diff options
| author | Paolo Bonzini <pbonzini@redhat.com> | 2025-06-10 20:10:18 -0400 |
|---|---|---|
| committer | Paolo Bonzini <pbonzini@redhat.com> | 2025-06-12 00:43:39 -0400 |
| commit | aa2024c01a9afba6728b362626f868811ca872ee (patch) | |
| tree | b6f667cde3395a97721995715e73f8226e873b86 | |
| parent | 179a8427fcbffe36ccfed5e138d7c9b6180caff9 (diff) | |
KVM: x86/mmu: Embed direct bits into gpa for KVM_PRE_FAULT_MEMORY
Bug[*] reported for TDX case when enabling KVM_PRE_FAULT_MEMORY in QEMU.
It turns out that @gpa passed to kvm_mmu_do_page_fault() doesn't have
shared bit set when the memory attribute of it is shared, and it leads
to wrong root in tdp_mmu_get_root_for_fault().
Fix it by embedding the direct bits in the gpa that is passed to
kvm_tdp_map_page(), when the memory of the gpa is not private.
[*] https://lore.kernel.org/qemu-devel/4a757796-11c2-47f1-ae0d-335626e818fd@intel.com/
Reported-by: Xiaoyao Li <xiaoyao.li@intel.com>
Closes: https://lore.kernel.org/qemu-devel/4a757796-11c2-47f1-ae0d-335626e818fd@intel.com/
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-ID: <20250611001018.2179964-1-xiaoyao.li@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
| -rw-r--r-- | arch/x86/kvm/mmu/mmu.c | 6 |
1 files changed, 5 insertions, 1 deletions
diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c index cbc84c6abc2e..a4040578b537 100644 --- a/arch/x86/kvm/mmu/mmu.c +++ b/arch/x86/kvm/mmu/mmu.c @@ -4896,6 +4896,7 @@ long kvm_arch_vcpu_pre_fault_memory(struct kvm_vcpu *vcpu, { u64 error_code = PFERR_GUEST_FINAL_MASK; u8 level = PG_LEVEL_4K; + u64 direct_bits; u64 end; int r; @@ -4910,15 +4911,18 @@ long kvm_arch_vcpu_pre_fault_memory(struct kvm_vcpu *vcpu, if (r) return r; + direct_bits = 0; if (kvm_arch_has_private_mem(vcpu->kvm) && kvm_mem_is_private(vcpu->kvm, gpa_to_gfn(range->gpa))) error_code |= PFERR_PRIVATE_ACCESS; + else + direct_bits = gfn_to_gpa(kvm_gfn_direct_bits(vcpu->kvm)); /* * Shadow paging uses GVA for kvm page fault, so restrict to * two-dimensional paging. */ - r = kvm_tdp_map_page(vcpu, range->gpa, error_code, &level); + r = kvm_tdp_map_page(vcpu, range->gpa | direct_bits, error_code, &level); if (r < 0) return r; |