Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210224085915.28751-1-natet@google.com> <YDaOw48Ug7Tgr+M6@google.com>
In-Reply-To: <YDaOw48Ug7Tgr+M6@google.com>
From:   Steve Rutherford <srutherford@google.com>
Date:   Wed, 24 Feb 2021 19:55:51 -0800
Message-ID: <CABayD+f6q0q2v7pT-hjD=oP_+hAyEW5VA2WoTQNn=5-=OD1e1w@mail.gmail.com>
Subject: Re: [RFC] KVM: x86: Support KVM VMs sharing SEV context
To:     Sean Christopherson <seanjc@google.com>
Cc:     Nathan Tempelman <natet@google.com>,
        Paolo Bonzini <pbonzini@redhat.com>,
        Tom Lendacky <thomas.lendacky@amd.com>,
        X86 ML <x86@kernel.org>, KVM list <kvm@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        David Rientjes <rientjes@google.com>,
        Brijesh Singh <brijesh.singh@amd.com>,
        Ashish Kalra <Ashish.Kalra@amd.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Feb 24, 2021 at 9:37 AM Sean Christopherson <seanjc@google.com> wrote:
> > +     unsigned int asid;
> > +     int ret;
> > +
> > +     if (!sev_guest(kvm))
> > +             return -ENOTTY;
> > +
> > +     mutex_lock(&kvm->lock);
> > +
> > +     /* Mirrors of mirrors should work, but let's not get silly */
>
> Do we really care?
Yes, unless you reparent mirrors of mirrors to the original ASID
owner. If you don't do that, I think userspace could pump a chain of
mirrors to blow the kernel stack when it closes the leaf vm, since you
could build up a chain of sev_vm_destroys. Refcounting the ASIDs
directly would also fix this.

Nate's early implementation did the reparenting, but I pushed for the
simplification since it made the locking a bit hairy.
>
> > +     if (is_mirroring_enc_context(kvm)) {
> > +             ret = -ENOTTY;
> > +             goto failed;
> > +     }
> > +