Re: [Qemu-devel] [PATCH v2 16/16] hw/intc/nvic: Use object_initialize_child for correct reference counting

[Paolo Bonzini]

On 07/05/19 11:34, Philippe Mathieu-Daudé wrote:
> As explained in commit aff39be0ed97:
> 
>   Both functions, object_initialize() and object_property_add_child()
>   increase the reference counter of the new object, so one of the
>   references has to be dropped afterwards to get the reference
>   counting right. Otherwise the child object will not be properly
>   cleaned up when the parent gets destroyed.
>   Thus let's use now object_initialize_child() instead to get the
>   reference counting here right.
> 
> This patch was generated using the following Coccinelle script:
> 
>  @use_sysbus_init_child_obj_missing_parent@
>  expression child_ptr;
>  expression child_type;
>  expression child_size;
>  @@
>  -   object_initialize(child_ptr, child_size, child_type);
>      ...
>  -   qdev_set_parent_bus(DEVICE(child_ptr), sysbus_get_default());
>      ...
>  ?-  object_unref(OBJECT(child_ptr));
>  +   sysbus_init_child_obj(OBJECT(PARENT_OBJ), "CHILD_NAME", child_ptr,
>  +                         child_size, child_type);
> 
> We let NVIC adopt the SysTick timer.
> 
> While the object_initialize() function doesn't take an
> 'Error *errp' argument, the object_initialize_child() does.
> Since this code is used when a machine is created (and is not
> yet running), we deliberately choose to use the &error_abort
> argument instead of ignoring errors if an object creation failed.
> This choice also matches when using sysbus_init_child_obj(),
> since its code is:
> 
>   void sysbus_init_child_obj(Object *parent,
>                              const char *childname, void *child,
>                              size_t childsize, const char *childtype)
>   {
>       object_initialize_child(parent, childname, child, childsize,
>                               childtype, &error_abort, NULL);
> 
>       qdev_set_parent_bus(DEVICE(child), sysbus_get_default());
>   }
> 
> Suggested-by: Eduardo Habkost <address@hidden>
> Inspired-by: Thomas Huth <address@hidden>
> Signed-off-by: Philippe Mathieu-Daudé <address@hidden>
> ---
>  hw/intc/armv7m_nvic.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/hw/intc/armv7m_nvic.c b/hw/intc/armv7m_nvic.c
> index fff6e694e60..2334fe51426 100644
> --- a/hw/intc/armv7m_nvic.c
> +++ b/hw/intc/armv7m_nvic.c
> @@ -2568,9 +2568,9 @@ static void armv7m_nvic_realize(DeviceState *dev, Error 
> **errp)
>           * as we didn't know then if the CPU had the security extensions;
>           * so we have to do it here.
>           */
> -        object_initialize(&s->systick[M_REG_S], sizeof(s->systick[M_REG_S]),
> -                          TYPE_SYSTICK);
> -        qdev_set_parent_bus(DEVICE(&s->systick[M_REG_S]), 
> sysbus_get_default());
> +        sysbus_init_child_obj(OBJECT(dev), "systick-reg-s",
> +                              &s->systick[M_REG_S],
> +                              sizeof(s->systick[M_REG_S]), TYPE_SYSTICK);
>  
>          object_property_set_bool(OBJECT(&s->systick[M_REG_S]), true,
>                                   "realized", &err);
> 

Nice, the M_REG_NS is already using sysbus_init_child_obj. :)

Reviewed-by: Paolo Bonzini <address@hidden>

Paolo