What is MPU FAULT?

[Kc Owen]

Hello，

I am using a customized NRF52832 circuit board. During the Bluetooth connection test with a mobile phone, the connection suddenly disconnected, and then quickly reconnected automatically. The log print information is as follows. What is the problem?

[01:10:56.880,828] <err> os: ***** MPU FAULT *****
[01:10:56.880,828] <err> os: Stacking error (context area might be not valid)
[01:10:56.880,859] <err> os: r0/a1: 0x4a25d56f r1/a2: 0x951ed952 r2/a3: 0x1525aa78
[01:10:56.880,859] <err> os: r3/a4: 0x250b10d0 r12/ip: 0x2594b26d r14/lr: 0x40c69401
[01:10:56.880,889] <err> os: xpsr: 0x41000200
[01:10:56.880,889] <err> os: s[ 0]: 0x912650bf s[ 1]: 0x14e98a5a s[ 2]: 0x01cc2022 s[ 3]: 0x24c730f2
[01:10:56.880,920] <err> os: s[ 4]: 0x0153829f s[ 5]: 0x9bba7910 s[ 6]: 0xd334a7b7 s[ 7]: 0x08372267
[01:10:56.880,920] <err> os: s[ 8]: 0x991172d8 s[ 9]: 0xfa6bca3b s[10]: 0xfba7ea19 s[11]: 0xdf60daeb
[01:10:56.880,950] <err> os: s[12]: 0x00000000 s[13]: 0x007fffff s[14]: 0x20009782 s[15]: 0x200092e5
[01:10:56.880,950] <err> os: fpscr: 0x200092ed
[01:10:56.880,981] <err> os: Faulting instruction address (r15/pc): 0x00029870
[01:10:56.881,011] <err> os: >>> ZEPHYR FATAL ERROR 2: Stack overflow on CPU 0
[01:10:56.881,042] <err> os: Current thread: 0x20001928 (unknown)
[01:10:57.391,204] <err> fatal_error: Resetting system

[Vidar Berg]

Hello,

The crashlog indicates that you have a stack overflow in one of your threads (https://docs.nordicsemi.com/bundle/ncs-latest/page/zephyr/hardware/arch/arm_cortex_m.html#mpu-assisted_stack_overflow_detection). I recommend enabling CONFIG_THREAD_NAME in your project configuration file to get the name of the thread with the stack overflow.

Best regards,

Vidar

[Kc Owen]

I'm sorry, I'm not quite sure what you mean,does the link you posted mean to enable stack overflow detection CONFIG_MPU_STACK_GUARD?

[Vidar Berg]

The stack guard is enabled by default, and the error log indicates that it has detected a stack overflow in one of your threads. I suggest enabling CONFIG_THREAD_NAME to find out which thread the stack overflow occured in. Then you can increase the stack size for that thread to fix the issue.