[v2] KVM: x86/mmu: Assert on @mmu in the __kvm_mmu_invalidate_addr()

Commit Message

Like Xu May 23, 2023, 3:29 a.m. UTC

From: Like Xu <[e-mail]>

Add assertion to track that "mmu == vcpu->arch.mmu" is always true in the
context of __kvm_mmu_invalidate_addr(). for_each_shadow_entry_using_root()
and kvm_sync_spte() operate on vcpu->arch.mmu,  but the only reason that
doesn't cause explosions is because handle_invept() frees roots instead of
doing a manual invalidation.  As of now, there are no major roadblocks
to switching INVEPT emulation over to use kvm_mmu_invalidate_addr().

Suggested-by: Sean Christopherson <[e-mail]>
Signed-off-by: Like Xu <[e-mail]>
---
V1 -> V2 Changelog:
- Better to assert on @mmu instead of deleting it. (Sean)
V1: https://lore.kernel.org/kvm/[e-mail]/

 arch/x86/kvm/mmu/mmu.c | 8 ++++++++
 1 file changed, 8 insertions(+)


base-commit: f1fcbaa18b28dec10281551dfe6ed3a3ed80e3d6

Comments

Sean Christopherson June 2, 2023, 1:22 a.m. UTC | #1

On Tue, 23 May 2023 11:29:47 +0800, Like Xu wrote:
> Add assertion to track that "mmu == vcpu->arch.mmu" is always true in the
> context of __kvm_mmu_invalidate_addr(). for_each_shadow_entry_using_root()
> and kvm_sync_spte() operate on vcpu->arch.mmu,  but the only reason that
> doesn't cause explosions is because handle_invept() frees roots instead of
> doing a manual invalidation.  As of now, there are no major roadblocks
> to switching INVEPT emulation over to use kvm_mmu_invalidate_addr().
> 
> [...]

Applied to kvm-x86 mmu, thanks!

[1/1] KVM: x86/mmu: Assert on @mmu in the __kvm_mmu_invalidate_addr()
      https://github.com/kvm-x86/linux/commit/762b33eb90c9

--
https://github.com/kvm-x86/linux/tree/next
https://github.com/kvm-x86/linux/tree/fixes

Patch

diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
index c8961f45e3b1..258f12235874 100644
--- a/arch/x86/kvm/mmu/mmu.c
+++ b/arch/x86/kvm/mmu/mmu.c
@@ -5797,6 +5797,14 @@  static void __kvm_mmu_invalidate_addr(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu
 
 	vcpu_clear_mmio_info(vcpu, addr);
 
+	/*
+	 * Walking and synchronizing SPTEs both assume they are operating in
+	 * the context of the current MMU, and would need to be reworked if
+	 * this is ever used to sync the guest_mmu, e.g. to emulate INVEPT.
+	 */
+	if (WARN_ON_ONCE(mmu != vcpu->arch.mmu))
+		return;
+
 	if (!VALID_PAGE(root_hpa))
 		return;