Re: [PATCH v4 3/3] arm64: dts: realtek: Add RTD1319 SoC and Realtek Pym Particles EVB

["Andreas Färber" <afaerber@suse.de>]

Hi James,

Am 21.06.20 um 01:32 schrieb Andreas Färber:
> From: James Tai <james.tai@realtek.com>
> 
> Add Device Trees for Realtek RTD1319 SoC family, RTD1319 SoC and
> Realtek Pym Particles EVB.
> 
> Signed-off-by: James Tai <james.tai@realtek.com>
> Signed-off-by: Andreas Färber <afaerber@suse.de>
> ---
>   v3 -> v4:
>   * Updated Realtek copyright for 2 out of 3 files from v3
>   * Renamed from rtd1319-pymparticle.dts to rtd1319-pymparticles.dts
>   * Updated compatible from pymparticle to pym-particles
>   * Updated PMU compatible from armv8-pmuv3 to cortex-a55-pmu (Robin)
>   
>   v2 -> v3:
>   * Add virtual maintenance interrupt for architecture timer
>   * Correct the GIC redistributor address range
[...]
> diff --git a/arch/arm64/boot/dts/realtek/rtd13xx.dtsi b/arch/arm64/boot/dts/realtek/rtd13xx.dtsi
> new file mode 100644
> index 000000000000..8c5b6fc7b8eb
> --- /dev/null
> +++ b/arch/arm64/boot/dts/realtek/rtd13xx.dtsi
[...]
> +		gic: interrupt-controller@ff100000 {
> +			compatible = "arm,gic-v3";
> +			reg = <0xff100000 0x10000>,
> +			      <0xff140000 0x80000>;
> +			interrupts = <GIC_PPI 9 IRQ_TYPE_LEVEL_HIGH>;

In my testing this appears to cause the following error:

[    2.239858] irq: type mismatch, failed to map hwirq-25 for 
interrupt-controller@ff100000!
...
[    3.505649] kvm [1]: IPA Size Limit: 40bits
[    3.506051] kvm [1]: GICv3: no GICV resource entry
[    3.506058] kvm [1]: disabling GICv2 emulation
[    3.506081] kvm [1]: GIC system register CPU interface enabled
[    3.506175] kvm [1]: vgic interrupt IRQ1
[    3.506293] kvm [1]: Hyp mode initialized successfully

If I change it to IRQ_TYPE_LEVEL_LOW, that error goes away:

[    3.506030] kvm [1]: IPA Size Limit: 40bits
[    3.506430] kvm [1]: GICv3: no GICV resource entry
[    3.506437] kvm [1]: disabling GICv2 emulation
[    3.506459] kvm [1]: GIC system register CPU interface enabled
[    3.506551] kvm [1]: vgic interrupt IRQ1
[    3.506672] kvm [1]: Hyp mode initialized successfully

In-tree RTD1619 has it as HIGH, too, but doesn't show above error:

[    2.918973] kvm [1]: IPA Size Limit: 40bits
[    2.919345] kvm [1]: GICv3: no GICV resource entry
[    2.919352] kvm [1]: disabling GICv2 emulation
[    2.919373] kvm [1]: GIC system register CPU interface enabled
[    2.919522] kvm [1]: vgic interrupt IRQ1
[    2.919700] kvm [1]: Hyp mode initialized successfully

RTD1619 doesn't show an error either if I change it to LOW though:

[    2.918843] kvm [1]: IPA Size Limit: 40bits
[    2.919212] kvm [1]: GICv3: no GICV resource entry
[    2.919218] kvm [1]: disabling GICv2 emulation
[    2.919240] kvm [1]: GIC system register CPU interface enabled
[    2.919390] kvm [1]: vgic interrupt IRQ1
[    2.919567] kvm [1]: Hyp mode initialized successfully

The GICv3 bindings example does have it as 4 == HIGH, but so does the 
GICv2 binding example, and yet we used LOW == 8 for in-tree RTD139x, 
RTD129x and RTD1195. The downstream BSP uses value 4 == HIGH for both 
RTD16xx and RTD13xx - is it possible this was never actually tested?

Thanks in advance for clarifying the correct interrupt polarity.

> +			interrupt-controller;
> +			#interrupt-cells = <3>;
> +		};
[snip]

Regards,
Andreas

-- 
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)