Re: [PATCH] x86/microcode/AMD: Attempt applying on every logical thread

[Ashok Raj]

Hi Boris,

On Wed, Aug 17, 2022 at 08:13:15PM +0200, Borislav Petkov wrote:
> > Do we have a more scalable way to support it today?
> 
> You're not reading my mails. Lemme repeat: microcode loading is a

I do read them, but probably I'm not seeing your perspective. It's
unintentional. 

> dangerous business, especially the late thing. I'm certainly not going
> to expose that to people if there's no merit. The only merit for loading
> the same revision is for testing purposes.

For this specific patch in question, its not for testing though.. Its
required for functional purposes?

> 
> If you're about to test stuff, you can just as well patch the microcode
> loader to do what you want it to, like I just did.

I guess after this patch is merged, you would need no special patches out
of tree. True? I'm sorry if I missed something that is obvious.

apply_mirocode_amd() has no revid checks, and __apply_microcode_amd() has
no revid checks.. 

In effect you can test applying the same microcode over and over again
without having any special patches. 

I thought you could enforce revid only on the primary, and siblings you
can re-apply. 

Will that will satisfy the real need?


Cheers,
Ashok