From mboxrd@z Thu Jan 1 00:00:00 1970 From: Theodore Ts'o Subject: Re: Kernel Errors: Wrong Opcodes Date: Mon, 12 Nov 2012 09:54:33 -0500 Message-ID: <20121112145433.GE4895@thunk.org> References: <87zk2mvsag.wl%oehme.markus@gmx.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-ext4@vger.kernel.org To: Markus Oehme Return-path: Received: from li9-11.members.linode.com ([67.18.176.11]:60660 "EHLO imap.thunk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751887Ab2KLOyh (ORCPT ); Mon, 12 Nov 2012 09:54:37 -0500 Content-Disposition: inline In-Reply-To: <87zk2mvsag.wl%oehme.markus@gmx.de> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Mon, Nov 12, 2012 at 02:56:23PM +0100, Markus Oehme wrote: > Hello, > > I see lots of the following errors on my amd64 machine under heavy load > (specifically the gcc test suite). Since the trace mentions ext4 I decided > to send this to the ext4 mailing list. > > The scenario: I'm running a Hardened Gentoo ~amd64 (Intel Core2 Quad) box > with gcc-4.7.2 and Linux 3.6.6 with hardening patches (grsec). During a > build from source of gcc-4.7.2 the error prints to the console, specifically > this seems to occur during the test phase. There are multiple instances of > the error all with the same call trace. The computer stays usable, but the > build directory of the gcc build is somehow damaged. It is afterwards > impossible to delete the build directory (rm simply hangs forever) until I > reboot the machine. After the reboot everything seems to be fine again. Invalid opcode sure sounds like it's a compiler issue. Since you can relatively easily reproduce this, can you try recompiling a kernel w/o hardening, and see if the problem goes away? If it does, you can send a bug report to the GCC folks. Regards, - Ted