RV1103B: Attempt to upstream

RV1103B: Attempt to upstream

[Fabio Estevam]

Hi,

I have a custom board based on the Rockchip evb-rv1103 (RV1103B silicon),
booting from SPI-NAND.

I can successfully boot the vendor Rockchip U-Boot v2017.09 on this board,
from SPI-NAND

I am currently working on upstreaming U-Boot support for the RV1103B SoC.
My first RFC attempt is available here:

  https://github.com/fabioestevam/u-boot/commits/rv1103/

Build steps:

  make evb-rv1103_defconfig
  make

However, I am not yet able to boot U-Boot on the board. At the moment,
the board does not boot and only prints "RKUART".

My initial plan was to avoid SPL, similar to the evb-rv1108 approach
(described in board/rockchip/evb_rv1108/README).

Before going further, I would like to ask:

Is the evb-rv1108 “no SPL” boot flow a valid reference for RV1103B?

Any guidance or pointers would be greatly appreciated.

Thanks,

Fabio Estevam

Re: RV1103B: Attempt to upstream

[Yao Zi]

On Thu, Dec 18, 2025 at 11:45:11PM -0300, Fabio Estevam wrote:
> Hi,
> 
> I have a custom board based on the Rockchip evb-rv1103 (RV1103B silicon),
> booting from SPI-NAND.
> 
> I can successfully boot the vendor Rockchip U-Boot v2017.09 on this board,
> from SPI-NAND
> 
> I am currently working on upstreaming U-Boot support for the RV1103B SoC.
> My first RFC attempt is available here:

I think the upstreaming process should start from the kernel side,
then we will have stable devicetree ABI reference to work with, instead
of creating everything from scratch in U-Boot.

This will also makes it easier to enable OF_UPSTREAM.

>   https://github.com/fabioestevam/u-boot/commits/rv1103/
> 
> Build steps:
> 
>   make evb-rv1103_defconfig
>   make
> 
> However, I am not yet able to boot U-Boot on the board. At the moment,
> the board does not boot and only prints "RKUART".
> 
> My initial plan was to avoid SPL, similar to the evb-rv1108 approach
> (described in board/rockchip/evb_rv1108/README).
> 
> Before going further, I would like to ask:
> 
> Is the evb-rv1108 “no SPL” boot flow a valid reference for RV1103B?
> 
> Any guidance or pointers would be greatly appreciated.
> 
> Thanks,
> 
> Fabio Estevam

Best regards,
Yao Zi

Re: RV1103B: Attempt to upstream

[Fabio Estevam]

Hi Yao,

On Fri, Dec 19, 2025 at 12:23 AM Yao Zi <me@ziyao.cc> wrote:

> I think the upstreaming process should start from the kernel side,
> then we will have stable devicetree ABI reference to work with, instead
> of creating everything from scratch in U-Boot.
>
> This will also makes it easier to enable OF_UPSTREAM.

I can help upstream the RV1103 device trees in Linux.

However, at this point, I would like some guidance on the RV1103's
boot process with mainline U-Boot.

Re: RV1103B: Attempt to upstream

[Quentin Schulz]

Hi Fabio,

On 12/19/25 3:45 AM, Fabio Estevam wrote:
> Hi,
> 
> I have a custom board based on the Rockchip evb-rv1103 (RV1103B silicon),
> booting from SPI-NAND.
> 
> I can successfully boot the vendor Rockchip U-Boot v2017.09 on this board,
> from SPI-NAND
> 
> I am currently working on upstreaming U-Boot support for the RV1103B SoC.
> My first RFC attempt is available here:
> 
>    https://github.com/fabioestevam/u-boot/commits/rv1103/
> 
> Build steps:
> 
>    make evb-rv1103_defconfig
>    make
> 
> However, I am not yet able to boot U-Boot on the board. At the moment,
> the board does not boot and only prints "RKUART".
> 
> My initial plan was to avoid SPL, similar to the evb-rv1108 approach
> (described in board/rockchip/evb_rv1108/README).
> 

There is an SPL, it's just a blob from Rockchip.

> Before going further, I would like to ask:
> 
> Is the evb-rv1108 “no SPL” boot flow a valid reference for RV1103B?
> 

No clue.

> Any guidance or pointers would be greatly appreciated.
> 

Do you actually use the same SPL blob as in the v2017.09 image? Do you 
see something on UART on that image?

I would think the SPL blob from Rockchip would print a bunch of things 
on the UART (e.g. while initializing the DDR like it's used to on 
RK3588). If you are using a different baudrate than the one set in the 
blob (usually 1500000 but maybe it's different for RV?), serial 
controller or serial IO mux (the number after M in the pinmux, e.g. 
UARTXMY means controller X mux Y), you can modify it with 
https://github.com/rockchip-linux/rkbin/blob/master/tools/ddrbin_tool.py.

RV1103 is somewhat old, but I'm assuming it's not impacted by the bug 
that requires us to use rkspi (instead of rksd) format for SPI images? 
You would need to write 2KiB every 4KiB IIRC. You can check after 
flashing the downstream image what is the content of your flash I guess 
to check that.

Usually, the blobs from Rockchip end up returning to BootROM for the 
next stage, so U-Boot proper would need to be stored somewhere in DRAM 
and jumping to it by the BootROM for it to work? Not sure about those 
details, I have been fortunate enough so far that I didn't have to do 
any SoC bring-up but maybe Jonas or Kever can help here?

Good luck!
Quentin

Re: RV1103B: Attempt to upstream

[Quentin Schulz]

Hi Fabio,

On 12/19/25 3:45 AM, Fabio Estevam wrote:
> Hi,
> 
> I have a custom board based on the Rockchip evb-rv1103 (RV1103B silicon),
> booting from SPI-NAND.
> 
> I can successfully boot the vendor Rockchip U-Boot v2017.09 on this board,
> from SPI-NAND
> 
> I am currently working on upstreaming U-Boot support for the RV1103B SoC.
> My first RFC attempt is available here:
> 
>    https://github.com/fabioestevam/u-boot/commits/rv1103/
> 
> Build steps:
> 
>    make evb-rv1103_defconfig
>    make
> 
> However, I am not yet able to boot U-Boot on the board. At the moment,
> the board does not boot and only prints "RKUART".
> 
> My initial plan was to avoid SPL, similar to the evb-rv1108 approach
> (described in board/rockchip/evb_rv1108/README).
> 

There is an SPL, it's just a blob from Rockchip.

> Before going further, I would like to ask:
> 
> Is the evb-rv1108 “no SPL” boot flow a valid reference for RV1103B?
> 

No clue.

> Any guidance or pointers would be greatly appreciated.
> 

Do you actually use the same SPL blob as in the v2017.09 image? Do you 
see something on UART on that image?

I would think the SPL blob from Rockchip would print a bunch of things 
on the UART (e.g. while initializing the DDR like it's used to on 
RK3588). If you are using a different baudrate than the one set in the 
blob (usually 1500000 but maybe it's different for RV?), serial 
controller or serial IO mux (the number after M in the pinmux, e.g. 
UARTXMY means controller X mux Y), you can modify it with 
https://github.com/rockchip-linux/rkbin/blob/master/tools/ddrbin_tool.py.

RV1103 is somewhat old, but I'm assuming it's not impacted by the bug 
that requires us to use rkspi (instead of rksd) format for SPI images? 
You would need to write 2KiB every 4KiB IIRC. You can check after 
flashing the downstream image what is the content of your flash I guess 
to check that.

Usually, the blobs from Rockchip end up returning to BootROM for the 
next stage, so U-Boot proper would need to be stored somewhere in DRAM 
and jumping to it by the BootROM for it to work? Not sure about those 
details, I have been fortunate enough so far that I didn't have to do 
any SoC bring-up but maybe Jonas or Kever can help here?

Good luck!
Quentin

Re: RV1103B: Attempt to upstream

[Fabio Estevam]

Hi Quentin,

On Thu, Jan 8, 2026 at 1:33 PM Quentin Schulz <quentin.schulz@cherry.de> wrote:

> Do you actually use the same SPL blob as in the v2017.09 image? Do you
> see something on UART on that image?

Thanks for your reply.

The board boots with mainline U-Boot now.

I will submit the patches to the list soon.

Thanks

Re: RV1103B: Attempt to upstream

[Christopher Obbard]

Hi Fabio,

On Thu, 2025-12-18 at 23:45 -0300, Fabio Estevam wrote:
> Hi,
> 
> I have a custom board based on the Rockchip evb-rv1103 (RV1103B silicon),
> booting from SPI-NAND.
> 
> I can successfully boot the vendor Rockchip U-Boot v2017.09 on this board,
> from SPI-NAND
> 
> I am currently working on upstreaming U-Boot support for the RV1103B SoC.
> My first RFC attempt is available here:
> 
>   https://github.com/fabioestevam/u-boot/commits/rv1103/
> 
> Build steps:
> 
>   make evb-rv1103_defconfig
>   make
> 
> However, I am not yet able to boot U-Boot on the board. At the moment,
> the board does not boot and only prints "RKUART".
> 
> My initial plan was to avoid SPL, similar to the evb-rv1108 approach
> (described in board/rockchip/evb_rv1108/README).
> 
> Before going further, I would like to ask:
> 
> Is the evb-rv1108 “no SPL” boot flow a valid reference for RV1103B?
> 

Great to see the RV1103 being upstreamed!


I have a Rockchip RV1106 based board, the CoolPi 1B https://www.cool-pi.com/product-category/gateway/cp1b/ which I'd also like to get in upstream linux/u-boot.

The downstream kernel/devicetree is at https://github.com/coolpi-george/coolpi-kernel/blob/linux-5.10-gen-rkr8/arch/arm/boot/dts/rv1106-cp1b.dts with defconfig
https://github.com/coolpi-george/coolpi-kernel/blob/linux-5.10-gen-rkr8/arch/arm/configs/rv1106_cp1b_defconfig

The downstream u-boot is https://github.com/coolpi-george/coolpi-loader/blob/develop/configs/rv1106_coolpicp1b_defconfig

I wonder how similar the RV1106 SoC is to the RV1103 you're working on ?

I only ask since the RV1106 SoC has a similar prefix to the RV1103. I don't have any other information about the SoC.


Cheers!

Chris

Re: RV1103B: Attempt to upstream

[Yao Zi]

On Sat, Jan 10, 2026 at 03:20:39AM +0000, Christopher Obbard wrote:
> Hi Fabio,
> 
> On Thu, 2025-12-18 at 23:45 -0300, Fabio Estevam wrote:
> > Hi,
> > 
> > I have a custom board based on the Rockchip evb-rv1103 (RV1103B silicon),
> > booting from SPI-NAND.
> > 
> > I can successfully boot the vendor Rockchip U-Boot v2017.09 on this board,
> > from SPI-NAND
> > 
> > I am currently working on upstreaming U-Boot support for the RV1103B SoC.
> > My first RFC attempt is available here:
> > 
> >   https://github.com/fabioestevam/u-boot/commits/rv1103/
> > 
> > Build steps:
> > 
> >   make evb-rv1103_defconfig
> >   make
> > 
> > However, I am not yet able to boot U-Boot on the board. At the moment,
> > the board does not boot and only prints "RKUART".
> > 
> > My initial plan was to avoid SPL, similar to the evb-rv1108 approach
> > (described in board/rockchip/evb_rv1108/README).
> > 
> > Before going further, I would like to ask:
> > 
> > Is the evb-rv1108 “no SPL” boot flow a valid reference for RV1103B?
> > 
> 
> Great to see the RV1103 being upstreamed!
> 
> 
> I have a Rockchip RV1106 based board, the CoolPi 1B https://www.cool-pi.com/product-category/gateway/cp1b/ which I'd also like to get in upstream linux/u-boot.
> 
> The downstream kernel/devicetree is at https://github.com/coolpi-george/coolpi-kernel/blob/linux-5.10-gen-rkr8/arch/arm/boot/dts/rv1106-cp1b.dts with defconfig
> https://github.com/coolpi-george/coolpi-kernel/blob/linux-5.10-gen-rkr8/arch/arm/configs/rv1106_cp1b_defconfig
> 
> The downstream u-boot is https://github.com/coolpi-george/coolpi-loader/blob/develop/configs/rv1106_coolpicp1b_defconfig
> 
> I wonder how similar the RV1106 SoC is to the RV1103 you're working on ?

Given that the devicetree for RV1103 in Rockchip downstream kernel tree
includes RV1106's devicetree[1] and only adds some overrides, I'd like
to say RV1103 is basically only in a different package. They're likely
the same silicon, probably with different qualities.

> I only ask since the RV1106 SoC has a similar prefix to the RV1103. I don't have any other information about the SoC.
> 
> 
> Cheers!
> 
> Chris

Regards,
Yao Zi

[1]: https://github.com/rockchip-linux/kernel/blob/1ba51b059f25533c5529b7f68186190b47d6a7b3/arch/arm/boot/dts/rockchip/rv1103.dtsi#L6