md: be more relaxed about stopping an array which isn't started.
In general we don't allow an array to be stopped if it is in use.
However if the array hasn't really been started yet, then any
apparent use is an anomily, probably due to 'udev' or similar
having a look to see what is there.
This means that if something goes wrong while assembling an array
it cannot reliably be un-assembled - STOP_ARRAY could fail.
There is no value here, so change do_md_stop() to succeed
despite concurrent opens if the array has not yet been
activated. i.e. if ->pers is NULL.
Reported-by: "Baldysiak, Pawel" <pawel.baldysiak@intel.com>
Signed-off-by: NeilBrown <neilb@suse.de>
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 1294238..4c72e96 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -5307,7 +5307,7 @@
mddev_lock_nointr(mddev);
mutex_lock(&mddev->open_mutex);
- if (atomic_read(&mddev->openers) > !!bdev ||
+ if ((mddev->pers && atomic_read(&mddev->openers) > !!bdev) ||
mddev->sync_thread ||
(bdev && !test_bit(MD_STILL_CLOSED, &mddev->flags))) {
printk("md: %s still in use.\n",mdname(mddev));
@@ -5362,7 +5362,7 @@
mddev_lock_nointr(mddev);
mutex_lock(&mddev->open_mutex);
- if (atomic_read(&mddev->openers) > !!bdev ||
+ if ((mddev->pers && atomic_read(&mddev->openers) > !!bdev) ||
mddev->sysfs_active ||
mddev->sync_thread ||
(bdev && !test_bit(MD_STILL_CLOSED, &mddev->flags))) {
@@ -6454,7 +6454,7 @@
* and writes
*/
mutex_lock(&mddev->open_mutex);
- if (atomic_read(&mddev->openers) > 1) {
+ if (mddev->pers && atomic_read(&mddev->openers) > 1) {
mutex_unlock(&mddev->open_mutex);
err = -EBUSY;
goto abort;