[PATCH v1 00/12] Support OF_UPSTREAM for StarFive JH7110

[E Shattow]

Hi,

On Thu, Oct 3, 2024 at 10:23 AM Hal Feng <hal.feng at starfivetech.com> wrote:
>
> > On 01.10.24 01:56, Heinrich Schuchardt wrote:
> > Another approach would be to build all device-trees into U-Boot and use the
> > EEPROM information to select the correct one at runtime.
> >
> > This approach is selected by Wandboard defconfig and other boards by
> > specifying CONFIG_MULTI_DTB_FIT and CONFIG_OF_LIST. The choice of the
> > device-tree is made in board_fit_config_name_match().
> >
> > The advantage from the standpoint of a Linux distro would be that one distro
> > image can be provided that will run on all JH7110 boards.
>
> That is a good idea. I will try to do this after my five-day vacation.
> By the way, I wanna know whether jh7110-milkv-mars-cm.dts and
> jh7110-milkv-mars-cm-lite.dts are in the process of upstreaming?
> I don't see them in the kernel mainline.
>
> Thanks for your suggestions.
>
> Best regards,
> Hal

There is not any jh7110-milkv-mars-cm.dts and
jh7110-milkv-mars-cmlite.dts patch for Linux kernel upstream yet.

Now may be a good time to do so, since Star64 was accepted and makes
use of the common jh7110 dtsi that was introduced with devicetree of
Milk-V Mars to Linux kernel. First this can go in U-Boot as the
jh7110-milkv-mars.dts is good for Milk-V Mars CM and CM Lite, with
addition of some GPIO assignments. This can be stubbed out by
including jh7110-milkv-mars.dts until there are accepted upstream
devicetree files for Mars CM and CM Lite. There are different GPIO
assignments depending if it is U-Boot phase or Linux phase of the boot
process; per the vendor sources. It is enough to pass a loaded
jh7110-milkv-mars.dts to Linux on Mars CM and CM Lite hardware, if
this GPIO assignment was done at U-Boot time and since then there is
never any MMC reset while in Linux kernel:

https://github.com/milkv-mars/mars-buildroot-sdk/commit/880a249518f72ecf1e2947dfeb2c66e5035fce90

https://github.com/milkv-mars/mars-buildroot-sdk/commit/042ea06598995db99ddbbbb252ee439c42b9c1a9

Much of the above diff listings can be ignored as already implemented
or not needed, however, provided for reference what I am talking about
for GPIO assignments 22 24 62 depending on the phase of boot process.

-E