[BUG] Lockdep warning with XFS on 2.6.22-rc6

[Johannes Weiner <hannes-kernel@saeurebad.de>]

Hi,

this is what just hit the ring buffer when I was surfing with elinks on a
brand-new -rc6.

	Hannes

=======================================================
[ INFO: possible circular locking dependency detected ]
2.6.22-rc6 #14
-------------------------------------------------------
elinks/4461 is trying to acquire lock:
 (&(&ip->i_lock)->mr_lock/1){----}, at: [<c01ef3d0>] xfs_ilock+0x50/0xd0

but task is already holding lock:
 (&(&ip->i_lock)->mr_lock){----}, at: [<c01ef3d0>] xfs_ilock+0x50/0xd0

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #1 (&(&ip->i_lock)->mr_lock){----}:
       [<c0137e0d>] __lock_acquire+0xe0d/0xfc0
       [<c013802d>] lock_acquire+0x6d/0x90
       [<c012e4f1>] down_write_nested+0x41/0x60
       [<c01ef412>] xfs_ilock+0x92/0xd0
       [<c01eff98>] xfs_iget_core+0x398/0x5c0
       [<c01f027e>] xfs_iget+0xbe/0x120
       [<c02098bd>] xfs_trans_iget+0xfd/0x160
       [<c01f379c>] xfs_ialloc+0xac/0x500
       [<c020a46c>] xfs_dir_ialloc+0x6c/0x2a0
       [<c02109ae>] xfs_create+0x33e/0x630
       [<c021bf7f>] xfs_vn_mknod+0x19f/0x210
       [<c016fd12>] vfs_mknod+0xa2/0xf0
       [<c03c3dbf>] unix_bind+0x1bf/0x2e0
       [<c0373ba4>] sys_bind+0x54/0x80
       [<c037518d>] sys_socketcall+0x7d/0x260
       [<c01029ce>] sysenter_past_esp+0x5f/0x99
       [<ffffffff>] 0xffffffff

-> #0 (&(&ip->i_lock)->mr_lock/1){----}:
       [<c0137c85>] __lock_acquire+0xc85/0xfc0
       [<c013802d>] lock_acquire+0x6d/0x90
       [<c012e6e1>] down_read_nested+0x41/0x60
       [<c01ef3d0>] xfs_ilock+0x50/0xd0
       [<c020dd45>] xfs_lock_inodes+0x175/0x190
       [<c02054af>] xfs_lock_for_rename+0x22f/0x280
       [<c02055af>] xfs_rename+0xaf/0x8c0
       [<c021bb64>] xfs_vn_rename+0x34/0x80
       [<c0170237>] vfs_rename+0x307/0x370
       [<c0171e3b>] sys_renameat+0x1bb/0x1f0
       [<c0171e98>] sys_rename+0x28/0x30
       [<c01029ce>] sysenter_past_esp+0x5f/0x99
       [<ffffffff>] 0xffffffff

other info that might help us debug this:

3 locks held by elinks/4461:
 #0:  (&inode->i_mutex/1){--..}, at: [<c016f459>] lock_rename+0xe9/0xf0
 #1:  (&inode->i_mutex){--..}, at: [<c03c8a8c>] mutex_lock+0x1c/0x20
 #2:  (&(&ip->i_lock)->mr_lock){----}, at: [<c01ef3d0>] xfs_ilock+0x50/0xd0

stack backtrace:
 [<c010398a>] show_trace_log_lvl+0x1a/0x30
 [<c01046a2>] show_trace+0x12/0x20
 [<c0104765>] dump_stack+0x15/0x20
 [<c0135d4c>] print_circular_bug_tail+0x6c/0x80
 [<c0137c85>] __lock_acquire+0xc85/0xfc0
 [<c013802d>] lock_acquire+0x6d/0x90
 [<c012e6e1>] down_read_nested+0x41/0x60
 [<c01ef3d0>] xfs_ilock+0x50/0xd0
 [<c020dd45>] xfs_lock_inodes+0x175/0x190
 [<c02054af>] xfs_lock_for_rename+0x22f/0x280
 [<c02055af>] xfs_rename+0xaf/0x8c0
 [<c021bb64>] xfs_vn_rename+0x34/0x80
 [<c0170237>] vfs_rename+0x307/0x370
 [<c0171e3b>] sys_renameat+0x1bb/0x1f0
 [<c0171e98>] sys_rename+0x28/0x30
 [<c01029ce>] sysenter_past_esp+0x5f/0x99
 =======================