Re: [RFC PATCH 14/32] x86/fred: header file with FRED definitions

[H. Peter Anvin]

On December 22, 2022 5:03:57 AM PST, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>On Wed, Dec 21, 2022 at 02:58:06AM +0000, Li, Xin3 wrote:
>
>> > > +/* Flags above the CS selector (regs->csl) */
>> > > +#define FRED_CSL_ENABLE_NMI		_BITUL(28)
>> > > +#define FRED_CSL_ALLOW_SINGLE_STEP	_BITUL(25)
>> > > +#define FRED_CSL_INTERRUPT_SHADOW	_BITUL(24)
>> > 
>> > What's the state of IBT WAIT-FOR-ENDBR vs this? That really should also get a
>> > high CS bit.
>> 
>> FRED does provide more possibilities :)
>
>That's not an answer. IBT has a clear defect and FRED *should* fix it.

You are not wrong, of course. That being said, we have not wanted to hitch too many things to the FRED baseline, lest it ends up delayed for implementation/validation reasons. The important thing is that FRED *does* provide the mechanism for addressing that even if it does not make the first implementation.