Re: [dm-crypt] request for new API to found out is a volume is a luks volume or a tcrypt volume

[Milan Broz <gmazyland@gmail.com>]

On 12/07/2012 10:50 PM, .. ink .. wrote:
 
> finding out if a device is luks device is a bit clunky at the moment
> because a user will have to go through the crypt_init(),crypt_load()
> and finally crypt_free() to see if a device is luks device.

yes, this is how it is designed.

> Just added support for opening truecrypt volumes and the way to check
> if a volume is a true crypt volume seem to be clunkier since it
> involves doing
> crypt_init(),crypt_load(),crypt_activate_by_volume_key() and if on
> success, crypt_deavicate() and then finally crypt_free()

No. crypt_load is enough, exactly the same logic as in LUKS
(just need to provide password/keyfiles).
See action_isLuks and action_tcryptDump (in cryptsetup.c).

truecrypt-compatible  header support it is not yet stable API,
I am still playing with it and it can still change.
(but you already found it so thanks for testing ;-)

(And for curious - only activation will supported, no header
manipulation. I will write more later on pre-release time,
it will need a lot of testing.)

> It will be nice if these operations were simplified with something
> like:
> 
> crypt_is_luks( const char* device )
> crypt_is_tcrypt( const char* device,const char* key,size_t key_len )

Is it really problem to do 3 simple steps? (init, load, free)

Milan