Re: [PATCH] KVM: arm64: selftest: Add standalone test checking for KVM's own UUID

[Marc Zyngier <maz@xxxxxxxxxx>]

On Tue, 22 Jul 2025 10:18:10 +0100,
Andrew Jones <ajones@xxxxxxxxxxxxxxxx> wrote:
> 
> Hi Marc,
> 
> On Mon, Jul 21, 2025 at 04:51:36PM +0100, Marc Zyngier wrote:
> > Tinkering with UUIDs is a perilious task, and the KVM UUID gets
> > broken at times. In order to spot this early enough, add a selftest
> > that will shout if the expected value isn't found.
> > 
> > Signed-off-by: Marc Zyngier <maz@xxxxxxxxxx>
> > Link: https://lore.kernel.org/r/20250721130558.50823-1-jackabt.amazon@xxxxxxxxx
> > ---
> >  tools/testing/selftests/kvm/Makefile.kvm     |  1 +
> >  tools/testing/selftests/kvm/arm64/kvm-uuid.c | 67 ++++++++++++++++++++
> >  2 files changed, 68 insertions(+)
> >  create mode 100644 tools/testing/selftests/kvm/arm64/kvm-uuid.c
> > 
> > diff --git a/tools/testing/selftests/kvm/Makefile.kvm b/tools/testing/selftests/kvm/Makefile.kvm
> > index ce817a975e50a..e1eb1ba238a2a 100644
> > --- a/tools/testing/selftests/kvm/Makefile.kvm
> > +++ b/tools/testing/selftests/kvm/Makefile.kvm
> > @@ -167,6 +167,7 @@ TEST_GEN_PROGS_arm64 += arm64/vgic_irq
> >  TEST_GEN_PROGS_arm64 += arm64/vgic_lpi_stress
> >  TEST_GEN_PROGS_arm64 += arm64/vpmu_counter_access
> >  TEST_GEN_PROGS_arm64 += arm64/no-vgic-v3
> > +TEST_GEN_PROGS_arm64 += arm64/kvm-uuid
> >  TEST_GEN_PROGS_arm64 += access_tracking_perf_test
> >  TEST_GEN_PROGS_arm64 += arch_timer
> >  TEST_GEN_PROGS_arm64 += coalesced_io_test
> > diff --git a/tools/testing/selftests/kvm/arm64/kvm-uuid.c b/tools/testing/selftests/kvm/arm64/kvm-uuid.c
> > new file mode 100644
> > index 0000000000000..89d9c8b182ae5
> > --- /dev/null
> > +++ b/tools/testing/selftests/kvm/arm64/kvm-uuid.c
> > @@ -0,0 +1,67 @@
> > +#include <errno.h>
> > +#include <linux/arm-smccc.h>
> > +#include <asm/kvm.h>
> > +#include <kvm_util.h>
> > +
> > +#include "processor.h"
> > +
> > +/*
> > + * Do NOT redefine these constants, or try to replace them with some
> > + * "common" version. They are hardcoded here to detect any potential
> > + * breakage happening in the rest of the kernel.
> > + *
> > + * KVM UID value: 28b46fb6-2ec5-11e9-a9ca-4b564d003a74
> > + */
> > +#define ARM_SMCCC_VENDOR_HYP_UID_KVM_REG_0	0xb66fb428U
> > +#define ARM_SMCCC_VENDOR_HYP_UID_KVM_REG_1	0xe911c52eU
> > +#define ARM_SMCCC_VENDOR_HYP_UID_KVM_REG_2	0x564bcaa9U
> > +#define ARM_SMCCC_VENDOR_HYP_UID_KVM_REG_3	0x743a004dU
> > +
> > +static void guest_code(void)
> > +{
> > +	struct arm_smccc_res res = {};
> > +
> > +	smccc_hvc(ARM_SMCCC_VENDOR_HYP_CALL_UID_FUNC_ID, 0, 0, 0, 0, 0, 0, 0, &res);
> > +
> > +	__GUEST_ASSERT(res.a0 != SMCCC_RET_NOT_SUPPORTED, "a0 = %lx\n", res.a0);
> 
> Should this check res.a0 == SMCCC_RET_SUCCESS instead?

Yeah, probably.

> 
> > +	__GUEST_ASSERT(res.a0 == ARM_SMCCC_VENDOR_HYP_UID_KVM_REG_0 &&
> > +		       res.a1 == ARM_SMCCC_VENDOR_HYP_UID_KVM_REG_1 &&
> > +		       res.a2 == ARM_SMCCC_VENDOR_HYP_UID_KVM_REG_2 &&
> > +		       res.a3 == ARM_SMCCC_VENDOR_HYP_UID_KVM_REG_3,
> > +		       "Unexpected KVM-specific UID %lx %lx %lx %lx\n", res.a0, res.a1, res.a2, res.a3);
> > +	GUEST_DONE();
> > +}
> > +
> > +int main (int argc, char *argv[])
> > +{
> > +	struct kvm_vcpu *vcpu;
> > +	struct kvm_vm *vm;
> > +	struct ucall uc;
> > +	bool guest_done = false;
> > +
> > +	vm = vm_create_with_one_vcpu(&vcpu, guest_code);
> > +
> > +	while (!guest_done) {
> > +		vcpu_run(vcpu);
> > +
> > +		switch (get_ucall(vcpu, &uc)) {
> > +		case UCALL_SYNC:
> > +			break;
> > +		case UCALL_DONE:
> > +			guest_done = true;
> > +			break;
> > +		case UCALL_ABORT:
> > +			REPORT_GUEST_ASSERT(uc);
> > +			break;
> > +		case UCALL_PRINTF:
> > +			printf("%s", uc.buffer);
> > +			break;
> > +		default:
> > +			TEST_FAIL("Unexpected guest exit");
> > +		}
> > +	}
> 
> This is becoming a very common and useful pattern. I wonder if it's time
> for a ucall helper
> 
> static void ucall_vcpu_run(struct kvm_vcpu *vcpu,
>                            void (*sync_func)(struct kvm_vcpu *, void *),
>                            void *sync_data)
> {
>         bool guest_done = false;
>         struct ucall uc;
> 
>         while (!guest_done) {
>                 vcpu_run(vcpu);
> 
>                 switch (get_ucall(vcpu, &uc)) {
>                 case UCALL_SYNC:
>                         if (sync_func)
>                                 sync_func(vcpu, sync_data);
>                         break;
>                 case UCALL_DONE:
>                         guest_done = true;
>                         break;
>                 case UCALL_ABORT:
>                         REPORT_GUEST_ASSERT(uc);
>                         break;
>                 case UCALL_PRINTF:
>                         printf("%s", uc.buffer);
>                         break;
>                 default:
>                         TEST_FAIL("Unexpected guest exit");
>                 }
>         }
> }

Honestly, I don't know. My understanding is that the common kvm
selftest code is now mostly a pile of x86-specific stuff, and I've
made it a goal not to touch any of it.

Thanks,

	M.

-- 
Without deviation from the norm, progress is not possible.