From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261701AbVCUI5c (ORCPT ); Mon, 21 Mar 2005 03:57:32 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261703AbVCUI5c (ORCPT ); Mon, 21 Mar 2005 03:57:32 -0500 Received: from smtp106.mail.sc5.yahoo.com ([66.163.169.226]:37969 "HELO smtp106.mail.sc5.yahoo.com") by vger.kernel.org with SMTP id S261701AbVCUI5Z (ORCPT ); Mon, 21 Mar 2005 03:57:25 -0500 Message-ID: <423E8C78.5040405@yahoo.com> Date: Mon, 21 Mar 2005 03:57:28 -0500 From: Daniel Dickman User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: linux-kernel@vger.kernel.org Subject: question about build.c Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org x86_64 checks if we're building a big kernel to make sure that it's <= 4MB. Should we be doing something similar for i386? Or perhaps we shouldn't be imposing this limit on x86_64? Below is the relevant code diff that I'm asking about. Thanks, Daniel --- linux-2.6.12-rc1-bk1/arch/i386/boot/tools/build.c 2005-03-02 02:38:09.000000000 -0500 +++ linux-2.6.12-rc1-bk1/arch/x86_64/boot/tools/build.c 2005-03-02 02:38:37.000000000 -0500 @@ -150,8 +150,10 @@ sz = sb.st_size; fprintf (stderr, "System is %d kB\n", sz/1024); sys_size = (sz + 15) / 16; - if (!is_big_kernel && sys_size > DEF_SYSSIZE) - die("System is too big. Try using bzImage or modules."); + /* 0x40000*16 = 4.0 MB, reasonable estimate for the current maximum */ + if (sys_size > (is_big_kernel ? 0x40000 : DEF_SYSSIZE)) + die("System is too big. Try using %smodules.", + is_big_kernel ? "" : "bzImage or "); while (sz > 0) { int l, n;