qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
* [Qemu-devel] [PATCH v2] qdev-monitor.c: Add device id generation
@ 2015-08-27 23:14 Programmingkid
  2015-08-28  8:58 ` Daniel P. Berrange
  0 siblings, 1 reply; 2+ messages in thread
From: Programmingkid @ 2015-08-27 23:14 UTC (permalink / raw)
  To: Markus Armbruster
  Cc: Kevin Wolf, Peter Maydell, John Snow, Jeff Cody,
	qemu-devel qemu-devel, Paolo Bonzini, Andreas Färber

Add device ID generation to each device if an ID isn't given.

An auto-generated ID will begin with an underscore character. This will
distinguish it from user-made ID's. 

An user-made ID cannot begin with an underscore, so there is no
problem of collisions with auto-generated ID's. 

A management program like libvirt should have no problems with this
patch since the names for ID's it gives do not start with an underscore. 

Signed-off-by: John Arbuckle <programmingkidx@gmail.com>

---
 qdev-monitor.c |   17 ++++++++---------
 1 files changed, 8 insertions(+), 9 deletions(-)

diff --git a/qdev-monitor.c b/qdev-monitor.c
index f9e2d62..a9beb6d 100644
--- a/qdev-monitor.c
+++ b/qdev-monitor.c
@@ -574,18 +574,17 @@ DeviceState *qdev_device_add(QemuOpts *opts, Error **errp)
     id = qemu_opts_id(opts);
     if (id) {
         dev->id = id;
+    } else { /* create an ID for a device if none is provided */
+        static uint64_t deviceIDCount;
+        char *deviceIDString;
+
+        /* Add an underscore to show this ID is auto-generated */
+        deviceIDString = g_strdup_printf("_%" PRIu64, deviceIDCount++);
+        dev->id = (const char *) deviceIDString;
     }
 
-    if (dev->id) {
-        object_property_add_child(qdev_get_peripheral(), dev->id,
-                                  OBJECT(dev), NULL);
-    } else {
-        static int anon_count;
-        gchar *name = g_strdup_printf("device[%d]", anon_count++);
-        object_property_add_child(qdev_get_peripheral_anon(), name,
+    object_property_add_child(qdev_get_peripheral(), dev->id,
                                   OBJECT(dev), NULL);
-        g_free(name);
-    }
 
     /* set properties */
     if (qemu_opt_foreach(opts, set_property, dev, &err)) {
-- 
1.7.5.4

^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [Qemu-devel] [PATCH v2] qdev-monitor.c: Add device id generation
  2015-08-27 23:14 [Qemu-devel] [PATCH v2] qdev-monitor.c: Add device id generation Programmingkid
@ 2015-08-28  8:58 ` Daniel P. Berrange
  0 siblings, 0 replies; 2+ messages in thread
From: Daniel P. Berrange @ 2015-08-28  8:58 UTC (permalink / raw)
  To: Programmingkid
  Cc: Kevin Wolf, Peter Maydell, John Snow, Jeff Cody,
	Markus Armbruster, qemu-devel qemu-devel, Paolo Bonzini,
	Andreas Färber

On Thu, Aug 27, 2015 at 07:14:28PM -0400, Programmingkid wrote:
> Add device ID generation to each device if an ID isn't given.
> 
> An auto-generated ID will begin with an underscore character. This will
> distinguish it from user-made ID's. 
> 
> An user-made ID cannot begin with an underscore, so there is no
> problem of collisions with auto-generated ID's. 
> 
> A management program like libvirt should have no problems with this
> patch since the names for ID's it gives do not start with an underscore. 
> 
> Signed-off-by: John Arbuckle <programmingkidx@gmail.com>
> 
> ---
>  qdev-monitor.c |   17 ++++++++---------
>  1 files changed, 8 insertions(+), 9 deletions(-)
> 
> diff --git a/qdev-monitor.c b/qdev-monitor.c
> index f9e2d62..a9beb6d 100644
> --- a/qdev-monitor.c
> +++ b/qdev-monitor.c
> @@ -574,18 +574,17 @@ DeviceState *qdev_device_add(QemuOpts *opts, Error **errp)
>      id = qemu_opts_id(opts);
>      if (id) {
>          dev->id = id;
> +    } else { /* create an ID for a device if none is provided */
> +        static uint64_t deviceIDCount;
> +        char *deviceIDString;
> +
> +        /* Add an underscore to show this ID is auto-generated */
> +        deviceIDString = g_strdup_printf("_%" PRIu64, deviceIDCount++);
> +        dev->id = (const char *) deviceIDString;

The ID namespace is global to QOM, but this counter is scoped to the
qdev code, so we should have a further prefix, in case other bits of
code that also use QOM need to have auto-generated IDs.

IOW I think we should use

  deviceIDString = g_strdup_printf("_qdev%" PRIu64, deviceIDCount++);

>      }
>  
> -    if (dev->id) {
> -        object_property_add_child(qdev_get_peripheral(), dev->id,
> -                                  OBJECT(dev), NULL);
> -    } else {
> -        static int anon_count;
> -        gchar *name = g_strdup_printf("device[%d]", anon_count++);
> -        object_property_add_child(qdev_get_peripheral_anon(), name,
> +    object_property_add_child(qdev_get_peripheral(), dev->id,
>                                    OBJECT(dev), NULL);
> -        g_free(name);
> -    }

It looks like this is removing the only use of the /machine/peripheral-anon
part of the tre, so if we do this, then we should remove the refernces to
peripheral-anon from the rest of the codebase. Alternatively we could keep
devices with auto-generated IDs under /peripheral-anon, but I don't really
see much point. The only reason this was separated was to avoid the
auto-generated path name clashing with a user provided name, but we avoid
that now by using an underscore. So we might as well just kill off the
/peripheral-anon part of the tree and leave just /peripheral 

Regards,
Daniel
-- 
|: http://berrange.com      -o-    http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org              -o-             http://virt-manager.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org       -o-       http://live.gnome.org/gtk-vnc :|

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2015-08-28  8:58 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2015-08-27 23:14 [Qemu-devel] [PATCH v2] qdev-monitor.c: Add device id generation Programmingkid
2015-08-28  8:58 ` Daniel P. Berrange

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).