Attachments (0)
Nordic Case Info
  • Case ID: 293508
Options

how to decipher HARD FAULT / MPU FAULT on zephyr running on nrf52832

ziv123

hello Nordic

i work on nrf52832, trying to run it with zephyr application

one board runs fine (more/less, there are some MPU FAULT issue that i am facing with no luck yet and i brought it up also in this, waiting for answer, thread: 

 performance issue with multi threadn in ncs with nrf52832 )

when i add another board i get the following fail :

my question is, how to decipher this log, how can i know from this lines where my problem is, which isr, which thread ?

hope to get help on this matter soon

best regards

Ziv

Parents
  • 0

    Hi Ziv,

    If you set CONFIG_THREAD_NAME, the fault will usually list the thread name instead of (unknown).

    I usually use this together with "CONFIG_RESET_ON_FATAL_ERROR=n" , so I do not get spammed down by fault logs.

    Regards,
    Sigurd Hellesvik

  • 0 in reply to Sigurd Hellesvik

    hi Sigurd

     with the mentioned configs i see the following 

    also get this type of fail on a different branch

    still not sure how it directs me to where the problem is, beside me thinking that this is a different type of problem then the one in the beginning of this thread

    what am i missing to understand that better and how can i debug it ?

    hope to read you soon

    best regards

    Ziv

  • 0 in reply to ziv123

    Hi Ziv,

    As these configurations are for debugging, it is odd that they change the way your code fail.
    If you try to reproduce the issue 10 times in a row, do you always get the same issue?

    In your first error in this ticket, the error says that you should not create new thread inside your interrupts. Do you create any interrupts in your own code?

    Regards,
    Sigurd Hellesvik

  • 0 in reply to Sigurd Hellesvik

    hi Sigurd 

    yes i do create threads in my code but not from inside an ISR 

    i don't think the configs changed the type of error its just a different git branch failing on a different reason, just wanted to show that the configs did not add any input (including thread remains 'unknown'), as far as i understand, on both cases

    hope to read you soon

    best regards

    Ziv

Reply
  • 0 in reply to Sigurd Hellesvik

    hi Sigurd 

    yes i do create threads in my code but not from inside an ISR 

    i don't think the configs changed the type of error its just a different git branch failing on a different reason, just wanted to show that the configs did not add any input (including thread remains 'unknown'), as far as i understand, on both cases

    hope to read you soon

    best regards

    Ziv

Children
  • 0 in reply to ziv123

    Hi Ziv,

    To clarify, are you asking with help to either:

    A. Solve this issue you have, or

    B. Learn how you can use the Fault messages for debugging in general

    ?

    Regards,
    Sigurd Hellesvik

  • 0 in reply to Sigurd Hellesvik

    with both A and B

    A is a bit more urgent but B is crucial as well 

  • 0 in reply to ziv123

    Hi,

    B:
    In your PROJECT/build/zephyr folder, there is a set of files.
    Sometimes it is useful to have a look at the zephyr.elf file, and find the Faulting instruction address.
    However, in my experience, these tools are more useful for debugging:
    1. Eventual error messages or asserts from the crash.
    2. Debugging functionality, see Our Debugging with VS Code tutorial.
    3. Using the printk function to debug.

    A:

    I will help you debug some then. Lets start with some questions.

    You say you get the error with one board, but not the other.
    Are the two boards different?

    If you try to reproduce the issue 10 times in a row, do you always get the same issue?

    Do you create your own ISR?
    If so, can you post the code of these (only the ISR)?

    Regards,
    Sigurd Hellesvik

  • 0 in reply to Sigurd Hellesvik

    hi Sigurd

    regarding B:

    Sigurd Hellesvik said:
    Sometimes it is useful to have a look at the zephyr.elf file

    1. "the file is not displayed in the editor (vs code) because it is ether binary or uses an unsupported        

        encoding" that's the message i get when trying to see the zephyr.elf file

    2. sometimes when trying to start debug i get this in the debug terminal

        "Debugger requested to halt target..."

         and debugging does not start, seems like it is incapable to reset the device if it gets stuck or

         something like that, why is that and what can i do beside close the debug and try again

    3. my debug starts like this from assembly code, why is that, what can i learn from it, also there are unknown threads in the call stack as can be seen,

    how can i know which threads are those can/should/how do i give my threads names?

    and when i try to run i don't see anything change in my debugger but i see device is connecting and doing stuff so why ?

    regarding A:

    Sigurd Hellesvik said:
    You say you get the error with one board, but not the other.
    Are the two boards different

    i did not write this, and for simplicity lets leave the other board.

    Sigurd Hellesvik said:
    If you try to reproduce the issue 10 times in a row, do you always get the same issue?

    i have one board that has the following fails, on a certain operation at about 70% fail 30% works 

    the fail is the following:

    and it sometimes fails on init at restart but maybe 5 or 10% fail out of 90-95% pass 

    Sigurd Hellesvik said:
    Do you create your own ISR?
    If so, can you post the code of these (only the ISR)?

    not sure what you mean by create my own ISR, i implement interrupt handlers like in this case:

    i also have spi_done handler and use enable ppi with timer to send spi request from a sensor

    i do open a thread by a function cal of "trigger_set" and let it close gracefully when sampling is done

    the thread open looks like so:

    and the thread itself:

    any ideas ?

    hope to read you soon

    best regards

    Ziv

  • +1 in reply to ziv123

    Hi Ziv,

    ziv123 said:

    1. "the file is not displayed in the editor (vs code) because it is ether binary or uses an unsupported        

        encoding" that's the message i get when trying to see the zephyr.elf file

    Sorry, I gave you the wrong file ending. I meant zephyr.lst.

    ziv123 said:
    why is that and what can i do beside close the debug and try again

    From your explanation, I am unsure what this could be. If trying again works, it seems like a workaround for now.

    ziv123 said:
    my debug starts like this from assembly code, why is that, what can i learn from it, also there are unknown threads in the call stack as can be seen,

    Try to add a breakpoint in the top of your main code, and "continue" the debugger until it breaks at the breakpoint.

    ziv123 said:
    the fail is the following:

    "[00000000] <err> os:   Stacking error (context area might be not valid)"

    From these error codes, it looks like a stack overflow. Try to increase CONFIG_IDLE_STACK_SIZE to for example 4096.

    Do you still get the error then?

    EDIT: Another tip on how to debug: I copied the "Stacking Error (context area might not be valid)" into DevZone search.
    Among others,  found this case. That case looks like it has multiple good tips on how to debug stack overflows. I suggest you have a look at it,

    Regards,
    Sigurd Hellesvik

Terms of service