Re: [Qemu-devel] [PATCH v5 3/5] qobject: replace qobject_incref/QINCREF qobject_decref/QDECREF

[Eric Blake]

On 04/17/2018 08:36 AM, Marc-André Lureau wrote:
> Now that we can safely call QOBJECT() on QObject * as well as its
> subtypes, we can have macros qobject_ref() / qobject_unref() that work
> everywhere instead of having to use QINCREF() / QDECREF() for QObject
> and qobject_incref() / qobject_decref() for its subtypes.
> 
> Note that the new macros evaluate their argument exactly once, thus no
> need to shout them.
> 
> Signed-off-by: Marc-André Lureau <address@hidden>
> ---

> +++ b/monitor.c

> @@ -346,7 +346,7 @@ static void monitor_qmp_cleanup_req_queue_locked(Monitor 
> *mon)
>  static void monitor_qmp_cleanup_resp_queue_locked(Monitor *mon)
>  {
>      while (!g_queue_is_empty(mon->qmp.qmp_responses)) {
> -        qobject_decref(g_queue_pop_head(mon->qmp.qmp_responses));
> +        qobject_unref((QObject *)g_queue_pop_head(mon->qmp.qmp_responses));

Interesting that you have to add a cast here.  That is, the new macro is
more type-safe than the old version (since g_queue_pop_head() returns
void*); a quick audit of monitor.c shows we only ever pushed QObject*
onto the queue in the first place...

> @@ -494,9 +494,9 @@ static void monitor_json_emitter(Monitor *mon, QObject 
> *data)
>           * caller won't free the data (which will be finally freed in
>           * responder thread).
>           */
> -        qobject_incref(data);
> +        qobject_ref(data);
>          qemu_mutex_lock(&mon->qmp.qmp_queue_lock);
> -        g_queue_push_tail(mon->qmp.qmp_responses, (void *)data);
> +        g_queue_push_tail(mon->qmp.qmp_responses, data);

...but you did manage to get rid of the pointless cast to void* in the
process.

Reviewed-by: Eric Blake <address@hidden>

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org

signature.asc
Description: OpenPGP digital signature