On Fri, Aug 02, 2024 at 04:08:39PM +0530, Sumit Garg wrote:
Hi Andre,
On Fri, 2 Aug 2024 at 15:27, Andre Przywara andre.przywara@arm.com wrote:
On Fri, 2 Aug 2024 12:24:56 +0530 Sumit Garg sumit.garg@linaro.org wrote:
Hi Sumit,
Hi Chris,
On Fri, 2 Aug 2024 at 04:29, Chris Morgan macroalpha82@gmail.com wrote:
From: Chris Morgan macromorgan@hotmail.com
The Anbernic RG35XX series of devices are based around an Allwinner H700 SoC with 1GB of RAM, 2 SD cards, and multiple input buttons.
This bootloader has been tested on the Anbernic RG35XX-2024 and RG35XX-H, but should be suitable for the entire lineup of H700 based devices.
A future series of updates will add board selection logic to identify and load the correct device tree automatically, and will switch to utilizing a mainline device tree.
I can already see the upstream DT subtree containing DTS for this board as:
./dts/upstream/src/arm64/allwinner/sun50i-h700-anbernic-rg35xx-2024.dts
Are there any major differences that won't allow you to switch to OF_UPSTREAM for this board?
In this case there are no differences between the DT in the kernel repo and what we put in U-Boot. The main reason for me holding back the switch to OF_UPSTREAM for the H616 was indeed the rather slow update cadence, which somewhat slows down upstreaming of new boards like this. Doing this separately gives us a bit more control and freedom to take DT patches early, for instance when they appear in some Linux -rc1, or even before that (linux-next). I explained my position in more detail here: https://lore.kernel.org/u-boot/191A0F66-E2AE-4D1B-ACBE-E411E135BB61@arm.com/
I can see the reasoning, thanks.
I think Tom partly convinced me that regular updates together with cherry-picking should solve most of those problems, so I will probably do the switch for the next cycle.
Glad to hear that. Cherry-picking has been working really well for Rockchip platforms [1] [2] [3] as the changes appear in Linux -rc1. Also, it pretty much avoids maintainers to review the DT changes unless there are some bits added via *-u-boot.dtsi.
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=417233 [2] https://patchwork.ozlabs.org/project/uboot/list/?series=417377 [3] https://patchwork.ozlabs.org/project/uboot/list/?series=417353
-Sumit
The specific differences between this and the upstream tree is that the upstream tree is missing the switch from rsb to i2c for the PMIC. That said, if we can cherry pick the change (it was just accepted and should end up in 6.12) that would work and I can start from the get-go with an upstream tree. There may be other differences, but they are ultimately superficial for U-Boot. I'll still make this for the RK35XX-2024 only for now, because I need to finish some stuff in the detection logic.
Thank you. Chris