From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759417AbXGRVFG (ORCPT ); Wed, 18 Jul 2007 17:05:06 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752199AbXGRVEy (ORCPT ); Wed, 18 Jul 2007 17:04:54 -0400 Received: from one.firstfloor.org ([213.235.205.2]:44459 "EHLO one.firstfloor.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751911AbXGRVEx (ORCPT ); Wed, 18 Jul 2007 17:04:53 -0400 Date: Wed, 18 Jul 2007 23:04:52 +0200 From: Andi Kleen To: "H. Peter Anvin" Cc: Andi Kleen , Matt Mackall , Jonathan Campbell , linux-kernel@vger.kernel.org Subject: Re: Patches for REALLY TINY 386 kernels Message-ID: <20070718210452.GG3898@one.firstfloor.org> References: <469A8AED.7070207@nerdgrounds.com> <469E3806.4030804@zytor.com> <20070718194137.GG11166@waste.org> <20070718201043.GF3898@one.firstfloor.org> <469E7709.9070801@zytor.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <469E7709.9070801@zytor.com> User-Agent: Mutt/1.4.2.1i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Jul 18, 2007 at 01:24:41PM -0700, H. Peter Anvin wrote: > Andi Kleen wrote: > > I was waiting for someone to make that "point" ... > > > >> Every byte you can shave off the compressed kernel image is another > >> byte you can use for userspace on your FLASH. > > > > Now let's see if that 1MB 386 contains any flash at all. Guesses? > > > > CPUID is hardly something you want to give up for modern embedded stuff. > I haven't seen any new embedded stuff using anything less than a 586, > and even those are disappearing quickly in favor of Geode and VIA. > > However, compressed size reductions as an abstract thing is useful for > this market. Just not these particular ones. The first thing to get > there is probably an LZMA-based compressor instead of gzip. That would need more memory again. Better just write less bloated code. Perhaps mandatory bloatometer runs during -rc*s for kernels with minimal config with public code pig shame lists similar to the regression lists are useful. Anyone volunteering? I suspect there is also much more low hanging fruit of this around. I don't think eliminating cpuid is a step forward though; that's just madness. I killed about 20k in a i386 defconfig kernel in my upcomming patchkit (and add some other code instead, but not much) so I'm coming out well currently. -Andi