Ok not that simple, the running status is checked outside of
balance_mutex and there's one more assertion that does not expect the
balance_ctl to exist:
@@ -4031,16 +4032,16 @@ int btrfs_pause_balance(struct btrfs_fs_info *fs_info)
return -ENOTCONN;
}
- if (atomic_read(&fs_info->balance_running)) {
+ if (test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) {
atomic_inc(&fs_info->balance_pause_req);
mutex_unlock(&fs_info->balance_mutex);
wait_event(fs_info->balance_wait_q,
- atomic_read(&fs_info->balance_running) == 0);
+ !test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags));
here it's unlocked
mutex_lock(&fs_info->balance_mutex);
/* we are good with balance_ctl ripped off from under us */
- BUG_ON(atomic_read(&fs_info->balance_running));
+ BUG_ON(test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags));
and rewriting the code so this could be checked the same way is not a simple
fixup as I expected.
As there's still the extra balance mutex lock/unlock after the volume
mutex removal, I'll have a look how this could be cleaned up further.
atomic_dec(&fs_info->balance_pause_req);
} else {
ret = -ENOTCONN;
Makes sense.
Thanks, Anand
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
