Re: [PATCH/proposal] dm-crypt: add digest-based iv generation mode

[Christophe Saout <christophe@saout.de>]

James,

> The crypto_hmac_init part could be done in the dm target constructor
> once and then call crypto_hmac_update and crypto_hmac_final in the
> iv_generator but this would require some cryptoapi hacking which I'm
> not too happy about. I would like to make a copy of the tfm (including
> the private context) on the stack and then operate on that copy for
> crypto_hmac_update and crypto_hmac_final.
> 
> Can I have a function to return the tfm size so I can do a memcpy
> to a variable on the stack? I could get rid of the mutex too then.

What do you think of this?

I would like to copy the tfm onto the stack so that I can
a) compute the hmac on several CPUs at the same time without locking
b) reuse a precomputed tfm from just after crypt_hmac_init