From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rusty Russell Subject: Re: Early crash Date: Tue, 8 Feb 2011 13:42:48 +1030 Message-ID: <201102081342.49243.rusty@rustcorp.com.au> References: <20110207081933.GA11855@dtor-ws.eng.vmware.com> <20110207165829.GA13101@dtor-ws.eng.vmware.com> <20110207.112708.193718501.davem@davemloft.net> Mime-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: Received: from ozlabs.org ([203.10.76.45]:47089 "EHLO ozlabs.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752762Ab1BHDMw (ORCPT ); Mon, 7 Feb 2011 22:12:52 -0500 In-Reply-To: <20110207.112708.193718501.davem@davemloft.net> Sender: linux-arch-owner@vger.kernel.org List-ID: To: David Miller Cc: dtor@vmware.com, schwab@linux-m68k.org, geert@linux-m68k.org, linux-kernel@vger.kernel.org, linux-m68k@vger.kernel.org, linux-arch@vger.kernel.org On Tue, 8 Feb 2011 05:57:08 am David Miller wrote: > From: Dmitry Torokhov > Date: Mon, 7 Feb 2011 08:58:29 -0800 > > > But, theoretically speaking, nothing stops GCC to align pointers with > > "gaps" as well? Let's say having everything (or some) aligned on > > quadword boundary even though arch is 32 bit? > > The alignment business only applies to aggregates (ie. structs and > unions). > > This has been confirmed via several weeks of expermentation with > different GCC versions on different platforms as well. But OTOH, this is an old problem which was faced by module params since pre-git. And we use the-align-to-void*-size method there; I vaguely recall inserting it. You've now got me wondering whether these platforms have broken builtin module parameters, but I think it would crash iterating if you had any boot parameters at all if that were the case. So do we fix that now too, or wait for it to break? Rusty.