Re: [PATCH v7 02/10] x86/fred: Pass event data to the NMI entry point from KVM

[Sean Christopherson <seanjc@google.com>]

On Fri, Jun 20, 2025, H. Peter Anvin wrote:
> On 2025-06-19 15:57, Sohil Mehta wrote:
> > On 6/19/2025 3:45 PM, Xin Li wrote:
> > > On 6/19/2025 3:15 PM, Sohil Mehta wrote:
> > > > 
> > > > I want to say that the event data for IRQ has to be zero until the
> > > > architecture changes — Similar to the /* Reserved, must be 0 */ comment
> > > > in asm_fred_entry_from_kvm().
> > > > 
> > > 
> > > FRED spec says:
> > > 
> > > For any other event, the event data are not currently defined and will
> > > be zero until they are.
> > > 
> > > So "Event data not defined for IRQ thus 0."
> > 
> > I am fine with this. Not *defined* removes the ambiguity.
> > 
> 
> So I was thinking about this, and wonder: how expensive is it to get the
> event data exit information out of VMX? If it is not very expensive, it
> would arguably be a good thing to future-proof by fetching that information,
> even if it is currently always zero.

It's trivially easy to do in KVM, and the cost of the VMREAD should be less than
20 cycles.  So quite cheap in the grand scheme.  If VMREAD is more costly than
that, then we have bigger problems :-)