Re: [tip: x86/urgent] x86/CPU/AMD: Clear RDRAND CPUID bit on AMD family 15h/16h

[H. Peter Anvin]

On 8/24/19 11:19 AM, Pavel Machek wrote:
> On Fri 2019-08-23 01:10:49, tip-bot2 for Tom Lendacky wrote:
>> The following commit has been merged into the x86/urgent branch of tip:
>>
>> Commit-ID:     c49a0a80137c7ca7d6ced4c812c9e07a949f6f24
>> Gitweb:        https://git.kernel.org/tip/c49a0a80137c7ca7d6ced4c812c9e07a949f6f24
>> Author:        Tom Lendacky <thomas.lendacky@xxxxxxx>
>> AuthorDate:    Mon, 19 Aug 2019 15:52:35 
>> Committer:     Borislav Petkov <bp@xxxxxxx>
>> CommitterDate: Mon, 19 Aug 2019 19:42:52 +02:00
>>
>> x86/CPU/AMD: Clear RDRAND CPUID bit on AMD family 15h/16h
>>
>> There have been reports of RDRAND issues after resuming from suspend on
>> some AMD family 15h and family 16h systems. This issue stems from a BIOS
>> not performing the proper steps during resume to ensure RDRAND continues
>> to function properly.
> 
> There are quite a few unanswered questions here.
> 
> a) Is there/should there be CVE for this?
> 
> b) Can we perform proper steps in kernel, thus making RDRAND usable
> even when BIOS is buggy?
> 

The kernel should at least be able to set its internal "CPUID" bit, visible
through /proc/cpuinfo.

	-hpa