On 2020-05-06 09:45, André Przywara wrote:
On 05/05/2020 19:25, Marc Zyngier wrote:
On Tue, 05 May 2020 17:52:03 +0100,
Andre Przywara <andre.przywara@xxxxxxx> wrote:
The GIC DT binding only allows a certain combination of DT compatible
strings, mostly just consisting of one name.
Drop the combination of multiple names and go with the
"arm,cortex-a15-gic" name for GICv2, as this seems to be the most
widely
accepted string. "arm,gic-400" would be more correct, but was
introduced
much later into the kernel's GIC driver.
Signed-off-by: Andre Przywara <andre.przywara@xxxxxxx>
---
arch/arm64/boot/dts/arm/foundation-v8-gicv2.dtsi | 2 +-
arch/arm64/boot/dts/arm/juno-base.dtsi | 2 +-
arch/arm64/boot/dts/arm/rtsm_ve-aemv8a.dts | 2 +-
3 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/arch/arm64/boot/dts/arm/foundation-v8-gicv2.dtsi
b/arch/arm64/boot/dts/arm/foundation-v8-gicv2.dtsi
index 15fe81738e94..61a1750fcdd6 100644
--- a/arch/arm64/boot/dts/arm/foundation-v8-gicv2.dtsi
+++ b/arch/arm64/boot/dts/arm/foundation-v8-gicv2.dtsi
@@ -6,7 +6,7 @@
/ {
gic: interrupt-controller@2c001000 {
- compatible = "arm,cortex-a15-gic", "arm,cortex-a9-gic";
+ compatible = "arm,cortex-a15-gic";
#interrupt-cells = <3>;
#address-cells = <2>;
interrupt-controller;
diff --git a/arch/arm64/boot/dts/arm/juno-base.dtsi
b/arch/arm64/boot/dts/arm/juno-base.dtsi
index 3feefd61eb76..62392ab1f880 100644
--- a/arch/arm64/boot/dts/arm/juno-base.dtsi
+++ b/arch/arm64/boot/dts/arm/juno-base.dtsi
@@ -69,7 +69,7 @@
};
gic: interrupt-controller@2c010000 {
- compatible = "arm,gic-400", "arm,cortex-a15-gic";
+ compatible = "arm,cortex-a15-gic";
Why? GIC-400 is definitely the most correct compatible string. I'd
rather see this compatible being generalised to the models rather than
only referencing the A15 GIC.
I agree that gic-400 is the far better name, but it was only introduced
in v3.16. So omitting arm,cortex-a15-gic would break any kernels before
that, which I would like to avoid.
I am not talking about dropping the A15 GIC. I'm saying that both should
stay. Is there anything in the DT binding that forbids multiple names in
the compatible property?
It's actually a pity that we are so picky about the compatible
listings,
because the existing combination is actually quite nice: we get
compatibility with older DT consumers, but still can say what it
actually is.
I wonder if I should introduce this combination to the GIC DT binding
instead, it seems like there are other users in the tree as well.
What do you think?
I'd say that if the binding forbids multiple compatible strings, the
binding is likely to be wrong. We should fix it, and not make the DTs
worse as a result of a binding issue.
Thanks,
M.
--
Jazz is not dead. It just smells funny...
