Princeton security researchers purport successful attacks against
Microsoft Bitlocker, TrueCrypt, and other mainline disk encryption
products.

There is a video at the following URL that shows an actual attack, and
it looks to be trivial:

http://news.cnet.com/8301-1009_3-10003167-83.html

If you have a laptop and use one of these type encryption solutions,
you probably need to eyeball this one.

Maybe some closer scrutiny will show a method to prevent the attack.

Is secure to use RIPEMD-160 hash function? I know that there are not
known atacks about that algorithm, but I don't know if today is really
secure to use a hash of 160 bits.

Or if would be best to use a hash of 256 bits as minimum (as SHA-256)?
--althought has been created by N S A--

http://vault.telin.com/?dir=./public/Hash

sha256.exe

I'm going to use 95 UTF-8 characters (that occupies 2 bytes each one),
and I want to calculate its entropy for 80 bits --I think that is the
minimum recommend by NIST--.

------------------------
95 latin-1 chars. = 190 bytes

80 / lg(190 by) = 10 by (= 5 UTF-8 chars.)

I have a 32Bit function which calculates some kind of hash using three
secret 32 bit values. The question is how to reverse this or is it
possible to reverse (as i don´t know if this function is based on
something cryptographic algo)?

In my example I defined three SECRET values and generated 60000
plain/crypt pairs as a test system.
Bruteforce attack tooks very long time so how can i reverse such an
algorithm to get SECRET values from generated plain/crypt examples?

here´s a c-example of the algorithm:

unsigned long keycalc(unsigned long myseed)
{
unsigned long sc1,sc2,sc3;
unsigned long hlp;
unsigned short i,j;

sc1 = SECRETVALUE1;
sc2 = SECRETVALUE2;
sc3 = SECRETVALUE3;