From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jan Kara Subject: Re: [BUG][dioread_nolock] blocked for more than 120s when we run xfstests #269 Date: Thu, 7 Mar 2013 16:11:40 +0100 Message-ID: <20130307151140.GF6723@quack.suse.cz> References: <20130307124054.GE2800@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-ext4@vger.kernel.org To: Zheng Liu Return-path: Received: from cantor2.suse.de ([195.135.220.15]:40636 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754942Ab3CGPLm (ORCPT ); Thu, 7 Mar 2013 10:11:42 -0500 Content-Disposition: inline In-Reply-To: <20130307124054.GE2800@gmail.com> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Thu 07-03-13 20:40:54, Zheng Liu wrote: > Hi all, > > This bug will be triggered when dioread_nolock enables. In 3.8 kernel, > the test will be blocked for more than 120s, and we get the following > messages. *But* in dev branch the system will hang silently without any > messages. I need to run test 7 times to hit it. > > wenqing: run xfstest 269 > kernel: EXT4-fs (sda2): mounted filesystem with ordered data mode. Opts: > acl,user_xattr,dioread_nolock > kernel: INFO: task umount:3376 blocked for more than 120 seconds. > kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this > message. > kernel: umount D 0000000000000001 0 3376 3003 0x00000000 > kernel: ffff880117417cd8 0000000000000082 ffff880117417bd8 ffff880117416010 > kernel: ffff880112d12110 0000000000012080 ffff880117417fd8 0000000000004000 > kernel: ffff880117417fd8 0000000000012080 ffff88011331d950 ffff880112d12110 > kernel: Call Trace: > kernel: [] ? pagevec_lookup+0x22/0x2b > kernel: [] ? truncate_inode_pages_range+0x273/0x3a2 > kernel: [] schedule+0x64/0x66 > kernel: [] ext4_ioend_wait+0x87/0x9f [ext4] > kernel: [] ? wake_up_bit+0x2a/0x2a > kernel: [] ext4_evict_inode+0x4a/0x41d [ext4] > kernel: [] evict+0xa2/0x15b > kernel: [] ? __list_del_entry+0x51/0x98 > kernel: [] dispose_list+0x3e/0x50 > kernel: [] evict_inodes+0xdb/0xe7 > kernel: [] generic_shutdown_super+0x4c/0xd3 > kernel: [] kill_block_super+0x27/0x69 > kernel: [] deactivate_locked_super+0x26/0x52 > kernel: [] deactivate_super+0x45/0x4a > kernel: [] mntput_no_expire+0x110/0x118 > kernel: [] sys_umount+0x306/0x331 > kernel: [] ? sigprocmask+0x63/0x67 > kernel: [] system_call_fastpath+0x16/0x1b I think I see the problem Hum, either we missed a wakeup or ioend gets stuck for some reason and is never processed. Can you maybe post sysrq-w output once the kernel gets stuck? Thanks. Honza -- Jan Kara SUSE Labs, CR