Date: Sat, 9 Dec 2006 21:09:45 -0800
From: Randy Dunlap <randy.dunlap@oracle.com>
To: Nicolas Pitre <nico@cam.org>
Cc: Dmitry Torokhov <dtor@insightbb.com>, lkml <linux-kernel@vger.kernel.org>,
       akpm <akpm@osdl.org>
Subject: Re: [PATCH] ucb1400_ts depends SND_AC97_BUS
Message-Id: <20061209210945.5abe2d89.randy.dunlap@oracle.com>
In-Reply-To: <Pine.LNX.4.64.0612092212410.2630@xanadu.home>
References: <20061209003635.e778ff76.randy.dunlap@oracle.com>
	<200612092150.02940.dtor@insightbb.com>
	<20061209185737.1768315d.randy.dunlap@oracle.com>
	<200612092205.19358.dtor@insightbb.com>
	<Pine.LNX.4.64.0612092212410.2630@xanadu.home>
Organization: Oracle Linux Eng.
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3184
Lines: 79

On Sat, 09 Dec 2006 22:17:55 -0500 (EST) Nicolas Pitre wrote:

> On Sat, 9 Dec 2006, Dmitry Torokhov wrote:
> 
> > On Saturday 09 December 2006 21:57, Randy Dunlap wrote:
> > > On Sat, 9 Dec 2006 21:50:01 -0500 Dmitry Torokhov wrote:
> > > 
> > > > On Saturday 09 December 2006 03:36, Randy Dunlap wrote:
> > > > > From: Randy Dunlap <randy.dunlap@oracle.com>
> > > > > 
> > > > > This driver is an AC97 codec according to its help text.
> > > > > However, if SOUND is disabled, the "select SND_AC97_BUS"
> > > > > still inserts that into the .config file:
> > > > > 
> > > > > #
> > > > > # Sound
> > > > > #
> > > > > # CONFIG_SOUND is not set
> > > > > CONFIG_SND_AC97_BUS=m
> > > > > 
> > > > 
> > > > I consider this abug in kconfig - users of "select" should not know
> > > > full dependency chain for selected symbol.
> > > 
> > > Seems that I've heard that somewhere else.
> > > so I agree with that part.
> > > 
> > > > > Even if the config software followed dependency chains on selects,
> > > > > we should try to limit usage of "select" to library-type
> > > > > code that is needed (e.g., CRC functions) instead of bus-type
> > > > > support.
> > > > >
> > > > 
> > > > I do not agree here - the way our directory structure is laid out
> > > > "sound" comes after "Input device support" menuconfig entry.
> > > > Your patch makes user go back and forth in menuconfig, which is
> > > > awkward. I think using select is fine when an option depends on
> > > > something down the stream. If user already had a chance to select
> > > > necessary option then using "depends on" is preferred.
> > > 
> > > Traversing the menus is not difficult.
> > > (It's easier in xconfig or gconfig than menuconfig IMO,
> > > but not a big deal in any of them.)
> > >
> > 
> > I agree but many people use menuconfig and may not even be aware of
> > a driver if it is hidden because facility it depends on is not
> > selected. The same with oldconfig - unless you are closely monitor
> > all changelogs (I for example don't, I just pull from Linus) you
> > would not even get prompted for UCB1400 if you do not have sound
> > enabled.
> >  
> > > Anyway, are you saying that the only fix for this build error
> > > is to fix *config to handle select dependencies?
> > > or could propose another way to handle the build error?
> > > 
> > 
> > Would not adding "select SOUND" fix it? We could use it as a bandaid
> > until kconfig os fixed.
> 
> Please consider what SND_CONFIG_AC97_BUS corresponds to.  It is 
> sound/pci/ac97/ac97_bus.c and if you look into this file you'll see that 
> it is perfectly buildable even if sound is entirely configured out, just 
> like some lib code would be.

OK.  Should it (CONFIG_SND_AC97_BUS -> sound/pci/ac97/ac97_bus.c)
be buildable when sound is disabled?

If so, where should it be moved to (since afaik, make won't even
descend into sound/ if SOUND=n; I don't see that changing
any time soon).

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