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: <20220715192956.1873315-1-pgonda@google.com> <20220715192956.1873315-10-pgonda@google.com>
 <20220719154330.wnwnu23gagcya3o7@kamzik> <CAMkAt6rFO6J5heuwocmvb_wstOOwsf9WooXu9iEUOvK0wEDAhw@mail.gmail.com>
 <20220727135603.ld5torjrn4gatjb4@kamzik>
In-Reply-To: <20220727135603.ld5torjrn4gatjb4@kamzik>
From:   Peter Gonda <pgonda@google.com>
Date:   Wed, 27 Jul 2022 08:07:05 -0600
Message-ID: <CAMkAt6r9wrDZTxzGJMKV7VQBgjwdM5YouDw-=Ntv1xo+RCoxNw@mail.gmail.com>
Subject: Re: [RFC V1 08/10] KVM: selftests: Make ucall work with encrypted guests
To:     Andrew Jones <andrew.jones@linux.dev>
Cc:     kvm list <kvm@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Marc Orr <marcorr@google.com>,
        Sean Christopherson <seanjc@google.com>,
        Michael Roth <michael.roth@amd.com>,
        "Lendacky, Thomas" <thomas.lendacky@amd.com>,
        Joerg Roedel <joro@8bytes.org>,
        Mingwei Zhang <mizhang@google.com>,
        Paolo Bonzini <pbonzini@redhat.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Jul 27, 2022 at 7:56 AM Andrew Jones <andrew.jones@linux.dev> wrote:
>
> On Wed, Jul 27, 2022 at 07:38:29AM -0600, Peter Gonda wrote:
> > On Tue, Jul 19, 2022 at 9:43 AM Andrew Jones <andrew.jones@linux.dev> wrote:
> > > I'm not a big fan of mixing the concept of encrypted guests into ucalls. I
> > > think we should have two types of ucalls, those have a uc pool in memory
> > > shared with the host and those that don't. Encrypted guests pick the pool
> > > version.
> >
> > Sean suggested this version where encrypted guests and normal guests
> > used the same ucall macros/functions. I am fine with adding a second
> > interface for encrypted VM ucall, do you think macros like
> > ENCRYPTED_GUEST_SYNC, ENCRYPTED_GUEST_ASSERT, and
> > get_encrypted_ucall() ?
> >
>
> It's fine to add new functionality to ucall in order to keep the
> interfaces the same, except for initializing with some sort of indication
> that the "uc pool" version is needed. I just don't like all the references
> to encrypted guests inside ucall. ucall should implement uc pools without
> the current motivation for uc pools creeping into its implementation.

Ah that makes sense. So maybe instead of checking for 'if
(vm->memcrypt.enabled)' I should just add a new field in kvm_vm to
select for use of the uc pool? Something like kvm_vm.enable_uc_pool?

Thanks Drew!