From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S271715AbTHDOHF (ORCPT ); Mon, 4 Aug 2003 10:07:05 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S271750AbTHDOHF (ORCPT ); Mon, 4 Aug 2003 10:07:05 -0400 Received: from holomorphy.com ([66.224.33.161]:55019 "EHLO holomorphy") by vger.kernel.org with ESMTP id S271715AbTHDOHD (ORCPT ); Mon, 4 Aug 2003 10:07:03 -0400 Date: Mon, 4 Aug 2003 07:08:14 -0700 From: William Lee Irwin III To: Andrew Morton Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org, mingo@elte.hu Subject: Re: 2.6.0-test2-mm4 Message-ID: <20030804140814.GJ32488@holomorphy.com> Mail-Followup-To: William Lee Irwin III , Andrew Morton , linux-kernel@vger.kernel.org, linux-mm@kvack.org, mingo@elte.hu References: <20030804013036.16d9fa3a.akpm@osdl.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20030804013036.16d9fa3a.akpm@osdl.org> Organization: The Domain of Holomorphy User-Agent: Mutt/1.5.4i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Aug 04, 2003 at 01:30:36AM -0700, Andrew Morton wrote: > +4g4g-pmd-fix.patch If you're going to back out pgd preconstruction, at least back it out all the way so list poison isn't tripped over randomly on PAE. This is actually worse than before, since you're basically doing list_del() on whatever value of page->lru was handed to mm/slab.c from page_alloc.c in pgd_dtor() multiple times per-page and pounding the lock for no reason whatsoever on PAE. It's also degrading performance on non-PAE due to the fact no preconstruction is done there, though harmless due to the fact the only trace of pgd preconstruction left is the AGP fix. Someone please tell me they realize this is a backout because absolutely zero data structure initialization is done in ->ctor() and the entire thing is memcpy()'d and memset()'d over in the front end to the slab. I have no idea what, if anything has been absorbed from my prior posts on this subject. AFAICT I'm getting dead air (or worse) back from everyone else and no one's even bothering to read the code. i.e. either no one understands a word I'm saying or no one's listening. -- wli