| Message ID | 20201027231044.655110-1-oupton@google.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Some fixes and a test for KVM_CAP_ENFORCE_PV_CPUID | expand |
Friendly ping (this series contains bugfixes) On Tue, Oct 27, 2020 at 4:10 PM Oliver Upton <oupton@google.com> wrote: > > Patches 1-2 address some small issues with documentation and the kvm selftests, > unrelated to the overall intent of this series. > > Patch 3 applies the same PV MSR filtering mechanism to guest reads of KVM > paravirt msrs. > > Patch 4 makes enabling KVM_CAP_ENFORCE_PV_FEATURE_CPUID idempotent with regards > to when userspace sets the guest's CPUID, ensuring that the cached copy of > KVM_CPUID_FEATURES.EAX is always current. > > Patch 5 fixes a regression introduced with KVM_CAP_ENFORCE_PV_CPUID, wherein > the kvm masterclock isn't updated every time the guest uses a different system > time msr than before. > > Lastly, Patch 6 introduces a test for the overall paravirtual restriction > mechanism, verifying that guests GP when touching MSRs they shouldn't and > get -KVM_ENOSYS when using restricted kvm hypercalls. Please note that this test > is dependent upon patches 1-3 of Aaron's userspace MSR test, which add support > for guest handling of the IDT in KVM selftests [1]. > > This series (along with Aaron's aforementioned changes) applies to > commit 77377064c3a9 ("KVM: ioapic: break infinite recursion on lazy > EOI"). > > [1] http://lore.kernel.org/r/20201012194716.3950330-1-aaronlewis@google.com > > Oliver Upton (6): > selftests: kvm: add tsc_msrs_test binary to gitignore > Documentation: kvm: fix ordering of msr filter, pv documentation > kvm: x86: reads of restricted pv msrs should also result in #GP > kvm: x86: ensure pv_cpuid.features is initialized when enabling cap > kvm: x86: request masterclock update any time guest uses different msr > selftests: kvm: test enforcement of paravirtual cpuid features > > Documentation/virt/kvm/api.rst | 4 +- > arch/x86/kvm/cpuid.c | 23 +- > arch/x86/kvm/cpuid.h | 1 + > arch/x86/kvm/x86.c | 38 ++- > tools/testing/selftests/kvm/.gitignore | 2 + > tools/testing/selftests/kvm/Makefile | 1 + > .../testing/selftests/kvm/include/kvm_util.h | 2 + > .../selftests/kvm/include/x86_64/processor.h | 12 + > tools/testing/selftests/kvm/lib/kvm_util.c | 28 +++ > .../selftests/kvm/lib/x86_64/processor.c | 29 +++ > .../selftests/kvm/x86_64/kvm_pv_test.c | 234 ++++++++++++++++++ > 11 files changed, 364 insertions(+), 10 deletions(-) > create mode 100644 tools/testing/selftests/kvm/x86_64/kvm_pv_test.c > > -- > 2.29.0.rc2.309.g374f81d7ae-goog >
Patches 1-2 address some small issues with documentation and the kvm selftests, unrelated to the overall intent of this series. Patch 3 applies the same PV MSR filtering mechanism to guest reads of KVM paravirt msrs. Patch 4 makes enabling KVM_CAP_ENFORCE_PV_FEATURE_CPUID idempotent with regards to when userspace sets the guest's CPUID, ensuring that the cached copy of KVM_CPUID_FEATURES.EAX is always current. Patch 5 fixes a regression introduced with KVM_CAP_ENFORCE_PV_CPUID, wherein the kvm masterclock isn't updated every time the guest uses a different system time msr than before. Lastly, Patch 6 introduces a test for the overall paravirtual restriction mechanism, verifying that guests GP when touching MSRs they shouldn't and get -KVM_ENOSYS when using restricted kvm hypercalls. Please note that this test is dependent upon patches 1-3 of Aaron's userspace MSR test, which add support for guest handling of the IDT in KVM selftests [1]. This series (along with Aaron's aforementioned changes) applies to commit 77377064c3a9 ("KVM: ioapic: break infinite recursion on lazy EOI"). [1] http://lore.kernel.org/r/20201012194716.3950330-1-aaronlewis@google.com Oliver Upton (6): selftests: kvm: add tsc_msrs_test binary to gitignore Documentation: kvm: fix ordering of msr filter, pv documentation kvm: x86: reads of restricted pv msrs should also result in #GP kvm: x86: ensure pv_cpuid.features is initialized when enabling cap kvm: x86: request masterclock update any time guest uses different msr selftests: kvm: test enforcement of paravirtual cpuid features Documentation/virt/kvm/api.rst | 4 +- arch/x86/kvm/cpuid.c | 23 +- arch/x86/kvm/cpuid.h | 1 + arch/x86/kvm/x86.c | 38 ++- tools/testing/selftests/kvm/.gitignore | 2 + tools/testing/selftests/kvm/Makefile | 1 + .../testing/selftests/kvm/include/kvm_util.h | 2 + .../selftests/kvm/include/x86_64/processor.h | 12 + tools/testing/selftests/kvm/lib/kvm_util.c | 28 +++ .../selftests/kvm/lib/x86_64/processor.c | 29 +++ .../selftests/kvm/x86_64/kvm_pv_test.c | 234 ++++++++++++++++++ 11 files changed, 364 insertions(+), 10 deletions(-) create mode 100644 tools/testing/selftests/kvm/x86_64/kvm_pv_test.c