The rng-tools is a set of utilities related to random number generation in kernel. The main program is rngd, a daemon developed to check and feed random data from hardware device to kernel entropy pool.
This is mainly useful to increase the quantity of entropy in kernel to make
/dev/random faster. By default,
/dev/random is very slow since it only collects entropy from device drivers and other (slow) sources. rngd allows the use of faster entropy sources, mainly hardware random number generators (TRNG), present in modern hardware like recent AMD/Intel processors, Via Nano or even Raspberry Pi.
While Linux itself uses the result from TRNG in
/dev/random, if available, they are only used as a XOR after the entropy is collected by kernel. So
/dev/random, by default, is slow even if you do have a TRNG. rngd feeds
/dev/random itself, increasing the available entropy by far.
The configuration file is located in
/etc/conf.d/rngd. There is only one option though, that is
RNGD_OPTS, the parameters to be passed to the daemon when running it with the included
rngd.service. The default parameter (
"", or blank) should work in the majority of cases.
By default, rngd will try to automatically detect your TRNG and use it. This is reported to work for Raspberry Pi and Intel Ivy Bridge CPU using the lastest versions of rng-tools. If this does not work, you may manually pass the device file used by your TRNG, as in the below example:
RNGD_OPTS="-o /dev/random -r /dev/my_hw_random_device"
If your system does not have a TPM module, you may pass
RNGD_OPTS to suppress the following warning message from log:
Unable to open file: /dev/tpm0
By default rngd fills the entropy pool until at least 2048 bits of entropy are available. This is to avoid the TRNG to dominate the contents of the pool. You can override this setting if you really trust your TRNG. To do this, pass
RNGD_OPTS, for example (4096 is the maximum size of kernel's entropy pool by default, you shouldn't pass a value greater than the maximum either). Doing so may increase the performance of
/dev/random even further, at the expense of maybe lower random number quality. However, it should be noted that the default setting is already sufficient for the majority of user cases.
Testing and usage
You may test if rngd is working before enabling its service by running:
# rngd -f
A simple test to see if everything is working as it should is to run (in another terminal) the following command:
$ dd if=/dev/random of=/dev/null bs=1024 count=1 iflag=fullblock
Without rngd, the above command will take lots of time to run. With rngd working properly, the result should be almost instantaneous:
1+0 records in 1+0 records out 1024 bytes (1.0 kB) copied, 0.0199623 s, 51.3 kB/s
A speed of around 50 kB/s in dd's output shows that everything is working properly. For comparison, without rngd you probably would get 0.0 kB/s (since the speed is too low).
Another interesting test is to run rngtest, to check the randomness of data using FIPS 140-2 tests:
$ cat /dev/random | rngtest -c 1000
rngtest 5 Copyright (c) 2004 by Henrique de Moraes Holschuh This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. rngtest: starting FIPS tests... rngtest: bits received from input: 20000032 rngtest: FIPS 140-2 successes: 999 rngtest: FIPS 140-2 failures: 1 rngtest: FIPS 140-2(2001-10-10) Monobit: 1 rngtest: FIPS 140-2(2001-10-10) Poker: 0 rngtest: FIPS 140-2(2001-10-10) Runs: 0 rngtest: FIPS 140-2(2001-10-10) Long run: 0 rngtest: FIPS 140-2(2001-10-10) Continuous run: 0 rngtest: input channel speed: (min=301.394; avg=417.091; max=693.187)Kibits/s rngtest: FIPS tests speed: (min=64.656; avg=91.010; max=123.055)Mibits/s rngtest: Program run time: 47037492 microseconds
It is normal for any random number generator to fail in a small number of tests in 1000 passes, however if the number of failures is too great (like 10), probably there is something wrong.