USB2 - USB OTG (on stm32mp157x-DK2) - does it need Vbussx or CCxx connected to function as USB mass storage in U-Boot ?

[Arnas Celkys]

I have custom stm32mp1 board and when running it as USB mass storage in U-Boot (ums 0 mmc 0), the USB driver is not detected nor by Linux nor by Windows. I am also attaching shematc comparison of eval-board VS my custom board. Also the USB-ORG is working in Engineering mode as DFU device - so I am guessing the soldering and USB pinouts are correct.

Pic.1. Custom board schematics of USB-OTG2.Pic. Eval-board schematic of USB-OTG.

[PatrickF]

Hi @Arnas Celkys​ ,

I recommend to check your DT and set OTG node according to https://wiki.st.com/stm32mpu/wiki/How_to_configure_U-Boot_for_your_board#USB_OTG_node

With ecosystem V3.x, in order to force VBUS and ID pins to be ignored, it should have:

 &usbotg_hs {
 	u-boot,force-b-session-valid;
 	u-boot,force-vbus-detection;
 	dr_mode = "peripheral";
 };

Btw, the pull-down Rd resistor on CC lines should be 5.1k (but I think 4.7k might not explain why PC did not detect your device as 10% tolerance is allowed)

Regards.

[Arnas Celkys]

Hey, @PatrickF​ ,

so I followed the 'stm32mp157c-dk2-u-boot.dtsi' include to file 'stm32mp157a-dk1-u-boot.dtsi' located in :<development-packet>/stm32mp1-openstlinux-5.10-dunfell-mp1-21-11-17/sources/arm-ostl-linux-gnueabi/u-boot-stm32mp-v2020.10-stm32mp-r2-r0/u-boot-stm32mp-v2020.10-stm32mp-r2/arch/arm/dts. Made the following changes:

-----------------------------------------------------------------------------------------------------------------------------------------------------------

&usbotg_hs {               //changes made my my, so that OTG-USB does not need Vbus, ID pins

    u-boot,force-b-session-valid;

    u-boot,force-vbus-detection;

    dr_mode = "peripheral";

};

also made a small change to 'stm32mp1.c', to printout the version of u-boot - and indicate that changes were made.

-----------------------------------------------------------------------------------------------------------------------------------------------------------

Then generated new u-boot.img and u-boot-spl.stm32, and uploaded them via USB-as-massstorage to eval-board - board performs as it used to. Then I slid the SD from eval-board to custom board, but still the same problem the masstorage is not detected after ums 0 mmc 0. Infact no USB is detected

STM32MP> ums 0 mmc 0

UMS: LUN 0, dev 0, hwpart 0, sector 0x0, count 0x7400000

Please connect USB cable.

USB cable not detected.

Command exit.

-----------------------------------------------------------------------------------------------------------------------------------------------------------

After making changes to 'stm32mp157a-dk1-u-boot.dtsi' do I need to follow any procedure other than

make stm32mp15_basic_defconfig and DEVICE_TREE=stm32mp157c-dk2 make ?

[Olivier GALLIEN]

Hi @Arnas Celkys​ ,

Your process to build and update uboot looks far from the mainstream.

You are on stm32mp1-openstlinux-5.10-dunfell-mp1-21-11-17 aka DV3.1.

Uboot-spl (aka basic) is not supported in this version. Do not use stm32mp15_basic_defconfig.

Also BSP partitioning use FIP by default.

Except if you set back to legacy mode the whole image and tsv on SDCard, you can't dd the uboot binary as such.

I recommend to use default configuration and refer to :

How to cross-compile with the Developer Package - stm32mpu

Olivier

[Arnas Celkys]

Thanks, @Community member​ ,

so just to make sure i understood my mistake fully -

1.1.after saving changes to stm32mp157a-dk1-u-boot.dtsi in <development-packet>/stm32mp1-openstlinux-5.10-dunfell-mp1-21-11-17/sources/arm-ostl-linux-gnueabi/u-boot-stm32mp-v2020.10-stm32mp-r2-r0/u-boot-stm32mp-v2020.10-stm32mp-r2/arch/arm/dts.

1.2. and optionally making changes to  'stm32mp1.c'

1.3.I should use

make -f $PWD/../Makefile.sdk all UBOOT_CONFIG=trusted UBOOT_DEFCONFIG=stm32mp15_trusted_defconfig UBOOT_BINARY=u-boot.dtb DEVICETREE="<xxxxxxx-custom>"

----- and this will generate fip-<xxxx-custom>-trusted.bin, and this .bin file is the only file that I need to upload (to fip dir on the pcb) to make changes to U-boot source and DT in U-boot, right?

Also am I correct in assuming all these changes will only affect DT in U-boot? - and therefore if I want to make changes on running OS (for example to use the USB-OTG with no ID or Vbuss pins when Os is runing), I will have to update the DT on the TF-A and/or Kernel? :)

[Olivier GALLIEN]

Hi @Arnas Celkys​ ,

Sorry for my previous comment which have confused you.

uboot-spl is supported ONLY for DDR_INTERACTIVE support.

For such please follow strictly the wiki page :

U-Boot SPL: DDR interactive mode - stm32mpu

Else to manage USB mass storage - original question - it's recommended to use default Trusted/ FIP configuration.

In this case generation of proper uboot follow guideline in

How to cross-compile with the Developer Package - stm32mpu

Else, for your point :

1.1.after saving changes to stm32mp157a-dk1-u-boot.dtsi 

It's not recommanded to change .dtsi files. All customization might be done at board level ( .dts)

cf warning in page OTG device tree configuration - stm32mpu

And read here to get full structure of DT :

STM32MP15 device tree - stm32mpu

Olivier

[Arnas Celkys]

Thanks, @Community member​ 

Could you help me out with this issue:

First I would like to ended DDR interactive mode using "stm32mp15_trusted_defconfig" rather then "stm32mp15_basic_defconfig".

So I proceed like so :

>>make clean && make mrproper

<<

   CLEAN  scripts/basic

   CLEAN  scripts/kconfig

   CLEAN  .config

>>make stm32mp15_trusted_defconfig

<<

   #

   # configuration written to .config

   #

>>DEVICE_TREE=stm32mp157c-dk2 DDR_INTERACTIVE=1 make

<<

 CC     drivers/core/simple-bus.o

 CC     drivers/core/dump.o

 CC     drivers/core/regmap.o

 CC     drivers/core/syscon-uclass.o

 CC     drivers/core/of_extra.o

 CC     drivers/core/ofnode.o

drivers/core/ofnode.c:239:6: error: redefinition of ‘ofnode_is_enabled’

 239 | bool ofnode_is_enabled(ofnode node)

     |     ^~~~~~~~~~~~~~~~~

drivers/core/ofnode.c:229:6: note: previous definition of ‘ofnode_is_enabled’ was here

 229 | bool ofnode_is_enabled(ofnode node)

     |     ^~~~~~~~~~~~~~~~~

make[2]: *** [scripts/Makefile.build:266: drivers/core/ofnode.o] Error 1

make[1]: *** [scripts/Makefile.build:419: drivers/core] Error 2

make: *** [Makefile:1780: drivers] Error 2

and get this wierd error of double definition of 'ofnode_is_enabled', was my 'ofnode.c' file corrupted? Should I re-download it?

----------------------------------------------------------------------------------------------------------

Also after successful "DEVICE_TREE=stm32mp157c-dk2 DDR_INTERACTIVE=1 make", should I upload u-boot-spl.stm32 and u-boot.img to /fip + /fsbl1 + /fsbl2 ? Or should I follow different steps?

It does not define it in U-Boot SPL: DDR interactive mode - stm32mpu

----------------------------------------------------------------------------------------------------------

[Olivier GALLIEN]

Hi @Arnas Celkys​ 

DDR_INTERACTIVE=1 is only supported in "stm32mp15_basic_defconfig" not with "stm32mp15_trusted_defconfig" .

----------------------------------------------------------------------------------------------------------

Also after successful "DEVICE_TREE=stm32mp157c-dk2 DDR_INTERACTIVE=1 make", should I upload u-boot-spl.stm32 and u-boot.img to /fip + /fsbl1 + /fsbl2 ? Or should I follow different steps?

It does not define it in U-Boot SPL: DDR interactive mode - stm32mpu

----------------------------------------------------------------------------------------------------------

Yes it is. See chapter 3 Load the SPL into embedded RAM

You can ignore FIP in this case. Uboot spl will come in fsbl partition.

Olivier

[Arnas Celkys]

oh yeah, @Community member​  it does i missed it. So to enter DDR interactive I have to use the U-Boot SPL - got it now.

And getting back to the original questions I should be using the FIP - to make any other changes related to DT or software of U-boot.

--------------------------------------------------------------------------------------------------------------------------

I am getting the same  'ofnode.c' redefinition error when trying the U-Boot SPL - perhaps I have corrupted the uboot-source directory somehow - what else could cause it?

[Olivier GALLIEN]

Not clue for this specific issue.

But might be probably good to restart for a fresh installation of Uboot source directory.

Since you have made several trials, even clean and mrproper may not be enough to put back all things right.

Olivier