Subject: L2x0 OF properties do not include interrupt #

Will Deacon will.deacon at arm.com
Wed Aug 10 10:39:19 EDT 2011


On Wed, Aug 10, 2011 at 03:37:26PM +0100, Rob Herring wrote:
> On 08/10/2011 09:10 AM, Will Deacon wrote:
> > Hi Rob,
> > 
> > On Wed, Aug 10, 2011 at 02:59:12PM +0100, Rob Herring wrote:
> >> I think you should allow for either the single irq or individual irqs.
> >> You can specify that the event counter interrupt must be first, then the
> >> pmu driver could work either way ignoring the rest. The driver probably
> >> needs to mark the handler as shared if there is only the combined
> >> interrupt unless you expect all interrupts to be handled by 1 driver.
> > 
> > I much prefer having seperate, individual IRQs with no requirement on
> > ordering.
> > 
> > Now, the L2 binding also doesn't fit too well for the L2CC on Cortex-A15,
> > which is an inner cache like the one on Cortex-A8. Because of this, it
> > doesn't have a base address but it *does* have an IRQ which is how external
> > aborts are raised.
> 
> This is not a general L2 binding, but an L2x0/PL310 binding. A8/A15 L2
> is a completely different binding and driver though. You would do
> something like the current cpu pmu binding that is just interrupts.

Ok, I think that's a lot more sense to have a separate binding for inner and
outer L2 cache implementstions. I was just a bit concerned by the
documentation using the term 'l2cc' as it sounds like it's fairly catch-all.

Will



More information about the linux-arm-kernel mailing list