Re: [PATCH] btrfs: fix inconsistency during missing device rejoin

[Anand Jain <anand.jain@oracle.com>]

>>> @@ -715,7 +715,8 @@ static noinline int device_list_add(const char 
>>> *path,
>>>           ret = 1;
>>>           device->fs_devices = fs_devices;
>>> -    } else if (!device->name || strcmp(device->name->str, path)) {
>>> +    } else if (!device->name || strcmp(device->name->str, path) ||
>>> +            test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state)) {
>>>           /*
>>>            * When FS is already mounted.
>>>            * 1. If you are here and if the device->name is NULL that
>>>
>>
>> I read the comments below this and wonder if BTRFS_DEV_STATE_MISSING 
>> is set
>> when device->name is null and therefore the first condition can be 
>> removed.

  On the 2nd thought, actually can not remove device->name, there can
  be a case where degraded FS is unmounted, then device->name is NULL.

Thanks, Anand

>   device->name is null only when missing is set.
> 
>   Will do this optimize in a new patch, as its about clean up and
>   is not about fixing inconsistency as in here.
> 
> Thanks, Anand