Clarification on blkid_do_safeprobe

Clarification on blkid_do_safeprobe

[Tony Asleson]

>From looking at the documentation and the code for blkid_do_safeprobe
it's not totally apparent to me that when the function returns -2
(ambivalent) that there is any results which may be retrieved by
utilizing blkid_probe_numof_values and blkid_probe_get_value.

Any clarification on this would be appreciated.  Also, do you only get
the -2 return if BLKID_SUBLKS_BADCSUM is set?

Thanks!
-Tony

Re: Clarification on blkid_do_safeprobe

[Karel Zak]

On Tue, Mar 27, 2018 at 11:28:17AM -0500, Tony Asleson wrote:
> From looking at the documentation and the code for blkid_do_safeprobe
> it's not totally apparent to me that when the function returns -2
> (ambivalent) that there is any results which may be retrieved by
> utilizing blkid_probe_numof_values and blkid_probe_get_value.

After any error blkid_probe_get_value() behavior is undefined. If need
to rescan the device by blkid_do_probe() get all possible results. 

See for example misc-utils/blkid.c:print_udev_ambivalent() where we
compose ID_FS_AMBIVALENT= for udev.

> Any clarification on this would be appreciated.  Also, do you only get
> the -2 return if BLKID_SUBLKS_BADCSUM is set?

You can get -2 when there is more valid superblocks detected on the
same device. The flag BLKID_SUBLKS_BADCSUM may help, but it's not
requirement.

All depends on msfs-like tools, if the tool does not wipe the device
before it creates a new filesystem then you will see more superblocks.

It's unusual now, but years ago it was pretty common. For example it
was possible to create LUKS/VFAT and SWAP on the same device. You can
imagine what happen with our encrypted data after swapon.

    Karel

-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com