Re: [PATCH v7 07/11] arm64: dts: ti: k3-am62a-ti-ipc-firmware: Move wkup reserved memory

[Andrew Davis <afd@ti.com>]

On 8/25/26 7:43 AM, Markus Schneider-Pargmann wrote:
> Hi Andrew,
> 
> On Thu Jul 9, 2026 at 9:33 PM CEST, Andrew Davis wrote:
>> On 7/1/26 7:39 AM, Markus Schneider-Pargmann (TI) wrote:
>>> Move the reserved memory regions used for wkup_r5fss0_core0 to the
>>> k3-am62a-ti-ipc-firmware.dtsi. These are all the same for the other
>>> boards as well, so we can combine them here similar to what is already
>>> done for the mcu_r5fss0_core0 memory regions.
>>
>> The reason we left this region here in the base DTSI is this region
>> is for the DM firmware and is required for the SoC to function.
>> The other firmware regions are all optional.
> 
> I left this in the base dts before and Nishanth pointed out that it
> should be in the ipc-firmware.dtsi so it only needs to be changed in one
> place:
> 
>    That was the entire purpose of splitting ipc-firmware.dtsi out -
>    if the default firmware carveouts have changed, do the change in
>    ipc-firmware.dtsi. That way some other firmware people will have could
>    potentially have their own dtsi files and handle things appropriately
>    there, even if they are using the same evm OR if they use the exact
>    ipc-firmware on different boards, that will appropriately introduce
>    the change there as well. BUT, do check first.
> 
> https://lore.kernel.org/r/20260507155710.5zch4ghhzrj7hnlh@immovably
> 
> Maybe I am misunderstanding what you are suggesting. Also note that the
> memory-region list in wkup_r5fss0_core0 which references the memory
> regions defined in the base dts were always in the ipc-firmware.dtsi.
> 
> Could you please help me understand how this is modeled so I can update
> the patches appropriately?
> 

I think the issue with your v3 was that you updated the wkup_r5fss0_core0
node outside of the ipc-firmware.dtsi file where it was defined.

The core idea behind ipc-firmware.dtsi is it should be optional. If for
instance if I have a different set of remote firmware then I only need
to modify that .dtsi file. If I don't want any remote firmware and my
bootloader starts none of the remote cores, then I should be able to
simply not apply that .dtsi file and not lose any memory to unneeded
carveouts.

Now the wkup_r5fss0_core0 is a little different, that core runs the
DM firmware, that firmware is mandatory to even boot the system, it is
not optional. If I try to boot Linux without marking its memory region
as reserved then Linux will write over it can things will break. So,
its carveout (and only its carveouts) remains in the base dts.

Do as you did in v4, but update the wkup_r5fss0_core0 node in
ipc-firmware.dtsi, not only for k3-am62a7-sk.dts. Yes this means
you will need to make this change for all AM62A based boards, but
that will be the correct thing to do as all boards will be using
the same DM firmware anyway.

Andrew

> Thanks!
> 
> Best
> Markus