Re: [Qemu-devel] [PATCH v6 1/3] IOMMU: add option to enable VTD_CAP_CM to vIOMMU capility exposoed to guest

[Tian, Kevin]

> From: Michael S. Tsirkin
> Sent: Tuesday, November 22, 2016 11:10 PM
> 
> On Tue, Nov 22, 2016 at 12:11:21PM +0800, Jason Wang wrote:
> > Yes, that's the interesting point. The fault is not guaranteed but
> > conditional. And we have similar issue for IEC.
> >
> > So in conclusion (since I can't find a hardware IOMMU that have CM set):
> >
> > 1) If we don't cache fault conditions, we are still in question that whether
> > it was spec compatible.
> > 2) If we do cache fault conditions, we are 100% sure it was spec compatible.
> >
> > Consider 2) is not complicated, we'd better do it I believe?
> >
> > Thanks
> 
> IMO it's just a confusing jargon used by intel architects.
> CM is there exactly so you can shadow the tables, but
> they were trying hard to use wording that also makes
> sense to hardware/driver developers.
> 
> Nothing to worry about.
> 

Agree. Software shouldn't make any assumption that fault entries
will be cached when caching mode is enabled. It's worded just
because some very old generation did cache non-present/faulting
entries so IOMMU driver needs to be aware of such potential side
effect. while for virtualization the only purpose is to ask guest side 
to do IOTLB invalidation for any PTE change. It's completely fine 
to do simple thing here for vIOMMU here.

Thanks
Kevin