From mboxrd@z Thu Jan 1 00:00:00 1970 From: Samuel Thibault Subject: Re: stubdom: build failure Date: Wed, 24 Sep 2008 16:51:39 +0200 Message-ID: <20080924145139.GP4527@implementation.bordeaux.inria.fr> References: <200809241621.22646.Christoph.Egger@amd.com> <20080924143558.GO4527@implementation.bordeaux.inria.fr> <200809241643.35218.Christoph.Egger@amd.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable Return-path: Content-Disposition: inline In-Reply-To: <200809241643.35218.Christoph.Egger@amd.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Christoph Egger Cc: xen-devel@lists.xensource.com, Keir Fraser List-Id: xen-devel@lists.xenproject.org Christoph Egger, le Wed 24 Sep 2008 16:43:34 +0200, a =E9crit : > On Wednesday 24 September 2008 16:35:58 Samuel Thibault wrote: > > Keir Fraser, le Wed 24 Sep 2008 15:31:13 +0100, a =E9crit : > > > If it's due to pulling in too many /usr/include headers then perhap= s > > > we could copy the ones we actually need into a private local direct= ory > > > and add that to the search path instead? I guess it depends if thos= e > > > headers themselves #include any more; then it'd get messy. If they'= re > > > compiler intrinsics then perhaps they won't. > > > > In principles we only needs files that declare compiler intrinsics. >=20 > On NetBSD, compiler intrinsics don't exist. >=20 > I think, it's better to put an OS abstraction into stubdom where > only the Linux specific code uses compiler intrinsics and other Linux-o= nly=20 > stuff. > If you try (hard) to make every OS look like Linux instead, things beco= me=20 > messy. That's not the problem. We already do make mini-os have its own definitions etc. What we don't want to do is to define things like stdarg.h and such, thus including these gcc headers (what we call gcc intrinsics). Before cs 17969:433d1b26fd51, we were not using the host headers and compiler at all and building a full cross-compilation environment instead; but that takes quite some time, while we already have a compiler on the host... Samuel