From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1761049AbXF0M6H (ORCPT ); Wed, 27 Jun 2007 08:58:07 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1758699AbXF0M54 (ORCPT ); Wed, 27 Jun 2007 08:57:56 -0400 Received: from static-ip-62-75-166-246.inaddr.intergenia.de ([62.75.166.246]:43319 "EHLO vs166246.vserver.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758548AbXF0M5z (ORCPT ); Wed, 27 Jun 2007 08:57:55 -0400 From: Michael Buesch To: Matt Mackall Subject: Re: [PATCH RFC #2] hwrng: Add type categories Date: Wed, 27 Jun 2007 14:57:17 +0200 User-Agent: KMail/1.9.6 Cc: Andrew Morton , Henrique de Moraes Holschuh , linux-kernel References: <200706262021.51410.mb@bu3sch.de> <20070626224517.GG11115@waste.org> In-Reply-To: <20070626224517.GG11115@waste.org> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200706271457.17709.mb@bu3sch.de> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wednesday 27 June 2007 00:45:18 Matt Mackall wrote: > On Tue, Jun 26, 2007 at 08:21:51PM +0200, Michael Buesch wrote: > > Don't use the word "quality", as people seem to think of > > the entropy quality when hearing that word. > > Why do I so often feel compelled to respond with "did you read what I > wrote?" on this list? Because you ignored my explanations. > I object to your MEANINGLESS CATEGORIES. > > > This uses the word "type", which is probably better for > > understanding what the value really means. > > Please explain: > > a) how is bad different from pseudo? > b) how is onboard different than dedicated? Read the Kdoc help text in the patch. It explains it. > I maintain that there are exactly two categories on this axis: real > and fake. And I'd rather stomp all over this notion of other > categories you've invented now before people actually start trying to > use them. > > There are also two other axes that this approach neglects. Trusted > (another boolean) and bitrate (a scalar). No. Read this: I will explain it once again: It is _just_ to select a default policy of which RNG is initialized by default. The type does say _nothing_ about the entropy quality or something else. It _just_ judges about which RNG is preferred for default initialization. External RNG boards are preferred over internal onboard stuff; onboard is preferred over bad RNGs like bcm43xx; bcm43xx is preferred over devices that are not RNGs by definition but could be used as such (sensors). -- Greetings Michael.