Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964971AbWADAXn (ORCPT ); Tue, 3 Jan 2006 19:23:43 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S964990AbWADAXn (ORCPT ); Tue, 3 Jan 2006 19:23:43 -0500 Received: from mxout.hispeed.ch ([62.2.95.247]:18122 "EHLO smtp.hispeed.ch") by vger.kernel.org with ESMTP id S964971AbWADAXm (ORCPT ); Tue, 3 Jan 2006 19:23:42 -0500 Subject: Re: [2.6 patch] schedule obsolete OSS drivers for removal From: Thomas Sailer To: Jaroslav Kysela Cc: Lee Revell , LKML In-Reply-To: References: <20050726150837.GT3160@stusta.de> <200601031522.06898.s0348365@sms.ed.ac.uk> <20060103160502.GB5262@irc.pl> <200601031629.21765.s0348365@sms.ed.ac.uk> <20060103170316.GA12249@dspnet.fr.eu.org> <1136312901.24703.59.camel@mindpipe> <1136316640.4106.26.camel@unreal> <1136318187.4106.32.camel@unreal> Content-Type: text/plain Date: Wed, 04 Jan 2006 01:23:24 +0100 Message-Id: <1136334204.4106.43.camel@unreal> Mime-Version: 1.0 X-Mailer: Evolution 2.2.3 (2.2.3-2.fc4) Content-Transfer-Encoding: 7bit X-DCC-spamcheck-01.tornado.cablecom.ch-Metrics: smtp-02.tornado.cablecom.ch 32700; Body=3 Fuz1=3 Fuz2=3 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1477 Lines: 35 On Tue, 2006-01-03 at 21:06 +0100, Jaroslav Kysela wrote: > The "plugin" (or rather conversion, routing and resampling) system in the > OSS emulation can be turned off using the proc interface: Hm. IMO by including resampling and format conversion you're trying to "unbreak" broken OSS apps in the kernel. And by having this on by default you're rewarding writers of broken OSS apps while punishing those that write correct apps... But this is a sidetrack. Even though it's not optimal from the CPU usage point of view to have two sampling rate converters in sequence, and apart from the SNR loss by the overly simplistic linear interpolator, soundmodem should still work with ALSA's OSS emulation. But it doesn't. Well, it almost does: only every tenth or so bit is incorrect (which is inacceptable for a modem, though). This leads me to suspect there's something else wrong with the sample rate converter. In sound/core/oss/rate.c, resample_expand, line 111: if (src_frames1-- > 0) { What is this test for? Similar code is also in resample_shrink. Either it's never false, or I know why modem apps don't work with it: it would be "inventing samples out of the blue", thereby adding lots of jitter to the time axis... Tom - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/