STM32MP157 – SD card works in ROM/TF-A (FSBL) but fails in U-Boot with CMD55 retries

[Najib Kassab]

Problem description

I have a custom board based on STM32MP157FAA.
The SD card boots perfectly in:

• ROM boot

• TF-A BL2 (FSBL)

But once control reaches U-Boot, the SD card becomes completely unusable and U-Boot shows continuous:

stm32_sdmmc2 mmc@58005000: cmd 55 failed, retrying ...

U-Boot cannot read environment, cannot rescan the card, and reports “No block device”.

This is very confusing because the same SD card and hardware clearly work during ROM and TF-A stages.

---

Boot log

TF-A / BL2 (working)

INFO: Using SDMMC INFO: Instance 1 INFO: Boot used partition fsbl1 ... INFO: BL2: Loading image id 5 NOTICE: BL2: Booting BL32

No SD errors at all.

U-Boot (failing)

U-Boot 2023.10-stm32mp-r2
MMC: STM32 SD/MMC: 0, STM32 SD/MMC: 1
Loading Environment from MMC...
stm32_sdmmc2 mmc@58005000: cmd 55 failed, retrying ...
*** Warning - No block device, using default environment ...
Saving Environment to MMC... No block device

Also in U-Boot:

Clocks:
- MPU : 0 MHz
- MCU : 0 MHz
- AXI : 0 MHz
- PER : 0 MHz
- DDR : 0 MHz

---

Hardware notes

• No external level shifter

• No SD direction pin used

• SD is wired directly to SDMMC1

• Fixed 3.3V I/O

• No card detect pin

---

Linux / TF-A SDMMC1 DTS (works)

&sdmmc1 {
status = "okay";
bus-width = <4>;
no-1-8-v;
vmmc-supply = <&vdd_sd>;
disable-wp;
broken-cd; };

PMIC regulator:

vdd_sd:ldo5{
regulator-name = "vdd_sd";
regulator-min-microvolt = <2900000>;
regulator-max-microvolt = <2900000>;
regulator-boot-on;
};

No special SDMMC settings exist in the *-u-boot.dtsi.

You can check the included log for more infomations

[PatrickF]

Hi,

---

@Najib Kassab wrote:

Also in U-Boot:

Clocks:
- MPU : 0 MHz
- MCU : 0 MHz
- AXI : 0 MHz
- PER : 0 MHz
- DDR : 0 MHz

---

---

that's very strange to have all clocks stated at 0 with uBoot. Maybe root cause of failing SDMMC access.

As far as I know, BootROM and TF-A uses default SDMMC settings while uBoot reconfigure SDMMC according to Device Tree. Maybe an issue with OpTee which manage the clocks after TF-A.

Regards.

[Najib Kassab]

@PatrickF 

So what do you suggest to do now , if the problem is with optee?

[Najib Kassab]

@PatrickF 

Also I have question, I am generating Devixe tree using stm32cubemx, the problem is whenever I made and change on GUI it does not affect the device tree whatever I do it always generates the same device tree.

Sometimes it works fine, but most of the time it does not work as expected.

[PatrickF]

Hi  @Najib Kassab 

unfortunately, I'm not very SW savvy. 
What I know is that CubeMx generates only very partial Device Tree which should be tailored to your HW.
Usually using existing working Device Tree from board like Evaluation or Discovery together with using wiki guidelines for each IPs.

https://wiki.st.com/stm32mpu/wiki/STM32_MPU_device_tree#STM32CubeMX_generated_device_tree

Regards.