STM32 MCUs Security

HiI am implementing the ECDSA-P256 signing for a payload.I have checked the example code  https://github.com/STMicroelectronics/STM32CubeU3/tree/main/Projects/NUCLEO-U385RG-Q/Examples/CCB/CCB_Protected_ECDSA_SignatureThe example code shows that the software has an access to the private-key. Would it be possible to generate a private key using RNG which software/processor can not see and which can be used by the CCB to compute a signature. The same key can be used to get public-key.  can I use this API   HAL_CCB_ECDSA_GenerateWrapPrivateKey(). Would this key be persistence across reboot or do I need to store the wrapped key? 

Hi teams,CPU: STM32U0Current RDP level: Level 1OEM1 RDP lock: Enabled.I had set the 128bit OEM1 key to the OEM key registers  [FLASH_OEM1KEYWyR ]accordingly, and the written OEM1 key values was verified using their CRC values by reading  OEM1KEYCRC register.Now I am trying to make the regression from level 1 to level 0 according to the user manual description [section: 3.5.1 FLASH read protection (RDP)]. But the operation was failed using my source code. Which is attached below#define STM32U0_DBGMCU_BASE 0x40015800 // Accessible to the software via the APB access port AP0#define STM32U0_DBGMCU_DBG_AUTH_HOST (STM32U0_DBGMCU_BASE + 0x100) // Debug authentication mailbox host register Could you please clarify the following doubts,1. The unlock sequence says, debug access port 1 has to be used for programming the 128-bit OEM1 key in the DBGMCU_DBG_AUTH_HOST register,but when I checked the  DBGMCU_DBG_AUTH_HOST register base, it sa

Hello,I am trying to get the OEMiROT example projects running on STM32H563 from "STM32CubeH5\Projects\NUCLEO-H563ZI\Applications\ROT" and this page : https://wiki.st.com/stm32mcu/wiki/Security:How_to_start_with_OEMiRoT_on_STM32H573_and_563%E2%80%93TrustZone_enabled Provisioning works well, unfortunately, I'm blocked on the firmware update step through the internal bootloader. After the message 'Start download with STM32CubeProgrammer through supported interfaces (USART/SPI/I2C/USB)', when I connect CubeProgrammer with uart connection, the bootloader seems to work correctly as I received:Chip ID:0x484 Bootloader protocol version : 4.0Then, CubeProgrammer freezes for 2 mins and the device is disconnected with this message log: 09:54:32 : Error: RDP is set to level 1 or target is held under reset, try using Read UnProtect (-rdu) command in order to remove the Read Out Protection. 09:54:32 : UPLOADING OPTION BYTES DATA ... 09:54:32 : Bank : 0x00 09:54:32 : Address : 0x4

Device detailsMPU: STM32H753AII6Connection:Probe: J-Link V11.00 / SW version V7.98c STM toolset: STM32 Cube Programmer / Trusted Package Creator v2.20.0SFI programming fails with error:Error: SECURITY State FailError: Invalid state after SFI install !Error: RSS internal system error or License is invalid ! I also tried against the STM32H757I-EVAL board, which has a compatible CPU, with the same result.-----HSM provisioningHSM provisioningFor Personalization Data I chose the group with the H753 in it, and from the 4 options in dropdown list the combination of "SFI" and Device ID "45002001" which is how the STM32H753 identifies.SFI creationFor the simplified example, there is only one section of code starting from default address 0x08000000.Settings for RAM size and continuation token are default, and seem to make sense.Option bytes are, except for RDP level = 0xBB (level 1), identical to the defaults after mass erase.SFI creationSFI flashingSFI flashingComplete log:15:25:

Do all the STM32WL series MCUs support STSAFE ? If yes, how it can be done? 

Hi,is there a possibility to do the complete chain to product state "closed" by non secure user code ?I have a custom bootloader which usually did a set of the old RDP Level.Now, with a STM32H57 device we have to use the new product state. I'm able to set the state to provisioning. But when i use RSSLIB_PFUNC->NSC.DataProvisioning i end up in a hard fault. I tried to set the provisioning data by ST Cube programmer later on which succeeded. But unfortunately i can not regress to open because the password is wrong. Any hints about that ? Thanks in advance.

Does ST offer an implementation of hardware acceleration of mbedtls-2021?I found this:STM32CubeL5\Projects\STM32L562E-DK\Applications\TFM\TFM_Appli\Secure\Src\ecdsa_alt.cSTM32CubeL5\Projects\STM32L562E-DK\Applications\TFM\TFM_Appli\Secure\Src\ecdsa_curves_alt.cfrom https://github.com/STMicroelectronics/STM32CubeL5.gitThis is excactly what I want (a bridge between mbedtls and the PKA_HAL driver), but it's for a 2019 version of mbedtls, and the mbedtls folks since broke the interface, so it doesn't compile. (and it doesn't look like an quick/easy fix)Is there a newer version somewhere?/Troels

I try to understand in details the way the Hash peripheral works on the STM32L552. I need to make the HASH SHA-256 with 224 blocks of 512 bits. The Hash works perfectly until the block 223. Then I enter 16 values in the FIFO (in my code below I have a loop for 15 because I entered previously another value). Then I launch the final digest. I expect the automatic digest of block 224 and the automatic padding in a block 225 to the final hash value. But I always have the wrong value in the final step. Any idea about what I'm missing? 

Hello.I'm working with STM32H573.I have a question related with the use of STSAFE-A110 for TLS. I have to authenticate also the client.I've succeded stablishing TLS mutual authentication with HTTPS server and TCP server  (different applications, not related) using this type of functions from NetxDuo with the correct data and test certificates:/* TCP Server */ nx_secure_tls_session_create(); nx_secure_tls_ecc_initialize(); memset(&certificate, 0, sizeof(certificate)); // Server Local certificate nx_secure_x509_certificate_initialize(); memset(&certificate_, 0, sizeof(certificate_)); // Client certificate nx_secure_x509_certificate_initialize(); nx_secure_tls_local_certificate_add(); nx_secure_tls_trusted_certificate_add(); nx_secure_tls_session_start(); /* HTTPS */ memset(&certificate, 0, sizeof(certificate)); // Server Local certificate nx_secure_x509_certificate_initialize(); memset(&certificate_, 0, sizeof(certificate_)); // Client certificate nx_secure_x509_

I am using mbedtls module to verify the signature of firmware using RSA and its giving MBEDTLS_ERR_PK_INVALID_PUBKEY after parsing the public key. I am using OpenSSL for generating public and private keys the public key is generated using openssl genpkey -algorithm RSA -out private_key.pem -pkeyopt rsa_keygen_bits:2048andopenssl rsa -pubout -in private_key.pem -outform DER -out public_key.derin der format and then converted to a .h file using python -c "data=open('public_key.der','rb').read(); print('const unsigned char public_key_der[] = {'); print(', '.join(f'0x{b:02x}' for b in data), end=''); print('};'); print(f'const unsigned int public_key_der_len = {len(data)};');" > public_key.hIn the mbedtls_config.h file I have enabled the following things#define MBEDTLS_PKCS1_V15#define MBEDTLS_PK_C#define MBEDTLS_PK_PARSE_C#define MBEDTLS_RSA_C#define MBEDTLS_BIGNUM_C#define MBEDTLS_ASN1_PARSE_C#define MBEDTLS_OID_CHere is the main function I am using to parse the public

Hi, I am using ADC1 on stm32H563ZI nucleo board with trustzone enabled, I am not getting proper adc values (getting adc count of 822). When trust zone is disabled I am getting proper value of around 4095 with input of 3.3Volt .I am attaching the project.Please help with this issue.

I’m using STM32WL5MOCH6STR and want to add STSAFE-A110 for security in a LoRaWAN application.I want to know:How to use STSAFE-A110 for authentication, key storage, encryption and secure boot & secure firmware update.Which software packages are needed to implement this security feature(like X-CUBE-STSAFE, X-CUBE-LRWAN, or others)?Is there any example or application note for STM32WL + STSAFE-A110 integration?Kindly let me know if there are any ways or recommended methods to achieve security with help of STSAFE-A110 with STM32WL MCUs.Thanks in advance!

Hello STM32 Community,I’ve successfully built a Zephyr RTOS blinky application for the NUCLEO-H533RE board. Flashing the zephyr.hex using west flash or STM32CubeProgrammer works perfectly — the LED blinks and the serial terminal prints the expected status messages.To enhance security, I’m now trying to encrypt and sign the firmware using STM32TrustedPackageCreator and provision the board using STiROT. I followed the STM32CubeH5 GitHub examples and used the STiROT_Code_Init_Image.xml file, modifying it to point to my zephyr.bin. Provisioning was successful, and the board state was set to PROVISIONED.However, after flashing the generated zephyr_enc_sign.hex, the board does not blink, and the serial terminal remains silent — indicating the firmware is not executing.Here’s what I’ve done:Used STiROT/Image/STiROT_Code_Init_Image.xml and modified paths to point to zephyr.bin.Generated the encrypted and signed image using STM32TrustedPackageCreator.Successfully provisioned the board and set i

Hi,I come to you about the key provisioning process for the iROT in the STM32H5.Using TrustedPackageCreator, the keys provided in the file STiROT_Config.xml for the Secure Boot (field <Authentication key>) are both the private and public keys of the .pem file. However, only the public key is needed, and will be embedded in the device. This method of provisioning is limiting for us for two reasons:- on the one hand, it doesn't allow to subcontract the provisioning task, as we don't want to give the private key to the subcontractor- secondly, the private key may be generated by an HSM and cannot leave the vault. So, is there a way to provision only the public key at this stage? Best regards,Christine 

Hello,I'm using STM32F439 with LWIP, SNMPv3, and mbedTLS, without FreeRTOS. I've encountered a performance issue when using the hardware acceleration for AES and SHA1 provided by ST (aes_alt.c, sha1_alt.c).Specifically, when decoding SNMPv3 messages, the software implementation of AES128 + SHA1 from mbedTLS is significantly faster than the hardware-accelerated version. I measured the execution time of snmp_receive() using a timer:- mbedTLS software AES128 + SHA1: ~800 µs- mbedTLS + ST-provided hardware AES (aes_alt.c) + SHA1: ~2.4 msThis is a 3x slowdown when using the hardware path.Can anyone explain why the hardware implementation is slower? Is this expected behavior due to HAL overhead, lack of DMA, or something else?Thanks in advance for any insights!Best regards, Piotr

Hi everyone,We’re developing an IoT node for agricultural applications and plan to make it compliant with certification requirements (CE, CRA, RED).We’re using the X-CUBE-AZURE package as our starting point and want to use TF-M (Trusted Firmware-M) for security.We’re new to cybersecurity and TrustZone, so I’d like to ask the community:Which peripherals and parts of the application stack should be implemented on the secure side of TF-M in this case?In our design we have various interfaces: UART, I2C, SPI, Wi-Fi module, Digital I/O, Analog I/O, CANBus, Modbus, and SDI-12.Which of these should be configured as secure or non-secure, and which parts of the Azure IoT stack are typically expected to run in the secure vs non-secure world?Any practical examples or recommendations based on your experience with X-CUBE-AZURE and TF-M would be very helpful. Thanks in advance!

I’m working with a NUCLEO-H533RE board (STM32H5 series, TrustZone enabled) and using ST-LINK for debugging.Here is the issue:Debug Authentication succeeds (certificate-based).Immediately after, the tool disconnects and shows:Error: Unable to get core ID Error: Cannot connect to access port 1! If you are trying to connect to a device with TrustZone enabled please try to connect with HotPlug mode. If you are trying to connect to a device which supports Debug Authentication with certificate and password, please open your device using it.Log shows:SDMAuthenticate: Authentication successful Debug Authentication Success Disconnected from device. ST-LINK FW: V3J16M8 Board: NUCLEO-H533RE Voltage: 3.28VST-Link configuration:Port: SWDMode: HotPlugAccess Port: 0Reset Mode: Software ResetSWD Frequency: 8000 kHzShared: DisabledI've tried:HotPlug Mode as recommended for TrustZone.Debug Authentication with valid certificate and key.ST-Link firmware is up to date.Still, after authentication, I cannot

 Hello,My device contains both a bootloader and an application. I’m currently trying to protect the bootloader using PCROP on the STM32G0B1.As a first approach, I attempted to configure PCROP using STM32CubeProgrammer. However, this failed — once PCROP was activated for the bootloader region, the bootloader was unexpectedly erased.Here’s what I did:Flashed the bootloader to address range 0x08000000–0x0800BFFF.Activated PCROP by setting the start and end addresses for the bootloader region (see screenshot below). After this, the bootloader was erased.Even when I deactivated PCROP_END, the result remained the same — the bootloader was still erased.I would greatly appreciate any insights into why this is happening and how to properly activate PCROP without losing the bootloader.Best regards,Alexander     

Hi ST team,I am using STM32U5G9 as secondary MCU. It has simple firmware upgrade feature. It would be reset by main MCU and enter system memory bootloader using Boot0 pin. The system memory bootloader would receive upgrade image (application) from main MCU by USART and write the image into internal flash. (AN3155 and AN2606)This method is simple but it does not have any sercrity feature. I would like to ask if I could add MCUBoot and secure boot into this method with minimum effort?1. System memory bootloader would receive signed image (application) from main MCU and write it into non-secure area.2. Exit system memory bootloader and enter secure boot3. Secure boot verifies the authenticity and integrity of the application.4. Execute the application.Is this possible? 

Hi.During production, we use USB to test the board. Finally, the MCU is supposed to be password-protected against flash reads. Unfortunately, it turns out that as long as the USB cable is connected, it's impossible to program the OBK. Segger Device Provisioner returns an "Error writing OBK". Using STLink and CubeProgrammer also doesn't work. After disconnecting the USB, changing the state to Closed works. BOOT0 is connected to ground. Is there a solution?Artur

Hi, I have a doubt. I made some changes in the secure bootloader to achieve certain functionality, but I am unsure whether the code in the secure bootloader is developed by ST (such as drivers or APIs), or if it is intended for user modifications. Specifically, I modified the sfu_interface_crypto_scheme.c file to achieve my functionality.As far as I understand, apart from the Secure Engine project and the we can't modify Bootloader project HAL and LL APIs, CMSIS APIs, and middleware like Secure Engine, KMS, and Crypto. Please correct me if I am wrong.For reference, I am using the STM32WLE5CC controller.

Hi Team, I am working on project where we are implementing Secure upgrade feature using SBSFU package, we followed the steps as mentioned by ST community and to integrate SBSFU in our project. we have two modes Debug mode and Release mode.in Debug Mode: below Protection flags are disable#define SFU_WRP_PROTECT_ENABLE#define SFU_RDP_PROTECT_ENABLE#define SFU_TAMPER_PROTECT_ENABLE #define SFU_DAP_PROTECT_ENABLEand bootloader and application are working fine. we can see the software upgrade is happening. In Release mode:all above Protection flags are enable. bootloader is not working I think it is not coming up properly.I can see through STMCubeprogrammer the protection changes applied to flash.worst part is, I couldn't debug bootloader code.it would be great somebody help me in building SBSFU in Release mode.Thanks in Advance. 

Hello,In a project with STM32 using SBSFU in a U575 microcontroller, why can a local variable of type structure within a function cause failures or unexpected behaviour in the secure zone, while a global variable of the same type works correctly?

Section 6.1 of AN5056 (for X-CUBE-SBSFU) states that the projects are delivered with the highest level of compiler optimizations turned on for size aspects.If the Optimization Level was set to 'None', then how much larger is the generated code expected to grow for the mapped regions (SE, SBSFU, etc.) excluding the image regions? (% increase in general)Which of the regions will be impacted the most?Update: To aid debug via IAR EWARM, I prefer to develop without any optimizations enabled, understanding that the code output size will be larger. I have no issues if the mappings are allowed to be adjusted to accommodate the larger code outputs.

Hello , I am trying to calculate a MAC using `cmox_mac_compute()` from this lib on an STM32L562 in the Secure code.The code looks like this :auto r = cmox_initialize(NULL); /*default template with stm32l5xx.h included*/ auto ret = cmox_mac_compute(CMOX_CMAC_AES_ALGO, in, in_s, key, 16, NULL, 0, out, 16, NULL);This exact code works when run on non secure mode if everything is set as non secure in stm32CubeMX.I tried to debug the library to understand what was wrong and found out that the issue comes from the fact that cmox_mac_compute tries to access my crc peripheral from 0x40023000 instead of 0x50023000 (CRC non secure and secure addresses).I am also almost sure that my GTZ and SAU are correctly set because i was able to write at address 0x50023000 and not at 0x40023000 when running from Secure code.I also have :CRC_BASE_S => 0x50023000 CRC_BASE_NS => 0x40023000 CRC_BASE => 0x50023000So this looks fine. I also tried to forcefuly set CRC_BASE_NS to