Re: [patch V5 24/38] x86/entry: Provide IDTENTRY_SYSVEC

[Thomas Gleixner]

Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx> writes:

> On 5/12/20 5:01 PM, Thomas Gleixner wrote:
>> +
>> +/**
>> + * DEFINE_IDTENTRY_SYSVEC - Emit code for system vector IDT entry points
>> + * @func:	Function name of the entry point
>> + *
>> + * idtentry_enter/exit() and irq_enter/exit_rcu() are invoked before the
>> + * function body. KVM L1D flush request is set.
>
>
> This is used for entry points for Xen and hyperV as well. Even though
> it's harmless at the moment, do we still want to set this flag for non-KVM?

Right, it's pointless for !KVM, but we have set this unconditionally
since the l1tf mess was introduced. I'm not sure whether it's worth to
optimize the single store out.

Thanks,

        tglx