How can I tell if an UUID generated by QUuid::createUuid() is based on entrophy of cryptographic quality?

Question

From the Qt5 reference on QUuid:

QUuid QUuid::createUuid() [static]

On any platform other than Windows, this function returns a new UUID with variant QUuid::DCE and version QUuid::Random. If the /dev/urandom device exists, then the numbers used to construct the UUID will be of cryptographic quality, which will make the UUID unique. Otherwise, the numbers of the UUID will be obtained from the local pseudo-random number generator (qrand(), which is seeded by qsrand()) which is usually not of cryptograhic quality, which means that the UUID can't be guaranteed to be unique.

When on a platform with /dev/urandom, how can I tell if /dev/urandom in fact is available and will be used by this function when it runs?

I suspect it was not in my case, based on this little code that I ran mostly for curiosity's sake:

while(true){
        QUuid u=QUuid::createUuid();
        QString str=u.toString();
        if(str.contains("2222")){
            qDebug()<<"UUID:"<<str;
            return;
        }
    }

This produced the following output when I ran it repeatedly:

 1 UUID: "{8b42222b-bac3-4c93-b55d-0255a33115a2}"
 2 UUID: "{5122227e-473e-4885-8285-8780cc51f71a}"
 3 UUID: "{bfeed28f-f8da-4a69-9303-77388752222e}"
 4 UUID: "{0000beee-690c-4875-9589-9e222222cedc}"
 5 UUID: "{e5a2646d-1c81-4974-94ad-8b222265b67a}"
 6 UUID: "{bb77d756-6726-4e9f-94d8-3d4892222a2f}"

Notice the rather bizarre looking output on line #4, and the other apparent symmetries in the 3rd and 4th columns (except for the 4 in the start of column 3 which was to be expected).

Thanks!

Answer 1

Your output on line 4 is almost certainly just random chance. 6 UUIDs is not a large enough sample to provide any indication of randomness, if you really wanted to know for sure you would need to test a massive number of UUIDs (more than is realistically feasible).

Concerning column 3 and 4, as you've noted the leading "4" in the 3rd column is expected because it indicates the UUID version. The leading value in the 4th column (b 55d, 8 285, 9 303, ...) though is also expected, this value will always be 8, 9, a, or b. Outside of these values, I don't see any particular symmetries in the 3rd and 4th columns.

For checking if /dev/urandom or qrand() was used, the QUuid class does not make this information available, but if you really wanted to know you could always do what the QUuid class does and try to open and read data from /dev/urandom. I would argue that this is probably overkill though. While I certainly would not rely on qrand() in any situation where quality random numbers are needed for security, it's implementation (given how it's being seeded, ~line 959) is probably good enough for generating UUIDs.

Answer 2

In general you can try it on console this way

random="$(dd if=/dev/urandom bs=10 count=1)";
echo $random

If this gives you 10 characters, urandom is working. If everything is OK with your setup, Qt also will use it.

As far as I know, this is also just a psedo random number generator working with the entropy generated by the system. While /dev/random blocks when used entropy is too low, urandom continues to give you numbers.

So if you need to rely on absolute security, buy a real RNG. If you are OK with the "randomness" of /dev/random, use it. If you are not creating that many (millions maybe more) uuids and not running at system startup, urandom is fine but to use with caution.

Oh, why /dev/random and /dev/urandom is not a real RNG ?

Simply because you could reproduce its output by reproducing the situation. Well this will not be possible without exactly observing a system while its entropy is used for such fun... but in theory its possible. For real randomness you may want to have a device observing quantum states.