Re: [PATCH v14] Add cli and ioctl to forget scanned device(s)

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

On 2/6/19 11:33 PM, David Sterba wrote:
> On Fri, Jan 04, 2019 at 01:31:51PM +0800, Anand Jain wrote:
>> Adds cli and ioctl to forget a scanned device or forget all stale
>> devices in the kernel.
>>
>> Anand Jain (2):
>>    btrfs: refactor btrfs_free_stale_devices() to get return value
>>    btrfs: introduce feature to forget a btrfs device
> 
> The 2nd patch has been added to misc-next too, I've updated the
> changelog a bit.

Thanks.

> The usecase here is quite small so I don't think we've
> missed something crucial on the kernel side. 


> The userspace part can
> provide some convenience shortcuts like --forget-all 

The --forget is already equivalent to --forget-all. For example..

# btrfs dev scan /dev/sdb
Scanning for Btrfs filesystems in '/dev/sdb'
# btrfs dev scan --forget /dev/sdb
# btrfs dev scan
Scanning for Btrfs filesystems
# btrfs dev scan --forget
# echo $?
0
# btrfs dev scan --forget /dev/sdb
ERROR: Can't forget '/dev/sdb': No such file or directory
# echo $?
1
#

 > or specifying by
 > filesystem id where the translation would happen internally, but the
 > ioctl would cover that. Thanks.

Looks like that seeks the complete enhancement to scan cli.

IMO we could add..

  #btrfs dev scan --show

  For which looks like we need to enhance existing ioctl
  BTRFS_IOC_FS_INFO and BTRFS_IOC_DEV_INFO so that it works
  through btrfs-control.
  Which means 'btrfs fi show' can also show the scanned devices from
  the kernel, probably we should just add flags to the output to indicate
  if the fsid-device is mounted.

  OR a wrapper around the sysfs fs-dev tree implementation will suffice
  as well.

further, (fsid translation happens internally)

  #btrfs dev scan <fsid>
  #btrfs dev scan --forget <fsid>

In general, agree use case is limited.

Thanks, Anand