[PATCH v3 5/9] KVM: arm64: selftests: Add shadow_stage2 test

[Itaru Kitayama]

On Sat, May 16, 2026 at 07:29:59PM +0100, Wei-Lin Chang wrote:
> The shadow_stage2 test is aimed to exercise the shadow page table
> management code in KVM. In this first patch a basic test similar to
> hello_nested is created. Right now it doesn't turn on stage-2 for the
> nested guest (L2) yet, therefore the shadow page table code in KVM will
> only be triggered minimally now.
> 
> Signed-off-by: Wei-Lin Chang <weilin.chang at arm.com>
> ---
>  tools/testing/selftests/kvm/Makefile.kvm      |   1 +
>  .../selftests/kvm/arm64/shadow_stage2.c       | 128 ++++++++++++++++++
>  2 files changed, 129 insertions(+)
>  create mode 100644 tools/testing/selftests/kvm/arm64/shadow_stage2.c
> 
> diff --git a/tools/testing/selftests/kvm/Makefile.kvm b/tools/testing/selftests/kvm/Makefile.kvm
> index e8c108e0c487..c0fac2ba1339 100644
> --- a/tools/testing/selftests/kvm/Makefile.kvm
> +++ b/tools/testing/selftests/kvm/Makefile.kvm
> @@ -176,6 +176,7 @@ TEST_GEN_PROGS_arm64 += arm64/page_fault_test
>  TEST_GEN_PROGS_arm64 += arm64/psci_test
>  TEST_GEN_PROGS_arm64 += arm64/sea_to_user
>  TEST_GEN_PROGS_arm64 += arm64/set_id_regs
> +TEST_GEN_PROGS_arm64 += arm64/shadow_stage2
>  TEST_GEN_PROGS_arm64 += arm64/smccc_filter
>  TEST_GEN_PROGS_arm64 += arm64/vcpu_width_config
>  TEST_GEN_PROGS_arm64 += arm64/vgic_init
> diff --git a/tools/testing/selftests/kvm/arm64/shadow_stage2.c b/tools/testing/selftests/kvm/arm64/shadow_stage2.c
> new file mode 100644
> index 000000000000..cf76a2b0582d
> --- /dev/null
> +++ b/tools/testing/selftests/kvm/arm64/shadow_stage2.c
> @@ -0,0 +1,128 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/*
> + * shadow_stage2 - Test correctness of shadow stage 2
> + */
> +
> +#include "nested.h"
> +#include "processor.h"
> +#include "test_util.h"
> +#include "ucall.h"
> +
> +#define XLATE2GPA	(0xABCD)
> +#define L2STACKSZ	(0x100)
> +
> +#define L2SUCCESS	(0x0)
> +#define L2FAILED	(0x1)
> +#define L2SYNC		(0x2)
> +
> +/*
> + * TPIDR_EL2 is used to store vcpu id, so save and restore it.
> + */
> +static gpa_t ucall_translate_to_gpa(void *gva)
> +{
> +	gpa_t gpa;
> +	u64 vcpu_id = read_sysreg(tpidr_el2);
> +
> +	GUEST_SYNC2(XLATE2GPA, gva);
> +
> +	/* get the result from userspace */
> +	gpa = read_sysreg(tpidr_el2);
> +
> +	write_sysreg(vcpu_id, tpidr_el2);
> +
> +	return gpa;
> +}
> +
> +static void l2_guest_code(void)
> +{
> +	do_hvc(L2SYNC, 10, 0);
> +	do_hvc(L2SYNC, 20, 0);
> +	do_hvc(L2SYNC, 30, 0);
> +
> +	do_hvc(L2SUCCESS, 0, 0);
> +}
> +
> +static void guest_code(void)
> +{
> +	struct vcpu vcpu;
> +	struct hyp_data hyp_data;
> +	int ret, i = 0;
> +	gpa_t l2_pc, l2_stack_top;
> +	/* force 16-byte alignment for the stack pointer */
> +	u8 l2_stack[L2STACKSZ] __attribute__((aligned(16)));
> +
> +	GUEST_ASSERT_EQ(get_current_el(), 2);
> +	GUEST_PRINTF("vEL2 entry\n");
> +
> +	l2_pc = ucall_translate_to_gpa(l2_guest_code);
> +	l2_stack_top = ucall_translate_to_gpa(&l2_stack[L2STACKSZ]);
> +
> +	init_vcpu(&vcpu, l2_pc, l2_stack_top);
> +	prepare_hyp();
> +
> +	while (true) {
> +		GUEST_PRINTF("L2 enter\n");
> +		ret = run_l2(&vcpu, &hyp_data);
> +		GUEST_PRINTF("L2 exit\n");
> +		GUEST_ASSERT_EQ(ret, ARM_EXCEPTION_TRAP);
> +		GUEST_ASSERT_EQ(ESR_ELx_EC(read_sysreg(esr_el2)), ESR_ELx_EC_HVC64);
> +
> +		if (vcpu.context.regs.regs[0] == L2SYNC)
> +			GUEST_SYNC3(L2SYNC, i++, vcpu.context.regs.regs[1]);
> +		else
> +			break;
> +	}
> +
> +	if (vcpu.context.regs.regs[0] != L2SUCCESS)
> +		GUEST_FAIL("L2 failed\n");
> +
> +	GUEST_PRINTF("L2 success!\n");
> +	GUEST_DONE();
> +}
> +
> +int main(void)
> +{
> +	struct kvm_vcpu_init init;
> +	struct kvm_vcpu *vcpu;
> +	struct kvm_vm *vm;
> +	struct ucall uc;
> +	gpa_t gpa;
> +
> +	TEST_REQUIRE(kvm_check_cap(KVM_CAP_ARM_EL2));
> +	vm = vm_create(1);

This assumes the default P40V48 4KB guest creation, so perhaps you'd
want to change this to __vm_create(VM_SHAPE(), 1, 0) so the test runs too on other 
16KB and 64KB kernels? I've been testing it with your stage 2 unmapping
optimization series together with this v3 KVM selftest series.

Thanks,
Itaru.

> +
> +	kvm_get_default_vcpu_target(vm, &init);
> +	init.features[0] |= BIT(KVM_ARM_VCPU_HAS_EL2);
> +	vcpu = aarch64_vcpu_add(vm, 0, &init, guest_code);
> +	kvm_arch_vm_finalize_vcpus(vm);
> +
> +	while (true) {
> +		vcpu_run(vcpu);
> +
> +		switch (get_ucall(vcpu, &uc)) {
> +		case UCALL_SYNC:
> +			if (uc.args[0] == XLATE2GPA) {
> +				gpa = addr_gva2gpa(vm, (gva_t)uc.args[1]);
> +				vcpu_set_reg(vcpu, KVM_ARM64_SYS_REG(SYS_TPIDR_EL2), gpa);
> +			}
> +			if (uc.args[0] == L2SYNC)
> +				pr_info("L2SYNC, L1 info: %ld, L2 info: %ld\n", uc.args[1], uc.args[2]);
> +			break;
> +		case UCALL_PRINTF:
> +			pr_info("[L1] %s", uc.buffer);
> +			break;
> +		case UCALL_DONE:
> +			pr_info("DONE!\n");
> +			goto end;
> +		case UCALL_ABORT:
> +			REPORT_GUEST_ASSERT(uc);
> +			fallthrough;
> +		default:
> +			TEST_FAIL("Unhandled ucall: %ld\n", uc.cmd);
> +		}
> +	}
> +
> +end:
> +	kvm_vm_free(vm);
> +	return 0;
> +}
> -- 
> 2.43.0
>