[PATCH v2 6/9] spacemit: k1: Add multiple device tree support

[Aurelien Jarno]

Hi,

On 2026-05-20 06:45, Guodong Xu wrote:
> Enable multiple DTB support in the FIT image for the Spacemit K1 SoC,
> allowing a single U-Boot binary to support different board variants.
> 
> The SPL reads the board type from EEPROM and selects the corresponding
> device tree at runtime via board_fit_config_name_match(), ensuring the
> correct hardware description is passed to U-Boot proper.
> 
> Signed-off-by: Guodong Xu <guodong at riscstar.com>
> 
> ---
> v2:
> - Reworked.  Drop the v1 approach (new local k1-muse-pi-pro.dts,
>   enlarge SYS_MALLOC_F_LEN, MMODE switch).
> - Use binman --fit-multi-config so u-boot.itb packs multiple board
>   DTs from the upstream tree.
> ---
>  board/spacemit/k1/spl.c       | 42 ++++++++++++++++++++++++++++++++++--------
>  configs/spacemit_k1_defconfig |  2 ++
>  2 files changed, 36 insertions(+), 8 deletions(-)
> 
> diff --git a/board/spacemit/k1/spl.c b/board/spacemit/k1/spl.c
> index da4169fbc8c..d749e21a2d5 100644
> --- a/board/spacemit/k1/spl.c
> +++ b/board/spacemit/k1/spl.c
> @@ -13,6 +13,7 @@
>  #include <dm/device.h>
>  #include <dm/uclass.h>
>  #include <i2c.h>
> +#include <linux/ctype.h>
>  #include <linux/delay.h>
>  #include <log.h>
>  #include <power/regulator.h>
> @@ -55,6 +56,8 @@ struct ddr_cfg {
>  binman_sym_declare(ulong, ddr_fw, image_pos);
>  binman_sym_declare(ulong, ddr_fw, size);
>  
> +char product_name[I2C_BUF_SIZE] = "k1";
> +
>  static void i2c_early_init(void)
>  {
>  	struct udevice *bus;
> @@ -322,14 +325,8 @@ void nor_early_init(void)
>  	udelay(10);
>  }
>  
> -void *board_spl_fit_buffer_addr(ulong fit_size, int sectors, int bl_len)
> -{
> -	return (void *)CONFIG_SPL_LOAD_FIT_ADDRESS;
> -}
> -
>  void board_init_f(ulong dummy)
>  {
> -	u8 i2c_buf[I2C_BUF_SIZE] = { 0 };
>  	int ret;
>  
>  	ret = spl_early_init();
> @@ -344,11 +341,11 @@ void board_init_f(ulong dummy)
>  	preloader_console_init();
>  
>  	i2c_early_init();
> -	ret = read_product_name(i2c_buf, I2C_BUF_SIZE);
> +	ret = read_product_name(product_name, I2C_BUF_SIZE);
>  	if (ret)
>  		log_info("Fail to detect board:%d\n", ret);
>  	else
> -		log_info("Get board name:%s\n", (char *)i2c_buf);
> +		log_info("Get board name:%s\n", product_name);
>  	pmic_init();
>  
>  	ddr_early_init();
> @@ -363,3 +360,32 @@ u32 spl_boot_device(void)
>  void spl_board_init(void)
>  {
>  }
> +
> +int board_fit_config_name_match(const char *name)
> +{
> +	char fdt_name[I2C_BUF_SIZE];
> +	int i;
> +
> +	memset(fdt_name, 0, I2C_BUF_SIZE);
> +	if (!strncmp(product_name, "k1-x_", 5)) {
> +		snprintf(fdt_name, I2C_BUF_SIZE, "%s-%s", "k1",
> +			 &product_name[5]);
> +	}

I wonder if any normalisation would be better in board_init_f() as
board_fit_config_name_match() is basically called for each possible 
board. Same for the tolower() loop below.

> +	if (fdt_name[0] == '\0') {
> +		/* set default board name */
> +		sprintf(fdt_name, "k1-musepi-pro");
> +	}

I am not sure k1-musepi-pro is a good default. The experience is that 
the Banana-Pi F3 (known as k1-x_deb1 in the vendor code) doesn't  have a 
product name defined in the TLV EEPROM and thus should probably be the 
default.

> +	for (i = 0; i < I2C_BUF_SIZE; i++) {
> +		if (fdt_name[i] == '\0')
> +			break;
> +		fdt_name[i] = tolower(fdt_name[i]);
> +	}
> +	if (!strcmp(name, fdt_name))
> +		return 0;

I am not sure that all the vendor product names will match the linux dtb 
names, so maybe just using a table to do the mapping would be better.  
Anyway, as long as the musepi-pro is the only one supported that should 
be fine.

Regards
Aurelien

-- 
Aurelien Jarno                          GPG: 4096R/1DDD8C9B
aurelien at aurel32.net                     http://aurel32.net