LU-15404 ldiskfs: use per-filesystem workqueues to avoid deadlocks
Calling flush_scheduled_work() under s_umount is dangerous and may
cause deadlocks. This patch backports the fix from
https://lore.kernel.org/all/
20220402084023.1841375-1-anserper@ya.ru/
Fixes:
e239a14001 ("LU-15404 ldiskfs: truncate during setxattr leads to kernel panic")
Signed-off-by: Andrew Perepechko <andrew.perepechko@hpe.com>
Change-Id: Ia191b70166f94f34e96a282ec18bd8650871e108
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/50354
Reviewed-by: Alex Zhuravlev <bzzz@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>