]> git.baikalelectronics.ru Git - kernel.git/commit
ocfs2: ocfs2_inode_lock_tracker does not distinguish lock level
authorLarry Chen <lchen@suse.com>
Fri, 8 Jun 2018 00:04:43 +0000 (17:04 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 8 Jun 2018 00:34:33 +0000 (17:34 -0700)
commit743c4abb243ac2de2936c07b171fa93ec76713e6
tree63a5684bf45ad13498e7122a20f9136843439d46
parent217048b49f3e92df51181cbbe05478124436880b
ocfs2: ocfs2_inode_lock_tracker does not distinguish lock level

ocfs2_inode_lock_tracker as a variant of ocfs2_inode_lock, is used to
prevent deadlock due to recursive lock acquisition.

But this function does not distinguish whether the requested level is EX
or PR.

If a RP lock has been attained, this function will immediately return
success afterwards even an EX lock is requested.

But actually the return value does not mean that the process got a EX
lock, because ocfs2_inode_lock has not been called.

When taking lock levels into account, we face some different situations:

1. no lock is held
   In this case, just lock the inode and return 0

2. We are holding a lock
   For this situation, things diverges into several cases

   wanted     holding      what to do
   ex ex     see 2.1 below
   ex pr     see 2.2 below
   pr ex     see 2.1 below
   pr pr     see 2.1 below

   2.1 lock level that is been held is compatible
   with the wanted level, so no lock action will be tacken.

   2.2 Otherwise, an upgrade is needed, but it is forbidden.

Reason why upgrade within a process is forbidden is that lock upgrade
may cause dead lock.  The following illustrate how it happens.

        process 1                             process 2
ocfs2_inode_lock_tracker(ex=0)
                               <======   ocfs2_inode_lock_tracker(ex=1)

ocfs2_inode_lock_tracker(ex=1)

For the status quo of ocfs2, without this patch, neither a bug nor
end-user impact will be caused because the wrong logic is avoided.

But I'm afraid this generic interface, may be called by other developers
in future and used in this situation.

  a process
ocfs2_inode_lock_tracker(ex=0)
ocfs2_inode_lock_tracker(ex=1)

Link: http://lkml.kernel.org/r/20180510053230.17217-1-lchen@suse.com
Signed-off-by: Larry Chen <lchen@suse.com>
Reviewed-by: Gang He <ghe@suse.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Cc: Changwei Ge <ge.changwei@h3c.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/ocfs2/dlmglue.c
fs/ocfs2/dlmglue.h