Re: 2.6.21-rc5-mm2 - compile error on x86-64

From: Andrew Morton <akpm@linux-foundation.org>
To: ebiederm@xmission.com (Eric W. Biederman)
Cc: Helge Hafting <helgehaf@aitel.hist.no>, linux-kernel@vger.kernel.org
Subject: Re: 2.6.21-rc5-mm2 - compile error on x86-64
Date: Thu, 29 Mar 2007 15:23:17 -0700	[thread overview]
Message-ID: <20070329152317.56df80d0.akpm@linux-foundation.org> (raw)
In-Reply-To: <m1mz1vem51.fsf@ebiederm.dsl.xmission.com>

On Thu, 29 Mar 2007 16:10:50 -0600
ebiederm@xmission.com (Eric W. Biederman) wrote:

> Andrew Morton <akpm@linux-foundation.org> writes:
> >
> > Ah.  I assume you have CONFIG_SYSCTL=y, CONFIG_PROC_SYSCTL=n?
> >
> >
> >
> >
> > From: Andrew Morton <akpm@linux-foundation.org>
> >
> > We're using #ifdef CONFIG_SYSCTL, but we should be using CONFIG_PROC_SYSCTL,
> > so we get
> 
> I have no problem with the patch it is clearly more correct than what we
> are doing now but I didn't think it was possible to select
> CONFIG_SYSCTL and CONFIG_PROCFS without CONFIG_PROC_SYSCTL being
> selected...
> Do we have a constraint problem in Kconfig?  Or am I misremembering things?
> 

config PROC_SYSCTL
        bool "Sysctl support (/proc/sys)" if EMBEDDED
        depends on PROC_FS
        select SYSCTL
        default y

CONFIG_PROC_SYSCTL depends on CONFIG_PROC_FS
CONFIG_PROC_SYSCTL selects CONFIG_SYSCTL

So I don't see anything preventing CONFIG_SYSCTL=y, CONFIG_PROC_FS=y,
CONFIG_PROC_SYSCTL=n.  I assume we can make this combination compile, link
and run easily enough.  I dunno if it makes any actual sense though?

I'd have thought that PROC_SYSCTL should just depend on SYSCTL.  But
SYSCTL's Kconfig setup is weird.