SK-AM62B-P1: How to change defconfig files for SPI NAND boot

Hi Koichiro ,

SK-AM62B-P1 Does not support SPI-NAND FLASH. It support SPI-NOR FLASH only

Hi Dilna,

Dilna K said:

SK-AM62B-P1 Does not support SPI-NAND FLASH. It support SPI-NOR FLASH only

Yes. They knew it. They replace SPI-NOR with SPI-NAND by themselves.

Thanks and regards,
Koichiro Tashiro

Hi Koichiro 

Please make the below changes for am62x_evm_a53_defconfig,am62x_evm_r5_defconfig and try

(These are the defonfig changes for am62a which support spi-nand)

CONFIG_SPL_DM_SPI=y
CONFIG_SPL_SPI=y
CONFIG_SPL_NAND_SPI_SUPPORT=y
CONFIG_SPL_DM_SPI_FLASH=y
CONFIG_MTD=y
CONFIG_DM_MTD=y
CONFIG_MTD_SPI_NAND=y
CONFIG_SPI=y
CONFIG_DM_SPI=y
CONFIG_CADENCE_QSPI=y
CONFIG_CADENCE_QSPI_PHY=y
CONFIG_DM_SPI_FLASH=y

And also make modifications in am62b-p1 dts for spi-nand. for this you can refer the am62a spi-nand devicetree changes.

https://software-dl.ti.com/processor-sdk-linux/esd/AM62AX/latest/exports/docs/linux/Foundational_Components/U-Boot/UG-QSPI.html

Hi Dilna,

The customer tried these changes in configuration and built u-boot. He also confirmed SPI NAND access from u-boot worked by mtd command.
Then he made dts modifications as well.
And "tiboot3.bin", "tispl.bin" and "u-boot.ilmg" were written in SPI NAND.
But booting from SPI NAND failed with below message.

U-Boot SPL 2024.04-dirty (Jul 29 2025 - 00:52:39 +0000) SYSFW ABI: 4.0 (firmware rev 0x000a '10.0.8--v10.00.08 (Fiery Fox)') SPL initial stack usage: 13392 bytes Trying to boot from SPINAND mtd_probe_uclass_mtd_devs 0 MTD device spi-nand not found, ret -1

Hi Koichiro,

Is the issue resolved?

If not, hope ypu are loading binaries from SD card.  Boot via SD card boot and stop at U-boot prompt before proceeding.

1. please confirm you are using the below commands.

=> mtd list
=> fatload mmc 1 ${loadaddr} tiboot3.bin
=> mtd write spi-nand0 $loadaddr 0x0 $filesize
=> fatload mmc 1 ${loadaddr} tispl.bin
=> mtd write spi-nand0 $loadaddr 0x80000 $filesize
=> fatload mmc 1 ${loadaddr} u-boot.img
=> mtd write spi-nand0 $loadaddr 0x280000 $filesize

2. And also please confirm whether you have changed boot mode pins to boot with Serial NAND boot mode and reboot EVM.

Hi Dilna,

Now the customer is able to load U-Boot SPL from SPI NAND, but the code is looping inside SPL forever.
What the customer observes are;
- mtd_read is done inside SPL to load u-boot image from NAND.
- In "drivers/spi/cadence_qspi_apb.c", function "cadence_qspi_apb_indirect_read_execute", 
the "while (bytes_to_read != 0)" loop cannot be exited because return value  
bytes_to_read = cadence_qspi_get_rd_sram_level(priv); is always "129".

What is a reason the return value is always "129"?
Any suggestion to solve this?

Thanks and regards,
Koichiro Tashiro

Hi Koichiro,

I am redirecting your queries to  OSPI/QSPI expert. since this is an error from driver side.

Regards,

Dilna K

Hi Koichiro,

Koichiro Tashiro said:

They replace SPI-NOR with SPI-NAND by themselves.

1/. any HW modification when replacing SPI-NOR with SPI-NAND on SK-AM62B-P1?
2/. do we have the SPI-NAND datasheet?
3/. what is BOOTMODE[15:0] when booting from serial nand?

FYI. the serial nand W35N01JWTBAG is on TI SK-AM62-LP & SK-AM62A boards
Best,
-Hong

Hi Hong,

Hong Guan64 said:

1/. any HW modification when replacing SPI-NOR with SPI-NAND on SK-AM62B-P1?

I sent HW modification details by e-mail.

Hong Guan64 said:

2/. do we have the SPI-NAND datasheet?

You can download the datasheet from below.

https://urldefense.com/v3/__https://www.winbond.com/hq/support/documentation/levelOne.jsp?__locale=ja&DocNo=DA00-W25M02GV.1__;!!G3vK!TjiwHUfVwtbTFhySRKHo58wni9hnu_RnvrL2UdqgMLJX67PcxBgbi3EM0l_TzKimDXxYupI7BzJIxFFnnemmw2RgwTMlOQ$

Hong Guan64 said:

3/. what is BOOTMODE[15:0] when booting from serial nand?

BOOT[15:00] =  0x0683

Thanks and regards,
Koichiro Tashiro

Hi Koichiro,
1/. do we have the detailed steps (log file would help) on how the customer flashed the serial nand on the customer board?
2/. have the customer verified the serial nand content after flashing?
3/. do we have the full boot log when booting from serial nand?

FYI. this is the FAQ on how to flash/boot from serial nand on SK-AM62-LP board
[FAQ] How to flash and boot u-boot from serial-NAND on AM62x/AM62Ax 
Best,
-Hong

Hi Hong,

Hong Guan64 said:

1/. do we have the detailed steps (log file would help) on how the customer flashed the serial nand on the customer board?

Customer used SD card boot and used below commands in u-boot.

=> fatload mmc 1 ${loadaddr} tiboot3.bin
313854 bytes read in 54 ms (5.5 MiB/s)
=> mtd write spi-nand0 $loadaddr 0x0 $filesize
Size not on a page boundary (0x800), rounding to 0x4d000
Writing 315392 byte(s) (154 page(s)) at offset 0x00000000

=> fatload mmc 1 ${loadaddr} tispl.bin
1195387 bytes read in 100 ms (11.4 MiB/s)
=> mtd write spi-nand0 $loadaddr 0x80000 $filesize
Size not on a page boundary (0x800), rounding to 0x124000
Writing 1196032 byte(s) (584 page(s)) at offset 0x00080000

=> fatload mmc 1 ${loadaddr} u-boot.img
1449851 bytes read in 116 ms (11.9 MiB/s)
=> mtd write spi-nand0 $loadaddr 0x280000 $filesize
Size not on a page boundary (0x800), rounding to 0x162000
Writing 1449984 byte(s) (708 page(s)) at offset 0x00280000

Hong Guan64 said:

2/. have the customer verified the serial nand content after flashing?

Yes. Here is log at verify.

=> printenv loadaddr
loadaddr=0x82000000

=> fatload mmc 1 ${loadaddr} tiboot3.bin
313854 bytes read in 54 ms (5.5 MiB/s)
=> mtd read spi-nand0 0x80000000 0x0 $filesize
Size not on a page boundary (0x800), rounding to 0x4d000
Reading 315392 byte(s) (154 page(s)) at offset 0x00000000
=> cmp.b ${loadaddr} 0x80000000 $filesize
Total of 313854 byte(s) were the same

=> fatload mmc 1 ${loadaddr} tispl.bin
1195387 bytes read in 100 ms (11.4 MiB/s)
=> mtd read spi-nand0 0x80000000 0x80000 $filesize
Size not on a page boundary (0x800), rounding to 0x124000
Reading 1196032 byte(s) (584 page(s)) at offset 0x00080000
=> cmp.b ${loadaddr} 0x80000000 $filesize
Total of 1195387 byte(s) were the same

=> fatload mmc 1 ${loadaddr} u-boot.img
1449851 bytes read in 116 ms (11.9 MiB/s)
=> mtd read spi-nand0 0x80000000 0x280000 $filesize
Size not on a page boundary (0x800), rounding to 0x162000
Reading 1449984 byte(s) (708 page(s)) at offset 0x00280000
=> cmp.b ${loadaddr} 0x80000000 $filesize
Total of 1449851 byte(s) were the same

Hong Guan64 said:

3/. do we have the full boot log when booting from serial nand?

Here is the boot log.

U-Boot SPL 2024.04-dirty (Aug 07 2025 - 08:27:08 +0000)
SYSFW ABI: 4.0 (firmware rev 0x000a '10.0.8--v10.00.08 (Fiery Fox)')
SPL initial stack usage: 13392 bytes
Trying to boot from SPINAND

Hi Koichiro,
Thanks for sharting the debug logs.
There were changes on the WB serial-nand driver introduced in Linux SDK 11.x.
Is it possible for your customer to test with the latest Linux SDK 11.01.05.03
https://www.ti.com/tool/download/PROCESSOR-SDK-LINUX-AM62X/11.01.05.03
Best,
-Hong

Hi Hong,

Hong Guan64 said:

Is it possible for your customer to test with the latest Linux SDK 11.01.05.03

It is decided to use SDK10.x for their product, so upgrading SDK is not possible.
Could you let me know the solution with SDK10.x?

Thanks and regards,
Koichiro Tashiro

Hi Koichiro,
1/. here is the u-boot serial-nand DT node, where the AM62-LP-SK board has a 128 MB serial-nand flash W35N01JW (OSPI). The customer need to adapt the DT node for the serial-nand device on the customer board (QSPI).
https://git.ti.com/cgit/ti-u-boot/ti-u-boot/commit/arch/arm/dts/k3-am62-lp-sk.dts?h=10.00.07&id=6c2704e46073349030956d3cea04e63e90aeafb6
2/. The customer need to check if the WB serial-nand device is configured correctly
https://git.ti.com/cgit/ti-u-boot/ti-u-boot/tree/drivers/mtd/nand/spi/winbond.c?h=10.00.07#n188
Best,
-Hong