From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [PATCH] qemu-kvm: kvm headers update Date: Thu, 08 Oct 2009 15:39:56 +0200 Message-ID: <4ACDEBAC.10200@redhat.com> References: <20091005094612.GA30068@redhat.com> <20091005192720.GA3399@redhat.com> <4ACB1358.3000807@redhat.com> <200910061909.29977.arnd@arndb.de> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: "Michael S. Tsirkin" , kvm@vger.kernel.org To: Arnd Bergmann Return-path: Received: from mx1.redhat.com ([209.132.183.28]:40956 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755795AbZJHNke (ORCPT ); Thu, 8 Oct 2009 09:40:34 -0400 In-Reply-To: <200910061909.29977.arnd@arndb.de> Sender: kvm-owner@vger.kernel.org List-ID: On 10/06/2009 07:09 PM, Arnd Bergmann wrote: > On Tuesday 06 October 2009, Avi Kivity wrote: > >> On 10/05/2009 09:27 PM, Michael S. Tsirkin wrote: >> >>> >>> >>>> Or is this the output of 'make headers_install'? >>>> >>>> >>> yes. I just added include to kvm_types to add the CONFIG_ stuff. >>> Or maybe instead we should fix users to depend on the correct ifdefs. >>> >>> >>> >> I think 'make headers_install' results in non-portable headers since it >> unifdefs stuff out. >> > The point of 'make headers_install' is to turn kernel headers into > user space headers, e.g. removing address space annotations > that are undefined in user space but required in the kernel. > > Unifdef only removes parts under #ifdef __KERNEL__, and by extension, > anything under CONFIG_*, which is not allowed outside of __KERNEL__, > because it may have an entirely different meaning in user space. > > So, for building qemu, we should really use a copy of the 'installed' > version, but of course not for building the external module. > Yeah - kvm headers used to depend on CONFIG_*, but looks like this was fixed (by me - oh my memory). -- error compiling committee.c: too many arguments to function