[PATCH 2/3] dt-bindings: interrupt-controller: add UltraRISC DP1000 PLIC
Vivian Wang
wangruikang at iscas.ac.cn
Fri Oct 10 15:03:34 PDT 2025
On 10/10/25 21:57, Lucas Zampieri wrote:
> On Thu, Oct 9, 2025 at 9:39 AM Krzysztof Kozlowski <krzk at kernel.org> wrote:
>> On 09/10/2025 17:20, Lucas Zampieri wrote:
>>> From: Charles Mirabile <cmirabil at redhat.com>
>>>
>>> Add a new compatible string for UltraRISC DP1000 PLIC.
>>>
>>> Signed-off-by: Charles Mirabile <cmirabil at redhat.com>
>>> ---
>>> .../bindings/interrupt-controller/sifive,plic-1.0.0.yaml | 2 ++
>>> 1 file changed, 2 insertions(+)
>>>
>>> diff --git a/Documentation/devicetree/bindings/interrupt-controller/sifive,plic-1.0.0.yaml b/Documentation/devicetree/bindings/interrupt-controller/sifive,plic-1.0.0.yaml
>>> index 5b827bc24301..a419de50f5a8 100644
>>> --- a/Documentation/devicetree/bindings/interrupt-controller/sifive,plic-1.0.0.yaml
>>> +++ b/Documentation/devicetree/bindings/interrupt-controller/sifive,plic-1.0.0.yaml
>>> @@ -74,6 +74,8 @@ properties:
>>> - sophgo,sg2044-plic
>>> - thead,th1520-plic
>>> - const: thead,c900-plic
>>> + - items:
>>
>> Missing SoC specific front compatible, me thinks.
> Right, if I'm understanding this correctly, I would need to add the
> sifive,plic-1.0.0 fallback compatible to indicate register layout
> compatibility. However, the DP1000 PLIC has a claim register hardware
> bug that breaks interrupt handling without the driver quirk. Should we
> still include the fallback even though the generic driver doesn't
> work?
See what the thead,c900-plic thing is doing. The PLIC compatible should
be SoC-associated model, and then CPU-core-associated model.
(I guess *theoretically* the PLIC could be external to the CPU but it
doesn't really make sense to design a core this way.)
Supposedly [1] the cores are called UR-CP100, so it should be
compatible = "ultrarisc,dp1000-plic", "ultrarisc,cp100-plic";
And the driver should match on ultrarisc,cp100-plic instead.
Vivian "dramforever" Wang
[1]: https://ultrarisc.github.io/ultrarisc-devblog/2025/06/18/dp1000-spec/
More information about the linux-riscv
mailing list