Re: [libvirt] [PATCH 2/2] fs: Fix probing on no-overwrite of target device
On Mon, Nov 21, 2016 at 05:42:15PM -0500, John Ferlan wrote:
[...]
>
> [C] might be more user-friendly, but I'm not sure if we can change the
> meaning of the NO_OVERWRITE constant like that. On the bright side,
> if blkid can also identify partition tables, we could unify the probing
> code with the disk backend and stop parsing parted's output (also, catch
> those cases when someone puts the filesystem on the block device instead
> of its partition).
After some experimenting - the blkid_known_fstype would fail for every
one of the disk format types, so it'd be useless to help detect.
The equivalent for partition tables is blkid_known_pttype.
After enabling the partition detection chain and calling do_probe in a
loop I was able to get "gpt" in the PTYPE property.
Furthermore, blkid_do_probe would fail for every type except when a
pvcreate has been performed on the device path.
Thus it seems we're stuck with parsing parted's output for disk.
Depends on the portability requirements. It seems there is
e2fsprogs-libblkid even on FreeBSD.
Jan
And the fs pool usage of overwrite still makes no sense when compared
to
disk (I think we agree that logical's lack of usage/support for
--overwrite is incorrect too).
John
Attachments:
- signature.asc (application/pgp-signature — 833 bytes)