Re: [PATCH v2] [POWERPC] pasemi: Implement MSI support

[Michael Ellerman <michael@ellerman.id.au>]

[-- Attachment #1: Type: text/plain, Size: 1444 bytes --]

On Wed, 2007-12-12 at 01:23 -0600, Olof Johansson wrote:
> On Wed, Dec 12, 2007 at 05:49:39PM +1100, Michael Ellerman wrote:
> > Before the ugliness spreads .. I'm pretty sure we can get this from the
> > chip_data of the virq - I haven't fixed it though because I wasn't sure
> > if using chip_data was kosher or not.
> 
> Hmm, maybe.
> 
> I can forsee MSI and legacy interrupts being handled by different irq
> hosts on systems where they are cascaded, so I'm not 100% sure that's a
> workable way to do it (I assume you'd look it up through the pci_dev's
> legacy irq field, right?).

No you're right that won't work, what I had in mind would work for
teardown, we could get the mpic from the chip_data of each virq.

But for setup we need to know which mpic to create the mapping on
_before_ it's mapped (obviously) so using chip_data doesn't work. And
yeah, guessing based on the legacy irq of the pci_dev would not be safe.

The only safe way to do it I think would be to walk up the interrupt
tree and find the interrupt controller node, then do an irq_find_host(),
but that's a lot of fuss to save one static variable :)

Patch looks good otherwise.

cheers

-- 
Michael Ellerman
OzLabs, IBM Australia Development Lab

wwweb: http://michael.ellerman.id.au
phone: +61 2 6212 1183 (tie line 70 21183)

We do not inherit the earth from our ancestors,
we borrow it from our children. - S.M.A.R.T Person

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]