Re: [PATCH v1 1/3] riscv: dts: spacemit: k3: override reserved status of mimsic and maplic

[E Shattow <e@freeshell.de>]

Hi Anirudh, Troy, et. al.

On 10/29/26 07:22, Anirudh Srinivasan wrote:
> Hi Troy,
> 
> On Thu, Oct 29, 2026 at 1:37 AM Troy Mitchell
> <troy.mitchell@linux.spacemit.com> wrote:
>>
>> On Thu Oct 29, 2026 at 6:35 AM +08, E Shattow wrote:
>>> Hi Anirudh, CC: Troy Mitchell, Guodong Xu
>>>
>>> On 10/28/26 15:01, Anirudh Srinivasan wrote:
>>>> On Wed, Oct 28, 2026 at 11:45:05AM -0700, E Shattow wrote:
>>>>> Linux boot stalls when probing peripherals unless OpenSBI has the
>>>>> enabled mimsic and maplic nodes. Override these nodes so Linux can boot.
>>>>>
>>>>> Reported-by: Yixun Lan <dlan@kernel.org>
>>>>> Signed-off-by: E Shattow <e@freeshell.de>
>>>>> ---
>>>>>  arch/riscv/dts/spacemit-k3-u-boot.dtsi | 4 ++++
>>>>>  1 file changed, 4 insertions(+)
>>>>>
>>>>> diff --git a/arch/riscv/dts/spacemit-k3-u-boot.dtsi b/arch/riscv/dts/spacemit-k3-u-boot.dtsi
>>>>> index 352e85e7de87..278b36acc79f 100644
>>>>> --- a/arch/riscv/dts/spacemit-k3-u-boot.dtsi
>>>>> +++ b/arch/riscv/dts/spacemit-k3-u-boot.dtsi
>>>>> @@ -3,4 +3,8 @@
>>>>>   * Copyright (C) 2026 Yixun Lan <dlan@kernel.org>
>>>>>   */
>>>>>
>>>>> +/* Override reserved status of mimsic and maplic for OpenSBI */
>>>>> +&mimsic { status = "okay"; };
>>>>> +&maplic { status = "okay"; };
>>>>
>>>> This should be fixed in the DT in linux. I guess you could carry this
>>>> patch till it is fixed in linux.
>> I think status = "reserved" in k3.dtsi is correct and should stay.
>> The DT spec defines "reserved" as a device that is operational
>> but should not be used, typically because it is controlled by
>> another software component such as platform firmware.
>>
>> That is exactly the case for the M-level IMSIC and APLIC.
> 
> The DT should be a description of the hw, not just the bits that linux
> cares about. This is why I'm asking for this to be in the linux DT.
> The changes that E is proposing here will have effect in the DT that
> u-boot passes to linux anyway.
> 

The mainline OpenSBI series supporting K3 has now landed. As promised
I've sent a patch upstream to Linux so that we may continue discussion
about the required mimsic and maplic status there:

https://lore.kernel.org/lkml/20261029162343.1896134-1-e@freeshell.de/

>>>>
>>>>> +
>>>>>  #include "spacemit-k3-binman.dtsi"
>>>>> --
>>>>> 2.53.0
>>>>>
>>>
>>> The upstream commit that adds these nodes as reserved status is:
>>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=56f37e391a626
>>>
>>> Quote, "...Also add M-mode APLIC (maplic) and IMSIC (mimsic) nodes to
>>> represent the hardware topology and ready for potential firmware usage."
>>>
>>> When the mainline OpenSBI series lands adding SpacemiT K3 support:
>>>
>>> https://lore.kernel.org/opensbi/20261028-spacemit-k3-v6-0-01d984f60941@linux.spacemit.com/
>>>
>>> ...then I think we could send this upstream?
>> Acked-by: Troy Mitchell <troy.mitchell@linux.spacemit.com>
>>
>>>
>>> Thanks for the review, -E
>>
>>
>> --
>> Troy Mitchell
>>

Thanks everyone for advice about this, I think it will be okay in any
decision about whether to change this in devicetree-rebasing upstream,
to carry these changes here as-needed in U-Boot for passing to OpenSBI.
Let's get consensus with discussion on LKML and then if there is
something more narrow in scope needed that can be done from U-Boot for
changing these nodes to pass to OpenSBI, I am open to that possibility too

-E