On Sun, Sep 22, 2024 at 5:33 PM Marek Vasut marex@denx.de wrote:
On 9/17/24 8:21 AM, Lothar Rubusch wrote:
Introduce the support for three variants of Enclustra's Intel Mercury AA1 (with Intel Arria10) SoMs and additional configs. This is supposed to be the first step to upstream several of Enclustras SoC FPGA Modules. There are still things to be modified. So, this is supposed to start the discussion and also to clarify some of my open issues.
On the DT approach: I can agree that in parallel I may go for upstreaming the kernel DT. There is a fundamental difference, still: In order to represent the combination of module + baseboard + bootmode for the kernel DT we have to use a mechanism like DTO. For u-boot this can be simpler. Currently the few adjustments needed will be represented by the DT files as are. I propose to keep them, and in case remove them later, which I doubt, since the full DT for the kernel will cover more details. In case refactoring is needed I can update the DT also in u-boot.
I assume after the LPC discussion, there will be a V3 ?
Yes.
V3 affects mainly the DT question. As of the LPC discussions I had, currently I'm about to rework the DT material. The material will need to pass dtbs_check/dtschema. Functionality then needs to be verified on my hardware zoo, and eventually I upstream it directly to the Linux kernel ML. If successful, then, in u-boot I will build up on upstream DTs as proposed here initially.
V2: Please, feel free to already review V2 patch set. As I assume still some issues. Alternatively, we do this on the V3 patch set. No stress! Every feedback here is highly appreciate!!
I pretend putting the involved u-boot people here in CC on the kernel ML. Pls, let me know if I need to do the u-boot ML in CC as well.