Messages in this thread | | | Subject | Re: PPC64: vio_find_node removal? | From | Hollis Blanchard <> | Date | Thu, 01 Jul 2004 17:26:12 -0500 |
| |
On Thu, 2004-07-01 at 14:54, Dmitry Torokhov wrote:
> kset_find_obj that is now only used by vio_find_name/vio_find_node needs to > take kobject reference otherwise use of this function is generally unsafe. > > I was looking at vio_find_name users and it is only used in rpaphp hotplug > driver. When creating a hotplug slot the function first tries to find already > existing vio node and if unsuccessfull tries to create a new one. The only > time when vio node would already exist if previous call to register_vio_slot > failed (the function does not do cleanup of created vio device node and it's > the only place where vio devices are created). So it seems to me that if > register_vio_slot would do proper cleanup we can get rid of vio_find_name/ > vio_find_node.
At boot time, all the virtual IO devices are registered and matched with their drivers (or not). Later on (possibly when loading a module), rpaphp initializes. rpaphp needs a reference to the already-registered VIO devices so that it can hotplug-remove them later by calling vio_unregister_device().
-- Hollis Blanchard IBM Linux Technology Center
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |