hal/tools/linux hal_hotplug.c,1.14,1.15

[David Zeuthen]

Update of /cvs/hal/hal/tools/linux
In directory pdx:/tmp/cvs-serv30212a/tools/linux

Modified Files:
	hal_hotplug.c 
Log Message:
2004-07-22  David Zeuthen  <david at fubar.dk>

	* tools/linux/hal_hotplug.c (main): Apply the same hack we did to
	hal_dev.c some time ago; improves hotplugging speed; suggestion by
	Sjoerd Simons.

	* hald/linux/block_class_device.c (block_class_visit): Fixed a nasty
	bug where .udev.sysfs_path were set before other key properties. This
	resulted in lack of information in handle_udev_node_created_found_device
	that was invoked as soon as .udev.sysfs_path were set. Obviously
	this situation was only applicable when we got the DeviceEvent before
	the HotplugEvent. udev is indeed fast.

	* hald/linux/class_device.c (class_device_visit): -do-



Index: hal_hotplug.c
===================================================================
RCS file: /cvs/hal/hal/tools/linux/hal_hotplug.c,v
retrieving revision 1.14
retrieving revision 1.15
diff -u -d -r1.14 -r1.15
--- hal_hotplug.c	4 Jul 2004 20:36:12 -0000	1.14
+++ hal_hotplug.c	22 Jul 2004 19:06:02 -0000	1.15
@@ -272,6 +272,7 @@
 	DBusMessage *message;
 	DBusMessageIter iter;
 	DBusMessageIter iter_dict;
+	int rc;
 
 	if (argc != 2)
 		return 1;
@@ -279,6 +280,20 @@
 	if (get_sysfs_mnt_path () != 0)
 		return 1;
 
+	/* fork a new process so we exit instantly and child is handling
+	 * the processing; 
+	 *
+	 * TODO, FIXME, HACK, XXX : This is not the right way; we merely
+	 * work around the problem that D-BUS requires us to take a nap
+	 * before exiting because otherwise messages are lost
+	 */
+	rc = fork ();
+	if (rc == -1)
+		return 1;
+	if (rc != 0)
+		return 0;
+
+
 	openlog ("hal.hotplug", LOG_PID, LOG_USER);
 
 	/* Connect to a well-known bus instance, the system bus */