[PATCH 3/4] powerpc/mpic: Move internal interrupt source vector allocation to a separate function.

[Sethi Varun-B16395]

> -----Original Message-----
> From: Kumar Gala [mailto:galak at kernel.crashing.org]
> Sent: Tuesday, March 27, 2012 7:32 PM
> To: Sethi Varun-B16395
> Cc: Linuxppc-dev at lists.ozlabs.org
> Subject: Re: [PATCH 3/4] powerpc/mpic: Move internal interrupt source
> vector allocation to a separate function.
> 
> 
> On Mar 27, 2012, at 8:52 AM, Sethi Varun-B16395 wrote:
> 
> >
> >
> >> -----Original Message-----
> >> From: Kumar Gala [mailto:galak at kernel.crashing.org]
> >> Sent: Tuesday, March 27, 2012 6:55 PM
> >> To: Sethi Varun-B16395
> >> Cc: Linuxppc-dev at lists.ozlabs.org
> >> Subject: Re: [PATCH 3/4] powerpc/mpic: Move internal interrupt source
> >> vector allocation to a separate function.
> >>
> >>
> >> On Mar 27, 2012, at 7:16 AM, Varun Sethi wrote:
> >>
> >>> Allocate vector numbers for MPIC internal interrupt sources (IPIs
> >>> and
> >>> Timers) in a separate function.
> >>>
> >>
> >> Explain why you are making this change.
> >
> >
> > [Sethi Varun-B16395] With the current code it becomes fairly difficult
> > to add new internal interrupt sources. In my case I had to add 32
> > additional interrupt sources corresponding to the MPIC error
> > interrupts. It's more convenient doing the internal interrupt source
> allocation using a loop.
> 
> I think that is more due to how you added the MPIC error interrupts and
> issues w/that code.  If you are treating the MPIC error interrupts as a
> cascade than they should have a distinct linux IRQ space from the
> standard MPIC interrupts.  This is how the MSIs work (as an example).
In case of error interrupts we are depending on the mpic_host_maps
for mapping and interrupt specifier translations. There is no separate
initialization as in case of MSIs. That's the reason I am treating
error interrupts as mpic internal interrupt sources.

-Varun