Re: kobjects, sysfs and the driver model make my head hurt

[Greg KH]

On Sun, Jul 06, 2003 at 05:33:53PM +0100, Matthew Wilcox wrote:
> 
> struct kobject * kobject_get(struct kobject * kobj)
> {
> 	if (kobj) {
> 		WARN_ON(!atomic_read(&kobj->refcount));
> 		atomic_inc(&kobj->refcount);
> 	}
> 	return kobj;
> }

That's nice.  Remember, we used to have a lock in there, that's why the
code doesn't look that clean after it was removed.

> But why return anything?  Which looks clearer?
> 
> (a)	kobj = kobject_get(kobj);

This is the way to call kobject_get(), as the object we get after the
function returns is the one we can then safely use.

> The first one makes me think that kobject_get might return a different
> kobject than the one I passed in.  That doesn't make much sense.

Think of it as, "now we can use this kobject, not the one before calling
kobject_get()".

thanks,

greg k-h