On Thu, Sep 11, 2014 at 03:05:05PM +0200, Paolo Bonzini wrote: > Il 11/09/2014 13:30, Gleb Natapov ha scritto: > >> > + vmcs_write64(APIC_ACCESS_ADDR, page_to_phys(page)); > >> > + /* > >> > + * Do not pin apic access page in memory so that memory > >> > + * hotplug process is able to migrate it. > >> > + */ > >> > + put_page(page); > >> > } > > This code is in prepare_vmcs02() and is executed during L1->L2 vmentry. What happens > > when apic access page is migrated while L2 is running? It needs to be update somewhere. > > Before it is migrated, the MMU notifier is called and will force a > vmexit on all CPUs. The reload code will call GUP again on the page > again and swap it in. > This is how it will work without "if (!is_guest_mode(vcpu))". But, unless I am missing something, with this check it will not work while vcpu is in L2. Suppose vmcs01->APIC_ACCESS_ADDR = 0xf000. During L2 entry vmcs02->APIC_ACCESS_ADDR is set to 0xf000 too (by prepare_vmcs02). Now 0xf000 is migrated to 0x8000, mmu notifier is called, it forces vmexit, but vcpu is in a guest mode so vmcs02->APIC_ACCESS_ADDR is never updated to 0x8000 because of "if (!is_guest_mode(vcpu))" check. So what am I missing here? -- Gleb. -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
