Re: mcrypt_create_iv - why so slow?

Interesting, using MCRYPT_DEV_URANDOM instead of MCRYPT_DEV_RANDOM seems
practically instantaneous.

Another less elegant solution I've found is to simply str_pad to the length
returned by mcrypt_get_iv_size.

Still begs the question though, any idea what's holding up the show w/
MCRYPT_DEV_RANDOM?  #morbidcuriosity

-nathan


On Fri, May 31, 2013 at 12:40 AM, Nathan Nobbe <[email protected]>wrote:

> Hi folks,
>
> This code:
>
> <?php
> $iv = mcrypt_create_iv(mcrypt_get_iv_size(MCRYPT_RIJNDAEL_256,
> MCRYPT_MODE_CBC), MCRYPT_DEV_RANDOM);
> var_dump($iv);
>
> Takes just over a minute to run on my laptop and roughly 45 seconds on a
> capable server, any idea why?
>
> time php test-iv.php
> string(32) "????'???H??y?PJ?U?1O;6???ѧ"
>
> real 0m44.917s
> user 0m0.024s
> sys 0m0.036s
>
> Also, I've noticed the mcrypt_encypt & mcrypt_decrypt complain with
>
> The IV parameter must be as long as the blocksize
>
> when not using mcrypt_create_iv, however, if the value of the IV parameter
> is consistent in both calls, the decryption seems to succeed despite the
> warning.
>
> So wondering:
> * can the call to mcrypt_create_iv be sped up
> * is there an alternative (faster) way to create a proper iv
> * how big a risk is it to 'ride dirty' here and not use mcrypt_create_iv
>
> thanks,
>
> -nathan
>