[pkg-cryptsetup-devel] Bug#923513: Bug#923513: cryptsetup-bin: Can no longer luksFormat as non-root: "Not compatible PBKDF options."

[Guilhem Moulin]

Control: found -1 2:2.1.0-1

Hi Christoph,

On Fri, 01 Mar 2019 at 11:09:53 +0100, Christoph Biedl wrote:
> Declare usage of format 1 like in
> 
>   $ echo -n foo | cryptsetup luksFormat --type luks1 /tmp/blob -
> 
> and possibly some other ways.

FWIW the regression isn't directly tied to the new default LUKS format,
but to the PBKDF benchmark for Argon2i/id (the default PBKDF for LUKS2).
Here is an output snippet obtained with --debug:

    # Found area 32768 -> 290816
    # Running argon2i() benchmark.
    # PBKDF benchmark: memory cost = 32, iterations = 4, threads = 4 (took 4 ms)
    # PBKDF benchmark: memory cost = 512, iterations = 4, threads = 4 (took 1 ms)
    # PBKDF benchmark: memory cost = 8192, iterations = 4, threads = 4 (took 10 ms)
    Not compatible PBKDF options.
    […]
    Command failed with code -3 (out of memory).

and now as root:

    # Running argon2i() benchmark.
    # PBKDF benchmark: memory cost = 32, iterations = 4, threads = 4 (took 3 ms)
    # PBKDF benchmark: memory cost = 512, iterations = 4, threads = 4 (took 1 ms)
    # PBKDF benchmark: memory cost = 8192, iterations = 4, threads = 4 (took 7 ms)
    # PBKDF benchmark: memory cost = 131072, iterations = 4, threads = 4 (took 152 ms)
    # PBKDF benchmark: memory cost = 215578, iterations = 4, threads = 4 (took 256 ms)
    # PBKDF benchmark: memory cost = 1048576, iterations = 6, threads = 4 (took 1920 ms)
    # Benchmark returns argon2i() 6 iterations, 1048576 memory, 4 threads (for 512-bits key).
    # Calculating attributes for LUKS2 keyslot 0.
    […]
    Command successful.

Hence other possible workarounds include using PBKDF2 (default PBKDF for
LUKS1) 

    cryptsetup luksFormat --pbkdf pbkdf2

Alternatively, you can pass --pbkdf-memory $KBYTES with a low enough
value so the benchmark doesn't exceed getrlimit(RLIMIT_MEMLOCK,), to
avoid the failing large mmap(2).  Milan, perhaps the benchmark should
call getrlimit(RLIMIT_MEMLOCK,) and automatically reduce --pbkdf-memory
accordingly?

Anyway, IMHO in test scenari it makes sense to the PBKDF benchmark and
instead force parameters directly (via --pbkdf-force-iterations and
--pbkdf-memory) to speed up luksFormat and luksOpen.

Cheers,
-- 
Guilhem.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/pkg-cryptsetup-devel/attachments/20190301/5438c712/attachment.sig>