From: Shaohua Li <shli@xxxxxx>
The raid5-cache .quiesce assumes it's called with reconfig_mutex hold. It's
true at most time but there are some exceptions. Since raid5-cache .quiesce
could write superblock, it must be called with reconfig_mutex hold. We already
have facility in .quiesce to indicate if it's called with the mutex hold. This
patch makes raid5-cache .quiesce use the facility.
Signed-off-by: Shaohua Li <shli@xxxxxx>
---
drivers/md/raid5-cache.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/drivers/md/raid5-cache.c b/drivers/md/raid5-cache.c
index 551158c..e4070d2 100644
--- a/drivers/md/raid5-cache.c
+++ b/drivers/md/raid5-cache.c
@@ -97,6 +97,7 @@ struct r5l_log {
bool need_cache_flush;
bool in_teardown;
+ bool reconfig_mutex_locked;
};
/*
@@ -712,7 +713,7 @@ static void r5l_write_super_and_discard_space(struct r5l_log *log,
* md_check_recovery() fails, so the PENDING never get cleared. The
* in_teardown check workaround this issue.
*/
- if (!log->in_teardown) {
+ if (!log->in_teardown || !log->reconfig_mutex_locked) {
set_mask_bits(&mddev->flags, 0,
BIT(MD_CHANGE_DEVS) | BIT(MD_CHANGE_PENDING));
md_wakeup_thread(mddev->thread);
@@ -826,6 +827,7 @@ void r5l_quiesce(struct r5l_log *log, int state, bool reconfig_mutex_locked)
struct mddev *mddev;
if (!log || state == 2)
return;
+ log->reconfig_mutex_locked = reconfig_mutex_locked;
if (state == 0) {
log->in_teardown = 0;
/*
@@ -850,6 +852,7 @@ void r5l_quiesce(struct r5l_log *log, int state, bool reconfig_mutex_locked)
md_unregister_thread(&log->reclaim_thread);
r5l_do_reclaim(log);
}
+ log->reconfig_mutex_locked = false;
}
bool r5l_log_disk_error(struct r5conf *conf)
--
2.7.4
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
