Re: [RFC/PATCH] unregister_node() for hotplug use

[Greg KH <gregkh@suse.de>]

On Tue, May 10, 2005 at 11:58:36AM -0700, Matthew Dobson wrote:
> Greg KH wrote:
> > On Tue, May 10, 2005 at 11:15:29AM -0700, Matthew Dobson wrote:
> > 
> >>So I think it's probably a good idea to stick the __devinit on
> >>register_node() and unregister_node(), otherwise we have no marker to know
> >>which functions to remove for CONFIG_TINY.  Greg?
> > 
> > 
> > Like _anyone_ would have CONFIG_NUMA and CONFIG_TINY enabled at the same
> > time?  I don't think so...
> > 
> > I'll leave it as is for now.
> 
> No, it seems unlikely that anyone would build with CONFIG_NUMA and
> CONFIG_TINY both enabled.  But it is possible and reasonable to build with
> CONFIG_NUMA=y and CONFIG_HOTPLUG=n, which is the case I was trying to speak
> to.  If NUMA is on and HOTPLUG is off, then we're wasting kernel text
> (granted, it's a very small amount of space) for the register_node() &
> unregister_node() functions that we *know* will never be called after
> initial bootup.  That's why I suggested marking both of those functions as
> __devinit.  But it doesn't make a huge difference either way.

I do not think this is an issue, and I want to move CONFIG_HOTPLUG to be
under CONFIG_TINY anyway, so you could only disable it if TINY is
enabled.  But that's a different email thread...

thanks,

greg k-h