From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S264385AbUEIUzH (ORCPT ); Sun, 9 May 2004 16:55:07 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S264386AbUEIUzH (ORCPT ); Sun, 9 May 2004 16:55:07 -0400 Received: from bay1-f96.bay1.hotmail.com ([65.54.245.96]:46341 "EHLO hotmail.com") by vger.kernel.org with ESMTP id S264385AbUEIUzD (ORCPT ); Sun, 9 May 2004 16:55:03 -0400 X-Originating-IP: [212.179.89.182] X-Originating-Email: [rimaherm@hotmail.com] From: "Rimaher Molanski" To: linux-kernel@vger.kernel.org Subject: releasing page cached pages in 2.4.20 Date: Sun, 09 May 2004 20:55:02 +0000 Mime-Version: 1.0 Content-Type: text/plain; format=flowed Message-ID: X-OriginalArrivalTime: 09 May 2004 20:55:02.0602 (UTC) FILETIME=[E614F2A0:01C43607] Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi, I'm conducting a small analysis of cache algorithms, and have run into problems trying to implement cache eviction in 2.4.20 kernel. I call try_to_release_page (with GFP_NOIO flag) on pages that aren't referenced by anything other than page cache and sometimes also the buffer cache. I sometimes do this call from inside the scheduler tq (a task_schedule function), if this is relevant. Anyway, in certain scenarios I get deadlocked on the text.lock.buffer lock, and I have no idea which locks I must obtain (or any other prerequisites) before I can safely call try_to_release_page. Can anyone help on the requirements? Also, is there a simpler function I could call to get the job done? Thanks, Rimaher. _________________________________________________________________ FREE pop-up blocking with the new MSN Toolbar – get it now! http://toolbar.msn.com/go/onm00200415ave/direct/01/