On Fri, Jun 14, 2019 at 04:16:55PM +0200, Christian Schneider wrote: > Am 14.06.2019 um 15:05 schrieb Guenter Roeck: > >On 6/14/19 6:52 AM, Christian Schneider wrote: > >>From: Christian Schneider <cschneider@xxxxxxxxxxxxx> > >> To: linux-hwmon > > > >The above should not be part of the e-mail body. > > > >>Alarms on gpio-fan can't be received from sysfs_notify, because the > >>wrong kobject is passed. > >>This patch fixes this. > > > >Empty line here please. > > > >>Signed-off-by: Christian Schneider <cschneider@xxxxxxxxxxxxx> > >>--- > >>diff --git a/drivers/hwmon/gpio-fan.c b/drivers/hwmon/gpio-fan.c > >>index 84753680a4e8..5f3a35db18a1 100644 > >>--- a/drivers/hwmon/gpio-fan.c > >>+++ b/drivers/hwmon/gpio-fan.c > >>@@ -54,7 +54,7 @@ static void fan_alarm_notify(struct work_struct *ws) > >> struct gpio_fan_data *fan_data = > >> container_of(ws, struct gpio_fan_data, alarm_work); > >> > >>- sysfs_notify(&fan_data->dev->kobj, NULL, "fan1_alarm"); > >>+ sysfs_notify(&fan_data->hwmon_dev->kobj, NULL, "fan1_alarm"); > >> kobject_uevent(&fan_data->dev->kobj, KOBJ_CHANGE); > >> } > >> > >You'll want to make the same change for kobject_uevent(). > > Actually, uevents are send and can be received as is. I have tried changing > the kobject for kobject_uevent(), too, and only the content of the uevent > changes. > Since I don't know what would be the "right" one, I left it as is. > If you say, that it would be correct with the hwmon_dev kobject, I'll add > this to the patch. The point of uevent notifications is to tell userspace that something changed in the provided directory. It doesn't help if the uevent does not point there. Guenter
