Michael Schmidt via RT wrote:
> Richard Levitte via RT schrieb:
>
>>Well, that depends on what you mean with number of bits. Take a number
>>like 0x0432, how many bits does it have? You could view it as a word,
>>and say it has 16 bits, or you could look at it more closely, and find
>>out that it has 11 *significant* bits.
>>
>>BN_num_bits() counts significant bits.
>
>
> OK, now I get it...
>
>
>
>>Another way to find this out is to add the following line to your
>>program:
>>
>> printf ("The number: "); BN_print_fp(stdout, pPubKey); printf ("\n");
>>
>>Looking at the output, I counted 511 hex characters. 511 * 4 = 2044.
>>
>>I don't think this is a bug. If you want to know the size of pPubKey in
>>byte increments, use BN_num_bytes() and multiply with 8.
>
>
> I'd suggest to clarify the man page of 'BN_num_bits':
> People (such as me) are tempted to use 'BN_num_bits' to get the key size
> (class) of a private or public key, and expect that its size matches the
> size (class) of its counterpart (public or private key). With "size
> class" I mean 512, 1024, 2048 bit etc.
It was not quite clear exactly what part of public key was evaluated
with BN_num_bits().
It's often spoken of public key size to be number of bits of modulus
for RSA and Diffie-Hellman schemes.
The best chance to get some variable number of bits is to evaluate
public key (being generator power private key calculated with modular
arithmetics). With private key often being random and public key being
unpredictable less than modulus one could get 2047 easily and 2044 with
some probability. Yes, DH parameters (modulus and generator) could be
the same (resulting in the same size 2048) for quite alotof keys
> Now what I experience is that the size of the public key that was
> generated from a given private key (e.g. with 'DH_generate_key') does
> not always match the size of the private key, which is confusing...
Hope it's not anymore
Regards,
Vadim
>
>
> Michael
>
>
>>[EMAIL PROTECTED] - Thu Jul 1 10:03:21 2004]:
>>
>>
>>
>>>Hi,
>>>
>>>I think I've encountered a problem with BN_num_bits. I have
>>
>>experienced
>>
>>
>>>that BN_num_bits often returns 1 (sometimes more) bits less than the
>>>actual key size of a BIGNUM. With 2048 bit DH key pairs, I've often
>>
>>seen
>>
>>
>>>2047 bit returned by BN_num_bits (checking the public key). I'm seeing
>>
>>a
>>
>>
>>>case now where it returns 2044 bit only.
>>>
>>>My environment:
>>>OpenSSL 0.9.7d
>>>SuSE Linux 9.0
>>>compiled with gcc 3.3.1
>>>
>>>I have attached a test case (actually more or less a fragment of an
>>>application of mine) which shows the 2044 bit case. Also, the output
>>
>>of
>>
>>
>>>'make report' is attached.
>>>
>>>Please e-mail me directly in case of any questions or comments, since
>>>I'm not subscribed to any openssl-related mailing lists.
>>>
>>>
>>>Michael
>>>
>>
>
>
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
