Re: [PATCH v4 3/3] btrfs: free alien device due to device add

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

>>
>> Signed-off-by: Anand Jain <anand.jain@oracle.com>
>> Reviewed-by: Josef Bacik <josef@toxicpanda.com>
> 
> So I'm not sure how much the review holds when the code changes between
> iterations. The right way is to drop the rev-by and CC the person for
> any non-trivial or maybe for all changes.
> 

  only thing that changed is use of helper function
  btrfs_forget_devices() instead of the open-code, so I kept the rev-by.
  Yeah I kind of noticed late that cc list added during git format-patch
  didn't actually add cc in the git send <patch-file>


>> --- a/fs/btrfs/volumes.c
>> +++ b/fs/btrfs/volumes.c
>> @@ -2664,6 +2664,15 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
>>   
>>   	/* Update ctime/mtime for libblkid */
>>   	update_dev_time(device_path);
>> +
>> +	/*
>> +	 * Now that we have written a new sb into this device, check all other
>> +	 * fs_devices list if device_path alienates any other scanned device.
>> +	 * Ignore the return as we are successfull in the core task - add
>> +	 * the device.
>> +	 */
>> +	btrfs_forget_devices(device_path);
> 
> Actually this should go before update_dev_time, as this is the point
> when the file change could be detected by udev and rescanned, which
> increases the window between commit and removal from fs_devices.
> 

  You are right, theoretically btrfs_forget_devices() should be before
  update_dev_time()


Noted and agreed on rest of the stuffs.

Thanks, Anand