Re: [PATCH 2/2] kobjects: Transmit return value of kobject_uevent_env() to caller

[Wang Chen <wangchen@cn.fujitsu.com>]

Cornelia Huck said the following on 2008-6-24 22:24:
> > On Tue, 24 Jun 2008 16:59:06 +0800,
> > Wang Chen <wangchen@cn.fujitsu.com> wrote:
> > 
>> >> The error signal shouldn't be dropped.
>> >>
>> >> Signed-off-by: Wang Chen <wangchen@cn.fujitsu.com>
>> >> ---
>> >> diff --git a/lib/kobject.c b/lib/kobject.c
>> >> index 718e510..e5de71e 100644
>> >> --- a/lib/kobject.c
>> >> +++ b/lib/kobject.c
>> >> @@ -430,7 +430,7 @@ int kobject_rename(struct kobject *kobj, const char *new_name)
>> >>  	 * Some hotplug package track interfaces by their name and
>> >>  	 * therefore want to know when the name is changed by the user. */
>> >>  	if (!error)
>> >> -		kobject_uevent_env(kobj, KOBJ_MOVE, envp);
>> >> +		error = kobject_uevent_env(kobj, KOBJ_MOVE, envp);
>> >>
>> >>  out:
>> >>  	kfree(devpath_string);
>> >> @@ -482,7 +482,7 @@ int kobject_move(struct kobject *kobj, struct kobject *new_parent)
>> >>  	kobj->parent = new_parent;
>> >>  	new_parent = NULL;
>> >>  	kobject_put(old_parent);
>> >> -	kobject_uevent_env(kobj, KOBJ_MOVE, envp);
>> >> +	error = kobject_uevent_env(kobj, KOBJ_MOVE, envp);
>> >>  out:
>> >>  	kobject_put(new_parent);
>> >>  	kobject_put(kobj);
>> >>
>> >>
>> >>
> > 
> > This looks wrong. If everything went right except sending the uevent,
> > you'll make the caller believe that the whole operation failed, while
> > in reality the sysfs operation succeeded. Either just drop the error
> > again (or print a warning), or undo the previous operations on error
> > (which may fail).
> > 

It's depended on how important sending uevent is.
If the sysfs operation succeeds, but user-space don't
get the notification. Is it ok?