Re: [PATCH 16/33] arm64: dts: ti: k3-am62p-verdin: Add missing cfg for TI IPC Firmware

[Beleswar Prasad Padhi <b-padhi@ti.com>]

On 23/08/25 22:13, Francesco Dolcini wrote:
> On Fri, Aug 22, 2025 at 10:09:48PM +0530, Beleswar Prasad Padhi wrote:
>> On 8/21/2025 11:36 AM, Francesco Dolcini wrote:
>>> On Fri, Aug 15, 2025 at 04:08:22AM +0530, Beleswar Padhi wrote:
>>>> The wkup_r5fss0_core0_memory_region is used to store the text/data
>>>> sections of the Device Manager (DM) firmware itself and is necessary for
>>>> platform boot. Whereas the wkup_r5fss0_core0_dma_memory_region is used
>>>> for allocating the Virtio buffers needed for IPC with the DM core which
>>>> could be optional. The labels were incorrectly used in the
>>>> k3-am62p-verdin.dtsi file. Correct the firmware memory region label.
>>>>
>>>> Currently, only mailbox node is enabled with FIFO assignment. However,
>>>> there are no users of the enabled mailboxes. Add the missing carveouts
>>>> for WKUP and MCU R5F remote processors, and enable those by associating
>>>> to the above carveout and mailboxes. This config aligns with other AM62P
>>>> boards and can be refactored out later.
>>>>
>>>> Signed-off-by: Beleswar Padhi <b-padhi@ti.com>
>>>> ---
>>>> Cc: Francesco Dolcini <francesco.dolcini@toradex.com>
>>>> Cc: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
>>>> Cc: Parth Pancholi <parth.pancholi@toradex.com>
>>>> Cc: Jo_o Paulo Gon_alves <joao.goncalves@toradex.com>
>>>> Requesting for a review/test.
>>>>
>>>>   arch/arm64/boot/dts/ti/k3-am62p-verdin.dtsi | 42 ++++++++++++++++++++-
>>>>   1 file changed, 41 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/arch/arm64/boot/dts/ti/k3-am62p-verdin.dtsi b/arch/arm64/boot/dts/ti/k3-am62p-verdin.dtsi
>>>> index 6a04b370d149..0687debf3bbb 100644
>>>> --- a/arch/arm64/boot/dts/ti/k3-am62p-verdin.dtsi
>>>> +++ b/arch/arm64/boot/dts/ti/k3-am62p-verdin.dtsi
>>>> @@ -162,7 +162,25 @@ secure_ddr: optee@9e800000 {
>>>>   			no-map;
>>>>   		};
>>>> -		wkup_r5fss0_core0_memory_region: r5f-dma-memory@9c900000 {
>>>> +		mcu_r5fss0_core0_dma_memory_region: mcu-r5fss-dma-memory-region@9b800000 {
>>>> +			compatible = "shared-dma-pool";
>>>> +			reg = <0x00 0x9b800000 0x00 0x100000>;
>>>> +			no-map;
>>>> +		};
>>>> +
>>>> +		mcu_r5fss0_core0_memory_region: mcu-r5fss-memory-region@9b900000 {
>>> Node name should be generic, `memory@9b900000` ?
>>
>> Humm, that memory is reserved and has the 'no-map' property. So it
>> technically is only used by the node which references it (a particular
>> rproc in this case), and never used by Linux for any allocations. So it
>> is not generic memory per say...
>>
>> So I was inclined for putting the specific node name which uses the
>> carveout in the label. What do you think?
> My understanding is that the node name must be generic, as required by
> the DT specification. What matters is that this is memory, not what is used
> for.


Sorry, I misinterpreted node name with label. I will fix this in v3
Will wait for any other comments on v2:
https://lore.kernel.org/all/20250823160901.2177841-1-b-padhi@ti.com/

Thanks,
Beleswar

>
>
> So it should be something like
>
>
> mcu_r5fss0_core0_dma_memory: memory@9c900000 {
> 	compatible = "shared-dma-pool";
> 	reg = <0x00 0x9b800000 0x00 0x100000>;
> 	no-map;
> };
>
>
> Francesco
>