On Wed, Nov 09, 2022 at 02:19:59PM -0800, Catherine Hoang wrote:
> Add a new ioctl to retrieve the UUID of a mounted xfs filesystem.
I think it's worth mentioning that this is the precursor to trying to
implement SETFSUUID... but that's something for a future series, since
changing the uuid will upset the log, and we have to figure out how to
deal with that gracefully.
> Signed-off-by: Catherine Hoang <catherine.hoang@xxxxxxxxxx>
> ---
> fs/xfs/xfs_ioctl.c | 32 ++++++++++++++++++++++++++++++++
> 1 file changed, 32 insertions(+)
>
> diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
> index 1f783e979629..657fe058dfba 100644
> --- a/fs/xfs/xfs_ioctl.c
> +++ b/fs/xfs/xfs_ioctl.c
> @@ -1865,6 +1865,35 @@ xfs_fs_eofblocks_from_user(
> return 0;
> }
>
> +static int xfs_ioctl_getuuid(
Nit: function names should start on a new line.
> + struct xfs_mount *mp,
> + struct fsuuid __user *ufsuuid)
> +{
> + struct fsuuid fsuuid;
> + __u8 uuid[UUID_SIZE];
> +
> + if (copy_from_user(&fsuuid, ufsuuid, sizeof(fsuuid)))
> + return -EFAULT;
> +
> + if (fsuuid.fsu_len == 0) {
> + fsuuid.fsu_len = UUID_SIZE;
> + if (copy_to_user(ufsuuid, &fsuuid, sizeof(fsuuid.fsu_len)))
> + return -EFAULT;
> + return -EINVAL;
Ted and I were looking through the ext4_ioctl_getuuid function on this
morning's ext4 concall, and we decided that copying the desired uuid
buffer length out to userspace shouldn't result in an EINVAL return
here...
> + }
> +
> + if (fsuuid.fsu_len != UUID_SIZE || fsuuid.fsu_flags != 0)
...and that we shouldn't reject the case where fsu_len > UUID_SIZE.
Instead, we should copy the uuid and update the caller's fsu_len to
reflect however many bytes we copied out. I'll send patches to do that
shortly.
> + return -EINVAL;
> +
> + spin_lock(&mp->m_sb_lock);
> + memcpy(uuid, &mp->m_sb.sb_uuid, UUID_SIZE);
> + spin_unlock(&mp->m_sb_lock);
> +
> + if (copy_to_user(&ufsuuid->fsu_uuid[0], uuid, UUID_SIZE))
> + return -EFAULT;
The rest of this logic looks correct to me. Thanks for getting this out
there.
--D
> + return 0;
> +}
> +
> /*
> * These long-unused ioctls were removed from the official ioctl API in 5.17,
> * but retain these definitions so that we can log warnings about them.
> @@ -2153,6 +2182,9 @@ xfs_file_ioctl(
> return error;
> }
>
> + case FS_IOC_GETFSUUID:
> + return xfs_ioctl_getuuid(mp, arg);
> +
> default:
> return -ENOTTY;
> }
> --
> 2.25.1
>
