From: Allen Kistler on
Kurt wrote:
>> As I see you only have one key enabled in your luks setup
>> I wish you good luck with trying different types of keybs!
>> (or maybe trying harder to remember your password, it also happens!-)
>
> Just an idea - as it is impossible to brute-force the password as it only
> allows one try per second, is it then somehow possible to extract the key
> and try and bruteforce that... if I know the exact same way to find the
> crypted key? I guess that would go so much faster to compare those two
> crypted strings? Does that make sense at all?
>
> I _think: I know the most of my password, but the last digits I may think
> could be wrong so if I could try with maybe 90% of the password good, then
> perhaps I could get some luck with bruteforce?

It makes sense. Check out page 6 of

http://cryptsetup.googlecode.com/svn-history/r42/wiki/LUKS-standard/on-disk-format.pdf

or you could hack cryptsetup to remove the 1 sec delay. You'd have to
do that, anyway, even if you want to brute force an "offline" copy of
the headers. Either way, you can't lock yourself out by trying
infinitely many times.

BTW, one thing I didn't ask before ....

Since you've applied LUKS to the whole disk, rather than partitioning
the disk and applying LUKS to the partition (the usual thing), is it
possible you used fdisk on the raw disk? Or installed a boot loader on
the raw disk? Either one of those would have trashed the data structures.