Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752795Ab2FFM7H (ORCPT ); Wed, 6 Jun 2012 08:59:07 -0400 Received: from s15943758.onlinehome-server.info ([217.160.130.188]:42352 "EHLO mail.x86-64.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752304Ab2FFM7F (ORCPT ); Wed, 6 Jun 2012 08:59:05 -0400 Date: Wed, 6 Jun 2012 14:59:29 +0200 From: Borislav Petkov To: Christopher Yeoh Cc: Andrew Morton , Linus Torvalds , LKML , Randy Dunlap Subject: Re: [PATCH] CMA: Do no enable it by default Message-ID: <20120606125929.GD1644@aftab.osrc.amd.com> References: <20120531153512.GK14515@aftab.osrc.amd.com> <20120531154224.GL14515@aftab.osrc.amd.com> <20120601110647.2c446aa7@rockpopper> <20120605110541.GC13495@aftab.osrc.amd.com> <20120606105459.49430920@Gantu> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20120606105459.49430920@Gantu> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1661 Lines: 47 On Wed, Jun 06, 2012 at 10:54:59AM +0930, Christopher Yeoh wrote: > > CROSS_MEMORY_ATTACH is a MPI feature which shouldn't be enabled by > > default on every linux system simply because the majority of users do > > not need it. > > btw although CMA was primarily written for MPI it is used by more than > just MPI implementations. For example, recent versions of strace now > use it instead of PTRACE_PEEKDATA if its available: > > http://article.gmane.org/gmane.comp.sysutils.strace.devel/2467/match=process_vm_readv I see. Looks like process_vm_readv() is faster than PTRACE_PEEKDATA. You could add this to the config option text so people can know why they could/should enable CMA. Oh, and also the note about ptrace privileges below :-). > > Besides, in the config option it says "... which allow a process with > > the correct privileges to directly read from or write to to another > > process's address space.", which, if the reading process has somehow > > gained privileges (as that never happens) is your security issue right > > there. > > The privileges required are exactly the same as required to ptrace the > target. You're rather stuffed anyway if you have a hostile process with > those privileges. Ok. Thanks. -- Regards/Gruss, Boris. Advanced Micro Devices GmbH Einsteinring 24, 85609 Dornach GM: Alberto Bozzo Reg: Dornach, Landkreis Muenchen HRB Nr. 43632 WEEE Registernr: 129 19551 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/