Re: [Qemu-devel] [PATCH 1/2] hw/sysbus.h: New sysbus_init_child() helper function

[Philippe Mathieu-Daudé]

On 02/16/2018 01:28 PM, Igor Mammedov wrote:
> On Fri, 16 Feb 2018 13:45:15 +0000
> Peter Maydell <address@hidden> wrote:
...
>>  static void sysbus_dev_print(Monitor *mon, DeviceState *dev, int indent);
>>  static char *sysbus_get_fw_dev_path(DeviceState *dev);
>> @@ -372,6 +373,19 @@ BusState *sysbus_get_default(void)
>>      return main_system_bus;
>>  }
>>  
>> +void sysbus_init_child(Object *parent, const char *childname,
>> +                       void *child, size_t childsize,
>> +                       const char *childtype)
>> +{
> 
> 
>> +    object_initialize(child, childsize, childtype);
>> +    /* error_abort is fine here because this can only fail for
>> +     * programming-error reasons: child already parented, or
>> +     * parent already has a child with the given name.
>> +     */
>> +    object_property_add_child(parent, childname, OBJECT(child), 
>> &error_abort);
> It would be useful not only for sysbus devices.
> maybe we should extend object_initialize instead,
> something like this:
>    void object_initialize(void *data, size_t size, const char *typename,
>                           Object *parent, const char *name)
> and set parent in it.
> git counts about 152 uses, so it would be tree wide change
> but still not too much.

we can keep object_initialize() when no parent,
and add object_initialize_child(, const char *childname, Object *parent)
'parent' last because all previous args are child-related.

> 
> 
>> +    qdev_set_parent_bus(DEVICE(child), sysbus_get_default());
> and then assuming we don't create sysbus devices, nor should be able to,
> which are not attached to sysbus_get_default() this one can go sysbus' 
> instance_init()

good idea.

> 
> then there won't be need for sysbus specific helper,
> inheritance will do the rest of the job.
> 
>> +}
>> +