[PATCH RFC/RFT] vfio/pci: Create feature to disable MSI virtualization

[Jason Gunthorpe]

On Tue, Aug 13, 2024 at 07:30:41PM +0200, Thomas Gleixner wrote:
> > Howver, it will still not work in a VM. Making IMS and non-MSI
> > interrupt controlers work within VMs is still something that needs to
> > be done.
> 
> Sure, but we really want to do that in a generic way and not based on ad
> hoc workarounds.
>
> Did the debate around this go anywhere?

No, it got stuck on the impossible situation that there is no existing
way for the VM to have any idea if IMS will work or is broken. Recall
Intel was planning to "solve" this by sticking a DVSEC in their
virtual config space that said to turn off IMS :\

So using IMS in the real world looked impractical and interest faded a
bit.

But the underlying reasons for IMS haven't gone away and more work is
coming that will bring it up again...

Jason