From: Yang Bai Subject: Question about doing umount when disk fails Date: Thu, 11 Oct 2012 15:13:26 +0800 Message-ID: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Cc: linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org, linux-ext4@vger.kernel.org To: fengguang.wu@intel.com, viro@zeniv.linux.org.uk, tytso@mit.edu, adilger@sun.com Return-path: Received: from mail-bk0-f46.google.com ([209.85.214.46]:34299 "EHLO mail-bk0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755344Ab2JKHNs (ORCPT ); Thu, 11 Oct 2012 03:13:48 -0400 Sender: linux-ext4-owner@vger.kernel.org List-ID: HI all, We found a problem when umount an ext4 partition, if at the same time, disk fails, the umount process will stay at uninterrupt sleep status and flush thread will be blocked. As I know, when doing umount, the filesystem will sync all changes in memory back to disk through SYNC IO, if the disk fails, the io request can not finish. My question is: Could the sync operation notice the IO errors on disk and cancel itself. So the flush thread will not be blocked and umount process will exit with an error code. This may also affect the process of write back, so I add fengguang and viro. Thanks very much, Yang