shift percpu_counter_destroy() into destroy_super_work()
authorOleg Nesterov <oleg@redhat.com>
Wed, 22 Jul 2015 18:21:13 +0000 (20:21 +0200)
committerOleg Nesterov <oleg@redhat.com>
Sat, 15 Aug 2015 11:52:11 +0000 (13:52 +0200)
commit853b39a7c82826b8413048feec7bf08e98ce7a84
tree459b6afcc9d478a4beb6347627cc13da2e24b132
parentbf3eac84c42da7017610abc8cfba64921ea92c76
shift percpu_counter_destroy() into destroy_super_work()

Of course, this patch is ugly as hell. It will be (partially)
reverted later. We add it to ensure that other WIP changes in
percpu_rw_semaphore won't break fs/super.c.

We do not even need this change right now, percpu_free_rwsem()
is fine in atomic context. But we are going to change this, it
will be might_sleep() after we merge the rcu_sync() patches.

And even after that we do not really need destroy_super_work(),
we will kill it in any case. Instead, destroy_super_rcu() should
just check that rss->cb_state == CB_IDLE and do call_rcu() again
in the (very unlikely) case this is not true.

So this is just the temporary kludge which helps us to avoid the
conflicts with the changes which will be (hopefully) routed via
rcu tree.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reviewed-by: Jan Kara <jack@suse.com>
fs/super.c
include/linux/fs.h