Re: [PATCH] PARAM 2/4

[Rusty Russell <rusty@rustcorp.com.au>]

In message <Pine.LNX.4.44.0211170953140.4425-100000@home.transmeta.com> you wri
te:
> 
> On Sat, 16 Nov 2002, Rusty Russell wrote:
> > MODULE_PARAM is misleading and wrong.
> 
> Why is MODULE_PARAM() misleading and wrong? I think it's a lot more 
> descriptive, and these things are "modules" whether they are actually 
> compiled in or not. 

I've already conceded this to Jeff Garzik, but...

> The MM layer is just "another module".

In theory.  In practice our current build process in mm/ is not
compiled as a module.  eg: if we put in mm/readahead.c 

	module_param(debug, int, 0600);

It'd be called "readahead.debug" not "mm.debug".  Maybe that's fine,
but indicates the subtle difference.

> Also, can we please stop shouting? I'd much rather see
> 
> 	module_param(debug, int, 0600)

Sure.  MODULE_AUTHOR(), MODULE_DESCRIPTION(), MODULE_PARM_DESC() and
MODULE_LICENSE() fight the other way.  They were previously module
only, but that's an implmentation detail: plan is to expose the first
three at least through sysfs, or just leave them as stubs for
documentation purposes.

Am off to visit Extremadura to see the LinEx project: will update
patch and re-send within 24 hrs,
Rusty.
--
  Anyone who quotes me in their sig is an idiot. -- Rusty Russell.