> On Mar 13, 2023, at 10:50 PM, Amir Goldstein <amir73il@xxxxxxxxx> wrote:
>
> On Tue, Mar 14, 2023 at 6:27 AM Catherine Hoang
> <catherine.hoang@xxxxxxxxxx> wrote:
>>
>> Add a new ioctl to set the uuid of a mounted filesystem.
>>
>> Signed-off-by: Catherine Hoang <catherine.hoang@xxxxxxxxxx>
>> ---
>> fs/xfs/libxfs/xfs_fs.h | 1 +
>> fs/xfs/xfs_ioctl.c | 107 +++++++++++++++++++++++++++++++++++++++++
>> fs/xfs/xfs_log.c | 19 ++++++++
>> fs/xfs/xfs_log.h | 2 +
>> 4 files changed, 129 insertions(+)
>>
>> diff --git a/fs/xfs/libxfs/xfs_fs.h b/fs/xfs/libxfs/xfs_fs.h
>> index 1cfd5bc6520a..a350966cce99 100644
>> --- a/fs/xfs/libxfs/xfs_fs.h
>> +++ b/fs/xfs/libxfs/xfs_fs.h
>> @@ -831,6 +831,7 @@ struct xfs_scrub_metadata {
>> #define XFS_IOC_FSGEOMETRY _IOR ('X', 126, struct xfs_fsop_geom)
>> #define XFS_IOC_BULKSTAT _IOR ('X', 127, struct xfs_bulkstat_req)
>> #define XFS_IOC_INUMBERS _IOR ('X', 128, struct xfs_inumbers_req)
>> +#define XFS_IOC_SETFSUUID _IOR ('X', 129, uuid_t)
>
> Should be _IOW.
Ok, will fix that.
>
> Would you consider defining that as FS_IOC_SETFSUUID in fs.h,
> so that other fs could implement it later on, instead of hoisting it later?
>
> It would be easy to add support for FS_IOC_SETFSUUID to ext4
> by generalizing ext4_ioctl_setuuid().
>
> Alternatively, we could hoist EXT4_IOC_SETFSUUID and struct fsuuid
> to fs.h and use that ioctl also for xfs.
I actually did try to hoist the ext4 ioctls previously, but we weren’t able to come
to a consensus on the implementation.
https://lore.kernel.org/linux-xfs/20221118211408.72796-2-catherine.hoang@xxxxxxxxxx/
I would prefer to keep this defined as an xfs specific ioctl to avoid all of the
fsdevel bikeshedding.
>
> Using an extensible struct with flags for that ioctl may turn out to be useful,
> for example, to verify that the new uuid is unique, despite the fact
> that xfs was
> mounted with -onouuid (useful IMO) or to explicitly request a restore of old
> uuid that would fail if new_uuid != meta uuid.
I think using a struct is probably a good idea, I can add that in the next version.
>
> Thanks,
> Amir.
