Bug 16034

Summary: 2.6.34: dlm: possible circular locking dependency detected
Product: File System Reporter: Maciej Rutecki (maciej.rutecki)
Component: ocfs2Assignee: fs_ocfs2
Status: CLOSED CODE_FIX    
Severity: normal CC: cat, error27, maciej.rutecki, rjw
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 2.6.34 Tree: Mainline
Regression: Yes
Bug Depends on:    
Bug Blocks: 15310    

Description Maciej Rutecki 2010-05-23 06:33:26 UTC
Subject    : 2.6.34: dlm: possible circular locking dependency detected
Submitter  : CaT <cat@zip.com.au>
Date       : 2010-05-21 6:59
Message-ID : 20100521065933.GH2657@zip.com.au
References : http://marc.info/?l=linux-kernel&m=127442519608471&w=2

This entry is being used for tracking a regression from 2.6.33.  Please don't
close it until the problem is fixed in the mainline.
Comment 1 Cat 2010-05-23 23:53:13 UTC
This also happens with 2.6.33.4.
Comment 2 Dan Carpenter 2010-06-19 12:00:00 UTC
I don't know the policy in for the stable kernel or for bugzilla, but this was fixed in 2.6.35.  The commit is:

commit d9ef75221a6247b758e1d7e18edb661996e4b7cf
Author: Wengang Wang <wen.gang.wang@oracle.com>
Date:   Mon May 17 20:20:44 2010 +0800

    ocfs2:dlm: avoid dlm->ast_lock lockres->spinlock dependency break
    
    Currently we process a dirty lockres with the lockres->spinlock taken. While
    during the process, we may need to lock on dlm->ast_lock. This breaks the
    dependency of dlm->ast_lock(lock first) and lockres->spinlock(lock second).
    
    This patch fixes the problem.
    Since we can't release lockres->spinlock, we have to take dlm->ast_lock
    just before taking the lockres->spinlock and release it after lockres->spinlock
    is released. And use __dlm_queue_bast()/__dlm_queue_ast(), the nolock version,
    in dlm_shuffle_lists(). There are no too many locks on a lockres, so there is no
    performance harm.
    
    Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com>
    Signed-off-by: Joel Becker <joel.becker@oracle.com>
Comment 3 Rafael J. Wysocki 2010-07-09 23:26:15 UTC
The problem has been fixed in mainline, so closing.

Please send an inclusion request to stable@kernel.org so that it goes to -stable.