Re: [PATCHv5 3/7] arm64: dts: rockchip: Add base DT for rk3588 SoC

[Marc Zyngier <maz@kernel.org>]

On Mon, 05 Dec 2022 17:23:46 +0000,
Sebastian Reichel <sebastian.reichel@collabora.com> wrote:
> 
> From: Kever Yang <kever.yang@rock-chips.com>
> 
> This initial version supports (single core) CPU, dma, interrupts, timers,
> UART and SDHCI. In short - everything necessary to boot Linux on this
> system on chip.
> 
> The DT is split into rk3588 and rk3588s, which is a reduced version
> (i.e. with less peripherals) of the former.
> 
> Signed-off-by: Yifeng Zhao <yifeng.zhao@rock-chips.com>
> Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
> Signed-off-by: Sugar Zhang <sugar.zhang@rock-chips.com>
> Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
> [rebase, squash and reword commit message]
> Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com>
> ---
>  arch/arm64/boot/dts/rockchip/rk3588.dtsi  |   58 +
>  arch/arm64/boot/dts/rockchip/rk3588s.dtsi | 1672 +++++++++++++++++++++
>  2 files changed, 1730 insertions(+)
>  create mode 100644 arch/arm64/boot/dts/rockchip/rk3588.dtsi
>  create mode 100644 arch/arm64/boot/dts/rockchip/rk3588s.dtsi
> 
> diff --git a/arch/arm64/boot/dts/rockchip/rk3588s.dtsi b/arch/arm64/boot/dts/rockchip/rk3588s.dtsi
> new file mode 100644
> index 000000000000..ecdd2294cd42
> --- /dev/null
> +++ b/arch/arm64/boot/dts/rockchip/rk3588s.dtsi
> @@ -0,0 +1,1672 @@

[...]

> +	pmu-a55 {
> +		compatible = "arm,cortex-a55-pmu";
> +		interrupts = <GIC_PPI 7 IRQ_TYPE_LEVEL_HIGH>;
> +	};
> +
> +	pmu-a76 {
> +		compatible = "arm,cortex-a76-pmu";
> +		interrupts = <GIC_PPI 7 IRQ_TYPE_LEVEL_HIGH>;
> +	};

Two PMUs with the same PPI and no partition? This is totally
wrong. Please see how RK3399 does it, with each PMU having a PPI
partition allocated.

See the PMU binding for the details.

> +
> +	psci {
> +		compatible = "arm,psci-1.0";
> +		method = "smc";
> +	};
> +
> +	spll: clock-0 {
> +		compatible = "fixed-clock";
> +		clock-frequency = <702000000>;
> +		clock-output-names = "spll";
> +		#clock-cells = <0>;
> +	};
> +
> +	timer {
> +		compatible = "arm,armv8-timer";
> +		interrupts = <GIC_PPI 13 IRQ_TYPE_LEVEL_HIGH>,
> +			     <GIC_PPI 14 IRQ_TYPE_LEVEL_HIGH>,
> +			     <GIC_PPI 11 IRQ_TYPE_LEVEL_HIGH>,
> +			     <GIC_PPI 10 IRQ_TYPE_LEVEL_HIGH>;

You have a set of ARMv8.2 cores, which do have an interrupt for the
EL2 virtual timer. Please add this interrupt and while you're at it,
add the interrupt-names properties that are associated with them.

See the binding for the details.

[...]

> +	gic: interrupt-controller@fe600000 {
> +		compatible = "arm,gic-v3";
> +		reg = <0x0 0xfe600000 0 0x10000>, /* GICD */
> +		      <0x0 0xfe680000 0 0x100000>; /* GICR */
> +		interrupts = <GIC_PPI 9 IRQ_TYPE_LEVEL_HIGH>;
> +		interrupt-controller;
> +		mbi-alias = <0x0 0xfe610000>;
> +		mbi-ranges = <424 56>;
> +		msi-controller;
> +		#interrupt-cells = <3>;
> +
> +		ppi-partitions {
> +			interrupt-partition-0 {
> +				affinity = <
> +					&cpu_l0 &cpu_l1 &cpu_l2 &cpu_l3
> +					&cpu_b0 &cpu_b1 &cpu_b2 &cpu_b3
> +				>;
> +			};
> +		};

What is the purpose of having an interrupt partition that covers *all*
the CPUs? This makes zero sense. You need to:

- Bump #interrupt-cells to 4, as per the GIC binding

- Create PPI partitions to segregate the two CPU types

- Make the PMU devices use the corresponding PPI partition as per the binding

- Fix all the interrupt specifiers to use 4 cells instead of 3

Again, RK3399 got it right, and for once I'm advocating some sort of
copy/paste...

	M.

-- 
Without deviation from the norm, progress is not possible.