Re: [Qemu-devel] [RFC PATCH v2 09/21] ppc/xive: extend the interrupt presenter model for XIVE

[David Gibson <david@gibson.dropbear.id.au>]

[-- Attachment #1: Type: text/plain, Size: 1611 bytes --]

On Tue, Sep 19, 2017 at 09:28:45PM +0200, Cédric Le Goater wrote:
> On 09/19/2017 09:36 AM, David Gibson wrote:
> > On Mon, Sep 11, 2017 at 07:12:23PM +0200, Cédric Le Goater wrote:
> >> The XIVE interrupt presenter exposes a set of Thread Interrupt
> >> Management Areas, also called rings, one per different level of
> >> privilege (four in all). This area is used to handle priority
> >> management and interrupt acknowledgment among other things.
> >>
> >> We extend the ICPState object with a cache of the register data for
> >> XIVE. The integration with the sPAPR machine is much easier and we
> >> need a common framework to switch from one controller model to
> >> another: XICS <-> XIVE.
> > 
> > This sounds like an even worse idea than referencing the ICS state.
> 
> ok ok.
> 
> > The TIMA really needs to be managed by a different object than the ICP.
> 
> like an array under the machine indexed by the cpu index ? 

Or individual TIMA objects which the cpus point to using their intc
pointers.

> at some point, we will need to :
> 
>     PowerPCCPU *cpu = POWERPC_CPU(current_cpu);
>     ICPState *icp = ICP(cpu->intc);
> 
> and 
> 
>     icp = xics_icp_get(xive->ics->xics, target);
> 
> 
> isn't the cpu->intc pointer  the best option to hold that information ? 
> and it is migrated.

No, it shouldn't be migrated.  It's set up during machine construction.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]