oss-sec mailing list archives

Re: Some AMD cpus with RDRAND fail to produce random numbers after suspend/resume

From: John Haxby <john.haxby () oracle com>
Date: Fri, 17 Jan 2020 14:50:33 +0000

On 17 Jan 2020, at 08:10, Sven Schwedas <sven.schwedas () tao at> wrote:

On 17.01.20 05:21, Jeffrey Walton wrote:

I agree with Lennart Poettering. This seems CVE worthy given RDRAND is
often used to get the kernel generator (and other userland generators)
in good working order.


From my understanding it's harmless as far as linux's kernel generator
is concerned, as it's just xor'd to other entropy sources?

CVEs should only be needed on a case-by-case basis for userland
generators that aren't properly engineered.


Actually, the kernel does use rdrand directly for cases where a strong CPRNG not required.  Whether some of those cases 
result in an exploitable bug I wouldn't like to say.

jch

Attachment: signature.asc
Description: Message signed with OpenPGP

Current thread:

Some AMD cpus with RDRAND fail to produce random numbers after suspend/resume Jeffrey Walton (Jan 16)
- Re: Some AMD cpus with RDRAND fail to produce random numbers after suspend/resume Sven Schwedas (Jan 17)
  - Re: Some AMD cpus with RDRAND fail to produce random numbers after suspend/resume John Haxby (Jan 17)
- Re: Some AMD cpus with RDRAND fail to produce random numbers after suspend/resume Peter Kjellström (Jan 20)