Re: Infiniate systemd loop when power off the machine with multiple MD RAIDs

[Hannes Reinecke <hare@suse.de>]

On 8/23/23 03:28, Yu Kuai wrote:
> Hi,
> 
> 在 2023/08/23 3:13, Carlos Carvalho 写道:
>> Song Liu (song@kernel.org) wrote on Tue, Aug 22, 2023 at 03:56:04PM -03:
>>> >From systemd code, i.e. function delete_md(), this error:
>>>
>>> [ 205.957004] systemd-shutdown[1]: Stopping MD /dev/md124p1 (259:6).
>>> [ 205.964177] systemd-shutdown[1]: Could not stop MD /dev/md124p1:
>>> Device or resource busy
>>>
>>> is most likely triggered by ioctl(STOP_ARRAY).
>>>
>>> And based on the code, I think the ioctl fails here:
>>>
>>>          if (cmd == STOP_ARRAY || cmd == STOP_ARRAY_RO) {
>>>                  /* Need to flush page cache, and ensure no-one else 
>>> opens
>>>                   * and writes
>>>                   */
>>>                  mutex_lock(&mddev->open_mutex);
>>>                  if (mddev->pers && atomic_read(&mddev->openers) > 1) {
>>>                          mutex_unlock(&mddev->open_mutex);
>>>                          err = -EBUSY;
>>>                          goto out;        ////////////////////// HERE
> 
> Yes, I suspect here as well, but I do some test with error injection to
> gurantee that ioctl(STOP_ARRAY) always return -EBUSY, but I found that
> system reboot didn't hang, it'll try a few times but eventually reboot
> finished.
> 
>>>                  }
>>>                  if (test_and_set_bit(MD_CLOSING, &mddev->flags)) {
>>>                          mutex_unlock(&mddev->open_mutex);
>>>                          err = -EBUSY;
>>>                          goto out;
>>>                  }
>>>                  did_set_md_closing = true;
>>>                  mutex_unlock(&mddev->open_mutex);
>>>                  sync_blockdev(bdev);
>>>          }
>>
>> Probably. The problem is why doesn't it manage to flush the page 
>> cache? I find
>> strange that the problem appears only when trying to stop the array, I 
>> get it
>> when trying to umount the filesystem, where it also hangs because of 
>> the same
>> reason. The kworker thread runs continuously using 100% cpu of only 1 
>> core.
> 
> The kworker do you mean the daemon thread or the sync thread?  runs
> continuously using 100% cpu doen't sounds correct to me.
> 
That sounds more like kworker waiting on a mutex to be released.
Can you trigger a kdump while in that state?
Or do a sysrq-T to get a stack trace of all running processes?
That should tell you where exactly the 100% cpu time is being
wasted.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke                Kernel Storage Architect
hare@suse.de                              +49 911 74053 688
SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), Geschäftsführer: Ivo Totev, Andrew
Myers, Andrew McDonald, Martje Boudien Moerman