On 09/03/2014 04:53 PM, Andrew Bresticker wrote:
On Tue, Sep 2, 2014 at 5:50 PM, David Daney <ddaney.cavm@xxxxxxxxx> wrote:
[...]
Your comments don't really make sense to me in the context of my knowledge
of the GIC.
Of course all the CP0 timer and performance counter interrupts are per-CPU
and routed directly to the corresponding CP0_Cause[IP7..IP2] bits. We are
don't need to give them further consideration.
Here is the scenario you should consider:
o 16 CPU cores.
o 1 GIC routing interrupts from external sources to the 16 CPUs.
o 2 network controllers each with an interrupt line routed to the GIC.
Q: What would the GIC "interrupts" property look like?
Note that the GIC doesn't have a single "interrupt-parent", as it can route
interrupts to *all* 16 CPUs.
I propose that the GIC have neither an "interrupt-parent", nor "interrupts".
The fact that it is an "mti,global-interrupt-controller", means that the
software drivers for the GIC already know how to route interrupts, and any
information the device tree could contain is redundant.
Ok, I misunderstood your opposition to the binding.
My intention for the "interrupt-parent" and "interrupts" property of
the GIC was to express that GIC interrupts are routed to the CPU
interrupt vectors and that a certain set of these vectors are
available for use by the GIC. I would agree that these are mostly
redundant (obviously the GIC routes interrupts to CPU interrupt
vecotrs) and that it is not the most accurate description of the
GIC-CPU relationship (the CPU interrupt controller are per-CPU, not
global, and the GIC can route interrupts to any of them), though I'm
not sure that there's a better way of describing it in DT.
So that leaves us with something like this:
interrupt-controller@1bdc0000 {
compatible = "mti,global-interrupt-controller";
interrupt-controller;
#interrupt-cells = <2>;
available-cpu-vectors = <2>, <3>, ...
Exactly what I had in mind, except for the missing "reg" property.
This gives software the information it needs, but doesn't impose any policy.
I will defer to others on the exact name the "available-cpu-vectors"
should have.
};
DT folks, thoughts?
