Over the years we have seen various workarounds like disabling RDSEED for select AMD CPUs due to hardware bugs and early on in the Zen days were also some RdRand issues due to different problems. It turns out the newest AMD EPYC 5th Gen “Turin” processors have a new RDSEED issue.
RDSEED is principally used for seeding software psuedo random number generators with additional entropy where not needing high quality RDRAND. Meta engineer Gregory Price announced on the Linux kernel mailing list the uncovering of an RDSEED bug affecting AMD EPYC 9005 “Turin” processors.
Gregory Price wrote:
“Under unknown architectural conditions, Zen5 chips running rdseed can produce (val=0,CF=1) as a “random” result over 10% of the time (when rdseed is successful). CF=1 indicates success, while val=0 is typically only produced when rdseed fails (CF=0).
This suggests there is an architectural issue which causes rdseed to misclassify a failure as a success under unknown conditions.
This was reproduced reliably by launching 2-threads per available core, 1-thread per for hamming on RDSEED, and 1-thread per core collectively eating and hammering on ~90% of memory.”
The proposed patch is disabling RDSEED usage on AMD Zen 5 Turin processors, similar to what the Linux kernel did for Cyan Skillfish APUs.
No AMD engineer has yet commented on this matter on the Linux kernel mailing list.
