RDRAND on AMD Ryzen 9 5900X is flakey

Re:RDRAND on AMD Ryzen 9 5900X is flakey by tadfisher on Monday January 11, 2021 @02:58PM (#25735506)

It's not systemd's job to fix hardware bugs, it's the kernel's (or AMD's). The fix for the last time this happened was to mask the offending CPU IDs in the kernel's RNG from using RDRAND. systemd did have a temporary hack in place for their direct RDRAND implementation, but this current bug isn't the same (not just emitting 0 like last time, but eventually emitting duplicate values in pairs).

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by ginko on Monday January 11, 2021 @03:19PM (#25735945)

Even so, systemd should be able to deal with name collisions even with numbers coming from a true RNG. That causing errors is a design issue, IMO.

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by tadfisher on Monday January 11, 2021 @03:40PM (#25736390)

That would be a non-deterministic failure case scaling to O(n) in both space and time. But it's not just UUIDs, it's also required for seeding systemd's internal hash tables, which would degrade from O(1) lookups to O(n) should an attacker exploit known RNG flaws.

So, a PRNG with a semi-decent (not perfect) entropy pool is required at boot time, and systemd needs to run in environments where seeding that pool in software could take on the order of minutes. This is why RDRAND is used to seed the pool during boot.

https://github.com/systemd/systemd/blob/61bd7d1ed595a98e5fbf...

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by aidenn0 on Monday January 11, 2021 @05:22PM (#25738335)

1. UUIDs are stored somewhere so there's already O(n) space in use. Checking for collisions is only O(log n) in time.

2. Use a tree instead of a hash-table; or just use RDRAND and hope you aren't presented with malicious inputs.

The comment in the code[1] about RDRAND being sufficient for the use-case of UUIDs is empirically false. To continue to rely upon it is to insist that the rest of the world conform to your beliefs, rather than to adjust your beliefs to match the rest of the world.

1: https://github.com/systemd/systemd/blob/bcac754d66374782a85a...

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by tadfisher on Monday January 11, 2021 @05:48PM (#25738817)

> The comment in the code[1] about RDRAND being sufficient for the use-case of UUIDs is empirically false.

Ipse dixit. Can you provide an explanation?

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by aidenn0 on Monday January 11, 2021 @05:53PM (#25738935)

1. TFA

2. https://arstechnica.com/gadgets/2019/10/how-a-months-old-amd...

3. https://linuxreviews.org/RDRAND_stops_returning_random_value...

4. There are also comments elsewhere on this page that older Intel RDRAND implementations had similar issues, but they did not cite their source.

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by johncolanduoni on Monday January 11, 2021 @03:38PM (#25736351)

Most cryptographic constructs you feed CSPRNG data into break spectacularly if they receive colliding data so I guess they all have design issues too. Including the TLS connection you’re typing your comment through.

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by tremon on Monday January 11, 2021 @05:36PM (#25738590)

But it becomes systemd's job to fix/compensate for hardware bugs if they explicitly access the hardware directly instead of going through the kernel, no?

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by LukeShu on Monday January 11, 2021 @09:33PM (#25741518)

They do go through the kernel; it asks the kernel "can I use rdrand", and the kernel says "yep, go ahead, it works on this CPU". If that's not true, and it doesn't work on the CPU, then it's the kernel's fault for lying and saying that it does.

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by klmadfejno on Monday January 11, 2021 @03:04PM (#25735617)

That doesn't seem to answer the question of why use it

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by numpad0 on Monday January 11, 2021 @03:23PM (#25736024)

It generally makes more sense to trust CPU specifications and utilize it fully, than to assume everything is tampered with and only ever use `mov`

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by tadfisher on Monday January 11, 2021 @03:22PM (#25735995)

If they don't use it, the kernel does in order to initialize the entropy pool, which will also result in broken boots. To work around historically flaky kernel implementations, systemd calls RDRAND directly if it detects it needs to with a heuristic. This is why the fix for the current issue is to set both 'nordrand' and 'SYSTEMD_RDRAND=0'.

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by tedunangst on Monday January 11, 2021 @04:43PM (#25737696)

What happened to it's not systemd's job to fix kernel bugs?

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by tadfisher on Monday January 11, 2021 @05:36PM (#25738585)

I didn't type that, so I'm not sure what you're referring to. systemd is fundamentally tied to Linux and thus needs to implement workarounds from time-to-time when kernel facilities are lacking. This is complicated by the requirement that systemd continue working on older Linux versions.

[ Parent | on HN ]

Re:RDRAND on AMD Ryzen 9 5900X is flakey by effie on Monday January 11, 2021 @06:21PM (#25739495)

I think it is systemd's job to make sure it works for its users, even though that requires working around HW bugs. If RDRAND is so problematic and can be not used, why use it at all? Just abstain from using it, simplify the logic and solve this class of issues once and for all.

[ Parent | on HN ]