From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753808Ab0IWItd (ORCPT ); Thu, 23 Sep 2010 04:49:33 -0400 Received: from gir.skynet.ie ([193.1.99.77]:53300 "EHLO gir.skynet.ie" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752124Ab0IWItc (ORCPT ); Thu, 23 Sep 2010 04:49:32 -0400 Date: Thu, 23 Sep 2010 09:49:19 +0100 From: Mel Gorman To: Christoph Lameter Cc: Naoya Horiguchi , Andi Kleen , Andrew Morton , Wu Fengguang , "Jun'ichi Nomura" , linux-mm , LKML , tony.luck@intel.com Subject: Re: [PATCH 02/10] hugetlb: add allocate function for hugepage migration Message-ID: <20100923084919.GA5185@csn.ul.ie> References: <1283908781-13810-1-git-send-email-n-horiguchi@ah.jp.nec.com> <1283908781-13810-3-git-send-email-n-horiguchi@ah.jp.nec.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-15 Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.17+20080114 (2008-01-14) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Sep 22, 2010 at 04:05:47PM -0500, Christoph Lameter wrote: > On Wed, 8 Sep 2010, Naoya Horiguchi wrote: > > > We can't use existing hugepage allocation functions to allocate hugepage > > for page migration, because page migration can happen asynchronously with > > the running processes and page migration users should call the allocation > > function with physical addresses (not virtual addresses) as arguments. > > Ummm... Some arches like IA64 need huge pages fixed at certain virtual > addresses in which only huge pages exist. A vma is needed in order to be > able to assign proper virtual address to the page. > Are you sure about this case? The virtual address of the page being migrated should not changed, only the physical address. > How does that work with transparent huge pages anyways? > IA-64 doesn't support transparent huge pages. Even if it did, this change is about hugetlbfs, not transparent huge page support. > This looks like its going to break IA64 hugepage support for good. How? > Maybe > thats okay given the reduced significance of IA64? Certainly would > simplify the code. > Currently I'm not seeing how IA-64 gets broken. -- Mel Gorman Part-time Phd Student Linux Technology Center University of Limerick IBM Dublin Software Lab