Two bug fixes and a clean up involving the TDP MMU, found by inspection. Patch 1 fixes a bug where KVM yields, e.g. due to lock contention, without performing a pending TLB flush that was required from a previous root. Patch 2 fixes a much more egregious bug where it fails to handle TDP MMU flushes in NX huge page recovery. Patch 3 explicitly disallows yielding in the TDP MMU to prevent a similar bug to patch 1 from sneaking in. v2: - Collect a review. [Ben] - Disallowing yielding instead of feeding "flush" into the TDP MMU. [Ben] - Move the yielding logic to a separate patch since it's not strictly a bug fix and it's standalone anyways (the flush feedback loop was not). v1: - https://lkml.kernel.org/r/20210319232006.3468382-1-seanjc@xxxxxxxxxx Sean Christopherson (3): KVM: x86/mmu: Ensure TLBs are flushed when yielding during GFN range zap KVM: x86/mmu: Ensure TLBs are flushed for TDP MMU during NX zapping KVM: x86/mmu: Don't allow TDP MMU to yield when recovering NX pages arch/x86/kvm/mmu/mmu.c | 9 +++++---- arch/x86/kvm/mmu/tdp_mmu.c | 26 ++++++++++++++------------ arch/x86/kvm/mmu/tdp_mmu.h | 23 ++++++++++++++++++++++- 3 files changed, 41 insertions(+), 17 deletions(-) -- 2.31.0.291.g576ba9dcdaf-goog
