Re: [PATCH] ocfs2: reflink deadlock when clone file to the same directory simultaneously

From: Joseph Qi <joseph.qi@linux.alibaba.com>
To: Gang He <ghe@suse.com>,
	mark@fasheh.com, jlbec@evilplan.org,
	Wengang Wang <wen.gang.wang@oracle.com>
Cc: linux-kernel@vger.kernel.org, ocfs2-devel@oss.oracle.com,
	akpm@linux-foundation.org
Subject: Re: [PATCH] ocfs2: reflink deadlock when clone file to the same directory simultaneously
Date: Fri, 13 Aug 2021 17:54:47 +0800	[thread overview]
Message-ID: <71608a14-58f4-dba0-d695-fee65de89192@linux.alibaba.com> (raw)
In-Reply-To: <c7e1f0a7-e75c-d1e0-870d-dc480d070079@suse.com>

On 8/9/21 6:08 PM, Gang He wrote:
> Hi Joseph and All,
> 
> The deadlock is caused by self-locking on one node.
> There is three node cluster (mounted to /mnt/shared), the user run reflink command to clone the file to the same directory repeatedly,
> e.g.
>  reflink "/mnt/shared/test" \
>  "/mnt/shared/.snapshots/test.`date +%m%d%H%M%S`.`hostname`"
> 
> After a while, the reflink process on each node is hung, the file system cannot be listed.
> The problematic reflink command process is blocked by itself, e.g. the reflink process is hung at ghe-sle15sp2-nd2,
> kernel: task:reflink         state:D stack:    0 pid:16992 ppid:  4530
> kernel: Call Trace:
> kernel:  __schedule+0x2fd/0x750
> kernel:  ? try_to_wake_up+0x17b/0x4e0
> kernel:  schedule+0x2f/0xa0
> kernel:  schedule_timeout+0x1cc/0x310
> kernel:  ? __wake_up_common+0x74/0x120
> kernel:  wait_for_completion+0xba/0x140
> kernel:  ? wake_up_q+0xa0/0xa0
> kernel:  __ocfs2_cluster_lock.isra.41+0x3b5/0x820 [ocfs2]
> kernel:  ? ocfs2_inode_lock_full_nested+0x1fc/0x960 [ocfs2]
> kernel:  ocfs2_inode_lock_full_nested+0x1fc/0x960 [ocfs2]
> kernel:  ocfs2_init_security_and_acl+0xbe/0x1d0 [ocfs2]
> kernel:  ocfs2_reflink+0x436/0x4c0 [ocfs2]
> kernel:  ? ocfs2_reflink_ioctl+0x2ca/0x360 [ocfs2]
> kernel:  ocfs2_reflink_ioctl+0x2ca/0x360 [ocfs2]
> kernel:  ocfs2_ioctl+0x25e/0x670 [ocfs2]
> kernel:  do_vfs_ioctl+0xa0/0x680
> kernel:  ksys_ioctl+0x70/0x80
> 
> In fact, the destination directory(.snapshots) inode dlm lock was acquired by ghe-sle15sp2-nd2, next there is bast message from other nodes to ask ghe-sle15sp2-nd2 downconvert lock, but the operation failed, the kernel message is printed like,
> kernel: (ocfs2dc-AA35DD9,2560,3):ocfs2_downconvert_lock:3660 ERROR: DLM error -16 while calling ocfs2_dlm_lock on resource M0000000000000000046e0200000000
> kernel: (ocfs2dc-AA35DD9,2560,3):ocfs2_unblock_lock:3904 ERROR: status = -16
> kernel: (ocfs2dc-AA35DD9,2560,3):ocfs2_process_blocked_lock:4303 ERROR: status = -16
> 
> Then, the reflink process tries to acquire this directory inode dlm lock, the process is blocked, the dlm lock resource in memory looks like
> 
>     l_name = "M0000000000000000046e0200000000",
>     l_ro_holders = 0,
>     l_ex_holders = 0,
>     l_level = 5 '\005',
>     l_requested = 0 '\000',
>     l_blocking = 5 '\005',
>     l_type = 0 '\000',
>     l_action = 0 '\000',
>     l_unlock_action = 0 '\000',
>     l_pending_gen = 645948,
> 
> 
> So far, I do not know what makes dlm lock function failed, it also looks we do not handle this failure case in dlmglue layer, but I always reproduce this hang with my test script, e.g.
> 
>   loop=1
>   while ((loop++)) ; do
>         for i in `seq 1 100`; do
>           reflink "/mnt/shared/test" "/mnt/shared/.snapshots /test.${loop}.${i}.`date +%m%d%H%M%S`.`hostname`"
>         done
>         usleep 500000
>         rm -f /mnt/shared/.snapshots/testnode1.qcow2.*.`hostname`
>   done
> 
> My patch changes multiple acquiring dest directory inode dlm lock during in ocfs2_reflink function, it avoids the hang issue happen again.The code change also can improve reflink performance in this case.
> 
> Thanks
> Gang

'status = -16' implies DLM_CANCELGRANT.
Do you use stack user instead of o2cb? If yes, can you try o2cb with
your reproducer?

Thanks,
Joseph