Re: [patch 21/32] NTB/msi: Convert to msi_on_each_desc()

[Logan Gunthorpe]

On 2021-11-29 5:29 p.m., Thomas Gleixner wrote:
> At the conceptual level these interrupts are in separate irq domains:
> 
>      |   _______________________
>      |   |                      |
>      |   | NTB                  |
>      |   |                      |
>      |   | PCI config space     |
>      |   |     MSI-X space      | <- #1 Global or per IOMMU zone PCI/MSI domain
>      |   |_____________________ |
>      |---|                      |
>          | Memory window A      |
>          | Memory window ..     | <- #2 Per device NTB domain
>          | Memory window N      |
>          |______________________|
> 
> You surely can see the disctinction between #1 and #2, right?

I wouldn't say that's entirely obvious or even irrefutable. However, I'm
certainly open to this approach if it improves the code.

> I'm way too tired to come up with a proper solution for that, but that
> PCI_IRQ_VIRTUAL has to die ASAP.

I'm willing to volunteer a bit of my time to clean this up, but I'd need
a bit more direction on what a proper solution would look like. The MSI
domain code is far from well documented nor is it easy to understand.

Logan