From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755452Ab3KVCNq (ORCPT ); Thu, 21 Nov 2013 21:13:46 -0500 Received: from mx1.redhat.com ([209.132.183.28]:32256 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753754Ab3KVCNo (ORCPT ); Thu, 21 Nov 2013 21:13:44 -0500 Date: Thu, 21 Nov 2013 21:13:12 -0500 From: Vivek Goyal To: Kees Cook Cc: Greg KH , linux-kernel@vger.kernel.org, kexec@lists.infradead.org, ebiederm@xmission.com, hpa@zytor.com, mjg59@srcf.ucam.org Subject: Re: [PATCH 4/6] kexec: A new system call, kexec_file_load, for in kernel kexec Message-ID: <20131122021312.GD31921@redhat.com> References: <1384969851-7251-1-git-send-email-vgoyal@redhat.com> <1384969851-7251-5-git-send-email-vgoyal@redhat.com> <20131121190350.GC17070@kroah.com> <20131122010311.GZ5274@outflux.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20131122010311.GZ5274@outflux.net> 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 On Thu, Nov 21, 2013 at 05:03:11PM -0800, Kees Cook wrote: > Hi, > > On Thu, Nov 21, 2013 at 11:03:50AM -0800, Greg KH wrote: > > On Wed, Nov 20, 2013 at 12:50:49PM -0500, Vivek Goyal wrote: > > > This patch implements the in kernel kexec functionality. It implements a > > > new system call kexec_file_load. I think parameter list of this system > > > call will change as I have not done the kernel image signature handling > > > yet. I have been told that I might have to pass the detached signature > > > and size as part of system call. > > > > This could be done as we do with modules, and just tack the signature > > onto the end of the 'blob' of the image. That way we could use the same > > tool to sign the binary as we do for modules, and save the need for > > extra parameters in the syscall. > > As long as the system call passing in an fd, I'm all good. For those > of us that run from verified filesystems, we don't need the additional > signing overhead, but we do need the file descriptor to validate the > origin of the kernel. Yep, Greg had mentioned that keep the interface file descriptor based so that it can work well with LSM hooks and that's why I went with it. Thanks Vivek