Bug 7287

Summary:	INFO: possible recursive locking detected: mount/3232 is trying to acquire lock
Product:	File System	Reporter:	Martin Mokrejs (mmokrejs)
Component:	XFS	Assignee:	Dave Chinner (dgc)
Status:	CLOSED CODE_FIX
Severity:	normal
Priority:	P2
Hardware:	i386
OS:	Linux
Kernel Version:	2.6.18	Subsystem:
Regression:	---	Bisected commit-id:
Attachments:	dmesg dmesg-2.6.19.2

Description Martin Mokrejs 2006-10-08 14:02:38 UTC

Most recent kernel where this bug did not occur: 2.6.17.11
Distribution: Gentoo
Hardware Environment: ASUS L3C/S laptop
Software Environment:
Problem Description:

The kernel has compiled in some bultin tests and so that's how I found the
message in dmesg output

Steps to reproduce:

Comment 1 Martin Mokrejs 2006-10-08 14:03:04 UTC

Created attachment 9186 [details]
dmesg

Comment 2 Martin Mokrejs 2007-01-31 00:47:23 UTC

Created attachment 10234 [details]
dmesg-2.6.19.2

Still present.

Comment 3 Christian Kujau 2007-04-05 06:41:44 UTC

still there with 2.6.21-rc5, full dmesg & details here:
http://nerdbynature.de/bits/2.6.21-rc5/



[93422.248250] 
[93422.248254] =============================================
[93422.248268] [ INFO: possible recursive locking detected ]
[93422.248273] 2.6.21-rc5 #2
[93422.248276] ---------------------------------------------
[93422.248280] rm/32198 is trying to acquire lock:
[93422.248284]  (&(&ip->i_lock)->mr_lock){----}, at: [<c0246a31>]
xfs_ilock+0x71/0xa0
[93422.248299] 
[93422.248300] but task is already holding lock:
[93422.248304]  (&(&ip->i_lock)->mr_lock){----}, at: [<c0246a31>]
xfs_ilock+0x71/0xa0
[93422.248314] 
[93422.248314] other info that might help us debug this:
[93422.248320] 3 locks held by rm/32198:
[93422.248323]  #0:  (&inode->i_mutex/1){--..}, at: [<c0167026>]
do_unlinkat+0x96/0x160
[93422.248334]  #1:  (&inode->i_mutex){--..}, at: [<c0165305>] vfs_unlink+0x75/0xe0
[93422.248346]  #2:  (&(&ip->i_lock)->mr_lock){----}, at: [<c0246a31>]
xfs_ilock+0x71/0xa0
[93422.248356] 
[93422.248357] stack backtrace:
[93422.248362]  [<c0134b19>] __lock_acquire+0xa99/0x1010
[93422.248373]  [<c01350e7>] lock_acquire+0x57/0x70
[93422.248379]  [<c0246a31>] xfs_ilock+0x71/0xa0
[93422.248385]  [<c012e478>] down_write+0x38/0x50
[93422.248393]  [<c0246a31>] xfs_ilock+0x71/0xa0
[93422.248399]  [<c0246a31>] xfs_ilock+0x71/0xa0
[93422.248405]  [<c026bdb6>] xfs_lock_dir_and_entry+0xf6/0x100
[93422.248413]  [<c026c437>] xfs_remove+0x197/0x4e0
[93422.248419]  [<c016db79>] d_instantiate+0x19/0x40
[93422.248426]  [<c016d9e0>] d_rehash+0x20/0x50
[93422.248434]  [<c0165305>] vfs_unlink+0x75/0xe0
[93422.248442]  [<c0273e93>] xfs_vn_unlink+0x23/0x60
[93422.248449]  [<c03ccbaf>] __mutex_lock_slowpath+0x13f/0x280
[93422.248459]  [<c01339bb>] mark_held_locks+0x6b/0x90
[93422.248464]  [<c03ccbaf>] __mutex_lock_slowpath+0x13f/0x280
[93422.248471]  [<c03ccbaf>] __mutex_lock_slowpath+0x13f/0x280
[93422.248478]  [<c0133b49>] trace_hardirqs_on+0xb9/0x160
[93422.248484]  [<c0165305>] vfs_unlink+0x75/0xe0
[93422.248491]  [<c03ccba2>] __mutex_lock_slowpath+0x132/0x280
[93422.248498]  [<c0165305>] vfs_unlink+0x75/0xe0
[93422.248504]  [<c01648e1>] permission+0x91/0xf0
[93422.248512]  [<c0165319>] vfs_unlink+0x89/0xe0
[93422.248518]  [<c0167062>] do_unlinkat+0xd2/0x160
[93422.248526]  [<c0102894>] sysenter_past_esp+0x8d/0x99
[93422.248533]  [<c0133b49>] trace_hardirqs_on+0xb9/0x160
[93422.248541]  [<c0102864>] sysenter_past_esp+0x5d/0x99
[93422.248550]  =======================

Comment 4 Christian Kujau 2007-11-18 05:41:13 UTC

Last seen in 2.6.23-rc6 (09/2007), but I've not seen this in 2.6.23-rc7, 2.6.23-rc9, 2.6.23.1 and 2.6.24-rc2. So, I guess the lock annotations for XFS made it into the kernel? If so, I'd like to see this bug closed.

Thanks to all involved,
Christian.

# zgrep -E 'DEBUG|LOCK' /proc/config.gz | grep -v ^\#
[...]
CONFIG_LOCKDEP_SUPPORT=y
CONFIG_DETECT_SOFTLOCKUP=y
CONFIG_DEBUG_RT_MUTEXES=y
CONFIG_DEBUG_PI_LIST=y
CONFIG_DEBUG_SPINLOCK=y
CONFIG_DEBUG_MUTEXES=y
CONFIG_DEBUG_LOCK_ALLOC=y
CONFIG_PROVE_LOCKING=y
CONFIG_LOCKDEP=y
CONFIG_DEBUG_SPINLOCK_SLEEP=y
CONFIG_DEBUG_LOCKING_API_SELFTESTS=y
CONFIG_DEBUG_INFO=y
CONFIG_DEBUG_LIST=y
CONFIG_DEBUG_STACKOVERFLOW=y

Comment 5 Martin Mokrejs 2007-11-18 12:41:19 UTC

Too bad, I have stopped using XFS on my laptop, so I cannot comment on this anymore, sorry.

Comment 6 Dave Chinner 2007-11-18 15:29:00 UTC

Fixed with this commit:

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=0f1145cc18e970ebe37da114fc34c297f135e062

(reassigning so I can close the bug)