Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S271364AbUJVPfX (ORCPT ); Fri, 22 Oct 2004 11:35:23 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S271385AbUJVPek (ORCPT ); Fri, 22 Oct 2004 11:34:40 -0400 Received: from fmr04.intel.com ([143.183.121.6]:39335 "EHLO caduceus.sc.intel.com") by vger.kernel.org with ESMTP id S271374AbUJVPdY (ORCPT ); Fri, 22 Oct 2004 11:33:24 -0400 Message-Id: <200410221530.i9MFUWq30084@unix-os.sc.intel.com> From: "Chen, Kenneth W" To: "'Andrew Morton'" , "Christoph Lameter" Cc: , , Subject: RE: Hugepages demand paging V1 [3/4]: Overcommit handling Date: Fri, 22 Oct 2004 08:32:41 -0700 X-Mailer: Microsoft Office Outlook, Build 11.0.5510 Thread-Index: AcS4IiG30zNvKFEoQhu+wDZnTedqIAAKOB4Q In-Reply-To: <20041022032823.215bd95f.akpm@osdl.org> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1409 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 923 Lines: 25 Andrew Morton wrote on Friday, October 22, 2004 3:28 AM > Christoph Lameter wrote: > > > > * overcommit handling > > What does this do, and why do we want it? The name of "overcommit" is definitely misleading. It means the opposite. Since physical hugetlb backing page pool is fixed and controlled by sys admin, the fault handler can not allocate free pages when hugetlb page pool exhaust because of user over commits hugetlb page. Thus we enforce strict accounting upfront to guarantee that there will be hugetlb page available at fault time. The down side of not having it is we SIGBUS if the kernel doesn't have any free hugetlb pages left. - Ken - 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/