Re: [patch, -git] isdn: hysdn_procconf.c build fix

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 30 Apr 2008 23:03:05 +0400
"Denis V. Lunev" <den@openvz.org> wrote:

> On Wed, 2008-04-30 at 11:47 -0700, Andrew Morton wrote:
> > On Wed, 30 Apr 2008 20:30:51 +0200
> > Ingo Molnar <mingo@elte.hu> wrote:
> > 
> > > 
> > > * Andrew Morton <akpm@linux-foundation.org> wrote:
> > > 
> > > > On Wed, 30 Apr 2008 19:57:40 +0200
> > > > Ingo Molnar <mingo@elte.hu> wrote:
> > > > 
> > > > > Ob'grumpy: this file was not even build-tested by any of the parties 
> > > > > who turned it into upstream commit ac41cfd19b ("isdn: use non-racy 
> > > > > method for proc entries creation").
> > > > 
> > > > That's because this stupidity:
> > > > 
> > > > config HYSDN
> > > >         tristate "Hypercope HYSDN cards (Champ, Ergo, Metro) support (module only)"
> > > >         depends on m && PROC_FS && PCI && BROKEN_ON_SMP
> > > > 
> > > > breaks allmodconfig coverage testing.
> > > 
> > > yeah. There's a trick though that we use in the x86 tree (and which i'm 
> > > using in a script when queuing up new patches). The following command 
> > > will punch through any Kconfig black holes:
> > > 
> > >   make drivers/isdn/hysdn/hysdn_procconf.o
> > 
> > That's daft, sorry.  We shouldn't expect everyone to do stuff like that.
> > 
> > > (it wont always work as it needs a halfways reasonable .config, but it 
> > > works well enough to be part of my workflow.)
> > > 
> > > note that there's tons more stuff and dependencies that plain 
> > > allmodconfig does not catch - in this merge window alone i'm at around 
> > > 20 fixes so far.
> > 
> > allmodconfig is for one thing and one thing only: compilation coverage
> > testing.  If there is anything which is not compiled after `make
> > allmodconfig' then that's a bug.
> > 
> > But how to fix it?  Perhaps we need to add a CONFIG_ALLFOOCONFIG which can
> > be tested at the appropriate places.
> 
> I think that the same should be applied to allyesconfig

That's what ALLFOOCONFIG refers to ;)

There _might_ be a need to discriminate between allyesconfig and
allmodconfig, but usually not, I expect.

Perhaps we should implement CONFIG_ALLYESCONFIG, CONFIG_ALLMODCONFIG and
CONFIG_ALLFOOCONFIG, where

	ALLFOOCONFIG = ALLYESCONFIG || ALLMODCONFIG
	
(where FOO is renamed to something sensible, but what?)