> On Mar 29, 2022, at 6:21 AM, Stefan Hajnoczi <stefa...@redhat.com> wrote:
> 
> On Fri, Mar 25, 2022 at 03:19:36PM -0400, Jagannathan Raman wrote:
>  ##
>> diff --git a/include/hw/remote/machine.h b/include/hw/remote/machine.h
>> index 8d0fa98d33..2fcb9dada5 100644
>> --- a/include/hw/remote/machine.h
>> +++ b/include/hw/remote/machine.h
>> @@ -26,6 +26,12 @@ struct RemoteMachineState {
>>     bool vfio_user;
>> };
>> 
>> +struct RemoteMachineClass {
>> +    MachineClass parent_class;
>> +
>> +    bool auto_shutdown;
>> +};
>> +
>> /* Used to pass to co-routine device and ioc. */
>> typedef struct RemoteCommDev {
>>     PCIDevice *dev;
>> @@ -33,7 +39,7 @@ typedef struct RemoteCommDev {
>> } RemoteCommDev;
>> 
>> #define TYPE_REMOTE_MACHINE "x-remote-machine"
>> -OBJECT_DECLARE_SIMPLE_TYPE(RemoteMachineState, REMOTE_MACHINE)
>> +OBJECT_DECLARE_TYPE(RemoteMachineState, RemoteMachineClass, REMOTE_MACHINE)
>> 
>> void coroutine_fn mpqemu_remote_msg_loop_co(void *data);
>> 
>> diff --git a/hw/remote/machine.c b/hw/remote/machine.c
>> index a9a75e170f..70178b222c 100644
>> --- a/hw/remote/machine.c
>> +++ b/hw/remote/machine.c
>> @@ -78,25 +78,48 @@ static void remote_machine_set_vfio_user(Object *obj, 
>> bool value, Error **errp)
>>     s->vfio_user = value;
>> }
>> 
>> +static bool remote_machine_get_auto_shutdown(Object *obj, Error **errp)
>> +{
>> +    RemoteMachineClass *rmc = REMOTE_MACHINE_GET_CLASS(obj);
>> +
>> +    return rmc->auto_shutdown;
>> +}
>> +
>> +static void remote_machine_set_auto_shutdown(Object *obj, bool value,
>> +                                             Error **errp)
>> +{
>> +    RemoteMachineClass *rmc = REMOTE_MACHINE_GET_CLASS(obj);
>> +
>> +    rmc->auto_shutdown = value;
>> +}
>> +
>> static void remote_machine_class_init(ObjectClass *oc, void *data)
>> {
>>     MachineClass *mc = MACHINE_CLASS(oc);
>> +    RemoteMachineClass *rmc = REMOTE_MACHINE_CLASS(oc);
>>     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
>> 
>>     mc->init = remote_machine_init;
>>     mc->desc = "Experimental remote machine";
>> 
>> +    rmc->auto_shutdown = true;
>> +
>>     hc->unplug = qdev_simple_device_unplug_cb;
>> 
>>     object_class_property_add_bool(oc, "vfio-user",
>>                                    remote_machine_get_vfio_user,
>>                                    remote_machine_set_vfio_user);
>> +
>> +    object_class_property_add_bool(oc, "auto-shutdown",
>> +                                   remote_machine_get_auto_shutdown,
>> +                                   remote_machine_set_auto_shutdown);
>> }
>> 
>> static const TypeInfo remote_machine = {
>>     .name = TYPE_REMOTE_MACHINE,
>>     .parent = TYPE_MACHINE,
>>     .instance_size = sizeof(RemoteMachineState),
>> +    .class_size = sizeof(RemoteMachineClass),
> 
> Why is ->auto_shutdown a global RemoteMachineClass field instead of a
> RemoteMachineState instance field?
> 
> The getter/setter functions receive an object instance so they could
> access the value in RemoteMachineState instead of RemoteMachineClass.
> Moving the field to RemoteMachineState would allow multiple
> RemoteMachineState instances with different ->auto_shutdown values in
> case QEMU ever supports running multiple machines within a single
> process.

Hi Stefan,

Presently, qemu_system_shutdown_request() shuts down the entire
system. Therefore, even if a single machine requests shutdown,
the whole system would be shutdown.

You’re suggesting a future enhancement to QEMU where multiple machines
could run in the same process, and the shutdown API could presumably
target a selected machine.

Sounds good to me, will make it a RemoteMachineState instance field.

Thank you!
--
Jag

Reply via email to