Re: [patch] kernel sysfs events layer

[Daniel Stekloff]

On Fri, 2004-09-10 at 17:48, Greg KH wrote:
> On Fri, Sep 10, 2004 at 05:18:49PM -0700, Tim Hockin wrote:
> > On Fri, Sep 10, 2004 at 04:54:09PM -0700, Greg KH wrote:
> > > To send an event, the user needs to pass the kobject, a optional
> > > sysfs-attribute and the signal string to the following function:
> > >   
> > >   kobject_uevent(const char *signal,
> > >                  struct kobject *kobj,
> > >                  struct attribute *attr)
> > 
> > Sorry I missed the flare up of this topic.  What about events for which
> > there is no associated kobject?
> 
> Tough, no event for them :)
> 
> > why is the kobject argument not first?  Seems weird..
> 
> Yeah, it is a bit "odd", but it follows my old kobject_hotplug() way.
> 
> > What happened to a formatted string argument?  The signal argument can 
> > become the pre-formatted string, and someone can provide a wrapper
> > that takes a printf() like format and args.
> > 	kobject_uevent_printf(kobj, "something bad: 0x%08x", err);
> 
> Use an attribute, and have userspace read that formatted argument if
> need be.  This keeps the kernel interface much simpler, and doesn't
> allow you to abuse it for things it is not intended for (like error
> reporting stuff...)


Not to be cheeky, but cpu "overheating" isn't an error event? <grin>



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/