[PATCH v5 20/22] KVM: riscv: selftests: Add SBI PMU selftest

[Atish Kumar Patra]

On Tue, Apr 9, 2024 at 1:01 AM Andrew Jones <ajones at ventanamicro.com> wrote:
>
> On Mon, Apr 08, 2024 at 05:37:19PM -0700, Atish Patra wrote:
> > On 4/5/24 05:50, Andrew Jones wrote:
> > > On Wed, Apr 03, 2024 at 01:04:49AM -0700, Atish Patra wrote:
> > > ...
> > > > +static void test_pmu_basic_sanity(void)
> > > > +{
> > > > + long out_val = 0;
> > > > + bool probe;
> > > > + struct sbiret ret;
> > > > + int num_counters = 0, i;
> > > > + union sbi_pmu_ctr_info ctrinfo;
> > > > +
> > > > + probe = guest_sbi_probe_extension(SBI_EXT_PMU, &out_val);
> > > > + GUEST_ASSERT(probe && out_val == 1);
> > > > +
> > > > + num_counters = get_num_counters();
> > > > +
> > > > + for (i = 0; i < num_counters; i++) {
> > > > +         ret = sbi_ecall(SBI_EXT_PMU, SBI_EXT_PMU_COUNTER_GET_INFO, i,
> > > > +                         0, 0, 0, 0, 0);
> > > > +
> > > > +         /* There can be gaps in logical counter indicies*/
> > > > +         if (ret.error)
> > > > +                 continue;
> > > > +         GUEST_ASSERT_NE(ret.value, 0);
> > > > +
> > > > +         ctrinfo.value = ret.value;
> > > > +
> > > > +         /**
> > > > +          * Accesibillity check of hardware and read capability of firmware counters.
> > >
> > > Accessibility
> > >
> >
> > Fixed it.
> >
> > > > +          * The spec doesn't mandate any initial value. No need to check any value.
> > > > +          */
> > > > +         read_counter(i, ctrinfo);
> > > > + }
> > > > +
> > > > + GUEST_DONE();
> > > > +}
> > > > +
> > > > +static void run_vcpu(struct kvm_vcpu *vcpu)
> > > > +{
> > > > + struct ucall uc;
> > > > +
> > > > + vcpu_run(vcpu);
> > > > + switch (get_ucall(vcpu, &uc)) {
> > > > + case UCALL_ABORT:
> > > > +         REPORT_GUEST_ASSERT(uc);
> > > > +         break;
> > > > + case UCALL_DONE:
> > > > + case UCALL_SYNC:
> > > > +         break;
> > > > + default:
> > > > +         TEST_FAIL("Unknown ucall %lu", uc.cmd);
> > > > +         break;
> > > > + }
> > > > +}
> > > > +
> > > > +void test_vm_destroy(struct kvm_vm *vm)
> > > > +{
> > > > + memset(ctrinfo_arr, 0, sizeof(union sbi_pmu_ctr_info) * RISCV_MAX_PMU_COUNTERS);
> > > > + counter_mask_available = 0;
> > > > + kvm_vm_free(vm);
> > > > +}
> > > > +
> > > > +static void test_vm_basic_test(void *guest_code)
> > > > +{
> > > > + struct kvm_vm *vm;
> > > > + struct kvm_vcpu *vcpu;
> > > > +
> > > > + vm = vm_create_with_one_vcpu(&vcpu, guest_code);
> > > > + __TEST_REQUIRE(__vcpu_has_sbi_ext(vcpu, KVM_RISCV_SBI_EXT_PMU),
> > > > +                            "SBI PMU not available, skipping test");
> > > > + vm_init_vector_tables(vm);
> > > > + /* Illegal instruction handler is required to verify read access without configuration */
> > > > + vm_install_exception_handler(vm, EXC_INST_ILLEGAL, guest_illegal_exception_handler);
> > >
> > > I still don't see where the "verify" part is. The handler doesn't record
> > > that it had to handle anything.
> > >
> >
> > The objective of the test is to ensure that we get an illegal instruction
> > without configuration.
>
> This part I guessed.
>
> > The presence of the registered exception handler is
> > sufficient for that.
>
> This part I disagree with. The handler may not be necessary and not run if
> we don't get the ILL. Usually when I write tests like these I set a
> boolean in the handler and check it after the instruction which should
> have sent us there to make sure we did indeed go there.
>

Ahh I got your point now. That makes sense. Since it was just a sanity test,
I hadn't put the boolean check earlier. But you are correct about bugs
in kvm code which wouldn't
generate an expected ILL .

I have added it. Thanks for the suggestion :)

> >
> > The verify part is that the test doesn't end up in a illegal instruction
> > exception when you try to access a counter without configuring.
> >
> > Let me know if you think we should more verbose comment to explain the
> > scenario.
> >
>
> With a boolean the test code will be mostly self documenting, but a short
> comment saying why we expect the boolean to be set would be good too.
>
> Thanks,
> drew
>
> >
> > > > +
> > > > + vcpu_init_vector_tables(vcpu);
> > > > + run_vcpu(vcpu);
> > > > +
> > > > + test_vm_destroy(vm);
> > > > +}
> > > > +
> > > > +static void test_vm_events_test(void *guest_code)
> > > > +{
> > > > + struct kvm_vm *vm = NULL;
> > > > + struct kvm_vcpu *vcpu = NULL;
> > > > +
> > > > + vm = vm_create_with_one_vcpu(&vcpu, guest_code);
> > > > + __TEST_REQUIRE(__vcpu_has_sbi_ext(vcpu, KVM_RISCV_SBI_EXT_PMU),
> > > > +                            "SBI PMU not available, skipping test");
> > > > + run_vcpu(vcpu);
> > > > +
> > > > + test_vm_destroy(vm);
> > > > +}
> > > > +
> > > > +int main(void)
> > > > +{
> > > > + test_vm_basic_test(test_pmu_basic_sanity);
> > > > + pr_info("SBI PMU basic test : PASS\n");
> > > > +
> > > > + test_vm_events_test(test_pmu_events);
> > > > + pr_info("SBI PMU event verification test : PASS\n");
> > > > +
> > > > + return 0;
> > > > +}
> > > > --
> > > > 2.34.1
> > > >
> > >
> > > Thanks,
> > > drew
> >