From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262514AbTHaJV7 (ORCPT ); Sun, 31 Aug 2003 05:21:59 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262524AbTHaJV7 (ORCPT ); Sun, 31 Aug 2003 05:21:59 -0400 Received: from aneto.able.es ([212.97.163.22]:60550 "EHLO aneto.able.es") by vger.kernel.org with ESMTP id S262514AbTHaJVe (ORCPT ); Sun, 31 Aug 2003 05:21:34 -0400 Date: Sun, 31 Aug 2003 11:21:27 +0200 From: "J.A. Magallon" To: Marcelo Tosatti Cc: lkml Subject: [PATCH] correct AT_PLATFORM for HT cpus Message-ID: <20030831092127.GA21970@werewolf.able.es> References: Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Disposition: inline Content-Transfer-Encoding: 7BIT In-Reply-To: ; from marcelo@conectiva.com.br on Sat, Aug 30, 2003 at 17:48:22 +0200 X-Mailer: Balsa 2.0.14 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On 08.30, Marcelo Tosatti wrote: > > Hello, > > Here goes -pre2. It contains an USB update, PPC merge, m68k merge, IDE > changes from Alan, network drivers update from Jeff, amongst other fixes > and updates. > In shortest: Pentium4 with HT do not realize they are themselves an i686. Even, they do not know what they are. This makes P4HT select i386 libraries at runtime, instead of i686 ones. In short: binfmt_elf::create_elf_tables() copies the AT_PLATFORM in the place pointed by u_platform, and then copies the pointer to elf tables. But, in the HT case, u_platform gets changed in the middle, beacuse it is also used as the beginning of the stack, that is shifted for HT processors to avoid cache collisions. So then it puts a pointer to garbage in the elf tables: if (k_platform) { platform_len = strlen(k_platform) + 1; u_platform = p - platform_len; __copy_to_user(u_platform, k_platform, platform_len); } else u_platform = p; ... if(smp_num_siblings > 1) u_platform = u_platform - ((current->pid % 64) << 7); ... if (k_platform) { sp -= 2; NEW_AUX_ENT(0, AT_PLATFORM, (elf_addr_t)(unsigned long) u_platform); } Solution is to use separate variables to store u_platform and the stack_top: --- linux/fs/binfmt_elf.c.orig 2002-12-28 00:12:32.000000000 +0100 +++ linux/fs/binfmt_elf.c 2002-12-28 00:32:37.000000000 +0100 @@ -116,11 +116,14 @@ elf_caddr_t *argv; elf_caddr_t *envp; elf_addr_t *sp, *csp; + char *stack_top; char *k_platform, *u_platform; long hwcap; size_t platform_len = 0; size_t len; + stack_top = p; + /* * Get hold of platform and hardware capabilities masks for * the machine we are running on. In some cases (Sparc), @@ -135,8 +138,8 @@ platform_len = strlen(k_platform) + 1; u_platform = p - platform_len; __copy_to_user(u_platform, k_platform, platform_len); - } else - u_platform = p; + stack_top = u_platform; + } #if defined(__i386__) && defined(CONFIG_SMP) /* @@ -149,15 +152,14 @@ * processors. This keeps Mr Marcelo Person happier but should be * removed for 2.5 */ - if(smp_num_siblings > 1) - u_platform = u_platform - ((current->pid % 64) << 7); + stack_top -= ((current->pid % 64) << 7); #endif /* * Force 16 byte _final_ alignment here for generality. */ - sp = (elf_addr_t *)(~15UL & (unsigned long)(u_platform)); + sp = (elf_addr_t *)(~15UL & (unsigned long)(stack_top)); csp = sp; csp -= (1+DLINFO_ITEMS)*2 + (k_platform ? 2 : 0); #ifdef DLINFO_ARCH_ITEMS As other patches, tested by myself. And Mandrake kernel includes it. Comments ? -- J.A. Magallon \ Software is like sex: werewolf.able.es \ It's better when it's free Mandrake Linux release 9.2 (Cooker) for i586 Linux 2.4.22-jam1m (gcc 3.3.1 (Mandrake Linux 9.2 3.3.1-1mdk))