On Fri, 2025-05-16 at 11:39 +0800, Chao Gao wrote: > > @@ -7368,6 +7381,9 @@ fastpath_t vmx_vcpu_run(struct kvm_vcpu *vcpu, u64 run_flags) > > if (run_flags & KVM_RUN_LOAD_GUEST_DR6) > > set_debugreg(vcpu->arch.dr6, 6); > > > > + if (run_flags & KVM_RUN_LOAD_DEBUGCTL) > > + vmx_guest_debugctl_write(vcpu, vmx_guest_debugctl_read()); > > ... > > > + > > /* > > * Refresh vmcs.HOST_CR3 if necessary. This must be done immediately > > * prior to VM-Enter, as the kernel may load a new ASID (PCID) any time > > diff --git a/arch/x86/kvm/vmx/vmx.h b/arch/x86/kvm/vmx/vmx.h > > index 1b80479505d3..5ddedf73392b 100644 > > --- a/arch/x86/kvm/vmx/vmx.h > > +++ b/arch/x86/kvm/vmx/vmx.h > > @@ -416,6 +416,8 @@ static inline void vmx_set_intercept_for_msr(struct kvm_vcpu *vcpu, u32 msr, > > > > void vmx_update_cpu_dirty_logging(struct kvm_vcpu *vcpu); > > u64 vmx_get_supported_debugctl(struct kvm_vcpu *vcpu, bool host_initiated); > > +void vmx_guest_debugctl_write(struct kvm_vcpu *vcpu, u64 val); > > +u64 vmx_guest_debugctl_read(void); > > > > /* > > * Note, early Intel manuals have the write-low and read-high bitmap offsets > > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > > index 684b8047e0f2..a85078dfa36d 100644 > > --- a/arch/x86/kvm/x86.c > > +++ b/arch/x86/kvm/x86.c > > @@ -10752,7 +10752,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu) > > dm_request_for_irq_injection(vcpu) && > > kvm_cpu_accept_dm_intr(vcpu); > > fastpath_t exit_fastpath; > > - u64 run_flags; > > + u64 run_flags, host_debug_ctl; > > > > bool req_immediate_exit = false; > > > > @@ -11024,7 +11024,10 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu) > > set_debugreg(0, 7); > > } > > > > - vcpu->arch.host_debugctl = get_debugctlmsr(); > > + host_debug_ctl = get_debugctlmsr(); > > + if (host_debug_ctl != vcpu->arch.host_debugctl) > > + run_flags |= KVM_RUN_LOAD_DEBUGCTL; > > + vcpu->arch.host_debugctl = host_debug_ctl; > > IIUC, using run_flags here may only update the GUEST_DEBUGCTL field of a vmcs02, > leaving vmcs01 not updated. Hi, Thankfully this should not be a problem because when KVM exits from vmcs02 to vmcs01 the IA32_DEBUGCTL is reset to 0. If I’m not mistaken, this always happens regardless of the VM_ENTRY_LOAD_DEBUG_CONTROLS. Now since I wrapped the write of the with a helper function (vmx_guest_debugctl_write), it should pick up the new value. Best regards, Maxim Levitsky > > > > > guest_timing_enter_irqoff(); > > > > -- > > 2.46.0 > > > > >
