CPU random: ubusd: uninitialized urandom read (4 bytes read)

For Developers
1

why uninitialized urandom read?

[    6.247708] urandom-seed: Seeding with /etc/urandom.seed
[    6.271517] procd: - early -
[    6.916134] procd: - ubus -
[    6.925933] urandom_read: 4 callbacks suppressed
[    6.925937] random: ubusd: uninitialized urandom read (4 bytes read)
[    6.972441] random: ubusd: uninitialized urandom read (4 bytes read)
[    6.981194] random: ubusd: uninitialized urandom read (4 bytes read)
[    6.990533] procd: - init -
.....
[   99.472205] random: crng init done
2

The answer is contained in the log data you provided. It seems your embedded device was only powered on about 6 seconds. That doesn't seem to be long enough to initialize the device.

It seems it takes ~99 seconds to initialize your cryptographic random number generator (CRNG).

3

linux kernel DRNGļ¼ˆDigital Random Number Generatorļ¼‰
@ drivers/char/random.c
extract_entropy()
for (i = 0; i < LONGS(EXTRACT_SIZE); i++) {
unsigned long v;
if (!arch_get_random_long(&v))
break;
hash.l[i] ^= v;
}

4

Can you explain why you posted that code?

5

These messages are harmless so long as the reads from urandom are not being used to generate long lived cryptographic secrets, such as the ssh host key or a diffie-hellman modulus. Probably they're being used to do something like key a hash table or the like. I'm not sure what ubusd is doing with urandom but I doubt it's generating crypto keys.

1 Like
6

Found this on a search along with https://github.com/systemd/systemd/issues/4167
dlakelan you seemed to imply there that crypto should use getrandom() but according to https://www.2uo.de/myths-about-urandom/ using urandom is fine.

/dev/urandom is insecure. Always use /dev/random for cryptographic purposes.
Fact: /dev/urandom is the preferred source of cryptographic randomness on UNIX-like systems.
I don't fully understand this just curious if you disagree with that.