| Message ID | 20230508154804.30078-1-minipli@grsecurity.net (mailing list archive) |
|---|---|
| Headers | show |
| Series | KVM CR0.WP series backport | expand |
On Mon, May 08, 2023, Mathias Krause wrote: > This is a backport of the CR0.WP KVM series[1] to Linux v5.10. It > further extends the v5.15 backport by two patches, namely patch 5 (which > is the prerequisite for Lai's patches) and patch 8 which was already > part of the v5.15.27 stable update but didn't made it to v5.10. > > I used 'ssdd 10 50000' from rt-tests[2] as a micro-benchmark, running on > a grsecurity L1 VM. Below table shows the results (runtime in seconds, > lower is better): > > legacy TDP shadow > Linux v5.10.177 10.37s 88.7s 69.7s > + patches 4.88s 4.92s 70.1s > > TDP MMU is, as for v5.15, slower than shadow paging on a vanilla kernel. > Fortunately it's disabled by default. > > The KVM unit test suite showed no regressions. > > Please consider applying. NAK, same reasoning as the 5.15 backports. https://lore.kernel.org/all/ZF1a8xIGLwcdJDVZ@google.com
This is a backport of the CR0.WP KVM series[1] to Linux v5.10. It further extends the v5.15 backport by two patches, namely patch 5 (which is the prerequisite for Lai's patches) and patch 8 which was already part of the v5.15.27 stable update but didn't made it to v5.10. I used 'ssdd 10 50000' from rt-tests[2] as a micro-benchmark, running on a grsecurity L1 VM. Below table shows the results (runtime in seconds, lower is better): legacy TDP shadow Linux v5.10.177 10.37s 88.7s 69.7s + patches 4.88s 4.92s 70.1s TDP MMU is, as for v5.15, slower than shadow paging on a vanilla kernel. Fortunately it's disabled by default. The KVM unit test suite showed no regressions. Please consider applying. Thanks, Mathias [1] https://lore.kernel.org/kvm/20230322013731.102955-1-minipli@grsecurity.net/ [2] https://git.kernel.org/pub/scm/utils/rt-tests/rt-tests.git Lai Jiangshan (4): KVM: X86: Don't reset mmu context when X86_CR4_PCIDE 1->0 KVM: X86: Don't reset mmu context when toggling X86_CR4_PGE KVM: X86: Ensure that dirty PDPTRs are loaded KVM: x86/mmu: Reconstruct shadow page root if the guest PDPTEs is changed Mathias Krause (3): KVM: x86: Do not unload MMU roots when only toggling CR0.WP with TDP enabled KVM: x86: Make use of kvm_read_cr*_bits() when testing bits KVM: VMX: Make CR0.WP a guest owned bit Paolo Bonzini (1): KVM: x86/mmu: Avoid indirect call for get_cr3 Sean Christopherson (2): KVM: x86: Read and pass all CR0/CR4 role bits to shadow MMU helper KVM: x86/mmu: Refresh CR0.WP prior to checking for emulated permission faults arch/x86/kvm/kvm_cache_regs.h | 2 +- arch/x86/kvm/mmu.h | 42 +++++++++++++++++++++++++++++++++- arch/x86/kvm/mmu/mmu.c | 32 +++++++++++++++++++------- arch/x86/kvm/mmu/paging_tmpl.h | 2 +- arch/x86/kvm/pmu.c | 4 ++-- arch/x86/kvm/vmx/nested.c | 4 ++-- arch/x86/kvm/vmx/vmx.c | 6 ++--- arch/x86/kvm/vmx/vmx.h | 18 +++++++++++++++ arch/x86/kvm/x86.c | 32 +++++++++++++++++++++----- 9 files changed, 118 insertions(+), 24 deletions(-)