rng-tools 6.7: tests/rngtesturandom.sh fails randomly in same conditions about rng-tools HOT 7 CLOSED

they are revealing that your urandom source isn't producing very random data. Transient failures like this are possible (and thats not expressly problematic), but it does suggest that the entropy pool in your host is running low. Are you running rngd in the host?

from rng-tools.

rngd running on the builder host doesn't address the issue.
The entropy pool fill level is almost stable while running tests in build environment:

 $ while :; do cat /proc/sys/kernel/random/entropy_avail; done
3387
3387
3387
3387
3387
3387
3387
3387
3387
3387
3387
3387
3387

but random test failures still persist.

from rng-tools.

you said you are running in a chroot..Are you also running in a VM?

from rng-tools.

I'm running both. My experiments (with above resulting logs) are being performed in a VM. But the reason to start the experiments was a failure of auto test during build on real HW.

from rng-tools.

so, to be clear, this test failure doesn't appear to be a failing in the rng-tools suite itself, but a detection of non-random data in your urandom pool. That said, urandom is meant to be a best effort operation as far as random data is concerned, meaning that occasional failures like this may occur. Perhaps it would be better to document those transient unexpected outcomes as XPASS results, meaning they pass, but for unexpected reasons. Do you think that would be a reasonable solution?

from rng-tools.

If there is nothing left to be done to make the auto tests' results more stable in my described environments, and you consider this is not a rng-tools codebase bug, then introducing some border "XPASS" state as a solution seems reasonable to me.
But I wonder, whether you are going to change auto tests code to output some additional result code to represent XPASS state? Should I expect that in future releases to change my spec file appropriately? Or it's up to me to decide how to treat random /dev/urandom test failures?
Thank you for your time to discuss the issue.

from rng-tools.

Closing as following commit introduces the agreed XPASS state:
ce96ccb
Thank you.

from rng-tools.