From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261263AbUEFHnH (ORCPT ); Thu, 6 May 2004 03:43:07 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261472AbUEFHnH (ORCPT ); Thu, 6 May 2004 03:43:07 -0400 Received: from colin2.muc.de ([193.149.48.15]:8204 "HELO colin2.muc.de") by vger.kernel.org with SMTP id S261263AbUEFHnE (ORCPT ); Thu, 6 May 2004 03:43:04 -0400 Date: 6 May 2004 09:43:02 +0200 Date: Thu, 6 May 2004 09:43:02 +0200 From: Andi Kleen To: "R. J. Wysocki" Cc: Andrew Morton , Andi Kleen , linux-kernel@vger.kernel.org Subject: Re: 2.6.6-rc3-mm2 does not build on AMD64 + essential patch is missing Message-ID: <20040506074302.GA47323@colin2.muc.de> References: <200405052210.18074.rjwysocki@sisk.pl> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200405052210.18074.rjwysocki@sisk.pl> User-Agent: Mutt/1.4.1i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, May 05, 2004 at 10:13:52PM +0200, R. J. Wysocki wrote: > Hi, > > The 2.6.6-rc3-mm2 kernel does not buld on AMD64 w/ NUMA, it appears. Here's > what the gcc says: > > CC init/version.o > LD init/built-in.o > LD .tmp_vmlinux1 > arch/x86_64/ia32/built-in.o(.text+0xa164): In function `ia32_setup_arg_pages': > : undefined reference to `mpol_set_vma_default' > kernel/built-in.o(.text+0x9ba0): In function `do_exit': > : undefined reference to `mpol_free' > make: *** [.tmp_vmlinux1] Error 1 > > (attached is the .config). Also, IMHO, the patch: Just revert the broken small-numa-api-fixups.patch patch, which never seems to have been compile tested on anything. > > --- include/asm-x86_64/processor.h.orig 2004-05-05 21:35:55.890656408 +0200 > +++ include/asm-x86_64/processor.h 2004-05-05 21:41:15.930003032 +0200 > @@ -20,6 +20,8 @@ > #include > #include > > +#define ARCH_MIN_TASKALIGN 16 > + > #define TF_MASK 0x00000100 > #define IF_MASK 0x00000200 > #define IOPL_MASK 0x00003000 > > should be applied to it, so that it does not crash at init. AFAIK Andrew fixed this in a different way. -Andi