UUID anomaly

[Rei Shinozuka]

I've been burnt by this cache business before.  Using libblkid 2.17.0 
the blkid -p option bypasses cache.

-rei


On 07/09/2010 05:42 AM, Markus Schönhaber wrote:
> 09.07.2010 10:59, R Kimber:
>
>    
>> On Fri, 9 Jul 2010 07:39:41 +0200
>>      
>    
>> Nils Kassube wrote:
>>
>>      
>>> I suppose you used "blkid" and not "sudo blkid". Calling blkid without
>>> sudo doesn't necessarily tell you the current state but the one from the
>>> cache file /etc/blkid.tab which was written when blkid was last used
>>> with sudo.
>>>        
> That's interesting, indeed.
> I've never fallen into this trap since I wasn't aware that blkid uses a
> cache file. Therefore, for me, running blkid was associated with "must
> access raw device" and in turn "needs root privileges".
>
>    
>> Ah!  Thank you.  Now they match.  Gparted was right it seems. Maybe there's
>> a case for making blkid not useable without sudo (etc).  I found the blkid
>> command on a webpage that made no mention of root privileges.
>>      
> I guess, there's a reason for making blkid usable as unprivileged user.
> Or, to put it the other way round, something might break if it wasn't.
> I'm not exactly sure if this really is the case, though.
>
> Nevertheless, this possible pitfall might be documented more prominently.
> Information is there:
> the blkid(8) man page
> http://manpages.ubuntu.com/manpages/lucid/en/man8/blkid.8.html
> mentions the cache file and says
> | The blkid program is the command-line interface to working
> | with libblkid(3) library
> and if you look at the libblkid(3) man page
> http://manpages.ubuntu.com/manpages/lucid/en/man3/libblkid.3.html
> you'll find
> | Block device information is normally kept in a cache file
> | /etc/blkid.tab and is verified to still be valid before being
> | returned to the user (if the user has read permission on the raw
> | block device, otherwise not).
> So, if you read carefully and think about it, this essentially says "if
> device information has changed since the last run of blkid as privileged
> user and you run blkid as user without access to the raw device, the
> information returned will be wrong".
> But that's not what I'd call extremely and absolutely obvious.
>
> Maybe blkid, when run under an unprivileged account, should display a
> warning and tell the user that the information returned wasn't verified.
>
>