RE: [RFC] x86/mce: Add workaround for SKX/CLX/CPX spurious machine checks

[Luck, Tony]

> In that case, you can just as well test the MSR bit directly
> MSR_IA32_MISC_ENABLE_FAST_STRING_BIT. If it set, you clear it, done.

Yes. That would work. It's an extra MSR read instead of a memory read. But this
isn't a performance path.

>> Maybe this would be more human friendly?
>> 
>> 		pr_err("CPU%d: Performance now degraded after applying machine check workaround\n",
>> 			smp_processor_id());
>
> Well, is there an erratum you can refer to in it instead?

The erratum has made its way through to the public specification update yet :-(

> Explaining the whole deal in a single error message is hard and almost
> certainly insufficient.

Not ideal, but the message is a search tool to get to these e-mail discussions.

> Also, what's the use of that message issuing once on every CPU? Instead
> of being a _once() message?

pr_err_once() would be better.

-Tony