On Fri, Jan 15, 2010 at 10:56:11AM -0800, Linus Torvalds wrote:
> It's insane to probe for zfs, ufs or reiserfs after you have already found
> a vfat filesystem.
# mkswap /dev/loop0
Setting up swapspace version 1, size = 51196 KiB
no label, UUID=2553d8ed-9201-4273-854e-04143d454f3e
# cryptsetup luksFormat /dev/loop0
[...]
Command successful.
Is it swap or LUKS? Both superblocks on the device are _VALID_. Yeah,
this is bug -- mkfs, mkswap or cryptsetup should be able to remove
old signatures.
Now imagine that some insane init scripts are able to automatically
call swapon always when a swap header on the device is detected. This
is not paranoid fantasy, this is experience...
We have to be paranoid before we say that there is FAT or LUKS or
whatever.
This feature has been explicitly requested by udev folks. The
libblkid library also provides less paranoid probing method.
(Note that mkswap and cryptsetup should be fixed now.)
> Don't you see that? The whole and only point of blkid is to do the right
> thing automatically.
I understand your point of view, and I have no problem to add to
blkid all necessary functionality for compatibility with your
point of view. But I have doubts that udev will use it.
All this was discussed 2 (?) years ago. The current blkid behavior
is based on the results from the discussion.
I'll
- fix -S
- try to remove the large read()
- add an option to allow run blkid in less paranoid mode (return
the first detected superblock)
Karel
--
Karel Zak <kzak@xxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe util-linux-ng" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
