On Fri, Jan 17, 2020 at 11:10:45AM +0800, yukuai (C) wrote: > > > On 2020/1/17 10:20, yukuai (C) wrote: > > After adding "-o 4", I tested over 200 times, and blocktrash never > > failed to corrupt xattr anymore. > > Unfortunately, test failed with more attempts: > > _check_dmesg: something found in dmesg (see > /root/xfstests-dev/results//xfs/126.dmesg) > > [ 4597.649086] BUG: MAX_LOCKDEP_CHAIN_HLOCKS too low! > [ 4597.649709] turning off the locking correctness validator. That's a deficiency in lockdep not being able to follow XFS crazy locking. It's not a bug in XFS itself. --D > [ 4597.650363] CPU: 4 PID: 377 Comm: kworker/4:1H Not tainted 5.5.0-rc6 #197 > [ 4597.651027] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS > ?-20180531_142017-buildhw-08.phx2.fedoraproject.org-1.fc28 04/01/2014 > [ 4597.652276] Workqueue: xfs-log/sdb xlog_ioend_work > [ 4597.652803] Call Trace: > [ 4597.653109] dump_stack+0xdd/0x13f > [ 4597.653573] __lock_acquire.cold.46+0x7a/0x409 > [ 4597.654000] lock_acquire+0xf6/0x270 > [ 4597.654487] ? xlog_state_do_callback+0x1eb/0x4e0 > [ 4597.654921] _raw_spin_lock+0x45/0x70 > [ 4597.655250] ? xlog_state_do_callback+0x1eb/0x4e0 > [ 4597.655666] xlog_state_do_callback+0x1eb/0x4e0 > [ 4597.656123] xlog_state_done_syncing+0x8b/0x110 > [ 4597.656727] xlog_ioend_work+0x94/0x150 > [ 4597.657210] process_one_work+0x346/0x910 > [ 4597.657714] worker_thread+0x284/0x6d0 > [ 4597.658125] ? rescuer_thread+0x550/0x550 > [ 4597.658647] kthread+0x168/0x1a0 > [ 4597.658931] ? kthread_unpark+0xb0/0xb0 > [ 4597.659470] ret_from_fork+0x24/0x30 > > I wonder, could we increase the number of "-o 4" to fix this? > > Thanks! > Yu Kuai >
