AM3356: SDCard bootup without EEPROM access

Hi Miyazaki-san,

Takayuki Miyazaki said:

My customer is trying to do boot-up without reading EEPROM now. They are developing their system without EEPROM, However, first of all, They are evaluating u-boot function (no reading EEPROM) with AM3335x GPEVM currently. But, customer does not success it with GPEVM yet.

Some options to try out

a) Not read EEPROM by removing do_board_detect() call as listed below in “arch/arm/mach-omap2/am33xx/board.c”

early_system_init (){
…
// do_board_detect()
…
}

b) change code as listed below in “board/ti/am335x/board.h” to make board_is_gp_evm() call return "true"

static inline int board_is_gp_evm(void)
{
        return 1;
// return board_ti_is("A33515BB");
}

Best,

-Hong

Hi Hong,

Thanks for your advice on this. I'll share this with customer, first of all. Also, I'd like to share the modifications which customer has done, by off-line.

best regards, Miyazaki

Hi Hong,

Customer tried to do a) modification and b) modification, however, it still failed. Could you share any other solution, please ?

I shared summary-document of customer's modification previously, would you see any issue for customer's procedure?

It will be appreciated if you will share your other advice/comments on this.

best regards, Miyazaki

Hello Hong,

Can I have any update for this, please?

best regards,

Miyazaki

Hi Miyazaki,

Can we add “#define DEBUG” at the top of “/lib/initcall.c”?

This would add full “initcall xxxxxxxx” debug dump in u-boot log to have more visibility on where u-boot boot fails.

For u-boot (proper) boot flow:

- initcall_run_list() (/lib/initcall.c) is called by boot_init_f() (/common/board_f.c), which executes a sequence of initialization functions as defined in init_sequence_f[] = {…};

Similarly initcall_run_list() is also called in boot_init_r() (/common/board_r.c), which runs a sequence of initialization functions as defined in init_sequence_r[] = {…};

Best,

-Hong

Hi Hong,

I received the debug.log with “#define DEBUG” from my customer. I attached this (initcall_DEBUG.zip). I believe it will be helpful to do analysis.

Best regards, Miyazaki

HI Hong,

Thank you so much for your hard work. Our customer is developing their system with AM3356/WL1837MOD, however, customer is verifying this u-boot (non-EEPROM access) with AM335xGP-EVM, so, please let me confirm if customer connects AM335xGPEVM to WiFI board (WL1837MODCOM8I) in this verification. 

Also, please allow me to confirm  "MMC dtb mis-configuration". according to debug-log, there is possibility that u-boot stopped during SDIO-access . Is that correct?

At first,  I’d like to share your comments with customer. I really appreciate.

Best regards, Miyazaki

Hi Miyazaki-san,

yes, mmc3 node needs to be disabled when booting on TI EVM w/o SDIO wifi module connected on MMC3
I checked the decoded dtb after running "dtc -I dtb am335x-evm.dtb > am335x-evm.dtb.txt".
Note that mmc3 node is disabled as shown below:

		mmc@47810000 {
			compatible = "ti,omap4-hsmmc";
			ti,hwmods = "mmc3";
			ti,needs-special-reset;
			interrupts = <0x1d>;
			reg = <0x47810000 0x1000>;
			status = "disabled";
		...
		}

Best,

-Hong

am335x-evm.dtb.txt

Hi Hong,

Sorry for my late input. The begin of this week is national holiday in Japan. I received customer's feedback. Customer is verifying this u-boot with AM335x-GP-EVM+WL1835MODCOM8B. therefore, I don't believe that they should disable mmc3 node. Or, should I request customer to disable mmc3 node?

It will be appreciated if you will share more advice on this none-EEPROM boot.

Best regards,

Miyazaki

Hi Hong,

Thanks for your clarification. I had misunderstood. It. I will request customer to disable the status of mmc3 node.

After that, I’ll wait for customer’s feedback.

Best regards,

Miyazaki

Hi Hong,

Customer was able to confirm this none-EEPROM booting is working fine. 

Thanks for your kind help !

Best regards,

Miyazaki