Hi Randolph,

On Mon, 13 Nov 2023 at 23:26, Randolph Lin randolph@andestech.com wrote:

Hi Simon, Thanks a lot. On Fri, Nov 10, 2023 at 04:50:24AM -0700, Simon Glass wrote:

...

Hi Randolph,

On Wed, Nov 8, 2023, 20:15 Randolph randolph@andestech.com wrote:

...

The #address-cells is now equal to 2. The format of the load field for the Linux kernel doesn't match.

Signed-off-by: Randolph randolph@andestech.com

arch/riscv/dts/binman.dtsi | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/riscv/dts/binman.dtsi b/arch/riscv/dts/binman.dtsi index 6b4eb8dc7b..5117d7c8c9 100644 --- a/arch/riscv/dts/binman.dtsi +++ b/arch/riscv/dts/binman.dtsi @@ -50,7 +50,8 @@ os = "Linux"; arch = "riscv"; compression = "none";

•                                   load = <CONFIG_TEXT_BASE>;
  

•                                   load = <U64_TO_U32_H(CONFIG_TEXT_BASE)
  

•                                           U64_TO_U32_L(CONFIG_TEXT_BASE)>;
  

I just see the #address-cells changed from 1 to 2 in commit id: 5a348ccf0257. In my last commit for binman.dtsi (commit id: d311df8b3169), it is based on the value of #address-cells being 1. That's the reason for my patch submission.

...

Does this work?

load = /bits 64/ <CONFIG_TEXT_BASE>

Yes, it works. We use this way for the DDR memory start address above 4G platform. We find the example for 64bit address in the document tools/binman/binman.rst and use it. What is the method that we should continue to use in the binman.dtsi?

1. #address-cells = 2
2. append /bits/64

I believe these go together. Using (1) means you have a 2-cell value, which is 64 bits since each cell is 32-bits.

Or perhaps I misunderstand what you are asking?

...

...

                                    linux_blob: blob-ext {
                                            filename = "Image";

-- 2.34.1

Regards, Simon