From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bernhard Reutner-Fischer Date: Fri, 9 Oct 2009 13:39:02 +0200 Subject: [Buildroot] [PATCH 03/14] RESEND: handle MMU configuration In-Reply-To: <87r5tchkvv.fsf@macbook.be.48ers.dk> References: <87skduj4ut.fsf@macbook.be.48ers.dk> <20091009115343.0fae0659@surf> <87r5tchkvv.fsf@macbook.be.48ers.dk> Message-ID: <20091009113902.GM2266@mx.loc> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net On Fri, Oct 09, 2009 at 01:22:44PM +0200, Peter Korsgaard wrote: >>>>>> "Thomas" == Thomas Petazzoni writes: > > Thomas> Le Wed, 7 Oct 2009 23:27:47 +0200, > Thomas> Bernhard Reutner-Fischer a ?crit : > > >> +config BR2_HAVE_NOMMU > >> + def_bool n > >> + # assume that we have an MMU per default. > > Thomas> Maybe we should use BR2_HAVE_MMU, def_bool y. The > Thomas> non-not-negative-if-not-options are really hard to understand. > >I guess the idea is that the architectures can 'select' this option >behind the user's back. There's afaik nothing in kconfig to disable Exactly. e.g.: $ grep -B3 HAVE_NOMMU target/Config.in.arch config BR2_m68k_cpu_68000 bool "68000" select BR2_HAVE_NOMMU config BR2_m68k_cpu_68010 bool "68010" select BR2_HAVE_NOMMU -- config BR2_m68k_cpu_coldfire_51qe bool "51qe" select BR2_HAVE_NOMMU config BR2_m68k_cpu_coldfire_5202 bool "5202" select BR2_HAVE_NOMMU config BR2_m68k_cpu_coldfire_5204 bool "5204" select BR2_HAVE_NOMMU config BR2_m68k_cpu_coldfire_5206 bool "5206" select BR2_HAVE_NOMMU >another option if you enable something. > >You could ofcourse move the logic here instead of in the individual arch >handling (E.G. default y if BR2_ARCH_bfin, or what it now is). > > >> +config BR2_USE_MMU > >> + bool "Use MMU" > >> + default y > >> + depends on !BR2_HAVE_NOMMU > >> + help > >> + If your target has an MMU and you want to use it > >> + then say Y here. > > Thomas> What's the usage of these two new options ? > >Yes, I would like to know this as well. I would imagine properly >supporting nommu would take quite some more effort than this (E.G. most >packages should get hidden). why would you want to hide most of the package? For a start, it only affects the binary format.