We already do kobject_hotplug for cpu offline; this adds a
kobject_hotplug call for the online case. This is being requested by
developers of an application which wants to be notified about both
kinds of events.
Signed-off-by: Nathan Lynch <[email protected]>
Index: linux-2.6.12-rc2-mm3/drivers/base/cpu.c
===================================================================
--- linux-2.6.12-rc2-mm3.orig/drivers/base/cpu.c 2005-03-02 01:37:53.000000000 -0600
+++ linux-2.6.12-rc2-mm3/drivers/base/cpu.c 2005-04-14 10:56:29.000000000 -0500
@@ -37,6 +37,8 @@ static ssize_t store_online(struct sys_d
break;
case '1':
ret = cpu_up(cpu->sysdev.id);
+ if (!ret)
+ kobject_hotplug(&dev->kobj, KOBJ_ONLINE);
break;
default:
ret = -EINVAL;
Hi!
> We already do kobject_hotplug for cpu offline; this adds a
> kobject_hotplug call for the online case. This is being requested by
> developers of an application which wants to be notified about both
> kinds of events.
I'm afraid of bad interactions with swsusp/S3 on smp. We offline cpus there,
but we probably do not want userland to know...
Pavel
--
64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms
On Thu, Apr 21, 2005 at 08:46:56PM +0200, Pavel Machek wrote:
> Hi!
>
> > We already do kobject_hotplug for cpu offline; this adds a
> > kobject_hotplug call for the online case. This is being requested by
> > developers of an application which wants to be notified about both
> > kinds of events.
>
> I'm afraid of bad interactions with swsusp/S3 on smp. We offline cpus there,
> but we probably do not want userland to know...
The kobject_hotplug calls for cpu online and offline are present only
in the code paths where the operation is initiated through the sysfs
interface, so I don't think you have to worry.
Nathan