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;
Date:   Thu, 9 Jun 2022 17:26:19 +0000
From:   Sean Christopherson <seanjc@google.com>
To:     Andrew Jones <drjones@redhat.com>
Cc:     Marc Zyngier <maz@kernel.org>, Paolo Bonzini <pbonzini@redhat.com>,
        kvm@vger.kernel.org, Vitaly Kuznetsov <vkuznets@redhat.com>,
        David Matlack <dmatlack@google.com>,
        Ben Gardon <bgardon@google.com>,
        Oliver Upton <oupton@google.com>, linux-kernel@vger.kernel.org,
        anup@brainfault.org, Raghavendra Rao Ananta <rananta@google.com>,
        eric.auger@redhat.com
Subject: Re: [PATCH v2 000/144] KVM: selftests: Overhaul APIs, purge VCPU_ID
Message-ID: <YqItO2cbsGDSyxD8@google.com>
References: <20220603004331.1523888-1-seanjc@google.com>
 <21570ac1-e684-7983-be00-ba8b3f43a9ee@redhat.com>
 <93b87b7b5a599c1dfa47ee025f0ae9c4@kernel.org>
 <YqEupumS/m5IArTj@google.com>
 <20220609074027.fntbvcgac4nroy35@gator>
 <YqIPYP0gKIoU7JLG@google.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <YqIPYP0gKIoU7JLG@google.com>
Precedence: bulk

On Thu, Jun 09, 2022, Sean Christopherson wrote:
> On Thu, Jun 09, 2022, Andrew Jones wrote:
> > On Wed, Jun 08, 2022 at 11:20:06PM +0000, Sean Christopherson wrote:
> > > diff --git a/tools/testing/selftests/kvm/aarch64/get-reg-list.c b/tools/testing/selftests/kvm/aarch64/get-reg-list.c
> > > index b3116c151d1c..17f7ef975d5c 100644
> > > --- a/tools/testing/selftests/kvm/aarch64/get-reg-list.c
> > > +++ b/tools/testing/selftests/kvm/aarch64/get-reg-list.c
> > > @@ -419,7 +419,7 @@ static void run_test(struct vcpu_config *c)
> > > 
> > >         check_supported(c);
> > > 
> > > -       vm = vm_create_barebones();
> > > +       vm = vm_create(1);
> > 
> > Hmm, looks like something, somewhere for AArch64 needs improving to avoid
> > strangeness like this. I'll look into it after we get this series merged.
> 
> Huh, you're right, that is odd.  Ah, duh, aarch64_vcpu_add() allocates a stack
> for the vCPU, and that will fail if there's no memslot from which to allocate
> guest memory.
> 
> So, this is my goof in
> 
>   KVM: selftests: Rename vm_create() => vm_create_barebones(), drop param
> 
> get-reg-list should first be converted to vm_create_without_vcpus().  I'll also
> add a comment explaining that vm_create_barebones() can be used with __vm_vcpu_add(),
> but not the "full" vm_vcpu_add() or vm_arch_vcpu_add() variants.

Actually, I agree with your assessment.  A better solution is to open code the
calls to add and setup the vCPU.  It's a small amount of code duplication, but I
actually like the end result because it better documents the test's dependencies.

Assuming it actually works, i.e. the stack setup is truly unnecessary, I'll add a
patch like so before the barebones change.

diff --git a/tools/testing/selftests/kvm/aarch64/get-reg-list.c b/tools/testing/selftests/kvm/aarch64/get-reg-list.c
index ecfb773ec41e..7bba365b1522 100644
--- a/tools/testing/selftests/kvm/aarch64/get-reg-list.c
+++ b/tools/testing/selftests/kvm/aarch64/get-reg-list.c
@@ -418,7 +418,8 @@ static void run_test(struct vcpu_config *c)

        vm = vm_create(DEFAULT_GUEST_PHY_PAGES);
        prepare_vcpu_init(c, &init);
-       aarch64_vcpu_add_default(vm, 0, &init, NULL);
+       vm_vcpu_add(vm, vcpuid);
+       aarch64_vcpu_setup(vm, 0, &init);
        finalize_vcpu(vm, 0, c);

        reg_list = vcpu_get_reg_list(vm, 0);