From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Tue, 1 May 2001 19:44:02 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Tue, 1 May 2001 19:43:52 -0400 Received: from pa147.bialystok.sdi.tpnet.pl ([213.25.59.147]:1284 "EHLO localhost.localdomain") by vger.kernel.org with ESMTP id ; Tue, 1 May 2001 19:43:42 -0400 Date: Wed, 2 May 2001 01:42:16 +0200 From: Jacek =?iso-8859-2?Q?Pop=B3awski?= To: linux-kernel@vger.kernel.org Subject: reason for VIA performance drop since 2.4.2-ac21 Message-ID: <20010502014216.A604@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-2 Content-Disposition: inline User-Agent: Mutt/1.3.17i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org I have motherboard with VIA MVP3 chipset. I noticed big video slowdown since 2.4.2-ac21. Watching "divx" by avifile with new kernels is impossible, becouse very bad performance. Now, after few hours - I found the reason, and I don't understand it. It has nothing to do with mtrr or K6. In file arch/i386/kernel/pci-pc.c there is a pci_fixup_via691_2 function. It appeared in 2.4.2-ac21. And it works for my chipset - VIA_82C598. When I put "return" in body of this function, recompile and start kernel 2.4.4 - "x11perf -putimage100" shows that video works fast again. I see two possibilities: 1) this is just a debug code, and kernels >2.4.2-ac20 shouldn't be used by VIA MVP3 owners 2) this code fix crash possibility, and all kernels without it (including 2.2.19) are buggy with VIA MVP3 What is true, and is it safe to skip that function if I need 2.4 and fast video?