From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S263645AbUDMR3k (ORCPT ); Tue, 13 Apr 2004 13:29:40 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S263646AbUDMR3k (ORCPT ); Tue, 13 Apr 2004 13:29:40 -0400 Received: from mproxy.gmail.com ([216.239.56.253]:4831 "HELO mproxy.gmail.com") by vger.kernel.org with SMTP id S263645AbUDMR3i (ORCPT ); Tue, 13 Apr 2004 13:29:38 -0400 Message-ID: Date: Tue, 13 Apr 2004 10:29:05 -0700 From: Martin Peck To: linux-kernel@vger.kernel.org Subject: Question about dentry cache in VFS Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org I am working on a userspace filesystem and would like the ability to selectively decache inodes and their children from the VFS. I can do this on a global scale using invalidate_inodes() and shrink_dcache_sb(), however, I would also like to be able to decache for specific inodes/entries. One possibility (and perhaps the 'right' way) to do this which I had considered is using dentry_operations to trap d_revalidate as needed, however, I would like to avoid an invocation here for every dentry if there is another way to selectively decache. Thanks in advance,