The single-step trap #DB should be injected after the instruction completes, as 'ctxt->tf' already records the old value of 'X86_EFLAGS_TF'. Therefore, it's okay to move kvm_set_rflags() up before kvm_vcpu_do_single_step() to align it more closely with hardware behavior. No functional change intended. Suggested-by: Sean Christopherson Signed-off-by: Hou Wenlong --- arch/x86/kvm/x86.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 44c2886589d7..7352c2114bab 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -9593,10 +9593,10 @@ int x86_emulate_instruction(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, if (ctxt->is_branch) kvm_pmu_branch_retired(vcpu); kvm_rip_write(vcpu, ctxt->eip); + __kvm_set_rflags(vcpu, ctxt->eflags); if (r && (ctxt->tf || (vcpu->guest_debug & KVM_GUESTDBG_SINGLESTEP))) r = kvm_vcpu_do_singlestep(vcpu); kvm_x86_call(update_emulated_instruction)(vcpu); - __kvm_set_rflags(vcpu, ctxt->eflags); } /* -- 2.31.1