On Tue, Dec 17, 2024, Ivan Orlov wrote:
Detect unhandleable vectoring in check_emulate_instruction to prevent infinite loop on SVM and eliminate the difference in how intercepted #PF during vectoring is handled on SVM and VMX.
Signed-off-by: Ivan Orlov iorlov@amazon.com
V1 -> V2:
- Detect the unhandleable vectoring error in svm_check_emulate_instruction
instead of handling it in the common MMU code (which is specific for cached MMIO) V2 -> V3:
- Use more generic function to check if emulation is allowed when
vectoring
arch/x86/kvm/svm/svm.c | 6 ++++++ 1 file changed, 6 insertions(+)
diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c index dd15cc635655..e89c6fc2c4e6 100644 --- a/arch/x86/kvm/svm/svm.c +++ b/arch/x86/kvm/svm/svm.c @@ -4802,6 +4802,12 @@ static int svm_check_emulate_instruction(struct kvm_vcpu *vcpu, int emul_type, bool smep, smap, is_user; u64 error_code;
- /* Check that emulation is possible during event vectoring */
- if ((to_svm(vcpu)->vmcb->control.exit_int_info &
SVM_EXITINTINFO_TYPE_MASK) &&
Let this poke out. Alternatively, and probably preferably, capture "svm" locally and it fits nicely on one line (there's an existing user of to_svm() in this helper). My objection to a local variable was specifically about a local "is_event_vectoring", not about local variables in general.
!kvm_can_emulate_event_vectoring(emul_type))return X86EMUL_UNHANDLEABLE_VECTORING;- /* Emulation is always possible when KVM has access to all guest state. */ if (!sev_guest(vcpu->kvm)) return X86EMUL_CONTINUE;
-- 2.43.0