On Wed, Dec 07, 2022 at 10:35:59PM +0000, Edgecombe, Rick P wrote:

Yes, the suggestion was to have one for kernel and one for user. But I
was also thinking about how KVM could hypothetically support shadow
stack in guests in the non !CONFIG_X86_USER_SHADOW_STACK case (it only
needs CET_U xsave support). So that configuration wouldn't expose
user_shstk and since KVM's guest feature support is retrieved
programmatically, it could be nice to have some hint for KVM users that
they could try. Maybe it's simpler to just tie KVM and host support
together though. I'll remove "shstk".

Hmm, I don't have a clear idea how guest shstk support should do so
maybe this is all way off but yeah, if the host supports CET - the
*hardware* feature - then you can use the same logic to support that in
a VM.

I.e., if the guest sees CET - i.e., HV has advertized it - then guest
kernel behaves exactly the same as on the host.

But it is likely I'm missing something more involved...

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette