| Message ID | 20231212204647.2170650-28-sagis@google.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | TDX KVM selftests | expand |
On 12/12/2023 12:47 PM, Shashar, Sagi wrote: > > > -----Original Message----- > From: Sagi Shahar <sagis@google.com> > Sent: Tuesday, December 12, 2023 12:47 PM > To: linux-kselftest@vger.kernel.org; Ackerley Tng <ackerleytng@google.com>; Afranji, Ryan <afranji@google.com>; Aktas, Erdem <erdemaktas@google.com>; Sagi Shahar <sagis@google.com>; Yamahata, Isaku <isaku.yamahata@intel.com> > Cc: Sean Christopherson <seanjc@google.com>; Paolo Bonzini <pbonzini@redhat.com>; Shuah Khan <shuah@kernel.org>; Peter Gonda <pgonda@google.com>; Xu, Haibo1 <haibo1.xu@intel.com>; Chao Peng <chao.p.peng@linux.intel.com>; Annapurve, Vishal <vannapurve@google.com>; Roger Wang <runanwang@google.com>; Vipin Sharma <vipinsh@google.com>; jmattson@google.com; dmatlack@google.com; linux-kernel@vger.kernel.org; kvm@vger.kernel.org; linux-mm@kvack.org > Subject: [RFC PATCH v5 27/29] KVM: selftests: Propagate KVM_EXIT_MEMORY_FAULT to userspace > > Allow userspace to handle KVM_EXIT_MEMORY_FAULT instead of triggering TEST_ASSERT. > > From the KVM_EXIT_MEMORY_FAULT documentation: > Note! KVM_EXIT_MEMORY_FAULT is unique among all KVM exit reasons in that it accompanies a return code of '-1', not '0'! errno will always be set to EFAULT or EHWPOISON when KVM exits with KVM_EXIT_MEMORY_FAULT, userspace should assume kvm_run.exit_reason is stale/undefined for all other error numbers. If KVM exits to userspace with KVM_EXIT_MEMORY_FAULT, most likely it's because the guest attempts to access the gfn in a way that is different from what the KVM is configured, in terms of private/shared property. I'd suggest to drop this patch and work on the selftests code to eliminate this exit. If we need a testcase to catch this exit intentionally, we may call _vcpu_run() directly from the testcase and keep the common API vcpu_run() intact. > > Signed-off-by: Sagi Shahar <sagis@google.com> > --- > tools/testing/selftests/kvm/lib/kvm_util.c | 4 ++++ > 1 file changed, 4 insertions(+) > > diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c > index d024abc5379c..8fb041e51484 100644 > --- a/tools/testing/selftests/kvm/lib/kvm_util.c > +++ b/tools/testing/selftests/kvm/lib/kvm_util.c > @@ -1742,6 +1742,10 @@ void vcpu_run(struct kvm_vcpu *vcpu) { > int ret = _vcpu_run(vcpu); > > + // Allow this scenario to be handled by the caller. > + if (ret == -1 && errno == EFAULT) > + return; > + > TEST_ASSERT(!ret, KVM_IOCTL_ERROR(KVM_RUN, ret)); } > > -- > 2.43.0.472.g3155946c3a-goog >
diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c index d024abc5379c..8fb041e51484 100644 --- a/tools/testing/selftests/kvm/lib/kvm_util.c +++ b/tools/testing/selftests/kvm/lib/kvm_util.c @@ -1742,6 +1742,10 @@ void vcpu_run(struct kvm_vcpu *vcpu) { int ret = _vcpu_run(vcpu); + // Allow this scenario to be handled by the caller. + if (ret == -1 && errno == EFAULT) + return; + TEST_ASSERT(!ret, KVM_IOCTL_ERROR(KVM_RUN, ret)); }
Allow userspace to handle KVM_EXIT_MEMORY_FAULT instead of triggering TEST_ASSERT. From the KVM_EXIT_MEMORY_FAULT documentation: Note! KVM_EXIT_MEMORY_FAULT is unique among all KVM exit reasons in that it accompanies a return code of '-1', not '0'! errno will always be set to EFAULT or EHWPOISON when KVM exits with KVM_EXIT_MEMORY_FAULT, userspace should assume kvm_run.exit_reason is stale/undefined for all other error numbers. Signed-off-by: Sagi Shahar <sagis@google.com> --- tools/testing/selftests/kvm/lib/kvm_util.c | 4 ++++ 1 file changed, 4 insertions(+)