<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
According to the call trace, umount hangs because
ocfs2_commit_thread<br>
can't exit. So I think we'd better put the logic into
ocfs2_commit_thread,<br>
which can be align with kthread_should_stop to identify the case.<br>
<br>
Thanks,<br>
Joseph<br>
<br>
<div class="moz-cite-prefix">On 17/1/7 20:01, Gechangwei wrote:<br>
</div>
<blockquote
cite="mid:63ADC13FD55D6546B7DECE290D39E37342E739A1@H3CMLB12-EX.srv.huawei-3com.com"
type="cite">
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<p>Hi,</p>
<p>When journal flushing in ocfs2_commit_cache() fails, following
umount procedure at stage of shutting journal may be blocked due
to non-zero transactions number.</p>
<p>Once jbd2_journal_flush() fails, the journal will be marked as
ABORT state inner JBD2. There is no way to come back afterwards.</p>
<p>So there is no chance to set transactions number to zero, thus,
shutting journal may be blocked .<br>
</p>
<p><br>
</p>
<p>ocfs2_commit_thread()<br>
</p>
<p> ocfs2_commit_cache()</p>
<p> jbd2_journal_flush() -> failure takes journal into
ABORT state, thus, transaction number will never set to zero<br>
</p>
<p><br>
</p>
<p>The back trace is cited blow:</p>
<p>[<ffffffff8109de0c>] kthread_stop+0x4c/0x150<br>
[<ffffffffc056a887>] ocfs2_journal_shutdown+0xa7/0x400
[ocfs2]<br>
[<ffffffffc059fbde>] ocfs2_dismount_volume+0xbe/0x4a0
[ocfs2]<br>
[<ffffffffc059fff7>] ocfs2_put_super+0x37/0xb0 [ocfs2]<br>
[<ffffffff8120534e>] generic_shutdown_super+0x7e/0x110<br>
[<ffffffff81205410>] kill_block_super+0x30/0x80<br>
[<ffffffff81205669>] deactivate_locked_super+0x59/0x90<br>
[<ffffffff812062ee>] deactivate_super+0x4e/0x70<br>
[<ffffffff81222423>] cleanup_mnt+0x43/0x90<br>
[<ffffffff812224c2>] __cleanup_mnt+0x12/0x20<br>
[<ffffffff8109c247>] task_work_run+0xb7/0xf0<br>
[<ffffffff81016f7c>] do_notify_resume+0x8c/0xa0<br>
[<ffffffff817f8384>] int_signal+0x12/0x17<br>
[<ffffffffffffffff>] 0xffffffffffffffff</p>
<p>Through crash debug tool, It can seen that j_num_trans field is
set to 2.<br>
</p>
<p>struct ocfs2_journal {<br>
j_state = OCFS2_JOURNAL_IN_SHUTDOWN, <br>
j_journal = 0xffff8800b4926000, <br>
j_inode = 0xffff880122aba298, <br>
j_osb = 0xffff880093caa000, <br>
j_bh = 0xffff8800a09df888, <br>
j_num_trans = {<br>
counter = 0x2<br>
}, <br>
j_lock = {<br>
{<br>
rlock = {<br>
raw_lock = {<br>
{</p>
<p><br>
</p>
<p>To solve this issue, I propose a patch. Any comments will be
welcomed.</p>
<p>Since journal has been marked as ABORT and flushing journal
failure will free all corresponding buffer heads, it will be
safe to directly set transactions number to zero.<br>
</p>
<p><br>
</p>
<p>From 98f42f5f52851ed84eb372a3e09a413a30ea2664 Mon Sep 17
00:00:00 2001<br>
From: gechangwei <a moz-do-not-send="true"
class="moz-txt-link-rfc2396E"
href="mailto:ge.changwei@h3c.com">
<ge.changwei@h3c.com></a><br>
Date: Sat, 7 Jan 2017 19:48:13 +0800<br>
Subject: [PATCH] fix umount hang after flushing journal failure<br>
<br>
Signed-off-by: gechangwei <a moz-do-not-send="true"
class="moz-txt-link-rfc2396E"
href="mailto:ge.changwei@h3c.com">
<ge.changwei@h3c.com></a><br>
---<br>
fs/ocfs2/journal.c | 1 +<br>
1 file changed, 1 insertion(+)<br>
<br>
diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c<br>
index a244f14..dab2094 100644<br>
--- a/fs/ocfs2/journal.c<br>
+++ b/fs/ocfs2/journal.c<br>
@@ -326,6 +326,7 @@ static int ocfs2_commit_cache(struct
ocfs2_super *osb)<br>
status = jbd2_journal_flush(journal->j_journal);<br>
jbd2_journal_unlock_updates(journal->j_journal);<br>
if (status < 0) {<br>
+ atomic_set(&journal->j_num_trans, 0);<br>
up_write(&journal->j_trans_barrier);<br>
mlog_errno(status);<br>
goto finally;<br>
-- <br>
2.5.1.windows.1<br>
<br>
</p>
<p><br>
</p>
<span style="font-size:7.5pt;font-family:华文细黑;color:gray"><span
lang="EN-US">-------------------------------------------------------------------------------------------------------------------------------------<br>
</span>本邮件及其附件含有杭州华三通信技术有限公司的保密信息，仅限于发送给上面地址中列出<span
lang="EN-US"><br>
</span>的个人或群组。禁止任何其他人以任何形式使用（包括但不限于全部或部分地泄露、复制、<span
lang="EN-US"><br>
</span>或散发）本邮件中的信息。如果您错收了本邮件，请您立即电话或邮件通知发件人并删除本<span
lang="EN-US"><br>
</span>邮件！<span lang="EN-US"><br>
</span></span><span
style="font-size:7.5pt;font-family:"Arial","sans-serif";color:gray"
lang="EN-US">This e-mail and its attachments contain
confidential information from H3C, which is
<br>
intended only for the person or entity whose address is listed
above. Any use of the
<br>
information contained herein in any way (including, but not
limited to, total or partial
<br>
disclosure, reproduction, or dissemination) by persons other
than the intended <br>
recipient(s) is prohibited. If you receive this e-mail in error,
please notify the sender
<br>
by phone or email immediately and delete it!</span>
</blockquote>
<br>
</body>
</html>