+0
−14
+0
−1
+0
−7
Loading
Previously, when a withdraw occurred, we would wait for another node to recover our journal. This also meant that frozen filesystem needed to be thawed because otherwise, other nodes wouldn't be able to recover the filesystem. With the reversal of commit 601ef0d5 ("gfs2: Force withdraw to replay journals and wait for it to finish"), we are no longer waiting for journal recovery during a withdraw, so we no longer need to thaw frozen filesystems, either. This also fixes a potential deadlock reported by lockdep when running xfstest generic/108. In addition, there is nothing left in do_withdraw() that would require taking sd_freeze_mutex, so don't bother taking that lock there anymore. Signed-off-by:Andreas Gruenbacher <agruenba@redhat.com>