From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751916Ab2LTU2V (ORCPT ); Thu, 20 Dec 2012 15:28:21 -0500 Received: from mail-pb0-f49.google.com ([209.85.160.49]:34256 "EHLO mail-pb0-f49.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751496Ab2LTU2Q (ORCPT ); Thu, 20 Dec 2012 15:28:16 -0500 Date: Thu, 20 Dec 2012 12:28:13 -0800 (PST) From: David Rientjes X-X-Sender: rientjes@chino.kir.corp.google.com To: Sasha Levin cc: Andrew Morton , "Kirill A. Shutemov" , Xiao Guangrong , Andrea Arcangeli , linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH 04/15] mm/huge_memory: use new hashtable implementation In-Reply-To: <50D27707.7060309@oracle.com> Message-ID: References: <1355756497-15834-1-git-send-email-sasha.levin@oracle.com> <1355756497-15834-4-git-send-email-sasha.levin@oracle.com> <50D27707.7060309@oracle.com> User-Agent: Alpine 2.00 (DEB 1167 2008-08-23) MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, 19 Dec 2012, Sasha Levin wrote: > In this case, the downside is that you'll waste 8KB if hugepages aren't available, > but the upside is that you'll have one less dereference when accessing the > hashtable. > > If the 8KB saving is preferable here I'll drop the patch and come back when > dynamic hashtable is supported. > If a distro releases with CONFIG_TRANSPARNET_HUGEPAGE=y and a user is running on a processor that does not support pse then this just cost them 8KB for no reason. The overhead by simply enabling CONFIG_TRANSPARENT_HUGEPAGE is worse in this scenario, but this is whole reason for having the dynamic allocation in the original code. If there's a compelling reason for why we want this change, then that fact should at least be documented. Could you propose a v2 that includes fixes for the other problems that were mentioned?