Steven, at first this solution seemed to me pretty cool. As far as I understand seed will contain some value from stack. The only question is how probable is it, that seed is always initialized with a different value? Thanks, Ovanes On Thu, Feb 11, 2010 at 5:58 AM, Steven Watanabe wrote:

AMDG

uint64_t seed; memcpy(&seed, &randomBytes[0], sizeof(seed));

The memcpy should copy from randomBytes into seed. (Assuming I haven't forgotten the signature of memcpy.)

Seems, like I forgot the signature... Another way to transfer the bytes would be: union bytes2seed { uint64_t seed_; char bytes_[sizeof(uint64_t)]; }; bytes2seed converter; Now one can pass &converter.bytes_[0] to CryptGenRandom. This does not require an additional memcpy call. Ovanes.

Steven, I know at least 2 books which describe this possibility and there is nothing stated about undefined behavior. I just double checked the ISO 2003 C++ Standards and did not find anything which disallows the presented approach. Can you point me to the paragraph in the Standard? Thanks, Ovanes

...

Seems, like I forgot the signature... Another way to transfer the bytes would be:

union bytes2seed { uint64_t seed_; char bytes_[sizeof(uint64_t)]; };

bytes2seed converter;

Now one can pass &converter.bytes_[0] to CryptGenRandom. This does not require an additional memcpy call.

Casting through a union is undefined behavior. :(.

On Thu, Feb 11, 2010 at 7:39 PM, Steven Watanabe wrote:

The usual aliasing rules apply with only one exception (listed in 9.5):

[Note: one special guarantee is made in order to simplify the use of unions: If a POD-union contains several POD-structs that share a common initial sequence (9.2), and if an object of this POD-union type contains one of the POD-structs, it is permitted to inspect the common initial sequence of any of POD-struct members; see 9.2. ]

Yes I read the paragraph as well. But as far as I understand my use case is legal (assumed uint64_t is POD), which I think it is. Or am I mistaken? This is what 9.2 Clause 16 states (you have a snippet from 9.5 which just refers to that one): "If a POD-union contains two or more POD-structs that share a common initial sequence, and if the PODunion object currently contains one of these POD-structs, it is permitted to inspect the common initial part of any of them. Two POD-structs share a common initial sequence if corresponding members have layoutcompatible types (and, for bit-fields, the same widths) for a sequence of one or more initial members." I understand that that it is allowed to use union for byte conversions of POD types. Regards, Ovanes

Thank you Steven. While I do not have boost installed on this computer, the code below compiles and runs as I would expect. Upon each execution, the value of seed changes. I expect boost::mt19937 rng(seed); will work. And now I have a very good random seed thanks to your advice. Your friend, Ed #include <iostream> #include #include int main() { HCRYPTPROV hProvider = 0; BYTE randomBytes[8]; CryptAcquireContext(&hProvider, 0, 0, PROV_RSA_FULL, CRYPT_VERIFYCONTEXT); CryptGenRandom(hProvider, sizeof(randomBytes), randomBytes); CryptReleaseContext(hProvider, 0); uint64_t seed; memcpy(&seed, &randomBytes[0], sizeof(seed)); std::cout << seed << std::endl; return 0; } On Wed, Feb 10, 2010 at 11:58 PM, Steven Watanabe wrote:

AMDG

e m wrote:

...

I have a C++ program that uses /dev/urandom on Unix systems to seed boost::mt19937. I'm porting the program to Windows. I can get good, random seed data on Windows similar to random seed data produced from /dev/urandom on Unix systems like so:

HCRYPTPROV hProvider = 0; BYTE randomBytes[8]; CryptAcquireContext(&hProvider, 0, 0, PROV_RSA_FULL, CRYPT_VERIFYCONTEXT); CryptGenRandom(hProvider, sizeof(randomBytes), randomBytes); CryptReleaseContext(hProvider, 0);

However, I'm running into trouble attempting to pass these randomBytes to boost::mt19937. On Unix I do something similar to this:

uint64_t GetSeed() { uint64_t seed; std::ifstream urandom; urandom.open("/dev/urandom"); urandom.read(reinterpret_cast (&seed), sizeof(seed)); urandom.close(); return seed; }

boost::mt19937 rng(GetSeed());

That works OK on Unix I think. Could anyone suggest how to get the random data from the Windows CryptoAPI passed to boost::mt19937 as seed data? I don't normally program Windows systems, so this has been a bit of a challenge for me.

Try

uint64_t seed; memcpy(&seed, &randomBytes[0], sizeof(seed));

In Christ, Steven Watanabe

_______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users