Received-SPF: pass (google.com: domain of linux-kernel+bounces-158037-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Precedence: bulk
MIME-Version: 1.0
References: <20240404185034.3184582-1-pbonzini@redhat.com> <20240404185034.3184582-10-pbonzini@redhat.com>
 <ZimIfFUMPmF_dV-V@google.com>
In-Reply-To: <ZimIfFUMPmF_dV-V@google.com>
From: Paolo Bonzini <pbonzini@redhat.com>
Date: Thu, 25 Apr 2024 07:56:24 +0200
Message-ID: <CABgObfYgmuBbzQP+ZQhpm7BEgiwLTz6W0g7=EK-xwd9=CWUCOw@mail.gmail.com>
Subject: Re: [PATCH 09/11] KVM: guest_memfd: Add interface for populating gmem
 pages with user data
To: Sean Christopherson <seanjc@google.com>
Cc: linux-kernel@vger.kernel.org, kvm@vger.kernel.org, michael.roth@amd.com, 
	isaku.yamahata@intel.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Thu, Apr 25, 2024 at 12:32=E2=80=AFAM Sean Christopherson <seanjc@google=
com> wrote:
>
> On Thu, Apr 04, 2024, Paolo Bonzini wrote:
> > +long kvm_gmem_populate(struct kvm *kvm, gfn_t gfn, void __user *src, l=
ong npages,
> > +                    int (*post_populate)(struct kvm *kvm, gfn_t gfn, k=
vm_pfn_t pfn,
> > +                                         void __user *src, int order, =
void *opaque),
>
> Add a typedef for callback?  If only to make this prototype readable.
>
> > +long kvm_gmem_populate(struct kvm *kvm, gfn_t gfn, void __user *src, l=
ong npages,
> > +                    int (*post_populate)(struct kvm *kvm, gfn_t gfn, k=
vm_pfn_t pfn,
> > +                                         void __user *src, int order, =
void *opaque),
> > +                    void *opaque)
> > +{
> > +     struct file *file;
> > +     struct kvm_memory_slot *slot;
> > +
> > +     int ret =3D 0, max_order;
> > +     long i;
> > +
> > +     lockdep_assert_held(&kvm->slots_lock);
> > +     if (npages < 0)
> > +             return -EINVAL;
> > +
> > +     slot =3D gfn_to_memslot(kvm, gfn);
> > +     if (!kvm_slot_can_be_private(slot))
> > +             return -EINVAL;
> > +
> > +     file =3D kvm_gmem_get_file(slot);
> > +     if (!file)
> > +             return -EFAULT;
> > +
> > +     filemap_invalidate_lock(file->f_mapping);
> > +
> > +     npages =3D min_t(ulong, slot->npages - (gfn - slot->base_gfn), np=
ages);
> > +     for (i =3D 0; i < npages; i +=3D (1 << max_order)) {
> > +             gfn_t this_gfn =3D gfn + i;
>
> KVM usually does something like "start_gfn" or "base_gfn", and then uses =
"gfn"
> for the one gfn that's being processed.  And IMO that's much better becau=
se the
> propotype for kvm_gmem_populate() does not make it obvious that @gfn is t=
he base
> of a range, not a singular gfn to process.
>
>
> > +             kvm_pfn_t pfn;
> > +
> > +             ret =3D __kvm_gmem_get_pfn(file, slot, this_gfn, &pfn, &m=
ax_order, false);
> > +             if (ret)
> > +                     break;
> > +
> > +             if (!IS_ALIGNED(this_gfn, (1 << max_order)) ||
> > +                 (npages - i) < (1 << max_order))
> > +                     max_order =3D 0;
> > +
> > +             if (post_populate) {
>
> Is there any use for this without @post_populate?  I.e. why make this opt=
ional?

Yeah, it probably does not need to be optional (before, the
copy_from_user was optionally done from kvm_gmem_populate, but not
anymore).

Paolo