On 1/15/20 6:48 PM, Greg Kurz wrote: > Migration can potentially race with CAS reboot. If the migration thread > completes migration after CAS has set spapr->cas_reboot but before the > mainloop could pick up the reset request and reset the machine, the > guest is migrated unrebooted and the destination doesn't reboot it > either because it isn't aware a CAS reboot was needed (eg, because a > device was added before CAS). This likely result in a broken or hung > guest. > > Even if it is small, the window between CAS and CAS reboot is enough to > re-qualify spapr->cas_reboot as state that we should migrate. Add a new > subsection for that and always send it when a CAS reboot is pending. > This may cause migration to older QEMUs to fail but it is still better > than end up with a broken guest. > > The destination cannot honour the CAS reboot request from a post load > handler because this must be done after the guest is fully restored. > It is thus done from a VM change state handler. > > Reported-by: Lukáš Doktor <ldok...@redhat.com> > Signed-off-by: Greg Kurz <gr...@kaod.org>
Cédric Le Goater <c...@kaod.org> Nice work ! That was quite complex to catch ! Thanks, C. > --- > > This patch is supposed to fix the interrupt controller mode inconsistency > between QEMU and the guest reported in this BZ: > > https://bugzilla.redhat.com/show_bug.cgi?id=1781315 (requires auth) > > Even if interrupt controller selection doesn't involve CAS reboot anymore, > we still have other conditions that require CAS reboot. > --- > hw/ppc/spapr.c | 43 +++++++++++++++++++++++++++++++++++++++++++ > 1 file changed, 43 insertions(+) > > diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c > index 30a5fbd3bea6..bf2763aa16e5 100644 > --- a/hw/ppc/spapr.c > +++ b/hw/ppc/spapr.c > @@ -1959,6 +1959,31 @@ static const VMStateDescription vmstate_spapr_dtb = { > }, > }; > > +static bool spapr_cas_reboot_needed(void *opaque) > +{ > + SpaprMachineState *spapr = SPAPR_MACHINE(opaque); > + > + /* > + * This causes the "spapr_cas_reboot" subsection to always be > + * sent if migration raced with CAS. This causes older QEMUs > + * that don't know about the subsection to fail migration but > + * it is still better than end up with a broken guest on the > + * destination. > + */ > + return spapr->cas_reboot; > +} > + > +static const VMStateDescription vmstate_spapr_cas_reboot = { > + .name = "spapr_cas_reboot", > + .version_id = 1, > + .minimum_version_id = 1, > + .needed = spapr_cas_reboot_needed, > + .fields = (VMStateField[]) { > + VMSTATE_BOOL(cas_reboot, SpaprMachineState), > + VMSTATE_END_OF_LIST() > + }, > +}; > + > static const VMStateDescription vmstate_spapr = { > .name = "spapr", > .version_id = 3, > @@ -1992,6 +2017,7 @@ static const VMStateDescription vmstate_spapr = { > &vmstate_spapr_dtb, > &vmstate_spapr_cap_large_decr, > &vmstate_spapr_cap_ccf_assist, > + &vmstate_spapr_cas_reboot, > NULL > } > }; > @@ -2577,6 +2603,21 @@ static PCIHostState *spapr_create_default_phb(void) > return PCI_HOST_BRIDGE(dev); > } > > +static void spapr_change_state_handler(void *opaque, int running, > + RunState state) > +{ > + SpaprMachineState *spapr = opaque; > + > + if (running && spapr->cas_reboot) { > + /* > + * This happens when resuming from migration if the source > + * processed a CAS but didn't have time to trigger the CAS > + * reboot. Do it now. > + */ > + qemu_system_reset_request(SHUTDOWN_CAUSE_SUBSYSTEM_RESET); > + } > +} > + > /* pSeries LPAR / sPAPR hardware init */ > static void spapr_machine_init(MachineState *machine) > { > @@ -2970,6 +3011,8 @@ static void spapr_machine_init(MachineState *machine) > > kvmppc_spapr_enable_inkernel_multitce(); > } > + > + qemu_add_vm_change_state_handler(spapr_change_state_handler, spapr); > } > > static int spapr_kvm_type(MachineState *machine, const char *vm_type) >