[RFC/PATCH 14/16] MPIC MSI backend

[Segher Boessenkool]

> MPIC MSI backend. Based on code from Segher, heavily hacked by me.
> Renamed to mpic_htmsi, as it only deals with MSI over Hypertransport.

More exactly: it only deals with MSIs that are translated
by some whatever-to-HT bridge into a normal HT interrupt.

> We properly discover the HT magic address by reading the config space.

...config space of that bridge.

> Now we have an irq allocator we can support > 1 MSI, and we don't  
> reuse
> the LSI.

Right, that's what I asked about in the other thread, so
the FIXME in htmsi_alloc() can indeed go.

Why is MSI-X still unsupported?  Simply because of lack
of testing?  (See htmsi_check()).

Looks good,


Segher