From: Jan Kara Subject: [PATCH 0/3] fsnotify cleanups Date: Mon, 27 Jul 2015 20:45:24 +0200 Message-ID: <1438022727-10348-1-git-send-email-jack@suse.com> Cc: Andrew Morton , Lino Sanfilippo , Eric Paris , Jan Kara To: linux-fsdevel@vger.kernel.org Return-path: Received: from mx2.suse.de ([195.135.220.15]:46552 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753886AbbG0Spg (ORCPT ); Mon, 27 Jul 2015 14:45:36 -0400 Sender: linux-fsdevel-owner@vger.kernel.org List-ID: Hello, after understanding some of the fsnotify mark handling code when fixing the last race I've decided to use that knowledge to improve the documentation and somewhat cleanup the code. The first patch in the series documents locking rules for fsnotify marks. The second patch gets rid of special purpose free_list list head in the fsnotify mark structure and the third patch splits the fsnotify_destroy_mark_locked() function so that it doesn't have the awkward property of temporarily dropping mark_mutex which can cause unexpected issues in the caller. All the patches have survived beating with LTP inotify and fanotify tests (including the test excercising inode & group destruction races). Honza