From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: fsnotify_unmount_inodes Date: Fri, 15 Oct 2010 14:02:23 -0400 Message-ID: <20101015180223.GA6970@infradead.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-fsdevel@vger.kernel.org To: Eric Paris Return-path: Received: from canuck.infradead.org ([134.117.69.58]:35837 "EHLO canuck.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932269Ab0JOSCY (ORCPT ); Fri, 15 Oct 2010 14:02:24 -0400 Content-Disposition: inline Sender: linux-fsdevel-owner@vger.kernel.org List-ID: Hi Eric, I just stumbled over fsnotify_unmount_inodes when auditing some inode list code. For one thing the idea to walk the whole inode list when we already do this again just later seems odd. Did you consider having a per-inode call from invalidate_list instead? That would also help to get rid of the rather messy refcount games. Second it sends a FS_UNMOUNT notification. But only one of the callers actually is for umount. Arguably the other calls are rather obscure (weird ioctl, cluster filesystems, etc), but the name suggests that the users might not be prepared for it. And even the the umount we talk about here is the final reference count drop of the superblock, not the detachment from namespace - I can't really see how an application can use this notification in a meaningful way.