google.c=
om> wrote:
>
> ...
> My preference is to make it a VM-scoped ioctl(), if it ends up being a KV=
M ioctl()
> and not a common syscall.  If the file isn't tightly coupled to a single =
VM, then
> punching a hole is further complicated by needing to deal with invalidati=
ng multiple
> regions that are bound to different @kvm instances.  It's not super compl=
ex, but
> AFAICT having the ioctl() be system-scoped doesn't add value, e.g. I don'=
t think
> having one VM own the memory will complicate even if/when we get to the p=
oint where
> VMs can share "private" memory, and the gmem code would still need to dea=
l with
> grabbing a module reference.

Copyless migration would be a scenario where "private" memory may need
to be shared between source and target VMs depending on how migration
support is implemented.

Regards,
Vishal