[dm-crypt] inconsistent behavior btw a binary and library call when closing truecrypt volume made up of multiple algorithms

Milan Broz gmazyland at gmail.com
Sun Mar 17 19:56:40 CET 2013


On 19.2.2013 12:37, Milan Broz wrote:
> On 02/18/2013 05:58 AM, .. ink .. wrote:
>>
>> test:
>>
>> 1. create a truecrypt volume using multiple algorithms,i went with AES-twofist-serpent
>> 2. open the volume using cryptsetup with "cryptsetup tcryptOpen ttt ttt"
>> 3. ls -l /dev/mapper/ will give:
>>
>> crw------- 1 root root 10, 236 Feb 17 22:37 /dev/mapper/control
>> lrwxrwxrwx 1 root root       7 Feb 17 23:45 /dev/mapper/ttt -> ../dm-2
>> lrwxrwxrwx 1 root root       7 Feb 17 23:45 /dev/mapper/ttt_1 -> ../dm-1
>> lrwxrwxrwx 1 root root       7 Feb 17 23:45 /dev/mapper/ttt_2 -> ../dm-0
>>
>> It looks like 3 mapper will be created,each per algorithm.
>>
>> closing the mapper with " cryptsetup tcryptClose ttt" will remove all three mappers expected.
>>
>> closing the mapper through the library "crypt_deactivate( NULL,"ttt"
>> )" will not remove all 3,just the "ttt" one.With the library call,
>> all 3 seem to require to be removed individually.>
>> Is this behavior of the library expected?
>
> Kind of... if NULL is in context. If you use context (init_by_name) it should work.

Fixed in commit
http://code.google.com/p/cryptsetup/source/detail?r=f2521889c2315b8f2bd812a92fa37057fc382877#

Thanks,
Milan


More information about the dm-crypt mailing list