Mathias' series to add explicit CR0.WP toggling tests, along with related cleanups from me to allow enabling forced emulation for the "core" of the access test. The potentially controversial change is to make the extra-slow variants, including forced emulation, nodefault, i.e. so called "manual" testcases. Using nodefault is the most elegant solution I could find, e.g. makes it relatively simple to run all tests on bare metal without triggering a pile of timeouts when running in a VM. v4: - Make FEP manual-only, but included in the core loop. - Mark the extra-slow (4+ minutes each in a VM) VPID-based variants manual-only - Use the standard reporting machinery so that SKIPs show up correctly. v3: - https://lore.kernel.org/all/20230403105618.41118-1-minipli@xxxxxxxxxxxxxx - Rewrite changelogs. - Use helper for guts of CR0.WP test. v2: https://lore.kernel.org/kvm/20230331135709.132713-1-minipli@xxxxxxxxxxxxxx Mathias Krause (4): x86: Use existing CR0.WP / CR4.SMEP bit definitions x86/access: CR0.WP toggling write to r/o data test x86/access: Add forced emulation support x86/access: Try forced emulation for CR0.WP test as well Sean Christopherson (5): x86/access: Replace spaces with tabs in access_test.c x86/access: Use standard pass/fail reporting machinery x86: Drop VPID invl tests from nested reduced MAXPHYADDR access testcase x86: Mark the VPID invalidation nested VMX access tests nodefault nVMX: Add forced emulation variant of #PF access test x86/access.c | 114 +++++++++++++++++++++++++++++++++++++++------- x86/access.h | 2 +- x86/access_test.c | 28 ++++++------ x86/pks.c | 5 +- x86/pku.c | 5 +- x86/unittests.cfg | 20 ++++++-- x86/vmx_tests.c | 27 ++++++++--- 7 files changed, 154 insertions(+), 47 deletions(-) base-commit: 5b5d27da2973b20ec29b18df4d749fb2190458af -- 2.40.0.348.gf938b09366-goog
