Inject a #GP if the memory operand received by INVCPID is non-canonical.
The APM clearly states that the intercept takes priority over all #GP
checks except the CPL0 restriction.

Of course, that begs the question of how the CPU generates a linear
address in the first place.  Tracing confirms that EXITINFO1 does hold a
linear address, at least for 64-bit mode guests (hooray GS prefix).
Unfortunately, the APM says absolutely nothing about the EXITINFO fields
for INVPCID intercepts, so it's not at all clear what's supposed to
happen.

Add a FIXME to call out that KVM still does the wrong thing for 32-bit
guests, and if the stack segment is used for the memory operand.

Cc: Babu Moger <babu.moger@amd.com>
Cc: Jim Mattson <jmattson@google.com>
Fixes: 4407a797e941 ("KVM: SVM: Enable INVPCID feature on AMD")
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 arch/x86/kvm/svm/svm.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
index b8aa0f36850f..63268d940ce1 100644
--- a/arch/x86/kvm/svm/svm.c
+++ b/arch/x86/kvm/svm/svm.c
@@ -3272,6 +3272,17 @@ static int invpcid_interception(struct kvm_vcpu *vcpu)
 	type = svm->vmcb->control.exit_info_2;
 	gva = svm->vmcb->control.exit_info_1;
 
+	/*
+	 * FIXME: Perform segment checks for 32-bit mode, and inject #SS if the
+	 *        stack segment is used.  The intercept takes priority over all
+	 *        #GP checks except CPL>0, but somehow still generates a linear
+	 *        address?  The APM is sorely lacking.
+	 */
+	if (is_noncanonical_address(gva, vcpu, 0)) {
+		kvm_queue_exception_e(vcpu, GP_VECTOR, 0);
+		return 1;
+	}
+
 	return kvm_handle_invpcid(vcpu, type, gva);
 }
 

base-commit: fed48e2967f402f561d80075a20c5c9e16866e53
-- 
2.48.1.658.g4767266eb4-goog

On Mon, 24 Feb 2025 09:45:22 -0800, Sean Christopherson wrote:
> Inject a #GP if the memory operand received by INVCPID is non-canonical.
> The APM clearly states that the intercept takes priority over all #GP
> checks except the CPL0 restriction.
> 
> Of course, that begs the question of how the CPU generates a linear
> address in the first place.  Tracing confirms that EXITINFO1 does hold a
> linear address, at least for 64-bit mode guests (hooray GS prefix).
> Unfortunately, the APM says absolutely nothing about the EXITINFO fields
> for INVPCID intercepts, so it's not at all clear what's supposed to
> happen.
> 
> [...]

Applied to kvm-x86 svm, thanks!

[1/1] KVM: SVM: Inject #GP if memory operand for INVPCID is non-canonical
      https://github.com/kvm-x86/linux/commit/d4b69c3d1471

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