From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261168AbULWMjn (ORCPT ); Thu, 23 Dec 2004 07:39:43 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261184AbULWMjn (ORCPT ); Thu, 23 Dec 2004 07:39:43 -0500 Received: from ebiederm.dsl.xmission.com ([166.70.28.69]:12930 "EHLO ebiederm.dsl.xmission.com") by vger.kernel.org with ESMTP id S261168AbULWMjj (ORCPT ); Thu, 23 Dec 2004 07:39:39 -0500 To: Vivek Goyal Cc: lkml , fastboot , Linus Torvalds , Andrew Morton , dipankar sarma Subject: Re: [PATCH] Secondary cpus boot-up for non defalut location built kernels References: <1103802944.8123.114.camel@2fwv946.in.ibm.com> From: ebiederm@xmission.com (Eric W. Biederman) Date: 23 Dec 2004 05:38:42 -0700 In-Reply-To: <1103802944.8123.114.camel@2fwv946.in.ibm.com> Message-ID: User-Agent: Gnus/5.0808 (Gnus v5.8.8) Emacs/21.2 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Vivek Goyal writes: > Hi, > > This patch fixes the problem of secondary cpus boot up. This situation > is faced when kernel is built for non default locations like 16MB and > onwards. In this configuration, only primary cpu (BP) comes and > secondary cpus don't boot. > > Problem occurs because in trampoline code, lgdt is not able to load the > GDT as it happens to be situated beyond 16MB. This is due to the fact > that cpu is still in real mode and default operand size is 16bit. Which means that the cpu will load a 24bit linear address (3 bytes) because it is acting like a 286. > This patch uses lgdtl instead of lgdt to force operand size to 32 > instead of 16. Sounds sane looking at the trampoline I suspect that people thought it was using a 32bit address all along. And the code only worked because the data was stored little endian. We probably want to apply the same fix to the ldt instruction just about it. Just in case we ever decide to populate an idt at that point. But the fix certainly looks good from here. Eric > Signed-off-by: Vivek Goyal > --- > > linux-2.6.10-rc3-mm1-changes-root/arch/i386/kernel/trampoline.S | 8 +++++++- > > 1 files changed, 7 insertions(+), 1 deletion(-) > > diff -puN arch/i386/kernel/trampoline.S~boot_ap_for_nondefault_kernel > arch/i386/kernel/trampoline.S > > --- > linux-2.6.10-rc3-mm1-changes/arch/i386/kernel/trampoline.S~boot_ap_for_nondefault_kernel > 2004-12-22 16:36:50.000000000 +0530 > > +++ linux-2.6.10-rc3-mm1-changes-root/arch/i386/kernel/trampoline.S 2004-12-22 > 16:47:54.000000000 +0530 > > @@ -51,8 +51,14 @@ r_base = . > movl $0xA5A5A5A5, trampoline_data - r_base > # write marker for master knows we're running > > + /* GDT tables in non default location kernel can be beyond 16MB and > + * lgdt will not be able to load the address as in real mode default > + * operand size is 16bit. Use lgdtl instead to force operand size > + * to 32 bit. > + */ > + > lidt boot_idt - r_base # load idt with 0, 0 > - lgdt boot_gdt - r_base # load gdt with whatever is appropriate > + lgdtl boot_gdt - r_base # load gdt with whatever is appropriate > > xor %ax, %ax > inc %ax # protected mode (PE) bit > _