[lm-sensors] [patch 2.6.20-git] hwmon should use subsys_init

[lm-sensors] [patch 2.6.20-git] hwmon should use subsys_init

[David Brownell]

The hwmon subsystem registers itself excessively late when it's not linked
as a module.  That can cause problems when multifunction chips, that happen
to have hwmon capabilities and live outside the hwmon subdirectory, need to
initialize earlier than on various platforms.

This patch makes the hwmon subsystem code register itself appropriately early:
as a subsystem, like other subsystems.

Signed-off-by: David Brownell <dbrownell at users.sourceforge.net>

Index: osk/drivers/hwmon/hwmon.c
=================================--- osk.orig/drivers/hwmon/hwmon.c	2006-05-02 19:18:31.000000000 -0700
+++ osk/drivers/hwmon/hwmon.c	2006-05-02 19:18:46.000000000 -0700
@@ -102,7 +102,7 @@ static void __exit hwmon_exit(void)
 	class_destroy(hwmon_class);
 }
 
-module_init(hwmon_init);
+subsys_initcall(hwmon_init);
 module_exit(hwmon_exit);
 
 EXPORT_SYMBOL_GPL(hwmon_device_register);

[lm-sensors] [patch 2.6.20-git] hwmon should use subsys_init

[Jean Delvare]

Hi David,

On Thu, 15 Feb 2007 16:03:13 -0800, David Brownell wrote:
> The hwmon subsystem registers itself excessively late when it's not linked
> as a module.  That can cause problems when multifunction chips, that happen
> to have hwmon capabilities and live outside the hwmon subdirectory, need to
> initialize earlier than on various platforms.
> 
> This patch makes the hwmon subsystem code register itself appropriately early:
> as a subsystem, like other subsystems.

This patch is already in my tree and planned for inclusion in kernel
2.6.21. I asked Linus to pull the hwmon updates (including this one)
from my tree two days ago but he did not yet. I'll ask him again
tomorrow if needed.

Thanks,
-- 
Jean Delvare

[lm-sensors] [patch 2.6.20-git] hwmon should use subsys_init

[David Brownell]

> > This patch makes the hwmon subsystem code register itself appropriately early:
> > as a subsystem, like other subsystems.
>
> This patch is already in my tree and planned for inclusion in kernel
> 2.6.21.

OK ... I thought I'd already sent that along, but didn't have a
note that I'd done so.


>	I asked Linus to pull the hwmon updates (including this one)
> from my tree two days ago but he did not yet. I'll ask him again
> tomorrow if needed.

He has to pull ARM patches too.  Right now ARM can't build, so I
can't test the latest I2C stuff ... ;)

- Dave