Re: [PATCH] KVM: x86/mmu: Add capability to zap only sptes for the affected memslot

[Paolo Bonzini]

On 09/07/20 23:12, Sean Christopherson wrote:
>> It's bad that we have no clue what's causing the bad behavior, but I
>> don't think it's wise to have a bug that is known to happen when you
>> enable the capability. :/

(Note that this wasn't a NACK, though subtly so).

> I don't necessarily disagree, but at the same time it's entirely possible
> it's a Qemu bug.

No, it cannot be.  QEMU is not doing anything but
KVM_SET_USER_MEMORY_REGION, and it's doing that synchronously with
writes to the PCI configuration space BARs.

> Even if this is a kernel bug, I'm fairly confident at this point that it's
> not a KVM bug.  Or rather, if it's a KVM "bug", then there's a fundamental
> dependency in memslot management that needs to be rooted out and documented.

Heh, here my surmise is that  it cannot be anything but a KVM bug,
because  Memslots are not used by anything outside KVM...  But maybe I'm
missing something.

> And we're kind of in a catch-22; it'll be extremely difficult to narrow down
> exactly who is breaking what without being able to easily test the optimized
> zapping with other VMMs and/or setups.

I agree with this, and we could have a config symbol that depends on
BROKEN and enables it unconditionally.  However a capability is the
wrong tool.

Paolo