This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AM2434: Configuring MPU Settings but Instance limit has been reached

Part Number: AM2434
Other Parts Discussed in Thread: SYSCONFIG, AM2634,

Tool/software:

Hi Team,

My customer wants to configure MPU regions, but it says that the Instance limit has been reached. when they try to add more.

Is their understanding that since MPU is at the L1 cache level, they should be able to configure 16 regions per core? 

If it is possible to configure, could you tell me how?

If not, what are the restrictions on this?

Best regards,

Mari Tsunoda

  • Greetings Mari,

    You are correct, each individual R5 core can have a maximum of 16 MPU regions. Based on the screenshot, are you saying this tool is limiting the customer to using only 6 regions instead of 16?

    Sincerely,

    Lucas

  • Hi Lucas,

    Yes, sysconfig is limiting the configurable region to only 6.

    Best regards,

    Mari

  • Hi Mari,

    If it is possible to configure, could you tell me how?

    I can see you have added 26 instances out of 16.

    Can you please share the example.syscfg file with us?

    Is this a core specific project or System project?

    Which version of MCU+SDK are you using?

    Regards,

    Tushar

  • Hi Tushar,

    I realize that this is for a multi-core system project. I see there is a similar issue on the AM2634 and has been filed as a bug.

    Does this also apply to AM2434? If so, is it resolved? I can send you the project zip file separately via email. 

    (1) AM2634: SYSCONFIG: MPU Region Denominator Is Incorrect For A Multicore Project - Arm-based microcontrollers forum - Arm-based microcontrollers - TI E2E support forums

    Best regards,

    Mari Tsunoda

  • Hello Mari,

    Tushar is out of office.

    The above bug is valid for AM64X/AM243 devices.

    Each core supports 16 MPU regions and when we open the multicore project, then the tool is supposed to show the 16*4 MPU region settings for all 4 R5F cores .

    I need to check the history of this bug.

    Please let me know if a customer gets stuck on this requirement.

    Regards,

    Anil.

  • Hi Anil,

    Thanks for the response. I believe they will get stuck on this because they need to configure more regions.

    Also, they are using Motor Control SDK which is using MCU+SDK v.9.02 I believe, so they cannot use newest version of MCU+SDK unless they can somehow upgrade the MCU+SDK version that is included in Motor Control SDK. Is this possible?

    Best regards,

    Mari Tsunoda 

  • Hello Mari,

    If they get stuck, rather than adding an MPU region to the syscfg, we can manually write code for this. We need to make some changes to the Application.

    Is customers OK to add manual changes to the application to add more MPU regions ?

    ​Is it OK to tell these procedures by Friday or Monday as I am busy with other escalations?

    Does the customer want to upgrade from the new MCU+SDK version to the older Motorcontrol SDK?

    Regards,

    Anil.

  • Hi Anil,

    My customer is seeing similar issues with DPL_CFG for the multi-core project so I am thinking it might be easier to upgrade the MCU+SDK packaged with the Motor Control SDK to avoid all the bugs that might come with the multi-core project. But I am worried that since Motor Control SDK has not been tested with the newest MCU+SDK version, they may run into issues. Which would be the best choice from your viewpoint between the two?

    • Keep MCU+SDK v.9.02 included in Motor Control SDK and make manual changes to allow MPU region selection and DPL_CFG selection
    • Upgrade from MCU+SDK v.9.02 -> v.10.01 to avoid the bugs but risk other issues that might arise from untested version of MCU+SDK for Motor Control SDK 

    Here is the screenshot for DPL_CFG issue:

    They were able to configure DPL_CFG by adding the following line manually to the .syscfg file.

    const dpl_cfg         = scripting.addModule("/kernel/dpl/dpl_cfg");

    Also, have you seen other issues with the multi-core project debug like where shared regions are being over-counted?

    Best regards,

    Mari Tsunoda

  • Hello Mari,

    My suggestion is, please go with the new MCU+SDK version and do the Testing.

    If this testing take more time and is not able to meet any SW deadlines, then customers should stick with the old version and make the manual changes to the MPU and DPL CFG settings.

    And one more suggestion is that when a customer is working on the multicore project instead of taking a system project, take individual 4 R5F projects and integrate the customer application and compile them individually for temporally .

    When customers work on individual projects then there is no problem with the MPU settings and DPL settings and all can be configured till 16 regions for each one core.

    But only one problem is when customer is selecting the same peripheral in other creates an issue  and this is need be take care by customer .

    And, generating the multicore app image we need to take the all individual app images  and generate the multicore app image .

    For the compilation we need to run individual projects.

    Regards,

    Anil.

  • Hi Anil,

    customers should stick with the old version and make the manual changes to the MPU and DPL CFG settings.

    Do you know if there are any bugs that were fixed with regards to similar issues like these in the system project?

    I will suggest that they upgrade the MCU+SDK, but I want to know if there are any other issues as well.

    And, generating the multicore app image we need to take the all individual app images  and generate the multicore app image .

    Could you give steps on how to generate a multicore app from the individual app images?

    Best regards,

    Mari Tsunoda

  • Hello Mari,

    To generate multicore app images, please follow the link below.

    https://software-dl.ti.com/mcu-plus-sdk/esd/AM64X/10_01_00_32/exports/docs/api_guide_am64x/TOOLS_BOOT.html

    Do you know if there are any bugs that were fixed with regards to similar issues like these in the system project?

    I have looked at the confluence page and could not find anything related to DPL cfg.

    These issues need to be fixed.

    Regards,

    Anil.

  • Hi Anil,

    If this testing take more time and is not able to meet any SW deadlines, then customers should stick with the old version and make the manual changes to the MPU and DPL CFG settings.

    So my customer has decided to move forward by sticking to the old version. Could you tell me how to manually add the MPU regions to the system project without exceeding the limit?

    Best regards,

    Mari

  • Hello Mari,

    My  suggestion is that instead of taking a system project, take individual 4 R5F projects and integrate the customer application and compile them individually for temporally .

    When customers work on individual projects then there is no problem with the MPU settings and all cores can be configured till 16 regions for each one core.

    But only one problem is when the customer is selecting one peripheral in the one core and in other cores should not be use the same Peripheral and this is need be taken care by customer.

    Regards,

    Anil.