Re: [PATCH v2] btrfs: allocate raid type kobjects dynamically
This patch + the fix to add/remove links on the respective device
operation seem to work with the previous test that used to catch the
slab corruptions, so far so good.
After this has been running fine for a few hours, I've tried to do a
simple check
# cd /sys/fs/btrfs
# find . -type f -exec cat '{}' \;
that produced this lockdep warning:
[ 4943.485655] run xfstest btrfs/004
[ 4944.053971] BTRFS: device fsid 6b90f3ce-cd83-44da-88e2-c89a4607b782 devid 1
transid 4 /dev/sda9
[ 4944.067637] BTRFS info (device sda9): disk space caching is enabled
[ 4944.075621] BTRFS: flagging fs with big metadata feature
[ 4944.093954] BTRFS: creating UUID tree
[ 4950.077597] BTRFS info (device sda9): setting 8 feature flag
[ 4950.084961] BTRFS info (device sda9): disk space caching is enabled
[ 4952.992993]
[ 4952.996020] ==
[ 4952.996020] [ INFO: possible circular locking dependency detected ]
[ 4952.996020] 3.15.0-rc7-default+ #146 Tainted: GW
[ 4952.996020] ---
[ 4952.996020] cat/3822 is trying to acquire lock:
[ 4952.996020] (found-groups_sem){..}, at: [a004babe]
raid_bytes_show+0x3e/0xd0 [btrfs]
[ 4952.996020]
[ 4952.996020] but task is already holding lock:
[ 4952.996020] (s_active#149){.+}, at: [811ef1c9]
kernfs_seq_start+0x39/0xb0
[ 4952.996020]
[ 4952.996020] which lock already depends on the new lock.
[ 4952.996020]
[ 4952.996020]
[ 4952.996020] the existing dependency chain (in reverse order) is:
[ 4952.996020]
- #1 (s_active#149){.+}:
[ 4952.996020][810b0662] lock_acquire+0x92/0x120
[ 4952.996020][811edc07] __kernfs_remove+0x2b7/0x380
[ 4952.996020][811ee787] kernfs_remove+0x27/0x40
[ 4952.996020][811f0b3a] sysfs_remove_dir+0x5a/0x90
[ 4952.996020][813ba028] kobject_del+0x18/0x90
[ 4952.996020][a0023cb2] btrfs_remove_block_group+0x442/0x580
[btrfs]
[ 4952.996020][a005caf4] btrfs_relocate_chunk+0x624/0x770
[btrfs]
[ 4952.996020][a005f652] btrfs_balance+0x902/0xf50 [btrfs]
[ 4952.996020][a006b090] btrfs_ioctl_balance+0x1e0/0x350
[btrfs]
[ 4952.996020][a006d4a9] btrfs_ioctl+0xc39/0x1830 [btrfs]
[ 4952.996020][8118c381] do_vfs_ioctl+0x91/0x560
[ 4952.996020][8118c8a3] SyS_ioctl+0x53/0x80
[ 4952.996020][81a16cd2] system_call_fastpath+0x16/0x1b
[ 4952.996020]
- #0 (found-groups_sem){..}:
[ 4952.996020][810afc5c] __lock_acquire+0x1c4c/0x1fa0
[ 4952.996020][810b0662] lock_acquire+0x92/0x120
[ 4952.996020][81a0bedc] down_read+0x4c/0xa0
[ 4952.996020][a004babe] raid_bytes_show+0x3e/0xd0 [btrfs]
[ 4952.996020][813b9ef6] kobj_attr_show+0x16/0x20
[ 4952.996020][811f04e9] sysfs_kf_seq_show+0xd9/0x230
[ 4952.996020][811eec56] kernfs_seq_show+0x26/0x30
[ 4952.996020][8119ec8f] seq_read+0xef/0x410
[ 4952.996020][811efa35] kernfs_fop_read+0x125/0x180
[ 4952.996020][81179fe4] vfs_read+0xb4/0x180
[ 4952.996020][8117a269] SyS_read+0x59/0xd0
[ 4952.996020][81a16cd2] system_call_fastpath+0x16/0x1b
[ 4952.996020]
[ 4952.996020] other info that might help us debug this:
[ 4952.996020]
[ 4952.996020] Possible unsafe locking scenario:
[ 4952.996020]
[ 4952.996020]CPU0CPU1
[ 4952.996020]
[ 4952.996020] lock(s_active#149);
[ 4952.996020]lock(found-groups_sem);
[ 4952.996020]lock(s_active#149);
[ 4952.996020] lock(found-groups_sem);
[ 4952.996020]
[ 4952.996020] *** DEADLOCK ***
[ 4952.996020]
[ 4952.996020] 3 locks held by cat/3822:
[ 4952.996020] #0: (p-lock){+.+.+.}, at: [8119ebdf]
seq_read+0x3f/0x410
[ 4952.996020] #1: (of-mutex){+.+.+.}, at: [811ef1c1]
kernfs_seq_start+0x31/0xb0
[ 4952.996020] #2: (s_active#149){.+}, at: [811ef1c9]
kernfs_seq_start+0x39/0xb0
[ 4952.996020]
[ 4952.996020] stack backtrace:
[ 4952.996020] CPU: 0 PID: 3822 Comm: cat Tainted: GW
3.15.0-rc7-default+ #146
[ 4952.996020] Hardware name: Intel Corporation Santa Rosa platform/Matanzas,
BIOS TSRSCRB1.86C.0047.B00.0610170821 10/17/06
[ 4952.996020] 82828a50 880065aedb88 81a07898
0001
[ 4952.996020] 82821230 880065aedbd8 810acb74
001d4500
[ 4952.996020] 880065aedc58 0002 880066ab2c08
0002
[ 4952.996020] Call Trace:
[ 4952.996020] [81a07898] dump_stack+0x51/0x71
[ 4952.996020] [810acb74] print_circular_bug+0x214/0x310
[ 4952.996020] [810afc5c] __lock_acquire+0x1c4c/0x1fa0
[ 4952.996020] [a004babe] ?
Re: [PATCH v2] btrfs: allocate raid type kobjects dynamically
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 5/26/14, 1:21 PM, David Sterba wrote:
This patch + the fix to add/remove links on the respective device
operation seem to work with the previous test that used to catch
the slab corruptions, so far so good.
After this has been running fine for a few hours, I've tried to do
a simple check
# cd /sys/fs/btrfs # find . -type f -exec cat '{}' \;
that produced this lockdep warning:
Thanks David. That's easy enough to fix.
- -Jeff
[ 4943.485655] run xfstest btrfs/004 [ 4944.053971] BTRFS: device
fsid 6b90f3ce-cd83-44da-88e2-c89a4607b782 devid 1 transid 4
/dev/sda9 [ 4944.067637] BTRFS info (device sda9): disk space
caching is enabled [ 4944.075621] BTRFS: flagging fs with big
metadata feature [ 4944.093954] BTRFS: creating UUID tree [
4950.077597] BTRFS info (device sda9): setting 8 feature flag [
4950.084961] BTRFS info (device sda9): disk space caching is
enabled [ 4952.992993] [ 4952.996020]
== [
4952.996020] [ INFO: possible circular locking dependency detected
] [ 4952.996020] 3.15.0-rc7-default+ #146 Tainted: GW [
4952.996020]
--- [
4952.996020] cat/3822 is trying to acquire lock: [ 4952.996020]
(found-groups_sem){..}, at: [a004babe]
raid_bytes_show+0x3e/0xd0 [btrfs] [ 4952.996020] [ 4952.996020]
but task is already holding lock: [ 4952.996020]
(s_active#149){.+}, at: [811ef1c9]
kernfs_seq_start+0x39/0xb0 [ 4952.996020] [ 4952.996020] which
lock already depends on the new lock. [ 4952.996020] [ 4952.996020]
[ 4952.996020] the existing dependency chain (in reverse order) is:
[ 4952.996020] - #1 (s_active#149){.+}: [ 4952.996020]
[810b0662] lock_acquire+0x92/0x120 [ 4952.996020]
[811edc07] __kernfs_remove+0x2b7/0x380 [ 4952.996020]
[811ee787] kernfs_remove+0x27/0x40 [ 4952.996020]
[811f0b3a] sysfs_remove_dir+0x5a/0x90 [ 4952.996020]
[813ba028] kobject_del+0x18/0x90 [ 4952.996020]
[a0023cb2] btrfs_remove_block_group+0x442/0x580 [btrfs]
[ 4952.996020][a005caf4]
btrfs_relocate_chunk+0x624/0x770 [btrfs] [ 4952.996020]
[a005f652] btrfs_balance+0x902/0xf50 [btrfs] [
4952.996020][a006b090]
btrfs_ioctl_balance+0x1e0/0x350 [btrfs] [ 4952.996020]
[a006d4a9] btrfs_ioctl+0xc39/0x1830 [btrfs] [
4952.996020][8118c381] do_vfs_ioctl+0x91/0x560 [
4952.996020][8118c8a3] SyS_ioctl+0x53/0x80 [
4952.996020][81a16cd2]
system_call_fastpath+0x16/0x1b [ 4952.996020] - #0
(found-groups_sem){..}: [ 4952.996020] [810afc5c]
__lock_acquire+0x1c4c/0x1fa0 [ 4952.996020] [810b0662]
lock_acquire+0x92/0x120 [ 4952.996020] [81a0bedc]
down_read+0x4c/0xa0 [ 4952.996020] [a004babe]
raid_bytes_show+0x3e/0xd0 [btrfs] [ 4952.996020]
[813b9ef6] kobj_attr_show+0x16/0x20 [ 4952.996020]
[811f04e9] sysfs_kf_seq_show+0xd9/0x230 [ 4952.996020]
[811eec56] kernfs_seq_show+0x26/0x30 [ 4952.996020]
[8119ec8f] seq_read+0xef/0x410 [ 4952.996020]
[811efa35] kernfs_fop_read+0x125/0x180 [ 4952.996020]
[81179fe4] vfs_read+0xb4/0x180 [ 4952.996020]
[8117a269] SyS_read+0x59/0xd0 [ 4952.996020]
[81a16cd2] system_call_fastpath+0x16/0x1b [ 4952.996020]
[ 4952.996020] other info that might help us debug this: [
4952.996020] [ 4952.996020] Possible unsafe locking scenario: [
4952.996020] [ 4952.996020]CPU0CPU1 [
4952.996020] [ 4952.996020]
lock(s_active#149); [ 4952.996020] lock(found-groups_sem); [
4952.996020] lock(s_active#149); [ 4952.996020]
lock(found-groups_sem); [ 4952.996020] [ 4952.996020] ***
DEADLOCK *** [ 4952.996020] [ 4952.996020] 3 locks held by
cat/3822: [ 4952.996020] #0: (p-lock){+.+.+.}, at:
[8119ebdf] seq_read+0x3f/0x410 [ 4952.996020] #1:
(of-mutex){+.+.+.}, at: [811ef1c1]
kernfs_seq_start+0x31/0xb0 [ 4952.996020] #2:
(s_active#149){.+}, at: [811ef1c9]
kernfs_seq_start+0x39/0xb0 [ 4952.996020] [ 4952.996020] stack
backtrace: [ 4952.996020] CPU: 0 PID: 3822 Comm: cat Tainted: G W
3.15.0-rc7-default+ #146 [ 4952.996020] Hardware name: Intel
Corporation Santa Rosa platform/Matanzas, BIOS
TSRSCRB1.86C.0047.B00.0610170821 10/17/06 [ 4952.996020]
82828a50 880065aedb88 81a07898 0001
[ 4952.996020] 82821230 880065aedbd8 810acb74
001d4500 [ 4952.996020] 880065aedc58 0002
880066ab2c08 0002 [ 4952.996020] Call Trace: [
4952.996020] [81a07898] dump_stack+0x51/0x71 [
4952.996020] [810acb74] print_circular_bug+0x214/0x310 [
4952.996020] [810afc5c] __lock_acquire+0x1c4c/0x1fa0 [
